Asthma is a major public health problem 12. It is the most common chronic disease of childhood 34, resulting in a significant medical burden and the resultant healthcare costs 56. The burden of this disease is increasing rapidly, with the most striking increases seen among children, the prevalence rates in some populations being greater than 30% 7. Although a multitude of causes contribute to childhood asthma, maternal smoking during pregnancy is a well-established contributor 8910111213, and it is a major modifiable risk factor, the elimination of which could significantly reduce the prevalence of childhood asthma. However, given that the tobacco industry continues to spend billions of dollars on advertising to attract young smokers, it is unlikely that the problem of smoking during pregnancy will go away any time soon 14. Approximately 250,000,000 women smoke daily world-wide. Twelve percent of US women still continue to smoke during pregnancy, resulting in the births of at least 400,000 smoke-exposed infants per year in the US 1516. This aspect of the smoking-induced asthma etiology is particularly important since there is emerging evidence that, following in utero exposure to maternal smoke, asthma can be transmitted multigenerationally. For example, a questionnaire in the Children's Health Study from Southern California reported that grandmaternal smoking during pregnancy increases the risk of asthma in grandchildren regardless of the presence of maternal smoking 17. Yet there is neither experimental evidence nor any mechanistic explanation for this phenomenon. Using a well-established rat model of in utero nicotine exposure for childhood asthma 181920, we aimed to determine if in utero nicotine exposure would transmit asthma to the second generation offspring, and if epigenetic mechanism(s) could be involved in this transmission.

The mechanism linking smoke/nicotine exposure during pregnancy to multigenerational (MG) transmission of asthma is unknown. Since transmission via the germline is the most likely explanation for the non-genetic MG transmission phenomenon, we rationalized that epigenetic modifications of the germline explain MG transmission of perinatal nicotine-induced asthma. For example, a change in the organism's environment, that is, smoke/nicotine exposure (of the F0 generation), can result in modifications in gene expression through alterations in DNA methylation, histone (H) modification, non-coding RNA, and/or protein structure and assembly without changing the DNA sequence of the F1 germline, which can be transmitted to subsequent generations 2122. However, whether epigenetic marks such as alterations in DNA methylation and/or H modifications acquired in one generation can be inherited by the next generation is not clear. While one study reported that the acquired DNA methylation marks are not transmitted to subsequent generations 23, another study came to the opposite conclusion 24. We hypothesize that smoke/nicotine-induced epigenetic marks can become permanently programmed and transferred through the germline to subsequent generations, resulting in altered phenotypic changes at the cellular and consequent organismal levels (for example, asthma in the case of in utero smoke/nicotine exposure) in the offspring over multiple subsequent generations, for example, F2, F3, and so on.

Furthermore, the effect of the child's gender on asthma risk following perinatal exposure to cigarette smoke is not clear since an increased risk in both boys 25 and girls 26 has previously been reported. However, since there is emerging evidence to suggest that finely-tuned developmental programs, such as that of the lung, may be sensitive to specific environmental challenges in a sex-specific manner, particularly during the developmental programming and gametogenesis stages 27, and since maternal smoking is known to affect fetal growth more profoundly in the male fetus 28, we also hypothesized that there would be a sexual dimorphism in asthma risk following in utero smoke exposure, with males being more susceptible than females.

We initially determined the effect of nicotine on pulmonary function in response to a graded methacholine challenge. Compared to the control group, with perinatal nicotine exposure to F0 dams (Figure 1) there was a significant increase in total airway Rrs, and a decrease in total airway Cdyn of the respiratory system, not only in F1 rats 20, but also in F2 rats (P < 0.01 versus control for both Rrs and Cdyn), even though the F2 rats were not exposed to nicotine. In accord with our recently published data on F1 rats 20, RGZ treatment blocked the nicotine-induced increase in Rrs and decrease in Cdyn in F2 rats (P < 0.01 versus nicotine for both Rrs and Cdyn) (Figure 2).

We subsequently analyzed the effects of nicotine exposure on PFT's in male and female offspring separately (Figure 3a and 3b). Following a methacholine challenge, perinatal nicotine exposure only to F0 dams caused a significant increase in Rrs and a decrease in Cdyn in both the males and females of the F1 and F2 generations (P < 0.05, males versus females, for both Rrs and Cdyn).

Having determined the effect of in utero nicotine exposure on the F1 and F2 offspring whole respiratory systems, we examined the effect of nicotine on tracheal rings isolated from F1 and F2 offspring (Figure 4). Nicotine administration caused a significant increase in tracheal constriction in response to acetylcholine only in the males of both the F1 and F2 generations (P < 0.01 versus control).

Since we had previously determined that the nicotine exposure of F0 dams increased the expression of lung contractile proteins and inhibited PPARγ expression in both the whole lung and isolated alveolar interstitial fibroblasts of F1 rats 20, we next determined this effect in the F2 offspring. We found that the protein levels of fibronectin, α-SMA, calponin, and collagens I and III, and nicotinic acetylcholine receptors α3 and α7 were all increased significantly in the whole lung lysates of both the male (Figure 5a and 5b) and female F2 (P < 0.05 versus control for all) (Figure 6a and 6b) rats. In contrast, at the tracheal level the expression of fibronectin, α-SMA, calponin, collagen I, nicotinic acetylcholine receptors (nAChR) α3 and α7

were increased only in males

To determine the likely mechanism of nicotine-induced transmission of asthma to the second generation offspring, we next determined the effect of F0 nicotine exposure on global DNA methylation and histone acetylation in the lungs and gonads of the F1 rats. We observed that with nicotine administration global DNA methylation increased in the testes (P < 0.01), decreased in the ovaries (p < 0.05), but did not change in the lungs (Figure 9a); H3 acetylation increased in the lungs (P < 0.01) and testes (P < 0.05), but did not change in the ovaries (Figure 9b); H4 acetylation decreased in the lungs (P < 0.01) while it increased in the testes (P < 0.01) and ovaries (P < 0.05) (Figure 9c). Further, we determined if the nicotine effects on histone methylation and acetylation were affected by RGZ, which has been shown to normalize the structural and functional effects of nicotine on the offspring lung phenotype at F1 20. The nicotine-induced increase in lung H3 acetylation was blocked (Figure 9b) by RGZ treatment, whereas there was no effect on global DNA methylation (Figure 9a) or on the nicotine-induced decrease in H4 acetylation (Figure 9c), suggesting a central role of H3 acetylation in mediating nicotine's effects on the pulmonary phenotype. Finally, it is reassuring to note that perinatal exposure of the gonads to RGZ by itself did not result in any significant changes in global gonadal DNA methylation or in H3 and H4 acetylation (Figure 10).

In the present series of experiments we have observed significant effects of nicotine treatment on lung function in the next two generations, affecting both the male and female offspring. In sharp contrast, nicotine treatment only affected the tracheal contractility of the male offspring. The functional effects of nicotine on the naive offspring were accompanied by increased expression of contractile proteins in the whole lung, as well as in the associated isolated lung fibroblasts, accompanied by decreased PPARγ expression. These nicotine-induced changes in lung function and mesenchymal protein expression, accompanied by decreased PPARγ expression, are consistent with the effect of nicotine on myofibroblast differentiation 32. Even more importantly, along with the normalization of the asthma phenotype in the F1 and F2 offspring, most of the nicotine-induced lung and gonadal epigenetic changes were also normalized. For example, nicotine-induced increases in H3 acetylation in the lung, DNA methylation and H4 acetylation in the testis, and the decrease in DNA methylation and increase in H4 acetylation in the ovaries of F1 offspring were normalized by RGZ treatment, but it had no effect on lung H4 acetylation, providing further mechanistic specificity regarding the nature of the epigenetic mechanism. Given these insights, we will extend these studies to F3 and F4 generation offspring in future studies.

Since ruling out genetic and environmental confounders is extremely difficult in humans, there is only scant evidence for MG epigenetic effects for any condition in humans 33; and in fact, there is none for asthma. Recently emerging evidence suggests that the phenotype of an individual is the cumulative result of complex interactions between the genotype and its current, past and ancestral environments 2734. Therefore, it is logical to speculate a role for ancestral cigarette smoke exposure in the child's asthma predisposition. Yet, as pointed out above, there is no evidence for this, other than the data from the Children's Health Study from Southern California 17.

In contrast to other species, the evidence for fetal programming as a mode of MG transmission of traits in humans is very limited. For example, mothers from the Dutch Hunger Winter who were exposed to famine as fetuses delivered offspring of lower birth weights than those with no fetal exposure to famine, although this was not confirmed in a subsequent study 35. There is also evidence of increased morbidity and mortality associated with parental and grandparental nutritional status, suggesting a role for fetal programming, possibly via epigenetic mechanisms to account for the MG effects 213637. In contrast to the very limited data in humans, in a variety of animal models gestational exposure to carcinogens, endocrine disruptors, and other toxins has been shown to have MG effects 383940.

The present study is groundbreaking in our understanding of the mechanisms potentially involved in the transmission of epigenetic human diseases, which to date have only been speculated, albeit based on strong epidemiologic grounds 41. The observation that nicotine exposure in utero affects the F1 offspring DNA methylation of the testes and ovaries, and that H3 acetylation was increased in the lungs and testes of F1 male offspring in association with male-specific increased tracheal constriction is the first demonstration of an epigenetic effect of nicotine on both gametocytes and somatocytes. Furthermore, the specific inhibition of this effect of nicotine on H3 acetylation by RGZ, which we have previously shown to block the nicotine-induced asthma phenotype in F1 offspring 20, provides a unique molecular genetic insight to the MG mechanism of nicotine action. It must be borne in mind, however, that these offspring's developing gonads had been exposed to nicotine in utero, leaving open the possibility that the effect was not a bona fide transgenerational effect, but was rather only MG. However, the transmission of the asthma phenotype to the F2 generation, both structurally and functionally, and its prevention by a specifically-targeted molecular intervention is the first unequivocal demonstration of MG transmission of an epigenetically-mediated effect on the offspring phenotype. The fact that the phenotypic effect was on asthma, a well-recognized epidemiologic example of epigenetic transmission of the cause of a public health epidemic, makes this series of experiments all the more significant and noteworthy. This, and the recent finding that even 'thirdhand smoke' can induce the asthma phenotype 42, portends new and rational ways of thinking about effectively coping with the health hazards that abound all around us 4344.

At first glance, it might seem surprising that the nicotine effect on the asthma phenotype is sex-specific. However, there is a documented association between gender and airway size, first referred to as dysanapsis by Mead 45, in which he showed that boys and women had self-similar airway structure that was distinctly different from that of men. Those data suggest that androgens may differentially affect airway development. We have previously shown that androgens affect the rate of lung development in rabbits and rats by inhibiting the glucocorticoid-induced differentiation of lung fibroblasts 46, consistent with their effect on lipofibroblast differentiation 47, which determines lung development 48. The airway narrowing of dysanapsis has also been shown to be associated with asthma 49. Therefore, androgens may precipitate asthma through a common genetic mechanism, since they, like nicotine, stimulate the Wnt pathway 50 and down-regulate PPARγ expression in lung fibroblasts 1819203132.

The compelling epigenetic data presented here potentially shift the current paradigm for our understanding of childhood asthma, and for the first time implicate epigenetics as the underlying cause for increased MG asthma following in utero exposure to maternal smoking. In fact, the data provided herein not only provide novel mechanistic information underlying the MG asthma risk, but also pave the way for studying the molecular mechanisms underlying MG effects for a host of other environmental toxins, agonists and antagonists as well.

It is important to point out that although there are many agents in cigarette smoke that may be detrimental to the developing lung, there is plenty of evidence to support nicotine as the main agent that alters fetal lung development: nicotine crosses the human placenta with minimal biotransformation 51; it accumulates in fetal blood, maternal milk, amniotic fluid, and several fetal tissues, including the respiratory tract 525354, and has been shown to have direct effects on pulmonary alveolar epithelial cells and interstitial fibroblasts isolated from the developing lung 3243555657585960. Therefore, it is not surprising that nicotine exposure during pregnancy is an extensively utilized and well-accepted model to study the effects of cigarette smoke on the developing lung in general, and on asthma in particular 1820. It is also important to emphasize that the main effects of in utero nicotine exposure on lung growth and development are due to specific alterations in signaling pathways involved in lung development 18203132555657585960, rather than being due to irreversible disruption by teratogenic or toxicological effects. This feature offers the opportunity for targeted interventions to modulate the effects of in utero nicotine exposure, in contrast to toxic or teratologic effects, which would be unlikely to be effectively blocked by any intervention.

It could be argued that the epigenetic marks in the gonads and lung are epiphenomena, and that the RGZ 'rescue' of the normal lung phenotype is an artifact. However, there is experimental evidence that methylation of H3/H4 results in down-regulation of PPARγ expression 61, mechanistically linking the epigenetic effect of nicotine with decreased lipofibroblast expression 18 and asthma 20.

From the data included here we conclude that the pulmonary effects of nicotine exposure during pregnancy are not restricted only to the offspring of the exposed pregnancy, but are also transmitted to subsequent generation, possibly through germline epigenetic alterations, and, even more importantly, these effects can be blocked by targeted molecular interventions. Moreover, these data not only provide novel mechanistic information underlying the multigeneration transmission of asthma risk following exposure to maternal smoke during pregnancy, but also set precedence for studying other such environmental toxins that might have multigenerational or transgenerational effects.

BSA: bovine serum albumin; Cdyn: dynamic compliance; F0, F1, F2: parent, first, and second generations; FBS: fetal bovine serum; HAT: histone acetyl transferase; HBSS: Hanks' balanced salt solution; HDAC: histone deacetylase; H3, H4: histones 3 and 4; MEM: minimal essential medium; MG: multigenerational; PFT: pulmonary function test; PND: postnatal day; PPARγ: peroxisome proliferator activated receptor gamma; RGZ: rosiglitazone maleate; Rrs: lung resistance; RT-PCR: reverse transcriptase-polymerase chain reaction; α-SMA: α-smooth muscle actin.

The authors declare that they have no competing interests.

VKR conceived the study, participated in study design, helped and coordinated the acquisition of data and its interpretation, and wrote the first draft the manuscript. JL carried out the animal work, molecular and pulmonary function studies, and made the data figures. EN and JT helped with DNA methylation and histone acetylation studies. RS performed fibroblast isolation and RT-PCR studies. KK helped with pulmonary function studies. OA supervised pulmonary function studies and the interpretation of pulmonary function data. JST helped in data interpretation, study design, and manuscript writing and revising. All authors read and approved the final manuscript.