Genomic assessment of renal cancer
Ronald M. Bukowski , Robert A. Figlin, Robert J. Motzer (Eds). Renal cell carcinoma. Molecular targets and clinical applications. 3 Ed. Springer Science+Business Media New York (2015)
Perhaps as much or more than any other category of disease, cancer is fundamentally a genomic disorder. With the advent of a variety of high-throughput techniques for investigating the cancer genome, most notably massively parallel sequencing (MPS) or next-generation sequencing (NGS), the genomics era in cancer medicine would appear finally to have arrived at the threshold of enormous, real possibility. One can trace the arc of clinical observation and the discovery of disease from the identification of families with multiple cases of renal cancer along with the abnormalities at the chromosomal level within the renal cancers themselves to the discovery of pivotal genetic lesions. The exemplar of such success started with the finding that retinal neoplasia occurred in families reported by investigators at the turn of the nineteenth century, which were later found to also have multiple cases of renal cancer [1, 2]. Eventually, the families with von Hippel-Lindau (VHL) disease led to the discovery of 3p loss and, ultimately, to the characterization of the VHL gene, its function, and its role in the pathophysiology of clear cell renal cell carcinoma (ccRCC) [3–5].
Renal cell carcinoma represents a collection of distinct diseases that can be distinguished histologically and genetically. Discovery-oriented work has focused primarily on clear cell renal cell carcinoma, the most common subtype of RCC, characterized by inactivating mutations in VHL which led to the initiation of the hypoxia response elements through HIF-mediated transcription resulting in alteration to the intracellular metabolic program. Recently, a number of discoveries also have implicated epigenetic changes as central to the molecular pathophysiology of disease, with mutations in genes responsible for chromatin remodeling and histone methylation. While distinct genetically, less is known regarding papillary renal cell carcinoma subtypes and chromophobe renal cell carcinoma. In this chapter, we will focus primarily on describing that which is now known regarding the ccRCC tumor genome. After a section describing the methods of genomic analysis, consideration of the RCC tumor genome will be organized into sections pertaining to cytogenetic changes, somatic mutations, epigenetic alterations, and the RCC tumor transcriptome.
Methods for genomic characterization
Methods for characterizing the genomics of renal cell carcinoma have focused predominantly on the tumor genome of clear cell renal cell carcinoma (ccRCC). Techniques have evolved, rapidly so, over the last 10 years, to allow for analysis of cytogenetics and copy number changes, DNA sequence and somatic mutations, epigenetic alterations, and expression profiling. Technologic advances have made it possible to perform these techniques readily in a massively parallel fashion and at comparatively low cost. Moreover, it has become possible in the context of a single unified project to use multiple analytical techniques together, called integrative genomic analysis, to expand the possibilities for discovery immeasurably. In this section, we will briefly review select methods available in each analytical category.
Cytogenetics and copy number change
The term “genomics” in the context of human cancer research refers to an evaluation of the entire set of genetic information contained across all 23 chromosome pairs of a human cell, including genes, gene-modifying regions, and all other areas in between. Cytogenetics, the oldest genomic method of analysis, emerged early in the twentieth century as a way of detecting changes at the chromosome level of detail using the karyotype. The earliest genomic discovery in RCC pertained to loss of the short arm of chromosome 3 [3, 4]. Clinical cytogenetics analyses still are performed and reported routinely on RCC nephrectomy specimens.
More commonly used in integrative studies today, however, are array-based comparative genomic hybridization (aCGH) or SNP (single-nucleotide polymorphism) arrays, which are high-throughput techniques for determining the relative copy number of thousands or more genes or specific loci. Two sources of DNA—typically tumor and normal—are isolated, denatured, and labeled with respective ?uorophores via nick translation so that their relative frequencies can be compared based on competitive hybridization to known primer sequences. aCGH techniques represent an important way to determine copy number information at the chromosome level, an important first place to look for large-scale and big picture changes in the tumor genome at hand, and this data often cannot be determined using conventional sequencing technologies such as Sanger. The level of resolution of these techniques, however, represents a limitation; although it may be possible to determine that there has been a copy number gain in a chromosomal region, it is not always possible to tell exactly which genes are affected.
DNA sequence and somatic mutations
Massively parallel sequencing (MPS) or next-generation sequencing (NGS) emerged as a commercially available analytic technique in the mid-2000s . Throughput has been optimized to the point that up to 100 million reads of short segments, typically ranging from 50 to up to several hundred bases, can be performed in hours to days at relatively affordable costs. Sensitivity also has been dramatically improved because DNA transcripts can be read many times, with greater than 1,000-fold per given locus, depending on the application, in a given sequencing run which enables the detection of variants that might be present at lower prevalence in a particular sample because of suboptimal tumor-normal admixture or because of tumor genetic heterogeneity. Although the platforms for performing MPS vary and continue to evolve, many share common core processes such as (1) template preparation and library construction, (2) sequencing reactions, and (3) paired-end analysis .
MPS represents an important advance for the aforementioned reasons—higher throughput, greater sensitivity, and aligned with continually decreasing costs. However, perhaps its greatest strength lies in its ability to integrate the detection of multiple categories of genomic aberrancy into one methodology. With MPS, it is possible not only to detect mutations or variants but also to detect structural changes, copy number changes, and small insertions or deletions throughout the genome. Although some of these variations require differing and more sophisticated analysis, they are all possible. Limitations of this technology include the fact that massively parallel sequencing generates massive data output which in turn poses real challenges in terms of data storage, manipulation, analysis, and interpretation. Along with this, there can be important implications to consider when potentially thousands of genomic events in a given specimen can be compared with, e.g., clinical outcome. And the discovery of many “new” genomic aberrancies poses a challenge to clinicians in bringing MPS data into the clinic—is the variation in question really important? What does it mean and what should be done about it?
A number of recent discoveries in the domain of RCC genomics, described in the sections that follow, pertain to alterations in epigenetic control, and many of these have come from MPS work. Thus, MPS itself represents an important method for elucidating the degree to which epigenetic alteration may contribute to the disease. Numerous methodologies have emerged for detecting DNA methylation, for examining gene silencing, and for identifying other regulatory marks on DNA. Bisulfite sequencing is one strategy, in which DNA is treated with Bisulfite (which converts cytosine residues to uracil unless the cytosine has been methylated). DNA can then be sequenced to allow for a determination of methylation status. This method has been used to determine the rate of VHL promoter methylation in ccRCC, an important cause of inactivation in cancers, typically about 8 % . Other methods will be described below as they are introduced in the data overview.
Several key papers have used transcript analysis to show that ccRCC can be classified by RNA expression; for example, HIF expression (HIF1α and HIF2α expressing versus HIF2α expressing ccRCCs) influences different transcriptional programs with implications for disease phenotype [9, 10]. Conventional approaches for expression profiling feature high-throughput techniques wherein tens of thousands of probes for specific genes of interest, typically mounted on a glass or on a silicon chip, are allowed to interact with purified mRNA from fresh frozen tissue. Transcript hybridization to probes can be detected and quantified. Data often is analyzed using clustering analysis as a way of identifying, e.g., active signal transduction pathways or the activity of key genes of interest such as Myc. MPS techniques also can be used to analyze RNA expression, so-called RNA sequencing, which uses similar end-labeling of cDNA as DNA-based massively parallel sequencing, and sequencing of single or paired-end transcripts using the same read lengths. Strengths of this line of analysis include the abilities to detect altered transcript expression levels, as well as altered allele-specific expression, and differential alternative splicing. With sufficient read depth, sequence alterations can also be detected in expressed genes.
The technologies for explorations in the human genome, epigenome, and transcript space have become greatly expanded and will continue to evolve at a rapid pace for the foreseeable future. The advantages are increased capability for discovery of events or changes, improvements in speed and cost, and provision of opportunity for integrated analysis. However, the component data have become bioinformatically enormous and complex to analyze and interpret. It is essential for the consumer of this information to have at least a fundamental knowledge of the data platforms, to recognize limitations and opportunities for errors to exist in the data, as well as to interpret and recognize interesting or novel findings. The role of bioinformatics expertise in this space is increasingly important, and these individuals play a critical role in interfacing between cancer biologists, physician scientists, and emerging genomic data.
Cytogenetics and copy number changes in renal cancer
As with all cancers, the initial studies falling into the genetics and genomics realm focused on examining the karyotype of renal cancers using standard cytogenetic techniques. It has been recognized for some time that ccRCC has chromosomal aberrations at significantly fewer sites when compared to other tumor types;  however, those that are present are very commonly observed. Copy number alterations in ccRCC also are more likely to involve full chromosome arms than to target them . Hyperploidy (ploidy >2.5) of ccRCC has been associated with a higher rate of metastases and poor prognosis . Multiple types of renal cancer were used together as an important initial model to show that copy number aberrations as defined by array-based comparative genomic hybridization (aCGH) could differentiate among cancer types. Using only 40 renal cancers in total, Waldman and colleagues were able to differentiate between ccRCC, papillary renal cancer, chromophobe renal cancer, and oncocytoma, demonstrating how different these subtypes of renal cancer are in copy number profile . This study not only was able to easily discriminate among types of renal cancer with small numbers but underscored the usefulness of aCGH.
Copy number changes in ccRCC
Multiple studies of ccRCC have reported regions of gains on chromosomes 1q, 5q, 7, 8q24, 11q, 12q, and 20q and regions of losses on chromosomes 1p, 3p, 4q, 6q, 8p, 9p, 9q, and 14q [14–17]. Chromosome 3p losses (60–90 %) and 5q gains (33–67 %) are the most prevalent genetic abnormalities in sporadic ccRCC tumors [11, 14, 16, 18, 19]. The four most commonly mutated genes in ccRCC—VHL (von HippelLindau), PBRM1 (polybromo 1), SETD2 (SET domain containing 2) and BAP1 (BRCA-associated protein 1)—are located on 3p [20–23]. Biallelic loss of VHL has long been known to occur in the vast majority of ccRCC, with loss of heterozygosity accompanied by either VHL mutation in most cases or methylation . Although the gain of 5q has been commonly observed for many years, the driver gene(s) has not been elucidated. In the tumors included for the Cancer Genome Atlas (TCGA) effort, focal amplifications were found which narrowed the region to 5q35, encompassing 60 genes . Within the region are several genes of interest, one of which also was investigated by Dondeti et al.  STC2 (stanniocalcin 2), a secreted glycoprotein, is upregulated under hypoxia and is thought to help cells adapt to the stress of the tumor microenvironment. Using siRNA experiments, the authors were able to show that STC2 promotes tumor growth by inhibiting cell death in ccRCC cells . This gene also has been found to be hypomethylated in ccRCC, supporting a tumorigenic role . The loss of 14q, containing HIF1Α, also is commonly observed (30–50 %)  and associated with prognosis, as discussed below.
Additional common amplifications and deletions are described in ccRCC. Targeted copy number changes include CDKN2A/B (9p) and the Myc oncogene (8q), which are deleted and amplified respectively;  however, Myc amplification appears to be more important in renal cancer cell lines than in tumors. The region that includes TP53 also has been observed to be recurrently deleted and that encompassing EPAS1 (HIF2α) is amplified . Additional common regions of amplification and deletion observed in the TCGA, by descending frequency, included deletions of 6q26 (QKI, ARID1B), 8p11, 10q23 (PTEN), 1p36 (ARID1A), and 4q35; amplification of 3p26 (MECOM; MDC1 and EVI1 complex locus); and deletions of 13q21 (RB1), 15q21, and 2q37 (CUL3) [11, 12]. The TCGA also identified several additional regions of focal amplification and deletion targeting specific genes including amplifications of MDM4 (1q32), PRKCI (name), and JAK2 (9p24) and deletions of NEGRI (1p31), CADM2 (3p12), PTPRD (9p23), and NRXN3 (14q24). Many of these findings explain previously described deletions and amplifications identified through less precise methodologies such as karyotyping and aCGH.
Prognostic associations with copy number changes in ccRCC
Several studies have examined whether karyotypic and copy number changes in ccRCC are associated with prognostic differences in ccRCC. Standard karyotyping has been done in 282 ccRCCs in patients with nephrectomies to examine whether cytogenetic changes were prognostic; this study remains influential in the field . The deletion of 3p was associated with a better prognosis (p = 0.03), whereas deletions of 4p (p < 0.001), 9p (p < 0.01), and 14q (p < 0.01) were associated with a decrease in disease-specific survival. In multivariate analysis, loss of 9p remained, along with stage and grade as independently associated with survival. CDKN2A is located on 9p, loss of which is associated with poor prognosis in other tumor types as well, such as melanoma . Additionally, 1p, 9q, and 13q loss and 12q gain have been associated with stage and grade . Some of these findings may come down to single gene changes, which have now been better delineated by the TCGA effort, such as RB1 on 13q.
Of particular interest in regard to worsened prognosis is the deletion of chromosome 14q, which contains the HIF1Α locus [18, 26]. When tumors expressing HIF1α and HIF2α (H1H2) are compared to those expressing HIF2α (H2) alone, losses in 9p and 14q are more significant in the H2 group compared to the H1H2 group . In patient samples, frequent targeted deletion of HIF1Α has been observed, particularly in those renal cancers associated with more aggressive disease .
Taken together, these findings by multiple investigators provide evidence that the loss of HIF1Α is a poor prognostic marker in ccRCC and support multiple other
avenues of evidence suggesting that HIF2α is the major HIF driver in this cancer type. Of note, copy number analyses of ccRCCs, sporadic and associated with VHL disease in several studies, have been compared and show generally a similar profile between groups, although the sporadic tumors are more heterogeneous and consistently demonstrate more copy number aberrations per tumor, significantly so in one study [14, 17, 24].
Familial renal cancer due to chromosome 3 translocations
Multiple families with inherited susceptibility due to balanced translocations involving chromosome 3 have been described [28–33]. The mechanism behind the increased risk of multifocal clear cell renal cancer is thought to be the loss of the rearranged chromosome during mitosis, which requires a quadrivalent (four chromosomes coming together), leading to greater errors during chromosomal segregation. As multiple genes involved in the pathogenesis of clear cell renal cancer are located on chromosome 3p, including VHL, PBRM1, BAP1, and SETD2 , it is not surprising that a mechanism of increased loss of one allele leads to an increased risk of clear cell renal cancer.
Somatic genetics of renal cancer
With the advent of massively parallel sequencing, as with other cancer types, the somatic genetic and genomic profiles of renal cancers have become increasing well detailed. Multiple studies focusing on ccRCC using whole exome sequencing (usually in fact covering 85–90 % of the genome) have been published [21, 23], and more comprehensive studies also including copy number analysis, methylation, RNA sequencing, and some whole-genome sequencing have been done [11, 12]. These studies have greatly contributed to our understanding of ccRCC, which had been poorly characterized compared to various other cancer types. The mutational profile of renal cancer is characterized by an enrichment of T > C/A > G transitions, followed by C > T/G > A transitions . In the ccRCC TCGA effort, 1.1 ± 0.5 nonsilent mutations per megabase were identified .
Mutated genes in ccRCC
In the TCGA dataset of 417 patients, 19 genes were identified as significantly mutated (q < 0.05) . Among those, 8 emerged at the highest level (q < 0.00001), whereas the remaining 11 remained significant but several orders of magnitude less so (q < 0.01–0.05). The eight genes included VHL, PBRM1, SETD2, KDM5C (lysine (K)-specific demethylase 5C), PTEN, BAP1, MTOR, and TP53. Mutations in the histone-modifying genes SETD2, KDM5C, and KDM6A (lysine (K)-specific demethylase 6A) and the tumor suppressor NF2 (neuro?bromin 2) had been previously emerged as important in ccRCC in a whole exome-sequencing study from Futreal and colleagues . Varela et al. had identified truncating mutations in PBRM1 (polybromo 1), a SWI/SNF complex member, also using massively parallel sequencing . Of these targets identified as mutated by massively parallel sequencing, only PBRM1 is involved in a large proportion (30–40 %) of ccRCC tumors. Most recently somatic mutations in BAP1 (BRCA-associated protein 1) also were identified through whole exome-sequencing studies . Interestingly, in the whole exome sequencing of clear cell renal cancer, which required both tumor and germline samples, mutations of BAP1 were found to originate from the germline in a few patients. Thus, two recent studies have suggested that BAP1 mutations predispose to familial clear cell renal cancer, along with uveal and cutaneous melanoma, and mesothelioma, the known tumor types associated with germline BAP1 mutations [34, 35]. Infrequently, as compared to the other genes, mutations in the known tumor suppressor genes, TET2, KEAP1, NRF2, CUL3, and TP53, also have been identified [11, 12].
Recurrent mutations in TCEB1, which encodes elongin C, part of the pVHL complex that ubiquinates the HIFs , have been recently identified . Although these mutations are relatively infrequent (3 % of cases), they are found only in VHL mutation negative ccRCC and accompanied by loss of the wild-type allele at 8q21. The missense mutations are found at Tyr79 and Alal100, more frequently at the former. The identified mutations are within the binding domain for pVHL and are predicted to abolish the interaction between elongin C and pVHL, resulting in accumulation of the HIF proteins, similar to VHL inactivation. Tumors containing TCEB1 mutations demonstrate increased HIF1α staining by immunohistochemistry.
Activation of the PI3K/AKT signaling pathway in ccRCC
Mutations and copy number changes affecting multiple genes within the PI3K/AKT signaling pathway are found in ccRCC, totaling ~30 % of cases, and are generally mutually exclusive with each other. Activating mutations in MTOR have been identified in 6 % of ccRCC, with recurring mutations at Phe1888 within the FAT domain [11, 12]. Additionally, rare activating mutations are found in PIK3CA and AKT1/2/3 and inactivating mutations in TSC1, TSC2, and PTEN, with the latter being more frequent with homozygous deletions observed as well. Amplifications of FGFR4, GNB2L1 (RACK1), and SQSTM1 (p62) also have been observed which are associated with activation of PI3K signaling [37, 38]. These data provide insight into the clinical activity of MTOR inhibitors (temsirolimus, everolimus) in ccRCC, perhaps opening the door for molecular stratification of patients and optimization of selection for therapy based on tumor genetic profile.
ccRCC tumorigenesis and prognosis in relation to genetic mutation
Mutations in the non-VHL 3p target genes, PBRM1, BAP1, and SETD2, all occur within the background of VHL mutations [11, 12]. In the 421 and 188 ccRCCs from the TCGA and MSKCC, respectively, mutations were present in PBRM1 33.5 % and 30.3 % of the time, SETD2 11.6 % and 7.4 %, and BAP1 9.7 % and 6.4 % . The mutation profiles can include all combinations of the genes on 3p, but they tend to be negatively correlated with each other. Recent studies suggest that SETD2 and BAP1 mutations may be acquired during progression, whereas PBRM1 mutations may be early or initiating mutations . Independent studies from UTSW and MSKCC, both of which were validated using the TCGA dataset demonstrated that BAP1 mutations were associated with a higher tumor grade and decreased overall survival, as compared to those with PBRM1 mutations (hazard ratio 2.8 (95 % CI 1.4–5.9 in the TCGA dataset), which are negatively correlated [11, 39, 40]. The few patients whose ccRCCs had mutations in both BAP1 and PBRM1 had the worst survival . SETD2 mutations also have been associated with decreased overall survival [12, 39].
Epigenetic regulation in ccRCC
Renal cell carcinoma has recently emerged as a paradigm shaping cancer owing to several recent discoveries linking epigenetic regulation with clear cell (conventional) renal cell carcinoma. These discoveries build on an existing body of evidence documenting that gene regulation in RCC occurs commonly via altered DNA methylation. The specific genes are described in detail elsewhere in this text and recently in a review of renal cell carcinoma . In general, as overviewed in other portions of this chapter, renal cell carcinomas display a low mutation frequency and relatively consistent copy number alterations . These findings are in stark contrast to tumors driven by defects in DNA repair or other hypermutable scenarios. As a result, it should be not surprising that epigenetic regulation should emerge as a major mechanism promoting tumorigenesis.
The landscape of epigenetic regulation in renal cell carcinomas is only now beginning to be understood, and much of this fascinating tumor biology remains to be discovered . A flood of new data is likely to appear in the next several years which will elucidate the mechanisms by which these changes promote cancer growth, along with perhaps greater insights as to why renal cell carcinoma, at this point in time, predominantly the clear cell subtype, favors this strategy of tumorigenesis. In this chapter, we will summarize “what” we currently know about epigenetic features and profiles in kidney cancer. The “how” and “why” will remain to be discovered.
Mutations in epigenetic regulatory genes
Key to this discussion is the recent observation that mutations in quintessential chromatin-modifying genes are among the most frequently altered genes in clear cell renal cell carcinoma [22, 23]. These mutation frequencies, discussed elsewhere in this chapter, suggest a potentially strong causal association with cancer progression. Several genes fall into this category, and their frequency of association with clear cell renal cell carcinoma is second only to mutations in VHL. This group of genes were initially identified by a series of deep sequencing studies and have been independently verified by the Cancer Genome Atlas data [11, 21–23]. Collectively these genes, including PBRM1, BAP1, and a set of histone-modifying genes, are mutated in up to 25 % of tumors. In addition to mutations in these genes, hypermethylation has also been detected specifically to reduce expression . We will examine each group separately, although it is important to note that at the time of this chapter, their specific roles in tumorigenesis have not been well described.
PBRM1 (polybromo-1), also known as BRG1-associated factor 180 (BAF180), is a component of the SWI/SNF-B (PBAF) chromatin-remodeling complex, which contains at least SMARCA4/BRG1, SMARCB1/SNF5/INI1/BAF47, ACTL6A/BAF53A or ACTL6B/BAF53B, SMARCE1/BAF57, SMARCD1/BAF60A, SMARCD2/BAF60B, and actin . The SWI/SNF complex functions as a nucleosome-remodeling complex . In packaged chromatin, nucleosome positioning is key to regulating available DNA sequences for sequence-specific transcription factor, enhancer, or repressor protein binding and for assembling DNA packaging properly for cell function. In simple terms, this complex uses ATP hydrolysis to unwind and rewind DNA around assembled nucleosomes . Although discovered with much excitement and fanfare in the mid-1990s, this essential process of nucleosome repositioning remains relatively poorly understood. In addition, mutations in several members of the SWI/SNF complex have been associated with various cancers . For reasons that remain unclear, within the complex, only PBRM1 inactivating mutations, with accompanying loss of heterozygosity, are associated with renal cell carcinoma. Although a complex coordinating DNA packaging might intuitively be associated with cancer suppression, the exact mechanism by which disruption of this complex by PBRM1 or other mutations promotes cancer remains poorly understood.
BAP1, also known as the BRCA1 associated protein-1, is a deubiquitinating enzyme that is a member of the polycomb group proteins that act as transcriptional repressors. BAP1 is the catalytic subunit of the polycomb repressive deubiquitinase (PR-DUB) complex, which controls gene regulation by titrating the amount of ubiquitinated histone H2A present in nucleosomes at the promoters of key developmental genes . It also serves as an adapter molecule for a variety of transcription factors that associate with chromatin-modifying complexes. The effect of mutations in BAP1 to remodel chromatin or affect chromatin-mediated transcriptional processes in tumors is not known. However, among the mutated genes implicated in a chromatin regulatory function, BAP1 is most closely associated with clinical outcome. As discussed above, BAP1 mutations have been linked to a new familial form of renal cell carcinoma [34, 35] and also are associated with a class of ccRCCs typified by poor outcome and aggressive disease [21, 40, 48].
Histone-modifying genes (HMGs)
Massively parallel sequencing of renal tumors has identified an increased rate of mutation in genes associated with modifying histones. Although individually many of these genes are mutated in a minority of ccRCC tumors (<5 %), collectively, mutations in this set of genes may contribute to nearly 30 % of tumors. The impact of mutations in genes that modify histones has potential to dramatically alter cellular dynamics, as the histone modifications of methylation, acetylation, and other alterations program the chromatin for efficient and proper “reading” by interacting proteins of the “histone code.”  This code provides an important sequence agnostic level of the regulation of genes for effective transcription control. The most commonly mutated gene in this set is SETD2 [22, 50]. This factor is well known to have a nonredundant role as a histone methyltransferase. SETD2 trimethylates histone 3 on lysine 36, placing a repressive mark on actively transcribed genes. The loss of SETD2 causes histones to lose this mark. The predicted effect of losing this activity would be to permit RNA polymerase II reentry on already transcribed genes or to miss exon and splicing cues. Human tumors were recently analyzed, demonstrating accumulated alternatively spliced transcripts, intron retention, and alternatively used transcriptional start sites and termination cues. In addition, a massive increase in accessible, non-nucleosome-bound DNA is observed, suggesting a global chromatin reprogramming effect. The net effect in human tumors has not yet been established, although mutations in SETD2 also are associated with poor outcome . Other genes mutated in this group include JARID1C (KDM5C, an H3K4 demethylase) and UTX (KMD6A, an H3K27 demethylase). Mechanistically, the link to advancing the tumor phenotype of these mutations remains to be discovered, but ultimately the high frequency of these events is provocative to consider them as a whole as a key step in the evolution of clear cell renal cell carcinoma.
Genomic assessments of chromatin
Ongoing studies to examine gene level changes in histone marks and the resultant alteration in the histone code will be essential to these advances. The future of epigenetic assessment of chromatin in cancer will require increasingly bioinformatically intensive processes to compile short read maps of the genome, essentially “decorating” the genome with regional information. Several tools are being actively applied to characterize the genomic in this way: Chromatin immunoprecipitation sequencing (ChIP-seq) uses high-specificity antibodies to capture regions of DNA bound by proteins, which are amenable to massive parallel sequencing, and may be a valuable tool going forward for delineating the function of chromatin-interacting proteins, such as BAP1, as well as to map the regions of the genome displaying specific histone marks (such as methylation, acetylation) using epitope-specific antibodies. Widely used to localize transcription factors in the genome, these technologies will create a cancer genome model very unlike from current versions and will hopefully provide insights regarding the derangements of epigenetic marks and programs occurring as a result of these mutations. Genomic studies that have the capacity to map nucleosome placement, such as micrococcal nuclease sequencing (MNase-seq), and regulatory element occupancy, such as formaldehyde-assisted isolation of regulatory elements sequencing (FAIRE-seq), are complementary technologies that capture fragmented regions of the genome to either localize nucleosomes genome-wide or expose open regulatory regions (promoters, enhancers, etc.). These tools have been applied in cell lines but are being developed for use in the complex tumor tissue to examine changes in the epigenome in ways previously impossible.
DNA methylation phenotypes
Perhaps the most well-studied epigenetic mark in cancer biology is CpG island DNA methylation. The relationship between mutations in chromatin epigenetic regulators as those discussed above and DNA methylation remains unknown. However, renal tumors consistently demonstrate differences in DNA methylation compared with normal tissue. In the TCGA analysis, hypermethylation was observed using Bisulfite sequencing at a variety of tumor suppressor loci . Gene mutation-specific differences in DNA methylation, such as changes associated with SETD2 mutation, suggest that DNA methylation change may result directly or indirectly from this mutational event. In particular, loss of DNA methylation was found in non-promoter regions in SETD2-mutated tumors, potentially suggesting a role in maintaining the heterochromatic state . This “reprogramming” of available promoters for gene expression can provide a powerful mechanism to repress or enhance gene expression. High-throughput Bisulfite sequencing has become a standard tool in the armamentarium of cancer genome scientists and can not only complement the marks indicated above by ChIP-seq or other methods but also provide specific information that augments gene expression profile information.
Renal cancer transcript assessments
Transcriptional dysregulation exists at the heart of clear cell type renal cell carcinoma. This is largely owing to the classical association with deregulated hypoxia signaling, covered in detail elsewhere in this text, although many other factors contribute to transcript variance in this cancer. Assessments of mRNA signatures have been examined in a variety of platforms. As indicated above, because it was recognized more than 30 years ago that clear cell renal cell carcinoma was associated with key hypoxia-regulated genes, transcription profiles have been studied from a time when such profiles were only evolving [52–54]. The hypoxia signature consists of massive upregulation of over 100 genes now known to be induced because of activation by the transcription factors HIF1α and HIF2α, each present in a complex with a ubiquitous nuclear transporter HIF1β, also known as ARNT. The induction of these transcripts ranges from subtle increases of twofold or less to genes that are activated more than tenfold in expression. Genes induced transcriptionally as a part of the HIF-driven hypoxia response include genes involved in: angiogenesis, glucose metabolism, cell survival, and cell migration/invasion properties. All tumors that harbor VHL mutation or loss display stabilization of one or both of these HIF factors. Although the consensus binding site for these transcription factors is the same, the factors themselves have overlapping but not identical sets of target genes [31, 55, 56]. Notably, gene expression profiling using array platforms identified that HIF1α specifically targets enzymes involved in glycolysis, which was verified by PCR . Subsequent detailed analysis of human tumors demonstrated that tumors could be classified for expression of HIF1α and HIF2α (H1H2), HIF2α alone (H2), or VHL wild type . The transcript profile analysis of these classifications confirmed this distinction, as well as demonstrating evidence of increased MTOR signaling in H1H2 and wild-type VHL tumors.
In addition to the hypoxia response transcription factors, other features may participate in adapting the transcriptome. Many gene expression mRNA factors have been identified by association with outcomes. The power of gene expression profiling is in the massive numbers of genes that can be simultaneously analyzed for level of expression at a single point in time. By examining genes or gene sets associated with poor outcome tumors, several gene sets have been identified that can aid in the classification of tumors according to the risk for disease progression or death [57–61]. Clustering methods have also emerged which use very high level pattern recognition algorithms to find inherent subgroups within groups of tumors. Clear cell renal cell carcinoma conforms to these pattern recognition algorithms by sorting into two dominant groups, designated as ccA and ccB [10, 62], which are also associated with disease outcome. Using a platform of metadata, which combines the majority of available gene expression data, these two dominant classifications were again observed . In addition, this analysis revealed a small group of variant tumors, not readily classifiable as clear cell tumors, and on histologic assessment these tumors were more readily classified as the rare clear cell papillary subtype , demonstrating the power of molecular assessment to define groups and variants.
Overall, these transcript analyses provide a powerful and ready tool to measure gene expression by a variety of means, from PCRand hybridization-based platforms to massively parallel sequencing, which has brought RNA sequencing into the mainstream. This tool, utilized in the TCGA dataset, provides isotype agnostic expression data as well as opportunities to identify gene fusions or mutations that are not detected by arrays or other means. For example, several recurrent key fusions were identified by the Cancer Genome Atlas . These indicate that some translocations more commonly found in pediatric cancers may be relevant in adult forms of ccRCC as well . Future work will discern the biological relevance of these fusions to promote kidney cancer.
Non-clear Cell Disease
As has been the case for ccRCC, genomic insights into non-clear cell disease have been driven by discoveries related to tumors arising in a familial or inherited context. Germline mutations in the Met proto-oncogene have been shown to be associated with inherited risk for the development of multiple, synchronous papillary RCC type 1 (pRCC-1). Hereditary leiomyomatosis and renal cell carcinoma (HLRCC), characterized by a solitary papillary RCC type 2 (pRCC-2) tumor, appears to arise from germline mutations in the fumarate hydratase gene (FH) [66–69]. It is important to note that HLRCC may only define one subset of pRCC-2. Germline mutations in folliculin (FLCN) have been characterized as the key genetic alterations with Birt-Hogg-Dubé syndrome associated with chromophobe RCC as well as, perhaps, oncocytoma and ccRCC . The FLCN protein has no homology to previously identified proteins, and its function has been controversial. Most recently, it has been suggested that it is a ciliopathy that is involved in cell polarity, regulates cell-cell adhesion, and negatively regulates rRNA synthesis [71, 72]. These genes and their respective genetic changes have been implicated in disrupting core metabolic programming. Understanding the relationship between these changes in tumor cell metabolism and tumorigenesis and progression across RCC subtypes remains a critical area of need for future work.
Little of the somatic genetics of the sporadic (non-inherited) non-clear cell RCC has been elucidated. On a genomic level, non-ccRCCs are known to vary considerably from ccRCC. For example, although chromosome 3p deletion is commonly present in ccRCC and often encompasses VHL, PBRM1, BAP1, and SETD2, chromophobe RCC tumors often accumulate heterozygous losses of multiple whole chromosomes, including chromosomes 1, 2, 6, 10, 13, 17, and 21 [73, 74]. Although chromosome 1 also is often lost in the oncocytoma variant of renal cancer, this pattern of copy number alterations provides a discrete footprint for assigning the diagnosis of chromophobe RCC on the basis of cytogenetic analysis.
Histologically, papillary RCC takes on two forms, type 1 and type 2, but, genetically, it is not clear that the distinction will remain clear cut. Cytogenetically, patterns of gains and losses include common gains of chromosome 17 in pRCC-1 and losses of chromosomes 8, 11, and 18 in pRCC-2 , but the driving mutations in any of these rarer variants of sporadic renal cancer, however, remain largely unknown. The genetics and genomics of these tumor types are being further elucidated, and the inclusion of both papillary and chromophobe-type RCC in TCGA projects will enable dramatic clarification of these diseases at a genomic level.
The methods of genomic assessment have evolved rapidly over the recent past creating remarkable possibility. Beyond the mutations and hypermethylation involving VHL which have been known for some time, the use of these modern methods of genomic analysis in ccRCC has identified a number of important additional changes related to tumor cell genetics, genomics, gene expression, and epigenetic control. Progress in the elucidation of the non-clear cell tumor genomes has lagged behind discoveries in ccRCC; however, new findings are emerging here too. Key challenges that we face now in this work pertain to refining the storyline of how these new findings are interrelated in the pathophysiology of RCC and to determining how best to leverage these findings therapeutically. We have crossed the threshold of enormous possibility in our understanding of these tumors. We hope with some optimism now that these findings and the many others that soon will follow can be translated into real improvements in the care of our patients with renal cancer.
- Collins ET. Intra-ocular growths (two cases, brother and sister, with peculiar vascular new growth, probably retinal, affecting both eyes). Trans Ophthalmol Soc UK. 1894;14:141–9.
- von Hippel E. Ueber eine sehr seltene Erkrankung der Nethaut. Graefe Arch Ophthalmol. 1904;59:83–106.
- Zbar B, Brauch H, Talmadge C, Linehan M. Loss of alleles of loci on the short arm of chromosome 3 in renal cell carcinoma. Nature. 1987;327(6124):721–4.
- Seizinger BR, Rouleau GA, Ozelius LJ, et al. Von Hippel-Lindau disease maps to the region of chromosome 3 associated with renal cell carcinoma. Nature. 1988;332(6161):268–9.
- Iliopoulos O, Kibel A, Gray S, Kaelin Jr WG. Tumour suppression by the human von HippelLindau gene product. Nat Med. 1995;1(8):822–6.
- Margulies M, Egholm M, Altman WE, et al. Genome sequencing in microfabricated highdensity picolitre reactors. Nature. 2005;437(7057):376–80.
- MacConaill LE. Existing and emerging technologies for tumor genomic profiling. J Clin Oncol. 2013;31(15):1815–24.
- Nickerson ML, Jaeger E, Shi Y, et al. Improved identification of von Hippel-Lindau gene alterations in clear cell renal tumors. Clin Cancer Res. 2008;14(15):4726–34.
- Gordan JD, Lal P, Dondeti VR, et al. HIF-αlpha effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clear cell renal carcinoma. Cancer Cell. 2008;14(6):435–46.
- Brannon AR, Reddy A, Seiler M, et al. Molecular stratification of clear cell renal cell carcinoma by consensus clustering reveals distinct subtypes and survival patterns. Genes Cancer. 2010;1(2):152–63.
- Cancer Genome Atlas Research Network. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature. 2013;499(7456):43–9.
- Sato Y, Yoshizato T, Shiraishi Y, et al. Integrated molecular analysis of clear-cell renal cell carcinoma. Nat Genet. 2013;45(8):860–7.
- Wilhelm M, Veltman JA, Olshen AB, et al. Array-based comparative genomic hybridization for the differential diagnosis of renal cell cancer. Cancer Res. 2002;62(4):957–60.
- Beroukhim R, Brunet JP, Di Napoli A, et al. Patterns of gene expression and copy-number alterations in von-Hippel Lindau disease-associated and sporadic clear cell carcinoma of the kidney. Cancer Res. 2009;69(11):4674–81.
- Girgis AH, Iakovlev VV, Beheshti B, et al. Multilevel whole-genome analysis reveals candidate biomarkers in clear cell renal cell carcinoma. Cancer Res. 2012;72(20):5273–84.
- Dondeti VR, Wubbenhorst B, Lal P, et al. Integrative genomic analyses of sporadic clear cell renal cell carcinoma define disease subtypes and potential new therapeutic targets. Cancer Res.2012;72(1):112–21.
- Moore LE, Jaeger E, Nickerson ML, et al. Genomic copy number alterations in clear cell renal carcinoma: associations with case characteristics and mechanisms of VHL gene inactivation. Oncogenesis. 2012;1:e14.
- Klatte T, Rao PN, de Martino M, et al. Cytogenetic profile predicts prognosis of patients with clear cell renal cell carcinoma. J Clin Oncol. 2009;27(5):746–53.
- Monzon FA, Alvarez K, Gatalica Z, et al. Detection of chromosomal aberrations in renal tumors: a comparative study of conventional cytogenetics and virtual karyotyping with singlenucleotide polymorphism microarrays. Arch Pathol Lab Med. 2009;133(12):1917–22.
- Young AC, Craven RA, Cohen D, et al. Analysis of VHL gene alterations and their relationship to clinical parameters in sporadic conventional renal cell carcinoma. Clin Cancer Res. 2009;15(24):7582–92.
- Pena-Llopis S, Vega-Rubin-de-Celis S, Liao A, et al. BAP1 loss defines a new class of renal cell carcinoma. Nat Genet. 2012;44(7):751–9.
- Dalgliesh GL, Furge K, Greenman C, et al. Systematic sequencing of renal carcinoma reveals inactivation of histone modifying genes. Nature. 2010;463(7279):360–3.
- Varela I, Tarpey P, Raine K, et al. Exome sequencing identifies frequent mutation of the SWI/ SNF complex gene PBRM1 in renal carcinoma. Nature. 2011;469(7331):539–42.
- Shuib S, Wei W, Sur H, et al. Copy number profiling in von Hippel-Lindau disease renal cell carcinoma. Genes Chromosomes Cancer. 2011;50(7):479–88.
- Conway C, Beswick S, Elliott F, et al. Deletion at chromosome arm 9p in relation to BRAF/ NRAS mutations and prognostic significance for primary melanoma. Genes Chromosomes Cancer. 2010;49(5):425–38.
- Kroeger N, Choueiri TK, Lee JL, et al. Survival outcome and treatment response of patients with late relapse from renal cell carcinoma in the era of targeted therapy. Eur Urol. 2013;65(6):1086–92.
- Shen C, Beroukhim R, Schumacher SE, et al. Genetic and functional studies implicate HIF1α as a 14q kidney cancer suppressor gene. Cancer Discov. 2011;1(3):222–35.
- Bodmer D, Eleveld M, Ligtenberg M, et al. Cytogenetic and molecular analysis of early stage renal cell carcinomas in a family with a translocation (2;3)(q35;q21). Cancer Genet Cytogenet. 2002;134(1):6–12.
- Bonne AC, Bodmer D, Schoenmakers EF, van Ravenswaaij CM, Hoogerbrugge N, van Kessel AG. Chromosome 3 translocations and familial renal cell cancer. Curr Mol Med. 2004;4(8): 849–54.
- Eleveld MJ, Bodmer D, Merkx G, et al. Molecular analysis of a familial case of renal cell cancer and a t(3;6)(q12;q15). Genes Chromosomes Cancer. 2001;31(1):23–32.
- Foster RE, Abdulrahman M, Morris MR, et al. Characterization of a 3;6 translocation associated with renal cell carcinoma. Genes Chromosomes Cancer. 2007;46(4):311–7.
- Kanayama H, Lui WO, Takahashi M, et al. Association of a novel constitutional translocation t(1q;3q) with familial renal cell carcinoma. J Med Genet. 2001;38(3):165–70.
- Koolen MI, van der Meyden AP, Bodmer D, et al. A familial case of renal cell carcinoma and a t(2;3) chromosome translocation. Kidney Int. 1998;53(2):273–5.
- Farley MN, Schmidt LS, Mester JL, et al. Germline BAP1 mutation predisposes to familial clear-cell renal cell carcinoma. Mol Cancer Res. 2013;11(9):1061–71.
- Popova T, Hebert L, Jacquemin V, et al. Germline BAP1 mutations predispose to renal cell carcinomas. Am J Hum Genet. 2013;92(6):974–80.
- Kibel A, Iliopoulos O, DeCaprio JA, Kaelin Jr WG. Binding of the von Hippel-Lindau tumor suppressor protein to Elongin B and C. Science. 1995;269(5229):1444–6.
- Wang F, Yamauchi M, Muramatsu M, Osawa T, Tsuchida R, Shibuya M. RACK1 regulates VEGF/Flt1-mediated cell migration via activation of a PI3K/Akt pathway. J Biol Chem. 2011;286(11):9097–106.
- Komatsu M, Kageyama S, Ichimura Y. p62/SQSTM1/A170: physiology and pathology. Pharmacol Res. 2012;66(6):457–62.
- Hakimi AA, Ostrovnaya I, Reva B, et al. Adverse outcomes in clear cell renal cell carcinoma with mutations of 3p21 epigenetic regulators BAP1 and SETD2: a report by MSKCC and the KIRC TCGA research network. Clin Cancer Res. 2013;19(12):3259–67.
- Kapur P, Pena-Llopis S, Christie A, et al. Effects on survival of BAP1 and PBRM1 mutations in sporadic clear-cell renal-cell carcinoma: a retrospective analysis with independent validation. Lancet Oncol. 2013;14(2):159–67.
- Jonasch E, Futreal PA, Davis IJ, et al. State of the science: an update on renal cell carcinoma. Mol Cancer Res. 2012;10(7):859–80.
- Maher ER. Genomics and epigenomics of renal cell carcinoma. Semin Cancer Biol. 2013;23(1):10–7.
- Ibragimova I, Maradeo ME, Dulaimi E, Cairns P. Aberrant promoter hypermethylation of PBRM1, BAP1, SETD2, KDM6A and other chromatin-modifying genes is absent or rare in clear cell RCC. Epigenetics. 2013;8(5):486–93.
- Pazin MJ, Bhargava P, Geiduschek EP, Kadonaga JT. Nucleosome mobility and the maintenance of nucleosome positioning. Science. 1997;276(5313):809–12.
- Wilson BG, Roberts CW. SWI/SNF nucleosome remodellers and cancer. Nat Rev Cancer. 2011;11(7):481–92.
- Whitehouse I, Flaus A, Cairns BR, White MF, Workman JL, Owen-Hughes T. Nucleosome mobilization catalysed by the yeast SWI/SNF complex. Nature. 1999;400(6746):784–7.
- Scheuermann JC, de Ayala Alonso AG, Oktaba K, et al. Histone H2A deubiquitinase activity of the Polycomb repressive complex PR-DUB. Nature. 2010;465(7295):243–7.
- Choueiri TK, Pomerantz MM, Signoretti S. Renal-cell carcinoma: a step closer to a new classification. Lancet Oncol. 2013;14(2):105–7.
- Strahl BD, Allis CD. The language of covalent histone modifications. Nature. 2000;403(6765):41–5.
- Duns G, van den Berg E, van Duivenbode I, et al. Histone methyltransferase gene SETD2 is a novel tumor suppressor gene in clear cell renal cell carcinoma. Cancer Res. 2010;70(11):4287–91.
- Wagner EJ, Carpenter PB. Understanding the language of Lys36 methylation at histone H3. Nat Rev Mol Cell Biol. 2012;13(2):115–26.
- Caldwell MC, Hough C, Furer S, Linehan WM, Morin PJ, Gorospe M. Serial analysis of gene expression in renal carcinoma cells reveals VHL-dependent sensitivity to TNFalpha cytotoxicity. Oncogene. 2002;21(6):929–36.
- Higgins JP, Shinghal R, Gill H, et al. Gene expression patterns in renal cell carcinoma assessed by complementary DNA microarray. Am J Pathol. 2003;162(3):925–32.
- Jiang Y, Zhang W, Kondo K, et al. Gene expression profiling in a renal cell carcinoma cell line: dissecting VHL and hypoxia-dependent pathways. Mol Cancer Res. 2003;1(6):453–62.
- Hu CJ, Iyer S, Sataur A, Covello KL, Chodosh LA, Simon MC. Differential regulation of the transcriptional activities of hypoxia-inducible factor 1 alpha (HIF-1αlpha) and HIF-2αlpha in stem cells. Mol Cell Biol. 2006;26(9):3514–26.
- Hu CJ, Wang LY, Chodosh LA, Keith B, Simon MC. Differential roles of hypoxia-inducible factor 1alpha (HIF-1αlpha) and HIF-2αlpha in hypoxic gene regulation. Mol Cell Biol. 2003;23(24):9361–74.
- Furge KA, Lucas KA, Takahashi M, et al. Robust classification of renal cell carcinoma based on gene expression data and predicted cytogenetic profiles. Cancer Res. 2004;64(12):4117–21.
- Kosari F, Parker AS, Kube DM, et al. Clear cell renal cell carcinoma: gene expression analyses identify a potential signature for tumor aggressiveness. Clin Cancer Res. 2005;11(14):5128–39.
- Takahashi M, Rhodes DR, Furge KA, et al. Gene expression profiling of clear cell renal cell carcinoma: gene identification and prognostic classification. Proc Natl Acad Sci U S A. 2001;98(17):9754–9.
- Yao M, Huang Y, Shioi K, et al. A three-gene expression signature model to predict clinical outcome of clear cell renal carcinoma. Int J Cancer. 2008;123(5):1126–32.
- Zhao H, Ljungberg B, Grankvist K, Rasmuson T, Tibshirani R, Brooks JD. Gene expression profiling predicts survival in conventional renal cell carcinoma. PLoS Med. 2006;3(1):e13.
- Reddy A, Brannon AR, Seiler M, et al. A predictor for survival in intermediate grade clear cell renal cell carcinoma. J Supercomputing. 2009;WORLDCOMP ‘09 Supplement.
- Brannon AR, Haake SM, Hacker KE, et al. Meta-analysis of clear cell renal cell carcinoma gene expression defines a variant subgroup and identifies gender influences on tumor biology. Eur Urol. 2012;61(2):258–68.
- Rohan SM, Xiao Y, Liang Y, et al. Clear-cell papillary renal cell carcinoma: molecular and immunohistochemical analysis with emphasis on the von Hippel-Lindau gene and hypoxiainducible factor pathway-related proteins. Mod Pathol. 2011;24(9):1207–20.
- Malouf GG, Monzon FA, Couturier J, et al. Genomic heterogeneity of translocation renal cell carcinoma. Clin Cancer Res. 2013;19(17):4673–84.
- Gunawan B, von Heydebreck A, Fritsch T, et al. Cytogenetic and morphologic typing of 58 papillary renal cell carcinomas: evidence for a cytogenetic evolution of type 2 from type 1 tumors. Cancer Res. 2003;63(19):6200–5.
- Gontero P, Ceratti G, Guglielmetti S, et al. Prognostic factors in a prospective series of papillary renal cell carcinoma. BJU Int. 2008;102(6):697–702.
- Okon K, Sinczak-Kuta A, Stachura J. Renal papillary carcinoma classification into subtypes may be reproduced by nuclear morphometry. Anal Quant Cytol Histol. 2009;31(2):109–17.
- Antonelli A, Arrighi N, Tardanico R, et al. Prognostic value of cytogenetic analysis in clear cell renal carcinoma: a study on 131 patients with long-term follow-up. Anticancer Res. 2010;30(11):4705–9.
- Nickerson ML, Warren MB, Toro JR, et al. Mutations in a novel gene lead to kidney tumors, lung wall defects, and benign tumors of the hair follicle in patients with the Birt-Hogg-Dube syndrome. Cancer Cell. 2002;2(2):157–64.
- Luijten MN, Basten SG, Claessens T, et al. Birt-Hogg-Dube syndrome is a novel ciliopathy. Hum Mol Genet. 2013;22(21):4383–97.
- Medvetz DA, Khabibullin D, Hariharan V, et al. Folliculin, the product of the Birt-Hogg-Dube tumor suppressor gene, interacts with the adherens junction protein p0071 to regulate cell-cell adhesion. PLoS One. 012;7(11):e47842.
- Yusenko MV, Kuiper RP, Boethe T, Ljungberg B, van Kessel AG, Kovacs G. High-resolution DNA copy number and gene expression analyses distinguish chromophobe renal cell carcinomas and renal oncocytomas. BMC Cancer. 2009;9:152.
- Monzon FA, Hagenkord JM, Lyons-Weiler MA, et al. Whole genome SNP arrays as a potential diagnostic tool for the detection of characteristic chromosomal aberrations in renal epithelial tumors. Mod Pathol. 2008;21(5):599–608.
- Yu W, Zhang W, Jiang Y, et al. Clinicopathological, genetic, ultrastructural characterizations and prognostic factors of papillary renal cell carcinoma: new diagnostic and prognostic information. Acta Histochem. 2013;115(5):452–9.