Despite extensive studies, the molecular mechanisms of Tau binding to microtubules (MTs) and its consequences on MT stability still remain unclear. It is especially true in cells where the spatiotemporal distribution of Tau–MT interactions is unknown. Using Förster resonance energy transfer (FRET), we showed that the Tau–MT interaction was distributed along MTs in periodic hotspots of high and low FRET intensities. Fluorescence recovery after photobleaching (FRAP) revealed a two-phase exchange of Tau with MTs as a rapid diffusion followed by a slower binding phase. A real-time FRET assay showed that high FRET occurred simultaneously with rescue and pause transitions at MT ends. To further explore the functional interaction of Tau with MTs, the binding of paclitaxel (PTX), tubulin acetylation induced by trichostatin A (TSA), and the expression of non-acetylatable tubulin were used. With PTX and TSA, FRAP curves best fitted a single phase with a long time constant, whereas with non-acetylatable α-tubulin, curves best fitted a two phase recovery. Upon incubation with PTX and TSA, the number of high and low FRET hotspots decreased by up to 50% and no hotspot was observed during rescue and pause transitions. In the presence of non-acetylatable α-tubulin, a 34% increase in low FRET hotspots occurred, and our real-time FRET assay revealed that low FRET hotspots appeared with MTs recovering growth. In conclusion, we have identified, by FRET and FRAP, a discrete Tau–MT interaction, in which Tau could induce conformational changes of MTs, favoring recovery of MT self-assembly.
In eukaryotic cells the microtubule cytoskeleton consists of a functional network involved in a diverse range of cellular activities such as mitosis, meiosis, motility, morphogenesis and intracellular trafficking of both macromolecules and organelles. The core component of microtubules (MTs) is a heterodimer of the α- and β-tubulin proteins. MTs are highly dynamic, and exhibit a non-equilibrium behavior termed dynamic instability (Mitchison and Kirschner, 1984). Indeed, MTs undergo rapid stochastic transitions between growth and shortening, as a result of the association and/or dissociation of tubulin dimers at its extremities. In this process, the microtubule-associated protein Tau promotes tubulin polymerization and stabilizes MTs both in vitro and when microinjected into cells (Weingarten et al., 1975; Cleveland et al., 1977; Drubin and Kirschner, 1986). In eukaryotic cells, Tau protein is a family of six isoforms, each with either three or four MT-binding repeats located in the C-terminal half of the protein, and zero to two inserts located in the N-terminal portion (Goedert and Jakes, 1990).
Much of the interest in Tau comes from its presence in the neurofibrillary tangles of age-related neurodegenerative pathologies such as Alzheimer's disease (see review Buée et al., 2000). Experimental models of tauopathies strongly suggest that Tau-mediated neurodegeneration results from a combination of toxic gain of function and the loss of normal Tau function (Lee and Rook, 1992; Preuss et al., 1997; Bunker et al., 2004; Feinstein and Wilson, 2005; LeBoeuf et al., 2008). In fact, most studies have focused on the disruption of normal Tau activity, through, for example, post-translational modifications such as hyper-phosphorylation (Trinczek et al., 1995), to highlight its inability to properly regulate MT dynamics.
To understand Tau functions, many workers have investigated its interaction with tubulin, mainly from three-dimensional reconstructions of images obtained by cryo-electron microscopy (Ackmann et al., 2000; Goode et al., 2000; Al-Bassam et al., 2002; Krebs et al., 2004; Santarella et al., 2004). However, no consensus exists about binding model(s), in part because of the unfolded states of Tau either when free or when complexed with MTs. Most binding models have been proposed based on experiments carried out with purified proteins, however, it appears increasingly evident that the Tau–MT interaction depends on both the ratio of Tau/tubulin concentrations and the concentrations of MT-stabilizing compounds such as paclitaxel (PTX) (Panda et al., 1995; Kar et al., 2003; Sillen et al., 2007; Park et al., 2008). The question of the molecular mechanisms of Tau binding to MTs and its consequences on MT stability, especially in the cell microenvironment remains open.
Here, we focus on the characterization of the Tau–MT interaction in live cells by Förster resonance energy transfer (FRET) and fluorescence recovery after photobleaching (FRAP) coupled to confocal imaging. Real-time FRET analysis provided spatiotemporal resolution of the Tau–MT interaction, in particular details of the MT-stabilizing function of Tau. Our approach highlighted a discrete interaction of Tau along MTs in relation to a diffusion-binding model. Focusing on single MTs in cells, our time-series of FRET provided evidence of a FRET hotspot at the end of MTs during rescue and pause transitions. To investigate the relationship between the interactions and the effect of Tau onto MTs we altered MT structure using the MT-stabilizing drug PTX, trichostatin A (TSA)-induced acetylation of α-tubulin and expression of a non-acetylatable α-tubulin. These factors are well known to cause conformational changes of tubulin dimers resulting in modulated MT dynamics (L'Hernault and Rosenbaum, 1985; Derry et al., 1995; Piperno et al., 1987; Nogales et al., 1995; Honore et al., 2004; Dompierre et al., 2007; Zilberman et al., 2009). We showed that PTX and the acetylation level of α-tubulin affect differently both the distribution of FRET hotspots and the diffusion-binding model. A real-time FRET analysis of the MT dynamics revealed that MT modifications regulate interaction between Tau and MTs.
FRET identified that Tau–microtubule interactions occur in discrete spots
We determined, by sensitized FRET imaging, the localization of the Tau–MT interaction in A549 cells expressing EGFP–Tau (as donor) and mCherry–tubulin (as acceptor; Fig. 1A). FRET images were obtained using Wouters normalization (see Materials and Methods for details). The simple observation of EGFP–Tau and mCherry–tubulin colocalization clearly showed a continuous distribution of Tau all along MTs. In contrast, the FRET signal is distributed discontinuously in clusters of pixels along the length of MTs that we refer to as FRET hotspots. Since the distance between the donor/acceptor couple is the main factor determining FRET efficiency (Stryer and Haugland, 1967), such FRET hotspots show that Tau is in close contact with the MT lattice. We distinguished two populations of FRET hotspots: ones with low FRET efficiency (normalized FRET intensity between 0.1 and 0.5) and ones with high FRET efficiency (values higher than 0.5). Moreover, the fluorescence emissions of EGFP–Tau and mCherry–tubulin were measured on 2-µm long sections of MTs (n = 10) and compared with the intensity profiles of FRET (Fig. 1B). Although we observed only a weak variation of the donor and acceptor intensities (upper graph), each peak of FRET represents a hotspot (lower graph). In addition, the average distance between FRET hotspots (n = 90 measured close to MT plus-ends) was 0.47±0.02 µm. This corresponds to a periodicity of about 60 tubulin dimers. Such a punctuated interaction along MTs would be sufficient to ensure stabilization of MTs. To validate the significance of the measured FRET signals in spots, two controls were performed. First, cells expressing the two-colored α-tubulin isotypes (as a positive control for the interaction) showed that the FRET signal was regularly distributed along the MTs with minor variations in the FRET intensity (supplementary material Fig. S1A,B). Second, the FRET signal from cells expressing the free EGFP/mCherry–tubulin pairs (as a negative control for the interaction) was very low intensity in the cytosol (<0.1), and negligible on the 2-µm long sections of MTs (supplementary material Fig. S1C,D).
As an alternative method to confirm our findings, we compared the quenched and unquenched EGFP–Tau emission after specific photobleaching of the mCherry–tubulin fluorophore (Fig. 1C, panels in false-color). Prior to acceptor photobleaching, MTs were quite uniformly labeled with EGFP–Tau (upper panel in Fig. 1C). After photobleaching (lowest panel in Fig. 1C), the fluorescence emission of EGFP–Tau increased in spots all along the length of MTs (black arrowheads in the ‘recovery’ area). No significant fluorescence intensity fluctuation was observed inside the control region. This experiment confirmed the distribution of the Tau–MT interaction in FRET hotspots all along MTs.
We also looked at Caco-2 and SK-N-SH cells to see whether FRET hotspots could be observed because these two cell lines have different endogenous expression levels of Tau (supplementary material Fig. S2A). We confirmed this distribution of FRET hotspots in both these cell lines (supplementary material Fig. S2B). Note that the expression of fluorescent Tau in A549 cells (‘A549 + Tau’ in supplementary material Fig. S2A) was comparable to the endogenous level of Tau in wild-type SK-N-SH cells. Our results indicate that the interaction between EGFP–Tau and MTs is independent of the endogenous expression of Tau in cells.
For the first time, we clearly demonstrate in several cell lines that Tau interacts discontinuously along the length of MTs. The various FRET hotspot intensities could highlight structural modifications of MTs requiring a specific binding of Tau. The dynamics of exchange and binding of Tau with MT interaction was then investigated by FRAP.
Tau–microtubule interaction evidenced and characterized by FRAP analysis
The rate of fluorescence recovery indicates how fast neighboring fluorescent molecules arrive to fill a bleached zone. This mobility of proteins depends on both diffusion and potential binding interactions. First, FRAP experiments were performed to monitor the recovery of EGFP–Tau fluorescence in the presence of MTs versus in the absence of MT (Fig. 2). In the presence of the MT network (open squares), recovery curves of EGFP–Tau was best fitted by a two-exponential function. Indeed, analysis of data with two exponentials gave a much smaller residual least-squares sum and a better spreading of residuals (inset a, black curve) than with one exponential (gray curve in inset a). To disassemble MTs, cells were treated with nocodazole (Jordan et al., 1992). Without MTs (Fig. 2, black squares), the FRAP recovery curve of EGFP–Tau best fitted a mono-exponential equation (as also shown in inset b). Analyzing FRAP curves enabled us to determine parameters such as time constants τ1 and τ2 associated with the diffusion and binding phases, respectively (supplementary material Table S1). In the presence of MTs, the calculated τ1 was 1.8 seconds. This value is similar to the one found in the absence of MT (τ1 = 2.6 seconds, as reported in supplementary material Table S1). After this rapid diffusion phase, fluorescence recovery due to bound EGFP–Tau molecules occurred with a second time constant which was 10-fold longer (τ2 = 18.6 seconds).
FRAP enabled us to identify that the binding of EGFP–Tau to MT impacts directly on the mobility of Tau in cells, even if we cannot rule out some additional molecular sieving effects within the MT network. Next, we carried out a real-time FRET assay to obtain the spatiotemporal distribution of FRET hotspots. This method was designed to show details of the Tau–MT interaction during MT growth and shrinkage.
Real-time FRET reveals that hotspots appear during rescue and pause phases
We acquired time-lapse image sequences coupled with FRET analysis to determine the spatiotemporal position of FRET hotspots during shrinkage and growth phases of MTs. A representative example is reported in Fig. 3. During the MT shortening (time 0 to 11.5 seconds on Fig. 3A), high and low FRET hotspots vanished one after the other, suggesting that hotspots did not avoid the shortening phase. When the MT shrinkage stopped (called rescue of catastrophe), a high FRET hotspot appeared (time 11.5–16.1 seconds, white arrow on Fig. 3A). Then, once the growth phase was robust, this FRET hotspot disappeared (at time 18.4 seconds). This seems to indicate a possible structural arrangement of the MT plus-end following Tau–MT interactions supporting the MT polymerization. Other hotspots appeared throughout the growing phase and finally, a last hotspot marked the plus-end when MT growth stopped (time 41.4 seconds to the end of record). This may correspond to a pause transition before a catastrophe event (white arrowhead). A time-resolved FRET layout of images in a kymograph provides a better visualization of FRET hotspots at the plus-end of the MT recovering a growth phase (white arrows on Fig. 3B) and in pause (white arrowheads). We noticed from the kymograph that the FRET hotspots were not very long-lived and so the 0.47 µm spacing did not lead to a characteristic banded pattern.
This spatiotemporal analysis of FRET showed that hotspots appeared at the plus-end of MT initiation (rescue) and cessation (pause) of growth. During a rescue transition, Tau could contribute to a molecular rearrangement of the tubulin dimers at the plus-end of the MT, creating a stable region to favor the assembly of MTs. During a pause transition, the FRET hotspot at the plus-end of MTs could delay a catastrophe phase but without preventing it. The interaction of Tau in hotspots along the length of MT could help to slow down MT shortening in cells.
Microtubule modifications strongly affect the Tau–microtubule interaction
Apart from endogenous proteins such as MAPs, MT structure and stability are known to be regulated by post-translational modification such as acetylation of α-tubulin on the K40 residue, or by the binding of drugs such as PTX (Nogales et al., 1995; Nogales et al., 1998; Snyder et al., 2001). In order to gain further insight into the molecular mechanism of Tau binding to MTs in cells, we studied the influence of these modifications on Tau–MT interactions by FRET, FRAP and real-time FRET. For this purpose, we either incubated cells with PTX or TSA as an inhibitor of HDAC6 to enhance the level of acetylated α-tubulin, or we expressed fluorescent non-acetylatable α-tubulin (see Material and Methods section).
Two quantitative parameters were determined: the overall intensity of FRET in cells (Fig. 4A) and the number of pixels belonging to high and low FRET hotspots (Fig. 4B). To determine overall intensity, cells were outlined on FRET images and values were expressed per square micrometer to compare experimental conditions. From Fig. 4A, cells treated with PTX and TSA displayed a 30% decrease of overall FRET intensity compared with untreated cells. In contrast, cells expressing the non-acetylatable α-tubulin showed a 20% increase of overall FRET intensity. FRET images in Fig. 4C show that in cells treated with PTX or TSA (middle panels) hotspots were mainly of the low FRET class, whereas the presence of non-acetylatable α-tubulin led to a higher FRET intensity (lower panel). As control, we confirmed that differences of FRET intensities were not due to modulator-induced variations of donor and acceptor emission (supplementary material Fig. S3).
In Fig. 4B, we compared the percentage of pixels in the high and low FRET classes following treatment with modulators. For untreated cells (black bars), 28% and 7% of pixels were of low and high FRET levels, respectively. In the presence of PTX or TSA (Fig. 4B, red and blue bars, respectively), there was a loss of about a quarter of the low FRET pixels and about half of the high FRET pixels. In contrast, the expression of non-acetylatable α-tubulin (Fig. 4B, white bar) resulted in a 34% increase in low FRET pixels, whereas the number of high FRET pixels was unchanged. Furthermore, we measured an average distance of 0.57±0.02 µm between hotspots in cells treated with PTX or TSA, confirming that there were fewer high and low FRET hotspots than in untreated cells. In cells expressing mutant tubulin an inter-hotspot distance of 0.28±0.01 µm was obtained, which is related to the higher number of FRET hotspots along the length of MTs.
As supplementary controls, we checked the acetylation-independent effect of PTX on Tau–MT interactions (supplementary material Fig. S4). Indeed, it has been described that PTX can promote the acetylation of α-tubulin (Piperno et al., 1987). First, western blotting of tubulin acetylation in cells treated with PTX showed no relevant acetylation increase of soluble and incorporated α-tubulin in comparison with untreated cells in our experimental conditions. Note the drastic increase of tubulin acetylation in cells incubated with TSA. Second, for cells expressing non-acetylatable tubulin and treated with PTX, the overall intensity and the amounts of low and high FRET pixels decreased about 25% compared with untreated cells (supplementary material Fig. S5). No relevant change was measured in cells treated with TSA.
Our observations highlighted that PTX and TSA decrease the number of both high and low FRET hotspots, indicating a decrease in Tau–MT interactions. In contrast, expression of non-acetylatable tubulin led mainly to an increase of low FRET hotspots, which suggests additional interaction of Tau on MTs. Next, we investigated how MT modifications can modulate the dynamic exchange of Tau with MTs.
Microtubule modifications modulate the diffusion interaction model of Tau
FRAP experiments were performed to determine the effects of PTX and tubulin acetylation on EGFP–Tau mobility (Fig. 5). Following an incubation with PTX or TSA (triangle and diamond, respectively), FRAP recovery curves of EGFP–Tau were best fitted by a one-exponential equation (insets a and b in Fig. 5). Nevertheless, time constants τ1 were higher (∼8 seconds) compared with that obtained in cells treated with nocodazole (∼2.6 seconds; see supplementary material Table S1). Although we could not separate FRAP recovery curves into two phases, PTX and TSA curves reflect both diffusion and binding of EGFP–Tau. Therefore, in cells treated with PTX or TSA, diffusion and binding are probably intermixed during FRAP recovery, with a time constant τ1 value between the two τ1 and τ2 values of EGFP–Tau in untreated cells. This demonstrates that the accessibility and/or the three-dimensional structure of the binding loci of Tau are strongly modified by the binding of PTX and the acetylation of tubulin.
In cells expressing non-acetylatable tubulin, FRAP recovery curves were best fitted by the sum of two exponentials (square symbol in Fig. 5; see also inset c). Moreover, we found a short first time constant τ1 (1.5 seconds) and a long second time constant τ2 (11.6 seconds). These values are comparable to those calculated for untreated cells. The expression of non-acetylatable tubulin supported a diffusion-binding model with two separated phases.
Our results showed that the binding of PTX and acetylation levels result in different FRAP recovery curves. Nevertheless it still corresponded to Tau–MT binding equilibrium in all conditions. This indicates that regions where PTX and acetylation occurs may impact on the binding of Tau on MTs. We then tested whether modifications of MTs alter the effects of Tau on MT dynamics.
Microtubule modifications affect the microtubule-stabilizing effect of Tau
Our real-time FRET assay was performed in cells treated with PTX or TSA, or expressing non-acetylatable tubulin (Fig. 6). As already pointed out for Fig. 4, there was a global decrease of FRET in the presence of PTX and TSA (Fig. 6A,B), whereas the level of FRET was significantly higher in cells expressing non-acetylatable tubulin. Moreover, in the presence of PTX or TSA, we did not observe FRET hotspots at the plus-end of MTs in rescue and pause transitions (dashed circles in Fig. 6A,B). In contrast, cells expressing non-acetylatable tubulin showed low FRET hotspots on MTs recovering a growth phase (white arrows in Fig. 6C) but not during the pause transition (dashed circles in Fig. 6C).
Next, parameters of MT dynamic instability were determined by video-microscopy (Fig. 7). These parameters were measured in cells expressing EGFP–Tau (black bars) and compared with cells expressing EGFP–tubulin (white bars). We focused on duration of the pause (Fig. 7A) and rescue (Fig. 7B) transitions and on catastrophe (Fig. 7C) and rescue (Fig. 7D) frequencies. Compared with control cells, Tau increased by 50% and 40% the duration of pause and rescue transitions, respectively. We also noticed that Tau induced lower catastrophe frequency (−35%). All these data are consistent with the MT-stabilizing effect of Tau, as previously reported (Bunker et al., 2004; Bunker et al., 2006).
Then, we examined how PTX, TSA and the expression of non-acetylatable tubulin affected MT dynamics in cells with EGFP–Tau. In the presence of PTX, we observed roughly similar variation regardless of the presence or absence of EGFP–Tau. However, catastrophe and rescue frequencies remained relatively high (∼1.5–2 events/µm; Fig. 7C,D). Moreover, in cells treated with TSA, parameters of MT dynamics were similar between control and Tau-expressing cells. All durations of pause and rescue phases remained generally high in the presence of TSA (Fig. 7A,B). These results indicated that changes of MT dynamics are mainly due to TSA and not to the binding of Tau to MTs. Therefore, following PTX and TSA treatment, our data are in agreement with expected results for a drug that induces MT stabilization (Schiff et al., 1979; Schiff and Horwitz, 1980). This also indicates that the MT-stabilizing role of Tau is not in addition to that of PTX and acetylation. In contrast, in cells containing non-acetylatable α-tubulin, Tau caused a 15% longer rescue compared to that in control cells (Fig. 7B), and mainly increased by two rescue frequencies (Fig. 7D). In cells where we had overexpressed both non-acetylatable tubulin and EGFP–Tau, we revealed again the MT-stabilizing effects of Tau.
In living cells treated with PTX or TSA, MT dynamics were not affected by the presence of EGFP–Tau. As a consequence, there was no (stable) hotspot at the plus-end of MTs, both in rescue and pause transitions. In cells with non-acetylatable tubulin, we observed an increase of the MT dynamic instability that EGFP–Tau can counterbalance. This is consistent with the high FRET hotspot occurrence during rescue transition and numerous FRET hotspots along the length of MTs. To conclude, these data proved evidence that the control of MT dynamics by Tau is related to a domain of higher stability as shown by the presence of FRET hotspots.
Tau, and its active role in the assembly of tubulin, was identified in the 1970s (Weingarten et al., 1975; Cleveland et al., 1977), but details of Tau–MT interactions still remain unclear. It is well-documented that the binding of Tau to MTs is mediated through the repeat domains (Lee et al., 1988; Himmler et al., 1989), in combination with the adjacent proline-rich flanking regions (Kanai et al., 1992; Gustke et al., 1994). In the current work, we demonstrated by FRET imaging in diverse cell lines a discrete distribution of Tau–MT interactions with hotspots that represents one Tau every ∼60 tubulin dimers, and contrasting with their co-localization all along the length of MTs. This discontinuous distribution of the Tau–microtubule interaction could explain the efficiency of Tau in suppressing MT dynamics at a very low molar ratios (Tau:tubulin ratio of 1:175) (Panda et al., 1995). Moreover, we characterized two classes of hotspots based on their FRET intensities. We recently demonstrated using isothermal titration calorimetry that Tau binds to tubulin at a high affinity binding site with a stoichiometry of 0.2 and at a low affinity binding site with a stoichiometry of 0.8 (Tsvetkov et al., 2012). These two classes of binding sites might be associated with the high and low FRET hotspots found in the present study.
To further investigate Tau–MT interactions in living cells, we used the MT-targeting agent PTX and modulated the level of tubulin acetylation. It is well documented that the binding site of PTX is inside the pocket between the M-loop, the S9–S10 loop, and H6 and H7 helices and that the acetylatable K40 is near the N-loop in α-tubulin. Both are located in the luminal side of MTs, between adjacent protofilaments (Nogales et al., 1995; Downing and Nogales, 1998; Li et al., 2002). Electron microscopy and molecular modeling studies demonstrate that the M-loop and H1′-S2 and H2-S3 loops are major contacts in MTs for lateral interactions between protofilament (Nogales et al., 1999; Li et al., 2002). Therefore, any ligand that binds near the loop domains may have substantial effects on local interactions between neighboring protofilaments. In the literature, several models for Tau binding to tubulin have been proposed. Kar et al. proposed that Tau would bridge several parallel protofilaments with Tau repeat domains binding on consecutive protofilaments (Kar et al., 2003). Other studies show that Tau stabilizes MT by binding along individual protofilaments, bridging the tubulin interfaces rather that adjacent protofilaments (Chau et al., 1998; Al-Bassam et al., 2002). Santarella et al. observed by cryo-electron microscopy that GFP-fused Tau was randomly distributed along the outer surface of the MTs (Santarella et al., 2004). They concluded that Tau binds along and across protofilaments. Our FRET data supports these models of Tau binding to MTs, for example, that Tau can bind along or across the MTs, even stabilized by PTX (Al-Bassam et al., 2002; Kar et al., 2003; Santarella et al., 2004) or by acetylated K40 MTs. However, we showed that the binding of PTX and the TSA-induced acetylation of tubulin drastically decreased the quantities of high and low FRET hotspots, whereas accumulation of non-acetylatable α-tubulin in MTs increased mainly the amount of low FRET hotspots. Therefore, our results indicate that MT modifications have substantial effects on the accessibility and/or the three-dimensional structure of the binding loci of Tau with the MT surface.
We investigated binding interactions of Tau with MTs by comparing the fit of FRAP recovery curves of EGFP–Tau with one- or two-exponential equations. Our analytical approach highlighted that Tau mobility follows two models in cells: (1) a free diffusion in the absence of MTs; and (2) a diffusion-binding model in the presence of MTs. However, we cannot assign one-exponential to exclusive diffusion and two-exponentials to diffusion and binding of Tau. Indeed, the binding of PTX and the acetylation of the K40 residue in α-tubulin led also to a diffusion-binding model of Tau with two inseparable phases (one exponential). Moreover, non-acetylatable α-tubulin promoted a diffusion-binding model with two separate phases fitting with the sum of two exponentials. Our results are consistent with those of Ross et al. who explored the importance of a potential binding site between Tau and PTX on MTs using FRAP on in vitro polymerized MTs (Ross et al., 2004). Without PTX, their FRAP recovery curves were best fitted with a bi-exponential equation, in contrast to with PTX, when a simple exponential curve was obtained. In our FRAP experiment, we also fitted our FRAP recovery curves to the sum of two exponentials in the presence of MTs giving short and long time constants (1.8 seconds and 18.6 seconds, respectively). Our time constants are consistent with previous data (Samsonov et al., 2004; Konzack et al., 2007). Moreover, Hinrichs et al., recently showed, using an in vitro system and TIRF microscopy, that single Tau molecules can move bi-directionally along the MT lattice (Hinrichs et al., 2012). Our cellular FRAP data are in agreement with their results and indicate a diffusion–binding model of Tau along MTs. In addition, with modifications of the MT structure by PTX and the acetylation level of tubulin, our findings could also indicate different accessibility of Tau to its binding site on the MTs.
Furthermore, we identified a functional aspect of this specific Tau–MT interaction. Overall, the dramatic changes in MT organization throughout the cell cycle reflect a high degree of spatial and temporal regulation of MT dynamic instability. In this process, post-translational modifications of tubulin such as acetylation (Tran et al., 2007; Zilberman et al., 2009) and/or MT-interacting molecules such as Tau (Drubin and Kirschner, 1986; Lee and Rook, 1992) or PTX (Yvon et al., 1999; Gonçalves et al., 2001) tightly regulate MT assembly in cells. From our real-time FRET assay, we succeeded in demonstrating the emergence of a FRET hotspot at the plus-end of MTs during rescue and pause phases. Regarding the parameters of MT dynamics, we noticed that Tau induced longer periods of pauses and rescues. Moreover, the binding of PTX and the acetylation of α-tubulin did not induce FRET hotspots during either the rescue or the pause phases. This could be correlated with no difference of MT dynamics between the control and Tau-expressing cells. This also shows that the MT-stabilizing function of PTX and acetylated K40 largely mask that of Tau. In the presence of non-acetylatable α-tubulin, we again found the appearance of a FRET hotspot during the rescue phase, which could be related to the long period of rescue phases and high rescue frequencies. Our findings indicate that the discrete Tau–MT interaction corresponds to a point of high MT stability necessary for the recovery of MT assembly.
Mechanistically, Tau could induce conformational changes of MTs during rescue events that subsequently help the recruitment of plus-end tracking proteins (+TIPS) responsible for closing the MT (Bratman and Chang, 2007; Brouhard et al., 2008; Vitre et al., 2008; Al-Bassam et al., 2010). Conversely, the Tau-induced changes of MT conformation may also be crucial for catastrophe-promoting proteins such as stathmin (Gavet et al., 1998; Manna et al., 2009). Indeed, MTs are ordinarily quite rigid structures (Mizushima-Sugano et al., 1983; Cross and Williams, 1991) and the binding of Tau to MTs could directly enhance inter-protofilament contacts and their stabilizing structures. This would modify the mechanical properties of MT, facilitating the interactions of +TIPS, which promote the MT growth, but disturbing the binding of stathmin, which supports MT shrinkage.
In conclusion, using FRET and FRAP measurements, we have identified for the first time in living cells a discrete Tau–MT interaction. Our data also provide significant support for a mechanism in which the binding of Tau induces necessary conformational changes of MTs, especially for the recovery of MT assembly. This study presents the molecular basis for an understanding of the stabilizing effects of Tau on the MT cytoskeleton, opening the way for the study of the effects of post-translational Tau modifications and/or mutations in these interactions between Tau and tubulin.
Materials and Methods
A gene coding for the longest human isoform of Tau (hTau40) was subcloned into pEGFP-C1 vector (ClonTech, CA, USA) between the XhoI site (forward primer: 5′-CGTCGTCTCGAGGCTGAGCCCCGCCAGGAGTTC-3′) and the KpnI site (reverse primer: 5′-GTAGGAGGTACCTCACAAACCCTGCTTGGC-3′). PCR was performed using high-fidelity Taq DNA polymerase (Biolabs Inc., MA, USA). The cDNA clone was confirmed by DNA sequence analysis. The pEGFP-tubulin plasmid was purchased from Invitrogen (Cergy-Pontoise, France). The pmCherry-tubulin and pmCherry-tubulin K40A plasmids coding, respectively, for wild-type α1B isotype and non-acetylatable α-tubulin mutated at lysine 40 to alanine were a gift from Dr F. Saudou (Dompierre et al., 2007).
Cell culture and transfection protocol
All cell lines were purchased from ATCC, MD, USA. Cells from human non-small lung carcinoma (clone A549; CCL2), colorectal adenocarcinoma (clone Caco-2; HTB-37) and neuroblastoma (clone SK-N-SH; HTB-17) were routinely grown at 37°C in a humidified atmosphere of 5% CO2. Cells were maintained by regular passage in a complete medium composed of RPMI 1640 (Lonza, Belgium) for A549 and SK-N-SH cells and with DMEM for the Caco-2 line, supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine and 100 U /ml penicillin and 50 U/ml streptomycin (Invitrogen). Cells were free from mycoplasma as determined by mycoalert tests (Lonza). Two days prior to experiments, cells were seeded at 105 cells per well in 1 ml complete medium in a four-well Lab-Tek chamber coverglass (Nunc, France). The lipofection of cells was carried out with Lipofectamine 2000 according to Invitrogen's instructions and 0.4 µg plasmid was used, in total, for every experiment. Moreover, a set of A549 cells expressing Lipofectamine Tau and/or α-tubulin proteins was incubated with 2 nM PTX (Mr = 853.91 in DMSO; Sigma, France) or 40 ng/µl TSA (Mr = 302.37 in DMSO; Sigma) for 4 hours at 37°C before FRET and FRAP experiments. For the FRAP assay, nocodazole (Mr = 301.32; Becton Dickinson, UK) was used to disassemble MTs with 1 µg/ml for 30 minutes at 37°C.
The following antibodies were used: mouse monoclonal anti-α-tubulin (clone DM1A; 1∶1000 from 1 mg/ml; Sigma), anti-acetylated α-tubulin (clone 6-11B-1; 1∶1000 from 1 mg/ml; Sigma) and anti-Tau (clone T1029; 1∶500 from 1 mg/ml; USbio–Euromedex, France). Immunoblotting of acetylated and total α-tubulin from A549 cells was performed as described previously (Galmarini et al., 2003). Cell pellets were lysed by resuspension in 100 ml of low-salt buffer (20 mM Tris-HCl pH 6.8; 1 mM MgCl2; 2 mM EGTA) and lysates centrifuged at 14,000 rpm for 10 minutes at room temperature. The supernatant containing soluble α-tubulin was transferred to a separate centrifuge tube and kept on ice. The pellet, which corresponds to the fraction of polymerized α-tubulin before the cell lysis was resuspended in Ling's buffer (10 mM Tris-HCl pH 7.5; 1.5 mM MgCl2; 10 mM KCl) to the same volume as the supernatant. Equivalent quantities of each fraction were subjected to western blot analyses: one to detect acetylated α-tubulin and a second to quantify total α-tubulin. Furthermore, immunoblots of Tau in A549, Caco-2 and SK-N-SH lines were also performed. Cells were lysed in RIPA buffer (50 mM Tris, pH 8; 150 mM NaCl; 1% Triton X-100; 0.1% SDS; 1 mM EDTA; 1 mM TCEP) and lysates were centrifuged at 14,000 rpm for 20 minutes at 4°C. Equivalent quantities of proteins from the supernatant fraction were subjected to western blot analysis to detect Tau, and quantified by densitometry with ImageJ.
Instrumentation and image acquisition
Cells were placed in appropriate medium supplemented with 1% FBS and 10 mM Hepes to reduce pH variations, and maintained at 37°C. To analyze MT dynamics, time-lapse series of transfected cells were acquired with a fluorescence microscope (Leica Microsystems, Mannheim, Germany) equipped with a 100× objective. Forty-one frames of cell were acquired every 3 seconds using a CCD camera (CoolsnapFX, Princeton Instruments, Trenton, USA) driven by Metamorph software (Universal Imaging Co., Downingtown, USA). For FRET experiments, the cytoskeleton was imaged using a Leica SP5 confocal laser scanning microscope (CLSM) with a Leica inverted microscope, equipped with a Plan-Apochromat 63× oil immersion objective (NA = 1.4). Each image was recorded with the spectral mode of the CLSM selecting specific domains of the emission spectrum. The donor EGFP was excited at 488 nm with an argon laser and its fluorescence emission was collected between 496 nm and 535 nm (corresponding to the donor filter) and between 580 nm and 650 nm (corresponding to the FRET filter). The acceptor mCherry was excited at 543 nm with an helium–neon laser and its fluorescence emission was collected between 580 nm and 650 nm. The donor and the acceptor fluorophores were excited sequentially. The public-domain ImageJ software was used for image analysis (Rasband, 1997). FRAP experiments were performed on a CLSM Fluoview FV 1000 (Olympus, Japan) with an IX81 inverted microscope, equipped with Olympus UPLSAPO 60× oil immersion objective (NA = 1.35). The microscope was controlled using Olympus Fluoview 2.0c software. Excitation (488 nm) was provided by a multi-line argon laser and emission collected using a 505–550 nm wavelength band pass filter. As is common in FRAP, the experiment was divided in three sequences: (i) a pre-bleaching period during which 10 frames were recorded to define the initial level of fluorescence; (ii) a photobleaching step, which was carried out on a 2 µm radius circular area of the cytoplasm using the 488 nm wavelength laser at 30% power with two iterations of 2 µseconds/pixel according to the circular scanning (Tornado) mode; (iii) a postbleaching period during which fluorescence recovery was monitored every 0.43 seconds for 45 seconds.
The NFRET intensities and the distribution of NFRET pixels inside cells (n = 90, three independent experiments) were determined as follows: cells expressing EGFP–Tau, mCherry–tubulin and NFRET were concomitantly outlined in images with ImageJ software. The total NFRET intensities of cells were reported per unit area (micrometers square) and the ratio of donor to acceptor emission was simultaneously determined. The percentage of pixels belonging to low FRET (between 0.1 and <0.5) and high FRET (≧0.5) classes was also calculated.
FRAP data manipulation and fitting
Analysis of the microtubule dynamic instability
This analysis was done as previously described (Honore et al., 2003). Series of 16-bit images were transferred to ImageJ software to determine the position of the plus-end of individual MTs during time. Changes in length of ≧0.5 µm were considered as growth or shortening events. Changes in length of <0.5 µm were considered as phases of attenuated dynamics (catastrophe rescues or pauses). Means and standard error were calculated on 90 MTs from three independent experiments. Mean durations of pauses and rescues of catastrophe were determined using ImageJ. The distance-based catastrophe frequency was calculated by dividing the number of transitions from growth or pauses to shortening by the total growth length for each monitored MT. The distance-based rescue frequency was calculated similarly, dividing the total number of transitions from shortening to rescue by the total shortening length.
We are very grateful to Dr Yannick Gachet and Dr Sylvie Tournier (LBCMCP, Toulouse, France) for helpful discussions. We thank Dr Saudou (CNRS UMR146, Orsay, France) for plasmid coding the non-acetylatable α-tubulin K40A fused to mCherry. V.P. thanks F. H. Boulay for his help.
G.B, P.H, R.N. and T.D.B. carried out the experimental work; G.B., P.H., F.D., P.B. and V.P. participated in the data analysis and interpretation; G.B., F.D., P.B., J.N.S. and V.P. participated in the elaboration of the concept and design of the research; G.B., R.N., P.B., F.D., J.N.S. and V.P. participated in the writing of the article.
We acknowledge the financial support of INCa (Institut National du Cancer); Inserm (Institut National de la Santé et de la Recherche Médicale); AMU (Aix-Marseille Université) and DGOS [SIRIC label (Site de Recherche Intégré sur le Cancer)]. R.N. received fellowships from Région Provence Alps-Côte d’Azur and from ONET Technologies.