Matrix Metallopeptidase-2 Gene rs2287074 Polymorphism is Associated with Brick Tea Skeletal Fluorosis in Tibetans and Kazaks, China

Brick tea skeletal fluorosis is still a public health issue in the north-western area of China. However its pathogenesis remains unknown. Our previous study reveals that the severity of skeletal fluorosis in Tibetans is more serious than that in Kazaks, although they have similar fluoride exposure, suggesting the onset of brick tea type skeletal fluorosis might be genetically influenced. Here we show that MMP-2 rs2287074 SNP (G/A), but not rs243865, was associated with Brick tea type fluorosis in Tibetans and Kazaks, China. The trend test reveals a decline in probability for skeletal fluorosis with increasing number of A alleles in Tibetans. After controlling potential confounders, AA genotype had about 80 percent lower probability of developing skeletal fluorosis than GG genotype in Tibetans (odds ratio = 0.174, 95% CI: 0.053, 0.575), and approximately 53 percent lower probability in Kazaks (odds ratio = 0.462, 95% CI: 0.214, 0.996). A meta-analysis shows that the AA genotype had approximately 63 percent lower odds (odds ratio = 0.373, 95% CI: 0.202, 0.689) compared with GG genotype within the two ethnicities. A significant correlation was also found between the genotype of MMP2 rs2287074 and skeletal fluorosis severity. Therefore, the A allele of MMP2 rs2287074 could be a protective factor for brick tea skeletal fluorosis.

suggesting that the prevalence rate of brick tea type skeletal fluorosis might be genetically influenced 3,4 . We also found that fluoride exposure is similar between Tibetans and Kazaks, however the severity of skeletal fluorosis in Tibetans is more advanced than in Kazaks 3 . The results suggest that the prevalence rate of brick tea type skeletal fluorosis might be genetically linked. Our findings and that of another group have shown gene polymorphism in GSTP1 3 and myeloperoxidase gene 5 might be associated with skeletal fluorosis, suggesting that some genetic factors that play a role in the pathogenesis of skeletal fluorosis. Unfortunately, they do not account for all cases. Hence it is important to identify additional candidate genes that might influence the risk of disease for the development of more effective preventive and treatment measures.
Skeletal fluorosis is a chronic metabolic bone and joint disease. There are growing evidences that MMP-2 plays an important role in the pathogenesis of bone metabolism disorders. MMP2, also known as gelatinase A, is a membrane-bound protein that is responsible for extracellular matrix degradation 6 . It is primarily expressed and secreted by osteoblasts and osteocytes in bone tissue 7,8 . Several research papers have shown that MMP-2 deficient mice display bone metabolism imbalance including loss of bone volume, mineralization abnormalities and joint erosion [8][9][10] . Osteoblast and osteoclast numbers were significantly decreased and their differentiation was restricted in MMP-2 deficient mice 9 . Bone marrow cells from MMP-2 −/− mice are unable to effectively support osteoblast and osteoclast growth and differentiation in culture 9 . Clinical studies revealed that MMP-2 expression has been shown to be up-regulated in human degenerative disc and the implant-prosthetic rehabilitation 11,12 . Other studies have shown a significant negative correlation between serum concentration of MMP-2 and bone mineral density (BMD) in postmenopausal Chinese women 13 , and MMP-2 could be used to evaluate bone remodelling and bone turnover 13,14 . Alternatively, the MMP-2 rs243865 single nucleotide polymorphism (SNP) (C/T) has been well characterized, and the C allele is associated with increased gene expression 15 . This SNP of MMP-2 has been shown to be associated with vertebral fracture 16 . The other SNP of MMP-2, rs2287074, has been suggested to be associated with stroke, obesity and maculopathy [17][18][19] . These studies suggest that MMP-2 plays a pivotal role in skeletal development and bone cell growth and proliferation, and that, some genetic viariants may be associated with the increased or decreased susceptibility to some diseases.
Recent studies have revealed that fluoride can affect the expression of MMP-2 in cell and animal experiment 20,21 . An epidemiological survey reveals that an increase of MMP-2 is positively correlated with fluoride exposure and the severity of dental fluorosis in adults 22 . These results suggest that MMP-2 might be a contributory factor for the onset of fluorosis. Recent observations points to a relationship between MMP-2 SNPs and the risk of diseases [16][17][18][19] , and ethnicity differences in which genetic susceptibility to diseases have been proven 23,24 . Therefore, we hypothesize that MMP-2 SNPs may be associated with the ethnic difference of skeletal fluorosis between Tibetans and Kazaks who have similar fluoride intake. In our study, two SNPs in the MMP-2 gene (rs243865 and rs2287074) were investigated for association with skeletal fluorosis. The statistical analyses show a significant lower OR in skeletal fluorosis associated with MMP-2 rs2287074 allele A. MMP-2 rs2287074 was also correlated with skeletal fluorosis severity, and the presence of A allele in Kazaks were siginificantly higher than that in Tibetans. The A allele of MMP2 rs2287074 was a protective factor for brick tea type skeletal fluorosis, which may be the reason for the differences in the severity of skeletal fluorosis between Tibetans and Kazaks.

Subjects.
A cross sectional study was conducted in seven villages from two provinces (Qinghai, Sinkiang), People's Republic of China, where brick-tea type fluorosis is prevalent from July to August 2012. The brick-tea type fluorosis village was identified as one in which people aged 16 years or older took the tea fluoride above 3.5 mg daily, and had skeletal fluorosis confirmed by X-ray (GB17018-2011, China). The subjects enrolled in this cross sectional study were older than 16 years, born and bred in the named villages. The subjects were investigated using a questionnaire which was designed to obtain name, address, sex, age, nationality, disposable income per capita, calcium (Ca) supplement, past medical history, personal history of brick tea consumption, and the volume of brick tea consumed daily. The face-to-face interview was performed by well-trained staff. Every subject received clinical examination which included physical examination and X-ray diagnosis (Beijing Longsafe Imaging Technology Co., Beijing City, China). In addition, brick tea water, blood and urine was collected from each participant.
Diagnosis of skeletal fluorosis. The radiograph of forearm, shank and pelvic of each participant was used to evaluate the skeletal fluorosis. Skeletal fluorosis was diagnosed and classified according to the Diagnostic Criteria of Endemic Skeletal Fluorosis (WS192-2008, China) as previously described 3 . Briefly, an X-ray of a skeletal fluorosis patient shows osteosclerosis, soft tissue calcification around the bone, acceleration of bone turnover, osteoporosis, osteomalacia and joint degeneration. Based on the results of the X-ray, skeletal fluorosis could be classified into three gradations: mild, moderate and severe. Genotyping methods. MMP-2 rs243865(C/T) and rs2287074 (A/G) are located in promoter region and in codon 460 of exon 9, respectively. Genomic DNA was extracted from whole blood with DNA extraction kit (Axygen Biosciences, Union City, USA). The DNA concentration was determined by TU1901 Spectrophotometry (Purkinje General Company, Beijing City, China) to ensure the DNA concentration was greater than 20 μ g/ml. The extracted genomic DNA was stored at − 80 °C. All gene sequencing were performed by the Shanghai Fenglin Clinical Laboratory Company (http://www.fenglinlab.com/index.asp) using the Sequenom MassARRAY system (Sequenom, Inc., San Diego, CA, USA).
Potential confounders. Except fluoride intake, age, gender, altitude, occupation, calcium supplement and income were also associated with skeletal fluorosis 4,25-29 . Fluoride intake was calculated according to the fluoride content of tea water, personal history of brick tea consumption and the volume of brick tea consumed daily. Age, gender and Ca supplement were investigated by a questionaire. Occupation of the two ethnicities included herdsman, farmer, teacher, monk, public servants and freelancer. They were classfied into three groups: herdsman, farmer and others, because there were relatively fewer teachers, monks, public servants and freelancers. Disposable income per capita was determined by asking subjects and divided into four groups: income ≤ 1000 RMB, 1000 RMB < income ≤ 2000 RMB, 2000 RMB < income ≤ 3000 RMB and income > 3000 RMB. The altitude of investigated sites was collected from local government.

Ethical approval and informed consent. The study was approved by the Ethical Review Board of Harbin
Medical University (HMUIRB20120021). All participants signed informed consent, and written informed consent was obtained from the guardians of minors. No specific permits were required for the locations or activities associated with the brick-tea water sample collection in this field study. The locations were not privately owned or protected in any way and this field study did not involve endangered or protected species. The methods were carried out in accordance with the approved guidelines. The association between MMP-2 rs2287074 and skeletal fluorosis was tested by the additive models. I 2 > 50% indicates significant heterogeneity. The overall OR with 95% CI was calculated using the fixed effects model, and stratified by age (age ≤ 45, 45 < age ≤ 65 and age > 65) because it was the common confounding factor in the two ethnicities. Correlation of two MMP-2 SNPs with skeletal fluorosis was assessed by Spearman test. P < 0.05 was considered statistically significant. All statistical analyses were performed with STATA (STATA, College Station, TX, Version 12.0). The data of this study had been provided in Supplementary information.

Results
A total of 598 subjects were enrolled in this study. 221 subjects were diagnosed with skeletal fluorosis, and the prevalence of skeletal fluorosis was 37.0%. The medians of fluoride intake and urine fluoride were 5.985 mg/L and 2.755 mg/L in enrolled participants, respectively. The prevalence of skeletal fluorosis and fluoride exposure between Tibetan and Kazak were shown in Table 1 The descriptive analysis of potential risk factors between skeletal fluorosis cases and controls were presented and stratified by ethnicity in Table 2. Fluoride intake and urine fluoride in the Tibetan cases were both significantly higher than those in Tibetan controls (Z = 2.354, p = 0.019; Z = 2.477, p = 0.013, respectively), but there were no significant difference in Kazak participants (Z = − 1.614, p = 0.106; Z = − 0.631, p = 0.528, respectively). The cases were significantly older than the controls in the two ethnicities (t = 7.078, p < 0.01 in Tibetans; t = 2.034, p = 0.043 in Kazaks, respectively). The difference of gender proportion between the cases and the controls was detected in Tibetans (Pearson Chi-square = 5.595, p = 0.018), but not found in Kazaks (Pearson Chi-square altitude was not different in Kazaks (Z = 0.379, p = 0.705). From the results, some confounding factors were identified in the two ethnicities: fluoride intake, urine fluoride, gender, age and altitude were identified among Tibetans, and age was a confounding factor for Kazaks. Age was the common confounding factor among the two ethnicities. The association between two SNPs of MMP-2 gene and skeletal fluorosis was investigated in the two ethnicities. The hypothesis of Hardy-Weinberg equilibrium could not be rejected for each of the two SNPs. Analysis of the genotype distribution shows that the genotype frequency of MMP-2 rs243865 between the cases and the controls were not statistically different in the two ethnicities (Likehood Ratio = 1.270, p = 0.530 in Tibetans; Likehood Ratio = 0.533, p = 0.766 in Kazaks). However, the genotype frequency of MMP-2 rs2287074 differed significantly between the cases and controls in Tibetans (Likehood Ratio = 8.704, p = 0.013), but not significantly in Kazaks (Likehood Ratio = 3.738, p = 0.154). The Cochran Armitage trend test and the logistic regression analysis were further performed to estimate the association between MMP-2 rs2287074 and the risk of skeletal fluorosis. The trend test revealed a declining probability for skeletal fluorosis with increasing number of A allele in Tibetans (Chi-square = 7.196, p = 0.007). In the logistic regression analysis, we adjusted for fluoride intake, urine fluoride, age, gender and altitude in Tibetans, and adjusted for age in Kazaks. After controlling potential confounders, there were significantly lower odds of skeletal fluorosis in participants with AA homozygotes. AA genotype had about 80 percent lower odds of having skeletal fluorosis than GG genotype in Tibetans (Odds Ratio = 0.174, 95% CI: 0.053, 0.575), and approximately 53 percent lower odds in Kazaks (Odds Ratio = 0.462, 95% CI: 0.214, 0.996) ( Table 3).
A meta-analysis was conducted to assess the association between MMP-2 rs2287074 and skeletal fluorosis risk in the two ethnicities. We defined the inheritance model by treating allele A as the 'risk' allele. The meta-analysis was stratified by age which was the common confounding factor in the two ethnicities. Table 4 shows that the AG genotype were not significantly associated with the risk of skeletal fluorosis compared with the GG genotype. However, the AA genotype was associated with a significant decrease in risk of skeletal fluorosis compared with GG genotype (Odds Ratio = 0.373, 95% CI: 0.202, 0.689). The stratified analysis shown that the A allele was not associated with the risk of skeletal fluorosis in subjects with age ≤ 45. However the A allele was significantly associated with skeletal fluorosis in subjects with 45 < age ≤ 65. The AA genotype was significantly associated with a decrease in risk of skeletal fluorosis compared with GG genotype in subjects aged 45 years and above (Odds Ratio = 0.403, 95% CI: 0.191, 0.850; Odds Ratio = 0.168, 95% CI: 0.031, 0.902, respectively).
Furthermore, a significant correlation between the allele of MMP2 rs2287074 and skeletal fluorosis severity was found in these participants (r = − 0.151, p < 0.01). The correlation was separately detected in Tibetans and Kazaks (r = − 0.148, p < 0.01; r = − 0.122, p = 0.038). Subjects with AG or AA genotypes had less severity skeletal fluorosis. Particularly, there was only one moderate, and none severe skeletal fluorosis among subjects with AA genotype (Table 5).

Discussion
Skeletal fluorosis is a chronic metabolic bone disease, which is caused by excessive fluoride intake 30 . Bone matrix degradation is a key step in bone metabolism in which MMPs play an important role. MMPs, zinc-dependent proteases, are able to degrade bone matrix proteins such as collagen and elastin 31 Table 4. Pooled estimates of ORs (95% CI) for the association between MMP-2 Rs2287074 and skeletal fluorosis.
(MMP-2), isolated from the culture media of rheumatoid arthritis synovial tissue 32 , is important for extracellular matrix degradation 33 . It has proteolytic activity against components of the basement membrane, preferentially cleaving collagen types I, IV, V, VII, and XI and gelatin 6,34,35 . Recently, the association between MMP-2 and fluorosis has been explored 22 . An in vitro study shown that MMP-2 activity of human saliva significantly inhibited by excessive fluoride (above 50 mg/L) 20 . However, an increase of MMP-2 expression in gingival connective tissue and periodontal ligament was found in rabbits treated with 40 mg/L of fluoride 21 . In a word, though the effect of fluoride on MMP-2 remains assumptive, excessive fluoride exposure could alter MMP-2 activity and expression. An investigation in human population also shown that a positive dose-response relationship between MMP-2 and fluoride exposure and the severity of dental fluorosis was separately observed in adults 22 , suggesting that MMP-2 could be involved in the pathogenesis of skeletal fluorosis. Previous researches have shown that MMP-2 rs243865 was associated with the increased expression of MMP-2 15 , and white women with CT genotype of MMP-2 rs243865 had a lower rate of vertebral fracture in USA 16 . The other synonymous SNP of MMP-2 rs2287074 has been found for associations with stroke, obesity and maculopathy [17][18][19] , but has no association with fracture risk and BMD 16 . This is the first epidemiologic study that evaluates the association between MMP-2 polymorphisms and skeletal fluorosis. In our study, the MMP-2 rs2287074 was associated with significantly lower odds of skeletal fluorosis, but the association between MMP-2 rs243865 and skeletal fluorosis was not seen. After controlling for potential confounders, the AA genotype of MMP-2 rs2287074 was separately associated with significantly lower odds of skeletal fluorosis in Tibetans and Kazaks. We further performed a meta-analysis to estimate the overall OR of the two ethnicities and OR stratified by age. The AA genotype of MMP-2 rs2287074 remained significantly associated with lower odds of skeletal fluorosis in all subjects. It has been reported that the prevalence of skeletal fluorosis was significantly associated with age 36 , and it was the common confounding factor in the two ethnicities. So we investigated the potential interactions between MMP2 rs2287074 and age. This protective effect of AA genotype of MMP2 rs2287074 was detected only in subjects with age > 45, but not in subjects with age ≤ 45. In present study, we also found that the prevalence of skeletal fluorosis in subjects with age ≤ 45 was lowest than two other older groups. Therefore we speculate large sample is required to get difference while the prevalence rate of fluorosis is low in the subjects with age ≤ 45. Also, there is a significant correlation of the MMP-2 rs2287074 variant with skeletal fluorosis severity, and subjects with the AA genotype had less risk to suffer from moderate and severe skeletal fluorosis in the two ethnicities. These results suggest that MMP-2 rs2287074 was not only associated with skeletal fluorosis, but also correlated with skeletal fluorosis severity. The more severe of skeletal fluorosis in Tibetans might be associated with the lower frequency of MMP-2 rs2287074 A allele.
Although we found that genotype AA of MMP2 rs2287074 was a protective factor for brick tea skeletal fluorosis, the specific function of the MMP2 rs2287074 SNP in skeletal fluorosis is unknown. This SNP is synonymous, resulting in the same amino acid (threonine) at codon 460 regardless of the allele present. Although this kind of variant does not appear to create a new splice site or alter an existing one, it has been shown that variation at synonymous sites could alter RNA secondary structures, affecting RNA stability, which result in the changes of protein expression and function [37][38][39] . Nevertheless, the specific functionality of MMP-2 rs2287074 in skeletal fluorosis needs to be further clarified.
There were several limitations in this study. Sample size of our study was too small to affect observed power, such as the frequency of MMP-2 rs 2287074 AA genotype was fewer in the two ethnic groups. The association could be as a result of confounding by unknown factors. Moreover, skeletal fluorosis is complex disease and it is likely that several genes and/or polymorphic sites influence its malformations. Thus, the nature of the association of this polymorphism with skeletal fluorosis needs to be further clarified.

Conclusions
In summary, in this study, we found there was significantly lower chances of skeletal fluorosis developing in subjects with the A allele of MMP2 rs2287074. The AA genotype of MMP2 rs2287074 was also shown to be   inversely associated with skeletal fluorosis severity. The differences in skeletal fluorosis severity between Tibetans and Kazaks are possibly induced by the different frequency of MMP2 rs2287074 genotypes. These results need to be replicated in future studies, particularly with sufficiently larger sample sizes, to confirm this association in other ethnicities. In addition, the direct or indirect role of SNPs in skeletal fluorosis pathogenesis should be further investigated.