Chromosome 8q24–Associated Cancers and MYC

Abstract

Genome-wide association studies (GWAS) have successfully identified common polymorphisms that are strongly associated with many traits, including cancer. A gene desert located on chromosome 8q24 is associated with multiple cancer types. One of the closest genes is the MYC proto-oncogene. Investigations are now turning toward a mechanistic understanding of these (and other) risk loci. Recent studies demonstrate that the 8q24 loci are enhancers and that they physically interact with MYC. A still unresolved issue is the absence of a consistent association between genotype status at the risk loci and steady-state MYC expression levels in adult human tissues. Clarifying the function of the 8q24 variants and their link to MYC regulation by further in vivo and in vitro functional studies will allow a deeper understanding of the mechanisms underlying human cancer development.

MYC and SNP Disease Association

One of the primary goals of human genetics is to understand the genetic basis of traits. The methods used to map genes are strongly dependent on inheritance patterns. Until recently, the only traits that were accessible to gene mapping were Mendelian disorders—that is, diseases that demonstrated a clear-cut pattern of inheritance, such as autosomal dominant, recessive, and sex linked. Linkage analysis has been tremendously successful in identifying numerous genes involved in various diseases. These studies allowed localization and identification of many highly penetrant genes.^1-4 The fine mapping of Mendelian disorders is relatively straightforward, sequencing the exons within the interval implicated by linkage mapping. The great majority of the time, one will discover a DNA variant that causes an amino acid–altering change with a predictable effect on protein structure.⁵ Although these studies continue to provide a valuable contribution to our understanding of inheritance of diseases and traits, the gene alterations explain only a small fraction of the total number of cases in the general population.^6-8

Only recently has it become possible to identify the genetic basis of non-Mendelian, or complex, diseases (i.e., diseases driven by multiple genes and environmental and behavioral factors). Understanding the genetic contribution to these more common disorders had to await the sequencing of the human genome and the subsequent cataloguing of common variants that exist in the human population.^9-12 These developments led to the successful implementation and execution of association studies. Association studies identify risk alleles by comparing frequencies of genetic variants in cases and controls.

Genome-wide association studies (GWAS) have identified thousands of variants associated with hundreds of traits.¹³ Similar to linkage studies, association studies are typically conducted in 2 phases: an initial phase, which localizes the disease region to a sub-chromosomal region, and a fine mapping phase, which localizes the causal variant. As mentioned above, most Mendelian disorders result from a DNA sequence change in the protein coding region; by contrast, the majority of common risk alleles discovered to date map outside of known protein coding regions (e.g., intronic and intergenic regions). This situation presents a challenge, as our ability to annotate and understand the non–protein coding region of the genome is rudimentary compared to our understanding of the protein coding portion of the genome.

A particularly interesting set of risk loci is clustered within a large gene desert on chromosome 8q24. Multiple-cancer GWAS have demonstrated an association at this region with prostate, breast, colon, ovarian, and bladder cancers and chronic lymphocytic leukemia^14-25 (including glioma,²⁶ but not in the portion of the 8q24 locus discussed here). To date, a total of 14 independent risk polymorphisms associated with various cancers have been mapped to this region (Table 1). Notably, the nearest annotated genes are FAM84B and MYC (Figure 1). The proximity of the MYC proto-oncogene to the variants, coupled with its longstanding history in cancer biology, makes this gene a particularly compelling candidate for driving tumorigenesis. A hypothesis consistent with the GWAS data is that the risk regions are regulatory elements of MYC.

Table 1.

Summary of Association Results for Genetic Variants in the 8q24 Region

#	SNP	Disease	Position	P Value	Reference
1	rs1016343	PC	chr8:128162479	1 × 10−7	18
2	rs10505477	OC; CC	chr8:128476625	2 × 10−3; 3 × 10−11	17; 24
3	rs9642880	BlC	chr8:128787250	7 × 10−12	17
4	rs13281615	BC	chr8:128424800	5 × 10−12	20
5	rs1447295	PC	chr8:128554220	2 × 10−19	14; 15; 19
6	rs1562430	BC	chr8:128457034	6 × 10−7	21
7	rs16901979	PC	chr8:128194098	3 × 10−14	13; 19
8	rs16902094	PC	chr8:128389528	6 × 10−15	19
9	rs2456449	CLL	chr8:128262163	8 × 10−10	16
10	rs4242382	PC	chr8:128586755	3 × 10−19	22
11	rs4242384	PC	chr8:128587736	2 × 10−24	25
12	rs445114	PC	chr8:128392363	5 × 10−10	19
13	rs6983267	OC, CC, PC	chr8:128482487	9.9 × 10−3; 1 × 10−14; 9 × 10−13	15; 17; 18; 22; 23
14	rs7014346	CC	chr8:128493974	9 × 10−26	23

Reported SNPs and shown P values originate from various recent genome-wide association studies. hg18 assembly was used to report chromosomal location of the variants. CC = colon cancer; OC = ovarian cancer; PC = prostate cancer; BC = breast cancer; BlC = bladder cancer; CLL = chronic lymphocytic leukemia; SNP = single-nucleotide polymorphism.

Figure 1.

Schematic view of 8q24 region. Modified and downloaded from of UCSC Genome Browser, hg18 assembly. Color coding: black, bladder cancer; green, prostate cancer; dark blue, chronic lymphocytic leukemia; red, breast cancer; light blue, colorectal cancer.

Multiple lines of evidence have revealed that the 8q24 risk regions possess regulatory activity. Recently, it has been demonstrated that particular constellations of epigenetic marks (e.g., histone modifications) and/or proteins (e.g., p300) can robustly annotate genomic regions.^27-29 Using a ChIP-on-chip approach, Jia et al.³⁰ constructed a dense epigenetic map of the 8q24 region in multiple cell types. Many of the known prostate, colon, and breast risk regions were enriched for marks that were consistent with enhancers.^31,32 More focused experiments have used ChIP to support the notion that the risk regions are regulatory elements. Two independent ChIP studies^33,34 have shown that the colorectal cancer risk locus, rs6983267, bears acetylated and methylated histone marks that are consistent with regulatory activity (H3K4me1, H3K4me2, H3K4me3, H3-Ac, H4-Ac). Similar findings were observed for 3 prostate cancer risk loci as well as a breast cancer locus.³⁵ In addition, reporter assays from various groups support the ChIP findings, showing that the prostate^36,37 and colon risk regions^33,36 can drive the expression of reporter genes. Moreover, 3 studies showed that the rs6983267 influences expression of a reporter gene in an allele-specific manner in colorectal cancer cell lines^33,34 and in the prostate of transgenic mice.³⁷ Lastly, these regions are largely devoid of transcriptional activity, as demonstrated by high-resolution tiling arrays.^30,33 In aggregate, these studies demonstrate that 8q24 contains multiple regulatory regions.

The rs6983267 risk variant falls in a transcription factor consensus binding sequence. Members of the TCF transcription factor family and the transcriptional coactivator β-catenin have been shown to regulate the Wnt signaling pathway and consequently regulate the expression levels of candidate genes, including MYC.³⁸ Interestingly, recent ChIP and reporter assay studies demonstrate that in colorectal carcinoma, these transcription factors bind to the region containing rs6983267.^39-41 In line with these findings, several groups^33,42 demonstrate that the colon cancer risk locus, rs6983267, differentially binds to transcription factor 7–like 2 (TCF7L2) in colon cancer cell lines; that is, TCF7L2 binds more avidly to the risk allele than to the non–risk allele. Experimental data also reveal that TCF7L2 binds to rs6983267 (a risk allele in common to both colon and prostate cancer) in a prostate cancer cell line.³⁶

Is MYC a target of the 8q24 enhancers? One method to measure interactions between regulatory elements and their targets is with the chromosome conformation capture (3C) assay. The 3C method is a powerful experimental technique that allows the detection of regions that are engaged in a physical interaction.^43,44 Several studies have shown that chromatin looping plays an important role in gene regulation.^45-47

Recent studies demonstrate that many of the 8q24 risk loci form long-range interactions with MYC. Two independent studies^33,34 showed that rs6983267 interacts near the MYC promoter in colorectal cell lines. Another group observed that in a colorectal carcinoma cell line, the MYC promoter interacts with 2 regulatory regions, 5′ (335 kb away) and 3′ (1.4 kb away) under the control of β-catenin.⁴⁸ The risk locus containing rs6983267 has also been shown to interact with the MYC promoter in prostate cancer cell lines.³⁶ More recently, Ahmadiyeh et al.³⁵ showed that the risk loci in 8q24 physically interact with MYC in a tissue-specific fashion by chromosomal looping. Specifically, the prostate risk loci interact with MYC in a prostate cancer cell line but not in colon and breast cancer cell lines, whereas the breast cancer risk locus shows physical interaction with MYC in a breast cancer cell line but not in a prostate cancer cell line.

These observations suggest that the 8q24 region harbors regulatory elements that regulate the expression of MYC. Furthermore, they show not only that interaction between regulatory elements and MYC is tissue specific but also that even within the same tissue type, the interactions may depend on the cellular state.

Since the 8q24 region contains enhancers that physically interact with the MYC proto-oncogene, a natural extension is to directly investigate if these regions are influencing MYC transcript levels. Multiple studies demonstrate that transcript abundance is a highly heritable trait; that is, RNA transcript levels are under genetic control (see Cookson et al.⁴⁹ and references therein). Transcripts that are associated with risk allele status are strong candidate genes for mediating the effect of risk alleles on disease.⁴⁹ By evaluating expression levels in normal prostate tissue, Solé et al.⁵⁰ analyzed normal tissue from 32 individuals for rs1447295. They observed that MYC overexpression is correlated to the genotype status of rs1447295. Subsequently, Pomerantz et al.⁵¹ evaluated 213 normal and 188 tumor tissue samples for rs1447295. These results showed that in both normal and prostate tissue, there is no convincing association between steady-state MYC transcript abundance and risk allele status of 6 SNPs, including rs1447295 and rs6983267. The inconsistent results between these 2 studies may be due to differences in statistical power, tissue purity, and/or differences in MYC RNA quantification. Interestingly, in a recent in vivo study, Wasserman et al.³⁷ demonstrated that the prostate cancer risk allele of the rs6983267 variant correlates with endogenous MYC expression during early prostate organogenesis, suggesting that risk variants might influence prostate cancer risk significantly before tumor formation.

Association studies between risk allele status and MYC transcript levels have also been carried out in colorectal tissue. Specifically, 3 independent studies evaluated normal and/or tumor colorectal tissue samples^24,33,42 and showed that no significant dependence could be observed, confirming previous observations.⁵² However, in a more recent study, Wright et al.³⁴ observed an increased expression of the cancer risk–associated allele of rs6983267 using an allelic ratio approach—that is, in a heterozygote cell line measuring the amount of transcribed MYC from each chromosome. This study suggests that MYC’s expression may be allele specific.

To date, most studies do not show a consistent association between risk allele status and MYC expression levels. Does the absence of clear-cut expression data rule out MYC involvement? Various reasons could explain why MYC transcript levels are not consistently associated with risk allele status. Perhaps the differences are too subtle to be picked up by the platforms being used. In addition, the association may be detected only at a particular developmental time point, in a particular set of cells (e.g., stem cells), when analyzed in the context of the complex cellular environment of specimens (e.g., due to dependence on a specific pathway or cellular interaction), or specifically measuring rate of synthesis and/or degradation (i.e., non-steady-state levels). Further studies will be necessary to address and clarify this issue.

GWAS and functional follow-up studies are beginning to shed light on the pathways that are driving human traits. It appears that the 8q24 risk variants act (at least in part) in a tissue-specific manner as regulatory elements of MYC, by physical interaction between the regions. Yet, at this time, it is unclear what the correlation is between MYC expression levels and risk allele status. It has been suggested that modest alterations in gene expression can result in trait variation, including disease pathogenesis.⁵³ In support of this hypothesis, a recent study demonstrates that subtle downregulation of PTEN can result in cancer susceptibility and progression in mice.⁵⁴ When and where the risk loci are influencing MYC expression levels will likely require more sophisticated modeling. Furthermore, it cannot be ruled out that the 8q24 risk variants may influence genes other than MYC.

Notably, this chromosomal location has been shown to be involved in somatic copy number alterations across cancer types.^55,56 Somatic amplification of 8q is one of the most prevalent copy number gains in cancer.⁵⁷ Potential connections between germline and somatic genomes are just beginning to be explored and will likely yield fresh insights into cancer biology.^58-62

Clarification of the function of these single-nucleotide polymorphisms and their link to MYC regulation, by resequencing and further in vivo and in vitro functional studies, may yield several important biological and clinical implications. It will allow a deeper understanding of the mechanisms underlying cancer development. Ultimately, the annotation of risk loci and the connection with their target genes will inform the pathophysiology of disease and may reveal opportunities to more rationally intervene in treatment and prevention.