Osteoporotic Fractures Are Associated with an 86-Base Pair Repeat Polymorphism in the Interleukin-1-Receptor Antagonist Gene But Not with Polymorphisms in the Interleukin-1β Gene

Authors


Abstract

Interleukin-1β (IL-1β) is a potent stimulator of bone resorption, and has been implicated in the pathogenesis of high bone turnover and osteoporosis. IL-1 receptor antagonist (IL-1ra) is a competitive inhibitor of IL-1β effects and the biological effects of IL-1β are therefore proportional to the ratio IL-1β/IL-1ra. The coding regions of IL-1β were examined for sequence variations by SSCP and sequencing after polymerase chain reaction (PCR) of genomic DNA. Three previously described polymorphisms (C−511-T, G3877-A and C3954-T) in the IL-1β gene were determined by restriction fragment length polymorphism (RFLP) using Ava I, Aci I, and Taq I after PCR. The 86-base pair repeat polymorphism in IL-1ra was examined by PCR and electrophoresis and the T11100-C polymorphism in the IL-1ra gene was examined by RFLP using MspA1I after PCR. All polymorphisms were related to bone mass, biochemical markers of bone turnover, and presence of fracture in a study including 389 osteoporotic patients with vertebral fractures and normal controls. Two normal women were heterozygous for a shift from cytosine to thymine (C3263-T) in exon 4 of the IL-1β gene. This substitution did not affect the amino acid sequence. We did not find other sequence variations in the IL-1β gene apart from the already known polymorphisms. The distribution of C−511-T, G3877-A, and C3954-T genotypes was similar in the osteoporotic and the normal controls. No significant differences could be shown in bone mass or bone turnover. In the IL-1ra gene almost complete linkage was confirmed between the already known polymorphisms: G1731-A, G1821-A, A1868-G, G1887-C, T8006-C, C8061-T, 86 base pair variable number tandem repeat (VNTR), A9589-T, and a new polymorphism: T1934-C. The A1A1/A3 genotypes of the IL-1ra VNTR polymorphism were significantly more frequent in osteoporotic patients (56.2%) compared with age-matched normal controls (433%) (χ2 = 4.09; p = 0.043). The relative risk of osteoporotic fractures was increased to 1.68 (95% CI, 1.01–2.77) in individuals with A1A1/A3 genotypes. Bone mineral density (BMD) of the lumbar spine was reduced in individuals with A1A1/A3 genotypes (p = 0.014, analysis of variance [ANOVA]). The difference in bone mass between A1A1/A3 and A2A1/A2 tended to increase with increasing age. T11100-C genotypes were distributed similarly in osteoporotic patients and normal controls and the polymorphism was without effect on bone mass and biochemical markers of bone turnover. In conclusion, an 86-base pair repeat polymorphism in the IL-1ra gene is associated with increased risk of osteoporotic fractures. Other polymorphisms in the IL-1ra and the IL-1β genes are not associated with osteoporotic fractures or alterations in bone mass or bone turnover.

INTRODUCTION

Osteoporosis is, in spite of new drugs, still not a curable disease. To increase the impact of prevention, a better detection of individuals at risk is needed. Osteoporosis is characterized by a combination of low bone mass and deteriorated microarchitecture in the bone tissue. Twin studies have suggested that up to 75% of the interindividual variation in bone mineral density (BMD) can be explained by genetic factors.(1) However, the genes involved are not sufficiently characterized. Recent work from our laboratory and other laboratories has brought up several candidate genes: transforming growth factor β1 (TGF-β1), collagen Iα1, vitamin D receptor, and estrogen receptor.(2–5) However, these polymorphisms or sequence variations cannot explain all cases of osteoporosis. In fact, the Sp1 binding site polymorphism in the collagen type Iα1 gene, which doubles the risk of osteoporotic fractures, has been shown to be responsible for only a few percent of the interindividual variation in bone mass.(6,7)

Figure FIG. 1..

Structure of the IL-1ra gene with previously described polymorphisms in the gene and the polymorphism described by the authors (T1934-C). *Linked polymorphisms according to Clay et al.(16) §Linked polymorphisms according to Guasch et al.(17) Nucleotides are numbered in accordance with Butcher at al. and Lennard et al.(13,15)

It is known from analysis of bone biopsies that interleukin-1β (IL-1β) is expressed in bone obtained from osteoporotic and nonosteoporotic postmenopausal women, but not in estrogen-substituted postmenopausal women.(8) IL-1β is a potent stimulator of bone resorption itself and stimulates the production of IL-6, which also has been shown to stimulate bone resorption in rodent and human osteoblasts.(9,10) The effect of IL-1β is competitively inhibited by the IL-receptor antagonist (IL-1ra), which binds to the IL-1 receptor without inducing any effects. It is therefore the ratio of IL-1β/IL-1ra that is responsible for the biological effects of IL-1β. In North American osteoporotic women with increased bone turnover, the ratio of IL-1/IL-1ra in serum tended to increase compared with normal women.(11) In ovariectomized rats, neutralization of IL-1β activity reduces bone loss.(12) Thus, IL-1β and IL-1ra seem to be involved in high turnover bone loss after menopause.

Both IL-1β and IL-1ra genes are located on chromosomes 2q13–21. The IL-1ra gene comprises four exons, with two alternative first exons; one encodes secreted IL-1ra (exon 1s), the other encodes intracellular IL-1ra (exon 1ic).(13–15) The coding regions have been examined for polymorphisms and several polymorphisms have been found.(16) They are located in the promoter region leading up to exon 1ic, G1731-A; in exon 1ic, G1821-A, A1868-G, and G1887-C; in exon 2, T8006-C; in intron 2, C8061-T and an 86-base pair repeat polymorphism; in intron 3. A9589-T; and finally in exon 4, T11100-C (Fig. 1). Furthermore, it has been shown that these polymorphisms are in linkage disequilibrium. Clay et al. found that the polymorphisms in the promoter, exon 1ic, exon 2, and the 86-base pair repeat polymorphism in intron 2 are linked and Guasch et al. found that the polymorphisms in exon 2, intron 2 (C8061-T), and exon 3 are linked.(16,17) Because T8006-C in exon 2 is included in both studies, all the polymorphisms except T11100-C in exon 4 are linked. However, linkage is population dependent and we therefore investigated if the linkage disequilibrium previously shown also could be found in a Danish population. Because this was the case, it was only necessary to examine T11100-C and one of the other polymorphisms to investigate the impact of these polymorphisms on bone metabolism. Danis et al. have found that the 86-base pair repeat polymorphism in intron 2 is associated with leukocyte production of IL-1ra.(18) We therefore chose to examine this polymorphism and T11100-C in exon 4.

The IL-1β gene comprises seven exons.(19,20) The IL-1β gene so far has only been examined for polymorphisms in a small number of individuals.(12,17) Three polymorphisms have been shown. They are located in the promoter region (C−511-T), intron 4 (G3877-A), and exon 5 (C3954-T) (Fig. 2).(17,21) Pocoit et al. have shown that the C3954-T polymorphism is associated with monocyte production of IL-1β in vitro.(22) Therefore the aim of the present study was to examine the IL-1β gene for sequence variations in the coding regions in a larger population and to evaluate the effect of possible sequence variations and the known polymorphisms in the IL-1β and IL-1ra genes on bone mass, bone turnover, and prevalence of osteoporosis.

Figure FIG. 2..

Structure of the IL-1β gene with previously described polymorphisms and the sequence variation (C3263-T), described by the authors. Nucleotides are numbered in accordance with Clark et al.(19)

MATERIAL AND METHODS

Study subjects

The study was a case control study. The osteoporotic group consisted of 163 women (mean age, 65.0 years ± 8.3 years; range, 33–79 years; premenopausal/postmenopausal, 10/153; and mean Z score [BMD lumbar spine], −1.88 ± 1.06) and 31 men (mean age, 55.0 years ± 10.8 years; range, 22–77 years; and mean Z score (BMD lumbar spine), −2.36 ± 1.47 with primary spinal osteoporosis defined by the presence of at least one nontraumatic fracture of the spine, referred to the Endocrinology Department of Aarhus University Hospital, Aarhus Amtssygehus, Denmark. The diagnosis of primary osteoporosis was made after extensive examination for secondary causes. Fracture was defined as 20% or more reduction of the anterior, posterior, or central height of a vertebra. The normal control group comprised 134 normal women (mean age, 47.2 years ± 13.7 years; range, 21–79 years; premenopausal/postmenopausal, 78/156; and mean Z score (BMD lumbar spine), 0.02 ± 1.17) and 73 normal men (mean age, 52.3 years ± 16.0 years; range, 22–78 years; and mean Z score (BMD lumbar spine), −0.17 ± 1.39) without diseases or medications that could influence bone mass or turnover. The normal controls were recruited from the local community by invitations posted at places of work, senior citizens clubs, schools, educational institutions, hospitals, and in general practitioners' offices. For screening the IL-1β gene for mutations we chose subgroups of 51 osteoporotic women (mean age, 64.1 years ± 9.3 years; range, 38–77 years) and 44 normal women (mean age, 53.3 years ± 9.5 years; range, 32–79 years). For comparison of genotype frequencies of the polymorphisms in the IL-1β and the IL-1ra genes we selected age-matched subgroups from the osteoporotic and normal groups. The age-matched groups comprised 79 osteoporotic women (mean age, 58.4 years ± 6.5 years; range, 33–65 years), 80 normal women (mean age, 56.4 years ± 7.9 years; range, 46–79 years), 30 osteoporotic men (mean age, 55.9 years ± 1 0.8 years; range, 28–78 years), and 73 normal men (mean age, 52.3 years ± 16.0 years; range, 22–78 years). The study was approved by the local ethical committee and conducted according to Helsinki Deklaration II.

Bone mass measurements

BMD of the lumbar spine and the hip were assessed using dual-energy X-ray absorptiometry (DXA) on a Hologic 1000 (Hologic Europe, Belgium) or a Norland (Gammatec, Denmark) bone densitometer. The coefficients of interassay and intra-assay variation on Hologic 1000 were 0.5% and 0.4% for the lumbar spine and 3.1% and 2.4% for the femoral neck. The figures for the Norland were 0.9% and 1.1% for the lumbar spine and 1.1% and 0.8% for the femoral neck. The results obtained on the Norland densitometer was corrected for the known difference in BMD values obtained in Norland and Hologic densitometers.(23)

Biochemical markers of bone turnover

Serum and urinary samples were available in 295 and 228 participants, respectively. Serum samples were collected in the morning after an 8-h fasting period. All samples from osteoporotic women were collected before institution of antiosteoporotic treatment. Urine samples were 24-h samples.

S–cross-linked collagen I carboxyterminal telopeptide (s-ICTP) was measured by an equilibrium radioimmunoassay; the intra-assay CV was 5% and the interassay CV was 6%.(24) Hydroxyproline (dU-OHP) was measured spectrophotometrically with p-dimethylaminobenz-aldehyde substrate according to the manufacturer's instruction (Organon Tecnica, B.V. Boxtel, Holland). To avoid contribution of dietary collagen, participants had to keep a diet free of gelatin 12 h before and during the 24 h of collection. To compensate for sampling error, OHP excretion was expressed as a ratio relative to creatinine. Bone-specific alkaline phosphatase (s-BAP) was lectin-precipitated and analyzed spectrophotometrically using p-nitrophenyl-phosphate as substrate according to recommendations from the Scandinavian Committee on Enzymes.(25) The intra-assay CV was 8% and the interassay CV was 25%. Osteocalcin (s-BGP) was determined by a radioimmunoassay using rabbit antiserum against bovine bone-gla-protein.(26) The intra-assay CV was 5% and the interassay CV was 10%. S-Carboxyterminal propeptide of human type I procollagen (s-PICP) was measured by a radioimmunoassay from Farmos Diagnostica (Oulunsalo, Finland). The intra-assay CV was 3%, and the interassay was CV 5% (detection limit =1.2 μg/liter).(27)

Table Table 1.. Primer Sequences and Length of the PCR Products
ExonPrimer positionPrimer sequenceLength of PCR product
  1. Primer positions are given relative to exon boundaries. Downstream, –; upstream, +.

2−88 to −67CAT GAC TGC CAT GCA CTG GAT G 
 + 80 to +59CTG GTC TCC AAG CAC AGA TAA C230
 3−127 to −106GTT CTA GGC AGC TTT GAG AGG C
 +93 to +72ACC ATG GCA TCA AAG TGG CCC A272
 4−61 to −40GTC AAA GCA TGG TTC CTG GCA G
 +49 to +28TTC CAA GAG GAC ACA AGT GGA G312
 5−104 to −83CCG TAT ATG CTC AGT GTC CTC C
 +53 to +32AAT TAG CAA GCT GCC AGG AGG C323
 6−44 to −23TGC ACT GCT GTG TCC CTA ACC A
 +49 to +28AGT GGT AGC AGG AGG CTG AGA A224
 7−92 to −71TCA GTT TCC TTT CTG GCC AAC C
 +40 to +19AGT CCA CAT TCA GCA CAG GAC T345

Single stranded conformation polymorphism

DNA was isolated from whole blood leukocytes as described by Kunkel et al.(28) For the six coding exons (exons 2–7) of the IL-1β gene pairs of upstream and downstream primers were constructed located in the intron sequences approximately 20 base pairs apart from the particular exon (Table 1).(19) Polymerase chain reactions (PCRs) were performed in a final volume of 100 μl containing genomic DNA, 200 ng; AmpliTaq DNA polymerase (PE Biosystems, Denmark), 2.5 U; mononucleotides (deoxyadenosine triphosphate [dATP], deoxycytosine triphosphate [dCTP], deoxyguanosine triphosphate [dGTP], and deoxythymidine triphosphate [dTTP]; PE Biosystems, Denmark), 0.2 μmol each; redivue-32P-dATP, 1 μCi (Amersham Corp., Amersham, U.K.); 10× amplification buffer (15 mM MgCl2, 500 mM KCl, 100 mM Tris-HCl, pH 8.3, and 0.01% [wt/vol] gelatin [PE Biosystems, Denmark]), 10 μl; and primers, 20–40 pmol of each. The mixture was overlaid with 30 μl mineral oil. PCR was conducted in an automated Perkin-Elmer Thermocycler 480. PCR conditions were denaturation at 95°C for 5 minutes, followed by 35 cycles of denaturation at 95°C for 1 minute, annealing at 58°C for 1 minute, primer extension at 74°C for 1 minute, and finally primer extension at 74°C for 8 minutes. The amplified DNA was denatured by adding a solution of 95% formamide and 10 mM NaOH and placed on ice for 5 minutes. The denatured DNA was then loaded in a mutation detection enhancent (MDE)-polyacrylamide gel (FMC BioProducts Europe, Denmark), using 0.6× trisborate EDTA (TBE) as a buffer. Electrophoresis was performed at 4 W constant power for 16 h.(29) The gel was dried on filter paper before autoradiography. Semiautomated solid-phase sequencing was performed using AmpliTaq FS (PE Biosystems, Denmark) and four fluorescent dideoxynucleotides on an automated ABI 377 (PE Biosystems, Denmark) sequencer as previously described.(2) Sequencing was performed in all patients or controls with altered SSCP patterns and in five patients/controls with normal SSCP pattern. In the exons without altered SSCP patterns, sequencing was performed in five randomly selected patients or controls. The number of bases and amino acids is in accordance with the published sequence by Clark et al.(20)

Sequencing of IL-1ra exon 1c

Sequencing of exon 1c was performed using AmpliTaq FS, four fluorescent dideoxynucleotides on an automated ABI 377 sequencer as previously described.(2,16)

86-Base pair polymorphism in IL-1ra

The 86-base pair repeat polymorphism in the second exon of the IL-1ra gene was examined by a PCR-based method as previously described(30). In short, PCRs were performed in a final volume of 50 μl containing 200 ng genomic DNA and AmpliTaq DNA polymerase using standard conditions on a Perkin-Elmer Thermocycler 2400 (PE Biosystems, Denmark): 35 cycles of 94°C for 1 minute, 60°C for 1 minute, and 70°C for 2 minutes. Before the first cycle initial denaturation was performed at 95°C for 2 minutes and the last cycle was followed by an extension step of 12 minutes at 70°C. The PCR products were analyzed on a 3% agarose gel. Three different alleles were identified: A1, 4 repeats; A2, 2 repeats; and A3, 5 repeats.

Table Table 2.. Distribution of Ava I, BsoFI, and Tag I Restriction Site Genotypes in Age-Matched Osteoporotic Patients and Normal Individuals
 MenWomenAll subjects
 OsteoporosisControlOsteoporosisControlOsteoporosisControl
Number30737980109153
Age55.9 ± 10.852.3 ± 16.058.4 ± 6.556.4 ± 7.957.5 ± 8.154.4 ± 12.7
AA4 (13.3%)8 (11.0%)9 (11.4%)11 (13.8%)13 (11.9%)18 (12.4%)
Aa9 (30.0%)33 (45.2%)38 (48.1%)32 (40.0%)47 (43.1%)65 (42.5%)
aa17 (56.7%)32 (43.8%)32 (40.5%)37 (46.2%)49 (45.0%)69 (45.1%)
χ2χ2 = 2.01χ2 = 1.05χ2 = 0.01
p valuep = 0.37p = 0.59p = 0.99
BB17 (56.7%)29 (39.7%)30 (38.0%)32 (40.0%)47 (43.1%)61 (39.9%)
Bb10 (33.3%)36 (49.3%)39 (49.4%)37 (46.3%)49 (45.0%)73 (47.7%)
bb3 (10.0%)8 (11.0%)10 (12.7%)11 (13.8%)13 (11.9%)19 (12.4%)
χ2χ2 = 2.60χ2 = 0.16χ2 = 0.28
p valuep = 0.27p = 0.92p = 0.87
TT3 (10.0%)4 (5.5%)6 (7.6%)7 (8.8%)9 (8.3%)11 (7.2%)
Tt13 (43.3%)28 (38.4%)26 (32.9%)26 (32.5%)39 (35.8%)54 (35.3%)
tt14 (46.7%)41 (56.2%)47 (59.5%)47 (58.8%)61 (56.0%)88 (57.5%)
χ2χ2 = 1–13χ2 = 0.07χ2 = 0.13
p valuep = 0.57p = 0.97p = 0.94

Restriction site polymorphisms in IL-1β and IL-1ra

The polymorphisms (C−511-T, G3877-A, and C3954-T) in the IL-1β gene and T8006-C, C8061-T, A9589-T, and T11100-C in the IL-1ra gene were examined by PCR-based methods as previously described.(17,21) In short, PCR was performed in a final volume of 25 μl containing 150 ng genomic DNA and AmpliTaqGold DNA polymerase using standard conditions on a Perkin-Elmer Thermocycler 2400. The PCR products were subsequently digested with the restriction enzymes according to manufacturers' instructions and analyzed on a 3% agarose gel. Absence and presence of the restriction site is denominated with upper-case and lowercase letters.

Statistical analysis

Differences in prevalence of the genotypes between osteoporotic patients and age-matched normal controls were tested using the χ2-test. The effect of genotype and combined genotypes (haplotypes) on BMD and levels of biochemical markers were evaluated by analysis of variance (ANOVA) and linear regression. Differences in BMD and levels of biochemical markers between groups were tested using Students t-test for unpaired data. Linkage between the different polymorphisms was examined by the χ2-test of actual and expected distribution of haplotypes. Expected prevalence was calculated from the prevalence of each genotype (prevalence [AABB] = prevalence [AA] × prevalence [BB]). The level of significance was set at 0.05.

RESULTS

IL-1β screening for sequence variations

PCR-SSCP analysis of exon 4 and surrounding intron sequences revealed two different patterns: A and B. Sequencing of the pattern A revealed a shift from cytosine to thymine at position 72 in exon 4 (C3263-T) (corresponding to position 3263 in the IL-1β complementary DNA [cDNA]) (Fig. 2). C3263 is the last in the triplet: TAC encoding tyrosine and a base change to thymine does not alter the amino acid sequence because both TAC and TAT encodes tyrosine. Sequencing of pattern B revealed the wild-type sequence.

We found C3263-T in 2 of 50 normal women and but not in any of the osteoporotic women. The two normal women were both heterozygous for the sequence variation. There were no differences in BMD of the lumbar spine and the femoral neck or biochemical markers of bone turnover between these two women and the normal women without C3263-T.

Apart from the previously described C3954-T in exon 5 we did not find other sequence variants in the coding exons of IL-1β. In the intron sequence leading up to exon 7, we confirmed the sequence published by Clark et al. with a cytosine in position 6188.(20) This base was not included in the first sequence published by Bensi et al.(19)

IL-1β polymorphisms

Distribution of Ava I restriction site (C−511-T), BsoFI restriction site (G3877-A), and Taq I restriction site (C3954-T) genotypes in osteoporotic patients and normal controls is shown in Table 2. The allele frequencies of the three polymorphisms are in Hardy-Weinberg equilibrium (χ2 = 2.03, p = 0.36; χ2 = 0.01, p = 0.99; and χ2 = 0.11, p = 0.95, respectively). There was no difference between osteoporotic patients and normal individuals in prevalence of these genotypes.

Table Table 3.. BMD and Biochemical Markers of Bone Turnover in Ava I, BsoFI, and Tag I Genotypes
 NumbersaAge (years)BMD lumbar spine (g/cm2)BMD femoral neck (g/cm2)s-BAP (U/l)s-BGP (ug/l)s-PICP (ug/l)dU-OHP/Creatinine (umol/mmol)s-ICTP (ug/l)
  1. Values are given as means ± SD.

  2. aNumber of individuals who had BMD/serum/urinary markers of bone turnover determined.

  3. *p = 0.036 compared with individuals with BB genotype.

AA56/46/3256.2 ± 13.30.835 ± 0.2230.694 ± 0.16060 ± 2315.7 ± 7.7137 ± 4421.7 ± 7.23.4 ± 1.1
Aa154/117/9455.8 ± 14.70.849 ± 0.2100.688 ± 0.31867 ± 2916.4 ± 6.6141 ± 5120.9 ± 8.83.1 ± 1.1
aa168/126/9854.4 ± 16.50.848 ± 0.2220.706 ± 0.15470 ± 3017.0 ± 7.1143 ± 4421.4 ± 8.33.2 ± 1.1
ANOVA  p = 0.92p = 0.81p = 0.13p = 0.53p = 0.77p = 0.90p = 0.39
BB169/127/9455.1 ± 15.40.862 ± 0.2250.692 ± 0.30562 ± 2515.8 ± 6.5138 ± 4920.8 ± 8.43.1 ± 0.9
Bb173/138/10456.4 ± 13.70.834 ± 0.2080.700 ± 0.15771 ± 3117.0 ± 7.4144 ± 4421.4 ± 8.03.3 ± 1.3
bb46/29/3057.1 ± 12.50.832 ± 0.2120.701 ± 0.14571 ± 26*17.3 ± 7.1142 ± 4822.3 ± 9.33.2 ± 0.9
ANOVA  p = 0.43p = 0.95p = 0.04p = 0.38p = 0.65p = 0.65p = 0.26
TT28/25/2156.8 ± 12.30.884 ± 0.2510.720 ± 0.15869 ± 2916.5 ± 6.9142 ± 3620.6 ± 6.73.1 ± 1.0
Tt146/112/6956.2 ± 15.10.846 ± 0.2250.698 ± 0.15468 ± 3116.5 ± 7.0138 ± 4220.9 ± 8.13.1 ± 1.0
tt214/161/7555.6 ± 14.10.841 ± 0.2050.693 ± 0.28366 ± 2716.5 ± 7.1143 ± 5121.6 ± 8.73.2 ± 1.1
ANOVA  p = 0.62p = 0.84p = 0.85p = 1.00p = 0.76p = 0.78p = 0.54
Table Table 4.. Distribution of Combined IL-1β Genotypes
 Ava I–BsoFIAva I–Taq IBsoFI–Taq I
Combined genotypesaOsteoporosisNormalExpectedbOsteoporosisNormalExpectedbOsteoporosisNormalExpectedb
  1. a(1) uppercase-uppercase, (2) uppercase-lowercase, and (3) lowercase-lowercase letter.

  2. bThe expected number of individuals is calculated from genotype prevalences of the individual genotypes.

111820138928108
12212915212711019
1310184203018102
21182544028252737
22243750142238142442
2321133039610011
3191447109122259
3233534240344467
33101481565121917
Osteoporosis vs. normalχ2 = 3.45 χ2 = 5.34 χ2 = 2.09 
 p = 0.90 p = 0.72 p = 0.84 
Actual vs. expectedχ2 = 114.3 χ2 = 104.8 χ2 = 35.8 
 p = 4.85 × 10−21 p = 4.47 × 10−19 p = 1.90 × 10−5 

In subjects with AA-genotype BMD of the lumbar spine was 0.835 g/cm2 ± 0.223 g/cm2, 0.849 g/cm2 ± 0.210 g/cm2 in subjects with Aa genotype and 0.848 g/cm2 ± 0.222 g/cm2 in subjects with aa genotype (NS). Comparison of BMD of the femoral neck between the three genotypes revealed a similar result (Table 3). Biochemical markers of bone turnover were not different between the three genotypes.

BMD of the lumbar spine was 0.862 g/cm2 ± 0.225 g/cm2 in subjects with BB genotype, 0.834 g/cm2 ± 0.208 g/cm2 in subjects with Bb genotype, and 0.832 g/cm2 ± 0.212 g/cm2 in subjects with bb genotype (NS). BMD of the femoral neck also was comparable between the genotypes. s-BAP was reduced in subjects with BB genotype 62 U/liter ± 25 U/liter compared with 71 U/liter ± 31 U/liter, and 71 U/liter ± 26 U/liter in subjects with Bb and bb genotypes, respectively (p = 0.039, ANOVA). No differences in the other four markers of bone turnover could be shown (Table 3).

Table Table 5.. Genotypes in 35 Normal Individuals Screened for All Known Polymorphisms in the IL-1ra Gene
NumberG1731AG1812AA1868GG1887CT1934CT8006CC8061TVNTRA9589TT11100C
2G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, C
6G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
8G, AG, AA, GG, CT, CT, CC, TA3, A2A, TT, T
9G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
10G, GG, GA, AG, GT, TT, TC, CA1, A1A, AC, C
11G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, T
14G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
27G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, C
28G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
29G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
30G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, T
51G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
59G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, T
83G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
87G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, C
89G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, T
90A, AA, AG, GC, CC, CC, CT, TA2, A2T, TT, T
91G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
92G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
93G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
100G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
103G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
106G, AG, AA, GG, CT, CT, CC, TA1, A2A, TC, C
131A, AA, AG, GC, CC, CC, CT, TA2, A2T, TT, T
132G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
133G, GG, GA, AG, GT, TT, CC, TA1, A2A, TT, T
136G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
171G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
172G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, T
174G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, T
178G, AG, AA, GG, CT, CT, CC, TA1, A3A, TT, T
186G, AG, AA, GG, CT, CT, CC, TA1, A2A, TT, C
188G, GG, GA, AG, GT, TT, TC, CA1, A1A, AC, C
189G, GG, GA, AG, GT, TT, TC, CA1, A1A, AT, C
208A, AA, AG, GC, CC, CC, CT, TA2, A2T, TT, T

BMD of the lumbar spine was 0.884 g/cm2 ± 0.251 g/cm2 in individuals with TT genotype, 0.846 g/cm2 ± 0.225 g/cm2 in individuals with Tt genotype, and 0.841 g/cm2 ± 0.205 g/cm2 in individuals with tt genotype (NS). BMD of the femoral neck was comparable between individuals with TT, Tt, and tt genotypes. Furthermore, no differences could be found between genotypes in biochemical markers (Table 3).

Genotypes of the three polymorphisms were combined two and two and all three. There was no difference in prevalence of these combined genotypes or haplotypes between osteoporotic patients and normal controls (Table 4). Furthermore, no differences could be shown in BMD or biochemical markers of bone turnover between these combined genotypes (data not shown). The distribution of the combined genotypes (haplotypes) is not random (χ2 = 35.8–114.3; p = 1.90 × 10−5 to 4.85 × 10−21); the polymorphisms are in linkage disequilibrium. When all three genotypes were combined a similar result was obtained (χ2 = 109.8; p = 5.98 × 10−19).

IL-1ra polymorphisms

Linkage disequilibrium of the nine known polymorphisms in IL-1ra, G1731-A, G1821-A, A1868-G, G1887-C, T8006-C, C8061-T, 86-base pair variable number tandem repeat (VNTR), A9589-T, and T11100-C was examined in a random sample of 35 normal individuals (Table 5). Sequencing of exon 1c revealed a new polymorphism: T1934-C. All polymorphisms except T11100-C are in complete linkage disequilibrium in 34 of the 35 normal individuals examined. The 86-base pair VNTR allele A1 (4 repeats) and A3 (5 repeats) are associated with the same alleles of the other polymorphisms; allele A2 (2 repeats) is associated with the other alleles. In individual number 133 a crossover has occurred between T1934-C and T8006-C.

The three different alleles of the 86-base pair repeat in IL-1ra (A1 = 4 repeats, A2 = 2 repeats, and A3 = 5 repeats) resulted in five different genotypes, which were distributed as shown in Fig. 3. The allele frequencies were in Hardy-Weinberg equilibrium (χ2 = 4.23; p = 0.37). Because the A1 and A3 alleles can be considered similar in evolutionary terms we combined these alleles before further analysis. In the osteoporotic population 56.2% were A1A1/A3 compared with 43.3% in the normal controls (χ2 = 4.32; p = 0.12). Another way to interpret repeats is to look at them as a continuous varible and if individuals were divided into two groups with a total number of repeats below or above average, group 1 (low number of repeats [4–7]) comprised the A2A2, A2A1, and A2A3 genotypes (A2A1/A2/A3) and group 2 (high number of repeats [8–10]) comprised the A1A1 and A1A3 genotypes (A1A1/A3). In the osteoporotic population 56.2% had a number of repeats above average compared with 43.3% in the normal controls (χ2 = 4.09; p = 0.043). This difference also was reflected in a significantly increased odds ratio of osteoporotic fractures in individuals with A1A1 or A1A3 genotypes of 1.68 (95% CI, 1.01–2.77).

Figure FIG. 3..

Distribution of an 86-base pair repeat polymorphism in the IL-1ra genotypes in osteoporotic patients and normal individuals.

BMD of the lumbar spine was higher in individuals with A2A1/A3 genotypes (0.879 ± 0,228; p = 0.014, ANOVA) compared with both A1A1/A3 (0.828 ± 0.205; p = 0.03) and A2A2 (0.763 ± 0.202, p = 0.02) (Table 6). BMD of the femoral neck showed the same tendency. The s-ICTP was significantly increased in individuals with the A2A2 genotype (p = 0.04, ANOVA); the other biochemical markers were unaffected.

Distribution of MspA1I restriction site (T11100-C) genotypes in osteoporotic patients and normal controls was comparable (Table 7). The allele frequencies are in Hardy-Weinberg equilibrium (χ2 = 0.11; p = 0.94). There were no significant differences between the genotypes in BMD or in biochemical markers of bone turnover (Table 8).

Combining the genotypes of the MspA1I restriction site polymorphism and the 86-base pair repeat polymorphism in the IL-1ra gene resulted in nine different genotypes distributed as shown in Table 9. There was no difference in prevalence of these combined genotypes or haplotypes between osteoporotic patients and normal individuals.

Table Table 6.. BMD and Biochemical Markers of Bone Turnover in IL-1ra 86-Base Pair Repeat Genotypes
 A1A1, A1A3, and A3A3A2A1 and A2A3A2A2ANOVA
  1. Values are given as means ± SD.

  2. aNumber of individuals who had BMD/serum/urinary markers of bone turnover determined.

  3. bp = 0.022 compared with individuals with A2A2 genotype and p = 0.027 compared with individuals with A1A1, A1A3, and A3A3 genotypes.

  4. cp = 0.019 compared with individuals with A1A1, A1A3, and A3A3 genotypes andp = 0.046 compared with individuals with A2A1 and A2A3 genotypes.

Numbera188/124/121169/122/8223/16/15 
Age56.8 ± 13.654.5 ± 15.459.8 ± 12.9 
BMD lumbar spine (g/cm2)0.828 ± 0.2050.879 ± 0.228b0.763 ± 0.202p = 0.014
BMD femoral neck (g/cm2)0.679 ± 0.2860.719 ± 0.1650.655 ± 0.185p = 0.217
s-BGP (μg/l)17.6 ± 9.816.5 ± 6.814.9 ± 7.7p = 0.364
s-BAP (U/l)66 ± 2967 ± 2867 ± 29p = 0.994
s-PICP (μg/l)147 ± 77143 ± 45127 ± 52p = 0.336
s-ICTP (μg/l)3.1 ± 1.03.2 ± 1.03.8 ± 1.9cp = 0.040
U-OHP/Cr (μmol/mmol)22.0 ± 10.521.8 ± 10.326.0 ± 10.2p = 0.338
Table Table 7.. Distribution of MspA1I Restriction Site Genotypes in Age-Matched Osteoporotic Patients and Normal Individuals
 MenWomenAll subjects
 OsteoporosisControlOsteoporosisControlOsteoporosisControl
Number30737980109153
Age55.7 ± 11.052.3 ± 16.058.2 ± 6.456.1 ± 7.657.5 ± 7.954.3 ± 12.4
MM12 (40.0%)36 (49.3%)38 (48.1%)47 (58.8%)50 (45.9%)83 (54.2%)
Mm16 (53.3%)32 (43.8%)34 (46.8%)25 (31.3%)50 (45.9%)57 (37.3%)
mm2 (6.7%)5 (6.8%)7 (8.9%)8 (10.0%)9 (8.3%)13 (8.5%)
χ2χ2 = 0.81χ2 = 2.39χ2 = 2.04
p-valuep = 0.67p = 0.30p = 0.36
Table Table 8.. BMD and Biochemical Markers of Bone Turnover in the MspA1I Genotypes
 MMMmmmANOVA
  1. Values are given as means ± SD.

  2. aNumber of individuals who had BMD/serum/urinary markers of bone turnover determined.

Numbera190/154/113161/116/9238/25/23 
Age (years)54.7 ± 14.257.5 ± 14.355.6 ± 14.9 
BMD lumbar spine0.865 ± 0.2170.825 ± 0.2180.831 ± 0.195p = 0.20
BMD femoral neck0.719 ± 0.1590.669 ± 0.3070.704 ± 0.150p = 0.13
s-BAP (U/l)64 ± 2769 ± 3072 ± 31p = 0.22
s-BGP (μg/l)16.2 ± 6.716.8 ± 7.117.6 ± 8.7p = 0.61
s-PICP (μg/l)139 ± 46143 ± 48143 ± 51p = 0.83
U-OHP/Cr (μmol/mmol)20.2 ± 8.022.1 ± 8.922.7 ± 6.9p = 0.17
s-ICTP (μg/l)3.2 ± 1.23.1 ± 1.13.0 ± 0.8p = 0.53

BMD of the lumbar spine differed between haplotypes (ANOVA = 0.046; Table 10). Multiple linear regression (using the five most common haplotypes) revealed that neither of the two polymorphisms was an independent predictor of BMD, but together they were significant predictors of lumbar spine BMD (p = 0.032). BMD of the femoral neck was not affected. Significant difference in levels of the biochemical markers could only be shown for s-ICTP (p = 0.016, ANOVA); however, s-BAP and s-BGP tended to be different between haplotypes (p = 0.085, ANOVA, and p = 0.056, ANOVA, respectively). Multiple linear regression revealed that urinary excretion of hydroxy-proline was correlated both to the T11100-C polymorphism and to both polymorphisms in combination (p = 0.027 and 0.042, respectively). None of the other biochemical markers of bone turnover were correlated to the two polymorphisms. The two polymorphisms are not linked but are in complete linkage equilibrium (χ2 = 4.86; p = 0.90).

Table Table 9.. Distribution of the IL-1ra Combined Genotypes in Osteoporotic Patients and Normal Individuals
 All subjects
 OsteoporosisControl
  1. Age is shown as mean ± SD and genotype frequencies as number (%).

  2. aχ2 calculated after exclusion of A2A2-Mm, A2A2-mm, and A2A1/A3-mm.

Number102151
Age57.8 ± 8.054.3 ± 12.5
A2A2-MM4 (3.9%)7 (4.6%)
A2A2-Mm0 (0.0%)0 (0.0%)
A2A2-mm1 (1.0%)0 (0.0%)
A2A1/A3-MM19 (18.6%)46 (31.5%)
A2A1/A3-Mm19 (18.6%)30 (19.9%)
A2A1/A3-mm2 (2.0%)3 (2.0%)
A1A1/A3-MM26 (25.5%)29 (19.2%)
A1A1/A3-Mm25 (24.5%)27 (17.9%)
A1A1/A3-mm6 (5.9%)9 (6.0%)
χ2aχ2 = 6.05
p valuep = 0.30

DISCUSSION

In this study, we have identified a new sequence variation in the IL-1β gene. We have found that two women without osteoporosis were heterozygous for C3263-T This substitution is without effect on the mature protein.

Previously described polymorphisms in the IL-1β gene showed comparable genotype distribution in the osteoporotic patients and the age-matched normal individuals.(17,21) The allele frequencies were comparable with previously published frequencies.(21,22,31) No effect of these polymorphisms could be shown on bone mass or bone turnover. Previously, Pociot et al. isolated mononuclear cells from peripheral blood and found that the lipopolysaccharide (LPS)-induced production of IL-1β was different in cells with different genotypes of the C3954-T polymorphism. Monocytes, homozygous for T in position 3954 secreted 22% more than heterozygous cells and 45% more than cells that were homozygous for C(22). This finding was followed by investigations of the hypothesis that this polymorphism is involved in diseases in which IL-1β plays a role in pathogenesis. However, so far only one study has been able to show linkage to this polymorphism, Huang et al. found that myasthenia gravis is associated with this polymorphism, the tt genotype (homozygous for T) was significantly more frequent in patients compared with controls.(32) Other studies have shown that this polymorphism is not associated with inflammatory bowel disease diabetes mellitus or diabetic nephropathy and the present study has shown that this as well as other polymorphisms in the IL-1β are not associated with the development of osteoporosis.(22,31,33)

It has previously been shown that an 86-base pair repeat polymorphism in the IL-1ra gene is associated with alterations in granulocyte-macrophage colony-stimulating factor (GM-CSF)-stimulated production of IL-1ra and IL-1α in monocytes.(18) Monocytes with the A2A2/A1 genotypes produced 78% more IL-1ra and 42% less IL-1α than monocytes with A1A1/A3 genotypes. The 86-base pair repeat has been shown to contain three protein binding sites: an α-interferon silencer A, a β-interferon silencer B, and an acute phase response element. These binding sites with their potential regulating effects could cause the difference in IL-1ra production.(30) However, Clay et al. showed that the polymorphism had no effect on messenger RNA (mRNA) production, whereas Mandrup-Poulsen et al. found that healthy individuals with the A1A1 genotype tended (p = 0.06) to have higher levels of IL-1ra in plasma compared with individuals with the A1A2 genotype.(16,34) These contradicting results perhaps are caused by the many differences between in vitro and in vivo situations; in vivo the intact regulation of many cytokines simultaneously may overcome a difference in the genetically determined production rate of IL-1ra. This is illustrated by the fact that monocytes with the A2A2/A1 genotypes secrete less IL-1α and more IL-1β.(35,36) Also, another possibility is that intracellular IL-1ra is important for the development of a disease and not just the secreted form that can be measured in plasma. In our normal individuals the allele frequencies were comparable with the frequencies published previously.(32,37) We found that the genotypes associated with low IL-1ra production (A1A1/A3) were significantly more frequent in patients with osteoporotic fractures (56.2%) compared with normal individuals (43.3%). Furthermore, we showed that individuals with A2A1/A3 had higher BMD of the lumbar spine compared with individuals with A1A1/A3. Keen et al. have reported that women with the A1A1 genotype displayed increased bone loss at the lumbar spine.(38) In our study we observed an increased difference in BMD of the lumbar spine between individuals with A1A1/A3 and A2A1/A3 genotypes with increasing age. The difference was −0.4% (women below 50 years), −2.2% (51–60 years), −6.6% (61–70 years), and −4.1% (above 70 years), respectively. Although these differences were not significant, they are consistent with the theory that A1A1/A3 is associated with increased postmenopausal or age-related bone loss. In early postmenopausal women, Abrahamsen et al. have shown that production of IL-1ra by monocytes is inversely correlated to bone loss and Khosla et al. found that osteoporotic women with increased bone turnover tended to have an increased IL-1/IL-1ra ratio.(11,39) Thus, our findings of increased prevalence of A1A1/A3 genotypes in osteoporotic patients and reduced bone mass in individuals with the same genotypes, mainly in the older individuals, corroborate the hypothesis that the genotype A1A1 causes reduced levels of IL-1ra and reduced levels of IL-1ra are correlated to increased postmenopausal bone loss.

Table Table 10.. BMD and Biochemical Markers of Bone Turnover in the IL-1ra-Combined Genotypes
 A2A2-MMA2A1-MMA2A1-MmA1A1-MMA1A1-MmA1A1-mmANOVA
  1. aNumber of individuals who had BMD/serum/urinary markers of bone turnover determined.

  2. *p < 0.016 compared with A2A2-MM and p < 0.009 compared with A1A1-Mm.

  3. p = 0.007 compared with A1A1-Mm and p = 0.029 compared with A1A1-mm.

  4. *p = 0.030 compared with A2A1-MM, p = 0.013 compared with A1A1-Mm, and p = 0.027 compared with A1A1-mm.

  5. §p = 0.020 compared with A2A1-Mm, p = 0.013 compared with A1A1-Mm, p = 0.004 compared with A1A1-mm, and p = 0.001 compared with A2A2-MM.

  6. p = 0.029 compared with A1A1-MM.

Numbera22/17/1484/71/3972/56/3572/56/5380/57/5129/19/18 
Age59.7 ± 13.252.2 ± 15.257.2 ± 15.556.4 ± 12.558.3 ± 13.356.2 ± 15.0 
BMD lumbar spine0.776 ± 0.2060.890 ± 0.233*0.856 ± 0.2270.851 ± 0.1980.807 ± 0.2140.822 ± 0.194p = 0.046
BMD femoral neck0.656 ± 0.1910.735 ± 0.1660.696 ± 0.1630.706 ± 0.1410.641 ± 0.4080.695 ± 0.146p = 0.239
s-BAP (U/l)67 ± 2867 ± 2865 ± 2660 ± 2474 ± 3275 ± 33p = 0.085
s-BGP (μg/l)14.9 ± 7.717.4 ± 7.315.4 ± 5.614.8 ± 5.118.1 ± 7.918.6 ± 8.6p = 0.056
s-PICP (μg/l)127 ± 52143 ± 46143 ± 41136 ± 41144 ± 55150 ± 53p = 0.702
U-OHP/Cr (μmol/mmol)26.0 ± 10.620.4 ± 8.422.2 ± 8.318.4 ± 6.4§22.4 ± 9.423.7 ± 6.9p = 0.016
s-ICTP (μg/l)3.8 ± 1.93.2 ± 0.93.1 ± 0.93.0 ± 1.03.2 ± 1.23.0 ± 0.8p = 0.160

We were not able to find an allele-dose effect on BMD; in fact BMD of the lumbar spine was lower in individuals with the A2A2 genotype than in the A2A1 genotype. Furthermore, the genotype A2A2 was found with the same frequency in osteoporotic patients as in controls. There could be more explanations for these findings; the percentage of individuals with A2A2 genotype is rather small (6%), which could justify leaving this genotype out of the analyses, because the average values of BMD and biochemical markers in this group are determined with lower accuracy than in the two larger groups. Another explanation could be that this polymorphism is not a disease-causing polymorphism but is linked to a mutation or another polymorphism in the IL-1ra or surrounding genes. This last explanation is emphasized further by the fact that several other studies have examined the impact of this polymorphism on diseases in which inheritance and the IL-1 system are involved in the pathogenesis with contradicting results. Some authors found an association between the A2 allele and diabetes, systemic lupus erytromatosus, multiple sclerosis, psoriasis, Schönlein Henoch nephritis, and alopecia, whereas others did not.(33,35,37,40–45) Furthermore, no associations have been demonstrated between the A2 allele and inflammatory bowel disease, malignant haematological diseases, myasthenia gravis or thyroid diseases(31,32,46,47).

Bioque et al. showed that the repeat polymorphism in IL-1ra is not in itself associated with inflammatory bowel disease; however, the polymorphism was in linkage disequilibrium with the C3954-T polymorphism in the IL-1β gene, and the combined genotypes predicted the disease.(48) We could not show linkage between the two polymorphisms (χ2 = 10.0; p = 0.44) or any association between the combined genotypes and prevalence of osteoporosis that could not be found in the IL-1ra genotypes alone. Combining the two IL-1ra genotypes generally reflected the differences shown by the IL-1ra genotypes alone.

In conclusion, we have shown that an 86-base pair repeat polymorphism in the IL-1ra gene is associated with osteoporotic fractures, as reflected in an odds ratio for osteoporotic fractures of 1.68 in the A1A1/A3 genotypes and reduced bone mass. The three polymorphisms in the IL-1β gene and the T11100-C in the IL-1ra gene are not associated with osteoporotic fractures, alterations in bone mass, or bone turnover.

Acknowledgements

The authors thank The Novo Nordisk Fonden, The Institute of Experimental Clinical Research, University of Aarhus, The Fonden til Lægevidenskabens Fremme, and The Danish Centre for Molecular Gerontology for financial support.

Ancillary