Transforming growth factor-β (TGF-β) is both abundant in bone and an important regulator of bone metabolism. A T→C transition at nucleotide 29 in the signal sequence region of the TGF-β1 gene results in a Leu→Pro substitution at amino acid position 10. The possible association of this polymorphism with bone mass and the prevalence of osteoporosis has now been investigated in a total of 287 postmenopausal women from two regions (Obu City, Aichi Prefecture, and Sanda City, Hyogo Prefecture) of Japan. A significant association of TGF-β1 genotype with bone mass was detected in both populations; bone mineral density (BMD) at the lumbar spine was greater in individuals with the CC genotype than in those with the TT or TC genotype. The frequency of vertebral fractures was significantly lower in individuals with the CC genotype than in those with the TC or TT genotypes. For each region, multivariable logistic regression analysis revealed that the frequency of the T allele was significantly higher in subjects with osteoporosis than in controls. Also, the serum concentration of TGF-β1 in individuals with the CC genotype was significantly higher than that in age-matched subjects with the TC or TT genotype in osteoporotic or osteopenic as well as healthy control groups. These results suggest that the T/C polymorphism of the TGF-β1 gene is one of the genetic determinants of bone mass and that the T allele is an independent risk factor for the genetic susceptibility to osteoporosis in postmenopausal Japanese women. Thus, analysis of the TGF-β1 genotype may be useful in the prevention and management of osteoporosis.
OSTEOPOROSIS, A COMMON DISEASE that is characterized by reduced bone mass and an increased risk of fracture,1 has a strong genetic component.2–4 Although polymorphisms of several genes, including those encoding the vitamin D receptor,5–8 type I collagen,9 the estrogen receptor,10 and apolipoprotein E,11 have been associated with bone mineral density (BMD), the genetic susceptibility to osteoporosis and the responsible genes have not been fully determined.
Transforming growth factor-β (TGF-β) is abundant in bone and has been implicated as an important regulator of both bone formation and resorption.12 Langdahl et al.13 showed that a one-base deletion in intron 4 (713–8delC) of the TGF-β1 gene is more frequent in individuals with osteoporosis than in normal controls. These researchers also showed that the deletion is associated with very low bone mass in osteoporotic women and with increased bone turnover in both osteoporotic and normal women. Several other polymorphisms in the TGF-β1 gene, including a T→C transition at nucleotide 29 in the region encoding the signal sequence,14,15 which results in a Leu→Pro substitution at amino acid 10, have been described. We have now examined the possible association of the T29→C polymorphism of the TGF-β1 gene with genetic susceptibility to osteoporosis in postmenopausal Japanese women.
MATERIALS AND METHODS
The study population comprised 239 consecutive unrelated postmenopausal Japanese women who either visited outpatient clinics of or were admitted to National Chubu Hospital (Obu City, Aichi Prefecture) or National Hyogo-Chuo Hospital (Sanda City, Hyogo Prefecture) between April 1996 and September 1997. Individuals with disorders known to cause abnormalities of bone metabolism, including diabetes mellitus; renal diseases; rheumatoid arthritis; and thyroid, parathyroid, and other endocrinological diseases, were excluded from the study. Women who had taken drugs such as estrogen, progesterone, glucocorticoids, bisphosphonates, and alfacalcidol were also excluded. Forty-eight healthy volunteers who did not have any serious diseases or take drugs known to affect bone and calcium metabolism were also recruited in the Obu City area. The study protocol was approved by the Committee on the Ethics of Human Research of National Chubu Hospital and National Hyogo-Chuo Hospital, and informed consent was obtained from all subjects.
Measurement of BMD
BMD at the lumbar spine (L2–L4) was measured by dual-energy X-ray absorptiometry with a DPX instrument (Lunar, Madison, WI, U.S.A.) at National Chubu Hospital or with an XR-26 machine (Norland Medical Systems, Fort Atkinson, WI, U.S.A.) at National Hyogo-Chuo Hospital. The coefficient of variance (CV) of both instruments for the lumbar spine was <1.5%. All scans were reviewed by orthopedic surgeons to exclude aortic calcification and osteoarthritic changes. The diagnosis of osteoporosis was based on the criteria recommended by the Committee for Diagnostic Criteria of Osteoporosis (Japanese Society of Bone and Mineral Research), which are either an L2–L4 BMD of <80% of the young adult (20–44 years old) reference mean in the presence of nontraumatic vertebral fracture, or an L2–L4 BMD of <70% of the young adult reference mean in the absence of nontraumatic vertebral fracture. Individuals with L2–L4 BMD of 70–80% of the young adult reference mean in the absence of vertebral fracture were diagnosed with osteopenia. According to these criteria, 129 subjects were diagnosed with osteoporosis and 69 with osteopenia. The L2–L4 BMD of 89 control subjects was >80% of that of the young adult reference mean. Vertebral fractures were diagnosed using lateral roentogenograms of the thoracic and lumbar spine on the basis of the criteria described by Riggs et al.16
Genotyping of the TGF-β1 gene
Venous blood (5 ml) was collected into tubes containing EDTA (disodium salt, 50 mM), and genomic DNA was isolated with a DNA extraction kit (Qiagen, Chatsworth, CA, U.S.A.) or an automated genomic DNA isolation system (NA-1000; Kurabo, Osaka, Japan). The genotype for the TGF-β1 gene was determined with an allele-specific polymerase chain reaction (PCR) with two sense primers (sense primer 1, 5′-CTCCGGGCTGCGGCTGCTGCT-3′; sense primer 2, 5′-CTCCGGGCTGCGGCTGCTGCC-3′) and an antisense primer (5′-GTTGTGGGTTTCCACCATTAG-3′). Reactions were performed in a total volume of 50 μl containing 0.5 μg of genomic DNA, 20 pmol of each primer, 0.2 mM each of dCTP, dTTP, dGTP, and dATP, 1 U of Taq DNA polymerase (Amplitaq Gold; Perkin Elmer, Foster City, CA, U.S.A.), 50 mM KCl, 1.5 mM MgCl2, 1.4% dimethyl sulfoxide, 0.01% gelatin, and 10 mM Tris-HCl (pH 8.3). The thermocycling procedure consisted of initial denaturation at 94°C for 5 minutes; five cycles of denaturation at 94°C for 1 minutes, annealing at 60°C for 1 minutes, and extension at 72°C for 1 minutes; 30 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 30 s; and a final extension at 72°C for 5 minutes. The PCR products were analyzed by 1% agarose gel electrophoresis and visualized by ethidium bromide staining. The expected size of the specific amplification product was 346 bp.
DNA sequencing of the polymorphic region of the TGF-β1 gene
To confirm the detection of the T/C polymorphism by the allele-specific PCR, we amplified exon 1 of the TGF-β1 gene by PCR with sense (5′-TCCTACCTTTTGCCGGGAGAC-3′) and antisense (5′-GTTGTGGGTTTCCACCATTAG-3′) primers. The reaction mixture (100 μl) contained 1 μg of genomic DNA; 100 pmol of each oligonucleotide primer; 0.2 mM each of dCTP, dTTP, dGTP, and dATP; 2 mM MgSO4; 10 mM KCl; 10 mM (NH4)2SO4; 0.1% Triton X-100; 0.1 mg/ml bovine serum albumin; 20 mM Tris-HCl (pH 8.8); and 5 U of Pfu DNA polymerase (Stratagene, La Jolla, CA, U.S.A.). The thermocycling procedure was identical to that described above. The PCR products were purified with Centricon-100 filtration devices (Amicon, Beverly, MA, U.S.A.) and directly sequenced on both strands with a fluorescence-based automated DNA sequencer (Applied Biosystems, Foster City, CA, U.S.A.).
Measurement of biochemical markers of bone turnover
Venous blood and urine samples were collected in the early morning after an overnight fast. After blood samples were centrifuged at 1600g for 15 minutes at 4°C, serum was separated and stored at −30°C until assay. Urine samples were collected in plain tubes and stored at −30°C.
As markers of bone resorption, urinary pyridinoline (Pyr) and deoxypyridinoline (Dpyr) were measured by high-performance liquid chromatography with a fluorescence detector.17 The minimun detection limit for both substances was 0.166 pmol/ml, and the intra- and interassay CVs were <3.5% and <11.7% for Pyr, and <8.4% and <10.5% for Dpyr, respectively. The values were corrected for urinary creatinine (Cr) and expressed as picomoles per micromole of Cr. Urinary Cr was enzymatically measured with a Cr test kit (Wako Chemical, Osaka, Japan).
The serum concentration of osteocalcin was measured with an immunoradiometric assay kit (Mitsubishi Chemical, Tokyo, Japan). The detection limit of this assay system was 1 ng/ml, and the intra- and interassay CVs were <3.7% and <5.1%, respectively. The activity of bone-specific alkaline phosphatase (B-ALP) in serum was measured with an immunoassay kit (Metra Biosystems, Mountain View, CA, U.S.A.). The cross-reactivity of the assay with serum isoenzymes of ALP from liver, placenta, and intestine was 3–8, 0, and 0.4%, respectively. The detection limit of the assay system was 0.7 U/l, and the intra- and interassay CVs were <5.8% and <7.6%, respectively.
Measurement of serum TGF-β1 concentration
The serum concentration of TGF-β1 was determined with an enzyme-linked immunosorbent assay kit (Amersham, Little Chalfont, Buckinghamshire, U.K.). The detection limit of this assay was 4 pg/ml, and the intra- and interassay CVs were <3.9% and <13.4%, respectively. The assay showed essentially no cross-reactivity (<1%) with TGF-β2, TGF-β3, or other cytokines.
Data are presented as means ± SD. Clinical and laboratory data were compared among TGF-β1 genotypes by one-way analysis of variance and Scheffe's multiple range test. Unpaired Student's t-test was used for comparisons between the control subjects and the patient group. Qualitative data were compared by the chi-square test. Allele frequencies were estimated by the gene counting method, and the chi-square test was used to identify significant departures from Hardy-Weinberg equilibrium. We performed multivariable logistic regression analysis to adjust risk factors; osteoporosis was a dependent variable and independent variables included age, height, body weight, years after menopause, smoking status (0 = nonsmoker, 1 = smoker), and TGF-β1 genotype (0 = CC, 1 = TC + TT). The odds ratio and 95% confidence interval (CI) were also calculated. A p value of <0.05 was considered statistically significant.
DNA sequencing of the polymorphic region in exon 1 of the TGF-β1 gene revealed the T/C polymorphism at nucleotide 29 that results in a Leu/Pro polymorphism at amino acid 10 (Fig. 1A). This nucleotide polymorphism was accurately detected by the allele-specific PCR assay (Fig. 1B). The frequencies of the TT, TC, and CC genotypes in the study population were 30.4, 53.8, and 15.8%, respectively, for individuals in Obu City, and 36.2, 50.0, and 13.8%, respectively, for those in Sanda City (Table 1). The genotype distributions in both cities were in Hardy-Weinberg equilibrium.
Table TABLE 1. BMD AND OTHER CHARACTERISTICS ACCORDING TO TGF-β1 GENOTYPE OF THE STUDY SUBJECTS IN OBU CITY AND SANDA CITY
For the subjects in both locations, age, height, body weight, years after menopause, and smoking status did not differ among the TGF-β1 genotypes, with the exception that the body weight of women with the CC genotype was significantly greater than that of those with the TC genotype in Sanda City (Table 1). The frequency of vertebral fractures in individuals with the CC genotype (9.3%) was significantly lower than that in those with the TC or TT genotypes (27.0%, p = 0.013), when data from two cities were combined. No statistically significant differences in the serum concentrations of calcium, B-ALP, or osteocalcin, or in urinary excretion of Pyr or Dpyr, were detected among genotypes, although, in both cities, there was a trend for the T allele to be associated with increased serum B-ALP and urinary excretion of Pyr and Dpyr, reflecting increased bone turnover (Table 1).
In Obu City, L2–L4 BMD in individuals with the CC genotype was significantly greater than that in those with the TT genotype (Table 1). Similarly, in Sanda City, L2–L4 BMD in women with the CC genotype was significantly greater than that in those with the TC or TT genotype. Analysis of combined data for subjects from Obu and Sanda revealed that L2–L4 BMD expressed as a percentage of the young adult reference mean was significantly higher in individuals with the CC genotype (85.1 ± 18.9%) than in individuals with the TC genotype (72.2 ± 15.6%, p < 0.0001) or in those with the TT genotype (69.7 ± 16.4%, p < 0.0001). L2–L4 BMD expressed as a percentage of the age-matched reference mean was also significantly higher in individuals with the CC genotype (107.1 ± 21.8%) than in those with the TC genotype (90.5 ± 17.8%, p < 0.0001) or the TT genotype (88.8 ± 17.3%, p < 0.0001).
Next, we compared the frequencies of the TGF-β1 genotypes between osteoporotic women and age-matched controls in Obu City (Table 2) and Sanda City (Table 3). The distribution of the TGF-β1 genotypes in control subjects in both Obu and Sanda was in Hardy-Weinberg equilibrium. In Obu, neither age nor years after menopause differed significantly between controls and patients (either osteoporosis or osteopenia-osteoporosis groups). Body weight was significantly smaller and the frequency of habitual smoking was significantly higher in patients with osteoporosis than in controls. Multivariable logistic regression analysis revealed that the frequency of the T allele was significantly higher in subjects with osteopenia-osteoporosis than in control individuals (p = 0.044; odds ratio, 2.9; CI, 1.0–8.1) and that the significance of the difference between controls and patients with osteoporosis was even greater (p = 0.0005; odds ratio, 33.1; CI, 5.4–280.6).
Table TABLE 2. DISTRIBUTION OF TGF-β1 GENOTYPES AMONG HEALTHY CONTROLS AND PATIENT GROUPS IN OBU CITY
Table TABLE 3. DISTRIBUTION OF TGF-β1 GENOTYPES AMONG HEALTHY CONTROLS AND PATIENT GROUPS IN SANDA CITY
Similar results were obtained with the population of Sanda. Age, height, years after menopause, and smoking status did not differ significantly between the controls and patients, whereas body weight in the control individuals was greater than that in patients with osteoporosis. Multivariable logistic regression analysis revealed that the frequency of the T allele was significantly higher in subjects with osteopenia-osteoporosis (p = 0.001; odds ratio, 22.9; CI, 4.5–216.3) and in those with osteoporosis (p = 0.017; odds ratio, 15.1; CI, 2.3–308.4) than in control individuals.
Finally, the serum concentration of TGF-β1 was determined to see if it changes as a function of the genotypes. As shown in Fig. 2A, serum TGF-β1 levels were significantly higher in healthy control subjects with the CC genotype than in age-matched healthy individuals with the TC or TT genotype. Among these subgroups, L2–L4 BMD with the TT genotype (0.939 ± 0.076 g/cm2) was significantly lower than that in those with the CC (1.052 ± 0.110 g/cm2, p = 0.010) or TC (1.035 ± 0.169 g/cm2, p = 0.021) genotype. Similarly, when serum TGF-β1 concentrations were assessed in patients with osteoporosis or osteopenia, there was a statistically significant difference among the genotypes (Fig. 2B). In these subgroups, L2–L4 BMD in patients with the TT genotype (0.707 ± 0.121 g/cm2) was significantly lower than that in those with the CC genotype (0.835 ± 0.101 g/cm2, p = 0.029).
Bone mass is determined by a variety of mechanisms that are affected by multiple genetic loci as well as various environmental factors. Family and twin studies have shown that genetic factors account for most of the variance in BMD in the general population,2–4 and the inheritance of bone mass is thought to be under polygenic control.18 Since the initial study describing the association of polymorphisms of the vitamin D receptor gene with BMD,5 several genes have been shown to be associated with bone mass and fracture risk.6–11 However, conflicting results have been obtained,19–22 and the genetic susceptibility to osteoporosis is not fully understood.
We have now examined the possible association of variants of the gene for TGF-β1, an important regulator of bone metabolism, with bone mass and the prevalence of oeteoporosis in postmenopausal women. Our study population was derived from two different areas in Japan: Obu City in Aichi Prefecture and Sanda City in Hyogo Prefecture. Individuals from these regions are thought to share the same ethnic ancestry and to possess a homogeneous genetic background. The distributions of TGF-β1 genotypes among control subjects in both cities were in Hardy-Weinberg equilibrium. Thus, our study population is likely to be genetically homogeneous and not biased. Our results indicate that the T29→C polymorphism of the TGF-β1 gene is associated with L2–L4 BMD in the total study population and that the T allele is significantly more prevalent in subjects with osteoporosis. Importantly, the same association was apparent when individuals from the two cities were analyzed separately. These observations suggest that TGF-β1 genotype is a genetic determinant of bone mass and that the T allele is an independent risk factor for genetic susceptibility to osteoporosis in Japanese women. Furthermore, we detected a significant difference in the serum concentration of TGF-β1 among healthy subjects and patients with osteoporosis or osteopenia with different TGF-β1 genotypes.
Langdahl et al.13 have studied the association of a C788→T polymorphism and 713–8delC in the TGF-β1 gene with osteoporosis and shown that the prevalence of the latter is significantly higher in osteoporotic women than in healthy women. As far as we examined in 104 Japanese subjects (84 patients with osteoporosis and 20 healthy individuals), we did not detect either of these genetic alterations. Thus, it is unlikely that there is linkage disequilibrium between the T29→C polymorphism and either 713–8delC or C788→T.
TGF-β1 is synthesized as a latent protein composed of 390 amino acids.23,24 The active TGF-β1 comprises two identical disulfide-linked polypeptide chains each corresponding to the 112 amino acids of the COOH-terminal portion of the precursor protein.23 The amino acid sequence of the active form of TGF-β1 is highly conserved across mammalian species, indicating a strong selection against variant forms of the protein. However, variations in the constitutive or induced expression of the protein as a consequence of variability in the TGF-β1 gene sequence might be associated with different effects of TGF-β1 on cellular functions.15 The Leu10→Pro polymorphism of TGF-β1 is located in the 29-residue signal peptide sequence that is cleaved from the precursor protein. The signal sequence, which functions to translocate newly synthesized proteins across the membrane of the endoplasmic reticulum,25 consists of three regions: a positively charged NH2-terminal region, a central hydrophobic core, and a polar COOH-terminal region.26 Amino acid residue 10 of TGF-β1 is located in the hydrophobic core. The association of TGF-β1 genotype with the serum concentration of TGF-β1 suggests that the Leu/Pro polymorphism may affect the function of the signal peptide, possibly resulting in a difference in export efficiency of the preproprotein. Thus, our results are consistent with the hypothesis that the T29→C (Leu10→Pro) polymorphism affects the extracellular concentration of TGF-β1 and thereby influences bone remodeling.
TGF-β, which is produced by osteoblasts and stored in substantial amounts in the bone matrix, is released during bone resorption and subsequently activated by the acidic microenvironment created by bone-resorbing osteoclasts.27,28 TGF-β plays crucial roles both in bone remodeling, determined by the balance between bone resorption and formation, and in skeletal development. TGF-β is important in proliferation and differentiation, as well as in matrix production by, osteoblasts,12 and it stimulates bone formation in vivo when injected locally into the periosteum of rats.29 In addition, TGF-β has been implicated as a mediator of the skeletal effects of estrogen.30 The production of TGF-β1 by human osteoblastic cells is stimulated by 17β-estradiol,31 and the TGF-β1 contributes to the estrogen-induced apoptosis of osteoclasts, which results in reduced bone resorption.32 Thus, TGF-β may mediate the local beneficial effects of systemic hormones, such as estrogen, on bone remodeling and cutaneous wound healing, and reduced TGF-β concentrations due to estrogen deficiency may predispose postmenopausal women to osteoporosis as well as delayed wound repair.33 Our observation that postmenopausal women with the CC genotype showed both the highest BMD and the highest serum concentrations of TGF-β1 is consistent with this concept. However, since the level and/or action of TGF-β are regulated at multiple points, including its release during bone resorption, activation by enzymatic cleavage, and binding to a carrier protein, the significance of the observed difference in circulating TGF-β1 concentrations among individuals with different genotypes is not clear. Further studies are required to determine whether the T29→C polymorphism is associated with differences in TGF-β1 concentration in the microenvironment of bone and whether it affects the response to hormone replacement therapy.
In conclusion, we have provided evidence for the association of the T29→C polymorphism of the TGF-β1 gene with L2–L4 BMD, the prevalence of osteoporosis, and the serum concentration of TGF-β1 in postmenopausal Japanese women. We propose that this polymorphism represents one of the genetic determinants of bone mass and that the T allele is an independent risk factor for genetic susceptibility to postmenopausal osteoporosis, at least in Japan. The potential usefulness of TGF-β1 genotyping in the prevention and management of osteoporosis warrants further study in populations with different ethnic and racial backgrounds.
This work was supported in part by Research Grants for Longevity Sciences from the Ministry of Health and Welfare of Japan (to Y.Y. and K.I.).