Negative Regulation of β₄ Integrin Transcription by Homeodomain-Interacting Protein Kinase 2 and p53 Impairs Tumor Progression

Altered β₄ expression is found in several tumors of epithelial origin where the integrin activates signaling pathways that modulate cell proliferation and survival (1, 2). α₆β₄ integrin associates with ErbB2 in mammary cells and cooperates with ErbB2 to promote phosphatidylinositol 3-kinase-dependent invasion and survival (3, 4). In MMTV-Neu mice, the introduction of a targeted deletion of the β₄ cytoplasmic domain revealed that β₄ integrin promotes mammary tumor progression by cooperating with ErbB-2 signaling (5). This is due, at least in part, to the integrin ability of regulating, at the translation level, the expression of the ErbB-3 receptor that results in an increase of ErbB2/ErbB3 heterodimerization and, consequently, in the activation of the phosphatidylinositol 3-kinase survival pathway (6). In mammary tumors, in patients with low disease-free survival, α₆β₄ expression is functionally associated with ErbB3, phosphorylated Akt, and estrogen receptor-β1 negativity (7). β₄ expression was also shown to be strongly associated with mammary basal-like tumors and their aggressive behavior (8, 9). In agreement with a functional role of increased α₆β₄ expression in tumor progression, depletion of α₆β₄ integrin inhibits tumor cell growth both in vitro and in vivo (10–12) and strongly reduces the activity of the phosphatidylinositol 3-kinase pathway, favoring responsiveness to tamoxifen treatment in MCF7 cells (11).

The increase of α₆β₄ expression in epithelial cancers and the peculiarity of its distribution have suggested that β₄ expression might be regulated, at least in part, at the transcriptional level (13). In follicular and papillary thyroid carcinomas, α₆β₄ integrin is transcriptionally activated on cell transformation (14). In contrast, in skin basal cell carcinoma (15) and prostate cancers (16), the expression of α₆β₄ is reduced compared with normal tissues, further indicating a transcriptional regulation of β₄ in tumors.

Recently, the transcription factor p63 was shown to be a key regulator of the ITGB4 gene (the β₄ integrin subunit) transcription in mammary MCF10A cells (17). In the present study, we show that homeodomain-interacting protein kinase 2 (HIPK2) controls β₄ integrin transcription. HIPK2 is a nuclear serine/threonine kinase originally identified as corepressor for homeodomain transcription factors (18). It binds p53 regulating its localization, phosphorylation, acetylation, and transcriptional activity (19–21). DNA damage-inducing agents, such as UV or Adriamycin, activate HIPK2 to induce p53-mediated apoptosis. This process is also associated with transcriptional repression or others genes, involved in tumor progression, such as galectin-3 (22). HIPK2 can also physically interact with the p53 family members, p73 and p63, and modulate the activity of these proteins whose functions can be directly or indirectly related to apoptosis (23).

Here, we show that HIPK2 requires wild-type p53 (wtp53) to repress β₄ transcription and that, in the absence of HIPK2 and/or p53 activity, β₄ transcription is strongly activated by TAp63 and TAp73. These findings were strongly supported by in vivo data. Indeed, analysis of 67 wtp53-carrying breast carcinomas revealed a strong correlation between β₄ overexpression and inactivation of HIPK2 by its cytoplasmic relocalization (24), supporting the hypothesis that the loss of HIPK2 and/or p53 function can favor the up-regulation of β₄ integrin in human cancers.

Cell lines and treatments. MCF7, RKO, LoVo, HT29, SW480, and HCT116 were obtained from the American Type Culture Collection; HCT116 p53^/- (25) cells were kindly provided by B. Vogelstein, MIP, DLD-1, WiDr, and GEO were provided by F. Guadagni, and LNCaP and PC3 were provided by A. Farsetti. RKO/shHIPK2 subclones were described previously (26). For DNA damage, cells were treated with lethal (3 μmol/L) or sublethal (0.5 μg/mL) doses of Adriamycin (Pharmacia) for 24 h or irradiated with 75 J/m² of UV light and collected at the indicate time points. Flow cytometry, soft agar, and invasion assays were done as described (7, 11).

Transfections and plasmids. Vectors were transfected by Lipofectamine Plus (Invitrogen). pRC/CMV-β₄ vector was kindly provided by F. Giancotti. β₄ depletion was obtained by pSuper.retro vector containing β₄-shRNA-specific sequence (6). pcDNA3 carrying HA-tagged TAp73 (27) or myc-tagged TAp63 were kindly provided by S. Strano. pCMV-Flag-HIPK2 (19) and pSuperHIPK2 (28) were used to overexpress or deplete HIPK2. Transient HIPK2 depletion was obtained by HIPK2-specific Stealth RNAi (Invitrogen) with Lipofectamine RNAiMAX. p53 protein was overexpressed by pCAG3.1-wtp53 vector and silenced by pRetroSuper-wtp53 vector (28).

The promoter-less luciferase reporter plasmid pGL-3 basic and pGL-3 L5.5K, carrying the 5′-flanking region of the β₄ integrin gene (L5.5K fragment: -5197, +333), were kindly provided by T. Yamada and S. Hirohashi. Deletion mutants were generated by restriction endonucleases as follows: L2.2K (SacI, -1854, +333) and L0.7K (SmaI, -315, +333). Luciferase activity was determined with the dual-luciferase reporter assay system (Promega). Results were expressed as relative luciferase activity or folds of induction obtained by the ratio of the firefly/Renilla luminescence.

Western blot analysis and antibodies. Total cell extracts were prepared as described (7). The following antibodies were employed: anti-β₄ 439-9B and 450-11A monoclonal antibody (11); anti-HA (3F10; Calbiochem); anti-p53 (FL-393) and anti-mitogen-activated protein kinase (MAPK; Santa Cruz Biotechnology); anti-Akt (Cell Signaling); anti-HIPK2 antiserum (29); anti-α-tubulin (TU-01), actin (JLA20), and HSP70 (N27F3-4; Immunological Science); and FITC- and peroxidase-conjugated anti-IgG (Cappel).

Reverse transcription-PCR. Total RNA was prepared using RNazol B (Invitrogen). Reverse transcription-PCR analyses for β₄ and HIPK2 were done using the following primers: human β₄ 5′-GAATTCGTTCTACGCTCTCC-3′ and 5′-GAATTCTGAGAGATGTGGGC-3′ and human HIPK2 5′-AGTCCACGACTCCCCCTACT-3′ and 5′-ATGGTGGGAGTGATGTAGGC-3′.

The housekeeping aldolase mRNA was used as an internal control (11).

Chromatin immunoprecipitation analysis. Chromatin immunoprecipitation was done as described (22). Immunoprecipitations were carried out with anti-p53 (Ab-7), p63 monoclonal antibody (4A4), p73 antibody (C-20 and H-79), HADC1, and p300 antibody (Santa Cruz Biotechnology) and anti-H3-K9 and H4-K20 acetylated (Upstate Biotechnology).

The oligonucleotides used for detection of the p53RE-bearing fragments of the human ITGB4 gene are fragment p2 5′-TGGGAGCTAAAAGGCAAAGA-3′ and 3′-AGCACCAACAATGTGACCAA-5′.

Patients and tissue specimens. The 67 stage I, IIa, and IIb wtp53 breast cancer patients included in this study have been already described (30) Tumors were staged according to the International Union Against Cancer tumor-node-metastasis system 2002 and graded according to Bloom and Richardson (31). The study was reviewed and approved by the ethics committee of the Regina Elena Cancer Institute, and written informed consent was obtained from all patients.

Immunohistochemistry. Immunohistochemical analysis of β₄ integrin and HIPK2 was done as described (7, 24). HIPK2 protein was considered overexpressed when >10% of the neoplastic cells presented a strong immunoreaction. β₄ expression was scored according to the following criteria: 0, no reaction; 1, low reaction (1-10% of positive cells; score +/++); and 2, high reaction (>10-50% of positive cells; score ++/+++).

Statistics. χ² test was used to test the relationship between integrin β₄ overexpression and HIPK2 nuclear or cytoplasmic compartmentalization. P < 0.05 was considered statistically significant. The tests indicate how much of the association between β₄ and HIPK2 nuclear and cytoplasmic localization is accounted by linear trend. In vitro experiments were analyzed with Student's t test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

β₄ transcription is induced by HIPK2 interference in β₄-null RKO cells. The colon cancer RKO cells were shown previously to carry wtp53 and be negative for β₄ expression (32). In these cells, HIPK2 depletion causes a strong increase of in vivo tumor uptake and tumor growth (26). Because expression of β₄ integrin is strictly related to tumor aggressiveness, we investigated whether the corepressor activity of HIPK2 was related to β₄ transcription. To test this hypothesis, we analyzed the expression of β₄ by reverse transcription-PCR and flow cytometry (fluorescence-activated cell sorting) in RKO/mock and RKO/shHIPK2 cells. We found that β₄ transcript is absent in RKO/mock but strongly expressed in RKO/shHIPK2 cells (Fig. 1A^,, left), suggesting that HIPK2 corepresses β₄ transcription in these cells. De novo expression of β₄ protein in RKO/shHIPK2 cells was confirmed by fluorescence-activated cell sorting (Fig. 1A^,, right). Regulation of β₄ transcription was further evaluated by testing the β₄ promoter activity by transfecting RKO/mock and RKO/shHIPK2 with a vector carrying L5.5K fragment of the β₄ promoter. As shown in Fig. 1B, β₄ promoter activity was strongly increased on depletion of HIPK2 (P < 0.01, shHIPK2 versus mock RKO cells). This result was confirmed in LNCaP prostate cancer cells, another β₄-negative cell line. The transient depletion of HIPK2 causes de novo expression of β₄ protein but, as soon as the HIPK2 interference is lost, LNCaP cells return to be negative for β₄ expression (Fig. 1C^,, left). Comparable results were also obtained in MCF7 and LoVo cells silencing HIPK2 by the use of Stealth siRNA, with different HIPK2-interfering sequences (Fig. 1C , middle and right). The data support the results obtained in RKO cells and exclude the possibility of selection of secondary effects. In agreement with this, data from arrays revealed up-regulation of β₄ transcription in HIPK2-depleted cells.⁵

Rescue of β₄ protein on HIPK2 interference increases MAPK and Akt phosphorylation, anchorage-independent growth, and invasion. We next analyzed the signaling pathways affected by the β₄ molecule. As shown in Fig. 2A,, de novo expression of β₄ induces a strong activation of MAPK phosphorylation in RKO/shHIPK2 compared with the basal level observed in RKO/mock cells. Based on our previous results (6, 7), we confirmed that expression of β₄ integrin induces a strong increase of ErbB3 synthesis (Fig. 2B), resulting in increase of Akt phosphorylation on heregulin β1 stimulation. The specificity of HIPK2-induced β₄-dependent signaling was confirmed by the finding that transient β₄ interference in RKO/shHIPK2 cells inhibits the level of MAPK and Akt activation (Fig. 2A and B) and causes a decrease of ErbB3 expression (Fig. 2B). The activation of MAPK and phosphatidylinositol 3-kinase signaling pathways determines a strong increase of invasion and anchorage-independent growth of shHIPK2 cells compared with RKO/mock cells (P < 0.001 and P < 0.05, respectively; Fig. 2C and D). Depletion of β₄ in RKO/shHIPK2 cells abolishes the increment of anchorage-independent growth and invasion indicating that HIPK2 controls through β₄-transcription cell growth and invasion (P < 0.001 and P < 0.05, respectively; Fig. 2C and D). Representative soft-agar colony formation is shown in Fig. 2D (right).

β₄ transcriptional repression by HIPK2 requires the presence of wtp53. To investigate whether p53 is involved in HIPK2-mediated β₄ transcriptional regulation, we analyzed a panel of human cancer cell lines endogenously expressing β₄ integrin and carrying wild-type or mutated p53 or p53-null. The cells were treated either with UV or Adriamycin, and the levels of HIPK2, p53 and β₄ were evaluated by reverse transcription-PCR and Western blot in the presence of HIPK2 or on its interference. Adriamycin treatment abrogated β₄ expression at the RNA and protein levels in wtp53-expressing MCF7 cells but not on HIPK2 depletion (Fig. 3A,, left) as well as in RKO/shHIPK2 cells (Fig. 3A,, right). Comparable results were obtained by UV treatment of MCF7 and HCT116 cells (Fig. 3B). In contrast, UV treatment did not affect β₄ expression in HT29 cells that express a mutated form of p53 (Fig. 3B,, right). In addition, HIPK2 overexpression down-regulated β₄ expression in wtp53 HCT116 cells but not in their p53-null derivatives (Fig. 3C,, left and middle), indicating that HIPK2 requires a wtp53 protein to inhibit β₄ expression. These results were strongly supported by the finding that depletion of HIPK2 increased β₄ expression only in HCT116 cells but not in their p53-null derivatives (Fig. 3C,, right). In agreement, several colon cancer cells expressing mutated p53 (WiDr, HT29, SW480, and GEO) show high level of β₄ expression, whereas wtp53-carrying cells (MIP, DLD-1, and LoVo) show low level of β₄. Consistent, prostate wtp53-carrying LNCaP and p53-null PC3 cells were negative and positive for β₄ expression, respectively (Fig. 3D,, top). The correlation between β₄ expression and p53 status was further confirmed by the basal transcriptional activity of β₄ promoter we found in the same cell lines (Fig. 3D , bottom).

Integrin β₄ overexpression strongly associates with HIPK2 cytoplasmic relocalization in breast carcinomas. To evaluate whether the relationship between β₄ overexpression and HIPK2 functionality we defined in tumor cell lines can occur in vivo, integrin β₄ expression and HIPK2 nuclear and cytoplasmic compartmentalization were analyzed in 67 stage I and II wtp53-carrying breast cancer patients whose clinical characteristics have been described previously (30). In these patients, it was shown previously that overexpression of high-mobility group A1 protein correlates with HIPK2 cytoplasmic localization and a low spontaneous apoptotic index (24). Thus, we asked whether a similar correlation occurred between HIPK2 localization and β₄ overexpression. Among the tumors analyzed, β₄ exhibited a strong homogeneous immunoreaction when HIPK2 was relocalized in the cytoplasm (62.5%). In contrast, when the tumors were β₄-negative (73.3%), HIPK2 was localized in the nucleus (Fig. 4A). Representative immunohistochemically positive and negative cases for HIPK2 and β₄ are shown in Fig. 4B. We also observed that HIPK2 was present in 39 (58.2%) tumors, with a cytoplasmic pattern of reactivity in 24 (35.8%) patients and a nuclear pattern in 15 (22.4%) patients. In the integrin β₄-positive samples, HIPK2 showed a cytoplasmic pattern in 62.5% of the cases and a nuclear one in 26.7% (P = 0.03, χ² test; Supplementary Table S1).

In tumor cells, the p53 family members divergently regulate β₄ transcription. To investigate in detail the role of p53 in the regulation of β₄ transcription, we overexpressed HIPK2 and p53 in LoVo colon cancer cells that endogenously express low level of β₄ integrin. Overexpression of wtp53 inhibits, although to a lesser extent than HIPK2, the transcriptional activity of the β₄ promoter (P < 0.01; Fig. 5A). The β₄ promoter region cloned into the L5.5K construct includes three putative p53-binding elements: p1 (-3141, -3117), p2 (-1088, -1064), and p3 (-107, -84). To assess their participation in the p53-dependent repression of β₄ transcription, we generated L2.2K and 0.7K deletion mutants, lacking one (p1) and two (p1 + p2) p53-responsive elements, respectively (Fig. 5B,, top). Overexpression of the three reporter constructs in LoVo cells revealed that the basal level of β₄ promoter activity is enhanced when the cells were transfected with L2.2K construct (P < 0.05) and further activated in the absence of two p53-responsive elements (L0.7K; P < 0.01; Fig. 5B). However, the concomitant expression of p53 down-regulates the activity of all three β₄ constructs (P < 0.05), indicating that p53 can use each of the three p53-responsive elements to negatively regulate β₄ transcription.

It has been shown previously that p63, one of the p53 family members, activates β₄ transcription in MCF10A normal epithelial cells (17). Both p63 and p73 are known to bind to canonical p53 DNA-binding sites and transactivate several p53-responsive promoters (33). However, novel target genes that are differentially regulated by various p53 family members were recently identified (34). To analyze the involvement of p63 and p73 in the regulation of β₄ transcription in tumors, we cotransfected the L5.5K reporter construct with TAp63 or TAp73 in RKO/mock and RKO/shHIPK2 cells. Overexpression of either TAp63 or TAp73 activates β₄ transcription in RKO/mock cells and even strongly in RKO/shHIPK2 cells (P < 0.01, and P < 0.001, respectively; Fig. 5C). Re-expression of HIPK2 in RKO/shHIPK2 significantly reduces (P < 0.01) TAp63 and TAp73-dependent activation of β₄ transcription, indicating that HIPK2 contrasts the transcriptional activity of p63 and p73 on β₄ promoter (Fig. 5C). In addition, TAp73 overexpression in RKO and LoVo cells on p53 interference further increased β₄ transcription (Fig. 5D,, left and middle). These results were consistent with the increase of the endogenous β₄ protein we found in the p53-depleted LoVo cells (Fig. 5D , right). Taken together, these data suggest that HIPK2 and p53 inhibit β₄ transcription counteracting the transcriptional activity of TAp63 and TAp73 on the β₄ promoter.

β₄ transcriptional modulation by HIPK2 and p53 family members involves HDAC and p300. HIPK2 can interact with homeobox factors forming corepressor complexes that contain histone deacetylases (35, 36). To verify this corepressor activity on the β₄ promoter, we performed chromatin immunoprecipitation assays in RKO/mock and RKO/shHIPK2 cells by amplifying the p2 region β₄ promoter as indicated in the cartoon (Fig. 6A). In this region, histones H3 and H4 are acetylated in RKO/shHIPK2 cells but not in RKO/mock cells (Fig. 6B,, top left). In agreement, the HDAC1 deacetylase, which can form corepressor complex with HIPK2 reinforcing its corepressor activity (35), is present on the β₄ promoter only in the HIPK2 positive RKO/mock cells (Fig. 6B,, top right). Given that the histone acetyltransferase p300/CBP can mediate p63- and p73-dependent transcription (37), we evaluated the presence of p300 on the β₄ promoter and found that it was absent in HIPK2 expressing cells but recruited on HIPK2 interference (Fig. 6B,, top right). In addition, p53 is present on the β₄ promoter in RKO/mock cells, whereas HIPK2 depletion strongly reduces the amount of p53 bound to the promoter (Fig. 6B,, bottom). By contrast, a greater amount of p63 and p73 is detectable in the β₄ promoter in RKO/shHIPK2 cells compared with RKO/mock cells (Fig. 6B,, bottom). These results were confirmed in LoVo cells by transient HIPK2 depletion by siRNA (Fig. 6B,, bottom). Interestingly, overexpression of TAp73 in mutated p53-carrying HT29 or p53-null HCT116 cells strongly enhances the activity of β₄ promoter (P < 0.01; Fig. 6C). As expected, both p63 and p73 were present on the β₄ promoter, whereas p53 was not (Fig. 6D).

Overall, these data indicate that although the detailed molecular mechanism of β₄ transcription by p53 family members is not yet clear, p53 and its family members directly interact with the β₄ promoter to repress or activate, respectively, β₄ transcription. In addition, although both p63 and p73 activate β₄ transcription, our results also indicate that TAp73 poses the stronger activity.

We have identified a mechanism by which the β₄ integrin subunit is regulated at the transcriptional level during tumor progression. In normal epithelial tissue, the presence of α₆β₄ integrin is restricted to the basal layer of the cells where it is required for hemidesmosome formation, cell adhesion, and cell survival (38). Many epithelial tumors lack the hemidesmosome and α₆β₄ expression is strongly increased and redistributed on the entire cell surface, indicating that, in these tumors, the integrin plays different role besides adhesion (39). These and other studies have suggested the existence of different transcriptional regulation of α₆β₄ integrin during tumor progression (14-5). Our study is the first one showing that α₆β₄ integrin is regulated in epithelial tumors at the transcriptional levels by HIPK2 and p53 that contributing to its regulation control β₄-induced tumor progression. This mechanism involves a functional pathway that links HIPK2 to the ability of p53 to suppress the survival function induced by β₄ integrin. We have defined a critical role for HIPK2 in the regulation of β₄ integrin expression. Knockdown of endogenous HIPK2 induced de novo expression of β₄ in RKO colon cancer cells and increased β₄-dependent signaling and tumor progression. These results are in strong agreement with previous findings showing that depletion of HIPK2 in RKO cells enhances in vivo tumorigenicity and tumor uptake (26).

However, HIPK2 acts as corepressor in concert with transcription factors, and among them, the p53 tumor suppressor represents one major HIPK2 target (36). In response to genotoxic stress, HIPK2 binds and activates p53 to induce apoptosis by activation of a large number of proapoptotic factors and the repression of antiapoptotic factors (36).

In contrast to p53, β₄ is considered a prosurvival factor and it is widely assumed that the distinct signaling properties of the α₆β₄ integrin account for its ability to enhance functions such as invasion and resistance to apoptotic stimuli (11, 12). Further, α₆β₄ expression is strictly correlated with tumor progression (39) leading to hypothesize that p53 could control β₄ transcription in the early phases of transformation or until p53 is functionally active. Indeed, our results support the hypothesis that a transcriptionally active p53 controls β₄ expression maintaining its level low in cancer cells. Microarray analysis revealed that RITA treatment in wtp53-carrying HCT116 cells activating p53 apoptotic function causes a down-regulation of β₄ mRNA level, further supporting our finding that wtp53 controls β₄ expression level (40).⁶

An important observation is that p63 regulates adhesion program and cell survival in normal mammary epithelial cells and that the β₄ gene is a direct target of p63 in these cells (17). These observations had added new highlights to the mechanism that regulates β₄ expression in normal epithelial tissues and suggested the possibility that others p53 family members participate to the regulation of its transcription. In line with this idea are our findings showing that overexpression of TAp63 or TAp73 strongly enhances β₄ transcription.

Although the p53-independent functions of p63 and p73 during development are clear, their role in tumorigenesis is less evident (33). In this respect, there is now increasing evidence supporting their participation to tumor formation and progression (33, 41). The expression of p63 and p73 reported in these studies strongly correlates with the expression of β₄ integrin previously analyzed in the same types of tumor (2, 7, 9, 42). Our results are apparently in contrast with recent finding showing that TAp73 knockout mice show genetic instability associated with enhanced aneuploidy and increased incidence of spontaneous tumors, indicating that TAp73 possess the tumor-suppressive ability (43). However, the tumor phenotype shown in TAp73^-/- mice occurs in a wtp53 background and it has been shown previously that the absence of p63 or p73 impairs the induction of p53-dependent apoptosis in response to DNA damage (44). These data indicate a specific role of p63 and p73 molecules in potentiating the apoptotic function of p53 and suggest a different role for these molecules in the loss of function of p53. We also cannot exclude that in the absence of HIPK2 or p53 function a different balance among the p53 family members might generate the formation of different transcription complexes that exert opposite effects on the same promoter as previously hypothesized for the p53 family members (33). However, both TAp63 and TAp73 belong to a complex family of factors with opposite functions, the relationships of which need further studies to understand their role during tumor progression.

Eventually, our results are strongly supported by the in vivo data showing a statistically significant correlation between β₄ overexpression and HIPK2 relocalization to the cytoplasm and are in agreement with recent finding showing that the stable depletion of HIPK2 leads to wtp53 misfolding causing a down-regulation of its transcriptional activity (45).

In summary, although the detailed molecular mechanisms need to be further clarified, our data indicate that HIPK2 and wtp53 corepress β₄ promoter impairing β₄-inducing tumor progression and that, in the absence of HIPK2 or when p53 becomes transcriptionally inactive, the TA members of the p53 family become able to transcribe the β₄ gene.

No potential conflicts of interest were disclosed.

Grant support: Associazione Italiana per la Ricerca sul Cancro and Ministero della Salute (R. Falcioni) and Fondazione Italiana Ricerca sul Cancro fellowship (G. Bon and V. Folgiero).

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

We thank Galina Selivanova for sharing results on microarray; all the people cited in the text for the gift of reagents; Drs. Giulia Fontemaggi, Aymone Gurtner, Cinzia Rinaldo, and Giulia Piaggio for critical discussion; and Silvia Bacchetti for critical reading of the article.