https://orcid.org/0000-0002-3364-3729
ABSTRACT
Mistimed sleep/wake and eating patterns put shift workers at increased risk for cardiometabolic disease, and epigenetic modification of circadian genes has been proposed as a mechanism. Although not as extreme as shift workers, adolescents often have delayed sleep timing and irregular eating patterns. The aim was to assess whether sleep midpoints – median of bed and wake time – and dietary patterns in adolescents were associated with DNA methylation of circadian genes. The study population included 142 Mexican youth (average age of 14.0 (SD = 2.0) years, 49% male). Average sleep midpoint over weekdays was estimated with actigraphy. Diet was assessed with a semi-quantitative food frequency questionnaire, and three dietary patterns were derived from principal component analysis, a Plant-based & lean proteins pattern, a Meat & starchy pattern, and an Eggs, milk & refined grain pattern. DNA methylation was quantified in blood leukocytes with the Infinium MethylationEPIC BeadChip, and data from 548 CpG sites within 12 circadian genes were examined. Linear regression analyses, adjusted for sex, age, and % monocytes, showed that later sleep timing was associated with higher DNA methylation of several circadian genes, notably with RORB, PER1, CRY2, and NR1D1. Each of the dietary patterns examined was also related to circadian gene DNA methylation, but the Eggs, milk & refined grain pattern (‘breakfast’ pattern) had the clearest evidence of relationships with circadian genes, with inverse associations (lower DNA methylation) across all 12 genes. Findings suggest that timing-related sleep and eating behaviours among adolescents could result in epigenetic modification of clock genes.
Introduction
Circadian disruption refers to a misalignment in timing of daily activities, including eating and sleeping, with the internal circadian rhythm. Shift workers display the most extreme example of circadian disruption, but there are a range of circadian disruptions present in non-shift work populations. Adolescents are one population who may be particularly susceptible to circadian disruption, since there are multiple factors that delay bedtimes (e.g., social media and screen use [Citation1]) and increase variability in sleep timing [Citation2] (e.g., sleeping in late on weekends, napping during the day). Eating patterns may also contribute to misalignment of daily activities among adolescents, as the pubertal transition often coincides with changes in eating behaviours, particularly with alterations in timing (e.g., late-night eating [Citation3]) and stability [Citation4,Citation5] (e.g., irregular snacking rather than regular meals).
Research among shift workers consistently shows that sleeping and eating at non-optimal times is associated with cardiometabolic disease risk [Citation6,Citation7]. Moreover, mounting evidence suggests that circadian disruption observed within non-shift work populations, including adolescents, may also increase risk for cardiometabolic disease. For example, sleep timing (a typical proxy for circadian disruption), independent of sleep duration, was associated with higher metabolic syndrome risk among girls in a cohort of adolescents from the US [Citation8]. In addition, a study of adolescents from Morelos, Mexico showed that both later (after 11 PM) and earlier (before 9 PM) bedtimes, as compared to those between 9 and 10 PM, were associated with higher incidence of elevated blood pressure over a 14-month period [Citation9].
One proposed mechanism to explain a link between disrupted circadian rhythms and poor cardiometabolic health is through altered expression of circadian genes. Circadian, or clock genes, are responsible for maintaining 24-h rhythmicity of sleep and wake processes, including the diurnal rhythms of metabolism [Citation10]. One pathway by which gene expression can be altered by environmental cues such as sleep and diet is through the epigenetic mechanism of DNA methylation [Citation11,Citation12], i.e., the addition of a methyl group to the 5ʹ-carbon of cytosine in a Cytosine-phosphate-Guanine (CpG) dinucleotide. Recent evidence shows that shift workers in comparison to those on day shifts have differential DNA methylation patterns of certain circadian genes [Citation13,Citation14], which in turn is likely related to differences in gene expression and function [Citation15]. Nonetheless, the degree to which circadian disruption could alter DNA methylation of circadian genes within adolescent populations remains unknown. Adolescence represents an important developmental window where the epigenome may be especially vulnerable to environmental and behavioural influence. Further, uncovering associations between sleep timing and dietary patterns with circadian genes could have important implications for later disease risk, particularly cardiometabolic health, as circadian genes play an important role in metabolism [Citation16]. Within a pilot study of 142 Mexican adolescents, the primary aim was to assess whether weekday sleep midpoint, defined as the median of bed and wake time, was associated with DNA methylation at multiple CpG sites of 12 selected core circadian genes, quantified via the Infinium MethylationEPIC. These genes were selected for study due to their known function in regulating circadian rhythms. A secondary aim was to examine associations between dietary patterns and circadian gene methylation.
Results
Mean (SD) age of the sample was 14.0 (2.0) years, and 49% were male. Mean (SD) weekday midpoint was 3:41 AM (1.3 h). Later sleep midpoint was associated with older age and longer sleep duration (). Dietary pattern scores were not statistically significantly associated with midpoints, although adolescents with later midpoints tended to have lower intake of the Plant-based and lean proteins pattern and the Eggs, milk & refined grain pattern.
Table 1. Characteristics of the 142 Mexican adolescents
Overall, we examined associations with DNA methylation at 548 CpG sites within the 12 selected circadian genes. The statistically significant associations (at P < 0.05) between individual loci and the sleep midpoint and dietary pattern exposures are shown in .
Table 2. Associations between sleep and dietary patterns and DNA methylation of circadian genes in Mexican adolescents, P < 0.05
Sleep midpoint and DNA methylation results
There were 24 associations with P < 0.05, representing loci across 8 different genes, and there were 3 associations at the Bonferroni-corrected P value of <0.004. These included a positive association between midpoint and cg14170313, a CpG site within the island region of RORB (P = 0.001), a positive association with cg15603424, a CpG site within an island of BMAL1 at the 5ʹUTR (P = 0.003), and an inverse association with cg03301481, a site found within the gene body of PER1 (P = 0.0036).
Due to the fact that some circadian genes had a higher coverage (RORA, in particular, had 192 CpG sites while the remaining genes had between 19 and 64), we next calculated the proportion of loci within a particular gene that had statistically significant associations (P < 0.05). Several genes were found to have >5% of CpG sites with statistically significant associations (). To illustrate, later midpoint was associated with differential methylation of 11% of the CpG sites measured within PER1 and 9% with both CRY2 and NR1D1; and the majority of these associations were in the positive direction.
Table 3. Associations between sleep and dietary patterns and DNA methylation of circadian genes in Mexican adolescents, overall description
Next, we averaged DNA methylation levels across CpG sites within annotated CpG islands (some genes had coverage at multiple CpG islands). For sleep midpoint, there were positive associations between DNA methylation of RORA island 3 (β = 0.018; P = 0.045) and for RORB (β = 0.04; P = 0.003). In general, most associations between sleep midpoint and island regions were positively correlated ().
Table 4. Associations between sleep and dietary patterns and DNA methylation of circadian gene CpG islands in adolescents
To ascertain whether any of the sleep midpoint findings could be attributed to sleep duration (since sleep midpoint and sleep duration were positively correlated), we ran sensitivity analyses using sleep duration as the outcome. We found essentially no overlap in the associations, indicating that sleep midpoint results were independent of sleep duration.
Diet pattern results
For Plant-based and lean proteins pattern, there were 12 associations at P < 0.05 across 7 genes with 1 at the Bonferroni-corrected threshold; the Meat & starchy foods diet pattern had 15 associations at P < 0.05 across 7 genes and 2 at the Bonferroni-corrected threshold; and the Eggs, milk & refined grain pattern had 62 associations at P < 0.05 from all 12 genes, and 4 at the Bonferroni-corrected threshold. Overall, the two most statistically significant associations (smallest P values) were for cg04917262 (found in the gene body of PER3) and the Plant-based and lean proteins diet pattern (P = 0.0006); and for cg19170589 (found in an island region of RORA) with the Meat & starchy foods diet pattern (P = 0.0007). Focusing in on the loci found in islands, higher adherence to the Plant-based and lean proteins diet pattern was associated with differential methylation in 10% of the PER2 CpG island loci, associations that were generally in the inverse direction. Multiple genes were highlighted in the Eggs, milk & refined grain pattern: BMAL1, NR1D1, NR1D2, PER1, and PER2 each had higher than 10% of CpG sites that were associated at P < 0.05, and in an inverse direction. Of these, 19% of the loci within NR1D1 were associated with the Eggs, milk & refined grain pattern.
Next, examining the associations between dietary patterns and the averaged CpG sites within island regions (), we found one positive association between the Plant-based and lean proteins diet pattern and PER1. There were no associations for Meat & starchy foods pattern. For the Eggs, milk & refined grain pattern, there were inverse associations for PER2, CRY2, RORA island 1, RORA island 3, and NR1D1 island 1. In general, most associations between the Eggs, milk & refined grain and island regions were inversely correlated, while directions of associations with the other dietary patterns were mixed.
Post-hoc analyses
Supplemental exploratory sex-stratified analyses revealed that many of the associations between sleep/dietary exposures and circadian gene CpG islands varied by sex (). For example, associations with sleep midpoint were more evident among girls. For dietary patterns, some associations (e.g., island of PER1 with the Plant-based and lean proteins pattern) were more evident among boys while others were more apparent among girls (e.g., island of PER2 with the Eggs, milk & refined grain pattern). Nonetheless, these analyses were underpowered.
Finally, we conducted post-hoc analysis among a subsample of participants who had RNA-seq data (Supplemental Table 1). We found weak inverse correlations, although not statistically significant, between island DNA methylation and gene expression in PER2 (−0.13), PER3 island 1 (−0.11), and RORB (−0.12). We found weak positive correlations between island DNA methylation and gene expression in BMAL1 (0.19), RORA island 3 (0.36), NR1D1 island 2 (0.14), and NR1D2 (0.27). The only correlation that was statistically significant was the positive correlation with RORA island 3.
Discussion
Within this cohort of Mexican adolescents, delayed sleep timing was generally associated with higher DNA methylation of core clock genes, whereas higher intake of a dietary pattern composed of breakfast foods was associated with lower methylation of core clock genes. In particular, actigraphy-assessed sleep timing was associated with DNA methylation of several core circadian genes, notably with RORB, PER1, CRY2, and NR1D1. Each of the three dietary patterns examined was also related to DNA methylation patterns, but the Eggs, milk & refined grain pattern (i.e., breakfast pattern) had the clearest evidence of relationships with circadian genes, with associations across all 12 genes. Within the CpG island regions, there were inverse associations between the Eggs, milk & refined grain pattern and PER2, CRY2, RORA, and NR1D1.
Prior studies examining the role of sleep on circadian gene DNA methylation patterns are rare, although a few adult studies exist. In line with the direction of the present findings, DNA methylation of core clock genes (candidate gene analysis) showed that sleep deprivation among nurses and midwives was associated with higher DNA methylation of PER2 [Citation17], although it was associated with lower methylation of CRY2. The study did not examine the role of sleep timing, although sleep deprivation is likely correlated with misaligned sleep times, especially in this population of workers who often work night shifts. Another study of 124 health-care workers found higher methylation of PER3 (at three loci within the gene body) in nightshift workers compared to dayshift workers [Citation14]. An experimental investigation of sleep deprivation and circadian gene methylation and expression among 15 young men found that in response to sleep deprivation, there was higher methylation of CRY1 and PER1 in adipose tissue, but no differentially methylated sites in skeletal tissue [Citation18]. Differences in DNA methylation across tissues may help explain inconsistencies between studies that use leukocyte DNA and DNA from other tissues.
Associations between the Eggs, milk & refined grain pattern and circadian gene DNA methylation were particularly notable as they were more numerous than associations with sleep or the other dietary patterns. Furthermore, these associations were consistently inverse, which was opposite the direction observed for sleep timing. There may be a few explanations as to why the associations with this dietary pattern were robust and consistent. The Eggs, milk & refined grain pattern contains breakfast foods, and thus could be a proxy for the regular consumption of breakfast. The daily timing of eating, which begins with breakfast, is an important circadian regulator and may also have effects on metabolic health [Citation19,Citation20]. The nutrients abundant within foods comprising the Eggs, milk & refined grain dietary pattern may also play a role. Eggs and fortified grains are sources of folate, a nutrient that can play a role in maintaining DNA methylation profiles following mitosis [Citation21,Citation22]. In addition, eggs and dairy contain nutrients that feed the one-carbon metabolism pathway which is also important for maintaining DNA methylation, including vitamin B12, choline, and methionine [Citation23]. Associations with the other two dietary patterns, the Plant-based and lean proteins and the Meat & starchy pattern, were much fewer in number. Yet, the direction of the associations with the Plant-based and lean proteins pattern, also rich in methyl donors such as folate and vitamin B12, was in the same direction as the Eggs, milk & refined grain pattern (inverse), while the associations with the Meat & starchy pattern were more mixed.
Overall, there was evidence that delayed sleep timing and diet patterns (in particular the breakfast pattern) were associated with DNA methylation of circadian genes in adolescents. The fact that these associations could be detected in a healthy population of adolescents is enlightening, because it indicates that even relatively constrained variability in sleep times and dietary patterns (as opposed to complete circadian shifts found in shift work populations) are associated with differences in circadian gene DNA methylation patterns. Due to the cross-sectional study design, it cannot be determined whether lifestyle behaviours preceded DNA methylation or vice versa. However, these findings provide preliminary support for a potential pathway whereby delayed sleep timing and regular consumption of breakfast foods are related to DNA methylation of circadian genes, which could ultimately affect gene expression and phenotypic outcomes such as BMI. Increasing evidence exists for associations between circadian gene methylation and BMI [Citation24,Citation25], although no studies have yet linked sleep, diet, DNA methylation, and BMI in a longitudinal manner. Another unique finding is that the associations may differ by sex. Of note was that associations with sleep midpoint were more pronounced (larger effect sizes and/or higher precision) among girls. Although speculative, it is worth mentioning that later midpoints among girls could be reflective of chronic insomnia, a sleep disorder that is much more prevalent among females than males [Citation26], and which might have independent associations with DNA methylation [Citation27].
In exploratory post-hoc analysis, we examined whether DNA methylation of these circadian genes was associated with gene expression in a subset of the participants. While this analysis was statistically underpowered, we observed correlations in the expected inverse direction for PER2, PER3 island 1, and RORB. In contrast, we also observed some positive correlations, with the strongest for RORA island 3 (the only statistically significant correlation) and NR1D2. These results provide some preliminary evidence that DNA methylation of these circadian genes could affect gene expression, although the magnitudes of these correlations are relatively low. The inconsistent directions of the correlations could also have to do with the fact that not all of the CpG islands were located within promoter regions (e.g., RORA island 3 was found in the gene body).
Strengths of the study include the use of objective data (from actigraphy) to characterize weekday sleep timing, and the ability to examine multiple important circadian genes. This pilot analysis had a limited sample size; thus, findings warrant replication in larger study populations and in populations from different ethnic backgrounds. Given that the study sample was reflective of a low to middle-SES population in Mexico City, the findings may also not be broadly generalizable to the entire Mexican population. Further, there may be unmeasured confounders, such as mental health and stress-related factors, which can be associated both with DNA methylation and with disrupted sleep and eating patterns [Citation28]. The ability to complete fully powered sex-stratified analyses would also elucidate whether findings indeed differed for boys versus girls (as they appeared in exploratory analyses). The fact that the samples were leukocytes from whole blood also means we could not investigate DNA methylation within other tissues of interest for cardiometabolic health and/or sleep regulation such as brain, liver, or adipose tissue which were not feasible to collect in our study population. Finally, there was not access to chronotype information (whether adolescents were more ‘morning’ types or ‘evening’ types).
In summary, within a sample of Mexican adolescents, there were cross-sectional associations between sleep timing and eating patterns with DNA methylation of core circadian genes. Future work that links sleep, diet, and DNA methylation with longitudinal change in metabolic outcomes will be pivotal to understanding any potential effects of misaligned sleep and eating on long-term metabolic health in adolescents.
Methods
Study population
The study sample included adolescent participants from 2 of 3 sequentially enrolled cohorts of the Early Life Exposure in Mexico to ENvironmental Toxicants (ELEMENT) study [Citation29]. Between 1997 and 2004, 1012 mother/child dyads were recruited from prenatal clinics of the Mexican Social Security Institute in Mexico City, which serves low- to middle-income populations formally employed in the private sector. At baseline, mothers reported sociodemographic and health information. Children were followed from birth approximately every 6–12 months to 5 years of age and then periodically over mid-childhood and adolescence to gain information on growth and relevant environmental and nutritional exposures. In 2015, a subset of 550 participants from the original birth cohorts 2 and 3, who were in the midst of pubertal transition (ages 9 to 17 years), were selected to participate in a follow-up visit. The present analysis includes adolescents who took part in this 2015 follow-up visit (n = 550), held in a research facility of the ABC hospital of Mexico City. Within a pilot sample of 145 adolescents (selected for analysis based on availability of biospecimens and completeness of prior visits), epigenome-wide analyses were conducted with blood leukocyte DNA taken during the follow-up visit. Of these, 142 adolescents also had 7-day actigraphy collected for sleep estimation, taken in the week following the in-clinic visit. The institutional review boards at the Mexico National Institute of Public Health and the University of Michigan approved the research protocols. Informed consent was obtained from parents for all participants in addition to participant assent.
Sleep timing
At the end of the clinical follow-up, adolescents were given an actigraph (ActiGraph GT3X+; ActiGraph LLC, Pensacola, FL) to wear on their non-dominant wrist continuously for 7 days. Nightly sleep parameters were estimated from the actigraphic data with the use of a fusedLASSO (least absolute shrinkage and selection operator)-based calculator package developed in R (R Foundation for Statistical Computing, Vienna) [Citation30]. The primary exposure of the study was average weekday midpoint. Midpoint refers to the median of sleep onset and wake time (reported in decimal hours), and a later sleep midpoint is considered a proxy for circadian misalignment.
Diet patterns
During the adolescent follow-up visit, a trained interviewer administered a 116-item semi-quantitative food frequency questionnaire to the adolescents to obtain information on typical consumption habits. The questionnaire asked adolescents to recall how often they had consumed one serving of a standard portion size of each food item over the previous week [Citation31]. Photographs of food items were provided as a visual aid. Response options ranged from never to ≥6 times per day. We converted the raw response values (1–9) into servings/day. Dietary patterns were identified in a previous analysis using principal component analysis [Citation32]. Each adolescent receives a score representing adherence to the dietary patterns, with higher scores representing higher similarity to the pattern. Three dietary patterns were identified in this cohort; the first pattern, called the Plant-based & lean proteins pattern, was characterized by high intake frequency of vegetables, fruit, soup, fish, water and unsweetened drinks, and high-fat dairy. The second pattern, called Meat & starchy, was marked by high intake frequency of ‘Western’ processed foods including chips, refined grains, sugar-sweetened beverages, processed meat, and high-fat dairy, as well as consumption more in line with ‘traditional’ foods such as the Mexican food group (e.g., tacos and quesadillas), potatoes and fried plantains, soup, legumes, and corn tortillas. The third pattern, the Eggs, milk & refined grain pattern, was characterized by high intake frequency of refined grains, milk, sweetened milk, mayonnaise or margarine, and eggs.
DNA methylation
Whole blood was collected via venipuncture and stored in EDTA-preservative at −80°C prior to processing. DNA was isolated from blood leukocytes via the Flexigene kit (Qiagen). For epigenome-wide analysis of DNA methylation, blood leukocyte DNA samples were first treated with sodium bisulphite, which converts unmethylated cytosines to uracils leaving methylated cytosines unchanged [Citation33]. DNA methylation profiles were then assessed with the Infinium MethylationEPIC BeadChip (Illumina). The BeadChip interrogates methylation at >850,000 cytosine-guanine dinucleotides (CpG sites) in gene promoters, enhancers, within gene bodies, and in intergenic regions [Citation34]. Bisulphite converted samples were randomized across chips and chip positions, hybridized to BeadChips, and signals read and processed by experienced personnel in the University of Michigan Advanced Genomics Core. Data produced by the BeadChip were processed according to a pipeline designed for this technology and previously utilized[Citation35]. Briefly, raw data consisting of average betas, the proportion of methylated cytosines at each site, were read into R Project for Statistical Computing using minfi [Citation36]. Background correction, dye bias correction, and functional normalization to remove unwanted technical variation estimated from control probes included on each chip were performed [Citation37,Citation38]. Poorly detected probes (in at least 5% of samples), probes known to be cross-reactive [Citation39], and probes with polymorphisms in the CpG site or the single base extension site were excluded. Cell type proportions from the blood leukocyte samples were estimated based on data from tissue-specific differentially methylated regions included on the BeadChip [Citation40,Citation41]. Final data consisted of beta values which are representative of the proportion methylated (from 0 to 1) at each CpG site. Data were extracted from all probes annotated of 12 circadian genes, including BMAL1, CLOCK, CRY1, CRY2, NR1D1, NR1D2, PER1, PER2, PER3, RORA, RORB, and RORC, with 548 total CpG sites (average of 46 CpG sites/gene, with a range of 19 to 192). These genes play an important role in establishing circadian rhythms [Citation42]. BMAL1 and CLOCK genes form the positive limb of a feedback loop and are expressed at night, while PER and CRY genes form the negative feedback loop and are expressed during the day. ROR genes and NR1D1/NR1D2 are nuclear receptors that are part of the stabilizing loop for the core feedback loop. Mutations in these genes can result in altered circadian phases and abnormal sleep patterning [Citation43].
Gene expression analysis via RNA-sequencing
For a subset of participants with adequate RNA quality and quantity (n = 72), next-generation sequencing of RNA (‘RNA-Seq’) was completed to obtain relative expression data for all genes, and results were utilized from the 12 circadian genes of interest to this study. Following collection of whole blood into EDTA-containing tubes, white blood cells were extracted, preserved in RNALater, and stored frozen (−80°C) until further processing. RNA was isolated via the All-Prep kit (Qiagen). Quality and quantity were assessed via a Bioanalyzer Tapestation (Agilent). Libraries were prepared with Universal Plus mRNA-Seq with Human globin AnyDeplete (NuGEN Technologies, Inc.) which removes globin transcripts that are highly abundant in blood samples. Library preparation and sequencing were performed at the University of Michigan Advanced Genomics Core. Paired-end 50 cycle sequencing on an Illumina HiSeq 4000 was performed. Quality of the raw reads data for each sample was checked using FastQC (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc/; version 0.11.3). The Tuxedo Suite software package was used for alignment [Citation44–46]. Briefly, reads were aligned to the reference mRNA transcriptome (hg19) (http://genome.ucsc.edu/) using TopHat (version 2.0.13) and Bowtie2 (version 2.2.1.) followed by a second round of post-alignment quality control in FastQC. One sample was dropped due to low alignment rates. All samples used in downstream analysis (n = 71) had at least 16.8 million good quality aligned reads. Prior to analysis, read counts were normalized by the trimmed mean of M-values method [Citation47].
Covariates
Other covariates included in the present analysis were sex, age, BMI-for-age Z scores, maternal education, physical activity and percentage of monocytes (as a proportion of total leukocytes, described above). Trained research assistants measured height in cm (Tonelli E120 A) and weight in kg (InBody230). BMI Z scores (BMIz) accounting for age and sex were calculated based on the World Health Organization reference [Citation48]. Maternal education was reported by mothers at the original cohort enrolment visit and was categorized as <9 years, 9 to 11 years, 12 years, or >12 years of schooling. Physical activity was estimated from the accelerometers as average minutes of moderate/vigorous activity per day as previously described [Citation49].
Statistical analysis
Means ± SD of sociodemographic and lifestyle characteristics were first estimated according to earlier versus later sleep midpoint (split at the median). Next, linear regression analyses were used to evaluate the associations between midpoint and each circadian gene CpG site, adjusting for age, sex, and % monocyte. Adjustment for other covariates did not affect results and thus were not retained. Sensitivity analyses with sleep duration as the outcome were also conducted, to determine whether results were similar for sleep midpoint and duration. Similar regression models were run for each dietary pattern score (all dietary patterns in separate models). Next, DNA methylation at annotated CpG island regions within these genes was examined. To do so, data from all of the CpG sites within a specific island region (some genes had more than one island region) were averaged (see Supplemental Table 1 for location of island sites and description of the location, i.e., promoter region versus gene body). Linear regression analysis was then conducted in the same manner as for the individual CpG sites, but with the averaged island regions as the outcomes. For these models, exploratory sex-stratified analyses were also run.
Finally, whether the % DNA methylation in the island regions corresponded to gene expression (RNA-seq) was examined. Using a subset of participants (N = 71) that also had RNA-seq data, the correlations between the % methylation of the islands with RNA-seq data of the corresponding genes were obtained.
Due to the exploratory nature of the analyses, any findings at P < 0.05 are reported. A Bonferroni-corrected P-value threshold of P < 0.004 (0.05 divided by 12 genes) was also applied. All analyses were performed in Stata 14.0.
Supplemental Material
Download MS Word (21.2 KB)Acknowledgments
We gratefully acknowledge the ELEMENT research staff and the American British Cowdray Medical Center (ABC) in Mexico for providing research facilities.
Disclosure statement
Dr Chervin has served on member boards or advisory board for the American Academy of Sleep Medicine, Associated Professional Sleep Societies, International Pediatric Sleep Association, Sweet Dreamzzz, and the Pajama Program; and received royalties from UpToDate and Cambridge University Press. Authors report no conflicts of interest.
Supplementary material
Supplemental data for this article can be accessed here.
Additional information
Funding
References
- Hale L, Guan S. Screen time and sleep among school-aged children and adolescents: a systematic literature review. Sleep Med Rev. 2015;21:50–58.
- Becker SP, Sidol CA, Van Dyk TR, et al. Intraindividual variability of sleep/wake patterns in relation to child and adolescent functioning: a systematic review. Sleep Med Rev. 2017;34:94–121.
- Hernandez E, Kim M, Kim WG, et al. Nutritional aspects of night eating and its association with weight status among Korean adolescents. Nutr Res Pract. 2016;10(4):448–455.
- Savige G, MacFarlane A, Ball K, et al. Snacking behaviours of adolescents and their association with skipping meals. Int J Behav Nutr Phys Act. 2007;4. DOI:10.1186/1479-5868-4-36
- Siega-Riz AM, Carson T, Popkin B. Three squares or mostly snacks - what do teens really eat?: a sociodemographic study of meal patterns. J Adolesc Health. 1998;22(1):29–36.
- Kervezee L, Kosmadopoulos A, Boivin DB. Metabolic and cardiovascular consequences of shift work: the role of circadian disruption and sleep disturbances. Eur J Neurosci. 2020;51(1):396–412.
- Molzof HE, Wirth MD, Burch JB, et al. The impact of meal timing on cardiometabolic syndrome indicators in shift workers. Chronobiol Int. 2017;34(3):337–348.
- Cespedes Feliciano EM, Rifas-Shiman SL, Quante M, et al. Chronotype, social jet lag, and cardiometabolic risk factors in early adolescence. JAMA Pediatr. 2019;173(11):1049–1057.
- Jansen EC, Dunietz GL, Matos-Moreno A, et al. Bedtimes and blood pressure: a prospective cohort study of mexican adolescents. Am J Hypertens. 2020;33(3):269–277.
- Poggiogalle E, Jamshed H, Peterson CM. Circadian regulation of glucose, lipid, and energy metabolism in humans. Metabolism. 2018;84:11–27.
- Wu Y, Sánchez BN, Goodrich JM, et al. Dietary exposures, epigenetics and pubertal tempo. Environ Epigenetics. 2019;5(1). DOI:10.1093/eep/dvz002
- Jansen EC, Dolinoy DC, O’Brien LM, et al. Sleep duration and fragmentation in relation to leukocyte DNA methylation in adolescents. Sleep. 2019;42(9). DOI:10.1093/sleep/zsz121
- Reszka E, Wieczorek E, Przybek M, et al. Circadian gene methylation in rotating-shift nurses: a cross-sectional study. Chronobiol Int. 2018;35(1):111–121.
- Bhatti P, Zhang Y, Song X, et al. Nightshift work and genome-wide DNA methylation. Chronobiol Int. 2015;32(1):103–112.
- Resuehr D, Wu G, Johnson RL, et al. Shift work disrupts circadian regulation of the transcriptome in hospital nurses. J Biol Rhythms. 2019;34(2):167–177.
- Cedernaes J, Waldeck N, Bass J. Neurogenetic basis for circadian regulation of metabolism by the hypothalamus. Genes Dev. 2019;33(17–18):1136–1158.
- Bukowska-Damska A, Reszka E, Kaluzny P, et al. Sleep quality and methylation status of core circadian rhythm genes among nurses and midwives. Chronobiol Int. 2017;34(9):1211–1223.
- Cedernaes J, Osler ME, Voisin S, et al. Acute sleep loss induces tissue-specific epigenetic and transcriptional alterations to circadian clock genes in men. J Clin Endocrinol Metab. 2015;100(9):E1255–61.
- St-Onge MP, Ard J, Baskin ML, et al. Meal timing and frequency: implications for cardiovascular disease prevention: a scientific statement from the American heart association. Circulation. 2017;135(9):e96–e121.
- Lopez-Minguez J, Gómez-Abellán P, Garaulet M. Timing of breakfast, lunch, and dinner. Effects on obesity and metabolic risk. Nutrients. 2019;11(11). DOI:10.3390/nu11112624
- Caffrey A, Irwin RE, McNulty H, et al. Gene-specific DNA methylation in newborns in response to folic acid supplementation during the second and third trimesters of pregnancy: epigenetic analysis from a randomized controlled trial. Am J Clin Nutr. 2018;107(4):566–575.
- Orjuela MA, Mejia-Rodriguez F, Quezada AD, et al. Fortification of bakery and corn masa-based foods in Mexico and dietary intake of folic acid and folate in Mexican national survey data. Am J Clin Nutr. 2019;110(6):1434–1448.
- Mahmoud AM, Ali MM. Methyl donor micronutrients that modify DNA methylation and cancer outcome. Nutrients. 2019;11(3). DOI:10.3390/nu11030608
- Ramos-Lopez O, Samblas M, Milagro FI, et al. Circadian gene methylation profiles are associated with obesity, metabolic disturbances and carbohydrate intake. Chronobiol Int. 2018;35(7):969–981.
- Samblas M, Milagro FI, Mansego ML, et al. PTPRS and PER3 methylation levels are associated with childhood obesity: results from a genome-wide methylation analysis. Pediatr Obes. 2018;13(3):149–158.
- Meers J, Stout-Aguilar J, Nowakowski S. Sex differences in sleep health. Sleep and Health. 2019. DOI:10.1016/B978-0-12-815373-4.00003-4
- Carroll JE, Irwin MR, Levine M, et al. Epigenetic aging and immune senescence in women with insomnia symptoms: findings from the women’s health initiative study. Biol Psychiatry. 2017;81(2):136–144.
- Ämmälä AJ, Urrila AS, Lahtinen A, et al. Epigenetic dysregulation of genes related to synaptic long-term depression among adolescents with depressive disorder and sleep symptoms. Sleep Med. 2019;61:95–103.
- Perng W, Tamayo-Ortiz M, Tang L, et al. Early life exposure in mexico to environmental toxicants (ELEMENT) project. BMJ Open. 2019;9(8). DOI:10.1136/bmjopen-2019-030427
- Jansen EC, Dunietz GL, Chervin RD, et al. Adiposity in adolescents: the interplay of sleep duration and sleep variability. J Pediatr. 2018;203:309–316.
- Denova-Gutiérrez E, Ramírez-Silva I, Rodríguez-Ramírez S, et al. Validity of a food frequency questionnaire to assess food intake in Mexican adolescent and adult population. Salud Publica Mex. 2016;58(6):617–628.
- Jansen EC, Marcovitch H, Wolfson JA, et al. Exploring dietary patterns in a Mexican adolescent population: a mixed methods approach. Appetite. 2020;147. DOI:10.1016/j.appet.2019.104542
- Grunau C. Bisulfite genomic sequencing: systematic investigation of critical experimental parameters. Nucleic Acids Res. 2002;29(13):65e- 65.
- Moran S, Arribas C, Esteller M. Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences. Epigenomics. 2016;8(3):389–399.
- Goodrich JM, Reddy P, Naidoo RN, et al. Prenatal exposures and DNA methylation in newborns: a pilot study in Durban, South Africa. Environ Sci Process Impacts. 2016;18(7):908–917.
- Aryee MJ, Jaffe AE, Corrada-Bravo H, et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363–1369.
- Fortin JP, Labbe A, Lemire M, et al. Functional normalization of 450k methylation array data improves replication in large cancer studies. Genome Biol. 2014;15(11). DOI:10.1186/s13059-014-0503-2
- Fortin JP, Triche TJ, Hansen KD. Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics. 2017;33(4):558–560.
- Chen YA, Lemire M, Choufani S, et al. Discovery of cross-reactive probes and polymorphic CpGs in the Illumina Infinium HumanMethylation450 microarray. Epigenetics. 2013;8(2):203–209.
- Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13(1). DOI:10.1186/1471-2105-13-86
- Houseman EA, Molitor J, Marsit CJ. Reference-free cell mixture adjustments in analysis of DNA methylation data. Bioinformatics. 2014;30(10):1431–1439.
- Koike N, Yoo SH, Huang HC, et al. Transcriptional architecture and chromatin landscape of the core circadian clock in mammals. Science. 2012;(80–). DOI:10.1126/science.1226339
- Andreani TS, Itoh TQ, Yildirim E, et al. Genetics of circadian rhythms. Sleep Med Clin. 2015;10(4):413–421.
- Langmead B, Trapnell C, Pop M, et al. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009;10(3). DOI:10.1186/gb-2009-10-3-r25
- Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics. 2009;25(9):1105–1111.
- Trapnell C, Hendrickson DG, Sauvageau M, et al. Differential analysis of gene regulation at transcript resolution with RNA-seq. Nat Biotechnol. 2013;31(1):46–53.
- Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol. 2010;11(3). DOI:10.1186/gb-2010-11-3-r25
- de Onis M, Onyango AW, Borghi E, et al. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007;85(9):660–667.
- Wu YUE, Goodrich JM, Dolinoy DC, et al. Accelerometer-measured physical activity, reproductive hormones, and DNA methylation. Med Sci Sports Exerc. 2020;52(3):598–607.