TYPE Original Research

PUBLISHED 14 March 2024

DOI 10.3389/fbrio.2023.1322165

Oral origin of the placenta

OPEN ACCESS microbiome in pregnant women
EDITED BY
Rajan P. Adhikari,
Integrated BioTherapeutics, Inc., United States
with preeclampsia
REVIEWED BY
Roger M. Arce,
Shontreal M. Cooper 1*, Adam Borgida 2, Sejal Thacker 3,
University of Texas Health Science Center at Erica Hammer 2, Amirtha Hariharan 3, ChiaLing Kuo 4,
Houston, United States
Christina Megli, Nyle Blanck 3, Hanshu Yuan 5, Hunter Panier 5, Qingqi Lin 5,
University of Pittsburgh, United States
Kendra Maas 6, Winston Campbell 1 and Yanjiao Zhou 1,5
*CORRESPONDENCE
Shontreal M. Cooper
1
Division of Maternal Fetal Medicine, UConn Health, Farmington, CT, United States, 2Division of Maternal
[email protected] Fetal Medicine, Hartford Hospital, Hartford, CT, United States, 3Department of Periodontology, University
of Connecticut School of Dental Medicine, UConn Health, Farmington, CT, United States, 4Connecticut
RECEIVED 15 October 2023 Convergence Institute for Translation in Regenerative Engineering, Department of Public Health
ACCEPTED 08 December 2023 Sciences, UConn Health, Farmington, CT, United States, 5Department of Medicine, UConn Health,
PUBLISHED 14 March 2024 Farmington, CT, United States, 6Microbial Analysis Research Services, University of Connecticut,
Storrs, CT, United States
CITATION
Cooper SM, Borgida A, Thacker S, Hammer E,
Hariharan A, Kuo C, Blanck N, Yuan H,
Panier H, Lin Q, Maas K, Campbell W and Preeclampsia (PE) is a leading cause of morbidity and mortality in pregnancy
Zhou Y (2024) Oral origin of the placenta
microbiome in pregnant women with elusive etiology. Patients with PE are thought to be associated with a
with preeclampsia. higher rate of periodontal diseases (PDs) and changes of oral bacteria with
Front. Bacteriol. 2:1322165.
targeted PCR techniques. However, few studies have investigated the
doi: 10.3389/fbrio.2023.1322165
associations between oral microbiome dysbiosis and secondarily
COPYRIGHT
© 2024 Cooper, Borgida, Thacker, Hammer, disseminated microbes in the placenta simultaneously in patients with PE.
Hariharan, Kuo, Blanck, Yuan, Panier, Lin, Maas, The association between detected microbiome in placenta and systemic
Campbell and Zhou. This is an open-access
inflammation in PE is also unclear. We enrolled 54 pregnant patients with
article distributed under the terms of the
Creative Commons Attribution License (CC BY). and without PE and PD, and profiled the subgingival and placenta microbiome
The use, distribution or reproduction in other by the V4 region of 16S rRNA gene sequencing. Systemic inflammatory
forums is permitted, provided the original
author(s) and the copyright owner(s) are
markers tumor necrosis factor-alpha (TNF-a), C-reactive protein (CRP),
credited and that the original publication in lipopolysaccharide binding protein (LBP), and interleukins 6 and 8 (IL-6 and
this journal is cited, in accordance with IL-8) in blood were measured by ELISA. We found that PD significantly
accepted academic practice. No use,
distribution or reproduction is permitted increased the risk of PE after adjusting for age and smoking status (OR =
which does not comply with these terms. 2.26, 95% CI = 1.14–4.48, p = 0.024). A combined group of oral-associated
bacteria Veillonella, Fusobacterium, Haemophilus, Granulicatella,
Streptococcus, Gemella, and Neisseria in placenta had a significantly higher
prevalence in women with PE compared to women without PE (53.8% vs.
19.0%, p = 0.018), with the highest prevalence in patients with both PE and PD
(58.8%). The relative abundance of Haemophilus, Veillonella, and
Fusobacterium in subgingival samples was significantly higher in patients
with PE than those without PE. The relative abundance of Haemophilus in
subgingival samples was associated with increased risk of PE (OR = 2.11, 95%
CI = 1.11–4.52, p = 0.032). Proinflammatory cytokine analysis showed that PE
patients with PD had higher blood IL-8 levels than PE patients without PD (p =
0.026). CRP, LBP, and TNF-a showed no statistical difference in patients with
and without PE or PD. Blood IL-6 levels were significantly higher in patients

Frontiers in Bacteriology 01 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

with detectable placenta microbiome compared to those without placenta

microbiome (p = 0.028). Together, our data suggest a potential oral origin of
the placental microbiota present in patients with PE, and the microbiota
detected in placenta is associated with increased IL-6 level in the blood.

KEYWORDS

preeclampsia, placenta, microbiome, gut, pregnancy, oral, periodontal

disease, periodontitis

Introduction disease (PE−/PD−). We determined that few bacteria on the placenta

with a potential oral origin are associated with PE and heightened
systemic inflammatory responses. We also identified specific oral
Preeclampsia (PE) is a leading cause of morbidity and mortality microbiota that were associated with an increased risk of PE.
in pregnancy, complicating approximately 2%–5% of pregnancies Outcomes from this study will help strengthen our understanding
(Gestational hypertension and preeclampsia, 2020; Mustafa et al., of the pathogenesis of PE and offer a novel preventive or treatment
2012). The etiology of PE remains unclear. The placenta has been strategy for hypertensive disorders in pregnancy.
considered as a central organ in the pathogenesis of PE.
Periodontal disease (PD) has been shown to be positively
correlated with PE. The prevalence of PD among pregnant Materials and methods
women is approximately 40% with a higher prevalence of 60%–
70% among racial and ethnic minorities (Lieff et al., 2004; Eke et al., Study participants
2012; Azofeifa et al., 2016). PD-associated bacteria can potentiate an
aggressive inflammatory response, even in extra-oral sites (Krohn We conducted a prospective, cross-sectional cohort study
et al., 1993; Offenbacher et al., 2008). Hematogenous spread of the (Supplementary Figure 1). The study was conducted at two
oral bacteria to the placenta was theorized as an important source of academic centers from October 2019 to December 2020. Both
amniotic fluid and placenta microbiome. It is possible that oral hospitals are considered high-volume maternal referral centers. The
pathogens can transmit to the blood and disseminate to the placenta protocol was approved by the institutional review boards at both sites
where the blood flow is slow. Indeed, oral bacteria Porphyromonas (IRB #: 19-140H-1). Pregnant women admitted to the maternal unit
gingivalis and Actinobacillus actinomycetemcomitans have been for delivery were approached regarding participation in the study.
detected in amniotic fluid of pregnant women with periodontitis Women were eligible if they were 18–50 years old, they carried a
(Esra et al., 2013), and reported to be associated with placental singleton gestation, and gestational age was greater than 28 weeks.
infections in hypertensive women (Swati et al., 2012). A study Exclusion criteria were pregestational or gestational diabetes, pre-
showed patients with PE had higher levels of A. existing chronic hypertension, known or diagnosed HIV infection in
actinomycetemcomitans, Fusobacterium nucleatum spp., P. pregnancy, hepatitis B or C positive status, inflammatory bowel
gingivalis, Prevotella intermedia, Tannerella forsythensis, and disease, major fetal anomalies, renal disease, chronic antibiotic use
Treponema denticola in the placenta (Barak et al., 2007; in pregnancy, chronic steroid use (≥2 months use), multiple
Amarasekara et al., 2015). However, the study had a small sample gestation, and inability to collect specimens within 72 h after birth.
size and detected a limited number of PD-associated bacteria in Owing to the possibility that labor may influence risk of infection and
placenta using a targeted PCR approach. To date, there is a lack of concern for bacterial contamination of the placenta, all cesarean
non-targeted global characterization of placenta microbiome and deliveries for this study were confined to women who underwent
oral microbiome simultaneously in the same patients to track the elective cesarean delivery prior to undergoing spontaneous labor.
origin of microbiome detected in the placenta of patients with PE. PE was defined as new-onset hypertension: BP ≥140 mmHg
Furthermore, the relationship between placenta microbiome and systolic or ≥90 mmHg diastolic (sustained on two occasions at least
systemic inflammation in PE is unclear. 6 h apart) in the presence or absence of proteinuria (at least 300 mg
To answer such unresolved questions, we prospectively enrolled per 24 h, a score of 0.3 on protein/creatinine ratio during the study
patients according to their maternal diagnosis at the time of period (Gestational hypertension and preeclampsia, 2020).
admission: preeclampsia/periodontal disease (PE+/PD+), Serum blood collection: Blood was collected on admission. The
preeclampsia/no periodontal disease (PE+PD−), no preeclampsia/ samples were centrifuged and serum was stored at −80°C for
periodontal disease (PE−/PD+), and no preeclampsia/no periodontal cytokine analysis (Supplementary Figure 2).

Frontiers in Bacteriology 02 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

Placenta specimen collection: At the time of delivery, the extractions. The 16S rRNA gene sequences were processed with the
placenta was placed in a sterile container and immediately DADA2 pipeline to generate amplicon sequence variants (ASVs)
handed off to trained study personnel who wore facial masks and that represent the lowest taxonomy unit in 16S rRNA gene
used sterile gloves, scalpel, and tissue forceps (Supplementary sequencing. Data processing included paired-end reads merging,
Figure 2). Two 1-cm × 1-cm × 1-cm cuboidal sections were denoising, identification and removal of chimeric sequences, and
circumferentially excised under the surface area of the placenta assigned sequences to bacterial taxonomies. Sequences detected in
from specified areas (chorionic plate to basal plate); one area any negative control were removed from all samples. Bacterial
was located 4 cm proximal to cord insertion and one area taxonomic identification of ASVs was done using the RDP
was located 4 cm from the placental edge site (Faul et al., 2007; Bayesian classifier against the Silva nr_v119 taxonomy database
Aagaard et al., 2014). The specimens were stored at −80°C for (Wang et al., 2007; Quast et al., 2012). Sequences that were
microbiome assessment. unidentified at the domain level and those identified as
Oral cavity specimen collection. All patients were evaluated for mitochondria were also removed. We employed rigorous
the presence or absence of PD by a trained periodontist. The dental procedures to remove potential placenta microbiome
exam was a full mouth periodontal examination, with positioning of contamination from various sources as illustrated in
a periodontal probe parallel to the long axis of the tooth at each site Supplementary Figure 3.
(Chapple et al., 2018). The exam was performed on postpartum day
1 (Supplementary Figure 2). Measurements were taken at six sites
around each tooth, namely, mesiobuccal, midbuccal, distobuccal, Analysis of demographic and clinical data
mesiolingual, midlingual, and distolingual on postpartum day 1
(Eke et al., 2018). The dental provider was blinded to all sample Descriptive statistics were used for demographic and clinical
groups to decrease the risk of selection bias. The following information. We first summarized women by PD and PE status for
parameters were evaluated: periodontal probing depth (PPD), continuous variables using mean and standard deviation;
bleeding on probing (Bop), Recession—measured (when present) categorical variables were analyzed using frequencies and
from the cement–enamel junction (CEJ) to free gingival margin, percentages (Table 1). The differences between the groups were
and clinical attachment loss (CAL) (Lang et al., 1990). tested using two-sample t-tests for continuous variables after proper
PD was defined as the dental exam finding ≥2 non-adjacent transformation and Fisher’s exact tests for categorical variables. The
interproximal sites with CAL ≥3 mm, and ≥2 non-adjacent prevalence of PD between women with and without preeclampsia
interproximal sites with PPD ≥ 4 mm, or ≥1 site with PPD ≥5 was then determined by Chi-square testing. A logistic regression
mm (Eke et al., 2018). For the purpose of this study, severe model was fitted to evaluate the associations between PE and PD,
periodontitis was defined as ≥2 interproximal sites with CAL ≥6 adjusting for age and smoking variables (Table 2).
mm and ≥1 interproximal site with PPD ≥5 mm. Non-severe
periodontitis was categorized into moderate (≥2 interproximal
sites with CAL ≥4 mm or ≥2 sites with PPD ≥5 mm) and mild Statistical analysis of placenta and
≥2 interproximal sites with CAL ≥3 mm and ≥2 interproximal sites oral microbiome
PPD ≥4 mm or ≥1 site with PPD ≥5 mm. Subgingival dental plaque
samples were collected by swiping the tooth surface with a dental Descriptive analysis of the placenta microbiome includes
explorer, and were placed in DNA Genotek media and stored at prevalence and relative abundance of the placenta microbiome in
−80°C until sequencing. each group. Prevalence refers to the proportion of participants who
have a given taxon. Data in two placenta sites were combined to
represent placenta microbiome for a given patient. Kruskal–Wallis
DNA extraction, PCR amplification, and 16S test or Fisher’s exact test and Wilcox on sum rank test were used to
rRNA gene sequencing test differences of the prevalence and relative abundance of the
placenta microbiome across groups, respectively. Permutational
To determine the microbiome profile in placenta and multivariate analysis of variance (PERMANOVA) was performed
subgingival plaque, we performed V4 regions of 16S rRNA gene to evaluate the global microbiome difference between two groups
sequencing (Maas and GitHub: Bioinformatics, [[NoYear]]). In after accounting for potential confounding variables including age,
brief, a maximum of 0.25 g of placenta and 200 mL of subgingival ethnicity, smoking status, and history of PE (Kelly et al., 2015).
samples were used for DNA extraction using a Qiagen DNA mini Wilcox on sum rank testing and DESeq2 were performed to identify
kit for tissue and blood, according to the manufacturer’s protocol differential bacteria between groups (Corcoll et al., 2017). A logistic
(The Human Microbiome Project Consortium, 2012). Genomic regression model was used to determine the risk of oral microbiome
DNAs were amplified using primers targeting the V4 region of the in the development of PE. p-values from multiple comparisons were
16S rRNA gene, followed by 2× 250-bp paired-end sequencing adjusted using the false discovery approach. Adjusted p-value <0.05
using the Illumina MiSeq sequencing platform. To control potential was considered as statistically significant. All above statistical
microbiome contaminations, negative controls including extraction analyses of the microbiome were performed with R 3.3.2 (Vienna,
controls and operation room controls were included for all sample Austria) software.

Frontiers in Bacteriology 03 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

TABLE 1 Clinical characteristics of the 54 participants by subgroups.

Variable PE+/PD+ PE+/PD− PE−/PD+ PE−/PD− p-value

(n = 19) (n = 10) (n = 7) (n = 18)
Age 29.32 ± 5.39 29.2 ± 5.71 31.57 ± 7.79 27.67 ± 5.77 0.517

Ethnicity 0.712

Hispanic 7 (36.8%) 2 (20%) 3 (42.9%) 8 (44.4%)

Non-Hispanic 12 (63.1%) 8 (80%) 4 (57.1%) 10 (55.6%)

Gestational age at the time of delivery/enrollment (weeks) 35.84 ± 2.71 35.5 ± 3.5 38.14 ± 1.68 38.11 ± 1.37 0.008

Mode of Delivery 0.247

C-section 10 (52.6%) 5 (50%) 3 (42.9%) 4 (22.2%)

Vaginal 9 (47.4%) 5 (50%) 4 (57.1%) 14 (77.8%)

NICU Admission 0.151

No 12 (63.2%) 8 (80%) 7 (100%) 16 (88.9%)

Yes 7 (36.8%) 2 (20%) 0 (0%) 2 (11.1%)

Pre-pregnancy BMI 31.53 ± 4.4 30.4 ± 7.71 32.43 ± 7.79 32.28 ± 7.09 0.885

Preterm 0.073

No 10 (52.6%) 8 (80%) 6 (85.7%) 16 (88.9%)

Yes 9 (47.4%) 2 (20%) 1 (14.3%) 2 (11.1%)

Race 0.650

Asian 0 (0%) 2 (20%) 0 (0%) 0 (0%)

Black 7 (36.8%) 4 (40%) 2 (28.6%) 5 (27.8%)

Caucasian 5 (26.3%) 2 (20%) 2 (28.6%) 5 (27.8%)

Hispanic 7 (36.8%) 2 (20%) 3 (42.9%) 8 (44.4%)

Smoking 0.291

Former 5 (26.3%) 1 (10%) 2 (28.6%) 1 (5.6%)

No 13 (68.4%) 9 (90%) 5 (71.4%) 17 (94.4%)

Yes 1 (5.3%) 0 (0%) 0 (0%) 0 (0%)

Cytokine analysis Results

Cytokine serum levels were determined using commercially Study population
available kits according to the manufacturer’s instructions. The
cytokine serum levels, expressed as medium fluorescence intensity A total of 126 women were screened from October 2019 to
(MFI), were log transformed and compared among groups by using December 2020. Seventy-two patients did not meet inclusion criteria,
Kruskal–Wallis, followed by Dunn’s post-test where the p-value was leaving 54 women eligible for the study (Supplementary Figure 1).
adjusted for multiple testing using the Benjamini–Hochberg method. Patient characteristics and demographics (PE and periodontal status)
are presented in Tables 1, 2. A subgroup analysis was performed to
TABLE 2 Logistic regression evaluating periodontal disease
evaluate the prevalence of PD among patients with and without PE
and preeclampsia.
(Supplementary Table 1). All clinical variables were comparable across
Variables Odds Ratio (OR) p-Value four groups (PE+/PD+, PE+/PD−, PE−/PD+, and PE−/PD−) except
a
that birth weight (p = 0.001) and gestational age (p = 0.008) were
Positive Periodontal Disease 2.26 (1.14–4.48) 0.024
significantly lower in patients with PE. Interestingly, PD was confirmed
Delivered Preterma 1.47 (1.08–2.00) 0.019 in 19 (65.5%) women with PE compared to 7 (28%) women without
Smoking Status (Former)a 1.22 (0.84–1.77) 0.298 PE (p = 0.007). Further logistic regression analysis showed that PD
a
Based on 54 women with oral examination on enrollment.
increased the risk of PE after adjustment for age, preterm delivery, and
*Reference groups (no periodontal disease, full term delivery, non-smoker). smoking status (OR = 2.26, 95% CI = 1.14–4.48, p = 0.024).

Frontiers in Bacteriology 04 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

Track the oral origin of the previous placenta microbiome studies (Eke et al., 2012; de Goffau
placenta microbiome et al., 2019). This yielded a final set of 23 ASVs from 10 genera
including Streptococcus, Haemophilus, Veillonella, Fusobacterium,
Ninety-six out of 99 placental samples from 54 patients were Neisseria, Granulicatella, Gemella, Actinomyces, Campylobacter,
successfully sequenced, resulting in 2,241 ASVs. DNA extraction and Leptotrichia in placenta. The majority of these microbiota are
and procedure room control accounted for a large proportion of well-known PD-associated bacteria. Streptococcus was the most
sequenced reads of placenta (81.7%). After contamination removals prevalent bacteria in placenta, accounting for 27.7% of patients in
from DNA extraction controls and procedure room controls, 2,140 our cohort (13/47 = 27.7%), with the rest of nine genera each
ASVs were left (Supplementary Figure 3). accounting for 2%–8% of total patients (Figure 1A).
To address the potential microbial contamination during
delivery, difference in placenta microbiome between cesarean and
vaginal delivery were evaluated. The microbiome in placenta after The placenta microbiome differs by PD
removing extraction and procedure room controls was significantly and/or PE status
different between cesarean and vaginal delivery by PERMANOVA (p
< 0.05) (Supplementary Figure 4) suggesting that delivery mode We analyzed the placenta microbiome in patients with and
contributes to the microbiota detected in placenta. Wilcox on without PE (PE+ and PE−) and four subgroups: PE+/PD+, PE+PD
sum rank test showed that vagina-related microbiome represented −, PE−/PD+, and PE−/PD−. We did not find overall difference in
by Lactobacillus, Prevotella, Gardnerella, and Megasphaera and alpha and beta diversity of the microbiome detected in the placenta
common stool microbiota represented by Bacteroides and based on PE and PD status (Supplementary Figure 5). Among 10
Faecalibacterium were significantly higher among vaginal deliveries. genera identified in placenta, 7 genera, namely, Veillonella,
Skin-related bacteria Bacilli were significantly higher in women who Fusobacterium, Haemophilus, Granulicatella, Streptococcus,
had cesarean deliveries. In total, ASVs from 67 genera noted to be Gemella, and Neisseria, were present in higher proportion in the
different between delivery modes were removed from placenta PE+ group compared to the PE− group. However, none of the
microbiome, resulting in 1,482 ASVs in 96 placental samples. individual genus showed statistical difference in prevalence or
Next, we filtered the placental microbiota by keeping ASVs abundance among PE+ and PE− groups or the four subgroups.
identified in oral samples and with >1% of averaged abundance in This is likely due to different patients having different bacterial
oral samples (Supplementary Figure 4). The relative abundance of composition in their placenta microbiome. However, because
1% was considered a biologically reliable or relevant cutoff for different bacteria may confer similar function and the identified
microbiome research in low biomass samples and was used for placenta microbiome were all associated with PD, we combined the

A B

C D

FIGURE 1
Prevalence of the placenta microbiome in different groups. Ten bacterial genera are identified in placenta after rigorous bioinformatic processing.
Proportions of women who have a given detectable placental microbiota are illustrated by barplots (A). Prevalence of patients with and without
placental microbiota among PE+ vs. PE− women (p = 0.018, Fisher exact test) (B). Prevalence of patients with and without placenta microbiota
among four subgroups (p = 0.035, Fisher exact test) (C). The total relative abundance of placenta microbiome across four groups, with PE−PD−
being significantly lower than PE+PD+ (p = 0.024, Wilcox on test) (D).

Frontiers in Bacteriology 05 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

prevalence or abundance of the seven bacteria genera and tested the (Supplementary Figures 7E–G). Differential taxa analysis by
differences of combined microbiome among groups. Interestingly, DESeq2 showed that Haemophilus (p = 0.015), Veillonella (p =
the prevalence of combined placenta microbiome in PE+ patients 0.020), and Fusobacterium (p = 0.046) had higher relative
(14/26 = 53.84%) was significantly higher than that of PE− patients abundance in the PE+ than in the PE− groups (Figure 2A). The
(4/21 = 19.04%) (p = 0.018) (Figure 1B). The combined placenta increase of Haemophilus in PE+ was confirmed by LefSe analysis
microbiome showed the highest prevalence in the PE+/PD+ group (Supplementary Figure 7H). Subgroup analysis showed that the
(10/17 = 58.82%), and it was significantly higher than the PE−/PD− relative abundance of Haemophilus and Fusobacterium was
group (3/15 = 20%) (p = 0.035). (Figure 1C). The abundance of significantly higher in PE+PD+ than in PE−PD+ patients
combined placenta microbiome showed a significant difference in (Figure 2B), indicating that these two bacteria may be specific to
the PE−/PD− and PE+/PD+ groups (p = 0.024) (Figure 1D). PE. Importantly, logistic regression analysis showed that
Together, these data support our hypothesis that PD-related Haemophilus but not Veillonella and Fusobacterium in
bacteria are high in prevalence and abundance in PE patients. subgingivae was associated with increased risk of PE (OR = 2.11,
95% CI = 1.11–4.52, p = 0.032). This raises the possibility that the
oral microbiome may potentially serve as a marker for PE.
The subgingival microbiome differs by
PE status
Associations between systemic
Thirty-five out of 48 oral samples were successfully sequenced inflammation and placenta microbiome
and included in the analysis. After data processing, we totally
yielded 2,087,144 reads, with a median of 51,382 and an IQR of The blood levels of CRP, LBP, TNF-a, and IL-6 showed no
64,570. As expected, Streptococcus was the most abundant genus in statistical difference in patients with or without PE or PD
subgingival samples (Supplementary Figure 6). Alpha diversity (Supplementary Tables 2, 3) even after stratification for term
analysis did not find differences in richness and Shannon deliveries with Kruskal–Wallis. However, IL-8 levels were
diversity between the PE+ and PE− groups and subgroups significantly higher in the PE+/PD+ group compared to the PE
(Supplementary Figures 7A–D). PERMANOVA analysis showed +/PD− group (p = 0.26, Figure 3A). To examine the relationship
that the global subgingival microbiome profile had a marginally between proinflammatory cytokine and placenta microbiome, we
significant difference between the PE+ and PE− groups (p = 0.09) compared cytokine levels in patients who had a detected and non-
and between the PE+PD+ and PE−PD− groups (p = 0.07) detected placenta microbiome. Interestingly, we found that there

FIGURE 2
Relative abundance of three microbiota in subgingival samples that are significantly different between PE+ and PE− patients (A) and among four
subgroups (B).

Frontiers in Bacteriology 06 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

A B

FIGURE 3
Comparison of inflammatory cytokines in four subgroups and in patients with and without detectable placental microbiota. IL-8 levels in blood are
significantly higher in PE+/PD+ patients than in PE+/PD− patients (p = 0.028, Wilcox on sum rank) (A). IL-6 levels in blood are significantly higher in
patients with detectable placenta microbiota than those without (p = 0.028, Wilcox on sum rank test) (B).

were elevated levels of blood IL-6 in placentas with detected to confirm the exact origin of the bacteria. Despite identification of
microbiome than placentas without detected microbiome (p = the potential oral origin of placenta microbiome, our study does not
0.028, Figure 3B), indicating that the presence of placenta exclude the possibility of other sources of the placenta microbiome
microbiome is associated with elevated systemic inflammation. such as the vagina and the gut. In fact, we found a common gut and
vaginal microbiome in the placenta of patients who had a vaginal
delivery, as compared to those who underwent a cesarean delivery,
Discussion suggesting that the microbiome from multiple origins may contribute
to the placenta microbiome in PE.
Overall, the placenta microbiome continues to present challenges Fusobacterium was previously found to be prevalent in the
in characterization due to the low microbial biomass and the placenta of PE patients by targeted PCR amplification (Barak et al.,
potential for DNA contamination through sampling techniques, 2007; Chen et al., 2020). Our results confirmed this finding and
reagents, and mode of delivery. Although recent studies with additionally revealed other PE-associated placenta microbiome
rigorous decontamination approaches do not support the presence through 16S sequencing. The underlying reasons why these
of placental microbiota, transient bacteria in circulation in pregnancy specific taxa can disseminate to placenta merit future
may lead to the detection of a small number of oral bacteria (de investigation. It is possible that both host factors and microbiota
Goffau et al., 2019). By simultaneously profiling the oral and placenta can contribute to microbiome seeding in placenta since hormonal
microbiome in the same pregnant women, we identified oral-origin changes during pregnancy in women and the invading capacity of
bacteria in placenta from more than half of the patients with PE, FadA adhesin from F. nucleatum (Richardson et al., 2020) can
compared to less than 20% of patients without PE. The presence of increase endothelial permeability.
placenta microbiome was associated with increased systemic Although the relative abundance of Haemophilus, Veillonella,
inflammation in PE. High abundance of Haemophilus in the and Fusobacterium in the oral samples were significantly higher in
subgingivae was associated with increased risk of PE. PE+ patients than in PE− patients, they were similar between PD+
PE has long been considered as a placental disorder. Despite the and PD− patients. This suggests that these oral bacteria are
lack of cohesive evidence of healthy placenta microbiome (Thies potential markers for PE independent of PD status. Interestingly,
et al., 2019; Sterpu et al., 2021), there is a consensus that the when analyzing the oral microbiome, we found that Haemophilus
microbiome may be present in placenta in maternal or fetal but not Veillonella and Fusobacterium in subgingivae was a risk
diseases. After rigorous potential contamination removal, we found factor for PE. A prior study showed that Haemophilus was
that the placenta microbiome was associated with PE. Sequencing the important in early biofilm formation but was not necessarily
oral and placenta microbiome in parallel in our study is advantageous more prominent in caries (Schoilew et al., 2019). This suggests
compared to any previous studies that only focused on either placenta that, in addition to bacteria associated with caries, non-caries-
or the oral microbiome. In addition, among the seven placenta- associated microbiome may also be important for PE. Since oral
associated microbiota, the relative abundance of Haemophilus, samples were collected when PE was diagnosed, we could not
Veillonella, and Fusobacterium was significantly higher in the oral determine whether the increased abundance of these oral bacteria
samples of PE+ patients than PE− patients, further supporting an oral is a consequence or a driver of PE.
origin of these placenta-associated bacteria. Whole genome shotgun Previous studies have shown that proinflammatory cytokines
sequencing with higher-resolution taxonomic classification is needed produced in pregnancy may be responsible for the exacerbated

Frontiers in Bacteriology 07 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

endothelial damage seen in women who develop PE in pregnancy informed consent for participation in this study was provided by
(Delassus et al., 1994; Chaouat et al., 1995). However, our study did the participants’ legal guardians/next of kin.
not observe any statistical difference in circulating cytokines in either
of the two groups (PE and PD) even after removing the potential
confounder, preterm delivery, and evaluating only the term deliveries. Author contributions
However, subgroup analysis found that within the PE+ group, PE
+/PD+ patients had higher IL-8 levels than PE+/PD− patients. We SC: Conceptualization, Data curation, Investigation, Methodology,
showed that detectable microbiota in placenta was associated with Resources, Visualization, Writing – original draft, Writing – review &
increased circulating IL-6 levels, suggesting that the presence of editing. AB: Conceptualization, Investigation, Supervision, Validation,
placenta microbiome may increase proinflammatory response. Writing – review & editing. ST: Data curation, Investigation,
This study has several strengths. Special attention was given to Methodology, Supervision, Writing – review & editing. EH:
the protocol development for collection of placenta specimens, and a Supervision, Writing – review & editing. AH: Data curation,
rigorous bioinformatic pipeline was employed to decrease the risk of Investigation, Methodology, Writing – review & editing. CK: Data
contamination. This study is also one of the first studies to curation, Formal analysis, Methodology, Software, Writing – review &
investigate the origins of the placenta microbiome in women with editing. NB: Data curation, Investigation, Methodology, Writing –
PE and PD by comparing the bacterial taxonomic levels for multiple review & editing. HY: Data curation, Formal analysis, Investigation,
body sites. Prior to our study, this has not been investigated in detail. Methodology, Software, Writing – review & editing. HP: Data curation,
One major limitation of this study is the absence of vagina and Investigation, Methodology, Project administration, Supervision,
stool sampling collection, which restricts the analysis of the placental Writing – review & editing. QL: Data curation, Formal analysis,
microbial origins to only oral microbiota. Second, because the Investigation, Methodology, Writing – review & editing. KM: Data
placenta in the study was not subjected to removal of maternal curation, Formal analysis, Investigation, Methodology, Software,
blood, it is possible that the microbiota we identified in the placenta Validation, Writing – original draft. WC: Conceptualization,
represents the disseminating maternal microbiota. Rinsing maternal Investigation, Methodology, Supervision, Writing – original draft.
blood from placenta and inclusion of additional sample types such as YZ: Conceptualization, Data curation, Formal analysis, Investigation,
maternal blood and cord blood will provide more information on the Methodology, Project administration, Resources, Supervision,
placenta microbiota during PE. Third, prophylactic antibiotic use Validation, Writing – original draft, Writing – review & editing.
during cesarean section and group B Streptococcus infection in the
study may alter the oral or placental microbiome, which may
complicate the interpretation of the results. However, we found no Funding
statistical difference in PE+ and PE− participants in antibiotics usage.
Thus, results from group comparisons remain meaningful. In The author(s) declare that no financial support was received for
addition, our study is an association study, and we are not able to the research, authorship, and/or publication of this article.
show a causative link between the microbiome in women with PD
and/or PE. Nevertheless, our findings provided important evidence
that there exists higher prevalence and abundance of bacteria in the Conflict of interest
placenta of patients with PE compared to controls, and there is an
oral microbiota dysbiosis in PE. Future studies should include early The authors declare that the research was conducted in the
and multiple sampling time points through the whole pregnancy, absence of any commercial or financial relationships that could be
which allows one to determine whether there is a causal role of the construed as a potential conflict of interest.
oral microbiome in PE. If this is confirmed, oral microbiome, with
easy sampling in clinic, may be used as an alternative strategy to
predict or prevent PE. Publisher’s note
All claims expressed in this article are solely those of the authors
Data availability statement and do not necessarily represent those of their affiliated organizations,
or those of the publisher, the editors and the reviewers. Any product
The datasets presented in this study can be found on the NCBI that may be evaluated in this article, or claim that may be made by its
database with accession number: PRJNA1077338. manufacturer, is not guaranteed or endorsed by the publisher.

Ethics statement Supplementary material

The studies involving humans were approved by University of The Supplementary Material for this article can be found online
Connecticut Health. The studies were conducted in accordance with at: https://www.frontiersin.org/articles/10.3389/fbrio.2023.1322165/
the local legislation and institutional requirements. Written full#supplementary-material

Frontiers in Bacteriology 08 frontiersin.org

Cooper et al. 10.3389/fbrio.2023.1322165

References
Aagaard, K., Ma, J., Antony, K. M., Ganu, R., Petrosino, J., and Versalovic, J. (2014). Gestational hypertension and preeclampsia. (2020). ACOG Practice Bulletin No. 222.
The placenta harbors a unique microbiome. SciTransl. Med. 6 (237), 237ra65. American College of Obstetricians and Gynecologists. Obstet Gynecol. 135, e237–e260.
doi: 10.1126/scitranslmed.3008599 doi: 10.1097/AOG.0000000000003891
Amarasekara, R., Jayasekara, R. W., Senanayake, H., and Dissanayake, V. H. (2015). Kelly, B., Gross, R., Bittinger, K., Sherrill-Mix, S., Lewis, J. D., Collman, R. G., et al.
Microbiome of the placenta in pre-eclampsia supports the role of bacteria in the (2015). Power and sample-size estimation for microbiome studies using pair wise
multifactorial cause of pre-eclampsia. J. Obstet Gynaecol. Res. 41 (5), 662–669. distances and PERMANOVA. Bioinformatics 31 (15), 2461–2468. doi: 10.1093/
doi: 10.1111/jog.12619 bioinformatics/btv183
Azofeifa, A., Yeunf, L., and Alverson, C. (2016). Dental caries and periodontal disease Krohn, M., Rabe, L., Klebanoff, S., and Eschenbach, D. (1993). The normal vaginal
among U.S. pregnant women and nonpregnant women of reproductive age, National flora, H2O2-producing lactobacilli, and bacterial vaginosis in pregnant women. Clin.
Health and Nutrition Examination Survey, 1999–2004. J. Public Health Dent. 76 (4), Infect. Dis. 16, S273–8110.
320–329. doi: 10.1111/jphd.12159 Lang, N. P., Adler, R., Joss, A., and Nyman, S. (1990). Absence of bleeding on probing
Barak, S., Oettinger-Barak, O., Machtei, E., Sprecher, H., and Ohel, G. (2007). An indicator of periodontal stability. J. Clin. Periodontol. doi: 10.1111/j.1600-
Evidence of periopathogenic microorganisms in placentas of women with 051X.1990.tb01059.x
preeclampsia. J. Periodontol. 78, 670. doi: 10.1902/jop.2007.060362 Lieff, S., Boggess, K. A., Murtha, A. P., Jared, H., Madianos, P. N., Moss, K., et al.
Chaouat, G., Assal Meliani, A., Martal, J., Raghupathy, R., Elliott, J. F., Mosmann, T., (2004). The oral conditions and pregnancy study: periodontal status of a cohort of
et al. (1995). IL-10 prevents naturally occurring fetal loss in the CBA x DBA/2 mating pregnant women. J. Periodontol. 75, 116–126. doi: 10.1902/jop.2004.75.1.116
combination, and local defect in IL-10 production in this abortion-prone combination Maas, K.. GitHub: Bioinformatics. https://github.com/krmaas/bioinformatics/blob/
is corrected by in vivo injection of IFN-tau. J. Immunol. 154 (9), 4261–4268. master/mothur.batch.
doi: 10.4049/jimmunol.154.9.4261
Mustafa, R., Ahmed, S., Gupta, A., and Venuto, R. C. (2012). A comprehensive
Chapple, I. L. C., Mealey, B. L., Van Dyke, T. E., Bartold, P. M., Dommisch, H., review of hypertension in pregnancy. J. Pregnancy. 2012, 105918. doi: 10.1155/2012/
Eickholz, P., et al. (2018). Periodontal health and gingival diseases and conditions on an 105918
intact and a reduced periodontium: Consensus report of workgroup 1 of the 2017
World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Offenbacher, S., Barros, S. P., and Beck, J. D. (2008). Rethinking periodontal
Conditions. J. Clin. Periodontol. inflammation. J. Periodontol. 79, 1577–1584. doi: 10.1902/jop.2008.080220
Chen, X., Li, P., Liu, M., Zheng, H., He, Y., Chen, M. X., et al. (2020). Gut dysbiosis Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al. (2012). The
induces the development of pre-eclampsia through bacterial translocation. Gut 69 (3), SILVA ribosomal RNA gene database project: improved data processing and web-based
513–522. doi: 10.1136/gutjnl-2019-319101 tools. Nucleic Acids Res. 41 (D1), D590–D596. doi: 10.1093/nar/gks1219
Corcoll, N., Osterlund, T., Sinclair, L., Eiler, A., Kristiansson, E., Backhaus, T., et al. Richardson, M., Ren, J., Rubinstein, M. R., Taylor, J. A., Friedman, R. A., and Han, B.
(2017). Comparison of four DNA extraction methods for comprehensive assessment of S. Y. (2020). Analysis of 16S rRNA genes reveals reduced Fusobacterial community
16S rRNA bacterial diversity in marine biofilms using high-throughput sequencying. diversity when translocating from saliva to GI sites. Gut Microbes 12, 1. doi: 10.1080/
FEMS Microbiol. Lett. 364 (14), fnx139. doi: 10.1093/femsle/fnx139 19490976.2020.1814120

de Goffau, M. C., Lager, S., Sovio, U., Gaccioli, F., Cook, E., Peacock, S. J., et al. (2019). Schoilew, K., Ueffing, H., Dalpke, A., Wolff, B., Frese, C., Wolff, D., et al. (2019).
Human placenta has no microbiome but can contain potential pathogens. Nature 572, Bacterial biofilm composition in healthy subjects with and without caries experience. J.
329–334. doi: 10.1038/s41586-019-1451-5 Oral. Microbiol. 11, 1. doi: 10.1080/20002297.2019.1633194
Sterpu, I., Fransson, E., Hugerth, L. W., Du, J., Pereira, M., Cheng, L., et al. (2021). No
Delassus, S., Coutinho, G. C., Saucier, C., Darche, S., and Kourilsky, P. (1994).
evidence for a placenta microbiome in human pregnancies at term. Am. J. Obstet
Differential cytokine expression in maternal blood and placenta during murine
Gynecol. 224 (3), 296.e1–296.e23.
gestation. J. Immunol. 152 (5), 2411–2420. doi: 10.4049/jimmunol.152.5.2411
Swati, P., Thomas, B., Vahab, S. A., Kapaettu, S., and Kushtagi, P. (2012).
Eke, P. I., Dye, B. A., Wei, L., Thornton-Evans, G. O., Genco, R. J.CDC Periodontal Simultaneous detection of periodontal pathogens in subgingival plaque and placenta
Disease Surveillance workgroup, et al. (2012). Prevalence of periodontitis in adults in of women with hypertension in pregnancy. Arch. Gynecol. Obstet 285 (3), 613–619. doi:
the United States: 2009–2010. J. Dent. Res. 91 (10), 914–920. doi: 10.1177/ 10.1007/s00404-011-2012-9
0022034512457373
The Human Microbiome Project Consortium (2012). Structure, function and diversity
Eke, P. I., Thornton-Evans, G. O., Wei, L., Borgnakke, W. S., Dye, B. A., and Genco, of the healthy human microbiome. Nature 486, 207–214. doi: 10.1038/nature11234
R. J. (2018). Periodontitis in US adults: national health and nutrition examination
survey 2009-2014. J. Am. Dental Assoc. Thies, K. R., Romero, R., Winters, A. D., Greenberg, J. M., Gomez-Lopez, N.,
Alhousseini, A., et al. (2019). Does the human placenta delivered at term have a
Esra, E., Kenan, E., Ozrur, D., Deniz, G., Ozgur, O., Belgin, A., et al. (2013). microbiota? Results of cultivation, quantitative real-time PCR, 16S rRNA gene
Evaluation of periodontal pathogens in amniotic fluid and the role of periodontal sequencing, and metagenomics. Am. J. Obstet Gynecol. 220, 267.e1–267.e39. doi:
disease in preterm birth and low birth weight. Acta Odontol. Scand. 71 (3–4), 553–559. 10.1016/j.ajog.2018.10.018
Faul, F., Erdfelder, E., Lang, A.-G., and Buchner, A. (2007). G*Power 3: a flexible Wang, Q., Garrity, G. M., Tiedje, J. M., and Cole, J. R. (2007). Naive Bayesian
statistical power analysis program for the social, behavioral, and biomedical sciences. Classifier for Rapid Assignment of rRNA Sequences into the New Bacterial Taxonomy.
Behav. Res. Methods 39, 175–191. doi: 10.3758/BF03193146 Appl. Environ. Microbiol. 73 (16), 5261–5267. doi: 10.1128/AEM.00062-07

Frontiers in Bacteriology 09 frontiersin.org