Re:비타민 B 복합체 대기오염의 미립자물질을 막는 매력적인 약물 - 논문 읽어야

[문형철]

beyond reason

어떤 사람에게는 좋은 음식이 어떤 사람에게는 독이 된다. 

비타민 b 대기오염의 미립자물질을 막는 매력적인 약물

(비타민 B9, 6, 12) - 호모시스테인 염증물질 억제!!

Proc Natl Acad Sci U S A. 2017 Mar 28; 114(13): 3503–3508. 

Published online 2017 Mar 13. doi: 10.1073/pnas.1618545114

Medical Sciences

B vitamins attenuate the epigenetic effects of ambient fine particles in a pilot human intervention trial

Jia Zhong,a,1 Oskar Karlsson,b,c Guan Wang,d Jun Li,e,f Yichen Guo,g Xinyi Lin,h Michele Zemplenyi,gMarco Sanchez-Guerra,i Letizia Trevisi,j Bruce Urch,k,l,m,n Mary Speck,k Liming Liang,g Brent A. Coull,gPetros Koutrakis,j Frances Silverman,k,l,m,n Diane R. Gold,j,o Tangchun Wu,e,f and  Andrea A. Baccarellia

Author information Copyright and License information Disclaimer

This article has been corrected. See Proc Natl Acad Sci U S A. 2017 April 18; 114(16): E3367.

See letter "B vitamins and pollution, an interesting, emerging, yet incomplete picture of folate and the exposome" in volume 114 on page E3878.

See "Reply to Lucock et al.: Significance of interpretation and misinterpretation of a small mechanistic study" in volume 114 on page E3880.

This article has been cited by other articles in PMC.

Associated Data

Data Availability Statement

Go to:

SIGNIFICANCE

Air pollution is a major public health concern worldwide. The molecular mechanistic underpinnings of the health effects of air pollution are not fully understood, and the lack of individual-level preventative options represent a critical knowledge gap. Our study demonstrated the epigenetic effects of air pollution and suggested that B vitamins might be used as prevention to complement regulations to attenuate the impact of air pollution on the epigenome. Our study inaugurated a line of research for the development of preventive interventions to minimize the adverse effects of air pollution on potential mechanistic markers. Because of the central role of epigenetic modifications in mediating environmental effects, our findings might be extended to other toxicants and environmental diseases.

Keywords: air pollution, B vitamins, DNA methylation, mitochondria

Go to:

ABSTRACT

Acute exposure to fine particle (PM2.5) induces DNA methylation changes implicated in inflammation and oxidative stress. We conducted a crossover trial to determine whether B-vitamin supplementation averts such changes. Ten healthy adults blindly received a 2-h, controlled-exposure experiment to sham under placebo, PM2.5 (250 μg/m3) under placebo, and PM2.5 (250 μg/m3) under B-vitamin supplementation (2.5 mg/d folic acid, 50 mg/d vitamin B6, and 1 mg/d vitamin B12), respectively. We profiled epigenome-wide methylation before and after each experiment using the Infinium HumanMethylation450 BeadChip in peripheral CD4+ T-helper cells. PM2.5 induced methylation changes in genes involved in mitochondrial oxidative energy metabolism. B-vitamin supplementation prevented these changes. Likewise, PM2.5depleted 11.1% [95% confidence interval (CI), 0.4%, 21.7%; P = 0.04] of mitochondrial DNA content compared with sham, and B-vitamin supplementation attenuated the PM2.5 effect by 102% (Pinteraction = 0.01). Our study indicates that individual-level prevention may be used to complement regulations and control potential mechanistic pathways underlying the adverse PM2.5 effects, with possible significant public health benefit in areas with frequent PM2.5 peaks.

Historical episodes of air pollution peaks were shown to be associated with up to >10 times increased death rates (1). According to the WHO, 92% of the world’s population currently lives in places where air quality levels exceed the WHO limits (2). Ambient PM2.5 (particles with an aerodynamic diameter of <2.5 µm) pollution is one of the most prominent air pollutants (3), because they deposit in the respiratory bronchioles and the alveoli and stimulate local and systemic inflammation and oxidative stress (4). Over the past few decades, substantial lowering of ambient PM2.5 levels has been achieved through large-scale emissions control policies (5). However, exposure peaks with adverse health consequences are still frequently recorded (6, 7), even in areas typically exhibiting low levels (5). The molecular mechanisms underlying PM2.5’s health effects are not fully understood, and the lack of preventative options at the individual level adds complexity to tackling this major public health challenge.

Recent studies in environmental epigenetics provide opportunities to understand the mechanistic underpinnings of exposure-related health effects and to develop novel individual-level interventions. DNA methylation, a potentially modifiable epigenetic mechanism, can regulate gene expression‎ and chromosome integrity via addition of methyl groups to cytosine residues (8). The dynamic DNA landscape can be rapidly altered in peripheral leukocytes following PM2.5 exposure; indeed, such changes are postulated to underlie PM-induced systemic inflammation and oxidative stress (9, 10). Most evidence of this phenomenon in humans is based on a heterogeneous mixture of leukocytes (9–12), but a loss of methylation in inflammatory genes and subsequent inflammatory responses, specifically in circulating Th cells, are observed in vivo after environmental challenge (13, 14). Notably, DNA methylation is dependent on a biochemical cycle that supplies methyl groups (CH3) while relying on methyl nutrients (i.e., B vitamins including folic acid, vitamin B6 and B12; and amino acids including methionine, betaine, and choline) (8, 15, 16). In animal studies, a methyl-nutrient-deficient diet leads to aberrant DNA methylation patterns (17), and administration of methyl nutrients enables restoration of epigenetic status (15, 18–20). Likewise, human studies show that dietary methyl nutrient intervention influences the plasticity of DNA methylation (21). The potential for epigenetic modulation has also been demonstrated in the presence of environmental stressors in animal models—Dolinoy et al. successfully used methyl nutrients to avert the DNA hypomethylation induced by bisphenol A exposure (22). These findings opened new avenues for the application of epigenetic intervention to reduce the health effects of air pollution. However, to date, epigenetic intervention in humans in the context of air pollution has not been described.

The present study is a placebo-controlled crossover pilot intervention trial with controlled human exposure experiments to concentrated ambient fine particles (PM2.5). We hypothesized that acute PM2.5 exposure can rapidly modify the DNA methylation profile in peripheral CD4+ Th cells—the most prolific cytokine producer mediating PM2.5-induced inflammatory responses—and that PM2.5-induced DNA methylation changes can be reduced by B vitamins (i.e., folic acid, vitamin B6, and B12), the primary source of methyl groups (Fig. 1).

Open in a separate window

Fig. 1.

Proposed conceptual model linking fine particulate matter (PM2.5), systemic oxidative stress and inflammation, and altered DNA methylation landscape in Th cells. We hypothesized that PM2.5 inhalation triggers local and systemic inflammation and oxidative stress, which alters the DNA methylation landscape in circulating CD4+ Th cells and further stimulates CD4+ Th cells. In return, stimulated CD4+ Th cells undergo more epigenetic remodeling—possibly due to modulated methyl group availability—therefore creating a vicious circle which amplifies the inflammatory and oxidative effects of PM2.5.

Go to:

RESULTS

Study Population, Plasma B-Vitamin Concentrations, and Exposure Levels. 

In the present crossover placebo-controlled trial, 10 volunteers completed 30 controlled exposure experiments following the same order (Fig. S1): seven volunteers aged 19–29 y and three aged 30–49 y. Four volunteers were white, three were Asian, and three were other races. Six volunteers were female and three volunteers had a body mass index (BMI) ≥25. All volunteers maintained consistent dietary patterns throughout the trial.

Fig. S1.

Study design: a single-blind, cross-over intervention trial with controlled exposure experiments in 10 healthy volunteers.

The targeted concentration of PM2.5 exposures was 250 µg/m3. However, the actual PM2.5 concentrations varied among controlled exposure experiments to PM2.5 (100.6–287.5 µg/m3; median: 234.0 µg/m3). There was no significant difference (P = 0.38) in PM2.5 concentration between PM2.5 experiment under placebo [median: 219.1 µg/m3; interquartile range (IQR): 33.1 µg/m3] and PM2.5 experiment under B vitamins (median: 237.2 µg/m3; IQR: 48.7 µg/m3).

To confirm‎ that supplementation affected circulating nutrient levels, we measured plasma B vitamins before and after placebo and supplementation. The median plasma concentrations of folic acid and vitamins B6 and B12 were 35 nmol/L (IQR: 14 nmol/L), 41 nmol/L (IQR: 16 nmol/L), and 292 pmol/L (IQR: 72 pmol/L) before sham experiment, respectively. After volunteers took placebos for 4 wk, their median plasma concentrations were similar: 39 nmol/L (IQR: 24 nmol/L) for folic acid (P = 0.82), 37 nmol/L (IQR: 18 nmol/L) for vitamin B6 (P = 0.75), and 262 pmol/L (IQR: 214 pmol/L) for vitamin B12(P = 0.42). B-vitamin supplementation significantly increased the median plasma concentrations of folic acid (56 nmol/L; IQR: 13; P = 0.02), vitamin B6 (428 nmol/L; IQR: 321; P = 0.004), and vitamin B12 (511 pmol/L; IQR: 85; P = 0.01). The abovementioned P values were based on Wilcoxon signed-rank test, a nonparametric paired difference test.

CD4+ Th Cell Purity. 

All CD4+ Th cell samples’ purity were over 80% (Table S1), with only minor contamination from CD8 T cells, B cells, granulocytes, and natural killer cells. The median purity of samples collected at sham experiment, PM2.5 experiment under placebo, and PM2.5 experiment under B vitamins was 96.9% (IQR: 5.1), 94.8% (IQR: 7.1), and 96.1% (IQR: 4.7), respectively.

Table S1.

The Houseman estimates of different leukocyte types in isolated CD4+ Th cells

Cell type	Median, %	IQR, %	Minimum, %	Maximum, %
Exposure 1
CD4 Th cells	96.9	5.1	81.9	100.0
CD8 T cells	2.7	3.7	0.0	5.9
B cells	0.3	1.1	0.0	8.2
Granulocytes	0.0	0.2	0.0	4.7
Monocytes	0.0	0.0	0.0	0.0
Nature killer cells	0.0	0.0	0.0	4.7
Exposure 2
CD4 Th cells	94.8	7.1	81.0	99.7
CD8 T cells	3.1	2.5	0.00	6.0
B cells	0.8	2.9	0.00	5.2
Granulocytes	0.0	0.87	0.00	4.9
Monocytes	0.0	0.00	0.00	0.0
Nature killer cells	0.0	0.00	0.00	7.7
Exposure 3
CD4 Th cells	96.1	4.7	85.1	100.0
CD8 T cells	2.4	4.0	0.0	5.8
B cells	0.4	1.0	0.0	3.9
Granulocytes	0.0	0.2	0.0	6.1
Monocytes	0.0	0.0	0.0	0.0
Nature killer cells	0.0	0.0	0.0	3.1

Open in a separate window

IQR, interquartile range.

Effect of PM2.5 and B-Vitamin Supplementation on DNA Methylation. 

Two-hour PM2.5 exposure substantially modified DNA methylation in CD4+ Th cells, and these changes were prevented by B-vitamin supplementation (Fig. 2). Because our study is limited in power, we present only the top 10 loci, selected following the method proposed by Maccani et al. (23)—first based on effect size and then P value (Table S2 and Fig. 2A). In the absence of B-vitamin supplementation, PM2.5 exposure either increased or decreased DNA methylation levels, compared with sham, at these loci (Fig. 2A). Quantile-Quantile plots for expected vs. observed distribution of P values showed minimal genomic inflation with a lambda of 1.03 (Fig. 2B).

Open in a separate window

Fig. 2.

The effect of 2-h exposure to PM2.5 on DNA methylation landscape and the reduction of PM2.5 effect by B-vitamin supplementation. A is the Volcano plot depicting the distribution of estimated effect of PM2.5 exposure on the epigenome. Each dot represents the estimated PM2.5 effect for one CpG. The vertical lines indicate suggestive threshold based on effect size, whereas the horizontal line reflects the suggestive threshold based on statistical significance. B is the quantile–quantile plot for associations of PM2.5 with DNA methylation in circulating CD4+Th cells. C represents the top 10 loci associated with PM2.5 and the reduction of PM2.5 effect by B vitamins. Bar height indicates PM2.5 effect, whereas the gray part indicates the magnitude of effect attenuation by B vitamins. Dis the Manhattan plot representing the chromosome location of each loci. The dashed horizontal line reflects the suggestive threshold for statistical significance. Analyses were adjusted for season, chamber humidity, and temperature.

Table S2.

Top 10 loci associated with 2-h PM2.5 exposure and the reduction in PM2.5 effects by B-vitamin supplement

Illumina CpG designation	Genomic position	Relation to CpG islands	Gene symbol	Change in methylation levels*	P	Reduction in PM2.5 effects by B vitamins (%)	P
Increased methylation with PM2.5 exposure
cg06194186	Chr 2: 207803693		CPO	1.74	2.3 × e−5	−1.00 (57)	0.003
cg07689821	Chr 6: 33290736	Island	DAXX	1.72	1.3 × e−5	−0.84 (49)	0.01
cg00068102	Chr 2: 121619261		GLI2	1.67	4.9 × e−5	−1.22 (73)	0.0007
cg00647528	Chr 18: 32957127	Island	ZNF396	1.66	4.7 × e−5	−0.52 (31)	0.13
cg15426626	Chr 12: 132984261†	Island		1.60	7.6 × e−6	−0.73 (45)	0.01
Decreased methylation with PM2.5 exposure
cg10719920	Chr 21: 46724843†	S_Shelf		−1.77	1.0 × e−6	0.49 (28)	0.07
cg21986027	Chr 6: 169238138†			−1.68	1.2 × e−5	1.27 (76)	0.0002
cg17157498	Chr 19: 1383493	Island	NDUFS7	−1.67	1.3 × e−5	1.24 (74)	0.0004
cg08075528	Chr 19: 1961017	Island	CSNK1G2	−1.63	5.8 × e−5	1.03 (63)	0.004
cg26995744	Chr 2: 200322034	Island	SATB2	−1.61	3.0 × e−5	1.14 (71)	0.001

Open in a separate window

*To calculate P values, DNA methylation β values were normal-transformed based on rank.

†Illumina array annotation indicates that these loci are not located within an annotated CpG region and are not associated with any gene.

Table S2 presents the genomic position, relation to CpG islands, and gene symbol for the 10 loci. The top two loci that are associated with known genes were cg06194186 and cg17157498. Locus cg06194186 is located in the promoter region (TSS1500) of the carboxypeptidase O (CPO) gene, and locus cg17157498 is located in the promoter region (TSS1500) of the NADH dehydrogenase (ubiquinone) Fe-S protein 7 (NDUFS7) gene (Figs. S2 and ​andS3).S3). Fig. 3 presents the shift in methylation level distributions and corresponding IQRs pre and post each exposure experiment for these two loci.

Open in a separate window

Fig. 3.

Distribution of DNA methylation levels of cg06194186 and cg17157498 before and after each exposure experiment. Red solid lines and green dashed lines indicate the distribution of methylation levels measured before and after exposure, respectively.

Open in a separate window

Fig. S2.

Genome location of cg06194186 and related region. A displays the location of cg06194186 on chromosome two. B shows the region covering chromosome 2: 207303694–208303693. Reference genes are noted in different colors, classified by protein-coding function. The red line indicates that cg06194186 is related to the CPO gene, a protein-coding gene. C is the zoomed in view for cg06194186. There are no short variants (SNPs and indels) for this region genotyped in admixed American (AMR), East Asian (EAS), and European (EUR) individuals by the 1000 Genomes project (phase 3) with frequency of at least 1%.

Fig. S3.

Genome location of cg17157498 and related region. A displays the location of cg17157498 on chromosome 19. Bshows the region covering chromosome 19: 883494–1883493. Reference genes are noted in different colors, classified by protein-coding function. The red line indicates that cg17157498 is related to the NDUFS7 gene, a protein-coding gene. C is the zoomed in view for cg17157498. There are no short variants (SNPs and indels) for this region genotyped in admixed American (AMR), East Asian (EAS), and European (EUR) individuals by the 1000 Genomes project (phase 3) with frequency of at least 1%.

Four-week B-vitamin supplementation attenuated the PM2.5 effect by 28–76% at the top 10 loci. Supplementing B vitamins resulted in a reduction in effect size by 57% for cg06194186, 49% for cg07689821, 73% for cg00068102, 31% for cg00647528, 45% for cg15426626, 28% for cg10719920, 76% for cg21986027, 74% for cg17157498, 63% for cg08075528, and 71% for cg26995744, respectively (Table S3 and Fig. 2C).

Table S3.

Effect of 2-h exposure to sham or fine particles (PM2.5) on mitochondrial DNA copy number and the protective effect due to B-vitamin supplementation

Model	Change,* %	P	95% CI
Original model†
Postexposure
Effect of PM2.5, compared with sham, without supplementation	−0.3	0.94	−10.1–9.5
Effect of PM2.5, compared with sham, with supplementation	−1.1	0.79	−9.9–7.6
PM2.5 effect reduction by B vitamins	−0.8	0.79	−7.2–5.6
24-h postexposure
Effect of PM2.5, compared with sham, without supplementation	−11.1	0.04	−21.7 to −0.4
Effect of PM2.5, compared with sham, with supplementation	0.2	0.96	−8.3–8.8
PM2.5 effect reduction by B vitamins	11.3	0.01	2.0–20.6
Sensitivity analysis†
Postexposure
Effect of PM2.5, compared with sham, without supplementation	−0.8	0.85	−9.1–7.6
Effect of PM2.5, compared with sham, with supplementation	−0.5	0.88	−7.5–6.5
PM2.5 effect reduction by B vitamins	0.3	0.93	−5.9–6.4
24-h postexposure
Effect of PM2.5, compared with sham, without supplementation	−13.2	0.008	−22.3 to −4.1
Effect of PM2.5, compared with sham, with supplementation	0.8	0.78	−5.5–7.1
PM2.5 effect reduction by B vitamins	14.0	0.002	6.4–21.6

*Represents the percentage of change in mitochondrial DNA copy number associated with PM2.5 exposure.

†The original model was adjusted for chamber humidity, chamber temperature, and season (spring/summer/fall/winter). In the sensitivity analysis, we adjusted additionally for age, BMI, and race.

Effect of PM2.5 and B-Vitamin Supplementation on Mitochondrial DNA Content. 

CPO and NDUFS7are both involved in mitochondrial oxidative energy metabolism—a pivotal function with substantial impact on mitochondrial biogenesis and clearance (24, 25). In the secondary exploratory analysis, we further tested the associations of PM2.5 with mitochondrial DNA content, as well as the potential protective effect of B-vitamin supplementation. In the absence of B-vitamin supplementation, compared with sham, 2-h exposure to PM2.5 was estimated to be nonsignificantly associated with a −0.3% change [95% confidence interval (CI): −10.1%, 9.5%; P = 0.94] in mitochondrial DNA content. However, 24 h after exposure experiments, we observed substantial reduction in mitochondrial DNA content associated with PM2.5: 2-h exposure to PM2.5 significantly depleted mitochondrial DNA content by 11.1% (95% CI: 0.4%, 21.7%; P = 0.04) (Table S3). B-vitamin supplementation completely attenuated such effect of PM2.5 by 102% (Pinteraction = 0.01). With B-vitamin supplementation, 2-h exposure to PM2.5 was not associated with mitochondrial DNA content (0.2%; 95% CI: −8.3%, 8.8%; P = 0.96) (Table S3).

Exploratory Mediation Analysis and External Supporting Data. 

We further deconstructed total PM2.5effects on mitochondrial DNA content into direct and indirect (i.e., mediated) effects to investigate whether DNA methylation levels at loci cg06194186 and cg17157498 mediate PM2.5–mitochondrial DNA content relationship. Our result indicated that 16.0% (95% CI: 4.1%, 27.9%) and 18.4% (95% CI: 9.9%, 26.9%) of the PM effect on mitochondrial DNA content was mediated by cg06194186 and cg17157498 methylation, respectively. Consistent with our hypothesis, we observed correlation between mitochondrial DNA content and the methylation levels of cg06194186 (r = −0.45; P = 0.06) and cg17157498 (r = 0.63; P = 0.01) in CD4+ Th cells, in an independent external dataset.

Sensitivity Analysis. 

B vitamins have a long biological half-life (26), therefore requires a washout period longer than four months. We designed the trial without randomizing the treatment order (placebo vs. B vitamins) to avoid long washout periods, which would have made exposure experiments on the same volunteer less comparable. Lifestyle factors may vary over several months, particularly in relation to seasonality, which also may directly affect DNA methylation (8, 10). To rule out the potential impact of temporal trend on our results, we adjusted for date-since-entry, and this adjustment did not affect our conclusion. Furthermore, we conducted permutation test on the two top loci to ensure the robustness of our analysis (Ppermulation < 0.001). In the analysis involving the mitochondrial DNA content, we additionally adjusted for age (continuous), BMI (continuous), and race (categorical) to examine if our results are sensitive to covariates specification. Our results were robust and consistent (Table S3).

Go to:

DISCUSSION

This crossover intervention trial with controlled exposure experiments demonstrated that 2-h exposure to concentrated ambient PM2.5 (250 μg/m3) affects the dynamic epigenetic landscape in circulating CD4+ Th cells among healthy adults. We showed that these effects can be prevented with B-vitamin supplementation (i.e., folic acid and vitamins B6 and B12). Furthermore—as the top loci suggested potential modulation of mitochondrial metabolism—we followed up these findings by showing that exposure to PM2.5significantly altered mitochondrial DNA content in circulating CD4+ Th cells, and B-vitamin supplementation nearly completely prevented these effects.

Air pollution has been consistently associated with adverse health outcomes in epidemiological studies (1, 3, 4, 9). Although the biological mechanism underlying the health effects of PM2.5 remains not fully understood, systemic inflammation and oxidative stress have been proposed as essential biological pathways (4, 27). Furthermore, PM2.5 can disturb DNA methylation profiles (9, 10, 13), which might exacerbate oxidative and inflammatory responses following exposure. A previous human exposure study demonstrated that Toll-like receptor 4 (TLR4) gene hypomethylation in leukocytes mediates a part of PM effect on blood pressure (28). Recently, an epigenome-wide association study (EWAS) showed that low-concentration air pollution alters DNA methylation profiles in whole blood (9). However, the interpretation of those findings is limited, considering potential bias due to cell-type heterogeneity within whole blood, and by the correlational nature of observational studies (11, 12). In addition, these findings might not be generalizable to areas with frequent air pollution peaks. The present study—a cell type-specific EWAS using controlled exposure experiments—has the unique advantage of providing unbiased insight on the novel epigenetic underpinnings of the proinflammatory and prooxidative effects of PM2.5 exposure peaks.

In line with our hypothesis, we demonstrated acute effects of PM2.5 inhalation on DNA methylation in the promoter region of genes related to mitochondrial function and oxidative metabolism (24, 25): CPO, a member of the metalloenzyme family (25), is involved in metal ion binding, metallopeptidase, and metallocarboxypeptidase activities—which are essential in regulation of the steady-state concentration of O2− in the intermembrane space of mitochondria (29); NDUFS7 encodes one of the subunits of the mitochondrial respiratory chain complex I that transfers electrons from NADH to coenzyme Q, and NDUFS7 mutations were of etiological significance in mitochondrial complex I deficiency (30). Although mitochondria have their own genetic material distinct from the nuclear DNA, the majority of mitochondrial proteins are encoded by the nuclear genome (31). The observation that PM2.5 exposure substantially altered DNA methylation of nuclear genes in mitochondrial pathways indicate that mitochondria—the specialized organelles that regulate cellular-redox-balance and supplies energy—are a primary target of PM-induced cellular responses (32).

Our results on mitochondrial DNA content supplemented the findings from the epigenome-wide DNA methylation scan: exposure to PM2.5 for 2 h was followed by reduced mitochondrial DNA content 24-h postexposure. The cellular mitochondrial genomic content is stringently regulated by biogenesis/degradation machinery (33), which is vital in the determination of cell survival and function. Compensatory mitochondria biogenesis can buffer an intracellular reactive oxygen species (ROS) challenge, as an adaptive stress response to eliminate cellular oxidative damage (34). However, persistent oxidative stress may eventually overwhelm the adaptive response system and lead to mitochondrial DNA depletion via mitophagy (35). Our results support this hypothesis by demonstrating that exposure to high-concentration PM2.5 can reduce the mitochondrial DNA contents in circulating CD4+ Th cells. Consistent with our results, a recent study reported that a 10 µg/m3 increase in coarse PM (PM10–2.5) exposure during pregnancy was associated with a 16.1% decrease in placental mitochondrial DNA content (36). Our exploratory mediation analysis indicates that short-term exposure to high-concentration PM2.5 depletes mitochondrial DNA content, likely via—at least in part—modulating DNA methylation levels of genes in mitochondrial pathways. Future studies are warranted to investigate the potential for targeted epigenetic interventions.

DNA methylation is a modifiable biochemical process relying on methyl-group supplying nutrients such as B vitamins, which is postulated to increase DNA methylation levels (8, 15, 16). This feature renders B-vitamin supplementation an attractive pharmaceutical intervention to counteract the PM effects, which has been associated with loss of DNA methylation on inflammatory genes (28). Landmark experiments on the Agouti Avy mice and other models have shown that dietary methyl nutrients, added during gestation (15–18) or later even in adult life (20, 21), can be used to modulate DNA methylation status. In human studies, intake of a folic acid-depleted diet for several weeks promotes hypomethylation of lymphocyte genomic DNA among postmenopausal women, and this hypomethylation is reversible with folic acid replacement (21). Among patients with colorectal adenomatous polyps, folic acid supplementation led to a 31% increase in leukocyte DNA methylation and a 25% increase in DNA from the colonic mucosa (37). Potential for human translation is also demonstrated in animal models, as methyl group-supplying nutrients can be used to prevent the loss of DNA methylation induced by environmental pollutants in rodents (22). However, to the best of our knowledge, whether B vitamins can be used to limit adverse effects from PM pollution has not previously been tested in human. Our research provided the experimental evidence showing that the epigenetic effects of PM2.5 can be reduced using 4-wk B-vitamin supplementation. Remarkably, in our data, the B-vitamin supplementation not only prevented decreased DNA methylation but also increased DNA methylation following acute exposure to high-concentration PM. These findings suggest that B vitamins might protect against DNA hypomethylation as methyl group-supplying nutrients. On the other hand, B vitamins might also minimize DNA hypermethylation through interactions with regulatory pathways mediated by essential enzymes (such as DNA methyltransferases and methylenetetrahydrofolate reductase).

A major innovation of the present study over previous human epigenome-wide studies is the use of isolated CD4+ Th cells—an essential cell type modulating human immunity through both its own immune activities and regulation of other leukocytes’ proliferation, apoptosis, migration, and other functions via cytokine signaling (31). Therefore, the epigenetic effects of PM2.5 and protective effects of B vitamins observed in CD4+ Th genome might indicate subsequent modulation of essential cellular functions of other blood cell types. The Houseman cell proportion estimates indicated high purity of the analyzed samples. Although our study is subject to residual influence from differential CD4+ Th subsets, the observed effect of PM2.5or B vitamins on DNA methylation is unlikely to be surrogate for leukocyte composition variation. Our plating scheme for DNA methylation analysis is independent of the exposure and treatment status, and was designed to minimize potential bias due to technical variables. Thus, the measurement error of DNA methylation can be assumed to be nondifferential and, therefore, likely to bias the results toward null. We conducted sensitivity analysis to robustly eval‎uate the impact of PM2.5 on top loci, and further supported the EWAS results with a widely accepted mitochondrial marker—mitochondrial DNA content—with a highly reproducible quantitative real-time PCR method. Our crossover design controlled for time-invariant factors such as sex, race, BMI, SNP, etc. In addition, all exposure experiments were conducted at the same time of the day to eliminate any impact due to diurnal variation.

We acknowledge several limitations in the present pilot study. Although our EWAS is limited in power to meet the stringent Bonferroni threshold for significance with only 10 volunteers (30 exposure experiments), our sample size is comparable to previous controlled exposure studies that succeeded in demonstrating health effects of PM exposure (28, 38–40). As suggested by previous study (23), we selected the top loci based on both effect size and statistical significance, because those loci are more likely to infer biological significance. The short study duration was implemented to reduce the impact of with-in volunteer seasonal variations. Therefore, we could not randomize on the treatment (placebo vs. B vitamins) order due to long biological half-life of body stores of B vitamins (26), which might create potential confounding due to a temporal trend or learning effect (i.e., the volunteers might be more tolerant with the PM2.5 effects at the second PM2.5 exposure). In the sensitivity analysis, we adjusted for the amount of time passing because the study entry, and our results were consistent. Although residual confounding is possible, considering the magnitude of our effect estimates and the consistency of our findings, it is unlikely that the observed association reflected bias resulting from confounding. In addition, actual PM2.5concentration in PM2.5 experiments under placebo was nonsignificantly lower than PM2.5 experiments under B-vitamin supplementation, which might bias our results on B vitamins’ protective effects toward the null. Finally, future validation studies are warranted because our unique study design using CD4+ Th cells created major challenge to identify a suitable replication cohort, and our findings might not be generalizable to other cell types due to cell-type specificity of DNA methylation and mitochondrial DNA content.

The unclear molecular mechanistic underpinning of PM2.5’s health effects remains the major gap in current knowledge—thereby creating challenges in developing preventative strategies. The present study is a pilot intervention trial in the investigation of mechanistic pathways underlying the adverse health effects of air pollution, and potential targeted preventive approaches. We demonstrated that ambient PM2.5 exposure peak has unfavorable effects on epigenetic and pro-oxidative markers that can be neutralized by B-vitamin supplementation. Our findings suggest promising opportunities to aid the development of novel intervention strategies—which is particularly important for pathologies related to ubiquitous exposures such as PM2.5 pollution. Future trials with larger sample sizes are warranted to shed light on the precise pathophysiological processes of PM-induced inflammatory and oxidative responses, the mechanism underlying the protective effect of B vitamins, and potential clinical application.

Go to:

METHODS

Study Population. 

We recruited 10 healthy, 18- to 60-y-old, nonsmoking volunteers who were not taking any medicines or vitamin supplements, from the University of Toronto campus and surrounding area. The trial protocols were approved by all participating institutional review board (University of Toronto, St. Michael’s Hospital, and Harvard T.H. Chan School of Public Health) and registered (ClinicalTrials.govidentifier NCT01864824; date of registration: May 8, 2013). All methods were performed in accordance with the relevant guidelines and regulations. We obtained written informed consent from every volunteer before enrollment.

Study Design. 

We conducted a single-blind, crossover intervention trial with controlled exposure experiments to concentrated ambient PM2.5 (July 2013 to February 2014). The design (Fig. 1) started with a 2-wk run-in period with placebo, followed by the baseline sham experiment (2 h, particle-free medical air, exposure one). After sham experiment, each volunteer took placebo for 4 wk and was then exposed to PM2.5 (2 h, target concentration: 250 µg/m3, exposure two). Volunteers started the 4-wk B-vitamin supplementation after exposure two, and then were exposed again to PM2.5 (2 h, target concentration: 250 µg/m3, exposure three). All volunteers received three exposure experiments following the same order.

Exposure Facility. 

Ambient particles were drawn in from an inlet 1.5 m high, beside a busy (>1,000 vehicles per hour) street in downtown Toronto. We used the Harvard ambient particle concentrator to generate concentrated ambient PM2.5 (41), and delivered PM2.5 air stream to the volunteer seated inside a 4.9-m3 (1.1 × 1.9 × 2.0 m) lexan enclosure via an “oxygen type” facemask. The sham exposure with medical air were generated as previously described (38). During each exposure experiment, PM2.5 mass was collected on a 47-mm, 2-µm Teflon filter (Teflo, R2PJ047; Pall Corp.) and was monitored using the gravimetric determination of PM2.5 exposure mass concentration (micrograms per cubic meter).

Folic Acid, Vitamin B6, and Vitamin B12 Supplement. 

We administered one placebo or B-vitamin supplement (2.5 mg of folic acid, 50 mg of vitamin B6, and 1 mg of vitamin B12) daily to each volunteer. Previous human trials showed that these doses rapidly increased plasma B-vitamin levels, modified methyl-cycle metabolite levels, and ameliorated cardiovascular measurements (42, 43). Preparation, packaging, and coding of the placebo and supplement were done by an external laboratory (Jamieson Laboratories) and was blinded to the volunteers. Before each exposure experiment, we measured volunteers’ plasma folic acid and vitamin B12 levels using competitive-binding immunoenzymatic assay (A98032 and 33000; Beckman Coulter), and vitamin B12 levels using HPLC with fluorescence detection. At the first and the last visit, we assessed typical daily B-vitamin intake with a self-administered validated (44), semiquantitative food-frequency questionnaire used in the Nurses' Health Study.

CD4+ Th Cell Isolation and DNA Extraction. 

We collected blood samples via venous phlebotomy (preexposure, immediately postexposure, and 24 h postexposure), and within 4 h, isolated CD4+ Th cells by removing unwanted cells using RosetteSep Human CD4+ T Cell Enrichment Mixture (no. 15062; Stem Cell Technologies). DNA was then extracted using a Promega Maxwell 16 instrument with tissue DNA purification kit (Promega). We monitored the concentration and quality of extracted DNA using NanoDrop ND-1000 spectrophotometer (Nanodrop Technologies). Unsatisfactory DNA samples were discarded and DNA was extracted again. We estimated the proportions of major leukocyte types (CD4 T cells, CD8 T cells, B cells, granulocytes, monocytes, and natural killer cells) to assess the purity of isolated CD4+ Th cells using the Houseman method, a statistical deconvolution technique based on the 450K data (11).

Epigenome-Wide DNA Methylation Scan. 

We measured the epigenome-wide DNA methylation profile using the Infinium Human Methylation 450K BeadChip (Illumina) (45), which allows the assessment of approximately half a million CpG sites across 99% of RefSeq genes within the genome (46). Because of the within-volunteer, cross-over design, we plated all samples from one volunteer in one chip, with pre- and postexposure samples randomly loaded onto each column on the same row. All samples were processed by one technician and analyzed in one batch to minimize batch effect. Sample preparation and quality control details are explained in SI Methods.

Mitochondrial DNA Content in CD4+ Th Cells. 

We measured mitochondrial DNA content in CD4+ Th cells through the mtDNA/nDNA ratio, a widely used biomarker representing the mitochondrial DNA copy number versus the nuclear DNA copy number (34). Mitochondrial DNA copy number was analyzed pre, immediately after, and 24 h after each exposure experiment using multiplex quantitative real-time PCR, as previously reported (34). The mtDNA/nDNA is used in the statistical analysis—a ratio value of 1 indicates that the mtDNA/nDNA of the test sample is equal to the mtDNA/nDNA in the reference DNA pool used in the assay.

CD4+ Th Mitochondrial DNA Content and DNA Methylation in External Dataset. 

To strengthen our findings, we identified an external dataset based on 15 de-identified volunteers’ CD4+ T cells—which were purified from fresh blood samples through magnetic-activated cell sorting using anti-CD4 antibody coupled paramagnetic microbeads (Miltenyi Biotec). Epigenome-wide DNA methylation profiles and mitochondrial DNA content were measured using the same methods of the present study.

Statistical Methods. 

We used linear mixed-effects models (SI Methods) with random intercepts assigned to each volunteer to account for correlation among within-volunteer measurements. The crossover design minimized the influence from time-invariant factors. In all models, we adjusted for time-varying covariates with potential influences on DNA methylation, selected based on prior knowledge and the existing literature [i.e., season (fall/winter/spring/summer), chamber temperature, and chamber relative humidity (38)]. Rank-based normal transformation was performed on all DNA methylation measures to improve normality and stabilize the variance. We further performed permutation tests on the observed top two loci to ensure the robustness of our results, and conducted exploratory mediation analysis to eval‎uate whether DNA methylation mediates the effect of PM2.5 on mitochondrial DNA content (SI Methods). Analyses were performed using SAS 9.4 (SAS Institute) and R statistical computing software (R Foundation for Statistical Computing).

Data Availability. 

Data are available on request due to privacy or other restrictions.

Go to:

SI METHODS

Epigenome-Wide DNA Methylation Scan. 

We diluted purified DNA samples to a concentration of 30 ng/µL, with a total volume of 40 µL (total amount: 1,200 ng) and sent the deidentified samples to an external laboratory (Genomics Core Facility, Northwestern University, Chicago) for bisulfite conversion and hybridization on the Infinium Human Methylation 450K BeadChip (Illumina). The β values for each CpG were derived from the fluorescent intensity ratio [β = intensity of the methylated allele/(intensity of the unmethylated allele + intensity of the methylated allele + 100)].

Systematic quality assurance and quality control were performed, including assessment of assay repeatability and chip/plate effects using 10 across-chip technical replicates. We preprocessed the “*.idat” files using minfi package (47). We screened for signal outliers using multidimensional scaling analysis and no outlier was identified. We dropped the probes if (i) 20% of samples had detection P values >0.01 (n = 441); (ii) the SNP probes rather than CpG probes (n = 62); or (iii) probes located on X or Y chromosome (n= 11,341). After excluding these probes, as well as control probes, a total of 473,730 CpGs were analyzed. All samples had >95% sites with detection of P < 0.01. We checked concordance of sex and did not observe any sample mismatch. We minimized color channel bias and difference in Infinium chemistry by using dasen method in wateRmelon package to normalize (48), which equalized type I and type II backgrounds first, and then quantile-normalized type I and type II intensities separately.

Epigenome-Wide Association Statistical Analysis. 

We used linear mixed-effects models (Eq. S1) to account for within-volunteer correlation in methylation measures. Random intercepts were assigned to each volunteer.

Yij = β0 +  β1PMij + β2PMij ∗ Treatmentij + β3X1ij +  … + βpXpij + bi + εij.

[S1]

In the above model, Yij was the β value of the measured CpG probe for volunteer i at exposure occasion j, β0 was the overall intercept, and bi was the separate random intercept for volunteer i with, bi ∼ N(0, Θ), εij∼ N(0, σ2). X1ij–Xpij were the covariates, for volunteer i at measurement j. PMij (binary) indicated exposure to PM2.5 or medical air; Treatmentij (binary) indicated placebo or B-vitamin supplement. The main effect of B-vitamin supplementation was not included in the model, given volunteers did not receive any medical air exposure while on B-vitamin supplementation. β1 represented the effect of PM2.5 exposure without B-vitamin supplementation and β1 + β2 represented the effect of PM2.5 exposure with B-vitamin supplementation; β2 thus represented the intervention effect of B-vitamin supplementation (i.e., the attenuation of PM2.5 effect due to B-vitamin supplementation). The linear mixed-effects model was implemented using the lme4 (49) and lmerTest (50) packages.

Permutation Test. 

We conducted 1,000 permutation tests on the observed top two loci, which randomly paired the exposure and treatment status with methylation data within each volunteer. We calculated the permutation P value based on the observed t statistic (t.actual) and the distribution of the t statistics obtained from 1,000 permutation simulations (t.nulls) (Ppermutation < 0.001 for both top loci).

Exploratory Mediation Analysis. 

To deconstruct total PM2.5 effects on mitochondrial DNA content into direct and indirect (i.e., mediated) effects, we simultaneously fitted two linear mixed-effects models (Eqs. S2 and S3) with random intercepts assigned to each individual, using a previously described method (51).

Mij  =  β0  + ui  + β1PMij + β2PMij ∗ Treatmentij + β3X1ij  + … + βpXpij  +  εij, 

[S2]

Yij  = γ0  + g0i + γ1PMij + γ2PMij ∗ Treatmentij + γ3X1ij + … + γpXpij  + γmMij  + ηij.

[S3]

β0 and γ0 represented the intercept for the population mean; ui and g0i represented the subject-specific random intercept. Mij was the mediator (cg06194186 and cg17157498 methylation) and Yij was mitochondrial DNA content measurement. X1ij–Xpij were the covariates, for volunteer i at measurement j. εij and ηij were the within-subject error terms. γ1 represented the natural direct effect of PM2.5, and the product of β1*γm was the natural indirect effect (i.e., mediated effect). We used the delta method (52) to calculate the variance (and 95% CI) of the natural indirect effect, and estimated proportion mediated as the percentage of natural indirect effect over the sum of natural direct and natural indirect effect.

Go to:

ACKNOWLEDGMENTS

This study was supported by NIH Grants R21ES021895, R01ES015172, R01ES021733, R01ES020836, R01ES021357, P30ES009089, and P30ES000002; US Environmental Protection Agency funding (Grant RD-83479801); Environment Canada; AllerGen Networks of Centres of Excellence; and National Natural Science Foundation of China Grant 91643202.

Go to:

FOOTNOTES

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618545114/-/DCSupplemental.

Go to:

REFERENCES

1. Bell ML, Davis DL. Reassessment of the lethal London fog of 1952: Novel indicators of acute and chronic consequences of acute exposure to air pollution. Environ Health Perspect. 2001;109(Suppl 3):389–394. [PMC free article] [PubMed] [Google Scholar]

2. World Health Organization  (September 27, 2016) WHO releases country estimates on air pollution exposure and health impact. Press release. Available at  www.who.int/mediacentre/news/releases/2016/air-pollution-estimates/en/. Accessed February 24, 2017.

3. Lelieveld J, Evans JS, Fnais M, Giannadaki D, Pozzer A. The contribution of outdoor air pollution sources to premature mortality on a global scale. Nature. 2015;525(7569):367–371. [PubMed] [Google Scholar]

4. Brook RD, et al. American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association. Circulation. 2010;121(21):2331–2378. [PubMed] [Google Scholar]

5. World Health Organization . Air Quality Guidelines: Global Update 2005: Particulate Matter, Ozone, Nitrogen Dioxide, and Sulfur Dioxide. World Health Organization; Copenhagen: 2006.  [Google Scholar]

6. Richmond-Bryant J, Saganich C, Bukiewicz L, Kalin R. Associations of PM2.5 and black carbon concentrations with traffic, idling, background pollution, and meteorology during school dismissals. Sci Total Environ. 2009;407(10):3357–3364. [PubMed] [Google Scholar]

7. Bender A, et al.  Understanding Episodes of High Airborne Particulate Matter in Iowa. Bistate Regional Commission in Eastern Iowa; Iowa City, IA: 2009.  [Google Scholar]

8. Ramchandani S, Bhattacharya SK, Cervoni N, Szyf M. DNA methylation is a reversible biological signal. Proc Natl Acad Sci USA. 1999;96(11):6107–6112. [PMC free article] [PubMed] [Google Scholar]

9. Panni T, et al. A genome-wide analysis of DNA methylation and fine particulate matter air pollution in three study populations: KORA F3, KORA F4, and the Normative Aging Study. Environ Health Perspect. 2016;124(7):983–90. [PMC free article] [PubMed] [Google Scholar]

10. Baccarelli A, et al. Rapid DNA methylation changes after exposure to traffic particles. Am J Respir Crit Care Med. 2009;179(7):572–578. [PMC free article] [PubMed] [Google Scholar]

11. Houseman EA, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86. [PMC free article] [PubMed] [Google Scholar]

12. Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nat Rev Genet. 2011;12(8):529–541. [PMC free article] [PubMed] [Google Scholar]

13. Liu J, et al. Combined inhaled diesel exhaust particles and allergen exposure alter methylation of T helper genes and IgE production in vivo. Toxicol Sci. 2008;102(1):76–81. [PMC free article] [PubMed] [Google Scholar]

14. Bruniquel D, Schwartz RH. Selective, stable demethylation of the interleukin-2 gene enhances transcription by an active process. Nat Immunol. 2003;4(3):235–240. [PubMed] [Google Scholar]

15. Cropley JE, Suter CM, Martin DI. Methyl donors change the germline epigenetic state of the A(vy) allele. FASEB J. 2007;21(12):3021, author reply 3021–3022. [PubMed] [Google Scholar]

16. Dolinoy DC, Weidman JR, Waterland RA, Jirtle RL. Maternal genistein alters coat color and protects Avy mouse offspring from obesity by modifying the fetal epigenome. Environ Health Perspect. 2006;114(4):567–572. [PMC free article] [PubMed] [Google Scholar]

17. Sinclair KD, et al. DNA methylation, insulin resistance, and blood pressure in offspring determined by maternal periconceptional B vitamin and methionine status. Proc Natl Acad Sci USA. 2007;104(49):19351–19356. [PMC free article] [PubMed] [Google Scholar]

18. Waterland RA, et al. Maternal methyl supplements increase offspring DNA methylation at Axin Fused. Genesis. 2006;44(9):401–406. [PubMed] [Google Scholar]

19. Weaver IC, et al. Reversal of maternal programming of stress responses in adult offspring through methyl supplementation: Altering epigenetic marking later in life. J Neurosci. 2005;25(47):11045–11054.[PMC free article] [PubMed] [Google Scholar]

20. Kotsopoulos J, Sohn KJ, Kim YI. Postweaning dietary folate deficiency provided through childhood to puberty permanently increases genomic DNA methylation in adult rat liver. J Nutr. 2008;138(4):703–709.[PubMed] [Google Scholar]

21. Jacob RA, et al. Moderate folate depletion increases plasma homocysteine and decreases lymphocyte DNA methylation in postmenopausal women. J Nutr. 1998;128(7):1204–1212. [PubMed] [Google Scholar]

22. Dolinoy DC, Huang D, Jirtle RL. Maternal nutrient supplementation counteracts bisphenol A-induced DNA hypomethylation in early development. Proc Natl Acad Sci USA. 2007;104(32):13056–13061.[PMC free article] [PubMed] [Google Scholar]

23. Maccani JZ, et al. Placental DNA methylation related to both infant toenail mercury and adverse neurobehavioral outcomes. Environ Health Perspect. 2015;123(7):723–729. [PMC free article] [PubMed] [Google Scholar]

24. Hyslop SJ, Duncan AM, Pitkänen S, Robinson BH. Assignment of the PSST subunit gene of human mitochondrial complex I to chromosome 19p13. Genomics. 1996;37(3):375–380. [PubMed] [Google Scholar]

25. Wei S, et al. Identification and characterization of three members of the human metallocarboxypeptidase gene family. J Biol Chem. 2002;277(17):14954–14964. [PubMed] [Google Scholar]

26. Butler CC, et al. Oral vitamin B12 versus intramuscular vitamin B12 for vitamin B12 deficiency: A systematic review of randomized controlled trials. Fam Pract. 2006;23(3):279–285. [PubMed] [Google Scholar]

27. Ghio AJ, Carraway MS, Madden MC. Composition of air pollution particles and oxidative stress in cells, tissues, and living systems. J Toxicol Environ Health B Crit Rev. 2012;15(1):1–21. [PubMed] [Google Scholar]

28. Bellavia A, et al. DNA hypomethylation, ambient particulate matter, and increased blood pressure: Findings from controlled human exposure experiments. J Am Heart Assoc. 2013;2(3):e000212.[PMC free article] [PubMed] [Google Scholar]

29. Turrens JF. Mitochondrial formation of reactive oxygen species. J Physiol. 2003;552(Pt 2):335–344.[PMC free article] [PubMed] [Google Scholar]

30. Smeitink J, van den Heuvel L. Human mitochondrial complex I in health and disease. Am J Hum Genet. 1999;64(6):1505–1510. [PMC free article] [PubMed] [Google Scholar]

31. Cooper G.  The Cell: A Molecular Approach. 2nd Ed Sinauer Associates; Sunderland, MA: 2000. [Google Scholar]

32. Lee HC, Wei YH. Mitochondrial biogenesis and mitochondrial DNA maintenance of mammalian cells under oxidative stress. Int J Biochem Cell Biol. 2005;37(4):822–834. [PubMed] [Google Scholar]

33. Attardi G, Schatz G. Biogenesis of mitochondria. Annu Rev Cell Biol. 1988;4:289–333. [PubMed] [Google Scholar]

34. Zhong J, et al. Traffic-related air pollution, blood pressure, and adaptive response of mitochondrial abundance. Circulation. 2015;133(4):378–87. [PMC free article] [PubMed] [Google Scholar]

35. Kim I, Rodriguez-Enriquez S, Lemasters JJ. Selective degradation of mitochondria by mitophagy. Arch Biochem Biophys. 2007;462(2):245–253. [PMC free article] [PubMed] [Google Scholar]

36. Janssen BG, et al. Placental mitochondrial DNA content and particulate air pollution during in utero life. Environ Health Perspect. 2012;120(9):1346–1352. [PMC free article] [PubMed] [Google Scholar]

37. Pufulete M, et al. Effect of folic acid supplementation on genomic DNA methylation in patients with colorectal adenoma. Gut. 2005;54(5):648–653. [PMC free article] [PubMed] [Google Scholar]

38. Behbod B, et al. Endotoxin in concentrated coarse and fine ambient particles induces acute systemic inflammation in controlled human exposures. Occup Environ Med. 2013;70(11):761–767.[PMC free article] [PubMed] [Google Scholar]

39. Brook RD, et al. Inhalation of fine particulate air pollution and ozone causes acute arterial vasoconstriction in healthy adults. Circulation. 2002;105(13):1534–1536. [PubMed] [Google Scholar]

40. Urch B, et al. Relative contributions of PM2.5 chemical constituents to acute arterial vasoconstriction in humans. Inhal Toxicol. 2004;16(6-7):345–352. [PubMed] [Google Scholar]

41. Demokritou P, Gupta T, Ferguson S, Koutrakis P. Development of a high-volume concentrated ambient particles system (CAPS) for human and animal inhalation toxicological studies. Inhal Toxicol. 2003;15(2):111–129. [PubMed] [Google Scholar]

42. Lonn E, et al. Heart Outcomes Prevention Eval‎uation (HOPE) 2 Investigators Homocysteine lowering with folic acid and B vitamins in vascular disease. N Engl J Med. 2006;354(15):1567–1577. [PubMed] [Google Scholar]

43. Ullegaddi R, Powers HJ, Gariballa SE. B-group vitamin supplementation mitigates oxidative damage after acute ischaemic stroke. Clin Sci (Lond) 2004;107(5):477–484. [PubMed] [Google Scholar]

44. Willett WC, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122(1):51–65. [PubMed] [Google Scholar]

45. Lowe R, Morris T. Report on the 2nd annual Infinium Humanmethylation450 array workshop: 15 April 2013 QMUL, London, UK. Epigenetics. 2013;8(10):1123–1124. [PMC free article] [PubMed] [Google Scholar]

46. Sandoval J, et al. Validation of a DNA methylation microarray for 450,000 CpG sites in the human genome. Epigenetics. 2011;6(6):692–702. [PubMed] [Google Scholar]

47. Aryee MJ, et al. Minfi: A flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363–1369. [PMC free article][PubMed] [Google Scholar]

48. Pidsley R, et al. A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics. 2013;14(1):293. [PMC free article] [PubMed] [Google Scholar]

49. Bates D, Maechler M, Bolker B. lme4: Linear mixed-effects models using S4 classes. J Stat Software. 2012;67(1) [Google Scholar]

50. Kuznetsova A, Brockhoff PB, Christensen RHB. 2013 lmerTest, R Package Version 2(6). Available at https://cran.r-project.org/web/packages/lmerTest/lmerTest.pdf. Accessed February 24, 2017.

51. Peng C, et al. Particulate air pollution and fasting blood glucose in nondiabetic individuals: Associations and epigenetic mediation in the Normative Aging Study, 2000–2011. Environ Health Perspect. 2016;124(11):1715–1721. [PMC free article] [PubMed] [Google Scholar]

52. Bind M-AC, Vanderweele TJ, Coull BA, Schwartz JD. Causal mediation analysis for longitudinal data with exogenous exposure. Biostatistics. 2016;17(1):122–134. [PMC free article] [PubMed] [Google Scholar]