|Home | About | Journals | Submit | Contact Us | Français|
Regulated centrosome biogenesis is required for accurate cell division and for maintaining genome integrity1. Centrosomes consist of a centriole pair surrounded by a protein network known as pericentriolar material (PCM)1. PCM assembly is a tightly regulated, critical step that determines a centrosome’s size and capability2–4. Here, we report a role for tubulin in regulating PCM recruitment via the conserved centrosomal protein Sas-4. Tubulin directly binds to Sas-4; together they are components of cytoplasmic complexes of centrosomal proteins5,6. A Sas-4 mutant, which cannot bind tubulin, enhances centrosomal protein complex formation and has abnormally large centrosomes with excessive activity. These suggest that tubulin negatively regulates PCM recruitment. Whereas tubulin-GTP prevents Sas-4 from forming protein complexes, tubulin-GDP promotes it. Thus, tubulin’s regulation of PCM recruitment depends on its GTP/GDP-bound state. These results identify a role for tubulin in regulating PCM recruitment independent of its well-known role as a building block of microtubules7. Based on its guanine bound state, tubulin can act as a molecular switch in PCM recruitment.
Centrosome biogenesis is a multi-step process that begins with centriole formation followed by PCM recruitment to form a functional organelle4. PCM recruitment begins with the formation of cytoplasmic protein complexes and requires Sas-4/CPAP3,5,8,9. Recently, we reported that Sas-4, a protein known to have a role in centriole and PCM formation3,10,11, scaffolds centrosomal protein complexes (S-CAP complexes) which include Cnn, Asl, D-PLP, CP-190, and tubulin (αβ–tubulin dimer), and tethers the S-CAP complexes to centrosomes5. Sas-4 also exists in complexes with γ-tubulin9 and γ-tubulin ring proteins (S-γ-tubulin complexes), suggesting that Sas-4 may also be associated with the assembly intermediates of γ-tubulin ring complexes (Figs. 1). S-γ-tubulin complexes are recruited to developing centrosomes in a Sas-4-dependent manner (Fig. 2). Together, these suggest that Sas-4 regulates PCM recruitment via several protein complex types. Interestingly, it appears that tubulin exists in the multiple Sas-4 complex types (Fig. 1b–d). Since tubulin is significantly more abundant than other centrosomal proteins, Sas-4 likely interacts with free tubulin prior to formation of the centrosomal protein complexes. If the Sas-4-tubulin interaction is a first step, then tubulin may regulate centrosomal complex formation and PCM recruitment.
We began testing this hypothesis, by comparing the abilities of Sas-4, which can bind tubulin, to a mutated version of Sas-4, which cannot bind tubulin. For this, we used an N-terminal fragment of Sas-4 (Sas-4-N) that includes Sas-4’s tubulin binding site12,13; we also used a mutated version of Sas-4-N (Sas-4-NΔT), which lacks the two amino acids essential for tubulin binding12,13. As expected, Sas-4-NΔT failed to pull-down tubulin from embryonic high-speed lysates (HSLs). Surprisingly, Sas-4-NΔT pulled-down significantly more Cnn, Asl, D-PLP, γ-tubulin, and Grip128 than was pulled-down by Sas-4-N (Fig. 3a). We then tested the effects of tubulin on the ability of Sas-4-N or Sas-4-NΔT to bind centrosomal proteins. Increasing amounts of tubulin progressively inhibited Sas-4-N’s binding to centrosomal proteins but did not inhibit Sas-4-NΔT’s binding (Fig. 3b–c). These suggest that tubulin can negatively regulate the formation of centrosomal protein complexes.
To test this hypothesis in vivo, we generated transgenic Drosophila that express full-length Sas-4ΔT in the sas-4s2214 null (sas-4ΔT::sas-4s2214). Sas-4ΔT failed to fully rescue the sas-4s2214 phenotype of uncoordination14: sas-4ΔT::sas-4s2214 flies stood but could barely walk (Supplementary Movie). Furthermore, sas-4ΔT::sas-4s2214 sperm axonemes were abnormal (Supplementary Fig. S1a). These phenotypes suggest defects in centrosome biogenesis14,15 and that tubulin binding to Sas-4 is essential for centrosome function and proper cilia formation. However, sas-4ΔT::sas-4s2214 flies had correct numbers of centrosomes, indicating that Sas-4ΔT rescued this aspect of the sas-4s2214 phenotype14 (Supplementary Fig. S1b). Thus, the Sas-4-tubulin interaction is not essential for maintaining centrosome number. Yet, sas-4ΔT::sas-4s2214 spermatocyte centrioles were slightly shorter (Supplementary Fig. S1c). This is consistent with reports that Sas-4, and in particular, the tubulin-Sas-4 interaction is required for centriole elongation13,16–18.
In addition to its well-known role in centriole formation, Sas-4 plays an important role in PCM formation and in regulating centrosome size3,10,11. Achieving proper centrosome size and capability requires Sas-4 and Cnn2,3. Since Sas-4 scaffolds centrosomal complexes that include Cnn, regulation of Sas-4 complex formation may indirectly control centrosome size. Indeed, although Cnn is normally detected only in mitotic or meiotic centrosomes, interphase spermatogonium and spermatocytes centrosomes of sas-4ΔT::sas-4s2214 contained Cnn (Fig. 3d–e)19. Moreover, mitotic and meiotic centrosomes of sas-4ΔT::sas-4s2214 contained twice the Cnn immunolabelling as control centrosomes (Fig. 3f and Supplementary Fig. S2a). Thus, tubulin can negatively regulate the timing, distribution and quantity of protein recruitment to centrosomes, via Sas-4.
In Drosophila, interphase centrosomes do not nucleate microtubules20. Since sas-4ΔT::sas-4s2214 centrosomes prematurely contain Cnn and Cnn’s human ortholog stimulates microtubule nucleation21, we tested whether the sas-4ΔT mutation affects microtubule nucleation. Interphase sas-4ΔT::sas-4s2214 centrosomes had premature microtubule nucleation (Fig. 3g) and their meiotic centrosomes had massive microtubule asters, which could fill a significant fraction of a cell (Supplementary Fig. S2b–c). Similarly, in cultured cells, Sas-4ΔT produced massive asters (Supplementary Fig. S2d–e). These results suggest that the tubulin present in wild-type Sas-4 complexes is not a building block of microtubule asters, but instead appears to be essential in the regulation of PCM recruitment.
To gain insight into how disruption of the Sas-4-tubulin interaction affects meiosis and mitosis, we analyzed spermatids and larval brain cells. We found that over 95% of sas-4ΔT::sas-4s2214 round spermatids exhibit normal morphology, suggesting that meiotic cell division can conclude normally (Supplementary Fig. S2f). In larval brain cells, unlike control cells, which recruit significant amounts of Cnn and form robust asters only during mitosis22, some sas-4ΔT::sas-4s2214 cells recruited Cnn and formed asters before entry into mitosis (Supplementary Fig. S3a–c). During mitosis, control larval brain cells have Cnn enrichment in only one centrosome22, 23, but in sas-4ΔT::sas-4s2214 cells, Cnn was distributed more evenly to both centrosomes (Supplementary Fig. S3d). Finally, spindle orientation relative to Bazooka’s crescent (a polarity establishment marker), were abnormal in sas-4ΔT::sas-4s2214, suggesting that these centrosomes have difficulty properly aligning their spindles (Supplementary Fig. S3e–f). Taken together, these results suggest that the interaction of tubulin with Sas-4 is essential for normal PCM recruitment and centrosome function in larval brain cells.
To better understand how tubulin operates in the regulation of PCM recruitment, we focused on the biochemical properties of the Sas-4-tubulin interaction. Tubulin is a guanine binding protein having GTPase activity, which hydrolyzes tubulin-GTP into tubulin-GDP7. Tubulin has a different conformation when present as tubulin-GTP versus tubulin-GDP and tubulin’s confirmation acts as a molecular switch that regulates microtubule dynamics7. Therefore, we speculated that tubulin’s confirmation might also regulate the formation of Sas-4 complexes. For this, we analyzed tubulin’s binding to Sas-4-N in the presence of GDP or GMPCPP (a non-hydrolyzable GTP analog)24. Sas-4-N, which includes Sas-4’s tubulin binding site, prevents microtubule polymerization when present in excess6,12,25. Tubulin-GMPCPP at 0.5μM (which is below the concentration necessary for microtubule polymerization26) had four-fold less binding to Sas-4-N than tubulin-GDP at the same concentration (Fig. 4a). Similar results were obtained by isothermal titration calorimetry experiments, indicating that tubulin-GDP has a higher affinity for Sas-4 than tubulin-GMPCPP has (Supplementary Fig. S4). However, since the affinity of tubulin to Sas-4 appears to be high (Fig. 3b–c and Supplementary Fig. S4) relative to the cytoplasmic concentration of free tubulin (~10 μM27), it is likely that cytoplasmic Sas-4 is bound to either tubulin-GDP or tubulin-GTP. Therefore, it is possible that conformation of this bound tubulin (depending on which guanine is present) regulates the formation of Sas-4-containing complexes.
To test this, we purified and analyzed Sas-4 complexes from HSLs exposed to GDP or GMPCPP. Although the quantity of Sas-4 present in the purified complexes was unaffected by GDP or GMPCPP exposure, the amounts of other centrosomal proteins in the Sas-4 complex were affected. More specifically, HSLs exposed to GDP had 6 to 12 fold-increases in the amounts of particular centrosomal proteins relative to HSLs exposed to GMPCPP (Fig. 4b–c). Therefore, when bound to tubulin-GDP, Sas-4 acts similar Sas-4ΔT in that it accumulatesexcess centrosomal proteins in its complexes. Perhaps, tubulin-GTP’s binding to Sas-4 sterically hinders Sas-4’s binding to other centrosomal proteins and tubulin-GDP reverses the steric hindrance, allowing Sas-4 to bind the centrosomal proteins. Together, it appears that Sas-4’s binding to tubulin-GDP (but not tubulin-GTP) favors formation of centrosomal protein complexes.
To confirm this, we first tested whether Sas-4 complexes preferentially contain GDP. We immuno-purified Sas-4 complexes from embryonic HSLs treated with [α32p]GTP and analyzed the complexes using thin-layer chromatography. Tubulin that was not bound to Sas-4 contained [α32p]GTP, whereas purified Sas-4 complexes instead contained mostly [α32p]GDP, which is the hydrolyzed product of [α32p]GTP (Fig. 4d). Sas-4N, but not Sas-4NΔT, was able to pull-down GDP (Fig. 4e). Accordingly, when in a Sas-4 complex, tubulin binds GDP.
Second, we tested the effects on the composition of Sas-4 complexes of treatments with Griseofulvin, a compound that changes tubulin’s conformation and induces hydrolysis of tubulin’s bound GTP into GDP28. Griseofulvin increased the quantity of centrosomal proteins in purified Sas-4 complexes (Fig. 4f–g); this is consistent with our data of HSLs exposed to GDP (Fig. 4b–c). Together, these suggest that tubulin’s conformation can regulate the formation of cytoplasmic Sas-4 complexes.
We then studied how tubulin modulates PCM recruitment. Typical GTP-binding proteins (G-proteins), i.e., heterotrimeric G-proteins and the small GTPases belonging to the Ras superfamily, act as molecular switches whose function depends on its GTP- or GDP-bound state. G-proteins have both low intrinsic GTPase and guanine exchange activities and require GTPase activating proteins (GAPs) and guanine exchange factors (GEFs) as catalysts29, 30. Accordingly, we tested whether tubulin can exhibit the characteristics of a typical G-protein during PCM recruitment by acting as a molecular switch.
It is known that free tubulin has low intrinsic GTPase activity and exists as tubulin-GTP7. Therefore, if a tubulin switch is involved in PCM recruitment, it is expected that a GAP exists which induces tubulin to hydrolyze its bound GTP into GDP. We tested whether Sas-4 functions as a tubulin GAP and found that Sas-4-N enhanced the intrinsic GTPase activity of tubulin, as measured by the release of inorganic phosphate (Fig. 5a). This suggests that Sas-4 can function as a tubulin GAP.
Tubulin is known to have high guanine exchange activity and readily exchanges its GDP with GTP31. On the other hand, although tubulin-GTP disfavors the formation of centrosomal protein complexes (Fig. 4b), Sas-4 complexes are quite stable regardless of whether they are exposed to GDP or GMPCPP (Supplementary Fig. S5a–c). Therefore, for a Sas-4 complex to remain stable, tubulin’s guanine exchange activity must remain low. To assay the effects of Sas-4 on tubulin’s guanine exchange activity, we added [α32p]GTP to tubulin bound to Sas-4 (Sas-4-N) or not bound to Sas-4 (Sas-4-NΔT). The amount of exchanged [α32p]GTP was then determined by scintillation counter and thin-layer chromatography (Fig. 5b; Supplementary Fig. S5d). Consistent with previous reports, tubulin had a high rate of guanine exchange in the absence of bound Sas-431. In contrast, GTP exchange was not observed in the presence of Sas-4. These suggest that Sas-4 inhibits tubulin’s guanine exchange activity and can stabilize the Sas-4-tubulin complex.
Eventually, Sas-4 complexes are recruited to centrosomes. So, we tested how centrosomes affect guanine exchange and the stability of Sas-4 complexes. In the presence of centrosomes, tubulin did not have an increase in guanine exchange when tubulin is unbound to Sas-4 (Fig. 5b; Supplementary Fig. S5d). This indicates that centrosomes cannot increase the intrinsic guanine exchange activity of tubulin. In contrast, in the presence of centrosomes, tubulin’s guanine exchange was significantly increased when tubulin is bound to Sas-4. Therefore, centrosomes appear to undo Sas-4’s inhibition of tubulin’s guanine exchange activity. Consistently, centrosomes also destabilize the Sas-4-tubulin interaction (Fig. 5c). In the absence of centrosomes or in the presence of centrosomes exposed to GDP, Sas-4-tubulin complexes remained stable; however, in the presence of centrosomes exposed to GMPCPP, Sas-4-tubulin complexes were destabilized and dissociated, potentially allowing Sas-4 to be released into the cytoplasm (Fig. 5c).
To further test this, we mixed isolated centrosomes, purified Sas-4 complexes, and either GMPCPP or GDP. The reaction mixture was subjected to velocity sedimentation, which pelleted centrosomes along with their bound proteins5. When exposed to GDP, Sas-4-complex proteins, Sas-4, and tubulin were in the pellet (Fig. 5d), indicating that the Sas-4 complexes were bound to centrosomes. However, when exposed to GMPCPP, the Sas-4-complex proteins were in the pellet, yet some Sas-4 and tubulin were released into the supernatant. This indicates that centrosomes have guanine exchange activity that releases Sas-4 and tubulin from Sas-4 complexes, whereas other complex proteins remain in the centrosome (Fig. 5d). This is consistent with the observation that Sas-4 traffics between centrosomes and cytoplasm8.
PCM recruitment is tightly coupled to the cell cycle32. Mathematical models33 and analyses of global cytoskeleton remodeling34 predict that microtubule breakdown releases tubulin-GDP causing tubulin-GDP’s concentration to increase when cells enter mitosis. Given our above observations that tubulin-GDP promotes complex formation, this increase in tubulin-GDP concentration may promote PCM recruitment. Currently, there are no tools to determine this directly. Therefore, we analyzed centrosomes of cells treated with taxol, a compound that stabilizes microtubules and reduces tubulin-GDP release into the cytoplasm35. As expected, taxol-treated mitotic centrosomes of Sas-4-GFP transfected cells had significantly less Cnn, whereas, mitotic centrosomes of Sas-4ΔT transfected cells were less sensitive to taxol (Fig. 5e). Although taxol may affect centrosomes via multiple mechanisms, our data suggests that cytoskeleton remodeling regulates recruitment of Sas-4 complexes to centrosomes. Furthermore, treating cells with Griseofulvin, which enhances Sas-4 complex formation, increased Cnn incorporation into the centrosome (Fig. 5f). Together, the taxol and Griseofulvin experiments show that modulating tubulin in cells affects PCM formation.
Our findings reveal a previously unknown function of tubulin. We show that tubulin can negatively control Sas-4 complex formation and, thereby regulate PCM recruitment. Tubulin is a molecular switch that can regulate the formation of Sas-4 complexes and the recruitment of centrosomal proteins to a developing centrosome. The data described above was used to formulate a model whereby tubulin coordinates normal PCM recruitment (Fig. 5g). In the cytoplasm, tubulin-GTP binds Sas-4, which prevents Sas-4 from forming complexes with centrosomal proteins (Fig. 5gi). When Sas-4 activates tubulin’s GTPase, hydrolysis of tubulin-GTP into tubulin-GDP takes place; tubulin-GDP can initiate Sas-4 complex formation (Fig. 5gii). Additionally, Sas-4 complex formation may be enhanced when the tubulin-GDP concentration in the cytoplasm is increased due to microtubule depolymerization (Fig. 5giii). Sas-4 binding to tubulin-GDP stabilizes the Sas-4-tubulin complex by blocking the exchange of GDP with GTP. Sas-4-tubulin-GDP then interacts with other centrosomal proteins to form one of the various types of Sas-4-containing complexes (Fig. 5giv). When a Sas-4 complex tethers to a centrosome, tubulin’s guanine exchange activity is induced by the centrosome, causing the release of tubulin and Sas-4 and allowing the recruitment of centrosomal proteins to the centrosome (Fig. 5gv).
Typical G-proteins act as molecular switches whose function depends on their GTP- or GDP-bound state. Here, during PCM recruitment, we show that tubulin acts as a molecular switch whose function depends on its GTP- or GDP-bound state. Therefore, in PCM recruitment, tubulin acts like a typical G-protein. By manipulating this switch, it may be possible to target cancerous cells, which are known to have abnormal centrosomes and PCM1.
We would like to thank Dr. J. Iwasa for scientific illustrations; Drs. T. Mitchision A. Johnson, I. Cheeseman and J. Malicki for their scientific discussions; Dr. T. Kaufman, Dr. J. Raff, and Dr. B. Raynaud-Messina, Dr. T. K. Tang for reagents; Drs. Eric, Hari, Rodriguez for technical help with biophysical experiments, Dr. E. Koundakjian scientific editing and discussions, and EM facility staff at HMS for help with EM analyses. This work was supported by a grant (R01GM098394) from National Institute of General Medical Sciences.
Author Contributions J.G. and T.A.R. conceived the project. J.G. performed most of the experiments described herein. T.A.R. supervised the project. Y.F.C. performed phase and electron microscopy analyses. A.H. performed biochemical complex analyses. M.L.B. generated constructs and took part in the biochemical purification of recombinant proteins. D.A.L. and N.M.R. advised and discussed larval brain analyses.
Author Information: The authors declare that they have no competing financial interests.