Bacterial diversity and functional analysis of severe early childhood caries and recurrence in India

Dental caries is the most prevalent oral disease affecting nearly 70% of children in India and elsewhere. Micro-ecological niche based acidification due to dysbiosis in oral microbiome are crucial for caries onset and progression. Here we report the tooth bacteriome diversity compared in Indian children with caries free (CF), severe early childhood caries (SC) and recurrent caries (RC). High quality V3–V4 amplicon sequencing revealed that SC exhibited high bacterial diversity with unique combination and interrelationship. Gracillibacteria_GN02 and TM7 were unique in CF and SC respectively, while Bacteroidetes, Fusobacteria were significantly high in RC. Interestingly, we found Streptococcus oralis subsp. tigurinus clade 071 in all groups with significant abundance in SC and RC. Positive correlation between low and high abundant bacteria as well as with TCS, PTS and ABC transporters were seen from co-occurrence network analysis. This could lead to persistence of SC niche resulting in RC. Comparative in vitro assessment of biofilm formation showed that the standard culture of S. oralis and its phylogenetically similar clinical isolates showed profound biofilm formation and augmented the growth and enhanced biofilm formation in S. mutans in both dual and multispecies cultures.

Biodiversity of microbiota among SC, RC and CF micro ecosystem through high quality sequencing. A total of 15,270,361 high quality reads were generated from 28 samples with an average of 391,573 per sample ranging between 110,367 and 772,880 after trimming out low quality reads (Table 1). High richness in species diversity using Shannon, Chao and ACE index with 52,628 unique OTU's after trimming out the low-quality reads were observed. RC exhibits significant difference in Chao 1 index with CF (P = 0.039), SC (P = 0.043) and in ACE with CF (0.012) and SC (P = 0.009) (Fig. 1A). Out of 52,628 OTU's obtained in the analysis SC, CF and RC are characterized by 19,984,12,473 and 8306 unique OTU's respectively. SC and CF shared 7314 OTU, CF and RC shared 4556, RC and SC shared 5320 OTU's. In common 3682 OTU's were shared by all three study groups. Representative Venn diagram illustrates the number of unique and shared OTUs in each study group (Fig. 1B). To investigate the relatedness of microbiome composition, separation of beta Taxonomical analysis of bacterial diversity and its phylogenetic relationship in SC, RC and CF group of Indian children. Upon aligning the 52,628 unique OTU's with the HOMD database at a 97% similarity level a total of 780 species from 10 phyla were classified. Percentage distribution of relative abundance of bacterial diversity at genus level was given in Fig. 2A. Among the 5 most abundant phyla Firmicutes dominated in all the three groups (CF-46.56%, SC-3.25%, RC-42.02%) followed by Proteobacteria (CF-36.99%, SC-32.98%, RC-29.55%). The next dominant group was Actinobacteria followed by Fusobacteria in SC, whereas CF was dominated by unassigned (9.3%) and RC was dominated by Fusobacteria (12.43%) and Bacteroidetes (6.43%). The abundance level of Actinobacteria was diminished in CF with 2.6% and it was slightly higher in RC with 5.1%. The Candidate Phyla Radiation (CPR) phyla TM7 was found to occur in all three groups with low abundance whereas SR1 and GN02 were represented only in the CF group. All 10 phyla constituted around 93.3% of aligned taxonomy while the remaining 6.63% constitute unassigned bacteria (Fig. S1). Col- Outliers are represented by dots. Significant differences between groups are shown by lines and the following notation. P < 0.05 (*), P < 0.01 (**), P < 0.001 (***). (B)Venn diagram illustrated the number of shared OTUs between SC, CF and RC at 97% similarity. Coloured circles represented each group and intersection part between circles represented the number of shared OTUs. (C) Principal coordinates analysis (PCoA) based on weighted UniFrac distance of community structure among all the individuals of three treatment groups. Each color circle represents one sample: pink for SC group (n = 10); blue for CF group (n = 8); violet for RC group (n = 10). (D) Box plot comparison of beta diversity analysis across treatment groups at genera level. The X-axis denotes the population studied while Y-axis denotes the corresponding Shannon-Weaver index representing the beta diversity of treatment and gender group. Interestingly the tooth biofilm of the CF group was dominated by Negativicutes followed by four different Gra-cilibacteria_GN02, three different Actinomyces sp., Veillonella sp._HMT_780, Rothia mucilaginosa and Haemophilus parainfluenzae (Fig. 3A). Thus, the significant differences observed in the abundance of Bifidobacteriales, Granulicatella, Micrococcaceae, Carnobacteriacea, Enterococcaceae and Lactobacillus between the CF and SC groups may be related to the disease condition in affected children. Additionally, the significant differences observed in the abundance of Propionibacteriales, Pseudomonadales, Lachnospiraceae, Bergeyella, Prevotella, Slackia could be related to the secondary caries formation in RC groups (Fig. 3B). It has been observed that Fusobacterium nucleatum and Lachnospiraceae_G_3_bacterium HMT 100 might be the key organism in the severity of the disease conditions (Fig. 3C).
Mutual relationship among biofilm forming caries microbiota exhibited in SC and RC ecological niche. Co-occurrence network among the core microbiome was constructed with the organisms that occur in 90% of the study group which consists of 80 nodes and 357 edges. Both positive and negative correlation among the interacting groups were shown as edges with green and red colour respectively. This network was characterized with two major clusters one with early colonizers and other with middle and late colonizers. All the late colonizers were found to interact mutually with each other. Separate clusters among the bacteria Carnobacteriaceae, Aerococcaceae, Granulicatella elegans, Enterococcus and Abiotrophia defective that belongs to the class Bacilli were observed. (Fig. 4A). The network of disease conditions constitutes 127 nodes with 2115 interactions indicating a strong association of complex networks among the bacterial community. Analysis on interacting microbes reveals that the Coriobacteria, Micrococceae was found to occur exclusively in caries condition exhibiting positive interaction with most of the early and late colonizers. S. oralis subsp. tigurinus clade 071 that shows high abundance in all three study groups had negative interaction with Kingella_sp._HMT_012 in the core microbiome. However, in the disease network S. oralis subsp. tigurinus clade 071 exhibited positive inter- Central interaction mainly occurs among Corynebacterium durum, Haemophilus parainfluenzae, Oribacterium, Enterococcaceae, Micrococcaceae, Capnocytophaga sputigena, Capnocytophaga leadbetteri, Prevotella and Lachnospiraceae. The interaction that occurs among Pasteurellaceae, Corynebacterium matruchotii, Eikenella, Fusobacterium, Veillonella tobetsuensis and Bifidobacterium dentium shows positive co-occurrence and Veillonella parvula and Gemella shows mutual exclusion with Bifidobacterium dentium (Fig. S3). Under disease condition three different strains of Saccharibacteria_TM7_(G-1,C-1,F-1) found to interact with most of the middle and late colonizers like Selenomonas, Tannerella, Prevotella, Eikenella, Lachnoanaerbaculum, Clostridia, Camphylobacter, Epsilonproteobacteria, Capnocytophaga and Aggregatibacter positively while there seems to be no interaction among the three strains of Saccharibacteria_TM7 (Fig. S4). Gracillibacteria GN02 was found to have mutual interaction with the majority of biofilm forming microbes in caries free condition.
Functional prediction of the predominant taxa among three study groups. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) was used to understand and infer the metabolic interactions between the microbiota of CF, SC and RC groups. Essential carbohydrate metabolisms that are related to caries like starch and sucrose metabolism, amino sugar metabolism, TCA cycle, fructose and mannose metabolism, galactose metabolism, glycolysis, pentose phosphate pathway in SC and RC exhibit highly significant difference from CF group (Fig. S5A). A distinct difference was not observed in pyruvate and inositol phosphate metabolism. Analysis on cell motility related pathway shows an increased level of bacterial toxins, bacterial chemotaxis, bacterial motility protein and flagellar assembly in RC and SC groups. The major transport systems related to quorum sensing Phosphotransferase System (PTS), ABC transport system show distinct differences between RC with that of the CF group. Glycan biosynthesis which is the prime function of oral bacterial taxon towards plaque formation found to be high in the RC group (Fig. 5).

Network analysis of bacterial community with functional traits. Inter and intra relationship
among the biofilm forming microbes and its predictive potential functional molecule was constructed to reveal the role of functional molecules for the ecological interactions among the bacteria in all three study groups. A www.nature.com/scientificreports/ total of 113 nodes with 2717 mutual interactions was obtained using Spearman correlation and P values were adjusted with Benjamini Hochberg multiple test correction to 0.05. More number of positive interactions was observed among different biofilm forming bacteria with that of the predicted functional traits. The network association was clustered based on the interaction among bacteria, cellular processes, environmental information processing, metabolism and organismal systems (Fig. 6). In the case of starch and sucrose metabolism we could observe a positive interaction among major bacteria like Streptococcus, Cardiobacterium hominis, Lachnospiraceae, Selenomonas, Leptotrichia, Atopobium, Prevotella oris, Campylobacter, Coriobacteria, Rothia aeria, Actinobacteria and Corynebacterium. Starch and sucrose metabolism is also positively correlated with other major functional traits like bacterial chemotaxis, pentose phosphate pathway, TCA cycle, bacterial motility proteins, flagellar assembly, N-Glycan biosynthesis, PTS, Two component system and other transport systems (Fig. S7).
Of the three transport mechanism ABC transport mechanism exhibited positive interaction with almost all the carbohydrate metabolising bacteria. Along with the above mentioned bacteria in starch and sucrose metabolism Kingella sp. HMT-012, Veillonella parvula, Betaproteobacteria, S. oralis subsp. tigurinus clade 071, Haemphilus shows positive correlation with the TCS (Fig. S8) OP 38) and all standard cultures showed positive results by producing dark black colonies in the agar plate indicating the production of exo-polysaccharide which is a main factor for biofilm formation. Representative CRA assay was given in (Fig. S10A). Phylogenetic comparison between the biofilm forming clinical isolates and S. oralis (MTCC) by BOX PCR showed 5 OPs closer to S. oralis grouped in the same clade. OP32 was the closest to S. oralis (MTCC) while OP12 and OP 29 showed branches near to the standard culture (Fig. 7A). OP32 was used for further analysis. Co-culture of CRA positive (OP32) with CRA negative along with standard colonies showed no antagonistic interaction. This indicated OP32 did not show negative growth impact on non-biofilm forming bacteria and vice-versa (Fig. S10B).

Effect of sucrose on biofilm formation on dual and multispecies biofilm growth over 24 h and 48 h.
Biofilm forming potential of the standard culture as planktonic, dual and mixed species at various time intervals were performed. In all the cultures, exponential growth was observed from 1 hr to 10 h and declined growth in 24 and 48 h culture which could have happened due to deficit in nutrient availability (Fig. 7B). When compared to the planktonic cells, the co-cultures (dual cultures) showed increased biofilm formation especially for C. albicans (MTCC) and S. mutans (MTCC). Both exhibited increased production of biomass in the presence of S. oralis (MTCC). Similarly, significant increase in growth of biomass in mixed species was observed at  (Fig. 7D). The pH of the biofilm was observed to decline over 48 h of growth period ranging below 5 in all the planktonic and multispecies cultures in the presence of sucrose when compared to the control media. S. mutans (MTCC): S. oralis (MTCC) and OP32: S. mutans (MTCC) biofilms had the largest decrease in supernatant pH. In the case of C. albicans (MTCC) dual culture we could observe slightly elevated pH above 5 (Fig. 7E).

Discussion
Severe early childhood caries (SC) and its recurrence (RC) is a critical health concern affecting both physical and psychological health of a child. The oral microbiome dysbiosis and formation of conducive cariogenic niche results in severe caries 16,17 . Poor oral hygiene and generalized treatment methods do not completely eradicate the caries forming bacteria that might lead to persistence of these bacteria causing recurrence in Indian children. www.nature.com/scientificreports/ High-throughput sequencing methods were used in this study to untap the hidden diversity 13 and its varied abundance among CF, SC and RC children in India. In general, we observed varied diversity of bacteria among the study groups. We could observe high bacterial diversity in SC as compared to CF group contradicting earlier reports 4,18-20 (Fig. 1B) while RC group was found to exhibit low diversity of bacteria enriched with late colonizers. Notably we also encountered very low diversity of bacteria in sample P19 from RC that was found to be enriched with Firmicutes (98.30%). The low diversity in RC is in corroboration with previous studies 18,20 . However, differences in beta diversity analysis revealed there could be varied proportion of taxa in RC form that of CF and SC (Fig. 1B). This might be the consequence of significantly higher metabolic activities of microbiota that contribute to host-microbiome interaction when compared with SC and CF, triggering its survival in the unique ecological niche in RC. However, significant correlation was not found between gender and any other oral demographic status with caries occurrence, similar to earlier finding 21 .
Most of the caries microbiome related studies are based on saliva samples 22 because it is a patient friendly non-invasive approach and is easy to collect. However, this may not address the complete details of potential microbial community composition, as it varies at different intraoral sites due to the diversified ecological niche present in oral cavity 23 . In this study, the samples collected from the lesions of carious dentine gives an insight into the variation in diversity of bacteria as well as its notable community interaction in both SC and RC condition. To the best of our knowledge, the current study provides a preliminary snapshot of the tooth biofilm microbiome variation of Indian children affected by severe early childhood caries and its recurrence. About 80% of OTU based taxonomical identification constitutes to the core microbiome (Fig. 5A) of the study group which includes www.nature.com/scientificreports/ the representation from almost all the reported phyla except for Fusobacteria found abundance in RC and SC groups. This clearly indicates that presence of Fusobacteira plays a vital role in caries formation and progression. The decrease in abundance of dominant phyla like Firmicutes, Proteobacteria in such a way CF > SC > RC corroborated with the reported low diversity of bacteria occurring upon caries progression 19,20 . The observed correlation of interspecies between the low abundant heterogenic oral microbiome and the resultant niche pH might play a determining role in caries formation and its recurrence. Our study showed significant variations in low abundant species among all the study groups which may act as a potential biomarker. Late colonizers like Prevotella saccharolytica, Campylobacter gracilis, Pseudomonas, Neisseria oralis, Coriobacteriaceae, Atopobium, Bergeylla, Fusobacterium HMT 203, Scardovia wiggsiae, Selenomonas flueggei, Stomatobaculum, Oribacterium parvum, Parvimonas and Peptostreptococcaceae in SC and its maintenance in RC (Fig. S6) shows its capability of being key pathogens in the severity of the disease. This augments the keystone hypothesis, that relative abundance of prevalent taxon may not be strongly attributed to the progression of caries 15 . Co-occurrence network analysis exhibits positive association among these microbes and also with that of the early colonizers emphasizing their role in severity of caries, predicting an incomplete elimination of these microbes from the caries niche with current therapeutic methods 20,24 . Among the oral commensals the dysbiosis of Streptococcus plays a pivotal role in caries formation and progression. However, the relative abundance of S. mutans was very low in our observation on Indian children, though it has been reported as a pioneer in caries forming bacteria 25,26 . We could find only 5% of S. mutans in the SC microbiome and 2% in RC (Fig. S1B,C) which clearly indicates the scarcity of S. mutans in advanced caries, as seen before 6,19,27 . Whereas the non-mutan Streptococcus and S. oralis subsp. tigurinus clade 071 was found to be abundant in all the study groups subjected in this report, particularly high in RC (Fig. S2). The specific biochemical activities in Streptococcus oralis subsp. tigurinus clade 071 is not clearly understood, however, its significant abundance in caries depicts a possible overlap in biochemical properties of acting as a flux and AI 24 activity that helps in quorum sensing similar to S. oralis.
In vitro analysis of the clinical isolates on biofilm formation supports this. The clinical isolates OP32, OP12 and OP20 were found closely related to S. oralis (MTCC) by genetic fingerprinting analysis using BOX PCR that helps to identify the bacterial strains at species level 28,29 . Clinical isolates closer to S. oralis MTCC might be S. oralis subsp. tigurinus clade 071 as seen from the NGS data analysis and closer phylogenetic clade. These oral isolates are supposed to be the subgroup of S. oralis as seen earlier 30 . The positive growth interaction between the OP32 with the non-biofilm forming clinical isolates show that the interspecies interaction is significant in the SECC microbiome. Further, the increased biofilm formation in co-culture of S. oralis and OP32 with the major cariogen S. mutans under high sucrose environment show the cariogenic interrelation between the species, as observed earlier 31,32 . Also high pH reduction in the dual culture S. mutans: S. oralis and S. mutans: OP32 co-culture suggests its role in acidification of oral niche.
Additionally the interaction of interkingdom microbe C. albicans enhances the biofilm forming capability in both standard and clinical isolates. Recent report states that the glucans produced by gtfR mediates crosskingdom interaction between C. albicans and S. oralis which in turn have positive effect on stronger biofilm matrix formation 31 . Similarly, we observed increased biofilm formation by S. oralis MTCC and OP32 in the presence of C. albicans both in dual and multispecies especially in the presence of sucrose and positively correlated with incubation time.
Increased level of starch and sucrose metabolism along with metabolites like ABC transporters, Phosphotransferase system, Two Component system in functional prediction analysis and its positive interaction with S. oralis subsp. tigurinus clade 071 (Fig. 6) supports our in vitro findings. These processes support an acidic environment that might allow growth of low abundant bacteria like Micrococceae, Abiotrophia defectiva, Granulicaterlla elegans, Aerococcaceae and Carnobacteriaceae. Moreover, the interaction network analysis on Two Component System which is one of the key player of biofilm formation (Fig. S8) clearly shows that S. oralis subsp. tigurinus clade 071 positively correlates with TCS and anaerobic late colonizers like Capnocytphaga leadbetteri, Capnocytphaga granulosa and Aggrgatibacter. These in turn may exhibit positive interaction with signal transduction mechanisms inviting other bacteria to interact with the biofilm more firmly and might cause the severity of the conditions. However high abundance of other Streptococcus genera in SC and RC groups and its co-occurrence with other disease causing microbes and metabolic pathways exemplifies that the untapped species level diversity of Streptococcus still exists in the oral cavity that contributes to a greater extent in caries formation and progression. Based on the observations we envisage that Streptococcus oralis subsp. igurinus clade 071 initially colonize the tooth surface, with its carbohydrate metabolism with EPS production resulting in increased quorum sensing signals that invite other unidentified Streptococcus making the environment more acidic and further allows all the other middle and late colonizers to form a strong network that leads to demineralization and rupturing of teeth which may cause Severe early childhood caries.
Leptotrichia the third largest significant bacteria in RC group (Fig. S2) could be one of the key pathogens in RC because of its reported property to metabolize sucrose and its isomers with the help of PTS to lactic acid in the absence of S. mutans 33 . Significantly high PTS in RC group (Fig. 6) and its positive interaction with Leptotrichia in the network analysis (Fig. 7) in the current report gave a piece of clear evidence that Leptotrichia could create acidic niche, thereby inviting other acid tolerant bacteria such as Capnocytophaga, Aggregatibacter, Fusobacterium, Campylobacter, Granulicatella, Gemella and Prophyromonas (Fig. S9). Association of these bacteria develops a strong network of complex microbial associations that may resist any treatment methods resulting in recurrence.
Unusually, among other abundant species, we found a potent cariogen V. parvula [34][35][36][37] high in CF individuals compared to RC and SC (Figs. 3, S2). The inability of V. parvula to enhance caries formation in these CF individuals could be linked with the high abundance of non-cariogenic arginolytic bacteria Neisseria and V. dispar 38 in the CF oral microbiome that might neutralize the acidification of the niche through arginine metabolism 19 .
Scientific Reports | (2020) 10:21248 | https://doi.org/10.1038/s41598-020-78057-z www.nature.com/scientificreports/ The PICRUST analysis showed high arginine metabolism in the CF group, possibly from the profound Neisseria and V. dispar in CF (Fig. S5B). Occurrence of Micrococcaceae and Bifidobacteriaceae in SC and RC groups, and not in CF shows its intrinsic role in caries formation and especially recurrence condition. We found that Micrococcaceae interacts positively with most of the reported facultative anaerobes and with the major predicted carbohydrate metabolism like starch and sucrose, fructose and mannose and amino acid metabolism. This could play a critical role in metabolism of carbohydrate even in the anaerobic environment of SC children that could maintain the acidic niche. The abundance of Bifidobacterium in RC corroborated with its fluoride resistance property with the help of bifid shunt, a bypass pathway, under fluoride inhibition of enolase to reduce the pH 39 that could lead to recurrent caries. The co-aggregation of microbes is of prime importance among oral commensals dysbiosis that leads to complex network of biofilm formation resulting in caries manifestation 24 . Most of the core microbiome exhibited positive interaction with each other, indicting the microbes are mutually benefiting from each other's metabolites. Co-occurrence network analysis showed that the interaction between microbes resulting in recurrence is not a single large hub, instead, several inter-related hubs of co-occurrence and mutual exclusion. Significant hubs like Pseudomonas, Bergeyella show mutual exclusion and Neisseria shows positive interaction with most of the bacteria. Prominent bacteria like Actinomyces, Neisseria, Fusobacterium, Corynebacterium, Prevotella, Bifidobacteria, Olesenella, Selenomonas, Oribacterium and Gracillibacter form a complete network of early, middle and late colonizers both in high and low abundant with different kinds of interactions depicting their role of heterogeneity in recurrence as well as severe caries (Fig. S3A). V. tobetuensis, one of the early colonizers possess very high autoinducer 2 activity, a strong quorum sensor 40 , co-aggregated to Fusobacterium in recurrent groups which could be a potent quorum sensing bacteria in the microbiome of Indian children especially in RC. This is supported by the increased sugar metabolism that resulted in increased Glycan biosynthesis in RC (Fig. 6) forming a strong matrix of biofilm with the help of bridging bacteria like Fusobacteria and Corneybacerium to link the early and late colonizers in the biofilm (Fig. S3B) 1,41 . A different bridging bacteria Corynebacterium matruchotii and Corynebacterium durum which produce acid from mannitol and galactose 42 family were found high in RC. This has a long filamentous structure that acts as an anchoring site for other microbes 1,2 . Early colonizers like Capnocytophaga leadbetteri, Haemophilus, Streptococcus and late colonizers like Oribacterium, Prevotella, Capnocytophaga sputigena were interacting with Corynebacterium rather than Fusobacterium, while the abundance of Fusobacterium was more. This might add a demographic effect in the Indian population, as well as synergistic or competitive effect between Fusobacteria and Corynebacteria (Fig. S3).
As most of the earlier studies on dental caries have focused on the dysbiosis in oral microbiome 20,36 , wherein functional properties of these species in causing caries and its recurrence is contentious 6 . Positive interactions among the predicted carbohydrate metabolism and quorum sensing systems like ABC transport system, TCS, PTS with that of middle and late colonizers like Prevotella, Cardiobacterium, Carnobacteriaceae, Selenomonas, Prophyromonas, Micrococcaceae, Rothia, Bifidobacterium, Corynebacterium, Oribacterium, Lachnoanaerobaculum, Actinomyces and Saccharibacteria (TM07) could be correlated to the pathway of energy flow among these microbes that could enhances the severity of the condition by enriching more number of aciduric or acidogenic bacteria and evading the treatment method due to complexity of the bacteria present at that particular ecological niche. Significant difference in flagellar assembly between caries free and caries condition could be related to the abundance of late colonizer Selenomonas and also other flagellated organisms. Positive associations of three members of CPR phyla members Gracillibacteria GN02, Absconditabacteria SR1 and Saccharibacteria TM7 with that of three study groups CF and SC, RC respectively was observed. These findings are in line with the studies that proved the role of SR1 with H 2 S production 43 and TM7 as obligate symbiont with Actinomyces species in oral disease conditions 44 . From this study we could infer that GN02 could be considered as a potential biomarker for the healthy status and TM7 for caries status of oral cavity (Fig. 4).
Thus, from the results, we observed the significant interaction among overrepresented and underrepresented bacteria and the functional metabolic interrelationship is essential in SC as well as its recurrence. In Indian children, we observed different combinations of middle and late colonizers in the polymicrobial system with the presence or absence of the so far reported early colonizers. This tends to develop a strong network of microbial association in caries condition that could develop resistance against most of the treatment methods and also nullifies the effect of resident bacteria that are trying to regain the acid environment which leads to the recurrent condition.

Conclusion
Severe Early Childhood Caries is a chronic and prevalent disease that affects the overall growth of the children, and recurrence after treatment persists also in adulthood. This study reports the nature and heterogeneity of microbiome in the dentine micro-ecological niche among the healthy and caries affected children in Indian subpopulations. This will help to develop some simple and cost effective solutions to treat this disease. Increased depth of high throughput sequencing analysis allowed us to identify significant high and low abundant bacteria and its co-occurrence towards the aetiology of caries formation and its recurrence in Indian children. From our study, we understood that S. oralis subsp. tigurinus clade 071 is the prevalent and significant among caries affected Indian children instead of S. mutans and is a potent cariogen. Detailed investigation of S. oralis subsp. tigurinus clade 071 towards caries etiology and its inter-kingdom ecological interactions would deepen the knowledge about its persistence towards treatment method and how it evades the host immune system. The biomarker and co-occurrence analysis have given a better insight on the contribution of low abundant bacteria towards the severity of the disease. Our findings significantly mitigate the understanding of changes in bacterial profile in response to interaction with the predictive functional traits especially quorum sensing signals at different stages of childhood caries. This has opened up new dimensions into the oral microbiome at the community level Scientific Reports | (2020) 10:21248 | https://doi.org/10.1038/s41598-020-78057-z www.nature.com/scientificreports/ under different micro-ecological niche to support future metabolomics and transcriptomic studies, coupled with functional assays, thereby contributing for the development of novel strategies to identify and manage the risk of caries in children. There was no genetic relationship among the individuals of both genders and they do not share a homogeneous living environment. Children treated with antibiotics within 3 months of the study visit were excluded. The dental health and hygiene of every subject in the study was assessed by a team of professional dentists satisfying the DMFT index standards 45 . A questionnaire on child's food habits, brushing habit, caries prevalence in their family, especially intake frequency of fermentable carbohydrates excluding their regular interval of meals like breakfast, lunch and dinner were also collected. A total of 55 children were divided into three groups, 1. No clinical evidence of caries experience with zero DMFT was considered as Caries-Free (CF): n = 15, 2. Children with lesion and decayed teeth with DMFT ≥ 4 were considered as Severe Early Childhood Caries (SC): n = 20. 3. Children with secondary caries and filled teeth for primary caries were considered as Recurrent Early Childhood Caries (RC): n = 20. After thorough dental examination of the children by the dentist, the teeth were gently wiped with cotton rolls followed by gentle air stream to avoid saliva contamination and biofilm was obtained using soft ended swab (Himedia, India) according to the prescribed manual of procedures for Human Microbiome (http://hmpda cc.org/resou rces/tools _pot.cols. php) with minor modifications and transported to lab within 4 h of sample collection 46 . DNA extraction from swab samples for metagenomic studies. Bacterial DNA was extracted using QIAamp DNA microbiome Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions with minor modification in the pathogen lysis step to enhance better lysis Gram's positive and Gram's negative organisms. The enriched microbial DNAs were purified by ethanol precipitation. The DNA concentration was estimated spectrophotometrically using a NanoDrop spectrophotometer (Thermo Electron Corporation, USA), and molecular size was estimated by agarose gel electrophoresis. DNA samples that had passed the QC with NanoDrop concentration > 10 ng/µL was used for further processing. DNA samples were then stored at − 20 °C until further use. Sequence analysis. The reads obtained from Illumina platform were filtered to discard sequences with an average Phred score > 30 and sequences containing incorrect barcodes and/or lacking primer sequences. Filtered reads were de-replicated, followed by subsequent removal of singletons and identified chimeras and subjected to further analysis using QIIME pipeline (version 1.9.1) 48 . The processed reads were then clustered at 97% similarity using UCLUST to identify the species-level Operational Taxonomic Units (OTUs) and the reads were mapped to the filtered OTUs to determine the exact count of each OTU in each sample 49,50 . To determine the bacterial genera the OTUs were BLAST against the HOMD database for human oral bacteria 51 . The species or phylotypes were identified by their HOMD identity. OTUs with less than 10 sequences were excluded from the analysis 37 . The sequences which could not be assigned to a genus in HOMD (at 97% identity) were considered as the probable candidates that might be unique or novel in the oral cavity of the Indian populations.

Materials and methods
Biomarker and network analysis. In order to identify the potential biomarker, LDA effect size (LEFSe) was performed to find out the differentially enriched taxa among the groups, the threshold for discriminative features was set to 3.0 52 and the results were displayed in a cladogram and a bar graph 53 . Krona plot was performed to find out the percentage of enriched taxa between groups 54 . The functional prediction of microbiota was done with PICRUST. Only reads identified in closed reference picking were used for PICRUST analysis and aligned to the Kyoto Encyclopedia of Genes and Genomes (KEGG) data base 47 . Co-occurrence network analysis was performed to understand the inter and intrarelationship of oral microbiome with its predicted functional potentials obtained from PICRUST analysis among the study groups using CoNet 55 and the model was visualized using Cytoscape 56 57 . Co-culture of the clinical isolates and standard colonies were performed in BHI agar. The co-culture growth was observed at 37 °C growth for 24 h 58 . Biofilm was formed by pipetting out 10 µL of OD cultures for planktonic cells and equal concentration for dual and mixed cells on the wells of a flat-bottom tissue-culture treated 96-well microtiter plate containing 100μL of BHI broth. For some experiments the media were supplemented with 1% W/V sucrose. The plates were incubated statically for 24 h at 37 °C for biofilm formation. After incubation, the biofilm biomass by crystal violet assay and pH assay was performed as described previously 59 .
BOX PCR. The  Statistical analysis. OTU table that was created by QIIME analysis contains raw counts were normalized to a relative abundance OTU table. Using this relative abundance, similar types of taxa were aggregated at the phylum, class, order, family, genus and species level 48 . Biodiversity between classified groups were examined using non-parametric Mann-Whitney and/chi-squared, Fisher's exact test. Significant differences in the alpha diversity indexes between the different groups (P < 0.05) were identified by Student's t-test 46 . Weighted Unifrac distance of OTU samples were used to perform PCoA to analyze the differences among groups (beta diversity). Differences in the Unifrac distances for pairwise comparisons among groups were determined using Multiple t-test and visualized by the construction of a box and whiskers plot. The correlation was evaluated by the Spearman Correlation. The graphical representation of the results was performed by using GraphPad Prism version 6.01 (GraphPad Software, La Jolla California USA).

Data availability
The raw reads generated in this study were deposited into the NCBI Sequence Read Archive (SRA) database under Bioproject Accession Number PRJNA454811.