Cks1 Regulates cdk1 Expression: A Novel Role during Mitotic Entry in Breast Cancer Cells

The cdk2 inhibitor and pocket protein, p130/Rb2, is abundant in G₀ and in early G₁, and a distinguishing feature of cell cycle arrest induced by the antiestrogen ICI 182780 (fulvestrant) in estrogen receptor–positive (ER+) MCF-7 breast cancer cells is that they accumulate in a “G₀-like” G₁ state characterized by increases in p130/Rb2 (1). The levels of p130/Rb2, and another cdk2 inhibitor, p27^Kip1, are mainly regulated through degradation by ubiquitin-dependent proteolysis and involves the SCF^Skp2-Cks1 ubiquitin ligase complex (2–4). Skp2, the specificity component of the SCF^Skp2-Cks1 ligase, binds to p27^Kip1 in a Cks1-dependent manner (5–7). Recent studies in MCF-7 cells show that forced expression of Skp2 could confer acute resistance to antiestrogens like tamoxifen and ICI 182780 (8, 9). We and others have shown that certain growth factors like heregulin β1 (HRGβ1) could also induce ICI 182780–resistant proliferation in MCF-7 cells (10–12). However, whether this involves alterations in the regulation of SCF^Skp2-Cks1 or the cdk2 inhibitors is unknown.

Targeted deletion of Skp2 in mice leads to increased levels of both p130/Rb2 and p27^Kip1, consistent with a role for Skp2 in the turnover of these cdk2 inhibitors in certain cells (4). Inappropriate expression of Skp2 in G₀ cells also promotes S phase entry concomitant with loss of p27^Kip1 (4, 13, 14). Importantly, Cks1−/− mice share certain similarities with Skp2 knockouts, and p130/Rb2 levels in Cks1−/− cells were much higher than wild-type controls, consistent with its role in p130/Rb2 degradation (4, 7). Cks1 mRNA and protein also fluctuate in parallel with the levels of Skp2 (9, 15, 16). Thus, alterations in Cks1 expression may also provide a regulatory mechanism for the degradation of cdk2 inhibitors and other cell cycle proteins.

Overexpression of Cks1 is also strongly associated with aggressive breast tumors and decreased disease-free and overall survival, and in general, correlates with decreased p27^Kip1 levels (17). Low levels of p27^Kip1 have been shown in a number of studies to be associated with reduced overall and disease-free survival in patients with breast cancer (18–20). In contrast, other studies show that p27^Kip1 is overexpressed in a subset of highly proliferative breast carcinomas, and this was associated with strong expression of cyclin D1 and ER positivity (21, 22). Therefore, it is unclear whether “Cks1-Skp2” represents a good therapeutic target for breast cancer because Cks1-independent mechanisms are also involved in the degradation of p27^Kip1 and p130/Rb2, and also because Cks1−/− mice are viable (although smaller than wild-type), and normal cell types from these mice do traverse the cell cycle despite various abnormalities. Nonetheless, it is possible that unlike normal cells, Cks1 might play essential roles in specific growth pathways in cancer cells and therefore be a target that could be exploited in therapy.

Cks1 has also been shown to play Skp2- and p27^Kip1-independent roles. The Schizosaccharomyces pombe and Saccharomyces cerevisiae Cks1 were initially identified as subunits of cdc2 and cdc28 (budding and fission yeast cdk1), respectively, although they do not directly affect catalytic phosphorylation by cdks (15). Subsequently, Cks proteins were also shown to have roles in stimulating the regulation of xenopus cdk1 kinases Wee1 and Myt1, and the phosphatase cdc25 (23–25). Also, high expression of Cks1 not related to p27^Kip1 levels has been reported in non–small cell lung carcinomas (26). Thus, it is possible that Cks1 is also essential in other cell cycle phases, and in certain mammalian cells, apart from its role in G₁-S.

In this report, we show that depletion of Cks1 in MCF-7–derived cells blocks cell cycle progression induced by both estrogen-dependent and growth factor–dependent pathways. Cks1 depletion not only slows progression through G₁-S, but also blocked their entry into M phase. Cks1 silencing led to a rapid loss of Skp2, concomitant increases in p130/Rb2, a rather progressive accumulation in p27^Kip1, and marked reduction in the level of cdk1, which is essential for M phase entry. To our knowledge, these are the first studies that link Cks1 to the regulation of cdk1 expression. This being rather unexpected, given Cks1's dispensability for normal cell proliferation, suggests that functional abrogation of Cks1 could be an effective and potentially safe therapeutic strategy in addition to adjuvant hormonal therapies in breast cancer.

Chemicals and growth factors. Recombinant human HRGβ1 (residues 176-246 corresponding to the epidermal growth factor–like domain) was from R&D Systems. ICI 182780 (Faslodex, fulvestrant) was a gift from AstraZeneca. Improved modified Eagle's medium was obtained from Mediatech, Inc.

Cell lines and cell cycle studies. MCF-7 cells (American Type Culture Collection) or MCF-7/LacZ, an estrogen-responsive MCF-7–derived line with a stably transfected LacZ which has growth properties similar to parental MCF-7 were used and maintained in improved modified Eagle's medium containing 5% fetal bovine serum (27–29). T47D or ZR75-1 are ER+ breast tumor lines, and were used in certain studies, and were maintained in medium with 10% fetal bovine serum. Human mammary epithelial cells (HMEC; Cambrex), were maintained in mammary epithelial growth medium which contained the growth factors human epidermal growth factor, insulin, and hydrocortisone. For cell cycle studies, cells were exposed to propidium iodide. Samples were then treated with 5 μg/mL of RNase (Calbiochem), and then analyzed on a fluorescence-activated cell sorting (FACS) scan using laser excitation at 488 nm (Becton Dickinson). Cell cycle analysis was done using ModFit (Verity Software House). For quantifying mitotic cells, samples were stained with an antibody specific for phosphorylation at Ser¹⁰ in histone H3, which occurs exclusively during mitosis in mammalian cells.

RNA interference studies. For Cks1 knockdown, four separate small interfering RNA (siRNA) duplexes were screened (Dharmacon, Inc.). Their sense sequences are as follows: no. 1, CGACG AGGAGUUUGAGUAUUU; no. 3, ACCAGAACCUCACAUCUUGUU; no. 4, UCUGAU GUCUGAAUCUGAAUU; and no. 5 CAAAUUUACUAUUCGGACAUU. Cells plated in 60 mm dishes were transfected with either a Cks1-specific siRNA or a control duplex with no known homology to mammalian genes using DharmaFECT reagent (Dharmacon) in serum-free medium. After 24 h of siRNA incubation, cultures were either harvested, or re-fed with the medium and conditions indicated for each experiment, and then harvested at various time points. Harvested cells were either subjected to FACS scans to determine the proportion in each cell cycle phase, or were lysed for immunoblotting analysis. In certain experiments, total RNA was prepared for Northern blot analysis of cdk1 expression.

Immunoblotting. Fifty micrograms of cell lysate was run on Criterion precast SDS-PAGE gels (Bio-Rad). Proteins were transferred onto Transblot nitrocellulose filters (Bio-Rad) by electroblotting. Blots were probed with the following antibodies: Cks1 (Santa Cruz), Skp2 (Zymed), p130/Rb2 (Santa Cruz), pRb (BD PharMingen), p107 (Santa Cruz), p27^Kip1 (BD PharMingen), p21^Cip1 (BD PharMingen), pThr¹⁸⁷-p27^Kip1 (Santa Cruz), cdk1 (Cell Signaling), cdk2 (Santa Cruz), cyclin B1 (Neomarkers), securin (Neomarkers), and actin (clone AC40, Sigma). Enhanced chemiluminescence detection with the Supersignal kit was used for band detection (Pierce).

Quantitative TaqMan PCR analysis of Cks1 expression. Total RNA was isolated from cells, partially synchronized with a 72-h ICI 182780 treatment, at various time points following their release into HRGβ1, estradiol (E₂), or control medium. Cells were lysed using the TRI reagent (Molecular Research Center) and total RNA made according to the manufacturer's protocol. First-strand cDNA was made using the Superscript III kit (Invitrogen) according to the manufacturer's protocol. Aliquots from the cDNA were subject to qPCR analysis for Cks1 and Cks2 expression using 300 nmol/L of 5′ forward primer, 300 nmol/L of 3′ reverse primer, and 100 nmol/L of double-fluorescently labeled minor groove–binding probe. For Cks1, the forward and reverse primers were 5′ ATGTCT GAATCTGAATGGAGG and 5′ TCATTTCTTTGG TTTCTTGGG, respectively, and the probe 5′ CCATGAACC AGAACCTCACA. For Cks2, the forward and reverse primers were 5′ GAAGAGGAG TGGAGGAGACTT and 5′ TTTTGG AAGAGGTCGTCTAAA, respectively, and the probe 5′ TCATGAGCCAGAACC ACATATTCTT. The reactions were carried out in an ABI PRISM TaqMan 7900 HT Sequence Detector (Applied Biosystems) according to the manufacturer's instructions under the following conditions: 40 cycles of 15 s at 95°C and 1 min at 60°C. The comparative C_T method was used to represent the relative expression level of either Cks1 or Cks2 transcripts, at various time points following release into HRGβ1 or E₂ containing medium, with respect to a calibrator sample (0 h time point). The relative expression level is expressed as a unitless number and calculated as 2^−ΔΔT as previously described (30).

Northern blot analysis of cdk1 expression. Total RNA was isolated from cells transfected with either CT siRNA or Cks1 siRNA. Cells were harvested either immediately after completion of the 24-h transfection procedure (i.e., 0 h), or 24 or 48 h subsequent to that. RNA was electrophoresed on agarose-formaldehyde gels, blotted onto nylon membranes, and probed with ³²P-labeled human cdc2 cDNA (excised from plasmid provided by Dr. Clare H. McGowan, Department of Molecular Biology, The Scripps Research Institute, La Jolla, CA).

cdk1 protein turnover studies. The decay of constitutive levels of cdk1 was measured after inhibiting new protein synthesis by adding cycloheximide at a concentration of 75 μg/mL to the cells after the 24-h siRNA application (i.e., 0 h time point). At increasing time intervals after the addition of cycloheximide, lysates were prepared for immunoblotting.

Construction of cdk1 expression plasmid and transfection. A plasmid (pTet 3XHA-cdc2) provided by Dr. Clare H. McGowan, that contained an HA-tagged human cdk1 was excised as an EcoRI fragment and subcloned into a plasmid upstream of an IRES sequence followed by a blasticidin resistance gene previously described by us (31). This plasmid, designated pIIB-HA cdc2 or the control vector pIIB was transfected into MCF-7 cells using Fugene (Invitrogen). Twenty-four hours later, cells from both transfections were retransfected with either CT siRNA or Cks1 siRNA duplexes. Following 24 h of siRNA incubation, cultures were re-fed with 5% FBS medium and then harvested after 48 h. Harvested cells were either subjected to FACS scans to determine the proportion in each cell cycle phase, or were lysed for immunoblotting analysis.

Classical ER function assays using transient ERE-Luc transfections. Cells were cultured in FBS-containing medium and transfected with either Cks1-specific siRNA or a control duplex for 24 h. Following the 24-h siRNA transfection, cells were estrogen-deprived by quick stripping in two changes of charcoal-stripped calf serum (CCS) medium, and transfected with an ERE-Luc construct in the same medium. Cells were then treated with ethanol vehicle or E₂ (10⁻⁸ mol/L) for 24 h, harvested, lysed, and luciferase activities determined.

ICI 182780 decreases Cks1 and Skp2 levels while concomitantly increasing p130/Rb2 in ER+ breast cancer cells. To understand the roles of Cks1 in human ER+ breast carcinoma cells, initial experiments were designed to determine the changes in cell cycle regulatory molecules in response to treatment of MCF-7/LacZ cells to ICI 182780. In cells growing exponentially in 5% FBS, the S phase proportion decreased from ∼32% to reach a minimum of 11% in response to ICI 182780 by 72 h (Fig. 1A and B). Decreases in S phase were mainly compensated for by increases in G₁ phase. Previous studies from other laboratories have shown a rapid decline in cdk2 kinase activity in MCF-7 cells treated with ICI 182780, which precedes the decline in the proportion of actively cycling cells, and is accompanied by a marked accumulation of the cdk2 inhibitor p130/Rb2 (1). Because p130/Rb2 levels are regulated by phosphorylation-dependent proteolysis following its ubiquitylation by the SCF^Skp2-Cks1 ubiquitin ligase, we examined the levels of p130/Rb2, Skp2, and Cks1 following treatment of MCF-7/LacZ cells with ICI 182780 (4, 32). Levels of p130/Rb2 were enhanced by antiestrogen treatment in these cells concomitant with marked decreases in Cks1 and Skp2 (Fig. 1A). Interestingly, although p27^Kip1 is the canonical SCF^Skp2-Cks1 substrate, its increases, unlike p130/Rb2, were modest at best (Fig. 1A).

HRGβ1 induces ICI 182780–resistant S phase entry and increases Cks1 protein and mRNA. We and others have shown that HRGβ1 can induce ICI 182780–resistant growth in MCF-7–derived cells (10–12). Because the cell cycle is the ultimate recipient of all mitogenic signals, we asked whether HRGβ1 could stimulate cell cycle progression in these cells in the continuous presence of ICI 182780. Cells were partially synchronized with a 72-h treatment with ICI 182780 resulting in ∼12% cells in S phase (80% in G₁; Fig. 1B and C). The continued presence of ICI 182780 for another 48 h led to further gradual decreases in S phase cells to ∼8% to 9% (Fig. 1B), which was accompanied by further decreases in Cks1, coupled with p130/Rb2 maintained at high levels (Fig. 1C). The continuous presence of ICI 182780 also led to further substantial decreases in another pocket protein p107 (Fig. 1C). Cell cycle progression by both stimuli markedly increased Cks1 protein, which was accounted for largely by Cks1 mRNA increases (Fig. 1C and D). Although Cks2 is expressed in these cells, changes in its mRNA were not significant (data not shown). Both stimuli also increased Skp2 protein levels moderately but consistently in separate experiments (24 and 48 h lanes, Fig. 1C; data not shown). Interestingly, despite their well-established roles in p27^Kip1 degradation, increases in Cks1 and Skp2 were accompanied by only minor decreases in p27^Kip1 abundance in these cells during cell cycle reentry by both stimuli (Fig. 1C).

Cell cycle progression by both stimuli also led to the appearance of hyperphosphorylated forms of p130/Rb2 and marked decreases in total p130/Rb2 (Fig. 1C). This effect is consistent with the fact that hyperphosphorylation of p130/Rb2 by cyclin D/cdk4/6 complexes in mid-G₁ phase leads it to its ubiquitination and proteolysis (4, 32). Unlike p130/Rb2 levels which were down-regulated, release into either HRGβ1- or E₂-containing medium led to substantial increases in p107 pocket protein (Fig. 1C). S phase entry by either stimulus was also accompanied by the increased appearance of the hyperphosphorylated form of the pRb pocket protein (Fig. 1C). Total pRB levels, unlike p130/Rb2, did not decrease upon S phase entry (Fig. 1C).

Cks1-depletion led to rapid decline in Skp2, reduction in cdk1, accumulation of p130/Rb2, p27^Kip1, and hypophosphorylated pRb. RNA interference was used to determine the consequences of Cks1 loss in MCF-7–derived cells. A 24-h siRNA application of cells with either of four different siRNA duplexes was sufficient to cause a marked reduction of Cks1 protein (Fig. 2A). All four duplexes were roughly equivalent in terms of knockdown. In subsequent experiments, we used siRNAs 1, 4, and 5, and identical results were obtained from the use of these duplexes. We also found that Skp2 levels were also almost undetectable after the 24-h Cks1 siRNA transfection procedure (Fig. 2A). Skp2 has been shown to be autoubiquitinated in the absence of Cks1, which suggests that its stability is closely dependent on Cks1 levels, which would explain this result (33, 34). However, p27^Kip1 levels were almost equal in CT siRNA versus Cks1 siRNA-transfected cells at this point (Fig. 2A). We surmised that because steady-state p27^Kip1 reflects the balance of ubiquitination/turnover versus translation rates, perhaps p27^Kip1 translation in these cells was not rapid enough to exhibit immediate accumulation. Therefore, we assessed the effects of Cks1 depletion at various times after the 24-h siRNA transfection protocol. We found that both Cks1 and Skp2 were undetectable for at least another 48 h after the completion of siRNA transfection, which was accompanied by a gradual accumulation of p27^Kip1 (Fig. 2B).

Unlike p27^Kip1 however, p130/Rb2 exhibited marked increases immediately after the 24-h Cks1 siRNA application (0 h time point, Fig. 2B). Also, Cks1-depleted cells exhibited hypophosphorylated forms of pRb, beginning at 24 h after completion of the siRNA transfection protocol, a likely indirect effect of cdk2 inactivation due to the increases in the two cdk2 inhibitor levels (Fig. 2B). Most interesting however, was the finding that cdk1 levels were markedly diminished in Cks1-depleted cells, although not as rapidly as Skp2. Cks1 depletion also led to gradual decreases in securin, and modest cyclin B1 decreases (Fig. 2B).

Cks1-depleted MCF-7–derived cells initially exhibit slowed G₁ progression, and are later blocked in G₂-M: E₂ or HRGβ1 cannot sustain cycling in Cks1-depleted cells. Because these cells have a doubling time of ∼36 to 38 h, and because some of the effects of Cks1 depletion on cell cycle regulatory proteins were not immediate, we assessed the effects of Cks1 depletion on the cell cycle at 24 and 48 h harvest time points after the 24 h siRNA application (Table S1; Fig. 3A). Both CT siRNA and Cks1 siRNA transfections were initially compared in cells made quiescent with estrogen-deprived CCS versus cells that are rapidly cycling in response to E₂ or HRGβ1 (Table S1; Fig. 3A). The S phase fraction declines after incubation in CCS, and CT siRNA-transfected cells have 12.3% after 48 h in CCS. The S phase fraction increases in medium supplemented with E₂ or HRGβ1, and are 30% and 24%, respectively, at 48 h (Table S1; Fig. 3A). On the other hand, Cks1 siRNA-transfected cells incubated in CCS exhibited an even lower S phase fraction as compared with CT transfectants with marked accumulation in G₁ at the 24-h time point and also a slight accumulation in G₂-M (Table S1). At 48 h posttransfection, Cks1-depleted cells exhibited even higher accumulation in G₂-M, as compared with the 24-h time point (Table S1; Fig. 3A). Even more remarkable was the finding that Cks1 siRNA-transfected cells incubated with E₂ or HRGβ1 also exhibited marked accumulation in G₂-M as compared with CT (Table S1; Fig. 3A).

In parallel, another set of siRNA-transfected cultures were exposed to complete serum (5% FBS), containing the normal estrogen component. Some of these samples received ICI 182780, plus supplements of E₂ or HRGβ1. As expected, cells in S phase in CT siRNA-transfected cultures treated with ICI 182780 gradually decreased (23% at 24 h and 15% by 48 h), with a marked accumulation in G₁ (Table S1). However, cells in Cks1-depleted cultures were markedly accumulated in G₂-M by 48 h in both CT and ICI 182780 treatments (Table S1; Fig. 3A). Similarly, when E₂ or HRGβ1 was added to override the effects of ICI 182780, Cks1-depleted cells exhibited an even higher accumulation in G₂-M as compared with CT siRNA-transfected cells (Table S1; Fig. 3A). When we quantified cells in mitosis, we found that they were significantly reduced in Cks1 siRNA-transfected cells suggesting that Cks1 depletion predominantly results in defect(s) in transit into M phase (Fig. 3B). In sum, these results suggested that Cks1 depletion leads to an initial transient accumulation in G₁. However, Cks1-depleted cells that have exited G₁-S phase experience a more severe G₂-M blockade, which although apparent at the 24-h harvest point, is clearer at the 48-h point as more cells transit into G₂-M phase from the preceding phases. This is consistent with the loss of cdk1 protein in these cells which is delayed with respect to the increases in p130/Rb2.

Reintroduction of cdk1 in Cks1 depleted cells prevents G₂-M accumulation. To assess whether reduction in cdk1 in Cks1-depleted cells played the pivotal role in the G₂-M block, we ectopically expressed cdk1 in MCF-7 cells by transient transfection. Stable cdk1 overexpression is not tolerated by these cells (data not shown). MCF-7 cells were transiently transfected either with a vector expressing a HA-tagged cdk1 cDNA or a CT vector. Cultures from both transfections were then retransfected with either a Cks1-specific siRNA or control duplexes as in previous experiments. Both CT transfectants and cdk1 transfectants exhibited similar decreases in endogenous cdk1 and Skp2, and also exhibited increased p27^Kip1 following Cks1 depletion, although ectopic cdk1 remained unaltered (Fig. 4A). Following Cks1 depletion, CT transfectants also exhibited a >3-fold accumulation of cells in G₂-M. On the other hand, the G₂-M accumulation in Cks1-depleted cdk1 transfectants was substantially reduced, with partial restoration of G₁ and S phases (Fig. 4B). This suggests that despite alterations in other cell cycle proteins, the cdk1 reduction in Cks1-depleted cells is the proximate cause of the G₂-M blockade. These studies were performed with MCF-7 cells in order to show that the effects of Cks1 depletion are not unique to MCF-7/LacZ cells.

Cks1 regulates cdk1 expression at the mRNA level. In unperturbed cycling cells, cdk1 protein remains constant throughout the cell cycle due to a coordinated synthesis and degradation of both cdk1 protein and mRNA (35–37). After mitosis, the translation of cdk1 mRNA is shut off with a concurrent increase in cdk1 mRNA decay. However, the stability of the cdk1 protein made in the previous interphase was relatively high, resulting in its levels remaining constant. Activation of cdk1 transcription at the G₁-S transition in each cell cycle results in the accumulation of newly synthesized protein. To investigate whether cdk1 expression was dependent on Cks1, we isolated total RNA from CT and Cks1 siRNA-treated cells, immediately after the 24-h siRNA application, i.e., 0 h, and at the 24 and 48 h time points after the completion of transfection (Fig. 5A). Interestingly, we find that cdk1 mRNA was drastically diminished concomitant with Cks1 knockdown, and remained very low at the 24 and 48 h harvest time points as well. To test whether the steady-state cdk1 protein turnover is altered after Cks1 depletion, we treated CT siRNA or Cks1 siRNA-treated cells with cycloheximide to block protein synthesis after 24 h of siRNA application and harvested cells at various time points. Although the cycloheximide block approach could affect the degradation machinery itself, it can be concluded that at least under these conditions, the turnover of the total cellular cdk1 pool after Cks1 depletion was not significantly different from the turnover rate in CT siRNA-treated cells (Fig. 5B). The apparent half-life of cdk1 in CT versus Cks1-depleted cells was 7.2 and 7.4 h, respectively. Thus, Cks1 positively regulates cdk1 mRNA levels, and therefore, new cdk1 protein synthesis is blocked when Cks1 is knocked down. This gradually results in the depletion of steady-state levels of cdk1 protein without altering cdk1 turnover.

E₂ or HRGβ1 cannot restore cdk1 levels in Cks1-depleted cells. To further investigate why Cks1-depleted cells are blocked in G₂-M and are incapable of cycling even when exposed to potent mitogens like E₂ or HRGβ1, we examined the levels of cdk1, Skp2, p27^Kip1, p130/Rb2, and securin. We show that, in general, both E₂ and HRGβ1 are unable to negate the effects of Cks1 depletion on all of these variables, except that Cks1 depletion–mediated increases in p130/Rb2 are less prominent in E₂, and particularly, HRGβ1-treated cultures (Fig. 6). Because SCF^Skp2-Cks1–mediated ubiquitination and proteolytic degradation of p27^Kip1 is dependent on its phosphorylation at Thr¹⁸⁷, we also showed that the inability of E₂ or HRGβ1 to degrade it was not due to lack of phosphorylation at this site (Fig. 6). We also show that the effects of Cks1 depletion on cdk1 levels are rather specific because cdk2 levels were not affected significantly (Fig. 6). We also tested whether the effects of Cks1 depletion on cdk1 levels occur in other ER+ cells by knocking down Cks1 in T47-D cells. A 24-h Cks1 siRNA application in these cells led to complete depletion of Cks1 protein with somewhat delayed kinetics as compared with what was previously observed in MCF-7–derived cells (Supplementary Fig. S1). However, in these cells, Cks1 depletion led to a marked reduction in cdk1 levels as well (Supplementary Fig. S1). We also found similar results in ZR75-1, another ER+ line (data not shown). We also showed that Cks1 depletion did not alter basal or E₂-inducible luciferase activities in cells transiently transfected with an ERE-Luc reporter plasmid, nor did it alter ERα protein levels (Supplementary Fig. S2).

Cks1 depletion in HMECs does not alter cell cycle progression. For Cks1 abrogation to be an effective intervention, the selectivity of its effects would be an important issue. We therefore assessed the effects of Cks1 depletion on HMECs. HMECs are grown in a defined medium containing insulin and epidermal growth factor, and in these conditions, they express moderate amounts of Cks1 constitutively (data not shown). However, siRNA transfections are done in serum and growth factor–free conditions for 24 h, and in these conditions, Cks1 levels are low but detectable (0 h time point, Supplementary Fig. S3A). Cks1 siRNA duplexes are effective in these cells and induce marked decreases in expression. After the 24 h transfection, cells are re-fed with growth factor containing defined medium, which causes a rapid increase in Cks1 expression (24 and 48 h lanes, Supplementary Fig. S3A). Cks1 siRNA duplexes are however very effective in inhibiting Cks1 expression. Cks1 depletion in these cells also led to decreases in Skp2 levels as in the tumor cells. Interestingly, cdk1 levels are barely detectable in HMECs, and did not exhibit major alterations following Cks1 depletion (data not shown). Similarly, Cks1 depletion did not alter p27^Kip1 as well, which increases after incubation in serum-free medium, and then returns to lower levels in defined medium. The cell cycle profiles of HMECs after 48 h of Cks1 depletion were also not altered (Supplementary Fig. S3B). These results suggest that Cks1 depletion has cell type–specific effects and that Cks1 depletion in normal cells might not be very deleterious.

A large proportion of breast cancers that are initially estrogen-responsive, eventually progress to exhibit resistance to antiestrogens, including pure antiestrogens such as ICI 182780. HRGβ1, a growth factor ligand for the Her3/4 dimerization partners of the Her2 receptor, has been shown to induce both tamoxifen- and ICI 182780–resistant growth in MCF-7 cells (10–12, 38, 39). We surmised that identifying an essential nexus in cell cycle progression might provide interventional targets for both estrogen-dependent as well as growth factor–dependent proliferation of ER+ breast cancer cells. “Cks1-Skp2” and their downstream pathways seem to be such a nexus. In this report, we have shown that both E₂ and HRGβ1 stimulate S phase progression in ER+ breast carcinoma cells, accompanied by the up-regulation of both Cks1 and Skp2, with concomitant decreases in the p130/Rb2 protein (Fig. 1B and C; Supplementary Fig. S4). However, Cks1 plays important roles not only in the G₁-S phase, but also in the G₂-M phase in these cells.

Cks1 and G₁-S progression. The p130/Rb2 protein in addition to being a cdk2 inhibitor, is also a pocket protein and forms transcriptional repressor complexes with E2F4/5 that inhibit transcription of S phase genes (40). Cks1 was recently shown to play a role in p130/Rb2 ubiquitination and degradation (4). This is consistent with our finding that Cks1 depletion led to significant and rapid increases in p130/Rb2. Cks1 depletion also led to p27^Kip1 accumulation, although with somewhat delayed kinetics. This could reflect either a slower translation rate for p27^Kip1 in these cells or the simultaneous operation of other p27^Kip1-degradative pathways. The accumulation of both cdk2 inhibitors p27^Kip1 and p130/Rb2, and the appearance of hypophosphorylated pRb in Cks1-depleted cells, presumably due to cdk2 inactivation, correlates with the initial G₁ accumulation. The reason for an incomplete arrest in G₁ is unclear, although it is reasonable to speculate that even a complete inhibition of cdk2 activity is not sufficient in breast cancer cells, as is also the case in other cancer cells such as colorectal carcinoma cells (41).

Cks1-dependent regulation of cdk1 and mitotic entry. The block in entry into mitosis induced by Cks1 depletion in the present study is significant. It has been shown that in mammalian cells, cdk1 activity is essential for successful entry into mitosis (42). Although reported nearly 13 years ago, a lesser known fact is that p27^Kip1 is also an inhibitor of cdk1, apart from its well-appreciated inhibitory effects on cdk2 (43). Furthermore, it was also reported recently that in Skp2−/− mouse embryo fibroblasts, there is increased association of p27^Kip1 with both cdk1 and cdk2, resulting in decreased kinase activities of cyclin A-cdk2 and cyclin A/B cdk1 complexes (44). However, our novel finding that Cks1 abrogation almost completely depletes cdk1 protein itself, and that ectopic cdk1 can reverse the G₂-M blockade, suggests that Cks1 regulation of cdk1 expression is an indispensable function in G₂-M in breast cancer cells.

Cks1 could regulate cdk1 at the level of transcription, or during mRNA turnover, or both. An E2F-binding repressor element at the −20 position in the cdk1 promoter binds p130/Rb2-E2F4 complexes and represses transcription. This complex comes off at the beginning of the S phase (45). Because p130/Rb2 levels increase in Cks1-depleted cells, this could contribute to transcriptional repression. However, ICI 182780 treatment of MCF-7 cells also causes an increase in cellular abundance of p130/Rb2-E2F4 complexes (1), but does not decrease cdk1 (this study). This suggests that nearly complete loss of Cks1, as in the siRNA-treated cells, could trigger other mechanisms that are obligatory for cdk1 promoter occupancy and transcriptional repression by these complexes (e.g., nuclear entry of E2F4, complexation with DP-1, etc.). These mechanisms, perhaps coupled with increased cdk1 mRNA turnover, could drastically diminish cdk1 mRNA, and eventually protein, in Cks1-depleted cells.

Tsai et al. showed that Cks1 depletion in H358 lung cancer cells resulted in a decrease in cdk1 activity and a modest accumulation in G₂-M (46). Unlike the present findings, however, they did not find a decrease in cdk1 expression (46). During the transition through cell cycle phases, cdk activity is also regulated by reversible phosphorylation at conserved sites such as Thr^160/161. Cks proteins are predicted to prevent access to this conserved threonine in cdk1/2 to both the activating kinase and inactivating phosphatase (47, 48). It is therefore possible that Cks1 could prevent the phosphorylation of Thr¹⁶¹ on cdk1 in certain cell types and block entry into M phase. It is also possible that decreases in cyclin B1 following Cks1 depletion could also contribute to the decreases in cdk1 activity reported by Tsai et al. and subsequent accumulation of cells in G₂-M (46).

Cks1 regulation of cyclin B1 and securin. In our system, Cks1 depletion led to moderate, although significant decreases in cyclin B1. In contrast, xenopus Cks immunodepletion from egg extracts increases cyclin B1 levels, as a result of decreased activation of the APC ubiquitin ligase, and prevents exit from mitosis (25, 49). Cks1 depletion in MCF-7–derived cells also led to decreases in the steady-state levels of another APC substrate, securin (Figs. 2B and 5). These findings are in contrast to findings in budding yeast in which Cks1 inactivation led to a stabilization of securin (49). Whether premature activation of the APC or subsequent proteasome functions occur following Cks1 depletion in our system, and are secondary to cdk1 decreases, will be tested in the future.

Concluding remarks. Both orthologues, Cks1 and Cks2, which are 81% identical, were expressed in the cells used in our study. Our studies, following specific depletion of Cks1, therefore suggest that Cks2 cannot substitute for certain Cks1 roles in G₂-M in these cells, just as Cks2 cannot substitute for Cks1 in its Skp2-dependent function of p27^Kip1 ubiquitination (50). It is also likely that some effects seen in this study following Cks1 depletion are indirect effects of Skp2 down-regulation, due to its autoubiquitination in the absence of Cks1, accounting for the inability of Cks2 in these cells to override these effects (Fig. 2; refs. 33, 34). In conclusion, our studies show that Cks1 contributes to multiple essential roles during cell cycle progression in ER+ breast cancer cells (depicted schematically Fig. S4). It is remarkable that despite the panoply of pathways activated by them, neither E₂ nor HRGβ1 could bypass the requirement for Cks1 in these cells. This suggests that Cks1 plays certain nonredundant roles in these proliferative pathways in breast cancer cells and could therefore be a target for therapy in addition to adjuvant antiestrogens. Although ICI 182780 reduces Cks1 expression, it does not cause G₂-M arrest, possibly because a near-complete eradication of Cks1 is probably essential for marked reduction of cdk1. Also p27^Kip1 increases in ICI 182780–treated cells are not substantial enough to inhibit cdk1 activity. However, p130/Rb2, which exclusively inhibits cdk2, increases markedly in ICI 182780–treated cells. As a result of time, due to slowed G₁ progression, cells steadily accumulate in this phase instead of arresting in G₂-M. Growth factors such as HRGβ1 can override these effects in part by stimulating Cks1 expression and could stimulate antiestrogen-resistant progression. Because Cks1 is indispensable for cell cycle progression by both ER-dependent and independent pathways in breast cancer cells, unlike its redundancy in HMECs, abrogation of functional Cks1 might be a safe and effective therapeutic strategy even in antiestrogen-resistant ER+ disease.

Grant support: Susan G. Komen Breast Cancer Foundation grant BCTR00-456 (J.V. Thottassery), IR&D grant 1094 (J.V. Thottassery), National Cancer Institute Breast Specialized Programs of Research Excellence Developmental grant (J.V. Thottassery), NIH grant CA50376 (F.G. Kern), Adolph Weil Endowed Chair in Cancer Biology (F.G. Kern), and DAMD 17-01-1-0400 (N.R. Estes).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Clare McGowan for providing the human cdc2 cDNA, and Dr. Michele Pagano for helpful discussions.