Individual risk of type 2 diabetes (T2D) is modified by perturbations to the mass, distribution and function of adipose tissue. To investigate the mechanisms underlying these associations, we explored the molecular, cellular and whole-body effects of T2D-associated alleles near KLF14. We show that KLF14 diabetes-risk alleles act in adipose tissue to reduce KLF14 expression and modulate, in trans, the expression of 385 genes. We demonstrate, in human cellular studies, that reduced KLF14 expression increases pre-adipocyte proliferation but disrupts lipogenesis, and in mice, that adipose tissue–specific deletion of Klf14 partially recapitulates the human phenotype of insulin resistance, dyslipidemia and T2D. We show that carriers of the KLF14 T2D risk allele shift body fat from gynoid stores to abdominal stores and display a marked increase in adipocyte cell size, and that these effects on fat distribution, and the T2D association, are female specific. The metabolic risk associated with variation at this imprinted locus depends on the sex both of the subject and of the parent from whom the risk allele derives.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This study was supported by MRC grant MR/J010642/1 to K.S.S., R.D.C., F.K. and M.I.M. K.S.S. is supported by an MRC New Investigator Award (MR/L01999X/1). M.I.M. is a Wellcome Senior Investigator and is supported by Wellcome (090532, 106130, 098381, 203141), NIDDK (U01-DK105535) and the MRC (MR/L020149/1). F.K. is supported by the British Heart Foundation (RG/17/1/32663). M.C. is supported by NIH Award R00 HL121172. A.L.G. is a Wellcome Senior Fellow in Basic Biomedical Science (095101/Z/10/Z and 200837/Z/16/Z). K.M. is supported by NIH award R01-DK099571, and A.J.L. is supported by NIH award NIH P01HL28481. R.D.C. is supported by MRC MC_U142661184. J.F. was a Marie Curie Fellow. M.L. is supported by Academy of Finland grants 77299 and 124243, the Finnish Heart Foundation, the Finnish Diabetes Foundation, and Commission of the European Community HEALTH-F2-2007-201681. A.V. and A.B. were supported by the European Union Framework Programme 7 grant EuroBATS (259749). Some computations were performed at the Vital-IT center for high-performance computing of the Swiss Institute of Bioinformatics (SIB; http://www.vital-it.ch/). The TwinsUK study was funded by Wellcome and European Community's Seventh Framework Programme (FP7/2007-2013). The TwinsUK study also receives support from the National Institute for Health Research (NIHR)-funded BioResource, Clinical Research Facility and Biomedical Research Centre based at Guy's and St Thomas’ NHS Foundation Trust in partnership with King's College London. We acknowledge excellent technical support for animal husbandry (Mary Lyon Centre), genotyping, histology and pathology. We thank the volunteers from the Oxford Biobank, NIHR Oxford Biomedical Research Centre, for their participation. The Oxford Biobank (http://www.oxfordbiobank.org.uk) is also part of the NIHR National Bioresource, which supported the recalling process of the volunteers. GERA data came from a grant, the Resource for Genetic Epidemiology Research in Adult Health and Aging (RC2 AG033067; C. Schaefer and N. Risch, principal investigators), awarded to the Kaiser Permanente Research Program on Genes, Environment, and Health (RPGEH) and the UCSF Institute for Human Genetics. The RPGEH was supported by grants from the Robert Wood Johnson Foundation, the Wayne and Gladys Valley Foundation, and the Ellison Medical Foundation. This research utilized data from the UK Biobank Resource under application 9161.
Integrated supplementary information
QQ-plots of association between rs473102 and all adipose probes in the deCODE dataset (female N=376 and male N=265). Top row displays results from standard genotypic tests, bottom row displays results for test of association to maternally inherited allele
The ancestral T2D risk allele C is colored in orange, the derived allele T in blue. Figure generated from the HGDP Selection Browser at the Pritchard Lab (http://hgdp.uchicago.edu/cgi-bin/gbrowse/HGDP/). rs4731702 does not display evidence of positive selection in the HGDP – with global FST in the 80th percentile genome-wide and non-significant iHS and XP-EHH scores in all populations71. Tests for recent positive selection at rs4731702 in samples from the United Kingdom using the Singleton Density Score (SDS) were not significant (P=0.20), but the trend was towards a recent increase of the non-risk allele T72. By contrast, the KLF14 transcript does exhibit evidence of intolerance to variation in the ExAC exome aggregation dataset73 (ExAC constraint scores: Missense z= 2.32; Synonymous z = 2.41) suggesting coding changes in KLF14 are under purifying selection
Supplementary Figure 3 Distinct 450K methylation probes at the KLF14 locus are associated to rs4731702 vs age.
Figure shows results for two 450K probes assayed in the TwinsUK adipose samples (N=603). cg02385110, which is ~3KB upstream of KLF14 is shown on the left, cg08097417 which is at the KLF14 transcription start site is shown on the right. The beta distributions (top row), association to rs4731702 (T2D risk allele homozygotes = 2) (middle row) and association to age (bottom row) differ between the two probes. cg02385110 has a higher mean beta value, is associated to rs4731702 (P=2.1x10-7), and is not associated to age (P=0.64). In contrast, cg08097417 is not associated to rs4731702 (P=0.99) but is highly associated to age (P=3.6x10-61). Associations were tested with linear mixed effect models adjusting for batch effects, BMI and famlly structure
Supplementary Figure 4 Clinical chemistry analysis of heterozygous Klf14tm1(KOMP)Vlcg knockout mice.
Clinical chemistry parameters were measured in male knockout C57BL/6N Klf14 mice and their wildtype controls. Since Klf14 is mono-allelically maternally expressed in mouse and human, we compared heterozygous mice that had inherited the deletion allele from their mother (heterozygous-MAT, expressing the deletion) with heterozygous mice that had inherited the deletion allele from their father (heterozygous-PAT, not expressing the deletion). We also compared the two groups to their own homozygous wildtype colonymate controls (wildtype-MAT and wildtype-PAT). This was because we used two separate crosses to produce the MAT and PAT carrier cohorts; one stock from heterozygous mothers and one from heterozygous fathers, each crossed to wildytpe C57BL/6N mice. Mice were fed a standard diet and then switched at 18-weeks of age to a 45kcal% high fat diet. Comparing across the timecourses for (a), HDL-C; (b), insulin; (c), glucose; (d), IPGTT by calculating area under the curve, base-lined to t=0 values (data not shown), and analysing with a 1-way ANOVA non-parametric Kruskal-Wallis test and Dunns multiple comparison test we found: that HDL-C (a) was lower in MAT compared to either PAT or wildtype-MAT (p= 0.0055 and 0.0086 respectively) and that wildtype-PAT compared to PAT was not significantly (p=0.56) different; that insulin (b) was lower in MAT compared to either PAT or wildtype-MAT (p= <0.0003 and 0.0020 respectively) and that wildtype-PAT compared to PAT was not significantly (p>0.99) different; that Glucose (c) was lower in MAT compared to PAT (p= <0.0001) and that wildtype–MAT and wildtype-PAT compared to MAT and PAT respectively were not significantly (p=0.79 and >0.99 respectively) different; and that for an IPGTT (d) none of the group comparisons were significantly different (p>0.99) to each other. Then to examine effects at specific times for (a) HDL-C, (b) insulin, (c) glucose and (d) IPGTT individual pairwise comparisons were made at each timepoint using a Mann-Whitney 2-tailed t-test and were significantly reduced in the MAT group compared to PAT group at 8, 22 and 27 weeks in (a,b,c) and at all timepoints in (d). The MAT groups were also significantly lower compared to wildtype-MAT for HDL-C and insulin at 22 and 27 weeks. Values are expressed as mean ± SD and in (a,b,c) wildtype-MAT n=16, wildtype-PAT n=9, heterozygous-MAT n=15, heterozygous-PAT n=16 and in (d) wildtype-MAT n =19, wildtype-PAT n=10, MAT n=16, PAT n=16. For (e), HDL-C; (f), LDL-C and (g), total cholesterol was measured in a 33-week blood sample collected under terminal anaesthetic and was significantly, using an unpaired 2-tailed t-test, reduced in heterozygous-MAT mice compared to PAT mice. Values are expressed as mean ± SD (PAT N =8, MAT N =8). Wildtype MAT grey, wildtype PAT black, MAT KO red, PAT KO blue lines and fill
Clinical chemistry parameters were measured in female and male CRISPR-Cas9 knockout (KO) C57BL/6J Klf14 mice and their wildtype (Wt) controls. Mice were fed a standard diet throughout their lifetimes. (a), HDL-C at 12 weeks of age was significantly reduced in female and reduced with borderline significance in male (unpaired two-tailed t-test). (b), triglycerides at 12 weeks of age were not significantly different in females or males (unpaired two-tailed t-test). Comparing across the IPGTT timecourses (c and d) by calculating area under the curve, base-lined to t=0 values, (data not shown) and analysing using a Mann-Whitney two-tailed t-test showed there was no significant difference between male or female KO and wildtype mice (p=0.23 and 0.93 respectively). Then in order to make comparisons at each timepoint for (c and d), individual pairwise Mann-Whitney 2-tailed t-tests were carried out, and as for AUC no differences were found in the female data, although in males glucose levels were nominally higher at 20, 60 and 120 minutes of the test. (e and f), ITT at 16 weeks was not significantly changed between wildtype and KO groups of either female (e) or male (f) mice, analysed by 2-way ANOVA with repeated measures and Bonferroni correction. Values are expressed as mean ± SD and in (a,b) female wildtype n=9, female KO n=9, male Wt n=9, male KO n=9 and in (c,d,ef) female wildtype n=9, female KO n=8, male Wt n=9, male KO N=9. Wildtype mice in blue lines and fill, KO mice in red lines and fill
Supplementary Figure 6 Gluteal adipose tissue of T2D risk allele homozygotes contains fewer, larger mature adipocytes compared to non-risk allele homozygotes.
Adipocyte cell area in histological sections of subcutaneous gluteal adipose biopsies from the Oxford BioBank. a, Median cell area in female (N=18, mean ± SEM) and male (N=18, mean ± SEM) gluteal biopsies stratified by genotype at rs4731702. b, Cumulative frequency distribution of adipocyte cell area in females (N=18). c, Cumulative frequency distribution of adipocyte cell area in males (N=18). Statistical significance was assessed using a Wilcoxon signed-rank two-sided test