EFFECTS OF MATERNAL PHYSICAL ACTIVITY ON METHYLATION PATTERNS IN
OFFSPRING BLOOD SPOTS
By
Mallory Marshall

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Kinesiology – Doctor of Philosophy
2014

ABSTRACT
EFFECTS OF MATERNAL PHYSICAL ACTIVITY ON METHYLATION
PATTERNS IN OFFSPRING BLOOD SPOTS
By
Mallory Marshall
A number of studies have suggested that lifestyle factors during
pregnancy, including diet and cigarette smoking, may influence the long term
health of offspring. Whether maternal leisure-time physical activity (LTPA) during
pregnancy might have a similar effect is unknown. One mechanism by which
LTPA might influence offspring health is DNA methylation, where methyl groups
are added to cytosine bases in DNA particularly at areas in the genome rich in
cytosine and guanine bases (CpG islands). The purpose of this dissertation was
to determine the relationship between maternal LTPA during pregnancy and
offspring DNA methylation, both globally and in metabolism-related candidate
genes, in infant blood spots.
We used the Archive for Research on Child Health (ARCH) study data set,
which has been ongoing at Michigan State University (MSU) since 2007. At
enrollment, subjects completed questionnaires on demographic information as
well as LTPA. We matched highly active subjects (averaged 637.5 minutes per
week of moderate or vigorous physical activity (MVPA); n = 14) to low active
subjects (averaged 60.4 minutes per week MVPA; n = 28) based on maternal
age and race. Subjects were contacted via telephone to ask for consent to link
physical activity data to their child’s blood spot which was obtained at birth.

Following DNA isolation and bisulfite treatment, we used pyrosequencing to
determine methylation levels of long interspersed nucleotide elements (LINE-1)
(global methylation) and PPARγ, PGC1-α, IGF2, PDK4, and TCF7L2.
We found no differences between high active and low active groups for
LINE-1 methylation, an indicator of global methylation (Aim 1). For Aim 2, the
only differences in candidate gene methylation between the two groups was at
two CpG sites in the P2 promoter of IGF2; in both instances the low active group
had significantly higher DNA methylation than the high active group (p = 0.020
and 0.047). However, the meaning of these differences is not clear. Also, mean
methylation over eight P2 promoter sites for IGF2 was significantly higher in the
low active group (p = 0.045), but there were no other differences in mean
methylation levels for any candidate gene.
Overall these results suggest no effect of maternal LTPA on global and
candidate gene methylation profiles, with the exception of IGF2 methylation. This
research is a unique pilot study of maternal LTPA in pregnancy and its effects on
DNA methylation, in humans. Possible reasons for the mostly null findings
include our choice of biologic sample (blood spots), lack of transfer of
methylation profile from mother to offspring, measurement of methylation rather
than gene expression, and inadequate LTPA stimulus. Future studies should
include examination of the role of higher levels of maternal LTPA on both
maternal and infant umbilical cord blood samples, and determination of the role
of maternal LTPA on both DNA methylation and gene expression.

Copyright by
MALLORY MARSHALL
2014

ACKNOWLEDGEMENTS

This dissertation would not have been possible without the guidance and
encouragement of many. Most of all I would like to thank my PhD advisor Dr. Jim
Pivarnik for his many hours of assistance with this project. I am extremely
grateful for his dedication to both students and to research, and am fortunate to
have had the opportunity to train with him at MSU. I would also like to thank my
other committee members. Dr. John Gerlach permitted me to use his laboratory
space for analyses and provided much feedback regarding the laboratory work
portion of this dissertation. Dr. Lanay Mudd provided excellent feedback on study
design and statistical analyses. Dr. Nigel Paneth oversees the ARCH study and
has permitted me to use this data. The work would not have been possible
without each of them.
Many others have assisted in making this work possible as well. Thank
you to Lynette Biery, ARCH study coordinator, for her assistance working with
this data. Carrie Langbo, BioTrust Coordinator for the State of Michigan, worked
tirelessly with me to obtain consent for use of the infant blood spots. Nancy
Christ at the Michigan Neonatal Biobank was also heavily involved in this
process.
I am also grateful to my fellow PhD students at MSU who have made my
four years here fun, memorable, and a great learning experience. I was fortunate
to start the program with an incredible group of people: these include Chris
Connolly, Alex Montoye, Larissa True, and Kimbo Yee. I am thankful to have

v

found a wonderful church family here in East Lansing and am indebted to many
friends at University Reformed Church for the support and prayers provided over
the last several years. To my daughter Addy, thank you for providing much
distraction and many opportunities for breaks from studying and writing—these
were needed! And finally, many thanks to my husband Dave who moved to
Michigan, to a poor job market and cold winters, so I could pursue this degree.
He is a great blessing to me and to many others as well.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................. ix
LIST OF FIGURES ................................................................................................ x
KEY TO ABBREVIATIONS .................................................................................. xi
INTRODUCTION ...................................................................................................1
Research Aims ...........................................................................................5
Specific Aim 1 .............................................................................................5
Specific Aim 2 .............................................................................................5
Organization of Dissertation .......................................................................6
REVIEW OF LITERATURE ...................................................................................7
Physical Activity and Pregnancy .................................................................7
Benefits of Physical Activity During Pregnancy – Fetal ....................8
Benefits of Physical Activity During Pregnancy – Maternal ............10
Physical Activity Quantification and Measurement ........................14
Physical Activity Recommendations for Pregnancy .......................18
Fetal Origins .............................................................................................20
Maternal Weight and Weight Gain .................................................20
Diet ................................................................................................26
DNA Methylation.......................................................................................30
Environmental Factors Influence DNA Methylation ........................31
DNA Methylation is Heritable .........................................................34
Factors that Influence DNA Methylation.........................................36
Diet......................................................................................36
Sex ......................................................................................37
Race....................................................................................38
Maternal Age .......................................................................38
Body Mass Index (BMI) .......................................................39
Physical Activity ..................................................................40
Metabolism-related Genes Subject to DNA Methylation ................41
Proliferator-activated receptor-γ co-activator α (PGC1-α) and
Peroxisome proliferator-activated receptor-γ (PPAR-γ) ......42
Insulin-like growth factor 2 (IGF2) .......................................42
Pyruvate dehydrogenase kinase 4 (PDK4) .........................43
Transcription factor 7-like 2 (TCF7L2).................................44
Summary ..................................................................................................44
METHODS ..........................................................................................................46
Study Design ............................................................................................46
Study Population ......................................................................................46
vii

Subject Matching ......................................................................................48
Sample Size .............................................................................................51
Laboratory Analysis ..................................................................................52
Aim 1..............................................................................................52
Aim 2..............................................................................................53
Statistical Analysis ....................................................................................54
RESULTS............................................................................................................56
Sample .....................................................................................................56
Aim 1 ........................................................................................................58
Aim 2 ........................................................................................................59
Correlations with Physical Activity .................................................60
DISCUSSION ......................................................................................................64
Future Directions ......................................................................................74
APPENDICES .....................................................................................................77
Appendix A: Questionnaires .....................................................................78
Appendix B: Oral Consent Script ..............................................................81
Appendix C: PCR and Pyrosequencing Conditions ..................................84
REFERENCES ....................................................................................................91

viii

LIST OF TABLES

Table 1: Candidate Gene Descriptions................................................................55
Table 2: Mean MVPA and BMI by the Three Originally Planned Groups ............57
Table 3: Descriptive Characteristics for Two Groups ..........................................57
Table 4: LINE-1 Methylation ................................................................................58
Table 5: PPARγ Methylation ...............................................................................60
Table 6: PGC1-α Methylation ..............................................................................61
Table 7: IGF2 Methylation ...................................................................................61
Table 8: PDK4 Methylation..................................................................................62
Table 9: TCF7L2 Methylation ..............................................................................62
Table 10: Candidate Gene Mean Methylation .....................................................62
Table 11: Correlation Coefficients for LTPA and Candidate Genes ....................63
Table 12: PPARγ Promoter PCR Conditions.......................................................85
Table 13: PPARγ Intron 2 PCR Conditions .........................................................85
Table 14: PGC1α Proximal Promoter PCR Conditions .......................................86
Table 15: PGC1α Exon 1 to Intron 1 PCR Conditions .........................................86
Table 16: IGF2 P2 Promoter PCR Conditions .....................................................87
Table 17: IGF2 P4 Promoter PCR Conditions .....................................................87
Table 18: IGF2 Distal Promoter PCR Conditions ................................................88
Table 19: PDK4 Promoter PCR Conditions .........................................................88
Table 20: TCF7L2 5’UTR and 5’UTR-Promoter PCR Conditions ........................89

ix

LIST OF FIGURES

Figure 1: ARCH Questionnaire ............................................................................79

x

KEY TO ABBREVIATIONS

ACOG: American College of Obstetricians and Gynecologists
ACSM: American College of Sports Medicine
AGA: Average for gestational age
ANOVA: Analysis of variance
ARCH: Archive for Research on Child Health
BMI: Body Mass Index
CVD: Cardiovascular disease
eNOS: endothelial nitric oxide synthase
GDM: Gestational diabetes mellitus
GR: Glucocorticoid receptor
GWG: Gestational weight gain
HEI: Healthy Eating Index
IGF2: Insulin-like growth factor 2
IOM: Institute of Medicine
LBW: Low birth weight
LDL: Low-density lipoprotein
LGA: Large for gestational age
LINE-1: Long interspersed nucleotide elements
LTPA: Leisure-time physical activity
MC4r: Melanocortin 4 receptor
MET: Metabolic Equivalent

xi

MSU: Michigan State University
MVPA: Moderate or vigorous physical activity
Ng: Nanograms
OPA: Occupational physical activity
PCR: Polymerase chain reaction
PDK4: Pyruvate dehydrogenase lipoamide kinase 4
PGC1-α: Peroxisome proliferator-activated receptor gamma coactivator 1-alpha
PPARγ: Peroxisome proliferator-activated receptor gamma
RXRA: Retinoid-x receptor-α
SES: Socioeconomic status
SGA: Small for gestational age
SNP: Single nucleotide polymorphism
TACSTD2: Tumor-associated calcium-signal transducer 2
TCF7L2: Transcription factor-7 like 2

xii

INTRODUCTION

The number of overweight and obese adults and children is increasing
worldwide. In the United States, 16.9 percent of children and adolescents ages 219 were classified as obese and another 31.7 percent were classified as
overweight based on national survey data from 2007-2008 (Ogden, Carroll,
Curtin, Lamb, & Flegal, 2010), and two thirds of American adults were classified
as either overweight or obese (Flegal, Carroll, Ogden, & Curtin, 2010). Because
obese children are more likely to become obese adults compared to normal
weight children (Dietz, 1998), identifying the causes of the rise in obesity and
implementing preventative and treatment programs is important from a public
health perspective. Traditionally the “big two”—poor diet and physical inactivity,
have been blamed for the rising obesity epidemic in both children and adults.
However, recent evidence suggests that numerous other factors, including the
prenatal and maternal environment, may also play a role (Poston, Harthoorn, &
Van Der Beek, 2011).
Cardiovascular disease (CVD) is responsible for approximately 34% of all
deaths in the United States each year (Lloyd-Jones et al., 2009). Atherosclerosis
begins early in life (Berenson et al., 1998; Mahoney et al., 1996), and there is
also evidence that CVD risk factors track moderately well from childhood into
adulthood and likely contribute to CVD morbidity (Katzmarzyk et al., 2001). The
prevalence of CVD risk factors in children has increased significantly over the
past three decades and thus, it is important to address this issue. However,

1

interventions designed to reduce CVD risk factors in children and adolescents
have been only moderately successful at best (van Sluijs, McMinn, & Griffin,
2008). Investigators have begun to consider alternative strategies, such as
determining whether a child's CVD risk factor profile is determined before birth,
and if so, whether positive genetic alterations in offspring can occur through
altered maternal behaviors.
That CVD and obesity risk factors can be traced to fetal origins is believed
to be related to a phenomenon known as “programming” (Barker, 1992).
Programming posits that critical developmental periods for the heart and blood
vessels exist during embryonic and fetal life. If, during these periods, the
maternal-fetal unit is unduly stressed, permanent developmental alterations can
occur. This has been shown experimentally in animal models (Hahn, 1984;
Langley & Jackson, 1994; Winick & Noble, 1966). Epidemiological evidence has
shown that programming may occur in humans, primarily in low birth weight
offspring and those at the lower end of the normal birth weight range (Law et al.,
2002). Most studies on fetal programming have focused on birth weight and
offspring health, but few have considered the relationships between maternal
behaviors during pregnancy and offspring health.
Results from a number of animal studies of maternal nutrition during
pregnancy and epigenetic programming effects in the offspring (reviewed in
(Burdge & Lillycrop, 2010b; Heerwagen, Miller, Barbour, & Friedman, 2010),
suggest that the maternal environment during development influences the
offspring phenotype, particularly in terms of metabolic health. It is reasonable to

2

pursue the study of the effects of maternal physical activity in the same manner,
yet this has not been done to date.
Numerous long-term benefits of maternal activity for the offspring have
been suggested, but few well designed studies have been conducted. The
majority of research in this area has been focused on risk of development of
obesity and co-morbidities in the offspring of exercising pregnant mothers, and
much of that work has been conducted in animal models due to the ease of
controlling both diet and genetic predisposition to obesity. In one recent human
study, Hopkins and colleagues published work demonstrating that a moderateintensity cycling program during the second half of pregnancy resulted in
offspring who weighed less and were less fat at birth (Hopkins, Baldi, Cutfield,
McCowan, & Hofman, 2010).They also found that serum IGF-I and IGF-II were
lower in the cord blood of the offspring of exercising mothers compared to
controls, suggesting that hormonal stimulation of fetal growth was less in the
exercise group.
A mechanism by which offspring phenotype might be affected by maternal
exposures or behaviors is epigenetics, which are heritable chemical changes in
DNA that may influence the expression of the gene. In terms of lifestyle and its
influence on the epigenome, the most well-studied epigenetic change is DNA
methylation (Alegría-Torres, Baccarelli, & Bollati, 2011), where methyl groups are
added to cytosine bases in CpG islands. High levels of methylation are
associated with low promoter activity of the particular gene (and lower gene
expression) and low levels, then, are associated with high promoter activity and

3

high amounts of gene expression. Whether high or low methylation level is
advantageous in terms of programming depends on the gene in question and
may depend on the tissue being studied as well. In humans, only one study has
shown varying levels of DNA methylation with a lifestyle exposure; individuals
who were in utero during the Dutch Famine of 1944 had lower levels of DNA
methylation genome-wide compared to their same-sex siblings who were not
prenatally exposed to the famine (Heijmans et al., 2008). Other studies in animal
models have shown DNA methylation differences in animals fed normal diets
compared to protein restricted diets; the direction of difference depended on the
gene being studied (Burdge & Lillycrop, 2010a).
Though the rationale for studying DNA methylation as a mechanism of
how fetal programming might occur is strong, to date, no studies have measured
methylation differences in human tissue from offspring of exercising or otherwise
physically active mothers compared to low active controls. The purpose of this
dissertation study is to evaluate the differences in global DNA methylation and
methylation of candidate genes with known metabolic function in blood spots
from infants born to physically active compared to low active mothers.
Specifically, leisure time physical activity (LTPA), defined as activity performed
during leisure time that resulted in a rise in energy expenditure above baseline
levels, was assessed via questionnaire. Physical activity data have been
collected on over 300 pregnant women as part of the Archive for Research on
Child Health (ARCH) study. These data, as well as possible covariates, were
used to select the most and least active subjects whose infants’ blood spots were

4

then requested from the Michigan Neonatal Biobank. Using the ARCH data and
DNA extracted from the requested blood spots, this dissertation addresses the
following Research Aims.
Research Aims
Specific Aim 1: Investigate global DNA methylation of newborn dried blood
spots of infants born to pregnant women reporting high levels of LTPA compared
to low active controls, in order to determine overall methylation in repetitive DNA
stretches in humans and how they may be affected by high levels (greater than
three times minimum recommended levels) of maternal LTPA. The MethylFlash
assay and long interspersed nucleotide element (LINE-1) methylation will be
used to assess global DNA methylation.
Hypothesis 1.1: Global DNA methylation will be greater in blood spots of
offspring of pregnant women reporting high LTPA compared to offspring of
low active controls.
Specific Aim 2: Determine whether methylation levels of metabolism-related
candidate genes are different in newborn dried blood spots of infants born to
pregnant women reporting high levels of LTPA compared to offspring of low
active controls. Candidate genes are peroxisome proliferator-activated receptor-γ
(PPAR-γ), proliferator-activated receptor-γ co-activator α (PGC1-α), insulin-like
growth factor 2 (IGF2), pyruvate dehydrogenase kinase 4 (PDK4), and
transcription factor 7-like 2 (TCF7L2).
Hypothesis 2.1: Methylation levels will be different for each of the five
candidate genes in blood spots of infants born to pregnant women

5

reporting high levels of LTPA compared to low active controls. Specifically,
we hypothesize that, compared to low active controls, women with high
LTPA will have lower methylation levels for PPAR-γ, PGC1-α,and PDK4,
but higher methylation levels for IGF2 and TCF7L2.
Organization of Dissertation
Chapter one is the introduction and specific aims while the literature
review is presented as Chapter two. The methods section of the dissertation is
Chapter three and all results are presented in Chapter four. Lastly, Chapter five
is a summary of the results and discussion of how these results should be
interpreted, as well as future research directions. References and appendices of
relevant supplemental materials are also included in the final dissertation.

6

REVIEW OF LITERATURE
Physical Activity and Pregnancy
Leisure time physical activity (LTPA) is activity that causes a rise in energy
expenditure above baseline levels and occurs during an individual’s leisure time
rather than as part of their occupation or transportation (Dishman, Heath, &
Washburn, 2004). Occupational physical activity (OPA) is energy expenditure
above baseline that is used to perform vocational or work activities, often in an
eight hour work day, and other types of activity that may, depending on the
researcher, include transportation, care for children, pets, and other individuals,
and household chores (Howley, 2001). Exercise is planned, purposeful physical
activity, often done with the goal of improving physical fitness and health
(Dishman, et al., 2004). The intensity, duration, and total volume of physical
activity or exercise that a person engages in influences health and well-being.
Therefore, it is important to consider physical activity when studying public health
outcomes.
LTPA levels, as assessed by both accelerometer and self-report, decline
over the course of pregnancy (Rousham, Clarke, & Gross, 2006). The reasons
for the decline are many, but include physical limitations, response to medical
advice to reduce or cease exercising, or perceived risks associated with activity,
particularly late in pregnancy (Clarke & Gross, 2004). Similar reasons for
pregnant women reducing LTPA or stopping completely were found by
Duncombe et al., but their study participants added that feeling tired or unwell
and lack of time kept them from maintaining LTPA levels throughout pregnancy

7

(Duncombe, Wertheim, Skouteris, Paxton, & Kelly, 2009). In all, the evidence is
quite consistent that LTPA levels decrease as pregnancy progresses, and the
reasons for this decline are numerous. More data supporting the safety and
efficacy for improved maternal and birth outcomes of LTPA during pregnancy
may encourage pregnant women to engage in LTPA and may inspire their
physicians to spend more time discussing the benefits of LTPA for both mother
and baby.
Benefits of Physical Activity During Pregnancy--Fetal
Historically there has been concern for pregnant women participating in
physical activity, particularly in strenuous activities. Concerns include shunting of
blood away from the placental perfusion toward exercising skeletal muscle,
hyperthermia that might result from activity and how this might influence the
fetus, and the impact of repetitive movements such as running or
jumping(Hopkins & Cutfield, 2011). However, the relatively limited research to
date have not substantiated these concerns.
Maternal pregnancy physical activity does not appear to place the fetus at
additional risk for adverse outcomes such as LBW or preterm birth. For example,
Kardel and Kase showed that maternal exercise did not increase risk of LBW, nor
did it affect Apgar scores or duration of labor (Kardel & Kase, 1998). In another
study, Sternfeld et al. found that gestational age and birth weight were unrelated
to exercise intensity during pregnancy (Sternfeld, Quesenberry, Eskenazi, &
Newman, 1995b). There is limited evidence that LTPA is related to preterm birth,
but these data are not strong enough or of sufficient quantity to make a

8

conclusive statement in favor of a protective effect of LTPA. Owe and colleagues
found a protective effect of maternal exercise against preterm birth in a large
cohort study (n = 61,098) (Owe, Nystad, Skjaerven, Stigum, & Bø, 2011) while a
similar but smaller study in Brazil also showed a protective effect of LTPA on
preterm birth risk (n = 4,147) (Domingues, Barros, & Matijasevich, 2008). It is
important to keep in mind that there are a number of possible confounders to the
relationship between preterm birth and LTPA (as well as any birth or maternal
outcome and LTPA) including maternal age, smoking, race and so on, and
different studies control for some but not all of them. This may explain, at least in
part, differences in results among studies.
The risk for large for gestational age (LGA) birth appears to be reduced
with maternal physical activity during gestation. Mothers who performed at least
two hours per week of physical activity during pregnancy had a 70% reduced risk
of giving birth to a LGA infant compared to those who were physically active less
than two hours per week (Alderman, Zhao, Holt, Watts, & Beresford, 1998). In
the same study, the infants born from active mothers were not at increased risk
of small for gestational age (SGA) birth. Similarly, the authors also found a link
between LGA odds and maternal size, and suggested that maternal glucose
tolerance may play a role in predicting LGA risk. In a study utilizing the Danish
National Birth Cohort data, exercising mothers had a slightly decreased risk of
having either a SGA or LGA infant compared to non-exercisers (hazard ratios
0.87 and 0.93, respectively), even after controlling for gestational age and
mother’s pre-pregnancy body mass index (BMI) (Juhl, Olsen, Andersen, Nøhr, &

9

Andersen, 2010). Mudd et al. found that maternal LTPA during pregnancy was
related to significantly lower odds of LGA (odds ratio = 0.3, 95% CI = 0.14 –
0.64) but not SGA (Mudd, 2009; Mudd et al., 2012). The study also showed a
positive association between maternal BMI and LGA risk. Overall, the literature
shows mixed results as to whether SGA risk is reduced or unaffected, but lower
LGA risk is consistently related to maternal LTPA during pregnancy. It is
important, however, to additionally study the effects of maternal size and glucose
handling on offspring size, as these factors also appear to influence LGA risk. At
worst, it appears that physical activity during pregnancy has no deleterious effect
on birth weight. At best, maternal physical activity appears to reduce the risk of
birth weight falling outside the normal range.
Benefits of Physical Activity During Pregnancy--Maternal
In addition to not causing harm, physical activity during pregnancy has
been shown to have a number of beneficial effects for the mother. First, physical
activity is associated with a reduced risk of preeclampsia, a condition
characterized by high blood pressure and proteinuria that can develop in the
second half of pregnancy. A case-control study that included women who had a
prior preeclamptic pregnancy found that women who participated in light or
moderate intensity activity prior to pregnancy had reduced risk of a subsequent
preeclamptic pregnancy by 24 percent, while those who performed vigorous
activity had a reduced risk of 54 percent (Sorensen et al., 2003). In another
study, Saftlas et al. asked women to recall leisure time and work physical activity
after their pregnancy and found that women who regularly participated in LTPA

10

or who had occupations requiring physical activity were at decreased risk of
preeclampsia, yet the confidence intervals included one and thus the effect is not
significant (Saftlas, Logsden-Sackett, Wang, Woolson, & Bracken, 2004).
Though the study had a large number of controls (n = 2,422), there were many
fewer cases (n = 44) which resulted in limited statistical power. LTPA may also
reduce risk for development of gestational diabetes (GDM) during pregnancy. In
a randomized control trial where pregnancy women were assigned to either a
pregnancy-long physical activity program consisting of light resistance and toning
exercises or a control group, the physically active women performed better on a
glucose screen at 24 to 28 weeks gestation and had lower incidence of GDM
(Barakat, Cordero, Coteron, Luaces, & Montejo, 2011), but there were design
flaws in the study. Another group instigated a 12-week physical activity
intervention in the second half of pregnancy and found no effect of the
intervention on insulin resistance of GDM incidence (Stafne et al., 2012). Metaanalyses of studies of either dietary or physical activity intervention during
pregnancy on GDM risk have shown that most interventions are ineffective in
preventing this outcome (Oostdam, van Poppel, Wouters, & van Mechelen,
2011). Many studies on this topic are either small or poorly designed, but better
designed randomized control trials are currently being conducted.
Few studies have addressed whether physical activity or exercise during
pregnancy is related to a mother’s risk for needing a caesarean section for
delivery. The largest of these studies (n = 1,342), was conducted using the
PRAMS database in North Carolina. The authors found no relationship between

11

maternal physical activity and delivery mode, though there was a trend toward a
protective effect of physical activity against caesarean section (Bovbjerg & SiegaRiz, 2009). Bungum and colleagues showed that low active women had a 4.48
(95% CI = 1.2 – 16.2) times higher odds of having a caesarean delivery
compared to women who reported LTPA during their pregnancy, even with a
small sample size of 137women (Bungum, Peaslee, Jackson, & Perez, 2000).
However, another group did not find that maternal LTPA predicted delivery mode
(Sternfeld, Quesenberry, Eskenazi, & Newman, 1995a). More research needs to
be conducted to determine whether physical activity during pregnancy might
reduce need for medically indicated caesarean deliveries. All three studies
included a cohort design, but Bovbjerg et al. used frequency of exercise (days
per week) as the independent variable while the other two studies included total
volume of physical activity. The differing results may also be explained by
different timings of physical activity exposure. The Bovbjerg et al. study only
evaluated exercise in the last three months of pregnancy while Bungum et al.
looked at only the first two trimesters and Sternfeld et al. collected data over all
trimesters.
The role of physical activity on miscarriage risk is not well studied, and
resulting evidence is equivocal. Of six published studies on this topic, two found
that physical activity during pregnancy was associated with increased risk of
miscarriage (Hjollund et al., 2000; Madsen et al., 2007), two found no relationship
(Clapp, 1989; Rose, Haddow, Palomaki, & Knight, 1991), and two found a
protective effect of physical activity (Latka, Kline, & Hatch, 1999; X. Zhang et al.,

12

2011). The Madsen study was a large cohort study that found more activity to be
associated with higher miscarriage risk. But, when the authors excluded
participants with potential recall bias due to physical activity ascertainment after
miscarriage, the relationship between physical activity and miscarriage was
attenuated (Madsen, et al., 2007). Two other cohort studies on this topic were
prospective while the Madsen cohort was primarily retrospective interviews with a
subset of the cohort interviewed prior to miscarriage. The other three studies on
this topic are case-control studies: two found maternal physical activity was
associated with decreased miscarriage risk (Latka, et al., 1999; X. Zhang, et al.,
2011) and the Clapp study results showed no difference in risk between cases
and controls (Clapp, 1989). It appears that study design may be related to the
study findings, with case-controls favoring a finding of a protective effect of LTPA
and cohort studies favoring a finding of an elevated risk. Because of the small
number of studies, though, it is difficult to know for certain. Once the Madsen et
al. study is removed from the list of studies finding a positive relationship
between LTPA and miscarriage (due to the recall bias issue), though, the effect
of study type is less obvious. Another possible reason for differences is that two
of the cohort studies (Rose et al. 1991 and Hjollund et al. 2000) assessed LTPA
differently than did the case-control study authors. Rose and colleagues used
total activity (including OPA) as their dependent variable and the Hjollund group
assessed intensity rather than quantity of activity. The current consensus by the
American College of Obstetricians and Gynecologists (ACOG) and American
College of Sports Medicine (ACSM) is that maternal physical activity performed

13

by women with low-risk pregnancies does not increase risk of miscarriage
(ACSM, 2006).
Physical Activity Quantification and Measurement
There are both objective and subjective methods for assessment of
physical activity. Subjective measures include questionnaires, surveys, and
diaries while objective techniques include pedometers, which count the number
of steps an individual takes, and accelerometers, which quantify accelerations of
the hip and are reported in ‘counts’ where a high number of counts indicates a
high amount of activity (Trost, McIver, & Pate, 2005). Direct observation involves
a researcher observing an individual participating in physical activity and using a
designed system to quantify that activity (McKenzie, 2002). Other techniques for
assessment of physical activity include indirect calorimetry which requires the
collection of expired gases, but this technique is not feasible in a free-living
environment or on large samples (Dishman, et al., 2004; McKenzie, 2002), and
doubly-labeled water, which is useful for quantifying energy expenditure over a
period of several days or weeks but can thus not separate physical activity from
resting metabolic rate (Lifson, Gordon, & McClintock, 1955). It is important to
point out that pedometers and accelerometers measure total daily physical
activity rather than solely LTPA or OPA, so it is necessary to have another
means, such as a questionnaire, to assess what percentage of an individual’s
total physical activity meets the definition of LTPA versus OPA.
Selection of physical activity assessment technique depends upon the
research question being asked, the number of subjects to be assessed, and the

14

type of physical activity to be assessed. For example, doubly labeled water is
useful for assessing total energy expenditure, rather than just physical activity,
over a period of time, often a week or more. It is very costly and therefore is
impractical for large samples, but can be very useful as a criterion method
against which other physical activity assessment techniques are validated.
Similarly, direct observation is not realistic for large sample sizes because of the
time-consuming, and thus expensive, nature of the data collection process.
Accelerometers produce large volumes of data and can be expensive in terms of
up-front cost, but provide objective measurement that takes into account intensity
of the activity, which pedometers are not able to do. Surveys and questionnaires
vary widely in the types of questions that they ask; some are lengthy and have
been validated for use in particular populations, and some may include a single
question such as “are you physically active?” The type of data that obtained
depends on the questions asked, and these surveys or questionnaires may be
designed for a specific research question.
Physical activity can be quantified in terms of frequency (how many days
per week or month), intensity (light, moderate, or vigorous), and time (minutes or
hours per day). In order to obtain a total volume of physical activity that takes into
account all of these variables, the units MET·min·week may be used. MET
stands for metabolic equivalent and one MET is equal to the value, at rest, of the
average individual’s energy expenditure. Three METs, then, is equal to three
times resting energy expenditure, and so on. Generally moderate physical
activities are considered those that raise metabolic rate to between three and six

15

METs and vigorous activities are those that have intensity greater than six METs.
By taking into account the intensity, in METs, for any given activity (see the
Compendium of Physical Activity for activities and their MET values (Ainsworth et
al., 2011)) and multiplying by the number of minutes per week the activity was
performed, a total physical activity volume can be calculated in MET·min·week.
One difficulty when performing physical activity research is being able to
separate LTPA from other types of physical activity. For example, it may be that
some investigators would classify child care and household chores as LTPA but
others would call it occupational. Additionally, objective measurements of
physical activity such as accelerometers and pedometers do not provide any
information regarding the type or purpose of the activity; they simply provide a
value for total physical activity over the time the device was worn. This may
influence their ability to accurately measure certain types of activities. For
example, in adult females ages 25 to 74, pedometer step counts were found to
correlate well to LTPA but not OPA (Sequeira, Rickenbach, Wietlisbach, Tullen,
& Schutz, 1995). Another issue with monitors is subject compliance, which may
change throughout pregnancy. Rousham et al. measured physical activity
throughout pregnancy by both accelerometry and interview, and found that the
correlation between the two methods decreased significantly as pregnancy
progressed (Rousham, et al., 2006). For these reasons, it may be beneficial to
measure physical activity during pregnancy using both subjective and objective
techniques when feasible. If the investigator’s interest is in LTPA alone, it is most
realistic to assess total physical activity with objective measures and use

16

questionnaires or diaries to determine what type of activities are being done or
what percentage of total physical activity can be classified as LTPA.
It is important for many research questions to be able to differentiate
between LTPA and OPA. Some studies have shown that strenuous physical
work is associated with intrauterine growth restriction (OR 2.46 95% CI 1.36 to
4.21 for women who report moderate to vigorously strenuous versus light)
(Spinillo et al., 1996) and premature delivery (OR 2.72 95% CI 1.24 to 5.95 for
women whose jobs required prolonged standings compared to those whose did
not) (Teitelman, Welch, Hellenbrand, & Bracken, 1990). One reason for the
possible adverse effects of OPA is the psychological as well as physical stress
that is associated with these working conditions (Clapp, 1996). Due to these
physiologic differences in the effects of OPA and LTPA, it is often important to
study them separately, and therefore methods to assess one or both separately
should be developed.
Depending on the research question, it may be important to assess
physical activity levels during pregnancy and determine if women are meeting the
recommendations. Ideally, this assessment should be repeated across trimesters
as the exposure may change over time as pregnancy progresses. In addition,
accurate measurement helps us determine which types of physical activity, and
how much, might result in the greatest benefit to the maternal-fetal unit. In the
first trimester women tend to report feelings of fatigue and/or nausea as reasons
for not participating in activity, and in the last trimester, change in body size and
center of gravity is more likely to prevent them from meeting physical activity

17

recommendations (Leiferman, Swibas, Koiness, Marshall, & Dunn, 2011).
Therefore, methods of physical activity assessment may need to be adapted for
use in pregnant women. In addition, assessment of activity via devices such as
pedometers or accelerometers may be influenced by pregnancy in terms of
monitor tilt due to gestational weight gain. For example, DiNallo and colleagues
found that during a controlled exercise bout on a treadmill, pedometer steps,
Actigraph accelerometer counts per minute, and minutes of moderate or vigorous
physical activity (MVPA) all were lower at 32 weeks gestation compared to 20
weeks (DiNallo, Downs, & Le Masurier, 2012). Therefore, considerations for
maternal weight gain, monitor tilt, and gestational age must be taken into account
when designing studies for assessing physical activity during pregnancy.
Physical Activity Recommendations for Pregnancy
The first generally accepted recommendations for pregnancy physical
activity in the US were introduced in 1985 and were quite conservative; pregnant
women were advised to keep their heart rates under 140 beats per minute at all
times and avoid strenuous activity for more than 15 minutes at a time (ACOG,
1985). Since then, much research has been conducted on the topics of safety
and benefit of physical activity or exercise throughout pregnancy, and the
overwhelming consensus is that it is both safe and effective at improving birth
outcomes in the majority of pregnant women (ACSM, 2006). Recommendations
for physical activity during pregnancy have also been updated, and are now very
similar to recommendations for the general population. In 2002, ACOG published
their expert committee opinion that most women can benefit from physical activity

18

during pregnancy and that in the absence of obstetric complications, they should
aim for 30 minutes of moderate intensity exercise on most if not all days of the
week (ACOG, 2002). The ACOG guidelines stipulate, however, that activities
potentially involving abdominal trauma, as well as scuba diving, which places the
fetus at risk for decompression sickness, should be avoided. The 2008 Physical
Activity Guidelines for Americans provide very similar recommendations; they
suggest 150 minutes per week of moderate intensity physical activity and
indicate that the 30 minutes per day can be accumulated in bouts over the
course of the day ("Physical Activity Guidelines Advisory Committee report, 2008.
To the Secretary of Health and Human Services. Part A: executive summary,"
2009).
It is important to note that most pregnancy physical activity
recommendations are for aerobic exercise, and that the safety of resistance
training has been studied very little in pregnancy. To our knowledge, only one
group has conducted a randomized control trial of resistance training in
pregnancy. Pregnant women were randomly assigned to either a resistance
training group, where they did light resistance and toning exercises three times a
week for 35 minutes per session, or a control group. There were no differences
between the groups in gestational age at delivery (Barakat, Stirling, & Lucia,
2008), mode of delivery (Barakat, Ruiz, Stirling, Zakynthinaki, & Lucia, 2009), or
birth size (Barakat, Lucia, & Ruiz, 2009). These studies were conducted on
relatively small samples (70 to 80 interventions and the same number of
controls), so it is difficult to determine whether resistance training might have

19

positive effects on birth outcomes or offspring health. Another group recently
examined the association between resistance training during pregnancy and birth
outcomes using a cross-sectional study design; they found no difference in
gestational age at delivery or delivery mode between women who participated in
resistance training and those who did not (E. White, Pivarnik, & Pfeiffer, 2013). In
addition to no effects on gestational age or delivery mode, the authors found a
protective effect of resistance training on gestational diabetes and hypertensive
disorders during pregnancy, suggesting that, at the least, resistance training
during pregnancy is not detrimental to the health of the fetus. Taken together,
these studies suggest that light resistance training is safe during pregnancy. The
effects of more strenuous activity are still unknown.
Fetal Origins
Maternal Weight and Weight Gain
The Fetal Origins of Adult Disease hypothesis, also termed the Barker
hypothesis, was developed based upon an ecological model that showed
correlations between geographic regions with high infant mortality rates and high
death rates, in adults, from cardiovascular and respiratory disease(Barker &
Osmond, 1986). Studies revealed that both placental size and infant birth weight
were related to systolic and diastolic blood pressure in adulthood (Barker, Bull,
Osmond, & Simmonds, 1990), and growth patterns in early life were related to
incidence of coronary events and insulin resistance in adulthood (Barker,
Osmond, Forsen, Kajantie, & Eriksson, 2005). However, there is controversy
surrounding some of the statistical analyses that Barker used, particularly in the

20

study of the relationship between birth weight and blood pressure. For example,
Paneth and colleagues argue that controlling for adult weight in such an analysis
is inappropriate (Paneth, Ahmed, & Stein, 1996), and Kuzawa demonstrates by
investigating a supply-and-demand hypothesis that prenatal nutrition is related to
low-density lipoproteins (LDL) and systolic blood pressure—not diastolic blood
pressure or other cholesterol measures (Kuzawa, 2004). Still, other outcomes,
including hypertension and endothelial dysfunction, do have stronger evidence
supporting an association with prenatal nutrition (Burdge & Lillycrop, 2010a;
Langley-Evans & McMullen, 2010), and certainly there are ample data
suggesting that various events during fetal development, not limited to nutrition,
influence the offspring phenotype. For instance, maternal smoking is strongly
related, in a dose-response manner, to low birth weight risk (reviewed in (Walsh,
1994)) and heavy alcohol consumption during pregnancy has a robust
association with central nervous system impairment and diminished fetal growth
in offspring (reviewed in (Little & Wendt, 1991)).
The prenatal environment may predispose the fetus to chronic disease
later in life. Maternal pre-pregnancy weight status and gestational weight gain
(GWG) are two relatively easy-to-assess maternal characteristics that may alter
the prenatal environment and may predict future poor maternal health and
elevated chronic disease risk (Linné, Dye, Barkeling, & Rössner, 2004;
Whiteman et al., 2011). Pre-pregnancy obesity and excessive or inadequate
GWG may reflect poor diet, physical activity, socioeconomic status, or other
stressors in the mother and these environmental factors potentially contribute to

21

a less than optimal environment in utero (Viswanathan et al., 2008). Numerous
studies have identified a relationship between pre-pregnancy weight status and
offspring weight status at both birth and throughout childhood and adolescence
(Gale et al., 2007; He, Ding, Fong, & Karlberg, 2000; Koupil & Toivanen, 2008;
Laitinen, Power, & Jarvelin, 2001; Reynolds, Osmond, Phillips, & Godfrey, 2010).
Rooney et al. found that in their cohort, maternal obesity was the strongest
predictor of offspring obesity in childhood and adolescence (Rooney, Mathiason,
& Schauberger, 2010), and another longitudinal study demonstrated that
maternal obesity was associated with weight and length at birth but not at ages 1
and 3 months (Regnault et al., 2010). There are a large number of studies on the
topic of maternal obesity and offspring weight status (at various time points) with
mixed results; a meta-analysis of 45 medium or high-quality studies on the topic
showed an odds ratio of 1.53 (95% CI 1.44 to 1.63) for LGA and 1.95 (95% CI
1.77 to 2.13) for subsequent offspring overweight/obesity in mothers who were
overweight or obese prior to pregnancy (Yu et al., 2013). Thus, it appears that in
general there is a positive association between maternal and offspring obesity,
but study findings might be affected by the timing of offspring measurements,
and maternal weight status may be more influential at some time points than
others. Taken together, these results suggest that environmental factors
influencing the offspring after birth are more strongly at play than is the prenatal
environment, but this is not yet clear from the literature.
Additional studies have found independent effects of maternal BMI prior to
pregnancy on other adverse health outcomes in offspring. Obese but not

22

overweight women are more likely to give birth to an infant with a congenital
heart defect, with the most obese women having the greatest risk (Mills,
Troendle, Conley, Carter, & Druschel, 2010). Another study showed that
maternal BMI is negatively associated with cardiorespiratory fitness in 9-year old
offspring, as measured by maximal cycle ergometer test, even after controlling
for child body fatness. However, there was no relationship between maternal BMI
and other risk factors for cardiovascular disease, including blood pressure, total
cholesterol, glucose and insulin levels, and triglycerides (Labayen et al., 2010). In
another analysis, researchers showed a higher odds (1.87) of development of
severe asthma in the first seven years of life in children whose mothers were
obese (BMI > 35) during pregnancy (Harpsøe et al., 2013). A review article on
the topic of maternal overweight/obesity during pregnancy and offspring health
articulated increased risk of congenital anomalies, small for gestational age,
LGA, preterm birth, neonate mortality, reduced breastfeeding success, long-term
obesity risk, and metabolic and cardiovascular abnormalities later in life (RuagerMartin, Hyde, & Modi, 2010). Overall, there is significant evidence that a mother’s
weight status prior to pregnancy affects the health of her child in a variety of
ways throughout his or her life.
Despite a fairly large amount of published data on maternal weight status
in pregnancy and effects on offspring, it is not well known whether or how
maternal pre-pregnancy obesity or weight gain during pregnancy contributes
toward a more risky prenatal environment. In 2009, the Institute of Medicine
(IOM) updated their guidelines for weight gain during pregnancy based on

23

evidence that inadequate or excess gestational weight gain may be related to
adverse health outcomes for the offspring (IOM, 2009). Recommended weight
gain is now based upon the mother’s BMI prior to becoming pregnant.
Underweight women (BMI less than 18.5) should gain 28 to 40 pounds over the
course of the pregnancy, normal weight women (BMI 18.5 to 24.9) should gain
25 to 35 pounds, overweight women (BMI 25 to 29.9) should gain 15 to 25
pounds, and obese women (BMI greater than or equal to 30) should gain 11 to
20 pounds (IOM, 2009).
Since the IOM update, research has indicated that weight gains outside of
the recommended ranges may affect the offspring adversely. In a study from the
Copenhagen Perinatal Cohort, investigators examined the relationship between
gestational weight gain (GWG) and child obesity in a group of 4,234 motheroffspring dyads. The authors found that GWG was related to offspring BMI at all
ages, including childhood, adolescence, and adulthood (Schack-Nielsen,
Michaelsen, Gamborg, Mortensen, & Sorensen, 2010). They also demonstrated
a linear relationship between GWG and odds of offspring adult obesity. This
suggests that inadequate weight gain based on IOM cut points might actually be
beneficial in regards to offspring weight status because lower GWG, even if less
than recommended by the IOM, is associated with low risk of becoming an obese
adult. Another study showed that the offspring of women gaining less than
recommended weight based on pre-pregnancy BMI had lower levels of adiposity
compared to normal and excess gain offspring, but the cardiovascular risk
profiles were similar among groups (Fraser et al., 2010). In the same study,

24

women who gained greater than recommended weight, additionally, had
offspring who had higher adiposity and greater cardiovascular disease risk at age
nine compared to normal and inadequate gain offspring. Together, these data
indicate that offspring health may be protected by weight gain within or below
IOM recommendations.
Several investigators have used gestational weight gain as a continuous
independent variable to investigate offspring health outcomes. In infants,
maternal weight gain was a significant predictor of weight at birth and at age 3
months (Regnault, et al., 2010). In another study, subjects (offspring) were
followed up at several time points; GWG was a significant predictor of offspring
BMI at adolescence and early adulthood but not childhood (Rooney, et al., 2010),
though the reason is unclear. In 21-year old offspring, mother’s GWG was
significantly and directly related to BMI (Mamun et al., 2009). Mamun and
colleagues also found that systolic blood pressure was greater in offspring whose
mothers had higher GWG, though the relationship was not significant. Another
study showed that GWG was independently associated with offspring percent fat,
measured via skinfolds, at age 30 years (Reynolds, et al., 2010). These studies
all followed individuals to varying ages, and therefore it is difficult to compare the
results. Both the Regnault et al. and Mamun et al. study were large cohorts (n =
1,418 and n = 2,432, respectively) and the other two studies were considerably
smaller. Also, a wide possible number of covariates, including mother’s prepregnancy BMI, breastfeeding or formula feeding, offspring dietary behaviors and
physical activity, and so on, potentially affect the results. Each study used these

25

and other covariates in different combinations, making results difficult to
compare.
Diet
Like maternal BMI and GWG, maternal diet too may affect offspring
health. Natural experiments of famine, either by food shortage (Japan) or war
(Dutch Hunger Winter, Seige of Leningrad), have provided evidence that food
restriction during pregnancy influences offspring health. Data from the Dutch
Hunger Winter show that those offspring prenatally exposed to famine have
diminished glucose tolerance as adults compared to controls who were not
exposed (de Rooij et al., 2006). Also, offspring prenatally exposed to famine in
early gestation had higher risk of coronary heart disease and had adverse lipid
profiles compared to those unexposed (Roseboom, de Rooij, & Painter, 2006).
Effects of famine exposure even affected the children of those individuals who
were prenatally exposed to the Dutch Famine. As infants, the offspring of those
who were in utero during the famine had higher adiposity and as adults, had
poorer overall health compared to controls whose parents were not exposed
(Painter et al., 2008).
Infant hormone levels are affected by mothers’ diet even when the mother
does not have GDM (Butte, 2000). Mothers with lower Healthy Eating Index (HEI)
(i.e., poor diet) scores delivered infants with higher cord blood insulin levels
(Gesteiro, Bastida, & Sánchez Muniz, 2011).The offspring also had a relative risk
of 7.6 for glycemia (cutoff for high glucose was 81.25 mg/dl for males and 82.5
mg/dl for females) and 6.7 insulinemia (cutoff was 4.8 mUI/I for males and 6.4

26

mUI/I for females) (95% CI not given; p = 0.008 and 0.011, respectively)
compared to infants of high-HEI moms (Gesteiro, Rodríguez Bernal, Bastida, &
Sánchez-Muniz, 2012).These hormone levels that influence in utero conditions
may set the stage for infants to develop chronic disease later in their life,
regardless of their own dietary behaviors because their cells are programmed to
deal with the perpetually high levels glucose experienced during fetal
development . It is not known whether some type of intervention, at the maternal
level during pregnancy, might attenuate or even prevent this cellular
programming that places the fetus at risk for disease.
Maternal metabolic health may also contribute to the health of offspring.
Gestational diabetes is a condition that affects 3 to 7 percent of pregnancies and
is associated with maternal obesity (Ferrara, Kahn, Quesenberry, Riley, &
Hedderson, 2004). Mothers with high plasma glucose are more likely to have a
baby born with hypoglycemia (OR = 2.61, 95% CI = 0.99 – 6.92) or macrosomia
(OR = 3.4, 95% CI = 1.55 – 7.43) than pregnant women who have normal
plasma glucose levels (Ferrara et al., 2007). In children, a combination of
maternal GDM and LGA birth weight resulted in a 10.4 times greater odds of
insulin resistance at age 11 compared to children whose mothers did not have
GDM and were not LGA (Boney, Verma, Tucker, & Vohr, 2005).There appears to
be a dose-response relationship between maternal glucose and offspring disease
risk, as the higher the glucose levels the fetus is exposed to in utero, the higher
the risk of metabolic disease in adulthood (Silverman, Metzger, Cho, & Loeb,
1995).

27

Interventions to consider during pregnancy include maternal weight loss
(or behavioral interventions that promote weight loss or body composition
change, such as physical activity). It is possible, but not clear whether maternal
weight loss (in overweight or obese prospective mothers) might improve birth
outcomes and ultimately offspring health later in life. There are limited data on
this topic, particularly on the offspring at any follow-up period after birth. Marceau
and colleagues studied women who underwent biliopancreatic diversion surgery,
a form of bariatric weight loss surgery, and evaluated subsequent pregnancies
and live births (Marceau et al., 2004). The authors found that incidence of
macrosomia in the offspring decreased from 34.8 to 7.7 percent and coincided
with an increase in normal-weight offspring from 62.1 to 82.7 percent. However,
women who had surgery had a higher than normal miscarriage rate (26.0%)
which did not decline with weight loss. The same research group studied children
of women who had bariatric surgery at a time point between two pregnancies.
The authors compared children born prior to and after maternal surgery. They
found that children born after surgery had lower weight at birth (but still in normal
range), and at follow up at ages 2.5 to 26 years. Children born after maternal
surgery also had higher insulin sensitivity, better lipid profiles, and lower Creactive protein (Smith et al., 2009). In another study utilizing the same design,
prevalence of offspring obesity was 52 percent less in children born after surgery
compared to those born before, though the authors did not present data on ages
of the children when height and weight measurements were taken (Kral et al.,
2006).

28

Taken together, these data suggest that maternal weight loss somehow
influences offspring disease risk, presumably independent of child lifestyle
behaviors. It is possible that the mother’s lifestyle changed considerably following
bariatric surgery. The result, theoretically, is that her child born after surgery was
exposed to different dietary and physical activity behaviors, which might partially
explain this phenomenon. Paternal factors and offspring diet, physical activity,
and screen time were not accounted for in either bariatric surgery study. The
authors proposed that the in utero exposures to the fetus after maternal weight
loss somehow program better metabolic health that persists for many years after
birth. Though no studies have shown what mechanism might be driving these
findings, epigenetic changes are suspected (Smith, et al., 2009). Epigenetic
modifications are chemical changes in the DNA that then are passed to the
offspring and may alter gene expression but without changing DNA sequence
(Tost, 2009). In the bariatric surgery studies, the theory is that the maternal
weight loss results in epigenetic changes in the maternal genome that are
passed on to her offspring whose disease risk is lower, compared to a sibling
born before the mother’s weight loss, due to the epigenome alterations.
Other lifestyle behaviors related to metabolic health, including physical
activity, might also drive epigenetic change that predisposes toward or protects
from disease. The role of physical activity in epigenetic effects on DNA has not
been well-studied, especially in humans. There are a number of animal studies of
maternal nutrition during pregnancy and epigenetic programming effects in the
offspring (reviewed in (Burdge & Lillycrop, 2010b; Heerwagen, et al., 2010)), with

29

results suggesting that the maternal environment during development does
indeed influence the offspring phenotype. It thus seems reasonable to pursue the
study of the effects of maternal physical activity in the same manner. However,
we found no studies evaluating the role of physical activity during pregnancy on
epigenetic processes.
DNA Methylation
The basis of the Fetal Origins Hypothesis is that nutrient supply during
fetal development and early infancy influences later health, and epigenetics is a
mechanism by which this may occur. Epigenetics is a heritable change in a gene
that does not result from a change in the DNA sequence (Tost, 2009). The most
commonly studied form of epigenetic variation is DNA methylation, in which a
methyl group binds to cytosine bases, often in the regulatory region of the gene,
altering the expression of that gene (Weber et al., 2007). The idea of
‘developmental programming’ is based upon the concept of epigenetics, where
environmental influences, including diet and physical activity, might alter the
epigenome and then be passed to subsequent generations, predisposing
offspring to particular phenotypes or even chronic diseases. Much methylation
occurs in CpG islands, which are regions of the genome rich in cytosine and
guanine bases (Jones & Takai, 2001). Promoter region of genes that are mostly
methylated are less transcriptionally active and promoters with low methylation
levels are more transcriptionally active (Weber, et al., 2007).
It would be expected, then, that DNA methylation levels correspond to
gene expression, but this is not always the case. For instance, Turan and

30

colleagues identified 23 candidate genes whose methylation levels were able to
explain considerable (70 to 87 percent) variance in birth weight in humans (Turan
et al., 2012). The gene expression levels for these same genes, however,
predicted far less of the variance in birth weight, and correlations between
methylation levels and gene expression was low and for the most part, not
statistically significant. Thus, it is not safe to assume that methylation levels
correlate with gene expression levels of the same gene. Gene expression
measurements are also known to be specific to the timing of the exposure of
interest and thus this outcome can be difficult to study. Methylation levels are
more stable than gene expression and are less affected by measurement and
exposure timing (Tost, 2009). Another issue in the study of DNA methylation is
tissue specificity; some genes have similar methylation levels across a number of
tissues (brain, gonads, heart, intestine, liver, lung, muscle, cord blood, and so on)
while other genes have methylation profiles that vary considerably among the
tissues (Murphy, Huang, & Hoyo, 2012). It is important to be aware of this issue
when reading and interpreting literature on this topic.
Environmental Factors Influence DNA Methylation
Because DNA methylation is mitotically stable, it follows that
environmental factors do not play a large, if any, role in determining DNA
methylation patterns. However, recent evidence suggests otherwise (Lim &
Song, 2012). Schar and Fritsch have described DNA methylation as possessing
‘dynamic stability’ because some methylation patterns are modeled after a
template and other patterns are developed de novo (Schär & Fritsch, 2011). Twin

31

studies have shown that older twin pairs have greater divergence of methylation
patterns compared to younger twins, suggesting that environmental factors affect
these patterns (Fraga et al., 2005). Most evidence for developmental
programming comes from animal studies, due to ethical limitations regarding the
manipulation of the human environment. Several animal studies have shown that
environmental changes in pregnant animals affect their offspring despite no
direct exposure of the offspring to the particular environment.
In mice fed either a standard or obesogenic (high fat, high sugar) diet for
six weeks prior to breeding, offspring of the obese dams, who were all fed
standard chow, ate more, had lower locomotor activity, and higher abdominal fat
pad mass compared to the offspring of the control dams(Samuelsson et al.,
2008). Additionally, offspring of the obese dams demonstrated adipocyte
hypertrophy as well as altered mRNA expression of several genes expressed in
adipocytes. In another mouse study, Kavanagh et al. examined the effects of
neonatal and fetal exposure to trans-fatty acids by developing a complex study
design that exposed offspring to trans-fat, either in utero, through breast milk, or
both. Experimental animals were compared to controls exposed to an identical
diet except for the replacement of trans-fats with cis-fats (Kavanagh et al., 2010).
Exposure to trans-fatty acids via breast milk resulted in slower growth, elevated
fasting glucose, and higher plasma levels of IGF-1, though insulin sensitivity did
not differ. Mice exposed to trans-fatty acids in utero but fed breast milk free of
trans-fatty acids experienced rapid early neonatal growth and had impaired
insulin sensitivity as well as higher abdominal fat pad mass. In another animal

32

study, George et al. examined the effects of overfeeding and maternal weight
gain on offspring growth and development in sheep (George et al., 2010). The
authors found that at mid-gestation, fetal crown to rump length, thoracic and
abdominal girths, and perinatal fat mass were all higher in the offspring of
overweight and obese ewes compared to offspring of controls who gained normal
weight. Additionally, fetal heart, pancreas, and liver weights were greater in the
offspring of obese compared to overweight and control animals. Taken together,
these data show that a variety of environmental exposures during pregnancy, in
animals, can influence offspring characteristics related to metabolic health.
In addition to animal studies, a few human studies have focused on
dietary alterations during pregnancy and influences on offspring. Khulan et al.
(2012) studied umbilical cord blood of offspring born to Gambian mothers who
were given micronutrient supplements during pregnancy (compared to controls
given placebos). The authors found different methylation patterns in the groups in
a large number of genes (108 in males, 106 in females), some of which are
imprinted genes (Khulan et al., 2012). The imprinted genes, which are genes in
which epigenetic mechanisms ‘silence’ either maternal or paternal expression of
the gene, tended to be different not at birth but at 9 months of age, suggesting
that these genes are not directly influenced by pregnancy micronutrient
supplementation. Instead, the authors suggested that a small number of genes
that are affected by supplementation might in turn affect a larger number of
genes later in postnatal development. In another study, individuals prenatally
exposed to famine during the Dutch Famine had less DNA methylation of the

33

IGF2 gene compared to siblings of the same sex, who were not exposed to the
famine (Heijmans, et al., 2008). Low methylation of this gene, which is related to
growth, should correlate to higher gene expression and thus more growth than in
the unexposed sibling. These studies demonstrate that human offspring are
affected by the maternal environment and that environmental stimuli such as diet
may impact the newborn throughout his or her lifetime.
Though few studies have directly measured methylation levels in the DNA
of experimental animals, evidence for the developmental programming of these
animals is strong. When all else is held constant, maternal nutrition appears to
alter fetal environment in such a way that the offspring experiences altered
cellular growth. Further research is required to understand how this might occur
in humans and what role interventions (i.e., physical activity, weight loss) might
have in preventing the passage of adverse epigenetic modifications to the next
generation. Not surprisingly, data supporting developmental programming
suggests that the obesity epidemic is a complex one that will not be solved in one
or two generations.
DNA Methylation is Heritable
Burdge and colleagues conducted the first study that specifically
measured DNA methylation differences as a result of intergenerational
transmission (2007).In this study, the grand-offspring of male rats who were fed a
protein-restricted diet had lower methylation of glucocorticoid receptor (GR) and
PPAR-α gene promoter regions, even though individual animals measured were
all fed normal chow diets their entire lives (Burdge et al., 2007). In another

34

experiment, rodents were overexposed to glucocorticoids which affected both
fetal and placental weights in two subsequent generations (Drake, Liu, Kerrigan,
Meehan, & Seckl, 2011).
Gervin et al. studied heritability of DNA methylation in humans by
comparing the similarities of monozygotic versus dizygotic twins. The authors
found low heritability (2 to 16 percent), suggesting that the environment
influences DNA methylation to a greater extent than do genetic factors (Gervin et
al., 2012). In contrast, Dunn and Bale studied mice and demonstrated that body
length and insulin sensitivity phenotypes were heritable across two generations
(i.e., the “grandchildren” were affected by the diets of two generations previous)
(Dunn & Bale, 2009). Though specific methylation patterns were not measured,
studies of women who were pregnant or gave birth during or shortly after the
Dutch Famine of 1944-1945 showed that some phenotypes were altered in
offspring. The offspring (F1; the first generation of offspring of those directly
exposed to a stimulus) of those exposed to the famine in utero (F0) had greater
adiposity as infants, and poorer health later in life, compared to F1 generation
whose mothers were not exposed to famine in utero (Painter, et al., 2008).
However, prevalence of Metabolic Syndrome and cardiovascular disease was
not affected, nor was birthweight. Timing of the exposure may also be important;
birthweight was lower in the F1 generation if the mother was exposed to caloric
deficit in the first or second trimester, but there was no difference in birthweight if
the exposure came in the 3rd trimester (Lumey & Stein, 1997). Overall, DNA

35

methylation appears to be heritable in both humans and rodents, though the type
and timing of exposure influence the outcome in terms of phenotype.
Factors that Influence DNA Methylation
Diet
Diet-induced cellular changes (in the mother or child) may also influence
phenotypic variance among infants and children. Methyl groups come mainly
through consumption in the diet, and these methyl groups are needed for DNA
methylation, a heritable change in the chemical composition but not sequence of
a gene (Choi, Corrocher, & Friso, 2009). It is plausible, then, that diet affects
DNA methylation, which in turn affects gene expression and potentially alters the
phenotype that is expressed. Evidence of this in humans in scarce, but animal
models suggest it likely occurs (Dominguez-Salas, Cox, Prentice, Hennig, &
Moore, 2012). In one of the few studies of epigenetic modifications in healthy
human children, newborn methylation patterns were found to explain 25% of the
variance in adiposity amongst 9 year olds. Some of these methylation pattern
differences were related to lower maternal carbohydrate intake in early
pregnancy (Godfrey et al., 2011). This is significant because the lower maternal
carbohydrate metabolism was also associated with elevated adiposity in
childhood. In a very recent study, Zhang and colleagues distributed selfadministered questionnaires on diet and physical activity to 165 healthy subjects,
ages 18-78 years. White blood cell global methylation patterns were not
associated with any diet or physical activity variables except for dietary folate
intake from fortified foods (Zhang et al., 2012).

36

Because there are so few data available, it is difficult to achieve
consensus on the role of diet on human DNA methylation. Still, it appears that
early pregnancy diet, particularly carbohydrate and folate intake, affects DNA
methylation patterns in offspring. In animal models, maternal protein restriction
during gestation has been associated with hypermethylation of total DNA in fetal
liver (Rees, Hay, Brown, Antipatis, & Palmer, 2000) and also in PPAR-α promoter
of the offspring’s liver (Lillycrop et al., 2008) and X-receptor promoter (van
Straten et al., 2010). A high fat diet resulted in altered methylation of the
melanocortin-r receptor (Mc4r) in the brain tissue of two lines of mice (Widiker,
Karst, Wagener, & Brockmann, 2010). This receptor has been implicated in
weight regulation. Animal model studies have shown that maternal diet
influences methylation of several different genes in a variety of tissues, and the
alteration appears to be very specific to the stimulus. Further study is required to
better understand the mother’s diet in pregnancy and how it may affect her
offspring not only at birth, but beyond.
Sex
Little is known about methylation pattern differences between the sexes,
but one recent study found that, with the exception of imprinted genes, CpG sites
studied were significantly more methylated in males compared to females (ElMaarri et al., 2007). The DNA for this study was isolated from blood samples of
96 healthy males and 96 healthy females. The authors selected another sample
of 48 individuals from each gender to confirm their findings; they found very
similar results. In another study (which was not designed to examine differences

37

between sexes, but the results stood out to the investigators), white blood cells
were extracted from whole blood of 134 male and 157 female subjects (Sarter et
al., 2005). They investigated methylation patterns on four autosomal genes, and
males had higher levels of methylation compared to females in three of those
genes (no difference in methylation in the fourth gene). It appears that males
tend to have higher methylation levels than females and therefore, depending on
the study design, it may be important to use subject sex as a covariate or to
stratify by sex for analysis.
Race
Racial differences in methylation patterns also exist. African-American and
Caucasian newborns have different methylation patterns in their cord blood DNA;
13.7% of CpG sites investigated (out of 26,000+ sites) were found differentially
methylated according to race, while only 2% were differentially methylated
according to gender (Adkins, Krushkal, Tylavsky, & Thomas, 2011). Another
study found that global DNA methylation, measured as LINE-1 methylation in
peripheral blood in adults, is lower in non-Hispanic blacks compared to nonHispanic whites (F. F. Zhang, R. Cardarelli, J. Carroll, K. G. Fulda, et al., 2011).
Like sex, race might be an important potential confounder to consider during
design and analysis in DNA methylation studies.
Maternal Age
Another factor that has been found to be associated with different DNA
methylation patterns at birth is maternal age. Methylation levels of umbilical cord
blood samples at 144 CpG sites on 142 genes were significantly correlated to the

38

mother’s age at birth, and a weaker correlation was found with paternal age; the
direction of the correlation varied by gene but the majority of correlations were
inverse (i.e., greater age was associated with lower methylation of the gene)
(Adkins, Thomas, Tylavsky, & Krushkal, 2011). However, parity was not used as
a covariate in the analysis and therefore it is not clear if the effect is the result of
age or of accumulation of subsequent pregnancies. Assuming it is indeed an age
effect, the clinical significance of this finding is not yet clear, however, and to our
knowledge, no additional studies have investigated this relationship. Adkins et al.
speculated that their findings might have something to do with the fact that
greater maternal age is related to offspring risk of type 1 diabetes (Cardwell,
Carson, & Patterson, 2005) and childhood cancers such as leukemia (Johnson et
al., 2009; Maule et al., 2007; Podvin, Kuehn, Mueller, & Williams, 2006).
Possibly, increased parental age results in methylation changes that are passed
on to offspring and then contribute to disease predisposition, though the degree
of speculation on this mechanism is great.
Body Mass Index (BMI)
Zhang and colleagues found no difference in peripheral blood DNA
methylation in adults by BMI, waist circumference, fat deposition location, or
percent body fat (F. F. Zhang, R. Cardarelli, J. Carroll, K. G. Fulda, et al., 2011).
In another study of children, DNA methylation in peripheral blood (studied on a
panel of 24 genes, with one to three sites per gene) was not associated with BMI,
fat mass, or lean mass at average age 12.35 years, after controlling for multiple
comparisons (Relton et al., 2012). In contrast, Godfrey et al. found that umbilical

39

cord blood DNA methylation at two gene promoters, retinoid-X receptor-α
(RXRA) and endothelial nitric oxide synthase (eNOS), is associated with that
child’s adiposity at age nine years (Godfrey, et al., 2011). Most recently, Groom
et al. found that tumor-associated calcium signal transducer 2 (TACSTD2) gene
expression and methylation was associated with fat mass at age nine to 15
years, but they were not able to replicate the finding in a second cohort (Groom
et al., 2012).
Variability in the findings of the relationships between BMI/body weight
and DNA methylation speaks to the complexity of this topic. Some data are from
children but the ages are varied, and some data are from adults. Two of these
four studies found a relationship between DNA methylation at birth and later body
composition or BMI, but the role of the mother’s BMI and adiposity on her child’s
epigenetic profile was not assessed. To our knowledge, no data exists to address
the effects of a mother’s BMI or body fatness on the epigenetic profile of her
offspring. Maternal physical activity is known to influence body weight and GWG,
but how a woman’s LTPA during pregnancy might alter the offspring epigenome
has also not been studied. More research is needed to clarify these mixed results
and provide further understanding of how maternal weight and body composition
may contribute to the programming of offspring phenotypes.
Physical Activity
The beneficial effects of physical activity for pregnant women are fairly
clear, but the effects on the developing fetus are less clear and not as wellstudied. Hopkins and colleagues have recently published work demonstrating

40

that a moderate-intensity cycling program during the second half of pregnancy
resulted in offspring who weighed less and were less fat at birth, though there
was no effect on maternal insulin sensitivity (Hopkins, et al., 2010). The authors
also found that serum IGF-I and IGF-II (both growth factors) were lower in the
cord blood of the offspring of exercising mothers compared to controls,
suggesting that hormonal stimulation of fetal growth was less in the exercise
group. Mothers in this study were normal weight and healthy, but Hopkins et al.
speculated that effects of maternal exercise on the offspring of overweight or
obese women might be even more profound. To our knowledge, no other studies
have examined the effects of maternal PA on offspring blood markers that may
predict future offspring health. More specifically, no studies have been designed
to examine the effects of maternal physical activity on epigenetic changes in
humans. Though they did not assess gene expression or DNA methylation in
their blood samples, the Hopkins et al. study findings, along with others, suggest
that gestational physical activity influences fetal growth, and it is worth exploring
the mechanism driving this occurrence (Hopkins, et al., 2010).
Metabolism-related Genes Subject to DNA Methylation
There is evidence that metabolic function in adulthood may, at least in
some cases, be ‘programmed’ by exposures in utero (Gabory, Attig, & Junien,
2011). Because physical activity is a known regulator of metabolic function, it
follows that physical activity might alter genes with known metabolic functions
that have the potential to be regulated epigenetically. The following genes have

41

some function in metabolism and the rationale for studying each is expressed in
the following paragraphs.
Proliferator-activated receptor-γ co-activator α (PGC1-α) and Peroxisome
proliferator-activated receptor-γ (PPAR-γ)
PGC1-α is involved in glucose and fat oxidation and gluconeogenesis (in
liver) and PPAR-γ is involved in adipogenesis and insulin signaling. Gemma and
colleagues found a positive correlation between mother’s BMI prior to pregnancy
and methylation level of PGC1-α promoter, though they found no relationship
between maternal BMI and PPAR-γ (Gemma et al., 2009). In another study,
PGC1-α promoter was hypomethylated in a dose-dependent manner in adults
after acute exercise (Barrès et al., 2012). No exercise component was included
in the Gemma et al. study, yet physical activity is known to affect insulin
signaling. PPAR-γ, then, may be epigenetically altered by physical activity. In a
comparable study, subjects were monozygotic twins discordant for type 2
diabetes. The authors assessed methylation levels of a large number of
candidate genes; there was a significantly higher methylation level in the skeletal
muscle of the diabetic twin compared to the non-diabetic (Ribel-Madsen et al.,
2012). Because physical activity is a well-known treatment for type 2 diabetes,
methylation levels of this gene might be influenced by physical activity behavior.
Whether this effect is passed on to offspring is unknown.
Insulin-like growth factor 2 (IGF2)
IGF2 is a maternally imprinted gene, meaning that the maternal copy of
the gene is silenced during embryonic development (via DNA methylation), and

42

only the paternal copy is expressed in the normal human phenotype. BeckwithWiedemann Syndrome is a condition in which silencing of the maternal IGF2
gene copy fails, and the primary phenotypic outcome is overgrowth. Because
IGF2 is related to human growth and known to be epigenetically regulated, this
gene has been well-studied in the DNA methylation literature.
Some maternal attributes have been linked to IGF2 methylation in
offspring. Folate supplementation after twelve weeks of pregnancy was found to
be associated with higher IGF2 methylation in both maternal red blood cells and
offspring cord blood at delivery (Haggarty et al., 2013). Circulating level of IGF2
in maternal circulation during the third trimester of pregnancy was associated
with placental IGF2 methylation, and placental methylation status was also
associated with birth weight (St-Pierre et al., 2012). In another study, cord blood
IGF2 methylation was associated with IGF2 protein concentration, particularly in
the offspring of obese women (Hoyo et al., 2012). Together, these data suggest
that maternal characteristics such as weight status and diet may influence
methylation of the IGF2 gene and in turn, offspring growth. It is plausible then,
that maternal physical activity might influence IGF2 methylation as well.
Pyruvate dehydrogenase kinase 4 (PDK4)
Another candidate gene is pyruvate dehydrogenase kinase 4
(PDK4).Methylation levels of this gene promoter were lower in Type 2 Diabetes
patients compared to non-diabetic controls, and physical activity intervention
(increase of 5 hours per week of activity) resulted in elevated mRNA expression
of PDK4 in individuals with normal glucose tolerance (Kulkarni et al., 2012).

43

Acute exercise also resulted in decreased PDK4 methylation in non-pregnant
adults (Barrès, et al., 2012), but this gene has not been studied in terms of
maternal transmission of methylation to the offspring. There may be differences,
then, in methylation levels in non-pregnant individuals who exercise (chronically)
compared to those who don’t, and this effect may extend to the offspring of
exercising pregnant women.
Transcription factor 7-like 2 (TCF7L2)
Six CpG sites in this gene were found to have altered methylation levels in
adipose tissue, in humans, following a six month exercise intervention (Rönn et
al., 2013). Five of the six genes had increased methylation in response to the
intervention while the other had decreased methylation. This gene has been
associated with Type 2 Diabetes; in a recent study TCF7L2 methylation of
pancreatic islet cells was associated with the TCF7L2 single nucleotide
polymorphism (SNP) genotype, suggesting that the gene regulates islet cell
function to some extent (Dayeh et al., 2013). The combined role of this gene in
etiology of diabetes and its exercise-induced alteration in the Ronn study suggest
it is a reasonable candidate gene for this study.
Summary
In summary, physical activity during pregnancy has known benefits to both
mother and offspring, though the effects on the offspring in particular are less
well understood. Fetal programming occurs when an exposure in the mother is
passed along to the fetus and influences the offspring phenotype; this
programming can be either advantageous or predisposing for the offspring. This

44

dissertation will a) seek to identify if and how the specific exposure of maternal
physical activity programs the fetus and b) determine if metabolism-related
candidate genes or global DNA methylation is affected.

45

METHODS
Study Design
We used a nested exposure-control design for this study, and selected our
participants from an ongoing cohort study described in detail in the Study
Population section below. Three groups of women were selected: one group was
highly active, one was considerably less active and was matched to the high
active group on maternal age, race, and BMI, and one was also low active but
was matched to the high active group on maternal age and race only. We
matched infant blood spots taken at birth to the study subjects and used the
spots to compare total global methylation as well as candidate gene methylation
among the groups.
Study Population
The Michigan State University Department of Epidemiology & Biostatistics
has, in collaboration with the College of Human Medicine, been operating the
ARCH Study since 2007. Questionnaire data were collected on pregnant women
at a mean gestational age of 12.8 weeks, and a copy of the questionnaire is
included in Appendix I. Basic demographic data were collected at enrollment:
name, date of birth, social security number, race, ethnicity, education level,
marital status, household income, home and car ownership, ownership of
stocks/bonds, height, pre-pregnancy weight, and whether pregnancy was
planned.
ARCH participants are recruited from one of three mid-Michigan clinics
that provide prenatal health care to primarily low-income pregnant women. After

46

enrollment, the women and their offspring are followed longitudinally. We opted
to use only subjects enrolled since 2010 due to the need for infant blood spots
(described in more detail later) as our biologic sample. Blood spots maintained
by the State of Michigan have been stored frozen since 2010. In addition to
demographic information, we have collected prospective physical activity data on
more than 300 women. Study participants were queried regarding their
participation in moderate and vigorous LTPA. Those who responded in the
affirmative provided additional information including days per week and minutes
per day. The LTPA questionnaire completed by participants is included in the
Appendix I. Specifically, moderate physical activity was defined as an activity that
caused a small increase in breathing and heart rate, and vigorous activity was
any that caused a large increase in breathing and resulted in sweating.
On average, the most active subjects in the ARCH data set, excluding
extreme outliers, reported 1,056 minutes of moderate or vigorous physical
activity per week (~150 minutes per day). This represents roughly seven times
the minimum amount of LTPA recommended in the 2008 Physical Activity
Guidelines for Americans (Physical Activity Guidelines for Americans, 2008), and
suggests that these women are very dissimilar to low active participants, with
respect to physical activity behaviors. Women who reported participating in more
than 2,500 minutes per week (approximately 6 hours per day) of moderate or
vigorous LTPA were not included in this analysis due to our concerns about
reporting accuracy. We were unable to use all of the most active subjects (for

47

whom the average LTPA was greater than 1,000 minutes per week in the
analysis) due to consent issues and the need for subject matching.
Subject Matching
We contacted prospective subjects by phone to request verbal consent to
utilize infant blood spots for our study. The script for consent phone calls is
included in the Appendix II. Phone numbers were available for approximately 150
subjects who had completed the physical activity portion of the questionnaire. We
called all subjects and after excluding those who did not answer despite multiple
attempts, had disconnected phone lines, or who refused, we obtained consent
from a total of 59 subjects. From these, we selected the most active subjects as
the ‘active’ group and matched on maternal BMI (for BMI group only), age, and
race as closely as possible to each individual ‘active’ woman. We did not use
pre-defined criteria for matching due to the limited (n = 59) number of subjects
from whom we could choose. In order to ensure the largest matched sample
possible, we matched as closely as we could for each variable. For instance, we
matched a high active woman of age 30 and BMI of 25 to a control with as similar
an age as we could find who had a BMI within a range we deemed reasonably
close. The average difference for age between the high active and combined low
active group was 6.0 years (range was 0.1 to 14.4 years) and the average BMI
difference for the BMI matched low active was 5.4 (range was 0.0 to 17.3). The
average BMI difference for the unmatched low active group compared to the high
active group was 10.8 (range was 3.5 to 21.0). Some high active women were
matched to women with a higher BMI and some were matched to women with

48

lower BMI, and the same was true for age. In total, 42 subjects were included in
the study.
We selected the 14 most active subjects in the database for whom we
were able to identify a similar control and from whom we received consent to
participate in the study. Another consideration in choosing the 14 most active
women was how recently they participated in the ARCH study; we sought to
select women who participated in the study within the last three years in hopes of
being more successful at contacting them for verbal consent to release their
infant’s blood spot for analysis.
The 14 most physically active women (termed ‘high active’) were
individually matched on maternal age and race/ethnicity to 14 low active controls
(i.e. each active woman was matched to a single control similar in age and
race/ethnicity). Matching criteria were selected because these factors are known
to alter DNA methylation (Adkins, Krushkal, et al., 2011; Adkins, Thomas, et al.,
2011; Borghol et al., 2012; El-Maarri, et al., 2007; Gemma, et al., 2009; Michels,
Harris, & Barault, 2011). Other research indicates that maternal smoking is a
strong predictor of DNA methylation (Knopik, Maccani, Francazio, & McGeary,
2012) so we attempted to exclude women who indicated that they smoked during
pregnancy. Due to the difficulty of matching subjects, we did include a total of five
smokers, but the number of smokers is not different across the three comparison
groups. Previous research also suggests that maternal BMI influences DNA
methylation of candidate genes (Gemma, et al., 2009; Liu et al., 2014) but not
global methylation (Michels, et al., 2011) so we opted to select a third group of 14

49

subjects who were individually matched to the 14 high active women for age,
race/ethnicity, and BMI. Thus, our total sample included 42 subjects (3 groups x
14 subjects per group). The purpose of the BMI match group was to distinguish
the effects of physical activity, which has a known role in weight management,
from BMI itself. Once we calculated group means for BMI, however, we found
that the three groups had very similar BMI values (26.5 ± 6.0 for high active, 26.3
± 5.5 for low active BMI match, 26.9 ± 9.4 for low active), and therefore we opted
to analyze the data as high active (n = 14) compared to low active (n = 28), which
was the combination of the BMI matched and unmatched low active groups. In
addition, we know that other potential confounders are related to DNA
methylation. Due to limitations with the total sample size, we were unable to
match for baby’s birth weight, and due to known difficulty with dietary recall,
particularly several years after the time frame of interest, we did not assess or
match on mother’s diet during pregnancy despite evidence that these covariates
might influence DNA methylation (Haggarty, Hoad, Campbell, et al., 2013; Lan et
al., 2013; Michels, et al., 2011). DNA methylation also varies according to gender
(El-Maarri, et al., 2007), but because there is no evidence to suggest that
maternal physical activity influences the baby’s gender, we did not match on this
variable. Finally, we opted not to match on socioeconomic status (SES) despite
its relationship with DNA methylation (Borghol, et al., 2012) because the ARCH
cohort is fairly homogeneous in this variable, as subjects are recruited from
clinics that serve primarily low-income women. We did, however, analyze
demographic information between the two groups and found no statistically

50

significant differences in any of these variable including baby’s gender, mother’s
household income, mother’s education (trend toward statistical significance was
observed), prevalence of maternal smoking during pregnancy, mother’s marital
status, or birthweight (see Table 3).
Sample Size
While sample size is small, a power analysis for this proposal was not
computed because no human data of a similar nature have been published to
date. However, others have performed similar studies using small samples. For
example, Aagaard-Tillery et al. (Aagaard-Tillery et al., 2008) found that in age
and weight matched adult female Japanese macaques, a high fat diet resulted in
an alteration of the fetal chromatin structure. The authors compared 10
experimental primates with 9 controls, similar to our proposed sample size. Thus,
we believed that an exploratory analysis of 14 active and 28 controls would
provide pilot data so that future studies can use our resulting effect sizes to
design studies to answer more specific research questions.
Identification numbers of the 42 selected subjects (14 high active matched
with 28 low active controls) were matched to subject names by the study
investigator so that blood spots for their infants could be requested from the
Michigan Neonatal Biobank. The Biobank was able to retrieve all 42 of our
requested spots; some were stored at the Detroit location and others at the
Lansing location. Though samples that have been in storage since 2010 are
frozen, we received four, three mm punches from each blood spot that had been
thawed. Samples were mailed overnight from the Biobank. DNA was extracted

51

using the QIAamp DNA Micro Kit (Qiagen) according to manufacturer
recommendations. Upon extraction, DNA was stored at -20 degrees Celsius.
Laboratory Analysis
Aim 1
Our first aim was to determine whether there are differences in global
DNA methylation in the offspring of physically active compared to low active
pregnant women. After extraction, DNA was quantified via spectrophotometry
(NanoDrop). Next, we examined total 5-methyl cytosine content by using a
colorimetric ELISA-like assay called MethylFlash (Epigentek, Farmingdale,
NY),which assesses total content of 5-methyl cytosine in the genome using a
standard curve metric. The assay includes positive controls for dilution into a
standard curve as well as a negative control.
In addition to the MethylFlash assay, we sent isolated genomic DNA to a
commercial laboratory (EpigenDX, Worcester, MA) where they conducted
pyrosequencing analysis for LINE-1 methylation. We assayed four regions of
repetitive elements throughout the genome; methylation of these long regions of
repetitive DNA stretches are commonly used as indicators of methylation of the
entire genome (Haggarty, Hoad, Campbell, et al., 2013; Michels, et al., 2011; A.
J. White et al., 2013). Methylation of these repetitive elements is similar between
mother (peripheral blood) and her newborn (umbilical cord blood); thus we
believe that differences observed between offspring of physically active women
versus low active can be attributed to maternal exposures (Kile et al., 2010).

52

Aim 2
Our secondary aim was to evaluate gene-specific DNA methylation
differences in newborn dried blood spot samples of infants whose mothers
reported high LTPA levels during pregnancy compared to infants of low active
mothers, either matched or unmatched for BMI. We selected five candidate
genes that have been shown to be altered epigenetically by exercise/physical
activity or that, because of their function in metabolism, might likely be affected.
None had been studied previously in terms of maternal transmission of the
methylation pattern to the offspring.
PGC1-α and PPAR-γ are genes involved in metabolism of glucose and fat.
PDK4 gene expression is affected by physical activity, and IGF2 is an imprinted
gene that has been frequently associated with fetal growth. TCF7L2 is a gene
with a known role in etiology of Type 2 Diabetes. Table 1 shows the descriptions
of each candidate gene and the regions within these genes that were assessed.
These genes are described in greater detail in the literature review portion of this
dissertation (Chapter 2).
To assess methylation levels of selected candidate genes, we sent
isolated genomic DNA to EpigenDX (Worchester, MA), a commercial laboratory
specializing in methylation analyses. Upon receipt of DNA samples, EpigenDX
treated samples with bisulfite and used polymerase chain reaction (PCR) to
amplify the bisulfite treated DNA. Primers used for the PCR reactions are
considered proprietary information by EpigenDX and thus are unavailable for
publication. Cycling conditions are included in the Appendix III. Pyrosequencing

53

was then used to sequence the regions of candidate genes shown in Table 1;
these conditions are also presented in Appendix III. Each pyrosequencing assay
has been validated by EpigenDX by use of low, medium, and highly methylated
DNA as well as a no template control. Results from each analysis are a
percentage of methylated DNA at the region of interest.
It is important to note that to our knowledge, no studies have examined
how a mother’s LTPA might influence methylation of offspring cells in any tissue.
Thus, this dissertation research should be considered a pilot project designed to
generate hypotheses for future studies in the area of pregnancy physical activity
and influences on the offspring’s metabolic health.
Statistical Analysis
Descriptive characteristics of the sample are reported for the high active
and both low active groups in Table 2. These data are presented as means ±
standard deviations. Methylation data results are provided as % methylation of
the genome (for global assays in Aim 1) or a given gene (Aim 2). No covariates
were included in the analysis due to the matching design.
We checked for normality of LINE-1 data using the Kolmogorov-Smirnoff
test. The LINE-1 data were normally distributed and therefore analysis of
variance (ANOVA) was used to assess methylation differences between groups.
Finally, we calculated a Spearman correlation coefficient to assess agreement
between the LINE-1 data and minutes per week MVPA within the total sample
(n=42).

54

We checked all candidate gene data for normality using KolmogorovSmirnoff tests and found that none of the data met criteria for normal distributions
(p < 0.05). All data analyses, then, were performed using non-parametric tests.
We compared the differences in percent methylation between groups, for each
gene, using Mann Whitney U tests. Additionally we evaluated Spearman
correlations for each mean methylation value for each gene region to number of
minutes of MVPA per week within the total sample (n=42).
Table 1: Candidate Gene Descriptions. Description of each candidate gene
region assessed for percent methylation in this dissertation.
Gene
PPARγ
PPARγ
PGC1α
PGC1α
IGF2
IGF2
IGF2
PDK4
TCF7L2
TCF7L2

Description
Promoter
Intron 2
Proximal promoter
Exon 1 to intron 1
P2 promoter
P4 promoter
Distal promoter
Promoter
5’UTR
5’UTR to promoter

PCR Size
132 bp
281 bp
238 bp
174 bp
271 bp
130 bp
270 bp
149 bp
150 bp
150 bp

55

# CpGs
14
3
3
4
8
5
3
7
2
5

RESULTS
Sample
A total of 42 subjects were included in the analyses. Of these, 14 were
classified as ‘high active’ and reported 360 to 1,020 minutes per week of MVPA,
with mean participation in MVPA of 637.5 minutes per week. We planned
originally to use a three group design with 14 low active subjects matched for
BMI, race, and maternal age and another 14 low active subjects matched for
race and age only. However, there were no differences in BMI among the low
active groups matched and unmatched for BMI (see Table 2), so we opted to
group these 28 subjects into one group to increase statistical power. The
expanded low active group reported an average of 59.46 minutes per week
MVPA.
Table 3 shows the sample characteristics for the two groups. As expected,
MVPA was significantly higher in the high active group compared to either low
active group (p = 0.001). There were no differences in maternal BMI, maternal
age, or infant birth weight between groups. Also, we found no difference in the
percentage of white participants, male offspring, smoking prevalence, income,
marital status, or planned pregnancies between groups. While not statistically
significant (p = 0.060), the active group was more highly educated than the low
active group.

56

Table 2: Mean MVPA and BMI by the Three Originally Planned Groups.
Mean ± standard deviation plus ranges are presented for MVPA min/week and
means ± standard deviation are presented for Maternal BMI.
High Active
Low Active
Low Active
(n=14)
– BMI Match
(n=14)
(n=14)
MVPA/week
637.5 ±
80.7 ± 93.9
38.2 ± 60.6
220.7** (360
(0 – 285)
(0 – 180)
– 1020)
Maternal BMI
26.5 ± 6.0
25.8.3 ± 5.1
27.4 ± 9.5
Table 3: Descriptive Characteristics for Two Groups. Means ± standard
deviations for MVPA, age, BMI, and birth weight were compared between two
groups using analysis of variance (ANOVA). Mean percentages smoking were
compared using chi-square tests. For categorical variables, data are presented
as % (n). *indicates significant difference between high active group and low
active group
High
Low
P-value
Active
Active
(n=14)
(n=28)
MVPA per week
637.5 ±
59.5 ±
*0.000
220.7
79.1
Maternal age (years)
26.4 ±
27.4 ±
0.562
8.2
5.2
Maternal BMI
26.5 ±
26.6 ±
0.971
6.0
7.4
Infant Birth Weight (grams)
3247 ±
3409 ±
0.280
419
470
Maternal race (% white)
92.9 (13)
71.4 (20)
0.111
Sex (% male)
57.1 (8)
57.1 (16)
1.000
Smoking (% reported)
14.3 (2)
10.7 (3)
0.736
Income (% <$25,000)
57.1 (8)
39.3 (11)
0.273
Marital Status (%
24.1 (3)
21.4 (6)
1.000
unmarried)
Marital Status (%
35.7 (5)
39.3 (11)
0.822
married/living with baby’s
father)
Education (% attended at
85.7 (12)
64.3 (18)
0.060
least some college)
Planned Pregnancy (%
50.0 (7)
53.6 (15)
0.827
unplanned)

57

Aim 1
We used the MethylFlash assay to assess percent global methylation. We
successfully ran a practice assay with a standard curve equation of y = 17.03x –
0.958. The R-squared value for the line of best fit was 0.99. While this initial
standard curve was acceptable, we were unable to replicate a useful standard
curve on subsequent assays when samples were analyzed. Two attempts were
made to use the kit but were unsuccessful; we did not have enough isolated DNA
to attempt a third assay. Therefore, no data from the MethylFlash assay are
shown.
In addition to the conducting the MethylFlash assay, we also had
pyrosequencing for LINE-1 performed, in order to estimate global DNA
methylation. We used a Kolmogorov-Smirnoff test to determine that the LINE-1
data was normally distributed. Results from this assay are presented in Table 4.
There were no differences between groups for LINE-1 percent methylation. We
also assessed the relationship between MVPA minutes per week and LINE-1
percent methylation, but found no significant relationship based on Spearman
correlation coefficients (data not shown).
Table 4: LINE-1 Methylation. Percent methylation for LINE-1 at four different
CpG sites plus the overall mean methylation of LINE-1 for two groups. Data are
presented as mean ± standard deviation. Significance was set at an alpha level
of P<0.05.
LINE-1
CpG
High Active
Low Active
P-value
CpG 1
84.57 ± 2.97
85.06 ± 2.85
0.602
CpG 2
77.86 ± 1.23
78.21 ± 1.62
0.482
CpG 3
75.09 ± 2.24
75.66 ± 2.05
0.415
Cpg 4
71.46 ± 2.36
71.23 ± 1.96
0.738
Mean
77.25 ± 1.44
77.54 ± 1.58
0.558

58

Aim 2
Five candidate genes were analyzed for percent methylation at various
CpG sites. Table 5 shows data for PPARγ, a gene located on chromosome 3.
Two gene regions were analyzed; the first region, in the promoter, contains 14
CpG sites and the second region, in intron 2, contains three. We found no
differences in percent methylation at any site within the gene. Data for PGC1-α,
on chromosome 4, is presented in Table 6; two regions were examined via
pyrosequencing. One region, in the proximal promoter, has three CpG sites and
the other, covering an area ranging from exon one to intron one has four. We
found no differences in percent methylation in this gene.
Results for IGF2, on chromosome 11, are presented in Table 7. We
analyzed three regions of interest within the gene: the first region we examined is
in the P2 promoter and has eight CpGs, another is in the P4 promoter with five
CpGs, and the last is in the distal promoter with three CpGs studied. There were
two sites within the P2 promoter that were significantly different between the two
groups; for CpG 3 and 4, percent methylation was higher in the low active group
compared to high active. Although not significantly different, it is also worth
noting that we found the same directional effect (low active group had higher
methylation than high active group) for CpGs 1 and 4 through 6 for the P2
promoter for IGF2. We found no differences between physical activity groups for
PDK4, a gene located on chromosome seven (see Table 8). The analysis for this
gene was conducted on a region of the promoter that included seven CpG sites.

59

Table 9 shows methylation for TCF7L2. Seven total CpGs were studied and
there were no differences between groups at any site.
We calculated total percent methylation for each gene region (two regions
for PPARγ, three for IGF2, two for PGCA-α, one for PDK4, two for TCF7L2) in
order to compare average methylation between groups. Data are presented in
Table 10. The only statistically significant difference was in the P2 promoter for
IGF2. We found that the low active group had a significantly higher percent
methylation than the high active group (p = 0.045).
Correlations with Physical Activity
We calculated Spearman correlation coefficients to assess the relationship
between total MVPA per week and percent methylation of each gene region.
These data are shown in Table 11. No significant correlations were found.
Table 5: PPARγ Methylation. Percent methylation for PPARγ at each CpG site
(column 2) within each studied gene region (column 1). Data are presented in
mean ± standard deviation for each group. P-values are presented (alpha =
0.05).
PPARγ
CpG
High
Low
P-value
Active
Active
Promoter
CpG 1
0.72 ± 0.58
0.58 ± 0.57
0.320
Promoter
CpG 2
1.11 ± 0.60
0.87 ± 0.60
0.468
Promoter
CpG 3
0.79 ± 0.53
0.60 ± 0.63
0.143
Promoter
CpG 4
0.57 ± 0.65
0.54 ± 0.58
0.966
Promoter
CpG 5
0.53 ± 0.87
0.60 ± 0.70
0.506
Promoter
CpG 6
0.73 ± 0.85
0.82 ± 0.71
0.581
Promoter
CpG 7
1.02 ± 0.71
1.13 ± 0.70
0.809
Promoter
CpG 8
1.55 ± 0.24
1.41 ± 0.71
0.739
Promoter
CpG 9
0.64 ± 0.89
0.54 ± 0.68
0.976
Promoter
CpG 10
0.44 ± 0.47
0.30 ± 0.47
0.265
Promoter
CpG 11
0.33 ± 0.46
0.44 ± 0.58
0.561
Promoter
CpG 12
0.71 ± 0.77
0.74 ± 0.78
0.732
Promoter
CpG 13
0.46 ± 0.64
0.18 ± 0.46
0.118
Promoter
CpG 14
0.64 ± 0.96
0.25 ± 0.63
0.137
Intron 2
Cpg 1
75.47 ± 8.80
76.03 ± 7.77
0.749

60

Table 5 (cont’d).
Intron 2
Cpg 2
Intron 2
Cpg 3

62.13 ± 6.92
57.84 ± 12.66

62.87 ± 4.79
59.49 ± 3.60

0.423
0.423

Table 6: PGC1-α Methylation. Percent methylation for PGC-1α at each CpG
site (column 2) within each studied gene region (column 1). Data are presented
in mean ± standard deviation for each of the three groups. P-values are
presented (alpha = 0.05).
PGC1-α
CpG
High
Low
P-value
Active
Active
Proximal
CpG 1
0.69 ± 1.04
0.72 ± 1.20
0.861
promoter
Proximal
Cpg 2
0.59 ± 0.84
0.17 ± 0.44
0.069
promoter
Proximal
CpG 3
1.81 ± 1.77
2.09 ± 1.83
0.619
promoter
Exon 1 to
CpG 1
3.22 ± 1.65
2.79 ± 1.58
0.416
intron 1
Exon 1 to
Cpg 2
4.85 ± 1.52
5.26 ± 2.14
0.531
intron 1
Exon 1 to
CpG 3
7.26 ± 2.85
7.45 ± 2.71
0.622
intron 1
Exon 1 to
Cpg 4
7.50 ± 1.92
6.81 ± 1.84
0.364
intron 1
Table 7: IGF2 Methylation. Percent methylation for IGF2 at each CpG site
(column 2) within each studied gene region (column 1). Data are presented in
mean ± standard deviation for each group. P-values are presented (alpha =
0.05). *indicates significant difference between the two groups
IGF2
CpG
High
Low
P-value
Active
Active
P2 promoter
CpG 1
73.30 ± 4.84
74.66 ± 4.32
0.431
P2 promoter
CpG 2
87.22 ± 5.24
86.29 ± 4.35
0.393
P2 promoter
CpG 3
52.98 ±6.67
57.95 ± 5.61
*0.020
P2 promoter
CpG 4
69.86 ± 3.95
72.35± 2.89
*0.047
P2 promoter
CpG 5
80.78 ± 3.82
82.23 ± 2.63
0.182
P2 promoter
CpG 6
39.87 ± 3.76
41.68 ± 3.71
0.165
P2 promoter
CpG 7
98.19 ± 2.41
98.00 ± 2.23
0.581
P2 promoter
CpG 8
60.05 ± 2.23
60.77 ± 1.75
0.423
P4 promoter
Cpg1
1.42 ± 1.49
2.10 ± 4.97
0.540
P4 promoter
CpG 2
1.91 ± 5.24
1.86 ± 3.91
0.604
P4 promoter
CpG 3
0.30 ± 1.11
1.39 ± 3.67
0.137
P4 promoter
CpG 4
1.45 ± 2.47
1.43 ± 3.64
0.370
P4 promoter
CpG 5
0.21 ± 0.78
0.86 ± 2.61
0.477

61

Table 8: PDK4 Methylation. Percent methylation for PDK4 at each CpG site
(column 2) within each studied gene region (column 1). Data are presented in
mean ± standard deviation for each group. P-values are presented (alpha =
0.05).
PDK4
CpG
High
Low
PActive
Active
value
Promoter
CpG 1
1.94 ± 1.51
1.09 ± 1.12
0.086
Promoter
CpG 3
1.16 ± 1.24
1.02 ±1.37
0.668
Promoter
CpG 4
0.14 ± 0.54
0.16 ± 0.60
0.952
Promoter
CpG 5
1.02 ± 1.40
1.07 ± 1.32
0.906
Promoter
CpG 6
0.41 ± 0.82
0.85 ± 1.20
0.234
Promoter
CpG 7
1.16 ± 1.26
0.62 ± 1.10
0.181
Table 9: TCF7L2 Methylation. Percent methylation for TCF7L2 at each CpG
site (column 2) within each studied gene region (column 1). Data are presented
as mean ± standard deviation for each group. P-values are presented (alpha =
0.05).
TCF7L2
CpG
High
Low
P-value
Active
Active
5’UTR
CpG 1
0.00 ± 0.00
0.00 ± 0.00
1.000
5’UTR
CpG 2
0.08 ± 0.31
0.06 ± 0.31
0.638
5’UTR to
CpG 1
0.09 ± 0.34
0.00 ± 0.00
0.157
Promoter
5’UTR to
CpG 2
0.35 ± 0.70
0.08 ± 0.44
0.079
Promoter
5’UTR to
Cpg 3
0.00 ± 0.00
0.00 ± 0.00
1.000
Promoter
5’UTR to
CpG 4
0.12 ± 0.44
0.00 ± 0.00
0.157
Promoter
5’UTR to
CpG 5
0.00 ± 0.00
0.00 ± 0.00
1.000
Promoter
Table 10: Candidate Gene Mean Methylation. Percent methylation across all
gene regions within 5 candidate genes. Data are presented in mean ± standard
deviation for each region of the genes. P-values are presented (alpha = 0.05).
*indicates significantly difference between the two groups.
Gene
Gene
High
Low
PRegion
Active
Active
value
PPARγ
Promoter
0.73 ±
0.64 ±
0.463
0.35
0.31
Intron 2
65.15 ±
66.13 ±
0.729
7.29
3.94
IGF2
P2
72.83 ±
74.70 ±
*0.045
Promoter
2.85
2.25

62

Table 10 (cont’d).

PDK4

P4
Promoter
Distal
Promoter
Proximal
Promoter
Exon 1 to
Intron 1
Promoter

TCF7L2

5’UTR

PGC1α

5’UTR to
Promoter

1.06 ±
1.29
55.31 ±
9.12
1.03 ±
0.96
5.71 ±
1.12
0.85 ±
0.58
0.04 ±
0.16
0.11 ±
0.24

1.53 ±
2.43
52.48 ±
6.14
1.16 ±
0.94
5.61 ±
1.38
0.69 ±
0.61
0.03 ±
0.15
0.02 ±
0.09

0.724
0.483
0.947
0.683
0.445
0.638
0.063

Table 11: Correlation Coefficients for LTPA and Candidate Genes.
Spearman correlation coefficients between gene regions and MVPA per week. Pvalues are presented (alpha = 0.05).
Gene
Gene Region
Correlation
P-value
Coefficient
PPARγ
Promoter
0.089
0.574
Intron 2
-0.008
0.958
IGF2
P2 Promoter
-0.176
0.266
P4 Promoter
-0.057
0.720
Distal Promoter
0.165
0.302
PGC1α
Proximal Promoter
-0.060
0.706
Exon 1 to Intron 1
0.005
0.973
PDK4
Promoter
0.110
0.489
TCF7L2
5’UTR
0.058
0.713
5’UTR to Promoter
0.155
0.327

63

DISCUSSION
The Developmental Origins of Adult Disease Hypothesis, sometimes
termed the Barker Hypothesis, says that events and exposures that occur during
fetal development affect health later in life. Though the exact mechanisms are
not understood, there is ample evidence that this phenomenon does occur, at
least to some extent, in humans (Burdge & Lillycrop, 2010a). The stimuli that
elicit these changes that later affect disease risk are not clear. For instance, the
Dutch Famine Study provides strong evidence that maternal diet during
pregnancy, particularly caloric restriction, impacts offspring metabolic health (de
Rooij, et al., 2006; Heijmans, et al., 2008; Painter, et al., 2008). Yet the effect of
physical activity during pregnancy has been studied sparsely with respect to the
Barker Hypothesis. Therefore, we designed our study to assess the effects of
maternal LTPA during pregnancy on one known mechanism of fetal
programming, called DNA methylation. We examined methylation throughout the
genome (global methylation) as well as methylation of five candidate genes
known to either be impacted (though not necessarily in a pregnant population) by
physical activity, or with a known role in metabolism. This work is novel in that it
is the first to examine the effects of pregnancy physical activity on DNA
methylation and its potential transfer of methylation patterns to offspring in
humans.
Our results suggest that there is no effect of physical activity during
pregnancy on offspring global DNA methylation. To our knowledge, this is the
first study to examine this topic in humans. Others have examined the role of

64

chronic physical activity on global methylation in non-pregnant populations. White
and colleagues found that in adult women, higher physical activity levels were
associated with higher levels of LINE-1 methylation in peripheral blood samples
(A. J. White, et al., 2013). Sample size was n = 647 and the median physical
activity level in the study was 12.5 hours per week. The authors adjusted for age,
smoking, alcohol consumption, and BMI but none of the variables affected the
statistical model. Another group found the same positive relationship between
DNA methylation and physical activity in 165 subjects, but the relationship was
attenuated after adjustment for confounders including race, gender, and age (F.
F. Zhang, R. Cardarelli, J. Carroll, S. Zhang, et al., 2011). However, physical
activity levels even in the most active group were low, with the most active group
participating in no more than 30 minutes per day of physical activity. In Swedish
older adults (n = 1,016, all age 70 years), however, physical activity was
associated with lower levels of methylation (Luttropp, Nordfors, Ekström, & Lind,
2013). There was no adjustment for confounders. All three of these studies used
blood samples from the individuals actually participating in the physical activity,
whereas we examined the role of a mother’s LTPA on her offspring’s global
methylation. We found no other studies of global methylation being transferred to
offspring as a result of maternal LTPA, though the effects of adverse pregnancy
conditions including gestational diabetes and preeclampsia have been shown to
be related to lower placenta DNA methylation in offspring (Nomura et al., 2014).
It seems plausible that some stimuli may certainly impact offspring DNA
methylation but it is not clear from our one study, given the lack of other data on

65

the topic, whether this might occur with LTPA during pregnancy. Further research
is needed to clarify this question.
For our candidate gene analysis, we hypothesized that the active group
would have lower DNA methylation levels for PPARγ, PGC1-α, and PDK4while
the active group would have the highest DNA methylation levels for IGF2 and
TCF7L2. PPARγ and PGC1-α both have roles in metabolism and there is a
known relationship between high BMI and higher PGC1-α methylation (Gemma,
et al., 2009). PPARγ methylation is higher in diabetic versus non-diabetic twin
pairs (Ribel-Madsen, et al., 2012), suggesting that high PPARγ methylation is not
a ‘good’ thing. Methylation of PDK4 is also lower in diabetics compared to nondiabetics (Kulkarni, et al., 2012) and acute exercise resulted in an decrease in
PDK4 and PGC1-α methylation (Barrès, et al., 2012). For all three of these
genes, data suggest that high methylation represents a ‘bad’ phenotype.
However, for IGF2 and TCF7L2, the opposite appears to be true. Higher IGF2
methylation is associated with gene silencing, which would turn off production of
IGF2 protein, associated with growth. We found two studies examining the
relationship between birth weight and IGF2 methylation; one found no
relationship (Burris et al., 2013) and the other found that higher methylation was
related to higher weight, within the normal birth weight range (St-Pierre, et al.,
2012). Finally, TCF7L2 is a gene with a known relationship to Type 2 Diabetes.
A six-month exercise intervention resulted in an increase in methylation in five of
six studied sites within the gene (Rönn, et al., 2013). Thus it seems that for these
two genes, high methylation levels represent a ‘good’ phenotype.

66

Despite previous results that helped formulate our hypotheses, we found
few differences in DNA methylation among groups, and those we did find were
opposite of what was expected. We found no influence of physical activity for
methylation of PPARγ, PGC1-α, PDK4, or TCF7L2. There were two CpG sites in
the P2 promoter region of IGF2 that did have different methylation levels.
However, it was the low active group that had the higher DNA methylation levels,
which was the opposite of what we expected to find.
Similar to our findings, Gemma and colleagues found no differences in
PPARγ methylation among SGA, LGA and average for gestational age (AGA)
infants; in fact, the promoter was nearly completely methylated in all of their
samples (Gemma, et al., 2009). However, the investigators used umbilical cord
blood as their biologic sample rather than peripheral blood. In the same study,
infant DNA methylation of PGC1-α was also studied and was significantly
correlated to maternal BMI. We were not able to replicate this finding in the
present study.
Several exercise studies have been designed to examine PGC1-α
methylation in lab animals, but data were collected using maternal, not offspring
samples. In mice, maternal exercise during pregnancy prevented increased
PGC1-α methylation associated with high-fat diet (Laker et al., 2014) and
promoter DNA methylation decreased with a bout of acute exercise in a mouse
study (Barrès, et al., 2012). In humans, an exercise intervention resulted in
decreased PGC1-α methylation in skeletal muscle (Nitert et al., 2012) but not in
adipose tissue (Rönn, et al., 2013). In another skeletal muscle biopsy study, bed

67

rest resulted in increased PGC1-α methylation and decreased gene expression
(Alibegovic et al., 2010). Taken together, these data suggest that low levels of
PGC1-α methylation represent a ‘good’ phenotype with possible better metabolic
health. We did not see this effect in offspring of high active versus low active
women, which could be due to our choice of biologic sample (infant peripheral
blood taken at birth) or the lack of transfer of the methylation profile, at least for
this gene, from mother to child.
Unlike PPARγ and PGC1-α, we expected the offspring of the active group
to have the highest DNA methylation for IGF2. This is because IGF2 is involved
in growth regulation, and we know that active women tend to have smaller babies
(albeit within a healthy range) compared to sedentary women. High IGF2
methylation should, at least theoretically, result in less gene expression and
therefore less growth. A study of folate supplementation in pregnancy resulted in
higher methylation levels of IGF2 in offspring cord blood, suggesting that nutrition
and perhaps other maternal lifestyle factors are influencing this gene’s
methylation status in offspring (Haggarty, Hoad, Campbell, et al., 2013). St-Pierre
and colleagues found that placental DNA methylation was significantly correlated
with infant birth size, but there was no association with mother’s BMI (St-Pierre,
et al., 2012). In another study, IGF2 methylation was shown to be inversely
related to birth weight (p = 0.052) (Haggarty, Hoad, Horgan, & Campbell, 2013)
and therefore we suspected that there might be birth weight differences between
our groups. However, we found no significant difference in birth weights between
high active and low active women (see Table 3).

68

IGF2 is a maternally imprinted gene, meaning that it is regulated by a
process called imprinting, which involves silencing one parent copy of the gene
via methylation. Because of the known regulation of this gene via methylation
and its role in growth, we were surprised to find few effects of maternal physical
activity on IGF2 methylation in offspring. But, we know of no studies that have
examined the relationship between IGF2 methylation and exercise/physical
activity, so it is not possible to directly compare our results to those of others.
However, Hopkins et al. found that exercise training in pregnancy was not related
to maternal serum IGF1 or IGF2 concentrations, which may mean that exercise
is an insufficient stimulus to alter this particular gene’s expression (Hopkins,
Baldi, Cutfield, McCowan, & Hofman, 2011). Also, we found an effect of LTPA at
only one of the promoters we studied (P2 promoter). Therefore, it may be that
there is not a large enough effect for a net change in IGF2 hormone to be
measured identified in serum. This may explain why the Hopkins et al. study,
which assessed serum levels rather than gene expression or methylation found
no effect of the physical activity intervention. Because the authors did not directly
assess methylation or gene expression, it is difficult to know whether their
findings should be compared directly to ours.
For PDK4, we expected the high active group to have lower DNA
methylation compared to the low active group. Other data have shown lower
levels of PDK4 methylation in Type 2 Diabetes patients, and higher levels of
PDK4 gene expression in individuals with higher compared to lower BMI
(Kulkarni, et al., 2012). Acute exercise, in mice and in humans (skeletal muscle),

69

resulted in decreased PDK4 methylation (Barrès, et al., 2012), so we expected
chronic LTPA to do the same, even in offspring blood. Still, we found no
relationship between physical activity and PDK4 methylation. We also found no
effect of physical activity on TCF7L2 methylation. This was unexpected because
a six month exercise intervention study found increased DNA methylation in
TCF7L2 in adipose tissue in humans (Rönn, et al., 2013), but like for the other
candidate genes, it may be that tissue specificity, maternal transmission, or
amount of exercise stimulus impacted our findings.
The general lack of differences among the activity groups may be due to
one or a combination of several reasons. First, we chose to study methylation
rather than gene expression, which is, at least in theory, the effect of the
methylation profile. There is not always direct correlation between methylation
and gene expression, however. The location of the methylation within the
genome likely influences the impact that the methylation change has on gene
expression (Jones, 1999). For instance, methylation within certain gene areas
including promoters of imprinted genes and DNA repair genes does seem to
permanently silence promoter activity (resulting in a positive relationship between
methylation level and expression) while methylation at other gene regions does
not necessarily equate to less gene expression (Jones, 1999). It is plausible that
our results might have been different had we measured mRNA levels (indicative
of gene transcription) rather than the methylation at particular CpG sites. To our
knowledge, few others have examined both gene expression and DNA
methylation simultaneously. Barres and colleagues examined DNA methylation

70

of PGC1-α, PDK4, and PPAR-δ, in skeletal muscle samples, while
simultaneously examining gene expression, and they found an inverse
relationship between methylation and expression for each of these three genes
(Barrès, et al., 2012). However, in the same study, the authors examined the
effects of a 3-week exercise training intervention on PGC1-α and TFAM
methylation and found no effect on DNA methylation but increased gene
expression for both genes. This suggests that there is a timing effect in terms of
the physical activity (i.e. acute versus chronic). Possibly we would have found
different results for gene expression than we did for methylation since we
examined habitual LTPA rather than an acute bout of exercise.
Another possible explanation for lack of differences among groups is that
the methylation profile may not transfer from mother to child, at least in regard to
physical activity. We found no studies showing heritability of methylation patterns
related to maternal physical activity, even in animal models. Based upon the
Fetal Origins of Adult Disease Hypothesis, however, it would be expected that
such a stimulus might impact offspring health risk by transfer of methylation
patterns. Other lifestyle-related stimuli have been shown to induce methylation
changes that are passed to offspring. These include smoking (after 18 weeks
gestation) (Joubert et al., 2014), mother’s BMI (Liu, et al., 2014), and folate
intake (Haggarty, Hoad, Campbell, et al., 2013; Steegers-Theunissen et al.,
2009). Future research should involve a direct comparison of mother to offspring
methylation patterns to assess similarity due to environment and/or transferred

71

epigenetic profile. If high active mothers are more comparable to offspring than
are low active mothers, transfer of epigenetic markings can be assumed.
Yet another possibility is that the physical activity stimulus in this study
was insufficient to elicit change in DNA methylation in the mother. However, the
average woman in our high active group reported over 600 minutes per week of
MVPA, and of those subjects, eight of 14 reported significant amounts of
vigorous physical activity (i.e. more than one day per week).The total exercise
stimulus in our study is considerably more than that used in Zhang and
colleagues study; after controlling for confounders they found no relationship
between global methylation and physical activity study when comparing
individuals who exercised 26 to 30 minutes per day to those who were active less
than ten minutes per day (F. F. Zhang, R. Cardarelli, J. Carroll, S. Zhang, et al.,
2011). In the global methylation and physical activity study that did find a
relationship between higher activity and more global methylation, the physical
activity stimulus used in our study would be equivalent to their second lowest
quartile of activity (A. J. White, et al., 2013); the highest active group reported
more than 18 hours per week of activity. It may be, then that the total amount of
activity was inadequate, or more likely, that the intensity was not great enough to
elicit changes in methylation profiles. Barres and colleagues found that higher
intensity (80% VO2peak) acute exercise resulted in greater methylation changes
than did lower intensity (40% VO2peak) exercise (Barrès, et al., 2012).
It is also worth noting that the low active groups in our study reported an
average of approximately 60 minutes MVPA per week. While neither group met

72

the recommendations for physical activity from the Physical Activity Guidelines
for Americans (health.gov, 2008), they did report doing some activity. Ideally we
would have compared the highly active group to two completely sedentary
groups, who reported no physical activity whatsoever. Due to the need to match
on other variables, however, this was not possible.
Another issue inherent to this type of study is recall bias. Our questions
regarding physical activity asked about activity in the past month. It is possible
that subjects did not properly recall the amount of LTPA they participated in, and
such problems would affect study findings. In a study using the same ARCH
sample, Schlaff and colleagues found no relationship between pregnancy LTPA
and pre-pregnancy BMI or between LTPA and risk for gaining excess weight
during pregnancy. Rather, BMI prior to pregnancy predicted gestational weight
gain (Schlaff, Holzman, Maier, Pfeiffer, & Pivarnik, 2014). This suggests a
possible combination of a recall bias issue and insufficient LTPA stimulus that
may be leading toward our mostly null findings.
Another possible reason for our findings is that we have used a different
biologic sample (blood spots) for analysis compared to most of the similar
literature. Most research on methylation changes passed from one generation to
the next has been performed on cord blood (Gemma, et al., 2009; Haggarty,
Hoad, Campbell, et al., 2013; Joubert, et al., 2014) while another study used
placenta (St-Pierre, et al., 2012). The majority of literature relating exercise or
physical activity to DNA methylation has used either fat (Rönn, et al., 2013) or
muscle biopsy (Barrès, et al., 2012; Kulkarni, et al., 2012; Nitert, et al., 2012;

73

Ribel-Madsen, et al., 2012; Wang, Psilander, Tonkonogi, Ding, & Sahlin, 2009)
as the biologic sample. One research group examined methylation changes in a
large number of genes following a six month exercise intervention. About 75
percent of the genes affected by exercise in skeletal muscle had decreased DNA
methylation while most of the genes that were affected in adipose tissue had
increased methylation (Nitert, et al., 2012; Rönn, et al., 2013), suggesting that
different tissues are affected in unique ways by physical activity. It is less clear
whether use of peripheral blood (blood spots) from birth compared to umbilical
cord blood could influence results, and to what extent. We found no studies
directly comparing methylation patterns of these two types of blood samples,
though gene expression was highly correlated (r = 0.941) for cord blood and
blood spots for three sets of genes (Slaughter et al., 2013). Due to known
variation in DNA methylation patterns by tissue type, it is reasonable to suspect
that cell type selection might lead to at least some differential findings across
studies (more so for muscle, fat, and placenta rather than cord blood), and thus it
is difficult to determine whether it is our sample selection or other factors that
explain our mostly null findings.
Future Directions
There are several limitations to this study that should be addressed in
future research. First, sample size was small, as we considered this to be pilot
work. It is also important to compare methylation levels of genes in mother and
her offspring in order to address the issue of transferability of these methylation
profiles from one generation to another. Simply taking a maternal blood sample

74

at delivery (or shortly after, around the time of the infant’s heel stick) would
provide for this opportunity. It would be interesting to compare the DNA
methylation levels of cord blood to blood spots to determine whether or not the
biologic sample being used for the infant is influencing results. It is not likely that
skeletal muscle biopsies would be plausible in infants or children due to ethical
reasons.
Other future directions could include comparisons of high active, moderate
active, and sedentary groups. It is possible that we see no differences in high
active and low active groups because both ends of the activity spectrum result in
a ‘negative’ epigenetic profile while the true effect lies with those who are
moderately active. However, White et al. used physical activity quartiles and did
not find this effect (A. J. White, et al., 2013). Still, similar u-shaped relationships
are seen with exercise and immunity (Gleeson, 2007) and with exercise and
telomere length (Ludlow et al., 2008), where moderate activity/exercise results in
the ‘good’ phenotype and the low active and high activity groups have the ‘bad’
outcome. Finally, it would be useful to examine a greater selection of genes,
possibly by use of a microarray or similar technology, to determine whether these
candidate genes we selected might actually be the wrong place to look.
While there are important limitations to the study, there are also several
strengths that should be acknowledged. This study is novel and is the first to
examine the effects of pregnancy physical activity on offspring DNA methylation
in humans. Though we found no differences in global methylation levels in
offspring of active compared to sedentary women, we have been able to highlight

75

important issues to address in future research in this area. The groups were
relatively well matched for known confounders, so future work can emphasize
selection of appropriate tissue type for analysis, avoidance of recall bias, and
adequate physical activity stimulus to further understand this issue and
whether/how pregnancy LTPA might influence offspring susceptibility to noncommunicable disease.

76

APPENDICES

77

APPENDIX A

Questionnaires

78

Figure 1: ARCH Questionnaire. Questionnaire is completed by subjects upon
enrollment.

79

Figure 1 (cont’d).

80

APPENDIX B

Oral Consent Script

81

ARCH Sub-Study: Effects of Maternal Physical Activity on Methylation Patterns
in Offspring Blood Spots
PI: Jim Pivarnik and Mallory Marshall
Date:
Investigator (Caller) Name:
Subject Name:
Oral Consent Script:
Hello, my name is _____________, a student from MSU; may I speak with
(subject’s first name) please?
If she answers:
Hello, my name is _________ and I am calling to follow up with you because you
participated in the research studyArchive for Research on Child Health at MSU
beginning in ____ (year) when you were pregnant with __________ (child’s first
name). May I ask you a question relating to your participation in the study?
If no: Thank you for your time. Goodbye.
If yes:
Researchers at MSU would like to use health information that you have already
provided to us and request the Michigan Department of Community Health link it
with your baby’s blood spot sample that was taken from your baby’s heel for
newborn screening when ___ (he/she) was born at Sparrow. This is for a
research study at MSU. The Department of Community Health oversees storage
of remaining blood spots from newborn screening for potential future personal
use or use in approved research. If you agree, we will request a small sample of
your baby’s blood spot, and we will use it for some genetic testing. The genetic
testing will not tell us about any diseases or conditions your child might have.
Very simply, the genetic testing will help us see whether certain genes which are
related to the way the body uses energy are active or inactive (on or off). Your
child’s identity will be kept private; we will not put your child’s name on any blood
samples or other materials, and we will store the samples in a locked box in a
laboratory where only study investigators can access the samples. Because no
subjects will be identified, research results cannot be returned but a summary of
the findings will be posted on-line at www.michigan.gov/biotrust when available.
Would you agree to allow us to use your child’s blood spot sample for this
research? Please note that participation is voluntary and you can withdraw your
consent at any time up until the research is completed, which we expect will be
approximately April 2014. If you do agree, we will document that you consented
over the phone, and you don’t need to complete any other paperwork. However,
if you would like a written copy of any materials including the original consent
form you signed when you enrolled in the study or the information just provided
to you via phone, I would be happy to mail that to you. If you have any questions

82

you may ask me or contact the study investigator, Mallory Marshall, at (517) 3557696 or email mrm@msu.edu.
______ Yes, my child’s blood spot can be used.
______ No, my child’s blood spot may not be used.
Thank you for your time! Goodbye.

83

APPENDIX C

PCR and Pyrosequencing Conditions

84

PCR Cycling Conditions
PPARγ
Table 12: PPARγ Promoter PCR Conditions
Component
Per 30µ
µl reaction
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
5X Q-Solution
3 µl (0.5x)
1.8 µl (3.0 mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6 µl (6 pmol)
(ADS2016FPre)
10 uM Rev primer
0.6 ul (6 pmol)
(ADS2016RPB)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1 µl of bisulfite treated
DNA
DNA
Water
Adjust to 30 µl
Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 53⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
Table 13: PPARγ Intron 2 PCR Conditions
Component
Per 30µ
µl reaction
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8 µl (3.0 mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6 µl (6 pmol)
(ADS2020FPB)
10 uM Rev primer
0.6 ul (6 pmol)
(ADS2020RP)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1 µl of bisulfite treated
DNA
DNA
Water
Adjust to 30 µl

85

Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 57⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
PGC1α
Table 14: PGC1α Proximal Promoter PCR Conditions
Component
Per 30µ
µl reaction
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8µl (3mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6µl (6pmol)
0.6 ul (6pmol)
10µM Rev primer
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl

Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 53⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
Table 15: PGC1α Exon 1 to Intron 1 PCR Conditions
Component
Per 30µ
µl reaction
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8µl (3mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6µl (6pmol)
0.6 ul (6pmol)
10µM Rev primer
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl

Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 56⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
86

IGF2
Table 16: IGF2 P2 Promoter PCR Conditions
Component
µl reaction
Per 30µ
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8 µl (3.0 mM
25 mM MgCl2
final conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6 µl (6 pmol)
(ADS3602FP)
10 uM Rev primer
0.6 ul (6 pmol)
(ADS3602RPB)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1 µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl
Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 65⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
Table 17: IGF2 P4 Promoter PCR Conditions
Component
µl reaction
Per 30µ
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
0 µl (1.5 mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.3 µl (3 pmol)
(ADS173FP)
2.5 µM Rev primer
0.12 ul (0.3 pmol)
(ADS173RPT)
10 uMUniversal RPB
0.27 ul (2.7pmol)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl
Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 56⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞

87

Table 18: IGF2 Distal Promoter PCR Conditions
Component
µl reaction
Per 30µ
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8 µl (3.0 mM
25 mM MgCl2
final conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6 µl (6 pmol)
(ADS1051FPB)
10 uM Rev primer
0.6 ul (6 pmol)
(ADS1051RP)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1 µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl
Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 60⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞
PDK4
Table 19: PDK4 Promoter PCR Conditions
Component
Per 30µ
µl reaction
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8µl (3mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6µl (6pmol)
0.6 ul (6pmol)
2.5 µM Rev primer
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl

Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 53⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞

88

TCF7L2
Table 20: TCF7L2 5’UTR and 5’UTR-Promoter PCR Conditions
Component
µl reaction
Per 30µ
10X PCR buffer
3 µl (1x)
(Contains 15mM MgCl2)
1.8µl (3mM final
25 mM MgCl2
conc.)
0.6 µl (200 µM of
10 mMdNTPs
each)
10 µM Fwd primer
0.6µl (6pmol)
(ADS173FP)
2.5 µM Rev primer
0.6 ul (6pmol)
(ADS173RPT)
HotStar Taq Polymerase
0.15 µl (0.75 U)
(5 U/µl)
1µl of bisulfite
DNA
treated DNA
Water
Adjust to 30 µl

Cycling Conditions: 95⁰C 15 min; 45 x (95⁰C 30sec; 53⁰C 30sec; 72⁰C
30sec); 72⁰C 5 min; 4⁰C ∞

Pyrosequencing Protocol
Pyrosequencing is performed using PSQ 96HS system or PSQ 96HSA system.
1. Prepare master Binding Solution: Components for 1 reaction of binding
solution: 2.2 uL streptavidin, 40 uL 2X Binding buffer, 25 uLMilli-Q-water
2. Add 65 uL binding solution to each 15 uL PCR sample.
3. Capture the PCR product as manufactory’s protocol
4. Release the sepharose beads into annealing buffer containing 0.5 uM of a
sequencing primer
5. Anneal the sequencing primer to the template by heating the plate to 85 ºC
for 2 minutes
6. Turn off the heating block and leave the Pyro plate on the heating block for
10 minutes
89

7. Remove the Pyro plate from the heating block and allow the plate continuing
to cool for 5 minutes
8. Run the Pyrosequencing as the manufactory’s instruction.

90

REFERENCES

91

REFERENCES

Aagaard-Tillery, K. M., Grove, K., Bishop, J., Ke, X., Fu, Q., McKnight, R., et al.
(2008). Developmental origins of disease and determinants of chromatin
structure: maternal diet modifies the primate fetal epigenome. J Mol
Endocrinol, 41(2), 91-102.
ACOG. (1985). Technical Bulletin: Exercise during pregnancy and the postnatal
period. Washington DC: American College of Obstetricians and
Gynecologists.
ACOG. (2002). ACOG committee opinion. Exercise during pregnancy and the
postpartum period.: American College of Obstetricians and Gynecologists.
ACSM. (2006). ACSM Roundtable Consensus Statement: Impact of physical
activity during pregnancy and the postpartum period on chronic disease
risk. Med Sci Sports Exerc, 38(5), 989-1006.
Adkins, R. M., Krushkal, J., Tylavsky, F. A., & Thomas, F. (2011). Racial
differences in gene-specific DNA methylation levels are present at birth.
Birth Defects Res A Clin Mol Teratol, 91(8), 728-736.
Adkins, R. M., Thomas, F., Tylavsky, F. A., & Krushkal, J. (2011). Parental ages
and levels of DNA methylation in the newborn are correlated. BMC Med
Genet, 12, 47.
Ainsworth, B. E., Haskell, W. L., Herrmann, S. D., Meckes, N., Bassett, D. R.,
Tudor-Locke, C., et al. (2011). 2011 Compendium of Physical Activities: a
second update of codes and MET values. Med Sci Sports Exerc, 43(8),
1575-1581.
Alderman, B. W., Zhao, H., Holt, V. L., Watts, D. H., & Beresford, S. A. (1998).
Maternal physical activity in pregnancy and infant size for gestational age.
Ann Epidemiol, 8(8), 513-519.
Alegría-Torres, J. A., Baccarelli, A., & Bollati, V. (2011). Epigenetics and lifestyle.
Epigenomics, 3(3), 267-277.
Alibegovic, A. C., Sonne, M. P., Højbjerre, L., Bork-Jensen, J., Jacobsen, S.,
Nilsson, E., et al. (2010). Insulin resistance induced by physical inactivity
is associated with multiple transcriptional changes in skeletal muscle in
young men. Am J Physiol Endocrinol Metab, 299(5), E752-763.

92

Barakat, R., Cordero, Y., Coteron, J., Luaces, M., & Montejo, R. (2011). Exercise
during pregnancy improves maternal glucose screen at 24-28 weeks: a
randomised controlled trial. Br J Sports Med 46(9), 656-61.
Barakat, R., Lucia, A., & Ruiz, J. R. (2009). Resistance exercise training during
pregnancy and newborn's birth size: a randomised controlled trial. Int J
Obes (Lond), 33(9), 1048-1057.
Barakat, R., Ruiz, J. R., Stirling, J. R., Zakynthinaki, M., & Lucia, A. (2009). Type
of delivery is not affected by light resistance and toning exercise training
during pregnancy: a randomized controlled trial. Am J Obstet Gynecol,
201(6), 590.e591-596.
Barakat, R., Stirling, J. R., & Lucia, A. (2008). Does exercise training during
pregnancy affect gestational age? A randomised controlled trial. Br J
Sports Med, 42(8), 674-678.
Barker, D. J. (1992). The fetal origins of adult hypertension. J Hypertens Suppl,
10(7), S39-44.
Barker, D. J., Bull, A. R., Osmond, C., & Simmonds, S. J. (1990). Fetal and
placental size and risk of hypertension in adult life. BMJ, 301(6746), 259262.
Barker, D. J., & Osmond, C. (1986). Infant mortality, childhood nutrition, and
ischaemic heart disease in England and Wales. Lancet, 1(8489), 10771081.
Barker, D. J., Osmond, C., Forsen, T. J., Kajantie, E., & Eriksson, J. G. (2005).
Trajectories of growth among children who have coronary events as
adults. [Research Support, Non-U.S. Gov't]. The New England journal of
medicine, 353(17), 1802-1809.
Barrès, R., Yan, J., Egan, B., Treebak, J. T., Rasmussen, M., Fritz, T., et al.
(2012). Acute exercise remodels promoter methylation in human skeletal
muscle. Cell Metab, 15(3), 405-411.
Berenson, G. S., Srinivasan, S. R., Bao, W., Newman, W. P., Tracy, R. E., &
Wattigney, W. A. (1998). Association between multiple cardiovascular risk
factors and atherosclerosis in children and young adults. The Bogalusa
Heart Study. N Engl J Med, 338(23), 1650-1656.
Boney, C. M., Verma, A., Tucker, R., & Vohr, B. R. (2005). Metabolic syndrome
in childhood: association with birth weight, maternal obesity, and
gestational diabetes mellitus. Pediatrics, 115(3), e290-296.

93

Borghol, N., Suderman, M., McArdle, W., Racine, A., Hallett, M., Pembrey, M., et
al. (2012). Associations with early-life socio-economic position in adult
DNA methylation. Int J Epidemiol, 41(1), 62-74.
Bovbjerg, M. L., & Siega-Riz, A. M. (2009). Exercise during pregnancy and
cesarean delivery: North Carolina PRAMS, 2004-2005. Birth, 36(3), 200207.
Bungum, T. J., Peaslee, D. L., Jackson, A. W., & Perez, M. A. (2000). Exercise
during pregnancy and type of delivery in nulliparae. J Obstet Gynecol
Neonatal Nurs, 29(3), 258-264.
Burdge, G. C., & Lillycrop, K. A. (2010a). Nutrition, epigenetics, and
developmental plasticity: implications for understanding human disease.
Annu Rev Nutr, 30, 315-339.
Burdge, G. C., & Lillycrop, K. A. (2010b). Nutrition, epigenetics, and
developmental plasticity: implications for understanding human disease.
Annu Rev Nutr, 30, 315-339.
Burdge, G. C., Slater-Jefferies, J., Torrens, C., Phillips, E. S., Hanson, M. A., &
Lillycrop, K. A. (2007). Dietary protein restriction of pregnant rats in the F0
generation induces altered methylation of hepatic gene promoters in the
adult male offspring in the F1 and F2 generations. Br J Nutr, 97(3), 435439.
Burris, H. H., Braun, J. M., Byun, H. M., Tarantini, L., Mercado, A., Wright, R. J.,
et al. (2013). Association between birth weight and DNA methylation of
IGF2, glucocorticoid receptor and repetitive elements LINE-1 and Alu.
Epigenomics, 5(3), 271-281.
Butte, N. F. (2000). Carbohydrate and lipid metabolism in pregnancy: normal
compared with gestational diabetes mellitus. Am J Clin Nutr, 71(5 Suppl),
1256S-1261S.
Cardwell, C. R., Carson, D. J., & Patterson, C. C. (2005). Parental age at
delivery, birth order, birth weight and gestational age are associated with
the risk of childhood Type 1 diabetes: a UK regional retrospective cohort
study. Diabet Med, 22(2), 200-206.
Choi, S. W., Corrocher, R., & Friso, S. (2009). Nutrients and DNA Methylation,
Nutrients and Epigenetics. Boca Raton, FL: CRC Press Taylor and
Francis Group.
Clapp, J. F. (1989). The effects of maternal exercise on early pregnancy
outcome. Am J Obstet Gynecol, 161(6 Pt 1), 1453-1457.
94

Clapp, J. F. (1996). Pregnancy outcome: physical activities inside versus outside
the workplace. Semin Perinatol, 20(1), 70-76.
Clarke, P. E., & Gross, H. (2004). Women's behaviour, beliefs and information
sources about physical exercise in pregnancy. Midwifery, 20(2), 133-141.
Dayeh, T. A., Olsson, A. H., Volkov, P., Almgren, P., Rönn, T., & Ling, C. (2013).
Identification of CpG-SNPs associated with type 2 diabetes and differential
DNA methylation in human pancreatic islets. Diabetologia, 56(5), 10361046.
de Rooij, S. R., Painter, R. C., Phillips, D. I., Osmond, C., Michels, R. P.,
Godsland, I. F., et al. (2006). Impaired insulin secretion after prenatal
exposure to the Dutch famine. Diabetes Care, 29(8), 1897-1901.
Dietz, W. H. (1998). Childhood weight affects adult morbidity and mortality.
[Comparative Study Research Support, U.S. Gov't, P.H.S. Review]. The
Journal of nutrition, 128(2 Suppl), 411S-414S.
DiNallo, J. M., Downs, D. S., & Le Masurier, G. (2012). Objectively assessing
treadmill walking during the second and third pregnancy trimesters. J Phys
Act Health, 9(1), 21-28.
Dishman, R. K., Heath, G. W., & Washburn, R. (2004). Physical Activity
Epidemiology (1 ed.). Champaign IL: Human Kinetics.
Domingues, M. R., Barros, A. J., & Matijasevich, A. (2008). Leisure time physical
activity during pregnancy and preterm birth in Brazil. Int J Gynaecol
Obstet, 103(1), 9-15.
Dominguez-Salas, P., Cox, S. E., Prentice, A. M., Hennig, B. J., & Moore, S. E.
(2012). Maternal nutritional status, C(1) metabolism and offspring DNA
methylation: a review of current evidence in human subjects. Proc Nutr
Soc, 71(1), 154-165.
Drake, A. J., Liu, L., Kerrigan, D., Meehan, R. R., & Seckl, J. R. (2011).
Multigenerational programming in the glucocorticoid programmed rat is
associated with generation-specific and parent of origin effects.
Epigenetics, 6(11), 1334-1343.
Duncombe, D., Wertheim, E. H., Skouteris, H., Paxton, S. J., & Kelly, L. (2009).
Factors related to exercise over the course of pregnancy including
women's beliefs about the safety of exercise during pregnancy. Midwifery,
25(4), 430-438.

95

Dunn, G. A., & Bale, T. L. (2009). Maternal high-fat diet promotes body length
increases and insulin insensitivity in second-generation mice.
Endocrinology, 150(11), 4999-5009.
El-Maarri, O., Becker, T., Junen, J., Manzoor, S. S., Diaz-Lacava, A., Schwaab,
R., et al. (2007). Gender specific differences in levels of DNA methylation
at selected loci from human total blood: a tendency toward higher
methylation levels in males. Hum Genet, 122(5), 505-514.
Ferrara, A., Kahn, H. S., Quesenberry, C. P., Riley, C., & Hedderson, M. M.
(2004). An increase in the incidence of gestational diabetes mellitus:
Northern California, 1991-2000. Obstet Gynecol, 103(3), 526-533.
Ferrara, A., Weiss, N. S., Hedderson, M. M., Quesenberry, C. P., Selby, J. V.,
Ergas, I. J., et al. (2007). Pregnancy plasma glucose levels exceeding the
American Diabetes Association thresholds, but below the National
Diabetes Data Group thresholds for gestational diabetes mellitus, are
related to the risk of neonatal macrosomia, hypoglycaemia and
hyperbilirubinaemia. Diabetologia, 50(2), 298-306.
Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2010). Prevalence and
trends in obesity among US adults, 1999-2008. JAMA : the journal of the
American Medical Association, 303(3), 235-241.
Fraga, M. F., Ballestar, E., Paz, M. F., Ropero, S., Setien, F., Ballestar, M. L., et
al. (2005). Epigenetic differences arise during the lifetime of monozygotic
twins. Proc Natl Acad Sci U S A, 102(30), 10604-10609.
Fraser, A., Tilling, K., Macdonald-Wallis, C., Sattar, N., Brion, M. J., Benfield, L.,
et al. (2010). Association of maternal weight gain in pregnancy with
offspring obesity and metabolic and vascular traits in childhood.
[Comparative Study Research Support, N.I.H., Extramural Research
Support, Non-U.S. Gov't]. Circulation, 121(23), 2557-2564.
Gabory, A., Attig, L., & Junien, C. (2011). Developmental programming and
epigenetics. Am J Clin Nutr, 94(6 Suppl), 1943S-1952S.
Gale, C. R., Javaid, M. K., Robinson, S. M., Law, C. M., Godfrey, K. M., &
Cooper, C. (2007). Maternal size in pregnancy and body composition in
children. [Research Support, Non-U.S. Gov't]. The Journal of clinical
endocrinology and metabolism, 92(10), 3904-3911.
Gemma, C., Sookoian, S., Alvariñas, J., García, S. I., Quintana, L., Kanevsky, D.,
et al. (2009). Maternal pregestational BMI is associated with methylation
of the PPARGC1A promoter in newborns. Obesity (Silver Spring), 17(5),
1032-1039.
96

George, L. A., Uthlaut, A. B., Long, N. M., Zhang, L., Ma, Y., Smith, D. T., et al.
(2010). Different levels of overnutrition and weight gain during pregnancy
have differential effects on fetal growth and organ development.
[Evaluation Studies Research Support, N.I.H., Extramural]. Reproductive
biology and endocrinology : RB&E, 8, 75.
Gervin, K., Vigeland, M. D., Mattingsdal, M., Hammerø, M., Nygård, H., Olsen, A.
O., et al. (2012). DNA methylation and gene expression changes in
monozygotic twins discordant for psoriasis: identification of epigenetically
dysregulated genes. PLoS Genet, 8(1), e1002454.
Gesteiro, E., Bastida, S., & Sánchez Muniz, F. J. (2011). Effects of maternal
glucose tolerance, pregnancy diet quality and neonatal insulinemia upon
insulin resistance/sensitivity biomarkers in normoweight neonates. Nutr
Hosp, 26(6), 1447-1455.
Gesteiro, E., Rodríguez Bernal, B., Bastida, S., & Sánchez-Muniz, F. J. (2012).
Maternal diets with low healthy eating index or mediterranean diet
adherence scores are associated with high cord-blood insulin levels and
insulin resistance markers at birth. Eur J Clin Nutr 66(9), 1008-15.
Gleeson, M. (2007). Immune function in sport and exercise. J Appl Physiol
(1985), 103(2), 693-699.
Godfrey, K. M., Sheppard, A., Gluckman, P. D., Lillycrop, K. A., Burdge, G. C.,
McLean, C., et al. (2011). Epigenetic gene promoter methylation at birth is
associated with child's later adiposity. Diabetes, 60(5), 1528-1534.
Groom, A., Potter, C., Swan, D. C., Fatemifar, G., Evans, D. M., Ring, S. M., et
al. (2012). Postnatal growth and DNA methylation are associated with
differential gene expression of the TACSTD2 gene and childhood fat
mass. Diabetes, 61(2), 391-400.
Haggarty, P., Hoad, G., Campbell, D. M., Horgan, G. W., Piyathilake, C., &
McNeill, G. (2013). Folate in pregnancy and imprinted gene and repeat
element methylation in the offspring. Am J Clin Nutr, 97(1), 94-99.
Haggarty, P., Hoad, G., Horgan, G. W., & Campbell, D. M. (2013). DNA
methyltransferase candidate polymorphisms, imprinting methylation, and
birth outcome. PLoS One, 8(7), e68896.
Hahn, P. (1984). Effect of litter size on plasma cholesterol and insulin and some
liver and adipose tissue enzymes in adult rodents. J Nutr, 114(7), 12311234.

97

Harpsøe, M. C., Basit, S., Bager, P., Wohlfahrt, J., Benn, C. S., Nøhr, E. A., et al.
(2013). Maternal obesity, gestational weight gain, and risk of asthma and
atopic disease in offspring: a study within the Danish National Birth
Cohort. J Allergy Clin Immunol, 131(4), 1033-1040.
He, Q., Ding, Z. Y., Fong, D. Y., & Karlberg, J. (2000). Risk factors of obesity in
preschool children in China: a population-based case--control study.
[Clinical Trial Multicenter Study Randomized Controlled Trial Research
Support, Non-U.S. Gov't]. International journal of obesity and related
metabolic disorders : journal of the International Association for the Study
of Obesity, 24(11), 1528-1536.
health.gov. (2008). Physical Activity Guidelines for Americans.
Heerwagen, M. J., Miller, M. R., Barbour, L. A., & Friedman, J. E. (2010).
Maternal obesity and fetal metabolic programming: a fertile epigenetic soil.
Am J Physiol Regul Integr Comp Physiol, 299(3), R711-722.
Heijmans, B. T., Tobi, E. W., Stein, A. D., Putter, H., Blauw, G. J., Susser, E. S.,
et al. (2008). Persistent epigenetic differences associated with prenatal
exposure to famine in humans. Proc Natl Acad Sci U S A, 105(44), 1704617049.
Hjollund, N. H., Jensen, T. K., Bonde, J. P., Henriksen, T. B., Andersson, A. M.,
Kolstad, H. A., et al. (2000). Spontaneous abortion and physical strain
around implantation: a follow-up study of first-pregnancy planners.
Epidemiology, 11(1), 18-23.
Hopkins, S. A., Baldi, J. C., Cutfield, W. S., McCowan, L., & Hofman, P. L.
(2010). Exercise training in pregnancy reduces offspring size without
changes in maternal insulin sensitivity. J Clin Endocrinol Metab, 95(5),
2080-2088.
Hopkins, S. A., Baldi, J. C., Cutfield, W. S., McCowan, L., & Hofman, P. L.
(2011). Effects of exercise training on maternal hormonal changes in
pregnancy. Clin Endocrinol (Oxf), 74(4), 495-500.
Hopkins, S. A., & Cutfield, W. S. (2011). Exercise in pregnancy: weighing up the
long-term impact on the next generation. Exerc Sport Sci Rev, 39(3), 120127.
Howley, E. T. (2001). Type of activity: resistance, aerobic and leisure versus
occupational physical activity. Med Sci Sports Exerc, 33(6 Suppl), S364369; discussion S419-320.

98

Hoyo, C., Fortner, K., Murtha, A. P., Schildkraut, J. M., Soubry, A., DemarkWahnefried, W., et al. (2012). Association of cord blood methylation
fractions at imprinted insulin-like growth factor 2 (IGF2), plasma IGF2, and
birth weight. Cancer Causes Control, 23(4), 635-645.
Impact of physical activity during pregnancy and postpartum on chronic disease
risk. (2006). Med Sci Sports Exerc, 38(5), 989-1006.
IOM. (2009). Institute of Medicine Guidelines, Weight gain during pregnancy:
reexamining the guidelines. Washington DC: National Academic Press.
Johnson, K. J., Carozza, S. E., Chow, E. J., Fox, E. E., Horel, S., McLaughlin, C.
C., et al. (2009). Parental age and risk of childhood cancer: a pooled
analysis. Epidemiology, 20(4), 475-483.
Jones, P. A. (1999). The DNA methylation paradox. Trends Genet, 15(1), 34-37.
Jones, P. A., & Takai, D. (2001). The role of DNA methylation in mammalian
epigenetics. Science, 293(5532), 1068-1070.
Joubert, B. R., Håberg, S. E., Bell, D. A., Nilsen, R. M., Vollset, S. E., Midttun, O.,
et al. (2014). Maternal smoking and DNA methylation in newborns: In
utero effect or epigenetic inheritance? Cancer Epidemiol Biomarkers Prev,
23(6), 1007-17.
Juhl, M., Olsen, J., Andersen, P. K., Nøhr, E. A., & Andersen, A. M. (2010).
Physical exercise during pregnancy and fetal growth measures: a study
within the Danish National Birth Cohort. Am J Obstet Gynecol, 202(1),
63.e61-68.
Kardel, K. R., & Kase, T. (1998). Training in pregnant women: effects on fetal
development and birth. Am J Obstet Gynecol, 178(2), 280-286.
Katzmarzyk, P. T., Pérusse, L., Malina, R. M., Bergeron, J., Després, J. P., &
Bouchard, C. (2001). Stability of indicators of the metabolic syndrome
from childhood and adolescence to young adulthood: the Québec Family
Study. J Clin Epidemiol, 54(2), 190-195.
Kavanagh, K., Sajadian, S., Jenkins, K. A., Wilson, M. D., Carr, J. J., Wagner, J.
D., et al. (2010). Neonatal and fetal exposure to trans-fatty acids retards
early growth and adiposity while adversely affecting glucose in mice.
[Research Support, N.I.H., Extramural]. Nutrition research, 30(6), 418-426.
Khulan, B., Cooper, W. N., Skinner, B. M., Bauer, J., Owens, S., Prentice, A. M.,
et al. (2012). Periconceptional maternal micronutrient supplementation is
associated with widespread gender related changes in the epigenome: a
99

study of a unique resource in the Gambia. Hum Mol Genet, 21(9), 20862101.
Kile, M. L., Baccarelli, A., Tarantini, L., Hoffman, E., Wright, R. O., & Christiani,
D. C. (2010). Correlation of global and gene-specific DNA methylation in
maternal-infant pairs. PLoS One, 5(10), e13730.
Knopik, V. S., Maccani, M. A., Francazio, S., & McGeary, J. E. (2012). The
epigenetics of maternal cigarette smoking during pregnancy and effects
on child development. Dev Psychopathol, 24(4), 1377-1390.
Koupil, I., & Toivanen, P. (2008). Social and early-life determinants of overweight
and obesity in 18-year-old Swedish men. [Research Support, Non-U.S.
Gov't]. International journal of obesity, 32(1), 73-81.
Kral, J. G., Biron, S., Simard, S., Hould, F. S., Lebel, S., Marceau, S., et al.
(2006). Large maternal weight loss from obesity surgery prevents
transmission of obesity to children who were followed for 2 to 18 years.
Pediatrics, 118(6), e1644-1649.
Kulkarni, S. S., Salehzadeh, F., Fritz, T., Zierath, J. R., Krook, A., & Osler, M. E.
(2012). Mitochondrial regulators of fatty acid metabolism reflect metabolic
dysfunction in type 2 diabetes mellitus. Metabolism, 61(2), 175-185.
Kuzawa, C. W. (2004). Modeling fetal adaptation to nutrient restriction: testing
the fetal origins hypothesis with a supply-demand model. J Nutr, 134(1),
194-200.
Labayen, I., Ruiz, J. R., Ortega, F. B., Loit, H. M., Harro, J., Veidebaum, T., et al.
(2010). Intergenerational cardiovascular disease risk factors involve both
maternal and paternal BMI. [Research Support, Non-U.S. Gov't]. Diabetes
care, 33(4), 894-900.
Laitinen, J., Power, C., & Jarvelin, M. R. (2001). Family social class, maternal
body mass index, childhood body mass index, and age at menarche as
predictors of adult obesity. [Research Support, Non-U.S. Gov't]. The
American journal of clinical nutrition, 74(3), 287-294.
Laker, R. C., Lillard, T. S., Okutsu, M., Zhang, M., Hoehn, K. L., Connelly, J. J.,
et al. (2014). Exercise prevents maternal high-fat diet-induced
hypermethylation of the Pgc-1α gene and age-dependent metabolic
dysfunction in the offspring. Diabetes, 63(5), 1605-11.
Lan, X., Cretney, E. C., Kropp, J., Khateeb, K., Berg, M. A., Peñagaricano, F., et
al. (2013). Maternal Diet during Pregnancy Induces Gene Expression and
DNA Methylation Changes in Fetal Tissues in Sheep. Front Genet, 4, 49.
100

Langley, S. C., & Jackson, A. A. (1994). Increased systolic blood pressure in
adult rats induced by fetal exposure to maternal low protein diets. Clin Sci
(Lond), 86(2), 217-222; discussion 121.
Langley-Evans, S. C., & McMullen, S. (2010). Developmental origins of adult
disease. Med Princ Pract, 19(2), 87-98.
Latka, M., Kline, J., & Hatch, M. (1999). Exercise and spontaneous abortion of
known karyotype. Epidemiology, 10(1), 73-75.
Law, C. M., Shiell, A. W., Newsome, C. A., Syddall, H. E., Shinebourne, E. A.,
Fayers, P. M., et al. (2002). Fetal, infant, and childhood growth and adult
blood pressure: a longitudinal study from birth to 22 years of age.
Circulation, 105(9), 1088-1092.
Leiferman, J., Swibas, T., Koiness, K., Marshall, J. A., & Dunn, A. L. (2011). My
baby, my move: examination of perceived barriers and motivating factors
related to antenatal physical activity. J Midwifery Womens Health, 56(1),
33-40.
Lifson, N., Gordon, G. B., & McClintock, R. (1955). Measurement of total carbon
dioxide production by means of D2O18. J Appl Physiol, 7(6), 704-710.
Lillycrop, K. A., Phillips, E. S., Torrens, C., Hanson, M. A., Jackson, A. A., &
Burdge, G. C. (2008). Feeding pregnant rats a protein-restricted diet
persistently alters the methylation of specific cytosines in the hepatic
PPAR alpha promoter of the offspring. Br J Nutr, 100(2), 278-282.
Lim, U., & Song, M. A. (2012). Dietary and lifestyle factors of DNA methylation.
Methods Mol Biol, 863, 359-376.
Linné, Y., Dye, L., Barkeling, B., & Rössner, S. (2004). Long-term weight
development in women: a 15-year follow-up of the effects of pregnancy.
Obes Res, 12(7), 1166-1178.
Little, R. E., & Wendt, J. K. (1991). The effects of maternal drinking in the
reproductive period: an epidemiologic review. J Subst Abuse, 3(2), 187204.
Liu, X., Chen, Q., Tsai, H. J., Wang, G., Hong, X., Zhou, Y., et al. (2014).
Maternal preconception body mass index and offspring cord blood DNA
methylation: exploration of early life origins of disease. Environ Mol
Mutagen, 55(3), 223-230.

101

Lloyd-Jones, D., Adams, R., Carnethon, M., De Simone, G., Ferguson, T. B.,
Flegal, K., et al. (2009). Heart disease and stroke statistics--2009 update:
a report from the American Heart Association Statistics Committee and
Stroke Statistics Subcommittee. Circulation, 119(3), 480-486.
Ludlow, A. T., Zimmerman, J. B., Witkowski, S., Hearn, J. W., Hatfield, B. D., &
Roth, S. M. (2008). Relationship between physical activity level, telomere
length, and telomerase activity. Med Sci Sports Exerc, 40(10), 1764-1771.
Lumey, L. H., & Stein, A. D. (1997). Offspring birth weights after maternal
intrauterine undernutrition: a comparison within sibships. Am J Epidemiol,
146(10), 810-819.
Luttropp, K., Nordfors, L., Ekström, T. J., & Lind, L. (2013). Physical activity is
associated with decreased global DNA methylation in Swedish older
individuals. Scand J Clin Lab Invest, 73(2), 184-185.
Madsen, M., Jørgensen, T., Jensen, M. L., Juhl, M., Olsen, J., Andersen, P. K.,
et al. (2007). Leisure time physical exercise during pregnancy and the risk
of miscarriage: a study within the Danish National Birth Cohort. BJOG,
114(11), 1419-1426.
Mahoney, L. T., Burns, T. L., Stanford, W., Thompson, B. H., Witt, J. D., Rost, C.
A., et al. (1996). Coronary risk factors measured in childhood and young
adult life are associated with coronary artery calcification in young adults:
the Muscatine Study. J Am Coll Cardiol, 27(2), 277-284.
Mamun, A. A., O'Callaghan, M., Callaway, L., Williams, G., Najman, J., & Lawlor,
D. A. (2009). Associations of gestational weight gain with offspring body
mass index and blood pressure at 21 years of age: evidence from a birth
cohort study. [Research Support, N.I.H., Extramural Research Support,
Non-U.S. Gov't]. Circulation, 119(13), 1720-1727.
Marceau, P., Kaufman, D., Biron, S., Hould, F. S., Lebel, S., Marceau, S., et al.
(2004). Outcome of pregnancies after biliopancreatic diversion. Obes
Surg, 14(3), 318-324.
Maule, M. M., Vizzini, L., Merletti, F., Magnani, C., Pastore, G., & Richiardi, L.
(2007). Parental age and risk of acute lymphocytic leukaemia and
embryonal tumours in the Piedmont Region, Italy. Int J Epidemiol, 36(3),
691-692.
McKenzie, T. L. (2002). Use of direct observation to assess physical activity. In
G. J. Welk (Ed.), Physical activity assessments for health-related research
(pp. 179-195). Champaign, IL: Human Kinetics.

102

Michels, K. B., Harris, H. R., & Barault, L. (2011). Birthweight, maternal weight
trajectories and global DNA methylation of LINE-1 repetitive elements.
PLoS One, 6(9), e25254.
Mills, J. L., Troendle, J., Conley, M. R., Carter, T., & Druschel, C. M. (2010).
Maternal obesity and congenital heart defects: a population-based study.
[Research Support, N.I.H., Extramural]. The American journal of clinical
nutrition, 91(6), 1543-1549.
Mudd, L. M. (2009). Leisure time physical activity during pregnancy and offspring
size. Retrieved from ProQuest Dissertations and Theses: Doctoral
Dissertation.
Mudd, L. M., Pivarnik, J., Holzman, C. B., Paneth, N., Pfeiffer, K., & Chung, H.
(2012). Leisure-time physical activity in pregnancy and the birth weight
distribution: where is the effect? J Phys Act Health, 9(8), 1168-1177.
Murphy, S. K., Huang, Z., & Hoyo, C. (2012). Differentially methylated regions of
imprinted genes in prenatal, perinatal and postnatal human tissues. PLoS
One, 7(7), e40924.
Nitert, M. D., Dayeh, T., Volkov, P., Elgzyri, T., Hall, E., Nilsson, E., et al. (2012).
Impact of an exercise intervention on DNA methylation in skeletal muscle
from first-degree relatives of patients with type 2 diabetes. Diabetes,
61(12), 3322-3332.
Nomura, Y., Lambertini, L., Rialdi, A., Lee, M., Mystal, E. Y., Grabie, M., et al.
(2014). Global methylation in the placenta and umbilical cord blood from
pregnancies with maternal gestational diabetes, preeclampsia, and
obesity. Reprod Sci, 21(1), 131-137.
Ogden, C. L., Carroll, M. D., Curtin, L. R., Lamb, M. M., & Flegal, K. M. (2010).
Prevalence of high body mass index in US children and adolescents,
2007-2008. JAMA : the journal of the American Medical Association,
303(3), 242-249.
Oostdam, N., van Poppel, M. N., Wouters, M. G., & van Mechelen, W. (2011).
Interventions for preventing gestational diabetes mellitus: a systematic
review and meta-analysis. J Womens Health (Larchmt), 20(10), 15511563.
Owe, K. M., Nystad, W., Skjaerven, R., Stigum, H., & Bø, K. (2011). Exercise
During Pregnancy and the Gestational Age Distribution: A Cohort Study.
Med Sci Sports Exerc, 44(6), 1067-74.

103

Painter, R. C., Osmond, C., Gluckman, P., Hanson, M., Phillips, D. I., &
Roseboom, T. J. (2008). Transgenerational effects of prenatal exposure to
the Dutch famine on neonatal adiposity and health in later life. BJOG,
115(10), 1243-1249.
Paneth, N., Ahmed, F., & Stein, A. D. (1996). Early nutritional origins of
hypertension: a hypothesis still lacking support. J Hypertens Suppl, 14(5),
S121-129.
Physical Activity Guidelines Advisory Committee report, 2008. To the Secretary
of Health and Human Services. Part A: executive summary. (2009). Nutr
Rev, 67(2), 114-120.
Podvin, D., Kuehn, C. M., Mueller, B. A., & Williams, M. (2006). Maternal and
birth characteristics in relation to childhood leukaemia. Paediatr Perinat
Epidemiol, 20(4), 312-322.
Poston, L., Harthoorn, L. F., & Van Der Beek, E. M. (2011). Obesity in
pregnancy: implications for the mother and lifelong health of the child. A
consensus statement. Pediatric research, 69(2), 175-180.
Rees, W. D., Hay, S. M., Brown, D. S., Antipatis, C., & Palmer, R. M. (2000).
Maternal protein deficiency causes hypermethylation of DNA in the livers
of rat fetuses. J Nutr, 130(7), 1821-1826.
Regnault, N., Botton, J., Forhan, A., Hankard, R., Thiebaugeorges, O., Hillier, T.
A., et al. (2010). Determinants of early ponderal and statural growth in fullterm infants in the EDEN mother-child cohort study. [Research Support,
Non-U.S. Gov't]. The American journal of clinical nutrition, 92(3), 594-602.
Relton, C. L., Groom, A., St Pourcain, B., Sayers, A. E., Swan, D. C., Embleton,
N. D., et al. (2012). DNA methylation patterns in cord blood DNA and body
size in childhood. PLoS One, 7(3), e31821.
Reynolds, R. M., Osmond, C., Phillips, D. I., & Godfrey, K. M. (2010). Maternal
BMI, parity, and pregnancy weight gain: influences on offspring adiposity
in young adulthood. [Research Support, Non-U.S. Gov't]. The Journal of
clinical endocrinology and metabolism, 95(12), 5365-5369.
Ribel-Madsen, R., Fraga, M. F., Jacobsen, S., Bork-Jensen, J., Lara, E.,
Calvanese, V., et al. (2012). Genome-wide analysis of DNA methylation
differences in muscle and fat from monozygotic twins discordant for type 2
diabetes. PLoS One, 7(12), e51302.

104

Rooney, B. L., Mathiason, M. A., & Schauberger, C. W. (2010). Predictors of
Obesity in Childhood, Adolescence, and Adulthood in a Birth Cohort.
Maternal and child health journal, 15(8), 1166-75.
Rose, N. C., Haddow, J. E., Palomaki, G. E., & Knight, G. J. (1991). Self-rated
physical activity level during the second trimester and pregnancy outcome.
Obstet Gynecol, 78(6), 1078-1080.
Roseboom, T., de Rooij, S., & Painter, R. (2006). The Dutch famine and its longterm consequences for adult health. Early Hum Dev, 82(8), 485-491.
Rousham, E. K., Clarke, P. E., & Gross, H. (2006). Significant changes in
physical activity among pregnant women in the UK as assessed by
accelerometry and self-reported activity. Eur J Clin Nutr, 60(3), 393-400.
Ruager-Martin, R., Hyde, M. J., & Modi, N. (2010). Maternal obesity and infant
outcomes. Early Hum Dev, 86(11), 715-722.
Rönn, T., Volkov, P., Davegårdh, C., Dayeh, T., Hall, E., Olsson, A. H., et al.
(2013). A six months exercise intervention influences the genome-wide
DNA methylation pattern in human adipose tissue. PLoS Genet, 9(6),
e1003572.
Saftlas, A. F., Logsden-Sackett, N., Wang, W., Woolson, R., & Bracken, M. B.
(2004). Work, leisure-time physical activity, and risk of preeclampsia and
gestational hypertension. Am J Epidemiol, 160(8), 758-765.
Samuelsson, A. M., Matthews, P. A., Argenton, M., Christie, M. R., McConnell, J.
M., Jansen, E. H., et al. (2008). Diet-induced obesity in female mice leads
to offspring hyperphagia, adiposity, hypertension, and insulin resistance: a
novel murine model of developmental programming. [Research Support,
Non-U.S. Gov't]. Hypertension, 51(2), 383-392.
Sarter, B., Long, T. I., Tsong, W. H., Koh, W. P., Yu, M. C., & Laird, P. W. (2005).
Sex differential in methylation patterns of selected genes in Singapore
Chinese. Hum Genet, 117(4), 402-403.
Schack-Nielsen, L., Michaelsen, K. F., Gamborg, M., Mortensen, E. L., &
Sorensen, T. I. (2010). Gestational weight gain in relation to offspring body
mass index and obesity from infancy through adulthood. International
journal of obesity, 34(1), 67-74.
Schlaff, R. A., Holzman, C., Maier, K. S., Pfeiffer, K. A., & Pivarnik, J. M. (2014).
Associations among gestational weight gain, physical activity, and prepregnancy body size with varying estimates of pre-pregnancy weight.
Midwifery, In press.
105

Schär, P., & Fritsch, O. (2011). DNA repair and the control of DNA methylation.
Prog Drug Res, 67, 51-68.
Sequeira, M. M., Rickenbach, M., Wietlisbach, V., Tullen, B., & Schutz, Y. (1995).
Physical activity assessment using a pedometer and its comparison with a
questionnaire in a large population survey. Am J Epidemiol, 142(9), 989999.
Silverman, B. L., Metzger, B. E., Cho, N. H., & Loeb, C. A. (1995). Impaired
glucose tolerance in adolescent offspring of diabetic mothers. Relationship
to fetal hyperinsulinism. Diabetes Care, 18(5), 611-617.
Slaughter, J., Wei, C., Korzeniewski, S. J., Lu, Q., Beck, J. S., Khoo, S. K., et al.
(2013). High correlations in gene expression between paired umbilical
cord blood and neonatal blood of healthy newborns on Guthrie cards. J
Matern Fetal Neonatal Med, 26(18), 1765-1767.
Smith, J., Cianflone, K., Biron, S., Hould, F. S., Lebel, S., Marceau, S., et al.
(2009). Effects of maternal surgical weight loss in mothers on
intergenerational transmission of obesity. J Clin Endocrinol Metab, 94(11),
4275-4283.
Sorensen, T. K., Williams, M. A., Lee, I. M., Dashow, E. E., Thompson, M. L., &
Luthy, D. A. (2003). Recreational physical activity during pregnancy and
risk of preeclampsia. Hypertension, 41(6), 1273-1280.
Spinillo, A., Capuzzo, E., Baltaro, F., Piazza, G., Nicola, S., & Iasci, A. (1996).
The effect of work activity in pregnancy on the risk of fetal growth
retardation. Acta Obstet Gynecol Scand, 75(6), 531-536.
St-Pierre, J., Hivert, M. F., Perron, P., Poirier, P., Guay, S. P., Brisson, D., et al.
(2012). IGF2 DNA methylation is a modulator of newborn's fetal growth
and development. Epigenetics, 7(10), 1125-1132.
Stafne, S. N., Salvesen, K., Romundstad, P. R., Eggebø, T. M., Carlsen, S. M., &
Mørkved, S. (2012). Regular exercise during pregnancy to prevent
gestational diabetes: a randomized controlled trial. Obstet Gynecol,
119(1), 29-36.
Steegers-Theunissen, R. P., Obermann-Borst, S. A., Kremer, D., Lindemans, J.,
Siebel, C., Steegers, E. A., et al. (2009). Periconceptional maternal folic
acid use of 400 microg per day is related to increased methylation of the
IGF2 gene in the very young child. PLoS One, 4(11), e7845.

106

Sternfeld, B., Quesenberry, C. P., Eskenazi, B., & Newman, L. A. (1995a).
Exercise during pregnancy and pregnancy outcome. Med Sci Sports
Exerc, 27(5), 634-640.
Sternfeld, B., Quesenberry, C. P., Eskenazi, B., & Newman, L. A. (1995b).
Exercise during pregnancy and pregnancy outcome. Med Sci Sports
Exerc, 27(5), 634-640.
Teitelman, A. M., Welch, L. S., Hellenbrand, K. G., & Bracken, M. B. (1990).
Effect of maternal work activity on preterm birth and low birth weight. Am J
Epidemiol, 131(1), 104-113.
Tost, J. (2009). DNA methylation: an introduction to the biology and the diseaseassociated changes of a promising biomarker. Methods Mol Biol, 507, 320.
Trost, S. G., McIver, K. L., & Pate, R. R. (2005). Conducting accelerometerbased activity assessments in field-based research. Med Sci Sports
Exerc, 37(11 Suppl), S531-543.
Turan, N., Ghalwash, M. F., Katari, S., Coutifaris, C., Obradovic, Z., & Sapienza,
C. (2012). DNA methylation differences at growth related genes correlate
with birth weight: a molecular signature linked to developmental origins of
adult disease? BMC Med Genomics, 5, 10.
van Sluijs, E. M., McMinn, A. M., & Griffin, S. J. (2008). Effectiveness of
interventions to promote physical activity in children and adolescents:
systematic review of controlled trials. Br J Sports Med, 42(8), 653-657.
van Straten, E. M., Bloks, V. W., Huijkman, N. C., Baller, J. F., van Meer, H.,
Lütjohann, D., et al. (2010). The liver X-receptor gene promoter is
hypermethylated in a mouse model of prenatal protein restriction. Am J
Physiol Regul Integr Comp Physiol, 298(2), R275-282.
Viswanathan, M., Siega-Riz, A. M., Moos, M. K., Deierlein, A., Mumford, S.,
Knaack, J., et al. (2008). Outcomes of maternal weight gain. Evid Rep
Technol Assess (Full Rep)(168), 1-223.
Walsh, R. A. (1994). Effects of maternal smoking on adverse pregnancy
outcomes: examination of the criteria of causation. Hum Biol, 66(6), 10591092.
Wang, L., Psilander, N., Tonkonogi, M., Ding, S., & Sahlin, K. (2009). Similar
expression of oxidative genes after interval and continuous exercise. Med
Sci Sports Exerc, 41(12), 2136-2144.

107

Weber, M., Hellmann, I., Stadler, M. B., Ramos, L., Pääbo, S., Rebhan, M., et al.
(2007). Distribution, silencing potential and evolutionary impact of
promoter DNA methylation in the human genome. Nat Genet, 39(4), 457466.
White, A. J., Sandler, D. P., Bolick, S. C., Xu, Z., Taylor, J. A., & Deroo, L. A.
(2013). Recreational and household physical activity at different time
points and DNA global methylation. Eur J Cancer.
White, E., Pivarnik, J., & Pfeiffer, K. (2013). Resistance Training During
Pregnancy and Pregnancy and Birth Outcomes. J Phys Act Health, Epub
ahead of print.
Whiteman, V. E., Aliyu, M. H., August, E. M., McIntosh, C., Duan, J., Alio, A. P.,
et al. (2011). Changes in prepregnancy body mass index between
pregnancies and risk of gestational and type 2 diabetes. Arch Gynecol
Obstet, 284(1), 235-240.
Widiker, S., Karst, S., Wagener, A., & Brockmann, G. A. (2010). High-fat diet
leads to a decreased methylation of the Mc4r gene in the obese BFMI and
the lean B6 mouse lines. J Appl Genet, 51(2), 193-197.
Winick, M., & Noble, A. (1966). Cellular response in rats during malnutrition at
various ages. J Nutr, 89(3), 300-306.
Yu, Z., Han, S., Zhu, J., Sun, X., Ji, C., & Guo, X. (2013). Pre-pregnancy body
mass index in relation to infant birth weight and offspring
overweight/obesity: a systematic review and meta-analysis. PLoS One,
8(4), e61627.
Zhang, F. F., Cardarelli, R., Carroll, J., Fulda, K. G., Kaur, M., Gonzalez, K., et al.
(2011). Significant differences in global genomic DNA methylation by
gender and race/ethnicity in peripheral blood. Epigenetics, 6(5), 623-629.
Zhang, F. F., Cardarelli, R., Carroll, J., Zhang, S., Fulda, K. G., Gonzalez, K., et
al. (2011). Physical activity and global genomic DNA methylation in a
cancer-free population. Epigenetics, 6(3), 293-299.
Zhang, F. F., Santella, R. M., Wolff, M., Kappil, M. A., Markowitz, S. B., &
Morabia, A. (2012). White blood cell global methylation and IL-6 promoter
methylation in association with diet and lifestyle risk factors in a cancerfree population. Epigenetics, 7(6), 606-614.
Zhang, X., Li, J., Gu, Y., Zhao, Y., Wang, Z., & Jia, G. (2011). A pilot study on
environmental and behavioral factors related to missed abortion. Environ
Health Prev Med, 16(4), 273-278.
108