Comprehensive Molecular Characterization Identifies Distinct Genomic and Immune Hallmarks of Renal Medullary Carcinoma

Abstract

Renal medullary carcinoma (RMC) is a highly lethal malignancy that mainly afflicts young individuals of African descent and is resistant to all targeted agents used to treat other renal cell carcinomas. Comprehensive genomic and transcriptomic profiling of untreated primary RMC tissues was performed to elucidate the molecular landscape of these tumors. We found that RMC was characterized by high replication stress and an abundance of focal copy-number alterations associated with activation of the stimulator of the cyclic GMP-AMP synthase interferon genes (cGAS-STING) innate immune pathway. Replication stress conferred a therapeutic vulnerability to drugs targeting DNA-damage repair pathways. Elucidation of these previously unknown RMC hallmarks paves the way to new clinical trials for this rare but highly lethal malignancy.

Keywords: SMARCB1; cGAS-STING pathway; molecular profiling; renal medullary carcinoma; replication stress.

Figure 1:. Somatic Genomic Alterations in RMC

Oncoplot showing the clinical characteristics, assays used, the number and types of somatic single-nucleotide variations (SNVs), as well as selected genomic alterations detected in renal medullary carcinoma (RMC) samples. Each column represents a different patient. CR, complete response with long term remission following perioperative chemotherapy and nephrectomy; FFPE, formalin-fixed, paraffin-embedded; PR, partial response by the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1; SD, stable disease by RECIST 1.1; PD, progressive disease by RECIST 1.1; PDX, patient-derived xenograft. See also Figure S1 and Table S1.

Figure 2:. Mutational and Copy Number Landscape of RMC

(A) Tukey boxplots of non-synonymous mutation load per genome for different tumor types. Tumor types are ordered by their median mutation load. RMC samples are highlighted in red. For each boxplot, the central rectangle spans the interquartile range (IQR), the segment within the rectangle shows the median, and the upper and lower whiskers respectively extend the upper and lower hinges of the rectangle by 1.5 * IQR. Black dots represent outliers outside 1.5 * IQR from each hinge. Abbreviations are detailed in the STAR Methods sections. (B) Arm-level copy number alterations in untreated primary RMC tumors. Blue corresponds to loss of one copy, red corresponds to a gain, and dark gray corresponds to more complex alterations shown in detail in Figure S2. (C) Genome plot of RMC4T. In the bottom two panels, the thick black line indicates the median value, blue bars indicate the interquartile range, and red lines indicate segmented values. Loss of heterozygosity is noted on chromosome 22 encompassing the SMARCB1 locus. (D & E) Regions of focal deletion (left) and amplification (right) identified by GISTIC analysis in untreated primary RMC (D) and rhabdoid (E) tumors. G-scores (top X axis) and q values (bottom X axis) are shown. Regions with q values of less than 0.20 (as delineated by the vertical green line) are considered to be significantly aberrant. Only focal copy number alterations (shorter than half the length of a chromosome arm) are shown. (F) Gene Ontology (GO) analysis of genes within regions of recurrent copy number alterations in RMC. See also Figure S1–S3, and Table S2.

Figure 3:. Integrative Characterization of the Mechanisms of SMARCB1 Loss

(A) WES chromosome plot showing chromosome 22 monosomy in sample MED1T. In the bottom two panels, the thick black line indicates the median value, blue bars indicate the interquartile range, and red lines indicate segmented values. (B) MLPA analysis of MED1T confirmed the heterozygous deletion present around the SMARCB1 locus. The heterozygous deletions noted on chromosomes 15 and 16 (CSK and FANCA probes, respectively) were also detected in the WES analysis (Figure S2). (C) CGH profile of MED1T. (D) Break-apart FISH of MED1T confirmed the presence of chromosome 22 monosomy and revealed the presence of a disruptive translocation around the SMARCB1 locus as shown by the separation of the green and orange probes (white arrows) seen inside RMC tumor cells (left image). Two yellow fusion signals (yellow arrows) representing two intact SMARCB1 alleles are noted within the nuclei of normal kidney cells (right image). Scale bar: 10 μm. (E) Sanger sequencing confirmation of the fusion RNA product between exon 3 of SMARCB1 and intron 23 of DCDC2C in the MED1T sample (untreated primary tumor), and of the fusion RNA product exon 1 of SMARCB1 and exon 23 of MYOM1 on both untreated primary tumor (RMC32T) and untreated liver metastasis (RMC32TL) from patient RMC32. (F) Agarose gel electrophoresis of the SMARCB1 fusion products using cDNA from samples RMC32T, RMC32TL, and MED1T. (G) Predicted amino acid sequences of the SMARCB1-DCDC2C fusion product in patient MED1 and of the SMARCB1-MYOM1 fusion product in patient RMC32. See also Table S3.

Figure 4:. Transcriptomic Signature Distinguishes RMC from Other Renal Malignancies

(A) Unsupervised hierarchical clustering of protein-coding gene expression from RMC, CDC, and UTUC. (B) Unsupervised hierarchical clustering of protein-coding gene expression from kidney malignancies. (C) A cartoon of the nephron regions (left; the dashed line separates the renal cortex from the medulla) and heat maps (right) showing intersample correlations (Pearson’s r) between expression profiles of kidney malignancies (arranged by subtype) and expression profiles of kidney nephron sites. S1 and S3, initial and terminal portions of the proximal tubule; mTAL, medullary thick ascending limb of Henle’s loop; cTAL, cortical thick ascending limb of Henle’s loop; DCT, distal convoluted tubule; CCD, cortical collecting duct; OMCD, outer medullary collecting duct. (D) Volcano plot showing the differential expression of genes involved in replication stress and innate immunity (interferon signaling and cGAS-STING pathways). The secondary horizontal line corresponds to a p value of 0.01. (E & F) Pathway diagrams representing differential expression patterns in core metabolic pathways (E), as well as hypoxia-induced genes and EMT (F) between RMC tissues and adjacent normal kidney. See also Figure S3, Figure S4, and Table S2, S4.

Figure 5:. RMC has a Distinct Immune Profile

(A) MCP-counter estimates of infiltrating immune and stromal cells in RMC compared with other carcinomas of the kidney. (B) Immune checkpoint pathway diagram showcasing the interactions of T cells with tumor cells and professional antigen-presenting cells based on the differential RNA expression patterns between RMC tumors and adjacent normal kidney tissues. (C) Representative immunohistochemistry microphotographs for CD3, CD4, CD8, CD20, CD68, FOXp3, PD-L1, and PD-1 in RMC tumor tissues and adjacent normal collecting tubules. Scale bar: 50 μm. (D) Representative immunohistochemistry microphotographs for STING in RMC tumor tissues, adjacent normal collecting tubules, and MRT tumor tissues. Scale bar: 50 μm. (E) Violin plots of the IHC quantification levels for STING in RMC tumor tissues (n = 20), adjacent normal kidney (n = 12) and MRT tumor tissues (n = 12). The width of each violin plot is proportional to the density of observed data points in each region. Dashed and dotted lines correspond to the median and interquartile values, respectively. The upper and lower lines correspond to the highest and lowest observed values, respectively. See also Figure S5 and Table S4, S5.

Figure 6:. SMARCB1 Loss Promotes MYC-induced Replication Stress

GSEA revealed a significant enrichment for the ATR DNA damage repair pathway in response to replication stress in RMC compared with (A) adjacent normal kidney tissues or (B) CDC. ES, enrichment score; NES, normalized enrichment score; FDR, false discovery rate. (C) Hallmark pathways significantly altered (FDR < 0.1) between RMC and CDC by GSEA analysis. (D & E) Western blots of replication stress and DNA damage response pathways following SMARCB1 rescue (D) or direct siRNA inhibition of c-MYC (E) in RMC2C, RMC219, and other SMARCB1-negative cell lines (G401 and VA-ES-BJ). (F) c-MYC peak differences on the promoter site (boxed in red) of the CDK4 gene in G401 MRT cells re-expressing SMARCB1 or EGFP control. The y-axis represents ChIP-seq read counts normalized to 1 million mapped reads. (G) Fold enrichment in c-MYC relative to negative control (normal rabbit IgG) and normalized with input DNA in RMC2C cells following re-expression of SMARCB1 or empty vector control. CCNE2, CDK4, and ATF4 are established c-MYC transcriptional targets, whereas PRM1 is a spermatogenesis-specific gene that is not regulated by c-MYC and serves as negative control. The values are expressed as mean fold change +/− SEM from triplicates. (H) Dot plot of DNA fiber tract lengths indicating a replication speed of ~0.39 kb/min in HEK293-control gRNA cells compared with ~0.51 kb/min in SMARCB1 knock-out cells. Bars (pink) represents the mean of replication tracts (n=187–291, from biological replicas). Top, experimental labeling scheme. Bottom, representative fibers (original magnification x40). See also Figure S6, Table S4, and Table S6.

Figure 7:. RMC is Vulnerable to Drugs Targeting Replication Stress In Vitro and In Vivo

(A) Viability curves and half maximal inhibitory concentrations (IC₅₀) of SMARCB1-negative (RMC2C, RMC219, G401, CHLA-06-ATRT) and SMARCB1-positive (786-O, RCC4, A-498) cell lines after 120-hour exposure to the PARP inhibitors olaparib and niraparib. (B) Viability curves and IC50 of SMARCB1-negative cell lines after exposure to the ATR inhibitors VX970 and AZD6738 and to the WEE1 inhibitor adavosertib. (C & D) Viability of RMC2C, RMC219, G401, and VA-ES-BJ cells expressing doxycycline-induced SMARCB1 or empty vector control (C) or treated with siRNA against c-MYC or sham control (D) followed by 120-hour exposure to olaparib (10 μM), niraparib (10 μM), VX970 (1 μM), AZD6738 (1 μM), or adavosertib (1 μM). * p < 0.05, ** p < 0.01 by unpaired two-tailed Welch’s t-test. All results in A-D are presented as means ± SEM from triplicates. (E) In vivo antitumor effect of niraparib, AZD6738, and their combination in the RMC2X PDX mouse model (n=5 mice / group). Plots represent mean percentage tumor volume change from baseline ± SEM. (F) In vivo antitumor effect of cisplatin alone or in combination with niraparib in RMC tumors (n=10 mice / group). Plots represent mean percentage tumor volume change from baseline ± SEM. (G) Schematic model of the interplay between SMARCB1 loss and CNAs in inducing replication stress and inflammatory responses in RMC. Loss of SMARCB1 and gain of 8q promote MYC-induced replication stress which renders RMC cells susceptible to DNA damaging agents such as platinum salts, topoisomerase inhibitors, and nucleoside analogs. DNA damage repair (DDR) pathways induced by replication stress can be directly targeted by DDR inhibitors. The inflammatory responses activated via cGAS-STING signaling in RMC upregulate immune checkpoints that can be therapeutically targeted. See also Figure S7 and Table S7.