IDENTIFICATION OF S6K2 AS A CENTROSOME-LOCATED KINASE

Abstract

Ribosomal S6 kinase 2 (S6K2) acts downstream of the mammalian target of rapamycin (mTOR). Here we show that some S6K2 localize at the centrosome throughout the cell cycle. S6K2 is found in the pericentriolar area of the centrosome. S6K2 centrosomal localization is unaffected by serum withdrawal or treatment with rapamycin, wortmannin, U0126, or phorbol-12-myristate-13-acetate (PMA). Unlike S6K2, S6 kinase 1 (S6K1) does not localize at the centrosome, suggesting the two kinases may also have non-overlapping functions. Our data suggest that centrosomal S6K2 may have a role in the Phosphoinositide-3-kinase (PI3K)/Akt/mTOR signaling pathway that has also been detected in the centrosome.

INTRODUCTION

Rapamycin is an immunosuppressant used in organ transplantation and more recently in some cardiac and anti-cancer therapies [1]. Rapamycin blocks or delays cell proliferation of many different cell types [1]. Its mammalian cellular target, mTOR, is a regulator of nutrient-and growth factor-sensing mechanisms and controls many cellular processes such as translation, cell cycle progression, cell size regulation, transcription, and cytoskeleton regulation [1]. Several proteins are activated downstream of mTOR, two of which are S6K1 and S6K2. S6K1 and S6K2 both phosphorylate the 40S ribosomal subunit protein S6 [2,3], a process that was thought to increase translation of mRNAs with a 5' terminus oligopyrimidine tract (5'TOP mRNA). Many 5'TOP mRNAs encode the translational machinery, leading to an increase in cellular protein synthesis capacity in preparation for cell division. However, recent studies showed that cells from S6K1 and S6K2 double knockout mice have impaired S6 phosphorylation but maintain mTOR-dependent 5'TOP mRNA translation, putting into question the function of S6 phosphorylation by S6K1 and S6K2 [2]. S6K1, but not S6K2, regulates cell size; mice lacking S6K1 have smaller cells and this cannot be compensated by the presence of S6K2 [4]. The full biological functions of S6K2 are unknown at this time. Understanding how these signaling molecules contribute to mTOR function would yield better insights into the mechanism of cell growth and/or proliferation.

S6K2 was initially identified as a homolog of S6K1 [4–8]. Evidence points to some common functions shared by the two; activities of both are regulated by the same upstream activating pathways such as mTOR, PI3K, and MEK pathways, and both S6K1 and S6K2 phosphorylate S6 [2–8]. However, several lines of evidence suggest that the two kinases have differential regulation and may have non-overlapping cellular function(s). The non-catalytic domains of the two kinases are distinct, and mutational studies show that equivalent mutants in the two kinases do not always behave the same [3,9–11], and that the MEK pathway plays a more important role for regulation of S6K2 than that of S6K1 through the C terminus of S6K2 [9,10]. The phenotypes of S6K1-null and S6K2-null mice are different in that only S6K1 plays a role in cell size regulation, indicating differential cellular functions for the two [2]. S6K1 has at least one substrate, SKAR, that is not phosphorylated by S6K2, suggesting that the two kinases have distinct subsets of substrates [12]. The full spectrum of S6K2 substrates is yet to be identified.

There have been reports showing that S6K2 is a nuclear protein with nuclear localization signals [4,7] and that the kinase may shuttle to the cytoplasm upon PMA stimulation [13]. There have also been reports of S6K2 staining both cytoplasmic and nuclear compartments in human tissues [14–16]. Some of these studies have noted that S6K2 is seen in a punctate pattern, and in order to further extend this finding, and in order to also better elucidate possible cellular function of S6K2, we set out to assess whether S6K2 co-localizes to any known subcellular components.

In this report we show that a fraction of S6K2 is found in the centrosome in all cell cycle stages. S6K2 localization to the centrosome is not inhibited by serum-starvation or treatment with rapamycin, wortmannin, U0126, or PMA. Interestingly, unlike S6K2, S6K1 does not localize to the centrosome. Finally, we show that S6K2 is a pericentriolar rather than a core centrosomal protein. Our study opens a possibility that the mTOR signaling pathway may also play a role in cytoskeleton regulation and/or cell division processes.

MATERIALS AND METHODS

Cell culture and transfection

HeLa cells or RPE-1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum, penicillin (250 units/ml), streptomycin (250 μg/ml), and L-Glutamine (292 μg/ml) at 37°C with 5.5% CO₂. KE-37 cells were cultured in RPMI media supplemented in the same manner as above. When required, cells were treated with rapamycin (20 ng/ml), wortmannin (50 nM), U0126 (10 μM), or PMA (20 ng/ml) (all from Calbiochem) for 30 minutes. Mouse myoblasts from wildtype or S6K1/S6K2 double-knockout mice (kind gifts from Dr. M. Pende, INSERM, Paris) [17] were cultured in DMEM-F12 media containing 20% fetal calf serum, 2% Ultroser G, penicillin (250 units/ml), streptomycin (250 μg/ml), and L-Glutamine (292 μg/ml) at 37 C with 5.5% CO₂. RPE-1 cells were transfected with a GFP-centrin 1 construct (a kind gift from Dr. M. Bornens, Institut Curie, Paris) using FuGene (Roche) per manufacturer's instructions.

Immunofluorescence

HeLa cells grown on coverslips were either fixed in cold methanol or in 2% paraformaldehyde and permeabilized in cold methanol. Soluble proteins were extracted with CSK buffer with Triton X-100 as previously described [18] for figure 1A; the rest of the figures used no triton extraction. RPE-1 cells were fixed with cold methanol. Primary antibodies: 07-173 sheep anti-S6K2 antibody (Upstate Biotechnology); 167-4 rabbit polyclonal antiserum to S6K2 (a kind gift from Dr. J. Blenis, Harvard Medical School); sc230 anti-S6K1 antibody (Santa Cruz); anti-γtubulin mAb and CTR453, a monoclonal antibody against AKAP450 (both gifts from Dr. M. Bornens) [19]. Secondary antibodies: RRX-conjugated anti-rabbit IgG; RRX-conjugated anti-sheep IgG; FITC-conjugated anti-mouse IgG; Cy3-conjugated anti-mouse Ig (all from Jackson ImmunoResearch Lab); and Alexa 488-conjugated anti-rabbit Ig (Molecular Probes). DNA was stained with 0.1 μg/ml 4’,6-diamidino-2-phenylindole (DAPI; Sigma). Confocal and conventional epifluorescence microscopy was performed as previously described [18].

Figure 1.

S6K2 co-localizes to the centrosome in HeLa cells. (A) Exponentially growing HeLa cells were fixed, extracted with Triton X-100, and stained for S6K2 using 07-173. DNA was stained with DAPI. (B) HeLa cells were fixed with methanol and co-stained with anti-γ-tubulin antibody (green) and with 07-173 (i) or 167-4 (ii) anti-S6K2 antibodies (red). (C) M-phase HeLa cells were co-stained with anti-γ-tubulin (green) and with 07-173 (red). (D) RPE-1 cells were stained with CTR453 (red), or167-4 (green), and DNA was stained with DAPI. Confocal laser images of the same cell were taken and merged for B and C. Bar, 10 μm for A-C, 20 μm for D.

Immunoblot

Cell lysates were immunoblotted as previously described [3]. Cell lysates were adjusted to have equal protein concentrations prior to gel loading. Primary antibodies: 07-173; 167-4; MP CT anti-S6K2 mouse mAb (gift from Dr. M. Pende) [2]; C3 rabbit anti-S6K1 antibody (gift from Dr. J. Blenis); anti-α-tubulin antibody (Sigma); anti-γ-tubulin antibody (gift form Dr. M. Bornens) [19]; anti-phospho-Erk antibody; anti-phospho-Akt substrate antibody; anti-phospho-S6 antibody; and anti-Erk antibody (last four from Cell Signaling Technology)

Purification of centrosome

Centrosomes were purified from KE-37 cells and immunofluorescence on purified centrosomes was carried out as previously described [20].

RESULTS

We carried out immunofluorescence on HeLa cells to study subcellular localization of endogenous S6K2. We used antibodies to endogenous S6K2 rather than using ectopically-expressed fluorescent S6K2 in order to eliminate potential complications from over-expression or epitope interference. For our immunofluorescence studies we used two different anti-S6K2 antibodies, 07-173 sheep anti-S6K2 antibody and 167-4 rabbit anti-S6K2 antibody [5]. When asynchronously-growing HeLa cells were stained with 07-173, we observed a speckled pattern both in the nucleus and the cytoplasm with a dot of S6K2 accumulating within the cytoplasm, located near the nucleus (Fig. 1A). During mitosis the protein was found in the cell body and did not co-localize with the condensed chromosomes (Fig. 1A). Similar data was obtained with 167-4 (data not shown and Fig. 1B)

The dot-like structure observed in the cytoplasm (Fig. 1A) was reminiscent of the centrosome. To see if S6K2 co-localizes with the centrosome we performed a confocal microscopy analysis of HeLa cells co-stained with anti-S6K2 antibodies and with an antibody against γ-tubulin, a centrosome marker. Figure 1B shows that S6K2 and γ-tubulin indeed co-localized at the centrosome. This was observed whether we used 07-173 (i) or 167-4 (ii) anti-S6K2 antibody. During mitosis S6K2 also associated with the mitotic asters (Fig. 1C), indicating that S6K2 also associates with the centrosome components during mitosis. We next verified localization of S6K2 to the centrosome in another cell type. RPE-1 cells, a human retinal epithelial cell line, were stained using 167-4 anti-S6K2 antibody and CTR453, a centrosome-specific monoclonal antibody against AKAP450 protein [21]. Figure 1D shows that S6K2 is localized throughout the cell in RPE-1 cells, and that a fraction of S6K2 accumulated to the centrosome in RPE-1 cells as well. Equivalent results were seen when HEK293 cells were used (data not shown).

Next we verified centrosomal localization of S6K2 using biochemical methods. Centrosomes were purified via sucrose gradient and analyzed by immunoblotting with antibodies against S6K2. We observe that S6K2 is detected in purified centrosomes when immunoblotted with anti-S6K2 antibodies, whether we use 167-4 or MP CT antibody (Fig. 2A). Figure 2B shows that the purified centrosome was indeed enriched in γ-tubulin, a centrosome-specific marker, but did not contain non-centrosomal protein such as Erk. An aliquot of the purified centrosomes used in immunoblotting was also stained with CTR453 and anti-γ-tubulin antibodies in order to verify via immunofluorescence that centrosome purification was indeed successful (Fig. 2B bottom panel). We also asked whether centrosome localization is unique to S6K2 or can be generalized to both S6K1 and S6K2. When we immunoblotted purified centrosomes with an antibody against S6K1, we did not observe any S6K1, indicating that only S6K2, and not S6K1, co-localizes to the centrosome (Fig. 2C). The lack of S6K1 in the centrosome also indicates that our centrosome preparation was free of contaminating cytosolic or nuclear proteins, as S6K1 is shown to reside in both nucleus and cytosol [22]. Our data indicate that S6K2, but not S6K1, is co-purified with the centrosome.

Figure 2.

Further verification of S6K2 localization to the centrosome. (A) Purified centrosomes or total cell lysates were immunoblotted with 167-4 or MP CT anti-S6K2 antibodies. (B) Purified centrosomes or total cell lysates were immunoblotted with an antibody against γ-tubulin, a centrosome marker, or against Erk, which is not found in the centrosome. Purified centrosomes were also fixed onto a coverslip and stained with CTR453 (red) or anti-γtubulin antibody (green) (bottom panel). Bar, 20 μm. (C) Purified centrosomes or total cell lysates were immunoblotted for S6K1. (D) Myoblast cells isolated from wildtype (WT) or S6K1/2 double knockout (KO) mice were stained with CTR453 (red) and 167-4 or 07-173 anti-S6K2 antibodies (green). Bar, 20 μm. (E) Cell lysates from wildtype or knockout myoblasts were immunoblotted for S6K2 or α-tubulin (loading control).

In order to further verify that immunofluorescence of centrosomal S6K2 in figure 1 was due to S6K2-specific binding of the antibodies and not due to cross-reaction with another protein, we used cells from S6 kinase knockout mice in immunofluorescence studies. We cultured myoblasts from wildtype or S6K1/S6K2 double-knockout mice and carried out immunofluorescence using CTR453 and anti-S6K2 antibodies (Fig. 2D). In cultured myoblasts from wild type mice we again observed the dot-like staining of S6K2 at the centrosome, and this was seen with both anti-S6K2 antibodies. On the contrary, knockout cells showed no centrosomal localization of S6K2. We observed dimmed homogenous staining in knockout cells, which indicates that both antibodies have a low-level non-specific background staining. Immunoblotting cell lysates from the knockout and wildtype myoblasts showed that the knockout cells are indeed lacking S6K2 (Fig. 2E). Our data indicate that the dot-like centrosomal staining we observe with both 167-4 and 07-173 anti-S6K2 antibodies are specific to S6K2 since such staining disappears when cells do not express S6K2.

In order to characterize the conditions in which S6K2 localizes to the centrosome, we treated RPE-1 cells with various stimuli or inhibitors and assessed S6K2 centrosomal localization using immunofluorescence. First we withdrew serum from the growth media of RPE-1 cells for 16 hours and carried out immunofluorescence. Figure 3A shows that co-localization of S6K2 with the centrosome was still preserved whether the cells were serum-starved or cultured in serum-containing media. Cell lysates from the serum-starved vs. serum-containing culture conditions were immunoblotted with anti-phospho-Erk antibody to show that cells were indeed serum-starved. We next assessed whether known upstream regulators of S6K2 activity such as mTOR, PI3K, and MEK play a role in centrosomal localization of S6K2. We used three different inhibitors: rapamycin, a pharmacologic inhibitor of mTOR; wortmannin, an inhibitor of the PI3K pathway; and U0126, an inhibitor of the Map kinase kinase (MEK) pathway. When RPE-1cells were treated with rapamycin (Fig. 3B), wortmannin (Fig. 3C), or U0126 (Fig. 3D), centrosomal localization of S6K2 was still observed. Immunoblotting of cell lysates from each condition showed that inhibitors functioned properly (Fig. 3B-D right panels): rapamycin inhibited phosphorylation of S6, a downstream event of mTOR/S6 kinases; wortmannin reduced the activity of Akt, a downstream target of PI3K, as shown by reduction of phosphorylation of substrates of Akt; and U0126 inhibited phosphorylation of Erk, a target of MEK. We conclude that inhibition of mTOR, PI3K, and MEK pathways had no effect on S6K2 localization to the centrosome.

Figure 3.

Characterization of centrosomal localization of S6K2. (A) RPE-1 cells were cultured in serum-free media for 16 hours (−) or in media containing fetal calf serum (+) and stained with CTR453 (red) or 167-4 (green). Right panels show an immunoblot of lysates for phospho-Erk or α-tubulin (loading control). (B-D) RPE-1 cells cultured in serum-containing complete media were treated without (−) or with (+) rapamycin (B), wortmannin (C), or U0126 (D) for 30 minutes and stained with CTR453 (red) or167-4 (green). Right panels show cell lysates from the experiments immunoblotted with antibodies against phospho-S6 (B), phospho-Akt substrates (C), or phospho-Erk (D). α-tubulin immunoblots are shown for loading control. (E) RPE-1 cells were cultured in serum-free media for 16 hours (−) and then treated with PMA (+) for 30 min. and stained with CTR453 (red) or 167-4 (green). Right panels show an immunoblot with anti-phospho-Erk or α-tubulin (loading control). Bars, 20 μm.

We next assessed whether PMA plays a role in S6K2 centrosomal localization. It has been shown that S6K2 can shuttle to the cytoplasm when stimulated with PMA [13]. We serum-starved RPE-1 cells and then treated them with or without PMA for 30 minutes. We observed in serum-starved cells both cytoplasmic and nuclear staining of S6K2 as seen in figure 3A, but when cells were stimulated with PMA, staining of S6K2 intensifies in the cytosol, indicating that S6K2 migrates to the cytosol upon PMA treatment (Fig. 3E). Our data confirm the previous findings that PKC regulates cytosolic shuttling of S6K2 [13]. However, we observed no change in centrosomal localization of S6K2 with PMA treatment.

We next assessed whether localization of S6K2 is a cell cycle-dependent event. We have already shown that during mitosis S6K2 is found in the spindle poles (Fig. 1C). We asked whether S6K2 is found in the centrosome in G1, S or G2 phases. RPE-1 cells were transfected with a construct for GFP-centrin 1 and stained with 167-4. Centrin 1 can specifically stain each centriole and therefore the number of centrioles as well as the distance between them can be used to determine the cell cycle stage of a given cell. Figure 4A shows that S6K2 localizes to the centrosome in G1, S and G2 phases of the cell cycle in RPE-1 cells. Finally, we carried out immunofluorescence on purified centrosomes using CTR453 and 167-4. Figure 4B confirms that S6K2 is found in the biochemically-purified centrosome as well. The staining shows that the S6K2 and CTR453 overlap but not completely, indicating that S6K2 is not a core integral centrosomal protein but rather is found in the pericentriolar area.

Figure 4.

S6K2 is found at the centrosome throughout the cell cycle and is found in the pericentriolar area. (A) RPE-1 cells were transfected with GFP-centrin 1 and stained with 167-4 (red). (B) Purified centrosomes were fixed onto coverslips and stained with CTR453 (red) and 167-4 (green). Bar, 20 μm.

DISCUSSION

In this paper we show for the first time that S6K2, but not S6K1, localizes to the centrosome throughout the cell cycle, more specifically to the pericentriolar area of the centrosome complex. There are several implications to our findings. First, this provides one more piece of evidence that S6K1 and S6K2 probably have non-overlapping cellular functions, and that, although both kinases phosphorylate the S6 protein, they may also have distinct function and substrates. Second, there have been several recent papers showing that proteins in the PI3K/mTOR signaling pathway such as Akt [23,24], TSC1 [25], and PI3K [26,27] localize to the centrosome. PI3K/mTOR pathway may play a yet-to-be-defined role in broader cytoskeleton regulation; most recently two proteins in the TOR pathway, Raptor and RheB, have been shown to play a role in mitotic spindle assembly in Drosophila [28] . We now have evidence that S6K2, one of the downstream targets of the PI3K/mTOR pathway, is also found in the centrosome, whereas another target, S6K1, does not. The full biological function for localization of the PI3K/mTOR pathway to the centrosome remains to be determined, although our data show that it is unlikely that the centrosome-relevant signaling from this pathway would be exerted via S6K1. Astrinidis et al. showed that TSC1 localizes to the centrosome and plays a role in centrosome duplication via the mTOR pathway [25]. It would be interesting to determine if S6K2 plays a role in regulation of centrosome duplication by TSC1.

Since S6K2 is a kinase it could be assumed that it phosphorylates a yet-to-be-identified substrate(s) at the centrosome that may affect centrosomal function. Identification of possible centrosomal substrates for S6K2 is underway. Current data suggest that S6K2 is not likely to be an absolute requirement for centrosome function; treatment of RPE-1 cells with rapamycin up to 6 hours did not alter the number or gross phenotype of the centrosome (Fig. 3B and data not shown), and S6K2 knockout mice do not have an obvious phenotype, suggesting that, if S6K2 plays a role in centrosome function, it is likely to be more subtle and readily compensated by another protein. It is also possible that S6K2 migrates to the centrosome not to affect centrosome's function but rather in order to receive certain signals from the centrosomal complex. This opens an interesting possibility that cytoskeletal regulation may play a role in S6K2 function. Our data suggest a possible interplay between centrosome function and the mTOR pathway. Further investigation is underway to delineate the detailed nature of this interplay.

Abbreviations

S6K1

Ribosomal S6 kinase 1

S6K2

Ribosomal S6 kinase 2

mTOR

mammalian target of rapamycin