Human amniotic mesenchymal stem cells inhibit hepatocellular carcinoma in tumour‐bearing mice

Abstract Hepatocellular carcinoma (HCC) is the third leading cause of the cancer‐related death in the world. Human amniotic mesenchymal stem cells (hAMSCs) have been characterized with a pluripotency, low immunogenicity and no tumorigenicity. Especially, the immunosuppressive and anti‐inflammatory effects of hAMSCs make them suitable for treating HCC. Here, we reported that hAMSCs administrated by intravenous injection significantly inhibited HCC through suppressing cell proliferation and inducing cell apoptosis in tumour‐bearing mice with Hepg2 cells. Cell tracking experiments with GFP‐labelled hAMSCs showed that the stem cells possessed the ability of migrating to the tumorigenic sites for suppressing tumour growth. Importantly, both hAMSCs and the conditional media (hAMSC‐CM) have the similar antitumour effects in vitro, suggesting that hAMSCs‐derived cytokines might be involved in their antitumour effects. Antibody array assay showed that hAMSCs highly expressed dickkopf‐3 (DKK‐3), dickkopf‐1 (DKK‐1) and insulin‐like growth factor‐binding protein 3 (IGFBP‐3). Furthermore, the antitumour effects of hAMSCs were further confirmed by applications of the antibodies or the specific siRNAs of DKK‐3, DKK‐1 and IGFBP‐3 in vitro. Mechanically, hAMSCs‐derived DKK‐3, DKK‐1 and IGFBP‐3 markedly inhibited cell proliferation and promoted apoptosis of Hepg2 cells through suppressing the Wnt/β‐catenin signalling pathway and IGF‐1R‐mediated PI3K/AKT signalling pathway, respectively. Taken together, our study demonstrated that hAMSCs possess significant antitumour effects in vivo and in vitro and might provide a novel strategy for HCC treatment clinically.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC), which accounts for 80%-90% of primary liver cancer, is the third leading cause of cancer-related death worldwide. 1 In recent years, although the level of screening, diagnosis and treatment of liver cancer has increased, the incidence and mortality of liver cancer have been increasing. 2 Furthermore, intrahepatic metastasis of primary liver cancer can lead to tumour recurrence, making the treatment of liver cancer very difficult.
Therefore, it is urgent to develop new methods to control the growth and metastasis of liver cancer clinically.
Mesenchymal stem cells (MSCs) are a multi-potential cell that can differentiate into osteocytes, chondrocytes and adipocytes.
MSCs also have the abilities of self-renew and have low tumorigenicity. 3 MSCs exhibit relatively low immunogenicity and do not stimulate lymphocyte proliferation, thus avoiding immune rejection. 4 These characteristics of MSCs make them very suitable for cell therapy. The most attractive feature of MSCs is that it can specifically migrate to the tumour site, which indicates its application in tumour therapy. To control tumour growth, MSCs can be genetically engineered to express therapeutic genes, 5-7 loaded with drugs 8 or be infected with oncolytic viruses. 9,10 Numerous studies shown that MSCs can directly and/or indirectly affect the development of multiple types of cancer. 11,12 They are recruited to tumorigenic sites and secrete factors, which have been proved to have antitumour and/or protumour effects. 13 14,15 In recent years, human amniotic mesenchymal stem cells (hAMSCs) have been considered to be the stem cell with the most application prospect clinically. However, it has not been fully understood how hAMSCs affect HCC cell properties in vivo and in vitro, for example proliferation, apoptosis, migration and invasion.
Recently, we have systemically investigated the morphology, phenotype, pluripotency, tumorigenicity and growth potency of hAMSCs. 4 Here, we established Hepg2 or Hepg2/hAMSCs co-injected xenografted BALB/c nude mouse models, and therapeutic effects of hAMSCs in Hepg2 xenografted model mouse were evaluated. We observed that hAMSCs or hAMSCs-CM had the similar antitumour effects in vivo and in vitro by inducing cell cycle arrest and promoting cell apoptosis, suggesting that the secreted cytokines from hAMSCs might be involved in their antitumour effects.
Antibody array assay showed that hAMSCs highly expressed DKK-3, DKK-1 and IGFBP-3, which inhibited liver cancer growth through suppressing Wnt/β-catenin and IGF-1R signalling pathways. Taken together, we demonstrated that hAMSCs-derived trophic factors possess anti-HCC effect in vitro and in vivo, suggesting that hAMSCs may provide a novel therapeutic strategy for the treatment of liver cancer.

| Isolation, culture and expansion of hAMSC S
After obtaining the oral consent of the donor, fresh amniotic membrane tissue was collected from the department of the obstetrics and gynaecology, the first affiliated hospital, Nanchang University. hAMSCs were prepared as previously described. 4,16
Subsequently, puromycin (3 mg/mL; Sigma-Aldrich) was used to select GFP-positive cells. The percentage of GFP-labelled hAMSCs was determined by immunofluorescence every day. Once the percentage of GFP-labelled hAMSCs was higher than 95%, puromycin was removed from the culture medium.

| Flow cytometry analysis, adipogenic and osteogenic differentiation
The surface molecular markers and differentiation potential of cultured hAMSCs and GFP-labelled hAMSCs were determined by flow cytometry analysis and adipogenic and osteogenic differentiation as previously described. 4 The antibodies used for flow cytometry analysis are listed in Table S1.

| Animal models
8-week-old NOD-SCID male mice and BALB/c nude mice were obtained from Changsha SLAC Laboratory Animal Company (Changsha, China, http://www.hnsja.com/). All animal experiments were performed according to institutional guidelines and approved by the Animal Care and Use Committee of Nanchang University.

| Hepg2 cell xenograft model and fluorescent imaging in vivo
The human hepatocarcinoma cell line Hepg2 was obtained from ATCC (Manassas, VA) and maintained in H-DMEM (Thermo Fisher) containing 100 U/mL penicillin, 100 μg/mL streptomycin and 10% FBS. Cells were incubated in a 5% CO 2 -humidified incubator at 37°C.
The longest size (a) and the shortest size of (b) tumour were measured with vernier calliper (Mitutoyo Co., Tokyo, Japan) every day for 24 days. The tumour volume calculation formula is V= (1/2)ab 2 .
Eighteen days after cell injection, the mice were anaesthetized first, and the mice were visualized with whole-body fluorescent imaging system (LB983; Berthold, Germany). Then, the mice were killed, and the tissues of tumour, heart, liver, brain, kidney, spleen, lung and pancreas were isolated and visualized with whole-body fluorescent imaging system.

| Histopathology and TUNEL assay
Tumour tissues were processed for paraffin by slicing into 5-μmthick section. Immunohistochemistry was performed on tumour sections to detect PCNA (1:1000, mouse monoclonal antibody, Abcam, Nanchang, China). Apoptosis assay was performed on tumour tissue sections by TUNEL assay kit (Millipore, USA) according to the manufacturer's instruction.

| Production of CM and experiment of coculture in vitro
hAMSC-CM was prepared as previously described. 4 Briefly, hAM-SCs were grown in normal culture medium, once the cells reached to 80% confluency, the medium was changed with H-DMEM containing 100 U/mL penicillin and 100 μg/mL streptomycin. CM was collected after 48 hours and centrifuged at 1500 rpm for 5 minutes to ensure complete removal of cellular debris. CM was then con-

| Tumour cell proliferation analysis
Cell proliferation of different groups was determined using a CCK-8 kit (Dojindo Laboratories, Kumamoto, Japan), according to the manufacturer's instructions from 24 to 48 hours after cells were plated.
Absorbance values were measured at a wavelength of 450 nm using a microplate spectrophotometer (Bio-Rad).

| Western blot analysis and immunofluorescence
Western blot analysis and immunofluorescence were preformed as previously described 4 ; the primary antibodies are listed in Table S1, in-

| Antibody protein array
Human cytokine arrays (QAH-CAA-440-1, RayBiotech, Norcross, USA) were used to measure the expression of 440 cytokines in serum-free culture supernatants at 48 hours from hAMSCs, according to the manufacturer's instructions.

| Statistical analysis
All data are presented as mean ± standard deviation (SD). Data were analysed by Student's t test or one-way analysis of variance (ANOVA).
Differences between values were considered significant at P < .05.

| Identification and characterization of hAMSCs and GFP-labelled hAMSCs
The GFP-labelled hAMSCs (GFP-hAMSCs) were prepared by lentiviral infection for cell tracking. As shown in Figure 1A, more than 95% of infected hAMSCs were GFP-positive after puromycin selection. Compared with hAMSCs, the morphology of GFP-hAMSCs did not change significantly; it was spindle-shaped and fibroblast-like and grew in adherent monolayer. In the medium containing bFGF, hAMSCs and GFP-hAMSCs proliferated rapidly with an average doubling time of two days ( Figure 1A). Flow cytometry showed that both hAMSCs and GFP-hAMSCs expressed MSCs marker proteins CD105, CD73, CD90, CD29 and HLA-ABC, a major histocompatibility protein, but did not express CD34 and CD45, the hematopoietic stem cell marker proteins. hAMSCs and GFP-hAMSCs also negative for major histocompatibility proteins HLA-DR and HLA-ABC co-stimulate molecules CD80, CD86 and CD40 ( Figure 1B). In vitro, both hAMSCs and GFP-hAMSCs can be induced to differentiate into osteoblasts and adipocytes under osteogenic and adipogenic differentiation conditions ( Figure 1C). The above results show that hAMSCs and GFP-hAMSCs both express specific molecular markers of MSCs and have low immunogenicity and multi-differentiation potential, the transfection of GFP does not affect the characteristics and proliferation ability of hAMSCs. In addition, our previous research results show that hAMSCs had no tumorigenicity in vitro and in vivo. All these advantages make hAMSCs and GFP-hAMSCs have great clinical application potential.   Figure 2I). These results indicated that the hAMSCs injected intravenously were able to efficiently home to the tumorigenic sites and in turn markedly inhibited the tumours growth through suppressing proliferation and promoting apoptosis of the tumour cells.

| hAMSCs or hAMSC-CM inhibited proliferation and promoted apoptosis of Hepg2 cells via a paracrine manner in vitro
Next, to determine whether the antitumour effects were directly  Figure 3D). Compared to control group, hAMSCs co-culture and hAMSC-CM treatment decreased the expression of PCNA by 55.5 ± 8.5% and 59.5 ± 6.3%, respectively, as well as 57 ± 3% and 67 ± 2% for Ki67 ( Figure 3E). Western blot analysis also showed that the expressions of cyclin B1 and cyclin A2 were significantly Taken together, these results demonstrated that the antitumour effects of hAMSCs were dependent on inhibiting cell proliferation and promoting cellular apoptosis in a paracrine manner.

| DKK-3 and DKK-1 derived from hAMSCs inhibited Hepg2 cell proliferation through blocking canonical Wnt/β-catenin signalling pathway
The hAMSCs and hAMSC-CM have a similar antitumour effect, suggesting that the secreted cytokines from hAMSCs were responsible for the antitumour effect. To elucidate the possible mechanism of the antitumour effects of hAMSCs, the antibody array assay was performed to examine the cytokine levels in the supernatants of hAMSCs. Among 440 cytokines evaluated, the 20 factors with the highest expression amount are shown in Table 1. Interestingly, we found that IGFBP-3, DKK-3 and DKK-1 were highly secreted by hAMSCs; the concentrations were 45 401.20 pg/mL, 37 200.47 pg/ mL and 11 930.42 pg/mL, respectively ( Figure 5A and Table 1). The concentrations of around 200 cytokines were more than 20 pg/mL and the detail information is shown in Table S2 F I G U R E 2 hAMSCs exhibit tumour-homing and tumour-suppressive properties in HCC mouse model. A, Experimental schematic of Hepg2-induced hepatocellular carcinoma in the BALB/c nude mice model. The experiments were conducted in two groups including the PBS group and GFP-labelled hAMSC group. B, Monitoring of hAMSCs tracking to tumour lesions in HCC mouse model on day 24 after Hepg2 transplantation by whole-body fluorescent imaging assay. The results showed that GFP-positive cells were detected around tumour site and in the tumour tissues. C and D, Gross observation of subcutaneous xenografts of Hepg2/PBS and Hepg2/hAMSCs intravenous-injected nude mice. The red circles represent the tumour location. E, The average volume of hAMSC-injected tumours was significantly smaller than that of PBS-injected tumours at day 18 and day 24 (n = 5). F, Gross observation of subcutaneous xenografts of Hepg2 alone and Hepg2/ hAMSCs co-injected nude mice. The red circles represent the tumour location. G, The average volume of Hepg2/hAMSCs subcutaneous coinjected tumours was significantly smaller than that of Hepg2 alone injected tumours at day 6, 12, 18 and 24 (n = 5). H, The average volume of Hepg2/hAMSCs subcutaneous co-injected tumours was significantly smaller than that of hAMSCs intravenous-injected tumours at days 6, 12, 18 and 24. I, Proliferation and apopsis of Hepg2 cells were tested by immunohistochemistry using antibodies against PCNA and TUNEL staining in PBS and hAMSCs intravenous-injected tumour tissues. Results are shown as mean ± SD in hAMSCs. As shown in Figure 6E 26,30 In this study, we reported that hAMSCs were multi-potent MSCs and showed a long fusiform morphology. hAMSCs met the criteria for MSCs identification by The Association of International Cell Therapy, such as more than 95% of hAMSCs were positive for MSC markers and negative for hematopoietic stem cell markers.

| D ISCUSS I ON
In addition, we also observed that there were no expressions of HLA-DR and the co-stimulatory molecules of HLA-AB although there was a low expression of HLA-ABC in hAMSCs, suggesting that the cells have a weak immunogenicity and potential immune tolerance for transplantation.
It has been reported that hAMSCs possess significant therapeutic effects in multiple disease models, including skin injury, 31 premature ovarian insufficiency, 32 acute liver injury, 33 lung injury, 34 T1 diabetes, 35 chronic kidney disease 36 and so on. MSCs also have shown tremendous potential in anticancer applications because of their innate ability to migrate the tumorigenic sites. 37 However, there was no study yet to reveal the function of hAMSCs on HCC.
In addition, hAMSCs have strong proliferation ability and can be expanded to 250 generation; it is easy to obtain enough number of stem cells for clinical applications. We also observed that hAMSCs have no tumorigenicity both in vivo and in vitro, suggesting that the cells might be safe for cancer therapy. Therefore, we choose hAM-SCs as the therapeutic cells for cancer therapy especially for HCC therapy. To elucidate the homing and therapeutic potential of hAM-SCs on HCC in mice, hAMSCs labelled with GFP were injected into the mice with HCC model via the tail vein. Whole-body fluorescent imaging analysis showed that injected hAMSCs were homed to the sites of tumorigenesis, and some of the hAMSCs were migrated into tumour tissues, indicating that hAMSCs have the ability to home to the tumour sites in vivo. The tumour-homing ability of hAMSCs provides the promising potential to use them as vehicles to transport  The IGF signalling pathway is activated by binding of IGF-1/IGF-2 to IGF-1R in the plasma membrane, 48 which triggers a rapid autophosphorylation of the receptor, followed by phosphorylation of intracellular targets, ultimately leading to activation of PI3K/AKT pathway. The activation of IGF-1R regulates several cellular processes including motility, proliferation and inhibition of apoptosis. 17 In the serum and the extracellular fluids, IGF-1 and IGF-2 are usually bound to IGFBPs which regulate the IGF-1/IGF-1R interaction through controlling the bioavailability of IGFs. 48 The IGFBP3 levels were very low or undetectable in human HCC samples compared with non-neoplastic liver tissue which are 4-to 100-fold higher than in HCC 49 and IGFBP-1 and IGFBP-2 were also decreased in hepatoblastoma. 50 It has been reported that the alterations in the expression of the molecules in the Wnt/β-catenin and IGF pathways have been observed during hepatocarcinogenesis. 17,40 In the present study, we shown that hAMSCs and hAMSC-CM markedly reduced F I G U R E 6 hAMSC-derived DKK-3 and DKK-1 inhibit the proliferation of Hepg2 cells through blocking Wnt/β-catenin signalling pathway. hAMSC-CM was pre-treated with monoclonal antibody against DKK-1, DKK-3 and IGFBP-3 for 24 hours. After which, the Hepg2 cells were treated with normal culture medium, ICG001 (20 μM), CM, CM + DKK-3 antibody, CM + DKK-1 antibody and CM + IGFBP-3 antibody for 48 h. A, The expression of PCNA, Ki67, β-catenin, N-cadherin, cyclin A2, cyclin B1, cyclin D1 and cyclin E1 in Hepg2 cells in each group was analysed by Western blot. B, Quantitative analysis of the expression of PCNA, Ki67, β-catenin, N-cadherin, cyclin A2, cyclin B1, cyclin D1 and cyclin E1 in Hepg2 cells of different groups as in (A). C, The expression of PCNA, KI67, cyclin A2, cyclin B1, cyclin D1 and cyclin E1 in Hepg2 cells in control (normal medium), CM, CM + IGFBP-3 antibody group were analysed by Western blot. D, Quantitative analysis of the expression of PCNA, KI67, cyclin A2, cyclin B1, cyclin D1 and cyclin E1 in Hepg2 cells of different groups as in C. The RNAi efficiency of DKK-3 (E), DKK-1 (F) and IGFBP-3 (G) in hAMSCs was assayed by Western blot before co-culture with Hepg2 cells. H, Western blot analysis showed that the accumulation of β-catenin and the expression of PCNA were increased in hAMSC-siDKK3 group and hAMSC-siDKK1 group when compared with hAMSC-siMOCK group. I, Quantitative analysis of the expression of PCNA and β-catenin in Hepg2 cells of different groups as in (H). J, Western blot analysis showed that the expression of PCNA was no significant difference observed in Hepg2 cells cocultured with hAMSC-siMOCK and hAMSC-siIGFBP-3. K, Quantitative analysis of the expression of PCNA in Hepg2 cells of different groups as in J. Results are shown as mean ± SD | 10539 LIU et aL.

| CON CLUS IONS
In our study, we demonstrated that hAMSCs inhibit proliferation and promote apoptosis of Hepg2 cells through blocking both Wnt/βcatenin and IGF-1R/PI3K/AKT signalling pathways, suggesting that administration of hAMSCs and hAMSC-CM may be a novel strategy for treatment of HCC clinically.

ACK N OWLED G EM ENTS
The authors are grateful to the donors for kindly providing pla-

CO N FLI C T O F I NTE R E S T
The authors have declared that no competing interests exist.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the finding of this study are available from the corresponding author upon reasonable request.