Tumor heterogeneity in VHL drives metastasis in clear cell renal cell carcinoma

Abstract

Loss of function of the von Hippel-Lindau (VHL) tumor suppressor gene is a hallmark of clear cell renal cell carcinoma (ccRCC). The importance of heterogeneity in the loss of this tumor suppressor has been under reported. To study the impact of intratumoral VHL heterogeneity observed in human ccRCC, we engineered VHL gene deletion in four RCC models, including a new primary tumor cell line derived from an aggressive metastatic case. The VHL gene-deleted (VHL-KO) cells underwent epithelial-to-mesenchymal transition (EMT) and exhibited increased motility but diminished proliferation and tumorigenicity compared to the parental VHL-expressing (VHL⁺) cells. Renal tumors with either VHL⁺ or VHL-KO cells alone exhibit minimal metastatic potential. Combined tumors displayed rampant lung metastases, highlighting a novel cooperative metastatic mechanism. The poorly proliferative VHL-KO cells stimulated the proliferation, EMT, and motility of neighboring VHL⁺ cells. Periostin (POSTN), a soluble protein overexpressed and secreted by VHL non-expressing (VHL^-) cells, promoted metastasis by enhancing the motility of VHL-WT cells and facilitating tumor cell vascular escape. Genetic deletion or antibody blockade of POSTN dramatically suppressed lung metastases in our preclinical models. This work supports a new strategy to halt the progression of ccRCC by disrupting the critical metastatic crosstalk between heterogeneous cell populations within a tumor.

Fig. 1

Human ccRCC tumor specimen showed intratumoral heterogeneity in VHL expression. H&E stain and VHL IHC performed on parallel sections of tumor from different cases of human RCC are shown in (a) case #20, (b) case #21, (c) case #22 and (d) case #23. Scale bar for low magnification field: 2 mm; for high magnification field: 100 μm. e Mutational data from WES of common oncogenic driver genes in ccRCC are shown for four loci of the patient’s tumor and the derived cell line of case #22. Point sizes represent variant allele frequencies. Values above 0.4 represent likely clonal mutations and above 0.9 represent clonal mutations combined with loss of heterozygosity. Colors represent log2 CNRs for each gene, with DNA gains in red and losses in blue. f UMAP dimensionality reduction plot of single-cell sequencing from the case #22 primary tumor tissue and disassociated primary tumor cells colored by VHL expression positivity in each cell. A total of 1343 P0 (passage 0) #22 tumor cells were analyzed. g Scatterplot of sequencing depth vs. VHL imputed gene expression across non-cycling single cells from single cell sequencing in the case #22 with the number of cells at each expression level of VHL listed. h Density plot showing the CNRs of the VHL locus in the TCGA-KIRC cohort (n = 459) after adjusting for both tumor purity and ploidy with consensus estimates of purity. A CNR value of −1.1 represents a two-copy loss of VHL (dotted line) and −0.4 indicates one copy loss. i Density plot showing the VAF of somatic VHL mutations in the TCGA-KIRC cohort (n = 148) after adjusting for both tumor purity and ploidy (with consensus estimates of tumor purity)

Fig. 2

VHL-KO cells cooperate with VHL-WT cells to cause metastasis. a Mice were implanted with a total of 1 × 10^6 tumor cells, either RC-VHL-WT, RC-VHL-KO, or a 1:1 mixture of the two cell types, into their left kidney (n = 6 per group). Images of three representative animals assessed by BLI on week 4 post-implantation are shown. b Primary tumor indicated by BLI in different groups are shown at the end point. c Lung metastases are indicated by elevated BLI signals in the thoracic cavity in the mixed tumor group. One-way ANOVA was used to test the statistical significance in (b) and (c) with n = 6, presented with mean ± SD. d H&E-stained sections (low magnification) of lung and heart of each animal in the RC-VHL-WT tumor group and mixed (1:1) tumor group. Arrowhead indicates the heart. Scale bar: 3 mm. e RC tumor growth was assessed in the CAM tumor system through longitudinal observation. Day 0 is the day of tumor cell implantation, which occurred on day 7 postfertilization. Scale bar: 1 cm. One-way ANOVA was used in the comparison with n = 9, presented with mean ± SD. f Flow cytometric analyses of CTC from chick embryos, bearing either VHL-WT cells (marked with mStrawberry) or 1:1 VHL-WT and VHL-KO (marked with EGFP) mixed CAM tumor, harvested on day 21 post-fertilization. g In the CAM tumor system, the metastasis in the duck liver can only be seen in mixed implantation of primary tumor cells from #22 and its VHL-KO counterpart as analyzed by TaqMan probes for human alu sequence and avian β-actin. n = 5 with “no signal” sample dots not shown and Ct value differences are presented with mean ± SEM. h Heatmap of the top 200 genes based on conserved gene expression patterns across comparisons of #22 VHL⁺ vs. VHL⁻ cells in single-cell sequencing (as noted in Fig. 1f), VHL KO vs. VHL WT ACHN models, and VHL KO vs. VHL WT RENCA models. Heatmap colors represent log2 fold changes z-score scaled by column. Detailed information is listed in Supplementary Table 2. i Top 30 gene sets based on conserved enrichment scores across the same three comparisons are listed. The color of points represents the differential expression comparison (ACHN VHL-KO vs VHL-WT as red, RENCA VHL-KO vs VHL-WT as green, and the single cell comparison of #22-VHL⁺ vs VHL⁻ as blue) and the size of the points corresponds to the adjusted p-value from GSEA, in which the bigger circles indicate up/down-regulation with lower p-values. (*p < 0.05, **p < 0.01) (not consistent with adjusted p vals in Fig. 2i)

Fig. 3

VHL⁻ cells induce the proliferation and the metastasis of VHL⁺ cells. a The growth rate of RC-VHL-WT (red line), RC-VHL-KO cells (green line), and a 1:1 mixture of the two cells (orange line) in a transwell setting. b The growth rate of ACHN VHL-WT cells (red line), its VHL-KO counterpart (green line) or VHL-WT cells with the addition of VHL-KO condition media (orange line). c The growth rate of the Caki-1 -VHL-WT (red line), Caki-1-VHL-KO (green line), and the Caki-1-VHL-WT in the conditioned media from Caki-1-VHL-KO (orange line). d The growth rate of the VHL + primary cell line from case #22 (red line), its VHL-KO derivative (green line) and VHL + co-culture with its VHL-KO derivative in a transwell setting. One-way ANOVA was used in the comparisons in (a), (b), (c) and (d) with triplicate repeats, presented with mean ± SD. e A section of primary tumor derived from implantation of 1:1 RC-VHL-KO:RC-VHL-WT cells was stained with IF to detect VHL (red), Ki67 (green), and nuclei (DAPI, blue). The dash lines demarcate VHL-negative areas with intact nuclei. Scale bar: 200 μm. f VHL IHC and Ki-67 IF in serial sections of human ccRCC (case #22). High-magnification images show cytoplasmic VHL expression specifically in area (#) and not in area (*). The bar graph shows the average percentage of Ki-67 positivity in the VHL-positive and VHL-negative regions. Scale bar for low magnification field: 3 mm; for high magnification field: 50 μm. Student t-test was used in the comparisons (e) and (f) with triplicate repeats and presented with mean ± SD. g The spatial relationship between VHL and Ki-67 expression in case #22 was assessed with IF. Fluorescent images were analyzed with HALO software. In the first and fifth panels, VHL-positive cells and Ki-67-positive cells are represented by blue dots, respectively. The second panel shows a heatmap of VHL-positive cell density. The third panel shows a boundary map of the VHL-positive tumor regions as topographic contour lines indicating the distance from the tumor boundary. For distance measurements of Ki67-positive cells, contour lines were placed up to 2000 μm from the tumor edge toward the inside of the tumor and up to 4000 μm away from the tumor edge of VHL-positive tumor regions. Regions between the contour lines are shown as different colors from the innermost red to farthest blue. Ki-67-positive cells in each region were counted, normalized to the area, and plotted in a histogram that is shown in the fourth panel. Scale bar: 3 mm. h IF staining of the CD31 (yellow), VHL (red), Ki-67(white) as well as DAPI (blue) in mixed implanted RENCA-VHL-WT and VHL-KO cells mouse tissues. Left: the primary tumor. Right: lung metastasis. Scale bar for the low magnification field: 100 μm; for high magnification field: 30 μm. i VHL IHC staining of a large, lung metastasis from the mixed implanted group with RC-VHL-WT and RC-VHL-KO cells. Scale bar for low magnification field: 1 mm; for high magnification field: 200 μm. j IF of a small, lung metastatic nodule from a mouse implanted with the mixture with HA-positive RC-VHL-WT cells shown in red and the few flag-positive VHL-KO cells shown in green. Scale bar: 100 μm. k Flow cytometry analysis of the lung metastasis showing the relative proportion of RC-VHL-WT (TRITC+) and RC-VHL-KO(FITC+) cells. l IF of CD31(yellow), VHL (red) and DAPI (blue) on the primary tumor tissue from the case #22. Scale bar for low magnification field: 100 μm; for high magnification field: 30 μm. m H&E staining and (n) VHL IHC staining of the lung metastasis from the case #22. Scale bar for the low magnification field: 5 mm; for high magnification field: 100 μm. (*p < 0.05, **p < 0.01)

Fig. 4

POSTN is overexpressed in VHL-KO or VHL⁻ cells in tumors. a Western blot assessing the expression of VHL, HIF1A and POSTN protein in RENCA VHL-WT, VHL-KO and VHL/HIF1A double knockout cells. b qRT-PCR of POSTN RNA in RENCA, ACHN and the patient-derived primary cells from case #22 and the VHL-deleted derivative (VHL-KO) of each. Student t-test was used in the comparison with triplicate repeats and presented with mean ± SD. c Serial sections from a large, lung metastatic lesion resulting from implantation of a 1:1 mixture of VHL-WT and VHL-KO cells were stained with H&E and IHC against HA tag (VHL-WT cells), flag tag (VHL-KO cells), and anti-POSTN. Scale bar: 1 mm. d Serial sections of the primary tumor tissue from case #22 were stained for VHL and POSTN. Dotted encircled areas are VHL-positive cell aggregated areas (left) and POSTN positive aggregated areas (right). Higher magnification of the boxed areas (a) and (b) are shown on the right. Scale bar for low magnification field: 3 mm; for high magnification field: 100 μm. e Multiplex IF stain, analyzed by HALO software, showing the cellular distribution of VHL⁺ POSTN⁻ and VHL⁻POSTN⁺ cells in case #22 primary tumor. Scale bar: 3 mm. f HALO analysis in (e) was used to generate heatmap of VHL⁺ POSTN⁻ and VHL⁻POSTN⁺ cells. Warmer colors (orange) identify areas of denser cells, and cooler colors (dark blue) signify areas with sparser cells. The rightmost plot scored the VHL⁺ POSTN⁻ and VHL⁻POSTN⁺ cells in each evenly divided area with respect to the interface border of the VHL⁺ POSTN⁻ area. It shows VHL⁺ POSTN⁻ cells (red curve, left y-axis) and VHL⁻POSTN⁺ cells (green curve, right y-axis) do not co-localize in the same area. Scale bar: 3 mm. g The lung metastatic lesion of case #22 and (h) the retroperitoneal lymph node metastatic lesion of case #17 were stained by H&E and IF to detect VHL (red), POSTN (green), and nuclei, (DAPI, blue). Scale bar for the low magnification field: 1 mm; for the high magnification field: 100 μm. (*p < 0.05, **p < 0.01)

Fig. 5

VHL-KO cells induce the EMT and motility of VHL-WT cells by POSTN. a EMT markers (E-cadherin, N-cadherin, MMP-9 and alpha-SMA) assessed by qRT-PCR in RC-VHL-WT cells alone or co-cultured with VHL-KO cells, separated by a transwell plate (VHL-WT + VHL-KO). b–g The motility of fluorescently marked RC cells was measured in a 2D scratch assay by time lapse live-cell microscopy, monitored over 20 h. b The motility of mStrawberry marked RC-VHL-WT cells and EGFP marked RC-VHL-KO cells were measured. c The motility of a 1:1 mixed culture of RC-VHL-WT and RC-VHL-KO cells (top panels). TRITC channel revealed the motility of RC-VHL-WT cells in the mixed culture (lower panel). Respective videos are in Supplementary Movie S1. d The quantified migration speed of the three cell groups is shown. e The migration speed of VHL-WT cells with (right) or without (left) the addition of conditioned medium of VHL-KO cells, cultured for 2 days at 90% confluence. Respective videos are shown in Supplementary Movie S3. f The migration speed of VHL-WT cells (mStrawberry⁺) co-cultured with VHL/POSTN-KO cells (EGFP⁺) or VHL-KO cells (EGFP⁺) was shown (also see Supplementary Movie S5). g Anti-POSTN mAb MPC5B4 (1 µg/mL) was added to VHL-WT cells co-cultured with VHL-KO cells. The migration speed of VHL-WT cells without and with MPC5B4 was shown (see Supplementary Movie S6). h The migration speed of VHL-WT cells alone or with the addition of recombinant POSTN protein or with the addition of POSTN and the integrin inhibitor cilengitide was assessed. i Scratch assay of a 1:1 mixture of VHL-WT and VHL-KO cells was assessed at 18 h. The migration speed of VHL-WT cells in the mixture was measured with 0, 2, 5 or 10 μM cilengitide added (see Supplementary Movie S7). Scale bar: 100 μm. Student t-test was used in (a), (e), (f) and (g) and one-way ANOVA was used in (d), (h) and (i) with triplicate repeats and presented with mean ± SD. (*p < 0.05, **p < 0.01)

Fig. 6

Blocking POSTN’s pro-metastatic activities, including vascular destruction, inhibited lung metastases. a HUVECs were co-cultured in a transwell setting with RC-VHL-WT (labeled “VHL (+)” here) or RC-VHL-KO (labeled “VHL(−)” here) cells without or with the anti-POSTN mAb MPC5B4. HUVECs were harvested after 48 h of culturing, and cell extracts were analyzed by western blot for necroptosis- and apoptosis-associated proteins. b HUVECs co-cultured with VHL-WT or VHL-KO cells for 48 h were assessed for necroptosis with a reporter assay by scoring the count of EthD-III(+) cells, which was normalized to the Hoechst 33342-positive nuclei count. c Apoptosis was evaluated with the Caspase-Glo 3/7 luminescence reporter assay in HUVECs co-cultured with VHL-WT cells, VHL-KO cells, or VHL-KO cells plus 1 µg/mL of anti-POSTN MPC5B4 mAb. d Tumor vascular leakage was assessed with the Miles assay on CAM tumors from VHL-WT cells or a 1:1 mixture of VHL-WT and VHL-KO cells. Evans Blue dye was injected intravenously into the chick embryo. The extent of tumor vascular leakage was scored by the amount of dye that leaked into the tumor. e Mice received intrarenal implantation of 1 × 10^6 total cells consisted of 1:1 mixture of RC-VHL-WT cells and RC-VHL-KO cells, or a 1:1 mixture of RC-VHL-WT and RC-VHL/POSTN-KO cells. BLI at 4 weeks post-implantation is shown. f Mice that received renal implantation of a 1:1 mixture of RC-VHL-WT and RC-VHL-KO cells were treated with either control IgG or anti-POSTN MPC5B4 mAb (n = 6). Scale bar: 1 cm. g Primary tumors and lungs harvested at 4 weeks post-implantation are shown with lung weights. h H&E stain of lung lobes and heart from control- or MPC5B4-treated tumor-bearing animals. Scale bar: 3 mm. i IF stain for POSTN in red, VHL in green, and DAPI in blue of the same tumor sections in (h). White arrows indicate selected lung metastases. Scale bar: 3 mm. j primary tumor weights for each group are shown. k Gross view of the primary tumor on CAM treated with either control or MPC5B4, and tumors were established from mixed implantation of the primary cell line from case #22 and its VHL-KO counterpart. Scale bar: 1 cm. The qRT-PCR relative quantification of the chicken liver metastasis is shown in (l), and the primary tumor weight analysis is shown in (m) as mean ± SEM. Student t-test was used in (b) and (d) with triplicate repeats, in (g), (j) and (l) with n = 6 (“no signal” sample dots not shown, Ct threshold set at 30), as well as in (m) with n = 8. One-way ANOVA was used in (c) with triplicate repeats. All results mentioned above except (l) and (m) are presented with mean ± SD. (*P < 0.05, **P < 0.01)

Fig. 7

Paracrine action of POSTN in the collaborative metastasis model between VHL- and VHL + cells. a A summary of the pro-metastatic paracrine functions mediated by POSTN. b The cooperative metastatic mechanism uncovered paracrine-promoting interactions between VHL-KO and VHL-WT cells, mediated by POSTN, at the tumor proper and the intravasation step (1) of the metastatic cascade. The potential impact of the cooperative metastatic mechanism on downstream steps of the cascade, such as cell survival in the circulation (step 2), and metastatic colonization (step 3) requires further investigation