Transplacental Antimicrobial Antibodies and Childhood Asthma: Maternal Nativity as an Upstream Factor

Abstract

Background:

Early-life exposure to microbes may play a role in asthma development. This study tests the hypothesis that mothers born in low-income countries might have been exposed to diverse microbes - leading to greater diversity and higher levels of transplacental anti-microbial antibodies, thereby conferring protection against asthma in offspring.

Methods

In the prospective Boston Birth Cohort, IgG antibody reactome against microbes (2740 species; 1311 genera) were profiled in cord blood, using Phage ImmunoPrecipitation Sequencing. Multinomial regression models were applied to examine the associations of cord blood IgG reactome with child risk of physician-diagnosed atopy and asthma. Mediation analysis was performed to examine the inter-relationships among maternal nativity, IgG reactome and risk of developing childhood asthma.

Results

This report includes 943 mother-child dyads enrolled at birth and followed prospectively. Compared to children of US-born mothers, children of foreign-born mothers had a lower prevalence of asthma (17.5% vs 30.5%) and greater diversity of cord blood IgG antibodies against hundreds of microbes (FDR < 0.05). Cord blood seropositivity to peptides or proteins from 6 microbes were inversely associated with risk of asthma in children, and children with more seropositivity of these microbes were at a lower risk of developing asthma (P < 0.001). IgG seropositivity to A. Actinomycetemcomitans, H. pylori, S. flexneri, and T. parva each mediated 17-62% of the association between maternal nativity and child risk of asthma.

Conclusion

In this US prospective birth cohort, maternal transplacental IgG reactivity to four microbes (A. Actinomycetemcomitans, H. pylori, S. flexneri and T. parva) were associated with lower risk of childhood asthma, and partly explain the lower risk of asthma in children of mothers born outside US.

Introduction

Childhood asthma is a major public health problem in the US, affecting one in 12 school-age children.¹ Epidemiologic research has identified many risk factors for childhood asthma, including genetic susceptibility, ^2-4 infections, ^5,6 other environmental factors ^7-10, and their interactions. Despite major progress in experimental and clinical research, the prevalence of asthma has increased over the past decades, particularly in developed countries.

Previous studies have suggested that immigrants from countries less affluent than the US had lower risk of asthma,^11-14 but the underlying mechanisms are largely unknown. There is a particular lack of prospective longitudinal studies to examine whether the reduced risk of asthma in first generation of immigrants extends to their offsprings born in the US. Evidence has shown that mothers from less industrialized countries are likely exposed to more diverse environmental microbiota, which can shape the development and function of the maternal immune system. Maternal prenatal immune status may influence asthma development in her child by altering the epigenome and trained innate immunity at birth.¹⁵ The maternal immune system is known to play a fundamental role in the induction, training, and function of the child’s immune system.^16-18 Because neonates produce little IgG antibodies immediately after birth,¹⁹ they rely on passive immunity from maternal IgG until 4-6 months.^20,21 Studies have also shown that early life exposure to microbiota and bacteria (such as H. pylori) may protect again asthma development, ^22-24 aligning with the “hygiene hypothesis”. However, it remains largely unclear whether prenatal exposure to such microbes or passive immunity against these microbes at birth play a role in asthma development.

We hypothesize that compared to US-born mothers, mothers born in less-industrialized countries might have higher diversity and higher levels of certain circulating IgG antibodies prenatally that are transferred to the fetus via the placenta and protective against offspring asthma risk. This study was conducted in the prospective Boston Birth Cohort (BBC), an ideal population for this type of investigation with ~65% of immigrant mothers, all of whom gave birth (second generation) at the Boston Medical Center in Boston, MA. To our knowledge, this report is the first large prospective birth cohort study to systematically assess the role of maternally transferred IgG reactome against thousands of microbial toxins and virulence factors in cord blood by employing Phage ImmunoPrecipitation Sequencing (PhIP-Seq) in relation to maternal nativity and offspring risk of asthma.

Methods

Study population

The BBC, consisting of predominantly urban, low-income, multi-ethnic mother-child pairs, was initiated in 1998 with rolling enrollment at the BMC in Boston, MA. Details about this parent cohort have been reported elsewhere.²⁵ After providing written informed consent, data on maternal sociodemographic characteristics and lifestyle were obtained based on standardized questionnaire interview. Maternal and newborn clinical information, including birth outcomes, was abstracted from electronic medical records (EMRs). Postnatal follow-up has been ongoing since 2004. Trained research staff interviewed mothers using standardized questionnaires to gather important data on postnatal health. The study team worked with data warehouse specialists to abstract EMRs of children annually, including dates of clinical visits, diagnoses (International Classification of Diseases [ICD] codes) and lab results from all types of clinical encounters. The study protocol received initial and annual approval from the Institutional Review Boards of BMC and the Johns Hopkins Bloomberg School of Public Health.

A total of 3,416 mother-child pairs have been followed in the BBC, including 2,888 pairs enrolled between 2000 and 2011. Among them, 986 children were enrolled for antibody profiling in cord blood, with priority given to those children with plasma samples available both at birth and during early childhood. After excluding 15 children with failed antibody profiling and 28 with an unknown diagnosis of asthma and atopy, 943 children remained for subsequent data analyses (Fig S1). Population characteristics of these 943 children were similar to the overall 2888 children, except that the subset had a higher gestational age and a lower proportion of preterm births (PTBs, Table S1).

IgG Antibody Profiling

IgG antibody profiling was performed using PhIP-Seq which has been previously described in detail.²⁶ Briefly, the ToxScan peptide library, applied in this study, included 14,429 unique protein sequences (2740 species; 1311 genera; 95,601 peptides) with “toxin” or “virulence factors” included as keywords in the UniProt databases to represent a set of proteins more likely to elicit an immune response. After sequencing, raw reads were processed as previously described,²⁷ with two modifications. First, enrichment of peptide reactivities (in fold changes or FC) in samples were defined via comparison to all negative control immunocaptures from the run (rather than only those on the same plate). Second, peptide enrichment was assessed in EdgeR using tagwise dispersion estimation as opposed to estimation of a common dispersion factor. Sequencing reads attributed to phage library members were counted via exact sequence matching, and EdgeR was used to assess the significance of enrichment for each peptide compared to the negative control immunocaptures. After Benjamin-Hochberg false discovery rate (FDR) adjustment, peptides were considered as reactive (or seropositive) if the adjusted p-value was <0.003. This threshold was determined empirically by examining replicate samples and identifying the adjusted p-value cutoff below which concordance began to exceed 50%. For a protein represented by multiple peptides in the library, it was considered as seropositive if one or more of its peptides were seropositive.

Definition of Outcomes

Asthma was diagnosed at five years of age or older, based on physician diagnoses and ICD codes from EMRs (Table S2). If a child had an ICD code for asthma only at <5 years, the asthma status for this child was coded as unknown. Diagnosis of other atopic diseases including atopic dermatitis and atopic rhinitis during childhood were also based on ICD codes (Table S2); and food allergy was defined based on clinical symptoms on food exposure and the existence of food sensitization, as reported.²⁸ A three-categorical outcome was then classified as: 0= having neither atopy nor asthma (the reference group); 1 = atopy only referring to those who had other atopic diseases but not asthma, and 2 = asthma with or without atopy, for subsequent data analyses.

Maternal Nativity and Covariates

At enrollment, mothers were interviewed to collect demographic characteristics before and during pregnancy. Maternal nativity was coded as US-born vs foreign-born. Foreign-born mothers were further classified as “recent immigrants” (residing in the US for <10 years) and “established immigrants” (residing in the US for ≥10 years). Pre-pregnancy body mass index (BMI) was calculated as self-reported pre-pregnancy weight in kilograms divided by self-reported height in meters squared. Maternal history of atopy and asthma was defined if the mother ever had physician-diagnosed other atopic disease rather than asthma and asthma, respectively. PTB was defined as gestational age at delivery <37 weeks. At each visit during early childhood, maternal questionnaire interview was applied to collect feeding modality (“exclusively formula-fed”, “exclusively breast-fed” for at least the first four months, and “both formula-fed and breast-fed”), which was further classified as “exclusively formula-fed” vs “any breastfed”. Lower respiratory tract infection during the first 5 years of life (LRTI), including bronchitis / bronchiolitis / pneumonia, was defined based on physician diagnosis and ICD-codes as documented in the EMRs.

Statistical Analysis

Cord blood IgG profiles were visualized using T-distributed stochastic neighbor embedding (TSNE), and their differences by maternal nativity were tested using permutational multivariate ANOVA (PERMANOVA). The overall diversity of seropositive peptides , calculated as the absolute number of peptides having IgG seropositivity in cord blood regardless of sequence overlap or prevalence, were presented in box plots.

Analysis of individual IgG reactivity was focused on 2327 immuno-prevalent peptides from 383 species of 246 genera with seropositivity at ≥5% prevalence. Peptide-level IgG was analyzed as both a binary variable (seropositive vs seronegative to a peptide) and a continuous variable (or inversely normalized FC) for its association with maternal nativity by fitting a logistic regression or linear regression model, respectively, adjusting for covariates including race/ethnicity, maternal age, pre-pregnancy BMI, maternal smoking, delivery type, prematurity and child’s sex. For those peptides whose seropositivity and IgG FC levels both significantly differed by maternal nativity, their IgG associations with risk of atopy and asthma (the three-categorical outcome) were analyzed using multinomial regression models, adjusting for the above-mentioned covariates and child’s age at last visit (Model 1). These analyses were repeated with additional adjustment of maternal history of atopy and asthma (Model 2) to compare whether the associations were dependent of maternal history of atopy and asthma. To account for multiple testing involving higher correlated exposure,^29,30 the Benjamin-Hochberg FDR correction was applied based on the total number of peptides and outcomes being analyzed in each model, with FDR < 0.05 as the significance level. Protein-level seropositivity was also analyzed using similar regression models as mentioned above.

As a sensitivity analysis, we applied more stringent criteria to define asthma: 1) the child was diagnosed as having asthma at ≥ 5 years by physicians for three or more times, or the child was diagnosed as having asthma for twice and had records on asthma medication; 2) all children have been followed for ≥ 5 years. A total of 188 children who did not meet this stringent criterion were excluded from sensitivity analyses.

For microbes with peptide-level or protein-level IgG FC or seropositivity associated with both maternal nativity and child risk of asthma, their mediation effects on the association between maternal nativity and child risk of asthma were performed, adjusting for covariates as mentioned above. In mediation analyses, each IgG seropositivity was analyzed as a binary potential mediator. Using “medplex” package, the total impact of maternal nativity on asthma risk (based on logistic analyses) was separated into the natural direct effect (NDE) and natural indirect effect (NIE, or mediating effect). The joint mediation effects by different combination of the identified potential mediators were also analyzed, based on the hypothesis that the total effect of maternal nativity on child asthma risk can be decomposed into the effect transmitted through seropositivity against multiple microbes simultaneously, and these potential mediators are not linked in a sequential causal chain (Figure S2). We considered a two-sided P<0.05 as the statistical significance cutoff in the mediation analysis. All analyses were performed in R4.3.1.

Results

Among the 943 children included in this study, 208 developed asthma at age ≥5 years, 77.8% of whom (n=162) also had other atopic diseases (“asthmatic children”), and 306 developed other atopic diseases but not asthma (“atopy-only children”). Compared to non-atopic non-asthmatic children, atopic-only children were more likely to be born to Black mothers, less likely to be bottle-fed, and had a longer follow-up period; asthmatic children were more likely to be born preterm, had a longer follow-up period, and their mothers were more likely to be Black, US-born, and have a history of atopy and asthma (P <0.05, Table 1).

Table 1.

Maternal and child characteristics by subgroups defined by child atopy and asthma in the Boston Birth Cohort.

Population characteristics a		Non-asthmatic and non-atopic children N=429	Atopic-only children d N=306	Asthmatic children (+/− atopy) d N=208
Maternal age at delivery, years (Mean ± SD)		28.5±6.2	28.9±6.2	27.9± 6.7
Gestational age, week (Mean ± SD)		38.6±2.4	38.8±2.1	37.7±3.0**
Maternal race and ethnicity, African American		231 (53.8)	191 (62.4) *	133 (63.9) *
Maternal nativity
US-born		138 (32.2)	92 (30.1)	101 (48.6) **
Established immigrant (residing for ≥10 years)		85 (19.8)	61 (19.9)	24 (11.5)
Recent immigrant (residing for <10 years		178 (41.5)	137 (44.8)	77 (37.0)
Immigrant with no info about years in the U.S		28 (6.5)	16 (5.2)	6 (2.9)
Maternal marital status b	married	145 (33.8)	113 (36.9)	56 (26.9)
Maternal educational status b	< High school	124 (28.9)	68 (22.2)	64 (30.8)
Pre-pregnancy body mass index category
Normal		215 (50.1)	142 (46.4)	82 (39.4) #
Overweight		113 (26.3)	79 (25.8)	64 (30.8)
Obese		87 (20.3)	72 (23.5)	50 (24.0)
Unknown		14 (3.3)	13 (4.2)	12 (5.8)
Nulliparity		257 (59.9)	184 (60.1)	117 (56.2)
Maternal smoking during pregnancy b
Never		356 (83.0)	271 (88.6) #	161 (77.4)
Quitter		28 (6.5)	18 (5.9)	19 (9.1)
Continuous		45 (10.5)	17 (5.6)	28 (13.5)
Delivery mode, c-section		141 (32.9)	111 (36.3)	83 (39.9) #
Maternal history of atopy and asthma c
None		331 (77.2)	229 (74.8)	114 (54.8) **
Other atopic disease		48 (11.2)	39 (12.7)	62 (29.8)
Asthma		50 (11.6)	38 (12.4)	32 (15.4)
Breastfeeding
Bottle-fed exclusively		103 (24.3)	52 (17.0)*	58 (27.9)
Both bottle- and breast-fed		209 (49.4)	173 (56.5)	101 (48.5)
Exclusively breastfed during the first 4 months		11 (26.2)	81 (26.5)	49 (23.6)
Child’s sex, Male		205 (47.8)	149 (48.7)	119 (57.2) #
Child's age at last visit, years (Mean ± SD)		10.0±5.7	12.4±4.9 **	14.6±2.9 **

^aMean±SD and n (%) are shown for continuous and categorical variables, respectively.

^bThese variables have missing data in < 2% samples, which are imputed as the most-frequent categories.

^cMaternal history of atopy and asthma was defined based on both medical record and ICD code.

^dPopulation characteristics in children with asthma (with or without atopy) or in atopy-only children were compared to non-atopic non-asthmatic children, using the t-test and chi square test for continuous and categorical variables, respectively.

^**P < 0.01

^*P < 0.05

^#P < 0.1

Maternal Nativity and Child’s Risk of Atopy and Asthma

There were 612 mothers (64.9%) born out of US, predominantly in developing or low-income countries (~92% of the foreign-born mothers), including Haiti (35.5%) and Cape Verde (8.7%), who themselves had lower prevalence of asthma (9.3%) than US-born mothers (19.0%, P< 0.001). Children of foreign-born mothers also had a lower asthma prevalence (17.5%) than peers of US-born mothers (30.5%, P< 0.0001), but asthma rates were not significantly different between children of established immigrants (14.1%) and of recent immigrants (19.6%, P=0.27). Using multinomial regression models to analyze the multiple-categorical outcome (none / atopy-only / asthma) and adjust for covariates, children of immigrant mothers had 54% (OR=0.46, 95%CI=0.31-0.68) lower odds of developing asthma compared with children of US-born mothers, or 38% (OR=0.62, 95%CI = 0.41-0.96) lower odds when further adjusting for maternal atopy and asthma. In comparison, no significant associations were found for risk of atopy only. The joint effect of maternal nativity and maternal history of atopy and asthma were shown in Table S3. Compared with children of US-born asthmatic mothers, those of foreign-born asthmatic mothers had lower, but not statistically significant, odds of asthma (OR=0.47, 95%CI=0.20-1.11); Children of foreign-born non-asthmatic mothers had 82% lower odds of asthma (OR=0.18, 95%CI=0.10-0.32). No significant interaction was observed between maternal nativity and maternal history of atopy and asthma.

Maternal Nativity and IgG Profiles in Cord Blood Samples

We detected cord blood seropositivity to an average of 800 (SD: 230) peptides or an average of 459 (SD: 136) proteins (Fig S3). The most frequently recognized microbes included S. pyogenes, S. pneumoniae and S. aureus, with seropositivity in > 95% cord blood samples.

As shown in Fig S4, cord blood IgG reactome differed significantly by maternal nativity, which explained 7% variance (R²=0.07, P<0.001). The overall diversity of seropositive peptides was the lowest among children of US-born mothers, followed by those of established immigrants and then by those of recent immigrants (Fig S5). After controlling for maternal nativity, other maternal /newborn factors, such as, maternal race/ethnicity, newborn sex, and PTB, showed no associations with the diversity of seropositive peptides (Fig S5). Fig 1 presents a heatmap for cord blood IgG reactome against 2,327 immuno-prevalent peptides, revealing marked differences by maternal nativity, especially for peptides from Shigella flexneri (S. flexneri) and Toxoplasma gondii (T. gondii). Given that IgG reactome in children of established immigrants and in those of recent immigrants were largely comparable, maternal nativity was analyzed as foreign-born vs US-born in subsequent analyses.

Figure 1.

Heatmap for cord blood IgG levels (fold changes) against the 2,327 immuno-prevalent peptides, stratified by maternal nativity and years residing in the U.S. in the Boston Birth Cohort. The top microbes with 3 or more peptides whose IgG reactivities were significantly associated with maternal nativity were labeled in the plot.

Adjusting for covariates, we observed 583 peptides (132 species,86 genera) with cord blood seropositivity significantly varying by maternal nativity. Of these, 451 peptides (96 species, 63 genera) showed significant difference in IgG levels (analyzed as continuous variables) by maternal nativity at FDR < 0.05. As labelled in Fig 1, 24 microbes had more than 3 peptides with seropositivity and IgG levels associated with maternal nativity. At protein level, cord blood seropositivity to 166 (out of 1,829) proteins were significantly associated with maternal nativity, including 123 proteins with higher seropositivity and 43 proteins with lower seropositivity in children of foreign-born mothers than those of US-born mothers.

Cord blood IgG Profiles in Association with Asthma during Childhood

Among the 451 peptides with both cord blood seropositivity and IgG levels significantly varying by maternal nativity, there were 19 peptides (16 from S. flexneri, 1 from S. boydii serotype 18, 1 from Helicobacter pylori [H. pylori] and 1 from Theileria parva [T. parva]) with seropositivity significantly associated with lower risk of asthma (FDR < 0.05, Table 2, Figure S6a). The strongest association was observed for a S. flexneri peptide (OR=0.38, 95%CI=0.25-0.58) located in the invasion plasmid antigen C (ipaC, residues 224-280). When IgG was analyzed as a continuous variable, significant associations were found for 19 peptides, including two annotated to Aggregatibacter Actinomycetemcomitans [A. Actinomycetemcomitans] and T. gondi (Table 2, Figure S6b). These two peptides showed similar association trends but statistically insignificant when seropositivity was analyzed as a binary variable . IgG profiles to these identified peptides did not significantly differ between 162 asthmatic children with other atopic diseases and 46 asthmatic children without other atopic diseases (Data not shown). No significant associations were found for other atopic diseases rather than asthma. In sensitivity analyses, the identified associations between cord blood IgG antibodies and child risk of asthma remained comparable although with somewhat reduced effect sizes with further adjustment of maternal history of asthma (Table S4), breastfeeding and LRTI (Table S5); or when more stringent criteria were applied for asthma diagnosis (Table S6). Similar findings were also observed when asthma was analyzed as the time-to-event outcome using the cox regression models (Figure S7, Table S7). At the protein level, IgG reactivity was analyzed as a binary exposure. We identified 4 proteins (one from S. boydii and three from S. flexneri) with seropositivity significantly associated with lower risk of asthma at FDR < 0.05 (Table S8).

Table 2.

Odd Ratio (OR, 95%CI) for the associations between cord blood IgG reactivity against peptides of microbes and child’s risk of incident asthma in the Boston Birth Cohort.

Microbial peptides a,b	Inversely normalized IgG (Continuous variable) c		Inversely normalized IgG (Continuous variable) c
	Atopy only	Asthma	Atopy only	Asthma
A.actinomycetemcomitans: leukotoxin ∣ 896-952	1.07 (0.72-1.60)	0.51 (0.33-0.79)	0.93 (0.79-1.09)	0.72 (0.59-0.87) *
H.pylori: vacA ∣ 896-952	0.91 (0.66-1.25)	0.49 (0.32-0.73) *	1.00 (0.86-1.17)	0.81 (0.67-0.98)
S.boydii serotype 18: ipaH9.8 ∣ 196-252	1.00 (0.72-1.38)	0.51 (0.34-0.76) *	0.94 (0.80-1.11)	0.72 (0.59-0.87) *
S.flexneri: ipaH7.8 ∣504-560	1.20 (0.87-1.67)	0.61 (0.41-0.91)	1.04 (0.88-1.23)	0.70 (0.57-0.85) *
S.flexneri: ipaH4.5 ∣ 224-280	1.00 (0.72-1.37)	0.57 (0.38-0.84)	0.95 (0.81-1.11)	0.72 (0.59-0.87) *
S.flexneri: ipaH4.5 ∣ 252-308	1.03 (0.74-1.42)	0.50 (0.33-0.74) *	0.95 (0.81-1.12)	0.71 (0.58-0.86) *
S.flexneri: ipaH4.5 ∣ 504-560	1.07 (0.76-1.50)	0.54 (0.36-0.81)	1.03 (0.87-1.21)	0.68 (0.55-0.83) *
S.flexneri: ipaH4.5 ∣ 520-574	1.17 (0.84-1.63)	0.61 (0.41-0.92)	0.99 (0.84-1.17)	0.65 (0.53-0.80) *
S.flexneri: ipaA ∣ 196-252	0.89 (0.65-1.23)	0.49 (0.33-0.74) *	0.95 (0.81-1.11)	0.68 (0.56-0.83) *
S.flexneri: ipaA ∣ 224-280	0.88 (0.64-1.22)	0.47 (0.31-0.72) *	0.94 (0.80-1.10)	0.72 (0.59-0.87) *
S.flexneri: ipaA ∣ 252-308	1.08 (0.77-1.51)	0.49 (0.33-0.74) *	1.01 (0.86-1.19)	0.72 (0.59-0.87) *
S.flexneri: ipaA ∣ 280-336	1.04 (0.75-1.45)	0.48 (0.32-0.73) *	0.92 (0.78-1.08)	0.68 (0.55-0.84) *
S.flexneri: ipaC ∣ 0-56	0.86 (0.62-1.21)	0.43 (0.29-0.65) *	1.02 (0.87-1.20)	0.76 (0.63-0.93)
S.flexneri: ipaC ∣ 28-84	0.85 (0.60-1.20)	0.48 (0.32-0.73) *	0.95 (0.81-1.12)	0.68 (0.56-0.83) *
S.flexneri: ipaC ∣ 56-112	0.88 (0.63-1.24)	0.42 (0.28-0.63) *	0.89 (0.76-1.05)	0.67 (0.55-0.82) *
S.flexneri: ipaC ∣ 140-196	0.96 (0.68-1.34)	0.49 (0.33-0.74) *	0.92 (0.78-1.08)	0.67 (0.54-0.81) *
S.flexneri: ipaC ∣ 168-224	0.87 (0.60-1.25)	0.40 (0.26-0.60) *	0.97 (0.82-1.14)	0.75 (0.61-0.92)
S.flexneri: ipaC ∣ 196-252	0.84 (0.59-1.21)	0.40 (0.26-0.61) *	0.93 (0.79-1.09)	0.70 (0.58-0.86) *
S.flexneri: ipaC ∣ 224-280	0.90 (0.63-1.29)	0.38 (0.25-0.58) *	0.97 (0.82-1.14)	0.72 (0.59-0.88) *
S.flexneri: ipaC ∣ 252-308	0.86 (0.61-1.21)	0.46 (0.31-0.70) *	0.99 (0.84-1.16)	0.75 (0.61-0.91)
S.flexneri: ipaC ∣ 280-336	0.97 (0.69-1.38)	0.47 (0.32-0.71) *	1.00 (0.85-1.18)	0.67 (0.55-0.82) *
S.flexneri: ipaB ∣ 140-196	0.94 (0.65-1.36)	0.39 (0.26-0.60) *	0.95 (0.80-1.12)	0.67 (0.54-0.82) *
S.flexneri: ipaB ∣ 476-532	1.04 (0.75-1.44)	0.54 (0.36-0.80)	0.96 (0.82-1.13)	0.72 (0.59-0.87) *
S.flexneri: ipgC ∣ 111-155	0.89 (0.64-1.24)	0.44 (0.29-0.65) *	0.93 (0.79-1.10)	0.75 (0.61-0.91)
T.gondii: dense granule protein 1∣141-190	1.00 (0.70-1.44)	0.67 (0.42-1.07)	0.88 (0.76-1.03)	0.72 (0.60-0.87) *
T.parva: vapC ∣ 84-140	0.95 (0.67-1.37)	0.40 (0.24-0.67) *	0.98 (0.84-1.15)	0.77 (0.64-0.94)

^aLimit to 26 peptides with cord blood seropositivity or inversely normalized IgG fold changes significantly associated with maternal nativity and with child’s risk of asthma at FDR < 0.05.

^*FDR < 0.05

^bEach peptide is expressed as taxa genus (first initial). species: protein ∣ peptide location.

^cMultinomial regression models were applied with the three-categorical atopy / asthma phenotype as the outcome, adjusting for maternal race and ethnicity (non-Hispanic Black vs others), maternal age at delivery, pre-pregnancy BMI, smoking during pregnancy, delivery type, prematurity, child's sex, and child’s age at last visit.

There were 26 peptides and 4 proteins from 6 microbes whose IgG seropositivity or IgG levels significantly associated with both maternal nativity and child risk of asthma (Table 2 & S8). The majority of them were from S. flexneri (21 peptides and 3 proteins), with their genomic location mapped to Shigella virulence plasmid (Figure S8). Because IgG reactivities to these S. flexneri peptides or proteins were significantly correlated with each other (Figure S9), we selected the top S. flexneri peptide (spanning the 224-280 region of the ipaC protein, with the most significant association with risk of asthma) for subsequent analyses. Besides, the identified S boydii serotype 18 peptide shared a 13-AA fragment (LQRLTSSPDYHG) with two S. flexneri peptides located in ipa4.5, and its IgG level was highly correlated with that of the top S. flexneri peptide (r = 0.76), so we excluded the S boydii serotype 18 peptide for subsequent analyses to minimize overlap.

Table 3 presented the joint effect of seropositivity to peptides from the rest five microbes (one from each microbe) on asthma risk. We observed that children with more seropositive microbial peptides were at a lower risk of developing asthma in a dose-responsive manner: compared to those with zero seropositive peptides, children with 2 or 3 seropositive peptides had 70% (OR=0.30, 95%CI=0.16-0.56) or 78% lower odds (OR=0.22, 95%CI=0.12-0.43) ; and those with 5 seropositive microbial peptides were at the lowest risk (OR=0.13, 95%CI=0.04-0.46). The associations remained robust when further adjusting for maternal history of atopy and asthma.

Table 3.

Odd Ratio (OR, 95%CI) for the associations between number of seropositive microbial peptides and child’s risk of incident asthma in the Boston Birth Cohort

N. of seropositive microbial peptides a	N	Model 1b		Model 2c
		Atopy only	Asthma	Atopy only	Asthma
0	127	ref	ref	ref	ref
1	197	1.06 (0.59-1.90)	0.58 (0.32-1.06)	1.09 (0.60-1.95)	0.64 (0.34-1.20)
2	213	0.80 (0.45-1.43)	0.30 (0.16-0.56) ***	0.83 (0.46-1.51)	0.41 (0.21-0.78) **
3	232	0.94 (0.53-1.67)	0.22 (0.12-0.43) ****	0.98 (0.55-1.77)	0.32 (0.16-0.63) **
4	138	0.87 (0.46-1.63)	0.19 (0.09-0.39) ****	0.91 (0.48-1.72)	0.25 (0.12-0.55) ***
5	38	0.97 (0.41-2.28)	0.13 (0.04-0.46) **	1.01 (0.43-2.41)	0.17 (0.05-0.61) **
Linear trend		0.98 (0.87-1.10)	0.65 (0.56-0.76) ****	0.99 (0.87-1.11)	0.71 (0.61-0.83) ****

^aLimit to one peptide for each of the five microbes (A. actinomycetemcomitan, H. pylori, S. flexneri, T. gondii and T. parva).

^bModel 1: Multinomial regression models were applied with the three-categorical atopy / asthma phenotype as the outcome, adjusting for maternal race and ethnicity (non-Hispanic Black vs others), maternal age at delivery, pre-pregnancy BMI, smoking during pregnancy, delivery type, prematurity, child's sex, and child’s age at last visit.

^*P < 0.05

^**P < 0.01

^***P < 0.001

^****P < 0.0001

^*****P < 1.0×10⁻⁶

^cModel 2: With further adjustment of maternal history of atopy and asthma.

Maternal Nativity, IgG Antibodies in Cord Blood and Risk of Asthma in Children

Cord blood seropositivity to the five microbes was analyzed for potential mediation effects. As shown in Table 4, seropositivity to A. actinomycetemcomita, H. pylori, S. flexneri, and T. parva each significantly mediated the inverse association between maternal foreign-born status (compared to US-born status) and child’s risk of asthma by 19.5%, 17.3%, 61.8% and 25.1%, respectively, and the mediation diagram was shown in Fig 2. Moreover, concurrent seropositivity to peptides from 2 to 4 microbes jointly accounted for 33.2-88.4% of the association between maternal nativity and child asthma risk. The mediation effects remained comparable with further adjustment of maternal history of asthma (Table S9).

Table 4.

Proportion ^a mediated by cord blood seropositivity to five microbes on the association between maternal nativity and child risk of asthma in the Boston Birth Cohort.

Potential mediators a	NDE OR (95% CI)	NIE OR (95% CI)	% mediatedb
Single mediator
A. actinomycetemcomitans (leukotoxin ∣ 896-952)	0.48 (0.30-0.76)	0.84 (0.71-0.98)	19.5
H. pylori (vacA ∣ 896-952)	0.47 (0.30-0.73)	0.85 (0.76-0.96)	17.3
S. flexneri (ipaC ∣ 224-280)	0.70 (0.37-1.35)	0.56 (0.35-0.91)	61.8
T. gondii (dense granule protein 1∣ 141-190)	0.42 (0.26-0.66)	0.96 (0.84-1.09)	4.9
T. parva (vapC ∣ 84-140)	0.50 (0.31-0.80)	0.79 (0.67-0.94)	25.1
Multiple mediators c
A. actinomycetemcomitans, H. pylori	0.54 (0.33-0.87)	0.74 (0.61-0.89)	33.2
A. actinomycetemcomitans, S. flexneri	0.77 (0.41-1.45)	0.52 (0.33-0.82)	71.2
A. actinomycetemcomitans, T. parva	0.60 (0.37-0.97)	0.67 (0.53-0.83)	44.1
H. pylori, S. flexneri	0.75 (0.40-1.43)	0.53 (0.33-0.84)	69.3
H. pylori, T. parva	0.58 (0.36-0.93)	0.69 (0.56-0.84)	40.9
S. flexneri, T. parva	0.77 (0.41-1.47)	0.51 (0.32-0.82)	72.1
A. actinomycetemcomitans, H. pylori, S. flexneri	0.81 (0.43-1.53)	0.49 (0.31-0.78)	77.4
A. actinomycetemcomitans, H. pylori, T. parva	0.67 (0.41-1.09)	0.59 (0.46-0.76)	56.7
A. actinomycetemcomitans, S. flexneri, T. parva	0.85 (0.45-1.59)	0.47 (0.30-0.74)	82.0
H. pylori, S. flexneri, T. parva	0.83 (0.44-1.57)	0.48 (0.30-0.76)	79.9
A. actinomycetemcomitans, H. pylori, S. flexneri, T. parva	0.90 (0.48-1.68)	0.44 (0.28-0.70)	88.4

^aFocused on five microbes whose protein-level or peptide-level seropositivity were significantly associated with both maternal nativity and childhood asthma. One peptide from each microbe was analyzed, expressed as genus (first initial). species ∣ protein ∣ peptide location.

^bProportion of the associations between maternal nativity (foreign-born mothers compared with US-born mothers) and risk of asthma in children mediated by seropositivity to each microbe was calculated as the percentage of natural indirect effect (NIE, in logOR scale) over the total effect (NIE + NDE, where NDE or the natural direct effect represents the unmediated effect. Both NIE and NDE are in logOR scale). Covariates adjusted in the model included maternal race and ethnicity, maternal pre-pregnancy BMI category, smoking during pregnancy, delivery type, prematurity, child's sex, and child’s age at last visit.

^cAnalyses of multiple mediators were performed for different combination of peptides from the four microbes (A. actinomycetemcomitans, H. pylori, S.flexneri, T. parva).

Figure 2. Mediation diagrams to illustrate the inter-relationships between maternal nativity, cord blood seropositivity to each microbe peptide, and child incident risk of asthma.

The figure illustrated the mediation effects by seropositivity to peptides of A. actinomycetemcomitans, H. pylori, T. parva, and S. flexneri, respectively. In each diagram, OR (95%CI) was shown for the associations between maternal nativity and incident asthma risk, between maternal nativity and cord blood seropositivity to each peptide, and between cord blood seropositivity to each peptide and incident asthma risk associations, adjusting for maternal race and ethnicity (non-Hispanic Black vs others) and maternal pre-pregnancy BMI category, maternal smoking during pregnancy, delivery type, prematurity, child's sex, maternal history of asthma, child’s age at last visit (when asthma was analyzed as the outcome) and maternal nativity (only for the associations between cord blood seropositivity and incident asthma risk in children).

Discussion

This study represents the first to elucidate the complex inter-relationships among maternal nativity, transplacental antibodies against microbes in cord blood, and child risk of asthma in a racially diverse prospective birth cohort. We reported that children of foreign-born mothers were associated with lower risk of developing asthma and a greater diversity of microbe-specific IgG antibodies in cord blood, compared to children of US-born mothers. Cord blood seropositivity to peptides from different microbes (e.x.,A. Actinomycetemcomitans, H pylori, S. flexneri, T. gondi and T. parva) were significantly associated with lower risk of asthma during childhood, and there was a dose-responsive effect between the number of seropositive microbes and reduced risk of asthma. More interestingly, seropositivity to peptides from A. Actinomycetemcomitans, H. pylori, S. flexneri and T. parva may jointly mediate 88% of the associations between maternal nativity and risk of asthma in children.

Our findings underscore the importance of considering maternal exposures—including the microorganisms she has encountered during early life—when evaluating a child's risk of asthma. The observed association between seropositivity to H. pylori (a bacterium colonizing the stomach lining) and reduced asthma risk in this study is consistent with prior reports. ^23,24 In animal models, in utero exposure to H. pylori extract showed substantial protection against allergic disease manifestations.³¹ The vacuolating cytotoxin (vacA), an immunomodulatory molecule produced by all H pylori strains, has been implicated in these protective effects.³² In line with this evidence, we found that seropositivity to a VacA peptide from H. pylori was significantly associated with lower asthma risk. Our data also suggested S. flexneri as the top microbe whose IgG reactivity in cord blood was inversely associated with child asthma risk. S. flexneri is an intracellular bacterium that infects the intestinal epithelium and causes diarrheal disease. While both the respiratory and gastrointestinal microbiomes have been linked to altered asthma pathophysiology²², few studies have examined immune-response to S. flexneri in relation to asthma.^23,33 In our analyses, most of the identified S. flexneri peptides or proteins reside on the Shigella virulence plasmid with important functionality.³⁴ Further studies are warranted to validate our findings and to elucidate the underlying mechanism.

Our data also indicate that cord blood seropositivity to peptides from A. actinomycetemcomitans, and T. parva was associated with lower risk of asthma. Consistently, previous studies have suggested that early-life exposure to oral pathogens, including A. actinomycetemcomitans,^35,36 may suppress susceptibility to asthma. In contrast, T. parva is a parasitic species that causes East Coast fever in cattle, and is primarily endemic to Eastern, Central, and Southern Africa. There is no direct evidence linking T. parva infection or antibodies with asthma risk in humans. Because > 40% foreign-born mothers in this cohort are from Haiti and Cape Verde, direct exposure to T. parva seems unlikely. A more plausible explanation is that the observed signal may reflect IgG cross-reactivity with other, more geographically relevant apicomplexan parasites to which these mothers may have been exposed.

Although the mechanisms underlying the observed inverse associations between cord blood IgG reactivity to microbes and child risk of asthma are largely unknown, one hypothesis is that maternally derived passive immunity might modulate the newborn developing immune system during a critical early developmental window, by shaping antigen exposure and T cell differentiation, or by altering microbial colonization patterns. Specifically, transplacental IgG may influence the differentiation of T-regulatory cells (Treg) and inhibit the TH2-skewed allergic response in children, leading to decreased risk of asthma. ^16-18,37-39 Maternal antibodies might also confer protection by reducing early-life susceptibility to microbial infections implicated in asthma pathogenesis, indirectly lowering long-term disease risk. These early immunomodulatory effects—despite the transient presence of maternal IgG—may have lasting consequences on immune system development and asthma susceptibility. Within the context of the hygiene hypothesis, the most relevant microbial exposures are thought to occur during early childhood. However, increasing evidence also suggests that prenatal microbial exposure may contribute to reduced allergic airway inflammation^40,41 and confer protection against asthma,^42,43 which is consistent with our findings. Alternatively, the identified antibodies in this study may simply serve as biomarkers of maternal socioeconomic status or microbiome composition, and are thus predictive of a qualitative difference in the child’s early life immune education (eg. sanitation conditions). Furthermore, the human antibody profile is shaped by both environmental exposure and host genetic background (especially the HLA region).⁴⁴ Shared genetic risk factors may contribute to the observed associations between cord blood IgG to microbes and childhood asthma risk in this study. Future research should examine the interaction between genetic susceptibility and microbial exposure (or immune response to microbes) on asthma risk, and clarify whether the protective effect of microbial exposure (or immune response to microbes) against asthma is modified the genetic background of the mother and child.

This study suggests that transplacental IgG profiles in cord blood may substantially mediate the impact of maternal nativity on child asthma risk. Differences in cord blood IgG profiles by maternal nativity likely reflect regional factors in maternal countries of origin, such as vaccination programs, local microbial type and prevalence.⁴⁵ Alternatively, the quantity and type of IgG antibodies transferred to the child may differ by subclass (e.g., IgG1, IgG2, IgG4) and sugar modifications, and may be influenced by maternal nativity which await further investigation. It is also important to note that statistical mediation analysis relies on key assumptions such as a clear temporal sequence among exposure, mediator and outcome, absence of unmeasured confounding, and no mediator–outcome confounders influenced by the exposure. Violation of these assumptions could bias estimates of the indirect effect, potentially overstating or understating the mediation associations. Accordingly, these findings should be interpreted as suggestive rather than definitive evidence of mediation.

Several limitations should be acknowledged. First, technical limitations relate to the PhIP-Seq technology including the lack of conformational epitopes and post-translational modifications of the peptides included in the library, as well as absence of microbial glycans and lipids from the library.²⁶ Second, this study is hypothesis-generating and observational in nature, and causal relationships cannot be determined. Third, asthma was defined using ICD-code based physician diagnoses, which may lead to misclassification. However, such misclassification, which is independent of maternal nativity and/or IgG reactome at birth, likely biased effect size toward the null. Fourth, because distinct microbes can share highly similar peptide fragments, some IgG signals may reflect cross-reactivity rather than pathogen-specific responses. Additional validation using other assays such as enzyme-linked immunosorbent assay (ELISA) will be an important next step, although previous studies have reported that PhIP-Seq has high sensitivity (~85-95%) and specificity of (~95-100% ) in detecting common viral exposures ⁴⁶, and has strong concordance with orthogonal binding assays ^26,47,48. Fifth, although a recent study reported a high rate of maternal IgG antibody transfer to their newborns, which did not vary by gestational age,²¹ future work is still needed to directly quantify maternal IgG reactome in maternal biospecimen and examine maternal – newborn correlations to confirm that cord blood IgG truly reflects maternal immune exposures in this cohort. Lastly, this study was conducted in a unique high-risk population. Caution is needed when generalizing our findings to populations with different social-demographic and clinical characteristics.

In summary, this study reported that cord blood seropositivity to four microbes were significantly and inversely associated with asthma risk and they explained partly lower risk of asthma in offspring of foreign-born mothers, suggesting that maternal prenatal microbial environment and antibody profiles may serve as valuable biomarkers of offspring asthma. More studies are needed to confirm our findings in other study cohorts and elucidate the independent and joint role of maternal transplacental antibodies and antibodies arising from postnatal microbe exposure on child risk of asthma. Research along this line may provide novel targets for mechanistic investigation and treatment strategies for preventing childhood asthma.

Data availability

Data will be made available upon reasonable request and after institutional IRB review and approval.