Methylmercury Exposure, PON1 Gene Variants and Serum Paraoxonase Activity in Eastern James Bay Cree Adults

ORIGINAL ARTICLE Methylmercury exposure, PON1 gene variants and serum paraoxonase activity in Eastern James Bay Cree adults

Olivia Drescher1, Eric Dewailly1,2, Caroline Diorio2,3, Nathalie Ouellet1, Elhadji Anassour Laouan Sidi1, Belkacem Abdous1,2, Beatriz Valera1 and Pierre Ayotte1,2,4

There is growing evidence that cardiovascular health can be affected by exposure to methylmercury (MeHg), by a mechanism involving oxidative stress. Paraoxonase 1 (PON1) is a high-density lipoprotein-bound enzyme that hydrolyzes toxic oxidized lipids and protects against cardiovascular diseases. Evidence from in vitro studies indicates that MeHg can inhibit PON1 activity but little is known regarding this effect in humans. We investigated whether increased blood mercury levels are associated with decreased serum PON1 activity in Cree people who are exposed to MeHg by fish consumption. We conducted a multi-community study of 881 Cree adults living in Eastern James Bay communities (Canada). Multivariate analyses considered sociodemographic, anthropometric, clinical, dietary and lifestyle variables and six PON1 gene variants (rs705379 ( À 108C/T), rs662 (Q192R), rs854560 (L55M), rs854572 ( À 909C/G), rs854571 ( À 832C/T) and rs705381 ( À 162C/T)). In a multiple regression model adjusted for all potential confounding factors and the rs854560 PON1 variant, a statistically significant MeHg*rs705379 interaction was observed. Blood mercury levels were inversely associated with serum PON1 activities in individual homozygous for the À 108T allele (P ¼ 0.009). Our results suggest a gene–environment interaction between the rs705379 polymorphism and MeHg exposure on PON1 activity levels in this aboriginal population. This finding will need to be replicated in other population studies.

Journal of Exposure Science and Environmental Epidemiology (2014) 24, 608–614; doi:10.1038/jes.2013.96; published online 22 January 2014 Keywords: Cree; gene–environment interaction, methylmercury; paraoxonase; PON1; selenium.

INTRODUCTION Most of the anti-inflammatory and antioxidant properties of Mercury (Hg) has always been present in the environment HDL appear to be associated with paraoxonase 1 (PON1), an because of its natural deposition but during the last century, the enzyme synthesized primarily in the liver and almost exclusively 7,8 environmental load has increased as a result of industrialization. bound to HDLs. A number of epidemiological studies has Hg is converted by microorganisms present in sediments to reported the anti-atherogenic properties of PON1 and its role in methylmercury (MeHg), which biomagnifies in the aquatic food protecting against CVDs. In the Caerphilly prospective study, men chain. Humans are exposed to MeHg primarily by the con- in the highest quintile of PON1 activity had decreased odds ratios sumption of contaminated fish, particularly large predatory fish of suffering a coronary event.9 Another large prospective study species.1 A spectrum of adverse health effects has been observed found evidence to further support PON1 anti-inflammatory pro- following exposure to MeHg, the most toxic form of mercury, perties through its role in reverse cholesterol transport, which particularly on the nervous system, the developing brain and the contributes to HDL protection against atherosclerotic diseases.7 In kidneys.2 Neonatal neurodevelopmental toxicity is the most addition, PON1 knockout mice fed a pro-atherogenic diet were sensitive end point of MeHg exposure; however, a recent litera- more likely to develop atherosclerosis than wild-type mice.10 ture review has presented substantial evidence of an association PON1 activity is partly genetically determined by a variety of between MeHg exposure and the risk for cardiovascular diseases polymorphic forms of the protein, which vary among individuals (CVD).3 and among populations of different ethnic origins.11,12 Genetic The first report of this association was from a prospective study variations of the PON1-coding region occur at position 192 (rs662), of healthy Finnish men by Salonen et al., who observed that where a glutamine (Q) is replaced by an arginine (R) because of an mercury body burden from the consumption of non-fatty fish amino-acid substitution, and at codon 55 (rs854560) where a (generally poor in n-3 polyunsaturated fatty acids (PUFAs)) was methionine (M) replaces a leucine (L). The Q192R polymorphism associated with excess risks of cardiac events and all-cause has been shown to shape the enzyme’s in vivo capacity to mortality.4 MeHg and inorganic mercury induce cellular toxicity by hydrolyze oxidized lipids.13,14 Concurrently, the L55M polymor- interacting with antioxidant glutathione molecules, forming bonds phism has been reported to affect plasma protein levels.15 In with thiol groups of cysteine residues.2 MeHg also induces oxida- addition, the À 108 (rs705379), À 162 (rs705381) and À 909 tion of low-density lipoproteins (LDLs),4 whereas high-density (rs854572) polymorphisms of the regulatory/promoter region lipoproteins (HDLs) destroy biologically active oxidized lipids.5,6 were found to alter PON1 expression.16 In particular, plasma

1Axe Sante´ des populations et pratiques optimales en sante´, Centre de recherche du CHU de Que´bec, Que´bec, QC, Canada; 2De´partement de me´dicine sociale et pre´ventive, Universite´ Laval, Que´bec, Quebec, Canada; 3Centre des maladies du sein Descheˆnes-Fabia, Centre de recherche du CHU de Que´bec, Que´bec, Quebec, Canada and 4Laboratoire de toxicologie, Institut national de sante´ publique du Que´bec, Que´bec, Quebec, Canada. Correspondence to: Professor Pierre Ayotte, Laboratoire de toxicologie, Institut national de sante´ publique du Que´bec, 945, Avenue Wolfe, 4e e´tage, Que´bec, QC, Canada G1V 5B3. Tel.: þ 418 650 5115 (ext 4654). Fax: þ 418 654 2148. E-mail: [email protected] Received 14 May 2013; revised 27 September 2013; accepted 6 December 2013; published online 22 January 2014 MeHg and serum PON1 activity in Cree people Drescher et al 609 PON1 concentrations were found to be twice as high in individuals at À 80 1C until they were analyzed at the toxicology laboratory of the with the À 108C allele compared with those bearing the À 108T Institut Nationale de Sante´ Publique du Que´bec (INSPQ). Paraoxonase allele.17 activity was measured in 5 ml of serum, diluted to a final concentration In addition to PON1 genotypes, environmental agents, lifestyle of 2%, using the EnzCheck Paraoxonase Assay Kit (Molecular Probes, factors (smoking, alcohol, diet, drugs), age and diseases have been Invitrogen, Carlsbad, CA, USA). This highly sensitive homogenous 18 fluorometric assay for the organophosphatase activity of paraoxonase is found to modulate the enzyme activity. Gonzalvo et al. were the based on the hydrolysis of a fluorogenic organophosphate analog first to show that mercurial compounds were the most potent 19 (7-diethylphospho-6,8-difluor-methylumbelliferyl). The intensity of emitted inhibitors of paraoxonase activity in vitro. However, exposure of light was measured with a Victor2 plate reader (Perkin Elmer–Cetus 18 mice to MeHg did not induce a significant inhibitory effect. Life Sciences, Boston, MA, USA) using 355 and 460 nm filters for exci- Results from a subsequent in vitro study revealed that mercury tation and emission, respectively. Enzymatic activity was calculated by induced a greater inhibitory effect on human serum PON1R192 converting fluorescence intensity into activity using a standard curve. 20 Activity was expressed as kilo units per liter (kU/l), where 1 U of activity than on PON1Q192 activity. Some aboriginal populations that rely on seafood for suste- paraoxonase is defined as the amount of enzyme that produces 1 nmol 1 nance may be highly exposed to MeHg, which could impact on of fluorescent product per minute at 37 C. Aliquots of a serum sample obtained from a volunteer were stored frozen at À 80 1C along with the their cardiovascular health. We hypothesize that inhibition of study samples; this ‘‘control’’ sample was analyzed in each analytical batch. PON1 activity might constitute a biomarker of adverse cardio- The mean value was 16.4 kU/l and the between-day coefficient of variation vascular effect induced by MeHg exposure in these populations. was 10.4% (n ¼ 16). Recently, we reported a negative association between blood Mercury, selenium and lead concentrations were determined in whole mercury levels and plasma PON1 activity in Inuit adults who blood samples collected in a 6-ml plastic vacutainer containing potassium participated to a large cross-sectional study conducted in Nunavik EDTA as the anticoagulant (BD Medical) and stored frozen at À 80 oC until (Northern Quebec, Canada).21 In the present study, we examined time of analysis. Determination was performed by inductively coupled the mercury–PON1 activity relationship in the Cree population of plasma mass spectrometry on an ELAN DRC II instrument for mercury and Eastern James Bay, another Northern Quebec aboriginal popula- selenium and an Elan 6000 instrument for lead (Perkin-Elmer SCIEX, Concord, ON, Canada). Blood samples were diluted in ammonium hydro- tion exposed to MeHg by fish consumption. We also investigated xide and metals were brought to their elementary form by passing through the potential protective role of selenium, which has been shown 22 argon plasma before being identified by mass spectrometry. The limits to counteract the toxicity of mercurials, as well as the possible of detection for mercury, selenium and lead were 0.5 nmol/l, 0.1 mmol/l confounding or modifying role of several factors, including PON1 and 1 nmol/l, respectively. The between-day coefficients of variation for gene variants. mercury, lead and selenium measurements were, respectively, 2.1%, 2.8% and 6.1%. Lipid analyses were performed at the Centre de Recherche sur les MATERIALS AND METHODS Maladies Lipidiques (Centre de Recherche du CHU de Que´bec, Que´bec, QC, Study Population and Sampling Canada). Cholesterol and triacylglycerol levels were determined by enzymatic methods on the Vitros 950 Chemistry Station (Ortho-Clinical Diagnostics, More than 16,000 Cree live in nine communities located in the James Raritan, NJ, USA) using the manufacturer’s reagents and calibrators. For Bay region of Quebec, Canada (Figure 1). From 2005 to 2011, the cholesterol and triglycerides measurements, samples were analyzed by Nituuchischaayihtitaau Aschii Multi-Community Environment-and-Health multilayer film dry-slide chemistry with colorimetric detection. Study was conducted in seven of these communities. The study protocol was HDL-cholesterol (HDL-C) was measured using the Vitros direct HDL-C slide approved by the Research Ethics Boards of Laval University, McGill University assay (Ortho-Clinical Diagnostics), based on the precipitation of apo- and McMaster University, as well as the Research Ethics Committee of the lipoprotein B-containing lipoproteins by sulfate dextran/magnesium Cree Board of Health and Social Services of James Bay. A consent form was chloride using magnetic beads. Cholesterol level was then measured by signed by all individuals who accepted to participate in the study. enzymatic methods on the Vitros 950 Chemistry Station. LDL-cholesterol Within each community, sampling of the population followed a stratified (LDL-C) was calculated using Friedewald’s formula.24 PUFAs were random design, considering the following age categories: 0–7, 8–14, 15–39 measured in erythrocyte membrane phospholipids. The fatty acids were and Z40 years old. The selection of participants within each age stratum was transmethylated and the resulting fatty acid methyl esters analyzed by done using simple random sampling without replacement in order capillary-column gas chromatography on a Hewlett-Packard 5890 series II to build a list of potential participants to be contacted by recruiters. Weights gas chromatograph (Hewlett-Packard, Mississauga, ON, Canada) equipped were attributed for each stratum in order to ensure population representation. with a fused silica column (DB-23; 30 m Â 0.25 mm ID Â 0.25 mm thickness; A total of 2267 persons Z15 years of age were eligible and were con- Agilent Technologies) and a flame ionization detector. Total concentra- tacted, of whom 971 (43%) accepted to participate. As CVDs are consi- tions of omega-3 fatty acids (n-3 PUFAs: C18:3 þ C18:4 þ C20:3 þ C20:4 þ dered an ‘‘adult’’ ailment, we excluded 90 participants who were younger C20:5 þ C22:5 þ C22:6) were expressed in percent of total fatty acids than 18 years. Therefore, our population sample for this study comprised (weight basis). 881 participants Z18 years of age. Genotyping Data Collection Six PON1 gene variants (rs705379 ( À 108C/T), rs662 (Q192R), rs854560 Data pertaining to the variables of interest were gathered using several (L55M), rs854572 ( À 909C/G), rs854571 and rs705381) were analyzed self-administered and interviewer-completed questionnaires, medical using Sequenom or TaqMan technologies at the McGill University records and during a clinical session. Questionnaires administered by a and Genome Quebec Innovation Center, Montreal, Canada. The call local interviewer, either in Iiyiyiuyimuwin or in English, were used to collect rates were 499.5% for all SNPs. The error rate of Sequenom technology is information on demographics, lifestyle factors, health and consumption lower than 0.5% according to duplicate and control samples tested at patterns of traditional and store-bought foods. Biological samples were Genome Quebec Innovation Centre. TaqMan technology has an average collected during the clinical evaluation session. Height, weight and waist error rate of 0.1%. All genotypes were in Hardy–Weinberg Equilibrium circumference were also measured during the clinical session. Waist (P40.01). circumference was determined at the end of exhalation using a measuring tape horizontally placed between the last floating rib and the iliac crest.23 Body mass index (BMI) was calculated by dividing the weight (kg) by the Statistical Analysis squared height (m2). Information on diabetes status and medication was First, a descriptive analysis of all variables was carried out. The geometric extracted from the medical records. mean (GM) was presented for variables that were not normally distributed. The best transformation for each variable was then selected in order to meet normality assumptions. With the help of the BoxCox Macro Laboratory Analyses developed by Michael Friendly (http://www.math.yorku.ca/SCS/sasmac/ Blood was collected from a cubital vein in a 6-ml plastic vacutainer without boxcox.html), these transformations were selected based on both anticoagulant (BD Medical, NJ, USA). Serum samples were stored normality test (Kolmogorov–Smirnov) and bivariate relationships between

& 2014 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 MeHg and serum PON1 activity in Cree people Drescher et al 610

Figure 1. Map of northern Quebec showing the location of the seven Cree communities that participated in the study.

variables. Subsequently, log transformation was retained for PON1 the residual values of the model fitted to an acceptable range of ±1 the activity, blood mercury, n-3 PUFAs, HDL, LDL and fish consump- regression line. tion frequency. Blood selenium on the other hand was best fitted as an inverse function. All statistical analyses were performed using SAS 9.2 software. RESULTS Homozygous individuals for the rare allele were grouped with heterozygous individuals for rs705381 and rs854560 variants. In addition, all Selected characteristics of our study group that comprised 881 pairwise combinations of SNPs were evaluated for linkage disequilibrium Cree adults living in seven Eastern James Bay communities are (LD) using Lewontin’s D’ statistic. listed in Table 1. There were 369 men and 512 women with an Spearman correlation coefficient was used to assess the crude overall average age of 40 years. Over 70% of the participants were relationship between blood concentration of mercury (nmol/l) and obese (BMI exceeding 30 kg/m2), and 90% were smokers or self-reported fish consumption frequency (monthly meals, averaged ex-smokers. Diabetes was reported for 17% of the men and for over a 1-year period). Crude, adjusted and adjusted-standardized 22% of women. More than 97% reported eating fish with an regression coefficients were calculated for PON1 activity (kU/l) with average consumption rate of 4.9 meals per month. The GM of blood mercury concentration (nmol/L), potential confounding variables PON1 activities was 11.5 kU/l, with a 20-fold factor between the (selenium (mmol/l), age (years), BMI (kg/m2), HDL-C (mmol/l), LDL-C (mmol/l), n-3 PUFAs (%), fish consumption frequency (monthly meals, lowest and highest values (1.6–31.8 kU/l). GM of blood mercury averaged over a 1-year period), gender, smoking status (everyday, levels was 16.7 nmol/l and values ranged from 0.3 to 200 nmol/l. occasionally, ex-smokers, never smoked), diabetes and PON1 genotypes GM of blood selenium concentrations was 2.2 mmol/l with values (rs705379 ( À 108C/T), rs662 (Q192R), rs854560 (L55M), rs854572, rs854571, ranging from 1.3 to 4.9 mmol/l. The number of fish meals per rs705381). month was significantly correlated to blood mercury concentra- To investigate mercury–gene interactions, multiple linear regression tion (r ¼ 0.59; Po0.0001). models were built that included blood mercury concentration, the PON1 The following SNPs were in complete linkage disequilibrium variant of interest, the mercury–PON1 variant interaction term and all (LD): rs854560 and rs662, both in the coding region; rs705379, potentially confounding variables. To account for missing values of several rs854572, rs705381 and rs854571 all in the promoter region. Two confounding factors, we used the multiple imputation procedure in 0 accordance to the fully conditional specification method. A collinearity SNPs (rs854560 and rs705379) were also in partial LD (D ¼ 0.7). diagnosis was performed between variables that were included in the final Crude and adjusted regression analyses were used to assess the model. Verification of linear regression postulates was performed by relation between blood mercury concentration and PON1 activity, examining the linearity, normality and homoscedasticity. Five influential while adjusting for the influence of potential confounding outliers, which were weakening the model, were identified. Upon removal, variables listed in Table 2. There were no significant associations

Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 & 2014 Nature America, Inc. MeHg and serum PON1 activity in Cree people Drescher et al 611 between PON1 activity and gender, smoking, diabetic status, BMI potential confounders. However, in additional multivariate ana- or n-3 PUFA content of erythrocyte membrane phospholipids. lyses, we found a significant interaction between blood mercury PON1 activity decreased with age (P ¼ 0.007) and with fish levels and the rs705379 variant. A negative relation between consumption (P ¼ 0.003). When adjusted for all potential con- blood mercury levels and serum PON1 activity was noted in founding variables, the association was no longer significant for individuals homozygous for the rare allele (TT). We speculate that fish consumption. PON1 activity was also positively associated to this subgroup could be at risk of adverse cardiovascular effects HDL-C (P ¼ 0.0001) and to LDL-C (P ¼ 0.01) and the associations induced by MeHg. remained statistically significant upon adjustment for the other In our previous study conducted in the Inuit population of variables. Five of the six SNPs (rs705379, rs854560, rs854572, Nunavik,21 we had found no evidence of such a gene– rs854571 and rs705381) were significantly associated with PON1 environment interaction on plasma PON1 activity. However, a activity (Po0.0001). Upon adjustment for all potential confoun- negative association of blood mercury levels to PON1 activity was ders, the associations with rs705379 and rs854560 remained observed in a multivariate model that included age, blood statistically significant. selenium, HDL-cholesterol, n-3 PUFAs and three genetic poly- We found a statistically significant interaction between blood morphisms (rs705379, rs662 and rs854560) as covariates (Pr0.001 Hg concentration and the rs705379 ( À 108C/T) variant (P ¼ 0.002) for blood mercury and all covariates). An important difference in a regression model adjusted for all potential confounding between the two aboriginal groups is the mean blood Hg level, factors (Table 3). Other models were built, which included which was threefold lower in the Cree group compared with interaction terms between blood mercury level and other gene the Inuit group (GM: 16.7 nmol/l vs 53.2 nmol/l). Therefore, the variants; no other statistically significant interaction was noted difference in findings may be attributable in part to the (data not shown). considerably lower Hg levels observed in the Cree population. The interaction between blood mercury concentration and Hg levels in the Inuit population may have been elevated enough rs705379 polymorphism on the adjusted PON1 activity is shown in to inhibit PON1 activity independently of gene polymorphisms Figure 2. Blood mercury concentration was inversely associated that are known to modulate the enzyme activity/expression level. with PON1 activity in homozygous TT participants (b ¼À0.06, In addition, the difference in allele frequency distribution for the P ¼ 0.004). Heterozygous (CT) subjects also showed a negative rs705379 variant may also explain in part the difference in PON1 mercury–PON1 association (b ¼À0.03, P ¼ 0.04). In TT homo- models between the two populations. The higher prevalence of zygotes, the mean enzyme activity was 8.1 kU/l at a blood Hg level individuals homozygous for the T allele in the Cree population of 1 nmol/l (5th percentile), and decreased to 7.2 kU/l for a blood compared with that of the Inuit study population (13% vs 4%) Hg level of 180 nmol/l (95th percentile). could have impacted on the statistical power of finding such an interaction in the Inuit group. Following results from in vitro studies, which reported that DISCUSSION metals including Hg were more effective inhibitors of the 20 In this study, we measured serum PON1 activity, a biomarker of PON1R192 isoenzyme, we were particularly interested in investi- CVD, in a large sample of the Cree adults living in Eastern James gating whether this could be observed in vivo in our population Bay, who are exposed to MeHg through their fish-based diet. sample. We did not find an interaction between blood Hg level Blood mercury levels were not associated with PON1 activity when and the rs662 (Q192R) polymorphism, but rather with the b coefficients were adjusted for six PON1 variants and several rs705379 variant, with individuals bearing the T allele showing a

Table 1. Selected characteristics of Cree adults participating in the Multi-community Environment-and-Health Study in Iiyiyiu Aschii.

Characteristics Men Women All

N Geo-mean 95% CI N Geo-mean 95% CI N Geo-mean 95% CI

Age (years)a 369 41.1 39.4, 42.8 512 39.3 38, 40.7 881 40.1 39, 41.1 BMI (kg/m2) 346 31.9 31.2, 32.5 487 35.2 34.6, 35.9 833 33.8 33.4, 34.3 Blood mercury (nmol/l) 360 19.6 16.8, 23.0 494 14.9 13, 16.9 854 16.7 15.1, 18.5 Blood selenium (mmol/l) 360 2.2 2.2, 2.3 494 2.1 2.1, 2.2 854 2.2 2.2, 2.2 HDL-C (mmol/l) 360 1.2 1.1, 1.2 496 1.3 1.2, 1.3 856 1.2 1.2, 1.2 LDL-C (mmol/l) 357 2.7 2.6, 2.8 492 2.4 2.4, 2.5 849 2.5 2.5, 2.6 n-3 PUFA (%) 361 6.2 6.1, 6.4 496 6.3 6.2, 6.4 857 6.3 6.2, 6.4 PON1 activity (kU/l) 351 11.4 11, 11.8 494 11.5 11.1, 11.9 845 11.5 11.2, 11.7

Smoking status (%)a Everyday 130 36.4 31.4, 41.4 197 39.0 34.7, 43.3 327 38.0 34.7, 41 Occasionally 41 11.5 8.2, 14.8 62 12.3 9.4, 15.1 103 12.0 9.8, 14.1 Ex-smokers 150 42.0 36.9, 47.2 191 37.8 33.6, 42.1 341 39.5 36.3, 42.8 Never smoked 36 10.1 6.9, 13.2 55 10.9 8.2, 13.6 91 10.5 8.5, 12.6

Diabetes (%)a No 288 83.0 79, 87 358 74.6 70.7, 78.5 646 78.1 75.3, 80.9 Yes 59 17.0 13, 21 122 25.4 21.5, 29.3 181 21.9 19.1, 24.7

Fish consumptiona (meals/month) 358 6.5 4.7, 8.3 503 3.9 3.1, 4.7 861 4.9 4.1, 5.9 Abbreviations: BMI, body mass index; CI, conﬁdence interval; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; PUFAs, polyunsaturated fatty acids; PON1, paraoxonase 1. n-3 PUFAs: Sum of (C18:3, C18:4, C20:3, C20:4, C20:5, C22:5, C22:6) expressed as % of total fatty acids. aArithmetic mean.

& 2014 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 MeHg and serum PON1 activity in Cree people Drescher et al 612 Table 2. Crude and adjusted b coefﬁcients for the effect of selected variables on PON1 activity (log kU/l) in Cree adults.

Crude Adjusteda

Continuous variables b 95% CI P-value b 95% CI bstd P-value

Blood mercury (nmol/l) À 0.0008 À 0.012, 0.015 0.9 0.020 À 0.001, 0.042 0.08 0.07 Age (years) À 0.002 À 0.004, À 0.001 0.007 À 0.005 À 0.007, À 0.003 À 0.21 o0.0001 BMI (kg/m2) 0.0008 À 0.003, 0.004 0.6 0.001 À 0.002, 0.005 0.02 0.5 Blood selenium (mmol/l) 0.27 À 0.137, 0.680 0.3 0.43 0.034, 0.821 0.07 0.03 HDL-C (mmol/l) 0.22 0.124, 0.310 0.0001 0.22 0.123, 0.308 0.15 o0.0001 LDL-C (mmol/l) 0.10 0.020, 0.176 0.01 0.10 0.026, 0.170 0.08 0.008 n-3 PUFAs (%) À 0.08 À 0.197, 0.032 0.1 À 0.007 À 0.149, 0.135 À 0.004 0.9 Fish consumption (meals/month) À 0.01 À 0.024, 0.0003 0.003 À 0.01 À 0.024, 0.004 À 0.06 0.15

Categorical variables

Gender (ref: men) Women 0.008 À 0.040, 0.056 0.6 À 0.011 À 0.059, 0.037 À 0.02 0.6

Smoking (ref: never smoked) Everyday 0.009 À 0.077, 0.095 0.8 0.011 À 0.066, 0.088 0.02 0.8 Occasionally 0.032 À 0.071, 0.135 0.4 0.009 À 0.082, 0.100 0.008 0.8 Ex-smokers À 0.027 À 0.113, 0.059 0.5 0.030 À 0.044, 0.104 0.04 0.4

Diabetes (ref: no) Yes 0.02 À 0.038, 0.083 0.5 À 0.04 À 0.020, 0.097 0.04 0.2

Variants and genotypes

b b N b 95% CI P-value b 95% CI bstd P-value

rs705379 ( À 108C/T) CC 315 Ref. o0.0001 Ref. o0.0001 CT 394 À 0.26 À 0.299, À 0.211 À 0.70 À 1.003, À 0.400 À 0.67 TT 106 À 0.58 À 0.647, À 0.517 À 0.37 À 0.580, À 0.164 À 0.52

rs662 (Q192R) RR 262 Ref. 0.8 Ref. 0.2 QR 394 À 0.018 À 0.073, 0.037 0.047 À 0.024, 0.117 0.05 QQ 163 À 0.007 À 0.076, 0.061 0.051 À 0.002, 0.103 0.07

rs854560 (L55M) MM 674 Ref. o0.0001 Ref. o0.0001 LM&LL 145 À 0.36 À 0.414, À 0.297 À 0.27 À 0.338, À 0.207 À 0.29

rs854572 (-909C/G) CC 305 Ref. o0.0001 Ref. 0.3 CG 399 À 0.25 À 0.296, À 0.208 0.18 À 0.131, 0.482 0.17 GG 116 À 0.57 À 0.636, À 0.509 0.17 À 0.048, 0.380 0.23

rs854571 ( À 832C/T) CC 289 Ref. o0.0001 Ref. 0.2 CT 386 0.14 0.090, 0.193 À 0.06 À 0.144, 0.034 À 0.06 TT 137 0.29 0.225, 0.361 À 0.03 À 0.090, 0.028 À 0.04

rs705381 ( À 162C/T) CC 622 Ref. o0.0001 Ref. 0.7 CT&TT 189 0.03 0.056, 0.169 0.01 À 0.056, 0.078 0.01

Abbreviations: bstd, Standardized coefficient; BMI, body mass index; CI, confidence interval; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; PUFA, polyunsaturated fatty acid; PON1, paraoxonase 1. aCoefficients adjusted for all other variables in the table. bOverall P-value.

decreased PON1 activity with increasing blood mercury con- mercuric ions, through an interaction with Cys2His2 zinc-binding centration. The rs705379 ( À 108C/T) polymorphism is located domains.27 We speculate that in people bearing the À 108T in the promoter region of the gene, more specifically in a allele, for which interaction between the Sp1 and the promotor region responsive to transcription factors.25 The polymorphism sequence is already compromised, mercury ions would further disrupts a recognition sequence for Sp1,26 a zinc-finger transcrip- lower PON1 expression by decreasing Sp1 affinity for its DNA- tion factor whose DNA-binding activity can be inhibited by binding sequence.

Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 & 2014 Nature America, Inc. MeHg and serum PON1 activity in Cree people Drescher et al 613 variance in PON1 activity in the latter studies; their contribution Table 3. Multiple linear regression modela of the relationship in our study might have been too weak to reach statistical between serum PON1 activity (log kU/l) and blood mercury significance. concentration (log nmol/l), including the blood mercury–rs705379 Our study benefits from a large number of participants and interaction term. results from this research can be said to represent the general Variables b 95% CI P-value population of seven of the nine Cree communities. The relatively low participation rate (43%) of this study could introduce a Blood mercury 0.038 0.015, 0.061 0.001 selection bias. However, we have no reason to believe that the rs705379 ( À 108C/T) individuals who refused to participate differed in any way from CT À 0.134 À 0.215, À 0.053 o0.0001 our study participants. Steps were taken to ensure proper internal TT À 0.350 À 0.481, À 0.219 validity. Information bias was minimized by using direct measure- Mercury–rs705379 ments of all biological and clinical determinants and by the use of CT À 0.028 À 0.054, À 0.002 0.002 trained nurses. Recall bias could be present for the variables such TT À 0.059 À 0.099, À 0.019 as tobacco-use or fish consumption, but unlikely as participants Abbreviations: CI, confidence interval; PON1, paraoxonase 1. were not aware of the specific objectives regarding self-reported aThe model was adjusted for: blood selenium concentration, age, variables at the time of data collection. We limited confounding body mass index, high-density lipoprotein cholesterol concentration, biases by including a large set of biological, clinical and socio- low-density lipoprotein cholesterol concentration, total n-3 PUFA content demographic determinants that could have potentially impacted of red blood cell membrane phospholipids, monthly fish consumption on PON1 levels. Most of the confounding variables known at the frequency, gender, smoking status, diabetes and rs854560 (L55M) genotype. present time to influence PON1 levels were included; however, other confounding variables will surely be discovered in the future. Hg and selenium analyses were carried out by the INSPQ Toxicology Laboratory, which is accredited ISO 17025 by the Standards Council of Canada and is an international leader in analytical toxicology as well as a reference institution for inter- laboratory comparison programs. Using only one substrate for the analysis of PON1 activity could be considered a weakness, as PON1 activity is also determined by the nature of the substrate itself. Rainwater et al. reported that common environmental determinants impacted on activity differently depending on the substrate used (reflecting paraoxonase, arylesterase or lactonase). The covariates explained 5.2% of the phenotypic variance with paraoxonase substrate, whereas they had a larger impact when using arylesterase (25.1%) or lactonase substrates (21.2%).31 The use of only one substrate in our study could have resulted in an underestimation of the strength of covariate associations. Although there is no consensus on the optimal substrate to measure PON1 activity, using only one substrate may not reveal the complete spectrum of enzyme function. According to a recent literature review, there is sufficient robust evidence to indicate that MeHg exposure can lead to increased Figure 2. Relation of blood mercury concentration (log nmol/l) to 3 paraoxonase 1 (PON1) activity (log kU/l) by rs705379 (-108C/T) PON1 risks of adverse cardiovascular effects. It is tempting to speculate genotypes. PON1 activity was adjusted for all potential confounding that the decreased PON1 activity found among individuals who factors listed in Table 2 and rs854560 (L55M) PON1 variant. are exposed to MeHg through their diet and who are homozygous for the T allele of the rs705379 ( À 108C/T) variant may be less protected against LDL oxidation and in turn at increased risk of The positive, statistically significant association of PON1 activity atherosclerosis. However, the finding of a statistically significant with HDL-C levels was expected considering that the enzyme is interaction between blood mercury and rs705379 will need to be almost exclusively bound to HDLs. However, as reported by replicated in other studies. Future research using longitudinal Watson et al. and by Ansell and colleagues, the ability of HDL-C to follow-up data will focus on the association between CVD attenuate the oxidation of LDL is largely attributable to PON1, so occurrences and low levels of PON1 enzyme brought-on the protective effects are relative to the abundance of HDL through MeHg exposure. particles, which contain the protective enzyme rather than plasma HDL level per se.6,28 In the adjusted model, age was negatively associated with CONFLICT OF INTEREST PON1 activity. This association is in agreement with previous The authors declare no conflict of interest. studies, which have shown that PON1 activity decreases with age.29–31 This trend is thought to occur as a result of increased oxidative stress with age, known to deactivate the free sulfhydryl ACKNOWLEDGEMENTS groups of the cysteine-284 residue, thereby hindering the This scientific communication is a report from the Nituuchischaayihtitaau Aschii Multi- protein’s antioxidant functions.19,32,33 Community Environment-and-Health Longitudinal Study in Iiyiyiu Ashii supported by In contrast to results from some previous human studies,34,35 the Cree people of Northern Quebec, the Cree First Nations and the Cree Board of Health we did not find an association between diabetes and reduced and Social Services of James Bay through financial contributions from Niskamoon Corporation. We gratefully acknowledge the support of the Natural Sciences and PON1 activity levels. Similarly, the smoking status was not asso- Engineering Research Council of Canada’s Metals in the Human Environment Strategic ciated with reduced PON1 activity levels, although this relation- Network. A complete list of sponsors is available at http://www.ncbi.nlm.nih.gov/ 18,31 ship was previously reported in other studies. It is noteworthy pubmed/20077280. We also acknowledge the support of FRQS Environmental Health that these covariates accounted for a very small proportion of the Research Network.

& 2014 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 MeHg and serum PON1 activity in Cree people Drescher et al 614 REFERENCES 18 Costa LG, Vitalone A, Cole TB, Furlong CE. Modulation of paraoxonase (PON1) 1 Goyer R, Vasken Apshian H, Arab L, Bellinger D, Burbacher T, Burke T et al. activity. Biochem Pharmacol 2005; 69: 541–550. Toxicological Effects of Methylmercury. National Academy Press: Washington DC, 2000. 19 Gonzalvo CM, Gil F, Hernandez AF, Villanueva E, Pla A. Inhibition of paraoxonase 2 Clarkson TW, Magos L. The toxicology of mercury and its chemical compounds. activity in human liver microsomes by exposure to EDTA, metals and mercurials. Crit Rev Toxicol 2006; 36: 609–662. Chem Biol Interact 1997; 105: 169–179. 3 Roman HA, Walsh TL, Coull BA, Dewailly E, Guallar E, Hattis D et al. Evaluation of 20 Gencer N, Arslan O. Purification human PON1Q192 and PON1R192 isoenzymes the cardiovascular effects of methylmercury exposures: current evidence supports by hydrophobic interaction chromatography and investigation of the inhibition development of a dose-response function for regulatory benefits analysis. Environ by metals. J Chromatogr B Analyt Technol Biomed Life Sci 2009; 877: 134–140. Health Perspect 2011; 119: 607–614. 21 Ayotte P, Carrier A, Ouellet N, Boiteau V, Abdous B, Laouan Sidi EA et al. Relation 4 Salonen JT, Seppanen K, Nyyssonen K, Korpela H, Kauhanen J, Kantola M et al. between methylmercury exposure and plasma paraoxonase activity in Inuit Intake of mercury from fish, lipid peroxidation, and the risk of myocardial adults from Nunavik. Environ Health Perspect 2011; 119: 1077–1083. infarction and coronary, cardiovascular, and any death in eastern Finnish men. 22 Khan M, Wang F. Mercury-selenium compounds and their toxicological signi- Circulation 1995; 91: 645–655. ficance: toward a molecular understanding of the mercury-selenium antagonism. 5 Costa LG, Cole TB, Jarvik GP, Furlong CE. Functional genomics of the paraoxonase Environ Toxicol Chem 2009; 28: 1567–1577. (PON1) polymorphisms: effects on pesticide sensitivity, cardiovascular disease, 23 Pouliot M, Despres J, Lemieux S, Moorjani S, Bouchard C, Tremblay A et al. Waist and drug metabolism. Annu Rev Med 2003; 54: 371–392. circumference and abdominal sagittal diameter: best simple anthropometric 6 Watson AD, Berliner JA, Hama SY, La Du BN, Faull KF, Fogelman AM et al. indexes of abdominal visceral adipose tissue accumulation and related cardio- Protective effect of high density lipoprotein associated paraoxonase. Inhibition of vcascular risk in men and women. Am J Cardiol 1994; 73: 460–468. the biological activity of minimally oxidized low density lipoprotein. J Clin Invest 24 Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of 1995; 96: 2882–2891. low-density lipoprotein cholesterol in plasma, without use of the preparative 7 Bhattacharya T, Nicholls SJ, Topol EJ, Zhang R, Yang X, Schmitt D et al. ultracentrifuge. Clin Chem 1972; 18: 499–502. Relationship of paraoxonase 1 (PON1) gene polymorphisms and functional 25 Deakin S, Guernier S, James RW. Pharmacogenetic interaction between para- activity with systemic oxidative stress and cardiovascular risk. JAMA 2008; 299: oxonase-1 gene promoter polymorphism C-107T and statin. Pharmacogenet 1265–1276. Genomics 2007; 17: 451–457. 8 Mackness B, Hine D, Liu Y, Mastorikou M, Mackness M. Paraoxonase-1 inhibits 26 Deakin S, Leviev I, Brulhart-Meynet MC, James RW. Paraoxonase-1 promoter oxidised LDL-induced MCP-1 production by endothelial cells. Biochem Biophys Res haplotypes and serum paraoxonase: a predominant role for polymorphic position Commun 2004; 318: 680–683. -107, implicating the Sp1 transcription factor. Biochem J 2003; 372(Pt2): 643–649. 9 Mackness B, Durrington P, McElduff P, Yarnell J, Azam N, Watt M et al. Low 27 Rodgers JS, Hocker JR, Hanas RJ, Nwosu EC, Hanas JS. Mercuric ion inhibition of paraoxonase activity predicts coronary events in the Caerphilly Prospective Study. eukaryotic transcription factor binding to DNA. Biochem Pharmacol 2001; 61: Circulation 2003; 107: 2775–2779. 1543–1550. 10 Shih NA, Xia YR, Wang XP, Miller E, Castellani LW, Subbanagounder G. 28 Ansell BJ, Navab M, Hama S, Kamranpour N, Fonarow G, Hough G et al. Inflam- Combined serum paraoxonase knockout/apolipoprotein E knockout mice exhibit matory/antiinflammatory properties of high-density lipoprotein distinguish patients increased lipoprotein oxidation and atherosclerosis. J Biol Chem 2000; 275: from control subjects better than high-density lipoprotein cholesterol levels and 17527–17535. are favorably affected by simvastatin treatment. Circulation 2003; 108: 2751–2756. 11 Costa LG, Cole TB, Furlong CE. Paraoxonase (PON1): from toxicology to cardio- 29 Cakatay U, Kayali R, Uzun H. Relation of plasma protein oxidation parameters and vascular medicine. Acta Biomed 2005; 76(Suppl 2): 50–57. paraoxonase activity in the ageing population. Clin Exp Med 2008; 8: 51–57. 12 Furlong CE, Suzuki SM, Stevens RC, Marsillach J, Richter RJ, Jarvik GP et al. Human 30 Cherki M, Berrougui H, Isabelle M, Cloutier M, Koumbadinga GA, Khalil A. Effect PON1, a biomarker of risk of disease and exposure. Chem Biol Interact 2010; 187: of PON1 polymorphism on HDL antioxidant potential is blunted with aging. 355–361. Exp Gerontol 2007; 42: 815–824. 13 Mackness B, Mackness MI, Arrol S, Turkie W, Durrington P. Effect of the human 31 Rainwater DL, Rutherford S, Dyer TD, Rainwater ED, Cole SA, Vandeberg JL et al. serum paraoxonase 55 and 192 genetic polymorphisms on the protection by Determinants of variation in human serum paraoxonase activity. Heredity 2009; high density lipoprotein against low density lipoprotein oxidative modification. 102: 147–154. FEBS 1998; 423: 57–60. 32 Jaouad L, de Guise C, Berrougui H, Cloutier M, Isabelle M, Fulop T. Age-related 14 Aviram M, Billecke S, Sorenson R, Bisgaier C, Newton R, Rosenblat M et al. Para- decrease in high-density lipoproteins antioxidant activity is due to an alteration in oxonase active site required for protection against LDL oxidation involves its the PON1’s free sulfhydryl groups. Atherosclerosis 2006; 185: 191–200. free sulfhydryl group and is different from that required for its arylesterase/ 33 Aviram M, Billecke S, Sorenson R, Bisgaier C, Newton R, Rosenblat M et al. paraoxonase activityes: selective action of human paraoxonase allozymes Q Paraoxonase active site required for protection against LDL oxidation involves its and R. Arterioscler Thromb Vasc Biol 1998; 18: 1617–1624. free sulfhydryl group and is different from that required for its arylesterase/ 15 Brophy VH, Jarvik GP, Richter RJ, Rozek LS, Schellenberg GD, Furlong CE. Analysis paraoxonase activities: selective action of human paraoxonase allozymes Q and R. of paraoxonase (PON1) L55M status requires both genotype and phenotype. Arterioscler Thromb Vasc Biol 1998; 18: 1617–1624. Pharmacogenetics 2000; 10: 453–460. 34 Mackness B, Mackness M, Arrol S, Turkie W, Julier K, Abuasha B et al. Serum para- 16 Brophy VH, Hastings MD, Clendenning JB, Richter RJ, Jarvik GP, Furlong CE. oxonase (PON1) 55 and 192 polymorphism and paraoxonase activity and concentra- Polymorphisms in the human paraoxonase (PON1) promoter. Pharmacogenetics tion in non-insulin dependent diabetes mellitus. Atherosclerosis 1998; 139: 341–349. 2001; 11: 77–84. 35 Abbot CA, Mackness M, Kumar S, Boulton AJM, Durrington PN. Serum para- 17 Brophy VH, Jampsa RL, Clendenning JB, McKinstry LA, Jarvik GP, Furlong CE. oxonase activity, concentration, and phenotype distribution in diabetes mellitus Effects of 50 regulatory-region polymorphisms on paraoxonase-gene (PON1) and its relationship to serum lipids and lipoproteins. Atheroscler ThrombVasc Biol expression. Am J Hum Genet 2001; 68: 1428–1436. 1995; 15: 1812–1818.

Journal of Exposure Science and Environmental Epidemiology (2014), 608 – 614 & 2014 Nature America, Inc.