The Gene INPPL1, Encoding the Lipid Phosphatase SHIP2, Is a Candidate for Type 2 Diabetes In Rat and Man

Fluorescent in situ hybridization and radiation hybrids mapping were performed as described (6). The following forward and reverse primers were used to amplify a 140-bp DNA fragment of the rat SHIP2 gene from hybrid DNA: 5′-CCAGGGGTGAAAGTTTTGAG-3′ and 5′-CCTGACCCTGGGCCTAAAAG-3′.

Consent was obtained from all subjects after the nature of the procedure was explained, and the investigation was conducted according to the principles expressed in the Declaration of Helsinki. All diabetic subjects were >35 years of age at diagnosis and met the World Health Organization’s criteria defining diabetes status. The control subjects were randomly and anonymously chosen in a DNA library isolated from a large population of women consulting for a genetic diagnosis of mutation in the CFTR gene. The SHIP2 cDNA and gene sequences were obtained after PCR amplification. The sequencing products were run on an Applied Biosystem sequencer.

CHO cells expressing the human insulin receptor were transfected as described (5), with the pcDNA3 vector (Invitrogen) or with the same vector containing either the wild-type or the R1142C-mutated hemagglutinin A (HA)–tagged SHIP2 cDNA (nucleotides 264–4360 from AF162781). After 2 days, cells were stimulated with 10 nmol/l insulin during 2 min and then lysed. The supernatant was assayed for Akt/ protein kinase B (PKB) and mitogen-activated protein (MAP) kinase activities as described (5). Each activity is calculated from triplicate ± SE.

Rat, mouse, simian virus 40 (SV40), and human 3′UTR were amplified by PCR and subcloned in pGEM-T easy. A DNA fragment containing the SV40 promoter and luciferase cDNA was excised from the pGL3 vector and introduced into the above plasmids. Each plasmid (0.8 μg/10 cm² well) was co-transfected in triplicate with 0.2 μg pSV–β-galactosidase transfection efficiency plasmid. After 24 h (human 293 cells) or 48 h (rat L6 myoblasts), the cells were lysed and processed either for luciferase activity according to manufacturer’s instructions (Promega) or for total RNA extraction. The amount of β galactosidase present in the lysate was determined by enzyme-linked immunoassay (Boehringer Mannheim). Total RNA (1 μg/lane) was loaded on a 1% agarose gel and transferred to a nylon membrane. A 1.6-kb fragment of the luciferase cDNA excised from the pGL3 vector was used as a radiolabeled probe. Quantification of the signals was performed on a phosphorimager.

We first localized the SHIP2 gene on rat chromosomes because chromosomal regions suspected to contain one or several genes predisposing to type 2 diabetes or insulin resistance have been identified using rat polygenic models (7–9). Fluorescent in situ hybridization (FISH) and radiation hybrid mapping revealed that the rat SHIP2 gene is localized on the long arm of chromosome 1, at q33-36, between markers D1Got150 and D1Mit4 (Fig. 1A). It co-localizes with Nidd/gk1 quantitative trait locus (QTL), a QTL for glucose homeostasis identified in the diabetic GK rat strain (8). SHIP2 mRNA expression was investigated by Northern blotting in skeletal muscles from GK rats, but no significant difference was observed when compared with the diabetic and obese Zucker-fatty or control Brown-Norway strains (data not shown). Sequencing of the complete SHIP2 cDNA coding region in the above three rat strains revealed the presence of a unique mutation substituting an arginine to a cysteine at amino acid 1142 in the GK rat, within the proline-rich region of the protein (Fig. 1B, Table 1). The R1142C mutation identified in the GK rat was also present in the insulino-resistant and spontaneously hypertensive (SHR) rat, but not in 17 other rat strains tested (Table 1). This arginine is conserved in SHIP1 and SHIP2 proteins and might play an important functional role. Indeed, the surrounding amino acids match the consensus sequence for SH3 domain class II ligand (Table 1) (10,11). It has been shown that the conserved arginine present in class II ligands interacts with the third specificity pocket of the SH3 domain and that substitution by alanine results in decreased affinity and specificity (12). The COOH-terminal region of SHIP1 and SHIP2 is known to mediate interactions with protein partners and/or plasma membrane translocation (13–15).

To test the effects of the R1142C mutation on SHIP2 function in insulin signaling, wild-type and mutated SHIP2 cDNAs were introduced in CHO cells expressing the human insulin receptor (CHO-IR) by transfection. Because PKB/Akt and MAP kinase (ERK-2) are two signaling molecules downstream of the insulin receptor, their activities were analyzed in CHO-IR cells after hormone stimulation. Expression of wild-type or mutated SHIP2 cDNAs in CHO-IR cells resulted in decreased PKB and MAP kinase activities after insulin stimulation, as compared with vector-transfected CHO-IR cells (Fig. 1C). However, a slightly but significantly more important decrease was observed when the mutated protein was expressed.

Altogether, these data indicate that the R1142C mutation specifically identified in GK and spontaneously hypertensive (SHR)rats slightly impairs insulin signaling and may contribute to the genetic predisposition to type 2 diabetes and/or insulin resistance in these rat strains.

The presence of a relevant mutation associated with diabetes and/or insulino-resistance in two rat strains led us to investigate the human SHIP2 gene. The SHIP2 cDNA from eight unrelated Caucasian subjects with type 2 diabetes and from four control subjects of the same origin was sequenced. No difference was detected in the SHIP2 coding region. To our surprise, one of the type 2 diabetic subjects (SMH10),but none of the control subjects, exhibited a heterozygositic deletion in the proximal part of the SHIP2 3'UTR (Fig. 2A). The 16-bp deleted sequence contains an ATTTA pentamer potentially belonging to an adenylate/uridylate-rich element (ARE) (Figs. 2B and C). AREs are conserved sequence elements implicated in the regulation of messenger RNA (mRNA) stability and translation efficiency (16,17). Sequence comparisonbetween mouse, rat, and human SHIP2 3'UTR identified ATTTA pentamers or related sequences, like ATTTTA hexamer, embedded in very conserved regions (Figs. 2B and C). In the rat and mouse 3'UTR, two pentamers and one hexamer sequences were detected (Fig. 2B). In man, a 16-bp fragment containing the first ATTTA pentamer was exactly duplicated, as compared with rat and mouse, and a total of three pentamers and one hexamer were detected (Figs. 2B and C). In addition, a very conserved thymidylate was mutated in adenylate in the hexamer, potentially affecting its function (Fig. 2B). To test the capacity of the SHIP2 rat,mouse, and human 3'UTR to regulate protein expression, plasmids containing the SV40 promoter followed by the luciferase cDNA and one of the 3'UTR described above were constructed and introduced in rat L6 myoblasts by transfection (Fig. 3A). The SV40 3'UTR was used as control, because it does not contain any AREs. The presence of the SV40 3'UTR downstream of the luciferase cDNA resulted in a 10- to 70-fold higher luciferase activity, as compared with SHIP2 rat, mouse, and human 3'UTR. In rat L6 cells, the lowest luciferase activity was detected with the rat 3'UTR, followed by the mouse and human 3'UTRs. A 4.5- and 7.7-fold increase in luciferase activity were respectively observed with these 3'UTR, as compared with the rat 3'UTR (Fig. 3A). Altogether, these results suggest that the AREs present in the SHIP2 3'UTR are functional and lead to decreased expression of the reportergene. Thus, mutations in the SHIP2 3'UTR sequence could affect ARE function, SHIP2 expression, and insulin sensitivity, contributing to type 2 diabetes pathogenesis. To test the capacity of the mutated 3'UTR found in SMH10 type 2 diabetic subjects to regulate the gene reporter expression, the wild-type and Δ16 SHIP2 3'UTR were placed downstream of the luciferase cDNA, and the plasmids were introduced in human 293 cells by transfection (Fig. 3B). The presence of the Δ 16 mutation resulted in a fourfold increase in luciferase activity (range 2.8–5.8; mean 4.0; n = 4; P < 10^-4 by Student's t test), as compared with the wild-type 3'UTR. The increased luciferase activity seen with the Δ16 mutant was not dependent on the presence of insulin: a similar ratio of the luciferase activities found with the wild-type and the Δ 16 SHIP2 3'UTR was observed when human 293 cells were cultured with or without 100 nmol/l insulin (data not shown). The increased luciferase activity observed with the Δ16 mutant was associated with an equal increase in the luciferase mRNA level (Fig. 3C). Indeed, quantification of the luciferase mRNA signals found in transfected human 293 cells revealed a 4.5-fold increased level with the Δ16 mutant, as compared with the wild-type 3'UTR (range 3.4–5.1; mean 4.3; n = 3; P < 0.02 by Student's t test). Our results suggest that the Δ16 deletion identified in the SHIP2 3'UTR of one type 2 diabetic subject results in decreased ARE function and increased mRNA and protein expressions, which may result in decreased insulin signaling, i.e., insulin resistance in vivo. We next determined the accurate frequency of the Δ16 mutated allele inCaucasian populations of type 2 diabetic and control subjects. DNA samples from type 2 diabetic subjects originating from the U.K. (n = 246) and Belgium (n = 169), as well as from anonymous healthy individuals collected by the Genetics Department of the Erasme Hospital in Brussels (n = 567), were included in the experiment. A total of 9 Δ16 mutations (5 in the Belgiancohort and 4 in the English cohort) were identified in 415 unrelated type 2 diabetic subjects (frequency 2.16%), whereas only 3 mutations were identified in 567 control subjects (frequency 0.52%). Statistical analysis using the Pearson Χ² test revealed a significant association between the Δ16 mutation and type 2 diabetes (P =0.021). The nine diabetic subjects with a Δ16 mutation had a typical type 2 diabetes history: the age at diagnosis was always >40 years (range 44–67) and their treatment did not required insulin injections. BMI ranged from 23.9 to 38.1 kg/m². Five of the nine diabetic subjects had a familial history of type 2 diabetes,and seven of nine were hypertensive.