Oncogenic role of PMEPA1 and its association with immune exhaustion and TGF-β activation in HCC

Abstract

Background & aims: Transforming growth factor β (TGF-β) plays an oncogenic role in advanced cancer by promoting cell proliferation, metastasis and immunosuppression. PMEPA1 (prostate transmembrane protein androgen induced 1) has been shown to promote TGF-β oncogenic effects in other tumour types. Thus, we aimed to explore the role of PMEPA1 in hepatocellular carcinoma (HCC).

Methods: We analysed 1,097 tumours from patients with HCC, including discovery (n = 228) and validation (n = 361) cohorts with genomic and clinicopathological data. PMEPA1 levels were assessed by qPCR (n = 228), gene expression data (n = 869) and at the single-cell level (n = 54). Genetically engineered mouse models overexpressing MYC+PMEPA1 compared to MYC were generated and molecular analyses were performed on the HCCs obtained.

Results: PMEPA1 was overexpressed in 18% of HCC samples (fold-change >2; n = 201/1,097), a feature associated with TGF-β signalling activation (p <0.05) and absence of gene body hypomethylation (p <0.01). HCCs showing both TGF-β signalling and high PMEPA1 levels (12% of cases) were linked to immune exhaustion, late TGF-β activation, aggressiveness and higher recurrence rates after resection, in contrast to HCCs with only TGF-β signalling (8%) or PMEPA1 overexpression (9%). Single-cell RNA sequencing analysis identified PMEPA1 expression in HCC and stromal cells. PMEPA1-expressing tumoural cells were predicted to interact with CD4⁺ regulatory T cells and CD4⁺ CXCL13⁺ and CD8⁺ exhausted T cells. In vivo, overexpression of MYC+PMEPA1 led to HCC development in ∼60% of mice and a decreased survival compared to mice overexpressing MYC alone (p = 0.014). MYC+PMEPA1 tumours were enriched in TGF-β signalling, paralleling our human data.

Conclusions: In human HCC, PMEPA1 upregulation is linked to TGF-β activation, immune exhaustion, and an aggressive phenotype. Overexpression of PMEPA1+MYC led to tumoural development in vivo, demonstrating the oncogenic role of PMEPA1 in HCC for the first time.

Impact and implications: PMEPA1 can enhance the tumour-promoting effects of TGF-β in cancer. In this study, we demonstrate that PMEPA1 is highly expressed in ∼18% of patients with hepatocellular carcinoma (HCC), a feature associated with poor prognosis, TGF-β activation and exhaustion of immune cells. Similarly, in mouse models, PMEPA1 overexpression promotes HCC development, which demonstrates its oncogenic role. The identification of PMEPA1 as oncogenic driver in HCC and its role in immune exhaustion and poor clinical outcomes enhances our understanding of HCC pathogenesis and opens new avenues for targeted therapeutic interventions.

Keywords: GEMM; HCC; STAG1; TGF-β signalling; TMEPAI; genetically engineered mouse model; immune exhaustion; oncogene.

Graphical abstract

Fig. 1

PMEPA1 overexpression in human HCC. (A) Cell viability after 3 days of incubation with recombinant human TGF-β1. (B) Volcano plot of genes differentially expressed (FDR <0.05) in HCC cell lines responsive to TGF-β1 stimulation (green) when compared to cells that are resistant to TGF-β1 stimulation (red). Top signalling pathways enriched based on differentially upregulated genes in the TGF-β-resistant condition (n = 194) are depicted in red. (C) Molecular characteristics of PMEPA1^High human HCCs in the Heptromic cohort (n = 228 patients). Statistical test: Student’s t test, Wilcoxon rank-sum test or Fisher’s exact test, as appropriate. (D) Pathways and biological processes enriched based on n = 562 differentially upregulated genes in PMEPA1^High tumours compared to PMEPA1^Low tumours in both Heptromic and TCGA cohorts (n = 228 and n = 361 patients, respectively), using Comparative Marker Selection analysis. (E) Methylation array levels in CpGs located within the PMEPA1 promoter or gene body in PMEPA1^High (purple) and PMEPA1^Low (grey) samples in the Heptromic cohort. Values represent the mean β-value in each CpG and bars represent the SD between samples. FC is normalized to 1 (mean expression value in non-tumour liver). DEGs, differentially expressed genes; FC, fold-change; FDR, false discovery rate; GO, gene ontology; HCC, hepatocellular carcinoma; TCGA, The Cancer Genome Atlas; TSS, transcription start site.

Fig. 2

PMEPA1 overexpression in the context of Wnt-TGF-β signalling in human HCC. (A) Molecular features of Wnt-TGF-β^Present/PMEPA1^High tumours in the Heptromic cohort when compared to tumours classified as Wnt-TGF-β^Present/PMEPA1^High or Wnt-TGF-β^Absent/PMEPA1^High. Statistical test: Student’s t test, Wilcoxon rank-sum test or Fisher’s exact test, as appropriate. (B) AR protein expression levels assessed by TCGA protein array (RPPA) in samples from the TCGA cohort according to their Wnt-TGF-β/PMEPA1 status. Statistical test: Wilcoxon rank-sum test. Adjusted p values, computed using the Benjamini-Hochberg method, are depicted. (C) Kaplan-Meier estimates of recurrence in patients with HCC from the Heptromic cohort based on PMEPA1^High status (left) or Wnt-TGF-β^Present/PMEPA1^High status (right). Statistical test: log-rank test. EMT, epithelial-mesenchymal transition; HCC, hepatocellular carcinoma; TCGA, The Cancer Genome Atlas.

Fig. 3

Single-cell RNA sequencing assessment of PMEPA1 expression in HCC. (A) UMAP projection of the cell types present in the scRNA-seq dataset comprising 16 patients with HCC (left panel) and with the overlaid PMEPA1 expression level (right panel). (B) Boxplot representation of the average PMEPA1 expression in each cell type of the TME. (C) Bar plot depicting the relative information flow for each ligand-receptor pair between tumour cells (source), coloured according to PMEPA1⁺ status, and CD8⁺ exhausted T cells, CD4⁺ Tregs, and CD4⁺ CXCL13⁺ T cells (target cells) in a scRNAseq cohort of 16 patients with HCC. Ligand-receptor pairs in blue text present a significantly higher information flow in PMEPA1⁺ tumour cells, while beige text denotes significance in their PMEPA1^- counterparts (p <0.01). p values were calculated using a two-sided Wilcoxon test. HCC, hepatocellular carcinoma; pDCs, plasmacytoid dendritic cells; scRNAseq, single-cell RNA sequencing; TME, tumour microenvironment; Tregs, regulatory T cells; UMAP, uniform manifold approximation and projection.

Fig. 4

PMEPA1 promotes HCC development in vivo. (A) Schematic of vectors injected into the mice of the two experimental arms. (B) Representative images of livers from MYC alone or MYC;PMEPA1 models at study endpoint. (C) Median survival of MYC and MYC;PMEPA1 animals. Statistical test: log-rank test. Number of mice per group is also shown. (D) Representative images of PMEPA1 protein levels in MYC;PMEPA1 and MYC-luc;CTNNB1 murine tumours (20X; 100 mm). HCC, hepatocellular carcinoma.

Fig. 5

Transcriptomic and histopathological characterization of MYC;PMEPA1 tumours. (A) Heatmap representing molecular features and signalling pathways of MYC;PMEPA1 tumours in comparison with MYC-luc;CTNNB1 and MYC-lucOS;p53 tumours and healthy liver tissue. Statistical test: Student’s t test, Wilcoxon rank-sum test or Fisher’s exact test, as appropriate. (B) Boxplots depicting AR expression levels (TPM) in healthy liver tissue, MYC-luc;CTNNB1 and MYC;PMEPA1 tumours. Statistical test: Wilcoxon rank-sum test. Adjusted p values, computed using the Benjamini-Hochberg method, are depicted. (C) Barplot depicting the proportion of each tumour differentiation degree category in MYC;PMEPA1, MYC-luc;CTNNB1 and MYC-lucOS;p53 GEMMs. GEMMs, genetically engineered mosaic mouse models; TPM, transcripts per million.