Examining kinesin processivity within a general gating framework

Abstract

Kinesin-1 is a dimeric motor that transports cargo along microtubules, taking 8.2-nm steps in a hand-over-hand fashion. The ATP hydrolysis cycles of its two heads are maintained out of phase by a series of gating mechanisms, which lead to processive runs averaging ~1 μm. A key structural element for inter-head coordination is the neck linker (NL), which connects the heads to the stalk. To examine the role of the NL in regulating stepping, we investigated NL mutants of various lengths using single-molecule optical trapping and bulk fluorescence approaches in the context of a general framework for gating. Our results show that, although inter-head tension enhances motor velocity, it is crucial neither for inter-head coordination nor for rapid rear-head release. Furthermore, cysteine-light mutants do not produce wild-type motility under load. We conclude that kinesin-1 is primarily front-head gated, and that NL length is tuned to enhance unidirectional processivity and velocity.

Keywords: E. coli; biophysics; human; molecular motor; optical trap; single molecule; structural biology.

Figure 1.. A general gating framework based on mechanical states of dimeric motors.

(A) The kinesin mechanochemical cycle. Kinesin starts from the one-head-bound (1-HB) ATP-waiting state [α], characterized by a strongly bound, nucleotide-free (Ø) front head (red) and an unbound, ADP-containing tethered head (blue). ATP binding induces a force-dependent transition involving partial NL docking, shifting the tethered head past the bound head [β₁]. ATP hydrolysis completes NL docking and facilitates tethered-head binding [β₂]. At this point, kinesin may access a dissociated state [Off], induced by premature phosphate release from the bound head, leading to dimer detachment. However, if the tethered head reaches the forward MT binding site and completes the step before the bound head can dissociate, kinesin enters the two-heads-bound (2-HB) state [γ]. Rear-head release returns the dimer to the ATP-waiting state [α], having moved forward by 8.2 nm. (B) A simplified general gating framework, based on the cycle in (A). Stepping, binding, and unbinding gates are shown with associated rate constants between each of the three gated states, [A], [B], and [C]. The cycle begins at the 1-HB ATP-waiting state [A], where the stepping gate promotes processivity by inhibiting rear-head (blue) rebinding and premature bound-head (red) release. Following a force-dependent step that shifts the tethered head past the bound head [B], the binding gate promotes binding of the tethered head at the forward MT binding site while inhibiting release of the bound head. Also shown is a competing dissociated state [Off], arising from premature release of the bound head from the 1-HB state, accessible from either [A] or [B]. Tethered-head binding leads to the 2-HB state [C], where the unbinding gate promotes rear-head release while inhibiting front-head release, returning the motor to the start of the cycle [A]. DOI: http://dx.doi.org/10.7554/eLife.07403.003

Figure 2.. Kinesin with as many as 3 AA inserted in the NL maintains a 1-HB ATP-waiting state.

(A) Half-site mantADP release measurements as a function of NL insert length (mean ± SE; N = 3). Upon MT binding, both DmK-WT and DmK-3AA, pre-incubated with mantADP (mantD), lose ∼50% of their initial fluorescence. The fluorescence loss exceeds 50% for constructs containing NL inserts longer than 3 AA. Inset, A 50% loss of fluorescence corresponds to dimers binding to MTs in a 1-HB state, whereas a 100% fluorescence decrease is consistent with the release of all bound mantADP (mantD) upon MT binding. (B) MantADP exchange by the tethered head as a function of NL insert length (mean ± SE; N = 3), measured by rapidly diluting mantADP·kinesin·MT complexes into nucleotide-free buffer via stopped flow. The cartoon depicts the measured reaction. The exchange rate increased significantly for constructs with NL inserts of 4 AA or more. In the insets (A and B), white shading indicates non-fluorescent, nucleotide-free heads (Ø); yellow indicates fluorescent, mantADP-bound heads. DOI: http://dx.doi.org/10.7554/eLife.07403.004

Figure 3.. Extending the NL by a single AA compromises processivity.

Mean run lengths as a function of applied force (mean ± SE; N = 49–818) for the constructs studied, acquired with an optical force clamp in the presence of 2 mM ATP (solid circles; color-coded according to the legend). DmK-WT (Milic et al., 2014) and hindering load data sets for DmK-3AA (Andreasson et al., 2015) are reproduced from our previous work. For all constructs, mean run lengths exhibited significant asymmetry, depending upon the direction of load. To obtain the unloaded run length (L₀) and characteristic distance parameter (δ_L) for each construct, run length (L) data for hindering (−6 to 0 pN) and assisting loads (+2 to +20 pN) were separately fit to exponentials (solid lines; color-coded according to the legend) of the form $L\left(F\right)=L_0\text{exp}\left[-\left|F\right|{\delta }_L/k_{\text{B}}T\right]$, where F is the force applied by the optical trap and k_BT is Boltzmann's constant times the absolute temperature; parameter values are in Table 1. Inset cartoon, a graphical representation of the experimental geometry of the single-molecule assay (not to scale). DOI: http://dx.doi.org/10.7554/eLife.07403.005

Figure 4.. Added phosphate enhances the processivity of DmK-WT and DmK-6AA.

(A) Mean run lengths (mean ± SE; N = 84–210) under moderate assisting load (+4 pN), in the presence of nucleotide analogs (green), 100 mM potassium chloride (KCl; purple), or 100 mM potassium phosphate (KP_i; orange). Run lengths for DmK-WT (shaded bars, data from Milic et al., 2014) are shown paired with DmK-6AA data (unshaded bars). Decreasing the ATP concentration, replacing ATP by ATPγS, or adding KCl elicited no statistically significant change in run length relative to the baseline run length for saturating ATP (2 mM) in the absence of added salt. The mean run length increased significantly in the presence of phosphate for both DmK-WT (p < 10⁻⁴; t-test) and DmK-6AA (p < 10⁻⁴; t-test). (B) Run-length data from (A), normalized to the baseline run length value for each construct. DOI: http://dx.doi.org/10.7554/eLife.07403.007

Figure 5.. Assisting load can rescue the velocity of mutants with extended NLs.

Velocity (mean ± SE; N = 49‒818) as a function of force for constructs (solid circles; color-coded according to the legend). Data were collected under the same conditions as Figure 3. DmK-WT velocity was not affected by assisting loads, but the velocities of all mutant constructs could be increased by larger assisting loads. Solid lines show the global fit to a minimal 3-state model (inset) for WT (blue) and mutant (red) constructs, with parameters in Table 2. Data sets for both DmK-WT and DmK-3AA under hindering loads are from (Andreasson et al., 2015). Force–velocity data are compared to other mutant constructs in Figure 7. DOI: http://dx.doi.org/10.7554/eLife.07403.008

Figure 6.. Kinesin unbinding rates are asymmetric with respect to the direction of load.

Single-molecule measurements of the rate of MT unbinding for DmK-WT (mean ± SE; N = 75‒818) at 2 mM ATP (solid circles). The unloaded release rate (k_off) and the associated distance parameter (δ_off) were obtained from exponential fits to unbinding data acquired under hindering loads (−), −25 to 0 pN, and assisting loads (+), +2 to +20 pN. Fits (solid lines) and associated parameters (legend; mean ± SE) are shown. DOI: http://dx.doi.org/10.7554/eLife.07403.010

Figure 7.. Kinesin motility characteristics depend upon parent species, NL length, NL sequence, and cysteine mutations.

Force–velocity relations of constructs that differ by parent species, NL length, NL insert sequence, or cysteine mutations (mean ± SE; N = 25–818; solid circles, color-coded according to the legends). (A) DmK-6AA exhibited lower velocity under all loads than WT, but was faster and less force-dependent than HsK-CL-6AA, a human CL construct with a different 6-AA NL insert. (B) Side-by-side comparison of human and Drosophila CL mutants, along with corresponding WT constructs. Unloaded velocities of CL constructs with WT-length NL domains were similar, but CL constructs were systematically slower than WT under hindering loads. Under assisting loads, only the HsK-CL construct could be sped up beyond WT velocities. (C and D) Comparisons of constructs carrying NL insert sequences LQASQT (C) and AEQKLT (D). Force–velocity data for WT Drosophila and human kinesin were indistinguishable, but human constructs with 6-AA extensions of the NL moved at lower velocities than corresponding Drosophila motors under all loads. (E and F) Comparisons of all Drosophila (E) and human (F) constructs. Constructs with the NL insert AEQKLT were slower for all forces than those with LQASQT, for both Drosophila and human kinesin. HsK-CL-6AA data (Clancy et al., 2011) and hindering-load velocities for DmK-WT (Andreasson et al., 2015) are reproduced from previous work. Fits to DmK-WT and DmK-6AA data sets (solid lines; color-coded according to the legends) correspond to the model of Figure 5, with parameters values from Table 2. The remaining force–velocity data were fit to polynomials (solid and dashed lines; color-coded according to legends), provided to guide the eye. DOI: http://dx.doi.org/10.7554/eLife.07403.011