Genetic and other factors determining mannose-binding lectin levels in American Indians: the Strong Heart Study

Background Mannose-binding lectin (MBL) forms an integral part of the innate immune system. Persistent, subclinical infections and chronic inflammatory states are hypothesized to contribute to the pathogenesis of atherosclerosis. MBL gene (MBL2) variants with between 12 to 25% allele frequency in Caucasian and other populations, result in markedly reduced expression of functional protein. Prospective epidemiologic studies, including a nested, case-control study from the present population, have demonstrated the ability of MBL2 genotypes to predict complications of atherosclerosis,. The genetic control of MBL2 expression is complex and genetic background effects in specific populations are largely unknown. Methods The Strong Heart Study is a longitudinal, cohort study of cardiovascular disease among American Indians. A subset of individuals genotyped for the above mentioned case-control study were selected for analysis of circulating MBL levels by double sandwich ELISA method. Mean MBL levels were compared between genotypic groups and multivariate regression was used to determine other independent factors influencing MBL2 expression. Results Our results confirm the effects of variant structural (B, C, and D) and promoter (H and Y) alleles that have been seen in other populations. In addition, MBL levels were found to be positively associated with male gender and hemoglobin A1c levels, but inversely related to triglyceride levels. Correlation was not found between MBL and other markers of inflammation. Conclusion New data is presented concerning the effects of known genetic variants on MBL levels in an American Indian population, as well as the relationship of MBL2 expression to clinical and environmental factors, including inflammatory markers.


Background
Early innate human defenses against microbial invaders include mannose-binding lectin (MBL), a serum protein, which acts as a microbial opsonin and initiates activation of the so-called lectin complement activation pathway. Deficiencies in innate immunity have been hypothesized to allow chronic inflammatory states that may contribute to the pathogenesis of atherosclerosis [1]. Common variations in the MBL gene (MBL2), between 12 to 25% allele frequencies in Caucasian and other populations [2], result in markedly lower levels of circulating MBL [3] and are associated with both an increased susceptibility to infections [4] and cardiovascular disease in many [5][6][7][8], but not all studies [9]. An independent association between variant MBL2 genotypes and coronary artery disease (CAD) was previously confirmed in the American Indian, Strong Heart Study (SHS) cohort (OR = 3.2, 95% CI 1.5-7.0, p = 0.004, adjusted for 11 CAD covariates) [10]. On the other hand, genetypes resulting in high levels of MBL may be involved in complications related to cardiovascular disease (CVD). Thus, the role of MBL in cardiovascular disease appears to be a double-edged sword, due to hitherto only partially resolved mechanisms [11][12][13].
Although considerable effort has focused on the genotypic determination of MBL levels, our understanding of the role of various promoters and structural genetic variations remains incomplete. New promoter variants continue to be identified [14,15] and genetic background effects in specific populations are largely unknown.
MBL is also known to increase during an acute-phase reaction [16]. Samples collected during follow-up of a subset of the SHS cohort provide a valuable opportunity to test the correlation of both genotype and environmental exposures (e.g. incident CAD and inflammation, as measured by C-reactive protein (CRP) or fibrinogen) with MBL levels in this population.

Methods
The previously mentioned SHS case/control study matched cases and controls on age, gender and recruitment center (Arizona (AZ), Dakotas (DK) and Oklahoma (OK)). From this group a subset was chosen to include 186 of the controls without regard to genotype and an additional 51 cases limited to those known to be homozygous for high expressing "HYA" alleles.
Ascertainment of fatal and nonfatal cardiovascular events occurring between examinations was accomplished by medical record review and/or yearly participant contact [17]. Cases were defined by evidence of definite myocardial infarction (MI), definite coronary artery disease (CAD) without MI, definite evidence of MI by Minnesota ECG coding [17], or mortality codes indicating either def-inite MI, sudden death due to CAD, or definite coronary heart disease occurring between 5/18/89 and 12/31/99 [18]. Participants with only a diagnosis of possible CAD, "other CAD", stroke, congestive heart failure, or peripheral vascular disease were excluded. Controls were those individuals without any of the above diagnoses. MBL levels, MBL2 genotypes and other covariates were determined from both of these groups. American Diabetes Association criteria [19] were used to classify participant diabetic status in table 1. Participants were considered hypertensive if they were taking anti-hypertensive medications, had a systolic blood pressure greater than 140 mm Hg, or a diastolic blood pressure greater than 90 mm Hg.
Determination of MBL levels was done at Rigshositalet in Copenhagen, Denmark, using a double sandwich ELISA method as previously described [20].
The MBL2 gene was assessed for the presence of the B (G54D), C (G57E) or D (R52C) structural variations, and two promoter polymorphisms, one a G/C transition at -550 bp (the H/L alleles) [21] and the other, a G/C transition at -221 (X/Y alleles). These structural variations have typically been labeled "O" alleles in contrast to the most common "A" allele. Genotypes were determined by the oligonucleotide ligation assay as described by Nickerson and colleagues [22]. Quality control, duplicate, genotyping was performed by direct DNA sequencing. The structural variations were assumed to occur on opposing chromosomes. A number of promoter variants and structural alleles have been found to be in complete linkage disequilibrium and genotypes were checked against these established relationships. [23]. Genotyping was conducted at the University of Pittsburgh, Pittsburgh, Pennsylvania.
Composite categories of genotypes predicted to result in reduced expression of MBL2 similar to previous reports [3], primarily A_A, A_O, O_O, YA_O, XA_O, HA_O, LA_O, HYA_O, LYA_O, LXA_O, along with selected specific haplotypes of adequate prevalence, were used to guide analysis of genotypic effects on MBL levels. More inclusive categories used in the previous SHS MBL2 analysis [10] were also considered. Low expression of MBL2 (LOW_1) was assumed from the presence of an O_O or LA_O genotype. The LOW_2 category consisted of either O_O, LA_O or LA_LA genotypes. These risk categories were compared with reference groups ALL_1 or ALL_2, consisting of all genotypes not included in either LOW_1 or LOW_2 respectively; or the HIGH category (either HA_HA or HA_LA). There is ample documentation of the biologic effect of these various genotypes on basal MBL levels [23], although unidentified background genetic influences in unique populations are of interest.
The Chi square statistic was used to evaluate proportions between groups. Multivariate regression models and comparison of means (t test of independent samples) were used to understand the contribution of the various factors to MBL2 expression using SPSS version 10.1.0 software. The distributions of all subgroups in table 2 were examined and no deviation from the normal distribution was detected. A nominal p value of 0.05 was used to determine statistical significance and the Bonferroni correction was applied in situations of multiple testing.
Approval consistent with the Helsinki Declaration was obtained from all relevant Institutional Review Boards and tribes, and all participants gave informed consent. Table 1 summarizes the characteristics of both cases and controls. Please note that cases for this study were chosen on the basis of genotype and since cases and controls were initially matched for gender, age, and recruitment center, comparison between aggregated case and control samples should be interpreted with caution.

Results
There were no significant differences in the prevalence of O (15.3%) and H (67.2%) alleles in this control sample, compared with an earlier SHS report [10].
These findings confirm low levels of MBL among individuals with composite genotypic groups (eg O_O, XA_O, LA_O) [2], as well as those previously associated with CAD in this SHS population (bolded) [10], as seen in table 2.
Among controls only, comparing means of composite genotypes showed highly significant differences (p < 0.001 for all comparisons) between either the HYA_HYA or HIGH groups and any of the YA_O, HA_O, LA_O, O_O, LOW1, or LOW2 groups. Figure 1 shows these comparisons in the familiar "box-plot" format. The difference in mean MBL between HYA_HYA and HYA_LYA (unad-  expression and function will continue to be uncovered. The worldwide high population prevalence of these variants suggests an apparently ambivalent role of MBL2 in a number of pathogenetic and homeostatic processes [35][36][37]. In this report we have generally confirmed the previously reported influence of the three main structural variants and the two most widely investigated promoter variants (H/L and X/Y). This is of importance due to the demonstrated association between CVD and these variants in this particular, Strengths of this study include confirmation of the presumed link between genotype and expressed protein, increasing confidence in a possibly causal association between MBL and cardiovascular disease in this population. These results also extend the previously identified genotype/phenotype correlations to another population with a unique genetic background. A weakness is the previous matching structure of the sample and the need to MBL levels for various genotypes Figure 1 MBL levels for various genotypes. Comparison of mean HIGH and HYA_HYA MBL levels with all other groups is significant at p < 0.001. Outliers (between 1.5 and 3 box lengths from upper and lower margin) and extreme values (greater than 3 box lengths from box margin) are denoted by "o" and "*" respectively.
compare HYA/HYA case genotypes with the same control genotypes. There were an insufficient number of previously matched pairs with this genotype to allow a matched analysis, thus the aggregate comparison between these two groups needs to be interpreted with caution.

Conclusion
The findings from our study corroborate previous research that has demonstrated the important effects of genetic variants and the association of metabolic abberations with MBL levels. Further understanding of MBL2 expression may improve our ability to understand and disrupt the pathogenetic mechanisms involved in cardiovascular disease.