Chronic Activation of LXRα Sensitizes Mice to Hepatocellular Carcinoma.

The oxysterol receptor liver X receptor (LXR) is a nuclear receptor best known for its function in the regulation of lipid and cholesterol metabolism. LXRs, both the α and β isoforms, have been suggested as potential therapeutic targets for several cancer types. However, there was a lack of report on whether and how LXRα plays a role in the development of hepatocellular carcinoma (HCC). In the current study, we found that systemic activation of LXRα in the VP-LXRα knock-in (LXRαKI) mice or hepatocyte-specific activation of LXRα in the VP-LXRα transgenic mice sensitized mice to liver tumorigenesis induced by the combined treatment of diethylnitrosamine (DEN) and 3,3',5,5'-tetrachloro-1,4-bis (pyridyloxy) benzene (TCPOBOP). Mechanistically, the LXRα-responsive up-regulation of interleukin-6 (IL-6)/signal transducer and activator of transcription 3 (STAT3) signaling pathway and the complement system, and down-regulation of bile acid metabolism, may have contributed to increased tumorigenesis. Accumulations of secondary bile acids and oxysterols were found in both the serum and liver tissue of LXRα activated mice. We also observed an induction of monocytic myeloid-derived suppressor cells accompanied by down-regulation of dendritic cells and cytotoxic T cells in DEN/TCPOBOP-induced liver tumors, indicating that chronic activation of LXRα may have led to the activation of innate immune suppression. The HCC sensitizing effect of LXRα activation was also observed in the c-MYC driven HCC model.
Conclusion: Our results indicated that chronic activation of LXRα promotes HCC, at least in part, by promoting innate immune suppressor as a result of accumulation of oxysterols, as well as up-regulation of the IL-6/Janus kinase/STAT3 signaling and complement pathways.

cholangiocarcinoma

cytochrome P450

CXC chemokine receptor 2

diethylnitrosamine

fatty acid binding protein

farnesoid X receptor

granulocyte and macrophage progenitor cell

gene‐set enrichment analysis

20‐hydroxycholesterol

hepatocellular carcinoma

hyodeoxycholic acid

interleukin‐6

Janus kinase

liver‐enriched activator protein

liver X receptor

myeloid‐derived suppressor cell

messenger RNA

polymerase chain reaction

RNA‐sequencing

signal transducer and activator of transcription 3

The Cancer Genome Atlas

3,3',5,5'‐tetrachloro‐1,4‐bis (pyridyloxy) benzene

taurodeoxycholic acid

tauroursodeoxycholic acid

ultra‐performance liquid chromatography

wild‐type

ω‐muricholic acid

Hepatocellular carcinoma (HCC) is the major form of primary liver cancers and the second‐most lethal cancer after the pancreatic cancer. The classical risk factors for HCC include viral infection and liver toxins.( , ) The increase in nonalcoholic fatty liver disease (NAFLD), together with metabolic syndrome and obesity, amplifies the risk of liver cancer. NAFLD is expected to become a leading cause of liver cancer in Western countries,( ) suggesting the critical role of lipid disorder in the development of HCC. Commonly used rodent models of HCC include those induced by the combined treatment of the tumor initiator diethylnitrosamine (DEN) and tumor promoter 3,3',5,5'‐tetrachloro‐1,4‐bis (pyridyloxy) benzene (TCPOBOP),( ) or the c‐MYC oncogene.( )

The oxysterol receptors liver X receptor α (LXRα) and LXRβ play a central role in the regulation of lipid and cholesterol metabolism. LXRα is highly expressed in the liver, but it is also found in adipose tissue, intestines, kidneys and macrophages, whereas LXRβ expression is detectable in most tissues.( ) LXRs have been proposed to be therapeutic targets for several cancer types, including breast cancer,( , ) prostate cancer,( ) and HCC,( , , , , ) through multiple pathways, ranging from promoting cholesterol metabolism( , , ) to cross‐talk with transforming growth factor β,( , ) and dampening innate immune responses.( ) More specifically, it has been reported that activation of LXRs inhibited tumor cell proliferation by promoting cholesterol catabolism and reducing intracellular cholesterol content.( , ) A recent study showed that activation of LXRβ reduced myeloid‐derived suppressor cells (MDSCs), an immature myeloid heterogeneous population markedly expanded and accumulated in the tumor microenvironment, in murine models and in patients treated the small‐molecule LXR agonist RGX‐104.( ) However, the cancer inhibitory effect of LXR activation remains controversial. For example, a recent report showed that pharmacological inhibition of LXR using the synthetic LXR inverse agonist SR9243 induced tumor destruction, primarily through stimulation of CD8+ T‐cell cytotoxic activity and mitochondrial metabolism in vitro and in vivo.( )

LXRs are also known as anti‐inflammatory transcription factors and physiological regulators of innate and adaptive immune responses. Activation of LXRs resulted in transcriptional silencing of the pro‐inflammatory transcription factor nuclear factor kappa B( ) and dampening of the antitumor responses of dendritic cells (DCs).( ) LXRα null mice were reported to be more susceptible to bacterial infection and showed accelerated apoptosis.( ) Knowing chronic inflammation is a risk factor for cancers, the anti‐inflammatory activity of LXRs may have also contributed to their antitumor activities.

Most of the reported effects of LXRs on cancers have relied on the short‐term use of LXR agonists or antagonists. Because LXR modulators are not typical cytotoxic chemotherapeutic agents, it is conceivable that their use in the clinical settings is likely to be chronic. As such, it is necessary to evaluate the effect of chronic activation of LXRs on cancers, including HCC.

In this study, we were surprised to find that chronic activation of LXRα systemically, or specifically in hepatocytes, sensitized mice to chemical and oncogenic models of HCC.

Materials and Methods

Chemicals

DEN and TCPOBOP were purchased from Sigma‐Aldrich (St. Louis, MO).

Animals and Mouse Models of HCC

The creation and characterization of VP‐LXRα knock‐in (LXRαKI)( ) and fatty acid binding protein (FABP)–VP‐LXRα transgenic mice( ) were previously reported. To establish chemical‐induced HCC, 5‐week‐old male mice were injected with a single dose of DEN (90 mg/kg body weight, intraperitoneally). A week after the DEN injection, mice received biweekly intraperitoneal injections of TCPOBOP (3 mg/kg body weight) for 14 weeks.( ) Mice were sacrificed and tissues were harvested 2 weeks after the last injection of TCPOBOP. The tet‐off liver‐enriched activator protein (LAP)–MYC transgenic mice have been described previously.( ) LAP‐MYC/LXRαKI mice were generated by breeding the LAP‐MYC transgene into the LXRαKI background. LAP‐MYC and LAP‐MYC/LXRαKI mice were maintained with drinking water containing 2 mg/mL of doxycycline (dox) to silence the transgene, and the expression of MYC was induced by withdrawing dox. All mice were maintained in C57/BL6 background, and only male mice were used in this study due to their higher sensitivity to experimental HCC. The use of mice in this study has complied with all relevant federal guidelines and institutional policies, and has been approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Histology and Immunohistochemistry

The liver tissues were freshly harvested and fixed in 10% neutral‐buffered formalin for 24 hours. The tissues were histologically processed, embedded in paraffin, sectioned at 4 μm, and stained with hematoxylin and eosin (H&E) for general histology. For CD45 and Ki67 immunostaining, de‐paraffinized sections were incubated with anti‐CD45 antibody (Cat#70257) from Cell Signaling (Danvers, MA) at 1:100 dilution, or anti‐Ki67 antibody from Abcam (Cambridge, MA) at 1:200 dilution overnight at 4°C. The antibody signals were visualized by peroxidase reaction using 3,3′‐diaminobenzidine as the chromogen. Hematoxylin was used as a nuclear counterstain. At least three mice were used for each treatment group, and for each sample at least four noncontiguous regions were photographed and analyzed. Quantification of Ki67 positive stain was calculated by ImageJ.( )

RNA‐Sequencing Analysis

RNA‐sequencing (RNA‐seq) was performed at the Health Sciences Sequencing Core at the Children’s Hospital of Pittsburgh. Gene expression was analyzed by gene‐set enrichment analysis.( )

Real‐Time Reverse‐Transcription Polymerase Chain Reaction

Total RNA was extracted from tissues using TRIzol reagent. Total RNA was treated with RNase‐free DNase I and reverse‐transcribed into single‐stranded complementary DNA. SYBR Green–based real‐time polymerase chain reaction (PCR) was performed with the ABI 7300 real‐time PCR System. Data were normalized against the housekeeping gene cyclophilin. Relative gene expression was calculated using the ΔΔCT method, in which fold difference was calculated using the expression 2−ΔΔCT. The primer sequences are provided in Supporting Table S1.

Flow Cytometry and Fluorescence‐Activated Cell Sorting

Tumor dissection and digestion were performed as described.( ) In brief, tissue samples were ground and digested with 0.25 mg/mL Liberase TL (Roche, Indianapolis, Indiana) and 0.3 mg/mL DNase (Fisher Scientific, Hampton, NH) for 30 minutes at 37°C. Single‐cell suspensions were filtered through a 100‐μm cell strainer. Multiparameter staining was used to identify the immune cell populations as followings: (1) CD8+ T cells (CD45+ CD8+) and CD4+ T cells (CD45+ CD4+); (2) macrophage M1 type (CD45+CD11b+ F4/80+ CD206−) and macrophage M2 type (CD45+ CD11b+ F4/80+ CD206+); (3) DCs (CD45+MHCII+ CD11b+ Gr‐1− F4/80− CD103+); (4) MDSCs (monocytic myeloid–derived suppressor cells [Mo‐MDSCs]: CD45+ CD11b+ Gr‐1int; granulocytic‐MDSCs [G‐MDSCs]: CD45+ CD11b+ G‐1high); and (5) granulocyte and macrophage progenitor cells (GMPs) (CD45+Lin−c‐KithighSca1−FcγRII/IIIhighCD34+). The stained cells were applied for data acquisition using Cytek Aurora flow cytometer (Cytek Biosciences, Fremont, CA) and analyzed by FlowJo software (TreeStar Inc., San Carlos, CA). The fluorescence‐activated cell sorting analysis of MDSCs was conducted as previously reported.( ) Single cells were labeled with anti‐CD11b‐APC, anti‐Ly6C–fluorescein isothiocyanate, and anti‐GR‐1‐PE. G‐MDSCs are defined as CD11b+Gr‐1highLy6Clow, and Mo‐MDSCs are defined as CD11b+Gr‐1intLy6Clow. All of the antibodies used for flow cytometry are listed in Supporting Table S2.

Ultra‐Performance Liquid Chromatography–Mass Spectrometry Analysis of Bile Acids and Oxysterols

The serum and liver bile acids were extracted and measured following the method described previously.( ) Briefly, 50 μL of serum sample was mixed with 150 μL of methanol, followed by vortexing for 30 seconds and centrifugation at 15,000g for 10 minutes. Liver samples were homogenized in water (100 mg tissue in 400 μL water), and then a 200‐μL aliquot of methanol was added to 100 μL of liver homogenate. After vortex and centrifugation at 15,000g for 20 minutes, the supernatant was transferred to a new Eppendorf tube for a second centrifugation. A total of 2 μL of the supernatant was injected onto the ultra‐performance liquid chromatography (UPLC) and quadrupole time‐of‐flight mass spectrometry system for metabolite analysis. Chromatographic separation of bile acids was performed on an Acquity UPLC BEH C18 column (2.1 × 100 mm, 1.7 μm; Waters Corporation, Milford, MA) using acetonitrile/water containing 0.1% formic acid as the mobile phase.( ) The detailed parameters for mass spectrometry were the same as previously reported.( )

The liver contents of oxysterols were measured by the Duke University Proteomics and Metabolomics Shared Resources using a customized UPLC electrospray ionization tandem mass spectrometry method, allowing chromatographic resolution of the isobaric hydroxycholesterol species. The extracting and measurement method of oxysterols involve alkaline hydrolysis of lipid esters followed by solid phase extraction (HyperSep C18 SPE [200 mg]) before liquid chromatography–mass spectrometry analysis. The detailed methods for oxysterol measurements were described previously.( ) Liver concentrations of 20‐hydroxycholesterol (HC), 22(R)‐HC, 22(S)‐HC, 24(R/S)‐HC, 25‐HC, 27‐HC, 7α/β‐HC, and cholesterol concentrations were semi‐quantitative, calculated based on a ratio to each internal standard (stable‐isotope dilution).

Statistical Analysis

Results are expressed as mean ± SEM. Differences between two individual groups were determined by Student’s t test. Differences between multiple groups were evaluated using two‐way analysis of variance followed by post‐hoc multiple comparison according to the Tukey’s test. Pearson chi‐square and Fisher exact were used for between‐group comparisons for tumor incidence. Statistical significance was accepted at P < 0.05.

Results

Increased Expression of LXRα in Patients With HCC Is Associated With Poor Survival

To determine the expression of LXR in human HCC, we analyzed The Cancer Genome Atlas (TCGA) HCC data sets, including gene‐expression profile and clinical features through GEPIA at http://gepia.cancer‐pku.cn/.( ) The transcripts of both LXRα and LXRβ tended to be higher in HCC compared with normal liver tissues, but the difference did not reach statistical significance (Fig. 1A). The higher expression of both LXRα and LXRβ was associated with lower survival rates of patients with HCC, but the association with LXRα expression was statistically more significant (Fig. 1B). To determine the expression of LXRα and LXRβ in different liver cancer types and disease stages, two online Gene Expression Omnibus data sets downloaded from Oncomine (https://oncomine.org) were analyzed. Analysis of GSE15765 revealed that the messenger RNA (mRNA) expression of LXRα was significantly higher in HCC tumors than in cholangiocarcinoma (CC) tumors (Fig. 1C, top panel), but the expression of LXRβ was not different between HCC and CC (Fig. 1C, bottom panel). Analysis of GSE6764 showed that the expression of LXRα decreased in cirrhosis and pre‐neoplastic hyperplastic livers compared with normal livers, after which the expression of LXRα increased as the disease progresses, leading to elevated expression of LXRα in advanced HCC compared to hyperplastic livers (Fig. 1D, left panel). In the same cohort of patients, the disease stages had little effect on the expression of LXRβ (Fig. 1D, right panel). These results suggest that elevated expression of LXRα was associated with the pathogenesis and poor prognosis of human HCC.

FIG. 1

Increased expression of LXRα in patients with HCC is associated with poor survival. (A) Analysis of LXRα and LXRβ gene expression in the TCGA HCC data set (normal control n = 50; HCC n = 369). (B) The overall survival curves of patients with HCC of high (n = 182) (high cutoff > 50% median) or low (n = 182) (low cutoff < 50% median) expression of LXRα or LXRβ. (C) Comparative expression of LXRα and LXRβ in patients with HCC (n = 70) and CC (n = 13). (D) Comparative expression of LXRα and LXRβ in patients with HCC of different stages (normal liver tissue n = 10; cirrhotic liver tissue n = 13; low‐grade dysplastic liver tissue n = 10; high‐grade dysplastic liver tissue n = 7; very early HCC n = 4; early HCC n = 8; advanced HCC n = 4; and very advanced HCC n = 11). *P < 0.05; **P < 0.01; the comparisons are labeled.

Chronic Activation of LXRα Sensitizes Mice to Chemical‐Induced HCC

We then used LXRα gain‐of‐function models, VP‐LXRα knock‐in (LXRαKI) mice and VP‐LXRα transgenic mice, to investigate the role of LXRα activation in liver carcinogenesis. As outlined in Fig. 2A, the LXRαKI mice were created by knocking the constitutively activated VP‐LXRα into the endogenous LXRα gene locus,( ) whereas the VP‐LXRα transgenic mice bear hepatocyte‐specific expression of VP‐LXRα under the control of the FABP gene promoter.( ) VP‐LXRα was constructed by fusing the VP16 activation domain of the herpes simplex virus to the amino terminus of mouse LXRα sequence.

FIG. 2

Chronic activation of LXRα sensitizes mice to chemical‐induced HCC. (A) Schematic representation of the VP‐LXRα knock‐in (LXRαKI) mice, in which VP‐LXRα was knocked into the endogenous LXRα gene locus, and FABP‐VP‐LXRα transgenic mice expressing VP‐LXRα in the hepatocytes under the control of hepatocyte‐specific FABP gene promoter. (B) Scheme of the DEN/TCPOBOP model of HCC. Five‐week‐old male mice were injected with a single dose of DEN (90 mg/kg body weight, intraperitoneally). A week after the DEN injection, mice received biweekly intraperitoneal injections of TCPOBOP (3 mg/kg body weight) for 14 weeks, and mice were sacrificed 2 weeks after the last injection of TCPOBOP. (C) The expression of VP‐LXRα and LXRβ in nontumor and tumor tissues was measured by real‐time PCR. (D,E) Liver tumor incidence (D) and multiplicity (E) were calculated in WT, LXRαKI, and VP‐LXRα mice. (F) Gross appearance of the livers following the completion of the DEN/TCPOBOP treatment. Dashed circles indicate tumor nodules. (G) Liver histology was analyzed by H&E staining. Shown on the right are magnified areas of corresponding smaller boxes on the left. Bar is 100 μm. (G) Tumor cell proliferation was analyzed by immunohistochemistry staining of Ki67. Shown on the bottom are magnified areas of corresponding smaller boxes on the top. Arrowheads indicate Ki67‐positive cells. Bar is 100 μm. Shown on the right is the quantification of the Ki67‐positive cells within the tumor areas. *P < 0.05; **P < 0.01. Abbreviations: ATG, anti‐thymocyte globulin; ns, statistically not significant, compared with the WT group, or the comparisons are labeled; and TGA, transglutaminase IgA.

We initially used the 36 weeks DEN‐induced liver carcinogenesis model( ) as outlined in Supporting Fig. S1A, but found the DEN alone regimen was not efficient to induce HCC in wild‐type (WT), LXRαKI, or VP‐LXRα mice (Supporting Fig. S1B). The lack of tumor formation in the DEN alone model was consistent with a previous report.( ) TCPOBOP, a constitutive androstane receptor (CAR) agonist that promotes hepatocyte proliferation, is an established liver‐cancer promoter following the initiation of DEN.( ) We then subjected mice to the DEN/TCPOBOP model of liver cancer.( ) In this model, mice were injected with a single intraperitoneal dose of DEN (90 mg/kg body weight) at 5 weeks of age, followed by biweekly intraperitoneal injections of TCPOBOP (3 mg/kg body weight) for 16 weeks as outline in Fig. 2B. The expression of the transgenic VP‐LXRα in the nontumor and tumor tissues of LXRαKI and VP‐LXRα mice was confirmed by real‐time PCR, whereas the expression of LXRβ was not affected by the transgene (Fig. 2C). Compared with their WT counterparts, DEN/TCPOBOP‐treated LXRαKI and VP‐LXRα mice showed higher tumor incidence (Fig. 2D) and multiplicity (Fig. 2E), as supported by the gross appearance of the liver (Fig. 2F) and H&E staining of liver sections (Fig. 2G). Immunohistochemical staining of Ki67 also showed a more robust proliferation of tumor cells in LXRαKI and VP‐LXRα mice (Fig. 2H). The tumor multiplicity (Fig. 2D) and Ki67 staining (Fig. 2G) were not statistically different between the LXRαKI and VP‐LXRα mice. As expected, the hepatic expression of Cyp2b10, the primary target gene of CAR, was up‐regulated in TCPOBOP‐treated mice of all three genotypes (Supporting Fig. S1C).

Interestingly, LXRα ablation had little effect on animal’s sensitivity to the DEN/TCPOBOP model of HCC, as evidenced by unchanged liver to body weight ratio (Supporting Fig. S2A), tumor incidence (Supporting Fig. S2B), and tumor multiplicity (Supporting Fig. S2C). It remains to be determined whether there will be a compensatory regulation of LXRβ in the LXRα null mice, and if so, whether this compensatory regulation of LXRβ will have functional consequence in the absence of exogenously added LXR agonists in vivo.

Chronic Activation of LXRα Up‐Regulates Pathways Associated With Innate Immune Suppression, but Down‐Regulates the Bile Acid Metabolism Pathway in Chemical‐Induced HCC

To understand the mechanism by which activation of LXRα sensitizes mice to DEN/TCPOBOP‐induced HCC, we performed RNA‐seq analysis on liver tumors derived from WT, LXRαKI, and VP‐LXRα mice. There were eight pathways commonly up‐regulated in both LXRαKI and VP‐LXRα mice that are known to be involved tumor progression and immune responses (Fig. 3A). Among the commonly up‐regulated pathways, our gene‐set enrichment analysis (GSEA) analysis validated the activations of interleukin‐6 (IL‐6)/Janus kinase (JAK)/signal transducer and activator of transcription 3 (STAT3) signaling (Fig. 3B) and complement (Fig. 3C) pathways, both of which have been reported to play important roles in innate immune suppression, including the recruitment of MDSCs to reduce cytotoxic T‐cell responses, as well as induction of tumor metastasis and proliferation.( , ) The up‐regulation of representative genes in the IL6/JAK/STAT3 signaling pathway (Fig. 3D) and complement pathway (Fig. 3E) were validated by real‐time reverse‐transcription PCR.

FIG. 3

Chronic activation of LXRα up‐regulates pathways associated with innate immune suppression, but down‐regulates the bile acid metabolism pathway in chemical‐induced HCC. (A) Venn plot derived from RNA‐seq analysis shows the pathways up‐regulated in tumors from LXRαKI and VP‐LXRα transgenic mice as compared with WT mice. Eight commonly up‐regulated pathways are shown on the right. (B,C) GSEA for HALMARK_IL6_JAK_STAT3_SIGNALING (B) and HALMARK_COMPLEMENT (C) in tumors derived from LXRαKI (left panels) and VP‐LXRα (right panels) mice as compared with WT mice. (D,E) The tumor mRNA expression of representative genes in the IL‐6/JAK/STAT3 signaling pathway (D) and complement pathway (E). (F) Venn plot derived from RNA‐seq analysis shows the pathways down‐regulated in tumors from LXRαKI and VP‐LXRα mice as compared with WT mice. Five pathways commonly down‐regulated are shown on the right. (G) GSEA for HALMARK_BILE_ACID_METABOLISM in tumors from LXRαKI (left) and VP‐LXRα (right) mice (n = 6 for each group). *P < 0.05, compared with WT groups in (D) and (E).

Meanwhile, there were five metabolism‐related pathways commonly down‐regulated in DEN/TCPOBOP‐treated LXRαKI and VP‐LXRα mice, among which the down‐regulation of bile acid metabolism pathway was ranked at the top (Fig. 3F). GSEA analysis further confirmed the down‐regulation of the bile acid metabolism pathway in LXRαKI and VP‐LXRα mice (Fig. 3G), suggesting a potential role of LXRα responsive regulation of the bile acid metabolism in the development of HCC.

Chronic Activation of LXRα Accumulates Pro‐HCC Bile Acid Species and Oxysterols in Chemical‐Induced HCC

Because our RNA‐seq results suggested a dysregulation of the bile acid metabolism pathway (Fig. 4A), we wanted to validate the LXRα responsive changes in the expression of key enzymes involved in the bile acid metabolism. The up‐regulation of cytochrome P450 (Cyp) 7a1 and down‐regulation of Cyp7b1 and Cyp8b1 in tumors derived from VP‐LXRα mice were verified by real‐time PCR. A similar pattern of regulation was observed in LXRαKI mice (Fig. 4B, top panel). The LXRα‐responsive induction of Cyp7a1 and suppression of Cyp7b1 were consistent with our previous report.( , ) Interestingly, besides a robust induction of Cyp7a1, a significant induction of Cyp8b1 was also observed in adjacent nontumor tissue of LXRαKI and VP‐LXRα mice (Fig. 4B, bottom panel). Analysis of the TCGA data sets by GEPIA showed a significant up‐regulation of CYP7A1, down‐regulation of CYP8B1, and a trend of down‐regulation in CYP7B1 in human HCC (Fig. 4C), suggesting that the dysregulation of bile acid metabolism was conserved in patients with HCC.

FIG. 4

Chronic activation of LXRα accumulates pro‐HCC bile acid species and oxysterols in chemical‐induced HCC. (A) Heatmap of gene expressions in the bile acid metabolism pathway. Each column represents individual mice. (B) The tumor (top) and nontumor (bottom) tissue mRNA expression of Cyp7a1, Cyp7b1, and Cyp8b1 was measured by real‐time PCR. (C) Analysis of CYP7A1, CYP7B1, and CYP8B1 gene expression from the TCGA HCC data set (normal control n = 50; HCC n = 369). (D,E) Relative levels of bile acids in the serum (D) and liver tissues (E) of tumor‐bearing WT, LXRαKI, and VP‐LXRα mice. (F,G) The liver concentrations of cholesterol (F) and oxysterols (G) (n = 3‐6 for each group). *P < 0.05; **P < 0.01. Abbreviation: a‐MCA, a‐muricholic acid; BA, bile acid; b‐MCA, b‐muricholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; G‐CA, glycocholic acid; HC, Hydroxycholesterol; T‐murideoxycholic acid; T‐aMCA, tauro‐a‐muricholic acid; T‐b‐MCA, tauro‐b‐muricholic acid; T‐CA, taurocholic acid; T‐CDCA, taurochenodeoxycholic acid; T‐HDCA, taurohyodeoxycholic acid; T‐LCA, taurolithocholic acid; T‐MDCA, T‐murideoxycholic acid; and UDCA, ursodeoxycholic acid.

When bile acid levels in the serum (Fig. 4D) and nontumor liver tissues (Fig. 4E) were measured, we found increased concentrations of multiple species of unconjugated and conjugated bile acids in LXRαKI and VP‐LXRα mice, consistent with the induction of the bile acid synthesis rate–limiting enzyme Cyp7a1, as well as the suppression of a battery of genes involved in the bile acid metabolism, such as farnesoid X receptor (FXR, encoded by gene NR1H4) and small heterodimer partner (encoded by gene NR0B2), which are transcriptional factors mediating the negative‐feedback regulation of bile acid synthesis (Fig. 4A) Among the bile acid species whose levels were elevated in LXRαKI and VP‐LXRα mice, ω‐muricholic acid (ω‐MCA), hyodeoxycholic acid (HDCA), taurodeoxycholic acid (T‐DCA), and tauroursodeoxycholic acid (T‐UDCA) are secondary bile acids known to promote HCC.( , ) The accumulation of FXR antagonist bile acids, including T‐α‐MCA (Fig. 4D,E),( ) may have also contributed to the inhibition of FXR signaling and thereby promoting the progression of HCC.

Knowing oxysterols can be accumulated as a result of LXR activation and Cyp7b1 suppression,( ) and oxysterols can promote the recruitment of tumor‐promoting myeloid cells and inhibition of DCs,( ) we went on to measure the concentrations of cholesterol and oxysterols, including 20‐HC, 22(R)‐HC, 22(S)‐HC, 24(R/S)‐HC, 25‐HC, 27‐HC, and 7α/β‐HC in adjacent normal liver tissues of tumor‐bearing mice. No significant change was observed in liver cholesterol level among the three genotypes (Fig. 4F). Among the oxysterol species, the hepatic levels of 27‐HC, a chemotactic factor for tumor‐promoting myeloid cells,( ) were increased in both LXRαKI and VP‐LXRα mice (Fig. 4G). Interestingly, the hepatic concentrations of several oxysterols, including 22R‐HC and 7α/β‐HC, were decreased in both genotypes (Fig. 4G).

Chemical‐Induced HCC in LXRα‐Activated Mice Is Accompanied by Up‐Regulation of Mo‐MDSCs and Down‐Regulation of Cytotoxic T Cells and Dendritic Cells

Because activation of the IL‐6/JAK/STAT3 signaling pathway and the complement system are closely related to the recruitment of MDSCs and escape of immune surveillance,( ) and the hepatic accumulation of 27‐HC can function as a chemotactic factor to recruit tumor‐promoting myeloid cells,( ) we went on to determine whether activation of LXRα affected tumor infiltration of immune cells. Compared with WT mice, LXRαKI and VP‐LXRα mice showed increased tumor infiltration of CD45+ cells as shown by immunohistochemistry (Supporting Fig. S3A). We then used flow cytometry to profile the intratumor immune cells with the gating strategies outlined in Supporting Fig. S3B. Our flow cytometry results revealed that both LXRαKI and VP‐LXRα − mice showed (1) a reduced number of cytotoxic CD8+ T cells, but no change in CD4+ T cells (Fig. 5A); (2) a decreased number of total DCs (Fig. 5B) and CD103+DCs (Fig. 5C) who transport intact antigens to the lymph nodes and prime tumor‐specific CD8+ T cells,( , ) but no decrease in the total number of macrophages (Fig. 5B) or changes in the number of CD206+macrophages (also known as alternatively activated macrophage, or M2 macrophages)( ) (Fig. 5D); and (3) an induction of the tumor‐promoting Mo‐MDSCs, but not the G‐MDSCs (Fig. 5E).

FIG. 5

Chemical‐induced HCC in LXRα‐activated mice is accompanied by up‐regulation of Mo‐MDSCs and down‐regulation of cytotoxic T cells and DCs. (A‐G) Flow cytometry analysis of tumor tissues derived from WT, LXRαKI, and VP‐LXRα mice. Shown are representative density plots (left) and quantification (right) of CD45+CD8+CD4− T cells and CD45+CD8−CD4+ T cells in tumor tissues (A); representative density plots (left) and quantification (right) of CD45+ Gr‐1lowMHCIIhighCD11b+F4/80− myeloid cells (DCs) and CD45+ Gr‐1lowMHCIIhighCD11b+F4/80+ myeloid cells (macrophages) (B); quantification of CD103+ DCs (C); quantification of CD206+ macrophages (D); representative density plots (left) and quantification (right) of CD45+CD11b+Gr‐1+ myeloid cells (total MDSCs), CD45+CD11b+Gr‐1high myeloid cells (G‐MDSCs), and CD45+CD11b+Gr‐1int myeloid cells (Mo‐MDSCs) (E); quantification of CXCR2+ Mo‐MDSCs in tumors (top) and nontumor tissues (bottom) (F); and quantification of CXCR2+ GMPs (CD45+Lin−c‐KithighSca1−FcγRII/IIIhighCD34+) in the bone marrows (G). (H) Proposed role of LXRα in promoting oxysterol accumulation and innate immune suppression. Results are presented as mean ± SEM (n = 3 for each group). *P < 0.05; **P < 0.01. Abbreviation: FITC, fluorescein isothiocyanate; G‐MDSC, granulocyte‐like myeloid derived suppressor cells; and Mo‐MDSC, monocytic myeloid‐derived suppressor cells.

Next, we wanted to determine the mechanism by which activation of LXRα promotes Mo‐MDSC recruitment. It has been reported that the oxysterol–CXC chemokine receptor 2 (CXCR2) axis plays a key role in the recruitment of tumor‐promoting CD11b+Gr‐1+ myeloid cells. Specifically, tumor‐derived oxysterols, such as 27‐HC, function as chemotactic factors for the migration of CXCR2‐expressing CD11b+Gr‐1+ myeloid cells.( ) Our flow cytometry analysis revealed an increased number of CXCR2+ Mo‐MDSCs in the liver tumor tissues, but not in adjacent nontumor tissues (Fig. 5F), consistent with the accumulation of 27‐HC in LXRαKI and VP‐LXRα mice (Fig. 4G). Because Mo‐MDSCs are derived from CXCR2+ GMPs,( ) we measured the CXCR2+ GMPs isolated from the bone marrow of tumor‐bearing LXRαKI and VP‐LXRα mice by flow cytometry. As shown in Fig. 5G, increased numbers of CXCR2+ GMPs were observed in both LXRαKI and VP‐LXRα mice.

Taken together and as summarized in Fig. 5H, our current study has shown that chronic activation of LXRα promotes DEN/TCPOBOP‐induced HCC by (1) promoting innate immune suppression by increased recruitment of MDSCs and reduced number of cytotoxic CD8+ T cells and DCs by enhancing the oxysterols‐CXCR2 signaling pathway; and (2) accumulation of pro‐HCC secondary bile acids, including ω‐MCA, HDCA, T‐DCA, and T‐UDCA, as a result of Cyp7a1 activation.

Chronic Activation of LXRα Sensitizes Mice to MYC‐Driven HCC

We then used the LAP‐MYC transgenic mice to determine whether chronic activation of LXRα also sensitizes mice to oncogene‐driven liver carcinogenesis. The LAP‐MYC transgenic mice express the c‐MYC oncogene in hepatocytes under the control of the hepatocyte‐specific LAP gene promoter.( ) In this experiment, the LXRαKI allele was bred into the LAP‐MYC transgenic background as outlined in Fig. 6A, and the liver cancer phenotype in the resultant LAP‐MYC/LXRαKI mice was compared with that of the LAP‐MYC transgenic mice. The LAP‐MYC and VP‐LXRαKI alleles were independently genotyped by PCR (data not shown). The tumor growth in vivo was monitored by ultrasound imaging starting from 5 weeks of age. At 14 weeks of age, LAP‐MYC/LXRαKI male mice exhibited severe abdominal distension indicative of extensive tumor burden, as shown by representative sonograph in Fig. 6B. Mice were sacrificed at 14 weeks of age, at which the liver to body weight ratio was higher in LAP‐MYC/LXRαKI mice (Fig. 6C). Increased liver tumorigenesis in LAP‐MYC/LXRαKI mice was confirmed by gross appearance of the liver (Fig. 6D), as well as quantifications of tumor incidence (Fig. 6E) and tumor multiplicity (Fig. 6F).

FIG. 6

Chronic activation of LXRα sensitizes mice to MYC‐driven HCC. (A) Schematic representation of the crossbreeding between LAP‐MYC transgenic mice and LXRαKI mice to generate the LAP‐MYC/LXRαKI mice. (B) Representative sonographs at indicated time points. Dotted circles indicate tumor nodules. Arrowheads indicate acoustic haloes surrounding the large tumor nodules. (C) Liver to body weight ratio. (D) Representative gross appearance of tumor‐bearing livers of mice at 14 weeks of age. (E,F) Liver tumor incidence (E) and multiplicity (F) were calculated. (G) GSEA for HALMARK_IL6_JAK_STAT3_SIGNALING (left) and HALMARK_COMPLEMENT (right). (H) Heatmap of gene expression of Cyp7a1, Cyp7b1, and Cyp8b1. (I) GSEA for HALMARK_BILE_ACID_METABOLISM. Results are presented as mean ± SEM (n = 3‐5 for each group). *P < 0.05; **P < 0.01, compared with the LAP‐MYC group.

RNA‐seq analysis on liver tumor tissues revealed that compared with the LAP‐MYC mice, LAP‐MYC/LXRαKI mice had up‐regulation of the IL‐6‐JAK‐STAT3 and complement pathways (Fig. 6G), consistent with the DEN/TOPOBOP model. Interestingly, the intratumor expression of Cyp7a1 and Cyp7b1/Cyp8b1 was suppressed and induced, respectively (Fig. 6H), a pattern opposite to the chemical model. GSEA analysis showed the bile acid metabolism pathway was up‐regulated (Fig. 6I), which was also opposite to the chemical model.

Discussion

The pathophysiological function of LXRα and LXRβ in the development of HCC remains controversial. A better understanding of the role of LXRs in liver carcinogenesis will help to develop HCC therapeutics that target LXRs. Activation of LXRβ was reported to inhibit cancers by reducing MDSC recruitment in multiple murine cancer models and patients.( ) Based on the widely held presumption that LXRα and LXRβ share similar functions, we initially speculated that chronic activation of LXRα may inhibit hepatocarcinogenesis. To our surprise, we found that the expression of LXRα, but not LXRβ was elevated in patients with HCC compared to patients with CC. Further studies found that the expression of LXRα in the development of HCC was dynamic, and increased expression of LXRα was found in advanced HCC compared with pre‐neoplastic hyperplastic livers. In our preclinical models, chronic activation of LXRα systematically or hepatocyte specifically was sufficient to sensitize mice to HCC induced by DEN/TCPOBOP or the c‐MYC oncogene.

LXRs have been proposed to be therapeutic targets for several cancer types, including HCC.( , , ) We reason that the discrepancies between our results and previous reports are likely due to the isoform specific effect of LXR on carcinogenesis. Most of the LXR ligands that were used to show the anticancer effects of LXR, such as RGX104, T0901317 and GW3965, can activate both the α and β isoforms. In the reported HCC studies, the authors either did not design experiments that differentiate the effect of the two LXR isoforms,( , ) or the isoform effect was shown on the regulation of certain genes, but not the HCC growth.( ) In the non‐HCC cancer studies, the authors used either RGX104 in LXRα null mice to conclude the effect of LXRβ activation,( ) or they used melanoma cells that predominantly express LXRβ.( ) In the current study and using our genetic models that bear the exclusive activation of LXRα, we showed that chronic activation of LXRα sensitized mice to hepatocarcinogenesis. The expression of LXRβ was not affected by the transgenic expression of VP‐LXRα, suggesting that the phenotype was not due to dysregulation of LXRβ in our LXRα‐activated models. Future studies are necessary to determine whether ligand‐dependent activation of LXRα will have the same sensitizing effect as the genetic activation. For example, we could conduct studies on LXRβ null mice challenged with the TCPOBOP/DEN model in the presence or absence of RGX‐104 treatment, or we could introduce the Myc transgene into the LXRβ knockout background before treating them with RGX‐104. The mechanisms for the isoform specific effect of LXRs on HCC or other cancer types remain to be understood.

LXRs were previously shown to promote cholesterol catabolism to form bile acids in mice by inducing Cyp7a1 and suggested to promote the accumulation of oxysterols by suppressing the expression of Cyp7b1.( ) In our current models, we found the expression of Cyp7a1 was indeed induced, which may explain the accumulation of bile acids, including several secondary bile acid species known to be pro‐HCC. We also found the expression of Cyp7b1 was suppressed by chronic activation of LXRα when mice were challenged with the DEN/TCPOBOP model. The suppression of Cyp7b1 may have explained the accumulation of oxysterols, including 27‐HC, in DEN/TCPOBOP‐treated LXRαKI and VP‐LXRα mice.

The immune cell profiling results in our LXRα activation models were intriguing. Activation of LXRβ was reported to reduce MDSC recruitment in multiple cancer types, which was reasoned to be responsible for the cancer inhibitory activity of LXRβ.( ) However, we showed that activation of LXRα was associated with a higher abundance of Mo‐MDSCs and reduced numbers of cytotoxic T cells and DCs in tumors of LXRαKI and VP‐LXRα mice. Further analysis revealed a higher CXCR2 gene expression and a higher number of CXCR2+ Mo‐MDSCs in LXRαKI and VP‐LXRα tumors. The increased recruitment of Mo‐MDSCs to LXRα‐activated liver tumors may be explained by the accumulation oxysterols, such as 27‐HC, which can function as a chemotactic factor to recruit tumor‐promoting myeloid cells.( ) The up‐regulation of IL‐6/JAK/STAT3 signaling and complement pathways in tumor‐bearing LXRα‐activated livers may have also contributed to the induction of innate immune suppression.( , ) Meanwhile, the accumulation of oxysterols may have contributed to the activation of the IL‐6/JAK/STAT3 signaling pathway, as oxysterols, such as 27‐HC, have been reported to promote the accumulation of cellular reactive oxygen species and subsequent activation of the IL‐6/STAT3 signaling pathway.( ) Nevertheless, we observed an increased number of CXCR2+ GMPs, precursors of Mo‐MDSCs,( ) in the bone marrow of LXRαKI and VP‐LXRα mice, which helped to explain the elevation of Mo‐MDSCs but not G‐MDSCs. However, the mechanism linking LXRα activation and increased GMP surface expression of CXCR2 remains to be defined. Future studies are necessary to understand the discrepancy between the effects of LXRα activation and LXRβ activation on MDSC recruitment. In our models, LXRα was specifically and constitutively activated. In contrast, the reported LXRβ effect on MDSC recruitment relied on a pharmacological activation of LXRβ in LXRα knockout mice, in which the effect of LXRα ablation on the phenotypic exhibition cannot be excluded.( , )

Although chronic activation of LXRα sensitized mice to both the DEN/TOPOBOP and c‐MYC models of HCC, these two models exhibited overlapping yet distinct mechanistic insights. The liver tumor tissues from both models showed shared up‐regulation of the IL‐6‐JAK‐STAT3 and complement pathways. However, the pattern of Cyp7a1 and Cyp7b1/Cyp8b1 regulation was opposite between these two models. GSEA analysis showed that the bile acid metabolism pathway was suppressed and up‐regulated in the DEN/TOPOBOP and c‐MYC models, respectively. The differences may be explained by the fact that these are two HCC models of distinct mechanisms. Driven by a potent oncogene, the LAP‐MYC model is more aggressive than the DEN/TOPOBOP model in terms of tumor development and overall tumor burden. Our analysis of the clinical samples suggested that the effect of LXRα on the development of HCC could be stage‐specific.

Among the limitations, we recognized that our findings of increased IL‐6/STAT3 and complement pathways, and altered bile acid metabolism in the chemical model of HCC, are associations. Although the increased IL‐6/STAT3 and complement pathways were also observed in the c‐Myc model, the dependence of these pathway changes on the HCC phenotype remains to be experimentally verified. For example, it will be interesting to know whether IL‐6 deletion or neutralization will abolish the LXRα activation–responsive hepatocarcinogenesis in both the chemical and c‐Myc models.

In summary, we have uncovered a role of LXRα in the development of HCC. Chronic activation of LXRα promotes HCC, at least in part, by promoting innate immune suppression as a result of accumulation of oxysterols, as well as up‐regulation of the IL‐6/JAK/STAT3 signaling and complement pathways. Our results suggest that cautions need to be applied when LXR‐activating drugs are explored for their use in HCC treatment.

Fig S1

Click here for additional data file.

Fig S2

Click here for additional data file.

Fig S3

Click here for additional data file.