The p38 mitogen-activated protein kinase (p38 MAPK) family, which is comprised of four protein isoforms, p38α, p38β, p38γ and p38δ, forms one of the key MAPK pathways. The p38 MAPKs are implicated in many cellular processes including inflammation, differentiation, cell growth, cell cycle and cell death. The function of p38 MAPKs in mitotic entry has been well established, but their role in mitotic progression has remained controversial. We identify p38γ MAPK as a modulator of mitotic progression and mitotic cell death. In HeLa cells, loss of p38γ results in multipolar spindle formation and chromosome misalignment, which induce a transient M phase arrest. The majority of p38γ-depleted cells die at mitotic arrest or soon after abnormal exit from M-phase. We show that p38 MAPKs are activated at the kinetochores and spindle poles throughout mitosis by kinase(s) that are stably bound to these structures. Finally, p38γ is required for the normal kinetochore localization of polo-like kinase 1 (Plk1), and this contributes to the activity of the p38 MAPK pathway. Our data suggest a link between mitotic regulation and the p38 MAPK pathway, in which p38γ prevents chromosomal instability and supports mitotic cell viability.
The mitogen-activated protein kinases (MAPKs) work in fundamental cellular processes and participate in signaling networks composing four distinct pathways: the extracellular signal-regulated kinases (ERK) pathway, the Jun N-terminal or stress-activated protein kinase (JNK–SAPK) pathway, the ERK5–BMK1 (big MAP kinase 1) pathway, and the p38 MAPK pathway. The p38 MAPK cascade is involved in several crucial cellular signaling events, including inflammatory and stress responses, cell growth and differentiation, and regulation of cell cycle and cell death (Aouadi et al., 2006; Thornton and Rincon, 2009). Four p38 MAPKs, namely p38α (MAPK14), p38β (MAPK11), p38γ (MAPK12/ERK6/SAPK3) and p38δ (MAPK13/SAPK4) have been characterized in mammals (Martín-Blanco, 2000).
The role of the p38 MAPK pathway in the regulation of mitotic entry has been well established in nonstress conditions (Cha et al., 2007) and in response to various genotoxic stresses such as DNA damage (Bulavin et al., 2002; Mikhailov et al., 2005). However, evidence for direct roles of the p38 MAPKs in mitotic progression after the G2–M transition has remained controversial and is very limited. Takenaka and co-workers (Takenaka et al., 1998) presented evidence that the p38 MAPK cascade is activated in NIH3T3 cells in response to spindle damage induced by the microtubule drug, nocodazole, but not during unperturbed mitosis. Moreover, the authors reported that the nocodazole-induced mitotic block was alleviated by SB203580, a putative inhibitor of p38 MAPK. In contrast to these results, p38 MAPK pathway activation was found to be essential for unperturbed mitosis in HeLa and NIH3T3 cells (Tang et al., 2008) and in rat retinal cells (Campos et al., 2002). Moreover, treatment with SB203580 led to mitotic arrest in both studies. Last, p38γ but not p38α or p38β promoted the G2–M transition in Xenopus oocytes by activating XCdc25C (Perdiguero et al., 2003). To get new insights into p38 MAPK signaling during mitosis we have individually silenced each of the four p38 MAPK isoforms in HeLa cells and monitored the cell progression through mitosis. We find that depletion of p38γ results in severe mitotic defects including multipolar spindle formation, chromosome misalignment and massive cell death from mitosis. This paper represents the first comprehensive analysis of p38γ function during mitosis and identifies a new link between the p38 MAPK pathway and the mitotic signaling network.
p38γ-depleted cells exhibit a transient M phase arrest followed by massive cell death
To address how the lack of p38 MAPK affects mitotic progression in cells, we depleted individual p38 MAPK isoforms from human HeLa cells using RNAi and followed their progression through the cell cycle. First, the relative amounts of mRNA encoding the individual p38 MAPK isoforms were determined using RT-PCR; mRNAs encoding p38α and p38γ were much more abundant in HeLa cells compared with those encoding p38β and p38δ (Fig. 1A,B). Transfection of 100 nM siRNA targeting p38α or p38γ reduced the mRNA levels by 89.4±12.6% and 82.5±10.4%, respectively, compared with mock-transfected cells (Fig. 1B). The silencing of p38δ (66.2±23.0% reduction) and p38β (27±14.0% reduction) was not as successful. The results were confirmed with quantitative PCR, which indicated reduction of mRNA encoding p38α and p38γ by 87.4±6.3% and 89.5±4.1%, whereas p38β and p38δ mRNAs were down by 59.3±9.1% and 76.9±5.2%, respectively (Fig. 1C). We also tested two other siRNAs targeting p38α and p38γ. The second p38α-targeting siRNA (α-siRNA2) did not markedly reduce the p38α mRNA levels, whereas the second p38γ-targeting siRNA (γ-siRNA2) resulted in a 58.8±4.1% reduction of p38γ mRNA (Fig. 1C). Therefore, the silencing of p38 (α-siRNA1) and p38γ (γ-siRNA1) was effective and these isotypes were selected for further analyses. The specificity of the siRNAs selected for further analysis were confirmed by Q-PCR analysis, demonstrating that siRNAs targeting p38α or p38γ did not affect the quantity of mRNA encoding the other p38 MAPKs (Fig. 1D).
Next, we monitored the cell cycle progression in HeLa cells transfected with the potent p38α siRNA (α-siRNA1) or with the two individual p38γ siRNAs (γ-siRNA1 and γ-siRNA2) and compared results with those from cells transfected with nontargeting control siRNA (Fig. 2A,B). Both p38γ-targeting siRNAs produced the same phenotype described below. However, the more potent one (γ-siRNA1) was selected for further analyses. Live cell imaging (Fig. 2A–C) and FACS (Fig. 2E) analyses showed normal proliferation rates (mitotic indices 4.1±1.8% and 6.2±2.3%), cell cycle progression and low cell death indices (4.8% and 4.0% at 72 hours) after transfections with control or p38α-targeted siRNA, respectively. By sharp contrast, p38γ silencing resulted in a significant accumulation of cells in mitosis (27.4±8.3%, P<0.001) and an increase in the cell death rate (31.6% at 72 hours, Fig. 2E). Transfection of HeLa cells with the p38γ-targeting γ-siRNA1 resulted in a notable reduction (39%) in the p38γ protein levels 30 hours after transfection (Fig. 2D), which most likely is an underestimation due to the induced cell death. Since FACS might underestimate cell death because of loss of fragmented nuclei, we confirmed p38γ siRNA-induced cell death using transmitted light microscopy. Various cell populations were filmed using time-lapse imaging for 72 hours starting immediately after transfection. The cell death index was determined by classification of cells into living and dead cell categories using the criteria defined by Häcker (Häcker, 2000) by scoring a total of 500 cells per sample and time point indicated. Significantly more of the p38γ-depleted cells (69.0±13.0%) died by 72 hours after transfection compared with p38α-depleted cells (19±3%) and cells transfected with control siRNA (16.4±1.0%) (Fig. 2F, P<0.001). Apo-ONE® assay indicated rapid and statistically significant elevation of caspase-3 and caspase-7 activity in the p38γ-depleted cells and in the All Death® control cells at 48 hours and 72 hours compared with control and p38α-depleted cells (Fig. 2G, P<0.01 at 48 hours and P<0.001 at 72 hours after transfection for p38α-depleted cells). In addition, 20 μM pan-caspase inhibitor, zVAD.fmk, was used to inhibit caspase activity in the p38γ-depleted cell population to examine the mode of cell death. Inhibitor was added 40 hours after siRNA transfection and cells were followed by time-lapse imaging. The cell death index was determined as described above. Significantly fewer p38γ-depleted cells died in zVAD.fmk treated population compared with controls; 26±7% vs 44±9% at 24 hours and 67±5% vs 79±4% at 48 hours after drug addition, respectively (P<0.05, supplementary material Fig. S1). These data suggest that depletion of p38γ induces caspase-dependent apoptotic cell death. During the course of the time-lapse filming, it became evident that the majority of p38γ-depleted cells die at M phase. Therefore, we examined the fate of mitotic cells using separate sets of siRNA-transfected cells that were subjected to time-lapse filming starting 48 hours after transfection. The mitotic cells were classified into living and dead cell categories using the same morphological criteria as above. The viability of mitotic cells in the p38γ-depleted population declined dramatically during the siRNA-induced M phase arrest. An average of 50% of the cells (n=100) were dead 7 hours after they had entered M phase and the maximum death index of mitotic cells (96%) was achieved 12 hours after entry into M phase (Fig. 2H). This is significantly different (P<0.001) from p38α and control siRNA populations, in which fewer than 5% of the mitotic cells died during the filming session (Fig. 2H). The majority of mitotic cells (72%, n=100) analyzed in the p38γ-depleted population died without clear signs of exit from mitosis whereas 24% of the cells exited M-phase with cytokinesis, but all daughter cells died within 4 hours of the mitotic exit.
p38γ-depleted cells fail to maintain metaphase chromosome alignment and show anaphase defects
To determine the effects of p38γ silencing on chromosome segregation in more detail, we used HeLa cells stably expressing histone H2B–GFP. Populations of HeLa H2B–GFP cells were transfected with p38γ, p38α or nontargeting control siRNA, and their mitotic progression was monitored with live cell imaging (Fig. 3, supplementary material Movies 1 and 2). As before, loss of p38γ resulted in a transient mitotic arrest that was followed by cell death or occasionally abnormal exit from M phase (Fig. 3A,B). The average duration of mitosis in the p38γ-depleted population was significantly longer (275±130 minutes, range 49–910 minutes, P<0.001) compared with p38α-depleted cells (54±28 minutes) and control cell population (59±19 minutes, Fig. 3C, n=30–50 cells per group in three separate assays). The p38γ-depleted cells typically exhibited many misaligned chromosomes. In the majority of the cells, chromosomes failed to move to the spindle equator, whereas in some cells, the chromosomes first congressed to the metaphase plate but failed to maintain their alignment (Fig. 3B). Quantification of mitotic phases confirmed that loss of p38γ led to a significant accumulation of prometaphase cells and a decrease in the proportion of anaphase cells (P<0.001) compared with p38α-depleted and control cells (Fig. 3B). Many of the p38γ-depleted cells that evaded cell death and succeeded to enter anaphase exhibited unaligned chromosomes or chromatid bridges (Fig. 3C). To investigate whether silencing of p38γ leads to an override of the mitotic arrest induced by microtubule drugs, we filmed p38γ-depleted and control HeLa-H2B cells that entered mitosis in the presence of Taxol (50 cells per group were analyzed). No notable differences were observed in the length of M phase arrest or in the number of cells escaping the taxol block. Only 2.5% and 4.5% of the taxol-treated p38γ-depleted cells and control cells, respectively, exited mitosis within 10 hours of entering mitosis.
Lack of p38γ leads to the formation of extra-microtubule-emanating foci and reduced interkinetochore tension
To examine spindle morphology in p38γ-depleted and control cells, we processed the cells for immunolabeling with anti-pericentrin and anti-tubulin antibodies 48 hours after transfection. Multipolarity was significantly increased in p38γ-depleted cells (65.5±4%, n=100, P<0.001) compared with controls (3±3%), whereas no monopolarity was observed in either population (Fig. 4A). Typically, loss of p38γ resulted in the formation of one or more extra-microtubule-emanating foci near the main spindle poles and elongation of the spindle apparatus without disturbance to the general spindle architecture (Fig. 4A). The average pole-to-pole distance at prometaphase was significantly increased in p38γ-depleted cells to 16.3±1.82 μm from the 12.7±0.68 μm of control cells (P<0.01, n=25 spindles per group). Loss of p38γ also reduced the interkinetochore distance of metaphase chromosomes (Fig. 4B) to 0.92±0.09 μm (range 0.77–1.20 μm, n=5 sister kinetochore pairs in each of five cells analyzed), which was almost identical to the distance in nocodazole-treated cells with no tension (0.83±0.08 μm, range 0.65–0.96 μm, n=5 sister kinetochore pairs in each of five cells analyzed). These values were both significantly different (P<0.001) from the control (1.59±0.14 μm, range 1.42–2.0 μm, n=5 sister kinetochore pairs in each of five cells analyzed) (Fig. 4B). We also tested whether the loss of p38γ affected the stability of kinetochore–microtubule attachments by analyzing the resistance of kinetochore–microtubule fibers (k-fibers) to treatment with cold Ca2+ buffer, which removes unstable microtubules (Yao et al., 2000). This treatment abolished astral microtubules in both control and p38γ-depleted cells, but did not have any major effect on the k-fibers (Fig. 4C).
Eg5 is required for multipolar spindle formation in p38γ-depleted cells
To test whether p38γ-depleted cells exhibit mono-, bi- or multipolar spindles upon reactivation of Eg5, a mitotic kinesin that is required for bipolar spindle assembly (Kapoor et al., 2000), we treated p38γ-depleted and control cells with 100 μM monastrol (an Eg5 inhibitor) for 4 hours and then removed the drug by extensive washing. The proteasome inhibitor MG132 was added to the culture medium after monastrol washout to maintain mitotic arrest. In the p38γ-depleted population 37±8% of the cells (n=100) had arrested in mitosis with monopolar spindles during the 4 hour drug incubation, whereas 95±1% of the control cells (n=100) were monopolar (Fig. 4D). This difference was due to the large population of p38γ-negative cells that were already arrested at M phase at the time monastrol was added, and these cells remained bi- or multipolar during the 4 hour incubation (Fig. 4D). The majority of control cells (83±3%, n=100) subjected to monastrol washout into MG132-containing culture medium formed bipolar spindles with full chromosome alignment within an hour. In contrast to this, only 24±4% of the p38γ-depleted cells formed bipolar spindles upon release from monastrol, whereas the majority of cells formed multipolar spindles (76±3%, n=100. This indicated that p38γ-negative monopolar cells became multipolar upon Eg5 reactivation (Fig. 4D). We also observed that most of the p38γ-depleted cells with bipolar spindles exhibited several unaligned chromosomes after monastrol washout (Fig. 4E,F, yellow arrows in bottom panel).
Active p38 MAPK localizes to kinetochores and spindle poles and is created by bound mitotic kinases
To investigate localization of active MAPKs in mitotic cells, we stained cells with an antibody that recognizes the dually phosphorylated p38 MAPKs (Thr180,Tyr182; phosphorylated p38) (Fig. 5A-E). Upon staining cells in different mitotic phases, we detected activated p38 MAPK at the kinetochores from prophase to late telophase, at the spindle poles from prometaphase to late telophase and at the mid-body in telophase. Also, a notable cytoplasmic fraction of phosphorylated p38 was present throughout mitosis. To confirm that the antibody against phosphorylated p38 recognized the p38γ isoform, we immunoprecipitated p38γ from taxol-arrested HeLa cells and performed western blotting. A strong protein band corresponding to the expected size of the phosphorylated p38 MAPK was present below the heavy chain in the p38γ pull-down fraction, whereas no detectable amount of phosphorylated p38 remained in the supernatant (Fig. 5F). The immunoprecipitation assay was also performed reciprocally with similar results (data not shown). To exclude the possibility that the anti-p38γ antibody either recognizes or pulls down the p38α isoform we blotted the p38γ immunoprecipitate with an antibody specific to p38α (ms p38α). There were no detectable levels of p38α present in the p38γ immunoprecipitate, but instead all of the p38α isoform remained in the supernatant fraction. We conclude that the antibody against phosphorylated p38 also recognizes the p38γ isoform in mitotic cells and that the p38γ does not interact with the p38α isoform.
To investigate which kinases are involved in p38 MAPK activation in early mitosis, we used the lysed cell assay (Ahonen et al., 2005; Daum and Gorbsky, 2006). In this assay, cells growing on coverglasses are extracted with a detergent and incubated in a buffer without the phosphatase inhibitor microcystin-LR (MC) or PhosSTOP phosphatase inhibitor cocktail allowing removal of the soluble cytoplasm and dephosphorylation of phosphoepitopes at kinetochores and other subcellular structures by the active phosphatases. Supplementation of the lysed cells with a buffer containing MC or PhosSTOP and ATP allows the re-creation of those phosphoepitopes that are targeted by adjacent stably bound kinase(s) (endogenous rephosphorylation). As expected, phosphorylated p38 signal was abolished to background levels after dephosphorylation (Fig. 6A). However, bound endogenous kinase(s) were able to re-create the epitope on kinetochores and spindle poles during rephosphorylation without any external source of kinase activity than the ATP in rephosphorylation buffer (Fig. 6B). Therefore, kinases that are responsible for creating phosphorylated p38 were stably bound to these subcellular structures. To investigate which mitotic kinases control phosphorylation of p38, we tested a number of mitotic kinase inhibitors in the rephosphorylation assay. Interestingly, the inhibition of Polo-like kinase 1 (Plk1) by ZK-thiazolidinone (TAL) (Santamaria et al., 2007) during the rephosphorylation step significantly reduced phosphorylated p38 signal intensity at kinetochores and spindle poles (Fig. 6B) whereas inhibition of Aurora B by ZM447439 (Ditchfield et al., 2003) did not affect the phosphorylated p38 epitope. For quantification of phosphorylated p38 at the kinetochores, we measured the integrated fluorescence intensities minus the background per 10 kinetochores in each of seven cells analyzed per group as described elsewhere (Beardmore et al., 2004). When the activities of the bound enzymes were abolished with N-ethylmaleimide (NEM) and the lyzed cells were supplemented with exogenous kinases present in M phase extract prepared from nocodazole-blocked cells, the phosphorylated p38 epitope reappeared at kinetochores and spindle poles (Fig. 7A,C). By contrast, the addition of S phase extract to the lyzed cells after NEM treatment did not lead to the re-phosphorylation of phosphorylated p38. This indicates that kinase activity required for creating phosphorylated p38 was present in mitotic but not in interphase cells. When M phase extract depleted of Plk1 was added together with ATP or when Plk1-depleted M phase extract was washed out before addition of ATP on NEM pretreated cells, the phosphorylated p38 signal was not regenerated (Fig. 7B,C). This strengthens the notion that Plk1 is required for creation of the Thr180, Tyr182 epitope of p38 MAPK.
p38γ is needed to target Plk1 to the kinetochore
A recently reported mitotic substrate of the p38 MAPK pathway is Plk1, which is phosphorylated at Ser326 by MAPK-activated protein kinase 2 (MK2) in vitro and in vivo (Tang et al., 2008). To analyze the effect of p38γ depletion on Plk1 levels at poles and kinetochores, we stained p38γ-depleted and control cells with antibodies against Plk1. A dramatic reduction was observed in the kinetochore level of Plk1 (down by 93±9%, P<0.001) after p38γ depletion in comparison with control cells (Fig. 8A). In these cells, the loss of kinetochore-bound Plk1 was accompanied with an increase in the cytosolic fraction of Plk1 (Fig. 8A). Plk1 levels at the main spindle poles were also decreased by 41±7% in the p38γ-depleted cells compared with levels in the control (P<0.01, Fig. 8A), but unexpectedly the signal intensity of phosphorylated Plk1 (Ser326) was not affected (Fig. 8B). This suggests that p38 MAPK isoforms other than p38γ are responsible for phosphorylation of Plk1 at Ser326 and that only a proportion of pole-located Plk1 is phosphorylated at this residue. Further support for this notion was provided by the observation that the phosphorylation of MK2 at Thr334 was not reduced in the p38γ-depleted cells (data not shown).
The p38 MAPK pathway consists of four kinase isoforms (p38α, p38β, p38γ, and p38δ) that have been implicated in inflammation, cell differentiation, cell growth, cell death, senescence and tumorigenesis (Aouadi et al., 2006; Thornton and Rincon, 2009). The pathway is also an interesting therapeutic opportunity for the treatment of various diseases and over 20 different p38 MAPK inhibitors have entered clinical trials (Goldstein and Gabriel, 2005). p38α was the first p38 MAPK isolated (Han et al., 1994) and is now the most thoroughly studied isoform, whereas much less is known about the other three family members. In addition to the functions mentioned above, the p38 MAPK pathway also controls the cell cycle by regulating both G1–S and G2–M transitions (MacCorkle and Tan, 2005). A role for the pathway in mitotic entry has been well established (Cha et al., 2007; Bulavin et al., 2002; Mikhailov et al., 2005), but data regarding the control of mitotic events by p38 MAPKs is very limited and controversial. Here, we present evidence that p38γ acts as important modulator of mitotic spindle architecture and that its loss of function has an impact on mitotic cell survival.
Role of p38γ in mitotic progression and mitotic cell survival
Our studies revealed that p38α and p38γ were the most abundant p38 MAPK isoforms expressed in HeLa cells. The presence of p38γ in cycling HeLa cells was an unexpected finding because earlier work demonstrated that this isoform is predominantly expressed in skeletal muscle (Lechner et al., 1996; Li et al., 1996), whereas p38α is more ubiquitously expressed (Jiang et al., 1996). Analyses after silencing individual p38 MAPKs revealed isoform-specific mitotic effects. In sharp contrast to p38α siRNA treatment, which did not yield any notable cell cycle defects or induction of apoptosis, the specific depletion of p38γ led to a transient accumulation of mitotic cells starting 24 hours after transfection. The p38γ-depleted mitotic cells exhibited spindle anomalies that prevented normal chromosome alignment and exit from mitosis. The spindle checkpoint remained activated because these cells spend several hours in mitosis and no notable escape from taxol-induced block was observed. The viability of the mitotic cells was dramatically reduced in the p38γ-depleted populations and the cells were eliminated via caspase-dependent apoptosis.
Our live cell assays showed that the p38γ-depleted cells had several unaligned chromosomes. This could be explained by the spindle anomalies observed in the p38γ-depleted cells. The cells exhibited several microtubule-nucleating foci that perturbed the normal spindle architecture. The multipolar phenotype was dependent on the normal function of Eg5. Furthermore, the Eg5 reactivation assays demonstrated that loss of p38γ impaired normal chromosome congression, even in cells with bipolar spindles, suggesting a role for p38γ in the correction of erroneous microtubule–kinetochore attachments or motor functions that move the chromosomes. However, as the k-fiber attachments in p38γ-depleted cells were stable according to their resistance to the cold Ca2+ buffer treatment, chromosome alignment defects were probably not due to mistakes in microtubule–kinetochore attachments. Interestingly, when p38γ was depleted, the interkinetochore tension was reduced to the same level as that seen in the absence of microtubules. Together, these phenotypes resemble the situation in cells after treatment with low doses of microtubule drugs that diminish interkinetochore tension but not the k-fiber occupancy (Waters et al., 1998; Skoufias et al., 2001) and therefore suggest that loss of p38γ interferes with normal microtubule dynamics.
Mitotic substrates of the p38 MAPK pathway
In the literature, only a few mitotic targets of the p38 MAPK cascade, such as Plk1 (Tang et al., 2008) and cytoplasmic dynein (Whyte et al., 2008), have been described. Both Plk1 and dynein are implicated in several mitotic processes including spindle organisation, chromosome motion and spindle checkpoint signaling (Karki and Holzbaur, 1999; Petronczki et al., 2008). MK2 colocalizes with activated p38 MAPK and Plk1 at spindle poles, where it phosphorylates Plk1 at Ser326 to promote normal mitotic progression (Tang et al., 2008). The expression of a Plk1 phosphomutant (S326A) causes spindle pole defects and the accumulation of cells at M phase (Tang et al., 2008).
Another phenotypic link between Plk1 and p38γ is the elimination of the M-phase-arrested cells after RNAi. Depletion of Plk1 causes mitotic arrests, albeit with monopolar spindle, and a dramatic decrease in cell viability (Spänkuch-Schmitt et al., 2002; Liu and Eriksson, 2003; Reagan-Shaw and Ahmad, 2005; Guan et al., 2005; Bu et al., 2008). Moreover, induction of apoptosis by Plk1 RNAi is associated with p53 malfunction (Guan et al., 2005), but whether this is linked to p38γ function remains to be determined. Interestingly, mislocalization of kinetochore-associated Plk1 by RNAi of Polo-box domain-binding protein PBIP1 has been reported to impair kinetochore–microtubule attachment and the generation of tension, thus causing severe chromosome misalignment (Kang et al., 2006). This is a reminiscent of the p38γ-depleted phenotype, although PBIP1 RNAi also caused other mitotic defects and compromised the spindle assembly checkpoint. Nevertheless, the observed requirement of p38γ for the normal kinetochore localization of Plk1 suggests a potential connection between the p38 MAPK cascade and Plk1 as determinants of cell fate at M phase.
Control of p38 MAPK activity during mitosis
In earlier studies, the active form of p38 MAPK was found in the cytosol of mitotic cells (Fan et al., 2005) or localized to the mitotic spindle poles (Tang et al., 2008). Here, we extend these findings by reporting kinetochore signals after immunofluorescent detection with an antibody recognizing the conserved Thr180,Tyr182 phosphoepitope present in all p38 MAPK isoforms. Activated p38 MAPK localized on kinetochores and spindle poles during mitosis. Silencing of p38γ alone was not sufficient to abolish the phoshoepitope signals, suggesting that p38γ is not the only active p38 isoform present at these subcellular foci (unpublished data). Using the lyzed cell model (Ahonen et al., 2005; Daum and Gorbsky, 2006), we tested the conditions that are required for creating phosphorylated p38 in mitotic cells. First, kinetochore and spindle-pole-bound kinases were able to re-phosphorylate the epitope pointing to the presence of immediate upstream kinases of p38 MAPK at these subcellular locations, or TAB1-dependent autophosphorylation mechanism of p38 MAPKs (Ge et al., 2002; Salvador et al., 2005). Second, the capability of the M phase but not the S phase extract to re-create phosphorylated p38 epitope demonstrated the mitotic origin of the elements required for p38 MAPK activation. Plk1 was implicated in the process by the observation that chemical perturbation of Plk1 but not Aurora B kinase or loss of Plk1 from the mitotic cell extracts during the rephosphorylation reaction was sufficient to suppress phosphorylated p38 epitope phosphorylation by the bound kinases. Together, the data strengthen the notion that Plk1 communicates with p38 MAPK to control its mitotic activity. Alternatively, Plk1 might stimulate the autophosphorylation of the Thr180,Tyr182 phosphoepitope of p38 MAPKs in the TAB1-dependent reaction.
In summary, the pleiotropic mitotic errors and massive cell death caused by loss of p38γ suggest that the kinase has several mitotic substrates whose perturbed function contributes to the observed phenotypes. Moreover, our results provide the first evidence for functional links between the p38 MAPK cascade and Plk1 in the control of mitotic progression and in promotion of cell viability at M phase. Details of such mitotic signaling reactions are subject to further experiments in various model systems.
Materials and Methods
HeLa and U2OS (human osteosarcoma) cells obtained from the American Type Culture Collection (Manassas, VA) were maintained in DMEM (Invitrogen, Carlsbad, CA) and Ptk1 (rat kangaroo kidney epithelial) cells (a gift from Gary Gorbsky, OMRF, Oklahoma City, OK) in MEM (Invitrogen) supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 0.1 mg/ml penicillin-streptomycin, 1 mM sodium pyruvate, 0.1 mM non-essential amino acids and 20 mM HEPES. The medium of stable H2B–GFP HeLa cell line (a gift from Geoffrey Wahl, Scripps Research Institute, La Jolla, CA) included 2 ng/ml blasticidin. Cells were grown in +37°C with 5% CO2.
Antibodies and reagents
In the study we used antibody against phosphorylated p38 (Cell Signaling Technology, Danvers, MA, 1:500), CREST autoimmune serum (Antibodies Incorporated, Davis, CA, 1:200), anti-Plk1 (Abcam, Cambridge, UK, 1:100), antibody against phosphorylated CenpA (Upstate Biotechnology Incorporated, Lake Placid, NY, 1:200), antibody against phosphorylated Plk1 (S326, 1:200), anti-α-tubulin (DM1A, Abcam, 1:200), anti-pericentrin (Abcam, 1:2000), anti-p38γ (Cell Signaling Technology, Danvers, MA, 1:100 in IF or 1:1000 in WB) and anti-p38α (1:1000 in WB) antibodies. The secondary antibodies anti-Cy3 (1:1000), anti-FITC (1:600) and anti-Cy5 (1:400) were from Jackson ImmunoResearch (Newmarket, UK) and Alexa Fluor antibodies from Invitrogen. Other reagents used in this study were purchased from Sigma if not stated otherwise.
Oligofectamine (Invitrogen) was used according to the manufacturer's instructions using the following siRNA targeting sequences: AAC TGC GGT TAC TTA AAC ATA (p38α-oligo 1), CAG AGA ACT GCG GTT ACT TAA (p38α-oligo 2), CAG GAT GGA GCT GAT CCA GTA (p38β), CTG GAC GTA TTC ACT CCT GAT (p38γ-oligo 1), TGG AAG CGT GTT ACT TAC AAA (p38γ-oligo 2), CCG GAG TGG CAT GAA GCT GTA (p38δ) and AAGATCACCCTCCTTAAATAT (Plk1). All Death (AllStars Cell Death Control siRNA) was purchased from Qiagen (Stanford, CA).
HeLa, U2OS and Ptk1 cells growing on coverslips were fixed and permeabilized in 2% paraformaldehyde in 60 mM PIPES, 25 mM HEPES (pH 7.0), 10 mM EGTA, 4 mM MgSO4 (PHEM) containing 0.5% Triton X-100 and 400 nM microcystin-LR for 15 minutes. For microtubule staining, 0.2% glutaraldehyde was added to the fixative. Cells on coverslips were rinsed in 10 mM MOPS (pH 7.4), 150 mM NaCl and 0.05% Tween 20 (MBST) and blocked with 20% boiled normal goat serum (BNGS) for 1 hour at room temperature (RT). Cells were stained with primary antibodies for 1 hour at RT. Cells on coverslips were washed with MBST and treated with secondary antibodies for 1 hour at RT. Antibodies were diluted to MBST containing 5% BNGS. Cells on coverslips were washed with MBST and DNA was stained with DAPI (4′,6-diamidino-2-phenylindole, 10 ng/ml in water). Cells on coverslips were washed with distilled water and mounted on microscope slides with Vectashield mounting medium (Vector Laboratories).
Lyzed cell assay
HeLa cells on coverslips were rinsed briefly in rinse buffer (50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 5 μg/ml protease inhibitor cocktail) and extracted in extraction buffer (50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 0.5% Triton X-100, 1 mM DTT, 5 μg/ml protease inhibitor cocktail) for 5 minutes. Samples were dephosphorylated in dephosphorylation buffer (50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 1 mM DTT, 5 μg/ml protease inhibitor cocktail) for 7 minutes (in some samples λ-phosphatase was added to dephosphorylation buffer), rinsed twice in rinse buffer and treated with either rephosphorylation buffer [50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 1 mM DTT, 400 nM Microcystin-LR or PhosSTOP (Roche), 1 mM ATP, 5 μg/ml protease inhibitor cocktail] for 20 minutes or with buffer without ATP (50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 1 mM DTT, 400 nM Microcystin-LR or PhosSTOP, 5 μg/ml protease inhibitor cocktail). In some assays 10 μM ZK-Thiazolidinone (TAL) or 20 μM ZM447439 was added to the rephosphorylation buffer before incubation of the samples for 20 minutes. In exogenous rephosphorylation assay cells were treated with N-ethylmaleimide (NEM) buffer (5 nM NEM in 50 mM Tris-HCl, pH 7.5, 4 mM MgSO4, 5 μg/ml protease inhibitor cocktail) for 10 minutes and washed in rinse buffer. Rephosphorylation buffer ± ATP was added together with mitotic or S phase HeLa cell extract or mitotic Plk1-depleted extract (5×106 cells/ml in extraction buffer; 1:20 in rephosphorylation buffer) for 40 minutes at RT. In some samples extracts were added in rephosphorylation buffer without ATP for 40 minutes, washed out in presence of phosphatase inhibitor followed by ATP addition for 40 minutes. Samples were fixed and subjected for immunofluorescence as described above.
Cold calcium lysis
Cells on coverslips were rinsed in 0.1 M PIPES (pH 6.95) and placed in ice cold (0–2°C) 0.1 M PIPES, pH 6.95, 80 μM CaCl2, 1% Triton X-100 for 5 minutes on ice bath. After the lysis, cells were rinsed in PIPES before fixation and immunofluorescence as described above.
Monastrol washout experiment
HeLa cells transfected with scrambled siRNA or siRNA to knockdown p38α or p38γ were treated with 100 μM monastrol for 4 hours. The cells were washed and released into fresh growth medium containing 20 μM MG132 for 1 hour before fixation and staining for tubulin and pericentrin. Some samples were also fixed before monastrol treatment or directly after the drug treatment before washout.
Microscopy and image analysis
Fixed samples were imaged with a Zeiss Axiovert microscope equipped with a 63× objective, Hamamatsu Orca-ER camera (Hamamatsu Photonics Norden AB, Solna, Sweden) and Metamorph imaging software (Molecular Devices, Downingtown, PA). Quantification of signal intensities was done using Metamorph imaging software. All values were corrected with background deduction.
Live cell imaging
Asynchronous or double thymidine synchronized HeLa and HeLa H2B–GFP cells growing on a well plate or μ-Slide eight-well plates (Ibidi, Martinsried, Germany) were imaged 48 hours after siRNA transfection for 16 hours. Images were captured using a Zeiss Axiovert microscope equipped with a 40× or 63× objective, Hamamatsu Orca-ER camera (Hamamatsu Photonics Norden) and Metamorph imaging software (Molecular Devices) or with the 3I-spinning disc confocal microscope (Intelligent Imaging Innovations, Goettingen, Germany) equipped with 63× objective, Hamamatsu Orca-ER camera (Hamamatsu Photonics Norden) and Slidebook imaging software (Intelligent Imaging Innovations). We also time-lapse imaged asynchronous HeLa cell populations on 384-well plates immediately after the siRNA transfection using incucyte imaging system (Essen Instruments, Welwyn Garden City, UK) for a total of 66 hours. In this study, duration of mitosis was defined as the time from nuclear envelope breakdown to end of telophase, or to the time of mitotic cell death. To calculate the cell death indices we used the morphological criteria of apoptosis described by Hacker (Hacker, 2000) and scored 1000 cells per each time point from three replicate assays.
HeLa cells transfected with scrambled, p38α or p38γ siRNA were harvested after 48 or 72 hours after transfection. Cells were fixed with −20°C ethanol followed by brief incubation in −20°C. Cells were centrifuged at 1300 rpm for 4 minutes. Cell pellets were washed and resuspended in PBS in round-bottom 96-well plates. RNase (100 μg/ml) and propidium iodide (20 μg/ml) diluted in PBS were added to the resuspended cells and the cells were incubated for 30 minutes with agitation at RT in a light-protected box. The samples were measured with LSR II system (Becton Dickinson, Franklin Lakes, NJ) and results were analyzed with FCS Express program.
Nondepleted HeLa cells or cells depleted for p38 MAPK isoforms were harvested 48 hours after the siRNA transfection. RNA was isolated with the RNeasy Midi Kit (Qiagen) and cDNA was prepared according to the manufacturer's instructions using iScript™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA). Different p38 MAPK isoforms were amplified from cDNAs using specific primers for different isoforms (see below). PCR products were separated on 1% agarose gel including 0.5 μg/ml ethidium bromide and imaged with UV (Gene Genius Bio Imaging System by Syngene, Cambridge, UK). Band intensities were quantified using Metamorph imaging software (Molecular Devices).
RNA was isolated from depleted HeLa cells and treated with DNaseI (Invitrogen) followed by cDNA isolation. p38 MAPK isoforms were amplified using isoform specific primers with q-PCR (TaqMan®, Applied Biosystems, Foster City, CA).
Statistical analysis was performed with GraphPad Prism and Excel software using Student's t-test or two-factor ANOVA test. The significance threshold accepted was P<0.05.
Caspase activity assay and inhibition
The Apo-ONE homogeneous caspase-3/7 assay (Promega, Madison, WI) was used to evaluate the activities of caspase-3 and caspase-7. HeLa cells were depleted with negative control, All Death control, p38α and p38γ siRNAs in 384-well format using oligofectamine reagent in a reverse transfection according to the manufacturer's instructions. After 24, 48 or 72 hours of incubation at 37°C, part of the medium was aspirated and Apo-ONE substrate buffer mix was added to each well. The plate was incubated for 30–60 minutes with agitation and was protected from light. EnVision 2100 multilabel reader (Perkin Elmer) was used to measure fluorescence (excitation wavelength 485 nm, emission wavelength 535 nm). Caspase activity was inhibited with 20 μM pan-caspase inhibitor zVAD.fmk. Inhibitor was added 40 hours after transfection to cells treated with siRNA against p38γ in 96-well plates.
Immunoprecipitation and western blot
Dynabeads (protein G) were washed twice with citrate-phosphate buffer (pH 5) containing 0.01% Tween-20 and left in citrate-phosphate buffer (pH 5). p38γ antibody was added on beads and incubated for 40 minutes at RT in rotation. Bead–antibody complexes were washed three times with citrate-phosphate buffer (pH 5) containing 0.01% Tween-20. Mitotic cell extract (5×106 cells/ml in PBS-based IP buffer; 1× PBS, 1 mM EDTA, 2 mM MgCl2, 0.2 mM CaCl2, 0.5% Triton X-100, 150 mM NaCl) arrested for 16 hours was then incubated with 0.6 μM Taxol at +4°C for 3 hours. IPs were washed three times with PBS-based IP buffer and bound proteins were eluted by adding SDS buffer and boiling for 5 minutes. Proteins from IP supernatant were precipitated with acetone and dried pellet was dissolved in SDS buffer and boiled for 5 minutes. SDS-PAGE was carried out using 4–12% gradient gels (Lonza) and proteins transferred to nitrocellulose membrane. Membrane was blocked in 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 0.05% Tween 20 (TBST) containing 5% nonfat dry milk for 1 hour. Membranes were incubated with primary antibodies overnight at 4°C, washed three times with TBST and incubated with secondary antibodies for 1 hour at RT protected from light. After washing, the Odyssey Infrared Imaging System (LI-COR Biotechnology) was used for detection of proteins.
We thank Bayer Schering Pharma AG, Xiaoqi Liu and Geoff Wahl for providing reagents for the study and Gary Gorbsky for discussions and critical reading of the manuscript. This study was supported by grants to M.J.K. from the Academy of Finland (120804), EUFP6 (Marie Curie EXT grant 002697), the Centre of Excellence for Translational Genome-Scale Biology, the Finnish Cancer Organisations and the Foundation for the Finnish Cancer Institute.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/124/2/216/DC1
- Accepted September 12, 2010.
- © 2011.