|Home | About | Journals | Submit | Contact Us | Français|
The Ncd kinesin-14 motor is required for meiotic spindle assembly in Drosophila oocytes and produces force in mitotic spindles that opposes other motors. Despite extensive studies, the way the motor binds to the spindle to perform its functions is not well understood. By analyzing Ncd deleted for the conserved head or the positively charged tail, we find that, strikingly, the tail is essential for binding to spindles and centrosomes, but both the head and tail are needed for normal spindle assembly and function. Fluorescence photobleaching assays to analyze binding interactions with the spindle yielded data for headless and full-length Ncd that, unexpectedly, did not fit well to previous recovery models. We report a new model that accounts for Ncd transport towards the equator revealed by fluorescence flow analysis of early mitotic spindles and gives rate constants that confirm the dominant role the Ncd tail plays in binding to the spindle. By contrast, the head binds weakly to spindles based on analysis of the tailless fluorescence recovery data. Minus-end Ncd thus binds tightly to spindles and is transported in early metaphase towards microtubule plus ends, the opposite direction as it moves, to produce force in the spindle later in mitosis.
Ncd, a minus-end kinesin-14 motor, plays an essential role in assembly of the anastral Drosophila oocyte meiosis I spindle (Matthies et al., 1996; Endow and Komma, 1997; Endow and Komma, 1998), where it is thought to crosslink and bundle microtubules to form foci or asters that migrate towards the chromosomes and nucleate microtubules for spindle assembly (Sköld et al., 2005). The motor also acts during oocyte meiosis I spindle assembly to form lateral interactions between microtubule-coated bivalent chromosomes and stabilizes the interactions by sliding microtubules against one another (Sköld et al., 2005). The Ncd motor binds to centrosomes and microtubules throughout the mitotic spindle (Hatsumi and Endow, 1992; Endow et al., 1994; Endow and Komma, 1996), and is thought to function in the mitotic spindle by binding to microtubules and hydrolyzing ATP to produce force to maintain spindle length or elongate the spindle (Saunders and Hoyt, 1992). In particular, Ncd has been proposed to act as a brake in the spindle by opposing outward acting forces (Sharp et al., 1999; Sharp et al., 2000; Tao et al., 2006). The motor also plays an important role in attaching centrosomes to mitotic spindle poles and chromosomes to spindles in early embryos (Endow et al., 1994; Endow and Komma, 1996).
Binding by the kinesin microtubule motors to the spindle is assumed to be via the highly conserved motor domain or head, which contains an invariant microtubule-binding motif and can bind to microtubules in vitro when expressed by itself (Chandra et al., 1993b; Hirose et al., 1995; Lockhart et al., 1995). The Ncd tail has also been reported to contain regions that bind tightly to microtubules in vitro and has been speculated to mediate interactions of the motor with microtubules in vivo (Karabay and Walker, 1999a), although tests of this hypothesis have not been reported. Information regarding the way in which Ncd and other spindle motors bind to the spindle is essential to understand how the motors perform their functions and produce force in the spindle.
Here we analyze Ncd head and tail binding interactions with the spindle using tailless and headless motors expressed in Drosophila oocytes and embryos as fusions to a bright green fluorescent protein. Unexpectedly, we find that the tailless Ncd motor does not bind to oocyte meiosis I spindles and binds weakly to mitotic spindles and centrosomes. By contrast, headless Ncd binds to both oocyte meiotic and mitotic spindles. Thus, rather than the conserved motor domain playing a dominant role, Ncd binding to the spindle and centrosomes is mediated predominantly by the tail. Tight binding by the motor to spindle microtubules results in motor transport from the poles to the equator during spindle assembly in early metaphase, as revealed by analysis of fluorescence flow in the spindle - opposite to the direction of movement of the minus-end Ncd motor along spindle microtubules.
To determine the role of the Ncd head and tail in binding to the spindle, we recovered germline transformants that express headless (HL) or tailless (TL) Ncd, regulated by the native ncd promotor and fused to Venus, an enhanced green fluorescent protein (Nagai et al., 2002) (Fig. 1A). HLncdVenus encodes the Ncd tail and coiled-coil stalk, and TLncdVenus, the stalk and conserved motor domain, both joined at the C terminus to Venus. FLncdVenus, expressing full-length Ncd fused to Venus, was made as a control.
HLncdVenus and TLncdVenus flies homozygous for the transgene and the ncd null mutant, cand, were tested for rescue of cand embryo inviability caused by elevated levels of chromosome nondisjunction and loss. Embryo viability of HLncdVenus; cand (0.114, n=19, total=167) and TLncdVenus; cand lines #M23M4 (0.187, n=25, total=134) and #F26F2 (0.101, n=17, total=168) was similar to cand (0.113, n=144, total=1,277) (Endow and Komma, 1997). By contrast, FLncdVenus cand (0.698, n=113, total=162) and a transgenic line expressing 4 copies of ncd fused to S65T gfp (Endow and Komma, 1997; Heim et al., 1995), ncdgfp* cand #4121 (0.868, n=79, total=91), showed much higher embryo viability. HLncdVenus (χ2=116.6, 1 d.f., p<0.0001) and TLncdVenus lines #M23M4 (χ2=75.0, 1 d.f., p<0.0001) and #F26F2 (χ2=121.9 , 1 d.f., p<0.0001) differ significantly from FLncdVenus; they also differ significantly from ncdgfp* #4121. Thus, Ncd deleted for either the tail or head fails to rescue embryo inviability of an ncd null mutant caused by frequent chromosome mis-segregation during meiosis in oocytes and mitosis in early embryos.
Mature HLncdVenus; cand oocytes expressing Ncd without the head showed multi-polar or multiple small meiosis I spindles (n=11, total=11) (Fig. 1B), consistent with previous findings that Ncd is required for spindle assembly in oocytes (Kimble and Church, 1983; Matthies et al., 1996; Endow and Komma, 1997; Sköld et al., 2005). The spindles did not resemble the mature metaphase-arrested FLncdVenus oocyte meiosis I spindle, but exhibited abnormalities typical of null or severe loss-of-function mutants, such as ncd2, which is blocked in the initial stages of spindle assembly (Sköld et al., 2005). The multi-polar or multiple small oocyte spindles demonstrate that the Ncd tail does not rescue cand for spindle assembly. They show that the Ncd tail is sufficient to bind to meiotic spindles and does not require the head, but the head is needed for normal assembly of the oocyte meiosis I spindle. The multi-polar and multiple small spindles indicate that the head is needed to form and stabilize lateral interactions between microtubule-associated bivalent chromosomes during assembly of the meiosis I spindle (Sköld et al., 2005).
Immature TLncdVenus; cand oocytes showed cytoplasmic fluorescence but the germinal vesicle was dark, indicating that TLNcdVenus was expressed but was excluded from the nucleus. Low fluorescence was also observed in mature oocytes following germinal vesicle breakdown, but meiosis I spindles were not observed. Oocytes expressing wild-type Ncd fused to a monomeric red fluorescent protein (mRFP) (Campbell et al., 2002) together with TLNcdVenus were examined to provide a marker for the meiosis I spindle. The spindles fluoresced red, but showed no green fluorescence, indicating little or no TLNcdVenus binding to the spindle (n=8, total=8) (Fig. 1B). The lack of binding by TLNcdVenus to the meiosis I spindle is consistent with the high frequency of TLNcdVenus; cand embryo lethality, similar to the cand null mutant.
HLncdVenus embryos exhibited bright spindle and centrosome fluorescence throughout the early cleavage divisions, like FLncdVenus and ncdgfp* embryos (Endow and Komma, 1996), but abnormal mitotic spindles were frequently observed in HLncdVenus embryos (n=29, total=29). The abnormalities included spindle spurs or branches caused by mis-segregating chromosomes detaching from the spindle or moving off the metaphase plate (Endow and Komma, 1996), bent anaphase spindles due to unequal forces in the spindle, and poorly formed midzones in telophase (Fig. 1C, Movie S1). Free centrosomes were present in some embryos that arose from failure of spindles to form or centrosome loss from a spindle pole, based on analysis of the time-lapse sequences. A few embryos (n=3, total=29) showed highly aberrant divisions caused by abnormal oocyte meiotic divisions, but most embryos (n=26, total=29) displayed some normal appearing spindles and others with abnormalities that increased in frequency in later divisions, similar to embryos of the severe loss-of-function mutant, ncd2 (Endow and Komma, 1996). The time required for HLncdVenus; cand cycle 10 embryos to progress from nuclear envelope breakdown (NEB) in prometaphase to midzone formation in telophase was 248±5 s (n=15), not significantly different from FLncdVenus cand (263±9 s, n=9) or ncdgfp* cand #4121 (245±3 s, n=17).
TLncdVenus; cand embryos showed dark interphase nuclei prior to mitosis and uniform cytoplasmic fluorescence, indicating that TLNcdVenus was expressed but was excluded from the nucleus, as in oocytes. The nuclei became fluorescent immediately following NEB and faintly fluorescent spindles were apparent soon after (Fig. 1C). TLncdVenus mitotic spindles showed a gap in fluorescence at the metaphase plate, similar to the gap at the chromosomes observed in rhodamine-tubulin labeled spindles (Endow and Komma, 1996), and centrosomes showed faint fluorescence (Fig. 1C, S1), differing from HLncdVenus and FLncdVenus. The pole-to-pole Ncd fluorescence in mitotic spindles at metaphase is thought to be caused by Ncd fibers or filaments that traverse the chromosomes (Endow and Komma, 1996). The gap at the chromosomes was not observed in HLncdVenus spindles; its presence in TLncdVenus spindles is presumably caused by dependence of fiber or filament formation on the Ncd tail. The gap disappeared in early anaphase as the spindle elongated (Fig. 1C, Movie S2). TLncdVenus ;cand embryos frequently displayed abnormally spurred or bridged spindles, nuclei that failed to form spindles, or different-sized spindles or nuclei, an indication of chromosome loss or mis-segregation (n=12, total=12). The time required for cycle 10 embryos to progress from NEB to midzone formation (357±7 s, n=14) was significantly longer than HLncdVenus, FLncdVenus or ncdgfp*. The TLncdVenus; cand embryos demonstrate that the Ncd head without the tail binds weakly to the mitotic spindle and frequently causes abnormal spindles to form.
HLncdVenus; cand mitotic spindles were longer by 1-2 μm than FLncdVenus cand spindles in prometaphase and metaphase; the difference in length increased to 3-4 μm in anaphase and telophase, reaching ~4 μm in mid-telophase (HLncdVenus, 23.7±0.3 μm, n=32; FLncdVenus, 19.9±0.2 μm, n=38) (Fig. 2, Table 1). The longer than normal HLncdVenus spindles imply that Ncd regulates spindle length during anaphase and telophase, presumably by opposing other motors in the spindle. These results parallel previous reports that cand spindles lacking Ncd are longer than wild type in late cleavage divisions (Sharp et al., 1999; Sharp et al., 2000). TLncdVenus spindles showed little or no centrosome fluorescence, making it difficult to measure pole-to-pole spindle lengths during mitosis. Measurements made in mid-telophase, the time at which the greatest difference was observed between HLncdVenus and FLncdVenus spindles, and where the centrosomes were faintly visible against the dark telophase nuclei, gave a length of 18.8±0.3 μm (n=26), close to the value for FLncdVenus spindles of 19.9±0.2 μm, indicating that the TLNcdVenus motor can function to regulate spindle length.
Photobleaching assays were performed to estimate kinetic constants for Ncd binding interactions with the mitotic spindle. Two different-sized regions of interest (ROIs) at the equator (radius, w=2.66 μm and 1.3 μm) were photobleached at high laser power in spindles of cycle 9-11 embryos and images were recorded rapidly (~165 ms/frame) at low laser power to monitor recovery. The fluorescence intensity in the bleach spot was normalized, corrected for loss during imaging, averaged with data from replicate assays, and plotted versus time. Assays of FLNcdVenus showed slower fluorescence recovery in the large ROI (n=11) than the small ROI (n=5), as predicted if recovery is dependent on diffusion (Fig. 3A,B; Movies S3, S4). However, the large ROI showed a greater delay than expected due to effects of diffusion alone. HLNcdVenus assays showed an even larger difference between the large (n=10) and small ROI (n=8), and similarly shaped recovery curves (Fig. 3C,D), indicating that the delayed recovery of the large ROI could be attributed to Ncd tail effects. TLNcdVenus assays showed a much smaller difference between the large (n=11) and small ROI (n=8) recovery curves and more rapid recovery than FLNcdVenus or HLNcdVenus for both ROIs (Fig. 3E,F), providing further evidence that the delayed recovery of the FLNcdVenus large ROI is caused by the Ncd tail, rather than the head.
Attempts to fit models for FRAP recovery to the FLNcdVenus and HLNcdVenus data based on binding or both diffusion and binding (Sprague et al., 2004) gave poor fits (Fig. S2). By contrast, the TLNcdVenus FRAP recovery data fit well to the diffusion-binding model reported previously, which accounts for recovery by both diffusion and binding interactions (Sprague et al., 2004). Concurrent fits of the TLNcdVenus data for the large and small ROI to this model (Hallen et al., 2008) yielded values for k*on, the pseudo-first order binding rate constant; koff, the dissociation rate constant; Ceq, the fraction of the motor bound at equilibrium; Feq, the fraction free at equilibrium; and Deff, the effective diffusion coefficient in the spindle (Table 2). The TLNcdVenus dissociation and binding rate constants were koff=0.06±0.01 s-1 and k*on=0.005±0.001 s-1. The low koff and k*on values and koffk*on indicate weak binding both kinetically and thermodynamically. The low Ceq=0.07±0.02 and correspondingly high Feq=0.93 explain the low level of fluorescence of TLncdVenus mitotic spindles.
Of the previous models, the FLNcdVenus and HLNcdVenus FRAP recovery data fit best to a two-state binding model (r2=0.98) and the diffusion-binding model (r2=0.98), but the data still exhibited strong deviations from the models in several phases of the curve, including the beginning, which is strongly influenced by rapid diffusional effects, a quasi-linear middle region nonexistent in the model, and the end, which is dominated by binding interactions (Fig. S2). Consequently, we sought to modify the diffusion-binding model in order to reduce these deviations and account for other factors that contribute to the recovery. The marked difference in shape of the large and small ROI recovery curves for FLNcdVenus and HLNcdVenus indicate a pronounced delay in recovery of the large ROI, suggesting depletion throughout recovery of the motor that is free to diffuse, possibly due to tight binding by FLNcd or HLNcd to spindle microtubules. This raised the possibility that recovery occurs by growth or transport of microtubules with bound Ncd into the bleach spot.
As current FRAP models do not account for recovery by transport, we derived a model that assumes rapid diffusional equilibration of unbleached free Ncd in the bleach spot and recovery by Ncd binding to microtubules, followed by growth or transport of microtubules with bound motor into the bleach spot. The new model fit the data for both FLNcdVenus and HLNcdVenus well (r2=0.99), exhibiting a quasi-linear middle region, like the data, that can be attributed to the effects of transport. The model assumes that microtubules in the ROI are growing or sliding towards the spindle equator with constant velocity, although each microtubule moves with a different velocity. The distribution of velocities is approximated as a half-Gaussian with a peak at zero and width determined by fitting to the recovery data. This is a tractable approximation to a Poisson distribution, which applies if microtubule growth rates form a Poisson distribution and microtubule velocity is proportional to the number of motors propelling them, since the number of bound motors is expected to follow a Poisson distribution. Net growth of spindle microtubules towards the equator occurs in early through mid-metaphase as spindles assemble, when the FRAP assays were performed. Sliding of microtubules towards the chromosomes is thought to be mediated by plus-end motors, e.g., the KLP61F kinesin-5 motor acting on parallel microtubules, or minus-end motors - Ncd or cytoplasmic dynein - sliding anti-parallel microtubules (Tao et al., 2006).
The new diffusion-binding-transport model was fit simultaneously to the mean data sets for the large and small bleach spots using MATLAB. The curve fits yielded estimated values for kinetic parameters that included v, the mean velocity of Ncd-bound microtubule growth or transport into the bleach spot, koff, k*on and Deff(Table 2). The FLNcdVenus dissociation rate constant from the spindle and binding rate constant were koff =0.028±0.004 s-1 and k*on =0.019±0.003 s-1 . The low koff and k*on ≈ koff indicate tight binding to the spindle by the Ncd motor. HLNcdVenus also showed a low koff=0.008±0.004 s-1 and k*on=0.006±0.003 s-1, both lower by ~3-4 fold compared to FLNcdVenus, with k*on ≈ koff, indicating that the tight binding by the FLNcd motor could be attributed to the Ncd tail. FLNcdVenus and HLNcdVenus Deff was lower than TLNcdVenus by ~2-fold. The mean velocity of Ncd-bound microtubule growth or transport into the bleach spot was v=0.026±0.002 μm/s for FLNcdVenus and v=0.042±0.002 μm/s for HLNcdVenus.
The velocities correspond to the net microtubule growth, including depolymerization, poleward motion and flux, and transport towards the spindle equator. The slower velocity of FLNcdVenus movement towards the equator, compared to HLNcdVenus, could be due to poleward motor activity. The velocities are close to one another and within a 2-3 fold range of the microtubule growth rate reported for interphase Drosophila S2 cells of 0.063 μm/s (Rogers et al., 2002); they are also close to the velocity of the plus-end spindle motor KLP61F (Tao et al., 2006) of 0.04 μm/s. The velocities of Ncd (0.1-0.2 μm/s) (Chandra et al., 1993b) and yeast cytoplasmic dynein (0.085±0.030 μm/s) (Reck-Peterson et al., 2006) are higher in vitro, however, motor-mediated microtubule sliding in the spindle is thought to be slowed by plus-end motors (Tao et al., 2006).
The net growth or transport of microtubules bound to Ncd in early mitotic spindles was analyzed by determining the median position of FLNcdVenus fluorescence in each spindle half - the point along the spindle axis where half of the florescence is closer to the equator and half is further away - in image sequences of cycle 10 spindles (n=7) from early metaphase through anaphase recorded at 2.32 s/frame. Spindles were rotated and aligned manually, then cropped to eliminate centrosomes, astral microtubules, and cytoplasm. The image sequences from spindles in different embryos (n=4) were aligned in time by phase of mitosis and the median position of fluorescence in each spindle half over time was determined relative to a stationary point outside the spindle using a MATLAB routine, fluorescence_medians. The median position values were normalized so that each data set starts at zero and peaks at one to offset differences in spindle size and length, then the normalized values were averaged (Fig. 4A).
The median fluorescence position changed rapidly with time during the initial ~60 s, moving an average of 1.5±0.2 μm towards the equator. This initial phase corresponds to rapid microtubule growth towards the equator during spindle assembly and was followed by a plateau, roughly correlating with mid-metaphase (Fig. 4B, Movie S5). The median position then retreated from the metaphase plate as the spindle elongated. The time at which the FRAP assays were performed corresponds to the phase of rapid fluorescence flow towards the equator (Fig. 4A,B).
The velocity during the initial ~60 s was 0.027±0.004 μm/s, not significantly different from the velocity of growth or transport towards the equator estimated from curve fits to the FLNcdVenus FRAP data of v=0.026±0.002 μm/s. The analysis shows that net flow of FLNcdVenus fluorescence occurs in the early metaphase spindle and is directed towards the spindle equator. The fluorescence flow is presumed to result from growth or transport of microtubules with bound Ncd towards the equator, opposite to the direction of movement of the minus-end motor along spindle microtubules.
A large number of kinesin motor proteins have now been shown to function in meiosis and mitosis, performing essential roles in spindle assembly and length regulation (Manning et al., 2007; Saunders et al., 2007; Zou et al., 2008), kinetochore to spindle attachment (Grishchuk et al., 2007), and chromosome oscillation and congression (Mayr et al., 2007; Stumpff et al., 2008). These motors include minus-end Ncd, a kinesin-14 motor required for spindle assembly in Drosophila oocytes (Kimble and Church, 1983; Matthies et al., 1996; Endow and Komma, 1997; Sköld et al., 2005) and chromosome segregation in early embryos (Endow et al., 1994; Sharp et al., 2000). Although several studies of the biochemical and motile properties of partial or truncated Ncd head (Chandra et al., 1993a; Chandra et al., 1993b; Furuta and Toyoshima, 2008) and tail proteins (Chandra et al., 1993b; Karabay and Walker, 1999a; Karabay and Walker, 1999b) have been reported, the roles of the Ncd head and tail in the spindle are not well understood. By analyzing proteins that contain the coiled-coil stalk but lack either the conserved head, which is needed for ATP hydrolysis and force generation, or the basic proline-rich tail, including the microtubule-binding sites identified by Karabay and Walker (Karabay and Walker, 1999a), we find that the tail is essential for motor binding to the spindle, but both the head and tail are needed for normal spindle assembly and motor function in the spindle.
Neither the Ncd motor protein deleted for the head nor the tail was sufficient to rescue embryo inviability of the ncd null mutant, cand, caused by high levels of chromosome mis-segregation in oocytes and early embryos. TLncdVenus females produced a low frequency of viable embryos, comparable to the cand null mutant, and TLNcdVenus did not bind to the oocyte meiosis I spindle - thus, the Ncd tail is necessary for spindle binding and motor function in the meiosis I spindle. HLNcdVenus binds to spindles, but the spindles are abnormal. Mature oocytes that express HLNcdVenus exhibit abnormal multi-polar meiosis I spindles or multiple small spindles in which lateral interactions have failed to form (Sköld et al., 2005), typical of cand or severe loss-of-function ncd mutants, indicating that the Ncd head is required for normal meiosis I spindle assembly. The multi-polar and multiple small spindles imply that the head is needed to form lateral interactions between spindle-associated bivalent chromosomes for bipolar spindle assembly (Sköld et al., 2005), presumably by sliding and crosslinking microtubules.
HLNcdVenus binds to mitotic spindles, but the spindles are frequently bent, indicating unequal forces in the spindle, or spurred or bridged to other spindles, evidence of mis-segregating chromosomes (Endow and Komma, 1996), and the midzones frequently fail to form. The longer than normal HLncdVenus spindles in late anaphase through telophase indicates that the Ncd head plays a role in regulating spindle length by acting as a brake and opposing forces that cause spindle elongation (Sharp et al., 1999; Sharp et al., 2000; Saunders et al., 2007). HLNcdVenus shows that the tail alone can bind to the spindle, but is not sufficient to prevent chromosomes from detaching from the mitotic spindle, resulting in spurs and bridges. The bent anaphase and telophase spindles in HLncdVenus embyros indicate that HLNcd is defective in counteracting other spindle forces, resulting in longer than normal spindles in late anaphase and telophase.
TLncdVenus embryos showed only faint mitotic spindle fluorescence and little or no centrosome staining, and mitotic spindles were frequently abnormal. TLncdVenus cell cycle progression was also markedly delayed - 90 to >100 s longer than FLncdVenus or HLncdVenus embryos, indicating that the Ncd head without the tail is not sufficient for motor function in the spindle.
Tight binding to spindle microtubules by the Ncd tail was confirmed by estimating binding constants for FLNcdVenus and HLNcdVenus in the mitotic spindle. Data from FRAP assays were fit to a new mathematical model that accounts for fluorescence recovery by diffusion into the bleach spot of unbleached free Ncd and binding to spindle microtubules, together with growth or transport of microtubules with bound Ncd into the ROI. Analysis of FLNcdVenus gave a low koff that was comparable to k*on, consistent with tight binding to the mitotic spindle. The HLNcdVenus koff and k*on were both lower than those for FLNcdVenus, but did not differ significantly from one another. By contrast, the TLNcdVenus koff was much higher than FLNcdVenus or HLNcdVenus, while k*on was lower, giving koffk*on, explaining the much faster recovery and weaker binding by TLNcdVenus. Deff for TLNcdVenus was also larger than FLNcdVenus or HLNcdVenus, indicating greater diffusional movement in the mitotic spindle. Tight binding by Ncd to the spindle can thus be attributed to the tail. A tailless Ncd has been reported previously to be a weak-binding, diffusional motor in vitro (Chandra et al., 1993a); the weak fluorescence of TLncdVenus mitotic spindles and absence of fluorescence in meiosis I spindles, together with FRAP kinetic constants showing koffk*on and Deff greater than FLNcdVenus or HLNcdVenus, indicate that TLNcdVenus binds a weak, diffusional manner to the spindle in vivo. Weak binding by the Ncd motor domain to microtubules in vitro has been reported previously (Kd,ATP=5.8±1.1 μM) (Song and Endow, 1998), compared to the tight binding by Ncd tail regions (<1 μM) (Karabay and Walker, 1999a). Ncd binding to the mitotic spindle and centrosomes thus requires the tail; the head binds little, if at all, to the oocyte meiosis I spindle and shows only weak, diffusional binding interactions with early embryo mitotic spindles.
Analysis of FLNcdVenus in the mitotic spindle demonstrates a net flow of fluorescence from the pole towards the equator in early metaphase with a rate of movement that decreases in mid-metaphase. The velocity is the same as the microtubule growth or transport rate towards the equator estimated from curve fits of the new diffusion-binding-transport model to our photobleaching recovery data. The data indicate that transport of FLNcdVenus occurs during early metaphase due to tight binding by the motor to growing or sliding microtubules that move into the bleach spot. The net flow towards the spindle equator is opposite to the direction of movement of the minus-end Ncd motor. Tight binding by Ncd to spindle microtubules during early metaphase thus results in transport of the motor towards the equator, positioning it in the spindle for function later in mitosis.
Transport of Ncd towards the spindle equator in early metaphase to position the motor in the spindle for function later in mitosis is correlated with evidence that Ncd is required in late anaphase to balance forces in the spindle and regulate spindle length. Mitotic spindles of HLncdVenus cycle 9-11 embryos are abnormally bent and longer than wild type in late anaphase through telophase, consistent with the proposal that the Ncd motor produces force in the spindle in the later stages of mitosis to regulate length by opposing other spindle forces. By contrast, mid-telophase TLncdVenus spindles are comparable in length to those of FLncdVenus embryos, indicating that the Ncd motor without the tail can function to regulate spindle length. Previous studies have shown that Ncd produces force in the mitotic spindle in cycle 12-13 cleavage divisions that suppresses spindle collapse and regulates centrosome spacing during prometaphase and metaphase, but has little effect in anaphase or later stages of mitosis (Sharp et al., 1999; Sharp et al., 2000). Ncd is a maternally expressed protein, thus these differences in observations could be due to reduced or altered activity in the later cleavage divisions as the number of spindles increases, compared to the earlier stages examined in the present work. Further studies to clarify how Ncd function is regulated in the spindle and the mechanism by which motor deficiency causes mitotic chromosome loss should be highly informative.
pCaSpeR (Thummel et al., 1988) plasmids encoding full-length, headless and tailless Ncd fused to the Venus fluorescent protein (Nagai et al., 2002), regulated by the native ncd promotor, were constructed using conventional methods. FLncdVenus codes for the Ncd tail, coiled-coil stalk and motor domain (M1-K700) with a change of D699L at the C terminus to create an AflII site to fuse Venus. HLncdVenus codes for the Ncd tail and coiled-coil stalk (M1-G347 + N348) with an added A349 to create an NcoI site to join Venus, and TLncdVenus codes for the stalk and conserved motor domain (MES-L209-K700) with a change of D699L to fuse Venus. Plasmids were injected into w;Δ2-3 Sb/+ embryos and w+ transformants were selected and made homozygous for the ncd null mutant cand, or heterozygous for cand and the TM3 balancer chromosome. Homozygous flies from different lines were analyzed by PCR using primers that give different length products for the cDNA transgene and native ncd to confirm the absence of ncd+. FLncdVenus #F28M1 cand, HLncdVenus #M6M1; cand, TLncdVenus #M23M4; cand and TLncdVenus #F26F2; cand were tested for viability by arraying embryos on grape juice agar plates and monitoring hatching over 3-4 days.
Transgenic flies expressing full-length Ncd fused at the C terminus to a monomeric red fluorescent protein (mRFP) (Campbell et al., 2002) were similarly recovered, transferred into a cand background, and tested by PCR.
Live oocytes or embryos expressing Venus or mRFP fusion proteins were prepared as described (Sciambi et al., 2005; Sköld et al., 2005; Zou et al., 2008) and images were collected at 20°C using a Bio-Rad Radiance2100 confocal scanhead mounted on an Axioskop2 microscope (Carl Zeiss, Inc.) with a 40X/1.3 NA Plan-NeoFluar oil immersion objective and recorded with Bio-Rad LaserSharp 2000 software. Image analysis, including centrosome-to-centrosome spindle length measurements, were performed in ImageJ v 1.39t (NIH). Images were cropped and assembled into montages in ImageJ, brightness and contrast were adjusted in Adobe Photoshop v 8.0, and final changes in size were made in Adobe Illustrator v 11.0.0. Adjustments to image contrast and brightness were linear.
FRAP assays at the equator of cycle 9-11 mitotic spindles were performed at 22-25°C using an LSM510 confocal microscope and LSM510 software (Carl Zeiss, Inc.), a 40X/1.3 NA Plan-NeoFluar oil immersion objective and the 488-nm line of a 30-mW Ar laser operating at 75% power. Briefly, six prebleach images were recorded, 3 or 4 photobleaching scans were performed at 100% laser power in an ROI of radius w=2.66 μm or 1.3 μm, followed by 494 rapid recovery images at ~165 ms/image (recovery time=82.5 s) and low laser power (1-3%) to minimize photobleaching. Data were normalized and corrected as described (Zou et al., 2008; Hallen et al., 2008), then fit concurrently in MATLAB (The Mathworks) (Hallen et al., 2008) to kinetic models that account for FRAP recovery by binding, or both diffusion and binding (Sprague et al., 2004). None of the available models for fluorescence recovery gave good fits to both the large and small ROI recovery data. Because of the unusually slow recovery of the large ROI, we hypothesized the existence of a binding state that depletes the free, diffusing Ncd, markedly affecting recovery of the large ROI compared to the small ROI. Tight binding by Ncd to the spindle would result in fluorescence recovery by microtubule growth or transport with bound Ncd into the bleached ROI. We derived a model that accounts for this, together with recovery by diffusion and binding to microtubules in the ROI. The fluorescence recovery as a function of time, frap(t), normalized to converge to 1 at large t, is given by
where Feq is the fraction of Ncd that is free at equilibrium, Ceq is the fraction bound to microtubules, Feq+Ceq=1, w is the bleach spot radius, D is the diffusion cofficient, I0 and I1 are modified Bessel functions, koff is the dissociation rate constant, and σ is the width of the velocity distribution. The mean velocity is given by . Fits of the FLNcdVenus and HLNcdVenus FRAP data at the spindle equator p to the new model, denoted the diffusion-binding-transport model, yielded the kinetic parameters shown in Table 2. The coefficient of multiple determination, r2, was calculated using the MATLAB routine leasqr.m (Shrager et al., 1994) as a measure of the goodness of fit by different recovery models to the data.
Like the diffusion-binding model it expands upon, the diffusion-binding-transport model requires that k*onw2/D1. It reduces to the diffusion-binding model for small v or large koff, and to the Soumpasis pure-diffusion model (Soumpasis, 1983) for small Ceq. In these domains, the effect of transport on recovery is small and determination of velocity may be difficult. The data for TLNcdVenus, which exhibit a large koff, are an example of this reduction; fits of the tailless data to the diffusion-binding-transport model gave values that were not significantly different from the diffusion-binding model, and v≈0.
Fluorescence in FLncdVenus mitotic spindles was analyzed in time-lapse sequences acquired at 2.32 s intervals using the Radiance2100 confocal system described above. Early metaphase through anaphase spindles from cycle 10 embryos were rotated and aligned as necessary in ImageJ and cropped to eliminate centrosomes, astral microtubules, and cytoplasm. Image sequences were visually aligned in time by phase of mitosis, based on the microtubule density and bundling in the spindle. Alignments based on the ratio of pole to equator fluorescence or the image with maximal fluorescence displacement from the poles gave a higher overall s.e.m. after averaging and normalization. The equator was defined as the midpoint between poles in the first and last image of the sequence, and the median fluorescence position in each spindle half was calculated using the MATLAB routine, fluorescence_medians. Data for the half-spindles from different embryos were averaged, the median fluorescence positions were normalized from zero to one, and the mean normalized data were plotted versus time in Kaleidagraph v. 3.6.4 (Synergy Software).
We thank V. Bennett for the use of a Zeiss LSM510 confocal microscope, R. Tsien for an mRFP plasmid, and K. Su for performing HLncdVenus embryo viability tests. This work was supported by grants to S.A.E. from the National Institutes of Health and March of Dimes Foundation. M.A.H. is a 2008 Goldwater Scholar.