Design and regulation of the AAA+ microtubule motor dynein

Dyneins are highly complex molecular motors that transport their attached cargo towards the minus end of microtubules. These enzymes are required for many essential motile activities within the cytoplasm and also power eukaryotic cilia and flagella. Each dynein contains one or more heavy chain motor units that consist of an N-terminal stem domain that is involved in cargo attachment, and six AAA+ domains (AAA1-6) plus a C-terminal globular segment that are arranged in a heptameric ring. At least one AAA+ domain (AAA1) is capable of ATP binding and hydrolysis, and the available data suggest that one or more additional domains also may bind nucleotide. The ATP-sensitive microtubule binding site is located at the tip of a 10nm coiled coil stalk that emanates from between AAA4 and AAA5. The function of this motor both in the cytoplasm and the flagellum must be tightly regulated in order to result in useful work. Consequently, dyneins also contain a series of additional components that serve to define the cargo-binding properties of the enzyme and which act as sensors to transmit regulatory inputs to the motor units. Here we describe the two basic dynein designs and detail the various regulatory systems that impinge on this motor within the eukaryotic flagellum.     

The third P-loop domain in cytoplasmic dynein heavy chain is essential for dynein motor function and ATP-sensitive microtubule binding

Sequence comparisons and structural analyses show that the dynein heavy chain motor subunit is related to the AAA family of chaperone-like ATPases. The core structure of the dynein motor unit derives from the assembly of six AAA domains into a hexameric ring. In dynein, the first four AAA domains contain consensus nucleotide triphosphate-binding motifs, or P-loops. The recent structural models of dynein heavy chain have fostered the hypothesis that the energy derived from hydrolysis at P-loop 1 acts through adjacent P-loop domains to effect changes in the attachment state of the microtubule-binding domain. However, to date, the functional significance of the P-loop domains adjacent to the ATP hydrolytic site has not been demonstrated. Our results provide a mutational analysis of P-loop function within the first and third AAA domains of the Drosophila cytoplasmic dynein heavy chain. Here we report the first evidence that P-loop-3 function is essential for dynein function. Significantly, our results further show that P-loop-3 function is required for the ATP-induced release of the dynein complex from microtubules. Mutation of P-loop-3 blocks ATP-mediated release of dynein from microtubules, but does not appear to block ATP binding and hydrolysis at P-loop 1. Combined with the recent recognition that dynein belongs to the family of AAA ATPases, the observations support current models in which the multiple AAA domains of the dynein heavy chain interact to support the translocation of the dynein motor down the microtubule lattice.     

Novel structural and functional insights into the MoxR family of AAA+ ATPases

The MoxR family of AAA+ ATPases is widespread among bacteria and archaea, although their cellular functions are not well characterized. Based on recent studies, MoxR ATPases are proposed to have chaperone-like function for the maturation of specific protein complexes or for the insertion of cofactors into proteins. MoxR proteins have been found to be important modulators of multiple stress response pathways in different organisms. For example, the respective MoxR proteins have been found to play important roles in the cell envelope stress response in Rhizobium leguminosarum, in the oxidative stress, acid stress, and heat stress responses in Francisella tularensis, in the acid stress and stringent responses in Escherichia coli, in viral tail formation in the crenarchaeal Acidianus two-tailed virus, and in the utilization of carbon monoxide as the sole carbon source by the Gram-negative chemolithoautotrophe Oligotropha carboxidovorans. Recent structural studies on the MoxR proteins from E. coli and Cytophaga hutchinsonii show the unique spatial arrangement of the αβα and all-α subdomains of the AAA+ domain in these proteins compared to the typical arrangement found in canonical AAA+ proteins such as HslU. The spatial organization of the subdomains in the AAA+ domain of MoxR proteins is similar to that found in the ATPase component of the magnesium chelatase complexes, possibly suggesting a similar mechanism of function. In this review, we provide an overview of the newly identified functions and the newly obtained structures of MoxR AAA+ ATPases.     

Antibodies to cytoplasmic dynein heavy chain map the surface and inhibit motility

Polyclonal antibodies have been raised against four 16 residue peptides with sequences taken from the C-terminal quarter of the human cytoplasmic dynein heavy chain. The sites are downstream from a known microtubule-binding domain associated with the "stalk" that protrudes from the motor domain. The antisera were assayed using bacterially expressed proteins with amino acid sequences taken from the human cytoplasmic dynein heavy chain. Every antiserum reacted specifically with the appropriate expressed protein and with pig brain cytoplasmic dynein, whether the protein molecules were denatured on Western blots or were in a folded state. But, whereas three of the four antisera recognized freshly purified cytoplasmic dynein, the fourth reacted only with dynein that had been allowed to denature a little. After affinity purification against the expressed domains, whole IgG molecules and Fab fragments were assayed for their effect on dynein activity in in vitro microtubule-sliding assays. Of the three anti-peptides that reacted with fresh dynein, one inhibited motility but the others did not. The way these peptides are exposed on the surface is compatible with a model whereby the dynein motor domain is constructed from a ring of AAA protein modules, with the C-terminal module positioned on the surface that interacts with microtubules. We have tentatively identified an additional AAA module in the dynein heavy chain sequence, which would be consistent with a heptameric ring.     

Head-head coordination is required for the processive motion of cytoplasmic dynein, an AAA+ molecular motor

Cytoplasmic dynein is an AAA(+)-type molecular motor whose major components are two identical heavy chains containing six AAA(+) modules in tandem. It moves along a single microtubule in multiple steps which are accompanied with multiple ATP hydrolysis. This processive sliding is crucial for cargo transports in vivo. To examine how cytoplasmic dynein exhibits this processivity, we performed in vitro motility assays of two-headed full-length or truncated single-headed heavy chains. The results indicated that four to five molecules of the single-headed heavy chain were required for continuous microtubule sliding, while approximately one molecule of the two-headed full-length heavy chain was enough for the continuous sliding. The ratio of the stroking time to the total ATPase cycle time, which is a quantitative indicator of the processivity, was approximately 0.2 for the single-headed heavy chain, while it was approximately 0.6 for the full-length molecule. When two single-headed heavy chains were artificially linked by a coiled-coil of myosin, the processivity was restored. These results suggest that the two heads of a single cytoplasmic dynein communicate with each other to take processive steps along a microtubule.     

Identification of a Manduca sexta NSF ortholog, a member of the AAA family of ATPases

Transport between intracellular compartments requires the activity of an N-ethylmaleimide-sensitive fusion protein (NSF). NSF is a member of a growing family of ATPases regulating several membrane fusion reactions. We have cloned the NSF ortholog from the moth, Manduca sexta (MsNSF). MsNSF is highly conserved in domains critical for NSF function in vertebrates. MsNSF codes for a protein of 745 amino acids, translating to a M_r of 83 kDa in vitro. MsNSF is 72% and 61% similar in amino acid sequence to Drosophila and vertebrate NSFs, respectively. We expressed the D1 ATP domain of MsNSF toward which antibodies selective to MsNSF were generated. Affinity purified -MsNSF antibodies detect a 83 kDa protein which is highly enriched in nervous tissues. Levels of MsNSF expression are substantially lower in other tissues examined. Anti-MsNSF antibodies are capable of inhibiting vertebrate intra-Golgi transport of a cargo protein in vitro. The identification of NSF ortholog from Manduca, whose neuroendocrine system is well studied, should facilitate isolation of complexes involved in protein trafficking from insect models. Phylogenetic analysis of NSF and related proteins suggests that the members of the AAA family arose from different ancestors, since the ingroup was not monophyletic. Proteasomal subunits and p97 homologs form two distinct subfamilies, while NSF homologs branch in to the third.     

Recombinant human cytoplasmic dynein heavy chain 1 and 2: observation of dynein-2 motor activity in vitro

Cytoplasmic dynein is a microtubule (MT) motor protein comprising two classes: dynein-1 and dynein-2. We purified recombinant human dynein-1 and dynein-2 from HEK-293 cells by expressing the streptavidin-binding peptide-tagged human cytoplasmic dynein-1 and dynein-2 heavy chains (HCs), respectively. Electron microscopy of the purified molecules revealed a two-headed structure composed of characteristic dynein motor domains. In an in vitro MT gliding assay, both dynein-1 and dynein-2 showed minus-end-directed motor activities. This is the first demonstration of dynein-2 motor activity, which supports the retrograde intraflagellar transport role of dynein-2. Our expression system of dynein HCs provides a useful means to investigate dynein functions.     

Light chain 1 from the Chlamydomonas outer dynein arm is a leucine-rich repeat protein associated with the motor domain of the gamma heavy chain.

The LC1 light chain from Chlamydomonas outer arm dynein is tightly bound to the gamma heavy chain. Molecular cloning revealed that LC1 is a member of the SDS22+ subclass of the leucine-rich repeat protein family and as such is likely involved in mediating interactions between dynein and the components of a signal transduction pathway. Through the combination of covalent cross-linking and vanadate-mediated photolysis, LC1 was found to associate with that portion of the gamma HC that is C-terminal to the P1 loop. This region comprises most of the globular head domain of the heavy chain and includes the stalk-like structure that is involved in microtubule binding. Attachment of LC1 to this region represents the only known example of an accessory polypeptide directly associated with a dynein motor domain. Additional cross-linking experiments revealed that LC1 also interacts directly in situ with an approximately 45 kDa axonemal component; this interaction is disrupted by the standard high salt treatment used to remove the outer arm from the axoneme. These data suggest that LC1 acts to mediate the association between this 45 kDa axonemal polypeptide and the motor unit of the gamma HC.     

Distinct but overlapping sites within the cytoplasmic dynein heavy chain for dimerization and for intermediate chain and light intermediate chain binding

Cytoplasmic dynein is a molecular motor complex consisting of four major classes of polypeptide: the catalytic heavy chains (HC), intermediate chains (IC), light intermediate chains (LIC), and light chains (LC). Previous studies have reported that the ICs bind near the N terminus of the HCs, which is thought to correspond to the base of the dynein complex. In this study, we co-overexpressed cytoplasmic dynein subunits in COS-7 cells to map HC binding sites for the ICs and LICs, as well as HC dimerization. We have found that the LICs bind directly to the N terminus of the HC, adjacent to and overlapping with the IC binding site, consistent with a role for the LICs in cargo binding. Mutation of the LIC P-loop had no detectable effect on HC binding. We detected no direct interaction between the ICs and LICs. Using triple overexpression of HC, IC and LIC, we found that both IC and LIC are present in the same complexes, a result verified by anti-IC immunoprecipitation of endogenous complexes and immunoblotting. Our results indicate that the LICs and ICs must be located on independent surfaces of cytoplasmic dynein to allow each to interact with other proteins without steric interference.     

Redox-based control of the gamma heavy chain ATPase from Chlamydomonas outer arm dynein

The outer dynein arm from Chlamydomonas flagella contains two redox-active thioredoxin-related light chains associated with the alpha and beta heavy chains; these proteins belong to a distinct subgroup within the thioredoxin family. This observation suggested that some aspect of dynein activity might be modulated through redox poise. To test this, we have examined the effect of sulfhydryl oxidation on the ATPase activity of isolated dynein and axonemes from wildtype and mutant strains lacking various heavy chain combinations. The outer, but not inner, dynein arm ATPase was stimulated significantly following treatment with low concentrations of dithionitrobenzoic acid; this effect was readily reversible by dithiol, and to a lesser extent, monothiol reductants. Mutational and biochemical dissection of the outer arm revealed that ATPase activation in response to DTNB was an exclusive property of the gamma heavy chain, and that enzymatic enhancement was modulated by the presence of other dynein components. Furthermore, we demonstrate that the LC5 thioredoxin-like light chain binds to the N-terminal stem domain of the alpha heavy chain and that the beta heavy chain-associated LC3 protein also interacts with the gamma heavy chain. These data suggest the possibility of a dynein-associated redox cascade and further support the idea that the gamma heavy chain plays a key regulatory role within the outer arm.     

Structure of the gamma heavy chain of the outer arm dynein from Chlamydomonas flagella

We describe here the vanadate-dependent photocleavage of the gamma heavy chain from the Chlamydomonas outer arm dynein and the pathways by which this molecule is degraded by endoproteases. UV irradiation in the presence of ATP, Mg2+, and vanadate cleaves the gamma chain at a single site (termed V1) to yield fragments of Mr 235,000 and 180,000. Irradiation in the presence of vanadate and Mn2+ results in cleavage of the gamma chain at two other sites (termed V2a and V2b) to yield fragment pairs of Mr 215,000/200,000 and 250,000/165,000. The mass of the intact chain is therefore estimated to be 415,000 D. We have located the major tryptic and staphylococcal protease cleavage sites in the gamma chain, determined the origins of the resulting fragments, and identified the regions which contain the epitopes recognized by two different monoclonal antibodies. Both antibodies react with the smaller V1 fragment; the epitope recognized by antibody 25-8 is within 9,000-52,000 D of the original gamma-chain terminus contained in that fragment, whereas that recognized by antibody 12 gamma B is within 16,000 D of the V1 site. The data permit the construction of a linear map showing the structural organization of the polypeptide. The substructure of the gamma chain is similar to that of the alpha and beta chains of the outer arm dynein with regard to polarity as defined by the sites of vanadate-dependent photocleavage, and to that of the beta chain with regard to a highly sensitive protease site located approximately 10,000 D from the original terminus contained in the smaller V1 fragment.     

Identification of a Manduca sexta NSF ortholog, a member of the AAA family of ATPases

Transport between intracellular compartments requires the activity of an N-ethylmaleimide-sensitive fusion protein (NSF). NSF is a member of a growing family of ATPases regulating several membrane fusion reactions. We have cloned the NSF ortholog from the moth, Manduca sexta (MsNSF). MsNSF is highly conserved in domains critical for NSF function in vertebrates. MsNSF codes for a protein of 745 amino acids, translating to a M_r of 83 kDa in vitro. MsNSF is 72% and 61% similar in amino acid sequence to Drosophila and vertebrate NSFs, respectively. We expressed the D1 ATP domain of MsNSF toward which antibodies selective to MsNSF were generated. Affinity purified α-MsNSF antibodies detect a 83 kDa protein which is highly enriched in nervous tissues. Levels of MsNSF expression are substantially lower in other tissues examined. Anti-MsNSF antibodies are capable of inhibiting vertebrate intra-Golgi transport of a cargo protein in vitro. The identification of NSF ortholog from Manduca, whose neuroendocrine system is well studied, should facilitate isolation of complexes involved in protein trafficking from insect models. Phylogenetic analysis of NSF and related proteins suggests that the members of the AAA family arose from different ancestors, since the ingroup was not monophyletic. Proteasomal subunits and p97 homologs form two distinct subfamilies, while NSF homologs branch in to the third.     

Relaxation-based structure refinement and backbone molecular dynamics of the dynein motor domain-associated light chain

The light chain 1 (LC1) polypeptide is a member of the leucine-rich repeat protein family and binds at or near the ATP hydrolytic site within the motor domain of the gamma heavy chain from Chlamydomonas outer arm dynein. It consists of an N-terminal helix, a central barrel formed from six leucine-rich repeats that fold as beta beta alpha units, and a C-terminal helical domain that protrudes from the main axis defined by the leucine-rich repeats. Interaction with the gamma heavy chain is likely mediated through a hydrophobic patch on the larger beta sheet face, and the C-terminal region is predicted to insert into the dynein ATP hydrolytic site. Here we have used 1H-15N heteronuclear relaxation measurements obtained at 500 and 600 MHz to refine and validate the LC1 solution structure. In this refined structure, the C-terminal helix is significantly reoriented by more than 20 degrees as compared to the control and provides a more precise understanding of the potential regulatory role of this domain. We also employed the refined structure to perform a dynamic analysis of LC1 using the 600 MHz data set. These results, which were cross validated using the 500 MHz data set, strongly support identification of the predicted LC1 binding surfaces and provide additional insight into the interaction mechanisms of leucine-rich repeat proteins.     

Point mutations in the stem region and the fourth AAA domain of cytoplasmic dynein heavy chain partially suppress the phenotype of NUDF/LIS1 loss in Aspergillus nidulans

Cytoplasmic dynein performs multiple cellular tasks but its regulation remains unclear. The dynein heavy chain has a N-terminal stem that binds to other subunits and a C-terminal motor unit that contains six AAA (ATPase associated with cellular activities) domains and a microtubule-binding site located between AAA4 and AAA5. In Aspergillus nidulans, NUDF (a LIS1 homolog) functions in the dynein pathway, and two nudF6 partial suppressors were mapped to the nudA dynein heavy chain locus. Here we identified these two mutations. The nudAL1098F mutation resides in the stem region, and nudAR3086C is in the end of AAA4. These mutations partially suppress the phenotype of nudF deletion but do not suppress the phenotype exhibited by mutants of dynein intermediate chain and Arp1. Surprisingly, the stronger DeltanudF suppressor, nudAR3086C, causes an obvious decrease in the basal level of dynein's ATPase activity and an increase in dynein's distribution along microtubules. Thus, suppression of the DeltanudF phenotype may result from mechanisms other than simply the enhancement of dynein's ATPase activity. The fact that a mutation in the end of AAA4 negatively regulates dynein's ATPase activity but partially compensates for NUDF loss indicates the importance of the AAA4 domain in dynein regulation in vivo.     

Characterization of the carp myosin heavy chain multigene family

We isolated partial coding sequences for 29 carp myosin heavy chain genes (MyoHCs) and determined the nucleotide sequences around the region encoding the loop 2 of the myosin molecule. The predicted amino acid sequences from the isolated genes all showed very high similarity to those of skeletal and cardiac muscles from higher vertebrates, but not to those of smooth and non-muscle counterparts. Among all clones isolated, carp MyoHC10, MyoHCI-1-3 and MyoHC30 showed exon-nucleotide sequences identical to those of cDNAs encoding the loop 2 region of the 10 degrees C-, intermediate- and 30 degrees C-type fast skeletal isoforms [Hirayama and Watabe, Euro. J. Biochem. 246 (1997) 380-387]. The loop 2 of 28 types of carp MyoHCs was encoded by two exons separated by an intron corresponding to that of the 16th in higher vertebrate MyoHCs, whilst this intron was not found in carp MyoHC30. Although carp MyoHC30 had a gene organization different from those of higher vertebrates and other carp MyoHCs, its predicted amino acid sequence for loop 2 showed the highest homology to those of higher vertebrates among carp MyoHCs. In the 28 carp MyoHCs containing the intron, a combination of different nucleotide sequences for the two resulted in 14 distinct series for the combined coding sequence. These different nucleotide sequences encoded nine distinct amino acid sequences. Phylogenetic analysis for the present loop 2 and light meromyosin previously reported for carp MyoHCs [Imai et al., J. Exp. Biol. 200 (1997) 27-34] revealed that carp MyoHCs have recently diverged and are more closely related to each other than to MyoHCs from other species.     

Three members of the LC8/DYNLL family are required for outer arm dynein motor function

The highly conserved LC8/DYNLL family proteins were originally identified in axonemal dyneins and subsequently found to function in multiple enzyme systems. Genomic analysis uncovered a third member (LC10) of this protein class in Chlamydomonas. The LC10 protein is extracted from flagellar axonemes with 0.6 M NaCl and cofractionates with the outer dynein arm in sucrose density gradients. Furthermore, LC10 is specifically missing only from axonemes of those strains that fail to assemble outer dynein arms. Previously, the oda12-1 insertional allele was shown to lack the Tctex2-related dynein light chain LC2. The LC10 gene is located approximately 2 kb from that of LC2 and is also completely missing from this mutant but not from oda12-2, which lacks only the 3' end of the LC2 gene. Although oda12-1 cells assemble outer arms that lack only LC2 and LC10, this strain exhibits a flagellar beat frequency that is consistently less than that observed for strains that fail to assemble the entire outer arm and docking complex (e.g., oda1). These results support a key regulatory role for the intermediate chain/light chain complex that is an integral and highly conserved feature of all oligomeric dynein motors.     

Mutations in the heavy chain of cytoplasmic dynein suppress the nudF nuclear migration mutation of Aspergillus nidulans

To identify proteins that interact directly or indirectly with the NUDF protein, which is required for nuclear migration in Aspergillus nidulans, we initiated a screen for extragenic suppressors of the heat-sensitive nudF6 mutation. Suppressor mutations in at least five genes, designated snfA–snfE, caused improved growth and nuclear migration at high temperatures compared to the nudF6 parent. Two snfC mutations mapped near the nudA gene, which encodes the cytoplasmic dynein heavy chain, and could be repaired by transformation with wild-type nudA DNA, demonstrating that they are mutations in nudA. The snfC mutations are bypass suppressors of nudF and genetic evidence indicated that NUDA and NUDF act in the same nuclear migration pathway. Taken together, our data suggests that NUDF affects nuclear migration by acting on the dynein motor system.     

Effect of extraction time on ability of calmodulin to activate 30S and 14S dynein ATPases

Cilia from the protozoan Tetrahymena pyriformis were demembranated and then extracted for 5 min with a buffer containing 0.5 M NaCl. The briefly extracted axonemal pellet was then reextracted for about 20 hr. The soluble material obtained from each extraction was resolved into 14S and 30S dynein ATPases by sedimentation on sucrose density gradients and tested for sensitivity to added calmodulin. The 14S dynein obtained by a 5-min extraction was generally insensitive to added calmodulin, whereas that obtained by 20-hr extraction of the 5-min extracted axonemes was activated by calmodulin, the activation being much larger in the “light” 14S fractions than in the “heavy” fractions. The 30S dynein ATPase obtained by a 5-min extraction was generally activated over 1.6-fold by added calmodulin, whereas that obtained by the subsequent long extraction was usually activated only 1.3-fold. After further purification of the 5-min extracted 30S dynein and of the 5-min to 20-hr-extracted 14S dynein on DEAE-Sephacel, these dyneins retained much of their calmodulin activatability. The ATPase activity of both 14S and 30S dyneins was inhibited more strongly by erythro-9-[3-(2-hydroxynonyl)] adenine and by vanadate in the presence of added calmodulin than in its absence. These data suggest that the only ATPase activity present in the fractions studied is that of the dyneins and demonstrate that both the 14S and 30S dynein ATPases may be obtained in forms mat are activated by added calmodulin as well as in forms that are insensitive to added calmodulin.     

Analysis of immunoglobulin heavy chain V-region genes belonging to the V NP-gene family.

A method was devised to clone immunoglobulin VH-region genes located on selected restriction fragments from genomic DNA directly into M13 vectors for subsequent nucleotide sequence analysis. Ten recombinant M13 clones representing four so far unknown VH-region genes of the VNP-gene family have been analysed. Sequence comparison shows that these genes are closely related to other VH-genes of the VNP gene family. One of the VH-genes exhibits a so far unobserved unusual length of 100 2/3 codons and appears to be functional. Analysis of the variation of the isolated VH-genes suggests that framework and complementarity determining regions are exposed to separate types of selective pressures.