Content uploaded by Antonio Gasparrini
Author content
All content in this area was uploaded by Antonio Gasparrini on Dec 03, 2018
Content may be subject to copyright.
Effects of Temperature and Relative Humidity on DNA
Methylation
Marie-Abele Binda,b, Antonella Zanobettia, Antonio Gasparrinic, Annette Petersa,d, Brent
Coullb, Andrea Baccarellia, Letizia Tarantinie, Petros Koutrakisa, Pantel Vokonasf, and Joel
Schwartza
aDepartment of Environmental Health, Harvard School of Public Health, Boston, MA
bDepartment of Biostatistics, Harvard School of Public Health, Boston, MA
cDepartment of Medical Statistics, London School of Hygiene and Tropical Medicine, London,
United Kingdom
dInstitute of Epidemiology, Helmholtz Zentrum München German Research Center for
Environmental Health, Neuherberg, Germany
eCenter of Molecular and Genetic Epidemiology, University of Milan, Milan, Italy
fVA Boston Healthcare System and the Department of Medicine, Boston University School of
Medicine, Boston, MA
Abstract
Background—Previous studies have found relationships between DNA methylation and various
environmental contaminant exposures. Associations with weather have not been examined.
Because temperature and humidity are related to mortality even on non-extreme days, we
hypothesized that temperature and relative humidity may affect methylation.
Methods—We repeatedly measured methylation on long interspersed nuclear elements (LINE-1),
Alu, and 9 candidate genes in blood samples from 777 elderly men participating in the normative
aging Study (1999–2009). We assessed whether ambient temperature and relative humidity are
related to methylation on LINE-1 and Alu, as well as on genes controlling coagulation,
inflammation, cortisol, DNA repair, and metabolic pathway. We examined intermediate-term
associations of temperature, relative humidity, and their interaction with methylation, using
distributed lag models.
Results—Temperature or relative humidity levels were associated with methylation on tissue
factor (F3), intercellular adhesion molecule 1 (ICAM-1), toll-like receptor 2 (TRL-2), carnitine O-
acetyltransferase (CRAT), interferon gamma (IFN-γ), inducible nitric oxide synthase (iNOS), and
glucocorticoid receptor, LINE-1, and Alu. For instance, a 5°c increase in 3-week average
temperature in ICAM-1 methylation was associated with a 9% increase (95% confidence interval:
Copyright © 2014 by Lippincott Williams & Wilkins
Correspondence: Marie-Abele Bind, Department of Environmental Health, 401 Park Drive, Landmark Center, Suite 415, Boston, MA
02115. ma.bind@mail.harvard.edu.
The authors report no conflicts of interest.
Europe PMC Funders Group
Author Manuscript
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Published in final edited form as:
Epidemiology. 2014 July ; 25(4): 561–569. doi:10.1097/EDE.0000000000000120.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
3% to 15%), whereas a 10% increase in 3-week average relative humidity was associated with a
5% decrease (−8% to −1%). The relative humidity association with ICAM-1 methylation was
stronger on hot days than mild days.
Conclusions—DNA methylation in blood cells may reflect biological effects of temperature and
relative humidity. Temperature and relative humidity may also interact to produce stronger effects.
High- and low-ambient temperatures, extreme or not, have been associated with increased
risk for cardiovascular mortality,1,2 especially in the elderly.3 Biological mechanisms of the
adverse effects of temperature have not been fully understood. Temperature changes may
induce inflammatory responses,4-7 change blood viscosity,7,8 change heart-rate variability,9
affect blood pressure,10,11 cause myocyte injury,6 or modify cholesterol levels.8,12 The
biological pathways for the possible effects of humidity on mortality have received less
attention.1,2 Koken et al1 related dew-point temperature to coronary atherosclerosis and
congestive heart failure. Other studies have focused on the adverse effects of apparent
temperature; however, their results were inconsistent.13-17 In these studies, apparent
temperature was constructed to reflect the physiological experience of exposure to both
temperature and humidity, which may capture health effects better than using temperature
alone. To the best of our knowledge, a health effect of humidity independent of temperature
has not been demonstrated.
Recent research has linked epigenetics to cardiovascular disease.18,19 We hypothesized that
epigenetics may also play a role in the adverse effects of weather on the cardiovascular
system. The most studied epigenetic mechanism is DNA methylation, which refers to the
addition of a methyl group on cytosine bases. Defects in the methylation machinery
resulting in hypomethylation or hypermethylation, depending on the gene it regulates, can
have devastating consequences including serious diseases.20 DNA methylation can act as a
switch in gene expression because its distribution is usually bimodal, with regions either
highly methylated or unmethylated.21 Moreover, although protein levels can change within
an hour, epigenetic marks are more stable and possibly a better choice for examining
associations with environmental exposures averaged over longer periods such as weeks.22,23
Previous studies have observed changes in DNA methylation also occurred after exposure to
air pollution.24,25 In addition, Madrigano et al26 demonstrated that gene-specific
methylation changes with time. Increased age was related to tissue factor (F3), interferon
gamma (IFN-γ) carnitine O-acetyltransferase (CRAT), and 8-oxoguanosine DNA
glycosylase-1 (OGG1) hypermethylation, and inducible nitric oxide synthase (iNOS), toll-
like receptor 2 (TLR-2), and glucocorticoid receptor (GCR) hypomethylation.
Epigenetic control also plays an important role in inflammatory gene expression.27
Therefore, we hypothesized that temperature and relative humidity (independently or
synergistically) may change methylation on interspersed nuclear elements (LINE-1), Alu,
and 9 candidate genes related to coagulation, inflammation, cortisol, DNA repair, and
metabolic pathway. Because the elderly are more susceptible to the adverse effects of
temperature,3 we focused our investigation on an elderly population having repeated
measurements of DNA methylation.
Bind et al. Page 2
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
METHODS
Study Population
The study population consisted of participants from the Normative Aging Study, an
investigation of aging community-dwelling men from the Greater Boston area.27 Our
analyses included 777 participants who visited the examination site every 3 to 5 years to
undergo physical examinations (n = 1798 observations). We started to measure DNA
methylation on blood samples collected after 1999. We restricted our analyses to
participants with C-reactive protein concentrations less than 10 mg/L so that the results are
not confounded by infection state. This study was approved by the review board of all
participating institutions.
Weather and Air Pollution Assessments
We obtained measurements of dew-point and ambient temperatures, relative humidity, and
barometric pressure from the Boston Logan Airport weather station located 8 km from the
study center. Because study participants lived throughout the metropolitan area, we assumed
that the monitored temperature and humidity can serve as surrogates of their exposures. We
calculated apparent temperature, vapor pressure, and absolute humidity using the following
formulae:
In this cohort, Halonen et al28 found intermediate-term associations between temperature
averaged up to 1 month and C-reactive protein, intercellular adhesion molecule 1 (ICAM-1)
and vascular cell adhesion molecule 1 (VCAM-1). Therefore, we focused our analyses on a
similar intermediate-term exposure window and considered averages of temperature and
relative humidity over 1–3 weeks preceding each participant’s visit.
We monitored fine particles concentrations at the Harvard supersite located in downtown
Boston and approximately 1 km from the examination center. PM2.5 concentrations were
measured with a Tapered Element Oscillation Microbalance (Model 1400a, Rupprecht and
Pastaschnick, East Greenbush, NY).
Bind et al. Page 3
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Epigenetics Assessment
We collected participant’s blood at every visit and isolated bisulfite-treated DNA with the
Wizard DNA Clean-Up System (Promega, Madison, WI). We assessed DNA methylation
using highly quantitative analysis based on polymerase chain reaction pyrosequencing.29
We chose to measure methylation on genes that are expressed in blood leukocytes. We
measured F3, ICAM-1, TLR-2, CRAT, OGG1, IFN-γ, GCR, and iNOS methylation at 1–5
CpG positions within each gene’s promoter region and calculated the mean values of the
position-specific measurements (eAppendix 1). Interleukin-6 (IL-6) methylation levels were
quantified outside the gene’s promoter region. The assay measuring gene-specific
methylation was developed by locating the promoters using genomatix Software
(Genomatix Software Inc, Ann Arbor, MI). The degree of methylation was expressed as a
percentage of methylated cytosines over the sum of methylated and unmethylated cytosines
at position 5 (%5mC).
Statistical Methods
Assumptions—Because we had repeated measures of methylation for 71% of the
participants, we fit generalized mixed-effects models with random intercepts to investigate
whether levels of temperature and relative humidity weekly averaged over the 3-week
period before the jth visit of the ith participant were associated with mean methylation of the
ith participant assessed at the jth visit (Yij). Because F3, ICAM-1, TLR-2, CRAT, and OGG1
methylation had a point mass at zero and their residuals’ distribution showed important
deviation from a normal density (right skewness), we assumed a tweedie distribution for
methylation (eAppendix 2) with a log-link and reported multiplicative effects. For IFN-γ,
IL-6, iNOS, GCR, LINE-1, and Alu methylation, we assumed Gaussian distributions for the
regression residuals and present our results on the additive scale.
We adjusted for potential confounders (vector C1) such as: batch of methylation
measurement, season (winter/spring-fall/summer), and seasonal sine and cosine with a 365-
day cycle. When we modeled the date-methylation relationship using a penalized spline, we
obtained a similar seasonal cycle. Therefore, we chose to use the sine and cosine terms to
estimate both the amplitude and the phase of the seasonal cycle with only 2 degrees of
freedom. We also controlled for the following risk factors of DNA methylation (vector C2):
age, race, diabetes, body mass index, smoking status, statin use, as well as percentages of
neutrophil, lymphocyte, monocytes, and basophils in differential blood count (because the 5-
methylcytosine distribution in the genome of differentiated somatic cells varies by cell type).
We included C2 in the models for statistical efficiency and to block any potential backdoor
path through unmeasured variable U that would be a common cause of the exposures of
interest and C2.30 We thus assumed no unmeasured confounding between the exposures of
interest and DNA methylation, given the random intercepts and the C1 and C2 vectors.
Moreover, we assumed the missing mechanisms to be at random conditional on the
covariates. We also assumed that secondary air pollutants may play the role of intermediate
variables between temperature and methylation rather than being confounding variables.
Therefore, we chose not to include any air pollutant in our main regression models. We
checked this assumption in a secondary analysis in which we controlled for PM2.5
Bind et al. Page 4
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
concentrations. We used akaike’s information criterion (AIC) to decide whether an
alternative model including only apparent temperature instead of both temperature and
relative humidity was more appropriate. AIC favored the model including both temperature
and relative humidity.
We checked for nonlinear dose-response relationships of temperature and relative humidity,
with methylation using generalized additive models and cubic splines. We found no
deviation from linear dose-response relationships with respect to methylation. We explored
the nature of the temperature and relative humidity associations with methylation over time
by fitting distributed lag linear models (lags 0–20 days), and we examined how the
relationship between lagged exposures and methylation outcomes changes across lags. This
methodology, previously developed for the analysis of time-series data,31 is extended here in
the context of individual longitudinal data. We chose natural splines with 3 degrees of
freedom to model the nonlinear shape of the distributed lag. Because the temperature and
relative humidity relationships with methylation varied over the exposure lags, we
calculated from the distributed lag model the cumulative temperature and relative humidity
effects over 1-week periods (lags 0–6, 7–13, and 14–20 days) preceding the jth visit of the ith
participant, as well as the cumulative effects over the entire 3-week period.
Main Regression Models
The distributed lag model for F3, ICAM-1, TLR-2, CRAT, and OGG1 methylation
(multiplicative scale) is as follows:
with Yij~Tweedie and ui~N(0,σu2).
The distributed lag model for IFN-γ, IL-6, iNOS, GCR, LINE-1, and Alu methylation
(additive scale) is as follows:
with εij~N(0,σ2) and ui~N(0,σu2)
In both of the models, f1 and f2 represent the distributed lag functions with sets of
coefficients γ1 and γ2 constrained by a natural spline (with 3 degrees of freedom) that
correspond to the temperature and relative humidity effects at lags 0–20 days. C1 and C2
correspond to sets of variables for which we adjusted.
Bind et al. Page 5
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Temperature and Humidity Interaction
Focusing only on genes associated with both temperature and relative humidity, we
examined whether these 2 weather variables interact to modify methylation. We used a 2-
covariate penalized thin plate spline to investigate the combined effect of 3-week average
temperature and relative humidity on methylation. We obtained 3-dimensional (3D)
perspective plots and examined the shapes of the surface. A plane would suggest no
interaction between the 2 variables, whereas a deformed surface would suggest an
interaction. In addition to a 3D-perspective plot, we examined interactions by estimating the
association between relative humidity and the outcome for 3 temperature levels (determined
with the 25th and 75th percentiles of the temperature distribution).
Sensitivity Analyses
Because we chose to model the distributed lag functions with natural splines (with 3 degrees
of freedom), we also considered another choice of distributed lag functions that assumes
constant lag effect within weeks. Note that this assumption is equivalent to fitting a model
that simultaneously includes 4 consecutive weekly exposure moving averages.
For simplicity and to limit the number of tests, our additional secondary analyses examined
only the cumulative exposure effect over the entire 3-week period. We adjusted for PM2.5
concentrations and barometric pressure to evaluate the role of temperature and humidity in a
more complex weather setting. We conducted some multiple testing corrections using
Bonferroni adjustment based on 66 tests (ntests= 2 exposures × 11 outcomes × 3 lag times).
RESULTS
Descriptive Statistics
Demographic characteristics of the study participants (median age 72 years) varied
according to their number of visits (Table 1 and eAppendix 3). Boston’s climate is
continental, with direct influences from the ocean. Summers are mostly warm and humid,
and winters are usually cold and dry. Temperature and relative humidity distributions
showed variability among participants (eAppendix 4). The Pearson correlation coefficient
between temperature and absolute humidity was 0.96, and between temperature and relative
humidity 0.22 (eAppendix 5). The distributions of gene-specific methylation varied by gene:
for instance, the distribution for IFN-γ methylation was wider than for TLR-2 methylation
(eAppendices 6a and 6b). The distributions of gene-specific methylation were concentrated
around low or high values, except for IL-6 and GCR methylation.
Temperature and Relative Humidity Effects
Temperature (eAppendix 7) and relative humidity (eAppendix 8) were associated with
methylation on F3, ICAM-1, TRL-2, CRAT, IFN-γ, iNOS, GCR, LINE-1, and Alu. For
instance, a 5°c increase in temperature (1st week) was related to a 6% decrease in F3
methylation (95% confidence interval [CI] = −11% to −1%) and a 10% increase in relative
humidity (3rd week) was associated with a 0.25%5mC decrease in IFN-γ (−0.48 to −0.01).
Temperature was not related to OGG1, IFN-γ, IL-6, iNOS, and GCR methylation, and
relative humidity was not associated with TLR-2, IL-6, and OGG1 methylation. The
Bind et al. Page 6
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
distributed lag models indicated different time windows for the associations of temperature
and relative humidity with methylation, depending on the genes. For instance, although the
temperature association with F3 methylation was observed during the first week of
exposure, the signals on ICAM-1 were stronger after 3 weeks of exposure.
The associations between temperature and relative humidity exposures over the 3-week
period preceding each medical visit and methylation are summarized in Table 2. For
example, a 5°c increase in temperature and a 10% increase in relative humidity (3 week)
were associated with an 9% increase (95% CI = 3% to 15%) and a 5% decrease (−8% to
−1%) in ICAM-1 methylation, respectively.
Interactions Between Temperature and Relative Humidity
We investigated whether exposures to temperature and relative humidity (3 week) interact to
modify levels of ICAM-1 and LINE-1 methylation. After fitting a 2-dimensional smoothing
of temperature and relative humidity on ICAM-1 methylation, we obtained a 3D surface that
indicated an interaction between temperature and relative humidity during humid and hot
days (Figure 1). The relative humidity association with ICAM-1 methylation was stronger
during hot days (Figure 2). We did not find any evidence of an interaction between
temperature and relative humidity that modified LINE-1 methylation.
Sensitivity Analyses
We found similar results when we assumed a distributed lag model with constant lag effects
within weeks for temperature and relative humidity versus distributed lag model constrained
by natural splines with 3 degrees of freedom (eAppendices 7 and 8).
When we added barometric pressure to our models, the results did not change (Figure 3 and
eappendix 9a and 9b for numerical values). We also adjusted for PM2.5 in our regression
model, with few changes in our primary findings (Figure 3 and eAppendix 9a and 9b for
numerical values); except for the temperature estimates on iNOS and GCR methylation,
which became significant, and for the temperature signal on LINE-1 methylation, which
disappeared. After controlling for PM2.5, a 5°c increase in temperature (3 week) was
associated with an increase of 1.68%5mc in iNOS methylation (95% CI = 0.53 to 2.83) and
of 1.19%5mc in GCR methylation (0.20–2.19). After adjusting for multiple comparisons
using the Bonferroni correction, P values were 0.04 for the association of relative humidity
with LINE-1 methylation, 0.03 for the association with Alu methylation.
DISCUSSION
Our results suggest associations of ambient temperature and relative humidity with DNA
methylation on LINE-1, Alu, and genes related to coagulation, inflammation, cortisol, and
metabolic pathway. Our findings support previous studies showing an influence of ambient
temperature on biomarkers of inflammation4,5,28 and heart failure,6 blood viscosity,7 heart
rate variability,9 blood pressure,10,11 and cholesterol.8,12 A correlation between temperature
and methylation on repetitive elements has been reported in a previous study focusing on the
association of air pollution with LINE-1 and Alu methylation.24 Associations of temperature
and relative humidity exposures with methylation have not been previously examined.
Bind et al. Page 7
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
A temperature decrease was associated with ICAM-1 hypomethylation. ICAM-1 encodes a
cell surface glycoprotein that is overexpressed during inflammatory responses. Temperature
decrease has previously been related to higher C-reactive protein levels,5,6,28 a general
inflammatory marker, as well as to ICAM-1 and VCAM-1.28 In our data, a 1%5mc decrease
in ICAM-1 methylation was associated with 0.7% increase in ICAM-1 protein level (95% CI
= 0.0 to 1.4). The negative slope we found and the results observed by Zhang et al21 suggest
that ICAM-1 hypomethylation is related to ICAM-1 gene desilencing and thus ICAM-1
protein overexpression. Therefore, low levels of ICAM-1 methylation may be responsible for
increased ICAM-1 expression in response to cold temperature.
Similarly, a decrease in temperature was associated with CRAT hypermethylation, as well as
GCR and iNOS hypomethylation (after adjustment for PM2.5). CRAT is an important
enzyme for the cellular cycle and participates in the metabolic pathway in peroxisomes,
mitochondria, and endoplasmic reticulum. Changes in CRAT have been related to increased
risk of cardiovascular outcomes.32 in the same cohort, aging was also associated with CRAT
hypermethylation.26 CRAT hypermethylation may consist of a cellular metabolic response
because of decreased temperatures and may be a mechanism that explains the relationship
between cold temperatures and cardiovascular-related death. GCR is a receptor to which
glucocorticoids, such as cortisol, bind. An animal study has related cold weather conditions
to increased cortisol secretion in sheep.33 Therefore, cold temperature exposure may induce
GCR hypomethylation, which in turn regulates the binding of released cortisol.
iNOS is an enzyme that generates nitric oxide (NO) from the amino acid l-arginine. NO
plays an important role in blood pressure regulation, host defense mechanisms, and
inflammation.34 Oxidative-stress-related inflammatory response has also been associated
with nuclear factor kappa B (NF-κB) and inOS activity in mice.35 In an oxidative
environment in which inOS is usually formed, NO reacts with superoxide (O2−) leading to
peroxynitrite (ONOO−) production and thus cell toxicity. When playing a role in host
immunity, iNOS participates in eliminating foreign compounds such as reactive oxidative
species. Madrigano et al25 demonstrated that air pollution exposure, which also involves
oxidative stress reactions, was associated with iNOS hypomethylation in the same elderly
cohort. Cold weather conditions may lead to the formation of reactive oxidative species and
iNOS hypomethylation, enhancing iNOS protein expression.
Temperature increases were associated with TLR-2 hypomethylation. TLR-2 is a gene
coding a protein that is involved in the activation of innate immunity. TLR-2 and IL-6 may
be interrelated, as suggested by a study that found greater effects of air pollution on IL-6 for
wildtype mice compared with knockout mice for TLR-2.36 Higher IL-6 levels have also been
related to increased C-reactive protein levels,37 which is also an important player in immune
responses as an early defense system against infections. TLR-2 hypomethylation may
activate TLR-2 gene expression and induce biologic processes enhancing IL-6 and C-
reactive protein after exposure to warm temperatures. Although our results do not prove that
these methylation pathways are the primary reasons for higher C-reactive protein with
higher temperatures reported in previous studies,4,6,7 they do suggest a plausible role in
inflammatory responses because of high temperature.
Bind et al. Page 8
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
We also observed associations between relative humidity and methylation. The lower AIC
for the models with both temperature and relative humidity indicates that apparent
temperature does not capture the weather relationship with methylation as well as the 2
covariates, suggesting that relative humidity may affect methylation independently of
temperature. An increase in relative humidity (1 week) was associated with ICAM-1
hypomethylation and was suggestively related to F3 hypomethylation. Although we have
already mentioned that ICAM-1 levels are high during inflammatory responses, it is
important that tissue factor upregulation and increased fibrin production were observed in
hypercoagulable and inflammatory states.38 Proinflammatory responses are also
accompanied by increased tissue factor expression and fibrinogen.38 Humid weather
conditions may lead to F3 and ICAM-1 hypomethylation that would increase tissue factor
and ICAM-1 expression, respectively. We also observed a negative association between
relative humidity (3rd week) and IFN-γ methylation. IFN-γ is a cytokine that plays an
important role in innate and adaptive immune responses against the intrusion of foreign
compounds.39 Days with high relative humidity could cause changes in IFN-γ methylation
that would, in turn, regulate the production of the IFN-γ protein. Previous studies have
investigated the association between humidity and inflammation.1,2
Relative humidity was also associated with LINE-1 hypomethylation and Alu
hypermethylation. Similar changes in repetitive element methylation in response to air
pollution exposure have been reported in the same cohort.24 Changes in LINE-1 and Alu
methylation have been related to cardiovascular outcomes29 and could also be potential
molecular mechanisms linking weather to cardiovascular mortality.
Our work provides evidence of a potential health effect of relative humidity independent of
temperature. Some previous studies examined humidity using dew-point temperature or
apparent temperature (which depends on dew-point temperature).1,13,14,17 Koken et al1
suggested that humidity has an impact on health and disease, especially on cardiovascular
disease. A dew-point temperature increase (from the 25th to the 75th percentile) was related
to a 9% increase in risk of hospitalization for coronary atherosclerosis and a 16% increase in
congestive heart failure. During cold periods, Wichmann et al15 observed associations
between maximum apparent temperature and admissions for respiratory and cardiovascular
diseases only for some susceptible groups, such as the elderly and men. Increases in
maximum apparent temperature during the colder months have also been related to a
decrease in admissions for acute myocardial infarction.17 In addition, apparent temperature
has been associated with respiratory and circulatory deaths in an elderly population in
Vancouver.14
The different time windows observed for the associations of temperature and of relative
humidity on DNA methylation suggest that molecular responses to temperature and relative
humidity may happen either independently from each other or sequentially. Further work is
required to explore this (and perhaps other hypotheses) for the difference in time windows.
Another interesting finding was the interaction between temperature and relative humidity,
which gave rise to stronger decreases in ICAM-1 methylation during hot and humid days,
suggesting a potential overexpression of the ICAM-1 protein after those episodes. Because
high ICAM-1 levels have been related to increased risk of cardiovascular events,40 this
Bind et al. Page 9
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
synergistic effect may contribute to cardiovascular morbidity and mortality in the elderly
during hot and humid days.
Limitations and Strengths
Our analysis is limited to methylation on 9 genes and 2 repetitive elements. Methylation on
other genes and histone modifications might be other important variables to evaluate. We
chose 9 genes based on their expression in leukocyte. Because blood is a heterogeneous
tissue, it has many leukocytes types and subtypes. Even though we control for the
percentages of neutrophils, lymphocytes, monocytes, and basophils in the blood count, there
may be some residual confounding. However, when we regressed temperature and relative
humidity on percentages of neutrophils, lymphocytes, monocytes, and basophils in blood
count in our data, we found no associations. Confounding by cell-type variation is therefore
unlikely to explain our findings. We presented results averaged over 1-week to 3-week
periods and chose a specific distributed lag function. First, we expect lag-specific estimates
to contain more measurement error than estimates using averages. Indeed, when averaging
exposure measurements having random noise, some measurement error may be averaged
out. Secondly, the distributed lag function we chose may be mis-specified and therefore
would bias the estimates. However, because we obtained similar estimates using 2
distributed lag functions, this issue is unlikely to explain our positive results. Not all of our
findings were robust to Bonferroni correction. This adjustment is very conservative and
assumes independence among the tests, which is unlikely in this study.
Our approach has several strengths. The prospective study design with repeated measures of
methylation permitted us to perform a well-powered analysis. High-precision
pyrosequencing yields more accurate methylation results than are available from array data.
Our findings suggest some independent temperature and humidity relationships with
methylation. To our knowledge, these associations have not been reported previously.
Furthermore, we fitted distributed lag models and could identify an exposure time window
for the temperature and humidity associations with gene-specific and repetitive elements
methylation. Our sensitivity analysis indicates that our findings were mostly not the result of
confounding by barometric pressure and PM2.5 concentrations. We checked for model
misspecification by permitting the dose-response relationship between exposures and
methylation to be nonlinear and by allowing temperature and humidity to interact. We
expect temperature and relative humidity conditions at participants’ homes to correlate with
conditions at logan airport. The correlation between daily temperatures at logan airport,
which exhibits coastal influence, and Worcester Airport, located inland and 80 km away,
was 0.95 indicating that weather conditions measured at Logan Airport are reasonable
surrogates for residents in Eastern Massachusetts. Previous studies have suggested that
temperature changes are associated with relevant cardiovascular events, such as changes in
blood viscosity, blood pressure, heart rate variability, and cholesterol. Our study highlights
DNA methylation as a biological mechanism that could mediate the health effects of
temperature and humidity.
Bind et al. Page 10
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
Acknowledgments
We thank the Normative Aging Study participants and Tania Kotlov.
This work was supported by the following US EPA grants RD-827353 and RD-832416; NIEHS grants RO1-
ES015172, 2RO1-ES015172, R21 AG0H0027, ES014663, ES00002, and Clean Air Act grant RD83479701, and by
a grant from Medical Research Council-UK G1002296. The Normative Aging Study is supported by the
Cooperative Studies Program/Epidemiology Research and Information Center of the U.S. Department of Veterans
Affairs and is a component of the Massachusetts Veterans Epidemiology Research and Information Center, Boston,
MA.
REFERENCES
1. Koken PJ, Piver WT, Ye F, Elixhauser A, Olsen LM, Portier CJ. Temperature, air pollution, and
hospitalization for cardiovascular diseases among elderly people in Denver. Environ Health
Perspect. 2003; 111:1312–1317. [PubMed: 12896852]
2. Schwartz J, Samet JM, Patz JA. Hospital admissions for heart disease: the effects of temperature and
humidity. Epidemiology. 2004; 15:755–761. [PubMed: 15475726]
3. Kim Y, Joh S. A vulnerability study of the low-income elderly in the context of high temperature
and mortality in Seoul, Korea. Sci Total Environ. 2006; 371:82–88. [PubMed: 17007909]
4. Hampel R, Breitner S, Rückerl R, et al. Air temperature and inflammatory and coagulation
responses in men with coronary or pulmonary disease during the winter season. Occup Environ
Med. 2010; 67:408–416. [PubMed: 19884649]
5. Schneider A, Panagiotakos D, Picciotto S, et al. AIRGENE Study group. Air temperature and
inflammatory responses in myocardial infarction survivors. Epidemiology. 2008; 19:391–400.
[PubMed: 18414085]
6. Wilker EH, Yeh G, Wellenius GA, Davis RB, Phillips RS, Mittleman MA. Ambient temperature
and biomarkers of heart failure: a repeated measures analysis. Environ Health Perspect. 2012;
120:1083–1087. [PubMed: 22588803]
7. Schäuble CL, Hampel R, Breitner S, et al. Short-term effects of air temperature on blood markers of
coagulation and inflammation in potentially susceptible individuals. Occup Environ Med. 2012;
69:670–678. [PubMed: 22833664]
8. Keatinge WR, Coleshaw SR, Easton JC, Cotter F, Mattock MB, Chelliah R. Increased platelet and
red cell counts, blood viscosity, and plasma cholesterol levels during heat stress, and mortality from
coronary and cerebral thrombosis. Am J Med. 1986; 81:795–800. [PubMed: 3776986]
9. Ren C, O’Neill MS, Park SK, Sparrow D, Vokonas P, Schwartz J. Ambient temperature, air
pollution, and heart rate variability in an aging population. Am J Epidemiol. 2011; 173:1013–1021.
[PubMed: 21385834]
10. Hoffmann B, Luttmann-Gibson H, Cohen A, et al. Opposing effects of particle pollution, ozone,
and ambient temperature on arterial blood pressure. Environ Health Perspect. 2012; 120:241–246.
[PubMed: 22020729]
11. Halonen JI, Zanobetti A, Sparrow D, Vokonas PS, Schwartz J. Relationship between outdoor
temperature and blood pressure. Occup Environ Med. 2011; 68:296–301. [PubMed: 20864465]
12. Halonen JI, Zanobetti A, Sparrow D, Vokonas PS, Schwartz J. Outdoor temperature is associated
with serum HDL and LDL. Environ Res. 2011; 111:281–287. [PubMed: 21172696]
13. Almeida SP, Casimiro E, Calheiros J. Effects of apparent temperature on daily mortality in lisbon
and Oporto, Portugal. Environ Health. 2010; 9:12. [PubMed: 20219128]
14. Krstić G. Apparent temperature and air pollution vs. Elderly population mortality in Metro
Vancouver. PLoS One. 2011; 6:e25101. [PubMed: 21980381]
Bind et al. Page 11
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
15. Wichmann J, Andersen Z, Ketzel M, Ellermann T, Loft S. Apparent temperature and cause-
specific emergency hospital admissions in Greater Copenhagen, Denmark. PLoS One. 2011;
6:e22904. [PubMed: 21829550]
16. Wichmann J, Andersen ZJ, Ketzel M, Ellermann T, Loft S. Apparent temperature and cause-
specific mortality in copenhagen, Denmark: a case-crossover analysis. Int J Environ Res Public
Health. 2011; 8:3712–3727. [PubMed: 22016711]
17. Wichmann J, Ketzel M, Ellermann T, Loft S. Apparent temperature and acute myocardial
infarction hospital admissions in Copenhagen, Denmark: a case-crossover study. Environ Health.
2012; 11:19. [PubMed: 22463704]
18. Kim M, Long TI, Arakawa K, Wang R, Yu MC, Laird PW. DNA methylation as a biomarker for
cardiovascular disease risk. PLoS One. 2010; 5:e9692. [PubMed: 20300621]
19. Baccarelli A, Wright R, Bollati V, et al. Ischemic heart disease and stroke in relation to blood
DNA methylation. Epidemiology. 2010; 21:819–828. [PubMed: 20805753]
20. Robertson KD, Wolffe aP. DNA methylation in health and disease. Nat Rev Genet. 2000; 1:11–19.
[PubMed: 11262868]
21. Zhang Y, Rohde C, Tierling S, et al. DNA methylation analysis of chromosome 21 gene promoters
at single base pair and single allele resolution. PLoS Genet. 2009; 5:e1000438. [PubMed:
19325872]
22. Bollati V, Baccarelli A. Environmental epigenetics. Heredity. 2010; 105:105–112. [PubMed:
20179736]
23. Byun HM, Nordio F, Coull BA, et al. Temporal stability of epigenetic markers: sequence
characteristics and predictors of short-term DNA methylation variations. PLoS One. 2012;
7:e39220. [PubMed: 22745719]
24. Baccarelli A, Wright RO, Bollati V, et al. Rapid DNA methylation changes after exposure to
traffic particles. Am J Respir Crit Care Med. 2009; 179:572–578. [PubMed: 19136372]
25. Madrigano J, Baccarelli A, Mittleman MA, et al. Prolonged exposure to particulate pollution,
genes associated with glutathione pathways, and DNA methylation in a cohort of older men.
Environ Health Perspect. 2011; 119:977–982. [PubMed: 21385671]
26. Madrigano J, Baccarelli A, Mittleman MA, et al. Aging and epigenetics: longitudinal changes in
gene-specific DNA methylation. Epigenetics. 2012; 7:63–70. [PubMed: 22207354]
27. Halonen JI, Zanobetti A, Sparrow D, Vokonas PS, Schwartz J. Associations between outdoor
temperature and markers of inflammation: a cohort study. Environ Health. 2010; 9:42. [PubMed:
20653951]
28. Bind MA, Baccarelli A, Zanobetti A, et al. Air pollution and markers of coagulation, inflammation,
and endothelial function: associations and epigene-environment interactions in an elderly cohort.
Epidemiology. 2012; 23:332–340. [PubMed: 22237295]
29. Baccarelli A, Tarantini L, Wright RO, et al. Repetitive element DNA methylation and circulating
endothelial and inflammation markers in the VA normative aging study. Epigenetics. 2010; 5:222–
228. [PubMed: 20305373]
30. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology.
1999; 10:37–48. [PubMed: 9888278]
31. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med. 2010;
29:2224–2234. [PubMed: 20812303]
32. Brunner S, Kramar K, Denhardt DT, Hofbauer R. Cloning and characterization of murine carnitine
acetyltransferase: evidence for a requirement during cell cycle progression. Biochem J. 1997;
322(pt 2):403–410. [PubMed: 9065756]
33. Panaretto BA, Vickery MR. The rates of plasma cortisol entry and clearance in sheep before and
during their exposure to a cold, wet environment. J Endocrinol. 1970; 47:273–285. [PubMed:
5453336]
34. Lechner M, Lirk P, Rieder J. Inducible nitric oxide synthase (iNOS) in tumor biology: the two
sides of the same coin. Semin Cancer Biol. 2005; 15:277–289. [PubMed: 15914026]
35. Ali F, Sultana S. Repeated short-term stress synergizes the ROS signalling through up regulation of
NFkB and iNOS expression induced due to combined exposure of trichloroethylene and UVB
rays. Mol Cell Biochem. 2012; 360:133–145. [PubMed: 21947658]
Bind et al. Page 12
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
36. Shoenfelt J, Mitkus RJ, Zeisler R, et al. Involvement of TLR2 and TLR4 in inflammatory immune
responses induced by fine and coarse ambient air particulate matter. J Leukoc Biol. 2009; 86:303–
312. [PubMed: 19406832]
37. Van Gaal LF, Mertens IL, De Block CE. Mechanisms linking obesity with cardiovascular disease.
Nature. 2006; 444:875–880. [PubMed: 17167476]
38. Chu AJ. Role of tissue factor in thrombosis. Coagulation-inflammation-thrombosis circuit. Front
Biosci. 2006; 11:256–271. [PubMed: 16146730]
39. Lundborg M, Johansson A, Lâstbom L, Camner P. Ingested aggregates of ultrafine carbon particles
and interferon-gamma impair rat alveolar macrophage function. Environ Res. 1999; 81:309–315.
[PubMed: 10581109]
40. Luc G, Arveiler D, Evans A, et al. PRIME Study Group. Circulating soluble adhesion molecules
ICAM-1 and VCAM-1 and incident coronary heart disease: the PRIME Study. Atherosclerosis.
2003; 170:169–176. [PubMed: 12957696]
Bind et al. Page 13
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
FIGURE 1. Three-dimensional plot obtained after fitting a 2-covariate penalized thin plate
spline of temperature and relative humidity on ICAM-1 methylation.
Bind et al. Page 14
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
FIGURE 2. Associations between a 10% increase in relative humidity (over a 3-week period) and
ICAM-1 methylation, on cold, mild, and hot days (temperatures below the 25th percentile,
between the 25th and 75th percentiles, and above the 75th percentile of the distribution).
Vertical bars indicate 95% CI.
Bind et al. Page 15
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
FIGURE 3. Associations between temperature and methylation on specific genes, LINE-1, and
Alu, controlling for barometric pressure and PM2.5 concentrations (sensitivity analysis).
Vertical bars indicate 95% CI.
Bind et al. Page 16
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Bind et al. Page 17
TABLE 1
Demographic Characteristics of the Normative Aging Study Participants by Number of
Visits
Age (Years) Percentiles Obesity Statin User Diabetic Smoking
5th 50th 95th % % % Never (%) Former (%) Current (%)
Baseline (n = 777) 62 72 84 27 36 14 29 67 4
Among participants having 1 visit (n1 = 221)
Visit 1 64 76 88 30 40 18 26 70 4
Among participants having 2 visits (n2 = 217)
Visit 1 60 73 83 28 35 16 26 69 5
Visit 2 66 77 86 27 54 19 26 70 4
Among participants having 3 visits (n3 = 216)
Visit 1 62 71 82 25 36 9 29 68 3
Visit 2 66 74 86 26 52 13 28 69 3
Visit 3 69 78 89 25 62 17 27 71 2
Among participants having 4 visits (n4 = 120)
Visit 1 60 69 77 22 29 10 38 58 4
Visit 2 63 72 81 22 42 11 38 58 4
Visit 3 66 75 84 18 59 16 38 59 3
Visit 4 70 78 87 17 65 18 38 60 2
Among participants having 5 visits (n5 = 3)
Visit 1 62 66 66 33 33 0 33 67 0
Visit 2 65 68 70 33 33 0 33 67 0
Visit 3 68 70 72 33 0 0 33 67 0
Visit 4 71 73 74 33 0 0 33 67 0
Visit 5 73 76 77 33 33 0 33 7 0
Epidemiology. Author manuscript; available in PMC 2015 January 01.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Bind et al. Page 18
TABLE 2
Associations with DNA Methylation of Temperature and of Relative Humidity Over the 3-Week Period Preceding Medical Examination
Methylation mean ratio for a Δ* increase in temperature and relative humidity (95% CI)
F3 ICAM-1 TLR-2 CRAT OGGI
Temperature 0.945 (0.874–1.021) 1.092 (1.034–1.154) 0.933 (0.872–0.999) 1.053 (1.004–1.104) 1.013 (0.904–1.134)
Relative humidity 0.967 (0.921–1.015) 0.952 (0.920–0.985) 0.978 (0.938–1.020) 0.966 (0.920–1.014) 0.971 (0.903–1.043)
Change in methylation (% 5mC) for a Δ* increase in temperature and relative humidity (95% CI)
IFN-
γ
IL-6 iNOS GCR LINE-1 Alu
Temperature 0.396 (−0.256 to 1.048) −0.736 (−1.810 to 0.338) 0.863 (−0.174 to 1.900) 0.845 (−0.053 to 1.743) −0.497 (−0.915 to −0.080) 0.074 (−0.114 to 0.262)
Relative humidity −0.289 (−0.684 to 0.106) 0.390 (−0.264 to 1.043) 0.913 (0.253 to 1.572) 0.328 (−0.222 to 0.877) −0.464 (−0.719 to −0.210) 0.199 (0.083 to 0.314)
Δ* corresponds to increments of 5°C and 10% for temperature and relative humidity, respectively.
Variables included in the models: f1(temperature), f2(relative humidity), age, body mass index, smoking status, diabetes status, statin use, % neutrophils in blood count, % lymphocytes in blood count, %
monocytes in blood count, % basophils in blood count, seasonal sine and cosine, season, and batch.
f1 (temperature) and f2 (relative humidity) represent the distributed lag functions with sets of coefficients constrained by a natural spline (with 3 degrees of freedom) that correspond to the temperature and
relative humidity effects at lags 0 and 20.
Epidemiology. Author manuscript; available in PMC 2015 January 01.