Cardiac Autonomic Dysfunction: Effects from Particulate Air Pollution and Protection by Dietary Methyl Nutrients and Metabolic Polymorphisms

Abstract

Background

Particulate air pollution is associated with cardiovascular mortality and morbidity. To help identify mechanisms of action and protective/susceptibility factors, we evaluated whether the effect of particulate matter <2.5 µm in aerodynamic diameter (PM_2.5) on heart rate variability (HRV) was modified by dietary intakes of methyl nutrients (folate, vitamin B6, B12, methionine) and related gene polymorphisms (C677T MTHFR and C1420T cSHMT).

Methods and Results

HRV and dietary data were obtained between 2000–2005 from 549 elderly men from the Normative Aging Study. In carriers of [CT/TT] MTHFR genotypes, the standard deviation of normal-to-normal intervals (SDNN) was 17.1% (95% CI, 6.5, 26.4; p=0.002) lower than in CC MTHFR subjects. In the same [CT/TT] MTHFR subjects, each 10 µg/m³ increase in PM_2.5 in the 48 hours before the examination was associated with a further 8.8% (95%CI: 0.2, 16.7; p=0.047) decrease in SDDN. In [CC] cSHMT carriers, PM_2.5 was associated with a 11.8% (95%CI: 1.8, 20.8; p=0.02) decrease in SDDN. No PM_2.5-SSDN association was found in subjects with either [CC] MTHFR or [CT/TT] cSHMT genotypes. The negative effects of PM_2.5 were abrogated in subjects with higher intakes (>median levels) of B6, B12, or methionine. PM_2.5 was negatively associated with HRV in subjects with lower intakes, but no PM_2.5 effect was found in the higher intake groups.

Conclusions

Genetic and nutritional variations in the methionine cycle affect HRV, either independently or by modifying the effects of PM_2.5.

INTRODUCTION

Reductions in Heart Rate Variability (HRV), a noninvasive measure of cardiac autonomic dysfunction that independently predicts cardiovascular mortality,¹ have been related to short-term exposure to particulate air pollution (PM), particularly to fine-particulate air pollution of <2.5 µM in aerodynamic diameter (PM_2.5).^2–6 This relation has been investigated to clarify mechanisms underlying the increased risk of cardiovascular disease (CVD) associated with PM_2.5 exposure observed in multiple investigations.^{7, 8}

Dietary methyl nutrients, including folate, the B-vitamins pyridoxine (B₆), cyanocobalamin (B₁₂), and methionine, are coenzymes or substrates in the methionine cycle that contribute to controlling biological processes^{9, 10} — such as methyl-group transfers, homocysteine synthesis, and redox states — that may be affected by PM exposure.^11–14 The activity of the methionine cycle is dependent on availability of dietary methyl nutrients^{15, 16} and is modified by genetic variations in metabolic genes.^{17, 18} In particular, the CT and TT genotypes of the C677T methylenetetrahydrofolate reductase (MTHFR) polymorphism have been associated with reduced enzyme activity,^{17, 19} and linked, though not consistently, with increased risk of CVD.²⁰ Conversely, the TT genotype of the C1420T cytoplasmic serine hydroxymethyltransferase (cSHMT) polymorphism has been associated with higher homocysteine levels²¹ and has been found to interact with MTHFR polymorphisms in determining increased CVD risk.¹⁸

Whether differences in dietary intakes of methyl nutrients or genetic variation in the methionine cycle modify the effects of PM_2.5 exposure on cardiovascular outcomes has never been tested. In the present study, we examined how the association of PM_2.5 with HRV in the Normative Aging Study, a repeated measure investigation of elderly subjects from the Boston metropolitan area, was affected by C677T MTHFR and C1420T cSHMT polymorphisms and by variation in dietary intakes of folate, vitamin B6, vitamin B12, and methionine.

MATERIALS AND METHODS

Study Population

Our study population consisted of 549 white males from the Normative Aging Study (NAS), a longitudinal study of aging established in 1963 by the U.S. Veterans Administration.²² Between November 2000 and June 2005, all participants still presenting for examination (n=676) were evaluated for HRV. Of these, 127 subjects were excluded because of heart arrhythmias, measurement time <3.5 minutes, or missing potential confounding data. The remaining 549 subjects had HRV measured either in one (n=363) or two visits (n=186). None of the subjects had a recent (<=6 months) myocardial infarction. The study participants were all male and 539 of them (97.3%) were Caucasians. This study was approved by the Institutional Review Boards of all participating institutions and all participants gave written informed consent to the study.

HRV Measurement

HRV was measured for seven minutes using a two-channel (five-lead) ECG monitor (Model: Trillium 3000; Forest Medical, East Syracuse, NY) while the subject was seated. Standard deviation of normal-to-normal intervals (SDNN), high frequency (HF) (0.15 to 0.4 Hz), and low frequency (LF) (0.04 to 0.15 Hz) were computed with a Fast-Fourier transform using software (Trillium-3000 PC-Companion Software; Forest Medical) complying with established guidelines.²³ We selected for the analysis the 4-consecutive minutes of ECG reading with the lowest number of artifacts.

Air Pollution and Weather Data

Continuous PM_2.5 was measured at a stationary monitoring site on the roof of Countway Library, Harvard University in downtown Boston, MA using a Tapered Element Oscillating Microbalance (TEOM) (Model 1400A, Rupprecht & Pataschnick Co., East Greenbush, NY). Meteorological data were obtained from the Boston-airport weather station. The 48-hour moving average of PM_2.5 was used as the exposure index, as this exposure period has shown the strongest association in previous studies.⁶

Semiquantitative Food-Frequency Questionnaires

Study subjects completed a Food-Frequency Questionnaire (FFQ) referring to intake in the prior year at every visit.¹⁸ FFQ data were available for 713 of the total 735 visits. Estimates of dietary intake, including folate, vitamin B₆, vitamin B₁₂, and methionine, were derived from the frequency and dosage information on the FFQ using software developed by the Nurses’ Health Study (NHS)²⁴ and processed by NHS operators. Validity and reliability of this FFQ for estimating daily vitamin intakes have been previously described.^{24, 25}

Genotyping methods

We performed genotyping of the C677T MTHFR (rs1801133) and C1420T cSHMT (rs1979277) polymorphisms on a subset of 362 of the 549 subjects included in the study. These 362 subjects were part of a prior analysis of a nested case-control study of CVD and controls selected by risk-set sampling.¹⁸ We estimated that data from the 362 subjects with available genotyping provided us with statistical power to detect effect modifications in the association between PM_2.5 and HRV equal to 86% for C677T MTHFR and 77% for C1420T cSHMT. These power calculations were performed on potential PM_2.5-effects on the HF component of HRV, which was associated with PM_2.5 exposure in a previous study on this population,¹³ assuming the same HF standard deviation and effect modification size as those observed in our recent work on hemochromatosis (HFE) gene polymorphisms.²⁶

DNA was extracted from stored frozen buffy coat of 7 mL whole blood, using the QiAmp DNA blood kits (QIAGEN). Genotypes of C677T MTHFR and C1420T cSHMT were determined by the TaqMan procedure using the allelic discrimination technique (ABI Prism 7900 Sequence Detection System, Applied Biosystems). Details of the genotyping are described elsewhere.¹⁸

Statistical Analysis

HRV measurements were log₁₀-transformed to improve normality. The following potential confounders were chosen a priori and included in the analysis: age, past/current coronary heart disease (CHD), BMI, mean arterial pressure, fasting blood glucose, cigarette smoking (never/former/current), alcohol consumption (2+ drinks/day, yes/no), use of beta-blockers, calcium-channel blockers, ACE inhibitors, room temperature, season, and 48h-moving average of outdoor apparent temperature. Potential nonlinearity between apparent temperature and HRV was accounted for using linear and quadratic terms. All independent variables were fitted as time-varying covariates.

Because our data included repeated measures of HRV for many participants, our data may lack independence. To deal with this, we fit a mixed effects model (PROC MIXED in SAS V9.0). We assumed:

$$Y_{\mathit{\text{it}}}=b_o+u_i+b_l{X}_{\mathit{\text{lit}}}+\dots +b_p{X}_{\mathit{\text{pit}}}+\beta {\mathit{\text{Pollution}}}_{\mathit{\text{it}}}+{\epsilon }_{\mathit{\text{it}}}$$

where Y_it is the logarithm of HRV in subject i at time t, b_o is the overall intercept, and u_i is the separate random intercept for subject i. In the above X_lit—X_pit are the covariates. We used this model to assess the effect of PM_2.5 on HRV. To evaluate the effect modification of PM_2.5 effect by gene polymorphisms or dietary intakes, we added interaction terms to a model that also included the main effects for both PM2.5 and the genetic/dietary factors. Dietary factors were entered in the model as time-varying variables.

Because MTHFR and SHMT genotype data were obtained from a convenience sample that represented a subset of our study population, stratum-weighted regression was used to obtain unbiased estimates, as indicated in a recent work on the use of extant case-control data for the analysis of additional outcomes.²⁷ All results including genotype data presented throughout the paper are obtained from mixed-models that used weights equal to 1 for cases and equal to the reciprocal of the probability of being sampled into the study for controls.²⁷ As a sensitivity analysis, we fitted non-weighted mixed-models that also included the original case-status variable or that were restricted only to the original control series, with no notable differences in the results.

All regression analyses presented in the paper were repeated after including heart rate as independent variable. Such adjustment by heart rate did not modify, compared with the results presented in this paper, the significance of the gene polymorphism main effects, as well as of the interactions of PM_2.5 with gene polymorphisms or dietary intakes.

The Authors had full access to the data and take responsibility for its integrity. All Authors have read and agreed to the manuscript as written.

RESULTS

Table 1 shows the characteristics of the entire study group, as well as of the subjects with and without genotype data. Genotype distributions were in Hardy-Weinberg equilibrium for both C677T MTHFR (p=0.95) and C1420T cSHMT (p=0.35). Subjects with genotype data, which included CVD cases and age-matched controls from a previous study on CVD,¹⁹ had older age, higher fasting blood glucose, and, presumably because of tighter control or other changes in cases after the CVD event, lower blood pressure and total cholesterol, as also suggested by the more frequent use of beta blockers (Table 1). Consistently, controls from the original study did not show differences when compared with subjects without genotype data, with the exception of older age and, as BMI is negatively associated with age in healthy elderly individuals, moderately lower BMI (Supplementary Table 1). Table 2 shows dietary intakes of methyl nutrients, HRV measures, and environmental data for the 735 visits included in the study. No differences were observed in the subset of visits (n=485) with available C677T MTHFR and C1420T cSHMT genotyping, compared with the entire study population.

Table 1.

Anthropometric and clinical characteristics^* [mean±SD or n(%)] of the study population in the Normative Aging Study

	All Subjects	Subjects with genotype data†	Subjects without genotype data†	p-value‡
	(n=549)	(n = 362)	(n = 187)
CVD incident cases (1961–1998)§, n (%)	141 (25.7%)	141 (39.0%)	0 (0.0%)	<0.001
Age (years)	72.8±6.7	74.1±6.6	70.4±6.0	<0.001
Body mass index (kg/m2)	28.1±4.1	28.0±4.0	28.4±4.1	0.06
Systolic blood pressure (mm Hg)	130.4±16.3	130.2±16.6	130.8±15.8	0.75
Diastolic blood pressure (mm Hg)	74.8±9.7	74.0±9.8	76.3±9.3	0.001
Mean arterial pressure (mm Hg)	93.3±10.5	92.7±10.6	94.4±10.2	0.02
Heart rate (beats/min)	70.7±6.8	70.7±6.5	70.9±7.5	0.79
Fasting blood glucose (mg/dL)	107.5±26.8	109.0±27.1	104.7±25.9	0.02
Total cholesterol (mg/dL)	194.9±37.8	192.4±38.0	199.8±36.8	0.002
HDL (mg/dL)	49.6±13.5	50.0±12.6	50.9±14.9	0.31
Triglyceride (mg/dL)	130.9±71.3	133.2±77.7	126.5±56.9	0.35
Smoking status, n (%)
Never smoker	173 (31.6%)	114 (31.5%)	59 (31.7%)
Current smoker	29 (5.3%)	14 (3.4%)	15 (8.1%)
Former smoker	346 (63.1%)	234 (64.6%)	112 (60.2%)	0.11
Alcohol intake (≥ 2 drinks/day), n (%)	102 (18.6%)	69 (19.1%)	33 (17.7%)	0.73
Diabetes mellitus, n (%)	77 (14.0%)	61 (16.9%)	16 (8.6%)	0.01
History of coronary heart disease, n (%)	156 (28.4%)	137 (37.9%)	19 (10.2%)	<0.001
History of Stroke, n (%)	35 (6.4%)	28 (7.7%)	7 (3.7%)	0.10
Hypertension, n (%)	383 (69.8%)	259 (71.6%)	124 (66.3%)	0.24
Use of β-blocker, n (%)	183 (33.3%)	132 (36.5%)	51 (27.3%)	0.04
Use of Ca-channel blocker, n (%)	73 (13.3%)	55 (15.2%)	18 (9.6%)	0.08
Use of ACE inhibitor, n (%)	116 (21.1%)	82 (22.7%)	34 (18.2%)	0.27
MTHFR 677C>T genotype
CC	NA	137 (37.8%)	NA
CT	NA	170 (47.0%)	NA
TT	NA	55 (15.2%)	NA
cSHMT 1420 C>T genotype
CC	NA	171 (47.2%)	NA
CT	NA	149 (41.2%)	NA
TT	NA	42 (11.6%)	NA

^*Information collected at the time of the first measurement of heart rate variability.

^†Subjects with or without genotype data for the C677T methylenetetrahydrofolate reductase (MTHFR) or C1420T cytoplasmic serine hydroxymethyltransferase (cSHMT) polymorphisms.

^‡P-value for differences between groups with and without genotype data from Fisher's exact test or Wilcoxon (Mann-Withney) test.

^§Incident cases of cardiovascular disease (CVD), including CHD and stroke, diagnosed between 1961–1998 were included in the nested case-control study on CVD¹⁸ for which genotypes were originally determined. Controls were sampled from the remaining cohort subjects by risk-set sampling.

Table 2.

Dietary intakes and environmental variables of the study population in the Normative Aging Study

	All visits		Visits in subjects with genotype data*		Visits in subjects without genotype data*		p-value§
	(n=735)		(n = 485)		(n=250)
Nutrient intake, geometric mean (95% CI)
Folate (µg/day)	460.2	(438.9–482.5)	467.3	(441.8–494.3)	447.4	(410.3–487.9)	0.33
Vitamin B6 (mg/day)	4.15	(3.83–4.50)	4.17	(3.78–4.60)	4.12	(3.59–4.72)	0.63
Vitamin B12 (µg/day)	11.9	(11.1–12.7)	11.7	(10.8–12.7)	12.1	(10.8–13.6)	0.83
Methionine (g/day)	1.86	(1.80–1.92)	1.83	(1.76–1.91)	1.90	(1.79–2.01)	0.38
Daily intake lower than recommended,34, 35 n (%)
Folate (<400 µg/day)	270	(38.2%)	173	(37.8%)	97	(39.0%)	0.63
Vitamin B6 (<1.7 mg/day)	97	(13.6%)	65	(14.0%)	32	(12.0%)	0.79
Vitamin B12 (<2.4 µg/day)	14	(2.0%)	11	(2.4%)	3	(1.2%)	0.29
Methionine (<19 mg/kg/day)	234	(33.3%)	158	(34.7%)	76	(30.9%)	0.31
Heart Rate Variability, geometric mean (95% CI)
SDNN,† msec	32.6	(31.2–34.0)	32.6	(30.9–34.4)	32.6	(30.4–34.9)	0.85
HF,† msec2	77.9	(69.8–86.9)	80.5	(70.2–92.3)	73.1	(60.9–87.7)	0.54
LF,† msec2	94.5	(86.5–103.2)	91.5	(82.0–102.0)	100.6	(86.5–117.0)	0.33
Environmental Variables, geometric mean (95% CI)
PM2.5,‡ (µg/m3), geometric mean (95% CI)	10.5	(10.0–10.9)	10.4	(9.9–11.0)	10.5	(9.8–11.4)	0.81
Outdoor temperature,‡ (°C), mean (95% CI)	11.1	(10.4–11.9)	11.7	(10.8–12.6)	10.1	(8.9–11.3)	0.08
Room temperature, (°C), mean (95% CI)	24.0	(23.9–24.1)	24.0	(23.9–24.2)	24.0	(23.8–24.2)	0.99

^*Subjects with genotype data for the C677T methylenetetrahydrofolate reductase (MTHFR) or the C1420T cytoplasmic serine hydroxymethyltransferase (cSHMT) polymorphisms

^†Difference in standard deviation of normal-to-normal intervals (SDNN) and power in high frequency (HF) (0.15 to 0.4 Hz) or low frequency (LF) (0.04 to 0.15 Hz) computed using a Fast-Fourier transform algorithm

^‡Average of hourly measurements of PM_2.5 and outdoor apparent temperature during the 48 hours before the heart rate variability measurement

^§Test for differences between groups with or without genotype data from mixed models or logistic regression with generalized estimating equations.

We calculated in multivariate models the adjusted percent change in HRV associated with the C677T MTHFR and C1420T cSHMT genotypes (Table 3). Subjects carrying the MTHFR 677 CT/TT genotypes exhibited a reduction of 17.1% in SDNN (95% CI: −25.4, −6.5; p=0.002), 33.6% in HF (95% CI: −50.7, −10.4; p=0.008), and 36.2% in LF (95% CI: −50.1, −18.3; p<0.001), relative to the CC genotype. cSHMT genotypes were not associated with HRV (Table 3).

Table 3.

Adjusted percent change (95% Confidence interval) in heart rate variability (HRV) associated with MTHFR 677C>T and cSHMT 1420C>T genotypes

HRV component*	Genotype	% change	(95% CI)	p-value
Main Effect of MTHFR 677C>T genotype on HRV*
SDNN	CC	Reference	-
	CT/TT	−17.1	(−26.4,−6.5)	0.002
HF	CC	Reference
	CT/TT	−33.6	(−50.7,−10.4)	0.008
LF	CC	Reference
	CT/TT	−36.2	(−50.1,−18.3)	<0.001
Main Effect of cSHMT 1420C>T on HRV*
SDNN	CC	Reference	-
	CT/TT	4.0	(−7.5,17.0)	0.51
HF	CC	Reference	-
	CT/TT	6.3	(−20.7,42.5)	0.68
LF	CC	Reference	-
	CT/TT	2.3	(−19.8,30.5)	0.85

^*Standard deviation of normal-to-normal intervals (SDNN), power in high frequency (HF) (0.15–0.4 Hz) and low frequency (LF) (0.04–0.15 Hz) computed using a Fast-Fourier transform algorithm.

The association between MTHR genotypes and HRV remained significant after ambient PM2.5 was added as independent variable to the models (Supplementary Table 3).

We estimated the association of PM_2.5 with HRV, overall and by C677T MTHFR and C1420T cSHMT genotypes (Table 4). In all subjects with genotype data, a 10 µg/m³ increase in ambient PM_2.5 level in the 48h before the HRV measurement was negatively associated, though non significantly, with SDNN, HF, and LF. In the full data set (Table 5), PM_2.5 was significantly associated with SDNN (−7.1%; −13.2, −0.6; p=0.03) and HF (−18.7%; −31.1, −4.0; p=0.01).

Table 4.

Adjusted percent change (95% Confidence interval) in heart rate variability (HRV) for each 10 µg/m³ of PM_2.5 in the 48h before the measurement, by MTHFR 677C>T and cSHMT 1420C>T genotype

HRV component*	Genotype	% change	(95% CI)	p-value	p-interaction
Main Effect of PM2.5 on HRV, all subjects with genotype data
SDNN	All Subjects	−6.0	(−13.5,2.0)	0.14
HF	All Subjects	−17.1	(−32.3,1.6)	0.07
LF	All Subjects	−8.2	(−22.1,8.2)	0.31
Effect of PM2.5 on HRV, by MTHFR 677C>T genotype
SDNN	CC	3.7	(−9.8,19.2)	0.61
	CT/TT	−8.8	(−16.7,−0.2)	0.047	0.19
HF	CC	4.7	(−25.7,47.5)	0.79
	CT/TT	−22.8	(−38.2,−3.5)	0.02	0.37
LF	CC	6.4	(−19.3,40.1)	0.66
	CT/TT	−12.1	(−26.6,5.4)	0.16	0.27
Effect of PM2.5 on HRV, by cSHMT 1420C>T genotype
SDNN	CC	−11.8	(−20.8,−1.8)	0.02
	CT/TT	−0.1	(−10.4,11.2)	0.98	0.02
HF	CC	−30.8	(−46.9,−9.8)	0.007
	CT/TT	−0.8	(−24.0,29.3)	0.95	0.03
LF	CC	−17.1	(−33.2,2.9)	0.09
	CT/TT	1.4	(−18.2,25.8)	0.90	0.10

^*Standard deviation of normal-to-normal intervals (SDNN), power in high frequency (HF) (0.15–0.4 Hz) and low frequency (LF) (0.04–0.15 Hz) computed using a Fast-Fourier transform algorithm.

Table 5.

Adjusted percent change (95% Confidence interval) in heart rate variability (HRV) for each 10 µg/m³ of PM_2.5 in the 48h before the measurement, by folate, methionine, and vitamin B6 and B12 intake

HRV component*	Intake Group	number of visits	% change	(95% CI)	p-value	p-interaction
Main effect of PM2.5 on HRV, all subjects
SDNN	All subjects	713	−7.1	(−13.2,/−0.6)	0.03	-
HF	All subjects	713	−18.7	(−31.1,/−4.0)	0.01	-
LF	All subjects	713	−11.8	(−23.2,/1.3)	0.08	-
Effect of PM2.5 on HRV, by folate intake‡
SDNN	< 495.8 µg/day	353	−8.8	(−16.4,/−0.4)	0.04
	≥ 495.8 µg/day	354	−5.7	(−13.8,/3.1)	0.20	0.57
HF	< 495.8 µg/day	353	−18.9	(−34.6,/0.6)	0.06
	≥ 495.8 µg/day	354	−20.0	(−35.7,/−0.3)	0.05	0.92
LF	< 495.8 µg/day	353	−15.7	(−29.5,/0.9)	0.06
	≥ 495.8 µg/day	354	−9.9	(−24.9,/8.2)	0.27	0.57
Effect of PM2.5 on HRV, by Vitamin B6 intake†
SDNN	< 3.65 mg/day	357	−13.1	(−20.0,/−5.5)	0.001
	≥ 3.65 mg/day	356	1.9	(−7.1,/11.7)	0.69	0.006
HF	< 3.65 mg/day	357	−27.4	(−40.9,/−10.6)	0.003
	≥ 3.65 mg/day	356	−5.2	(−24.3,/18.8)	0.65	0.06
LF	< 3.65 mg/day	357	−20.0	(−32.6,/−4.9)	0.01
	≥ 3.65 mg/day	356	0.2	(−17.0,/20.9)	0.99	0.06
Effect of PM2.5 on HRV, by Vitamin B12 intake†
SDNN	< 11.1 µg/day	356	−12.2	(−19.1,/−4.7)	0.002
	≥ 11.1 µg/day	357	1.3	(−7.8,/11.4)	0.78	0.01
HF	< 11.1 µg/day	356	−27.3	(−40.6,/−11.0)	0.002
	≥ 11.1 µg/day	357	−4.9	(−24.6,/20.1)	0.68	0.06
LF	< 11.1 µg/day	356	−19.2	(−31.8,/−4.4)	0.01
	≥ 11.1 µg/day	357	0.2	(−17.5,/21.6)	0.99	0.07
Effect of PM2.5 on HRV, by methionine intake†
SDNN	< 1.88 g/day	352	−11.9	(−18.9,/−4.1)	0.003
	≥ 1.88 g/day	351	3.0	(−6.3,/13.1)	0.54	0.007
HF	< 1.88 g/day	352	−25.7	(−39.6,/−8.7)	0.005
	≥ 1.88 g/day	351	0.2	(−20.4,/26.0)	0.99	0.03
LF	< 1.88 g/day	352	−20.9	(−33.4,/−6.0)	0.008
	≥ 1.88 g/day	351	4.1	(−14.1,/26.2)	0.68	0.02

^*Standard deviation of normal-to-normal intervals (SDNN), power in high frequency (HF) (0.15–0.4 Hz) and low frequency (LF) (0.04–0.15 Hz) computed using a Fast-Fourier transform algorithm.

^†Subjects were divided in the high and low intake groups using the median values of folate, methionine, vitamin B6, and vitamin B12 intakes of the study population

In subjects carrying the MTHFR 677 CT/TT genotypes (Table 4), PM_2.5 level was associated with significant decreases in both SDNN (−8.8%; 95% CI −16.7, −0.2; p=0.047) and HF (−22.8%; 95% CI −38.2, −3.5; p=0.02), while no PM_2.5-related change was found in MTHFR 677 CC subjects. However, the statistical interactions between PM_2.5 level and C1420T cSHMT genotypes were not statistically significant (p≥0.19).

In subjects carrying the cSHMT 1420 CC genotype, PM_2.5 level was associated with significant decreases in SDNN (−11.8%; 95% CI −20.8, −1.8; p=0.02) and HF (−30.8%; 95% CI −46.9, −9.8; p=0.007), while no significant PM_2.5-related change was found in CT/TT subjects (Table 4). The statistical interactions between PM_2.5 level and C1420T cSHMT genotypes were statistically significant for both SDNN (p=0.02) and HF (p=0.03).

In our data, the MTHFR 677 CT/TT genotypes were associated with increased risk of CVD (Supplementary Table 4), as defined for the original case-control study.¹⁸

When subjects were divided according to their dietary intakes of folate, vitamin B6, vitamin B12, or methionine, we found that the negative effect of PM_2.5 on HRV was abrogated in subjects with B6, B12, or methionine higher than the median daily intake of the study population (Table 5). In particular, the association of PM_2.5 with SDDN was significantly modified by B6, B12, and methionine intakes (p-interaction<0.05). The modification of the association of PM_2.5 with HF and LF was statistically significant for differences in methionine intake (p-interaction<= 0.03), and only borderline significant for B6 and B12 (p-interaction between 0.06–0.07). When we evaluated the association of dietary intakes with HRV regardless of PM_2.5 exposure (Supplementary Table 5), vitamin B6, vitamin B12, and methionine intakes exhibited positive associations — generally non-significant — with HRV. A significant increase in SDNN was found in association with methionine intake above the median. (11.4%; 95% CI 2.0, 21.7; p=0.02) Other potential modifiers of the PM_2.5-HRV association, such as CHD, obesity, diabetes or hypertension, were not associated with methyl-nutrient intakes in this population (Supplementary Table 6).

Throughout the manuscript we have presented modifications in HRV as percent changes. Changes on the original scale are presented in Supplementary Tables 7–9.

DISCUSSION

The present study, based on an elderly population in Boston, Massachusetts, showed that genetic and nutritional variations in the methionine-cycle metabolism modified HRV, either independently or by modifying the effects of PM_2.5. We demonstrated that HRV outcomes were affected by multiple components of the methionine cycle, including polymorphisms in genes coding for enzyme proteins (MTHFR and cSHMT), as well as dietary intakes of enzyme cofactors (B6, B12) and cycle substrates (methionine).

We found that subjects with MTHFR 677 CT/TT genotypes had lower HRV than subjects with the CC genotypes. This finding is in the same direction as the results of a comprehensive meta-analysis²⁰ — based on 11,162 CHD cases and 12,758 controls from 40 different studies — that showed that TT carriers had significantly higher risk of CHD. However, the MTHFR 677 TT genotype appeared to be associated with increased CHD risk only in European populations, and speculation has been brought about on whether dietary or other characteristics abrogated in North American populations the C677T MTHFR effect.²⁰ Our findings in this elderly population suggest that at least some age-groups of the U.S. population may not be protected against the negative effects of C677T MTHFR on cardiac function.

The negative association between PM_2.5 and heart rate variability was modified by both C677T MTHFR and C1420T cSHMT polymorphisms, though the effect modification was significant for the C1420T cSHMT polymorphism only. PM_2.5 effects on HRV were stronger in subjects with the MTHFR 677 CT/TT and cSHMT 1420 CC genotypes, which have been associated with reduced enzyme activity,^{17, 19, 21} and increased risk of CVD.^{18, 20} MTHFR, though not directly part of the methionine cycle, is the key limiting enzyme required for the conversion of 5–10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, the methyl-group donor required for the remethylation of homocysteine to methionine.²⁸ cSHMT produces the MTHFR substrate — 5–10-methylenetetrahydrofolate — from tetrahydrofolate in a B6-dependent reaction. Thus the effects of both the MTHFR 677 CT/TT and cSHMT 1420 CC genotypes may be mediated through a reduction in the methionine cycle activity.

The reduction in HRV associated with PM_2.5 level was abrogated in subjects with intakes of B6, B12, and methionine higher than the median level of our study population. Conversely, PM_2.5 level was negatively associated with all measures of HRV in subjects with lower intakes. These results suggest that lower availability of B6, B12, or methionine, as well as genetically-determined reductions in key enzymatic activities, may reduce methionine-cycle dependent cell functions that counteract PM_2.5 effects. The methionine cycle communicates with different pathways presiding over varied cell functions, including DNA methylation and glutathione synthesis.^{9, 15} The methyl donors produced by the methionine cycle contribute to DNA methylation as substrates of DNA methyl transferases. Global DNA methylation content in blood leukocytes and other tissues decreases with aging, a finding that has been related to the age-associated increase in cardiovascular risk,²⁹ and oxidative DNA damage such as that following PM_2.5 exposure may interfere with methylation processes,³⁰ thus also resulting in genomic hypomethylation. Genetic or dietary factors that increase the production of methyl donors may prevent the potential loss of DNA methylation that may be caused by PM_2.5 exposure.

In our previous work in the Normative Aging Study, we showed that the association of PM_2.5 with reduced HRV was stronger in subjects with GST deletion, a common polymorphism that impairs glutathione-related responses to oxidative stress.¹³ Glutathione is synthesized from homocysteine, also a substrate of the methionine cycle, through additional B6-dependent reactions.⁹ Rodents exposed to concentrated urban particles evinced increased reactive oxygen species in both the lung and the heart,³¹ an effect muted by pre-administration of N-acetyl cysteine, a glutathione precursor and potent antioxidant.³² Experimental evidence has shown that a meal rich in methionine shifts the cycle activity toward glutathione production.⁹ Thus, reduction in methionine cycle activity may represent an additional mechanism modulating particle effects by reducing oxidative stress defenses.

The median intakes in our populations were above the current Dietary Reference Intakes (DRI) of folate, B6, B12, and methionine. However, a relatively high percentage of individuals had daily intakes of methionine lower than the DRI. The nutritional profile of the study subjects, that also included a large majority showing adequate B6 and B12 intakes, indicate that the usual diet in the study population was rich in meat with less intake of vegetables and dairy products. Our findings indicate that differences in B6 and B12 intakes in the range above the current DRI levels may modify the cardiovascular effects of air pollution. This finding, together with the increased particle-related risk in the subjects with lower methionine intake, would warrant, if confirmed, a reassessment of current strategies for methyl nutrient supplementation.

A potential limitation of this study is that we used ambient PM_2.5 concentrations from a single monitoring site as a surrogate for recent exposure to PM_2.5. A recent study comparing ambient concentrations at this site with personal exposures in Boston has shown a high longitudinal correlation³³ between the two measurements; the study also reported that PM_2.5 concentrations were spatially homogeneous over the Boston area. This suggests that our use of ambient concentrations is reasonable and the resulting exposure error is likely to be non-differential. In our analyses, we considered several potential confounding factors that may have influenced HRV measures, as we adjusted our models for age, existing diagnosis of CHD, body mass index, mean arterial pressure, fasting blood glucose, cigarette smoking, alcohol consumption, room temperature, outdoor apparent temperature, season, and use of beta-blockers, calcium channel blockers, and ACE inhibitors. Therefore, chances that the observed associations reflected bias due to confounders are minimized.

Our analyses on C677T MTHFR and C1420T cSHMT polymorphisms were based on a subset of the study population that had been previously included in a case-control study on CVD nested in the Normative Aging Study cohort.¹⁸ The use of such convenience sample with extant genotype data did not appear to have produced bias in our results, as analyses restricted to controls from the original case-control series, or adjusted by CVD case-status confirmed our findings.²⁷

Our results can only be generalized to an aged population that consists of older males who are almost all white. The effect on women and children as well as different ethnic groups should be addressed in future studies, particularly in relation to the exposure of different population groups to PM_2.5 with varying geographical location, occupation, socioeconomic status and behavioral characteristics. Other health outcomes of PM_2.5 including respiratory responses may also be modified by genetic variations in the methionine pathway or differences in B6, B12, and methionine intake. Our findings provide novel hypotheses to pursue further research to investigate the mechanisms of action of air particles and ultimately to identify measures to prevent CVD and reduce the effects of air pollution in human populations.