Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2002 Mar 19;99(7):4614–4619. doi: 10.1073/pnas.062059799

HNF-1α G319S, a transactivation-deficient mutant, is associated with altered dynamics of diabetes onset in an Oji-Cree community

Barbara L Triggs-Raine *, Robert D Kirkpatrick *, Sherrie L Kelly *, Lisa D Norquay †, Peter A Cattini †, Kazuya Yamagata ‡, Anthony J G Hanley §, Bernard Zinman §, Stewart B Harris , P Hugh Barrett ‖, Robert A Hegele **,‡‡
PMCID: PMC123696  PMID: 11904371

Abstract

The prevalence of type 2 diabetes mellitus in the Oji-Cree of northwestern Ontario is the third highest in the world. A private mutation, G319S, in HNF1A, which encodes hepatic nuclear factor-1α (HNF-1α), was associated with Oji-Cree type 2 diabetes and was found in ≈40% of affected subjects. The G319S mutation reduced the in vitro ability of HNF-1α to activate transcription by ≈50%, with no effect on DNA binding or protein stability. There was no evidence of a dominant negative effect of the mutant protein. The impact of the G319S mutation at the population level was assessed by classifying subjects with type 2 diabetes according to HNF1A genotype and plotting the cumulative age of onset of diabetes. Disease onset was modeled satisfactorily by two-parameter sigmoidal functions for all diabetic subjects and all three HNF1A genotypes. Pairwise statistical comparisons showed significant between-genotype differences in t50 (all P < 0.00001), corresponding to the age at which half the subjects had become diabetic. Each dose of G319S accelerated median disease onset by ≈7 years. Thus, the transactivation-deficient HNF1A G319S mutation affects the dynamics of disease onset. The demonstration of a functional consequence for HNF1A G319S provides a mechanistic basis for its strong association with Oji-Cree type 2 diabetes and its unparalleled specificity for diabetes prediction in these people, in whom diabetes presents a significant public health dilemma. The findings also show that HNF1A mutations can be associated with typical adult-onset insulin-resistant obesity-related diabetes in addition to maturity-onset diabetes of the young.

Type 2 diabetes mellitus has reached epidemic proportions in many aboriginal communities. For instance, the prevalence of type 2 diabetes in the population of ≈30,000 Oji-Cree of northwestern Ontario and Manitoba is the third highest in the world (1). In an isolated community from this region called Sandy Lake, ≈40% of adults had either typical obesity-related type 2 diabetes or impaired glucose tolerance (1). Both environmental and genetic factors are considered to be important contributors to this diabetes epidemic (1, 2). Defining the genetic determinants of diabetes susceptibility could help to develop strategies to prevent or delay disease onset, thereby delaying the onset of complications such as blindness, vascular disease, and/or nephropathy, which are difficult to manage in far northern locales. A search for genetic determinants resulted in the identification of the G319S mutation in HNF1A (also called TCF1; Mendelian Inheritance in Man 142410.0008), that encodes hepatic nuclear factor-1α (HNF-1α; ref. 2). HNF1A G319S was strongly associated with Oji-Cree type 2 diabetes (2), was found in ≈40% of affected subjects, and had a very high diagnostic specificity for diabetes prediction (3).

HNF-1α is a 631-aa atypical homeodomain-containing protein, which was identified first through its ability to bind to critical regulatory cis elements present in the 5′-flanking region of liver-specific genes such as those encoding albumin, fibrinogen, transthyretin (TTR), and α1-antitripsin (4, 5). HNF-1α is expressed in the liver, kidney, intestine, and pancreas (4, 5) and contains an N-terminal dimerization domain, a bipartite DNA binding domain, and a C-terminal transactivation domain (4–7). In induced-mutant mice, HNF-1α was shown to be essential for the expression of the glut2 glucose transporter and L-type pyruvate kinase genes in pancreatic β-cells (8, 9). Humans with mutations in HNF1A have pancreatic β-cell dysfunction (10–13). These findings indicate that HNF-1α has an important role in differentiation and coordinate regulation of gene transcription in pancreatic β-cells (7–9), providing a pathophysiological link to diabetes.

HNF1A mutations have been reported mainly in the maturity-onset diabetes of the young (MODY) type 3 form of diabetes, which is characterized by an absolute deficiency in insulin secretion and nonobese anthropometry (10). Some HNF1A mutations completely abolished the ability of HNF-1α to promote transcription, and a few resulted in partial loss of activity and had a dominant negative effect on the activity of wild-type HNF-1α (10–13). All these defects presumably led to an absolute deficiency in insulin secretion, which is the biochemical hallmark of MODY3. However, Oji-Cree type 2 diabetes does not resemble MODY, because affected Oji-Cree subjects are obese and insulin-resistant with elevated plasma insulin concentrations, which clearly were insufficient to prevent diabetes onset. We evaluated the in vitro function of HNF1A G319S to both confirm that the mutation had a functional impact and determine whether the functional impact was distinct from the complete loss of function and dominant negative mutations seen in the MODY3 phenotype. We also evaluated the impact of the HNF1A G319S mutation on the dynamics of type 2 diabetes onset in the whole Sandy Lake Oji-Cree community.

Methods

Strategy for Transactivation Experiments and Reagents.

To evaluate the functional impact of the G319S mutation, we used an HNF-1α-dependent luciferase reporter construct. Plasmid constructs expressing either wild-type or mutant HNF-1α were cotransfected into HeLa cells with a firefly luciferase reporter construct. The reporter construct contained a minimal TTR promoter that included an HNF-1α binding site and had been used previously for similar studies (12). A plasmid expressing Renilla luciferase was also cotransfected to provide an internal control for transfection efficiency. The control vector expressing Renilla luciferase, pRL-TK, was purchased from Promega. The anti-Xpress antibody was from Invitrogen, anti-HNF-1α from Santa Cruz Biotechnology, and goat anti-mouse IgG-horseradish peroxidase from The Jackson Laboratory. Unless stated otherwise, all restriction and modifying enzymes were purchased from New England Biolabs. Cell culture reagents were purchased from Life Technologies (Mississauga, ON). Other chemicals and reagents were from Fisher Scientific unless stated otherwise.

Construction of Expression Vectors.

The construction of the pcDNA3.1/HisC-wild-type human HNF-1α expression vector was described previously (13). The G319S mutation was introduced into HNF-1α using PCR overlap extension mutagenesis. Briefly, PCR was used to generate overlapping fragments, A and B, containing the 955G → A (G319S) substitution. Fragment A was generated with sense primer WPG221 (5′-AGA GCC CAC AGG TGA TGA G-3′) and antisense primer WPG213 (5′-AGC GCA CAC TGT GGA CCT TAC T-3′). Fragment B was generated with sense primer WPG214 (5′-AGT AAG GTC CAC AGT GTG CGC T-3′) and antisense primer WPG222 (5′-GTG TCT GTG ATG AGC ATA G-3′). Fragments A and B were used as PCR templates with primers WPG221 and WPG222, and the resulting 1.2-kb fragment was cloned into pCRII using the TA cloning kit (Invitrogen). A 305-bp AspI/BsmI fragment was subcloned into an HNF-1α cDNA in pBluescript (+). The entire HNF-1α cDNA then was removed with EcoRI/BamHI and subcloned into pcDNA3.1/HisC to regenerate pcDNA3.1/HisC-G319S HNF-1α. The absence of additional changes in the PCR-amplified region was confirmed by sequencing. The construction of the luciferase reporter plasmid, pGL3-TTR, was reported previously (12). It contains the minimal promoter of human TTR in the vector pGL3-basic. DNA for transfection was isolated using the Concert Maxiprep system (Life Technologies) and quantified carefully by gel electrophoresis.

Transient Expression Analysis of Wild-Type and Mutant HNF-1α.

Cultured HeLa cells (American Type Culture Collection) were maintained in αMEM containing 10% FBS and 100 units/ml penicillin-streptomycin. Cells were subcultured into 60-mm dishes (400,000 per dish) and transfected at ≈80% confluence. Transfections were performed with the indicated amount of HNF-1α expression vector, and reporter plasmids (TTR and pRL-TK) using DOSPER reagent (Roche, Laval, QC) at a ratio of 8:1 (Dosper/DNA). At 48 h posttransfection, cells were collected in Passive lysis buffer (Promega Dual luciferase reporter assay system) and subjected to two rounds of freezing and thawing. Sample aliquots then were assayed by using the Dual luciferase reporter assay system (Promega) and following the instructions of the manufacturer. Luminescence was measured using an EG & G Berthold Lumat BL9507 luminometer (Bundoora, Australia).

Western Blotting.

Protein lysates from transfections were separated on a 7.5% SDS/PAGE gel and transferred to nitrocellulose (14). Blots were blocked in Tris-buffered saline containing 0.1% Tween 20 and 5% skim milk powder; incubations with antibodies were in the same buffer but without the skim milk powder. Reactive bands were detected by ECL (Amersham Pharmacia) and quantitated with a Bio-Rad Fluor-S MultiImager equipped with a charge-coupled device camera and the manufacturer's software.

Preparation of Nuclear Extracts.

Nuclear protein extracts from HeLa cells (nontransfected and transfected with 1 μg of wild-type or G319S-containing cDNA expression vector) were prepared as reported (15). Protein concentration of the extracts was assessed using the Bradford (16) protein assay with BSA as a standard. Extracts were stored as aliquots at −70°C.

Electrophoretic Mobility-Shift Assay (EMSA).

Double-stranded DNA elements were generated by synthesizing and annealing sense and antisense oligonucleotides (Life Technologies). The sense strand for each element is given: HNF-1α, 5′-TAT GGG TTA CTT ATT CTC TCT TT-3′, and RF-1, 5′CTC ATC AAC TTG GTG TGG ACG GC-3′. EMSA was performed essentially as described (17). Reactions carried out in the presence of 3 μg of poly (dI-dC) in reaction buffer (10 mM Hepes-KOH, pH 7.9 at 4°C/12.5% glycerol/210 mM NaCl/7 mM MgCl2/0.1 mM EDTA/0.25 mM DTT/0.1 mM PMSF) in a 20-μl final volume. HeLa cell nuclear protein (10 μg) was preincubated for 10 min at room temperature with competitor oligonucleotides (i.e., 25-, 50-, and 100-fold mass excess of the probe), HNF-1α antibodies (1 μl), or normal rabbit serum (1 μl). Radiolabeled HNF-1α probe was added, and the reactions were incubated a further 10 min at room temperature before electrophoresis through 5% polyacrylamide in Tris-borate buffer.

Dynamics of Diabetes Onset in the Sandy Lake Community.

Descriptions of the clinical, biochemical, and genetic attributes of study subjects are presented elsewhere (1–3). For the analysis of dynamics of diabetes onset, all 121 subjects (74 women) from the community with physician-diagnosed type 2 diabetes were studied. The mean (± SD) present age of these subjects was 44.4 ± 15.2 years, and the mean age of onset of type 2 diabetes was 39.8 ± 12.7 years. The age of onset was plotted on the abscissa against the cumulative proportion of subjects with type 2 diabetes on the ordinate. Because this plot appeared sigmoidal (Fig. 4A), a nonlinear regression analysis, using SAAMII software (SAAM Institute, Seattle) was performed to determine whether the relationship could be explained by a sigmoidal function of the general form: y = xb/(xb + ab).

Figure 4.

Figure 4

Open in a new tab

Dynamics of conversion to diabetes in the Sandy Lake Oji-Cree. (A) Cumulative proportion of subjects from Sandy Lake with type 2 diabetes plotted against age of onset. The solid line represents regression line with the parameters shown in Table 1. (B) Cumulative proportion of subjects from Sandy Lake with type 2 diabetes mellitus plotted against age of onset. The solid circles are subjects with HNF1A S319/S319 genotype, open circles are subjects with the S319/G319 genotype, and triangles are subjects with G319/G319 genotype. The solid lines represent regression lines with the parameters shown in Table 1.

Parameter a is equivalent to t50, or the age at which half the subjects had become diabetic. Parameter b, also called the Hill constant, is a direct index of the sigmoidicity of the function. To determine whether there was a relationship between the age of onset of type 2 diabetes and the HNF1A genotype, subjects were stratified by genotype. Similar plotting and nonlinear regression analyses were performed for the three genotype subgroups. Comparisons of the differences in the sigmoidal parameter estimates between these three subgroups were made using the t test for independent samples assuming unequal variances.

Results

Functional Impact of HNF-1α G319S on Transactivation.

Luciferase reporter experiments reproducibly showed that the HNF1A G319S mutation decreased the transcriptional activity of HNF-1α by 54% (P < 0.0001) when compared with the wild-type protein (Fig. 1 A and B). Because several mutations that decrease HNF-1α activity also impart dominant negative properties when coexpressed with wild-type HNF-1α, we tested this possibility for HNF-1α that contained G319S. When equal amounts of wild-type and G319S HNF-1α-expressing plasmids were transfected, the levels of activity were found to be additive (Fig. 1B), indicating that the G319S HNF-1α had no negative effect on the activity of the wild-type protein.

Figure 1.

Figure 1

Open in a new tab

Transcriptional activity of wild-type (WT) and G319S HNF-1α. HeLa cells were transfected with 400 ng of pcDNA3.1/HisC wild-type HNF-1α, pcDNA3.1/HisC G319S HNF-1α, or the vector pcDNA3.1/HisC together with 1 μg of the reporter vector pGL3-TTR and 100 ng of pRL-TK (A; *P < 0.0001 compared with wild type) and 400 ng of pcDNA3.1/HisC wild-type HNF-1α, 400 ng of pcDNA3.1/HisC G319S HNF-1α, or a mixture of 200 or 400 ng each of pcDNA3.1/HisC wild-type HNF-1α and pcDNA3.1/HisC G319S HNF-1α together with 1 μg of the reporter vector pGL3-TTR and 100 ng of pRL-TK (B). Luciferase activity from pGL3-TTR was normalized to the activity expressed from pRL-TK. The ratio of wild-type HNF-1α activity to the control was set at 100%, and all other samples were expressed as a percentage of the wild type. Each variable was studied in triplicate, and each experiment was repeated three times (except #, which was performed twice). Means ± SE are shown.

The decreased activity resulting from the HNF-1α G319S mutation could have been caused by a decrease in transcriptional activation of the reporter construct or to decreased stability of the mutant protein and thus a decrease in steady-state protein levels. The latter alternative was assessed by comparing the amount of expressed wild-type and G319S HNF-1α by using Western blot analysis; samples were corrected for transfection efficiency through loading on the basis of equivalent units of Renilla luciferase activity. The protein was detected with an antibody directed to an N-terminal Xpress (Invitrogen) epitope tag that was introduced during construction of the expression construct. As shown in Fig. 2, the levels of wild-type and mutant protein were similar, indicating that the G319S mutation had no significant impact on the stability of HNF-1α.

Figure 2.

Figure 2

Open in a new tab

Steady-state levels of wild-type and G319S HNF-1α protein. HeLa cells were transfected with 4 μg of pcDNA3.1/HisC WT-HNF-1α or pcDNA3 1/HisC G319S HNF-1α and 200 ng of pRL-TK. Equivalent units of Renilla luciferase (expressed from pRL-TK) activity were loaded from each of two replicates that were performed for each construct. HNF-1α was detected with a 1:5,000 dilution of the primary antibody toward the Xpress epitope and a 1:10,000 dilution of the secondary anti-mouse antibody. The ECL signal was detected using Kodak MS x-ray film. In addition, each lane was quantitated (1–268, 2–231, 3–262, and 4–227) as described in Methods. The average signal for the G319S mutant protein was 98% of that of the signal from the wild-type protein. An arrow indicates the HNF-1α protein. This gel is representative of three experiments.

DNA Binding Experiments.

The HNF1A G319S mutation is located at the N-terminal end of the transactivation domain, adjacent to the DNA binding domain. To determine if the mutation affected binding to the recognition sequence, EMSA was performed by using the expressed wild-type and mutant protein. As shown in Fig. 3, extracts from both wild type (A) and G319S (C) transfected cells could specifically shift a 32P-labeled oligonucleotide harboring the HNF-1α-binding site. Furthermore, binding of both the mutant and wild-type proteins was competed with unlabeled oligonucleotide harboring the HNF-1α site (indicated as HNF-1α in Fig. 3) but not with a nonspecific competitor oligonucleotide (RF-1). In addition, the HNF-1α–oligonucleotide complexes could be competed by preincubation with an antibody against HNF-1α, indicating that the complex resulted from binding of HNF-1α. Both the wild-type and mutant HNF-1α–oligonucleotide complex appeared to be supershifted by the antibody (Fig. 3 B and D), confirming that both proteins bind DNA. Together, these data show that, as predicted, the G319S mutation does not alter the in vitro DNA binding ability of HNF-1α.

Figure 3.

Figure 3

Open in a new tab

Direct in vitro DNA binding by wild-type and G319S HNF-1α. EMSA was performed with a radiolabeled oligonucleotide probe containing a binding site for HNF-1α and extracts prepared from HeLa cells that were transfected with HNF-1α wild type (A and B) or the HNF-1α G319S mutant (C and D). A–D also contain EMSA performed by using nontransfected HeLa cell nuclear proteins. B and D are insets from longer exposures of A and C, respectively, to allow for better visualization of the less abundant specific complexes. Competition using specific (HNF-1α) and nonspecific (RF-1) double-stranded oligonucleotides or HNF-1α antibodies and normal rabbit serum (NRS) are indicated. Competitor oligonucleotides were used at 25-, 50-, and 100-fold mass excess compared with probe. Specific complexes are indicated by open arrowheads, and supershifted, less mobile bands are indicated by closed arrowheads. Complex II is present in all HeLa nuclear extracts and is not competed by the HNF-1α Ab, whereas complex I is visible only in EMSAs performed with transfected HeLa extracts and is competed with the HNF-1α antibody.

HNF-1α G319S and Community Dynamics of Diabetes Onset.

The S319 allele was shown previously to be significantly more prevalent in diabetic than nondiabetic Oji-Cree (0.21 vs. 0.09, P = 0.000001), and 37.6% of Oji-Cree subjects with diabetes had at least one copy of the S319 allele (2). The cumulative proportion of subjects from Sandy Lake with type 2 diabetes plotted against the age of onset is shown in Fig. 4A. The data could be modeled satisfactorily using a sigmoidal function with an adjusted r2 > 0.99 (P < 0.00001) as shown in Table 1. Parameter a, namely 38.5 ± 0.11 years, is equivalent to t50, or the age at which half the subjects had become diabetic. Parameter b, namely 4.3 ± 0.1, is an index of sigmoidicity (Hill constant).

Table 1.

Sigmoidal age-of-onset dynamics in Oji-Cree type 2 diabetes overall and according to the HNF1A genotype

All subjects S319/S319 S319/G319 G319/G319
Number 121 5 40 76
Present age, years 44.4  ± 15.2 39.3  ± 16.3 42.2  ± 15.2 45.7  ± 15.1
Body mass index, kg/m2 30.1  ± 4.8 27.7  ± 6.4 29.4  ± 4.1 30.7  ± 4.9
Serum insulin, pmol/l 178  ± 15 177  ± 51 183  ± 39 177  ± 13
a (t50 in years) 38.5  ± 0.11 26.7  ± 0.72 34.7  ± 0.18 40.8  ± 0.21
b (Hill constant) 4.3  ± 0.1 5.7  ± 1.0 4.7  ± 0.1 4.5  ± 0.1
Adjusted model r2 0.99 0.92 0.99 0.98
Model P value <0.00001 <0.00001 <0.00001 <0.00001
Open in a new tab

The cumulative proportion with type 2 diabetes plotted against the age of onset for subjects stratified by the HNF1A genotype is shown in Fig. 4B. These two curves each were modeled satisfactorily using sigmoidal functions (each r2 > 0.9 and P < 0.00001) as shown in Table 1. The defining parameters of these curves were compared against each other. For parameter a, or t50, all three pairwise between-genotype comparisons indicated significant differences (each P < 0.00001). For parameter b (the Hill constant), pairwise comparisons of G319 homozygotes with heterozygotes and with S319 homozygotes showed significant differences (Table 2), and pairwise comparison of heterozygotes with S319 homozygotes tended to show a difference (Table 2).

Table 2.

Summary of between-genotype comparisons of sigmoidal parameters

P values
a (t50 in years) b (Hill constant)
S319/S319 vs. S319/G319 <0.00001 0.084
S319/S319 vs. G319/G319 <0.00001 0.044
S319/G319 vs. G319/G319 <0.00001 <0.00001
Open in a new tab

Discussion

HNF1A G319S is unique among diabetes mutations, because it is private and occurs in such a high proportion of diabetic subjects in a closed human community. We report herein that HNF1A G319S has an impact on biological end points in two disparate model systems, namely in vitro analysis of expression and in vivo evaluation of diabetes onset in an Oji-Cree community. Specifically, the mutation decreased the transactivation potential of HNF-1α by ≈50% with no effect on DNA binding or stability. There was no evidence of a dominant negative effect of the mutant protein. Sigmoidal modeling showed that the HNF1A G319S mutation had a codominant influence on the onset of type 2 diabetes in Sandy Lake with ≈7 years earlier median diabetes onset for each copy of the G319S allele. These data support the idea that the G319S allele is a causative mutation in the pathogenesis of diabetes in the Oji-Cree population and provide a new example of how allelism in a MODY gene can be associated either with typical MODY3 or typical adult-onset insulin-resistant obesity-related diabetes.

The demonstration of functional consequences for HNF1A G319S provides a mechanistic basis for its strong association with Oji-Cree type 2 diabetes and its unparalleled specificity for diabetes diagnosis and prediction in these people (3). In the Oji-Cree, HNF1A G319S behaves as a susceptibility allele for type 2 diabetes. The odds of developing diabetes in homozygotes and heterozygotes compared with wild-type homozygotes were ≈4 and 2, respectively (both P < 0.0001; ref. 2). However, ≈60% of Oji-Cree subjects with type 2 diabetes were homozygous for wild-type HNF1A. This result suggests that other genetic and/or environmental factors must be involved in the development of Oji-Cree type 2 diabetes.

These observations confirm that HNF1A mutations can cause either type 2 diabetes or MODY3, which are very different clinical phenotypes. Subjects with Oji-Cree type 2 diabetes are obese, with high plasma insulin and C-peptide concentrations, indicating insulin resistance. In contrast, subjects with MODY3 are lean, with low plasma insulin caused by defective secretion. The in vitro analysis of transactivation function hints at the possible basis for this in vivo difference. The HNF1A G319S mutation resulted in diminished but not absent residual transcriptional activity of the HNF-1α protein. Thus, Oji-Cree with the G319S mutation would have relatively lower but not absent in vivo expression of HNF-1α-dependent genes, including insulin. A similar HNF1A missense mutation with partial dysfunction, G415R, was found in a Japanese patient with early-onset type 2 diabetes (12). This type of mutation contrasts sharply with the other HNF1A mutations that cause MODY3. A severe or complete loss of function would explain why the insulin-secretion defect in MODY3 does not require the additional stress of obesity-induced insulin resistance for disease expression. However, the partial loss of function of HNF1A G319S would require an additional stress to create a diabetes phenotype.

Because MODY patients typically are young and lean, the underlying severe HNF1A defects in MODY3 generally lead to diabetes in the absence of insulin resistance (9). In contrast, the modest transactivation defect resulting from HNF1A G319S has no clinical consequences in young, lean Oji-Cree subjects (2). Among nondiabetic Oji-Cree, fasting plasma insulin concentration was reduced significantly in HNF1A G319S carriers compared with noncarriers, suggesting that the partial impairment of function is tolerated when there is no insulin resistance (2). However, among Oji-Cree with type 2 diabetes, both carriers and noncarriers of HNF1A G319S had elevated plasma insulin concentration compared with nondiabetic Oji-Cree (2). The stress of obesity-induced insulin resistance seemed to expose the partial defect in HNF1A G319S carriers, causing the expression of the disease (2). Homozygotes for the G319S mutation would be expected to have a greater in vivo deficiency in HNF-1α activity than heterozygotes and thus greater susceptibility to diabetes, which was reflected by the differences between the age-of-onset curves (Fig. 4B).

The observations also demonstrate that a complex disease outcome, namely the onset of type 2 diabetes, in a higher order biological system, namely an aboriginal community, can be modeled mathematically with a nonlinear function. In particular, a sigmoidal model with 2-parameters described the dynamics of onset (or conversion to disease) for the overall Sandy Lake diabetic community, with r2 > 0.99 (P < 0.00001). Analysis using general linear models such as linear regression analysis would not have captured the sigmoidal relationships seen in Fig. 4. Also, the sigmoidal parameters that defined the different curves for the three HNF1A genotypes at the community level lent themselves to statistical comparisons and insinuated biological differences. In the case of the curves in Fig. 4B, these biological differences had clinical relevance, because they showed that the G319S allele accelerated the onset of disease in a dose-related manner. The earlier onset of diabetes would prolong the duration of disease in affected subjects, resulting in more time for the development of diabetes complications. Furthermore, the earlier onset implied that subjects with G319S required a shorter duration of exposure to those causative factors. These in vivo results are consistent with the in vitro expression results, which suggested that mutation carriers would be more susceptible to the development of diabetes (by becoming ill at a younger age) compared with noncarriers.

Sigmoidal relationships are common in biology and denote changing tendency for an outcome depending on contextual changes and/or accumulation of determinants such as the hemoglobin–oxygen equilibrium (18), cell death caused by cumulative damage (19), and dose-effect relationships in pharmacodynamics (20). Our findings extend the applicability of sigmoidal models to the mathematical description of disease onset in a closed human community. Other modeling approaches such as kinetic simulation, metabolic control analysis, biochemical systems theory, metabolic pathway analysis, and network analysis have been used to study cellular metabolism and signaling (21–23), allowing for indirect visualization of otherwise imperceptible molecular events. The findings from Sandy Lake indicate that under appropriate circumstances, the use of descriptive mathematical functions can extend to biological data collected from higher levels of organization such as closed human communities. The model parameters then can be compared statistically and interpreted, such as for sigmoid curves defining the three HNF1A genotypes. Mathematical models may be helpful tools to understand the mechanisms in complex biological systems that underlie emergent properties such as diseases.

In summary, the in vitro and in vivo experiments in this study confirmed the functional impact of the HNF1A G319S mutation in the Oji-Cree. The observations help to explain the previously observed statistical association with diabetes. HNF1A G319S was not the sole factor that predisposed to Oji-Cree type 2 diabetes, because many diabetic subjects were not carriers of this allele. However, the observations indicated that the partial loss of in vitro transactivation function was associated with altered in vivo dynamics of conversion to diabetes in a community whose disease risk has many determinants.

Acknowledgments

This work was supported by grants from the Canadian Genetic Diseases Network of Centres of Excellence, the Canadian Institutes for Health Research, and the Canadian Diabetes Association. S.L.K and L.D.N. were supported by Manitoba Health Research Council studentships. L.D.N. received support from the Women's Health Research Foundation. R.A.H. holds a Canada Research Chair in Human Genetics and a Career Investigator Award from the Heart and Stroke Foundation of Ontario. P.H.B. is a National Heart Foundation Career Development Fellow and was supported also by National Institutes of Health Grant RR12609.

Abbreviations

HNF-1α

hepatic nuclear factor-1α

TTR

transthyretin

MODY

maturity-onset diabetes of the young

EMSA

electrophoretic mobility-shift assay

References

  • 1.Harris S B, Gittelsohn J, Hanley A, Barnie A, Wolever T M, Gao J, Logan A, Zinman B. Diabetes Care. 1997;20:185–187. doi: 10.2337/diacare.20.2.185. [DOI] [PubMed] [Google Scholar]
  • 2.Hegele R A, Cao H, Harris S B, Hanley A J G, Zinman B. J Clin Endocrinol Metab. 1999;84:1077–1082. doi: 10.1210/jcem.84.3.5528. [DOI] [PubMed] [Google Scholar]
  • 3.Hegele R A, Cao H, Hanley A J G, Zinman B, Harris S B, Anderson C. Diabetes Care. 2000;23:775–778. doi: 10.2337/diacare.23.6.775. [DOI] [PubMed] [Google Scholar]
  • 4.Baumhueter S, Mendel D, Conley P B, Kuo C, Turk C, Graves M, Edwards C, Courtois G, Crabtree G. Genes Dev. 1990;4:372–379. doi: 10.1101/gad.4.3.372. [DOI] [PubMed] [Google Scholar]
  • 5.Blumenfeld M, Maury M, Chouard T, Yaniv M, Condamine H. Development (Cambridge, UK) 1991;113:589–599. doi: 10.1242/dev.113.2.589. [DOI] [PubMed] [Google Scholar]
  • 6.Cereghini S. FASEB J. 1996;10:267–282. [PubMed] [Google Scholar]
  • 7.Wang H, Antinozzi P A, Hagenfeldt K A, Maechler P, Wollheim C B. EMBO J. 2000;19:4257–4264. doi: 10.1093/emboj/19.16.4257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Lee Y-H, Sauer B, Gonzalez F J. Mol Cell Biol. 1998;18:3059–3068. doi: 10.1128/mcb.18.5.3059. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Parrizas M, Maestro M A, Boj S F, Paniagua A, Casamitjana R, Gomis R, Rivera F, Ferrer J. Mol Cell Biol. 2001;21:3234–3243. doi: 10.1128/MCB.21.9.3234-3243.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Ellard S. Hum Mutat. 2000;16:377–385. doi: 10.1002/1098-1004(200011)16:5<377::AID-HUMU1>3.0.CO;2-2. [DOI] [PubMed] [Google Scholar]
  • 11.Yamada S, Tomura H, Nishigori H, Sho K, Mabe H, Iwatani N, Takumi T, Kito Y, Ogata T, Onigata K, et al. Diabetes. 1999;48:645–648. doi: 10.2337/diabetes.48.3.645. [DOI] [PubMed] [Google Scholar]
  • 12.Vaxillaire M, Abderrahman A, Boutin P, Bailleul B, Froguel P, Yaniv M, Pontoglio M. J Biol Chem. 1999;274:35639–35646. doi: 10.1074/jbc.274.50.35639. [DOI] [PubMed] [Google Scholar]
  • 13.Yamagata K, Yang Q, Yamamoto K, Iwahashi H, Miyagawa J, Okita K, Yoshiuchi I, Miyazaki J, Noguchi T, Nakajima H, et al. Diabetes. 1998;47:1231–1235. doi: 10.2337/diab.47.8.1231. [DOI] [PubMed] [Google Scholar]
  • 14.Towbin H, Staechelin T, Gordon J. Proc Natl Acad Sci USA. 1979;76:4350–4354. doi: 10.1073/pnas.76.9.4350. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Andrews N C, Faller D V. Nucleic Acids Res. 1991;19:2499. doi: 10.1093/nar/19.9.2499. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Bradford M M. Anal Biochem. 1976;72:248–254. doi: 10.1016/0003-2697(76)90527-3. [DOI] [PubMed] [Google Scholar]
  • 17.Jin Y, Surabhi R M, Fresnoza A, Lytras A, Cattini P A. Mol Endocrinol. 1999;13:1249–1266. doi: 10.1210/mend.13.8.0332. [DOI] [PubMed] [Google Scholar]
  • 18.Neet K E. Methods Enzymol. 1995;249:519–567. doi: 10.1016/0076-6879(95)49048-5. [DOI] [PubMed] [Google Scholar]
  • 19.Clarke G, Collins R A, Leavitt B R, Andrews D F, Hayden M R, Lumsden C J, McInnes R R. Nature (London) 2000;406:195–199. doi: 10.1038/35018098. [DOI] [PubMed] [Google Scholar]
  • 20.Goldstein A, Aronow L, Kalman S M. Principles of Drug Action: The Basis of Pharmacology. New York: Wiley; 1974. pp. 14–20. [Google Scholar]
  • 21.Gershenfeld N. The Nature of Mathematical Modeling. Cambridge, U.K.: Cambridge Univ. Press; 1999. pp. 10–12. [Google Scholar]
  • 22.Wildermuth M C. Genome Biol. 2000;1:1031.1–1031.5. doi: 10.1186/gb-2000-1-6-reviews1031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Poolman M G, Fell D A, Thomas S. J Exp Bot. 2000;51:319–328. doi: 10.1093/jexbot/51.suppl_1.319. [DOI] [PubMed] [Google Scholar]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES