Determination of cyclin D1 localization by sub-cellular fractionation
We have previously shown that TSA induces cyclin D1 degradation and the apparent nuclear exclusion of ectopically expressed GFP-cyclin D1 in human breast cancer cell lines. These results coupled with previous findings suggested that TSA induces cyclin D1 degradation through mediating its export to the cytoplasm. To test this idea, we determined by sub-cellular fractionation, the effect of TSA on endogenous cyclin D1 localization in asynchronous human cancer cells. Surprisingly, we observed the majority of cyclin D1 is detected in the cytoplasmic fraction of untreated, asynchronous exponentially growing MCF-7 and MDA-MB231 breast cancer cells (Figures 1A and 1B). In contrast, the levels of GSK3α and GSK3β were similar in both the cytoplasmic and nuclear fractions. We confirmed efficient cytoplasmic and nuclear fractionation by localization of Sp1 and Hsp60 (or mitochondrial Hsp70 [mHsp70]) that localized predominantly to the nucleus and cytoplasm respectively. Culture of MCF-7 breast cancer cells with 1 μM TSA for 6 h resulted in a reduction of cyclin D1 level within both fractions, while treatment with 50 μM MG132 (a proteasome and cathepsin inhibitor) resulted in increased levels of the cyclin D1 within the nuclear fraction. We also observed in these studies, that MG132 abolished the TSA induced loss of cyclin D1 in both fractions but did not affect the level or localization of GSK3α and GSK3β (Figure 1A). In contrast, retinoblastoma protein (pRb), the cyclin D1-CDK4/6 substrate was detected largely within the nuclear fraction (Figure 1B). Similar results were observed in MDA-MB231 cells, with the exception that co-culture with MG132 also resulted in reduced GSK3α and GSK3β levels within both fractions (Figure 1C).
We also wished to determine if similar results could be observed in non-breast cancer cell lines and tissues. Firstly, we performed similar experiments in the SK-UT-1B uterine cancer cell line that has been shown previously to be defective in cyclin D1 degradation [12, 13]. As expected, TSA treatment had little effect on cyclin D1 levels [11] although co-culture with MG132 did result in a slight increase in cyclin D1 accumulation within the nuclear fraction. This observation suggested that low levels of cyclin D1 degradation do occur in this cell line (Figure 1D). The results observed in KNRK transformed rat kidney fibroblasts were similar to those observed in MDA-MB231 cells (Figure 1E), with the exception that MG132 that a more pronounced effect on cyclin D1 stability. Additional experiments in U2OS osteosarcoma and HeLa cervical cancer cells indicated that cyclin D1 was also localized predominantly within the cytoplasmic fraction of these cells. In addition, cyclin D1 was not detected in the insoluble fraction obtained from these cells following the separation of the cytoplasmic and nuclear fractions (Figure 2A). Analysis of the cytoplasmic and nuclear fractions from murine heart, kidney, liver and lung tissue, similarly demonstrated that the highest levels of cyclin D1 were detectable within the cytoplasmic fraction (Figure 2B). These observations are in agreement with those from a recent study by De Falco et al., on cyclin D1 localization in murine tissues [14].
In order to verify these results, we used two additional protocols to obtain cytoplasmic and nuclear fractions from MCF-7 cells. Firstly, a second commercially available kit was used to sequentially isolate cytoplasmic, membrane/organelle, nuclear and cytoskeletal fractions. In these fractions, the highest levels of cyclin D1 were again detected in the cytoplasmic fractions and to a lesser extent, in the membrane/organelle fractions (Figure 2C). Sp1 was detected solely in the nuclear fraction while equivalent levels of GSK3α and GSK3β levels were detected in the cytoplasmic, membrane/organelle and nuclear fractions. Only GSK3α was detectable in the cytoskeletal fraction, while the highest levels of Hsp60 were detected in the cytoplasmic and membrane/organelle fractions. Lastly, we used an in-house protocol to prepare cytoplasmic and nuclear fractions from MCF-7 cells. In these experiments, the highest levels of cyclin D1 were again detected within the cytoplasmic fraction (Figure 2D). Our data thus strongly suggested that cyclin D1 is localized predominantly within the cytoplasm, at least in the various cell lines and tissues tested.
Effect of cell cycle phase on cyclin D1 levels and sub-cellular localization
The treatment of MCF-7 cells with the pure antiestrogen ICI182,780 (100 nM) for 48 h results in reduced cyclin D1 levels and G1 cell cycle arrest in this cell line [15–17]. Treatment of these cells with 17β-oestrodiol following the removal of ICI182,780, thus results in cell cycle progression that is accompanied by and dependent on the increased expression of cyclin D1. We employed this system in order to investigate the effect of 17β-oestrodiol on cyclin D1 accumulation and localization in cytoplasmic and nuclear fractions from MCF-7 cells. Increased levels of cyclin D1 were detectable in both fractions at 3 h and were maximal at ~8 h as previously reported [17]. In our system however, the highest levels of cyclin D1 were again detectable within the cytoplasmic fraction, irrespective of the time point (Figure 3A).
We next examined the effect of cell cycle arrest at the G1/S-phase boundary on cyclin D1 localization and levels in MCF-7, MDA-MB231 and SK-UT-1B cells. Cells were left untreated or incubated in the presence of 1 mM hydroxyurea for 24 h. Under these conditions, >90% of hydroxyurea treated cells arrested in late G1 or early S-phase (results not shown). We observed a marked reduction in cyclin D1 levels in MCF-7 and MDA-MB231 treated cells (Figures 3B and 3C). Reduced cyclin D1 levels were accompanied by increased nuclear levels of the Skp2 F-box protein that has been previously linked to the ubiquitin-dependent degradation of this cyclin [12, 18]. The degree of cyclin D1 downregulation was clearly proportional to the increase in Skp2 levels (compare Figures 3B and 3C). The co-culture of hydroxyurea pre-treated cells with 1 μM TSA for 6 h, lead to further reductions in cyclin D1 levels in MCF-7 but not MDA-MB231 cells. Co-culture of both cell lines with TSA and MG132 (50 μM) however, lead to increased levels of cyclin D1 within both the cytoplasmic and nuclear fractions of both cell lines (Figures 3B and 3C). This observation suggested that the rate of cyclin D1 degradation is increased in both cell lines under these conditions. Hydroxyurea did not affect the levels or localization of GSK3α and GSK3β in MCF-7 cells but induced the downregulation of GSK3β in MDA-MB231 cells. Interestingly, co-culture with hydroxyurea did not result in a significant reduction of cyclin D1 levels in SK-UT-1B cells, despite a rise in nuclear Skp2 levels (Figure 3D). Since the reduction of nuclear cyclin D1 levels has been reported to be a prerequisite for S-phase progression, our observations raise interesting questions about how these cells might progress into S-phase in the absence of cyclin D1 degradation.
Analysis of cyclin D1 localization by confocal microscopy
Based on immunofluorescent studies, cyclin D1 has been previously localized to the nucleus of asynchronously growing cells [1]. Our observation that cyclin D1 localization was mainly restricted to the cytoplasmic fraction of the cell lines and tissues examined in this study was therefore surprising. Indeed, experiments from our own laboratory using an FITC-conjugated anti-cyclin D1 antibody also suggested a predominantly nuclear localization for this cyclin (Figure 4A). We observed however, that in MCF-7 cells transiently expressing GFP-Cyclin D1, the recombinant protein appeared to be localized to both the cytoplasm and nucleus when examined by direct immunofluorescence microscopy (Figure 4B). When these cells were subjected to sub-cellular fractionation and analysed by immunoblot analysis with a monoclonal antibody against GFP, the recombinant protein was again detected predominantly within the cytoplasmic fraction (Figure 4C). Single or combined mutations of Thr-286 or Thr-288 to alanine resulted in only a slight increase of the recombinant protein within the nuclear fraction. These observations strongly suggested that the apparent nuclear localization of cyclin D1 as determined by indirect immunofluorescence microscopy may be an artefact of this particular technique. We thus decided to examine the localization of both endogenous and transiently expressed GFP-Cyclin D1 in MCF-7 cells by confocal microscopy. These experiments confirmed that cyclin D1 does not indeed, localize predominantly within the nucleus of mammalian cells (Figures 4D and 4E).
Combined proteasome and CRM1 inhibition does not result in the nuclear localization of cyclin D1
The inhibition of protein synthesis by cycloheximide results in rapid disappearance cyclin D1 in several cell lines [19]. According to the current model for cyclin D1 posttranslational regulation and our data presented above, cytoplasmic cyclin D1 would need to rapidly enter the nucleus, where its phosphorylation and subsequent ubiquitylation would target it for nuclear export and degradation within the cytoplasm [12, 20]. We thus wished to determine, if we could induce the predominant nuclear localization of cyclin D1 in cycloheximide treated cells by simultaneously inhibiting nuclear export and degradation of the cyclin. Co-culture of asynchronous MCF-7 cell populations with 50 μM cycloheximide for 2 h resulted in the loss of cyclin D1 in both fractions (Figure 5A). This effect could be abolished by MG132 (50 μM), leading to the stabilization of the cyclin in both fractions. Co-culture of cycloheximide treated cells with MG132 and LMB (an inhibitor of CRM1), lead to further increases in the level of cyclin D1 within both fractions but not in the predominant nuclear localization of the protein. Similar treatments in SK-UT-1B cells did not affect the levels or localization of cyclin D1 (Figure 5B). Inhibition of GSK3 activity by SB216763 resulted in modest nuclear accumulation of cyclin D1 in MCF-7 cells (Figure 5C). Our data thus clearly demonstrates, that i) cyclin D1 localizes predominantly to the cytoplasm in all cell lines and tissues examined and ii) that the nuclear exclusion/cytoplasmic sequestration of the cyclin as opposed to its accumulation appears to be the rule. Cyclin D1 downregulation has been shown to be a prerequisite for S-phase progression [2]. Our findings thus offer a novel explanation for the continued proliferation of cells that are either defective in cyclin D1 downregulation [12, 13], or express variants lacking the C-terminal Thr-286 residue required for nuclear export and degradation within the cytoplasm [21]. The failure of cyclin D1 to relocalize to the nucleus when both its degradation and nuclear export are inhibited, suggests that the nuclear import of this protein is inhibited in these cell lines.