S100A9 Derived from Chemoembolization‐Induced Hypoxia Governs Mitochondrial Function in Hepatocellular Carcinoma Progression

Abstract Transarterial chemoembolization (TACE) is the major treatment for advanced hepatocellular carcinoma (HCC), but it may cause hypoxic environment, leading to rapid progression after treatment. Here, using high‐throughput sequencing on different models, S100 calcium binding protein A9 (S100A9) is identified as a key oncogene involved in post‐TACE progression. Depletion or pharmacologic inhibition of S100A9 significantly dampens the growth and metastatic ability of HCC. Mechanistically, TACE induces S100A9 via hypoxia‐inducible factor 1α (HIF1A)‐mediated pathway. S100A9 acts as a scaffold recruiting ubiquitin specific peptidase 10 and phosphoglycerate mutase family member 5 (PGAM5) to form a tripolymer, causing the deubiquitination and stabilization of PGAM5, leading to mitochondrial fission and reactive oxygen species production, thereby promoting the growth and metastasis of HCC. Higher S100A9 level in HCC tissue or in serum predicts a worse outcome for HCC patients. Collectively, this study identifies S100A9 as a key driver for post‐TACE HCC progression. Targeting S100A9 may be a promising therapeutic strategy for HCC patients.

A) KEGG enrichment analysis of DEGs in HCC tissues in patients between the early recurrence and late recurrence groups. B) Heatmap showing the DEGs between early recurrence and late recurrence (showing only the 10 most upregulated genes in each group) (tumor recurrence within two years after surgical resection). C) Heatmap showing the DEGs between the ischemia group and the control group (showing only the 10 most upregulated genes in each group). D) Heatmap showing the DEGs between day 0 and day 14 from the CRISPR/Cas9 library screen in PLC/PRF/5 cells (showing only the 10 most upregulated genes in each group). E) Flow cytometry showing the proportion of EPCAM+/S100A9+ and CD45+/S100A9+ cells among total S100A9+ cells in the HCC tissues of patients who received only surgery (top) or patients who received TACE followed by surgery (bottom). F) TACE upregulated the proportion of EPCAM+/S100A9+ cells and downregulated the proportion of CD45+/S100A9+ cells among the total S100A9+ cells in the HCC tissues of patients who received TACE followed by surgery compared to patients who received surgery only (n=6 each group). G-J) Multiple immunofluorescence staining showing DAPI (gray), S100A9 (red), EPCAM (green, Fig G), CD45 (yellow, Fig H), CD68 (bright blue, Fig I) and CD3 (blue, Fig J) expression, coexpression (double-positive cells) and quantification of each kind of cell proportion in the tumor regions in patients treated with or without TACE (three random areas for five different samples). K) RNA expression of S100A9 in orthotopic liver xenograft tumors after different treatments in vivo. Epi (epirubicin), Cis (cisplatinum), Isc (ischemia). L) RNA expression of S100A9 in MHCC-97H cells after different treatments in vitro. H (hypoxia) (1% oxygen 24 hours). M) Hypoxia stabilized HIF1A and induced S100A9 expression in Hep-3B cells. N) Silencing HIF1A suppressed the expression of S100A9 induced by hypoxia in MHCC-97H cells. O) Schematic view of the luciferase reporter constructs containing various lengths of the 5′-flanking regions of the S100A9 promoter. Detailed characterization of the S100A9 promoter by 5′-deletion and site-specific deletion analyses was performed. P) S100A9 promoter luciferase activity in PLC/PRF/5 cells or 293T cells cotransfected with the vector or the indicated luciferase reporter and pcdna3.1-HIF1A constructs for 48 h. Q) Representation of S100A9 promoter regions and the indicated primers for ChIP assay. R) ChIP assay for HIF1A occupancy at the S100A9 promoter in PLC/PRF/5 cells or Huh7 cells. Precipitated DNA was purified and subjected to semiquantitative PCR. Data in (F)-(L) and (P) is presented as mean ± SEM, * P<0.05, ** P < 0.01, *** P < 0.001, by two-tailed unpaired Student t-test.

Supplementary figure 2 S100A9 promotes the growth and metastasis of HCC cells in vitro and in vivo.
A) Real-time PCR (left) and Western blotting assays (right) showing the expression of S100A9 in hepatoma cell lines(n=3). B-C) Real-time PCR and Western blot assays showing the efficiency of S100A9 knockout/overexpression in PLC/PRF/5 (B) and MHCC-97H (C) cells(n=3). D) Tas inhibited the expression of S100A9 in HCC cells. PLC/PRF/5 cells were treated with Tas at the indicated concentration for 24 h before harvest. MHCC-97H-OE and Huh7 cells were treated with 100 µM Tas for 24 h before harvesting. E) Tas inhibited the proliferation of MHCC-97H-OE (left) and Huh7 (right) cells, as indicated by CCK-8 assays(n=3). F-G) Knockout of S100A9 suppressed and overexpression of S100A9 enhanced the migration of HCC cells, as indicated by wound-healing assays. H-J) Knockout of S100A9 or 100 µM Tas inhibited the cell migration of HCC cells, as indicated by Transwell assays(n=3). K) Western blot analysis showing the expression of S100A9 and EMT markers in Huh7/S100A9-sg cells. L) Overexpression of S100A9 enhanced the metastatic ability of tumor cells, even after neutralizing A.) S100A9 knockout reduced intracellular ROS production in Huh7 cells, as indicated by DCFH-DA fluorescence assay. B-D) Tas reduced intracellular ROS production in PLC/PRF/5 (B), Huh7 (C) and

MHCC-97H (D) cells. E-F) Mdivil-1(10μM) reduced intracellular ROS production in PLC/PRF/5 (E) and
MHCC-97H (F) cells. G) Effects of S100A9 on oxidative phosphorylation (as determined by Seahorse XF analyzers) in PLC/PRF/5 (top) and MHCC-97H cells (bottom). The oxygen consumption rate (OCR) over time were shown. H) Effects of S100A9 on glycolytic rates (as determined by Seahorse XF analyzers) in PLC/PRF/5 (top) and MHCC-97H cells (bottom). The extracellular acidification rate (ECAR) (left) over time and the calculated glycolytic rates(right) were shown. I) S100A9 knockout reduced whereas S100A9 overexpression increased ATP production in HCC cells. J-K) Knockout of S100A9 or overexpression of S100A9 did not consistently change the RNA levels of the NOX family, as indicated by real-time PCR. L) Mito-tempo (20μM)  ** P < 0.01, *** P < 0.001 by two-tailed unpaired Student t-test.

Supplementary figure 5
Supplementary figure 5 S100A9 affects HCC growth and metastasis through a PGAM5-dependent pathway.

A) Western blot analysis showing the PGAM5 overexpression efficiency in PLC/PRF/5 cells. B) Western blot analysis showing the PGAM5 knockdown efficiency in MHCC-97H cells. C) Western blot analysis
showing the PGAM5 overexpression efficiency in Huh7 cells. D) Ectopic expression of PGAM5 enhanced the growth of S100A9-sg Huh7 cells, as indicated by EdU assay. E) Ectopic expression of PGAM5 enhanced the migration of S100A9-sg Huh7 cells, as indicated by the Transwell assay. F) Ectopic expression of PGAM5 enhanced the intracellular ROS level of Huh7-S100A9-sg cells, as indicated by DCFH-DA fluorescence assay. G) Ectopic expression of PGAM5 enhanced the intramitochondrial ROS level of Huh7-S100A9-sg cells, as indicated by the MitoSOX fluorescence assay. Data in (E)-(G) are presented as mean ± SEM, n = 3. *** P < 0.001 by two-tailed unpaired Student t-test.

Supplementary figure 6 Clinical significance of S100A9 inhibition in HCC.
A) Stacked histogram for AFP levels in the S100A9-High group and S100A9-Low group in the SYSUCC cohort. B) Correlation between S100A9 expression and pathological stage in the SYSUCC cohort (n=172) (grouping by median S100A9 expression based on IHC score). C) Correlation between S100A9 expression and tumor size in the SYSUCC cohort (n=172) (grouping by median S100A9 expression based on IHC score). D-F) Correlation between S100A9 expression and T stage, tumor stage and tumor grade in the TCGA-LIHC cohort. Data in (D)-(F) are presented as mean ± SEM. n=352. Significant differences were determined by one-way ANOVA. G) Kaplan-Meier curve analysis of OS in HCC patients by the expression of S100A9 in the TCGA-LIHC cohort (n=352) (grouping by median S100A9 presented as mean ± SEM, n = 6. * P < 0.05 by two-tailed unpaired Student t-test.

Supplementary Table 1 Correlation between S100A9 and clinicopathological parameters
Characteristics S100A9