Evaluating the Role of LPIN1 Variation on Insulin Resistance, Body Weight and Human Lipodystrophy in UK Populations

Diabetes Publish Ahead of Print, published online June 30, 2008

Evaluating the role of LPIN1 variation on insulin resistance, body weight and human lipodystrophy in UK populations

Katherine A. Fawcett BSc1, Neil Grimsey BSc2, Ruth J.F. Loos PhD3, Eleanor Wheeler PhD1, Allan Daly BSc1, Maria Soos PhD4, Robert Semple MD PhD4, Holly Syddall MSc5, Cyrus Cooper DM5, Symeon Siniossoglou PhD2, Stephen O’Rahilly MD4, Nicholas J. Wareham PhD3, Inês Barroso PhD1

1Metabolic Disease Group, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SA, UK 2Cambridge Institute for Medical Research, University of Cambridge, Wellcome Trust/Medical Research Council (MRC) Building, Hills Road, Cambridge CB2 2XY, UK 3MRC Epidemiology Unit, Institute of Metabolic Science, Cambridge, UK 4Department of Clinical Biochemistry, Addenbrooke's Hospital, University of Cambridge, Cambridge, UK 5MRC Epidemiology Resource Centre, University of Southampton, Southampton, UK

Inês Barroso Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SA E-mail: [email protected]

Received 27 March 2008 and accepted 17 June 2008.

Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org

This is an uncopyedited electronic version of an article accepted for publication in Diabetes. The American Diabetes Association, publisher of Diabetes, is not responsible for any errors or omissions in this version of the manuscript or any version derived from it by third parties. The definitive publisher-authenticated version will be available in a future issue of Diabetes in print and online at http://diabetes.diabetesjournals.org.

OBJECTIVE: Loss of Lpin1 activity causes lipodystrophy and insulin resistance in the fld mouse, and LPIN1 expression and common genetic variation were recently suggested to influence adiposity and insulin sensitivity in humans. We aimed to conduct a comprehensive association study to clarify the influence of LPIN1 common variation on adiposity and insulin sensitivity in UK populations, and to examine the role of LPIN1 mutations in insulin resistance syndromes. RESEARCH DESIGN AND METHOD: Twenty-two SNPs tagging LPIN1 common variation were genotyped in MRC Ely (N = 1709) and Hertfordshire (N = 2901) population-based cohorts. LPIN1 exons, exon/intron boundaries and 3’UTR were sequenced in 158 patients with idiopathic severe insulin resistance (including 23 lipodystrophic patients), and 48 controls. RESULTS: We found no association between LPIN1 SNPs and fasting insulin, but report a nominal association between rs13412852 and BMI (P = 0.042) in a meta-analysis of 8504 samples from in-house and publicly available studies. Three rare nonsynonymous variants (A353T, R552K and G582R) were detected in severely insulin resistant patients. However, these did not co-segregate with disease in affected families and Lipin1 protein expression and phosphorylation in patients with variants was indistinguishable from controls. CONCLUSIONS: Our data do not support a major effect of LPIN1 common variation on metabolic traits and suggest that mutations in LPIN1 are not a common cause of lipodystrophy in humans. The nominal associations with BMI and other metabolic traits in UK cohorts require replication in larger cohorts.

LPIN1, insulin resistance and adiposity

ipin 1, a multi-functional protein associated with BMI in a Finnish obesity highly expressed in mouse and human case-control and in the Quebec Family Study L adipose tissue, has been shown to (12; 14) but not in a German population-based influence adipose tissue development and cohort (the MONICA study) (13). Moreover, function. Null mutations in the murine lipin 1 LPIN1 haplotypes were strongly associated gene (Lpin1) result in impaired adipocyte with traits underlying metabolic syndrome in differentiation leading to a severe reduction in the MONICA study but these haplotypes adipose tissue mass, insulin resistance, and often had the opposite effect on the same progressive peripheral neuropathy in the fld traits in a replication cohort (13). This and fld2J mouse models (1). In contrast, inconsistency suggests that further studies are transgenic mice with adipose tissue-specific needed to clarify the role of LPIN1 variation overexpression of Lpin1 exhibit diet-induced on human metabolic traits. In this study we obesity and enhanced insulin sensitivity have taken complementary approaches to compared to wild-type littermates (2). In study the role of LPIN1 variation in human humans, LPIN1 expression in adipose tissue metabolic traits in UK populations: appears to be inversely correlated with measures of adiposity such as Body Mass a. We genotyped 22 SNPs that tag Index (BMI), and positively correlated with common LPIN1 variation (Minor Allele insulin sensitivity (3-6). Frequency (MAF) >0.01) in two white UK population-based cohorts (N = 4610) and The mechanism through which lipin 1 tested for association with fasting serum influences adiposity and insulin sensitivity in insulin levels, BMI, and a number of mice and humans is not entirely known. additional metabolic traits with previously However, recent data indicates that lipin 1 is a reported association with LPIN1. magnesium-dependent phospatidate phosphatase responsible for catalysing the b. We sequenced LPIN1 in a cohort of penultimate step in triacylglyceride synthesis, patients with syndromes of severe insulin explaining why Lpin1 deficient fld mice resistance (N=135) and lipodystrophy (N=23) cannot accumulate fat in mature adipocytes to identify potentially pathogenic mutations. (7). Lipin 1 is also thought to regulate RESEARCH DESIGN AND METHODS transcription of genes involved in adipocyte differentiation (PPARγ, C/EBPα), fat Definition of cohorts synthesis and storage (DGAT, ACC-1, PEPCK, FAS, SCD1), and fatty acid oxidation ELY Cohort. The Medical Research Council (CPT-1, AOX, PPARα) (2; 6-10). (MRC) Ely Study is a population-based cohort study of the aetiology and There has only been one study pathogenesis of type 2 diabetes and related sequencing LPIN1 in lipodystrophic patients metabolic disorders in the UK (15). It (N=15), with no pathogenic mutation being comprises white men and women aged 35-79 reported (11). Furthermore, although a years without diagnosed diabetes. number of studies have evaluated the role of Measurements of anthropometric and common variation in LPIN1 and metabolic metabolic data analysed in this study have quantitative phenotypes (12-14), the results been described previously (16). Informed have been inconsistent across studies, and consent was obtained from all participants sometimes within the same study. For and ethical approval for the study was granted example, rs2716610 and a SNP in high by the Cambridge Local Research Ethics linkage disequilibrium, rs2716609, were Committee.

LPIN1, insulin resistance and adiposity

Hertfordshire Cohort. The Hertfordshire informed consent with approval of the local Cohort Study was recruited from the cohort of research ethics committee in Cambridge, U.K. people born in Hertfordshire between 1931 and 1939. The cohort details and Selection of SNPs to tag LPIN1 common measurements of metabolic traits analysed in variation. TagSNPs were selected to cover this study have been described previously variation with MAF greater than 1% present (17). in LPIN1 and its flanking 4 kb regions (Chromosome 2, coordinates 11800212- EPIC-Obesity Study. The EPIC-Obesity 11889941 (NCBI B36 assembly)). The study is nested within the EPIC-Norfolk International HapMap (Rel 20, PhaseII) study, a population based cohort study of reports 61 SNPs for CEU samples in this 25663 white European men and women aged chromosomal region, while we identified 5 39-79 recruited in Norfolk, UK between 1993 novel SNPs during resequencing of 48 CEPH and 1997 (18). Height and weight were individuals (Supplementary Table 1). measured using standard anthropometric Twenty-five tagSNPs were selected to cover techniques (18). All samples were genotyped all 66 novel and known HapMap using the Affymetrix GeneChip Human polymorphisms with a pairwise r2 ≥ 0.8 using Mapping 500K Array Set that contained Tagger (21) as a stand-alone program in genotype information for 5 of our LPIN1 Haploview (22) (Supplementary Figure 1). tagSNPs, each of which had a call rate >90%. Twenty-two of these tagSNPs passed assay In total, 2415 individuals with height and design and pre-screening (Supplementary weight measures and quality-controlled Table 2). The three failed SNPs tagged only genotype data were available for analyses. five intronic SNPs. We calculated that we have >80% power to detect a per allele effect HGDP-CEPH. The HGDP-CEPH Human on BMI of >1.33 kg/m2 with MAF=0.01, and Genome Diversity Cell Line Panel is a >0.27kg/m2 with MAF=0.5. For logged resource of 1064 DNA samples from fasting insulin data this range is >1.04 to individuals distributed around the world and >1.22. has been described previously (19). Genotyping. Genotyping was performed by CEPH. 48 unrelated individuals from CEPH the genotyping facility within the Genetics of families supplied by Coriell Cell Repositories Complex Traits in Humans team at the (20) are control individuals of North and West Wellcome Trust Sanger Institute. LPIN1 European origin. mutations A353T, R552K, and G582R were genotyped on the CEPH human diversity Severe Insulin Resistance Cohort. All panel as stand-alone assays using the patients had severe insulin resistance, defined Sequenom MassArray hME platform as fasting insulin above 150 pmol/l, or peak according to the manufacturer’s instructions. insulin on oral glucose tolerance testing above Twenty-two LPIN1 tagging SNPs 1,500 pmol/l in non-diabetic patients. In (Supplementary Table 2) were genotyped complete insulin deficiency it was defined as using the Sequenom MassArray iPlex an insulin requirement above 3U/kg/day. 2 platform according to the manufacturer's Most patients had a BMI <30 kg/m and at instructions ( Sequenom, San Diego, CA). least 58 had BMI>30. Those with partial beta Primers, probes and conditions available on cell decompensation and clinical features request. All SNPs, except for rs17603755, including acanthosis nigricans, and those with 2 were successfully genotyped in > 85% of BMI >30 kg/m were included at the samples (call rates for each SNP are presented investigators’ discretion. All patients gave 4

LPIN1, insulin resistance and adiposity

in Supplementary Table 2) and did not deviate PCR and sequencing. Genomic DNA from from Hardy-Weinberg Equilibrium (p > 0.01). patients was randomly preamplified in a The average call rate across Ely and GenomiPhi reaction (GE Healthcare UK, Hertfordshire samples pooled together was Chalfont St. Giles, UK) prior to amplification 92.1%. with gene-specific primers (designed using Primer3 software (http://frodo.wi.mit.edu/cgi- Statistical analysis. Deviation of LPIN1 bin/primer3/primer3_www.cgi)) covering all tagSNP genotype from Hardy-Weinberg coding exons, splice junctions, and 3’UTR equilibrium was assessed using a goodness- 2 (sequences and cycling conditions available of-fit χ test. Linear regression analysis was on request). PCR was performed using used to assess the association between standard conditions and products purified individual SNPs and BMI, log-transformed using exonuclease I and shrimp alkaline fasting plasma insulin, and log-transformed phosphatase (USB Corporation, Cleveland, additional metabolic traits (systolic and OH, USA). Bi-directional sequencing was diastolic blood pressure, HDL and LDL performed using the Big Dye Terminator 3.1 cholesterol, plasma triglycerides, waist kit (Applied Biosystems, Foster City, CA, circumference, and HbA1c levels) in Ely and USA). Sequencing reactions were run on Hertfordshire cohorts using Stata v9 (Stata ABI3730 capillary machines (Applied Corporation, Texas, USA). All analyses were Biosystems) and analysed using Mutation adjusted for age and sex and, in the case of Surveyor version.2.20 (SoftGenetics LLC, the joint analysis, included an indicator term State College, PA, USA). All non- for study. Logistic regression in Stata was synonymous variants with MAF<0.01 were used to test for association between LPIN1 confirmed in a second PCR and sequencing SNPs and risk of hypertension and diabetes. reaction using patient genomic DNA. DNA Chi-squared analysis was performed to test from family members used for co-segregation for significant differences (P<0.01) in call analysis was genomic. Only one amplicon rate between cases and controls. The joint within the 3’UTR failed sequencing, and all Ely and Hertfordshire cohort analysis of others passed on >85% of samples (with an additional traits underlying metabolic average pass rate of 95%). PANTHER was syndrome comprised 189 tests so the P value used to predict the functional impact of non- threshold adjusted for multiple testing using synonymous mutations the Bonferroni correction is 0.0003. Fixed (http://www.pantherdb.org/tools/csnpScoreFo effect meta-analysis using the inverse rm.jsp). Sequencing of pre-amplified CEPH variance method was performed by using the samples was done in the same way. metan command in Stata (23). Heterogeneity among studies was assessed using the Q Western blotting. Patient fibroblast cells were statistic. IMPUTE software maintained in DMEM medium supplemented (http://www.stats.ox.ac.uk/~marchini/softwar with 10% fetal bovine serum (FBS) and 2mM e/gwas/impute.html) was used to impute L-glutamine in a humidified 37°C incubator genotypes for rs17603420 in the EPIC cohort. with 5% CO2. All cells were routinely Plink assessed for and protected against (http://pngu.mgh.harvard.edu/~purcell/plink/) mycoplasma infection using VenorGeM® was used to perform hapotype analysis (24), mycoplasma detection kit (Minerva biolabs, and Ely and Hertfordshire cohorts were meta- CamBio, VGM-025) and BM-cyclin (Roche; analysed using METAL 799050) respectively. Fibroblasts were (http://www.sph.umich.edu/csg/abecasis/meta collected by trypsin EDTA release, washed l/index.html). with PBS and then lysed in a 50mM HEPES

LPIN1, insulin resistance and adiposity

pH7.4 buffer containing 150mM NaCl, 1% mounted onto glass slides and then visualised Triton X-100, 100µM AEBSF, Protease with a 63X or 100X Plan Apochromat inhibitor cocktail 1, Phosphatase inhibitor objective (numerical aperture,1.4) on a Ziess cocktail II, 1µg/ml Dnase and 4mM MgCl2 Axiovert 200M inverted microscope with an chilled to 4°C. Each sample was LSM 510 confocal laser Scanning attachment. homogensised by passing through a 25G needle 10 times. Insoluble debris was RESULTS removed by a 16,000g centrifugation step at Association studies of LPIN1 tagSNPs. 4°C. Sample protein concentration was Twenty-one LPIN1 tagSNPs were measured by a comparative Bradford protein successfully genotyped in two white UK assay. Samples were then suspended in 1 x population-based cohorts, the MRC Ely study SDS sample buffer and boiled at 95°C for 5 (N = 1709) and the Hertfordshire cohort study minutes, loaded onto 7% SDS-PAGE, (N = 2901). In the MRC Ely cohort, the transferred onto nitrocellulose, and blocked in minor allele of rs13412852 is nominally PBS with 1% TX-100 and 5% milk. These associated with lower fasting insulin levels (P were then probed with protein specific = 0.041) and the minor allele of rs17603350 primary antibodies: Lipin 1, Lipin 2, anti- is nominally associated with higher BMI (P = Mab414 (Covance; MMS-120P), anti-laminB 0.031) but these associations are not (Santa-Cruz; sc-6217). Lipin 1 and 2 replicated in the Hertfordshire cohort (Table polyclonal antibody production will be 1). Conversely, in the Hertfordshire cohort, described elsewhere (Grimsey and rs17603420 and rs2577261 are nominally Siniossoglou, in preparation). This was associated with BMI (P = 0.01 and P = 0.006 followed by species specific secondary respectively), but not in the MRC Ely cohort antibodies coupled to Horse Radish (Table 1). In a joint analysis of the pooled Peroxidase (HRP): anti-rabbit IgG (Jackson Ely and Hertfordshire cohorts (Supplementary immuno research; 211-032-171), anti-goat Table 3), no SNPs were associated with IgG (Novus Biologicals; NB 710-H), anti- fasting insulin levels but rs13412852, mouse IgG (H & L) highly cross-adsorbed rs17603420 and rs2577261 were nominally (Molecular probes; A11029). Proteins were associated with BMI (p ≤ 0.05). Two of these then detected using standard SNPs, rs13412852 and rs2577261, overlapped electrochemiluminescence techniques with SNPs on the Affymetrix 500k and (Amersham ECL-reagents). Illumina 300k SNP chips, and rs17603420 Indirect immunofluorescence by confocal could be imputed. Consequently we were microscopy. Fibroblasts were fixed with 3% able to increase the power of our study to Formaldehyde, permiablised with 0.1% Triton detect modest effects of these SNPs on BMI X-100, and blocked with 1mg/ml BSA in by performing meta-analyses with in-house PBS. Each cover slip was labelled with data (EPIC-Obesity study, N = 2415) and, in primary mouse −αMab414 (nuclear pore the case of rs13412852 and rs2577261, data marker) and secondary anti-mouse conjugated deposited by the WTCCC and WTSI and to FITC (green), primary rabbit published online from the British 1958 DNA −αCalreticulin (Endoplasmic reticulum collection (N = 1479) calcium binding protein) (Calbiochem; (http://www.b58cgene.sgul.ac.uk/, accessed 208910), and secondary anti-rabbit January 2008). SNPs rs2577261 and conjugated to Alexa fluor 594 (red) rs17603420 were not associated with BMI in (Molecular probes; A11037), DNA was the meta-analysis (P = 0.114 and 0.071 stained with DAPI (blue). Each slide was respectively). However, the association 6

LPIN1, insulin resistance and adiposity

between rs13412852 and BMI remained lipin 1 protein by PANTHER. DNA from the marginally statistically significant (P = 0.042) patient’s mother, maternal aunts, and maternal in the meta-analysis (Figure 1). grandparents was sequenced and demonstrated that the A353T variant did not In analyses of pooled Ely and segregate with the hallmarks of insulin Hertfordshire cohorts, a number of nominal resistance in the family (Figure 3A). A353T associations (P<0.05) were detected between was also genotyped in 1064 participants of the LPIN1 tagSNPs and traits previously reported HGDP-CEPH Human Genome Diversity Cell to be associated with LPIN1 variation (13). Line Panel (Diversity Panel) but was not These data are presented in Supplementary detected. Tables 4 and 5. For SNPs overlapping with the Affymetrix 500k SNP chip, meta-analyses R552K was detected in two unrelated were performed on continuous traits with white European females, but not in 1064 publicly available data from the Broad controls from the Diversity Panel. The first Institute proband presented with severe insulin (http://www.broad.mit.edu/diabetes/scandinav resistance and femorogluteal lipodystrophy at s/metatraits.html) (Supplementary Table 4). 15 years old. The lipodystrophy progressed to For rs2577256 meta-analysis was performed become generalised in conjunction with the with publicly available WTCCC data to test appearance of aggressive haemolytic anemia for association with diabetes and hypertension and autoimmune liver disease. Liver failure (Supplementary Table 5). led to her demise at 24 years old. The other proband was diagnosed with insulin resistant Mutation screening in the severe insulin diabetes at 32 years old, and subsequently resistance (SIR) cohort. A total of 44 variants required in excess of 4U/day exogenous were detected in insulin resistant or insulin to maintain satisfactory glycemic lipodystrophic patients (Supplementary Table control. She had no clinical evidence of 1), eight of which were present in the coding lipodystrophy, and her BMI was sustained sequence (Figure 2A). Coding sequence above 30 kg/m2. R552 is within a highly variants that did not alter the amino acid conserved tract (Figure 2B) and mutation to sequence and/or that were also present in lysine is predicted by PANTHER to have controls were considered unlikely to be deleterious effects on lipin 1 function. Family pathogenic (shown below the schematic in DNA was not available for co-segregation Figure 2A). This left three rare analysis for either patient. nonsynonymous variants detected as heterozygous (shown above the schematic) G582R was identified in a white, that did not fall within any known functional European male with a complex syndrome. domains within LPIN1. This included severe insulin resistance and severe, early onset sensorimotor neuropathy A353T was detected in a female which confined him to a wheelchair, a patient with a Pakistani father and British combination reminiscent of white mother. She presented with clinical lipodystrophy/insulin resistance and features of severe insulin resistance at 8 years neuropathy in the fld mouse. This patient also old, which worsened with weight gain in the underwent allogeneic bone marrow second decade, before improving dramatically transplantation in childhood for acute with weight loss in adult life. She had no lymphoblastic leukemia, and had a cerebral evidence of lipodystrophy. A353T was cavernous hemangioma. All genomic predicted to have no functional impact on the analyses were undertaken on DNA extracted

LPIN1, insulin resistance and adiposity from cultured skin fibroblasts. G582 is a well membrane morphology by staining a nuclear conserved residue within the protein (Figure pore marker (Supplementary Figure 2B). 2B), and mutation to arginine is predicted by There was no discernable difference in PANTHER to have deleterious effects on morphology between patient and control lipin 1 function. Co-segregation analysis was fibroblasts. performed using DNA from first-degree relatives of the patient (Figure 3B). The DISCUSSION father also carried the variant but although In this study we performed a diagnosed with diabetes at age 69 years, he comprehensive analysis of LPIN1 variants had no peripheral neuropathy, nor clinical or and their effects on metabolic quantitative biochemical evidence of insulin traits and syndromes of insulin resistance resistance/lipodystrophy. Subsequently, the (including lipodystrophy). Analysis of LPIN1 G582R variant was genotyped in the common variation (MAF>0.01) in two UK Diversity Panel and detected in a Bedouin population-based cohorts (N = 4610) revealed control individual from Nedev, Israel. nominal significant associations with BMI, In summary, we identified 3 rare and rs13412582 remained marginally LPIN1 missense variants in a cohort of insulin associated with BMI (P = 0.042) in a meta- resistant patients. G582R and R552K are analysis of UK population-based samples predicted to be deleterious by PANTHER and from in-house and publicly available genome- the proband carrying the G582R variant had a wide studies (N = 8504). We also detected syndrome reminiscent of the fld mouse. nominal associations between our tagSNPs Thus, despite the failure of this variant to co- and metabolic traits previously reported to be segregate with disease in the kindred, and associated with LPIN1 variation (13). despite the absence of available family Sequencing of 23 patients with lipodystrophy members from the R552K kindred, we elected and 135 patients with syndromes of insulin to investigate the possibility of impaired Lipin resistance revealed that mutations in LPIN1 1 function in primary skin fibroblasts from the are not a common cause of these diseases in probands. As A353T was predicted benign, humans. did not segregate with disease in the family, To our knowledge, neither rs13412582 and as no fibroblasts were available, this nor any highly correlated SNPs have been variant was not investigated further. tested in other association studies published to Assessing the functional impact of LPIN1 date. Further replication in larger cohorts will mutations. To investigate the impact of be required to confirm the association R552K and G582R mutations on lipin 1 between rs13412582 and BMI. protein levels and phosphorylation status, Seven of our twenty-one tagSNPs total cell extracts from patient fibroblasts were directly genotyped in at least one of the were probed with lipin 1, lipin 2, a nuclear other association studies (12-14). All pore marker (Mab414), and lamin B-specific analyses, including our own, found no primary antibodies to detect protein levels association between rs4669781, rs1050800, (Supplementary Figure 2A). The resulting and rs2577256 and insulin levels and Western blot shown in Supplementary Figure measures of adiposity. Results for the other 2A shows similar intensities of all four four SNPs are inconsistent between studies. proteins in patient fibroblasts compared to For example, rs2716610 was associated with control cells. Immunocytochemistry was BMI in lean Finnish men (12) and with employed to detect changes in nuclear quantitative measures of adiposity in French- 8

LPIN1, insulin resistance and adiposity

Canadian families in the Quebec Family HapMap CEU trios) against metabolic traits Study (14). Here, the highly correlated SNP in Ely and Hertfordshire but only found rs2716609 (r2 = 1.0 in HapMap trios) was nominal associations with hypertension associated with skinfolds and waist (Supplementary Table 6). We did detect a circumference, and BMI showed the same number of nominal associations between trend. Given our sample size of 4130 these traits and other SNPs in our study individuals with full rs2716609 genotype and (Supplementary Tables 4 and 5). However, BMI data we had >80% power to detect the none of these reached statistical significance effect size observed in the Quebec Family after adjustment of the P value threshold for study. Nevertheless, we did not replicate the multiple testing using the Bonferroni association between rs2716609 and BMI or correction and require further replication. waist circumference in Ely and Hertfordshire cohorts (Table 1). Our results agree with the There are several possible reasons MONICA study Augsburg (N = 1416), a why we could not replicate previously German population-based cohort, which reported associations between LPIN1 variants found no association between rs2716610 and and metabolic quantitative traits. Firstly, we BMI in men or women (13). may have reported false negative results. However, where effect sizes were reported in Two other SNPs, rs893346 and previously published studies we were able to rs2577262, were associated with BMI in lean calculate that our study had >80% power to Finnish men (12) but showed no statistical detect them. Secondly, previous studies association with BMI in 1873 lean men from might have reported false positive results. In Ely and Hertfordshire cohort studies (P = particular, as a consequence of multiple 0.631 and 0.253 respectively). Similarly, testing, detection of false positive associations rs2278513 and rs2577262 were associated becomes more likely when analyses are with BMI in obese Finnish men but not in performed in subsets of samples and on many obese men from the UK (P = 0.780 and 0.676 traits. Furthermore, one has to expect false respectively). Our data agree with the positive findings amongst previously reported MONICA study which found no association disease associations given the low prior of rs893346 and two SNPs highly correlated probability of detecting a true association with rs2577262 in HapMap CEU trios (r2 = with a complex trait (25). Alternatively, the 1.0 and 0.96 for rs6744682 and rs6708316 discrepancy in results between studies may be respectively) with BMI in men (13). due to genetic and/or environmental differences between the populations The MONICA study reported strong genotyped. For example, the degree of associations between haplotypes of linkage disequilibrium between LPIN1 tag rs33997857, rs6744682 and rs6708316 and SNPs and the putative unmeasured true hypertension-, obesity-, and diabetes-related functional variant(s) may vary between the traits (13). Several of the traits were also cohorts. Also, LPIN1 SNPs may interact with statistically associated with the same other genetic and/or environmental risk haplotypes in a replication study, but the factors in different studies. effect was always in the opposite direction compared to the original cohort. To attempt In the fld mouse model Lpin1 null replication of the MONICA study data we mutations cause lipodystrophy, insulin tested haplotypes of rs33997857 and resistance and peripheral neuropathy (1). rs2577262 (highly correlated with rs6744682 However, of the three rare (MAF<0.01) (r2 = 1.0) and rs6708316 (r2 = 0.96) in nonsynonymous LPIN1 variants detected

LPIN1, insulin resistance and adiposity

within our cohort of patients with syndromes We conclude that LPIN1 coding of severe insulin resistance, none are likely to variants are not a common cause of be pathogenic in isolation in heterozygous lipodystrophy and severe insulin resistance in form: family co-segregation analysis showed humans, and that polymorphisms in LPIN1 that A353T and G582R did not segregate with are unlikely to importantly contribute to disease in a fully penetrant manner and insulin sensitivity and waist circumference in G582R was also detected in one Bedouin UK populations. Nominal associations control. between LPIN1 variants and BMI, blood pressure, cholesterol, triglycerides, HbA1c, Western blotting of patient fibroblasts and risk of hypertension need replicating in showed that G582R and R552K had no larger cohorts. discernable impact on lipin 1 protein levels. Proteins orthologous to lipin 1 in yeast are ACKNOWLEDGEMENTS proposed to be involved in nuclear membrane growth and morphology (26-28). However, We would like to thank Susannah staining of a nuclear pore marker in patient Bumpstead and Andrew Keniry, members of fibroblasts with R552K and G582R variants the genotyping facility within the Genetics of revealed no abnormalities in membrane Complex Traits in Humans team at the morphology compared to control fibroblasts. Wellcome Trust Sanger Institute, for genotyping LPIN1 variants in Ely and To date, our study (N=23) and Hertfordshire cohorts and the HGDP CEPH previously published work (N=15) (11) have panel. We acknowledge use of genotype data demonstrated that LPIN1 coding mutations from the British 1958 Birth Cohort DNA are unlikely to be a common cause of human collection, funded by the Medical Research lipodystrophy. However, we cannot rule out Council grant G0000934 and the Wellcome the possibility that LPIN1 mutations interact Trust grant 068545/Z/02. We also with other genetic defects to cause disease, or acknowledge use of genotype data from that they are rarer causes of these disorders. WTCCC and DGI. KF, EW, AD and IB are The methods used to screen for mutations funded by the Wellcome Trust. IB and EW would not have detected copy number also acknowledge support from EU FP6 variations affecting large regions, or funding (contract no.LSHM-CT-2003- mutations affecting regulatory regions, 503041). NG and SS are supported by a therefore we cannot exclude these types of Wellcome Trust Career Development LPIN1 variation as causes of human Fellowship. RS, MS, and SO are grateful for lipodystrophy and insulin resistance. the support of the Wellcome Trust (RS: Furthermore, the in vitro assays used to assess Intermediate Clinical Fellowship the functional impact of LPIN1 non- 080952/Z/06/Z and Programme Grant synonymous variants might have missed some 078986/Z/06/Z) and the U.K. NIHR functional effects, such as phosphatidic acid Cambridge Biomedical Research Centre. phosphatase (PAP) activity.

LPIN1, insulin resistance and adiposity

REFERENCES 1. Peterfy M, Phan J, Xu P, Reue K: Lipodystrophy in the fld mouse results from mutation of a new gene encoding a nuclear protein, lipin. Nat Genet 27:121-124, 2001 2. Phan J, Reue K: Lipin, a lipodystrophy and obesity gene. Cell Metab 1:73-83, 2005 3. Croce MA, Eagon JC, LaRiviere LL, Korenblat KM, Klein S, Finck BN: Hepatic lipin 1beta expression is diminished in insulin-resistant obese subjects and is reactivated by marked weight loss. Diabetes 56:2395-2399, 2007 4. Donkor J, Sparks LM, Xie H, Smith SR, Reue K: Adipose tissue lipin-1 expression is correlated with ppar{alpha} gene expression and insulin sensitivity in healthy young men. J Clin Endocrinol Metab, 2007 5. Lindegaard B, Larsen LF, Hansen AB, Gerstoft J, Pedersen BK, Reue K: Adipose tissue lipin expression levels distinguish HIV patients with and without lipodystrophy. Int J Obes (Lond) 31:449-456, 2007 6. Yao-Borengasser A, Rasouli N, Varma V, Miles LM, Phanavanh B, Starks TN, Phan J, Spencer HJ, 3rd, McGehee RE, Jr., Reue K, Kern PA: Lipin expression is attenuated in adipose tissue of insulin-resistant human subjects and increases with peroxisome proliferator-activated receptor gamma activation. Diabetes 55:2811-2818, 2006 7. Han GS, Wu WI, Carman GM: The Saccharomyces cerevisiae Lipin homolog is a Mg2+- dependent phosphatidate phosphatase enzyme. J Biol Chem 281:9210-9218, 2006 8. Finck BN, Gropler MC, Chen Z, Leone TC, Croce MA, Harris TE, Lawrence JC, Jr., Kelly DP: Lipin 1 is an inducible amplifier of the hepatic PGC-1alpha/PPARalpha regulatory pathway. Cell Metab 4:199-210, 2006 9. Phan J, Peterfy M, Reue K: Lipin expression preceding peroxisome proliferator-activated receptor-gamma is critical for adipogenesis in vivo and in vitro. J Biol Chem 279:29558-29564, 2004 10. Xu J, Lee WN, Phan J, Saad MF, Reue K, Kurland IJ: Lipin deficiency impairs diurnal metabolic fuel switching. Diabetes 55:3429-3438, 2006 11. Cao H, Hegele RA: Identification of single-nucleotide polymorphisms in the human LPIN1 gene. J Hum Genet 47:370-372, 2002 12. Suviolahti E, Reue K, Cantor RM, Phan J, Gentile M, Naukkarinen J, Soro-Paavonen A, Oksanen L, Kaprio J, Rissanen A, Salomaa V, Kontula K, Taskinen MR, Pajukanta P, Peltonen L: Cross-species analyses implicate Lipin 1 involvement in human glucose metabolism. Hum Mol Genet 15:377-386, 2006 13. Wiedmann S, Fischer M, Koehler M, Neureuther K, Riegger G, Doering A, Schunkert H, Hengstenberg C, Baessler A: Genetic variants within the LPIN1 gene, encoding lipin, are influencing phenotypes of the metabolic syndrome in humans. Diabetes, 2007

LPIN1, insulin resistance and adiposity

14. Loos RJF, Rankinen T, Pérusse L, Tremblay A, Després J-P, Bouchard C: Association of Lipin 1 Gene Polymorphisms with Measures of Energy and Glucose Metabolism. Obesity 15:2723-2732, 2007 15. Wareham NJ, Hennings SJ, Byrne CD, Hales CN, Prentice AM, Day NE: A quantitative analysis of the relationship between habitual energy expenditure, fitness and the metabolic cardiovascular syndrome. Br J Nutr 80:235-241, 1998 16. Ekelund U, Franks PW, Sharp S, Brage S, Wareham NJ: Increase in physical activity energy expenditure is associated with reduced metabolic risk independent of change in fatness and fitness. Diabetes Care 30:2101-2106, 2007 17. Syddall HE, Aihie Sayer A, Dennison EM, Martin HJ, Barker DJ, Cooper C: Cohort profile: the Hertfordshire cohort study. Int J Epidemiol 34:1234-1242, 2005 18. Day N, Oakes S, Luben R, Khaw KT, Bingham S, Welch A, Wareham N: EPIC-Norfolk: study design and characteristics of the cohort. European Prospective Investigation of Cancer. Br J Cancer 80 Suppl 1:95-103, 1999 19. Cann HM, de Toma C, Cazes L, Legrand MF, Morel V, Piouffre L, Bodmer J, Bodmer WF, Bonne-Tamir B, Cambon-Thomsen A, Chen Z, Chu J, Carcassi C, Contu L, Du R, Excoffier L, Ferrara GB, Friedlaender JS, Groot H, Gurwitz D, Jenkins T, Herrera RJ, Huang X, Kidd J, Kidd KK, Langaney A, Lin AA, Mehdi SQ, Parham P, Piazza A, Pistillo MP, Qian Y, Shu Q, Xu J, Zhu S, Weber JL, Greely HT, Feldman MW, Thomas G, Dausset J, Cavalli-Sforza LL: A human genome diversity cell line panel. Science 296:261-262, 2002 20. Dausset J, Cann H, Cohen D, Lathrop M, Lalouel JM, White R: Centre d'etude du polymorphisme humain (CEPH): collaborative genetic mapping of the human genome. Genomics 6:575-577, 1990 21. de Bakker PI, Yelensky R, Pe'er I, Gabriel SB, Daly MJ, Altshuler D: Efficiency and power in genetic association studies. Nat Genet 37:1217-1223, 2005 22. Barrett JC, Fry B, Maller J, Daly MJ: Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263-265, 2005 23. Bradburn MJ, Deeks JJ, Altman DG: Metan - an alternative meta-analysis command. Stata Technical Bulletin Reprints 8:86-100, 1999 24. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC: PLINK: a tool set for whole-genome association and population- based linkage analyses. Am J Hum Genet 81:559-575, 2007 25. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N: Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst 96:434-442, 2004 26. Santos-Rosa H, Leung J, Grimsey N, Peak-Chew S, Siniossoglou S: The yeast lipin Smp2 couples phospholipid biosynthesis to nuclear membrane growth. Embo J 24:1931-1941, 2005

LPIN1, insulin resistance and adiposity

27. Siniossoglou S, Santos-Rosa H, Rappsilber J, Mann M, Hurt E: A novel complex of membrane proteins required for formation of a spherical nucleus. Embo J 17:6449-6464, 1998 28. Tange Y, Hirata A, Niwa O: An evolutionarily conserved fission yeast protein, Ned1, implicated in normal nuclear morphology and chromosome stability, interacts with Dis3, Pim1/RCC1 and an essential nucleoporin. J Cell Sci 115:4375-4385, 2002

LPIN1, insulin resistance and adiposity

TABLE 1. Mean Fasting insulin levels and mean BMI of study participants by LPIN1 tagSNP genotype in the MRC Ely cohort and the Hertfordshire cohort study

Ely Hertfordshire

Insulin (pmol/l) BMI (kg/m2) Insulin (pmol/l) BMI (kg/m2)

P P P P SNP 0 1 2 value 0 1 2 value 0 1 2 value 0 1 2 value

rs893346 49.18 ± 1.02 47.65 ± 1.05 57.57 ± 1.08 0.609 27.23 ± 0.13 27.5 ± 0.39 26.3 ± 1.7 0.303 70.6 ± 1.01 72.33 ± 1.04 78.17 ± 1.2 0.486 27.36 ± 0.09 27.43 ± 0.24 26.14 ± 1.08 0.993

rs4669778 49.99 ± 1.03 48.88 ± 1.02 47.23 ± 1.03 0.216 27.42 ± 0.24 27.41 ± 0.16 27.11 ± 0.27 0.109 71.28 ± 1.02 72.11 ± 1.02 69.06 ± 1.03 0.409 27.41 ± 0.16 27.37 ± 0.12 27.33 ± 0.16 0.719

rs893345 48.17 ± 1.03 48.69 ± 1.02 49.76 ± 1.03 0.424 27.27 ± 0.28 27.75 ± 0.32 25.87 ± 1.57 0.967 71.15 ± 1.03 71.73 ± 1.02 69.14 ± 1.03 0.406 27.21 ± 0.17 27.38 ± 0.12 27.49 ± 0.17 0.234

rs7595221 47.75 ± 1.03 50.08 ± 1.02 48.59 ± 1.04 0.55 27.11 ± 0.22 26.99 ± 0.24 26.09 ± 0.92 0.815 70.01 ± 1.02 71.45 ± 1.02 70.3 ± 1.03 0.708 27.34 ± 0.14 27.44 ± 0.12 27.23 ± 0.19 0.734

Novel1 48.46 ± 1.02 49.79 ± 1.03 49.67 ± 1.08 0.409 27.15 ± 0.15 27.49 ± 0.21 27.01 ± 0.64 0.393 71.71 ± 1.02 70.63 ± 1.02 71.25 ± 1.06 0.575 27.34 ± 0.1 27.52 ± 0.16 27.72 ± 0.45 0.172

rs16857866 49.1 ± 1.01 53.35 ± 1.08 47.7 ± 0 0.398 27.23 ± 0.12 28.02 ± 0.64 29.65 ± 0 0.215 70.96 ± 1.01 72.11 ± 1.06 0 ± 0 0.734 27.36 ± 0.09 27.8 ± 0.4 0 ± 0 0.306

rs13412852 50.18 ± 1.02 48.63 ± 1.02 45.1 ± 1.05 0.042 27.42 ± 0.18 27.36 ± 0.18 26.44 ± 0.35 0.054 70.57 ± 1.02 71.63 ± 1.02 69.18 ± 1.04 0.89 27.47 ± 0.13 27.36 ± 0.12 27.15 ± 0.2 0.249

rs2278513 47.41 ± 1.03 49.45 ± 1.02 48.82 ± 1.04 0.37 27.06 ± 0.22 27.28 ± 0.16 27.45 ± 0.26 0.467 69.47 ± 1.02 71.46 ± 1.02 72.34 ± 1.03 0.234 27.45 ± 0.14 27.32 ± 0.12 27.42 ± 0.2 0.753

rs3795974 48.59 ± 1.02 49.57 ± 1.02 49.93 ± 1.04 0.441 27.21 ± 0.19 27.76 ± 0.96 27.86 ± 3.19 0.844 71.04 ± 1.02 70.16 ± 1.02 72.91 ± 1.04 0.602 27.31 ± 0.13 27.41 ± 0.13 27.42 ± 0.22 0.633

rs33997857 48.95 ± 1.01 46.11 ± 1.1 56.18 ± 1.9 0.627 27.24 ± 0.12 27.7 ± 0.67 28.76 ± 0.57 0.401 70.82 ± 1.01 78.33 ± 1.07 73.83 ± 0 0.166 27.36 ± 0.08 27.91 ± 0.44 25.2 ± 0 0.259

rs17603350 48.99 ± 1.02 47.89 ± 1.05 99.5 ± 1.37 0.986 27.34 ± 0.13 26.3 ± 0.36 27.52 ± 1.28 0.031 71.27 ± 1.01 67.78 ± 1.05 60.16 ± 2.08 0.282 27.4 ± 0.09 27.15 ± 0.36 30.9 ± 4.96 0.725

rs17603420 50.63 ± 1.03 47.75 ± 1.02 49.14 ± 1.03 0.271 27.33 ± 0.2 27.39 ± 0.18 26.9 ± 0.25 0.269 70.43 ± 1.02 71.41 ± 1.02 70.26 ± 1.03 0.971 27.58 ± 0.16 27.42 ± 0.12 26.92 ± 0.17 0.01

rs6729430 48.95 ± 1.02 45.77 ± 1.1 49.99 ± 1.47 0.61 27.24 ± 0.12 27.39 ± 0.34 29.79 ± 1.28 0.508 70.98 ± 1.01 71.68 ± 1.08 0 ± 0 0.849 27.34 ± 0.08 28.37 ± 0.53 0 ± 0 0.053

rs2577264 48.61 ± 1.02 48.91 ± 1.02 50.6 ± 1.04 0.368 27.19 ± 0.2 27.37 ± 0.18 27.21 ± 0.3 0.733 71.27 ± 1.02 69.8 ± 1.02 73.51 ± 1.04 0.644 27.28 ± 0.13 27.39 ± 0.12 27.48 ± 0.23 0.462

rs2577262 49.25 ± 1.02 48.96 ± 1.02 48.72 ± 1.05 0.639 27.24 ± 0.17 27.61 ± 0.28 27.99 ± 0.86 0.857 70.76 ± 1.02 71.58 ± 1.02 69.77 ± 1.04 0.95 27.47 ± 0.12 27.37 ± 0.13 26.98 ± 0.25 0.148

rs2577261 49.52 ± 1.02 46.67 ± 1.03 48.09 ± 1.12 0.15 27.18 ± 0.12 27.38 ± 0.18 27.23 ± 0.3 0.256 71.08 ± 1.02 71.39 ± 1.03 73.5 ± 1.08 0.719 27.23 ± 0.09 27.67 ± 0.22 29.06 ± 0.72 0.006

rs4669781 48.91 ± 1.02 49.56 ± 1.04 43.59 ± 1.21 0.844 27.22 ± 0.12 27.19 ± 0.18 27.25 ± 0.29 0.780 71.03 ± 1.01 69.44 ± 1.03 66.11 ± 1.11 0.426 27.41 ± 0.09 27.02 ± 0.24 26.81 ± 2.07 0.126

rs2716609 48.99 ± 1.02 48.25 ± 1.05 44.85 ± 1.2 0.61 27.13 ± 0.15 27.34 ± 0.17 26.82 ± 0.24 0.073 71.19 ± 1.01 69.03 ± 1.04 78.39 ± 1.22 0.597 27.35 ± 0.1 27.42 ± 0.17 27.82 ± 0.66 0.465

Novel2 48.11 ± 1.02 50.44 ± 1.04 52.4 ± 1.17 0.117 27.25 ± 0.12 27.25 ± 1.26 0 ± 0 0.987 71.65 ± 1.02 68.38 ± 1.03 76.59 ± 1.08 0.34 27.38 ± 0.08 25.71 ± 0.72 0 ± 0 0.103

rs1050800 49.01 ± 1.01 47.15 ± 1.19 0 ± 0 0.728 27.21 ± 0.14 27.34 ± 0.23 26.55 ± 0.62 0.993 71.02 ± 1.01 69.3 ± 1.15 0 ± 0 0.848 27.4 ± 0.1 27.34 ± 0.16 27.14 ± 0.55 0.585

rs2577256 48.81 ± 1.02 49.32 ± 1.03 49.09 ± 1.09 0.862 27.02 ± 0.24 27.41 ± 0.19 26.94 ± 0.38 0.208 70.48 ± 1.02 71.77 ± 1.02 74.09 ± 1.07 0.386 27.19 ± 0.16 27.42 ± 0.12 27.47 ± 0.18 0.219

Data are means ± standard error. The P value indicates the results of a regression analysis assuming an additive model of gene action (nominally significant values, p<0.05, are highlighted in bold). For fasting insulin the analysis was performed on log- transformed data, and the table shows geometric means and standard errors. 0 = homozygous for the major allele (refer to Supplementary table 2), 1 = heterozygous, 2 = homozygous for the minor allele

LPIN1, insulin resistance and adiposity

FIG. 1. Association between rs13412852 and BMI in individual studies (Ely = MRC Ely study, Hertfordshire = Hertfordshire cohort study, EPIC = EPIC Obesity cohort, and 1958 = 1958 British Birth cohort) and combined effect size (Overall), P = 0.045

Effect size Study (95% CI) % Weight

Ely -0.35 (-0.70, 0.01) 15.2

Hertfordshire -0.14 (-0.38, 0.10) 32.6

EPIC -0.11 (-0.33, 0.12) 37.8

1958 -0.02 (-0.38, 0.34) 14.5

Overall -0.14 (-0.28,-0.00) 100.0

-.7 0 .7

Effect size

LPIN1, insulin resistance and adiposity

FIG. 2. A) Schematic of the lipin 1 protein showing exons in alternating black and white and known domains among lipin family proteins in grey. Arrows indicate the location of coding SNPs detected in LPIN1 by sequencing 23 patients with partial lipodystrophy and 135 patients with other syndromes of severe insulin resistance. Rare (MAF<1%) non-synonymous mutations (above) were considered potentially pathogenic. NLIP (amino acids 1-114) = N-terminal lipin domain, NLS (amino acids 153-158) = nuclear localisation signal, and CLIP (amino acids 674- 830) = C-terminal lipin domain, also referred to as the LNS2 (Lipin/Ned1/Smp2) domain. HAD = haloacid dehalogenase domain. B) Multiple sequence alignments (using ClustalW) showing conservation of LPIN1 amino acids A353, R552, and G582. Straight lines indicate hidden sequence.

G582R (G>A) A R552K (G>A)

HAD domain A353T (G>A) 678-682

1 114153-158 674 830 NLIP NLS CLIP 1 890

I184I (C>T) V494M (G>A) P610S (C>T)

S232S (G>C) G750G (G>C)

Homo sapiens Pan troglodytes Mus musculus Rattus norvegicus Xenopus tropicalus Gallus gallus Monodelphis domestica Danio rerio

LPIN1, insulin resistance and adiposity

FIG. 3. A) A family pedigree demonstrating that the A353T mutation does not segregate with disease in a fully penetrant manner. +/- represents a heterozygous genotype and +/+ represents the wild-type genotype. The patient (indicated by the arrow) has hyperinsulinemia (diagonal stripes), hirsutism (spots) and acanthosis nigricans (dashes). The grandfather has diabetes (diamonds). There is no fasting insulin data for the grandmother or the youngest maternal aunt. B) A family pedigree demonstrating that the G582R mutation does not segregate with disease in a fully penetrant manner. +/- represents a heterozygous genotype and +/+ represents the wild- type genotype. The patient (indicated by the arrow) has hyperinsulinemia (diagonal stripes), severe peripheral neuropathy (black), previous bone marrow transplant for AML (horizontal stripes) and an intracerebral cavernous haemangioma (vertical stripes). His father has diabetes (diamonds).

+/+ +/-

+/- +/- +/+

+/-

+/- +/+

+/++/+ +/-