CBX7 is a tumor suppressor in mice and humans (original) (raw)
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 _Eco_RI-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.
Generation of _Cbx7_-KO mice. (A) Endogenous WT allele, targeting vector, and resulting KO allele. E, _Eco_RI; 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).
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 _CC_NE1.
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.
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.
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.
_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.
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).
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.
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.