Milk Fat Globule-EGF Factor 8 Contributes to Progression of Hepatocellular Carcinoma

Milk fat globule-EGF factor 8 (MFG-E8) is an anti-inflammatory glycoprotein that mediates a wide spectrum of pathophysiological processes. MFG-E8 has been studied as a key regulator of cancer cell invasion, migration, and proliferation in different tissues and organs. However, potential roles of MFG-E8 in the growth and progression of liver cancer have not been investigated to date. Here, we analyzed 33 human hepatocellular carcinoma (HCC) samples and found that levels of MFG-E8 expression were significantly higher in HCC cells than in normal liver tissues. In addition, our in vitro gain-of-function study in three different HCC cell lines revealed that overexpression of MFG-E8 promoted the proliferation and migration of HCC cells, as determined by RT-qPCR, MTT assays, and wound healing analyses. Conversely, an MFG-E8 loss-of function study showed that proliferation capacity was significantly reduced by MFG-E8 knockdown in HCC cells. Additionally, MFG-E8 activity-neutralizing antibodies profoundly inhibited both migration and proliferation of HCC cells, attenuating their tumorigenic properties. These reductions in migration and proliferation were rescued by treatment of HCC cells with recombinant MFG-E8 protein. Furthermore, an in vivo HCC xenograft study showed that the number of proliferating HCC cells and tumor volume/weight were all significantly increased by MFG-E8 overexpression, compared to control mice. These results clearly show that MFG-E8 plays an important role in HCC progression and may provide a basis for future mechanistic studies and new strategies for the treatment of liver cancer.


Introduction
Hepatocellular carcinoma (HCC) is the fifth most common cancer worldwide and the third-most common cause of cancer mortality [1]. Although multidisciplinary therapeutic strategies have been proven effective for HCC treatments, the five-year survival of HCC patients remains poor due to frequent tumor recurrences after surgery [2]. Therefore, elucidating the molecular mechanism 2. Results

MFG-E8 Expression is Significantly Upregulated in Liver Tumor Tissues and HCC Cell Lines
The parenchyma of normal liver tissues showed diffuse and moderate positive cytoplasmic immunostaining for MFG-E8 ( Figure 1A, left). The area of the stroma in portal tract was negative for MFG-E8 expression, but the bile ducts showed strong positive signals for MFG-E8 ( Figure 1A, dashed circle in left panel and inset). Notably, a profound increase in MFG-E8 was detected in HCC tissues ( Figure 1A, right). Strong MFG-E8 signals were primarily confined to tumor lobules surrounded by thick fibrous septa that were negative for MFG-E8. Based on double immunofluorescence staining, we observed that normal tissues showed strong positive signals of MFG-E8 in ductal areas ( Figure 1B: Normal), as observed in Figure 1A. The expression levels of MFG-E8 and α-fetoprotein (AFP), a marker of HCC, were both significantly increased in HCC tissues ( Figure 1B,C) and the immunofluorescence signals were overlapped in areas of high malignancy ( Figure 1B: HCC high). Prominent MEG-E8 signals were also seen in metastatic HCC tissues, but low signals of AFP were detected in these areas ( Figure 1B: Meta HCC). RT-qPCR analysis also revealed that the expression levels of MFG-E8 in human HCC tissues were significantly upregulated compared to levels in normal liver tissues ( Figure 1D). We also investigated the expression of MFG-E8 in three different HCC cell lines (Huh7, HepG2, and Hep3B). Consistent with the tissue data, RT-qPCR and ELISA results demonstrated that the levels of gene expression ( Figure 1E) and protein secretion ( Figure 1F) of MFG-E8 were both significantly increased in HCC cell lines compared to primary normal hepatocytes, although the levels differed between the cell lines. In particular, Huh7 cells showed the highest level of MFG-E8 expression and secretion among the three HCC cell lines tested.
Cancers 2020, 12,403 3 of 16 lines (Huh7, HepG2, and Hep3B). Consistent with the tissue data, RT-qPCR and ELISA results demonstrated that the levels of gene expression ( Figure 1E) and protein secretion ( Figure 1F) of MFG-E8 were both significantly increased in HCC cell lines compared to primary normal hepatocytes, although the levels differed between the cell lines. In particular, Huh7 cells showed the highest level of MFG-E8 expression and secretion among the three HCC cell lines tested. The mRNA expression levels of MFG-E8 in normal tissues and HCC tissues (D) and in human primary hepatocytes (Phh) and HCC cell lines (Huh7, HepG2, and Hep3B) (E). mRNA expression data were normalized to GAPDH levels and expressed as relative values. (F) The levels of MFG-E8 protein secreted from human Phh and HCC cells. Data represent the mean ± S.D. * p < 0.05 versus normal tissue (C and D) and Phh (E and F) by a two-tailed Student's t-test.

MFG-E8 Promotes Proliferation of HCC Cells
We next investigated the effects of MFG-E8 on the proliferation of HCC cells in vitro in three different hepatic tumor cell lines (Huh7, HepG2, and Hep3B) using both gain-and loss-of-function approaches. The gain-of-function study was performed by overexpressing MFG-E8 in three HCC cell lines using a lentiviral vector (LV-MFG-E8). RT-qPCR and ELISA results confirmed that the levels of MFG-E8 expression and secretion significantly and profoundly increased after MFG-E8 overexpression in all three cell lines compared to both the control cell line transfected with a were measured using ImageJ software and expressed as relative values to those in normal tissues. (D,E) The mRNA expression levels of MFG-E8 in normal tissues and HCC tissues (D) and in human primary hepatocytes (Phh) and HCC cell lines (Huh7, HepG2, and Hep3B) (E). mRNA expression data were normalized to GAPDH levels and expressed as relative values. (F) The levels of MFG-E8 protein secreted from human Phh and HCC cells. Data represent the mean ± S.D. * p < 0.05 versus normal tissue (C,D) and Phh (E,F) by a two-tailed Student's t-test.

MFG-E8 Promotes Proliferation of HCC Cells
We next investigated the effects of MFG-E8 on the proliferation of HCC cells in vitro in three different hepatic tumor cell lines (Huh7, HepG2, and Hep3B) using both gain-and loss-of-function approaches. The gain-of-function study was performed by overexpressing MFG-E8 in three HCC cell lines using a lentiviral vector (LV-MFG-E8). RT-qPCR and ELISA results confirmed that the levels of MFG-E8 expression and secretion significantly and profoundly increased after MFG-E8 overexpression in all three cell lines compared to both the control cell line transfected with a nontargeting control lentivirus vector (LV-NTC) and untransfected wild-type (WT) cells (Figure 2A,B). Importantly, a cell proliferation assay revealed that the MFG-E8 gain of function (LV-MFG-E8) led to significant increases in tumor cell proliferation in all cell lines tested compared to control (LV-NTC) and WT cells ( Figure 2C).
Cancers 2020, 12, 403 4 of 16 nontargeting control lentivirus vector (LV-NTC) and untransfected wild-type (WT) cells (Figure 2A and B). Importantly, a cell proliferation assay revealed that the MFG-E8 gain of function (LV-MFG-E8) led to significant increases in tumor cell proliferation in all cell lines tested compared to control (LV-NTC) and WT cells ( Figure 2C). The loss-of-function study was conducted using a siRNA that silences the expression of MFG-E8 and an anti-MFG-E8 antibody that directly inhibits MFG-E8 activity. In all three HCC cell lines, knockdown of MFG-E8 by siRNA interference (siRNA-MFG-E8) significantly and markedly reduced the expression levels of MFG-E8 compared to control (siRNA-NTC) and WT cells ( Figure 3A). The silencing of MFG-E8 expression in HCC cell lines resulted in a significant attenuation of cell growth 3 days post-infection compared with control (WT and siRNA-NTC) treatments ( Figure 3B). Interestingly, direct inhibition of MFG-E8 activity using anti-MFG-E8 antibodies showed a more potent antiproliferative effect in HCC cell lines, almost completely abolishing HCC cell growth 24-72 h after antibody treatment ( Figure 3C). The loss-of-function study was conducted using a siRNA that silences the expression of MFG-E8 and an anti-MFG-E8 antibody that directly inhibits MFG-E8 activity. In all three HCC cell lines, knockdown of MFG-E8 by siRNA interference (siRNA-MFG-E8) significantly and markedly reduced the expression levels of MFG-E8 compared to control (siRNA-NTC) and WT cells ( Figure 3A). The silencing of MFG-E8 expression in HCC cell lines resulted in a significant attenuation of cell growth 3 days post-infection compared with control (WT and siRNA-NTC) treatments ( Figure 3B). Interestingly, direct inhibition of MFG-E8 activity using anti-MFG-E8 antibodies showed a more potent antiproliferative effect in HCC cell lines, almost completely abolishing HCC cell growth 24-72 h after antibody treatment ( Figure 3C).

MFG-E8 Promotes Migration of HCC Cells
Cell migration is involved in many pathological processes, including invasion, angiogenesis, and metastasis, which are essential activities of tumor cells. Wound healing assays are widely used methods for investigating of carcinoma cell migration [12]. We examined the effects of MFG-E8 on the migration of HCC cells in response to a mechanical scratch wound in vitro. The results showed that MFG-E8 overexpression (LV-MFG-E8) significantly increased the migratory ability of Huh7 cells, closing the open scratched area faster than control LV-NTC and WT cells ( Figure 4A and B; data obtained from HepG2 and Hep3B are provided in Figure S1).

MFG-E8 Promotes Migration of HCC Cells
Cell migration is involved in many pathological processes, including invasion, angiogenesis, and metastasis, which are essential activities of tumor cells. Wound healing assays are widely used methods for investigating of carcinoma cell migration [12]. We examined the effects of MFG-E8 on the migration of HCC cells in response to a mechanical scratch wound in vitro. The results showed that MFG-E8 overexpression (LV-MFG-E8) significantly increased the migratory ability of Huh7 cells, closing the open scratched area faster than control LV-NTC and WT cells ( Figure 4A,B; data obtained from HepG2 and Hep3B are provided in Figure S1). In contrast, the motility of Huh7 cells was steadily and markedly impaired by treatment with anti-MFG-E8 antibodies (1 μg/mL) during the 72 h of measurement ( Figure 5A and B; data obtained from HepG2 and Hep3B are provided in Figure S2 and S3). Silencing of MFG-E8 expression using siRNA (siRNA-MFG-E8) also significantly decreased the migratory capability of Huh7 cells compared with both control siRNA-NTC and WT cells 48 h after transfection ( Figure 5C and D; data obtained from HepG2 and Hep3B are provided in Figure S4). However, the reduction in motility of MFG-E8 siRNA-transfected HCC cells was rescued by the addition of recombinant human MFG-E8 protein (rhMFG-E8, 500 ng/mL) ( Figure 5C and D). In contrast, the motility of Huh7 cells was steadily and markedly impaired by treatment with anti-MFG-E8 antibodies (1 µg/mL) during the 72 h of measurement ( Figure 5A,B; data obtained from HepG2 and Hep3B are provided in Figures S2 and S3). Silencing of MFG-E8 expression using siRNA (siRNA-MFG-E8) also significantly decreased the migratory capability of Huh7 cells compared with both control siRNA-NTC and WT cells 48 h after transfection ( Figure 5C,D; data obtained from HepG2 and Hep3B are provided in Figure S4). However, the reduction in motility of MFG-E8 siRNA-transfected HCC cells was rescued by the addition of recombinant human MFG-E8 protein (rhMFG-E8, 500 ng/mL) ( Figure 5C,D).

MFG-E8 Activities Are Involved in the Integrin-Akt Signaling Pathway in HCC Cells.
We next investigated the potential signaling pathways mediated by MFG-E8 in HCC cells. Western blot analyses showed that phosphorylation of Akt was markedly increased in the HCC cells overexpressing MFG-E8 (LV-MFG-E8), compared with the control cells (LV-NTC) ( Figure 6A and B). However, the enhanced phosphorylation of Akt was decreased by inhibiting MFG-E8 activity using anti-MFG-E8 antibodies in MFG-E8 overexpressing HCC cells ( Figure 6A and B). Interference with MFG-E8 binding to integrins using a synthetic integrin-blocking RGD-based peptide (GRGDSP) also reduced the phosphorylation of Akt ( Figure 6A and B), suggesting that the Akt phosphorylation was mediated by the interaction between the RGD modif of MFG-E8 and integrins. In addition, the expression of cyclin D1 (CCND1) was upregulated in HCC cells overexpressing MFG-E8, compared with the control cells (NTC) ( Figure 6C). However, the upregulation of CCND1 was inhibited by the treatment of HCC cells with MFG-E8 antibodies or RGD-peptides. These data suggest that the action of MFG-E8 on HCC cells is mediated through the integrin-Akt signaling pathway.

MFG-E8 Activities Are Involved in the Integrin-Akt Signaling Pathway in HCC Cells
We next investigated the potential signaling pathways mediated by MFG-E8 in HCC cells. Western blot analyses showed that phosphorylation of Akt was markedly increased in the HCC cells overexpressing MFG-E8 (LV-MFG-E8), compared with the control cells (LV-NTC) ( Figure 6A,B). However, the enhanced phosphorylation of Akt was decreased by inhibiting MFG-E8 activity using anti-MFG-E8 antibodies in MFG-E8 overexpressing HCC cells ( Figure 6A,B). Interference with MFG-E8 binding to integrins using a synthetic integrin-blocking RGD-based peptide (GRGDSP) also reduced the phosphorylation of Akt ( Figure 6A,B), suggesting that the Akt phosphorylation was mediated by the interaction between the RGD modif of MFG-E8 and integrins. In addition, the expression of cyclin D1 (CCND1) was upregulated in HCC cells overexpressing MFG-E8, compared with the control cells (NTC) ( Figure 6C). However, the upregulation of CCND1 was inhibited by the treatment of HCC cells with MFG-E8 antibodies or RGD-peptides. These data suggest that the action of MFG-E8 on HCC cells is mediated through the integrin-Akt signaling pathway.

MFG-E8 Promotes HCC Tumor Formation in Vivo
The data obtained from clinical samples and in vitro assays prompted us to investigate whether MFG-E8 promotes tumor formation in a mouse xenograft model. BALB/c nude mice were injected with Huh7 cells transduced with either control lentivirus (LV-NTC) or MFG-E8-overexpressing lentivirus (LV-MFG-E8) into the right flank (n = 3 for each group). Tumor size was measured up to 9 weeks after grafting ( Figure 7A and B). As expected, MFG-E8 overexpression resulted in a marked increase in both tumor volume and weight, compared to the control group ( Figure 7B-D). Histological analyses of HCC xenografts revealed that the number of Ki-67-positive proliferating cells was significantly greater in MFG-E8-overexpressing HCC xenografts compared with control HCC xenografts ( Figure 8A and B). We also observed that cells positive for CD31, a blood endothelial cell marker, were significantly increased in HCC xenografts overexpressing MFG-E8 over control HCC xenografts ( Figure 8C and D). The growth of tumor in xenograft animals was also analyzed after injections of two additional HCC cell lines, Hep3B and HepG2, that overexpressed MFG-E8. A similar pattern of increased tumor growth by the MFG-E8 overexpression was observed in the additional xenograft studies (Data obtained from Hep3B and HepG2 are provided in Figure S5).

MFG-E8 Promotes HCC Tumor Formation In Vivo
The data obtained from clinical samples and in vitro assays prompted us to investigate whether MFG-E8 promotes tumor formation in a mouse xenograft model. BALB/c nude mice were injected with Huh7 cells transduced with either control lentivirus (LV-NTC) or MFG-E8-overexpressing lentivirus (LV-MFG-E8) into the right flank (n = 3 for each group). Tumor size was measured up to 9 weeks after grafting ( Figure 7A,B). As expected, MFG-E8 overexpression resulted in a marked increase in both tumor volume and weight, compared to the control group ( Figure 7B-D). Histological analyses of HCC xenografts revealed that the number of Ki-67-positive proliferating cells was significantly greater in MFG-E8-overexpressing HCC xenografts compared with control HCC xenografts ( Figure 8A,B). We also observed that cells positive for CD31, a blood endothelial cell marker, were significantly increased in HCC xenografts overexpressing MFG-E8 over control HCC xenografts ( Figure 8C,D). The growth of tumor in xenograft animals was also analyzed after injections of two additional HCC cell lines, Hep3B and HepG2, that overexpressed MFG-E8. A similar pattern of increased tumor growth by the MFG-E8 overexpression was observed in the additional xenograft studies (Data obtained from Hep3B and HepG2 are provided in Figure S5).

Discussion
While previous data revealed a significant correlation of MFG-E8 with the progression of multiple tumor types [9], no available data have been published regarding the tumorigenic activity of MFG-E8 in HCC. High levels of MFG-E8 expression have been correlated to tumor progression via

Discussion
While previous data revealed a significant correlation of MFG-E8 with the progression of multiple tumor types [9], no available data have been published regarding the tumorigenic activity of MFG-E8 in HCC. High levels of MFG-E8 expression have been correlated to tumor progression via

Discussion
While previous data revealed a significant correlation of MFG-E8 with the progression of multiple tumor types [9], no available data have been published regarding the tumorigenic activity of MFG-E8 in HCC. High levels of MFG-E8 expression have been correlated to tumor progression via multiple pathways in various cancer types in different tissues, including the salivary gland, thyroid, pancreas, ovary, brain, skin, bladder, and blood [17]. A previous study demonstrated that gene expression of MFG-E8 was significantly increased in tumor tissues from patients with cholangiocarcinoma and suggested that MFG-E8 is a promising biomarker for the management of cholangiocarcinoma [18]. A recent study showed that serum MFG-E8 levels were significantly lower in patients with HCC than in healthy controls, suggesting that serum MFG-E8 could be a feasible biomarker for HCC [19]. In the present study, we compared HCC tissues and adjacent normal tissues from the same patients and showed that MFG-E8 expression was significantly increased in primary and metastatic HCC tissues compared to normal liver tissues. We also observed that the expression of MFG-E8 was upregulated in all three HCC cell lines tested compared to primary hepatocytes. These results are in good agreement with the above-mentioned studies demonstrating a positive correlation of MFG-E8 expression in tissues with tumor progression in other organs [9]. Patients with melanoma who exhibited a high level of MFG-E8 expression had significantly shorter survival periods than those without MFG-E8 expression [17]. Therefore, these findings altogether strongly suggest that MFG-E8 expression may serve as a promising tissue biomarker for both the diagnosis and prognosis of HCC. As described above, a previous study demonstrated that serum levels of MFG-E8 were lower in HCC patients compared with healthy control [19]. However, this study obtained contradictory results from liver tissues, showing that the expression levels of MFG-E8 were higher in HCC tissues than those in normal tissues [19], as we observed in the present study. Our previous study also demonstrated similar conflicting results showing that serum MFG-E8 levels were comparable in normal and cirrhotic patients, while tissue expression of MFG-E8 was reduced profoundly in the liver tissues of patients with cirrhosis [8]. The discrepancy in MFG-E8 levels between liver tissues and serum was possibly due to the production of MFG-E8 from various tissues other than the liver [3]. Currently, there is limited information regarding the expression kinetics of MFG-E8 during tumor progression, not only for HCC but also for other cancer types. Therefore, it will be interesting to investigate the kinetics of MFG-E8 expression in tissues and its levels in serum during the different stages of acute and chronic liver diseases and tumorigenesis.
Our in vitro loss-and gain-of-function analyses clearly showed that MFG-E8 is actively involved in HCC tumor progression. We also showed that MFG-E8 has tumor-promoting effects in HCC using a mouse xenograft model in vivo. In particular, the in vitro proliferation and migration of HCC cells were more profoundly reduced by the direct targeting of MFG-E8 activity using antibodies, compared to knockdown of MFG-E8 expression using siRNA. Previous studies showed that the administration of MFG-E8-specific monoclonal antibodies effectively blocked the tumor-promoting effects of MFG-E8 in ovarian and breast carcinomas [20]. In addition, the systemic blocking of MFG-E8 activities using MFG-E8 antibodies enhanced the therapeutic effects of anticancer regimens in colon carcinoma, melanoma, thymoma, and fibrosarcoma [16]. Thus, our data suggest that anti-MFG-E8 antibodies could serve as a potential therapeutic regimen against HCC progression and metastasis.
The exact mechanism by which MFG-E8 affects tumor progression is not fully understood and remains to be investigated. MFG-E8 plays diverse cellular roles by binding to integrins via its RGD motif. Our previous study also showed that the binding of MFG-E8 to integrin αvβ3 suppressed TGFβ-induced activation of hepatic stellate cells [8]. Integrin αvβ3 is highly expressed in primary liver cancer and is involved in HCC progression [21,22]. In the present study, our data suggest that MFG-E8 induced the integrin-mediated phosphorylation of Akt and upregulated the expression of cyclin D1 in HCC cells. Therefore, MFG-E8 may bind to integrins and activate Akt/Twist signaling pathways, promoting HCC cell growth and migration, as previously suggested in patients with melanoma [13]. Previous studies demonstrated that MFG-E8 promotes vascular endothelial growth factor-dependent neovascularization and angiogenesis, possibly through αvβ3/5 integrin-mediated AKT phosphorylation [23,24]. In the present study, we found that the number of CD31-postive endothelial cells significantly increased in HCC tumors overexpressing MFG-E8 compared with control HCC tumor tissues in mouse xenografts. Thus, it is possible that MFG-E8 contributes to blood vessel formation in HCC tissues, which is a key process in tumor progression. Other potential mechanisms may be involved in the exosome-mediated tumor progression. Exosomes play important roles in the exchange and delivery of mRNA, miRNA, and proteins between cells, thus modulating the microenvironment [25,26]. Tumor-derived exosomes promote the epithelial-mesenchymal transition, tumor angiogenesis, and metastasis [20,27,28]. MFG-E8 is secreted from cells in association with exosomes and may play an important role in exosome secretion and delivery [25,29,30]. Thus, it is possible that MFG-E8 may regulate the HCC microenvironment by mediating exosome trafficking. Further studies regarding cellular and molecular events are needed to elucidate the exact role of MFG-E8 in the progression of HCC.
Our in vivo study showed that MFG-E8 overexpression significantly increased the tumor volume/weight and the number of proliferating cells as well as endothelial cells in the tumor grafts. Although the size and histochemical differences of tumors were significantly different between control and MFG-E8-overexpressing groups, the small number of animals is a major limitation in the present study. Further studies involving large numbers of animals in different stages of HCC are needed to elucidate the expression kinetics and the exact mechanism of MFG-E8 in HCC progression.

Tissue Preparation and Cell Culture
Thirty-three pairs of HCC and adjacent non-tumor liver tissues were obtained from HCC patients (n = 33) who underwent surgical resection without any radiotherapy and chemotherapy before the surgery at Kyung Hee University Hospital at Gangdong (Seoul, Korea; IRB No., KHNMC IRB 2012-065). The total of 33 HCC samples included 20 primary and 13 metastatic specimens. The diagnosis of HCC and pathological grades were confirmed by certified pathologists. Pathologic differentiation of primary HCC was classified as grade I-IV according to the Edmondson-Steiner's histopathological classification [31] and the grades were divided into two groups, the low (grade I and II, n = 10) and high grade (grade III and IV, n = 10) groups. Based on the WHO classification, 1, 17, and 2 of the 20 primary tumor samples were graded as well-, moderately-, and poorly-differentiated tumors, respectively. Human HCC cell lines (Huh7, HepG2, and Hep3B) were obtained from the American Type Culture Collection (USA). All cell lines were cultured in Dulbecco's modified Eagle's medium (Hyclone, USA) containing 10% fetal bovine serum (Gibco-BRL, USA) and 1% penicillin/streptomycin (Gibco-BRL) at 37 • C in a humidified atmosphere with 5% CO 2 . Human primary hepatocytes were purchased from Lonza (USA) and cultured in Lonza hepatocyte culture media according to the manufacturer's instructions.

Lentiviral Transduction
Lentiviral pcLV-CMV-MFG-E8-EF1a-Neo-eGFP vector (Sirion Biotech GmbH, Germany) was used to express GFP-tagged MFG-E8 (MFG-E8-GFP) in HCC cell lines. Cells were cultured at 5 × 10 4 cells/well into 6-well tissue culture plates overnight. The viral supernatant was then added to cells at a multiplicity of infection of 10 with 5 µg/mL polybrene. The expression of GFP was evaluated by fluorescence microscopy to estimate infection efficiency. HCC cells infected with an MFG-E8-expressing lentiviral vector and non-targeting control lentivirus vector were named 'LV-MFG-E8' and 'LV-NTC,' respectively.

Immunostaining
For colorimetric staining, tissue samples were fixed in formalin, embedded in paraffin, and cut into 4-µm-thick sections. Tissue sections were deparaffinized twice in xylene and microwaved for antigen retrieval in low-pH buffer (pH 6.0) for 20 min. After treatment with 3% hydrogen peroxide for 15 min to block endogenous peroxidases, the sections were incubated with the corresponding primary antibodies at room temperature for 15 min, followed by incubation with HRP-labeled polymer-conjugated goat anti-mouse IgG (Dako Cytomation Envision + System, DAKO, Carpinteria, CA, USA) for 30 min. Subsequently, the sections were developed using 3,3'-diaminobenzidine chromogen and substrate buffer (DAKO). All sections were counterstained with hematoxylin. Primary antibodies were purchased from Santa Cruz Biotechnology ( Santa Cruz, CA, USA) and used at the following dilutions: anti-MFG-E8 (sc8029, 1:100), Ki67 (sc23900, 1:100), and AFP (sc8108, 1:50).
For immunofluorescence staining, cells and tissues were fixed in cold 4% paraformaldehyde (Sigma-Aldrich, St Louis, MO, USA) in phosphate-buffered saline (PBS, Gibco-BRL, Gaithersburg, MD, USA). Tissues were embedded in O.C.T. compound (Tissue Tek, Sakura Finetek USA, Torrance, CA, USA) after fixation and then sectioned into 4-µm slices. After blocking and permeabilization in 0.3% Triton X-100 (Sigma-Aldrich) and 10% donkey serum (Sigma-Aldrich) in 0.1% bovine serum albumin (Sigma-Aldrich)/PBS, the sections were incubated at 4 • C overnight with primary antibodies. For MFG-E8, AFP, and CD31 staining, Alexa Fluor 488-conjugated donkey antibodies against rabbit or goat IgG, or Alexa Fluor 594-conjugated donkey antibody against mouse IgG (Invitrogen, Carlsbad, CA, USA) were used as secondary antibodies. An Apotome-Axiovert 200M fluorescence microscope (Carl Zeiss, Oberkochen, Germany) was used to visualize cells after counterstaining with 4 ,6-diamidino-2-phenylindoledihydrochlorid (DAPI, 1:1,000, Sigma-Aldrich). To quantify the immunofluorescence intensity of MFG-E8 and AFP expression, ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used to assess the mean fluorescence intensity of greyscale images in five random fields of tumor and normal regions from each tissue section obtained from the 33 patients. The value was expressed as relative values to those in normal tissues, which were arbitrarily set as 1.

ELISA
Culture media were collected from primary hepatocytes and from Huh7, HepG2, and Hep3B cells after 24 h of culture, centrifuged to obtain the supernatants, and assessed for MFG-E8 levels using ELISA kits (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. Absorbance was immediately read at 450 nm using a microplate reader (Bio-Tek Instruments, Bad Friedrichshall, Germany).

RT-qPCR
For gene expression analyses, HCC and adjacent non-tumor liver tissues from patients (n = 10) were provided by the Biobank of Ajou University Hospital and Korea University Guro Hospital. Total RNA was extracted from two groups of tissues (HCC and non-tumor tissues) frozen in liquid nitrogen or cultured cells using Trizol reagent (Invitrogen, USA) and then reverse-transcribed into cDNA using the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA). For RT-qPCR, cDNA was mixed with the appropriate primers and the SYBR-Green Super mix (Bio-Rad, Hercules, CA, USA) and then run on the CFX96 Real-Time System (Bio-Rad). The reaction conditions were 96 • C for 1 min, and 40 cycles of 96 • C for 10 s, 57 • C for 10 s, and 72 • C for 30 s. The 2 -∆∆Ct method was used to calculate the relative expression of MGF-E8, and the data obtained were normalized to GAPDH levels . The primer  pairs used were: MFG-E8 forward primer, 5 -CCGTAACTTTGGCTCTGTCC-3 and reverse primer,  5 -TCTTGTGGGAGTGGTTGTCC-3 ; Cyclin D1 forward primer, 5 -GCTGCGAAGTGGAAACCATC-3  and reverse primer, 5 -CCTCCTTCTGCACACATTTGAA-3 ; GAPDH forward primer, 5 -AGGGCTGCTTTTAACTCTGGT-3 and reverse primer, 5 -CCCCACTTGATTTTGGAGGGA-3 .

Cell Proliferation Assay
Cells were seeded in each well of a 96-well plate at a density of 1 × 10 4 cells/well. Following incubation at 37 • C for different periods of time (0, 24, 48, and 72 h), the culture media were removed, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich, 5 mg/mL) was added to each well. After incubation at 37 • C for another 4 h, the MTT solution was removed and replaced with dimethyl sulfoxide (Sigma-Aldrich). Absorbance was measured at 570 nm using a microplate reader (Bio-Tek Instruments).

Wound Healing Assay
To determine the role of MFG-E8 in the regulation of HCC cell migration, a wound scrape assay was performed. Briefly, artificial vertical scratches were made on confluent monolayers in 60-mm culture dishes with a 200-µL yellow pipette tip. The process by which the scratches were filled by migrating cells was monitored under a phase-contrast microscope equipped with a digital camera at 0, 24, 48, 72, and 96 h after plating. The cell migration rate was calculated using the following formula: Cell migration distance/original wound width × 100 for each time point.

Mouse Xenograft Model
The animal protocol was approved by the Animal Care and Use Committee of Korea University (IRB No., KUIACUC-2018-0019). Male BALB/c nude mice (6 weeks of age) were purchased from Orient Bio (Gyonggi-Do, Korea). Huh7 cells transfected with LV-NTC (n = 3) or LV-MFG-E8 (n = 3) (both groups: 5 × 10 6 cells per mouse) were subcutaneously injected into the flanks of BALB/c nude mice. An additional 10 animals were injected with Hep3B (LV-NTC, n = 3; LV-MFG-E8, n = 3) and HepG2 cells (LV-NTC, n = 2; LV-MFG-E8, n = 2). The lengths and widths of tumors were measured using a caliper every week. All mice were euthanized at 9 weeks post-injection, and the tumor nodules were removed and weighed. The tumor volume was calculated according to the following formula: tumor volume (mm 3 ) = length (mm) × width (mm) 2 /2.

Statistical Analysis
All the results except in vivo data were obtained from at least three independent experiments unless otherwise indicated and expressed as the mean ± standard deviation (S.D.). Statistical analyses were performed using SPSS software, version 25.0 (IBM SPSS, Armonk, NY, USA). Comparisons between different groups were made using a two-tailed Student's t-test. P values less than 0.05 were considered statistically significant.

Conclusions
HCC is a malignant tumor with high mortality, poor prognosis, and frequent relapse. The results of the present study, to the best of our knowledge, provide the first direct evidence that MFG-E8 substantially contributes to the progression of HCC. All data obtained from human HCC tissues, HCC cell lines, and a HCC xenograft model support this finding. Although the exact mechanism of action remains to be elucidated, our study suggests that targeting MFG-E8 may serve as a promising therapeutic approach against HCC.