CBX7 is a tumor suppressor in mice and humans

The CBX7 gene encodes a polycomb group protein that is known to be downregulated in many types of human cancers, although the role of this protein in carcinogenesis remains unclear. To shed light on this issue, we generated mice null for Cbx7. Mouse embryonic fibroblasts derived from these mice had a higher growth rate and reduced susceptibility to senescence compared with their WT counterparts. This was associated with upregulated expression of multiple cell cycle components, including cyclin E, which is known to play a key role in lung carcinogenesis in humans. Adult Cbx7-KO mice developed liver and lung adenomas and carcinomas. In in vivo and in vitro experiments, we demonstrated that CBX7 bound to the CCNE1 promoter in a complex that included HDAC2 and negatively regulated CCNE1 expression. Finally, we found that the lack of CBX7 protein expression in human lung carcinomas correlated with CCNE1 overexpression. These data suggest that CBX7 is a tumor suppressor and that its loss plays a key role in the pathogenesis of cancer.

CBX7 is a chromobox family protein and a member of the polycomb repressive complex 1 (PRC1) that, together with PRC2, maintains developmental regulatory genes in a silenced state (1–3). Mouse Cbx7 associates with facultative heterochromatin and with the inactive X chromosome, thereby implicating the CBX7 protein in the repression of gene transcription (4, 5). We previously reported that the CBX7 gene is drastically downregulated in thyroid carcinomas and that its expression progressively decreases with malignant grade and neoplastic stage (6). Indeed, CBX7 protein levels decrease in an increasing percentage of cases from benign adenomas to papillary, follicular, and anaplastic thyroid carcinomas. Moreover, restoration of CBX7 expression in thyroid cancer cells reduces growth rate, with retention of cells in the G1 phase, which suggests that CBX7 is critical for the regulation of transformed thyroid cell proliferation (6).

The correlation of the loss of CBX7 with a highly malignant phenotype and a consequent poor prognosis appears to be a general event in oncology. In fact, loss of CBX7 expression has been recently shown to be also associated with increasing malignancy grade in bladder (7), pancreatic (8), breast (9), gastric (10), and colon (11) carcinoma, whereas maintenance of CBX7 expression generally correlates with a longer survival (11). These data suggest that the loss of CBX7 expression may be strictly correlated with the acquisition of invasiveness accompanied by the loss of epithelial features and the acquisition of a mesenchymal phenotype, a process known as the epithelial-mesenchymal transition (EMT). Consistent with this hypothesis, we recently demonstrated that CBX7 is able to positively regulate CDH1 expression, which plays a critical role in maintaining normal epithelial cell morphology, by interacting with histone deacetylase 2 (HDAC2) and inhibiting its activity on the CDH1 promoter (12). The positive correlation between CBX7 and E-cadherin expression has been shown in thyroid (12) and pancreatic (8) carcinomas.

It is noteworthy that, in seeming contrast with our previous data, CBX7 has been described as an oncogene (13). Indeed, it cooperates with c-Myc to produce highly aggressive B cell lymphomas and can initiate T cell lymphomagenesis (3). Moreover, CBX7 extends the lifespan of a wide range of normal human cells and immortalizes mouse fibroblasts by downregulating expression of the Ink4a (also known as Arf) locus (13, 14), and its ablation, by shRNA treatment, inhibits growth of normal cells through induction of the Ink4a locus (14). Therefore, in an attempt to better understand the role of CBX7 in the regulation of cell growth and its potential involvement in carcinogenesis, as well as to define its physiologic function in vivo, we examined the consequences of disrupting the Cbx7 gene in mice.

Here, we report that both heterozygous and homozygous Cbx7-KO mice (Cbx7^+/– and Cbx7^–/–, respectively) showed increased susceptibility to developing benign and malignant liver and lung neoplasias. Consistently, mouse embryonic fibroblasts (MEFs) derived from the Cbx7-KO mice had a higher growth rate and delayed senescence compared with their WT counterparts. The ability of CBX7 to counteract activation of CCNE1 expression by the HMGA proteins probably accounts for some phenotypic features of the Cbx7-KO mice. We also found an inverse correlation between CBX7 and cyclin E expression in human lung carcinomas, which indicates that the loss of CBX7 gene expression plays a relevant role in lung carcinogenesis.

Generation of Cbx7-KO mice. We used gene targeting techniques in ES cells to generate a null mutation at the murine Cbx7 genomic locus. The targeting vector was designed to replace exons 5 and 6 of the mouse Cbx7 gene with a neomycin cassette (Figure 1A). Exons 5 and 6 code for the C-terminal region of the protein, including the Pc-box domain required for its functions. Progeny of chimeric animals were identified by Southern blot analysis of EcoRI-digested tail DNA (Figure 1B), and matings were established to produce mice heterozygous or homozygous for the Cbx7-null allele. To verify the absence of Cbx7 expression in Cbx7^–/– mice, total RNA from the kidney, liver, lung, and MEFs of WT and Cbx7-KO mice was assayed by Northern blot. As expected, Cbx7^–/– mutants did not express the Cbx7 transcript that was present in tissues from WT mice, whereas Cbx7^+/– mutants expressed an intermediate amount of Cbx7 mRNA (Figure 1C). Northern blot did not show any band of 380 bp, corresponding to the potential transcript of exons 1–4 of the Cbx7 gene that are not disrupted by the targeting construct. Moreover, Western blot performed using antibodies able to recognize the 5′ region of the CBX7 protein did not show the expression product of the potential truncated transcript (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI58620DS1). It is likely that this short messenger is degraded and does not code for any protein.

Figure 1

Generation of Cbx7-KO mice. (A) Endogenous WT allele, targeting vector, and resulting KO allele. E, EcoRI; Neo, neomycin; Ex, exon. (B) Southern blot of representative Cbx7^+/+ and Cbx7^+/– ES cell clones. (C) Northern blot analysis of total RNA from kidneys, lungs and MEFs of Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– mice. The lengths of the WT Cbx7 transcript (2,900 bp) and the predicted truncated form after homologous recombination with the Cbx7-KO construct (380 bp) are shown. The absence of a 380-bp signal in both Cbx7^+/– and Cbx7^–/– mice is likely the result of instability of the short transcript. Actb expression analysis of the same blot is shown as control of RNA loaded. (D) Gross appearance of a representative 1-year-old Cbx7^–/– mouse and a sex-matched Cbx7^+/+ sibling. (E) Naso-anal length of cohorts of 20 mice, males or females, was measured at 12 months of age. Values are mean ± SD. *P < 0.05.

Heterozygote matings yielded Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– offspring at roughly the expected Mendelian ratio, indicative of no marked embryonic lethality. Interestingly, Cbx7-KO mice showed a significant increase in naso-anal body length: 70% of females and 46% of males showed average increases of 6.5% (P = 0.041) and 16% (P = 0.044), respectively, compared with their WT counterparts (Figure 1, D and E).

Cbx7^–/– MEFs grow faster and senesce later than do their Cbx7^+/+ counterparts. To investigate the role of CBX7 in cellular proliferation, we analyzed the growth rate and cell cycle distribution of MEFs from Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– embryos at 12.5 dpc. As shown in Figure 2A, growth rate was significantly higher in Cbx7^–/– MEFs than in Cbx7^+/+ controls, and intermediate in Cbx7^+/– MEFs. To assess whether the higher growth rate of Cbx7^–/– MEFs was caused by deranged progression through the phases of the cell cycle, we examined asynchronously growing MEFs by flow cytometry. The number of Cbx7^–/– MEFs was lower in G₁ and higher in S phase of the cell cycle compared with Cbx7^+/+ MEFs (Figure 2B). Consistently, the Cbx7^–/– MEFs transiently transfected with a Myc-Hys–tagged Cbx7 expression vector showed a decreased number of cells in the S phase and an increased cell population in the G₁ phase of the cell cycle compared with backbone vector–transfected Cbx7^–/– MEFs (Figure 2C).

Figure 2

Growth alterations in Cbx7-KO MEFs. (A) MEFs were prepared from Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– embryos at 12.5 dpc. At culture passage 4, they were plated and counted daily for 13 days to extrapolate growth curves. Shown are mean ± SEM of 3 different cell clones (each originating from a different embryo) for each genotype. (B and C) Propidium iodide flow cytometry of asynchronous growing WT and Cbx7-KO MEFs. (B) Percent cells in each phase of the cell cycle (mean ± SEM). *P < 0.05. (C) FACS analysis of Cbx7^–/– MEFs transiently transfected with different amounts of a Cbx7 expression vector or the backbone vector (BV). (D–I) Light microscopy of representative Cbx7^+/+ (D and G), Cbx7^+/– (E and H), and Cbx7^–/– (F and I) MEF clones stained for β-galactosidase activity at culture passages 4 (P4; D–F) and 7 (G–I). (J) Expression of cell cycle inhibitors Cdkn2a, tp53, and Cdkn1a in representative MEFs from each genotype was determined by qRT-PCR at culture passages 4 and 7. *P < 0.05. Passage 4 Cbx7^–/– MEFs transiently transfected with the backbone vector (C) or whose Cbx7 expression had been restored (R) are also shown. (K) Expression of cell cycle and senescence regulators in representative MEFs from each genotype was determined by Western blot at culture passage 7. Normalized protein levels, evaluated by densitometric analysis, are indicated above immunoblots.

We next examined the susceptibility to senescence of the MEFs at different culture passages by measuring senescence-associated β-gal (SA–β-gal) activity. Starting from culture passage 7, SA–β-gal activity was present in WT MEFs, as evidenced by the presence of green cells (Figure 2G). Conversely, SA–β-gal activity was absent from most of the Cbx7-KO counterparts (Figure 2, H and I). These findings suggest a role of CBX7 in the induction of cellular senescence. Senescent MEFs express elevated levels of p21 and p16 consequent to replication and culture stress (15). These 2 genes may cooperate to inhibit Rb phosphorylation and maintain growth arrest in an irreversible state. Therefore, we evaluated mRNA and protein levels of p16, p53, and p21 in WT and Cbx7-KO MEFs by quantitative RT-PCR (qRT-PCR) and Western blot analysis. The mRNA and protein expression levels of p16 (starting from passage 4) and of p53 and p21 (at passage 7) were significantly lower in Cbx7-KO than in WT MEFs (Figure 2K). Consistently, the levels of S15-phosphorylated p53 paralleled those of p21, and both cyclin A and cyclin E were upregulated in Cbx7^–/– versus Cbx7^+/+ MEFs (Figure 2K). Moreover, Cbx7^–/– MEFs in which Cbx7 expression had been restored — but not those transfected with empty vector — showed Cdkn2a and trp53 transcript levels comparable to those observed in WT MEFs, and significantly increased Cdkn1a levels compared with WT (Figure 2J).

Interestingly, MEFs isolated from 2 transgenic mouse lines overexpressing Cbx7 under the transcriptional control of a cytomegalovirus promoter (Supplemental Figure 2A) showed a behavior opposite to that of Cbx7-KO MEFs: a lower proliferation rate (Supplemental Figure 2B) that was Cbx7 dose dependent, with a reduced cell population in the S phase of the cell cycle (data not shown). Consistently, decreased Ccne1, Ccna2, and Ccnb1 expression was observed in MEFs of both transgenic mouse lines compared with the WT MEFs (Supplemental Figure 2A).

CBX7 binds and negatively regulates the CCNE1 promoter. In an attempt to understand the mechanism by which CBX7 negatively regulates the G₁/S transition, we focused on the CCNE1 gene for various reasons: (a) it is crucial for the G₁/S transition during the mammalian cell cycle (16); (b) its promoter contains an E-box that is also present in the region recognized by CBX7 on the CDH1 promoter (12); (c) it is positively regulated by E2F3, whose activity is negatively modulated by chromatin-regulating and -remodeling proteins, including HDAC and PcG proteins (17, 18); and (d) its overexpression is frequently associated with lung cancer (19), the most frequent tumor type in Cbx7-KO mice (see below). Therefore, we asked whether CBX7 could negatively regulate CCNE1 expression and consequently modulate cell proliferation, thereby accounting for the increased cell growth rate of Cbx7-KO fibroblasts. First, we measured Ccne1 expression in MEFs and tissues (kidney and lung) of Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– mice by qRT-PCR. As shown in Figure 3A, Ccne1 expression levels were higher in Cbx7-KO MEFs and tissues than in the WT controls. Western blot analysis yielded the same results (Figure 2K and data not shown). Interestingly, restoration of Cbx7 expression resulted in decreased Ccne1 expression levels in Cbx7^–/– MEFs, comparable to those observed in WT MEFs (Figure 3A). Moreover, Ccne1 expression was reduced in MEFs derived from mice overexpressing Cbx7 in a manner dependent on Cbx7 level (Supplemental Figure 2A). Together, these results support the hypothesis that CBX7 negatively regulates CCNE1.

Figure 3

CBX7-dependent regulation of Ccne1 expression. (A) qRT-PCR on MEFs and tissues from WT and Cbx7-KO mice, including Cbx7^–/– MEFs transiently transfected with the backbone vector (Cbx7^–/–C) or whose Cbx7 expression had been restored (Cbx7^–/–R), to detect Ccne1 expression. (B) EMSA performed with a radiolabeled CCNE1 promoter oligonucleotide incubated with 5 and 20 ng of the recombinant GST-CBX7 (lanes 2 and 3), GST-CBX7-CHROMO (lanes 5 and 6), or GST-CBX7-NOCHROMO (lanes 8 and 9) proteins. Where indicated, a 400-fold molar excess of unlabeled CCNE1 promoter oligonucleotide was added (lanes 4, 7, and 10). Further negative controls were obtained by incubating the recombinant proteins with a radiolabeled GAPDH promoter oligonucleotide (lanes 11 and 12). (C–E) Results of ChIP assays. CDKN2A and CDH1 promoters were used as positive controls, whereas GAPDH promoter and nonspecific IgG instead of anti-CBX7 were used as negative controls. (C) ChIP assay on HEK 293 cells transfected with CBX7-expressing or empty vector for binding of CBX7 to the CCNE1 promoter. (D) ChIP assay on MEFs for binding of endogenous Cbx7 to the Ccne1 promoter. (E) ChIP assay on MEFs for binding of endogenous Cbx7 and Hdac2 to the Ccne1 promoter. (F) CCNE1 promoter–driven luciferase activity, relative to activation of empty vector–transfected cells, in HEK 293 cells. Where indicated, 1 and 2 μg of CBX7 was cotransfected with the cyclin E–luc plasmid. *P < 0.05; **P < 0.01.

We next evaluated binding of the CBX7 protein to the CCNE1 promoter in vitro by EMSA with an oligonucleotide spanning –51 to –12 bp upstream of the transcriptional starting site (TSS) of the human CCNE1 promoter region, including an E2F-responsive element and an HMGA-responsive AT-rich stretch. Increasing amounts (5 and 20 ng) of the recombinant glutathione-S-transferase–CBX7 (GST-CBX7) protein bound to the ³²P-end-labeled double-strand oligonucleotide in a dose-dependent manner (Figure 3B, lanes 2 and 3). Binding specificity was demonstrated by competition experiments that showed loss of binding with the addition of a 400-fold molar excess of the specific unlabeled oligonucleotide (Figure 3B, lane 4). No binding was obtained when an oligonucleotide corresponding to a region of the GAPDH promoter was used as a negative control (Figure 3B, lanes 11 and 12).

To identify the region of CBX7 required for CCNE1 promoter binding, we performed the same experiment with 2 recombinant mutated forms of the CBX7 protein: one containing only the chromodomain (GST-CBX7-CHROMO) and another lacking the chromodomain (GST-CBX7-NO CHROMO). Only GST-CBX7-CHROMO specifically bound to the CCNE1 promoter (Figure 3B, lanes 5–10), which indicates that the CBX7 protein is able to bind the CCNE1 promoter in vitro through its chromodomain.

We next evaluated whether the CBX7 protein could bind to a human CCNE1 promoter region, including the E-box, by performing ChIP assays. HEK 293 cells were transiently transfected with HA-tagged CBX7 expression vector or empty vector. Cells were then crosslinked, and DNA-chromatin complexes were subjected to IP with anti-HA or IgG antibodies. The IP chromatin was subsequently analyzed by quantitative PCR (qPCR), using primers spanning the region of the human CCNE1 promoter (–850 to –700 bp upstream of the TSS) covering an E-box region. As shown in Figure 3C, HA-tagged CBX7 showed co-IP with the CCNE1 promoter in cells transfected with CBX7. Moreover, CBX7 also bound to CDH1 and CDKN2A, which have previously been identified as promoters (12, 20). Conversely, no amplification was observed with anti-IgG precipitates and when primers for the GAPDH promoter were used, indicative of the specificity of CBX7 binding to the CCNE1 promoter.

The same results were obtained when we performed ChIP assays on Cbx7^+/+ and Cbx7^–/– MEFs (Figure 3D) and lung tissues (Supplemental Figure 3), which also indicated that the endogenous CBX7 protein was able to bind the CCNE1 promoter. Furthermore, the binding of Hdac2 to the Ccne1 promoter was reduced in Cbx7^–/– compared with Cbx7^+/+ cells (Figure 3E), which indicates that the presence of Cbx7 favors Hdac2 binding to the Ccne1 promoter. Finally, to assess the functional consequences of CBX7 binding to the CCNE1 promoter, we demonstrated that CBX7 dose-dependently reduced the transcriptional activity of the CCNE1 promoter (Figure 3F).

HMGA1 displaces the HDAC2/CBX7 complex from the CCNE1 promoter. We previously demonstrated that the expression of HMGA proteins increases E2F activity by displacing HDACs from E2F-responsive promoters, including CCNE1 (21). Therefore, we hypothesized that CBX7 influences CCNE1 expression by interacting with HMGA1 and then antagonizing its activity. To test this hypothesis, we transiently transfected HEK 293 cells with CBX7 and HMGA1b expression constructs, subjected protein extracts to IP with anti-CBX7 or anti-HMGA1 antibodies, and immunoblotted them with the reciprocal antibodies. As shown in Figure 4A, coexpression of CBX7 and HMGA1b resulted in reciprocal co-IP of the 2 proteins. Co-IP was performed in the presence of ethidium bromide to exclude that the co-IP of CBX7 and HMGA1b is dependent on contaminating DNA. The same result was obtained when co-IP was performed with the endogenous proteins extracted from WT mouse embryos (Figure 4B). Conversely, no interaction was detected when nonspecific IgGs were used (Figure 4, A and B). To verify this interaction, we carried out a pulldown assay using a GST-CBX7 recombinant protein; as shown in Figure 4C, GST-CBX7, but not GST, was able to pull down HMGA1b. Then, because of the ability of HMGA1b to interact with CBX7 (Figure 4) and of CBX7 to interact with HDAC2 (12), we investigated whether the physical interactions between CBX7 and either HDAC2 or HMGA1b take place on the CCNE1 promoter. We performed ChIP and re-ChIP analysis on HEK 293 cells transiently transfected with HA-tagged CBX7, HMGA1b, or both expression vectors. Re-ChIP experiments were performed on exogenously expressed proteins because of the difficulty in obtaining good results when endogenous proteins are not highly expressed. The cells were crosslinked and subjected to IP with anti-HA antibodies recognizing the exogenous CBX7 protein. The CBX7 complexes were subjected to re-IP with anti-HMGA1, anti-HA, and anti-HDAC2 antibodies and then analyzed by qPCR for CCNE1 promoter amplification using the same primers as above. The results demonstrated that CBX7, HMGA1b, and HDAC2 occupied the same region on the CCNE1 promoter (Figure 5A, top). Moreover, HMGA1b expression led to a reduction of the CBX7/HDAC2 complex levels present at this promoter. The reciprocal experiment, using anti-HDAC2 antibodies for the first ChIP, yielded comparable results (data not shown), whereas the ChIP for HMGA1 showed that CBX7 expression reduced the amount of HMGA1 protein bound to the CCNE1 promoter, suggestive of mutual competition between CBX7 and HMGA1 for the binding to this promoter. Interestingly, the amount of HDAC2 present in the CCNE1 promoter–bound HMGA1b complex increased from 17% to 40% in the presence of CBX7 (Figure 5B). Taken together, these results indicate that CBX7 binds the CCNE1 promoter and participates in the DNA-bound multimeric complex containing HDAC2 and HMGA1. These data also suggest that CBX7 recruits HDAC2 on the CCNE1 promoter, whereas HMGA1 displaces the CBX7/HDAC2 complex from this promoter.

Figure 4

CBX7 interacts with HMGA1b. (A) HEK 293 cells were transfected with both CBX7-V5 and HMGA1b expression plasmids. Cellular lysates were prepared, and equal amounts of proteins were subjected to IP with anti–CBX7-V5, anti-HMGA1, or nonspecific IgG, as indicated. The immunocomplexes were immunoblotted with reciprocal antibodies. As a positive control, 50 μg of transfected cell lysates were separated on the polyacrylamide gel as input. Et.Br., ethidium bromide. (B) Tissue lysates were prepared from WT embryos, and equal amounts of proteins were subjected to IP with anti-HMGA1 antibodies or nonspecific IgG. The immunocomplexes were immunoblotted with anti-CBX7 antibodies, for detection of the co-IP, or with anti-HMGA1 antibodies, for the IP control. As a positive control, 100 μg tissue lysates were used as input. Lanes were run on the same gel but were noncontiguous (white lines). (C) GST-HMGA1b or GST proteins immobilized on glutathione beads were used to bind CBX7-V5 from HEK 293 cells. The filter was probed with the anti–CBX7-V5 antibody.

Figure 5

Competition between HMGA1 and CBX7 for binding, regulation, and acetylation of the CCNE1 promoter. (A) ChIP and re-ChIP assay, revealed by qRT-PCR, on HEK 293 cells transfected with HMGA1b expression vector, HA-tagged CBX7 expression vector, or both to detect the HMGA1/CBX7, CBX7/HDAC2, and HMGA1/HDAC2 interactions on the CCNE1 promoter. GAPDH promoter was also analyzed as a negative control. (B) Percent HDAC2 protein co-IP with HMGA1, as detected in the ChIP and re-ChIP assay in A, comparatively reported in cells transfected with HMGA1b with or without CBX7. **P < 0.01. (C) EMSA, performed as in Figure 3B, using 5 ng GST-CBX7 (lanes 4–8) and 5 and 20 ng His-HMGA1b (lanes 1–3 and 6–8). A 400-fold molar excess of unlabeled probe (lanes 3, 5, and 8) was added as a specific competitor, and 5 ng of a recombinant GST protein (lane 9) was used as a negative control. (D) CCNE1 promoter–driven luciferase activity, expressed relative to activation of empty vector–transfected cells, in HEK 293 cells. Where indicated, 1 μg CBX7 and/or 1, 2, and 5 μg HMGA1b expression vectors were cotransfected with the cyclin E–luc plasmid. (E) ChIP assay, revealed by qRT-PCR, on HEK 293 cells transfected with CBX7 expression vector, HMGA1b expression vector, both, or empty vector for binding of acetylated histone H4 (AcH4) to the CCNE1 promoter. Equal amounts of proteins were subjected to IP with anti-AcH4 antibodies or nonspecific IgG, as indicated.

HMGA1 competes with CBX7 for binding and regulation of the CCNE1 promoter. To investigate possible functional interactions between HMGA1b and CBX7 on the CCNE1 promoter, we first analyzed the in vitro binding of both proteins to this promoter by EMSA, using the GST-CBX7 protein with increasing amounts of a histidine-tagged HMGA1b (HIS-HMGA1b) recombinant protein. Both proteins specifically bound to the oligonucleotide used as probe (Figure 5C, lanes 1–5). Importantly, the GST-CBX7/DNA complex was dose-dependently inhibited by HIS-HMGA1b (Figure 5C, lanes 6 and 7), which confirmed the competition of these proteins for binding to the CCNE1 promoter. Consistently, analysis of CCNE1 promoter activity in the presence of CBX7 and increasing amounts of HMGA1b demonstrated that HMGA1b counteracted the negative activity of CBX7 on the CCNE1 promoter (Figure 5D), thereby causing a switch from repression to activation.

CBX7 expression results in increased histone acetylation of the CCNE1 promoter. Chromatin remodeling and histone modifications have recently emerged as the main mechanisms whereby gene expression is controlled. Since the link between DNA methylation and histone deacetylation in gene silencing is well established (22–26), we evaluated the lysine acetylation of histone tails at the CCNE1 promoter. HEK 293 cells were transfected with V5-tagged CBX7 expression vector (referred to herein as CBX7-V5) or with empty vector, cells were crosslinked, and DNA-chromatin was subjected to IP with anti–H4 acetylated or anti-IgG antibodies. IP chromatin was analyzed by qPCR with primers spanning the CCNE1 promoter region. As shown in Figure 5E, the amounts of H4 acetylated tails were lower in the CCNE1 promoter of CBX7-transfected cells than in backbone vector controls, which suggests that CBX7 decreases histone acetylation. No IP occurred with anti-IgG precipitates. Moreover, CBX7 expression competed with the H4 acetylation–increasing activity of HMGA1b. Our data therefore suggest that CBX7 binds the CCNE1 promoter and inhibits CCNE1 transcription, counteracting the enhancement of CCNE1 transcription by HMGA1b and modifying histone acetylation at its promoter.

Cbx7-KO mice develop liver and lung neoplasias. Neither Cbx7^–/– nor Cbx7^+/– mice showed any evidence of illness up to 12 months of age. Conversely, histopathological analysis revealed the presence of solid tumors in aged (17- to 22-month-old) Cbx7^+/– and Cbx7^–/– mice, but not in their Cbx7^+/+ littermates (Figure 6A). These tumors included lung and liver neoplasias, ranging from adenoma to carcinoma (Figure 6B). In the case of liver neoplasias, an equal number of benign lesions was present in Cbx7^+/– and Cbx7^–/– mice, but there was a higher percentage of hepatocellular carcinomas in Cbx7^–/– than in Cbx7^+/– mice (32% vs. 14%; Table 1 and Figure 6C). Lung adenomas and adenocarcinomas occurred in Cbx7^+/– and Cbx7^–/– mice (28% and 24%, respectively); however, adenomas were detected only in Cbx7^+/– mice, whereas carcinomas were prevalent in Cbx7^–/– mice (Table 1 and Figure 6D). Immunohistochemical and qRT-PCR analyses of hepatocellular and lung carcinomas in Cbx7^–/– mice confirmed the absence of Cbx7 expression and showed cyclin E overexpression (data not shown). We assessed Cdkn2a and Cdkn2d levels qRT-PCR in the lung carcinoma samples and found increased Cdkn2a expression in the neoplastic tissues, in contrast to no clear differences in Cdkn2d expression observed between normal and carcinoma lung tissues (Supplemental Figure 4). This result is consistent with our previously published results showing that p16 is moderately overexpressed in papillary thyroid cancer (where CBX7 is moderately downregulated) and highly overexpressed in anaplastic thyroid cancer (where CBX7 is heavily downregulated) compared with normal thyroid tissues (6). The cellular context might play a critical role in the regulation of p16 by CBX7, and it is also likely that p16 is not effective in the neoplastic tissues: it might be just an attempt of the cells to contrast the proliferation signals.

Figure 6

Cbx7-KO mice develop lung and liver tumors. (A) Frequency of spontaneous solid tumor development in 17- to 22-month-old mice of each genotype (n = 11 [Cbx7^+/+]; 34 [Cbx7^+/–]; 24 [Cbx7^–/–]). Ad, adenoma; Ca, carcinoma. (B) Representative liver and lung neoplasias in Cbx7-KO mice. Liver and lung tissues from Cbx7^+/+ animals are shown on the left as normal controls. Original magnification, ×10. (C and D) Tumor incidence in livers (C) and lungs (D) from Cbx7^+/+, Cbx7^+/–, and Cbx7^–/– mice.

Table 1

Neoplastic findings in Cbx7-KO mice

CBX7 is drastically downregulated in human lung carcinomas. Because cyclin E overexpression plays a critical role in lung carcinogenesis (19, 27), we focused our attention on human lung tumors. We first evaluated CBX7 protein expression in a panel of human lung adenocarcinomas by immunohistochemistry. As shown in Table 2, CBX7 was not expressed in any of the lung carcinomas analyzed, whereas it was expressed in normal lung tissue. Moreover, analysis of CBX7- and CCNE1-specific mRNA expression by qRT-PCR revealed an inverse correlation between CBX7 and CCNE1 expression in lung carcinomas (Supplemental Figure 5).

Table 2

CBX7 LOH and expression of CBX7 and cyclin E in lung normal and carcinoma tissues

Loss of heterozygosity (LOH) analysis at the CBX7 locus revealed LOH in 50% of the informative carcinomas (Table 2). Interestingly, when the morphologically normal lung area adjacent to the tumor was analyzed for CBX7 expression, no staining occurred in 4 of 8 cases analyzed (Figure 7). In other cases, staining was weaker compared with the normal lung (data not shown). Moreover, the 4 samples negative for CBX7 expression showed LOH when we analyzed the DNA extracted from the same microdissected areas (Table 2), which suggests that the reduced CBX7 expression might be involved in the transition of lung cells to a completely transformed malignant stage.

Figure 7

Immunohistochemical analysis of CBX7 and cyclin E expression in human normal and neoplastic lung tissues. (A and B) Normal lung had intense immunoreactivity for CBX7 (A, arrows), whereas it was negative for cyclin E expression (B). Arrowhead in B denotes positive histiocytes, a positive internal control. (C and D) Lung carcinoma was negative for CBX7 staining (C), but strongly positive for cyclin E expression (D). Higher-magnification views of A–D are provided in Supplemental Figure 7. (E and F) In a morphologically normal lung carcinoma–adjacent tissue, the tumoral part (top left) and the adjacent tissue (bottom right) did not express CBX7 (E), whereas both expressed high levels of cyclin E (F). Original magnification: ×63 (A–D); ×10 (E and F).

Since, as shown above, CBX7 negatively regulates CCNE1 promoter activity, we also evaluated cyclin E expression in the same lung neoplastic samples. In line with previous data (19, 27), cyclin E was overexpressed in all cases (Table 2 and Figure 7). Interestingly, cyclin E expression was higher in samples with LOH at the CBX7 locus. Moreover, cyclin E was overexpressed in the areas adjacent to the lung carcinoma only when CBX7 expression was absent and LOH was detected at the CBX7 locus (Table 2 and Figure 7).

To verify that CBX7 is a negative regulator of CCNE1 expression in lung cells, we restored CBX7 expression in 2 lung carcinoma cell lines (A549 and H1299) that did not show any expression of CBX7 (Supplemental Figure 6A). Restoration of CBX7 expression resulted in a significant decrease in CCNE1 expression, verified by qRT-PCR (Supplemental Figure 6B). Moreover, to further confirm that CBX7 downregulation could be a causative event in lung cell proliferation, we performed a colony-forming assay on the same lung carcinoma cell lines. As shown in Supplemental Figure 6C, the cells transfected with CBX7 expression vector gave rise to a lower number of colonies compared with cells transfected with empty vector.

Polycomb group proteins function as multiprotein complexes and are part of a gene regulatory mechanism that determines cell fate during normal and pathogenic development. They form transcriptional repressor modules that can be functionally divided into at least 2 distinct complexes: (a) the initiation complex PRC2, the core of which in humans consists of EZH2, EED, and SUZ12; (b) the maintenance complex PRC1, whose components are the mammalian homologs of Drosophila polycomb (Pc), posterior sex combs (Psc), sex combs extra (Sce), and polyhomeiotic (Ph) proteins (28). CBX7 is a Pc homolog consisting of a conserved chromodomain near the N-terminus and a Pc box in the C-terminal region. Deregulation of PcG proteins contributes to tumorigenesis where aberrant silencing of PcG target genes is frequent (29). Indeed, EZH2 is amplified and highly expressed in many tumor types, and its expression correlates with high proliferation rate and poor outcome in breast, prostate, and other cancers (30, 31). Conversely, loss of CBX7 expression correlates with a highly malignant phenotype (6, 8, 9, 11, 12). The fact that EZH2 and CBX7 can physically associate with DNA methyltransferases (DNMTs) (32, 33) suggests a mechanism whereby PcG proteins directly contribute to the altered DNA methylation profiles that are observed in multiple cancer types. Indeed, PcG target genes are as much as 12 times more likely to be aberrantly silenced by DNA methylation in cancer than are non-PcG target genes (34–36). However, the role of CBX7 in the regulation of cell growth and tumorigenesis is still controversial, since there is also evidence that its overexpression leads to cellular immortalization in vitro and tumor development in vivo, thereby indicating that CBX7 is an oncogene (3, 14). To address this issue, we generated and characterized Cbx7-KO mice.

The data reported here propose CBX7 as a tumor suppressor gene. Consistent with our earlier finding that restoration of CBX7 expression in thyroid carcinoma cells blocks the cells in G₁ phase (6), the proliferation rate of Cbx7-KO MEFs was higher than that of WT MEFs, and Cbx7^–/– mice developed neoplasias. However, in contrast to a report that CBX7 extends cellular lifespan (14), Cbx7-KO MEFs went to senescence later than did WT MEFs. This suggests that the cellular context plays a critical role in the effect of the CBX7 protein on cell growth. In the case of Cbx7-KO MEFs, the context may be considered more physiological than experiments in which CBX7 was ectopically overexpressed in HprECs and WI38 cells (14). It is possible that CBX7 overexpression deranges the composition of the PRC complexes, thereby causing unbalanced cell DNA methylation. Moreover, the same authors report reduced growth of both MEFs and human lung fibroblasts (WI38) when CBX7 expression is silenced (14). The different experimental approach, ours in vivo and the previous one in vitro, may account for these contradictory results. Consistent with this hypothesis, it has been previously shown that in vivo data show significant discrepancies compared with in vitro data and transfection approaches when p53 pathway has been studied (37). Perhaps it is difficult to faithfully reproduce the relative ratios of CBX7 and its regulators and/or partners using transfection protocols. We also cannot exclude the hypothesis that the shRNA constructs used in the previous studies may have interfered with the expression of other genes, maybe of the Pc family. The behavior of MEFs derived from Cbx7-KO mice and Cbx7-overexpressing transgenic mice described herein is supported by the drastic downregulation of CBX7 expression previously reported by several groups in malignant neoplasias — including thyroid (6), pancreatic (8), colon (11), lung (present study), gastric (10), bladder (7), and breast (9) carcinomas — and by a reduced growth rate achieved by restoration of CBX7 expression in carcinoma cells of different origin (6, 9, 11, 38).

The inhibition of Ccne1 expression by Cbx7 seems to have a critical role in the phenotype of the Cbx7-KO mice. Indeed, Ccne1 was expressed at higher levels in Cbx7-KO mice than in the WT controls, and CBX7 binds to the CCNE1 promoter in a region that includes the E-box, thereby repressing its activity. The HMGA1b protein was a competitor of this action. The antagonistic effect of CBX7 versus HMGA1b protein on the CCNE1 promoter, and perhaps other CBX7 target genes, probably accounts for most of the phenotypic features of the Cbx7-KO mice. Interestingly, HMGA and CBX7 act in an opposite manner in human malignancies: HMGA overexpression and CBX7 downregulation correlate with a poor prognosis (6, 8, 11, 39, 40).

We also report that benign liver and lung adenomas were moderately frequent in Cbx7^+/– mice, whereas liver and lung carcinomas appeared in Cbx7^–/– mice, which suggests that the grade of malignancy depends on the number of Cbx7 functional alleles. To further probe the susceptibility of Cbx7-KO mice to neoplasias and evaluate the role of Cbx7 expression in cancer progression, we plan to cross these animals with transgenic mice carrying activated oncogenes and/or induce experimental neoplasias in Cbx7-KO mice.

Interestingly, all the human lung adenocarcinomas we analyzed showed the absence of CBX7 expression, and LOH at the CBX7 locus was frequent in these neoplasias. Interestingly, LOH at the CBX7 locus also occurred in the morphologically normal tissue adjacent to lung tumor, in which CBX7 expression was significantly lower than in the normal lung tissues and cyclin E was overexpressed. These results suggest that the reduced CBX7 expression might facilitate the complete malignant lung cell transformation. Cyclin E overexpression in lung cells is probably the mechanism whereby loss of CBX7 expression contributes to lung carcinogenesis. Indeed, aberrant cyclin E expression is frequent in pulmonary dysplasia and lung cancer (19), and transgenic mice overexpressing cyclin E develop dysplasia and multiple pulmonary adenocarcinomas (41). However, in order to unequivocally demonstrate that cyclin E overexpression accounts for the phenotype of the Cbx7-KO mice, further experiments are required, specifically the generation of mice carrying the disruption of both Cbx7 and Ccne1 genes; the lack of development of lung carcinomas in these mice would validate our hypothesis.

In conclusion, our analysis of the phenotype of Cbx7-KO mice supports the concept that CBX7 is a tumor suppressor gene, in seeming contrast with other published reports that CBX7 behaves as an oncogene. It is conceivable that CBX7 can exert both oncogenic and antioncogenic functions, depending on the nature of other cellular events and on the presence of interacting proteins. Moreover, beyond defining a tumor suppressor role of the CBX7 gene, the data reported here indicate a critical role of the loss of CBX7 expression in lung carcinogenesis as a likely precursor to CCNE1 overexpression.

View Supplemental data

This work was supported by grants from AIRC (IG 5346) and the Ministero dell’Università e della Ricerca Scientifica e Tecnologica – MIUR (PRIN 2008). We thank Jean Gilder for editing the text.

Address correspondence to: Alfredo Fusco, Istituto di Endocrinologia ed Oncologia Sperimentale del CNR, Via Pansini 5, 80131 Naples, Italy. Phone: 39.081.7463602; Fax: 39.081.2296674; E-mail: afusco@napoli.com or alfusco@unina.it.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J Clin Invest. 2012;122(2):612–623. doi:10.1172/JCI58620.

1. Schuettengruber B, Chourrout D, Vervoort M, Leblanc B, Cavalli G. Genome regulation by polycomb and trithorax proteins. Cell. 2007;128(4):735–745.
   View this article via: PubMed CrossRef Google Scholar
2. Wu JL, Lessard J, Crabtree GR. Understanding the words of chromatin regulation. Cell. 2009;136(2):200–206.
   View this article via: PubMed CrossRef Google Scholar
3. Scott CL, et al. Role of the chromobox protein CBX7 in lymphomagenesis. Proc Natl Acad Sci U S A. 2007;104(13):5389–5394.
   View this article via: PubMed CrossRef Google Scholar
4. Bernstein E, Duncan EM, Masui O, Gil J, Heard E, Allis CD. Mouse polycomb proteins bind differentially to methylated histone H3 and RNA and are enriched in facultative heterochromatin. Mol Cell Biol. 2006;26(7):2560–2569.
   View this article via: PubMed CrossRef Google Scholar
5. Lund AH, van Lohuizen M. Polycomb complexes and silencing mechanisms. Curr Opin Cell Biol. 2004;16(3):239–246.
   View this article via: PubMed CrossRef Google Scholar
6. Pallante P, et al. Loss of the CBX7 gene expression correlates with a highly malignant phenotype in thyroid cancer. Cancer Res. 2008;68(16):6770–6778.
   View this article via: PubMed CrossRef Google Scholar
7. Hinz S, et al. Expression parameters of the polycomb group proteins BMI1, SUZ12, RING1 and CBX7 in urothelial carcinoma of the bladder and their prognostic relevance. Tumour Biol. 2008;29(5):323–329.
   View this article via: PubMed CrossRef Google Scholar
8. Karamitopoulou E, et al. Loss of the CBX7 protein expression correlates with a more aggressive phenotype in pancreatic cancer. Eur J Cancer. 2010;46(8):1438–1444.
   View this article via: PubMed CrossRef Google Scholar
9. Mansueto G, et al. Identification of a new pathway for tumor progression: microRNA-181b up-regulation and CBX7 down-regulation by HMGA1 protein. Genes Cancer. 2010;1(3):210–224.
   View this article via: PubMed CrossRef Google Scholar
10. Jiang Z, et al. Increased expression of miR-421 in human gastric carcinoma and its clinical association. J Gastroenterol. 2010;45(1):17–23.
    View this article via: PubMed CrossRef Google Scholar
11. Pallante P, et al. The loss of the CBX7 gene expression represents an adverse prognostic marker for survival of colon carcinoma patients. Eur J Cancer. 2010;46(12):2304–2313.
    View this article via: PubMed CrossRef Google Scholar
12. Federico A, et al. Chromobox protein homologue 7 protein, with decreased expression in human carcinomas, positively regulates E-cadherin expression by interacting with the histone deacetylase 2 protein. Cancer Res. 2009;69(17):7079–7087.
    View this article via: PubMed CrossRef Google Scholar
13. Bernard D, et al. CBX7 controls the growth of normal and tumor-derived prostate cells by repressing the Ink4a/Arf locus. Oncogene. 2005;24(36):5543–5551.
    View this article via: PubMed CrossRef Google Scholar
14. Gil J, Bernard D, Martinez D, Beach D. Polycomb CBX7 has a unifying role in cellular lifespan. Nat Cell Biol. 2004;6(1):67–72.
    View this article via: PubMed CrossRef Google Scholar
15. Sherr CJ, DePinho RA. Cellular senescence: mitotic clock or culture shock? Cell. 2000;102(4):407–410.
    View this article via: PubMed CrossRef Google Scholar
16. Gray-Bablin J, Zalvide J, Fox MP, Knickerbocker CJ, DeCaprio JA, Keyomarsi K. Cyclin E, a redundant cyclin in breast cancer. Proc Natl Acad Sci U S A. 1996;93(26):15215–15220.
    View this article via: PubMed CrossRef Google Scholar
17. Kim S, et al. Histone deacetylase inhibitor apicidin induces cyclin E expression through Sp1 sites. Biochem Biophys Res Commun. 2006;342(4):1168–1173.
    View this article via: PubMed CrossRef Google Scholar
18. Leone G, et al. E2F3 activity is regulated during the cell cycle and is required for the induction of S phase. Genes Dev. 1998;12(14):2120–2130.
    View this article via: PubMed CrossRef Google Scholar
19. Lonardo F, Rusch V, Langenfeld J, Dmitrovsky E, Klimstra DS. Overexpression of cyclins D1 and E is frequent in bronchial preneoplasia and precedes squamous cell carcinoma development. Cancer Res. 1999;59(10):2470–2476.
    View this article via: PubMed Google Scholar
20. Bracken AP, et al. The Polycomb group proteins bind throughout the INK4A-ARF locus and are disassociated in senescent cells. Genes Dev. 2007;21(5):525–530.
    View this article via: PubMed CrossRef Google Scholar
21. Fedele M, et al. HMGA2 induces pituitary tumorigenesis by enhancing E2F1 activity. Cancer Cell. 2006;9(6):459–471.
    View this article via: PubMed CrossRef Google Scholar
22. Marks P, Rifkind RA, Richon VM, Breslow R, Miller T, Kelly WK. Histone deacetylases and cancer: causes and therapies. Nat Rev Cancer. 2001;1(3):194–202.
    View this article via: PubMed CrossRef Google Scholar
23. Bolden JE, Peart M, Johnstone RW. Anticancer activities of histone deacetylase inhibitors. Nat Rev Drug Discov. 2006;5(9):769–784.
    View this article via: PubMed CrossRef Google Scholar
24. Minucci S, Pelicci PG. Histone deacetylase inhibitors and the promise of epigenetic (and more) treatments for cancer. Nat Rev Cancer. 2006;6(1):38–51.
    View this article via: PubMed CrossRef Google Scholar
25. Smith CL. A shifting paradigm: histone deacetylases and transcriptional activation. Bioessays. 2008;30(1):15–24.
    View this article via: PubMed CrossRef Google Scholar
26. Kouzarides T. Chromatin modifications and their function. Cell. 2007;128(4):693–705.
    View this article via: PubMed CrossRef Google Scholar
27. Fukuse T, Hirata T, Naiki H, Hitomi S, Wada H. Prognostic significance of cyclin E overexpression in resected non-small cell lung cancer. Cancer Res. 2000;60(2):242–244.
    View this article via: PubMed Google Scholar
28. Valk-Lingbeek ME, Bruggeman SW, van Lohuizen M. Stem cells and cancer; the polycomb connection. Cell. 2004;118(4):409–418.
    View this article via: PubMed CrossRef Google Scholar
29. Bracken AP, Helin K. Polycomb group proteins: navigators of lineage pathways led astray in cancer. Nat Rev Cancer. 2009;9(11):773–784.
    View this article via: PubMed CrossRef Google Scholar
30. Yu J, et al. Integrative genomics analysis reveals silencing of beta-adrenergic signalling by polycomb in prostate cancer. Cancer Cell. 2007;12(5):419–431.
    View this article via: PubMed CrossRef Google Scholar
31. Yu J, et al. A polycomb repression signature in metastatic prostate cancer predicts cancer outcome. Cancer Res. 2007;67(22):10657–10663.
    View this article via: PubMed CrossRef Google Scholar
32. Vire E, et al. The Polycomb group protein EZH2 directly controls DNA methylation. Nature. 2006;439(7078):871–874.
    View this article via: PubMed CrossRef Google Scholar
33. Mohammad HP, et al. Polycomb CBX7 promotes initiation of heritable repression of genes frequently silenced with cancer-specific DNA hypermethylation. Cancer Res. 2009;69(15):6322–6330.
    View this article via: PubMed CrossRef Google Scholar
34. Ohm JE, Baylin S-B. Stem cell chromatin patterns: an instructive mechanism for DNA hypermethylation? Cell Cycle. 2007;6(9):1040–1043.
    View this article via: PubMed CrossRef Google Scholar
35. Schlesinger Y, et al. Polycomb-mediated methylation on Lys27 of histone H3 pre-marks genes for de novo methylation in cancer. Nat Genet. 2007;39(2):232–236.
    View this article via: PubMed Google Scholar
36. Widschwendter M, et al. Epigenetic stem cell signature in cance. r Nat Genet. 2007;39(2):157–158.
    View this article via: PubMed CrossRef Google Scholar
37. Toledo F, Wahl GM. Regulating the p53 pathway: in vitro hypotheses, in vivo veritas. Nat Rev Cancer. 2006;6(12):909–923.
    View this article via: PubMed CrossRef Google Scholar
38. Li Q, et al. Polycomb CBX7 directly controls trimethylation of histone H3 at lysine 9 at the p16 locus. PLoS One. 2010;5(10):e13732.
    View this article via: PubMed CrossRef Google Scholar
39. Fusco A, Fedele M. Roles of HMGA proteins in cancer. Nat Rev Cancer. 2007;7(12):899–910.
    View this article via: PubMed CrossRef Google Scholar
40. Fedele M, Fusco A. HMGA and cancer. Biochim Biophys Acta. 2010;1799(1–2):48–54.
    View this article via: PubMed Google Scholar
41. Ma Y, et al. Transgenic cyclin E triggers dysplasia and multiple pulmonary adenocarcinomas. Proc Natl Acad Sci U S A. 2007;104(10):4089–4094.
    View this article via: PubMed CrossRef Google Scholar
42. Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: A Laboratory Manual . 2nd ed. Cold Spring Harbor, New York, USA: Cold Spring Harbor Laboratory Press; 1989.
43. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001;25(4):402–408.
    View this article via: PubMed CrossRef Google Scholar
44. Pierantoni GM, Rinaldo C, Esposito F, Mottolese M, Soddu S, Fusco A. High mobility group A1 (HMGA1) proteins interact with p53 and inhibit its apoptotic activity. Cell Death Differ. 2006;13(9):1554–1563.
    View this article via: PubMed CrossRef Google Scholar
45. Melillo RM, et al. Critical role of the HMGI(Y) proteins in adipocytic cell growth and differentiation. Mol Cell Biol. 2001;21(7):2485–2495.
    View this article via: PubMed CrossRef Google Scholar
46. Esposito F, Tornincasa M, Chieffi P, De Martino I, Pierantoni GM, Fusco A. High-mobility group A1 proteins regulate p53-mediated transcription of Bcl-2 gene. Cancer Res. 2010;70(13):5379–5388.
    View this article via: PubMed CrossRef Google Scholar