The mechanisms underlying disease-associated genetic variants. (A) Location of a non-coding and a coding genetic variant relative to three genes: A, B, and C. The coding variant is likely to exert its influence on disease pathogenesis through an effect on the function of gene C. Several different mechanisms, including amino acid substitution and protein truncation, could be driving the association. The non-coding variant, on the contrary, could be acting through any of the three nearby genes, as shown by the dotted arrows indicating potential cis-regulatory links. The variant could, for example, affect either expression levels or splicing patterns, but identifying the causative transcript can be challenging. (B) The schematic depicted in panel A with additional functional annotations ( for details). The binding sites of three β cell specific transcription factors (TF 1, 2 and 3) are shown by lines connected to the respective proteins. Three binding sites cluster to form a stretch enhancer, which is likely to have islet-specific cis-regulatory activity (see the section ‘Understanding the impact of regulatory variants on islet gene function’). The non-coding variant is seen to overlap with the binding site of TF2 within the stretch enhancer, and could therefore plausibly act through disruption of TF2 binding. The solid arrow shows an established cis-regulatory link between the variant and gene 1. This relationship could be supported by different types of functional annotation, including a correlation between the non-coding variant and expression of gene 1 (established by cis-eQTL studies) or a physical interaction between the enhancer and the promoter region of gene 1 (demonstrated by chromatin-conformation capture, 3C) (). With such annotations at hand, identifying the causative transcript at a disease-associated locus is a more tractable problem.

Schematic representation of the pancreatic β cell and the location of key components implicated by human genetics in diabetes pathogenesis. Solid arrows indicate direct mechanisms of action and dotted arrows indicate mechanisms with intermediate events not displayed in the schematic. (A) Consensus model of glucose-stimulated insulin secretion (GSIS), including several key proteins involved in Mendelian forms of diabetes. Proteins depicted: GCK (encoded by GCK), GLUT1/2/3 (SLC2A1/2/3), insulin (INS), Kir6.2 (KCNJ11), SUR1 (ABCC8), and voltage-gated Ca²⁺ channels (CaVs), which represent both P/Q type and L-type channels ( for details). Briefly, glucose enters the cell through one of the glucose transporters GLUT1/2/3 and is phosphorylated by GCK to form glucose-6-phosphate (G6P). In the mitochondria, aerobic metabolism of glycolytic products results in the generation of ATP. Changes in the ATP:ADP ratio inactivates the ATP-sensitive potassium channel K_ATP, which consists of four SUR1 and four Kir6.2 subunits. Closure of K_ATP in turn causes depolarization of the membrane opening voltage-gated Ca²⁺ channels. Finally, the influx of Ca²⁺ triggers exocytosis of insulin granules. (B) Location and proposed role in β cell function of proteins encoded by genes discussed in this review. Proteins depicted: ANKRD15 (encoded by KANK1), CDK4/6 (CDK4 and CDK6), cyclin D2 (CCND2), KvLQT1 (KCNQ1), Mel-1B (MTNR1B), PAM (PAM), TBC1D30 (TBC1D30), ZnT-8 (SLC30A8), and the transcription factors GATA-6 (GATA6), MNX1 (MNX1), Nkx-2.2 (NKX2-2), PDX-1 (PDX1), PTF1-p48 (PTF1A), and TCF4 (TCF7L2). See main text for details of individual genes.

From association signal, through causative transcript, to biology. Beige boxes contain key experimental procedures; blue boxes indicate the entity of focus along the process of translating genetic associations to biology; and red boxes are divisions of such entities. As summarised in , human genetics can identify genomic regions implicated in diabetes risk. Subsequently, fine-mapping approaches can refine the genetic associations and identify the causative variant. Physiological characterisation of individuals carrying those risk alleles can in turn reveal effects indicative of the tissue(s) affected, but only indirect evidence for the underlying mechanisms (dotted line). Instead, identifying the transcripts driving the association of β cell relevant variants provides more direct insight into disease biology. For coding variants this can be achieved by prediction or validation of pathogenicity, whereas for non-coding signals consideration of prior candidate gene biology for nearby genes or variant transcript links can prioritize regional transcripts. Once a causative transcript has been identified, several different approaches, including human models, animal models, and functional studies, can be used to interrogate the function of the protein and ascertain its role in disease biology. Abbreviations: 3C, chromatin conformation capture; cis-eQTL, cis-expression quantitative trait loci ().