Summary:
Xp11 translocation renal cell carcinoma is a rare, female-predominant cancer driven by a fusion between the TFE3 gene on chromosome Xp11.2 and a partner gene on either chrX or an autosome. It remains unknown what types of rearrangements underlie TFE3 fusions, whether fusions can arise from both the active (chrXa) and inactive X (chrXi) chromosomes, and whether TFE3 fusions from chrXi translocations account for the female predominance of tRCC. To address these questions, we performed haplotype-specific analyses of chrX rearrangements in tRCC whole genomes. We show that TFE3 fusions universally arise as reciprocal translocations and that oncogenic TFE3 fusions can arise from chrXi:autosomal translocations. Female-specific chrXi:autosomal translocations result in a 2:1 female-to-male ratio of TFE3 fusions involving autosomal partner genes and account for the female predominance of tRCC. Our results highlight how X chromosome genetics constrains somatic chrX alterations and underlies cancer sex differences.
Keywords: Cancer sex bias, cancer genomics, X-inactivation, X chromosome, XIST, gene fusions, kidney cancer, renal cell carcinoma, translocation renal cell carcinoma, tRCC, TFE3, MITF
Graphical Abstract
In Brief
Whole genome analysis of a female predominant kidney cancer reveals the mechanisms by which X chromosome genetics constrains somatic X chromosome alterations and underlies sex bias in tumorigenesis
Introduction:
Xp11 translocation renal cell carcinoma (tRCC) is a rare and aggressive subtype of kidney cancer with no effective therapies1–6. tRCC is defined by oncogenic fusions between the Transcription Factor binding to IGHM Enhancer 3 (TFE3) gene on the short arm of chromosome X (chrX) and any one of several recurrent partners that may be encoded on either chrX or on autosomes7. Apart from the driver fusion, tRCCs harbor few other genetic alterations8–14.
An intriguing clinical feature of tRCC is that it displays a strong female predominance, in stark contrast to the male predominance in other subtypes of kidney cancer1,8. As TFE3 is located on chrX, an open question is whether the female sex bias in tRCC incidence1 can be explained by the capacity to generate oncogenic TFE3 fusions from both chrX homologs in females, as opposed to only one chrX homolog in males.
In females, the two chrX homologs have distinct epigenetic and transcriptional states due to X-chromosome inactivation (XCI). XCI occurs during early embryonic development and achieves dosage compensation of chrX gene expression between males and females15. While either chrX homolog can be marked for XCI, once established, the active X-chromosome (chrXa) and the inactive X-chromosome (chrXi) are stably inherited in progeny cells. For cancer cells, which are generated by clonal expansion, the chrXa and chrXi identities in the ancestral tumor cell are therefore preserved in all daughter cells.
A salient feature of chrX (as opposed to autosomes) is that most chrX transcription occurs from the chrXa homolog (with the exception of a few escapee genes that are expressed biallelically). Therefore, large segmental deletions encompassing essential genes on chrXa are intolerable16,17. However, in female cells, the deletion of the same regions on chrXi could be tolerated. Therefore, the interrogation of haplotype-specific somatic chrX deletions in females provides a strategy to infer the identity of chrXi and chrXa in a cancer cell.
In this study, we leverage tRCC as a prototypical sex-biased cancer driven by an oncogene on chrX in order to investigate the largely unexplored impact of chrX genetics on cancer sex differences18,19. We provide definitive evidence that oncogenic TFE3 fusions can arise from translocations between chrXi and autosomes and elucidate their transcriptional consequences including partial reversal of XCI and silencing of juxtaposed autosomal regions.
Results:
Landscape of genomic alterations in tRCC from whole-genome sequencing
We performed whole-genome sequencing (WGS) on tumor and matched germline DNA from 15 individuals (5 males and 10 females) with tRCC. Eleven samples were collected from treatment-naïve patients and 18 samples were collected after patients had received systemic therapy for tRCC (total of 29 tumor samples). Bulk RNA-Sequencing (RNA-Seq) was performed on a subset of samples with available material. In one patient (TRCC18), primary tumor tissue was collected prior to the initiation of systemic therapy with a tyrosine kinase inhibitor (cabozantinib) and immunotherapy (nivolumab), and 13 metastatic sites were collected at autopsy. In this case, both bulk and single-nucleus RNA-Seq (sNuc-Seq) were performed on the primary tumor and bulk RNA-Seq was additionally performed on a liver metastasis (Figure 1A–B; Table S1).
Figure 1: Overview of TFE3 fusions in tRCC cohort.
(A) Study design. (B) Summary of tRCC cohort (see Table S1). (C-D) TFE3 fusion partners (C) and breakpoints (D) at the TFE3 locus in the tRCC cohort. Each fusion (one arc in C) and breakpoint (vertical line in D) is colored by sex. (E-F) Left: IGV snapshots of RNA-Seq reads showing an oncogenic fusion between ASPSCR1 (exons 1–7) and TFE3 (exons 6–10) in TRCC8 (E) and an oncogenic fusion between SFPQ (exons 1–6) and TFE3 (exons 2–10) in TRCC18 (F); right: Schematic diagram of the fusion transcript and the fusion gene sequence.
We performed variant discovery for somatic single nucleotide variants (sSNV), indels, and rearrangements. Additionally, we employed our previously described method for haplotype-specific copy-number analysis20. For tumors from TRCC18, we reconstructed the phylogeny based on a joint analysis of sSNV and haplotype-specific copy-number alterations (Methods). The average burden of sSNV was 0.88 per Mb (range 0.28–2.7, excluding FFPE samples due to the high frequency of artifacts in FFPE DNA libraries21,22) (Figure S1; Table S2). This observed SNV burden in tRCC is lower than in clear cell renal cell carcinomas23–25 but comparable to Ewing sarcoma, another translocation-driven solid tumor with a median sSNV burden less than 1 per Mb26,27. Of note, two samples (both from one individual: RCC16) that had been initially diagnosed as tRCC were subsequently inferred to be misclassified ccRCC based on the absence of a detectable TFE3 fusion but the presence of hallmark ccRCC alterations including chr3p loss (Figure S2) and inactivating VHL mutation; these two samples also showed a significantly higher sSNV burden compared to tRCCs.
We observed a broad range in the burden of somatic copy-number alterations (SCNA) across our cohort (Figure S2). While several samples displayed pervasive segmental SCNAs on a number of chromosomes (e.g. TRCC8, TRCC10, TRCC15), several samples had no SCNAs (TRCC3, TRCC11, TRCC17; notably, all three were treatment-naïve primary tumors) or only arm-level alterations (TRCC9) (Figure S2). These observations imply that TFE3 fusions are sufficient to induce tRCC development, consistent with animal models14. This also suggests that, while some tRCC samples may be genomically unstable, genome instability is not essential for cancer progression, in contrast to other cancers20,28. Consistent with this notion, we also did not identify mutations that are clear drivers of therapeutic resistance in the post-treatment sample from TRCC12 or in the metastases of TRCC18, in comparison to the paired pre-treatment specimens (Figure S2; Table S3).
The size of our WGS cohort limits power to identify recurrently disrupted cancer genes in tRCC apart from the TFE3 fusion. We assessed the prevalence of pathogenic mutations in canonical cancer genes (Tier 1 group of the COSMIC Cancer Gene Census) and in genes previously reported to be recurrently altered in tRCC8,12,13,29 (Figure S1A,B; Table S3). The most frequent alterations were deletions spanning CDKN2A (6/15 patients), including homozygous CDKN2A deletions in 4/15 cases (26%), consistent with findings reported in our prior study of tRCC8. We further identified mutations in CDKN2C and KMT2C in two samples, with at least one mutation predicted to be of high pathogenic impact. While neither of these genes have been reported to be mutated in tRCC yet, they could contribute to tRCC pathogenesis: CDKN2A/2C inactivation may be functionally similar and recurrent CDKN2C mutations have been reported in glioblastoma30,31; KMT2D mutations were previously identified in tRCC13 and KMT2C mutations may be functionally similar. Finally, one sample (TRCC15) showed biallelic inactivation of ATM, a mutation that has also previously been reported in a minority of tRCCs9.
TFE3 fusions universally arise by reciprocal translocations
We identified TFE3 fusion breakpoints from WGS data (Table S2) and validated them in RNA-Seq data when available (Methods; Figure S1C). There were seven distinct partners across 15 patients (Figure 1C), including 2 on chrX and 5 on autosomes, all of which have been previously reported8,12,13. DNA breakpoints in TFE3 were most commonly located between exons 4 and 6, which would result in fusion proteins that preserve the dimerization and DNA-binding domains of TFE3. However, in two cases (TRCC9 and TRCC18), DNA breakpoints were upstream of the first exon (Figure 1D). Analysis of RNA-Seq data from TRCC18 indicated that the expressed fusion transcript began in exon 2 of TFE3 with skipping of exon 1 due to the absence of a splice acceptor site; in the fusion partner SFPQ, the DNA breakpoint resided in exon 7 and this exon was also skipped in the fusion product (Figure 1E–F,S1C); similar exon skipping events have been reported in other cancers32,33. By interrogating other published RNA-Seq data of tRCC, we identified several additional instances of TFE3 fusions that began with exon 2 of TFE312, likely caused by translocations with breakpoints that are upstream of TFE3. In three cases (TRCC11, TRCC12, TRCC14), DNA breakpoints in the TFE3 gene were exonic but breakpoints in the partner genes were intronic. Interrogation of RNA-Seq data revealed that in TRCC12, the fusion transcript skipped the exon that contained the DNA breakpoint (exon 4). In TRCC11, the fusion transcript indicated an in-frame splicing with a splice acceptor (AG) within exon 3 that is 6 bp downstream of the DNA breakpoint (Figure S1C). A similar outcome was observed in TRCC14 with a breakpoint in exon 3 of TFE3. Altogether, these observations highlight how the intricate interplay between splicing and DNA rearrangements drives fusion gene expression in tRCC.
TFE3 fusions arose by four distinct mechanisms of rearrangement, which are all reciprocal (Figure 2A–B): (1) paracentric inversion on chrXp (2 cases); (2) pericentric inversion on chrX (1 case); (3) reciprocal chrX:autosome translocations (11 cases); and (4) chromoplexy (1 case). Notably, the chromoplexy event in TRCC15 involved reciprocal breakpoints on chr1, chrX, and a ribosomal repeat sequence mapped to chr14_GL00019v1_random that may be located on the short arm of any acrocentric chromosome (chr13, chr14, chr15, chr21, or chr22). This case also contained multiple other reciprocal translocations that may be related to ATM inactivation34. In contrast to oncogenic fusions in other cancers35–40, no TFE3 fusion arose from complex/unbalanced rearrangements. Moreover, copy-number alterations to either chrX or the partner chromosomes were rare. A notable exception was TRCC18 where we detected complex SCNAs on the partner chromosome (chr1p) in a subset of metastatic samples; the absence of these alterations in other metastatic lesions or the primary tumor suggests that the TFE3 fusion was still generated by a simple reciprocal translocation in the common ancestor.
Figure 2: Etiology of TFE3 fusions revealed by whole-genome sequencing.
(A) Balanced DNA copy number across chrX and chr17 in TRCC14 (LUC7L3-TFE3 fusion). Fusion partners positions are marked by vertical lines and are joined by reciprocal translocations (red arcs), with allelic balance of both homologs as indicated by the complete overlap of black and gray dots corresponding to allele A and allele B copy number. (B) Rearrangement types giving rise to TFE3 fusions: (Left) Intra-chrX inversions: paracentric (inversion within the chrXp arm) and pericentric (inversion spanning the centromere). (Middle) Inter-chromosomal translocations. chrX:autosome translocations with partners on autosomal p-arm (e.g. SFPQ) or q-arm (e.g. PRCC); (Right) chromoplexy, i.e., balanced translocations between three or more pairs of break ends (see Table S2). (C) Sequence features of non-homologous end-joining at TFE3 and partner loci illustrated by the TFE3-RBM10 fusion in TRCC7. Left: minimum DNA loss or gain at each partner locus (no DNA loss or gain at the TFE3 breakpoint and 3bp deletion at the RBM10 breakpoint). Right: minimum homology or insertion at rearrangement junctions (2bp microhomology at the RBM10-TFE3 junction and 3 bp insertion at the TFE3-RBM10 junction). (D) Summary of DNA loss/gain at TFE3 and fusion partner loci and junction homology across the tRCC WGS cohort. The x- and y-axes indicate the number of duplicated (+) or deleted (−) nucleotides at breakpoints in TFE3 (y) and the TFE3 translocation partner (x). The upper right and lower left triangles of each square show the classification of microhomology or insertion at reciprocal junctions (Partner-TFE3 or TFE3-Partner). (E) Mechanisms that result in no change (i), deletion (ii), or duplication (iii) between opposite break ends generated by a double-strand break. (F) Histogram showing the length of microhomology (left) or insertion (right) at the junctions between TFE3 and Partner breakpoints (including both reciprocal junctions). An outlier with 24bp insertion at the TFE3-MED15 junction detected in TRCC10 was the only instance with a significant deletion at both TFE3 (27bp) and the partner (MED15, 6bp), potentially indicating microhomology-mediated end-joining.
In summary, the DNA copy-number and rearrangement data from our WGS cohort suggest that TFE3 fusions universally arise by reciprocal translocations.
TFE3 fusion junctions are consistent with canonical non-homologous end joining
The preservation of both reciprocal translocations in TFE3 fusion events enabled us to directly analyze breakpoints and junctional homology at the TFE3 locus and at the partner locus (Figure 2C). We observed minimal DNA deletion or duplication at both the TFE3 locus and the partner loci, including at all three loci in the case of chromoplexy (TRCC15). Among 31 pairs of reciprocal breakpoints (3 in TRCC15 and 2 in each of the other 14 cases), 22 had deletions or duplications of no more than 10 base pairs; the largest deletion was 71bp (ASPSCR1 in TRCC8) and the largest duplication was 38bp (PRCC in TRCC12). The absence of significant deletion or duplication between reciprocal breakpoints (Figure 2D) suggest that they were derived from DNA ends that were either blunt or near blunt (Figure 2E)41.
Blunt DNA ends are most commonly repaired via canonical non-homologous end joining (c-NHEJ)41,42. Consistent with this prediction, 26/31 (84%) of junctions displayed ≤ 2bp of microhomologies or insertions (Figure 2F), as expected for c-NHEJ but not alternative end joining (a-EJ) or single-strand annealing (SSA). Both the absence of sequence changes between reciprocal breakpoints and the absence of sequence homology at rearrangement junctions are consistent with prior observations of experimentally generated translocations by nucleases43–45. To further demonstrate that TFE3 fusions arise from c-NHEJ of double-strand breaks (DSBs), we used CRISPR/Cas9 to generate DSBs at TFE3 and PRCC (a recurrent TFE3 fusion partner located on chr1q) in 293T human embryonic kidney cells (Methods; Table S4). Sequencing of the amplicons across DNA breakpoints in the resultant fusions revealed a tight distribution of small insertions and deletions (median indel size: PRCC-TFE3: 3 bp, TFE3-PRCC: 3.5 bp) that recapitulates observations in primary tRCCs. The nearly identical sequence features of TFE3 fusions in primary tRCCs and experimentally engineered fusions together support an origin of TFE3 fusions from c-NHEJ of near blunt DSBs (Figure S3),
In summary, TFE3 fusions exemplify the canonical model of reciprocal translocations from DSBs and c-NHEJ and our results suggest that the same mechanism underlies chromoplexy.
Deletions of chrX upon translocation implicate a chrXi origin for TFE3 fusions
The universal generation of reciprocal TFE3 fusions raises the question of whether both fusions (Partner-TFE3 and TFE3-Partner) are required for oncogenesis. This is disfavored by the observation of subclonal deletions of the der(X) translocation (which encodes the TFE3-Partner fusion) in TRCC9 and TRCC18 (Note: Derivative chromosomes are named based on the centromere that is preserved, c.f. Figure 2B). Additionally, breakpoints are located upstream of the TFE3 gene body in both of these cases and so the reciprocal fusion genes do not contain coding sequences from TFE3 (Figure 1D).
An alternative possibility is that the preservation of both translocated chromosomes is not due to positive selection of both fusions but due to negative selection against deletions of chrX. For males with only one chrX, deletion of either translocated chromosome would be intolerable, as it would result in the deletion of essential genes on chrX16. In females with one active and one inactive chrX, if translocations originate from chrXa, the same constraint against deletions applies. By contrast, deletions of translocations involving chrXi could be tolerated since most genes from chrXi are not actively transcribed. However, for oncogenic TFE3 fusions to arise from chrXi translocations, the TFE3 gene on chrXi must be fused to a partner gene on an autosome with an active promoter. Therefore, the distinct epigenetic states of chrXi and chrXa impose different constraints on both the translocation partner and the copy-number status of TFE3 translocations (Figure 3A).
Figure 3: Female predominance in chrX:autosome translocations underlies sex bias in tRCC.
(A) Sex-specific constraints on each class of chrX rearrangement leading to TFE3 fusions. (B) Summary of aggregate tRCC cohort assembled from 11 independent studies. Right, sex distribution of tRCC cases in the aggregate cohort (N = 203). (C) Sex disparity of chrX:autosome translocations and intra-chrX inversions resulting in oncogenic TFE3 fusions: chrX:autosome translocations display a 2:1 female-to-male ratio while intra-chrX inversions are enriched in males. The enrichment of intra-chrX inversions in males is largely attributed to RBM10-TFE3 fusions, which may be related to the higher mutation frequency of RBM10 in males49. (D) Landscape of copy-number imbalance of chrX and TFE3 fusion partner chromosomes broken down by sex and rearrangement type. Copy number imbalance resulting from deletions on chrX (blue) is exclusive to females. Other types of copy number imbalance (green), including chrX gains and gains/losses of the partner chromosome are seen in both females and males. See Figure S4. (E) An example of copy-number imbalance due to deletion of the translocated chromosome der(X)t(1;X) containing the reciprocal TFE3 fusion. This example is a tRCC sample profiled via whole-exome sequencing in a previous study12. Shown are the minor allele fraction (top) and normalized copy ratio (bottom) of tumor DNA. Data points in the deleted regions are orange.
By a systematic analysis of TFE3 fusions and the copy number of translocated chromosomes, we sought to test the hypothesis that some TFE3 fusions may originate from chrXi, thus providing a genetic explanation for the female predominance of tRCC. We assembled a cohort of 203 tRCC cases from 11 different studies (Figure 3B; Table S5; Methods), including 15 cases from the current study and 35 cases that had originally been misclassified as other types of kidney cancer. This cohort, the largest to date, shows an approximately 3:2 female-to-male ratio (61.1% females [N=124]; 38.9% males [N=79]) (Figure 3B). We then separated TFE3 fusions with partners on chrX from those with autosomal partners. This revealed a near-perfect 2:1 ratio (103 females, 50 males) amongst chrX:autosome fusions in contrast to a male dominance (21 males, 8 females) amongst TFE3 fusions arising from chrX inversions (Figure 3C). When excluding fusions involving RBM10, which have previously been reported to show a male bias in mutation frequency49, there is no significant sex bias of chrX inversions (11 males, 7 females; P=0.24 by binomial test). The absence of sex bias amongst fusions from chrX inversions is consistent with the expectation that fusions generated by chrXi inversion would remain transcriptionally silenced and are therefore not functional. By contrast, the 2:1 female-to-male ratio of chrX:autosomal translocations provide a compelling argument for both chrXa and chrXi translocations contributing to oncogenic TFE3 fusions in females.
To test the hypothesis that TFE3 fusions arising from chrXi can tolerate copy-number losses of the translocated chromosome encoding the non-oncogenic fusion transcript (TFE3-Partner), we analyzed DNA copy number in 160 cases with whole-exome or whole-genome sequencing using a harmonized workflow (Methods). While a vast majority of TFE3 translocations are devoid of copy-number alterations or imbalance on both chrX and the partner chromosome, deletions spanning the breakpoint on the reciprocal TFE3 fusion were found in 6/141 female cases (“Unbalanced with Deletion” in Figure 3D; an example shown in Figure 3E). We also identified 11 instances of copy-number imbalance due to alterations other than chrX deletions (Figure S4), including copy-number gains of either translocated chromosome (Figure S4A,B) or losses on the autosomal partner chromosome without copy-number losses on chrX (Figure S4D). The latter instance may reflect losses on one or multiple translocated chromosomes generated by chromoplexy.
In summary, sex differences in the incidence of autosomal:TFE3 fusions and sex-specific patterns of copy-number alterations associated with TFE3 translocations implicate oncogenic TFE3 fusions as arising from both chrXa and chrXi in females. This ascribes a genetic basis to the female predominance of tRCC.
Paired genomic and transcriptomic analyses establish chrXi involvement in TFE3 fusions
We next turned our attention to TRCC18, where we observed secondary copy-number losses on the chr1p-terminus telomeric to the translocation partner (SFPQ) in multiple metastatic lesions. We inferred the phylogeny of the metastatic lesions based on sSNVs and refined phylogenetic inference using copy-number alterations (Figure 4; Table S6; Methods). sSNVs were assigned to phylogenetic branches after accounting for DNA deletions (Figure 4B–C; Figure S5; Figure S6; Table S6). The sSNV-derived phylogeny corroborated the timing of initiating events (i.e. oncogenic TFE3 fusions) and subsequent events (i.e. whole-genome duplication and deletion of TFE3 translocations).
Figure 4: tRCC evolution revealed by multiregional sequencing.
(A) Treatment history and anatomic sites of 14 tRCC samples from subject TRCC18. (B) Summary of truncal (left), shared (middle), and private (right) point mutations detected across all tumors. Presence/absence of mutations in each sample was determined by the mutant allele frequency (MAF) cutoff estimated from truncal mutations. See Methods and Figure S5 for more details. (C) Phylogenetic tree of metastatic lesions inferred from shared mutations in (B). Two samples inferred to have near tetraploid genomes (Liver-4 and MedLN) are underlined. Each mutation was assigned to a branch (a-k) based on the maximum likelihood calculated from a constant false positive rate (0.01) and sample-specific false negative rate estimated as described in Figure S5, with the final mutational burden shown on the right. Note that the three groups of mutations assigned to branch b in panel B are almost all due to the downstream deletion of der(X)t(1;X) in the Lung metastases and in RPLN-2; these lost mutations are ancestral. Similar patterns of mutations were found on chr4, chr11, and chr22 that are deleted in MedLN/Liver-4, Lung, and RPLN-2/RPLN-3 samples. See Table S6 for details about mutation assignment considering DNA deletions and losses.
The SFPQ:TFE3 fusion was generated by a reciprocal translocation between chrX and chr1 (Figure 5A) and both translocations were preserved in five liver metastases (Liver-1,2,3,5,6) and two metastases involving the retroperitoneal lymph nodes (RPLN-1 and RPLN-3). Therefore, the reciprocal translocation was truncal. By consequence, deletions that led to copy-number imbalance between the oncogenic translocation, der(1)t(1;X)(p34.2;p11.23) [hereafter, der(1)t(1;X)], and the reciprocal translocation, der(X)t(1;X)(p34.2;p11.23) [hereafter, der(X)t(1;X)] in the remaining samples were secondary (Figure 4C; Figure S5D).
Figure 5: Deletions and losses of der(X)t(1;X) in metastases from TRCC18 indicate that the TFE3 fusion arose via a chr1:chrXi translocation.
(A) Summary of ancestral copy-number alterations and rearrangements in all tumor biopsies using data from Liver-2. Left: CIRCOS plot showing haplotype-specific DNA copy number (blue and red) and rearrangements (green for intra-chromosomal, purple for inter-chromosomal), for chromosomes with copy number alterations or rearrangements. The reciprocal chr1:chrX translocations are highlighted as a thick magenta line. Right: Distributions of allelic depths of chromosomes with altered DNA copy number (chrs.6,8,9,14,17, and 18). ‘A’ refers to the homolog with altered DNA copy number and ‘B’ to the intact homolog; for chr9, the A homolog shows focal deletions and the B homolog undergoes complete deletion. The allelic depths of the deleted 9B homolog (CN=0), the gained 8A homolog (CN=2), and regions with normal DNA copy number (CN=1) determine the integer copy-number states and the clonal fractions of subclonal copy-number alterations (e.g., a minor subclonal loss of the 4B homolog in the Liver-2 sample). See Figure S6 for other instances of copy-number evolution in this case. (B-D) CIRCOS and allelic-depth distribution plots of selected chromosomes showing downstream evolution after ancestral alterations shown in A. (B) Lung-1 (shown as the representative of all three lung metastases) displays a near clonal deletion of most of der(X)t(1;X) (see Figure S6E), a near clonal deletion of 22A, and gains of 17A and 18A. (C) MedLN (shown as the representative of MedLN and Liver-4) is inferred to be near tetraploid (4N) based on the presence of 4A, 7A, and 11A at the median copy-number state (CN=1) between 22A (CN=0) and the basal copy-number state (allelic depth=1; CN=2). Based on this inference, der(X)t(1;X) underwent whole-chromosome loss after whole-genome duplication. (D) RPLN-2 has several chromosomes (4A, 13A, 20A, 22A) showing non-integer copy-number states. Although we cannot determine the number of subclones or their ploidy states, the mean allelic depths of der(X)t(1;X), 4A, and the p-arm of 20A are all below the median between complete deletion (CN=0 inferred from 9B) and the basal copy number state (CN=1 in 2N genomes or CN=2 in 4N genomes); this reduction cannot be solely attributed to chromosome losses after whole-genome duplication, but implies complete chromosome deletion in one or more subclones. (E) Box plots of allelic fractions of somatic mutations on chrXq in all metastatic biopsies. Each mutation is phased to either the retained (open boxes) or the deleted (filled boxes) homolog based on their allelic fractions in all samples (Figure S5D). The Lung-1 and Lung-3 samples show complete deletion of mutations on the deleted homolog. Liver-4 and MedLN show lower allelic fractions of these mutations consistent with losses but not deletion; the 2:1 ratio of the average allelic fraction of mutations on the retained versus those on the deleted homolog is also consistent with the 2:1 copy-number ratio between the two homologs. Lung-2 and RPLN-2 are both polyclonal. The mean allelic fraction of mutations in each sample is consistent with the clonal fraction of tumor cells estimated from the allelic depth of 9B that is inferred to be deleted in the founding clone. The ratio between the average allelic fraction of mutations on the retained versus those on the deleted homolog in MedLN and Liver-4 is also consistent with a 2:1 copy-number ratio between the two homologs.
We further inferred that the loss or deletion of der(X)t(1;X) occurred independently in different branches (Figure 5B–D; Figure S6E). In the clade of lung metastases that are all near-diploid, we determined that there was complete deletion of der(X)t(1;X) containing the entire chrXq arm (Figure 5B). In the Liver-4/mediastinal lymph node (MedLN) clade (Figure 5C), we inferred a single-copy loss of der(X)t(1;X) after whole-genome duplication. We further inferred the presence of a dominant near-diploid subclone in the RPLN-2 metastasis (Figure 5D) with near complete deletion of der(X)t(1;X). The loss or complete deletion of der(X)t(1;X) was independently validated by the allelic fractions of somatic mutations phased to the deleted chrXq (Figure 5E).
While loss of chrXq after WGD could be tolerated regardless of whether the loss was of chrXa or chrXi, the complete deletion of chrXq in the 2N lung metastases and the near complete deletion in a dominant clone in RPLN-2 provided strong evidence that the translocation was generated on chrXi. To definitively demonstrate this, we assessed haplotype-specific expression of genes on chrXq in the primary tumor and a liver metastasis (liver-2), where both chrXa and chrXi were preserved (Figure 6A), using the haplotype phase determined based on allelic imbalance in a lung metastasis (Lung-1). The chrXi origin of the translocation was verified by near complete mono-allelic expression of the retained haplotype (chrXa), with an exception being the XIST gene (which is monoallelically expressed from chrXi).
Figure 6: chrXi:autosome translocations can lead to autosomal silencing and chrXi reactivation.
(A-C) Haplotype-specific transcription of chrX (A), chr1 (B), and chr7 (C) in the primary tumor and the Liver-2 metastasis of TRCC18. The first row shows allelic imbalance in Lung-1 (chrX) and MedLN (chr1 and chr7) that was used to determine the haplotype phase of these chromosomes. Rows 2 and 3 show the phased DNA allelic depths and RNA allelic fractions in the primary tumor; rows 4 and 5 show the DNA allelic depths and RNA allelic fractions in the Liver-2 metastasis. Note that the monoallelic expression on chrXq establishes HapA (altered homolog) as chrXi and this is further corroborated by the monoallelic expression of XIST solely from chrXi. Allelic imbalance telomeric to the SFPQ locus in chr1 indicates silencing of the A homolog that is in cis with XIC on the translocated chrXi (P=1.2e-2 in the primary and P=1.78e-5 in Liver-2 by a Mann-Whitney test comparing the allelic fractions of transcripts on either side of the breakpoint). Biallelic transcription of chr7 serves as a negative control. (D) Predicted transcriptional consequences of chrXi:autosome translocations. (E) Partial reactivation of chrXi expression revealed by an increase in biallelic Xp expression. Shown are the cumulative fractions of biallelically expressed polymorphic sites in comparison to the cumulative fractions of all polymorphic sites on all disomic chromosomes (from p-terminus to q-terminus). (F) Partial reactivation of chrXi expression inferred by an increase in total transcriptional output in tumor cells, as assessed by single nucleus RNA sequencing. Shown is the average ratio of transcripts in tumor nuclei (N=4118) relative to normal nuclei (N=172) from the TRCC18 primary tumor in different chromosomal regions. A statistically significant difference is only observed for the Xp region that is telomeric to the breakpoint (P = 6.3e-3; Mann-Whitney U test) but not the Xq region (P = 0.48) or disomic autosomes (P = 0.79). (G) Additional examples of chrXi reactivation inferred from increased biallelic expression of genes telomeric to the TFE3 breakpoint. Shown are the cumulative fractions of biallelically expressed heterozygous genotypes (red) and of all heterozygous sites in expressed genes (black) on chrX for selected female samples of the following categories: (top row) three tRCC samples with autosome-TFE3 fusions and chrXp reactivation (P < 0.05 by Mann-Whitney U test); (bottom row from left to right) a tRCC sample, a ccRCC sample, and a tRCC cancer-adjacent normal sample, all showing no evidence of reactivation (P > 0.05). See Figure S7 for additional examples. (H) Summary of 15 female tRCC samples with or without biallelic chrXp expression as assessed by RNA-Seq.
Together, these data unambiguously demonstrate that chrXi was involved in the oncogenic TFE3 fusion in TRCC18 and further show that the non-functional translocation der(X)t(1;X) from chrXi underwent frequent secondary losses and deletions during tumor evolution.
Transcriptional consequences of Xi:autosomal translocations in tRCC
Having identified a chrXi:chr1 translocation in TRCC18, we sought to further analyze the transcriptional and epigenetic consequences of chrXi:autosomal translocations. X-chromosome silencing is orchestrated by the cis-spreading of the non-coding RNA XIST. It has been reported that XCI may extend to autosomal regions in autosome:chrXi translocations50 and that ectopically inserted XIST can also induce epigenetic silencing of entire autosomes in experimental systems51,52. Consistent with these observations, we detected reduced transcription in the chr1 region to the left of the translocation breakpoint (Figure 6B, Primary and Liver-2). The reduction was specific to der(X)t(1;X) based on the haplotype determined from allelic imbalance in the MedLN metastasis. The biallelic expression pattern on a diploid autosome (chr7) with allelic balance in both the primary tumor and the liver-2 metastasis serves as a negative control for this analysis (Figure 6C).
The expression of oncogenic TFE3 fusion transcripts from der(1)t(1;X) implies that the silenced chrXp arm is at least partially reactivated (Figure 6D), in contrast to the stable maintenance of XCI with normal somatic cell division15. To determine the extent of X-reactivation, we assessed allelic expression of genes subject to X-inactivation (non-escapees). We observed a significant increase of biallelic gene expression on the chrXp arm telomeric to the TFE3 breakpoint in comparison to the chrXq arm (Figure 6A; Primary: P=1.05e-3, Liver-2: P=9.04e-15; Mann-Whitney U test). We further performed this analysis across all disomic chromosomes and found that only chrX displayed a positional bias in biallelic expression, with the transition point located precisely at the TFE3 breakpoint (Figure 6E).
To further demonstrate partial reactivation of chrXi, we performed single nucleus RNA-Seq of the TRCC18 primary tumor (4,290 cells). As X-inactivation is clonal in tumor cells (i.e. the same chrX homolog inactivated in all tumor cells) but mosaic in adjacent normal cells (different homologs inactivated in different cells), we compared total gene expression from chrX in tumor cells to normal cells. Consistent with the prediction of increased total transcription due to partial chrXi reactivation, we observed significantly higher transcriptional output in tumor cells relative to non-tumor cells for the chrXp arm, but not chrXq or disomic autosomes (Figure 6F). The observations of increased biallelic expression and increased total transcription output on chrXp but not chrXq provide orthogonal evidence for the partial reactivation of genes located on the telomeric side of the TFE3 breakpoint in der(1)t(1;X).
Having established biallelic expression on chrXp as a signature of chrXi reactivation in tRCC, we sought to use this feature to detect chrXi:autosome translocations directly from tRCC bulk RNA-Seq data (Methods). We restricted this analysis to RNA-Seq data generated from frozen samples from female tRCCs, all with autosomal fusion partners (FFPE samples were excluded due to lower quality of allelic expression data53,54; N=15 total from our cohort or external cohorts). In 9/15 (60.0%) cases, we observed evidence of chrXp reactivation (Figure 6G–H; Table S5) as indicated by an enrichment of biallelic expression telomeric to the TFE3 breakpoint (Figure S7A; P<0.05 by Mann-Whitney U test) and by increased transcriptional output of reactivated genes in chrXi-rearranged samples (Figure S7B; TCGA: P=2.71e-4, Institutional cohort: P=4.54e-6 by Mann Whitney U-test). Importantly, we did not observe any significant enrichment of biallelic expression in either adjacent normal samples (which display consistent biallelic expression across chrX due to mosaicism of XCI), female ccRCC samples from TCGA, or the remaining six female tRCCs (Figure 6G; Figure S7A). We note that the assessment of chrXp reactivation by allelic expression in bulk tumor RNA-Seq may be confounded by the presence of normal cells (which have a mixture of chrXi and chrXa alleles) and is also limited by the number of reactivated genes with informative SNVs. We therefore explored several orthogonal strategies to infer whether chrXi or chrXa were rearranged in tRCC and found them to be highly consistent with the use of biallelic chrXp expression as a metric (Table S5 and caption).
Finally, given evidence for the cis-spreading of chrX silencing to autosomal regions in the TRCC18 case (Figure 6B), we interrogated samples inferred to have chrXi translocations as described above, and identified potential autosomal silencing in TRCC17 and another case in the external cohort (Figure S8A).
In summary, our RNA-Seq analysis demonstrated that chrXi:autosomal translocations can lead to long-range epigenetic changes including both partial silencing of the autosomal region that is in cis to the X inactivation center on chrXq and partial reactivation of chrXp that becomes fused to an autosome (Figure 6D).
Epigenetic changes due to chrXi rearrangements in tRCC
Under normal conditions, X-inactivation is stable and a majority of the genes on chrXi are epigenetically silenced via DNA methylation in their promoter region55. To determine whether the chrXi:autosomal translocations are capable of reversing these stable and heritable epigenetic marks, we interrogated 9 tRCC samples from TCGA cohort (originally misclassified as ccRCCs) that have DNA methylation, RNA-Seq, and WES data. Using our method above, we inferred that four samples had chrXa translocations and five had chrXi translocations.
We observed no significant difference in DNA methylation at the promoters of non-escaping genes to the left versus to the right of the TFE3 breakpoint in all samples, regardless of chrXi/chrXa rearrangement status (Figure S7C); notably, differences in chrX methylation between males and females on this assay are typically readily apparent56. As a recent study suggested that a subset of chrX genes with promoter hypomethylation are most XIST-dependent57, we next assessed DNA methylation specifically at promoters of reactivated genes. Indeed, in tumors displaying chrXp reactivation, reactivated genes on chrXp (defined as RNA MAF ≥ 0.2) were flanked by relatively hypomethylated regions, as compared to non-reactivated genes (MAF < 0.2) (P=8.69e-4 by Mann-Whitney U test, Figure S7D). Overall, our data suggest that X-reactivation due to somatic chrXi rearrangement may preferentially impact loci with a lower basal level of methylation, leading to increased expression of a subset of chrXp genes.
chrXi rearrangements in other tumor types
The discovery of chrXi rearrangements underlying oncogenic TFE3 fusions prompted us to investigate whether similar mechanisms may be active in other cancers. We first reviewed the sex distribution of TFE3 fusions in other rare tumor types – most notably alveolar soft part sarcoma (ASPS, typified by ASPSCR1-TFE3 fusions)58 and YAP1-TFE3 hemangioendothelioma59. Both tumor types display a near 2:1 female sex bias, suggesting that the same mechanisms ascribed to TFE3 rearrangements in tRCC apply to these distinct malignancies. Moreover, we identified cell line models of both tRCC and ASPS that displayed features suggestive of chrXi rearrangements and X-reactivation (Figure S8B–D). Notably, the more pronounced female bias in these cancers than tRCC may be explained by the predominance of autosomal partners in these cancers (rather than a mix of autosomal and chrX partners, as in tRCC).
We further noted that a rare subtype of neuroblastoma driven by activation of the FOXR2 gene on chrXp (CNS NB-FOXR2) has been reported to have a strong female sex bias in multiple cohorts (26/44 [59%] females60; 8/11 [73%] females61; 15/20 [75%] females62). In CNS NB-FOXR2, FOXR2 can be activated through amplification and/or complex structural rearrangements that result in fusion transcripts retaining the C-terminal region of FOXR2 driven by juxtaposed promoters from fusion partners60,62,63. Notably several cases of CNS NB-FOXR2 have been reported to have large segmental deletions on chrX (c.f. Figure S5 from Sturm et al.60); by the same reasoning of essentiality of chrX-linked genes16,17, these deletions likely occurred on chrXi.
Rearrangements of chrXi and disruption of chrXi silencing also occur in other cancers. In our recent study of esophageal cancers20, we identified segmental alterations on both chrXa and chrXi based on haplotype-specific DNA copy number alterations. The interrogation of RNA-Seq data from TCGA and the Cancer Cell Line Encyclopedia (CCLE) also revealed examples of biallelic chrX expression accompanied by chrX rearrangements64 (Figure S8E–F). Together, these observations indicate that oncogenic chrXi translocations may contribute to female sex bias in multiple fusion-driven cancers; and both chrXi rearrangements and the consequent epigenetic/transcriptional changes occur in other cancer types beyond tRCC.
Discussion:
Our integrative genomic and transcriptomic analyses of tRCC in combination with the interrogation of data from multiple prior studies8,12,13,65 provides insight into the etiology of TFE3 fusions and reveals important features of chrX rearrangements.
We found that TFE3 fusions almost uniformly arise via reciprocal rearrangements with minimal loss or gain of genetic material at break points on each chromosome involved in the rearrangements (including in a case of chromoplexy). This contrasts with oncogenic fusions in other tumors (e.g. EWSR1-ETS fusions in Ewing sarcoma26, TMPRSS2-ERG fusions in prostate cancer66, the EML4-ALK fusion in lung cancer67,68, or BCR-ABL fusions in chronic myeloid leukemia36–40), where a varying fraction of the fusions arise from chromoplexy, chromothripsis, or other complex rearrangements involving multiple chromosomes26,35,66,69,69. Our observation of minimal microhomology/insertion and DNA loss/gain at reciprocal fusion breakpoints suggests an origin in c-NHEJ after near-blunt double-strand breaks42,70. The precise mechanisms that generate near blunt DSBs in tRCC precursor cells in vivo remain to be elucidated. This observation also raises the question of whether cancers in which fusions arise via reciprocal translocations or chromoplexy with 10–100kb deletions between paired break ends are subject to different mechanisms of DNA breakage and repair26,35,66.
We suggest that frequent preservation of the reciprocal derivative chromosome encoding the non-oncogenic TFE3 fusion reflects constraints imposed by chrX transcription. Due to monoallelic transcription from chrXa in both males and females and the presence of many essential genes on chrX16, large deletions on chrXa are intolerable (although losses of chrXa after whole-genome duplication may be tolerated). By contrast, segmental or whole-chromosome deletions of chrXi are tolerable and may occur frequently in female cancers. Given the drastically different consequences of copy-number alterations on chrXa vs. chrXi, haplotype-specific copy-number analysis is essential to uncovering the functional implications of alterations involving chrX. The interrogation of chrXi or chrXa alterations requires long-range haplotype information that cannot be directly obtained from short-read sequencing data. Our study demonstrates several strategies to overcome this difficulty by combinations of statistical phasing, allelic imbalance, and monoallelic chrX expression in tumor clones. These approaches are generalizable to other cancers and their incorporation into future cancer genomic analyses will generate further knowledge about how chrX alterations influence cancer evolution.
We provide definitive evidence that TFE3 fusions can arise from chrXi and that such translocations result in at least partial reactivation of the p-arm of chrXi. These findings are in line with recent studies reporting somatic perturbation of XCI in cancer cells56,71,72. Moreover, the fact that chrXi can be recurrently accessed for TFE3 fusions through chrXi:autosome translocations presents a genetic explanation for the observed female sex bias in tRCC, with a striking 2:1 female-to-male ratio of chrX:autosome translocations. Many cancers show differences in incidence or outcomes between males and females that cannot be fully explained by differences in lifestyle, environment, or hormonal factors alone73,74. We propose that chrXi:autosomal translocations represent a female-specific oncogenic process that may underlie female predominance in other cancers driven by X-linked oncogenes58–60. More broadly, baseline genetics of the sex chromosomes likely underlie sex-specific phenotypes across cancers.
We also show that chrXi:autosomal translocations are linked to partial silencing of autosomal genes in some cases. The latter class of events may promote tumorigenesis by transcriptional repression of autosomal tumor suppressor genes. For example, we found that many tRCCs with SFPQ-TFE3 fusions have deletion of the chr1p region. This region encompasses ARID1A, a SWI/SNF component and tumor suppressor gene whose haploinsufficiency and/or alteration has been linked to tumorigenesis in kidney cancer and other cancer types72–75. We speculate that ARID1A downregulation via chr1:chrXi fusions may be another mechanism of inactivation of this tumor suppressor gene, that could occur concurrently with the generation of SFPQ-TFE3 fusions. It is also conceivable that translocations involving chrXi may be co-opted in the inactivation of other tumor suppressor genes across cancer types through long-range epigenetic silencing79, although the prevalence of this phenomenon remains to be systematically explored.
Prior studies in experimental systems have shown that disruption of the X-inactivation center (XIC) usually has modest effects on maintaining XCI, though the phenotype varies based on context57,80–82. tRCC represents a physiologic context in which a large segment of chrX is physically disconnected from the XIC while autosomal segments are brought in cis with the XIC. tRCC therefore represents a genetically defined model in which to study the mechanisms underlying somatic plasticity of XCI in cancer. Given the distinct features of chrXi, including its epigenetic state, chromatin ultrastructure, physical location and DNA replication timing83–87, it is also conceivable that somatic chrXi rearrangements may evince selective dependencies. More broadly, somatic rearrangements involving chrXi may underlie sex-specific transcriptional differences in other cancers.
Limitations of the Study:
The inference of chrXa/chrXi translocation based on biallelic transcription may be influenced by sample quality and by normal tissue contamination (as XCI is mosaic in normal tissues), amongst other factors. This was mitigated by prioritizing samples with higher tumor purity, frozen tissue-derived data, and by validating results in cell line models and by validation with genomic sequencing data. Future whole genome sequencing studies that span cancer types and incorporate multiple modalities (including DNA methylation and other epigenetic marks) will be required to comprehensively define the implications of chrX rearrangements across cancers. Additionally, it is difficult to quantitatively assess the extent of and mechanisms underlying X-reactivation in primary tumor samples, and this will require future experimental studies.
STAR METHODS
RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact Srinivas R. Viswanathan (srinivas_viswanathan@dfci.harvard.edu).
Materials availability
This study did not generate any new reagents. Material availability is specified in the key resources table.
Key resources table
| REAGENT or RESOURCE | SOURCE | IDENTIFIER |
|---|---|---|
| Antibodies | ||
| Rabbit anti-TFE3 mAb | Cell Signaling Technology | #14779 |
| Rabbit anti-β-Actin (13E5) mAb | Cell Signaling Technology | #4970 |
| IRDye 680LT goat anti-Rabbit IgG | LI-COR | #926–68021 |
| Bacterial and virus strains | ||
| N/A | ||
| Biological samples | ||
| N/A | ||
| Chemicals, peptides, and recombinant proteins | ||
| SuperScript IV | Thermo Fisher Scientific | 18090010 |
| SYBR™ Green PCR Master Mix | Thermo Fisher Scientific | 11756050 |
| Critical commercial assays | ||
| KAPA Hyper Prep with Library Amplification Primer Mix | Roche | KK8504 |
| AllPrep DNA/RNA Mini Kit | Qiagen | 80204 |
| RNeasy Plus Mini Kit | Qiagen | 74316 |
| Deposited data | ||
| dbGAP | This study | phs003008.v1.p1 |
| SRA (WES) | This study | PRJNA865001 |
| GEO | Separate study | GSE269466 |
| GEO | Separate study | GSE266517 |
| Public Data | ||
| TCGA cohort | TCGA (GDC portal) | https://gdc.cancer.gov/about-data/publications/pancanatlas |
| Sun et al. cohort | Sun et al.12 | https://www.biosino.org/node/project/detail/OEP002535 |
| Qu et al. cohort | Qu et al.13 |
https://ngdc.cncb.ac.cn/gsa-human/browse/HRA002855
https://ngdc.cncb.ac.cn/gsa-human/browse/HRA003190 |
| PCAWG cohort | PCAWG106 | https://dcc.icgc.org/releases/PCAWG |
| Sturm et al. cohort | Sturm et al.60 |
https://ega-archive.org/datasets/EGAD00001001927 GEO: GSE73801 |
| Cancer Cell Line Encyclopedia | CCLE (GDC legacy archive, moved to EGA) | https://www.omicsdi.org/dataset/ega/EGAD00001001039 |
| Malouf et al. cohort | Malouf et al.10 | https://clincancerres.aacrjournals.org/content/20/15/4129.long |
| MSK cohort (tRCC) | MSK9 | https://clincancerres.aacrjournals.org/content/early/2020/05/28/1078-0432.CCR-19-3283.figures-only?versioned=true |
| Sato et al. cohort | Sato et al.109 | https://ega-archive.org/studies/EGAS00001000509 |
| Human Reference Genome NCBI build 38, GRCh38 | Genome Reference Consortium | http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/human/ |
| Experimental models: Cell lines | ||
| 293T | ATCC | CRL-3216 |
| 786-O | ATCC | CRL-1932 |
| S-TFE | RIKEN BioResource Research Center (BRC) Cell Bank | RCB4699 |
| UOK146 | Dr. W. Marston Linehan’s laboratory (National Cancer Institute) | N/A |
| UOK109 | Dr. W. Marston Linehan’s laboratory (National Cancer Institute) | N/A |
| ASPS-1 | Dr. R. H. Shoemaker’s laboratory (National Cancer Institute) | N/A |
| ASPS-KY | Dr. K. Shinoda’s laboratory (Niigata University) | N/A |
| Experimental models: Organisms/strains | ||
| N/A | ||
| Oligonucleotides | ||
| Primers, sgRNAs, and shRNAs | This study | Table S4 |
| Recombinant DNA | ||
| pLentiCRISPRv2 | Addgene | 52961 |
| Software and algorithms | ||
| BWA v0.7.17 | Li and Durbin121 | https://bio-bwa.sourceforge.net/ |
| Eagle2 | Loh et al.99 | https://alkesgroup.broadinstitute.org/Eagle/ |
| Longranger 2.2.0 | 10x Genomics | https://support.10xgenomics.com/genome-exome/software/pipelines/latest/installation |
| GATK4’s Picard tools | GATK | https://github.com/broadinstitute/picard |
| GATK4’s Mutect2 v4.3.0 | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570422171291-Mutect2 |
| GATK4’s CalculateContamination | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570322332315-CalculateContamination |
| GATK4’s GetPileupSummaries | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570416554907-GetPileupSummaries |
| GATK4’s FilterMutectCalls | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570331605531-FilterMutectCalls |
| GATK4’s SelectVariants | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570332289307-SelectVariants |
| GATK4’s FilterAlignmentArtifacts | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570389354011-FilterAlignmentArtifacts-EXPERIMENTAL- |
| GATK4’s CollectReadCounts | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570293421723-CollectReadCounts |
| GATK4’s HaplotypeCaller | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570334998171-HaplotypeCaller |
| GATK4’s ASEReadCounter | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/360037428291-ASEReadCounter |
| GATK4’s GenotypeGVCFs | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/360037057852-GenotypeGVCFs |
| Picard CollectWgsMetrics | GATK | https://gatk.broadinstitute.org/hc/en-us/articles/9570389458331-CollectWgsMetrics-Picard- |
| mLinker | Tourdot et al.95 | https://github.com/rwtourdot/mlinker |
| Vep106 | Ensembl124 | RRID:SCR 002344 |
| BioMart-Ensembl | Ensembl126 | RRID:SCR 019214 |
| Samtools v1.2.0 | Danecek et al.97 | https://www.htslib.org/ |
| Bcftools v1.2.0 | Danecek et al.97 | https://samtools.github.io/bcftools/bcftools.html |
| CoMut | Crowdis et al.98 | https://github.com/vanallenlab/comut |
| Cancer Gene Consensus | CGC | https://cancer.sanger.ac.uk/census |
| Snakemake | Mölder et al.101 | https://snakemake.readthedocs.io/en/stable/ |
| Python v3.7, v3.9 | Python Software Foundation | https://www.python.org/ |
| Matplotlib (v3.5.2) | https://matplotlib.org/ | https://matplotlib.org/ |
| R v3.6, v4.3.2 | R Development Core Team | https://www.r-project.org/ |
| R package ggplot2 | https://cran.r-project.org/web/packages/ggplot2/index.html | https://cran.r-project.org/web/packages/ggplot2/index.html |
| TITAN (v1.15.0) | Ha et al.89 | https://genome.cshlp.org/content/24/11/1881 |
| Circos (v0.69.9) | Krzywinski et al.100 | http://circos.ca/ |
| IGV (v2.12.3) | Robinson et al.125 | https://software.broadinstitute.org/software/igv/ |
| SvABA | Wala et al.116 | https://github.com/walaj/svaba |
| STAR (v2.7.10a) | Dobin et al.117 | https://github.com/alexdobin/STAR |
| STAR-Fusion (v1.7.0) | Github | https://github.com/STAR-Fusion/STAR-Fusion |
| Tophat2 2.0.10 | Kim et al.120 | https://ccb.jhu.edu/software/tophat/index.shtml |
| InferCNV package | Tickle et al115 | https://bioconductor.org/packages/release/bioc/html/infercnv.html |
| PureCN 2.10.0 | Riester et al.102 | https://github.com/lima1/PureCN |
| RSEM | Li et al.118 | https://github.com/deweylab/RSEM |
| Other | ||
| Codes deposition | This study | https://github.com/SViswanathanLab/WGS-TRCC/tree/main |
Data and code availability:
De-identified whole genome sequencing data reported in this study have been deposited in the NCBI Database of Genotypes and Phenotypes (dbGaP) under the study accession phs003008.v1.p1 and are publicly available with controlled access in accordance with an IRB-reviewed protocol. This study also analyzes publicly available existing data and the links or accessions to these datasets are listed in the key resources table. Whole exome data of tRCC cell lines have been deposited as a linked study at Sequence Read Archive (SRA) under the accession PRJNA865001.
Algorithms and code implemented in this study are uploaded at https://github.com/SViswanathanLab/WGS-TRCC.
Any additional information required to reanalyze the data is available upon reasonable request from the lead contacts.
EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Human subjects
A total of 31 tRCC tumors from 16 individuals (including two tumors from one misclassified ccRCC case, RCC16), each with matched germline control (either peripheral blood or adjacent normal tissue) were collected for whole-genome sequencing and RNA sequencing (excluding formalin-fixed paraffin embedded samples). Two samples (TRCC1 and TRCC2) failed DNA extraction and are not included in this study. For subject TRCC18, primary tumor and adjacent normal were collected at the time of initial nephrectomy and the remaining 15 samples (13 metastases and two normal tissues) were collected at rapid autopsy (see Figure 4). One subject (TRCC12) had both pre- and post-progression biopsies taken from the same metastatic site. Age range of the cohort was 25–68 years old; N=10 females; N=6 males. Clinical features of subject cohort are provided in Table S1.
All individuals who donated samples provided written informed consent consistent with genomic analysis of freshly collected or banked tumor samples with blood/normal tissue control. Rapid autopsy for subject TRCC18 was conducted under a genitourinary oncology rapid autopsy protocol approved by the Institutional Review Board (IRB) of the Dana Farber/Harvard Cancer Center. Under this protocol, patients expected to die of a genitourinary malignancy provide written informed consent to a research autopsy performed within 3–6 hours of death. Visceral and lymph node tumor deposits are identified via correlation of most recent clinical imaging with grossly visible tumor deposits at autopsy and collected for downstream profiling.
Cell Lines
293T and 786-O were obtained from American Type Culture Collection (ATCC, CRL-3216 & CRL-1932). S-TFE was obtained from the RIKEN BioResource Research Center (BRC) Cell Bank (RCB4699). UOK146 was obtained from Dr. Marston Linehan’s laboratory (National Cancer Institute). ASPS-1 was obtained from Dr. Robert H Shoemaker’s laboratory (National Cancer Institute). All cells were cultured in DMEM supplemented with 10% FBS, 100 IU/mL penicillin, 100 μg/mL streptomycin, and 100 mg Ml−1 Normocin at 37°C in 5% CO2.
METHOD DETAILS
Sequence data generation
Whole Genome Sequencing (WGS)
Tumor and matching germline tissue samples were subjected to whole genome sequencing in one of three categories: (1) standard PCR-free WGS performed on frozen tissue (25 samples); (2) standard WGS with PCR performed on FFPE tissue (5 samples); (3) linked-read WGS performed on frozen tissue (4 samples). DNA and RNA extraction, library preparation, and sequencing were performed by the Genomics Platform at the Broad Institute of MIT and Harvard using the standard procedures as described below. For single nucleus RNA sequencing, extraction, library preparation, and sequencing were performed at the Center for Cancer Genomics (DFCI).
Standard PCR-free WGS
For each tumor or germline reference sample, an aliquot of genomic DNA (100ng in 50μL) was used for library construction. Genomic DNA was fragmented to a target size of 385bp by shearing using a Covaris focused-ultrasonicator, with additional size selection performed using a SPRI cleanup. Library preparation was performed using a commercially available kit provided by KAPA Biosystems (KAPA Hyper Prep with Library Amplification Primer Mix, product KK8504), and with palindromic forked adapters using unique 8-base index sequences embedded within the adapter (Roche). DNA libraries were sequenced on HiSeqX instruments (2 × 151bp mode).
Standard WGS with PCR
For FFPE samples, the DNA libraries were further amplified by 10 cycles of PCR. Following sample preparation, libraries were quantified using quantitative PCR (kit purchased from KAPA Biosystems) with probes specific to the ends of the adapters; the qPCR quantification was performed on Agilent’s Bravo liquid handling platform. Based on qPCR quantification, libraries were normalized to 2.2nM and pooled into 24-plexes. Multiplexed DNA libraries were sequenced on NovaSeq 6000 instruments (2 × 151bp).
For standard WGS with or without PCR, sequence reads were aligned and processed following the best practice workflow recommended by the GATK without base score recalibration.
10x Genomics WGS
Linked-read DNA libraries were constructed using the standard protocol from 10X Genomics, starting with 1.2 ng of DNA for each sample as previously described93. Library fragment sizes were determined using the DNA 1000 Kit and 2100 BioAnalyzer (Agilent Technologies) and quantified using qPCR (KAPA Library Quantification Kit, Kapa Biosystems). Sequencing was performed on HiSeqX instruments (2 × 151bp mode). Raw sequencing reads were processed using the LongRanger software (10X Genomics) for barcode-aware alignment and variant calling.
Cell Line Whole Exome Sequencing (WES)
For whole exome sequencing of S-TFE and UOK146 cell lines, exome capture was performed using the Agilent SureSelect Human All Exon V6 Kit and paired-end sequencing (2×150 bp) was performed on an Illumina NovaSeq6000 instrument. The genomic DNA (> 300 ng) used for library construction was randomly sheared into short fragments with the size of 180–280 bp. The DNA fragments were end-repaired, A-tailed, and ligated with Illumina adapters for PCR amplification, size selection and purification. The prepped libraries were then hybridized in the buffer with biotin-labeled probes, and magnetic beads with streptavidin were used to capture the exons of genes. Non-hybridized fragments were washed out and probes were digested. The captured libraries were enriched by PCR amplification with real-time PCR and size distribution was assessed using Qubit and a bioanalyzer. Pooled and quantified libraries were sequenced on Illumina sequencers.
RNA Sequencing
Tumor Bulk RNA-seq
Bulk RNA sequencing libraries were constructed for 10 frozen tumor samples. One sample (TRCC13) with low sequencing depth and poor library quality was excluded from further analysis. RNA extraction was performed using AllPrep (Qiagen). Total RNA was quantified using the Quant-iT™ RiboGreen® RNA Assay Kit and normalized to 5ng/ul. Following plating, 2ul of ERCC controls (using a 1:1000 dilution) were spiked into each sample. An aliquot of 200ng for each sample was used for library preparation with the TruSeq™ Stranded mRNA Sample Preparation Kit. The resultant 400bp cDNA then went through dual-indexed library preparation (‘A’ base addition, adapter ligation using P7 adapters, and PCR enrichment using P5 adapters). Libraries were quantified using Quant-iT PicoGreen (1:200 dilution) and normalized to 5ng/ul. Pooled libraries were quantified using the KAPA Library Quantification Kit for Illumina platforms.
Pooled libraries were normalized to 2nM and denatured using 0.1 M NaOH prior to sequencing. Sequencing was performed on an Illumina NovaSeq 6000 instrument (2 × 101bp, single 8bp index) at the Broad Institute of MIT and Harvard. Sequencing data were processed as described below.
Single-Nucleus RNA-seq
Single-nucleus RNA-Seq was performed on the nephrectomy sample from TRCC18. Nuclei isolation was performed as previously described94. Briefly, OCT-embedded tissue fragments were quickly thawed in PBS on wet ice to remove excess OCT, tissue was next mechanically dissociated by pipetting and homogenized in TST solution, filtered through a 30 μm MACS SmartStrainer (Miltenyi Biotec), and pelleted by centrifugation for four minutes at 500g at 4C. The nuclei pellet was resuspended in 200 μL of ST-SB buffer, and nuclei were counted by eye using INCYTO C-Chip Neubauer Improved Disposable Hemacytometers (VWR International Ltd.). Low-retention microcentrifuge tubes (Fisher Scientific) were used throughout the procedure to minimize nuclei loss.
Approximately 8,000 nuclei were loaded into a single channel of the Chromium Next GEM Chip G for library construction on the 10x Chromium Controller (10x Genomics) following manufacturer’s instructions (Chromium Next GEM Single Cell 3ʹ Reagent Kits v3.1 User Guide, Rev D). Libraries were normalized and pooled for sequencing on a NovaSeq SP-100 flow cell at the Center for Cancer Genomics at Dana-Farber Cancer Institute.
DNA Sequence Data Processing and Variant Calling
All sequencing data were aligned to GRCh38. Linked-reads data were aligned by the LongRanger program from 10X Genomics. Standard WGS data were aligned using bwa.
Germline and somatic variant calling by HaplotypeCaller
Variant calling was performed jointly on tumor and matched normal samples collected from each patient using HaplotypeCaller from Genome Analysis Toolkit (GATK) as described previously90. The same parameters were used for variant calling for all libraries; differences in downstream processing for each data type is described below.
To select germline heterozygous variants for allelic DNA and RNA analysis, we imposed three criteria: (1) variant is included in the 1000 genomes reference data; ftp.ebi.ac.uk/1000genomes/ftp.1000genomes.ebi.ac.uk/vol1/ftp/data_collections/1000G_2504_high_coverage/working/20201028_3202_phased/, (2) heterozygous genotype emitted by HaplotypeCaller; (3) five or more reads of both the reference and alternate genotypes in the germline reference sample. We recognized the presence of contaminating DNA (from unrelated individuals) in the germline reference of TRCC15 and the tumor sample of TRCC17 based on the presence of many common heterozygous variants with low allelic fractions. To eliminate these variants from DNA copy-number analysis, we additionally imposed the following criteria: (4) the allelic fractions of both reference and alternate genotypes need to be within the range of 0.35 to 0.65; (5) the combined allelic depth needs to be equal to or greater than 20. These additional criteria were unnecessary for the other samples.
For linked-read data, we directly performed haplotype phasing using mLinker, which can purify false heterozygous variants based on haplotype linkage95.
For standard WGS data, we selected somatic variants with the following criteria: (1) variant sites not among common polymorphisms in the 1000 Genomes Phase 3 reference haplotype panel; (2) homozygous genotype in the germline reference; (3) five or more reads of the alternate genotype across all tumor samples. The resulting somatic variant callsets were used to intersect with the somatic callsets generated by Mutect2 to improve variant calling accuracy.
Somatic mutation detection with Mutect2
Somatic mutation detection was performed on all WGS samples using Mutect2 v4.3.0 from Genome Analysis Toolkit (GATK). All samples were run in paired tumor-normal mode.
We imposed the following read filters to control false variant detection due to misalignment:
--minimum-mapping-quality 30 (excluded reads having low mapping quality)
--read-filter OverclippedReadFilter
--filter-too-short 25
--downsampling-stride 20 (truncate regions of extreme depth due to mapping error)
--max-reads-per-alignment-start 6
--max-suspicious-reads-per-alignment-start 6
To filter false variants due to recurrent alignment artifacts, we used a reference panel of common false variants provided by GATK:
gs://gatk-test-data/mutect2/M2PoN_4.0_WGS_for_public.vcf
To filter false variants due to rare alignment artifacts and exclude germline variants that were missed in the matching germline reference, we used a germline resource consisting of >10,000 genomes from gnomAD provided by GATK:
gs://gatk-best-practices/somatic-hg38/af-only-gnomad.hg38.vcf.gz
For three patients with multiple tumors (TRCC12, RCC16, and TRCC18) or multiple normal references (TRCC18), Mutect2 was run in the joint-calling mode with similar filtering parameters as described above.
Filtering of somatic mutations
For FFPE samples (TRCC3, TRCC4, TRCC7, TRCC9 and TRCC15), we did not perform further analysis of sSNVs or INDELs given the high frequency of errors introduced during library construction resulting from DNA degradation21,22. Variants with functional impact detected in CGC Tier 1 genes from FFPE samples are provided in Figure S1 for informative purposes.
For linked-read data (TRCC5, TRCC6, TRCC8 and TRCC10), we intersected variant callsets generated by the LongRanger pipeline with callsets generated by Mutect2 using bcftools96, 97 (bcftools isec). This intersection was performed to exclude low-confidence variants generated by LongRanger and false positive variants generated by Mutect2 (which does not incorporate molecular barcode information present in linked-read data).
For callsets generated from standard WGS data, we performed the following filtering steps. (1) FilterMutectCalls to filter false variants due to contaminating DNA and artifacts that may introduce orientation bias (such as 8-oxoguanine artifacts) by specifying tumor contamination, segmentation and read orientation models. (2) FilterAlignmentArtifacts to remove alignment artifacts from poorly mappable regions of the genome. All variants located in centromeres were also removed.
We imposed a hard filter of variant AD ≥ 3 for all detected mutations. For TRCC18, we intersected variants generated by Mutect2 with those generated by HaplotypeCaller and selected bi-allelic single nucleotide variants for phylogenic inference. We further removed variants with missing genotype calls (‘./.’) in one or more samples in the Mutect2 callset and variants with either missing genotype calls or called as homozygous reference by HaplotypeCaller.
Variant annotation with VEP
All somatic variants that passed filters were annotated using Ensembl Variant Effector Predictor (version 106) with default parameter settings. Given the cohort size, we limited to somatic variants within Tier 1 genes from Cancer Gene Census (CGC; GRCh38 COSMIC v96). For downstream analyses, these Tier 1 filtered genes were assessed for impact and pathogenicity prediction from VEP106. Mutations were visualized with python package CoMut98. For the CoMut plot, genes are shown if they were mutated with 1) frameshift HIGH 2) HIGH/MODERATE impact with at least one mutation predicted to be of high pathogenic impact in the cohort.
Somatic DNA Copy-Number and Rearrangement Analysis
DNA copy number was calculated using a haplotype-specific copy-number workflow90. For standard WGS, haplotype phasing was performed using eagle299 and reference haplotype data from the 1000 genomes consortium (phased variants include both single-nucleotide variants and insertion/deletion variants); for linked-reads data, haplotype phasing was performed using mLinker95 on germline variants detected in the germline reference linked-reads data (only sSNVs). Local haplotype fractions were calculated based on allelic depths at heterozygous variants and their haplotype phase in either 10kb intervals (samples with high purity and more SCNAs: TRCC12, TRCC13, TRCC14, TRCC15, TRCC16, TRCC18) or 30kb intervals (samples with few SCNAs or low purity: TRCC3, TRCC4, TRCC7, TRCC9, TRCC11, TRCC17; linked-reads samples: TRCC5, TRCC6, TRCC8, and TRCC10). The larger interval size was chosen to either maximize the detection sensitivity of low-clonality SCNAs in four FFPE samples (TRCC3/4/7/9), in samples with no large-scale SCNAs (TRCC3/11/17), and in linked reads samples with more variable total sequence coverage at 10kb intervals (TRCC5/6/8/10). For linked-read data, allelic depth was calculated from the number of unique molecular barcodes at each variant site using mLinker; for standard WGS data, allelic depth was generated by HaplotypeCaller.
The combined sequence coverage (from both haplotypes) was calculated from the normalized sequencing coverage in 10kb intervals (or aggregated into 30kb intervals); for linked reads data, the combined sequence coverage was calculated from the number of unique molecular barcodes using mLinker.
The allelic depth was calculated from the combined sequence coverage and the haplotype fractions. We then performed haplotype phasing based on allelic imbalance as described previously90. We calculated and reviewed 100kb-interval DNA copy number generated from the allelic-depth phased chromosomal haplotype, and manually corrected rare long-range switching errors (usually 0–3 per sample). The final haplotype and haplotype-specific DNA copy number (in 100kb intervals) were used for downstream DNA copy-number analysis.
Rearrangement detection
We identified rearrangements based on discordant (including split mapped) reads as described previously90. Candidate rearrangements were emitted when there were at least two distinct discordant reads within the maximum insert size of the DNA library (~1000bp) at each candidate breakpoint region. All TFE3 fusion breakpoints were manually reviewed (breakpoint coordinates listed in Table S2).
For three samples with FFPE DNA (TRCC3/4/9), we only considered TFE3 fusion breakpoints as they were assessed to have either low tumor cell fractions (TRCC3/4) based on the number of reads supporting the TFE3 fusions or displayed few large segmental SCNAs (TRCC9). We manually reviewed copy-number changepoints and identified 11 rearrangements in TRCC7.
For linked-reads data, we imposed the following filtering criteria on candidate rearrangements: (1) The breakpoints need to be at least 100kb away from each other or reside on different chromosomes (due to the prevalence of short-range chimeras in linked-reads data); (2) The number of unique molecular barcodes of all supporting reads needs to pass a sample-specific threshold (10 for TRCC6 and TRCC10, 6 for TRCC8); (3) The number of unique molecular barcodes of supporting reads in the normal sample needs to be lower than 5% of the number of unique molecular barcodes of supporting reads in the tumor sample. Only rearrangements with one or both breakpoints co-localized with copy-number changepoints are included in the CIRCOS100 plots. There were no long-range breakpoints detected in TRCC5 when reviewed against DNA copy-number data.
For other standard WGS data, we required at least 5 supporting discordant reads (10 reads for TRCC12 or RCC16 with two tumor biopsies) for each candidate rearrangement. We further imposed the following criteria: (1) Supporting reads only come from each individual sample (this is not imposed on TFE3 fusions as the breakpoints may be in close proximity); (2) There are at least two non-split discordant reads and at least one split read; (3) The total number of supporting reads from all normal reference samples is less than three. Last, we intersected the rearrangement breakpoints with copy-number changepoints and excluded short insertions, deletions, or tandem duplications (<100kb). We manually reviewed all breakpoints against the DNA copy-number plots.
Inference of Purity/Ploidy
We performed tumor purity/ploidy inference using TITAN89 with a Snakemake workflow101 and verified the purity/ploidy solution based on the haplotype-specific allelic depths generated for each chromosome (46 for female, 45 for male). For samples with few or no SCNAs, we further used the AF distribution of somatic mutations to corroborate the purity estimates (for non-FFPE samples). The final results are presented in Table S1.
Phylogenetic inference in TRCC18
Due to the low overall burden of both sSNVs and SCNAs, we performed phylogenetic inference on all 13 tumor samples (excluding the primary) from the TRCC18 patient by a joint analysis of sSNVs and SCNAs.
Somatic SNV-based Phylogenetic Reconstruction in TRCC18
To infer the phylogenetic relationship based on sSNVs in TRCC18, we first determined the appropriate variant allelic fraction (VAF) cutoff for clonal sSNVs in each tumor sample. We performed k-means (k = 2) clustering on the VAF data from each tumor sample to determine the cutoff separating high and low VAF variants. We verified these sample-specific cutoffs against the tumor cell fractions estimated from the allelic depth of the deleted chr9 haplotype that was inferred to be a truncal event shared by all tumors (and must be clonal in each sample). We further determined a set of truncal sSNVs that passed the VAF threshold in 9 or more tumor samples based on the histogram of sample presence of all sSNVs; based on these truncal variants, we independently verified the sample-specific VAF cutoffs (nearly all truncal variants passed the cutoff in each sample) and estimated the rate of false negative detection based on the number of truncal variants that do not meet the VAF cutoff.
We next calculated the number of sSNVs shared by two or more tumor samples and constructed the phylogenetic tree by neighbor-joining. We started by grouping sample pairs with the most shared sSNVs (e.g., MedLN and Liver-4), merged the pairs into one node in the phylogenetic tree, and iteratively constructed a binary tree from all samples.
There are several noteworthy caveats. First, there are 28 mutations shared by all samples except the lung tumors (Lung-1/2/3) and RPLN-2, 9 mutations present in all but Liver-4 plus the same lung and RPLN samples, and 14 mutations shared by all samples except Lung-1/3 and RPLN-2. Nearly all these variants (51 total) reside on the translocated chromosome der(X)t(1;X) that was deleted in the lung samples, in a major subclone of RPLN-2, and underwent a single-copy loss after whole-genome duplication in the common ancestor of MedLN and Liver-4. Therefore, we conclude that the absence of these mutations in the Lung/RPLN/Liver-4 is due to secondary deletions/losses of this chromosome, instead of de novo mutations in the remaining samples.
Second, the Lung-2 sample shares 39 mutations with the other two Lung biopsies, but also contains 14 mutations that are absent in Lung-1/3 but present in the Liver and MedLN samples, mostly on der(X)t(1;X). A close examination of the somatic copy-number data revealed that the Lung-2 sample is polyclonal and has low tumor cell fractions. We therefore conclude that the conflicting sSNV evidence is due to the mixing of tumor DNA (possibly during rapid autopsy) and grouped Lung-2 with the other two lung biopsies because of their genetic similarity as well as anatomical proximity.
Finally, the RPLN-2 sample has very few shared mutations with the other samples (excluding the truncal mutations). This is not due to low tumor cell fraction in the RPLN-2 sample (the tumor cell fraction is estimated to be ~60%). We further detect a large number of private mutations in this sample (second only to Liver-3) and a private chromothripsis event on der(X)t(1;X). We group RPLN-2 with RPLN-3 based on 8 mutations shared between them but note that RPLN-2 likely diverged early from the other tumors and also acquired more genetic alterations during downstream evolution.
Phylogenetic inference of sSNVs
After determining the phylogeny, we then assigned each mutation to a branch in the phylogenetic tree based on its allelic fractions in all tumor samples. For each mutation classified as not clonal or not present in a given sample (based on the VAF cutoff), we assigned a false-negative probability based on the fraction of truncal mutations that were missed in this sample; for each mutation classified as clonal in a given sample, we assigned a constant false-positive probability (0.01). We then calculated the likelihood of assigning a mutation to a branch based on the presence or absence of this mutation in all the samples and assigned this mutation to the branch with the maximum likelihood. We grouped mutations by clonality patterns and further reviewed the genome-wide distribution of mutations with each pattern to identify mutations whose absence arose from chromosomal deletions or losses. We manually adjusted their branch assignment as needed (Table S6).
Consistency between sSNV based phylogeny and SCNA-based phylogeny
We inferred most segmental SCNA breakpoints to be ancestral, which were not useful for phylogenetic inference. Both MedLN and Liver-4 samples were inferred to be tetraploid based on the allelic depth distribution; this inference is consistent with their adjacency in the phylogenetic tree. In addition, Liver-4 and MedLN also share breakpoints on 17p and 22q that are distinct from the other samples. We observed independent SCNAs in different phylogenetic branches, including (1) deletion of chr22A; (2) subclonal gains of segments from chr17 and chr18 that were joined together by an ancestral rearrangement; (3) loss of chr4A; (4) deletion of the translocation der(X)t(1;X). We verified that deletions of chr22A were either of the entire chromosome (which may occur independently in different branches), or showed distinct deletion boundaries (e.g., between Lung and Liver-4). The same pattern was observed for deletions of der(X)t(1;X). In summary, there are no segmental SCNA breakpoints shared by two tumor samples that are not immediately adjacent in the phylogenetic tree.
Cell Line Copy Number Analysis
WES reads were processed similarly tos WGS reads as described above. For s-TFE and UOK146 cell lines, PureCN 2.8.12.8.1102 was used to analyze copy number without a paired normal sample with recommended segmentation PSCBS (Parent Specific Copy Number). To contain both somatic and germline variants in the VCFs as input to PureCN, variant calling with Mutect2 was run with tumor only mode as
--genotype-germline-sites true –genotype-pon-sites true
To filter sequencing artifacts and library-specific bias, Coverage.R from PureCN was used to calculate GC-normalized coverage of sTFE and UOK146 cell lines. We created a normal database from a pool of 34 adjacent normal samples from a published study13 that used same SureSelect Human All Exon V6 bait design. Mapping bias file was generated from these 34 samples as recommended for PureCN run without a matched normal. Copy number analysis was performed with PureCN against hg38 bait intervals that SureDesign used to select the probes and excluded blacklisted regions of simple_repeats.bed file. To search for optimal solutions without normal contamination, the analysis was calibrated for cell line run as
–max-copy-number 8, --min-purity 0.9, --max-purity 0.99 and –model-homozygous
Analysis of External tRCC Data
Public Datasets
Data from external tRCC cohorts were obtained as previously described8 for TCGA, CCLE, PCAWG, Durinck et al., Malouf et al., MSK Panel, MSK WES, Oncopanel, Sato et al9,10,25,65,103–109. Raw tRCC sequencing data and metadata for Sun et al.12 and Qu et al.13 were obtained from the respective studies. For TFE3 tRCC samples in Sun et al. with RNA-seq data but lacking fusion calls, we ran STAR fusion110 (v1.7.0) on the RNA-seq to infer TFE3 fusion partners. As Sun et al. did not report patient sex annotations, we assessed chrX heterozygosity in the matched normal samples to determine sex, as previously described56.
Copy Number Analysis
tRCC samples from published cohorts were uniformly processed with the TitanCNA pipeline89 using the baitsets specified in the source publications and converted to hg38 (UCSC LiftOver) if needed, with the following parameters TitanCNA_maxNumClonalClusters: 3; TitanCNA_maxPloidy: 4; ichorCNA_normal: c(0.25, 0.5, 0.75); ichorCNA_ploidy:c(2,3,4); ichorCNA_includeHOMD:TRUE; ichorCNA_minMapScore: 0.75; ichorCNA_maxFracGenomeSubclone: 0.5; ichorCNA_maxFracCNASubclone:0.7. The optimal cluster solution calculated by TitanCNA was taken as the most likely solution in the absence of manual review.
Copy number imbalance (CNI) was determined by manual review of allelic copy number on chrX and the autosomal fusion partner in all cases. Deletion or gain from the breakpoint to the telomere on at least one arm was considered CNI. Samples with CNI are shown in Figure S4 and/or annotated in Table S5. Even though the TFE3 rearrangement junctions cannot be identified in WES data, the reciprocality of TFE3 rearrangements can be indirectly inferred based on the DNA copy number near the TFE3 locus on chrX and near the partner locus. Fusions arising via simple reciprocal translocations or inversions on chrX will have no large segmental or arm-level SCNAs. By contrast, copy-number imbalance at either the TFE3 locus or at the partner locus could indicate either (1) a translocation that was unbalanced at the onset (e.g., resulting from large terminal deletion or retention generated by chromosome bridge resolution) or (2) a balanced translocation that was followed by subsequent copy number change, such as loss of the translocated chromosome not containing the oncogenic fusion88,90.
RNA-Sequencing Analysis
Bulk RNA-Seq Analysis
Sequence reads were aligned to GRCh38 using STAR (v2.7.10a), followed by fusion analysis using STAR-Fusion (v1.7.0) with the GRCh38 Gencode v37 CTAT library (vMar012021). The identified gene fusions were subsequently reviewed manually using Integrative Genomics Viewer (IGV)112. We further reviewed chimeric RNA-Seq reads supporting the TFE3 fusion transcripts and DNA-Seq reads supporting the fusion junction jointly to infer splicing events of fusion genes.
snRNA-Seq analysis
Demultiplexing of sequencing reads, barcode processing, read alignment, and UMI counting were performed using the 10x Cell Ranger analysis pipeline. The reads were aligned to Gencode (GRCh38) and extracted RNA counts were analyzed using Seurat V4113. Doublet detection and removal on gene-barcode matrices were performed to exclude data from droplets containing more than one cell using Scrublet114. Cells with fewer than 200 genes detected or more than 5% of counts attributed to mitochondrially-encoded transcripts were removed. Genes detected in fewer than three cells were also excluded. Tumor cells were identified based on known tRCC gene markers (TRIM63 and GPNMB) and inferred transcriptional copy number variations (CNVs) estimated via the inferCNV package115 that were consistent with WGS-based copy number profile116. Normal cell types were determined through manual annotation via known marker genes. For arm-level expression analysis, the mean expression level of all genes within a specific chromosome region was calculated for each malignant and normal (non-tumor compartment) cell. For autosome analysis, disomic autosomes were considered (chromosomes 1, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 20, and 21). The mean expression of respective chromosome arm per cell was then normalized to the average expression of that arm across all non-malignant cells. Statistical significance was identified using a Mann-Whitney U test between tumor vs normal of individual chromosome arms.
Normalized Expression Analysis
STAR (v.2.7.2b)/RSEM (v.1.3.1) was utilized to align and quantify paired-end RNA-seq reads to the GENCODE v38 transcript reference117–119. For each cohort, the mean expression of a gene (in TPM) was calculated in reference samples (those without chrXp-reactivation). The sample-level expression of the gene was then normalized using this value.
Haplotype-Specific Expression in TRCC18
The haplotype phase of variants in regions of allelic imbalance was determined in the DNA copy-number analysis. The allele-specific expression pipeline was then applied to obtain haplotype expression fractions, using the aforementioned assigned haplotypes.
Allele-Specific Expression of external tRCC data
Tumor RNA-seq data were uniformly re-aligned using TopHat2120 and DNA-seq data of the matching germline reference were re-aligned using bwa mem121 to GRCh38. For CCLE, we utilized tumor DNA-seq for the identification of heterozygous variants. We called germline variants using HaplotypeCaller with the standard workflow122 and selected heterozygous SNVs. We then used ASEReadCounter to count reference and alternative RNA reads across the heterozygous sites122. We restricted to sites with total RNA read count and total DNA read count of at least 10 and further imposed the constraint that the germline DNA variant allele fraction is within one standard deviation from the mean, to select biallelically expressed variants; the VAF filter was not applied for cancer cell lines (CCLE) without normal reference. For analysis of chrX, we filtered pseudoautosomal regions (PAR1: 1–2781479 and PAR2: 155701383–156030895; GRCh38 coordinates), escapee exons, and variable escapee exons as defined by kidney-specific methylation123 and coordinates determined from Ensembl BioMart124. Heterozygous sites were classified as biallelically expressed if the minor allele fraction of RNA reads was ≥ 0.2. We determined whether chrXp reactivation was present by comparing the distributions of minor allele fractions of expressed SNVs to the left versus to the right of TFE3 using a two-tailed Mann-Whitney U test. We determined whether autosomal silencing was present by comparing the distributions of minor allele fraction for expressed SNVs to the left versus to the right of the autosomal breakpoint using a two-tailed Mann-Whitney U test.
Analysis of 3’ SNVs in TFE3 and Autosomal Fusion Partner
SNVs in 3’ TFE3 or the autosomal fusion partner are possibly informative for inference of translocated haplotype (chrXi, chrXa). One expects 3’-TFE3 will be monoallelically expressed during chrXa fusion, but biallelically expressed during chrXi fusion. On the other hand, the 3’ autosomal fusion partner will be biallelically expressed during chrXa fusion, but monoallelically expressed during chrXi fusion. We manually inspected every sample for expressed germline heterozygous SNVs in these regions, and computed RNA MAF if such site was present using the output from the allelic expression analysis using ASEReadCounter. If RNA MAF ≥ 0.2, we considered the SNV biallelic; otherwise, it was considered monoallelic. If a sample had multiple SNVs in one region, we considered the SNV with the highest RNA MAF.
Analysis of Reciprocal Fusion Supporting Reads
Reads supporting the TFE3-Partner reciprocal fusion should be detectable for chrXa fusions. We manually counted split/discordant reads spanning at the TFE3-Partner and Partner-TFE3 breakpoints by visualizing GRCh38 aligned RNA-seq .bam files in IGV125.
Analysis of chrX expression in non-tRCC tumors cancers with chrX rearrangements
To identify additional cases of X-reactivation in the TCGA, we focused on tumors with X:autosome rearrangements in an orientation analogous to the X:autosome TFE3 rearrangements in tRCC. We used the TCGA driver fusion callset64, removed samples with multiple chrX rearrangements, and selected samples with rearrangements that retain the C-terminus of a chrX gene with expression driven by the promoter of an autosomal gene. We further restricted to cases where the chrX gene is anti-sense if it is to the left of the X-inactivation center (XIC) and sense if it is to the right of XIC (such that the autosomal promoter is in trans with the XIC as for TFE3 fusions). After excluding tRCCs from the final dataset, we applied the previously described allele-specific expression pipeline. For analysis of CCLE, we ran the allele-specific expression pipeline (using DNA minor allele fraction ≥ 0.2 to identify heterozygous sites) on 273 cell lines with deposited RNA-seq and exome sequencing in the GDC legacy archive on July 2022 (Note: the legacy archive has been retired as of May 2023, so one must now obtain this data through EGAD00001001039). Of these, 115 were female, 147 were male, and 11 had sex unspecified. We focused on female cell lines with loss of XIST expression (N=48), as we have previously identified56. This can occur from whole chromosome loss (of chrXi, N=28) or focal events (deletion of the XIST locus [N=4] or repression of XIST expression [N=16]). We focused on the latter classes, as they have the potential to lead to X-reactivation on the retained portion(s) of chrXi.
Methylation Analysis
Methylation Left Versus Right of the TFE3 Breakpoint
TCGA methylation array data for female tRCCs was downloaded from the GDC portal (https://portal.gdc.cancer.gov/). We identified gene start coordinates for chrX genes using Ensembl BioMart126. We excluded escapees (including PARs) as defined by kidney-specific methylation123. We restricted the methylation array to probes in promoters (regions ±1 kb from the gene starts) of the remaining genes. We compared the distribution of beta at probes left of TFE3 (chrX:1–49,028,726) to those right of TFE3 (chrX:49,043,410–156,040,895).
Methylation at Biallelically Expressed Genes
For each chrXi-rearranged female TCGA tRCC (determined by biallelic chrXp expression), we identified biallelic SNVs (RNA MAF ≥ 0.2) left of TFE3 using the output from the allele-specific expression analysis. We matched these SNVs to genes using Ensembl BioMart126 and additional manual review in the UCSC Genome Browser127. If a gene had at least one biallelic SNV (RNA MAF ≥ 0.2), it was considered reactivated. After identifying reactivated genes, we subsequently generated a list of beta values for probes in the promoters (regions ±1 kb from the gene starts) of reactivated genes.
Experimental Work
Generation of TFE3 Fusion via CRISPR/Cas9
sgRNAs targeting intron 1 of PRCC and intron 3 of TFE3 were cloned into lentiCRISPR v2 (Plasmid #52961 on Addgene, Table S4)129. The combination of sgRNAs were transfected into 293T cells using 4:1 polyethylenimine (PEI):DNA. After 72 hours, total RNA was isolated with the RNeasy Plus Mini Kit (Qiagen, #74136). cDNA was synthesized from 1 μg of total RNA using SuperScript IV VILO Master Mix (Thermo Fisher Scientific, #11756050). Genomic DNA was isolated using the DNeasy Blood & Tissue kit (Qiagen, #69504). cDNA or gDNA was amplified using appropriate primers (see Table S4) for 40 cycles and run on a 1% agarose gel containing SYBR Safe DNA gel stain (1:10000). Gels were imaged using UV light.
Amplicon Sequencing and Analysis
Amplicons were excised from gels and DNA was isolated using QIAquick Gel Extraction Kit (Qiagen, #28706X4). Amplicons were submitted for linear/amplicon long-read sequencing (PlasmidSaurus). We first aligned the reads (using bwa mem) to a fasta reference containing both TFE3 and PRCC sequences to determine the breakpoints on both fusion products (chr1:156777745 and chrX:49038604 for the PRCC-TFE3 fusion; chrX:49038605 and chr1:156777746 for the TFE3-PRCC fusion; GRCh38 coordinates). We then created references for each fusion gene by concatenating 2kb up- and down-stream sequences from each breakpoint and realigned the reads. Breakpoint-spanning reads were manually reviewed to determine the number of inserted or deleted nucleotides in each amplicon.
Immunoblotting of wildtype and fusion TFE3 proteins
The cells were lysed on ice using RIPA lysis buffer (Thermo Fisher Scientific, #89901) supplemented with protease inhibitors (Roche,ac #11836170001). The amount of protein in lysates was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, #23225). Equal quantities of protein were then loaded onto NuPAGE 4–12% Bis-Tris Protein Gels (Thermo Fisher Scientific, #NP0335) and separated by SDS-PAGE. The proteins were transferred from the gels to nitrocellulose membranes using an iBlot2 (Thermo Fisher Scientific) and incubated overnight at 4°C with rabbit anti-TFE3 (CST, #14779S, 1:1000). Afterward, the membranes were washed in TBS-T and then incubated at room temperature for 1 hour with IRDye 680LT goat anti-Rabbit IgG: LI-COR (#926–68021, 1:20,000). Immunoblots were imaged using the Odyssey CLx Infrared Imaging System (LI-COR Biosciences).
QUANTIFICATION AND STATISTICAL ANALYSIS
Pairwise comparisons of distributions were performed using two-tailed Mann-Whitney U tests (α=0.05) unless otherwise specified. P-values are shown in Figures, text, and/or Supplementary Tables. The statistical tests used to generate P-values are indicated in the Figure Legends and text. The exact value of N is indicated in the Figures, Figure Legends and/or text. Boxes in boxplots extend from the lower to upper quartile values of the data with a line at the median.
Supplementary Material
Figure S1. Recurrent mutations and copy number alterations in tRCCs revealed by WGS. Related to Figure 1. (A) Recurrent single-nucleotide substitution and insertion/deletion mutations of Tier 1 cancer genes in the cancer gene census (CGC) in the tRCC WGS cohort. Only cancer genes with mutations predicted to have high functional impact in at least one sample, or high/moderate impact in two or more samples are included. FFPE samples (TRCC3/4/7/9/15) were excluded in the calculation of mutation frequency (NA) due to the prevalence of artifacts in FFPE libraries21,22 and the low clonal fractions of tumor cells in some FFPE samples (e.g. TRCC3/4), but potentially pathogenic mutations of CGC genes are provided for informative purposes. We used both copy-number alterations and point mutations to assess the clonal fractions of tumor cells (Purity). For TRCC3 and TRCC9, purity was not assessed due to the absence of large copy-number alterations and the uncertainty of point mutation detection in FFPE DNA. (B) Genome-wide distribution of copy number gains (red) and losses (blue) in 100 kb bins. Known cancer drivers in peak/valley regions are annotated; two insets highlight recurrent focal deletions on chr1p (spanning CDKN2C, ARID1A) and 9p (spanning CDKN2A/2B). (C) Diagrammatic view of TFE3 fusions in 9 cases with both DNA- and RNA-sequencing (Table S2). In each sample, the exons included in the oncogenic fusion transcripts (Partner-TFE3) as detected by RNA-seq are highlighted in gray-shaded boxes of the TFE3 gene (top, 3’ of the fusion transcript) and its partner (bottom, 5’ of the fusion transcript). Exons included in the reciprocal fusion (TFE3-Partner) when detected are outlined with dotted boxes (5’ of TFE3 and 3’ of the partner). TFE3 is anti-sense. Fusion partners with shaded boxes on the left are sense, those with shaded boxes on the right are anti-sense. DNA breakpoints are denoted by vertical lines with dark blue lollipops. Note for SFPQ, isoform NR_136703.1 is used based on the RNA-Seq data.
Figure S2. tRCC genome complexity and evolution revealed by WGS. Related to Figure 1. (A) CIRCOS plots showing haplotype-specific DNA copy number (red and blue for each parental haplotype), inter-chromosomal translocations (purple arcs), and long-range intrachromosomal rearrangements (green arcs, breakpoints >1Mb apart). For TRCC18 (boxed), the primary tumor sample is shown, with haplotype-specific DNA copy number across all chromosomes (top CIRCOS) or selected chromosomes displaying segmental copy number alterations (bottom CIRCOS); rearrangements are indicated on the latter CIRCOS. Other representative metastases from TRCC18 are shown separately in Figures 4–5 and Figures S5–S6. In both the primary tumor and the resected metastatic lesion from subject RCC16, the absence of a detectable TFE3 fusion suggests that it may have been misclassified (CIRCOS for primary shown as representative). The following copy-number changes/rearrangement of this genome are consistent with features of a clear-cell renal cell carcinoma (ccRCC): (1) The presence of 3p loss is a hallmark feature of ccRCC; moreover, the 3p loss is coupled to a terminal deletion on 1p accompanied with chromothripsis, a frequent rearrangement pattern seen in ccRCCs24. (2) The clustering of breakpoints on chr4, 8, 10, 11, 15, and 17 near the boundaries of large segmental SCNAs is consistent with rearrangement outcomes generated by the breakage of multi-chromosome bridges, including both regional chromothripsis and junctions with multiple tandem short insertions24,88. Although a similar pattern was observed in TRCC18 between chr6 and chr14, such events are rare in tRCCs and we never observed multi-chromosomal (3 or more) rearrangements as seen in this case. (3) The tumor genome is inferred to be 4N based on the distribution of allelic depths showing four different peaks at equal spacing indicating allelic copy-number states of 0, 1, 2, and 3; moreover, the presence of many segmental SCNAs at odd copy-number states suggests downstream SCNA evolution after ancestral whole-genome duplication. The sequence of early WGD followed by downstream evolution contrasts with tRCC genomes that are usually 2N or show few alterations either before or after WGD (see Figures 4–5). (B) Haplotype-specific coverage for chr5 and chr6 in the pre- (left) and post-treatment (right) samples from TRCC12. The integer copy-number states are determined from truncal deletions of chr3q and chr17p shared by both samples. Allelic DNA copy number of chr5 and chr6 indicates clonal deletion of 6q of the B haplotype (6Bq) and duplication of 5q of the A haplotype (5Aq) in the post-treatment clone (right), but subclonal losses of 6q and gains of 5q of both haplotypes in the pre-treated clone (left). One possible explanation is that the pre-treatment tumor has two subclonal populations: one containing an extra copy of 5Aq but missing 6Bq, the other containing an extra copy of 5Bq but missing 6Aq; and the post-treatment tumor only consists of the first. Interestingly, both subclones have disomic 6p, monosomic 6q, and trisomic 5q (from both homologs) that match the average DNA copy number of these arms across all tRCCs (Figure S1B).
Figure S3. CRISPR/Cas9 engineering of TFE3 fusions in 293T cells. Related to Figure 2. (A) Genomic locations of sgRNAs used to engineer PRCC-TFE3 fusions via CRISPR/Cas9 and PCR primers for detecting both PRCC-TFE3 and TFE3-PRCC fusions. (B) DNA gel showing PRCC-TFE3 fusions in the bulk edited population of 293T cells. (C) Distribution of the number of deleted or inserted nucleotides in each PCR amplicon relative to the predominant PRCC-TFE3 breakpoints (chr1:156777745 and chrX:49038604) inferred from the population. (D) DNA gel showing TFE3-PRCC fusion in the edited bulk population. (E) Distribution of the number of deleted or inserted nucleotides in each PCR amplicon relative to the predominant TFE3-PRCC breakpoints (chrX:49038605; chr1:156777746). Two amplicons containing insertions of sequences of Cas9 DNA (~75bp) were excluded. (F) Expressed PRCC-TFE3 fusion transcripts detected via RT-PCR across the predicted RNA fusion breakpoint in the edited bulk population. (G) Sanger sequencing trace confirming PRCC-TFE3 fusion junction.
Figure S4. Copy number imbalance of TFE3 translocations in tRCC. Related to Figure 3. (A) Minor allele fraction (top) and normalized read depth (tumor:normal) from the whole-genome sequencing data of a tumor (TCGA BP-4756, female) showing copy-number gain of the translocated chromosome der(1)t(1;X) containing an oncogenic SFPQ-TFE3 fusion. Regions with copy number imbalance in chr1 (top) and chrX (bottom) are highlighted in orange. (B) A tRCC tumor (TCGA BQ-5887, male) showing copy-number gain of der(X)t(1;X) with the reciprocal TFE3-PRCC fusion. Plots are similar to A. Schematic diagrams of the duplicated chromosome segments are shown below. (C-E) Three tRCCs with SFPQ-TFE3 fusions showing (C) deletion of both segments in the reciprocal translocation der(X)t(1;X); (D) deletion of the segment telomeric to SFPQ; and (E) duplication of both segments in the oncogenic translocation der(1)t(1;X). Note that deletion of the TFE3 fusion is never observed. Shown for each sample are the minor allele fractions, normalized read depths, and segmented allelic copy number of both chrX and the partner chromosome (chr1). In all cases, DNA copy number was calculated by TITAN89, which does not perform local haplotype inference. For cases with copy-number imbalance on one breakpoint (SFPQ) but not the other (TFE3) as shown in D, a plausible explanation is that the TFE3 fusion arises from chromoplexy (multi-way balanced translocations) as shown in Figure 1B (TRCC15), and that the copy-number imbalance reflects the loss of one translocated chromosome; direct validation of this explanation requires whole-genome sequencing.
Figure S5. Clonality analysis and phasing of somatic mutations in TRCC18 metastases. Related to Figures 4–5. (A) Determination of the variant allele fraction (VAF) threshold for clonal mutations in each tumor sample. The VAF cutoff is first determined by a k-means clustering to separate subclonal mutations (low VAF) from clonal mutations (large VAF) and then validated using truncal mutations (defined as those passing the minimum VAF cutoff in 9 or more tumor samples). The percentage of truncal mutations (clonal by definition) that do not meet the VAF cutoff determines the false negative detection rate. (B) Histograms showing the number of somatic mutations passing the minimum VAF cutoffs in a certain number of tumor samples. We operationally choose truncal mutations as those present in 9 or more samples; this threshold does not substantially affect the results of phylogenetic inference as most truncal mutations pass the cutoff in nearly all (13 or 14) samples, either considering all mutations or only those on disomic chromosomes for which no truncal mutations are lost by secondary deletions. (C) Validation of the VAF cutoff in each tumor sample based on truncal mutations identified on chr2 (no allelic imbalance in any autopsy sample). While there is a minor DNA copy-number gain of chr2 in the primary tumor, this does not impact the analysis as the primary tumor is not used for phylogenetic inference. (D) Phasing of truncal mutations to the deleted chrX and the retained chrX based on the VAFs in all tumor samples. Mutations are ordered by their locations on chrXq. Shown for each mutation are the variant allele fractions (gray dots) in all tumor samples; outlined dots indicate mutations whose VAF passes the sample-specific VAF cutoff to be classified as clonal. Mutations on the retained X are preserved in all tumors (bottom); those on the deleted X have near zero VAFs in Lung-1/Lung3 (clonal deletion), low VAFs in RPLN-2 and Lung-2 (subclonal deletion), and reduced VAFs in Liver-4/MedLN (clonal chromosome loss after whole-genome duplication). The VAF distributions of all truncal mutations on either chrX in each tumor sample are shown on the right.
Figure S6. Segmental copy-number alterations and evolution in TRCC18. Related to Figures 4–5. (A) Haplotype-specific DNA copy number (gray and black dots, 100kb intervals) and rearrangements of chr9. Shown are the data for the Liver-1 metastatic lesion (representative of ancestral alterations of the founding tumor clone), the normal kidney and normal liver samples taken at rapid autopsy showing minor chr9 losses indicating the presence of tumor cells (gray dots lower than the black), and the normal adjacent kidney tissue taken at the time of nephrectomy (primary tumor resection) showing complete allelic balance (gray dots overlaying on top of black dots). We excluded the two normal autopsy samples from mutation analysis due to the presence of contaminating tumor cells. (B) Independent deletions of chr22q detected in the MedLN sample (complete deletion), Liver-4 (terminal deletion), Lung-1 (terminal deletion with a different boundary), and RPLN-2 (complete deletion). These deletions are absent in the remaining liver metastases and have a low clonal fraction in the primary. Interestingly, all the deletions occurred to the same 22 homolog; we cannot rule out the possibility that these deletions may have arisen from progressive deletions of an unstable chr22q and therefore are related. (C) Chromothripsis of chr6 that is linked to a terminal deletion on 14q, both of which inferred to be ancestral (see Figure 4). Black arcs represent intra-chromosomal junctions, magenta vertical lines represent chr6 and chr14 breakpoints joined by inter-chromosomal junctions. The clustering of breakpoints on chr6 is very similar to the clustering of breakpoints seen in RCC16 and in our prior studies of breakage-fusion-bridge cycles88. This pattern contrasts with the simple outcome of reciprocal translocations associated with TFE3 fusions between chr1 and chrX. (D) Large segmental deletions on 17p (left) and 18p (right) inferred to be present in the founding tumor clone. The same breakpoints on 17p and 18p are seen in Liver-2 (and other Liver metastases except Liver-4), Lung metastases, and RPLN-1, but there is copy-number gain of the translocated chromosome der(17)t(17;18) in multiple metastases that may be due to positive selection; the Liver-4/MedLN lesions show a larger deletion on 17p, indicating downstream evolution of the translocated chromosome. Notably, there is sloping copy-number variation near the 17p terminus (black) in RPLN-1 that is absent in the intact homolog (gray); this pattern suggests ongoing evolution of the translocated chromosome that is consistent with varying copy-number states and breakpoints seen in the different metastatic samples90. (E) Independent deletions or losses of der(X)t(1;X) in the lung metastases (using Lung-1 as representative), MedLN/Liver-4 (using MedLN as representative), and RPLN-2. In the Lung samples, the deletion spans the entire Xq segment (centromeric to the translocation breakpoint), but a smaller 1p segment; the reciprocal fusion junction is deleted by small focal deletions near the translocation breakpoint. In the MedLN/Liver-4 samples, the loss spanned the entire der(X)t(1;X) but occurred after whole-genome duplication; therefore, this must be independent from the deletion in Lung metastases. Finally, the deletion in RPLN-2 indicates chromothripsis of der(X)t(1;X), with partial retention on chrXq and complex rearrangements between chr1p and chrX; this event is independent from MedLN/Liver-4, since it is inferred to be clonal or have occurred before WGD; it is also independent from the deletion in Lung metastases as they do not share any breakpoint.
Figure S7. Evidence for chrXp reactivation in tRCC samples. Related to Figure 6. (A) Cumulative fractions of biallelically expressed heterozygous sites (top), minor allelic fractions of transcripts from only non-escaping genes (middle), and segmented chrX copy number of samples shown in Figure 6G (tRCCs with chrXp reactivation: TRCC18, TCGA-G7–7501, TRCC17; tRCCs without chrXp reactivation: TCGA-B0–5705; ccRCC: TCGA-T7-A92I; tRCC cancer-adjacent-normal: TCGA-B0–5705-N). The cumulative distribution of biallelic expression (red) is shown together with the cumulative distribution of all transcribed genes with heterozygous polymorphisms (black). P-values are calculated by Mann-Whitney U tests comparing distributions of minor allele fraction for heterozygous sites to the left and right of the TFE3 breakpoint. (B) Normalized expression level of biallelically expressed genes (RNA MAF ≥ 0.2) in samples inferred to have chrXp reactivation and in samples without chrXp reactivation. Gene expression is normalized by the mean expression in the non-reactivated cohort. Left: TCGA, Right: current study. P-values calculated by Mann-Whitney U-test. (C) Methylation (quantified by beta value) at probes in the promoters of all non-escaping genes to the left (red, N=109 probes) or right (brown, N=316 probes) of TFE3 in each sample. Samples are grouped by chrXp reactivation status. P-values are computed by Mann-Whitney U-test. (D) Methylation at the promoters (±1kb from exon 1 start) of genes showing biallelic expression (RNA MAF ≥ 0.2) to the left of TFE3 in samples with evidence of chrXp reactivation (N=5) compared to genes showing monoallelic expression (RNA MAF < 0.2) across the entire chrX for all tRCCs (brown). P-values are computed by Mann-Whitney U test.
Figure S8. Additional examples of long-range transcriptional changes associated with XCI perturbation. Related to Figure 6. (A) An example of autosomal silencing in a female tRCC. Summary of allelic RNA fraction for chr1 at expressed germline heterozygous sites (SNVs), plotted similarly to Figure S7A and Figure 6G. Segmented DNA copy number is also shown across the chromosome. P-values calculated by Mann-Whitney U test between distributions of minor allele fraction for heterozygous sites to the left and right of the autosomal breakpoint. (B) Western blot of wildtype and TFE3 fusion proteins in a panel of cell lines with TFE3 fusions (tRCC: UOK146 [PRCC-TFE3], S-TFE [ASPSCR1-TFE3]; alveolar soft part sarcoma: ASPS-1 [ASPSCR1-TFE3], ASPS-KY [ASPSCR1-TFE3]) and TFE3 fusion-negative ccRCC cell line 786O. A single band in the tRCC lines is suggestive of only one TFE3 protein species (i.e., chrXa fusion); a double band suggests two different TFE3 proteins, one from the fusion transcript generated by chrXi translocation and the other corresponding to WT TFE3 from chrXa. (C) TFE3 western blot with doxycycline-inducible shRNA-mediated knockdown of TFE3 in ASPS-KY and S-TFE cells with target sequence in the TFE3 3’UTR, demonstrating that both bands represent TFE3 species. (D) Allelic chrX expression assessed at common polymorphic sites in S-TFE (left) and UOK146 (right) with segmented DNA copy number shown at the bottom. In the 23Mb region (chrX:26,139,329–49,028,726) where both cell lines show heterozygosity, we observe biallelic expression in sTFE but not UOK146, indicating a possible chrXi fusion in s-TFE that is accompanied by partial Xp reactivation. (E) Two examples of biallelic chrXp expression associated with chrX:autosome fusions, suggesting partial Xp reactivation (TCGA-24–1424 and TCGA-N8-A4PN), and two representative samples without chrX-reactivation (TCGA-X6-A7WB and TCGA-FU-A3TX). Data are shown as in Figure S7A. Biallelic expression was assessed by the following criteria: RNA minor allele fraction ≥ 0.2, threshold ≥ 10 DNA reads and ≥ 10 RNA reads. (F) Two examples of biallelic chrX expression associated with deletion of the X-inactivation center in cell lines from the Cancer Cell Line Encyclopedia (HCC-366 and LMSU). Deleted regions on chrX are shaded in blue. A representative example of monoallelic chrX expression is shown as reference (CL-40)..
Table S1. Sample information, clinical annotations, and sequencing metrics for tRCC WGS cohort. Related to Figure 1. The following information is listed for each sample: (A) Subject (B) Sample ID; (C) biological sex (male/female); (D) Race/ethnicity (E) sample type (FFPE or frozen); (F) TFE3 fusion partner; (G) sequencing library type (standard or linked-read WGS); (H) tumor site (primary or metastasis) (I) tumor tissue acquisition method (resection or biopsy); (J) source of normal (tumor-adjacent normal or blood); (K) treatment status at the time of tumor sample acquisition (treatment-naïve or post-treatment); (J) prior therapy (if post-treatment); In one pre-treated case (TRCC12), tumor tissue was sampled from the same metastatic site both prior to, and following progression on, systemic therapy with a tyrosine kinase inhibitor (lenvatinib) and mammalian target of rapamycin (mTOR) inhibitor (everolimus); (L) mean sequencing depth of tumor sample; (M) mean sequencing depth of the matching normal; (N,Q) tumor purity; (O,P) tumor ploidy.
Table S2. Summary of key genomic features of tRCCs, including breakpoints of TFE3 and fusion partners. Related to Figures 2 and S1.
Table S3. Somatic mutations in cancer driver genes detected in the tRCC WGS cohort. Related to Figure 1 and Figure S1.
Table S4. Primers, sgRNAs, and shRNA sequences used in this study. Related to Figure 2, Figure S3, and Figure S8.
Table S5. Catalog of biological sex, TFE3 fusions, and copy-number imbalance status of 203 tRCC cases (tab 1), and summary of allele-specific expression analysis (tab 2 and tab 3). Related to Figure 3, Figure 6, Figure S4, S7, S8. Tab 1 lists 203 tRCC samples from 11 datasets (this study; TCGA; Sun et al.; Qu et al, Durinck et al., Malouf et al., MSK Panel, MSK WES, OncoPanel, PCAWG, and Sato et al.). Tab 2 lists the mean minor allele fractions of transcripts of genes to the left or to the right of the TFE3 breakpoint, and the P-value computed by Mann-Whitney tests. Tab 3 summarizes the inference of chrXi or chrXa translocations from the RNA-Seq data of frozen tumor samples by Xp allelic expression and the following features of TFE3 or fusion partner transcripts. (1) allelic expression of SNVs within the 3’ end of TFE3 (expected to be monoallelic for chrXa fusions; biallelic for chrXi fusions, due to co-expression of wild type TFE3 from chrXa and fusion TFE3 from chrXi), (2) allelic expression of SNVs in the 3’ end of the autosomal fusion partner (expected to be biallelic for chrXa fusions; monoallelic for chrXi fusions due to silencing of the 3’ end of the autosomal partner); (3) RNA-seq reads spanning the reciprocal (TFE3-Partner) fusion junction (expected to be detectable only for chrXa fusions). These features were further integrated with the copy-number status of the reciprocal translocated chromosome (loss only tolerated for chrXi fusions) or chromosome-wide chrX loss-of-heterozygosity (only tolerated for chrXa fusions). Altogether, we were able to assign translocated haplotype (chrXi/chrXa) using one or more of these criteria in 14/15 samples. In 11 of these 14 samples, the orthogonal inferences using different features were concordant. In the remaining 3 samples (TRCC8, TRCC10, TCGA-J7–8537), we observed biallelic chrXp expression (suggestive of a chrXi fusion), but also identified junction-spanning RNA-seq reads suggesting expression of the reciprocal fusion (Partner-TFE3). TRCC8 displayed chrXq deletion and lacked XIST expression, which could indicate XIC erosion that explains the expression of the reciprocal fusion. Reciprocal fusion expression in the remaining two cases may be caused by somatic erosion of XCI91 or the incomplete silencing (partial escape) of the fusion gene92.
Table S6. Somatic mutations in TRCC18 and their co-occurrence patterns used for phylogenetic inference. Related to Figure 4 and Figures S5–6. Tab1 lists the mutant allele fractions (MAFs) of all somatic single-nucleotide substitutions in the metastatic lesions. Tab2 lists the clonality status of each mutation in each sample, and the pattern (pattern_id). Tab3 lists the the assignment of mutations with each clonality pattern (pattern_id) to different evolutionary branches, with explanations of the rationale for re-assignment during manual review (Comments).
Highlights.
TFE3 driver fusions usually arise via reciprocal, balanced rearrangements in Xp11 tRCC.
Female sex bias of tRCC explained by rearrangements of both the active and inactive X.
Active vs. inactive X rearrangements impose sex-specific evolutionary constraints.
Inactive X:autosome translocations linked to X-reactivation and/or autosome silencing.
Acknowledgements:
We are deeply grateful to the patients and their families for their generosity in contributing to this research project. We thank Dr. David Papke, M.D., Ph.D. at the Brigham and Women’s Hospital for discussions related to TFE3 fusions in rare cancers. Sample preparation and sequencing was performed at the Genomics Platform of the Broad Institute of MIT and Harvard and at the Center for Cancer Genomics at Dana-Farber Cancer Institute. Graphical abstract and illustrations in Figure panels 1A and 4A were created with BioRender.com.
Funding:
S.R.V: Doris Duke Charitable Foundation (Clinician-Scientist Development Award grant number: 2020101), Department of Defense Kidney Cancer Research Program (DoD KCRP) (W81XWH-19-1-0815), Damon Runyon-Rachleff Innovation Award (Grant Number 71-22), Claudia Adams Barr Program for Innovative Cancer Research, NCI (R01CA286652 and R01CA279044), DF/HCC Kidney SPORE (2P50CA101942-16). C-Z.Z: Claudia Adams Barr Program for Innovative Cancer Research, Breakthrough Cancer, NCI (K22CA216319 and R01CA286652). T.K.C: Dana-Farber/Harvard Cancer Center Kidney SPORE (2P50CA101942-16) and Program 5P30CA006516-56, the Kohlberg Chair at Harvard Medical School and the Trust Family, Michael Brigham, Pan Mass Challenge, Hinda and Arthur Marcus Fund and Loker Pinard Funds for Kidney Cancer Research at DFCI. P.K.: DoD KCRP Postdoctoral and Clinical Fellowship (HT94252310066). D.J.E.: Prostate Cancer Foundation Young Investigator Award, Prostate SPORE (P50 CA254861-01A1); DF/HCC Kidney SPORE (2P50CA101942-16). J.L.: DoD KCRP Postdoctoral and Clinical Fellowship Award (W81XWH-22-1-0399)
Declaration of Interests:
S.R.V.: has consulted for Jnana Therapeutics within the past 3 years; receives research support from Bayer. C.-Z.Z.: is a co-founder, consultant, and equity holder of Pillar Biosciences, a for-profit company specialized in assay development for targeted DNA sequencing. C.B: currently an employee of Inocras Inc., a for-profit company specialized in whole-genome sequencing of cancer and rare diseases. T.K.C.: institutional and/or personal, paid and/or unpaid support for research, advisory boards, consultancy, and/or honoraria past 5 years, ongoing or not, from: Alkermes, Arcus Bio, AstraZeneca, Aravive, Aveo, Bayer, Bristol Myers-Squibb, Calithera, Circle Pharma, Deciphera Pharmaceuticals, Eisai, EMD Serono, Exelixis, GlaxoSmithKline, Gilead, HiberCell, IQVA, Infinity, Ipsen, Jansen, Kanaph, Lilly, Merck, Nikang, Neomorph, Nuscan/PrecedeBio, Novartis, Oncohost, Pfizer, Roche, Sanofi/Aventis, Scholar Rock, Surface Oncology, Takeda, Tempest, Up-To-Date, CME events (Peerview, OncLive, MJH, CCO and others), outside the submitted work; institutional patents filed on molecular alterations and immunotherapy response/toxicity, and ctDNA; Equity: Tempest, Pionyr, Osel, Precede Bio, CureResponse, InnDura Therapeutics, Primium; Committees: NCCN, GU Steering Committee, ASCO (BOD 6-2024-, ESMO, ACCRU, KidneyCan; Medical writing and editorial assistance support may have been funded by Communications companies in part; No speaker’s bureau; Mentored several non-US citizens on research projects with potential funding (in part) from non-US sources/Foreign Components. The institution (Dana-Farber Cancer Institute) may have received additional independent funding of drug companies or/and royalties potentially involved in research around the subject matter. D.J.E.: Research funding to institution from Bristol-Myers Squibb, Cardiff Oncology, MiNK Therapeutics, Novartis, Sanofi, Puma. Discounted research sequencing from Foundation Medicine.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References:
- 1.Zhuang W, Liu N, Guo H, Zhang C, and Gan W (2020). Gender difference analysis of Xp11.2 translocation renal cell carcinomas’s attack rate: a meta-analysis and systematic review. BMC Urol 20, 130. 10.1186/s12894-020-00696-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Argani P (2015). MiT family translocation renal cell carcinoma. Seminars in Diagnostic Pathology 32, 103–113. 10.1053/j.semdp.2015.02.003. [DOI] [PubMed] [Google Scholar]
- 3.Choueiri TK, Lim ZD, Hirsch MS, Tamboli P, Jonasch E, McDermott DF, Dal Cin P, Corn P, Vaishampayan U, Heng DYC, et al. (2010). Vascular endothelial growth factor-targeted therapy for the treatment of adult metastatic Xp11.2 translocation renal cell carcinoma. Cancer 116, 5219–5225. 10.1002/cncr.25512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Malouf GG, Camparo P, Oudard S, Schleiermacher G, Theodore C, Rustine A, Dutcher J, Billemont B, Rixe O, Bompas E, et al. (2010). Targeted agents in metastatic Xp11 translocation/TFE3 gene fusion renal cell carcinoma (RCC): a report from the Juvenile RCC Network. Ann Oncol 21, 1834–1838. 10.1093/annonc/mdq029. [DOI] [PubMed] [Google Scholar]
- 5.Boilève A, Carlo MI, Barthélémy P, Oudard S, Borchiellini D, Voss MH, George S, Chevreau C, Landman-Parker J, Tabone M-D, et al. (2018). Immune checkpoint inhibitors in MITF family translocation renal cell carcinomas and genetic correlates of exceptional responders. j. immunotherapy cancer 6, 159. 10.1186/s40425-018-0482-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Chanzá NM, Xie W, Bilen MA, Dzimitrowicz H, Burkart J, Geynisman DM, Balakrishnan A, Bowman IA, Jain R, Stadler W, et al. (2019). Cabozantinib in advanced non-clear-cell renal cell carcinoma: a multicentre, retrospective, cohort study. Lancet Oncol 20, 581–590. 10.1016/S1470-2045(18)30907-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kauffman EC, Ricketts CJ, Rais-Bahrami S, Yang Y, Merino MJ, Bottaro DP, Srinivasan R, and Linehan WM (2014). Molecular genetics and cellular features of TFE3 and TFEB fusion kidney cancers. Nat Rev Urol 11, 465–475. 10.1038/nrurol.2014.162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bakouny Z, Sadagopan A, Ravi P, Metaferia NY, Li J, AbuHammad S, Tang S, Denize T, Garner ER, Gao X, et al. (2022). Integrative clinical and molecular characterization of translocation renal cell carcinoma. Cell Rep 38, 110190. 10.1016/j.celrep.2021.110190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Marcon J, DiNatale RG, Sanchez A, Kotecha RR, Gupta S, Kuo F, Makarov V, Sandhu A, Mano R, Silagy AW, et al. (2020). Comprehensive Genomic Analysis of Translocation Renal Cell Carcinoma Reveals Copy-Number Variations as Drivers of Disease Progression. Clin Cancer Res 26, 3629–3640. 10.1158/1078-0432.CCR-19-3283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Malouf GG, Su X, Yao H, Gao J, Xiong L, He Q, Compérat E, Couturier J, Molinié V, Escudier B, et al. (2014). Next-generation sequencing of translocation renal cell carcinoma reveals novel RNA splicing partners and frequent mutations of chromatin-remodeling genes. Clin Cancer Res 20, 4129–4140. 10.1158/1078-0432.CCR-13-3036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Malouf GG, Monzon FA, Couturier J, Molinié V, Escudier B, Camparo P, Su X, Yao H, Tamboli P, Lopez-Terrada D, et al. (2013). Genomic heterogeneity of translocation renal cell carcinoma. Clin Cancer Res 19, 4673–4684. 10.1158/1078-0432.CCR-12-3825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sun G, Chen J, Liang J, Yin X, Zhang M, Yao J, He N, Armstrong CM, Zheng L, Zhang X, et al. (2021). Integrated exome and RNA sequencing of TFE3-translocation renal cell carcinoma. Nat Commun 12, 5262. 10.1038/s41467-021-25618-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Qu Y, Wu X, Anwaier A, Feng J, Xu W, Pei X, Zhu Y, Liu Y, Bai L, Yang G, et al. (2022). Proteogenomic characterization of MiT family translocation renal cell carcinoma. Nat Commun 13, 7494. 10.1038/s41467-022-34460-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Baba M, Furuya M, Motoshima T, Lang M, Funasaki S, Ma W, Sun H-W, Hasumi H, Huang Y, Kato I, et al. (2019). TFE3 Xp11.2 Translocation Renal Cell Carcinoma Mouse Model Reveals Novel Therapeutic Targets and Identifies GPNMB as a Diagnostic Marker for Human Disease. Mol Cancer Res 17, 1613–1626. 10.1158/1541-7786.MCR-18-1235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Heard E (2004). Recent advances in X-chromosome inactivation. Current Opinion in Cell Biology 16, 247–255. 10.1016/j.ceb.2004.03.005. [DOI] [PubMed] [Google Scholar]
- 16.Wang T, Birsoy K, Hughes NW, Krupczak KM, Post Y, Wei JJ, Lander ES, and Sabatini DM (2015). Identification and characterization of essential genes in the human genome. Science 350, 1096–1101. 10.1126/science.aac7041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Blomen VA, Májek P, Jae LT, Bigenzahn JW, Nieuwenhuis J, Staring J, Sacco R, van Diemen FR, Olk N, Stukalov A, et al. (2015). Gene essentiality and synthetic lethality in haploid human cells. Science 350, 1092–1096. 10.1126/science.aac7557. [DOI] [PubMed] [Google Scholar]
- 18.Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, Donovan J, Barretina J, Boehm JS, Dobson J, Urashima M, et al. (2010). The landscape of somatic copy-number alteration across human cancers. Nature 463, 899–905. 10.1038/nature08822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Taylor AM, Shih J, Ha G, Gao GF, Zhang X, Berger AC, Schumacher SE, Wang C, Hu H, Liu J, et al. (2018). Genomic and Functional Approaches to Understanding Cancer Aneuploidy. Cancer Cell 33, 676–689.e3. 10.1016/j.ccell.2018.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bao C, Tourdot RW, Brunette GJ, Stewart C, Sun L, Baba H, Watanabe M, Agoston AT, Jajoo K, Davison JM, et al. (2023). Genomic signatures of past and present chromosomal instability in Barrett’s esophagus and early esophageal adenocarcinoma. Nat Commun 14, 6203. 10.1038/s41467-023-41805-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Haile S, Corbett RD, Bilobram S, Bye MH, Kirk H, Pandoh P, Trinh E, MacLeod T, McDonald H, Bala M, et al. (2019). Sources of erroneous sequences and artifact chimeric reads in next generation sequencing of genomic DNA from formalin-fixed paraffin-embedded samples. Nucleic Acids Res 47, e12. 10.1093/nar/gky1142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Do H, and Dobrovic A (2015). Sequence Artifacts in DNA from Formalin-Fixed Tissues: Causes and Strategies for Minimization. Clinical Chemistry 61, 64–71. 10.1373/clinchem.2014.223040. [DOI] [PubMed] [Google Scholar]
- 23.Turajlic S, Litchfield K, Xu H, Rosenthal R, McGranahan N, Reading JL, Wong YNS, Rowan A, Kanu N, Al Bakir M, et al. (2017). Insertion-and-deletion-derived tumour-specific neoantigens and the immunogenic phenotype: a pan-cancer analysis. Lancet Oncol. 18, 1009–1021. 10.1016/S1470-2045(17)30516-8. [DOI] [PubMed] [Google Scholar]
- 24.Mitchell TJ, Turajlic S, Rowan A, Nicol D, Farmery JHR, O’Brien T, Martincorena I, Tarpey P, Angelopoulos N, Yates LR, et al. (2018). Timing the Landmark Events in the Evolution of Clear Cell Renal Cell Cancer: TRACERx Renal. Cell 173, 611–623.e17. 10.1016/j.cell.2018.02.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cancer Genome Atlas Research Network (2013). Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 499, 43–49. 10.1038/nature12222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Anderson ND, de Borja R, Young MD, Fuligni F, Rosic A, Roberts ND, Hajjar S, Layeghifard M, Novokmet A, Kowalski PE, et al. (2018). Rearrangement bursts generate canonical gene fusions in bone and soft tissue tumors. Science 361. 10.1126/science.aam8419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Crompton BD, Stewart C, Taylor-Weiner A, Alexe G, Kurek KC, Calicchio ML, Kiezun A, Carter SL, Shukla SA, and Mehta SS (2014). The genomic landscape of pediatric Ewing sarcoma. Cancer discovery 4, 1326–1341. 10.1158/2159-8290.CD-13-1037. [DOI] [PubMed] [Google Scholar]
- 28.Baslan T, Morris JP, Zhao Z, Reyes J, Ho Y-J, Tsanov KM, Bermeo J, Tian S, Zhang S, Askan G, et al. (2022). Ordered and deterministic cancer genome evolution after p53 loss. Nature 608, 795–802. 10.1038/s41586-022-05082-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Sondka Z, Bamford S, Cole CG, Ward SA, Dunham I, and Forbes SA (2018). The COSMIC Cancer Gene Census: describing genetic dysfunction across all human cancers. Nat Rev Cancer 18, 696–705. 10.1038/s41568-018-0060-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wiedemeyer WR, Dunn IF, Quayle SN, Zhang J, Chheda MG, Dunn GP, Zhuang L, Rosenbluh J, Chen S, Xiao Y, et al. (2010). Pattern of retinoblastoma pathway inactivation dictates response to CDK4/6 inhibition in GBM. Proc. Natl. Acad. Sci. U.S.A 107, 11501–11506. 10.1073/pnas.1001613107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Solomon DA, Kim J-S, Jenkins S, Ressom H, Huang M, Coppa N, Mabanta L, Bigner D, Yan H, Jean W, et al. (2008). Identification of p18INK4c as a Tumor Suppressor Gene in Glioblastoma Multiforme. Cancer Research 68, 2564–2569. 10.1158/0008-5472.CAN-07-6388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Calabrese C, Davidson NR, Demircioğlu D, Fonseca NA, He Y, Kahles A, Lehmann K-V, Liu F, Shiraishi Y, Soulette CM, et al. (2020). Genomic basis for RNA alterations in cancer. Nature 578, 129–136. 10.1038/s41586-020-1970-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Shlien A, Raine K, Fuligni F, Arnold R, Nik-Zainal S, Dronov S, Mamanova L, Rosic A, Ju YS, Cooke SL, et al. (2016). Direct Transcriptional Consequences of Somatic Mutation in Breast Cancer. Cell Rep 16, 2032–2046. 10.1016/j.celrep.2016.07.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lukaszewicz A, Lange J, Keeney S, and Jasin M (2021). De novo deletions and duplications at recombination hotspots in mouse germlines. Cell 184, 5970–5984.e18. 10.1016/j.cell.2021.10.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lee JJ-K, Park S, Park H, Kim S, Lee J, Lee J, Youk J, Yi K, An Y, Park IK, et al. (2019). Tracing Oncogene Rearrangements in the Mutational History of Lung Adenocarcinoma. Cell 177, 1842–1857.e21. 10.1016/j.cell.2019.05.013. [DOI] [PubMed] [Google Scholar]
- 36.Valencia A, Cervera J, Such E, Barragán E, Bolufer P, Fuster O, Collado R, Martínez J, and Sanz MA (2009). Complex Variant t(9;22) Chromosome Translocations in Five Cases of Chronic Myeloid Leukemia. Adv Hematol 2009, 187125. 10.1155/2009/187125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.FU G-N, FAN H-Y, HAN X-J, and XIN C-L (2016). Complex chromosomal rearrangements involving five chromosomes in chronic myelogenous leukemia: A case report. Oncol Lett 11, 2651–2653. 10.3892/ol.2016.4275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Esan OA, Senft JR, and Wenger SL (2012). Patterns of BCR/ABL Gene Rearrangements in Chronic Myeloid Leukemia with Complex t(9;22) Using Fluorescence In Situ Hybridization (FISH). J Assoc Genet Technol 38, 5–7. [PubMed] [Google Scholar]
- 39.Reddy KS, and Sulcova V (2000). A FISH Study of Variant Philadelphia Rearrangements. Cancer Genetics and Cytogenetics 118, 121–131. 10.1016/S0165-4608(99)00187-9. [DOI] [PubMed] [Google Scholar]
- 40.Fitzgerald PH, and Morris CM (1991). Complex chromosomal translocations in the Philadelphia chromosome leukemias. Cancer Genetics and Cytogenetics 57, 143–151. 10.1016/0165-4608(91)90145-K. [DOI] [PubMed] [Google Scholar]
- 41.Robbiani DF, Bunting S, Feldhahn N, Bothmer A, Camps J, Deroubaix S, McBride KM, Klein IA, Stone G, Eisenreich TR, et al. (2009). AID produces DNA double-strand breaks in non-Ig genes and mature B cell lymphomas with reciprocal chromosome translocations. Mol Cell 36, 631–641. 10.1016/j.molcel.2009.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ghezraoui H, Piganeau M, Renouf B, Renaud J-B, Sallmyr A, Ruis B, Oh S, Tomkinson AE, Hendrickson EA, Giovannangeli C, et al. (2014). Chromosomal Translocations in Human Cells Are Generated by Canonical Nonhomologous End-Joining. Molecular Cell 55, 829–842. 10.1016/j.molcel.2014.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wang P, Zhou R-H, Zou Y, Jackson-Cook CK, and Povirk LF (1997). Highly conservative reciprocal translocations formed by apparent joining of exchanged DNA double-strand break ends. Proceedings of the National Academy of Sciences 94, 12018–12023. 10.1073/pnas.94.22.12018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Torres R, Martin MC, Garcia A, Cigudosa JC, Ramirez JC, and Rodriguez-Perales S (2014). Engineering human tumour-associated chromosomal translocations with the RNA-guided CRISPR–Cas9 system. Nat Commun 5, 3964. 10.1038/ncomms4964. [DOI] [PubMed] [Google Scholar]
- 45.Brunet E, and Jasin M (2018). Induction of chromosomal translocations with CRISPR-Cas9 and other nucleases: Understanding the repair mechanisms that give rise to translocations. Adv Exp Med Biol 1044, 15–25. 10.1007/978-981-13-0593-1_2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Culver-Cochran AE, and Chadwick BP (2012). The WSTF-ISWI Chromatin Remodeling Complex Transiently Associates with the Human Inactive X Chromosome during Late S-Phase Prior to BRCA1 and γ-H2AX. PLOS ONE 7, e50023. 10.1371/journal.pone.0050023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Chadwick BP (2007). Variation in Xi chromatin organization and correlation of the H3K27me3 chromatin territories to transcribed sequences by microarray analysis. Chromosoma 116, 147–157. 10.1007/s00412-006-0085-1. [DOI] [PubMed] [Google Scholar]
- 48.Kuznetsova AV, Kurinov AM, and Aleksandrova MA (2014). Cell Models to Study Regulation of Cell Transformation in Pathologies of Retinal Pigment Epithelium. J Ophthalmol 2014, 801787. 10.1155/2014/801787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Yuan Y, Liu L, Chen H, Wang Y, Xu Y, Mao H, Li J, Mills GB, Shu Y, Li L, et al. (2016). Comprehensive Characterization of Molecular Differences in Cancer between Male and Female Patients. Cancer Cell 29, 711–722. 10.1016/j.ccell.2016.04.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Sharp AJ, Spotswood HT, Robinson DO, Turner BM, and Jacobs PA (2002). Molecular and cytogenetic analysis of the spreading of X inactivation in X;autosome translocations. Hum Mol Genet 11, 3145–3156. 10.1093/hmg/11.25.3145. [DOI] [PubMed] [Google Scholar]
- 51.Jiang J, Jing Y, Cost GJ, Chiang J-C, Kolpa HJ, Cotton AM, Carone DM, Carone BR, Shivak DA, Guschin DY, et al. (2013). Translating dosage compensation to trisomy 21. Nature 500, 296–300. 10.1038/nature12394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Hall LL, Byron M, Sakai K, Carrel L, Willard HF, and Lawrence JB (2002). An ectopic human XIST gene can induce chromosome inactivation in postdifferentiation human HT-1080 cells. PNAS 99, 8677–8682. 10.1073/pnas.132468999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.von Ahlfen S, Missel A, Bendrat K, and Schlumpberger M (2007). Determinants of RNA quality from FFPE samples. PLoS One 2, e1261. 10.1371/journal.pone.0001261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Bossel Ben-Moshe N, Gilad S, Perry G, Benjamin S, Balint-Lahat N, Pavlovsky A, Halperin S, Markus B, Yosepovich A, Barshack I, et al. (2018). mRNA-seq whole transcriptome profiling of fresh frozen versus archived fixed tissues. BMC Genomics 19, 419. 10.1186/s12864-018-4761-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Disteche CM, and Berletch JB (2015). X-chromosome inactivation and escape. Journal of genetics 94, 591. 10.1007/s12041-015-0574-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Sadagopan A, Nasim IT, Li J, Achom M, Zhang C-Z, and Viswanathan SR (2022). Somatic XIST activation and features of X chromosome inactivation in male human cancers. Cell Syst 13, 932–944.e5. 10.1016/j.cels.2022.10.002. [DOI] [PubMed] [Google Scholar]
- 57.Yu B, Qi Y, Li R, Shi Q, Satpathy AT, and Chang HY (2021). B cell-specific XIST complex enforces X-inactivation and restrains atypical B cells. Cell 184, 1790–1803.e17. 10.1016/j.cell.2021.02.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Fujiwara T, Nakata E, Kunisada T, Ozaki T, and Kawai A (2022). Alveolar soft part sarcoma: progress toward improvement in survival? A population-based study. BMC Cancer 22, 891. 10.1186/s12885-022-09968-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Dermawan JK, Azzato EM, Billings SD, Fritchie KJ, Aubert S, Bahrami A, Barisella M, Baumhoer D, Blum V, Bode B, et al. (2021). YAP1-TFE3-fused hemangioendothelioma: a multi-institutional clinicopathologic study of 24 geneticallyconfirmed cases. Modern Pathology 34, 2211–2221. 10.1038/s41379-021-00879-7. [DOI] [PubMed] [Google Scholar]
- 60.Sturm D, Orr BA, Toprak UH, Hovestadt V, Jones DTW, Capper D, Sill M, Buchhalter I, Northcott PA, Leis I, et al. (2016). New Brain Tumor Entities Emerge from Molecular Classification of CNS-PNETs. Cell 164, 1060–1072. 10.1016/j.cell.2016.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Tauziède-Espariat A, Figarella-Branger D, Métais A, Uro-Coste E, Maurage C-A, Lhermitte B, Aline-Fardin A, Hasty L, Vasiljevic A, Chiforeanu D, et al. (2023). CNS neuroblastoma, FOXR2-activated and its mimics: a relevant panel approach for work-up and accurate diagnosis of this rare neoplasm. Acta Neuropathol Commun 11, 43. 10.1186/s40478-023-01536-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Korshunov A, Okonechnikov K, Schmitt-Hoffner F, Ryzhova M, Sahm F, Stichel D, Schrimpf D, Reuss DE, Sievers P, Suwala AK, et al. (2021). Molecular analysis of pediatric CNS-PNET revealed nosologic heterogeneity and potent diagnostic markers for CNS neuroblastoma with FOXR2-activation. Acta Neuropathol Commun 9, 20. 10.1186/s40478-021-01118-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Poh B, Koso H, Momota H, Komori T, Suzuki Y, Yoshida N, Ino Y, Todo T, and Watanabe S (2019). Foxr2 promotes formation of CNS-embryonal tumors in a Trp53-deficient background. Neuro-Oncology 21, 993–1004. 10.1093/neuonc/noz067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Gao Q, Liang W-W, Foltz SM, Mutharasu G, Jayasinghe RG, Cao S, Liao W-W, Reynolds SM, Wyczalkowski MA, Yao L, et al. (2018). Driver Fusions and Their Implications in the Development and Treatment of Human Cancers. Cell Reports 23, 227–238.e3. 10.1016/j.celrep.2018.03.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Ricketts CJ, De Cubas AA, Fan H, Smith CC, Lang M, Reznik E, Bowlby R, Gibb EA, Akbani R, Beroukhim R, et al. (2018). The Cancer Genome Atlas Comprehensive Molecular Characterization of Renal Cell Carcinoma. Cell Rep 23, 313–326.e5. 10.1016/j.celrep.2018.03.075. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Baca SC, Prandi D, Lawrence MS, Mosquera JM, Romanel A, Drier Y, Park K, Kitabayashi N, MacDonald TY, Ghandi M, et al. (2013). Punctuated evolution of prostate cancer genomes. Cell 153, 666–677. 10.1016/j.cell.2013.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Xia P, Zhang L, Li P, Liu E, Li W, Zhang J, Li H, Su X, and Jiang G (2021). Molecular characteristics and clinical outcomes of complex ALK rearrangements identified by next-generation sequencing in non-small cell lung cancers. J Transl Med 19, 308. 10.1186/s12967-021-02982-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Kodama T, Motoi N, Ninomiya H, Sakamoto H, Kitada K, Tsukaguchi T, Satoh Y, Nomura K, Nagano H, Ishii N, et al. (2014). A Novel Mechanism of EML4-ALK Rearrangement Mediated by Chromothripsis in a Patient-Derived Cell Line. Journal of Thoracic Oncology 9, 1638–1646. 10.1097/JTO.0000000000000311. [DOI] [PubMed] [Google Scholar]
- 69.Lee JJ-K, Jung YL, Cheong T-C, Espejo Valle-Inclan J, Chu C, Gulhan DC, Ljungström V, Jin H, Viswanadham VV, Watson EV, et al. (2023). ERα-associated translocations underlie oncogene amplifications in breast cancer. Nature, 1–9. 10.1038/s41586-023-06057-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Lieber MR, Gu J, Lu H, Shimazaki N, and Tsai AG (2010). Nonhomologous DNA End Joining (NHEJ) and Chromosomal Translocations in Humans. Subcell Biochem 50, 279–296. 10.1007/978-90-481-3471-7_14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Richart L, Picod-Chedotel M-L, Wassef M, Macario M, Aflaki S, Salvador MA, Héry T, Dauphin A, Wicinski J, Chevrier V, et al. (2022). XIST loss impairs mammary stem cell differentiation and increases tumorigenicity through Mediator hyperactivation. Cell 185, 2164–2183.e25. 10.1016/j.cell.2022.04.034. [DOI] [PubMed] [Google Scholar]
- 72.Weiner AC, Williams MJ, Shi H, Vázquez-García I, Salehi S, Rusk N, Aparicio S, Shah SP, and McPherson A (2023). Single-cell DNA replication dynamics in genomically unstable cancers. bioRxiv, 2023.04.10.536250. 10.1101/2023.04.10.536250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Edgren G, Liang L, Adami H-O, and Chang ET (2012). Enigmatic sex disparities in cancer incidence. Eur J Epidemiol 27, 187–196. 10.1007/s10654-011-9647-5. [DOI] [PubMed] [Google Scholar]
- 74.Dorak MT, and Karpuzoglu E (2012). Gender differences in cancer susceptibility: an inadequately addressed issue. Front Genet 3, 268. 10.3389/fgene.2012.00268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Wu JN, and Roberts CWM (2013). ARID1A Mutations in Cancer: Another Epigenetic Tumor Suppressor? Cancer Discovery 3, 35–43. 10.1158/2159-8290.CD-12-0361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76.Sun X, Wang SC, Wei Y, Luo X, Jia Y, Li L, Gopal P, Zhu M, Nassour I, Chuang J-C, et al. (2017). Arid1a Has Context-Dependent Oncogenic and Tumor Suppressor Functions in Liver Cancer. Cancer Cell 32, 574–589.e6. 10.1016/j.ccell.2017.10.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 77.Linehan WM, and Ricketts CJ (2019). The Cancer Genome Atlas of renal cell carcinoma: findings and clinical implications. Nat Rev Urol 16, 539–552. 10.1038/s41585-019-0211-5. [DOI] [PubMed] [Google Scholar]
- 78.Fukagawa A, Hama N, Totoki Y, Nakamura H, Arai Y, Saito-Adachi M, Maeshima A, Matsui Y, Yachida S, Ushiku T, et al. (2023). Genomic and epigenomic integrative subtypes of renal cell carcinoma in a Japanese cohort. Nat Commun 14, 8383. 10.1038/s41467-023-44159-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 79.Tsutsumi M, Hattori H, Akita N, Maeda N, Kubota T, Horibe K, Fujita N, Kawai M, Shinkai Y, Kato M, et al. (2019). A female patient with retinoblastoma and severe intellectual disability carrying an X;13 balanced translocation without rearrangement in the RB1 gene: a case report. BMC Medical Genomics 12, 182. 10.1186/s12920-019-0640-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 80.Yang L, Kirby JE, Sunwoo H, and Lee JT (2016). Female mice lacking Xist RNA show partial dosage compensation and survive to term. Genes Dev. 30, 1747–1760. 10.1101/gad.281162.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 81.Yang L, Yildirim E, Kirby JE, Press W, and Lee JT (2020). Widespread organ tolerance to Xist loss and X reactivation except under chronic stress in the gut. PNAS 117, 4262–4272. 10.1073/pnas.1917203117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 82.Yildirim E, Kirby JE, Brown DE, Mercier FE, Sadreyev RI, Scadden DT, and Lee JT (2013). Xist RNA Is a Potent Suppressor of Hematologic Cancer in Mice. Cell 152, 727–742. 10.1016/j.cell.2013.01.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 83.Aladjem MI, and Fu H (2014). A new light on DNA replication from the inactive X chromosome. Bioessays 36, 591–597. 10.1002/bies.201400021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.Giorgetti L, Lajoie BR, Carter AC, Attia M, Zhan Y, Xu J, Chen CJ, Kaplan N, Chang HY, Heard E, et al. (2016). Structural organization of the inactive X chromosome in the mouse. Nature 535, 575–579. 10.1038/nature18589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 85.Wang C-Y, Jégu T, Chu H-P, Oh HJ, and Lee JT (2018). SMCHD1 Merges Chromosome Compartments and Assists Formation of Super-Structures on the Inactive X. Cell 174, 406–421.e25. 10.1016/j.cell.2018.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 86.Bauer M, Vidal E, Zorita E, Üresin N, Pinter SF, Filion GJ, and Payer B (2021). Chromosome compartments on the inactive X guide TAD formation independently of transcription during X-reactivation. Nat Commun 12, 3499. 10.1038/s41467-021-23610-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 87.Casas-Delucchi CS, Brero A, Rahn H-P, Solovei I, Wutz A, Cremer T, Leonhardt H, and Cardoso MC (2011). Histone acetylation controls the inactive X chromosome replication dynamics. Nat Commun 2, 222. 10.1038/ncomms1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Umbreit NT, Zhang C-Z, Lynch LD, Blaine LJ, Cheng AM, Tourdot R, Sun L, Almubarak HF, Judge K, Mitchell TJ, et al. (2020). Mechanisms generating cancer genome complexity from a single cell division error. Science 368, eaba0712. 10.1126/science.aba0712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 89.Ha G, Roth A, Khattra J, Ho J, Yap D, Prentice LM, Melnyk N, McPherson A, Bashashati A, Laks E, et al. (2014). TITAN: inference of copy number architectures in clonal cell populations from tumor whole-genome sequence data. Genome Res. 24, 1881–1893. 10.1101/gr.180281.114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Bao C, Tourdot RW, Brunette GJ, Stewart C, Sun L, Baba H, Watanabe M, Agoston AT, Jajoo K, Davison JM, et al. (2023). Genomic signatures of past and present chromosomal instability in Barrett’s esophagus and early esophageal adenocarcinoma. Nat Commun 14, 6203. 10.1038/s41467-023-41805-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 91.Chaligné R, Popova T, Mendoza-Parra M-A, Saleem M-AM, Gentien D, Ban K, Piolot T, Leroy O, Mariani O, Gronemeyer H, et al. (2015). The inactive X chromosome is epigenetically unstable and transcriptionally labile in breast cancer. Genome Res 25, 488–503. 10.1101/gr.185926.114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 92.Carrel L, and Willard HF (2005). X-inactivation profile reveals extensive variability in X-linked gene expression in females. Nature 434, 400–404. 10.1038/nature03479. [DOI] [PubMed] [Google Scholar]
- 93.Viswanathan SR, Ha G, Hoff AM, Wala JA, Carrot-Zhang J, Whelan CW, Haradhvala NJ, Freeman SS, Reed SC, Rhoades J, et al. (2018). Structural Alterations Driving Castration-Resistant Prostate Cancer Revealed by Linked-Read Genome Sequencing. Cell 174, 433–447.e19. 10.1016/j.cell.2018.05.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 94.Slyper M, Porter CBM, Ashenberg O, Waldman J, Drokhlyansky E, Wakiro I, Smillie C, Smith-Rosario G, Wu J, Dionne D, et al. (2020). A single-cell and single-nucleus RNA-Seq toolbox for fresh and frozen human tumors. Nature Medicine 26, 792–802. 10.1038/s41591-020-0844-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 95.Tourdot RW, Brunette GJ, Pinto RA, and Zhang C-Z (2021). Determination of complete chromosomal haplotypes by bulk DNA sequencing. Genome Biology 22, 139. 10.1186/s13059-021-02330-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 96.Li H (2011). A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data. Bioinformatics 27, 2987–2993. 10.1093/bioinformatics/btr509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 97.Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, Whitwham A, Keane T, McCarthy SA, Davies RM, et al. (2021). Twelve years of SAMtools and BCFtools. GigaScience 10, giab008. 10.1093/gigascience/giab008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 98.Crowdis J, He MX, Reardon B, and Van Allen EM (2020). CoMut: visualizing integrated molecular information with comutation plots. Bioinformatics 36, 4348–4349. 10.1093/bioinformatics/btaa554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 99.Loh P-R, Danecek P, Palamara PF, Fuchsberger C, A Reshef Y, K Finucane H, Schoenherr S, Forer L, McCarthy S, Abecasis GR, et al. (2016). Reference-based phasing using the Haplotype Reference Consortium panel. Nat Genet 48, 1443–1448. 10.1038/ng.3679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 100.Krzywinski M, Schein J, Birol İ, Connors J, Gascoyne R, Horsman D, Jones SJ, and Marra MA (2009). Circos: An information aesthetic for comparative genomics. Genome Res. 19, 1639–1645. 10.1101/gr.092759.109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 101.Mölder F, Jablonski KP, Letcher B, Hall MB, Tomkins-Tinch CH, Sochat V, Forster J, Lee S, Twardziok SO, Kanitz A, et al. (2021). Sustainable data analysis with Snakemake. Preprint at F1000Research, https://doi.org/10.12688/f1000research.29032.1 10.12688/f1000research.29032.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 102.Riester M, Singh AP, Brannon AR, Yu K, Campbell CD, Chiang DY, and Morrissey MP (2016). PureCN: copy number calling and SNV classification using targeted short read sequencing. Source Code Biol Med 11, 13. 10.1186/s13029-016-0060-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 103.Cancer Genome Atlas Research Network, Linehan WM, Spellman PT, Ricketts CJ, Creighton CJ, Fei SS, Davis C, Wheeler DA, Murray BA, Schmidt L, et al. (2016). Comprehensive Molecular Characterization of Papillary Renal-Cell Carcinoma. N Engl J Med 374, 135–145. 10.1056/NEJMoa1505917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 104.Davis CF, Ricketts CJ, Wang M, Yang L, Cherniack AD, Shen H, Buhay C, Kang H, Kim SC, Fahey CC, et al. (2014). The somatic genomic landscape of chromophobe renal cell carcinoma. Cancer Cell 26, 319–330. 10.1016/j.ccr.2014.07.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 105.Ghandi M, Huang FW, Jané-Valbuena J, Kryukov GV, Lo CC, McDonald ER, Barretina J, Gelfand ET, Bielski CM, Li H, et al. (2019). Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569, 503–508. 10.1038/s41586-019-1186-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 106.ICGC/TCGA Pan-Cancer Analysis of Whole Genomes Consortium (2020). Pan-cancer analysis of whole genomes. Nature 578, 82–93. 10.1038/s41586-020-1969-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 107.Durinck S, Stawiski EW, Pavía-Jiménez A, Modrusan Z, Kapur P, Jaiswal BS, Zhang N, Toffessi-Tcheuyap V, Nguyen TT, Pahuja KB, et al. (2015). Spectrum of diverse genomic alterations define non-clear cell renal carcinoma subtypes. Nat Genet 47, 13–21. 10.1038/ng.3146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 108.Garcia EP, Minkovsky A, Jia Y, Ducar MD, Shivdasani P, Gong X, Ligon AH, Sholl LM, Kuo FC, MacConaill LE, et al. (2017). Validation of OncoPanel: A Targeted Next-Generation Sequencing Assay for the Detection of Somatic Variants in Cancer. Arch Pathol Lab Med 141, 751–758. 10.5858/arpa.2016-0527-OA. [DOI] [PubMed] [Google Scholar]
- 109.Sato Y, Yoshizato T, Shiraishi Y, Maekawa S, Okuno Y, Kamura T, Shimamura T, Sato-Otsubo A, Nagae G, Suzuki H, et al. (2013). Integrated molecular analysis of clear-cell renal cell carcinoma. Nat Genet 45, 860–867. 10.1038/ng.2699. [DOI] [PubMed] [Google Scholar]
- 110.Haas BJ, Dobin A, Li B, Stransky N, Pochet N, and Regev A (2019). Accuracy assessment of fusion transcript detection via read-mapping and de novo fusion transcript assembly-based methods. Genome Biol 20, 213. 10.1186/s13059-019-1842-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 111.Zehir A, Benayed R, Shah RH, Syed A, Middha S, Kim HR, Srinivasan P, Gao J, Chakravarty D, Devlin SM, et al. (2017). Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients. Nat Med 23, 703–713. 10.1038/nm.4333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 112.Thorvaldsdottir H, Robinson JT, and Mesirov JP (2013). Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Briefings in Bioinformatics 14, 178–192. 10.1093/bib/bbs017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 113.Hao Y, Hao S, Andersen-Nissen E, Mauck WM, Zheng S, Butler A, Lee MJ, Wilk AJ, Darby C, Zager M, et al. (2021). Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e29. 10.1016/j.cell.2021.04.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 114.Wolock SL, Lopez R, and Klein AM (2019). Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst 8, 281–291.e9. 10.1016/j.cels.2018.11.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 115.Tickle Timothy, Tirosh Itay, Georgescu Christophe, Brown Maxwell, Brian Haas infercnv. (Bioconductor). https://doi.org/10.18129/B9.BIOC.INFERCNV 10.18129/B9.BIOC.INFERCNV. [DOI] [Google Scholar]
- 116.Wala JA, Bandopadhayay P, Greenwald NF, O’Rourke R, Sharpe T, Stewart C, Schumacher S, Li Y, Weischenfeldt J, Yao X, et al. (2018). SvABA: genome-wide detection of structural variants and indels by local assembly. Genome Res. 28, 581–591. 10.1101/gr.221028.117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 117.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, and Gingeras TR (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 118.Li B, and Dewey CN (2011). RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 12, 323. 10.1186/1471-2105-12-323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 119.Frankish A, Diekhans M, Ferreira A-M, Johnson R, Jungreis I, Loveland J, Mudge JM, Sisu C, Wright J, Armstrong J, et al. (2019). GENCODE reference annotation for the human and mouse genomes. Nucleic Acids Res 47, D766–D773. 10.1093/nar/gky955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 120.Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, and Salzberg SL (2013). TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14, R36. 10.1186/gb-2013-14-4-r36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 121.Li H, and Durbin R (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760. 10.1093/bioinformatics/btp324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 122.McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, et al. (2010). The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 20, 1297–1303. 10.1101/gr.107524.110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 123.Cotton AM, Price EM, Jones MJ, Balaton BP, Kobor MS, and Brown CJ (2015). Landscape of DNA methylation on the X chromosome reflects CpG density, functional chromatin state and X-chromosome inactivation. Hum Mol Genet 24, 1528–1539. 10.1093/hmg/ddu564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 124.Cunningham F, Allen JE, Allen J, Alvarez-Jarreta J, Amode MR, Armean IM, Austine-Orimoloye O, Azov AG, Barnes I, Bennett R, et al. (2022). Ensembl 2022. Nucleic Acids Res 50, D988–D995. 10.1093/nar/gkab1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 125.Robinson JT, Thorvaldsdóttir H, Winckler W, Guttman M, Lander ES, Getz G, and Mesirov JP (2011). Integrative genomics viewer. Nat Biotechnol 29, 24–26. 10.1038/nbt.1754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 126.Martin FJ, Amode MR, Aneja A, Austine-Orimoloye O, Azov AG, Barnes I, Becker A, Bennett R, Berry A, Bhai J, et al. (2023). Ensembl 2023. Nucleic Acids Res 51, D933–D941. 10.1093/nar/gkac958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 127.Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, and Haussler D (2002). The human genome browser at UCSC. Genome Res 12, 996–1006. 10.1101/gr.229102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 128.Qin Q, Mei S, Wu Q, Sun H, Li L, Taing L, Chen S, Li F, Liu T, Zang C, et al. (2016). ChiLin: a comprehensive ChIP-seq and DNase-seq quality control and analysis pipeline. BMC Bioinformatics 17, 404. 10.1186/s12859-016-1274-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 129.Sanjana NE, Shalem O, and Zhang F (2014). Improved vectors and genome-wide libraries for CRISPR screening. Nat Methods 11, 783–784. 10.1038/nmeth.3047. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Recurrent mutations and copy number alterations in tRCCs revealed by WGS. Related to Figure 1. (A) Recurrent single-nucleotide substitution and insertion/deletion mutations of Tier 1 cancer genes in the cancer gene census (CGC) in the tRCC WGS cohort. Only cancer genes with mutations predicted to have high functional impact in at least one sample, or high/moderate impact in two or more samples are included. FFPE samples (TRCC3/4/7/9/15) were excluded in the calculation of mutation frequency (NA) due to the prevalence of artifacts in FFPE libraries21,22 and the low clonal fractions of tumor cells in some FFPE samples (e.g. TRCC3/4), but potentially pathogenic mutations of CGC genes are provided for informative purposes. We used both copy-number alterations and point mutations to assess the clonal fractions of tumor cells (Purity). For TRCC3 and TRCC9, purity was not assessed due to the absence of large copy-number alterations and the uncertainty of point mutation detection in FFPE DNA. (B) Genome-wide distribution of copy number gains (red) and losses (blue) in 100 kb bins. Known cancer drivers in peak/valley regions are annotated; two insets highlight recurrent focal deletions on chr1p (spanning CDKN2C, ARID1A) and 9p (spanning CDKN2A/2B). (C) Diagrammatic view of TFE3 fusions in 9 cases with both DNA- and RNA-sequencing (Table S2). In each sample, the exons included in the oncogenic fusion transcripts (Partner-TFE3) as detected by RNA-seq are highlighted in gray-shaded boxes of the TFE3 gene (top, 3’ of the fusion transcript) and its partner (bottom, 5’ of the fusion transcript). Exons included in the reciprocal fusion (TFE3-Partner) when detected are outlined with dotted boxes (5’ of TFE3 and 3’ of the partner). TFE3 is anti-sense. Fusion partners with shaded boxes on the left are sense, those with shaded boxes on the right are anti-sense. DNA breakpoints are denoted by vertical lines with dark blue lollipops. Note for SFPQ, isoform NR_136703.1 is used based on the RNA-Seq data.
Figure S2. tRCC genome complexity and evolution revealed by WGS. Related to Figure 1. (A) CIRCOS plots showing haplotype-specific DNA copy number (red and blue for each parental haplotype), inter-chromosomal translocations (purple arcs), and long-range intrachromosomal rearrangements (green arcs, breakpoints >1Mb apart). For TRCC18 (boxed), the primary tumor sample is shown, with haplotype-specific DNA copy number across all chromosomes (top CIRCOS) or selected chromosomes displaying segmental copy number alterations (bottom CIRCOS); rearrangements are indicated on the latter CIRCOS. Other representative metastases from TRCC18 are shown separately in Figures 4–5 and Figures S5–S6. In both the primary tumor and the resected metastatic lesion from subject RCC16, the absence of a detectable TFE3 fusion suggests that it may have been misclassified (CIRCOS for primary shown as representative). The following copy-number changes/rearrangement of this genome are consistent with features of a clear-cell renal cell carcinoma (ccRCC): (1) The presence of 3p loss is a hallmark feature of ccRCC; moreover, the 3p loss is coupled to a terminal deletion on 1p accompanied with chromothripsis, a frequent rearrangement pattern seen in ccRCCs24. (2) The clustering of breakpoints on chr4, 8, 10, 11, 15, and 17 near the boundaries of large segmental SCNAs is consistent with rearrangement outcomes generated by the breakage of multi-chromosome bridges, including both regional chromothripsis and junctions with multiple tandem short insertions24,88. Although a similar pattern was observed in TRCC18 between chr6 and chr14, such events are rare in tRCCs and we never observed multi-chromosomal (3 or more) rearrangements as seen in this case. (3) The tumor genome is inferred to be 4N based on the distribution of allelic depths showing four different peaks at equal spacing indicating allelic copy-number states of 0, 1, 2, and 3; moreover, the presence of many segmental SCNAs at odd copy-number states suggests downstream SCNA evolution after ancestral whole-genome duplication. The sequence of early WGD followed by downstream evolution contrasts with tRCC genomes that are usually 2N or show few alterations either before or after WGD (see Figures 4–5). (B) Haplotype-specific coverage for chr5 and chr6 in the pre- (left) and post-treatment (right) samples from TRCC12. The integer copy-number states are determined from truncal deletions of chr3q and chr17p shared by both samples. Allelic DNA copy number of chr5 and chr6 indicates clonal deletion of 6q of the B haplotype (6Bq) and duplication of 5q of the A haplotype (5Aq) in the post-treatment clone (right), but subclonal losses of 6q and gains of 5q of both haplotypes in the pre-treated clone (left). One possible explanation is that the pre-treatment tumor has two subclonal populations: one containing an extra copy of 5Aq but missing 6Bq, the other containing an extra copy of 5Bq but missing 6Aq; and the post-treatment tumor only consists of the first. Interestingly, both subclones have disomic 6p, monosomic 6q, and trisomic 5q (from both homologs) that match the average DNA copy number of these arms across all tRCCs (Figure S1B).
Figure S3. CRISPR/Cas9 engineering of TFE3 fusions in 293T cells. Related to Figure 2. (A) Genomic locations of sgRNAs used to engineer PRCC-TFE3 fusions via CRISPR/Cas9 and PCR primers for detecting both PRCC-TFE3 and TFE3-PRCC fusions. (B) DNA gel showing PRCC-TFE3 fusions in the bulk edited population of 293T cells. (C) Distribution of the number of deleted or inserted nucleotides in each PCR amplicon relative to the predominant PRCC-TFE3 breakpoints (chr1:156777745 and chrX:49038604) inferred from the population. (D) DNA gel showing TFE3-PRCC fusion in the edited bulk population. (E) Distribution of the number of deleted or inserted nucleotides in each PCR amplicon relative to the predominant TFE3-PRCC breakpoints (chrX:49038605; chr1:156777746). Two amplicons containing insertions of sequences of Cas9 DNA (~75bp) were excluded. (F) Expressed PRCC-TFE3 fusion transcripts detected via RT-PCR across the predicted RNA fusion breakpoint in the edited bulk population. (G) Sanger sequencing trace confirming PRCC-TFE3 fusion junction.
Figure S4. Copy number imbalance of TFE3 translocations in tRCC. Related to Figure 3. (A) Minor allele fraction (top) and normalized read depth (tumor:normal) from the whole-genome sequencing data of a tumor (TCGA BP-4756, female) showing copy-number gain of the translocated chromosome der(1)t(1;X) containing an oncogenic SFPQ-TFE3 fusion. Regions with copy number imbalance in chr1 (top) and chrX (bottom) are highlighted in orange. (B) A tRCC tumor (TCGA BQ-5887, male) showing copy-number gain of der(X)t(1;X) with the reciprocal TFE3-PRCC fusion. Plots are similar to A. Schematic diagrams of the duplicated chromosome segments are shown below. (C-E) Three tRCCs with SFPQ-TFE3 fusions showing (C) deletion of both segments in the reciprocal translocation der(X)t(1;X); (D) deletion of the segment telomeric to SFPQ; and (E) duplication of both segments in the oncogenic translocation der(1)t(1;X). Note that deletion of the TFE3 fusion is never observed. Shown for each sample are the minor allele fractions, normalized read depths, and segmented allelic copy number of both chrX and the partner chromosome (chr1). In all cases, DNA copy number was calculated by TITAN89, which does not perform local haplotype inference. For cases with copy-number imbalance on one breakpoint (SFPQ) but not the other (TFE3) as shown in D, a plausible explanation is that the TFE3 fusion arises from chromoplexy (multi-way balanced translocations) as shown in Figure 1B (TRCC15), and that the copy-number imbalance reflects the loss of one translocated chromosome; direct validation of this explanation requires whole-genome sequencing.
Figure S5. Clonality analysis and phasing of somatic mutations in TRCC18 metastases. Related to Figures 4–5. (A) Determination of the variant allele fraction (VAF) threshold for clonal mutations in each tumor sample. The VAF cutoff is first determined by a k-means clustering to separate subclonal mutations (low VAF) from clonal mutations (large VAF) and then validated using truncal mutations (defined as those passing the minimum VAF cutoff in 9 or more tumor samples). The percentage of truncal mutations (clonal by definition) that do not meet the VAF cutoff determines the false negative detection rate. (B) Histograms showing the number of somatic mutations passing the minimum VAF cutoffs in a certain number of tumor samples. We operationally choose truncal mutations as those present in 9 or more samples; this threshold does not substantially affect the results of phylogenetic inference as most truncal mutations pass the cutoff in nearly all (13 or 14) samples, either considering all mutations or only those on disomic chromosomes for which no truncal mutations are lost by secondary deletions. (C) Validation of the VAF cutoff in each tumor sample based on truncal mutations identified on chr2 (no allelic imbalance in any autopsy sample). While there is a minor DNA copy-number gain of chr2 in the primary tumor, this does not impact the analysis as the primary tumor is not used for phylogenetic inference. (D) Phasing of truncal mutations to the deleted chrX and the retained chrX based on the VAFs in all tumor samples. Mutations are ordered by their locations on chrXq. Shown for each mutation are the variant allele fractions (gray dots) in all tumor samples; outlined dots indicate mutations whose VAF passes the sample-specific VAF cutoff to be classified as clonal. Mutations on the retained X are preserved in all tumors (bottom); those on the deleted X have near zero VAFs in Lung-1/Lung3 (clonal deletion), low VAFs in RPLN-2 and Lung-2 (subclonal deletion), and reduced VAFs in Liver-4/MedLN (clonal chromosome loss after whole-genome duplication). The VAF distributions of all truncal mutations on either chrX in each tumor sample are shown on the right.
Figure S6. Segmental copy-number alterations and evolution in TRCC18. Related to Figures 4–5. (A) Haplotype-specific DNA copy number (gray and black dots, 100kb intervals) and rearrangements of chr9. Shown are the data for the Liver-1 metastatic lesion (representative of ancestral alterations of the founding tumor clone), the normal kidney and normal liver samples taken at rapid autopsy showing minor chr9 losses indicating the presence of tumor cells (gray dots lower than the black), and the normal adjacent kidney tissue taken at the time of nephrectomy (primary tumor resection) showing complete allelic balance (gray dots overlaying on top of black dots). We excluded the two normal autopsy samples from mutation analysis due to the presence of contaminating tumor cells. (B) Independent deletions of chr22q detected in the MedLN sample (complete deletion), Liver-4 (terminal deletion), Lung-1 (terminal deletion with a different boundary), and RPLN-2 (complete deletion). These deletions are absent in the remaining liver metastases and have a low clonal fraction in the primary. Interestingly, all the deletions occurred to the same 22 homolog; we cannot rule out the possibility that these deletions may have arisen from progressive deletions of an unstable chr22q and therefore are related. (C) Chromothripsis of chr6 that is linked to a terminal deletion on 14q, both of which inferred to be ancestral (see Figure 4). Black arcs represent intra-chromosomal junctions, magenta vertical lines represent chr6 and chr14 breakpoints joined by inter-chromosomal junctions. The clustering of breakpoints on chr6 is very similar to the clustering of breakpoints seen in RCC16 and in our prior studies of breakage-fusion-bridge cycles88. This pattern contrasts with the simple outcome of reciprocal translocations associated with TFE3 fusions between chr1 and chrX. (D) Large segmental deletions on 17p (left) and 18p (right) inferred to be present in the founding tumor clone. The same breakpoints on 17p and 18p are seen in Liver-2 (and other Liver metastases except Liver-4), Lung metastases, and RPLN-1, but there is copy-number gain of the translocated chromosome der(17)t(17;18) in multiple metastases that may be due to positive selection; the Liver-4/MedLN lesions show a larger deletion on 17p, indicating downstream evolution of the translocated chromosome. Notably, there is sloping copy-number variation near the 17p terminus (black) in RPLN-1 that is absent in the intact homolog (gray); this pattern suggests ongoing evolution of the translocated chromosome that is consistent with varying copy-number states and breakpoints seen in the different metastatic samples90. (E) Independent deletions or losses of der(X)t(1;X) in the lung metastases (using Lung-1 as representative), MedLN/Liver-4 (using MedLN as representative), and RPLN-2. In the Lung samples, the deletion spans the entire Xq segment (centromeric to the translocation breakpoint), but a smaller 1p segment; the reciprocal fusion junction is deleted by small focal deletions near the translocation breakpoint. In the MedLN/Liver-4 samples, the loss spanned the entire der(X)t(1;X) but occurred after whole-genome duplication; therefore, this must be independent from the deletion in Lung metastases. Finally, the deletion in RPLN-2 indicates chromothripsis of der(X)t(1;X), with partial retention on chrXq and complex rearrangements between chr1p and chrX; this event is independent from MedLN/Liver-4, since it is inferred to be clonal or have occurred before WGD; it is also independent from the deletion in Lung metastases as they do not share any breakpoint.
Figure S7. Evidence for chrXp reactivation in tRCC samples. Related to Figure 6. (A) Cumulative fractions of biallelically expressed heterozygous sites (top), minor allelic fractions of transcripts from only non-escaping genes (middle), and segmented chrX copy number of samples shown in Figure 6G (tRCCs with chrXp reactivation: TRCC18, TCGA-G7–7501, TRCC17; tRCCs without chrXp reactivation: TCGA-B0–5705; ccRCC: TCGA-T7-A92I; tRCC cancer-adjacent-normal: TCGA-B0–5705-N). The cumulative distribution of biallelic expression (red) is shown together with the cumulative distribution of all transcribed genes with heterozygous polymorphisms (black). P-values are calculated by Mann-Whitney U tests comparing distributions of minor allele fraction for heterozygous sites to the left and right of the TFE3 breakpoint. (B) Normalized expression level of biallelically expressed genes (RNA MAF ≥ 0.2) in samples inferred to have chrXp reactivation and in samples without chrXp reactivation. Gene expression is normalized by the mean expression in the non-reactivated cohort. Left: TCGA, Right: current study. P-values calculated by Mann-Whitney U-test. (C) Methylation (quantified by beta value) at probes in the promoters of all non-escaping genes to the left (red, N=109 probes) or right (brown, N=316 probes) of TFE3 in each sample. Samples are grouped by chrXp reactivation status. P-values are computed by Mann-Whitney U-test. (D) Methylation at the promoters (±1kb from exon 1 start) of genes showing biallelic expression (RNA MAF ≥ 0.2) to the left of TFE3 in samples with evidence of chrXp reactivation (N=5) compared to genes showing monoallelic expression (RNA MAF < 0.2) across the entire chrX for all tRCCs (brown). P-values are computed by Mann-Whitney U test.
Figure S8. Additional examples of long-range transcriptional changes associated with XCI perturbation. Related to Figure 6. (A) An example of autosomal silencing in a female tRCC. Summary of allelic RNA fraction for chr1 at expressed germline heterozygous sites (SNVs), plotted similarly to Figure S7A and Figure 6G. Segmented DNA copy number is also shown across the chromosome. P-values calculated by Mann-Whitney U test between distributions of minor allele fraction for heterozygous sites to the left and right of the autosomal breakpoint. (B) Western blot of wildtype and TFE3 fusion proteins in a panel of cell lines with TFE3 fusions (tRCC: UOK146 [PRCC-TFE3], S-TFE [ASPSCR1-TFE3]; alveolar soft part sarcoma: ASPS-1 [ASPSCR1-TFE3], ASPS-KY [ASPSCR1-TFE3]) and TFE3 fusion-negative ccRCC cell line 786O. A single band in the tRCC lines is suggestive of only one TFE3 protein species (i.e., chrXa fusion); a double band suggests two different TFE3 proteins, one from the fusion transcript generated by chrXi translocation and the other corresponding to WT TFE3 from chrXa. (C) TFE3 western blot with doxycycline-inducible shRNA-mediated knockdown of TFE3 in ASPS-KY and S-TFE cells with target sequence in the TFE3 3’UTR, demonstrating that both bands represent TFE3 species. (D) Allelic chrX expression assessed at common polymorphic sites in S-TFE (left) and UOK146 (right) with segmented DNA copy number shown at the bottom. In the 23Mb region (chrX:26,139,329–49,028,726) where both cell lines show heterozygosity, we observe biallelic expression in sTFE but not UOK146, indicating a possible chrXi fusion in s-TFE that is accompanied by partial Xp reactivation. (E) Two examples of biallelic chrXp expression associated with chrX:autosome fusions, suggesting partial Xp reactivation (TCGA-24–1424 and TCGA-N8-A4PN), and two representative samples without chrX-reactivation (TCGA-X6-A7WB and TCGA-FU-A3TX). Data are shown as in Figure S7A. Biallelic expression was assessed by the following criteria: RNA minor allele fraction ≥ 0.2, threshold ≥ 10 DNA reads and ≥ 10 RNA reads. (F) Two examples of biallelic chrX expression associated with deletion of the X-inactivation center in cell lines from the Cancer Cell Line Encyclopedia (HCC-366 and LMSU). Deleted regions on chrX are shaded in blue. A representative example of monoallelic chrX expression is shown as reference (CL-40)..
Table S1. Sample information, clinical annotations, and sequencing metrics for tRCC WGS cohort. Related to Figure 1. The following information is listed for each sample: (A) Subject (B) Sample ID; (C) biological sex (male/female); (D) Race/ethnicity (E) sample type (FFPE or frozen); (F) TFE3 fusion partner; (G) sequencing library type (standard or linked-read WGS); (H) tumor site (primary or metastasis) (I) tumor tissue acquisition method (resection or biopsy); (J) source of normal (tumor-adjacent normal or blood); (K) treatment status at the time of tumor sample acquisition (treatment-naïve or post-treatment); (J) prior therapy (if post-treatment); In one pre-treated case (TRCC12), tumor tissue was sampled from the same metastatic site both prior to, and following progression on, systemic therapy with a tyrosine kinase inhibitor (lenvatinib) and mammalian target of rapamycin (mTOR) inhibitor (everolimus); (L) mean sequencing depth of tumor sample; (M) mean sequencing depth of the matching normal; (N,Q) tumor purity; (O,P) tumor ploidy.
Table S2. Summary of key genomic features of tRCCs, including breakpoints of TFE3 and fusion partners. Related to Figures 2 and S1.
Table S3. Somatic mutations in cancer driver genes detected in the tRCC WGS cohort. Related to Figure 1 and Figure S1.
Table S4. Primers, sgRNAs, and shRNA sequences used in this study. Related to Figure 2, Figure S3, and Figure S8.
Table S5. Catalog of biological sex, TFE3 fusions, and copy-number imbalance status of 203 tRCC cases (tab 1), and summary of allele-specific expression analysis (tab 2 and tab 3). Related to Figure 3, Figure 6, Figure S4, S7, S8. Tab 1 lists 203 tRCC samples from 11 datasets (this study; TCGA; Sun et al.; Qu et al, Durinck et al., Malouf et al., MSK Panel, MSK WES, OncoPanel, PCAWG, and Sato et al.). Tab 2 lists the mean minor allele fractions of transcripts of genes to the left or to the right of the TFE3 breakpoint, and the P-value computed by Mann-Whitney tests. Tab 3 summarizes the inference of chrXi or chrXa translocations from the RNA-Seq data of frozen tumor samples by Xp allelic expression and the following features of TFE3 or fusion partner transcripts. (1) allelic expression of SNVs within the 3’ end of TFE3 (expected to be monoallelic for chrXa fusions; biallelic for chrXi fusions, due to co-expression of wild type TFE3 from chrXa and fusion TFE3 from chrXi), (2) allelic expression of SNVs in the 3’ end of the autosomal fusion partner (expected to be biallelic for chrXa fusions; monoallelic for chrXi fusions due to silencing of the 3’ end of the autosomal partner); (3) RNA-seq reads spanning the reciprocal (TFE3-Partner) fusion junction (expected to be detectable only for chrXa fusions). These features were further integrated with the copy-number status of the reciprocal translocated chromosome (loss only tolerated for chrXi fusions) or chromosome-wide chrX loss-of-heterozygosity (only tolerated for chrXa fusions). Altogether, we were able to assign translocated haplotype (chrXi/chrXa) using one or more of these criteria in 14/15 samples. In 11 of these 14 samples, the orthogonal inferences using different features were concordant. In the remaining 3 samples (TRCC8, TRCC10, TCGA-J7–8537), we observed biallelic chrXp expression (suggestive of a chrXi fusion), but also identified junction-spanning RNA-seq reads suggesting expression of the reciprocal fusion (Partner-TFE3). TRCC8 displayed chrXq deletion and lacked XIST expression, which could indicate XIC erosion that explains the expression of the reciprocal fusion. Reciprocal fusion expression in the remaining two cases may be caused by somatic erosion of XCI91 or the incomplete silencing (partial escape) of the fusion gene92.
Table S6. Somatic mutations in TRCC18 and their co-occurrence patterns used for phylogenetic inference. Related to Figure 4 and Figures S5–6. Tab1 lists the mutant allele fractions (MAFs) of all somatic single-nucleotide substitutions in the metastatic lesions. Tab2 lists the clonality status of each mutation in each sample, and the pattern (pattern_id). Tab3 lists the the assignment of mutations with each clonality pattern (pattern_id) to different evolutionary branches, with explanations of the rationale for re-assignment during manual review (Comments).
Data Availability Statement
De-identified whole genome sequencing data reported in this study have been deposited in the NCBI Database of Genotypes and Phenotypes (dbGaP) under the study accession phs003008.v1.p1 and are publicly available with controlled access in accordance with an IRB-reviewed protocol. This study also analyzes publicly available existing data and the links or accessions to these datasets are listed in the key resources table. Whole exome data of tRCC cell lines have been deposited as a linked study at Sequence Read Archive (SRA) under the accession PRJNA865001.
Algorithms and code implemented in this study are uploaded at https://github.com/SViswanathanLab/WGS-TRCC.
Any additional information required to reanalyze the data is available upon reasonable request from the lead contacts.