Fig 2: MRPS31 suppression induces mitoribosomal dysfunction, OXPHOS defect, and cell invasiveness.MRPS31_high type cells (JHH4 and HepG2) were transfected with MRPS31 siRNA for 3 days. A Western blots. B Western blots after sucrose gradient sedimentation analysis of mitoribosome. MRPS29 (a small subunit protein) and MRPL13 and MRPL48 (large subunit proteins) were used as indicators of each subunit. Purple, blue and pink colored box indicated 55 S monosome, 39 S large subunit, and 28 S small subunit, respectively. C Mitochondrial translation activity. De novo synthesized proteins using L-homopropargylglycine (HPG)-conjugated Alexa Fluor 488 were detected by Western blot analysis. To distinguish cytosolic and mitochondrial translation activities, emetine (a cytosolic translation inhibitor) and chloramphenicol (CAP, a mitochondrial translation inhibitor) were used. The effect of MRPS31 knockdown on mitochondrial translation is shown in bottom panel. D Cellular OCR of JHH5 (left) and HepG2 (right) and their quantifications (basal and maximal rates, bottom). Nine independent experiments were performed. **p < 0.01 vs. siNC by student t test. E, F Cell invasion activity (left, N = 5) and cell growth (right, N = 3) of MRPS31-knockdowned JHH5 (E) and cell invasion activity (left, N = 3) and cell growth (right, N = 3) of MRPS31-knockdowned HepG2 (F). *p < 0.05 and **p < 0.01 vs. siNC.

Fig 3: MRPS31 loss is closely associated with poor clinical outcome of HCC.A Forrest plot of the hazard ratios for the key SCNA-dependent MRPs by the univariate Cox-regression survival analysis. Their confidence interval (CI, 95%) and p values are indicated. B Boxplot of MRPS31 mRNA levels (left) and overall survival times (OS) using the KM survival analysis (right). Based on MRPS31 mRNA level, TCGA-LIHC samples were stratified into MRPS31_high (n = 93) or MRPS31_low (n = 93) groups (upper or lower quartile). C KM survival curve. Based on RB1 DCN (threshold of −0.2 for DCN loss), TCGA-LIHC samples were divided into RB1 wild type (n = 210) and RB1 deleted type (n = 93) group and then the RB1 wild type group was subdivided into MRPS31_high (n = 52) and MRPS31_low (n = 52) groups (upper and lower quartile, respectively). The statistical p value by the Cox-Mantel log-rank test was depicted for (B) and (C). D Heatmap (top) of HCC classifier expressions and proportions of the predicted classifiers (bottom) obtained by performing NTP algorithm for the high and low MRPS31 groups. The statistical p values by Fisher’s exact test are noted. E Enrichment score (ES) by gene-set enrichment analysis (GSEA) using the Lee’s Poor prognosis A (left) and Woo’s CLHCC_UP (right) gene-sets. Normalized enrichment score (NES) and false discovery rate (FDR) for each gene set are noted. (F) Expression heatmap (left) and ES plots (right) of up (n = 641) or downregulated (n = 788) genes in the high and low MRPS31 groups. NES and FDR for each gene set are indicated. G Barplots of GO analysis using the biological process (BP) gene sets.

Fig 4: Identification of key SCNA-dependent MRPs in HCC.A Schematic flow of the analysis to identify key SCNA-dependent MRPs. B Histograms of the association between DE and SCNA of total genes, RPs, and MRPs. C Heatmaps of DE (left), SCNA (middle), and Cor. R (right) of eight SCNA-dependent MRPs. Average fold changes (FC) of the HCC samples’ DE and SCNA values are included. D Commonly reported SCNA-associated locations and genes and their link to the MRPs. E Whole chromosome SCNA frequencies of high and low MRPS31-DCN groups and their fold differences (bottom). Chromosome boundaries (perpendicular solid lines) and centromere positions (dashed) are indicated and chromosomal region with substantial frequency differences between the two groups was colored yellow. F Correlation estimates between expressions of MRPS31 and the genes located in the yellow-colored region of (E). The top 20 genes with high correlation are listed with their location. Scatter plots (lower panel) of the association between expressions of MRPS31 and neighboring genes (MTRF1 and WBP4) or RB1. The statistical r and p values by Pearson’s product-moment correlation test are depicted.

Fig 5: MMP7 and COL1A1 are downstream effectors of MRPS31 loss to enhance hepatoma cell invasiveness.A Volcano plot of total transcripts (n = 58,387) in TCGA-LIHC, using fold difference and p values of the expression level, based on the permutated student t test between the MRPS31_high (n = 93) or _low (n = 93) group. Names of highly upregulated genes (fold difference > 1.5 and permutated p < 10−4, blue region) are listed (right). B, C MRPS31_high type cells (JHH4 and HepG2) were transfected with MRPS31 siRNA for 3 days. B Messenger RNA levels by qRT-PCR. Fold changes against siNC are displayed (N = 4). *p < 0.05 and **p < 0.01 vs. siNC by student t test. C Western blots. D Western blots of JHH5 cell exposed to the mitochondrial translation inhibitors, chloramphenicol (CAP) and doxycycline (Doxy), for 3 days. E, F JHH5 cells were transfected with MRPS31 siRNA in the absence or presence of siMMP7 (N = 3) (E) or siCOL1A1 (N = 4) (F) for 3 days. Western blots (left) and cell invasiveness (right). *p < 0.05 and **p < 0.01 vs. siNC; #p < 0.05, and ##p < 0.01 vs. siMRPS31.