bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1 Longitudinal profiling of the macaque vaginal microbiome reveals similarities to
2 diverse human vaginal communities: implications for use as a pre-clinical model
3 for bacterial vaginosis.
4
5 Nicholas S. Rhoades1#, Sara M. Hendrickson2#, Danielle R. Gerken1, Kassandra
6 Martinez1, Ov D. Slayden3, Mark K. Slifka2*, Ilhem Messaoudi1*
7
8 1Department of Molecular biology and Biochemistry, University of California Irvine, Irvine,
9 CA, USA
10 2 Division of Neuroscience, Oregon National Primate Research Center, Beaverton OR,
11 USA
12 3Division of Reproductive & Developmental Sciences, Oregon National Primate
13 Research Center, Beaverton, OR, USA
14 USA
15 # These authors contributed equally to this work
16
17 * To whom correspondence should be addressed
18 Ilhem Messaoudi, PhD
19 Molecular Biology and Biochemistry
20 University of California Irvine
21 2400 Biological Sciences III
22 Irvine, CA 92697
23 Phone: 949-824-3078
24 Email: [email protected]
25 Or bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
26 Mark Slifka, PhD
27 Division of Neuroscience
28 Oregon National Primate Research Center
29 Oregon Health & Science University
30 505 NW 185th Avenue
31 Beaverton, OR 97006
32 Phone: (503) 346-5483
33 Email: [email protected]
34
35 This PDF file includes:
36 Main Text
37 Table 1
38 Figures 1 to 5
39 Supplemental figures 1 to 4
40 Supplemental tables 1 to 2
41 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
42 ABSTRACT
43 The vaginal microbiota plays an important role in women’s reproductive and
44 urogenital health. Disturbances in this microbial community can lead to several adverse
45 outcomes including pelvic inflammatory disease, bacterial vaginosis (BV) as well as
46 increased susceptibility to sexually transmitted infections, miscarriage, and pre-term
47 births. It is now well accepted that while the microbiome of healthy women in the
48 developed world is dominated by Lactobacillus species, vaginal communities in
49 asymptomatic women, especially those in the developing world, can be comprised of a
50 diverse set of micro-organisms. The presence of a diverse vaginal microbiome has been
51 associated with increased susceptibility to HIV infection but their implications for
52 women’s health remain poorly understood. Rhesus macaques are an excellent
53 translational animal model due to significant physiological and genetic homology with
54 humans. In this study, we performed a longitudinal analysis of clinical and microbiome
55 data from 16 reproductive age female rhesus macaques. Many animals showed
56 hallmarks of BV, including Nugent scores above 7 and high vaginal pH. At both the
57 taxonomic and functional level, the rhesus macaque vaginal microbiome was most
58 similar to that of women who harbor a diverse vaginal community associated with
59 asymptomatic/symptomatic bacterial vaginosis. Specifically, rhesus macaque vaginal
60 microbiomes harbored a diverse set of anaerobic gram-negative bacteria, including;
61 Snethia, Prevotella, Porphyromonas, and Mobilluncus. Interestingly, some animals were
62 transiently colonized by Lactobacillus and some with Gardnerella. Our in-depth and
63 comprehensive analysis highlights the importance of the model to test interventions for
64 manipulating the vaginal microbiome.
65 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
66 IMPORTANCE
67 It is widely accepted that the “healthy” vaginal microbiome of the majority
68 of women in the developed world is dominated by Lactobacillus species.
69 However, in the developing world, a majority of women are colonized by diverse
70 microbial communities, typically associated with bacterial vaginosis, but remain
71 asymptomatic. Many questions remain about the drivers of this disparity and
72 potential interventions to alter the vaginal microbiome. Rhesus macaques
73 provide an excellent translational model due to significant physiological and
74 genetic homology with humans. In this study, we performed a longitudinal
75 analysis of clinical and microbiome data from a large cohort of reproductive age
76 rhesus macaques. At the taxonomic, genomic, and functional level, the rhesus
77 macaque vaginal microbiome was most similar to that of humans who harbor a
78 diverse vaginal community associated with asymptomatic/symptomatic bacterial
79 vaginosis. Our in-depth and comprehensive analysis highlights the utility of
80 macaques to test interventions for manipulating the vaginal microbiome.
81
82
83
84 85 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
86 INTRODUCTION
87 The vaginal microbial community modulates several critical physiological
88 functions and protect the host from infection with pathogenic organisms. The “healthy”
89 vaginal microbiome (VM) in women in the developed world is typically dominated by
90 Lactobacillus species, with L. crispatis, L. iners, L. johnsonii, and L. gasseri. These
91 microbes are considered keystone microbes that rarely co-occur in individuals (1, 2).
92 Lactobacilli inhibit the growth of other vaginal microbes via the production of lactic acid
93 through fermentation which decreases the pH of the vaginal environment (pH 3.5-5.5)
94 inhibiting the growth of other microbes (3). Lactobacillus species also competitively
95 exclude genitourinary pathogens thereby preventing infection. Disruptions of this
96 community have significant implications for women’s health (4-6). However,
97 approximately ~20% of women in the developed world and ~50% in the developing
98 world harbor a diverse VM with reduced abundance or absence of Lactobacilli and a
99 high abundance of taxa such as Gardnerella, Prevotella, Snethia, and Mobiluncus (1, 4).
100 This diverse community is often associated with a heightened inflammatory state,
101 increased susceptibility to sexually transmitted diseases notably HIV, and the
102 development of bacterial vaginosis (BV) (7).
103 BV is an inflammatory disorder and the most common vaginal infection in the US
104 affecting ~ 30% of women ages 15-44 (8). However, the prevalence of BV can vary
105 widely based on geographic location and racial background (8). BV is caused by
106 bacterial overgrowth, disruption of the commensal microbial community, and a pro-
107 inflammatory environment (9). While typically asymptomatic, BV can result in discomfort
108 and more importantly increased susceptibility to sexually transmitted diseases, preterm
109 birth, and pelvic inflammatory disease (6, 10, 11). The antibiotics Metronidazole and/or bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
110 Clindamycin are the current standard treatment for BV. However, BV recurs in ~50% of
111 women within 12 months of antibiotic treatment (12). Additionally, the production of
112 biofilms and the development of other antimicrobial resistance mechanisms will most
113 likely reduce the effectiveness of antibiotics treatments over time (13, 14). Recently, VM
114 transplants have been shown to be effective in the treatment of intractable BV (15).
115 However, microbiome transfer procedures have recently come under scrutiny due to
116 safety concerns (16, 17).
117 Another potential intervention is the use of pre-biotics to shift a microbial
118 community to a more “beneficial” state. Pre-biotics are molecules and/or nutrients that
119 are metabolically accessible to microbes with the goal of enriching for beneficial
120 microbes and/or depleting undesired microbes. An example of such an intervention is
121 the consumption or supplementation of dietary fiber to enrich the gut microbiome in
122 beneficial short-chain fatty acid-producing microbes (18, 19). The use of pre-biotics has
123 also been proposed as a cheap and easily accessible alternative to antibiotic treatment
124 for BV (20, 21). Of interest is the intravaginal application of di- or polysaccharides that
125 are preferentially metabolized and fermented by Lactobacillus to produce lactic acid and
126 hydrogen peroxide, thereby generating a low pH vaginal environment and inhibiting the
127 growth of BV-associated bacteria. Clinical trials are currently underway to test the use of
128 intravaginal Lactose (NTC03878511), Glucose (NCT03357666), and Glycogen
129 (NCT02042287) to treat BV symptoms. Sucrose in particular has shown potential for
130 improving clinical markers of BV in humans (22) and shifting the VM of rhesus
131 macaques (23). However, the clinical study only examined changes in clinical symptoms
132 and Amsel criteria at one-time point after 14 days of treatment and did not interrogate
133 changes in microbial communities (22). The preclinical study in macaques utilized
134 animals with a vaginal microbial community that had >1% Lactobacillus; did not identify bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
135 the Lactobacillus species present; nor did it report individual animal data (23). Therefore,
136 it is still unclear whether a sucrose intervention could improve clinical outcomes when
137 the frequency of Lactobacillus species is extremely low, which Lactobacillus strains can
138 respond to sucrose treatment, and how effective is this intervention.
139 To address these questions, we utilize a combination of clinical tests, 16S rRNA
140 amplicon sequencing, and shotgun metagenomics to characterize the taxonomic and
141 functional landscape of the rhesus macaque VM before and after intra-vaginal sucrose
142 treatment. Additionally, we determined the relatedness of rhesus macaque and human
143 vaginal microbes with whole-genome resolution. Previous studies have defined the
144 taxonomic landscape of pigtail (24), cynomologus (25), and rhesus (26) macaques VM.
145 Together these studies have shown that macaques harbor a diverse vaginal community
146 (24-26) that shares taxa with the diverse community state type associated with BV in
147 humans (25). These patterns include a low abundance of Lactobacillus spp. and a high
148 abundance of Snethia, Prevotella, and Mobiluncus among others. However, these
149 previous studies have been limited to amplicon sequencing techniques that are limited in
150 resolution and did not address the functional potential of the macaque vaginal
151 community or examine this similarity at a whole-genome level. The data presented
152 herein further strengthen the case for using macaques to test interventions for altering
153 the VM.
154 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
155 METHODS
156 Sample collection and cohort information
157 All macaque studies were reviewed and approved by the OHSU/ONPRC
158 Institutional Animal Care and Use Committees (IACUC). The animals were socially
159 housed indoors at the Oregon National Primate Research Center (ONPRC) following
160 standards established by the US Federal Animal Welfare Act and The National
161 Research Council Guide for the Care and Use of Laboratory Animals. All animals were
162 tested annually for simian viruses (Simian Immunodeficiency Virus, Simian Retrovirus 2,
163 Macacae herpesvirus 1, and Simian T lymphotropic virus) and received a mammalian
164 old tuberculin test semi-annually. The monkeys underwent pre-assignment evaluations
165 prior to initiation of the study. Qualifications of assignment to the study included normal
166 menstrual cycle, age 5-8 years, and a healthy weight of 4-8kgs, and all had to be void of
167 gastrointestinal issues and antibiotic use for greater than three months. At the pre-
168 assignment screening, Nugent score and microbiome samples were collected.
169 Samples were collected for 16S rRNA amplicon sequencing, metagenomics, and
170 Nugent scoring. Samples used for Nugent scoring, including pH analysis and whiff test,
171 were collected using polyester swabs (Fisher Scientific; item 23-400-122). Samples used
172 for microbiome analysis were collected with Copan Swabs (Fisher Scientific; 23-600-
173 957) and stored in 20% glycerol (shotgun metagenomics) or immediately snap-frozen
174 upon collection and stored at -80C until DNA extraction (16S rRNA amplicon
175 sequencing). To collect the vaginal and rectal swab samples, animals were sedated and
176 placed in a ventral recumbency with their pelvis slightly elevated. A sterile nasal
177 speculum was inserted vaginally to ensure a mid-vaginal sample collection without rectal
178 or vaginal introitus contamination. Preliminary screening identified 3/20 (15%) of adult
179 rhesus macaque females with vaginal pH = 4, absence of Clue cells, and negative Whiff bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
180 test. These animals were not included in the longitudinal study. The animals chosen for
181 further study had high pH (pH >5) and other characteristics of vaginal microbiome
182 dysbiosis.
183 To minimize the natural variability that may occur over the menstrual cycle,
184 animals received 21 days of combination oral contraceptives (Portia; Teva
185 Pharmaceuticals USA, Inc.) to synchronize the group. One day after the last dose of
186 oral contraceptives, samples were collected (Nugent scores, Clue cells, Whiff test,
187 microbiome, and metagenomics), and animals menstruated approximately 24-48 hours
188 later. Next, animals underwent post-menstruation sample collection (Nugent scores,
189 Clue cells, Whiff test microbiome, metagenomics), and then began five days of gel
190 treatment administration.
191 Animals were randomly designated to one of two treatment groups, Control
192 (vehicle only; n=8) or Sucrose (10% sucrose; n=8). The sucrose treatment consisted of
193 1.6% xanthan gum (Xantural 180; CP Kelco) added to a 10% sucrose solution (Sigma
194 Aldrich) to create a gel property and brought to a pH of 5.0 using lactic acid (Fisher
195 Scientific). The control treatment did not contain sucrose. Vaginal administration of the
196 gel occurred daily using a 5ml slip-tip syringe, with control animals received the xanthan
197 gum gel only.
198
199 Hormone quantification
200 Serum concentrations of estradiol (E2) and progesterone (P4) were assayed by
201 the ONPRC Endocrine Technologies Core. Hormone concentrations were determined
202 using a chemiluminescence-based automated clinical platform (Roche Diagnostics
203 Cobas e411, Indianapolis, IN). Serum E2 and P4 assay ranges were between 5-3000 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
204 pg/ml and 0.05-60 ng/ml, respectively. E2 and P4 intra-assay CVs were 8.3 and 4.2%,
205 whereas inter-assay CVs were 5.9 and 6.3%, respectively.
206
207 Clinical data generation
208 Nugent’s scoring was carried out by a trained microbiologist with 4 years of experience
209 performing Nugent scores in a CLIA-approved clinical laboratory using established
210 criteria for clinical studies (27). To complement Nugent scoring, three of four Amsel
211 criteria (28) were also assessed including vaginal pH, presence of Clue cells, and Whiff
212 test. Vaginal discharge, the fourth Amsel criteria, was not measured. The presence of
213 Clue cells (i.e., vaginal epithelial cells coated with bacteria, resembling a “sandy”
214 appearance) was determined and samples with 20% Clue cells were considered
215 positive for this test. Whiff tests were performed by adding 10% KOH to a sample of
216 vaginal fluid and the presence of fishy odor was interpreted as a positive test while its
217 absence was determined as a negative test result. A full breakdown of clinical
218 measurements can be found in supplemental table 1 (Supp. Table 1).
219
220 16S rRNA gene library construction and sequencing
221 Total DNA was extracted from vaginal swabs using the Blood and Tissue DNA
222 Isolation KIT (MO BIO Laboratories, Carlsbad, CA, USA). Rectal swabs were extracted
223 using the PowerSoil DNA Isolation KIT (MO BIO Laboratories, Carlsbad, CA, USA). This
224 DNA was used as the template to amplify the hypervariable V4 region of the 16S rRNA
225 gene using PCR primers (515F/926R with the forward primer containing a 12-bp
226 barcode) in duplicate reactions containing: 12.5 ul GoTaq master mix, 9.5 ul nuclease-
227 free H20, 1 ul template DNA, and 1 ul 10uM primer mix. Thermal cycling parameters
228 were 94C for 3 minutes; 35 cycles of 94C for 45 seconds, 50C for 1 minute, and 72C bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
229 for 1 minute and 30 seconds; followed by 72C for 10 minutes. PCR products were
230 purified using a MinElute 96 UF PCR Purification Kit (Qiagen, Valencia, CA, USA).
231 Libraries were sequenced (1 x 300 bases) using Illumina MiSeq.
232
233 16S rRNA gene sequence processing
234 Raw FASTQ 16S rRNA gene amplicon sequences were uploaded and processed
235 using the QIIME2 version 2019.10 (29) analysis pipeline as we have previously
236 described (30). Briefly, sequences were demultiplexed and quality filtered using DADA2
237 (31), which filters chimeric sequences and generated sequence variants were then
238 aligned using the MAFFT (32) and a phylogenetic tree was constructed using FastTree2
239 (33). Taxonomy was assigned to sequence variants using q2-feature-classifier against
240 the SILVA database (release 119) (34). To prevent sequencing depth bias, samples
241 were rarified to 10,000 sequences per sample before alpha and beta diversity analysis.
242 QIIME 2 was also used to generate the following alpha diversity metrics: richness (as
243 observed taxonomic units), Shannon evenness, and phylogenetic diversity. Beta
244 diversity was estimated in QIIME 2 using weighted and unweighted UniFrac distances
245 (35).
246 Analysis of the unweighted Unifrac distance (based on the presence/absence of
247 microbes) revealed that 9 vaginal samples were very similar to the fecal microbiome
248 (Supp. Figure 1A). These 9 samples had a significantly higher number of observed
249 Amplicon sequence Variants (ASVs) than the group average of all other vaginal samples
250 (Supp. Figure 1B) and a high relative abundance of taxa typically found in the fecal
251 microbiome (Supp. Figure 1C). Therefore, these nine VM samples were removed from
252 future analysis along with eight additional samples that did not meet our minimum
253 sequencing depth threshold of 10,000 reads (Supp. Figure 1D). bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
254 16S rRNA gene amplicon sequencing data obtained from vaginal swabs from
255 236 women from Umlazi, South Africa aged 18 to 23 was obtained from Gossman et al.
256 (4). These samples were imported into QIIME 2 and rarified to 13,000 reads per sample.
257 Taxonomy was assigned using the full-length SILVA database (release 119) at the 99%
258 OTU cutoff. Genus-level (L6) taxonomy tables were merged, and Bray-Curtis
259 dissimilarity matrices were generated using QIIME 2.
260
261 Shotgun metagenomics
262 Shotgun metagenomic libraries were prepared for vaginal samples obtained from
263 all animals at the initiation of oral contraceptives using the Illumina Nextera Flex library
264 prep kit, per the manufacturer's recommended protocol, and sequenced on an Illumina
265 HiSeq 4000 2 × 100. Raw demultiplexed reads were quality filtered using Trimmomatic
266 (36), and potential host reads were removed by aligning trimmed reads to the Macaca
267 mulatta genome (Mmul 8.0.1) using BowTie2 (37). After quality filtering and
268 decontamination, an average of 6.08 million reads (min 0.517, max 30 million reads) per
269 sample were used for downstream analysis. Samples with less than 1 million reads after
270 quality filtering were excluded from downstream analysis. Trimmed and decontaminated
271 reads were then annotated using the HUMAnN2 pipeline using the default setting with
272 the UniRef50 database and assigned to GO terms (38). Functional annotations were
273 normalized using copies per million (CPM) reads before statistical analysis (39-41).
274 Trimmed and decontaminated reads were assembled into contigs using meta-
275 SPAdes with default parameters (42). Assembled contigs <1kb were also binned into
276 putative metagenomically assembled genomes (MAGs) using MetaBat (43). Genome
277 completeness/contamination was tested using CheckM (44), and all bins with a bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
278 completeness > 80% and contamination < 2% were annotated using PATRIC (45). The
279 taxonomy of draft genomes was determined using PATRICs' similar genome finder.
280 Shotgun metagenomic sequencing data was obtained from Lev-Sagie et.al (15)
281 and Oliver et. al (46). Human samples were classified as " Lactobacillus dominated”
282 samples if the relative abundance of Lactobacillus >90%. Samples classified as “Diverse
283 human (BV symptomatic)” were the pre-transplant samples from Lev-Sagie et.al (15).
284 Samples classified as “Diverse human (asymptomatic)” were obtained from Oliver et. al
285 (46) with a Lactobacillus relative abundance < 90%. To avoid pseudoreplication only
286 samples from the earliest time-point were used. Sequences were annotated using the
287 HUMAnN2 pipeline as described above for rhesus vaginal samples.
288
289 Statistical analysis
290 All statistical analyses were conducted used PRISM (V8). QIIME 2 was used to
291 calculate alpha-diversity metrics; observed OTUs, Shannon evenness, and beta
292 diversity; and weighted and unweighted UniFrac distances. Bray-Curtis dissimilarity
293 matrices were constructed for species-level relative abundance. Unpaired t-tests and
294 one-way ANOVAs with post hoc correction were implemented using PRISM (V8) to
295 generate p values. The LEfSe algorithm was used to identify differentially abundant taxa
296 and pathways between groups with logarithmic linear discriminant analysis (LDA) score
297 cutoff of 2.
298 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
299 RESULTS
300 Rhesus macaques display the clinical hallmarks of BV and are colonized by a
301 diverse vaginal microbiome.
302 To determine if Rhesus macaques can serve as a model of BV, we screened 9
303 reproductive age female rhesus macaques. Of these animals, six (66%) had a Nugent
304 score >7 and the presence of Clue cells (Table 1). Additionally, 5 of these 6 animals had
305 a vaginal pH of 6.0 (Table 1). These data suggest that the majority of rhesus macaque
306 females display the clinical hallmarks of BV. Profiling of the vaginal microbial (VM)
307 communities using 16S rRNA gene amplicon sequencing showed a high relative
308 abundance of bacterial taxa associated with BV, including; Snethia, Mobiliuncus,
309 Prevotella, and Gardnerella (Table 1). Interestingly, 3 animals (33%) included in this
310 screen had a high relative abundance of Lactobacillus (43-84%). These data suggest
311 that a subset of RM can naturally harbor a Lactobacillus dominated vaginal community.
312
313 Intra-Vaginal Sucrose treatment does not alter the vaginal microbiome of rhesus
314 macaques
315 We screened an additional 20 animals for inclusion in a longitudinal study to
316 determine the efficacy of vaginal sucrose gel to improve BV clinical indicators and
317 increase the abundance of Lactobacillus in the VM. Initial screening of these 20 animals
318 indicated that 4 animals did not display the clinical indicators of BV (pH <5 and/or
319 Nugent score <4) and were excluded from the longitudinal study. The remaining 16
320 animals were split evenly into two groups. Animals in group 1 received seven daily
321 applications of a sucrose gel vaginally, while animals in group 2 received seven daily
322 applications of the gel alone. Since hormonal levels can influence VM composition (25,
323 47, 48), animals were administered an oral contraceptive for 21 days (1-month prior to bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
324 sucrose treatment) to synchronize the menstrual cycles of the animals (Figure 1A).
325 Contraceptive treatment resulted in a significant decrease in the levels of progesterone
326 and estradiol (Figure 1 B, C). Seven days after cessation of the contraceptive treatment,
327 the animals were then treated with either sucrose or placebo gel intravaginally for 7
328 days. These animals were sampled at 5 time-points across 35 days, to collect clinical
329 (Vaginal pH, Nugent scores, whiff test, and Clue cells), circulating hormone levels
330 (Progesterone and Oestrodiol), and microbiome (16S amplicon sequencing) data
331 (Figure 1A).
332 Sucrose treatment did not alter clinical measurements (pH, Nugent score, Whiff
333 test, or presence of Clue cells) or levels of progesterone or estradiol up to 28 days after
334 treatment (Supp. Figure 2 A-F). Additionally, we observed no differences in VM
335 diversity, overall community composition (Supp. Figure 3 A-C). Sucrose treatment also
336 did not increase the relative abundance of Lactobacillus (Supp. Figure 3D). Since
337 sucrose treatment did not lead to measurable changes in clinical or microbiome
338 measurements, we combined the two groups to generate a longitudinal dataset to further
339 our understanding of rhesus macaque VM community dynamics.
340 Analysis of the clinical markers of BV in all 16 animals showed association with
341 menstrual cycle. Specifically, menses were associated with reduced Nugent scores and
342 a smaller proportion of animals that meet all three measured Amsel criteria while pH
343 levels increased (Figure 1D-F). The association between menstrual cycle and these
344 clinical markers is more evident when comparing Nugent scores and vaginal pH in
345 animals with or without menses (Figure 1 G, H).
346
347 The taxonomic landscape of the rhesus macaque vaginal microbiome. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
348 Using 16S rRNA gene amplicon sequencing, we characterized the taxonomic
349 landscape of the rhesus VM across eight time-points (Figure 2A). The rhesus VM is a
350 low diversity community with an average of ~50 ASVs per sample across all time-points
351 (Figure 2B). Interestingly, the overall community composition remained stable across
352 time-points in 7 animals (Figure 2C). These “Stable” communities were dominated by
353 either Prevotella, Prophymonas, or Snethia and were associated with a higher vaginal
354 pH (Figure 2 A, C). The VM of the remaining 9 animals was more variable, transitioning
355 between states and communities dominated by less common microbes such as
356 Gardnerella, and Lactobacillus (Figure 2C). Finally, we found that individual
357 (PERMANOVA: p=0.009) rather than time point (PERMANOVA: p=0.42) was the best
358 predictor of community composition.
359 We next explored the most abundant taxa within rhesus macaques VM. A small
360 number of samples (13/120) were dominated (>50% relative abundance) by a single
361 microbe (Figure 2A), including 9 samples dominated by Snethia, 2 by Gardnerella, and
362 4 by Lactobacillus. The remaining samples contained communities composed of multiple
363 anaerobic bacteria including Snethia, Porphyromonas, Prevotella, Fastidiosipila,
364 Cantonella, Mobiluncus, and Atopobium (Figure 2A). Additional taxa found in lower
365 abundance across all samples include Parvimonas, Dialister, Fusobacterium,
366 Treponema, Peptoniphilus, and Campylobacter (Figure 2A). The relative abundance of
367 the most abundant 25 taxa did not cluster by time of sample collection or menstruation
368 status; however, samples from some animals with a “Stable” VM did cluster together
369 (Figure 2E). Interestingly Lactobacillus was detectable in 51 samples and found in a
370 relative abundance above 30% in 5 samples. These five samples came from two
371 monkeys across 4 time-points. bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
372 We next explored the relationship between clinical measures, hormone levels,
373 and the relative abundance of vaginal microbes using Spearman rank correlations
374 (Figure 2F). As expected, Nugent score and vaginal pH were positively correlated, while
375 the relative abundance of Lactobacillus was negatively correlated with both vaginal pH
376 and Nugent score (Figure 2F). Nugent score was positively correlated with systemic
377 Progesterone levels. The relative abundance of Porphyomonas, Mobiluncus mulieris,
378 and Campylobacter sputorum were also positively correlated with Nugent scores (Figure
379 2F). Levels of estradiol did not correlate with clinical markers or relative abundance of
380 specific microbes. Surprisingly Gardnerella was also negatively correlated with vaginal
381 pH (Figure 2F). However, the vaginal pH of those animals was in the 5-7 range, and
382 those animals harbored a low abundance of Lactobacillus.
383
384 The rhesus macaque vaginal microbiome is comparable to human non-
385 Lactobacillus dominated vaginal communities.
386 Next, we compared our longitudinal 16S rRNA gene amplicon data to those
387 reported in humans by Gosmann et. al (4). This study was selected for the large number
388 of vaginal samples examined and the presence of both Lactobacillus dominated (N =83)
389 and diverse communities (N =117). Rhesus macaque samples were divided into
390 “Diverse” (N=105) and “High Lactobacillus” (N=5). Due to differences in differences in
391 the methodology, a comparison could only be made qualitatively at the genus level. A
392 principal coordinates analysis shows that the “Diverse” rhesus VM was more similar to
393 “Diverse” than “Lactobacillus dominated” human vaginal communities (Figure 3A, B).
394 Six bacterial genera were shared between at least one human and one macaque group
395 with 1% relative abundance: Lactobacillus, Prevotella, Gardnerella, Snethia, Atopbium,
396 and Mobiluncus (Figure 3C). Megasphera, Veillonella, and Bacteroidales were only bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
397 found in the diverse human vaginal microbiome (Figure 3B). On the other hand,
398 Porphytomonas, Fastidiosipila, Parvimonas, Dialister, Campylobacter, and Catonella
399 were only found in the “Diverse” macaque VM (Figure 2B).
400
401 Metagenomic genome assembly reveals taxa in the rhesus macaque vaginal
402 microbiome similar to human urogenital bacteria
403 Shotgun metagenomics provides higher resolution taxonomic information and
404 functional potential of microbial communities not attainable using 16S rRNA amplicon
405 sequencing. Shotgun metagenomic libraries were prepared from vaginal samples
406 collected before and 7 days after sucrose/placebo treatment (Figure 1A). We eliminated
407 any samples with less than 1 million reads after host decontamination resulting in the
408 loss of 8 samples at the pre-sucrose and 3 samples from the post-sucrose time points
409 (Supp. Figure 4A). As noted for 16s rRNA amplicon sequencing, sucrose treatment did
410 not exert a significant impact on the functional potential of the VM (Supp. Figure 4C).
411 Although we were unable to identify the Lactobacillus colonizing the rhesus VM at the
412 species level using 16S amplicon sequencing or genome assembly, annotation of
413 shotgun metagenomic reads using MetaPhlan revealed the presence of Lactobacillus
414 johnsonii, L. amylovarus, and L. acidophilus (Supp, Figure 4B).
415 We employed metagenomic genome assembly to generate a higher resolution
416 picture of which bacterial taxa were present in the rhesus macaque VM, and how they
417 relate to the genomes of bacterial strains isolated from the human VM. We constructed a
418 total of 78 metagenomically assembled genomes (MAGs) with >80% genome
419 completeness and <2% contamination as measured by CheckM (Table 2, Supp. Figure
420 4D). The MAGs were largely representative of dominant taxa identified in our 16S rRNA
421 amplicon sequencing data, including 9 Gardnerella, 6 Mobiluncus, 8 Prevotella, 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
422 Campylobacter and 3 Snethia genomes (Table 2, Supp. Figure 4D). We next
423 determined the relationship between our MAGs and human isolates. Macaques
424 Mobiluncus and Snethia genomes were most closely related to the common human
425 bacteria Mobiluncus mulieris, and Snethia sanguinegens respectively (Figure 4 A, B).
426 Macaque Gardnerella genomes were most closely related to Gardnerella vaginalis,
427 commonly detected in non-lactobacillus dominated communities and often associated
428 with BV in humans (Figure 4C). Additionally, assembled vaginal Prevotella and
429 Campylobacter genomes were distinct from those we previously assembled from rhesus
430 macaque fecal samples (Figure 4D, E). Our assembled Prevotella genomes were most
431 closely related to Prevotella timonensis and Prevotella bucallis previously isolated from
432 human vaginal samples (Figure 4D). The assembled Campylobacter were most similar
433 to Campylobacter sputorum which is most commonly associated with livestock
434 urogenital samples (Figure 4E).
435
436 The rhesus macaque vaginal microbiome is functionally more similar to women
437 with a diverse vaginal microbiome.
438 To determine the functional potential of the rhesus VM compared to that of
439 human VM, we compared our shotgun metagenomic data to those obtained from clinical
440 studies by Oliver et. al (46) and Lev-Sagie et. al (15). These studies analyzed samples
441 from human vaginal communities that were classified as “Lactobacillus dominated”
442 (>90% Lactobacillus), “Diverse asymptomatic” (<90% Lactobacillus), and “Recurrent
443 BV”. We used supervised random forest modeling to determine if the overall functional
444 capacity of the VM could be used to distinguish between samples collected from
445 asymptomatic women (either “Lactobacillus dominated” or “Diverse Asymptomatic”),
446 women with BV, and rhesus macaques. The overall random forest model was 83% bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
447 accurate at classifying samples into the four groups with overlap between the VMs from
448 women with BV and asymptomatic women with a diverse VM (Figure 5A). The VM of
449 rhesus macaques was more closely related to that of asymptomatic women with diverse
450 communities than to VM of women with BV (Figure 5B).
451 We then extracted the functional GO terms that distinguished between the four
452 groups (Supp. Figure 4E). We found that ‘D-gluconate metabolic process’,
453 ‘phosphoglycerate mutase activity’, and ‘lipopolysaccharide biosynthetic process’ were
454 among the most significant GO terms for classifying samples into the four groups (Supp.
455 Figure 4E). Several pathways were unique to each group. For example, “palmitoyl-[acyl-
456 carrier-protein] hydrolase activity” and “heme transport” pathways were common in
457 “Lactobacillus dominated” human VM communities but absent in vaginal communities
458 from rhesus macaques and from asymptomatic women with diverse communities or
459 those with recurrent BV (Figure 5C). Other terms such as ‘D-gluconate metabolic
460 process’ and ‘UDP-glucose metabolic process’ were found in human vaginal
461 communities regardless of health status, but absent in the rhesus VM (Figure 5C). We
462 also found GO-terms shared between Lactobacillus dominated humans and diverse
463 asymptomatic humans, but absent in women with BV and rhesus macaques such as
464 “protein-N(PI)-phosphohistidine-sugar phosphotransferase activity” and
465 “phosphoenolpyruvate-dependent sugar phosphotransferase system” that were absent
466 in vaginal communities from women with BV and rhesus macaque. We also found that
467 the "oxidation-reduction process” pathway was abundant in rhesus and human diverse
468 vaginal communities but not in Lactobacillus dominated human vaginal communities.
469 Finally, we also identified GO terms that were unique to the rhesus macaque vaginal
470 microbiome including “antioxidant activity”, “arginine catabolic process”, and “regulation
471 of pentose phosphate shunt” (Figure 5C). bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
472
473 DISCUSSION
474 Approximately 50% of women globally have a diverse vaginal microbiome
475 defined by the absence or low abundance of gram-positive Lactobacillus that typically
476 dominates the human vaginal microbiome in developed countries (1, 2, 4, 11). Instead,
477 this diverse community is composed of gram-negative anaerobic bacteria such as
478 Gardnerella, Prevotella, Snethia, and Mobiluncus (1). The prevalence of this diverse
479 community can vary widely by geographical location, with a prevalence between 20 and
480 30% in the developed world (1, 46) and between 50-80% in the developing world (4, 49).
481 Prevalence of this diverse community state has also been connected to multiple host
482 factors including ethnicity (50), genetics (51), and sexual habits (52). These diverse
483 vaginal communities are largely considered to be a dysbiotic state of the vaginal
484 microbiome, which can result in an inflammatory state such as BV, and increased
485 susceptibility to sexually transmitted diseases (4, 53). However, important questions
486 about the implications of this community remain unanswered since many women are
487 asymptomatically colonized by these diverse communities. These questions are
488 challenging to ask in the clinical setting and require the availability of a robust animal
489 model.
490 In this study, we carried out a comprehensive analysis of the rhesus macaque
491 VM using a combination of 16s rRNA amplicon as well as shotgun metagenomic
492 sequencing. We also measured key clinical parameters associated with BV. Our
493 analysis showed that most animals displayed clinical and microbial markers of BV
494 notably high vaginal pH and high Nugent scores. At the 16S rRNA gene amplicon level,
495 the rhesus VM harbors a diverse set of anaerobic microbes including Snethia,
496 Prevotella, Mobiliuncus, and Fusobacterium that are often associated with BV. Although bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
497 the rhesus macaque VM shares a core set of highly abundant genera with humans, it
498 also harbors a diverse set of anaerobic genera less commonly found in human vaginal
499 samples including; Porphyromonas, Fastidiosipila, Catonella, Peptostreptococcaceae,
500 Spriochaetaceae, and Campylobacter (54, 55).
501 Our observations are in line with previous studies that have reported a high
502 prevalence and abundance of these genera in other macaque species (24-26). However,
503 our study is the first to report the presence of Gardnerella, a key pathogen in human BV,
504 and an indicator of a dysbiotic community (24-26). Specifically, Gardnerella was present
505 in 62 of 120 samples and at 10 to 66% relative abundance in 15 samples analyzed in
506 this study. Additionally, using shotgun metagenomics, we assembled 9 Gardnerella
507 genomes from 9 different animals that were closely related to human Gardnerella
508 vaginalis. Moreover, we assembled genomes from a diverse set of vaginal microbes
509 including Mobiliuncus, Sneathia, and Prevotella that were closely related to genomes
510 assembled from the human urogenital tract.
511 As others have previously reported (26), we found that Lactobacillus spp. was
512 largely absent in the rhesus macaque vaginal microbiome. In our initial screen, a
513 Lactobacillus dominant microbiome was found in 3 of the 10 animals. However, in our
514 longitudinal study, a Lactobacillus dominant microbiome (33-87% relative abundance)
515 was only transiently observed in animal RM14 at one time-point and animal RM3 at four
516 time-points. Lactobacillus was also detected at between 1-15% relative abundance in 13
517 samples across 7 animals. In humans, Lactobacillus spp. are the major producers of
518 lactic acids that results in low vaginal pH (~3.5). In accordance with those observations,
519 the presence of Lactobacillus in these animals correlated with lower vaginal pH and
520 Nugent scores at these time points. Our shotgun metagenomics analysis revealed the
521 presence of L. amylovarus and L. acidophilus. Although these species have not been bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
522 identified in samples from women in the developed world, future studies should
523 investigate the presence of these Lactobacillus species in VM from women in developing
524 countries.
525 Previous studies reported that the macaque VM is highly variable over time and
526 that some of this variability was correlated to hormonal cycling (25). We observed a
527 similar trend with multiple taxa being correlated to Progesterone levels including
528 Mobiluncus mulleris and Peptoniphilus. In contrast to previous longitudinal studies, we
529 found that the vaginal microbiome of some animals remained relatively stable while that
530 of others was variable over time. While further studies are needed to determine the
531 causes and consequences of these two unique communities’ states, one potential
532 explanation for this pattern is fecal contamination. Indeed, some samples obtained from
533 6 of the 9 animals with “variable” communities had fecal contamination while samples
534 from only 1 of 7 animals with “stable” communities exhibited fecal contamination. This
535 may suggest that “variable” VM are responding to community perturbation caused by
536 infiltration of fecal material and microbes. It is also possible that we did not sample often
537 enough to capture the true variability of the rhesus VM. Daily variability has been
538 observed in human samples and was associated with diet, exercise, and natural
539 hormonal cycling (56).
540 We also compared the functional capacity of the rhesus vaginal microbiome of
541 three distinct human vaginal microbiome communities; “Lactobacillus-dominated”,
542 “Asymptomatic diverse” and “Recurrent BV”. This analysis revealed that the rhesus
543 macaque and human VM are functionally distinct. However, the rhesus VM is
544 functionally most similar to that of asymptomatic women with a diverse vaginal
545 community. We found differences in metabolic potential between the four groups with all
546 three human VM functionally enriched in a variety of carbohydrate metabolism GO terms bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
547 that were not found in the rhesus macaque vaginal microbiome. Host derived glycogens
548 are the major energy source in the vaginal microbiome and driver of Lactobacillus
549 dominance (57). Differences in host glycogen production between humans and other
550 mammals is hypothesized to contribute to the unique Lactobacillus dominated
551 communities found in humans (7, 58). Although a small percentage of animals can
552 transiently harbor a Lactobacillus dominated VM, microbial communities in rhesus
553 macaques have significantly less vaginal glycogen than humans which may contribute to
554 their inability to sustain a Lactobacillus dominated vaginal community (59).
555 We tested an intravaginal prebiotic sucrose intervention to drive the vaginal
556 microbiome to a lactobacillus dominated state and improve clinical markers of BV.
557 Despite previously reported positive results for this intervention (22, 23), we found no
558 clinical or microbial indications of improvement. In our hands, intravaginal sucrose gel
559 did not lower vaginal pH, Nugent scores, or Amsel criteria positivity. Additionally, we did
560 not observe an increase in the relative abundance of Lactobacillus or a shift in overall
561 community composition. This intervention could have failed because the relative
562 abundance of Lactobacillus in these particular animals was below detection at the start
563 of the intervention. Moreover, shotgun metagenomics indicated that the “protein-N(PI)-
564 phosphohistidine-sugar phosphotransferase activity” pathway, a key step in the bacterial
565 import of sucrose, was largely absent in VM of humans with BV and rhesus macaques.
566 A microbial community lacking genes within this pathway is less likely to import and
567 metabolize sucrose. It is also worth noting that since the original goal of this study was to
568 determine if intravaginal sucrose treatment could improve clinical markers and increase
569 the abundance of Lactobacillus, we excluded animals with low vaginal pH and high
570 Lactobacillus abundance. In future studies, these animals should be identified and bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
571 studied longitudinally to determine if this represents a stable or transient community
572 state.
573 It has been well established that humans in the developing world have a distinct
574 microbiome from those of the developed nations (60). This has been particularly well
575 characterized in the context of the gut microbiome. For example, individuals in the
576 developing world harbor a more diverse gut microbiome colonized by microbes that have
577 all but disappeared in the western microbiome (61). This is due to a combination of key
578 environmental factors such diet, and antibiotic use among other factors (62-64). In
579 humans, there is also a clear distinction between the vaginal microbiomes of individuals
580 of the developed and developing worlds. Specifically, women in the developing world
581 have a much higher prevalence of diverse non-Lactobacillus dominated communities (4,
582 49). Understanding the drivers of these distinct microbial community states is critical to
583 developing VM targeted therapies. As we have previously shown for the gut microbiome,
584 we found that the rhesus VM is also more reflective of women in the developing world
585 (30).
586 In summary, we show that macaques harbor a diverse vaginal microbial
587 community similar to that detected in women with a non-Lactobacillus dominated
588 community or symptomatic BV at the 16S rRNA gene amplicon level. Additionally, we
589 show genomes assembled from the rhesus macaque vaginal microbiome are closely
590 related to human pathobionts associated with BV. Future studies should focus on the
591 immunological landscape of the rhesus urogenital tract to determine if this diverse
592 community state results in local inflammation as seen in humans with BV and
593 asymptomatic women with diverse communities. There are many unanswered questions
594 about why this community is maintained in some women and populations. While this
595 community type has been shown to increase the risk for some sexually transmitted bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
596 diseases such as HIV and HPV, these communities may be important for defense
597 against pathogens that are more prevalent in the developing world. Rhesus macaques
598 offer a unique opportunity to explore the importance of a diverse vaginal community that
599 is highly prevalent in women on a global scale.
600 601 602 603 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
604 FIGURE LEGENDS 605 606 607 Figure 1: Longitudinal assessment of clinical markers associated with BV. (A) 608 Study design and timeline. (B-E) Scatter plot of measured systemic (B) Progesterone, 609 (C) Oestrodiol, (D) Nugent scores, and (E) pH. Each dot represents an individual sample 610 with solid lines connecting samples from the same individual across time. The bold black 611 line represents the mean value across time. Significance was measured by non- 612 parametric 1-way repeated measures ANOVA (Friedman test) with Dunn's posthoc 613 comparisons between time points. Overall p-value denoted on each graph and bars 614 denoting significance of post hoc tests, * p < 0.05, ** p < 0.01, *** p < 0.001. (F) Stacked 615 bar graph denoting the percentage of animals that meet Amsel criteria. Colors 616 correspond to those used in panels B-E. (G, H) Scatter plot of (G) Nugent scores and 617 (H) vaginal pH comparing samples collected in animals with and without menses. 618 Significance determined by unpaired t-test, ** p < 0.01. 619 620 621 Figure 2: Longitudinal changes in the rhesus macaque vaginal microbiome. (A) 622 Heatmap of the 25 most abundant taxa across all samples ordered from left to right by 623 average abundance. Samples clustered by average linkage of Bray-Curtis distance 624 between samples. Metadata associated with each sample including animal ID, time- 625 point, Nugent score, vaginal pH, whiff test positivity, and menstruation status is provided 626 as well as a heatmap of Spearman rank correlation rho values between vaginal pH, 627 Nugent scores and systemic progesterone levels against the abundance of top 25 628 microbes. (B) Scatter plot of absolute sequence variants (ASVs) at each time point. 629 Each dot represents an individual sample with solid lines connecting samples from the 630 same individual across time. The bold black line represents the mean value across time. 631 (C, D) Principal coordinate analysis of unweighted UniFrac distance colored by 632 individuals with lines connecting samples collected from the same individual over time, 633 with density plots of key microbial taxa along the PCoA1 and PCoA2 axis. 634 635 636 Figure 3: Comparison of human and rhesus macaque vaginal microbiomes. (A) 637 Principal coordinate analysis of Bray-Curtis distance between rhesus macaque vaginal 638 microbiome, Lactobacillus dominated human vaginal microbiome, and diverse non- 639 Lactobacillus dominated human vaginal microbiome samples. (B) Bar graph of the 640 average Bray-Curtis distance between the rhesus macaque vaginal microbiome and the 641 two representative human communities. Significance determined using 1-way ANOVA p 642 < 0.001, with Bonferroni posthoc comparisons, ** p < 0.01, *** p < 0.001. (C) Table of 643 shared and exclusive bacterial genera between the three microbial community types. To 644 be included in this analysis genera had to be present in 10% of samples within a group 645 at < 0.1% relative abundance. 646 647 Figure 4: Genomes assembled from the rhesus macaque vaginal microbiome. 648 Phylogenetic tree built using 500 randomly selected conserved Cross-genus gene 649 families (PGfams) for assembled genomes from (A) Mobiluncus, (B) Snethia, (C) 650 Gardnerella, (D) Prevotella and (E) Campylobacter. Genomes in bold were assembled in 651 this study. Genome ID’s colored by host source. 652 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
653 Figure 5: Comparison of the functional potential of the rhesus and human vaginal 654 microbiome. (a) NMDS plot colored by sample source, outer circles denote if a sample 655 was misclassified at any point during the random forest model generation. The color of 656 the outer circle represents which group that sample was misclassified as. (B) Heatmap 657 of 50 most important GO terms as predicted by random forest. (C) Bar graph of the 658 average random forest proximity between the rhesus macaque vaginal microbiome and 659 the three human communities. Significance determined using 1-way ANOVA p < 0.001, 660 with Bonferroni posthoc comparisons, * p < 0.05. 661 662 Supplemental Figure 1: Removal of vaginal samples with potential fecal 663 contamination. (A) Principal coordinate analysis of unweighted UniFrac distance 664 generated from vaginal and concurrently collected fecal samples colored by sample 665 source. Vaginal samples circled in black denote samples with suspected fecal 666 contamination. (B) Scatter plot of measured absolute sequence variants (ASVs) for each 667 sample type including vaginal samples with suspected sequence contamination. 668 Significance determined using 1-way ANOVA p < 0.001, with Bonferroni posthoc 669 comparisons, *** p < 0.001. (C) Stacked bar plot highlighting three taxa that were shared 670 between fecal samples and vaginal samples with fecal contamination but absent in all 671 other vaginal samples and the sequenced microbial community standard. (D) Table of all 672 vaginal samples used for 16S rRNA gene amplicon sequencing. Boxes highlighted in red 673 were samples eliminated due to suspected fecal contamination, boxes in white were 674 eliminated from alpha and beta-diversity analysis due to low sequencing depth. Boxes in 675 green denote samples that were retained for further analysis. 676 677 Supplemental Figure 2: Impact of intravaginal sucrose treatment on clinical 678 markers of BV in rhesus macaques. (A-F) Longitudinal measurements of (A) Nugent 679 scores, (B) vaginal pH, (C) Whiff positivity, (D) clue cell positivity, (E) levels of estradiol, 680 and (G) progesterone. 681 682 Supplemental Figure 3: Impact of intravaginal sucrose treatment on the vaginal 683 microbiome in rhesus macaques. (A) Principal coordinate analysis of weighted 684 UniFrac distance colored by treatment group at baseline (open circle) and 14-days post- 685 treatment (filled circle). (B, C) Longitudinal measurement of (B) ASVs and (C) Shannon 686 evenness divided by treatment group. (D) The relative abundance of Lactobacillus 687 separated by treatment group at baseline and 4 time-points post-treatment. 688 689 Supplemental Figure 4: Shotgun metagenomic analysis of rhesus macaque 690 vaginal microbiome. (A) Scatter plot of sequences remaining after removal of host 691 reads. Red indicates samples that were removed from downstream analysis due to low 692 sequencing depth. (B) Table of genomes assembled based on assigned genus. (C) 693 NMDS of shotgun metagenomic samples collected 7-days post sucrose treatment 694 colored by treatment group, outer circles denote if a sample was misclassified at any 695 point during the random forest model generation. The color of the outer circle indicates 696 which group that sample was misclassified. (D) Heat map of top 50 GO terms ordered by 697 the overall contribution to random forest prediction accuracy, colored by importance for 698 defining each group and their contribution to overall model accuracy. 699 700 701 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
702 References 703 704 1. J. Ravel et al., Vaginal microbiome of reproductive-age women. Proc Natl 705 Acad Sci U S A 108 Suppl 1, 4680-4687 (2011). 706 2. P. Gajer et al., Temporal dynamics of the human vaginal microbiota. Sci Transl 707 Med 4, 132ra152 (2012). 708 3. D. E. O'Hanlon, T. R. Moench, R. A. Cone, In vaginal fluid, bacteria associated 709 with bacterial vaginosis can be suppressed with lactic acid but not hydrogen 710 peroxide. BMC Infect Dis 11, 200 (2011). 711 4. C. Gosmann et al., Lactobacillus-Deficient Cervicovaginal Bacterial 712 Communities Are Associated with Increased HIV Acquisition in Young South 713 African Women. Immunity 46, 29-37 (2017). 714 5. R. R. Spurbeck, C. G. Arvidson, Lactobacilli at the front line of defense against 715 vaginally acquired infections. Future Microbiol 6, 567-582 (2011). 716 6. J. M. Fettweis et al., The vaginal microbiome and preterm birth. Nat Med 25, 717 1012-1021 (2019). 718 7. E. A. Miller, D. E. Beasley, R. R. Dunn, E. A. Archie, Lactobacilli Dominance and 719 Vaginal pH: Why Is the Human Vaginal Microbiome Unique? Front Microbiol 720 7, 1936 (2016). 721 8. K. Peebles, J. Velloza, J. E. Balkus, R. S. McClelland, R. V. Barnabas, High Global 722 Burden and Costs of Bacterial Vaginosis: A Systematic Review and Meta- 723 Analysis. Sex Transm Dis 46, 304-311 (2019). 724 9. M. N. Anahtar et al., Cervicovaginal bacteria are a major modulator of host 725 inflammatory responses in the female genital tract. Immunity 42, 965-976 726 (2015). 727 10. R. B. Ness et al., A cluster analysis of bacterial vaginosis-associated microflora 728 and pelvic inflammatory disease. Am J Epidemiol 162, 585-590 (2005). 729 11. Y. A. You et al., Vaginal microbiome profiles of pregnant women in Korea 730 using a 16S metagenomics approach. Am J Reprod Immunol 82, e13124 731 (2019). 732 12. C. S. Bradshaw et al., High recurrence rates of bacterial vaginosis over the 733 course of 12 months after oral metronidazole therapy and factors associated 734 with recurrence. J Infect Dis 193, 1478-1486 (2006). 735 13. P. Nagaraja, Antibiotic resistance of Gardnerella vaginalis in recurrent 736 bacterial vaginosis. Indian J Med Microbiol 26, 155-157 (2008). 737 14. J. Castro, D. Machado, N. Cerca, Unveiling the role of Gardnerella vaginalis in 738 polymicrobial Bacterial Vaginosis biofilms: the impact of other vaginal 739 pathogens living as neighbors. ISME J 13, 1306-1317 (2019). 740 15. A. Lev-Sagie et al., Vaginal microbiome transplantation in women with 741 intractable bacterial vaginosis. Nat Med 25, 1500-1504 (2019). 742 16. Z. DeFilipp et al., Drug-Resistant E. coli Bacteremia Transmitted by Fecal 743 Microbiota Transplant. N Engl J Med 381, 2043-2050 (2019). 744 17. A. A. Weil, E. L. Hohmann, Fecal microbiota transplant: benefits and risks. 745 Open Forum Infect Dis 2, ofv005 (2015). bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
746 18. F. De Filippis et al., High-level adherence to a Mediterranean diet beneficially 747 impacts the gut microbiota and associated metabolome. Gut 65, 1812-1821 748 (2016). 749 19. F. Bishehsari et al., Dietary Fiber Treatment Corrects the Composition of Gut 750 Microbiota, Promotes SCFA Production, and Suppresses Colon 751 Carcinogenesis. Genes (Basel) 9, (2018). 752 20. V. Rousseau, J. P. Lepargneur, C. Roques, M. Remaud-Simeon, F. Paul, 753 Prebiotic effects of oligosaccharides on selected vaginal lactobacilli and 754 pathogenic microorganisms. Anaerobe 11, 145-153 (2005). 755 21. I. Coste, P. Judlin, J. P. Lepargneur, S. Bou-Antoun, Safety and efficacy of an 756 intravaginal prebiotic gel in the prevention of recurrent bacterial vaginosis: a 757 randomized double-blind study. Obstet Gynecol Int 2012, 147867 (2012). 758 22. S. Khazaeian et al., Comparing the effect of sucrose gel and metronidazole gel 759 in treatment of clinical symptoms of bacterial vaginosis: a randomized 760 controlled trial. Trials 19, 585 (2018). 761 23. K. T. Hu et al., Directed shift of vaginal microbiota induced by vaginal 762 application of sucrose gel in rhesus macaques. Int J Infect Dis 33, 32-36 763 (2015). 764 24. G. T. Spear et al., Longitudinal assessment of pigtailed macaque lower genital 765 tract microbiota by pyrosequencing reveals dissimilarity to the genital 766 microbiota of healthy humans. AIDS Res Hum Retroviruses 28, 1244-1249 767 (2012). 768 25. M. T. Nugeyre et al., Dynamics of Vaginal and Rectal Microbiota Over Several 769 Menstrual Cycles in Female Cynomolgus Macaques. Front Cell Infect Microbiol 770 9, 188 (2019). 771 26. G. T. Spear et al., Identification of rhesus macaque genital microbiota by 16S 772 pyrosequencing shows similarities to human bacterial vaginosis: 773 implications for use as an animal model for HIV vaginal infection. AIDS Res 774 Hum Retroviruses 26, 193-200 (2010). 775 27. R. P. Nugent, M. A. Krohn, S. L. Hillier, Reliability of diagnosing bacterial 776 vaginosis is improved by a standardized method of gram stain interpretation. 777 J Clin Microbiol 29, 297-301 (1991). 778 28. R. Amsel et al., Nonspecific vaginitis. Diagnostic criteria and microbial and 779 epidemiologic associations. Am J Med 74, 14-22 (1983). 780 29. E. Bolyen et al., Reproducible, interactive, scalable and extensible 781 microbiome data science using QIIME 2. Nat Biotechnol 37, 852-857 (2019). 782 30. N. Rhoades et al., Maturation of the infant rhesus macaque gut microbiome 783 and its role in the development of diarrheal disease. Genome Biol 20, 173 784 (2019). 785 31. B. J. Callahan et al., DADA2: High-resolution sample inference from Illumina 786 amplicon data. Nat Methods 13, 581-583 (2016). 787 32. K. Katoh, D. M. Standley, MAFFT multiple sequence alignment software 788 version 7: improvements in performance and usability. Mol Biol Evol 30, 772- 789 780 (2013). bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
790 33. M. N. Price, P. S. Dehal, A. P. Arkin, FastTree 2--approximately maximum- 791 likelihood trees for large alignments. PLoS One 5, e9490 (2010). 792 34. C. Quast et al., The SILVA ribosomal RNA gene database project: improved 793 data processing and web-based tools. Nucleic Acids Res 41, D590-596 (2013). 794 35. C. Lozupone, M. E. Lladser, D. Knights, J. Stombaugh, R. Knight, UniFrac: an 795 effective distance metric for microbial community comparison. ISME J 5, 169- 796 172 (2011). 797 36. A. M. Bolger, M. Lohse, B. Usadel, Trimmomatic: a flexible trimmer for 798 Illumina sequence data. Bioinformatics 30, 2114-2120 (2014). 799 37. B. Langmead, S. L. Salzberg, Fast gapped-read alignment with Bowtie 2. Nat 800 Methods 9, 357-359 (2012). 801 38. S. Abubucker et al., Metabolic reconstruction for metagenomic data and its 802 application to the human microbiome. PLoS Comput Biol 8, e1002358 (2012). 803 39. B. E. Suzek, Y. Wang, H. Huang, P. B. McGarvey, C. H. Wu, UniRef clusters: a 804 comprehensive and scalable alternative for improving sequence similarity 805 searches. 31, 926-932 (2015). 806 40. P. D. Karp, M. Riley, S. M. Paley, A. Pellegrini-Toole, The MetaCyc Database. 807 Nucleic Acids Research 30, 59-61 (2002). 808 41. N. Segata et al., Metagenomic biomarker discovery and explanation. Genome 809 Biol 12, R60 (2011). 810 42. A. Bankevich et al., SPAdes: a new genome assembly algorithm and its 811 applications to single-cell sequencing. J Comput Biol 19, 455-477 (2012). 812 43. D. D. Kang, J. Froula, R. Egan, Z. Wang, MetaBAT, an efficient tool for 813 accurately reconstructing single genomes from complex microbial 814 communities. PeerJ 3, e1165 (2015). 815 44. D. H. Parks, M. Imelfort, C. T. Skennerton, P. Hugenholtz, G. W. Tyson, 816 CheckM: assessing the quality of microbial genomes recovered from isolates, 817 single cells, and metagenomes. Genome research 25, 1043-1055 (2015). 818 45. A. R. Wattam et al., Assembly, Annotation, and Comparative Genomics in 819 PATRIC, the All Bacterial Bioinformatics Resource Center. Methods Mol Biol 820 1704, 79-101 (2018). 821 46. A. Oliver et al., Cervicovaginal Microbiome Composition Is Associated with 822 Metabolic Profiles in Healthy Pregnancy. mBio 11, (2020). 823 47. C. Zhao et al., Characterization of the Vaginal Microbiome in Women with 824 Infertility and Its Potential Correlation with Hormone Stimulation during In 825 Vitro Fertilization Surgery. mSystems 5, (2020). 826 48. J. Xu et al., Fertility factors affect the vaginal microbiome in women of 827 reproductive age. Am J Reprod Immunol 83, e13220 (2020). 828 49. R. N. Lois Bayigga, Alfred Ssekagiri, David P. Kateete, Musa Sekikubo, 829 Deborah J. Anderson, Jiawu Xu, Douglas S. Kwon, Damalie Nakanjako, Diverse 830 vaginal microbiome was associated with pro-inflammatory vaginal milieu 831 among pregnant women in Uganda. Human Microbiome Journal 18, (2020). 832 50. R. L. Goldenberg et al., Bacterial colonization of the vagina during pregnancy 833 in four ethnic groups. Vaginal Infections and Prematurity Study Group. Am J 834 Obstet Gynecol 174, 1618-1621 (1996). bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
835 51. R. B. Ness et al., Can known risk factors explain racial differences in the 836 occurrence of bacterial vaginosis? J Natl Med Assoc 95, 201-212 (2003). 837 52. M. A. Klebanoff et al., Race of male sex partners and occurrence of bacterial 838 vaginosis. Sex Transm Dis 37, 184-190 (2010). 839 53. V. L. Edwards et al., The Cervicovaginal Microbiota-Host Interaction 840 Modulates Chlamydia trachomatis Infection. mBio 10, (2019). 841 54. K. Diop, J.-C. Dufour, A. Levasseur, F. Fenollar, Exhaustive repertoire of 842 human vaginal microbiota. Human microbiome journal 2019 v.11, (2019). 843 55. Y. Chen et al., Association between the vaginal microbiome and high-risk 844 human papillomavirus infection in pregnant Chinese women. BMC Infect Dis 845 19, 677 (2019). 846 56. S. D. Song et al., Daily Vaginal Microbiota Fluctuations Associated with 847 Natural Hormonal Cycle, Contraceptives, Diet, and Exercise. mSphere 5, 848 (2020). 849 57. S. S. Witkin, I. M. Linhares, Why do lactobacilli dominate the human vaginal 850 microbiota? BJOG 124, 606-611 (2017). 851 58. P. Mirmonsef et al., Free glycogen in vaginal fluids is associated with 852 Lactobacillus colonization and low vaginal pH. PLoS One 9, e102467 (2014). 853 59. P. Mirmonsef et al., A comparison of lower genital tract glycogen and lactic 854 acid levels in women and macaques: implications for HIV and SIV 855 susceptibility. AIDS Res Hum Retroviruses 28, 76-81 (2012). 856 60. T. Yatsunenko et al., Human gut microbiome viewed across age and 857 geography. Nature 486, 222-227 (2012). 858 61. E. Angelakis et al., Treponema species enrich the gut microbiota of traditional 859 rural populations but are absent from urban individuals. New Microbes New 860 Infect 27, 14-21 (2019). 861 62. C. De Filippo et al., Diet, Environments, and Gut Microbiota. A Preliminary 862 Investigation in Children Living in Rural and Urban Burkina Faso and Italy. 863 Front Microbiol 8, 1979 (2017). 864 63. C. De Filippo et al., Impact of diet in shaping gut microbiota revealed by a 865 comparative study in children from Europe and rural Africa. Proceedings of 866 the National Academy of Sciences of the United States of America 107, 14691- 867 14696 (2010). 868 64. L. C. Bailey et al., Association of antibiotics in infancy with early childhood 869 obesity. JAMA pediatrics 168, 1063-1069 (2014). 870 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Table 1. Animal 1 2 3 4 5 6 7 8 9 10 Nuggent score 9 9 8 5 10 8 7 2 5 N.C. Clue cells >20% Yes Yes Yes No Yes Yes Yes No No N.C. Vagninal pH 7 7.5 5.5 7.5 7 7.5 5.5 6 4.5 8 16S amplicon relative abundance Leptotrichiacece (Snethia) 27.26 21.26 30.52 89.78 9.7 9.37 0.36 2.78 2.13 0.05 Lactobacillus 0.02 0 0 0.13 0.03 0.03 42.98 55.75 84.44 0.1 Prevotella 14.84 5.02 13.58 0 7.46 8.41 4.97 0.25 0.02 0.02 Mobiluncus 10.53 14.97 3.07 0 7.01 8.16 0.46 0.62 0.01 0 Gardnerella 0 0 0 0 1.17 0 0 6.01 8.29 0 Porphyromonas 6.66 10.78 11.08 0.29 2.95 5.18 0.32 0.36 0 8.7 Trichococcus 0.31 2.18 1.1 0.18 1.18 0.62 10.76 0.67 0.01 61.01 Campylobacter 3.74 3.79 7.1 0 5.74 15.94 0.07 0.27 0 0.02 Anaerococcus 0.43 0.07 1.75 0.35 0.27 0.37 15.77 0.75 0.11 15.1 Catonella 0 0 0 0 8.91 9.87 0 0.14 0 0 Fusobacterium 1.71 0.3 16.22 0.02 0 0 0.01 0.11 0.01 0 Peptoniphilus 4.14 3.81 1.62 0.24 1.56 3.91 0.27 0.58 0.01 1.7 N.C. = Not Collected
A 7-day Timepoint 1.1 1.2 2 3 4Sucrose 5 6 7 8 Oral Contraception Menses or Placebo Elapsed Time 24-days 4-days 21-days 7-days 10-days 4-days 7-days 14-days
Clinical Measurement Hormone Measurement Vaginal 16S Fecal 16S
VaginalbioRxiv shotgunpreprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint Animals with Menses(which was3/16 not certified by peer review)N.C. is the 4/16author/funder. All rights reserved.10/16 No 2/16 reuse allowed without 0/16 permission. 0/16 5/16 1/16
Animal ID RM1 RM5 RM9 RM13 RM2 RM6 RM10 RM14 RM3 RM7 RM11 RM15 RM4 RM8 RM12 RM16 B C 100 1000
10 100
1
10 0.1 Oestrodiol pg/ml Oestrodiol Progesterone pg/ml Progesterone 1-way RM ANOVA p < 0.0001 1-way RM ANOVA p = 0.0001 0.01 1 Timepoint 1.1 2 3 4 5 6 7 8 1.1 2 3 4 5 6 7 8
D E 10 9
8 8
6 7 pH 4 6 Nugent Score Nugent 2 5 1-way RM ANOVA p = 0.0002 1-way RM ANOVA p = 0.012 0 4 Timepoint 1.1 2 3 4 5 6 7 8 1.1 2 3 4 5 6 7 8
G H F 100 10 10
8
50 5
Vaginal pH Vaginal 6 Nugent Score Nugent % Animals Amsel + Amsel Animals %
0 0 4 Timepoint 1.1 2 3 4 5 6 7 8
Mense Mense Figure 1 Non-Mense Non-Mense B 125
A 100
75
50
AnimalTimepoint IDpH NugentWhiffMense Observed ASVs 25
Animal ID 0 RM1 Timepoint 1.2 2 3 4 5 6 7 8 RM2 C 0.4 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint RM3 Stable (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. RM4 0.3 RM5
RM6 0.2 RM7
RM8 0.1 RM9
RM10 15.91 Weighted2 -0.0 RM11 RM12 -0.1 RM13 RM14 -0.2
RM15 -0.5 -0.4 -0.3 -0.2 -0.1 -0.0 0.1 0.2 0.3 0.4 RM16 Weighted1 33.3 Timepoint Snethia 1 Prevotella timonesis 2 Porphyromonas 3 Shannon Evenness 4 Porphyromonas Porphyromonas bennonis Porphyromonas 5 D 0.4 Variable 6 7 0.3 8 0.2 Whiff/Menses
Yes/Positive 0.1 No/Negative
No data 15.91 Weighted2 -0.0 pH/Nugent -0.1 1 2 3 4 5 6 7 8 9 10 E Progesterone Nugent -0.2 pH Spearman Rho pH
GardnerellaLactobacillus -0.6 −0.3 0.0 0.3 0.6 Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A B
0.4 Lactobacillus dominated human Diverse human Diverse human 0.3 Diverse rhesus macaque High Lactobacillus High Lactobacillus rhesus macaque rhesus macaque Lactobacillus 0.2 dominated human
0.1 High Lactobacillus rhesus macaque -0.0
Bray2 22.2% Diverse Diverse human -0.1 rhesus macaque
-0.2 Lactobacillus dominated human
-0.3 Lactobacillus Diverse human dominated human -0.4
-0.5 0.00.20.4 0.5 0.6 0.7 0.8 0.9 1.0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 Bray-Curtis Distance Bray1 31.1%
C Lactobacillus High Lactobacillus Diverse Diverse dominated human rhesus macaque human rhesus macaque (N=83) (N=5) (N=117) (N=105) Lactobacillus 90.10 66.39 6.21 0.37 Gardnerella 6.19 15.36 35.11 2.96 Prevotella 0.69 1.17 17.66 11.01 Porphyromonas 2.38 0.18 14.97 Leptotrichiaceae (Snethia) 0.15 3.43 7.09 23.60 Megasphaera 0.84 6.42 Mobiluncus 0.01 0.19 2.34 4.11 Fastidiosipila 0.45 0.12 5.35 Dialister 0.03 0.87 0.76 4.34 Atopobium 0.28 4.29 Catonella 0.35 4.36 Parvimonas 0.01 0.66 0.21 3.27 Atopobium 2.22 3.16 Peptoniphilus 0.02 0.51 0.28 2.34 Spirochaetaceae 0.18 0.00 2.57 Veillonella 0.08 2.13 Peptostreptococcaceae 0.11 2.35 Fusobacterium 0.01 0.03 0.03 2.28 Bacteroidales 0.04 0.01 1.83 0.21 Campylobacter 0.14 1.73 Figure 3 Other 1.54 5.56 15.32 11.02 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A Mobiluncus mulieris strain FDAARGOS_303 B Sneathia RM10_T4_B1 Mobiluncus mulieris ATCC 35243 Sneathia RM15_T6_B4 Mobiluncus mulieris 28-1 Mobiluncus mulieris strain NCTC11497 Sneathia RM4_T6_B9 Mobiluncus RM8_T4_B1 Sneathia sp. Sn35 Mobiluncus RM7_T4_B5 Sneathia sanguinegens strain CCUG41628 Mobiluncus RM3_T6_B2 Mobiluncus RM4_T6_B2 Streptobacillus moniliformis DSM 12112 Mobiluncus RM14_T6_B1 Fusobacterium nucleatum ATCC 25586 Mobiluncus RM2_T6_B2 0 . 0 8 Fusobacterium periodonticum ATCC 33693 Mobiluncus curtisii subsp. curtisii strain NCTC11656 Mobiluncus curtisii ATCC 43063 Fusobacterium perfoetens ATCC 29250 0 . 5 Actinomyces radicidentis strain CCUG 36733 Arcanobacterium phocae strain DSM 10002 Prevotella RM1_T6_B6 D Prevotella RM4_T4_B7 Prevotella RM11_T6_B2 Prevotella RM2_T6_B1 C Gardnerella RM5_T6_B1 Prevotella RM7_T4_B3 Gardnerella RM12_T6_B6 Prevotella RM12_T6_B2 Prevotella timonensis CRIS 5C-B1 Gardnerella RM11_T6_B3 Prevotella buccalis ATCC 35310 Gardnerella RM2_T4_B2 Prevotella RM4_T6_B1 Gardnerella RM6_T6_B1 Prevotella RM10_T6_B1 Gardnerella RM4_T6_B6 Prevotella oralis ATCC 33269 Gardnerella RM3_T6_B9 Prevotella amnii CRIS 21A-A Gardnerella RM13_T4_B8 Prevotella bivia JCVIHMP010 88 Gardnerella RM16_T6_B1 Prevotella melaninogenica ATCC 25845 Prevotella denticola F0289 Gardnerella vaginalis strain DSM 4944 Prevotella disiens FB035-09AN Gardnerella vaginalis ATCC 14018 Prevotella OR4B2M1 Gardnerella vaginalis HMP9231 Prevotella CA6B3M1 Gardnerella vaginalis strain N160 Prevotella stercorea DSM 18206 Bi dobacterium longum s ubsp. inf ant i s R0033 Prevotella OR10B2M1 Bi dobacterium breve strain NRBB52 Prevotella copri DSM 18205 0 . 2 Bi dobacterium bi dum BGN4 Bact eroides f ragilis YCH46 Bi dobacterium animalis subsp. lactis Bi-07 Bi dobacterium tsurumiense strain JCM 13495 0 . 2 Alloscardovia omnicolens DSM 21503 E Campylobacter sputorum INTA08/209 Campylobacter sputorum RM3237 Campylobacter RM14_T6_B2 Campylobacter RM10_T6_B6 Rhesus Vaginal (This Study) Rhesus Fecal Campylobacter ureolyticus RIGS 9880 Campylobacter OR3B6M8 Human Vaginal Animal Other Campylobacter CA6B5M8 Outgroup Campylobacter CA5B6M8 Human Fecal Campylobacter hyointestinalis LMG 9260 Human Oral Campylobacter fetus 82-40 Campylobacter coli 76339 Campylobacter jejuni F38011 Campylobacter lari LMG 11760 Figure 4 0.2 Helicobacter cinaedi ATCC BAA-847 Arcobacter butzleri RM4018 bioRxiv preprint doi: https://doi.org/10.1101/2020.12.14.422805; this version posted December 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
A B −4 0 2 4 Row Z−Score amino.acid.transmembrane.transporter.activity 0.6 83% correct protein.N.PI..phosphohistidine.sugar.phosphotransferase.activity dipeptidase.activity phosphoenolpyruvate.dependent.sugar.phosphotransferase.system amino.acid.transport RNA.secondary.structure.unwinding RNA.helicase.activity negative.regulation.of.G0.to.G1.transition 0.4 transcription.regulatory.region.sequence.speci c.DNA.binding oxidoreductase.activity..acting.on.paired.donors..with.incorporation.or.reduction.of.molecular.oxygen..reduced.pteridine.as.one.donor..and.incorporation.of.one.atom.of.oxygen lipopolysaccharide.biosynthetic.process voltage.gated.chloride.channel.activity ATPase.coupled.organic.phosphonate.transmembrane.transporter.activity FtsZ.dependent.cytokinesi 0.2 D.gluconate.metabolic.process tRNA.5..leader.removal aspartate.tRNA.Asn..ligase.activity
Dimension 2 chromosome.segregation protein.serine.threonine.kinase.activity UDP.glucose.metabolic.process UTP.glucose.1.phosphate.uridylyltransferase.activity 0.0 ribosomal.large.subunit.binding palmitoyl..acyl.carrier.protein..hydrolase.activity myristoyl..acyl.carrier.protein..hydrolase.activity heme.transport Lactobacillus dominated human N.acetyldiaminopimelate.deacetylase.activity N.acetyltransferase.activity −0.2 Diverse human (BV symptomatic) sporulation.resulting.in.formation.of.a.cellular.spore ATP.metabolic.process Diverse human (Asymptomatic) proton.transmembrane.transport Rhesus macaque antioxidant.activity arginine.catabolic.process transferase.activity..transferring.alkyl.or.aryl..other.than.methyl..groups −0.6 −0.4 −0.2 0.0 0.2 0.4 phosphoglycerate.mutase.activity oxidoreductase.activity..acting.on.the.CH.CH.group.of.donors Dimension 1 oxidoreductase.activity..acting.on.the.aldehyde.or.oxo.group.of.donors..NAD.or.NADP.as.acceptor aminoacyl.tRNA.ligase.activity tRNA.aminoacylation C oxidation.reduction.process alanine.tRNA.ligase.activity organic.substance.metabolic.process aspartate.tRNA.ligase.activity Diverse human response.to.cadmium.ion (BV symptomatic) nucleobase.containing.compound.kinase.activity intein.mediated.protein.splicing ATPase.coupled.polyamine.transmembrane.transporter.activity propionyl.CoA.carboxylase.activity Lactobacillus regulation.of.pentose.phosphate.shunt dominated human myelin.sheath hydroxymethyl...formyl..and.related.transferase.activity
Diverse human (Asymptomatic)
0.0 0.2 0.4 0.6 0.8 1.0 Average distance to rhesus macaque Figure 5