MicroRNA miR-29a Inhibits Colon Cancer Progression by Downregulating B7-H3 Expression: Potential Molecular Targets for Colon Cancer Therapy

MiR-29a belongs to one of the subtypes of miRNAs known as non-coding single-stranded RNAs and is preferentially expressed in normal tissues. B7-H3, a member of the B7/CD28 immunoglobulin superfamily, was shown to be overexpressed in several solid malignant tumors, including colon cancer. In addition, it is associated with tumor progression and poor prognosis. We used immunohistochemical and Western blotting to assess B7-H3 protein expression levels in colon cancer and adjacent normal tissues and then compared their relationships with clinicopathological factors. Quantitative real-time reverse-transcription PCR was used to assess B7-H3 and miRNA-29a mRNA expression levels, and then their relationship and clinical significance were evaluated. In addition, colon cancer Caco-2 cells, which constitutively overexpress B7-H3, were transfected with lentivirus particles for miR-29a upregulation. Invasion and migration assays were carried out in vitro along with the establishment of a subcutaneous xenograft model in vivo to determine the role of miRNA-29a in colon cancer progression. The B7-H3 protein showed elevated expression in colon carcinoma and was relevant to TNM staging, lymph node metastasis, and reduced survival. Meanwhile, miR-29a was preferentially expressed in normal colon tissues, while B7-H3 transcript levels had no marked differences between tumor and normal tissue specimens. In vitro, miR-29a upregulation resulted in reduced B7-H3 expression. Furthermore, miR-29a upregulation reduced the invasive and migratory abilities of colon carcinoma cells. In animal models, upregulation of miR-29a slowed down the growth of subcutaneous xenotransplanted tumors and resulted in prolonged survival time. MiR-29a downregulates B7-H3 expression and accordingly inhibits colon cancer progression, invasion, and migration, indicating miR-29a and B7-H3 might represent novel molecular targets for advanced immunotherapy in colon cancer.


Introduction
Colon carcinoma ranks fourth among diagnosed malignancies, and it's mortality ranks fifth worldwide [1,2]. The improving quality of life, changes in eating habits, and pollution of the environment have resulted in increased incidence of colon cancer in recent years [3]. Surgery, chemotherapy, radiotherapy, and immunotherapy are the four major therapeutic modalities [4]. Although the survival rate of colon cancer patients has increased thanks to advanced and improved early diagnostic and therapeutic strategies, the five-year survival rate is only about 50% worldwide [5]. To obtain early prediction and effective therapeutic strategies, novel biological markers are urgently required for subsequent application in colon cancer patients.
Interestingly, elevated miRNA-29a amounts in colon cancer cases limited to stageIIconfer longer recurrencefree survival, indicating miR-29a constitutes a potential biomarker of recurrence risk in such patients [21]. However, how miRNA-29a regulates B7-H3 in colon cancer and the relationship between the two molecules as well as their clinical significance in colon cancer progression remain largely unknown. Here, B7-H3 and miRNA-29a expression levels were assessed in colon carcinoma to determine their relationship and clinical significance. In addition, invasion and migration assays were carried out in vitro along with the establishment of a subcutaneous xenograft model in vivo to determine the role of miRNA-29a in colon cancer progression.

Colon Cancer Cases and Clinical Specimens
Clinical specimens were obtained from 80 colon cancer patients who were surgically treated from January to December 2010 in the Department of General Surgery, The First Affiliated Hospital of Soochow University, Suzhou, China. Inclusion criteria were: patients who were diagnosed with colon cancer before surgery and the diagnosis was confirmed by postoperative pathology. Exclusion criteria were: (i) treatment by chemotherapy or radiotherapy; (ii) other colon surgery prior to the surgical treatment of colon cancer. Colon cancer tissues were collected following signed written informed consent from the patients. In this study, colon cancer cases included 44 men and 36 women, with ages between 46 and 93 years (mean, 71.6 years). A total of 41 and 39 tumors were found in left and right colons, respectively. And there were 32 and 48 stage I ~ II and III-IV cases according to the seventh edition of TNM staging criteria, respectively.
Tumor tissue samples and paired adjacent normal colon tissue specimens (obtained from colon tissues at over 5 cm away from the tumor edge) were simultaneously resected from the same patients. A portion of each tissue specimen was immediately immersed in 10% buffered formalin for fixation followed by paraffin embedding for IHC, and the other part was stored at − 80 °C for qRT-PCR. The study protocol had approval from the Ethics Committee of First Affiliated Hospital of Soochow University.

Survival Analysis
Survival was assessed from initial colon cancer surgery to colon carcinoma associated death. Clinical follow-up lasted 6 years until December 1, 2016, and data were obtained from telephone calls or return visits. After the primary surgery, follow-up was carried out every 3 months in the first 2 years, every 6 months in years 2 ~ 5 and yearly thereafter.

Immunohistochemical Staining
After paraffin embedding, 80 cases of tissue specimens were serially sectioned at 4 μm with a microtome. Deparaffinization was performed with xylene, and an ethanol gradient was employed for dehydration. The activity of endogenous peroxidase was inactivated by 3% H 2 O 2 for 15 min, and 3% BSA was used to block non-specific binding at 37℃ for 30 min. Then, The sections were incubated overnight with anti-B7-H3 (dilution, 1:200; mouse. no. ab105922; Abcam, China) primary antibody at 4 °C, and further incubated with a secondary antibody (dilution, 1:1000; rabbit anti-mouse. no. ab6728, Abcam, China) at 37˚C for 30 min. The tissue sections were counterstained with hematoxylin, dehydrated by gradient concentrations of ethanol. and sealing with neutral resin.
For evaluating B7-H3 expression after immunohistochemical staining, ten high power fields were randomly selected under optical microscope to detect staining range and intensity. The staining range of cells was scored as follows: no staining, 0; light yellow staining, 1; yellow staining, 2; yellowish-brown staining, 3. Staining intensity was scored as follows: < 5%, 1; 5% ~ 25%, 1; 26% ~ 50%, 2; and > 50%, 3. Then, range and intensity scores were multiplied to obtain the final score (≤ 1 indicated negative expression while > 1 reflected positive expression). The immunohistochemically stained sections were evaluated by two observers blinded to the patients' clinicopathological parameters and prognoses.

Western Blotting
Total protein extraction from 80 cases of tissue specimens (cancerous or non-cancerous) was performed with protein lysis buffer. Equal amounts of protein were resolved by 10% SDS-PAGE followed by transfer onto PVDF membranes. Upon blocking with 5% non-fat milk at room temperature for 1 h, the membranes were incubated overnight with anti-B7-H3 (dilution, 1:200; mouse. no. ab105922; Abcam, China) and anti-GAPDH (dilution, 1:500; mouse. no. Ab8245; Abcam, China) primary antibodies at 4˚C. A secondary antibody (dilution, 1:2000; rabbit anti-mouse. no. ab6728, Abcam, China) was added for 1 h under shaker conditions at room temperature. After three PBST washes, immunoreactive bands were finally visualized on an ECL imaging system (BIO-RAD, USA). Band intensities were semi-quantitatively assessed with Image J (Rawak Software, Germany). Assays were run in triplicate.

Cell Lines, Cell Culture and Stable-Transfected Cell Line Establishment
Caco-2 (human colon carcinoma) cells were provided by the Institute of Clinical Immunology of Jiangsu Province, Suzhou, China. Caco-2 cells were cultured in complete medium containing DMEM (Hyclone, USA), 10% heatinactivated fetal bovine serum (FBS) (Sijiqing Biological Manufacturer, HangZhou, China) and 1% penicillin-streptomycin (provided by the Institute of Clinical Immunology of Jiangsu Province), in a humid environment at 37℃containing 5% CO 2 .
The lentivirus stably over-expressing miR-29a (LV-miR-29a), anti-miR-29a (LV-anti-miR-29a, which can downregulate miR-29a) and negative control (LV-NC) were provided by Shanghai GeneChem CO., Ltd (Shanghai, China). The plasmid vector of the lentivirus carried the GFP sequence, and could appear green under a fluorescent microscope. Caco-2 cells prepared for transfection were in the logarithmic growth phase, and a cell suspension containing 5 × 10 4 cells was seeded in a 6-well cell culture plate. Then, 20 μl of lentiviral particles (MOI = 100; concentration, 1 × 10 8 TU/ ml) was added to transfect Caco-2 cells at a confluency of 30%. Four cell groups were obtained: Blank control group (Control), not transfected by lentivirus; over-expressing miR-29a group (LV-miR-29a), transfected by lentivirus overexpressing miR-29a; Negative control group (LV-NC), transfected by the control lentivirus without miR-29a expression; downregulate miR-29a group (LV-anti-miR-29a), transfected by the lentivirus stably downregulating miR-29a. After confirming that the cell growth was not inhibited by lentiviruses 12 h after transfection, the culture medium was replaced by the normal culture medium, followed by incubation in a medium containing puromycin (5 µg/ml) for selecting stably transfected cells. Fluorescence microscopy was carried out to evaluate GFP expression 3 days after transfection. Finally, miR-29a levels in stable-transfected cells were assessed by qRT-PCR.

Wound Healing Assay
Four groups of infected Caco-2 cells (Control, LV-miR-29a, LV-NC and LV-anti-miR-29a) were incubated in 6-well plates at 1 × 10 6 cells/well in presence of 5% CO 2 at 37 °C for 24 h. Then, a 200 µl pipette tip was employed to generate two parallel straight lines in confluent monolayers followed by two PBS washes. Next, DMEM without FBS was added for further incubation. An inverted microscope (provided by the Institute of Clinical Immunology of Jiangsu Province) was used to capture wound healing images at 0, 6, 12, and 18 h, respectively, for measuring wound healing widths. Each assay was performed in triplicate. Then, three groups of infected Caco-2 cells (Control, LV-NC and LV-anti-B7-H3) were implemented for wound healing assay using the same experimental method as above.

Transwell Invasion Assay
Transwell matrigel invasion chambers were prepared by adding 20 µl matrigel diluted at 1:3 in DMEM without FBS into the upper chambers for incubation at 37 °C for 4 h. The lower chambers were added 600 µl FBS-free DMEM. Then, single-cell suspensions (100 µl; 5 × 10 5 /ml) incubated for 12 h with FBS-free DMEM were added into transwell matrigel invasion chambers. After incubation with 5% CO 2 at 37 °C for 48 h, the upper chambers were washed with PBS and wiped with cotton swabs to remove non-invading cells. Next, 4% methanol was used to fix cells for 30 min, followed by staining with 1% crystal violet for 20 min. Invading cells were counted in 5 randomly selected fields under a microscope. Each assay was performed in triplicate.

Mouse Subcutaneous Xenograft Model
Six-week-old female BALB/c mice (Suzhou Supusi Biological Technology co., SuZhou, China) were assigned to four groups, including control, LV-miR-29a, LV-NC, and LV-anti-miR-29a groups (five mice per group). A total of 1 × 10 6 wild-type and stable-transfected (LV-miR-29a, LV-NC and LV-anti-miR-29a) Caco-2 cells, respectively, were subcutaneously administered on the right flank of the animals. Mice were checked every day, and the subcutaneous tumor formation rate was 100%. Tumor diameters were measured with Vernier calipers every two days starting from the fifth day after injection, and the survival time of each mouse was recorded. This study was approved by the local institutional animal care and use committee.

Statistical Analysis
The GraphPad Prism 7.0 software (GraphPad Software, USA) was used for all analyses. Enumeration data were presented as rate, with the χ 2 test employed for group comparisons. Measurement data are mean ± standard deviation (SD), and were assessed by Student's t test. Kaplan-Meier curves were employed to determine survival, which was assessed by the log rank test. P < 0.05 indicated statistical significance.

B7-H3 Levels in Colon Cancer and Adjacent Normal Colon Tissues
Representative Immunohistochemical images are shown in Fig. 1. B7-H3 immunohistochemical signals were mainly localized to the membrane, cytoplasm, and/or nucleus in colon cancer cells, appearing yellowish-brown (Fig. 1). B7-H3 levels in non-cancerous specimens were lower compared with those of tumor tissues (Fig. 1). B7-H3 amounts in colon cancer and paired adjacent non-cancerous colon tissue specimens are shown in Table 1. There were 58 (72.5%) cases with B7-H3 expression, while 22 (27.5%) cases were negative in the colon cancer group. Meanwhile, 26 (32.5%) and 54 (67.5%) cases, respectively, were positive and negative for B7-H3 in non-cancerous samples. B7-H3 levels in colon cancer samples were significantly higher compared with those of adjacent non-cancerous counterparts (P < 0.001, Table 1).

Associations of B7-H3 with Clinicopathological Factors
B7-H3 in colon carcinoma was tightly associated with TNM stage as well as lymph node metastasis, suggesting B7-H3 participated in tumor progression. There were more patients with positive B7-H3 signals (81.3%) in stage III-IV disease compared with 59.4% obtained for stage I-II cases (P = 0.032), indicating that B7-H3 promoted cancer invasion. Similarly, more patients with lymph node metastasis had positive B7-H3 signals (83.3%) in comparison with 60.5% found in cases with no nodal metastasis (P = 0.023), which suggested that B7-H3 contributed to cancer migration (Table 2). Meanwhile, B7-H3 in colon carcinoma showed no associations with gender, age, and tumor location, differentiation, and distant metastasis (P > 0.05, Table 2).

Associations of B7-H3 with Patient Survival
The patient follow-up rate was 92.5%. And five-year survival rates were 56.9% and 81.8% in colon carcinoma cases with positive and no B7-H3 signals, respectively. According to univariate analysis, there were tight associations of B7-H3 expression with survival in colon carcinoma, with cases expressing B7-H3 showing reduced survival compared with those displaying no B7-H3 expression (P = 0.022, Fig. 2).

MiR-29a Expression and B7-H3 mRNA and Protein Levels in Colon Carcinoma
A total of 80 colon carcinoma-and paired adjacent noncancerous tissue specimens were assessed for B7-H3 mRNA and miR-29a levels in order to detect B7-H3 and miR-29a expression patterns. B7-H3 mRNA expression levels in colon carcinoma were similar to those of non-cancerous samples (P > 0.05, Fig. 3A). However, B7-H3 protein levels were obviously higher in colon cancer compared with normal colon tissues, in which no overt signals were obtained by Western blotting (Fig. 3C). At the mRNA level, miR-29a amounts in normal tissue samples were significantly elevated

Associations of miR-29a with Clinicopathological Factors
MiR-29a in colon carcinoma was obviously associated with TNM stage as well as lymph node metastasis, suggesting miR-29a was related to tumor progression. There were more patients with positive miR-29a signals (37.5%) in stage I-II disease compared with 14.6% obtained for stage III-IV cases (P = 0.018), indicating that miR-29a inhibited cancer invasion. Similarly, more patients without lymph node metastasis had positive miR-29a signals (34.2%) in comparison with 14.3% found in cases with nodal metastasis (P = 0.037), which suggested that miR-29a aslo inhibited cancer migration ( Table 3). The results above showed that miR-29a inhibited tumor progression which was opposite to B7-H3. Meanwhile, miR-29a in colon carcinoma showed no associations with gender, age, and tumor location, differentiation, and distant metastasis (P > 0.05, Table 3).

Effects of miR-29a on Caco-2 Cell Invasion and Migration
In the wound healing assay, wound healing widths at 6, 12, and 18 h, respectively, showed differences (P < 0.05, The average value of mRNA expression levels in adjacent normal colon tissues was set as 1, to derive relative amounts in colon cancer tissues. There was no significant differences between colon cancer and adjacent normal colon tissues in B7-H3 mRNA expression levels (P > 0.05). (B) MiR-29a expression levels in colon cancer tissues were obviously higher than those of colon cancer tissues. (C) B7-H3 protein levels in colon cancer tissues were remarkably higher than those of adjacent normal colon tissues. *P < 0.05, **P < 0.01, ***P < 0.001 Fig. 6A, B). Wound healing were significantly reduced in LV-miR-29a transfected Caco-2 cells compared with the LV-NC group (P < 0.05, Fig. 6A, B). Conversely, wound healing were increased in LV-anti-miR-29a transfected Caco-2 cells compared with the LV-NC group (P < 0.05, Fig. 6A, B). LV-NC harboring Caco-2 cells and controls showed no differences (P > 0.05, Fig. 6A, B). Wound healing assay data suggested miR-29a might inhibit colon cancer cell migration.

Effects of miR-29a on Subcutaneous Xenotransplanted Tumor and Mouse Survival
The growth rates of mouse subcutaneous xenotransplanted tumors in the four groups differed, and survival was analyzed by Kaplan-Meier curves. The LV-miR-29a group showed overtly smaller tumors compared with the LV-NC group (P = 0.027, Fig. 7A). Meanwhile, tumors in mice administered LV-anti-miR-29a expressing cells were much larger than those of the LV-NC group (P = 0.020, Fig. 7A). The LV-NC and control groups showed no differences (P > 0.05, Fig. 7A). The LV-miR-29a group showed longer survival time while the LV-anti-miR-29a presented shorter survival time, compared with LV-NC mice (P = 0.024, Fig. 7B). No differences between LV-NC mice and controls were found (P = 0.004, Fig. 7B).

Discussion
Previously, B7-H3 has been described as an immunostimulatory molecule that stimulates CD4 + and CD8 + T cell proliferation, specifically promoting IFN-γ secretion [23]. However, B7-H3 was recently demonstrated to be an immuno-inhibitory molecule that could inhibit the body's immune surveillance toward tumors by suppressing Th1 cell mediated immune reactions [24]. High B7-H3 levels in multiple malignant tumors, including colon cancer, and its anti-tumor effects have attracted growing attention. Multiple reports suggest elevated B7-H3 levels are associated with tumor stage, lymphatic metastasis, and postoperative recurrence and metastasis [13][14][15][16]. Chen and collaborators even revealed that B7-H3 serves as an tumor-associated antigen, directly controlling tumor invasion and metastasis [25]. MiR-29a as an important miR-29 subtype is lowly expressed in different types of solid tumors. B7-H3 protein levels are negatively correlated with miR-29a expression in solid tumors and cancer cell lines [20]. It is known that miR-29a targets B7-H3 3'UTR, which contains a single conserved miR-29a binding site [20,26]; this suggested that miR-29a may downregulate B7-H3 protein expression.
As shown above, B7-H3 displayed elevated amounts in colon carcinoma samples. In addition, high B7-H3 levels were positively associated with TNM stage, nodal metastasis, and poor prognosis, corroborating findings by Ingebrigtsen and colleagues [27]. We also demonstrated that miR-29a levels were reduced in colon cancer tissue samples, and miR-29a expression levels were positively associated with TNM stage and nodal metastasis, suggesting miR-29a may exert anti-tumor effects. Recently, many studies supported the anti-tumor role of miR-29a in cancer progression. Pasqualini L, et al. showed that in primary prostate cancer, miR-29a levels are reduced in malignant tissues in comparison with  [29]. Ling C, et al. indicated miR-29a may suppress gastric cancer growth [30]. However, some studies showed that serum miR-29a was significantly higher in colon cancer patients than in healthy individuals and in colorectal liver metastasis (CRLM) patients than in CRC patients [31][32][33]. Besides, it was also reported that miR-29a was significantly higher in colorectal adenomas compared to normal colonic mucosa Fig. 4 MiR-29a downregulates B7-H3. (A) MiR-29a expression levels were determined by qRT-PCR after Caco-2 cell transfection with lentiviral particles harboring miR-29a, anti-miR-29a, and negative control, respectively. The average value of miR-29a expression levels in the LV-NC group was set as 1, to derive the relative expression levels in the other groups. The miR-29a expression levels in the LV-miR-29a group were higher compared with those of the LV-NC group; the anti-miR-29a group showed the opposite trend. (B) B7-H3 expression levels were determined by flow cytometry and were lower in the LV-miR-29a group compared with the LV-NC group, but higher in the anti-miR-29a group. A non-specific MoAb served as the staining control. (C) B7-H3 expression levels were determined by Western blot and were lower in the LV-miR-29a group compared with the LV-NC group, but higher in the anti-miR-29a group.*P < 0.05, **P < 0.01, ***P < 0.001 1 3 [34]. The above contrary results may be that the effects and regulatory mechanisms of miR-29a are complex and different in serum and colon tissues or other tissues, and the actual ways of how miR-29a act as regulators need to be further verified. Nevertheless, few studies have researched and confirmed miR-29a's role in colon cancer in vitro and in vivo. By wound healing and transwell invasion assays, we demonstrated that miR-29a delayed Caco-2 cell invasion and migration in vitro. These results suggested miR-29a plays a critical role in colon cancer cell invasion and migration. Besides, we also demonstrated that B7-H3 promoted Caco-2 cell invasion and migration in vitro, suggesting that miRNA-29a may work by downregulating B7-H3. Moreover, by establishing a subcutaneous xenotransplanted tumor model in mice, miR-29a could inhibit the growth of xenotransplanted tumors, thereby resulting in improved survival. These findings suggested that miR-29a may affect colon cancer malignancy in vivo.

Conclusion
In summary, B7-H3 promotes colon cancer progression resulting in poor prognosis, and miR-29a inhibits colon cancer cell invasion and migration both in vitro and in vivo. Moreover, miR-29a exerts anti-tumor effects by downregulating B7-H3 expression. These findings suggest miR-29a and B7-H3 may be potential molecular targets for colon cancer therapy.