Keywords

1 Nuclear Receptor Genomic Interactions Are Highly Integrated and Sense a Wide Variety of Inputs

The collective transcriptional actions of nuclear receptors (NRs) form a central conduit for hormonal, dietary and environmental compounds to signal to the genome. Specifically, NRs act as sensors that respond to both the presence and absence of a diverse array of ligands and in turn initiate and fine-tune transcriptional events. The impact of NR gene regulatory complexes is evident in development, metabolism, circadian rhythm and cell fate decisions including differentiation phenotypes. Reflecting this widespread importance, there is clear evidence for their disruption acting as disease drivers for various syndromes including cancer [1,2,3,4,5].

The classical sex steroids bind cognate receptors with high affinity; estradiol binds estrogen receptor, NR3A1/ERα, and dihydrotestosterone binds the androgen receptor, NR3C4/AR. Beyond these ligands seco-steroids, retinoid derivatives and bioactive dietary-derived factors such as fatty acids, oxysterols, heme, and bile acids act as ligands and regulate the genomic interactions of a broader group of the NRs. More broadly, these integrated and environmentally-driven NR-genomic interactions are central to concepts such as nutrigenomics and provide the rationale for positioning a wider panel of NRs as promising therapeutic targets in cancer [6,7,8]. Finally, other NRs, without known ligands, have also been identified, known as orphan receptors [9]. Collectively, the interaction of all these NRs allows for the highly dexterous transcriptional outputs, underpinned by the dynamic and mobile NR-genomic interactions, known as NR cistromes. In turn, the NR cistrome gene-regulatory functions are regulated by NR-associated coregulators including coactivators, corepressors and other transcription factors (TFs) and thereby provide a further level of control to regulate transcription [10,11,12,13].

NRs are classified based on mode of action as Type I, II, III, or IV [14]. Steroid NRs are Type I and in the absence of ligand these receptors are often largely cytoplasmic associated with heat shock proteins. Ligand binding results in their dissociation from heat shock proteins and NR homo-dimerization and translocation to the nucleus. Type II NRs, in contrast, reside in the nucleus as heterodimers (for example with RXRs) and bound to genome even in the absence of ligands [15]. Types III and Type IV are orphan receptors, for which ligands are unknown, or possibly don’t exist, and are also generally located in the nucleus and bind DNA as homodimers (Type III) or monomers (Type IV).

The impact of NRs is highly evident across many high-profile and impactful hormone-dependent cancers, including not only prostate cancer (PCa) and breast cancer (BrCa), but also other cancers including ovarian, endometrium, testis, thyroid, and pancreas. An appreciation of the relationship between steroids and cancers of the reproductive system was pioneered by the work of Sir George Beatson in the nineteenth century, who began to define the relationship between estrogen and BrCa risk [16]. Subsequently, in the 1940s this concept was echoed by the work of Dr. Charles Huggins and colleagues who established the endocrine synthesis of androgens and the relationship to PCa [17]. As a result, the genomic functions of AR and ERα in PCa and BrCa, respectively, are highly studied and these are well understood TFs. Additionally, there is a parallel and, in many cases, emerging appreciation of how these cancers are impacted by non-steroidal NRs, and the potential for the genomic cross-talk between steroidal and non-steroidal NRs. For example, there are physiological and gene regulatory studies that strongly support the concept that Type I and Type II NRs function in a range of cooperative and antagonist cross-talk signaling mechanisms, for example centered around AR [18,19,20,21,22,23], and ERα [24,25,26,27,28,29].

The focus of the current chapter is to summarize genomic insights into the Type II NRs in hormone-dependent cancer including the vitamin D receptor (NR1I1VDR), retinoic acid receptors (NR1B1/RARα, NR1B2/RARβ, and NR1B3/RARγ), and peroxisome proliferator-activated receptors (NR1C1/PPARα, NR1C2/PPARδ, and NR1C3/PPARγ) [9], summarized in Table 13.1. Clearly, orphan receptors, given they have no identified ligands, also fall under the classification of non-steroidal receptors. In parallel, the understanding of adopted nuclear orphans and orphan NRs is evolving, and reveal further insights into NR functions in terms of genomic distribution and cross-talk with signaling pathways including those that are key targets for pharmacological pathways [30].

Table 13.1 Nonsteroidal nuclear receptors
Full size table

2 Genomic Interactions of Non-steroidal Nuclear Receptors in PCa and BrCa

2.1 The Vitamin D Receptor

Supporting an anti-tumorigenic role for the VDR men whose prostate tumors have higher VDR expression have significantly lower prostate-specific antigen, lower Gleason score and less advanced tumor stage [31]. The circulating pre-hormone vitamin D3 is the precursor to the active hormone calcitriol (1alpha,25dihydroxyvitaminD3 (1α,25(OH)2D3)) that binds to the VDR. Epidemiological approaches have identified relationships between low circulating vitamin D3 and cancer incidence, and that 1α,25(OH)2D3 suppresses early prostate carcinogenesis by regulating genes involved in proliferation, differentiation and apoptosis [32]. Underscoring the potential importance of this signaling axis, genomic studies in murine VDR knockout cells as well as human studies have suggested that 1α,25(OH)2D3 can regulate as much as 3% of the mouse or human genome directly and/or indirectly [33].

Several studies have assessed the VDR cistrome in PCa [34, 35] by VDR chromatin immunoprecipitation sequencing (ChIP-Seq). Work by Fleet et al [34] identified binding at ~3400 protein-coding genes, ~680 long noncoding RNAs, and ~ 470 miRNAs. This included VDR-bound peaks at known VDR target genes including CYP24A1 and IGFBP3. Peak distribution was evenly divided between intergenic and intronic regions, supporting both long-range and proximal regulation. These studies also suggested that 1α,25(OH)2D3 amplifies signals mediated through other TFs including NF-Kappa-B Inhibitor Alpha (NFKBIA) and FOXO1, and some peaks near immune response related genes (e.g., L1R2) hint towards VDR regulation of immune processes.

A further VDR-ChIP Seq study in non-malignant prostate cells (PrEc) [35] identified ~5000 VDR binding sites, again including well-known targets (e.g., CYP24A1) and, interestingly, ligand activation led to a significant decrease in the number of VDR-ChIP peaks, reflecting perhaps an active role for the basal VDR in gene expression. Sites with loss of peaks include aminoacyl tRNA synthetase genes, which in turn leads to decreased proliferation. VDR also binds near genes regulating neural differentiation, which raises a possibility that itmay also be linked to neuroendocrine trans differentiation in PCa.

Finally, a recent study from our lab [doi.org/10.1101/2022.01.31.478573] addressed VDR function in the context of PCa health disparities by examining a panel of European American (EA) (HPr1-AR and LNCaP) and African American (AA) cell lines (RC43N, RC43T, RC77N and RC77T). These analyses lent strong evidence to the concept that the VDR is a significantly more potent transcriptional regulator in AA than EA prostate cells, and that in PCa this signaling is distorted and suppressed. In non-malignant RC43N cells, VDR ChIP-Seq identified significant basal and 1α,25(OH)2D3 dependent VDR binding sites, with ~1300 in total associated with transcriptional responses enriched for circadian rhythm and inflammation networks. In parallel, 1α,25(OH)2D3-dependent ATAC-Seq also revealed the greatest impact on chromatin accessibility in RC43N cells, with significant gain of nucleosome-free regions at enhancers. By contrast, in malignant EA and AA cell models 1α,25(OH)2D3 led to a loss of VDR binding. Motif prediction identified a diverse set of enriched motifs within peaks, including the VDR motif and other NRs including the AR and RARs. The suppressed transcriptional responses in AA PCa cells associated with reduced expression of Bromodomain adjacent to zinc finger domain protein 1A (BAZ1A), a component of the human SWI/SNF complex, and restored expression of this protein led to significantly enhanced 1α,25(OH)2D3-regulated transcriptome.

There are also equally compelling epidemiological associations between vitamin D3 and breast cancer incidence. For example, the Vdr −/− mouse [36, 37] displays a range of mammary gland phenotypes in terms of disrupted development of the gland, and then changing sensitives to the control of programmed cell death within epithelial cells. In parallel there are a wide range of pre-clinical studies which all support a potentially anti-tumorigenic role in BrCa [38].

Two studies have examined VDR genomic interactions which revealed that in MCF-7 BrCa cells, VDR has ~2300 VDR-binding sites in the absence of 1,25(OH)2D3, and ~7,400 sites following ligand stimulation (4 h). Out of these, ~700 sites remained unchanged in both presence and absence of ligand. A significant numbers of VDR-binding sites were detected in intergenic regions, and distal from promoters, and VDR-bound enhancers were enriched in apoptotic and metabolic pathways. In a series of comprehensive studies led by Kevin White and coworkers [39, 40] multi-cistrome analyses were undertaken for a range of more than 20 NRs including non-steroidal ones in BrCa cancer cell lines [39,40,41]. Within these studies VDR binding was analyzed in MCF-7 cells and also reported ~7000 binding regions, which were more distal to TSS regions than many of the other NRs, and in terms of network topology demonstrated lower interconnectedness compared to NRs such as the retinoic acid receptors. These workers were able to undertake integrative regions.

Together these data strongly support the VDR playing an important role in the biology of the prostate and mammary glands, and suggest disruption of VDR signalling is carcinogenic by disrupting a wide number of gene regulatory mechanisms including overlap with other NRs.

2.2 Retinoic Acid Receptors

The NR1B1/RARα represents one of the earliest examples of targeted cancer therapy, involving all-trans retinoic acid in acute promyelocytic leukemia [42, 43]. This was a major catalyst for the development of the field of differentiation therapy, whereby compounds such as retinoic acid would be in cancers to limit their proliferation and induce either differentiation or programmed cell death [44, 45]. In part, these actions were the motivations for cistromic studies on the VDR, RARs and multiple NRs in PCa, and BrCa [44, 45].

In the prostate, retinoic acid regulates normal differentiation and the Rarγ knockout mouse exhibits prostate metaplasia [46, 47], both suggesting the receptor plays a role in control of cell growth. Reflecting this, NR1B3/RARγ is commonly down-regulated in PCa3, for example because of up-regulated miR-96-5p, and this leads to significant changes to AR signaling [48]. In a non-malignant prostate cell line, RWPE-1, under basal conditions the RARγ cistrome is ~1250 peaks and interestingly the addition of a RARγ-selective ligand (CD437) restricts the number of peaks to ~350, which are mostly shared with the basal state (only ~50 appear unique). These data also revealed that RARγ significantly enhanced AR function, and regulation of AR target genes, and that the RARγ cistrome significantly overlapped with AR binding at active enhancers. In turn, reduced expression genes that were annotated RARγ binding was associated with aggressive PCa [48].

In MCF-7 BrCa cells, RARα/γ and ERα form a genomic antagonism [40] in a so-called “Yin and Yang” manner to regulate proliferation and survival. These NRs balance expression of shared gene targets in part because RARs overlaps significantly with ERα binding in a genome wide fashion. These co-occupied regions are in the vicinity of genes for which estrogen and retinoic acid regulate antagonistically. The number of peaks in the presence of selective RARα (AM580) and RARγ (CD437) ligands was ~7300 for RARα and ~ 3200 for RARγ sites, and using a generous distance cut-off of 1 kb between the center of the peaks there was a significant overlap of sites; it is unclear how many of the peaks actually overlap as opposed to being closely adjacent. This therefore suggests convergence at the level of gene-regulatory actions rather than perhaps direct chromatin-accessibility [40]. Together, these data suggest significant genomic interactions between RARs and both AR and ERα in PCa and BrCa.

Interestingly, the related paralog, RARβ, appears to be a bona fide tumor suppressor in BrCa and PCa. For example, methylation patterns of the CpG islands associated with the RARβ promoter are exploited in algorithms to predict tumor grade and progression risks in these tumors [49,50,51,52]. Against this backdrop it is perhaps surprising that there are no cistrome data for this receptor in these cancers, although it has been undertaken in brain tissues [53].

2.3 RAR Related Orphan Receptor C

NR1F3/RORC encodes RORγ and is amplified and upregulated in metastatic recurrent PCa tumors following androgen deprivation therapy. It acts as an upstream regulator of AR and appears to drive AR expression, as well as to facilitate recruitment of coactivators such as Nuclear Receptor Coactivator 1 and 3 (NCOA1/3, SRC1/3). Furthermore, pharmacological targeting with an antagonist to RORγ reduces expression of AR as well as the oncogenic AR splice variant 7 and reduces AR genomic binding, and as a result reduced expression of various AR target genes. This regulation appears to be a targeted AR event, as inhibiting RORγ does not alter genome-wide histone modifications associated with chromatin accessibility [54].

Studies on RORγ in BrCa have suggested that its function is an essential activator of the cholesterol-biosynthesis program, as it binds to cholesterol-biosynthesis genes, and it facilitates the genomic recruitment of Sterol regulatory element-binding protein 2 (SREBP2) in Triple-negative BrCa [55]. From a genome-wide perspective there appear to be a massive number of RORγ binding sites in the HCC70 BrCa cell line, in excess of 30,000, and these are highly shared with SREBP2 binding sites. Again, similarly to PCa, a RORγ antagonist very potently inhibits BCa tumor growth in vitro and in xenografts [55]. Similarly, the related RORα is also a potential tumor suppressor and a therapeutic target for BrCa [56, 57] but as yet cistromic studies have not been undertaken and so the extent of genomic cooperation between these two receptors remains unknown. RORγ therefore plays a paramount role in regulating cholesterol-biosynthesis through its own genomic binding leads to the recruitment of SREBP2 at the gene targets to stimulate the cholesterol-biosynthesis.

2.4 Peroxisome Proliferator-Activated Receptors

PPARs regulate energy production, lipid metabolism, and inflammation [58]. In triple negative BrCa MDA-MB-231 cells, ChIP-Seq and transcriptomic analyses identified ~500 PPARδ peaks and, amongst these, the hormone ANGPTL4 was a significant PPARδ target [59]. In another study, using a transformed variant of the non-malignant breast epithelial cell, MCF10A-NeuT cells, PPARγ binds to a large number of sites and regulates genes and notably EphA-Amphiregulin as well as genes involved in chemokine signaling [60]. Similarly, PPARα, PPARδ and PPARγ bind to ~2230, ~3250 and ~ 6300 genomic regions respectively in MCF7 cells with PPARγ binding as sites at a greater distal distance to TSS [39] than the other PPARs. Interestingly, the PPARδ cistrome shared a significant proportion (~70%) of its binding sites with RARα and RARγ, and in part this led to the concept of high occupancy target (HOT) regions in the genome. Specifically, these are regions that are significantly shared by multiple NRs and other TFs, and appear to be found disproportionately associated with genes associated with cancer development and progression. The functional significance of these sites is illustrated by shared PPARδ and RARs binding sites at target genes, which in turn are associated with poor prognosis in BrCa. More widely these genomic findings also support a concept of selectively targeting RARs and PPARδ to inhibit synergistically BrCa growth.

Set against these interesting data, to date there are no cistromic studies of PPARs in PCa. This is all the more striking given that there is a considerable literature on PPARs [61,62,63,64,65] and the PPAR coregulator PPARGC1α [66,67,68] playing significant roles in PCa carcinogenesis. Such studies would also be able to address the concept of HOT regions in PCa, and how these cistromic patterns impact AR signaling.

2.5 Hepatocyte Nuclear Factor 4 α and γ

In PCa, NR2A1/HNF4γ appears to function as a pioneer factor that generates and maintains enhancer landscape at lineage genes, for example those associated with neuronal lineages, and which impacts AR signaling in a more nuanced manner. For example, restoring HNF4γ expression reduces AR sensitivity towards androgen deprivation therapy [69], and increased HNF4γ expression does not alter the AR cistrome or AR signaling directly, but increased FOXA1 binding at a subset of HNF4γ sites. Approximately 35% of HNF4γ peaks share binding FOXA1, and a smaller proportion of HNF4γ peaks directly overlap with AR peaks. Therefore, HNF4γ binding sites appear to cooperate with FOXA1 to establish and maintain enhancers that facilitate lineage-specific transcriptomes in the prostate; this is potentially corrupted in PCa progression [69]. Similarly, NR2A2/HNF4α appears to exert a tumor suppressor function and has reduced expression in PCa tissues, cell lines, and xenografts of androgen deprivation therapy recurrent PCa [70] through epigenetic mechanisms. For example, HNF4α binds constitutively to binding sites in the promoter of CDKN1A, which guides AR to bind upon dihydrotestosterone stimulation. Indeed, the motifs of HNF4α are over-represented within unique AR-binding loci, and the cistrome shows significant overlap with AR-binding sites [71]. Again, given these potent cooperative actions between HNF4 receptors with a principal steroid hormone receptor, it is perhaps surprising that similar studies haven’t yet been undertaken in BrCa.

2.6 COUP Transcription Factor I and II

NR2F1/COUP-TF I is one of the earliest cloned NRs, first being identified in the late 1980s [72], and subsequently led to the discovery of NR2F2/COUP-TF II [73]. Several studies [39, 74, 75] have analyzed the COUP-TF II cistrome in BrCa. High expression of COUP-TF II is related with better survival in ERα-positive BrCa patients but not in ERα-negative patients, and COUP-TF II cooperates with pioneer factors such as FOXA1 and GATA3 to promote ERα function [74, 75]. These findings suggest a cooperativity between ERα and COUP-TF-11, and although estradiol is not required for COUP-TF II binding, inhibition of COUP-TF II decreases ERα binding, chromatin accessibility (ATAC-Seq peaks were reduced by 70% after COUP-TF II depletion), and estradiol-dependent cell growth suggesting a protein-protein interaction. Together, these data suggest a complex interdependency between estradiol, ERα and COUP-TF II. In MCF-7 cells, approximately, 40% of ERα binding sites overlap with FOXA1, 60% with COUP-TF II and 70% with GATA3, and there is evidence for shared binding at super-enhancers on a wide-spread scale which directly leads to high de novo transcription. Indeed, this integration also impacts other NRs downstream, including RARβ [76]. These roles for COUP-TF II in regulating ERα-mediated transcription make it an interesting potential therapeutic target in BCa. In parallel studies COUP-TF I-specific agonists suppress metastasis supporting a wider role for COUP-TFs to interact with ERα and to regulate anticancer actions [77].

2.7 NUR77

NR4A1/NUR77 is an orphan NR that acts in a ligand-independent manner. In a recent study [78], NUR77 was reported to regulate immediate early genes, suppressing replication stress in BCa and acting as a master regulator through a transcriptional processing checkpoint. Genome-wide analyses revealed that NUR77 binds the gene body and 3’ UTR of immediate early genes, inhibits transcriptional elongation, generating R-loops and accessible chromatin domains. Under stress, dissociation of NUR77 leads to a burst of expression of these transcriptionally poised genes thereby suggesting a role for NUR77 in governing transcriptional responses to chronic replication stress. Although there are no genome-wide cistrome studies of NUR77 in PCa, there is strong evidence for it regulating programmed cell death in this cancer [79, 80].

3 Mechanisms of NR Cooperation: Bookmarking Functions by Non-steroidal NRs

Mitotic bookmarking functions to retain epigenetic states throughout the cell cycle at gene loci that are poised for immediate reactivation post-mitotically (Fig. 13.1). This involves the retention of histone variants, regulatory proteins and modifications, and some selected TFs. Bookmarking mechanisms prevent the spreading of heterochromatin into genomic regions which are pre-marked for TF future actions. In this manner, these epigenetic mechanisms regulate genes that coordinately control cell growth and lineage maintenance following mitosis. Furthermore, it is clear these mechanisms are corrupted in carcinogenesis and tumor maintenance leading to deregulated proliferation and compromised control of differentiation [81,82,83,84,85,86].

Fig. 13.1
An image illustrates a cycle that highlights G 1, G 0, S, G 2, Mitosis bookmaking pe ii N R, bound transcription factor, degradation, and displacement.

A model for bookmarking function by Type II nuclear receptors for Type I nuclear receptors. As cells go through the cell cycle and division chromatin assumes different conformations, becoming most compacted during metaphase of mitosis. Prior to this, many proteins to chromatin associations are lost as a result of degradation and displacement. However, a number of transcription factors are retained such that transcription, or the marking of sites for transcription, can be activated rapidly in G1. This function is termed bookmarking and there is evidence that Type II nuclear receptors that are nuclear resident in both the presence and absence of ligand (mitosis, purple open symbols) can serve this function for other nuclear receptors (G1, purple solid symbol)

Full size image

Several Type II NRs have been reported to have bookmarking properties independent of ligand exposure, again reflecting their predominant nuclear location. NR1I2/PXR remains constitutively associated with mitotic chromatin specifically at the CYP3A4 promoter during mitosis [88]. A region of PXR contains a ‘mitotic chromatin binding-determining region’ which exerts these functions. The bookmarking property of PXR is impeded by direct interaction with the orphan NR small heterodimer partner (SHP) perhaps underscoring the importance of this function [89]. Other examples of NRs appearing to play a bookmarking function include NR3B2/ESRBB, which is a major pluripotency TF that remains bound to key regulatory regions during mitosis [90]; it is bound widely with at least 10,000 binding sites and maintains nucleosome positioning during mitosis to ensure the rapid post-mitotic re-establishment of functional regulatory complexes at selected enhancers and promoters [91]. Similarly, HNF4α bookmarks specific genomic regions and keeps them competent for future activation during liver development [92].

The raises an interesting question of whether this bookmarking property is a generalized feature of NRs, and specifically those NRs that are nuclear resident independent of ligand exposure maybe retaining enhancer access through mitosis for other NRs. This concept is supported by examples above of Type II and Type I NR co-regulation of gene expression programs. Given that non-steroidal NRs in PCa and BrCa are frequently disrupted for example with decreased expression (e.g., RARγ), this may suggest that Type II NRs bookmark and regulate the actions of AR and ERα. However, there is also evidence of ligand activated (and therefore nuclear resident) AR and ERα being associated with mitotic chromatin although it is unclear if these complexes are the cause or consequence of other NRs/TFs serving as bookmarking factors [87].

More generally, there are clear examples of AR and ERα being genomically relocated to other sites during cancer initiation and progression, and in response to NR-targeted therapies. For example, the AR is reprogrammed specifically to genomic sites that are normally regulated in development only in the transition to metastatic PCa by reactivating latent regulatory elements active in fetal prostate organogenesis [93]. It is a tantalizing prospect that the interactions between Type II and Type I NRs is in part underpinned by Type II NR bookmarking enhancers and regulatory regions that are regulated by Type I NR binding to promote cell fate decisions such as differentiation. Furthermore, disruption of these Type II NR complexes potentially disrupts these functions.

4 Genomic Approaches to Defining Type I and II NR Cistromes and Interactions

Methods to map histone and TF genomic interactions emerged in the 1990s with the development of ChIP approaches [94,95,96], and became genome-wide with the advent of microarray technologies giving rise to so-called ChIP–chip [97] approaches, and then subsequently ChIP-Seq [98]. This key technology has been profoundly improved and diversified to tackle limitations such as protein abundance, cross-linking efficiency and antibody availability and specificity. For example, Cleavage Under Targets & Release Using Nuclease (CUT&RUN) and CUT&Tag has made it easier to study TF binding and histone modifications at genome scale [99, 100]. Similarly, the development of ATAC-Seq (Transposase Accessible Chromatin followed by high-throughput sequencing) [101] has enabled the measurement of chromatin accessibility and has also been refined to address single cells and to improve accuracy. More widely, genomic approaches are advancing rapidly to encompass single cell resolution, which allows ever more complex biological questions to be addressed [102]. In parallel, CRISPR technologies are enabling the tagging of proteins, and DNA and epigenome editing, to more establish conditional cell contexts with which to test NR functions more accurately [103, 104].

Matching these wet-lab advancements has been an equally explosive growth in the dry-lab to develop and refine the analyses of cistromic data and combine it with parallel transcriptomic data. This challenge of integrating cistrome to transcriptome data is surprisingly complex. For example, defining NR:enhancer:gene interactions that are driven by NRs is challenging because of the large number of NR and coregulator interactions, which are altered by diverse and interdependent genetic and epigenetic mechanisms, and are further controlled by the 3-D genome [105,106,107]. Thus, NR:enhancer:gene relationships are dynamic and non-linear, with each gene regulated by multiple enhancers in a time- and signal-dependent manner [108, 109], and occur over large genomic distances [110].

Defining the statistical significance of NR:gene relationships, or even NR:NR:gene relationships, is a question of whether a NR signal-to-gene-expression relationship is occurring more than predicted by chance, which in turn requires defining the background of NR:gene relationships. Random sampling methods such as bootstrapping can be used to simulate the distribution of NR:gene relationships changes across the genome for statistical comparison [111, 112], and parsimonious annotation of the genome, for example with the ChromHMM algorithm [113] to define epigenetic states, or the ROSE algorithm to define super-enhancers [114] [115] can refine these statistical challenges. Furthermore, testing the overlap of target NR ChIP-Seq data with comprehensive data sets, such as contained in Cistrome DB [116], allows co-enrichment testing of hundreds of TF and histone modification ChIP-Seq datasets to reveal the extent of enrichment with other NRs and their coregulators. RNA-Seq undertaken in parallel treatments can be matched with these highly annotated cistromic data to define cistrome-transcriptome relationships and test their phenotypic associations for example using Kolmogorov–Smirnov tests to examine differences in cumulative distribution plots for cistrome binding sites with respect to nearest gene, and again using bootstrapping approaches to measure how the specific cistrome-relationships associate with gene expression patterns [117].

Thus, there are many routes through testing NR:gene relationships and this most likely underpins the frequently divergent findings in the literature. On top of this there are multiple methods for cistrome [118, 119] or transciptome [120, 121] analyses and as yet there are few commonly accepted protocol standards, in contrast, for example to the MIAME-compliant protocols for microarray analyses [122]. Therefore, it is unsurprising that for a given NR there is little consensus on the number of significant binding sites, what motifs are most enriched, what the genomic distribution is and how it relates to transcription.

5 Conclusion

Non-steroidal NRs have been somewhat neglected from a genomic perspective, although it is clear their actions and interactions with steroidal NRs are biologically impactful. In this chapter we attempted to provide a broad overview of the advances in understanding non-steroidal nuclear receptor cistromes and their interaction with other AR and ERα in PCa and BrCa and highlighted the expanding impact of the genome wide studies in NR biology. These NRs are potential therapeutic targets in cancer and may be exploited to augment traditional therapeutic approaches. Cistromic studies are rapidly advancing and revealing unprecedented insights into the interactions between Type I and Type II NRs, even with some methodological ambiguities.