Heterodimerization of Kinesin-2 KIF3AB Modulates Entry into the Processive Run*

Mammalian KIF3AB is an N-terminal processive kinesin-2 that is best known for its roles in intracellular transport. There has been significant interest in KIF3AB to define the key principles that underlie its processivity but also to define the mechanistic basis of its sensitivity to force. In this study, the kinetics for entry into the processive run were quantified. The results show for KIF3AB that the kinetics of microtubule association at 7 μm−1 s−1 is less than the rates observed for KIF3AA at 13 μm−1 s−1 or KIF3BB at 11.9 μm−1 s−1. ADP release after microtubule association for KIF3AB is 33 s−1 and is significantly slower than ADP release from homodimeric KIF3AA and KIF3BB, which reach 80–90 s−1. To explore the interhead communication implied by the rate differences at these first steps, we compared the kinetics of KIF3AB microtubule association followed by ADP release with the kinetics for mixtures of KIF3AA plus KIF3BB. Surprisingly, the kinetics of KIF3AB are not equivalent to any of the mixtures of KIF3AA + KIF3BB. In fact, the transients for each of the mixtures overlay the transients for KIF3AA and KIF3BB. These results reveal that intermolecular communication within the KIF3AB heterodimer modulates entry into the processive run, and the results suggest that it is the high rate of microtubule association that drives rebinding to the microtubule after force-dependent motor detachment.

The kinesin-2 subfamily members are ubiquitously expressed and act as major transporters of intracellular cargoes (for reviews, see Refs. [1][2][3][4][5]. In mammals, KIF3A, KIF3B, and KAP 4 expression yields a heterotrimeric complex (6 -9). The motor polypeptides KIF3A and KIF3B form a heterodimeric motor, KIF3AB, and the nonmotor polypeptide KAP associates at the C terminus of KIF3AB. KAP is a distinctive adaptor pro-tein in that it is largely composed of armadillo repeats (10,11), and it is these motifs that provide the specificity of interaction between KIF3AB and KAP and between KIF3AB-KAP and its specific cargo. A heterotrimeric kinesin-2 complex with KAP has been described in many species since its discovery in sea urchin eggs including Mus musculus, Chlamydomonas, Caenorhabditis elegans, Drosophila, Xenopus, and Tetrahymena (8 -10, 12-19).
KIF3A, KIF3B, and KAP are essential genes (20 -25). Knockout mice for KIF3A or KIF3B have revealed the absence of cilia and a randomized left-right body axis (20 -22). These nodal cilia are crucial for the proper mesodermal patterning during embryogenesis to establish the left-right body asymmetry. Other studies have linked KIF3AB-KAP to cilia-dependent signal transduction cascades including the Hedgehog signaling pathway (26,27). The role of intraflagellar transport in ciliogenesis is considered the basis for KIF3AB-KAP to be an essential protein for development. KIF3AB-KAP is also implicated in a variety of cytoplasmic transport events including organelles, melanosomes, mRNA granules, and membrane-bound vesicles (2, 4, 25, 28 -30). KIF3AB-KAP is also essential for axon elongation, and its transport of fodrin-associated vesicles may provide membrane components to the tips of neurites (28).
Kinesin-2 KIF3AB is clearly implicated in diverse but specific transport events as is conventional kinesin-1, suggesting that the mechanochemical properties of these two kinesins are in some way tuned differently and specifically for their cellular roles because one is not redundant for the other. One distinctive characteristic of KIF3AB and its orthologs is their response to hindering loads (18,(31)(32)(33)(34). Conventional kinesin-1 continues to step processively until approaching its stall force at 6 pN (35)(36)(37)(38)(39), but Xenopus Xklp3a/Xklp3b (32) and C. elegans KLP11/KLP20 (18) tended to detach at hindering loads of ϳ4 pN, shortening their run length capability dramatically. More recently, Andreasson et al. (34) reported that the processivity of murine KIF3AB is strongly force-dependent and dropped precipitously when hindering force was applied such that runs consisted of less than 25 steps at 4 pN. However, the other property observed for each of these kinesin-2 motors regardless of species was that upon detachment they were able to rebind the microtubule quickly and initiate another processive run. These observations have resulted in the hypothesis that kinesin-2 motors are tuned specifically to navigate microtubule (MT) roadblocks such as microtubule-associated proteins like tau in neurons (40) and cross-links along the transition zone when loading and transporting intraflagellar particles into cilia (41).
A presteady-state ATPase kinetics analysis of the steps for entry into the processive run was pursued to ask whether there are intrinsic characteristics within KIF3AB that account for its rapid rebinding to the microtubule after detachment. For these studies, we used motors that were co-expressed from mouse KIF3A and KIF3B constructs that included the native sequence for the motor domain and native helix ␣7 to initiate coiled coil formation followed by a synthetic heterodimerization domain (SHD) as an extension of the native helix to stabilize the native dimer (42,43).
Our recent single molecule studies (43) revealed that in the absence of load KIF3AB with its native neck linker and helix ␣7 is highly processive with run lengths of 1.62 m, which exceeded those of kinesin-1 K560 at 1.26 m. The run lengths of KIF3AA and KIF3BB showed that both were highly processive, although the run length of KIF3BB at 1.51 m is greater than KIF3AA at 1 m and more similar to the run length of KIF3AB (Table 1). Homodimers of KIF3AA and KIF3BB do not appear to occur in vivo, but they provide a tool to assess the motor properties of KIF3A and KIF3B separately. Therefore, we engineered similarly designed constructs for their expression (43). Note that expression of KIF3AA and KIF3BB required a different dimerization motif. The dimerization motif used is a portion of the homodimeric coiled coil and four-helix bundle motif of EB1 (44,45) and was reported previously not to interact with MTs (43,46,47). Control experiments with KIF3AC also confirmed that the dimerization motif (SHD or EB1) used to stabilize the helix ␣7 coiled coil did not alter the processivity (43). The experiments presented here include an analysis of KIF3AB, KIF3AA, and KIF3BB to ask specifically whether heterodimerization alters the intrinsic properties of KIF3A and KIF3B and to define the characteristics of KIF3AB that enable this kinesin to rebind the microtubule so quickly after detachment.

Results
Scheme 1 represents the KIF3AB ATPase cycle, and Fig. 1 presents the proposed steps for entry into the processive run. These were used to design the presteady-state experiments presented. In Table 1, the experimentally determined kinetic and equilibrium constants are reported for heterodimeric KIF3AB and homodimeric KIF3AA and KIF3BB.
Microtubule Association Followed by MantADP Release-Albracht et al. (42) published a presteady-state kinetic analysis of KIF3AB, but at the time it was not known that KIF3AB was as processive as we discovered soon thereafter (43). Therefore, we wanted to take a fresh look at the ATPase cycle and design different experiments to explore entry into the processive run. The first experiments tested the hypothesis of whether there is an intrinsic bias resulting from heterodimerization of KIF3A and KIF3B that will lead to one head preferentially starting the processive run. We began by measuring the presteady-state kinetics of MT collision followed by mantADP release for homodimeric KIF3AA and KIF3BB in comparison with results for KIF3AB (42). For these experiments ADP at the active site was exchanged with mantADP (1:6 ratio of KIF3 nucleotide sites to mantADP). The KIF3-mantADP complex was rapidly mixed in the stopped-flow instrument with varying concentrations of MTs plus 2 mM MgATP. The transients in Fig. 2A show a biphasic decrease in fluorescence as a function of time due to fluorescence quenching when mantADP is released from the active site to the aqueous buffer. The observed rates of the initial fast exponential phase were plotted as a function of MT concentration, and the results in Fig. 2B reveal that the maxi-SCHEME 1 FIGURE 1. KIF3AB stepping model. KIF3AB in solution and detached from the microtubule holds ADP tightly bound at each head (E0). The processive run begins upon MT collision followed by ADP release (E0 and E1), and it is assumed that there is an equal probability of either KIF3A or KIF3B initiating the processive run. After MT collision, the leading head is nucleotide-free and the trailing head is detached from the MT with ADP tightly bound. ATP binding at the leading head triggers a series of structural transitions including neck linker docking that promotes the trailing head to move forward to its next MT binding site (E2-E4). ADP released upon MT collision of the second head results in the E5 two-head-bound state. Strain develops between the two heads and inhibits ATP binding to the nucleotide-free leading head (E5). ATP hydrolysis on the trailing head followed by phosphate (P i ) release results in the weak affinity of the trailing head to the MT and its subsequent detachment (E7). This series of events relieves the interhead tension (E4 -E7). The leading head can now bind ATP, which initiates the next step coupled to ATP turnover without detachment from the MT. In this model, no intrinsic bias is assumed for KIF3A versus KIF3B to start the processive run. Ø, no nucleotide. mum rate constant for mantADP from KIF3BB at 21.3 s Ϫ1 is faster than this rate constant for KIF3AA at 14.4 s Ϫ1 with man-tADP release for KIF3AB reported at 12.8 s Ϫ1 (42). At low MT concentrations, mantADP release is limited by MT collision; therefore, the observed rates plotted as a function of increasing MT concentrations can provide the second-order rate constant for MT association. Fig. 2C shows that the MT association constant for KIF3BB at 12.7 M Ϫ1 s Ϫ1 is faster than this constant observed for KIF3AA at 8.3 M Ϫ1 s Ϫ1 in comparison with 5.7 M Ϫ1 s Ϫ1 observed for KIF3AB using the same experimental approach (42).
Subsequently, using equilibrium approaches Chen et al. (48) reported that mantADP binds more tightly than ADP to their homodimeric KIF3AA motor that was fused to the Drosophila kinesin-1 stalk (KIF3AA-KHC). Therefore, we designed a new experiment to test the hypothesis that mantADP may not capture the kinetics of native ADP release from KIFAB, KIF3AA, and KIF3BB. A similar experimental design was reported by Zhang et al. (49) for KIF3AC.
Microtubule Association and Release of Native ADP Are Significantly Faster than Determined by MantADP-To measure the kinetics of MT collision and ADP release ( Fig. 1, E0 and E1), we took advantage of the fluorescence signal of mantATP binding as a readout of MT collision followed by native ADP release. For this experiment, the KIF3 motors with ADP tightly bound were rapidly mixed in the stopped-flow instrument with varying MT concentrations plus 50 M mantATP for a final concentration of 25 M mantATP. The second-order rate constant for KIF3AB mantATP binding is 7.5 M Ϫ1 s Ϫ1 (42); therefore, mantATP binding at these conditions is Ͼ150 s Ϫ1 and significantly faster than ADP release at the conditions of the assay. Fig. 3A includes representative transients for KIF3AB at varying MT concentrations, which reveal an exponential rise in fluorescence as mantATP binds the hydrophobic environment of the active site. Fig. 3B shows the observed rates of the initial exponential phase plotted as a function of MT concentration, and the hyperbolic fit to the data provides the maximum rate constant of ADP release k ϩ2 at 33.5 s Ϫ1 , which was significantly faster than reported previously for mantADP release (Fig. 3E and Table 1). Note too that at low MT concentrations (Fig. 3B, inset) the second-order rate constant k ϩ1 for MT association at 7 M Ϫ1 s Ϫ1 was somewhat faster than the constant determined previously by mantADP at 5.7 M Ϫ1 s Ϫ1 (42). This experiment was repeated for KIF3AA (Fig. 3C) and KIF3BB (Fig. 3D). The maximum rate constant for ADP release for KIF3AA was somewhat faster at 89.9 s Ϫ1 than k ϩ2 for KIF3BB at 80.2 s Ϫ1 , but both constants were significantly faster than k ϩ2 determined for mantADP release. The insets for Fig. 3, C and D, show that MT association was significantly faster for KIF3AA at 13 M Ϫ1 s Ϫ1 and KIF3BB at 11.9 M Ϫ1 s Ϫ1 than determined for KIF3AB at 7.0 M Ϫ1 s Ϫ1 (Fig. 3E and Table 1). Therefore, these results support the interpretation of the mantADP equilibrium binding results reported by Chen et al. (48) for Drosophila KIF3AA-KHC and show conclusively that mantADP release did not capture the kinetics of MT association and ADP release from KIF3AB, KIF3AA, and KIF3BB. Moreover, the MT association data for KIF3AA and KIF3BB do not reveal a significant difference. Therefore, we conclude that there is an equal probability such that either KIF3A or KIF3B can initiate the processive run. Note too that the KIF3AB rate constant for ADP release at 33.5 s Ϫ1 is consistent with the V max for KIF3AB at 31 s Ϫ1 based on the velocity of single molecule stepping at 246 nm/s with one ATP turnover per 8-nm step (Table 1 and Ref. 43).
Heterodimerization of KIF3AB Controls Entry into the Processive Run-To test the hypothesis that heterodimerization of KIF3AB alters the intrinsic kinetics of KIF3A and KIF3B as revealed by their homodimers, we pursued additional experiments measuring MT association followed by native ADP release (Fig. 4). KIF3AB, KIF3AA, KIF3BB, and mixtures of KIF3AA ϩ KIF3BB at 5 M KIF3 site concentration were rapidly mixed in the stopped-flow instrument with 40 M MTs plus 50 M mantATP (syringe concentrations). The final concentrations were 2.5 M KIF3 sites, 20 M MTs, and 25 M mantATP. Fig. 4A shows the comparison of the transients for 100% KIF3AA, 100% KIF3BB, and 100% KIF3AB. The transients of KIF3AA and KIF3BB overlay each other as one would expect because their kinetics are so similar as documented in Fig. 3. In contrast, however, the transient for KIF3AB is clearly displaced and reveals a slower observed rate of ADP release. The kinetics of ADP release were also measured with mixtures of KIF3AA plus KIF3BB at varying ratios (Fig. 4B). Note that the transients for each of the mixtures, regardless of composition, were equivalent to the transients of 100% KIF3AA or 100% KIF3BB, which we attribute to the kinetics of KIF3AA and KIF3BB being so similar (Figs. 3 and 4). The fits for each transient revealed an observed ADP release rate at 20 M MTs that was very similar for each mixture at 61-65 s Ϫ1 (Fig. 4C). In contrast, the kinetics were significantly slower for ADP release from KIF3AB at

Discussion
The investigation of microtubule association followed by ADP release has revealed important insights. First, the fluorescent analog mantADP, which has been used successfully for many kinesins (50 -70), did not accurately capture the kinetics of MT association and ADP release for KIF3AB, KIF3AA, and KIF3BB (Figs. 2 and 3). The data in Fig. 3 clearly show that MT association followed by ADP release is significantly faster when quantifying native ADP release. If one viewed the mantADP release kinetics in isolation, one may have assumed that it was KIF3B that initiated the processive run because both MT association and mantADP release were faster than the rates observed for KIF3AA. In contrast, the kinetics of KIF3AA and KIF3BB are similar and support the hypothesis that either KIF3A or KIF3B can begin the processive run through MT collision.
The second important insight was the impact of KIF3A on KIF3B and of KIF3B on KIF3A that occurs because of heterodimerization. The rate constants of MT association for homodimeric KIF3AA (13 M Ϫ1 s Ϫ1 ) and KIF3BB (11.9 M Ϫ1 s Ϫ1 ) were similar, but these were significantly faster than MT association for KIF3AB at 7 M Ϫ1 s Ϫ1 , indicating that the intrinsic kinetics of KIF3A and KIF3B as captured by the homodimers are altered due to intermolecular communication that results upon heterodimerization. The mixture experiments of KIF3AA ϩ KIF3BB in Fig. 4 reinforce this conclusion because no combination of KIF3AA ϩ KIF3BB homodimers was able to recapitulate the ADP release kinetics of KIF3AB.
There was another intriguing observation when the kinetics of MT association for KIF3AB were compared with the MT association kinetics for kinesin-2 KIF3AC published previously (49). MT association for KIF3AC was reported at 6.6 M Ϫ1 s Ϫ1 , and this constant for KIF3AC is comparable with the secondorder rate constant for KIF3AB at 7 M Ϫ1 s Ϫ1 reported here. Although Zhang et al. (49) reported a similar KIF3AA MT association rate constant at 11.4 M Ϫ1 s Ϫ1 , MT association by KIF3CC was exceedingly slow at 2.1 M Ϫ1 s Ϫ1 followed by ADP release at 7.6 s Ϫ1 and very different from MT association and ADP release for KIF3BB, supporting the argument that the motor heads of KIF3B and KIF3C are very different catalytically.
These results indicate that it is the heterodimerization of KIF3AB and KIF3AC that modulates the rate of MT association for entry into a processive run rather than the intrinsic catalytic capability of KIF3A, KIF3B, or KIF3C. Additional evidence in support of this hypothesis is that the sequence of the neck linker followed by helix ␣7 for KIF3B and KIF3C are identical with the exception of a single residue. Furthermore, the structures of the neck linkers and ␣7 for KIF3A and KIF3C, and hence KIF3B, are very similar (71). Thus, it is likely that the modulation arises from interactions between the motor domains themselves that cannot be achieved by the homodimers despite how catalytically similar KIF3AA and KIF3BB are. Note too that even though the kinetics of KIF3A and KIF3B are very similar there is only 69% sequence identity between these proteins, suggesting that KIF3AB has evolved its particular set of kinetic properties. Therefore, we conclude that heterodimerization of KIF3A with KIF3B as well as KIF3A and KIF3C results in intermolecular communication outside of the neck linker that controls the rate constant of MT association regardless of which head initiates the processive run.
Although these results inform us of entry into the processive run and the regulation by heterodimerization, they cannot predict the impact of heterodimerization on other steps in the  OCTOBER 28, 2016 • VOLUME 291 • NUMBER 44

KIF3AB Heterodimerization Controls Microtubule Association
ATPase cycle as outlined in the stepping model in Fig. 1 One aim of this study was to define the mechanistic basis of the ability of KIF3AB to rebind to the MT after detachment. This question is compelling for KIF3AB in particular because of its sensitivity to force, resulting in motor detachment rather than stalling as observed for kinesin-1. Our interpretation of the KIF3AB rapid reassociation is based on the high secondorder rate constant for MT association at 7 M Ϫ1 s Ϫ1 . The rate of rebinding by KIF3⅐ADP is determined by the local MT concentration, which was estimated previously at 1 mM near the MT lattice (72). At this concentration, KIF3AB could rebind to the MT at 7,000 s Ϫ1 , resulting in a very short detached state of ϳ143 s. Therefore, the catalytic capability of KIF3AB is opti-mized for reassociation with the MT to initiate another processive run. Based upon the comparisons of KIF3AB with KIF3AA and KIF3BB whose second-order rate constants for MT association are even higher than KIF3AB, one may ask why heterodimerization? Note, however, that homodimeric KIF3AA and KIF3BB are not native motors. Therefore, what is significant in these results is the very fast MT association rate constant for the physiologically relevant KIF3AB that results in rapid rebinding to the MT for intracellular transport.

Experimental Procedures
Constructs of Homodimeric and Heterodimeric KIF3-The M. musculus KIF3A and KIF3B plasmids for expression of KIF3AB heterodimer (42) and the KIF3AA and KIF3BB homodimers (43) were described previously in detail as well as their expression and purification. Briefly, to generate a stable heterodimer of KIF3AB, an SHD motif containing either an acidic (AHD) or basic fusion helix (BHD) was used (73,74). The KIF3A-AHD polypeptide consisted of the KIF3A motor domain, neck linker, and three heptads of native helix sequence followed by the AHD (bold font), a TEV protease site (italic For the KIF3AA and KIF3BB homodimers, each native N-terminal motor domain sequence, neck linker, and ␣7 helix were C-terminally fused to an in-register segment of the dimerization motif from EB1 (bold font) followed by the TEV protease site (italic font) with linker residues (plain font) and a His 8 tag (underlined font): for KIF3A, KIF3A(Met 1 -Leu 374 )-DFYFGKLRNIELICQENEGENDPVLQRIVDILYATDETT-SENLYFQGASHHHHHHHH (predicted molecular weight, 48,502), and for KIF3B, KIF3B(Met 1 -Leu 369 )-DFYFGKLRNI-ELICQENEGENDPVLQRIVDILYATDETTSENLYFQGAS-HHHHHHHH (predicted molecular weight, 48,011). Note that the EB1 motif is a dimerization domain only and not an MT binding domain as shown previously (43,46,47).
KIF3 Protein Expression and Purification-KIF3 motors were expressed in Escherichia coli BL21-CodonPlus (DE3)-RIL cell line (Stratagene, La Jolla, CA) for purification as described previously (42,43). The predicted molecular weights based on amino acid sequence of KIF3AB, KIF3AA, and KIF3BB are 97,118, 97,004, and 96,022, respectively. Experiments were designed by using the KIF3 dimer concentration but converted to the ATP site concentration (two sites per dimer) to report here.
Experimental Conditions-The experiments were performed at 25°C in ATPase buffer: 20 mM Hepes, pH 7.2 with KOH, 5 mM magnesium acetate, 0.1 mM EDTA, 0.1 mM EGTA, 50 mM potassium acetate, 1 mM dithiothreitol, 5% sucrose. The morning of each experiment bovine brain tubulin was cold depolymerized, clarified, and polymerized with 1 mM MgGTP at 37°C. The MTs were stabilized with 40 M paclitaxel, and the MT concentrations reported represent those of the paclitaxel-stabilized tubulin polymer. The reported concentrations of ATP, GTP, and the nucleotide analogs mantATP and mantADP include an equivalent concentration of magnesium acetate. Note that the fluorescent analogs mantATP and mantADP were purchased from Invitrogen as the isomeric mixture.
Experimental Approach for Data Analysis-The presteadystate kinetics experiments were repeated 8 -12 times at varying motor and microtubule concentrations as well as time domains for data collection to acquire experimental results that merge to common data sets. For Figs. 2A and 3A, each representative transient is an average of 8 -10 transients and is from a single experimental day. The plots of observed rates as a function of MT concentration (Figs. 2, B and C, and 3, B-D) include data from multiple experiments on different days to cover the entire MT concentration range. The mean of three to five data sets is reported with errors as S.E. For Fig. 4, the experiments were repeated four to six times on different days, and each transient represents the average of 8 -10 individual transients. The transients in Fig. 4 are from a single day of experiments with the rates of ADP release reported in Fig. 4C as the mean of four data sets ϮS.E.
MantADP Release upon MT Collision-To measure the kinetics of microtubule association followed by mantADP release (Fig. 1, E0 and E1), ADP tightly bound at the active sites of KIF3 was exchanged with the fluorescent analog mantADP at a concentration of 5 M KIF3 nucleotide sites and 30 M mantADP (1:6 ratio). The KIF3⅐mantADP complex was subsequently mixed in the stopped-flow instrument (KinTek SF2003, KinTek Corp., Austin, TX) with varying MT concentrations plus 2 mM MgATP (syringe concentration). The change in fluorescence was monitored as a function of time ( ex ϭ 360 nm, em ϭ 450 nm, detected using a 409-nm long pass filter). A double exponential function was fit to the biphasic transients, and the observed rates of the initial fast exponential phase were plotted as a function of increasing MT concentrations.
At low MT concentrations (0.25-2.25 M tubulin polymer), the rate of mantADP release is limited by MT collision. Therefore, in this MT concentration range, the linear fit to the data (Equation 1) provides the second-order rate constant for MT association k ϩ1 with the KIF3 off-rate, k Ϫ1 , as defined by the y-intercept.
At higher concentrations of MTs, mantADP release after MT collision becomes rate-limiting, and the hyperbolic fit to the data provides the maximum rate constant of mantADP release, k ϩ2 (see Fig. 2 and Scheme 1).
Release of ADP upon MT Collision-To measure the kinetics of native ADP release upon MT collision, the experimental design mimicked the reaction condition at the beginning of the cycle (Fig. 1, E0 and E1). KIF3 motors were rapidly mixed in the stopped-flow instrument with MTs plus mantATP. After MT collision, under these assay conditions, mantATP binding was significantly faster than ADP release from the KIF3 active site. Therefore, the enhanced fluorescence upon mantATP binding was used as a readout of ADP release. KIF3 at varying site concentrations (0.5, 1, or 5 M) was rapidly mixed with 0.5-50 M MTs plus 50 or 100 M mantATP in the stopped-flow instrument. Final concentrations were 0.25 or 0.5 M KIF3 sites/25 M mantATP for 0.25-2.25 M MTs or 2.5 M KIF3 sites/50 M mantATP for 2.5-25 M MTs. A double exponential function was fit to each transient. The observed rates of the initial exponential phase were plotted as a function of MT concentration, and the hyperbolic fit to the data provided the maximum rate constant of ADP release. At low concentrations of MTs, the observed rate of ADP release is limited by MT collision; therefore, the linear fit (Equation 1) to these data provided the second-order rate constant of MT association (see Fig. 3, Scheme 1, and Table 1).
Does Heterodimerization Modulate the Kinetics for KIF3AB Entry into the Processive Run?-To test this hypothesis, additional ADP release experiments were performed as described in the section above. KIF3AB, KIF3AA, KIF3BB, and mixtures of KIF3AA ϩ KIF3BB at 5 M KIF3 site concentration were rapidly mixed with 40 M MTs plus 50 M mantATP (syringe concentrations). Final concentrations were 2.5 M KIF3 site concentration, 20 M MTs, and 25 M mantATP. The mixtures corresponded to 25% KIF3AA ϩ 75% KIF3BB, 50% KIF3AA ϩ 50% KIF3BB, and 75% KIF3AA ϩ 25% KIF3BB. A double exponential function fit to each transient provided the rate constants of each reaction. See Fig. 4.