(Received for publication, December 4, 1996, and in revised form, January 15, 1997)
From the Department of Biochemistry and the
§ Medical Research Council Group in Protein Structure and
Function, University of Alberta, Edmonton, Alberta T6G 2H7, Canada and
the ¶ Howard Hughes Medical Institute and Departments of
Pharmacology and Biochemistry/Biophysics, University of California, San
Francisco, California 94143
Kinesin is a dimeric motor protein that can move
for several micrometers along a microtubule without dissociating. The
two kinesin motor domains are thought to move processively by operating in a hand-over-hand manner, although the mechanism of such
cooperativity is unknown. Recently, a ~50-amino acid region adjacent
to the globular motor domain (termed the neck) has been shown to be
sufficient for conferring dimerization and processive movement. Based
upon its amino acid sequence, the neck is proposed to dimerize through a coiled-coil interaction. To determine the accuracy of this prediction and to investigate the possible function of the neck region in motor
activity, we have prepared a series of synthetic peptides corresponding
to different regions of the human kinesin neck (residues 316-383) and
analyzed each peptide for its respective secondary structure content
and stability. Results of our study show that a peptide containing
residues 330-369 displays all of the characteristics of a stable,
two-stranded -helical coiled-coil. On the other hand, the
NH2-terminal segment of the neck (residues
~316-330) has the capacity to adopt a
-sheet secondary structure.
The COOH-terminal residues of the neck region (residues 370-383) are
not
-helical, nor do they contribute significantly to the overall
stability of the coiled-coil, suggesting that these residues mark the
beginning of a hinge located between the neck and the extended
-helical coiled coil stalk domain. Interestingly, the two central
heptads of the coiled-coil segment in the neck contain conserved,
"non-ideal" residues located within the hydrophobic core, which we
show destabilize the coiled-coil interaction. These residues may enable
a portion of the coiled-coil to unwind during the mechanochemical
cycle, and we present a model in which such a phenomenon plays an
important role in kinesin motility.
Understanding how motor proteins generate force and movement from the chemical hydrolysis of ATP remains one of the most intriguing problems in biophysics. At present, there are three separate families of motor proteins found within eukaryotic cells: myosins, which move along actin filaments, and kinesins and dyneins, which move along microtubules (1). The motor domains that typify each of these superfamilies exhibit little or no amino acid sequence similarity, and hence it was believed that they had evolved separately and were structurally unrelated. However, the recently determined crystal structure of kinesin revealed an unexpected structural similarity to the core of the myosin motor domain, particularly in the nucleotide binding pocket (2, 3). Hence, myosin and kinesin may share some similarities in how they generate unidirectional movement and force, although the precise mechanistic details remain to be elucidated for both types of motors.
Kinesin has proven to be an excellent model system for investigating
the mechanism of motility, in part due to the small size of its motor
domain (>2-fold smaller than myosin's). Kinesin purified from tissue
sources exists as an 2
2 heterotetramer,
in which two
subunits (heavy chains) and two
subunits (light
chains) associate to form a highly elongated molecule with globular
termini (4, 5). The kinesin heavy chains are organized into four domains (listed from NH2 to COOH terminus): (i) a
~325-amino acid residue globular motor domain head that contains the
ATP and microtubule binding sites, (ii) a ~50-amino acid residue
region adjacent to the globular motor domain (termed the neck region)
that is sufficient for allowing dimerization of the motor domains (6)
and contains a sequence that is predicted to form an
-helical coiled
coil (6, 7), (iii) a long (~450-amino acid residue)
-helical coiled-coil domain (termed the stalk), and (iv) a small globular COOH
terminus (termed the tail) (8-11). Flexible "hinge" regions are
found between the neck and the stalk and in the center of the stalk.
The light chains (
subunits) of kinesin, which are not necessary for
force-generation, are associated with the smaller globular COOH
terminus of the heavy chain and may be involved in determining cargo
specificity (8).
A single kinesin molecule can move continuously along a microtubule for several micrometers in a series of 8-nm steps, which corresponds to the distance between tubulin binding sites along the microtubule protofilament (12). Such processive movement, which is not displayed by muscle myosin or ciliary dynein, very likely represents a specialized adaptation that enables a few kinesin motors to transport membrane organelles efficiently within cells. Functional studies on recombinantly expressed kinesin heavy chains have begun to uncover regions that are necessary for kinesin motility. Bacterial expression of the first 340 amino acids of the Drosophila kinesin heavy chain (which contains the core NH2-terminal globular motor domain and the first ~10 amino acids of the neck) produces a monomeric protein that generates directed motility when many motors are interacting simultaneously with a single microtubule in gliding motility assays (7, 13). However, these monomeric kinesins do not exhibit processive movement when assayed as individual motors in a single molecule fluorescence motility assay (14) or a bead assay (15). A kinesin motor containing the complete motor and neck domains, on the other hand, forms a dimer and also exhibits processive movement (14, 16). Collectively, these studies suggest that the dimeric structure of kinesin is not essential for force-generation per se, although it does appear to be required for processive movement. This raises the possibility that processive movement may involve a hand-over-hand coordination of the two kinesin heads.
The proposed -helical coiled-coil domain in the kinesin neck may be
structurally important for coordinating the activities of the two
kinesin heads during processive movement. The existence of a
coiled-coil structure in close proximity to the motor domains, however,
raises important questions concerning its exact boundaries and
stability, since the connection between the heads must be sufficiently
extensible to allow the two motor domains to span the distance between
two tubulin dimers during movement. It is also important to determine
the thermodynamic properties of the neck coiled-coil to ascertain if it
could partially or totally "un-coil" during the generation of a
power stroke. Unfortunately, the atomic resolution structure of the
neck domain is unknown, since the segment between residues 323 and 349 is disordered and hence invisible in the present electron density maps
of human kinesin (hK349) (2). Thus, to gain insight into the structure of the kinesin neck region and its possible functional roles, we have
investigated the secondary structure of the human kinesin heavy chain
neck region using several synthetic peptides in conjunction with CD
spectroscopy.
In the present study, we report that a two-stranded, -helical
coiled-coil dimerization domain exists between residues 330-369 within
the human kinesin neck region, as predicted from previous work.
Residues located to the COOH terminus of this region, 370-383, appear
to be unstructured and are not significantly involved in further
stabilization of the dimerization domain. Residues located to the
NH2 terminus of the proposed coiled-coil dimerization
domain may adopt a
-sheet secondary structure. Analysis of the
stability of each peptide indicates that the heptads required to form a stable coiled-coil domain are arranged in a strong-weak-strong manner.
Loss of two heptads from either the NH2 or COOH terminus significantly affects dimer stability. These results suggest that a
conformational change in the motor domain, driven by a free energy
change associated with ATP hydrolysis, could be transmitted in a manner
that affects the stability and/or conformation of the adjacent neck
region. We propose a model for kinesin motility in which unwinding of a
portion of the coiled-coil domain plays an important role in the
mechanochemical cycle.
Synthetic kinesin
peptides were prepared by solid-phase synthesis methodology using a
4-benzylhydrylamine hydrochloride resin with conventional
N-t-butyloxycarbonyl chemistry on an Applied Biosystems
model 430A peptide synthesizer as described by Sereda et al.
(17). Peptides were cleaved from the resin by reaction with hydrogen
fluoride (20 ml/g resin) containing 10% anisole and 2%
1,2-ethanedithiol for 1 h at 5 °C, washed with cold ether several times, extracted from the resin with glacial acetic acid, and
then lyophilized. Purification of each peptide was performed by
reversed-phase high performance liquid chromatography
(RP-HPLC)1 on a SynChropak semi-preparative
C-8 column (250 × 10 mm, inner diameter, 6.5-µm particle size,
300-Å pore size; SynChrom, Lafayette, IN) with a linear AB gradient
(ranging from 0.2 to 1.0% B/min) at a flow rate of 2 ml/min, where
solvent A is aqueous 0.05% trifluoroacetic acid and solvent B is
0.05% trifluoroacetic acid in acetonitrile. Homogeneity of the
purified peptides were verified by analytical RP-HPLC, amino acid
analysis, and electrospray quadrapole mass spectrometry.
Oxidation of kinesin peptides (formation of a disulfide bond to form a homo-two-stranded molecule) was carried out by dissolving 5 mg of peptide into 2 ml of 100 mM NH4HCO3, pH 8 buffer and stirring overnight in an open reaction vessel. The oxidized peptides were then re-purified by RP-HPLC and characterized by mass spectrometry (as described above).
Circular Dichroism SpectroscopyCircular dichroism
(CD) spectra were recorded on a Jasco J-720 spectropolarimeter (Jasco
Inc., Easton, MD) interfaced to an Epson Equity 386/25 computer running
the Jasco DP-500/PS2 system software (version 1.33a). The
temperature-controlled cuvette holder was maintained at 20 °C with a
Lauda model RMS circulating water bath (Lauda, Westbury, NY). The
instrument was calibrated with an aqueous solution of re-crystallized
d-10-(+)-camphorsulfonic acid at 290.5 nm. Results are
expressed as mean residue molar ellipticity []
(deg·cm2·dmol
1) calculated from
Equation 1.
![]() |
(Eq. 1) |
Denaturation midpoints, slopes, and free energy of unfolding values for the various kinesin peptides (see Table I) were determined by following the change in molar ellipticity at 222 nm using a Jasco J-720 spectropolarimeter (as described above). Ellipticity readings were normalized to the fraction of the peptide folded (ff) or unfolded (fu) using the standard equations shown (Equations 2 and 3).
![]() |
(Eq. 2) |
![]() |
(Eq. 3) |
![]() |
(Eq. 4) |
|
Molecular weights of the peptides in aqueous solution were determined by size-exclusion chromatography (SEC) with laser light scattering. SEC was carried out on a Superose 12 column (1.0 cm × 30.0 cm) from Pharmacia at a flow rate of 0.5 ml/min at room temperature. The eluent was a 100 mM KCl, 50 mM K2HPO4, pH 7, buffer. The effluent from the column was monitored using either a Hewlett Packard UV-visible spectrophotometer at 210 nm, or a Dawn F multiangle laser light scattering photometer connected in series with a Optilab 903 refractometer. Determination of molecular weights (by laser light scattering) was carried out according to the methodology described by Farrow et al. (21).
Helical Propensity/Hydrophobicity AnalysisThe score for
-helical propensity and hydrophobicity (occurring in a 3-4
repeating pattern) was calculated for residues 280-420 of the kinesin
protein. Each
-helical propensity data point was obtained by an
iterative process involving the summing of 11 individual
-helical
propensity scores (22) for the sequence starting at residue position
280. Subsequent data points were then obtained by shifting 1 residue
toward the COOH terminus and repeating the process. Hydrophobicity
values, occurring in a 3-4 repeating pattern, were calculated in a
similar manner by summing the hydrophobicity scores (22) for the first
6 residues that occur in a "3-4" repeating pattern
(i.e. residues 1, 4 8, 11, 15, and 18). Subsequent data points for the same face were then calculated by shifting first 3 residues (4, 8, 11, 15, 18, and 22), then 4 residues (8, 11, 15, 18, 22, and 25), and so forth throughout the sequence to maintain the same
reading frame. Calculation of the other six faces was carried out by
starting +1 residue toward the COOH terminus (2, 5, 9, 12, 16, and 19)
etc. to obtain data for the second face and another +1 residue shift
(3, 6, 10, 13, 17, and 20), etc., for face 3. This process is repeated
to obtain data for faces 4 to 7.
To examine the
structural characteristics of the human kinesin neck region, several
synthetic peptides (see Fig. 1 for sequences) that span
various regions between residues 316 and 383 were prepared and analyzed
by circular dichroism (CD) spectroscopy. Fig. 2
(panels A and B) shows the far UV CD spectra for
the kinesin peptides K1-K5. Kinesin peptides K2, K3, K4, and K5 all
show characteristic -helical spectra with double minima at 208 and
222 nm (23, 24). The K2 peptide, which represents residues 330-369 of
the kinesin neck region, displayed the greatest molar ellipticity (
24,600°), corresponding to ~70%
-helical content or ~28
helical residues (see Table I). The addition of two
heptads (14 residues) onto the COOH terminus of this region (K5,
330-383) results in a decrease in the molar ellipticity to
19,500°, indicating that the COOH-terminal residues 370-383 added
are not
-helical. The calculation of the helical content for K5 is
54% or ~29 helical residues, which indicates no loss of the existing
helical residues from the region 330-369. The addition of two heptads
onto the NH2 terminus of this region (K4, 316-369) also
results in a significant decrease in the molar ellipticity to
13,300°, again indicating the addition of non-helical residues onto
the helical domain located between 330 and 369. Calculation of the
helical content for the peptide K4 is 37% or ~20 helical residues,
which represents a loss of ~8 helical residues from the region
330-369. Analysis of the kinesin peptide K3 (344-383), which
represents a 40-residue peptide shifted two heptads (14 residues)
toward the COOH terminus, shows a molar ellipticity at 222 nm of
16,500°, corresponding to a helical content of 46% or ~19
helical residues. This represents a loss of ~9 helical residues for
the deletion of the two heptads 330-343, indicating that the two
NH2-terminal heptads of 330-343 are very important for the
helical content observed in the region 330-369. One can also see from
the spectra of the kinesin peptides in Fig. 2A that the
ratios of [
]222 to [
]208, which are
often used as an indication of coiled-coil formation, are >1 for K2 and K5, but <1 for K3 and K4, suggesting a transition from a possible two-stranded
-helical coiled-coil to a single-stranded
-helix (25-30).
CD analysis of the kinesin peptide K1 (residues 316-355), which
represents a 40-residue peptide shifted two heptads toward the
NH2 terminus (from the region 330-369), reveals a complete absence of -helical content. In fact the secondary structure of this
peptide now displays a
-sheet pattern (23). It is important to note,
however, that the spectrum of this peptide could not be acquired under
the same benign conditions like those used for the other peptides. At
pH 2.5, K1 is highly soluble and shows only a random coil spectrum.
Increasing the pH successively from 2.5 to 5.5 results in a major
transition from a largely random coil spectrum to that of a
-sheet
spectrum, with the greatest transition occurring between pH 4.5 and
5.5. Analysis of the peptide above pH 5.5 was not possible due to the
complete gelation of the solution. The pH dependence of this transition
(pH 4.5-5.5) suggests that the ionization of glutamic acid residues
are involved. Unfortunately, it is not possible to ascertain from these
results whether formation of the
-sheet secondary structure is a
result of intramolecular or intermolecular interactions, or a
combination of both. However, our observations with K4 (which contains
this region) showing a loss in helical content of ~8 residues but no substantial loss in stability compared with K2 (discussed below) suggests that their is some intramolecular formation at the
NH2 terminus of this peptide. The gelation of the solution
also suggests the formation of intermolecular association as well.
Interestingly, in the presence of 50% TFE, a helix-inducing
solvent (26, 31), the K1 peptide reverts to a fully -helical spectrum (equal to that calculated for a 40-residue peptide (23, 24).
This result indicates that this region of kinesin has the intrinsic
ability to adopt either
-sheet or helical secondary structures
depending on the environment. It should be noted that the first few
amino acids of this peptide (316-320) are in a helical configuration
in the kinesin crystal structure (2).
Although the -helical
content of the K2 peptide (residues 330-369) is significantly greater
than that of the other native kinesin peptides, it still does not
represent a fully helical structure as calculated theoretically for a
40-residue peptide (~
36,500°) (24). We therefore determined if
oxidation of the NH2-terminal cysteine (Cys330)
to form a disulfide bridge could increase the helical content to that
of the theoretical value by stabilizing the ends of the proposed
coiled-coil as well as by making the coiled-coil dimerization domain
concentration independent (29, 32, 33). When the peptide was oxidized,
a change in its molar ellipticity was observed (see Fig.
3). Oxidation only slightly increased the molar
ellipticity at 222 nm by approximately 2000°, which is similar to
that obtained for the reduced K2 peptide in the presence of 50% TFE, a
helix-inducing solvent (26, 31). This degree of ellipticity indicates
that the reduced (monomeric) K2 peptide is almost fully helical (93%) if judged by the maximal amount of helical content that can be induced
either in an oxidized state or in 50% TFE. Although previous studies
have shown that theoretical maximum values are not always observed for
helical peptides, the observation of a lower molar ellipticity than the
theoretical value may also be indicating that there is a region within
residues 330-369 that cannot be induced into a fully
-helical
structure by either a helix-inducing solvent or oxidation. The central
region containing residues Tyr344, Glu347, and
Asn351 in the hydrophobic core is a good candidate for such
a region and will be discussed further below.
Two further points can be made regarding the oxidation results. First,
the finding that the -helical secondary structure for the K2 peptide
is enhanced instead of disrupted by disulfide bond formation indicates
that the two peptides interact in a parallel and in-register manner.
Second, while disulfide bond formation of Cys330 can occur
in a peptide, this may not be possible when the neck is joined to the
globular motor domain. This cysteine is also not conserved among
conventional kinesin motor proteins from different species.
An important question raised regarding the existence of
a coiled-coil within the neck region was whether it alone is
sufficiently stable to account for the dimerization of the kinesin
motor domain heads or whether other subunit-subunit interactions are
also involved. To address this question, we determined the stability of
the five kinesin peptides by GdnHCl denaturation (Fig.
4A). The K2 peptide (330-369), which showed
the greatest -helical content (Fig. 2), displayed a GdnHCl midpoint
of 3.61 and an extrapolated
GuH2O of
unfolding of 10.42 kcal/mol, indicating a very stable
-helical structure. The K4 peptide (316-369), which showed a decrease in the
helical content and calculated
-helical residues, showed a similar
GdnHCl midpoint and
GuH2O of
unfolding (compare K2 and K4, Table I). Thus the apparent loss of
helical residues at the NH2 terminus of the coiled-coil region is apparently compensated by the formation of an alternative structure or interaction of similar stability.
The K5 peptide (330-383), the COOH-terminal residues (370-383) of
which are non-helical, also showed a similar GdnHCl midpoint and
GuH2O of unfolding (compare K2 and
K5, Fig. 3A, and Table I), indicating that the COOH-terminal
residues do not significantly affect the stability of the
-helical
structure. The small difference that is observed in the GdnHCl
midpoints may be due to end effect stabilization (32). The stability of
the K3 peptide (344-383), which showed a decrease in the helical
content and loss of helical residues, was significantly destabilized
(GdnHCl midpoint of 1.17) by the loss of the two
NH2-terminal heptads (330-343), indicating that these two
heptads are very important in the stability of the proposed
-helical
coiled-coil structure. Oxidation of both the K2 and K5 peptides
resulted in significant increases in their GdnHCl midpoints (from 3.61 to 5.22 and 3.93 to 5.54, respectively), which indicates that the
formation of a disulfide bonds stabilizes the
-helical structure as
seen in other studies (29, 32, 33).
The concentration dependence of the -helical content, which
can also be used as an indicator of the stability of the associated coiled-coils, was determined for kinesin peptides K2, K3, and K5. Fig.
4B shows that the helical content for K2 and K5 is largely unaffected by concentration over the range tested, indicating that they
are very tightly associated
-helical structures. The equilibrium
association constants from the GdnHCl denaturation plots for K5 and K2
are estimated to be 2.7 × 108
M
1 and 2.3 × 108
M
1, respectively. The similar values for K2
and K5 indicate that the the stability and association of the structure
resides principally within residues 330-369. The effect of deleting
the two NH2-terminal heptads (residues 330-343) from K5
(giving peptide K3) shows a greater dependence of the
-helical
content with peptide concentration. Quite surprisingly, however, the
difference in the concentration dependence of K3 and K2 was not as
dramatic as expected from the difference in their stability (Fig.
4A). This result may indicate that electrostatic
interactions also play a significant role in the association between
the two
-helices in K3 (predominantly electrostatic interactions are
quickly quenched by GdnHCl denaturation and thus not seen as a major
stabilizing factor).
To
determine the oligomerization state of the -helical structures in an
aqueous solution, size-exclusion chromatography with laser light
scattering detection was conducted. Representative SEC chromatograms
for the K5 and K3 peptides are shown in Fig. 5. Analysis
of kinesin peptide K5 in the oxidized and reduced states showed only a
single eluting peak with an apparent molecular weight obtained from
light scattering of 11,702 Da. This value is close to that predicted
for a dimeric structure (13,460 Da). The observation of only a single
species indicates that K5 peptide forms a very stable, dimeric
structure, which agrees with the data presented previously. Similar
studies conducted with K2 (data not shown) also produced only one peak
corresponding to a dimeric molecular weight. On the other hand, the K3
peptide eluted in two peaks: the first dominant peak having an apparent
molecular mass of 9,285 Da and the second smaller peak having a
molecular weight (5,854 Da) close to that of the monomeric peptide
(5,074 Da). The observation of dimer and monomer peaks is consistent with previous data showing that the K3 peptide exhibits the greatest concentration dependence for helical content. Thus, SEC, CD
spectroscopy, and stability studies all suggest that a stable,
two-stranded
-helical coiled-coil can form between two chains of the
kinesin neck region and that residues 330-369 are the ones of primary importance for the formation of this structure.
Destabilizing Effects of Tyr344, Glu347, and Asn351 in the Hydrophobic Core
Previous studies
have shown that the stability of two-stranded -helical coiled-coils
is dependent upon the helical propensity of the region, hydrophobicity
of the residues in the core, packing of residues in the core,
electrostatic interactions adjacent to the core, and chain length
effects (27-29, 34-40). In the kinesin neck region, several of the
residues that are predicted to exist within the hydrophobic core are
considered to be "non-ideal" for generating a stable coiled-coil.
Particularly noticeable are residues Tyr344,
Glu347, and Asn351 that score relatively low by
hydrophobicity analysis and thus are not expected to contribute
significantly to stability. In particular, the ionized carboxyl group
of glutamic acid has been shown to be extremely destabilizing in model
coiled-coils.2 To examine the effects of
residues Tyr344, Glu347, and Asn351
on coiled-coil stability, we prepared and analyzed three analog peptides (see Fig. 1 for sequences). First, we prepared a kinesin peptide analog (K6) in which the four heptads of the native kinesin sequence between residues 344-370 were replaced by a model coiled-coil sequence that has been previously characterized (39). Second, in the K7
analog, the three "destabilizing" kinesin hydrophobic core
residues, Tyr344, Glu347, and
Asn351, were substituted into the above model coiled-coil
sequence. Finally, in K8, three high stability hydrophobic core
residues from the "model" coiled-coil (leucine and isoleucine) were
substituted into the native kinesin sequence into positions 344, 347, and 351.
Fig. 6A shows the secondary structure content
of the three kinesin analogs relative to the unsubstituted kinesin
peptide K3 (344-383). All three of the kinesin analogs show
characteristic double minimas at 208 and 222 nm typical for -helical
protein structures (23, 24). As expected, K6 (the model coiled-coil sequence) displayed the greatest molar ellipticity (39). Introduction of the three kinesin hydrophobic core residues Tyr344,
Glu347, and Asn351 into the K6 sequence caused
a significant decrease in the molar ellipticity (compare K6 and K7,
Fig. 6A), suggesting a disruption between the two associated
-helices as well as a decrease in coiled-coil stability. Conversely,
replacement of Tyr344, Glu347, and
Asn351 with the "ideal" hydrophobes from the model
sequence in K8 resulted in an increase in molar ellipticity, suggesting
an increase in coiled-coil stability and association compared with the
native kinesin sequence (K8 versus K3 in Fig.
6A).
To verify that the changes in helical content observed are in fact a result of changes in the stability of the respective coiled-coils, we determined the stability of K6, K7 and K8 by GdnHCl denaturation (Fig. 6B and Table I). The introduction of kinesin Tyr344, Glu347, and Asn351 into the model coiled-coil sequence caused a dramatic decrease in the stability (compare K6 and K7, GdnHCl midpoints of >8 and 2.73, respectively). Correspondingly, the introduction of the three ideal model hydrophobic residues into the kinesin sequence dramatically increased the stability of the kinesin peptide K3 by 2.74 kcal/mol (GdnHCl midpoints of 1.17 and 3.58 for K3 and K8, respectively).
The changes in stability observed in Fig. 6B are also reflected in the concentration dependence of the helical content, as measured at 222 nm (Fig. 6C). The model coiled-coil sequence K6 exhibited practically no concentration dependence, which is consistent with its high degree of stability. Introduction of kinesin residues Tyr344, Glu347, and Asn351 into this sequence caused the coiled-coil now to dissociate upon dilution (compare K6 and K7, Fig. 6C). Conversely, introduction of the ideal hydrophobes into the native kinesin sequence (K8) dramatically decreased its concentration dependence, which is consistent with its increased stability (compare K3 and K8, Fig. 6B). Collectively, these data indicate that residues Tyr344, Glu347, and Asn351 destabilize the central region of the coiled-coil domain in kinesin. It is intriguing that both positions Tyr344 (Tyr or Phe) and Glu347 are very well conserved in the neck domains of the conventional kinesin, bimC/Eg5, and Kif3 (heterotrimeric) subfamilies of kinesin motors.3
Predictions of Helical Propensity and Hydrophobicity in the Kinesin NeckFinally, we wished to determine whether the observed
location of the coiled-coil dimerization domain (residues 330-369)
agrees with a predictive method developed in our own laboratory. The criteria of our method for predicting coiled-coil regions, which is
very similar to that used by others, is based on the fact that coiled-coil regions are typically high in -helical propensity (which
exists over a minimum of 21 successive residues) as well as
amphipathic, with hydrophobic residues occurring in a 3-4 repeating pattern. Analysis of the helical propensity of human kinesin sequence between residues 280 and 420 shows that there are three regions of high
helical propensity that are well above our statistically determined
cut-off value of 440.2 These regions are indicated by the
three connected boxes shown above the plot in Fig.
7A. Of the three helical sections, only the
region spanning residues 332-369 meets the minimum 21 successive residue cut-off. This prediction agrees well with the experimentally observed stability of peptide K2 (residues 330-369). The residues adjacent to 369 dramatically drop in helical propensity, which is in
agreement with our experiments showing that residues 370-383 do not
add any helical content to that of the region 330-369. Residues
NH2-terminal to 332 also drop in helical propensity, which
is again consistent with our results. It is interesting that the region
from residue 325 to 340, which shows a large dip in helical propensity
centered around residue 330, was predicted by Huang et al.
(6) to be high in
-sheet propensity based upon the secondary
structure prediction program of Holly and Karplus (41). The peptide K1
(316-355), which spans this region, appears to adopt
-sheet
secondary structure at pH 5.5.
Analysis of hydrophobicity occurring in a 3-4 repeating pattern (Fig. 7B) shows that there is only one dominant hydrophobic face in the region of high helical propensity. This face (B) contains the following hydrophobic residues (see also Fig. 1) (Sequence 1).
330 347 368 |
-C--V---A--W---Y--E---N--L---I--L---L--W |
![]() |
Using several
synthetic peptides that overlap within the human kinesin heavy chain
neck region (residues 316-383), we have been able to identify distinct
subdomains with different secondary structure using CD spectroscopy.
The central region from residue 330 to 369 of human kinesin heavy chain
shows all of the characteristics of a stable two-stranded -helical
coiled-coil. This region shows a significant
-helical spectrum with
double minima at 222 and 208 nm, as well as a ratio of
[
]222/[
]208 of 1.06, which is often
indicative of such structures (29-30). Furthermore, the location of
this "coiled-coil" region correlates well with our own coiled-coil prediction method based on helical propensity and hydrophobicity. Stability, concentration dependence, and gel filtration analyses have
indicated that the 330-369 coiled-coil region forms a very stable and
tightly associated dimer. Hence, we refer to the 330-369 region of the
neck as the "dimerization domain." We have also been able to
establish that the dimer is most likely oriented in a parallel fashion,
since oxidation of cysteine 330 preserves the helical content and only
enhances the stability. This would occur only if a disulfide bond was
formed between two parallel-oriented chains. These results agree with
previous reports that the kinesin heavy chains are arranged in a
parallel and in-register orientation (11). Hence, this work
conclusively shows that the kinesin neck contains a region that is
capable of forming a two-stranded
-helical coiled-coil, as had been
previously suggested by sequence predictions (6, 7).
The determined location of the dimerization domain also agrees with the results of Huang et al. (6) and Correia et al. (43), who showed that Drosophila kinesin heavy chains truncated to residue 392 and 401, respectively, form stable dimers (these residues are equivalent to human residues 384 and 393). In addition, similar findings with synthetic kinesin neck peptides to those described here have now also been obtained by Morii et al. (44).
One of the questions raised from the previous work on truncated kinesin
heavy chains was whether the putative coiled-coil in the neck region
was sufficient on its own to account for dimerization, or whether other
interactions (e.g. possibly between the head domains) were
also involved. The estimated equilibrium dissociation constant for
peptide 330-369 (4.3 × 109 M) is
comparable to the value obtained for Drosophila K401
(~4 × 10
8 M) based upon equilibrium
sedimentation studies (43). Thus, our data indicate that the
dimerization domain located within the neck is sufficiently stable on
its own and thus could account for the dimerization observed in these
recombinantly expressed motor domain constructs.
In contrast to the mostly helical dimerization domain, our work also
indicates that the adjacent COOH-terminal (370-383) and NH2-terminal (320-332) segments show little helical
propensity. The COOH-terminal residues 370-383 contribute little to
dimer stability, and we suspect that these residues are part of a
flexible hinge that is thought to exist between the neck and the stalk regions. The NH2-terminal segment of the neck, on the other
hand, appears to have the capability of adopting a -sheet secondary structure. This notion is based upon the finding that deletion of two
COOH-terminal heptads from the coiled-coil dimer containing residues
316-369 (K5) causes the residual peptide (316-355) to display a CD
spectrum characteristic of a
-sheet at pH 5.5 and to form a gel at
neutral pH. We also observed that the addition of 14 NH2-terminal residues (316-329) onto the coiled-coil
dimerization domain (330-369) caused a net loss of ~8 helical
residues without decreasing the stability of the dimeric structure.
This would suggest that a nonhelical intermolecular interactions
(e.g. a small
-sheet secondary structure) can occur in
the region between ~325-335. Our observation of
-sheet secondary
structure also agrees with a secondary structure predictive algorithm
of Holly and Karplus for the NH2-terminal portion of the
neck by Huang et al. (6). Since the NH2-terminal
segment appears to be primarily non-helical and connects helix 6 in the
crystal structure of the globular motor domain (2) to the helical
dimerization domain, we refer to this segment of the neck as the "
linker region."
The linker region of the neck is of considerable interest, since it
appears to play an important role in motility and is highly conserved
in many NH2-terminal motor proteins in the kinesin superfamily. Truncation at Drosophila kinesin residue 340 (human kinesin 332) eliminates the dimerization domain, and yet the
motor still produces directional movement in a multiple motor
microtubule gliding assay (7, 13-15,). Amino acid mutants in the
linker region of the neck also yield kinesin proteins that are severely defective in motility.2 Whether a structural change occurs
in the linker region during the force-generation cycle is unknown.
However, the finding that this region can revert to a fully helical
structure in the presence of 50% TFE indicates that this region has
the intrinsic ability to adopt both
-strand and helical structures
depending on the external environment. Thus, one could imagine that the
linker region could undergo a structural transition during the
ATPase cycle, as will be discussed below.
GdnHCl denaturation
studies of the kinesin peptides indicate that the coiled-coil
dimerization domain is arranged in a strong-weak-strong pattern. Full
stability of the dimerization domain is achieved only when all six of
the spanning heptads (from residues 330-369) are present. Deletion of
two heptads (14 residues) from the COOH terminus (K1 peptide) results
in a complete loss of all -helical content, which is surprising
considering that four of the six heptads still remain. Deletion of the
two NH2-terminal heptads (K3 peptide), on the other hand,
drastically decreases stability and ellipticity, although a dimeric
structure can still be observed by gel filtration. These observations
indicate that both the N- and COOH-terminal heptads are important for
the stability of the structure, with the COOH-terminal heptads
appearing to be the most important. These results agree with those of
Corriea et al. (43), who showed that truncation of the
COOH-terminal 1.5 heptads of the proposed dimerization domain in
Drosophila kinesin produces a protein (K366) that fails to
dimerize at concentrations up to 4 µM.
Interestingly, the central portion of the dimerization domain contains
three residues, Tyr344, Glu347, and
Asn351, that destabilize the coiled-coil structure.
Introducing these amino acids into a model coiled-coil sequence has a
significant destabilizing effect, and conversely, substituting these
three residues in a kinesin neck peptide with ideal hydrophobic
residues increases the stability of the coiled-coil interaction.
Residues Tyr344 and Glu347 are highly conserved
among several classes of NH2-terminal kinesin motors,
suggesting that their role in destabilizing the central region of the
coiled-coil may serve an important function. It is interesting to note
that the Glu347 residues within the "hydrophobic" core
are surrounded by opposite charged lysine and arginine residues in
adjacent e and g positions (Fig.
8). Glover et al. (45) observed in a
c-Fos/c-Jun coiled coil crystal structure that a lysine residue located
within a core position could potentially form hydrogen bonds and/or
electrostatic interaction with adjacent residues in the e
and g positions and speculated that this could stabilize the
structure. Therefore the devastating effect of packing glutamic acid
residues into the hydrophobic core in the kinesin neck could be
mitigated to some extent by the formation of salt bridges with nearby
residues. Such electrostatic interactions could possibly explain the
higher than expected equilibrium association constant of the K3
peptide, even though it is very unstable in GdnHCl (which disrupts salt bonds).
Another intriguing feature of the model representation of the
coiled-coil dimerization domain in Fig. 8 is that the majority of
electrostatic interactions across the core (e-g) are
repulsive (indicated by the arrows in Fig. 8). Previous
studies with model coiled-coil sequences have shown that attractive
electrostatic interactions can be used to increase coiled-coil
stability, control dimer orientation (parallel versus
antiparallel), and govern homo- versus heterodimerization
(27, 29, 35, 36, 40, 46, 47) (for recent reviews, see also Refs.
48-52). The lack of significant attractive electrostatic interactions,
taken together with our data showing an instability within the
hydrophobic core, may indicate that the stability of the neck domain is
optimized not only for its structure but also for its function as
discussed below.
Studies on the kinesin dimer have indicated that the enzymatic cycles of the two kinesin motor domains may be coupled during processive movement. The strongest evidence for this idea comes from Hackney (55), who showed that a kinesin dimer containing two bound ADP molecules releases ADP rapidly from one site and slowly from a second site after mixing with microtubules in a nucleotide-free buffer. Since microtubule interaction stimulates ADP release, Hackney suggested that after one head bound to the microtubule, the partner head had restricted access to a microtubule binding site. This idea is also consistent with recent cryo-EM images of microtubules decorated with dimeric kinesin, which show one kinesin head bound to the microtubule and the second head detached and oriented toward the plus-end of the microtubule (53, 54). Hackney (55) and Ma and Taylor (56) have also shown that ATP binding and hydrolysis by the microtubule-bound kinesin head enables the partner head to bind to the microtubule and release its ADP. These results have led to the suggestion that the two kinesin heads in a dimer are predominantly in different states: one head strongly binds to the microtubule and weakly binds to ADP, while the second head weakly binds to the microtubule and strongly binds to ADP (55, 56). The two heads are suggested to alternate between these two states during processive movement.
Our results on the thermodynamic properties of kinesin neck peptides
suggest a model that could provide an explanation for the results
described above. The model shown in Fig. 9 begins with
one head (without nucleotide) tightly bound to the microtubule, while
the second head (containing ADP) is detached and directed toward the
microtubule plus-end (Fig. 9, step A-B). In this state, the
6 heptad repeats in the neck domain are proposed to exist in a
coiled-coil structure, which constrains the detached head from reaching
an adjacent microtubule binding site. After the bound head binds and
hydrolyzes ATP, a conformational change occurs in the globular motor
domain, which is propagated to the nearby neck domain (Fig. 9,
steps B-C). As a consequence, the linker region of the
neck domain changes its structure or interacts in a new manner with the
core motor domain, and thereby acts like a contracting spring. This
conformational change would account for the observation that monomeric
kinesin containing the
linker region, although not processive, can
elicit force and directional movement. In the kinesin dimer, the
conformational change in the
linker region results in a loss of
-helical residues at the NH2 terminus of the coiled-coil
dimerization domain. This, in turn, could result in a cooperative
unzippering of the majority of the coiled-coil up to the last two
stable heptad repeats (since the middle two heptads are inherently
unstable). Unwinding of the coiled-coil would create a flexible linker
between the heads, which would allow the detached head to reach an
adjacent microtubule binding site (step B-C). Directional
movement to a forward binding site could be favored by the initial
positioning of the detached head closer to the plus-end of the
microtubule (53, 54). Upon docking to the microtubule, the forward head
would then release its ADP and enter a strong microtubule-binding
state, while the rearward head would have progressed in the cycle to a
weak-binding ADP state (Fig. 9, steps C-D). These events
would also reverse the structural change in the linker region of the
neck, which would permit the coiled-coil in the dimerization domain to
"re-zip." In the process, the rearward head would rotate by
~180° (57), returning the enzyme to its initial state in the cycle
and thereby completing one mechanical step.
The above model makes several predictions concerning the roles of different regions within the neck domain. Most notably, enhancing the stability of the middle heptad repeats in the dimerization domain should impair processivity and alternating head ATP catalysis as a result of increasing the energetic requirement for unraveling the coiled-coil as a prelude to separating the heads. Experiments are currently under way to examine this question.
We thank Lorne Burke, Paul Semchuck, and Kim Oikawa for technical assistance in peptide synthesis, purification, and CD spectroscopy. We also are grateful to C. T. Mant, C. Coppin, and L. Romberg for helpful discussions and comments on this manuscript. We also thank Drs. T. Shimizu and H. Morii for sharing their unpublished data on kinesin neck region peptides during the course of this work.