The Pro387Leu variant of protein tyrosine phosphatase-1B is not associated with diabetes mellitus type 2 in a German population
Dr Ioanna Gouni-Berthold, Department of Internal Medicine II, University of Cologne, Joseph-Stelzmann-Str. 9, 50924 Cologne, Germany.
(fax: +49 221 478 4179; e-mail: firstname.lastname@example.org).
Objectives. Diabetes mellitus type 2 (DM-2) is a complex disorder with a strong genetic background. Protein tyrosine phosphatase-1B (PTP-1B) dephosphorylates various receptor protein kinases in vitro, including the β subunit of the insulin receptor, therefore representing a potential candidate to be involved in the polygenic pathogenesis of DM-2. Recently a Pro387Leu variant of the PTP-1B gene has been associated with an increased risk of DM-2 in a Danish population. Reports from China and Finland failed to confirm this association.
Design, setting and subjects. The purpose of the present study was to examine the possible association between the presence of DM-2 and the Pro387Leu polymorphism in a German Caucasian population. A total of 836 subjects (age 20–92 years) participated in the study. The presence of the Pro387Leu variant of the PTP-1B gene was investigated using polymerase chain reaction (PCR) restriction fragment-length polymorphism in 402 subjects with DM-2 (231 men, 171 women, age 63.1 ± 10.8 years, BMI 28.7 ± 5.1 kg m−2) and in 434 normoglycemic age- and sex-matched control subjects (248 men, 186 women, age 64.4 ± 6.5 years, BMI 26.5 ± 3.7 kg m−2).
Results. Nine subjects in the control group and nine in the diabetic group (allelic frequency 0.99% in both groups) carried the Pro387Leu polymorphism. A meta-analysis on published data of >3000 subjects including our own data did not find an association between the polymorphism and DM-2. In addition, the polymorphism had no significant influence on the presence of atherosclerotic disease, whilst the influence of other known cardiovascular risk factors was confirmed. Furthermore, the impact of the mutation on metabolic and anthropometric parameters in both groups was examined. Amongst the controls there were no significant differences in BMI, HDL and LDL cholesterol or blood pressure between the two groups with or without the Pro387Leu polymorphism. The same was true for the diabetic group. Interestingly, in both diabetics and controls, Pro387Leu carriers had significantly higher triglycerides. In a logistic regression model only BMI and family history but not polymorphism were predictors of DM-2.
Conclusions. In conclusion, the present data suggest that in a German Caucasian population the Pro387Leu polymorphism of the PTP-1B gene is not associated with DM-2 but may play a role in other metabolic phenotypes.
Diabetes mellitus type 2 (DM-2) is a complex disease with a strong genetic background . It is caused by defective insulin secretion and action . The binding of insulin to the insulin receptor (IR) mediates the tyrosine phosphorylation of downstream signalling proteins such as the IR substrate-1 and -2 . Whilst a substantial amount of knowledge exists regarding the events that initiate and propagate insulin signalling, events that lead to signal termination are not sufficiently understood. The level of receptor activation is determined by the opposing actions of receptor phosphorylation and dephosphorylation, the latter mediated through protein tyrosine phosphatases (PTP) . The first mammalian PTP identified, PTP-1B, is widely expressed and localizes predominantly to the endoplasmic reticulum (ER) through a cleavable proline-rich C-terminal segment [5, 6]. The C-terminal 35 amino acids of PTP-1B are both necessary and sufficient for its targeting to the ER . Cleavage of this segment appears to release the enzyme from the ER and to increase its specific activity . In vitro, PTP-1B dephosphorylates many receptor tyrosine kinases, including the β-subunit of the IR [8–10], thus inactivating it. It is considered to be a negative regulator of insulin signalling [11–13] by dephosphorylating the phosphotyrosine residues of the IR kinase activation segment . For this reason a significant amount of effort has gone into creating PTP-1B-specific inhibitors for the treatment of DM-2 .
PTP-1B was originally purified from the cytosolic fraction of human placenta as a 37-kDa protein . Subsequent molecular cloning and sequencing demonstrated that the full-length molecule is composed of 435 amino acids having a molecular mass of approximately 50 kDa. It contains a conserved PTP catalytic domain within residues 30–278 and a COOH-terminal noncatalytic segment with potential regulatory function [5, 16]. A cysteine at position 215 in the catalytic domain is necessary for its enzymatic activity [17, 18]. Taking into account the crucial role of PTP-1B in regulating insulin signalling, PTP-1B seems to be involved in the polygenic pathogenesis of DM-2. A Pro387Leu variant in PTP-1B causing impaired serine phosphorylation of PTP-1B in vitro was associated with DM-2 in a Danish population , whilst other studies failed to replicate this association [20–22]. The proline at position 387 in the carboxyterminal regulatory region of the protein is conserved between mouse, rat and humans . It is part of a consensus sequence phosphorylated in vitro by the proline-directed kinase p34cdc2 [23, 24], in vivo in states of mitosis [25, 26] and in response to various stress stimuli . Replacing the proline by a leucine reduces the phosphorylation of the serine by 70%in vitro . However, the effect of Ser 386 phosphorylation on PTP-1B activity is unclear as it has been associated with both increased  and decreased  PTP-1B enzymatic activity. Other variants of the PTP-1B gene have also been identified, such as the silent variant P303P, the insertion variant 3′UTR + 104insG and the missense variant G381S. All of them had no significant association with DM-2 [19, 28]. Conflicting results regarding the clinical relevance of this enzyme have been observed in human studies relating PTP-1B mRNA levels and phosphatase activity to insulin sensitivity and obesity [29–33]. As in vitro studies suggest that the expression levels of PTPases may not be the primary determinant of their IR phosphorylation activity , other factors such as accessibility to intracellular substrates might be involved in regulating PTP-1B activity. Furthermore, a recent study showed that increased phosphoserine content of PTP-1B in vivo is followed by an increased PTP-1B activity and indicated that protein kinase A and insulin play a critical role in the regulation of PTP-1B .
PTP-1B is also a potential candidate in the polygenic pathogenesis of DM-2. First, PTP-1B-deficient mice have increased insulin sensitivity, maintain euglycaemia in the fed state with one-half the level of insulin observed in wild-type mice and are resistant to diet-induced obesity when fed a high-fat diet . Secondly, impaired PTP-1B activity is associated with insulin resistance in obese human subjects with and without DM-2 . Thirdly, the PTP-1B gene is located on the long arm of chromosome 20 , in a region (q13.1–q13.2) linked to the quantitative trait loci of obesity [38, 39], fasting serum insulin concentrations , as well as DM-2 [40–44]. Recent genetic evidence has shown that human PTP-1B gene variants are associated with changes in insulin sensitivity [19, 45, 46]. In this regard, Mok et al.  suggested that genomic variations in PTP-1B (specifically the single nucleotide polymorphism 981C→T) might be associated with a reduced risk of DM-2 in Canadian subjects. Furthermore, two recent studies performed in Danish and Chinese populations showed conflicting results regarding the relative risk for DM-2 in patients with the Pro378Leu variation. Echwald et al.  found a relative risk of DM-2 of 3.7 (CI 1.26–10.93, P = 0.02) in Danish Caucasian carriers of the Pro378Leu polymorphism whilst Weng et al.  found no association between the aforementioned polymorphism and DM-2 in a Chinese Han population. Two subsequent studies showed no association in French  and in Finnish  subjects. Considering that studies involving associations between genetic variations and DM-2 are often inconsistent in different ethnic populations, we examined the association between the Pro387Leu variation of the PTP-1B gene with DM-2 in a German Caucasian population.
A total of 836 subjects (age 20–92 years) participated in the study. Their data were obtained from the LIANCO database (Lipid Analytic Cologne). In brief, LIANCO was designed to assess the relationship between genetic mutations, serum lipoproteins, other biochemical parameters, and clinical data on atherosclerotic disease. Approval of the study protocol was obtained from the university ethics committee. Between spring 1999 and March 2002 a total of about 5000 patients were recruited in the Cologne (Germany) area by hospitals and office-based doctors. Patients’ data were recorded, their lipoproteins analysed and samples frozen for DNA extraction. The Pro387Leu variation of the PTP-1B gene was detected using PCR restriction fragment length polymorphism in 402 subjects with DM-2 (231 men and 171 women, age 63.1 ± 10.8 years, BMI 28.7 ± 5.1 kg m−2) and in 434 subjects without DM (248 men and 186 women, age 64.4 ±6.5 years, BMI 26.5 ± 3.7 kg m−2) matched according to sex and age. Subjects were defined as diabetic through OGTT or if receiving antidiabetic treatment. Sulphonylureas, metformin, acarbose, thiazolidinediones, glinides and insulin were considered as antidiabetic medications. Control subjects had normal fasting glucose according to American Diabetes Association Criteria .
Genomic DNA was isolated from whole blood using QiAmp Blood Maxi Kit (Qiagen, Hilden, Germany). The C→T transversion that causes an amino acid exchange from proline to leucine at codon 387 (Pro387Leu) was detected by using polymerase chain reaction (PCR) and restriction length polymorphism followed by electrophoresis on a 3% agarose gel. For the detection of the PTP-1B variant a PCR was performed with a forward primer (5′-CATCTCTGCCCTCTGATTCC-3′) and a reverse primer (5′-GTGCATCTGAGCCAGTCTCA-3′). The PCR product contains two BslI recognition sites, one at position 101 and one at position 290 in the PCR product. The mutation removes the BslI restriction site at position 101. The enzyme digestion of the PCR product from wild-type carriers produce three fragments: 37, 101 and 189 bp, whereas homozygous mutant carriers produce only a 37- and a 290-bp fragment. Heterozygous individuals show all fragments.
PCR was performed with 50 ng genomic DNA in a total volume of 25 μL, 2.5 pmol of each primer, 2.5 nmol dNTPs und 0.5 U HotStarTaq DNA Polymerase (Qiagen) and 2.5 mmol L−1 MgCl2 in a buffer supplied by the manufacturer. PCR conditions were as follows: initial denaturation (95 °C for 15 min), followed by 35 cycles of denaturation (95 °C for 1 min), annealing (65 °C for 1 min) and extension (72 °C for 2 min), with the final extension at 72 °C for 10 min. The PCR products were digested with 5 U BslI (New England Biolabs, Frankfurt, Germany), for 8 h at 55 °C using the buffer recommended by the manufacturer. The fragments were separated on 3% agarose gels with ethidium bromide and were visualized under ultraviolet light.
The PCR products were automatically sequenced (Abi Prism Genetic Analyzer model 310; Applied Biosystems, Foster City, CA, USA). The sequence of both strands was determined.
Statistical analysis was carried out by using StatView version 5.0 (SAS Institute, Inc., Cary, NC, USA). Statistical significance was defined as P < 0.05. Descriptive statistics are given, unless otherwise indicated, as mean values ± standard deviation. Comparison of mean values was performed by the unpaired Student's t-test or by Mann–Whitney U-test, as appropriate. Fisher's exact test was used to test for differences in allele and genotype frequencies. Allele frequencies were estimated by gene counting. The observed genotype frequencies were in Hardy–Weinberg equilibrium. Two-way analysis of variance was used to establish the influence of categorical variables on continuous parameters. A logistic modelling approach was used to investigate parameters that may be associated with the presence of DM-2 or atherosclerotic disease. A meta-analysis was conducted using data from previously published studies (Table 4) to increase the power of calculating genotype–phenotype associations. Power calculations were performed separately for the available studies including Caucasian subjects ([19, 47] and the present study) or for all studies (including also ).
Table 4. Genotype distribution of the Pro387Leu variant of the PTP-1B in three different populations
|Diabetic patients, n (% male)||528 (58)||329 (43.5)||257 (52.1)||402 (67.9)||1187 (56.6)||1516 (53.8)|
|Control subjects, n (% male)||542 (47.6)||238 (42.0)||285 (50.9)||434 (72.8)||1261 (51.6)||1499 (50.1)|
|Age diabetic patients (years)||60.0 ± 11.0||59.4 ± 9.9||58.1 ± 11.4||63.1 ± 10.8|| || |
|Age controls (years)||56.0 ± 10.0||57.5 ± 8.3||55.7 ± 3.4||64.4 ± 6.5|| || |
|BMI diabetic patients (kg m−2)||29.0 ± 5.0||23.9 ± 3.5||28.0 ± 4.8||28.7 ± 5.1|| || |
|BMI controls (kg m−2)||25.0 ± 8.0||23.8 ± 3.1||26.6 ± 3.9||26.5 ± 3.7|| || |
|Pro387Leu variant in controls, n (%)||5/542 (0.92)||2/238 (0.84)||3/285 (1.05)||9/434 (2.07)||17/1261 (1.35)||19/1499 (1.27)|
|Allelic frequency in controls (%)||0.46||0.42||0.53||0.99||0.67||0.63|
|Pro387Leu variant in diabetic patients, n (%)||14/528 (2.65), P = 0.0373*||2/329 (0.61), P > 0.9999||1/257 (0.389), P = 0.63||9/402 (2.24), P > 0.9999||24/1187 (2.02), P = 0.21||26/1516 (1.72), P = 0.37|
|Allelic frequency in diabetic patients (%)||0.987||0.3||0.2||0.99||1.0||0.86|
In the population studied nine subjects carried the Pro387Leu polymorphism in the control group (allelic frequency of 0.99%). Interestingly, in the diabetic group also nine subjects were Pro387Leu carriers (allelic frequency of 0.99%); Fisher's exact P-value for difference >0.9999 (Table 1).
Table 1. Genotype and allele frequencies of the Pro387Leu genotypes in PTB-1B
|Pro/Pro||818 (97.85)||393 (97.76)||425 (97.93)|
|Pro/Leu|| 18 (2.15)|| 9 (2.24)|| 9 (2.07)|
|Allelic frequency (%)|| 0.99|| 0.99|| 0.99|
|P-value|| ||>0.9999|| |
There was no difference between age (P = 0.72) and sex distribution (P = 0.94) between the diabetic and the control group. Smokers were more frequent in the diabetic group (18.7% vs. 13.4%, P = 0.038). Body mass index was significantly higher in diabetics than in controls (28.7 ± 5.1 vs. 26.5 ± 3.7 kg m−2; P < 0.0001). Diabetics had significantly higher systolic (143 ± 21 vs. 138 ± 17 mmHg; P = 0.003) but not diastolic blood pressure (83 ± 11 vs. 83 ± 11 mmHg; P = 0.65). The differences within the groups between carriers of the Pro387Leu variant are shown in Table 2. There were no statistically significant differences between genotypes in BMI, cholesterol concentrations and blood pressure. Triglyceride levels were significantly higher in Pro387Leu carriers, both in diabetics (P < 0.05) and in nondiabetic controls (P < 0.05).
Table 2. Clinical and biochemical characteristics of patients with diabetes mellitus type 2 and age- and sex-matched control subjects classified according to genotype of the Pro387Leu variant of the PTP-1B gene
|Male/female (n)||393 (226/167)||9 (5/4)||>0.9999||425 (240/185)||9 (8/1)||0.0847|
|Nonsmokers/smokers, n (%)||320/73 (81.4/18.6)||7/2 (77.8/22.2)||0.68||369/56 (86.8/13.2)||7/2 (77.8/22.2)||0.34|
|Age (years)||63.2 ± 10.8||58.1 ± 6.0||–||64.4 ± 6.5||65.3 ± 6.1||–|
|BMI (kg m−2)||28.7 ± 5.0||31.2 ± 6.5||0.21||26.4 ± 3.7||27.5 ± 2.6||0.22|
|Positive family history for diabetes, n (%)||188 (47.8)||2 (22.2)||0.18||127 (29.9)||1 (11.1)||0.29|
|Blood pressure systolic (mmHg)||143 ± 21||136 ± 15||0.36||138 ± 17||145 ± 14||0.23|
|Blood pressure diastolic (mmHg)||83 ± 11||90 ± 10||0.095||83 ± 11||91 ± 6||0.034|
|Total cholesterol (mg dL−1)||249 ± 84||277 ± 88||0.25||269 ± 64||249 ± 48||0.54|
|LDL cholesterol (mg dL−1)||139 ± 45||159 ± 80||0.83||172 ± 67||156 ± 35||0.53|
|HDL cholesterol (mg dL−1)||50 ± 17||42 ± 17||0.28||61 ± 17||51 ± 11||0.14|
|Triglycerides (mg dL−1)||258 ± 219||420 ± 263||0.038||178 ± 135||207 ± 54||0.0423|
|Lipoprotein a (mg dL−1)||32.6 ± 49.7||34.4 ± 47.6||0.84||39.9 ± 69.1||52.2 ± 112.1||0.40|
|Atherosclerotic disease, n (%)a||137 (34.9)||6 (66.7)||0.074||37 (8.7)||1 (11.1)||0.57|
|Positive family history for premature atherosclerotic disease, n (%)||58 (14.8)||3 (33.3)||0.14||76 (17.9)||2 (22.2)||0.67|
As expected, diabetics had a significantly higher rate of atherosclerotic disease (35.6% vs. 8.8%; P < 0.0001). There was a tendency, which did not reach significance (P = 0.074), for a higher rate of atherosclerotic disease in Pro387Leu carriers within the diabetic cohort, which might be due to the limited number of subjects. Based on this tendency we included the polymorphism in a multiple regression model analysing the factors influencing atherosclerotic disease. The rate of a positive family history for premature atherosclerotic disease was not different either between diabetics and controls (15.2% vs. 18.0%; P = 0.31) or between carriers and noncarriers of the aforementioned polymorphism.
To analyse the impact of the Pro387Leu variant on diabetes and atherosclerotic disease, we used the logistic modelling approach. A first model was designed to investigate parameters that may be associated with the presence of DM-2. We chose the parameters positive family history for diabetes, BMI and the Pro387Leu variant. Only family history (OR 2.1; 95% CI 1.6–2.8; P < 0.0001) and BMI (OR 1.13 kg m−2; 95% CI 1.087–1.165; P < 0.0001) had a significant influence. The Pro387Leu polymorphism was not associated with the presence of DM-2 (OR 1.05; 95% CI 0.39–2.84; P = 0.92).
In a second model we identified parameters associated with the presence of atherosclerotic disease. Diabetes, hypertension, age, sex, a positive family history for premature atherosclerotic disease and the Pro387Leu polymorphism were included in the model. Lipoproteins were not included in the model because some patients were receiving lipid-lowering medications. With a total R2 of 18.4% all the above parameters except for polymorphism exerted a significant influence on the presence of atherosclerotic disease. The respective relative risks and P-values are given in Table 3.
Table 3. Logistic regression analysis of parameters that influence the presence of atherosclerotic disease in 402 patients with diabetes mellitus type 2 and 434 controls without DM-2
|Diabetes mellitus type 2||5.9||3.9–8.9||< 0.0001|
|Age (per year)||1.059||1.037–1.083||< 0.0001|
|Positive family history for premature atherosclerotic disease||2.15||1.33–3.49||0.0018|
PTP-1B is an ubiquitously and abundantly expressed enzyme, which negatively regulates insulin signal transduction [13, 49]. In this context, a Pro387Leu variant in PTP-1B causing impaired serine phosphorylation of PTP-1B in vitro has been associated with an increased risk for DM-2 in a Danish population , whilst other studies failed to replicate this association [20–22].
The purpose of the present study was to examine a potential association between the presence of the polymorphism and DM-2 in a German Caucasian population. We found no association between the polymorphism and the presence of DM-2, the results of which are in agreement with studies examining French , Finnish  and Chinese subjects , but are contradictory to the results of Echwald et al. in a Danish population . The reasons for nonreplication of association studies are numerous . Many factors such as population heterogeneity, ethnic stratification, population-specific linkage disequilibrium between markers and causal variants, sample size, variation in study design, confounding sampling bias, misclassification of phenotypes, and gene–gene and gene–environment interactions may contribute to variable association results [51, 52]. Different diagnostic criteria for diabetes, and a lack of agreement between the diagnostic criteria for diabetes between ethnically different populations might contribute to the contradictory results. Furthermore, there is no doubt that a marked publication bias towards positive association studies also exists, where the observed size of the genetic effect in the first positive report is biased upwards . Indeed, only 16% of genetic associations identified are subsequently replicated with formal statistical significance . In this context, for genetic analysis of complex traits, nominal P-values of <0.05 should be published but with clear indication that they are preliminary . Therefore, replication of associations before declaring the evidence as convincing is of utmost importance. Replication studies tend to show effect sizes smaller and closer to the true effect size than the initially published reports of association .
The allelic frequencies in the diabetic versus nondiabetic groups were 0.99% vs. 0.99% (German population), 0.99% vs. 0.46% (Danish population), 0.20% vs. 0.53% (Finnish population) and 0.30% vs. 0.42% (Chinese population) respectively (Table 4). As the Caucasian populations of Denmark, Finland and Germany could be genetically similar, we examined the association between the presence of DM-2 and the aforementioned polymorphism after combining the data of all three studies, eventually including approximately 2448 subjects (Table 4). Genotype frequencies of the Pro387Leu variant thus seem to be rather similar in diabetics and controls, reaching 2.02% and 1.35% respectively (P = 0.21). Combining all four studies (3015 subjects) yielded in a frequency of the variant of 1.27% in controls and 1.72% in diabetics (P = 0.37). The overall power of this difference was about 17%. This result does not exclude that if even larger numbers of populations are investigated, statistically significant differences could be potentially found. However, to show an association with a power of 80%, more than 11 000 subjects in each group would be needed. In conclusion, it is safe to say that there are no differences between PTP-1B genotype frequency between control subjects and patients with DM-2.
PTP-1B regulates adiposity and expression of genes involved in lipogenesis such as sterol regulatory element binding protein (SREBP) 1, spot 14, fatty acid synthase, lipoprotein lipase (LPL) and peroxisome proliferator activated receptor (PPAR) γ .
We found that the Pro387Leu carriers had elevated triglyceride (TG) levels in comparison with noncarriers both in diabetics and in controls. This finding is novel and could be due to a downregulation of LPL, as decreased PTP-1B protein is associated with LPL downregulation in PTP-1B antisense-treated ob/ob mice . Furthermore, PTP-1B overexpression upregulates SREBP-1 expression and mediates hepatic lipogenesis and postprandial hypertriglyceridemia . Thus the observed Pro387Leu-associated hypertriglyceridemia could result from modification of PTP-1B activity. A recently published study of obese French subjects, both diabetics and nondiabetics, also showed an association between polymorphism and higher TG levels . Similar to our findings there was also a lack of association between polymorphism and DM-2, BMI, cholesterol levels, and blood pressure. They also found no association with insulin resistance indices, insulinemia and glycaemia. The allelic frequency was very close to that described in the German (present study) and Danish population  (Table 4).
Increased vascular smooth muscle cell (VSMC) motility is associated with the pathogenesis of atherosclerosis. PTP-1B is a negative regulator of VSMC motility via modulation of phosphotyrosine levels . Therefore, we examined the possible association of the presence of the Pro387Leu polymorphism with the presence of atherosclerotic disease. There was a significant association between age, hypertension, diabetes mellitus and positive family history and atherosclerosis, but no association with the presence of the aforementioned polymorphism.
In conclusion, the data of the present study suggest that the Pro387Leu polymorphism in the PTP-1B gene is not associated with the presence of DM-2 in a German Caucasian population. Our meta-analysis of >3000 subjects further supports our conclusion. More studies and a subsequent meta-analysis, as carried out for the PPARγ Pro12Ala polymorphism , are needed to clarify the functional consequences of the Pro387Leu variant and to establish the relationship between genomic variation in PTP-1B and metabolic phenotypes or atherosclerotic disease.
Conflict of interest statement
No conflict of interest was declared.
Lipid Analytic Cologne (LIANCO) was supported by Bayer Vital GmbH, Leverkusen (Germany). This study was funded by the Köln Fortune Programme, University of Cologne (Germany). We thank Ms Nadine Spenrath and Ms Doris Vollmar for their excellent technical assistance.