Diabetes Care 1 IBTSCARE DIABETES SYMPOSIUM The TCF7L2 Locus: A Genetic Struan F.A. Grant1–5 Window Into the Pathogenesis of Type 1 and Type 2 Diabetes https://doi.org/10.2337/dci19-0001

Over the past ∼15 years there has been great progress in our understanding of the genetics of both type 1 diabetes and type 2 diabetes. This has been driven principally by genome-wide association studies (GWAS) in increasingly larger sample sizes, where many distinct loci have now been reported for both traits. One of the loci that dominates these studies is the TCF7L2 locus for type 2 diabetes. This genetic signal has been leveraged to explore multiple aspects of disease risk, including develop- ments in genetic risk scores, genetic commonalities with cancer, and for gaining insights into diabetes-related molecular pathways. Furthermore, the TCF7L2 locus has aided in providing insights into the genetics of both latent autoimmune diabetes in adults and various presentations of type 1 diabetes. This review outlines the knowledge gained to date and highlights how work with this locus leads the way in guiding how many other genetic loci could be similarly used to gain insights into the pathogenesis of diabetes.

There is a general need to identify better drug targets with greater efficacy for common diseases, with gene discovery efforts presenting a specific possibility for novel therapeutic options. Given that a genetic component is well-known to exist for both type 1 diabetes and type 2 diabetes, genomics offers an opportunity to 1 uncover new high-value targets for these diseases. Divisions of Human Genetics and Endocrinology, Children’s Hospital of Philadelphia, Philadelphia, The evidence for a genetic component to type 2 diabetes is largely undisputed, PA given the combination of prevalence differences across different populations, familial 2Center for Spatial and Functional Genomics, clustering, and high concordance among monozygotic twins when contrasted with Children’s Hospital of Philadelphia, Philadelphia, dizygotic twins; indeed, the sibling risk for type 2 diabetes is now estimated to be ;3.5 PA 3Department of Genetics, Perelman School of (1), albeit varying somewhat between populations. In contrast to the relatively mild Medicine, University of Pennsylvania, Philadel- genetic component to type 2 diabetes, type 1 diabetes risk is far more clearly in- phia, PA fluenced by genetic factors, with it being most apparent within the MHC. 4Institute for Diabetes, Obesity and Metabolism, Perelman School of Medicine, University of Penn- GENOME-WIDE ASSOCIATION STUDIES sylvania, Philadelphia, PA 5Department of Pediatrics, Perelman School of Genome-wide association studies (GWAS) started to emerge in the literature around Medicine, University of Pennsylvania, Philadel- 2005, and such efforts have substantially advanced the field of diabetes genetics. phia, PA Indeed, prior to GWAS, only a miniscule number of loci were established for type 1 Corresponding author: Struan F.A. Grant, grants@ and type 2 diabetes, including the MHC, INS, PTPN22, PPARG, and KCNJ11. chop.edu The resulting data from GWAS of complex traits are empirically interpreted Received 22 March 2019 and accepted 12 June through a search for signals breaching the specific P value threshold of 5 3 1028.This 2019 bar is considered the gold standard for genome-wide significance and is derived from © 2019 by the American Diabetes Association. the required correction for a very large number of multiple comparisons due to Readers may use this article as long as the work is properly cited, the use is educational and not assaying many hundreds of thousands of single nucleotide polymorphisms (SNPs) (Fig. for profit, and the work is not altered. More infor- 1). As a consequence, it is becoming increasingly clear that diseases like type 1 and mation is available at http://www.diabetesjournals type 2 diabetes are highly polygenic, with a plethora of additional signals achieving this .org/content/license. Diabetes Care Publish Ahead of Print, published online August 13, 2019 2 TCF7L2 Locus and Pathogenesis of Diabetes Diabetes Care

Figure 1—Cartoon representing a simplified plot of a type 2 diabetes GWAS result, showing that the TCF7L2 locus achieves genome-wide significance with multiple SNPs in linkage disequilibrium.

level of significance as cohort sizes grow. subsequent larger sequencing efforts multiple Icelandic families with the GWAS have now revealed tens of loci initially did not identify additional coding disease, we followed up this region with for type 1 diabetes (2,3), while .240 loci variants (6) until reaching critical mass localized association efforts using highly have been reported for type 2 diabetes, this year with the exome sequencing polymorphic microsatellite markers. The corresponding to .400 independent as- of .20,000 case and .24,000 control associated microsatellite was rapidly found sociation signals (4). This is a truly out- subjects to reveal many more findings to be in linkage disequilibrium with a series standing feat by the genetics field, where (7). Furthermore, the TBC1D4 Arg684Ter of SNPs, any of which could have been the only less than two decades ago type 2 nonsense variant confers very high risk much sought after actual underlying causal diabetes in particular was seen as the of developing type 2 diabetes, but only variant. “geneticist’snightmare” and it was specifically among Greenlanders (8). De- It rapidly became clear from the first widely considered that finding genes spite these studies indisputably revealing reported efforts in type 2 diabetes stud- for the disease was comparable to exciting targets for pursuit, such genetic ies that the TCF7L2 locus was in fact the searching for the “holy grail.” variants are generally extremely rare strongest GWAS signal (10), with sub- After extensive international efforts and/or specific to a very narrow geo- sequent independent studies readily rep- driving increasingly larger sample sizes graphic window, making their potential licating the association in populations of to characterize more and more of the application to generalized genetic risk multiple ancestries. This finding remains genetic component to both diseases, the assessment somewhat limited currently. the common locus most strongly asso- community has delivered not only a set of Conversely, GWAS has uncovered var- ciated with the disease (4). variants contributing to the susceptibility iants that are common in the population. Researchers went on to study, through for each trait but, more importantly, the Although the risk these variants confer is association analyses, the influence of the vast majority of them are readily replicat- relatively subtle, it is very statistically TCF7L2 genetic signal on progress from able by independent research groups. The significant and much more readily de- prediabetes to diabetes, on drug response, community can therefore move forward tectable in population-based studies. and on possible association with other confidently by working with an agreed Such results observed in GWAS have traits (see below), with varying degrees “truth” that can inform subsequent char- proved useful for investigating specific of positive outcomes. Interestingly, this acterization efforts by which type 1 and traits more closely. variant has been reported to also be type 2 diabetes loci operate. The most statistically significant asso- strongly associated with cystic fibrosis– Another strategy for seeking genetic ciation signal for type 2 diabetes in non- related diabetes (11), the association ap- variants associated with common com- isolated populations is at the TCF7L2 pearing to be even stronger than that plex traits is through the utilization of locus (4). Given it has been known since observed for type 2 diabetes itself. exome and whole-genome sequencing, 2006 and that its relative strength is whichhavealsorevealedanumberofkey higher than most other GWAS-implicated CAUSAL VARIANT variants. Although much rarer, compared common loci reported to date, this locus Large-scale genotyping approaches are with individual GWAS signals these var- has proved a highly leveraged signal in excellent at interrogating the bulk of the iants are relatively impactful for those follow-up studies. The signal was first common diversity in the genome, but specific subjects harboring them. A no- reported by myself and colleagues (9) the identification of the actual underlying table discovery in recent years was the in Iceland and was subsequently repli- causal variant “tagged” by each signal is protein-truncating variants in SLC30A8 cated in additional cohorts from Den- not actually integral to the design of this (Arg138* and Lys34Serfs*50), which are mark and the U.S. Initially detected method. loss of function in nature but actu- within a linkage region for type 2 di- We refined the type 2 diabetes TCF7L2 ally confer protection (5), although abetes, i.e., a region of sharing across association in West African cohorts down care.diabetesjournals.org Grant 3

to a smaller region of the genome (12), type for incretin action, but given that cascade. As highlighted above, there concluding that from the variants tagging TCF7L2 exonic mutations and gene fu- are a number of lines of evidence that the association the T allele of intronic sions lead to colorectal cancer, there is TCF7L2 influences type 2 diabetes risk SNP rs7903146 was very likely to be the some compelling evidence that at least through the influence of GLP-1 produc- ancestral allele and served as the best one site of action is intestinal in nature. tion in the intestine (9,14). TCF7L2 knock- proxy for the underlying mutation. But the type 2 diabetes community is out mice do not live longer than 1 day Subsequent transethnic analyses and very focused on the b-cell as a primary as a specific consequence of faults in the functional efforts have further con- site, and despite very strong evidence intestinal epithelium, with nothing ob- verged on this same SNP (4). Situated for many of the other type 2 diabetes served in any other organ systems, sug- within the fourth intron of the TCF7L2 signals operating at this location, there gesting that TCF7L2 plays a specific key gene, it is the only variant that transcends is less compelling evidence for TCF7L2, role in the development of the intestinal ethnicity and captures the association which is delineated further below. This epithelium (22). with type 2 diabetes across various pop- is further supported by site-specifichu- However, a less direct correlation with ulations. It has similar allele frequencies manized mice, which have revealed that insulin secretion via the TCF7L2 locus in around the world, except for East Asia, there is indeed a role for this transcrip- another culprit cell type cannot be ruled where the risk allele is relatively rare but tion factor in non–b-cell types (15). out, given that early studies (23,24) still captures the disease association. Adipose tissue has been strongly im- suggest an influence on the insulinogenic Furthermore, the open chromatin land- plicated, with multiple splice isoforms of index, implicating this locus in actually scape across the genome, which is an TCF7L2 being implicated as playing a role operating via insulin secretion mecha- indication of the level of enhancer ac- in conferring risk for type 2 diabetes (16). nisms. Indeed, it has been shown that the tivity, has been shown to differ across Furthermore, a compelling study in mu- risk variant is associated with increased rs7903146 depending on the allele stud- rine liver strongly supported a possible TCF7L2 expression at the mRNA level ied (13). As such, this SNP has garnered hepatic role for the gene product (17). along with a decrease in insulin secretion general agreement as being the ac- With respect to therapeutic response, in the pancreatic b-cell, implicating this tual functional event at this locusdone Pearson et al. (18) showed that carriers tissue type (25). of the first identified for a complex trait. of the rs7903146 T allele present with In contrast to the lack of expression of a lower response rate to sulfonylureas TCF7L2 in murine pancreatic islets (14), it THE TCF7L2 LOCUS AND TYPE 2 compared with individuals carrying ei- has been reported that significant ex- DIABETES THERAPY ther homozygous genotype; however, pression is observed in human purified There is continued excitement that novel they did not observe association be- pancreatic b-cells (26), further support- type 2 diabetes genes offer new thera- tween metformin response and the var- ing the concept that this transcription peutic opportunities, as has been iant after adjustment for baseline HbA1c. factor is involved in the function of this seen previously with the peroxisomal Further studies are therefore warranted cell type. In addition, Schafer¨ et al. (27) proliferator–activated receptor g (PPARG) to confirm these observations. demonstrated that the TCF7L2 risk allele gene in the context of thiazolidinediones specifically impacts insulin secretion in- and with the KCNJ11 gene (encoding TCF7L2 AND FUNCTION duced by GLP-1, suggesting disruption of Kir6.2) in relation to sulfonylurea treat- Understanding the actual mechanism by GLP-1 signaling in pancreatic b-cells, as ment. However, glycemic control com- which the TCF7L2 locus confers its in- opposed to impacting secretion of GLP-1 monly involves a combination of oral fluence on type 2 diabetes risk holds itself. agents, which in turn can lead to sub- great promise to gain insights into the The work of Lyssenko et al. (25) res- stantial side effects. In addition, the disease. TCF7L2 has an intriguing con- onated with other reports of the TCF7L2 efficacy of existing drugs is regularly nection to cancer and may offer up some locus conferring its effect via pancreatic suboptimal and often ultimately requires clues on the underlying molecular mech- b-cells by showing an impact on insulin insulin supplementation. anisms driving type 2 diabetes patho- secretion. In addition to oral glucose, Effortshavegoneontoattemptto genesis. Common variants in the TCF7L2 they also showed that islet tissue har- determine where the TCF7L2 gene prod- gene have now been reported in GWAS boring the risk variant had a reduced uct, transcription factor 7-like 2, exerts its discovery efforts in the context of co- response to external stimuli, such as effect. It continues to be speculated (9) that lorectal cancer, while a multicancer sig- intravenous glucose and arginine. They it regulates the proglucagon gene (GCG) nal on chromosome 8q24 has been also observed a reduced b-cell response (14), which in turn yields a posttransla- shown to drive the expression of MYC to incretins. Interestingly, TCF7L2 gene tional product, glucagon-like peptide distally via a distant TCF7L2 regulatory expression was actually five times higher 1 (GLP-1). As this key incretin influences element (19,20). Indeed, it is already in type 2 diabetes islets, contrary to what blood glucose homeostasis (an active known that exonic mutations within one might expect, while nondiabetic area of treatment for type 2 diabetes), TCF7L2 (formerly TCF4) confer strong islets harboring the risk variant corre- one theory is that the TCF7L2 locus exerts risk for colorectal cancer, while a recur- lated with higher TCF7L2 gene expres- its influence on type 2 diabetes risk via rent VTI1A-TCF7L2 gene fusion is known sion. Furthermore, overexpression of the gut and is supported by observations to drive pathogenesis of colorectal TCF7L2 in human islets led to a decrease that incretin levels are associated with adenocarcinoma (21). in the secretion of insulin. this genetic signal. Of course, the a-cell The encoded transcription factor op- One obvious issue for functional follow- would be another key candidate cell erates at the end of the Wnt signaling up efforts is as follows: if the causal 4 TCF7L2 Locus and Pathogenesis of Diabetes Diabetes Care

SNP, rs7903146, is in an intron, it is likely especially as it remains harbored within substantial reduction in ACSL5 protein that it operates within a form of a yet the most strongly associated common expression. As such, we are working to be defined enhancer element. To that locus for type 2 diabetes. with the hypothesis that rs7903146 end, we explored what protein factor lies in a distal regulatory region for bound across this region. Our oligo pull- IMPLICATING EFFECTOR GENES AT ACSL5 and is a mechanism that at least down and subsequent mass spectropho- THE TCF7L2 LOCUS partially influences type 2 diabetes risk at tometry approach suggested that PARP-1 The knockout mouse model of the this locus (35). However, although a re- interacted with this putative element GWAS-implicated obesity-associated cent international expression quantita- (28). Despite the fact that we know FTO gene (33) resulted in a clear BMI- tive trait loci (eQTL) effort in pancreatic PARP-1 is a known promiscuous factor related phenotype, despite the fact that islets did support TCF7L2 as an effector that binds to fragmented DNA and fea- the originally uncovered variant does not gene at this locus, they did not see any tures in the “CRAPome” (29), we were account for a large proportion of the evidence for ACSL5 in that particular intrigued by the finding. This was par- predicted genetic component for obe- tissue type (37). ticularly the case given that PARP-1 is a sity. So despite the magnitude of risk To add further complexity, there are cancer therapeutic target and the fact conferredbyagivenlocus, it does appear predicted to be multiple splice isoforms that type 2 diabetes and cancer genetics that GWAS presents us with novel in- of the gene transcript, so it is likely that appear to show a degree of overlap; in sights into the biology of complex traits this feature is confounding functional addition, it is known that PARP-1 acts as and thus opens up novel therapeutic follow-up analyses of TCF7L2. Despite an intermediate for driving diabetic mi- opportunities. However, the narrative on there being less isoforms in endocrine- crovascular complications. By bathing that association got more complicated relevant cell types when compared with Caenorhabditis elegans in glucose, which when the GWAS signal, harbored within cancerous ones, there is still reason to is known to shorten the life span of this an FTO intron, was studied more care- believe this could cloud possible associ- organism, along with a PARP-1 inhibitor fully. Specifically, the signal is now ations with gene expression. For in- we observed a rescue of this feature, known to influence the expression of stance, given it has been relatively which in turn was ablated when we nearby genes, IRX3 and IRX5, rather than challenging to observe significant eQTL knocked down the expression of the principally the “host” gene itself (34). These between this SNP and TCF7L2 gene ex- TCF7L2 homolog (30). Despite our in- discoveries suggest that the associated pression up until recently, this splicing triguing results, there is a lack of con- variants can be residing in one gene but feature may be one of the contributing sensus on what factors bind across this actually influence the expression of other factors. region, with FOXA2 and HMGB1 both putatively causal effector genes some Despite the ongoing efforts to char- being suggested to play a role. This still distance away. acterize the precise functional behavior remains to be fully resolved. Inspired by this FTO work, and given of this locus, there is one undisputed fact: In addition to exploring what binds to that the T allele of rs7903146 at the the type 2 diabetes genetic signal at the this locus, TCF7L2 is itself a transcription TCF7L2 locus is also within an intronic TCF7L2 locus is beyond doubt and can be factor, so there is also interest in its binding region and has been widely implicated as leveraged in other types of studies. In- pattern across the genome. Characterizing the causal variant, we similarly elected deed, with larger and larger GWAS being its occupancy would provide insight into to implicate genes whose expression is published (4), there are now more and its downstream targets, which in turn influenced by the putative noncoding more independent signals, albeit sub- could present more therapeutically at- enhancer element coinciding with this stantially weaker, being observed at tractive avenues. However, binding pre- key variant. To that end, we conducted a this locus, highlighting how important dictions are hampered by the fact that chromatin conformation capture cam- this region is in conferring susceptibility the DNA binding motif for TCF7L2 is very paign combined with sequencing in to type 2 diabetes and potentially other degenerate. It is therefore clear that two carcinoma cell lines derived from diseases. functional studies, such as with chroma- human colon (motivated by the known tin immunoprecipitation (ChIP), should biology of TCF7L2 with respect to the INDEX EVENT BIAS be leveraged to physically characterize intestine), along with expression analy- One possible mechanism by which the the portfolio of genes transcriptionally ses (35). In addition to observed phys- TCF7L2 locus confers its type 2 diabetes regulated by TCF7L2. Indeed, TCF7L2 has ical contacts and correlation with TCF7L2 susceptibility effect is by influencing been recognized as a master regulator gene expression, we also observed a BMI and therefore insulin resistance. (31), and when we employed ChIP-seq to strong relationship with the ACSL5 gene, Interestingly, a very large GWAS of adult investigate its genome-wide occupancy which encodes acyl-CoA synthetase long BMI implicated the TCF7L2 locus in influ- (32), we found that the genes it bound chain family, member 59, and is known encing risk for higher BMI (38). However, were indeed in pathways related to di- to influence fatty acid metabolism. what was intriguing was that it was the abetes and comorbidities; most inter- Indeed, a recent study reported that nonrisk C allele of rs7903146 that con- estingly, these genes were enriched at the ACSL5 knockout mouse presented ferred higher BMI, which seemed at odds GWAS loci, highlighting that the binding with increased insulin sensitivity (36). with the T allele conferring risk for type 2 of TCF7L2 was far from random. Subsequent genome editing in the im- diabetes. The authors of the study did Overall, the therapeutic opportunities mediate neighborhood around the pre- outline the fact that there was a relatively that the TCF7L2 gene product poten- sumed enhancer coinciding with the high degree of heterogeneity between tially presents have still to be resolved, location of rs7903146 led to a very cohorts, so the observation could have care.diabetesjournals.org Grant 5

been due to issues related to differential diabetes loci, namely, at the MHC and to treatment for what is increasingly accounting for comorbidities. Further- INS (43). looking like a spectrum of diabetes sub- more, sometime later it was shown The roles of the type 1 diabetes MHC types. The TCF7L2 locus remains a potent that the TCF7L2 locus was an example locus and the type 2 diabetes TCF7L2 genetic factor to inform such studies of “index event bias” in the context of locus have suggested that LADA sits at going forward. obesity risk (39), where stratification can the “genetic meeting point” of these two lead to a false positive association with traits (44,45). Indeed, one study sug- SUMMARY BMI. As such, there is still debate as to gested that the TCF7L2 locus could be Although TCF7L2 is by far not the only whether the association with obesity is used in young patients to distinguish locus associated with type 2 diabetes, it a true phenomenon. autoimmune from nonautoimmune does represent one of the first and And the issue of confounding does causes of diabetes; however, it was strongest genetic signals for this trait. not stop there; understanding the inter- not applicable to older patients (46). As such, it has proved useful for some play between the various forms of di- Such observations of course do not time in exploration of the role of type 2 abetes is extremely important in such a exclude the possibility that LADA just diabetes genetics in various settings, context. represents a mixture of poorly defined including revealing clues to a common type 1 diabetes cases and type 2 diabetes genetic etiology with cancer, revealing a TCF7L2, LATENT AUTOIMMUNE cases, but given that the TCF7L2 locus has role in the presentation of type 1 di- DIABETES IN ADULTS, AND TYPE 1 been reported to be associated with LADA abetes and LADA, and providing lessons DIABETES in a number of cohorts, these findings do on how to best functionally follow up a It has been widely thought that type 1 suggest an intriguing role for type 2 di- GWAS locus. Given the observations and type 2 diabetes have distinct un- abetes genetics in a disease with many made with this locus, it bodes well derlying disease mechanisms with very features of type 1 diabetes pathogenesis. when one comparably leverages the little evidence for genetic overlaps be- As with LADA, type 1 diabetes cases other established type 2 diabetes loci tween the two diseases. This is what has present with varying repertoires of pos- either individually or combined in the been seen generally with the TCF7L2 itivity for the four to five autoantibodies form of genetic risk scores. locus, where it does not obviously play that are hallmarks of the trait. Maria an overall role in the pathogenesis of Redondo and colleagues have spent juvenile-onset type 1 diabetes (40,41). much time leveraging the TCF7L2 locus Funding. S.F.A.G. is supported by the National However, we and others have ob- to explore a possible role in the patho- Institutes of Health (R01 DK085212) and the served that the TCF7L2 locusisassociated genesis of type 1 diabetes, primarily as a Daniel B. Burke Endowed Chair for Diabetes with latent autoimmune diabetes in first pass, given it is the strongest com- Research. fl adults (LADA). This relatively common mon type 2 diabetes locus and could Duality of Interest. No potential con icts of interest relevant to this article were reported. disease, that progresses to insulin de- guide the community on exploring fur- pendency more slowly than type 1 di- ther with other such loci. Intriguingly, References abetes, is considered by the World Health although the TCF7L2 locus is not associ- 1. Rich SS. Mapping genes in diabetes. Genetic Organization as a distinct diabetes sub- ated with type 1 diabetes overall, it does epidemiological perspective. Diabetes 1990;39: group (42). There is very little consensus show a role when one focuses on type 1315–1319 on the precise definition of the disease, 1 diabetes cases only positive for one 2. Onengut-Gumuscu S, Chen WM, Burren O, et al.; Type 1 Diabetes Genetics Consortium. Fine with disagreement even existing on the autoantibody (47). In this setting, the mapping of type 1 diabetes susceptibility loci and typical age of onset, which ranges from association is quite apparent and reso- evidence for colocalization of causal variants with 25 to 40 years old. Overall, the disease nates with the LADA observations, with lymphoid gene enhancers. Nat Genet 2015;47: often masquerades as type 2 diabetes, as this particular subgroup residing in a gray 381–386 these adults have no need for insulin zone between the two diseases. Further- 3. Barrett JC, Clayton DG, Concannon P, et al.; Type 1 Diabetes Genetics Consortium. Genome- treatment for a minimum of 6 months more, the same research group went on wide association study and meta-analysis find following diagnosis of diabetes. As such, to show that autoantibody cases prog- that over 40 loci affect risk of type 1 diabetes. Nat LADA presents with clinical features of ress to full type 1 diabetes presentation Genet 2009;41:703–707 both type 1 diabetes and type 2 diabe- less rapidly when the type 2 diabetes 4. Mahajan A, Taliun D, Thurner M, et al. Fine- tes, thus being often termed “type 1.5 TCF7L2 risk allele is carried as well (48). mapping type 2 diabetes loci to single-variant ” resolution using high-density imputation and diabetes. So, this presents us with an intriguing islet-specific epigenome maps. Nat Genet 2018; Such cases within a given apparent question: does the genetics of type 2 50:1505–1513 type 2 diabetes cohort could lead to diabetes help “push” individuals over 5. Flannick J, Thorleifsson G, Beer NL, et al.; Go- spurious genetic associations, especially into diabetes presentation if they do T2D Consortium; T2D-GENES Consortium. Loss- when one is pursuing increasingly mod- not harbor the full repertoire of of-function mutations in SLC30A8 protect against type 2 diabetes. Nat Genet 2014;46:357–363 est signals in larger and larger cohorts. type 1 diabetes genetic risk factors, 6. Fuchsberger C, Flannick J, Teslovich TM, et al. Indeed, in relatively recent GWAS efforts, i.e., does the former act as a catalyst The genetic architecture of type 2 diabetes. type 2 diabetes loci involved in autoim- to the latter? In short, that question still Nature 2016;536:41–47 mune traits are beginning to emerge; for remains to be answered, and further 7. Flannick J, Mercader JM, Fuchsberger C, et al.; Broad Genomics Platform; DiscovEHR Collabo- instance, a GWAS of type 2 diabetes in follow-up is certainly warranted. These ration; CHARGE; LuCamp; ProDiGY; GoT2D; ESP; African Americans revealed two novel observations could have fundamental SIGMA-T2D; T2D-GENES; AMP-T2D-GENES. signals that are both classic type 1 implications for personalized approaches Exome sequencing of 20,791 cases of type 2 6 TCF7L2 Locus and Pathogenesis of Diabetes Diabetes Care

diabetes and 24,440 controls. Nature 2019;570: TCF7L2 polymorphisms and progression to di- 2 diabetes [article online], 2019. Available from 71–76 abetes in the Diabetes Prevention Program. N https://www.biorxiv.org/content/10.1101/ 8. Moltke I, Grarup N, Jørgensen ME, et al. A Engl J Med 2006;355:241–250 655670v1. Accessed 9 June 2019 common Greenlandic TBC1D4 variant confers 24. Saxena R, Gianniny L, Burtt NP, et al. Com- 38. Locke AE, Kahali B, Berndt SI, et al.; LifeLines muscle insulin resistance and type 2 diabetes. mon single nucleotide polymorphisms in TCF7L2 Cohort Study; ADIPOGen Consortium; AGEN-BMI Nature 2014;512:190–193 are reproducibly associated with type 2 diabetes Working Group; CARDIOGRAMplusC4D Consor- 9. Grant SF, Thorleifsson G, Reynisdottir I, et al. and reduce the insulin response to glucose in tium; CKDGen Consortium; GLGC; ICBP; MAGIC Variant of transcription factor 7-like 2 (TCF7L2) nondiabetic individuals. Diabetes 2006;55:2890– Investigators; MuTHER Consortium; MIGen Con- gene confers risk of type 2 diabetes. Nat Genet 2895 sortium; PAGE Consortium; ReproGen Consor- 2006;38:320–323 25. Lyssenko V, Lupi R, Marchetti P, et al. Mech- tium;GENIEConsortium;InternationalEndogene 10. Sladek R, Rocheleau G, Rung J, et al. A genome- anisms by which common variants in the TCF7L2 Consortium. Genetic studies of body mass index wide association study identifies novel risk loci for gene increase risk of type 2 diabetes. J Clin Invest yield new insights for obesity biology. Nature type 2 diabetes. Nature 2007;445:881–885 2007;117:2155–2163 2015;518:197–206 11. Blackman SM, Hsu S, Ritter SE, et al. A 26. CauchiS,MeyreD,DinaC,etal.Transcription 39. Yaghootkar H, Bancks MP, Jones SE, et al. susceptibility gene for type 2 diabetes confers factor TCF7L2 genetic study in the French pop- Quantifying the extent to which index event substantial risk for diabetes complicating cystic ulation: expression in human b-cells and adipose biases influence large genetic association stud- fibrosis. Diabetologia 2009;52:1858–1865 tissue and strong association with type 2 di- ies. Hum Mol Genet 2017;26:1018–1030 12. Helgason A, Palsson´ S, Thorleifsson G, et al. abetes. Diabetes 2006;55:2903–2908 40. Qu HQ, Grant SF, Bradfield JP, et al. Asso- Refining the impact of TCF7L2 gene variants on 27. Schafer¨ SA, Tschritter O, Machicao F, et al. ciation analysis of type 2 diabetes loci in type 1 type 2 diabetes and adaptive evolution. Nat Impaired glucagon-like peptide-1-induced insu- diabetes. Diabetes 2008;57:1983–1986 Genet 2007;39:218–225 lin secretion in carriers of transcription factor 41. Field SF, Howson JM, Smyth DJ, Walker NM, 13. Gaulton KJ, Nammo T, Pasquali L, et al. A map 7-like2 (TCF7L2) gene polymorphisms [published Dunger DB, Todd JA. Analysis of the type 2 of open chromatin in human pancreatic islets. correction appears in Diabetologia 2009;52:557; diabetes gene, TCF7L2, in 13,795 type 1 diabetes Nat Genet 2010;42:255–259 Diabetologia 2008;51:208]. Diabetologia 2007; cases and control subjects. Diabetologia 2007; 14. Yi F, Brubaker PL, Jin T. TCF-4 mediates cell 50:2443–2450 50:212–213 type-specific regulation of proglucagon gene 28. Xia Q, Deliard S, Yuan CX, Johnson ME, Grant 42. Alberti KG, Zimmet PZ. Definition, diagnosis expression by beta-catenin and glycogen syn- SF. Characterization of the transcriptional ma- and classification of diabetes mellitus and its thase kinase-3beta. J Biol Chem 2005;280:1457– chinery bound across the widely presumed complications. Part 1: diagnosis and classifica- 1464 type 2 diabetes causal variant, rs7903146, within tion of diabetes mellitus. Provisional report of a 15. Bailey KA, Savic D, Zielinski M, et al. Evidence TCF7L2. Eur J Hum Genet 2015;23:103–109 WHO Consultation. Diabet Med 1998;15:539– of non-pancreatic beta cell-dependent roles of 29. Mellacheruvu D, Wright Z, Couzens AL, et al. 553 Tcf7l2 in the regulation of glucose metabolism The CRAPome: a contaminant repository for 43. Ng MC, Shriner D, Chen BH, et al.; FIND in mice. Hum Mol Genet 2015;24:1646–1654 affinity purification-mass spectrometry data. Consortium; eMERGE Consortium; DIAGRAM 16. Kaminska D, Kuulasmaa T, Venesmaa S, et al. Nat Methods 2013;10:730–736 Consortium; MuTHER Consortium; MEta-analysis Adipose tissue TCF7L2 splicing is regulated 30. Xia Q, Lu S, Ostrovsky J, McCormack SE, Falk of type 2 DIabetes in African Americans Consor- by weight loss and associates with glucose MJ, Grant SFA. PARP-1 inhibition rescues short tium. Meta-analysis of genome-wide association and fatty acid metabolism. Diabetes 2012;61: lifespan in hyperglycemic C. Elegans and im- studies in African Americans provides insights into 2807–2813 proves GLP-1 secretion in human cells. Aging Dis the genetic architecture of type 2 diabetes. PLoS 17. Boj SF, van Es JH, Huch M, et al. Diabetes risk 2018;9:17–30 Genet 2014;10:e1004517 gene and Wnt effector Tcf7l2/TCF4 controls 31. Zhou Y, Park SY, Su J, et al. TCF7L2 is a master 44. Cousminer DL, Ahlqvist E, Mishra R, et al.; hepatic response to perinatal and adult meta- regulator of insulin production and processing. Bone Mineral Density in Childhood Study. First bolic demand. Cell 2012;151:1595–1607 Hum Mol Genet 2014;23:6419–6431 genome-wide association study of latent auto- 18. Pearson ER, Donnelly LA, Kimber C, et al. 32. Zhao J, Schug J, Li M, Kaestner KH, Grant SF. immune diabetes in adults reveals novel insights Variation in TCF7L2 influences therapeutic re- Disease-associated loci are significantly over- linking immune and metabolic diabetes. Diabe- sponse to sulfonylureas: a GoDARTs study. Di- represented among genes bound by transcrip- tes Care 2018;41:2396–2403 abetes 2007;56:2178–2182 tion factor 7-like 2 (TCF7L2) in vivo. Diabetologia 45. Cervin C, Lyssenko V, Bakhtadze E, et al. 19. Pomerantz MM, Ahmadiyeh N, Jia L, et al. 2010;53:2340–2346 Genetic similarities between latent autoimmune The 8q24 cancer risk variant rs6983267 shows 33. Fischer J, Koch L, Emmerling C, et al. In- diabetes in adults, type 1 diabetes, and type 2 long-range interaction with MYC in colorectal activation of the Fto gene protects from obesity. diabetes. Diabetes 2008;57:1433–1437 cancer. Nat Genet 2009;41:882–884 Nature 2009;458:894–898 46. Bakhtadze E, Cervin C, Lindholm E, et al. 20. Tuupanen S, Turunen M, Lehtonen R, et al. 34. Claussnitzer M, Dankel SN, Kim KH, et al. FTO Common variants in the TCF7L2 gene help to The common colorectal cancer predisposition obesity variant circuitry and adipocyte browning differentiate autoimmune from non-autoimmune SNP rs6983267 at chromosome 8q24 confers in humans. N Engl J Med 2015;373:895–907 diabetes in young (15-34 years) but not in middle- potential to enhanced Wnt signaling. Nat Genet 35. Xia Q, Chesi A, Manduchi E, et al. The type 2 aged (40-59 years) diabetic patients. Diabetologia 2009;41:885–890 diabetes presumed causal variant within TCF7L2 2008;51:2224–2232 21. Bass AJ, Lawrence MS, Ramos AH, et al. resides in an element that controls the expres- 47. Redondo MJ, Geyer S, Steck AK, et al.; Type 1 Genomic sequencing of colorectal adenocarci- sion of ACSL5. Diabetologia 2016;59:2360–2368 Diabetes TrialNet Study Group. TCF7L2 genetic nomas identifies a recurrent VTI1A-TCF7L2 fu- 36. Bowman TA, O’Keeffe KR, D’Aquila T, et al. variants contribute to phenotypic heterogeneity sion. Nat Genet 2011;43:964–968 Acyl CoA synthetase 5 (ACSL5) ablation in mice of type 1 diabetes. Diabetes Care 2018;41:311–317 22. Korinek V, Barker N, Moerer P, et al. De- increases energy expenditure and insulin sensi- 48. Redondo MJ, Steck AK, Sosenko J, et al.; Type pletion of epithelial stem-cell compartments in tivity and delays fat absorption. Mol Metab 1 Diabetes TrialNet Study Group. Transcription the small intestine of mice lacking Tcf-4. Nat 2016;5:210–220 factor 7-like 2 (TCF7L2) gene polymorphism and Genet 1998;19:379–383 37. Vinuela~ A, Varshney A, van de Bunt M, et al. progression from single to multiple autoantibody 23. Florez JC, Jablonski KA, Bayley N, et al.; Influence of genetic variants on gene expression positivity in individuals at risk for type 1 diabetes. Diabetes Prevention Program Research Group. in human pancreatic islet – implications for type Diabetes Care 2018;41:2480–2486