LOXL2 promotes vasculogenic mimicry and tumour aggressiveness in hepatocellular carcinoma

Abstract Lysyl oxidase‐like 2 (LOXL2) has shown to promote metastasis and poor prognosis in hepatocellular carcinoma (HCC). Also, we have previously reported that vasculogenic mimicry (VM) is associated with invasion, metastasis and poor survival in HCC patients. In the present study, we investigated molecular function of LOXL2 in HCC and VM. We used the immunohistochemical and CD31/periodic acid‐Schiff double staining to detect the relationship between LOXL2 and VM formation. We performed the gain and loss of function studies and analysed the migratory, invasion and tube formation in HCC cell lines. We analysed the function of LOXL2 in VM formation and HCC metastasis both in vitro and in vivo. We have showed that LOXL2 was overexpression in HCC and was positively correlated with tumour grade, metastasis, VM formation and poor survival in 201 HCC patients. Secondly, our studies have showed that LOXL2 overexpression in HCC cells significantly promoted migration, invasion and tube formation. Finally, we found that LOXL2 may increase SNAIL expression, thereby enabling VM. Our study indicated that LOXL2 may promote VM formation and tumour metastasis by collaborating with SNAIL in HCC. What's more, the overexpression of LOXL2 indicated a poor prognosis in HCC patients.

metastasize. 6,10,13,14 Current studies of the mechanism of VM mainly focus on the remodelling of the ECM and tumour microenvironments. 11,14 Our previous work has shown that VM is associated with the invasive and metastatic potential of tumour cells and with poor clinical outcomes. 15 To date, three microcirculation patterns have been identified in HCC: endothelium-dependent vessels, VM and mosaic vessels. 10,12 However, because the tumour blood supply patterns are so complex, the effects of anti-angiogenic therapy on HCC patients are unsatisfactory. Therefore, more research on the tumour blood supply is needed.
Lysyl oxidases constitute a family of secreted copper-dependent amine oxidases that catalyse the oxidative deamination of the ε-amino group in peptidyl lysine residues, thus promoting covalent protein cross-linkages. [16][17][18] Lysyl oxidase-like 2 (LOXL2) is a member of the lysyl oxidase (LOX) family, of which five members have far been identified to date, including one LOX, and four LOX-like proteins, LOXL1-4. LOXL2 play an important role in several processes, such as cell adhesion, cell migration and invasion, epithelial cell plasticity and the epithelial-mesenchymal transition (EMT). [16][17][18][19][20][21] Moreover, LOXL2 plays a key role in the stabilization of collagen and elastin fibres in ECM remodelling, which plays a major role in the development of a functional vascular system. 20,[22][23][24] Bignon et al recently showed that LOXL2 is involved in angiogenesis, which is driven by hypoxic tumour microenvironments. 25,26 LOXL2 can affect endothelial cell proliferation and migration, which are necessary for capillary formation. 23,[25][26][27] More interestingly, accumulating evidence indicates that LOXL2 can promote invasion and metastasis in basallike breast cancer cells by collaborating with SNAIL and lethal giant larvae (LLGL2) 28,29 ; LOXL2 was also described as a prognostic marker in larynx squamous cell carcinomas. 30 SNAIL is a transcription factor that is involved in the EMT 31,32 ; Our previous research data revealed that the EMT may be involved in VM in HCC. 14,33 Moreover, LLGL2, which is a component of the cell polarity complex (Scribble complex), has previously been shown to be downregulated by EMT factors such as Snail1/Snail2 and/or ZEB1. 14,33 Although the expression of LOXL2 and its implications have been shown in HCC, the correlation between LOXL2 and VM in HCC and their relevance as clinical parameters remain unclear. In this study, we attempted to identify the effects of LOXL2 on tumour VM.

| Patients
Tissue specimens were obtained from the Tumor Tissue Bank of the

| Immunohistochemical and double histochemical staining methods
The sections were pre-treated with microwaves, blocked and incubated with a series of antibodies (Table S2). The staining systems used in this study were PicTure PV6000 (Zhongshan Chemical Co., Beijing, China) and Elivision Plus (Zhongshan Chemical Co.). Finally, the sections were counter-stained with haematoxylin or periodic acid-Schiff (PAS). Phosphate-buffered saline was used in place of the primary antibodies for the negative control.

| Counting methods
The results of immunohistochemical (IHC) staining were assessed through microscope by two pathologists. The assessing method was described in the supplement material. VM was identified by the presence of red blood cells in vessels lined by tumour cells, and not by endothelial cells, and by the absence of necrosis and inflammatory cells infiltrating the area surrounding the channels.

| Cell culture and transfection
The HCC cell lines HepG2, HepG3B and SMMC7221 we used were obtained from the American Type Culture Collection (Rockville, MD, USA) in 2012 and authenticated using short tandem repeat (STR) analysis by Genewiz Inc. in 2014. STR analysis showed that the submitted samples were in good agreement with the reference cell lines.

| Expression plasmids
The full-length LOXL2 and SNAIL complementary DNAs (cDNAs) were generated from normal human embryo total cDNAs, digested with XhoI/EcoRI and subcloned into pcDNA3.1 vectors. The resulting constructs were confirmed by DNA sequencing. SNAIL and LOXL2 Gene Silencing used the small interfering RNA (siRNA) kit (pGP-Twist1-shRNA) purchased from GeneCopoeia (US). Puromycin was used as the stable cell line selection marker.

| RNA extraction and quantitative reverse transcription-PCR
Total RNA was extracted using TRNzol A+ Reagent (TaKaRa Biotechnology Co., Ltd., Japan), according to the manufacturer's instructions; cDNAs were prepared using the Quantscript RT Kit (Tiangen Biotech). Quantitative PCR (qPCR) was performed with a 7500/7500 Fast Real-Time PCR System (Applied Biosystems), Tli RNaseH Plus (RR820A; TaKaRa). Quantitative reverse transcription-PCR (qRT-PCR) was performed as previously described. 12 The primers used for qRT-PCR are listed in Table S8.

| Western blot analysis
The whole cell lysates were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes (Millipore). The blots were blocked and incubated with the appropriate antibody (Table S2)

| Immunofluorescence staining
Cells were plated onto chamber slides and fixed in ice-cold methanol. The primary antibodies against LOXL2 and SNAIL were used at a 1:400 working dilution. Fluorescein isothiocyanate-and tetramethylrhodamine isothiocyanate-conjugated mouse and rabbit immunoglobulin G antibodies (Santa Cruz Biotechnology) were used as labels for the immunofluorescence assay. After immunolabelling, the cells were washed, stained with DAPI (Sigma), mounted and then viewed with a fluorescent microscope (Nikon, Japan).

| 3D cultures
Tumour cells were transfected, incubated 24 hours and mixtureseeded with Matrigel (Collaborative Biomedical), and the matrix was allowed to polymerize. The addition of conditioned media with 10% foetal bovine serum (Hyclone) allowed us to perform pre-treatment and continuous treatment regimens during the 10-day incubation period in 3D cultures. The cells were collected from the Matrigel with trypsin, to which Trizol or RIPA buffer was added to isolate the total RNA or protein from the cells.

| Invasion and wound healing assay
The cell migration assay was performed with Transwell cell culture inserts (Invitrogen). The transfected cells were maintained for 48 hours and allowed to migrate for another 24 hours. The migrated cells were stained with a crystal violet solution and its absorbance was determined at 595 nm. In the wound healing assays, cell motility was assessed by measuring the movement of cells into a scrape. The speed of wound closure was monitored after 12 and 24 hours by measuring the ratio of the distance of the wound at 0 hours. Migration/invasion assays were performed as reported. 34 Each experiment was also performed in triplicate.

| Murine xenograft model
The orthotopic transplantation tumour model and the HCC metastatic model was made by injecting 5 × 10 6 HCC cell into the 4-to 6-week-old nude mice. Then the mice were monitored for 4-5 weeks and tumour sizes were measured daily using a caliper. After the observations were complete, the mice were killed, and then used for the sequent histological examination. The detail method was described in the supporting material.

| Statistical analysis
In this study, we evaluated all data using SPSS 17.0. All P-values were two-sided, and P < 0.05 was considered significant. The significant groups are marked with an asterisk in the figures. Pearson's chisquared test was then used to compare the differences in protein expression between the metastatic and non-metastatic groups.
Moreover, forward selection of the stepwise discriminant analysis was then used to establish a combination formula to predict the metastatic potential of the second set of HCC samples. Pearson's chi-squared test was then used to compare the difference between the predicted and actual results and to validate the predictive value of the differentially expressed proteins. For the survival analysis, survival curves were produced with the Kaplan-Meier method. Differences in the survival curves were assessed by the log rank test.

| LOXL2 cytoplasm expression was significantly correlated with VM in HCC and its upregulation is associated with poor patient prognosis
To assess the relationship between LOXL2 and VM in HCC, we first performed IHC staining on 201 HCC tissue sections. The results showed that LOXL2 is located in both the nuclei and in the cytoplasm of HCC cells. We assessed both the intensity and the percentage of positive cells using previously described criteria (see Section 2 or Supplement Material). Then, using CD31/PAS double staining, vascular-like patterns, which were formed by HCC cells and contained When comparing the VM + /VM − and LOXL2 expression (Table S2), we found that LOXL2-c expression differed significantly between the VM + and VM − groups (P = 0.002) but that LOXL2-n expression did not. Furthermore, we found that LOXL2-c expression was SHAO ET AL.
| 1365 positively correlated with VM in the HCC samples by Spearman analysis; however, the LOXL2-n expression was not.
By co-staining of LOXL2 and VM, we found that the tissue which had VM formation, usually expressed high level of LOXL2-c, while, the tissue without VM formation, had low expression of LOXL2-c ( Figure 1A). The analysis of the relationship between LOXL2 and the clinicopathological characteristics showed that the cytoplasmic LOXL2 overexpression was not correlated with the patients' age or tumour size but was correlated with clinical metastasis (P = 0.002; and tumour grade (P = 0.04) ( Table 1).
To validate the prediction value of LOXL2-c expression for metastasis, discriminant analysis was performed to establish some predictive formulas from 160 HCC samples. By Pearson chi-squared test, we compared differences between the predicted and actual metastasis in a second set of HCC cases, and found that LOXL2-c expression had a good prediction value for HCC metastasis (Table S3).
The results of the Kaplan-Meier survival analysis showed that patients with VM and LOXL2-c expression had a shorter survival period than those without VM or LOXL2-c expression ( Figure 1C), and that patients with LOXL2-n expression did not show a short survival period than those without LOXL2-n expression ( Figure S1). LOXL2-c overexpression and VM were independent predictors of HCC in the Cox analysis (Tables S7 and S8), while LOXL2-n overexpression was not.
Together, these results indicated that LOXL2-c expression is significantly associated with VM, metastasis, tumour grade and a shorter survival of HCC patients.

| LOXL2 overexpression increased HCC cell invasion and migration and promoted the formation of tube-like structures by HCC cells in vitro
In this section, we investigated the influence of LOXL2 on VM and metastasis in the HCC cell lines. First, the expression levels of (C) Patients with VM and LOXL2-c expression had a shorter survival period than those without VM or LOXL2-c expression. The left two panels showed that patients with LOXL2-c expression had a shorter survival period than those without LOXL2-c expression; the right two panels showed that patients with both VM formation and LOXL2-c expression had a shorter survival period than those without VM formation or LOXL2-c expression exhibited lower LOXL2 expression, whereas the HCC cells with a mesenchymal phenotype, such as Bel7402 cells, exhibited higher LOXL2 expression (Figure 2A). Thus, the above HCC cell lines were selected and transfected as recipient cells: HepG2 cells were transfected with the LOXL2 plasmid, and Bel7402 cells were transfected with the LOXL2 shRNA.
Stable cell lines that over-expressed or downregulated LOXL2 were established and tentatively designated as HepG2-LOXL2 and Bel7402-shLOXL2 cells ( Figure 2B). The expression of LOXL2 in these cells was confirmed by western blot assay, immunofluorescence analyses and qRT-PCR ( Figure 2B-D).
Previous studies have shown that VM was associated with cell migration and invasion. 35,36 Therefore, the migration and invasion potential of the HepG2-LOXL2 and Bel7402-shLOXL2 cells were examined in vitro using a wound healing assay and Transwell migration assay. The results indicated that the stable overexpression of LOXL2 significantly increased the migration and invasion abilities of the HepG2 cells in vitro. Moreover, a significant decrease in cell migration and invasion was observed following the silencing of LOXL2 in the Bel7402-shLOXL2 cells ( Figure 3A,B). These data indicate that LOXL2 overexpression in HCC cells increased their migration and invasion abilities.
Next, to examine the influence of LOXL2 on VM, three-dimensional cultures were used, and the results showed that VM was significantly increased in the HepG2-LOXL2 cells compared with the control cells (P < 0.05) ( Figure 3C)  Recently, Peinado H's research revealed that LOXL2 interact and cooperate with Snail to downregulate E-cadherin expression, then induce an EMT progress. 29 Also, in our study, we found that there is a positive association between LOXL2 and SNAIL. Giving that SNAIL is a very key factor of LOXL2, we have also studied the relationship between SNAIL expression and LOXL2 expression in vitro. The result showed that when LOXL2 expression level is higher in HepG2-LOXL2 cells, comparing with control cells, the SNAIL expression level is higher.
While LOXL2 expression level is lower in Bel7402-shLOXL2 cells, comparing with control cells, the SNAIL expression level is lower ( Figure 3A, B). This finding also gives us a clue to take our mechanism research.

| LOXL2 overexpression promoted tumour growth, metastasis and VM formation in vivo
We   Figure 4B). What's more, the metastasis array results showed that the migration rate in Bel7402-control group is 22%, which is higher than the migration rate (0%) in Bel7402-shLOXL2 group (Figure 4C). Taken together, these results strongly suggest that LOXL2 is required for the efficient growth, VM formation, metastasis and progression of malignant tumours.

| LOXL2 expression promoted VM by collaborating with SNAIL
Recently, the research revealed that the collaboration of LOXL2 with SNAIL and LLGL2 influences EMT and cell polarity respectively. 20,[22][23][24] To gain insight into the mechanism by which LOXL2 promotes VM, we studied the expression of LOXL2-asso-    Figure 5D). Therefore, we propose that LOXL2 may increase SNAIL expression, thereby enabling VM.
All of these data from the HCC tissue samples, HCC cell lines and xenograft model preliminarily verify the mechanism by which LOXL2 promotes VM.

| DISCUSSION
Tumour growth and invasion are dependent on a persistent blood supply. 5 Here, we showed that LOXL2 overexpression promotes VM, which is a form of angiogenesis, and has a good predictive value for HCC metastasis and prognosis; thus, these results can provide deeper insights into tumour progression.
Vasculogenic mimicry has increasingly been considered to function simultaneously with angiogenesis. 9 tumour types that are aggressive, highly metastatic and inclined to poor differentiation. 6,10,13,14 The term VM refers to the ability of aggressive tumour cells to form periodic acid-Schiff-positive and CD31-negative cells that line VM networks in vivo and form tubular structures and patterned networks in three-dimensional (3D) cultures in vitro. 10,12 The results of the three-dimensional cultures of HCC cells and endomucin/PAS double staining in the mouse tumour tissues showed that LOXL2 promoted VM. 6,14,38,39 This result agrees with other reports showing that LOXL2 is involved in angiogenesis. Due to the special function of VM, tumour cells could be exposed to blood flow, allowing them to enter the microcirculation and metastasize to other organs. 9,27 Thus, VM is an important factor in tumour progression that not only leads to a poor prognosis but also explains at least part of the poor targeting of endothelial cells by anti-angiogenic treatments. 38 Our previous research data have revealed that patients with VM + tumours tend to have poorer outcomes than patients with VM − tumours. 11,15 Furthermore, the presence of VM in highly invasive tumours is linked to a high tumour grade, invasiveness, metastasis and shorter survival. 15 The relationships between LOXL2-c expression and tumour size and grade were detected in our study. We found that LOXL2-c expression was significantly correlated with tumour grade and metastasis, which is consistent with a previous study. 25 29 We proposed that LOXL2 may promote VM by collaborating with SNAIL.
As shown in a previous study, EMT regulators and VM are correlated, and many EMT regulators, such as Twist and Slug, can promote VM. 14,33,41 Similarly, in the present study, we found that in another EMT regulator, the expression of SNAIL correlates with VM in HCC, and SNAIL itself exerts a positive effect on cell migration and invasion and capillary tube formation. Interestingly, we found that SNAIL can promote VM in HCC cell which is consistent to Sun D's research. 33 All of these results indicated that LOXL2 promotes VM in HCC by collaborating with SNAIL. Our results also showed that LOXL2 positively correlated with VE-cadherin expression and VM. VE-cadherin is one of the first molecules that was identified as a VM promoter, and it is critical in VM. 14,35,42,43 Together, these results indicate that LOXL2 collaborated with SNAIL, further promoted the expression of the VM marker VE-cadherin, and increased VM in HCC. In addition, the results that the LOXL2 protein and mRNA levels were negatively correlated with LLGL2 expression in HCC tumour tissues and HCC cell lines were detected in this study.  Tieju Liu, Bing Shao and Yong Wang wrote the paper.

CONF LICT OF I NTEREST
No conflicts of interest were disclosed.