Re-examination of the proposed SAC function of PICH
PICH could effectively be depleted from HeLa cells using any one of three different siRNA oligonucleotides, referred to as PICH-1, PICH-2, and PICH-CC. Live-cell imaging performed on HeLa cells stably expressing histone H2B–green fluorescent protein (GFP) showed that all three siRNA treatments also caused remarkably similar mitotic phenotypes. Compared to control siRNA (siGL2)-treated cells, PICH-depleted cells failed to form proper metaphase plates and instead displayed premature onset of anaphase (Supplementary Fig. 1a); moreover, PICH-depleted cells failed to arrest in response to nocodazole (Supplementary Fig. 1b), confirming and extending previous results (Baumann et al. 2007). To strengthen the conclusion that PICH is the critical target of siRNA in these experiments, we asked whether normal mitotic timing could be restored in PICH-1 siRNA-treated cells by overexpression of an siRNA-resistant PICH protein tagged with mCherry. Whereas control cells spent an average time of 43.8 min between nuclear envelope breakdown (NEBD) and anaphase onset, PICH-depleted cells expressing mCherry vector alone expeditiously initiated anaphase with an average time of 26.8 min (Fig. 1). In contrast, the overexpression of mCherry-tagged PICH wild type (WT) successfully restored the mitotic timing to near-normal levels (42.6 min), although some cells died, in line with the previously observed toxicity of excess PICH (Fig. 1; Baumann et al. 2007). The concordant phenotypes produced by three independent siRNA oligonucleotides and the apparently successful rescue initially bolstered our confidence that PICH plays an essential role in SAC signaling. However, several observations subsequently challenged this conclusion, as described in the following sections.
A first unexpected result emerged when we generated a 293T cell line (293T TREX-shPICH), which allowed the inducible expression of a short hairpin RNA (shRNA) coding for the PICH-1 siRNA sequence. Three stable clones were obtained, all of which showed efficient PICH depletion after 72 h of incubation with tetracycline. This depletion was reversible, as PICH protein was partially restored upon removal of tetracycline and incubation of cells in tetracycline-free medium for 72 h (Fig. 2a). Much to our surprise, none of these three 293T cell lines showed any sign of SAC failure (such as aberrant mitoses or micronucleation; data not shown), although PICH was efficiently depleted (Fig. 2a, b). Furthermore, neither the KT localization of Mad2 nor its levels were detectably affected upon PICH depletion (Fig. 2a, c). Taken at face value, these results suggested the existence of cell-type-specific differences in the response of HeLa and 293T cells to PICH depletion, challenging the idea that PICH is an essential component of the SAC in all cell types.
A second disturbing result emerged when we tried to restore SAC activity in PICH-depleted HeLa cells by re-expressing PICH protein at physiological levels. Specifically, we examined the consequences of depleting PICH from a HeLa cell line stably expressing a LAP-tagged mouse homolog of PICH (Poser et al. 2008; Wang et al., manuscript submitted). Although mouse PICH (Ercc6L) shares extensive sequence similarity with its human ortholog (72% amino acid identity; 81% amino acid similarity), expression of LAP-mouse PICH failed to restore SAC functionality in PICH-1 siRNA-treated HeLa cells, as indicated by both extensive micronucleation (Fig. 3a) and accelerated mitosis (Fig. 3b). Yet, LAP-mouse PICH was clearly resistant to PICH-1 siRNA treatment; in fact, LAP-mouse PICH showed enhanced expression upon depletion of endogenous human PICH, suggesting the existence of a feedback mechanism controlling total PICH protein levels (Fig. 3c). These results suggested that either mouse PICH is not able to provide SAC functionality in human cells, that LAP tagging interfered with functionality, or that PICH is not the only target of PICH-1 siRNA.
Although earlier observations had suggested that PICH depletion affected Mad2 localization to KTs rather than Mad2 protein levels (Baumann et al. 2007), in the course of the above studies, we noticed that treatment of HeLa cells with the siPICH-1 oligonucleotide caused a reduction in the level of Mad2 protein (Fig. 3a). Thus, the expression of Mad2 protein in response to various PICH siRNA treatments was carefully re-examined. Remarkably, depletion of PICH by all three published PICH-directed siRNAs showed detectable, albeit partial, reduction of Mad2 protein levels (Fig. 4a). Virtually identical results were obtained regardless of whether cells were arrested by thymidine or collected by shake off from an MG132-arrest (data not shown), demonstrating that the observed decrease in Mad2 protein was independent of cell cycle stage. Transfection of cells with siRNAs targeting several genuine SAC components did not cause a reduction of Mad2 protein (Fig. 4b). This rules out general effects related to siRNA transfection per se and demonstrates that the observed reduction of Mad2 levels in PICH siRNA-treated cells is not simply a consequence of SAC abrogation.
If the previously observed loss of Mad2 protein from KTs in PICH-depleted cells reflected a depletion of Mad2 protein rather than a selective disruption of the Mad1–Mad2 interaction at KTs, we reasoned that it might be possible to restore Mad2 localization to KTs by overexpression of exogenous Mad2 protein in PICH siRNA-treated cells. This was indeed the case, as a Mad2–GFP fusion protein readily localized to KTs in a PICH-1 siRNA background (Supplementary Fig. 2a). In contrast, overexpressed Mad2–GFP did not localize to KTs in cells depleted of Aurora B (Supplementary Fig. 2b), consistent with previous reports (Ditchfield et al. 2003; Vigneron et al. 2004). These observations thus support the conclusion that the PICH siRNA phenotype reported previously might reflect a change in Mad2 abundance rather than localization.
To investigate the mechanism underlying the reduction in Mad2 protein, we next asked whether this reduction was also reflected at the mRNA level. To this end, PICH, Mad1, and Mad2 mRNAs were measured by quantitative real-time polymerase chain reaction (qRT-PCR) in HeLa cells depleted of either PICH or Mad1 for control. As summarized in Fig. 4c, all three PICH siRNAs displayed almost identical knockdown efficiencies with regard to PICH mRNA. Remarkably, all three PICH siRNAs also caused a significant decrease of Mad2 mRNA when compared to the siGL2 control, whereas Mad1 mRNA was not detectably affected (Fig. 4c). As expected, Mad1 siRNA depleted Mad1 mRNA but did not reduce either PICH or Mad2 mRNA (Fig. 4c). Taken together, these results suggest that treatment of HeLa cells with any one of the three previously published PICH siRNA oligonucleotides reduces not only PICH but also Mad2 mRNA levels, suggesting that the previously observed loss of checkpoint functionality might actually reflect a reduction, albeit partial, of cellular Mad2 protein. This in turn implies that either PICH protein somehow regulates Mad2 expression or, alternatively, that the three PICH-directed siRNA oligonucleotides all display off-target effects that cause a lowering of Mad2 expression.
To distinguish between these possibilities, we sought to identify novel siRNA oligonucleotides that would effectively deplete PICH without lowering Mad2 mRNA and protein levels. Indeed, four new siRNAs, referred to as PICH-3, PICH-4, PICH-5, and PICH-6, as well as a corresponding pool (PICH-SP; comprising all four duplexes), were found to deplete PICH effectively. In a direct comparison, PICH-3, PICH-4, and PICH-5 showed comparable efficiency to PICH-1, PICH-2, and PICH-CC siRNA, whereas PICH-6 and the pooled PICH-SP duplexes were slightly less efficient (Fig. 5a). Similar results were obtained regardless of the duration of treatment, indicating that the kinetics of depletion were also comparable (Supplementary Fig. 3). Most importantly and in striking contrast to PICH-1, PICH-2, and PICH-CC, none of the new PICH-directed siRNA oligonucleotides detectably affected Mad2 expression, regardless of whether Mad2 protein (Fig. 5a) or mRNA (Fig. 5b) was monitored. Furthermore, none of the new oligonucleotides detectably interfered with KT localization of Mad2, in contrast to cells treated with PICH-1, PICH-2, and PICH-CC siRNA (Fig. 5c, d). Most importantly, we also examined the influence of the new PICH-directed oligonucleotides on both nocodazole-induced SAC activation (Fig. 5e, f) and the timing of mitotic progression (Fig. 5g). In addition to determining the mitotic index after nocodazole treatment (Fig. 5e), live-cell imaging was performed on HeLa cells stably expressing histone H2B–GFP (Fig. 5f, g). Both mitotic timing (NEBD to anaphase onset) and the duration of the nocodazole-induced SAC arrest were very similar in cells treated with the new PICH-directed siRNAs and in GL2-treated controls, in striking contrast to cells treated with the previously described PICH siRNA oligonucleotides, which advanced anaphase onset and abolished SAC activity (Fig. 5f, g). In our view, the most plausible interpretation of these results is that the originally described PICH-directed siRNA duplexes interfere with SAC activity because they lower Mad2 transcript and protein levels through an off-target effect. We recognize, however, that in principle one could also attribute an off-target effect to those siRNA duplexes that do not lower Mad2 expression. In particular, one could argue that PICH plays a physiological role in controlling Mad2 expression and that the failure of some duplexes to affect Mad2 levels (and hence checkpoint functionality) might reflect off-target effects interfering with this (hypothetical) regulation. A genetic PICH knockout will be required to formally exclude this alternative interpretation.
Re-examination of the role of Tao1 kinase in the spindle checkpoint
Recently, a genome-wide siRNA screen identified a requirement for the Tao1 kinase (also known as microtubule affinity-regulation kinase kinase (MARKK); Johne et al. 2008; Timm et al. 2003) in the spindle checkpoint (Draviam et al. 2007). Moreover, Tao1 depletion was reported to cause a selective loss of Mad2 but not Mad1 from KTs (Draviam et al. 2007), highly reminiscent of the early data obtained after PICH siRNA (Baumann et al. 2007). Intrigued by this similarity, we originally suspected that PICH and Tao1 cooperate in a regulatory step controlling Mad2 localization. However, having obtained evidence that the apparent requirement for PICH in Mad2 localization is likely to reflect an off-target effect, we wondered whether a similar explanation might pertain to Tao1. To explore this possibility, we compared different Tao1-directed siRNA oligonucleotides for their efficiency to deplete Tao1, their effects on Mad2 localization, protein, and transcript levels, as well as their ability to abolish SAC activity. This survey included the most effective previously published Tao1 siRNA oligonucleotide (Tao1-NCB3) as well four new oligonucleotides (Tao1-2 to Tao1-5) and a smart pool comprising Tao1-2 to Tao1-5 (Tao1-SP).
All Tao1-directed siRNAs exhibited a similar ability to deplete Tao1 from HeLa cells, and curiously, PICH-1 siRNA also lowered Tao1 levels (Fig. 6a). Most importantly, siTao1-NCB3 additionally caused a significant reduction of Mad2 protein, comparable to the reduction of Mad2 seen after PICH-1 or PICH-CC siRNA (Fig. 6a). We cannot presently explain the discrepancy between this observation and the previous report, indicating that Tao1 siRNA does not lower Mad2 levels (Draviam et al. 2007). In any case, however, Mad2 levels remained unaffected upon treatment of cells with the other Tao1 siRNAs, except for Tao1-4, which produced a marginal effect (Fig. 6a). Similar results were also obtained upon depletion of Tao1 from U2OS and RPE-1 cells (Supplementary Fig. 4a, b). Quantitative Western blot analysis (three independent experiments; normalized to α-Tubulin) revealed that Tao1-NCB3 siRNA lowered Mad2 protein levels substantially, whereas the other Tao1-directed siRNAs barely affected Mad2, although they depleted Tao1 with similar efficiency as Tao1-NCB3 (Fig. 6b). In line with these observations, the KT localization of Mad2 was abolished only in cells treated with the Tao1-NCB3 siRNA but not in cells treated by any of the other Tao1-directed siRNAs (Fig. 6c, d).
Next, we used qRT-PCR to analyze the mRNA levels of Mad1, Mad2, and Tao1 after treatment of HeLa cells with different siRNAs targeting Tao1 or PICH. While Mad1 mRNA remained constant in all cases, Tao1 mRNA levels were similarly reduced by all Tao1 siRNAs and, to a lesser extent, PICH-1 siRNA (Fig. 7a). Most importantly, Mad2 mRNA was significantly reduced upon treatment with siPICH-1 and siTao1-NCB3 siRNAs and, to a lesser extent, Tao1-4 siRNA. No changes in Mad2 mRNA levels were detectable in cells treated with siPICH-3, siTao1-2, siTao1-3, or control siRNA (siGL2; Fig. 7a). Parallel investigation of cellular phenotypes by time-lapse microscopy (Fig. 7b) revealed that cells depleted of Tao1 by Tao1-NCB3 siRNA went through mitosis more rapidly (NEBD to anaphase; average 14 min) than control cells (average 33 min). Mitotic timing was also reduced in cells treated with the Tao1-4 siRNA, albeit to a lesser extent (average 23 min), but not in response to any of the other Tao1-directed siRNA duplexes (Fig. 7b). Similarly, in line with the original report implicating Tao1 in SAC function (Draviam et al. 2007), an override of the nocodazole-induced SAC arrest could be observed in response to siTao1-NCB3 (average duration of arrest 25 min), whereas a robust checkpoint arrest (often followed by cell death after 12–15 h) was seen upon depletion of Tao1 by siTao1-2 or siTao1-3 (Fig. 7c). Cells treated with Tao1-4 siRNA showed a partial phenotype in that they usually exited mitosis after about 4 h (Fig. 7c). Collectively, we interpret the above data to suggest that the apparent role of the Tao1 kinase in SAC signaling might in reality reflect an off-target effect produced by particular Tao1-directed siRNA oligonucleotides on Mad2, similar to the situation described above for certain PICH siRNA duplexes.
Rescue of PICH and Tao1 siRNA phenotypes by Mad2 expression from a bacterial artificial chromosome
If, as argued above, the abrogation of the SAC in cells treated by some PICH- and Tao1-directed siRNA oligonucleotides were reflecting an off-target effect that causes a reduction of Mad2 levels, it should in principle be possible to rescue this phenotype by re-expressing Mad2. To test this prediction, we used bacterial artificial chromosome transgenesis which is expected to restore protein expression to near physiological levels (Poser et al. 2008). Specifically, a pool of HeLa cells expressing a LAP-tagged murine Mad2 was subjected to PICH, Tao1, and Mad2 siRNA, and SAC functionality was then examined by live-cell imaging (Fig. 8a). In these experiments, GFP-positive cells (expressing murine Mad2) could directly be compared to GFP-negative cells (devoid of murine Mad2). In parallel, the levels of PICH, Tao1, human Mad2, and murine Mad2 were monitored by Western blotting (Fig. 8b). As expected, the mitotic timing (NEBD to anaphase onset) was accelerated in GFP-negative cells subjected to PICH-1, Tao1-NCB3, or Mad2 siRNA, when compared to GL2 controls. In contrast, mitotic timing was similar in GFP-positive cells, regardless of whether they were treated with GL2 control or Mad2 siRNA, attesting to the ability of murine Mad2 to compensate for the depletion of endogenous human Mad2. In the case of PICH siRNA, GFP-positive cells showed a delay in the traverse of mitosis, suggesting that loss of PICH in cells harboring (near) physiological levels of Mad2 actually triggers a mitotic delay. GFP-positive Tao1 siRNA-treated cells also showed a slowing of mitotic traverse compared to GFP-negative cells, although passage through mitosis was still faster than in GL2-treated controls. Taken at face value, this would be consistent with the notion that Tao1 is required for full SAC functionality even in the presence of physiological levels of Mad2 (Draviam et al. 2007). However, we note that all GFP-positive, murine Mad2-expressing cells showed a prolonged arrest in response to nocodazole, regardless of whether they were treated with PICH, Tao1, or (human) Mad2 siRNA (data not shown). Collectively, these data make a strong case that the observed reduction of Mad2 levels is largely, although in the case of Tao1 perhaps not exclusively, responsible for the abrogation of SAC activity in the PICH- and Tao1-siRNA-treated cells.
Uncovering of a regulatory influence of Plk1 on Mad2 function
A critical role for PICH and Tao1 in the SAC had originally been supported not only by consistent siRNA phenotypes but also apparently successful rescue experiments (this study, Fig. 1; Draviam et al. 2007). In particular, the SAC abrogation produced by Tao1-NCB3 siRNA could successfully be rescued by active but not inactive kinase, strongly supporting a role for Tao1 in the SAC (Draviam et al. 2007). However, the published evidence suggests that this rescue involved overexpression of siRNA-resistant Tao1 (Draviam et al. 2007), raising the possibility of a “bypass suppression” (in genetic terms) rather than a genuine rescue. In particular, we consider it possible that excess exogenous Tao1 activity might have interfered with chromosome attachment to the spindle apparatus in such a way that even reduced levels of Mad2 might have been sufficient to restore an SAC response. So it would seem critical to demonstrate restoration of SAC activity in Tao1-depleted cells by physiological levels of Tao1 kinase.In the case of PICH, our results clearly demonstrate that overexpression of PICH protein restored SAC functionality in PICH-1 siRNA-treated cells (Fig. 1), and yet, physiological levels of LAP-mouse PICH provided no rescue (Fig. 3b). A first clue for how to explain this discrepancy emerged when we tested two PICH mutants for their ability to restore normal mitotic timing in PICH-1 siRNA-treated cells. In fact, a PICH protein carrying a mutation predicted to abolish its ATPase activity (K128A, Wang et al., submitted; Leng et al. 2008) readily restored SAC activity in PICH-1 siRNA-treated cells, whereas a mutant that cannot bind Plk1, due to a mutation within the so-called Polo-box domain (T1063A), failed to restore proper mitotic timing (Fig. 9a). These unexpected results could not be attributed to different expression levels of the rescuing constructs, as neither the transfection efficiencies nor the levels of overexpression achieved revealed significant differences (Fig. 9b). Instead, they demonstrate that restoration of SAC functionality in PICH-1 siRNA-treated cells depends on the ability of the overexpressed PICH protein to sequester Plk1, rather than its ATPase activity. One straightforward interpretation of this result is that residual Mad2 protein, as it persists after PICH-1 siRNA, was sufficient to provide SAC activity, provided that Plk1 was sequestered by exogenous wild-type or ATPase-deficient PICH protein. Thus, the observed restoration of SAC activity would represent a case of bypass suppression rather than genuine rescue.
A corollary of the above interpretation is that Plk1 activity contributes to regulate the efficacy of Mad2 in SAC signaling, so that less Mad2 is needed for SAC functionality in the absence of Plk1 than in its presence. If this were the case, one would predict that siRNA-mediated depletion of Plk1 or inhibition of Plk1 by the small molecule inhibitor TAL (Santamaria et al. 2007) should also be able to restore SAC activity in PICH-1 siRNA-treated cells. To directly test this prediction, we performed live-cell imaging on HeLa cells stably expressing histone H2B–GFP and monitored mitotic progression under various treatments. Cells were depleted of either PICH or Tao1, using PICH-1 or Tao1-NCB3 siRNA, respectively, and some samples were subjected to additional depletion of Plk1 by siRNA or treated with TAL. Whereas cells treated with PICH-1 siRNA alone showed no mitotic arrest in response to nocodazole (Fig. 5f), codepletion of Plk1 or inhibition of Plk1 by TAL resulted in mitotic delays of 120 and 160 min, respectively (Fig. 10a), before cells exited mitosis without chromosome segregation and often formed micronuclei (data not shown). Single depletion of Plk1 or inhibition of Plk1 caused a prolonged prometaphase delay, mostly followed by apoptosis (Fig. 10a), as expected (Santamaria et al. 2007; Liu and Erikson 2003; Petronczki et al. 2008).
We also analyzed the effect of Plk1 depletion or inhibition on cells that were treated with Tao1-NCB3 siRNA (Fig. 10b). Whereas control cells exhibited the expected mitotic timing, Tao1-NCB3 siRNA-treated cells separated their chromosomes in less than 20 min without forming a metaphase plate. The simultaneous depletion of Plk1 by siRNA or its inhibition by TAL increased the timing in Tao1-NCB3 siRNA-treated cells to about 35 min, similar to the duration seen in control cells but far below the duration of the mitotic arrest seen after depletion of Plk1 alone (Fig. 10b). This suggests that Plk1 depletion or inhibition only marginally restored SAC activity in Tao1-NCB3 siRNA-treated cells, presumably because residual Mad2 levels were too low (see below).
To demonstrate that residual Mad2 was critical for the restoration of SAC activity by Plk1 depletion/inhibition in PICH-1 or Tao1-NCB3 siRNA-treated cells, we also examined the consequences of combining Plk1 inhibition with direct siRNA-mediated depletion of Mad2. As shown in Fig. 10c, TAL addition could not restore normal mitotic timing to cells that were extensively depleted of Mad2, arguing that the absence of Plk1 activity cannot rescue SAC functionality in cells from which the bulk of Mad2 has been depleted. These results support the view that residual Mad2, as it persists after PICH-1 or Tao1-NCB3 siRNA treatment, is critical for the restoration of some SAC activity upon depletion or inhibition of Plk1. As a final test of this conclusion, different amounts of Mad2 siRNA oligonucleotides were used to reduce cellular Mad2 to levels within the range of those seen as a consequence of PICH-1 or Tao1-NCB3 siRNA treatment (Fig. 10d). Then, live-cell imaging was used to analyze mitotic timing in TAL-treated cells as a function of Mad2 levels. As shown in Fig. 10e, the mitotic timing (NEBD to anaphase onset) observed in response to Plk1 inhibition showed a strong correlation with Mad2 levels. Furthermore, Tao1-NCB3 siRNA resulted in a more severe depletion of Mad2 than PICH-1 siRNA and (Fig. 10d), concomitantly, a more pronounced advancement of mitotic timing (Fig. 10e), in excellent agreement with the data shown above (Fig. 10a, b). On the basis of these results, we propose that restoration of SAC activity by depletion or inhibition of Plk1 depends on residual levels of Mad2, implying that depletion or inhibition of Plk1 enhances the functionality of Mad2. In turn, these observations point to the conclusion that—under physiological circumstances—Plk1 antagonizes that SAC signaling function of Mad2.