Concerted actions of DnaA complexes with DNA-unwinding sequences within and flanking replication origin oriC promote DnaB helicase loading

Unwinding of the replication origin and loading of DNA helicases underlie the initiation of chromosomal replication. In Escherichia coli, the minimal origin oriC contains a duplex unwinding element (DUE) region and three (Left, Middle, and Right) regions that bind the initiator protein DnaA. The Left/Right regions bear a set of DnaA-binding sequences, constituting the Left/Right-DnaA subcomplexes, while the Middle region has a single DnaA-binding site, which stimulates formation of the Left/Right-DnaA subcomplexes. In addition, a DUE-flanking AT-cluster element (TATTAAAAAGAA) is located just outside of the minimal oriC region. The Left-DnaA subcomplex promotes unwinding of the flanking DUE exposing TT[A/G]T(T) sequences that then bind to the Left-DnaA subcomplex, stabilizing the unwound state required for DnaB helicase loading. However, the role of the Right-DnaA subcomplex is largely unclear. Here, we show that DUE unwinding by both the Left/Right-DnaA subcomplexes, but not the Left-DnaA subcomplex only, was stimulated by a DUE-terminal subregion flanking the AT-cluster. Consistently, we found the Right-DnaA subcomplex–bound single-stranded DUE and AT-cluster regions. In addition, the Left/Right-DnaA subcomplexes bound DnaB helicase independently. For only the Left-DnaA subcomplex, we show the AT-cluster was crucial for DnaB loading. The role of unwound DNA binding of the Right-DnaA subcomplex was further supported by in vivo data. Taken together, we propose a model in which the Right-DnaA subcomplex dynamically interacts with the unwound DUE, assisting in DUE unwinding and efficient loading of DnaB helicases, while in the absence of the Right-DnaA subcomplex, the AT-cluster assists in those processes, supporting robustness of replication initiation.

The initiation of bacterial DNA replication requires local duplex unwinding of the chromosomal replication origin oriC, which is regulated by highly ordered initiation complexes. In Escherichia coli, the initiation complex contains oriC, the ATP-bound form of the DnaA initiator protein (ATP–DnaA), and the DNA-bending protein IHF (Fig. 1, A and B), which promotes local unwinding of oriC (1, 2, 3, 4). Upon this oriC unwinding, two hexamers of DnaB helicases are bidirectionally loaded onto the resultant single-stranded (ss) region with the help of the DnaC helicase loader (Fig. 1B), leading to bidirectional chromosomal replication (5, 6, 7, 8). However, the fundamental mechanism underlying oriC-dependent bidirectional DnaB loading remains elusive.Open in a separate windowFigure 1Schematic structures of oriC, DnaA, and the initiation complexes. A, the overall structure of oriC. The minimal oriC region and the AT-cluster region are indicated. The sequence of the AT-cluster−DUE (duplex-unwinding element) region is also shown below. The DUE region (DUE; pale orange bars) contains three 13-mer repeats: L-DUE, M-DUE, and R-DUE. DnaA-binding motifs in M/R-DUE, TT(A/G)T(T), are indicated by red characters. The AT-cluster region (AT cluster; brown bars) is flanked by DUE outside of the minimal oriC. The DnaA-oligomerization region (DOR) consists of three subregions called Left-, Middle-, and Right-DOR. B, model for replication initiation. DnaA is shown as light brown (for domain I–III) and darkbrown (for domain IV) polygons (right panel). ATP–DnaA forms head-to-tail oligomers on the Left- and Right-DORs (left panel). The Middle-DOR (R2 box)-bound DnaA interacts with DnaA bound to the Left/Right-DORs using domain I, but not domain III, stimulating DnaA assembly. IHF, shown as purple hexagons, bends DNA >160° and supports DUE unwinding by the DnaA complexes. M/R-DUE regions are efficiently unwound. Unwound DUE is recruited to the Left-DnaA subcomplex and mainly binds to R1/R5M-bound DnaA molecules. The sites of ssDUE-binding B/H-motifs V211 and R245 of R1/R5M-bound DnaA molecules are indicated (pink). Two DnaB homohexamer helicases (light green) are recruited and loaded onto the ssDUE regions with the help of the DnaC helicase loader (cyan). ss, single stranded.The minimal oriC region consists of the duplex unwinding element (DUE) and the DnaA oligomerization region (DOR), which contains specific arrays of 9-mer DnaA-binding sites (DnaA boxes) with the consensus sequence TTA[T/A]NCACA (Fig. 1A) (3, 4). The DUE underlies the local unwinding and contains 13-mer AT-rich sequence repeats named L-, M-, and R-DUE (9). The M/R-DUE region includes TT[A/G]T(A) sequences with specific affinity for DnaA (10). In addition, a DUE-flanking AT-cluster (TATTAAAAAGAA) region resides just outside of the minimal oriC (Fig. 1A) (11). The DOR is divided into three subregions, the Left-, Middle-, and Right-DORs, where DnaA forms structurally distinct subcomplexes (Fig. 1A) (8, 12, 13, 14, 15, 16, 17). The Left-DOR contains high-affinity DnaA box R1, low-affinity boxes R5M, τ1−2, and I1-2, and an IHF-binding region (17, 18, 19, 20). The τ1 and IHF-binding regions partly overlap (17).In the presence of IHF, ATP–DnaA molecules cooperatively bind to R1, R5M, τ2, and I1-2 boxes in the Left-DOR, generating the Left-DnaA subcomplex (Fig. 1B) (8, 17). Along with IHF causing sharp DNA bending, the Left-DnaA subcomplex plays a leading role in DUE unwinding and subsequent DnaB loading. The Middle-DOR contains moderate-affinity DnaA box R2. Binding of DnaA to this box stimulates DnaA assembly in the Left- and Right-DORs using interaction by DnaA N-terminal domain (Fig. 1B; also see below) (8, 12, 14, 16, 21). The Right-DOR contains five boxes (C3-R4 boxes) and cooperative binding of ATP–DnaA molecules to these generates the Right-DnaA subcomplex (Fig. 1B) (12, 18). This subcomplex is not essential for DUE unwinding and plays a supportive role in DnaB loading (8, 15, 17). The Left-DnaA subcomplex interacts with DnaB helicase, and the Right-DnaA subcomplex has been suggested to play a similar role (Fig. 1B) (8, 13, 16).In the presence of ATP–DnaA, M- and R-DUE adjacent to the Left-DOR are predominant sites for in vitro DUE unwinding: unwinding of L-DUE is less efficient than unwinding of the other two (Fig. 1B) (9, 22, 23). Deletion of L-DUE or the whole DUE inhibits replication of oriC in vitro moderately or completely, respectively (23). A chromosomal oriC Δ(AT-cluster−L-DUE) mutant with an intact DOR, as well as deletion of Right-DOR, exhibits limited inhibition of replication initiation, whereas the synthetic mutant combining the two deletions exhibits severe inhibition of cell growth (24). These studies suggest that AT-cluster−L-DUE regions stimulate replication initiation in a manner concerted with Right-DOR, although the underlying mechanisms remain elusive.DnaA consists of four functional domains (Fig. 1B) (4, 25). Domain I supports weak domain I–domain I interaction and serves as a hub for interaction with various proteins such as DnaB helicase and DiaA, which stimulates ATP–DnaA assembly at oriC (26, 27, 28, 29, 30). Two or three domain I molecules of the oriC–DnaA subcomplex bind a single DnaB hexamer, forming a stable higher-order complex (7). Domain II is a flexible linker (28, 31). Domain III contains AAA+ (ATPase associated with various cellular activities) motifs essential for ATP/ADP binding, ATP hydrolysis, and DnaA–DnaA interactions in addition to specific sites for ssDUE binding and a second, weak interaction with DnaB helicase (1, 4, 8, 10, 19, 25, 32, 33, 34, 35). Domain IV bears a helix-turn-helix motif with specific affinity for the DnaA box (36).As in typical AAA+ proteins, a head-to-tail interaction underlies formation of ATP–DnaA pentamers on the DOR, where the AAA+ arginine-finger motif Arg285 recognizes ATP bound to the adjacent DnaA protomer, promoting cooperative ATP–DnaA binding (Fig. 1B) (19, 32). DnaA ssDUE-binding H/B-motifs (Val211 and Arg245) in domain III sustain stable unwinding by directly binding to the T-rich (upper) strand sequences TT[A/G]T(A) within the unwound M/R-DUE (Fig. 1B) (8, 10). Val211 residue is included in the initiator-specific motif of the AAA+ protein family (10). For DUE unwinding, ssDUE is recruited to the Left-DnaA subcomplex via DNA bending by IHF and directly interacts with H/B-motifs of DnaA assembled on Left-DOR, resulting in stable DUE unwinding competent for DnaB helicase loading; in particular, DnaA protomers bound to R1 and R5M boxes play a crucial role in the interaction with M/R-ssDUE (Fig. 1B) (8, 10, 17). Collectively, these mechanisms are termed ssDUE recruitment (4, 17, 37).Two DnaB helicases are thought to be loaded onto the upper and lower strands of the region including the AT-cluster and DUE, with the aid of interactions with DnaC and DnaA (Fig. 1B) (25, 38, 39). DnaC binding modulates the closed ring structure of DnaB hexamer into an open spiral form for entry of ssDNA (40, 41, 42, 43). Upon ssDUE loading of DnaB, DnaC is released from DnaB in a manner stimulated by interactions with ssDNA and DnaG primase (44, 45). Also, the Left- and Right-DnaA subcomplexes, which are oriented opposite to each other, could regulate bidirectional loading of DnaB helicases onto the ssDUE (Fig. 1B) (7, 8, 35). Similarly, recent works suggest that the origin complex structure is bidirectionally organized in both archaea and eukaryotes (1, 46). In Saccharomyces cerevisiae, two origin recognition complexes containing AAA+ proteins bind to the replication origin region in opposite orientations; this, in turn, results in efficient loading of two replicative helicases, leading to head-to-head interactions in vitro (46). Consistent with this, origin recognition complex dimerization occurs in the origin region during the late M-G1 phase (47). The fundamental mechanism of bidirectional origin complexes might be widely conserved among species.In this study, we analyzed various mutants of oriC and DnaA in reconstituted systems to reveal the regulatory mechanisms underlying DUE unwinding and DnaB loading. The Right-DnaA subcomplex assisted in the unwinding of oriC, dependent upon an interaction with L-DUE, which is important for efficient loading of DnaB helicases. The AT-cluster region adjacent to the DUE promoted loading of DnaB helicase in the absence of the Right-DnaA subcomplex. Consistently, the ssDNA-binding activity of the Right-DnaA subcomplex sustained timely initiation of growing cells. These results indicate that DUE unwinding and efficient loading of DnaB helicases are sustained by concerted actions of the Left- and Right-DnaA subcomplexes. In addition, loading of DnaB helicases are sustained by multiple mechanisms that ensure robust replication initiation, although the complete mechanisms are required for precise timing of initiation during the cell cycle.     

Conformational States of Melittin at a Bilayer Interface

The distribution of peptide conformations in the membrane interface is central to partitioning energetics. Molecular-dynamics simulations enable characterization of in-membrane structural dynamics. Here, we describe melittin partitioning into dioleoylphosphatidylcholine lipids using CHARMM and OPLS force fields. Although the OPLS simulation failed to reproduce experimental results, the CHARMM simulation reported was consistent with experiments. The CHARMM simulation showed melittin to be represented by a narrow distribution of folding states in the membrane interface.Unstructured peptides fold into the membrane interface because partitioned hydrogen-bonded peptide bonds are energetically favorable compared to free peptide bonds (1–3). This folding process is central to the mechanisms of antimicrobial and cell-penetrating peptides, as well as to lipid interactions and stabilities of larger membrane proteins (4). The energetics of peptide partitioning into membrane interfaces can be described by a thermodynamic cycle (Fig. 1). State A is a theoretical state representing the fully unfolded peptide in water, B is the unfolded peptide in the membrane interface, C is the peptide in water, and D is the folded peptide in the membrane. The population of peptides in solution (State C) is best described as an ensemble of folded and unfolded conformations, whereas the population of peptides in State D generally is assumed to have a single, well-defined helicity, as shown in Fig. 1 A (5). Given that, in principle, folding in solution and in the membrane interface should follow the same basic rules, peptides in state D could reasonably be assumed to also be an ensemble. A fundamental question (5) is therefore whether peptides in state D can be correctly described as having a single helicity. Because differentiating an ensemble of conformations and a single conformation may be an impossible experimental task (5), molecular-dynamics (MD) simulations provide a unique high-resolution view of the phenomenon.Open in a separate windowFigure 1Thermodynamic cycles for peptide partitioning into a membrane interface. States A and B correspond to the fully unfolded peptide in solution and membrane interface, respectively. The folded peptide in solution is best described as an ensemble of unfolded and folded conformations (State C). State D is generally assumed to be one of peptides with a narrow range of conformations, but the state could actually be an ensemble of states as in the case of State C.Melittin is a 26-residue, amphipathic peptide that partitions strongly into membrane interfaces and therefore has become a model system for describing folding energetics (3,6–8). Here, we describe the structural dynamics of melittin in a dioleoylphosphatidylcholine (DOPC) bilayer by means of two extensive MD simulations using two different force fields.We extended a 12-ns equilibrated melittin-DOPC system (9) by 17 μs using the Anton specialized hardware (10) with the CHARMM22/36 protein/lipid force field and CMAP correction (11,12) (see Fig. S1 and Fig. S2 in the Supporting Material). To explore force-field effects, a similar system was simulated for 2 μs using the OPLS force field (13) (see Methods in the Supporting Material). In agreement with x-ray diffraction measurements on melittin in DOPC multilayers (14), melittin partitioned spontaneously into the lipid headgroups at a position below the phosphate groups at similar depth as glycerol/carbonyl groups (Fig. 2).Open in a separate windowFigure 2Melittin partitioned into the polar headgroup region of the lipid bilayer. (A) Snapshot of the simulation cell showing two melittin molecules (MLT1 and MLT2, in yellow) at the lipid-water interface. (B) Density cross-section of the simulation cell extracted from the 17-μs simulation. The peptides are typically located below the lipid phosphate (PO4) groups, in a similar depth as the glycerol/carbonyl (G/C) groups.To describe the secondary structure for each residue, we defined helicity by backbone dihedral angles (φ, ψ) within 30° from the ideal α-helical values (–57°, –47°). The per-residue helicity in the CHARMM simulation displays excellent agreement with amide exchange rates from NMR measurements that show a proline residue to separate two helical segments, which are unfolded below Ala⁵ and above Arg²² (15) (Fig. 3 A). In contrast, the OPLS simulation failed to reproduce the per-residue helicity except for a short central segment (see Fig. S3).Open in a separate windowFigure 3Helicity and conformational distribution of melittin as determined via MD simulation. (A) Helicity per residue for MLT1 and MLT2. (B) Corresponding evolution of the helicity. (C) Conformational distributions over the entire 17-μs simulation.Circular dichroism experiments typically report an average helicity of ∼70% for melittin at membrane interfaces (3,6,16,17), but other methods yield average helicities as high as 85% (15,18). Our CHARMM simulations are generally consistent with the experimental results, especially amide-exchange measurements (15); melittin helicity averaged to 78% for MLT1, whereas MLT2 transitioned from 75% to 89% helicity at t ≈ 8 μs, with an overall average helicity of 82% (Fig. 3 B). However, in the OPLS simulation, melittin steadily unfolds over the first 1.3 μs, after which the peptide remains only partly folded, with an average helicity of 33% (see Fig. S3). Similar force-field-related differences in peptide helicity were recently reported, albeit at shorter timescales (19). Although suitable NMR data are not presently available, we have computed NMR quadrupolar splittings for future reference (see Fig. S4).To answer the question asked in this article—whether the conformational space of folded melittin in the membrane interface can be described by a narrow distribution—the helicity distributions for the equilibrated trajectories are shown in Fig. 3 C. Whereas MLT1 in the CHARMM simulation produces a single, narrow distribution of the helicity, MLT2 has a bimodal distribution as a consequence of the folding event at t ≈ 8 μs (Fig. 3 C). We note that CHARMM force fields have a propensity for helix-formation and this transition might therefore be an artifact. We performed a cluster analysis to describe the structure of the peptide in the membrane interface. The four most populated conformations in the CHARMM simulation are shown in Fig. 4. The dominant conformation for both peptides was a helix kinked at G12 and unfolded at the last 5–6 residues of the C-terminus. The folding transition of MLT2 into a complete helix is visible by the 48% occupancy of a fully folded helix.Open in a separate windowFigure 4Conformational clusters of the two melittin peptides (MLT1 and MLT2) from the 17-μs CHARMM simulation in DOPC. Clustering is based on Cα-RMSD with a cutoff criterion of 2 Å.We conclude that the general assumption when calculating folding energetics holds: Folded melittin partitioned into membrane interfaces can be described by a narrow distribution of conformations. Furthermore, extended (several microsecond) simulations are needed to differentiate force-field effects. Although the CHARMM and OPLS simulations would seem to agree for the first few hundred nanoseconds, the structural conclusions differ drastically with longer trajectories, with CHARMM parameters being more consistent with experiments. However, as implied by the difference in substate distributions between MLT1 and MLT2, 17 μs might not be sufficient to observe the fully equilibrated partitioning process. The abrupt change in MLT2 might indicate that the helicity will increase to greater than experimentally observed in a sufficiently long simulation. On the other hand, it could be nothing more than a transient fluctuation. Increased sampling will provide further indicators of convergence of the helix partitioning process.     

Mitochondrial Protein Synthesis,Import, and Assembly

The mitochondrion is arguably the most complex organelle in the budding yeast cell cytoplasm. It is essential for viability as well as respiratory growth. Its innermost aqueous compartment, the matrix, is bounded by the highly structured inner membrane, which in turn is bounded by the intermembrane space and the outer membrane. Approximately 1000 proteins are present in these organelles, of which eight major constituents are coded and synthesized in the matrix. The import of mitochondrial proteins synthesized in the cytoplasm, and their direction to the correct soluble compartments, correct membranes, and correct membrane surfaces/topologies, involves multiple pathways and macromolecular machines. The targeting of some, but not all, cytoplasmically synthesized mitochondrial proteins begins with translation of messenger RNAs localized to the organelle. Most proteins then pass through the translocase of the outer membrane to the intermembrane space, where divergent pathways sort them to the outer membrane, inner membrane, and matrix or trap them in the intermembrane space. Roughly 25% of mitochondrial proteins participate in maintenance or expression of the organellar genome at the inner surface of the inner membrane, providing 7 membrane proteins whose synthesis nucleates the assembly of three respiratory complexes.TO think about how mitochondrial proteins are synthesized, imported, and assembled, it is useful to have a clear picture of the organellar structures that they, along with membrane lipids, compose and the functions that they carry out. As almost every schoolchild learns, mitochondria carry out oxidative phosphorylation, the controlled burning of nutrients coupled to ATP synthesis. Since Saccharomyces cerevisiae prefers to ferment sugars, respiration is a dispensable function and nonrespiring mutants are viable [although they cannot undergo meiosis (Jambhekar and Amon 2008)]. However, mitochondria themselves are not dispensable. A substantial fraction of intermediary metabolism occurs in mitochondria (Strathern et al. 1982), and at least one of these pathways, iron–sulfur cluster assembly, is essential for growth (Kispal et al. 2005). Thus, any mutation that prevents the biogenesis of mitochondria by, for example, preventing the import of protein constituents from the cytoplasm, is lethal (Baker and Schatz 1991).The mitochondria of S. cerevisiae are tubular structures at the cell cortex. While the number of distinct compartments can range from 1 to ∼50 depending upon conditions (Stevens 1981; Pon and Schatz 1991), continual fusion and fission events among them effectively form a single dynamic network (Nunnari et al. 1997). The outer membrane surrounds the tubules. The inner membrane has a boundary domain closely juxtaposed beneath the outer membrane and cristae domains that project internally from the boundary into the matrix (Figure 1A). The matrix is the aqueous compartment surrounded by the inner membrane. The aqueous intermembrane space lies between the membranes and is continuous with the space within cristae.Open in a separate windowFigure 1 Overview of mitochondrial structure in yeast. (A) Schematic of compartments comprising mitochondrial tubules. The outer membrane surrounds the organelle. The inner membrane surrounds the matrix and consists of two domains, the inner boundary membrane and the cristae membranes, which are joined at cristae junctions. The intermembrane space lies between the outer membrane and inner membrane. (B) Electron tomograph image of a highly contracted yeast mitochondrion observed en face (a) with the outer membrane (red) and (b) without the outer membrane. Reprinted by permission from John Wiley & Sons from Mannella et al. (2001).Inner membrane cristae are often depicted as baffles emanating from the boundary domain. However, electron tomography of mitochondria from several species, including yeast, shows that cristae actually emanate from the boundary membrane as narrow tubular structures at sites termed “crista junctions” and expand as they project into the matrix (Frey and Mannella 2000; Mannella et al. 2001) (Figure 1B). It seems clear that the boundary and cristae domains of the inner membrane have distinct compositions with respect to the respiratory complexes that are embedded preferentially in the cristae membrane domains, as well as other components (Vogel et al. 2006; Wurm and Jakobs 2006; Rabl et al. 2009; Suppanz et al. 2009; Zick et al. 2009; Davies et al. 2011).The outer and inner boundary membranes are connected at multiple contact sites, at least some of which are involved in protein translocation and may be transient (Pon and Schatz 1991). In addition, there appear to be firm contact sites, not directly involved with protein translocation, preferentially colocalized with crista junctions (Harner et al. 2011a).Overall, there appear to be ∼1000 distinct proteins in yeast mitochondria (Premsler et al. 2009). One series of proteomic studies on highly purified organelles identified 851 proteins thought to represent 85% of the total number of species (Sickmann et al. 2003; Reinders et al. 2006; Zahedi et al. 2006). Another study identified an additional 209 candidates (Prokisch et al. 2004). A computationally driven search for candidates involved in yeast mitochondrial function, coupled with experiments to assay respiratory function and maintenance of mitochondrial DNA (mtDNA), identified 109 novel candidates, although many of these may not be mitochondrial per se (Hess et al. 2009). Taking the boundary and cristae domains together, the inner membrane is the most protein-rich mitochondrial compartment, followed by the matrix (Daum et al. 1982).Only eight of the yeast mitochondrial proteins detected in proteomic studies are encoded by mtDNA and synthesized within the organelle. They are hydrophobic subunits of respiratory complexes III (bc₁ complex or ubiquinol-cytochrome c reductase), IV (cytochrome c oxidase), and V (ATP synthase), as well as a hydrophilic mitochondrial small subunit ribosomal protein. The remaining ∼99% of yeast mitochondrial proteins are encoded by nuclear genes, synthesized in cytoplasmic ribosomes, and imported into the organelle.An overview of known nuclearly encoded mitochondrial protein functions (Figure 2) reveals that ∼25% of them are involved directly in genome maintenance and expression of the eight major mitochondrial genes (Schmidt et al. 2010). The functions of ∼20% of the proteins are not known. Fifteen percent are involved in the well-known processes of energy metabolism. Protein translocation, folding, and turnover functions occupy ∼10% of mitochondrial proteins.Open in a separate windowFigure 2 Classification of identified mitochondrial proteins according to function. Reprinted by permission from Nature Publishing Group from Schmidt et al. (2010).The following discussion reviews our understanding of the biogenesis of mitochondria starting on the outside, the cytoplasm, and working inward through the mitochondrial compartments.     

Crystal structure of Bax bound to the BH3 peptide of Bim identifies important contacts for interaction

The BH3-only protein Bim is a potent direct activator of the proapoptotic effector protein Bax, but the structural basis for its activity has remained poorly defined. Here we describe the crystal structure of the BimBH3 peptide bound to BaxΔC26 and structure-based mutagenesis studies. Similar to BidBH3, the BimBH3 peptide binds into the cognate surface groove of Bax using the conserved hydrophobic BH3 residues h1–h4. However, the structure and mutagenesis data show that Bim is less reliant compared with Bid on its ‘h0'' residues for activating Bax and that a single amino-acid difference between Bim and Bid encodes a fivefold difference in Bax-binding potency. Similar to the structures of BidBH3 and BaxBH3 bound to BaxΔC21, the structure of the BimBH3 complex with BaxΔC displays a cavity surrounded by Bax α1, α2, α5 and α8. Our results are consistent with a model in which binding of an activator BH3 domain to the Bax groove initiates separation of its core (α2–α5) and latch (α6–α8) domains, enabling its subsequent dimerisation and the permeabilisation of the mitochondrial outer membrane.The intrinsic pathway to apoptosis is regulated by interactions between members of three factions of the Bcl-2 protein family: the BH3-only proteins such as Bim and Bid, which initiate the process, the essential effectors Bax and Bak, and the prosurvival members, which oppose the action of both other factions.¹ The interactions between prosurvival Bcl-2 family members and BH3 peptides have been well characterised as the earliest studies with Bcl-x_L and a BakBH3 peptide.² Such complexes are readily formed in solution by incubating the C-terminally (ΔC) truncated prosurvival Bcl-2 protein with a BH3 peptide. The absence of the C-terminal segment that can anchor the Bcl-2 protein in a membrane apparently has little effect on the ensuing complex. That complex is believed to be responsible for the antiapoptotic function of Bcl-2, by sequestration of the BH3 motif either of the so-called BH3-only proteins such as Bim (''mode 1'') or of Bax or Bak (''mode 2'').³Although proapoptotic Bax and Bak have very similar three-dimensional structures to their prosurvival relatives,^{4, 5, 6} until recently^{7, 8} no structure of a complex of either Bax or Bak with a BH3 peptide had been captured, despite an accumulation of evidence that Bax and Bak could be activated directly by interaction with the BH3-only proteins Bid, Bim and possibly others.^{9, 10, 11, 12, 13}Unlike Bak, which is constitutively anchored in the mitochondrial outer membrane (MOM) via its C-terminal segment, Bax is largely cytosolic in healthy cells and accumulates at the MOM only upon a death signal.^{14, 15} There it is believed to display at least two different conformers,^{16, 17} one loosely associated with the MOM and another in which its membrane anchor (helix α9) is inserted into the MOM. In striking contrast to the antiapoptotic relatives of Bcl-2, a construct of Bax lacking its C-terminal membrane anchor, BaxΔC21, has no measurable interaction with BH3 peptides. However, in the presence of the detergent octylglucoside binding is detected by surface plasmon resonance (SPR) for the BH3 peptides of Bim, Bid, Bak and Bax itself with IC50s in the range of 0.1–1μM,^{7, 18} some 100-fold weaker compared with those measured similarly with (for example) Bcl-x_LΔC, where no detergent is required. Weaker interactions between BidBH3 or BimBH3 and BaxΔC as compared with Bcl-x_LΔC are not inconsistent with various models for the function of the Bcl-2 protein family whereby the prosurvival molecules sequester BH3 motifs with high affinity and long half-lives, but proapoptotic Bax and Bak are activated by transient (‘hit-and-run'') interactions with BH3 motifs.^{19, 20, 21}Complexes of BaxΔC21 bound to BH3 peptides from Bid and Bax have been prepared by coincubation of the protein with CHAPS and an excess of the peptides.⁷ Under these conditions, the protein undergoes a conformational change and dimerises via domain swapping of helical segments α2–α5 and α6–α8, dubbed ‘core'' and ‘latch'' domains, respectively. Although this ‘core/latch dimer'' is thought to be an in vitro artefact, its formation is diagnostic for the core and latch separation, which is required for membrane-associated Bax to dimerise via its core domains and then to permeabilise the MOM.⁷ If the latch domain is absent, as in a recombinant construct of GFP fused to Bax α2–α5, the core domain forms BH3:groove symmetric dimers,⁷ which, consistent with a wide body of evidence,^{21, 22, 23, 24, 25} are present in apoptotic pores.Previous work⁷ highlighted the importance of two hydrophobic ‘h0'' residues (Figure 1) in the peptide (I82/I83 in BidBH3) in governing Bid''s ability to activate Bax. Similar to Bid, Bim is also a potent direct activator of Bax, and the ‘h0'' amino acids in Bim are proline and glutamic acid. In the absence of a structure of BimBH3:BaxΔC, it remained unclear how these ‘h0'' residues were accommodated. Here we describe the crystal structures of BimBH3 26- and 20-mer peptides bound to BaxΔC26. Comparison with the structure of BidBH3:BaxΔC21 allows a dissection of the critical contacts between these two peptides and BaxΔC. The binding profiles of mutant BH3 peptides illustrate that BimBH3 binding to Bax is less dependent on the ‘h0'' residues compare with that in the case for BidBH3. The BimBH3 complex displays a similar cavity adjacent to Bax α1, α2, α5 and α8 as seen in the BidBH3 complex. We also describe a structure of BidBH3 bound to a BaxΔC21 mutant, I66A, which is more typical of the BH3 signature of antiapoptotic Bcl-2 family proteins^{7, 26}Open in a separate windowFigure 1BimBH3 binds BaxΔC. (a) BH3 peptide sequences used in this study, indicating the 5 hydrophobic amino-acid positions ‘h0''–‘h4''. (b) The core/latch dimer of BaxΔC26 bound to BimBH3. The two Bax polypeptides, shown here as cartoons, are coloured yellow and grey, and the two Bim peptides cyan and orange. A crystallographic dyad symmetry axis passes through the centre of this particle. (c) Structure of BimBH3:BaxΔC26 complex. The globular unit depicted comprises Bax residues 1–128 from one polypeptide and 129–166 from the other, together with the associated Bim peptide. Bax is represented by its surface and colour coded according to surface charge (blue, positive potential (4kT/e); red, negative potential (−2kT/e); calculated using the Adaptive Poisson–Boltzmann Solver.⁴¹ The trace of the Bim peptide (cyan) is shown with ‘h0'' (P144, E145), ‘h1'' (I148), ‘h2'' (L152), ‘h3'' (I155) and ‘h4'' (F159) represented as sticks. (d) Overlay of BimBH3:BaxΔC26 with BidBH3:BaxΔC21 (PDB:4BD2). Structures represented as cartoon ribbons, yellow for Bax in the Bim complex and magenta for Bax in the Bid complex. The peptides (Bim cyan and Bid blue) stand vertically in the foreground in this view (similar to Figure 1c), with their N termini at the bottom of the figure     

Interplay between Velocity and Travel Distance of Kinesin-based Transport in the Presence of Tau

Although the disease-relevant microtubule-associated protein tau is known to severely inhibit kinesin-based transport in vitro, the potential mechanisms for reversing this detrimental effect to maintain healthy transport in cells remain unknown. Here we report the unambiguous upregulation of multiple-kinesin travel distance despite the presence of tau, via decreased single-kinesin velocity. Interestingly, the presence of tau also modestly reduced cargo velocity in multiple-kinesin transport, and our stochastic simulations indicate that the tau-mediated reduction in single-kinesin travel underlies this observation. Taken together, our observations highlight a nontrivial interplay between velocity and travel distance for kinesin transport, and suggest that single-kinesin velocity is a promising experimental handle for tuning the effect of tau on multiple-kinesin travel distance.Conventional kinesin is a major microtubule-based molecular motor that enables long-range transport in living cells. Although traditionally investigated in the context of single-motor experiments, two or more kinesin motors are often linked together to transport the same cargo in vivo (1–4). Understanding the control and regulation of the group function of multiple kinesins has important implications for reversing failure modes of transport in a variety of human diseases, particularly neurodegenerative diseases. Tau is a disease-relevant protein enriched in neurons (5,6). The decoration of microtubules with tau is known to strongly inhibit kinesin transport in vitro (7–9), but how kinesin-based transport is maintained in the presence of high levels of tau, particularly in healthy neurons, remains an important open question. To date, no mechanism has been directly demonstrated to reverse the inhibitory effect of tau on kinesin-based transport. Here we present a simple in vitro study that demonstrates the significant upregulation of multiple-kinesin travel distance with decreasing ATP concentration, despite the presence of tau.This investigation was motivated by our recent finding that single-kinesin velocity is a key controller for multiple-kinesin travel distance along bare microtubules (10). The active stepping of each kinesin motor is stimulated by ATP (11), and each kinesin motor remains strongly bound to the microtubule between successive steps (10,11). As demonstrated for bare microtubules (10), with decreasing ATP concentrations, each microtubule-bound kinesin experiences a decreased stepping rate per unit time and spends an increased fraction of time in the strongly bound state; additional unbound kinesins on the same cargo have more time to bind to the microtubule before cargo travel terminates. Thus, reductions in single-kinesin velocity increase the probability that at least one kinesin motor will remain bound to the microtubule per unit time, thereby increasing the travel distance of each cargo (10). Because this effect only pertains to the stepping rate of each individual kinesin and does not address the potential presence of roadblocks such as tau on the microtubules, we hypothesized in this study that single-kinesin velocity may be exploited to relieve the impact of tau on multiple-kinesin travel distance.We focused our in vitro investigation on human tau 23 (htau23, or 3RS tau), an isoform of tau that exhibits the strongest inhibitory effect on kinesin-based transport (7–9). Importantly, htau23 does not alter the stepping rate of individual kinesins (7,9), supporting our hypothesis and enabling us to decouple single-kinesin velocity from the potential effects of tau. We carried out multiple-kinesin motility experiments using polystyrene beads as in vitro cargos (8,10), ATP concentration as an in vitro handle to controllably tune single-kinesin velocity (10,11), and three input kinesin concentrations to test the generality of potential findings for multiple-kinesin transport. Combined with previous two-kinesin studies (10,12), our measurements of travel distance (Fig. 1 A) indicate that the lowest kinesin concentration employed (0.8 nM) corresponds to an average of ∼2–3 kinesins per cargo. Note that in the absence of tau, the observed decrease in bead velocity at the higher kinesin concentrations (Fig. 1 A) is consistent with a recent in vitro finding (13). At 1 mM ATP, htau23 reduced kinesin-based travel distance by a factor of two or more (Fig. 1, A and B). This observation is in good agreement with previous reports (7,8).Open in a separate windowFigure 1Distributions of multiple-kinesin travel distances measured at three experimental conditions, to verify the effect of tau (A and B) and to investigate the impact of single-kinesin velocity on the tau effect (B and C). Shaded bars at 8.7 μm indicate counts of travel exceeding the field of view. The mean travel distance (d; ± standard error of mean, SEM), sample size (n), and corresponding mean velocity (v; ± SEM) are indicated. MT denotes microtubule. Mean travel distance increased substantially at 20 μM ATP (C), despite the presence of htau23. This effect persisted across all three kinesin concentrations tested (left to right).Consistent with our hypothesis, reducing the available ATP concentration to 20 μM increased the multiple-kinesin travel distance by >1.4-fold for all three input kinesin concentrations (Fig. 1, B and C), despite the presence of htau23. The corresponding reduction in single-kinesin velocity with decreasing ATP concentration (10,11) is reflected in the ∼3.4-fold reduction in the measured bead velocities (Fig. 1, B and C). Therefore, the strong negative relationship between single-kinesin velocity and multiple-kinesin travel distance occurs not only for bare microtubules (10), but also for tau-decorated microtubules.What causes the observed increase in travel distance at the lower ATP concentration (Fig. 1, B and C)? In addition to the mechanism discussed above for the case of bare microtubules (10), an intriguing mechanism was suggested by recent studies of tau-microtubule interactions in which htau23 was observed to dynamically diffuse along microtubule lattices (14,15): reducing the stepping rate of a microtubule-bound kinesin may effectively increase the probability that a tau roadblock can diffuse away before the kinesin takes its next step.Perhaps surprisingly, although htau23 does not impact single-kinesin velocity (7,9), we observed a modest reduction in the average velocity of multiple-kinesin transport in experiments using tau-decorated microtubules (Fig. 1, A and B). This decreased velocity reflects a substantially larger variance in the instantaneous velocity for bead trajectories in the presence of htau23 (see Fig. S1 in the Supporting Material), as quantified by parsing each bead trajectory into a series of constant-velocity segments using a previously developed automatic software incorporating Bayesian statistics (16).To test the possibility that single-kinesin travel distance impacts multiple-kinesin velocity, we performed stochastic simulations (see the Supporting Material) that assumed N identical kinesin motors available for transport and included kinesin’s detachment kinetics (17). Previously, this model successfully captured multiple-dynein travel distances in vivo using single-dynein characteristics measured in vitro (18). In this study, we introduced one (and only one) free parameter to reflect the probability of each bound kinesin encountering tau at each step. When encountering tau, each kinesin has a 54% probability of detaching from the microtubule (interpolated from Fig. 2A of Dixit et al. (7)); the undetached kinesin is assumed to remain engaged in transport and completes its step along the microtubule despite the presence of tau.Remarkably, our simple simulation suggested that the tau-mediated reduction in single-kinesin travel is sufficient to reduce multiple-kinesin velocity (Fig. 2 A). The majority of the velocity decrease is predicted to occur at the transition from single-kinesin to two-kinesin transport (Fig. 2). Further decreases in cargo velocity with increasing motor number are predicted to be modest and largely independent of tau (Fig. 2 B). The results of our simulation remain qualitatively the same when evaluated at two bounds (40 and 65%) encompassing the interpolated 54% probability of kinesin detaching at tau (see Fig. S2).Open in a separate windowFigure 2Stochastic simulations predict a tau-dependent reduction in multiple-kinesin velocity, assuming that the only effect of tau protein is to prematurely detach kinesin from the microtubule (or, to reduce single-kinesin travel distance). (A) Average velocity of cargos carried by the indicated number of kinesins was evaluated at 1 mM ATP, and for four probabilities that a kinesin may encounter tau at each step. Mean velocity was evaluated using 600 simulated trajectories for all simulation conditions. Error bars indicate SEM. (B) Change in cargo velocity with each additional kinesin (ΔVel/kinesin) as a function of tau-encounter probability. These values were calculated from cargo velocities shown in panel A. Error bars indicate SEM.We note that our simple simulations do not consider the possibility that kinesin may pause in front of a tau roadblock, as previously reported in Dixit et al. (7). We omitted this consideration because the interaction strength between kinesin and the microtubule in such a paused state is unknown. In a multiple-motor geometry, could a paused kinesin be dragged along by the other motors bound to the same cargo? Could a tau roadblock be forcefully swept off the microtubule surface by the collective motion of the cargo-motor complex? Significant experimental innovations are necessary to specifically address these questions in future multiple-motor assays and to guide modeling efforts. Nonetheless, our simple simulation demonstrates that reducing single-kinesin travel distance is sufficient to decrease multiple-kinesin travel distance.Taken together, our observations highlight a nontrivial interplay between velocity and travel distance for kinesin-based transport in the presence of tau. We uncover a previously unexplored dual inhibition of tau on kinesin-transport: in addition to limiting cargo travel distance, the tau-mediated reduction in single-kinesin travel distance also leads to a modest reduction in multiple-kinesin velocity. We provide what we believe to be the first demonstration of the unambiguous upregulation of multiple-kinesin travel distance despite the presence of tau, via reducing single-kinesin velocity, suggesting a mechanism that could be harnessed for future therapeutic interventions in diseases that result from aberrant kinesin-based transport.     

Voltage Sensor Gating Charge Transfer in a hERG Potassium Channel?Model

Relaxation of a hERG K⁺ channel model during molecular-dynamics simulation in a hydrated POPC bilayer was accompanied by transitions of an arginine gating charge across a charge transfer center in two voltage sensor domains. Inspection of the passage of arginine side chains across the charge transfer center suggests that the unique hydration properties of the arginine guanidine cation facilitates charge transfer during voltage sensor responses to changes in membrane potential, and underlies the preference of Arg over Lys as a mobile charge carrier in voltage-sensitive ion channels.The response of voltage-sensitive ion channels to changes in membrane potential is mediated by voltage sensor domains (VSD) containing a transmembrane helical segment (S4) with a repeating motif of positively charged and hydrophobic amino acids (Fig. 1) (1,2). Changes in membrane potential drive the S4 helix through the membrane plane with the charged side chains (largely arginine) on S4 swapping Glu/Asp carboxylate partners that lie on less mobile elements of the VSD (2). Movement of S4 is coupled to the ion-conducting pore to transmit changes in membrane potential to channel gating (3).Open in a separate windowFigure 1Structures of the VSD of membrane domains before MD in a POPC bilayer. The S2 (pink) and S4 (blue) helices of the VSD of the hERG model (A) and Kv1.2/2.1 chimera structure (B) are highlighted. (C) Sequence alignment of S2 and S4 among homologous voltage-sensitive K⁺ channels.The VSD charge-pairing motif of K⁺ and Na⁺ channels is best represented in VSD states at zero membrane potential (S4 helix up) for which crystal structures exist for Kv1.2 (4), Kv1.2/2.1 chimera (5), and Na_v channels (6,7). In these states, positively charged residues on the intra- and extracellular sections of the S4 helix are separated by a hydrophobic charge-transfer center (CTC) (1) or plug (8) containing a highly conserved Phe residue (Fig. 1). This plug restricts water incursion across the VSD, focusing the electric field across a narrow region near the bilayer center. In voltage-driven transitions between S4 down- and up-states, positively charged S4 side chains move across the CTC.The ether-à-go-go (eag) and eag-related family of voltage-sensitive K⁺ channels likely share similar charge pairing interactions with VSDs in other channels (9,10). However, eag VSDs contain an extra negative charge on S2 (underlined in Fig. 1 C) so that in hERG, Asp residues (D460 and D466) lie approximately one helical turn above and below the conserved charge-transfer center Phe (F463) (Fig. 1). This eag-specific motif might be expected to facilitate transfer of Arg side chains through the CTC and to stabilize the voltage sensor (VS) in the up state. We recently described an open state (VS-up) hERG model built on the crystal structure template of the Kv1.2/2.1 chimera and molecular-dynamics (MD) simulation of this model in a hydrated POPC bilayer (11). We have inspected an extended version of this simulation and identified transitions of a gating charge into the CTC despite the absence of a membrane potential change. These transitions are absent in equivalent MD simulations of the chimera structure in a POPC bilayer.Fig. 1 shows a single VS from starting structures of the hERG model and the chimera structure in a hydrated POPC bilayer, after restrained MD to anneal the protein-lipid interface (see Methods in the Supporting Material). Because the hERG model is constructed on the chimera structure according to the alignment in Fig. 1 the pattern of pairing between S4 charges and acidic VS side chains is equivalent in the hERG model and chimera structure.The arrangement of charge-paired side chains remains constant during MD in all subunits of the chimera (e.g., Fig. 2 E and see Fig. S2 in the Supporting Material). However, in two subunits of the hERG model the R534 side chain moves toward the extracellular side of the bilayer, sliding into the CTC to form a charge interaction with the extra Asp residue (D460 in hERG) that lies just above F463 (Fig. 2, A–C). This transition is facilitated by changes in side-chain rotamers of R534 and F463 as the planar Arg guanidine group rotates past the F463 ring, and the availability of D460 as a counterion for the R534 guanidine (Fig. 2). Movement of an Arg guanidine past the Phe side chain of the CTC is similar to that described in steered MD of an isolated VS domain (12).Open in a separate windowFigure 2Movement of the R534 side chain across the CTC in chain a of the hERG model simulation (A). Similar transitions are observed in chains a and b (panels B and C), but not chains c (D) or d (not shown), where the R534 side chain remains close to D466. In all subunits of the Kv1.2/2.1 chimera simulation, charge pairing of the starting structure (Fig. 1B) was maintained throughout (e.g., panel E and see Fig. S2 in the Supporting Material). (Black and blue lines) Distances from the Arg CZ or Lys ε atom to the two O atoms, respectively, of Asp or Glu.Mason et al. (13) have shown, using neutron scattering, that the low charge density guanidine cation (Gdm⁺) corresponding to the Arg side chain is poorly hydrated above and below the molecular plane. This property may underlie the universal preference for Arg (over Lys) in voltage sensor charge transfer. Although the poorly-hydrated surfaces of Gdm⁺ interact favorably with nonpolar (especially planar) surfaces (14,15), Gdm⁺ retains in-plane hydrogen bonding (13). In the transition of R534 across the CTC, in-plane solvation of the guanidine side chain is provided initially by D466, D501, and water molecules below the CTC, and during and after the transition by D501 and D460 side chains and waters above the CTC (Fig. 3, A and B). Complete transfer of the R534 side chain across the CTC was not observed, but would be expected to involve movement of the guanidine group away from H-bonding distance with D501.Open in a separate windowFigure 3In-plane solvation of R534 guanidine in the charge transfer center during the hERG model MD (A). (Dotted lines) H-bond distances of <2.5 Å. The right-hand group consists of top-down (B) and end-on (C) views of the distribution of oxygen atoms around the side chain of hERG R534 at 20-ns intervals during MD (subunit a). (D) End-on view of equivalent atom distributions around the K302 side chain during the Kv1.2/2.1 chimera MD (subunit c). (Red spheres, water O; pink, Asp OD1 and OD2; purple:, Glu OE1 and OE2.)The atom distribution around the R534 side chain during MD (Fig. 3, B and C) conforms to the experimental Gdm⁺ hydration structure (13), with H-bonding to waters and side-chain Asp O atoms exclusively in the guanidine plane. The passage of Gdm⁺ through the CTC is facilitated by the hydrophobic nature of Gdm⁺ above and below the molecular plane (13), which allows interaction with the nonpolar groups (especially F463) in the CTC (Fig. 3 A and see Fig. S3). This contrasts with the solvation properties of the Lys amino group (e.g., K302 of the Kv1.2/2.1 chimera (Fig. 1), which has a spherical distribution of H-bonding and charge-neutralizing oxygen atoms (Fig. 3 D and see Fig. S4).To further test these interpretations, we ran additional MD simulations of the isolated hERG VS domain model and an R534K mutant in a hydrated POPC bilayer. Again, the R534 side chain entered the CTC in the wild-type model simulation whereas the K534 side chain did not (see Fig. S5). Inspection of the atom distributions in Fig. 3 D (and see Fig. S4) indicates that the pocket below the conserved Phe of the CTC is particularly favorable for a Lys side chain, with waters and acidic side chains that satisfy the spherical solvation requirements of the terminal amino group, and nonpolar side chains that interact with the aliphatic part of the side chain.The occurrence of transitions of the R534 side chain through the CTC in the hERG model, in the absence of a change in membrane potential, indicates a relaxation from a less-stable starting structure. However, the path of the R534 side chain provides useful molecular-level insight into the nature of charge transfer in voltage sensors. How do these observations accord with broader evidence of charge transfer in voltage-sensitive channels in general, and hERG in particular? Studies with fluorinated analogs of aromatic side chains equivalent to F463 of hERG or F233 of the chimera indicate the absence of a significant role for cation-π interactions involving the CTC aromatic group in K⁺ and Na_v channels, although a planar side chain is preferred in some cases (1,16). In hERG, F463 can be replaced by M, L, or V with small effects on channel gating (17), indicating that the hERG CTC requires only a bulky nonpolar side chain to seal the hydrophobic center of the VS and allow passage of the Arg side chain through the CTC. Both absence of requirement for cation-π interactions, and accommodation of nonplanar hydrophobic side chains in a functional hERG CTC, are broadly consistent with the interpretation that it is the poorly-hydrated nature of the Arg guanidine group above and below the molecular plane (together with its tenacious proton affinity (18)) that governs its role in carrying gating charge in voltage sensors.While the simulations suggest that R534 may interact with D460 in the open channel state, the possibility that the extra carboxylate side chain above the CTC might facilitate gating charge transfer is seemingly inconsistent with the slow activation of hERG, although hERG D460C does activate even more slowly than the WT channel (9). However, S4 movement in hERG occurs in advance of channel opening (19), and slow gating is partly mediated by interactions involving hERG cytoplasmic domains (20); thus, slow S4 movement may not be an inherent property of the hERG voltage sensor. Recent studies show that when hERG gating is studied at very low [Ca²⁺] (50 μM) and low [H⁺] (pH 8.0), the channel is strongly sensitized in the direction of the open state; this effect is reduced in hERG D460C (and hERG D509C) (10). These observations support a role for the extra hERG Asp residues in binding Ca²⁺ (and H⁺) (10), allowing the channel to be allosterically responsive to changes in pH and [Ca²⁺]. A true comparison of a hERG model with experimental channel gating might involve studies on a channel lacking cytoplasmic domains that modulate gating, and using conditions (high pH and low [Ca²⁺]) that leave the eag-specific Asp residues unoccupied. This could reveal the inherent current-voltage relationships and kinetics of the hERG voltage sensor.     

The tubulin code: Molecular components,readout mechanisms,and functions

Microtubules are cytoskeletal filaments that are dynamically assembled from α/β-tubulin heterodimers. The primary sequence and structure of the tubulin proteins and, consequently, the properties and architecture of microtubules are highly conserved in eukaryotes. Despite this conservation, tubulin is subject to heterogeneity that is generated in two ways: by the expression of different tubulin isotypes and by posttranslational modifications (PTMs). Identifying the mechanisms that generate and control tubulin heterogeneity and how this heterogeneity affects microtubule function are long-standing goals in the field. Recent work on tubulin PTMs has shed light on how these modifications could contribute to a “tubulin code” that coordinates the complex functions of microtubules in cells.

Introduction

Microtubules are key elements of the eukaryotic cytoskeleton that dynamically assemble from heterodimers of α- and β-tubulin. The structure of microtubules, as well as the protein sequences of α- and β-tubulin, is highly conserved in evolution, and consequently, microtubules look alike in almost all species. Despite the high level of conservation, microtubules adapt to a large variety of cellular functions. This adaptation can be mediated by a large panel of microtubule-associated proteins (MAPs), including molecular motors, as well as by mechanisms that directly modify the microtubules, thus either changing their biophysical properties or attracting subsets of MAPs that convey specific functions to the modified microtubules. Two different mechanism can generate microtubule diversity: the expression of different α- and β-tubulin genes, referred to as tubulin isotypes, and the generation of posttranslational modifications (PTMs) on α- and β-tubulin (Figs. 1 and ​and2).2). Although known for several decades, deciphering how tubulin heterogeneity controls microtubule functions is still largely unchartered. This review summarizes the current advances in the field and discusses new concepts arising.Open in a separate windowFigure 1.Tubulin heterogeneity generated by PTMs. (A) Schematic representation of the distribution of different PTMs of tubulin on the α/β-tubulin dimer with respect to their position in the microtubule lattice. Acetylation (Ac), phosphorylation (P), and polyamination (Am) are found within the tubulin bodies that assemble into the microtubule lattice, whereas polyglutamylation, polyglycylation, detyrosination, and C-terminal deglutamylation take place within the C-terminal tubulin tails that project away from the lattice surface. The tubulin dimer represents TubA1A and TubB2B (Fig. 2), and modification sites for polyglutamylation and polyglycylation have been randomly chosen. (B) Chemical structure of the branched peptide formed by polyglutamylation and polyglycylation, using the γ-carboxyl groups of the modified glutamate residues as acceptor sites for the isopeptide bonds. Note that in the case of polyglutamylation, the elongation of the side chains generates classical peptide bonds (Redeker et al., 1991).Open in a separate windowFigure 2.Heterogeneity of C-terminal tails of tubulin isotypes and their PTMs. The amino acid sequences of all tubulin genes found in the human genome are indicated, starting at the last amino acid of the folded tubulin bodies. Amino acids are represented in single-letter codes and color coded according to their biochemical properties. Known sites for polyglutamylation are indicated (Eddé et al., 1990; Alexander et al., 1991; Rüdiger et al., 1992). Potential modification sites (all glutamate residues) are indicated. Known C-terminal truncation reactions of α/β-tubulin (tub) are indicated. The C-terminal tails of the yeast Saccharomyces cerevisiae are shown to illustrate the phylogenetic diversity of these domains.

Tubulin isotypes

The cloning of the first tubulin genes in the late 1970’s (Cleveland et al., 1978) revealed the existence of multiple genes coding for α- or β-tubulin (Ludueña and Banerjee, 2008) that generate subtle differences in their amino acid sequences, particularly in the C-terminal tails (Fig. 2). It was assumed that tubulin isotypes, as they were named, assemble into discrete microtubule species that carry out unique functions. This conclusion was reinforced by the observation that some isotypes are specifically expressed in specialized cells and tissues and that isotype expression changes during development (Lewis et al., 1985; Denoulet et al., 1986). These high expectations were mitigated by a subsequent study showing that all tubulin isotypes freely copolymerize into heterogeneous microtubules (Lewis et al., 1987). To date, only highly specialized microtubules, such as ciliary axonemes (Renthal et al., 1993; Raff et al., 2008), neuronal microtubules (Denoulet et al., 1986; Joshi and Cleveland, 1989), and microtubules of the marginal band of platelets (Wang et al., 1986; Schwer et al., 2001) are known to depend on some specific (β) tubulin isotypes, whereas the function of most other microtubules appears to be independent of their isotype composition.More recently, a large number of mutations in single tubulin isotypes have been linked to deleterious neurodevelopmental disorders (Keays et al., 2007; Fallet-Bianco et al., 2008; Tischfield et al., 2010; Cederquist et al., 2012; Niwa et al., 2013). Mutations of a single tubulin isotype could lead to an imbalance in the levels of tubulins as a result of a lack of incorporation of mutant isoforms into the microtubule lattice or to incorporation that perturbs the architecture or dynamics of the microtubules. The analysis of tubulin disease mutations is starting to reveal how subtle alterations of the microtubule cytoskeleton can lead to functional aberrations in cells and organisms and might provide novel insights into the roles of tubulin isotypes that have so far been considered redundant.

Tubulin PTMs

Tubulin is subject to a large range of PTMs (Fig. 1), from well-known ones, such as acetylation or phosphorylation, to others that have so far mostly been found on tubulin. Detyrosination/tyrosination, polyglutamylation, and polyglycylation, for instance, might have evolved to specifically regulate tubulin and microtubule functions, in particular in cilia and flagella, as their evolution is closely linked to these organelles. The strong link between those modifications and tubulin evolution has led to the perception that they are tubulin PTMs; however, apart from detyrosination/tyrosination, most of them have other substrates (Regnard et al., 2000; Xie et al., 2007; van Dijk et al., 2008; Rogowski et al., 2009).

Tubulin acetylation.

Tubulin acetylation was discovered on lysine 40 (K40; Fig. 1 A) of flagellar α-tubulin in Chlamydomonas reinhardtii (L’Hernault and Rosenbaum, 1985) and is generally enriched on stable microtubules in cells. Considering that K40 acetylation per se has no effect on the ultrastructure of microtubules (Howes et al., 2014), it is rather unlikely that it directly stabilizes microtubules. As a result of its localization at the inner face of microtubules (Soppina et al., 2012), K40 acetylation might rather affect the binding of microtubule inner proteins, a poorly characterized family of proteins (Nicastro et al., 2011; Linck et al., 2014). Functional experiments in cells have further suggested that K40 acetylation regulates intracellular transport by regulating the traffic of kinesin motors (Reed et al., 2006; Dompierre et al., 2007). These observations could so far not be confirmed by biophysical measurements in vitro (Walter et al., 2012; Kaul et al., 2014), suggesting that in cells, K40 acetylation might affect intracellular traffic by indirect mechanisms.Enzymes involved in K40 acetylation are HDAC6 (histone deacetylase family member 6; Hubbert et al., 2002) and Sirt2 (sirtuin type 2; North et al., 2003). Initial functional studies used overexpression, depletion, or chemical inhibition of these enzymes. These studies should be discussed with care, as both HDAC6 and Sirt2 deacetylate other substrates and have deacetylase-independent functions and chemical inhibition of HDAC6 is not entirely selective for this enzyme (Valenzuela-Fernández et al., 2008). In contrast, acetyl transferase α-Tat1 (or Mec-17; Akella et al., 2010; Shida et al., 2010) specifically acetylates α-tubulin K40 (Fig. 3), thus providing a more specific tool to investigate the functions of K40 acetylation. Knockout mice of α-Tat1 are completely void of K40-acetylated tubulin; however, they show only slight phenotypic aberrations, for instance, in their sperm flagellum (Kalebic et al., 2013). A more detailed analysis of α-Tat1 knockout mice demonstrated that absence of K40 acetylation leads to reduced contact inhibition in proliferating cells (Aguilar et al., 2014). In migrating cells, α-Tat1 is targeted to microtubules at the leading edge by clathrin-coated pits, resulting in locally restricted acetylation of those microtubules (Montagnac et al., 2013). A recent structural study of α-Tat1 demonstrated that the low catalytic rate of this enzyme, together with its localization inside the microtubules, caused acetylation to accumulate selectively in stable, long-lived microtubules (Szyk et al., 2014), thus explaining the link between this PTM and stable microtubules in cells. However, the direct cellular function of K40 acetylation on microtubules is still unclear.Open in a separate windowFigure 3.Enzymes involved in PTM of tubulin. Schematic representation of known enzymes (mammalian enzymes are shown) involved in the generation and removal of PTMs shown in Fig. 1. Note that some enzymes still remain unknown, and some modifications are irreversible. (*CCP5 preferentially removes branching points [Rogowski et al., 2010]; however, the enzyme can also hydrolyze linear glutamate chains [Berezniuk et al., 2013]).Recent discoveries have brought up the possibility that tubulin could be subject to multiple acetylation events. A whole-acetylome study identified >10 novel sites on α- and β-tubulin (Choudhary et al., 2009); however, none of these sites have been confirmed. Another acetylation event has been described at lysine 252 (K252) of β-tubulin. This modification is catalyzed by the acetyltransferase San (Fig. 3) and might regulate the assembly efficiency of microtubules as a result of its localization at the polymerization interface (Chu et al., 2011).

Tubulin detyrosination.

Most α-tubulin genes in different species encode a C-terminal tyrosine residue (Fig. 2; Valenzuela et al., 1981). This tyrosine can be enzymatically removed (Hallak et al., 1977) and religated (Fig. 3; Arce et al., 1975). Mapping of tyrosinated and detyrosinated microtubules in cells using specific antibodies (Gundersen et al., 1984; Geuens et al., 1986; Cambray-Deakin and Burgoyne, 1987a) revealed that subsets of interphase and mitotic spindle microtubules are detyrosinated (Gundersen and Bulinski, 1986). As detyrosination was mostly found on stable and long-lived microtubules, especially in neurons (Cambray-Deakin and Burgoyne, 1987b; Robson and Burgoyne, 1989; Brown et al., 1993), it was assumed that this modification promotes microtubule stability (Gundersen et al., 1987; Sherwin et al., 1987). Although a direct stabilization of the microtubule lattice was considered unlikely (Khawaja et al., 1988), it was found more recently that detyrosination protects cellular microtubules from the depolymerizing activity of kinesin-13–type motor proteins, such as KIF2 or MCAK, thus increasing their longevity (Peris et al., 2009; Sirajuddin et al., 2014).Besides kinesin-13 motors, plus end–tracking proteins with cytoskeleton-associated protein glycine-rich (CAP-Gly) domains, such as CLIP170 or p150/glued, specifically interact with tyrosinated microtubules (Peris et al., 2006; Bieling et al., 2008) via this domain (Honnappa et al., 2006). In contrast, kinesin-1 moves preferentially on detyrosinated microtubules tracks in cells (Liao and Gundersen, 1998; Kreitzer et al., 1999; Konishi and Setou, 2009). The effect of detyrosination on kinesin-1 motor behavior was recently measured in vitro, and a small but significant increase in the landing rate and processivity of the motor has been found (Kaul et al., 2014). Such subtle changes in the motor behavior could, in conjunction with other factors, such as regulatory MAPs associated with cargo transport complexes (Barlan et al., 2013), lead to a preferential use of detyrosinated microtubules by kinesin-1 in cells.Despite the early biochemical characterization of a detyrosinating activity, the carboxypeptidase catalyzing detyrosination of α-tubulin has yet to be identified (Hallak et al., 1977; Argaraña et al., 1978, 1980). In contrast, the reverse enzyme, tubulin tyrosine ligase (TTL; Fig. 3; Raybin and Flavin, 1975; Deanin and Gordon, 1976; Argaraña et al., 1980), has been purified (Schröder et al., 1985) and cloned (Ersfeld et al., 1993). TTL modifies nonpolymerized tubulin dimers exclusively. This selectivity is determined by the binding interface between the TTL and tubulin dimers (Szyk et al., 2011, 2013; Prota et al., 2013). In contrast, the so far unidentified detyrosinase acts preferentially on polymerized microtubules (Kumar and Flavin, 1981; Arce and Barra, 1983), thus modifying a select population of microtubules within cells (Gundersen et al., 1987).In most organisms, only one unique gene for TTL exists. Consequently, TTL knockout mice show a huge accumulation of detyrosinated and particularly Δ2-tubulin (see next section). TTL knockout mice die before birth (Erck et al., 2005) with major developmental defects in the nervous system that might be related to aberrant neuronal differentiation (Marcos et al., 2009). TTL is strictly tubulin specific (Prota et al., 2013), indicating that all observed defects in TTL knockout mice are directly related to the deregulation of the microtubule cytoskeleton.

Δ2-tubulin and further C-terminal modification.

A biochemical study of brain tubulin revealed that ∼35% of α-tubulin cannot be retyrosinated (Paturle et al., 1989) because of the lack of the penultimate C-terminal glutamate residue of the primary protein sequence (Fig. 2; Paturle-Lafanechère et al., 1991). This so-called Δ2-tubulin (for two C-terminal amino acids missing) cannot undergo retyrosination as a result of structural constraints within TTL (Prota et al., 2013) and thus is considered an irreversible PTM.Δ2-tubulin accumulates in long-lived microtubules of differentiated neurons, axonemes of cilia and flagella, and also in cellular microtubules that have been artificially stabilized, for instance, with taxol (Paturle-Lafanechère et al., 1994). The generation of Δ2-tubulin requires previous detyrosination of α-tubulin; thus, the levels of this PTM are indirectly regulated by the detyrosination/retyrosination cycle. This mechanistic link is particularly apparent in the TTL knockout mice, which show massive accumulation of Δ2-tubulin in all tested tissues (Erck et al., 2005). Loss of TTL and the subsequent increase of Δ2-tubulin levels were also linked to tumor growth and might contribute to the aggressiveness of the tumors by an as-yet-unknown mechanism (Lafanechère et al., 1998; Mialhe et al., 2001). To date, no specific biochemical role of Δ2-tubulin has been determined; thus, one possibility is that the modification simply locks tubulin in the detyrosinated state.The enzymes responsible for Δ2-tubulin generation are members of a family of cytosolic carboxypeptidases (CCPs; Fig. 3; Kalinina et al., 2007; Rodriguez de la Vega et al., 2007), and most of them also remove polyglutamylation from tubulin (see next section; Rogowski et al., 2010). These enzymes are also able to generate Δ3-tubulin (Fig. 1 A; Berezniuk et al., 2012), indicating that further degradation of the tubulin C-terminal tails are possible; however, the functional significance of this event is unknown.

Polyglutamylation.

Polyglutamylation is a PTM that occurs when secondary glutamate side chains are formed on γ-carboxyl groups of glutamate residues in a protein (Fig. 1, A and B). The modification was first discovered on α- and β-tubulin from the brain (Eddé et al., 1990; Alexander et al., 1991; Rüdiger et al., 1992; Mary et al., 1994) as well as on axonemal tubulin from different species (Mary et al., 1996, 1997); however, it is not restricted to tubulin (Regnard et al., 2000; van Dijk et al., 2008). Using a glutamylation-specific antibody, GT335 (Wolff et al., 1992), it was observed that tubulin glutamylation increases during neuronal differentiation (Audebert et al., 1993, 1994) and that axonemes of cilia and flagella (Fouquet et al., 1994), as well as centrioles of mammalian centrosomes (Bobinnec et al., 1998), are extensively glutamylated.Enzymes catalyzing polyglutamylation belong to the TTL-like (TTLL) family (Regnard et al., 2003; Janke et al., 2005). In mammals, nine glutamylases exist, each of them showing intrinsic preferences for modifying either α- or β-tubulin as well as for initiating or elongating glutamate chains (Fig. 3; van Dijk et al., 2007). Two of the six well-characterized TTLL glutamylases also modify nontubulin substrates (van Dijk et al., 2008).Knockout or depletion of glutamylating enzymes in different model organisms revealed an evolutionarily conserved role of glutamylation in cilia and flagella. In motile cilia, glutamylation regulates beating behavior (Janke et al., 2005; Pathak et al., 2007; Ikegami et al., 2010) via the regulation of flagellar dynein motors (Kubo et al., 2010; Suryavanshi et al., 2010). Despite the expression of multiple glutamylases in ciliated cells and tissues, depletion or knockout of single enzymes often lead to ciliary defects, particularly in motile cilia (Ikegami et al., 2010; Vogel et al., 2010; Bosch Grau et al., 2013; Lee et al., 2013), suggesting essential and nonredundant regulatory functions of these enzymes in cilia.Despite the enrichment of polyglutamylation in neuronal microtubules (Audebert et al., 1993, 1994), knockout of TTLL1, the major polyglutamylase in brain (Janke et al., 2005), did not show obvious neuronal defects in mice (Ikegami et al., 2010; Vogel et al., 2010). This suggests a tolerance of neuronal microtubules to variations in polyglutamylation.Deglutamylases, the enzymes that reverse polyglutamylation, were identified within a novel family of CCPs (Kimura et al., 2010; Rogowski et al., 2010). So far, three out of six mammalian CCPs have been shown to cleave C-terminal glutamate residues, thus catalyzing both the reversal of polyglutamylation and the removal of gene-encoded glutamates from the C termini of proteins (Fig. 3). The hydrolysis of gene-encoded glutamate residues is not restricted to tubulin, in which it generates Δ2- and Δ3-tubulin, but has also been reported for other proteins such as myosin light chain kinase (Rusconi et al., 1997; Rogowski et al., 2010). One enzyme of the CCP family, CCP5, preferentially removes branching points generated by glutamylation, thus allowing the complete reversal of the polyglutamylation modification (Kimura et al., 2010; Rogowski et al., 2010). However, CCP5 can also hydrolyze C-terminal glutamate residues from linear peptide chains similar to other members of the CCP family (Berezniuk et al., 2013).CCP1 is mutated in a well-established mouse model for neurodegeneration, the pcd (Purkinje cell degeneration) mouse (Mullen et al., 1976; Greer and Shepherd, 1982; Fernandez-Gonzalez et al., 2002). The absence of a key deglutamylase leads to strong hyperglutamylation in brain regions that undergo degeneration, such as the cerebellum and the olfactory bulb (Rogowski et al., 2010). When glutamylation levels were rebalanced by depletion or knockout of the major brain polyglutamylase TTLL1 (Rogowski et al., 2010; Berezniuk et al., 2012), Purkinje cells survived. Although the molecular mechanisms of hyperglutamylation-induced degeneration remain to be elucidated, perturbation of neuronal transport, as well as changes in the dynamics and stability of microtubules, is expected to be induced by hyperglutamylation. Increased polyglutamylation levels have been shown to affect kinesin-1–mediated transport in cultured neurons (Maas et al., 2009), and the turnover of microtubules can also be regulated by polyglutamylation via the activation of microtubule-severing enzymes such as spastin (Lacroix et al., 2010).Subtle differences in polyglutamylation can be seen on diverse microtubules in different cell types. The functions of these modifications remain to be studied; however, its wide distribution strengthens the idea that it could be involved in fine-tuning a range of microtubule functions.

Polyglycylation.

Tubulin polyglycylation or glycylation, like polyglutamylation, generates side chains of glycine residues within the C-terminal tails of α- and β-tubulin (Fig. 1, A and B). The modification sites of glycylation are considered to be principally the same as for glutamylation, and indeed, both PTMs have been shown to be interdependent in cells (Rogowski et al., 2009; Wloga et al., 2009). Initially discovered on Paramecium tetraurelia tubulin (Redeker et al., 1994), glycylation has been extensively studied using two antibodies, TAP952 and AXO49 (Bressac et al., 1995; Levilliers et al., 1995; Bré et al., 1996). In contrast to polyglutamylation, glycylation is restricted to cilia and flagella in most organisms analyzed so far.Glycylating enzymes are also members of the TTLL family, and homologues of these enzymes have so far been found in all organisms with proven glycylation of ciliary axonemes (Rogowski et al., 2009; Wloga et al., 2009). In mammals, initiating (TTLL3 and TTLL8) and elongating (TTLL10) glycylases work together to generate polyglycylation (Fig. 3). In contrast, the two TTLL3 orthologues from Drosophila melanogaster can both initiate and elongate glycine side chains (Rogowski et al., 2009).In mice, motile ependymal cilia in brain ventricles acquire monoglycylation upon maturation, whereas polyglycylation is observed only after several weeks (Bosch Grau et al., 2013). Sperm flagella, in contrast, acquire long glycine chains much faster, suggesting that the extent of polyglycylation could correlate with the length of the axonemes (Rogowski et al., 2009). Depletion of glycylases in mice (ependymal cilia; Bosch Grau et al., 2013), zebrafish (Wloga et al., 2009; Pathak et al., 2011), Tetrahymena thermophila (Wloga et al., 2009), and D. melanogaster (Rogowski et al., 2009) consistently led to ciliary disassembly or severe ciliary defects. How glycylation regulates microtubule functions remains unknown; however, the observation that glycylation-depleted axonemes disassemble after initial assembly (Rogowski et al., 2009; Bosch Grau et al., 2013) suggests a role of this PTM in stabilizing axonemal microtubules. Strikingly, human TTLL10 is enzymatically inactive; thus, humans have lost the ability to elongate glycine side chains (Rogowski et al., 2009). This suggests that the elongation of the glycine side chains is not an essential aspect of the function of this otherwise critical tubulin PTM.

Other tubulin PTMs.

Several other PTMs have been found on tubulin. Early studies identified tubulin phosphorylation (Eipper, 1974; Gard and Kirschner, 1985; Díaz-Nido et al., 1990); however, no specific functions were found. The perhaps best-studied phosphorylation event on tubulin takes place at serine S172 of β-tubulin (Fig. 1 A), is catalyzed by the Cdk1 (Fig. 3), and might regulate microtubule dynamics during cell division (Fourest-Lieuvin et al., 2006; Caudron et al., 2010). Tubulin can be also modified by the spleen tyrosine kinase Syk (Fig. 3; Peters et al., 1996), which might play a role in immune cells (Faruki et al., 2000; Sulimenko et al., 2006) and cell division (Zyss et al., 2005; Sulimenko et al., 2006).Polyamination has recently been discovered on brain tubulin (Song et al., 2013), after having been overlooked for many years as a result of the low solubility of polyaminated tubulin. Among several glutamine residues of α- and β-tubulin that can be polyaminated, Q15 of β-tubulin is considered the primary modification site (Fig. 1 A). Polyamination is catalyzed by transglutaminases (Fig. 3), which modify free tubulin as well as microtubules in an irreversible manner, and most likely contribute to the stabilization of microtubules (Song et al., 2013).Tubulin was also reported to be palmitoylated (Caron, 1997; Ozols and Caron, 1997; Caron et al., 2001), ubiquitinated (Ren et al., 2003; Huang et al., 2009; Xu et al., 2010), glycosylated (Walgren et al., 2003; Ji et al., 2011), arginylated (Wong et al., 2007), methylated (Xiao et al., 2010), and sumoylated (Rosas-Acosta et al., 2005). These PTMs have mostly been reported without follow-up studies, and some of them are only found in specific cell types or organisms and/or under specific metabolic conditions. Further studies will be necessary to gain insights into their potential roles for the regulation of the microtubule cytoskeleton.

Current advances and future perspectives

The molecular heterogeneity of microtubules, generated by the expression of different tubulin isotypes and by the PTM of tubulin has fascinated the scientific community for ∼40 years. Although many important advances have been made in the past decade, the dissection of the molecular mechanisms and a comprehensive understanding of the biological functions of tubulin isotypes and PTMs will be a challenging field of research in the near future.

Direct measurements of the impact of tubulin heterogeneity.

The most direct and reliable type of experiments to determine the impact of tubulin heterogeneity on microtubule behavior are in vitro measurements with purified proteins. However, most biophysical work on microtubules has been performed with tubulin purified from bovine, ovine, or porcine brains, which can be obtained in large quantities and with a high degree of purity and activity (Vallee, 1986; Castoldi and Popov, 2003). Brain tubulin is a mixture of different tubulin isotypes and is heavily posttranslationally modified and thus inept for investigating the functions of tubulin heterogeneity (Denoulet et al., 1986; Cambray-Deakin and Burgoyne, 1987b; Paturle et al., 1989; Eddé et al., 1990). Thus, pure, recombinant tubulin will be essential to dissect the roles of different tubulin isoforms and PTMs.Attempts to produce recombinant, functional α- and β-tubulin in bacteria have failed so far (Yaffe et al., 1988), most likely because of the absence of the extensive tubulin-specific folding machinery (Yaffe et al., 1992; Gao et al., 1993; Tian et al., 1996; Vainberg et al., 1998) in prokaryotes. An alternative source of tubulin with less isotype heterogeneity and with almost no PTMs is endogenous tubulin from cell lines such as HeLa, which in the past has been purified using a range of biochemical procedures (Bulinski and Borisy, 1979; Weatherbee et al., 1980; Farrell, 1982; Newton et al., 2002; Fourest-Lieuvin, 2006). Such tubulin can be further modified with tubulin-modifying enzymes, such as polyglutamylases, either by expressing those enzymes in the cells before tubulin purification (Lacroix and Janke, 2011) or in vitro with purified enzymes (Vemu et al., 2014). Despite some technical limitations of these methods, HeLa tubulin modified in cells has been successfully used in an in vitro study on the role of polyglutamylation in microtubule severing (Lacroix et al., 2010).Naturally occurring variants of tubulin isotypes and PTMs can be purified from different organisms, organs, or cell types, but obviously, only some combinations of tubulin isotypes and PTMs can be obtained by this approach. The recent development of an affinity purification method using the microtubule-binding TOG (tumor overexpressed gene) domain of yeast Stu2p has brought a new twist to this approach, as it allows purifying small amounts of tubulin from any cell type or tissue (Widlund et al., 2012).The absence of tubulin heterogeneity in yeast has made budding and fission yeast potential expression systems for recombinant, PTM-free tubulin (Katsuki et al., 2009; Drummond et al., 2011; Johnson et al., 2011). However, the expression of mammalian tubulin in this system has remained impossible. This problem was then partially circumvented by expressing tubulin chimeras that consist of a yeast tubulin body fused to mammalian C-terminal tubulin tails, thus mimicking different tubulin isotypes (Sirajuddin et al., 2014). Moreover, detyrosination can be generated by deleting the key C-terminal residue from endogenous or chimeric α-tubulin (Badin-Larçon et al., 2004), and polyglutamylation is generated by chemically coupling glutamate side chains to specifically engineered tubulin chimeras (Sirajuddin et al., 2014). These approaches allowed the first direct measurements of the impact of tubulin isotypes and PTMs on the behavior of molecular motors in vitro (Sirajuddin et al., 2014) and the analysis of the effects of tubulin heterogeneity on microtubule behavior and interactions inside the yeast cell (Badin-Larçon et al., 2004; Aiken et al., 2014).Currently, the most promising development has been the successful purification of fully functional recombinant tubulin from the baculovirus expression system (Minoura et al., 2013). Using this system, defined α/β-tubulin dimers can be obtained using two different epitope tags on α- and β-tubulin, respectively. Although these epitope tags are essential for separating recombinant from the endogenous tubulin, they could also affect tubulin assembly or microtubule–MAP interactions. Thus, future developments should focus on eliminating these tags.Current efforts have brought the possibility of producing recombinant tubulin into reach. Further improvement and standardization of these methods will certainly provide a breakthrough in understanding the mechanisms by which tubulin heterogeneity contributes to microtubule functions.

Complexity of tubulin—understanding the regulatory principles.

The diversity of tubulin genes (isotypes) and the complexity of tubulin PTMs have led to the proposal of the term “tubulin code” (Verhey and Gaertig, 2007; Wehenkel and Janke, 2014), in analogy to the previously coined histone code (Jenuwein and Allis, 2001). Tubulin molecules consist of a highly structured and thus evolutionarily conserved tubulin body and the unstructured and less conserved C-terminal tails (Nogales et al., 1998). As PTMs and sequence variations within the tubulin body are expected to affect the conserved tubulin fold and therefore the properties of the microtubule lattice, they are not likely to be involved in generating the tubulin code. In contrast, modulations of the C-terminal tails could encode signals on the microtubule surface without perturbing basic microtubule functions and properties (Figs. 1 A and ​and4).4). Indeed, the highest degree of gene-encoded diversity (Fig. 2) and the highest density and complexity of PTMs (Fig. 1) are found within these tail domains.Open in a separate windowFigure 4.Molecular components of the tubulin code. Schematic representation of potential coding elements that could generate specific signals for the tubulin code. (A) The length of the C-terminal tails of different tubulin isotypes differ significantly (Fig. 2) and could have an impact on the interactions between microtubules and MAPs. (B) Tubulin C-terminal tails are rich in charged amino acid residues. The distribution of these residues and local densities of charges could influence the electrostatic interactions with the tails and the readers. (C) Although each glutamate residue within the C-terminal tails could be considered a potential modification site, only some sites have been found highly occupied in tubulin purifications from native sources. This indicates selectivity of the modification reactions, which can participate in the generation of specific modification patterns (see D). Modification sites might be distinguished by their neighboring amino acid residues, which could create specific modification epitopes. (D) As a result of the large number of modification sites and the variability of side chains, a large variety of modification patterns could be generated within a single C-terminal tail of tubulin. (E) Modification patterns as shown in D can be distinct between α- and β-tubulin. These modification patterns could be differentially distributed at the surface of the microtubule lattice, thus generating a higher-order patterning. Tub, tubulin. For color coding, see Fig. 2.Considering the number of tubulin isotypes plus all potential combinations of PTMs (e.g., each glutamate residue within the C-terminal tubulin tail could be modified by either polyglutamylation or polyglycylation, each of them generating side chains of different lengths; Fig. 4), the number of distinct signals generated by the potential tubulin code would be huge. However, as many of these potential signals represent chemical structures that are similar and might not be reliably distinguished by readout mechanisms, it is possible that the tubulin code generates probabilistic signals. In this scenario, biochemically similar modifications would have similar functional readouts, and marginal differences between those signals would only bias biological processes but not determine them. This stands in contrast to the concept of the histone code, in which precise patterns of different PTMs on the histone proteins encode distinct biological signals.The concept of probabilistic signaling is already inscribed in the machinery that generates the tubulin code. Polyglutamylases and polyglycylases from the TTLL family have preferential activities for either α- or β-tubulin and for generating different lengths of the branched glutamate or glycine chains. Although under conditions of low enzyme concentrations, as found in most cells and tissues, the enzymes seem to selectively generate their preferential type of PTM, higher enzyme concentrations induce a more promiscuous behavior, leading, for instance, to a loss of selectivity for α- or β-tubulin (van Dijk et al., 2007). Similarly, the modifying enzymes might prefer certain modification sites within the C-terminal tails of tubulin but might be equally able to modify other sites, which could be locally regulated in cells. For example, β-tubulin isotypes isolated from mammalian brain were initially found to be glutamylated on single residues (Alexander et al., 1991; Rüdiger et al., 1992), which in the light of the comparably low sensitivity of mass spectrometry at the time might rather indicate a preferential than a unique modification of these sites. Nevertheless, the neuron-specific polyglutamylase for β-tubulin TTLL7 (Ikegami et al., 2006) can incorporate glutamate onto many more modification sites of β-tubulin in vitro (Mukai et al., 2009), which clearly indicates that not all of the possible modification events take place under physiological conditions.Several examples supporting a probabilistic signaling mode of the tubulin code are found in the recent literature. In T. thermophila, a ciliate without tubulin isotype diversity (Gaertig et al., 1993) but with a huge repertoire of tubulin PTMs and tubulin-modifying enzymes (Janke et al., 2005), tubulin can be easily mutagenized to experimentally eliminate sites for PTMs. Mutagenesis of the most commonly occupied glutamylation/glycylation sites within the β-tubulin tails did not generate a clear decrease of glycylation levels nor did it cause obvious phenotypic alterations. This indicates that the modifying enzymes can deviate toward alternative modification sites and that similar PTMs on different sites can compensate the functions of the mutated site. However, when all of the key modification sites were mutated, glycylation became prominently decreased, which led to severe phenotypes, including lethality (Xia et al., 2000). Most strikingly, these phenotypes could be recovered by replacing the C-terminal tail of α-tubulin with the nonmutated β-tubulin tail. This α–β-tubulin chimera became overglycylated and functionally compensated for the absence of modification sites on β-tubulin. The conclusion of this study is that PTM- and isotype-generated signals can fulfill a biological function within a certain range of tolerance.But how efficient is such compensation? The answer can be found in a variety of already described deletion mutants for tubulin-modifying enzymes in different model organisms. Most single-gene knockouts for TTLL genes (glutamylases or glycylases) did not result in prominent phenotypic alterations in mice, even for enzymes that are ubiquitously expressed. Only some highly specialized microtubule structures show functional aberrations upon the deletion of a single enzyme. These “tips of the iceberg” are usually the motile cilia and sperm flagella, which carry very high levels of polyglutamylation and polyglycylation (Bré et al., 1996; Kann et al., 1998; Rogowski et al., 2009). It thus appears that some microtubules are essentially dependent on the generation of specific PTM patterns, whereas others can tolerate changes and appear to function normally. How “normal” these functions are remains to be investigated in future studies. It is possible that defects are subtle and thus overlooked but could become functionally important under specific conditions.A tubulin code also requires readout mechanisms. The most likely “readers” of the tubulin code are MAPs and molecular motors. Considering the probabilistic signaling hypothesis, the expected effects of the signals would be in most cases rather gradual changes, for instance, to fine-tune molecular motor traffic and/or to bias motors toward defined microtubule tracks but not to obliterate motor activity or MAP binding to microtubules. An in vitro study using recombinant tubulin chimeras purified from yeast confirmed this notion (Sirajuddin et al., 2014). By analyzing which elements of the tubulin code can regulate the velocity and processivity of the molecular motors kinesin and dynein, these researchers found that the C-terminal tails of α- and β-tubulin differentially influence the kinetic parameters of the tested motors; however, the modulation was rather modest. One of their striking observations was that a single lysine residue, present in the C-terminal tails of two β-tubulin isotypes (Figs. 2 and ​and4),4), significantly affected motor traffic and that this effect can be counterbalanced by polyglutamylation. These observations are the first in vitro evidence for the interdependence of different elements of the tubulin code and provide another indication for its probabilistic mode of signaling.

Future directions.

One of the greatest technological challenges to understanding the function of the tubulin code is to detect and interpret subtle and complex regulatory events generated by this code. It will thus be instrumental to further develop tools to better distinguish graded changes in PTM levels on microtubules in cells and tissues (Magiera and Janke, 2013) and to reliably measure subtle modulations of microtubule behavior in reconstituted systems.The current advances in the field and especially the availability of whole-organism models, as well as first insights into the pathological role of tubulin mutations (Tischfield et al., 2011), are about to transform our way of thinking about the regulation of microtubule cytoskeleton. Tubulin heterogeneity generates complex probabilistic signals that cannot be clearly attributed to single biological functions in most cases and that are not essential for most cellular processes. Nevertheless, it has been conserved throughout evolution of eukaryotes and can hardly be dismissed as not important. To understand the functional implications of these processes, we might be forced to reconsider how we define biologically important events and how we measure events that might encode probabilistic signals. The answers to these questions could provide novel insights into how complex systems, such as cells and organisms, are sustained throughout difficult and challenging life cycles, resist to environmental stress and diseases, and have the flexibility needed to succeed in evolution.     

Capture and delivery of tail-anchored proteins to the endoplasmic reticulum

Tail-anchored (TA) proteins fulfill diverse cellular functions within different organellar membranes. Their characteristic C-terminal transmembrane segment renders TA proteins inherently prone to aggregation and necessitates their posttranslational targeting. The guided entry of TA proteins (GET in yeast)/transmembrane recognition complex (TRC in humans) pathway represents a major route for TA proteins to the endoplasmic reticulum (ER). Here, we review important new insights into the capture of nascent TA proteins at the ribosome by the GET pathway pretargeting complex and the mechanism of their delivery into the ER membrane by the GET receptor insertase. Interestingly, several alternative routes by which TA proteins can be targeted to the ER have emerged, raising intriguing questions about how selectivity is achieved during TA protein capture. Furthermore, mistargeting of TA proteins is a fundamental cellular problem, and we discuss the recently discovered quality control machineries in the ER and outer mitochondrial membrane for displacing mislocalized TA proteins.

IntroductionThe protein components of biological membranes expand their functionality beyond physical barriers by acting as gateways, allowing intercompartment communication as well as facilitating transport and other membrane-associated processes. Membrane proteins collectively constitute ∼30% of the proteome of most organisms (Krogh et al., 2001), and their biogenesis represents a major challenge for cells. Their hydrophobic transmembrane domains (TMDs), essential for integration into the lipid bilayer and functionality, render such proteins inherently prone to aggregation in the aqueous cytosolic environment. Dedicated targeting strategies for chaperoning such proteins to their target membranes are therefore necessary. Proteins destined for the ER that carry short signal sequences at their N-terminal end and/or internal TMDs are typically recognized cotranslationally by the signal recognition particle (SRP). This arrests translation and induces relocalization of the ribosome nascent chain complex (RNC) to the ER-bound Sec61 translocon, where the newly synthesized protein is channeled directly into the ER lumen and/or membrane (reviewed in Akopian et al., 2013; Rapoport et al., 2017). In both yeast and mammals, a macromolecular ER membrane protein complex (EMC) cooperates with the translocon by assisting the cotranslational folding and biogenesis of polytopic membrane proteins in the ER as well as itself acting as a membrane insertase, mediating the correct topological insertion of the first TMD of specific G-coupled receptors (Bai et al., 2020; Chitwood et al., 2018; Shurtleff et al., 2018). Furthermore, an SRP-independent ER targeting pathway (SND) has recently been revealed in yeast, where proteins containing TMDs in their central regions are captured by Snd1 and directed toward a Sec61 translocon associated with Snd2 and Snd3 (Aviram et al., 2016). This pathway appears to have a broad client spectrum and has been suggested to functionally compensate when other ER targeting pathways are impaired.Tail-anchored (TA) proteins represent a specific class of membrane proteins characterized by a single TMD close to their C-terminus (reviewed in Kutay et al., 1993). The TA protein family contains >50 members in yeast (Beilharz et al., 2003), >500 in plants (Kriechbaumer et al., 2009), and >300 in humans (Kalbfleisch et al., 2007), which populate different membranes (ER, Golgi, and inner nuclear, outer mitochondrial, and peroxisomal membranes). These proteins fulfill diverse membrane-associated functions ranging from regulating intracellular vesicular trafficking (SNARE proteins) to apoptosis, autophagy, lipid biosynthesis, and protein degradation. The topology of TA proteins dictates their posttranslational targeting, as translation termination occurs concurrent with emergence of the TMD from the polypeptide exit tunnel of the ribosome. A major route to the ER for TA proteins is the evolutionarily conserved guided entry of TA proteins (GET) pathway in yeast and the homologous transmembrane recognition complex (TRC) pathway in mammals (Figs. 1 and ​and2;2; Borgese et al., 2019; Schuldiner et al., 2008; Stefanovic and Hegde, 2007). The established view of the GET/TRC pathway (reviewed in Borgese et al., 2019; Chio et al., 2017) involves the posttranslational capture of TA proteins by a pretargeting complex composed of the homodimeric, cytosolic chaperone Sgt2 (yeast)/SGTA (mammals) and the Get4-Get5 heterodimer (yeast)/TRC35-UBL4A-BAG6 complex (mammals; Jonikas et al., 2009; Mariappan et al., 2011; Wang et al., 2010). Interaction of the pretargeting complex with an ATP-bound form of the ATPase Get3 (yeast)/TRC40 (mammals) results in transfer of the TA protein from Sgt2/SGTA to Get3/TRC40 such that the TMD is shielded within a hydrophobic pocket of Get3/TRC40 (Bozkurt et al., 2009; Mateja et al., 2009; Stefanovic and Hegde, 2007; Suloway et al., 2009). ATP hydrolysis triggered by interaction of Get3/TRC40 with the TA protein, coupled with conformational changes in Get3/TRC40 induced by interaction with Get4-Get5/TRC35-UBL4A-BAG6, drives dissociation of TA protein-bound Get3/TRC40 from the pretargeting complex, allowing delivery to the ER-bound GET receptor composed of Get1 and Get2 (yeast)/tryptophan-rich basis protein (WRB) and calcium-modulating cyclophilin ligand (CAML; mammals; Mariappan et al., 2011; Mateja et al., 2015; Stefer et al., 2011). Receptor binding triggers ADP release and conformational rearrangement of Get3/TRC40, allowing TA protein insertion into the membrane and recycling of Get3/TRC40 (reviewed in Mateja and Keenan, 2018). Although much less is currently known about the GET pathway in plants, homologues of the GET pathway components have been identified or are predicted (Xing et al., 2017; Srivastava et al., 2017; Asseck et al., 2021), and two TA SNARE proteins have been shown to be affected by lack of Arabidopsis thaliana (At)GET3 (Xing et al., 2017). A growing body of evidence indicates functional redundancy between different pathways for targeting TA proteins to the ER (Figs. 1 and ​and2;2; Casson et al., 2017). Although some TA proteins of the secretory pathway fulfill essential cellular functions, deletion of GET pathway components is not lethal in yeast or plants, and for many TA proteins, lack of GET pathway components reduces but does not abolish ER targeting. Consistent with this notion, both the EMC insertase (Guna et al., 2018) and the SND pathway (Aviram et al., 2016) have been shown to support the ER targeting of TA proteins.Open in a separate windowFigure 1.TA protein targeting to the ER in yeast. Nascent TA proteins emerging from the ribosome can be captured by alternative ER-targeting machineries. A major route to the ER is via the GET pathway, involving ribosome-associated capture by Sgt2, followed by Get4/Get5-mediated handover to the Get3 ATPase and insertion into the ER membrane by a heterotetrameric GET receptor complex composed of Get1 and Get2.Open in a separate windowFigure 2.TA protein targeting to the ER in mammals. Mammalian TA proteins are predominantly targeted to the ER by the TRC pathway. After capture by SGTA, together with the BAG6 complex (BAG6, UBL4A, and TRC35), the TA protein is passed to the TRC40 chaperone for delivery to the ER-bound receptor complex formed by WRB and CAML. BAG6 has dual functions bridging ER targeting and ubiquitination of TA proteins and can be antagonized by SGTA. Ubiquitinated TA proteins can be deubiquitinated by ER-associated UPS20/UPS33.Despite a wealth of knowledge on some aspects of TA protein targeting to the ER by the GET/TRC pathway, other features and mechanistic details remain enigmatic. How TA proteins can be captured posttranslationally but also reliably avoid aggregation during handover from the ribosome to the pretargeting complex or other chaperones is an inherent conundrum of their biogenesis. Knowledge on the architecture of the GET/TRC receptor complex and mechanistic understanding of how the TA protein, delivered to the receptor by Get3/TRC40, is inserted into the ER membrane, have been limited by the technically challenging nature of structural analyses of membrane-bound complexes. Furthermore, how the fidelity of TA protein targeting to different membranes is ensured is poorly understood, and little is currently known about how the actions of TA protein targeting and quality control are coordinated. In this review, we describe recent advances that address these key aspects of the GET/TRC pathway and TA protein biogenesis.Capture of nascent TA proteinsIn contrast to cotranslational protein targeting, where capture of client proteins and their delivery to the ER-bound receptor complex is performed by the SRP complex, posttranslational targeting of TA proteins to the ER by the GET/TRC pathway is a more stepwise process, involving a dynamically assembling pretargeting complex that mediates initial capture but then hands over substrates to another chaperone for delivery to the membrane-bound receptor. The existence of a modular pretargeting complex appears to be a feature of the GET/TRC pathway conserved throughout eukaryotes, but compositional and structural differences between species are apparent. In yeast, Get4-Get5 form an obligate heterodimer, whereas the homologous TRC35 and UBL4A interact via an additional component BAG6 (Chang et al., 2010; Chartron et al., 2010; Mariappan et al., 2010; Mock et al., 2015). Biochemical evidence shows that a Get4/TRC35 homologue exists in plants, and homologues of Get5/UBL4A and Sgt2/SGTA are predicted from in silico analyses (Srivastava et al., 2017; Xing et al., 2017). A BAG6 homologue has also been identified in plants (https://www.arabidopsis.org/servlets/TairObject?id=35038&type=locus); however, it remains unknown if this protein associates with components of the GET pathway and/or contributes to TA protein targeting. These differences likely reflect subtle variations in the mechanism of TA protein capture between species and/or the greater need for regulation and surveillance in multicellular organisms.Ribosome binding of pretargeting complex componentsDue to the position of the TMD at the C-terminus of TA proteins, the GET/TRC pathway must capture clients posttranslationally. However, posttranslational capture has the inherent caveat of protein aggregation in the narrow window between emergence of the TMD from the ribosome exit tunnel and protein capture by a chaperoning factor. The first hint how this issue may be overcome came with the intriguing discovery of Get5 in a high-throughput screen for ribosome-associated proteins in yeast (Fleischer et al., 2006). This observation raised the possibility of a physical connection between the upstream components of the GET pathway and the translation machinery, despite the posttranslational nature of the final capture event. Detection of the Get4-Get5 heterodimer associated with polysomes importantly confirmed recruitment of these proteins to actively translating ribosomes (Zhang et al., 2016), further supporting that the GET pathway pretargeting complex might be poised on the ribosomes ready to shield nascent TA proteins directly as they emerge from the exit tunnel. In vitro reconstitution confirmed a high-affinity interaction between Get4-Get5 and ribosomes, and protein–protein and protein–RNA cross-linking analyses pinpointed the polypeptide exit tunnel of the ribosome as the Get4-Get5 binding site (Fig. 1; Zhang et al., 2021). Get4-Get5 bridge interactions between Sgt2 and Get3 to facilitate handover of the TA protein to the downstream chaperone, and therefore functional significance of Get4-Get5 ribosome binding must be coupled to TA protein capture by Sgt2. Indeed, it was recently shown that Get4-Get5 act as a binding platform for recruitment of Sgt2 to ribosomes and that the presence of Get4-Get5 on ribosomes enhances TA protein capture by Sgt2 (Zhang et al., 2021). Crystal structures of GET pathway subcomplexes demonstrate that the N-terminal domains of the Sgt2 homodimer interact with the central UBL domain of Get5, while the N-terminal region of Get5 mediates interaction with C-terminal region of Get4 (Chang et al., 2010; Chartron et al., 2012a; Simon et al., 2013). As Get5 appears to simultaneously interact with Get4, Sgt2, and ribosomes, it is tempting to speculate that ribosome binding of the pretargeting complex may be mediated by the C-terminal region of Get5. However, due to the multimeric nature of the pretargeting complex and evidence that the Get5 C-terminal region also mediates homodimerization (Chartron et al., 2012b), structural analyses of Get4-Get5-Sgt2–bound ribosomes will be necessary to resolve in detail the architecture of GET pathway pretargeting complex–bound ribosomes.The mammalian pretargeting complex components UBL4A, TRC35, BAG6, and SGTA also associate with RNC complexes (Fig. 2; Leznicki and High, 2020; Mariappan et al., 2010), implying that the mechanism of ribosome-associated capture of TA proteins is also employed in mammalian cells to circumvent protein aggregation upon the TMD encountering the aqueous cytosol during targeting. However, the mammalian-specific pretargeting complex component BAG6, rather than the Get5-homologous UBL4A, appears to act as the key ribosome adaptor, and SGTA can also interact with RNC complexes independently of TRC35-UBL4A (Leznicki and High, 2020; Mariappan et al., 2010).Notably, Get3/TRC40, which receive the TA protein from the pretargeting complex, are not ribosome associated (Mariappan et al., 2010; Zhang et al., 2016). This implies that following Get4-Get5–facilitated capture of the TA protein by ribosome-associated Sgt2, the pretargeting complex should dissociate from the ribosome to encounter Get3. It is possible that a conformational change in the pretargeting complex, induced by TA protein binding, triggers release from the ribosome.Recognition of ribosomes synthesizing TA proteinsThe discovery of ribosome-associated populations of the GET pretargeting complex and the BAG6 complex and SGTA in yeast and mammals, respectively, raises the question of how these complexes, which are significantly less abundant than cytosolic ribosomes, are able to identify ribosomes synthesizing TA proteins. Analogous to SRP, which probes the translating ribosome population, preferentially binding ribosomes translating proteins with a signal sequence (Berndt et al., 2009; Holtkamp et al., 2012; Ogg and Walter, 1995; Voorhees and Hegde, 2015), yeast Get4-Get5 show increased association with RNCs containing a TA/TMD in the exit tunnel (Zhang et al., 2021). Mammalian SGTA is likewise selectively recruited to ribosomes synthesizing membrane proteins (Leznicki and High, 2020). This implies that a similar substrate-scanning mechanism, involving transient association events followed by high-affinity docking only onto appropriate ribosomes, is employed in both the co- and posttargeting pathways. The underlying mechanism of how the GET/TRC pretargeting components sense the presence of the TA/TMD in the exit tunnel still remains to be elucidated. Intriguingly, the newly identified ribosome-binding site of the GET pathway pretargeting complex overlaps with that of SRP, implying mutually exclusive ribosome occupancy. Consistent with this, the presence of Get4-Get5 on RNCs with a TMD in the exit tunnel reduces the amount of SRP bound, and strong binding of SRP to an exposed TMD leads to displacement of Get4-Get5 (Zhang et al., 2021). This suggests a compelling model where the GET/TRC pretargeting complex, SRP, and potentially ribosome-bound Snd1 (Fleischer et al., 2006) constantly screen the translating pool of ribosomes and outcompete each other for ribosome binding upon encountering one synthesizing an optimal substrate.Additional players in initial TA protein captureIt is well established both in vitro and in vivo that TA protein targeting by the GET/TRC pathways involves a chaperone exchange in which the TMD, initially shielded by a hydrophobic patch within the C-terminal region of Sgt2, is handed over to Get3, where it is protected within a dedicated hydrophobic groove formed in the ATP-bound state. While Sgt2 has long been considered the upstream component of the GET pathway, it has recently emerged that the abundant, Hsp70-like chaperone Ssa1 can also act as a highly efficient nascent TA protein chaperone and that its effectiveness in TA protein trapping is enhanced by the J domain–containing cochaperone protein Ydj1 (Fig. 1; Cho et al., 2021; Cho and Shan, 2018). Ssa1 and numerous other chaperones physically interact with Sgt2 via its tetratricopeptide repeat domain (Cho and Shan, 2018; Krysztofinska et al., 2017). As this domain of Sgt2 is important for ER targeting of TA proteins in vivo, this supports a potential role of other chaperones alongside Sgt2. Transfer of TA protein cargoes from Ssa1 to Sgt2 is energetically favorable and stimulated by the J domain proteins Ydj1 and Sis1. Interestingly, Ydj1 and Sis1 appear to function redundantly in targeting of a reporter TA protein to the ER (Cho et al., 2021), perhaps suggesting that the chaperone cascade protecting TA proteins during their biogenesis can be more extensive than initially anticipated. However, it still remains to be determined how much of a contribution these proteins make to endogenous TA protein targeting within the native cellular environment. Interestingly, human cells lacking SGTA or BAG6 are viable, and TA protein targeting to the ER can still be accomplished in the absence of these factors (Culver and Mariappan, 2021). It is possible, therefore, for capture by the pretargeting complex to be bypassed, and perhaps in this context, other chaperones, analogous to those characterized in yeast, contribute to the initial protection of nascent TA proteins before their association with TRC40. Along this line, TA proteins with less hydrophobic TMDs, which use the EMC rather than the TRC pathway, have been shown to be chaperoned through the cytosol by calmodulin (Fig. 2; Guna et al., 2018). The existence of alternative, partially redundant, targeting pathways, supported by the nonlethality of yeast, plant, and human GET pretargeting factor knockouts (Stefanovic and Hegde, 2007; Schuldiner et al., 2008; Jonikas et al., 2009; Srivastava et al., 2017; Xing et al., 2017; Culver and Mariappan, 2021), likely helps ensure high-fidelity and robust targeting of TA proteins in vivo.Selectivity of TA protein captureIt has become increasingly clear that nascent TA proteins emerge into a crowded environment where encounters with different machineries can direct them toward different fates, such as targeting to the ER via different routes, targeting to different organelles, or potentially, degradation. These observations suggest a finely tuned process of nascent TA protein capture to direct different proteins to the appropriate destinations and highlight the question of how TA proteins are selectively captured. While all TA proteins share the common characteristic of a single TMD close to the C-terminus that can serve as a targeting signal, the length and hydrophobicity of this TMD, as well as the net charge of the downstream sequence (also termed C-terminal element [CTE]), vary significantly, and these features are important determinants of the ultimate destination of the protein (Beilharz et al., 2003; Borgese et al., 2019). TA proteins of the outer mitochondrial membrane (OMM) and peroxisomes are typified by short, less hydrophobic TMDs, and positively charged CTEs are characteristic features of peroxisomal TA proteins (Horie et al., 2002). In contrast, ER and Golgi TA proteins generally have relatively long, hydrophobic TMDs, and the charge of their CTEs varies considerably (Rao et al., 2016; Borgese et al., 2019). The physiochemical properties of ER TMDs favor capture by Sgt2/SGTA and binding by Get3/TRC40, whereas TA proteins with less hydrophobic TMDs are poor substrates (Coy-Vergara et al., 2019; Guna et al., 2018). This implies that an important layer of capture selectivity is already encoded within the proteins themselves. Strict categorization of different organellar TA proteins based on physiochemical properties is not possible, however, as differently localized TA proteins have partially overlapping properties, indicating that this criterion alone is insufficient to ensure correct targeting.TA proteins not only need to be directed to different target membranes, but they also need to be recognized as bona fide membrane proteins. The hydrophobic sequences that are integral features of membrane proteins must be distinguished from exposed hydrophobic patches of misfolded, nonmembrane proteins that serve as signals for recruitment of the protein quality control machinery. Interestingly, the BAG6 component of the mammalian pretargeting complex sits at the nexus between the alternative fates of targeting and degradation; a minimal C-terminal BAG domain in BAG6 scaffolds interactions between TRC35 and UBL4A and is sufficient for TA protein targeting to the ER (Mock et al., 2015), while the N-terminal UBL domain of the protein promotes recruitment of the ubiquitination machinery to mediate quality control of aberrant proteins (Fig. 2; Rodrigo-Brenni et al., 2014). In this context, BAG6 has been implicated in rerouting SGTA-bound TA proteins that are not efficiently relayed to TRC40 toward the degradation pathway (Shao et al., 2017). SGTA has emerged as another central player in determining the fate of TA proteins, as it is able to antagonize BAG6-facilitated protein ubiquitination. Interestingly, SGTA not only reduces the likelihood of protein ubiquitination by shielding the hydrophobic TMD but also promotes active deubiquitination of BAG6 complex–associated proteins (Leznicki and High, 2012). In this way, SGTA could contribute to the recovery of nascent TA proteins aberrantly marked for degradation by BAG6-associated ubiquitin ligases. Notably, the interplay between the BAG complex and SGTA in determining protein fate extends beyond TA proteins to other membrane proteins targeted cotranslationally (Leznicki and High, 2020). Intriguingly, it was recently shown that ubiquitination of TA proteins can occur independently of BAG6, and that ubiquitinated TA proteins can still be handled by TRC40 and directed to the ER, where they are fully deubiquitinated by USP20/USP33 before or after membrane insertion (Fig. 2; Culver and Mariappan, 2021). It remains unclear mechanistically how exactly ubiquitinated TA proteins evade proteosome-mediated degradation in the cytosol, but it is possible that either the nature of the ubiquitination and/or rapid capture by TRC40 enables them to efficiently reach their destination and be deubiquitinated. It will be interesting to determine if this ubiquitination-deubiquitination cycle simply represents a futile mislabeling and recovery process or whether it fulfils a specific function during TA protein biogenesis.Delivery of TA proteins into the ERBoth the GET/TRC receptor and the EMC complex have emerged as gateways into the ER for TA proteins (Guna et al., 2018; Schuldiner et al., 2008; Vilardi et al., 2011; Yamamoto and Sakisaka, 2012). TA proteins destined for ER insertion via the GET/TRC receptor converge on the homodimeric cytosolic chaperone Get3/TRC40, which escorts them to the ER-bound GET/TRC insertase (McDowell et al., 2020; Wang et al., 2014). Docking of ADP and TA protein–bound Get3/TRC40 onto the GET receptor allows transfer of the TA protein to the receptor, which subsequently mediates their insertion into the membrane (Wang et al., 2014). Then, upon ADP release, Get3/TRC40 is recycled for another round of ATP binding and TA protein targeting (reviewed in detail in Chio et al., 2017).Evolutionary conservation of the GET receptor complexThe Get1 component of the GET receptor is a member of the Oxa1 superfamily of insertase proteins, which includes bacterial YidC and eukaryotic EMC3 (Anghel et al., 2017; McDowell et al., 2021), and it shares its three-TMD topology with other members of this family. Get1 sequence conservation among different phyla readily facilitated the identification of homologues in mammals (WRB) and plants (AtGet1; Srivastava et al., 2017; Xing et al., 2017). In contrast, Get2 homologues are more divergent, and sequence conservation is limited to the functionally essential N-terminal Get3-binding sequence and, to a lesser extent, the three TMDs (Borgese, 2020). CAML has nevertheless been identified as a functional and structural homologue of Get2 in mammals (Yamamoto and Sakisaka, 2012), and demonstrated to complement phenotypes associated with loss of Get1/Get2 when coexpressed with WRB in budding yeast (Vilardi et al., 2014). The existence of Get2 orthologues or functional homologues in other phyla were uncertain for a long time. The recent identification of the archaeplastidic Get2 homologue AtGet2(Asseck et al., 2021) and in silico prediction of Get2 homologues in mollusks and arthropods (Borgese, 2020), however, now strongly support that not only Get1/WRB, but the GET receptor complex as a whole is conserved among eukaryotes.Architecture and stoichiometry of the GET receptorThe interaction between Get1/WRB and Get2/CAML is mediated by their TMDs, which are also necessary for the insertase function of the complex (Vilardi et al., 2014; Wang et al., 2014). Moreover, mammalian WRB and CAML, and likely other homologues as well, are thought to exist as an obligate complex, mutually stabilizing each other. Indeed, the expression levels of WRB or CAML decrease in the absence of the other subunit, likely because of destabilization of the remaining subunit (Colombo et al., 2016; Rivera-Monroy et al., 2016). However, the effects of the individual components seem to be asymmetric on each other. Namely, WRB can insert into the ER membrane correctly in the absence of CAML and is later degraded as an orphan subunit, whereas WRB is required for CAML to assume a correct topology in the first place (Carvalho et al., 2019; Inglis et al., 2020). More specifically, the second TMD of CAML remains exposed to the ER lumen in the absence of WRB, where it acts as a degron, triggering protein degradation. Interestingly, its topology can be corrected posttranslationally, and its degradation prevented when WRB is available in the membrane (Inglis et al., 2020). It remains to be seen whether a similar inter-subunit interplay also occurs in other species; however, results from A. thaliana imply that correct assembly of the GET receptor may be differently controlled in plants, as ectopically expressed AtGet2 remains stable in the absence of AtGet1 (Asseck et al., 2021).Although formation of a Get1/Get2 heterodimer is recognized as a prerequisite for a minimal functional receptor complex, a higher-order stoichiometry of the insertion-competent GET receptor complex has long been actively discussed. Due to the symmetric, homodimeric nature of Get3 in the TA protein–loaded complex, two analogous binding sites for both Get1 and Get2 are offered (Stefer et al., 2011). Despite a partial overlap of the Get1 and Get2 interaction sites on Get3, simultaneous binding of Get3 by Get1 and Get2 has been observed in crystal structures of Get3, together with the cytosolic domains of Get1 and Get2 (Stefer et al., 2011). This gave rise to the notion of a heterotetrameric structure of the GET receptor composed of two Get1 and two Get2 subunits, which could bind a single Get3 dimer with high affinity (Mariappan et al., 2011; Stefer et al., 2011). However, results obtained with in vitro reconstituted proteoliposomes demonstrated that a single dimer of Get1/Get2 can be sufficient for insertion of TA substrates into the membrane (Zalisko et al., 2017). Nevertheless, the heterotetrameric arrangement is supported by recent high-resolution cryo-EM structures of Get3/TRC40 homodimers docked onto the yeast and mammalian GET receptors, which reveal the formation of a heterotetrameric receptor complex upon Get3/TRC40 binding (Figs. 1, ​,2,2, and ​and3;3; McDowell et al., 2020). These new structures further consolidate the previously proposed model (Mariappan et al., 2011; Stefer et al., 2011) that Get3/TRC40 is initially captured by the extended cytosolic domains of Get2/CAML before contacting Get1/WRB (Figs. 1, ​,2,2, and ​and3).3). This arrangement, coupled with the finding that Get1-Get2 can simultaneously bind two Get3 molecules (McDowell et al., 2020), opens the possibility for a relay system where translocation of a first Get3/TRC40-TA protein complex to Get1 immediately allows capture of a second substrate complex by Get2, potentially increasing the efficiency of receptor complex loading and minimizing the risk of mistargeting. Interestingly, the GET/TRC receptor seems to be stabilized at the heterotetramer interface by not only protein–protein but also protein–lipid interactions, indicating that the lipid environment of the ER membrane may also influence the oligomeric state of the receptor. Importantly, complementation experiments in yeast demonstrate that disrupting lipid binding and thus the formation of the heterotetramer leads to in vivo loss of function of the receptor manifesting in TA protein mislocalization (McDowell et al., 2020). Therefore, although a Get1/Get2 dimer appears to be sufficient for the insertase function of the receptor in vitro, it is highly likely that a tetrameric complex is required to ensure efficient and accurate TA protein targeting within the cellular environment.Open in a separate windowFigure 3.Architecture of the Get1/WRB and Get2/CAML in the GET/TRC receptor complex. Get1/WRB and Get2/CAML both possess three TMDs (labeled 1–3) and rely on each other for stability and correct assembly within the ER membrane. The cytosolic regions of Get2/CAML (N-terminus and a loop between TMD 2 and 3) and a cytosolic region of Get1/WRB between TMDs 1 and 2 are docking sites for Get3/TRC40 carrying a TA protein cargo. A hydrophilic groove formed by the Get1/WRB TMDs and Get2/CAML TMD3 is proposed to serve as an insertion route for TA proteins to enter the membrane. Assembly of the heterodimer shown here into the final heterotetrameric structure of the receptor upon Get3/TRC40 binding involves protein–lipid interactions.Mechanistic view of TA protein insertion into the lipid bilayerRecent structural advances provide exciting mechanistic insights into the details of TA insertion, both by the GET/TRC receptor (McDowell et al., 2020) and the EMC complex (Bai et al., 2020; Miller-Vedam et al., 2020; O’Donnell et al., 2020; Pleiner et al., 2020). In the case of the GET/TRC receptor, within the membrane, the TMDs of WRB, together with TMD3 of CAML are arranged such that a hydrophilic groove, sealed at the luminal face but accessible from the cytosol, is assembled (Fig. 3). This likely serves as a substrate entry point with the charged, extreme C-terminus of the TA protein drawn in by interactions with the numerous hydrophilic residues of the receptor channel, thus bringing the TMD in close proximity to the destabilized bilayer, allowing insertion. This corroborates previous results obtained by cross-linking nascent TA substrates to the receptor in vitro, which pinpointed the region around the hydrophilic groove as the entry point for TA proteins into the membrane (Wang et al., 2014). Interestingly, assembly of the GET receptor as a heterotetrameric complex means that tandem hydrophilic grooves generated by each Get1/WRB-Get2/CAML pair represent two alternative routes into the membrane. The asymmetric binding of the TA protein within the Get3 dimer (Mateja et al., 2015), means that depending on the orientation of the docking, insertion via a particular channel will be favored. Dynamic modeling of interactions between the receptor and the TRC40 dimer in different conformations/nucleotide-bound states (Mateja et al., 2015; Stefer et al., 2011) suggest that transition of Get3/TRC40 to the open conformation leads to rearrangement of complex such that the C-terminus of the released TA protein is oriented toward the hydrophilic groove (McDowell et al., 2020). Complementary structural analyses of the EMC complexes (Bai et al., 2020; Miller-Vedam et al., 2020; O’Donnell et al., 2020; Pleiner et al., 2020) reveal analogous hydrophilic groove features, indicating that a common insertion mechanism is used by evolutionarily distant membrane receptors to accomplish insertion of a diverse set of membrane proteins (Bai and Li, 2021; McDowell et al., 2021). For the GET/TRC receptor, it is not yet clear how exactly the TA protein transits from this hydrophilic groove to become fully immersed in the membrane, but an amphipathic helix of Get1/WRB that lies close to the membrane has been suggested to cause membrane distortions that could facilitate TA protein integration.Another key feature of the GET receptor revealed by the new structures is a short helix, α3′, present in the cytosolic domains of both yeast Get2 and human CAML, which binds the TA-binding domain of Get3/TRC40 and serves a gating function for the hydrophilic groove (Fig. 3; McDowell et al., 2020). Contact between the Get2/CAML helix and the Get3/TRC40 TA-binding domain induce conformational rearrangements crucial to the release and insertion of TA substrates in vivo, further underlining the role the GET receptor plays in stimulating substrate release from Get3. It remains to be determined whether, mechanistically, this helix facilitates TA protein insertion by preventing backsliding of the TMD out of the hydrophilic groove or actively driving the TA protein into the channel.Quality control and rescue of TA protein targetingThe multiple possible destinations for TA proteins, together with the broad spectrum of physical properties of secretory pathway TA proteins, the existence of alternative ER targeting and insertion strategies, and the complexity of ensuring optimal capture, renders targeting of TA proteins to the ER inherently prone to errors. Defects in this process can manifest as either redirection of ER TA proteins to other membranes or the spurious ER insertion of non-ER TA proteins. Given the importance of TA protein functions and the toxic effects of cytosolic protein aggregates, either of these scenarios can have seriously detrimental effects on cells, necessitating robust surveillance, recovery, and degradation pathways (Jiang, 2021).Removal of TA proteins misdirected to the OMMIn contrast to ER TA proteins, some TA proteins of the OMM have been proposed to be inserted directly, potentially due to the specific lipid composition of the OMM (Figueiredo Costa et al., 2018). The hydrophobicity of their TMDs are lower than that of secretory pathway TA proteins (Borgese et al., 2001; Beilharz et al., 2003), which, considering that high TMD hydrophobicity is necessary for recognition by the GET pathway (Guna et al., 2018), helps explain how they avoid capture and subsequent delivery to the ER by Get3. In contrast, peroxisomal TA proteins can either use the Pex19-Pex3 machinery to directly reach peroxisomes (reviewed in Mayerhofer, 2016) or be first targeted to the ER by the GET pathway (Schuldiner et al., 2008) before reaching peroxisomes (Fig. 4).Open in a separate windowFigure 4.Quality control machineries regulating distribution of mislocalized TA proteins between the ER and other organelles. ER-destined and peroxisome (Perox.)-destined TA proteins can mislocalize to the OMM, and OMM TA proteins can mislocalize to the ER. ATP-dependent machines in these membranes can recognize and displace mistargeted protein while correctly localized proteins are retained, for example by interactions with binding partners (indicated by unnamed, colored circles on the OMM). Nomenclature is as in yeast.It has been observed that ER-inserted TA proteins can mislocalize to mitochondria when the functionality of the GET pathway is impaired (Schuldiner et al., 2008). A salient example is yeast Pex15, which is a peroxisome-destined TA protein first inserted into the ER but that mislocalizes to mitochondria not only when the C-terminal 30 amino acids or Pex19 are lacking, but also in the absence of Get3 (Schuldiner et al., 2008; Okreglak and Walter, 2014; Li et al., 2019). Furthermore, a basal level of mistargeting of secretory pathway TA proteins, such as yeast Gos1, to the OMM is observed even in the presence of a functional GET/TRC pathway, implying that some level of mistargeting is unavoidable (Chen et al., 2014). This therefore raises the question of how mistargeted TA proteins are recognized and cleared while the appropriately localized proteins are retained.A mechanism by which TA proteins incorrectly inserted into the OMM can be extracted has recently been described; the highly evolutionarily conserved mitochondrial and peroxisomal AAA-ATPase Msp1 (ATAD1 in mammals) has been shown to be essential for the removal of mitochondrially mislocalized Pex15 and is thought to act as a general dislocase of TA proteins mislocalized to the mitochondria (Fig. 4; Chen et al., 2014; Okreglak and Walter, 2014). This explains why double-mutant yeast strains lacking both GET pathway components and Msp1 mislocalize TA proteins noticeably to mitochondria (Li et al., 2019). Accordingly, GET pathway components and Msp1 show a synthetic negative genetic effect (Chen et al., 2014; Okreglak and Walter, 2014), emphasizing the importance of high-fidelity protein targeting and quality control. Msp1 forms hexamers in the OMM, and in vitro, its ATPase activity is sufficient to drive the removal of TA proteins from proteoliposomes (Wohlever et al., 2017), demonstrating that Msp1 alone can both recognize its substrates and drive their subsequent extraction from the membrane. Recent visualizations of Msp1-substrate complexes provide insights into the mechanism of membrane extraction; the TA protein substrate enters via a single, hydrophobic site and is then anchored within the hydrophobic pore by a network of aromatic amino acids (Wang et al., 2020). ATP hydrolysis, the driving force for membrane extraction, is coordinated with subunit positioning within the complex via specific elements at the subunit interfaces (Wang et al., 2020). Interestingly, based on in vitro experiments with MBP-tagged Pex3, it has been suggested that the unfoldase activity of Msp1 may be regulated by Pex3 on peroxisomes (Castanzo et al., 2020), but it remains to be determined whether a similar regulatory mechanism exists in the OMM as well.The stringency of recognition of mitochondrially mislocalized TA proteins relies on at least a twofold recognition mechanism by Msp1. First, basic residues typical of the luminal tails of peroxisomal TAs are recognized when exposed in the mitochondrial intermembrane space (Li et al., 2019). Second, exposed hydrophobic amino acids close to the membrane, present in various secretory pathway and peroxisomal TA proteins, also act as a recognition signal for Msp1 (Li et al., 2019). Furthermore, there is evidence indicating that orphan TA proteins lacking binding partners within the OMM are more readily recognized and displaced by Msp1 (Weir et al., 2017; Dederer et al., 2019). This indicates that besides the biophysical properties of Msp1 substrates, their ability to form functional protein–protein interactions with other membrane proteins is an important discriminating factor that determines whether a TA protein is retained in or removed from the OMM. Mechanistically, lack of a binding partner may render mislocalized TA proteins less stably anchored within the membrane, or, when separated from their normal interaction partners, mistargeted proteins may be more likely to expose hydrophobic patches to the cytosol, both of which would increase the chance of expulsion by Msp1.It is possible that after removal from the OMM by Msp1, secretory pathway TA proteins returned to the cytosol could be retargeted if they are captured by Get3/TRC40 or other chaperones capable of directing them to the ER-bound insertase machineries. However, a cellular machinery also needs to exist that degrades excess TA proteins in the cytosol or at the ER to ensure protein homeostasis. Indeed, the E3 ubiquitin ligase Doa10, an important player in ER-associated degradation, has recently emerged as a quality control factor responsible for sensing and ubiquitination of spurious and excess TA proteins ejected from the OMM. It has been suggested that Doa10-mediated ubiquitination may take place either in the cytosol (Dederer et al., 2019) or after retargeting to the ER, where their extraction and degradation is facilitated by the AAA-ATPase Cdc48 (Matsumoto et al., 2019; Fig. 4).Mistargeting of mitochondrial TA proteins to the ERIt is not only the case that nonmitochondrial TA protein are misdirected to the OMM; also, reciprocal events can occur as OMM TMD-containing proteins are observed in the ER, especially when mitochondrial targeting signals are masked, when the mitochondrial import machineries are overloaded, or upon mitochondrial dysfunction (Hansen et al., 2018; Xiao et al., 2021; Vitali et al., 2018). More specifically, the GET pathway itself has been shown to contribute to the ER mislocalization of OMM TA proteins (Vitali et al., 2018; Xiao et al., 2021). While the high efficiency of mitochondrial targeting appears to keep such mistargeting to a minimum, perturbation of the equilibrium between OMM targeting and ER mistargeting by the GET pathway can lead to aberrant accumulation of clients in the wrong membrane. Similar to nonmitochondrial TA proteins ejected from the OMM by Msp1, a parallel ATP-dependent mechanism for displacing TA proteins incorrectly introduced into the ER has recently been discovered (Fig. 4; McKenna et al., 2020; Qin et al., 2020). Structural and biochemical analyses of the ER-bound P5A-type ATPase Spf1 (yeast)/ATP13A1 (human)/CATP-8 (Caenorhabditis elegans) have identified it as a major quality control factor in the ER. Spf1 contains a substrate- binding pocket, laterally accessible from the membrane, that has been proposed to flip ER-associated proteins, promoting their release back into the cytosol or their topological rearrangement. Precisely how specificity for mislocalized/misinserted proteins is achieved remains unclear, but it is suggested that the lower hydrophobicity of non-ER TMDs may favor their dislocation. Furthermore, in cells lacking Spf1, the ergosterol content of the ER is altered to more closely resemble the OMM, implying that membrane composition may contribute to TA protein distribution and that Spf1 could also influence TA protein mislocalization by regulating the lipid composition of membranes (Krumpe et al., 2012). In yet another analogy to Msp1-mediated rejection of mislocalized TA proteins from the OMM, it is likely that mitochondrial proteins extracted from the ER may then be successfully retargeted to their desired destination. Indeed, the recently described ER–surface-mediated protein targeting pathway sets a precedent for such a route (Hansen et al., 2018).Concluding remarksThe initial momentum of the TA protein–targeting field following the discovery of the GET/TRC pathway has not abated. Recent years have seen not only a deepening mechanistic understanding of the intricacies of this pathway but also growing knowledge on its interplay with other targeting pathways and the safety nets present to ensure the fidelity of TA protein targeting. Building on the early structural analyses of individual proteins and protein domain complexes, recent advances in cryo-EM have now enabled visualization of larger complexes, such as the targeting factor–bound membrane insertase, allowing the route of TA proteins from the cytosol into the ER membrane to be anticipated. Alongside high-throughput microscopy–based screens, further analyses of TA protein targeting in cells have uncovered alternative routes that TA protein can take to reach the ER and have highlighted functional redundancies. The existence of different pathways for TA protein targeting and insertion into the ER, on the one hand, suggest a robust system with backup strategies in place in case a client protein escapes its normal targeting route, but on the other hand, emphasize the complexity of the capture process and selection of the optimal targeting pathway. In turn, these additional layers of complexity underscore the need for quality control pathways. The discovery of active removal of mislocalized TA proteins not only provides a mechanism for such quality control but perhaps also suggests a dynamic aspect to TA protein targeting wherein TA proteins within membranes can be removed and then either retargeted to their correct destination or reinserted if appropriate. Altogether, a picture emerges of an array of capture, insertion, and ejection machineries that are finely balanced in terms of substrate preferences to optimize TA protein localization within the cell.     

Mitofusins: ubiquitylation promotes fusion

Mitochondrial genes including Mfn2 are at the center of many diseases, underscoring their potential as a therapeutical target. The Chen group now identified 15-oxospiramilactone as a chemical inhibitor of the mammalian deubiquitylase USP30, acting on Mfn1 and Mfn2.Mitofusins, Fzo1 in yeast and Mfn1 and Mfn2 in mammals, are ubiquitylated and this post-translational modification has both positive and negative consequences on mitochondrial fusion¹. The process of ubiquitylation requires enzymes belonging to three classes of proteins called E1, E2 and E3, which catalyze a cascade of successive steps leading to the covalent attachment of the modifier to its target protein². Deubiquitylating enzymes render this modification reversible, thus offering further possibilities for regulation². Ubiquitylation of mitofusins leads to their proteolyic breakdown, inhibiting fusion of mitochondria that consequently undergo fragmentation (Figure 1, left panel)^1,3. For example in response to mitochondrial depolarization or apoptotic stimuli, E3 ligases like Parkin and Huwe1 ubiquitylate and target Mfn1 and Mfn2 to the proteasome (Figure 1, left panel)^3,4. However, ubiquitylation of mitofusins is a dual process and a non-proteolytic role of mitofusin ubiquitylation that promotes mitochondrial fusion is now emerging¹. This opposing mechanism was first described in yeast, where the isopeptidases Ubp12 and Ubp2 that deubiquitylate Fzo1 have been identified⁵. Inhibition and activation of mitochondrial fusion by ubiquitylation enable different morphologies of mitochondria ranging from a multitude of small organelles to a hyperconnected network (Figure 1)⁵. In a recent paper published in Cell Research, Yue et al.⁶ reveal that a similar process is present in mammalian cells. The authors report that the isopeptidase USP30 acts on ubiquitylated forms of Mfn1 and Mfn2 that stimulate mitochondrial fusion (Figure 1, right panel). This discovery identifies for the first time in mammals a positive role of ubiquitylation in the regulation of Mfn1 and Mfn2 fusion activity⁶.Open in a separate windowFigure 1Dual roles of ubiquitylation and deubiquitylation of mitofusins Mfn1 and Mfn2, the key effectors for mitochondrial fusion, in regulating mitochondrial fusion. On one hand, ubiquitylation of Mfn1 and Mfn2 by E3 ligases like Parkin or Huwe1 targets their proteasomal degradation and inhibits mitochondrial fusion, which results in mitochondrial fragmentation due to unopposed fission events. On the other hand, ubiquitylation of Mfn1 and Mfn2 by an unknown E3 ligase enhances their activity and promotes mitochondrial fusion. This positive regulation is counteracted by the deubiquitylase USP30, targeted by the small molecule inhibitor 15-oxospiramilactone.Moreover, Yue et al.⁶ identified the first small molecule inhibitor of mitochondrial fusion, 15-oxospiramilactone, which targets USP30 in both human and mouse cell lines. 15-oxospiramilactone is a semi-synthetic diterpene alkaloid of 330 Da that can be chemically synthetized through an oxidation reaction from spiramines extracted from the roots of a Chinese herbal medicine Spiraea japonica (Rosaceae). Inhibition of USP30 increased ubiquitylation of Mfn1 and Mfn2 and led to an elongation of the mitochondrial network (Figure 1, right panel)^6,7. USP30 is a cysteine ubiquitin isopeptidase N-terminally anchored to the outer membrane of mitochondria, which was previously shown to regulate mitochondrial morphology dependent on Mfn1 and Mfn2⁷. USP30 knockdown leads to mitochondrial elongation, a phenotype rescued by ectopic expression of wild-type USP30, while the catalytically inactive mutant C77S USP30 failed to revert⁷. Yue et al.⁶ show that 15-oxospiramilactone directly interacts with USP30, which also depends on its catalytically active cysteine, and inhibits the DUB activity of USP30 on tetraubiquitin chains. Moreover, they demonstrate that inhibition of USP30 and subsequent mitochondrial elongation are due to stimulated mitochondrial fusion activity, apparently with no influence on mitochondrial fission⁶. Concomitantly, cells showed increased ubiquitylation of Mfn1 and Mfn2 without significant changes in protein turnover of these two proteins⁶. Therefore, in analogy to findings in yeast, ubiquitylation of Mfn1 and Mfn2 can either signal them to activate mitochondrial fusion or in contrast promote their proteasomal degradation, resulting in mitochondrial fission (Figure 1).Importantly, 15-oxospiramilactone reverts the mitochondrial fragmentation phenotype of single Mfn-knockout (Mfn1^−/− or Mfn2^−/−) cells, suggesting that mitochondrial fusion depends on the ubiquitylation of both mitofusin proteins⁶. In yeast, the importance of ubiquitylation was proven by directly attaching a deubiquitylase to Fzo1, which resulted in a non-ubiquitylated and non-functional Fzo1 protein⁵. In addition, the identification and the subsequent mutagenesis study of the ubiquitylation sites in Fzo1 confirmed an interplay between ubiquitylation and oligomerization in mitochondrial fusion in S. cerevisiae⁵. Impairing the yeast E3 ligase SCF^Mdm30 inhibited mitochondrial fusion and, conversely, ablation of UBP12 led to more fusion events^5,8. Given this new identification of USP30 as the functional orthologue of the yeast Ubp12, future studies will certainly aim at the identification of the E3 ligase counterpart of SCF^Mdm30 and ubiquitylation sites in Mfn1 and Mfn2. In addition to USP30 inhibition, other conditions leading to mitochondrial hyperfusion have been previously observed, such as mild stress conditions that increase reactive oxygen species (ROS)⁹. Importantly, oxidative stress and mitochondrial fusion are directly linked as ROS induces disulphide switching of Mfn2 to oligomeric forms that promote mitochondrial fusion⁹. It would be interesting to investigate whether 15-oxospiramilactone also affects the generation of disulphide-mediated mitofusin oligomers, thus activating mitochondrial fusion.Mutations in Mfn2 are causative for the Charcot-Marie-Tooth type 2A neuropathy, an autosomal dominant disorder of the peripheral nervous system that mainly affects axons and lower extremities¹. Deficiencies in Parkin and Mfn2 ubiquitylation were also linked to Parkinson''s disease³. In addition to neuropathies, Mfn2 is associated to other diseases like cardiomyophathies and diabetes¹. Yue et al.⁶ found that 15-oxospiramilactone reverted phenotypes arising from the lack of Mfn1 or Mfn2. It restored the normal distribution of mtDNA, allowed recovery of the ΔΨm and increased the ATP levels and OXPHOS capacity of the rebuilt mitochondrial network. Therefore, this study potentiates 15-oxospiramilactone for therapeutical benefit. The anti-cancer properties of 15-oxospiramilactone, also named S3 or NC043, have been previously reported^10,11. It inhibits Wnt/β-catenin signaling and colon cancer cell tumorigenesis in a xenograft model¹⁰. Moreover, 15-oxospiramilactone increases Bim expression and apoptosis to inhibit tumor growth from Bax^−/−/Bak^−/− cells implanted in mice¹¹. However, Yue et al.⁶ show that the effect of 15-oxospiramilactone in mitochondrial fusion is independent of apoptosis and suggest that the difference is due to drug concentration. Indeed, previous anti-cancer studies used 15-oxospiramilactone at a concentration range of 3.75-15 μM^10,11, whereas 2 μM suffice to inhibit USP30⁶. Further studies are needed to address the clinical relevance of 15-oxospiramilactone and USP30 in Mfn2-associated diseases.     

Single-Particle Tracking Demonstrates that Actin Coordinates the Movement of the Ebola Virus Matrix Protein

The Ebola virus causes severe hemorrhagic fever and has a mortality rate that can be as high as 90%, yet no vaccines or approved therapeutics, to our knowledge, are available. To replicate and egress the infected host cell the Ebola virus uses VP40, its major matrix protein to assemble at the inner leaflet of the plasma membrane. The assembly and budding of VP40 from the plasma membrane of host cells seem still poorly understood. We investigated the assembly and egress of VP40 at the plasma membrane of human cells using single-particle tracking. Our results demonstrate that actin coordinates the movement and assembly of VP40, a critical step in viral egress. These findings underscore the ability of single-molecule techniques to investigate the interplay of VP40 and host proteins in viral replication.The actin cortex below the plasma membrane of mammalian cells is essential for maintenance of cell shape and for cell movement. This cortex has also been found to play an essential role in the replication process of a number of viruses including West Nile virus (1), respiratory syncytial virus (2), influenza (3), and vaccinia virus (4). Additionally, actin has been found to play a central role in the assembly and budding of HIV-1 (5) whereas Marburg virus has been shown to use actin-enriched filopodia to exit the host cell (6). Actin has also been found to be packaged into Ebola-virus-like particles (VLPs) (7). Ebola virus, which causes severe hemorrhagic fever, harbors a single-stranded negative-sense RNA genome encoding seven proteins. Of these seven proteins, VP40 is the most abundantly expressed and has been found to play a central role in the budding of the virus from the plasma membrane (8). Whereas actin has been found in Ebola VLPs (7), the role of actin in Ebola VP40 assembly is still seemingly unknown. Here, we have used Raster image correlation spectroscopy (RICS) (9) and three-dimensional single-particle tracking (see Fig. S1 in the Supporting Material) (10) to investigate the dynamics of Ebola VP40 and actin. We report that preassembled VLPs (pVLPs) of Ebola VP40 require actin for directed movement and assembly.Ebola VP40 has been demonstrated to colocalize with actin and actin is found in VP40 VLPs (7), suggesting an important role for actin in the replication cycle of the virus. To confirm the colocalization between VP40 and actin in HEK293 and CHO-K1 cells, we used confocal microscopy to examine the distribution of EGFP-VP40 and mCherry-actin. EGFP-VP40 and mCherry-actin displayed colocalization at the plasma membrane of HEK293 and CHO-K1 cells (see Fig. S2 A), which was markedly reduced in response to treatment with LAT-A (see Fig. S2 B and Fig. S3 A), an actin polymerization inhibitor. VP40 plasma membrane localization was not disrupted by LAT-A treatment (not unexpected, as VP40 is a lipid-binding protein (11) where high affinity for the PM drives its cellular localization (E. Adu-Gyamfi and R. V. Stahelin, unpublished)). To test whether this VP40-actin interaction is important to viral egress, we detected EGFP-VP40 with an anti-EGFP antibody used to measure VLPs formed from cells expressing EGFP-VP40. This was also performed to assess the effect of pharmacological treatment on EGFP-VP40-expressing cells with LAT-A or with the microtubule polymerization inhibitor nocodazole (see Fig. S3 B). LAT-A treatment led to a significant reduction in VLP formation whereas nocodazole did not display detectable effects.To test whether the VP40 and actin are engaged in synchronized movement, we performed time-lapse imaging in both the green and red channels. We observed that the pVLPs move with actin fibers extending from the plasma membrane (see Movie S1 in the Supporting Material). The movement was rapid, and caused smaller particles to merge into larger filamentous forms. To further demonstrate that the motion of actin and VP40 spatially overlapped, we used RICS to obtain correlation maps of EGFP-VP40 and mCherry-actin (Fig. 1). The spatial cross-correlation map indicated significant overlap of VP40 and actin movement (Fig. 2, A–C) at the plasma membrane (Fig. 1 and see Fig. S6), but not in the cytosol (see Fig. S5 and Fig. S7). In contrast, EGFP-VP40 and mCherry-α-tubulin (see Fig. S8, Fig. S9, and Fig. S10) displayed no significant spatial cross-correlation at the plasma membrane (Fig. S11) or other regions of the cell (see Fig. S12), supporting the VLP egress data where inhibition of microtubule polymerization did not influence viral egress.Open in a separate windowFigure 1EGFP-VP40 and mCherry-actin RICS analysis at the membrane. (A) HEK293 cells expressing EGFP-VP40 and mCherry-actin were imaged for 100 frames at 256 × 256 pixels. (White scale bar = 2 μm.) (B) Average intensity image of EGFP-VP40 across the 100 collected frames. (Pink box) Used to select a region of interest to yield the (C) average EGFP-VP40 intensity image. (D) Average intensity image of mCherry-actin taken for 100 frames at 256 × 256 pixels was used to select the same region of interest as in panel B (pink box) to yield the (E) average intensity image of the mCherry-actin signal in this region. (F) The two-dimensional spatial cross-correlation analysis of panels C and E demonstrates significant cross-correlation of VP40 and actin signals.Open in a separate windowFigure 2Three-dimensional RICS correlation maps of VP40 and actin cross-correlate at the plasma membrane. (A) EGFP-VP40 and (B) mCherry-actin (Fig. 1 and see Fig. S6 in the Supporting Material) RICS autocorrelation functions. (C) Appreciable cross-correlation is observed for EGFP-VP40 and mCherry-actin at the plasma membrane.To test whether the motion of the pVLPs is directed by actin, we applied the three-dimensional orbital tracking method first introduced by Levi et. al. (10). Tracking of isolated particles (Fig. 3 A) in five different cells allowed determination of the pVLPs trajectories (Fig. 3 D), which suggested that the VP40 particles undergo a directed motion. To verify this, we plotted the mean-square displacement (MSD) curves for the pVLPs (Fig. 3 C), which confirmed the trajectory was characteristic of directed motion. Analysis of the intensity profile of the dynamic VP40 particles suggested that the intensity of the particle changes with respect to time. Bleaching is expected if the molecule is exposed to the laser beam for an extended period of time; however, an increase in intensities was observed along the trajectory of the green channel due to addition of VP40 molecules. This suggests that the movement of the particles along actin fibers promote multimerization and maturation of the pVLPs. When actin polymerization was inhibited in four different cells with LAT-B, the rapid movement (see Fig. S13) and the directed trajectories of the pVLPs were lost (Fig. 3, E and F). This was reflected in a change from directed motion to movement indicative of random then constrained diffusion (Fig. 3, E and F).Open in a separate windowFigure 3Actin directs the movement of VP40 particles. HEK293 cells transfected with EGFP-VP40 were imaged with an electronic zoom of 2000 mV, corresponding to 72 nm/pixel in both X and Y. (A) An isolated and representative VP40 particle (highlighted by white box, inset) was tracked as described in the Supporting Material. (B) Intensity profile of the pVLP in A demonstrates increases in EGFP-VP40 intensity along the trajectory. (C) MSD of the pVLP, which follows a ballistic motion with a velocity of 0.067 ± 0.01 μm² s⁻¹. (D) The three-dimensional trajectory of the particle shown in panels A–C. (E) MSD curve of VP40 particles yields random then constrained diffusion after LAT-B treatment with a mean velocity of 0.017 ± 0.006 μm² s⁻¹ (p < 0.001). (F) Three-dimensional trajectory of the same particle shown in panel E displays a random then constrained diffusion.Taken together, our findings demonstrate that the movement of the pVLPs is driven by actin. Analysis of the pVLPs trajectories also suggests that the motion of pVLPs on actin enables further addition of VP40 molecules. These findings raise important questions regarding contemporary understanding of Ebola assembly and egress. VP40 lacks a consensus actin-binding motif, suggesting an adaptor protein such as an actin motor protein may function in this process. For instance, Myo10 has been found to be essential to Marburg virus release (6); however, Marburg VP40-Myo10 direct interactions were not observed, suggesting other cellular adaptor proteins may function in this process. Given the pathogenic nature of the Ebola virus and the necessity of VP40 to the assembly and egress of the virus (8), the VP40-actin coordination represents, to us, a novel target for therapeutic development.     

Novel Growth Regime of MDCK II Model Tissues on Soft Substrates

It is well established that MDCK II cells grow in circular colonies that densify until contact inhibition takes place. Here, we show that this behavior is only typical for colonies developing on hard substrates and report a new growth phase of MDCK II cells on soft gels. At the onset, the new phase is characterized by small, three-dimensional droplets of cells attached to the substrate. When the contact area between the agglomerate and the substrate becomes sufficiently large, a very dense monolayer nucleates in the center of the colony. This monolayer, surrounded by a belt of three-dimensionally packed cells, has a well-defined structure, independent of time and cluster size, as well as a density that is twice the steady-state density found on hard substrates. To release stress in such dense packing, extrusions of viable cells take place several days after seeding. The extruded cells create second-generation clusters, as evidenced by an archipelago of aggregates found in a vicinity of mother colonies, which points to a mechanically regulated migratory behavior.Studying the growth of cell colonies is an important step in the understanding of processes involving coordinated cell behavior such as tissue development, wound healing, and cancer progression. Apart from extremely challenging in vivo studies, artificial tissue models are proven to be very useful in determining the main physical factors that affect the cooperativity of cells, simply because the conditions of growth can be very well controlled. One of the most established cell types in this field of research is the Madin-Darby canine kidney epithelial cell (MDCK), originating from the kidney distal tube (1). A great advantage of this polarized epithelial cell line is that it retained the ability for contact inhibition (2), which makes it a perfect model system for studies of epithelial morphogenesis.Organization of MDCK cells in colonies have been studied in a number of circumstances. For example, it was shown that in three-dimensional soft Matrigel, MDCK cells form a spherical enclosure of a lumen that is enfolded by one layer of polarized cells with an apical membrane exposed to the lumen side (3). These structures can be altered by introducing the hepatocyte growth factor, which induces the formation of linear tubes (4). However, the best-studied regime of growth is performed on two-dimensional surfaces where MDCK II cells form sheets and exhibit contact inhibition. Consequently, the obtained monolayers are well characterized in context of development (5), mechanical properties (6), and obstructed cell migration (7–9).Surprisingly, in the context of mechanics, several studies of monolayer formation showed that different rigidities of polydimethylsiloxane gels (5) and polyacrylamide (PA) gels (9) do not influence the nature of monolayer formation nor the attainable steady-state density. This is supposedly due to long-range forces between cells transmitted by the underlying elastic substrate (9). These results were found to agree well with earlier works on bovine aortic endothelial cells (10) and vascular smooth muscle cells (11), both reporting a lack of sensitivity of monolayers to substrate elasticity. Yet, these results are in stark contrast with single-cell experiments (12–15) that show a clear response of cell morphology, focal adhesions, and cytoskeleton organization to substrate elasticity. Furthermore, sensitivity to the presence of growth factors that are dependent on the elasticity of the substrate in two (16) and three dimensions (4) makes this result even more astonishing. Therefore, we readdress the issue of sensitivity of tissues to the elasticity of the underlying substrate and show that sufficiently soft gels induce a clearly different tissue organization.We plated MDCK II cells on soft PA gels (Young’s modulus E = 0.6 ± 0.2 kPa), harder PA gels (E = 5, 11, 20, 34 kPa), and glass, all coated with Collagen-I. Gels were prepared following the procedure described in Rehfeldt et al. (17); rigidity and homogeneity of the gels was confirmed by bulk and microrheology (see the Supporting Material for comparison). Seeding of MDCK II cells involved a highly concentrated solution dropped in the middle of a hydrated gel or glass sample. For single-cell experiments, cells were dispersed over the entire dish. Samples were periodically fixed up to Day 12, stained for nuclei and actin, and imaged with an epifluorescence microscope. Details are described in the Supporting Material.On hard substrates and glass it was found previously that the area of small clusters expands exponentially until the movement of the edge cannot keep up with the proliferation in the bulk (5). Consequently, the bulk density increases toward the steady state, whereas the density of the edge remains low. At the same time, the colony size grows subexponentially (5). This is what we denote “the classical regime of growth”. Our experiments support these observations for substrates with E ≥ 5 kPa. Specifically, on glass, colonies start as small clusters of very low density of 700 ± 200 cells/mm² (Fig. 1, A and B), typically surrounded by a strong actin cable (Fig. 1, B and C). Interestingly, the spreading area of single cells (Fig. 1 A) on glass was found to be significantly larger, i.e., (2.0 ± 0.9) × 10⁻³ mm². After Day 4 (corresponding cluster area of 600 ± 100 mm²), the density in the center of the colony reached the steady state with 6,800 ± 500 cells/mm², whereas the mean density of the edge profile grew to 4,000 ± 500 cells/mm². This density was retained until Day 12 (cluster area 1800 ± 100 mm²), which is in agreement with previous work (9).Open in a separate windowFigure 1Early phase of cluster growth on hard substrates. (A) Well-spread single cells, and small clusters with a visible actin cable 6 h after seeding. (B) Within one day, clusters densify and merge, making small colonies. (C) Edge of clusters from panel B.In colonies grown on 0.6 kPa gels, however, we encounter a very different growth scenario. The average spreading area of single cells is (0.34 ± 0.3) × 10⁻³ mm², which is six times smaller than on glass substrates (Fig. 2 A). Clusters of only few cells show that cells have a preference for cell-cell contacts (a well-established flat contact zone can be seen at the cell-cell interface in Fig. 2 A) rather than for cell-substrate contacts (contact zone is diffusive and the shape of the cells appears curved). The same conclusion emerges from the fact that dropletlike agglomerates, resting on the substrate, form spontaneously (Fig. 2 A), and that attempts to seed one single cluster of 90,000 cells fail, resulting in a number of three-dimensional colonies (Fig. 2 A). When the contact area with the substrate exceeds 4.7 × 10⁻³ mm², a monolayer appears in the center of such colonies (Fig. 2 B). The colonies can merge, and if individual colonies are small, the collapse into a single domain is associated with the formation of transient irregular structures (Fig. 2 B). Ultimately, large elliptical colonies (average major/minor axis of e = 1.8 ± 0.6) with a smooth edge are formed (Fig. 2 C), unlike on hard substrates where circular clusters (e = 1.06 ± 0.06) with a ragged edge comprise the characteristic phenotype.Open in a separate windowFigure 2Early phase of cluster growth on soft substrates. (A) Twelve hours after seeding, single cells remain mostly round and small. They are found as individual, or within small, three-dimensional structures (top). The latter nucleate a monolayer in their center (bottom), if the contact area with the substrate exceeds ∼5 × 10⁻³ mm². (B) Irregularly-shaped clusters appear due to merging of smaller droplets. A stable monolayer surrounded by a three-dimensional belt of densely packed cells is clearly visible, even in larger structures. (C) All colonies are recorded on Day 4.Irrespective of cluster size, in the new regime of growth, the internal structure is built of two compartments (Fig. 2 B):

• 1.The first is the edge (0.019 ± 0.05-mm wide), a three-dimensional structure of densely packed cells. This belt is a signature of the new regime because on hard substrates the edge is strictly two-dimensional (Fig. 1 C).
• 2.The other is the centrally placed monolayer with a spatially constant density that is very weakly dependent on cluster size and age (Fig. 3). The mean monolayer density is 13,000 ± 2,000 cells/mm², which is an average over 130 clusters that are up to 12 days old and have a size in the range of 10⁻³ to 10 mm², each shown by a data point in Fig. 3. This density is twice the steady-state density of the bulk tissue in the classical regime of growth.Open in a separate windowFigure 3Monolayer densities in colonies grown on 0.6 kPa substrates, as a function of the cluster size and age. Each cluster is represented by a single data point signifying its mean monolayer density. (Black lines) Bulk and (red dashed lines) edge of steady-state densities from monolayers grown on glass substrates. Error bars are omitted for clarity, but are discussed in the Supporting Material.

Until Day 4, the monolayer is very homogeneous, showing a nearly hexagonal arrangement of cells. From Day 4, however, defects start to appear in the form of small holes (typical size of (0.3 ± 0.1) × 10⁻³ mm²). These could be attributed to the extrusions of viable cells, from either the belt or areas of increased local density in the monolayer (inset in Fig. 4). This suggests that extrusions serve to release stress built in the tissue, and, as a consequence, the overall density is decreased.Open in a separate windowFigure 4Cell nuclei within the mother colony and in the neighboring archipelago of second-generation clusters grown on 0.6 kPa gels at Day 12. (Inset; scale bar = 10 μm) Scar in the tissue, a result of a cell-extrusion event. (Main image; scale bar = 100 μm) From the image of cell nuclei (left), it is clear that there are no cells within the scar, whereas the image of actin (right) shows that the cytoplasm of the cells at the edge has closed the hole.Previous reports suggest that isolated MDCK cells undergo anoikis 8 h after losing contact with their neighbors (18). However, in this case, it appears that instead of dying, the extruded cells create new colonies, which can be seen as an archipelago surrounding the mother cluster (Fig. 4). The viability of off-cast cells is further evidenced by the appearance of single cells and second-generation colonies with sizes varying over five orders of magnitude, from Day 4 until the end of the experiment, Day 12. Importantly, no morphological differences were found in the first- and second-generation colonies.In conclusion, we show what we believe to be a novel phase of growth of MDCK model tissue on soft PA gels (E = 0.6 kPa) that, to our knowledge, despite previous similar efforts (9), has not been observed before. This finding is especially interesting in the context of elasticity of real kidneys, for which a Young’s modulus has been found to be between 0.05 and 5 kPa (19,20). This coincides with the elasticity of substrates studied herein, and opens the possibility that the newly found phase of growth has a particular biological relevance. Likewise, the ability to extrude viable cells may point to a new migratory pathway regulated mechanically by the stresses in the tissue, the implication of which we hope to investigate in the future.     

3D-SIM Super-resolution of FtsZ and Its Membrane Tethers in Escherichia coli Cells

FtsZ, a bacterial homolog of eukaryotic tubulin, assembles into the Z ring required for cytokinesis. In Escherichia coli, FtsZ interacts directly with FtsA and ZipA, which tether the Z ring to the membrane. We used three-dimensional structured illumination microscopy to compare the localization patterns of FtsZ, FtsA, and ZipA at high resolution in Escherichia coli cells. We found that FtsZ localizes in patches within a ring structure, similar to the pattern observed in other species, and discovered that FtsA and ZipA mostly colocalize in similar patches. Finally, we observed similar punctate and short polymeric structures of FtsZ distributed throughout the cell after Z rings were disassembled, either as a consequence of normal cytokinesis or upon induction of an endogenous cell division inhibitor.The assembly of the bacterial tubulin FtsZ has been well studied in vitro, but the fine structure of the cytokinetic Z ring it forms in vivo is not well defined. Super-resolution microscopy methods including photoactivated localization microscopy (PALM) and three-dimensional-structured illumination microscopy (3D-SIM) have recently provided a more detailed view of Z-ring structures. Two-dimensional PALM showed that Z rings in Escherichia coli are likely composed of loosely-bundled dynamic protofilaments (1,2). Three-dimensional PALM studies of Caulobacter crescentus initially showed that Z rings were comprised of loosely bundled protofilaments forming a continuous but dynamic ring (1–3). However, a more recent high-throughput study showed that the Z rings of this bacterium are patchy or discontinuous (4), similar to Z rings of Bacillus subtilis and Staphylococcus aureus using 3D-SIM (5). Strauss et al. (5) also demonstrated that the patches in B. subtilis Z rings are highly dynamic.Assembly of the Z ring is modulated by several proteins that interact directly with FtsZ and enhance assembly or disassembly (6). For example, FtsA and ZipA promote ring assembly in E. coli by tethering it to the cytoplasmic membrane (7,8). SulA is an inhibitor of FtsZ assembly, induced only after DNA damage, which sequesters monomers of FtsZ to prevent its assembly into a Z ring (9). Our initial goals were to visualize Z rings in E. coli using 3D-SIM, and then examine whether any FtsZ polymeric structures remain after SulA induction. We also asked whether FtsA and ZipA localized in patchy patterns similar to those of FtsZ.We used a DeltaVision OMX V4 Blaze microscope (Applied Precision, GE Healthcare, Issaquah, WA) to view the high-resolution localization patterns of FtsZ in E. coli cells producing FtsZ-GFP (Fig. 1). Three-dimensional views were reconstructed using softWoRx software (Applied Precision). To rule out GFP artifacts, we also visualized native FtsZ from a wild-type strain (WM1074) by immunofluorescence (IF).Open in a separate windowFigure 1Localization of FtsZ in E. coli. (A) Cell with a Z ring labeled with FtsZ-GFP. (B) Rotated view of Z ring in panel A. (C) Cell with a Z ring labeled with DyLight 550 (Thermo Fisher Scientific, Waltham, MA). (D) Rotated view of Z ring in panel C. (B1 and D1) Three-dimensional surface intensity plots of Z rings in panels B and D, respectively. (E) A dividing cell producing FtsZ-GFP. The cell outline is shown in the schematic. (Asterisk) Focus of FtsZ localization; (open dashed ovals) filamentous structures of FtsZ. Three-dimensional surface intensity plots were created using the software ImageJ (19). Scale bars, 1 μm.Both FtsZ-GFP (Fig. 1, A, B, and B1) and IF staining for FtsZ (Fig. 1, C, D, and D1) consistently localized to patches around the ring circumference, similar to the B. subtilis and C. crescentus FtsZ patterns (4,5). Analysis of fluorescence intensities (see Fig. S1, A and B, in the Supporting Material) revealed that the majority of Z rings contain one or more gaps in which intensity decreases to background levels (82% for FtsZ-GFP and 69% for IF). Most rings had 3–5 areas of lower intensity, but only a small percentage of these areas had fluorescence below background intensity (34% for FtsZ-GFP and 21% for IF), indicating that the majority of areas with lower intensity contain at least some FtsZ.To elucidate how FtsZ transitions from a disassembled ring to a new ring, we imaged a few dividing daughter cells before they were able to form new Z rings (Fig. 1 E). Previous conventional microscopy had revealed dynamic FtsZ helical structures (10), but the resolution had been insufficient to see further details. Here, FtsZ visualized in dividing cells by 3D-SIM localized throughout as a mixture of patches and randomly-oriented short filaments (asterisk and dashed oval in Fig. 1, respectively). These structures may represent oligomeric precursors of Z ring assembly.To visualize FtsZ after Z-ring disassembly another way, we overproduced SulA, a protein that blocks FtsZ assembly. We examined E. coli cells producing FtsZ-GFP after induction of sulA expression from a pBAD33-sulA plasmid (pWM1736) with 0.2% arabinose. After 30 min of sulA induction, Z rings remained intact in most cells (Fig. 2 A and data not shown). The proportion of cellular FtsZ-GFP in the ring before and after induction of sulA was consistent with previous data (data not shown) (1,11).Open in a separate windowFigure 2Localization of FtsZ after overproduction of SulA. (A) Cell producing FtsZ-GFP after 0.2% arabinose induction of SulA for 30 min. (B) After 45 min. (B1) Magnified cell shown in panel B. (C) Cell producing native FtsZ labeled with AlexaFluor 488 (Life Technologies, Carlsbad, CA) 30 min after induction; (D) 45 min after induction. (D1) Magnified cell shown in panel D. Scale bars, 1 μm. (Asterisk) Focus of FtsZ localization; (open dashed ovals) filamentous structures of FtsZ.Notably, after 45 min of sulA induction, Z rings were gone (Fig. 2, B and B1), replaced by numerous patches and randomly-oriented short filaments (asterisk and dashed ovals in Fig. 2), similar to those observed in a dividing cell. FtsZ normally rapidly recycles from free monomers to ring-bound polymers (11), but a critical concentration of SulA reduces the pool of available FtsZ monomers, resulting in breakdown of the Z ring (9). The observed FtsZ-GFP patches and filaments are likely FtsZ polymers that disassemble before they can organize into a ring.We confirmed this result by overproducing SulA in wild-type cells and detecting FtsZ localization by IF (Fig. 2, C, D, and D1). The overall fluorescence patterns in cells producing FtsZ-GFP versus cells producing only native FtsZ were similar (Fig. 2, B1 and D1), although we observed fewer filaments with IF, perhaps because FtsZ-GFP confers slight resistance to SulA, or because the increased amount of FtsZ in FtsZ-GFP producing cells might titrate the SulA more effectively.Additionally, we wanted to observe the localization patterns of the membrane tethers FtsA and ZipA. Inasmuch as both proteins bind to the same C-terminal conserved tail of FtsZ (12–14), they would be expected to colocalize with the circumferential FtsZ patches in the Z ring. We visualized FtsA using protein fusions to mCherry and GFP (data not shown) as well as IF using a wild-type strain (WM1074) (Fig. 3 A). We found that the patchy ring pattern of FtsA localization was similar to the FtsZ pattern. ZipA also displayed a similar patchy localization in WM1074 by IF (Fig. 3 B).Open in a separate windowFigure 3Localization of FtsA (A) and ZipA (B) by IF using AlexaFluor 488. (C) FtsA-GFP ring. (D) Same cell shown in panel C with ZipA labeled with DyLight 550. (C1 and D1) Three-dimensional surface intensity plots of FtsA ring from panel C or ZipA ring from panel D, respectively. (E) Merged image of FtsA (green) and ZipA (red) from the ring shown in panels C and D. (F) Intensity plot of FtsA (green) and ZipA (red) of ring shown in panel E. The plot represents intensity across a line drawn counterclockwise from the top of the ring around the circumference, then into its lumen. Red/green intensity plot and three-dimensional surface intensity plots were created using the software ImageJ (19). Scale bar, 1 μm.To determine whether FtsA and ZipA colocalized to these patches, we used a strain producing FtsA-GFP (WM4679) for IF staining of ZipA using a red secondary antibody. FtsA-GFP (Fig. 3 C) and ZipA (Fig. 3 D) had similar patterns of fluorescence, although the three-dimensional intensity profiles (Fig. 3, C1 and D1) reveal slight differences in intensity that are also visible in a merged image (Fig. 3 E). Quantitation of fluorescence intensities around the circumference of the rings revealed that FtsA and ZipA colocalized almost completely in approximately half of the rings analyzed (Fig. 3 F, and see Fig. S2 A), whereas in the other rings there were significant differences in localization in one or more areas (see Fig. S2 B). FtsA and ZipA bind to the same C-terminal peptide of FtsZ and may compete for binding. Cooperative self-assembly of FtsA or ZipA might result in large-scale differential localization visible by 3D-SIM.In conclusion, our 3D-SIM analysis shows that the patchy localization of FtsZ is conserved in E. coli and suggests that it may be widespread among bacteria. After disassembly of the Z ring either in dividing cells or by excess levels of the cell division inhibitor SulA, FtsZ persisted as patches and short filamentous structures. This is consistent with a highly dynamic population of FtsZ monomers and oligomers outside the ring, originally observed as mobile helices in E. coli by conventional fluorescence microscopy (10) and by photoactivation single-molecule tracking (15). FtsA and ZipA, which bind to the same segment of FtsZ and tether it to the cytoplasmic membrane, usually display a similar localization pattern to FtsZ and each other, although in addition to the differences we detect by 3D-SIM, there are also likely differences that are beyond its ∼100-nm resolution limit in the X,Y plane.As proposed previously (16), gaps between FtsZ patches may be needed to accommodate a switch from a sparse Z ring to a more condensed ring, which would provide force to drive ring constriction (17). If this model is correct, the gaps should close upon ring constriction, although this may be beyond the resolution of 3D-SIM in constricted rings. Another role for patches could be to force molecular crowding of low-abundance septum synthesis proteins such as FtsI, which depend on FtsZ/FtsA/ZipA for their recruitment, into a few mobile supercomplexes.How are FtsZ polymers organized within the Z-ring patches? Recent polarized fluorescence data suggest that FtsZ polymers are oriented both axially and circumferentially within the Z ring in E. coli (18). The seemingly random orientation of the non-ring FtsZ polymeric structures we observe here supports the idea that there is no strong constraint requiring FtsZ oligomers to follow a circumferential path around the cell cylinder. The patches of FtsZ in the unperturbed E. coli Z ring likely represent randomly oriented clusters of FtsZ filaments that are associated with ZipA, FtsA, and essential septum synthesis proteins. New super-resolution microscopy methods should continue to shed light on the in vivo organization of these protein assemblies.     

Real-Time Imaging of Electrical Signals with an Infrared FDA-Approved Dye

Clinical methods used to assess the electrical activity of excitable cells are often limited by their poor spatial resolution or their invasiveness. One promising solution to this problem is to optically measure membrane potential using a voltage-sensitive dye, but thus far, none of these dyes have been available for human use. Here we report that indocyanine green (ICG), an infrared fluorescent dye with FDA approval as an intravenously administered contrast agent, is voltage-sensitive. The fluorescence of ICG can follow action potentials in artificial neurons and cultured rat neurons and cardiomyocytes. ICG also visualized electrical activity induced in living explants of rat brain. In humans, ICG labels excitable cells and is routinely visualized transdermally with high spatial resolution. As an infrared voltage-sensitive dye with a low toxicity profile that can be readily imaged in deep tissues, ICG may have significant utility for clinical and basic research applications previously intractable for potentiometric dyes.Voltage-sensitive dyes provide a way to observe cellular electrical activity without the physical limitations imposed by electrodes. Although these dyes can monitor membrane potential with a resolution of a few microns from large populations of cells (1), there are three obstacles that prevent the use of these dyes in many research settings, including clinical research:

• 1.Most voltage-sensitive dyes use visible wavelengths of light that prevent imaging of tissues beneath the skin.
• 2.Many of these dyes produce significant toxicity or off-target effects (2).
• 3.Before this report, to our knowledge, no voltage-sensitive dyes have ever been available for administration in humans, which has limited their value in biomedically focused research.

Here, we show that indocyanine green (ICG), an FDA-approved fluorescent dye routinely used in many clinical tests, is voltage-sensitive. Our initial experimental system used Xenopus laevis oocytes. Changes in the membrane potential of the cell induced by two-electrode voltage-clamp resulted in robust, consistent changes in the fluorescence of ICG (Fig. 1, inset). All data in this work was obtained from single acquisitions with no averaging of multiple images. The voltage-dependent fluorescence changes were roughly linear with respect to membrane potential and had a magnitude of ∼1.9% of the baseline fluorescence per 100 mV of membrane potential change (Fig. 1). Additionally, ICG displayed a rapid response with a primary time constant of 4 ms (see Fig. S1 in the Supporting Material), suggesting that this dye could successfully monitor action potentials.Open in a separate windowFigure 1ICG-labeled oocytes showed that ICG’s fluorescence (blue points) is roughly linearly dependent (red line, fit to data) with voltage. (Inset) Oocyte membrane potential was held at −60 mV and then pulsed to potentials ranging from −120 mV (blue) to +120 mV (red). Ex: 780 nm, Em: 818–873 nm.To test this hypothesis, we transformed our oocytes into synthetic neurons, previously dubbed “excitocytes”, by coinjecting them with cRNA of voltage-gated sodium (Na_v) and potassium channel components (3). Under suitable current-clamp conditions, excitocytes fire trains of action potentials similar to those in naturally excitable cells. ICG’s fluorescence clearly recapitulated action potentials firing at speeds above 100 Hz (Fig. 2 A), faster than the physiological firing rates of most neurons (4).Open in a separate windowFigure 2ICG can monitor action potentials. (A) Oocytes coinjected with voltage-gated sodium and potassium channel cRNA fired action potentials (bottom, green) when held under current clamp. ICG fluorescence changes (top, blue) detected these action potentials at a rate of 107 Hz. Stimulus start (black arrow) and end (red arrow) are shown. (B and C) ICG fluorescence (blue, inverted) distinguished between healthy action potentials from wild-type sodium channels (B, green) and diseased action potentials from sodium channels with a myotonic substitution (C, green). Cells are stimulated for the entire time course of these panels. The delay between action potentials and the ICG signal is due to a low-pass filtering effect caused by the dye response time and the camera integration time. (D) In cells with myotonic sodium channels, a brief stimulus (top, black) was sufficient to elicit a train of action potentials (bottom, green) that only ceased upon significant hyperpolarization, as expected in a myotonia. ICG fluorescence (middle, blue) successfully followed each one of these action potentials.We extended the excitocyte technique from wild-type channels to evaluation of channelopathies and their effects on excitability to determine whether ICG could discriminate between normal and diseased action potentials based on shape. We compared excitocytes injected with wild-type Na_v channel cRNA to those injected with cRNA coding for a version of Na_v channel containing a point mutation, G1306E, which produces episodic myotonia (5). This disease is characterized by continued action potential firing in skeletal muscles after cessation of voluntary stimuli; the resulting prolonged muscle contractions are the hallmark of myotonia. Compared to the wild-type Na_v channel, the G1306E mutation causes a slowing of the fast inactivation of the Na_v channels, which in turn results in broadened action potentials (5). The electrical recordings and the ICG fluorescence response clearly distinguished the sharp action potentials produced by the healthy sodium channel (Fig. 2 B, and see Fig. S2) from the wider peaks produced by the myotonic sodium channel (Fig. 2 C, and see Fig. S2). Furthermore, a brief injection of current led to repetitive firing and hyperexcitability that persisted after the stimulus was stopped. ICG fluorescence clearly resolved every action potential of this myotonia-like behavior (Fig. 2 D). The successful recreation of disease-like action potentials validates the excitocyte system as a convenient method for investigating the electrophysiological effects of channelopathies.We next investigated whether ICG’s voltage sensitivity extended to excitable mammalian tissue. This validation was critical, inasmuch as other voltage-sensitive dyes have shown promise in invertebrate preparations but had much smaller signals in mammalian cells (6). We first measured ICG fluorescence from cultured rat dorsal root ganglion neurons. Under whole-cell current clamp, we observed neurons firing in the stereotypical fashion of the nociceptive C-type fiber, and these action potentials were clearly visible in the ICG fluorescence (Fig. 3 A, and see Fig. S3). We also examined syncytia of cultured cardiomyocytes from neonatal rats (7) to further validate ICG’s utility; these cells beat spontaneously and showed changes in ICG fluorescence indicative of changes in membrane potential (Fig. 3 B). Although we cannot formally exclude the possibility that the cardiomyocytes’ physical motion produced fluorescence changes, several observations suggested that these effects were minimal (see Fig. S4). Taken together, our results in frog and rat cells confirmed that ICG voltage sensitivity was broadly applicable across a range of tissues and not confined to a particular animal or cell lineage.Open in a separate windowFigure 3ICG follows electrical activity in living mammalian tissue. (A) Rat cultured dorsal root ganglion cells under current-clamp (black arrow, pulse start; red arrow, pulse end) fired action potentials (green), that ICG fluorescence tracked (blue, inverted, low-pass-filtered at 225 Hz; blue arrow, relative fluorescence change). (B) ICG fluorescence sensed spontaneous membrane potential changes in cardiomyocyte syncytia. (C) In rat brain slices, ICG responds differently to no stimulus (black) and stimuli of increasing intensity (magenta, cyan, green, and blue, increasing amplitude; scale bar shows relative fluorescence change). Weaker stimuli traces (e.g., magenta) show complete fluorescence recovery whereas larger stimuli (e.g., blue) do not fully recover within this time course; traces are vertically offset for clarity. (D) Tetrodotoxin (TTX) reduced the ICG response to a stimulus over 12 min (green, pre-TTX; cyan, magenta, and black, increasing time post-TTX; low-pass-filtered at 40 Hz; black arrow, stimulus).Finally, we tested whether ICG voltage sensitivity could be detected in a complex tissue. Rat hippocampal slice cultures comprise a well-described organotypic preparation in which the three-dimensional architecture, neuronal connections, and glial interactions are maintained (8,9). Using these rat brain explants, we found that brain excitation produced by field electrode stimulation was clearly accompanied by ICG fluorescence changes (see Fig. S5). Additionally, ICG discriminated between electrical responses caused by differing excitation intensities and durations (Fig. 3 C, and see Fig. S5). To confirm that the fluorescence changes originated from changes in excitable cell activity, we used the Na_v channel blocker tetrodotoxin (TTX). When applied to brain slices, electrical excitability was clearly inhibited (Fig. 3 D, and see Fig. S5) and partial recovery was observed upon subsequent TTX removal (see Fig. S5). These signals measuring brain slice activity were similar in shape and magnitude to those reported using other voltage-sensitive dyes (10,11). This demonstrates that ICG can report on electrical activity even in a physiological architecture with many nonexcitable cells.To our knowledge, this is the first report that a clinically approved fluorescent dye is voltage-sensitive. Our results demonstrate that indocyanine green can accurately detect action potentials at firing rates common in mammalian neurons, and that it is sensitive enough to distinguish between healthy and diseased action potentials in a model system. ICG can measure electrical activity in mammalian neurons, cardiomyocytes, and explanted brain tissue. This voltage sensitivity was observed with both monochromatic and broad-band illumination sources (data not shown), under labeling conditions that differed in solution composition, duration, and dye concentration (see Methods in the Supporting Material), and at temperatures ranging from 19°C to 30°C. ICG’s water solubility further extends its potential utility. This robustness suggests that ICG can be used to measure voltage in many environments and tissues.ICG has been FDA-approved for use in ophthalmic angiography, as well as in tests of cardiac output and hepatic function (12) and is additionally used off-label in a number of surgical applications (13). Interestingly, ICG has been shown to clearly label retinal ganglion cells in human patients (see Fig. S6) (14). This provides immediate motivation for biomedical investigations, because laboratory findings with ICG can potentially be translated to humans. Although many other voltage-sensitive dyes have been described, some with similar structures to ICG (15) and others with faster or larger signals (15,16), as of this writing none of these are FDA-approved. Additionally, ICG utilizes wavelengths further into the infrared spectrum than other available fast potentiometric dyes (17) and can thus be imaged in tissues up to 2 cm deep (18). This presents the possibility of optically imaging electrical activity deeper inside tissues than is feasible today. Although two-photon excitation with voltage-sensitive dyes can improve imaging depth, it remains intrinsically limited by the unaffected emission wavelength (19). Finally, ICG has been used in patients for more than 50 years and is known to have low toxicity (18,20). These properties suggest that ICG voltage sensitivity could extend the capabilities of modern electrophysiological techniques for disease diagnosis and monitoring in the clinic, and allow for the investigation of previously inaccessible experimental systems in basic research.     

Dendritic cells are stressed out in tumor

A recently paper published in Cell reports that dendritic cells (DCs) are dysfunctional in the tumor environment. Tumor impairs DC function through induction of endoplasmic reticulum stress response and subsequent disruption of lipid metabolic homeostasis.Tumors develop diverse strategies to escape tumor-specific immunity. Tumor-infiltrating dendritic cells (tDCs) are dysfunctional and/or mediate immune suppression¹. Cubillos-Ruiz et al.² showed in their recent Cell paper that tDCs exhibit an activation of the unfolded protein response (UPR), as indicated by the presence of high levels of spliced XPB1, and this may be attributed to reactive oxygen species (ROS) in tumor, which induces lipid peroxidation, leading to the endoplasmic reticulum (ER) stress in tDCs. Furthermore, they demonstrated that UPR activation in tDCs results in poor DC function, which is accompanied by impaired lipid metabolism and subsequent reduction of T cell anti-tumor immunity. Thus, these observations present a novel mechanism for tDC malfunction.ER stress is evoked by the presence of unfolded or chemically modified proteins. In short, the presence of damaged proteins is sensed by the proteins in the ER membrane. Of these, IRE1α can remove a short nucleotide sequence from mRNA encoding XBP1 protein. This splicing event facilitates XBP1 translation. XBP1 protein binds to consensus sequences in target genes and activates their expression³. Many XBP1 target genes are fatty acid synthesis enzymes. Enhanced production of fatty acids leads to the formation of lipid droplets inside the cytoplasm and extension of the ER compartment due to efficient intracellular membrane formation³. Therefore, this mechanism is a form of adaptation of the cell to the harsh environment, which sustains the production of functional proteins. In such a context, the article by Cuillos-Ruiz et al.² shows intriguing data on the XBP1 pathway in silencing DC function in the tumor environment.First, Cubillos-Ruiz et al.² observed that tDCs express high levels of spliced XBP1, its direct target genes and other markers for the ER stress response. Targeted deletion of XBP1 in CD11c⁺ DCs reveals an association of the XBP1-dependent ER stress response with immunosuppressive properties of tDCs. DC-specific deletion of XBP1 not only inhibits tumor growth and prolongs animal survival, but also reduces tumor peritoneal metastasis, ascites accumulation, and splenomegaly in an ovarian cancer-bearing mouse model.Next, the authors elucidated why the XBP1 pathway in tDCs is highly activated. Unexpectedly, neither typical tumor-associated cytokines nor hypoxia can efficiently stimulate XBP1 activation. Interestingly, tDCs contain high levels of lipid peroxidation byproducts bound to the proteins, which is associated with the production of ROS. Microarray analysis revealed that DC-specific XBP1 deletion downregulates both UPR pathway-dependent genes and lipid metabolism, which results in lower total lipid production, loss of lipid droplets in the cytoplasm and decreased production of triacylglycerides. This phenotype can be recapitulated by chemical inhibition of ROS formation or IRE1α and XBP1 signaling. Furthermore, XBP1-deficient DCs are potent stimulators of OT-1 T cells. Adoptive transfer of T cells isolated from metastatic tumor-bearing mice with DC-specific XBP1 deletion also shows their superior ability to control tumor growth.On the basis of these observations, the authors tested the effects of therapeutic intervention with the usage of nanocomplexes containing XBP1 siRNA. The size of lipid particles or nanocomplexes determines the anatomical location of specific drug delivery⁴. Additionally, they previously optimized the nanocomplexes for selective engulfing by DCs⁵. They found that administration of the nanoparticles containing XBP1 siRNA causes potent T cell activation, which is accompanied by reduced cancer metastatic foci and improved animal survival.The paper by Cubillos-Ruiz et al.² provides an important insight into DC biology in general. Since the identification of Toll-like receptor (TLR) signaling, the main direction in DC biology has been associated with PAMP and DAMP recognition in various degrees (Figure 1). However, the link between cellular metabolism in specific microenvironment and DC biology is poorly explored (Figure 1). The ER stress response has previously been observed in DCs^6,7. It is thought that in response to ER stress, XBP1 is crucial for DC generation, survival, and function^6,7. The conceptual link between XBP1 signaling and DC biology is as follows⁸: (i) stimulation of DCs leads to ER stress; (ii) ER stress activates UPR/XBP1 pathway; (iii) XBP1 activates lipid synthesis genes; (iv) lipids are used for the extension of the ER and Golgi compartment, which are of importance for cytokine production and secretion^9,10. Now, the authors challenged this concept in the context of cancer. The authors demonstrated that XBP1 signaling is strongly associated with poor T cell activation and limitation of XBP1 signaling leads to improved T cell function in the tumor environment (Figure 1). Thus, the paper sheds a new light on DC biology.Open in a separate windowFigure 1Double-faced role of XBP1 signaling pathway in DCs. Left: under immunostimulatory conditions, DCs receive signals from Toll-like receptors. NF-κB signaling induces a XBP1-dependent ER stress response, which enhances lipid metabolism. Formation of new membranes expands ER and Golgi compartments, which enhances cytokine production and secretion⁷. Right: In the tumor microenvironment, DCs are exposed to ROS, which also results in ER stress. However, in this case DC lipid metabolism is impaired and DCs acquire an immunosuppressive phenotype².Of course, as with any interesting work, the article raises more questions than answers. For example, which immune-related properties of DCs are “selectively” affected by the XBP1 pathway? Is there an actual cause-and-effect relationship between aberrant lipid accumulation and the immunosuppressive phenotype of tDCs? The authors observed no obvious change in PD-L1 (B7-H1) expression in DCs, but noticed reduced surface levels of peptide-loaded MHC-I complexes in wild-type tDCs as compared to XBP-deficient tDCs. However, it remains unknown whether and how DC cross-presentation is involved in tDC function regulated by ER stress, XBP1 activation, and lipid metabolism. Human ovarian cancer-associated DCs express high levels of B7-H1 and limited IL-12¹¹. It would be interesting to thoroughly examine the cytokine profile, and the B7 and TNF family members, along with lipid pathway manipulation in XBP1^−/− DCs. It is well known that ER morphology and function is substantial for the synthesis of membrane proteins and cytokines, which would be affected by the XBP1 pathway. Paradoxically, a restricted ER stress response can help immune reaction, which requires the expansion of the ER compartment. Another question is how and which lipid synthesis and metabolism pathway is targeted by XBP1 in DCs (or/and tumor cells in the same environment). Obviously, future studies are warranted to address these important questions.In conclusion, the paper by Cubillos-Ruiz et al.² opens an interesting chapter for scientifically and therapeutically exploring DC biology in a specific metabolic environment. Given that silencing XBP1 signaling in DCs enhances tumor immunity, and XBP1 is an intrinsic pro-tumor factor, it is reasonable to assume that targeting this pathway may be beneficial in patients with cancer and could kill two birds with one stone.     

ER as master regulator of membrane trafficking and organelle function

The endoplasmic reticulum (ER), which occupies a large portion of the cytoplasm, is the cell’s main site for the biosynthesis of lipids and carbohydrate conjugates, and it is essential for folding, assembly, and biosynthetic transport of secreted proteins and integral membrane proteins. The discovery of abundant membrane contact sites (MCSs) between the ER and other membrane compartments has revealed that, in addition to its biosynthetic and secretory functions, the ER plays key roles in the regulation of organelle dynamics and functions. In this review, we will discuss how the ER regulates endosomes, lysosomes, autophagosomes, mitochondria, peroxisomes, and the Golgi apparatus via MCSs. Such regulation occurs via lipid and Ca²⁺ transfer and also via control of in trans dephosphorylation reactions and organelle motility, positioning, fusion, and fission. The diverse controls of other organelles via MCSs manifest the ER as master regulator of organelle biology.

IntroductionThe endoplasmic reticulum (ER) is the cell’s largest organelle and Ca²⁺ reservoir with well-characterized roles in the biosynthesis of lipids, proteins, and glycoconjugates. The more recent discoveries of membrane contact sites (MCSs) between the ER and other organelles have revealed that the functions of the ER go far beyond biosynthesis. Here, we will review these “non-traditional” functions of the ER. Since the interplays between the ER and the plasma membrane and lipid droplets have been extensively reviewed (Crul and Maleth, 2021; Renne and Hariri, 2021), we will focus on the interactions between the ER and intracellular organelles with emphasis on molecular mechanisms that control membrane trafficking and organelle function.ER in control of endosomesThe endocytic pathway consists of numerous endocytic vesicles, endosomes, and lysosomes that receive the material taken up from the cell surface via endocytosis, including cargos such as nutrient receptors and activated growth factor and hormone receptors. Endocytic vesicles derived from the plasma membrane fuse with early endosomes, which mature and change their molecular composition as they move toward the cell interior guided by dynein-dependent transport along microtubules. As endosomes mature, they become gradually more acidic and acquire hydrolytic enzymes supplied by fusion with Golgi-derived vesicles. Finally, the resulting late endosomes fuse with lysosomes and their cargo is degraded (Huotari and Helenius, 2011; Scott et al., 2014). Although the endocytic and biosynthetic pathways have traditionally been considered to be highly separate, recent studies have revealed surprising connections between the ER and endosomes (Fig. 1).Open in a separate windowFigure 1.ER-mediated control of endosome dynamics. Overview of cell biological functions of ER-endosome contact sites and the involved molecules. The molecular composition of ER–endosome contact sites. OSBP, ORPs, and VAPs function as dimers or multimers. For simplicity, this is not displayed in the figure. VAP family members (see text box) are depicted as “VAP.” (1) Perinuclear vesicle tethering: The E2 ubiquitin-conjugating enzyme UBE2J1 activates the E3 ubiquitin ligase RNF26, which then ubiquitinates SQSTM1/p62. Ubiquitinated SQSTM1/p62 in turn binds to organelle-specific adaptor proteins, such as T6BP/TAX1BP1, on TGN vesicles and EPS15B or TOLLIP on endosomes. The release of the tethered vesicles is mediated by the deubiquitinase USP15. (2) Endosome translocation. (2a) The BORC complex recruits the small GTPase ARL8B to endosomes, which in turn recruits and activates the Kinesin-1 adaptor protein SKIP/PLEKHM2, resulting in plus-end directed movement of endosomes and lysosomes. Upon ER stress, IRE1 inhibits BORC-dependent anterograde endosome translocation. (2b) The ER-resident protein Protrudin contacts endosomes by binding to RAB7 and PtdIns3P. At these contact sites, Protrudin mediates the hand-over of Kinesin-1 to the endosomal adaptor protein FYCO1, allowing plus-end translocation of endosomes along microtubules. The activity of Protrudin can be regulated by CPT1C, which promotes anterograde endosome transport under nutrient-rich conditions and blocks it under cellular stress conditions. PDZD8 interacts with Protrudin and RAB7, also mediating ER-endosome contact. In addition, PDZD8 might mediate contact with mitochondria. (2c) Endosomes containing high levels of cholesterol move along microtubules in the minus-end direction by dynein/dynactin motor proteins, which connect to the endosome through RILP, RAB7, and ORP1L. Under low concentrations of cholesterol, ORP1L makes contact with VAP in the ER, which leads to the dissociation of dynein/dynactin and the HOPS complex. ER-endosome contact enables ORP1L to transfer cholesterol from the ER to endosomes. Sufficient levels of endosomal cholesterol are a prerequisite for ILV formation (see also legend to 4b). (3) Shaping of endosomal tubules: The formation of recycling tubules requires transient accumulation of PtdIns4P on endosomes to allow WASH-dependent actin nucleation and retromer function. OSBP interacts with PtdIns4P on endosomes via its PH domain and tethers endosomes to the ER via interaction with VAP. PtdIns4P is then dephosphorylated by the ER-resident lipid phosphatase SAC1, securing a transient PtdIns4P pool on endosomes. WASH is linked to the retromer by its subunit FAM21, which marks the site of tubule scission. The PtdIns3P-binding retromer subunit SNX2 is also able to interact with the ER through VAP. The ER protein TMCC1 and Coronin 1C on endosomes are required for contact site formation and fission of WASH-containing endosome tubules. It is not known whether Coronin 1C and TMCC1 interact directly, or if there are additional proteins required to generate these membrane contact sites. (4) Receptor dephosphorylation, ILV formation, and cholesterol transfer. (4a) EGFR-induced phosphorylation of Annexin A1 induces the formation of Annexin A1/S100A11-mediated ER-endosome contact sites, aided by the local increase in Ca²⁺ through the endosomal Ca²⁺ channel TPC1. PTP1B in the ER dephosphorylates EGFRs and ESCRT-0, facilitating the sorting of EGFRs into forming ILVs. (4b) In addition, Annexin A1/S100A11-mediated ER-endosome contact sites facilitate cholesterol transfer from ER to forming ILVs by ORP1L (see also legend to 2c). (4c) STARD3 and its paralog STARD3NL (not shown) mediate cholesterol transfer from ER to EGFR-negative endosomes. ORP5 facilitates cholesterol transport from endosomal membranes to the ER. The cholesterol is provided by NPC2 and NPC1, which interacts with ORP5, forming an ER–endosome contact. Direct shuttling of sterols using the ORD domain of ORP5 remains to be confirmed (Santos et al., 2020).Perinuclear retention of endosomesThe bulk of endosomes and lysosomes exhibit a perinuclear localization clustered around the microtubule-organizing center, together with vesicles of the trans-Golgi network (TGN). This localization enables efficient endosome maturation and cargo trafficking, important for endocytic pathway functions including nutrient uptake, receptor downregulation and cell signaling, host defense against pathogens, and control of cell polarity and cell migration (Alanko et al., 2016; Huotari and Helenius, 2011; Scott et al., 2014). Although the perinuclear clustering of endosomes and Golgi vesicles has been observed for decades, how they are organized and retained was not understood until recently. The ER plays a direct role in the maintenance of this endosomal architecture, orchestrated by two ER-resident ubiquitination enzymes (Cremer et al., 2021; Jongsma et al., 2016; Fig. 1, 1). The E2 ubiquitin conjugation enzyme UBE2J1 interacts with and activates the multimembrane spanning RING domain E3 ubiquitin ligase RNF26. This induces the recruitment and ubiquitination of SQSTM1/p62, a cytosolic ubiquitin adapter best known for its role in selective autophagy. Ubiquitinated SQSTM1/p62 in turn interacts with many ubiquitin-binding organelle-specific adaptor proteins, including T6BP/TAX1BP1 at the TGN (Morriswood et al., 2007) and EPS15B or TOLLIP on endosomes (Katoh et al., 2004; Roxrud et al., 2008). The localization of the E2/E3 pair UBE2J1/RNF26 is confined to the perinuclear ER, which ensures the perinuclear retention of vesicles until released. The ubiquitin-dependent vesicle tethering is released by the deubiquitination enzyme USP15, which is recruited by RNF26 (Jongsma et al., 2016). Although the perinuclear retention of RNF26 depends on its RING domain, it is not known how this mechanism is regulated or how it is coordinated with mechanisms that translocate vesicles to the cell periphery. Dysregulation of ER-UBE2J1/RNF26-mediated vesicle tethering leads to the increased half-life of phosphorylated epidermal growth factor receptors (EGFR) accompanied by prolonged AKT-S473 phosphorylation due to impaired endocytic downregulation (Cremer et al., 2021).Regulation of endosome translocation to the cell peripheryThe nutritional status as well as cellular stress responses influence how endosomes are positioned and utilized for cellular functions (Korolchuk et al., 2011; Raiborg, 2018). During stress and low nutrient conditions, endosomes and lysosomes cluster perinuclearly to facilitate cargo degradation for nutrient supply. When nutrients are available and in the absence of cellular stress, motile endosomes engage in a variety of cellular processes to support cell growth and development. The motile and dispersed late endosomes are less acidic and contain lesser hydrolytic enzymes than the perinuclear late endosomes (Johnson et al., 2016), consistent with their role in functions other than cargo degradation. Indeed, although some endosomes recycle cargo back to the cell surface, others are engaged in plasma membrane repair, protrusion formation, mTORC1 signaling, or secretion of exosomes (Ballabio and Bonifacino, 2020; Pu et al., 2016). Importantly many of these responses are coordinated through the ER. There are two established mechanisms that facilitate the centrifugal transport of late endosomes: the protrudin-dependent pathway, whose function depends on ER-resident proteins, and the BORC-dependent pathway, which is inhibited by ER stress (see below). Thus, the ER is a master regulator of endosome positioning through the control of mechanisms that promote their perinuclear or peripheral localization.Inhibition of BORC-dependent endosome translocation upon cellular stressThe eight-subunit protein BLOC-one-related complex (BORC) localizes to late endosomes. When nutrient supplies are rich, BORC recruits the small GTPase ARL8B, which through its effector SKIP/PLEKHM2 engages the plus-end-directed microtubule motor Kinesin-1, thus promoting late endosome translocation to the cell periphery (Fig. 1, 2 a). This mechanism is important for cell migration and axonal growth (Farías et al., 2017; Pu et al., 2015). Under cellular stress conditions, however, this pathway is turned off. When cells are deprived of amino acids and growth factors, the BORC complex binds to the endosomal Ragulator complex, making it unable to engage Kinesin-1 (Filipek et al., 2017; Pu et al., 2017). In addition, the ER-resident transmembrane nuclease inositol requiring enzyme 1 (IRE1) plays a role in shutting off BORC-dependent endosome translocation. One branch of the unfolded protein response triggered by ER stress goes through the activation of IRE1. Once activated, IRE1 cleaves and initiates the degradation of certain mRNAs, including the mRNA encoding Blos1, a subunit of the BORC complex. Thus, endosomes cluster perinuclearly, facilitating the lysosomal degradation and clearance of ubiquitinated protein aggregates by microautophagy during ER stress (Bae et al., 2019).Protrudin-mediated endosome translocationProtrudin is a transmembrane ER-resident protein that induces ER–endosome MCSs by binding to the late endosomal small GTPase RAB7 in combination with the endosomally enriched lipid phosphatidylinositol 3-phosphate (PtdIns3P; see text box for RAB GTPases and phosphoinositides). In such MCSs, Protrudin hands over Kinesin-1 to the endosomal adapter protein FYCO1, which also interacts with RAB7 and PtdIns3P. This facilitates the translocation of late endosomes along microtubules to the plasma membrane (Raiborg et al., 2015a; Fig. 1, 2 b). The ER-resident pseudoenzyme carnitine palmitoyltransferase 1C (CPT1C) is found in a complex with Protrudin and functions as a nutrient sensor. Under glucose-rich conditions, malonyl-CoA binds CPT1C, and this activates Protrudin-mediated Kinesin-1 handover, which is inhibited upon cellular stress by signaling from the 5'' AMP-activated protein kinase (AMPK; Palomo-Guerrero et al., 2019). It is not clear if the seemingly parallel BORC and Protrudin pathways are redundant. As they depend on different small GTPases, ARL8B and RAB7, respectively, they likely translocate different subpopulations of late endosomes (Jongsma et al., 2020). The Protrudin pathway is important for the formation of cellular protrusions like neurites or invadopodia, and this requires that endosomes fuse with the plasma membrane in a Synaptotagmin-VII-dependent manner (Palomo-Guerrero et al., 2019; Pedersen et al., 2020; Raiborg et al., 2015a; Shirane and Nakayama, 2006). In addition, the endosomes contain cargo, such as the metalloprotease MT1-MMP, and the overexpression of Protrudin increases the cell’s invasive behavior by facilitating exocytosis of MT1-MMP in growing invadopodia (Pedersen et al., 2020). Moreover, the Protrudin pathway facilitates mTORC1 signaling from late endosomes (Hong et al., 2017) and stimulates angiogenesis (Arora et al., 2022) and axon regeneration (Petrova et al., 2020).The vesicle-associated membrane protein-associated protein (VAP) family consists of five dimeric transmembrane ER proteins that contain a major sperm protein (MSP) domain, which binds FFAT (two phenylalanines in an acidic tract) or FFNT (two phenylalanines in a neutral tract) motifs present in proteins on the membranes of various organelles to form MCSs (Cabukusta et al., 2020; James and Kehlenbach, 2021; Loewen and Levine, 2005). Mammalian VAPs include the FFAT-binding VAP-A, VAP-B, and MOSPD2, and the FFNT-binding MOSPD1 and MOSPD3.RAB GTPases are small GTPases of the RAS superfamily, which act as molecular switches that are active in the GTP-bound form and inactive in the GDP-bound form (Stenmark, 2009). In their active conformation, RAB GTPases control membrane dynamics and intracellular transport by binding various effector proteins, including vesicle tethers, enzymes, and motor adaptors. Almost 70 different mammalian RAB GTPases have been identified, and they are known to associate with specific membranes such as the Golgi (RAB6), early endosomes (RAB5), or late endosomes/lysosomes (RAB7). Membrane association is mediated via C-terminal isoprenoid groups.Phosphoinositides (PIs) are phosphorylated derivatives of the abundant membrane phospholipid, phosphatidylinositol (PtdIns; Schink et al., 2016). Seven PIs exist in nature – PtdIns3P, PtdIns4P, PtdIns5P, PtdIns(3,4)P₂, PtdIns(3,5)P₂, PtdIns(4,5)P₂, and PtdIns(3,4,5)P₃, with numbers indicating the positions of phosphates in the inositol headgroup. PtdIns3P, PtdIns4P, and PtdIns(4,5)P₂ have been implicated in MCS formation and dynamics. Phosphorylations of the headgroup are mediated by isoform-specific PI kinases whereas dephosphorylations are catalyzed by specific PI phosphatases.ORP (oxysterol binding protein-related protein) is a family of proteins that has the capacity to bind and transfer sterols and phosphoinositides (Nakatsu and Kawasaki, 2021). ORPs are characterized by an OSBP-related domain, ORD, which contains a hydrophobic sterol binding pocket. Most ORPs also contain phosphoinositide-binding pleckstrin homology (PH) domains and FFAT motifs, which mediate their localization and functions at MCSs.Synchronization of endosome translocation and lipid transferLipid transfer between closely opposed organelles is mediated by lipid transfer proteins (Reinisch and Prinz, 2021). The ER-resident PDZ domain containing protein 8 (PDZD8) harbors lipid transfer activity and transfers glycerophospholipids and ceramide between membranes in vitro by the use of its synaptotagmin-like mitochondrial-lipid-binding (SMP) domain (Gao et al., 2022; Shirane et al., 2020). In vivo, the depletion of PDZD8 results in a decrease in the abundance of phosphatidylserine (PS) in neuronal endosomes (Shirane et al., 2020) and the accumulation of endosomal PtdIns(4,5)P₂ (Jeyasimman et al., 2021). PDZD8 is important for endosome maturation and their degradative capacity, neuronal integrity, and neurite outgrowth (Gao et al., 2022; Jeyasimman et al., 2021; Shirane et al., 2020).PDZD8 mediates ER–endosome contact sites by binding to RAB7 and interacts with Protrudin via its transmembrane domain (Elbaz-Alon et al., 2020; Gao et al., 2022; Guillén-Samander et al., 2019; Khan et al., 2021; Shirane et al., 2020; Fig. 1, 2 b). The potential functional relationship between Protrudin and PDZD8 is not completely understood. Since both proteins can form ER–endosome contact sites, why would they need to interact? It is tempting to speculate that these proteins cooperate in the regulation of endosome maturation and translocation, PDZD8, by mediating lipid transfer, and Protrudin by providing a microtubule motor protein. Thus, endosome maturation, function, and translocation can be coordinated efficiently by the ER. This might be especially important in neurons, which depend heavily on endosomal trafficking for their function.Coordination of microtubule-mediated retrograde and anterograde endosome transportEndosome positioning entails a constant balance between minus- and plus-end-directed transport along microtubules, mediated by dynein or kinesins, respectively (Bonifacino and Neefjes, 2017; Gennerich and Vale, 2009). With its widespread connection to endosomes, the ER constitutes a unique platform for the organization of the required motor proteins (Friedman et al., 2013). One such possible coordination point centers on the ER-resident protein VAP-A (see text box). Despite being a transmembrane ER protein, Protrudin harbors a VAP-binding FFAT motif, and VAP-A is important for the proper distribution of Protrudin in the ER and for the function of Protrudin in protrusion formation, suggesting that VAP-A facilitates Kinesin-1-dependent endosome translocation (Saita et al., 2009; Fig. 1, 2 b). VAP-A is also implicated in the loss of dynein from endosomal membranes. The dynein binding endosomal protein RILP forms a tripartite complex with RAB7 and the endosomal cholesterol sensor ORP1L, a member of the ORP family (see text box). Under low endosomal cholesterol concentration, the endosomes become tethered to the ER by ORP1L binding to VAP-A, leading to the dissociation of dynein from RILP (Rocha et al., 2009; Fig. 1, 2 c). Thus, although not yet experimentally verified, it is conceivable that VAP-A sites in the ER coordinate the loss of endosomal dynein with the gain of Kinesin-1 through ORP1L-RILP and Protrudin.Another clue to the role of ER as a coordinator of endosomal motor protein switching comes from the association between Protrudin and the long M1 isoform of the microtubule-severing AAA-ATPase, Spastin. Spastin interacts with Protrudin in the ER and inhibits Protrudin-dependent polarized membrane traffic (Connell et al., 2020). The inhibitory effect of Spastin on endosome translocation requires its ability to interact with the endosomal-sorting complex required for transport (ESCRT)-III proteins, IST1 and CHMP1B, in addition to its microtubule severing-activity. Although not completely understood, this effect might be related to the role of Spastin in the fission of endosomal recycling tubules, which requires the same functional properties as Spastin (Allison et al., 2013). The recruitment of dynein to Spastin-induced microtubule plus ends (Fassier et al., 2013; Lenz et al., 2006; Riano et al., 2009; Zhang et al., 2003) likely counteracts the Protrudin-mediated Kinesin-1-dependent movement of endosomes on microtubule rails toward the cell periphery (Wassmer et al., 2009). The interaction between Spastin and Protrudin in the ER could ensure that the microtubule severing is positioned in close proximity to Protrudin. Thus, the ER coordinates the recruitment of dynein and Kinesin-1 via Spastin and Protrudin, respectively.Shaping of endosomal tubulesEndocytic cargo that is not destined for lysosomal degradation is sorted into endosomal membrane tubules for recycling back to the plasma membrane or to the Golgi (Huotari and Helenius, 2011; Scott et al., 2014). This process involves membrane budding, tubule extension, and fission to generate cargo-containing vesicles. The ER appears to control both endosomal tubule formation and fission, involving different types of ER–endosome contact sites.The formation of endosomal recycling tubules requires actin polymerization by the WASH complex, which is coupled to the cargo-sorting retromer machinery by its subunit FAM21 (Derivery et al., 2009; Gomez and Billadeau, 2009; Harbour et al., 2012; Puthenveedu et al., 2010). The transient accumulation of PtdIns4P on endosomes is coupled to a transient burst of WASH-dependent actin nucleation to facilitate retromer function, and the ER is the master regulator of these dynamics (Dong et al., 2016). A type II PI 4-kinase localizes to the WASH complex and produces a local pool of PtdIns4P (Ryder et al., 2013). Endosomal OSBP interacts with PtdIns4P via its PH domain and tethers the endosome to the ER by interacting with VAP-A/B (Fig. 1 3). Here, OSBP transfers PtdIns4P to the ER-resident lipid-phosphatase SAC1, which dephosphorylates PtdIns4P, ensuring the transient PtdIns4P pool on the endosome required for tubule dynamics. In addition, the PtdIns3P binding retromer subunit SNX2 interacts with the ER through VAP-A/B. As actin nucleation by WASH is tightly coupled to retromer-dependent cargo sorting, the ER presumably coordinates their activities through the interaction between VAP-A/B (ER), SNX2 (retromer), and OSBP/PtdIns4P (WASH), all of which localize to the same intracellular hotspots. When this mechanism is perturbed by the depletion of VAP-A/B, SNX2, or OSBP, both PtdIns4P and actin hyper-accumulate on endosomes, and the traffic between endosomes and the Golgi complex is disrupted (Dong et al., 2016). Thus, by regulating endosomal PtdIns4P levels, the ER affects WASH-dependent actin nucleation and retromer function; however, it remains to be seen how PtdIns4P mechanistically interacts with WASH activity.In addition to regulating endosomal actin dynamics, the ER defines the position and timing of endosome fission (Hoyer et al., 2018; Rowland et al., 2014). Immediately prior to fission, contact sites are formed between ER tubules and endosome buds on sites marked by the WASH component FAM21. The organelles are tethered by the ER membrane protein TMCC1 and endosomal Coronin1C, which is connected to actin on the endosomal buds (Fig. 1 3). Both proteins are required for contact site formation and fission of WASH-containing endosome tubules. Depletion of TMCC1 disrupts recycling of the CI-MPR from endosomes to the Golgi to a similar extent as the depletion of FAM21 (WASH) or VPS35 (retromer), emphasizing the role of ER in this process (Hoyer et al., 2018). Coronin1C confines the localization of actin to bud necks, thereby defining membrane availability for ER–endosome contact sites (Striepen and Voeltz, 2022). How the ER promotes fission is, however, not understood. It will be important to investigate a possible connection with the PtdIns4P-regulated mechanism discussed above. It is tempting to speculate that the final fission step is facilitated by the ESCRT-III-related proteins, IST1 and CHMP1B, which are known to mediate positive membrane bending and constriction (Nguyen et al., 2020) and are connected to the ER by Spastin M1, which is indeed required for the fission of tubules and the recycling of endosomal cargo (Allison et al., 2013).Coordination of receptor dephosphorylation and formation of multivesicular endosomesUpon growth factor stimulation, activated growth factor receptors, such as EGFR, are internalized by endocytosis for their final degradation in lysosomes, a process referred to as receptor downregulation (Huotari and Helenius, 2011; Scott et al., 2014). This process ensures that signaling is switched off in a timely manner to prevent hyperproliferation. To attenuate EGFR signaling, the receptors are dephosphorylated and sorted into forming intraluminal vesicles (ILVs) of multivesicular endosomes (MVEs) on their way to the lysosome. It is interesting to note that EGF-stimulation itself induces this process by triggering the dephosphorylation of EGFR and at the same time stimulates ILV formation. Intriguingly, the ER is recruited to promote both tasks.The phospholipid-binding protein Annexin A1 associates with EGFR-containing MVEs, whereas its ligand S100A11 localizes to the ER (Futter et al., 1993; Gerke and Moss, 2002; Liu et al., 2012; Fig. 1, 4 a). EGFR-induced phosphorylation of Annexin A1 induces the formation of Annexin A1/S100A11-mediated ER–endosome contact sites (Eden et al., 2016). Both Annexin A1 and S100A11 are Ca²⁺ binding proteins, and the contact site formation is aided by the local increase in Ca²⁺, which is induced by the endosomal NAADP-sensitive two-pore Ca²⁺ channel TPC1 (Kilpatrick et al., 2017). These contact sites promote EGF-induced ILV formation (Eden et al., 2016; White et al., 2006; Wong et al., 2018). First, the protein tyrosine phosphatase 1B (PTP1B), which is embedded in the cytoplasmic face of the ER, dephosphorylates EGFRs on the endosomes, depending on Annexin A1/S100A11-mediated ER–endosome contact sites (Eden et al., 2010). At the same time, the EGFRs are sorted into forming ILVs by the ESCRT protein machinery, which interacts with the ubiquitinated EGFRs and mediates membrane deformation and scission to generate ILVs (Migliano et al., 2022). Interestingly, the ESCRT proteins HRS and STAM are dephosphorylated by PTP1B, implying that the Annexin A1/S100A11-mediated ER–endosome contact sites can regulate ESCRT function (Eden et al., 2010; Stuible et al., 2010). This functional relationship, which could facilitate the progression of cargo through downstream ESCRTs and ILV formation, needs further investigation.In addition to acting on the ESCRT machinery through PTP1B, the Annexin A1/S100A11-mediated ER–endosome contact sites can facilitate ILV formation through a different mechanism. ILVs are rich in cholesterol, and high levels of endosomal cholesterol are required to form ILVs (Möbius et al., 2003). Cholesterol can be supplied by the uptake of low-density lipoprotein (LDL) by receptor-mediated endocytosis (Anderson et al., 1977). To fuel ILV formation in the absence of LDL, cholesterol needs to come from internal sources such as the ER. When endosomal cholesterol levels are low, EGF-stimulated ILV formation depends on Annexin A1/S100A11-mediated ER–endosome contacts (Eden et al., 2016). As Annexin A1/S100A11 does not harbor sterol transfer properties, such delivery has to be coordinated with a lipid transfer protein. The endosomal cholesterol sensor ORP1L localizes to Annexin A1-dependent ER–endosome contact sites in the absence of LDL and is a plausible candidate for this activity. When endosomal cholesterol levels are low, a conformational change exposes the ORP1L FFAT-motif, inducing binding to VAP-A in the ER (Rocha et al., 2009; Fig.1, 4 b). Here, ORP1L facilitates the transfer of cholesterol from the ER to EGFR-containing endosomes and stimulates ILV formation in a manner that requires its interaction with VAP-A (Eden et al., 2016). Whether ORP1L is directly responsible for transfer or regulates another lipid transfer protein is unresolved. It is interesting to note that under cholesterol depletion, the interaction of ORP1L with VAP-A in the ER at the same time leads to the loss of dynein and the HOPS complex from the endosomes (van der Kant et al., 2013). This will inhibit the perinuclear translocation and fusion of MVEs when cholesterol levels are low, and could thus halt the maturation of endosomes to ensure proper sorting of EGFRs into ILVs by use of cholesterol from the ER.Control of endosome maturation and homeostasisTo maintain lipid homeostasis, the ER influences the transport of lipids from the ER to endosomes and vice versa. In addition to the lipid transporters PDZD8 and ORP1L mentioned above, STARD3 resides in EGFR-negative endosomes and facilitates cholesterol transport from the ER to the endosomes, which are anchored to the ER through VAP-A/B and MOSPD2 (Alpy et al., 2013; Voilquin et al., 2019; Wilhelm et al., 2017). Conversely, NPC1 facilitates the transfer of cholesterol from endosomes to the ER by interaction with ORP5 in the ER (Fig.1, 4 c; Du et al., 2011; Raiborg et al., 2015b). The ER plays a pivotal role in maintaining endosome homeostasis and maturation by regulating endosomal identity (Wu and Voeltz, 2021), and the sorting and trafficking of hydrolytic enzymes and endocytosed proteins and lipids, as exemplified above. Mutations in NPC1 lead to the accumulation of cholesterol in endosomes, causing the neurodegenerative disease Nieman-Pick (Mukherjee and Maxfield, 2004). Accumulation of endocytosed or cellular material caused by dysfunctional ER–endosome MCS proteins can thus lead to severe metabolic and developmental defects, as manifested by genetic diseases, collectively termed lysosomal storage disorders (Platt et al., 2012).ER as source and regulator of autophagosomesMacroautophagy (hereafter, autophagy) is a catabolic process that entails sequestration of portions of cytoplasm by a double-membrane structure known as the phagophore (Fig. 2). The phagophore closes to form an autophagosome, and when the autophagosome fuses with a lysosome to form an autolysosome, the sequestered material is degraded by lysosomal hydrolases (Melia et al., 2020; Mizushima and Komatsu, 2011) The catabolic functions of autophagy are used to supply cells with amino acids and other small molecules during conditions of low nutrient availability, but autophagy is also used to protect cells from potentially harmful cytoplasmic objects such as protein aggregates, pathogens, and damaged organelles.Open in a separate windowFigure 2.Biogenesis of the phagophore membrane via ER contacts. Autophagy is initiated by sequestration of cytoplasmic material by a double-membrane phagophore, whose seed is thought to be composed of ATG9-containing vesicles originating from the Golgi. The phagophore elongates and closes to form an autophagosome, and the sequestered material is degraded once the autophagosome fuses with a lysosome. Phagophore elongation is promoted by a flux of lipids from the ER to the phagophore membrane via the lipid channel transporter ATG2, which tethers subdomains of the ER to growing phagophores by interaction with the ER-localized lipid scramblases TMEM41B and VMP1, and the lipid scramblase ATG9 in the phagophore membrane (additional contacts between the membranes are likely). TMEM41B-VMP1 and ATG9 serve to maintain transbilayer lipid balance in the ER and phagophore membrane, respectively.Biogenesis of the phagophore membraneAlthough several cellular membranes have been proposed as the origin of phagophore membranes, there is little doubt that the ER is a major source (Lamb et al., 2013; Melia et al., 2020). The fact that autophagosome membranes, in contrast to other cellular membranes, are almost devoid of transmembrane proteins (Fengsrud et al., 2000) suggests that much of the phagophore could originate from de novo membrane synthesis rather than budding from existing membranes. In support of this, a large cytosolic protein required for autophagosome biogenesis, ATG2, is an elongated lipid transporter that contains a hydrophobic groove through which lipids can slide in an efficient way (Ghanbarpour et al., 2021; Maeda et al., 2019). ATG2 could thus function in MCSs that bridge the lipid-synthesizing ER and the forming phagophore.ATG2 interacts with two lipid scramblases in the ER membrane, TMEM41B and VMP1, and with a lipid scramblase on Golgi-derived vesicles, ATG9 (Ghanbarpour et al., 2021; Judith et al., 2019; Noda, 2021). Lipid scramblases transfer lipids from one membrane leaflet to the other, and it has been proposed that even a single ATG9-containing vesicle might act as a seed for phagophore biogenesis (Ghanbarpour et al., 2021). In this model, ATG2 mediates lipid transport from the ER membrane to the seeding vesicle, whereas TMEM41B and VMP1 re-equilibrate the leaflets of the ER during lipid extraction. In the seed vesicle, ATG9 scrambles ER-derived lipids upon their delivery to allow phagophore expansion (Fig. 2). Even though this is an attractive model that explains the requirement for lipid transporters and scramblases in autophagosome biogenesis, it still needs to be verified experimentally. Hybrid organelles consisting of membranes from endosomes and the cis-Golgi have recently been put forward as precursors of phagophores (Kumar et al., 2021), and it remains plausible that autophagosomes can originate from membranes other than the ER, at least under some conditions (Melia et al., 2020).Class III PI 3-kinase (PI3K-III), which phosphorylates PtdIns into PtdIns3P, is required for phagophore biogenesis, and it is conceivable that PtdIns3P contributes to defining the sites of phagophore initiation. Indeed, ATG14, a subunit of the autophagy-specific version of PI3K-III, localizes to ER sites, and this localization is required for autophagy (Matsunaga et al., 2010). The PtdIns3P-binding protein DFCP1 is recruited to PtdIns3P-containing ER subdomains in response to amino acid starvation, a classical way to induce autophagy, and is a likely PtdIns3P effector in autophagosome biogenesis. Due to their omega shape in light microscopy, DFCP1-containing ER subdomains are referred to as omegasomes (Axe et al., 2008). The exact spatial and functional relationships of omegasomes with phagophores are not known, but current evidence suggests that omegasomes could represent ER subdomains that are involved in the elongation and sculpting of the phagophore.PtdIns3P is not only found on ER subdomains but also on phagophore membranes (Cheng et al., 2014), suggesting the involvement of additional PtdIns3P-binding proteins in phagophore biogenesis. PtdIns3P-binding proteins of the WIPI family are good candidates as they interact with ATG2 and localize to the growing phagophore. WIPI4, which shows the highest affinity to ATG2, binds to one of the tips of ATG2, consistent with the idea that ATG2 could be recruited by WIPI4 to form a lipid-transporting bridge between the ER and the tip of the phagophore (Chowdhury et al., 2018). However, the spatiotemporal relationships between ATG2, ATG9, PtdIns3P, DFCP1, and WIPI4 during phagophore biogenesis remain to be defined.Control of autophagosome fusionAutophagy culminates in the fusion of autophagosomes with lysosomes. The membranes of late endosomes and lysosomes contain the small GTPase RAB7, and among the RAB7 effectors they recruit are ORP1L, RILP, and PLEKHM1. As described above, ORP1L is a cholesterol sensor that forms tripartite contacts with RAB7 and the dynein adaptor RILP, thereby promoting dynein-mediated transport of late endosomes and lysosomes toward the microtubule organizing center. The endolysosomal protein PLEKHM1, in concert with RILP, recruits the HOPS complex, which promotes fusion between lysosomes and autophagosomes (McEwan et al., 2015; Wijdeven et al., 2016). If the lysosomes have low cholesterol content, the FFAT motif of ORP1L is exposed and engages in interaction with VAP-A in the ER membrane. The cholesterol-free conformation of ORP1L not only prevents the interaction of RILP with dynein and HOPS, but also dissociates PLEKHM1, and this inhibits both lysosome motility toward the microtubule organizing center and fusion between lysosomes and autophagosomes (Wijdeven et al., 2016). Thus, autophagic flux is positively regulated by cholesterol and negatively controlled by the ER–lysosome MCSs. Since autolysosomes, like lysosomes, contain RAB7 and ORP1L, their motility is regulated in the same manner.Regulation of mitochondria by ERMCSs between ER and mitochondria are guided by bridging proteins that tether the two membranes. Such MCSs are important for several mechanisms, including mitochondrial homeostasis, lipid composition, nutrient sensing, and regulation of the apoptotic machinery. The MCSs affect mitochondria both through physical interactions between the membranes, as with mitochondrial fission, and also via Ca²⁺ release and signaling, as for regulation of the Krebs cycle (Marchi et al., 2014; Rowland and Voeltz, 2012).Ca²⁺ transfer between ER and mitochondriaThe ER contains a highly concentrated pool of intraluminal Ca²⁺, which is involved in the regulation of processes ranging from ATP production to the onset of apoptosis. Upon activation of the IP₃ gated Ca²⁺ channel (IP3R), the ER can release Ca²⁺ ions to the surrounding milieu. The close proximity of the ER–mitochondria MCSs allows for a directional flow of Ca²⁺ to enter the mitochondria through the voltage-dependent anion channel 1 (VDAC1) in the outer membrane (Gincel et al., 2001; Rapizzi et al., 2002) and the mitochondrial calcium uniporter (MCU1) in the inner membrane (Kirichok et al., 2004). The glucose-regulated protein 75 (GRP75) bridges the two organelles to form a stable “synapse” for the Ca²⁺ transfer by binding both VDAC1 and IP3R (Szabadkai et al., 2006; Fig. 3 a). This synapse is necessary for mitochondrial function, homeostasis, energy production, and viability. The concentration of Ca²⁺ inside the inner mitochondrial membrane has consequences for ATP production through the regulation of Ca²⁺-dependent enzymes in the Krebs cycle (Rossi et al., 2019). However, excessive levels of Ca²⁺ ions can induce apoptosis (Rasola and Bernardi, 2011; see below).Open in a separate windowFigure 3.Control of mitochondrial functions via contacts with ER. The figure shows an overview of some of the best-studied functional contacts between the ER and mitochondrial membranes. (a) Calcium transport for homeostasis or apoptosis. In healthy cells, Ca²⁺ flows from the lumen of ER via the IP3R and through the VDAC1 channel in the outer mitochondria membrane (OMM). GRP75 binds both channels to stabilize the synapse. Inside the mitochondria, ions pass the inner mitochondria membrane (IMM) via MCU1 where Ca²⁺ is needed for the Krebs cycle. Several protein–protein interactions are required to strengthen the contact site. Examples of such contacts are the ER proteins MFN2 and VAP-B which can interact with mitochondria-resident proteins MFN1/2 and PTPIP51, respectively. During apoptosis, a membrane complex consisting of BAP31, procaspase-8, CDIP1, and FIS1 tethers mitochondria and ER together in addition to the complex required for calcium transport. BAP31 from the ER bind both CDIP1 and procaspase-8, the latter is activated by interacting via its DED domain to bind a vDED domain on BAP31. FIS1 on the mitochondria interacts with BAP31 to bridge the two organelles. These apoptotic cues lead to increased Ca²⁺ levels in the mitochondria matrix and open the PTP. This disrupts the proton gradient and eventually leads to swelling and rupture of the mitochondria membrane, allowing cytochrome c to leak into the cytosol. APAF1 binds cytochrome c and assembles the apoptosome to execute apoptosis. (b) Mitochondria fission and fusion. ER marks the position for mitochondria fission or fusion by wrapping tubules around the mitochondria. Spire1C nucleates actin filaments and binds INF2 on the ER. INF2 stimulates the mitochondrial Ca²⁺ uptake and polymerizes actin filaments to further connect ER and mitochondria, allowing the IMM to divide first. DRP1 self assembles into a spiral guided by MFF and FIS1, and with the help of actin filaments constricts to separate the OMM. The final separation of the mitochondria can be aided by lysosomes or trans-Golgi network vesicles containing PtdIns4P at the ER–mitochondria contact site. Fusion is engaged by homodimerization between MFN1 or MFN2 in the OMM through their GTPase domain. Similarly, the GTPase domain on OPA1 interacts to fuse the inner membranes. Miro can bind motor proteins on both microtubules and actin filaments, possibly to strengthen the ER–mitochondria contact by reducing mitochondria movements.Several protein–protein interactions are involved in the flow of Ca²⁺ between these membranes, likely due to its important functions in the regulation of both cell growth and cell death. One such example is VAP-B on ER, which binds PTPIP51 on the outer mitochondrial membrane to ensure proper Ca²⁺ release from ER lumen to the mitochondria (De Vos et al., 2012). The interaction is mediated via the FFAT-like motif on PTPIP51 and the MSP motif on VAP-B. A mutated version of VAP-B, VAPB^P56S, and the dysregulation of Ca²⁺ flow between ER and mitochondria are both associated with amyotrophic lateral sclerosis, highlighting the physiological importance of this connection (Langou et al., 2010; Nishimura et al., 2004). Mitofusin (MFN) is another example of a protein bridge that supports Ca²⁺ transport from the ER to the mitochondria. The ER membrane carries MFN2, which can bind heterotypically or homotypically to MFN1 or MFN2, respectively, on the mitochondrial membrane. This interaction aids in forming a stable bridge between the organelles during mitochondria Ca²⁺ uptake (de Brito and Scorrano, 2008).The ER-resident lipid transfer protein PDZD8 (described under “Synchronization of endosome translocation and lipid transfer”) is a mammalian paralog of the yeast protein Mmm1, a member of the ER–mitochondrial encounter structure (ERMES) complex, implicated in the formation of ER–mitochondria contact sites (Hirabayashi et al., 2017; Wideman et al., 2018). In neurons, PDZD8 is necessary for the contact between ER and mitochondria during Ca²⁺ transport. This MCS is proposed to be utilized by neurons to regulate dendritic excitability and plasticity during signal transduction. PDZD8 establishes directionality of Ca²⁺ flux from IP3R and ryanodine receptors toward the mitochondrion. In the absence of PDZD8, the cytosolic Ca²⁺ concentration increases. The protein partner of PDZD8 on the mitochondria membrane has yet to be elucidated (Hirabayashi et al., 2017).Regulation of apoptosisIn the context of cell death, the ER-resident protein BAP31 is crucial for ER–mitochondrial tethering (Ng et al., 1997), and several BAP31 complexes are associated with cell death, including FIS1 (Iwasawa et al., 2011) and CDIP1 (Namba et al., 2013; Fig. 3 a). These protein–protein contacts regulate cell death by establishing signaling platforms, translating apoptotic cues, and engaging in the onset of apoptosis (Iwasawa et al., 2011; Mattson and Chan, 2003; Namba et al., 2013). The initiation of cell death will activate either the intrinsic or extrinsic apoptotic pathway. However, a few common events occur independently of the mode of action. These include the activation of caspases by cleavage and the release of cytochrome c into the cytosol from the mitochondria. These actions are downstream of an increased flow of Ca²⁺ from ER into the mitochondria. When the concentration of Ca²⁺ reaches a certain threshold, the permeability transition pore (PTP) opens and allows for water molecules and protons to travel freely over the inner mitochondria membrane. This disrupts the proton gradient and induces swelling and rupture of the outer membrane (Halestrap, 2009). As a result, cytochrome c leaks into the cytosol where it binds the apoptotic protease-activating factor 1 (APAF1) machinery (Hardingham and Bading, 2003; Mattson and Chan, 2003), causing the apoptosome to assemble and execute apoptosis (Rasola and Bernardi, 2011).Cell death can also be mediated through ceramide-induced apoptosis (Obeid et al., 1993). Ceramides synthesized by ER are highly regulated and are normally transported to the Golgi for further processing. A rise in ceramide levels followed by the recruitment of ceramide binding proteins on the outer mitochondria membrane can trigger apoptosis. The exact molecular mechanisms leading to this event remain elusive, but VDAC2 in the outer mitochondrial membrane has recently been shown to bind ceramides and acts as an effector of cell death signals (Dadsena et al., 2019).Interestingly, BAP31 in the ER–mitochondria MCSs is not restricted to cell death signaling. The mitochondrial membrane protein TOM40 can bind BAP31 to recruit NDUSF4. This protein complex is involved in stress sensing and cellular homeostasis. The lack of BAP31 activates autophagy and glycolysis while reducing mitochondrial oxygen consumption (Namba, 2019). Thus, BAP31 has a role in ER MCS during both self-preservation and self-destruction.Mitochondrial fissionHomeostasis of mitochondria is maintained by fission and fusion of the organelle. ER–mitochondrial contacts spatially define where the mitochondrion will divide (Abrisch et al., 2020; Friedman et al., 2011), as ER tubules wrap around the mitochondria to indicate the position for fission (Abrisch et al., 2020; Chakrabarti et al., 2018; Korobova et al., 2013; Fig. 3 b). The ER-associated formin, inverted formin-2 (INF2), polymerizes actin filaments to establish a close contact between the two organelles. The INF2-mediated actin polymerization stimulates mitochondria Ca²⁺ uptake. Spire1c is an actin nucleator and resides on the outer mitochondrial membrane during fission. Here it binds INF2 directly to connect the two organelles, as well as initiating nucleation of the actin filaments (Manor et al., 2015). ER tubules in the MCS release Ca²⁺, which enters the mitochondria through the VDAC1 channel. INF2 has been implied in the constriction of the inner mitochondria membrane indirectly by an increase in ER–mitochondria contact sites following an influx of Ca²⁺, which triggers the inner mitochondrial membrane to divide first (Chakrabarti et al., 2018). It is unknown exactly how the inner membrane divides, but the electron transport chain is required for the execution. Constriction of the outer membrane depends on the cytosolic GTPase dynamin related-protein 1 (DRP1) to self-assemble into a ring around the mitochondrion at the fission site. For the Drp1-spiral to form, the outer membrane receptors FIS1 and MFF guide the assembly (Elgass et al., 2013; Koch et al., 2005). Drp1 binds actin filaments and modulates actin bundles in vitro (Ji et al., 2015). It is therefore possible that the actin filaments on the mitochondria surface are anchored directly to the DRP1 spiral. The spiral constricts with the aid of actin–myosin filaments (Smirnova et al., 2001). Thus, the inner membrane scission is followed by constriction of the outer membrane as the Drp1 spiral tightens and ultimately results in the formation of two daughter organelles (Chakrabarti et al., 2018; Korobova et al., 2013).A recent contribution to the field of mitochondrial fission is the observation that PtdIns4P-containing vesicles derived from lysosomes or the TGN interact with ER–mitochondria MCSs, forming three-way contacts (Boutry and Kim, 2021; Nagashima et al., 2020). Within these contact sites, ORP1L was suggested to transfer PtdIns4P from lysosomes to the mitochondrion, promoting mitochondrial fission (Boutry and Kim, 2021). Likewise, inhibition of the formation of PtdIns4P microdomains on TGN vesicles results in branched and hyperfused mitochondria (Nagashima et al., 2020), thus indicating an important role of ER–mitochondria triple contact sites to finalize fission.When mitochondria divide, the daughter mitochondria must bear a copy of the mitochondrial DNA (mtDNA) found in the parental mitochondrion. Subpopulations of ER–mitochondria contact sites have been shown to be specifically reserved for and required for the synthesis of mtDNA (Lewis et al., 2016). Following duplication of the nucleoid, DRP1 regulates the mtDNA synthesis and the distribution to the daughter organelles during fission by altering the ER sheets that contact the mitochondria (Ilamathi et al., 2022; Lewis et al., 2016).Mitochondrial fusionIt has been proposed that mitochondrial fission and fusion events are coordinated to quickly respond to metabolic cues. Hence, as for mitochondrial fission, the ER marks the sites of mitochondrial fusion (Guo et al., 2018), and molecules involved in fusion and fission colocalize in ER–mitochondria contact sites (Abrisch et al., 2020). It has been proposed that ER tethering guides the position and timing of mitochondria fusion, but the exact role of ER–mitochondria MCS is still not clear (Gao and Hu, 2021). Fusion of the outer mitochondria membranes is executed by the mitofusins, MFN1 and MFN2 (Cao et al., 2017; Chen et al., 2003), while Opa1 regulates the fusion of the inner mitochondria membrane (Song et al., 2009; Fig. 3 b). Even though both hetero- and homotypic protein interactions can occur between MFN1 and MFN2, it is the homotypic interactions that are required for fusion (Chen et al., 2003). Interestingly, only a certain protein conformation of mitofusins allows for mitochondrial fusion (Franco et al., 2016). Both MFN1 and DRP1 puncta localize to the ER–mitochondria contact sites during the synchronized fusion and fission events, respectively (Abrisch et al., 2020). To maintain the ER–mitochondria contact during fusion, the Ca²⁺ sensitive molecule Miro will decrease mitochondria motility (Kornmann et al., 2011). Miro is also a motorprotein-adaptor involved in both actin filament and microtubule transport. It is unknown exactly how Miro regulates the translocation of mitochondria during fission and fusion; however, acetylation of Miro at specific Lysine residues has been implicated in the context of mitochondria transportation in neurons (Kalinski et al., 2019). More research is needed to further explore the role of Miro in mitochondria fission and fusion.Lipid transfer between ER and mitochondriaMitochondria rely on lipid transport proteins to maintain membrane homeostasis. The evolutionarily conserved ER protein LTC1 is found at MCSs between mitochondria and ER, depending on the mitochondrial import receptors Tom70/71. It has been suggested that the function of LTC1 is to transport and/or sense sterols to maintain correct lipid homeostasis and organelle function (Murley et al., 2015).Miro is not only involved in mitochondria mobility as discussed above but it also promotes lipid transfer. In this role, Miro has been shown to locate to the outer mitochondrial membrane to recruit the lipid channel transporter VPS13D, which is anchored to the ER by VAP-B. Analogously, Miro regulates lipid transfer in ER–peroxisome contact sites (Guillen-Samander et al., 2021; Kornmann et al., 2011).The interaction between the ER-resident protein VAP-B and the mitochondrial protein PTPIP51 is also important for lipid transfer (De Vos et al., 2012). The interaction between the proteins is provided by a FFAT-like motif on PTPIP51 and an MSP domain on VAP-B. The contact site can provide the transport of phosphatidic acid from ER to mitochondria via a TRP motif in PTPIP51 (Yeo et al., 2021). Interestingly, the same protein–protein interaction and MCS are also involved in regulating autophagy in mammalian cells. Loss of PTPIP51 or VAP-B results in increased autophagy, while the overexpression of either protein reduces the number of autophagic structures, likely in a Ca²⁺-dependent manner (Gomez-Suaga et al., 2017).Regulation of peroxisomes by ERPeroxisomes are organelles with diverse metabolic tasks, such as fatty acid turnover, lipid synthesis, and the generation of reactive oxygen species (He et al., 2021; Mast et al., 2020). These processes require a close interplay with other cellular organelles, in particular with the ER. ER and peroxisome interactions have long been observed in electron micrographs, where peroxisomes are localized in the vicinity or even enwrapped by ER (Fahimi et al., 1993; Grabenbauer et al., 2000; Novikoff and Novikoff, 1972; Zaar et al., 1987). The molecular composition and function of peroxisome–ER tethers involve the peroxisomal C-tail anchored proteins acyl-CoA binding domain proteins 4 and 5 (ACBD4 and ACBD5), which interact with the VAP-A and VAP-B proteins in the ER via their FFAT motifs (Costello et al., 2017a; Costello et al., 2017b; Hua et al., 2017; Fig. 4 a). The interaction between ACBD5 and VAP-A/B regulates the extent of ER–peroxisome contacts (Costello et al., 2017a; Hua et al., 2017). Disruption of the tether by depletion of either ACBD5 or VAP-A and VAP-B increased peroxisome motility, indicating that the MCS acts like an anchor for the peroxisomes to the ER (Costello et al., 2017a; Wang et al., 2018). Further, elongation and growth of the PO membrane are reduced, which implicates this contact in lipid transfer. Indeed, plasmalogen and cholesterol homeostasis was shown to be disrupted when the MCS was compromised (Hua et al., 2017). Not surprisingly, a disruption of the peroxisome–ER contact site is associated with pathologies in mice and humans: loss or mutation of ACBD5 causes retinal dystrophy and white matter disease, which are characterized by an increase in very long-chain fatty acids due to impaired lipid transfer and impaired peroxisomal β-oxidation (Bartlett et al., 2021; Darwisch et al., 2020; Ferdinandusse et al., 2017; Gorukmez et al., 2022; Hua et al., 2017; Yagita et al., 2017). Interestingly, more levels of regulation of MCS formation are currently identified, such as phosphorylation of FFAT motifs, which can either promote or inhibit contact formation: ACBD5 phosphorylation through GSK3b was shown to negatively regulate the ACBD5–VAP–B interaction and thus peroxisome–ER MCS formation, while phosphorylation of the STARD3 FFAT motif induces contact formation with the MSP domains of VAP–A and –B (Di Mattia et al., 2020; Kors et al., 2022).Open in a separate windowFigure 4.Contact sites with ER control peroxisome and Golgi functions. (a) The peroxisomal proteins ACBD4 and ACBD5 interact with the VAP proteins in the ER via their FFAT motifs, anchoring peroxisomes to the ER to facilitate lipid transfer. ER anchored E-Syts contact peroxisomal PtdIns(4,5)P₂ to allow cholesterol transport from lysosomes via peroxisomes to the ER. The ER protein BAP31 can potentially interact with FIS1 on peroxisomes, presumably required for peroxisome fission similar to mitochondria. (b) PtdIns4P, the signature phosphoinositide of the Golgi. A PtdIns4P gradient is maintained by phosphorylation of PtdIns by PI4KIIIβ in the TGN and dephosphorylation of PtdIns4P in the ER by the phosphatase SAC1. CERT recognizes PtdIns4P in the Golgi and is tethered to the ER by binding to VAP, where it uses its START domain to transfer ceramide from the ER to the trans-Golgi network. OSBP and OSBP-related proteins (ORPs) interact with PtdIns4P in the Golgi and VAP in the ER. Here, the transfer of PtdIns4P from the Golgi to the ER along the PtdIns4P gradient fuels the counter-transfer of cholesterol or PS to the Golgi. NIR2 binds and transfers PtdIns from the ER to the Golgi, thereby replenishing the PtdIns pool. FAPP1 promotes the activity of SAC1 to dephosphorylate PtdIns4P in trans in narrow membrane contact sites. CPT1C inhibits SAC1 activity to maintain normal levels of PtdIns4P in the Golgi under basal conditions.Another peroxisome–ER contact site is formed by ER-resident extended synaptotagmins (E-Syts-1, 2, and 3) which contact peroxisomal phosphatidylinositol-4,5-bisphosphate to allow cholesterol transport from lysosomes via peroxisomes to the ER (Xiao et al., 2019; Fig. 4 a).Likely, more mammalian ER–peroxisome contact sites will be discovered in the future. MOSPD2 has been suggested to function as an alternative to VAP–A/B, since it also contains an MSP domain shown to interact with FFAT motif proteins (Di Mattia et al., 2018); however its interaction with ACBD4 or 5 has not yet been experimentally proven (Schrader et al., 2020). Further, the ER protein BAP31 interacts with the mitochondrial protein FIS1, which is required for mitochondrial fission (Fig. 4 a). As FIS1 and other mitochondrial fission proteins (DRP1 and MFF) can also be found on peroxisomes, and ER–mitochondrion or ER–endosome MCSs have been shown to mark fission sites (Friedman et al., 2011; Rowland et al., 2014), an analogous ER–peroxisome contact site might assist in peroxisome fission.ER-mediated regulation of the GolgiDue to their collaborative roles in synthesis, modification, and transport of biomolecules, the ER and the Golgi require efficient ways to exchange molecules. Besides vesicular transport between the ER and cis-Golgi, biomolecules can also be exchanged directly at contact sites. To form contacts, PtdIns4P, the signature phosphoinositide of the Golgi, is indispensable as it governs the localization and regulation of lipid-exchange molecules.PtdIns4P fuels lipid transfer between ER and GolgiPtdIns4P is generated by the Golgi-localized phosphatidylinositol kinase 4β (PI4KIIIβ) through the phosphorylation of PtdIns (Balla and Balla, 2006). Oxysterol-binding protein (OSBP) and ORPs (see text box), such as ORP4L, ORP9, and ORP10, are lipid-exchange transporters, which depend on high levels of PtdIns4P in the Golgi, both to form ER–Golgi contact sites and to function in lipid transport. These cytosolic transport proteins contact the Golgi with their PH domain, which binds PtdIns4P and ARF1-GTP, and they tether to the ER via their VAP-binding FFAT motifs, which allows the exchange of PtdIns4P against sterols or PS (Mesmin et al., 2013; Ngo and Ridgway, 2009). The lipid transfer is mediated by the ORD domain and requires PtdIns4P in exchange. The high PtdIns4P levels in the Golgi fuel a counter-transfer of cholesterol (OSBP, ORP4L, and ORP9) or PS (ORP10) to the Golgi (Maeda et al., 2013; Pietrangelo and Ridgway, 2018; Venditti et al., 2019b; Fig. 4 b). The PtdIns4P gradient is maintained by PI4KIIIβ-mediated synthesis of PtdIns4P in the TGN and dephosphorylation of PtdIns4P in the ER by the phosphatase SAC1 (Mesmin et al., 2013; Mesmin et al., 2017).SAC1 regulates PtdIns4P levelsSAC1 is an ER-resident phosphatase that dephosphorylates PtdIns4P in cis in the ER (Mesmin et al., 2013). However, close contact at ER–Golgi–MCS may allow SAC1 to also act in trans to consume PtdIns4P in the Golgi (Manford et al., 2010), and this activity is promoted by the presence of phosphatidyl-four-phosphate-adaptor-protein-1 (FAPP1) within these contacts (Venditti et al., 2019a; Fig. 4 b). Another regulator of SAC1 phosphatase activity is the neuronally-expressed CPT1C (Sierra et al., 2008), which cooperates with Protrudin as described above. CPT1C senses metabolic changes through binding to malonyl-CoA, an intermediate in de novo long-chain fatty acid synthesis, whose levels correlate with the nutritional state. Under basal conditions, CPT1C inhibits SAC1 activity to maintain normal levels of PtdIns4P in the Golgi, allowing AMPA receptor trafficking to the plasma membrane. Under glucose deprivation, CPT1C releases SAC1 inhibition allowing SAC1 to dephosphorylate PtdIns4P at ER–Golgi–MCS in trans, which results in AMPA receptor retention at the TGN (Casas et al., 2020). In this way, the ER can affect neuronal function and cognition depending on energy status.Sphingolipid transfer between ER and GolgiPtdIns4P also plays an important role for sphingolipid transporters: Ceramide transfer protein (CERT) uses its START domain to transfer ceramide from the ER to the TGN for further processing into sphingomyelin (Hanada et al., 2003; Fig. 4 b). CERT forms a contact site by binding to ER-resident VAP proteins via its FFAT motif and its PH domain recognizes PtdIns4P in the Golgi (Hanada et al., 2003; Peretti et al., 2008). The activity of CERT is regulated by phosphorylation and by a negative feedback loop recognizing elevated DAG levels, resulting from sphingomyelin synthesis (Fugmann et al., 2007; Kumagai et al., 2014; Saito et al., 2008). In addition, the START domain can compete with the PH domain for PtdIns4P binding. When the levels of ceramide are high, the START domain is occupied by ceramide, and the shuttling of ceramide from ER to Golgi will occur. When the levels of ceramide are low, the START domain will bind to the PH domain, removing CERT from the Golgi (Prashek et al., 2017).Phosphatidylinositol-four-phosphate adapter protein 2 (FAPP2; also known as PLEKHA8) possesses a PtdIns4P- and ARF1-GTP-binding PH domain at the N-terminus and a glycolipid transfer protein homology domain at the C-terminus, responsible for glucosylceramide (GlcCer) transport (Godi et al., 2004). Depletion of FAPP2 disrupts GlcCer transport from cis- to trans-Golgi, resulting in a disturbed synthesis of complex glycosphingolipids (D''Angelo et al., 2007; D''Angelo et al., 2013; Halter et al., 2007). How FAPP2 aids in the transport of GlcCer from the cis to the trans-Golgi is not yet fully understood, but may involve ER–Golgi contact sites: FAPP2 has a putative FFAT motif (Backman et al., 2018) and has been suggested to transfer GlcCer retrogradely to the ER, where GlcCer translocates into the lumen. From the ER lumen, GlcCer could be anterogradely transported to the trans-Golgi for further glycosylation into complex glycosphingolipids (Halter et al., 2007). Alternatively or additionally, FAPP2 may mediate the direct transfer of GlcCer from cis- to trans-Golgi (D''Angelo et al., 2007).Control of Golgi homeostasisThe phosphatidylinositol-transfer protein NIR2 (PYK2 N-terminal domain- interacting receptor 2) contacts the ER through a classical FFAT motif and it has a PtdIns-transfer domain (PITD) that mediates Golgi localization (Amarilio et al., 2005; Kim et al., 2013). Through its PITD domain, NIR2 transfers PtdIns from the ER to the Golgi, and thereby replenishes the substrate for the PtdIns4-kinase. This closes the PtdIns4P cycle of phosphorylation, transfer, and dephosphorylation, which is necessary to fuel sterol and PS transport (Fig. 4 b). In addition, NIR2 has been shown to regulate DAG levels at the Golgi apparatus (Litvak et al., 2005) and it can affect OSBP and CERT localization and activity (Peretti et al., 2008). The action of the lipid-transfer proteins OSBP, CERT, and NIR2 is thus intricately connected and coordinated at ER–Golgi contact sites.Maintaining lipid homeostasis through lipid-transfer proteins is important for the physiologic function of the secretory pathway. Disturbances in lipid exchange and the resulting imbalances in PtdIns4P or cargo lipids disrupt Golgi morphology, lipid modifications, and anterograde cargo transport processes (Cruz-Garcia et al., 2013; Godi et al., 2004; Litvak et al., 2005; Peretti et al., 2008; Szentpetery et al., 2010; Wakana et al., 2021; Wakana et al., 2015). The ER maintains this balance not only through its direct function at lipid-transfer contact sites but also by anchoring TGN vesicles in the perinuclear area through T6BP/TAX1BP1–SQSTM1/p62–RNF26 as described above (Fig. 1, 1).Conclusions and perspectivesThe ER controls the synthesis and trafficking of molecules not only through its classical functions in protein biosynthesis but also directly through membrane contacts with other organelles. The wide distribution of the ER throughout the cytoplasm makes it well suited to control other organelles via MCSs. As exemplified in this review, ER MCSs have diverse functions that include lipid transfer, Ca²⁺ transfer, protein and lipid dephosphorylation in trans, energy sensing, and regulation of organelle fusion, fission, motility, and positioning. This means that MCSs should always be taken into account when investigating organelle biology. Likewise, organelle-associated diseases can sometimes be understood by considering the dysfunctions of specific ER MCSs. Disruption of ER-organelle MCSs can severely affect cellular homeostasis, causing diseases ranging from metabolic and developmental defects, lipid storage, and neuronal diseases to cancer (Castro et al., 2018; Henne, 2017; Schrader et al., 2020; Simoes et al., 2020; Xu et al., 2020).Even though we are getting a clearer picture of the protein compositions of many MCSs, most of them have not been characterized in full, and we know little about how the many different MCSs influence each other. Triple contacts of ER with endolysosomes, mitochondria, peroxisomes, Golgi apparatus, and lipid droplets have been described (Boutry and Kim, 2021; Elbaz-Alon et al., 2020; Guillen-Samander et al., 2021; Joshi et al., 2018; Nagashima et al., 2020), and it is conceivable that more tripartite MCSs will be detected as more are being characterized. The recent discovery of lipid channel proteins such as VPS13D and ATG2, which allow efficient lipid transport from the ER to other organelles via MCSs, has highlighted the involvement of MCSs in the composition and expansion of organelle membranes. Further progress in the burgeoning research field of ER MCSs will be spurred by combinations of molecular biological dissections of MCSs, structural analyses by cryo-electron microscopy, intracellular localization by advanced light and electron microscopy, and functional characterization by genetic approaches.     

Bridging the gap: Membrane contact sites in signaling,metabolism, and organelle dynamics

Regions of close apposition between two organelles, often referred to as membrane contact sites (MCSs), mostly form between the endoplasmic reticulum and a second organelle, although contacts between mitochondria and other organelles have also begun to be characterized. Although these contact sites have been noted since cells first began to be visualized with electron microscopy, the functions of most of these domains long remained unclear. The last few years have witnessed a dramatic increase in our understanding of MCSs, revealing the critical roles they play in intracellular signaling, metabolism, the trafficking of metabolites, and organelle inheritance, division, and transport.

Introduction

The compartmentalization of cells allows the segregation and regulation of the myriad reactions that occur within them. The tremendous benefits of intracellular compartmentalization also come at a price; to function optimally, cells must transmit signals and exchange material between compartments. Numerous mechanisms have evolved to facilitate these exchanges. One that has not been well appreciated until the last few years is the transmission of signals and molecules between organelles that occurs at regions where the organelles are closely apposed, often called membrane contact sites (MCSs). These sites were first characterized because of their critical roles in the intracellular exchange of lipids and calcium, which can be directly channeled between organelles via MCSs. More recently, it has also become apparent that MCSs are important sites for intracellular signaling, organelle trafficking, and inheritance, and that MCSs are specialized regions where regulatory complexes are assembled (English and Voeltz, 2013; Helle et al., 2013).A hallmark of MCSs is that membranes from two organelles (or compartments of the same organelle) are tethered to one another, but not all instances in which membranes interact with or are tethered to one another are considered MCSs. True MCSs have four properties: (1) membranes from two intracellular compartments are tethered in close apposition, typically within 30 nm, (2) the membranes do not fuse (though they may transiently hemi-fuse), (3) specific proteins and/or lipids are enriched at the MCS, and (4) MCS formation affects the function or composition of at least one of the two organelles in the MCS.This review will discuss what we know about proteins that tether organelles, the exchange of small molecules at MCSs, and other emerging functions of MCSs.

MCS tethers

An MCS tether is a protein or complex of proteins (Fig. 1) that simultaneously binds the two apposing membranes at an organelle contact site and plays a role in maintaining the site (English and Voeltz, 2013; Helle et al., 2013). In many cases it is not yet clear if these proteins and complexes are genuine tethers, which are necessary to maintain MCSs, or function at MCSs but are not necessary to sustain contacts. Distinguishing between these possibilities is an important challenge for the field, especially when more than one protein or complex of proteins independently hold together the membranes at an MCS.Open in a separate windowFigure 1.Proteins proposed to mediate tethering at MCSs. Mammalian proteins are shown on a yellow background, yeast proteins on a blue background, and proteins found in both mammals and yeast are on a green background. Tethering complexes not described in the text are indicated with red numbers: (1) StARD3-VAPs (Alpy et al., 2013), (2) NPC1-ORP5 (Du et al., 2011), (3) Psd2-Pdr17 (Riekhof et al., 2014), (4) Vac8-Nvj1 (Pan et al., 2000), (5) Nvj2 (Toulmay and Prinz, 2012), (6) PTPIP51-VAPs (De Vos et al., 2012), (7) Orai1-STIM1 (Nunes et al., 2012), (8) DGAT2-FATP1 (Xu et al., 2012), and (9) IncD-CERT-VAPs (Derré et al., 2011; Elwell et al., 2011).As a growing number of potential tethers are identified, three trends are emerging. First, most MCSs are maintained by several tethers. One of the best-characterized examples of this is the junction of the ER and plasma membrane (PM) in Saccharomyces cerevisiae. Recent work showed that it was necessary to eliminate six ER resident proteins to dramatically reduce the normally extensive interactions between the ER and PM (Manford et al., 2012; Stefan et al., 2013). This suggests that these six proteins mediate tethering independently of each other. Four of the six proteins (three calcium and lipid-binding domain proteins 1–3, also called Tcb1–3, and Ist2) are integral ER membrane proteins that have cytosolic domains that bind the plasma membranes (Fischer et al., 2009; Toulmay and Prinz, 2012). The other two proteins, Scs2 and Scs22 (Scs, suppressor of Ca²⁺ sensitivity), are homologues of mammalian VAPs (vesicle-associated membrane protein–associated proteins). VAPs are integral membrane tail-anchored proteins in the ER that bind proteins containing FFAT (phenylalanines in an acid tract) motifs (Loewen et al., 2003). A number of proteins that contain these motifs also have domains that bind lipids and proteins in the PM, allowing them to simultaneously bind and tether the ER and PM. For example, some oxysterol-binding protein (OSBP)–related proteins (ORPs) have FFAT motifs and pleckstrin homology (PH) domains that bind phosphoinositides (PIPs) in the plasma membrane (Levine and Munro, 1998; Weber-Boyvat et al., 2013). Thus, ORPs and other FFAT motif-containing proteins can mediate ER–PM tethering via VAPs. It should be noted that VAPs and proteins bound by VAPs also mediate tethering between the ER and organelles in addition to the PM. These are shown in Fig. 1.A second emerging trend is that tethering seems to be a dynamic, regulated process, and we are beginning to understand the mechanisms of dynamic apposition of membranes at MCSs by tethers. One example is ER–PM tethering mediated by proteins called extended synaptotagmins (E-Syts), which are homologues of the yeast Tcb tethers. The tethering of the ER and PM by E-Syts is regulated by Ca²⁺ and the PM-enriched lipid PI(4,5)P₂ (Chang et al., 2013; Giordano et al., 2013). Binding of these molecules by E-Syts may control both the extent of ER–PM contact and the distance between these organelles at MCSs. A second example of regulated MCS formation is provided by a recent study on OSBP. This protein and other FFAT motif-containing proteins have been thought to mediate ER–Golgi tethering by simultaneously binding VAPs in the ER and PIPs in the Golgi complex (Kawano et al., 2006; Peretti et al., 2008). In an elegant set of experiments, Mesmin et al. (2013) showed that OSBP regulates its own ability to mediate ER–Golgi tethering by modulating PI4P levels in the Golgi complex. When PI4P levels in the Golgi complex are high, OSBP tethers the ER and Golgi complex and also transports PI4P from the Golgi to the ER. When the PI4P reaches the ER, it is hydrolyzed by the phosphatase Sac1, preventing it from being transferred back to the Golgi. The reduction in Golgi complex PI4P levels by OSBP causes OSBP to dissociate from the Golgi, decreasing ER–Golgi tethering. Thus, OSBP negatively regulates its own tethering of the ER and Golgi membranes. Lipid transport by OSBP and similar proteins will be discussed in more detail in the section on lipid transport at MCSs.The third important feature of many MCS tethering complexes is that most have functions in addition to tethering. This is well illustrated by complexes proposed to mediate ER–mitochondria tethering in mammalian cells, where four such complexes have been described (Fig. 1). For example, Mfn2 (mitofusin-2) acts as a tether (de Brito and Scorrano, 2008), but the primary function of this dynamin-like protein is to mediate mitochondrial fusion. Although Mfn2 is largely in the outer mitochondrial membrane (OMM), a small fraction also resides the ER, and it has been proposed that the interaction of Mfn2 in the ER with Mfn2 in the OMM tethers the ER and mitochondria (de Brito and Scorrano, 2008). The other ER–mitochondria tethering complexes proposed in mammals (Fig. 1) also have additional functions—either Ca²⁺ signaling or apoptotic signaling between these organelles.

Tethers within organelles

MCSs may form not only between organelles but also between compartments of the same organelle. In two cases, proteins necessary for these intra-organelle contacts are known. The Golgi complex is divided into a number of cisternae that remain closely apposed in some cell types, forming stacked compartments. Two tethering proteins maintain connections between Golgi cisternae. Golgi reassembly stacking protein 65 (GRASP65) forms contacts between cis- and medial-Golgi cisternae and GRASP55 mediates medial- to trans-cisternal interactions (Fig. 1; Barr et al., 1997; Shorter et al., 1999). The Golgi stack disassembles when both GRASPs are depleted, indicating that they are the primary or sole tethers (Xiang and Wang, 2010). Tethering by these proteins is regulated by kinases to allow Golgi cisternal disassembly during the cell cycle. Whether the inter-Golgi contacts formed by GRASPs mediate signaling or lipid exchange between cisternae is not yet known (Tang and Wang, 2013).MCSs also form inside organelles with internal membranes: mitochondria, chloroplasts, and multivesicular bodies. These MCSs may form between membranes within these organelles or between internal membranes and the outer membrane of the organelle. Recently, three groups discovered a tethering complex involved in forming contacts between mitochondrial cisternae and between cisternae and the mitochondrial outer membrane (Harner et al., 2011; Hoppins et al., 2011; von der Malsburg et al., 2011). This complex, called the mitochondrial contact site and cristae organizing system (MICOS), is conserved from yeast to humans and contains at least six proteins (Fig. 1). It is necessary to maintain inner membrane organization and also interacts with protein complexes in the outer membrane, including the translocase of the outer membrane (TOM) complex and the sorting and assembly machinery (SAM) complex (van der Laan et al., 2012; Zerbes et al., 2012).

Lipid exchange at MCSs

Lipid exchange between organelles at MCSs may serve a number of important functions. One is that it allows cells to rapidly modulate the lipid composition of an organelle independently of vesicular trafficking. In addition, some organelles, such as mitochondria and chloroplasts, must obtain most of the lipids they require for membrane biogenesis by nonvesicular lipid trafficking that almost certainly occurs at MCSs (Osman et al., 2011; Wang and Benning, 2012; Horvath and Daum, 2013). Finally, and perhaps most importantly, lipid transfer at MCSs may play an important role in lipid metabolism by channeling lipids to or away from enzymes in different compartments.Some lipid exchange at MCSs is facilitated by soluble lipid transport proteins (LTPs), which can shuttle lipid monomers between membranes (Fig. 2 A). In other cases, known LTPs do not seem to be required and lipids may be exchanged at MCSs by other mechanisms (Fig. 2, B and C), which will be discussed next.Open in a separate windowFigure 2.Possible mechanisms of lipid exchange at MCSs. (A) Transfer by LTPs using CERT as an example. The targeting PH domain (pink) and FFAT motif (blue) are shown. CERT could shuttle between membranes (left) or transfer while binding both membranes (right). (B) Some transfer could occur through hydrophobic channels or tunnels (in green) bridging the two membranes at a MCS. (C) Lipid exchange between hemifused membranes. Hemifusion could be promoted and regulated by proteins (red).Most LTPs fall into at least five superfamilies that differ structurally but that all have a hydrophobic pocket or groove that can bind a lipid monomer, and often have a lid domain that shields the bound lipid from the aqueous phase (D’Angelo et al., 2008; Lev, 2010). This allows LTPs to shuttle lipid monomers between membranes. LTPs probably transfer lipids between organelles in cells most efficiently at MCSs, where they have only a short distance to diffuse between membranes. LTPs that may transfer lipids at contact sites are: OSBP, ceramide transport protein (CERT), the yeast OSBP homologues Osh6 and Osh7, protein tyrosine kinase 2 N-terminal domain–interacting receptor 2 (Nir2), and Ups1 (Hanada, 2010; Connerth et al., 2012; Chang et al., 2013; Maeda et al., 2013; Mesmin et al., 2013).LTPs could function by shuttling between membranes at MCSs or while simultaneously bound to both membranes (Fig. 2 A). Many LTPs have domains that target them to the two membranes at an MCS. For example, OSBP and CERT have FFAT motifs, which bind ER resident VAPs, and PH domains that bind PIPs in the Golgi complex or PM.Another important emerging aspect of lipid exchange by some LTPs is that it may be driven by their ability to exchange one lipid for another. For example, OSBP can transfer both cholesterol and PI4P. At ER–Golgi MCSs, OSBP may facilitate the net movement of cholesterol from the ER to the Golgi and PI4P in the opposite direction (Mesmin et al., 2013). The difference in the PI4P concentrations in the ER and Golgi (lower in the ER than in the Golgi) may drive the net transfer of cholesterol to the Golgi. The ability to exchange one lipid for another has been found for other LTPs (Schaaf et al., 2008; de Saint-Jean et al., 2011; Kono et al., 2013) and may be critical for driving directional lipid exchange at MCSs.Some lipid exchange at MCSs does not seem to be facilitated by LTPs. The best evidence for this comes from studies on lipid transfer between the ER and mitochondria. It has long been known that lipids are exchanged between these two organelles; mitochondria must acquire most of the lipid it requires for membrane biogenesis from the rest of the cell. Lipid exchange at ER–mitochondria MCSs occurs by a mechanism that does not require energy, at least in vitro, and does not require any cytosolic factors (Osman et al., 2011; Vance, 2014).How this lipid transfer occurs is not known, and two possible types of mechanism are shown in Fig. 2, B and C. One is that some MCS proteins form a hydrophobic channel that allows lipids to move between membranes. Such a channel would be similar to an LTP, but whereas lipids enter and exit LTPs by the same opening, they enter and exit channels by different openings. This difference could allow lipid exchange by a channel to be regulated and, if the channel could bind two different lipids simultaneously, it might couple the transfer of the lipids. A domain that may form channels at MCSs has been identified. Called the synaptotagmin-like mitochondrial lipid-binding protein (SMP) domain, it has been predicted to be part of a superfamily of proteins that includes cholesterol ester transfer protein (CETP; Kopec et al., 2010). CETP has a tubular lipid-binding domain that transfers lipids between high-density and low-density lipoproteins, probably while simultaneously bound to both (Qiu et al., 2007; Zhang et al., 2012). SMP domains could transfer lipids between membranes by a similar mechanism. Consistent with this possibility, all SMP-containing proteins in budding yeast localize to MCSs and many mammalian SMP-containing proteins do as well (Toulmay and Prinz, 2012). Interestingly, SMP domains are present in three of the five proteins in a yeast ER–mitochondria tethering complex called ERMES (Kornmann et al., 2009). Whether ERMES facilitates lipid exchange between the ER and mitochondria is not yet clear. Mitochondria derived from cells missing ERMES have altered lipid composition (Osman et al., 2009; Tamura et al., 2012; Tan et al., 2013), indicating that lipid exchange between the ER and mitochondria could be altered in these strains. On the other hand, little or no defect in the rates of phospholipid exchange between ER and mitochondria were found in ERMES mutants (Kornmann et al., 2009; Nguyen et al., 2012; Voss et al., 2012). Thus, whether proteins that contain SMP domains actually facilitate lipid exchange remains to be determined.As second possible mechanism of lipid transfer at MCSs that does not require LTPs is membrane hemifusion (Fig. 2 C), which could allow rapid exchange of large amounts of lipids between compartments. Recent indirect evidence suggests that hemifusion may occur between the ER and chloroplasts (Mehrshahi et al., 2013). This is consistent with an earlier study using optical tweezers that found the ER and chloroplasts remained attached to one another even when a stretching force of 400 pN was applied (Andersson et al., 2007). Whether hemifusion occurs at MCSs in animal cells remains to be determined.

Calcium signaling at MCSs

MCSs between the ER and PM and the ER and mitochondria play central roles in intracellular Ca²⁺ storage, homeostasis, and signaling in mammalian cells. MCSs between the ER and lysosomes may also be important, though they are less well understood (Helle et al., 2013; Lam and Galione, 2013).One of the best-characterized MCSs is the one formed between the PM and ER in muscle cells. In both cardiac and skeletal muscle cells, deep invaginations of the PM, called T (transverse)-tubules, allow it to form extensive contacts with the ER, called the sarcoplasmic reticulum (SR) in muscle cells. These contacts are essential for coupling excitation and contraction. Before excitation, Ca²⁺ levels in the cytoplasm of muscle cells are low, whereas the Ca²⁺ concentrations in the SR and outside muscle cells are high. During muscle excitation, Ca²⁺ rapidly flows into the cytosol through channels in the PM and the SR (Fig. 3 A). The channels in the PM, called dihydropyridine receptors (DHPRs), and those in the SR, known ryanodine receptors RyRs, directly interact with each other where the SR and PM are closely apposed, allowing the opening of both types of channels to be coordinated (Fabiato, 1983; Bannister, 2007; Beam and Bannister, 2010; Rebbeck et al., 2011).Open in a separate windowFigure 3.Ca²⁺ trafficking at ER–PM MCSs. (A) In muscle cells, the interaction of the RyR in the SR and with DHPR in the PM allows the coordinated release of Ca²⁺ during muscle excitation and contraction. See text for details. (B) When STIM1 senses low Ca²⁺ concentration in the ER, it undergoes a conformational change that allows it to oligomerize and bind to the PM, to the protein Orai1, and to accumulate at ER–PM MCSs. Ca²⁺ influx at these sites facilitates Ca²⁺ import into the ER by sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA). (C) Calcium channeling from the ER lumen to the mitochondrial matrix. Calcium exits the ER through the inositol trisphosphate receptor (IP3R) channel, enters mitochondria via VDAC, and then uses the mitochondrial Ca²⁺ uniporter (MCU) to move into the mitochondrial matrix.The extensive contacts between the SR and PM in muscle cells are largely maintained by tethering proteins called junctophilins, which have a single transmembrane domain in the SR and a large cytosolic domain that interacts with the PM. Expression of junctophilins in cells lacking them induces ER–PM contacts (Takeshima et al., 2000) and cells lacking junctophilins have abnormal SR–PM MCSs and defects in Ca²⁺ signaling (Ito et al., 2001; Komazaki et al., 2002; Hirata et al., 2006). Thus, junctophilins are both necessary and sufficient for generating functional SR–PM contacts. However, cells lacking junctophilins still maintain some SR–PM contacts, indicating that other proteins also tether the SR and the PM. Some of this residual tethering probably comes from the interaction of DHPRs and RyRs.ER–PM contacts also play a role in regulating intracellular Ca²⁺ levels in non-excitable cells. When the Ca²⁺ concentration in the ER lumen is low it triggers Ca²⁺ entry into the cytosol and ER from outside cells (Fig. 3 B), a process known as store-operated Ca²⁺ entry (SOCE). The PM channel responsible for Ca²⁺ entry is Orai1, and the sensor of Ca²⁺ concentration in the ER lumen is the integral membrane protein stromal interaction molecule-1 (STIM1). When STIM1 senses that the Ca²⁺ concentration in the ER is low, it oligomerizes and undergoes a conformational change that exposes a basic cluster of amino acids in its C terminus that binds PIPs in the PM (Stathopulos et al., 2006, 2008; Liou et al., 2007; Muik et al., 2011). STIM1 also binds to Orai1 in the PM and activates it (Kawasaki et al., 2009; Muik et al., 2009; Park et al., 2009; Wang et al., 2009). Activation of STIM1 causes it to shift from being relatively evenly distributed on the ER to forming a number of puncta, which are regions were the ER and PM are closely apposed. It seems likely that STIM1 accumulates at and expands preexisting ER–PM MCSs and may also drive the formation of new MCSs (Wu et al., 2006; Lur et al., 2009; Orci et al., 2009).The interaction of STIM1 and Orai1 at ER–PM contacts during SOCE is an elegant mechanism for channeling both signals and small molecules at an MCS. The signal that ER luminal Ca²⁺ concentration is low is transmitted directly from STIM1 in the ER to Orai1 in the PM. The close contact of PM and ER also allows Ca²⁺ to move from outside the cell into the lumen of the ER without significantly increasing cytosolic Ca²⁺ levels (Jousset et al., 2007). During SOCE, ER Ca²⁺ levels are restored by the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) pump (Sampieri et al., 2009; Manjarrés et al., 2011). This pump is enriched in ER–PM contacts with STIM1 and may interact directly with it, suggesting how Ca²⁺ can be effectively channeled from outside cells directly into the ER lumen at ER–PM MCSs (Fig. 3 B).Interestingly, it has become clear that proteins that are not part of the SOCE pathway also facilitate ER–PM connections during Ca²⁺ signaling. The E-Syts have multiple domains that probably bind Ca²⁺. They have been shown to regulate both the number of the ER–PM contacts and the distance between the ER and PM at MCSs during Ca²⁺ signaling (Chang et al., 2013; Giordano et al., 2013).MCSs between the ER and mitochondria similarly facilitate Ca²⁺ movement from the ER lumen to mitochondria (Rizzuto et al., 1998; Csordás et al., 2006). Ca²⁺ channels in the ER and OMM interact with each other at MCSs (Fig. 3 C). The channel in the ER is called the inositol trisphosphate receptor (IP3R), while the voltage-dependent anion channel (VDAC) in the outer mitochondrial membrane is a nonspecific pore that allows Ca²⁺ entry into mitochondria. These proteins, together with the cytosolic chaperone Grp75, form a complex that links the ER and mitochondria and facilitates Ca²⁺ exchange (Szabadkai et al., 2006).More evidence that Ca²⁺ transfer from the ER to mitochondria occurs at MCSs came from studies on the channel that allows Ca²⁺ to move across the inner mitochondrial membrane, called the mitochondrial Ca²⁺ uniporter (MCU). Surprisingly, this channel has an affinity for Ca²⁺ that is lower than the typical Ca²⁺ concentration in the cytosol (Kirichok et al., 2004). However, Ca²⁺ release by the ER at ER–mitochondrial MCSs suggests a solution to this puzzle; the local Ca²⁺ concentration at these MCSs is probably high enough for MCU to function (Csordás et al., 2010). Close contacts between the ER and mitochondria are therefore essential for channeling Ca²⁺ from the ER lumen to the mitochondrial matrix.It is thought that MCSs between the ER (or SR) and lysosomes regulate Ca²⁺ release by lysosomes, but the mechanism is not yet understood (Kinnear et al., 2004, 2008; Galione et al., 2011; Morgan et al., 2011).

Enzymes working in trans and signaling at MCSs

MCSs allow rapid and efficient signaling between intracellular compartments. We are still just beginning to understand the mechanisms and functions of this signaling. One way that signals are transmitted between the two compartments at an MCS is for an enzyme in one compartment to modify substrates in the second; that is, for the enzyme to work in trans. Although there are currently only a few examples of this, which are discussed here, it seems likely that many more will be uncovered.The protein tyrosine phosphatase PTP1B regulates a number of receptor tyrosine kinases. PTP1B resides on the surface of the ER with its active site in the cytosol, and yet the receptor tyrosine kinases it modifies are in the PM. Although this was initially puzzling, it was found that PTP1B probably encounters its substrates at MCSs, either at ER–PM junctions or at contacts between the ER and endocytic recycling compartments (Haj et al., 2002; Boute et al., 2003; Anderie et al., 2007; Eden et al., 2010; Nievergall et al., 2010). Interestingly, in some cases the interaction of PTP1B with substrates in the PM occurs on portions of the PM that are part of cell–cell contacts (Haj et al., 2012), suggesting that ER–PM contacts could play a role in signaling, not only between the ER and PM but between cells as well. Dephosphorylation of receptor tyrosine kinases by PTP1B at contact sites probably allows their kinase activity to be regulated in response to changes in the ER or changes in cellular architecture that alter MCSs. For example, the dephosphorylation of epidermal growth factor receptor (EGFR) by PTP1B occurs at regions of close contact between the ER and multivesicular bodies, causing EGFR to become sequestered with multivesicular bodies (Eden et al., 2010). This may provide a mechanism for cells to regulate EGFR levels on the PM in response to signals in the ER.Lipid metabolism enzymes can also work in trans at MCSs. In two cases, both in yeast, enzymes that reside in the ER have been found to modify lipids in the PM at MCSs. In one instance, the phosphatase Sac1, which is on the surface of the ER, can dephosphorylate PIPs in the PM (Stefan et al., 2011). In the second, the ER enzyme Opi3 methylates phosphatidylethanolamine in the PM, a reaction that is required for the conversion of phosphatidylethanolamine to phosphatidylcholine (Tavassoli et al., 2013). Remarkably, the PIP-binding protein Osh3 (Tong et al., 2013) regulates both reactions, suggesting that lipid metabolism at ER–PM junctions is regulated by PIPs. It seems likely that ER–PM junctions play important roles in integrating lipid metabolism in both organelles.

MCSs and organelle trafficking and inheritance

In addition to being sites at which signals and small molecules are exchanged between cellular compartments, there is growing evidence that MCS formation also regulates organelle trafficking and inheritance.In budding yeast, organelle transport is polarized from the mother cell to the growing bud and is required for proper organelle inheritance. The transport of peroxisomes and mitochondria to the bud is regulated by their association with the ER or PM.Knoblach et al. (2013) found that tethering of the ER to peroxisomes requires Pex3, an integral membrane protein that resides in both compartments, and Inp1, a cytosolic protein that binds to Pex3. This tether keeps peroxisomes in mother cells. When peroxisomes divide they are transferred to the bud by the myosin V motor Myo2 and become attached to the ER in the bud. In cells lacking the ER–peroxisome tether, peroxisomes accumulate in daughter cells. Thus, tethering plays a critical role in ensuring that some peroxisomes are retained in mother cells and that both cells inherit peroxisomes.Mitochondrial inheritance in yeast is regulated by close contacts with both the ER and PM. Mitochondria–PM contacts mediated by a complex containing Num1 and Mdm36 ensure that mitochondria are properly distributed between mother and daughter cells and seem to be particularly important for retaining mitochondria in the mother cells (Klecker et al., 2013; Lackner et al., 2013). Interestingly, Num1–Mdm36-mediated contacts also associate with the ER (Lackner et al., 2013), suggesting that three membranes may somehow associate at these MCSs. An ER–mitochondria tether containing the protein Mmr1, which anchors mitochondria to bud tips, also plays a role in mitochondrial inheritance (Swayne et al., 2011). Thus, the Num1-tethering complex and Mmr1-tethering complex seem to play antagonistic roles in mitochondrial distribution; the Num1 complex promotes mitochondrial retention in the mother, whereas the Mmr1 complex favors retention in the bud.MCSs also play a role in endosomal trafficking in mammalian cells. One of the complexes that tethers the ER to endosomes contains VAPs and ORP1L, which is an OSBP homologue that can bind cholesterol (Fig. 1). ORPlL can also binds the p150^Glued subunit of the dynein–dynactin motor that participates in endosome transport along microtubules (Johansson et al., 2007). When cellular cholesterol levels are high, ORP1L associates with p150^Glued but not VAPs and endosomes are transported on microtubules. However, when cholesterol levels decrease, ORP1L undergoes a conformation change that dissociates it from p150^Glued and allows it to bind to VAPs on the surface of the ER, thus forming a tether between endosomes and the ER (Rocha et al., 2009). Under these conditions, endosome transport on microtubules is blocked. ORPlL is therefore a cholesterol sensor that regulates a switch between the association of endosomes with either motors or the ER.

MCSs and organelle division

A groundbreaking study revealed a new and unexpected role for MCSs between the ER and mitochondria: the ER regulates mitochondrial fission (Friedman et al., 2011). Although a mechanistic understanding of how ER participates in mitochondrial fission is not yet available, the sequence of events is beginning to come into focus (Fig. 4). The ER encircles mitochondria at sites where scission will occur. The ERMES complex is present at these sites (Murley et al., 2013). Because mammalian cells lack ERMES, another tethering complex must perform the same function in higher eukaryotes. Mitochondrial division requires membrane scission by the dynamin-like protein Dnm1/Drp1, which multimerizes on the outer mitochondria membrane. Close contacts between the ER and mitochondria occur before Dnm1/Drp1 assembly, suggesting that these contacts promote or regulate the association of Dnm1/Drp1 with mitochondria and hence mitochondrial division. It is possible that when the ER encircles mitochondria it causes mitochondria to constrict to a diameter that allows Dnm1/Drp1 to assemble. The force necessary to drive constriction may come from actin polymerization. A recent study found that the ER protein, inverted formin-2, probably drives actin polymerization at these sites and is necessary for mitochondria fusion (Korobova et al., 2013).Open in a separate windowFigure 4.Model of ER-mediated regulation of mitochondrial fission at sites of contact. (A) The ER and mitochondria are tethered by ERMES in yeast (other tethers are used in higher eukaryotes). (B) The ER encircles mitochondria at sites where division will occur. (C) Actin polymerization facilitated by formin 2 may cause mitochondrial constriction. (D) The dynamin-like protein Drp1 is recruited to the mitochondrial surface, where it multimerizes and causes mitochondrial scission. (E) After fission, the ER remains associated with the mitochondrion that retains the ERMES complex.Understanding the assembly and regulation of the mitochondrial division machinery at ER–mitochondria MCSs and how this is linked to mitochondrial and perhaps ER function remain fascinating questions for the future. Another interesting question is whether other MCSs play roles in the fission of other organelles.

Proposed functions of ER–mitochondrial MCSs

A growing number of studies have suggested that ER–mitochondria MCSs play critical roles in autophagy, apoptosis, inflammation, reactive oxygen species signaling, and metabolic signaling. ER–mitochondria MCSs have also been implicated in Alzheimer’s disease, Parkinson’s disease, and some viral infections. These topics have been recently reviewed (Eisner et al., 2013; Raturi and Simmen, 2013; Marchi et al., 2014; Vance, 2014) and will not be discussed in detail here.One issue with most of the studies on the functions of ER–mitochondria junctions is that they rely, at least in part, on density gradient purification of the ER that associates with mitochondria. These operationally defined membranes, often called mitochondrial-associated membranes (MAMs), remain poorly defined. In fact, a significant number of proteins that are enriched in MAMs do not seem to be enriched at ER–mitochondria junctions when their localization is determined by other methods (Helle et al., 2013; Vance, 2014). Therefore, it remains unclear why some proteins and lipids are enriched in MAMs.Here, two interesting findings will be discussed that suggest the importance of ER–mitochondrial junctions in signaling in addition to their well-known role in Ca²⁺ signalling.The induction of apoptosis requires signal transmission between the ER and mitochondria. Part of this signaling process occurs through an interaction between the ER protein Bap31 and the mitochondrial fission protein Fission 1 homologue (Fis1; Iwasawa et al., 2011). This interaction occurs at ER–mitochondria MCSs and results in the cleavage of Bap31 by caspase-8 to form p20Bap31, which is pro-apoptotic. Both Bap31 and Fis1 are parts of larger complexes that are still being characterized. Interestingly, it has recently been found that a protein called cell death–involved p53 target-1 (CDIP1) binds to Bap31 during ER stress and promotes apoptotic signaling from the ER to mitochondria (Namba et al., 2013), suggesting how ER stress signals are transmitted from the ER to mitochondria through MCSs.Another important connection between ER–mitochondrial MCSs and signaling has to do with the target of rapamycin (TOR) kinase complexes, which are critical regulators of growth and metabolism. The mammalian TOR complex 2 (mTORC2) was found to interact with the IP3R–Grp75–VDAC complex that tethers the ER and mitochondria (Betz et al., 2013). Remarkably, this study presents evidence that mTORC regulates both the formation of ER–mitochondrial MCSs and mitochondrial function, suggesting an interesting new mechanism for how metabolic signaling can impact mitochondrial function via MCSs.

Conclusions and perspectives

The potential of MCSs to facilitate Ca²⁺ signaling and channel lipids between organelles was recognized some time ago (Levine and Loewen, 2006), but it has only been in the last few years that we have finally begun to have some mechanistic insight into how these processes occur and how MCSs are formed. Many fundamental questions remain to be addressed. How lipid exchange at MCSs that does not require soluble LTPs occurs or whether transient hemifusion of membranes at MCS ever occurs remain open questions. Another is the mechanisms by which Ca²⁺ regulates MCS formation between the ER and other organelles. One major challenge for the field will be devising better methods to visualize MCSs and identify proteins and lipids enriched at these sites. It is particularly important to better understand what the MAM fraction is and what it means for proteins and lipids to be enriched in this fraction.One of the most exciting developments in the study of MCSs in the last few years has been the discovery of the role of MCSs in organelle trafficking, inheritance, and dynamics. These studies have revealed that MCSs not only play critical roles in signaling and metabolism, but also modulate the intracellular distribution of organelles and organelle architecture. Understanding how MCSs perform these functions will probably shed light on the connection between the still murky relationship between organelle structure and function as well as the role of the ER as a regulator of other organelles. Given the current pace of discovery, it seems likely that in the next few years our knowledge of the functions of MCSs will grow dramatically.     

Chlorophyll f-driven photosynthesis in a cavernous cyanobacterium

Chlorophyll (Chl) f is the most recently discovered chlorophyll and has only been found in cyanobacteria from wet environments. Although its structure and biophysical properties are resolved, the importance of Chl f as an accessory pigment in photosynthesis remains unresolved. We found Chl f in a cyanobacterium enriched from a cavernous environment and report the first example of Chl f-supported oxygenic photosynthesis in cyanobacteria from such habitats. Pigment extraction, hyperspectral microscopy and transmission electron microscopy demonstrated the presence of Chl a and f in unicellular cyanobacteria found in enrichment cultures. Amplicon sequencing indicated that all oxygenic phototrophs were related to KC1, a Chl f-containing cyanobacterium previously isolated from an aquatic environment. Microsensor measurements on aggregates demonstrated oxygenic photosynthesis at 742 nm and less efficient photosynthesis under 768- and 777-nm light probably because of diminished overlap with the absorption spectrum of Chl f and other far-red absorbing pigments. Our findings suggest the importance of Chl f-containing cyanobacteria in terrestrial habitats.The textbook concept that oxygenic phototrophs primarily use radiation in the visible range (400–700 nm) has been challenged by several findings of unique cyanobacteria and chlorophylls (Chl) over the past two decades (Miyashita et al., 1996; Chen et al., 2010; Croce and van Amerongen, 2014) Unicellular cyanobacteria in the genus Acaryochloris primarily employ Chl d for oxygenic photosynthesis at 700–720 nm (Miyashita et al., 1996) and thrive in shaded habitats with low levels of visible light but replete of near-infrared radiation (NIR, >700 nm, Kühl et al., 2005; Behrendt et al., 2011, 2012). Furthermore, Chl f was recently discovered in filamentous (Chen et al., 2010; Airs et al., 2014; Gan et al., 2014) and unicellular cyanobacteria (Miyashita et al., 2014), enabling light harvesting even further into the NIR region up to ∼740 nm, often aided by employing additional far-red light-absorbing pigments such as Chl d and phycobiliproteins (Gan et al., 2014). Whereas the biochemical structure (Willows et al., 2013) and biophysical properties (Li et al., 2013; Tomo et al., 2014) of Chl f have been studied in detail, the actual importance of this new chlorophyll for photosynthesis is hardly explored (Li et al., 2014).Chlorophyll f has been found in cyanobacteria originating from aquatic/wet environments: the filamentous Halomicronema hongdechloris from stromatolites in Australia (Chen et al., 2012), a unicellar morphotype (Strain KC1) from Lake Biwa in Japan (Akutsu et al., 2011; Miyashita et al., 2014) and a filamentous Leptolyngbya sp. strain (JSC-1, Gan et al., 2014) from a hot-spring and in a unicellular Chlorogloeopsis fritschii strain from rice paddies (Airs et al., 2014). In this study, we report on a unicellular Chl f-containing cyanobacterium originating from a wet cavernous habitat and demonstrate its capability of NIR-driven oxygenic photosynthesis. Enrichments of the new cyanobacterium were obtained from a dense dark green-blackish biofilm dominated by globular morphotypes of Nostocaceae growing on moist limestone outside Jenolan Caves, NSW, Australia. The sampling site was heavily shaded even during mid-day with low irradiance levels of 400- to 700-nm light varying from 0.5 to 5 μmol photons m⁻² s⁻¹. Biofilms were carefully scraped off the substratum and kept in shaded zip-lock bags in a moist atmosphere until further processing. Samples were then incubated at 28 °C in a f/2 medium under NIR at 720 nm (∼10 μmol photons m⁻² s⁻¹) yielding conspicuous green cell aggregates after several months of incubation. Repeated transfer of the aggregates into fresh medium resulted in a culture predominated by green cell clusters (Figure 1a), exhibiting orange-red fluorescence upon excitation with blue light (Figure 1b). Transmission electron microscopy revealed that the green clusters consisted of slightly elongated unicellular cyanobacteria (∼1- to 2-μm wide and ∼2- to 3-μm long), with stacked thylakoids and embedded in a joint polymer matrix (Figure 1c). Hyperspectral microscopy (Kühl and Polerecky, 2008) of the clusters revealed distinct troughs in the transmission spectra at absorption maxima indicative of Chl a (675–680 nm) and Chl f (∼720 nm; Figure 1d, red line). In situ spectral irradiance measurements at the sampling site showed strong depletion of visible wavelengths in the 480- to 710-nm range (Figure 1d, gray line), whereas highest light levels were found in the near-infrared region of the solar spectrum at 710–900 nm. The presence of Chl a and f was further confirmed in enrichment cultures using high-performance liquid chromatography-based pigment analysis (Figure 1e, Supplementary Figure S1), while no Chl d was detected. In addition, weak spectral signatures of carotenoids and phycobilins, with absorption occurring at ∼495 and 665 nm, were evident in the hyperspectral data. Cyanobacteria, including those producing Chl d/f, are known to actively remodel their pigment content in response to the available light spectrum (Stomp et al., 2007; Chen and Scheer, 2013; Gan et al., 2014) and Chl d/f has almost exclusively been found in cyanobacteria grown under far-red light and not under visible light (Kühl et al., 2005; Chen et al., 2010; Airs et al., 2014; Gan et al., 2014; Li et al., 2014; Miyashita et al., 2014). Recent work describes this acclimation response as ‘Far-Red Light photoacclimation'' (FaRLiP), which, in strain JSC-1, comprises a global change in gene expression and structural remodeling of the PSII/PSI core proteins and phycobilisome constituents (Gan et al., 2014). The extent to which this arrangement results in optimized photosynthetic performance is only known for the NIR (=710 nm)-acclimated strain JSC-1, where exposure to wavelengths >695 nm resulted in 40% higher O₂ evolution rates as compared with cells that were previously adapted to red light (645 nm; Gan et al., 2014). Yet the discrimination of actinic wavelengths and their relative effect on gross photosynthesis in Chl f-containing cells needs further investigation. Using an O₂ microsensor and the light–dark shift method (Revsbech et al., 1983) on embedded Chl f-containing aggregates, we found maximal gross photosynthesis rates (∼1.06 μmol O₂ cm⁻³ s⁻¹) to occur at irradiances of ∼250 μmol photons m⁻² s⁻¹ of 742 nm (half-bandwidth, HBW, 25 nm, Figures 2a and b) with light saturation to occur very early at ∼35 μmol photons m⁻² s⁻¹. Further red-shifted actinic light, that is, 768 nm (HBW 28 nm) and 777 nm (HBW 30 nm), yielded lower O₂ evolution rates, which, in all likelihood, are an effect of the diminished overlap with far-red light-absorbing pigments, including Chl f (Figures 2a and b). As O₂ evolution rates were measured on non-axenic cell aggregates, 16S rDNA amplicon sequencing was employed to determine the microbial diversity found within the enrichment culture. This revealed the presence of a variety of bacterial types, including anoxygenic phototrophs, yet all sequences for known oxygenic phototrophs in the data set (∼9.3% of all reads on the order level, Supplementary Figure S2) formed a single operational taxonomic unit (OTU) closely affiliated with the Chl f-containing strain KC1 (Miyashita et al., 2014, Figure 2c).Open in a separate windowFigure 1Imaging and pigment analysis of Chl f-containing cyanobacteria isolated from a cavernous low-light environment. (a) Representative bright field microscope image of cultured cells grown under 720 nm NIR. (b) Fluorescence image of the same cells as in a, excited at 450–490 nm, with emission being detected at >510 nm. (c) Transmission electron microscopy of a Chl f-containing cyanobacterium with densely stacked thylakoid membranes. (d) Transmittance spectrum of cell aggregate determined by hyperspectral imaging (red line). Ambient light conditions at the site of isolation (gray line), as measured by a spectroradiometer. Note the Chl f-specific in vivo absorption at ∼720 nm in the transmittance spectrum (dotted line). Small insert picture denotes the cells and area of interest (black arrow) from which the spectrum was taken. (e) In vitro absorption spectrum of Chl f extracted from enrichment cultures and analyzed via high-performance liquid chromatography. The two Chl f-specific absorption peaks (404 and 704 nm in acetone:MeOH solvent) are indicated.Open in a separate windowFigure 2Taxonomic affiliation and O₂ evolution of Chl f-containing cells as determined by O₂ microelectrode measurements and 16 S rDNA amplicon sequencing. (a) Emission spectra of narrow-band light-emitting diodes (LEDs) used in this study, with peak emissions at 742, 768 and 777 nm indicated by a–c, respectively. (b) Gross photosynthesis measured via an O₂ microsensor placed in a clump of agarose-embedded Chl f-containing cells. Different NIR irradiance was administered by the LEDs in a and by altering the distance of the LEDs to the embedded cells. (c) Phylogenetic affiliation of known Chl f and/or Chl d-containing cyanobacteria (highlighted in gray) and their respective habitat/place of isolation. Taxonomy was determined by clustering all known oxygenic phototrophs found in enrichment cultures from this study (at order level) into a single OTU (=292 bp length, see Supplementary Materials for details). Phylogeny was inferred using Maximum-likelihood in conjunction with the GTR +I +G nucleotide substitution model, tree stability was tested using bootstrapping with 100 replicates. The analysis involved 39 nucleotide sequences each truncated to a length of 292 bp. Here, the green-sulphur bacterium Chlorobium tepidum TLS was chosen as the outgroup.This advocates that cells from our enrichment culture are related to KC1 cells and supports, in conjunction with further morphological-, physiological- and ultrastructural evidence, that Chl f is extending the usable light spectrum for oxygenic photosynthesis in a cavernous low-light environment. Given the lifestyle and known habitats of recognized Chl d/f-producing cyanobacteria (Figure 2c), we propose that many, if not all, surface-associated cyanobacteria are intrinsically capable of producing far-red light-absorbing pigments and to actively employ them in oxygenic photosynthesis as a result of FaRLiP or similar, yet unknown, mechanisms.