HOXA13 Promotes Cancer Cell Growth and Predicts Poor Survival of Patients with Esophageal Squamous Cell Carcinoma

Esophageal cancer is the sixth most common cancer in the world. The majority of esophageal squamous cell carcinoma (ESCC) cases occur in Asia, especially in north central China. Tremendous advances in diagnosis and treatment have been achieved recently; however, the survival rate has not been significantly improved. Among the reasons are the advanced stage at diagnosis and the lack of an efficacious system both in understanding its carcinogenic mechanism and clinical evaluation, especially the lack of sensitive and specific molecular markers for early detection.

It is known that the process of normal embryogenesis and tumorigenesis share many of the same pathways, and tumorigenesis is an aberrant form of organogenesis (1, 2). Homeobox genes represent classic examples of the intimate relationship between embryogenesis and tumorigenesis. The mammalian HOX genes, which are arranged in clusters (HOXA through HOXD) on four separate chromosomes, encode a subset of transcription factors which regulate axial regional specification during embryonic development and have recently been shown to be aberrantly expressed in a variety of solid tumors (3–5).

HOX genes are expressed with temporal and spatial colinearity in development (6–8). Accordingly, Yahagi and colleagues (9) showed that HOX genes with high expression in the foregut had a propensity for low expression in the hindgut, whereas those with a low expression in the foregut tended to have high expression in the hindgut. Specifically, paralogue 13 (HOXA13, HOXB13, and HOXD13) had a tendency for high expression in the hindgut region and weak expression in the foregut including the esophagus. However, in our previous study, HOXA13, a member of paralogue 13, was found to be abnormally expressed in esophageal cancerous tissues but not in noncancerous tissues using reverse transcription-PCR, suggesting that HOXA13 might play a role in the carcinogenesis of the esophageal epithelium (10). In the present study, we further supported the involvement of HOXA13 in esophageal cancer growth because of the following evidences: (a) the abnormal expression of HOXA13 was confirmed by the detection of its expression on protein level in an esophageal cancer cell line; (b) the rate of colony formation and the capability of tumor growth in cells without HOXA13 expression were much lower than in cells expressing HOXA13; (c) the expression of HOXA13 protein was more abundantly found in ESCC tissues than its noncancerous counterparts; (d) further statistical analysis showed that the expression of HOXA13 protein was a negative predictor of disease-free survival time of patients.

Cell culture and transfections. Human esophageal cancer cell line EC-109 was cultured in DMEM (Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum in a humidified atmosphere with 5% CO₂, at 37°C. EC-109 cells were plated at 70% confluence (∼2 × 10⁵ cells) in a 24-well plate and grown over night, and transfected with 0.8 μg of either short hairpin (shRNA) targeting HOXA13 gene or scrambled shRNA, or empty vector using the Lipofectamine Reagent (Invitrogen). Stably transfected cells were selected with 600 μg/mL of G418 (Promega).

Western blot analysis. Total cell extracts were prepared in 1× SDS loading buffer, separated by SDS-PAGE, blotted onto polyvinylidene difluoride, then immunoreacted with goat anti-HOXA13 polyclonal antibody (Santa Cruz Biotechnology) as primary antibody. Rabbit anti-goat polyclonal horseradish peroxidase–conjugated IgG was used as a secondary antibody. Immunoreactivity was detected with an enhanced chemiluminescence reaction kit (GE Healthcare). As a loading control, glyceraldehyde-3-phosphate dehydrogenase was detected using a goat polyclonal antibody (Santa Cruz Biotechnology).

RNA interference. For down-regulation of HOXA13 expression in EC-109, a plasmid-mediated shRNA was used. The candidate shRNA sequences targeting HOXA13 mRNA were generated by Genechem Com, which were cloned into plasmid pGCsi-U6NeoGFP-Blank. The oligonucleotide pairs used were listed in Supplementary Table S1. Empty vector and scrambled shRNA were used as controls.

Colony formation assay. Stable shRNA transfected and control cells (2 × 10³) were cultured in 90-mm plate for 3 weeks in regular culture medium with 600 μg/mL of G418. The formed colonies were then fixed, stained, and counted.

In vivo carcinogenesis. Stable shRNA transfected cells (1 × 10⁶) in 100 μL serum-free DMEM (Life Technologies) were injected s.c. into the right flank of female nude mice at 18 to 20 g of weight. Scrambled shRNA-transfected cells were used as controls. Each group included 12 mice, and was housed in pathogen-free conditions. After 3 weeks of observation, the mice were sacrificed. Tumors were excised, measured, and weighted. All experiments were done in accordance with institutional standard guidelines of Peking University School of Oncology for animal experiments.

Patients and tissue samples. To investigate the difference of HOXA13 protein expression between cancerous and noncancerous tissue, nine ESCC patients were recruited, including eight men and one woman, ages 52 to 71 years (median, 62 years), with stage IIa (n = 4), IIb (n = 2), and III (n = 3) diseases according to the criteria of the TNM classification system of malignant tumors (UICC, 1987). The patients underwent radical esophagectomy at the Department of Thoracic Surgery, Peking University School of Oncology from February to June, 2003. The samples of cancer tissues and paired noncancerous mucosa were obtained from the Department of Pathology of the hospital. None of these patients had received preoperative adjuvant therapy or suffered from severe postoperative complications, such as anastomotic leakage, hemorrhage, or chylothorax.

To study the association of HOXA13 protein expression and other clinicopathologic characteristics and disease-free survival time, 155 ESCC cases were included, which underwent radical esophagectomy from July 1996 to November 2002 at the Department of Thoracic Surgery, Peking University School of Oncology, and were selected using the same criteria described above. The selected cases involved 117 men and 38 women, ages 39 to 78 years (median, 63 years), with stage I (n = 6), IIa (n = 62), IIb (n = 13), III (n = 72), and IV (n = 2) diseases. All the specimens had been routinely formalin-fixed, paraffin-embedded, and serially sectioned at 4 μm in thickness.

The clinical data were obtained from hospital records because all postoperative patients were routinely scheduled for a regular physical examination (visiting our hospital) for life-long follow-up. The calculation of disease-free survival time began at the date of surgery and ended when any of the following events happened: recurrence, metastasis, or oncological death. All patients in this study had recorded disease-free survival times. This study was approved by both the Ethics and the Academic committees of Peking University School of Oncology, and informed consent was obtained from all participants.

Immunohistochemistry. After routine deparaffinization and hydration, tissue sections were treated with 3% hydrogen peroxide and then heated in EDTA (pH 8.0) for antigen retrieval. The HOXA13 antigen-antibody reaction took place overnight at 4°C, following goat serum blocking. The streptavidin/peroxidase amplification kit (Zymed) was applied to detect the signal of the HOXA13 antigen-antibody reaction. Peroxidase activity was developed with diaminobenzidine. All sections were counterstained with hematoxylin. The purified rabbit polyclonal antibody against human HOXA13 (Abcam) was used at 2 μg/mL and goat anti-rabbit biotin-conjugated IgG was used as secondary antibody. Immunohistochemical signals were scored by two independent observers. The scores were calculated as the number of stained cells divided by the total number of cancer cells counted. Four high-power fields (400×) per slide were calculated and the results were averaged. Unequivocal staining of the cytoplasm in >50% of cancer cells was considered positive.

Statistical analysis. SPSS 10.0 software was used to perform the statistical analyses. All in vitro experiments were performed at least thrice in triplicates. When the data from different groups were compared, normal analysis and homogeneity of variance were checked first, and then an unpaired two-tailed t test analysis was used. Bars and error bars on the graphs as well as data in the text represent the mean ± SD. The relationship between the expression of HOXA13 protein and clinicopathologic characteristics was tested by χ² test. Univariate survival analysis was carried out by Kaplan-Meier method, and subjected to the log rank test. The Cox proportional hazards model with a stepwise procedure was used for multivariate analysis. The variables in the multivariate analysis were age, sex, tumor location, tumor cell differentiation, TNM stage, and HOXA13 expression. P < 0.05 was considered significant.

Expression of HOXA13 protein in EC-109. We have previously reported 11 of 39 HOX genes, including HOXA13, abnormally expressed in esophageal cancerous tissues but not in noncancerous tissues using reverse transcription-PCR (10). In this study, we obtained an esophageal cancer cell line EC-109 and further examined the expression of HOXA13 protein by Western blotting. Endogenous HOXA13 protein expression was detected in EC-109 cells (Fig. 1A).

Down-regulation of HOXA13 protein in EC-109. We introduced plasmids expressing shRNA targeting three different regions of HOXA13 into EC-109 cells to knock down the expression of HOXA13 protein. The efficacy of knockdown of HOXA13 protein was examined by Western blotting. Of the three plasmids used, one worked well, whereas the other two did not. Meanwhile, one plasmid with scramble shRNA was used as a control. After screening by G418, we established a cell strain with stable knockdown of HOXA13 protein following three passages (Fig. 1B).

Down-regulation of HOXA13 protein decreased cell proliferation in vitro. To investigate if down-regulation of HOXA13 protein could modulate cell proliferation, a colony formation assay was performed. The number of colonies formed (456 ± 7) in cells with stable knockdown of HOXA13 protein was significantly reduced, as compared with that of the control (983 ± 78; Fig. 1C, P = 0.0003). The level of colony formation in parental cells (812 ± 100) showed no significant difference compared with that of the control (P = 0.495).

Down-regulation of HOXA13 protein decreased tumor growth in vivo. After studying the effect of HOXA13 protein on cell proliferation in vitro, the involvement of HOXA13 protein in carcinogenesis in vivo was investigated. HOXA13 knockdown cells, control cells, and parental cells were injected into nude mice. The tumors were removed 3 weeks after injection and were measured and weighted. The weight and volume of tumors from the HOXA13 knockdown cells were 154.3 ± 71.1 mg and 0.249 ± 0.142 cm³, respectively, which were significantly less than those of the control (482.2 ± 90.5 mg and 0.718 ± 0.209 cm³, respectively; Fig. 1D; P < 0.01). However, no difference was found between the weight and volume of tumors from parental cells and those of the control.

Expression of HOXA13 protein detected by immunohistochemistry in ESCC and paired noncancerous tissue. The above results suggested that HOXA13 protein might exert an effect on the cell proliferation of esophageal cancer cells. To further elucidate the expression of HOXA13 protein in esophageal cancer, we initially used formalin-fixed, paraffin-embedded noncancerous colon and rectum mucosa specimens as positive controls. It was noted that HOXA13 protein was localized in the cytoplasm of colorectal epithelium (data not shown). As a transcription factor, HOXA13 was reported to show nuclear staining by Knosp and colleagues (11). However, we consistently detected it in the cytoplasm of esophageal cancer cells (Fig. 2). It is speculated that the cytoplasmic localization may be due to the modulation of nuclear localization signals, nuclear export sequences (12, 13), and interaction with cytoplasmic anchoring factor (12, 14–17). In the current study, HOXA13 protein showed increased expression in cancerous tissue in seven of nine patients (77.8%), as compared with paired noncancerous tissues in one of nine patients with ESCC (11.1%; Fig. 2; P = 0.015).

Association of HOXA13 protein expression with clinical characteristics and disease-free survival. The expression of HOXA13 protein was not significantly associated with clinical characteristics (Table 1).

Univariate analysis using the log rank test showed that the expression of HOXA13 was significantly associated with disease-free survival time (P = 0.0006, Table 2; Fig. 3). The median survival time was 14 months for HOXA13-positive patients, which was significantly shorter than the 26 months for HOXA13-negative patients. Multivariate analysis showed that TNM stage and HOXA13 expression were independent poor predictors of disease-free survival time in this series of 155 patients (Table 3). In addition, we attempted to investigate the correlation between HOXA13 and its known downstream target genes, Enpp2 (Innis group) or Bmp2, Bmp7 (Stadler group), or Epha7 (Stadler or Zappavigna group). Seventy-eight of 155 cases were studied immunohistochemically. Twenty-seven of 78 cases were HOXA13-positive and 51 cases were HOXA13-negative. There is no concordant expression between HOXA13 and Bmp2, Bmp7, or Epha7 (Supplementary Table S2). The reason for this phenomenon is unknown. The expression of Enpp2 could not be determined because Enpp2 antibody is not commercially available.

Genes regulating normal cell proliferation and differentiation participate in tumorigenesis and/or tumor progression because of their deregulated function or interaction with deregulated target genes. Among the principal examples are homeobox genes, which encode transcriptional regulatory proteins. Those proteins play important roles during embryonic development and seize function in mature organs. However, the aberrant expression is often found in cancers (3). The products of homeobox genes are transcription factors that regulate gene expression. The homeodomain is deemed as the functional motif of homeobox genes, which bind to specific DNA sequences, including promoters of other HOX genes, to enhance or inhibit their transcription. It has been shown that HOX genes are abnormally expressed in many solid tumors and derivative cell lines, as well as the corresponding embryonic tissues from which these tumor cells are derived (18–20). However, there is no detailed report thus far describing the function of HOX genes in cancer. Do HOX genes play a role in the carcinogenesis or growth of ESCC? Is the expression of HOX genes linked to the survival of patients, and does it have potential to be used as a clinical biomarker of ESCC, and direct the patient-specific treatment strategy? To elucidate these questions, further and detailed studies of HOX genes function are warranted.

In our previous studies, we detected the expression of HOX gene family members in esophageal cancer and paired noncancerous tissue. Eight of 39 HOX genes were detected in cancerous tissue, but not in paired noncancerous tissue, which were HOXA10, HOXA13, HOXB7, HOXC4, HOXC8, HOXD9, HOXD10, and HOXD13 (10). Interestingly, according to the “spatial colinearity” of HOX genes which is displayed in development, HOXA13 and HOXD13 are expressed specifically in the cloacal mesoderm and also uniquely in the hindgut and cloacal endoderm, but not in the esophagus (21, 22). Thus, the up-regulation of HOXA13 in esophageal cancer tissue implies that it might play a crucial role in carcinogenesis.

In order to further investigate the role of HOXA13 in esophageal carcinogenesis, we evaluated the effect of HOXA13 expression on esophageal cancer cell proliferation. EC-109 is a HOXA13-expressing esophageal cancer cell line. Using RNA interference technique, the expression of HOXA13 was down-regulated. A cell strain with stable knockdown of HOXA13 protein was selected and used for the subsequent experiments. The colony formation ability of HOXA13 protein expression was examined in vitro. The results suggested that cells with a stable knockdown of HOXA13 proteins showed a significantly weaker colony formation ability, compared with the controls. In order to validate this observation, we injected the cells and control cells into nude mice to test their tumor formation ability and growth in vivo. Consistently, the tumor formation ability of the cells with HOXA13 protein knocked-down was significantly weaker than that of the controls. All of these results indicate that the expression of HOXA13 protein is involved in carcinogenesis and growth of ESCC. This result is consistent with those reports implying the role of HOX gene members in carcinoma. Whelan and colleagues (23) found that HOXA9 overexpression induced IGF-1R expression and subsequently promoted leukemia cell proliferation. Beslu and colleagues (24) reported that HOXB4 overexpression up-regulated the growth of hematopoietic stem cells. Transfection of HOXB7 into breast cancer cell line SkBr3 stimulated the amplification and colony formation ability of cancer cells (25), and HOXB7 overexpression increased the proliferation ability of immortalized ovary epithelial cells (26). There are reports demonstrating the role of HOXA13 in neoplasms other than solid tumors. Taketani and colleagues (27) found the NUP98-HOXA13 fusion gene in patients with acute myelogenous leukemia suggesting that HOXA13, like HOXA9, might play an important role in the pathogenesis of leukemia. In addition, the deregulated expression of HOXA13 was found in glioblastoma tissue (28), and melanoma (29). However, to date, there is no report regarding the function of HOXA13 in solid tumors. Our results showed for the first time that HOXA13 overexpression is involved in carcinogenesis and promoted tumor growth in ESCC.

Because the final goal of these diagnosis and treatment strategies is to improve the survival of patients, it is urgent to clarify if the promotion of carcinogenesis and growth from HOXA13 proteins exerts any effect on the survival of patients. To investigate this, we first confirmed that the overexpression of HOXA13 protein did exist in the cancer tissue of patients with ESCC, then we further testified that this overexpression did affect the survival; the survival of patients with HOXA13 overexpression was significantly shorter than that of patients without HOXA13 overexpression. HOXA13 expression, like TNM stage, was a negative, independent predictor of disease-free survival time of patients with ESCC. Thus, combination of HOXA13 expression and TNM stage classification might provide a more accurate prediction of the postoperative outcome of patients with ESCC. Because HOXA13 is involved in the tumorigenesis of ESCC, it may provide a new aspect for drug development.

No potential conflicts of interest were disclosed.

Grant support: National Natural Science Foundation of China (grant no. 30572130), and Capital Medical Development Research Fund (grant no. 2005-2020).

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.