ACVR2A attenuation impacts lactate production and hyperglycolytic conditions attracting regulatory T cells in hepatocellular carcinoma

Abstract

Although ACVR2A mutations are prevalent in non-viral hepatocellular carcinomas (HCCs), the underlying mechanism remains unelucidated. Our molecular investigation reveals that ACVR2A impairment induces hyperglycolysis through the inactivation of the SMAD signaling pathway. Using syngeneic transplantation models and human clinical samples, we clarify that ACVR2A-deficient HCC cells produce and secrete lactate via the upregulation of lactate dehydrogenase A (LDHA) and monocarboxylate transporter 4 (MCT4) expression levels, which promotes regulatory T (Treg) cell accumulation and then acquires resistance to immune checkpoint inhibitors. Remarkably, genetic knockdown and pharmacological inhibition of MCT4 ameliorate the high-lactate milieu in ACVR2A-deficient HCC, resulting in the suppression of intratumoral Treg cell recruitment and the restoration of the sensitivity to PD-1 blockade. These findings furnish compelling evidence that lactate attenuates anti-tumor immunity and that therapeutics targeting this pathway present a promising strategy for mitigating immunotherapy resistance in ACVR2A-deficient HCC.

Keywords: ACVR2A; LDHA; MCT4; PD-1; hepatocellular carcinoma; immunotherapy; lactate; regulatory T cell.

Graphical abstract

Figure 1

Impacts of ACVR2A attenuation on patient outcomes and biological functions of HCC (A) Cumulative numbers of gene mutations in human viral and non-viral HCC (N = 217 and 154, respectively). The 18 genes specifically mutated in non-viral HCC cases are ranked in ascending order of statistical significance. The p value was calculated by χ2 test. (B and C) Kaplan-Meier curves of OS in patients with ACVR2A-high and -low HCC in the TCGA (B) and Tokyo Medical and Dental University (TMDU) (C) cohorts. The p value was calculated by the log rank test. (D) Sequencing analysis (upper panel) and western blot analysis (lower panel) of ACVR2A in HuH7 and Hepa1-6 cells. β-Actin was used as an internal control. (E–J) Proliferation (E), wound healing (F), migration (G), invasion (H), colony formation (I), and sphere formation (J) assays of HuH7 KO and Hepa1-6 KO cells. Representative photo images in each assay were included. The p value was calculated by Welch’s t test, and data are the mean ± SD (E and F). The p value was calculated by Mann-Whitney U test, and boxes represent the 25th, 50th, and 75th percentiles (G–J). The scale bar represents 500 (F) or 200 μm (G, H, and J). n.s, not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

Glycolysis activation and lactate production in ACVR2A-KO HCC (A) Volcano plot of genes differentially expressed between Hepa1-6 NC and KO cells. Genes with |log2 fold-change| > 1 and p value < 0.05 were highlighted in red. (B) Enrichment plots of gene sets positively associated with Acvr2a knockout in Hepa1-6 cells. NES, normalized enrichment score; FDR, false discovery rate. (C) Quantitative RT-PCR analysis of genes upregulated in HuH7 KO and Hepa1-6 KO cells. Bars represent relative mRNA levels compared to the NC cells. The p value was calculated by ANOVA with Tukey-Kramer post hoc test. (D) Western blot analysis of genes associated with hypoxia and glycolysis. β-Actin was used as an internal control. (E and F) Extracellular lactate (E) and glucose (F) levels. The p value was calculated by Kruskal-Wallis test with Steel-Dwass post hoc test. (G and H) Intracellular lactate (G) and glucose (H) levels. The p value was calculated by Mann-Whitney U test. Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. n.s, not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 3

Recruitment of Treg cells under high-lactate conditions of ACVR2A-KO HCC (A) Tumorigenicity assay of HuH7 KO cells in immunodeficient mice (n = 6). Representative photo images of tumor specimens were included. The p value was calculated by Welch’s t test. (B) Representative immunohistochemical images of ACVR2A, LDHA, and endothelial and immune cell markers in tumors derived from HuH7 KO cells. Nuclei were stained with hematoxylin. The scale bar represents 200 μm (C) Tumorigenicity assay of Hepa1-6 KO cells in immunoproficient mice (n = 6). Representative photo images of tumor specimens were included. The p value was calculated by Welch’s t test. (D) Representative immunohistochemical images of ACVR2A, LDHA, and endothelial and immune cell markers in tumors derived from Hepa1-6 KO cells. The scale bar represents 200 μm (E) Intratumoral lactate levels in HuH7 KO and Hepa1-6 KO cells. The p value was calculated using Mann-Whitney U test. (F) Quantitative immunohistochemical analysis of CD8⁺ T cell and Foxp-3⁺ Treg cell infiltration. The p value was calculated using Mann-Whitney U test. (G) CD8⁺ T cell/Foxp-3⁺ Treg cell ratio. The p value was calculated using Mann-Whitney U test. (H) Quantitative flow cytometric analysis of Foxp-3⁺ Treg cells and CD8⁺ T cells with exhaustion and activation markers. The p value was calculated using Mann-Whitney U test. H&E, hematoxylin and eosin. Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 4

Immunohistopathological evaluation of ACVR2A attenuation, LDHA expression, and Treg cell infiltration in human HCC (A) Representative immunohistochemical images of ACVR2A and LDHA in human HCC samples, indicating the representative ACVR2A-high and -low cases. Nuclei were stained using hematoxylin. H&E, hematoxylin and eosin; C, cancerous tissues; N, adjacent liver tissues. The scale bar represents 100 or 200 μm. (B) Kaplan-Meier curves of OS in patients with the ACVR2A-high and -low HCC groups. The p value was calculated using the log rank test. (C) Multivariate analysis of clinicopathological factors associated with OS in the TMDU cohort. HR, hazard ratio; CI, confidence interval. (D) Representative immunofluorescent images of ACVR2A, CD8, and Foxp-3. Nuclei were stained using DAPI. The scale bar represents 100 μm (E) Quantitative analysis of CD8⁺ T cell and Foxp-3⁺ T cell infiltration. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (F) CD8⁺ T cell/Foxp-3⁺ Treg cell ratio. The p value was calculated using Mann-Whitney U test. (G) Expression analysis of LDHA (left) and enrichment analysis of the glycolytic pathway (right) in the ACVR2A-high and -low HCC groups using the TCGA dataset. The scale bar represents 200 μm. (H) Single-cell analysis of HCC samples. The left panels show UMAP plots of 185,469 live cells in six scRNA-seq datasets annotated by the CellTypist program with the healthy liver (upper) and immune cell (lower) models. The upper right panel shows UMAP plot of 31,794 hepatocytes with enrichment scores for the glycolytic pathway estimated by the ssGSEA program. The lower left panel presents the cytotoxic T cell/Treg cell ratio in the glycolysis-high and -low HCC groups (N = 28 and 28, respectively). Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. n.s, not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 5

Lactate production induced by LDHA overexpression via the activin/SMAD signaling pathway in ACVR2A-KO HCC (A and B) Western blot analysis of HIF1A and LDHA expression in HuH7 KO and Hepa1-6 cells with HIF1A knockout (A) and LDHA knockdown (B). β-Actin was used as an internal control. (C) Co-immunoprecipitation analysis of Hif1a and L-lactyl-lysine. (D and E) Tumorigenicity assay of HuH7 KO cells with LDHA knockdown in immunodeficient (D) and immunoproficient (E) mice (n = 6). Representative photo images of tumor specimens were included. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (F) Intratumoral lactate levels in tumors derived from Hepa1-6 KO cells with Ldha knockdown. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (G) Representative immunohistochemical images of CD8⁺ T cells and Foxp-3⁺ T cells. Nuclei were stained with hematoxylin. (H) Quantitative analysis of CD8⁺ T cell and Foxp-3⁺ T cell infiltration. The p value was calculated using Mann-Whitney U test. (I) Quantitative ChIP analysis of SMAD4 at the promoter region of LDHA. The p value was calculated by Welch’s t test. (J) Schematic representation of the promoter region of LDHA. H&E, hematoxylin and eosin. Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 6

Impairment of Treg cell infiltration by MCT inhibition in ACVR2A-KO HCC (A) Western blot analysis of MCT4 expression in HuH7 KO and Hepa1-6 KO cells. β-Actin was used as an internal control. (B) Dose-response curves of the MCT4 inhibitor VB124 in HuH7 KO and Hepa1-6 KO cells. (C and D) Proliferation (C) and colony formation (D) assays of HuH7 KO and Hepa1-6 KO cells treated with VB124. Representative photo images in each assay were included. The p value was calculated by Welch’s t test (C). The p value was calculated by Mann-Whitney U test (D). (E and F) Quantitative flow cytometric analysis of Foxp-3⁺ Treg cells co-cultured with HuH7 KO cells (E) and Hepa1-6 KO cells (F). The p value was calculated using Mann-Whitney U test or Kruskal-Wallis test with Steel-Dwass post hoc test. Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. n.s, not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 7

Synergistic anti-tumor effects of MCT inhibition and immune checkpoint blockade on ACVR2A-KO HCC (A) Tumorigenicity assay of Hepa1-6 KO cells with Mct4 knockdown in immunoproficient mice treated with anti-PD-1 antibody (n = 4). Representative photo images of tumor specimens were included. The p value was calculated by ANOVA with Tukey-Kramer post hoc test. (B) Quantitative immunohistochemical analysis of CD8⁺ T cell and Foxp-3⁺ T cell infiltration. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (C) CD8⁺ T cell/Foxp-3⁺ Treg cell ratio. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (D) Intratumoral lactate levels. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (E) Tumorigenicity assay of Hepa1-6 KO cells in immunoproficient mice treated with anti-PD-1 antibody and VB124 (N = 4). Representative photo images of tumor specimens were included. The p value was calculated by ANOVA with Tukey-Kramer post hoc test. (F) Quantitative immunohistochemical analysis of CD8⁺ T cell and Foxp-3⁺ T cell infiltration. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (G) CD8⁺ T cell/Foxp-3⁺ Treg cell ratio. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (H) Intratumoral lactate levels. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (I) Tumorigenicity assay of 3H3-Pten-KO cells with Acvr2a knockout in immunoproficient mice treated with anti-PD-1 antibody and VB124 (N = 4). Representative photo images of tumor specimens were included. The p value was calculated by ANOVA with Tukey-Kramer post hoc test. (J) Quantitative immunohistochemical analysis of CD8⁺ T cell and Foxp-3⁺ T cell infiltration. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (K) CD8⁺ T cell/Foxp-3⁺ Treg cell ratio. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (L) Intratumoral lactate levels. The p value was calculated using Kruskal-Wallis test with Steel-Dwass post hoc test. (M and N) Quantitative flow cytometric analysis of Foxp-3⁺ Treg cells and CD8⁺ T cells with exhaustion and activation markers in Hepa1-6 KO (M) and 3H3-Pten-KO with Acvr2a-knockout (N) xenografts. The p value was calculated using Mann-Whitney U test. (O) Representative case of HCC with high SUVmax. The left and right panels show an FDG-PET/CT image and representative immunohistochemical images of ACVR2A, LDHA and Treg cells, respectively. Nuclei were stained using hematoxylin. The scale bar represents 100 or 200 μm. Boxes represent the 25th, 50th, and 75th percentiles. Data are the mean ± SD. n.s, not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.