HOXB7 overexpression promotes cell proliferation and correlates with poor prognosis in gastric cancer patients by inducing expression of both AKT and MARKs

Increased expression of HOXB7 has been reported to correlate with the progression in many cancers. However, the specific mechanism by which it promotes the evolution of gastric cancer (GC) is poorly understood. In this study, we sought to investigate the role of HOXB7 in GC by assessing HOXB7 expression in patient tissue and its correlation to clinical characteristics. We found that GC tissues showed increased expression of HOXB7 and that the HOXB7 expression was significantly associated with Lauren classification, invasion depth, lymphatic metastasis and poor prognosis, and could serve as an independent prognostic factor. To further investigate the role of HOXB7 in GC, we generated stable GC cell lines and both over-expressed and knocked down HOXB7 expression. Over-expression of HOXB7 in GC cell lines enhanced cell proliferation, colony formation, migration and invasion ability, whereas the opposite trends were observed upon reduction of HOXB7 expression by knockdown. These findings were further supported by our in vivo studies which show that HOXB7 expression can affect the GC cells' subcutaneous growth and lung metastases. A Phospho-MAPK Array Kit was used to explore the possible mechanism of HOXB7-induced cell proliferation and invasion. We found that the AKT signaling pathway and the two members of the MAPK pathway, were involved in those promoting effects. In conclusion, our results showed that increased expression of HOXB7 might play an important role in promoting GC proliferation, migration and invasion by inducing both AKT and MAPK pathways, thus resulting in progression of, and poor prognosis in GC patients.


INTRODUCTION
Despite improvements in surgical technique and adjuvant chemotherapy, gastric cancer (GC) remains a highly lethal disease. In China, the five year survival rate of GC is only 40% and GC ranks as the second most frequent cause of cancer-related deaths [1][2]. Unfortunately, GC is often diagnosed at advanced stages (III-IV), when metastatic disease with lymphatic spread is common. Better understanding the molecular mechanisms of GC is urgently needed to identify novel therapeutic targets and identify biomarkers for GC prognosis.

Research Paper www.impactjournals.com/oncotarget
The homeobox genes encode a family of transcription factors that are essential for regulating growth and differentiation during embryonic development and maintaining adult tissue homeostasis. They are frequently dysregulated in cancer where they variably impact tumor cells proliferation, migration, invasion, apoptosis [3][4].Thirty-nine HOX genes were categorized into four chromosomal clusters (A, B, C and D) have been reported in human. These four chromosomal clusters, each about 100 kb in length, are located on chromosomes 7, 17, 2, and 12, respectively [5]. As a member of class I homeobox genes, the transcription factor homeobox B7 (HOXB7) is known to play an important role in tumorigenesis in several cancers, including melanoma [6], breast [7], lung [8], colorectal [9] and pancreatic cancers [10]. Overexpression of HOXB7 is correlated with poor prognosis of patients with many different cancers, such as esophageal squamous cell carcinoma [11][12][13], colorectal cancer [9] and oral cancer [14]. Studies have shown that in these cancers, HOXB7 overexpression promotes cell proliferation, DNA repair [15], angiogenesis, epithelial mesenchymal transition (EMT) and cell survival [16]. Recently, it was reported that HOXB7 plays a dual role in HER2 positive breast cancer progression by delaying tumor formation, but promoting lung metastasis [17]. Although many studies have shown that HOXB7 plays an important role in cancer development, the biological functions of HOXB7 in GC tumorigenesis, progression and prognosis have not been well characterized. Here, we aimed to investigate the prognostic significance and possible functional mechanisms of HOXB7 in GC.

HOXB7 was up-regulated in GC tissues and cell lines
Expression of HOXB7 was analyzed by qPCR in 36 GC patients' tissues, in both tumor and paired adjacent noncancerous regions of the tissues. Comparative analysis indicated that expression of HOXB7 was significantly increased in GC tumor tissue relative to adjacent noncancerous gastric tissues (P<0.05, Figure 1A). Only 11% (4/36) of GC patient tissues showed lower HOXB7 expression in cancer tissues relative to paired adjacent noncancerous tissue ( Figure 1B). The expression levels of HOXB7 were much higher in GC (0.003344 ±0.0004176) tissues than in noncancerous tissues (0.0009040 ±0.0001908; P<0.01; Figure 1A). Furthermore, increased levels of HOXB7 protein was found in GC tissue relative to adjacent normal tissue levels in seven GC patient cases by Western blot ( Figure 1C). Additionally, Western blot analysis of six GC cell lines, including BGC-823, MKN-45, 7901, AGS, MKN-28 and MGC-803, exhibited increased levels of HOXB7 in comparison to the normal gastric mucosa cell line GES-1, and among the six GC lines, the HOXB7 expression was the lowest in MGC-803 and the highest in BGC-823 ( Figure 1D).

HOXB7 expression promotes proliferation and invasiveness of GC cell lines in vitro
To evaluate the possible role of HOXB7 in the tumorigenesis and invasiveness of human GC cells, we generated a HOXB7 shRNA knockdown GC cell line BGC-823-shB7, and a HOXB7 overexpression GC cell line, MGC-803-B7 (Figure 2A). MGC-803 and BGC-823 were selected for further study because these two GC cell lines were found to have relatively lower and higher endogenous HOXB7 expression, respectively, than other GC cell lines ( Figure 1D).
Overexpression of HOXB7 in the GC line MGC-803 resulted in significantly increased proliferation relative to the negative control cell line (MGC-803-NC) by the MTT assay ( Figure 2B; P<0.05). The colony formation assay revealed that these MGC-803-B7 also cells formed significantly more and larger colonies than control cells ( Figure 2C; P<0.01). The MGC-803 lines also showed significantly increased migration and invasive ability with overexpression of HOXB7 relative to control the control line by the Transwell assay ( Figure 2D; P <0.05). In order to further validate the phenotypic differences we saw upon expression of HOXB7 expression in GC cell lines, we knocked down endogenous HOXB7 in BGC-823, which normally expresses elevated levels of HOXB7, by shRNA-mediated knockdown ( Figure 2A). Knockdown of HOXB7 in BGC-823 resulted in significantly decreased proliferation rate, decreased colony formation ability, and decreased migration and invasion by the MTT assay, colony formation assay, migration and invasion assays, respectively (Figure 2B-2D; P<0.05).

HOXB7 expression promotes tumorigenesis and invasiveness of GC in vivo
The in vitro data indicated that the expression of HOXB7 conferred features of tumorigenesis and invasiveness to GC cell lines. To test whether expression of HOXB7 could promote these features of GC in vivo, MGC-803-B7 and BGC-823-shB7 and their respective controls (MGC-803-NC and BGC-823-shNC) cell lines were studied both through subcutaneous injection and also by tail vein injection in immunodeficient female nude mice.
In subcutaneous implantation models, shRNAmediated reduction in HOXB7 expression in the BGC-823 cell line caused significant reduction tumor growth relative to controls (1301.38 ± 294.25 mm 3 in BGC-823-shNC group vs. 255.79 ± 72.35 mm 3 in BGC-823-shB7 group, Figure 3A-3B; t-test, P<0.05). Implantation of the HOXB7-overexpressing cell line MGC-803-B7 resulted in significantly faster tumor growth than in the MGCwww.impactjournals.com/oncotarget paired (tumor and adjacent normal tissue) GC tissues by real-time PCR relative to GAPDH expression. A, HOXB7 mRNA expression levels are increased in GC tumor tissue relative to adjacent normal tissues (Cancer vs. Normal, P<0.05). B, HOXB7 mRNA expression level in 36 paired GC tissues (Log2 Cancer vs. Normal). C. Expression of HOXB7 protein in primary GC tissues (T) and the paired tissues adjacent noncancerous tissues (N) from the same patient by western blotting, and in each of the T and N tissues β-Actin was used as an internal control. D. The protein expression level of HOXB7 in GC cell lines. Expression levels were normalized with β-Actin. In the tail vein injection models, increased expression of HOXB7 resulted in significantly greater numbers of macroscopic lung metastatic cancer nodules in the mouse lung tissues, with the MGC-803-B7 (100 ± 21.36) and BGC-823-shNC (112.5 ± 17.25) groups showing increased foci relative to MGC-803-NC (15.83 ± 4.84) and BGC-823-shB7 group (17.5 ± 5.88) (t-test, P<0.05). These results indicate that HOXB7 expression level could promote GC subcutaneous growth and lung metastases in vivo.
Moreover, IHC staining of the subcutaneous implantation and lung metastases of the MGC-803-B7 and BGC-823-shNC groups displayed significantly elevated Ki-67 index than those in the MGC-803-NC vector and BGC-823-shB7 groups, as well as the expected stronger HOXB7 staining (P<0.05, Figure 3B, 3D, 3F and 3H). were not affected by HOXB7 expression (Figure 4). Furthermore, in the MGC-803-B7 group compared to the empty vector control cell line (MGC-803-NC), the expression levels of downstream targets of the AKT pathway and EMT markers, including FOXO1, BAX and E-cadherin, were down-regulated, whereas BCL-2, MDM2, N-cadherin and Vimentin were increased ( Figure  5A). In contrast, the opposite trends were seen in the BGC-823-shB7 group, compared with the corresponding negative controls ( Figure 5A). FOXO1 plays a pivotal role in tumor suppression by inducing growth arrest and apoptosis, and is an important target in AKT signaling pathways. AKT negatively regulates FOXO1 by direct phosphorylation, resulting in its inactivation and sequestration into the cytoplasm [18]. In this study, we showed that AKT could be activated by HOXB7 overexpression in GC cell lines, so we further sought to determine whether this difference in FOXO1 expression was mediated by a difference subcellular localization by immunofluorescence. The cells overexpressing HOXB7 (MGC-803-B7) demonstrate more cytoplasmic localization of FOXO1 than the negative control MGC-803-NC cells. In contrast, more nucleus localization of FOXO1 was observed in HOXB7 knock-down cell (BGC-823-shB7) than the negative control BGC-823-shNC cells (Supplementary Figure S1). These data indicate that a cytoplasmic subcellular location of FOXO1 was associated with increased HOXB7 expression. Because tumor epithelial mesenchymal transformation plays an important role in tumor invasiveness, a series of EMT markers were also examined by Western blot. When HOXB7 was overexpressed in MGC-803 cells, the epithelial marker E-cadherin was decreased and the mesenchymal markers N-cadherin and vimentin were increased. When the expression of HOXB7 in BGC-823 was knocked down by HOXB7-shRNA, the opposite trend in expression was observed ( Figure  5A). Additionally, the MGC-803-B7 line overexpressing HOXB7 displayed a more elongated, spindle cellular morphology, which may be consistent with the increased EMT markers observed (Supplemental Figure S1).
These findings indicate that HOXB7 likely promotes GC cells proliferation and survival in part through the AKT1/2-FOXO1-BCL-2-BAX axis. Two major family members of the MAPK pathway, ERK1/2 and p38α, which were increased in the context of HOXB7 overexpression, may play an important role in GC invasiveness and spread, in part by promoting the cancer cell epithelial mesenchymal transformation ( Figure 5B).
In order to further investigate the role of HOXB7 and its effect on AKT/MAPK signaling and cell migration and proliferation, three commercially available kinase inhibitors (p38 inhibitor IV, FR180204, Akt1/2 inhibitor) were selected to block AKT and MAPK signaling. Addition of these inhibitors significantly reduced the invasion and proliferation ability of both MGC-803-B7 and control MGC-803 cell lines both in the presence and absence of HOXB7 expression. However, treatment with these AKT/MAPK inhibitors resulted in less inhibitory effects on cell invasion and proliferation in the GC lines overexpressing HOXB7 (Supplementary Figure S2). These data indicate that the AKT/MAPK signaling pathway accounts for at least part of the HOXB7-mediated increase in invasiveness and proliferation, and that targeting this axis can inhibit these features in GC cell lines.

HOXB7 expression is increased in GC patient tissue and associated with poor prognosis
To further study the relationship between HOXB7 expression, clinical pathological parameters and prognosis, we first sought to validate the trend of increased HOXB7 expression in gastric tumor tissue. We used immunohistochemical staining for HOXB7 in a GC tissue microarray (TMA) containing 330 patient samples (Table 1). Based on the HOXB7 immunoreactivity scores, elevated levels of HOXB7 were detected in 59.1% (195/330) of cases, and decreased levels of HOXB7 were detected in 40.9% (135/330) of cases ( Figure 6F-6I). HOXB7 staining was confined to cancerous portions of tissues examined (Supplementary Figure S3).
The HOXB7 positive staining rate in the patients with lymph node metastasis was greater (73.8%, or 158/214) than in patients without lymph node metastasis (31.9%, or 37/116, P<0.05). Additionally, the HOXB7 positive rate in patients with distant metastasis (83.9%, or 26/31) was also higher than that in patients without distant metastasis (56.5%, or 169/299; P<0.05). HOXB7 positivity was also correlated with tumor size, as 72.8% of tumors from GC patients (size ≥5cm) showed positive HOXB7 staining, compared with 50.7% of GC tumors < 5cm (P<0.05). HOXB7 positive staining was also correlated with tumor grade, as 83.9% of Grade IV tumors showed HOXB7 staining while only 15.3% of Grade I GC tumors were positive (P<0.05). Finally, higher levels of tumors bearing HOXB7 positive-staining was also found to be correlated with Lauren classification, invasion depth (T Grade), and lymphatic metastasis (N Grade) (P<0.05, Table 1), but were not associated with age, sex, location, and histological type of GC (P>0.05, Table 1).
We also analyzed the relationship between HOXB7 expression and the prognosis of GC patients. Based on HOXB7 staining of tumors from this cohort of GC patients (n= 330), the mean survival time in patients with HOXB7 tumors was significantly shorter than that of the HOXB7 negative tumors (34. 83 ± 13.26 months vs. 50.91 ± 10.10 months, respectively, P<0.05), as was the 5-year survival rate (5.1% vs. 37.0%, P<0.05, Figure 6A), which was consistent across each TNM stage ( Figure 6B-6E, P<0.05). Cox multivariate analysis showed that TNM stage, Lauren classification, Vascular invasion and HOXB7 expression were independent prognostic factors ( Table 2).

DISCUSSION
Previous studies have demonstrated that cancers often exhibit aberrant expression of genes with functions in embryogenesis, particularly the Homeobox group. Expression of homeobox genes can induce tissue growth that contributes to tumor onset and progression [19]. HOXB7, a member of the HOX family of homeodomain transcription factors, is a critical developmental regulator of cancer cells. Increased expression of HOXB7 has been reported in several malignancies and has been implicated in influencing a number of cellular processes, including cell invasion, DNA repair, metastasis, and angiogenesis, and is thought to contribute to tumorigenesis and poor survival in many cancers [6-10, 15, 20]. In this study, we studied the effect of HOXB7 on cell proliferation and invasiveness by both in vitro and in vivo experiments with GC cell lines, and also investigated the prognostic significance of HOXB7 expression in a larger series of GC patient samples. In GC patients, increased expression of HOXB7 was observed in cancer tissue compared with adjacent normal tissues. Increased expression of HOXB7 in GC cell lines significantly increased the proliferation, colony formation and migration ability of GC cells in vitro. Our in vivo results show that over-expression of HOXB7 significantly accelerates subcutaneous tumor growth and increases the number of lung metastases. In contrast, these promotion effects were reversed by decreased expression of HOXB7 expression in GC cell lines (mediated by HOXB7-targeting shRNA), which was observed both in vitro and in vivo. Furthermore, in a cohort of 330 patient GC tumors, our study showed that HOXB7 immuno-positivity was observed in 59.1% of GC patients and was associated with increased tumor size, depth of invasion, lymph node metastasis, distant metastasis, and TNM stage. Moreover, the patients with positive HOXB7 staining had shorter survival time and poorer prognosis across all tumor stages.
AKT is a primary signaling transducer of the phosphoinositide 3-kinase (PI3K) pathway and contributes to cancer progression by inhibiting apoptosis, promoting cell proliferation and regulating migration and invasion in many cancer types [21][22]. Previous studies have shown that the PI3K/AKT pathway is frequently activated in GC and revealed that PI3K/AKT signal transduction pathway participates in cell proliferation with regulators such as FOXO1, BCL-2 and BAX [23][24]. Sasaki et al [22] reported that activation of the AKT pathway in GC could promote cancer cell proliferation and mediate cancer cell migration. Our results indicate that increased the expression of HOXB7 promoted the proliferation and invasion ability of GC cell lines and that these lines exhibited increased AKT1 (S473) and AKT2 (S474) phosphorylation. Furthermore, the downstream AKT pathway molecules associated with cell apoptosis and survival, such as FOXO1, BCL-2 and BAX, were also www.impactjournals.com/oncotarget involved in facilitating this effect. We therefore speculate that the HOXB7-mediated increase in GC cell line proliferation and invasiveness is at least partly dependent on effects on the AKT signaling pathway The MAPKs signaling pathway, including P38, ERK1/2 and JNK1/2, plays a crucial role in tumor cell proliferation, adhesion and migration [25][26][27][28][29][30][31] and these pathways are known to be activated in GC [32][33]. Epithelial to mesenchymal transition plays an important role in cancer invasion and metastasis and is characterized by a reduction in expression of epithelial markers, such as E-cadherin, while other mesenchymal markers and transcription factors, such as N-cadherin, snail, slug, and, vimentin, are up-regulated [34]. Previous studies showed that MAPKs signaling pathway was involved in EMT procession [35][36], include the ERK, p38 or JNK kinase pathway. Herein, we examined the effect of HOXB7 expression on MAPKs pathway-associated key functional protein kinases in GC cell lines, using the Human Phospho-MAPK Array Kit and Western blotting assays, as well as the effect on key EMT associated proteins. We found that ERK1 (T202/Y204), ERK2 (T185/Y187) and p38α (T180/Y182) phosphorylation were involved in the dynamic expression of HOXB7 in GC cell lines, while our data indicate that in this context the JNK pathway members, such as JNK1 (T183/Y185), JNK2 (T183/Y185), JNK2 (T221/Y223), are likely not involved. Additionally, we found that HOXB7 expression increased the expression of the key functional proteins of EMT, such as N-cadherin and vimentin, and these cell lines also exhibited increased invasion and proliferation. Therefore, HOXB7 enhancement of the migration and invasion ability of GC cells may be mediated by ERK and p38α pathways, two members of MAPKs pathway, at the phosphorylation level, resulting in a series of enhanced invasive phenotype changes, including EMT. However, the mechanism how HOXB7 affects AKT or MAPK signaling and whether HOXB7 could directly regulate the upstream molecules of them, such as PTEN, PI3K, MKK3/6 and RTKs, still need to be further explored.
In conclusion, our findings suggest that increased expression of HOXB7 might be a valuable prognostic

RNA extraction and qRT-PCR
RNA was extracted and purified from cell or tissue homogenates by using Trizol method. cDNA was prepared using Superscript cDNA synthesis kit (TAKARA, Japan) following the manufacturer's protocol. qRT-PCR was carried out using SYBR Premix Ex Taq (Takara, Japan) with cDNA-specific primers. GAPDH functioned as internal control and relative expression level was calculated using the 2 −∆∆ CT method. The mRNA specific primers for qRT-PCR as follows: HOXB7-F: GTCCCTGCCTACAAATCATC and HOXB7-R: GAAG CAAACG CACAAGAAGT; GAPDH-F: TGAAGGT CGGAGTCAACGG and GAPDH-R: CTGGAAGATG GTGATGGGATT. The PCR parameters were as follows: 95°C for 4 min, followed by 40 cycles of 95 °C for 10 s, 58 °C for 30 s and 72 °C for 30 s. At the end of the PCR cycles, melting curve analysis was performed.

MTs assay
Cellular proliferation was assayed using The Cell Proliferation MTs Assay Kit (G3580, Promega, Japan) following the manufacturer's protocol. All cell lines (stable over-expression and down-regulation of HOXB7 cells and their corresponding negative control cells) were seeded into 96-well plates with 3×10 3 cells in 200 μL culture medium per well. After attachment, 20μL of MTs reagents were added to each well every 24 hours. After incubating for an additional 4 hours, the absorbance was measured at 490 nm.

Colony formation assays
All cell lines (stable over-expression and downregulation of HOXB7 cells and their corresponding negative control cells) were seeded in 6-well plates with 100 cells and a final volume of 2 mL culture medium per well. The cells were maintained at 37°C in 5% CO 2 and the culture medium was changed every four days for the following three weeks. At the end of the experiment, the colony formation cells of each group were stained with H&E staining.

Migration and invasion assays
The cell migration and invasion assays were done in 24-well Transwell plates with or without pre-coated Matrigel, as described previously [37]. For migration assay, the stable cells and the corresponding negative control cells were seeded in the upper chamber of the Transwell system (8.0 mm, pore size; 3422, Millipore, USA) with 1×10 4 cells/well in 100 μL of serum-free RPMI-1640 medium, and the lower chamber was filled with 600μL 30% FBS RPMI-1640 culture medium. For invasion assay, 2×10 5 cells/well of the stable cells and the corresponding negative control cells were seeded in the upper chamber with pre-coated Matrigel (8.0 mm, pore size; ECM554, Millipore, USA), and the lower chamber was filled with 600 μL 30% FBS RPMI-1640 culture medium. After 24h or 48h of incubation, cells remaining on the top layers of the inserts were removed by cotton swab scrubbing, and cells on the lower surface of the membrane were fixed and stained with H&E staining. The cell numbers in five random fields (200×) were counted for each chamber, and the average value was calculated.

Phospho-MAPK array detection and westernblotting
The Human Phospho-MAPK Array Kit (ARY002B, R&D Systems, USA) was employed to detect the relative levels of phosphorylation of 26 kinases of all three major families of mitogen activated protein kinases (MAPKs), the extracellular signal-regulated kinases (ERK1/2), c-Jun N-terminal kinases (JNK1-3), and different p38 isoforms (α/β/δ/γ) to understand the effect of HOBX7 on these signaling pathways.
Cells were collected and lysed in cell lysates buffer and pipetted up and down to re-suspend and rock the lysates gently at 2-8 °C for 30 min, centrifuged at 14000g for 5 min, and the supernatant was transferred into a clean test tube. Sample protein concentrations were quantitated using the BCA method and then the extractions were aliquoted and stored at ≤-80°C.
The Array Detection steps were followed according to the manufacturer's protocol. Array membranes were blocked by using Array Blocking Buffer for one hour. A total of 200μg protein of each sample were added to separate tubes and adjust to a final volume of 1.5 mL with Array Buffer 1, and then 20 μL of Detection Antibody Cocktail was added to every separate tubes and incubated at room temperature for one hour. After blocking the membranes, the prepared Sample/Detection Antibody Cocktail Antibody mixtures were added into the 4-Well Multi-dish and incubated overnight at 2-8°C. After washing the membranes for three times, the diluted Streptavidin-HRP buffer was added and incubated for 30 minutes. After washing three times, the membranes were incubated with Chemi Reagent Mix and theChemiluminescence signal was detected using the Chemiluminescence Gel Imaging System (Bio-Rad, USA).

AKT/MAPK signaling validation
In order to validate whether HOXB7 truly regulates the migration and proliferation process via AKT/MAPK signaling, three commercially available inhibitors, Akt1/2 kinase inhibitor( A6730, SIGMA, USA), FR180204 (ERK inhibitor II, SML0320, SIGMA, USA) and p38 MAP Kinase Inhibitor IV (SML0543, SIGMA, USA), were choosenfor blocking AKT, ERK and p38α kinase activations, respectively, according the manufacturer's instructions. After MGC-803 cells adhered to the sixwell plate, 200 nM of Akt1/2 kinase inhibitor, 200 nM of FR180204 and 130 nM of p38 MAP Kinase Inhibitor IV, were added into the well for blocking the corresponding kinase activation, respectively. After 24h later, the MGC-803 cells were transfected with HOXB7 plasmid, to increase the expression of HOXB7, followed by collected for doing invasion and proliferation assays.

Xenograft analysis
Four to five-week-old female BALB/c athymic nude mice were purchased from Slac Laboratory Animal Co. Ltd. (Shanghai, China). All mice were housed and maintained under specific pathogen-free conditions and used in accordance with institutional guidelines and approved by the Use Committee for Animal Care.
To evaluate the effect of HOXB7 on gastric tumor progression, mice were inoculated subcutaneously in the right flank with a total of 2 × 10 6 cells which were suspended in 100 μL PBS, of MGC-803-B7, BGC-823-shB7, MGC-803-Vector or BGC-823-shNC cells. Tumor size was measured by a slide caliper and tumor volume was calculated as (length×width 2 )/2 every week.
To evaluate the effect of HOXB7 on gastric tumor metastasis, a total of 1×10 6 cells of BGC-823-shNC, BGC-823-shB7, MGC-803-Vector or MGC-803-B7 cells were injected into the lateral tail vein. Metastatic lung nodules were quantified after H&E staining using a dissecting microscope at endpoint.
After mice were sacrificed, subcutaneous tumor tissues or Lung metastastic tissues were rapidly taken out, a part of the tissue was put into liquid nitrogen frozen immediately and the rest of it was fixed in 10% neutral buffered formalin and embedded in paraffin. Sections of 4 μm were cut and stained with H&E and IHC staining using antibodies of HOXB7 and Ki-67 (#GA62661, Clone MIB-1, Ready to use, DAKO, USA). The Ki-67 proliferation index was determined by counting 1000 cells in hot spots and calculated as the percentage of positive nuclei by one senior pathologist.

Patient samples and tumor tissue microarray (TMA)
Thirty-six paired GC and adjacent non-tumor tissues were obtained from patients with primary gastric adenocarcinoma without any history of radiation or chemotherapy treatment prior to surgery were obtained at Zhejiang Provincial People's Hospital, Hangzhou, Zhejiang Province, China. After surgical removal, the cancer tissues were frozen immediately in liquid nitrogen and stored until extraction of RNA and protein.
In addition, 330 cases of GC samples were collected from gastrointestinal surgery and the department of pathology of Zhejiang Provincial People's Hospital, from January 1998 to January 2004. All of the patients have been followed up for over 5 years with the deadline of December 2009. The survival time was counted from the date of surgery to the follow-up deadline or date of death, which was mostly caused by carcinoma recurrence or metastasis. No patients received any radiotherapy or chemotherapy prior to surgery, and written informed consent was obtained before analysis. The GC patients' age ranged from 17 to 80 (with median as 58.0 years old) and all cases were classified according to the World Health Organization's pathological classification (2010) of tumors. The clinicopathological characteristics of the GC patients are summarized in Table 1.
The core tumor area (tumor occupying >50%) of every GC tissue wax block was determined and labeled by the pathologist through H&E stained sections. Then, every 35 cases of the core cancer tissues (about 2mm diameter) were taken from individual paraffin embedded GC blocks and were arranged in recipient paraffin microarray blocks (tissue array blocks) using a trephine. Finally 10 tissue array blocks were made, containing a total of 330 cases of the GC samples described above. Each block contained more than three internal controls consisting of normal gastric mucosa.

IHC staining and evaluation
The GC patient tumor tissue microarray (TMA) as described above was used for HOXB7 immunohistochemical detection. Each 4 μm TMA section was deparaffinized, rehydrated and then rinsed with PBS. Antigen retrieval was carried out in 0.01 M citrate buffer (pH 6.0) for 3 min using high pressure retrieval method. Then the sections were incubated with 3% H 2 O 2 for 10 min followed by 10% normal goat serum for 15 min at room temperature, in order to block endogenous peroxidase and non-specific antigen. Then the sections were incubated with rabbit anti-human HOXB7 polyclonal antibody (1:400 dilutions in PBS, H00003217-D01P, Abnova, Taiwan) overnight at 4°C. After rinsing with PBS for three times, the sections were incubated with biotin labeled secondary antibody for 20 min at room temperature, and then incubated with horseradish peroxidase conjugate polymer (Invitrogen, USA) for another 20 min at room temperature. Finally, 3, 3-diaminobenzidine (DAB) was used to visualize the signal development, and then the sections were counterstained with hematoxylin.
The immunoreactivity levels of each case were estimated under a light microscope by assessing the average signal intensity (on a scale of 0-3) and the proportion of cells showing positive staining (0, <5%; 1, 5-25%; 2, 26-50%; 3, 51-75%; 4, 76-100%) and were independently carried out by two pathologists without knowledge of the clinical data, as described previously [38]. The intensity and proportion scores were then multiplied to obtain a composite score; 0-3 was defined as negative and 4-12 as positive.

Statistical analysis
All statistical analyses were performed using the SPSS 13.0 statistical software. Comparisons between groups were performed with a 2-tailed paired Student's t test. The relationships between HOXB7 expression and clinicopathologic characteristics were tested using the Chisquare test. Survival curves were plotted by Kaplan-Meier method and compared by log-rank test. The significance of various survival related variables was assessed by Cox regression model in the multivariate analysis. P<0.05 was considered statistically significant.

CONFLICTS OF INTEREST
We declare that no conflicts of interest in relation to this article exists.