Aberrant expression of homeobox genes is common in cancers [5
], and recent studies have started to elucidate the causal role of these genes in tumorigenesis. In general, homeobox genes regulate developmental programs that coordinate different cell behaviors during embryogenesis, but their misexpression in differentiated cells can result in the acquisition of tumor-promoting properties, including proliferation, dedifferentiation, migration, invasion and survival [17
]. Good examples of the causal effects of homeobox genes were obtained in animal models with Oct-4, which promoted testicular germ cell tumors when expressed inappropriately [18
]; NKX3.1, which induced prostatic intraepithelial neoplasia and enhanced prostate cancer progression in collaboration with loss of function of the PTEN tumor-suppressor gene [19
]; and SIX1, the overexpression of which induced breast cancer in mammary epithelial cells [20
In the present study we identified differentially expressed genes belonging to cluster A of the HOX family of homeobox genes in OSCC compared to healthy oral mucosa. In healthy oral mucosa derived from patients without risk factors for oral cancer, only HOXA1, HOXA2 and HOXA4 were expressed, whereas the abundance of transcripts in histologically normal-looking oral mucosa near OSCC and in OSCC samples was higher. Among the transcripts detected in this study, the expression of HOXA1 was significantly higher in OSCC compared with healthy oral mucosa. In addition to the high transcriptional levels, we showed that HOXA1 homeoprotein expression was limited to the basal and suprabasal cells of the normal epithelium, while OSCCs showed broad immunoreactivity in the tumor nests and a significantly higher number of HOXA1-positive cells than that of controls. HOXA1 misexpression has been reported in leukemia, and carcinomas of the cervix, breast and oral cavity [21
]. In the latter, similar to our findings, it was demonstrated that the expression of HOXA1 was significantly higher in OSCCs when compared with healthy oral mucosas [3
]. The present study further revealed that the expression of HOXA1 was correlated with Ki67 immunohistochemical expression in both control and OSCC tissues and that the forced expression of HOXA1 in the non-tumorigenic cell line HaCAT was able to stimulate cell cycle progression, whereas the downregulation of HOXA1 levels in the OSCC cell line SCC9 decreased the proliferation.
Evidence from both in vitro
and in vivo
studies has suggested an oncogenic role for HOXA1 based on its effects on the promotion of anchorage-independent growth of normal epithelial cells and on induction of tumors in mice [22
]. In hematopoietic cells, overexpression of HOXA1 blocked differentiation, leading to transformation by colony formation in soft agar assays and to the development of acute myeloblastic leukemia in lethally irradiated mice [21
]. In immortalized mammary epithelial cells, expression of HOXA1 resulted in a dramatic increase in anchorage-independent proliferation by the promotion of cell survival mediated by activation of the STAT pathway [13
] and by transcriptional upregulation of Bcl-2 [22
]. Furthermore, HOXA1 is a downstream effector of E-cadherin-directed signaling required for anchorage-independent proliferation of mammary carcinoma cells [9
]. In addition, in growth hormone-induced oncogenic transformation of immortalized human mammary epithelial cells, HOXA1 governs the transcriptional up-regulation of c-Myc, cyclin D1 and Bcl-2 that are required for this event [23
]. In contrast to those studies, overexpression of HOXA1 did not affect apoptosis, adhesion, invasion, EMT and anchorage-independent growth of HaCAT epithelial cells. Reasons for the lack of those phenotypes in HOXA1 overexpression cells could reside in the fact that many of the members of homeobox gene families required cofactors for their complete functional activity [24
]. HOXA1 is dependent on the MEIS, PREP or HTH protein cofactors to activate and/or repress transcription [25
]. Consistent with this observation, Meis1 augmented the efficacy of HOXA1 on tumorigenic induction of primary hematopoietic cells, resulting in a more intense growth and increased colony number in soft agar assays [21
]. Thus, it is possible that the availability of cofactors in the HaCAT cells, as well as their stoichiometry with HOXA1, was not conducive to stimulating those biological functions of the protein, thereby removing the ability of aberrant HOXA1 expression to induce anchorage-independent growth and inhibit apoptosis.
Next, we investigated the clinicopathological relevance of HOXA1 expression for OSCC patients. Although 100% of the samples demonstrated positivity for HOXA1, the levels of immunoexpression varied considerably. We have observed that higher levels of HOXA1 were associated with larger tumor size at diagnosis, histopathologic differentiation of the tumor with high positivity for HOXA1 in undifferentiated tumors, and higher proliferative potential of the tumor. These tumor progression hallmarks were consistent with what was already described as induced by HOXA1 overexpression in other cell lines [9
]. Our findings also show a significant association between high HOXA1 expression with the presence of lymph node metastasis and worse clinical outcome, where patients with a high number of HOXA1-positive cells had substantially shorter overall and cancer-specific survival than did patients with a low number of HOXA1-positive cells. Furthermore, multivariate analysis showed that HOXA1 overexpression was an independent marker for overall survival in the entire sample after adjusting for other prognostic factors. Although a strong association in vitro between HOXA1 and Ki-67 was observed, Ki-67 immunohistochemical expression was not correlated with the overall survival of patients with OSCC. Interestingly, this lack of association has been reported by other studies [26
]. We do not feel that HOXA1 was sufficiently studied in its role in tumor prognosis in OSCC or other malignancies, however, several studies have reported that lymph node metastasis is the most reliable marker of OSCC patient prognosis [28
], and our findings showed that the presence of lymph node metastasis significantly correlated with high HOXA1 expression and both were significant prognostic factors for overall survival by multivariate analysis. Since HOXA1 is expressed in a number of cancers, our findings encourage its verification and monitoring in other malignancies.
In addition, as master regulators of development and tissue homeostasis, HOX genes often rely on other homeobox genes, as well as closely related HOX members, to ensure tissue specificity via gene expression regulation [5
]. Therefore, it is common to have dysregulated expression of several HOX genes at the same time during the malignancy process [3
]. Our group has previously demonstrated the role of HOXB7 and its prognostic significance to OSCC [8
]. In the latter study, immunoexpression of HOXB7 was significantly associated with clinically important markers of OSCC behavior, including lymph nodal metastasis at diagnosis, vascular infiltration and proliferative potential of the tumor, resulting in significantly shortened overall survival. Interestingly, 110 samples used in the present study were also utilized to analyze HOXB7, and a strong and positive correlation between these two markers was observed in the OSCC samples (data not shown), highlighting the importance of several members of the HOX family to OSCC.