Reconstitution of human atlastin fusion activity reveals autoinhibition by the C terminus

ER network formation depends on membrane fusion by the atlastin (ATL) GTPase. In humans, three paralogs are differentially expressed with divergent N- and C-terminal extensions, but their respective roles remain unknown. This is partly because, unlike Drosophila ATL, the fusion activity of human ATLs has not been reconstituted. Here, we report successful reconstitution of fusion activity by the human ATLs. Unexpectedly, the major splice isoforms of ATL1 and ATL2 are each autoinhibited, albeit to differing degrees. For the more strongly inhibited ATL2, autoinhibition mapped to a C-terminal α-helix is predicted to be continuous with an amphipathic helix required for fusion. Charge reversal of residues in the inhibitory domain strongly activated its fusion activity, and overexpression of this disinhibited version caused ER collapse. Neurons express an ATL2 splice isoform whose sequence differs in the inhibitory domain, and this form showed full fusion activity. These findings reveal autoinhibition and alternate splicing as regulators of atlastin-mediated ER fusion.     

Comparison of human and Drosophila atlastin GTPases

Formation of the endoplasmic reticulum (ER) network requires homotypic membrane fusion, which involves a class of atlastin (ATL) GTPases. Purified Drosophila ATL is capable of mediating vesicle fusion in vitro, but such activity has not been reported for any other ATLs. Here, we determined the preliminary crystal structure of the cytosolic segment of Drosophila ATL in a GDP-bound state. The structure reveals a GTPase domain dimer with the subsequent three-helix bundles associating with their own GTPase domains and pointing in opposite directions. This conformation is similar to that of human ATL1, to which GDP and high concentrations of inorganic phosphate, but not GDP only, were included. Drosophila ATL restored ER morphology defects in mammalian cells lacking ATLs, and measurements of nucleotide-dependent dimerization and GTPase activity were comparable for Drosophila ATL and human ATL1. However, purified and reconstituted human ATL1 exhibited no in vitro fusion activity. When the cytosolic segment of human ATL1 was connected to the transmembrane (TM) region and C-terminal tail (CT) of Drosophila ATL, the chimera still exhibited no fusion activity, though its GTPase activity was normal. These results suggest that GDP-bound ATLs may adopt multiple conformations and the in vitro fusion activity of ATL cannot be achieved by a simple collection of functional domains.     

The hypervariable region of atlastin-1 is a site for intrinsic and extrinsic regulation

Atlastin (ATL) GTPases catalyze homotypic membrane fusion of the peripheral endoplasmic reticulum (ER). GTP-hydrolysis–driven conformational changes and membrane tethering are prerequisites for proper membrane fusion. However, the molecular basis for regulation of these processes is poorly understood. Here we establish intrinsic and extrinsic modes of ATL1 regulation that involve the N-terminal hypervariable region (HVR) of ATLs. Crystal structures of ATL1 and ATL3 exhibit the HVR as a distinct, isoform-specific structural feature. Characterizing the functional role of ATL1’s HVR uncovered its positive effect on membrane tethering and on ATL1’s cellular function. The HVR is post-translationally regulated through phosphorylation-dependent modification. A kinase screen identified candidates that modify the HVR site specifically, corresponding to the modifications on ATL1 detected in cells. This work reveals how the HVR contributes to efficient and potentially regulated activity of ATLs, laying the foundation for the identification of cellular effectors of ATL-mediated membrane processes.     

The dynamin-like GTPase Sey1p mediates homotypic ER fusion in S. cerevisiae

The endoplasmic reticulum (ER) forms a network of tubules and sheets that requires homotypic membrane fusion to be maintained. In metazoans, this process is mediated by dynamin-like guanosine triphosphatases (GTPases) called atlastins (ATLs), which are also required to maintain ER morphology. Previous work suggested that the dynamin-like GTPase Sey1p was needed to maintain ER morphology in Saccharomyces cerevisiae. In this paper, we demonstrate that Sey1p, like ATLs, mediates homotypic ER fusion. The absence of Sey1p resulted in the ER undergoing delayed fusion in vivo and proteoliposomes containing purified Sey1p fused in a GTP-dependent manner in vitro. Sey1p could be partially replaced by ATL1 in vivo. Like ATL1, Sey1p underwent GTP-dependent dimerization. We found that the residual ER-ER fusion that occurred in cells lacking Sey1p required the ER SNARE Ufe1p. Collectively, our results show that Sey1p and its homologues function analogously to ATLs in mediating ER fusion. They also indicate that S. cerevisiae has an alternative fusion mechanism that requires ER SNAREs.     

ER network formation requires a balance of the dynamin-like GTPase Sey1p and the Lunapark family member Lnp1p

Although studies on endoplasmic reticulum (ER) structure and dynamics have focused on the ER tubule-forming proteins (reticulons and DP1/Yop1p) and the tubule fusion protein atlastin, nothing is known about the proteins and processes that act to counterbalance this machinery. Here we show that Lnp1p, a member of the conserved Lunapark family, plays a role in ER network formation. Lnp1p binds to the reticulons and Yop1p and resides at ER tubule junctions in both yeast and mammalian cells. In the yeast Saccharomyces cerevisiae, the interaction of Lnp1p with the reticulon protein, Rtn1p, and the localization of Lnp1p to ER junctions are regulated by Sey1p, the yeast orthologue of atlastin. We propose that Lnp1p and Sey1p act antagonistically to balance polygonal network formation. In support of this proposal, we show that the collapsed, densely reticulated ER network in lnp1 Δ cells is partially restored when the GTPase activity of Sey1p is abrogated.     

The Atlastin C-terminal Tail Is an Amphipathic Helix That Perturbs the Bilayer Structure during Endoplasmic Reticulum Homotypic Fusion

Fusion of tubular membranes is required to form three-way junctions found in reticular subdomains of the endoplasmic reticulum. The large GTPase Atlastin has recently been shown to drive endoplasmic reticulum membrane fusion and three-way junction formation. The mechanism of Atlastin-mediated membrane fusion is distinct from SNARE-mediated membrane fusion, and many details remain unclear. In particular, the role of the amphipathic C-terminal tail of Atlastin is still unknown. We found that a peptide corresponding to the Atlastin C-terminal tail binds to membranes as a parallel α helix, induces bilayer thinning, and increases acyl chain disorder. The function of the C-terminal tail is conserved in human Atlastin. Mutations in the C-terminal tail decrease fusion activity in vitro, but not GTPase activity, and impair Atlastin function in vivo. In the context of unstable lipid bilayers, the requirement for the C-terminal tail is abrogated. These data suggest that the C-terminal tail of Atlastin locally destabilizes bilayers to facilitate membrane fusion.     

Membrane topology and quaternary structure of cardiac gap junction ion channels.

The membrane topology and quaternary structure of rat cardiac gap junction ion channels containing alpha 1 connexin (i.e. Cx43) have been examined using anti-peptide antibodies directed to seven different sites in the protein sequence, cleavage by an endogenous protease in heart tissue and electron microscopic image analysis of native and protease-cleaved two-dimensional membrane crystals of isolated cardiac gap junctions. Specificity of the peptide antibodies was established using dot immunoblotting, Western immunoblotting, immunofluorescence and immunoelectron microscopy. Based on the folding predicted by hydropathy analysis, five antibodies were directed to sites in cytoplasmic domains and two antibodies were directed to the two extracellular loop domains. Isolated gap junctions could not be labeled by the two extracellular loop antibodies using thin-section immunogold electron microscopy. This is consistent with the known narrowness of the extracellular gap region that presumably precludes penetration of antibody probes. However, cryo-sectioning rendered the extracellular domains accessible for immunolabeling. A cytoplasmic "loop" domain of at least Mr = 5100 (residues (101 to 142) is readily accessible to peptide antibody labeling. The native Mr = 43,000 protein can be protease-cleaved on the cytoplasmic side of the membrane, resulting in an Mr approximately 30,000 membrane-bound fragment. Western immunoblots showed that protease cleavage occurs at the carboxy tail of the protein, and the cleavage site resides between amino acid residues 252-271. Immunoelectron microscopy demonstrated that the Mr approximately 13,000 carboxy-terminal peptide(s) is released after protease cleavage and does not remain attached to the Mr approximately 30,000 membrane-bound fragment via non-covalent interactions. Electron microscopic image analysis of two-dimensional membrane crystals of cardiac gap junctions revealed that the ion channels are formed by a hexagonal arrangement of protein subunits. This quaternary arrangement is not detectably altered by protease cleavage of the alpha 1 polypeptide. Therefore, the Mr approximately 13,000 carboxyterminal domain is not involved in forming the transmembrane ion channel. The similar hexameric architecture of cardiac and liver gap junction connexins indicates conservation in the molecular design of the gap junction channels formed by alpha or beta connexins.     

Structure and dynamics of three-way DNA junctions: atomic force microscopy studies

We have used atomic force microscopy (AFM) to study the conformation of three-way DNA junctions, intermediates of DNA replication and recombination. Immobile three-way junctions with one hairpin arm (50, 27, 18 and 7 bp long) and two relatively long linear arms were obtained by annealing two partially homologous restriction fragments. Fragments containing inverted repeats of specific length formed hairpins after denaturation. Three-way junctions were obtained by annealing one strand of a fragment from a parental plasmid with one strand of an inverted repeat-containing fragment, purified from gels, and examined by AFM. The molecules are clearly seen as three-armed molecules with one short arm and two flexible long arms. The AFM analysis revealed two important features of three-way DNA junctions. First, three-way junctions are very dynamic structures. This conclusion is supported by a high variability of the inter-arm angle detected on dried samples. The mobility of the junctions was observed directly by imaging the samples in liquid (AFM in situ). Second, measurements of the angle between the arms led to the conclusion that three-way junctions are not flat, but rather pyramid-like. Non-flatness of the junction should be taken into account in analysis of the AFM data.     

The luminal N-terminus of yeast Nvj1 is an inner nuclear membrane anchor

The endoplasmic reticulum (ER) in Saccharomyces cerevisiae is largely divided between perinuclear and cortical compartments. Yeast Nvj1 localizes exclusively to small patches on the perinuclear ER where it interacts with Vac8 in the vacuole membrane to form nucleus-vacuole (NV) junctions. Three regions of Nvj1 mediate the biogenesis of NV junctions. A membrane-spanning domain targets the protein to the ER. The C-terminus binds Vac8 in the vacuole membrane, which induces the clustering of both proteins into NV junctions. The luminal N-terminus is required for strict perinuclear localization. Three-dimensional cryo-electron tomography reveals that Nvj1 clamps the separation between the two nuclear membranes to half the width of bulk nuclear envelope. The N-terminus contains a hydrophobic sequence bracketed by basic residues that resembles outer mitochondrial membrane signal-anchors. The hydrophobic sequence can be scrambled or reversed without affecting function. Mutations that reduce the hydrophobicity of the core sequence or affect the distribution of basic residues cause mislocalization to the cortical ER. We conclude that the N-terminus of Nvj1 is a retention sequence that bridges the perinuclear lumen and inserts into the inner nuclear membrane.     

Use of a beta-lactamase fusion vector to investigate the organization of penicillin-binding protein 1B in the cytoplasmic membrane of Escherichia coli

The coding region for the mature form of TEM beta-lactamase was fused to random positions within the coding region of the penicillin-binding protein 1B (PBP 1B) gene and the nucleotide sequences across the fusion junctions of 100 in-frame fusions were determined. All fusion proteins that contained at least the NH2-terminal 94 residues of PBP 1B provided individual cells of E. coli with substantial levels of ampicillin resistance, suggesting that the beta-lactamase moiety had been translocated to the periplasm. Fusion proteins that contained less than or equal to 63 residues of PBP 1B possessed beta-lactamase activity, but could not protect single cells of E. coli from ampicillin, indicating that the beta-lactamase moiety of these fusion proteins remained in the cytoplasm. The beta-lactamase fusion approach suggested a model for the organization of PBP 1B in which the protein is embedded in the cytoplasmic membrane by a single hydrophobic transmembrane segment (residues 64-87), with a short NH2-terminal domain (residues 1-63), and the remainder of the polypeptide (residues 88-844) exposed on the periplasmic side of the cytoplasmic membrane. The proposed model for the organization of PBP 1B was supported by experiments which showed that the protein was completely digested by proteinase K added from the periplasmic side of the cytoplasmic membrane but was only slightly reduced in size by protease attack from the cytoplasmic side of the membrane.     

Hydrophobic residues that form putative fusion loops of Epstein-Barr virus glycoprotein B are critical for fusion activity

To test the importance of the hydrophobic residues within the putative Epstein-Barr virus (EBV) glycoprotein B (gB) fusion loops in membrane fusion, WY(112-113) and WLIW(193-196) were mutated into alanine, glutamic acid, or the analogous residues from herpes simplex virus type 1 (HSV-1) gB (HR and RVEA). All gB variants exhibited cell surface expression, demonstrating that the substitutions did not perturb gB trafficking. None of six gB variants was, however, capable of mediating fusion with either epithelial or B cells. These data demonstrate that the bulky and hydrophobic EBV loop residues, which differ from the more hydrophilic HSV-1 residues and appear more compatible with membrane insertion, are essential for EBV gB-dependent fusion.     

The three-way DNA junction is a Y-shaped molecule in which there is no helix-helix stacking.

We have studied the structure of a number of three-way DNA junctions that were closely related in sequence to four-way junctions studied previously. We observe that the electrophoretic mobility of the species derived by selective shortening of one arm of a junction are very similar whichever arm is shortened, and that this remains so whether or not magnesium is present in the buffer. This suggests that the angles subtended between the arms of the three-way junctions are similar. All thymine bases located immediately at the junction are reactive to osmium tetroxide, indicating that out-of-plane attack is not prevented by helix-helix stacking, and this is also independent of the presence or absence of metal cations. The results suggest that the three-way junction cannot undergo an ion-induced conformational folding involving helical stacking, but remains fixed in a Y-shaped extended conformation. Thus the three- and four-way junctions are quite different in character in the presence of cations.     

Structure-function analysis of the coiled-coil and leucine-rich repeat domains of the RPS5 disease resistance protein

The Arabidopsis (Arabidopsis thaliana) RESISTANCE TO PSEUDOMONAS SYRINGAE5 (RPS5) disease resistance protein mediates recognition of the Pseudomonas syringae effector protein AvrPphB. RPS5 belongs to the coiled-coil-nucleotide-binding site-leucine-rich repeat (CC-NBS-LRR) family and is activated by AvrPphB-mediated cleavage of the protein kinase PBS1. Here, we present a structure-function analysis of the CC and LRR domains of RPS5 using transient expression assays in Nicotiana benthamiana. We found that substituting the CC domain of RPS2 for the RPS5 CC domain did not alter RPS5 specificity and only moderately reduced its ability to activate programmed cell death, suggesting that the CC domain does not play a direct role in the recognition of PBS1 cleavage. Analysis of an RPS5-super Yellow Fluorescent Protein fusion revealed that RPS5 localizes to the plasma membrane (PM). Alanine substitutions of predicted myristoylation (glycine-2) and palmitoylation (cysteine-4) residues affected RPS5 PM localization, protein stability, and function in an additive manner, indicating that PM localization is essential to RPS5 function. The first 20 amino acids of RPS5 were sufficient for directing super Yellow Fluorescent Protein to the PM. C-terminal truncations of RPS5 revealed that the first four LRR repeats are sufficient for inhibiting RPS5 autoactivation; however, the complete LRR domain was required for the recognition of PBS1 cleavage. Substitution of the RPS2 LRR domain resulted in the autoactivation of RPS5, indicating that the LRR domain must coevolve with the NBS domain. We conclude that the RPS5 LRR domain functions to suppress RPS5 activation in the absence of PBS1 cleavage and promotes RPS5 activation in its presence.     

Drug Uptake,Lipid Rafts,and Vesicle Trafficking Modulate Resistance to an Anticancer Lysophosphatidylcholine Analogue in Yeast

The ether-phospholipid edelfosine, a prototype antitumor lipid (ATL), kills yeast cells and selectively kills several cancer cell types. To gain insight into its mechanism of action, we performed chemogenomic screens in the Saccharomyces cerevisiae gene-deletion strain collection, identifying edelfosine-resistant mutants. LEM3, AGP2, and DOC1 genes were required for drug uptake. Edelfosine displaced the essential proton pump Pma1p from rafts, inducing its internalization into the vacuole. Additional ATLs, including miltefosine and perifosine, also displaced Pma1p from rafts to the vacuole, suggesting that this process is a major hallmark of ATL cytotoxicity in yeast. Radioactive and synthetic fluorescent edelfosine analogues accumulated in yeast plasma membrane rafts and subsequently the endoplasmic reticulum. Although both edelfosine and Pma1p were initially located at membrane rafts, internalization of the drug toward endoplasmic reticulum and Pma1p to the vacuole followed different routes. Drug internalization was not dependent on endocytosis and was not critical for yeast cytotoxicity. However, mutants affecting endocytosis, vesicle sorting, or trafficking to the vacuole, including the retromer and ESCRT complexes, prevented Pma1p internalization and were edelfosine-resistant. Our data suggest that edelfosine-induced cytotoxicity involves raft reorganization and retromer- and ESCRT-mediated vesicular transport and degradation of essential raft proteins leading to cell death. Cytotoxicity of ATLs is mainly dependent on the changes they induce in plasma membrane raft-located proteins that lead to their internalization and subsequent degradation. Edelfosine toxicity can be circumvented by inactivating genes that then result in the recycling of internalized cell-surface proteins back to the plasma membrane.     

Use of a β-lactamase fusion vector to investigate the organization of penicillin-binding protein 1B in the cytoplasmic membrane of Escherichia coli

The coding region for the mature form of TEM β–lactamase was fused to random positions within the coding region of the penicillin–binding protein 1B (PBP 1B) gene and the nucleotide sequences across the fusion junctions of 100 in–frame fusions were determined. All fusion proteins that contained at least the NH₂–terminal 94 residues of PBP 1B provided individual cells of E. coli with substantial levels of ampicillin resistance, suggesting that the β–lactamase moiety had been translocated to the periplasm. Fusion proteins that contained ≤ 63 residues of PBP 1B possessed β–lactamase activity, but could not protect single cells of E. coli from ampicillin, indicating that the 3–lactamase moiety of these fusion proteins remained in the cytoplasm. The β–lactamase fusion approach suggested a model for the organization of PBP 1B in which the protein is embedded in the cytoplasmic membrane by a single hydrophobic trans–membrane segment (residues 64–87), with a short NH₂–terminal domain (residues 1–63), and the remainder of the polypeptide (residues 68–844) exposed on the periplasmic side of the cytoplasmic membrane. The proposed model for the organization of PBP 1B was supported by experiments which showed that the protein was completely digested by proteinase K added from the periplasmic side of the cytoplasmic membrane but was only slightly reduced in size by protease attack from the cytoplasmic side of the membrane.     

A three-way junction and constituent stem-loops as the stimulator for programmed -1 frameshifting in bacterial insertion sequence IS911

Several signals are required for the programmed frameshifting in translation of IS911 mRNA. These include a Shine Dalgarno (SD)-like sequence, a slippery sequence of six adenine residues and a guanine residue (A6G) and a 3' secondary structure. The structure of the mRNA containing these elements was investigated using chemical and enzymatic probing. The probing data show that the 3' structure is a three-way junction of stems. The function of the three-way junction was investigated by mutagenesis. Disrupting the stability of the structure greatly affects frameshifting and transposition levels as tested by separate in vivo assays. Structural probing and thermal melting profiles indicate that the disrupted three-way junctions have altered structures.     

Connexin43 PDZ2 binding domain mutants create functional gap junctions and exhibit altered phosphorylation

Connexin43 (Cx43) is the most abundantly expressed gap junction protein. The C-terminal tail of Cx43 is important for regulation of gap junctions via phosphorylation of specific tyrosine and serine residues and through interactions with cellular proteins. The C-terminus of Cx43 has been shown to interact with the PDZ2 domain of the tight and adherens junction associated zona occludens 1 (ZO-1) protein. Analysis of the PDZ2 binding domain of Cx43 indicated that positions -3 and -2, and the final hydrophobic amino acid at the C-terminus, are critical for ZO-1 binding. In addition, the C-termini of connexins 40 and 45, but not Cx32, interacted with ZO-1. To evaluate the functional significance of the Cx43-ZO-1 interaction, Cx43 wild type (Cx43wt) and mutants lacking either the C-terminal hydrophobic isoleucine (Cx43deltaI382) or the last five amino acids (Cx43delta378-382), required for ZO-1 binding in vitro, were introduced into a Cx43-deficient MDCK cell line. In vitro binding studies and coimmunoprecipitation assays indicated that these Cx43 mutants failed to interact with ZO-1. Confocal and deconvolution microscopy revealed that a fraction of Cx43wt colocalized with ZO-1 at the plasma membrane. A similar colocalization pattern was observed for the Cx43deltaI382 and Cx43 delta378-382 mutants, which were translocated to the plasma membrane and formed functional gap junction channels. The wt and mutant Cx43 appeared to have similar turnover rates. However, the P2 and P3 phosphoisoforms of the Cx43 mutants were significantly reduced compared to Cx43wt. These studies indicated that the interaction of Cx43 with ZO-1 may contribute to the regulation of Cx43 phosphorylation.     

Mapping interactions between the CRAC activation domain and CC1 regulating the activity of the ER Ca2+ sensor STIM1

Stromal interaction molecule 1 (STIM1) is a widely expressed protein that functions as the endoplasmic reticulum (ER) Ca²⁺ sensor and activator of Orai1 channels. In resting cells with replete Ca²⁺ stores, an inhibitory clamp formed by the coiled-coil 1 (CC1) domain interacting with the CRAC-activation domain (CAD) of STIM1 helps keep STIM1 in a quiescent state. Following depletion of ER Ca²⁺ stores, the brake is released, allowing CAD to extend away from the ER membrane and enabling it to activate Orai1 channels. However, the molecular determinants of CC1–CAD interactions that enforce the inhibitory clamp are incompletely understood. Here, we performed Ala mutagenesis in conjunction with live-cell FRET analysis to examine residues in CC1 and CAD that regulate the inhibitory clamp. Our results indicate that in addition to previously identified hotspots in CC1⍺1 and CC3, several hydrophobic residues in CC2 and the apex region of CAD are critical for CC1–CAD interactions. Mutations in these residues loosen the CC1-CAD inhibitory clamp to release CAD from CC1 in cells with replete Ca²⁺ stores. By contrast, altering the hydrophobic residues L265 and L273 strengthens the clamp to prevent STIM1 activation. Inclusion of the inactivation domain of STIM1 helps stabilize CC1–CAD interaction in several mutants to prevent spontaneous STIM1 activation. In addition, R426C, a human disease–linked mutation in CC3, affects the clamp but also impairs Orai1 binding to inhibit CRAC channel activation. These results identify the CC2, apex, and inactivation domain regions of STIM1 as important determinants of STIM1 activation.     

Mutational analysis of different regions in the coxsackievirus 2B protein: requirements for homo-multimerization, membrane permeabilization, subcellular localization, and virus replication

The coxsackievirus 2B protein is a small hydrophobic protein (99 amino acids) that increases host cell membrane permeability, possibly by forming homo-multimers that build membrane-integral pores. Previously, we defined the functional role of the two hydrophobic regions HR1 and HR2. Here, we investigated the importance of regions outside HR1 and HR2 for multimerization, increasing membrane permeability, subcellular localization, and virus replication through analysis of linker insertion and substitution mutants. From these studies, the following conclusions could be drawn. (i) The hydrophilic region ((58)RNHDD(62)) between HR1 and HR2 is critical for multimerization and increasing membrane permeability. Substitution analysis of Asn(61) and Asn(62) demonstrated the preference for short polar side chains (Asp, Asn), residues that are often present in turns, over long polar side chains (Glu, Gln). This finding supports the idea that the hydrophilic region is involved in pore formation by facilitating a turn between HR1 and HR2 to reverse chain direction. (ii) Studies undertaken to define the downstream boundary of HR2 demonstrated that the aromatic residues Trp(80) and Trp(82), but not the positively charged residues Arg(81), Lys(84), and Lys(86) are important for increasing membrane permeability. (iii) The N terminus is not required for multimerization but does contribute to the membrane-active character of 2B. (iv) The subcellular localization of 2B does not rely on regions outside HR1 and HR2 and does not require multimerization. (v) Virus replication requires both the membrane-active character and an additional function of 2B that is not connected to this activity.     

NHE6 protein possesses a signal peptide destined for endoplasmic reticulum membrane and localizes in secretory organelles of the cell

The NHE6 protein is a unique Na(+)/H(+) exchanger isoform believed to localize in mitochondria. It possesses a hydrophilic N-terminal portion that is rich in positively charged residues and many hydrophobic segments. In the present study, signal sequences in the NHE6 molecule were examined for organelle localization and membrane topogenesis. When the full-length protein was expressed in COS7 cells, it localized in the endoplasmic reticulum and on the cell surface. Furthermore, the protein was fully N-glycosylated. When green fluorescent protein was fused after the second (H2) or third (H3) hydrophobic segment, the fusion proteins were targeted to the endoplasmic reticulum (ER) membrane. The localization pattern was the same as that of fusion proteins in which green fluorescent protein was fused after H2 of NHE1. In an in vitro system, H1 behaved as a signal peptide that directs the translocation of the following polypeptide chain and is then processed off. The next hydrophobic segment (H2) halted translocation and eventually became a transmembrane segment. The N-terminal hydrophobic segment (H1) of NHE1 also behaved as a signal peptide. Cell fractionation studies using antibodies against the 15 C-terminal residues indicated that NHE6 protein localized in the microsomal membranes of rat liver cells. All of the NHE6 molecules in liver tissue possess an endoglycosidase H-resistant sugar chain. These findings indicate that NHE6 protein is targeted to the ER membrane via the N-terminal signal peptide and is sorted to organelle membranes derived from the ER membrane.