Inducible nitric oxide synthase gene methylation and parkinsonism in manganese-exposed welders

Abstract

Introduction

Neurologist-assessed parkinsonism signs are prevalent among workers exposed to manganese (Mn)-containing welding fume. Neuroinflammation may possibly play a role. Inducible nitric oxide synthase, coded by NOS2, is involved in inflammation, and particulate exposure increases the gene’s expression through methylation of CpG sites in the 5′ region.

Methods

We assessed DNA methylation at three CpG sites in the NOS2 exon 1 from blood from 201 welders. All were non-Hispanic Caucasian men 25–65 years old who were examined by a neurologist specializing in movement disorders. We categorized the workers according to their Unified Parkinson Disease Rating Scale motor subsection 3 (UPDRS3) scores as parkinsonism cases (UPDRS3 ≥ 15; n = 49), controls (UPDRS3 < 6; n = 103), or intermediate (UPDRS3 ≥6 to <15; n = 49).

Results

While accounting for age, examiner and experimental plate, parkinsonism cases had lower mean NOS2 methylation than controls (p-value for trend = 0.04), specifically at CpG site 8329 located in an exonic splicing enhancer of NOS2 (p-value for trend = 0.07). These associations were not observed for the intermediate UPDRS3 group (both p-value for trend ≥ 0.59).

Conclusions

Inflammation mediated by inducible nitric oxide synthase may possibly contribute to the association between welding fume and parkinsonism, but requires verification in a longitudinal study.

1. Introduction

Historically, occupational exposure to manganese (Mn) has been associated with parkinsonism among highly exposed workers, such as Mn miners and ore crushers. Modern occupational Mn exposures are substantially lower, but welders are frequently exposed to Mn above current occupational standards [1]. While records-based studies provide little evidence that contemporary welders have an increased risk of Parkinson disease [2], studies that investigate parkinsonism more generally and employ neurologists in the assessment of this broader outcome demonstrate that signs of parkinsonism are prevalent among workers exposed to welding fume [3, 4] and other occupational sources of Mn [5]. Among welders, characteristic signs of this movement disorder are bradykinesia and rigidity [3].

Welders demonstrate increased signal on T1 weighted MRI in the basal ganglia [6], a finding generally believed to indicate Mn accumulation. Basal ganglia dysfunction likely contributes to the association between parkinsonism and Mn-containing welding fume [6–8]. Although the exact pathophysiologic mechanism of Mn neurotoxicity has not been fully elucidated, inflammation may play a role. In contemporary Mn mine workers, Mn exposure is associated with microglia cell density in the globus pallidus [8], the portion of the basal ganglia traditionally considered a target of Mn neurotoxicity. Glial cell activation enhances the uptake of Mn into neurons [9], which in turn may contribute to neuronal death [10]. In addition, activation of glial cells, particularly microglia, might damage adjacent neurons through increased expression of pro-inflammatory mediators [11, 12]. These include nitric oxide, which is produced by nitric oxide synthase [13]. There are three such enzymes in humans, including inducible nitric oxide synthase (iNOS). Inhaled welding fumes, at concentrations consistent with potential workplace exposures, increase levels of messenger RNA for microglial markers and iNOS in the striatum and midbrain of rats [14]. Moreover, mice without the gene coding for iNOS are less susceptible to the neurotoxic effects of Mn [15].

The gene coding for iNOS in humans, NOS2, is primarily regulated at the transcriptional level, at least in part via methylation of CpG dinucleotides [16]. Specifically, hypermethylation of CpG sites in the 5′ promoter region of the gene decreases its expression [16, 17] and is associated with lower breath nitric oxide [18] suggesting lower iNOS activity. Conversely, hypomethylation of NOS2 may be associated with greater iNOS activity. Most studies indicate that exposure to fine and coarse particulate, including metal-rich particulate, is associated with lower NOS2 methylation in or near the gene’s promoter region [19–23]. Therefore, in a sample of workers from a well-characterized cohort of welders with a relatively high prevalence of parkinsonism [3], we assessed the association between NOS2 methylation and parkinsonism. Our hypothesis was that compared to welders with normal neurological exams, parkinsonian welders would have lower NOS2 methylation (Figure 1).

Figure 1. Hypothesized mechanism for association between welding and parkinsonism.

Abbreviations: iNOS, inducible nitric oxide synthase; NOS2, gene coding for iNOS

2. Materials and Methods

2.1 Participants and assessment of parkinsonism

Prior to study conduct, we obtained Human Subjects approval from Washington University (St. Louis, MO) and the University of Washington (Seattle, WA), and written informed consent from each participant. All participants were part of an on-going study in the U.S. Midwest examining the association between welding and parkinsonism [3]. Recruitment has been detailed previously [24]. Briefly, we used a union membership list to contact employees and retirees from three welding worksites: two shipyards and one heavy equipment fabrication shop. One of two neurologists specializing in movement disorders (B.A.R. and S.R.C.) examined each participant using a standardized neurological exam that included the Unified Parkinson Disease Rating Scale motor subsection 3 (UPDRS3) [3, 25]. At the time of the exam, we asked each participant to complete a structured questionnaire [24] on demographics and work history, and to provide a blood sample. We stored all blood samples at −80°C.

At the time of laboratory analysis, 437 whole blood specimens were potentially available for the present work. We selected a subsample based on demographic characteristics to minimize the potential for confounding. Specifically, most participants in the cohort were non-Hispanic Caucasian men [3], and therefore we focused on this demographic group to avoid confounding by race, ethnicity and sex. In addition, as expected parkinsonism was very strongly associated with age, so we restricted our study to workers age 25–65 to minimize the potential for confounding by age. This also had the benefit of largely restricting to active workers exposed for a sufficient period of time to have developed parkinsonism. Because this was a pilot study, we then applied to the cohort a design similar to a nested case-control study: We excluded specimens obtained at a repeat exam and selected workers based on UPDRS3 category. In total we included 201 workers from one of three UPDRS3 groups: UPDRS3 score < 6 (n = 103, hereafter controls), UPDRS3 score > 8 to ≤ 12 (n = 49, hereafter intermediate UPDRS3 group), and UPDRS3 ≥ 15 (n = 49, hereafter parkinsonism cases). These categories parallel our previous classification [3] except that we were unable to include all workers in the intermediate category (UPDRS3 ≥ 6 to UPDRS3 <15) so we only retained workers with UPDRS3 scores most clearly distinct from those of both cases and controls.

2.2 Assessment of NOS2 methylation

We (P.L.S., F.M.F.) collected DNA from whole blood using the QIAamp DNA blood kit (Qiagen, Germantown, MD). We then bisulfite-treated 500ng of purified DNA using Qiagen’s EpiTect Fast DNA Bisulfite Kit and diluted to a concentration of 10ng/μl. We assessed NOS2 methylation at the three CpG sites as in a recent study [26] of apprentice welders: sites 8309 (CpG site 1), 8314 (CpG site 2) and 8329 (CpG site 3). These CpG sites are in exon 1 (Genbank: AF017634), bordering the 5′ promoter region and immediately adjacent to a transcription factor binding site and overlapping an exonic splicing enhancer.

We designed the assay using Qiagen’s Assay Design Software, and obtained primers from Eurofins MWG Operon (Huntsville, AL). The sequences were gggtgagtataaatattttttggttgttag (forward primer), biotin-taaaactacccaatcccctcat (reverse primer), and tggttgttagtgtgtttata (sequencing primer). Each 25μl PCR reaction consisted of 12.5μl 2× Pyromark PCR Master Mix (Qiagen), 5 pmol forward primer, 5 pmol reverse primer, 15 – 20ng of bisulfite-treated DNA and water. Thermocycling conditions were 15 min at 95°C followed by 40 cycles of 30 seconds at 94°C, 30 seconds at 56°C, and 30 seconds at 72°C, with a final extension of 10 minutes at 72°C. After visual determination of a single band on an agarose gel, 8μl of the PCR product was used in a Qiagen Q24 Pyrosequencing Assay according to the manufacturer’s protocol. DNA methylation status was determined using a Pyromark Q24 instrument and final results were analyzed using Pyromark Q24 software. We included duplicate sample(s) and positive and negative laboratory controls on each experimental plate. The inter-assay coefficient of variation for mean NOS2 methylation was 2.4% and the intra-assay coefficient of variation was 0.98%.

We ensured that all three UPDRS3 groups were represented on each experimental plate, while maintaining blinding of the lab. Complete methylation data were available for all participants; most had complete data in the initial attempt, but a repeat attempt was required for 4 cases, 5 controls, and 2 in the intermediate UPDRS3 group.

2.3 Statistical analysis

All statistical analyses were conducted in Stata Version 11 (College Station, TX). We constructed logistic regression models to separately compare two groups – parkinsonism cases and the intermediate UPDRS3 group – to controls with regard to percent NOS2 methylation. In our primary analysis, we defined percent NOS2 methylation as mean percent methylation across the three CpG sites [26]. In a secondary analysis, we considered percent methylation at individual CpG sites [20], as these sites have been observed to be differentially methylated in the brain despite their close proximity [27]. We modeled NOS2 methylation linearly, but to verify the appropriateness of this approach, we also constructed tertiles of NOS2 methylation. We calculated p-values for trend based on the former.

We adjusted a priori for age because of its association with both NOS2 methylation [28] and UPDRS3 score [3]. We also adjusted a priori for experimental plate for pyrosequencing reactions [20, 22], specifically conditional logistic regression while grouping on experimental plate. We examined whether inclusion of any of the following factors in the model further altered odds ratios (ORs) or 95% confidence intervals (CIs) by > 10%: age squared; examiner; current and former tobacco smoking; consumption of caffeine, coffee and alcohol; body mass index and timing of blood collection. This included year, day of the week, season and heating degree days in the study region on the day blood was collected. We accordingly adjusted all models for examiner in addition to age and experimental plate. Other considered factors, including smoking, did not alter results and therefore were not included in models. We also did not adjust for welding exposure because our hypothesis was that DNA methylation is in the causal pathway between welding fume exposure and parkinsonism (Figure 1), and therefore is inappropriate to include in the model as a potential confounder. We repeated all analyses while excluding workers who had been retired for > 1 year because NOS2 methylation appears to change relatively rapidly [26, 29, 30], and the presence of parkinsonism could affect a worker’s ability to work in a welding related job. Results from the larger cohort study [3] were consistent with such a potential healthy worker effect. In addition, we verified results were similar when we excluded all participants with co-morbidities associated with vascular parkinsonism (stroke, hypertension, heart disease and/or diabetes). Specifically, participants with a history of stroke were excluded from the original cohort [3], and then we conducted sensitivity analyses here in which we excluded participants with any of the other three conditions.

We also explored whether, as hypothesized (Figure 1), welding fume exposure was associated with NOS2 methylation. Using the workers’ self-reported work histories, we calculated cumulative duration (years) of occupational exposure to welding fume. This exposure metric is the one most strongly associated with parkinsonism among this cohort [3] and thus was our primary exposure of interest. We also considered the current job classification (welder, welder helper, around welding, not around welding) and retirement status to explore the influence of intensity and recentness of exposure on NOS2 methylation.

3. Results

3.1 Participants

On average, participants with DNA methylation data had worked nearly 20 years in a job with exposure to welding fume (mean 19.8 years, standard deviation 13.8 years). Parkinsonism cases were older than controls and were less likely than participants in the intermediate UPDRS3 group to be a welder at the current or most recent job (Table 1). Proportionally more controls had ever smoked tobacco regularly compared to the other UPDRS3 groups.

Table 1.

Characteristics of non-Hispanic men with DNA methylation data, by Unified Parkinson Disease Rating Scale motor subsection 3 (UPDRS3), Midwestern U.S. Welders Cohort, 2006–2013

	Parkinsonism	Intermediate	Control
	(UPDRS3 ≥ 15)	(UPDRS3 > 8 to ≤ 12)	(UPDRS3 < 6)
	n = 49	n = 49	n = 103
	n (%)	n (%)	n (%)
UPDRS3, mean (SD)	20.3 (3.6)	10.1 (1.1)	3.0 (1.8)
Age, years, mean (SD)	53.7 (10.0)	49.6 (10.4)	47.7 (11.4)
Welding Worksite
Worksite 1	25 (51)	23 (47)	36 (35)
Worksite 2	16 (33)	20 (41)	53 (51)
Worksite 3	8 (16)	6 (12)	14 (14)
Total duration welding exposure,	21.7 (13.5)	21.6 (12.6)	18.0 (14.4)
years, mean (SD)
Weldera	14 (29)	25 (51)	44 (43)
Retiree (> 1 year)	6 (12)	3 (6)	7 (7)
Ever smoked tobacco	30 (61)	28 (57)	78 (76)
Currently smoke tobacco	19 (39)	16 (33)	41 (40)
Typically drink coffee daily	33 (67)	36 (73)	75 (73)
> 12 years of educationb	12 (24)	11 (23)	20 (20)
First degree relative with	1 (2)	2 (4)	0 (0)
Parkinson diseaseb

SD, standard deviation.

^aWelder at current/most recent job vs. welder helper or other worker at welding worksite.

^bPercentages based on those with data (excludes 2 with missing education data, or 3 with missing data on family history of Parkinson disease).

3.2 NOS2 methylation and parkinsonism

The NOS2 CpG sites we assessed were highly methylated; all controls had > 90% methylation at each of the three CpG sites, and the overall mean NOS2 methylation was 95.5% (standard deviation 1.3%). Methylation appeared similar in cases (mean 95.4%, standard deviation 1.5%) as compared to the other groups (each with mean 95.5%, standard deviation 1.2–1.3%). However, after accounting for age, examiner and especially experimental plate, there was a clear inverse association between NOS2 methylation and parkinsonism (p-value for trend = 0.04, Table 2). Compared to the lowest tertile of NOS2 methylation, the prevalence of parkinsonism was 59% lower in the middle tertile of NOS2 methylation and 76% lower in the upper tertile of NOS2 methylation. A weaker inverse association was suggested when comparing the intermediate UPDRS3 group to controls, but there was no evidence of a dose-response association (p-value for trend = 0.59).

Table 2.

Parkinsonism and inducible nitric oxide synthase (NOS2) methylation,^a Midwestern U.S. Welders Cohort, 2006–2013

	All workers		Excluding retirees
	Intermediate	Parkinsonism	Intermediate	Parkinsonism
	(UPDRS3 > 8 to ≤ 12)	(UPDRS3 ≥ 15)	(UPDRS3 > 8 to ≤ 12)	(UPDRS3 ≥ 15)
	OR (95% CI)b	OR (95% CI)b	OR (95% CI)b	OR (95% CI)b
	n = 49	n = 49	n = 46	n = 43
Mean NOS2 methylationa
Methylation tertile 1	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
Methylation tertile 2	0.59 (0.22–1.54)	0.41 (0.13–1.29)	0.56 (0.21–1.50)	0.32 (0.09–1.09)
Methylation tertile 3	0.61 (0.23–1.61)	0.24 (0.07–0.83)	0.53 (0.19–1.51)	0.13 (0.03–0.56)
p-value for trendc	p = 0.59	p = 0.04	p = 0.50	p = 0.02
NOS2 CpG Site 1a
Methylation tertile 1	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
Methylation tertile 2	1.91 (0.74–4.93)	1.10 (0.34–3.53)	1.90 (0.72–5.06)	0.70 (0.19–2.61)
Methylation tertile 3	0.84 (0.28–2.54)	0.72 (0.20–2.58)	0.86 (0.27–2.73)	0.67 (0.16–2.72)
p-value for trendc	p = 0.92	p = 0.11	p = 0.88	p = 0.10
NOS2 CpG Site 2a
Methylation tertile 1	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
Methylation tertile 2	0.75 (0.28–2.02)	0.46 (0.15–1.41)	0.70 (0.25–1.90)	0.43 (0.14–1.37)
Methylation tertile 3	0.77 (0.30–1.96)	0.83 (0.29–2.37)	0.68 (0.25–1.85)	0.77 (0.24–2.53)
p-value for trendc	p = 0.53	p = 0.32	p = 0.42	p = 0.29
NOS2 CpG Site 3a
Methylation tertile 1	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
Methylation tertile 2	0.72 (0.28–1.85)	0.53 (0.17–1.64)	0.67 (0.25–1. 80)	0.34 (0.10–1.19)
Methylation tertile 3	0.78 (0.28–2.17)	0.40 (0.12–1.34)	0.72 (0.24–2.09)	0.22 (0.06–0.87)
p-value for trendc	p = 0.81	p = 0.07	p = 0.77	p = 0.01

^aMean and individual percent methylation at three CpG sites in NOS2 exon 1: CpG site 1 (8309), CpG site 2 (8314), CpG site 3 (8329).

^bOdds ratio and 95% confidence interval, relative to workers with UPDRS3 < 6 (N = 103 when including all workers, N = 96 when excluding retirees), adjusted for age (continuous), examiner and experimental plate.

^cBased on NOS2 methylation as a continuous measure.

The association between parkinsonism and NOS2 methylation was most evident for the third CpG site (8329). This association was not attenuated when we excluded retirees (Table 2, p-value for trend = 0.01), or conducted a variety of sensitivity analyses (exclusion of participants with co-morbidities associated with vascular parkinsonism, exclusion of samples for which the assay failed initially, application of the smooth function to the continuous methylation data, or adjustment for methylation at the other CpG sites in NOS2, data not shown in tables). Among the non-retired workers, each absolute increase of 1% of the CpG site 8329 was associated with a 33% (95% CI 8% to 51%) lower prevalence of parkinsonism.

3.3 Welding exposure and NOS2 methylation

Because results for the third CpG site (8329) were particularly robust, and methylation of this CpG site is perhaps the most plausibly related to NOS2 expression given its location in the gene, we explored whether welding exposure was associated with methylation at this CpG site. In linear regression models accounting for age and experimental plate, workers with recent exposure to welding fume had somewhat lower NOS2 8329 methylation than workers who had been retired for > 1 year (0.72% lower absolute methylation) or who were still at the worksite but not around welding fume in their current job (0.51% lower absolute methylation), although confidence intervals were wide because most workers remained exposed (Table 3). Locally-weighted scatterplot smoothing revealed an inverse association between total duration of welding fume exposure and NOS2 8329 methylation among workers with < 10 years of cumulative exposure, but no association otherwise. Accordingly, likelihood ratio test (p-value = 0.02) indicated that we must include a spline at 10 years duration (in addition to a linear term for duration) to adequately capture the association between duration of welding fume exposure and NOS2 8329 methylation. Therefore we stratified by duration of exposure, and observed 0.16% (95% CI 0.02% to 0.29%) lower methylation of NOS2 8329 per year of exposure among workers with < 10 years of exposure (p-value for trend = 0.03) and no association among workers who had already experienced ≥ 10 years of exposure (0.004%, 95% CI −0.03% to 0.04%, p-value for trend = 0.81).

Table 3.

Welding occupations and methylation of NOS2 CpG Site 8329, Midwestern U.S. Welders Cohort, 2006–2013

	n = 201 N	Mean % methylation	β (95% CI)a	β (95% CI)a
Classification of current/most recent job
Not around welding	15	95.6	Reference	Reference
Around welding	95	95.7	−0.48 (−1.44, 0.47)
Welder helper	8	95.5	−0.35 (−1.83, 1.14)	−0.51 (−1.43, 0.41)
Welder	83	95.5	−0.57 (−1.55, 0.41)
Currently working at
welding worksite
No (retiree > 1 year)	16	96.4	Reference
Yes	185	95.5	−0.72 (−1.66, 0.23)	–

^aLinear regression beta coefficient and 95% confidence interval, adjusted for age (continuous) and experimental plate

4. Discussion

In workers from welding worksites subject to contemporary guidelines for workplace exposures, we observed greater UPDRS3 scores in relation to lower NOS2 methylation. Interestingly, when we focused on participants who were employed at the welding worksite at the time of blood collection, this association was particularly evident for the third CpG site (8329), which is located in an exonic splicing enhancer of NOS2. This short motif is involved in the splicing of hetero-nuclear RNA or pre-mRNA into messenger RNA. It is also adjacent to the 5′ promoter region. Because lower NOS2 methylation in the promoter is associated with greater NOS2 expression [16, 17] and perhaps greater iNOS activity [18], these results are consistent with the hypothesis that parkinsonism associated with exposure to welding fume could be mediated, in part, through the nitric oxide pathway (Figure 1). Moreover, our results suggest that at least in the initial years of exposure, methylation at the NOS2 8329 CpG site continually drops with each additional year of on-the-job welding fume exposure. Although this contrasts with results in 38 welding school apprentices [26], our study adds to the mounting evidence [19–23] that particulate exposure reduces promoter region NOS2 methylation in humans. This would be expected if NOS2 methylation contributes to the occurrence of parkinsonism in workers exposed to welding fume (Figure 1). Thus, insofar as our results are not due to chance, inflammation, and iNOS in particular, may play a role in the high prevalence of parkinsonism among welders. As detailed above, experimental and animal studies support the hypothesis that iNOS plays a role in Mn neurotoxicity [11, 12] [14, 15].

In the present study we were not able to assess NOS2 methylation in brain tissue, nor did we assess NOS2 expression, iNOS activity or nitric oxide levels. However, we hypothesize that these are correlated with NOS2 methylation in blood (Figure 1). Blood DNA methylation may be a good surrogate for DNA methylation in the brain [27], and methylation of NOS2 is associated with NOS2 expression [16, 17]. Therefore, it is intriguing that using DNA obtained from blood we observed an association in the hypothesized direction even though this association may have been attenuated by the use of DNA from blood rather than brain, and the use of DNA methylation as a surrogate for gene expression.

We also note there are limitations to the cross-sectional study design. NOS2 methylation appears to change rather rapidly [26, 29, 30]. While this lends support to our results that suggest that NOS2 methylation may be noticeably altered with relatively recent changes in welding fume exposure, it also means that we cannot be certain that NOS2 methylation contributed to parkinsonism, or rather the reverse occurred. Greater UPDRS3 scores could be associated with reduced mobility, and we cannot rule out the possibility that mobility affects NOS2 methylation. In particular, it is known that exercise can affect methylation of a variety of genes in humans [31]. Another consideration when interpreting the results of the present work is that the results may have been influenced by a healthy worker effect, as evidenced by a U-shaped dose-response association between welding and parkinsonism [3], which would be expected since tremor and bradykinesia associated with parkinsonism would make highly demanding fine motor tasks difficult to perform. However, similar to the overall study findings [3], we would anticipate that any healthy worker effect may have attenuated the associations, here between NOS2 methylation and parkinsonism, and between duration of exposure and NOS2 methylation. Finally, we assessed DNA methylation from whole blood and it is possible that the differences we observed could be due to different proportions of blood cell types across UPDRS3 groups, but we have no reason to believe that blood cell types would differ according to UPDRS3 score.

Since parkinsonism in Mn neurotoxicity is presumably mediated through neuropathologic changes, studying DNA methylation of NOS2 in human brain tissue of welders and other Mn-exposed workers would provide the most definitive data. This type of specimen is not widely available in working populations in the U.S. since neuropathologic tissue can only be obtained at death. Future studies of parkinsonism and methylation of NOS2 and other potentially relevant genes, such as inflammatory and metal transporter genes, might benefit from the inclusion of DNA from both blood and brain. Only the former can be collected longitudinally, and a longitudinal assessment of NOS2 methylation will be required to strengthen our findings and those from any future studies utilizing brain tissue. While our study was focused on parkinsonism in welders, such studies might also be of interest in population-based studies of Parkinson’s disease. Parkinson’s disease shares some features of parkinsonism among Mn-exposed workers [3], and ambient Mn exposure may increase risk of Parkinson’s disease [32]. NOS2 has received considerable attention with regard to Parkinson’s disease, but results of numerous human studies, which considered NOS2 genotype but not methylation, have been inconclusive. One reason for that could have been a failure to well assess NOS2 expression by genotype alone, as might be expected, given that methylation of the gene clearly contributes to its expression [16, 17].

5. Conclusions

Lower methylation of the gene coding for iNOS was associated with greater signs of parkinsonism among workers from welding worksites, suggesting that inflammation mediated by iNOS may possibly contribute to the high prevalence of parkinsonism observed previously in workers exposed to welding fume.

Highlights.

• We assess methylation of NOS2 in 201 workers exposed to welding fume

• Exposure to manganese-containing welding fume is associated with NOS2 methylation

• NOS2 methylation is associated with parkinsonism among welders

• Inducible nitric oxide synthase may play a role in parkinsonism in welders