Sex-specific differences in the salivary microbiome of caries-active children

ABSTRACT Background and Objectives: Dental caries is a chronic disease affecting young children and has multi-factorial risk factors. The purpose of this work was to identify sex-specific differences in the salivary microbiota within caries-active children. Design: Saliva specimens were collected from 85 children (boys: 41; girls: 44) between the ages of 2-12 years. Salivary microbial DNA was subjected to PCR amplification using V3-V4 16S rDNA-specific primers and next-generation sequencing. Results: Significant sex differences in salivary microbiota were found between caries-active boys versus caries-active girls. Neisseria flavescens, Rothia aeria, and Haemophilus pittmaniae were found at significantly higher levels in caries-active boys. In contrast, Lactococcus lactis, Selenomonas species HOT 126, Actinobaculum species HOT 183, Veillonella parvula, and Alloprevotella species HOT 473 were found at significantly higher levels in caries-active girls. Conclusion: We have found the acid-generating, cariogenic Lactococcus lactis to be much more abundant in caries-active girls than caries-active boys, indicating that this microorganism may play a more significant role in shaping the cariogenic microbiome in girls. In addition, in caries-active girls, Alloprevotella species HOT 473 was the only species that exhibited both significant sex differences (4.4-fold difference; p=0.0003) as well as high abundance in numbers (1.85% of the total microbial population).

salivary flow rates in women, in addition to lower pH and IgA levels, potentially leading to higher caries incidence [12,14,15]. Genetic variations in factors influencing taste, enamel proteins, and enamel structure are also linked to this trend [12,13,16]. Regarding patterns of tooth eruption, female teeth generally erupt at an earlier age, exposing teeth to a potential caries-inducing environment for longer durations of time [12]. Caries rates in women increase progressively throughout lifetimes, particularly during the reproductive years. Several physiological changes occur during pregnancy including hormonal fluctuations, immune suppression, dietary changes, and salivary flow [12,13]. Social factors that potentially help determine sex differences in caries experience include lower literacy rates in women globally, responsibility for food preparation, experience of domestic violence, and higher rates of eating disorders [12].
In this paper, we identified sex-specific differences in the salivary microbiome in caries-active children and describe the cariogenic salivary microbiome to gain further understanding of caries progression in females that may account for the observed global sex dissimilarity in caries experience.

Participant selection and demographics
We selected study participants primarily from patients visiting the OHSU Pediatric Dental Clinic at Doernbecher Children's Hospital, or at the OHSU School of Dentistry at the Robertson Life Sciences Building on the South Waterfront Campus. Some cariesfree patients were obtained from an external practitioner's dental clinic operated in Albany, Oregon. The study protocol was reviewed and approved by the OHSU Institutional Review Board (IRB; protocol #6535), and permissions and consent to participate were obtained from the parents or guardians of all children selected for this study. For older children, age 7 years or more, child assent and permissions were also obtained. This study is part of a larger study that examines the efficacy and use of adjunctive clinical therapies in caries-active children, and involves the use of specimens obtained before the application of any adjunctive therapies or other procedures that were applied as part of routine dental care. The inclusion parameters for study participation were children ages 2-12 years, in good general health (ASA I or II). The ASA physical status classification system is used by the American Association of Anesthesiologists to assess the fitness of patients before surgery; ASA I and ASA II designate a healthy individual, or an individual with mild systemic disease, respectively. Children excluded from the study were those who had undergone antibiotic treatment within the previous three month-period or were undergoing current orthodontic therapies, as well as children with recent exposure to topical fluoride or antiseptic mouth rinses. Eighty-five children were enrolled in this study (41 boys and 44 girls). Of the 85 children, 64 children were caries-active and exhibited DMFT scores of 1-15 at the time of appointment (DMFT average: 8.6) and 21 children were caries-free individuals, who demonstrated no apparent evidence of carious lesions. The numbers of caries-active boys, caries-active girls, caries-free boys, and caries-free girls were 28, 33, 13, and 11 individuals, respectively. Caries status and prevalence were determined by clinical examination, both visual and tactile, and with the use of the decay-missing-filled (DMFT) index and confirmatory radiographs. For children, 5 years of age or younger, tooth surfaces were examined to assess caries status.
Saliva collection, DNA extraction, and human oral microbial identification Non-stimulated saliva specimens (1-2 ml) were collected from the oral cavity of enrolled participants using the OMNIgene-ORAL collection and stabilization kit for microbial DNA (DNA Genotek; OM-501). In some cases, especially for young children, sterile bulbs or sterile swabs were used to collect saliva from the oral cavity, prior to placement of saliva within the OMNIgene-ORAL stabilization reagent. OMNIgene-ORAL stabilizes samples at the point of collection until transport to the laboratory, and permits long-term storage up to 1 year under ambient temperatures. The OHSU Gene Profiling Shared Resources Laboratory extracted microbial genomic DNA from the saliva specimens, using kits compatible with the OMNIgene-ORAL stabilization reagent, and subsequently quantified purified DNA by absorbance at 260 nm. Microbial genomic DNA was then subjected to agarose gel electrophoresis, in the presence of ethidium bromide, and nucleic acid integrity was then documented using the Bio-Rad Gel Doc EZ Gel Documentation system (product 1,708,270; Bio-Rad Laboratories, Hercules, CA). Microbial DNA was then subjected to PCR amplification using V3-V4 16S rDNAspecific primers and next-generation sequencing with high-throughput Illumina sequencing in the Human Oral Microbial Identification Next Generation Sequencing (NGS) HOMINGS Laboratory at the Forsyth Institute (Cambridge, MA). HOMINGS as defined by the Forsyth Institute 'permit species-level identification of nearly 600 oral bacterial taxa and genuslevel identification of remaining sequences for 129 genera'. This methodology has been pioneered by Dr. Bruce J Paster, has been used extensively in microbiome studies, including oral microbiome analyses, and is described in representative publications [17][18][19]. Briefly, oral taxa identification and abundance of microorganisms were assessed with the ProbeSeq program, where sequence targets were matched initially with species probes and then sequentially with genus probes.
Sequence targets not uniquely matched to either species or genus probes were then assessed as unmatched.

Statistical analyses
The relative abundance of oral microorganisms was compared according to caries status and by sex. All non-zero NGS count data were included in computing the relative abundance within each participant, and then averaged by the group. The differential abundance analyses of NGS count data of oral microorganisms were performed by a zero-inflated negative binomial regression model by the zinbwave and edgeR packages in the R statistical language (http:// www.r-project.org) [20,21]. The ZINB-WaVE method implemented the zinbwave package and efficiently identified excess zero count data and provided microorganism and sample-specific weights. These weights were then incorporated into the dispersion estimates of the negative binomial regression model in edgeR, in order to determine or recover the power of statistical tests of relative abundance of microorganisms between caries status or by sex. Only 298 microorganisms detected in at least nine participants were kept in the analysis to stabilize dispersion parameter estimates. Two-sided p-values for testing relative abundance by caries status or by sex were adjusted by the false discovery rate (FDR) for multiple test correction [22]. Fold changes larger than 1.5, with the FDR adjusted p-value of < 0.05 were considered statistically significant [22].

Results
Abundance of salivary microorganisms within caries-active boys versus caries-active girls Table 1 ranks the abundance of microorganisms in caries-active girls, caries-active boys, caries-free girls, cariesfree boys, caries-active children, and caries-free children; using box representations, Figures 1 and 2 display the abundance of the top 20 microorganisms differentiated by caries status and sex, respectively. Tables 2 (cariesactive boys versus caries-active girls) and 3 (caries-free boys versus caries-free girls) compares the fold changes for each microorganism between boys versus girls and ranks the fold changes with high positive numbers representing higher proportional representation of microorganisms in boys and extremely low negative values representing higher proportional representation of microorganisms in girls. The abundance list in Table 1 is sorted according to rankings of microorganisms observed in caries-active children and then applied in the same sequence for all remaining groups, and lists only the top 100 microorganisms. In Tables 2 and 3, microorganisms found in significantly-higher proportions in boys versus girls are shaded in gray, and in general, only lists microorganisms with statistically significant differences according to sex. In Table 1, Streptococci are the most abundant microorganisms observed in all groups. Prevotella melaninogenica is the second most abundant microorganism in cariesactive children while it is the fourth most abundant in caries-free children. In caries-active boys, Neisseria flavescens and Haemophilus pittmaniae were the only two microorganisms (shaded in gray in Table 1) that exhibited both significant sex differences (> 1.5-fold changes with FDR adjusted p < 0.05) and abundance comprising > 0.5% of the microbial population (Tables 1 and 2). N. flavescens and H. pittmaniae were abundant in cariesactive boys, comprising 3.92% and 1.53%, respectively, of the salivary microbial population. In caries-active girls, Alloprevotella species HOT 473 was the only species that exhibited significant sex differences (4.4-fold difference between caries-active girls and caries-active boys; p = 0.0003) and high abundance in numbers (1.85% of the total microbial population in caries-active girls).
Conversely, for caries-free children, Lachnospiraceae [G-3] species HOT 100 and Bergeyella species HOT 900 were found in significantly higher levels in caries-free boys than cariesfree girls, exhibiting 39.5 and 21.8-fold differences, respectively (  28.7-, 28.6-, 26.8-, and 16.9-fold differences, respectively (Table 3). Microorganisms were also compared between cariesactive children as a combined cohort compared to microorganisms contained within caries-free children ( Table 4). Fold changes were also determined for each microorganism contained within caries-active children versus caries-free children and then ranked from high positive values (indicating a high prevalence of specific microorganisms within caries-active children) and extremely low negative values (indicating a high prevalence of specific microorganisms within caries-free children). Microorganisms displaying statistically significant fold changes (> 1.5 fold changes with FDR adjusted p < 0.05) are displayed in Table 4; gray shading indicates a high prevalence of microorganisms found in cariesactive children compared to caries-free children. This table lists only those microorganisms displaying significant fold changes according to caries status, and omits the large majority of species (hundreds) that exhibit nominal  Figure 1. Abundance of top 20 salivary microorganisms found in caries-active and caries-free children.
Plots were calculated based on select information displayed in Table 1.
or minor fold changes between caries-active children and caries-free children.
Rationale for examining the salivary microbiome and influence of saliva on the oral microbiome and dental caries The oral microbiome encompasses heterogeneous microorganisms from different sites of the mouth, including periodontal sulcus, dental plaque, tongue, buccal mucosa and saliva (48). Saliva contains little if any indigenous microbiota, instead acquiring bacteria shed from desquamating oral epithelial surfaces and oral biofilms, including dental plaque. Dental plaque contains Streptococcus mutans, one significant microbial contributor to dental caries and tooth decay, as well as other adherent microorganisms. Therefore, the   salivary microbiome is considered to be an important reservoir that contains microorganisms found throughout the oral cavity [48,49], and upon interaction with the gut microbiome, may influence the onset of systemic diseases including inflammatory bowel disease. Thus, defining the factors underlying the constituency of the salivary microbiome are critical in understanding the complete oral microbiome and its influence in oral and systemic health and disease.
Sex differences in the salivary microbiome of caries-active and caries-free children Our data demonstrate significant differences in the salivary microbiome between caries-active and cariesfree children, consistent with the known shifts in acidity of the oral cavity in the development of dental caries. For caries-active boys, Streptococci may represent more important determinants in the onset of caries, while for caries-active girls, Veillonella may be more important factors. The high prevalence of Actinomyces, of which some species are known to have cariogenic potential, in caries-free girls ( Table 3), indicates that this microorganism may not be a major influential determinant in dental caries. The detection of several periodontal microorganisms, including S. sputigena, S. noxia, and Aggregatibacter species HOT 949, in caries-free girls, is consistent with the more neutral pH found in the non-cariogenic oral cavity. In caries-active girls, Alloprevotella species HOT 473 was the only species that exhibited significant sex differences and moderate abundance within the total microbial population (Tables 1 and 2).

Other non-microbial factors influencing sex differences in dental caries
Shaffer et al. [1] have indicated that genetics may significantly influence variability in dental caries experience between individuals. In genome-wide scans, the genetic region signifying the potential for 'higher caries experience' was located at 14q24.3, which is in close proximity to ESRRB (estrogen-related receptor beta) [16]. ESRRB is similar to the estrogen receptor and contributes to various biological functions in physiology and pathology [50]. High expression of ESRRB in females may lead to higher caries rates due to increased enamel surface sensitivity towards demineralization in the acidic oral environment [51]. Genes expressed in enamel mineralization, AMELX (amelogenin) and AMNB, as well as ESRRB were found to be associated with calcium levels in saliva. Genetic mutations in enamel mineralization are directly responsible for X-linked Amelogenesis Imperfecta, which supports their contribution in enamel matrix formation and caries vulnerability [52]. Deeley et al. [53] found that 'higher caries experience' was associated with irregular amelogenin allele markers. This may occur in the presence of X-inactivation and mosaicism and may contribute to female caries susceptibility [53].
Depending on the type and frequency of sugar introduced in the diet, a cariogenic oral environment may also develop via fermentation of sugars by acidogenic microorganisms, resulting in enamel decalcification [12]. Vegetarian diets, which contain sugar-containing fruits and berries, have been shown to increase caries incidence [54,55]. Higher caries prevalence was found in participants with vegetarian diets, while protein-meat consumption was suggested to be caries-protective because of the development of a more alkaline oral environment due to protein metabolism [56]. Females were noted to engage more in vegetarianism than males [57], with western societies containing more vegetarian women than vegetarian men, and women consuming substantially less meat in comparison to men [54].
Throughout the female lifespan, various biological stages are influenced by reproductive hormone fluctuations that coincide with changes in body structure. Estrogen fluctuations that occur during pregnancy influence suppression of the immune system and have known impact on salivary composition and flow rate contributing to a less protective oral environment and greater caries susceptibility. The parotid, buccal and labial salivary glands demonstrated lower flow rates in females than males, with elderly participants demonstrating substantial differences [15,58]. The minor salivary gland in women consistently contained substantially lower IgA concentrations among all tested salivary glands, and the higher IgA concentration found in men may provide greater protective anti-cariogenic effects in the oral cavity when compared to women [15].
Oral health disparities in men and women can be influenced by cultural and social differences [59]. Overall, dental caries unequally affects ethnic minorities and low socioeconomic status populations, with women in these categories being more affected than men [60]. Female-inclined behaviors, such as labor and domestic role in food preparation, may contribute to increased caries rates known to be associated with women. This role provides easy access to food supplies and a greater tendency towards frequent snacking [61]. Some countries exhibit this trend in caries incidence due to social and religious involvement, including son preference, fasting, and pregnancy-related dietary restrictions [62].

Limitations of the study and concluding remarks
One limitation of the study concerns the difficulty in obtaining caries-free children, which was primarily due to the referral-based, region-wide patient acquisition system at the OHSU School of Dentistry. This referral system resulted in the predominant availability of more difficult and problematic dental cases, including advanced caries, and very few caries-free individuals. We supplemented the caries-free cohort with additional patients obtained from an external practitioner's clinic, but we do acknowledge that increasing the caries-free control population may increase the statistical power of the study, and would have the potential to further discriminate borderline differences in microbial species between the sexes. Another limitation was the use of participants in different developmental dental stagesindividuals with primary dentition or mixed dentition. We initially started the study attempting to identify only young individualsages 2-6with only primary dentition, but were not able to identify sufficient numbers of participants, and later expanded our age range to age 12 years. The majority of participants in this study were ages 2-6 years (30 boys and 34 girls) and the smaller remainder (11 boys and 13 girls) were within the 7-12 years age range, many with mixed dentition and eruption of permanent central and lateral incisors, first molars and cuspids. The limitations of different developmental stages were best mitigated with the use of saliva specimens instead of plaque specimens from specific teeth, many of which were missing or replaced with permanent teeth. The use of saliva also facilitated investigation of the salivary microbiome, perhaps the most reflective of the entire repertoire of microorganisms contained within the oral cavity. Any potential limitations were also mitigated with the use of our advanced multi-stage statistical analyses, which use the criterion of abundance and statistical differences in fold changes to identify the most relevant microbial species based on caries-status and by sex.
As the primary finding in our studies, in caries-active boys, N. flavescens and H. pittmaniae were the only two microorganisms (shaded in gray in Table 1) that exhibited both significant sex differences (> 1.5-fold changes with FDR adjusted p < 0.05) and abundance comprising > 0.5% of the microbial population (Tables 1 and 2). In caries-active girls, Alloprevotella species HOT 473 was the only species that exhibited both significant sex differences (4.4-fold difference between caries-active girls and caries-active boys; p = 0.0003) as well as high abundance in numbers (1.85% of the total microbial population). Alloprevotella species are associated with neutral pH in the oral cavity and has been found in cases of caries progression, while Neisseria species are considered to be acid producers and are more conducive to the presence within a cariogenic environment. Determining the causative reason why Alloprevotella species are found more predominantly in caries-active girls, or conversely why N. flavescens and H. pittmaniae are found more predominantly in caries-active boys, is beyond the scope of this study, and could be potentially dictated by genetic, nutritional, hormonal or immune factors or mechanisms, all of which could represent avenues for future investigation.
Acknowledgments SO, EH and CL are equal contributors to this work. EH is a recipient of the Dean's Student Research Fellowship Award from the OHSU School of Dentistry, and was the OHSU-SCADA student representative at the 2019 IADR Meeting in Vancouver, BC, Canada (June [19][20][21][22]2019). KR is also the current recipient of the AADR Student Research Day Award, supporting travel to the 2020 IADR Meeting in Washington, DC (March 18-21, 2020). SO, EH, KR, BK, and MS are students in the academic DMD degree program at the OHSU School of Dentistry. CL is the Clinical Study Coordinator for our project and AP is a graduating resident in the Pediatric Dentistry program. TM and CM receive salary support from the OHSU School of Dentistry and AF has an adjunct appointment at the School of Dentistry. This project and CM were also supported in part by NIH DE024317. We thank Dr. Kim Kutsch, who provided several caries-free study participants from his private dental practice in Albany, Oregon, and for establishing and supporting with Carifree (Albany, Oregon), the Bob Bowers Memorial Award for Student Research in Dental Caries. The Bob Bowers Memorial Award helped support SO for travel to present research at the 2019 IADR Meeting. We also thank Dr. Bruce J Paster and the HOMINGS Laboratory for providing commentary on next generation sequencing procedures, and Dr. Elizabeth Palmer for assistance in patient recruitment, and Drs. Tom Shearer, Bill Knight, Bob Steelman, John Engle, John Peterson, and Dena Fischer (NIDCR) for many discussions and support.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the NIH -NIDCR [NIH DE024317].