|Home | About | Journals | Submit | Contact Us | Français|
D-type cyclins form complexes with cyclin dependent kinases (CDK4/6), and promote cell cycle progression. Although cyclin D functions appear largely tissue specific, we demonstrate that cyclin D3 has unique functions in lymphocyte development and cannot be replaced by cyclin D2, which is also expressed during blood differentiation. We show that only combined deletion of p27Kip1 and Rb is sufficient to rescue the development of Ccnd3−/− thymocytes. Furthermore, we show that a small molecule targeting the kinase function of cyclin D3:CDK4/6 inhibits both cell cycle entry in human T cell acute lymphoblastic leukemia (T-ALL) and disease progression in animal models of T-ALL. These studies identify unique functions for cyclin D3:CDK4/6 complexes and suggest potential therapeutic protocols for this devastating blood tumor.
D-type cyclins (D1, D2, and D3) bind cyclin dependent kinases 4 and 6 (CDK4/6), and the activity of cyclin D:CDK complexes promotes entry into cell cycle (Sherr, 1995; Sherr and Roberts, 2004). Cyclin D:CDK complexes are believed to promote cell cycle progression through at least two functions: by interacting with cell cycle inhibitors such as p21Cip1 and p27Kip1 and by the phosphorylation of the retinoblastoma tumor suppressor (Rb). Cyclin D:CDK4/6 are thought to form ternary complexes that bind cyclin dependent kinase inhibitors (CDKIs) of the p21Cip/p27Kip1 family (Sherr and Roberts, 2004). This facilitates downstream cyclin E:CDK2 complex activity that, along with cyclin D:CDK4/6, inactivates Rb and allows activation of E2F transcription factors and progression through the cell cycle.
The functions of D-type cyclins have been studied using germ-line gene deletion. Each knock-out mouse was viable, but displayed distinct tissue-specific defects (Ciemerych et al., 2002; Kozar et al., 2004; Sicinska et al., 2003; Sicinska et al., 2006; Sicinski et al., 1996; Sicinski et al., 1995). When these deficiencies were combined, complete hematopoietic failure was observed demonstrating the absolute requirement for D-type cyclins within the hematopoietic system (Kozar et al., 2004). Cyclin D2-deficient (Ccnd2−/−) mice display reduced proliferation of mature splenic B cells and lack CD5+ peritoneal B cells (Solvason et al., 2000). Cyclin D3 knock-out (Ccnd3−/−) animals show defects in early B and T cell differentiation, as well as impaired proliferation of granulocytes (Cooper et al., 2006; Peled et al., 2010; Sicinska et al., 2003; Sicinska et al., 2006). Cyclin D1 was recently suggested to play a key role in HSC quiescence and self-renewal (Zou et al., 2011); however, Ccnd1−/− mice do not display striking hematopoietic effects, most likely due to redundancy with D2 and D3 (Sicinski et al., 1995).
Previous work has suggested that defects associated with individual cyclin D deficiency stem from their tissue-specific expression and that D-type cyclins are largely functionally redundant. For example, high expression of cyclin D1 protein, but not D2 or D3, is observed in both the retina and mammary tissue, and Ccnd1−/− animals correspondingly have reduced proliferation of both the cells that contribute to the retina and breast epithelium compartment (Sicinski et al., 1995). Genetic studies in which endogenous Ccnd1 was substituted with Ccnd2 cDNA have demonstrated that cyclin D2 can largely replace cyclin D1 function in mammary and retina tissue development (Carthon et al., 2005). However, these tissues typically express a single D-type cyclin, so whether D-type cyclins can functionally replace one another in cells that express more than one cyclin, such as developing lymphocytes, remains unclear.
Aberrant cell cycle regulation is a common thread to all forms of cancer (Hunter and Pines, 1994). Deregulated expression of all D-type cyclins is frequently observed in hematopoietic malignancies (Bergsagel et al., 2005; Motokura and Arnold, 1993). We have previously shown that induction of T cell acute lymphoblastic leukemia (T-ALL), a disease caused by transformation of lymphocyte progenitors, requires cyclin D3, as expression of the oncogenic intracellular domain of Notch1 (ICN1) in Ccnd3−/− bone marrow progenitors fails to initiate disease. Consistent with these animal studies, cyclin D overexpression is commonly seen in human T-ALL, with specific cyclin D expression associated with distinct T-ALL subsets (Li et al., 2008; Sicinska et al., 2003). Early thymocyte progenitor (ETP)-ALL is characterized by cyclin D2 overexpression (Coustan-Smith et al., 2009), whereas more mature forms of T-ALL are associated with D3 overexpression (Joshi et al., 2008; Li et al., 2008). Finally, previous data have suggested that Notch signaling directly regulates cyclin D3 expression, and blocking cyclin D3 expression by γ-secretase inhibition of Notch signaling prevents cell cycle progression in human T-ALL cell lines in vitro (Joshi et al., 2008). These data suggested that D-type cyclins and/or their downstream interacting partners could be attractive therapeutic targets in this type of disease.
We have previously shown that cyclins D2 and D3 are both expressed during early stages of lymphocytic differentiation; however, only loss of cyclin D3 leads to significant effects on cell differentiation (Cooper et al., 2006; Sicinska et al., 2003). To genetically test the ability of cyclin D2 to replace cyclin D3 function, we generated mice in which Ccnd2 cDNA was targeted to the Ccnd3 locus, such that Ccnd2 was regulated by the Ccnd3 5′ and 3′ UTR (Figure S1). The unique Ccnd3/2 transcript generated from the knock-in allele was not detected in wild type, Ccnd2−/−, or Ccnd3−/− cells using qPCR analysis (Figure 1A). This unique transcript was specifically produced in Ccnd3+/Ccnd2-Neo lymphocytes at low levels, but deletion of the neomycin resistance cassette resulted in a significant increase in mRNA expression in Ccnd3+/Ccnd2 cells. Analysis of total Ccnd2 mRNA showed comparable expression in Ccnd3+/Ccnd2 cells to that of wild type cells. As expected, Ccnd2 transcripts were decreased in Ccnd2+/− thymocytes and diminished in Ccnd2−/− cells (Figure 1B). To further confirm knock-in allele expression, we generated Ccnd2−/−Ccnd3+/Ccnd2 animals and analyzed cyclin D2 expression in lymphocytes. Although the protein was not detected in Ccnd2−/−Ccnd3+/Ccnd2-Neo cells that retained the neomycin resistance cassette, cyclin D2 was readily detected in cells from knock-in animals (Ccnd2−/−Ccnd3+/Ccnd2) that had deleted the selection cassette (Figure 1C). Furthermore, cyclin D2 expression was increased in Ccnd3Ccnd2/Ccnd2 (hereafter Ccnd3D2/D2) cells compared to both wild type and Ccnd3−/− cells, demonstrating that cyclin D2 protein was specifically generated from the knock-in allele.
Given the role of cyclin D3 in early thymopoiesis (Sicinska et al., 2003), we investigated T cell development in Ccnd3D2/D2 animals (Figure 1D). We found that thymus size and total thymus cellularity of Ccnd3D2/D2 animals was reduced similarly to that of Ccnd3−/− mice, and both were significantly decreased from those of controls. Loss of cyclin D3 is associated with an increase in the percentage of CD4−CD8− double negative (DN) thymocytes and a corresponding decrease in CD4+CD8+ double positive (DP) cells (Sicinska et al., 2003); although the overall number of DN cells was not altered, the number of Ccnd3−/− DP was significantly reduced. We found that the total number of Ccnd3D2/D2 DN and DP cells was similar to that observed in Ccnd3−/− animals (Figure 1E). Although loss of cyclin D3 did not affect the absolute number of DN cells generated, the percentage of DN cells was increased while the percentage of DP cells was decreased compared to controls (Ccnd3−/− DN 10.2% and DP 62.4%; control DN 2.3% and DP 83.8%). As observed in Ccnd3−/− thymocytes, the percentage of Ccnd3D2/D2 DN cells was also increased and the percentage of DP cells was reduced compared to controls (Figure 1F). Finally, we quantified proliferation of Ccnd3D2/D2 pro- and pre-T cells, which was comparable to that observed in Ccnd3−/− pro- and pre-T cells (Figure 1G); however, both of these populations showed a decrease in rates of proliferation in comparison to wild type cells. Even with high cyclin D2 expression, cyclin D:CDK4/6 specific phosphorylation of Rb (S807/811) was reduced in Ccnd3D2/D2 thymocytes compared to wild type and was similar to that observed in Ccnd3−/− cells (Figure 1H). These data demonstrated that in early Ccnd3D2/D2 thymocytes, despite the abundance of cyclin D2, Rb remained hypo-phosphorylated and the cells could not efficiently pass into S-phase.
In addition to its role in T cell development, cyclin D3 is also required for early B lymphopoiesis (Cooper et al., 2006). We found that the overall bone marrow cellularity in Ccnd3D2/D2 animals was reduced to that found in Ccnd3−/− mice (Figure S1), while control bone marrow contained a significantly higher number of cells. Furthermore, knock-in animals had a decreased number of pre-B cells compared to controls, and knock-in pre-B cells had an impaired ability to proliferate compared to controls. Collectively, these results indicated that cyclin D2 expression could not rescue the defects in lymphocyte development caused by the lack of cyclin D3.
Although Ccnd2 expression from the Ccnd3 locus did not rescue the defect during the normal development of Ccnd3−/− early lymphocytes, we wanted to test its ability to replace the requirement for cyclin D3 in oncogenic transformation, specifically in T-ALL (Sicinska et al., 2003). We selected a Notch-driven model of T-ALL as more than 90% of human T-ALL shows signs of constitutive Notch pathway activation and the majority of human T-ALL lines are addicted to Notch1 function (Palomero et al., 2007; Weng et al., 2004). We transduced lineage−c-Kit+ bone marrow progenitors from wild type and Ccnd3D2/D2 mice with retrovirus encoding the constitutively active intracellular domain of Notch1-IRES-EGFP (ICN1-EGFP) as previously described (Espinosa et al.; Vilimas et al., 2007; Walkley et al., 2005). These cells were transplanted into lethally irradiated congenic wild type mice, and animals were monitored for the presence of CD4+CD8+ DP leukemic cells in the peripheral blood. At 2 weeks after transplant ICN1-EGFP+ leukemic cells were detected in the blood of control animals. Furthermore, one month post-transplant, the peripheral blood of control animals contained a significant number of ICN1-EGFP+ cells, of which 71% were DP cells, while only a minute number of ICN1-EGFP+ cells were observed in recipients of Ccnd3D2/D2 cells (Figure 2A). The development of disease in hosts that received wild type cells transduced with ICN1 retrovirus was rapid, with all control animals succumbing to the disease 4–6 weeks after transplant (Figure 2B). In contrast, animals that received transduced Ccnd3D2/D2 cells displayed significant protection from disease, with all animals remaining disease-free for the entire 6-month period of observation. These results demonstrated a specific requirement for cyclin D3, but not cyclin D2, in the induction of T-ALL.
Considering that Ccnd2 knock-in into the Ccnd3 locus did not rescue the specific requirements for cyclin D3, we further investigated the regulation of expression of these molecules in early lymphocytes. We readily detected cyclin D2 protein in purified wild type pro-T (CD4−8−25+44−); however, this expression decreased in wild type pre-T (CD4−8−25low/neg44−) cells (Figure S1). However, in Ccnd3−/− cells, cyclin D2 protein was significantly increased in pro- and pre-T cells as well as total thymocytes. This finding suggested that “physiological” cyclin D2 over-expression was not able to rescue the developmental block caused by cyclin D3 deficiency, in agreement with our observations from the knock-in described in Figure 1. Cyclin D2 over-expression appears to be post-transcriptional, as no differences in Ccnd2 mRNA levels were detected from wild type and Ccnd3−/− total thymocytes (Figure S1). Furthermore, immunofluorescence analysis of cyclin D2 and D3 protein from purified DN wild type thymocytes confirmed previous findings as it showed preferential induction of cyclin D2 protein in Ccnd3−/− thymocyte progenitors (Figure S1). Cyclin D3 expression remained unchanged in response to Ccnd2 deletion (not shown and Figure S1). To further investigate regulation of cyclin D2 protein, we treated thymocytes with cycloheximide, which inhibits de novo protein synthesis, and analyzed expression over time. Cyclin D2 half-life was significantly increased in Ccnd3−/− cells (2.5–4 fold) when compared to wild type thymocytes (Figure S3d). We also observed that cyclin D2 protein was stabilized upon treatment with the proteasome inhibitor MG-132, suggesting regulation by the ubiquitin-proteasome system (Figure S1). These studies further support our previous hypothesis as they demonstrate that “physiological” upregulation of cyclin D2 protein is unable to “rescue” normal lymphocytic differentiation and induction of leukemia.
We next sought to further define the functions of cyclin D3 in regulating cell cycle during early lymphocyte development. Of the downstream cell cycle regulators, we focused on the cell cycle inhibitor p27Kip1 as mice deficient for p27Kip1 show increased thymic cellularity (Fero et al., 1996; Nakayama et al., 1996), and p27Kip1 is the only CDKI dynamically regulated at early stages of T cell development (I.A. unpublished data). To genetically test the interaction between cyclin D3 and p27Kip1, we crossed Ccnd3−/− and Cdkn1b−/− mice. Analysis of cyclin D3 and p27 protein from total thymocytes of Ccnd3−/−Cdkn1b−/− mice confirmed the genotypes of the mice (Figure 3A). We measured total thymic cellularity and found that the average number of Ccnd3−/−Cdkn1b−/− cells was significantly increased compared to that of Ccnd3−/− thymocytes, but remained significantly reduced compared to wild type controls (Figure 3B). Differences in total thymocyte numbers were associated with differences in percentages and average number of DP cells (Figure 3C,D). While Ccnd3−/−Cdkn1b−/− animals displayed significantly higher percentages and numbers of DP thymocytes than Ccnd3−/− animals, they had significantly fewer DP cells than control mice. Taken together, these observations suggested that ablation of p27Kip1 only partially restores development of Ccnd3−/− thymocytes.
We have previously shown that the Ccnd3−/− T cell defect stems from a reduction in S-phase entry and cell cycle progression (Sicinska et al., 2003), thus we next assessed the cell cycle status of Ccnd3−/− Cdkn1b−/− early lymphocytes. We evaluated cell cycle using DAPI and observed an increase from 11.6% to 17.5% of pre-T cells in S-G2-M phases of cell cycle from Ccnd3−/− and Ccnd3−/−Cdkn1b−/− animals respectively (Figure 3E). In contrast, the percentage of Ccnd3−/−Cdkn1b−/− proliferating pre-T cells was still reduced compared to controls. To further investigate cell cycle, we measured CDK2-associated activity using an in vitro kinase assay. CDK2-containing complexes immunoprecipitated from Ccnd3−/− total thymocytes showed little kinase activity towards exogenous Rb (Figure 3F). In contrast, Ccnd3−/−Cdkn1b−/− thymocytes showed an increase in CDK2-associated activity, but this activity did not reach the levels of control CDK2 activity. These combined studies indicated that loss of p27Kip1 only partially restored the cell cycle and developmental defects of early Ccnd3−/− T cells.
We also tested the ability of Ccnd3−/−Cdkn1b−/− cells to be transformed by NOTCH activation. Peripheral blood analysis 2 weeks after transplant revealed a small population of ICN1-EGFP+ DP cells in hosts that received Ccnd3−/−Cdkn1b−/− transduced cells (Figure S2A). Similar to control mice that received wild type ICN1-transduced cells, all recipients of Ccnd3−/−Cdkn1b−/− transduced cells developed disease, albeit with slightly delayed kinetics, and succumbed to disease by 10 weeks post-transplant (Figure S2B). These data show that additional loss of p27Kip1 was sufficient to rescue transformation and progression to disease of Ccnd3−/− cells.
As loss of p27 did not completely rescue Ccnd3−/− lymphopoiesis at the steady-state, we hypothesized that remaining activities of additional key cell cycle regulators could prevent cells from properly entering cell cycle in the absence of cyclin D3. We observed that Rb was expressed and phosphorylated in wild type pre-T cells and that its expression, specifically the cyclin D:CDK4/6-phosphorylated pRb (S807/811) species, was reduced in Ccnd3−/− cells (Figure 4A). To genetically test the importance of Rb regulation we generated cyclin D3/Rb doubly deficient animals by crossing Ccnd3−/− mice to Rb1F/F mice carrying the Mx1-Cre transgene. Administration of the double strand RNA mimic poly(I:C) to these animals induced Cre recombinase expression, mediating deletion of Rb1, and these animals hereafter are referred to as Ccnd3−/−Rb1−/−. Rb was not detectable in total thymocytes from Ccnd3−/−Rb1−/− mice, demonstrating efficient deletion of the floxed Rb1 alleles (Figure 4B).
As with Ccnd3−/−Cdkn1b−/− mice, the thymic cellularity of Ccnd3−/−Rb1−/− animals was significantly increased compared to Ccnd3−/− animals, yet significantly decreased compared to controls (Figure 4C). Ccnd3−/−Rb1−/− animals displayed trends in the percentages and average number of DP cells that were similar to those seen in Ccnd3−/−Cdkn1b−/− mice (Figure 4D,E), indicating that deletion of Rb1 only partially restored Ccnd3−/− early T cell development. We next analyzed cell cycle of Ccnd3−/−Rb1−/− pre-T cells and detected approximately 21% of cells in S-G2-M phases, whereas only 11.5% of Ccnd3−/− pre-T cells had transitioned beyond G1 (Figure 4F). However, the percentage of Ccnd3−/−Rb1−/− proliferating pre-T cells was significantly lower than that of controls (approximately 30%). We also tested CDK2-associated kinase activity in Ccnd3−/−Rb1−/− total thymocytes by in vitro kinase assay. CDK2-containing complexes isolated from Ccnd3−/−Rb1−/− T cells showed partially restored ability to phosphorylate exogenous Rb compared to Ccnd3−/− thymocytes (Figure 4G). However, the CDK2 kinase activity in control thymocytes was significantly higher, providing an explanation for the only partial restoration of progenitor cell proliferation.
Our genetic analyses showed that neither p27Kip1 nor Rb loss was sufficient to completely rescue Ccnd3−/− developmental defects. Although there could be several alternative explanations for this incomplete rescue (including redundancy with other CDKI or Rb pocket proteins), we hypothesized that concomitant inactivation of p27Kip1 and Rb could provide efficient rescue suggesting that Ccnd3 acts by simultaneously altering pRb activity and binding p27Kip1. We thus generated Ccnd3−/−Cdkn1b−/−Rb1−/− compound mutant animals and analyzed early T cell development. We found that ablation of both p27Kip1 and Rb in Ccnd3−/− animals resulted in a significant increase in total thymocyte numbers, comparable to that of controls (Figure 5A). The number of DN cells in compound mutant animals was also similar to wild type thymuses (Figure 5B). Moreover, Ccnd3−/−Cdkn1b−/−Rb1−/− and control thymuses contained an almost identical number of DP cells (Figure 5B,C). Finally, we measured pre-T cell proliferation. Only 13.0% of Ccnd3−/− cells were in S-G2-M phases (Figure 5D). In contrast, approximately 29% of Ccnd3−/−Cdkn1b−/−Rb1−/− pre-T cells were in S-G2-M phases, a percentage similar to that found in control littermate animals. These results indicated that only simultaneous genetic deletion of p27Kip1 and Rb was sufficient to completely rescue cell cycle progression and the development of Ccnd3−/− early T cells.
Our studies suggested that cyclin D3 has unique functions in lymphocyte development and transformation, likely in conjunction with CDK4/6, to titrate CDKIs and inactivate Rb. We have previously shown that cyclin D3 expression is essential for the induction of Notch-driven T-ALL (Sicinska et al., 2003). However, these studies did not address the potential of cyclin D3:CDK4/6 targeting during disease progression, a question of significant clinical relevance. To address this question we used PD-0332991, a CDK4/6 specific small molecule inhibitor currently in clinical trials for multiple myeloma treatment (Baughn et al., 2006; Marzec et al., 2006; Menu et al., 2008). Initially, to confirm interaction between cyclin D3 and CDK4/6 in T cells, we performed immunoprecipitation of endogenous CDK6 in wild type thymocytes and observed specific interaction with cyclin D3 (Figure 6A). Similar cyclin D3:CDK4/6 interaction was also observed in human T-ALL cell lines. Having confirmed this interaction, we initiated in vitro treatments of mouse and human T-ALL cell lines with PD-0332991. All human T-ALL lines utilized carried NOTCH1 mutations, and the majority was absolutely dependent on Notch activity. Although the mouse T-ALL lines were driven by overexpression of TAL1, they also contained Notch1 truncating PEST mutations (O’Neil et al., 2006). We tested PD-0332991 at both 0.5 and 1 μM and found similar effects in vitro. PD-0332991 treatment efficiently inhibited S-phase entry of all cell lines within 15 hours, leading to accumulation of cells in G0/G1 phases (Figure 6B). The effects of drug treatment were reversible as removal of PD-0332991 led to efficient re-entry in cell cycle (Figure S3). To further expand these studies using primary leukemia samples, we in vitro treated cells isolated from two T-ALL patients. PD-0332991 treatment led to inhibition of cell cycle progression and accumulation in the G0/G1 cell cycle phase in the two T-ALL primary samples (Figure 6C). We also compared primary T-ALL leukemia cells at diagnosis to cells at relapse or after engraftment in immune-deficient mice. We found a strong inhibition in all conditions, despite higher cycling rates in the relapse and xenograft cells, suggesting that PD-0332991 could be efficient in the treatment of relapsed T-ALL (Figure 6C).
To gain a better molecular and biochemical understanding of PD-0332991 function on human T-ALL cells we have used immunoblotting to define expression and activation of known cell cycle regulators. PD-0332991 treatment efficiently suppressed pRb (S807/811) phosphorylation and increased the expression of the p27Kip1 CDKI, both hallmarks of a G0/G1 arrest (Figure 6D). Whole-transcriptome analysis using four human T-ALL lines led to similar results (Figure 6E,F). PD-0332991 treatment suppressed the expression of key mitosis regulators, including E2f2, Ccna2, Skp2, Cdc25a, Ccne2 and Cdt1. PD-0332991 treatment did not affect expression of Ccnd3 or Cdk4/Cdk6 mRNA. Gene set enrichment analysis (GSEA) showed that PD-0332991 treatment led to significant gene expression correlation with gene sets related to cell cycle progression including DNA replication, S-phase entry, and entry into mitosis (Figure 6F). Furthermore, after 4 days exposure to PD-0332991 we observed a significant increase in annexin V expression compared to controls indicating progression to cell death after treatment (Figure 6G). These combined studies suggested that by inducing cell cycle arrest and apoptosis of leukemic cells, PD-0332991-mediated inhibition of cyclin D3:CDK4/6 activity could be an attractive therapy for T-ALL.
To test the ability of PD-0332991 to suppress disease progression in preclinical models of T-ALL we treated animals that received cells expressing potent oncogenic forms of NOTCH1. Upon establishment of disease at week 3, we initiated treatment of leukemic mice by oral administration of PD-0332991 for 10 consecutive days. The effects of the treatment were rapid and significant as all untreated control mice died by week 12 while the majority of PD-0332991 treated mice survived during the period of observation (Figure 7A). Peripheral white blood counts significantly decreased, and ICN1-EGFP+ DP leukemic cells disappeared from the peripheral blood of the vast majority of the treated animals (Figure 7B,C). In contrast, control animals displayed splenomegaly, and analysis of control splenocytes showed that the majority of cells were ICN1-EGFP+ DP leukemic cells (Figure 7D,E). Histologic examination demonstrated a significant reduction of leukemic cell infiltrations in all tissues studied from PD-0332991 treated animals (Figure 7F). Furthermore, ICN1-EGFP+ cells showed increased annexin V expression in PD-0332991 treated animals compared to controls (Figure 7G). T-ALL cells were addicted to CDK4/6 function, as interruption of drug administration led to relapse of the disease (data not shown), suggesting that the disease can be re-initiated by a small number of non-cycling cells with leukemia-initiating abilities.
To extend these studies using a xenograft model, we transplanted lethally irradiated Rag2−/−Il2rg−/− mice with the human T-ALL CEM cell line and administered PD-0332991 3 weeks post-transplant for 14 consecutive days. Peripheral blood analysis at 4 weeks post-transplant showed 55% huCD45+ cells from untreated control mice compared to 6% huCD45+ cells from PD-0332991 treated mice. All untreated control mice died within 30 days of transplant while the majority of treated mice survived during the observation period (Figure S4). Collectively, these studies demonstrated that cyclin D3:CDK4/6 inhibition was able to efficiently suppress T-ALL progression leading to disease regression in vivo.
We demonstrate here that the cyclin D3:CDK4/6 complex has unique functions in the expansion of normally developing T cell progenitors and induction of T cell leukemia. We show that cyclin D2, a D-type cyclin also expressed in developing T cell progenitors, cannot replace cyclin D3. Although we detected cyclin D2 expression specifically generated by the knock-in allele, Ccnd2 is unable to rescue the developmental defects caused by Ccnd3 deficiency. Two additional findings support the notion that cyclin D2 is unable to sustain expansion of thymocyte progenitors. Initially, we found high expression of cyclin D2 protein in pro-T cells; however, the cells are largely quiescent, suggesting that elevated levels of D-type cyclins do not always correlate with high rates of proliferation. Moreover, we were able to demonstrate that in the absence of cyclin D3, D2 expression was significantly elevated. This finding suggested that differentiating progenitors respond to the loss of cyclin D3 by upregulating cylin D2; however, as with Ccnd3D2/D2 genetic replacement, this endogenous upregulation of cyclin D2 fails to rescue the Ccnd3−/− phenotypes. These experiments suggest that the two cyclins have different functions during early hematopoiesis, in agreement with previous reports using distinct tissue systems. Indeed it was demonstrated that D type cyclins could have distinct cellular localizations, and different abilities to bind CDK inhibitors such as p27 and p21 (Tamamori-Adachi et al., 2008). Moreover, biochemical studies showed differential substrate specificity between cyclin D1:CDK4 and cyclin D3:CDK4 complexes (Sarcevic et al., 1997). Finally, other studies suggested differential utilization of the LxCxE Rb-binding motif between Ccnd1 and Ccnd3 (Baker et al., 2005). Although we were able to demonstrate that cyclin D2 is unable to efficiently phosphorylate Rb in the absence of cyclin D3, further studies are required to elucidate the mechanistic differences among the three kinases in hematopoiesis.
In addition to its requirement in normal T cell development, cyclin D3 is required for the induction of T cell acute lymphoblastic leukemia (T-ALL), a disease with rapid kinetics and characterized by increased rates of cell division. The inhibition of T-ALL induction is not due to a complete inhibition of T cell progenitor differentiation as Ccnd3D2/D2 thymuses generate significant number of mature TCRαβ+ cells. Finally, the inability of cyclin D2 to rescue the Ccnd3−/− phenotypes is unlike previous models of cyclin “rescue” experiments (Carthon et al., 2005; Geng et al., 1999). This phenotypic disparity may reflect the requirement for precise regulation of cell cycle in lymphocytes, a cell type expressing more than one D-type cyclin, compared to cell types where a single D-type cyclin is expressed.
Since the Ccnd3−/− defects stemmed from the failure of lymphocyte progenitors to proliferate optimally, we focused on putative functions of cyclin D3 that directly regulate the cell cycle. Using genetic animal crosses, we demonstrated that the inhibitor p27Kip1 is only one of the critical regulators of cell cycle during T cell development. The p27Kip1 inhibitor appears to be a key regulator in early T cell development, as its protein expression is immediately decreased upon expression of the pre-T cell receptor, a prerequisite for both progression of T cell differentiation and induction of T-ALL (Aifantis et al., 2008). It would be interesting to determine the specific functions of p27Kip1 in developing T cells using specific mutants that disrupt the interaction between p27Kip1 and cyclin D:Cdk4/6. While elimination of p27Kip1 is sufficient to allow normal proliferation of Ccnd1−/− mammary and retina cells (Geng et al., 2001), loss of p27Kip1 does not restore optimal proliferation of Ccnd3−/− T cells. Indeed, phosphorylation and inactivation of Rb function by Ccnd3:Cdk4/6 complexes is another critical function, as demonstrated using the Ccnd3−/−Rb1−/− animals. p27Kip1 and Rb are thought to have both cell-autonomous and cell-nonautonomous effects (Chien et al., 2006; Walkley et al., 2007). Although we cannot exclude cell-nonautonomous effects from contributing to the rescue of Ccnd3−/− thymocyte development, we believe there is a cell-intrinsic component as well since ICN1-EGFP transduced Ccnd3−/−Cdkn1b−/− cells were able to be transformed and produce disease after transplant into wild type hosts (Figure S4). Our data provide genetic evidence suggesting that only simultaneous elimination of both layers of regulation sufficiently lowers the threshold for developing progenitors to enter cell cycle.
These data suggested that targeting cyclin D3:CDK4/6 complex function can directly target induction and progression of T cell leukemia and, at the same time, cause minimal hematologic side-effects, based on the phenotypes of animals lacking expression of cyclin D3 or CDK4/6 (Cooper et al., 2006; Malumbres et al., 2004; Rane et al., 1999; Sicinska et al., 2003; Sicinska et al., 2006; Tsutsui et al., 1999). To pharmacologically target this kinase complex we have used PD-0332991, a small molecule selectively inhibiting CDK4/6 function (Fry et al., 2004). PD-0332991 is a pyridopyrimidine which exhibits an IC50 value less than 0.01 μmol/L against cyclin D3:CDK4/6 complexes. PD-0332991 is currently in clinical trials for the treatment of multiple myeloma (Baughn et al., 2006; Menu et al., 2008), advanced adult solid tumors and refractory non-Hodgkin’s lymphoma (Schwartz et al., 2011). The drug has a relatively long half-life and is generally well tolerated with minimal side effects. Our results and studies performed by Yoon et al. now demonstrate that PD-0332991 could be an efficient treatment for pediatric and adult T cell leukemia. Indeed, we were able to show rapid induction of cell cycle arrest in both mouse and human T-ALL cell lines. Furthermore, we show that the treatment results in cell cycle arrest of primary human leukemia cells both at diagnosis and relapse, as well as loss of leukemic cells by apoptosis, and inhibition of disease progression. Previous data have shown that PD-0332991 effectively inhibits cell cycle in a breast cancer cell line and this is dependent on Rb, as chronic loss of Rb eventually leads to resistance to PD-0332991 (Dean et al., 2010). Further studies are required to determine whether Rb is required for PD-0332991 induced cell cycle arrest in T-ALL. PD-0332991 efficiency against relapsed disease is particularly exciting as such leukemia cells are hyper-proliferating, particularly aggressive and usually non-responsive to current therapeutic protocols. Although we have observed that PD-0332991 administration can efficiently suppress disease progression, it most likely does not target putative leukemia-initiating cells as the disease slowly relapses upon discontinuation of treatment. To further potentiate its activity and achieve sustained remission, PD-0332991 could be used in combinatorial treatment protocols together with either conventional chemotherapy or next generation T-ALL targeted therapeutic compounds, including γ-secretase inhibitors (Real et al., 2009) or Velcade (Menu et al., 2008; Vilimas et al., 2007).
For generation of Ccnd3D2/D2 mice, see Supplemental Experimental Procedures. Ccnd2−/−, Ccnd3−/−, and Rb1F/F mice were kindly provided by Piotr Sicinski and Kay Macleod respectively, and genotyped following published procedures (Ciemerych et al., 2002). Rb1F/F mice were subsequently crossed to Mx1-Cre+ mice (Jackson Laboratory). Deletion of floxed alleles was induced by intraperitoneal injection of 20 mg/kg polyI:C. Mice were injected 5 times, once every other day, and analyzed 2 weeks after the last injection. C57BL/6 and Cdkn1b−/− mice were purchased from the Jackson Laboratory. All animals were used between 4–10 weeks of age and housed in the sterile Smilow animal facility at NYU Medical Center (New York, NY). All experiments were performed according to the guidelines of the Institutional Animal Care and Use Committee of NYU Medical Center.
Retroviral supernatant was generated by transfection of HEK293T cells with pMIGR1-Notch1-IC-IRES-EGFP retroviral construct by calcium phosphate method. c-Kit+ bone marrow progenitors from wild type control, Ccnd3Ccnd2/Ccnd2, Ccnd3−/−, or Ccnd3−/−Cdkn1b−/− animals were purified by magnetic selection and cultured in complete OPTI-MEM supplemented with SCF and Flt3 at 50 ng/ml and IL-6 and IL-7 at 10 ng/ml. Cells were incubated with retroviral supernatant plus 8 μg/ml polybrene and subjected to 3 rounds of spinoculation prior to transplant. Prior to transplantation, GFP expression was analyzed by flow cytometry, and 3 × 105 ICN1-EGFP+ cells were injected retro-orbitally into lethally irradiated C57B6.SJL (CD45.1+) hosts. Mice were monitored for symptoms of disease by peripheral blood analysis for the presence of CD4+CD8+ DP cells.
RNA and protein expression analyses were performed using standard molecular biology procedures. Please refer to the Supplemental Experimental Procedures for further details.
The CDK4/6-specific inhibitor PD-0332991 was generously provided by Pfizer Global Research and Development. Human T-ALL lines (CEM, DND41, HPB-ALL, Jurkat, and TAL1) and mouse (#130, #720, and #5146) were treated with 0.5, 1, or 5 μM of PD-0332991, with concentrations of 0.5 and 1 μM yielding similar results in vitro. Total RNA was extracted from PD-0332991 treated cells using the RNEasy plus mini kit (Qiagen). RNA was amplified with the Ovation RNA Amplification System V2 (Nugen) for cRNA amplification and labeling. Labeled cRNA was then hybridized to Human Genome U133 Plus 2.0 GeneChips (Affymetrix) for microarray analysis. Affymetrix gene expression profiling data were normalized with the Robust Multi-array Average (RMA) algorithm using GeneSpring GX software (Agilent). The gene expression intensity presentation was generated with Multi Experiment Viewer software (http://www.tm4/org/mev/). Gene set enrichment analysis was performed using GSEA software (http://www.broadinstitute.org/gsea/) using phenotype as permutation type, 1,000 permutations and signal to noise ratio as metric for ranking genes (Mootha et al., 2003; Subramanian et al., 2005). Gene sets used in the analysis were taken from the MSig database of the Broad Institute (http://www.broadinstitute.org/gsea/msigdb/cards/).
For in vivo studies of PD-0332991 treatment, 3 weeks after transplant of Notch1-expressing cells, transplanted mice received either 150 mg/kg PD-0332991 in 50 mM sodium lactate or vehicle control by gavage daily for 10 consecutive days. 4 weeks after transplant, peripheral blood was analyzed for GFP, CD4, and CD8 expression. Animals were sacrificed 7–8 weeks post-transplant as control mice became moribund. Upon autopsy, tissues were harvested and prepared for FACS and histological analysis. Tissues were fixed for 24 hours in 10% buffered formalin, dehydrated, and then embedded in paraffin. Paraffin blocks were sectioned at 5 μm and stained with hematoxylin and eosin. For xenograft studies, sublethally irradiated Rag2−/−Il2rg−/− mice were transplanted with 5×105 human T-ALL CEM cells. 3 weeks post-transplant, mice were given either vehicle control or 150 mg/kg PD-0332991 by gavage for 14. Peripheral blood was analyzed for huCD45+ cells 4 weeks post-transplant and survival of mice was monitored.
The study was approved by the Institut Universitaire d’Hematologie Institutional Review Board (Saint-Louis Hospital, Paris, France) and informed consent was obtained from the patients. Human and xenografted T-ALL cell samples were thawed in T-ALL medium and seeded onto MS5-DLH1 stromal cells as previously described (Clappier et al., 2011). After 48 hours, blasts were reseeded onto new MS5-DLH1 (6mm multiwell at 106 cells/ml) and treated overnight with 0.5 μM of PD-0332991 inhibitor or solvent (DMSO), followed by 10 μM BrdU pulse for 30 minutes. Bulk cultures were harvested, and stromal cells were removed from the suspension by two rounds of filtration through a 70 μm strainer and panning for 30 minutes. The cell suspension was then incubated with anti-CD45 antibody, fixed and labeled with BrdU antibody.
All data for cell numbers, expression analysis by qPCR, and FACS analysis of cell cycle and apoptosis represent mean±SD. Statistical significance was calculated using Students t-test, with p<0.05 considered significant. Significance of survival differences was determined by logrank test.
While dispensable for proliferation of many tissues, D-type cyclins are absolutely required in the hematopoietic system, specifically for early lymphocyte development. Aberrant expression of D-type cyclins is associated with hematopoietic neoplasms, thus deciphering the functions of cyclin D is critical for the development of strategies to prevent oncogenic cell cycle. We present genetic evidence that combined deletion of p27Kip1 and Rb, which regulate progression through G1 phase of cell cycle, rescue the defect in early T cell development from deletion of cyclin D3. Furthermore, we show that inhibition of cyclin D:CDK4/6 activity abrogates proliferation in T-ALL cell lines and primary human cells as well as progression of disease in animal models of T-ALL.
We would like to thank members of the Aifantis Lab and C.W. Brains for discussions and comments. Dr. J. Zavadil and the NYU Genome Technology Center (supported in part by NIH/NCI P30 CA016087-30 grant) for expert assistance with microarray experiments. The NYU Flow Cytometry facility (supported in part by NIH/NCI 5 P30CA16087-31) for expert cell sorting, the NYU Histology Core (5P30CA16087-31), and the Transgenic Mouse Core (NYU Cancer Institute Center Grant (5P30CA16087-31). I.A. was supported by the National Institutes of Health (RO1CA133379, RO1CA105129, R21CA141399, RO1CA149655, and RO1GM088847). I.A. was also supported by the William Lawrence and Blanche Hughes Foundation, the Leukemia & Lymphoma Society, and The V Foundation for Cancer Research. We are grateful for the support by a Feinberg Lymphoma Grant. L.G. is supported by a grant from the Institute National du Cancer (INCa). T.T is supported by the NYU Molecular and Cellular Biology Training Program. A.S. is supported by NYSTEM institutional NYU Stem Cell Training Grant (C026880). I.A. is a Howard Hughes Medical Institute Early Career Scientist.
The gene expression data from treated T-ALL cell lines can be found at the GEO database (http://www.ncbi.nlm.nih.gov/geo/) using the accession number GSE40635.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.