| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Medicine (H.Z., Y.J., J.J.C., T.H., Z.Z., S.C.E.), University of Arkansas for Medical Sciences; and Endocrinology Section, Department of Medicine (S.C.E.), Central Arkansas Veterans Healthcare System, Little Rock, Arkansas 77205
Address all correspondence and requests for reprints to: Steven C. Elbein, M.D., Professor of Medicine, Central Arkansas Veterans Healthcare System, Endocrinology 111J/LR, 4300 West 7th Street, Little Rock, Arkansas 72205. E-mail: elbeinstevenc{at}uams.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Although most studies considered the GFPT protein as encoded by a single gene, a second gene was cloned recently in humans and mice (12). The two enzymes are encoded by nonallelic genes in humans, GFPT1 (chromosome 2p13) and GFPT2 (chromosome 5q34–q35). GFPT2 encodes a protein that is 75.6% homologous to GFPT1, and the mRNA sequences of GFPT1 and GFPT2 may be indistinguishable by many probes (12). Like GFPT1, GFPT2 is widely expressed, including in the pancreas, liver (12), and central nervous system. We hypothesized that a mild increase in GFPT activity, particularly in the presence of obesity, dietary excess, or mild hyperglycemia, could predispose to diabetes and diabetic complications and that common, naturally occurring sequence variants could result in either increased gene expression or increased enzymatic activity. We initiated our test of these hypotheses with GFPT2 because the full genomic sequence was available when the study was initiated, and we were able to demonstrate gene expression in transformed lymphocytes. We report here on the role of GFPT2 sequence variation in the risk of diabetes in both Caucasian and African-American populations, the risk of diabetic nephropathy in an African-American population, and the association of these variants with altered GFPT2 mRNA levels in transformed lymphocytes from these populations.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The study goal was to evaluate GFPT2 as a susceptibility gene for T2DM and diabetic nephropathy. Because of the increased prevalence of diabetic nephropathy in African-American populations, we focused on this population for the evaluation of GFPT2 in diabetic nephropathy. We screened for GFPT2 sequence variants in 40 individuals comprising 10 Caucasian individuals with T2DM and diabetic nephropathy, 10 Caucasian individuals with T2DM who did not have evidence of nephropathy, 10 African-American individuals with T2DM and diabetic nephropathy, and 10 African-American subjects with T2DM and no diabetic nephropathy. All subjects in the mutation detection study were included in the case-control study.
The role of GFPT2 variants identified in the screening study was then determined in four separate studies. In the first study, we tested for an association with T2DM in 384 unrelated Caucasian subjects comprising 196 individuals with diabetes and 196 control individuals. Subjects were ascertained from Utah for Northern European ancestry or from Arkansas for mixed Caucasian ancestry. Among diabetic subjects, 73 individuals were selected families used in previous studies (13), and 123 were similarly ascertained for a history of diabetes and diabetes in at least one-first degree relative. Control individuals had no family history of diabetes in any first-degree relative and either a normal 75-g glucose tolerance test or a random glucose less than 5.6 mmol/liter (100 mg/dl) and included 72 individuals ascertained from Arkansas for European ancestry, four samples from three generation Center for the Study of Human Polymorphism (CEPH) families (gender and diagnosis unknown but counted as nondiabetic), and 120 individuals from Utah with Northern European ancestry.
The role of GFPT2 in diabetes and diabetic nephropathy was tested in 333 unrelated African-American subjects, including 93 control individuals with no family history of diabetes and either a normal 75-g oral glucose tolerance test or a random glucose less than 5.6 mmol/liter, 105 individuals with known T2DM and urine albumin to creatinine ratios of less than 30 mg albumin/g creatinine (T2DM without nephropathy), and 135 individuals with T2DM and known or newly detected nephropathy (elevated creatinine, dialysis, proteinuria, or albumin to creatinine ratio of > 300 mg albumin/g creatinine). All individuals were unrelated and were ascertained from the same population in Arkansas. Individuals with intermediate values of albumin to creatinine ratios (microalbuminuria) were not included in this study.
To examine the impact of GFPT2 variants associated with T2DM on insulin sensitivity (SI) and secretion, we tested 126 nondiabetic members of 26 Northern European families ascertained for at least two diabetic siblings. The subjects underwent frequently sampled intravenous glucose tolerance tests for determination of insulin secretion [acute insulin response to glucose (AIRG)] and SI, as described in detail elsewhere (14, 15).
Subjects ascertained in Utah provided written informed consent under a protocol approved by the University of Utah Institutional Review Board. Subjects studied in Arkansas provided written informed consent under protocols approved by the University of Arkansas for Medical Sciences Human Research Advisory Committee. All human studies were performed in the General Clinical Research Centers of the University of Utah and the University of Arkansas for Medical Sciences.
Detection of sequence variants
We screened a total of 7170 bp, including 1 kb of 5' flanking sequence, 5' and 3' untranslated regions (UTRs), all 19 exons, and between 50 bp and 150 bp of sequence flanking each exon for mutations using 25 sets of primers (primer sequences available from authors). We used the public genome sequence data and the mRNA sequence (accession no. NM_005110) to determine the genomic structure of GFPT2. We relied primarily on single-strand conformation polymorphism analysis (16), as described in detail previously (17). Amplicons were under 300 bp for maximum sensitivity, and fragments were separated under the following two conditions: 5% polyacrylamide (49:1 of acrylamide to bis-acrylamide) with 10% glycerol and Mutation Detection Enhancement gels (FMC Bioproducts, Rockland, ME). The sensitivity of this method for detection of new variants in our laboratory exceeds 80% and outperforms alternative gel-based methods, including direct sequencing. Polymorphic fragments were characterized by bi-directional sequence analysis (18) using infrared dye-labeled primers (http://bio.licor.com/App_514/App514.htm; LI-COR Biotech, Lincoln, NE) and the DYEnamic Direct Cycle Sequencing Kit with 7-deaza-dGTP (Amersham Pharmacia Biotech, Piscataway, NJ). Two additional single nucleotide polymorphisms (SNPs) were identified by searching the National Center for Biotechnology Information SNP database (www.ncbi.nlm.nih.gov/SNP) for SNPs in the promoter (5' flanking) region.
SNP analysis
We typed SNP1 by Hinf1 (New England Biolabs, Beverly, MA) digestion of the amplified product followed by separation on 2% agarose gels. All other SNPs were typed using the Pyrosequencer PSQ-96 (Pyrosquencing, Inc, Uppsula, Sweden) according to the manufacturer’s methods. Primers for all assays are shown in Table 1
. All SNPs were in Hardy-Weinberg equilibrium except for SNP7 among African-American nondiabetic individuals (P = 0.015), which remained out of Hardy-Weinberg equilibrium after assay redesign.
|
Epstein-Barr virus-transformed lymphocytes from the subjects described above were grown to 0.5–1.0 x 106 cells/ml in RPMI 1640 media (Omega Scientific Inc, Tarzana, CA) with 10% fetal bovine serum and 11 mM glucose. Total RNA was isolated using Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA), and genomic contamination was removed by DNA-free reagent (Ambion, Inc, Austin, TX). Total RNA (1 µg) was reverse transcribed using random priming and Superscript II reverse transcriptase (Invitrogen Life Technologies). RT-PCR products from each allele of individuals heterozygous for SNP8 and SNP10 were quantified by Pyrosequencing using SNP Software AQ (Pyrosequencing, Inc). Validity of allelic ratios was determined by comparing peak heights for the two alleles in genomic DNA from heterozygous individuals and by mixing genomic DNA from homozygous individuals. Each measure was conducted in duplicate or triplicate, with six individuals for SNP8 and 16 individuals for SNP10. For SNP10, we tested cell lines derived from three Caucasian subjects without diabetes, six Caucasian subjects with T2DM, three African-American subjects without diabetes, two African-American subjects with diabetes and no nephropathy, and two African-American subjects with T2DM and diabetic nephropathy.
The ratios of GFPT2 mRNA levels to 18S rRNA in transformed lymphocytes were compared among 25 Caucasian individuals and 25 African-American individuals, comprising 10 Caucasian individuals and eight African-American individuals without diabetes, 10 Caucasian individuals and eight African-American individuals with T2DM and no nephropathy, and five Caucasian individuals and nine African-American individuals with T2DM and nephropathy. Primers were designed using Primer Express software (Applied Biosystems, Foster City, CA), and real-time PCR performed using the SYBR green real-time PCR reagents kit according to the manufacturer’s protocol (Applied Biosystems). The primers were 5'-GGACAGCACAACCTGCCTTT (GFPT2, forward), 5'-CAGCACTTGCATCAGAAGCAA (GFPT2, reverse), 5'-TTCGAACGTCTGCCCTATCAA (18S forward), and 5'-ATGGTAGGCACGGCGACTA (18S reverse). Standard curves were generated using pooled samples. To account for differences in gene expression, the ratio of real-time PCR product used to measure GFPT2 and 18S rRNA was adjusted to 1:8. GFPT2 and 18S rRNA levels were measured in separate reactions; reactions were performed in triplicate and detected on the ABI Prism 7700 (Applied Biosystems).
Statistical analysis
Allele frequencies between individuals with T2DM and controls were compared separately for each ethnic group using the Fisher’s exact test as implemented in the 2by2 program (19). To minimize problems of multiple testing, only allele frequencies were compared, which precludes the inclusion of covariates. SNP associations with diabetic nephropathy were examined similarly in African-American diabetic individuals with normal urine albumin to creatinine ratios and diabetic individuals with nephropathy. Deviations in allelic expression from the expected 50% were determined by the one-sample t test (SNP10) or by the independent two-tailed t test compared with unrelated DNA samples. GFPT2 mRNA expression between groups was compared using ANOVA by taking the natural logarithm of the mean of two or three measures of the ratio of GFPT2 mRNA to 18S rRNA levels after discarding obvious outliers. For all statistical tests, we considered P < 0.05 to be evidence of significance. Because many of the tests are correlated, we did not correct for multiple testing. All analyses were performed in SPSS for Windows, version 11.5 (SPSS, Inc, Chicago, IL). Pairwise linkage disequilibrium (LD) was calculated from combined case and control population data using the expectation maximization algorithm (19). Two statistics were calculated, D' (19) and r2, which is a better measure of how well the results from two SNPs will correlate (20).
We tested the impact of each variant on SI and insulin secretion using mixed effects regression models (21), as described elsewhere (15, 22). Insulin secretion (AIRG), SI, and disposition index (SI*AIRG) were determined as described previously (15, 22). All skewed variables were natural logarithmically transformed to normality before analysis. Analyses were performed in SPSS, version 11.5.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
and in Table 2. Only four of the 11 SNPs were also reported in the National Center for Biotechnology Information database (www.ncbi.nlm.nih.gov/SNP). Six SNPs were present within the coding sequence, of which one (SNP1) was in the 5' UTR, two (SNP11 and SNP10) were in the 3' UTR, and two SNPs were synonymous (SNP4 and SNP8). The remaining nonsynonymous SNP (I471V) in exon 14 changed codon 471 from isoleucine to valine. SNP3 and SNP11 were unusual in the screening samples and were not typed, and SNP4 and SNP6 were observed only in the African-American sample.
|
|
). We only typed variants in African-American nondiabetic controls to replicate findings in the Caucasian case-control study or if the variant was seen only in African-American individuals. Among Caucasians, SNPs 2, 5, 7, 9, and 10 were associated with T2DM (P < 0.05), with the minor allele overrepresented in diabetic individuals for each SNP (Table 2). For SNP8, a trend was noted but with overrepresentation of the common allele among diabetic individuals. Among African-Americans with diabetes, SNP10 was associated with nephropathy (P < 0.05), but no other SNP was significantly associated with nephropathy, and no SNP tested was associated with T2DM in African-American subjects (Table 2). Among Caucasians, SNPs 5, 7, 9, and 10 in intron 2 through the 3' UTR were in strong LD (D' > 9, r2 > 0.7). Although SNPs 1, 2, and 8 were also in strong LD with SNP10 by D', the correlation coefficient (r2), which better reflects whether the same disease association would be observed in both SNPs, was very low (r2 < 0.2). The extended haplotypes of all SNPs among diabetic and control populations are shown in Table 3. Of the potential 512 haplotypes, only 38 were observed, of which only 12 haplotypes exceeded a 1% frequency in either population. These 12 haplotypes accounted for 94.5% of control and 89.8% of all T2DM haplotypes. A single haplotype (TTGTGATCG) was more common among controls than among cases (P < 0.05), but the SNP10 T allele, which was associated with both T2DM in Caucasians and with altered gene expression (see next paragraph and Table 4), was present on several common haplotypes (bold type in Table 3). We observed much less LD among African-American subjects (data not shown). Only SNP12 and SNP13 were in strong LD (r2 = 0.94), and high correlations were not observed between other pairs of SNPs, including SNPs 5–10 in the 3' end of the gene.
|
|
). The ratio between the alleles was quantified using Allele Quantification software and Pyrosequencing (Pyrosequencing, Inc). We first confirmed that the allele ratio for these assays in genomic DNA did not deviate from the expected 50% for each allele (data not shown). In contrast, the T allele of SNP10 was nearly 2-fold overexpressed in GFPT2 RT-PCR product (mean, 65.8% vs. 34.2% in all subjects without regard to ethnicity or diagnosis; P < 0.0001; Fig. 2), and allele C of SNP8 was also overexpressed (55.7% vs. 44.3% for all subjects; P = 0.02; Fig. 2). To explore the possibility of differences of expression due to ethnicity or diagnosis, we also examined the allele-specific expression of SNP10 by ethnic group and by diagnosis (Table 4). We found a greater deviation from the expected equal expression ratio among African-American than among Caucasian subjects (P < 0.02), but no differences were seen by diagnosis (Table 4). The deviation from the expected expression ratio for SNP10 was significant for both ethnic groups (P < 0.001 in African-Americans, P = 0.005 in Caucasians by one-sample t test).
|
; P = 0.008). In exploratory analyses, cell lines derived from African-American individuals showed 2-fold higher mean ratios of GFPT2/18S RNA than Caucasian individuals (mean ratio, 18.62; 95% confidence interval, 7.5–28.7; and mean ratio, 8.99; 95% confidence interval, 4.03–56.5 in African-American and Caucasian individuals, respectively; P < 0.005). In contrast, when ethnicity was ignored, no significant differences were observed between diagnostic categories (Fig. 3).
|
). Thus, these findings are opposite of those expected under the hypothesis that increased GFPT2 mRNA levels increase the risk for T2DM. |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Based on these concepts, we undertook a thorough genetic analysis of GFPT2, one of two isoenzymes that accounts for GFPT activity in both humans and mice. To our knowledge, this is the first exploration of the role of genetic variation in this rate-limiting step in humans. In support of GFPT2 as a candidate gene for T2DM, this gene lies in a region of linkage on chromosome 5q34–5q35.2 recently identified in nonobese diabetic members of families from Iceland (26). We have identified one common amino acid polymorphism and several noncoding polymorphisms in the 3' end of the gene that show an association with T2DM in Caucasian subjects. Although GFPT1 is the major isoform in kidneys, we show evidence for an association of the same GFPT2 allele with diabetic nephropathy in African-American subjects. Because our Caucasian population with T2DM and nephropathy is small, we were unable to test the association of GFPT2 and diabetic nephropathy in Caucasian individuals.
We have examined a large number of SNPs for an association with two diseases, diabetes and diabetic nephropathy. The observed associations are modest, with differences in allele frequency of only 5–7% between cases and controls. Were we to apply a Bonferroni correction for multiple tests, no observed association would be significant. However, the SNPs were not independent due to strong LD in the Caucasian population, and multiple SNPs were associated with T2DM or diabetic nephropathy. Thus, a Bonferroni correction would be overly conservative and would likely lead to a type 2 error. Replication is an important means to guard against spurious associations, but as is often the case, we were unable to replicate our association findings in an Arkansas African-American population. This failure to replicate might suggest that the original findings were spurious, but several other possibilities seem more likely. Altshuler et al. (27) recently argued that smaller association studies often lack the power to replicate the initial findings. For example, our African-American population included 93 control (nondiabetic) individuals and 333 diabetic individuals. Over the range of allele frequencies observed for GFPT2 SNPs, which were less frequent in African-Americans than in Caucasians, we have 80% power to detect differences of 10% or greater between cases and controls, but we have only 50% power to replicate the difference observed in Caucasians. Additionally, the strong LD observed among SNPs in the Caucasian population was not seen in the African-American population. Thus, an undetected intronic regulatory SNP might not be detected by LD in African-American subjects.
Because of the difficulties in replicating association findings, we sought alternative strategies to support the association. We chose the following three methods: examination of intermediate phenotypes in individuals who had undergone assessment of SI and secretion, assessment of allele-specific expression, and assessment of mRNA levels among different ethnic and diagnostic groups. We found evidence of an association of only SNP8 with SI, and the reduced SI resulted from the genotype that was most common in nondiabetic individuals and thus might have been expected to be protective against T2DM. SNP8 showed only marginal evidence for an association in the Caucasian case-control study, although we did find differences in allele-specific expression. Considering that we examined three traits (AIRG, SI, and a ß-cell compensation index, AIRG*SI) for four SNPs, the finding of a decrease in SI with the TT genotype may represent a type 1 error. Alternatively, the effects of SNP8 or another SNP (SNP7) in LD with SNP8 might differ among tissues. Larger studies in individuals without a strong family history of diabetes are needed to reassess the role of GFPT2 on SI and insulin secretion.
A second method used to validate the association studies was to examine the influence of sequence variants on GFPT2 mRNA levels in transformed lymphocytes using the two variants identified within the transcribed sequence. By comparing allele-specific gene expression, we were able to quantify the amount of mRNA produced by each allele independent of potential confounders, such as trans-acting factors, that differ between individuals. A caveat of these studies is that transformed lymphocytes are clearly not a tissue involved in either diabetes or diabetic nephropathy. However, the assessment of gene variation in transformed lymphocytes was validated recently by demonstrating common alterations in allelic expression ratios in other genes (28) and by showing heritability of gene expression levels in three generation pedigrees (23). Because tissues of interest are often not available (pancreas) or cannot be obtained ethically (renal tissues before onset of nephropathy, liver), surrogate tissues are important to examine effects on gene expression. We found that the alleles associated with diabetes and diabetic nephropathy were also preferentially expressed, which is consistent with the hypothesis that increased GFPT activity would increase diabetes risk. Although we observed the allelic association only among Caucasian individuals, the allele-specific mRNA overexpression was seen in both Caucasian and African-American individuals regardless of diagnostic status. Therefore, our data support a role for these variants in both ethnic groups despite our failure to find an association with T2DM among African-American subjects.
A third way to evaluate the role of GFPT2 in diabetes and diabetic nephropathy independent of allelic association is to compare GFPT2 mRNA levels among lymphoblastoid cell lines derived from nondiabetic individuals, individuals with diabetes, and individuals with diabetes and nephropathy for both Caucasians and African-Americans. However, results of these studies are less direct in implicating sequence variation of the GFPT2 gene. Consistent with a study of 813 genes by Cheung et al. (23), we observed very large individual variations in gene expression among cell lines grown in identical conditions. The large interindividual variation made intergroup differences difficult to detect but suggested possible genetic influences on GFPT2 expression. These genetic influences might reflect underlying genotypes at the GFPT2 locus as well as unique trans-acting factors. Due to the frequency of the SNPs evaluated in this study, we were unable to stratify gene expression by GFPT2 genotype. However, we found a striking 2-fold increase in GFPT2 mRNA among African-Americans subjects compared with Caucasian individuals. Because the cell lines used for this study from both populations were treated similarly and are of approximately the same vintage, ethnicity appears to be the significant differentiating factor. No sequence variant identified in this study appears to account for these differences among individuals or ethnic groups. Such variants might exist in intronic regions not screened, in regions upstream or downstream of the region included in our analysis, or in other genes that regulate transcription.
In summary, we have identified a number of variants in the GFPT2 gene that are associated with T2DM in Caucasian subjects and with diabetic nephropathy but not diabetes in African-American individuals. Although combinations of SNPs (haplotypes) may alter disease susceptibility (29, 30), we found no extended haplotype that predicted disease susceptibility better than single SNPs in the 3' portion of the gene. The most likely candidate SNPs to account for these observations are SNP7, which alters a highly conserved amino acid in exon 14, and SNP10 in the 3' UTR, which shows the greatest difference in differential expression. Variants in 3' UTRs have been demonstrated recently to alter mRNA stability and thus mRNA levels (31). Based on the lack of association of SNPs in the 5' end of the gene and the lack of LD between 3' UTR SNPs and SNPs in the 5' region, variation in the promoter region is unlikely to account for our observations. However, SNP10 falls outside of the region that is conserved between mouse and human genes. Further study is needed to determine the functional roles of the I471V and SNP10 variants, to determine how these variants might interact, and to confirm the associations found in these studies.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: AIRG, Acute insulin response to glucose; GFPT, glutamine:fructose-6-phosphate amidotransferase; LD, linkage disequilibrium; SI, insulin sensitivity; SNP, single nucleotide polymorphism; T2DM, type 2 diabetes mellitus; UTR, untranslated region.
Received July 24, 2003.
Accepted October 22, 2003.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
genes. Diabetes Care 24:472–478 (TCF1) gene in typical familial type 2 diabetes in African Americans. Metabolism 49:280–284[CrossRef][Medline]
variation with lipid levels in familial type 2 diabetes. Mol Genet Metab 76:3–12
Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes. Nat Genet 26:76–80[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  R. Karchin Next generation tools for the annotation of human SNPs Brief Bioinform, January 1, 2009; 10(1): 35 - 52. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. C. Cooksey, S. Pusuluri, M. Hazel, and D. A. McClain Hexosamines regulate sensitivity of glucose-stimulated insulin secretion in {beta}-cells Am J Physiol Endocrinol Metab, February 1, 2006; 290(2): E334 - E340. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. G. Ewens, R. A. George, K. Sharma, F. N. Ziyadeh, and R. S. Spielman Assessment of 115 Candidate Genes for Diabetic Nephropathy by Transmission/Disequilibrium Test Diabetes, November 1, 2005; 54(11): 3305 - 3318. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Wang, H. Zhang, Y. Jia, Z. Zhang, R. Craig, X. Wang, and S. C. Elbein Adiponectin Receptor 1 Gene (ADIPOR1) as a Candidate for Type 2 Diabetes and Insulin Resistance Diabetes, August 1, 2004; 53(8): 2132 - 2136. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
