Notably, coding SNPs may interact with other SNPs to produce a stronger functional role ^[5]. The rs138649767 A allele (Figure 1b) located in the exon region of

TCF7L2 can activate the

MYC enhancer containing rs6983267 allele G to promote the expression of MYC ^[16]. SNPs occurring in the exons and introns of

SMAD7 may affect its regulation and jointly affect downstream signaling pathways involving SMAD7 and TGFβ ^[11]. As a result, while examining SNPs in coding regions, the interactions between them should be taken into account to better understand their functional processes.

Accumulating evidence shows that a SNP in non-coding regions is the most common type of genetic variation in the human genome, accounting for 90% of inter-individual variation ^[17][18]. Depending on the location, the region can harbor a response element that is either proximal (promoter, enhancer, or super-enhancer) or distal (intergenic or intra-genic). The risk loci identified by GWAS were located in the genomic regions of cell type-specific active chromatin, and most of them were quantitative trait loci, methylation quantitative trait loci and transcription factor (TF) binding related loci. Chromatin conformational studies have helped to link regulatory regions localized by SNPs to their respective target genes ^[17][19][20]. These loci may be involved in gene transcription, post-transcriptional processing, translation, post-translational modifications, and other processes to regulate gene expression. Many target genes have been identified using expression quantitative trait loci (eQTL) to detect the relationship between SNPs and gene expression. Non-coding SNPs can regulate the transcription of target genes by sequence-proximal (cis)- or distal (trans)-interactions. Studies have found that histone modifications in the regions of such risk SNPs are particularly abundant, especially those related to promoter and enhancer activities (H3K4me3, H3K4me1, H3K27ac). Most SNPs are predicted to destroy the binding motifs of specific transcription factors. For example, rs6983267 may change the binding of transcription factors such as MYC, CTCF, and TCF7L2 ^[21]. In addition to affecting gene transcription levels by altering transcription factor-binding sites (TFBS), non-coding SNPs also change epigenetic modifications and/or the chromatin structure to influence target gene expression. Through the above method, non-coding SNP participates in cell proliferation, apoptosis, migration, and invasion.

A promoter is a sequence of DNA that is recognized, bound, and serves to initiate transcription by RNA polymerase. Promoters contain variations of a conserved sequence required for the specific binding of RNA polymerase and transcription initiation. Most promoters are located upstream of the transcription initiation point of structural genes, and the promoter itself is not transcribed ^[22]. Promoters are located upstream of the 5’ end of a given structural gene, and they activate RNA polymerase to bind accurately to the template DNA with specificity for inducing the initiation of transcription ^[22]. Promoters do not control gene activity themselves; rather, gene activity is regulated by binding to proteins called transcription factors (TF). SNPs within promoter regions generally play a regulatory role by influencing the binding of such transcription factors. A recently reported example is that of the SNP rs13278062 located in the promoter of death receptor 4 (DR4) which confers an altered risk of colorectal cancer. The study revealed that the rs13278062 G>T variant changed the binding affinity of the transcription factor Sp1/NF1, increased the expression of DR4, and thus suppressed carcinogenesis and metastasis of colorectal cancer ^[23]. The MPO promoter SNP rs2333227 increases the malignant characteristics of colorectal cancer by changing the promoter’s affinity to AP-2α ^[24]. The variant SNP rs10993994 located in the upstream promoter of the gene MSMB is also found to be overrepresented in individuals with prostate cancer; this is attributed to stronger CREB binding and thus increased promoter activity ^[25]. Furthermore, the SNP rs11672691 is a risk locus associated with prostate cancer that is related to the lncRNA PCAT19. The non-risk variant rs11672691 and its linkage disequilibrium (LD) SNP rs887391 are more likely to bind the TFs NKX3.1 and YY1 to the PCAT19-short promoter, thereby leading to increased promoter but lower enhancer activity, which then activates PCAT19-short, and ultimately results in lower prostate cancer susceptibility ^[26]. SNPs in promoter regions of multiple genes, including TERT, KLHDC7A, PIDD1, and ESR1, have been discovered in breast cancer by GWAS, with reporter studies revealing that independent risk alleles change target promoter activity ^[27][28]. Most of the reported promoter changes exert their regulatory effects by altering TF binding. The SNP rs3824662 allele A (Figure 2a) increases chromatin accessibility by changing the TF GATA3 expression, promoting the binding of GATA3 with the CRLF promoter, and ultimately forming a chromatin loop ^[29].

Enhancers are regions of DNA sequence that can increase the cis-acting transcription of their target gene sequences. Enhancers each differ in their distance from their target promoter(s); in mammalian species, an enhancer can be 100 bp to Mb away from their target gene ^[30]. Enhancers, unlike promoters, can be found anywhere in a gene; they can be positioned either upstream or downstream of their target genes, or even within another gene’s gene body, and enhancer regulation can circumvent other genes irrespective of their orientation. Enhancers must bind to specific protein factors to enhance the transcription of their target. Enhancers generally have tissue or cell specificity, whereby they only show activity in certain cells or tissues, which is determined by the specific protein factors present in these cells or tissues ^[31]. Enhancers are typically recognized by the epigenetic marks H3K4me1 and H3K27ac, which are present in active enhancer elements. Conversely, H3K27me3 is regarded as a silent epigenetic mark associated with lower enhancer activity ^[32][33]. GWAS-identified risk loci for common illnesses are often found in non-coding areas, and many of these are thought to function as enhancers ^[34]. According to emerging data, these SNPs may influence gene regulation by changing the binding of important TFs to critical transcriptional enhancers ^[35].

Of all cancers, breast cancer has so far yielded the greatest number of discovered risk loci ^[36]. Understanding the driving mechanism(s) underlying malignant transformation provides the prospect of combating cancer recurrence and treatment resistance. Zhang et al. identified that the SNP rs4971059 resides in the sixth intron and within an active enhancer element of the TRIM46 gene. By using CRISPR/Cas9-mediated homologous recombination, they constructed the SNP rs4971059 with the allele G converted to allele A, thereby resulting in TRIM46 overexpression, boosting breast carcinoma cell growth, enhancing chemotherapy resistance in vitro, and hastening tumor development in vivo ^[37]. In addition, Yang and colleagues (Figure 2b) reported the noncoding regulatory variant rs11836367 at the NTN4 locus (12q22) and identified it to be associated with the risk of breast carcinoma as a causal variant. The rs11837367 protective T allele promotes GATA3 binding to the distal enhancer and increases NTN4 expression ^[38].

2.2.3. Colorectal Cancer

GWAS have identified numerous colorectal cancer risk loci, but only a fraction of the target genes of these loci have been systematically interrogated. For example, Yu et al. identified a common SNP (rs7198799) in the intron of the gene

CDH1. They demonstrated that the risk allele C of rs7198799 acts as an enhancer that can target the TF NFATC2 and remotely enhance ZFP90 expression ^[45]. A prominent mechanism by which SNP variants can affect cell-specific enhancer function is via altered TF binding, thus regulating the target gene’s expression. Tian et al. identified two risk SNPs (rs61926301 and rs79591129) located in the ATF1 promoter and first intron, respectively. These are enriched in enhancer regions and open chromatin, which are also associated with H3K4me1, H3K27ac, and ATAC-seq peaks. The two variants increase the expression of ATF1 through preferentially binding to the two TFs SP1 and GATA3 ^[46]. Rs174575 can act as a specific remote enhancer of FADS2 and lncRNA-AP002754.2 with the participation of the transcription factor E2F1. Interestingly, TF E2F1 can promote the expression of FADS2, form a chromatin loop, and affect the occurrence of colorectal cancer ^[47].

Promoter–enhancer interactions (PEIs) underlie differential transcriptional regulation. Several technologies (chromosome conformation capture (3C), Hi-c, and H3K27Ac-HiCHIP) allow for the study of long-range cis-regulation ^[48][49][50]. Promoter–enhancer interactions are essential events involved in the current theory of transcriptional control. So far, there is little evidence that PEIs are required for the transcriptional control of an enhancer’s target gene. The insertion or deletion of promoters, the absence of certain PEI-associated proteins, and the inclusion of PEI-disrupting insulators all have an effect on the expression of target genes. Tian et al. found two risk variants (rs1926301 and rs7959129) located in the ATF1 promoter and intron, respectively; the former binds the TF SP1 while the latter binds the TF GATA3 (Figure 2c). They found that these two risk sites increase the interaction between the promoter and enhancer by binding SP1 and GATA3, facilitating ATF1 expression, and conferring hereditary susceptibility to CRC ^[46]. Moreover, the SNP rs11672691 mediates promoter and enhancer switching in a manner dependent on different background genotypes. The risk is determined by the PCAT19-long enhancer interacting with the PCAT19-long promoter, thereby altering prostate cancer development through activating cell cycle genes ^[26].