Fig. 1

The plasma biomarkers are differentially expressed in ccRCC patients compared to healthy controls and are significantly correlated with TGF-β and HIF-α/pVHL pathway proteins in solid tumors. a The schematic illustration of the two parts of the study, which focuses on identifying novel biomarkers in the blood of patients with ccRCC. Additional analysis has been performed to investigate the associations between TGF-β and hypoxia pathway proteins from solid ccRCC tumors of the same patient cohort. The volcano plot summarizing the abundance of biomarkers significantly altered in ccRCC samples compared to healthy controls (two-sided, unpaired Wilcoxon Rank Sum test, p-values are FDR corrected. Only biomarkers with log2 fold-change greater than 0.15 or less than −0.15 with adjusted p-value < 0.05 are highlighted and labelled. Representative western blot image showing the protein bands of HIF1A, HIF2A, pSMAD2/3, TGFBR1-FL, TGFBR1-ICD, pVHL, and β-actin, in total protein extracted from 10 randomly chosen, different ccRCC solid tumor excisions. b The ROC curves for correctly classifying ccRCC patients and healthy controls, using gaussian linear models (GLM) and individual proteins one at a time (left) and in combination (right), show accuracy close to 100%. The red curves in both show the model performance when only ANXA1 data is used and each new curve shows the model performance with an additional biomarker, up to all seven. The model with all seven is a perfect classifier (AUC = 1). c Correlation matrix representing correlations between protein components of TGF-β and HIF-α/pVHL pathway proteins with serum biomarkers. pVHL was positively associated with CTSV, ICOSLG, ERBB2, GPNMB, and ITGAV, and negatively with ADAMTS15, SDC1, TNFSF13, NT5E, TXLNA, VEGFA, EGF, IL6, FCRLB, FURIN, SCAMP3, ADAM8, CDKN1A, and ABL1. For the TGF-β pathway, TGFBR1-Full length receptor (FL) and TGFBR1-intracellular domain (ICD), we report here, for the first time, positive associations with ADAMTS15 and FCRLB and negative associations with LY9, CD48, and XPNPEP2. The associations with SMAD2/3 -downstream effectors in the TGF-β canonical pathway- follow the observations for TGFBR1-FL and TGFBR1-ICD. * Denotes significance at p-values < 0.05, Spearman correlation, without investigation of contributions from potential confounders. The frame and arrows indicate some of the important associations described in the text. d Transient overexpression of TGFBR1 or stimulation with TGF-β1 induces NT5E in VHL-negative 786-O cells (immunoblots). Transient overexpression of HA-TGFBR1 in ccRCC cells, as indicated stimulated with TGF-β1 promotes binding of TGFBR1 to endogenous NT5E in two different ccRCC cell lines (immunoprecipitation). Transiently overexpressed TGFBR1 (HA-TGFBR1, green) and NT5E (red) proteins are colocalized in CAKI-1 cell membrane and cytoplasm (merged, yellow). DAPI (blue) stain represents nuclear regions. The scalebar in grey shows 20 μm. Endogenous TGFBR1 (red) and NT5E (green) are colocalized (merged, yellow) in the two ccRCC cell lines treated with TGF-β for 30 min, where the two proteins are likely to form protein-protein complexes. The scalebar in grey shows 20 μm (bottom left). Endogenous TGFBR1 (green) and NT5E (red) are colocalized (merged, yellow) in ccRCC tumors, while not in normal kidney cortex. DAPI and Hoechst (blue) stains represent nuclear regions. Hematoxylin-Eosin (H-E) show morphology. The scalebar in grey shows 50 μm (bottom right)