RUNX transcription factors: association with pediatric asthma and modulated by maternal smoking

Abstract

Intrauterine smoke exposure (IUS) is a strong risk factor for development of airways responsiveness and asthma in childhood. Runt-related transcription factors (RUNX1–3) have critical roles in immune system development and function. We hypothesized that genetic variations in RUNX1 would be associated with airway responsiveness in asthmatic children and that this association would be modified by IUS. Family-based association testing analysis in the Childhood Asthma Management Program genome-wide genotype data showed that 17 of 100 RUNX1 single-nucleotide polymorphisms (SNPs) were significantly (P < 0.03–0.04) associated with methacholine responsiveness. The association between methacholine responsiveness and one of the SNPs was significantly modified by a history of IUS exposure. Quantitative PCR analysis of immature human lung tissue with and without IUS suggested that IUS increased RUNX1 expression at the pseudoglandular stage of lung development. We examined these associations by subjecting murine neonatal lung tissue with and without IUS to quantitative PCR (N = 4–14 per group). Our murine model showed that IUS decreased RUNX expression at postnatal days (P)3 and P5 (P < 0.05). We conclude that 1) SNPs in RUNX1 are associated with airway responsiveness in asthmatic children and these associations are modified by IUS exposure, 2) IUS tended to increase the expression of RUNX1 in early human development, and 3) a murine IUS model showed that the effects of developmental cigarette smoke exposure persisted for at least 2 wk after birth. We speculate that IUS exposure-altered expression of RUNX transcription factors increases the risk of asthma in children with IUS exposure.

prenatal smoking remains among the most common preventable causes of infant morbidity and mortality. In utero smoke exposure (IUS) can lead to alterations in normal fetal lung development with subsequent childhood impacts such as lower pulmonary function, early-onset wheezing, and asthma (46, 55, 70), the leading cause of childhood hospitalizations. A mechanism for the effect of IUS on asthma susceptibility may be via its potential to increase airway remodeling (6, 16, 57), leading to increased airway responsiveness, possibly via decreases in lung cAMP levels (60), upregulation of the α7 nicotinic receptor (57), or oxidative stress (56). This model is supported by the multiple pediatric cohort studies that have reported the association of IUS with subsequent increases in airway responsiveness (20, 21, 63, 73).

The “fetal origins” or “developmental origins” hypothesis, first popularized by Barker and colleagues (5), proposes that specific in utero exposures at critical periods during organogenesis and maturation result in long-term physiological or metabolic changes, ultimately contributing to the development of disease in later life. Although Barker's original hypothesis focused on the consequences of maternal nutrition, there is now substantial evidence that other in utero exposures, including IUS, have similar effects (64).

The runt-related proteins (RUNX) are highly conserved transcription factors that modulate the embryological development and/or postnatal function of the nervous system, bone and cartilage, epithelial cells, angiogenesis, hematopoiesis, and immune responses (4). Three mammalian RUNX homologs have been identified thus far: RUNX1 (AML1/CBFPEBP2αB/CBFA2), RUNX2 (AML3/PEBP2αA/CBFA1), and RUNX3 (AML3/PEBP2αC/CBFA3) (4). Although all of the RUNX proteins form a heterodimer with a common binding protein, CBFa2, knockout mouse models have shown that each of the RUNX proteins have distinct roles during fetal development (4, 18, 37–38). Complete RUNX1 deficiency is embryonic lethal because of ineffective hematopoiesis (52, 68), whereas lack of RUNX3 causes more subtle leukocyte abnormalities with defective thymic development and dendritic cell hyperresponsiveness, causing spontaneous pulmonary eosinophilic infiltrates (18). These mice additionally have hyperplastic gastric epithelium, a premalignant lesion (41). In contrast, RUNX2-deficient mice have unremarkable hematopoiesis but demonstrate severe skeletal abnormalities (37, 54).

RUNX1 and RUNX3 mediate the normal maturation and function of several components of the postnatal immune system. Similar to the fetal requirement for RUNX1 for the initiation of hematopoiesis, RUNX1 is required for postnatal hematopoietic stem cell function: RUNX1-deficient bone marrow is unable to rescue lethally irradiated mice (50). Additionally, RUNX1 and RUNX3 both promote Th1 cell development from naive T helper by activating the IL-4 silencer, which leads to decreased IL-4 and increased interferon-γ expression in these cells (49). RUNX1 also modulates IL-17 expression in T helper cells in a context-dependent manner. In CD4+ cells, RUNX1 acts synergistically with RORγt to increase IL-17 expression (75). This increase is blocked by T-bet binding of RUNX1 (39). In contrast, in T regulatory cells, RUNX1 binds with FoxP3, resulting in decreased IL-17 and IL-2 expression (75). Recent studies show that RUNX3 activation depends on T-bet and that RUNX3 modulates interferon-γ expression in CD4 cells (12–13). Thus dysregulation of either RUNX1 or RUNX3 could promote increased Th17 and/or Th2 immune responses, which would favor an asthmalike phenotype.

Since RUNX1 is important in normal immune system development and function, several laboratories have conducted epidemiological studies to determine whether genetic variation in this transcription factor is associated with human disease. The association between RUNX1 and malignancies, particularly hematological malignancies such as acute myelogenous leukemia and lymphoma, was the first link between the RUNX transcription factors and human disease (34, 44–45). More recently, investigations have linked RUNX1 polymorphisms with multiple autoimmune diseases, including insulin-dependent (Type 1) diabetes mellitus, systemic lupus erythematosus (SLE), psoriasis, and rheumatoid arthritis (RA) (61, 66, 72).

Given the historical association of IUS with asthma and airway responsiveness as well as the roles of the RUNX transcription factors in normal immune function and their association with autoimmune processes, we hypothesized that genetic variation in the RUNX transcription factors would be associated with airway responsiveness in asthma and that this effect might be modified via IUS exposure. We focused on airway responsiveness since it has been demonstrated to be increased in children and young adults following IUS in humans and murine models (11, 20, 73). We tested this hypothesis in a population of asthmatic children participating in the Childhood Asthma Management Program (CAMP). To investigate whether IUS caused RUNX transcription factor abnormalities in a developmental context, we evaluated the effect of IUS on RUNX1 expression in the developing human and murine lung.

METHODS

Study protocols were approved by the Partners Human Research Committee and the Harvard Medical School Institutional Animal Care and Use Committee. Subjects gave written, informed consent for use of discarded surgical material and completion of a questionnaire regarding cigarette usage. For the CAMP clinical trial, each parent or legal guardian provided consent, with each child proband giving assent for the clinical trial; separate consent/assent was obtained for the genetics ancillary study in CAMP.

Animals.

Study animals were housed in pathogen-free conditions, with ad libitum food and water. Daily exposure of female C57Bl/6 mice (Charles River Laboratories, Waltham, MA) to the smoke of three 3R4F research cigarettes (University of Kentucky), five times per week, started at age 10 wk. Lung tissue samples from the offspring were collected at embryonic days (e) 15.5 and 17.5, the day of birth (P0), and postnatal day 3 (P3), P5, P7, P10, and P14.

Real-time qPCR.

RNA extraction was performed by using TriReagent (Molecular Research Center) according to manufacturer's instructions and was followed by reverse transcription to cDNA (Retroscript, Ambion/Applied Biosystems) as previously described (23). Two lung tissue samples were pooled for each N from the e15.5 samples. Single lung tissue samples were analyzed for all other time points. Real-time polymerase chain reaction (qPCR) using validated Taqman primers and probes (Applied Biosystems) was performed as previously described on murine and developing human lung tissue samples (23). The RUNX and T-bet primers (catalog numbers: human RUNX1 Hs 01021971_m1, murine RUNX1 Mm 01213405_m1, murine RUNX2 Mm 00501584_m1, murine RUNX3 Mm 00490666_m1, and murine T-bet Mm 00450960_m1) spanned an intron and normalized to the endogenous control, 18S (Hs 99999901_s1).

Immunostaining.

Immunostaining was carried out as previously described (23) on paraffin-embedded lung tissue sections by use of murine monoclonal anti-RUNX1 (Santa Cruz, at 10 μg/ml) and rabbit polyclonal anti-cytokeratin (used at 11 μg/ml, Dako Cytomation). The negative control antiserum was irrelevant IgG2b. Double staining to identify cell types immunopositive for RUNX1 was carried out as previously described (23), using rabbit polyclonal anti-cytokeratin to stain epithelial cells (used at 11 μg/ml, Dako Cytomation). Hematoxylin and eosin staining using standard protocols was performed to assess tissue architecture for 14 developing human lung tissue samples with a history of maternal smoking and 21 without a history of maternal smoking.

CAMP genotype analyses.

CAMP was a multicenter, randomized, double-blinded clinical trial. Trial design and methodology have been published (1). Information regarding maternal smoking was obtained by using self-report on a validated questionnaire. We examined the provocative concentration of methacholine required to cause a 20% decrement in forced expiratory volume in 1 s (FEV₁) (FEV₁PC₂₀), as measured during the run-in period of the clinical trial, a period during which no asthma controller medications were taken. The FEV₁PC₂₀ was chosen since this measurement has been shown to be affected by intrauterine cigarette smoke exposure (11).

Genomewide SNP genotyping was performed in the CAMP sample by Illumina (San Diego, CA) on the HumanHap550 v3 BeadChip as detailed previously (26). All markers had greater than 90% genotyping completion rate, with the average completion rate for each marker over 99%. From 561,466 markers present on the BeadChip, 547,645 markers (97.54%) passed all quality control metrics (26). No filtering was done based on Hardy-Weinberg equilibrium because of ascertainment of the cohort through affected probands.

Statistical methods.

Expression differences between groups were analyzed by Student's t-test for normally distributed data. Data that were not normally distributed were analyzed by nonparametric statistical tests (Mann-Whitney). Association of RUNX1, RUNX2, and RUNX3 SNPs with FEV₁PC₂₀ were analyzed by family-based association testing (FBAT), a generalization of the transmission disequilibrium test that allows for valid testing of association with any phenotype, sampling structure, and pattern of missing marker allele information (30). We focused on nominal levels of association (P < 0.05) using an additive genetic model. Each SNP meeting these criteria was then evaluated for significant interaction with IUS (binary variable), via a gene-environment interaction analysis in FBAT.

RESULTS

RUNX1 is associated with airway hyperresponsiveness in pediatric asthma.

Table 1 demonstrates the characteristics of the CAMP population evaluated in our genetic association study. Just over 10% of the genotyped subjects were reported as exposed to intrauterine smoke. At baseline, there were no significant differences between the smoke-exposed and non-smoke-exposed probands in regard to age, sex, duration of asthma, baseline lung function (FEV₁), or baseline airways responsiveness (FEV₁PC₂₀) (Table 1). DNA samples from 422 family trios yielded completed genotypes in 1,169 subjects including 403 CAMP probands and their parents. The average genotyping completion rate for each subject was 99.75%. From the genome-wide association SNP dataset, we extracted the SNPs located on chr21:35,076,968–35,348,465 (HG 18), a region encompassing the human RUNX1 gene and 10 kb, both 5′ and 3′ of the gene; 100 SNPs from the HumanHap550v3 Beadchip mapped to this region.

Table 1.

Comparison of CAMP subjects with and without a history of intrauterine smoke exposure

	Intrauterine Smoke Exposed	No Smoke Exposure	P Value
N	44	378
Age	8.6 ± 2.2	8.8 ± 2.1	NS
Number Female Sex (%)	15 (34%)	141 (37%)	NS
Duration of Asthma, yr	5.6 ± 2.4	5.7 ± 2.6	NS
Baseline FEV1%	94.1 ± 13.4	93.3 ± 14.0	NS
Baseline log FEV1PC2020	0.04 ± 1.05	0.02 ± 1.17	NS

Values are means ± SD;

N = number of subjects. CAMP, Childhood Asthma Management Program; FEV₁, forced expiratory volume in 1 s; FEV₁PC₂₀, dose of methacholine causing a 20% decrease in the FEV₁; NS, not significant.

Of the 100 RUNX1 SNPs examined, 17 were nominally associated with FEV₁PC₂₀ under the additive model, with P values ranging from 0.03 to 0.04 (Table 2). Analysis of linkage disequilibrium patterns of the 17 SNPs revealed only one pairing with significant linkage disequilibrium (r² ≥ 0.80), suggesting that these associations are independent or in linkage disequilibrium with an additional, untyped, RUNX1 variant. In contrast, an examination of 11 RUNX3 SNPs did not show any association with the FEV₁PC₂₀ (data not shown). Examination of 40 RUNX2 SNPs showed two SNPs that were associated with the FEV₁PC₂₀, but these associations were not modified by IUS (data not shown).

Table 2.

Initial significant RUNX1 SNP associations with FEV₁PC₂₀ in CAMP

SNP	Number of Informative Families	Chromosome	Position on Chromosome 21	Minor Allele Frequency	Association P Value
rs2070370	157	21	35079567	0.39	0.0097
rs9976946	17	21	35129931	0.04	0.016
rs2834651	132	21	35150229	0.33	0.011
rs9979015	87	21	35157437	0.22	0.012
rs2298352	89	21	35174491	0.22	0.025
rs2834662	99	21	35196045	0.25	0.043
rs2834670	71	21	35202245	0.18	0.041
rs2834684	138	21	35234216	0.34	0.028
rs9979153	17	21	35245376	0.04	0.023
rs9976122	19	21	35260175	0.05	0.033
rs1981392	164	21	35275529	0.41	0.036
rs2834714	141	21	35279221	0.35	0.022
rs9636887	75	21	35285971	0.18	0.013
rs11088302	195	21	35286031	0.48	0.031
rs11702779	144	21	35290528	0.35	0.040
rs9976900	172	21	35306232	0.43	0.040
rs2834736	136	21	35312810	0.34	0.0032

The single-nucleotide polymorphism (SNP) associated with airway hyperresponsiveness and modified by maternal smoking, rs11702779, is highlighted in boldface.

Given the consistent identification of maternal smoking as a risk factor for pediatric asthma, we next tested the 17 RUNX1 SNPs for significant interaction with IUS. This analysis showed that the association between airways hyperresponsiveness, and one SNP, rs11702779, was significantly modified by IUS (interaction P = 0.0009) (Fig. 1). This SNP, located in a noncoding region, results in an AA genotype instead of AG or GG. Compared with the major alleles, the AA genotype was associated with increased airway responsiveness (lower FEV₁PC₂₀) in children without tobacco exposure. In contrast, whereas children with the AA genotype who were exposed to cigarette smoke also demonstrated increased in airway responsiveness compared with tobacco-naive AA children, this increase was less marked than that observed in smoke-exposed children with the AG or GG genotypes (Fig. 1). The effects of the AA genotype on the degree of airway hyperresponsiveness compared with the wild-type AG or GG genotypes depended on the presence or absence of an environmental exposure, cigarette smoke. Thus the effects of the AA genotype on airway hyperresponsiveness showed a gene-by-environment interaction.

Fig. 1.

Genotypic analysis of CAMP genome-wide database showing that intrauterine tobacco smoke exposure alters the association between RUNX1 single-nucleotide polymorphisms (SNPs) and airway responsiveness, as assessed by methacholine responsiveness [provocative concentration of methacholine required to cause a 20% decrement (PC₂₀) in forced expiratory volume in 1 s (FEV₁) (FEV₁PC₂₀)]. In children without a history of intrauterine tobacco smoke exposure (no smoke), both the AG and GG SNPs are correlated with greater doses of methacholine required to cause a 20% decrease in FEV₁, consistent with protection against airway hyperresponsiveness. However, in children with a history of intrauterine tobacco smoke exposure (smoke), these wild-type genotypes are associated with greater sensitivity to methacholine, as shown by the decreased log FEV₁PC₂₀ in these subjects. In contrast, children with the AA minor allele SNP have heightened sensitivity to methacholine regardless of maternal smoking history, with the subjects who were positive for intrauterine tobacco smoke exposure showing greater sensitivity to methacholine than the subjects without prenatal tobacco smoke exposure. Thus, in the “no smoke” condition, the minor allele is associated with greater sensitivity to methacholine, but in the “smoke” condition the minor allele is somewhat protective regarding airway hyperresponsiveness. The effects of the minor allele SNP thus vary with the environmental exposure to tobacco smoke.

RUNX1 expression has been investigated in embryonic murine lung but has not been well studied in developing human lung. We therefore used qPCR analysis to examine the expression of mRNA for RUNX1 in 20 samples of developing human lung (gestational ages between 10 and 22 wk) obtained from discarded surgical material. Information regarding maternal cigarette smoking was obtained by maternal questionnaire. This analysis demonstrated that, similar to murine lung, RUNX1 is expressed in the human lung at both the pseudoglandular and canalicular stages of lung development. In the pseudoglandular stage (N = 3 without and 3 with IUS), gestational age 10–15 wk, maternal smoking was associated with a slight trend toward greater RUNX1 mRNA expression compared with tobacco-naive samples (Fig. 2).

Fig. 2.

Quantitative PCR analysis of developing human lung, normalized for 18S expression, showing expression of the mRNA for RUNX1 at both the pseudoglandular (N = 6, 3 each with and without intrauterine cigarette smoke exposure) and the canalicular (N = 14, 7 each with and without cigarette smoke exposure) stages of development. At the pseudoglandular stage, RUNX1 expression showed a slight trend toward increased expression in the samples with intrauterine tobacco smoke exposure. In the canalicular stage, intrauterine tobacco smoke exposure did not change the expression of RUNX1. Line indicates median normalized expression. Ps, pseudoglandular stage; Can; canalicular stage; NS, no maternal smoking history; IUS, positive maternal smoking history, resulting in intrauterine smoke by-product exposure.

Not all mRNAs expressed during fetal lung development are transcribed to protein (71). Additionally, maternal smoking was not associated with inflammatory infiltrates in developing human lung tissue samples (N = 21 without a history of maternal smoking, N = 14 with a history of maternal smoking) analyzed by H&E staining (Fig. 3, A and B). The lack of substantial inflammatory cell infiltrates raised the possibility that RUNX protein might not be transcribed in the fetal lung. Therefore, we next evaluated whether RUNX1 protein could be detected in paraffin-embedded developing human lung tissue samples. Immunostaining demonstrated cells immunopositive for RUNX1 at both the pseudoglandular and canalicular stages of lung development (n = 22), with immunopositive cells located predominantly in cartilage and epithelium. No staining was observed following substituting an irrelevant IgG for the primary antibody (Fig. 4, A and B). Double staining with the epithelial cell marker cytokeratins confirmed the expression of RUNX1 in these cells (Fig. 4, C–F). This pattern of RUNX1 expression was similar to that observed in murine lung development (40, 42).

Fig. 3.

Representative hematoxylin and eosin staining of developing human lung, gestational age 20–22 wk (canalicular stage; A). Maternal smoking is not associated with lung inflammatory cell infiltrates in the canalicular stage of lung development (B). Scale bar indicates 100 μm.

Fig. 4.

Immunoperoxidase staining of developing human lung showing abundant RUNX1 immunopositivity (A) in the cartilage (short arrows) and airway epithelium (long arrows). No staining was detected when an irrelevant IgG2b was substituted for the primary antisera (B). Double staining (C) for RUNX1 and the epithelial cell marker cytokeratins shows double-positive immunopositive epithelial cells (arrows) with only RUNX1-immunopositive cells in the cartilage (curved arrows) with high-power view of representative double staining in epithelial cells shown in D. Single-stain components of the double stain are shown in E and F. E shows RUNX1-immunopositive cells identified by using alkaline phosphatase chromogen, and F shows cells immunopositive for cytokeratins detected by a peroxidase chromogen, with immunopositive cells indicated by arrows. Scale bar indicating 100 μm shown for A–F.

Maternal smoking alters RUNX transcription factor expression in a murine model of prenatal cigarette smoke.

Our human studies raised the possibility that a history of maternal smoking was associated with a trend toward increased RUNX1 mRNA expression in developing lung in at least the pseudoglandular stage of lung development. However, these studies did not have the statistical power to fully analyze this question. Additionally, our human studies did not determine whether any relationship between maternal smoking and increased RUNX1 was causal. To address these questions, we studied the expression of the RUNX transcription factors in a murine model of prenatal cigarette smoke exposure. We adapted a well-established adult murine model (24) of mainstream cigarette smoke exposure to permit us to study the developmental effects of maternal smoking on the lungs of the offspring. The inhaled particulates from our model are 310.6 mg/m³, which is comparable with murine cigarette smoke exposure models used by other investigators studying mainstream cigarette smoke exposure in adult mice (60).

We tested the hypothesis that maternal smoking causes abnormalities in the RUNX transcription factors by comparing the expression of mRNA in lung tissue from mice with and without IUS. RNA was extracted from whole lung tissue samples from e15.5, e17.5, P0, P3, P5, P7, P10, and P14; n = 4–14 for each group at each time point. Each time point included mice from at least two litters. The RNA was reversed transcribed to cDNA and subjected to qPCR analysis. 18s expression was used to normalize mRNA levels in the samples; 18s was used since it is not regulated during lung development. This analysis showed that immature murine lung expresses the mRNA encoding the three RUNX transcription factors at all time points examined. Similar to the findings in developing human lung, there was a trend toward increased RUNX1 expression in the cigarette smoke-exposed samples at e17.5. IUS significantly decreased RUNX1 mRNA expression in the alveolar stage of lung development, at P3 and P5; trends toward decreased expression were noted at P0, P7, and P10 (Table 3).

Table 3.

Quantitative PCR analysis of RUNX1 transcription factor in developing murine lung with and without prenatal tobacco smoke exposure

	Tobacco Naive		Intrauterine Tobacco Smoke Exposure
Age	N	Fold Change mRNA Expression	N	Fold Change mRNA Expression	P Values
E15.5	4	1.5 (0.2)	5	1.4 (0.1)	NS
E17.5	12	1.9 (0.2)	9	2.3 (0.2)	NS
P0	13	5.6 (0.8)	14	4.2 (0.6)	NS
P3	4	13.9 (1.9)	5	3.6 (1.2)	P = 0.002*
P5	5	11.8 (8.2–14.9)	7	1.9 (1.6–4.3)	P = 0.048*
P7	9	31.1 (6.9)	10	24.1 (6.1)	NS
P10	7	5.8 (5.4–14.7)	4	2.8 (1.6–5.0)	NS
P14	3	1.6 (0.7)	6	3.4 (0.6)	NS

Table provides relative amounts of RUNX1 mRNA as assessed by quantitative polymerase chain reaction, normalized to 18S expression and then to lowest expressing sample to obtain fold changes in relative expression (ΔΔCt). Values are means (SE) if normally distributed and median (interquartile range) if not normally distributed; N = number of samples. Lung tissue was pooled from 2 mice for each sample at embryonic day (e)15.5; all other time points use lung tissue from 1 mouse for each sample.

P0, day of birth; P, postnatal age in days. NS = P ≥ 0.05;

^*P < 0.05.

We next used qPCR to determine whether tobacco smoke exposure during lung development caused abnormal mRNA expression for other RUNX transcription factors. Similar to the findings observed with RUNX1, mRNA for RUNX3 was significantly (P < 0.05) decreased at P3 and P5 in lung tissue from mice exposed to cigarette smoke during lung development compared with tobacco-naive mice (data not shown). In contrast, cigarette smoke exposure during lung development significantly (P < 0.05) decreased the mRNA for RUNX3 only at P3 (data not shown).

Given the association between both RUNX1 and RUNX3 and T-bet, we next examined the murine lung tissue samples to determine whether gestational cigarette smoke exposure altered T-bet mRNA expression. This quantitative PCR analysis did not show any differences in T-bet expression between the mice with and without cigarette smoke exposure during development (data not shown).

DISCUSSION

This study shows that a prenatal environmental exposure, IUS, modifies the association between genetic variations in the transcription factor RUNX1 and airway hyperresponsiveness in asthmatic children, alters RUNX transcription factor expression in murine developing lung, and raises the possibility that transcription is also altered in human developing lung. Our limited human studies did not have the statistical power to address this issue fully. FBAT in asthmatic children indicated an association between genetic variation in RUNX1 and increased responsiveness to methacholine as assessed by the FEV₁PC₂₀. This association was modulated by a history of prenatal exposure to tobacco smoke: the AA genotype correlated with increased sensitivity to methacholine in children without tobacco exposure, as evidenced by a lower FEV₁PC₂₀, and this was only slightly exacerbated by prenatal tobacco smoke exposure. In contrast, both the AG and GG genotypes did not have an association with the FEV₁PC₂₀ in tobacco-naive asthmatic children but were associated with increased methacholine responsiveness following prenatal tobacco smoke exposure. Thus the minor allele AA genotype was associated with increased airway responsiveness in tobacco-naive asthmatic children but was slightly protective in the setting of IUS. Our data therefore show a gene-by-environment interaction for RUNX1. Family-based methods, such as FBAT, are robust to population stratification and therefore spurious results from ancestral allele frequency differences in our population are highly improbable. Surprisingly, an analysis of 11 RUNX3 SNPs did not show any association between genotype and the FEV₁PC₂₀. Similarly, analysis of 40 RUNX2 SNPs showed that two SNPs had associations with the FEV₁PC₂₀, but neither of these were modified by a history of IUS.

We further investigated the association between the RUNX transcription factors and the immature lung by comparing the expression of RUNX1 in developing human lung tissue with and without prenatal tobacco smoke exposure. Our analysis showed that prenatal tobacco smoke exposure correlated with a slight trend toward increased RUNX1 expression in the pseudoglandular stage of lung development and confirmed that RUNX1 protein is robustly expressed in human fetal lung during both the pseudoglandular and canalicular stages. Taken together, these findings support the plausibility of the fetal origins of disease hypothesis that was proposed by Barker and colleagues (5).

The pseudoglandular and canalicular stages of lung development are important for branching morphogenesis, the establishment of the basic structure of the lung, and the differentiation of critical lung cell types. The RUNX transcription factors modulate the maturation of cartilage and muscle cells (59, 69). Of note, increased airway wall size is found in children exposed to prenatal maternal smoking (17). It is possible that abnormal RUNX expression during lung development could affect the lung structures such as airway wall thickness, which could contribute to airway obstruction after birth.

We investigated whether the changes in RUNX1 following prenatal tobacco smoke exposure could be causal by examining the expression of the RUNX transcription factors in a murine model of developmental exposure to cigarette smoke. The use of a murine model for these analyses allowed us both to determine whether prenatal exposure to cigarette smoke by-products caused abnormalities in RUNX transcription factor mRNA expression in the lungs of the offspring and to examine RUNX mRNA expression in the saccular and alveolar stages of lung development. As was suggested by the findings in our human studies, our murine experiments showed that maternal smoking resulted in abnormal RUNX transcription factor expression in the lungs of the offspring. As we found in human developing lung, IUS caused a trend toward increased RUNX1 expression in the embryonic murine lung, although this was seen in the canalicular stage in the murine samples and the pseudoglandular stage in the human lung tissue samples. Additionally, in our murine model, IUS caused decreased RUNX1 expression in lung tissue samples at P3 and P5, in the alveolar stage of lung development. These differences could be due either to differential expression between mouse and human immature lung tissue or to the human and murine studies sampled representing different stages of lung development. Of note, the greatest effects of IUS on murine lung Runx expression were observed at the alveolar stage of lung development, which occurs entirely after birth in the mouse. In humans, this stage of lung development occurs very late in gestation and is completed in infancy. Our findings suggest that tobacco smoke components might change the regulation of the RUNX transcription factors in the immature lung. This study therefore supports the concept that the RUNX transcription factors as targets of maternal smoking in the immature murine and human lung.

There are several possible mechanisms whereby gestational exposure to maternal cigarette smoking could alter RUNX expression. Components of cigarette smoke, including nicotine and lipid soluble mediators, directly modulate transcription factor expression in vitro (14, 27, 31–33, 67). Nicotine also impacts the epigenetic modulation of gene expression (47). Additionally, maternal smoking causes uterine artery constriction (2) and placental abnormalities (74) and so could indirectly affect fetal expression of transcription factors.

Our study extends prior work examining the RUNX transcription factors in the lung. Lian and colleagues (42) showed that RUNX1 is expressed in the fetal and adult murine pulmonary epithelium. Our study confirms this finding and demonstrates similar expression in developing human pulmonary epithelium. A study by Chae and colleagues (10) demonstrated an association between a RUNX promoter polymorphism and IgE levels in Korean asthmatic patients. Our study shows that a different polymorphism is associated with airway hyperresponsiveness in a large cohort of asthmatic children. Similar to the study by Chae et al., our study confirms that RUNX transcription factor genetic variation is associated with asthma phenotypes. The differences between the two investigations may be due to the different populations studied: Korean adult asthmatic patients and asthmatic children in the United States.

There are three RUNX transcription factors: RUNX1, RUNX2, and RUNX3. This family has important roles in immune function and cartilage and bone formation. Additionally, they can act as both oncogenes and tumor suppressors (9, 34, 38, 41, 45, 62). RUNX1, located on chromosome 21, has been best studied in relation to acute myelogenous leukemia (22, 44). It has been associated with several autoimmune diseases, such as RA and SLE (61, 66, 72). Lack of Runx1 prevents the formation of fetal hematopoietic cells and is embryonic lethal (51, 52). Runx2, located on chromosome 6, is essential for skeletal development (54). Mice lacking Runx2 die shortly after birth (54), and abnormal Runx2 expression has been associated with the human disease cleidocranial dysplasia (48, 53). Additionally, RUNX2 abnormalities have been reported in metastatic breast and prostate cancer and in T cell lymphoma (8). RUNX3, located on chromosome 1, modulates thymocyte maturation and promotes the development of CD8 cells by activating the CD4 silencer (49, 65). The Runx3-deficient mouse displays transient spontaneous pulmonary infiltrates and increased IgE (41). Thus these transcription factors have critical roles in modulating immune cell development and function.

There are several possible mechanisms whereby abnormalities in RUNX transcription factor expression could contribute to the asthma phenotype. RUNX transcription factors are required for thymocyte maturation and B cell function and modulate CD4 expression by binding to the IL-4 silencer (3, 7, 15, 49, 58, 65). Increased RUNX1 expression increases CD8 cell number (25); however, these cells do not function normally. Thus increased RUNX1 could result in impaired responses to viral infections, which can result in increased IL-13, which would, in turn, favor an asthmatic phenotype (36). Additionally, RUNX1 promotes IL-17 expression; therefore overexpression could increase this proinflammatory cytokine (75). Recent studies indicate that RUNX1 also has nonimmune cell functions that could also promote an asthma phenotype. For example, RUNX1 promotes angiogenesis and muscle cell growth (69). Thus it is feasible that abnormal RUNX1 expression could contribute to airway remodeling, which is a typical finding in both asthma and IUS (17). Our findings show that developing human lung epithelial cells express RUNX1. Dysregulation of epithelial cell cytokine expression has been shown in asthma (28, 29, 43); therefore it is feasible that abnormal epithelial cell RUNX1 expression could promote asthma. Abnormal expression of the other RUNX transcription factors could additionally promote asthma: RUNX3 deficiency in mice results in spontaneous eosinophilic lung infiltrates (19). Persistent eosinophilic infiltration is typical of the pathology of human asthma. We speculate that maternal smoking could promote asthma in the offspring by causing abnormal expression of transcription factors such as RUNX1 and RUNX3.

A limitation of this study is that our data do not permit us to determine whether the association of rs11702779 and airway hyperresponsiveness due to IUS is causal or whether that SNP in RUNX1 is in linkage disequilibrium with another variant causing the airway hyperresponsiveness. This polymorphism is in a noncoding region of the RUNX1 gene; however, recent investigations have determined that these areas may have important regulatory functions (35). Further molecular genetics studies to determine the functional SNP are warranted. Another limitation is that the size of the CAMP genotype database is also relatively small, which limits the power of our analysis. We would anticipate that additional associations would be identified in a larger population samples. However, despite its small size, the CAMP database has the advantage of a very detailed phenotypic characterization of the subjects who participated in this study for several years.

In summary, our findings support the concept that environmental exposures, such as cigarette smoke, during fetal development might contribute to lung dysfunction after birth. Our murine data show significant changes in postnatal RUNX expression in the lung tissue of mice with exposure to cigarette smoke by-products during lung development. We speculate that the changes in the three RUNX transcription factors following developmental tobacco smoke exposure and the known importance of this family in regulating immune function raise the possibility that the abnormal expression patterns of these transcription factors contributes to the increased susceptibility to asthma following prenatal tobacco smoke exposure. Our data show that IUS can contribute to postnatal lung disease by both direct and indirect mechanisms: IUS changes transcription factor expression, and modulates the impact of SNPs in lung genes. Further studies to investigate the interaction of IUS with the developing lung genome are needed to characterize the ways in which IUS predisposes children to develop asthma, with the goal of identifying new therapeutic targets for this important disease.

GRANTS

This study was funded by Flight Attendants Medical Research Institute (FAMRI) Clinical Innovator's Award CIA_062525 (K. J. Haley), NIH K08 HL7910 (K. J. Haley), NIH R01 HL097144 (K. J. Haley, K. Tantisira, S. T. Weiss), R03ES016399 (K. J. Haley) and R01 HL092197 (K. Tantisira). The CAMP Genetics Ancillary Study is supported by U01 HL075419, U01 HL65899, P01 HL083069, R01 HL086601, and T32 HL07427 from the National Heart, Lung and Blood Institute, National Institutes of Health.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).