Calcium tips the balance: a microtubule plus end to lattice binding switch operates in the carboxyl terminus of BPAG1n4

Abstract

Microtubules (MTs) are integral to numerous cellular functions, such as cell adhesion, differentiation and intracellular transport. Their dynamics are largely controlled by diverse MT-interacting proteins, but the signalling mechanisms that regulate these interactions remain elusive. In this report, we identify a rapid, calcium-regulated switch between MT plus end interaction and lattice binding within the carboxyl terminus of BPAG1n4. This switch is EF-hand dependent, and mutations of the EF-hands abolish this dynamic behaviour. Our study thus uncovers a new, calcium-dependent regulatory mechanism for a spectraplakin, BPAG1n4, at the MT plus end.

Introduction

Microtubule (MT)-associated proteins (MAPs) and plus end-binding proteins (+TIPs) have a principal role in modulating the organization and dynamic behaviour of MTs. +TIPs not only regulate the growth and shrinkage of MTs [1, 2], but also their attachment to other cellular structures [3]. Disruption of these interactions has been linked to many human diseases [4, 5, 6]. So far, phosphorylation is the most well-known mechanism for controlling MT-interacting proteins, and it has been implicated in the spatial and cell-cycle-dependent regulation of many MAPs and +TIPs [7, 8, 9, 10]. Despite these insights, the potential exists for more signals that modulate the rapid coordination of this dynamic interaction system, yet their discovery has remained elusive.

BPAG1n4, a 600-kDa neuronal protein is a member of the spectraplakin family, which includes ACF-7 in mammals, and shot and Vab-10 in invertebrates. Spectraplakins can bind to actin and intermediate filaments as well as MTs, making them master coordinators of the cytoskeleton [11, 12, 13, 14]. BPAG1n4 has an essential role in retrograde axonal transport in sensory neurons, by virtue of functional domains that associate with the p150_Glued dynactin complex and with retrolinkin, an endosomal receptor [11, 15]. In addition, on the basis of sequence homology, the C-terminus of BPAG1n4 might contain MT-binding sites [10, 16, 17, 18].

In this study, we identify an extremely rapid mechanism for switching between MT plus end and lattice-binding activities in the C-terminus of BPAG1n4. This regulation is EF-hand and calcium (Ca²⁺) dependent. Alteration of Ca²⁺ responsiveness, resulting from mutations in the EF-hands, completely abolishes the rapid switch while rendering permanent the binding interaction with either the MT lattice or the plus end. This newly identified, Ca²⁺-dependent regulatory mechanism in spectraplakins might have a critical role in a number of diverse MT-associated processes. We thus provide the first evidence for a mechanism through which Ca²⁺ can directly and rapidly regulate the dynamic interaction of MTs with a key regulatory protein.

Results and Discussion

BP4tail preferentially binds to the plus end

The MT-binding domain (MTBD) at the C-terminus of BPAG1n4 consists of a Gas2-related segment [10, 11, 16] (supplementary Fig S1 online). Consistent with its sequence, green fluorescent protein (GFP)-MTBD decorated the MT network when expressed in Cos-7 cells (supplementary Fig S2A online; supplementary video 1 online, n>30 cells). Notably, downstream of the MTBD, the C-terminus harbours a short Sx-IP homology motif that might constitute an EB1-binding domain (EBBD) [17, 18]. Using an in vitro binding assay, we showed that EBBD could directly bind to EB1 (supplementary Fig S2D online), with a binding affinity of 4.5±0.74 μM (±s.d., n=3, supplementary Fig S2E online). The EB1 family is a central player at the plus end, responsible for the localization of many +TIPs [19]. In Cos-7 cells, GFP-EBBD robustly tracked the plus end, mimicking the behaviour of an EB1 family member, EB3 (supplementary Fig S2B,C online; supplementary video 1 online, n>30). The velocity of EBBD tip tracking (0.392±0.01 μm/s, ±s.e.m., n=5 cells) was similar to that observed for EB3 (0.352±0.03 μm/s, n=5), and falls within the range documented for other +TIPs [20].

Our findings demonstrate that the C-terminus of BPAG1n4 has two independent MT-interaction domains, one that underlies interaction with the MT lattice (MTBD) and the other that mediates association with the plus end through binding with EB1 family proteins (EBBD).

Although certain proteins such as CLASP2 and Drosophila msps also contain both MT lattice and EB1-binding sites, the interaction of these proteins with MTs is spatially regulated, and they interact either with the lattice or the plus end in the cell interior versus periphery [7, 21]. It is important to consider how these two different binding activities within the same short C-terminus stretch of BPAG1n4 co-function without impeding one another.

To understand the behaviour of the C-terminus of BPAG1n4, we expressed a construct spanning both the MTBD and the EBBD (BP4tail, supplementary Fig S1 online). We found that despite containing the MTBD, GFP-BP4tail localized predominantly to plus ends (Fig 1A; supplementary video 1 online, n>30 cells) and tracked them with a velocity of 0.418±0.01 μm/s (n=5), behaviour similar to that of EBBD alone. This suggests that the lattice-binding activity of the MTBD is suppressed in the context of BP4tail, leading us to question how the preferential interaction with MT plus ends is determined. We hypothesized that an active regulatory mechanism might underlie this selective interaction.

Figure 1.

A Ca²⁺-regulated rapid switch between MT plus end and lattice association. (A) Cos-7 cell expressing GFP-BP4tail. Inset; magnified boxed region. Comet position; red arrow (t=0 s), green arrowhead (t=6 s) and blue arrowhead (t=12 s). Colour overlay images (red; t=0 s, green; t=6 s and blue; t=12 s on the extreme right. Scale bar, 10 μm. (Supplementary Video 1 online). (B,B′) GFP-BP4tail changes localization in response to ionomycin. A representative Cos-7 cell expressing GFP-BP4tail (B) before and (B′) 30 s after ionomycin addition. Ionomycin was added at t=0 s. Inset right, magnified boxed region showing frames at 4 s intervals. Red arrow=original comet position (t=−8/30 s), red arrowheads mark subsequent positions. Scale bar, 10 μm (supplementary Video 2 online). (C) Representative linescans plotting the normalized distribution of GFP-BP4tail along the MT, before (green) and 30 s after (red) addition of ionomycin. The graph represents the average normalized intensity (±s.e.m.) of 10–15 individual MTs from the cell in (B,B′), plotted against distance from the MT end. Fluorescence was normalized to the average fluorescence intensity of the cell, such that average fluorescence equalled 1. (D–F) n=15 cells, 10–15 MTs per cell, Student’s paired t-test. (D) Ionomycin treatment causes a significant decrease in GFP-BP4tail fluorescence at the MT tip (*P<0.05, ±s.e.m.). (E) Ionomycin treatment causes a significant increase in GFP-BP4tail fluorescence along the MT lattice (**P<0.01, ±s.e.m.). (F) GFP-BP4tail redistributes after ionomycin treatment. Ratio of normalized fluorescence at MT tip (0–2 μm) to that at MT lattice (3–5 μm) before and 30 s after ionomycin (**P<0.01). (G) In vitro pull-down assays using purified MBP-EB1 and Flag-BP4tail. 2 mM Ca²⁺ decreases the binding of Flag-BP4tail to MBP-EB1 (lane 5 versus lane 4), n>3,±s.e.m., *P<0.05. (H) Co-sedimentation of BP4tail with polymerized MTs. High Ca²⁺ results in a trend towards increased binding of Flag-BP4tail to MTs (lane 5 versus lane 4), n>3,±s.e.m., P=0.07. GFP, green fluorescent protein; MT, microtubule.

Ca²⁺-regulated switch from plus end to MT lattice

In a search for candidate regulatory mechanisms, we analysed the sequence of our expression construct, noting two Ca²⁺-binding EF-hand motifs immediately upstream of the MTBD (domain structure in supplementary Fig S1 online). Their proximity to both the MTBD and EBBD prompted us to consider that Ca²⁺ might have a role in controlling binding selectivity of the C-terminus of BPAG1n4. To investigate this possibility, we treated Cos-7 cells expressing GFP-BP4tail (EF-MTBD-EBBD) with the Ca²⁺ ionophore, ionomycin, to induce an increase in the cytosolic Ca²⁺ levels. Before treatment, GFP-BP4tail preferentially localizes to comets at MT plus ends and shows tip-tracking behaviour (Fig 1A,B). Fluorescence intensity linescan analysis verified that BP4tail forms a 2–3 μm long comet at the MT tip, similar to the reported localization of EB1 (Fig 1C, n=15 cells) [17]. Remarkably, following ionomycin treatment, we observed a striking translocation of BP4tail from the plus ends to the MT lattice, and both comet localization and tip tracking were mostly abolished (Fig 1B,B’; supplementary video 2 online, n=15). This switch could occur as quickly as 10 s, and overall was completed 30–40 s after treatment (supplementary Video 2 online). To obtain quantitative evidence, we conducted linescan analysis before and 30 seconds after ionomycin treatment. This analysis revealed a decrease of fluorescence at the MT tip (0–2 μm, Fig 1D, P<0.05), accompanied by an increase in fluorescence along the lattice (3–5 μm, Fig 1E, P<0.01), indicating that high Ca²⁺ might alter the binding preference of BP4tail, resulting in its redistribution along the MT (Fig 1E, P<0.01). These results demonstrate that BP4tail undergoes a switch between plus end and lattice binding in response to elevated Ca²⁺ levels.

To obtain more support, we examined the effect of Ca²⁺ on the binding of Flag-BP4tail to purified maltose binding protein (MBP)-EB1 in vitro (Fig 1G). We found that 2 mM Ca²⁺ significantly inhibited the binding of BP4tail to EB1 (P<0.05, n>3, Fig 1G). We next assessed whether Ca²⁺ could alter the interaction between BP4tail and MTs. Polymerized MTs were incubated with Flag-BP4tail in either 0 or 2 mM Ca²⁺ buffer, and then MTs and any bound proteins were co-sedimented. Although not statistically significant (P=0.09, n>3, Fig 1H), a clear trend indicates that increased MT binding by BP4tail occurs at 2 mM Ca²⁺. Together, these in vitro results indicate that Ca²⁺ regulation might act more strongly at the EB1-binding site, which might be sufficient to cause the observed switch. They support our in vivo finding that the interaction between BP4tail and EB1/MTs is Ca²⁺ dependent.

The MTBD is essential for the Ca²⁺-regulated switch

We hypothesized that Ca²⁺ binding to the EF-hands might underlie the observed regulation. To test this, we assessed whether Ca²⁺ could directly alter MT structure or EB1-binding sites. We examined the behaviour of GFP-EB3 and GFP-EBBD (without EF-hands). Both proteins track the MT plus end after ionomycin treatment (Fig 2A–D), indicating that the observed switch of BP4tail from MT tips to lattice in response to elevated Ca²⁺ is unlikely to be a result of structural changes or of a redistribution of EB1 family proteins. Next, we deleted the MTBD from BP4tail, creating an EF-EBBD construct (supplementary Fig S1 online). Like BP4tail, GFP-EF-EBBD tracked MT plus ends; yet, despite the presence of EF-hands, it remained unresponsive to ionomycin (Fig 2E–F; supplementary video 3 online, n=10). Linescan measurements taken before and 30 s after ionomycin treatment show no change in the tip localization of EF-EBBD (Fig 2F, n=10), and EF-EBBD continues to tip track for several minutes after treatment. Together, these data demonstrate that increased Ca²⁺ does not change EF-EBBD-specific binding activity at the plus end, and that the switch of BP4tail from the MT tip to the lattice requires the MTBD. Finally, as the presence of the MTBD appears essential for the switch from the plus end to the lattice, we examined the effect of increased Ca²⁺ on the localization of the MTBD. Both GFP-MTBD (Fig 2H) and GFP-EF-MTBD share similar lattice binding before and after ionomycin treatment (Fig 2G; supplementary video 4 online, n=10). Our data demonstrate that in order for the Ca²⁺-driven switch to be functional, both lattice and plus end-binding domains have to be present.

Figure 2.

The MTBD is essential for the Ca²⁺-regulated switch. Addition of ionomycin does not affect the tip tracking of (A,A′) EB3; (C′,C′) EBBD and (E,E′) EF-EBBD. Representative Cos-7 cells expressing GFP-EB3, GFP-EBBD and GFP-EF-EBBD (A,C,E) before and (A′,C′,E′) 30 s after ionomycin addition. Ionomycin was added at t=0 s. Inset right, magnified boxed region showing frames at 4 s intervals. Red arrow=original comet position (t=−8/30 s), red arrowheads mark subsequent positions. Scale bar, 10 μm. (E,E′) corresponds to supplementary Video 3 online. (B,D,F) Representative linescans plotting the normalized distribution of (B) GFP-EB3, (D) GFP-EBBD and (F) GFP-EF-EBBD along the MT, before (green) sand 30 s after (red) addition of ionomycin (10–15 MTs,±s.e.m.). (G) GFP-EF-MTBD localizes to MTs before and after the addition of ionomycin. Inset=magnified boxed region. Scale bar, 10 μm (supplementary Video 4 online). (H) Addition of ionomycin has no effect on GFP-MTBD-labelled MTs. Inset=magnified boxed region. Scale bar, 10 μm. EBBD, EB1-binding domain; GFP, green fluorescent protein; MT, microtubule; MTBD, MT-binding domain.

Suppression of lattice binding requires EF-hands

To define the role of the EF-hands in the Ca²⁺-regulated switch, we deleted both EF-hand motifs from BP4tail (ΔEF). GFP-BP4tail ΔEF localized exclusively to the MT lattice (Fig 3A, n>30). Similar to the effect of deletion, when three conserved Ca²⁺-binding aspartate residues in both EF-hands were replaced by alanine (Fig 3E), the mutated BP4tail (EF1,2MUT) mimicked the localization of BP4tail ΔEF, binding to the lattice with no observable plus end tracking (Fig 3B, n>30). We then further dissected the role of each EF-hand by introducing mutations in one hand at a time. We found that mutations in EF-hand 1 (BP4tail EF1MUT) also resulted in lattice localizatin (Fig 3C, n>20). In contrast, when EF-hand 2 was mutated (BP4tail EF2MUT), the mutant protein maintained specific association with the plus end, making it the only EF-hand mutant to retain the +TIP behaviour of the wild-type BP4tail construct at basal Ca²⁺ levels (Fig 3D, n>20). These results strongly suggest that the condition of the two EF-hands determine the selective interaction of BP4tail with the MT plus ends versus the lattice.

Figure 3.

The suppression of lattice interaction requires the EF-hands. Cos-7 cells expressing (A) BP4tail ΔEF, (B) BP4tail EF1,2MUT, (C) BP4tail EF1MUT and (D) BP4tail EF2MUT. Red arrow=comet position (t=0 s), red arrowheads mark subsequent positions (t=4 s and t=8 s). Coloured overlay images (red; t=0 s, green; t=4 s and blue; t=8 s) on the extreme right. Schematic of the domain structure and mutations shown on left of each image. Scale bar, 10 μm. (E) Domain structure of BP4tail and EF-hand mutants. EBBD, EB1-binding domain; MTBD, MT-binding domain.

Supporting evidence was gained from co-immunoprecipitation experiments in which Flag-EB1 was co-transfected in HEK cells with GFP-BP4tail or its mutants (supplementary Fig S3A online). GFP-BP4tail, but not GFP alone, was co-precipitated by Flag antibody. However, the co-precipitation of BP4tail by Flag-EB1 was impaired by deletion or mutation of both EF-hands (ΔEF, EF1,2MUT), or mutation of only EF-hand 1 (EF1MUT). Consistent with our live-imaging data, mutations in EF-hand 2 (EF2MUT) did not strongly affect the interaction of BP4tail with EB1. The immunoprecipitation was performed in a Ca²⁺-free environment, suggesting again that the condition of the EF-hands can directly affect BP4tail-binding activity, even in the absence of Ca²⁺.

Importantly, we found that mutation of the EF-hands had no major effect on the behaviour of EF-MTBD or EF-EBBD, which retain their localization to the MT lattice and plus end, respectively (supplementary Fig S4A,B online, n=10). Thus, although mutation of EF-hands can drastically alter the localization of BP4tail, their regulatory effect requires the presence of both the MTBD and the EBBD.

The Ca²⁺-regulated rapid switch is EF-hand dependent

As mutation of the EF-hands can affect the preferential localization of BP4tail, we questioned whether they also altered BP4tail’s ability to rapidly switch localization in response to increased Ca²⁺. To address this, we repeated the live-imaging and ionomycin experiments using the EF-hand mutants of BP4tail. As expected, mutation of both EF-hands (EF1,2MUT) or of EF-hand 1 (EF1MUT), which result in lattice binding under basal condition, also abolish any Ca²⁺ sensitivity, and their localization along MTs remained unaltered by ionomycin treatment (Fig 4A–B’, n=10). Remarkably, when we examined BP4tail EF2MUT, we found that its plus end localization was completely insensitive to ionomycin treatment (Fig 4C–G; supplementary video 5 online). Like EF-EBBD (supplementary Video 3 online), BP4tail EF2MUT continued to track the plus end, without any change in activity for minutes after ionomycin treatment. The fluorescent intensity at the tip and at the lattice remained unchanged on treatment (Fig 4E–F, n=10, P>0.05). Thus, although the tip-tracking behaviour of wild-type BP4tail and EF2MUT are alike, the inability of EF2MUT to switch from the plus end to the lattice on ionomycin treatment strongly suggests that a functional EF-hand 2 is required for this rapid switch.

Figure 4.

The Ca²⁺-regulated rapid switch from plus end to lattice binding is EF-hand dependent. Addition of ionomycin does not affect the localization of (A,A′) BP4tail EF1,MUT, (B,B′) BP4tail EF1MUT or (C,C′) BP4tail EF2MUT. Representative Cos-7 cells expressing GFP-BP4tail EF1,2MUT, EF1MUT or EF2MUT (A,B,C) before and (A′,B′,C′) 30 s after ionomycin addition. Ionomycin was added at t=0 s. Inset right, magnified boxed region showing frames at 4 s intervals. (C,C′) Red arrow=original comet position (t=−8/30 s), red arrowheads mark subsequent positions. Schematic of the domain structure and mutation shown on left of each image. Scale bar, 10 μm. (C,C′) corresponds to supplementary Video 5 online. (D) Representative linescans plotting the normalized distribution of GFP-BP4tail EF2MUT along the MT, before (green) and 30 s after (red) addition of ionomycin. (10–15 MTs, ±s.e.m., cell shown in C,C′). (E–G) n=10 cells, 10–15 individual MTs per cell, Student’s paired t-test. (E) Ionomycin treatment does not alter GFP-BP4tail EF2MUT fluorescence at the MT tip (P>0.05, ±s.e.m.). (F) Ionomycin treatment does not change GFP-BP4tail EF2MUT fluorescence along the MT lattice (P>0.05, ±s.e.m.). (G) Ratio of normalized fluorescence at MT tip (0–2 μm) to that at the lattice (3–5 μm) before and 30 s after ionomycin (P>0.05). (H) In vitro pull-down between EF-hand mutants of BP4tail and MBP-EB1. High Ca²⁺ does not significantly alter the interaction with EB1 (lane 5 versus lane 4). n>3, ±s.e.m., P>0.05. (I) Co-sedimentation of the EF-hand mutants of BP4tail with polymerized MTs. High Ca²⁺ does not alter the binding to MTs (lane 4 versus lane 3). n>3, ±s.e.m., P>0.05. GFP, green fluorescent protein; MT, microtubule.

Consistent with our live-imaging data, mutation of the EF-hands abolishes the Ca²⁺ regulation previously observed for wild-type BP4tail in vitro. Increased Ca²⁺ has no significant effect on the interaction of EF2MUT with EB1 (P>0.05, n>3, Fig 4H). Similarly, the co-sedimentation of EF1MUT with MTs is unaltered by high Ca²⁺ (P>0.05, n>3, Fig 4I). As expected, the interaction between EF1,2MUT and EB1 or MTs was also insensitive to high Ca²⁺ (P>0.05, n>3, Fig 4H,I). Thus, functional EF-hands are essential for the observed Ca²⁺ regulation.

A hypothetical model of the Ca²⁺-regulated switch

Taken together, our data uncover a new, Ca²⁺-regulated dynamic interaction of the C-terminus of BPAG1n4 with the MT plus end and lattice. We demonstrate that the EF-hands underlie the Ca²⁺ regulation of this preferential interaction. Here, we propose a model of the Ca²⁺-regulated switch (Fig 5).

Figure 5.

A hypothetical model of the Ca²⁺ switch. Model of the Ca²⁺-regulated switch in BP4tail. In the ‘closed’ conformation, lattice binding is suppressed allowing +TIP binding, while in the ‘open’ conformation lattice binding is dominant (upper). Wild-type BP4tail undergoes a transition from closed to open conformation in the presence of high Ca²⁺, EF2MUT remains locked in the closed conformation, while all other constructs (EF1MUT, EF1,2MUT and ΔEF) remain locked in the open conformation (lower). EBBD, EB1-binding domain; MTBD, MT-binding domain.

In the absence of the EF-hands, BP4tail localizes to the MT lattice (Fig 3A,B). Similar behaviour has been observed in Drosophila shot, in which the homologous domain, lacking the upstream EF-hands, was found to decorate the lattice [22]. However, when functional EF-hands are present, Ca²⁺ binding might alter the domain conformation to modulate the selectivity of the two binding activities. At basal or 0 mM Ca²⁺ levels, the proposed ‘closed’ conformation of BP4tail with the two EF-hands present might suppress the functional activity of the MTBD, preventing lattice interaction and leading to a preferential association with EB1 at plus ends (Fig 1A). When Ca²⁺ levels increase, Ca²⁺ binding to the EF-hands might modify the domain to its ‘open’ conformation, exposing the MTBD-binding site, leading to the observed rapid switch in localization (Fig 1B,B’; supplementary video 2 online). Mutation of EF-hand 1 led to lattice binding, strongly suggesting that EF-hand 1 might have a role in suppressing the MTBD activity (Fig 3C). In contrast, EF-hand 2 appears more important for the release of MTBD suppression following an increase in cytosolic Ca²⁺. Disruption of EF-hand 2 locked BP4tail EF2MUT in a ‘closed’ conformation, retaining all the +TIP features of wild-type BP4tail, even under high Ca²⁺ conditions (Fig 4C,C’; supplementary video 5 online). It is apparent that Ca²⁺ binding to EF-hand 1 alone is not sufficient to release the suppression of lattice interaction. Thus, the Ca²⁺-regulated switch is exquisitely sensitive to disruption of the EF-hands and requires both EF-hands to be present and functional.

While Ca²⁺ has been implicated in the regulation of MT dynamics through MAPs and +TIPs, it usually acts upstream of or directly on a kinase [23, 24, 25]. Although STIM1 is a +TIP that contains EF-hands, its association with EB1 does not seem to be directly Ca²⁺ regulated [26]. Our study presents the first evidence of Ca²⁺ directly and rapidly modulating the interaction of a protein at the plus end, uncovering a new mechanism through which Ca²⁺ signalling can directly regulate the dynamic interactions of MTs with regulatory proteins. This process might have an essential role in many cellular activities, including cellular adhesion, differentiation and axonal transport.

Like other spectraplakins, BPAG1n4 is a massive protein, containing several functional domains. At the ultrastructural level, BPAG1n4 mainly localizes to vesicles along MTs [11], suggesting that its interaction with the plus end might be dynamic and transient. Future investigations must overcome technical hurdles to examine the Ca²⁺ regulation of full-length BPAG1n4. The conservation of the C-terminal domain of BPAG1n4 in other members of the spectraplakin family, such as ACF7, shot and Vab-10, suggests that Ca²⁺ regulation of the plus end interaction might have a vital role in their functional activities as well [27]. Addressing the potential role of this Ca²⁺ switch might be crucial to fully understanding spectraplakins’ many diverse functions.

Methods

Ionomycin treatment. Cos-7 cells were imaged in extracellular solution containing 2 mM CaCl₂ under perfusion. Cells were imaged for 1–2 min after the addition of ionomycin (2 μM), acquiring one frame every 2 s.

Linescan analysis. Analysis was performed using MetaMorph v7.7. Briefly, 10–15 individual MTs were analysed before and 30 s after the addition of ionomycin. Fluorescence measurements were normalized to the average fluorescence of the cell (average fluorescence =1). The effect of ionomycin was determined using a paired Student’s t-test.

EB1-binding assay. MBP-EB1 was immobilized on amylose beads and used to pull down Flag-BP4tail (wild-type/mutants) from HEK cell lysate in buffer containing either 0 or 2 mM CaCl₂. Equivalent volumes of the supernatant and bound fractions were processed for immunoblotting. The Flag signal was normalized to MBP, and EB1 binding was calculated by dividing the intensity of the bound band by the summed intensities of the supernatant and bound bands. Binding at 2 mM Ca²⁺ was expressed as a fraction of binding in the absence of Ca²⁺ (n>3, t-test).

MT co-sedimentation. A modified in vivo co-sedimentation assay was used (BK038, Cytoskeleton). HEK cells transfected with Flag-BP4tail (wild-type/mutants) were treated with nocodazole to depolymerize MTs, and then lysed and processed to acquire the Flag protein and the soluble tubulin pool. MTs were then repolymerized by taxol, and subjected to co-sedimentation by standard protocols. The results were quantified as described above (n>3, t-test). For details, see supplementary methods online.

Supplementary information is available at EMBO reports online (http://www.emboreports.org).