The Complex Interplay between the Neck and Hinge Domains in Kinesin-1 Dimerization and Motor Activity^D⃞

Cloning of Constructs

The NcKin_hTail construct contains amino acid residues 1–436_NcKin from N. crassa conventional kinesin and a portion of the human kinesin tail (residues 432–546_HsKin), which promotes unspecific attachment of kinesin to surfaces in gliding assays. It was generated by standard PCR and cloning techniques and was shown to be indistinguishable from wild-type NcKin in gliding and steady state ATPase assays (Hahlen, unpublished results).

The chimeric constructs NcKin-379_hTail and NcKin-384_hTail contained NcKin sequences up to amino acid residue 379 or 384, respectively, followed by human kinesin sequences (Navone et al., 1992) up to residue 546 (Figure 1). Human kinesin-coding sequences were transferred to NcKin plasmids by PCR on a pT7-based vector that contained the human kinesin gene pK546_N₂K₂ with modified restriction sites (Kallipolitou, unpublished results). To facilitate cloning, silent mutations were introduced into the NcKin expression vector that 1) knocked out an endogenous KpnI site, 2) created a new one between the regions encoding neck and hinge domains, and 3) a NgoMIV restriction site between hinge and stalk. The vector, pNK433(C307A) also 4) contained the functionally unimportant, mutated cysteine residue C307A (Hahlen, unpublished results). A reverse primer at the 3′ end of the human kinesin gene with a downstream PstI site was used to generate PCR products of human kinesin in combination with forward primers that annealed at the homologous region of the introduced NcKin-KpnI site. For NcKin-379_hTail the forward primer fully matched the human kinesin sequence, for NcKin-384_hTail it encoded amino acid residues 378–384 from NcKin and the rest of human kinesin. Constructs lacking the human stalk (NcKin_hinge) were based on the respective NcKin_hTail pendants and were cloned by replacing the NgoMIV-PstI fragment with a short oligonucleotide dimer that encoded a cysteine tag (PSIVHRKCF; Funatsu et al., 1995) with stop codon and a downstream PstI site.

To introduce cross-links between the neck domains of the two polypeptide chains of a kinesin dimer, proline 342 was exchanged to cysteine, because this residue is expected to be in the first a-position of the α-helical neck coiled-coil (Schafer et al., 2003). Such an arrangement is known to form interstrand disulfide bonds readily (Zhou et al., 1993). P342 was mutated using a PCR-based mutagenesis protocol (Stratagene QuickChange, La Jolla, CA). The chimeric constructs were cloned into this background via the internal BamHI and NgoMIV restriction sites. The Stratagene Quickchange protocol was also used to generate the W384F point mutation. All plasmids were verified by sequencing.

Expression and Purification of Kinesin Constructs

Expression and purification of NcKin constructs was accomplished as described (Schafer et al., 2003). Briefly, 4 g of Escherichia coli cells were suspended in buffer A (20 mM Na-phosphate, pH 7.4, 50 mM NaCl, 5 mM MgCl₂, 1 mM EGTA, 1 mM dithiothreitol [DTT], 10 μM ATP, protease inhibitor mix (Roche Diagnostics, Penzberg, Germany), sonified, and centrifuged (35 min at 100,000 × g). The supernatant was loaded on a 5-ml HiTrap SP Sepharose column (Amersham Pharmacia Biotech, Piscataway, NJ) and eluted without DTT in a manual NaCl step gradient. The peak fractions were pooled, frozen in liquid N₂ with 10% glycerol, and stored at –70°C. Protein concentrations were evaluated by a Bradford assay.

Size Determination

To characterize the oligomerization state of NcKin constructs, sedimentation coefficients S_w,20 and Stokes' radii (r_stokes) of the constructs were determined by comparison with standard proteins.

S_w,20 coefficients were determined by centrifugation of 5–10 μM protein solutions on 5–18% sucrose gradient cushions in buffer B (20 mM Na-phosphate, pH 7.4, 200 mM NaCl, 5 mM MgCl₂, 10 μM ATP, 1 mM DTT) using the Beckman rotor SW 50.1 (Berkeley, CA) at ∼20,000 × g for 16 h at 4°C. As internal standards aldolase, bovine serum albumin (BSA), and carboanhydrase (S_w,20 = 7.4, 4.3, and 3.2, Roche Diagnostics) were included. The gradient was fractionated and analyzed on SDS-polyacrylamide gels. The position of the protein peaks were used to determine the S_w,20 values (NIH-Image program, Kaleidagraph software, Reading, PA).

The proteins' Stokes radii (r_stokes) were determined by gel filtration analysis. Protein solutions (8–15 μM) were loaded on a Sephadex 200 column (Amersham Pharmacia Biotechnol), equilibrated with buffer B. The elution volumes of the samples were compared with standard proteins (ferritin: 5.9 nm r_stokes, aldolase: 4.5 nm, BSA: 3.5 nm, carboanhydrase: 2.4 nm, cytochrome c: 1.6 nm; Roche Diagnostics and Sigma Chemical, St. Louis, MO). Stokes radii were determined from a plot of elution volumes versus standard sizes (Andrews, 1970). The molecular weights (M_r) of kinesin constructs were calculated according to Cantor and Schimmel (1980):

where n_A is the Avogadro number, η viscosity (10^–3 N/m^–1s^–1), ν is the specific volume of the protein (0.725 cm³/g), and ρ is the density of the medium (1.0 g/cm³).

Gliding Assays

To bind kinesin molecules on glass coverslips, NcKin_hTail constructs containing part of the human kinesin tail region (amino acid residues 432–546 in the human kinesin sequence) were used (Hahlen, unpublished results). After 1–3 min of incubation of 1.5 μg kinesin (20–30 μmol) on an 18 × 18-mm² glass coverslip, the assay was started by adding motility buffer (10 mM MgCl₂, 10 mM ATP, 10–20 mM KCl, and microtubules in BRB80+ [Pipes·KOH, 80 mM, 5 mM Mg-acetate, 0.5 mM EGTA, final pH 6.8 with 0.25–1 g/l Casein]) to a volume of 10 μl (Kallipolitou et al., 2001). Motility was monitored in a Zeiss Axiophot using video-enhanced phase-contrast microscopy (Thornwood, NY; Steinberg and Schliwa, 1996). To determine the gliding velocities, at least 20 microtubules were measured and averaged.

Steady State ATPase Activity

Microtubule-stimulated steady state ATPase rates were measured in a photometric assay that couples ATP hydrolysis to NADH oxidation by phosphoenolpyruvate, pyruvate kinase, and l-lactate dehydrogenase (Kallipolitou et al., 2001).

Chemical Cross-linking

Reversible disulfide bonds were formed between the two cysteines of dimeric NcKin_hTailP342C by incubation with 5,5′dithio-bis-(2-nitrobenzoic acid) (DTNB) or air (Tomishige and Vale, 2000; Hahlen, unpublished results). DTNB, 200 μM, was added to 10–30 μM kinesin containing the P342C mutation and incubated for 5 min. To verify the cross-link, aliquots were analyzed on SDS gels without reducing agents. To visualize air-oxidation, untreated kinesin constructs were mixed with sample buffer.

Disulfide bonds were broken by reduction with DTT. DTT, 2–5 mM, was added to 10–30 μM kinesin, incubated for 1–3 h, and then mixed with SDS sample buffer. For recovery after DTNB incubation (“rescue”), samples were incubated with at least 5 mM DTT for 4 h. SDS-PAGE was performed with all samples followed by Coomassie Blue staining.

To verify the behavior of the most sensitive mutant, NcKin-379_hTailP342C, 5 mM DTT was present in a control preparation. To minimize unspecific oxidation, this preparation was assayed immediately after purification. Oxidation was achieved by removal of DTT on a Sephadex G25 gel filtration column and air exposure or DTNB incubation.

To perform ATPase activity or gliding assays under reducing or oxidizing conditions, kinesin samples were treated as described above, starting the assays immediately after DTT or DTNB incubation. In gliding assays DTT or DTNB was present in the motility buffer.

Yeast Two-hybrid Assay

To test the interaction of kinesin's tail with the neck and/or hinge regions, the Matchmaker III system was used (Clontech/BD Biosciences, San Jose, CA). As bait construct, the NcKin tail domain from amino acid 725–928 was cloned into the plasmid pGBKT7 using the internal NcoI restriction site. N-terminal kinesin fragments were cloned into the prey vector pGADT7 by PCR on the plasmid pT12.1 (Henningsen and Schliwa, 1997; Seiler et al., 1999), yielding pNK343GADT7 (encoding the core motor domain, residues 1–343, NcKin_core), pNK378GADT7 (1–378, NcKin_neck), pNK400GADT7 (1–400, NcKin_hinge; due to cloning difficulties, this construct ends 33 codons earlier than NcKin_hinge from Figure 1), and pNK480GADT7 (1–480, NcKin_Tail; this construct contains the NcKin counterpart of human kinesin's stalk). DNA fragments encoding chimeric constructs were PCR-amplified from their respective expression plasmids, cut in restriction digestions, and used to replace the SacII/NgoMIV fragment of pNK480GADT7, encoding NcKin_TailW384F, NcKin-379_Tail, and NcKin-384_Tail. A second version of the NcKin_Tail prey construct was cloned this way (pNK480(C307A)GADT7), because all chimeric constructs were based on a modified expression vector with altered restriction sites and codon usage (pNK433(C307A), see above). The C307A point mutation did not change the behavior in the two-hybrid assay.

Bait and prey plasmids were cotransformed into the yeast strain AH109 using the PEG/LiAc method as described by BD Biosciences. Double transformants were selected on LW–plates. The interaction was screened on LWHA–plates where only colonies of strains with strongly interacting gene products grow or on LW-/X-Gal plates where interactions are indicated by blue colony coloration. These tests were performed on three independent transformants. To quantify the interaction, five independent yeast transformants were tested in an enzymatic assay for β-galactosidase activity using chlorophenol red-β-galactopyranoside (CPRG, Clontech Matchmaker manual) as a substrate. The activity was calculated as described by the manufacturer, which involves corrections for variable cell densities and assay times:

Conservation and Importance of W384

To investigate the influence of the proximal part of NcKin's hinge domain on neck coiled-coil formation, we constructed a series of mutants with altered hinges. Previous work had shown that Neurospora kinesin offers unique possibilities for such studies because, in contrast to animal kinesins, the presence of the hinge turned out to be a prerequisite for neck coiled-coil formation (Kallipolitou et al., 2001). A C-terminal deletion series had shown that truncated NcKin variants up to amino acid 383 are monomeric, but constructs comprising 391 or more residues are dimeric, demonstrating that the hinge residues 384–391 are crucial for neck coiled-coil formation. One of these truncated kinesins, NcKin_hinge, ending at the C-terminal end of the hinge after residue D433 and before the stalk coiled-coil, turned out to be a very useful construct with wild-type properties (stable dimer, gliding velocity typically 2.0–2.5 μm/s, steady state ATPase rate 60–80 s^–1). It is well expressed in E. coli and rather stable in vitro. We therefore used NcKin_hinge as the basis for our mutagenesis. We initially mutated a fungal-specific tryptophan, W384, located nine residues downstream of the end of the neck coiled-coil, into a phenylalanine. The exchange W→ F is the mildest possible natural side-chain substitution. The resulting mutant, NcKin_hingeW384F was monomeric and showed a decelerated ATPase rate (21 ± 6 vs. 66 ± 21 s^–1 for wild type, Table 1). This slow ATPase rate is typical for constructs that are inhibited by the residue Y362 in the neck domain (Schafer et al., 2003) and therefore suggests that the absence of W384 mimics the inactive state of NcKin where the coiled-coil conformation of the neck is disrupted.

To test the possible influence of the neighboring residues, we also constructed chimeric mutants consisting of NcKin head and neck domain, and variable amounts of the hinge domain. These N-terminal parts were fused to the rest of the human kinesin hinge as indicated in Figure 1. As expected, a chimera comprising NcKin 1–379 in front of the human hinge, and thus lacking the fungal-specific W384 (NcKin-379_hinge), failed to dimerize (Table 1). As the W384 F point mutant, the NcKin-379_hinge chimera showed a substantially decreased ATPase activity (27 s^–1), again consistent with the neck domain being in an unfolded state, as explained above. Together, the observations on NcKin_hingeW384F and NcKin-379_hinge were in full agreement with previous results from domain mapping studies that identified the region between amino acid 383 and 391 as a crucial element for NcKin dimerization. NcKin variants were monomeric when truncated before residue 384, but dimeric after 391 (Kallipolitou et al., 2001). The novel data on the W384F point mutant and the NcKin-379 chimera therefore suggested that W384 might be chiefly responsible for coiled-coil formation and thus dimerization of NcKin.

However, the chimera NcKin-384_hinge turned out to be a monomer as well, although it includes the important residue W384 (Table 1). As other monomeric mutants, it exhibited a slow ATP turnover (16 ± 3 s^–1). Therefore, W384 is not sufficient on its own to induce dimer formation and wild-type-like ATPase behavior in the NcKin_hinge protein background. Still, W384 seems to enhance or induce coiled-coil conformation of the neck domain, as indicated in a longer construct background.

The Human Kinesin Stalk Suppresses the Strict Requirement for W384

Further insights into the role of W384 and its vicinity were gained when we used constructs longer than 433 amino acids. To test NcKin mutants in microscopic gliding assays, we added part of the human kinesin stalk domain after the hinge (NcKin_hTail constructs, Figure 1; the part used for glass attachment did not comprise the tail region responsible for tail inhibition). The stalk portion causes unspecific adsorption to microscopic glass coverslips and thus facilitates multiple motor gliding assays. According to previous studies, we expected that mutants with a disrupted coiled-coil show slow and nonprocessive motility in these assays (Schafer et al., 2003). This was true for the NcKin-379_hTail chimera, which showed about one third of the wild-type gliding velocity (0.8 μm/s), and thus closely resembled the NcKin-Y362K point mutant with a disrupted neck coiled-coil (Table 1; Schafer et al., 2003). Unexpectedly, the NcKin_hTail W384F point mutant and the NcKin-384_hTail chimera both moved microtubules at wild-type velocity (2.3 and 2.2 μm/s), suggesting that the additional stalk sequence rescued the coiled-coil neck conformation. If this were true, W384 would behave like an element of a hydrophobic staple (Tripet and Hodges, 2002) in the NcKin_hTail background. However, the simple standard prove for neck coiled-coil formation— gel filtration/sucrose density centrifugation to test dimerization—is meaningless in this protein background, because the human stalk portion forms a coiled-coil and thus leads to a dimeric molecule regardless of the structural state of the neck domain (Vale and Fletterick, 1997). We therefore established an assay that probes neck coiled-coil formation in molecules that are dimeric due to the presence of additional domains.

Structural Information on the NcKin Neck Detected by Disulfide Bonds

To gain evidence for the structural integrity of kinesin's neck coiled-coil, we studied the distances of mutant neck domains between both subunits by introducing cysteine residues at the first coiled-coil a position (P342C mutations) and testing the capability of forming an intersubunit disulfide bridge. The formation of disulfide bonds requires close proximity of two –SH groups (∼2 Å) and has been shown to occur readily between opposing a positions of an intact coiled-coil (Zhou et al., 1993). To detect disulfide cross-links, we treated proteins with DTT (negative control), air, and DTNB and tried to revert the DTNB oxidation by adding excess DTT (rescue; Tomishige and Vale, 2000). Proteins incubated under these conditions were analyzed on nonreducing SDS gels, where uncrosslinked proteins appear at the calculated size of a single polypeptide, but proteins with intersubunit cross-links at multiples of this size.

Two negative controls were used in the cross-linking experiments: first, wild-type NcKin_hTail without introduced cysteines and second NcKin_hTail P342C, Y362K, which had been shown to contain a disrupted neck coiled-coil despite of being dimeric (Table 2; Schafer et al., 2003). As expected, both constructs were unaffected by oxidation and remained uncrosslinked under all conditions (Figure 2). By contrast, the NcKin_hTailP342C positive control showed bands at 205 kDa under untreated (i.e., air) and DTNB conditions that disappeared under DTT conditions, indicating formation of intersubunit disulfide bridges. The observed size of 205 kDa was higher than expected for a dimer, but gel filtration and sucrose density centrifugation experiments revealed a mass of 129 kDa, as expected for a dimer (Table 2). These results proved the feasibility of the method and agreed well with previous observations on synthetic peptides (Deluca et al., 2003; Schafer et al., 2003).

The chimeric mutants, however, behaved somewhat unexpectedly in this assay. Both the NcKin-384_hTail P342C and the -379_hTailP342C chimera showed formation of intersubunit cross-links in oxidizing environments, but only the former was expected to do so. As its cysteine-less parent, NcKin-384_hTailP342C behaved like constructs with an intact neck coiled-coil in motility assays (2.1 μm/s; Table 2). Notably, the P342C point mutation did not affect the gliding velocity in the NcKin_hTail background (2.1 μm/s). The NcKin-379_hTailP342C chimera also caused a similarly fast gliding velocity (2.1 μm/s), and SDS-PAGE on nonreducing gels confirmed that the neck of this mutant was cross-linkable. However, the cysteine-less NcKin-379_hTail parent showed slow motility and slow catalytic activity, reminiscent of constructs with a disrupted coiled-coil (Table 1; Kallipolitou et al., 2001; Schafer et al., 2003). The most intuitive interpretation, that the P342C mutation stabilized the coiled-coil conformation even in the reduced, disulfide-free situation, is unlikely because both the short NcKin-379_hingeP342C and the NcKin-384_hingeP342C chimera are monomeric in the presence of DTT (Supplementary Table S1).

Therefore, most likely the reduced activity of the NcKin-379_hTail parent construct is due to a disrupted neck coiled-coil structure. However, coiled-coil and non–coiled-coil structures seem to be very close to thermodynamic equilibrium. Support for this came from ATPase measurements on a control preparation of NcKin-379_hTailP342C performed in the presence of 5 mM DTT to exclude unintended autooxidation during protein purification. ATPase activity was tested immediately after the purification and without adding glycerol and in the presence of high levels of DTT (at least 5 mM). Under these reducing conditions the steady-state ATP turnover rate decreased to 15 s^–1 (even lower than 27 s^–1 for the NcKin-379_hTail parent) and was accelerated 2.5-fold to 37 s^–1 upon removal of DTT by gel filtration. This result suggests that 1) NcKin-379_hTailP342C oxidizes very easily, and has to be kept under high DTT concentrations to prevent disulfide bond formation. If kept reduced properly, 2) the ATPase is slow but can be accelerated by oxidation, which is most easily explained by the reformation of the previously disrupted neck coiled-coil conformation.

Unlike in ATPase assays, the NcKin-379_hTailP342C mutant was as fast as wild-type NcKin in gliding assays under all conditions, even in the control preparation in the presence of 5 mM DTT (Table 2). However, multiple motor gliding assays are somewhat deceiving because the gross action that is measured does not necessarily average the behavior over the entire population, but may be dominated by a small fraction of very active motors. To test this possibility, we mixed NcKin-379_hTail (0.8 μm/s) with wild-type kinesin (app. 2.5 μm/s) in variable amounts and measured the resulting gliding velocities (Supplementary Material). The results of these experiments suggested that a fraction of <20% of motor with intact neck coiled-coil was able to mask the mutant effect and to produce wild-type gliding velocity. We therefore think that the apparently undisturbed gliding velocity of NcKin-379_hTailP342C under reducing conditions is an experimental artifact that is due to a small fraction (5–10%) of oxidized mutant motor protein that is already present in the recombinant E. coli protein.

Together, the results described above suggest a complex interplay between the single amino acid residue W384, its immediate neighborhood, and the more distant neck and tail coiled-coil structures. 1) The replacement of the indole ring of W384 with a phenyl group (F384) leads to the disruption of the α-helical neck coiled-coil structure in the otherwise rock-solid NcKin_hinge dimer background. 2) The strict requirement of W384 for neck coiled-coil formation is suppressed when dimerization is induced by kinesin's stalk, and phenylalanine can replace tryptophan in this case. 3) Still, W384 and its close neighbors facilitate neck coiled-coil formation, because the NcKin-379_hTail chimera retains the non–coiled-coil neck structure, whereas the NcKin-384_hTail contains a functional neck coiled-coil. The fact that NcKin_hTailW384F also forms an active neck conformation (despite the lack of W384) may indicate that either the W→F exchange is mild because it retains an aromatic ring or that elements from the close vicinity (residues 385–391; Kallipolitou et al., 2001) contribute to the effect of W384. Thus, W384 is neither absolutely necessary nor completely sufficient for neck coiled-coil formation under all circumstances, but seems to cause its formation when present in its proper environment.

What may be the physiological role of this proximal part of the hinge domain? Previous studies strongly argue for intramolecular regulation where the tail folds back on the head/neck/hinge region to inactivate the motor when no cargo is bound (Coy et al., 1999; Friedman and Vale, 1999; Kirchner et al., 1999; Hackney and Stock, 2000; Seiler et al., 2000). To test whether the proximal hinge plays a role in this regulation mechanism, we studied the tail–neck/hinge interaction using yeast two-hybrid assays.

The Hinge near W384 May Be a Target of Motor Regulation by the Tail

To study the neck/hinge–tail interaction in a yeast two-hybrid assay, part of the tail domain (amino acids 725–928) was used as the bait construct and cotransformed into yeast with prey plasmids encoding different regions of the head/neck/hinge domains (Figure 3). The strongest interaction was found in wild-type prey constructs that that encoded NcKin_Tail, (amino acids 1–480_NcKin; cf. Material and Methods for details). Yeast transformants with these preys even grew under the most restrictive conditions on LWHA–plates. No interaction was seen when a shorter tail construct (amino acid 725–844) was used, consistent with the tail interaction site being close to the globular C-terminus (Kirchner et al., 1999; Seiler et al., 2000). All shorter NcKin prey constructs, as well as all kinesin chimera and point mutants were negative under these stringent growth conditions.

The NcKin_Tail prey construct and the truncated NcKin_hinge were also positive on LW–plates containing X-α-Gal, suggesting a robust tail interaction in constructs that contain the wild-type hinge domain (Figure 3). The shorter NcKin_core and NcKin_neck constructs were clearly negative on X-α-Gal plates; the point mutant NcKin_TailW384F, as well as the chimeric NcKin-379_Tail and -384_Tail constructs were highly variable in this assay. Therefore, five independent transformants of each bait-prey pair were compared in parallel in an enzymatic assay, using CPRG as a substrate. The result confirmed the observations of the previous assays and revealed a 3–5-fold higher β-galactosidase activity with the NcKin_Tail and NcKin_neck prey constructs than with the other mutants (Figure 3). In this assay, the W384F point mutant and both chimeras were clearly negative.

These findings indicate that the proximal region of the hinge, from amino acid 378 to amino acid 400, which includes W384, may be a target site for tail-mediated inhibition of motor activity. The fact that the NcKin-384 chimera was negative in all two-hybrid assays indicates that residue W384 alone is not sufficient for a strong interaction but requires additional motifs further C-terminal.

Sequence Requirements for Folding of the NcKin Neck Domain

The study of two chimeras, one retaining W384 (NcKin-384) and one lacking it (NcKin-379) shows that only the former displays normal gliding and ATPase activities. In addition to W384, the mutants differ in residues K381, E382, and K382, and based on the sequence alignments, both lysine residues might also contribute to the observed differences, although neither of these residues is strictly conserved in fungal kinesins (Figure 1). Cross-linking assays that probe the integrity of the neck coiled-coil indicate that only the chimera that comprises W384 is able to form an α-helical neck coiled-coil in the NcKin_hTail background. Data on the W384F point mutant NcKin_hTailW384F that does not contain W384 but still moves quickly and forms a functional neck domain seem to contradict this interpretation. However, the side chain of the introduced F384 still contains an aromatic ring and thus might still be able to contribute to possible π -stacking interactions (as e.g., observed by Zheng et al., 2004). Such stacking interactions have been found to increase the stability of the human kinesin neck, but at a completely different place (Tripet and Hodges, 2002). Alternatively, additional residues C-terminal of W384 may contribute to the stabilizing effect of the hinge on the neck coiled-coil and be sufficient for neck coiled-coil formation in constructs that are dimerized via the stalk (NcKin_hTail), but insufficient in those that lack this region (NcKin_hinge). We suspect that both stacking interactions and neighboring residues of W384 contribute to the stabilizing effect of the hinge domain.

Possible Mechanism of Neck Coiled-coil Stabilization

As far as known, all kinesin-1 neck domains are able to form coiled-coils. Formally, coiled-coils are stabilized by the hydrophobic interface between the strands, and the exclusion of water leads to a gain in entropy that drives dimerization. However, the sequences and spectroscopic data on kinesin-1 neck peptides show that these coils are far from being optimal. Therefore, even subtle mutations in the NcKin neck/hinge region apparently lead to less favorable entropy changes and prevent kinesin from dimerizing. However, these unfavorable thermodynamic changes can be overcome in some cases by the presence of the stalk coiled-coil that provides a positive offset for the entropy change. How do these general considerations apply to the case of NcKin?

There are several obvious deviations from the perfect heptad pattern that leads to an α-helical coiled-coil structure. Particularly, the N-terminal half of the human kinesin neck is known to contain several nonideal, hydrophilic or bulky residues in the a and d heptad positions that at first glance seem to exclude a coiled-coil conformation (Morii et al., 1997; Tripet et al., 1997; Thormahlen et al., 1998). However, a closer look reveals the presence of a novel set of so-called helix-capping motifs outside the coiled-coil that are able to compensate the destabilizing effect of the nonideal residues (Tripet and Hodges, 2002). Here, a tryptophan residue was found to be part of a “molecular sandwich” or “hydrophobic staple” that stabilizes the coiled-coil conformation. The term “sandwich” was used to describe layered aromatic amino acid side chains that are stabilized by hydrophobic or π -stacking interactions.

In fungi a typical subgroup of kinesin-1 motors evolved that shows obvious sequence dissimilarities to animal kinesin-1s in the N-terminal part of the neck, but is very similar toward the C-terminal end of the neck. Therefore the stabilizing effect of the proximal part of the fungal-specific hinge most likely involves interactions with the N-terminus or central part of the neck. Is such a long-distance interaction likely to occur?

NMR spectroscopy on rat kinesin hinge peptides show that the N-terminal portion of the hinge consists of short, well-defined secondary structure elements (Seeberger et al., 2000) that are likely to occur also in fungal kinesins because of substantial sequence similarities at the neck-hinge junction (Figure 1). According to the NMR data, the glycine-rich neck/hinge junction is highly flexible and allows the hinge to explore a large volume in space almost freely. In the C-terminal region of the neck-hinge peptides, several residues downstream of the neck coiled-coil, a relatively stiff motif (E374-Q381 in rat kinesin) is found that includes the counterpart of the NcKin W384 residue, Q381_rat (Figure 4). Apparently, the three-dimensional structure of kinesin's hinge allows interactions between hinge and neck coiled-coil.

Multiple Structural States of the NcKin Neck Domain

One of the most intriguing findings of this study is the evidence for multiple structural states of the NcKin neck domain. In our study we used three approaches to tackle the coiled-coil conformation of the NcKin neck: 1) kinetic evidence, based on previous experience (Schafer et al., 2003), 2) dimerization assays on truncated NcKin_hinge motors, and 3) the cross-linking behavior of a mutant with introduced cysteine residues (NcKin_hTailP342C). These three lines of evidence together give a rather complex, but informative picture on kinesin mechanism and regulation.

Kinesin-1 type motors (conventional kinesins) seem to move by a hand-overhand mechanism that produces fast, processive, stepwise motility. This mechanism relies on the coordinated action of two coupled motor heads, which, in turn, requires the mechanical coupling of the two heads via the coiled-coil neck domain. In one extreme model, the neck coiled-coil is static and does not need to “breathe” or “melt” during the stepping process (Romberg et al., 1998; Tomishige and Vale, 2000). Our motility data on cross-linked kinesin mutants support this simplistic model, because the position of the introduced disulfide bond (P342C) does not permit dissociation of the neck coiled-coil but still allows fast motility. Still, our experiments also indicate that the neck domain may assume conformations other than an α-helical coiled-coil. The solution is that the hand-overhand model only describes the mechanical cycle of kinesin, but does not exclude that the neck refolds for regulatory purposes.

In the present study, the NcKin-379_hTail chimera and the NcKin_hTailP342C,Y362K mutant obviously deviate from wild-type NcKin, and both do not form a coiled-coil neck, as indicated by their motility properties and their behavior in dimerization assays in the NcKin_hinge background (cf. Schafer et al., 2003). However, unlike the P342,Y362K double mutant, the NcKin-379_hTailP342C chimera can be cross-linked, indicating different structural states of the neck domain in these mutants. The strict conservation of Y362 among fungal kinesins and the fact that its absence cannot be suppressed indicates that the structure of the neck in NcKin-Y362K mutants may not be physiological. By contrast, the easy oxidation of reduced NcKin-379_hTailP342C that leads to a functional conformation suggests that its inactive, non–coiled-coil structure is not just “denatured” but represents a down-regulated state. Together, these observations suggest that stabilizing and destabilizing entropic effects of specific sequence elements are in such a close balance in NcKin that it is conceivable that different neck conformations occur and are easily induced by external influences.

Physiological Role of Different Neck Conformations

Our present data fit well in a model that attributes the switch between the coiled-coil and the non–coiled-coil neck conformation to the action of the proximal hinge region around W384. According to this model, the proximal part of the hinge would either interact with the tail domain, leading to the inhibitory, non–coiled-coil conformation of the neck, or with the neck, leading to the active coiled-coil structure (Figure 4).

The regulatory influence of the tail domain on the motor head has been described in several studies on animal and fungal kinesin-1 motors (Friedman and Vale, 1999; Seiler et al., 1999; Stock et al., 1999). Most likely, the ADP release is inhibited in the folded kinesin state where the conserved IAK tail sequence motif signals the absence of cargo to the motor domain (Hackney and Stock, 2000). In fungal kinesins, the inhibitory effect of the neck is largely due to the fungal-specific residue Y362 whose exchange leads to constitutively fast and active kinesins (Schafer et al., 2003). However, the exact binding target of the tail and details of the inhibitory mechanism have not been described so far. Our yeast two-hybrid assays show that the tail domain interacts with the N-terminal part of the hinge domain (379–400), suggesting that it may no longer be free to interact with the neck domain and to stabilize its coiled-coil conformation. This scenario predicts a conformation where hinge and tail interact, and the neck assumes a non–coiled-coil conformation that inhibits the motor domain via Y362 (Figure 4). The tail binding, however, seems not to rely solely on W384 because the yeast two-hybrid interaction was absent also in chimera that comprised W384. Apparently, the near environment is required for tail binding as well, and might supply stabilizing interactions or impose structural constraints.