The gene coding for the endotoxin receptor TLR4 (toll-like receptor 4) has been sequenced recently, and the polymorphic spectrum of the gene has been elucidated (1). A polymorphism in the gene affects the inflammatory response to lipopolysaccharide (LPS) (2,3), with impact on the risk of gram-negative infections and septic shock (4,5). We have analyzed two cosegregating mutations in the gene region coding for the extracellular domain of the endotoxin receptor TLR4, characterized by a substitution at amino acid position 299 (glycine for aspartate) and another at position 399 (isoleucine for threonine). The 299Gly form of TLR4 affects the inflammatory response to LPS by exhibiting attenuated signaling leading to a dampened response to LPS (4,5). Polymorphisms in the TLR4 protein show an increased risk to develop a septic shock with gram-negative microorganisms (6), premature birth (7), and graft versus host disease after hematopoietic stem cell transplantation (8).

Polymorphisms in the TLR4 gene, however, did not reveal any association with severity of menigococcal disease (9), risk of premature rupture of membranes caused by infections (10), or asthma- and atopy-related diseases (11).

TLR4 interacts with endogenous ligands such as the stress protein hsp60 (12). Since hsp60 is an important player in chronic inflammatory conditions (12), the TLR4 polymorphism may regulate the subclinical inflammation underlying the pathogenesis of arteriosclerosis.

Indeed, a recent study reported an association of the TLR4 polymorphism with atherogenesis (13). Subjects carrying the rare 299Gly allele exhibited a lower risk of carotid atherosclerosis and a smaller intima-media thickness in the common carotid artery (13).

Parameters of mild systemic inflammation observed in association with atherosclerosis are strikingly similar to what is seen in subjects with metabolic syndrome or type 2 diabetes, notably elevated serum levels of acute-phase proteins, some inflammatory cytokines, and soluble adhesion molecules (14–17). We therefore analyzed for an association of the TLR4 gene polymorphism with features of the metabolic syndrome or with overt type 2 diabetes.

Subjects.

The Cooperative Health Research in the Augsburg Region (KORA) Survey 2000 studied a population-based sample of 4,261 subjects aged 25–74 years during 1999–2001 (14). Each study participant signed a consent form to participate in genetic studies. All study methods were approved by the ethics committee of the “Bayerische Landesärtzekammer” Munich. The sampling design followed the guidelines of three previous surveys in the same region as part of the multinational World Health Organization (WHO)-MONICA (Monitoring Trends and Determinants of Cardiovascular Disease) study. In the age range 55–74 years, 1,653 people participated in a standardized interview followed by biochemical and clinical analyses. An oral glucose tolerance test and biochemical and immunological analyses were performed as described previously (18). Acute infections (fever) or gastrointestinal illness were an exclusion criterion for the oral glucose tolerance test. Diabetes was diagnosed according to 1999 WHO criteria (18). After exclusion of all subjects with self-reported type 1 diabetes, humoral autoimmunity to glutamic acid decarboxylase, or diabetes onset in the context of pancreatitis, a total of 236 individuals with type 2 diabetes and 242 individuals with impaired glucose tolerance (IGT) were available for analyses. There were 244 normoglycemic control subjects randomly selected after matching for age and sex. This totals 722 probands. Of the diabetic patients, 120 were newly detected and did not yet receive antidiabetic treatment, of the other 116, 33% were under insulin treatment; 57% took oral antidiabetic agents (18).

Biochemical analyses.

Concentrations of C-reactive protein (CRP), SAA (serum amyloid A), and fibrinogen in plasma were determined by nephelometric assays as described (14,19,20). Serum concentrations of cytokines and circulating receptors were determined by rigidly evaluated sandwich enzyme-linked immunosorbent assay; lipid diagnostics were done by standard procedures (14).

Genotyping.

Genomic DNA was extracted from leukocytes by using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN) according to the manufacturer’s recommendation. For polymorphism analysis DNA was subjected to PCR and homogenous MassEXTEND reaction (hME; Sequenom, San Diego, CA). The identical SNPs as described were analyzed (4).

Statistical analysis.

The nonparametrical Kruskal-Wallis test was performed to test for association with characteristics of subclinical inflammation. All other variables were analyzed with ANOVA for continuous end points and logistic regression for discrete end points. All models were adjusted for age and sex.

As observed by Kiechl et al. (13), the Asp299Gly and Thr399Ile polymorphisms were in linkage disequilibrium in >90% of cases. Therefore, we report the analyses with regard to the Asp299Gly polymorphism only. Results for the polymorphism at amino acid 399 provided were not different. However, the functional defect of TLR4 is largely caused by the amino acid substitution at position 299.

Genotype analysis of the polymorphism resulting in an Asp299Gly exchange was possible in 671 of the 722 subjects leading to a call rate of 92.9%. The Asp299/Asp299 was observed in 599 cases (89.3%), the heterozygous type Asp299/299Gly was found in 70 cases (10.4%), and the homozygous type 299Gly/299Gly in 2 cases (0.3%). A bias due to the genotyping rate of 92.9% appears unlikely because both single nucleotide polymorphisms were in Hardy Weinberg equilibrium and the data are very similar to previously published data (13).

The group with type 2 diabetes or IGT exhibited characteristics of subclinical inflammation with elevated systemic levels of IL-6, IL-6 receptor, CRP, SAA, or fibrinogen (Table 1). This higher state of activation of innate immunity was not accompanied by a bias toward the high LPS responder type (TLR4 major allele, Table 1). In this context it should be noted that individuals taking anti-inflammatory drugs were not excluded because, in the age-group studied, a large fraction were taking such substances, including statins, antihypertensive medication, and antidiabetic drugs. The impact of such therapy appears limited because systemic levels of inflammatory mediators are elevated in individuals with type 2 diabetes, metabolic syndrome, or atherosclerosis (14) despite their more frequent use of such drugs.

Next, we analyzed for an association of the TLR4 polymorphism with individual parameters of the metabolic syndrome or of subclinical inflammation. Subjects with one or two alleles causing the 299Gly TLR4 did not differ from carriers of TLR4 Asp homozygotes with regard to hypertension, BMI, waist circumference, or HDL cholesterol levels (Table 2). Differences were also not observed for systemic levels of IL-6, IL-6 receptor, CRP, SAA, or fibrinogen (Table 2). Of all parameters analyzed, only the prevalence of hypertension showed a trend (P = 0.07), leaving the possibility of a mild protective effect of the Gly299 TLR4 allele. A similar analysis only regarding control and IGT subjects also failed to show an association between TLR4 genotype and metabolic or immunologic phenotype (Table 3).

The data do not confirm TLR4 allele-dependent differences in systemic IL-6 or fibrinogen concentrations reported for the Bruneck Study (13). Both the Bruneck and our study share basic characteristics of population-based sampling and a similar age range (40–80 vs. 55–74 years) and number of subjects studied. Methodological aspects probably do not account for the difference because a number of inflammatory markers were assessed in both studies. None of these came close to the significance level in our study, while most parameters did so in the Bruneck Study. A reason for this difference could not be identified; in theory small differences may be expected with regard to the genetic background, environmental factors (such as low level infections), or the sampling procedure. In this context it is noteworthy that almost all carriers of TLR4 299Gly allele are heterozygous and hence coexpress the fully functional TLR4 Asp299. Monocytes heterozygous for TLR4 do not exhibit deficient responses to LPS (3,21). Independent of this issue, the important finding of our KORA study is that there is no strong impact of the TLR4 polymorphism on major features of the metabolic syndrome.

TABLE 1

No association of TLR4 polymorphism with type 2 diabetes or IGT

Type 2 diabetesIGTControlP value
Type 2 diabetes vs. controlIGT vs. control
IL-6 (pg/ml) 2.45 (1.25–4.39) 2.29 (1.28–3.51) 1.62 (0.65–2.85) <0.0001 <0.0001 
 n 214 223 227   
IL-6 receptor (ng/ml) 155.6 (119.8–194.3) 131.0 (95.3–174.8) 126.3 (91.5–166.0) <0.0001 0.24 
 n 214 223 226   
CRP (mg/l) 2.62 (1.13–5.96) 2.35 (1.22–4.53) 1.27 (0.70–2.98) <0.0001 <0.0001 
 n 212 224 228   
SAA (mg/l) 4.3 (2.8–7.5) 4.1 (2.6–7.6) 3.4 (2.2–5.5) <0.0001 <0.0001 
 n 212 224 228   
Fibrinogen (g/l) 3.05 (2.65–3.39) 2.96 (2.58–3.36) 2.82 (2.52–3.15) 0.0002 0.022 
 n 212 224 228   
Genotype (Asp299Asp/ Asp299Gly/Gly299Gly) 196/21/0 196/29/0 207/20/2 0.98 0.27 
 n 217 225 229   
Age (years) 65.1 ± 5.2 65.4 ± 5.3 65.1 ± 5.4 — — 
 n 217 225 229   
Sex (M/F) 129/88 124/101 128/101 — — 
 n 217 225 229   
Type 2 diabetesIGTControlP value
Type 2 diabetes vs. controlIGT vs. control
IL-6 (pg/ml) 2.45 (1.25–4.39) 2.29 (1.28–3.51) 1.62 (0.65–2.85) <0.0001 <0.0001 
 n 214 223 227   
IL-6 receptor (ng/ml) 155.6 (119.8–194.3) 131.0 (95.3–174.8) 126.3 (91.5–166.0) <0.0001 0.24 
 n 214 223 226   
CRP (mg/l) 2.62 (1.13–5.96) 2.35 (1.22–4.53) 1.27 (0.70–2.98) <0.0001 <0.0001 
 n 212 224 228   
SAA (mg/l) 4.3 (2.8–7.5) 4.1 (2.6–7.6) 3.4 (2.2–5.5) <0.0001 <0.0001 
 n 212 224 228   
Fibrinogen (g/l) 3.05 (2.65–3.39) 2.96 (2.58–3.36) 2.82 (2.52–3.15) 0.0002 0.022 
 n 212 224 228   
Genotype (Asp299Asp/ Asp299Gly/Gly299Gly) 196/21/0 196/29/0 207/20/2 0.98 0.27 
 n 217 225 229   
Age (years) 65.1 ± 5.2 65.4 ± 5.3 65.1 ± 5.4 — — 
 n 217 225 229   
Sex (M/F) 129/88 124/101 128/101 — — 
 n 217 225 229   

Data are means ± SD or median (interquartile range). The respective P values result from the nonparametric Kruskal-Wallis test, while associations with the genotype were searched with logistic regression. Essentially the same outcomes were noted when means ± SD were used for statistical evaluation. The groups were matched for age and sex. Only groups of subjects that were both genotyped and phenotyped were analyzed.

View Large
TABLE 2

No association of TLR4 polymorphism with components of the metabolic syndrome or of subclinical systemic inflammation

Asp299/Asp299299Gly/Asp299; 299Gly/299GlyP value
Hypertension* (yes/no) 410/187 57/15 0.083 
 n 597 72  
BMI (kg/m2) 29.3 ± 4.3 28.8 ± 4.3 0.31 
 n 592 72  
Waist circumference (cm) 98.5 ± 11.7 98.6 ± 11.5 0.86 
 n 597 72  
Total cholesterol (mg/dl) 241.6 ± 44.6 236.7 ± 39.9 0.43 
 n 598 72  
HDL (mg/dl) 55.0 ± 15.9 53.7 ± 15.3 0.59 
 n 597 72  
LDL (mg/dl) 153.1 ± 42.0 150.0 ± 39.4 0.60 
 n 597 72  
Triglycerides (mg/dl)† 142.7 ± 82.4 138.1 ± 98.6 0.75 
 n 495 61  
IL-6 (pg/ml) 2.12 (0.93–3.52) 2.23 (1.32–3.74) 0.53 
 n 594 70  
IL-6 receptor (ng/ml) 137.4 (102.2–180.5) 136.2 (96.4–176.9) 0.73 
 n 593 70  
CRP (mg/l) 1.98 (0.94–4.40) 2.30 (0.99–4.59) 0.58 
 n 592 72  
SAA (mgl/l) 3.9 (2.5–6.8) 4.0 (2.2–6.7) 0.81 
 n 592 72  
Fibrinogen (g/l) 2.94 (2.56–3.33) 2.98 (2.64–3.39) 0.37 
 n 592 72  
Asp299/Asp299299Gly/Asp299; 299Gly/299GlyP value
Hypertension* (yes/no) 410/187 57/15 0.083 
 n 597 72  
BMI (kg/m2) 29.3 ± 4.3 28.8 ± 4.3 0.31 
 n 592 72  
Waist circumference (cm) 98.5 ± 11.7 98.6 ± 11.5 0.86 
 n 597 72  
Total cholesterol (mg/dl) 241.6 ± 44.6 236.7 ± 39.9 0.43 
 n 598 72  
HDL (mg/dl) 55.0 ± 15.9 53.7 ± 15.3 0.59 
 n 597 72  
LDL (mg/dl) 153.1 ± 42.0 150.0 ± 39.4 0.60 
 n 597 72  
Triglycerides (mg/dl)† 142.7 ± 82.4 138.1 ± 98.6 0.75 
 n 495 61  
IL-6 (pg/ml) 2.12 (0.93–3.52) 2.23 (1.32–3.74) 0.53 
 n 594 70  
IL-6 receptor (ng/ml) 137.4 (102.2–180.5) 136.2 (96.4–176.9) 0.73 
 n 593 70  
CRP (mg/l) 1.98 (0.94–4.40) 2.30 (0.99–4.59) 0.58 
 n 592 72  
SAA (mgl/l) 3.9 (2.5–6.8) 4.0 (2.2–6.7) 0.81 
 n 592 72  
Fibrinogen (g/l) 2.94 (2.56–3.33) 2.98 (2.64–3.39) 0.37 
 n 592 72  

Data are means ± SD or median (interquartile range). The first two columns display the frequencies of hyptertensive and nonhypertensive subjects in both genotype groups in the first row. The P value for hypertension was taken from analysis with a logistic regression model; the remaining components of the metabolic syndrome were analysed with ANOVA. All models were adjusted for age and sex. The P values belonging to the components of subclinical systemic inflammation were obtained with the Kruskal-Wallis test. Essentially the same outcomes were noted when means ± SD were used for statistical evaluation. Separate analyses of men or women or of type 2 diabetic cases and IGT subjects only also showed no association with the TLR4 polymorphism.

*

Hypertension was defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or current antihypertensive treatment.

†

108 patients with diabetes provided non fasting blood samples and were excluded.

View Large
TABLE 3

Analysis of the influence of the TLR4 genotype on components of the metabolic syndrome or of subclinical systemic inflammation, regarding IGT subjects and control subjects solely

Asp299/Asp299299Gly/Asp299; 299Gly/299GlyP
Hypertension* (yes/no) 254/147 37/14 0.23 
 n 401 51  
BMI (kg/m2) 28.5 ± 4.0 27.9 ± 3.8 0.26 
 n 400 51  
Waist circumference (cm) 96.1 ± 11.0 96.5 ± 11.2 0.93 
 n 402 51  
Total cholesterol (mg/dl) 245.3 ± 43.9 235.5 ± 39.1 0.14 
 n 402 51  
HDL (mg/dl) 57.4 ± 16.1 56.2 ± 16.7 0.62 
 n 401 51  
LDL (mg/dl) 156.5 ± 41.8 151.0 ± 37.7 0.39 
 n 401 51  
Triglycerides (mg/dl) 136.5 ± 78.6 125.2 ± 60.7 0.40 
 n 398 49  
IL-6 (pg/ml) 1.95 (0.85–3.12) 1.87 (1.25–3.13) 0.74 
 n 400 50  
IL-6 receptor (ng/ml) 128.3 (94.7–169.7) 122.4 (89.5–171.7) 0.49 
 n 399 50  
CRP (mg/l) 1.73 (0.88–3.73) 2.11 (0.88–3.78) 0.51 
 n 401 51  
SAA (mgl/l) 3.7 (2.4–6.5) 3.6 (2.0–6.8) 0.77 
 n 401 51  
Fibrinogen (g/l) 2.88 (2.54–3.26) 2.96 (2.53–3.27) 0.52 
 n 401 51  
Asp299/Asp299299Gly/Asp299; 299Gly/299GlyP
Hypertension* (yes/no) 254/147 37/14 0.23 
 n 401 51  
BMI (kg/m2) 28.5 ± 4.0 27.9 ± 3.8 0.26 
 n 400 51  
Waist circumference (cm) 96.1 ± 11.0 96.5 ± 11.2 0.93 
 n 402 51  
Total cholesterol (mg/dl) 245.3 ± 43.9 235.5 ± 39.1 0.14 
 n 402 51  
HDL (mg/dl) 57.4 ± 16.1 56.2 ± 16.7 0.62 
 n 401 51  
LDL (mg/dl) 156.5 ± 41.8 151.0 ± 37.7 0.39 
 n 401 51  
Triglycerides (mg/dl) 136.5 ± 78.6 125.2 ± 60.7 0.40 
 n 398 49  
IL-6 (pg/ml) 1.95 (0.85–3.12) 1.87 (1.25–3.13) 0.74 
 n 400 50  
IL-6 receptor (ng/ml) 128.3 (94.7–169.7) 122.4 (89.5–171.7) 0.49 
 n 399 50  
CRP (mg/l) 1.73 (0.88–3.73) 2.11 (0.88–3.78) 0.51 
 n 401 51  
SAA (mgl/l) 3.7 (2.4–6.5) 3.6 (2.0–6.8) 0.77 
 n 401 51  
Fibrinogen (g/l) 2.88 (2.54–3.26) 2.96 (2.53–3.27) 0.52 
 n 401 51  

Data are means ± SD or median (interquartile range). Analyses are performed as in Table 2. The first two columns display the frequencies of hyptertensive and nonhypertensive subjects in both genotype groups in the first row. For further components of the metabolic syndrome, means ± SD are shown, and the median and quartiles are shown for markers of subclinical inflammation. P values were obtained with a logistic regression model for hypertension, with ANOVA for the remaining components of the metabolic syndrome and with the Kruskal-Wallis test for components of subclinical systemic inflammation. All models were adjusted for age and sex.

*

Hypertension was defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or current antihypertensive treatment.

View Large

This work was supported by grants from the German National Genome Research Net (NGFN) platform 6 and by the GSF Research Center, the Deutsche Forschungsgemeinschaft, the European Foundation for the Study of Diabetes, the Federal Ministry of Health, the Ministry of Science and Research of North Rhine-Westfalia, and the Department of Internal Medicine II-Cardiology at the University of Ulm.

This study was supported by the KORA study group: A. Döring, T. Illig, H. Löwel, C. Meisinger, B. Thorand, and H.E.-Wichmann from the GSF National Research Center for Environment and Health, Institute of Epidemiology; R. Holle and J. John from the GSF National Research Center for Environment and Health, Institute of Health Economics and Health Care Management.

We thank Monika Wimmer, Michaela Bunge, and Petra Weskamp for excellent technical assistance. The genetic part of the work was performed in the “Genome Analysis Center” of the GSF Research Center for Environment and Health.

1.
Smirnova I, Poltorak A, Chan EK, McBride C, Beutler B: Phylogenetic variation and polymorphism at the toll-like receptor 4 locus (TLR4).
Genome Biol
 
1
:
RESEARCH002
, Epub
2000
2.
Schwartz DA: TLR4 and LPS hyporesponsiveness in humans.
Int J Hyg Environ Health
 
205
:
221
–227,
2002
3.
Schmitt C, Humeny A, Becker CM, Brune K, Pahl A: Polymorphisms of TLR4: rapid genotyping and reduced response to lipopolysaccharide of TLR4 mutant alleles.
Clin Chem
 
48
:
1661
–1667,
2002
4.
Arbour NC, Lorenz E, Schutte BC, Zabner J, Kline JN, Jones M, Frees K, Watt JL, Schwartz DA: TLR4 mutations are associated with endotoxin hyporesponsiveness in humans.
Nat Genet
 
25
:
187
–191,
2000
5.
Agnese DM, Calvano JE, Hahm SJ, Coyle SM, Corbett SA, Calvano SE, Lowry SF: Human toll-like receptor 4 mutations but not CD14 polymorphisms are associated with an increased risk of gram-negative infections.
J Infect Dis
 
186
:
1522
–1525,
2002
6.
Lorenz E, Mira JP, Frees KL, Schwartz DA: Relevance of mutations in the TLR4 receptor in patients with gram-negative septic shock.
Arch Intern Med
 
162
:
1028
–1032,
2002
7.
Lorenz E, Hallman M, Marttila R, Haataja R, Schwartz DA: Association between the Asp299Gly polymorphisms in the toll-like receptor 4 and premature births in the Finnish population.
Pediatr Res
 
52
:
373
–376,
2002
8.
Lorenz E, Schwartz DA, Martin PJ, Gooley T, Lin MT, Chien JW, Hansen JA, Clark JG: Association of TLR4 mutations and the risk for acute GVHD after HLA-matched-sibling hematopoietic stem cell transplantation.
Biol Blood Marrow Transplant
 
7
:
384
–387,
2001
9.
Read RC, Pullin J, Gregory S, Borrow R, Kaczmarski EB, di Giovine FS, Dower SK, Cannings C, Wilson AG: A functional polymorphism of toll-like receptor 4 is not associated with likelihood or severity of meningococcal disease.
J Infect Dis
 
184
:
640
–642,
2001
10.
Ferrand PE, Fujimoto T, Chennathukuzhi V, Parry S, Macones GA, Sammel M, Kuivaniemi H, Romero R, Strauss JF III: The CARD15 2936insC mutation and TLR4 896 A>G polymorphism in African Americans and risk of preterm premature rupture of membranes (PPROM).
Mol Hum Reprod
 
8
:
1031
–1034,
2002
11.
Raby BA, Klimecki WT, Laprise C, Renaud Y, Faith J, Lemire M, Greenwood C, Weiland KM, Lange C, Palmer LJ, Lazarus R, Vercelli D, Kwiatkowski DJ, Silverman EK, Martinez FD, Hudson TJ, Weiss ST: Polymorphisms in toll-like receptor 4 are not associated with asthma or atopy-related phenotypes.
Am J Respir Crit Care Med
 
166
:
1449
–1456,
2002
12.
Ohashi K, Burkart V, Flohe S, Kolb H: Cutting edge: heat shock protein 60 is a putative endogenous ligand of the toll-like receptor-4 complex.
J Immunol
 
164
:
558
–561,
2000
13.
Kiechl S, Lorenz E, Reindl M, Wiedermann CJ, Oberhollenzer F, Bonora E, Willeit J, Schwartz DA: Toll-like receptor 4 polymorphisms and atherogenesis.
N Engl J Med
 
347
:
185
–192,
2002
14.
Muller S, Martin S, Koenig W, Hanifi-Moghaddam P, Rathmann W, Haastert B, Giani G, Illig T, Thorand B, Kolb H: Impaired glucose tolerance is associated with increased serum concentrations of interleukin 6 and co-regulated acute-phase proteins but not TNF-alpha or its receptors.
Diabetologia
 
45
:
805
–812,
2002
15.
Pickup JC, Crook MA: Is type II diabetes mellitus a disease of the innate immune system?
Diabetologia
 
41
:
1241
–1248,
1998
16.
Festa A, D’Agostino R, Howard G, Mykkanen L, Tracy RP, Haffner SM: Inflammation and microalbuminuria in nondiabetic and type 2 diabetic subjects: the Insulin Resistance Atherosclerosis study.
Kidney Int
 
58
:
1703
–1710,
2000
17.
Greenberg AS, McDaniel ML: Identifying the links between obesity, insulin resistance and beta-cell function: potential role of adipocyte-derived cytokines in the pathogenesis of type 2 diabetes.
Eur J Clin Invest
 (
32 Suppl. 3
):
24
–34,
2002
18.
Rathmann W, Haastert B, Icks A, Lowel H, Meisinger C, Holle R, Giani G: High prevalence of undiagnosed diabetes mellitus in Southern Germany: Target populations for efficient screening: the KORA survey 2000.
Diabetologia
 
46
:
182
–189,
2003
19.
Rifai N, Tracy RP, Ridker PM: Clinical efficacy of an automated high-sensitivity C-reactive protein assay.
Clin Chem
 
45
:
2136
–2141,
1999
20.
Hoffmeister A, Rothenbacher D, Bazner U, Frohlich M, Brenner H, Hombach V, Koenig W: Role of novel markers of inflammation in patients with stable coronary heart disease.
Am J Cardiol
 
87
:
262
–266,
2001
21.
Erridge C, Stewart J, Poxton IR: Monocytes heterozygous for the Asp299Gly and Thr399Ile mutations in the toll-like receptor 4 gene show no deficit in lipopolysaccharide signalling.
J Exp Med
 
197
:
1787
–1791,
2003
DIABETES
2003