Identification of differentially expressed genes in skeletal muscle of non-diabetic insulin-resistant and insulin-sensitive Pima Indians by differential display PCR
Whole body insulin resistance results largely from impaired insulin-stimulated glucose disposal into skeletal muscle. We carried out muscle gene expression profiling to identify differentially expressed genes associated with insulin resistance.
Skeletal muscle total RNA samples from six pairs of non-diabetic insulin-resistant and insulin-sensitive Pima Indians matched for percent body fat were analyzed by DDPCR with 90 primer combinations. The mRNA expression concentrations of selected 13 known genes and four expressed sequences tags were measured by quantitative real-time RT-PCR in 50 non-diabetic Pima subjects.
From over 6500 displayed DDPCR cDNA bands, 36 of the most differentially expressed cDNAs were identified, revealing 29 unique sequences: 16 known genes, 10 expressed sequences tags and three unknown transcripts. Multiple regression analyses indicated that whole body insulin-mediated glucose disposal rates of the subjects, independent of age, sex, and percent body fat, were negatively correlated with mRNA concentrations of an EST (DD23; r=−0.38, p=0.007), ATP1A2 (r=−0.27, p=0.05), MAP2K4 (r=−0.34, p=0.02), and PRPSAP1 (r=−0.37, p=0.008). Transcript concentrations of DD23 (r=0.27, p=0.05) and MTND4 (r=−0.29, p=0.05) were correlated with plasma insulin concentration, independent of age, sex, and percent body fat.
Altered expression concentrations of these genes might be causes or consequences of insulin resistance, and these genes serve as candidate susceptibility genes for insulin resistance.
KeywordsPima Indians insulin resistance Type 2 diabetes skeletal muscle gene expression differential display real-time PCR
differential display PCR
People with Type 2 diabetes are characterized by decreased action of insulin to mediate glucose disposal (insulin resistance), obesity, relatively reduced early insulin secretion, and excess rates of endogenous glucose production . Prospective studies in Pima Indians of Arizona, a population with the highest reported prevalence of Type 2 diabetes , indicate that the first three of these metabolic characteristics are predictive of the disease .
Insulin resistance resulting from obesity contributes to the relationship between obesity and Type 2 diabetes [4, 5]. Nevertheless, there is considerable variance in insulin action in vivo at any given degree of adiposity , and insulin resistance predicts the development of Type 2 diabetes independent of obesity in this population . This obesity-independent variance in insulin action in vivo has considerable familial aggregation , indicative of possible primary genetic determinants. In contrast, insulin elicits many biological responses in various cell types, and hyperinsulinaemia associated with insulin resistance results in secondary effects. Both the primary (etiologic) genetic determinants and secondary effects of insulin resistance could be reflected in dysregulated transcription of genes contributing to the development and/or complication of the condition.
Since most of the insulin-mediated glucose uptake in the whole body is determined by insulin action on skeletal muscle , we compared concentrations of gene expression using differential display PCR (DDPCR) in skeletal muscle tissue samples from paired equally obese, insulin-sensitive and non-diabetic insulin-resistant Pima Indians to determine the transcripts that are over- or under-expressed in insulin-resistant subjects. This approach complements the global microarray profiling method on non-paired samples from a different set of insulin-sensitive and insulin-resistant Pima Indian subjects . DDPCR uses random primers that could show the expression of transcripts that are not captured by the a priori design of the microarrays with defined sets of genes to interrogate.
Subjects and methods
Subjects and clinical procedures
This study was approved by the Tribal Council of the Gila River Indian Community and the Institutional Review Board of the National Institutes of Diabetes and Digestive and Kidney Diseases. All subjects provided written informed consent as participants in a longitudinal study of the development of Type 2 diabetes. They were in good health as determined by medical history, physical examination and routine blood and urine tests, and were not taking any prescribed or over-the-counter medication. Even though physical activity status was not measured in this study, the subjects were generally physically inactive.
Clinical characteristics of six pairs of insulin-sensitive (IS) and insulin-resistant (IR) Pima Indian subjects for differential display experiment
Body fat (%)
Fasting glucose (mmol/l)
2-h glucosea (mmol/l)
Fasting insulin (pmol/l)
2-h insulina (pmol/l)
Means ± SD
Clinical characteristics of the 50 non-diabetic Pima Indian subjects for real-time RT-PCR and genotyping experiments
Body fat (%)
Fasting glucose (mmol/l)
2-h glucose (mmol/l)
Fasting insulin (pmol/l)
2-h insulin (pmol/l)
Volunteers were admitted to the Clinical Research Unit for 8 to 10 days. They were fed a weight-maintaining diet (containing 50% of calories as carbohydrates, 30% as fat, and 20% as protein) for 2 to 3 days prior to metabolic testing. Body composition was measured by dual energy X-ray absorptiometry (DEXA) using a total body scanner (DPX-L, Lunar Radiation, Madison, Wis., USA) . A 3-h 75 g OGTT was carried out after a 12-h overnight fast, and diabetes was defined according to the WHO criteria . Insulin action was measured at physiologic and maximally stimulating insulin concentrations during a two-step hyperinsulinaemic-euglycaemic glucose clamp . After an overnight fast, a primed continuous intravenous insulin infusion was administered for 100 min at a constant rate of 40 mU·m−2 body surface area·min−1 (M-low), followed by a second insulin infusion for 100 min at 400 mU·m−2·min−1 (M-high). These infusions achieved steady-state plasma insulin concentrations of 840±250 pmol/l and 13,320±3480 pmol/l (Means±SD), respectively. Plasma glucose concentrations were maintained at 5.6 mmol/l with a variable infusion of a 20% glucose solution. Rates of insulin-stimulated glucose disposal at physiologic (M-low) and maximally stimulating (M-high) insulin concentrations were calculated for the last 40 min of each phase, and corrected for endogenous glucose output (EGO) . During the low dose and baseline, EGO was calculated using a primed (1.11×106 Bq·min−1), continuous (1.11×104 Bq·min−1) 3-3H-glucose infusion [3, 13]; during the high insulin dose, EGO was assumed to be zero. The glucose disposal rates during the clamp served as the selection criteria for insulin-sensitive or -resistant group. Subjects of similar percent body fat with highest or lowest glucose disposal rates were classified into the IS or IR group, respectively. All measurements derived from the glucose clamp were normalized to estimated metabolic body size (EMBS, which equals fat-free mass +17.7 kg) .
Percutaneous needle biopsies were carried out on the vastus lateralis muscle under local anesthesia with 1% lidocaine after a 12-h overnight fast, and the biopsy specimens were immediately frozen in liquid nitrogen. Total RNA was isolated from the frozen tissues homogenized in TRIzol Reagent (Life Technologies, Gaithersburg, Mass., USA).
Differential Display PCR
Differential display  was carried out as described in the Delta Differential Display kit user manual (Clontech, Palo Alto, Calif., USA). For the first screen, we used pooled equal quantities of total RNA from the IS vs the IR groups. Briefly, 4 µg of total RNA was reverse-transcribed in 20 µl reaction containing 1 mmol/l dNTP, 0.1 µmol/l oligo (dT) primers and 400 U of MMLV reverse transcriptase. Reverse transcription was carried out for 60 min at 42°C with a final denaturation step at 75°C for 10 min. Aliquots of 4 µl and 16 µl from the resulting cDNA sample were diluted to 1:10 and 1:40. A 1 µl of cDNA sample from each dilution was then amplified by PCR in a 20 µl of total PCR reaction mixture including Advantage Klen Taq DNA polymerase (Clontech) and [33P] dATP (PerkinElmer Life Sciences, Boston, Mass., USA). We used one of nine anchored oligo(dT) primers and one of 10 arbitrary upstream primers provided in the kit, resulting in 90 primer combinations in total.
The cycling parameters were one cycle at 94°C for 5 min, 40°C for 5 min, 68°C for 5 min, two cycles at 94°C for 30 s, 40°C for 30 s, 68°C for 5 min, 23 cycles at 94°C for 20 s, 60°C for 30 s, 68°C for 2 min and a final extension at 68°C for 7 min. The PCR products were separated on a polyacrylamide sequencing gel, SequaGel XR (National Diagnostics, Atlanta, Ga., USA). The gels were dried and exposed to Biomax MR film (Kodak, Rochester, N.Y., USA). Amplified cDNA bands with visually different signal intensities between the pooled IS vs IR samples were verified by differential display method in the six subject pairs individually. In this second screen, we determined a cDNA band to be differentially expressed if the signal intensity of that band visually differed in three or more of the six subject pairs.
Reamplification, cloning and sequencing
Differentially expressed PCR-bands were excised from dried gels, rehydrated in 40 µl H2O by boiling for 10 min and the eluted cDNA was reamplified by PCR using the same primer combination for 20 cycles in the presence of [33P] dATP. The reamplified cDNAs were separated again on a sequencing gel. After the second band recovery procedure, the reamplified cDNAs were cloned into the pT-Adv vector using the AdvanTage PCR Cloning kit (Clontech, Palo Alto, Calif., USA). Cloned cDNA fragments were sequenced with an ABI 377 or 3700 DNA sequencer using vector-specific T7 and M13 reverse primers and an ABI PRISM dye terminator cycle sequencing ready reaction kit (Applied Biosystems, Foster City, Calif., USA). We sequenced 10 to 30 representative clones of each transformed cDNA and the sequences were compared with the National Center of Biotechnology Information sequence database using the BLAST program.
Quantitative real-time RT-PCR
Probes and primers used in quantitative real-time RT-PCR
Statistical analyses were carried out using the statistical analysis system of the SAS Institute (Cary, N.C., USA). For continuous variables, general linear regression models were used to adjust for the effects of age, sex, and percent body fat. Plasma insulin concentrations and rates of glucose disappearance during the low-dose insulin infusion were log transformed to approximate normal distribution. The mRNA expression concentrations of the genes were normalized to that of human cyclophilin by taking their residuals from linear regression analysis. A p value less than or equal to 0.05 (unadjusted for multiple testing) was considered statistically significant.
Differential Display PCR
Differentially expressed genes with known functions identified by DD-PCR in six pairs of non-diabetic insulin-resistant and insulin-sensitive Pima Indians
ATP1A2 (ATPase, Na+/K+ transporting, alpha 2 (+) polypeptide)
Small molecule transport
ATP5A1 (ATP synthase, H+ transporting, mitochondrial F1 complex, alpha subunit, isoform 1)
HBB (haemoglobin, beta)
Small molecule transport
MAP2K4 (mitogen-activated protein kinase kinase 4)
MBNL (muscleblind (Drosophila)-like)
MTCO2 (cytochrome c oxidase II)
MTCYB (cytochrome b)
MTND3 (NADH dehydrogenase 3)
MTND4 (NADH dehydrogenase 4)
MYBPC1 (myosin-binding protein C, slow-type)
PRPSAP1 (phosphoribosyl pyrophosphate synthetase-associated protein 1)
Lipid, fatty-acid and sterol metabolism; Small molecule transport
RY-1 (putative nucleic acid binding protein)
TNNC1 (troponin C, slow)
TPI1 (triosephosphate isomerase 1)
Carbohydrate metabolism; Energy generation; Differentiation
FLJ10326 (hypothetical protein FLJ10326, Member of the class I tRNA synthetase family)
Extension of selected DDPCR results in a larger sample set
We sought to extend the DDPCR results using Real Time RT-PCR in a larger sample set of 50 non-diabetic Pima subjects including 8 of the 12 original samples used for the differential display. Out of the 16 known genes with differential expression as assessed by DDPCR, we chose to validate the expression concentrations of 13 genes that were expressed in muscle at least as abundantly as in some other tissues according to GeneCard database (http://nciarray.nci.nih.gov/cards/). The 13 selected genes were ATP1A2, ATP5A1, TNNC1, TPI1, PSAP, MBNL, MYBPC1, MTCYB, PRPSAP1, MAP2K4, MTND3, MTND4, and FLJ10326. We excluded MTCO2 and RY-1, since their differentially expressed cDNA bands were due to deletion mutation, and HBB that was most highly expressed in bone marrow. Out of the 13 differentially expressed ESTs in DDPCR, we further investigated four ESTs with the biggest visual differences between the IR and the IS pairs. GeneBank accession numbers of the four ESTs (DD11, DD14, DD23, and DD24) are AI204632, BE173380, BG436275, and W58230, respectively.
Significant correlations of five skeletal muscle mRNA transcripts with adjusted clinical parameters of 50 non-diabetic Pima Indians
Our study revealed differentially expressed genes in skeletal muscle of insulin-resistant human subjects that could constitute primary causes or secondary effects of this condition. Identification of abnormal skeletal muscle gene expression associated with insulin resistance is important in elucidating the pathogenesis of this condition. Altered gene expression in insulin resistance can be considered molecular markers that could lead to a better understanding of the functions of the genes in the progressive metabolic impairment. To this end, we applied mRNA Differential Display PCR techniques to identify differentially expressed genes in skeletal muscle tissues of six pairs of non-diabetic insulin-resistant and insulin-sensitive Pima Indians. To eliminate potential confounding complications from diabetes, we selected non-diabetic subjects with the biggest difference of glucose disposal rates as a measure of insulin sensitivity. Apart from being matched for sex, the paired subjects were also matched for percent body fat to minimize differences resulting from obesity. The effort to match each subject pair for percent body fat, especially in the Pima population that is generally obese, resulted in a collectively small (24%), though significant (p=0.03), difference of glucose disposal rate at physiological concentration of insulin during the clamp. Nevertheless, the collective average differences in other measures of insulin sensitivity were bigger and significant: 39% (p=0.002) for glucose disposal rates at maximally stimulating insulin concentration as well as 29% (p=0.02) and 64% (p=0.002) for fasting and 2-h plasma insulin concentrations post oral glucose bolus intake, respectively.
The DDPCR approach revealed 29 unique genes that are most differentially expressed in these samples; 45% of these transcripts are ESTs of unknown genes. The identified known genes are involved in energy metabolism, signal transduction, and cell structure. Verification of differential gene expression after mRNA Differential Display PCR is crucial to minimize false-positive bands. To do this, we used quantitative real-time PCR that required relatively low quantities of RNA suitable for our study involving limited samples of human tissues. In skeletal muscle tissues of 50 non-diabetic Pima subjects, including eight subjects from the differential display experiment, we analyzed mRNA expression concentrations of selected 17 genes and ESTs. Multiple regression analyses showed that the individual mRNA expression concentrations of five genes (DD23, ATP1A2, PRPSAP1, MAP2K4, and MTND4) are associated with several in vivo measures of insulin resistance of the donors [glucose disposal rates at physiologic (M-low) and maximally stimulating (M-high) insulin concentrations during clamp, as well as fasting plasma insulin concentration] independent of age, sex, and percent body fat. Some of the correlations between the gene expression concentrations and M-low were not present for the M-high correlations, presumably because the correlations were modest, and also because M-low and M-high measure slightly different aspects of insulin action in vivo. The validation rate of 29% might reflect not only the methodological differences between DDPCR and Real Time PCR, but also grouping differences between the six paired subjects for DDCPR and the extension sample set of 50 subjects.
This study was conducted to complement our previous microarray data on skeletal muscle gene expression in non-diabetic insulin-resistant compared with insulin-sensitive Pima Indians using microarray . Out of the 13 known genes differentially expressed in the DDPCR experiment, five were not arrayed on the Hu6800 arrays used in the previous study. In fact, one of the five genes, MTND4, turned out to be correlated to fasting plasma insulin in the extension sample set of this study. This result showed the utility of DDPCR in complementing microarrays that have defined sets of genes/ESTs to interrogate. Nevertheless, the other genes that correlated with measures of insulin sensitivity in the extension sample set of this study were not differentially expressed in the microarray data. Closer analysis showed that the microarray expression concentrations (average difference) of probe sets interrogating all of these genes, except ATP1A2, were below reliable detection level. The failure of microarray to identify the moderately abundant ATP1A2 muscle transcript as differentially expressed might represent a false negative case, which proved to be positive using DDPCR. Despite these advantages of DDPCR, the procedure is more time consuming than microarray method, as also reported by another study , and could be less amiable to technical improvement than microarrays.
One of the validated genes with differential expression in our study, DD23, represents a novel susceptibility gene for insulin resistance in muscle; increased transcript concentration of this EST correlates not only with insulin resistance measured as lower glucose disposal rates during hyperinsulinaemic euglycaemic clamp, but also with higher fasting plasma concentration of insulin. A current database search (as of 11/02) indicated that DD23 shares homology with a few ESTs (accession numbers BG542139, AW298235, AI379723), although none of these ESTs correspond to any known gene. DD23 does not seem to code for any peptide homologous to any known proteins. An in silico search placed the gene encoding DD23 on chromosome 3p, which has suggestive linkages to fasting plasma insulin concentration in Pima Indians . This chromosomal region also showed linkage to fasting C-peptide/glucose concentration in the FUSION study . Taken together, these results warrant further investigation on the identity of DD23 and its role in insulin sensitivity.
Insulin sensitivity was also negatively associated with skeletal muscle mRNA concentrations of ATP1A2 independent of age, sex, and percent body fat. The ATP1A2 gene encodes the α2 subunit of Na+/K+-ATPase, which is specifically expressed in skeletal muscle, heart and fat tissues . Several isoforms of Na+/K+-ATPase exist from different combinations of the α and β subunits . In the streptozotocin (STZ) rat model of diabetes, increased muscle α2 subunit mRNA  and protein concentrations  with no change or decrease of enzyme activity were observed in skeletal muscle. The increase in α2 mRNA concentration was thought to be a compensatory increase due to the decreased Na+/K+-ATPase activity under hypoinsulinaemic state, which occurred in the STZ-induced diabetes or after 48-h fasting in rats . The skeletal muscle mRNA expression of ATP1A2 could respond similarly to hypoinsulinaemic and insulin-resistant states. Furthermore, α2 subunit mRNA expression was increased by long-term insulin treatment of 3T3-L1 cells [22, 25]; the treatment presumably induced insulin resistance [26, 27]. Thus, the increased skeletal muscle mRNA expression of ATP1A2 in insulin-resistant subjects of this study might be a secondary effect of insulin resistance. Nevertheless, we cannot exclude the possibility that polymorphisms in the ATP1A2 gene might be the primary cause for insulin resistance. This gene maps to chromosome 1q21–23, a region previously reported to be linked to insulin resistance and Type 2 diabetes in Pima Indians , French population , and Utah Caucasians . Thus, the ATP1A2 gene can still be considered as a candidate susceptibility gene for insulin resistance.
Our results indicate that insulin resistance is also correlated with increased skeletal muscle mitogen-activated protein kinase kinase 4 (MAP2K4). MAP2K4 is a member of mitogen-activated protein kinase (MAPK) family generally involved in the regulation of cell growth, differentiation, and survival . Insulin increases glucose uptake by initially increasing glucose transporter (GLUT4) translocation to cell surface and then further activating the transporters. This GLUT4 activation seems to be regulated by a p38 MAPK (MAPK14)-dependent signalling pathway , which acts downstream of MAP2K4 . Thus, it is possible that the increased mRNA expression of MAP2K4 in skeletal muscle of insulin-resistant subjects in this study constitutes a compensatory mechanism to counter against the insulin-resistant condition of the tissue.
Fasting plasma insulin concentration (adjusted for age, sex, and percent body fat) was correlated with skeletal muscle transcript concentration of NADH dehydrogenase 4 (MTND4). MTND4 is encoded by the guanine-rich heavy (H) strand of the mitochondrial genome and located between nucleotide pair 10760 and 12137 . It is one of subunits of NADH-ubiquinone oxidoreductase (respiratory complex I) that transports electrons from NADH to ubiquinone. We observed decreased MTND4 transcript concentration in muscle of non-diabetic insulin-resistant Pima Indians, whereas it was previously reported to be upregulated in skeletal muscle tissues of human subjects with Type 1 or Type 2 diabetes despite decreased mitochondrial DNA copy number . We do not have any data on mitochondrial DNA copy number in our study, but the decreased expression of MTND4 transcript concentration in non-diabetic insulin-resistant subjects could reflect lower gene copy number. One can speculate that if subsequent diabetes is uncontrolled, the mRNA expression of MTND4 and perhaps other mitochondrially encoded genes of the oxidative phosphorylation pathway are increased to compensate for the decreased copy number, resulting in increased mitochondrial respiration.
Another gene transcript increased in insulin-resistant subjects codes for phosphoribosyl pyrophosphate synthetase-associated protein 1 (PRPSAP1). This protein is a putative regulatory subunit of the phosphoribosylpyrophosphate (PRPP) synthetase complex, which catalyzes the formation of PRPP from ATP and ribose 5-phosphate. It might play a negative regulatory role in PRPP synthesis . PRPSAP1 is encoded by a gene on chromosome 17q24-q25. The mechanistic details that might have resulted in the negative correlation of skeletal mRNA expression of PRPSAP1 with insulin sensitivity still need to be further elucidated.
Distinguished banding patterns on differential display gels in our study led to the genotyping of MTCO2, RY-1 and DD34 cDNA bands. The polymorphic MTCO2 cDNA bands resulted from insertion/deletion of a 9-base pair repeat sequence in position 8271–8290 of the mitochondrial genome, which is common in Asians, Polynesian and Native Americans [37, 38, 39]. None of the genotyped polymorphisms were related to insulin sensitivity in the 50 Pima subjects. Our results point to the possibility that some cDNA products amplified by the degenerate differential display primers could appear to be differentially expressed, when, in fact, the absence of such amplicons in particular samples are due to an insertion/deletion mutation within the amplicon of differential display or to the inability of the primers to bind to the template if the primer binding sites are polymorphic.
In summary, we carried out differential display to identify differential gene expression associated with insulin resistance in non-diabetic Pima Indians. We identified 29 unique, differentially expressed transcripts and sought to validate 17 of them with quantitative real-time PCR in 50 subjects. Multiple regression analyses showed that the mRNA expression concentrations of DD23, ATP1A2, MAP2K4, MTND4, and PRPSAP1 are associated with glucose disposal rate at physiological and maximally stimulating insulin concentration during clamp and fasting plasma insulin concentration of the donors independent of age, sex, and percent body fat. These differentially expressed muscle transcript concentrations might contribute to or result from the development of insulin resistance prior to frank diabetes. Thus, the genes encoding the aforementioned transcripts serve as candidate susceptibility genes for insulin resistance.
We would like to thank the members and leaders of the Gila River Indian Community, without whose cooperation these studies would have been impossible. We gratefully acknowledge M. Milner, PA, C. Massengill, RN, the nurses of the clinical research ward, and the metabolic kitchen staff for the care of the research volunteers. We also thank G. Luczy-Bachman for technical help.
- 1.Bogardus C (1996) Metabolic abnormalities in the development of non-insulin dependent diabetes mellitus. In: LeRoith D, Taylor SI, Olefsky JM (eds) Diabetes mellitus: a fundamental and clinical text. Lippincott Williams & Wilkins, Philadelphia, pp 459–467Google Scholar
- 14.Lillioja S, Bogardus C (1988) Obesity and insulin resistance: Lessons learned from the Pima Indians. Diabet Metab Rev 4:517–540Google Scholar
- 19.Pratley RE, Thompson DB, Prochazka M et al. (1998) An autosomal genomic scan for loci linked to prediabetic phenotypes in Pima Indians. J Clin Invest 101:1757–1764Google Scholar
- 29.Vionnet N, Hani E, Dupont S et al. (2000) Genomewide search for type 2 diabetes-susceptibility genes in French whites: evidence for a novel susceptibility locus for early-onset diabetes on chromosome 3q27-qter and independent replication of a type 2-diabetes locus on chromosome 1q21-q24. Am J Hum Genet 67:1470–1480PubMedGoogle Scholar