27‐Hydroxycholesterol promotes metastasis by SULT2A1‐dependent alteration in hepatocellular carcinoma

Abstract Oxysterol metabolism plays an important role in the initiation and development of various tumors. However, little is known that the metabolic alternation can promote the metastasis of hepatocellular carcinoma (HCC). In this study, we identify the sulfotransferase family 2A member 1 (SULT2A1) to 27‐hydroxycholesterol (27‐OHC) metabolic axis as playing a critical role in HCC metastasis. The level of 27‐OHC closely corresponded with HCC metastasis instead of proliferation in vitro and in vivo. Also, the expression of SULT2A1 is extremely downregulated in human HCC tissues and is correlated with poor prognosis and tumor metastasis. Gain‐ and loss‐of‐function studies reveal that SULT2A1 suppresses the metastasis of HCC by regulating the level of 27‐OHC. Further mechanistic studies indicated that SULT2A1‐dependent alternation of 27‐OHC activates the nuclear factor‐κB signaling pathway and promotes HCC metastasis by enhancing Twist1 expression and epithelial–mesenchymal transition. In conclusion, our findings indicate the relationship between the metabolism of 27‐OHC and the metastasis of HCC. Moreover, SULT2A1 could act as a potential prognostic biomarker and a therapeutic target for preventing HCC metastasis.

metabolisms is involved in the progression and metastasis of HCC. 6,7 Therefore, understanding the relationship between the metabolism and HCC is very important.
Oxysterols are oxidized forms of cholesterol or of its precursors.
Constituting a large family of lipids (i.e., the oxysterome), they are mainly produced through two pathways, enzymatic or nonenzymatic oxidation reaction. 8 Previously, it has been shown that dysfunction of oxysterol can lead to metabolic, inflammatory, and neurodegenerative diseases, [9][10][11] involving many physiological processes such as membrane fluidity, membrane protein activity, 12 vesicle trafficking, and cytoskeleton function. 12, 13 27-Hydroxycholesterol, an important member of the oxysterols family, has received less attention but is increasingly being recognized in cancer cells. It has been reported that 27-OHC can show pro-oncogenic effects through regulating inflammation and several signaling pathways, 14 or through oxysterol-binding proteins. 15 27-Hydroxycholesterol can mainly be metabolized by sulfation. Sulfotransferases are considered to be exclusive enzymes for sulfidic oxysterols. 16,17 However, their roles in oxysterol metabolism as well as in cancer development remains uncertain.
Metastasis is a significant hallmark of cancer and leads to most HCC-related deaths. 18 To date, several researchers have verified the relationship between specific metabolic alternations and HCC metastasis. 19,20 In this study, we identify the SULT2A1-27-OHC metabolic axis as a key player in HCC metastasis instead of proliferation.
Moreover, considered as a prognostic biomarker and a therapeutic target, its potential value might reduce the mortality caused by HCC metastasis.

| MATERIAL S AND ME THODS
Detailed Material and Methods are shown in Appendix S1. To verify these findings, the effects of 27-OHC were further evaluated on in vivo tumor growth and metastasis of HCC xenografts. 27-Hydroxycholesterol did not show an obvious effect on tumor growth of HCC xenografts in subcutaneous xenograft models ( Figure 1H). In orthotopic xenograft models, the corresponding HCC subcutaneous xenografts were isolated and implanted into the liver. The 27-OHC group was also treated with 27-OHC 20 mg/kg i.p. every 3 days. Mice was killed and the primary tumors and lungs were resected after 6 weeks ( Figure S1F). Serial sections of every lung tissue were taken and stained with H&E to determine lung metastasis.

| 27-Hydroxycholesterol can increase HCC metastasis without affecting cell proliferation
Almost no lung metastasis lesions were found in nude mice bearing orthotopic xenografts from the control group. However, we observed a significant increase in lung metastasis in nude mice bearing orthotopic xenografts from the 27-OHC treated group ( Figure 1I).
These data suggest a close relationship between 27-OHC and metastasis of HCC.

| Low expression of SULT2A1 is associated with poor prognosis of HCC patients
Next, we examined the possible influencing factor of 27-OHC in HCC. As previously mentioned, SULT, mainly SULT2A1, SULT2B1, and SULT1E1, are responsible for oxysterol metabolism. 16,22 We analyzed the former three enzymes at the transcriptional level using public databases. Data revealed that, compared to the other two enzymes, SULT2A1 showed significant decrease in tumor tissues In (I), arrows refer to lung metastatic lesions. Scale bar, 100 μm. Significance was determined by two-way ANOVA (Bonferroni post test) (H, bottom panel) or Wilcoxon test (I, bottom panel). *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant against normal control (Figures 2A and S2A,B). Furthermore, the expression of SULT2A1 was similarly reduced in HCC vascular metastasis tissues compared with nonvascular metastasis tissues ( Figures 2B and S2C,D), which might suggest the inherent association with HCC. We also analyzed the expression of SULT2A1 among different tumor grades and found that downregulation of SULT2A1 was correlated with highly malignant tumor, consistent with previous findings ( Figure 2C).
We then undertook survival analyses to determine whether downregulation of SULT2A1 might affect prognosis. Using Kaplan-Meier survival analysis, low expression of SULT2A1 was shown to be significantly correlated with worse OS (p = 0.015) and DSS (p = 0.0013) ( Figure 2D,E).
We also used another public cohort to further verify our previous findings. Based on the GEO database (GSE14520), HCC patients with lower expression of SULT2A1 were considered to have worse tumor types and poorer clinical outcomes, which were similar to the above results ( Figure S2E-G).
Subsequently, the clinical association of SULT2A1 expression was examined in our own patients' set by using quantitative real-time PCR and western blotting. It was revealed that both mRNA and protein levels of SULT2A1 were much lower in HCC tissues compared with the paired adjacent nontumor liver tissues ( Figure S3A,B). Furthermore, the cases were grouped according to the MVI grade. Data showed that low expression of SULT2A1 corresponded to a high probability of MVI occurrence ( Figure 2F,G). We also examined the expression of SULT2A1 in the previous tissue microarray 23 containing paired primary HCC tissues, adjacent nontumor liver tissues and PVTT tissues from our research group. Similarly, results showed that PVTT tissues, representing HCC metastasis, had lower expression of SULT2A1 compared with HCC and nontumor liver tissues ( Figure 2H,I).
We then measured the expression of SULT2A1 among HCC cell lines and found that the mRNA and protein levels of SULT2A1 were significantly associated with their metastatic potential ( Figure   S3C,D). Consequently, these results strongly suggest that downregulation of SULT2A1 is obviously correlated with poor prognosis of HCC patients and further indicate that SULT2A1 might play a vital role in the metastasis of HCC.

| Downregulation of SULT2A1 can increase the metastasis of HCC by modulating the level of 27-OHC
We undertook in vitro functional studies to determine the roles of SULT2A1 and 27-OHC in HCC cells. Two SULT2A1-specific shR-NAs were generated to silence SULT2A1 expression (shSULT2A1).
shSULT2A1#1, which induced a more significant knockdown effect, was adopted for knocking down the SULT2A1 expression in Huh7 cells that highly expressed SULT2A1 in previous studies (shown in Figures   S3 and S4A,B). We also overexpressed SULT2A1 in HCC-LM3 cells that expressed SULT2A1 at low levels. First, we verified the relationship be-

| Downregulation of SULT2A1 and increase of 27-OHC lead to EMT in HCC
Epithelial-mesenchymal transition is well known as a promoting mechanism in HCC metastasis. 24,25 We next evaluated whether SULT2A1dependent alteration of 27-OHC could affect EMT. We collected We also checked the EMT status in human HCC tissues with different SULT2A1 expressions. Similar results were found, in that human HCC with lower SULT2A1 expressions (MVI M1-2, based on Figure 2G) had higher EMT levels ( Figure 4F). More importantly, the expression of SULT2A1 and EMT markers was highly correlated by using gray analysis (SULT2A1 and E-cad: Pearson's r = 0.8425; SULT2A1 and N-cad: Pearson's r = −0.7055, p < 0.001) ( Figure 4G).
Collectively, these data show that SULT2A1-dependent alternation of 27-OHC promotes the migration of HCC by regulating EMT.

| Sulfotransferase 2A1-dependent alternation of 27-OHC is correlated with activation of NFκβ signaling pathway, together with elevated Twist1 expression in HCC
As inflammation is critically involved in the pathological progress between oxysterols and diseases, we evaluated the impact of 27-OHC on inflammation. Nuclear factor-κB signaling pathway is widely acknowledged as an essential role in inflammation. In recent years, it has also been increasingly realized as a crucial player in many steps of cancer initiation and progression. 26,27 Moreover, much research has been focused on the relationship between NF-κB and EMT. We next evaluated whether 27-OHC could affect EMT in HCC cells through the NF-κB signaling pathway. Two key factors, NF-κB p65 and phospho-NF-κB p65 (Ser536) (p-p65), were examined by western blotting with or without the addition of 27-OHC in two HCC cell lines. Data showed that when treated with 27-OHC, the expression of p65 and p-p65 were both increased. In Huh7 and HCC-LM3 modulated cells, the reduced SULT2A1 expression obviously increased the expression of p65 and p-p65, and the upregulation of SULT2A1 reversed this alternation following 27-OHC treatment ( Figure 5A,B).
We further evaluated several downstream genes of the NF-κB signaling pathway 28,29 that are connected with tumor migration, and found similar changes corresponding with p-65 and p-p65 as described in Figure 5A

| Upregulation of SULT2A1 inhibits HCC metastasis by suppressing NFκ B signaling pathway and EMT phenotype in vivo
A lung metastasis mouse model was adopted to verify our previous in vitro findings of metastasis of HCC, similar to that shown in Figure 1I We further constructed orthotopic xenograft models by using shSULT2A1 Huh7 cells to investigate the effect of NF-kB inhibitor in vivo. The shSULT2A1 Huh7 cells treated with 27-OHC showed a higher incidence of lung metastatic lesions compared to the shNT group, and the NF-kB inhibitor Bay11-7082 remarkably reduced the metastasis occurrence ( Figure 6G,H). Moreover, the increased status of EMT and the expression of p65 and Twist1, which resulted from 27-OHC and downregulation of SULT2A1, were significantly inhibited by Bay11-7082 ( Figure 6I,J). We also found a strong correlation between p-65 and Twist1 expression (Pearson's r = 0.9379, p < 0.001) ( Figure 6K). Once again, these results confirm that SULT2A1-dependent alternation of 27-OHC could affect EMT by regulating Twist1 through the NF-κB signaling pathway, contributing to HCC metastasis.

| Sulfotransferase 2A1 acts as a potential biomarker in predicting HCC clinical outcomes
To establish a clinically applicable method for predicting the prognosis of HCC patients, we attempted to use SULT2A1 as a potential biomarker to predict HCC relapse based on GEO databases (GSE14520).
Surprisingly, by analyzing the cumulative relapse event, we discovered an obvious trend in less probability of HCC relapse with high expression of SULT2A1 compared with low expression of SULT2A1 ( Figure 7A). We also carried out receiver operating characteristic analyses using SULT2A1 based on a GEO database (GSE9843) and found potential value in predicting HCC metastasis (AUC = 0.664) ( Figure 7B).
We established a prognostic nomogram to predict the survival probability at 1, 3, and 5 years based on the GEO databases (GSE14520). Eight independent prognostic parameters, including α-fetoprotein, alanine aminotransferase, gender, age, cirrhosis, expression of SULT2A1, tumor stage, and relapse, were enrolled in the prediction model ( Figure 7C). The calibration plots ( Figure 7D

| DISCUSS ION
As revealed by previous researchers, oxysterol is closely related to multiple cancer types, including cancers of the colon, lung, skin, breast, and bile ducts. It has been shown that oxysterol induced cell proliferation and metastasis by promoting oxidative stress and inflammation. 31 Other studies consider oxysterol a potential predictor of cancer risk. 32 Here, we discovered a close relationship be- reported that 25-OHC could promote HCC metastasis, contrary to our findings. 40 We consider the distinction might be attributed to differences in cell lines (i.e., HepG2 cells). The relationship between SULT and cancer progression has not been well determined. Several early studies discovered the distinct expression patterns of SULT in human cancers, 43,44 considering it as an independent influencing factor that can act on different signaling pathways. 45,46 Here, we first linked SULT with HCC metastasis in the metabolic pathway, specifically, that SULT2A1 can affect HCC metastasis by regulating the level of 27-OHC. Our study has also shown that the expression of SULT2A1 was significantly downregulated, especially in HCC tissues with high metastatic potential. In addition, SULT2A1 has a great capability to predict vascular metastasis and relapse. Therefore, activation of SULT2A1 could be a possible approach for antimetastasis in HCC.
Since it was first identified as an endogenous selective ER modulator and an agonist of LXR, 14,47 27-OHC has been considered to have a close relationship with receptors. By acting on ER, 27-OHC can increase tumor growth and metastasis in breast cancer, 48,49 together with functioning as a novel mechanism of resistance to endocrine therapy. 50 27-Hydroxycholesterol can also promote the growth of melanoma cells by activating ERα. 51 Moreover, LXR is a major target receptor of 27HC, given its purported role in maintaining cholesterol homeostasis by promoting the efflux and reverse transport of cholesterol to the outside of cells. 52 27-Hydroxycholesterol can play a pro-tumorigenic role by suppressing the expression of various inflammatory genes in macrophages and other immune cells through LXR activation. 53,54 Interestingly, we note that LXR activation might increase SULT2A1 mRNA levels in human LNCaP prostate cancer cells based on previous research, 55 and the formation of a range of oxysterol sulfates shows antagonistic effects on LXRs. 56 The role of LXR in the regulation of SULT expression is still not well determined.
Whether LXR functions in the SULT2A1-27-OHC metabolic axis and in turn impacts on HCC metastasis remains unknown and requires further studies.
In conclusion, our study indicates that the SULT2A1-27-OHC metabolic axis is a key player in HCC metastasis. Downregulation of SULT2A1 increases the levels of 27-OHC and promotes the F I G U R E 5 Sulfotransferase family 2A member 1 (SULT2A1)-dependent alternation of 27-hydroxycholesterol (27-OHC) is correlated with the activation of the nuclear factor-κB (NFκβ) signaling pathway, together with elevated Twist1 expression in hepatocellular carcinoma (HCC). (A, B) Expression of p65, p-p65, and epithelial-mesenchymal transition (EMT)-related transcription factors were determined by western blotting analysis. Huh7 (including shSULT2A1 group) (A) and HCC-LM3 (including SULT2A1 group) (B) were treated with or without 2 μM 27-OHC for 72 h. (C, D) Several downstream genes of the NF-κB signaling pathway which connected with tumor migration were evaluated by Quantitative RT-PCR analysis after treatment with 27-OHC in SULT2A1-shRNA stably expressed Huh7 (C) and SULT2A1 overexpressed HCC-LM3 (D) cells. *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant. (E, F) Expression of p65, p-p65, Twist1, and EMT markers in Huh7 (E) and HCC-LM3 (F) cells were determined by western blot analysis. All groups were treated with or without 2 μM 27-OHC and 5 μM Bay11-7082 for 72 h. (G, H) Expression of p65, p-p65, Twist1, and EMT markers were determined by western blot analysis. Huh7 (including shSULT2A1 group) (G) and HCC-LM3 (including SULT2A1 group) (H) were treated with or without 2 μM 27-OHC and 5 μM Bay11-7082 for 72 h F I G U R E 6 Upregulation of sulfotransferase family 2A member 1 (SULT2A1) suppresses the nuclear factor-κB (NF-κB) signaling pathway and epithelial-mesenchymal transition (EMT) phenotype in vivo, leading to the inhibition of hepatocellular carcinoma (HCC) metastasis. (A, B) SULT2A1overexpressed HCC-LM3 cells and the control group treated with 27-OHC or not were used to establish orthotopic xenograft models. Representative H&E staining images of lung tissues (A) and the average numbers of lung metastatic lesions (B) from six mice per group are shown. Arrows refer to lung metastatic lesions. (C-F) Levels of EMT markers (C, D) and p65 and Twist1 (E, F) from SULT2A1-overexpressed HCC tissues and the control group were analyzed by immunohistochemistry (IHC). Representative images of IHC staining from six mice per group and the IHC H scores are shown. (G, H) shSULT2A1 Huh7 cells and the shNT group treated with 27-OHC or Bay11-7082 were used to establish orthotopic xenograft models. Representative H&E staining images of lung tissues (G) and the average numbers of lung metastatic lesions (H) from six mice per group are shown. Arrows refer to lung metastatic lesions. (I, J) Levels of EMT markers and p65 and Twist1 from shSULT2A1 HCC tissues and the shNT group were analyzed by IHC. Representative images of IHC staining from six mice per group (I) and the IHC H scores (J) are shown. Scale bar, 100 μm (A, C, E, G, I). Significance was determined by Student's t-test (B, D, F, H, J). *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant. K, Correlation between p65 expressions and Twist1 expression in HCC tissues based on the IHC H score. The correlation was then analyzed by Pearson correlation analysis. ***p < 0.001. shNT, shRNA nontarget F I G U R E 7 Sulfotransferase family 2A member 1 (SULT2A1) is a potential biomarker in predicting hepatocellular carcinoma (HCC) clinical outcomes. (A) Cumulative relapse event to predict HCC relapse using SULT2A1 based on the Gene Expression Omnibus (GEO) dataset (GSE14520). The median SULT2A1 expression was used as a cut-off value. (B) Receiver operating characteristic (ROC) curves to predict metastasis (vascular invasion) using SULT2A1 based on the GEO dataset (GSE9843). (C-E) Prognostic nomogram to predict the survival of HCC patients based on The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) dataset (C). Red circles show clinical characteristics of one specific patient with low expression of SULT2A1 from the GEO dataset (GSE14520). Calibration curves (D) and ROC curves (E) of the nomogram for predicting survival at 1, 3, and 5 years in the GSE14520 dataset are shown. AFP, α-fetoprotein; ALT, alanine aminotransferase; AUC, area under the ROC curve; OS, overall survival metastasis of HCC by activation of the NFκβ signaling pathway together with elevating Twist1 expression and eventually affecting the EMT. Sulfotransferase 2A1 could become a novel prognostic marker for HCC metastasis, and discovering agonists for enhancing SULT2A1 activity could be a potential therapy to suppress the metastasis of HCC.

D I SCLOS U R E
The authors have no conflict of interest.