Background

Primary liver cancer is the sixth most commonly diagnosed cancer and the third-leading cause of cancer death, with the highest incidence in Asia.1 Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and accounts for approximately 75% of all cases.2 There are various treatments for HCC, including surgical resection, liver transplantation and local ablation, which are primarily used in the early stage, and systemic treatment is mostly used in advanced-stage patients.3 4 However, approximately 80% of patients have advanced HCC at the time of initial diagnosis, and the 5-year survival rate is less than 18% due to numerous factors, such as poor response to medication and susceptibility to recurrence.4–6 Immunotherapy, represented by immune checkpoint inhibitors (ICIs), such as nivolumab and pembrolizumab, has emerged as an alternative therapy for HCC, and these agents exert antitumor effects primarily by promoting T-cell initiation and activation.7 Clinical studies have shown that ICIs, such as nivolumab, significantly improve the overall survival (OS) of drug-resistant patients and have gradually become the preferred drug treatment option.8 9 However, the overall uneven therapeutic efficacy of ICIs and the obvious phenomenon of drug resistance have made the development of effective efficacy-predicting biomarkers a clinical problem that must be solved.

The immune system plays an important biological role in tumor progression.10 Impairment of the tumor immune system promotes tumor progression by generating a tumor-immunosuppressive microenvironment, which leads to immune escape.11 The tumor immune microenvironment (TIME) in HCC is primarily composed of tumor cells, immune cells and other components, and tumor-associated macrophages (TAMs) play important roles in the metastasis and drug resistance of HCC.12 TAMs primarily consist of M1 (induced by IFNγ alone or with Lipopolysaccharide) and M2 (induced by IL-4 and IL-13) macrophages, and M2 macrophages mediate the Th2 response and promote tumor progression in a direct and indirect manner, respectively.13–15 Therefore, targeting M2 macrophage recruitment and clearance may be a new tumor therapeutic strategy. Complex factors affect the polarization fate of TAMs, and further elucidation of the mechanism of TAM polarization has important clinical significance for HCC treatment.

Common fragile sites (CFSs) are normal chromosomal regions prone to forming gaps or breakages on partial inhibition of DNA synthesis, which correlate with chromosomal rearrangement and copy number variation.16 FRA3B and FRA16D are the two most classical fragile sites, most frequently expressed and best characterized in tumor cells, and both are located within fragile histidine triad and WW domain-containing oxidoreductase (WWOX) genes, respectively.17 Our previous study revealed that the WWOX rs9926344 polymorphism was associated with the recurrence of HCC.18 Previous studies showed that WWOX was a tumor suppressor that may be involved in a variety of biological processes, including cell proliferation, metastasis and metabolism.19 20 However, few studies focused on the biological role of CFS-related genes, especially classical genes, in HCC tumor immunity. By analyzing the transcriptome profiles of HCC patients with or without response to ICIs, we identified WWOX as an effective predictor of ICI response. Furthermore, we used in vitro experiments and a humanized mouse model of HCC and determined that the molecular mechanism of WWOX-mediated immunosensitization may originate from significant inhibition of the NME2/SCD5/oleic acid (OA) axis.

ResultsWWOX mediates the resistance of HCC to anti-PD-1 treatment

We first performed next-generation sequencing (NGS) on eight HCC biopsy samples from patients with different ICI responses (online supplemental table 1). Single-sample genomic enrichment analysis (ssGSEA) revealed differences in tumor immunoreactivity between the two groups (online supplemental figure 1A). Differential analysis revealed WWOX was the most upregulated CFS related gene in ICI response HCC patients (figure 1A). Gene Ontology and Kyoto Encyclopedia of Genomes (KEGG) analyses revealed that metabolism and immunity-related terms were significantly enriched (online supplemental figure 1B,C). WWOX was detected in the serum of patients receiving PD-1 antibody treatment using ELISA and was more abundant in the responder group (figure 1B, online supplemental table 2). Representative MRIs are shown in figure 1C. In addition, WWOX protein expression was examined in different PD-1-responsive HCC tissues, and the IHC results suggested that WWOX was significantly upregulated in immune-responsive HCC tissues (figure 1D). Patients were divided into two groups based on the serum levels of WWOX in patients treated with PD-1. K-M analysis showed that low levels of WWOX in serum were associated with poor OS (27.59 vs 43.08 months) and recurrence-free survival (RFS, 23.97 vs 39.70 months) (figure 1E). To further clarify whether WWOX was required to mediate PD-1 resistance in HCC, we constructed a WWOXlow humanized orthotopic HCC mouse model and compared PD-1 treatment responses. The results revealed that PD-1 treatment significantly reduced tumor growth in mice in the control but not the huh7-shWWOX group (figure 1F,G). Also, IHC results suggested that CD8+ T-cell infiltration was reduced in Huh7-shWWOX tumors (figure 1H). These results suggest that WWOX deficiency renders HCC resistant to PD-1 treatment.

Figure 1

WWOX mediates the resistance of HCC to anti-PD-1 treatment. (A) Heatmap of differentially expressed CFSs genes in HCC patients who were ICI responders (n=4) and non-responders (n=4). (B) ELISA of WWOX levels in the serum of ICI responder (n=28) and non-responder (n=29) HCC patients. (C) Representative abdominal MRIs of HCC patients who were ICI responders and non-responders. (D) Representative IHC staining images of WWOX in HCC tissues from the indicated patients. Scale bar, 100 µm. (E) Kaplan-Meier survival curves showing the differences in overall survival (left) and cumulative recurrence rates (right) between high-WWOX and low-WWOX level in our cohort. (F) (Top): Schematic diagram of the orthotopic HCC tumor constructs in humanized mice and the schedule of anti-PD-1 antibody treatment of WWOXlow tumors in humanized mice. (Bottom): Gross appearance of orthotopic HCC tumors in humanized mice in the indicated groups; (G) Weight and volume of orthotopic HCC tumors at the endpoint in the indicated groups. (H) (Left): Representative IF staining images of CD8 in HCC tumor tissues from the indicated group. Scale bar, 100 µm. (Right) Statistical analysis of the number of CD8+ T cells in the indicated groups. *p<0.05, **p<0.01, ***p<0.001; Student’s t-test. CFS, common fragile sites; HCC, hepatocellular carcinoma; ICI, immune checkpoint inhibitor; IF, immunofluorescent; TIME, tumor immune microenvironment; t-SNE, t-distributed stochastic neighbor embedding; WWOX, WW domain-containing oxidoreductase.

WWOX deficiency is associated with suppressive TIME

To further observe the relationship between WWOX alterations and the TIME, we evaluated eight HCC tissues using cytometry by time-of-flight (CyTOF) analysis. Eight HCC tissues were divided into two groups according to the protein level of WWOX (WWOXhigh and WWOXlow, online supplemental figure 1D). Immune cells were clustered into 32 cell clusters based on the expression of 41 cell markers (figure 2A). Among these cells, macrophage infiltration was increased in the WWOXlow subgroup, but no significant difference was observed in the infiltration of other immune cells (figure 2B,C). CD8+ T subgroup analysis showed increased infiltration of the C14 subgroup in the WWOXhigh subgroup (online supplemental figure 1E). We further compared differences in macrophage and CD8+ T-cell-related marker levels between the two groups. The results showed that M2 macrophage markers, such as CD68 and CD204, were elevated in the WWOXlow subgroup, and CD8+ T-cell-associated markers, such as PD-L1, Granzyme B and PD-1, were not different (figure 2D,E, online supplemental figure 1F). To further validate the above results of the effect of WWOX on the immune microenvironment, we performed multiple immunofluorescence staining of HCC tissues from eight cases treated with PD-1. The results revealed that WWOX was highly expressed in PD-1-responsive patients and was associated with high CD8 T cells and low macrophage infiltration (figure 2F). These results suggested that WWOX deficiency induced an immunosuppressive tumor microenvironment and attenuated the response to ICIs.

Figure 2

WWOX deficiency is associated with suppressive TIME. (A) (Left) t-SNE analysis of CyTOF data of immune cells in eight HCC tissues. (Right) Heatmap of the expression of 41 markers in all 32 subclusters. (B) t-SNE plots of immune cells in HCC tissues from the WWOX high/low subgroup. (C) Proportion of tumor-infiltrating immune cells in HCC tissues from the two groups. (D) (Left) t-SNE plots of CD68 expression levels in the indicated groups. (Right) CD68 expression levels in immune cells in both groups. (E) Expression of CD204, PD-L1, Granzyme B and PD-1 in CD45+ cells in the indicated groups. (F) (Left): The representative image of HCC tissue stained with WWOX (red), CD8 (green), CD68 (yellow). (Right): Statistical analysis of the number of CD8+ T cells, macrophages and WWOX expression in the indicated groups. *p<0.05, ***p<0.001; Student’s t-test. CyTOF, cytometry by time-of-flight; HCC, hepatocellular carcinoma; TIME, tumor immune microenvironment; t-SNE, t-distributed stochastic neighbor embedding; WWOX, WW domain-containing oxidoreductase.

WWOX deficiency induces macrophage recruitment and immunosuppressive macrophage polarization

We constructed lentiviral stably transfected cell lines based on WWOX protein and mRNA levels in HCC cells (figure 3A, online supplemental figure 2A–C). To further examine whether WWOX is valuable for macrophage polarization and recruitment, we constructed an HCC and macrophage co-culture system (figure 3B). Co-culture with WWOXlow HCC cells (HCCLM3-Vector and Huh7-shWWOX) upregulated CD163, MRC1, ARG-1, IL-10, and TGF-β mRNA levels and upregulated CD206 protein levels in macrophages (figure 3C,D). Chemotaxis assays showed that WWOXlow HCC cells promoted macrophage recruitment (figure 3E). Similar results were found in PLC/PRF/5 and MHCC97H cells (online supplemental figure 2D–F). To further elucidate whether macrophage M2 polarization mediates WWOX-induced PD-1 resistance, we used Clophosome to clear macrophages in mice. Remarkably, PD-1 treatment significantly reduced tumor growth in the huh7-shWWOX group of mice after macrophage clearance compared with the control group (figure 3F,G).

Figure 3

WWOX deficiency induces macrophage recruitment and immunosuppressive macrophage polarization. (A) Western blot analysis of WWOX levels in HCC cells. (B) Schematic diagram of HCC cells and macrophages (PMA-treated) co-cultured in vitro. (C) qPCR analysis of the mRNA levels of CD163, MRC1, ARG1, IL-10 and TGF-β in macrophages (PMA-treated) after co-culture with HCC cells. (D) Flow cytometry analysis of CD206 expression on the surface of macrophages (PMA-treated) after co-culture with HCC cells. (E) Chemotaxis assays were used to analyze the effect of Huh7-shWWOX and MHCCLM3-WWOX cell supernatants on macrophage (PMA-treated) migration. Scale bar, 50 µm. (F) (Left) Gross appearance of orthotopic HCC tumors with anti-PD-1 treatment in humanized mice in the indicated groups; (Right) Weight and volume of orthotopic HCC tumors at the endpoint in the indicated groups. (G): Representative IHC images of CD68 in HCC tumors of indicated groups. (H) (Top) Representative IHC images of WWOX, CD68 and CD206 in HCC tissues. Scale bar, 200 µm (left) and 50 µm (right). (Bottom) Kaplan-Meier survival curves for patients stratified according to WWOX/CD68 and WWOX/CD206 expression. *p<0.05, **p<0.01, ***p<0.001; Student’s t-test and log-rank test. HCC, hepatocellular carcinoma; IHC, immunohistochemistry; PMA, phorbol-12-myristate-13-acetate; qPCR, quantitative real-time PCR; WWOX, WW domain-containing oxidoreductase.

To further determine whether the effect of WWOX on macrophage polarization affected the clinical prognosis of HCC patients, we performed IHC staining of tissues from 176 HCC patients. IHC staining revealed that WWOX expression negatively correlated with CD68 and CD206 expression (figure 3H, online supplemental table 3). Prognostic analysis revealed that HCC patients with WWOXlowCD68high and WWOXlowCD206high tumors had the worst OS, and WWOXlowCD206high was an independent prognostic factor (figure 3H, online supplemental table 4). These results suggest that WWOX deficiency induces macrophage M2-like reprogramming and mediates ICIs resistance.

OA-derived tumor tissue from patients with WWOX deficiency promotes immunosuppressive macrophage polarization

WWOX is closely associated with tumor metabolism, and we analyzed metabolites in the supernatants of HCC cells from control and shWWOX groups using untargeted metabolomics to further elucidate the molecular mechanisms of the effects of WWOX on macrophages (figure 4A). The results of difference analysis suggested that OA was most significantly upregulated in the supernatants of HCC cells in the shWWOX group compared with the control group (figure 4B). ELISAs further confirmed that the OA concentration was elevated in the supernatants of the WWOXlow HCC cells compared with the control group (figure 4C, online supplemental figure 3B). Stimulation of PMA-treated THP-1 cells with OA significantly increased macrophage CD206 protein expression and M2-like macrophage marker mRNA levels (figure 4D). The transcriptomics results suggested that OA promoted M2 macrophage polarization via the mTOR signaling pathway (online supplemental figure 3A). Therefore, we hypothesized that WWOX affected macrophage polarization via OA.

Figure 4

NME2/SCD5 axis-mediated release of oleic acid from HCC cells promotes immunosuppressive macrophage polarization. (A) Untargeted metabolic analysis of the control and shWWOX groups of Huh7 cell supernatants. Groups 1 and 2 represent WWOX knockdown and control HCC cells, respectively. Classes I and II represent upregulated and downregulated metabolites. (B) Differential analysis of OA content in supernatants of WWOX knockdown and control HCC cells. (C) ELISA was used to analyze the OA concentration in the supernatants of the indicated HCC cells. (D) (Left) Flow cytometry analysis of CD206 expression on the surface of macrophages after OA treatment. (Right) qPCR analysis of the mRNA levels of M2 markers in macrophages after OA treatment. (E) Western blot and qPCR analyses of the protein and mRNA levels of SCD5 in the indicated HCC cells. (F) (Left) Flow cytometry analysis of CD206 levels in macrophages co-cultured with Huh7-shWWOX cells transfected with lentivirus-shSCD5. (Right) Flow cytometry analysis of CD206 levels in macrophages after co-culture with MHCCLM3-shWWOX cells transfected with lentivirus-SCD5. (G) The function of NME2 in suppressing SCD5 transcription was analyzed using dual luciferase assay in the SCD5 promoter region. (H) DNA pulldown assay of NME2 binding to the SCD5 promoter in the indicated HCC cells. (I) Western blot and qPCR analyses of the mRNA and protein levels of SCD5 after lentiviral NME2-mediated transfection of Huh7-shWWOX cells; (J) Western blot and qPCR analyses of the mRNA and protein levels of SCD5 after lentiviral shNME2-mediated transfection of HCCLM3-SWWOX cells. (K) A DNA pulldown assay was used to analyze the effect of WWOX on the binding affinity of NME2 to the SCD5 promoter in the indicated HCC cells. *p<0.05, **p<0.01, ***p<0.001; Student’s t-test or one-way ANOVA with a post hoc Least Significant Difference (LSD) test. ANOVA, analysis of variance; HCC, hepatocellular carcinoma; OA, oleic acid; WWOX, WW domain-containing oxidoreductase; qPCR, quantitative real-time PCR.

To examine the biological mechanisms of the effects of WWOX on OA, we detected key enzymes of OA metabolism, SCD1 and SCD5. Western blot assays and qPCR confirmed that WWOX deficiency upregulated SCD5 protein and mRNA levels without affecting SCD1 (figure 4E, online supplemental figure 3C–E). To determine whether upregulated SCD5 was associated with increased OA, we used lentiviral shSCD5-transfected Huh7-shWWOX cells and lentiviral SCD5-transfected HCCLM3-WWOX cells and found these cells significantly attenuated the promotion of WWOX deficiency-induced OA synthesis (online supplemental figure 3F–H). Therefore, we hypothesized that SCD5 was a key mediator of the effects of WWOX on OA synthesis. To further clarify whether the SCD5-OA axis mediated the effects of WWOX on macrophages, we performed rescue experiments. Notably, SCD5 knockdown reversed the WWOX deficiency-induced increase in CD206 protein expression in macrophages (figure 4F). These results suggest that WWOX deficiency promotes M2 macrophage polarization primarily via transcriptional activation of the SCD5-OA axis.

NME2 is a transcription factor for SCD5 that promotes immunosuppressive macrophage polarization

Because WWOX deficiency transcriptionally activates SCD5, we further screened for potential transcription factors related to SCD5. The direct binding of WWOX to transcription factors was reported in previous studies.23 Therefore, we detected SCD5 promoter-binding proteins and WWOX direct-binding proteins using liquid chromatography-tandem mass spectrometry (LC-MS/MS) and obtained four proteins using intersection analysis of the results with the transcription factor database (Cistrome, AnimalTFDB) (online supplemental figure 4A). Among these factors, overexpression of NME2 had the most significant effect on the SCD5 protein level (online supplemental figure 4B,C). Western blot assays showed that WWOX did not affect NME2 protein levels (online supplemental figure 4D). To determine whether NME2 was a transcription factor, we constructed a dual-luciferase reporter plasmid containing the SCD5 promoter region. Notably, NME2 significantly reduced SCD5 luciferase activity (figure 4G). A DNA pull-down assay using the SCD5-specific promoter confirmed that NME2 bound to the SCD5 promoter (figure 4H). These results indicate that NME2 binds to the SCD5 promoter region, which suggests that NME2 is a transcriptional repressor of SCD5.

To determine whether NME2 mediates WWOX deficiency-induced macrophage M2 polarization, we transfected Huh7-shWWOX cells with lentiviral NME2 and HCCLM3-WWOX cells with lentiviral shNME2 (online supplemental figure 4E). These cells significantly attenuated the promotion of SCD5 activation and WWOX deficiency-induced OA synthesis and reduced macrophage CD206 protein expression (figure 4I,J, online supplemental figure 4F,G). These findings suggest that the WWOX-NME2-SCD5 axis is involved in regulating M2 macrophage polarization.

To elucidate the intrinsic mechanism of the effects of WWOX on NME2 regulation of SCD5 transcription, we performed nuclear-cytoplasmic extraction experiments. The results revealed that WWOX did not affect the protein localization of NME2 (online supplemental figure 4H). Previous studies found that the binding of transcription factors to target gene promoters affected their transcriptional activity.24 We hypothesized that WWOX affected NME2 binding to the SCD5 promoter. By constructing an SCD5-specific promoter, a DNA pull-down assay showed that NME2 binding to the SCD5 promoter was reduced in Huh7-shWWOX cells, and overexpression of WWOX in HCCLM3 cells promoted NME2 binding to the SCD5 promoter (figure 4K). These results suggest that WWOX inhibits SCD5 transcription by promoting the binding of NME2 to the SCD5 promoter.

WWOX competes with KAT1 for binding to NME2 to inhibit the acetylation of NME2 at site 31 mediated by KAT1

To further investigate the specific mechanism of WWOX promotion of NME2 binding to the SCD5 promoter, we determined the direct interaction between WWOX and NME2 based on the results of LC-MS/MS. Immunoprecipitation and immunofluorescence analyses demonstrated the interaction and co-localization of WWOX with NME2 in HCC cells (figure 5A,B). We mapped the structural domains involved in the interaction between WWOX and NME2 and found that the WWOX (106-414) and NME2 (67-152) domains were necessary for their interaction (figure 5C).

Figure 5

WWOX binds NME2 in a competitive manner with KAT1 to inhibit the acetylation of NME2 at site 31 mediated by KAT1. (A, B) Co-IP and IF experiments showed the interaction and co-localization of WWOX and NME2 in HCC cells. Scale bar, 10 µm. (C) (Left) HA-WWOX FL or fragments were cotransfected with GFP-NME2 FL or fragments in 293 T cells, and Western blot detection of mutually interacting structural domains was performed; (Right): Schematic representation of the structural domains of WWOX and NME2. (D) Western blot analysis of NME2 acetylation levels in the indicated HCC cells. (E) Mass spectrometry was used to identify possible acetylation sites in NME2. (F) Huh7 cells were transfected with His-NME2 WT, His-NME2 K31R site mutant plasmids, and NME2 acetylation was detected via Co-IP. (G) Huh7 cells were cotransfected with the above plasmids, and the level of NME2 acetylation was detected using a Co-IP assay. (H) Co-IP assay was used to detect the amount of NME2 bound to KAT1 in the indicated HCC cells. (I, M) Top: Schematic showing the flow of sequential IP assays. Bottom: IP analyses showed interactions of NME2 with WWOX/KAT1 but no binding of WWOX to KAT1. (J) IP analyses of cell lysates from HEK293T cells transfected with GFP-NME2, Myc-KAT1, and HA-WWOX using the indicated tag antibodies. (K) IP analyses of cell lysates from HEK293T cells transfected with GFP-NME2 FL or fragments using the indicated tag antibodies. HCC, hepatocellular carcinoma; IF, immunofluorescence; IP, immunoprecipitation; KAT1, histone acetyltransferase 1; NME2, nucleoside diphosphate kinase 2; qPCR, quantitative real-time PCR; SCD5, stearoyl-CoA desaturase 5; WWOX, WW domain-containing oxidoreductase.

Post-translational modifications of transcription factors have important effects on transcription factor function.24 25 Acetylation modifications affect the binding of transcription factors to target gene promoters, and the presence of acetylation modifications has been reported for other family members of nucleoside diphosphate kinase.24 26 Therefore, we hypothesized that WWOX regulated the acetylation of NME2 and affected its binding to the SCD5 promoter. Notably, Western blot assays showed that the acetylation level of NME2 was elevated in Huh7-shWWOX cells, and the acetylation level of NME2 was reduced in HCCLM3-WWOX cells compared with their counterparts (figure 5D). This finding suggested that WWOX affected the acetylation of NME2. We used mass spectrometry to identify the major acetylation sites of NME2. The result showed that NME2 was primarily acetylated at lysine residues 31, and this site is highly conserved in different species (figure 5E, online supplemental figure 5A). To further confirm the acetylation site of NME2, we prepared lysine-to-arginine mutant NME2-K31R plasmids and transfected it into Huh7 cells. As shown in figure 5F, NME2 acetylation was significantly reduced in cells transfected with the NME2-K31R mutation.

To further identify the acetyltransferase responsible for NME2 acetylation, we used a mass spectrometry approach and found that KAT1 may be a candidate acetyltransferase for NME2 (online supplemental figure 5B). Further immunoprecipitation analysis demonstrated the interaction of endogenous and exogenous NME2 with KAT1 (online supplemental figure 5C). To demonstrate that KAT1 affects the acetylation of NME2, we cotransfected Huh7 cells with His-tagged NME2 and Flag-tagged KAT1 and found that KAT1 significantly increased the acetylation level of NME2 (figure 5G). Cotransfection with the shKAT1 plasmid significantly reduced the level of NME2 acetylation in NME2 WT-transfected cells, which was fully restored in NME2-K31Q-mutated cells (figure 5G). We investigated whether KAT1 mediated the WWOX-induced differences in NME2 acetylation. Transfection of the KAT1 and shKAT1 plasmids into Huh7-shWWOX and HCCLM3-WWOX cells, respectively, revealed that KAT1 completely restored the WWOX-induced differences in NME2 acetylation (online supplemental figure 5D). Further immunoprecipitation analysis revealed that transfection of WWOX reduced the binding capacity of NME2 to KAT1 (figure 5H). Therefore, KAT1 is an acetyltransferase of NME2 and is involved in the WWOX-induced reduction in NME2 acetylation at site 31. We next explored the interaction patterns of WWOX, NME2, and KAT1. Sequential IP assays showed that NME2 that bound WWOX was unable to rebind KAT1, which was validated in further Co-IP assays, suggesting that WWOX interacts with NME2 in a mutually exclusive manner with KAT1 (figure 5I,J). To further clarify the structural domain of NME2-WWOX/KAT1 binding, NME2 full-length and fragments were used to perform IP experiments. Our results indicated that both WWOX and KAT1 bind to the Nucleoside diphosphate kinase (67-152aa) domain of NME2 (figure 5K). Hence, WWOX interacts competitively with KAT1 to NME2 to suppress NME2 acetylation.

We investigated whether NME2 acetylation mediated the WWOX-mediated suppression of SCD5 transcription. The results of DNA pulldown experiments showed complete restoration of NME2 binding to the SCD5 gene promoter in KAT1-transfected HCCLM3-WWOX cells and shKAT1-transfected Huh7-shWWOX cells compared with controls (online supplemental figure 5E). The mRNA and protein levels of SCD5 were similarly fully restored in the transfected NME2-K31R group compared with the NME2 WT group (online supplemental figure 5F,G). These results suggest that NME2 acetylation attenuates the ability of NME2 to bind to the SCD5 gene promoter and mediates WWOX-mediated suppression of SCD5 transcription.

Pharmacological blockade of SCD5 signaling enhances the therapeutic effect of anti-PD-1 therapy in humanized WWOXlow tumor-bearing mice

We further examined whether blockade of SCD5 signaling improved the efficacy of anti-PD-1 therapy. A humanized orthotopic HCC mouse model was constructed with Huh7 cells, which was administered anti-PD-1 antibody, SCD1/5-IN-1 2 weeks later (figure 6A). The results revealed that blockade of SCD5 signaling in combination with anti-PD-1 treatment inhibited tumor growth and lung metastasis more than the control treatment or single treatment alone, and no difference in liver function was observed (figure 6B–D, online supplemental figure 6A–C). IHC analysis of the combination group revealed that M2 macrophage infiltration was reduced in tumor tissues (figure 6E, online supplemental figure 2D–G). These results suggest that WWOX deficiency promotes HCC growth and increases M2 macrophage infiltration in humanized mice and further blockade of SCD5 signaling improved the efficacy of anti-PD-1 therapy.

Figure 6

Pharmacological blockade of SCD5 signaling enhances the therapeutic effect of anti-PD-1 therapy in humanized WWOXlow tumor-bearing mice. (A) Schematic diagram of the orthotopic HCC tumor constructs in humanized mice and the schedule of anti-PD-1 antibody and SCD1/5-IN-1 treatment of WWOXlow tumors in humanized mice. (B) Gross appearance of orthotopic HCC tumors in humanized mice in the indicated groups; (C) Weight and volume of orthotopic HCC tumors at the endpoint in the indicated groups. (D) Representative images of HE-stained mouse lungs from the indicated groups (left) and differential analysis of the number of metastases per lung (right). Scale bar, 100 µm (top) and 50 µm (bottom). (E) Representative IHC staining images of CD68, CD206, NME2 and SCD5 in orthotopic HCC tumor tissues from the indicated groups. Scale bar, 100 µm. ***p<0.001; one-way ANOVA with a post hoc LSD test. ANOVA, analysis of variance; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; NME2, nucleoside diphosphate kinase 2; SCD5, stearoyl-CoA desaturase 5; WWOX, WW domain-containing oxidoreductase.

Correlation and clinical prognostic value of WWOX/NME2/SCD5 axis in patients with HCC

To verify the correlation and prognostic value of WWOX, NME2 and SCD5, we performed IHC staining using tissue microarrays (figure 7A). Correlation analyses suggested that WWOX negatively correlated with SCD5 expression but not with NME2 expression (figure 7B). Notably, survival analysis revealed that patients with WWOXlow/SCD5high HCC had the shortest OS (figure 7C). In addition, HCC patients with SCD5high and NME2low tumors had poorer OS (figure 7D,E). These results emphasize the important prognostic value of the WWOX/NME2/SCD5 axis in HCC patients.

Figure 7

Correlation and clinical prognostic value of WWOX/NME2/SCD5 axis in patients with HCC. (A) Representative images of IHC staining of WWOX, SCD5 and NME2 in HCC tissues. Scale bar, 200 µm (Left) and 40 µm (Right). (B) Correlation analysis of WWOX with SCD5 and NME2 expression. (C) Kaplan-Meier curve analysis of WWOX/SCD5 expression and HCC prognosis. (D) Kaplan-Meier curve analysis of SCD5 expression with HCC prognosis. (E) Kaplan-Meier curve analysis of NME2 expression and HCC prognosis. (F) Schematic illustration of the role of WWOX in regulating the immune microenvironment in HCC. WWOX inhibited the synthesis of oleic acid in HCC cells and immunosuppressive polarization in macrophages by competitively binding KAT1 to NME2, which inhibited the acetylation of NME2 at site 31 and promoted the binding of NME2 to the SCD5 promoter region. Blockade of SCD5 signaling enhances the therapeutic effect of anti-PD-1 therapy in HCC patients with low WWOX expression. χ2 test (log-rank test). HCC, hepatocellular carcinoma; IHC, immunohistochemistry; NME2, nucleoside diphosphate kinase 2; SCD5, stearoyl-Coa desaturase 5; WWOX, WW domain-containing oxidoreductase.

Discussion

ICI therapy has become an effective treatment for solid tumors. However, only some patients achieve therapeutic benefits due to clinical tolerance in HCC patients. The present study indicated that WWOX deficiency was related to poor prognosis and poor treatment response to ICIs. Specifically, we found that WWOX deficiency promoted acetylation of NME2 at site 31 via KAT1, which inhibited its binding to the SCD5 promoter region. This process promotes increased OA secretion from tumor cells and macrophage M2 polarization, which induces HCC immune escape. Therefore, the present study reveals the molecular mechanism of WWOX deficiency promotion