Introduction
Insulin resistance is a common consequence of obesity, and is frequently a prelude to frank type 2 diabetes [
1]. Uncommonly, severe insulin resistance is seen in lean people, often presenting with features of ovarian hyperandrogenism and ovulatory dysfunction in women before hyperglycaemia ensues [
2,
3]. In most cases, such severe insulin resistance is believed to have a genetic basis, and around 10–20% of such patients have loss-of-function mutations in the insulin receptor
INSR gene [
4].
Since the first description of
INSR mutations in patients with severe insulin resistance in 1988 [
5,
6], nearly 100 allelic variants have been described [
4], and many of these have been characterised in vitro. Such studies were the basis of the classification by Taylor and colleagues in 1992 of pathogenic
INSR mutations into five classes, defined by defects in receptor expression, trafficking to the cell surface, insulin binding, signal transduction or receptor recycling, respectively [
7]. Although many mutant receptors with defective trafficking and/or insulin binding and/or signal transduction have been described, mutations predominantly affecting receptor endocytosis or recycling are rare. The best studied examples are the Lys460Glu [
8] and Ser462Asn [
9] mutations, which perturb a critical hinge region of the insulin binding domain of the receptor [
10], leading to impaired dissociation of bound insulin in acidic conditions [
8,
9], and the Arg252Cys mutation, which impairs receptor endocytosis [
11].
Genetic defects in the insulin receptor produce a clinical spectrum of abnormality. Biallelic mutations most commonly produce Donohue or Rabson–Mendenhall syndromes, disorders with marked impairment of linear growth, soft tissue overgrowth and extreme metabolic derangements that lead to mortality between infancy and early adulthood [
2,
3]. The less severe end of the spectrum comprises severe insulin resistance usually presenting peripubertally in females with hyperandrogenism, oligomenorrhoea and acanthosis nigricans [
2,
3]. Heterozygous mutations in the intracellular domain of the insulin receptor are most commonly found in this setting, transmitting disease with autosomal dominant inheritance, but a minority of such patients have milder, biallelic mutations in the extracellular domain with autosomal recessive inheritance [
4].
One reported cause of such milder, autosomal recessive disease is the p.Ile119Met mutation in the alpha subunit, reported in a single Yemeni kindred [
12,
13]. Although a case was made for the pathogenicity of the mutant based on a two point logarithm (base 10) of the odds (LOD) score derived from this single family [
12], no functional studies of the mutant in vitro or ex vivo have been reported. Since the index report we have identified four unrelated patients of Somali ancestry with the same
INSR mutation, which did not occur in any other patients with severe insulin resistance. We thus set out to assess the mechanism of loss of function of the Ile119Met mutant insulin receptor, and, given the close geographical proximity of Somalia and the Yemen in the region of the Horn of Africa, to investigate a likely geographic founder effect for this mutation.
Methods
Molecular genetic studies
Genomic DNA was extracted from peripheral blood leucocytes before PCR amplification of all exons of the INSR and 50 bp of their flanking sequences. Primer sequences are available on request. PCR products were purified using Agencourt CleanSEQ reagents (Agencourt Bioscience Corporation, Beverly, MA, USA) and sequenced bidirectionally on an ABI 3730 DNA sequencer (Applied Biosystems, Warrington, UK). Sequence analysis was performed using Softgenetics Mutation Surveyor (State College, PA, USA).
DNA from 100 unrelated, healthy Somalian immigrants to Denmark was typed for the
INSR Ile119Met variant using PCR and single base extension (SBE) [
15]. Exon 2 of
INSR was amplified by PCR as described above and treated with ExoSAP-IT (USB, Cleveland, OH, USA) to remove single stranded nucleotides and excess dNTPs. The Ile119Met mutation was typed using SBE with the SNapShot kit (Applied Biosystems). The following primer was used for the SBE (the single nucleotide polymorphisms [SNPs] are underlined): Ile119Met (reverse): GAGCTCATTGTTCTTCTC-
G/C. The SBE reaction was performed in triplex. Excess ddNTPS were removed with shrimp alkaline phosphatase before samples were run on an ABI Prism ABI3130 genetic analyser (Applied Biosystems). Allele determination was undertaken using Genescan 3.7 and Genotyper 3.7 software (Applied Biosystems) [
15]. DNA from two patients was applied as positive control and both were homozygous for the Ile119Met mutation.
INSR sequence data from 179 individuals of different ethnic origins in the 1000 Genomes Project pilot 1 data release (April 2009, 100328;
ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/pilot_data/release/2010_03/pilot1/README_SRP000031.2010_03_snps) and exome-wide sequencing data from 354 Europid individuals (I. Barroso, unpublished data) were additionally interrogated for the presence of the p.Ile119Met and p.Arg1039X mutations.
Patients were genotyped for three microsatellite markers flanking
INSR (D19S916, D19S873, D19S216); an unannotated microsatellite consisting of a tandem AC repeat within intron 1 of the insulin receptor gene (position 19:7269388-7269437, bounded by rs67146219 and rs711177190, Ensembl assembly GRCh37, named 25AC in Table 2); and SNPs in introns 1 and 2 chosen for having a minor allele frequency ≥40% in the Yoruba population in Ibadan, Nigeria (YRI) (the International HapMap Project [
16] accessed 17 May 2010) (rs7254487, rs6210958, rs7248939, rs11671297, rs4804433, rs3852876, rs57930737, and rs4499341). Microsatellites were amplified using GoTaq Green chemistry (Promega, Madison, WI, USA) according to the manufacturer’s instructions using a FAM fluorescent tag on the reverse primer. Products were diluted and mixed with GeneScan 500LIZ Size Standard (Applied Biosystems) and run on an ABI 3730 DNA Analyser (Applied Biosystems). Allele determination was undertaken using GeneMapper software (Applied Biosystems). For SNP genotyping, DNA was amplified using GoTaq Green chemistry, sequenced using a BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems), run on the ABI 3730 DNA Analyser, and genotype calls were made using Sequencher software (Gene Codes Corporation, Ann Arbor, MI, USA). Primer sequences are available upon request.
Insulin dissociation assays
For insulin dissociation studies, 60,000 CHO cells were seeded per well of five 24 well plates and transfected with 400 ng of the appropriate plasmid using polyethylimine, 12 h after seeding, before dissociation studies, when cells were at 70–90% confluence.
To assess the insulin dissociation rate from heterologously expressed receptors, plates were placed on ice, before aspiration of medium and washing twice with PBS. 150 μl of binding buffer (120 mmol/l NaCl, 2.5 mmol/l KCl, 15 mmol/l Na acetate, 1 mmol/l EDTA, 1.2 mmol/l MgSO4, 10 mmol/l glucose, 1% (wt/vol). BSA, 50 mmol/l HEPES, pH 7.8) was added to all wells, followed by 100 μl binding buffer with or without cold insulin as appropriate. To the relevant wells 50 μl of 125I-labelled insulin (10–15,000 cpm) was then added before incubation at 4°C with shaking for 4–6 h. Plates were then put on ice, buffer was aspirated and cells were washed twice with ice cold PBS. Cells in one plate were then solubilised in 200 μl 0.03% (wt/vol.) SDS for 30 min at room temperature before counting in a γ-counter to determine insulin binding prior to dissociation.
For cooperativity studies, 250 μl of dissociation buffer (120 mmol/l NaCl, 4.5 mmol/l KCl, 1 mmol/l EDTA, 1.2 mmol/l MgSO4, 10 mmol/l glucose, 1% [wt/vol.] BSA, and 50 mmol/l HEPES, pH 7.8) was added to each well with additional cold insulin to final concentrations of 0, 0.1 and 10 μmol/l as appropriate, prior to incubation for 5 min at 4°C. Medium was then removed for counting before washing of cells with ice cold PBS and solubilisation and counting as before. For studies of the pH dependency of insulin dissociation buffers containing 120 mmol/l NaCl, 4.5 mmol/l KCl, 1 mmol/l EDTA, 1.2 mmol/l MgSO4, 10 mmol/l glucose, 1% (wt/vol.) BSA, and either 100 mmol/l MES (for pH 5.5), ACES (for pH 6.2), HEPES (for pH 7.0) or TAPS (for pH 7.8) were used, and no additional cold insulin was added during dissociation. All chemical reagents were purchased from Sigma unless otherwise indicated.
Discussion
Over the past 22 years, since the index reports of loss-of-function mutations in the human insulin receptor associated with extreme insulin resistance, more than 100 pathogenic mutations have been described [
4]. Relatively few recurring mutations have been reported, although in our own practice we have identified the Pro1178Leu variant in six unrelated Europid patients, adding to the index Japanese report, and the Pro193Leu variant in unrelated Saudi and Afghan families. No geographical founder effects have been demonstrated. Thus our finding of the p.Ile119Met variant in five unrelated patients, all of whom trace ancestry to Somalia or the Yemen, is striking. Indeed, this mutation has been found in all the Somali patients referred to us with extreme insulin resistance, but in no other patient. On these grounds alone, we believe this represents a founder effect in the region of the Horn of Africa. Although we sought to corroborate this by comparing the haplotype around the mutation between patients, lack of variability in markers around the mutation locus meant that we could only establish an upper size limit of 28 kb for the putative ancestral haplotype, suggesting that any such founder effect must be ancient.
The previous report of the p.Ile119Met mutation associated with extreme insulin resistance included determination of a LOD score of between 3.4 and 4.6, depending on the allele frequency in the population. Our demonstration here, that none of 100 Somali controls carries the mutation, suggests that the higher end of this range is appropriate, providing strong genetic evidence for the pathogenicity of the mutation. We have now determined that patients homozygous for the Ile119Met variant not only have extreme insulin resistance, but also have a biochemical phenotype—extreme hyperinsulinaemia, high adiponectin, SHBG and IGFBP1, and normal lipid profile—which we have found to be pathognomonic for insulin receptor dysfunction [
19,
20]. We thus believe that the collective clinical, genetic and biochemical evidence that the Ile119Met
INSR allele severely impairs insulin receptor function in vivo is incontrovertible.
Nevertheless, there have been no reports to date of in vitro characterisation of the Ile119Met mutant receptor. We now report such studies, and surprisingly the in vitro defect is very subtle. Most alpha subunit mutations that have been studied to date either impair proreceptor processing and thus cell surface expression of mature receptors, or severely affect insulin binding to the receptor [
4,
7]. However, insulin binding in both primary cells from an Ile119Met
INSR homozygous patient, and in cells heterologously expressing the mutant receptor, showed normal steady state insulin binding, although there was very subtle evidence of aberrant processing of the proreceptor in both situations, and insulin-stimulated receptor autophosphorylation was normal.
Study of the Lys460Glu and Ser462Asn mutant receptors has previously shown that some α subunit mutations result in impaired cooperativity in insulin binding, or in altered pH-dependence of insulin dissociation [
8,
9]. In the case of the Ile119Met mutant, insulin dissociation was slightly slower at baseline than that from wild-type receptor; however both positive and negative cooperativity were preserved. The impaired dissociation rate was accentuated, however, under acidic conditions. This would be predicted to favour degradation of insulin-receptor complexes in acidified endosomes after internalisation in vivo, and reduce the amount of receptor recycling to the cell surface, which requires insulin dissociation. Substitution of Ile119 by methionine is predicted to alter the packing within the barrel of the L1 domain only modestly, potentially leading to compensation in the sheet structure itself. However Ile119 is neighboured by Arg118 and Glu120, both of which have recently been shown to play a key role in creating so-called site 1 for insulin binding to the receptor through formation of a charge compensating cluster with residues Glu698 and Arg702from the C terminal alpha helix of the alpha subunit [
25]. In this key structural context it is plausible that substitution of Ile119 for methionine may perturb some aspect of the dynamic interaction of the alpha subunit with insulin. Nevertheless, our functional studies are not comprehensive, and it remains possible that a further, unstudied aspect of insulin receptor function is selectively affected by the Ile119Met mutation, or that this mutation may have a greater deleterious effect in the context of the longer ‘B’ isoform of the insulin receptor that includes the small exon 11. In particular, it would be of interest to study surface distribution of the insulin receptor and insulin-dependent receptor internalisation, processes that have been shown to be impaired by another alpha subunit mutation, albeit one affecting the cysteine-rich domain rather than the L1 domain [
11].
The current wealth of genome-wide association studies revealing statistical associations between disease traits and common genetic variants has led to keen interest in trying to assign specific functional consequences to each such variant. In this context it is sobering to reflect that, despite the compelling case for defective insulin receptor function in the patients studied, and the wealth of understanding of insulin receptor function, it has proved extremely difficult to demonstrate a defect in insulin signalling in vitro commensurate with the clinical severity of the phenotype. This emphasizes the challenges that lie ahead in translating pure genetic associations into cellular disease mechanism.
In summary, we provide evidence that the Ile119Met INSR mutation is a loss-of-function mutation that causes autosomal recessive extreme insulin resistance, with a strong founder effect in the region of the Horn of Africa. We thus suggest that any patient of Yemeni or Somali origin who presents with extreme insulin resistance should be screened first for this allele. The Ile119Met insulin receptor shows normal steady state insulin binding and autophosphorylation in response to insulin, but mildly perturbed proreceptor processing and insulin dissociation, particularly in acidic conditions.
Acknowledgements
This work was funded by the Wellcome Trust (E. Raffan, Clinical Research Training Fellowship 087678/Z/08/Z; I. Barroso, 077016/Z/05/Z; R.K. Semple, Intermediate Clinical Fellowship 080952/Z/06/Z; S. O‘Rahilly, Programme Grant 078986/Z/06/Z), the UK NIHR Cambridge Biomedical Research Centre, and the UK Medical Research Council Centre for Obesity and Related Metabolic Diseases.