Hypoxia induced metastasis of gastric cancer through HIF1α/ANXA1/MMP-2 pathway


 BackgroundThe aim of this study was to investigate the mechanism of hypoxia induced metastasis of gastric cancer (GC). MethodThis study explored the clinical significance of hypoxia inducible factor (HIF) 1α and annexin A1 (ANXA1) in human GC by immunohistochemistry of a tissue microarray with 76 surgically resected GC specimens and further examined the mechanism of the role of HIF1α in GC.ResultsOur data demonstrated that HIF1α expression was dramatically increased in GC tissues. High HIF1α expressions were significantly correlated with lymph node metastasis and with worse survival in GC patients. In addition, HIF1α expression was an independent prognostic factor in GC patients and the combination of HIF1α and ANXA1 might serve as useful prognostic markers in GC patients. Furthermore, HIF-1α could promote GC migration and invasion through HIF1α/ANXA1/MMP-2 (Matrix metal proteinase 2) pathway.ConclusionsOur findings indicated that HIF-1α may serve as a promising prognostic biomarker and therapeutic target for inhibiting GC metastasis.

Loss of HIF-1α could sustain tumor metabolism and cellular proliferation by upregulating ANXA1 and they expressed in a mutual exclusive manner to in 37 cell lines of GC [11]. ANXA1 is one of the important proteins of the downstream pathways under hypoxia environment regulated by HIF1 [11]. As an important marker and an attractive therapeutic target of malignant tumor, ANXA1 is a 37-kDa protein belonging to the annexin superfamily, besides its multiple physiological functions, recent evidences implicated ANXA1 strongly affects migration and invasion of tumor cells [12][13][14].
In this study, we aimed to investigate whether hypoxia could regulate metastasis of gastric cancer through HIF1α/ANXA1 pathway. We rstly investigated the expressions of HIF-1α and ANXA1 in GC tissues and analyzed the clinicopathological and prognostic signi cance of HIF-1α and ANXA1 in GC patients. Furthermore, we found that hypoxia could improve migration and invasion of GC cells through HIF-1α/ANXA1/MMP-2 pathway. Our results indicated that HIF-1α may act as diagnostic and prognostic marker and may be a potential therapeutic target for GC.   Positive HIF-1α and ANXA1 staining appeared brown in cytoplasm with or without in nucleus. We graded positive staining according to both the stain intensity and the percentage of stained cells. Two pathologists independently examined all slides. The intensities of HIF-1α and ANXA1 staining were scored 0 to 3 (0 = negative, 1 = moderate; 3 = strong). The percentage of stained cells was scored 1 to 4: 1 (0-25% cells positively stained), 2 (26-50%), 3 (51-75%) and 4 (76-100%). Final scores of both two proteins positive staining were evaluated by the immunoreactive score (IRS), which was calculated by multiplying the intensity score by the percentage score of positive cells. The negative staining of the specimens was IRS: 0.

Patients and specimens
Protein expressions were also evaluated by Western blotting. Equivalent amounts of total proteins from frozen samples of surgically resected GC tissues and their adjacent non-tumor tissues were extracted as according to the extraction protocol of the total protein kit, KeyGen Biotech, China). The concentration determined using BCA protein kit, Beyotime Biotechnology, Shanghai, China. The procedure of western blot analysis was performed as described previously [15]. Protein samples were denatured, electrophoresed in SDS/polyacrylamide gels and transferred into polyvinylidene di uoride membranes (Millipore). Antibodies against the following proteins were used: HIF-1α at a 1:2,000 dilution, ANXA1 at 1:1,000 and anti-β-actin at 1:5,000.
After incubation with secondary antibody: horseradish peroxidase-conjugated goat anti-rabbit IgG (ABL3012,AbSci Washington US) at a dilution of 1:10,000, protein bands were exposed to ECL system.

Cell lines
The human GC cell lines HGC27 and MGC803 were obtained from the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Science (Shanghai, China

Gelatin zymography
Gelatin zymography was used to detect the activity of MMP-2. Cells (2.5 × 10 6 ) were seeded in 100-mm plates and cultured for 3 h, 12 h and 24 h. Then, proteins were collected and concentrated by Amicon Ultra-4-30 k centrifugal lters (Millipore, Billerica, MA, USA), 7500 g, 20  In order to detect the interaction between HIF-1 and ANXA1, and between HIF-1 and MMP-2, we incubated with antibodies of ANXA1 and MMP-2 respectively during the Western blotting.
Immuno uorescence staining Wound healing assay Wound healing assay was used to detect the ability of cell migration. After transfecting HGC27 and MGC803 cells with ANXA1 siRNA and control siRNA, cells were grown to con uency, wound lines were made by scraping closed Pasteur pipette tips across the con uent cell layer. Then, cells were washed by PBS three times to remove detached cells and debris. The width of wound was observed and measured after 0 6 and 12 hours.

Cell migration and invasion assays
Cell migration and invasion assays were performed by using modi ed two chamber migration and invasion assays with a pore size of 8 µm. Transwell lters were inserted with or without Matrigel (BD Biosciences, USA) coating for invasion and migration assays, respectively. The detailed conditions have been described previously [16].

Statistical analysis
Data has been expressed as the means ± SDs from at least three independent repeats. One-way analysis of variance (ANOVA) and multi-factorial ANOVA were used for comparing the differences between groups. Statistical analysis for the GC tissue microarray was performed by SPSS 16.0. The association between the staining of HIF-1α, ANXA1 and the clinicopathological parameters of GC patients, including gender, age, tumor size, TNM stage, lymph node and hematogenous metastasis were evaluated by the two-sided Fisher's exact test. The correlation between HIF-1α, ANXA1 expressions and patient survival was assessed by Kaplan-Meier and log-rank tests. Analysis of univariate and multivariate was analyzed by Cox regression model. Difference was considered signi cant when P < 0.05.

Results
The expression of HIF-1αand ANXA1 in GC and non-tumor tissues We performed immunohistochemistry with the GC tissue microarray to study whether HIF-1α and ANXA1 expressions are changed in human GC tissues. According to immunohistochemistry results we found positive HIF-1α and ANXA1 were stained in brown in cytoplasm with or without nuclear staining (Fig. 1A).
Positive HIF-1α expression was observed in approximately 82.9% of the GC versus ANXA1 35.5% (Fig. 1A, Table 1). The result of immunoreactivity scores (IRS) of HIF-1α and ANXA1 staining available from the GC tissue microarray were consistent with the ndings of immunohistochemistry analysis (Fig. 1B).
For veri cation of HIF-1α and ANXA1 expressions in GC tissues, we next collected 10 pairs of frozen GC samples and none cancer gastric tissues for further Western blot analysis. As shown in Fig. 1C, the expression of HIF-1α was higher versus ANXA1 in same GC tissues which indicated that HIF-1α and ANXA1 may be negative regulatory relationship.
Increased HIF-1α correlates with clinicopathological parameters in GC patients Assessment of the association between HIF-1α expression and the clinicopathological data is presented in Table 2. Our data showed that increased expression of HIF-1α was signi cantly correlated with the diameter of GC (P = 0.026). We also found increased HIF-1α expression was signi cantly correlated with lymph node metastasis (P = 0.011). Whereas, we did not nd signi cant correlation between HIF-1α with other clinicopathologic variables, including age, gender, TNM stage, tumor differentiation and hematogenous metastasis.
However, we did not nd signi cant correlation between ANXA1 with any clinicopathologic variables.
Assessment of the association between ANXA1 expression and the clinicopathological data is presented in supplementary Table 1.

HIF-1α serves as a potential prognostic indicator for GC patients
To further study whether the expressions of HIF-1α and ANXA1 in GC patients correlate with the prognosis of GC patients, Samples with the IRS of 0-4 were classi ed as low-expressing, and samples with the IRS of 5-12 were identi ed as having high HIF-1α and ANXA1 expressions. Our results revealed that high HIF-1α expression was correlated with worse overall survival in GC patients (P = 0.000, Fig. 1D). However, there was no correlation of ANXA1 with the prognosis of GC patients (P = 0.107, Fig. 1D).
Furthermore, to determine whether the expression of HIF-1α was an independent prognostic marker for GC patients, we analyzed HIF-1α, gender, age, tumour differentiation, tumor diameter, lymph node metastasis, TNM stage and hematogenous metastasis by univariate and multivariate Cox regression analysis. Univariate analysis (Table 3) revealed that HIF-1α, lymph node metastasis and TNM stage were correlated with survival of GC patients. In the multivariate analysis (Table 3), only HIF-1α remained as a signi cant independent prognostic factor of decreased survival of GC patients. Taken together, HIF-1α may serve as an independent prognostic factor in GC. Relationship between expressions of HIF-1α and ANXA1 in GC tissues and GC patients We searched for the correlation between HIF-1α and ANXA1. We found that the expression of HIF-1α was negatively correlated with ANXA1 (P = 0.032) (Spearman's rank correlation test) (Table 4), which was consistent with WB result (Fig. 1C).
Kaplan-Meier analysis was further applied to compare overall survival of GC patients according to combinations of HIF-1α and ANXA1 (Fig. 1D). Patients with low HIF-1α plus low ANXA1 had the longest survival of GC patients, while those with high HIF-1α plus high ANXA1 had worse survival.

Hypoxia increases HIF-1α expression and promotes GC cell migration and invasion
To explore the effect of hypoxia on GC cells, we detected the expressions of HIF-1α in GC cells and the migration and invasion of GC cells. The results showed that HIF-1α was increased after 6 h and 12 h after co-cultured with CoCl2 ( Fig. 2A). As HIF-1α and HIF-2α expressions were induced under hypoxia in HepG2 cells [17], we also detected HIF-2α expression. There's no change of HIF-2α expression being detected under hypoxia in both HGC27 and MGC803 cells ( Fig. 2A), we concluded that HIF-2α was not involved in our experiment.
We continued to examine the effects of hypoxia on the migration and invasion of GC cells via wound healing and transwell assays. The width of the scratches in the wound healing assay at 0 h, 6 h and 12 h (Fig. 2B) was analyzed. The result showed that the speeds of wound healing in HGC27 and MGC803 cells co-cultured with CoCl2 were faster than that in the control groups. In transwell lter assays, we found that hypoxia signi cantly enhanced the ability of HGC27 and MGC803 cells to migrate through transwell lter inserts (Fig. 3A). We further examined the effects of hypoxia on the invasion ability of GC cells. We found that hypoxia signi cantly enhanced the ability of HGC27 and MGC803 cells to invade through the transwell lter inserts (Fig. 3B).

Hypoxia promotes the migration and invasion of GC cells through HIF1α/ANXA1/MMP-2 pathway
To investigate the mechanisms that hypoxia regulates the migration and invasion of GC cells, we detected HIF1α, ANXA1, MMP-2, TIMP-2, MMP-9 and TIMP-1 protein expressions in HGC27 and MGC803 cells by western blot. Our data showed that HIF1α increased, ANXA1 decreased, MMP-2 increased and TIMP-2 downregulated by hypoxia in GC cells (Fig. 4A). We further detected the activity of MMP-2 by gelatin zymography and we found that the activity of MMP-2 was positively regulated by hypoxia (Fig. 4B). Therefore, we supposed that hypoxia could promote the migration and invasion of GC cells by regulating HIF1α/ANXA1/MMP-2 pathway.
To further validate our hypothesis, immunoprecipitation analysis clearly indicated interactions of HIF1α, ANXA1 and MMP-2 in both HGC27 and MGC803 cells (Fig. 4C). In addition to the immunoprecipitation analysis, we determined the subcellular localization of HIF1α, ANXA1 and MMP-2 via immuno uorescence staining. The result showed that proteins (HIF1α, ANXA1 and MMP-2) are all localized in the cytosol (Fig. 5A, 5B). In addition, after ANXA1 siRNA was transfected into GC cells (Fig. 5C), the expression of MMP-2 was upregulated (Fig. 5D, 5E). Our results con rmed that under hypoxia condition, HIF1α regulated the expression of MMP-2 through ANXA1.

Discussion
Hypoxia is de ned as the increase in consumption of oxygen or reduction of oxygen relatively to the supply in cells, tissues and organs. Solid tumor cells survived in a microenvironment of hypoxia known as hypoxic adaptation. HIF1α plays the critical role in tumor hypoxic adaptation and it may mediate series of downstream signaling pathways involved in tumor metastasis including epithelial to mesenchymal transition [18], invasion the extracellular matrix and the secondary growth of metastases [19]. CoCl 2 has been demonstrated to enhance HIF1α expression under its induced tumor hypoxia environment in vivo [18,20]. In this study, hypoxia condition was induced by CoCl 2 and our study showed the expression of HIF1α was strongly increased. Furthermore, our results showed that hypoxia promoted the migration and invasion of GC cells.
An important protein of one of the downstream pathways under hypoxia environment regulated by increased HIF1α is ANXA1 [11]. Growing evidences have suggested that ANXA1 contributes to the progress of tumor metastasis of several human tumors. ANXA1 has been reported regulating miR-562 and miR-26b through targeting NF-κb and angiogenesis to control metastasis of breast cancer [21]. Our previous study has showed that ANXA1 knockdown inhibits cell proliferation by inducing G1 phase cell cycle arrest and increases migration and invasion of pancreatic ductal adenocarcinoma (PDAC) cells through up-regulating MMP-9 expression and activity [16]. Our hypothesis was that HIF1α play crucial role in migration and invasion of GC cells under hypoxia condition through regulating ANXA1/MMP pathway.
In our present study, our data showed that under hypoxia condition, the expression of HIF1α increased, ANXA1 decreased, MMP-2 increased and TIMP-2 downregulated in HGC27 and MGC803 cells. We continuously detected the activity of MMP-2 by gelatin zymography and we found that the activity of MMP-2 was positively regulated by hypoxia. Furthermore, our immunoprecipitation and immuno uorescence results illustrated the interaction among HIF1α, ANXA1 and MMP-2. In addition, under hypoxia condition when knock down ANXA1 (siANXA1), as we expected, the expression of MMP-2 signi cantly increased in HGC27 and MGC803 cells. Together, these data support that HIF1α regulated MMP-2 through ANXA1. As

Conclusion
In summary, based on our ndings, we conclude that increased HIF-1α expression was correlated with lymph node metastasis and poor survival rate of GC patient. HIF-1α may serve as an independent prognostic factor in GC and the combination of HIF-1α and ANXA1 might serve as a useful prognostic marker in GC. Furthermore, we have con rmed that HIF-1α can promote GC migration and invasion HIF-1α could promote GC migration and invasion through HIF1α/ANXA1/MMP-2 pathway. Our results support the potency of HIF-1α as novel biological marker for GC and its migration and invasion, thus may provide new insight for both diagnosis and cellular therapy of GC.  Tables   Table 4,