Intrauterine Development: An Important Critical Window

A number of epidemiological studies have explored the links between adverse prenatal conditions and increased risk for diseases, health problems, and psychological outcomes later in life. Such studies have examined historical examples, such as the Dutch Famine Birth Cohort, which consists of men and women born as term singletons in Amsterdam, The Netherlands (de Rooij, Wouters, Yonker, Painter, & Roseboom, Reference de Rooij, Wouters, Yonker, Painter and Roseboom2010; Stein, Susser, Saenger, & Marolla, Reference Stein, Susser, Saenger and Marolla1972). The formation of the Dutch Famine Birth Cohort, resulting from the Dutch Famine of 1944–1945, provided researchers with an opportunity to study the particular adverse exposure of starvation during pregnancy and a number of health and developmental outcomes potentially associated with this challenging adverse exposure. Associations have been reported between adverse intrauterine environment as influenced by famine and a number of diseases and conditions, such as increased risk for Type 2 diabetes mellitus, cardiovascular disease, and other metabolic disorders and decreased cognitive function later in life (Argente, Mehls, & Barrios, Reference Argente, Mehls and Barrios2010; Barker & Clark, Reference Barker and Clark1997; de Rooij et al., Reference de Rooij, Wouters, Yonker, Painter and Roseboom2010). In addition, researchers found that maternal weight loss or moderate to low weight gain was significantly associated with infant birth weight, length, and other measures of fetal growth status, as well as with trimester of exposure to famine (Stein, Ravelli, & Lumey, Reference Stein, Ravelli and Lumey1995). In terms of psychological outcomes, increased risk of affective disorders has been found in males exposed to famine during their second trimester (Brown, Susser, Lin, Neugebauer, & Gorman, Reference Brown, Susser, Lin, Neugebauer and Gorman1995). Findings such as these underscore the importance of considering timing of prenatal exposure to adverse conditions, such as famine, as well as potential confounding elements, such as the gender of the infant, when looking at outcomes across development. Birth cohorts such as the Dutch Famine Birth Cohort have provided researchers with many of the earliest tools necessary to investigate epidemiological associations between adverse intrauterine conditions and postnatal health and disease.

Maternal Cigarette Smoking During Pregnancy

We now move beyond the example of famine to consider one of the most common, potentially hazardous environmental exposures during pregnancy: maternal cigarette smoking. Studies have shown that there are more than 4,000 chemicals in cigarette smoke, including benzo(a)pyrene, nicotine, and carbon monoxide, and more than 40 of these chemicals are known carcinogens (Thielen, Klus, & Muller, Reference Thielen, Klus and Muller2008; US Department of Health and Human Services, 2010). Nicotine crosses the placenta, and fetal concentrations of nicotine can be 15% higher than maternal concentrations (Lambers & Clark, Reference Lambers and Clark1996). Despite a number of studies showing a decrease in the overall prevalence of smoking in women in the past 20 years, key studies have suggested that the prevalence of smoking in young pregnant women under 20 years of age has increased, with prevalence rates of 30%–40% (Jaakkola, Jaakkola, Gissler, & Jaakkola, Reference Jaakkola, Jaakkola, Gissler and Jaakkola2001; Mohsin & Bauman, Reference Mohsin and Bauman2005; US Department of Health and Human Services, 2004). Others have reported that 12%–15% of all women of childbearing age smoke while pregnant (Cnattingius, Reference Cnattingius2004; Goodwin, Keyes, & Simuro, Reference Goodwin, Keyes and Simuro2007). Considering that women who smoke during pregnancy are more likely to be nicotine dependent, less likely to quit, and have a partner who smokes (Agrawal et al., Reference Agrawal, Knopik, Pergadia, Waldron, Bucholz and Martin2008; Knopik et al., Reference Knopik, Sparrow, Madden, Bucholz, Hudziak and Reich2005), it is important to also keep the effects of second-hand smoke exposure in mind, which are difficult to tease apart from direct prenatal primary smoke exposure. Women who are pregnant may also be exposed to second-hand smoke in homes, vehicles, the workplace, or public areas. More than 126 million nonsmoking adults continue to be exposed to second-hand smoke and current estimates suggest that almost 60% of children, aged 3–11, are exposed to second-hand smoke (US Department of Health and Human Services, 2006). Thus, exposure to cigarette smoke remains a common and hazardous in utero exposure.

Previous work has suggested that maternal cigarette smoking during pregnancy is associated with increased risk for spontaneous abortion (Castles, Adams, Melvin, Kelsch, & Boulton, Reference Castles, Adams, Melvin, Kelsch and Boulton1999), preterm delivery (Castles et al., Reference Castles, Adams, Melvin, Kelsch and Boulton1999; Kaddar et al., Reference Kaddar, Rouault, Chien, Chebel, Gadoux and Salles2009; Shah & Bracken, Reference Shah and Bracken2000), respiratory disease (Cook & Strachan, Reference Cook and Strachan1999), immune system difficulties such as asthma and allergies (Prescott & Clifton, 2009), and cancer later in life (Doherty, Grabowski, Hoffman, Ng, & Zelikoff, Reference Doherty, Grabowski, Hoffman, Ng and Zelikoff2009). The results also suggest that there are a variety of placental complications linked to prenatal exposure to cigarette smoke, including alterations to the development and function of the placenta (Einarson & Riordan, Reference Einarson and Riordan2009).

A number of studies have suggested that prenatal tobacco exposure is associated with a number of serious neurodevelopmental and behavioral consequences in infants, children, and adolescents. For example, studies in infants have shown that maternal smoking during pregnancy is associated with delayed psychomotor and mental developmental scores (as measured by the Bayley Scales of Infant Development; Kiechl-Kohlendorfer et al., Reference Kiechl-Kohlendorfer, Ralser, Pupp Peglow, Reiter, Griesmaier and Trawoger2010). Kable, Coles, Lynch, and Carroll (Reference Kable, Coles, Lynch and Carroll2009) analyzed the impact of maternal smoking during pregnancy on auditory brainstem responses and determined that, after controlling for a number of potential confounders, maternal smoking during pregnancy was negatively related to auditory brainstem response latency in infants. It should be noted that because alterations in infants’ auditory processing have been previously found to be highly predictive of future reading and language difficulties, these findings suggest that maternal cigarette smoking during pregnancy may negatively affect a child's speech and language development (Benasich & Tallal, Reference Benasich and Tallal2002; Kable et al., Reference Kable, Coles, Lynch and Carroll2009; Molfese, Reference Molfese2000).

There are many examples of prenatal smoking exposure research in childhood (for a review, see Knopik, Reference Knopik, Jacob, Haber, Swenson and Howell2009), perhaps because of more sensitive assessments for this age group, relative to infants. For instance, Olds, Henderson, and Tatelbaum (Reference Olds, Henderson and Tatelbaum1994) conducted a study investigating associations between maternal cigarette smoking during pregnancy and offspring intelligence at age 4 and found that children whose mothers smoked 10 or more cigarettes per day during their pregnancy exhibited intellectual impairment relative to children whose mothers did not smoke cigarettes during pregnancy. Huijbregts, Seguin, Zoccolillo, Boivin, and Tremblay (Reference Huijbregts, Seguin, Zoccolillo, Boivin and Tremblay2007, Reference Huijbregts, Seguin, Zoccolillo, Boivin and Tremblay2008) found associations between maternal cigarette smoking during pregnancy and physical aggression during early childhood. This is consistent with multiple studies suggesting an association of prenatal smoke exposure and increased externalizing disorders, such as conduct disorder and attention-deficit/hyperactivity disorder (Abbott & Winzer-Serhan, Reference Abbott and Winzer-Serhan2012; Cornelius & Day, Reference Cornelius and Day2009; Knopik, Reference Knopik2009). Cognitive function has also been shown to be negatively affected by maternal smoking during pregnancy, with deficits in sustained attention (Fried, O'Connell, & Watkinson, Reference Fried, O'Connell and Watkinson1992; Fried, Watkinson, & Gray, Reference Fried, Watkinson and Gray1992), response inhibition, memory, and impulsivity, overall cognitive function, receptive language (Fried, O'Connell et al., 1992), verbal learning and design memory (Cornelius, Ryan, Day, Goldschmidt, & Willford, Reference Cornelius, Ryan, Day, Goldschmidt and Willford2001), problem solving (Cornelius et al., Reference Cornelius, Ryan, Day, Goldschmidt and Willford2001), speech and language (Makin, Fried, & Watkinson, Reference Makin, Fried and Watkinson1991), school performance (Lambe, Hultman, Torrang, Maccabe, & Cnattingius, Reference Lambe, Hultman, Torrang, Maccabe and Cnattingius2006), and auditory processing (McCartney, Fried, & Watkinson, Reference McCartney, Fried and Watkinson1994). Dose–response relationships, in which the smoking-related relative risk increases with the amount smoked, have also been reported for general cognitive ability (Sexton, Fox, & Hebel, Reference Sexton, Fox and Hebel1990), arithmetic, and spelling (Batstra, Hadders-Algra, & Neeleman, Reference Batstra, Hadders-Algra and Neeleman2003), suggesting the presence of vulnerable periods during fetal development (Ernst, Moolchan, & Robinson, Reference Ernst, Moolchan and Robinson2001).

Recent research has also suggested that second-hand smoke exposure during pregnancy is associated with a decrease in infant mental development index score and an increase in the risk of developmental delay in the infant (Lee et al., Reference Lee, Hong, Park, Ha, Kim and Chang2011), thereby extending the risk profile related to smoking exposure beyond just direct, primary smoke exposure during pregnancy. Overall, these observations suggest that smoke exposure during pregnancy remains a common exposure that can have major ramifications on the in utero growth and development of the fetus. Moreover, due to the plethora of scientific data suggesting negative consequences associated with smoking during pregnancy, pregnant women are cautioned against smoking while pregnant (Cornelius & Day, Reference Cornelius and Day2009; Shea & Steiner, Reference Shea and Steiner2008). However, despite the numerous research articles suggesting undesirable outcomes in children exposed to prenatal smoke exposure and warnings encouraging women to stop smoking while pregnant, the underlying biological processes in humans are not well understood. Thus, we turn our attention now to describing one potential piece of this puzzle, namely, fetal programming and how this may play an important role in future child development.

Fetal Programming

What is meant by fetal programming? The theory of fetal programming has been described as a model of gene–environment interaction that explains the influence of the in utero environment on the molecular character of development (Barker & Clark, Reference Barker and Clark1997; Hales & Barker, Reference Hales and Barker1992). One of the forefathers of this theory is David Barker. Starting in the 1980s and 1990s, he proposed and tested the hypothesis that an adverse fetal nutritional environment but plentiful food in adulthood might be a major factor associated with a number of adult diseases (Schulz, Reference Schulz2010). The Barker hypothesis further hypothesized that adverse intrauterine conditions may result in a negative, or poor, maternal forecast, commonly manifested in small for gestational age status or reduced infant birth weight. This poor maternal forecast “predicts” that the child will be born into a postnatal environment in which resources are scarce; thus, the child has been forecasted or programmed to thrive in such a poor environment. Poor maternal forecasts can prove incorrect if a child is born into what is or soon becomes a nutrient-rich environment.

We emphasize that current research expands the notion of an adverse intrauterine environment beyond the traditional, nutrient-poor environment first elucidated by Barker and colleagues in their analysis of infants from the Dutch Famine Birth Cohort (Barker & Clark, Reference Barker and Clark1997; Hales & Barker, Reference Hales and Barker1992). Adverse intrauterine conditions may also be the result of exposure to viruses (e.g., influenza); increased levels of stress during pregnancy; and maternal cigarette smoking and second-hand (passive) smoke exposure during pregnancy, which is of the most importance for this article. In this expanded view of adverse intrauterine conditions and the resulting effects on the fetus, an exposure such as prenatal exposure to cigarette smoke may lead to a negative maternal forecast for the fetus. Although prenatal smoke exposure certainly contributes to this adverse environment, evidence that maternal cigarette smoking during pregnancy is correlated with other potential contributors to adverse in utero environmental conditions can make causal attribution difficult (for a review, see Knopik et al., Reference Knopik, Jacob, Haber, Swenson and Howell2009). This is supported by recent work from our own group and others suggesting that maternal smoking during pregnancy is correlated with many risk factors, such as lower levels of maternal education (D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Grann2010), spousal/significant other substance dependence (Knopik et al., Reference Knopik, Sparrow, Madden, Bucholz, Hudziak and Reich2005, Reference Knopik, Heath, Jacob, Slutske, Bucholz and Madden2006), nicotine dependence (Agrawal et al., Reference Agrawal, Knopik, Pergadia, Waldron, Bucholz and Martin2008), as well as maternal attention-deficit/hyperactivity disorder (ADHD) and other psychopathology (D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Neiderhiser2010; Huizink & Mulder, Reference Huizink and Mulder2006; Knopik, Reference Knopik2009; Knopik et al., Reference Knopik, Jacob, Haber, Swenson and Howell2009), which may come with a host of additional influences on intrauterine environment that also predict poor offspring performance.

Figure 1 summarizes the concept of fetal programming and the contributions of the variety of stimuli that make up the intrauterine environment on maternal forecasts. As shown in the figure, in utero stimuli include many exposures that may influence the intrauterine environment, such as exposure to nutrients, stress, drugs, trauma, and smoking. A largely positive and healthy intrauterine environment may result in the mother imparting a rich maternal forecast on her developing fetus, predicting a rich postbirth environment in which resources are predicted to be plentiful and negative exposures are predicted to be at a minimum. A relatively negative or adverse intrauterine environment may result in the mother imparting a poor maternal forecast on her developing fetus, sometimes characterized by a “thrifty phenotype” (Hales & Barker, Reference Hales and Barker1992) or small for gestational age status, thereby preparing her child to survive in a poor postbirth environment, where resources are predicted to be scarce and/or negative postnatal exposures to be frequent and abundant. Maternal forecasts that do not accurately predict the postbirth environment have been hypothesized to lead to negative consequences for the health of the child over its life course, such as increased risk for metabolic diseases and decreased cognitive functioning in offspring that had been given a poor maternal forecast but were born into a rich environment (Argente et al., Reference Argente, Mehls and Barrios2010; Barker & Clark, Reference Barker and Clark1997; de Rooij et al., Reference de Rooij, Wouters, Yonker, Painter and Roseboom2010; Hochberg et al., Reference Hochberg, Feil, Constancia, Fraga, Junien and Carel2010).

Figure 1.

In utero stimuli-associated epigenetic alterations and maternal forecasts. In utero stimuli comprise the many exposures that may characterize the intrauterine environment, including exposure to nutrients, stress, and drugs. A largely positive and healthy intrauterine environment may result in the mother imparting a rich maternal forecast on her developing fetus, predicting a rich postbirth environment in which resources are predicted to be plentiful and negative exposures at a minimum level. A relatively negative or adverse intrauterine environment may result in the mother imparting a poor maternal forecast, often characterized by a “thrifty phenotype,” on her developing fetus, thereby preparing her child to survive in a poor postbirth environment in which resources are predicted to be scarce and/or negative exposures (e.g., second-hand smoke exposure) to be frequent and abundant. Maternal forecasts that do not accurately predict the postbirth environment have been hypothesized to lead to negative consequences for the health of the child over its life course, such as increased risk for metabolic diseases in offspring born into a rich environment with a poor maternal forecast. In utero stimuli may influence maternal forecasting through epigenetic mechanisms, both directly and indirectly. These epigenetic-mediated maternal forecasts may be accessible to measurement through techniques established for measuring changes to epigenetic modes of regulation. In addition, these epigenetic marks, such as DNA methylation profiles or miRNA expression, may have utility as diagnostic biomarkers capable of predicting increased risk for diseases or disease progression and as therapeutic targets. Scientific artwork by Jennifer Z. Joukhadar. [A color version of this figure can be viewed online at http://journals.cambridge.org/dpp]

Despite the large literature suggesting undesirable outcomes in children exposed to maternal smoking during pregnancy, the underlying biological processes in humans are not well understood. Work is under way to better understand the underlying mechanisms of fetal programming, and many researchers have focused on how epigenetic mechanisms may play a role in mediating the effects of environmental exposures on future outcomes, such as the physical and behavioral health of the individual in infancy, childhood, and adulthood (Maccani & Marsit, Reference Maccani and Marsit2009). Understanding how fetal programming may lead to future health consequences have caused some to adopt an even more expanded theory of fetal programming, namely the “developmental origins of health and disease” (Barker, Reference Barker2004; Gillman, Reference Gillman2005).

Epigenetics: Mechanisms by Which Fetal Programming May Occur

The study of epigenetics, or the study of changes in gene expression that are not caused by changes in the sequence of DNA (Bird, Reference Bird2007), has intrigued scores of scientists over the past few decades. It is these epigenetic mechanisms that can influence whether one's genes are switched on or off (i.e., gene expression). It may be through epigenetic mechanisms that environmental factors like diet, stress, prenatal nutrition, or prenatal drug exposure can lead to changes in gene expression from one cell to its daughter cells and, in some cases, from one generation to the next. Work in this area has focused on examining four main modes of epigenetic gene regulation: DNA methylation, imprinting, histone modification, and noncoding RNA-mediated gene regulation, especially by microRNA (miRNA). (For a more comprehensive review on these four modes of epigenetic regulation, as well as on the technological advances that have made it possible to measure such changes, see Maccani & Marsit, Reference Maccani and Marsit2009.)

As hinted above, new subfields have emerged to explore epigenetic effects in a variety of settings. The field of “environmental epigenetics” studies how environmental exposures affect epigenetic mechanisms, which was described by Reamon-Buettner, Mutschler, and Borlak (Reference Reamon-Buettner, Mutschler and Borlak2008). Many researchers are interested in uncovering how environmental exposures at sensitive periods of development, such as maternal cigarette smoking during pregnancy, might influence epigenetics and thus affect the developing fetus and fetal programming. Research in both human cohorts and model systems (e.g., mice) continues to characterize the influence of environmental exposures on epigenetics. Figure 1 describes how certain prenatal exposures may, both directly and indirectly, influence maternal forecasting through epigenetic mechanisms. In addition, these epigenetic mechanisms, such as DNA methylation profiles or miRNA expression, may have utility not only as diagnostic biomarkers capable of predicting increased risk for behavioral deficits, diseases, or disease progression, but also as therapeutic targets. This is similar to recent reports using gene-expression differences, which were identified via a blood test, to predict youth at risk for early onset depression (Pajer et al., Reference Pajer, Andrus, Gardner, Lourie, Strange and Campo2012).

As we highlight in this article, the research community has placed particular focus on investigating the influence of maternal cigarette smoking during pregnancy on DNA methylation patterns, whereas work on the effects of maternal cigarette smoking during pregnancy on miRNA has been much less comprehensive. Investigations of specific associations between prenatal smoke exposure and histone modifications or imprinting have yet to be described in the literature, but others have cited these two modes of epigenetic regulation as important future considerations for research in the context of in utero smoke exposure (Swanson, Entringer, Buss, & Wadhwa, Reference Swanson, Entringer, Buss and Wadhwa2009).

Maternal Cigarette Smoking During Pregnancy Is Associated With Atypical DNA Methylation Patterns

DNA methylation is one of the more widely studied and well characterized of the main modes of epigenetic regulation. DNA methylation of cytosine residues is performed by one of a number of DNA methyltransferases that add a methyl group to a specific cytosine residue. These cytosine residues often reside in cytosine- and guanine-rich stretches of DNA called cytosine nucleotide–phosphate–guanine nucleotide (CpG) islands. Sometimes DNA methylation is also referred to as CpG methylation. In general, a DNA methylation-regulated gene whose promoter region shows a great degree of methylation will be effectively silenced. When the same gene's promoter region is not methylated (i.e., the promoter is hypomethylated), the gene will likely not be silenced, and thus the gene will be expressed. Research has determined that the blocking of transcription in a methylated gene is not due to the methylation of DNA alone, but rather due to the irregular binding of a variety of proteins. In the presence of DNA methylation, proteins that normally bind DNA and enable transcription to proceed are unable to bind as well or at all, which effectively limits or stops transcription. Research is continuing to develop a better understanding of how DNA methylation in a gene's promoter region controls the complex regulatory environment necessary for transcription and, thus, gene expression.

During the period of embryonic development, methylation patterns of the germline and somatic cell lineages are established (Maccani & Marsit, Reference Maccani and Marsit2009). During the cleavage phase (i.e., the early cell divisions that occur as a fertilized egg begins to develop into an embryo), methylation in the zygote's genome is almost completely removed. After implantation, as the cells produced during the cleavage phase begin to organize themselves (i.e., gastrulation), the organism's methylation patterns are reestablished by de novo methylation (Jaenisch, Reference Jaenisch1997; Kafri et al., Reference Kafri, Ariel, Brandeis, Shemer, Urven and McCarrey1992; Monk, Boubelik, & Lehnert, Reference Monk, Boubelik and Lehnert1987; Razin & Shemer, Reference Razin and Shemer1995). Such patterning and repatterning of methylation marks also occurs in trophoblast lineages (i.e., the various specialized cells comprising the placenta; Jaenisch, Reference Jaenisch1997). Proper setting and resetting of methylation marks throughout development is crucial for the proper health and development of the embryo.

Recent studies have even reported initial findings suggesting that an additional mode of epigenetic regulation, namely, conversion of 5-methylcytosine to 5-hydroxymethylcytosine, may play a role in the reprogramming of the zygote (Wossidlo et al., Reference Wossidlo, Nakamura, Lepikhov, Marques, Zakhartchenko and Boiani2011). Further studies will be necessary to understand more completely the implications of these findings on the patterning of methylation marks during development, as well as the effect of exposures on the underlying mechanisms.

When considering epigenetic research in general, DNA methylation is the most heavily studied mode of epigenetic regulation (Bird, Reference Bird2007). Thus, it is not surprising that the majority of published articles on the epigenetics of maternal cigarette smoking during pregnancy describe associations with DNA methylation. A number of studies have characterized associations between prenatal smoke exposure and aberrant DNA methylation patterns in a variety of tissues. The following sections describe many of these efforts.

Maternal Cigarette Smoking During Pregnancy Is Associated With Differential miRNA Expression

Compared to the growing body of literature characterizing associations between DNA methylation patterns and maternal cigarette smoking during pregnancy, much less is known about associations between maternal cigarette smoking during pregnancy and miRNA expression. miRNA are small, ~22 nucleotide-long noncoding RNA molecules, which were previously shown to be highly ubiquitous and conserved across many species (Lee, Feinbaum, & Ambros, Reference Lee, Feinbaum and Ambros1993). miRNA control gene expression by base pairing to the 3′-untranslated region of a target messenger RNA (mRNA). In some cases, there will be an imperfect match, or imperfect complementarity, between the miRNA and its mRNA target, which results in problems translating the mRNA into a protein molecule (i.e., translational repression). In other cases of perfect complementarity between the miRNA and the target mRNA, the result is degradation of the target mRNA. Research has found that because partial or imperfect complementarity of a miRNA to a target mRNA can still result in translational repression, a single miRNA has the capability of regulating a number of genes (Du & Zamore, Reference Du and Zamore2007). Through the mechanism of negative regulation, miRNA have been shown to be involved in regulating a number of key biological processes, including cell proliferation, differentiation, and apoptosis (i.e., cell death; Maccani & Knopik, Reference Maccani and Knopik2012; Maccani & Marsit, Reference Maccani and Marsit2009; Miska, Reference Miska2005).

One of the few studies investigating maternal cigarette smoking during pregnancy and miRNA expression was conducted by Maccani et al. (Reference Maccani, Avissar-Whiting, Banister, McGonnigal, Padbury and Marsit2010). They used a candidate miRNA approach to investigate the changes in four human placental miRNA associated with maternal cigarette smoking during pregnancy. These four miRNA were selected for investigation because they had been reported to be expressed in the placenta and had been shown to be involved in the regulation of key cell processes. The work identified the association of maternal cigarette smoking during pregnancy with the downregulation of three specific miRNA: miR-16, miR-21, and miR-146a. Based on these findings, the research was expanded to consider three placental cell lines from different stages of placental development. The aim was to further investigate the effects of nicotine and benzo(a)pyrene, two components of cigarette smoke previously suggested to have negative effects on both the placenta and fetus, on miRNA expression in placental cells. Downregulation of miR-146a in T cell leukemia/lymphoma 1 cell treated with nicotine and benzo(a)pyrene suggested that miR-146a may be especially sensitive to agents of cellular stress, such as prenatal smoke exposure. Moreover, this result suggests that two specific components of cigarette smoke that affect the expression of miR-146a in term placentas may be nicotine and benzo(a)pyrene (Maccani et al., Reference Maccani, Avissar-Whiting, Banister, McGonnigal, Padbury and Marsit2010).

However, Maccani et al.'s (2010) observations were limited by a relatively small set of samples (n = 25), as well as a lack of data regarding the duration of cigarette smoking during pregnancy, daily cigarette usage, or more extensive environmental exposure information (e.g., environmental pollutant exposure or second-hand/passive cigarette smoke exposure). Despite these limitations, their data comprise an important first step in determining associations between maternal cigarette smoking during pregnancy and aberrant miRNA expression in the placenta. More work is needed to investigate associations between maternal cigarette smoking during pregnancy and abnormal miRNA expression in a larger sample set with more complete environmental exposure information, as well as potentially using a hypothesis-generating approach (e.g., using miRNA microarrays; Maccani et al., Reference Maccani, Avissar-Whiting, Banister, McGonnigal, Padbury and Marsit2010; Maccani & Marsit, Reference Maccani and Marsit2011). Future work assessing associations between exposure-influenced placental miRNA profiles and downstream neurobehavioral and developmental outcomes is also needed. For instance, longitudinal analyses across the life span will be critical for elucidating the impact of miRNA-mediated fetal programming on later health, behavior, and disease.

As reviewed by Maccani and Marsit (Reference Maccani and Marsit2011), future experiments using cigarette smoke condensate (Crane-Godreau et al., Reference Crane-Godreau, Maccani, Eszterhas, Warner, Jukosky and Fiering2009) may enable researchers to better understand effects of complex mixtures of components of cigarette smoke on placental cells and cells derived from other tissues. Of course, such experiments will also present challenges: determining the specific composition of the complex mixture as well as the physiological relevance of exposure to such a mixture on a given cell type. Other more extensive experiments aimed at better understanding the more complex effects of organism-wide cigarette smoke exposure might be designed using mouse (or other animal) models of chronic environmental tobacco smoke exposure as described previously (Maccani & Marsit, Reference Maccani and Marsit2011; Xiong, Leme, Ray, Shapiro, & Lee, Reference Xiong, Leme, Ray, Shapiro and Lee2011). In such an experimental design, pregnant mice might be exposed to environmental tobacco smoke throughout their entire gestation and euthanized at key time points in gestation, with placentas and other key organs harvested for measurement of miRNA expression and for determination of the effects of dysregulated miRNA expression on target gene protein levels and function. The findings from such experiments would elaborate on the work done to characterize the association of maternal cigarette smoking during pregnancy with changes in placental miRNA and expand investigations into new tissues. In addition, such findings using an in vivo model system may provide crucial mechanistic data that may further strengthen data from previous epidemiologic association studies with findings from a model system experimental paradigm.

Integration of Epigenetic Characterization Into Existing Human Research Paradigms

There is a growing literature on maternal cigarette smoking during pregnancy and epigenetic phenomena, such as DNA methylation and miRNA expression. How do we align these recent advances with the overall literature on smoking during pregnancy, particularly the reported associations between smoking during pregnancy and later child/adolescent behavior? There is a large body of literature suggesting undesirable outcomes in children exposed to maternal smoking during pregnancy. This evidence (in human cohorts) has been limited until recently by researchers’ frequent inability to separate prenatal exposure effects from other confounding environmental and genetic factors. The majority of studies have provided limited control for prenatal exposures possibly being correlated with parental behaviors that could act as more proximal risk factors that are in turn transmitted to their offspring (Knopik, Reference Knopik2009). In other words, mothers who smoke during their pregnancies share other risk factors with their children and it may be these other risk factors that are associated with the observed adverse outcomes rather than only maternal cigarette smoking during pregnancy per se (Kuja-Halkola, D'Onofrio, Iliadou, Langstrom, & Lichtenstein, Reference Kuja-Halkola, D'Onofrio, Iliadou, Langstrom and Lichtenstein2010). These factors can include unmeasured familial environmental influences, such as second-hand smoke, socioeconomic status, and parental education. Additional factors might be genetic influences, such as genetic variation influencing (a) neurocognitive phenotypes through nonexposure contingent mechanisms, (b) pharmacokinetic and/or pharmacodynamic aspects of nicotinic effects, or (c) metabolism of the 4,000+ other xenobiotics (e.g., foreign substances) found in cigarette smoke. Epigenetic modification has long been considered a possible influence in this regard; however, this article indicates that it largely remains to be empirically tested when it comes to behavioral or neurodevelopmental phenotypes.

Because of ethical and practical concerns about research considering prenatal smoke exposure, maternal cigarette smoking during pregnancy in humans cannot be randomized. The consequence of this is that prenatal smoke exposure effects on human offspring outcomes are quasiexperimental in nature (Shadish, Cook, & Campbell, Reference Shadish, Cook and Campbell2002). However, innovative designs such as full sibling pairs discordant for prenatal exposure (an extension of the case-crossover design; D'Onofrio, Singh, Iliadou, Lambe, Hultman, Grann, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Grann2010; D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Neiderhiser2010; D'Onofrio et al., Reference D'Onofrio, Van Hulle, Waldman, Rodgers, Harden and Rathouz2008; Knopik, Reference Knopik2009; Knopik, McGeary, Nugent, Francazio, & Heath, Reference Knopik, McGeary, Nugent, Francazio and Heath2010; Kuja-Halkola et al., Reference Kuja-Halkola, D'Onofrio, Iliadou, Langstrom and Lichtenstein2010; Lumley & Levy, Reference Lumley and Levy2000; Meyer, Williams, Hernandez-Diaz, & Cnattingius, Reference Meyer, Williams, Hernandez-Diaz and Cnattingius2004) and in vitro fertilization cross-fostering approaches (Rice et al., Reference Rice, Harold, Boivin, Hay, van den Bree and Thapar2009, Reference Rice, Harold, Boivin, van den Bree, Hay and Thapar2010; Thapar et al., Reference Thapar, Harold, Rice, Ge, Boivin and Hay2007, Reference Thapar, Rice, Hay, Boivin, Langley and van den Bree2009) have enabled better controls for confounding factors. Results from these studies suggest that some associations suggested between maternal cigarette smoking during pregnancy and offspring behavioral outcomes, a common finding in the wider literature, may indeed be influenced by design limitations, specifically, an inability to adequately control for shared familial influences, including genetic factors. This has been shown for offspring conduct problems (D'Onofrio et al., Reference D'Onofrio, Van Hulle, Waldman, Rodgers, Harden and Rathouz2008), criminality (D'Onofrio, Singh, Iliadou, Lambe, Hultman, Grann, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Grann2010), academic achievement (D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Neiderhiser2010; Lambe et al., Reference Lambe, Hultman, Torrang, Maccabe and Cnattingius2006), intellectual performance (Lundberg et al., Reference Lundberg, Cnattingius, D'Onofrio, Altman, Lambe and Hultman2009) and ADHD (Knopik et al., Reference Knopik, McGeary, Nugent, Francazio and Heath2010; Thapar et al., Reference Thapar, Rice, Hay, Boivin, Langley and van den Bree2009). Thus, the suggestion that maternal cigarette smoking during pregnancy is exclusively responsible for a broad range of later offspring behavioral outcomes must be tempered and reconsidered in light of these recent findings across multiple research groups.

Moreover, preliminary efforts to study the fetal programming hypothesis, as described above, from a phenotypic standpoint (i.e., in the absence of epigenetic data), using low birth weight as a proxy for a poor gestational environment, suggest that this mechanism does not wholly explain potential prenatal smoke exposure influences on later psychological functioning (Kuja-Halkola et al., Reference Kuja-Halkola, D'Onofrio, Iliadou, Langstrom and Lichtenstein2010). Kuja-Halkola et al. (Reference Kuja-Halkola, D'Onofrio, Iliadou, Langstrom and Lichtenstein2010) also suggest that possible effects of low birth weight on later offspring behavior may also be due to familial confounding. Thus, further biological characterization, such as specific genetic influences, as well as epigenetic modification resulting from prepregnancy smoking, which has an influence regardless of the decision to smoke during pregnancy, may further highlight influences beyond maternal cigarette smoking during pregnancy and add clarity to this complicated story.

The conclusions of this body of quasiexperimental approaches to this research question should not be overinterpreted to suggest that maternal cigarette smoking during pregnancy is a benign influence. Substantial evidence exists to suggest prenatal cigarette smoke exposure is associated with numerous adverse outcomes related to birth and infancy (Cnattingius, Reference Cnattingius2004; D'Onofrio, Singh, Iliadou, Lambe, Hultman, Grann, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Grann2010; D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Neiderhiser2010; D'Onofrio et al., Reference D'Onofrio, Turkheimer, Eaves, Corey, Berg and Solaas2003; Gilman, Gardener, & Buka, Reference Gilman, Gardener and Buka2008; Johansson, Dickman, Kramer, & Cnattingius, Reference Johansson, Dickman, Kramer and Cnattingius2009; Rice et al., Reference Rice, Harold, Boivin, Hay, van den Bree and Thapar2009; Sexton & Hebel, Reference Sexton and Hebel1984). Thus, there is a need for continued smoking cessation efforts in pregnant mothers. However, given that smoking during pregnancy is correlated with a host of other behaviors (Agrawal et al., Reference Agrawal, Knopik, Pergadia, Waldron, Bucholz and Martin2008; D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, et al., Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman and Grann2010; Knopik, Reference Knopik2009; Knopik et al., Reference Knopik, Sparrow, Madden, Bucholz, Hudziak and Reich2005, Reference Knopik, Heath, Jacob, Slutske, Bucholz and Madden2006), treatment and prevention efforts might need to also consider other factors, such as maternal alcohol dependence, maternal nicotine dependence, and maternal ADHD.

Given the above considerations, the potential influence of maternal cigarette smoking during pregnancy on epigenetic mechanisms, which may have consequences for downstream behavior, is considerable. First, there is the possibility of intergenerational transmission of smoking-related epigenetic consequences (i.e., grandmaternal smoking influences may affect germline cells or may escape epigenetic reprogramming during development). Second, we can begin to understand the extent to which epigenetic modification may alter critical neurobehavioral circuitry in the developing brain. Third, questions regarding the influence of smoke exposure induced methylation on the xenobiotic metabolism of cigarette smoke by-products can begin to be answered. These are just a few of the currently unanswered questions for which epigenetic research might begin to provide answers. Fortunately, at least for DNA methylation, existing data sets with collected DNA samples may provide some opportunities to investigate such possibilities. For example, methylomic characterization of samples collected in sibling pairs discordant for prenatal exposure might provide initial evidence of longer lasting affects of prepregnancy smoking, which may make maternal smoking during pregnancy per se less important than the overall maternal smoking history. Use of animal models in mutually informative translational research (especially in inbred lines where the genetic background is held constant) may further strengthen our ability as a field to investigate these tantalizing hypotheses.

Preliminary evidence already points to important genetic differences in xenobiotic metabolism with downstream effects on intrauterine growth (Price, Grosser, Plomin, & Jaffee, Reference Price, Grosser, Plomin and Jaffee2010) and behavior (Knopik, McGeary, Nugent, Francazio, & Heath, in press). Further characterization of these samples from an epigenetic standpoint could explain additional variance in behavioral outcomes while simultaneously identifying potentially targetable mechanisms to reduce such effects.