Cyclin D1 protein is destabilized specifically in S phase in cancer cells
To investigate the mechanism and importance of cyclin D1 proteolysis, we first assessed the expression profile of cyclin D1 during cell cycle progression from quiescence in three normal cell lines (NIH 3T3 & WI-38 fibroblasts, and CCD841 CoN colon epithelium cells) and in three cancer cell lines (HCT 116 and SW480 colon cancers and T98G glioblastomas). Normal cells ( and Fig. S1
[A]) and cancer cells ( and Fig. S1
[B]) were released from quiescence at G0/G1 phase and cell cycle profiles were determined by flow-cytometric cell cycle analyses. In both cell types, cyclin D1 expression gradually increased after re-entry into the cell cycle and reached a maximum at the G1-S transition. In all three normal cell lines, cyclin D1 levels remained constant during S phase ( and Fig. S1
[A]), although we observed a slight decrease in cyclin D1 expression after entry into S phase. This finding is consistent with previous observations 
. In contrast, all three cancer cell lines showed a dramatic reduction of cyclin D1 expression during S phase ( and Fig. S1
[B]). These observations suggest that cyclin D1 turnover is increased during S phase in these cells.
Figure 1 Cyclin D1 is destabilized during S phase through the ubiquitin-proteasome pathway in cancer cells. (A, B) Expression profile of cyclin D1 during cell cycle progression in cells released from quiescence. Normal cells (A) and cancer cells (B) were tested. (more ...)
To confirm this observation, NIH 3T3 mouse fibroblast and HCT 116 colon cancer cells were synchronized at the G0/G1 phase and released from quiescence and subjected to pulse-chase analysis. show pulse-chase analyses on 35
S metabolically labeled cyclin D1 at 9 hrs (NIH 3T3) or 6 hrs (HCT 116), when most of the cells were in G1 phase, and at 21 hrs (NIH 3T3) and 15 hrs (HCT 116) when most were in S phase (). Labeled cyclin D1 levels of were estimated by quantitative scanning (). There was no significant difference in the half-life of cyclin D1 during G1 and S phases in the NIH 3T3 cells. In contrast, there was a significant difference in the HCT 116 cells (S phase T1/2
11.8 min, compared to T1/2
27.5 min in G1 phase). Thus, cyclin D1 is destabilized specifically in S phase in HCT 116 cells.
Cyclin D1 is degraded in the cytoplasm during S phase through the ubiquitin-proteasome pathway in cancer cells
We next determined if cyclin D1 destabilization during S phase was due to increased proteolysis through the ubiquitin-proteasome pathway. We treated HCT 116 and NIH 3T3 cells with MG132, a proteasome inhibitor, for 2 hrs prior to harvesting at each time point during cell cycle progression (). In treated HCT 116 cells, cyclin D1 accumulated significantly in S phase, with no significant accumulation during G1 phase. In contrast, the difference in accumulated cyclin D1 between G1 and S phase in NIH 3T3 cells appeared to be much smaller. These findings suggest that, in these cancer cells, cyclin D1 is destabilized during S phase through the 26S proteasome pathway.
confirm that destabilization of cyclin D1 involves polyubiquitination. In , HCT 116 cells were transfected with (lanes 1–3) or without (lane 4) HA-tagged ubiquitin cDNA and then synchronized to S phase. Cells were treated with MG132 (lanes 3 and 4) or without it (lanes 1 and 2) for an hour. Lysates were immunoprecipitated with a cyclin D1 antibody (lanes 2–4) or control IgG (lane 1) and immunoblotted with a HA antibody. A group of slower migrating bands was detected by the HA antibody exclusively in the anti-cyclin D1 immunoprecipitates in the presence of ubiquitin (lanes 2 and 3) and the reduced mobility bands were enhanced further after exposure to MG132 (lane 3), indicating that these bands included polyubiquitinated cyclin D1. These observations confirmed that cyclin D1 is degraded during S phase through the ubiquitin-proteasome pathway in HCT 116 cells. shows that more than 70% of these cells were in S phase.
To identify where cyclin D1 degradation is increased during S phase, we extracted nuclear (N) and cytoplasmic (C) protein from cell lysates collected in . Histone H1 was exclusively detected in the nuclear fraction, whereas MEK1 was totally expressed in the cytoplasmic extract, suggesting that we successfully fractionated cell lysates (Fig. S1
[C]). The majority of cyclin D1 was localized in the cytoplasm (Fig. S1
[C]). Nuclear and cytoplasmic extracts were immunoprecipitated with antibodies to cyclin D1 () or IgG (lane 3) and immunoblotted with a HA antibody. Polyubiquitinated cyclin D1 bands were predominantly detected in the cytoplasmic extracts. Furthermore, inhibition of nuclear-to-cytoplasmic localization of cyclin D1 with Leptomycin B (LMB) did not enhance these bands significantly in the nucleus (Fig. S1
[D]). We conclude that cyclin D1 is degraded in the cytoplasm specifically in S phase by a proteasome-dependent mechanism in HCT 116 cells.
MAPK interacts with cyclin D1 through a D-domain and phosphorylates Thr286
MAPK activity is elevated in all three cancer cell lines examined (data not shown; [ref.4]
). We investigated the possibility that the Ras/Raf/MEK/ERK MAPK signaling cascade might regulate the phosphorylation of cyclin D1 residue Thr286, which is followed by proline (; [refs. 8]
Figure 2 MAPK phosphorylates cyclin D1 at Thr286, which triggers subsequent ubiquitination. (A) Identification of the D-domain in cyclin D1 (Cyc D1). The illustration of full-length human Cyc D1 shows the region of the D-domain (A.A. 179-193) and the MAPK phosphorylation (more ...)
ERK/MAPK is a proline (Pro)-directed protein kinase 
. It requires a kinase docking site (or D-domain) on its substrate to increase phosphorylation efficiency (; [ref. 27]
). D-domains have been found in various ERK substrates, such as Elk-1, Sap-1, Sap-2, Ets-1 and c-Myc (; [refs. 27]
). We searched for a D-domain on cyclin D1 with Motif Scan software (http://scansite.mit.edu
). Through a series of stringent searches, we identified a highly significant (within 0.041 percentile) D-domain in amino acids 179-193 (), suggesting that the Ras/Raf/MEK/ERK MAPK signaling cascade might be responsible for cyclin D1 phosphorylation.
To test the possibility that purified ERK/MAPK might phosphorylate recombinant cyclin D1, purified ERK2 was surely used to phosphorylate a glutathione S-transferase (GST)-cyclin D1 fusion protein (). Purified ERK2 efficiently phosphorylated GST-full-length wild type (WT) cyclin D1 (). In contrast, ERK2 failed to phosphorylate T286A, a cyclin D1 mutant (), suggesting that Thr286 is the major phosphorylation site of ERK/MAPK. Identical results were obtained in the presence of purified CDK4/6; ERK was able to phosphorylate cyclin D1 at Thr286, not only in the monomeric form but also within cyclin D1-CDK4/6 complexes (data not shown).
To determine if ERK/MAPK requires the D-domain for efficient cyclin D1 phosphorylation at Thr286, we performed in vitro kinase assays using a complete deletion of the D-domain (ΔD) from the GST-C-terminal cyclin D1 fusion protein. This fusion protein retains the MAPK binding site. shows that purified ERK2 effectively phosphorylated WT cyclin D1 (lane 2), but not the T286A and ΔD mutants (lanes 3 and 4).
These results strongly suggest that MAPK interacts with cyclin D1 through its D-domain to phosphorylate Thr286. To confirm this possibility, we performed an immunoprecipitation and immunoblotting (IP-IB) analysis following ectopic expression of Flag-tagged ERK2 with either HA-tagged WT or ΔD cyclin D1 in HCT 116 colon cancer cells (). ERK2 associated well with WT cyclin D1 (lane 2) and poorly with ΔD cyclin D1 (lane 3). To establish the importance of MAPK in the phosphorylation of cyclin D1 at Thr286 in the HCT 116 cells, we transfected the cells with various forms of cyclin D1 expression vectors (). Ectopic expression of cyclin D1 was distinguished from endogenous expression by the reduced mobility of HA-tagged cyclin D1. We analyzed the phosphorylation status of exogenous cyclin D1 expression at Thr286. Cyclin D1 phosphorylation was significantly reduced by the deletion of the D-domain (lane 4). These observations suggested that the majority of Thr286 phosphorylation in HCT 116 cells is through MAPK activity.
Ras/MAPK-mediated ubiquitination and degradation of cyclin D1 is directly linked to the association of MAPK/ERK with cyclin D1
We next investigated whether MAPK-mediated phosphorylation of cyclin D1 may lead ubiquitination of cyclin D1 in vitro
(). We used a ubiquitination assay system that uses Fraction II HeLa cell extracts as a source of the enzymes necessary to conjugate ubiquitin to substrates and ATP 
. Polyubiquitination of cyclin D1 () was enhanced further by ERK2 ( and ). Slower migrating bands were not detected in the absence of ubiquitin (), suggesting that they consist of polyubiquitinated forms of cyclin D1 (). We believe that polyubiquitination required direct interaction of ERK2 with cyclin D1 and the phosphorylation of cyclin D1 at Thr286, because ubiquitination was largely prevented in the D-domain deletion mutant form (ΔD) and the alanine for Thr286 substitution (T286A) of cyclin D1 ().
The stability of cyclin D1 protein is regulated by ERK/MAPK activities in HCT 116 cancer cells
We also determined the contribution of ERK/MAPK to the stability of cyclin D1 in cancer cells. We performed pulse-chase analysis on metabolically labeled-cyclin D1 after inhibiting MAPK activities (; [refs. 30]
). Exponentially growing HCT 116 cells were treated with U0126 for 30 min, which significantly depleted the phosphorylated form of ERK (pERK) and cyclin D1 at Thr286 (pCyc D1 Thr286) without affecting the cell cycle profile (). Levels of metabolically labeled-cyclin D1 were estimated by quantitative scanning as described above (). Reducing MAPK activity increased the half-life of cyclin D1 from 22.5 min to 54.6 min. These data indicate that the stability of cyclin D1 is regulated by ERK/MAPK activity in cancer cells.
Cyclin D1 stability is regulated through the SCF or the SCF-like pathway
Because of the strict specificity of E3 ligases and their substrates 
, cyclin D1 is likely to have its own E3 ligase. We hypothesized that cyclin D1 proteolysis is mediated by the SCF E3 ligases or an SCF-like complex of E3 ligases, where an F-box protein determines the specificity for its substrate. To test this idea, we performed IP-IB analysis (). Cyclin D1 from exponentially growing HCT 116 cells was immunoprecipitated and sequentially blotted with antibodies to cyclin D1, CDK4, SKP1, CUL1 and CUL7. Cyclin D1 associated with SKP1, CUL1 or CUL7, and CDK4, suggesting that its proteolysis is mediated by the SCF (SKP1-CUL1-F-box protein) or the SCF-like (SKP1-CUL7-FBXW8) complex of E3 ligases.
Figure 3 FBXW8 ubiquitinates cyclin D1 in a Thr286 phosphorylation-dependent manner. (A) IP-IB analysis (left). Protein from exponentially growing HCT 116 cells was precipitated with antibodies to cyclin D1 or IgG. Immunoprecipitates were subjected to SDS-PAGE (more ...)
To test whether levels of cyclin D1 are mainly regulated by the SCF or the SCF-like pathway, we performed an immunoblot analysis 48 hrs after depleting SKP1 expression with siRNA double-strand oligonucleotides in HCT 116 cells (). siRNA for SKP1 significantly reduced SKP1 expression and resulted in accumulation of cyclin D1. These observations strongly support the idea that cyclin D1 stability is regulated through the SCF or the SCF-like pathway.
FBXW8, an F-box protein, specifically associates with cyclin D1 in a Thr286 phosphorylation dependent manner
We next identified the protein responsible for cyclin D1 stability. We tested candidate human F-box protein genes to identify the unique E3 ubiquitin ligase for cyclin D1. Substrate specificity of SCF complexes occurs through protein-protein interaction domains that are often tryptophan-aspartic acid (WD) 40 motifs or leucine-rich repeats (LRR) within F-box proteins 
. We searched the NCBI databases for human F-box proteins with WD40 or LRR motifs. We found approximately 70 potential genes. Among these, 9 had WD40 repeat motifs and 17 had LRR motifs.
We obtained these genes by first performing reverse transcriptase-PCR (RT-PCR) using total RNA from HEK 293, HCT 116 or WI-38 cells. The full-length cDNAs we retrieved were cloned into V5 or Flag epitope tag expression vectors. To address whether any of the products of these genes could recognize cyclin D1, we transiently transfected DNA plasmids for the V5 or Flag-tagged F-box proteins into T98G cells with or without N-terminal HA-tagged cyclin D1 and CDK4 expression vectors (). After 24 hrs, the cells were collected and performed IP-IB analysis. The samples were precipitated with an HA epitope tag antibody and stained with Flag (FBXW7 and FBXL5) or V5 (others), and cyclin D1 antibodies. Panel C shows cyclin D1 associating with two F-box proteins, FBXW8 (lane 7) and FBXL12 (lane 15). FBXW8 possessed WD40 motifs and FBXL12 had LRR motifs.
Because F-box protein substrates must be phosphorylated 
, we tested whether FBXW8 and FBXL12 recognize cyclin D1 in a Thr286 phosphorylation-dependent manner. We transiently transfected T98G cells with V5-tagged F-box protein DNA plasmids and cyclin D1 (wild type or T286A mutant) and CDK4 expression vectors. Samples were precipitated with a HA epitope tag antibody and blotted with V5, HA and Thr286 phosphorylated cyclin D1 antibodies (). FBXW8 was associated with both cyclin D1 wild type and the T286A mutant, but the majority was bound to wild type, which was mostly phosphorylated at Thr286. In contrast, we did not see a significant difference between wild-type cyclin D1 and the mutant in association with FBXL12. These results suggest that FBXW8 recognizes cyclin D1 in a Thr286 phosphorylation-dependent manner, but FBXL12 does not. Consistent with this finding, FBXL12 was not involved in cyclin D1 polyubiquitination in vitro
(). We concluded that FBXW8 may play a role in cyclin D1 stability.
FBXW8 ubiquitinates cyclin D1 in a Thr-286 phosphorylation dependent manner
We investigated whether in vitro
ubiquitination of cyclin D1 requires FBXW8 (). We incubated each in vitro
-translated F-box protein with recombinant GST-cyclin D1 (Cyc D1), fraction II HeLa cell extracts with ATP, ubiquitin and ERK2, and either in vitro
-translated SKP1, RBX1 and CUL1, or SKP1, RBX1 and CUL7 proteins. Next, we blotted them with a cyclin D1 antibody. Cyclin D1 ubiquitination was detected in the combinations of FBXW8 with SKP1, CUL1, and RBX1 (lane 5), or FBXW8 with SKP1, CUL7, and RBX1 (lane 6). However, polyubiquitinated bands were not increased in other combinations. To confirm that the SCF complexes were assembled properly upon in vitro
translation, we performed immmunoprecipitation with each F-box protein in the 35
S-labeled in vitro
translated samples (not shown) and tested whether the complexes containing β-TRCP were functional for polyubiquitination of β-catenin (Fig. S1
[E]). Our results suggest that cyclin D1 ubiquitination involves FBXW8.
We next investigated whether in vitro ubiquitination of cyclin D1 through the SCF-like (SCFL) complex FBXW8 (SKP1-CUL7-FBXW8-RBX1/SCFLFBXW8) requires phosphorylation of cyclin D1 at Thr286 (). Polyubiquitination through SCFLFBXW8 was dramatically reduced by the depletion of ERK2 (lane 2). Furthermore, cyclin D1 polyubiquitination was largely prevented by the alanine-for-Thr286 substitution (T286A, lane 3), suggesting that phosphorylation of cyclin D1 at Thr286 is necessary for ubiquitination by SCFLFBXW8. These data are in good accordance with our observation that FBXW8 specifically associates with cyclin D1 in a Thr286 phosphorylation-dependent manner.
Finally, we reconstituted cyclin D1 polyubiquitination in vitro
using purified E1 and E2 enzymes (). SCFLFBXW8
promotes UbcH5C-catalyzed polyubiquitin chain assembly 
. Consistent with this fact, the V5 immunoprecipitates containing SCFLFBXW8
exhibited significant E3 activities for polyubiquitination of cyclin D1 in the presence of both E1 and E2/UbcH5C (lane 3), and no activity in the absence of E1 or both E1 and E2 (lanes 1 and 2). Taken together, these data indicate that 1) cyclin D1 can be ubiquitinated by FBXW8 and 2) that this process is dependent on Thr286 phosphorylation of cyclin D1 by ERK/MAPK.
Cyclin D1 levels are regulated by FBXW8
We tested whether ectopic expression of FBXW8 reduces levels of endogenous cyclin D1 in cultured cells. We infected HCT 116 cells with retroviruses expressing FBXW8, FBXW7, or GFP as a control ( and Fig. S1
Figure 4 The stability of cyclin D1 is regulated by FBXW8 complexes through the ubiquitin-proteasome pathway. (A, C) IB analysis. HCT 116 cells were infected with a retrovirus expressing FBXW8 (A), a ΔF mutant form (ΔF FBXW8, Panel C) or a control (more ...)
As shows, overexpression of FBXW8 reduced endogenous expression of cyclin D1, but did not significantly change expression profiles of cyclin E. In contrast, FBXW7 inhibited expression of cyclin E, but not cyclin D1 (Fig. S1
[F]; [ref. 32]
). Similar profiles were obtained from SW480, U-2 OS, and T98G cells (not shown).
We also investigated whether overexpression of a dominant-negative form of FBXW8 could cause cyclin D1 accumulation in exponentially growing cells. The F-box deletion ΔF) mutant form of FBXW8 is considered to be a dominant-negative because it can bind to cyclin D1 but barely associates with SKP1, CUL1 and CUL7 (), and therefore does not bring cyclin D1 into the ubiquitin-proteasome pathway. We infected HCT 116 cells with retroviruses expressing ΔF FBXW8 or GFP (). There was significant cyclin D1 accumulation following ΔF FBXW8 expression. In contrast, an ectopically expressed dominant-negative form of FBXW8 did not significantly change levels of cyclin E (). Similar observations were obtained from SW480 and T98G cells (not shown).
We confirmed this finding by determining whether siRNA-induced depletion of endogenous FBXW8 expression could cause cyclin D1 to accumulate in HCT 116 cells (). We treated cycling HCT 116 cells with control or FBXW8 siRNA for 48 hrs. FBXW8 inhibition was verified RT-PCR analysis. We observed approximately 95% inhibition of FBXW8 in comparison to the control sample (see Fig. S1
[G]). We observed significant cyclin D1 accumulation in the sample treated with FBXW8 siRNA () and no effect on cyclin E levels. We concluded that cyclin D1 levels are regulated by FBXW8 in the cancer cells tested here.
The stability of cyclin D1 is regulated through complexes containing FBXW8
FBXW8 associates with CUL1 or CUL7 and forms a complex with SKP1 and RBX1 (; [ref. 22]
), suggesting that CUL1 and CUL7 define the stability of cyclin D1 through FBXW8. Given that depleting FBXW8 from cultured cells increased cyclin D1 levels, reducing CUL1 or CUL7 should give the same result. We treated HCT 116 cells with siRNA for 48 hrs to knock down expression of CUL1 or CUL7 (). In parallel, we used RT-PCR to confirm that the siRNA transfection was working efficiently (Fig. S1
[G]). The siRNAs for CUL1, CUL7, or FBXW8 reduced expression of their respective genes, resulting in accumulation of cyclin D1, which was mostly phosphorylated at Thr286 (). The effect was achieved without affecting MAPK activities (pERK) in the first 48 hrs of siRNA treatment (). Comparable data were obtained from SW480, U-2 OS, and T98 cells (not shown).
To confirm that cyclin D1 accumulation was due to increased cyclin D1 stability, we performed pulse-chase analysis on metabolically labeled-cyclin D1 after using siRNA to deprive HCT 116 cells of FBXW8, CUL1, or CUL7 (). Levels of metabolically labeled-cyclin D1 were estimated as described (). Reducing FBXW8, CUL7 or CUL1 led to a stabilization of cyclin D1. The half-life of cyclin D1 was 27.8 min in control cell, and was extended by FBXW8, CUL7, or CUL1 siRNA treatment (T1/2
79.7, 58.7, or 46.2 min respectively). These data confirmed that accumulation of cyclin D1 through depletion of FBXW8, CUL1, or CUL7 was caused by increased cyclin D1 stability. We conclude that cyclin D1 stability is regulated by complexes containing FBXW8, through the ubiquitin-proteasome pathway.
Increased cyclin D1 degradation in cancer cells is linked to increased cyclin D1-E3 ligase association
We demonstrate that cyclin D1 undergoes increased degradation in the cytoplasm during S phase in a variety of cancer cells. We hypothesized that enhanced cyclin D1 degradation in S phase is associated with an increase in the association of E3 ligase with cyclin D1. To test this hypothesis, we determined the subcellular localizations of FBXW8 and cyclin D1 during the cell cycle in cancer cells (). HCT 116 cells were transfected with V5 epitope-tagged FBXW8. The cells were rendered quiescent 24 hours later by serum starvation for a period lasting a further 24 hours, and then stimulated by adding medium containing serum. This process allowed synchronization of cell cycle progression. Cell cycle profiles were determined by flow cytometry (). At times corresponding to the G1 and S phases, cells were fixed and labeled with fluorescent V5 epitope tags and cyclin D1 antibodies (). The majority of FBXW8 was expressed in the cytoplasm during G1 and S phase. In contrast, cyclin D1 accumulated in the nucleus during G1 phase and exited into the cytoplasm in S phase, in agreement with previous reports 
. The separation of FBXW8 and cyclin D1 during G1 phase suggested that ubiquitination and degradation of cyclin D1 is prevented in G1. Conversely, their colocalization during S phase demonstrated that cyclin D1 proteolysis could be increased in the cytoplasm as cells proceed into S phase ().
FBXW8-mediated cyclin D1 degradation in the cytoplasm is required for cancer cell proliferation
A recent report suggested that cyclin D1 degradation is necessary for efficient DNA synthesis in NIH 3T3 cells 
. We examined the biological significance of enhanced cyclin D1 degradation in the cytoplasm during S phase in HCT 116 cells. We inhibited cyclin D1 proteolysis in the cytoplasm by using siRNA to knock down E3 ligase components such as FBXW8, CUL1, or CUL7 and counted cells for five days (). siRNA for FBXW8, CUL1, or CUL7 significantly reduced cell numbers. These data indicate that rapid turnover of cyclin D1 is required for HCT 116 proliferation.
Figure 5 Cyclin D1 degradation in the cytoplasm is essential for cell proliferation. (A) Viable HCT 116 cells after siRNA-mediated knockdown of FBXW8, CUL1, or CUL7 expression. siRNAs were transfected on days 0, 1, 2, and 4. Cells were collected as indicated, (more ...)
A recent study demonstrated that cytoplasmically expressed cyclin D1 competes with nuclear cyclin D1 and translocates CDK4 from the nucleus to the cytoplasm 
. The consequence is a growth arrest. We tested whether reduced cell proliferation due to knocking down FBXW8 expression is caused by cyclin D1 accumulation and subsequent cytoplasm sequestration of CDK1 (). We treated HCT 116 cells with control (Cont) or FBXW8 (W8) siRNA for 48 hrs. Inhibition of FBXW8 expression was verified by a RT-PCR. We subsequently fractionated nuclear and cytoplasmic proteins. shows that depleting FBXW8 caused cytoplasmic cyclin D1 accumulation. Cyclin D1 was mostly phosphorylated at Thr286, suggesting that cyclin D1 degradation is linked to enhanced phosphorylation of cyclin D1 by MAPK (see ). This process resulted in relocalization of CDK1 from the nucleus to the cytoplasm. This caused a dramatic reduction of the nuclear CDK1 kinase activities, as assessed by a CDK1-associated Histone H1 in vitro
kinase assay (). These observations indicated that inhibiting rapid turnover of cyclin D1 induced growth arrest.
We examined whether constitutive expression of the nuclear protein cyclin D1 T286A could abrogate the block to cell proliferation caused by siRNA against FBXW8 (). We tested this cyclin D1 mutant because it is resistant to polyubiquitination, and also prevents the nuclear export of cyclin D1 during S phase resulting in constitutive nuclear localization 
. Importantly, this mutant is functional: ectopically expressed T286A assembled with CDK4 in cultured cells and showed similar levels of kinase activities to wild type cyclin D1 (data not shown; [ref. 7]
We generated a cyclin D1 ecdysone-inducible (IND) system in HCT 116 cells. Ponasterone A (Pon A) induced ectopic expression of HA-tagged T286A in physiological levels (). We subsequently counted viable cell numbers for five days () and performed a colony formation assay for 2 weeks () in the presence (+) or absence (−) of Pon A, and control (Cont) or FBXW8 siRNA. Cell colonies were stained with crystal violet. show that ectopically expressed physiological levels of nuclear protein cyclin D1 T286A dramatically rescued cells from growth arrest. Thus, FBXW8-mediated cyclin D1 degradation is essential for proliferation of HCT 116 cells.