University of Southern California Dissertations and Theses

Air pollution, mitochondrial function, and growth in children

(USC Thesis Other)

Air pollution, mitochondrial function, and growth in children

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content AIR POLLUTION, MITOCHONDRIAL FUNCTION, AND GROWTH IN CHILDREN

By

Ashley Yi Song

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for Qualifying Examination for
DOCTOR OF PHILOSOPHY
(EPIDEMIOLOGY)

December 2019

Copyright 2019 Ashley Yi Song

Table of Contents

CHAPTER 1 Introduction................................................................................................... 4
1.1 Significance ............................................................................................................... 4
1.2 Dissertation Project ................................................................................................... 8
CHAPTER 2 Mitochondria and Mitochondrial DNA ...................................................... 14
2.1 Mitochondria: Origin, Structure, Morphology, and Dynamics ............................... 14
2.2 Mitochondrial Biogenesis ....................................................................................... 15
2.3 Mitochondrial Response to Oxidative Stress .......................................................... 19
2.4 Important Metabolic Pathways Inside Mitochondria .............................................. 21
CHAPTER 3 Prenatal Air Pollution, MtDNA Methylation and Copy Number, and Infant
Growth .............................................................................................................................. 27
3.1 Abstract ................................................................................................................... 27
3.2 Mitochondrial DNA Methylation ............................................................................ 28
3.3 Proposed Study ........................................................................................................ 30
3.4 Results ..................................................................................................................... 37
3.5 Discussion ............................................................................................................... 47
CHAPTER 4 Prenatal Air Pollution, Mitochondrial-Associated MicroRNA, and Infant
Growth .............................................................................................................................. 70
4.1 Abstract ................................................................................................................... 70
4.2 Mitochondrial-Associated MiRNAs ....................................................................... 71
4.3 Proposed Study ........................................................................................................ 75
4.4 Results ..................................................................................................................... 83
4.5 Discussion ............................................................................................................... 91
CHAPTER 5 Mitochondrial Genetic Variants Are Associated with Growth in Children
......................................................................................................................................... 111
5.1 Abstract ................................................................................................................. 111
5.2 Mitochondrial Genetic Variations ......................................................................... 112
5.3 Proposed Study ...................................................................................................... 115
5.4 Results ................................................................................................................... 121
5.5 Discussion ............................................................................................................. 132

CHAPTER 6 Summary and Future Research ................................................................. 145
6.1 Summary and Conclusion ..................................................................................... 145
6.2 Implications and Future Directions ....................................................................... 149

4

CHAPTER 1
Introduction
1.1 Significance
In the United States, approximately 33% of children are overweight or obese [1, 2].
Due to its high prevalence and serious health consequences, such as cardiovascular disease,
type 2 diabetes and many cancers, obesity is considered to be one of the most challenging
public health issues of 21
st
century. Despite substantial policy efforts, there is no evidence
of a decline in childhood obesity prevalence in any age group [3]. Prevention strategies for
childhood obesity to date have been focused on changes in lifestyle, specifically diet and
physical activity. A growing body of literature has suggested that early life factors play a
crucial role in later risk of obesity in childhood [4, 5]. Exposure to chemical and
environmental hazards and stressors during pregnancy, infancy, and early childhood has
been associated with increased risk of obesity in childhood [4, 6, 7]. Yet, the mechanisms
remain largely unknown. Therefore, understanding of the mechanisms by which early life
environment contribute to early childhood growth and later childhood obesity is crucial
and necessary for developing prevention strategies for childhood obesity.
Mitochondria have been identified as one of the key players in the regulation of
early development and metabolism, as mitochondria are the primary energy producers of
adenosine-5’-triphosphate (ATP) via oxidative phosphorylation (OXPHOS).
Dysfunctional mitochondria have been implicated in various metabolic disorders including
obesity and type 2 diabetes [8-10]. Human mitochondria contain a small independent
circular genome of 16.5 kb that includes genes encoding for 13 proteins, 22 transfer RNAs

5

(tRNAs), and 2 ribosomal RNAs (rRNAs) [11]. Mitochondrial DNA (mtDNA) lacks
histones, chromatin structure, introns and has insufficient repair mechanisms, making it
susceptible to reactive oxygen species (ROS)-induced damage [12]. Changes in mtDNA
genetics including mitochondrial oxidative damage, DNA copy number, and DNA
mutations have been studied in relation to environmental exposure and disease outcomes.
Recent studies have shown that particulate matter (PM) can penetrate cells and damage the
mitochondria, including disruption of structure and function and altered mtDNA copy
number [13, 14]. Maternal stress and prenatal psychological functioning was also found to
be associated with decreased mtDNA copy number in placenta [15]. However, because of
their non-Mendelian inheritance, many mitochondrial diseases cannot be fully understood
solely by genetic studies. Epigenetic regulation of mitochondrial DNA adds another layer
of regulation that may be susceptible to environmental exposure and disease risk.
Emerging evidence has shown that mitochondria undergo epigenetic regulation
through mechanisms that are similar to nuclear DNA [16, 17]. The discovery of
mitochondrial DNA methyltransferase 1 (mtDNMT1) has confirmed the link between
oxidative stress and mitochondrial functions and provided new evidence that epigenetic
regulation of the mitochondrial genome by nuclear-encoded translocated mtDNMT1
relative to mitochondrial dysfunction [16]. Bellizzi et al. [18] confirmed that mtDNA is
methylated in the promoter region of the mitochondria displacement loop (D-loop), which
is required for transcription initiation. Moreover, ambient air pollutants have been reported
to be associated with placental mtDNA methylation in newborns as well as blood mtDNA
methylation in the elderly [19, 20]. Despite this recent interest in mtDNA methylation, the

6

effects of mtDNA methylation on early-life health outcomes in response to environmental
exposures have not been fully explored.
Non-coding RNA, and particularly, microRNA (miRNA), represent another
epigenetic mechanism that is involved in regulation of messenger RNAs of protein-coding
genes and control of gene expression at a post-transcriptional level. Studies have shown
that some miRNAs, derived from the nuclear genome, can translocate from the nucleus to
mitochondria and are likely to contribute to regulation of gene expression and various
metabolic pathways such as tricarboxylic acid cycle, electron transport chain, and lipid
metabolism. In addition, the discovery of mitochondrial miRNAs (mitomiRs), unique sets
of miRNAs that can localize in mitochondria, has raised the possibility of mitochondrial
RNA synthesis. Currently, little is known about the effect of air pollution on expression of
miRNAs targeting mitochondrial genes and its potential impact on development of
metabolic diseases in childhood. A growing literature suggests that miRNAs can respond to
environmental exposures, including air pollution, smoking, metals, and chemicals such as
endocrine disruptors [21, 22]. In vitro studies have shown that diesel exhaust particles (DEP),
the largest source of emitted airborne PM, can disrupt expression patterns of miRNAs in
human airway epithelial cells, indicating that DEP exposure may modify cellular processes
in response to inflammatory exposure and potentially lead to the development of a pathologic
disease state [23]. In addition, altered miRNA expression profiles have been observed in
various populations in response to air pollution exposure in population-based studies [24-
28]. In particular, a recent study in the ENVIRONAGE birth cohort reported an association
between prenatal exposure to PM2.5 and placental miRNA expression suggesting that in utero
PM2.5 affects miRNAs expression and its target genes [28].

7

Mitochondria have been largely studied by investigating their biochemistry and
morphology, which have been implemented by genetic studies. Mitochondrial genetics
studies have been focused on three main areas including the unique pattern of mtDNA
inheritance, mtDNA mutations yielding heteroplasmy, and high mutation rate due to high
oxidative stress. Genetic variation in mtDNA has been associated with congenital
anomalies at birth and metabolic diseases in both children and adults [29-34]. Genetic
variants in the mtDNA control region have been associated with metabolic phenotypes in
various populations [32, 33]. The control region of mtDNA is the most polymorphic region
of the human mtDNA genome and contains the origin of replication and transcription of
mtDNA. A common mitochondrial variant T16189C within the control region is associated
with type 2 diabetes in Caucasians and Asians, suggesting that the presence of T16189C
variant may alter mtDNA biosynthesis and affect mtDNA replication and transcription [35-
38]. In addition, it is known that mtDNA variation is under significant selection and
adaption to environment, leading to haplogroups as a result of accumulation of mutations
through maternal linkages[39]. Disease susceptibility may be captured by haplogroups,
which represent the geographic origin of populations and may contribute to the differences
in disease prevalence among racial/ethnic groups [39-41]. Emerging evidence also
suggests that genetic variation in mtDNA may modify the relationship between air
pollution exposure and various health endpoints in humans [42].

8

1.2 Dissertation Project
For my dissertation project I investigated the association between air pollution
exposure, mitochondrial function, and growth in infants and children. The mitochondrial
function studied include mitochondrial DNA methylation, patterns of miRNA expression
targeting mitochondrial genes, and genetic variation in mitochondrial genome.
The titles of my three chapters are: 1) association between prenatal exposure to air
pollution, epigenetic variations in mitochondrial DNA, and infant growth; 2) association
between prenatal exposure to air pollution, expression of mitochondria-associated miRNA
in maternal blood and cord blood, and newborn outcomes; 3) mitochondrial genetic
variants are associated with growth in children.
The first chapter of my dissertation is to establish the association between prenatal
exposure to air pollution and mitochondrial DNA methylation as well as the association
between mitochondrial DNA methylation and infant growth. The project is designed to
better understand how environmental exposure such as air pollution may alter
mitochondrial DNA methylation and in turn influence newborn birthweight and postnatal
growth. To accomplish this, I utilized data from Maternal and Child Health Study
(MACHS), which is designed to study DNA methylation in mixed cell population and to
prospectively examine the relationship between environmental obesogens and birthweight
and growth trajectories in infants. The results from the first project revealed that prenatal
exposure to traffic-related air pollution was positively associated with 12s rRNA (MT-
RNR1) methylation in peripheral blood mononuclear cells (PBMCs); ambient air pollution
was associated with displacement loop (D-loop) methylation in CD14+ cells; genetic
variations in mitochondrial DNA may modify the association between mitochondrial DNA

9

(mtDNA) methylation and birthweight. These findings suggest that air pollution alters the
methylation patterns of mtDNA and such an association together with genetic variation
may contribute to fetal and postnatal growth.
For the second chapter, I examined the relationship between in utero air pollution
exposure and expression patterns of microRNAs (miRNA) that target mitochondrial genes
or nuclear genes known to have mitochondrial functions as well as the relationship between
the mitochondria-associated miRNAs and newborn outcomes. To address the aims, I
utilized already collected data and blood samples from the ongoing Maternal and
Development Risks from Environmental and Social Stressors (MADRES) cohort
participants. MADRES is designed to study the impact of multiple pre- and postpartum
environmental exposures and psychosocial stressors on maternal and child health,
particularly obesity risk in low-income, urban minority communities in Los Angeles. We
observed significant associations of miR-23b, miR-29a, miR-199a, and miR-122 with NO2
and PM2.5 during third trimester. Sensitive windows of air pollution exposure in
relationship to acute changes of miRNA expression were also identified using distributed
lag models. This study provides evidence for the role that mitochondrial-related miRNAs
play in response to air pollution and in early life growth, and increases our knowledge of
the epigenetic mechanisms involving mitochondria and its susceptibilities to
environmental exposure and to the disease risk.
The third chapter of my dissertation project is to examine the relationship between
genetic variants in mitochondrial genome and childhood obesity risk. The project is aimed
to identify mitochondrial single nuclear polymorphisms (mtSNPs) that may influence body
mass index (BMI) in children and adolescents. This project is nested in the Children’s

10

Health Study (CHS), which has recruited and followed up children for over 20 years in 13
Southern California communities and collected extensive data on air pollution exposure,
health outcomes, and DNA and genotyping information. I investigated the relationship
between mtDNA genetic variants and BMI of children, and identified one mtSNP located
in mitochondriallly encoded cytochrome b (MT-CYB) gene significantly associated with
BMI growth over a 10-year timespan and five mtSNPs associated with BMI at age 10-18
in the primary CHS population. We then sought to replicate our findings in a separate
population of children from EVE Consortium on the Genetics of Asthma. We were able to
observe a significant association between one of the mtSNPs identified in the primary
population and BMI in EVE children at age 5-10. These findings of this research may
highlight the importance of mtDNA and its contribution to the risk of BMI growth in
children and provide evidence that genetic variants in the mitochondrial genome may play
a role in mitochondrial function and in turn influence metabolism.

11

Chapter 1: References
1. Ogden CL, Carroll MD, Kit BK, Flegal KM: Prevalence of childhood and adult
obesity in the United States, 2011-2012. JAMA 2014, 311(8):806-814.
2. Ogden CL, Carroll MD, Lawman HG, Fryar CD, Kruszon-Moran D, Kit BK, Flegal
KM: Trends in Obesity Prevalence Among Children and Adolescents in the United
States, 1988-1994 Through 2013-2014. JAMA 2016, 315(21):2292-2299.
3. Skinner AC, Perrin EM, Skelton JA: Prevalence of obesity and severe obesity in
US children, 1999-2014. Obesity (Silver Spring) 2016, 24(5):1116-1123.
4. Reilly JJ, Armstrong J, Dorosty AR, Emmett PM, Ness A, Rogers I, Steer C,
Sherriff A: Early life risk factors for obesity in childhood: cohort study. BMJ 2005,
330(7504):1357.
5. Taveras EM, Gillman MW, Kleinman KP, Rich-Edwards JW, Rifas-Shiman SL:
Reducing racial/ethnic disparities in childhood obesity: the role of early life risk
factors. JAMA Pediatr 2013, 167(8):731-738.
6. Toledo-Corral C AT, Habre R, Berhane K, Lurmann F, Weigensberg M, Goran M,
Gilliland F: Effects of air pollution exposure on glucose metabolism in Los Angeles
minority children. Pediatric Obesity 2016.
7. Oken E, Levitan EB, Gillman MW: Maternal smoking during pregnancy and child
overweight: systematic review and meta-analysis. Int J Obes (Lond) 2008,
32(2):201-210.
8. Rolo AP, Gomes AP, Palmeira CM: Regulation of mitochondrial biogenesis in
metabolic syndrome. Curr Drug Targets 2011, 12(6):872-878.
9. Rolo AP, Palmeira CM: Diabetes and mitochondrial function: role of
hyperglycemia and oxidative stress. Toxicol Appl Pharmacol 2006, 212(2):167-
178.
10. Nunnari J, Suomalainen A: Mitochondria: in sickness and in health. Cell 2012,
148(6):1145-1159.
11. Anderson S, Bankier AT, Barrell BG, de Bruijn MH, Coulson AR, Drouin J, Eperon
IC, Nierlich DP, Roe BA, Sanger F et al: Sequence and organization of the human
mitochondrial genome. Nature 1981, 290(5806):457-465.
12. Lee H WY: Mitochondrial role in life and death of the cell. Journal of Biomedical
Science 2000, 7(1).
13. Delgado-Buenrostro NL, Freyre-Fonseca V, Cuellar CM, Sanchez-Perez Y,
Gutierrez-Cirlos EB, Cabellos-Avelar T, Orozco-Ibarra M, Pedraza-Chaverri J,
Chirino YI: Decrease in respiratory function and electron transport chain induced
by airborne particulate matter (PM10) exposure in lung mitochondria. Toxicol
Pathol 2013, 41(4):628-638.
14. Hou L, Zhang X, Dioni L, Barretta F, Dou C, Zheng Y, Hoxha M, Bertazzi PA,
Schwartz J, Wu S et al: Inhalable particulate matter and mitochondrial DNA copy
number in highly exposed individuals in Beijing, China: a repeated-measure study.
Part Fibre Toxicol 2013, 10:17.
15. Brunst KJ, Sanchez Guerra M, Gennings C, Hacker M, Jara C, Bosquet Enlow M,
Wright RO, Baccarelli A, Wright RJ: Maternal lifetime stress and prenatal
psychological functioning are associated with decreased placental mitochondrial
DNA copy number in the PRISM study. Am J Epidemiol 2017.

12

16. Shock LS, Thakkar PV, Peterson EJ, Moran RG, Taylor SM: DNA
methyltransferase 1, cytosine methylation, and cytosine hydroxymethylation in
mammalian mitochondria. Proc Natl Acad Sci U S A 2011, 108(9):3630-3635.
17. Bellizzi D, D'Aquila P, Scafone T, Giordano M, Riso V, Riccio A, Passarino G:
The control region of mitochondrial DNA shows an unusual CpG and non-CpG
methylation pattern. DNA Res 2013, 20(6):537-547.
18. Bellizzi D DAP, Scafone T, Giordano M, Riso V, Roccio A, Passarino G: The
control region of mitochondrial DNA shows an unusual CpG and non-CpG
methylation pattern. DNA Research 2013, 20(6).
19. Janssen B BH, Gyselaers W, Lefebvre W, Baccarelli A, Nawrot T: Placental
mitochondrial methylation and exposure to airborne particulate matter in the early
life environment: An ENVIRONAGE birth cohort study. Epigenetics 2015, 10(6).
20. Byun H PT, Motta V, Hou L, Nordio F, Apostoli P, Bertazzi P, Baccarelli A: Effects
of airborne polutants on mitochondrial DNA methylation. Particle and Fibre
Toxiocology 2013, 10(18).
21. Vrijens K, Bollati V, Nawrot TS: MicroRNAs as potential signatures of
environmental exposure or effect: a systematic review. Environ Health Perspect
2015, 123(5):399-411.
22. Li Q, Kappil MA, Li A, Dassanayake PS, Darrah TH, Friedman AE, Friedman M,
Lambertini L, Landrigan P, Stodgell CJ et al: Exploring the associations between
microRNA expression profiles and environmental pollutants in human placenta
from the National Children's Study (NCS). Epigenetics 2015, 10(9):793-802.
23. Jardim MJ, Fry RC, Jaspers I, Dailey L, Diaz-Sanchez D: Disruption of microRNA
expression in human airway cells by diesel exhaust particles is linked to
tumorigenesis-associated pathways. Environ Health Perspect 2009, 117(11):1745-
1751.
24. Bollati V, Marinelli B, Apostoli P, Bonzini M, Nordio F, Hoxha M, Pegoraro V,
Motta V, Tarantini L, Cantone L et al: Exposure to metal-rich particulate matter
modifies the expression of candidate microRNAs in peripheral blood leukocytes.
Environ Health Perspect 2010, 118(6):763-768.
25. Levanen B, Bhakta NR, Torregrosa Paredes P, Barbeau R, Hiltbrunner S, Pollack
JL, Skold CM, Svartengren M, Grunewald J, Gabrielsson S et al: Altered
microRNA profiles in bronchoalveolar lavage fluid exosomes in asthmatic patients.
J Allergy Clin Immunol 2013, 131(3):894-903.
26. Hou L, Barupal J, Zhang W, Zheng Y, Liu L, Zhang X, Dou C, McCracken JP,
Diaz A, Motta V et al: Particulate Air Pollution Exposure and Expression of Viral
and Human MicroRNAs in Blood: The Beijing Truck Driver Air Pollution Study.
Environ Health Perspect 2016, 124(3):344-350.
27. Fossati S, Baccarelli A, Zanobetti A, Hoxha M, Vokonas PS, Wright RO, Schwartz
J: Ambient particulate air pollution and microRNAs in elderly men. Epidemiology
2014, 25(1):68-78.
28. Tsamou M, Vrijens K, Madhloum N, Lefebvre W, Vanpoucke C, Nawrot TS: Air
pollution-induced placental epigenetic alterations in early life: a candidate miRNA
approach. Epigenetics 2016:0.

13

29. Parle-McDermott A, Pangilinan F, O'Brien KK, Mills JL, Magee AM, Troendle J,
Sutton M, Scott JM, Kirke PN, Molloy AM et al: A common variant in MTHFD1L
is associated with neural tube defects and mRNA splicing efficiency. Hum Mutat
2009, 30(12):1650-1656.
30. Minguzzi S, Selcuklu SD, Spillane C, Parle-McDermott A: An NTD-associated
polymorphism in the 3' UTR of MTHFD1L can affect disease risk by altering
miRNA binding. Hum Mutat 2014, 35(1):96-104.
31. Hefti E, Quinones-Lombrana A, Redzematovic A, Hui J, Blanco JG: Analysis of
mtDNA, miR-155 and BACH1 expression in hearts from donors with and without
Down syndrome. Mitochondrial DNA A DNA Mapp Seq Anal 2016, 27(2):896-903.
32. Weng SW, Lin TK, Wang PW, Chen SD, Chuang YC, Liou CW: Single nucleotide
polymorphisms in the mitochondrial control region are associated with metabolic
phenotypes and oxidative stress. Gene 2013, 531(2):370-376.
33. Palmieri VO, De Rasmo D, Signorile A, Sardanelli AM, Grattagliano I, Minerva F,
Cardinale G, Portincasa P, Papa S, Palasciano G: T16189C mitochondrial DNA
variant is associated with metabolic syndrome in Caucasian subjects. Nutrition
2011, 27(7-8):773-777.
34. Flaquer A, Heinzmann A, Rospleszcz S, Mailaparambil B, Dietrich H, Strauch K,
Grychtol R: Association study of mitochondrial genetic polymorphisms in
asthmatic children. Mitochondrion 2014, 14(1):49-53.
35. Park KS, Chan JC, Chuang LM, Suzuki S, Araki E, Nanjo K, Ji L, Ng M, Nishi M,
Furuta H et al: A mitochondrial DNA variant at position 16189 is associated with
type 2 diabetes mellitus in Asians. Diabetologia 2008, 51(4):602-608.
36. Parker E, Phillips DI, Cockington RA, Cull C, Poulton J: A common mitchondrial
DNA variant is associated with thinness in mothers and their 20-yr-old offspring.
Am J Physiol Endocrinol Metab 2005, 289(6):E1110-1114.
37. Aral C, Akkiprik M, Caglayan S, Atabey Z, Ozisik G, Bekiroglu N, Ozer A:
Investigation of relationship of the mitochondrial DNA 16189 T>C polymorphism
with metabolic syndrome and its associated clinical parameters in Turkish patients.
Hormones (Athens) 2011, 10(4):298-303.
38. Casteels K, Ong K, Phillips D, Bendall H, Pembrey M: Mitochondrial 16189
variant, thinness at birth, and type-2 diabetes. ALSPAC study team. Avon
Longitudinal Study of Pregnancy and Childhood. Lancet 1999, 353(9163):1499-
1500.
39. Wallace DC: Mitochondrial DNA variation in human radiation and disease. Cell
2015, 163(1):33-38.
40. Mitchell S GR, Brown-Gentry K, Pendegrass S, Murdock D, Crawford D:
Characterization of mitochondrial haplogroups in a large population-based sample
from the United States. Human Genetics 2014, 133(7).
41. Salas A, Amigo J: A reduced number of mtSNPs saturates mitochondrial DNA
haplotype diversity of worldwide population groups. PLoS One 2010, 5(5):e10218.
42. Wittkopp S SN, Tjoa T, Gillen D, Daher N, Shafer M, Schauer J, Sioutas C, Delfino
R: Mitochondrial genetic background modifies the relationship between traffic-
related air pollution exposure and systemic biomarkers of inflammation. PLoS ONE
2013, 8(5).

14

CHAPTER 2
Mitochondria and Mitochondrial DNA
2.1 Mitochondria: Origin, Structure, Morphology, and Dynamics
The term “mitochondrion” originated from the Greeks word grains “chondria” and
filaments “mitos” reflecting the typical shapes of mitochondria first identified by light
microscopy more than 120 years ago [1]. Hypotheses proposing that mitochondria and
plastids evolved from free-living bacteria have been around since their discovery. The
detection of DNA in mitochondria as well as the generalization that all mitochondria in all
eukaryotes contained DNA greatly bolstered the hypothesis of endosymbiotic origin of
mitochondria [1, 2]. However, major progress in the characterization and understanding
of structure and morphology was not possible until the techniques for electron microscopy
and specimen preparation were perfected [1]. Mitochondria contains two membranes that
separate four distinct compartments, the outer member, intermembrane space, inner
member and the matrix (more). The inner membrane is highly folded into cristae, where
the complexes of electron transport chain and ATP synthase that control the basics of
cellular metabolism are housed [3]. It was discovered that the number of mitochondria per
cell varies significantly by cell types and tissues, which led to the consideration of
mitochondrial biogenesis in the context of cell differentiation and specialization. Individual
mitochondria can change their shape and size continuously through fission, fusion, and
other undefined processes such as branching/debranching [1, 4].

15

Among many types of mitochondrial shape and size changes, fission and fusion are
the most studied processes. Fission is postulated mechanism for mitochondrial
proliferation and relies on the activation of a dynamin-related protein-1 (Drp1) while fusion
is mediated through the action of at least three GTPases, known for their membrane
remodeling activities [1, 3]. Mitochondrial fusion contributes to the even distribution of
mitochondrial constituents, whereas fission increases the number of organelles and targets
damaged organelles to mitophagy. Emerging evidence suggests that the changes in shapes
and sizes of mitochondria regulated by specific fission and fusion proteins may have
important biological roles in the energy-producing activity of mitochondria [4]. Studies
have shown that mitochondrial fission and fusion are associated with mitochondrial
maintenance, energetic state of mitochondria, mitochondrial biogenesis and function.
Disrupted mitochondrial fusion and fission have been reported to result in impaired
mitochondrial function and cell function, which may eventually lead to apoptosis [3, 5]. In
addition, dysregulation of the fission and fusion has been reported to be associated with
many pathological conditions including neurodegeneration and metabolic diseases [4, 5].
2.2 Mitochondrial Biogenesis
2.2.1 Mitochondrial Genome
Human mitochondrial DNA (mtDNA) is a double-stranded, circular molecule of
16,569 bp and contains 37 genes that encodes for 13 polypeptides, 2 ribosomal RNAs
(rRNAs) and 22 transfer RNAs (tRNAs). Each human cell contains hundreds of
mitochondria and thousands of mtDNAs. In each cell, mtDNA only comprises 0.1-2 % of
the total DNA. The heavy (H; purine-rich) strand of mtDNA codes for 12 polypeptides, 2
rRNAs, and 14 tRNAs where the light (L; pyrimidine-rich) strand codes for a single

16

polypeptide and 8 tRNAs [6]. Seven mtDNA-encoded proteins represent subunits (ND1,
ND2, ND3, ND4, ND4L, ND5, and ND6) of Complex I, one (cytochrome b) is a subunit
of Complex III, three are subunits (CO1, CO2, and CO3) of Complex IV, and two are
subunits (ATP6 and ATP8) of Complex V [6]. A large proportion of mtDNA contains a
triple-stranded structure, called displacement loop or D-loop. The D-loop region has
evolved as major control site for mtDNA expression, containing the leading-strand origin
of replication and the major promoters for transcription [7].

Figure 2. 1. Map of the human mitochondrial genome (16,569 bp). The outer circle
represents the heavy (H) strand, containing the majority of the genes; the inner circle
represents the light (L) strand.

17

There are several unique features of mitochondrial genome. Mitochondrial genome
contains no introns and intergenetic sequences are absent or limited to a few bases. In
addition, mitochondrial genetic code differs from the universal genetic code, where a
simplified decoding mechanism is used. There is higher flexibility in paring of their base
of mitochondrial anticodons, allowing the 22 tRNAs species encoded by mtDNA being
sufficient to translate all 13 mitochondrial polypeptides [7]. Moreover, mtDNA is
maternally inherited exhibiting non-Mendelian inheritance pattern. Lastly, mitochondria
do not have histones. Studies have demonstrated that mtDNA is highly conserved and
packed into DNA-protein complexes in nucleoids.
2.2.2 Replication, Transcription and Translation of Mitochondrial DNA
The isolation and characterization of factors involved in mtDNA replication,
transcription, and translation have generally been hampered due to difficulties to obtain
highly purified mitochondrial preparations[1, 8]. For many years, maintenance of
mitochondrial genome was regarded as simplified compared to the nucleus. However,
major players have been identified and their functions in mtDNA have been subsequently
characterized in human cells.
The current established model of mtDNA replication maintains that leading-strand
replication initiates at a cluster of sites located downstream of the major light-strand
transcription promoter and proceeds unidirectionally with displacement of the parental
heavy strand to form a bubble (the D-loop) until approximately two-thirds of the closed
circular mtDNA has been copied [9, 10]. Moreover, currently there are several candidate
initiation sites for leading strand mtDNA synthesis [10]. Nothing is known about the
termination of mtDNA replication beyond its location. The only DNA polymerase found

18

in mitochondria is the DNA polymerase γ (Polγ) [11]. Polγ was said to replicate
mitochondrial DNA exclusively. The process and regulation of replication of mtDNA is
complex and there are still many gaps left in our knowledge. The transcription of mtDNA
occurs in both directions with the open reading frames on either strand. Heavy strand
transcription starts at the H-strand promoter (HSP) in the D-loop region and proceeds
counter-clockwise whereas light strand transcription starts at the L-strand promoter (LSP)
in the D-loop region and proceed clockwise [12]. Despite the close proximity of the HSP
and LSP, the initial in vitro transcription studies demonstrated that these elements are
functionally independent [13, 14].
2.2.5 Nuclear Genes Encoding Mitochondrial Proteins
Mitochondria consist approximately of 500-1,400 different proteins. Only a small
number of those proteins are encoded by mitochondria. Most mitochondrial translation
products are encoded by nuclear genes. Approximately 1,500 additional nuclear-encoded
genes have been identified to account for the remaining protein machinery responsible for
mitochondrial morphology, redox regulation and energetics [15]. The nuclear-encoded
mitochondrial proteins are translated on cytosolic ribosomes and selectively imported into
the mitochondrion through various mitochondrial protein import system [15]. Nuclear
DNA encodes for mitochondrial proteins including all four subunits of complex II, the
mitochondrial DNA polymerase γ (POLG) subunits, the mitochondrial RNA polymerase
components, the mitochondrial transcription factor (mtTFA), the mitochondrial ribosomal
proteins and elongation factors, and the mitochondrial metabolic enzymes. In addition, five
of the 10 nuclear-encoded subunits of complex IV (COX) have been identified to be tissue-
specific, developmental-specific, and species-specific (Oxidative phosphorylation). The

19

assembly of some of the complexes of mitochondria are believed to be partly encoded on
mtDNA and partly on nuclear DNA.

2.3 Mitochondrial Response to Oxidative Stress
2.3.1 Mitochondrial Oxidative Phosphorylation
In mammalian cells, mitochondrial respiratory chain oxidative phosphorylation
(OXPHOS) generates, under normal conditions, more than 80% of the ATP cell need.
OXPHOS, the process by which the mitochondrion generates energy through oxidation of
organic acids and fats with oxygen to create a capacitor across the mitochondrial inner
membrane [15, 16]. This proton gradient (∆P) is used as a source of potential energy to
generate adenosine triphosphate (ATP), transport substrates or ions, or produce heat.
OXPHOS encompasses five multi-polypepetide complexes I, II, III, IV and V. As a toxic
by-product of OXPHOS, the mitochondria generate most of the endogenous reactive
oxygen species (ROS). It has been hypothesized that the rate of mitochondrial and mtDNA
damage and thus decline is modulated by the extent of mitochondrial oxidative stress.
Mitochondrial ROS production, in turn, is increased by the availability of excess calories,
modulated by regional mtDNA genetic variation, and regulated by alterations in nDNA
expression of stress response genes [15, 16].
Studies have linked increased mitochondrial ROS production and decreased
mitochondrial OXPHOS to various biological processes and human diseases, such as
apoptosis and diabetes. The association of mitochondrial dysfunction with chronic diseases
may reflect, in part, the vulnerability of mitochondria to environmental stress.

20

2.3.2 Oxidative Stress and Mitochondrial Response
Oxidative stress occurs when the production of ROS is greater than the intrinsic
antioxidant defense. Augmented free radical generation in mitochondria has been regarded
as a result of two important phenomenon. The first one occurs when ATP production
exceeds energy demand of cell, and in this very situation, the electron transport gets
diminished [15]. On the other hand, impairment and uncoupling of specific respiratory
chain complexes under some stress conditions also favor the formation of free radicals
[17]. As mitochondria are the major generator of ROS, they often become the target of
elevated ROS exposure with deadly consequences, such as oxidative damage of
mitochondrial DNA.
Studies have investigated that the mechanism of cell damage induced by PM
exposure and the increase in ROS generation is one of the mechanisms involved in its
cytotoxicity [18]. A growing body of literature has shown that mitochondria are a central
target of PM-induced ROS. The mitochondrial alterations induced by PM are related to
cellular effects such as apoptosis, for example, depletion of mitochondrial DNA protected
cells from PM-induced apoptosis and these cells lack proteins involved in oxidative
phosphorylation [19]. In a previous study by Delgado-Buenrostro et al., results showed that
exposure to PM10 in isolated mitochondria from lung tissue caused enlarged intermembrane
spaces and shape alterations, disruption of cristae, and the decrease in dense granules [18].
Oxygraphic traces showed a concentration-dependent decrease in oxygen consumption and
RCI. In addition, mitochondrial membrane potential, ATP synthesis, and activity of
complexes II and IV showed an increase and decrease, respectively, after PM 10 exposure
[18].

21

Mitochondria are also susceptible to insult from multiple natural and synthetic
compounds that exert their toxicity by a) altering mtDNA integrity, b) inhibiting complexes
in the electron transport chain (ETC), c) modifying membrane potential, d) affecting Ca2+
transport, and e) activating proapoptotic signaling [20, 21]. Mitochondrial functions are
tightly integrated with cellular responses to damage in both mtDNA and nuclear DNA.
Given the significant generation of ROS during normal mitochondrial functions, it is not
surprising that base excision repair (BER), which repairs most oxidative DNA damage, is
a critical DNA repair pathway in the maintenance of mtDNA integrity [22, 23]. Other DNA
repair pathways that protect the nuclear genome— including mismatch repair as well as
repair of DNA double-strand breaks through homologous recombination or
nonhomologous end joining—may be active in mitochondria, but the specific roles for
these pathways, or the proteins involved in maintaining mtDNA stability, are not clear [24].
Nucleotide excision repair, which repairs damage resulting from many common
environmental genotoxicants including polycyclic aromatic hydrocarbons, mycotoxins,
and ultraviolet radiation, is not present in mitochondria [24].
2.4 Important Metabolic Pathways Inside Mitochondria
2.4.1 The Citric Acid Cycle (TCA Cycle)
The tricarboxylic acid (TCA) cycle is a central route for oxidative metabolism.
Besides being responsible for the production of NADH and FADH2, which fuel the
mitochondrial electron transport chain to generate ATP, the TCA cycle is also a robust
source of metabolic intermediates required for anabolic reactions [1]. In the catabolism of
carbohydrates, this begins with the transport of pyruvate from the cytosol into the
mitochondrion, and its subsequent oxidative decarboxylation to acetyl CoA by a soluble,

22

multienzyme pyruvate dehydrogenase complex, which is located in the mitochondrial
matrix [1, 25]. The oxidation of acetyl CoA is achieved by a cyclic process involving eight
catalytic steps. This process is known as either the citric acid or the tricarboxylic acid
(TCA) cycle. The respiratory substrates NADH and FADH 2 generated through the TCA
cycle are next oxidized in a process coupled to ATP synthesis. Electrons are transferred
from NADH and FADH 2 to oxygen via enzyme complexes located on the inner
mitochondrial membrane. Three of the electron carriers (complexes I, III and IV) are proton
pumps, and couple the energy released by electron transfer to the translocation of protons
from the matrix side to the external side of the inner mitochondrial membrane [25]. Energy
stored in the resulting proton gradient (i.e. the proton-motive force) is used to drive the
synthesis of ATP via the mitochondrial enzyme ATP synthetase (complex V) [1, 25]
2.4.2 The Fatty Acid Cycle
Fatty acid oxidation is another important metabolic activity located in the
mitochondria acid oxidation. Fatty acids undergo oxidative decarboxylation in the
mitochondrial matrix to give acetyl CoA, which is fed into the TCA cycle, and new acyl
CoA molecules that are successively shortened with each round of the cycle [25]. Under
certain conditions (e.g. fasting), acetyl CoA molecules are converted into ketones for use
as an alternative energy source. The beta-oxidation pathway involves four separate
enzymes that are soluble in the mitochondrial matrix and that function in a repetitive cycle.
With each round of the cycle, a fatty acid undergoes oxidative decarboxylation to produce
one molecule of acetyl CoA and one molecule of a new acyl CoA that is two carbons
shorter than the starting fatty acid [1, 25]. However, during conditions of prolonged fasting
and starvation, or in certain metabolic diseases (e.g. diabetes mellitus), the acetyl CoA

23

molecules generated by fatty acid oxidation are converted into ketones (e.g., ß-
hydroxybutyrate, acetoacetate and acetone) by enzymes also located in the mitochondrial
matrix [25]. These molecules are then transported through the blood to other tissues, such
as brain and heart, where they are used as an alternative energy source to glucose.
2.4.3 Urea cycle
Mitochondria are involved in the regeneration of cytosolic NAD (required for the
substrate-level phosphorylation step in glycolysis) and in the intracellular homeostasis of
inorganic ions such as calcium and phosphate. A wealth of studies show that mitochondria
also play an integral role in the cascade of intracellular events that lead to apoptosis, or
programmed cell death [25].

24

Figure 2. 2. Schematic of key mitochondrial metabolic pathways.(a) Carbohydrate
metabolism. Pyruvate produced from glycolysis undergoes oxidative decarboxylation to
acetyl CoA, which is then oxidized in an eight-step process known as the tricarboxylic
acid (TCA) cycle. (b) Fatty acid oxidation. (c) Urea cycle. Amino acid degradation
resulting in excretion of nitrogen as urea occurs partly in the mitochondrion. Reference:
Modica-Napolitano et al. 2007.

25

Chapter 2: References
1. Scheffler IE: Mitochondria, 2nd Edition edn: Hoboken, N.J.: Wiley-Liss 2008.
2. Gray MW: Mitochondrial evolution. Cold Spring Harb Perspect Biol 2012,
4(9):a011403.
3. McBride HM, Neuspiel M, Wasiak S: Mitochondria: more than just a powerhouse.
Curr Biol 2006, 16(14):R551-560.
4. Galloway CA, Lee H, Yoon Y: Mitochondrial morphology-emerging role in
bioenergetics. Free Radic Biol Med 2012, 53(12):2218-2228.
5. Kadenbach B: Mitochondrial Oxidative Phosphorylation: Nuclear-Encoded Genes,
Enzyme Regulation, and Pathophysiology: Springer New York Heidelberg
Dordrecht London; 2012.
6. Anderson S, Bankier AT, Barrell BG, de Bruijn MH, Coulson AR, Drouin J, Eperon
IC, Nierlich DP, Roe BA, Sanger F et al: Sequence and organization of the human
mitochondrial genome. Nature 1981, 290(5806):457-465.
7. Taanman JW: The mitochondrial genome: structure, transcription, translation and
replication. Biochim Biophys Acta 1999, 1410(2):103-123.
8. Marín-García J: Mitochondria and Their Role in Cardiovascular Disease: Springer
US; 2013.
9. Clayton DA: Mitochondrial DNA replication: what we know. IUBMB Life 2003,
55(4-5):213-217.
10. Holt IJ, Reyes A: Human mitochondrial DNA replication. Cold Spring Harb
Perspect Biol 2012, 4(12).
11. Bolden A, Noy GP, Weissbach A: DNA polymerase of mitochondria is a gamma-
polymerase. J Biol Chem 1977, 252(10):3351-3356.
12. Clayton DA: Transcription and replication of mitochondrial DNA. Hum Reprod
2000, 15 Suppl 2:11-17.
13. Wong TW, Clayton DA: Isolation and characterization of a DNA primase from
human mitochondria. J Biol Chem 1985, 260(21):11530-11535.
14. Wong TW, Clayton DA: DNA primase of human mitochondria is associated with
structural RNA that is essential for enzymatic activity. Cell 1986, 45(6):817-825.
15. Wallace DC: A mitochondrial paradigm of metabolic and degenerative diseases,
aging, and cancer: a dawn for evolutionary medicine. Annu Rev Genet 2005,
39:359-407.
16. Wallace DC: Mitochondrial DNA variation in human radiation and disease. Cell
2015, 163(1):33-38.
17. Khawaja NR, Carre M, Kovacic H, Esteve MA, Braguer D: Patupilone-induced
apoptosis is mediated by mitochondrial reactive oxygen species through Bim
relocalization to mitochondria. Mol Pharmacol 2008, 74(4):1072-1083.
18. Delgado-Buenrostro NL, Freyre-Fonseca V, Cuellar CM, Sanchez-Perez Y,
Gutierrez-Cirlos EB, Cabellos-Avelar T, Orozco-Ibarra M, Pedraza-Chaverri J,
Chirino YI: Decrease in respiratory function and electron transport chain induced
by airborne particulate matter (PM10) exposure in lung mitochondria. Toxicol
Pathol 2013, 41(4):628-638.

26

19. Mutlu EA, Engen PA, Soberanes S, Urich D, Forsyth CB, Nigdelioglu R, Chiarella
SE, Radigan KA, Gonzalez A, Jakate S et al: Particulate matter air pollution causes
oxidant-mediated increase in gut permeability in mice. Part Fibre Toxicol 2011,
8:19.
20. Shaughnessy DT, McAllister K, Worth L, Haugen AC, Meyer JN, Domann FE,
Van Houten B, Mostoslavsky R, Bultman SJ, Baccarelli AA et al: Mitochondria,
energetics, epigenetics, and cellular responses to stress. Environ Health Perspect
2014, 122(12):1271-1278.
21. Meyer JN, Leung MC, Rooney JP, Sendoel A, Hengartner MO, Kisby GE, Bess
AS: Mitochondria as a target of environmental toxicants. Toxicol Sci 2013,
134(1):1-17.
22. Mandal A, Mohindra V, Singh RK, Punia P, Singh AK, Lal KK: Mitochondrial
DNA variation in natural populations of endangered Indian feather-back fish,
Chitala chitala. Mol Biol Rep 2012, 39(2):1765-1775.
23. Maynard S, de Souza-Pinto NC, Scheibye-Knudsen M, Bohr VA: Mitochondrial
base excision repair assays. Methods 2010, 51(4):416-425.
24. Kazak L, Reyes A, Holt IJ: Minimizing the damage: repair pathways keep
mitochondrial DNA intact. Nat Rev Mol Cell Biol 2012, 13(10):659-671.
25. Modica-Napolitano JS, Kulawiec M, Singh KK: Mitochondria and human cancer.
Curr Mol Med 2007, 7(1):121-131.

27

CHAPTER 3
Prenatal Air Pollution, MtDNA Methylation and Copy Number, and Infant Growth
3.1 Abstract

Mitochondria, known to respond to environmental exposures that induce
oxidative stress, have been implicated in a variety of developmental diseases. We
investigated associations among prenatal air pollutant exposure, mitochondrial DNA
(mtDNA) methylation, mtDNA copy number, and mitochondrial SNPs (mtSNPs) in cord
blood of 178 newborns that were enrolled in the MACHS birth cohort. We also examined
the association of mtDNA methylation and copy number with birth weight and infant
growth. MtDNA methylation was measured using bisulfite-pyrosequencing in
mitochondrially encoded transfer RNA phenylalanine (MT-TF), 12S rRNA (MT-RNR1)
and the displacement loop (D-loop) control region in peripheral blood mononuclear cells
(PBMCs), CD4+ cells, and CD14+ cells. Relative mtDNA copy number was measured
by a quantitative real-time polymerase chain reaction. The associations, stratified by
mtSNPs and haplogroups, were assessed using mixed-effect models. A 1 SD increase in
non-freeway oxides of nitrogen (NOx) was associated with 0.16 [95% confidence interval
(CI): 0.00, 0.31] higher MT-RNR1 methylation in PBMCs. A 1 SD increase in PM2.5 was
associated with a 0.13 (95% CI: -0.03, -0.24) lower MT-RNR1 methylation and a 0.45
(95% CI: 0.08, 0.81) higher D-Loop methylation in CD14+ cells. MtSNP T16189C
modified the association between prenatal air pollutants and mtDNA methylation as well
as mtDNA methylation with birth weight. Prenatal air pollutant exposures may alter

28

mtDNA methylation and copy number and influence child growth by interaction with
mtSNP.
3.2 Mitochondrial DNA Methylation
The human mtDNA is made up of 16,569 bp with 435 CpG sites and 4747 cytosine
residues at non-CpG sites. Both cytosines at CpG sites and cytosines at non-CpG sites can
be methylated although with different frequency [1]. These CpG sites are almost evenly
dispersed, lacking the conventional ‘CpG islands’ that are found in nuclear DNA (nDNA).
Mitochondrial DNA methylation, by means of 5-methylcytosine and 5-
hydroxymethylcytosine, has been a topic of controversy in the last couple decades.
Recently, the presence of DNA methyltransferase enzyme 1 (mtDNMT1) in mitochondria
and its role in cytosine methylation has spurred renewed interest [2]. Bellizzi et al. [3]
confirmed that mtDNA is methylated in the promoter region of the mitochondria
displacement loop (D-loop), which is required for transcription initiation. Through a
mitochondrial targeting pre-peptide DNMT1 translocates into mitochondria sequences,
where it binds to mtDNA and modifies transcription of the mitochondrial genome [1].
Recent evidence suggests that not only nuclear DNA but also mitochondrial DNA
(mtDNA) may be subjected to epigenetic modification related to environmental exposure
and disease phenotypes. However, the role of mtDNA methylation and its clinical
significance has been long questioned. Recent studies have shown that mtDNA
methylation may have a role in controlling the expression of mitochondrial-encoded genes
similar to nuclear DNA. Considering the role of mitochondria in ATP production, any
epigenetic modification could potentially affect expression of respiratory chain complex
subunits, which in turn affect the energy production. Recently, methylated cytosines have

29

been found in the promoter region of the heavy strand located at the end of D-loop region
[4]. As mentioned in Chapter 2, D-loop region is considered to be implicated in the
processing of RNA primer during replication of the heavy strand [5]. mtDNA transcription
originates from single promoters, one per mtDNA strand [6]. The methylation at each CpG
dinucleotide in the mtDNA is likely to be more impactful than for the nuclear DNA and
thus more likely to undergo high turnover in response to cellular stress. In addition,
mitochondrial activity varies by cell and tissue type. Animal models have shown that
tissues with high demands for energy such as brain and heart may have different
methylation patterns and thus differential expression of mitochondrial genes [7, 8].
Environmental toxins have been directly or indirectly related to mitochondria and
mitochondrial function. As previously mentioned in Chapter 2, mtDNA lacks histones,
chromatin structure, introns and has insufficient repair mechanisms, making it susceptible
to reactive oxygen species (ROS)-induced damage. Oxidative stress in mitochondria can
be generated from environmental toxins including air pollutants, smoke, temperature, and
other factors. Ambient air pollutants have been reported to be associated with placental
mtDNA methylation in newborns as well as blood mtDNA methylation in the elderly [9,
10]. Despite this recent interest in mtDNA methylation, the effects of mtDNA methylation
on early-life health outcomes in response to environmental exposures have not been fully
explored.
MtDNA methylation assessment raises many challenges. Current mitochondrial
epigenetic studies are limited to measuring 5-methylcytsine and 5-hydroxymethcytosine at
CpG or non-CpG sites in mtDNA using techniques that are similar to those used for nuclear
DNA methylation measurement. One adapted method for mtDNA methylation

30

measurement is to use DNA from buffy coat cells extracted by a conventional method for
isolating DNA, which may result in a mixture of nuclear DNA and mtDNA. Many
homologous DNA sequences may be present in the mitochondrial genome and nuclear
genome. The presence of these nuclear-mitochondrial sequences (Numts), or pseudogenes
in the nuclear genome that do not transcribe, may be often confused with those originating
from mtDNA [8]. In addition, it is possible that every mitochondrial DNA copy may have
a different methylation pattern due to their mutation rate. Because there are multiple copies
of mtDNA in each mitochondrion, mutation rate may vary from 1 to 100%. It is generally
expected that mtDNA methylation levels would be lower than nuclear DNA methylation
levels. A recent gene-specific analysis by bisulfite-pyrosequencing showed that mtDNA
methylation in human buffy coat ranging between 3% and 12% [8]. The lower level
methylation of mtDNA is likely due to unique features of mtDNA, previously mentioned
in Chapter 2.
3.3 Proposed Study
Given the important role of mitochondria in metabolism and the suggested
associations between air pollutants and postnatal growth and child development, we
hypothesized that prenatal exposure to air pollution may be associated with mtDNA
methylation and copy number in cord blood in response to environmental-induced
oxidative stress and that haplogroups and mtSNPs may modify such relationships. Lastly,
we examined whether differences in mtDNA methylation level and copy number were
associated with birth weight and infant growth trajectories in the first six months of life.

31

3.3.1 Study Population
The Maternal and Child Health Study (MACHS) birth cohort consists of 178
mother-infant pairs recruited from the Los Angeles County + University of Southern
California (LAC+USC) Medical Center from September 2012 to August 2015. A subset of
infants was followed by phone and medical record abstraction until 6 months after birth
(n=52). Exclusion criteria included <18 years of age, HIV positive status, physical, mental,
or cognitive disabilities that prevented participation, current incarceration, or multiple
pregnancy. At study entry, a questionnaire was used to obtain data on socio-demographic
variables. Medical record information pertaining to pregnancy complications and delivery
was abstracted from electronic medical records. Infant medical records were requested
from pediatricians to obtain information on growth, well-child visits, developmental
milestones and other relevant health conditions. The institutional review board at Keck
School of Medicine, University of Southern California approved the study protocol, and
written informed consent was obtained from each pregnant woman prior to any testing.
3.3.2 Air Pollution Exposure Assessment
Traffic-related air pollution (TRAP) and ambient air pollution (AAP) exposures
were estimated based on participants’ residential addresses reported at birth. Participant
residence addresses were standardized and their locations were geocoded using the Tele
Atlas Geocoding Service. Nine-month average exposure to local traffic-related air
pollution was estimated using California Line-Source Dispersion Model (CALINE4) that
incorporated roadway geometry, traffic volumes, wind speed and direction, atmospheric
stability, mixing heights and vehicle emission rates [11]. Residential distance to major
freeways (FCC1) was classified as <500, ≥500 to <1000, ≥1000 to ≤1500, and >1500

32

meters. Residential distance to major roads (FCC3) was classified as <75, ≥75 to <150,
≥150 to ≤300, and >300 meters [12]. Prenatal ambient air pollution concentrations were
estimated using routine air monitoring data collected daily in California and available from
U.S. Environmental Protection Agency’s Air Quality System
(http://www.epa.gov/ttn/airs/airsaqs). Nine-month average ambient air pollution exposures
were spatially interpolated at each participant’s residence location using inverse distance-
squared weighting (IDW2) [13]. Data up to four nearest monitoring stations within 50 km
was included in each interpolation for NO2, O3, PM2.5, and PM10.
3.3.3 Measurements of mtDNA Methylation and Copy Number
A 15-cc cord blood sample for each participant was collected by hospital providers
in 2 EDTA tubes for plasma and DNA isolation. Blood samples were stored at room
temperature for up to 24 hours until transportation to the molecular biology laboratory at
the Southern California Environmental Health Sciences Center where samples were then
processed by one of two technicians. One EDTA tube was centrifuged at 1500 xg for 10
minutes. The buffy coat was collected and lysed for DNA extraction. Peripheral blood
mononuclear cells (PBMCs) were isolated from the 2
nd
EDTA tube by treating 5-10 ml of
whole blood with Huma Granulocyte Depletion Cocktail (STEMCELL Technologies) for
20 minutes. The blood was diluted with an equal volume of EasySep
TM
Buffer
(STEMCELL Technologies), mixed then overlaid into a 50 ml SepMate
TM
tube filled with
15 ml Lymphoprep (STEMCELL Technologies). The tube was centrifuged at 1200 xg for
10 minutes and the PBMCs layer was transferred into a 50 ml centrifuge tube using a
transfer pipette. The cells were washed with 20 ml of EasySep
TM
Buffer and centrifuged
for 8 minutes at 300xg. The cell pellet was resuspended in 2 ml of EasySep
TM
Buffer then

33

transferred into a 14 ml polystyrene tube. CD14+ monocytes cells and CD4+ T cells were
then separated by using the EasySep
TM
Human CD14 Selection Kit followed by the
EasySep
TM
Human CD4 Selection Kit according to manufacturer protocols. Lysed buffy
coat, CD14 and CD4 cells were suspended in Qiagen RLT+/2-mecaptoethanol and stored
at -20C for a couple of days before DNA extraction.
DNA was extracted from the buffy coat, CD14 and CD4 cells using the QIAamp
DNA Blood Midi Kit (Qiagen) and then bisulfite-treated using the EZ-96 DNA
Methylation Kit (Zymo Research) according to the manufacturer’s instructions.
Mitochondrial DNA methylation was assayed using Pyrosequencing. Methylation analyses
were performed by bisulfite-PCR. Pyrosequencing assays were performed using the
HotMaster Mix (Eppendorf, Hamburg, Germany) and the Pyrosequencing (PSQ) HS 96
Pyrosequencing System (Qiagen, Inc) as described in previous work (Byun et al. [14]). The
Pyrosequencing instrument includes built-in controls for assessing completion of bisulfite
conversion. The assays for measuring methylation in three mtDNA regions are the
mitochondrial encoded transfer RNA Phenylalanine (MT-TF), 12S ribosomal RNA (MT-
RNR1) and D-loop control region. The real-time polymerase chain reaction (PCR) and
pyrosequencing primers from Byun et al.[10] were used for measuring MT-TF methylation
only. For MT-RNR1 region, the new set of PCR primers was modified with biotin-labeled
forward primer, although both forward and reverse primer sequences were same as the
primers for MT-TF. One CpG in each of MT-TF and MT-RNR1, 3 CpGs from the D-loop
region, and one CpG from the light chain in D-loop region (LDLR2) [15] were analyzed.
The primer sequences and locations of targeted CpGs on the mitochondrial genome are
shown in Appendix 1. The correlation coefficient of inter-assay variation was less than 1

34

for all assays. For example, the coefficient of the MT-TF assay was 0.55 for PBMC cells.
CD4+ and CD14+ cells were sorted for 89 participants at the time of analysis, and thus
mtDNA methylation data for sorted cells exists for only 89 participants.
Relative mtDNA copy number was measured by a quantitative real-time
polymerase chain reaction (qPCR) assay that measures mtDNA copy number by
determining the ratio of mitochondrial copy number to single copy gene (human [beta]
globin: HBG) number in experimental samples relative to a reference obtained from Byun
et al [10]. The method is based on measurement of mitochondrial and gene quantities
expressed as cycle of threshold derived from a standard curve obtained from serial dilutions
of a reference DNA. The 178 samples were run in triplicate. The mean of the three
measurements of mitochondrial quantification was divided by the mean of the three
measurements of HBG quantification to calculate the mitochondrial / HBG ratio for each
sample. The measurements were considered acceptable if the standard deviations for the
threshold of the cycle was greater than 0.25, otherwise the samples were rerun. A control
DNA sample was included in each qPCR. The reference DNA, which is pooled DNA from
all 178 MACHS samples, was run in duplicates. A fresh standard curve, which ranged from
20 ng/μl to 0.625 ng/μl were generated in every qPCR run. The R
2
for each standard curve
was 0.99 or greater. The coefficient of variation for mtDNA copy number was 125%.
3.3.4 Assessments of Covariates
Maternal information, including maternal age, race/ethnicity, maternal education,
annual family income, maternal smoking status, and maternal body mass index (BMI) prior
to pregnancy calculated based on mother’s pre-pregnancy weight and height, was obtained
through baseline questionnaires completed at delivery. Gestational age at delivery, mode

35

of delivery, parity, infants’ birth weight, and season of birth (October-March, April-
September) were extracted from hospital medical record. Infant weight at 6 months of age
was extracted from pediatric medical record for 52 infants who had completed 6-month
well-child visits at analysis. Weight-for-age z-score was calculated based on the Centers
for Disease Control and Prevention 2000 growth chart at birth and at 6 months.
3.3.5 Statistical Analysis
Descriptive analyses were conducted to examine maternal and infant
characteristics, the distributions of TRAP and AAP exposure, and mtDNA methylation and
copy number of the entire cohort and the subsample. Spearman correlations were used to
study the pairwise correlation of air pollutants and mean levels of methylation between
adjacent CpGs in D-Loop and LDLR2. Because the loci in each region were highly
correlated, we used an average percent mtDNA methylation of three loci in D-Loop and
two loci in LDLR2 in all analyses. We also evaluated the correlation of mtDNA
methylation with mtDNA copy number as well as the correlation of mtDNA methylation
by cell types. Multiple linear mixed models were used to evaluate the association between
prenatal air pollution with percent mtDNA methylation and copy number in cord blood as
well as the association between percent mtDNA methylation and copy number in cord
blood with infant growth. Similar models were used to evaluate the association between
prenatal air pollution exposures with percent mtDNA methylation in CD14+ cells and
CD4+ cells. Plate number, a technical variable and indicator of potential batch effects, was
treated as a random effect in these models. Seven plates were run for methylation analysis.
Seventeen plates were run for mtDNA copy number measurement. Potential confounders
were retained in the final models if a 10% change in coefficient estimate was observed.

36

The following covariates were evaluated as potential confounders: maternal age,
gestational age at delivery, race/ethnicity, parity, family income, maternal education,
smoking status, maternal BMI, season of birth, year of birth, infant sex, and birth weight
of newborns. When evaluating the association between air pollution exposure and mtDNA
methylation, effect modification by mtSNPs was also examined, adjusting for multiple
testing at a false discovery rate (FDR) of 0.05. For the current analysis, one subject with
an extremely high mtDNA copy number was considered an outlier and removed. The final
models examining the association between traffic-related air pollution exposures with
mtDNA methylation and copy number adjusted for maternal age, gestational age,
race/ethnicity, parity, maternal smoking, and season of birth (n=178). The final models
examining the association between ambient air pollution exposures with mtDNA
methylation and copy number adjusted for maternal age, gestational age, race/ethnicity,
parity, maternal smoking, and year of birth (n=178). Multi-pollutant models were also
performed to further investigate for adjustment of a second pollutant. In order to account
for the unique feature of the mtDNA genome, that is, there are multiple copies of mtDNA
per mitochondrion and per cell, we further investigated the association between air
pollutants and mtDNA methylation additionally adjusting for mtDNA copy number. The
final models examining the association between mtDNA methylation and copy number
with infant growth adjusted for maternal age, gestational age, race/ethnicity, parity,
delivery type, infant sex, birth weight of newborns, and pre-pregnancy BMI (n=52).
Regression coefficients represent the absolute change in mtDNA methylation percent or
the relative change of mtDNA copy number for a one SD increase in the exposure. All tests

37

were two-sided with p-values less than 0.05 considered statistically significant. All
statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC).
3.4 Results
3.4.1 Participants’ Characteristics
Baseline characteristics of the 178 study participants and their average distribution
of TRAP and AAP exposures are shown in Table 1. No difference was observed between
study participants in the entire cohort and the 89 participants for whom we also had mtDNA
methylation levels measured in sorted CD14+ and CD4+ cells. Mothers were
predominantly Hispanic-White (90%) and had completed 12
th
grade or higher (71%) in the
entire cohort. Infants were 50% female; 32% of the infants were born via C-section. The
distribution of prenatal air pollutants was not statistically significantly different between
the entire cohort and the subsample except for O3 (Table 1). Nine-month pregnancy
averages of NO2, O3, and PM2.5 were moderately correlated with one other, whereas PM10
was not correlated with other ambient air pollutants (see Supplement Material, Table S1).
For 52 of the 178 participants, we had 6-month follow-up visit information including height
and weight. Those with a 6-month well-child follow-up were not significantly different
from the full cohort with respect to baseline characteristics (see Supplement Material,
Table S2).
Table 1. Maternal and infant characteristics from the Maternal and Child Health Study

Entire Cohort
(n=178)
Subsample
(n=89)

n (%) or mean
± SD
n (%) or mean
± SD
p-value
Maternal Characteristics
Age (mean, SD) 26.9 (6.5) 27.7 (70) 0.38

38

Entire Cohort
(n=178)
Subsample
(n=89)

n (%) or mean
± SD
n (%) or mean
± SD
p-value
BMI (mean, SD) 29.4 (7.7) 29.2 (6.0) 0.81
Race/ethnicity

Hispanic White 161 (90.4) 79 (88.8) 0.67
Other 17 (9.6) 10 (11.2)
Nulliparous 97 (48.9) 38 (42.7) 0.34
Education

Less than 12th grade 51 (28.8) 26 (29.6) 0.98
Completed grade 12 67 (37.9) 32 (36.4)
College or higher education 59 (33.3) 30 (34.1)
Family income (US $)

Less than 15,000 56 (31.5) 32 (36.0) 0.69
15,000 - 29,999 28 (27.0) 27 (30.3)
30,000 – 100,000 22 (12.3) 11 (12.3)
Don't know 52 (29.2) 19 (21.4)
Infant Characteristics

Gender

Male 94 (52.8) 49 (55.1) 0.73
Female 84 (47.2) 40 (44.9)
Gestational age at birth (weeks) 39 ± 1.7 38 ± 1.8 0.41
Birth weight (grams) 3246 ± 482.7 3229 ± 475.7 0.78
Mode of delivery

Vaginal delivery 120 (68.5) 54 (62.8) 0.47
C-section 55 (31.5) 32 (37.2)
Season of birth

October – March 83 (46.6) 52 (58.4) 0.07
April - November 95 (53.4) 37 (41.6)
Traffic-related air pollutants

Freeway NOx (ppb) 9.8 ± 8.3 9.7 ± 8.9 0.96
Non-Freeway NOx (ppb) 3.1 ± 1.5 3.0 ± 1.4 0.64
Total NOx (ppb) 12.8 ± 8.7 12.7 ± 9.2 0.90
Ambient air pollutants

NO2 (ppb) 19.0 ± 2.4 18.7 ± 2.7 0.43
O3 (ppb) 25.8 ± 2.7 26.8 ± 2.8 0.01
PM2.5 (μg/m
3
) 12.4 ±0.9 12.2 ± 1.0 0.13
PM10 (μg/m
3
) 30.8 ± 2.7 30.9 ± 2.5 0.68
Abbreviations: BMI, body mass index. Prenatal exposures to traffic-related and ambient
air pollution in the nine months prior to delivery are reported. Subsamples included
participants who had mtDNA methylation in CD4+ and CD14+ cells measured. P-values
for categorical characteristics derived using Pearson chi-square test (for 2 sample
comparison) and ANOVA test (for multi-group comparison) unless otherwise noted; p-
values for continuous characteristics derived using t-test.

39

Distribution of percent mtDNA methylation and copy number are also evaluated.
In general, MT-TF, MT-RNR1, D-Loop, and LDLR2 in PBMCs, CD14+ cells, and CD4+
cells were largely unmethylated. LDLR2 had the highest levels of methylation, with 10%
in PBMCs, 7% in CD14+ cells, and 7% in CD4+ cells. In the subsample of 89 subjects
with methylation in CD14+ cells and CD4+ cells measured, methylation levels of MT-TF,
MT-RNR1, D-Loop, and LDLR2 in PBMCs were not correlated with methylation levels of
MT-TF, MT-RNR1, D-Loop, and LDLR2 in CD14+ cells and CD4+ cells, respectively
(p>0.05).Mean mtDNA copy number was 1.6 with a standard deviation of 2.0. Those with
higher MT-TF methylation in PBMCs exhibited higher mtDNA copy number
(rspearman=0.34, p<0.001). There was no significant correlation of mtDNA copy number
with MT-RNR1 methylation, D-Loop methylation, and LDLR2 methylation in PBMCs.
3.4.2 Associations between Prenatal TRAP/AAP Exposures and mtDNA Methylation
Prenatal exposure to non-freeway NOx was positively associated with MT-RNR1
methylation in cord-blood but not in isolated CD14+ or CD4+ cells (Figure 1A). A 1 SD
(2 ppb) increase in non-freeway NOx was associated with 0.16 (95% CI: 0.00, 0.31) higher
MT-RNR1 methylation in cord blood. Living near major roads during pregnancy was also
associated with higher MT-RNR1 methylation in cord blood, consistent with the modeled
non-freeway NOx findings. Compared to those infants born to mothers who lived >300 m
from a major road during pregnancy, infants born to mothers who lived within 75 m of a
major road had 0.60 (95% CI: 0.16, 1.04) higher MT-RNR1 methylation in cord blood
(Figure 1B). Moreover, the trend test across categories of distance to a major road was
marginally significant (p=0.08) suggesting that the strength of the association increased as
distance to a major road decreased. Prenatal exposure to TRAP was not associated with

40

MT-TF, D-Loop, and LDLR2 methylation in cord blood, CD14+ cells, or CD4+ cells (see
Supplement Material, Figure S1, Figure S2).
Figure 1. Association between traffic-related air pollution (TRAP) exposures A) NOx and
B) distance to major freeways and roadways and MT-RNR1 methylation. Abbreviations:
MT-RNR1, 12S ribosomal RNA. Estimated exposures to prenatal air pollution correspond
to the 9-month average prior to delivery. Estimated percent difference in mtDNA
methylation are shown for a 1SD or one category difference in exposures. Distance to
major freeways (FCC1) and distance to major roadways (FCC3) are shown in meters.
Models adjusted for maternal age, gestational age, race/ethnicity, parity, and season of
birth. Plate numbers were included as random effects. The values for 1SD were 8 ppb for
freeway NOx, 2 ppb for non-freeway NOx, and 9 ppb for total NOx. Sample sizes for cord
blood, CD14+ and CD4+ cell methylation are 178, 89 and 89, respectively.

In contrast to TRAP, prenatal AAP exposure was associated with mtDNA
methylation in CD14+ cells, but not in cord blood and CD4+ cells (Figure 2, see
Supplement Material, Table S4, Table S5). For example, a one SD (3 μg/m
3
)

increase in
PM10 was associated with a 0.50 (95% CI:-0.15, -0.85) lower LDLR2 methylation in
CD14+ cells. Prenatal NO2 and O3 showed no association with mtDNA methylation in
CD14+ cells (Figure 2). Multi-pollutant models suggested that adjustment for a second
pollutant strengthened the association between PM2.5 and PM10 with D-loop and LDLR2
methylation in CD14+ cells even further (see Supplement Material, Table S6). For

41

example, a one SD (1 μg/m
3
) increase in PM2.5 associated with a 0.60 (95%: 0.05, 1.15)
higher D-loop methylation in CD14+ cells when adjusting for O3 and a 0.48 (95% CI: -
0.06, -0.90) lower LDLR2 methylation in CD14+ cells when adjusting for NO2 (see
Supplement Material, Table S6). We did not observe any substantial changes in the
associations between TRAP/AAP exposure and mtDNA methylation with further
adjustment of mtDNA copy number (data not shown), suggesting that the observed
associations are not confounded by mtDNA copy number.

Figure 2. Association between ambient air pollution (AAP) exposure and mtDNA
methylation in CD14+ cells from single-pollutant models. Abbreviations: MT-TF,
mitochondrially encoded transfer RNA phenylalanine; MT-RNR1, 12S ribosomal RNA; D-
loop, displacement loop region; LDLR, light chain in D-loop region. Estimated exposures
to prenatal air pollution correspond to the 9-month average prior to delivery. Estimated
percent difference in mtDNA methylation are shown for a 1SD difference in exposure.
Models adjusted for maternal age, gestational age, race, parity, and year of birth. Mixed
models were used with plate numbers included as random effects. The values for 1SD were
2 ppb for NO2, 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3 μg/m
3
for PM10. Sample size for
CD14+ cell methylation is 89.

42

3.4.3 Associations between Prenatal TRAP/AAP Exposures and mtDNA Copy Number
Non-freeway NOx was positively and significantly associated with mtDNA copy
number (Figure 3). A 1 SD (2 ppb) increase in non-freeway NOx was associated with a
0.24 (95% CI: 0.04, 0.44) higher mtDNA copy number (Figure 3). We did not observe a
statistically significant association between living near major roads and mtDNA copy
number. Prenatal AAP exposure was negatively associated with mtDNA copy number.
When adjusting for covariates, a lower mtDNA copy number of 0.24 (95% CI: -0.00, -
0.49) was observed for a one SD (1 μg/m
3
)

increase in PM10 (Figure 3).
Figure 3. Association between traffic-related air pollution (TRAP) and ambient air
pollution (AAP) exposure A) NOx, B) distance to major freeways and roadways, and C)
AAP exposure and mtDNA copy number. Estimated exposures to prenatal air pollution
correspond to the 9-month average prior to delivery. Estimated percent difference in
mtDNA methylation are shown for a 1SD or one category difference in exposure. Models
for traffic-related air pollutants were adjusted for maternal age, gestational age, race, sex,

43

parity, and season of birth. Models for ambient air pollutants adjusted for maternal age,
gestational age, race, parity, and year of birth. Plate numbers were included as random
effects. Mixed models were used with plate numbers included as random effects. The
values for 1SD were 8 ppb for freeway NOx, 2 ppb for non-freeway NOx, 9 ppb for total
NOx, 2 ppb for NO2, 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3 μg/m
3
for PM10. Sample size
is 178 for mtDNA copy number model.

3.4.4 Effect Modification by Haplogroups and mtSNPs
We also evaluated whether mitochondrial haplogroups and mtSNPs modify the
association between prenatal air pollution and mtDNA methylation and copy number in
PBMCs. Of the 178 participants, 69 participants were determined to belong to haplogroup
A, 35 belonged to haplogroup B, and 28 belonged to haplogroup C, whereas 46 were
inconclusive based on current haplogroup phylogenetic tree. We did not observe a
significant difference in effects of prenatal TRAP/AAP exposure on mtDNA methylation
and copy number among haplogroups (data not shown).
Of the 63 mtSNPs selected, 16 mtSNPs had an allele variant frequency >15% and
could be evaluated with sufficient sample size in each genotype. After FDR adjustment for
multiple testing of the 16 mtSNPs, one mtSNP T16189C remained statistically significant
when tested for the interaction between prenatal air pollutants and methylation and copy
number. There was no significant difference between subjects with the C allele of 16189
and subjects with the T allele of 16189 with respect to maternal and infant characteristics
(see Supplement Material, Table S7).
Stratified analysis suggested that associations between TRAP exposure and MT-
RNR1 methylation in cord blood as well as AAP exposure and LDLR2 methylation were
modified by mtSNP T16189C, though interaction terms were only marginally significant.
Specifically, a one SD (9 ppb) increase in total NOx was associated with a 0.50 (95% CI:

44

0.13, 0.86) higher MT-RNR1 methylation in subjects carrying the C allele and a 0.08 (95%
CI: -0.25, 0.09) lower MT-RNR1 methylation in subjects carrying the T allele (interaction
term p=0.09) (Figure 4). In addition, a 1 SD (2 ppb) increase in ambient NO2 was associated
with a 0.65 (95% CI:-1.59, 0.29) lower LDLR2 methylation in subjects carrying the C
allele and a 0.99 (95% CI: 0.28, 1.70) higher LDLR2 methylation in subjects carrying the
T allele (interaction term p=0.02) (Figure 4). A 1 SD (1 μg/m
3
) increase in PM2.5 was
associated with a 0.58 (95% CI:-1.58, 0.43) lower LDLR2 methylation in subjects carrying
the C allele and a 0.65(95% CI: -0.11, 1.42) higher LDLR2 methylation in subjects carrying
the T allele (interaction term p=0.01) (Figure 4). Associations between prenatal air
pollution and MT-TF methylation, D-Loop methylation, and mtDNA copy number were
not modified by T16189C (see Supplement Material, Table S8, Table S9).
Figure 4. Association between traffic-related air pollution (TRAP) and ambient air
pollution (AAP) exposure and mtDNA methylation and copy number in PBMCs by
T16189C A) NOx with MT-RNR1 methylation, B) AAP exposure with MT-RNR1

45

methylation, C) NOx with LDLR2 methylation, and D) AAP exposure with LDLR2
methylation. Abbreviations: MT-RNR1, 12S ribosomal RNA; LDLR, light chain in
displacement loop region. Estimated exposures to prenatal air pollution correspond to the
9-month average prior to delivery. Estimated effects in mtDNA methylation are shown for
a 1SD or one category difference in exposure. Models for traffic-related air pollutants were
adjusted for maternal age, gestational age, race, sex, parity, and season of birth. Models for
ambient air pollutants adjusted for maternal age, gestational age, race, parity, and year of
birth. Plate numbers were included as random effects. Pinteraction tests effect modification of
T16189C and adjusted for multiple testing at a false discovery rate (FDR) of 0.05. * marks
a pinteraction value <0.10. The values for 1SD were 8 ppb for freeway NOx, 2 ppb for non-
freeway NOx, 9 ppb for total NOx, 2 ppb for NO2, 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3
μg/m
3
for PM10. Sample sizes are 60 for C allele group and 118 for T allele group

3.4.5 Associations between mtDNA Methylation and Copy Number with Infant Growth
MtDNA methylation and copy number were further investigated with respect to
birth weight of newborns and their growth from birth to 6 months of age. No significant
associations between mtDNA methylation and copy number and infant birth weight and
their growth at 6 months were observed (see Supplement Material, Table S10). We did not
observe a significant association of T16189C with birth weight and infant growth at 6
months either. However, stratified analysis by T16189C revealed that a 1SD (0.5%)
increase in MT-TF methylation in PBMCs was associated with a 115 gram (95% CI: -226,
-5) lower birth weight in subjects carrying the C allele and a 42 gram (95% CI: -29, 113)
higher birth weight in subjects carrying the T allele (p-value for interaction =0.03, Table
2). We were unable to evaluate the stratum-specific association between mtDNA
methylation and infant growth at 6 months by T16189C due to the limited sample size
within each stratum.

46

47

3.5 Discussion
Results from this study demonstrate that prenatal TRAP and AAP exposures were
associated with mtDNA methylation and copy number changes in different cord blood cell
types. Furthermore, the strength of these associations differed by mtSNP T16189C
genotype in some cases. MtSNP T16189C genotype also modified the association between
mtDNA methylation level and infant birth weight. However, we did not observe an effect
of prenatal exposure to air pollution on mtDNA methylation and copy number stratified by
mitochondrial haplogroups.
Although the existence of mtDNA methylation has been a matter of debate for
many decades, the discovery of DNMT1 enzyme inside mitochondria and accumulating
evidence from studies of various human tissues firmly suggests that methylation of mtDNA
is a real biological phenomenon [2, 16]. However, the relevance of mtDNA methylation
and its clinical usefulness is still controversial. Recent studies have shown that mtDNA
methylation may have a role in controlling the expression of mitochondrially-encoded
genes in a similar way to nuclear DNA methylation [8]. Growing evidence suggests
mtDNA methylation is a novel and unexplored mitochondrial mechanism to cope with
changing environments [17]. Findings from this study add to a new body of literature that
suggests environmental exposures may alter mtDNA methylation and copy number, which
in turn may influence child health outcomes. Specific underlying mechanisms of action
remain to be determined.
Exposure to TRAP has been studied in association with gene-specific DNA
methylation and global DNA methylation of the nuclear genome [18-20]. Effects of TRAP
on the mitochondrial genome are far less studied. One previous study showed that black

48

carbon (BC), a combustion byproduct that often serves as a proxy for traffic-related air
pollutants, was associated with an increased mtDNA copy number in older men [21]. In
this study, we investigated the effect of prenatal TRAP exposure on mtDNA methylation
and copy number in newborns. We found that higher levels of NOx from major roadways
and living closer to major roadways were both positively associated with increased MT-
RNR1 methylation in PBMCs and increased mtDNA copy number, consistent with the
previous study despite the difference in participant age. TRAP exposure may activate
oxidative pathways through production of ROS, and mitochondria are known to be
particularly vulnerable to ROS-induced damage [21]. Accumulation of ROS during the
production of ATP through OXPHOS in mitochondria could lead to a cellular stress
response, apoptosis, and damage of mitochondria structure and function, which may be
partly reflected in mtDNA copy number [22]. In addition, many signaling pathways that
are involved in the regulation of mitochondrial biogenesis are activated in response to
excess production of ROS [23]. However, we only observed an association with non-
freeway NOx but not freeway or total NOx. It is possible that non-freeway NOx exposures
capture other neighbor characteristics that may introduce residual confounding.
Recent studies have also suggested an association between exposure to ambient
pollutants with methylation of mtDNA at MT-RNR1 and D-loop region [9, 10]. Janssen et
al. [9] observed a positive association between PM 2.5 exposure during gestation and MT-
RNR1 and D-loop methylation in placental tissue, whereas we found a positive association
between PM2.5 and MT-RNR1 and D-loop methylation in CD14+ cells of cord blood, but
not within PBMCs or CD4+ cells. Some may argue that increased methylation of mtDNA
in the presence of increased mtDNA copy number as reported in our study could simply

49

represent having more mtDNA in the sample. Therefore, we further adjusted for mtDNA
copy number in the analysis of air pollution exposure and mtDNA methylation association
and did not observe any significant changes in the results.
The MT-RNR1 gene encodes 12S rRNA, which is essential for the stability of the
small subunit of the mitochondrial ribosome. Studies in animal models have suggested that
disruption in 12s rRNA structure leads to impaired assembly of the mitochondrial ribosome
and abolished mitochondrial transcription [24]. This may be one mechanism through which
AAP impairs mitochondrial function. Together with increased mtDNA copy number,
increased methylation of mtDNA may represent a mechanism that maintains normal
cellular function and copes with increased respiratory demand for ROS clearance [10].
Furthermore, the 12S region of the mitochondrial genome also contains an open reading
frame that transcribes a newly discovered mitochondrial-derived peptide (MDP) known as
MOTS-c [25]. MOTS-c is a circulating MDP that acts as an exercise mimetic [26]. It
primarily targets the muscle cells to activate AMPK and prevent obesity caused by a high-
fat diet. Therefore, it is possible that prenatal AAP exposure may alter MOTS-c levels at
birth with the potential to influence infant growth.
In addition, given the role of the D-loop region in the regulation of mitochondrial
replication and transcription, an increase in methylation at the D-loop region may lead to a
decrease in mitochondrial biogenesis. Studies have demonstrated that the D-loop region is
highly susceptible to oxidative damage when compared with other regions of mtDNA and
suggest that persistent oxidative stress may alter the mtDNA replication rate, resulting in
decreased mitochondrial respiratory function [23, 27]. Furthermore, recent evidence shows
that the presence of cytosine methylation in the highly conserved sequences located at the

50

D-loop region may be implicated during the replication of mtDNA [2, 3]. Our findings of
higher methylation at D-loop region in association with prenatal air pollution exposure
suggest that oxidative stress induced by environmental exposure may result in a change in
mtDNA methylation and copy number. Many have suggested that mtDNA methylation
may affect the regulation of mitochondrial transcription through several proposed
mechanisms [17]. One of the mechanisms is that mtDNA methylation may affect the
binding of transcription factor A (TFAM) to the DNA either directly or indirectly. Some
have speculated that TFAM might bind to promotor region within the D-loop region and
prevent remethylation enabling transcription [17]. However, future in vitro studies are
warranted to support the hypothesis that higher methylation of mtDNA at D-loop may
impact transcription of mtDNA, subsequent mitochondrial gene expression, and thus could
have effects on fetal development.
Our study is one of the first to demonstrate an association between AAP exposure
during pregnancy and mtDNA methylation in CD14+ monocyte cells in cord blood. It is
known that the number of mitochondria varies greatly among different cell types. There is
speculation that tissue-specific and cell-type specific mitochondrial epigenetic patterns
may occur depending on the number of mitochondria and rates of cellular metabolism [8].
Animal studies have observed reduced mitochondrial function in circulating monocytes in
response to environmental-induced ROS [28]. In addition, mitochondrial monocytes also
showed increased apoptosis, decreased proliferation, and increased inflammatory cytokine
release in mice with point mutations and deletions in the mitochondrial genome [28]. This
evidence suggests that ROS-induced mtDNA damage may lead to mitochondrial
dysfunction in monocytes as well. However, the mechanism driving such an association

51

remains unclear. Further studies are warranted to evaluate the effect of cell-specific
mtDNA methylation on disease pathogenesis in response to environmental factors in larger
prospective studies, as this area of research has only just begun. Furthermore, we
acknowledge that mtDNA methylation and mtDNA copy number in various cell types may
be influenced by cellular contamination or differences in proportions of cell types,
including platelets. However, in a previous study by Janssen et al. [29], mtDNA content
did not correlate with blood platelets or white blood cells.
Although in utero air pollution exposure has been associated with adverse birth
outcomes, mechanisms through which prenatal air pollution exposures increase the risk of
adverse outcomes at birth have remained largely unknown. Recently, it has been
hypothesized that oxidative stress-induced mitochondrial damage and alteration of mtDNA
may play a role in mediating the association between prenatal air pollution exposure and
low birth weight [30]. Clement et al. [30] found that mtDNA content in placenta was
associated with reduced birth weight in a European birth cohort and 10% of the association
between prenatal NO2 exposure and low birth weight could be explained by placental DNA
contents through a mediation analysis. However, associations between mtDNA
methylation and birth weight as well as the mediation effects were not observed in a similar
birth cohort study [9]. In our study, we did not find an association between mtDNA
methylation and copy number in cord blood and birth weight and infant growth at 6 months
except when stratifying on genotype. Given our small sample size within genomic strata,
we were unable to test for mediation.
Genomic variants in mtDNA have been reported to be associated with obesity and
metabolic diseases [31, 32]. In a population of Taiwanese adults, mtSNP T16189C was

52

found to be more prevalent among patients with impaired fasting glucose and patients with
hypertension [31]. In other population-based studies, T16189C has been associated with
reduced birth weight, leanness in young adults, type 2 diabetes and other metabolic
syndromes among Caucasians and Asians [32-36]. These findings support the hypothesis
that alterations in the control region of mtDNA may contribute to alterations in replication
and transcription of mitochondria, and thus disruption of mitochondrial biogenesis, which
could lead to increased risk of obesity and metabolic syndrome. We did not observe a direct
association between T16189C and birth weight and infant growth in our study, possibly
due to our predominately Hispanic study population. However, we did observe that
T16189C alters individual susceptibility to prenatal air pollution exposure and mtDNA
methylation, and alters the association between mtDNA methylation with birth weight. In
other words, individuals with the C allele were much more susceptible to the effects of air
pollution on mtDNA methylation in the D-loop region, and individuals with the C allele
and higher methylation had lower birth weight. To our knowledge, no other study has
reported such differences in susceptibility based on mtDNA variation.
We acknowledge several limitations in the present study. Although we reported an
association between prenatal air pollution with mtDNA methylation in CD14+ cells,
measurements of methylation in CD14+ cells and CD4+ cells were not available for the
entire cohort. The sample size of the cell-specific methylation analysis may contribute to
the observed differences in which mtDNA loci are associated with prenatal exposure to air
pollution and with birth weight. We were not able to test formally for mediation by mtDNA
methylation in the association between air pollution with birth weight and infant growth
due to the sample size. Furthermore, we had limited power to detect significant false-

53

positive results when a large number of SNPs were tested in a relatively small cohort. In
addition, only three regions of mtDNA were selected and assayed in the present study. Our
findings may not extrapolate to the entire mtDNA genome. Another challenge that may
affect mtDNA methylation detection is presence of nuclear-mitochondrial sequences
(Numts), or pseudogenes in the nuclear genome that do not transcribe. Current bisulfite
treatment of total genomic DNA for methylation analysis may not be able to distinguish
mtDNA sequences and Numts even though we designed primers specific to the mtDNA
sequence and checked them against the nuclear genome for potential overlap.
In summary, the present study found changes in mtDNA methylation in multiple
cell types in relationship with prenatal exposure to TRAP and AAP. Our work also
identified that an mtSNP T16189C modified the association between air pollution and
mtDNA methylation as well as the association between mtDNA methylation and birth
weight. These findings may contribute to further understanding of underlying biological
mechanisms linking air pollution to adverse health outcomes.

54

Supplemental Material
Figure S1. Association between TRAP exposures and mtDNA methylation in PBMCs,
CD14+ and CD4+ cells. A) MT-TF methylation, B) MT-RNR1 methylation, C) D-loop
methylation, and D) LDLR2 methylation
Figure S2. Association between distance to major freeways and major roads with
mtDNA methylation in PBMCs, CD14+ and CD4+ cells. A) MT-TF methylation, B)
MT-RNR1 methylation, C) D-loop methylation, and D) LDLR2 methylation
Table S1. Spearman correlation coefficients between TRAP and AAP exposure
Table S2. Characteristics of infants with a 6-month follow-up
Table S3. Distribution of mtDNA methylation and copy number
Table S4. Association between ambient air pollution exposures and mtDNA methylation
in PBMCs from single-pollutant model
Table S5. Association between ambient air pollution exposures and mtDNA methylation
in CD4+ cells from single-pollutant model
Table S6. Association between ambient air pollution exposure and mtDNA methylation
in CD14+ cells from multi-pollutant model
Table S7. Maternal and infant characteristics by T16189C
Table S8. Association between prenatal air pollution exposure and mtDNA methylation
in PBMCs modified by T16189C
Table S9. Association between prenatal air pollution exposure and mtDNA copy number
modified by T16189C
Table S10. Association between mtDNA methylation in PBMCs and copy number with
infant growth
Appendix Figure 1. Locations of targets in human mitochondrial genome
Appendix Table 1. Primer sequences
Appendix Table 2. List of mtSNPs

55

Figure S1. Association between traffic-related air pollution exposures and mtDNA methylation in PBMCs,
CD14+ and CD4+ cells. A) MT-TF methylation, B) MT-RNR1 methylation, C) D-loop methylation, and D)
LDLR2 methylation. Abbreviations: PBMCs, peripheral blood mononuclear cells; MT-TF, mitochondrially
encoded transfer RNA phenylalanine; MT-RNR1, 12S ribosomal RNA; D-loop, displacement loop region; LDLR,
light chain in displacement loop region. Estimated exposures to prenatal air pollution correspond to the 9-month
average prior to delivery. Estimated percent difference in mtDNA methylation are shown for a one SD difference
in exposure. Models adjusted for maternal age, gestational age, race/ethnicity, parity, and season of birth. Plate
numbers were included as random effects. The values for 1SD were 8 ppb for freeway NOx, 2 ppb for non-
freeway NOx, and 9 ppb for total NOx.

56

Figure S2. Association between distance to major freeways and major roads with mtDNA methylation in
PBMCs, CD14+ and CD4+ cells. A) MT-TF methylation, B) MT-RNR1 methylation, C) D-loop methylation,
and D) LDLR2 methylation. Abbreviations: PBMCs, peripheral blood mononuclear cells; MT-TF,
mitochondrially encoded transfer RNA phenylalanine; MT-RNR1, 12S ribosomal RNA; D-loop, displacement
loop region; LDLR, light chain in displacement loop region. Estimated percent difference in mtDNA
methylation are shown for a one category change in exposure. Distance to major freeways (FCC1) and distance
to major roadways (FCC3) are shown in meters. Models adjusted for maternal age, gestational age,
race/ethnicity, parity, and season of birth. Plate numbers were included as random effects.

57

Table S1. Spearman correlation coefficients between traffic-related air pollution and ambient air pollution
exposure

Freeway
NOx
Non-
Freeway
NOx
Total
NOx
NO2 O3 PM2.5 PM10
Freeway NOx 1 0.21* 0.96* 0.07 -0.19* -0.02 0.30*
Non-Freeway NOx

1 0.45* -0.22* -0.07 -0.09 -0.09
Total NOx

1 -0.01 -0.19* -0.05 0.26*
NO2

1 -0.62* 0.58* 0.12
O3

1 -0.50* -0.22*
PM2.5

1 0.04
PM10

1
*Correlation coefficient had p values <0.05

58

Table S2. Characteristics of infants with a 6-month follow-up

No Follow-Up
(n=126)
6M Follow-up
(n=52)

n (%) or mean ± SD n (%) or mean ± SD p-value
Maternal Characteristics

Age 27.1 ± 6.6 26.2 ± 6.1 0.35
BMI 29.5 ± 7.6 29.2 ± 8.0 0.84
Race/ethnicity

Hispanic White 109 (88.6) 49 (94.2) 0.40
Other 14 (11.4) 3 (5.8)

Nulliparous 57 (46.3) 28 (53.9) 0.36
Education

Less than 12th grade 30 (24.4) 20 (39.2) 0.16
Completed grade 12 52 (42.3) 14 (27.4)

College or higher education 41 (33.3) 17 (33.4)

Family income (US $)

Less than 15,000 43 (35.0) 12 (23.1) 0.18
15,000 - 29,999 35 (28.5) 11 (21.2)

30,000 – 100,000 14 (11.3) 7 (13.5)

Don't know 31 (25.2) 21 (40.4)

Infant Characteristics

Gender

Male 66 (53.7) 25 (48.1) 0.50
Female 57 (46.3) 27 (51.9)

Gestational age (weeks) 38.8 (1.6) 38.3 (2.1 0.10
Birth weight (grams) 3249.1 ± 451.1 3243.1 ± 554.9 0.94
Mode of delivery

Vaginal delivery 81 (65.8) 39 (75.0) 0.27
C-section 42 (34.2) 13 (25.0)

Season of birth

October – March 61 (49.6) 21 (40.4) 0.26
April - November 62 (50.4) 31 (59.6)
Abbreviations: BMI, body mass index. Exposures to prenatal air pollution correspond to the 9-month average
prior to delivery. Pearson chi-square test (for 2 sample comparison) and ANOVA test (for multi-group
comparison) unless otherwise noted; p-values for continuous characteristics derived using t-test.

59

Table S3. Distribution of mtDNA methylation and copy number by T16189C

Full Cohort
(n=178)
C Allele
(n=60)
T Allele
(n=118)
p-value
mtDNA copy number 1.6 ± 2.0 1.6 ± 1.7 1.6 ± 2.2 0.92
PBMCs

MT-TF

1.0 ± 0.5 1.0 ± 0.6 1.0 ± 0.5 0.88
MT-RNR1 2.8 ± 1.1 2.9 ± 1.2 2.7 ± 1.0 0.19
D-loop 2.9 ± 1.4 2.8 ± 1.6 2.9 ± 1.3 0.95
LDLR2 10.0 ± 3.9 9.7 ± 3.4 10.1 ± 4.1 0.50
CD14+ cells
a

MT-TF 0.4 ± 0.4 0.5 ± 0.4 0.3 ± 0.3 0.04
MT-RNR1 1.9 ± 0.5 1.9 ± 0.4 1.9 ± 0.5 0.99
D-loop 2.3 ± 1.7 2.7 ± 1.8 2.1 ± 1.6 0.12
LDLR2 6.6 ± 1.5 6.6 ± 1.8 6.5 ± 1.3 0.85
CD4+ cells
a

MT-TF 0.6 ± 0.6 0.6 ± 0.4 0.6 ± 0.7 0.86
MT-RNR1 1.9 ± 0.7 2.0 ± 0.6 1.9 ± 0.7 0.52
D-loop 1.9 ± 0.8 2.0 ± 0.8 1.8 ± 0.8 0.50
LDLR2 7.4 ± 1.4 7.6 ± 1.1 7.2 ± 1.6 0.19
Abbreviations: PBMCs, peripheral blood mononuclear cells; MT-TF, mitochondrially encoded transfer RNA
phenylalanine; MT-RNR1, 12S ribosomal RNA; D-loop, displacement loop region; LDLR, light chain in
displacement loop region. Distributions of mtDNA methylation and copy number are shown as mean with
standard deviation.
a
Sample sizes for CD14+/CD4+ cells were 33 for C allele group and 56 for T allele group.

60

Table S4. Association between ambient air pollution exposures and mtDNA methylation in PBMCs from
single-pollutant model
MT-TF MT-RNR1 D-loop LDLR2

β 95% CI β 95% CI β 95% CI β 95% CI
NO2 0.008 (-0.08, 0.01) -0.13 (-0.32, 0.05) -0.10 (-0.35, 0.16) 0.37 (-0.26, 1.00)
O3 -0.07 (-0.16, 0.03) 0.07 (-0.12, 0.26) 0.04 (-0.19, 0.27) -0.44 (-1.09, 0.22)
PM2.5 0.00 (-0.08, 0.08) -0.05 (-0.21, 0.12) 0.00 (0.23, 0.22) 0.32 (-0.24, 0.89)
PM10 0.06 (-0.03, 0.15) -0.05 (-0.23, 0.14) 0.09 (0.15, 0.33) -0.13 (-0.46, 0.18)
Abbreviations: PBMCs, peripheral blood mononuclear cells; MT-TF, mitochondrially encoded transfer RNA
phenylalanine; MT-RNR1, 12S ribosomal RNA; D-loop, displacement loop region; LDLR, light chain in
displacement loop region. Estimated exposures to prenatal air pollution correspond to the 9-month average prior
to delivery. Estimated percent difference [β] in mtDNA methylation are shown for a one SD difference in
exposure. Models adjusted for maternal age, gestational age, race/ethnicity, parity, and season of birth. Plate
numbers were included as random effects. The values for 1SD were 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3
μg/m
3
for PM10.

61

Table S5. Association between ambient air pollution exposures and mtDNA methylation in CD4+ cells from
single-pollutant model
MT-TF MT-RNR1 D-loop LDLR2

β 95% CI β 95% CI β 95% CI β 95% CI
NO2 -0.04 (-0.22, 0.13) 0.02 (-0.17, 0.20) 0.02 (-0.22, 0.25) 0.06 (-0.33, 0.46)
O3 -0.04 (-0.18, 0.11) 0.01 (-0.14, 0.16) -0.03 (-0.22, 0.16) -0.04 (-0.37, 0.28)
PM2.5 -0.02 (-0.16, 0.11) -0.04 (-0.18, 0.10) -0.03 (-0.21, 0.15) 0.02 (-0.29, 0.33)
PM10 0.02 (-0.12, 0.16) -0.04 (-0.19, 0.11) 0.06 (-0.14, 0.25) -0.11 (-0.43, 0.21)
Abbreviations: MT-TF, mitochondrially encoded transfer RNA phenylalanine; MT-RNR1, 12S ribosomal RNA;
D-loop, displacement loop region; LDLR, light chain in displacement loop region. Estimated exposures to
prenatal air pollution correspond to the 9-month average prior to delivery. Estimated percent difference [β] in
mtDNA methylation are shown for a one SD difference in exposure. Models adjusted for maternal age,
gestational age, race/ethnicity, parity, and season of birth. Plate numbers were included as random effects. The
values for 1SD were 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3 μg/m
3
for PM10.

62

Table S6. Association between ambient air pollution exposure and mtDNA methylation in CD14+ cells from
multi-pollutant model
MT-TF MT-RNR1 D-loop LDLR2
β 95% CI β 95% CI β 95% CI β 95% CI
NO 2 + O 3
NO 2 -0.04 (-0.18, 0.09) -0.11 (-0.29, 0.07) 0.20 (-0.43, 0.84) 0.34 (-0.22, 0.89)
O 3 0.04 (-0.07, 0.15) -0.03 (-0.18, 0.12) 0.07 (-0.46, 0.61) 0.46 (0.01, 0.91)
NO 2 + PM 2.5
NO 2 -0.10 (-0.23, 0.04) 0.02 (-0.14, 0.19) -0.30 (-0.88, 0.28) 0.27 (-0.26, 0.80)
PM 2.5 0.03 (-0.08, 0.13) -0.15 (-0.28, 0.01) 0.59 (0.13, 1.05) -0.39 (-0.80, 0.03)
NO 2 + PM 10
NO 2
-0.08 (-0.20, 0.04) -0.16
(-0.31, -
0.01) 0.23 (-0.32, 0.78) 0.32 (-0.16, 0.78)
PM 10
0.00 (-0.09, 0.10) 0.12 (0.00, 0.25) -0.14 (-0.59, 0.30) -0.62
(-0.99, -
0.24)
O 3 + PM 2.5
O 3 0.09 (-0.02, 0.21) -0.12 (-0.27, 0.03) 0.53 (0.01, 1.05) 0.18 (-0.30, 0.65)
PM 2.5
0.04 (-0.07, 0.15) -0.22
(-0.36, -
0.07) 0.79 (0.30, 1.28) -0.14 (-0.60, 0.31)
O 3 + PM 10
O 3 0.06 (-0.03, 0.16) 0.07 (-0.06, 0.19) -0.08 (-0.53, 0.38) 0.10 (-0.27, 0.46)
PM 10
0.00 (-0.09, 0.09) 0.09 (-0.03, 0.22) -0.09 (-0.53, 0.35) -0.46
(-0.82, -
0.09)
PM 2.5 + PM 10
PM 2.5
-0.01 (-0.09, 0.07) -0.17
(-0.28, -
0.06) 0.53 (0.14, 0.91) -0.12 (-0.46, 0.21)
PM 10
-0.02 (-0.11, 0.07) 0.12 (0.01, 0.23) -0.24 (-0.64, 0.17) -0.46
(-0.81, -
0.10)
Abbreviations: MT-TF, mitochondrially encoded transfer RNA phenylalanine; MT-RNR1, 12S ribosomal RNA;
D-loop, displacement loop region; LDLR, light chain in displacement loop region. Estimated exposures to
prenatal air pollution correspond to the 9-month average prior to delivery. Estimated percent difference [β] in
mtDNA methylation are shown for a one SD difference in exposure. Models adjusted for maternal age,
gestational age, race, parity, and season of birth. Plate numbers were included as random effects. The values for
1SD were 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3 μg/m
3
for PM10.

63

Table S7. Maternal and infant characteristics by T16189C
C Allele T Allele

n (%) or
mean ± SD
n (%) or
mean ± SD
p-value
Maternal Characteristics

Age 27.2 ± 6.9 26.8 ± 6.4 0.70
BMI 29.1 ± 5.0 29.6 ± 8.8 0.67
Race/ethnicity

Hispanic White 51 (85.0) 110 (93.2) 0.08
Other 9 (15.0) 8 (6.8)

Nulliparous 29 (48.3) 58 (49.2) 0.92
Education

Less than 12th grade 19 (31.7) 32 (27.4) 0.72
Completed grade 12 23 (38.3) 44 (37.6)

College or higher education 18 (30.0) 41 (35.0)

Family income (US $)

Less than 15,000 14 (23.3) 42 (35.6) 0.37
15,000 - 29,999 19 (31.7) 29 (24.6)

30,000 – 100,000 9 (15.0) 13 (11.0)

Don't know 18 (30.0) 34 (28.8)

Infant Characteristics

Gender

Male 28 (46.7) 66 (55.9) 0.24
Female 32 (53.3) 52 (44.1)

Gestational age (weeks) 38.5 ± 1.6 38.7 (1.8) 0.39
Birth weight (grams)
3166.0 ±
472.2
3286.1 ±
484.8 0.12
Mode of delivery

Vaginal delivery 44 (73.3) 76 (66.1) 0.38
C-section 16 (26.7) 39 (33.9)

Season of birth

October – March 28 (46.7) 55 (46.6) 0.99
April - November 32 (53.3) 63 (53.4)

Traffic-related air pollutants

Freeway NOx (ppb) 8.8 ± 6.6 10.3 ± 9.0 0.26
Non-Freeway NOx (ppb) 3.4 ± 1.9 2.9 ± 1.3 0.09
Total NOx (ppb) 12.1 ± 6.7 13.2 ± 9.5 0.38
Ambient air pollutants

NO2 (ppb) 19.1 ± 2.4 18.9 ± 2.5 0.75
O3 (ppb) 25.7 ± 2.4 25.8 ± 2.8 0.94
PM2.5 (μg/m
3
) 12.5 ±0.8 12.4 ± 1.0 0.45
PM10 (μg/m
3
) 30.8 ± 2.8 30.7 ± 2.6 0.88
Abbreviations: BMI, body mass index. Prenatal exposures to traffic-related and ambient air pollution in the nine
months prior to delivery are reported. Pearson chi-square test (for 2 sample comparison) and ANOVA test (for
multi-group comparison) unless otherwise noted; p-values for continuous characteristics derived using t-test.

64

65

Table S9. Association between prenatal air pollution exposure and mtDNA copy number modified by T16189C
C Allele T Allele
β 95% CI β 95% CI pinteraction
Freeway NOx 0.15 (-0.35, 0.64) -0.06 (-0.36, 0.25) 0.92
Non-Freeway NOx 0.52 (0.20, 0.83) 0.15 (-0.25, 0.56) 0.79
Total NOx 0.37 (-0.15, 0.88) -0.04 (-0.35, 0.27) 0.90
NO2 -0.24 (-0.75, 0.26) 0.15 (-0.26, 0.56) 0.85
O3 -0.07 (-0.66, 0.52) -0.43 (-0.86, -0.01) 0.72
PM2.5 -0.47 (-0.16, 0.18) 0.002 (-0.38, 0.38) 0.72
PM10 -0.21 (-0.60, 0.19) -0.01 (-0.47, 0.46) 0.90
Estimated effects [β] are shown for a one SD increase in exposure. Models were adjusted for maternal age,
gestational age, race, sex, parity, BMI, and delivery type. Plate numbers were included as random effects.
Pinteraction tests effect modification of T16189C and adjusted for multiple testing at a false discovery rate (FDR)
of 0.05. The values for 1SD were 8 ppb for freeway NOx, 2 ppb for non-freeway NOx, 9 ppb for total NOx, 2
ppb for NO2, 3 ppb for O3, 1 μg/m
3
for PM2.5, and 3 μg/m
3
for PM10.

66

Table S10. Association between mtDNA methylation in PBMCs and copy number with infant growth
Weight at 6 months z-score change
β 95% CI β 95% CI
PBMCs methylation

MTTF 9.08 (-312.64, 330.81) -0.02 (-0.39, 0.35)
MTRNR1 -203.36 (-594.39, 187.67) -0.17 (-0.62, 0.28)
D-Loop 163.56 (-100.11, 427.22) 0.14 (-0.16, 0.45)
LDLR2 33.62 (-22.18, 289.04) 0.04 (-0.25, 0.33)
mtDNA copy number 111.72 (-176.86, 400.31) 0.13 (-0.20, 0.46)
Abbreviations: PBMCs, peripheral blood mononuclear cells; MT-TF, mitochondrially encoded transfer RNA
phenylalanine; MT-RNR1, 12S ribosomal RNA; D-loop, displacement loop region; LDLR, light chain in
displacement loop region. Estimated effects [β] are shown for a one SD increase in exposure. The sex-specific
weight-for-age z-score calculated based on CDC 2000 growth chart. Models were adjusted for maternal age,
gestational age, race/ethnicity, sex, parity, BMI, and delivery type. Plate numbers were included as random
effects. Sample size = 52.

67

Chapter 3: References
1. Iacobazzi V, Castegna A, Infantino V, Andria G: Mitochondrial DNA methylation
as a next-generation biomarker and diagnostic tool. Mol Genet Metab 2013, 110(1-
2):25-34.
2. Shock L TP, Peterson E, Moran R, Taylor S: DNA methyltransferase 1, cytosine
methylation, and cytosine hydroxymethylation in mammalian mitochondria. PNAS
2011, 106(9).
3. Bellizzi D DAP, Scafone T, Giordano M, Riso V, Roccio A, Passarino G: The
control region of mitochondrial DNA shows an unusual CpG and non-CpG
methylation pattern. DNA Research 2013, 20(6).
4. Bellizzi D, D'Aquila P, Scafone T, Giordano M, Riso V, Riccio A, Passarino G:
The control region of mitochondrial DNA shows an unusual CpG and non-CpG
methylation pattern. DNA Res 2013, 20(6):537-547.
5. Clayton DA: Mitochondrial DNA replication: what we know. IUBMB Life 2003,
55(4-5):213-217.
6. Mercer TR, Neph S, Dinger ME, Crawford J, Smith MA, Shearwood AM, Haugen
E, Bracken CP, Rackham O, Stamatoyannopoulos JA et al: The human
mitochondrial transcriptome. Cell 2011, 146(4):645-658.
7. Chan DC: Mitochondria: dynamic organelles in disease, aging, and development.
Cell 2006, 125(7):1241-1252.
8. Byun HM, Baccarelli AA: Environmental exposure and mitochondrial epigenetics:
study design and analytical challenges. Hum Genet 2014, 133(3):247-257.
9. Janssen B BH, Gyselaers W, Lefebvre W, Baccarelli A, Nawrot T: Placental
mitochondrial methylation and exposure to airborne particulate matter in the early
life environment: An ENVIRONAGE birth cohort study. Epigenetics 2015, 10(6).
10. Byun H PT, Motta V, Hou L, Nordio F, Apostoli P, Bertazzi P, Baccarelli A: Effects
of airborne polutants on mitochondrial DNA methylation. Particle and Fibre
Toxiocology 2013, 10(18).
11. Benson P: Caline4 - a dispersion model for predicting air pollution concentration
near roadways. . Sacramento: California Department of Transportation, Office of
Transportation Laboratory 1989.
12. Eckel SP, Berhane K, Salam MT, Rappaport EB, Linn WS, Bastain TM, Zhang Y,
Lurmann F, Avol EL, Gilliland FD: Residential traffic-related pollution exposures
and exhaled nitric oxide in the children's health study. Environ Health Perspect
2011, 119(10):1472-1477.
13. Wong DW, Yuan L, Perlin SA: Comparison of spatial interpolation methods for
the estimation of air quality data. J Expo Anal Environ Epidemiol 2004, 14(5):404-
415.
14. Byun HM, Siegmund KD, Pan F, Weisenberger DJ, Kanel G, Laird PW, Yang AS:
Epigenetic profiling of somatic tissues from human autopsy specimens identifies
tissue- and individual-specific DNA methylation patterns. Hum Mol Genet 2009,
18(24):4808-4817.
15. Armstrong DA GB, Blair BA, Guerin DJ, Litzky JF, Chavan NR, Pearson KJ,
Marsit CJ.: Maternal Smoking During Pregnancy is Associated with Mitochondrial
DNA Methylation. Environmental Epigenetics 2016, 2(3):1-9.

68

16. Ghosh S, Sengupta S, Scaria V: Comparative analysis of human mitochondrial
methylomes shows distinct patterns of epigenetic regulation in mitochondria.
Mitochondrion 2014, 18:58-62.
17. van der Wijst MG, Rots MG: Mitochondrial epigenetics: an overlooked layer of
regulation? Trends Genet 2015, 31(7):353-356.
18. Ding R, Jin Y, Liu X, Zhu Z, Zhang Y, Wang T, Xu Y: Characteristics of DNA
methylation changes induced by traffic-related air pollution. Mutat Res Genet
Toxicol Environ Mutagen 2016, 796:46-53.
19. Gruzieva O, Xu CJ, Breton CV, Annesi-Maesano I, Anto JM, Auffray C, Ballereau
S, Bellander T, Bousquet J, Bustamante M et al: Epigenome-Wide Meta-Analysis
of Methylation in Children Related to Prenatal NO2 Air Pollution Exposure.
Environ Health Perspect 2017, 125(1):104-110.
20. Baccarelli A, Wright RO, Bollati V, Tarantini L, Litonjua AA, Suh HH, Zanobetti
A, Sparrow D, Vokonas PS, Schwartz J: Rapid DNA methylation changes after
exposure to traffic particles. Am J Respir Crit Care Med 2009, 179(7):572-578.
21. Zhong J, Cayir A, Trevisi L, Sanchez-Guerra M, Lin X, Peng C, Bind MA, Prada
D, Laue H, Brennan KJ et al: Traffic-Related Air Pollution, Blood Pressure, and
Adaptive Response of Mitochondrial Abundance. Circulation 2016, 133(4):378-
387.
22. Shaughnessy DT, McAllister K, Worth L, Haugen AC, Meyer JN, Domann FE,
Van Houten B, Mostoslavsky R, Bultman SJ, Baccarelli AA et al: Mitochondria,
energetics, epigenetics, and cellular responses to stress. Environ Health Perspect
2014, 122(12):1271-1278.
23. Lee HC, Wei YH: Mitochondrial biogenesis and mitochondrial DNA maintenance
of mammalian cells under oxidative stress. Int J Biochem Cell Biol 2005,
37(4):822-834.
24. Metodiev MD, Lesko N, Park CB, Camara Y, Shi Y, Wibom R, Hultenby K,
Gustafsson CM, Larsson NG: Methylation of 12S rRNA is necessary for in vivo
stability of the small subunit of the mammalian mitochondrial ribosome. Cell
Metab 2009, 9(4):386-397.
25. Lee C, Kim KH, Cohen P: MOTS-c: A novel mitochondrial-derived peptide
regulating muscle and fat metabolism. Free Radic Biol Med 2016, 100:182-187.
26. Lee C, Zeng J, Drew BG, Sallam T, Martin-Montalvo A, Wan J, Kim SJ, Mehta H,
Hevener AL, de Cabo R et al: The mitochondrial-derived peptide MOTS-c
promotes metabolic homeostasis and reduces obesity and insulin resistance. Cell
Metab 2015, 21(3):443-454.
27. Armstrong DA, Green BB, Blair BA, Guerin DJ, Litzky JF, Chavan NR, Pearson
KJ, Marsit CJ: Maternal smoking during pregnancy is associated with
mitochondrial DNA methylation. Environ Epigenet 2016, 2(3).
28. Yu E, Calvert PA, Mercer JR, Harrison J, Baker L, Figg NL, Kumar S, Wang JC,
Hurst LA, Obaid DR et al: Mitochondrial DNA damage can promote
atherosclerosis independently of reactive oxygen species through effects on smooth
muscle cells and monocytes and correlates with higher-risk plaques in humans.
Circulation 2013, 128(7):702-712.

69

29. Janssen BG, Munters E, Pieters N, Smeets K, Cox B, Cuypers A, Fierens F, Penders
J, Vangronsveld J, Gyselaers W et al: Placental mitochondrial DNA content and
particulate air pollution during in utero life. Environ Health Perspect 2012,
120(9):1346-1352.
30. Clemente DB, Casas M, Vilahur N, Begiristain H, Bustamante M, Carsin AE,
Fernandez MF, Fierens F, Gyselaers W, Iniguez C et al: Prenatal Ambient Air
Pollution, Placental Mitochondrial DNA Content, and Birth Weight in the INMA
(Spain) and ENVIRONAGE (Belgium) Birth Cohorts. Environ Health Perspect
2016, 124(5):659-665.
31. Weng SW, Lin TK, Wang PW, Chen SD, Chuang YC, Liou CW: Single nucleotide
polymorphisms in the mitochondrial control region are associated with metabolic
phenotypes and oxidative stress. Gene 2013, 531(2):370-376.
32. Palmieri VO, De Rasmo D, Signorile A, Sardanelli AM, Grattagliano I, Minerva F,
Cardinale G, Portincasa P, Papa S, Palasciano G: T16189C mitochondrial DNA
variant is associated with metabolic syndrome in Caucasian subjects. Nutrition
2011, 27(7-8):773-777.
33. Mohlke KL, Jackson AU, Scott LJ, Peck EC, Suh YD, Chines PS, Watanabe RM,
Buchanan TA, Conneely KN, Erdos MR et al: Mitochondrial polymorphisms and
susceptibility to type 2 diabetes-related traits in Finns. Hum Genet 2005,
118(2):245-254.
34. Ong KK, Dunger DB: Birth weight, infant growth and insulin resistance. Eur J
Endocrinol 2004, 151 Suppl 3:U131-139.
35. Park KS, Chan JC, Chuang LM, Suzuki S, Araki E, Nanjo K, Ji L, Ng M, Nishi M,
Furuta H et al: A mitochondrial DNA variant at position 16189 is associated with
type 2 diabetes mellitus in Asians. Diabetologia 2008, 51(4):602-608.
36. Parker E, Phillips DI, Cockington RA, Cull C, Poulton J: A common mitchondrial
DNA variant is associated with thinness in mothers and their 20-yr-old offspring.
Am J Physiol Endocrinol Metab 2005, 289(6):E1110-1114.

70

CHAPTER 4
Prenatal Air Pollution, Mitochondrial-Associated MicroRNA, and Infant Growth
4.1 Abstract
Mitochondria, known to respond to environmental exposures that induce oxidative
stress, have been implicated in a variety of developmental diseases. We investigated
associations between in utero air pollution exposure, patterns of maternal circulating
microRNAs (miRNAs) that target mitochondrial genes or nuclear genes known to
influence mitochondrial function, and newborn outcomes in 137 pregnant women currently
enrolled in the ongoing Maternal and Developmental Risks from Environmental and Social
Stressors (MADRES) birth cohort. Maternal miRNAs were extracted and profiled from
maternal blood samples during first trimester and third trimester utilizing NanoString
nCounter miRNA Expression Assay Human v2 System. Residential timelines were
constructed on a daily level for all participants starting from two years prior to delivery.
Daily concentrations of the ambient pollutants NO2, PM10, PM2.5 and O3 were estimated
using inverse-distance squared spatial interpolation at each participant’s residence. A total
of 83 mitochondrial-associated miRNAs were identified and 31 of those were above limit
of detection and included for final analyses. Multiple linear regression models and
distributed lag models were fitted to determine the association between weekly air
pollutants and miRNA expressions. After correcting for multiple testing, miR-23b-3p
(β=9.86, 95% CI 3.84, 15.87, p-adj=0.02) and miR-29a-3b (β=8.79, 95% CI 3.77, 13.81,
p-adj=0.02) were positively associated with PM2.5 exposure during third trimester. Weekly
DLMs revealed that the changes of miRNAs are specific to vulnerable time windows of

71

exposure. We did not observe a significant association between mitochondrial-associated
miRNAs and birthweight. Our study provides evidence of the important functional role of
miRNAs in early fetal development. Further studies are warranted to determine the
underlying regulatory mechanism of mitochondria.
4.2 Mitochondrial-Associated MiRNAs
MiRNAs are non-coding RNA sequences known to be key regulators of gene
expression at the post-transcriptional level. miRNAs are short, single-stranded, ~21
nucleotide-long noncoding RNA molecules, that are highly ubiquitous and possess
conservation across many species [1]. miRNAs have emerged as the key post-transcriptional
regulators of expression of protein-coding gene, through repressing messenger RNA
(mRNA) translation or promoting their degradation. Each miRNA can affect multiple target
genes. Functional studies have shown that miRNAs participate in regulation of almost every
cellular process such as cell proliferation, cell death, differentiation and organogenesis [2].
Seven years after the initial identification of the founding member of the miRNA
family lin-4, the second miRNA, let-7, was discovered. let-7 encodes a temporally
regulated 21-nucleotide small RNA that controls the developmental transition [3]. Since
the discovery of lin-4 and let-7, hundreds of miRNAs have been identified in various
organisms by molecular cloning and bioinformatic approaches. Most miRNAs are
transcribed by RNA polymerase II (Pol II) to generate a stem loop containing primary
miRNA (pri-miRNA). The pri-miRNA is processed within the nucleus by a multiprotein
complex called the microprocessor, of which the core components are the RNase III
enzyme Drosha and the double-stranded RNA-binding domain (dsRBD) protein [4]. This
complex cleaves the pri-miRNA to produce the precursor miRNA (pre-miRNA). The

72

product of the cleavage, a ~70-nucleaotide pre-miRNA, is exported to the cytoplasm where
Dicer processes it to an ~20-bp miRNA duplex. One strand of this duplex, representing a
mature miRNA, is then incorporated into the miRNA-induced silencing complex
(miRISC). As part of miRISC, miRNAs base-pair to target mRNAs and induce their
translational repression or deadenylation and degradation [5].
Recently, miRNAs have been found to be present within mitochondria in different
cell lines and tissues [6-10]. In addition, nuclear-encoded mRNA and miRNAs were found
to be associated with other mitochondrial membrane and are depleted in mitoplast [9, 10].
This implies that miRNAs are more significantly associated with mitochondria rather than
their enrichment inside the double membrane bound organelle. However, the dynamics and
enrichment of miRNAs in various mitochondrial compartments have to be determined [11].
The discovery of mitochondria-localized miRNAs (mitomiRs) has raised the issue
of how they translocate from the nucleus to the mitochondria and the possibility of
mitochondrial RNA synthesis. Approximately 150 miRNAs have been detected in
mitochondrial specimens [12]. However, the physiology and pathology of miRNAs in
mitochondria remain largely unknown. One recent study has reported that the
mitochondrial-related miRNA, miR-143-3p was highly expressed in high-fat-diet induced
mice and further gain-and-loss-of-function study identified that miR-143-3p up-regulated
ATP production and ROS production of mitochondria by silencing phosphatase and tensin-
homolog (PTEN) gene [13].

73

Table 1. miRNA localized within mitochondria

Source of
mitochondria
miRNA Ref
Mouse liver mmu-miR-202-5p, mmu-miR-122, mmu-miR-223, mmu-miR-
134, mmu-miR-709, mmu-miR-720, mmu-miR-680, mmu-miR-
494, mmu-miR-155
Bian et al.,
2010
Rat liver rno-miR-130a, rno-miR-130b, rno-miR-140*, rno-miR-320, rno-
miR-494, rno-miR-671
Kren et al.,
2009
Myotubes hsa-miR-720, hsa-miR-133b, hsa-miR-1974, hsa-miR-24, hsa-
miR-133a, hsa-miR-125a-5p, hsa-miR-1979, hsa-miR-103, hsa-
miR-125b, hsamiR-103, hsa-miR-221, hsa-miR-23a, hsa-let-7b,
hsa-miR-423-3p, hsa-miR-106a, hsa-miR-23b, hsa-miR-92a, hsa-
miR-193b, hsa-miR-365,hsa-miR-93, hsa-miR-532-3p, hsa-miR-
20a, hsa-miR-149, hsa-miR-181a, hsa-miR-503, hsa-miR-210,
hsa-miR-107, hsa-miR-574-3p, has-miR-34a, hsa-let-7g, hsa-
miRPlus-D1033, hsa-miR-19b, hsa-miR-197, hsa-miR-324-3p,
hsa-miR-127-3p, hsa-miR-324-5p, hsa-miR-484, has-miR-151-5p,
hsa-miR-486-5p, hsa-miR-542-5p, hsa-miR-199a-5p, hsa-miR-
501-3p, hsa-miR-675*, hsa-miR-134, hsa-miR-490-3p, hsa-miR-
598
Barrey et
al., 2011
HeLa hsa-mir-1973, hsa-mir-1275, hsa-mir-494let-7b, let7g, hsa-miR-
107, hsa-miR-181a, hsa-miR-221, hsa-miR-320a
Sripada et
al., 2012
143-B hsa-mir-146a, hsa-mir-103, hsa-mir-16 Mercer et
al., 2011
HEK293 let-7b, let7g, hsa-miR-107, hsa-miR-181a, hsa-miR-221, hsa-miR-
320a
Sripada et
al., 2012

Several mitomiRs and miRNAs targeting mitochondria have been shown to influence lipid
metabolism and thus may contribute to the development of obesity [12, 14, 15]. The
majority of mitomiRs has been identified in only one or two studies. Only seven miRNAs
have been detected in mitochondria by three or more studies, among which the most
frequently studied mitomiR is miR-494-3p [12]. The mitomiRs that have been detected in
mitochondria are shown in Table 1.
miRNAs are both associated and localized in mitochondria. The association of few
miRNAs with mitochondria appears to be universal, while the association of some miRNAs
with mitochondria are cell type specific [11]. It is possible that the mitochondrial-

74

associated miRNAs regulate mitochondria encoded mRNA or nuclear encoded
mRNA/protein destined to mitochondria [11]. The identification of miRNAs in
mitochondria has raised new questions regarding biological function of miRNAs. It is
established that miRNAs, originated in the nuclear genome, are exported to cytosol where
they are processed and exert their function by inhibiting nuclear genome-derived mRNA
[16]. In addition, emerging evidence suggest that miRNAs associated with mitochondria
target both nuclear encoded and mitochondrial mRNA transcripts [6].
miRNAs are involved in various aspects of mitochondrial function and metabolic
pathways including electron transport chain, tricarboxylic acid cycle, amino acid
metabolism, lipid metabolism, one-carbon metabolism, and regulation of aging and
apoptosis [12, 16]. For example, miR-210, which is considered to be a paradigm of miRNA
regulating mitochondrial metabolism in response to hypoxic stress, also directly affects the
mitochondrial respiratory chain through targeting cytochrome c oxidase assembly protein
(COX10) [17]. miRNAs regulate mitochondrial gene expression and influence
mitochondrial function either by regulating mitochondrial genes directly or by regulating
nuclear genes that are known to influence mitochondrial function [16].
A growing literature suggests that miRNAs can respond to environmental exposures,
including air pollution, smoking, metals, and chemicals such as endocrine disruptors [18,
19]. In vitro studies have shown that diesel exhaust particles (DEP), the largest source of
emitted airborne PM, can disrupt expression patterns of miRNAs in human airway epithelial
cells, indicating that DEP exposure may modify cellular processes in response to
inflammatory exposure and potentially lead to the development of a pathologic disease state
[20]. In addition, altered miRNA expression profiles have been observed in various

75

populations in response to air pollution exposure in population-based studies [21-25]. In
particular, a recent study in the ENVIRONAGE birth cohort reported an association between
prenatal exposure to PM2.5 and placental miRNA expression suggesting that in utero PM2.5
affects miRNAs expression and its target genes [25].
There is increasing interest in determining how miRNAs could impact newborn
outcomes and growth trajectories. Circulating miRNAs have been found to be associated
with a number of maternal-placental conditions such as preeclampsia, fetal growth
restriction, and preterm delivery [26-31]. For example, miR-21, is reported to be associated
with DEP exposure and is also significantly associated with maternal body mass index (BMI)
and birthweight, suggesting that miRNA may underlie health effects of PM exposure [21,
29, 32]. Potential functional roles of miRNA that could represent the mechanistic effects of
environmental exposure on placental function and fetal and postnatal growth include
placental development, placental transport and metabolism, placental protective and
endocrine functions, and extracellular signaling to the mother or developing infant [33-36].
Previous works have demonstrated that expression of miR-21 in placenta tissue is associated
with increased growth and macrosomia potentially through regulation of the Jak-STAT and
mTOR signaling pathway [37, 38]. Conversely, miR-141 suppresses E2F3 and PLAG1,
which is, in turn, related to reduced IGF2 expression and linked to growth restriction [39].
4.3 Proposed Study
The proposed study aims to investigate whether in utero air pollution exposure
alters patterns of circulating miRNAs that target mitochondrial genes or nuclear genes
known to influence mitochondrial function, and additionally if these miRNAs can be
related to early life growth. Advanced statistical methods (e.g., distributed lag models

76

[DLMs]), were also applied to identify sensitive windows in relation to fetal growth. This
study addresses critical gaps in our understanding of impacts of air pollution on
mitochondrial-associated miRNA expression patterns and their influence on newborn
outcomes. It also addresses the potential role of mitochondria in metabolism during critical
fetal developmental windows.
4.3.1 Study Population and Design
The Maternal and Developmental Risks from Environmental and Social Stressors
(MADRES) study is designed specifically to address the impacts of environmental and
psychosocial stressors on maternal and child health, particularly obesity risk, in an urban,
low income Hispanic population in Los Angeles. One of the study goals is to be able to
study mechanistic questions in this underserved population by offering opportunities to
identify useful molecular biomarkers and understand the biological importance of these
molecular features in the context of children’s environmental health. MADRES study
subjects are currently recruited from the Obstetrics and Gynecology clinic at the Los
Angeles County (LAC) + USC Medical Center and from the Eisner Pediatric and Family
Medical Center. Inclusion criteria are: (1) < 30 weeks since the date of mother’s last
menstrual period at the time of enrollment, (2) currently a patient at either of the Centers,
and (3) singleton pregnancy. Exclusion criteria are (1) HIV positive status; (2) physical,
mental, or cognitive disabilities that prevent participation; and (3) current incarceration.
Subjects are followed using a combination of in-person clinic visits and telephone
interviews in either English or Spanish with mothers during pregnancy,
collection/abstraction of maternal and pediatric medical record data from birth through the
first year of life, and a final home visit when the infant is 12 months old. The ongoing

77

MADRES study aims to recruit ~1,000 mother-infant pairs and currently has an
approximately 750 pregnant women enrolled. The current study utilized already collected
maternal plasma samples from two timepoints during pregnancy (first trimester and third
trimester) on 225 women for miRNA expression profiling.
A wide range of potential covariates were assessed in MADRES, including self-
rated health status, health history, smoking, alcohol/drug use, maternal stress and
depression, sleep, pregnancy history (e.g., BMI, parity, gravidity, gestational age/preterm,
delivery type, pregnancy complications), neonatal complications, and demographics.
Informed consent and HIPAA authorization to access medical records was obtained
at study entry for each participant and her child. The University of Southern California’s
Institutional Review Board approved the protocol.
4.3.2 Exposure Assessment
The residential addresses of pregnant mothers and children from two years prior to
pregnancy through the first year of the infant’s life were collected using a standardized
protocol and geocoded for use in spatial analysis. Daily pollutant estimates of NO2, PM10,
PM2.5 and O3 were assigned to residential locations and time periods for each study subject,
accounting for any residential mobility or moving, by Sonoma Technology, Inc (STI) using
a time-weighting approach for subjects with multiple residences during a time period using
historical air quality data for 2014-2018 from the U.S. EPA’s Air Quality System.
Exposures were estimated using the inverse distance-squared weighting [40, 41].

78

4.3.3 Newborn Health Outcomes
The primary outcome of interest, birthweight of newborns, was extracted from
electronic medical records (EMR) at regular intervals. Pregnancy and reproductive health
outcomes including prenatal visit height and weight, pregnancy complications, APGAR
score etc., were obtained through the EMR system as well. Birthweight less than 2500 g in
infants is defined as low birthweight (LBW), while children in less than the 10th percentile
of predicted birthweight based on gestational age and sex in term infants are considered as
small for gestational age (SGA) [42].
4.3.5 Measurement of miRNA expression
Maternal blood samples were collected for the purpose of MADRES miRNA study.
A 46ml blood sample was collected from mothers by a certified phlebotomist at first
trimester and third trimester clinic visits. Samples were immediately transported to our
laboratory for processing and storage of plasma, trace metals free plasma, and serum and
stored at -80°C.
This study utilized the expression profiles of miRNAs measured for the MADRES
miRNA study, which aims to examine how psychosocial stress in the maternal environment
impacts the pattern of expression of miRNA in maternal blood and in placental tissue. We
have developed protocols to isolate exosomal RNA utilizing the Qiagen ExoRNeasy
Serum/Plasma Midi kit, which uses membrane affinity spin columns to efficiently isolate
RNA from exomes and other extracellular vesicles extracted from serum or plasma. To
assure high RNA purity, the laboratory at USC also perform an on-column DNase I
digestion. RNA quantity is assessed using the Nanodrop Spectrophotometer, and integrity

79

assessed using the Agilent Bioanalyzer. Isolated RNA is separated into single use aliquots
and kept in RNAse-free water at -80
o
C. Genome-wide miRNA expression profiling was
carried out using NanoString technology, specifically the nCounter miRNA Expression
Assay Human v2, on the nCounter Analysis System (Seattle, WA, USA). This approach
uses a novel digital technology (proprietary to NanoString) that is based on direct
multiplexed measurement of nucleic acids and offers the ability to quantify 800 human
miRNAs derived from miRBase v.18 in a single reaction with better precision than
microarray. The assay includes 6 positive miRNA controls which are probes and spiked
miRNA samples added to each sample, 6 negative miRNA assay controls which are probes
with no known human binding sites, and 5 housekeeping controls to aid in quality control
in each individual reaction. This profiling was performed in the Norris Cancer Center
Molecular Genomics Core Facility at USC. For all data from the Nanostring platform, raw
data was normalized using the geometric mean of the counts of the top 75 most highly
expressed miRNAs. MiRNA whose counts are below limit of detection (LOD), defined as
2 standard deviations above the mean of the negative control probes, in more than 50% of
the samples were excluded as potential background.
4.3.6 Selection of Mitochondrial-Associated miRNAs
Approximately 150 miRNAs that are localized in mitochondria were identified
through a literature search and review articles. Literature describing mitomiRs or
mitochondrial-associated miRNAs was searched in PubMed and Web of Science using
Medical Search Heading (MeSH) terms and keywords. Identified articles describing either
human studies or animal models were reviewed. Review articles were also included. In
addition, because miRNAs often have numerous predicted gene targets, some of genes may

80

not be necessarily specific to mitochondria or have mitochondrial function. A list of all
known nuclear-encoded mitochondrial genes and nuclear genes known to have
mitochondrial function was developed using 1) AmiGO 2, the gene product data generated
by the Gene Ontology Consortium [43], and 2) MitoCarta compendium, an inventory data
base of 1,158 human and mouse genes encoding proteins with mitochondrial localization
[44]. A total of 1,302 genes on either mitochondrial genome or nuclear genome were
identified through search. The list of genes was then cross-referenced with known targeted
genes of the 798 human miRNAs profiled from the Nanostring Platform generated using
two databases that catalogue experimentally validated miRNA-gene interactions, DIANA-
TarBase v7.0 and miRTarBase 7.0. Eighty-three mitochondrial-associated miRNAs were
identified through the extensive search.
4.3.7 Statistical Analysis
Descriptive analyses were conducted to examine maternal and infant characteristics
and to characterize the distribution of air pollutant exposures, miRNA expression levels,
and infant growth. Chi-square test was used for categorical variables and the t-test was
used for continuous variables. Fisher’s exact test and Wilcoxon-Mann-Whitney test were
used when appropriate. Spearman correlations were used to assess the pairwise correlation
of air pollutant exposures. The ambient air pollutant exposures were examined as
continuous variables, while miRNA expression levels were examined as continuous
variables. Of the 83 mitochondrial-associated miRNAs evaluated, expression of 31
miRNAs were above limit of detection in more than half of the total samples. Distribution
and correlation of the 31 miRNAs were also evaluated by time point.

81

To estimate the association between prenatal air pollutant exposures and miRNAs,
a linear model was fitted with miRNA expression level at each time point as the response
and prenatal ambient air pollutants as the exposure variable controlling for potential
confounders. Weekly residential air pollution exposure was assessed based on two
methods, one corresponding the time of miRNA collection, going back 12 weeks from that
date and one corresponding to the gestational week of pregnancy. In order to capture the
acute change of miRNA expression in response to ambient air pollution, only 12 weeks of
air pollution estimates at each time point were evaluated. For first trimester, air pollution
estimates corresponding to the gestational week obtained between gestational week 1 and
gestational week 12 were calculated. For third trimester, air pollution estimates obtained
between gestational week 20 and gestational week 31 were calculated. First trimester and
third trimester maternal blood samples were first modelled separately and then combined
in the same model with an indicator variable Potential confounders, including maternal
age, gestational age at miRNA collection, recruitment site, acculturation (Non-Hispanic,
US-born Hispanic, Foreign-born Hispanic), maternal education, parity, maternal pre-
pregnancy BMI, marital status, and season of birth (defined as warm season if baby was
born between March and September, or cool season otherwise), were chosen for inclusion
in models to evaluate potential confounding based on a priori knowledge and retained in
the models if a 10% change in coefficient estimate was observed. To ascertain the
independent effects of ambient pollutants, 2-pollutant models were also performed.
In order to investigate vulnerable windows of exposure, DLMs were fitted to
examine the time-varying association between weekly averages of air pollutants and
mitochondrial-associated miRNAs. Several functions of DLMs were considered: 1)

82

polynomials (e.g., a cubic order polynomial) 2) spline-based (e.g., natural cubic splines or
B-splines of various degrees). For example, a cubic lag function restricts the lag
coefficients as following:
𝛽 𝑠 = 𝜉 0
+ 𝜉 1
𝑠 + 𝜉 1
𝑠 2
+ 𝜉 2
𝑠 3
, s=0,1,…q
where 𝜉 0
, 𝜉 1
, 𝜉 2
are the parameters of the cubic function describing the lag weight.
The DLM model can be constructed as:
𝑦 𝑡 = 𝛼 + ∑ 𝛽 𝑠 𝑥 𝑡 −𝑠 + 𝑢 𝑡 𝑞 𝑠 =0

where 𝛽 𝑠 are the distributed lag weights, 𝑥 𝑡 −𝑠 are the lagged values of the air pollution
exposures, q is the length of the lag and 𝑢 𝑡 is the noise. The DLMs were compared using
the Akaike information criterion (AIC).
In addition, we fitted a linear model to evaluate the association between miRNA at
each time point and birthweight adjusting for potential confounders. Again, first trimester
and third trimester maternal blood samples were first modelled separately and then
combined in the same model with an indicator variable. The final model was adjusted for
maternal age, gestational age at miRNA collection, recruitment site, acculturation,
maternal education, parity, maternal pre-pregnancy BMI, marital status, infant gender, and
delivery methods. In sensitivity analyses, additional models were run 1) further adjusting
for chip variations, 2) further adjusting for storage time of miRNA samples, 3) excluding
self-reported birthweight. A type I error through the use of the false discovery rate (FDR)
of 0.05 were considered significant. Statistical analyses were performed using and R
version 3.5 software (R project for statistical computing).

83

4.4 Results

4.4.1 Characteristics of the Study Population and Air Pollution Exposures
As of August 29, 2019, a total of 187 samples from 138 MADRES study
participants who had ambient air pollution estimates available were included in the current
study. Baseline characteristics of the 137 study participants are shown in Table 2. Mothers
were predominantly Hispanic (80.5%) and had completed 12
th
grade or higher (72.2%) in
the full samples. Infants were 55.2% male; 28.6% of the infants were born via C-section.
Compared to those provided first trimester samples, mothers who completed a third
trimester visit were more likely to be older, English-speaking, and normal weight.
Table 2. Characteristics of participants by time point

All participants
(n=137)
First-trimester
samples (n=70)
Third trimester
samples (n=115)

N (%) or mean
(SD)
N (%) or mean
(SD)
N (%) or mean
(SD)
Mother characteristics

Maternal age (years) 29.4 (5.7) 28.9 (5.8) 29.6 (5.6)
Race/ethnicity

White, non-Hispanic 9 (6.8) 3 (4.3) 9 (8.0)
Black, non-Hispanic 12 (9.0) 6 (8.7) 9 (8.0)
Hispanic 107 (80.5) 56 (81.2) 92 (82.1)
Other, non-Hispanic 5 (3.8) 4 (5.8) 2 (1.9)
Acculturation

Non-Hispanic 26 (20.5) 13 (18.8) 20 (18.9)
US-born Hispanic 41 (32.3) 22 (31.9) 34 (32.1)
Foreign-born Hispanic 60 (47.2) 34 (49.3) 52 (49.1)
English-speaking 56 (40.9) 27 (38.6) 48 (41.7)
Marital status

Married 37 (30.3) 18 (26.5) 33 (31.7)
Living together 57 (46.7) 33 (48.5) 49 (47.1)
Never married, single 25 (20.5) 14 (20.6) 17 (16.3)
Other 3 (2.5) 3 (4.4) 5 (4.9)
Complete 12
th
grade or higher 96 (72.2) 51 (73.9) 83 (74.1)
First-born 52 (41.6) 30 (43.5) 40 (38.5)
Pre-pregnancy maternal BMI

84

All participants
(n=137)
First-trimester
samples (n=70)
Third trimester
samples (n=115)

N (%) or mean
(SD)
N (%) or mean
(SD)
N (%) or mean
(SD)
Normal/underweight 43 (36.4) 26 (38.2) 38 (39.2)
Overweight/obese 75 (63.6) 42 (61.8) 59 (60.8)
Infant characteristics

Gestational age (weeks) 38.7 (1.8) 38.7 (1.8) 38.9 (1.7)
Birthweight (grams) 3204 (615) 3220 (707) 3250 (620)
Male 74 (55.2) 46 (68.7) 65 (57.0)
Delivery method

Spontaneous vaginal
delivery 82 (61.7) 47 (69.1) 71 (63.4)
C-section 38 (28.6) 16 (23.5) 31 (27.7)
Other 13 (9.7) 5 (7.4) 10 (8.9)
Abbreviations: BMI, body mass index. Missing values were not included.

The distribution of overall 12-week ambient air pollution by trimester is presented
in Table 3. The air pollution estimates anchored to the miRNA collection data and the air
pollution estimates based on the actual gestational weeks were similar. We did not observe
significant deviation between first trimester samples and third trimester samples either. In
general, NO2 and O3 were inversely correlated (pearson coefficient=-0.70, p<0.05, Figure
S1), whereas PM2.5 and PM10 were positively correlated (pearson coefficient=0.90, p<0.05,
Figure S1).

85

Table 3. Distribution of ambient air pollution exposures (12-week average)
Mean SD Median IQR
Prior to MiRNA collection date

NO2 (ppb)

Trimester 1 18.7 6.8 17.8 10.4
Trimester 3 17.4 6.0 16.3 7.8
O3 (ppb)

Trimester 1 44.2 8.5 45.6 10.7
Trimester 3 44.3 8.0 46.3 8.0
PM2.5 (μg/m
3
)

Trimester 1 12.4 2.7 12.2 2.6
Trimester 3 12.2 2.3 11.6 2.2
PM10 (μg/m
3
)

Trimester 1 32.0 7.0 31.5 9.7
Trimester 3 31.7 5.9 31.4 6.0
Based on gestational weeks
NO2 (ppb)

Trimester 1 18.6 6.0 18.3 8.5
Trimester 3 17.7 6.1 16.8 8.6
O3 (ppb)

Trimester 1 44.1 8.6 46.0 12.3
Trimester 3 44.2 8.1 46.3 10.2
PM2.5 (μg/m
3
)

Trimester 1 12.2 2.2 12.1 2.1
Trimester 3 12.2 2.5 11.7 2.0
PM10 (μg/m
3
)

Trimester 1 31.0 7.2 31.6 9.2
Trimester 3 31.6 6.2 31.4 6.2
MiRNA collection date: 12-week air pollution average prior to miRNA collection date.
Gestational weeks: 12-week air pollution average based on gestational week. For first
trimester, gestational week 1-12 were calculated; for third trimester, gestational week 20-
31 were calculated.

4.4.2 Individual Models for Association of Air Pollution and Mitochondrial-Associated
MiRNAs
The correlation of 31 candidate miRNAs and ambient air pollution estimated using
both methods vary by miRNAs by time point. We observed different expression patterns
of first trimester miRNAs and third trimester miRNAs. In the multivariate analysis, PM2.5
exposure averaging 12 weeks prior to miRNA collection was most significantly associated

86

with miRNA expression profiled during third trimester. Two of the associations remained
significantly after correcting for multiple testing. As shown in Figure 1A, miR-23b-3p was
positively associated with PM2.5 exposure (per 5 μg/m
3
increase) during third trimester
(β=9.86, 95% CI 3.84, 15.87, p-adj=0.02). Additionally, at FDR adjustment of 0.05, one
miRNA miR-29a-3b was also found to be positively associated with PM2.5 during third
trimester (β=8.79, 95% CI 3.77, 13.81, p-adj=0.02). However, when we combined first
trimester samples and third trimester samples together, the strength of both associations
attenuated and the associations were no longer significant. Figure 1 shows the changes of
miRNA-23b and miR-29a in association with PM2.5 and NO2. More significant associations
were observed while using air pollution exposures estimated corresponding to the
gestational age. In addition to PM2.5, miR-23b and miR-29a were positively associated with
NO2 after correcting for multiple testing (p-adj<0.05, Figure 1B). A 12 ppb increase in NO2
was associated with higher expression of miR-199a (β=15.11, 95% CI 5.00, 25.21, p-
adj=0.04) and lower expression of miR-122 (β=-62.56, 95% CI -106.35, -18.77, p-
adj=0.04). In sensitivity analyses, we did not observe major changes of our significant
results when further adjusting for storage time of miRNA samples and excluding self-
reported birthweight participants. Further adjustment of chip variations attenuated the
effects of the associations and remained marginally significant.

87

Figure 1. Association between NO2 and PM2.5 exposures and miR-23b and miR-29a. A)
Estimated exposures to prenatal air pollution correspond to the 12-week average prior to
miRNA collection. B) Estimated exposures to prenatal air pollution correspond to the 12-
week of gestational age (gestational weeks 1-12 for first trimester and gestational week 20-
31 for third trimester). Estimated percent difference in miRNA expression are shown for a
12ppb increase in NO2 and a 5 μg/m
3
increase in PM2.5. Models adjusted gestational age at
sample collection, recruitment site, birth season, maternal age, parity, maternal pre-
pregnancy BMI, education, acculturation, infant gender, and marital status. Sample sizes
for first trimester samples and third trimester samples are 70 and 115, respectively.

4.4.3 Multipollutant Models for Association of Air Pollution and Mitochondrial-
Associated MiRNAs
We further evaluated the association between PM2.5 and NO2 and miR-23b, miR-
29a, miR-199a, and miR-122 in 2-pollutant models to ascertain independent effects of NO2
and PM2.5. Results from multi-pollutant models were similar to those from individual
pollutant models. Multi-pollutant models suggested that adjustment for a second pollutant

88

attenuated the association between PM2.5 and NO2 with miRNA expression. For example,
a 5 μg/m
3
increase in PM2.5 was associated with 7.52 (95%: 0.61, 14.43) higher miR-29b
expression when adjusting for NO2 (see Supplement Material, Table S2) while the single-
pollutant model showed an 8.79 increase in miR-29b expression per 5 μg/m
3
increase in
PM2.5.
4.4.4 Distributed Lag Models for Association of Air Pollution and Mitochondrial-
Associated MiRNAs
Natural cubic spline DLMs with 4 df (i.e. 2 knots placed at week 4 and week 9)
were used for primary analyses. As we did not observe a significant difference by using
the air pollution estimates calculated based on miRNA collection date and the air pollution
estimates calculated based on gestational weeks, we only included results from the DLMs
using the gestational weekly air pollution estimates. Results of DLMs for PM2.5 and NO2
with miR-23b, miR-29a, miR-199a, and miR-122 are shown in Figure 2. Exposure to PM2.5
during week 20-22 of pregnancy was positively and significantly associated with miR-23b
expression, while NO2 exposure was positively and significantly associated with miR-23b
expression during week 26. A similar trend for miR-29a was observed where PM2.5 and
NO2 were both positively associated with miR-29a in early third trimester. The strongest
association between NO2 and miR-199a was observed for exposure during week 23-25 of
pregnancy, and the strongest association between PM2.5 and miR-122 was observed for
exposure towards week 29-31. Results were similar for polynomial DLMs and natural
cubic spline DLMs with a knot placed at the week 6 for first trimester and week 25 for
third trimester (see Supplemental Material, Figure S2 and Figure S3).

89

Figure 2. Association of A) miR-23b, B) miR-29a, C) miR-199a, and D) miR-122 with
NO2 and PM2.5 exposures from distributed lag models using air pollution estimates
corresponding to gestational week 20-31. Estimated percent difference in miRNA
expression are shown for a 12ppb increase in NO2 and a 5 (μg/m
3
) increase in PM2.5.
Models adjusted gestational age at sample collection, recruitment site, birth season,
maternal age, parity, maternal pre-pregnancy BMI, education, acculturation, infant gender,
and marital status.

90

4.4.5 Association of Mitochondrial-Associated MiRNAs with Birthweight
We did not observe any significant associations between mitochondrial-associated
miRNAs and birthweight after correcting for multiple testing. One miRNA miR-107
profiled during first trimester was negatively associated with birthweight (β=-20.96, 95%
CI -36.53, -5.39, p-raw=0.01, p-adj=0.29). MiR-23b, miR-29a, miR-199a during third
trimester were all negatively associated with birthweight; however, none of the nominal
associations were significant. Results from the birthweight model are shown in Supplement
Material Table S3.
4.4.5 MiRNA Target Prediction
A list of predicted targets compiled from mirTarBase and DIANA-TarBase with
description of the target genes is provided in Table 3 and Table S4. Integrated score of
mitochondrial localization compiles utilizing Human MitoCarta 2.0 database was also
reviewed for the target genes in Table 3. The gene with strongest mitochondrial localization
evidence is UQCRFS1 targeted by miR-23b, followed by ATP5A1, UQCRC, and SDHD
all targeted by miR-23b.
Table 3. Target gene prediction of miR-23b and miR-29a
miRNA Target genes Description
miR-23b AARS2 Alanyl-tRNA synthetase 2, mitochondrial

ACAT1 Acetyl-CoA acetyltransferase 1

ACSS3 Acyl-CoA synthetase short-chain family member 3

ATP5A1 ATP synthase, H+ transporting, mitochondrial F1 complex, alpha
subunit 1, cardiac muscle

BCL2L1 BCL2-like 1

BCS1L BC1 (ubiquinol-cytochrome c reductase) synthesis-like

C2orf69 Chromosome 2 open reading frame 69

CCT7 Chaperonin containing TCP1, subunit 7 (eta)

COX1 Cytochrome c oxidase subunit I

CRLS1 Cardiolipin synthase 1

91

ECH1 Enoyl CoA hydratase 1, peroxisomal

GAPDH Glyceraldehyde-3-phosphate dehydrogenase

GHITM Growth hormone inducible transmembrane protein

HK1 Hexokinase 1

MPV17 MpV17 mitochondrial inner membrane protein

NDUFV3 NADH dehydrogenase (ubiquinone) flavoprotein 3, 10kDa

PANK2 Pantothenate kinase 2

PRDX3 Peroxiredoxin 3

SDHD Succinate dehydrogenase complex, subunit D, integral membrane
protein

SLC16A1 Solute carrier family 16 (monocarboxylate transporter), member 1

SOD2 Superoxide dismutase 2, mitochondrial

STARD7 StAR-related lipid transfer (START) domain containing 7

UQCRC2 Ubiquinol-cytochrome c reductase core protein II

UQCRFS1 Ubiquinol-cytochrome c reductase, Rieske iron-sulfur polypeptide 1
miR-29a ALDH5A1 Aldehyde dehydrogenase 5 family, member A1

BCL2 B-cell CLL/lymphoma 2

CASP8 Caspase 8, apoptosis-related cysteine peptidase

COX7A2L Cytochrome c oxidase subunit VIIa polypeptide 2 like

CYTB Cytochrome b

DIABLO Diablo, IAP-binding mitochondrial protein

EPHX2 Epoxide hydrolase 2, cytoplasmic

METTL15 Methyltransferase like 15

SLC16A1 Solute carrier family 16 (monocarboxylate transporter), member 1

VDAC1 Voltage-dependent anion channel 1

4.5 Discussion
In the current study, short-term prenatal exposure to ambient air pollutant PM2.5
and NO2 was associated with higher expression of miR-23b, miR-29a, and miR-199a and
lower expression of miR-122 in maternal blood during third trimester. Ozone and PM10
were not associated with differential patterns of mitochondrial-associated miRNAs after
correcting for multiple testing. Distributed lag models revealed that susceptible windows
of short-term ambient air pollutant to differential expression patterns of miRNAs vary by
miRNAs. No consistent time windows were observed among the significant miRNAs. In

92

addition, maternal blood miRNAs profiled during first trimester and third trimester were
not significantly associated with newborn birthweight. The findings of our current study
indicate that air pollutants can affect mitochondrial-associated miRNA expression in
maternal blood during pregnancy.
Previous in vitro and human studies have observed differential expression patterns
of miRNAs in response to environmental stressors and few have studied the effect of PM
on miRNAs [18, 20, 23, 25]. The most frequently studied miRNAs in response to PM
exposure are miR-16, miR-20a, miR-21, miR-34a, miR-146a and miR-222. Studies
focused on those candidate miRNAs have shown that higher levels of these miRNAs tend
to be associated with less expression of their target mRNAs [24]. In addition, some of the
studies have gone on to demonstrate that these miRNAs may alter pathways involved in
maternal decidua, migration and growth of placental cells [25]. During specific period of
pregnancy, air pollution exposures, specifically PM2.5 exposures have been linked to
alterations of candidate miRNA related to cell growth control in the placenta [25]. In our
study, we found a positive association between PM2.5 and novel maternal circulating
miRNAs during third trimester. Our study is one of the first to investigate the impact of
ambient air pollution on maternal miRNAs during pregnancy in a healthy study population.
In addition, we used DLMs, which account for both current and past exposure, and
observed that miRNA expression changes may be specific to time windows of exposure.
Previously, Tsamou et al. reported a trimester-specific effect of ambient air pollution on
placental miRNAs and hypothesized that each developmental time window has its own
physiological events regulated by different molecular process [25]. Our findings further
suggest that chemical stressors in the maternal environment can alter circulating miRNA

93

patterns during specific time windows of exposure, providing additional evidence of the
important functional role miRNAs may have during fetal development.
To our knowledge, this is the first study to examine the role of miRNAs localized
in mitochondria and its response to environmental exposure. A number of studies have
demonstrated possible localization of miRNAs in mitochondria [8, 11, 45]. It has been
hypothesized that the contribution of miRNAs to mitochondria can be considered as 1)
nuclear-encoded miRNAs acting the in nucleus or in the cytosol on genes encoding
mitochondria proteins, or 2) as nuclear-encoded miRNAs acting at mitochondria, or 3) as
mitochondrially encoded miRNAs acting within the organelle. However, the extent to
which miRNA localization in the mitochondria is regulated remains largely unknown.
Studies investigating the function of mitomiRs have suggested that mitomiRs may
modulate OXPHOS, contribute to post-transcriptional regulation of gene expression
related to mitochondrial energetics and apoptosis, control mitochondrial ATP synthesis
coupled to the respiratory chain, translation, metabolic processes, and progression of cell
cycle [8, 11, 45]. Our positive findings of mitomiRs miR-23b, miR-29a, miR-199a and
miR-122 with air pollution add to a new body of literature as one step forward in
understanding the possibility that environmental exposures may influence mitochondrial
related miRNAs by providing evidence of associations from an epidemiologic study.
There is a growing body of literature highlighting the regulation of mitochondrial-
associated miRNAs on mitochondrial metabolism and its possible pathophysiological
impact. Previous works have shown that miR-122 is highly abundant in liver and has major
effects on lipid metabolism [15, 46, 47]. In addition, over-expression of miR-199a has been
found to reduce mitochondrial content and increased fatty acid deposition [12].

94

Furthermore, several putative target genes of miR-23b, miR-29a, miR-199a, and miR-122
are known to influence mitochondrial metabolic pathways. For example, COXI, regulated
by miR-23b, is a subunit of the electron transport chain complex IV. Animal studies have
revealed that miR-181c-5p can repress expression of COXI gene by 30% [48]. One of the
target gene of miR-23b with strongest evidence of mitochondrial localization ATP5A1
gene encodes a subunit of mitochondrial ATP synthase. Mitochondrial ATP synthase
catalyzes ATP synthesis, using an electrochemical gradient of protons across the inner
membrane during oxidative phosphorylation. Recently, mutations in APT synthases have
been found to be present in severe metabolic disorders [49, 50]. Additionally, genome-
wide methylation study has shown gestational diabetes mellitus (GDM) associated
epigenetic changes in cord blood on ATP5A1 gene region, suggesting a role of ATP5A1
in regulation of fetal programming. However, we did not observe a direct effect of miRNAs
on newborn birthweight in current study. As the field of mitochondrial miRNAs is growing
and current investigations on mitomiRs are largely studied in cell culture models and
animal models, it is possible that the effect of mitomiRs on mitochondrial metabolism is
cell-specific or tissue-specific. Future investigation of mitomiRs in larger epidemiological
studies are warranted.
We acknowledge several limitations of the current study. Generalizability of the
study to other population is limited as our study population is predominantly healthy
Hispanic White women. In addition, expression level of mRNAs was not measured in this
study. We were not able to demonstrate the observed changes in miRNAs corresponds to
changes of target mRNAs. There are known challenges with regards to current
methodology in miRNA extraction and measurement, including the effect of blood draw

95

timing, the stability of miRNAs, and cross-hybridization of miRNAs[51]. Our study is also
subject to those methodological challenges.
In conclusion, we observed significant associations between ambient air pollution
exposure and maternal mitochondrial-associated miRNAs (miR-23b, miR-29a, miR-199a,
and miR-122) during third trimester. Different time windows of exposures were also
identified, suggesting the possible developmental specific changes of miRNAs in response
to air pollution during fetal growth. Lastly, given the suggested role of mitomiRs on
metabolic pathways, future studies should be warranted in larger epidemiological studies
to further evaluate the relationship between mitomiRs and newborn growth.

96

Supplemental Material
Figure S1. Spearman correlation between air pollution exposures estimated by miRNA
collection date and estimated by gestational age weeks
Figure S2. Results from polynomial distributed lag models
Figure S2. Results from natural cubic spline distributed lag models
Table S1. Nominal Association between ambient air pollution exposures and miRNAs
from single-pollutant model
Table S2. Association between ambient air pollution exposures and significant miRNAs
from multi-pollutant model
Table S3. Association between significant mitochondrial-associated miRNAs and
birthweight
Table S4. Target gene prediction of miR-199a and miR-122
Appendix I. List of mitochondrial-associated miRNAs

97

Figure S1. Spearman correlation between air pollution exposures estimated by miRNA
collection date and estimated by gestational age weeks. Exposures estimated by miRNA
collection date are annotated as _12weeks while exposures estimated by gestational
weeks are annotated as _GAweek.

98

Figure S2. Association of A) miR-23b, B) miR-29a, C) miR-199a, and D) miR-122 with
NO2 and PM2.5 exposures from polynomial distributed lag models (cubic function) using
air pollution estimates corresponding to gestational week 20-31. Estimated percent
difference in miRNA expression are shown for a 12ppb increase in NO2 and a 5 (μg/m
3
)
increase in PM2.5. Models adjusted gestational age at sample collection, recruitment site,
birth season, maternal age, parity, maternal pre-pregnancy BMI, education, acculturation,
infant gender, and marital status.

99

Figure S3. Association of A) miR-23b, B) miR-29a, C) miR-199a, and D) miR-122 with
NO2 and PM2.5 exposures from natural cubic spline distributed lag models (df=3, a knot
placed at week 25) using air pollution estimates corresponding to gestational week 20-31.
Estimated percent difference in miRNA expression are shown for a 12ppb increase in
NO2 and a 5 (μg/m
3
) increase in PM2.5. Models adjusted gestational age at sample
collection, recruitment site, birth season, maternal age, parity, maternal pre-pregnancy
BMI, education, acculturation, infant gender, and marital status.

100

Table S1. Nominal Association between ambient air pollution exposures and miRNAs
from single-pollutant model
Outcome Exposure Estimate
Lower
CI
Upper
CI p-raw p-adj
First trimester

miR-130a-3p PM10 -57.23 -92.22 -22.25 0.00 0.06
miR-127-3p NO2 10.58 1.86 19.30 0.02 0.57
miR-130a-3p PM2.5 -60.26 -115.38 -5.15 0.03 0.64
miR-127-3p PM10 6.07 0.08 12.06 0.05 0.42
Third trimester

miR-29a-3p PM2.5 7.74 2.94 12.55 0.00 0.06
miR-29a-3p NO2 7.55 2.85 12.26 0.00 0.03
miR-23b-3p NO2 8.91 3.32 14.50 0.00 0.03
miR-23b-3p PM2.5 8.62 2.88 14.37 0.00 0.06
miR-199a-5p NO2 15.11 5.00 25.21 0.00 0.04
miR-122-5p NO2 -62.56 -106.35 -18.77 0.01 0.04
Air pollution estimates for first trimester correspond to gestational week 1-12 and for
third trimester correspond to gestational week 20-31. Estimated percent difference in
miRNA expression are shown for a 12ppb increase in NO2 and a 5 (μg/m
3
) increase in
PM2.5. Models adjusted gestational age at sample collection, recruitment site, birth
season, maternal age, parity, maternal pre-pregnancy BMI, education, acculturation,
infant gender, and marital status. Sample sizes for first trimester samples and third
trimester samples are 70 and 115, respectively.

101

Table S2. Association between ambient air pollution exposures and significant miRNAs
from multi-pollutant model
Outcome Exposure Estimate
Lower
CI Upper CI p-raw
MiRNA collection date
miR-23b-3p
NO2 4.30
-3.38 11.99 0.27
PM2.5 9.86
3.84 15.87 0.00
miR29a-3p
NO2 1.74
-4.72 8.20 0.59
PM2.5 8.79
3.77 13.82 0.00
miR-199a-
5p
NO2 17.96
3.90 32.02 0.01
PM2.5 7.65
-3.65 18.95 0.18
miR-122-5p
NO2 -65.11
-126.28 -3.95 0.04
PM2.5 -36.95
-85.63 11.72 0.13
Gestational week

miR-23b-3p
NO2 5.92
-1.55 13.39 0.12
PM2.5 4.59
-3.05 12.23 0.24
miR29a-3p
NO2 4.51
-1.75 10.78 0.16
PM2.5 7.74
2.94 12.55 0.00
miR-199a-
5p
NO2 18.26
4.68 31.84 0.01
PM2.5 7.59
-3.11 18.29 0.16
miR-122-5p
NO2 -57.46
-116.45 1.54 0.06
PM2.5 -46.96
-92.63 -1.29 0.04
Estimated percent difference in miRNA expression are shown for a 12ppb increase in
NO2 and a 5 (μg/m
3
) increase in PM2.5. Models adjusted gestational age at sample
collection, recruitment site, birth season, maternal age, parity, maternal pre-pregnancy
BMI, education, acculturation, infant gender, and marital status. Sample sizes for first
trimester samples and third trimester samples are 70 and 115, respectively.

102

Table S3. Association between significant mitochondrial-associated miRNAs and
birthweight
miRNA Estimate Lower CI Upper CI p-raw p-adj
First trimester

miR-23b-3p -9.00 -22.16 4.16 0.18 0.97
miR-29a-3p 5.41 -11.44 22.27 0.52 0.97
miR-199a-5p -0.55 -7.90 6.81 0.88 0.97
miR-122-5p 1.38 -0.92 3.68 0.23 0.97
Third trimester

miR-23b-3p -2.48 -12.44 7.48 0.62 0.99
miR-29a-3p -4.95 -16.36 6.47 0.39 0.99
miR-199a-5p -1.14 -6.52 4.23 0.67 0.99
miR-122-5p 0.46 -0.79 1.71 0.46 0.99
Estimated percent difference in birthweight are shown for one unit change of miRNA
expression. Models adjusted gestational age at sample collection, recruitment site,
delivery method, maternal age, parity, maternal pre-pregnancy BMI, education,
acculturation, infant gender, and marital status. Sample sizes for first trimester samples
and third trimester samples are 70 and 115, respectively.

103

Table S4. Target gene prediction of miR-199a and miR-122
miRNAs Target genes Description
miR-
199a
CHCHD4 coiled-coil-helix-coiled-coil-helix domain containing 4

COX10 cytochrome c oxidase assembly homolog 10 (yeast)

COX15 cytochrome c oxidase assembly homolog 15 (yeast)

FXN frataxin

HK2 hexokinase 2

MCCC2 methylcrotonoyl-CoA carboxylase 2 (beta)

MTRF1L mitochondrial translational release factor 1-like

NDUFS2 NADH dehydrogenase (ubiquinone) Fe-S protein 2, 49kDa (NADH-
coenzyme Q reductase)

OCIAD2 OCIA domain containing 2

PLGRKT plasminogen receptor, C-terminal lysine transmembrane protein

PTCD2 pentatricopeptide repeat domain 2

SLC30A6 solute carrier family 30 (zinc transporter), member 6

SOD2 superoxide dismutase 2, mitochondrial

TFAM transcription factor A, mitochondrial

UNG uracil-DNA glycosylase
miR-122 ABCF2 ATP-binding cassette, sub-family F (GCN20), member 2

ACOT9 acyl-CoA thioesterase 9

ACP6 acid phosphatase 6, lysophosphatidic

AK3 adenylate kinase 3

ARL2 ADP-ribosylation factor-like 2

BAX BCL2-associated X protein

BCL2L1 BCL2-like 1

CEP89 centrosomal protein 89kDa

CHDH choline dehydrogenase

CISD2 CDGSH iron sulfur domain 2

CLIC4 chloride intracellular channel 4

CS citrate synthase

DBT dihydrolipoamide branched chain transacylase E2

FAM210B family with sequence similarity 210, member B

FUNDC2 FUN14 domain containing 2

GLOD4 glyoxalase domain containing 4

GRSF1 G-rich RNA sequence binding factor 1

HCCS holocytochrome c synthase

IBA57 IBA57, iron-sulfur cluster assembly homolog (S. cerevisiae)

KIAA0100 KIAA0100

LYRM4 LYR motif containing 4

MPV17 MpV17 mitochondrial inner membrane protein

104

MRPL17 mitochondrial ribosomal protein L17

MRPL52 mitochondrial ribosomal protein L52

MRPS23 mitochondrial ribosomal protein S23

MRPS25 mitochondrial ribosomal protein S25

MTO1 mitochondrial tRNA translation optimization 1

PDK4 pyruvate dehydrogenase kinase, isozyme 4

PDP2 pyruvate dehyrogenase phosphatase catalytic subunit 2

PTCD3 pentatricopeptide repeat domain 3

PXMP4 peroxisomal membrane protein 4, 24kDa

ROMO1 reactive oxygen species modulator 1

RPS15A ribosomal protein S15a

SERAC1 serine active site containing 1

SLC25A30 solute carrier family 25, member 30

SLC25A33 solute carrier family 25 (pyrimidine nucleotide carrier), member 33

SUCLA2 succinate-CoA ligase, ADP-forming, beta subunit

TIMM50 translocase of inner mitochondrial membrane 50 homolog (S.
cerevisiae)
YME1L1 YME1-like 1 ATPase

105

Appendix I. List of mitochondrial-associated miRNAs
Mitochondrial-
associated miRNAs
Number of
samples above
LOD
hsa-let-7b-5p 326
hsa-let-7c-5p 134
hsa-let-7d-5p 322
hsa-let-7f-5p 149
hsa-let-7g-5p 325
hsa-miR-1-3p 31
hsa-miR-107 202
hsa-miR-122-5p 304
hsa-miR-1224-5p 18
hsa-miR-1246 18
hsa-miR-125a-3p 62
hsa-miR-125a-5p 101
hsa-miR-125b-5p 123
hsa-miR-126-3p 326
hsa-miR-127-3p 199
hsa-miR-1275 25
hsa-miR-128-3p 12
hsa-miR-130a-3p 323
hsa-miR-130b-3p 18
hsa-miR-133a-3p 102
hsa-miR-133a-5p 128
hsa-miR-133b 15
hsa-miR-139-3p 88
hsa-miR-140-5p 130
hsa-miR-144-3p 279
hsa-miR-146a-5p 320
hsa-miR-148a-3p 264
hsa-miR-149-5p 78
hsa-miR-16-5p 326
hsa-miR-181a-5p 307
hsa-miR-181c-5p 36
hsa-miR-188-5p 188
hsa-miR-1908-5p 37
hsa-miR-1972 178
hsa-miR-1973 162
hsa-miR-1976 86
hsa-miR-199a-5p 319
hsa-miR-200c-3p 81
hsa-miR-210-5p 102
hsa-miR-211-3p 35

106

hsa-miR-212-3p 200
hsa-miR-22-3p 282
hsa-miR-221-5p 40
hsa-miR-23a-3p 326
hsa-miR-23b-3p 204
hsa-miR-24-3p 147
hsa-miR-26a-5p 247
hsa-miR-27a-3p 81
hsa-miR-27b-3p 42
hsa-miR-29a-3p 230
hsa-miR-29b-3p 314
hsa-miR-29c-3p 87
hsa-miR-30a-5p 59
hsa-miR-30d-5p 165
hsa-miR-30e-5p 270
hsa-miR-320a 22
hsa-miR-324-3p 29
hsa-miR-324-5p 12
hsa-miR-328-5p 169
hsa-miR-342-3p 213
hsa-miR-34a-5p 165
hsa-miR-361-5p 103
hsa-miR-365b-5p 45
hsa-miR-423-3p 28
hsa-miR-451a 326
hsa-miR-483-5p 16
hsa-miR-484 26
hsa-miR-490-3p 20
hsa-miR-494-3p 37
hsa-miR-497-5p 130
hsa-miR-501-3p 23
hsa-miR-503-5p 97
hsa-miR-513a-5p 18
hsa-miR-532-3p 109
hsa-miR-542-5p 120
hsa-miR-574-3p 65
hsa-miR-574-5p 219
hsa-miR-638 58
hsa-miR-652-5p 20
hsa-miR-671-5p 46
hsa-miR-675-5p 40
hsa-miR-93-5p 325
hsa-miR-99a-5p 52

107

Chapter 4: References
1. Lee RC, Feinbaum RL, Ambros V: The C. elegans heterochronic gene lin-4
encodes small RNAs with antisense complementarity to lin-14. Cell 1993,
75(5):843-854.
2. Bartel DP, Chen CZ: Micromanagers of gene expression: the potentially
widespread influence of metazoan microRNAs. Nat Rev Genet 2004, 5(5):396-400.
3. He L, Hannon GJ: MicroRNAs: small RNAs with a big role in gene regulation. Nat
Rev Genet 2004, 5(7):522-531.
4. Bushati N, Cohen SM: microRNA functions. Annu Rev Cell Dev Biol 2007, 23:175-
205.
5. Krol J, Loedige I, Filipowicz W: The widespread regulation of microRNA
biogenesis, function and decay. Nat Rev Genet 2010, 11(9):597-610.
6. Bian Z, Li LM, Tang R, Hou DX, Chen X, Zhang CY, Zen K: Identification of
mouse liver mitochondria-associated miRNAs and their potential biological
functions. Cell Res 2010, 20(9):1076-1078.
7. Kren BT, Wong PY, Sarver A, Zhang X, Zeng Y, Steer CJ: MicroRNAs identified
in highly purified liver-derived mitochondria may play a role in apoptosis. RNA
Biol 2009, 6(1):65-72.
8. Barrey E, Saint-Auret G, Bonnamy B, Damas D, Boyer O, Gidrol X: Pre-
microRNA and mature microRNA in human mitochondria. PLoS One 2011,
6(5):e20220.
9. Sripada L, Tomar D, Prajapati P, Singh R, Singh AK: Systematic analysis of small
RNAs associated with human mitochondria by deep sequencing: detailed analysis
of mitochondrial associated miRNA. PLoS One 2012, 7(9):e44873.
10. Mercer TR, Neph S, Dinger ME, Crawford J, Smith MA, Shearwood AM, Haugen
E, Bracken CP, Rackham O, Stamatoyannopoulos JA et al: The human
mitochondrial transcriptome. Cell 2011, 146(4):645-658.
11. Sripada L, Tomar D, Singh R: Mitochondria: one of the destinations of miRNAs.
Mitochondrion 2012, 12(6):593-599.
12. Geiger J, Dalgaard LT: Interplay of mitochondrial metabolism and microRNAs.
Cell Mol Life Sci 2017, 74(4):631-646.
13. Ji J, Qin Y, Ren J, Lu C, Wang R, Dai X, Zhou R, Huang Z, Xu M, Chen M et al:
Mitochondria-related miR-141-3p contributes to mitochondrial dysfunction in
HFD-induced obesity by inhibiting PTEN. Sci Rep 2015, 5:16262.
14. Rottiers V, Najafi-Shoushtari SH, Kristo F, Gurumurthy S, Zhong L, Li Y, Cohen
DE, Gerszten RE, Bardeesy N, Mostoslavsky R et al: MicroRNAs in metabolism
and metabolic diseases. Cold Spring Harb Symp Quant Biol 2011, 76:225-233.
15. Iliopoulos D, Drosatos K, Hiyama Y, Goldberg IJ, Zannis VI: MicroRNA-370
controls the expression of microRNA-122 and Cpt1alpha and affects lipid
metabolism. J Lipid Res 2010, 51(6):1513-1523.
16. Duarte FV, Palmeira CM, Rolo AP: The Role of microRNAs in Mitochondria:
Small Players Acting Wide. Genes (Basel) 2014, 5(4):865-886.
17. Chan SY, Zhang YY, Hemann C, Mahoney CE, Zweier JL, Loscalzo J: MicroRNA-
210 controls mitochondrial metabolism during hypoxia by repressing the iron-
sulfur cluster assembly proteins ISCU1/2. Cell Metab 2009, 10(4):273-284.

108

18. Vrijens K, Bollati V, Nawrot TS: MicroRNAs as potential signatures of
environmental exposure or effect: a systematic review. Environ Health Perspect
2015, 123(5):399-411.
19. Li Q, Kappil MA, Li A, Dassanayake PS, Darrah TH, Friedman AE, Friedman M,
Lambertini L, Landrigan P, Stodgell CJ et al: Exploring the associations between
microRNA expression profiles and environmental pollutants in human placenta
from the National Children's Study (NCS). Epigenetics 2015, 10(9):793-802.
20. Jardim MJ, Fry RC, Jaspers I, Dailey L, Diaz-Sanchez D: Disruption of microRNA
expression in human airway cells by diesel exhaust particles is linked to
tumorigenesis-associated pathways. Environ Health Perspect 2009, 117(11):1745-
1751.
21. Bollati V, Marinelli B, Apostoli P, Bonzini M, Nordio F, Hoxha M, Pegoraro V,
Motta V, Tarantini L, Cantone L et al: Exposure to metal-rich particulate matter
modifies the expression of candidate microRNAs in peripheral blood leukocytes.
Environ Health Perspect 2010, 118(6):763-768.
22. Levanen B, Bhakta NR, Torregrosa Paredes P, Barbeau R, Hiltbrunner S, Pollack
JL, Skold CM, Svartengren M, Grunewald J, Gabrielsson S et al: Altered
microRNA profiles in bronchoalveolar lavage fluid exosomes in asthmatic patients.
J Allergy Clin Immunol 2013, 131(3):894-903.
23. Hou L, Barupal J, Zhang W, Zheng Y, Liu L, Zhang X, Dou C, McCracken JP,
Diaz A, Motta V et al: Particulate Air Pollution Exposure and Expression of Viral
and Human MicroRNAs in Blood: The Beijing Truck Driver Air Pollution Study.
Environ Health Perspect 2016, 124(3):344-350.
24. Fossati S, Baccarelli A, Zanobetti A, Hoxha M, Vokonas PS, Wright RO, Schwartz
J: Ambient particulate air pollution and microRNAs in elderly men. Epidemiology
2014, 25(1):68-78.
25. Tsamou M, Vrijens K, Madhloum N, Lefebvre W, Vanpoucke C, Nawrot TS: Air
pollution-induced placental epigenetic alterations in early life: a candidate miRNA
approach. Epigenetics 2016:0.
26. Wang W, Feng L, Zhang H, Hachy S, Satohisa S, Laurent LC, Parast M, Zheng J,
Chen DB: Preeclampsia up-regulates angiogenesis-associated microRNA (i.e.,
miR-17, -20a, and -20b) that target ephrin-B2 and EPHB4 in human placenta. J
Clin Endocrinol Metab 2012, 97(6):E1051-1059.
27. Guo L, Yang Q, Lu J, Li H, Ge Q, Gu W, Bai Y, Lu Z: A comprehensive survey of
miRNA repertoire and 3' addition events in the placentas of patients with pre-
eclampsia from high-throughput sequencing. PLoS One 2011, 6(6):e21072.
28. Pineles BL, Romero R, Montenegro D, Tarca AL, Han YM, Kim YM, Draghici S,
Espinoza J, Kusanovic JP, Mittal P et al: Distinct subsets of microRNAs are
expressed differentially in the human placentas of patients with preeclampsia. Am
J Obstet Gynecol 2007, 196(3):261 e261-266.
29. Miura K, Higashijima A, Hasegawa Y, Abe S, Miura S, Kaneuchi M, Yoshiura K,
Masuzaki H: Circulating levels of maternal plasma cell-free miR-21 are associated
with maternal body mass index and neonatal birth weight. Prenat Diagn 2015,
35(5):509-511.

109

30. Elovitz MA, Brown AG, Anton L, Gilstrop M, Heiser L, Bastek J: Distinct cervical
microRNA profiles are present in women destined to have a preterm birth. Am J
Obstet Gynecol 2014, 210(3):221 e221-211.
31. Huang L, Shen Z, Xu Q, Huang X, Chen Q, Li D: Increased levels of microRNA-
424 are associated with the pathogenesis of fetal growth restriction. Placenta 2013,
34(7):624-627.
32. Yamamoto M, Singh A, Sava F, Pui M, Tebbutt SJ, Carlsten C: MicroRNA
expression in response to controlled exposure to diesel exhaust: attenuation by the
antioxidant N-acetylcysteine in a randomized crossover study. Environ Health
Perspect 2013, 121(6):670-675.
33. Gude NM, Roberts CT, Kalionis B, King RG: Growth and function of the normal
human placenta. Thromb Res 2004, 114(5-6):397-407.
34. Farrokhnia F, Aplin JD, Westwood M, Forbes K: MicroRNA regulation of
mitogenic signaling networks in the human placenta. J Biol Chem 2014,
289(44):30404-30416.
35. Manaster I, Goldman-Wohl D, Greenfield C, Nachmani D, Tsukerman P, Hamani
Y, Yagel S, Mandelboim O: MiRNA-mediated control of HLA-G expression and
function. PLoS One 2012, 7(3):e33395.
36. Ouyang Y, Mouillet JF, Coyne CB, Sadovsky Y: Review: placenta-specific
microRNAs in exosomes - good things come in nano-packages. Placenta 2014, 35
Suppl:S69-73.
37. Jiang H, Wu W, Zhang M, Li J, Peng Y, Miao TT, Zhu H, Xu G: Aberrant
upregulation of miR-21 in placental tissues of macrosomia. Journal of
perinatology : official journal of the California Perinatal Association 2014,
34(9):658-663.
38. Maccani MA, Padbury JF, Marsit CJ: miR-16 and miR-21 expression in the
placenta is associated with fetal growth. PLoS One 2011, 6(6):e21210.
39. Tang Q, Wu W, Xu X, Huang L, Gao Q, Chen H, Sun H, Xia Y, Sha J, Wang X et
al: miR-141 contributes to fetal growth restriction by regulating PLAG1
expression. PloS one 2013, 8(3):e58737.
40. Kunzli N, Jerrett M, Garcia-Esteban R, Basagana X, Beckermann B, Gilliland F,
Medina M, Peters J, Hodis HN, Mack WJ: Ambient air pollution and the
progression of atherosclerosis in adults. PloS one 2010, 5(2):e9096.
41. Kunzli N, Jerrett M, Mack WJ, Beckerman B, LaBree L, Gilliland F, Thomas D,
Peters J, Hodis HN: Ambient air pollution and atherosclerosis in Los Angeles.
Environ Health Perspect 2005, 113(2):201-206.
42. Fenton TR, Kim JH: A systematic review and meta-analysis to revise the Fenton
growth chart for preterm infants. BMC Pediatr 2013, 13:59.
43. Carbon S, Ireland A, Mungall CJ, Shu S, Marshall B, Lewis S: AmiGO: online
access to ontology and annotation data. Bioinformatics 2009, 25(2):288-289.
44. Calvo SE, Clauser KR, Mootha VK: MitoCarta2.0: an updated inventory of
mammalian mitochondrial proteins. Nucleic Acids Res 2016, 44(D1):D1251-1257.
45. Bandiera S, Mategot R, Girard M, Demongeot J, Henrion-Caude A: MitomiRs
delineating the intracellular localization of microRNAs at mitochondria. Free
Radic Biol Med 2013, 64:12-19.

110

46. Esau C, Davis S, Murray SF, Yu XX, Pandey SK, Pear M, Watts L, Booten SL,
Graham M, McKay R et al: miR-122 regulation of lipid metabolism revealed by in
vivo antisense targeting. Cell Metab 2006, 3(2):87-98.
47. Moore KJ, Rayner KJ, Suarez Y, Fernandez-Hernando C: The role of microRNAs
in cholesterol efflux and hepatic lipid metabolism. Annu Rev Nutr 2011, 31:49-63.
48. Das S, Ferlito M, Kent OA, Fox-Talbot K, Wang R, Liu D, Raghavachari N, Yang
Y, Wheelan SJ, Murphy E et al: Nuclear miRNA regulates the mitochondrial
genome in the heart. Circ Res 2012, 110(12):1596-1603.
49. Hejzlarova K, Mracek T, Vrbacky M, Kaplanova V, Karbanova V, Nuskova H,
Pecina P, Houstek J: Nuclear genetic defects of mitochondrial ATP synthase.
Physiol Res 2014, 63 Suppl 1:S57-71.
50. Jonckheere AI, Renkema GH, Bras M, van den Heuvel LP, Hoischen A, Gilissen
C, Nabuurs SB, Huynen MA, de Vries MC, Smeitink JA et al: A complex V
ATP5A1 defect causes fatal neonatal mitochondrial encephalopathy. Brain 2013,
136(Pt 5):1544-1554.
51. Moldovan L, Batte KE, Trgovcich J, Wisler J, Marsh CB, Piper M: Methodological
challenges in utilizing miRNAs as circulating biomarkers. J Cell Mol Med 2014,
18(3):371-390.

111

CHAPTER 5
Mitochondrial Genetic Variants Are Associated with Growth in Children
5.1 Abstract
Mitochondria have been identified as one of the key players in the regulation of
early development and have been implicated in various metabolic disorders including
diabetes and obesity. Emerging evidence has shown that mitochondrial single nucleotide
polymorphisms (mtSNPs) may contribute to the obesity risk. We investigated the
associations between 327 mtSNPs and BMI among children enrolled in Children’s Health
Study (CHS). For confirmation, we analyzed 226 mtSNPs in a separate population of
children from the EVE Consortium on the Genetics of Asthma. The genotyping data of
CHS study population were obtained from Illumina HumanExome BeadChip or a genome-
wide association study (GWAS), whereas the genotyping of EVE replication population
was performed on Illumina HumanExome BeadChip. We identified one significant mtSNP
mt15301 G→A associated with BMI at last year of CHS follow-up and BMI growth across
the ten-year follow-up period. Five additional mtSNPs were associated with BMI during
last year of their follow-up, one of which was also identified in the replication study.
Subgroup analysis further revealed that the observed association was likely to be driven by
ethnicity. These findings provide evidence that genetic variants in the mitochondrial
genome may play a role in the mitochondrial function and in turn influence metabolism.
Further studies should be warranted to investigate ethnic-specific effect of mtSNPs on
obesity risk of children and BMI.

112

5.2 Mitochondrial Genetic Variations
5.2.1 Genetic Variants in Mitochondria
In the 1980’s, it was discovered that mitochondrial DNA (mtDNA) point mutations
and deletion could result in human disease, a breakthrough for molecular medicine [1].
Since the landmark breakthrough, it has been established that inherited and acquired
mtDNA defects, in addition to mutations in autosomal mitochondrial genes in the nucleus,
are the origin of heterogeneous pediatric and adult diseases. Because a cell carries many
mitochondria, and also the mitochondrial genome has a higher mutation rate than the
nuclear genome, there is heterogeneity of the relative frequency of a mtDNA variant within
an individual tissue, cell, and even within the same mitochondrion [2]. A large number of
pathogenic mtDNA mutations have been identified and the more severe mutations are
frequently mixed with normal mtDNAs within the cell, a state known as heteroplasmy. As
the percentage of mutant mtDNAs increases, the resulting bioenergetic defect becomes
increasingly severe and the degree of severity of the clinical phenotype increases [2, 3].
mtDNA heteroplasmy has been found in almost every healthy individual who has been
studies, albeit at a very low level, raising the possibility that some of the mutations found
late in life are actually low-level heteroplasmies that been inherited from the maternal side
[3].
mtSNPs have been historically studied in haplogroups. A mtDNA haplogroup is a
collection of mtSNPs at certain genetic loci accumulated throughout human history that
could be attributed to genetic drift [4]. Various mtDNA haplogroups that arose from each
continent and geographical region are functionally different. The adaptations of regional
environment, migration, diet changes or new infant agents can result in different clinical

113

phenotypes [5]. Thus, haplogroups are important in normal physiological adaption as well
as in modulating risk of developing diseases [6-8]. For example, haplogroup I, J, K, T and
U have been found to be associated with increased risk of autism spectrum disorder (ASD)
[9]. It was also suggested that interaction between haplogroups, mtDNA mutations, and
environment factors that modulate bioenergetics altogether may contribute to ASD risk [9].
Figure 1. Regional distribution of human mtDNA haplogroup from their origin
(reproduced from MITOMAP, 2015).
5.2.2 Mitochondrial Genetic Variants and Disease Risk
Emerging evidence has shown that even mtDNA single nucleotide polymorphisms
(SNPs) can confer disease risk and influence development of human diseases [10]. A
previous case-control mitochondrial association analysis of 16,158 mtSNPs revealed a total
number of 36 mtSNPs to be significantly associated with asthma in children [11]. In
addition, a growing body of research has demonstrated that altered mitochondrial energy
production is a major anomaly capable of setting off a chain of metabolic events leading to

114

obesity [12, 13]. Studies have shown that particular mtDNA haplogroups and mtSNPs are
associated with inefficient energy expenditure, obesity and diabetes. For example, the D-
loop variant T16189C confers risk for type 2 diabetes in various population, but possibly
only in those with high BMI [14, 15]. Additionally, literature have suggested the variant
T16189C may create an uninterrupted poly-C tract located near a regulatory element
involved in replication of mtDNA [16]. Our previous mtDNA methylation study also found
T16189C modified the association between air pollution and mtDNA methylation as well
as the association between mtDNA methylation and birth weight of newborns. Flaquer A
et al., recently identified 2 mtSNPs located in the Cytochrome c oxidase subunit gene (MT-
CO1) and 3 mtSNPs located in the NADH dehydrogenase subunit genes (MT-ND1, MT-
ND2 and MT-ND4L) associated with higher BMI in adults [17].
Previous studies have suggested that increase in reactive oxygen species (ROS)
production in mitochondria may cause oxidative stress, which may in turn exacerbate the
inflammatory process in obesity [18-20]. Given the role of mitochondria in energy
metabolism and ROS production, compromised mitochondrial function could lead to
reduced mitochondrial biogenesis, reduced mitochondrial DNA content, decreased β-
oxidation rate, and altered adipocyte pathways, which may further contribute to changes in
insulin sensitivity[18, 21]. Furthermore, alteration in function of OXPHOS has been
associated with type 2 diabetes and obesity in in vitro studies [22, 23]. Kelley and
colleagues observed impaired bioenergetics capacity of skeletal muscle mitochondria in
diabetic patients compared to non-diabetic patients [23].

115

5.3 Proposed Study
The proposed study aims to examine the relationship between genetic variants in
mitochondrial genome and childhood obesity risk. The project aims to identify
mitochondrial single nuclear polymorphisms (mtSNPs) that may influence body mass
index (BMI) in children and adolescents. We also sought to replicate our findings in a
separate population of children from EVE Consortium on the Genetics of Asthma. These
findings of this research may highlight the importance of mtDNA and its contribution to
the risk of BMI growth in children and provide evidence that genetic variants in the
mitochondrial genome may play a role in mitochondrial functions and in turn influence
metabolism.
5.3.1 Study Population
This study was nested in Children’s Health Study (CHS). Twelve Southern
California communities were originally selected to represent a historically diverse
pollution profile of regional levels of nitrogen dioxide, PM10, ozone, and acid vapor. Three
successively recruited cohorts were used in the current study. In 1993, 1800 fourth graders,
aged 9 and 10 years, were recruited from schools across 12 communities and followed up
through high school graduation in 2001. In 1996, another cohort of approximately 2080
fourth graders from the same communities was recruited and followed up through high
school graduation in 2004. In 2003, a new cohort of 5600 either kindergarten or first
graders, aged 5 through 7 years, was recruited from 13 Southern California communities.
Eight communities (Alpine, Lake Elsinore, Long Beach, Mira Loma, Riverside, San
Dimas, Santa Maria, and Upland) had participants in all 3 cohorts (hereafter referred to as
the 1993-2001, 1996-2004, and 2003-2012 cohorts) with data collected with consistent

116

methods over the period of study. Two other 1993-1995 and 1993-1998 cohorts involved
in the California Children’s Health Study were not included in the current analysis because
they had relatively shorter follow-up.

Figure 1. Southern California Children’s Health Study (CHS) communities (obtained from
McConnell et al, 2015).
At baseline, a parent or guardian of each participating child provided written
informed consent and completed a written questionnaire that supplied detailed information
on family demographic characteristics, history of respiratory illness and associated risk
factors, indoor sources of exposure to pollutants, physical activity patterns of the children,
and household characteristics.

117

5.3.2 Measurement of BMI in Children
In spring of each study year, an update questionnaire was completed by each child,
and anthropometric measures such as height and weight were measured for every child
enrolled in the CHS by a trained technician at baseline and annually through the entire 8-
year follow-up. Technicians followed a standardized procedure that included details on
scale calibration and interaction with the children. These objective measures of height and
weight allow for precise and accurate calculation of BMI (kg of weight/height in m
2
). BMI
and questionnaire-based individual information were then linked to built-environment
measures around the children's homes and schools, along with physical features such as
topography and social environment variables in the census tract of residence (e.g., poverty).
Age- and sex-specific percentiles based on the Centers for Disease Control and Prevention
body mass index growth charts (http://
www.cdc.gov/NCCDPHP/dnpa/growthcharts/resources/sas.htm) were used to categorize
body mass index into underweight, normal, overweight, and obese categories.
5.3.3 Genotyping
Buccal samples were collected for genotyping. The CHS genotyped 5,617 DNA
samples on the Illumina HumanExome BeadChip, which contains >250K fixed-content
and >10K custom-content SNPs (>5K selected specifically for the CHS). The Illumina
HumanExome BeadChip genotyping array interrogates rare variants discovered from the
exome and genome sequences of ~12,000 individuals of European American, African
American, Latino and Asian Subjects from a range of common disorders. Study subjects
were expected to be unrelated and have either Hispanic or Non-Hispanic White ancestry.
Sample processing and genotyping was performed at the USC Cancer Center Molecular

118

Biology Core. In addition, a subset of the study’s subjects was previously genotyped as
part of a genome-wide association study (GWAS) on the Illumina HumanHap550,
HumanHap550-Duo, or Human610-Quad BeadChip platforms. In addition, a subset of
study subjects was genotyped as part of a candidate gene study using the Illumina
GoldenGate platform. Only single-nucleotide polymorphisms located in the mitochondrial
genome (mtSNPs) were considered in this study. The following quality control criteria
were applied: (1) remove samples with call rate < 90%, (2) remove SNPs with call rate <
90%, (3) iteratively remove samples with call rate <91%, SNPs <91%, samples<92%,
SNPs<92%, samples<93%, SNPs <93%, samples<94%, SNPs<94%, samples<95%,
SNPs<95%, (4) remove SNPs with replicate errors (concordance < 95%), (5) remove
replicates with lowest call rate. A total of 164 mtSNPs from GWAS chip and 163 mtSNPs
from Exome chip were included for analysis. The complete list of mtSNPs arrayed are
included in Appendix I.
5.3.4 Replication Study Population
A subset of EVE participants was evaluated for replication of our initial findings.
13 individual studies from eight investigators that comprise the EVE Consortium on the
Genetics of Asthma. Description of each of the studies can be found in prior published
work [24]. All participants, or parents of minors, provided written informed consent. Age,
weight, and height information at enrollment were obtained for individual studies. BMI of
the replications study subjects were then calculated. Three studies within EVE, a total of
556 children with BMI and genotyping information available, were included in the
replication study.

119

The replication samples from EVE were genotyped on the Illumina HumanExome
BeadChip. Variants with genotype call rates < 95%, Hardy-Weinberg equilibrium P-values
< 10
-4
in any of the studied populations and caution sites reported as problematic by the
exome array design group [25]. A total of 226 mtSNPs were genotyped in EVE replication
study. The mtSNP list of the replication study is shown in Appendix I.
5.3.5 Statistical Analysis
Allele frequency, commonly used index for genetic studies, denotes as the relative
frequency of a certain allele at a marker locus. Both individual-level allele frequency,
which represents the within-individual relative frequency of alleles, and population-level
allele frequency refers to the within-population relative frequency of alleles, can be
obtained from genotyping. The standard error of individual-level allele frequency
represents the inter-cell variability in an individual, whereas the standard error of
population-level allele frequency represents the inter-individual variability in a population
[26]. Both individual-level and population-level allele frequency can be estimated using an
allele counting approach (genotype-based) and using an intensity measuring approach
(intensity based).
Genotype calling algorithms may be controversial when applied to mtSNPs due to
the heteroplasmy effect. The mtDNA tend to be heterogenous in the sense that different
mitochondria of an individual may have different genotypes. This issue affects the
possibility of estimating genotypes and makes calling algorithms useless. Therefore, to
account for heteroplasmy, individual-level allele frequencies obtained from the intensity
values were used for the analysis. For every individual and SNP, two intensity
measurements (A1, B1), …, (An, Bn) with n = 1 where Ai and Bi represent the intensities

120

of the two alleles A and B were generated. The mtSNP intensity entered the model as the
response via a log 2 transformation log2 (A/B) [17, 27].
A multi-level growth curve model that employed a flexible linear-spline based
approach is used to characterize the nonlinear BMI trajectories during childhood. Similar
models have been used extensively for assessing lung function growth in our cohort and
are adapted for BMI growth [28, 29]. All BMI data are checked for outliers and internal
consistency in the growth curves. The BMI growth curves for boys and girls in CHS are
shown in Figure 2. The BMI trajectories are estimated using linear splines with breakpoints
(known as knots) at ages 12, 14, and 16, essentially fitting four straight lines for <12, (12–
14), (14–16) and >16 years of age for the 1993-2001 cohort and the 1996-2004 cohort. For
the 2003-2012 cohort, knots are placed at ages 8, 10, 12, 14, 16, essentially fitting six
straight lines for <8, (8-10), (10-12), (12-14), (14-16) and >16 years of age.
Figure 2. BMI growth curves for boys and girls in Children’s Health Study (obtained from
Jerrett et al., 2014).

121

Simple linear regression models were fitted for the association with BMI at baseline
and last year of follow-up. The linear regression model for mtSNPs and BMI association
were first evaluated in the whole cohort and then evaluated by race/ethnicity adjusted for
age, race/ethnicity (in the whole cohort), gender, communities, and platform of genotyping.
Effect modification by gender and race/ethnicity were also evaluated in the whole cohort
for the associations between mtSNPs and BMI at baseline, last year of follow-up and BMI
growth over the follow-up period.
In the replication study, simple linear regression models were also fitted to evaluate
the association between BMI at enrollment and significant loci identified in CHS
population adjusting for age at enrollment, gender, race/ethnicity, and study centers. We
were not able to assess growth of BMI in the replication study as EVE study only collected
weight and height information at enrollment. Replication study subjects were further
categorized into two subgroups based on their age enrollment to match the age distribution
of the primary population. A type I error through the use of the false discovery rate (FDR)
of 0.05 was considered significant. All statistical analyses were performed using SAS 9.4
(SAS Institute, Cary, NC) and R 3.5 (R Foundation for Statistical Computing).
5.4 Results
5.4.1 Population Characteristics
The study included 4,020 participants (617 from the 1993-2001, 871 from the 1996-
2004, and 2532 from the 2003-2012 cohorts) who had data from 2 or more annual follow-
up questionnaires and BMI measurements during the follow-up period. The study included
similar proportion of boys and girls for the 1993-2001 cohort and 1996-2004 cohort, where

122

the 2003-2012 cohort included 52.7% girls and 47.4% boys (Table 1). The proportion of
Hispanic White children increased from 8.9 % for the 1993-2001 cohort to 10.1% for the
1996-2004 cohort and to 57.7% for the 2003-2012 cohort. The 2003-2012 cohort had a
significantly lower proportion of exposure to secondhand smoke or history of in utero
exposure to maternal smoking. The mean (SD) age of children at study entry was 9.9 (0.5)
years for the 1993-2001 cohort, 9.5 (0.4) years for the 1996-2004 cohort, and 6.5 (0.6)
years for the 2003-2012 cohort. Average BMI at study entry and last year of follow-up
varied by gender and race/ethnicity across the cohorts (Table 1). The mean (SD) of BMI at
study entry decreased from 19.0 (3.3) to 16.2 (2.1) for Hispanic-White children. Similar
decreases were observed across the cohorts for Non-Hispanic White, boys and girls. Boys
showed a slightly greater increase for the 1993-2001 and 1996-2004 cohorts. We did not
observe a significant difference between the 1993-2001 and 1996-2004 cohorts in regards
to their characteristics; therefore, we combined participants from the 1993-2001 and 1996-
2004 cohorts for regression analysis.
73.6% of the EVE subjects included in the replication study were white and 61.9%
were boys. The mean (SD) age at enrollment was 9 (3). Average BMI at enrollment varied
by age and race/ethnicity across study centers. The flowchart of replication study
population is shown in Figure 3 and the characteristics of replication study participants are
shown in Table 2.

123

Table 1. Characteristics of study participants by cohort follow-up period

Cohort Follow-up Period, N (%)

1993-2001
(n=617)
1996-2004
(n=871)
2003-2012
(n=2532)
Age at baseline, mean (SD) 9.9 (0.5) 9.5 (0.4) 6.5 (0.6)
Gender

Male 287 (46.5) 427 (49.0) 1333 (52.7)
Female 330 (53.5) 444 (51.0) 1199 (47.4)
Race/ethnicity

Hispanic White 55 (8.9) 88 (10.1) 1461 (57.7)
Non-Hispanic White 431 (69.9) 559 (64.2) 997 (39.4)
Other/Don't know 131 (21.2) 224 (25.7) 74 (2.9)
Maternal education

Less tha high school 66 (11.0) 82 (9.8) 498 (20.7)
High school or higher 533 (89.0) 751 (90.2) 1906 (79.3)
Second hand smoke

Yes 81 (13.5) 118 (14.0) 155 (6.4)
No 519 (86.5) 728 (86.0) 2280 (93.6)
BMI at baseline, mean (SD)

Gender

Male
18.1 (2.9)
18.5 (3.8) 16.8 (2.8)
Female
18.3 (3.2)
18.0 (3.3) 16.6 (2.6)
Race/ethnicity

Hispanic White 19.0 (3.3)
18.2 (4.3) 16.2 (2.1)
Non-Hispanic White 17.9 (2.7)
18.0 (3.3) 17.1 (4.8)
BMI at last year of follow-up, mean
(SD)

Gender

Male
24.4 (4.8)
24.4 (4.8) 23.1 (5.0)
Female
23.1 (4.3)
23.5 (5.0) 23.3 (4.8)
Race/ethnicity

Hispanic White 24.8 (4.7)
24.1 (5.9) 23.9 (5.2)
Non-Hispanic White 23.2 (4.2)
23.5 (4.4) 22.3 (4.1)
Abbreviations: BMI, body mass index. Missing values were not included.

124

Table 2. Characteristics of replication study participants
N=556 N (%) or mean (SD)
Age at enrollment 9 (2.5)
5-10 years old 371 (66.7)
11-18 years old 185 (33.3)
Gender
Male 344 (61.9)
Female 212 (38.1)
Race/ethnicity
White 409 (73.6)
African American 54 (9.7)
Hispanic 93 (16.7)

Figure 3. Flowchart of replication study population

EVE study cohort with
phenotype information
available
(N=5441)
Children age from 5-
18 with BMI available
(N=1050)
mtSNPs genotyping
informaiton available
(N=556)
Excluded: age and
BMI information not
available, age <5 or
age >=18

125

5.4.2 Association of mtSNPs and BMI at Baseline and Last Year of Follow-up
To assess mtSNP association while taking heteroplasmy into account, we used the
raw signal intensity values that were measured on the microarray and applied the linear
regression method as described in the Statistical Methods section. A negative beta
estimate for the mtSNP (β<0) indicates that the risk allele is the minor allele while a β>0
indicates that the risk allele is the major allele. The resulting p-values after adjustment for
multiple testing for the 2003-2012 cohort are plotted in Figure 4A for BMI at baseline
model and Figure 4B for BMI at last year of follow-up. Subgroup analysis by
race/ethnicity are shown in Figure 5 for baseline BMI and Figure 6 for BMI at last year
of follow-up.
One mtSNP G15301A located in the mitochondrially encoded cytochrome b (MT-
CYB) was significant for both BMI at baseline and last year of follow-up for the 2003-
2012 cohort. Five additional mtSNPs were associated with BMI at last year of follow up
as shown in Figure 4B. The five mtSNPs are located in the NADH subunit
dehydrogenase genes of complex I (MT- ND4, MT-ND5, MT-ND6), Displacement loop
region (D-loop) and MT-CYB. The interactions by sex and race/ethnicity were significant
for all six significant mtSNPs. Subgroup analysis revealed that the significant
associations were only observed among Hispanic-White groups. All six significant
mtSNPs were present only on one array without overlapping. No significant mtSNPs
were observed when analyzing the combined 1993-2001 cohort and the 1996-2004 cohort
data.

126

Figure 4. A) Association of mtSNPs with BMI at baseline for the 2003-2012 cohort B)
Association of mtSNPs with BMI at last year of follow-up for the 2003-2012 cohort. White
circle represents log10(p-value) and blue circle indicate a significance p-vale of 0.05.
Models adjusted for age, race/ethnicity, gender, communities, and platform. Sample sizes
for association with BMI at baseline and last year of follow-up are 2012 and 1095,
respectively.
A)
B)

127

A)
B)

Figure 5. Association of mtSNPs with BMI at baseline for the 2003-2012 cohort A)
Hispanic White, B) Non-Hispanic White. White circle represents log10(p-value) and blue
circle indicate a significance p-vale of 0.05. Models adjusted for age, gender, communities,
and platform.

128

A)
B)

Figure 6. Association of mtSNPs with BMI at last year of follow-up for the 2003-2012
cohort A) Hispanic White, B) Non-Hispanic White. White circle represents log10 (p-value)
and blue circle indicate a significance p-vale of 0.05. Models adjusted for age, gender,
communities, and platform.

129

5.4.3 Association of mtSNPs and BMI growth
Results from linear spline models for BMI growth identified the mtSNP G15301A
was significant after adjusting for multiple testing for the cohort 2003-2012 (Figure 7A).
mtSNP G15301A is located in MT-CYB gene and was consistently associated with higher
BMI in males compared to females and in Hispanic-Whites compared to NHWs
throughout the follow-up period (Figure 7B). We did not observe such significant
association between G15301A and BMI at any time point for the 1993-2001 cohort and
the 1996-2004 cohort. For example, for mt15301 G→A, the beta coefficient estimate for
BMI growth was 0.39, which indicates that BMI increases when the ratio of alleles A and
G decreases, i.e., an increase in heteroplasmy with the G allele at this locus leads to an
increased BMI. The estimates of the model parameters for each significant mtSNP are
provided in Table 3.
5.4.4 Findings from replication study
In the replication study, one mtSNP T15672C located in MT-CYB gene identified
in CHS population at last year of follow-up was found to be significantly associated with
BMI among subjects aged 5-10 and marginally associated with BMI among subjects aged
11-18 in EVE consortium. The results from replication study in age group 5-10 are
provided in Table 3.

130

Figure 7. A) Association of mtSNPs with BMI growth for the 2003-2012 cohort B)
Association of mt15302 with BMI growth by gender C) Association of mt15302 with BMI
growth by race/ethnicity. White circle represent log10 (p-value) and blue circle indicate a
significance p-vale of 0.05. Models adjusted for age, race/ethnicity, gender, and
communities. Sample sizes for association with BMI at baseline and last year of follow-up
are 2012 and 1095, respectively.

A)
B) C)

131

132

5.5 Discussion

Our primary study, in which we examined association of mtSNPs and BMI in
children of different age group living in Southern California, identified one mtSNP
mt15301 G→A located in MT-CYB gene significantly associated with BMI growth, and
five additional mtSNPs significantly associated with BMI in adolescents. Findings from
our replication study also showed mtSNP mt15672 T→C located in MT-CYB gene can
influence childhood obesity risk.
The MT-CYB gene, located in complex III of mitochondria, provides instructions
for making a protein called cytochrome b, which plays a kay role in mitochondria structure
and OXPHOS process. Mutations in the MT-CYB gene can induce complex III deficiency
and reduction in oxidative phosphorylation. Previous studies have reported that a deletion
leading to an overexpression of MT-CYB gene is associated with increase ROS production
[30]. In addition, an in vivo study has discovered that production of mitochondrial complex
III ROS could result in induction of peroxisome proliferator-activated receptors (PPARγ)
transcriptional machinery which is known to be the master regulator of adipogenesis and
obesity and is required to initiate adipocyte differentiation [31-33]. In this study, we
identified for first time a significant association between genetic variants in MT-CYB
(mt15301 G→A) and BMI with an estimate effect size of 0.66 in adolescents, suggesting
that the G allele is the one to be related to higher BMI. Children with heteroplasmy having
a larger number of G alleles will have a higher BMI than those with a smaller number of
G alleles.

133

In addition, genetic variants in mitochondrially encoded NADH dehydrogenase
subunit genes have been identified in association with BMI in adult population [17].
Flaquer et al. previously reported two mtSNPs in MT-ND1 and MT-ND2 genes and one
mtSNP in MT-ND4L towards minor alleles increased the risk of developing high BMI in
adults [17]. In our study, we were not able to assess those association reported in Flaquer
et al. study with BMI in children population. However, we found three significant mtSNPs
located in NADH dehydrogenase subunit genes to be associated with BMI in children aged
10-18. Mitochondrially encoded NADH dehydrogenase is a part of a large enzyme
complex known as complex I, which is responsible for the first step in the electron transport
process, the transfer of electrons from NADH to ubiquinone [34]. The malfunction of
NADH dehydrogenase may result in a wide range of metabolic disorders with complex I
being one of several enzyme complexes necessary for OXPHOS. Defects in OXPHOS or
alternation in function of OXPHOS have been recognized consistently in type 2 diabetes
[22, 23]. In vitro studies also have shown reduced activity of complex I of mitochondria to
be associated with type 2 diabetes [23]. Previous association studies examining the
association between common mtSNPs and metabolic traits have found nominal
associations between mtSNPs of MT-ND and BMI in adult population but failed to reach
significance after correcting for multiple testing [35, 36]. However, both studies did not
address the role of mtDNA heteroplasmy and limited to Caucasian-dominant population.
Our positive findings together with previously inconsistent published work on
genetic variations in the mitochondrial genome and BMI suggest that there may be
population-specific or ethnic-specific variants that confer the altered risk of obesity and
other metabolic disorders [16, 17, 35, 36]. Our subgroup analysis further addressed this

134

hypothesis where we found significant mtSNPs associated with BMI in Hispanic-White
population only. In contrast, previous studies with null findings have mainly focused on
European population. Few studies have examined genetic variations in the mitochondrial
genome contribution to various disease risk in Hispanic or Latino population and reported
that Hispanics share certain genetic susceptibility loci with Whites and Asians to certain
diseases but may also have distinct genetic susceptibility factors [37, 38]. One of the studies
conducted within EVE consortium has reported that the effects of rare variants on asthma
risk is likely to be ethnic-specific [25]. To our knowledge, our study is one of the first to
examine the effects of mitochondrial genetic variants on BMI in Hispanic-dominant
population in children. Further studies, including replication in a large Hispanic children
population, will be needed to explore the population-specific or ethnic-specific hypothesis.
We acknowledge several limitations in the present study. Due to the analytic
method we used to take into account of mtDNA heteroplasmy, we were not able to
determine mtSNP alleles at individual-level and thus characterize haplogroups. It is
impossible to examine population stratification as a cause of false positive results.
Although we found nominal association of one of the significant mtSNPs in the replication
population, future studies should be warranted in a larger sample size.
In summary, our study identified genetic variants located in the cytochrome b and
in the NADH dehydrogenase subunits that were significantly associated with BMI in
Hispanic children. Our findings suggest that obesity or higher BMI in children may be in
part due to mitochondrial genetic variations. These findings highlight the important role of
mtDNA and its contribution to the higher BMI and risk of obesity in children and provide
evidence that genetic variants in the mitochondrial genome may play a role in

135

mitochondrial function and in turn influence metabolism. Further studies should be
warranted to investigate ethnic-specific effect of mtSNPs on obesity and BMI.

136

Appendix I: Position of mitochondrial genetic variants tested in primary population and
replication population
CHS
GWAS
CHS
Exome
EVE
Exome
182 217 150
217 228 217
228 285 228
247 295 285
250 489 295
295 606 418
458 709 489
464 710 499
479 951 606
491 1005 629
752 1018 709
829 1048 710
1020 1382 921
1050 1393 951
1191 1406 961
1440 1413 1005
1721 1438 1018
1738 1442 1048
2160 1664 1243
2354 1694 1382
2485 1700 1393
2708 1703 1406
2791 1706 1413
2887 1709 1438
3012 1717 1442
3198 1719 1664
3349 1811 1694
3395 1842 1700
3481 1888 1703
3548 2056 1706
3595 2259 1709
3667 2283 1717
3721 2332 1719
3916 2352 1811

137

3919 2358 1842
3971 2416 1888
3993 2706 2056
4025 2758 2218
4105 2768 2259
4337 2831 2283
4562 2885 2332
4640 3010 2352
4770 3027 2358
4821 3206 2416
4825 3210 2417
4884 3394 2706
4918 3398 2758
4978 3434 2768
5005 3547 2772
5047 3736 2831
5179 3746 2885
5264 3866 3010
5391 4093 3027
5443 4123 3116
5461 4129 3197
5496 4216 3200
5657 4312 3206
5774 4336 3210
5952 4363 3316
6027 4491 3338
6046 4561 3394
6072 4732 3398
6153 5331 3434
6222 5460 3497
6261 5814 3547
6366 6253 3571
6456 6261 3736
6681 6267 3746
6720 6480 3866
6735 6663 4021
6753 7028 4093
6777 7476 4123
7056 8414 4129

138

7176 8417 4216
7257 8460 4312
7275 8478 4336
7522 8654 4363
7769 8684 4491
8207 8701 4561
8270 8869 4732
8278 8875 4917
8617 8896 5331
8656 9098 5460
8870 9128 5814
9073 9477 6253
9094 9804 6261
9222 9903 6267
9378 9966 6480
9541 10007 6663
9668 10031 6982
9699 10034 7028
9717 10321 7476
9900 10345 8411
9951 10398 8414
10035 10410 8417
10045 10609 8460
10239 11025 8478
10311 11204 8654
10322 11253 8684
10399 11963 8701
10464 11969 8857
10551 11984 8860
10587 12236 8869
10590 12308 8875
10689 12397 8896
10874 12454 8923
10916 12820 9053
11252 12940 9098
11378 12950 9128
11468 13105 9142
11486 13135 9477
11723 13276 9682

139

11813 13477 9804
11900 13780 9855
11915 13880 9903
11970 13933 9966
12309 13942 10007
12373 13966 10031
12415 14053 10034
12631 14059 10084
12670 14148 10086
12706 14178 10143
12851 14258 10321
13106 14318 10345
13264 14502 10398
13651 14687 10410
13781 14927 10463
13790 15038 10609
13966 15071 11025
14179 15074 11172
14213 15077 11177
14234 15110 11204
14583 15119 11253
14767 15204 11696
14784 15236 11963
14799 15257 11969
15044 15263 11984
15111 15314 12236
15219 15323 12248
15245 15326 12308
15258 15381 12397
15302 15402 12454
15536 15431 12542
15671 15479 12820
15759 15662 12937
15785 15672 12940
15834 15693 12950
15905 15758 13105
15925 15849 13135
15929 15860 13276
15931 15889 13477

140

16130 15904 13708
16145 15905 13780
16146 15907 13789
16149 15924 13879
16163 15928 13880
16164 15939 13886
16184 15940 13924
16272 15941 13933
16280 15942 13934
16329 15946 13942
16392 15978 13958
16393 16017 13966

16390 13981

14000

14053

14059

14148

14178

14180

14258

14318

14502

14687

14766

14793

14798

14861

14927

14978

14979

15038

15071

15074

15077

15110

15119

15204

15218

15236

141

15257

15263

15311

15314

15323

15326

15381

15402

15431

15452

15479

15497

15662

15672

15693

15758

15812

15824

15849

15860

15889

15904

15905

15907

15924

15927

15928

15939

15940

15941

15942

15946

15951

15978

16017

16390

142

Chapter 5: References
1. Wallace DC, Zheng XX, Lott MT, Shoffner JM, Hodge JA, Kelley RI, Epstein CM,
Hopkins LC: Familial mitochondrial encephalomyopathy (MERRF): genetic,
pathophysiological, and biochemical characterization of a mitochondrial DNA
disease. Cell 1988, 55(4):601-610.
2. Wallace DC, Chalkia D: Mitochondrial DNA genetics and the heteroplasmy
conundrum in evolution and disease. Cold Spring Harb Perspect Biol 2013,
5(11):a021220.
3. Stewart JB, Chinnery PF: The dynamics of mitochondrial DNA heteroplasmy:
implications for human health and disease. Nat Rev Genet 2015, 16(9):530-542.
4. Wallace DC: A mitochondrial paradigm of metabolic and degenerative diseases,
aging, and cancer: a dawn for evolutionary medicine. Annu Rev Genet 2005,
39:359-407.
5. Wallace DC: Mitochondrial diseases in man and mouse. Science 1999,
283(5407):1482-1488.
6. Azar A, Devlin K, Mell JC, Giovannetti T, Pirrone V, Nonnemacher MR, Passic S,
Kercher K, Williams JW, Jacobson JM et al: Mitochondrial Haplogroup Influences
Motor Function in Long-Term HIV-1-Infected Individuals. PLoS One 2016,
11(10):e0163772.
7. Kofler B, Mueller EE, Eder W, Stanger O, Maier R, Weger M, Haas A, Winker R,
Schmut O, Paulweber B et al: Mitochondrial DNA haplogroup T is associated with
coronary artery disease and diabetic retinopathy: a case control study. BMC Med
Genet 2009, 10:35.
8. Ebner S, Mangge H, Langhof H, Halle M, Siegrist M, Aigner E, Paulmichl K,
Paulweber B, Datz C, Sperl W et al: Mitochondrial Haplogroup T Is Associated
with Obesity in Austrian Juveniles and Adults. PLoS One 2015, 10(8):e0135622.
9. Chalkia D, Singh LN, Leipzig J, Lvova M, Derbeneva O, Lakatos A, Hadley D,
Hakonarson H, Wallace DC: Association Between Mitochondrial DNA
Haplogroup Variation and Autism Spectrum Disorders. JAMA Psychiatry 2017,
74(11):1161-1168.
10. Picard M, Wallace DC, Burelle Y: The rise of mitochondria in medicine.
Mitochondrion 2016, 30:105-116.
11. Flaquer A, Heinzmann A, Rospleszcz S, Mailaparambil B, Dietrich H, Strauch K,
Grychtol R: Association study of mitochondrial genetic polymorphisms in
asthmatic children. Mitochondrion 2014, 14(1):49-53.
12. Wang Y, Mohsen AW, Mihalik SJ, Goetzman ES, Vockley J: Evidence for physical
association of mitochondrial fatty acid oxidation and oxidative phosphorylation
complexes. J Biol Chem 2010, 285(39):29834-29841.
13. Goldstein AC, Bhatia P, Vento JM: Mitochondrial disease in childhood: nuclear
encoded. Neurotherapeutics 2013, 10(2):212-226.
14. Poulton J, Luan J, Macaulay V, Hennings S, Mitchell J, Wareham NJ: Type 2
diabetes is associated with a common mitochondrial variant: evidence from a
population-based case-control study. Hum Mol Genet 2002, 11(13):1581-1583.

143

15. Liou CW, Chen JB, Tiao MM, Weng SW, Huang TL, Chuang JH, Chen SD,
Chuang YC, Lee WC, Lin TK et al: Mitochondrial DNA coding and control region
variants as genetic risk factors for type 2 diabetes. Diabetes 2012, 61(10):2642-
2651.
16. Knoll N, Jarick I, Volckmar AL, Klingenspor M, Illig T, Grallert H, Gieger C,
Wichmann HE, Peters A, Wiegand S et al: Mitochondrial DNA variants in obesity.
PLoS One 2014, 9(5):e94882.
17. Flaquer A, Baumbach C, Kriebel J, Meitinger T, Peters A, Waldenberger M,
Grallert H, Strauch K: Mitochondrial genetic variants identified to be associated
with BMI in adults. PLoS One 2014, 9(8):e105116.
18. de Mello AH, Costa AB, Engel JDG, Rezin GT: Mitochondrial dysfunction in
obesity. Life Sci 2018, 192:26-32.
19. Bhatraju NK, Agrawal A: Mitochondrial Dysfunction Linking Obesity and
Asthma. Ann Am Thorac Soc 2017, 14(Supplement_5):S368-S373.
20. Lahera V, de Las Heras N, Lopez-Farre A, Manucha W, Ferder L: Role of
Mitochondrial Dysfunction in Hypertension and Obesity. Curr Hypertens Rep
2017, 19(2):11.
21. Kusminski CM, Scherer PE: Mitochondrial dysfunction in white adipose tissue.
Trends Endocrinol Metab 2012, 23(9):435-443.
22. Ritov VB, Menshikova EV, He J, Ferrell RE, Goodpaster BH, Kelley DE:
Deficiency of subsarcolemmal mitochondria in obesity and type 2 diabetes.
Diabetes 2005, 54(1):8-14.
23. Kelley DE, He J, Menshikova EV, Ritov VB: Dysfunction of mitochondria in
human skeletal muscle in type 2 diabetes. Diabetes 2002, 51(10):2944-2950.
24. Torgerson DG, Ampleford EJ, Chiu GY, Gauderman WJ, Gignoux CR, Graves PE,
Himes BE, Levin AM, Mathias RA, Hancock DB et al: Meta-analysis of genome-
wide association studies of asthma in ethnically diverse North American
populations. Nat Genet 2011, 43(9):887-892.
25. Igartua C, Myers RA, Mathias RA, Pino-Yanes M, Eng C, Graves PE, Levin AM,
Del-Rio-Navarro BE, Jackson DJ, Livne OE et al: Ethnic-specific associations of
rare and low-frequency DNA sequence variants with asthma. Nat Commun 2015,
6:5965.
26. Yang HC, Huang MC, Li LH, Lin CH, Yu AL, Diccianni MB, Wu JY, Chen YT,
Fann CS: MPDA: microarray pooled DNA analyzer. BMC Bioinformatics 2008,
9:196.
27. Macgregor S, Visscher PM, Montgomery G: Analysis of pooled DNA samples on
high density arrays without prior knowledge of differential hybridization rates.
Nucleic Acids Res 2006, 34(7):e55.
28. Jerrett M, McConnell R, Chang CC, Wolch J, Reynolds K, Lurmann F, Gilliland
F, Berhane K: Automobile traffic around the home and attained body mass index:
a longitudinal cohort study of children aged 10-18 years. Prev Med 2010, 50 Suppl
1:S50-58.
29. Gauderman WJ, Vora H, McConnell R, Berhane K, Gilliland F, Thomas D,
Lurmann F, Avol E, Kunzli N, Jerrett M et al: Effect of exposure to traffic on lung
development from 10 to 18 years of age: a cohort study. Lancet 2007,
369(9561):571-577.

144

30. Wang HC, Choudhary S: Reactive oxygen species-mediated therapeutic control of
bladder cancer. Nat Rev Urol 2011, 8(11):608-616.
31. Tormos KV, Anso E, Hamanaka RB, Eisenbart J, Joseph J, Kalyanaraman B,
Chandel NS: Mitochondrial complex III ROS regulate adipocyte differentiation.
Cell Metab 2011, 14(4):537-544.
32. Shao X, Wang M, Wei X, Deng S, Fu N, Peng Q, Jiang Y, Ye L, Xie J, Lin Y:
Peroxisome Proliferator-Activated Receptor-gamma: Master Regulator of
Adipogenesis and Obesity. Curr Stem Cell Res Ther 2016, 11(3):282-289.
33. Gregoire FM, Smas CM, Sul HS: Understanding adipocyte differentiation. Physiol
Rev 1998, 78(3):783-809.
34. Weiss H, Friedrich T, Hofhaus G, Preis D: The respiratory-chain NADH
dehydrogenase (complex I) of mitochondria. Eur J Biochem 1991, 197(3):563-576.
35. Liu C, Yang Q, Hwang SJ, Sun F, Johnson AD, Shirihai OS, Vasan RS, Levy D,
Schwartz F: Association of genetic variation in the mitochondrial genome with
blood pressure and metabolic traits. Hypertension 2012, 60(4):949-956.
36. Saxena R, de Bakker PI, Singer K, Mootha V, Burtt N, Hirschhorn JN, Gaudet D,
Isomaa B, Daly MJ, Groop L et al: Comprehensive association testing of common
mitochondrial DNA variation in metabolic disease. Am J Hum Genet 2006,
79(1):54-61.
37. Sun Y, Gu J, Ajani JA, Chang DW, Wu X, Stroehlein JR: Genetic and intermediate
phenotypic susceptibility markers of gastric cancer in Hispanic Americans: a case-
control study. Cancer 2014, 120(19):3040-3048.
38. Restrepo NA, Mitchell SL, Goodloe RJ, Murdock DG, Haines JL, Crawford DC:
Mitochondrial variation and the risk of age-related macular degeneration across
diverse populations. Pac Symp Biocomput 2015:243-254.

145

CHAPTER 6
Summary and Future Research
6.1 Summary and Conclusion

The goal of this dissertation was to explore the epigenetic regulation and genetic
regulation of mitochondrial DNA in relation to prenatal exposure to air pollution and the
growth of infants and children. Exposure to chemical and environmental hazards and
stressors during pregnancy, infancy, and early childhood have been associated with an
increased risk of obesity in childhood [1-3]. However, the mechanisms thereof remain
largely unknown. Therefore, an understanding of the mechanisms by which the early-life
environment contributes to early childhood growth and later childhood obesity is crucial
and necessary in order to develop prevention strategies for childhood obesity.
Mitochondria have been identified as one of the key players in the regulation of early
development and metabolism, as mitochondria are the primary energy producers of
adenosine-5’-triphosphate (ATP) via oxidative phosphorylation (OXPHOS).
Mitochondrial DNA is sensitive to damage by exogenous reactive oxygen species. In
particular, environmental pollutants such as traffic-related air pollution (TRAP) and
ambient air pollution (AAP), known to generate oxidative stress, have been associated with
various forms of mitochondrial damage [4-6]. Furthermore, dysfunctional mitochondria
have been implicated in various metabolic disorders, including obesity and type 2 diabetes
[7-9].

146

Given the important role of mitochondria in human disease, the increasing
understanding of mito-regulatory mechanisms, and the suggested associations between air
pollutants and mitochondrial damage, we aim for this dissertation 1) to identify whether
patterns of differentially methylated mitochondrial DNA CpG sites, which occur in
response to air pollution exposure, are associated with newborn outcomes; 2) to determine
whether expression patterns of miRNAs, which are mitochondrial-associated in response
to air pollution exposure, are associated with newborn outcomes; and 3) to identify genetic
variations in the mitochondrial genome as potential risk factors for childhood obesity.
In Chapter 3, we evaluated the association of prenatal exposure to TRAP and APP
and targeted mtDNA methylation in mitochondrially encoded transfer RNA phenylalanine
(MT-TF), 12S rRNA (MT-RNR1), and the displacement loop (D-loop) control region in
peripheral blood mononuclear cells (PBMCs), CD4+ cells, and CD14+ cells. The
associations between mtDNA methylation in different cell types and birthweight were also
examined. We demonstrated that prenatal TRAP and AAP exposures were associated with
mtDNA methylation and copy number changes in different cord blood cell types.
Furthermore, the strength of these associations differed by mtSNP T16189C genotype in
some cases. The MtSNP T16189C genotype also modified the association between mtDNA
methylation level and infant birth weight. However, we did not observe an effect of
prenatal exposure to air pollution on mtDNA methylation and the copy number stratified
by mitochondrial haplogroups. Our findings are consistent with previous work that has
examined the effect of TRAP and AAP on mitochondrial DNA methylation; however, we
observed patterns of differential methylation in different cell population, suggesting that
tissue-specific and cell-type-specific mitochondrial epigenetic patterns may occur

147

depending on the number of mitochondria and rates of cellular metabolism [10].
Additionally, our study is one of the first to report that mitochondrial genetic variations
alters individual susceptibility to prenatal air pollution exposure and mtDNA methylation,
and alters the association between mtDNA methylation and birth weight.
In Chapter 4, we investigated associations among in utero air pollution exposure,
patterns of maternal-circulating mitochondrial-associated miRNAs profiled using the
nCounter miRNA Expression Assay Human v2 System, and newborn outcomes in 137
pregnant women currently enrolled in the ongoing Maternal and Developmental Risks from
Environmental and Social Stressors (MADRES) birth cohort. Distributed lag model
(DLMs) approaches allowed us to further examine the dynamic effect of air pollution on
maternal miRNAs at first and third trimesters. After correcting for multiple testing, we
identified four miRNAs that were thought to be localized in mitochondria that were
associated with short-term exposure to NO2 and PM2.5 during the trimester. Weekly DLMs
revealed the possibility of sensitive time windows of exposure that are specific to changes
of miRNA expressions. We did not observe a significant connection between
mitochondrial-associated miRNAs and birthweight. Our study is one of the first to
investigate the relationship among air pollution, mitochondrial-associated miRNAs in
maternal blood during pregnancy, and newborn outcomes. A number of studies have
demonstrated the possible localization of miRNAs in mitochondria; however, the extent to
which miRNA localization in the mitochondria is regulated remains largely unknown. Our
positive findings of mitomiRs miR-23b, miR-29a, miR-199a, and miR-122 with air
pollution add to a new body of literature and comprise a step toward understanding the

148

possibility that environmental exposures may influence mitochondrial-related miRNAs by
providing evidence of associations from an epidemiologic study.
Finally, in Chapter 5, we evaluated the longitudinal associations of genetic
variations in the mitochondrial genome and childhood obesity in a large cohort of children.
This project is nested in the Children’s Health Study (CHS), which has recruited and
followed up children for over 20 years in 13 southern-California communities. We
identified one mtSNP mt15301 G→A, located in the MT-CYB gene, that was significantly
associated with BMI growth and five additional mtSNPs associated with BMI in children
at their last year of follow-up. We also performed a replication study in a separate
population of children from the EVE Consortium on the Genetics of Asthma. Our positive
findings, together with previously inconsistent published work on genetic variations in the
mitochondrial genome and BMI, suggest that there may be population-specific or ethnicity-
specific variants that confer the altered risk of obesity and other metabolic disorders [11-
14]. The findings of this research highlight the importance of mtDNA and its contribution
to the risk of BMI growth in children and provide evidence that genetic variants in the
mitochondrial genome may play a role in mitochondrial function and, in turn, influence
metabolism. Further studies should investigate the ethnicity-specific effect of mtSNPs on
obesity and BMI.
Taken together, all three projects undertaken in this dissertation have contributed
to the literature linking environmental stressors, mitochondrial function, and growth in
children. We observed both long- and short-term effects of prenatal exposure to air
pollution on mitochondrial epigenetic changes and of the genetic regulation of
mitochondrial DNA on changes in BMI in children.

149

6.2 Implications and Future Directions

Collectively the three projects undertaken in this dissertation addressed several
gaps in the current literature on mitochondrial function in the context of epidemiological
observation studies, but it has also brought to light the need for continued research in this
area. Mitochondrial epigenetics and genetics is an emerging research field that shows great
potential on developing biomarkers of effects of environmental exposures on disease risk.
Additional studies investigating the biological function of mitochondria as the potential
underlying mechanism linking environmental exposures and fetal development and early
childhood health are needed. In particular, there is a need to determine how prenatal and
early life exposure to environmental toxins affect the regulation of genes through
epigenetic modifications in both nucleus and mitochondria, which could help better
understand the cross-walk between mitochondria and nucleus.
First, as we hypothesized in Chapters 4 and 5, some of our positive findings are
likely to be ethnicity-specific. More replications studies are needed to generalize our
findings. In particular, we would like to seek replications of our genetic variation study in
a larger cohort of children consisting of Hispanic-White participants, as only 17% our
original replication study consisted of Hispanic participants. Also, some of our preliminary
analyses of MADRES birth cohort have found that acculturation may contribute to mother
and children’s health outcomes. Evaluating the interaction of ethnicity and acculturation
together should be a new investigation direction for future studies.
In addition to replication studies, future studies could re-evaluate the efficiency and
adequacy of bisulfite pyrosequencing in the detection of mitochondrial DNA methylation
in the entire mitochondrial genome. In Chapter 3, only three regions of mitochondrial DNA

150

were selected and assayed. Our findings may not extrapolate to the entire mitochondrial
DNA genome. Another challenge that may affect mitochondrial DNA methylation
detection is the presence of nuclear-mitochondrial sequences (Numts) or of pseudogenes
in the nuclear genome that do not transcribe. The current method of bisulfite treatment of
the total genomic DNA for methylation analysis may not be able to distinguish between
mtDNA sequences and Numts, despite the fact that we designed primers specific to the
mtDNA sequence and checked them against the nuclear genome for potential overlap,
which greatly reduces this likelihood. Recently, studies have recommended the use of
bisulfite pyrosequencing primers that are highly selective to the converted DNA template
from linearized mitochondrial DNA [15].
Furthermore, the current sample size of our miRNA study (Chapter 4) is relatively
small. Although we were able to observe significant associations despite our small sample
size, the effects of short-term air pollution on mitochondrial-associated miRNAs in a larger
pregnancy cohort should be examined. Although we collected maternal blood samples
from both the first- and third-trimester visits, the number of overlapping samples between
both time points was much smaller, and we were thus unable to perform mixed-effects
models to evaluate the longitudinal association of air pollution and maternal circulating
miRNAs or the longitudinal association of maternal circulating miRNAs and birthweight.
In addition, due to the small sample size, we were not able to evaluate effect modifications
by potential modifiers. For example, maternal stress has been shown to be associated with
miRNA expression in maternal blood during pregnancy in our study population. Ideally,
with a larger sample size, we would like to evaluate the role of maternal stress and
incorporate such factors in the DLMs.

151

We acknowledge that this dissertation work on mitochondrial function is built upon
observed associations of epidemiological studies. In the current dissertation work, we did
not measure the target gene expressions regulated by mitochondrial DNA methylation and
mitochondrial-associated miRNAs. Therefore, we cannot validate whether the epigenetic
alterations affect the biological function of mitochondria. In addition, given the nature of
epidemiological studies, we cannot confirm which biological pathways of mitochondria
are involved in the association of air pollution and the growth of infants and children. In
an experimental study setting, we would like to further examine the interrelationship
among air pollution, mitochondrial epigenetics and genetics, mitochondrial function
assessed via oxidative phosphorylation methods (e.g., Seahorse), and newborn/children
health outcomes. The underlying biological mechanism by which mitochondria epigenetics
and genetics modify the relationship between air pollution and children’s health is beyond
the scope of this dissertation work and warrants future study.
To advance the field of environmental epigenetics and deepen our understanding
of mitochondrial function, we need to conduct future studies generating omics data, which
requires consideration of integration between exposomics, genetics, and epigenetics.
Research should focus on fetal programming and early-life reprogramming, windows of
susceptibility, and continuous alteration of epigenetics throughout life. In particular,
genetic variations associated with prenatal and early-life exposures, epigenetic markers, or
early-life phenotypes can be used to evaluate whether those early-life predictors can also
predict health outcomes throughout life. For example, genetic variants that are associated
with DNA methylation can be identified using methylation quantitative trait locus
(metQTL) mapping and potentially be the target sites of gene-environmental interactions.

152

Chapter 6: References
1. Toledo-Corral C AT, Habre R, Berhane K, Lurmann F, Weigensberg M, Goran M,
Gilliland F: Effects of air pollution exposure on glucose metabolism in Los Angeles
minority children. Pediatric Obesity 2016.
2. Oken E, Levitan EB, Gillman MW: Maternal smoking during pregnancy and child
overweight: systematic review and meta-analysis. Int J Obes (Lond) 2008,
32(2):201-210.
3. Reilly JJ, Armstrong J, Dorosty AR, Emmett PM, Ness A, Rogers I, Steer C,
Sherriff A: Early life risk factors for obesity in childhood: cohort study. BMJ 2005,
330(7504):1357.
4. Delgado-Buenrostro NL, Freyre-Fonseca V, Cuellar CM, Sanchez-Perez Y,
Gutierrez-Cirlos EB, Cabellos-Avelar T, Orozco-Ibarra M, Pedraza-Chaverri J,
Chirino YI: Decrease in respiratory function and electron transport chain induced
by airborne particulate matter (PM10) exposure in lung mitochondria. Toxicol
Pathol 2013, 41(4):628-638.
5. Gualtieri M, Ovrevik J, Mollerup S, Asare N, Longhin E, Dahlman HJ, Camatini
M, Holme JA: Airborne urban particles (Milan winter-PM2.5) cause mitotic arrest
and cell death: Effects on DNA, mitochondria, AhR binding and spindle
organization. Mutat Res 2011, 713(1-2):18-31.
6. Hou L, Zhu ZZ, Zhang X, Nordio F, Bonzini M, Schwartz J, Hoxha M, Dioni L,
Marinelli B, Pegoraro V et al: Airborne particulate matter and mitochondrial
damage: a cross-sectional study. Environ Health 2010, 9:48.
7. Rolo AP, Gomes AP, Palmeira CM: Regulation of mitochondrial biogenesis in
metabolic syndrome. Curr Drug Targets 2011, 12(6):872-878.
8. Rolo AP, Palmeira CM: Diabetes and mitochondrial function: role of
hyperglycemia and oxidative stress. Toxicol Appl Pharmacol 2006, 212(2):167-
178.
9. Nunnari J, Suomalainen A: Mitochondria: in sickness and in health. Cell 2012,
148(6):1145-1159.
10. Byun HM, Baccarelli AA: Environmental exposure and mitochondrial epigenetics:
study design and analytical challenges. Hum Genet 2014, 133(3):247-257.
11. Knoll N, Jarick I, Volckmar AL, Klingenspor M, Illig T, Grallert H, Gieger C,
Wichmann HE, Peters A, Wiegand S et al: Mitochondrial DNA variants in obesity.
PLoS One 2014, 9(5):e94882.
12. Flaquer A, Baumbach C, Kriebel J, Meitinger T, Peters A, Waldenberger M,
Grallert H, Strauch K: Mitochondrial genetic variants identified to be associated
with BMI in adults. PLoS One 2014, 9(8):e105116.
13. Liu C, Yang Q, Hwang SJ, Sun F, Johnson AD, Shirihai OS, Vasan RS, Levy D,
Schwartz F: Association of genetic variation in the mitochondrial genome with
blood pressure and metabolic traits. Hypertension 2012, 60(4):949-956.
14. Saxena R, de Bakker PI, Singer K, Mootha V, Burtt N, Hirschhorn JN, Gaudet D,
Isomaa B, Daly MJ, Groop L et al: Comprehensive association testing of common
mitochondrial DNA variation in metabolic disease. Am J Hum Genet 2006,
79(1):54-61.

153

15. Owa C, Poulin M, Yan L, Shioda T: Technical adequacy of bisulfite sequencing
and pyrosequencing for detection of mitochondrial DNA methylation: Sources and
avoidance of false-positive detection. PLoS One 2018, 13(2):e0192722.

Abstract (if available)

Abstract In the United States, approximately 33% of children are overweight or obese. Due to its high prevalence and serious health consequences, such as cardiovascular disease, type 2 diabetes and many cancers, obesity is considered to be one of the most challenging public health issues of 21st century. Prevention strategies for childhood obesity to date have been focused on changes in lifestyle, specifically diet and physical activity. A growing body of literature has suggested that early life factors play a crucial role in later risk of obesity in childhood. Exposure to chemical and environmental hazards and stressors during pregnancy, infancy, and early childhood has been associated with increased risk of obesity in childhood. Yet, the mechanisms remain largely unknown. Therefore, understanding of the mechanisms by which early life environment contribute to early childhood growth and later childhood obesity is crucial and necessary for developing prevention strategies for childhood obesity. ❧ Mitochondria have been identified as one of the key players in the regulation of early development and metabolism, as mitochondria are the primary energy producers of adenosine-5’-triphosphate (ATP) via oxidative phosphorylation (OXPHOS). Dysfunctional mitochondria have been implicated in various metabolic disorders including obesity and type 2 diabetes. Recent studies have shown that particulate matter (PM) can penetrate cells and damage the mitochondria, including disruption of structure and function and altered mtDNA copy number. However, because of their non-Mendelian inheritance, many mitochondrial diseases cannot be fully understood solely by genetic studies. Epigenetic regulation of mitochondrial DNA adds another layer of regulation that may be susceptible to environmental exposure and disease risk. ❧ Emerging evidence has shown that mitochondria undergo epigenetic regulation through mechanisms that are similar to nuclear DNA. The discovery of mitochondrial DNA methyltransferase 1 (mtDNMT1) has confirmed the link between oxidative stress and mitochondrial functions and provided new evidence that epigenetic regulation of the mitochondrial genome by nuclear-encoded translocated mtDNMT1 relative to mitochondrial dysfunction. Moreover, ambient air pollutants have been reported to be associated with placental mtDNA methylation in newborns as well as blood mtDNA methylation in the elderly. Despite this recent interest in mtDNA methylation, the effects of mtDNA methylation on early-life health outcomes in response to environmental exposures have not been fully explored. ❧ Non-coding RNA, and particularly, microRNA (miRNA), represent another epigenetic mechanism that is involved in regulation of messenger RNAs of protein-coding genes and control of gene expression at a post-transcriptional level. Studies have shown that some miRNAs, derived from the nuclear genome, can translocate from the nucleus to mitochondria and are likely to contribute to regulation of gene expression and various metabolic pathways such as tricarboxylic acid cycle, electron transport chain, and lipid metabolism. In addition, the discovery of mitochondrial miRNAs (mitomiRs), unique sets of miRNAs that can localize in mitochondria, has raised the possibility of mitochondrial RNA synthesis. Currently, little is known about the effect of air pollution on expression of miRNAs targeting mitochondrial genes and its potential impact on development of metabolic diseases in childhood. ❧ Mitochondria have been largely studied by investigating their biochemistry and morphology, which have been implemented by genetic studies. Mitochondrial genetics studies have been focused on three main areas including the unique pattern of mtDNA inheritance, mtDNA mutations yielding heteroplasmy, and high mutation rate due to high oxidative stress. Genetic variation in mtDNA has been associated with congenital anomalies at birth and metabolic diseases in both children and adults. Genetic variants in the mtDNA control region have been associated with metabolic phenotypes in various populations. In addition, disease susceptibility may be captured by haplogroups, which represent the geographic origin of populations and may contribute to the differences in disease prevalence among racial/ethnic groups. Emerging evidence also suggests that genetic variation in mtDNA may modify the relationship between air pollution exposure and various health endpoints in humans. ❧ For my dissertation project I investigated the association between air pollution exposure, mitochondrial function, and growth in infants and children. The mitochondrial function studied include mitochondrial DNA methylation, patterns of miRNA expression targeting mitochondrial genes, and genetic variation in mitochondrial genome. ❧ The titles of my three chapters are: 1) association between prenatal exposure to air pollution, epigenetic variations in mitochondrial DNA, and infant growth

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Prenatal air pollution exposure, newborn DNA methylation, and childhood respiratory health

PDF

Air pollution, smoking, and multigenerational DNA methylation Signatures: a study of two southern California cohorts

PDF

Air pollution and childhood obesity

PDF

Assessing the impact of air pollution on adverse birth outcomes in a low resource setting

PDF

The relationship between per- and polyfluoroalkyl substances, microRNA, and non-alcoholic fatty liver disease

PDF

Personal exposure to particulate matter PM2.5 sources during pregnancy and birthweight

PDF

Prenatal environmental exposures and fetal growth in the MADRES cohort

PDF

Ambient air pollution and lung function in children

PDF

Exploring serum and tear micro-RNA as biomarkers for early diagnosis of Sjögren’s Syndrome

PDF

Native American ancestry among Hispanic Whites is associated with higher risk of childhood obesity: a longitudinal analysis of Children’s Health Study data

PDF

Spatial analysis of PM₂.₅ air pollution in association with hospital admissions in California

PDF

Sensor-based mobile health approaches for personal air pollution and pediatric asthma studies

PDF

Identification of novel epigenetic biomarkers and microRNAs for cancer therapeutics

PDF

Assessment of the mortality burden associated with ambient air pollution in rural and urban areas of India

PDF

Genetic epidemiological approaches in the study of risk factors for hematologic malignancies

PDF

Genetic and environmental risk factors for childhood cancer

PDF

Airway inflammation and respiratory health in the Southern California children's health study

PDF

Spatial modeling of non-tailpipe emissions and its association with children's lung function

PDF

DNA methylation of NOS genes and carotid intima-media thickness in children

PDF

Detecting joint interactions between sets of variables in the context of studies with a dichotomous phenotype, with applications to asthma susceptibility involving epigenetics and epistasis

Asset Metadata

Creator Song, Ashley Yi (author)

Core Title Air pollution, mitochondrial function, and growth in children

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Epidemiology

Publication Date 12/05/2019

Defense Date 10/22/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Air pollution,birth outcome,childhood obesity,DNA methylation,genetic variation,microRNAs,mitochondrial DNA,OAI-PMH Harvest,prenatal exposure to air pollution

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Breton, Carrie (committee chair), Dubeau, Louis (committee member), Habre, Rima (committee member)

Creator Email ashleyyisong@gmail.com,song059@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-244486

Unique identifier UC11673165

Identifier etd-SongAshley-8001.pdf (filename),usctheses-c89-244486 (legacy record id)

Legacy Identifier etd-SongAshley-8001.pdf

Dmrecord 244486

Document Type Dissertation

Rights Song, Ashley Yi

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA