Features of increased malignancy in eosinophilic clear cell renal cell carcinoma

Abstract

Clear cell renal cell carcinoma (ccRCC) is the most common form of renal cancer. Due to inactivation of the von Hippel-Lindau tumour suppressor, the hypoxia-inducible transcription factors (HIFs) are constitutively activated in these tumours, resulting in a pseudo-hypoxic phenotype. The HIFs induce the expression of genes involved in angiogenesis and cell survival, but they also reset the cellular metabolism to protect cells from oxygen and nutrient deprivation. ccRCC tumours are highly vascularized and the cytoplasm of the cancer cells is filled with lipid droplets and glycogen, resulting in the histologically distinctive pale (clear) cytoplasm. Intratumoural heterogeneity may occur, and in some tumours, areas with granular, eosinophilic cytoplasm are found. Little is known regarding these traits and how they relate to the coexistent clear cell component, yet eosinophilic ccRCC is associated with higher grade and clinically more aggressive tumours. In this study, we have for the first time performed RNA sequencing comparing histologically verified clear cell and eosinophilic areas from ccRCC tissue, aiming to analyse the characteristics of these cell types. Findings from RNA sequencing were confirmed by immunohistochemical staining of biphasic ccRCC. We found that the eosinophilic phenotype displayed a higher proliferative drive and lower differentiation, and we confirmed a correlation to tumours of higher stage. We further identified mutations of the tumour suppressor p53 (TP53) exclusively in the eosinophilic ccRCC component, where mTORC1 activity was also elevated. Also, eosinophilic areas were less vascularized, yet harboured more abundant infiltrating immune cells. The cytoplasm of clear cell ccRCC cells was filled with lipids but had very low mitochondrial content, while the reverse was found in eosinophilic tissue. We herein suggest possible transcriptional mechanisms behind these phenomena. © 2020 The Authors. The Journal of Pathology published by John Wiley & Sons, Ltd. on behalf of The Pathological Society of Great Britain and Ireland.

Keywords: VHL; clear cell; electron microscopy; eosinophilic; granular; kidney cancer; mTOR; mitochondria; p53; renal cell carcinoma; vasculature.

Figure 1

Histology of ccRCC tissue. (A–F) Normal kidney (A, C, E) and ccRCC tissue (B, D, F) from the same patient were stained with haematoxylin and eosin (A, B), for RCC antigen (C, D) and carbonic anhydrase 9 (CAIX) (E, F). (G) Overview of a biphasic ccRCC tissue sample. Heterogeneity with areas dominated by clear cells and eosinophilic cells, respectively, is visible. In H and I, the marked areas in G are enlarged. DT, distal tubule; G, glomerulus; PT, proximal tubule. Scale bars = 100 μm.

Figure 2

(A, B) Hierarchical cluster analysis (A) and principal component analysis (B) of sequenced samples. (C) Heatmap illustrating significantly differentially expressed genes in ccRCC_Beo compared with ccRCC_Bcc. (D) Selected Gene Ontology terms correlating to the eosinophilic (eo up) or the clear cell (eo down) phenotype are shown.

Figure 3

(A) Expression levels of the KEGG cell cycle signature in normal kidney tissue and ccRCCs of different phenotypes. (B) Immunohistochemical staining of biphasic ccRCC for Ki67. Positive nuclei are more abundant in the eosinophilic (right) area. Scale bar = 100 μm. (C) Quantification of % cells positive for Ki67 in the double cores from ten cases of eosinophilic (eo) or clear cell (cc) ccRCC included in the validation cohort. p = 0.0032, Student's unpaired t‐test. (D) Relative expression of proximal tubule specific HNF‐regulated gene network in sequenced samples. (E, F) Correlation between the eosinophilic gene signature and the KEGG cell cycle signature (E) or HNF signature (F) in ccRCCs from the TCGA sample collection. (G, H) The eosinophilic expression signature is enriched in ccRCCs from the TCGA of higher stage (G) and is associated with worse patient overall survival (H), based on 495 ccRCCs stratified into quartiles based on each tumour's individual eo signature score. P value represents a log‐rank test.

Figure 4

(A) Immunohistochemical staining of biphasic ccRCC tissue for CD31. Note the difference in vessel density between the clear cell (left) and eosinophilic (right) areas. Scale bar = 100 μm. (B) Quantification of CD31‐positive objects per mm² in the validation cohort of eosinophilic and clear cell ccRCCs. p < 0.0001, unpaired Student's t‐test. (C) Relative expression levels of the ‘hallmark angiogenesis’ gene signature in normal kidney tissue and ccRCCs of different phenotypes as indicated. (D, E) Scoring of CD45 (D) and CD3 (E) staining in ccRCCs of clear or eosinophilic phenotype. Data are presented as % cores for each score, based on evaluation of clear cell or eosinophilic ccRCCs in the validation cohort. Score 1 equals few positive cells and score 3 widespread positivity. p < 0.0001 (CD31) and p = 0.0008 (CD45), χ² test. (F) Relative expression of the Th2 gene signature in normal kidney and ccRCCs of different phenotypes.

Figure 5

(A) GSEA plot showing enrichment of mTORC1 signalling in the eosinophilic signature. NES, normalized enrichment score. (B) Expression levels of mTOR pathway genes as indicated in normal kidney tissue and ccRCCs of different phenotypes. (C) Immunohistochemical staining of biphasic ccRCC tissue for S2448 phosphorylated mTOR. To the left, clear cell tissue; to the right, eosinophilic. Scale bar = 100 μm. (D) Quantification of S2448 phosphorylated mTOR in the validation cohort of clear cell or eosinophilic ccRCC. Data are presented as % cores for each score. 0 equals no positive cells and 2 equals widespread positivity. p < 0.0001, χ² test. (E) Lollipop plot showing localization of TP53 mutations identified in ccRCC_Beo samples. The mutations were in patient 1, H179R (c.536A>G); in patient 3, P151S (c.451C>T); and in patient 5, P152L (c.455C>T). TAD, transcriptional activation domain; Tetra, tetramerization domain. (F) Immunohistochemical staining for p53 in biphasic ccRCC. Eosinophilic area is to the right and clear cell to the left. Scale bar = 100 μm. (G) Quantification of nuclear p53 in eosinophilic and clear cell ccRCCs in the validation cohort, presented as % cores where staining was negative (0), scattered (1) or widespread (2). p < 0.0001, χ² test.

Figure 6

(A, B) Electron microscopy images of clear cell (A) or eosinophilic area (B) in a case of biphasic ccRCC. G, glycogen granules; LD, lipid droplets; M, mitochondria; N, nucleus. (C) Relative RNA levels of genes associated with lipid metabolism in normal kidney cortex and ccRCC samples of different phenotypes as indicated. (D) Immunohistochemical staining for MTCO2 in biphasic ccRCC, where the eosinophilic area is to the right. Scale bar = 100 μm. (E) Relative mitochondrial DNA levels in clear cell or eosinophilic tissue from biphasic ccRCC. n = 2. (F) Relative expression of the indicated genes in normal kidney and ccRCCs of different phenotypes.