Estrogen Induces LCAT to Maintain Cholesterol Homeostasis and Suppress Hepatocellular Carcinoma Development

The poor prognosis of hepatocellular carcinoma (HCC) urges the development of new therapeutic targets and strategies. Apart from the well-investigated risk factors like hepatitis virus infection and cirrhosis (1), HCC also exhibits a notable male predominance (2), and the protective effects of estrogen against HCC have inspired interest in its clinical translation (3). Unfortunately, canonical mechanism studies do not lead to clinical application and estrogen administration is clinically impractical (4). Full dissection of estrogen-regulated biological processes may help understand hepatocarcinogenesis and develop new treatments.

Estrogen profoundly regulates metabolism and involves sex differences in cholesterol homeostasis (5, 6). Cholesterol homeostasis is maintained through orchestrating cholesterol biosynthesis, uptake, efflux, transport, storage, utilization, and excretion among the liver, blood, and other tissues (7). The liver is the central organ controlling cholesterol homeostasis and estrogen inhibits de novo cholesterol biosynthesis but promotes hepatic uptake mainly via LDLR and SCARB1 receptors (8). SREBP2 governs the regulation of cholesterol homeostasis through a negative feedback loop (9), and disturbed cholesterol homeostasis plays critical roles in the development of HCC (6, 10–12). Particularly, increased cholesterol biosynthesis and reduced cholesterol uptake, exportation, and excretion have been shown to promote HCC (6, 10). Moreover, a subtype of HCC characterized by disrupted cholesterol homeostasis has the worst prognosis (13). Interestingly, a high level of serum cholesterol inhibits mouse liver cancer (14). LDLR downregulation promotes HCC by promoting de novo cholesterol biosynthesis in compensating for the reduced uptake and upregulation of PCSK9 as an LDLR suppressor promotes sorafenib resistance (15, 16). Thus, how cholesterol homeostasis affects hepatocarcinogenesis remains to be fully dissected.

It has been well-documented that sex differences in cholesterol homeostasis significantly correlate with the risk of HCC. For instance, pre-menopause women exhibit lower low-density lipoprotein cholesterol (LDL-C) and higher high-density lipoprotein cholesterol (HDL-C) than that in age- and diet-matched men (17–19). Pre-menopause women also exhibit lower HCC risk than age-matched men and estrogen replacement therapy reduces liver cancer risk in postmenopause women (20, 21). However, these retrospective studies are limited and fail to address the causal–effect relationship. Whether or not sex differences in cholesterol homeostasis account for estrogen suppression on HCC remains unclarified.

To address this, we utilized an ovariectomized mouse liver cancer model for integrative trans-species transcriptomic analyses and identified lecithin cholesterol acyltransferase (LCAT) as a key estrogen-upregulated factor. LCAT is specifically expressed in hepatocytes and secreted into the plasma, and primarily catalyzes cholesterol esterification with phospholipid-derived acyl chains within lipoprotein in blood and promotes HDL maturation and the reverse cholesterol transport that are essential for plasma transport and hepatic uptake of cholesterol (22). We systemically elucidated that the LCAT-HDL axis inhibited cholesterol biosynthesis by impairing SREBF2 maturation and mediated estrogen suppression on HCC. LCAT and HDL-C served as prognostic biomarkers. LCAT and HDL presented potently therapeutic efficacy in mouse liver cancer models. Our study unveils novel mechanisms underlying sex disparities in HCC, providing promising biomarkers and therapeutic strategies against HCC.

All animal experiments were approved by the Animal Care and Use Committee of the College of Life Sciences, Wuhan University (No.16010A), and we provided care in accordance with the criteria outlined in the Guide for the Care and Use of Laboratory Animals. All mice are housed in an SPF environment. To validate the therapeutic efficacy of HDL and LCAT, a xenograft nude mouse liver cancer model was established as previously described (23).

The primary liver cancer model was induced via hydrodynamic tail vein injection of transposon-based vectors overexpressing NRASV12 and myr-AKT, and a transient expression vector expressing transposase in male and female mice (C57BL/6, IMSR_JAX:000664) as previously described (23). For female mice, Sham or ovariectomized (OVX) surgery was performed 2 weeks before injection. To test the function of LCAT, a transposon-based vector overexpressing LCAT or vector expressing Cas9 and sgRNA specific for LCAT was added for hydrodynamic injection (24). AKT1 overexpression mimics the activation of the mTOR pathway observed in >50% of human HCC cases (25). Based on our dosage, mice typically developed liver cancer 6 to 8 weeks postinjection (≥90% observed in 8 weeks). PCR confirmed the integration of NRASV12 and AKT into the liver genome, ruling out incorrect hydrodynamic injection as a cause of failure in liver cancer development.

For the orthotopic mouse liver cancer model, we injected a 10-μL mixture containing an equal volume of H22 cell suspension (7.5 × 10⁷ cells/mL) and matrix gel into the spleens of BALB/C (IMSR_ORNL:BALB-CRL) mice. Mice were subjected to tail vein injection with vehicle, HDL, lenvatinib, or lenvatinib plus HDL 2 days postinjection every other day. Mice were sacrificed on day 14 for further biochemical and pathological analyses.

RNA sequencing was conducted by Wellanimal Gene Technology. The RNA sequencing data and detailed experimental information are deposited as GSE255753.

For mouse transcriptomic data, genes with Foldchange >1.5 and P value ≤ 0.05 were defined as differentially expressed genes (DEG). For liver-hepatocellular carcinoma–The Cancer Genome Atlas (LIHC-TCGA), we chose patients below the average menopause age of 52, and genes with Foldchange >1.3 and P value ≤ 0.05 were identified as DEGs. In the LIHC-TCGA dataset, we compared males to females using HCC tissues or nontumor (NT) tissues. DEGs between males and females in both HCC tissues and nontumor liver tissues (NT) tissues were merged as sex-biased genes in humans. Similarly, DEGs between Sham and OVX mice in liver cancer tissues and nontumor tissues were identified and merged as female-biased genes. The intersection of these two gene sets generated 316 genes in common as conserved female-biased genes for further analysis.

Liver cancer data from TCGA (LIHC-TCGA) were also used for analyses of LCAT and ESR1 expression and their correlations to the TNM stage, E-S grade, and survival rate. Liver expression profiles from GTEx and LIHC-TCGA were employed to calculate ESR1 and LCAT-expression relevance. GSE107170 and GSE76427 datasets were used to validate the transcriptional activities relevance between LCAT and cholesterol synthesis pathway. GSE22058, GSE9843, GSE19977, and GSE63898 were utilized for gene set enrichment analysis for comparison of high and low LCAT–expression groups.

Paired HCC and adjacent nontumor tissues from 35 patients (Supplementary Table S1) were analyzed for LCAT and ESR1 transcription levels. Clinical data and tumor samples from 254 patients (Supplementary Table S2) were utilized for validating prognostic significance and expression relevance. All clinical data and tumor samples were obtained from the Department of Hepatobiliary and Pancreatic Surgery of Zhongnan Hospital of Wuhan University (Wuhan, China). Written informed consent was obtained from each HCC patient, diagnosed clinically and histologically without prior chemotherapy or radiotherapy. Ethical approval was obtained from the Medical Ethics Committee of Zhongnan Hospital of Wuhan University (No.2018013). All procedures performed in this study were adhered to the principles of the Declaration of Helsinki.

Physical examination data of the nonHCC population was described previously (26). Out of 450 clinical HCC patients, 181 patients suffered from cardiovascular and metabolic diseases (hypertension, coronary artery disease, stroke, and chronic kidney disease), 121 patients had no HDL-c tests, and the clinical data of seven patients were incomplete. Ultimately, 141 patients (115 males and 26 females) were included. After 1:1 matching with healthy controls based on gender, age, body mass index, smoking, and alcohol consumption, a total of 282 participants were included in the analysis.

Huh7 (CVCL_0336), HCCLM9 (CVCL_A5CU), HEK293T (CVCL_0063), and H22 (CVCL_H613) cells were purchased from the China Center for Type Culture Collection in 2019. Short tandem repeat profiling was utilized to authenticate cell lines. Hoechst 33258 staining was used to detect Mycoplasma contamination. Huh7 cells originated from a Japanese male HCC case, and HCCLM9 originated from a Chinese male HCC patient. HEK293T, H22, Huh7, and HCCLM9 cells were passaged less than 40 generations. Mouse liver cancer cells H22 (BABL/C background) were cultured in RPMI-1640 medium whereas others were in DMEM. The cell culture medium was supplemented with 10% fetal bovine serum (Cegrogen) and 1% penicillin–streptomycin solution. All cells were maintained in a 37°C incubator with 5% CO₂.

Estradiol (10 µmol/L), purified LCAT protein (30 μg/mL), and HDL (20 μg/mL) were added to treat cells. To disrupt the ESR1-binding site in the LCAT gene promoter of Huh7 cells, two sgRNAs were designed using CRISPOR (http://crispor.tefor.net/), and the CRISPR/Cas9 lentivirus system was employed as previously described (27). Short hairpin RNA (shRNA) sequences to knock down LCAT, LDLR, SCARB1, and ESR1 were obtained from Sigma-Aldrich (https://www.sigmaaldrich.cn/). Sequences of shRNAs and sgRNAs were listed in Supplementary Table S3.

Western blotting was performed as previously described (24). Antibodies used in this study were listed in Supplementary Table S4.

The cell proliferation was measured following instructions provided by the manufacturer (CCK-8 reagent kit; Vazyme Biotech Co., Ltd.). For the colony formation assay, the same amount (400–700) of cells was seeded for 7-day incubation. Then, fixation and staining were performed for quantification by ImageJ.

Cell cycle analysis was performed as previously reported (27). Briefly, Huh7 cells (1 × 10⁶) were seeded, synchronized to G1 phase by serum deprivation, recultured in serum-contained medium, digested and fixed with ice-cold 75% ethanol, and stained with DAPI (2 μg/mL in PBS) for 60 minutes. Flow cytometry was performed and analyzed with FlowJo.

Quantitative RT-PCR was performed as previously described (27). The primer sets used for RT-PCR are listed in the Supplementary Table S5.

The His-tagged protein was purified as previously reported (24). Briefly, cell lysates from 6×His-tagged LCAT-overexpressing HEK293T cells were loaded into the nickel-immobilized resin. After wash, the His-tagged protein was eluted with a buffer containing a high concentration of imidazole. Imidazole was removed via repetitive ultrafiltration and wash. The purified His-tagged LCAT protein was analyzed via Coomassie Brilliant Blue staining and Western blotting.

Centrifuge porcine serum at 3,000 g for 10 minutes to remove any visible precipitates. Ultracentrifuge the density-adjusted serum (ρ = 1.063 g/mL) at 120,000 × g for 18 hours at 4°C, and carefully collect the HDL-containing fractions from the top gradient. Purified HDL was obtained after the removal of extra ions via multiple ultrafiltration and wash.

Huh7 cells stably overexpressing 3× Flag-tagged ESR1 were used for chromatin immunoprecipitation (ChIP) and PCR as previously described (28). Primer sets used for PCR were listed in Supplementary Table S5.

Free nitrobenzoxadiazole (NBD)-labeled cholesterol (Biofount) was added to the Huh7 cell culture for 4 hours. Collect and ultrafilter the culture medium to obtain lipoprotein-derived NBD-labeled cholesterol. Huh7 cells were then exposed to the ultrafiltered medium for 30 minutes and harvested for extraction of lipids. The NBD fluorescence intensity of the extracted lipids was measured, which indicated the lipoprotein-derived cholesterol uptake rate of the treated Huh7 cells.

We used Filipin staining to visualize cholesterol content in sections. Briefly, tissue sections were stained with Filipin (100 µg/mL in PBS containing 1% FBS) for 1 hour. After wash, tissue sections and cell slides were mounted with a glycerol-based medium. Images of the stained samples were captured and the fluorescence intensity and distribution of Filipin-stained cholesterol were analyzed using ImageJ.

Huh7 cells were cultured in phenol red-free DMEM for 24 hours, followed by medium collection and concentration via ultrafiltration. HDL-C and LDL-C levels as well as other metabolites were then measured using Jining Shiye kits (HDL-C: JN24782; LDL-C: JN24783; mevalonate: JN837240; HMG-CoA: JN18286; total cholesterol (TC): JN24807; β-hydroxybutyric acid: JN823913; total bile acid: JN1220W; Malonyl-CoA: JN20393). Gas chromatography was employed for fatty acid measurements following established protocols (29).

Liver sections fixed with formaldehyde and embedded with paraffin were used for hematoxylin and eosin staining as well as immunohistochemistry for Ki67 and HMGCR. Histopathological analyses were performed as previously described (23). Detailed information on used antibodies is in Supplementary Table S4.

All data were presented as mean ± SD and analyzed using SPSS Statistics. Student t test assessed differences between two groups for normally distributed data whereas one-way ANOVA with Bonferroni post hoc test (homogeneous variance) or Tamhane T2 post hoc test (heteroscedastic data) was used to compare multiple groups. Nonparametric tests were applied for nonGaussian distributed groups. Statistical methods and corresponding P values for each figure are provided in figure legends.

The transcriptomic data generated in this study are publicly available in Gene Expression Omnibus at GSE255753. The transcriptomic data analyzed in this study were obtained from Gene Expression Omnibus at GSE76427, GSE107170, GSE22058, GSE9843, GSE19977, and GSE63898, as well as from the liver-hepatocellular carcinoma subset in TCGA database (http://lifeome.net/database/hccdb/download.html) and the liver subset in GTEx database (https://www.gtexportal.org/home/aboutAdultGtex). All other raw data generated in this study are available upon request from the corresponding author.

We aimed to fully dissect estrogen-regulated biological processes for potential clinical translation. In the LIHC-TCGA dataset, DEGs between males and females in HCC tissues and NT tissues shared only 43 common genes (Supplementary Fig. S1A and Supplementary Table S6). Interestingly, ESR1—a well-recognized dominant estrogen receptor and factor mediating estrogen’s suppression of HCC—was differentially expressed in NT between females and males but not in HCC (Supplementary Fig. S1B). As expected, ESR1 was downregulated in HCC, negatively associated with HCC severity, served as a favorable prognosis factor, and suppressed HCC in vitro (Supplementary Fig. S1C–L). These observations suggest that DEGs in normal livers driving hepatocarcinogenesis may be lost in HCC and vice versa. Thus, new strategies are needed to identify the estrogen-regulated biological processes that truly drive hepatocarcinogenesis.

Liver cancer was induced in OVX mice to study estrogen deficiency in HCC. OVX mice showed exacerbated liver cancer symptoms, with paler and stiffer livers, increased liver weight and liver weight to body weight ratio, and tumor numbers (Fig. 1A and B). OVX mice also exhibited more pseudolobules formation, severe inflammatory infiltration, and cell vacuolation than Sham mice (Fig. 1C). Evaluated serum ALT and AST levels indicated more severe liver injury in OVX mice (Fig. 1D). Reduced serum HDL-C but comparable LDL-C was observed in OVX mice (Fig. 1D). TC but not triglyceride content of liver cancer tissues was higher in OVX mice than that in Sham mice (Fig. 1E). Thus, OVX mouse liver cancer model nicely reproduces the effect of estrogen deficiency on HCC and may serve as a tool to dissect estrogen-regulated biological processes in hepatocarcinogenesis.

A trans-species transcriptomic analysis was further performed. DEGs between males and females in HCC tissues and NT tissues (LIHC-TCGA dataset) were identified and merged as sex-biased genes (Fig. 1F). Similarly, DEGs between Sham and OVX mice in liver cancer tissues and nontumor tissues were identified and combined as female-biased genes (Fig. 1F; Supplementary Fig. S2A–D). The intersection of two gene sets yields 316 common genes (Fig. 1F). Consistent with previous reports, the common genes upregulated in OVX mice exhibited enrichments in innate immune, NFκB signaling, and cell cycle processes (Fig. 1G; Supplementary Fig. S2E), supporting the rationality of our analytical model. Unexpectedly, the common genes downregulated in OVX mice exhibited significant enrichment in multiple metabolic processes (Fig. 1H). These genes were further filtered with prognostic value (P < 0.05) in LIHC-TCGA and six genes were identified (Supplementary Table S7). In this study, we focused on LCAT for its lowest P value and unknown function in HCC. In support, female mice expressed higher levels of LCAT than male mice in normal liver samples (Fig. 1I). Like ESR1, LCAT expression was higher in female NT but comparable between male and female HCC tissues (Fig. 1J). LCAT expression was downregulated in HCC tissues, negatively correlated to E-S grade and TNM stage, and served as a favorable prognosis factor (Fig. 1K–O; Supplementary Fig. S3A and S3B). These findings demonstrate that LCAT is a potential estrogen-upregulated factor and its expression negatively correlates to HCC manifestations. LCAT may play an important role in the pathogenesis of HCC.

In vitro, LCAT overexpression significantly inhibited Huh7 cell proliferation and colony formation possibly through inducing cell cycle arrest at G1 phase (Fig. 2A–D). Conversely, LCAT knockdown showed opposite phenotypes (Fig. 2E–H). Similarly, LCAT overexpression was inhibited, whereas its downregulation promoted HCCLM9 cell proliferation (Fig. 2I and J). Moreover, the addition of the recombinant LCAT (20 μg/mL) to the medium significantly suppressed Huh7 cell proliferation and colony formation and offset the promotive effect of LCAT knockdown on Huh7 cells (Fig. 2K and L).

The in vivo function of LCAT was further tested in myr-AKT/NRASV12D-induced primary liver cancer model. LCAT overexpression decreased tumor burden and liver damage, and increased serum HDL-C levels but not LDL-C in male mice (Fig. 2M–Q). Consistently, liver cancer tissues from LCAT-overexpressing mice showed decreased TC levels but unchanged triglyceride levels (Fig. 2R). Taken together, LCAT plays an inhibitory role in HCC in vitro and in vivo.

LCAT downregulation in OVX mice suggests that estrogen and downstream signaling may upregulate LCAT expression. In vitro, estradiol (E2) treatment significantly elevated LCAT expression in Huh7 cells (Fig. 3A). Transcription factor prediction and gene expression correlation analyses predicted that ESR1 might regulate LCAT expression (Fig. 3B and C). Modulation of ESR1 expression altered LCAT expression accordingly (Fig. 3D and E). ChIP-PCR verified ESR1 binding to one of four putative ESR1-binding sites in the promoter region of the LCAT gene and estradiol enhances the binding (Fig. 3F and G). Using the CRISPR/Cas9 technique, the ESR1-binding site was disrupted in Huh7 cells (termed as LCAT-ΔE cells; Supplementary Fig. S4A). LCAT-ΔE cells exhibited decreased LCAT expression, and increased cell proliferation, colony formation, and G1/S transition (Fig. 3H–L; Supplementary Fig. S4B and S4C). Although estradiol boosted ESR1 expression in both Ctrl and LCAT-ΔE cells, it failed to stimulate LCAT expression, suppress proliferation, impair colony formation, or arrest cells at G1 phase in LCAT-ΔE cells (Fig. 3H and L; Supplementary Fig. S4B and S4C). Likewise, ESR1 overexpression phenocopied the estradiol effects in Ctrl and LCAT-ΔE cells (Fig. 3M–Q; Supplementary Fig. S4D and S4E). Notably, a significant positive correlation between LCAT and ESR1 gene expression was observed in HCC samples (Fig. 3R). These results illustrate that LCAT mediates the suppressive effect of estrogen-ESR1 signaling on HCC.

LCAT catalyzes cholesterol esterification and promotes HDL maturation. Indeed, LCAT overexpression increased whereas LCAT knockdown decreased HDL-C contents in the cell culture medium without altering LDL-C levels (Fig. 4A and B). Introducing human HDL (20 μg/mL) to the medium not only mimicked the effect of LCAT to suppress cell proliferation and colony formation but also counteracted the promotive effects of LCAT knockdown on Huh7 cells (Fig. 4C and D; Supplementary Fig. S4F). Cholesterol is dynamically exchanged between HDL and LDL, which are absorbed mainly by hepatocytes through LDLR and SCARB1 receptors. Simultaneous knockdown of LDLR and SCARB1 (DKD) promoted cell proliferation and colony formation and abrogated the inhibitory effects of HDL on Huh7 cells (Fig. 4E and F; Supplementary Fig. S4G). Similarly, DKD abolished the suppressive effect of LCAT overexpression (Fig. 4G and H; Supplementary Fig. S4H). Furthermore, LCAT overexpression increased the uptake of lipoprotein-derived cholesterol that was nullified by DKD (Fig. 4I). Notably, HDL treatment led to a comparable effect with lovastatin (5 µmol/L) treatment and their combination did not show any synergistic effect (Fig. 4J and K; Supplementary Fig. S4I). These results suggest that LCAT suppresses HCC by promoting HDL maturation and hepatic uptake of HDL-C through LDLR and SCARB1 pathways.

To assess the clinical significance of our findings, we obtained HDL-C and LDL-C data from nonHCC individuals (health examination) and HCC patients. NonHCC individuals exhibited sex disparities in levels of HDL-C but not LDL-C (Supplementary Fig. S4J). Reduced HDL-C levels in HCC patients were observed in both genders (Fig. 4L and M) and female HCC patients maintained higher levels of HDL-C than male HCC patients (Fig. 4N). Moreover, HDL-C was the sole blood lipid index that was positively correlated to patient’s overall survival rate (Fig. 4O). Furthermore, in HCC patients, LCAT–expression levels in tumor tissues positively correlated with serum HDL-C levels (Fig. 4P).

Considering that HDL and lovastatin exhibit comparable effects and no addictive effects on HCC (Fig. 4J and K), LCAT and HDL may suppress HCC by inhibiting cholesterol biosynthesis. To test this, gene set variation analysis was performed (Supplementary Fig. S5A and S5B). Besides suppressing procancer pathways, LCAT demonstrated tight associations with metabolic pathways, including amino acid metabolism, cholesterol metabolism, fatty acid degradation, β-oxidation, and bile acid synthesis (Supplementary Fig. S5B). Indeed, reduced cholesterol biosynthesis intermediates (HMG-CoA, mevalonate) and TC, as well as increased total bile acids, malonyl-CoA, β-hydroxybutyrate, and fatty acid oxidase expression were validated in LCAT-overexpressing Huh7 cells and mouse liver cancer tissues (Fig. 2R; Supplementary Fig. S5C–H and S5J–N). Notably, major fatty acid types remained unaffected (Supplementary Fig. S5I–O). In support, gene set enrichment analysis consistently enriched cell cycle and cholesterol biosynthesis pathways in the low LCAT–expression group across multiple HCC datasets (Supplementary Fig. S5P). These observations suggest that LCAT may suppress HCC by inhibiting cholesterol biosynthesis.

Taken together, these results suggest that LCAT suppresses HCC by promoting HDL maturation and hepatic uptake of HDL-C through LDLR and SCARB1 pathways to inhibit cholesterol biosynthesis.

We further explored how LCAT-enhanced HDL-C maturation and uptake inhibited cholesterol biosynthesis. LCAT overexpression downregulated the expression of multiple cholesterol biosynthesis genes including Hmgcr, Hmgcs1, and Sqle, Srebf2 but not Srebf1 in mouse liver tumors (Fig. 5A and B). Both the precursor SREBP2 (pSREBP2) and nuclear SREBP2 (nSREBP2) were decreased in LCAT-overexpressing mouse liver tumors (Fig. 5C). In contrast, LCAT knockdown increased HMG-CoA and mevalonate products and caused downregulation of these genes except SREBF1 in Huh7 cells (Fig. 5D and E). Both the precursor SREBP2 (pSREBP2) and nuclear SREBP2 (nSREBP2) were increased in LCAT knockdown Huh7 cells (Fig. 5F). Low cellular cholesterol activates SREBP2 via cleavage by S1P and S2P. Indeed, S1P silencing nullified LCAT-induced inhibition on SREBP2 activation (Fig. 5G). Moreover, SREBF2 overexpression not only significantly promoted the expression of cholesterol biosynthesis genes, cell proliferation, colony formation, and G1/S transition but also antagonized the suppressive effect of LCAT treatment (Fig. 5H–K; Supplementary Fig. S4K and S4L). Lovastatin treatment verified that the effect of SREBF2 originated from the enhanced cholesterol synthesis (Fig. 5L). In support, the reverse correlation of SREBF2 to the level of tumoral LCAT or serum HDL-C was observed in our HCC patients (Fig. 5M). These results strongly demonstrate that LCAT exerts suppression on HCC by impairing cholesterol biosynthesis via inhibiting SREBF2 expression and its protein maturation.

Our findings were also supported by other datasets. Cholesterol biosynthesis was negatively correlated to LCAT expression in the HCC transcriptome (Supplementary Fig. S6A and S6B). Moreover, cholesterol biosynthesis was significantly enriched in sex-differentiated pathways in mouse livers (Supplementary Fig. S6C and S6D). Consistently, female mice livers exhibited lower levels of free cholesterol, and estrogen administration reduced the hepatic contents of cholesterol and squalene (Supplementary Fig. S6E and S6F).

Supportively, APOA1 as an LCAT coactivator was downregulated and negatively correlated to HCC E-S grade and TNM stage, served as a favorable prognosis factor, and suppressed Huh7 cell proliferation (Supplementary Fig. S7A–E). Additionally, PCSK9 as an LDLR repressor was upregulated, served as an unfavorable prognosis factor in HCC, and promoted Huh7 cell proliferation (Supplementary Fig. S7F–J).

To validate whether estrogen-upregulated LCAT links sex differences of cholesterol homeostasis to estrogen suppression on HCC in vivo, liver cancer was induced in female mice with LCAT knockdown (LCAT-KD) using CRISPR/Cas9 technique and hydrodynamic tail vein injection (Supplementary Fig. S3C). As expected, both OVX and LCAT-KD exacerbated liver cancer (Fig. 6A–D). Importantly, both OVX and LCAT-KD reduced serum HDL-C but increased free cholesterol in liver cancer tissues (Fig. 6D and E). Upregulation of cholesterol biosynthesis genes by OVX and LCAT-KD was verified and immunohistochemistry confirmed the upregulation of the HMGCR protein (Fig. 6F and G). These results demonstrate that LCAT-KD replicates the effect of OVX on HCC.

Liver cancer was also induced in OVX mice with LCAT overexpression by hydrodynamic tail vein injection (Supplementary Fig. S3D). LCAT overexpression counteracted the detrimental effects of ovariectomy without altering estradiol levels (Fig. 6H–K; Supplementary Fig. S3E). Similarly, LCAT overexpression nullified the effect of OVX in reducing HDL-C and increasing HMGCR expression and cholesterol content (Fig. 6L and M). These results collectively prove that LCAT connects sexual differences in cholesterol homeostasis and estrogen suppression on HCC.

To assess the therapeutic efficacy of LCAT and HDL in HCC, a xenograft HCC model was established using HCCLM9 cells in male nude mice. Intratumor administration of LCAT and HDL significantly delayed tumor growth, decreased HMGCR and Ki67 expression, and reduced free cholesterol content (Fig. 7A–C). Moreover, reduction of SREBP2 maturation and downregulation of cholesterol biosynthesis genes were confirmed (Fig. 7D and E). These results substantiated that LCAT and HDL suppress cholesterol biosynthesis and HCC proliferation.

Orthotopic mouse liver cancer model was further generated by injecting H22 cells into the spleen of BABL/C mice and treated with lenvatinib, HDL, or their combination. Lenvatinib is a first-line treatment for patients with unresectable HCC and drug combination is an effective way to improve treatment efficacy. As expected, lenvatinib treatment potently suppressed liver tumor growth and reduced liver damage (Fig. 7F–I). These mice showed a reduction in tumor numbers and invasion area as well as an improvement in liver function (Fig. 7F–I). Moreover, HDL alone had comparable efficacy to lenvatinib, and their combination significantly enhanced therapeutic outcomes. Mice treated with the combination therapy exhibited the lowest liver weight and fewest tumors across all groups (Fig. 7F and G). The majority of these tumors were located near the edge of livers in this group, which might explain why there was no detectable invasion area in liver sections derived from the middle part of the medial lobe for analysis (Fig. 7H). These results demonstrate that LCAT and HDL are potent agents against HCC. Normalization of cholesterol homeostasis by LCAT and HDL may be novel strategies for HCC treatment.

Numerous clinical studies have demonstrated a potential regulatory link between estrogen signaling and hepatic cholesterol homeostasis (30), the existence of gender differences in HDL-C levels, and the distinctive characteristics and pathological roles of cholesterol metabolism in HCC (15, 31, 32). However, no systematic study has connected these isolated phenomena and the causal–effect relationship has never been clarified. Through our research, we identified the estrogen-ESR1-LCAT-HDL axis as a crucial mechanism underlying sexual differences of HCC, which may yield novel biomarkers for the prognosis and treatment of HCC.

Despite the role of estrogen and ESR1 in HCC gender differences, we observed diminished sexual differences in ESR1 and LCAT expression in HCC tumors possibly due to extensive DNA methylation observed in HCC. In contrast, sex differences in HDL-C levels persist among HCC patients. We attribute this phenomenon primarily to the substantial retention of normal liver tissue in liver cancer patients, which are still capable of differentially expressing LCAT in response to estrogen. Although estrogen protects females from HCC mainly via regulating inflammation, hepatitis B virus infection, replication, and cell cycle (33), our study demonstrates that sex differences in cholesterol homeostasis may substantially contribute to estrogen suppression on HCC. Moreover, many reports suggest a close relationship between cholesterol reverse transport and hepatic cholesterol biosynthesis rate. For instance, ABCA1 transfers phospholipids to apolipoproteins to form HDL, whose absence leads to the increased expression of SREBP2 and HMGCR (34). LCAT deficiency increased hepatic HMGCR mRNA and nuclear SREBP2 protein (35). These findings emphasize serum HDL may exert function on hepatic cholesterol metabolism. Our study further provides evidence that LCAT mediates estrogen suppression on HCC by promoting HDL maturation and uptake and reducing de novo cholesterol biosynthesis.

Our findings also provide novel strategies for targeting cholesterol metabolism in HCC treatment. Targeting cholesterol metabolism has been suggested to be a good intervention strategy. However, inhibition of cholesterol biosynthesis by statins only achieves modest benefits (6, 36). Recombinant LCAT and mimetic peptides targeting LCAT and ApoAI have been tested for atherosclerotic cardiovascular diseases (22), necessitating further investigation for potential application in HCC treatment. Moreover, we noticed that either LDLR or SCARB1 downregulation alone was not enough to obligate the inhibitory effect of LCAT, suggesting that both HDL and LDL may be involved in this process. Recently, mimetic HDL and reconstituted LDL have been reported to serve as drug carriers (37, 38) and artificial adiposomes with apoprotein may mimic HDL and LDL (39, 40). Nanoscale and liver-specificity make these artificial particles particularly suitable for targeted therapies for liver cancer. Further investigation is warranted to determine whether these particles, loaded with esterified cholesterol and chemotherapy drugs, exert therapeutic effects on HCC. It may potentially pave an alternative way for the clinic application of lipoproteins in HCC treatment.

In summary, estrogen-upregulated LCAT causes sex differences in cholesterol homeostasis via increasing HDL maturation and uptake to inhibit de novo cholesterol biosynthesis, which mediates estrogen suppression on HCC. LCAT and HDL-C may serve as novel biomarkers for targeting cholesterol homeostasis as novel strategies for HCC treatment.

W. He reports grants from National Nature Science Foundation of China during the conduct of the study; in addition, W. He has a patent for The Application of High-Density Lipoprotein in the Preparation of Antihepatocellular Carcinoma Drugs pending and a patent for The Application of the LCAT Gene and Purified Protein in the Preparation of Drugs for the Prevention, Alleviation, and/or Treatment of Primary Hepatocellular Carcinoma pending. X. Zhang reports grants from the National Natural Science Foundation of China during the conduct of the study. Y. Wang reports grants from the National Nature Science Foundation of China during the conduct of the study. D. Zhao reports grants from the National Nature Science Foundation of China during the conduct of the study. F. Lei reports grants from the National Nature Science Foundation of China during the conduct of the study. Z. Huang reports a patent for The Application of High-Density Lipoprotein in the Preparation of Antihepatocellular Carcinoma Drugs pending and a patent for The Application of the LCAT Gene and Purified Protein in the Preparation of Drugs for the Prevention, Alleviation, and/or Treatment of Primary Hepatocellular Carcinoma pending. No disclosures were reported by the other authors.

W. He: Investigation, methodology, writing–original draft. M. Wang: Investigation. X. Zhang: Validation. Y. Wang: Validation. D. Zhao: Validation. W. Li: Data curation. F. Lei: Software. M. Peng: Resources. Z. Zhang: Resources. Y. Yuan: Resources. Z. Huang: Supervision, funding acquisition, writing–review and editing.

We thank Dr. Baoliang Song (College of Life Sciences of Wuhan University) for providing the SREBP2 antibody and helping us measure cholesterol metabolism, Yan Wang (College of Life Sciences of Wuhan University) for the kind gift of PCSK9 expression plasmids, Jian Huang for advice on estrogen study from College of Life Sciences of Wuhan University, and Zhigang She for critical reading from Remin Hospital of Wuhan University. This research was supported by the National Natural Science Foundation of China (grant nos. 32270643 and 91957109 to Z. Huang), the Hubei Science Foundation for Distinguished Young Scholars (2023AFA079 to M. Peng), the Interdisciplinary Innovative Talents Foundation from Renmin Hospital of Wuhan University (JCRCFZ-2022-025 to M. Peng), and Large-scale Instrument and Equipment Sharing Foundation of Wuhan University.