Microdiversity of the vaginal microbiome is associated with preterm birth

Liao, Jingqiu; Shenhav, Liat; Urban, Julia A.; Serrano, Myrna; Zhu, Bin; Buck, Gregory A.; Korem, Tal

doi:10.1038/s41467-023-40719-7

Download PDF

Article
Open access
Published: 17 August 2023

Microdiversity of the vaginal microbiome is associated with preterm birth

Nature Communications volume 14, Article number: 4997 (2023) Cite this article

6848 Accesses
4 Citations
38 Altmetric
Metrics details

Subjects

Abstract

Preterm birth (PTB) is the leading cause of neonatal morbidity and mortality. The vaginal microbiome has been associated with PTB, yet the mechanisms underlying this association are not fully understood. Understanding microbial genetic adaptations to selective pressures, especially those related to the host, may yield insights into these associations. Here, we analyze metagenomic data from 705 vaginal samples collected during pregnancy from 40 women who delivered preterm spontaneously and 135 term controls from the Multi-Omic Microbiome Study-Pregnancy Initiative. We find that the vaginal microbiome of pregnancies that ended preterm exhibited unique genetic profiles. It was more genetically diverse at the species level, a result which we validate in an additional cohort, and harbored a higher richness and diversity of antimicrobial resistance genes, likely promoted by transduction. Interestingly, we find that Gardnerella species drove this higher genetic diversity, particularly during the first half of the pregnancy. We further present evidence that Gardnerella spp. underwent more frequent recombination and stronger purifying selection in genes involved in lipid metabolism. Overall, our population genetics analyses reveal associations between the vaginal microbiome and PTB and suggest that evolutionary processes acting on vaginal microbes may play a role in adverse pregnancy outcomes such as PTB.

A distinct Fusobacterium nucleatum clade dominates the colorectal cancer niche

Article Open access 20 March 2024

Genomic data in the All of Us Research Program

Article Open access 19 February 2024

Blimp-1 and c-Maf regulate immune gene networks to protect against distinct pathways of pathobiont-induced colitis

Article Open access 12 April 2024

Introduction

Preterm birth (PTB), childbirth at <37 weeks of gestation, is the leading cause of neonatal morbidity and mortality¹. Each year, approximately 15 million infants are born preterm globally, over 500,000 of them in the US². Preterm infants are at a high risk of respiratory, gastrointestinal and neurodevelopmental complications³. While a number of maternal, fetal, and environmental factors have been associated with PTB^1,4,5, its etiopathology remains largely unknown, and early diagnosis and effective therapeutics are still lacking.

Over the past decades, growing evidence has pointed to potential involvement of the vaginal microbiome in PTB^6,7,8,9,10. This involvement has so far been mostly characterized as an ecological process, meaning changes in microbial abundances and vaginal community states. For instance, increased richness and diversity of microbial communities and the presence of particular community state types (CST), have been repeatedly associated with PTB^{6,9,11,12,13,14,15}. In addition, vaginal microbiomes of women who delivered preterm appear to be less stable during pregnancy, with some studies reporting a significant decrease in the richness and diversity of these microbial communities during pregnancy^6,12.

Multiple endogenous factors, such as hormonal changes, nutrient availability and microbial interactions, and exogenous factors, such as genital infections, antibiotic treatment and exposure to xenobiotics, could trigger ecological processes and alter the vaginal microbial composition^16,17. These factors may also act as selective pressures that affect genetic variation in the microbial populations that make the vaginal microbiome. Such adaptive evolution in the vaginal environment, even during pregnancy, is highly plausible given the high mutation rates, short generation times, and large population sizes of microbes¹⁸. They are further supported by observations of rapid adaptation to environmental changes in other human-associated microbial ecosystems^19,20,21. The way by which vaginal microbes respond to various selective pressures may, in turn, affect the host, including pregnancy outcomes. Therefore, a comprehensive investigation of the genetic diversity of the vaginal microbiome at the population level, which we term “microdiversity”, and the underlying evolutionary forces that shape it, holds promise for a better understanding of the etiopathology of PTB.

Here, we performed an in-depth population genetics analysis and characterized the population structure of the vaginal microbiome along pregnancy and in the context of preterm birth. We used metagenomic sequencing data from 705 vaginal samples collected longitudinally during pregnancy as part of the Multi-Omic Microbiome Study-Pregnancy Initiative (MOMS-PI⁶). Our analyses include samples from 40 women who subsequently experienced spontaneous preterm birth (sPTB) and 135 women who had a term birth (TB). We show that the vaginal microbiome of pregnancies that ended preterm exhibits higher nucleotide diversity at the species level and higher antimicrobial resistance potential. We find that this higher nucleotide diversity is driven by Gardnerella spp., a group of central vaginal pathobionts, especially during the first half of pregnancy, and suggest that this may be related to optimization of growth rates in this taxon. We further identify a strong association between evolutionary signatures and sPTB in Gardnerella spp., including more frequent homologous recombination and stronger purifying selection. Overall, our results reveal strong associations between the vaginal microbiome and sPTB at the population genetics level, and suggest that evolutionary processes acting on the vaginal microbiome may play a critical role in sPTB, and potentially also in other adverse pregnancy outcomes.

Results

The phylogenetic composition of the vaginal microbiome associates with sPTB

We assembled a total of 1078 metagenome-assembled genomes (MAGs) with at least medium quality²² (>50% completeness, <10% contamination; Supplementary Data 1; Methods) from previously published⁶ metagenomic reads generated from 705 vaginal samples⁶. These samples were collected from 175 women visiting maternity clinics in Virginia and Washington at various time points along pregnancy⁶, with an average of 3.36 and 3.21 samples for women delivering preterm and at term, respectively (Supplementary Table 1). We clustered these MAGs into 157 species-level phylogroups at the level of 95% average nucleotide identity (ANI), which roughly corresponds to the species level²³; and selected the most complete MAG with the least contamination as the representative for each phylogroup. These representative MAGs were 86 ± 14% (mean ± SD) complete and 1.1 ± 1.8% contaminated, with 93 (59% of 157) of them estimated to have high quality²² (>90% completeness and <5% contamination; Supplementary Data 1). Taxonomic assignment of these representative MAGs (Methods) revealed that the phylogroups represent at least 8 phyla, with genome size (adjusted by completeness) ranging from 0.6 to 7.4 Mbps and GC content ranging from 25.3% to 69.7% (Fig. 1a, Supplementary Data 1). Actinobacteria had the most phylogroups detected in the samples, followed by Firmicutes and Bacteroidetes (Fig. 1a).

**Fig. 1: The composition of the vaginal microbiome associated with sPTB.**

Of note, 12 of these species-level phylogroups (PG042-PG053) were assigned to Gardnerella vaginalis according to CheckM²⁴, supporting the existence of multiple genotypes at the species level within the ‘species’ G. vaginalis^25,26. To better resolve the classification of these G. vaginalis phylogroups, we compared the average nucleotide identity (ANI) for the representative MAGs of these phylogroups against updated reference genomes for Gardnerella, including G. vaginalis, G. piotti, G. leopoldii, G. swidsinskii, and nine species remained to be characterized (gs2-3 and gs7-13)²⁵; gs-2-3 and gs7-13 correspond to group 2-3 and 7-13 shown in Fig. 1 in ref. ²⁵. The ANI analysis shows that PG043 represents G. vaginalis, PG044 represents G. swidsinskii, PG042 represents G. piotti, and PG046, PG049, PG051 and PG053 represent G. gs7, G. gs8, G. gs13 and G. gs12, respectively (Supplementary Fig. 1). The remaining phylogroups (PG045, PG047, PG048, PG050, and PG052) do not cluster with any reference species and may represent previously undocumented species of Gardnerella. Here, we refer to phylogroups PG042-PG053 as Gardnerella spp.

To understand if the temporal dynamics of the vaginal microbiome is associated with sPTB, we employed compositional tensor factorization (CTF)²⁷ to assess temporal changes to the composition of the microbiome during pregnancy. This analysis shows a significant separation of women by pregnancy outcomes (PERMANOVA F = 8.0492; p = 0.002; Fig. 1b-e) based on the dynamics of their microbiome composition over time (Fig. 1e), and specifically observed for component 1 in the CTF analysis (Mann-Whitney p = 0.015; Fig. 1c). We further found that the top features contributing to this difference belong to Lactobacillus helveticus (PG081), Lactobacillus crispatus (PG080), Lactobacillus gasseri (PG079), and Lactobacillus jensenii (PG076 and PG077) that are associated with TB and Megasphaera genomosp. (PG061), Gardnerella spp. (PG047, PG050, PG052), and Atopobium vaginae (PG041) that are associated with PTB (Fig. 1d); these species were previously found to be associated with pregnancy outcomes^6,9,28. These results suggest that the vaginal microbiome has a different temporal trajectory during pregnancies ending preterm, consistent with previous findings^6,7, and with Gardnerella as an important factor. Overall, our results demonstrate that de-novo metagenomic analysis replicates and expands previous findings with respect to associations between the composition of the vaginal microbiome and sPTB.

Next, we sought to examine the diversity of microbial strains detected within species, and its association with sPTB. We performed the analysis on all phylogroups and found that the strains of M. genomosp. showed significantly higher ANI between women who delivered preterm, compared to a null distribution calculated based on ANIs from any two randomly selected women (Permutation p = 0.002, adjusted P < 0.05; Methods), a relationship not observed between women who delivered at term (p = 0.208; Supplementary Fig. 2). This result indicates that M. genomosp. were more closely related than expected by chance across women who delivered preterm. It suggests that sPTB-associated vaginal conditions across women may be more conserved, harboring a group of significantly closely related M. genomosp. strains, compared to TB-associated vaginal conditions.

The microdiversity of the vaginal microbiome is higher in the first half of pregnancies that ended preterm, driven by Gardnerella species

Human microbes can adapt to host-induced environmental changes (e.g., diet, antibiotics) through genetic variations²⁹. Therefore, the microbial populations of the same species in different hosts can have a different genetic structure, which provides them a competitive advantage. These genetic differences, in turn, may be related to the phenotype of the host. To understand the genetic structure of microbial populations in the vaginal environment and its association with pregnancy outcomes, we calculated the nucleotide diversity for each identified phylogroup. Overall, vaginal microbial populations had a significantly higher genome-wide nucleotide diversity in sPTB than in TB (median along pregnancy; Mann-Whitney p = 0.0073; Fig. 2a). Stratifying by phylogroups, we found that this difference was mainly driven by Gardnerella spp. (p = 0.017; Fig. 2b). G. piotti (PG042), G. swidsinskii (PG044), and G. gs13 (PG051) and a potentially new Gardnerella spp. (PG045), along with a phylogroup of Atopobium vaginae, a suspected vaginal pathobiont³⁰, showed significantly higher genome-wide nucleotide diversity in sPTB (P < 0.05, adjusted P < 0.1 for all; Supplementary Fig. 3a). These results imply that microbial populations composed of more diverse strains from the same species, and particularly Gardnerella spp., are growing in the vaginal environment associated with sPTB.

**Fig. 2: Microdiversity patterns of the vaginal microbiome are associated with sPTB.**

To understand how the nucleotide diversity of Gardnerella spp. changes over time during pregnancy, we analyzed temporal trajectories of term and preterm pregnancies. To this end, we pooled the data from all women in each group, binned pregnancy weeks and used splines to smooth the temporal curves (Methods). We found a significant difference between the temporal trajectories of Gardnerella spp. nucleotide diversity in pregnancies ending at term and preterm (Permutation test <0.001 (ref. ³¹), Wilcoxon signed-rank P < 0.003; Methods; Fig. 2c). Specifically, we found that the nucleotide diversity of Gardnerella spp. increased at the beginning of pregnancies which ended preterm, with a peak at around gestational week 13, and then dropped to its initial value at around gestational week 20 (Fig. 2c). In comparison, nucleotide diversity of Gardnerella spp. in TB remained relatively stable (Fig. 2c). Given that gestational week 20 is the middle of a full-term pregnancy, we subsequently analyzed samples with respect to two time periods - first half (0–19 gestational week) and second half of pregnancy (20–37 gestational week; 37 was chosen to ensure a similar time range for both sPTB and TB). As expected, the nucleotide diversity of Gardnerella spp.in sPTB was also significantly higher in sPTB in the first half of pregnancy (median along first half; Mann-Whitney U p = 0.0091; Fig. 2d), but not in the second half (p = 0.71; Fig. 2e). We further found that nucleotide diversity had a significantly stronger correlation with synonymous mutations than with nonsynonymous mutations across Gardnerella spp. (paired t-test p = 0.0011; Supplementary Fig. 3b), suggesting a more important role of purifying selection in shaping genomic diversity. Overall, these results suggest that genetic diversity of Gardnerella spp. in the first half of pregnancy is important to birth outcomes, and could perhaps be used as a biomarker for early diagnosis of sPTB.

Analyzing microdiversity across all phylogroups in the two halves of pregnancy, we again found that it was significantly higher in sPTB in the first half of pregnancy (median along first half; Mann-Whitney U p = 0.0036; Fig. 2f), but not in the second half (p = 0.16; Fig. 2g). Notably, we were able to replicate this analysis using data from an additional dataset of vaginal metagenomic sequencing from ten pregnant individuals³² (median along first and second half; one-sided Mann-Whitney U p = 0.021 and p = 0.12, respectively; Fig. 2h, i). Finally, clinical interventions for risk of preterm birth (e.g., receiving cerclage or progesterone) may alter the microbiome composition, and thus confound our findings. To examine this potential bias, we repeated our analysis on 161 woman who received neither progesterone nor cerclage. Once more, we found that it was significantly higher in sPTB in the first half of pregnancy (p = 0.028; Supplementary Fig. 4a), but not in the second half (p = 0.12; Supplementary Fig. 4b). Overall, our results demonstrate an increased nucleotide diversity in the vaginal microbiome during pregnancies that ended preterm, in a way that replicates across studies and is not biased by common clinical interventions for prevention of preterm birth.

To understand if any particular genes are driving the association between sPTB and the microdiversity of Gardnerella spp. in the first half of pregnancy, we further analyzed nucleotide diversity at the gene level for these species. We identified 21 and 47 genes (out of 825 and 531) in G. swidsinskii (PG044) and G. vaginalis (PG043), respectively, that showed significantly different nucleotide diversity between sPTB and TB (median along the first half of pregnancy; Mann-Whitney P < 0.05, adjusted P < 0.1 for all). These genes included one gene encoding the putative tail-component of bacteriophage HK97-gp10 (p = 0.0012) and one gene encoding putative AbiEii toxin, Type IV toxin–antitoxin system (p = 5 × 10⁻⁴), which might be involved in the interaction with maternal health^33,34. To further identify what functions were related to these associations, we then performed functional enrichment analysis (Methods) using the eggNOG functional annotation of genes (Supplementary Data 2). We found that the KEGG pathway ‘drug metabolism - other enzymes’ (ko00983) was significantly enriched among genes from G. swidsinskii (PG044) that had significantly higher microdiversity (P < 0.05, adjusted P < 0.1; Fig. 2j). This result suggests that the more diverse gene pool in G. swidsinskii (PG044) detected in sPTB may be associated with adaptation to drugs present in the environment. This may be consistent with our recent finding that xenobiotics detected in the vaginal environment are strongly associated with sPTB³⁵.

To verify that the higher nucleotide diversity we observed in sPTB pregnancies was not caused by sampling or sequencing bias, we compared the read count and quality of MAGs obtained from sPTB and TB samples. If this higher diversity is the result of a higher read count in sPTB samples or more complete MAGs, we would expect read count and MAG completeness to be higher in sPTB samples. Instead, we found the completeness and contamination of MAGs assembled from sPTB samples were not significantly different from TB (Mann-Whitney p = 0.71 and 0.73, respectively; Supplementary Fig. 5a, b). Next, we assessed the correlation between the number of reads mapped to each phylogroup and its genome-wide diversity. If a higher diversity is caused by more reads mapped to the MAG representing the phylogroup, we would expect a positive correlation between these two measurements. However, only 3 phylogroups (PG064, a Dialister spp.; PG102, a Peptoniphilus spp.; and PG122, a Bradyrhizobium spp., which was likely either a contaminant or misidentified) had a significant positive correlation between read counts and nucleotide diversity (Spearman ρ = 0.70, 0.54, and 0.42, respectively; non-adjusted p = 0.035, 0.024, and 0.00015, respectively). In 98% of phylogroups, we did not observe a statistically significant positive correlation (median [IQR] Spearman correlation of −0.067 [−0.19, −0.13]). None of the four Gardnerella spp. Phylogroups that showed significantly higher nucleotide diversity in sPTB pregnancies in Supplementary Fig. 3 were significantly positively correlated (Spearman ρ = −0.00, −0.12, −0.02, and −0.35 and p = 0.96, 0.091, 0.77, and 0.00045 for PG042, PG044, PG045, and PG051, respectively; Supplementary Fig. 5c).

Finally, as we have observed a significantly higher read count in sPTB samples (Mann-Whitney p = 0.0004 and 0.061 for samples and subjects, respectively; Supplementary Fig. 5d, e, respectively), we subsampled an identical number of reads (10⁵) from each sample, retaining 75% of samples, and repeated our analyses of nucleotide diversity. As with the first analysis (Fig. 2), nucleotide diversity was significantly higher in sPTB pregnancies across all phylogroups, and particularly in Gardnerella spp.(Mann-Whitney p = 0.015 and p = 0.0043, respectively; Supplementary Fig. 5f, g, respectively). Similarly, in the first half of pregnancy, the nucleotide diversity of Gardnerella spp.was significantly higher in sPTB (p = 0.026; Supplementary Fig. 5h), while in the second half, there was no significant difference (p = 0.22; Supplementary Fig. 5i). We further subsampled an identical number of reads (5000) mapped to Gardnerella spp. from each sample and repeated the analysis in Fig. 2b. A significant higher nucleotide diversity in sPTB pregnancies in Gardnerella spp. is still detected (p = 0.028; Supplementary Fig. 5j), indicating this association is not biased by higher coverage of reads mapped to Gardnerella spp. Overall, these results confirm that the sPTB-associated nucleotide diversity we observed was not biased by technical artifacts.

Evolutionary forces acting on Gardnerella species are associated with pregnancy outcomes

Adaptation should increase the fitness of an organism, its ability to survive and reproduce in a given environment. To better understand if the Gardnerella spp. populations with higher genetic diversity grow better in the vaginal environment associated with sPTB, we inferred fitness using two measures: relative abundance and growth rate. Indeed, we found that nucleotide diversity in these species was positively correlated with relative abundance (Spearman ρ = 0.35, p = 0.0013; Fig. 3a). This correlation was not observed in other phylogroups (Supplementary Fig. 6a). L. crispatus (PG080) and L. iners (PG086) even showed a significantly negative correlation (ρ = −0.39 and −0.32, p = 0.026 and 0.0014, respectively; Supplementary Fig. 6b, c). We additionally used gRodon³⁶ to predict the maximal growth rate of microbes based on codon usage bias in highly expressed genes encoding ribosomal proteins. We found that in the first half of pregnancy, Gardnerella spp. had a somewhat higher, albeit not statistically significant, maximal growth rate in sPTB pregnancies (Mann-Whitney p = 0.057; Fig. 3b), while in the second half of pregnancy, the difference was diminished (p = 0.15; Fig. 3c). No significant difference in the maximal growth rate was observed in non-Gardnerella phylogroups (P > 0.05 for all). The in situ growth rates^37,38 of Gardnerella were not significantly different between the groups (p = 0.14). These results suggest that the sPTB-associated genetic diversity observed in Gardnerella spp. may be related to the optimization for faster growth in the sPTB-associated vaginal environment.

**Fig. 3: Evolutionary forces on *Gardnerella* spp.**

Microbial population structure is influenced by various evolutionary processes including selection and homologous recombination³⁹. Competence, a mechanism of horizontal gene transfer which involves homologous recombination, has been identified in Gardnerella spp⁴⁰. To better interpret the significant differences we observed in the microdiversity patterns of Gardnerella spp. between sPTB and TB, we quantified the degree of homologous recombination using the normalized coefficient of linkage disequilibrium between alleles at two loci, D’. A value of D’ closer to 0 indicates a higher degree of recombination⁴¹. Interestingly, we found that the median D’ of Gardnerella spp. was significantly smaller in sPTB pregnancies in both the first (Mann-Whitney p = 0.041; Fig. 3d) and second halves of pregnancy (p = 0.013; Fig. 3e), and the same was also observed for the D’ of three specific Gardnerella spp., G. piotti (PG042), G. gs7 (PG046), and PG047, in the first half of pregnancy (P < 0.05, adjusted P < 0.1 for all; Supplementary Fig. 7a). No significant difference in recombination was observed in non-Gardnerella phylogroups (adjusted P > 0.1 for all). These results suggest that Gardnerella spp. tends to have more frequent recombination in women who delivered preterm during both halves of pregnancy.

Next, we quantified the degree of selection using dN/dS in this species (Methods). This measure quantifies the ratio between synonymous and non-synonymous mutations, and hence offers insight into the type of selection, with values close to zero indicating purifying selection, and values higher than one indicating positive selection⁴². dN/dS is calculated in relation to the reference, and can therefore detect selection on mutations that have already been fixed within the population⁴³. Consistent with the gut and ocean microbiomes^44,45,46, purifying selection is predominant across all genes of the vaginal microbiome (dN/dS « 1; median [IQR] dN/dS of 0.17 [0.10, 0.29]; Supplementary Fig. 7b). While the median dN/dS of all Gardnerella spp. genes was not significantly different between sPTB and TB pregnancies (Mann-Whitney U p = 0.48), we detected some differences when examining high-level functions (COG categories⁴⁷) within each half of pregnancy. In the first half, the median dN/dS of Gardnerella spp. genes was somewhat lower in sPTB pregnancies for inorganic ion transport and metabolism, lipid transport and metabolism, secondary structure, and cell wall/membrane/envelope biogenesis, though this was not statistically significant after adjusting for multiple testing (COG categories “P”, “I”, “Q”, and “M”, respectively; Mann-Whitney P < 0.05, adjusted P > 0.1 for all; Supplementary Fig. 7c). In the second half of pregnancy, the median dN/dS was significantly lower in sPTB pregnancies for lipid transport and metabolism and cell motility (COG categories “I” “N”; Mann-Whitney p = 0.0040 and p = 0.04, adjusted p = 0.07 and 0.40, respectively; Fig. 3f). No significant difference in the selection based on dN/dS was observed in non-Gardnerella phylogroups (P > 0.05 for all). Our results suggest that Gardnerella spp. genes involved in lipid transport and metabolism may undergo stronger purifying selection in sPTB. As purifying selection maintains the fitness of organisms by constantly sweeping away deleterious mutations and conserving functions, Gardnerella spp. may benefit from this stronger purifying selection targeting lipid functioning when growing in the sPTB-associated vaginal environment during pregnancy.

sPTB-associated vaginal microbiomes have a higher antibiotic-resistance potential

Antibiotics are widely used during pregnancy, sometimes even topically in the vagina⁴⁸. This exposure may promote antimicrobial resistance (AMR). To assess if antibiotic-resistance potential in the vaginal microbiome is associated with sPTB, we subsampled an identical number of reads (10⁵) from each sample and mapped them to the Comprehensive Antibiotic Resistance Database⁴⁹. The total number of reads mapped to AMR reference genes was significantly higher in the first half of sPTB pregnancies (Mann-Whitney U p = 0.015; Fig. 4a), but not in the second half (p = 0.76; Fig. 4b). In addition, to assess the difference of specific AMR genes between the vaginal microbiomes of sPTB and TB, we identified AMR genes in the genomic assemblies. A significantly higher median count and Shannon-Wiener diversity of AMR genes were detected in vaginal microbes sampled at the first half of pregnancies that ended preterm (3-times higher on average; Mann-Whitney U p = 0.011 and p = 0.0078, respectively; Fig. 4c, e, respectively), yet this difference was not detected in the second half (p = 0.16 for both; Fig. 4d, f, respectively). Exploring the source of these genes, we found a significantly higher median fraction of phage-borne AMR genes in the microbiomes of women who delivered preterm (one-sided p = 0.016, adjusted p = 0.093.; Fig. 4g), suggesting transduction may promote the higher median count and diversity of AMR genes observed in the first half of sPTB pregnancies (Fig. 4c, e). Among the 9 AMR gene categories that had genes present in at least 10% of women, phenicol, aminoglycoside, glycopeptide, and MLS resistance genes showed a significantly higher median fraction in the sPTB microbiome (P < 0.05, adjusted P < 0.1 for all; Fig. 4h). These results suggest a unique antibiotic resistance profile associated with the first half of sPTB pregnancies, potentially indicative of usage of specific antibiotics. Indeed, we detected a somewhat higher richness of AMR genes along the first half of pregnancy in the 29 women who used antibiotics in the past 6 months before pregnancy than those who did not (Mann-Whitney U p = 0.079; Fig. 4i). This is also consistent with our observation that genes with sPTB-associated nucleotide diversity were enriched for drug metabolism in G. swidsinskii (PG044) (Fig. 2d).

**Fig. 4: Antimicrobial resistance (AMR) gene profiles of the vaginal microbiome are associated with sPTB.**

As a higher risk of preterm birth has been reported to be associated with bacterial vaginosis (BV)^50,51,52, and antibiotics used to treat BV may change the AMR genetic profiles of the vaginal microbiome, our findings in AMR genes might be potentially confounded by BV. To assess this potential bias, we compared AMR gene count between women with and without BV as well as between women with and without BV history. We found that the AMR gene count was not significantly different between women with and without BV for both halves of pregnancy (p = 0.38 and 0.5, respectively; Supplementary Fig. 8a). Similarly, it was not different between women with and without BV history (p = 0.45 and 0.1, respectively; Supplementary Fig. 8b). These results suggest that the association between AMR gene and pregnancy outcomes is independent of BV.

To check if the strong association between sPTB and the AMR potential of the vaginal microbiome is contributed by a particular phylogroup, we performed a similar analysis for each phylogroup. We found, however, that none of them showed a significant difference in the median count and diversity of AMR genes between sPTB and TB (Mann-Whitney U P > 0.05 for all). This result suggests that the higher AMR potential associated with sPTB may be a property of the vaginal microbiome as an ecosystem. However, this lack of association could also be driven by underestimation of AMR genes due to the limitation of MAG binning methods in recovering mobile genetic elements⁵³.

Discussion

Microbial genomes can exhibit large variations even within the same species, as a result of adaptation to various environments⁵⁴. Associations between the vaginal microbiome and preterm birth have been widely reported^7,8,12,35,52. However, there is still much left to explore regarding potential mechanisms underlying host-microbiome interactions in this context. Here, by leveraging publicly available metagenomic data⁶, we provide a population genetic view of the vaginal microbiome during pregnancy. We identify a number of microbial features including population nucleotide diversity, selection metrics, and antibiotic resistance potential that are associated with sPTB. Interestingly, we find that the higher population nucleotide diversity is driven by Gardnerella spp. during the first half of pregnancy. This species appears to undergo more intense changes in the population structure contributed by recombination and purifying selection in pregnancies which ended preterm. We also show evidence that this sPTB-associated genetic pattern of Gardnerella spp. may be related to optimization of growth rates in vaginal conditions linked to sPTB. Our results are indicative of adaptation of the vaginal microbiota to the host, which in turn may influence pregnancy outcomes.

Our findings regarding a relationship between ecological processes in the pregnancy vaginal microbiome and subsequent preterm birth are consistent with previous studies^{6,7,11,12,14,52,55}. We add to these studies by exploring an additional layer of microbial variability associated with sPTB - microbial genetic diversity. It is known that genomic variation within species can result in phenotypic diversity and adaptations to different environments⁵⁴. These adaptations, in turn, can affect host phenotypes such as disease outcomes⁵⁶. Such associations between microbial genomic variation and host phenotypes have been reported in the gut microbiome^45,57,58. Our study suggests that this phenomenon also occurs in the vaginal ecosystem, and that it may be associated with pregnancy outcomes. Nethertheless, the associations between microbial genetic diversity and pregnancy outcomes we detect might also be a consequence of different processes or unmeasured confounders that act on both variables (e.g., certain drugs or exogenous chemical compounds are that drive inflammation), and while we find this unlikely, this should be determined by future studies.

Interestingly, we found that the association of genetic diversity and sPTB was largely driven by Gardnerella spp., a group of species commonly associated with BV^50,51,52. A number of studies reported a higher abundance of these species in sPTB pregnancies^6,9,52,59,60. A recent preprint also demonstrated a higher number of Gardnerella clades in sPTB⁶¹. We show that Gardnerella spp. populations with more genetically diverse strains may also be associated with sPTB. In addition, we found that this taxon has the capacity to grow 1.5 times faster in pregnancies that ended preterm, consistent with an overall higher relative transcriptional rate of G. vaginalis which was previously reported⁶. These more genetically diverse strains appear to have adapted to the vaginal environment associated with sPTB, exhibiting higher fitness. Notably, the higher genetic diversity associated with sPTB in Gardnerella spp. was detected during the first half of the pregnancy (<20 gestational week) rather than the second half. The enriched nucleotide diversity during the first half of pregnancy might be related to the change of human chorionic gonadotropin (HCG), which peaks at roughly the same time as G. vaginalis microdiversity as we observed in Fig. 2c, and was suggested to play an immunomodulatory role in humans^62,63. To our knowledge, however, the effect of HCG on the vaginal ecosystem is not well established yet. Most potential biomarkers of sPTB (e.g., serum alpha-fetoprotein⁶⁴ were so far identified using samples from the second trimester of pregnancy (gestational week 14–27). Our results suggest that high resolution analysis of microbiome samples from even earlier stages of pregnancy (<week 20) may yield informative biomarkers of pregnancy outcomes.

As in the human gut microbiome^19,20,21, we show evidence that adaptive evolution also occurs in the vaginal microbiome. Several environmental factors affecting the vaginal ecosystem, such as pH, neutrophil levels, and xenobiotics, have been reported to be associated with sPTB^35,65. These environmental factors may act as selective stressors that lead to different evolutionary patterns in the vaginal microbiome. Indeed, we detected more frequent homologous recombination and stronger purifying selection within Gardnerella spp. during pregnancies that end preterm. Homologous recombination is a critical mechanism speeding adaptation by increasing fixation probability of beneficial mutations⁶⁶ and reducing clonal interference (i.e., competition between beneficial mutations) in bacteria⁶⁷. Purifying selection also contributes to adaptation by sweeping away deleterious mutations and conserving functions, such as in oligotrophic nutrient conditions^44,68. Notably, we found that sPTB-associated purifying selection is particularly strong on genes involved in lipid transportation and metabolism. This is consistent with previous identification of lipid metabolites (e.g., monoacylglycerols and sphingolipids) as signatures of sPTB^35,69. Whether this stronger purifying selection targeting lipid transportation and metabolisms in pregnancy that ended preterm leads to changes in the concentrations of lipid metabolites however requires further experimental testing. As both recombination and purifying selection can reduce genetic diversity, sPTB-associated recombination and purifying selection along pregnancy may explain the higher nucleotide diversity of Gardnerella spp. in sPTB in the first half of pregnancy compared to the second half.

Antibiotics are common selective stresses acting on the human microbiome²⁹ and have been associated with preterm birth⁷⁰. We detected higher count and diversity of AMR genes associated with sPTB, which our analysis suggests to be facilitated by prophages in preterm vaginal microbiomes. While multiple phages (e.g., Siphoviridae, Myoviridae, and Microviridae) have been detected in the vagina of pregnant women, their association with sPTB is rarely studied⁷¹. Our results imply a potentially important role of bacteria-phage interactions in pregnancy outcomes via transferring of AMR genes. We also found that genes related to phenicol and aminoglycoside resistance were more abundant in vaginal microbiomes during pregnancies that ended preterm. While both antibiotics have been frequently used to treat gynecologic infection for decades⁷², and some phenicols (e.g., chloramphenicol) are thought to be safe for use during pregnancy⁷³ aminoglycoside is teratogenic. Previous studies reported that exposure to antibiotics could change the composition of the vaginal microbiome^48,74, indicating an ecological effect. In comparison, our results may suggest adaptation of the vaginal microbiome to more frequent antibiotics usage in women who delivered preterm, leading to an enrichment of AMR genes as well as higher nucleotide diversity in Gardnerella spp. genes encoding enzymes for drug metabolism. While this hypothesis requires further study, it is further supported by the fact that a higher proportion of women who delivered preterm (31%) had used antibiotics in the past 6 months before pregnancy than women who delivered at term (23%).

Despite its findings, our study is limited by low sequencing depth (median bacterial read count <5 × 10⁵) and inconsistent sampling frequency during pregnancy (1 to 8 samples per pregnancy, with an average of 4). These limitations lead to high sparsity in the features analyzed, preventing a more in-depth temporal and predictive analysis of the link between population genetics of the vaginal microbiome and sPTB. Our results warrant a high-resolution investigation of the vaginal metagenome, with frequent sampling and high sequencing depth.

In summary, through in-depth population genomic analyses, our study identified genetic and functional associations between the vaginal microbiome and preterm birth. We revealed evidence of microbial genetic adaptation to the host environment linked to preterm birth and highlighted the importance of microbial evolutionary processes to adverse pregnancy outcomes, particularly in Gardnerella spp. Future investigation on the pressures driving the sPTB-associated microbial adaptation is warranted to fully understand the molecular mechanisms underlying preterm birth.

Methods

Sample selection and metagenomic data

We analyzed published metagenomic sequencing data from the Multi-Omic Microbiome Study: Pregnancy Initiative (MOMS-PI) PTB case-control study⁶ as well as publicly available data from ref. ³² for validation. The metagenomic sequencing data and clinical covariates from ref. ⁶ were obtained from dbGaP (study no. 20280; accession ID phs001523.v1.p1) and directly from Virginia Commonwealth University. MOMS-PI participants provided written informed consent allowing for their data to be shared and used in future research studies. The analyses performed here were approved by the IRB of Columbia University, approval number AAAS5367.

MOMS-PI recruited a majority of women identifying as Black as described in ref. ⁶. We obtained 135 vaginal samples collected longitudinally during pregnancy from 40 women who eventually delivered preterm spontaneously (sPTB; mean ± sd) age 26 ± 5.68) and 570 vaginal samples from 135 women who delivered at term (TB; age 25.9 ± 5.43). Samples were sequenced paired-end, to a mean ± sd depth of 717,887 ± 1,536,354 (mean ± s.d.) non-human reads. Fettweis et al. 2019⁶ only included in this dataset term births after 39 weeks of gestation, with the intention of avoiding complications associated with early term birth⁶. Our study therefore uses the same definitions: spontaneous preterm birth is defined as live birth between 23 and 37 gestational weeks without medical indication, and term birth is defined as live birth at or after 39 gestational weeks.

On average, 3.36 and 3.21 samples were collected for each sPTB and TB woman, respectively (Mann-Whitney U p = 0.83); 1.68 and 1.51 samples were collected for the first half of pregnancy for each sPTB and TB woman, respectively (p = 0.30); and 2.34 and 2.14 samples were collected for the second half of pregnancy for each sPTB and TB woman, respectively (p = 0.27) (Supplementary Table 1). The average gestational age at the first sample being collected for sPTB and TB women is 17.38 and 16.09, respectively (p = 0.45), while for the last sample being collected for sPTB and TB women is 31.23 and 32.31, respectively (p = 0.72) (Supplementary Table 1). In addition, none of these women delivered at <20 gestational weeks (Supplementary Fig. 9a). These indicate that dropping out due to late miscarriage/early PTB is not a concern to bias our findings in this study.

To check for the presence of some potential confounders for vaginal microbiome-sPTB associations, we calculated propensity scores⁷⁵ for each subject based on income, age, and race using a logistic regression model. We found that propensity scores for both sPTB and TB subjects exhibited a similar distribution (Kolmogorov–Smirnov test p = 0.21), suggesting the associations we detect are not likely to be confounded with these variables (Supplementary Fig. 9b). These results suggest a negligible confounding effect of income, age, and race in this study on microbiome-sPTB associations. We note, however, that population studies such as the one performed by Fettweis et al.⁶ can be exposed to selection bias, via access to medical care and other reasons. Experimental procedures for data generation are described by Fettweis et al⁶.

Metagenomic assembly, genomic binning, genome annotation, and relative abundance

Our analysis follows the accepted standards used in refs. ^76,77,78,79, using the ATLAS pipeline⁸⁰ (v. 2.4.4). Bases with quality scores <25, raw reads <50 bp lengths, and sequencing adapters were removed using Trimmomatic v.0.39⁸¹. Reads mapped to human and PhiX genome sequences were removed by mapping with Bowtie2 v.2.3.5.1⁸². Assembly and binning were done with ATLAS: filtered reads were assembled using metaSPAdes v.3.15.2⁸³, and contigs were binned into metagenome-assembled genomes (MAGs) using MetaBAT2 v.2.14.0 (ref. ⁸⁴) with a minimum contig length of 1500. Quality, GC content, genome size, and taxonomy of MAGs were estimated using CheckM v.1.0.9²⁴. MAGs were de-replicated using dRep v.3.2.0⁸⁵ with an average nucleotide identity (ANI) of 0.95, minimum completeness of 50%, and maximum genome contamination of 10%. The MAG with the highest dRep score within each 95% ANI cluster, termed here as a phylogroup, was selected as the representative MAGs for that phylogroup. Genes were predicted using Prodigal v.2.6.3⁸⁶ and annotated using EggNOG v.5.0⁸⁷. Filtered reads were mapped to representative MAGs using Bowtie2 v.2.3.5.1⁸⁸. The relative abundance of each representative MAG was calculated by dividing the number of reads that mapped to that MAG, corrected to the genome size and completeness, by the total number of reads in each sample.

Phylogeny, ANI, and dendrogram

Amino acid (AA) sequences of 120 marker genes were called and aligned for representative MAGs using GTDB-Tk v.1.5.1 (ref. ⁸⁹). MAGs with <60% of AA in the alignment were excluded in the phylogenetic tree construction. The best evolutionary model LG + G + I (the Le Gascuel model + gamma distribution + invariant sites) was identified using prottest3 v.3.4.2 (ref. ⁹⁰) and 500 bootstraps were used for tree construction using RAxML v.8.2.12 (ref. ⁹¹). The tree was rooted by midpoint and visualized in iTol v.6.3 (ref. ⁹²).

Pairwise ANI of MAGs for each phylogroup as well as for representative MAGs annotated as G. vaginalis (PG42-53) and reference genomes of 13 Gardnerella species defined in ref. ²⁵ was calculated using pyani v.0.2. Dendrogram of ANI was constructed using the complete-linkage clustering method in the vegan package (v.2.6-4) in R v.3.6.0.

Microdiversity profiling, growth rate estimation, and antimicrobial resistance genes

Population microdiversity metrics including genome-wide nucleotide diversity, gene-wide nucleotide diversity, linkage disequilibrium measures (D’) and selection measures (dN/dS), were calculated using InStrain v1.0.0⁴³ using the 157 representative MAGs as the reference database. Maximal growth rate was estimated for each MAG using gRodon³⁶ v.1. Antimicrobial resistance (AMR) genes were detected in assemblies and MAGs using PathoFact v.1.0 (ref. ⁹³) with default parameters.

Temporal analysis

To generate the trajectories representing the change in nucleotide diversity over time in term and preterm deliveries, we pooled the temporal data of Gardnerella spp. from all women in each group (term and preterm). When we had more than one observation per gestational week, we took the median value across samples. We then binned the temporal data into bins of 3 weeks, except for the first bin which spanned weeks 1-7, and took the median of each bin as a summary. To smooth the observed binned data we applied splines, which is a special function defined piecewise by polynomials for data smoothing. To compare between the temporal trajectories of preterm and term, we performed a permutation test, in which we generated a null distribution of euclidean distances by shuffling the these trajectories 10⁴ times and comparing to the euclidean distance in the original data³¹.

Functional enrichment analysis

To identify COG/KEGG pathways that were enriched in genes showing significant difference in nucleotide diversity between sPTB and TB, the frequency of each COG/KEGG category was first calculated from significant genes (observed frequency). Then, the frequency of each COG/KEGG category was calculated from an identical number of genes randomly selected from all genes (expected frequency). This process was repeated 10,000 times. The null hypothesis was that the observed frequency of COG category is smaller than the expectation. For each COG, probability P of the null hypothesis was calculated using the formula: p = |[x_i ∈ x: x_i > = k]| / 10000, where […] denotes a multiset, x = (x₁, x₂, …, x_n) is a list of expected values, and k is the observed value.

Statistics & reproducibility

This was a secondary analysis of data obtained from a case-control analysis of an observational cohort⁶. Sample size follows the original study and no sample size calculation was performed for this study. No data was excluded from this analysis. As this was a secondary analysis of an observational study, there was no allocation to intervention, and hence no randomization or blinding to allocation. A STORMS checklist is provided as Supplementary Data 3. A different number of samples was available for each woman in the database. In our analyses, we therefore used the median along pregnancy (or its first or second half). The false-discovery rate procedure (FDR) of Benjamini and Hochberg (BH)⁹⁴ was used to correct for multiple testing. Adjusted P < 0.1 was used as the significance cutoff.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The raw sequencing data and metadata were obtained from ref. ⁶. Data are available under restricted access for ethical and privacy concerns from dbGAP under accession phs001523. Access can be obtained by a data access request to dbGaP, under the purview of the data access committee of the Eunice Kennedy Shriver National Institute of Child Health and Human Development [HD-DAC@mail.nih.gov]. All processed features generated in this study are available⁹⁵ at Zenodo, https://doi.org/10.5281/zenodo.8150902. The validation dataset³² is available in SRA, under accession PRJNA288562.

Code availability

Code to replicate all analyses is available⁹⁶ from https://github.com/korem-lab/MOMs-PI_microdiversity_2023.

References

Tiensuu, H. et al. Risk of spontaneous preterm birth and fetal growth associates with fetal SLIT2. PLoS Genet. 15, e1008107 (2019).
Article CAS PubMed PubMed Central Google Scholar
Walani, S. R. Global burden of preterm birth. Int. J. Gynaecol. Obstet. 150, 31–33 (2020).
Article PubMed Google Scholar
Goldenberg, R. L., Culhane, J. F., Iams, J. D. & Romero, R. Epidemiology and causes of preterm birth. Lancet 371, 75–84 (2008).
Article PubMed PubMed Central Google Scholar
Hong, X. et al. Genome-wide approach identifies a novel gene-maternal pre-pregnancy BMI interaction on preterm birth. Nat. Commun. 8, 15608 (2017).
Article ADS PubMed PubMed Central Google Scholar
Hong, X. et al. Genome-wide association study identifies a novel maternal gene × stress interaction associated with spontaneous preterm birth. Pediatr. Res. 89, 1549–1556 (2021).
Fettweis, J. M. et al. The vaginal microbiome and preterm birth. Nat. Med. 25, 1012–1021 (2019).
Article CAS PubMed PubMed Central Google Scholar
DiGiulio, D. B. et al. Temporal and spatial variation of the human microbiota during pregnancy. Proc. Natl Acad. Sci. USA. 112, 11060–11065 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Ravel, J. et al. Vaginal microbiome of reproductive-age women. Proc. Natl Acad. Sci. USA. 108, Suppl 14680–4687 (2011).
Article ADS CAS PubMed Google Scholar
Tabatabaei, N. et al. Vaginal microbiome in early pregnancy and subsequent risk of spontaneous preterm birth: a case-control study. BJOG 126, 349–358 (2019).
Article CAS PubMed Google Scholar
Chu, D. M., Seferovic, M., Pace, R. M. & Aagaard, K. M. The microbiome in preterm birth. Best. Pract. Res. Clin. Obstet. Gynaecol. 52, 103–113 (2018).
Article PubMed Google Scholar
Freitas, A. C., Bocking, A., Hill, J. E. & Money, D. M. & VOGUE Research Group. Increased richness and diversity of the vaginal microbiota and spontaneous preterm birth. Microbiome 6, 117 (2018).
Article PubMed PubMed Central Google Scholar
Stout, M. J. et al. Early pregnancy vaginal microbiome trends and preterm birth. Am. J. Obstet. Gynecol. 217, 356.e1–356.e18 (2017).
Article PubMed Google Scholar
Hyman, R. W. et al. Diversity of the vaginal microbiome correlates with preterm birth. Reprod. Sci. 21, 32–40 (2014).
Article PubMed PubMed Central Google Scholar
Feehily, C. et al. Shotgun sequencing of the vaginal microbiome reveals both a species and functional potential signature of preterm birth. NPJ Biofilms Microbiomes 6, 50 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kosti, I., Lyalina, S., Pollard, K. S., Butte, A. J. & Sirota, M. Meta-Analysis of Vaginal Microbiome Data Provides New Insights Into Preterm Birth. Front. Microbiol. 11, 476 (2020).
Article PubMed PubMed Central Google Scholar
Gupta, P., Singh, M. P. & Goyal, K. Diversity of Vaginal Microbiome in Pregnancy: Deciphering the Obscurity. Front. Public Health 8, 326 (2020).
Article PubMed PubMed Central Google Scholar
Ceccarani, C. et al. Diversity of vaginal microbiome and metabolome during genital infections. Sci. Rep. 9, 14095 (2019).
Article ADS PubMed PubMed Central Google Scholar
Chase, A. B., Weihe, C. & Martiny, J. B. H. Adaptive differentiation and rapid evolution of a soil bacterium along a climate gradient. Proc. Natl. Acad. Sci. USA. 118, e2101254118 (2021).
Zhao, S. et al. Adaptive Evolution within Gut Microbiomes of Healthy People. Cell Host Microbe 25, 656–667.e8 (2019).
Article CAS PubMed PubMed Central Google Scholar
Garud, N. R. & Pollard, K. S. Population Genetics in the Human Microbiome. Trends Genet. 36, 53–67 (2020).
Article CAS PubMed Google Scholar
Garud, N. R., Good, B. H., Hallatschek, O. & Pollard, K. S. Evolutionary dynamics of bacteria in the gut microbiome within and across hosts. PLoS Biol. 17, e3000102 (2019).
Article PubMed PubMed Central Google Scholar
Murovec, B., Deutsch, L. & Stres, B. Computational Framework for High-Quality Production and Large-Scale Evolutionary Analysis of Metagenome Assembled Genomes. Mol. Biol. Evol. 37, 593–598 (2020).
Article CAS PubMed Google Scholar
Olm, M. R. et al. Consistent Metagenome-Derived Metrics Verify and Delineate Bacterial Species Boundaries. mSystems 5, e00731–19 (2020).
Article CAS PubMed PubMed Central Google Scholar
Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. & Tyson, G. W. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–1055 (2015).
Article CAS PubMed PubMed Central Google Scholar
Vaneechoutte, M. et al. Emended description of Gardnerella vaginalis and description of Gardnerella leopoldii sp. nov., Gardnerella piotii sp. nov. and Gardnerella swidsinskii sp. nov., with delineation of 13 genomic species within the genus Gardnerella. Int. J. Syst. Evol. Microbiol. 69, 679–687 (2019).
Article CAS PubMed Google Scholar
Hill, J. E. & Albert, A. Y. K. & the VOGUE Research Group. Resolution and Cooccurrence Patterns of Gardnerella leopoldii, G. swidsinskii, G. piotii, and G. vaginalis within the Vaginal Microbiome. Infect. Immun. 87, e00532–19 (2019).
Article CAS PubMed PubMed Central Google Scholar
Martino, C. et al. Context-aware dimensionality reduction deconvolutes gut microbial community dynamics. Nat. Biotechnol. 39, 165–168 (2021).
Article CAS PubMed Google Scholar
Mendz, G. L., Petersen, R., Quinlivan, J. A. & Kaakoush, N. O. Potential involvement of Campylobacter curvus and Haemophilus parainfluenzae in preterm birth. BMJ Case Rep. 2014, bcr2014205282 (2014).
Article PubMed PubMed Central Google Scholar
Suzuki, T. A. & Ley, R. E. The role of the microbiota in human genetic adaptation. Science 370, eaaz6827 (2020).
Ferris, M. J. et al. Association of Atopobium vaginae, a recently described metronidazole resistant anaerobe, with bacterial vaginosis. BMC Infect. Dis. 4, 5 (2004).
Article PubMed PubMed Central Google Scholar
Danielsson, P.-E. Euclidean distance mapping. Computer Graph. Image Process. 14, 227–248 (1980).
Article Google Scholar
Goltsman, D. S. A. et al. Metagenomic analysis with strain-level resolution reveals fine-scale variation in the human pregnancy microbiome. Genome Res. 28, 1467–1480 (2018).
Article CAS PubMed PubMed Central Google Scholar
Manrique, P., Dills, M. & Young, M. J. The Human Gut Phage Community and Its Implications for Health and Disease. Viruses 9, 141 (2017).
Article PubMed PubMed Central Google Scholar
Schwebke, J. R., Muzny, C. A. & Josey, W. E. Role of Gardnerella vaginalis in the pathogenesis of bacterial vaginosis: a conceptual model. J. Infect. Dis. 210, 338–343 (2014).
Article PubMed Google Scholar
Kindschuh, W. F. et al. Preterm birth is associated with xenobiotics and predicted by the vaginal metabolome. Nat. Microbiol 8, 246–259 (2023).
Article CAS PubMed PubMed Central Google Scholar
Weissman, J. L., Hou, S. & Fuhrman, J. A. Estimating maximal microbial growth rates from cultures, metagenomes, and single cells via codon usage patterns. Proc. Natl Acad. Sci. USA. 118, e2016810118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Korem, T. et al. Growth dynamics of gut microbiota in health and disease inferred from single metagenomic samples. Science 4, 1101–1106 (2015).
Article ADS Google Scholar
Joseph, T. A., Chlenski, P., Litman, A., Korem, T. & Pe’er, I. Accurate and robust inference of microbial growth dynamics from metagenomic sequencing reveals personalized growth rates. Genome Res. 32, 558–568 (2022).
Article PubMed PubMed Central Google Scholar
Achtman, M. & Wagner, M. Microbial diversity and the genetic nature of microbial species. Nat. Rev. Microbiol. 6, 431–440 (2008).
Article CAS PubMed Google Scholar
Bohr, L. L., Mortimer, T. D. & Pepperell, C. S. Lateral Gene Transfer Shapes Diversity of spp. Front. Cell. Infect. Microbiol. 10, 293 (2020).
Article CAS PubMed PubMed Central Google Scholar
Hudson, R. R. Linkage disequilibrium and recombination. Handbook of statistical genetics (John Wiley & Sons, Ltd, 2004).
Kryazhimskiy, S. & Plotkin, J. B. The population genetics of dN/dS. PLoS Genet. 4, e1000304 (2008).
Article PubMed PubMed Central Google Scholar
Olm, M. R. et al. inStrain profiles population microdiversity from metagenomic data and sensitively detects shared microbial strains. Nat. Biotechnol. 39, 727–736 (2021).
Article CAS PubMed PubMed Central Google Scholar
Shenhav, L. & Zeevi, D. Resource conservation manifests in the genetic code. Science 370, 683–687 (2020).
Article ADS CAS PubMed Google Scholar
Schloissnig, S. et al. Genomic variation landscape of the human gut microbiome. Nature 493, 45–50 (2013).
Article ADS PubMed Google Scholar
He, M. et al. Evolutionary dynamics of Clostridium difficile over short and long time scales. Proc. Natl Acad. Sci. USA. 107, 7527–7532 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Tatusov, R. L., Galperin, M. Y., Natale, D. A. & Koonin, E. V. The COG database: a tool for genome-scale analysis of protein functions and evolution. Nucleic Acids Res. 28, 33–36 (2000).
Article CAS PubMed PubMed Central Google Scholar
Stokholm, J. et al. Antibiotic use during pregnancy alters the commensal vaginal microbiota. Clin. Microbiol. Infect. 20, 629–635 (2014).
Article CAS PubMed Google Scholar
Alcock, B. P. et al. CARD 2020: antibiotic resistome surveillance with the comprehensive antibiotic resistance database. Nucleic Acids Res. 48, D517–D525 (2020).
CAS PubMed Google Scholar
Pace, R. M. et al. Complex species and strain ecology of the vaginal microbiome from pregnancy to postpartum and association with preterm birth. Med 2, 1027–1049 (2021).
Article CAS PubMed Google Scholar
Brown, R. G. et al. Vaginal dysbiosis increases risk of preterm fetal membrane rupture, neonatal sepsis and is exacerbated by erythromycin. BMC Med. 16, 9 (2018).
Article PubMed PubMed Central Google Scholar
Callahan, B. J. et al. Replication and refinement of a vaginal microbial signature of preterm birth in two racially distinct cohorts of US women. Proc. Natl Acad. Sci. USA. 114, 9966–9971 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Maguire, F. et al. Metagenome-assembled genome binning methods with short reads disproportionately fail for plasmids and genomic Islands. Microb Genom 6, mgen000436 (2020).
Liao, J. et al. Nationwide genomic atlas of soil-dwelling Listeria reveals effects of selection and population ecology on pangenome evolution. Nat. Microbiol 6, 1021–1030 (2021).
Article CAS PubMed Google Scholar
Haque, M. M., Merchant, M., Kumar, P. N., Dutta, A. & Mande, S. S. First-trimester vaginal microbiome diversity: A potential indicator of preterm delivery risk. Sci. Rep. 7, 16145 (2017).
Article ADS PubMed PubMed Central Google Scholar
Leung, J. M., Graham, A. L. & Knowles, S. C. L. Parasite-Microbiota Interactions With the Vertebrate Gut: Synthesis Through an Ecological Lens. Front. Microbiol. 9, 843 (2018).
Morowitz, M. J. et al. Strain-resolved community genomic analysis of gut microbial colonization in a premature infant. Proc. Natl Acad. Sci. USA. 108, 1128–1133 (2011).
Zeevi, D. et al. Structural variation in the gut microbiome associates with host health. Nature 568, 43–48 (2019).
Menard, J. P. et al. High vaginal concentrations of Atopobium vaginae and Gardnerella vaginalis in women undergoing preterm labor. Obstet. Gynecol. 115, 134–140 (2010).
Article PubMed Google Scholar
Kumar, S. et al. The Vaginal Microbial Signatures of Preterm Birth Delivery in Indian Women. Front. Cell. Infect. Microbiol. 11, 622474 (2021).
Article CAS PubMed PubMed Central Google Scholar
Berman, H. L., Aliaga Goltsman, D. S., Anderson, M., Relman, D. A. & Callahan, B. J. Gardnerella diversity and ecology in pregnancy and preterm birth. bioRxiv https://doi.org/10.1101/2023.02.03.527032 (2023).
Schumacher, A. et al. Human chorionic gonadotropin attracts regulatory T cells into the fetal-maternal interface during early human pregnancy. J. Immunol. 182, 5488–5497 (2009).
Article CAS PubMed Google Scholar
Polese, B. et al. The Endocrine Milieu and CD4 T-Lymphocyte Polarization during Pregnancy. Front. Endocrinol. 5, 106 (2014).
Article Google Scholar
Yuan, W., Chen, L. & Bernal, A. L. Is elevated maternal serum alpha-fetoprotein in the second trimester of pregnancy associated with increased preterm birth risk? Eur. J. Obstet. Gynecol. Reprod. Biol. 145, 57–64 (2009).
Article CAS PubMed Google Scholar
Simhan, H. N., Caritis, S. N., Krohn, M. A. & Hillier, S. L. Elevated vaginal pH and neutrophils are associated strongly with early spontaneous preterm birth. Am. J. Obstet. Gynecol. 189, 1150–1154 (2003).
Article PubMed Google Scholar
Otto, S. P. & Barton, N. H. The evolution of recombination: removing the limits to natural selection. Genetics 147, 879–906 (1997).
Article CAS PubMed PubMed Central Google Scholar
Cooper, T. F. Recombination speeds adaptation by reducing competition between beneficial mutations in populations of Escherichia coli. PLoS Biol. 5, e225 (2007).
Article PubMed PubMed Central Google Scholar
Martinez-Gutierrez, C. A. & Aylward, F. O. Strong Purifying Selection Is Associated with Genome Streamlining in Epipelagic Marinimicrobia. Genome Biol. Evol. 11, 2887–2894 (2019).
Article CAS PubMed PubMed Central Google Scholar
Gerson, K. D. et al. A non-optimal cervicovaginal microbiota in pregnancy is associated with a distinct metabolomic signature among non-Hispanic Black individuals. Sci. Rep. 11, 22794 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Terzic, M. et al. Periodontal Pathogens and Preterm Birth: Current Knowledge and Further Interventions. Pathogens 10, 730 (2021).
Article CAS PubMed PubMed Central Google Scholar
da Costa, A. C. et al. Identification of bacteriophages in the vagina of pregnant women: a descriptive study. BJOG 128, 976–982 (2021).
Article PubMed Google Scholar
Bargaza, R. A. & Cunha, B. A. Aminoglycosides in gynecology. Int. Urogynecol. J. 3, 197–207 (1992).
Article Google Scholar
Amstey, M. S. Chloramphenicol therapy in pregnancy. Clin. Infect. Dis. 30, 237 (2000).
Rick, A.-M. et al. Group B Streptococci Colonization in Pregnant Guatemalan Women: Prevalence, Risk Factors, and Vaginal Microbiome. Open Forum Infect. Dis. 4, ofx020 (2017).
Article PubMed PubMed Central Google Scholar
Rosenbaum, P. R. & Rubin, D. B. The central role of the propensity score in observational studies for causal effects. Biometrika 70, 41–55 (1983).
Article MathSciNet MATH Google Scholar
Gálvez, E. J. C. et al. Distinct Polysaccharide Utilization Determines Interspecies Competition between Intestinal Prevotella spp. Cell Host Microbe 28, 838–852.e6 (2020).
Article PubMed Google Scholar
Yang, S., Liebner, S., Svenning, M. M. & Tveit, A. T. Decoupling of microbial community dynamics and functions in Arctic peat soil exposed to short term warming. Mol. Ecol. 30, 5094–5104 (2021).
Article CAS PubMed Google Scholar
Kieser, S., Zdobnov, E. M. & Trajkovski, M. Comprehensive mouse microbiota genome catalog reveals major difference to its human counterpart. PLoS Comput. Biol. 18, e1009947 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Chevalier, C. et al. Warmth Prevents Bone Loss Through the Gut Microbiota. Cell Metab. 32, 575–590.e7 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kieser, S., Brown, J., Zdobnov, E. M., Trajkovski, M. & McCue, L. A. ATLAS: a Snakemake workflow for assembly, annotation, and genomic binning of metagenome sequence data. BMC Bioinforma. 21, 257 (2020).
Article Google Scholar
Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).
Article CAS PubMed PubMed Central Google Scholar
Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012).
Article CAS PubMed PubMed Central Google Scholar
Nurk, S., Meleshko, D., Korobeynikov, A. & Pevzner, P. A. metaSPAdes: a new versatile metagenomic assembler. Genome Res. 27, 824–834 (2017).
Article CAS PubMed PubMed Central Google Scholar
Kang, D. D. et al. MetaBAT 2: an adaptive binning algorithm for robust and efficient genome reconstruction from metagenome assemblies. PeerJ 7, e7359 (2019).
Article PubMed PubMed Central Google Scholar
Olm, M. R., Brown, C. T., Brooks, B. & Banfield, J. F. dRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication. ISME J. 11, 2864–2868 (2017).
Article CAS PubMed PubMed Central Google Scholar
Hyatt, D. et al. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinforma. 11, 119 (2010).
Article Google Scholar
Huerta-Cepas, J. et al. eggNOG 5.0: a hierarchical, functionally and phylogenetically annotated orthology resource based on 5090 organisms and 2502 viruses. Nucleic Acids Res. 47, D309–D314 (2019).
Article CAS PubMed Google Scholar
Danecek, P. et al. Twelve years of SAMtools and BCFtools. Gigascience 10, giab008 (2021).
Article PubMed PubMed Central Google Scholar
Chaumeil, P.-A., Mussig, A. J., Hugenholtz, P. & Parks, D. H. GTDB-Tk: a toolkit to classify genomes with the Genome Taxonomy Database. Bioinformatics 36, 1925–1927 (2019).
Article PubMed PubMed Central Google Scholar
Darriba, D., Taboada, G. L., Doallo, R. & Posada, D. ProtTest 3: fast selection of best-fit models of protein evolution. Bioinformatics 27, 1164–1165 (2011).
Article CAS PubMed Google Scholar
Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).
Article CAS PubMed PubMed Central Google Scholar
Letunic, I. & Bork, P. Interactive Tree Of Life (iTOL) v5: an online tool for phylogenetic tree display and annotation. Nucleic Acids Res. 49, W293–W296 (2021).
Article CAS PubMed PubMed Central Google Scholar
de Nies, L. et al. PathoFact: a pipeline for the prediction of virulence factors and antimicrobial resistance genes in metagenomic data. Microbiome 9, 49 (2021).
Article PubMed PubMed Central Google Scholar
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Soc. 57, 289–300 (1995).
MathSciNet MATH Google Scholar
Liao, J., Shenhav L. & Korem T. Processed data of vaginal microbiome microdiversity from the MOMS-PI dataset. Zenodo https://doi.org/10.5281/zenodo.8150902 (2023).
Liao, J., Shenhav L. & Korem T. Code for “Microdiversity of the Vaginal Microbiome is Associated with Preterm Birth”. GitHub https://doi.org/10.5281/zenodo.8150916 (2023).

Download references

Acknowledgements

We thank members of the Korem lab for useful discussions. This study was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health under award number R01HD106017, the Program for Mathematical Genomics at Columbia University, and the CIFAR Azrieli Global Scholarship in the Humans & the Microbiome Program (T.K.). The dataset used was obtained from dbGaP (phs001523), using data provided by Gregory A. Buck, Ph.D. and colleagues and supported by NICHD (U54 HD080784) (G.A.B.).

Author information

These authors contributed equally: Jingqiu Liao, Liat Shenhav.

Authors and Affiliations

Program for Mathematical Genomics, Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA
Jingqiu Liao, Julia A. Urban & Tal Korem
Department of Civil and Environmental Engineering, Virginia Tech, Blacksburg, VA, USA
Jingqiu Liao
Center for Studies in Physics and Biology, Rockefeller University, New York, NY, USA
Liat Shenhav
Department of Microbiology and Immunology, School of Medicine, Virginia Commonwealth University, Richmond, VA, USA
Myrna Serrano, Bin Zhu & Gregory A. Buck
Center for Microbiome Engineering and Data Analysis, Virginia Commonwealth University, Richmond, VA, USA
Myrna Serrano, Bin Zhu & Gregory A. Buck
Department of Computer Science, School of Engineering, Virginia Commonwealth University, Richmond, VA, USA
Gregory A. Buck
Department of Obstetrics and Gynecology, Columbia University Irving Medical Center, New York, NY, USA
Tal Korem
CIFAR Azrieli Global Scholars program, CIFAR, Toronto, ON, Canada
Tal Korem

Authors

Jingqiu Liao
View author publications
You can also search for this author in PubMed Google Scholar
Liat Shenhav
View author publications
You can also search for this author in PubMed Google Scholar
Julia A. Urban
View author publications
You can also search for this author in PubMed Google Scholar
Myrna Serrano
View author publications
You can also search for this author in PubMed Google Scholar
Bin Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Gregory A. Buck
View author publications
You can also search for this author in PubMed Google Scholar
Tal Korem
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

J.L. and T.K. designed the study. J.L., L.S., and J.A.U analyzed the data with input from T.K., M.S., B.Z. and G.A.B. J.L. wrote the manuscript with input from L.S., T.K., M.S., B.Z. and G.A.B. G.A.B. assisted with data access and acquisition.

Corresponding authors

Correspondence to Jingqiu Liao or Tal Korem.

Ethics declarations

Competing interests

G.A.B. is a member of the Scientific Advisory Board of Juno, LTD., a startup biotech firm focused on using the vaginal microbiome to address issues of women’s gynecologic and reproductive health. Juno had no involvement in the current study. Other authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Matthew Olm, Marina Sirota, and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1-3

Reporting Summary

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Liao, J., Shenhav, L., Urban, J.A. et al. Microdiversity of the vaginal microbiome is associated with preterm birth. Nat Commun 14, 4997 (2023). https://doi.org/10.1038/s41467-023-40719-7

Download citation

Received: 01 March 2023
Accepted: 07 August 2023
Published: 17 August 2023
DOI: https://doi.org/10.1038/s41467-023-40719-7

This article is cited by

Depth-dependent microbial metagenomes sampled in the northeastern Indian Ocean
- Xiaomeng Wang
- Muhammad Zain Ul Arifeen
- Qiang Zheng
Scientific Data (2024)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.