Lack of evidence for microbiota in the placental and fetal tissues of rhesus macaques

The prevailing paradigm in obstetrics has been the sterile womb hypothesis. However, some are asserting that the placenta, intra-amniotic environment, and fetus harbor microbial communities. The objective of this study was to determine if the fetal and placental tissues of rhesus macaques harbor viable bacterial communities. Fetal, placental, and uterine wall samples were obtained from cesarean deliveries without labor (∼130/166 days gestation). The presence of viable bacteria in the fetal intestine and placenta was investigated through culture. The bacterial burden and profile of the placenta, umbilical cord, and fetal brain, heart, liver, and colon were determined through quantitative real-time PCR and DNA sequencing. These data were compared with those of the uterine wall, as well as to negative and positive technical controls. Bacterial cultures of fetal and placental tissues yielded only a single colony of Cutibacterium acnes. This bacterium was detected at a low relative abundance (0.02%) in the 16S rRNA gene profile of the villous tree sample from which it was cultured, yet it was also identified in 12/29 background technical controls. The bacterial burden and profile of fetal and placental tissues did not exceed or differ from those of background technical controls. In contrast, the bacterial burden and profiles of positive controls exceeded and differed from those of background controls. Among the macaque samples, distinct microbial signals were limited to the uterine wall. Therefore, using multiple modes of microbiologic inquiry, there was not consistent evidence of viable bacterial communities in the fetal and placental tissues of rhesus macaques. IMPORTANCE Microbial invasion of the amniotic cavity (i.e. intra-amniotic infection) has been causally linked to pregnancy complications, especially preterm birth. Therefore, if the placenta and the fetus are typically populated by low biomass yet viable microbial communities, current understanding of the role of microbes in reproduction and pregnancy outcomes will need to be fundamentally reconsidered. Could these communities be of benefit by competitively excluding potential pathogens or priming the fetal immune system for the microbial bombardment it will experience upon delivery? If so, what properties (e.g. microbial load, community membership) of these microbial communities preclude versus promote intra-amniotic infection? Given the ramifications of the in utero colonization hypothesis, critical evaluation is required. In this study, using multiple modes of microbiologic inquiry (i.e. culture, qPCR, DNA sequencing) and controlling for potential background DNA contamination, we did not find consistent evidence for microbial communities in the placenta and fetal tissues of rhesus macaques.


INTRODUCTION
The development and widespread use of DNA sequencing technologies to characterize 88 host-associated microbial communities has increasingly led researchers to question the sterility 89 of body sites and fluids previously presumed to be free of resident microorganisms. For example, 90 researchers have recently proposed the existence of microbiota in the human blood (1-8), bladder 91 (9-16), uterus (17-30), placenta (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45), and fetus (36,(44)(45)(46). This has led to discussion in the 92 literature on the caveats associated with studies of the microbiota of very low microbial biomass, 93 or potentially sterile, body sites (47-54). In particular, there has been much debate over the 94 existence of a placental microbiota (31-45, 50, 55-69) and of in utero microbial colonization of 95 the human fetus (36, 44-46, 64, 70-72). 96 The primary focus of the debate is that most of the studies proposing the existence of 97 placental and fetal microbiota in humans have relied heavily, if not exclusively, on DNA 98 sequencing techniques (31-35, 37-42, 45), and the bacterial signals in these studies may be 99 background DNA contaminants from extraction kits, PCR and sequencing reagents, and general 100 laboratory environments (50,55,57,59,62). Furthermore, even if the bacterial DNA sequence 101 data are derived from placental and fetal tissues and not from background contamination, this 102 does not necessarily indicate that there are viable bacterial communities in the placenta or the 103 fetus. Specifically, the bacterial DNA sequence data may reflect bacterial products and 104 components rather than resident microbiota (73-77). 105 As a consequence, we and others (50, 62) have suggested criteria for establishing the 106 existence of placental and fetal microbiota. First, viability of the resident bacteria should be 107 established through culture or metatranscriptomic data from bacterial-specific genes within 108 placental and fetal tissues. Second, the bacterial load of placental and fetal tissues, as 109 demonstrated through quantitative real-time PCR (qPCR), should exceed those of background 110 technical controls. Third, the bacterial profiles of placental and fetal tissues should be distinct 111 from those of the technical controls. Fourth, the resident bacteria should be visualized in the 112 tissues through microscopy. Fifth, the taxonomic data of the detected bacteria should be 113 ecologically plausible (50,62). There have been many studies that may have met one or two of 114 these criteria (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(78)(79)(80), but no study has yet attempted to simultaneously meet all criteria 115 and ultimately conclude that there is widespread colonization of the placenta and/or fetus by 116 viable microbial communities (72).

117
Although most of the research evaluating the existence of placental and fetal microbiota 118 has been done with human subjects, animal models afford opportunities to surgically obtain 119 placental and fetal tissues before the process of labor. Tissues collected after the process of labor 120 could confound experimental results regarding in utero colonization due to potential microbial 121 invasion of the amniotic cavity (81)(82)(83). Several studies using rat and mouse models have 122 provided mixed evidence: while three studies detected placental and fetal microbiota through 123 DNA sequencing techniques following cesarean delivery (44,46,75), two other studies did not 124 (60, 84). In non-human primates, specifically rhesus and Japanese macaques, a unique placental 125 and/or fetal microbiota has been consistently detected through DNA sequencing following 126 cesarean delivery (85-89). However, these preliminary studies neither include culture or qPCR 127 components nor display the sequence data from background technical controls.

128
The objective of the current study was therefore to determine whether the fetal and 129 placental tissues of rhesus macaques harbor bacterial communities using bacterial culture, qPCR, 130 and 16S rRNA gene sequencing and by comparing the bacterial profiles of these tissues to those 131 of background technical controls.

133
Bacterial culture from fetal and placental samples 134 All negative culture controls were negative (no bacterial growth over seven days) and all 135 positive culture controls were positive (lawn of bacterial growth within 24 hours). The 96 total 136 cultures of fetal and placental samples from the four rhesus macaques yielded only a single 137 bacterial colony (Figure 1)  The bacterial burden of fetal and placental tissues did not exceed that of background 171 technical controls (Figure 2A,B). Among the swab samples, only the maternal myometrium had 172 a higher bacterial load than sterile swabs (Mann-Whitney U test: U = 0, p = 0.005; Figure 2A).

173
No fetal, placental, or uterine wall tissue samples consistently had higher bacterial loads than 174 blank DNA extraction kits ( Figure 2B). profiles of human urine also differed from those of sterile swab controls (F = 1.834, p = 0.0058).

195
The bacterial taxonomic data associated with the ASV profiles of the fetal, placental, and 196 uterine wall samples and controls are illustrated in Figure 4. There were only two prominent (≥ 197 5% relative abundance) ASVs among the fetal and placental swab and tissue samples: ASVs 001 198 (Staphylococcus) and 002 (Pelomonas). These two ASVs were also prominent in the profiles of 199 both the sterile swabs and the blank DNA extraction kits. ASV 001 was identified in the profiles 200 of 9/12 (75%) and 5/10 (50%) swab and extraction kit technical controls, respectively, while 201 ASV 002 was identified in 5/12 (42%) sterile swab and 6/10 (60%) extraction kit profiles. ASV 202 001, but not ASV 002, was identified as a contaminant among swab samples by the decontam 203 program (Figure 4). Among tissue samples, neither ASV 001 or ASV 002 were identified as 204 contaminants using decontam.

205
Aside from ASV 002, ASVs 003 (Acinetobacter), 007 (Ottowia), and 012 (uncl. 206 Obscuribacterales) were prominent (≥ 5% relative abundance) among both uterine wall swab and 207 tissue samples (Figure 4). None of these three ASVs were prominent among either sterile swab 208 or extraction kit technical controls, but ASV 012 (uncl. Obscuribacterales) was identified as a profile of the villous tree sample was very low (0.02%), but its relative abundance in the swab of 280 the maternal decidua sample from this subject was 29.4%. Given that this bacterium was cultured 281 from the villous tree, was identified in molecular surveys of the villous tree sample and other 282 placental and fetal samples from this subject, and was further detected at high relative 283 abundances in a maternal decidua sample for this subject, it is reasonable to consider whether 284 this isolate represents a viable bacterium that was transmitted from the mother to the fetus 285 through the placenta. Cutibacterium (Propionibacterium) acnes has also been cultured from the is also reasonable to consider whether this isolate and molecular signals of 296 Cutibacterium/Propionibacterium may simply represent microbial contamination from study 297 personnel.

298
In the current study, qPCR revealed that the quantities of 16S rRNA gene copies in the 299 placenta (i.e. basal plate, villous tree, and the subchorion, amnion-chorion interface, and amnion 300 of the chorionic plate), umbilical cord, and fetal organs (i.e. brain, heart, liver, colon) of rhesus

305
In this study, there were no fetal or placental sites whose 16S rRNA gene profiles  The only rhesus macaque samples with bacterial profiles distinct from those of 326 background technical controls were the myometrial swabs and the uterine wall biopsies. These 327 sample types also had the highest bacterial load, as assessed through qPCR. There were four 328 ASVs that were prominent (i.e. ≥ 5% average relative abundance) among all uterine wall 329 samplesthey were classified as Acinetobacter, Ottowia, Pelomonas, and a member of the order 330 Obscuribacterales. As discussed above, the ASV classified as Pelomonas is likely a DNA 331 contaminant. Also, the program decontam identified the ASV classified as Obscuribacterales as 332 another likely DNA contaminant. The data from Acinetobacter and Ottowia are more 333 compelling. The primary ASV classified as Acinetobacter was detected in 9/10 (90%) uterine 334 wall samples at an average relative abundance of 22.9%. In contrast, it was detected in only 1/22 335 (4.5%) background technical controls. Acinetobacter has been reported in prior sequence-based 336 investigations of the human endometrium (20,23,24,29,30), and it has been cultured from the 337 human endometrium as well (95). The primary ASV classified as Ottawia was detected in 6/10 338 (60%) uterine wall samples at an average relative abundance of 10.5%. It was not detected in any 339 background technical controls. Ottowia is a genus of bacteria that has been isolated from   Limitations of this study 361 First, given that the study was conducted on a non-human primate, the sample size was    The placenta and umbilical cord were placed in an autoclave-sterilized container and 391 covered. Rhesus macaque fetuses were euthanized with pentobarbital (100 mg/kg) prior to 392 necropsy. The fetal liver, heart, and brain were snap frozen in sterile 50 ml conical tubes. The

511
Raw sequence reads were processed using DADA2 (v 1.12) (105). An analysis of 16S 512 rRNA gene amplicon sequence variants (ASVs), defined by 100% sequence similarity, was 513 performed using DADA2 in R (v 3.5.1) (https://www.R-project.org), and the online MiSeq 514 protocol (https://benjjneb.github.io/dada2/tutorial.html) with minor modifications. These 515 modifications included allowing truncation lengths of 250 bp and 150 bp and a maximum 516 number of expected errors of 2 bp and 7 bp for forward and reverse reads, respectively. To allow 517 for increased power to detect rare variants, sample inference allowed for pooling of samples. 518 Additionally, samples in the resulting sequence table were pooled prior to removal of chimeric 519 sequences. Sequences were then classified using the "silva_nr_v132_train_set" database with a 520 minimum bootstrap value of 80%, and sequences that were derived from Archaea, Chloroplast, 521 or Eukaryota were removed. of swab and tissue samples after subsampling were 98.9 ± 0.8 SD and 99.1 ± 0.5 SD, 532 respectively. Heat maps of the 16S rRNA gene profiles of samples were generated using 533 heatmap.2 in the gplots library for R (version 3.5.1). The R package decontam (106) was utilized 534 to identify ASVs that were potential background DNA contaminants using the 535 "IsNotContaminant" method with a prevalence threshold of P = 0.5. The decontam analyses 536 were run separately for the swab and tissue samples.

537
The alpha diversity of sample ASV profiles was characterized using the Chao1 index to 538 address profile richness and the Shannon and Simpson (1 -D) indices to address profile 539 heterogeneity. Differences in alpha diversity between rhesus macaque and background technical 540 control samples were evaluated using Mann-Whitney U and t-tests with sequential Bonferroni 541 corrections applied.

542
The beta diversity of ASV profiles among fetal, placental and uterine wall samples and 543 background technical controls was characterized using the Bray-Curtis similarity index. Bray-544 Curtis similarities in sample ASV profiles were visualized using Principal Coordinates Analysis 545 (PCoA) plots and statistically evaluated using non-parametric multivariate ANOVA 546 (NPMANOVA). PCoA plots were generated using the vegan package (version 2.5.5) in R. All 547 statistical analyses were completed using PAST software (v 3.25) (107).