Clinical and sampling information for all infants
After applying inclusion and exclusion criteria, a total of 98 nasal swabs samples were collected from 54 preterm infants, 13 developed BPD infants and 41 control infants were finally involved in this study in the NICU at Children’s Hospital, Zhejiang University School of Medicine, from 2019 to 2020 (Table 1). Six premature infants were transferred to the other units before collecting the second time specimens, so we didn’t get second nasal swabs. Gestational age ranged from 25 2/7to 31 6/7. Birth weights ranged from 700 to 2,050 g. There are no significant differences between two groups in terms of gender, delivery mode, feeding, PDA, NEC, sepsis and antibiotics exposure (P > 0.05). Although the BPD group has a lower gestational age, it is still appropriate for the control to compare the nasal microbiome.
Microbial community characterization.
Eight of the samples had inadequate biomass for DNA sequencing and was excluded. A total of 5,581,298 high quality reads were obtained from the 94 samples, with a mean read count per sample of 59,376 (range 14,783 to 76,132). As shown in Figure 1A, Firmicutes, Bacteroidetes, Proteobacteria and Actinobacteria were dominant phylums and shown in Figure 1B, Muribaculaceae, Escherichina, Staphylococcus and Corynebacterium were dominant genus in all group. There was no significant difference between BPD group and control group both at first week and third week. PCoA was performed to study the similarities or differences in sample community composition. NMDS analysis was performed using the weighted UniFrac distance algorithm, and two coordinate axes that could reflect the differences between samples to the greatest extent were selected for graphical display by dimension reduction of the multidimensional data. Based on the beta diversity, including PCoA and NMDS, no significant differences were detected in comparisons (P > 0.05, Figure 2A, 2B). To evaluate which bacterial genera were involved in these observed temporal differences, we examined the relative abundances of the prevalent taxa. Some of the abundant genera changed significantly in relative abundance at first and third week (Kruskal–Wallis test, all P-values < 0.05) (Supplementary Figure 1A, 1B). We found increased in the expression of Prevotella, Marinomonas, Enterobacteriaceae, Weissella, Selenomonas, Oribacterium, Nubsella and Antricoccus in BPD group at both time points. Prevotella (phylum Bacteroidetes) was shown in Figure 3. We also found that Prevotella was correlated with the severity of BPD (Spearman r=0.361, P=0.000).
Reduced Coumarins And Mannosylglycerate Biosynthesis Associated With Bpd
To infer metabolic pathways associated with the nasal taxa identified as differentially abundant based on BPD, we used PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) and STAMP (STatistical Analysis of Metagenomic Profiles)17 to map microbial genes to metabolic databases to infer microbial functions differentially expressed by BPD. Coumarins and mannosylglycerate biosynthesis were the metabolic pathway that was differentially abundant (Kruskal-Wallis, all P <0.05) across the study groups. Compared to control group, coumarins and mannosylglycerate biosynthesis were reduced in BPD (Figure 4A, 4B).