Report

Stathmin-Tubulin Interaction Gradients in Motile and Mitotic Cells

See allHide authors and affiliations

Science  19 Mar 2004:
Vol. 303, Issue 5665, pp. 1862-1866
DOI: 10.1126/science.1094108

Abstract

The spatial organization of the microtubule cytoskeleton is thought to be directed by steady-state activity gradients of diffusible regulatory molecules. We visualized such intracellular gradients by monitoring the interaction between tubulin and a regulator of microtubule dynamics, stathmin, using a fluorescence resonance energy transfer (FRET) biosensor. These gradients were observed both during interphase in motile membrane protrusions and during mitosis around chromosomes, which suggests that a similar mechanism may contribute to the creation of polarized microtubule structures. These interaction patterns are likely to reflect phosphorylation of stathmin in these areas.

Microfilaments and microtubules are organized in a variety of patterns at specific sites in the leading edge of the cell during cell motility and mitotic spindle assembly. The reorganization of cytoskeleton components depends on the global and local activity of a number of proteins that affect the nucleation, dynamics, and arrangement of the filament systems. Stathmin–oncoprotein 18 (OP18) belongs to a class of proteins that negatively regulate microtubule dynamics (1). It is a 17-kD cytoplasmic protein with a complex phosphorylation pattern. In response to extracellular stimuli (2) or during mitosis (3), stathmin becomes phosphorylated on up to four residues. Stathmin binds two tubulin heterodimers per molecule (4, 5) and forms a trimeric complex (T2S-complex). Stathmin's inhibitory effect on microtubule growth is believed to derive from its ability to sequester tubulin, which decreases the concentration of free heterodimers available for polymerization (6). An alternative mechanism proposes a catastrophe-inducing activity of stathmin at microtubule tips (7). Phosphorylation inactivates the inhibitory effect of stathmin, probably by reducing its interaction with tubulin (4, 8, 9).

During mitosis, stathmin becomes highly phosphorylated because of one or several factors present on mitotic chromatin (3). One factor has been identified as Polokinase1. Depletion of Polokinase1 inhibits chromatin-induced stathmin hyperphosphorylation and spindle assembly in mitotic Xenopus egg extracts (10).

During interphase, stathmin is phosphorylated via the Rac1-Pak1 pathway (11). Both proteins, Rac1 and Pak1, become activated at the leading edges of cells made motile by a wound in the monolayer or by the exposure to growth factors (12, 13). The Rac1-Pak1 pathway mediates “pioneering” of microtubules into the leading edge (14). Thus, stathmin could be locally inactivated around mitotic chromosomes and at the leading edge of migrating cells, which would specifically promote localized microtubule growth.

To observe the spatial regulation of stathmin in cells, we constructed a fluorescent stathmin fusion protein COPY (CFP-OP18-YFP) (15), with cyan fluorescent protein (CFP) and the yellow fluorescent protein (YFP) variant citrine (16) fused to the N and C terminus, respectively. This fluorescence resonance energy transfer (FRET)–based sensor can report tubulin binding to stathmin, because the free stathmin molecule in solution is flexible, whereas the stathmin-tubulin complex is stiff and longitudinally extended (17). As a consequence, the two fluorophores should move apart from each other, leading to a decrease in FRET detected by a decrease in the YFP/CFP emission ratio (fig. S1). The emission spectra of stathmin fused to CFP and YFP (COPY-wt) (Fig. 1A) (15) confirmed this proposed behavior and indicated high FRET of the sensor in the tubulin-free state and apparently no FRET in the tubulin-bound state. Furthermore, COPY exhibited the binding properties reported for unmodified stathmin (4, 8, 9, 15), namely, the typical 2:1 binding ratio (Fig. 1B) and a decrease in tubulin affinity on phosphorylation (Fig. 1, C and D). The COPY-wt–tubulin binding isotherms fitted well with a model of two independent binding sites with different affinities [Kd (1) = 1.94 ± 0.63 μM, Kd (2) = 0.020 ± 0.015 μM (15)]. No change in the YFP/CFP emission ratio was detected during phosphorylation of COPY-wt in the absence of tubulin. For bacterial expression, we subcloned Xenopus stathmin cDNAs encoding wild-type (wt) protein and stathmins with triple mutations in the phosphorylation sites from Ser-to-Ala (aaa) and Ser-to-Glu (eee). Phosphorylated COPY-wt, as well as COPY-eee (8, 18, 19), exhibited a lower affinity for tubulin with two identical binding sites [Kd of P-wt = 3.78 ± 0.16 μM, Kd of eee = 6.25 ± 0.27 μM (15)]. The COPY-aaa mutant also exhibited identical binding sites with slightly decreased average affinity (Kd of aaa = 1.36 ± 0.06 μM) relative to COPY-wt.

Fig. 1.

(A) Emission spectrum of 2 μM COPY-wt before (red) and after addition of 30 μM tubulin (black). (B) Increasing amounts of tubulin were titrated into a 4 μM solution of COPY-wt (15). A loss of FRET was detected by a decrease of the YFP/CFP emission ratio (+, right axis). The changes in FRET are correlated with the fractional saturation (Y) defined as the ratio of occupied versus total binding sites on COPY (black circles, left axis). Error bars show the SEM of four data sets. (C) We subjected time-course aliquots (0 min, 30 min, 1 hour, 2 hours, 4 hours) of the phosphorylation reaction (15) to SDS–polyacrylamide gel electrophoresis and Western blot analysis using a stathmin antibody directed against an N-terminal epitope (top) and phosphorylated Ser16 (middle). Ponceau staining of the Western blot membrane confirms equal protein loading (bottom). The fraction of nonphosphorylated protein was quantified densitometrically (small numbers, top). (D) Tubulin binding of 2 μM pseudophosphorylated COPY-eee, phosphorylation site-deficient COPY-aaa, COPY-wt, and phosphorylated COPY-wt (P-COPY-wt) fitted with a model for two binding sites (15). Error bars show the SEM of at least two data sets. (E) Ratio images of COPY-wt (6 μM) added to mitotic Xenopus egg extract subjected to okadaic acid treatment (4 μM, bottom) versus untreated control extract (top). Color bar corresponds to the ratio values in the images.

To test phosphorylation-dependent tubulin release of the sensor under more physiological conditions, we compared ratio images of a mitotic Xenopus egg extract (15) supplemented with 6 μM COPY-wt after incubation with the phosphatase inhibitor okadaic acid (OA, 4 μM) and an untreated control extract (Fig. 1E). We found a ratio higher by 10.5 ± 0.9% (±SEM, n = 2) for the OA-treated sample than for the untreated control extract. Thus, inhibition of phosphatases induced a tubulin release from COPY that could be detected by our probe, reflecting the hyperphosphorylated state of stathmin in OA-treated extracts (3).

We transfected COPY into Xenopus A6 epithelial cells (15). As did endogenous stathmin, COPY exhibited a cytosolic localization. The presence of COPY-wt overexpressed about five times (15) did not affect the microtubule cytoskeleton seriously, although the tubulin network seemed slightly less dense (Fig. 2A, tubulin staining). The YFP and the CFP fluorescence images for each cell were acquired, and their ratios were calculated (15). Low COPY-tubulin interaction was most frequently detected within lamella, typical of migrating cells (Fig. 2, A and B). From those areas, the interaction increased toward the cell center.

Fig. 2.

Phosphorylation-dependent patterns of stathmintubulin interaction in Xenopus A6 cells. (A) CFP and YFP emission (left) and YFP/CFP ratio image of a fixed Xenopus A6 cell transfected with COPY-wt and stained with a tubulin-specific antibody (middle) acquired by confocal microscopy (15). The color bar corresponds to the ratio values in the YFP/CFP image. Scale bars, 20 μm. (B) CFP and YFP emission and YFP/CFP ratio images of live Xenopus A6 cells transfected with COPY-aaa (top) and COPY-wt (bottom) (15). Scale bars, 20 μm. (C) Statistical evaluation of the SIH of COPY-tubulin interaction in populations of live Xenopus A6 cells transfected with COPY-wt and COPY-aaa. (Left), Schematic representation of the SIH assay (15). Average SIH for cells transfected with COPY-wt and COPY-aaa (right). Error bars show the SEM of at least 17 cells scored per population.

A decrease of COPY-tubulin interaction could be attributed either to COPY phosphorylation or to a locally reduced tubulin or COPY concentration. To distinguish between these phenomena, we compared the spatial patterns of stathmin-tubulin interaction in cells transfected with COPY-wt with those of cells expressing mutant COPY-aaa that lacked the phosphorylation sites. This mutant should be responsive only to changes in local tubulin or COPY concentration but not to phosphorylation. We applied a method of data evaluation that produced a single average parameter, which reflected the degree of spatial inhomogeneity (SIH) of the interaction of COPY-wt or COPY-aaa with tubulin in a cell (Fig. 2C) (15).

Phosphorylation contributed to the SIH observed in cells transfected with COPY-wt, because it was significantly higher (∼2.8×, P < 0.003; t test) than that measured for the COPY-aaa population (Fig. 2C). However, we also observed spatial inhomogeneities in cells expressing COPY-aaa. These patterns could not have been phosphorylation-dependent. Local differences in tubulin or COPY concentration (20) could possibly affect the readout. In addition, other unknown factors may regulate stathmin-tubulin interaction independent of phosphorylation.

The local differences in stathmin-tubulin interaction observed in Xenopus A6 cells grown under normal serum conditions appeared to be related to the formation of leading edges in motile cells. 12-O-Tetradecanoylphorbol-13-acetate (TPA), a potent activator of protein kinase C, induces stathmin phosphorylation (21, 22) and stimulates membrane ruffling and cell motility. TPA-induced membrane ruffling is mediated by a Rac1-dependent pathway (23, 24), as is growth factor–triggered stathmin phosphorylation (11). We treated serum-starved Xenopus A6 cells expressing COPY-wt or COPY-aaa with TPA, fixed the cells, and analyzed the SIH of COPY-tubulin interaction (Fig. 3, A and B) (15). The serum-starved cells revealed about the same low SIH level, when transfected with either COPY-wt or COPY-aaa. In the presence of TPA, we observed a ∼2.1-fold increase in SIH (P < 0.0003, t test) within COPY-wt cells, but no increase within the COPY-aaa population. The creation of COPY-tubulin interaction gradients by TPA was accompanied by the massive induction of leading edges and membrane ruffles (Fig. 3C). The dynamic behavior of the gradients in TPA-treated live cells was followed by time-lapse ratio imaging [Fig. 4, A and B; Movie S1 (15)]. Phosphorylation of stathmin fluctuates with time but is found preferentially in the lamellum. Retraction of the lamellum at the end of the recording coincided with the disappearance of the gradient.

Fig. 3.

TPA induces phosphorylation-dependent gradients of COPY-tubulin interaction. (A) Xenopus A6 cells transfected with COPY-aaa (left column) and COPY-wt (right column). Cells were serum-starved for ∼16 hours (ΔS), stimulated 30 min with 100 nM TPA, and fixed. Images show the CFP and the YFP emission of COPY and their ratio. Scale bars, 20 μm. (B) SIH assay of Xenopus A6 cells expressing COPY-aaa or COPY-wt. Error bars show the SEM of at least 11 cells scored per population. (C) Induction of leading edges by TPA stimulation. Bars represent the fraction of cells transfected with COPY-wt or COPY-aaa that show significant membrane ruffling at the leading edges. Error bars show the SEM of two experiments.

Fig. 4.

(A) Time-lapse recording of YFP/CFP ratio in A6 cells transfected with COPY-wt and stimulated with TPA (15). Scale bars, 20 μm. (B) SIH analysis of ratio versus time. (C) Acceptor photobleaching (15) of A6 cells transfected with COPY-wt (top, left), COPY-aaa (middle, left), and COPY-eee (bottom, left) and stimulated with TPA (100 nM, 30 min). Cells transfected with COPY-wt were additionally treated with okadaic acid (OA, 6 μM, 30 min, top, right). Scale bars, 20 μm. (D) Mean differences in FRET efficiency between the central and peripheral regions of the cells were quantified for each population of cells. Error bars show the SEM of at least 18 cells per population.

As an independent measurement of FRET efficiency, we performed acceptor photobleaching (15, 25) of TPA-treated cells, transfected with COPY-wt and its mutants (Fig. 4, C and D). The maximal differences of FRET efficiency between central and lamellum-like regions of the cells were found in cells transfected with COPY-wt. In contrast, COPY-eee revealed the most homogeneous distribution of tubulin interaction, in agreement with the fact that this mutant could not be phosphorylated and could not interact with tubulin very efficiently. The COPY-eee signal was thus less sensitive to differences in local tubulin concentration within a cell than COPY-aaa. Inhibition of protein phosphatases PP1 and PP2A by OA (15) significantly flattened the intracellular gradients of COPY-wt–tubulin interaction compared with untreated cells transfected with COPY-wt (P = 0.004, t test). Thus, phosphorylation and, in particular, the balance of kinase and phosphatase activity modulate the spatial distribution of stathmin-tubulin interaction within cells.

Stathmin has also been reported to be hyperphosphorylated in mitotic Xenopus egg extracts in the presence of chromatin (3). This suggested the occurrence of a gradient of phosphorylation-inactivated stathmin around mitotic chromosomes. We tested this hypothesis by imaging COPY in mitotic Xenopus XL177 cells (15). A COPY-tubulin interaction gradient was detected around mitotic chromosomes where tubulin binding increased with rising distance from the chromosomes (Fig. 5, A and B). The interaction patterns observed within cells transfected with COPY-aaa were flat, which indicated that the COPY-wt–tubulin interaction gradient was produced by stathmin phosphorylation around mitotic chromosomes.

Fig. 5.

Gradient of COPY-tubulin interaction around mitotic chromatin. (A) Xenopus XL177 cell transfected with COPY-wt (top) and COPY-aaa (bottom) (15) in mitosis. The inset (OL) in lower right edge of each ratio image shows the microtubule staining (15) superimposed on the ratio image. Scale bars, 10 μm. (B) Statistical gradient analysis (15). Error bars show the SEM of at least four cells scored per population. (C) Spindle lengths of mitotic cells untransfected (control), or transfected with COPY-wt and COPY-aaa. The distance between the two spindle poles was measured. Error bars show the SEM of at least six spindles scored per population. (D) Hypothetical scheme summarizing the factors necessary for the maintenance of a steady-state gradient of stathmin phosphorylation. Nu, nucleus.

Spindles of untransfected XL177 cells were slightly longer than of COPY-wt cells (Fig. 5C), probably because of incomplete inactivation of overexpressed sensor. Apart from this, COPY-wt spindles looked normal. By contrast, COPY-aaa spindles were ∼30% shorter, and their morphology often looked perturbed (Fig. 5A) (4, 10, 26, 27). Thus, local phosphorylation and inactivation of stathmin around mitotic chromosomes seems to be necessary for correct spindle assembly.

A striking consequence of the observed stathmin-tubulin interaction gradients can be envisaged: They seem to produce a cytoplasmic environment with graduated microtubule-stabilizing activity that could provide intracellular guidance for microtubules before they actually reach a potential capture site in the cell cortex or at the chromosomes. This would result in preferential microtubule growth along these gradients (28, 29). Indeed, a subset of pioneering microtubules in the leading edge (30) seems to be stabilized selectively compared with those in the center of the cell. In line with our study, recent reports (14, 31) propose Rac1-Pak1–mediated inactivation of stathmin to be at least in part responsible for this localized stabilization.

In mitotic cells, we showed that a stathmin phosphorylation gradient is necessary for correct spindle formation. If this gradient reflects a steady-state pattern created by a kinase-phosphatase system, its theoretical extent can be calculated (32): Using realistic literature values for PP2A concentration and activity, as well as the diffusion constant of stathmin [DCOPY ≈ 13 to 18 μm2/s, as approximated by fluorescence recovery after photobleaching (FRAP) experiments], we estimated (15) that it should extend about 4 to 8 μm. Our observed mitotic gradients are in the same order of magnitude. Thus, it is plausible that they derive from a steady-state balance of localized kinase and counteracting, nonlocalized phosphatase activity (Fig. 5D).

Whereas gradients of diffusible morphogens are known to be crucial for the supracellular self-organization of tissues and organisms, our findings add to a rising body of evidence that similar principles may operate at the intracellular level.

Supporting Online Material

www.sciencemag.org/cgi/content/full/303/5665/1862/DC1

Materials and Methods

SOM Text

Fig. S1

References

Movie S1

References and Notes

View Abstract

Navigate This Article