Preterm birth and the early-life respiratory microbiota: implications for host epigenetics and respiratory health

Abstract

Research in general paediatric populations indicates that the development of the respiratory microbiome in early life may predict long-term respiratory health. Perinatal and neonatal exposures, including lack of breastfeeding and antibiotic treatment, have been linked to aberrant microbiome dynamics. Additional evidence indicates that the microbiome, in turn, is capable of modulating host epigenetics, suggesting that microbiome-host interactions in early life may influence ongoing (respiratory) health trajectories. Preterm-born infants (i.e. before 37 weeks of gestation) are at high risk of experiencing unbeneficial early-life exposures and developing respiratory diseases such as bronchopulmonary dysplasia (BPD). Yet, the respiratory microbiome formation and development in this vulnerable group are incompletely understood.

This PhD project therefore sought to investigate the impact of preterm birth (PTB) on respiratory microbiota development from birth until 2 years of age, its link to host DNA methylation (DNAm) and the development of BPD. To this end, I studied a subset of preterm- and term-born infants participating in Theirworld Edinburgh Birth Cohort. First, I investigated the nasopharyngeal microbiota composition using 16S rRNA sequencing at 1 week of life, term-equivalent age, 4.5 months, 9 months and 2 years of age in preterm- and term-born infants, aiming to characterise the microbiota features related to PTB (in Chapter 3). Second, I aimed to identify (1) the main perinatal, neonatal and environmental drivers of microbiota development and (2) microbiota features associated with BPD within preterm infants (in Chapter 4). Third, I characterised the functional genome-wide DNAm landscape in early life and the impact of PTB thereon (in Chapter 5). Host DNA methylation in saliva samples was analysed using the Illumina EPIC 850K microarray at term-equivalent/neonatal age (in preterm and term infants, respectively) and at 9 months of (corrected) age. The functional DNAm profiles (DNAm modules) were identified using weighted gene correlation network analysis. Finally, I investigated the associations between the main microbiota features and the activity of DNAm modules in term and preterm infants (in Chapter 6).

In Chapter 3, where I studied the impact of PTB on the nasopharyngeal microbiota, I observed a strong correlation between microbiota composition, diversity and maturation dynamics and PTB, which was most pronounced soon after birth. Preterm-born infants showed low alpha diversity compared to term infants, and more often a Staphylococcus-dominated microbiota profile. I also observed a slower microbial community maturation in preterm infants between birth and term-equivalent age, characterised by a lower microbiota age and enrichment of potential nosocomial pathogens: Escherichia/ Shigella, Klebsiella, Staphylococcus haemolyticus and Enterococcus. Microbiota composition and maturation trajectories of preterm and term infants progressively converged after preterm infants’ discharge from the hospital. In the later study period (4.5 months – 2 years of age), the microbiota of both groups was significantly, yet differently associated with the same general exposures (season and household crowding), suggesting that PTB may modify the effects of environmental factors on microbiome dynamics.

Upon further analysis within preterm infants, in Chapter 4, I identified postnatal age and delivery mode and, to a smaller extent, gestational age at birth (GAb) as the main drivers of microbiota composition. Interestingly, after hospital discharge, environmental exposures – household size and season – became the most important drivers of microbiota composition. However, associations between the microbiota composition and past (neonatal) co-exposures of PTB – antibiotics, pulmonary surfactant treatment and exclusive maternal breast milk consumption – were also visible with microbiota composition and maturation dynamics between 4.5 and 9 months of age. The microbiota in the first week of life was also moderately associated with BPD. Infants with a future BPD diagnosis had enrichment of (among others) S. haemolyticus, a potential nosocomial pathogen known to cause neonatal sepsis, soon after birth. Also following BPD diagnosis, a correlation with microbiota composition was found, with enrichment of Staphylococcus and oral-type bacteria (including Streptococcus, Gemella, Alloprevotella, Rothia and Veillonella) in infants with BPD, at 4.5 months of corrected age.

In Chapter 5, where I studied the DNA methylation landscape in term and preterm-born infants, I identified 9 functionally related DNAm modules mapped to genes implicated in various developmental, cellular, metabolic, neurological, and immunological functions. Age at sampling was overall the strongest driver of module activity and associated with the majority (7 out of 9) of DNAm modules; the data thus demonstrated a dynamic methylome development over time in early life. PTB (GAb) was associated with 5 modules, in particular one (module M7) implicated in neurodevelopmental and innate immunity functions. BPD was additionally associated with the activity of an adaptive immunity-related module, M9.

In Chapter 6, where I integrated the microbiota and DNAm data, I observed multiple associations between microbial community composition, diversity and maturity, and the activity of 4 of the 9 DNAm modules. The same 4 modules showed associations with the microbiota in both term and preterm infants, yet the microbiota features driving these associations differed between both study groups. The main consistent associations were between the neurological function-related modules M4 and M6 and the abundance of Streptococcus, Dolosigranulum pigrum and Corynebacterium in both groups. Microbiota features driven by Staphylococcus predominance, including community (im)maturity, were strongly associated with two DNAm modules (M6 and M9), in preterm, but not term infants. These modules were implicated in neurological (M6), adaptive immunity and antigen presentation (M9) functions.

In summary, the findings of this PhD research project show that PTB has a significant and complex impact on the developmental respiratory microbiota dynamics throughout infancy, in particular during the initial hospital stay. Despite the gradual convergence to a term-like microbiota following discharge, PTB co-exposures were persistently related to microbiota dynamics even after discharge from the hospital. Combined with the observations of microbiota features that are associated with the development of BPD, these findings suggest that the respiratory microbiome may be implicated in the pathogenesis of respiratory complications of PTB. The findings of significant associations between the community features unique to PTB and developmental and immunological DNAm profiles further support this conclusion. This work thus suggests that host epigenetics may act as a potential mediator between aberrant microbiota development in early life and future illness.

The findings arising from this PhD project support a strong link between the early-life exposures following PTB and the respiratory microbiota, as well as between the microbiota and host epigenetics. These observations warrant further mechanistic studies to understand and confirm the causality behind these associations. In addition, this PhD work supports the respiratory microbiota in early life as a new research direction in the study of BPD pathophysiology. Future research should aim to enhance the translational potential of these findings, especially regarding microbiota-targeted interventions to improve the respiratory health of PTB survivors.