The CXCL12 SNPs and their haplotypes are associated with serum lipid traits

The relationship among the single nucleotide polymorphisms (SNPs) of the C-X-C motif chemokine ligand 12 gene (CXCL12) and the serum lipid profiles in the Chinese population has rarely been described, especially in somewhat old-fashioned and isolated Maonan minority. The goal of the current study was to elucidate the connection among the CXCL12 rs501120 and rs1746048 SNPs, haplotypes, several environmental factors and serum lipid traits in the Maonan as well as Han populations. Genotyping of the two SNPs, gel electrophoresis and direct sequencing were accomplished in 1,494 distinct subjects (Maonan, 750 and Han, 744) using polymerase chain reaction and restriction fragment length polymorphism. The frequencies of genotypes as well as alleles of the two SNPs were not similar between the two ethnic groups. The rs501120 SNP was related with serum total cholesterol levels, while the rs1746048 SNP was related with serum apolipoprotein (Apo) B levels. Four haplotypes were identified, of which the rs501120A-rs1746048C haplotype was the most common. The haplotypes of rs501120A-rs1746048T increased and rs501120G-rs1746048C decreased the risk of hyperlipidemia (P < 0.001 for each), showing consistent association with the levels of serum triglyceride, ApoA1 and ApoB. These outcomes specify that the CXCL12 SNPs as well as their haplotypes are related to serum lipid levels. Different serum lipid levels between both populations may partially be related to the CXCL12 SNPs, their haplotypes along with several environmental factors.

Genotypes as well as serum lipid levels. Figures 1 and 2 demonstrate the relationship among genotypes and serum lipid concentrations in both ethnic groups. The concentrations of TC in Han were dissimilar between the rs501120 genotypes (AA vs. AG/GG, P < 0.025 signified statistical significance after the Bonferroni correction). The concentrations of LDL-C plus ApoB in Maonan males were not similar amongst the rs501120 genotypes (P < 0.025 for each); the rs501120G allele carriers had greater LDL-C plus lesser ApoB concentrations compared to the rs501120G allele non-carriers. The concentrations of TG in Han males were dissimilar between rs501120 genotypes (P < 0.025); the rs501120G allele carriers had lesser TG concentrations compared to the rs501120G allele non-carriers. The concentrations of TC and LDL-C were dissimilar in Han females between the rs501120 genotypes (P < 0.025 for each); the rs501120G allele carriers had lesser TC plus LDL-C concentrations compared to the rs501120G allele non-carriers.
In addition, we also found that the concentrations of serum ApoB in Maonan were dissimilar between the rs1746048 genotypes (TT vs. TC/CC, P < 0.025) and the rs1746048C allele carriers had higher ApoB levels compared to the rs1746048C allele non-carriers. The concentrations of LDL-C and ApoB, and the ratio of ApoA1/ ApoB were different in the Maonan males. Moreover, the rs1746048C allele carriers had greater LDL-C as well as ApoB levels, and lesser ApoA1/ApoB compared to the rs1746048C allele non-carriers. The ApoA1/ApoB Haplotypes and serum lipid levels. As presented in Figure 3, weak linkage disequilibrium (LD) was observed between the two SNPs (r 2 = 0.63). Therefore, a haplotype analysis was conducted (Table 3). Four haplotypes were identified in the subjects of the current study. The haplotype of rs501120A-rs1746048C was the most common one (62.2%). The haplotype of rs501120A-rs1746048T was related with an augmented risk of hyperlipidemia (OR = 2.337, 95% CI = 1.252-4.361, P < 0.001), whereas the haplotype of rs501120G-rs1746048C was related with a diminished risk of hyperlipidemia (OR = 0.402, 95%CI = 0.234-0.690, P < 0.001). It also showed consistent association between the two haplotypes and serum TG, ApoA1 and ApoB concentrations. Multivariate logistic regression investigation indicated that the rs501120A-rs1746048T and rs501120G-rs1746048C haplotypes were associated with the occurrence of hyperlipidemia in Maonan and Han as per the stratified risk analyses including gender, BMI, alcohol consumption, diabetes as well as blood pressure (BP; P< 0.05; Table 4). The differences between the haplotypes and serum lipid profiles in both ethnic groups were as followed: the rs501120A-rs1746048T haplotype carriers had greater ApoB concentration than the rs501120A-rs1746048T haplotype non-carriers in the Maonan population. The rs501120G-rs1746048C haplotype carriers had lesser TG plus ApoA1 concentrations in Maonan and Han; as well as lesser ApoA1 concentrations in Han compared to the rs501120G-rs1746048C haplotype non-carriers (P < 0.025; Figure 4). Associated factors for serum lipid parameters. By performing the multivariable linear regression analysis on genotypes, environmental factors as well as serum lipid parameters in both groups, we found that the concentrations of ApoA1 were related with the rs501120 genotypes in both ethnic groups (P = 0.014). The concentrations of LDL-C, ApoB as well as the ratio of ApoA1 to ApoB were related with the rs1746048 genotype in Maonan males. The concentrations of TG in Han males (P = 0.038) and the concentrations of TC and ApoA1 in Han females were related with the rs1746048 genotypes (P = 0.049; P = 0.025, respectively; The value of triglyceride was presented as median (interquartile range) for not meet the normal distribution, the difference between the two ethnic groups was determined by the Wilcoxon-Mann-Whitney test.

Discussion
This study indicated that the effects of the two CXCL12 SNPs on serum lipid profiles were dissimilar: (a) the concentrations of TC in the Han were dissimilar between the rs501120 genotypes, and the ApoB concentrations in the Maonan were dissimilar between the rs1746048 genotypes; (b) the concentrations of LDL-C as well as ApoB in Maonan males, TG level in Han males, TC and LDL-C concentrations in Han females differed between the rs5011120 genotypes; and (c) the concentrations of LDL-C plus ApoB along with ApoA1/ApoB ratio in the Maonan males and ApoA1/ApoB ratio in the Han males differed between the rs1746048 genotypes. Both rs501120 and rs1746048 SNPs were associated with serum LDL-C along with ApoB concentrations. On the other hand, we established that the rs501120G allele carriers had greater LDL-C as well as lesser ApoB concentrations than the rs501120G allele non-carriers in Maonan males but not in Han males, while the G allele carriers had lesser LDL-C levels than the G allele non-carriers in Han females. There was not any substantial change between the concentrations of HDL-C plus ApoA1 and the genotypes of the two SNPs in both groups. It is well-acknowledged that hyperlipidemia is a complicated syndrome influenced by environmental as well as genetic factors and their interactions 9 . Preceding reports about family and twins have revealed that in large populations around 50% variation of the serum lipid profiles is genetically measured [30][31][32][33][34] . Serum LDL-C concentration is related with the advancement of CAD and can cause atherosclerotic cardiovascular disease (ASCVD) 35 . Carr et al. showed that ApoB was the primary organizing protein component of some serum lipids and was required for the formation of these lipids and lipoproteins, and it could be recognized as a measure of the number of atherogenic lipoproteins 36 . In other words, both LDL-C and ApoB may have a relationship with CAD or ASCVD, which is consistent with the connection amongst the SNPs and serum lipid profiles in the current study.
In addition, the results of the genotypic frequencies about the two SNPs in the both Chinese populations were not analogous to those demonstrated in other populations acquired from the International 1000 Genomes www.nature.com/scientificreports www.nature.com/scientificreports/ database. The allelic frequencies of the two SNPs in different populations are shown in Supplementary Table 3. This study showed substantial changes in the genotypic incidences of the rs501120 and rs1746048 SNPs between both ethnic groups. The minor allele frequency (MAF) of the rs501120 and rs1746048 SNPs was lesser in the Maonan compared to the Han. According to these outcomes, we could infer that the incidence of the CXCL12 rs501120 and rs1746048 SNPs may display racial/ethnic dissimilarity. Therefore, we strongly believe that the dissimilarities in genotypic frequencies between both ethnic groups may occur due to their different genetic background as well as environmental exposure.
Diet, lifestyle, obesity, physical activity, hypertension as well as the environmental factors would have an essential involvement in changing serum lipid levels [37][38][39][40] . The Maonan and Han populations have different dietary habits. Rice is the Maonan people's primary food. They also like strong flavor. They adore eating spicy and acid or sour foods that contain lots of oil and salt. It is likely that the preference for this high carbohydrate diet contributes to Maonan people having a higher weight and waist circumference than in Han people. In the meanwhile, foods high in fat and salt can cause hypertension, higher levels of serum TG as well as ApoB in the Maonan compared to the Han. Many previous studies have confirmed that diet, by itself, is obligated to the inconsistency on serum lipid levels 41,42 . In the current study, some environmental factors, for example age, waist circumference, BMI, alcohol intake, cigarette smoking, BP, and blood glucose showed correlations with serum lipid parameters. It is widely believed that high-fat diets, particularly comprising great amounts of saturated fatty acids, produce elevated serum cholesterol levels and may make people susceptible to CAD 43 . In the present study, however, we didn't find difference in the percentage of alcohol consumption between both ethnic groups, but we found some relationships between serum lipid parameters and alcohol consumption in both ethnic groups. Moreover, alcohol consumption, an important factor that influences serum lipid levels has been a hot topic and popular focus in recent years 44 . Onat et al. also conducted a research about alcohol and serum lipids. It showed that increased TG, LDL-C as well as ApoB levels existed in the males of their subjects with alcohol, and alcohol consumption in females reduced TG but did not alter LDL-C or ApoB levels 45 . When the ethnic variable was excluded from the predictors in the multiple linear regression evaluation, we found that the data such as the standard coefficient had some changes. As for the mechanism of mutation causing serum lipid changes, it is not clear in our study. But the environmental factor has a great influence on blood lipid levels, and can even change the genetic effects of genes. For the CXCL12 and its expression, a previous study found that the CXCL12 was less hypomethylated in patients with CAD, and the CXCL12 methylation was negatively correlated with its expression 46 . Therefore, the further  www.nature.com/scientificreports www.nature.com/scientificreports/ modification of the association between genetic variation and serum lipid concentrations may attribute to the exposure to diverse lifestyle as well as environmental factors.
Vital intra-and inter-genetic linkage disequilibrium (LD) relations were also noted in the current study. The LD pattern of the two detected CXCL12 SNPs was weak and low in our study populations, and it specified the functional dependences of the encoded relevant proteins, even though the two SNPs were located in non-coding region. The relationship amongst the CXCL12 polymorphisms and serum lipid levels in both ethnic groups could be partly explained by haplotype analysis. The haplotype of rs501120A-1746048C was the most common and signified around 62.3% of the samples. The rs501120A-rs1746048T haplotype increased and the rs501120G-rs1746048C haplotype decreased the risk of hyperlipidemia. Moreover, carriers of rs501120A-rs1746048T haplotype had increased serum ApoB concentrations, whereas carriers of rs501120G-rs1746048C haplotype had decreased serum TG and ApoA1 concentrations. We should make a statement that we only described the rs501120 G allele in the specific research populations (the Maonan male and Han female) with the allele decreasing LDL-C levels (decreasing risk factors), not for all mankind; it might be the gender difference that led to these results. Whether other alleles could increase or decrease serum lipid levels, the current research results had not yet been strongly supported. The description in the haplotype analysis was only discussed from the results of statistical analysis. If more explanations are needed, such as biological aspects, more in-depth experimental support is needed. Although there was only a tendency to risk allele of the rs501120 SNP in European patients with Moyamoya disease 47 , we established that haplotypes of the two CXCL12 SNPs might partly elucidate considerable serum lipid variation compared to single SNP alone in our study populations.
Our analysis benefited from a strong hypothesis and the indicators from other researchers of our group, a good sample size and population for study, and compared with clinical samples, there was less sample related issue. Nevertheless, the shortcomings and limitations are: (1) the amount of the study samples is relatively small; (2) some general characteristics could influence the concentrations of serum lipids despite adjustment in the data analysis.
(3) information about other SNPs near CXCL12 may be missing; (4) the selection of the SNPs was limited; (5) our study was originally an observational finding, and it would be impossible to recollect blood samples for institutional research because of national protection policies.
Through the collection of samples, blood biochemical testing, genotyping and gene sequencing, and statistical analysis, we concluded that the genotypic as well as allelic incidences of the CXCL12 rs501120 and rs1746048 SNPs were not similar between Maonan and Han. There were ethnic-as well as sex-specific associations among the CXCL12 SNPs and serum lipid levels. We detected four haplotypes and the commonest one was the rs501120A-rs1746048C haplotype. Two haplotypes of rs501120A-rs1746048T and rs501120G-rs1746048C showed an association with serum lipid traits. All these differences may partly attribute to the CXCL12 SNPs, their haplotypes, environmental factors as well as gene-environment interactions.

Materials and Methods
Participants. We 48 . The clinical data were collected by the specialists (relevant doctors, postgraduate students and nurses) trained from the First Affiliated Hospital, Guangxi Medical University. Gender, age, medical history, smoking and drinking history, family history, history of medications related to CVD, occupation, education, workforce, eating habits, etc. of the participants were recorded in detail. According to the amount, degree and type of alcohol, the intake of alcohol consumption was collected in the past 12 months. Alcohol intake was classified into groups of grams of alcohol per day: 0 (non-drinker), ≤25 and >25 49 . Similarly, cigarettes smoking was also classified into three groups: 0 cigarette per day (non-smoker), ≤20 cigarettes per day and >20 cigarettes per day 50 . Other parameters such as height, weight, BMI, BP as well as waist circumference were also determined sequentially. The methods for measuring relevant parameters referred to previous research methods 51 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Biochemical measurements. Blood sample collection: All subjects fasted for more than 12 hours the day before blood collection. The subjects took the sitting position and evacuated the fasting venous blood in the morning. A well-trained professional nurse took the blood sample of 5 ml from the elbow vein, and 2 ml was placed in a non-anticoagulated tube to solidify and separate the serum for blood biochemical measurements. Another 3 ml was treated with a special anticoagulant (ACD) containing 13.20 g/L trisodium citrate, 14.70 g/L glucose and 4.80 g/L citric acid for further removal of the DNA and it was stored in a refrigerator at −20 °C for further experiment after testing its concentration and purity. Blood biochemical indices detection: The Hitachi 7170 automatic biochemical analyzer was utilized to measure the serum biochemical indices of the collected blood samples: (1) Serum TC and TG levels were determined using enzyme labeling; (2) Serum ApoA1 as well as ApoB levels were determined using immunoturbidimetry; and (3) Serum HDL-C as well as LDL-C levels were determined by an enzyme-linked immunoassay. The outcomes were verified in the Clinical Science Experiment Center of the First Affiliated Hospital, Guangxi Medical University 40,52-55 . In addition, fasting blood glucose from fingertip peripheral blood was also determined by a Roche blood glucose meter 54,55 . DNA amplification as well as genotyping. (    When the participant's serum TC concentration was more than 5.17 mmol/L and/or his or her serum TG level was beyond 1.70 mmol/L, he or she was diagnosed as hyperlipidemic 56 . Hypertension was diagnosed by the following criteria: SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg were determined twice in the absence of antihypertensive drugs and averaged after three consecutive measurements 57 . The value of BMI in normal weight, overweight and obesity ranged from <24, 24-28 and >28 kg/m 2 , respectively 58 . According to the WHO diagnostic criteria for diabetes, type 2 diabetes was diagnosed by the following criteria: (1) fasting glucose ≥7.0 mmol/L; (2) 2 hours postprandial glucose ≥11.1 mmol/L; or (3) self-reported opinion of diabetes or usage of anti-diabetic medicines 59 .
Statistical analyses. The analysis was done using SPSS 21.0 (USA). The quantitative variables of normal distribution were mentioned as mean ± standard deviation, for example age, height, weight, BMI, waist circumference, BP, glucose, the concentrations of serum TC, HDL-C, LDL-C, ApoA1, ApoB and the ratio of ApoA1/ ApoB; whereas serum TG concentrations were mentioned as medians plus interquartile ranges for its skewed distribution. Direct calculation of the allele frequency and the genotype distribution among groups was assessed by the chi-square test. The Student's unpaired t-test was utilized for the comparison of general characteristics between the two ethnic groups. The relationship among genotypes and serum lipid parameters was evaluated by covariance analysis (ANCOVA). When each SNP related with serum lipid profiles fitted the value of P < 0.025 (matching P < 0.05 after adjusting for two independent assessments by the Bonferroni correction), it reflected statistical significance. The method to evaluate the haplotype frequencies as well as pair-wise LD amongst the identified SNPs was Haploview (USA, version 4.2). The range of the r 2 values was 0.30-0.80 as significant LD. Haplotype identification and the association of haplotype and the risk of hyperlipidemia were analyzed in the entire study population, and the stratified risk analysis was conducted from the separated population of Maonan and Han. To evaluate the connection between the risk of hyperlipidemia and genotypes, we considered performing an unconditional logistic regression analysis. To investigate the association among the genotypes and numerous environmental factors with serum lipid levels, we conducted a multivariable linear regression examination with stepwise modeling (with or without ethnic variable), defining the common homozygote genotype as 1, heterozygote genotype as 2 and rare homozygote genotype as 3. P value < 0.05 signified statistical significance.