Scientists at the University of Cambridge have identified rare genetic variants – carried by one in 3,000 people – that have a larger impact on the risk of developing type 2 diabetes than any previously identified genetic effect.
Type 2 diabetes is thought to be driven in part by inherited genetic factors, but many of these genes are yet unknown. Previous large-scale studies have depended on efficient ‘array genotyping’ methods to measure genetic variations across the whole genome.
This approach typically does a good job at capturing the common genetic differences between people, though individually these each confer only small increases in diabetes risk.
Recent technical advances have allowed more comprehensive genetic measurement by reading the complete DNA sequences of over 20,000 genes that code for proteins in humans. Proteins are essential molecules that enable our bodies to function. In particular, this new approach has allowed for the first time a large-scale approach to study the impact of rare genetic variants on several diseases, including type 2 diabetes.
By looking at data from more than 200,000 adults in the UK Biobank study, researchers from the Medical Research Council (MRC) Epidemiology Unit at the University of Cambridge used this approach to identify genetic variants associated with the loss of the Y chromosome.
This is a known biomarker of biological aging that occurs in a small proportion of circulating white blood cells in men and indicates a weakening in the body’s cellular repair systems. This biomarker has been previously linked to age-related diseases such as type 2 diabetes and cancer.
In results published today in Nature Communications, the researchers identified rare variants in the gene GIGYF1 that substantially increase susceptibility to loss of the Y chromosome, and also increase an individual’s risk of developing type 2 diabetes six-fold. In contrast, common variants associated with type 2 diabetes confer much more modest increases in risk, typically much lower than two-fold.
Around 1 in 3,000 individuals carries such a GIGYF1 genetic variant. Their risk of developing type 2 diabetes is around 30%, compared to around 5% in the wider population. In addition, people who carried these variants had other signs of more widespread aging, including weaker muscle strength and more body fat.
GIGYF1 is thought to control insulin and cell growth factor signaling. The researchers say their findings identify this as a potential target for future studies to understand the common links between metabolic and cellular aging, and to inform future treatments.
Dr. John Perry, from the MRC Epidemiology Unit and a senior author on the paper, said: “Reading an individual’s DNA is a powerful way of identifying genetic variants that increase our risk of developing certain diseases. For complex diseases such as type 2 diabetes, many variants play a role, but often only increasing our risk by a tiny amount. This particular variant, while rare, has a big impact on an individual’s risk.”
Professor Nick Wareham, Director of the MRC Epidemiology Unit, added: “Our findings highlight the exciting scientific potential of sequencing the genomes of very large numbers of people. We are confident that this approach will bring a rich new era of informative genetic discoveries that will help us better understand common diseases such as type 2 diabetes.
By doing this, we can potentially offer better ways to treat – or even to prevent – the condition.”
Ongoing research will aim to understand how the loss of function variants in GIGYF1 lead to such a substantial increase in the risk of developing type 2 diabetes. Their future research will also examine other links between biomarkers of biological aging in adults and metabolic disorders.
Sequencing of large cohorts offers an unprecedented opportunity to identify rare genetic variants and to find novel contributors to human disease. We used gene-based collapsing tests to identify genes associated with glucose, HbA1c and T2D diagnosis in 363,977 exome-sequenced participants in the UK Biobank. We identified known associations with diabetes including variants in GCK, HNF1A and PDX1, genes involved in Mendelian forms of diabetes.
Novel associations were identified for GIGYF1 predicted loss of function (pLOF), TNRC6B pLOF and PFAS predicted damaging missense variants. Multiple rare variants contributed to these associations, including singleton variants. The most significant novel associations were seen for GIGYF1 pLOF which associated with increased levels of glucose (0.77 mmol/L increase, p = 4.42 × 10−12) and HbA1c (4.33 mmol/mol, p = 1.28 × 10−14) as well as T2D diagnosis (OR = 4.15, p= 6.14 x10−11).
GIGYF1 pLOF also associated with decreased cholesterol levels as well as an increased risk of hypothyroidism. An independent common variant association for glucose and T2D was identified at GIGYF1 which replicated in additional datasets. Our results highlight the role of GIGYF1 in regulating insulin signaling and protecting from diabetes.
Discussion
Our results highlight the power of whole exome sequencing to make novel discoveries relevant to human disease and to detect known associations of Mendelian disease genes. Gene-level aggregation and burden testing of rare pLOF and predicted damaging missense variants identified genes associating with diabetes and biomarkers of glycemic control. These included several genes not previously implicated in diabetes, GIGYF1, TNRC6B and PFAS, as well as GCK, HNF1A and PDX1, known MODY genes [14, 22-24].
We also identified PAM and G6PC2, genes identified by other rare-variant studies of T2D and glucose levels [5, 15]. Gene-level tests were needed to detect the majority of these associations owing to the rarity of damaging variants. For example, out of 363,977 individuals, just 40 carried a pLOF variant in GCK and 131 carried a pLOF variant in GIGYF1. In general, singleton variants contributed a large part of the signal arguing strongly, as others have done [4], for including such variants in gene-based collapsing tests.
Test statistic inflation can be a challenge when testing rare variants as statistical assumptions break down when the number of carriers expected to have the disease of interest is low [4, 25]. To avoid false positives in our analysis of diabetes, we initially examined associations with glucose and HbA1c because quantitative traits are less susceptible to inflation.
All of the variant sets that associated with T2D also affected HbA1c and/or glucose levels giving us confidence in these associations. In addition, T2D associations for all genes, apart from TNRC6B, were significant (p ≤ 1.46 × 10−6) using the linear mixed model implemented by SAIGE-Gene which can be more robust when dealing with low numbers of variant carriers [16]. Additional confidence in our results comes from the fact that we identified genes known to be involved in Mendelian forms of diabetes and previously reported genes.
We uncovered novel associations with T2D and biomarkers of glycemic control for aggregated variants in GIGYF1, TNRC6B and PFAS. TNRC6B encodes trinucleotide repeat-containing gene 6B protein which is involved siRNA- and miRNA-mediated gene silencing [26-28]. As TNRC6B is highly constrained, one must view pLOF variants in this gene with suspicion as they may be sequencing errors or not result in a true loss of function. Additional data will be needed to verify the association of TNRC6B pLOF with T2D. Interestingly, heterozygous protein-truncating variants in TNRC6B have been implicated in developmental delay and autism spectrum disorders [29].
Consistent with this, we see an association of TNRC6B pLOF with worse performance in a cognitive test as well as with hearing impairment. Damaging missense variants in PFAS associate with decreased HbA1c levels and decreased incidence of T2D diagnosis with a large part of the signal driven by two missense variants Glu434Lys and Ala466Thr. PFAS encodes phosphoribosylformylglycinamidine synthase which is involved in de novo synthesis of the purine inosine monophosphate (IMP), a process which is required by growing and proliferating cells [30]. A mechanistic link to glucose homeostasis or diabetes is not apparent but PFAS may have roles beyond purine biosynthesis as it interacts with and deamidates other proteins [31]. PFAS is ubiquitously expressed [19, 32].
We focused our analysis on understanding the consequences of GIGYF1 pLOF as it strongly associated with glucose, HbA1c and T2D. GIGYF1 encodes a protein that was initially identified for its binding to the adapter protein GRB10 which negatively regulates both the insulin and IGF-1 receptors [33]. Transfection of cells with GRB10-binding fragments of GIGYF1 lead to greater activation of both the insulin and IGF-1 receptors [34].
This supports a hypothesis whereby GIGYF1 enhances insulin signaling by reducing the negative regulation of the insulin receptor by GRB10. When GIGYF1 is reduced, as is the case in individuals carrying pLOF variants, GRB10 presumably inhibits insulin signaling to a greater degree thereby reducing the action of insulin in its target tissues and leading to increased risk of T2D. However, the exact mechanistic details of these interactions remain to be determined.
GRB10 variants have also been reported to associate with glycemic traits although interpretation of these results is complicated by imprinting [35]. GIGYF1 is broadly expressed with high levels observed in endocrine tissues, pancreas and brain [19, 32]. GIGYF1 and the related protein GIGYF2 have also been implicated in translational repression [36] and translation-coupled mRNA decay [37] suggesting biological roles beyond regulation of insulin and IGF-1 receptor signaling.
PheWAS of GIGYF1 pLOF revealed a strong association with decreased cholesterol levels reflecting altered energy homeostasis in carriers. An inverse relationship between glucose and cholesterol levels has been observed for variants in other genes [38]. We also observed several associations that could reflect complications of diabetes in GIGYF1 pLOF carriers including increased cystatin c levels and increased diagnosis of urinary disorders suggesting renal complications as well as syncope and collapse which may be a side-effect of hyperglycemia and/or hypoglycemia in diabetics.
Other associations may reflect poor health in carriers including decreased grip strength and decreased peak expiratory flow. GIGYF1 pLOF also associated with decreased mean corpuscular hemoglobin levels and increased diagnosis of anemia as well as increased emphysema diagnosis. The biological basis for these associations is not clear. GIGYF1 is highly expressed in lung [19, 32] although the emphysema association is driven by small numbers of individuals, so replication is required.
GIGYF1 pLOF associated with a 4.5-fold increased risk of hypothyroidism and GIGYF1 is highly expressed in thyroid [19, 32] consistent with a biological function in this tissue. IGF-1 and insulin have been implicated in the proliferation of thyroid cells which may, in part, explain the association with thyroid dysfunction [39-41].
An alternative explanation is that GIGYF1 contributes to thyroid function by affecting secretion of thyroid stimulating hormone in the anterior pituitary gland. Damaging variants in GIGYF1 have recently been implicated in conferring risk for developmental delay and autism spectrum disorders [42]. Consistent with this, we see an association of GIGYF1 pLOF with increased time to complete a cognitive test. It may be that metabolic aberrations in carriers affect cognitive performance, that brain development is altered due to perturbation of IGF-1 signaling, or that other functions of GIGYF1 such as regulation of mRNA expression and decay are responsible for cognitive phenotypes.
Replication of associations is a challenge in rare variant analysis. The UKBB dataset used consists of exomes and baseline biomarker measurements for 363,977 individuals which is, to our knowledge, the largest dataset of its kind available. We replicated the majority of our associations with glucose and HbA1c levels, including those for GIGYF1 pLOF, using independent measurements from primary care data.
Additionally, we turned to common genetic variants to further investigate the role of the GIGYF1 locus in diabetes. Non-coding variants at the GIGYF1 locus associated with glucose levels and T2D, and we replicated these findings in independent datasets. These variants associated with increased GIGYF1 expression but a lower risk of T2D. This direction of effect is consistent with what we see for the pLOF variants – reduced levels of GIGYF1 increases diabetes risk but increased levels of GIGYF1 are protective.
We assessed the impact of pLOF and damaging missense variants in approximately 17,000 genes on glycemic traits and uncovered a hitherto unappreciated role for GIGYF1 in regulating blood sugar and protecting from T2D. By highlighting the importance of GIGYF1 and GRB adapter proteins in modulating insulin signaling this finding may lead to new therapeutic approaches for the treatment of diabetes. Discoveries such as this are only possible by combining health-related data with the sequencing of rare variants on a biobank scale.
reference link : https://www.medrxiv.org/content/10.1101/2021.01.19.21250105v1.full
More information: Zhao, Y. et al. GIGYF1 loss of function is associated with clonal mosaicism and adverse metabolic health. Nature Communications 2021; 07 Jul 2021; DOI: 10.1038/s41467-021-24504-y