1186 Lif Helgerson et al J Dent Res 90(10)2011 saskia exigua was detected exclusively, and in LaciobacWus Custer i Streptococcus parasanguinis i and r prevalence, in infants delivered by C-section. These findings indicate that there were differences in the microbiota of the oral Bundobactenum animals ss. animals, Brdobactenwm laces delivery method, as has been reported for TM7]G-1)sp HoT 347 morea haemolysin the microbiota of the lower gastrointestinal tract(Penders et al, 2006; Domingue Bello et al., 2010). ahonen marty In the current study, 85 species or spe- Prewotewa loeschew, Prevotella sp HOT 472(clone GU027) cies clusters out of the approximately 300 taxa evaluated by the HOMIM Capnocytophaga granulosa, Cano aga sp, HOT 326 microarray were detected in the three- month-old infants. This is fewer bacteri taxa than reported for adults in whom Leptotmchia buccalis, L goodfellowi, Sneathia sanguinegens approximately 65 to 70% of the microar ray species were detected by the same assay( Colombo et al, 2009: Preza et al. 2009). While there are no direct com- Reactivity to 24 probes (out of 85 probe reactions) that differed significantly (p parisons between the numbers of taxa 0.005)or marginally (p <0.01)in three-month-old infants born vaginally or by Caesarian detected by HOMIM in infants and those section.***ps0.005,*p<0.01 tested by Chi-square. No indication for pvalues between in adults, a lower species diversity of 0.02and0.01 infants compared with individuals in consistent with separate PLS-DA Multivariate Modeling of HOMIM reports for infants and adults(Kononen, 2000; Hao and Lee, 16S rRNA-based Microarray Data 2004: Morelli, 2008). The lower number of taxa detected in oral biofilms from C-section compared with vaginally delivered A model with two significant components was obtained b infants is in accord with the lower diversity reported for the gut PLS-DA modeling with mode of delivery as outcome (y-variable) in samples taken immediately after C-section birth(Dominguez and dichotomized HOMIM microarray signals as the x-block, Bello et al. 2010) including selected individual characteristics(see statistics)as After birth, bacterial colonization of the gastrointestinal potential confounders. This model virtually clustered infants tract, including the mouth, is influenced by the transmission of delivered by C-section separately from those delivered vagi- bacteria from the environment and by genetic factors(Mandar ally(Appendix Fig. 2). The multivariate model, which had an and Mikelsaar, 1996; Dominguez-Bello et al., 2010). In the first explanatory and predictive capacity of 62%(R=0.618)and few months of life, the major influences on the oral microbial 44%(Q=0.0457), respectively, confirmed associations found succession are person-to-person transmission, composition of in the univariate analyses Species strongly associated (VIP> the infant's saliva, mode of feeding, and microbial cross-talkIn 1.5)with being born by C-section were Slackia exigua the neonate, oral bacterial colonization starts with streptococci Lactobacillus Cluster I, veillonella sp. EF509966, Veillonella from the viridans group(Pearce et al., 1995; Kononen, 2000) atypical, and K parvula, and those associated with being vagi- whereas significant colonization of anaerobes was not detected ally delivered were S sanguinis, Streptococcus sp HOT 058, in infants before 2 mos of age(Kononen, 2000).While there are hominis (Table 3). The model rema strong(R=0.693, Q2=0.496), and the same taxa remained no comparable data with HOMIM in infants, the present fre- quent detection of species in Firmicutes, and particularly within rongly influential when the two pre-term (gestational wk 35) the genus Streptococcus, is consistent with previous reports of and all formula-fed infants were excluded oral colonization by streptococci in infants(Pearce et al., 1995 Kononen, 2005). It is unlikely that these species are transients, DISCUSSION considering the detection threshold of about 10 cells for the HOMIM microarray, indicating that species detection in this The present study investigated the oral microbiota in infants assay likely reflects colonization and growth (Paster and ferences associated by birth delivery mode. Higher numbers of mothers with antibiotics during delivery was not influent, the delivered vaginally or by C-section to evaluate if there were dif- Dewhirst, 2009; Olson et al., 2011). Notably, treatment of taxa were detected among infants delivered vaginally, compared the oral microbiota in three-month-old infants. with those delivered by C-section, with probes to the 16s rrna The present dataset was characterized by a larger number of of cultivated and uncultivated oral bacteria in a microarray variables than study participants, and by the presence of species format(HOMIM: Paster and Dewhirst, 2009 ). Further, the that might be interdependent based on shared environmental esults indicated differences in the microbiota depending on needs or inter-species co-aggregation. Under these conditions, delivery method, including a novel finding that Slackia exigua the multivariate PLS-DA method is suitable to search for subject1186 Lif Holgerson et al. J Dent Res 90(10) 2011 PLS-DA Multivariate Modeling of HOMIM 16S rRNA-based Microarray Data A model with two significant components was obtained by PLS-DA modeling with mode of delivery as outcome (y-variable) and dichotomized HOMIM microarray signals as the x-block, including selected individual characteristics (see statistics) as potential confounders. This model virtually clustered infants delivered by C-section separately from those delivered vagi￾nally (Appendix Fig. 2). The multivariate model, which had an explanatory and predictive capacity of 62% (R2 = 0.618) and 44% (Q2 = 0.0.457), respectively, confirmed associations found in the univariate analyses. Species strongly associated (VIP ≥ 1.5) with being born by C-section were Slackia exigua, Lactobacillus Cluster I, Veillonella sp. EF509966, Veillonella atypical, and V. parvula, and those associated with being vagi￾nally delivered were S. sanguinis, Streptococcus sp. HOT 058, and Cardiobacterium hominis (Table 3). The model remained strong (R2 = 0.693, Q2 = 0.496), and the same taxa remained strongly influential when the two pre-term (gestational wk 35) and all formula-fed infants were excluded. Discussion The present study investigated the oral microbiota in infants delivered vaginally or by C-section to evaluate if there were dif￾ferences associated by birth delivery mode. Higher numbers of taxa were detected among infants delivered vaginally, compared with those delivered by C-section, with probes to the 16S rRNA gene of cultivated and uncultivated oral bacteria in a microarray format (HOMIM; Paster and Dewhirst, 2009). Further, the results indicated differences in the microbiota depending on delivery method, including a novel finding that Slackia exigua was detected exclusively, and in high prevalence, in infants delivered by C-section. These findings indicate that there were differences in the microbiota of the oral cavity depending on birth delivery method, as has been reported for the microbiota of the lower gastrointestinal tract (Penders et al., 2006; Dominguez￾Bello et al., 2010). In the current study, 85 species or spe￾cies clusters out of the approximately 300 taxa evaluated by the HOMIM microarray were detected in the three￾month-old infants. This is fewer bacterial taxa than reported for adults in whom approximately 65 to 70% of the microar￾ray species were detected by the same assay (Colombo et al., 2009; Preza et al., 2009). While there are no direct com￾parisons between the numbers of taxa detected by HOMIM in infants and those in adults, a lower species diversity of infants compared with individuals in later ages is consistent with separate reports for infants and adults (Könönen, 2000; Hao and Lee, 2004; Morelli, 2008). The lower number of taxa detected in oral biofilms from C-section compared with vaginally delivered infants is in accord with the lower diversity reported for the gut in samples taken immediately after C-section birth (Dominguez￾Bello et al., 2010). After birth, bacterial colonization of the gastrointestinal tract, including the mouth, is influenced by the transmission of bacteria from the environment and by genetic factors (Mandar and Mikelsaar, 1996; Dominguez-Bello et al., 2010). In the first few months of life, the major influences on the oral microbial succession are person-to-person transmission, composition of the infant’s saliva, mode of feeding, and microbial cross-talk. In the neonate, oral bacterial colonization starts with streptococci from the viridans group (Pearce et al., 1995; Könönen, 2000), whereas significant colonization of anaerobes was not detected in infants before 2 mos of age (Könönen, 2000). While there are no comparable data with HOMIM in infants, the present fre￾quent detection of species in Firmicutes, and particularly within the genus Streptococcus, is consistent with previous reports of oral colonization by streptococci in infants (Pearce et al., 1995; Könönen, 2005). It is unlikely that these species are transients, considering the detection threshold of about 104 cells for the HOMIM microarray, indicating that species detection in this assay likely reflects colonization and growth (Paster and Dewhirst, 2009; Olson et al., 2011). Notably, treatment of the mothers with antibiotics during delivery was not influential on the oral microbiota in three-month-old infants. The present dataset was characterized by a larger number of variables than study participants, and by the presence of species that might be interdependent based on shared environmental needs or inter-species co-aggregation. Under these conditions, the multivariate PLS-DA method is suitable to search for subject Figure. Reactivity to 24 probes (out of 85 probe reactions) that differed significantly (p < 0.005) or marginally (p < 0.01) in three-month-old infants born vaginally or by Caesarian section. ***p ≤ 0.005, **p < 0.01 tested by Chi-square. No indication for p-values between 0.02 and 0.01