Species-level characterization of saliva and dental plaque microbiota reveals putative bacterial and functional biomarkers of periodontal diseases in dogs

Abstract Periodontal diseases are among the most common bacterial-related pathologies affecting the oral cavity of dogs. Nevertheless, the canine oral ecosystem and its correlations with oral disease development are still far from being fully characterized. In this study, the species-level taxonomic composition of saliva and dental plaque microbiota of 30 healthy dogs was investigated through a shallow shotgun metagenomics approach. The obtained data allowed not only to define the most abundant and prevalent bacterial species of the oral microbiota in healthy dogs, including members of the genera Corynebacterium and Porphyromonas, but also to identify the presence of distinct compositional motifs in the two oral microniches as well as taxonomical differences between dental plaques collected from anterior and posterior teeth. Subsequently, the salivary and dental plaque microbiota of 18 dogs affected by chronic gingival inflammation and 18 dogs with periodontitis were compared to those obtained from the healthy dogs. This analysis allowed the identification of bacterial and metabolic biomarkers correlated with a specific clinical status, including members of the genera Porphyromonas and Fusobacterium as microbial biomarkers of a healthy and diseased oral status, respectively, and genes predicted to encode for metabolites with anti-inflammatory properties as metabolic biomarkers of a healthy status.


Introduction
T he oral ca vity of mammals r epr esents an extr emel y heter ogeneous ecological niche due to its multiple distinct micr oenvir onments, such as teeth, g ing ival sulcus , tongue , cheek, and hard and soft palate (Kilian et al. 2016, Lamont et al. 2018, Zhang et al. 2018, Ruparell et al. 2020a ).For such reason, the bacterial ecosystem inhabiting the oral cavity, generally known as oral microbiota, is considered, together with the gut microbiota, as one of the most complex and dynamic microbial communities of the mammalian body (Lamont et al. 2018, Zhang et al. 2018 ).Millions of years of coev olution betw een this indigenous micr obial comm unity and its host have led to the establishment and consequent consolidation of v arious tr ophic inter actions that ultimatel y influence the or al and systemic well-being of the host (Kilian et al. 2016, Gr av es et al. 2019, Bell et al. 2020 ).In this context, the or al micr obial ecosystem not only acts as a protective barrier competing with exogenous pathogens and supporting the oral tissue integrity, but it is also involved in the stimulation/regulation of the host mucosa and immune system (Kilian et al. 2016, Lamont et al. 2018 ).Ho w e v er, beyond these beneficial activities, the oral microbial ecosystem also plays a r ele v ant r ole in the onset of v arious or al diseases, including g ing ivitis , periodontitis , and dental caries (Costalonga and Herzberg 2014, Yamashita and Takeshita 2017, Zhang et al. 2018, Bell et al. 2020 ).
An alteration of the oral ecosystem equilibrium may rapidly e volv e to a dysbiosis status, generating a favorable environment for pathogen colonization and pr olifer ation with a subsequent stimulation and exacerbation of the host immune response contributing to c hr onic inflammation (Wallis et al. 2015, Bell et al. 2020, Santibanez et al. 2021 ).
Specificall y, the de v elopment of periodontal diseases is primarily caused by bacterial plaque accumulation on the periodontium through biofilm development (Dewhirst et al. 2012, Davis et al. 2013, Gr av es et al. 2019, Ruparell et al. 2020a, 2021, Oba et al. 2021a ).As the biofilm thickens through bacterial adhesion, the dental plaque extends into the subg ing ival sulcus with subsequent oxygen depletion pr omoting anaer obic bacterial proliferation (Oba et al. 2021a ).At this stage, the calcification of dental plaque may occur due to the exchange of calcium and phosphate ions in saliva leading to the accumulation of additional plaque and, consequently, further irritation of g ing ival tissues (Logan 2006, Carr eir a et al. 2015, Oba et al. 2021a ).Finally, the persistence/continuous growth of suprag ing ival and/or subg ing ival plaque can cause inflammation of the adjacent g ing ival tissues soliciting inflammatory response cascade (Oba et al. 2021a ,b ).
Gingivitis and periodontitis r epr esent the most common oral diseases in dogs, affecting up to 70% of canine patients, a percentage that is reported to dramatically increase for extrasmall and small dog breeds or with increasing age (Hoffmann TaG 1996, Kortegaard et al. 2008, Wallis and Holcombe 2020, Oba et al. 2021a,b , Wallis et al. 2021a ,b ).Because of its role in periodontal diseases, the assessment of the microbial community of saliva and dental plaque in dogs is of fundamental importance.In this context, the growing concern for the health of this companion animal has prompted the scientific community to characterize the canine oral microbiota (Ruparell et al. 2020a, Oba et al. 2021a ).Ho w e v er, despite the latter has been lar gel y studied through amplicon-based next-generation sequencing techniques and culture-dependent approaches (Davis et al. 2013, Wallis et al. 2015, Flancman et al. 2018, Ozavci et al. 2019, Ruparell et al. 2020a, Oba et al. 2021a, Wallis et al. 2021a ), an accurate and precise dissection of the canine oral microbiota composition down to the species le v el as well as the identification of taxonomic and functional microbial biomarkers related to periodontal pathologies are still far from being fully dissected.
In the current study, the species-level taxonomic composition of the oral microbiota of a total of 66 dogs, divided into 30 healthy dogs, 18 dogs with c hr onic g ing ival inflammation (CGI), and 18 dogs with periodontitis, was assessed through a shallow shotgun sequencing a ppr oac h.Specificall y, for eac h enr olled dog, a saliv a sample, as well as a dental plaque sample from both an anterior and posterior tooth of the same individual were collected.The obtained data were subsequently used to explore the oral bacterial composition of dogs, identifying the most pr e v alent and abundant species of the saliva and dental plaque microniches as well as possible taxonomic differences related to the location of dental plaques .Furthermore , possible taxonomic and functional differences wer e inv estigated in corr elation with health and disease to identify any bacterial biomarker for CGI and/or periodontitis.

Ethical statement
The study protocol was approved by the "Comitato Etico per la Sperimentazione Animale" (CESA) of the University of Parma (reference number: PROT.N. 22/CESA/2021) and conducted in compliance with the rules , regulations , and recommendations of the Ethical Committee of the University of Parma.All animal procedures were carried out in accordance with national guidelines (Decreto Legislativo 26/2014).Furthermore, signed informed consent was obtained from the owners of each dog involved in this study.

Sample collection
For the purpose of this study, a total of 66 dogs wer e enr olled, divided into 30 healthy dogs together with 36 dogs equally divided into dogs affected by CGI and periodontitis, r espectiv el y ( Table S1 ).All samples were collected within 2 months thanks to the collaboration with three veterinary clinics located in the North of Ital y.For eac h subject, a dental plaque sample was collected fr om maxillary teeth, including both anterior (tooth position C or I3) and posterior (tooth position P4 or M1) tooth together with a salivary sample ( Table S1 ).Specifically, when sufficient material could not be collected from anterior tooth C or posterior tooth P4, dental plaques were harvested from the adjacent teeth.Furthermore, while dental plaque samples were manually collected through a dental prophylaxis during a routine veterinary visit, saliva was retrie v ed by rolling a sterile swab under the tongue of each dog for at least 30 s.In detail, to determine the clinical status of each dog, a g ing ivitis score between 0 and 4 was recorded for each tooth using a combination of the g ing ival index and sulcus bleeding index, by e v aluating pr obing depth, g ing iv al r ecession, and furcation exposur e, as pr e viousl y described (Lobprise 2019, Rupar ell et al. 2021, Wallis et al. 2021a ).
Once collected, dental plaque and saliva were separately conserved into dedicated sterile tubes containing 1 ml and 3 ml of phosphate buffered saline, respecti vely, k e pt on ice and shipped to the laboratory under frozen conditions where they were preserved at −20 • C until they were processed.Specifically, since different taxonomic compositions characterize the various stages of dental plaque de v elopment, to be compar able, onl y matur e dental plaque samples were collected (Holcombe et al. 2014, Jiang et al. 2021 ).In detail, since early dental plaque turns into mature biofilms within 15/20 da ys , to collect only mature dental plaque, the latter were collected from dogs that had not undergone a veterinary dental hygiene pr ocedur e for at least 2 months (Herrmann et al. 2020 ).Furthermore , to be in volved in this study, dogs had to be free from any pathology except for the ones under examination, and not having undergone treatment with any probiotics, antibiotics, or other drugs during the month prior sample collections .Furthermore , for each dog, age , sex, breed, weight, diet, and possible use of dental c he ws or medical devices for canine dental cleaning were annotated ( Table S1 ).

DN A extr action and Illumina shallo w shotgun sequencing
DNA extr action fr om dental plaque samples was performed using the ZymoBIOMICS DNA Miniprep Kit (Zymo Researc h Cor por ation, USA), following the manufacturer's instructions.
Subsequentl y, the extr acted DN A w as pr epar ed using the Illumina Nexter aXT DNA Libr ary Pr epar ation Kit and following the Illumina Nexter aXT pr otocol.Briefly, DN A samples w ere enzymaticall y fr a gmented, barcoded, and purified encompassing magnetic beads .Furthermore , samples were quantified using the fluorometric Qubit quantification system (Life Technologies, USA), loaded on a 2200 Tape Station Instrument (Agilent Technologies, USA) and normalized to 4 nM.A paired-end sequencing was performed using an Illumina MiSeq sequencer with MiSeq Reagent Kit v3 for 600 cycles (Illumina Inc., San Diego, USA).

Shallow shotgun sequencing analysis
The obtained .fastqfiles were filtered to remove reads with a quality of < 25 and sequences of Canis lupus familiaris DNA, while reads with a length of > 149 bp were retained.Specifically, to remov e r eads with high sequence identity to the C. lupus f amiliaris DNA, the obtained reads wer e compar ed to a dog-based genomic database through a METAnnotatorX2 pipeline function (Milani et al. 2021 ).Quality-filtered data were used for taxonomic profile reconstruction by using the METAnnotatorX2 pipeline, as pr e viousl y described (Milani et al. 2021 ).Specificall y, when the softwar e fails to assign a sequence corr ectl y and accur atel y to an alr eady isolated and c har acterized species, the sequence is classified only down to the genus le v el (Milani et al. 2021 ).For each downstream anal ysis, onl y classified bacterial-r elated r eads wer e consider ed ( Table S2 ).Functional profiling of the sequenced reads was per-formed with the METAnnotatorX2 bioinformatic platform (Milani et al. 2018(Milani et al. , 2021 ) ). Functional classification of reads was performed to r e v eal metabolic pathways based on the MetaCyc database (release 24.1) (Caspi et al. 2016 ) through RAPSear ch2 softw are (Ye et al. 2011, Zhao et al. 2012 ).

Sta tistical anal ysis
Origin 2021 Pro ( https:// www.originlab.com/2021 ) and SPSS softw are ( www.ibm.com/software/ it/ analytics/ spss/ ) were used to compute statistical analyses.In detail, pairwise Kruskal-Wallis test analyses tested differences in alpha diversity that is calculated through species richness and Shannon index.Moreover, similarities between samples (beta-diversity) were calculated by Bray-Curtis dissimilarity matrix based on species abundance.Beta-diversity was represented through PCoA using the function "ape" of the R suite package .Furthermore , the fitting analysis was performed through "envfit" function in RStudio and results were plotted on a PCoA using the "plot" function in RStudio.Specifically, the "envfit" function calculated the fitting score for each tested envir onmental v ariable, and the empirical P -value was corrected through the Bonferroni test.The function "silhouette" in RStudio was used to perform the Silhouette analysis and the functions "hclust" and "cutree" in RStudio wer e used to calculate the hier archical clustering of samples, both based on species-level bacterial composition and in particular Bray-Curtis dissimilarity matrix.Mor eov er, the lme4 mixed linear model (10.18637/jss.v067.i01),whic h included a ge, gender, breed, diet, dental cleaning and type of sample as fixed variable, along with clinical status as a random effect, was calculated on Shannon index biodiversity through "lmer" function in RStudio.Furthermore, a specific RDA based on the species-le v el taxonomic pr ofiles, follo w ed b y an ANOVA permutation test, was performed using the "rda" function from the "v egan" pac ka ge in RStudio.
SPSS softw are w as also exploited for the calculation of the nonparametric test for two independent samples , i.e .the Mann-Whitney U-test, and the nonparametric test Kruskal-Wallis.Specifically, the Mann-Whitney U-test results were subjected to correction for multiple comparisons using the FDR through the EdgeR pac ka ge.Furthermor e, for Kruskal-Wallis statistic, the Bonferroni post hoc test was calculated.

Data deposition
Raw shallow shotgun sequencing data are accessible through SRA under BioProject number PRJNA863598.

Taxonomic classification of the oral bacterial community in dogs
To explore the species-level taxonomical composition of the oral microbiota of dogs, a total of 66 dogs were enrolled ( Table S1 ).For each dog, a saliva sample and two dental plaque samples, obtained from both an anterior and posterior tooth, were collected ( Table S1 ).Furthermore, to investigate possible differences in taxonomic composition and functional potential of these two or al micr onic hes in case of or al diseases, samples wer e collected from healthy dogs (Alessandri et al. 2020 ) as well as from dogs affected by one of the two major canine oral diseases , i.e .CGI (18 dogs) or periodontitis (18 dogs) ( Table S1 ).Bacterial DNA extracted fr om eac h sample w as submitted to shallo w shotgun sequencing and data obtained wer e anal yzed thr ough the METAnnotatorX2 pipeline (Milani et al. 2021 , Lugli andVentura 2022 ).Illumina se-quencing generated a total of 7 282 980 filter ed r eads with an aver a ge of 36 782 classified reads per sample, ranging from a minimum of 10 976 to a maximum of 98 200 reads ( Table S2 ).T hus , an adequate number of reads , i.e .at least 10 000, was r eac hed to predict reliable taxonomic profiles from each sample, through the METAnnotatorX2 software, as previously described (Milani et al. 2021 ).Details regarding the taxonomic classification at the phylum-and genus-le v el ar e r eported in the Supplementary text ( Tables S3 and S4 ).
The gener ated species-le v el taxonomic pr ofiles wer e used to explor e the α-div ersity among samples.In detail, no significant differ ences wer e observ ed in bacterial complexity among the various types of samples , i.e .saliva or dental plaque from canines or molars, neither through the species richness nor the Shannon index analysis (Kruskal-Wallis test P -value > .05for both statistics) (Fig. 1 A), suggesting a similar bacterial complexity among the considered oral microniches.At the same time, no differences in microbial richness w ere recor ded b y separating samples based on the clinical condition within each sample type (Kruskal-Wallis test P -value > .05),except for dental plaques collected from the molars (Fig. 1 A).Indeed, a significant increase in microbial complexity was recorded in the posterior dental plaques of dogs affected by CGI when compared to the healthy ones for both species richness (Kruskal-Wallis test Pvalue = .019)and Shannon index analysis (Kruskal-Wallis test Pvalue = .041)(Fig. 1 A).Furthermore, the species richness analysis also sho w ed a significant increment of the microbial complexity in the biofilms collected from the molars of CGI-affected dogs with respect to the dogs with periodontitis (Kruskal-Wallis test P -value = .0015)(Fig. 1 A).T herefore , o verall, α-diversity results indicate that, in gener al, micr obial complexity does not vary among oral ecological microniches nor among different clinical conditions, except for CGI that seems to be associated with a bacterial complexity increment in the dental plaque of posterior teeth in dogs.
In addition, the obtained species-le v el taxonomic profiles were further used to perform a β-div ersity anal ysis, r epr esented through a principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity matrix, to e v aluate the micr obial biodiv ersity of each canine oral microbiota sample (Fig. 1 B and Table S1 ).Furthermore, to assess whether and which factors ma y pla y a role in influencing the bacterial biodiversity of the canine oral micr onic hes, the numer ous collected metadata, including a ge, gender, breed, diet, dental cleaning, type of sample, and clinical status, together with the detected bacterial species, were plotted and integrated into the PCoA using an environmental fitting analysis (Fig. 1 B and Table S1 ).The latter highlighted a statistically significant difference in the taxonomic composition of saliva and dental plaque samples, suggesting that the two different ecological micr onic hes may be c har acterized by a different microbial community (envfit R 2 = 0.224 and Bonferroni post hoc test P -value = .003)(Fig. 1 B and Table S5 ).Specificall y, while saliv a samples strictly correlated with various bacterial species, including Conchiformibius spp ., Haemophilus spp ., Frederiksenia spp ., P asteurella spp ., P orph yromonas cangingivalis , dental plaque samples resulted to be characterized by Actinomyces spp., Corynebacterium canis , Corynebacterium spp., Desulfomicorbium orale , Desulfomicrobium spp., Desulfobulbus spp., Filifactor spp., P orph yromonas gulae , P orph yromonas gingivalis , P orph yromonas spp ., Tessaracoccus spp ., and Treponema spp.(Fig. 1 B and Table S5 ).An observation that was also confirmed by the different prevalence and/or average relative abundance of these bacterial species between saliva and dental plaque samples ( Table S6 ).For each plot, the x -axis reports the different considered groups (H, healthy; CGI, c hr onic g ing ival inflammation; and P, periodontitis), while the y -axis shows the number of species.Boxes are determined by the 25th and 75th percentiles .T he whiskers ar e determined by the maxim um and minim um v alues and corr espond to the box extr eme v alues.Lines inside the bo xes re present the a verage , while crosses correspond to the median.Panel B shows the two-dimensional Bray-Curtis dissimilarity matrix-based PCoA.In the gr a ph, onl y those bacterial species with a significant P -value and an average relative abundance > 0.5% were reported.In addition, only those metadata that significantly explain the biodiversity of samples were displayed through rhombus.
Conv ersel y, the differ ent clinical statuses did not significantl y impact on the bacterial composition (envfit R 2 = 0.0081 and Bonferroni post hoc test P -value = .754)(Fig. 1 B and Table S5 ).Pr obabl y, the presence of CGI or periodontitis does not cause a drastic alteration of the taxonomic composition of the oral microniches, but rather a modulation of only certain specific microbial biomarkers.Similarly, diet, sex, and dental cleaning did not appear to play a role in influencing the taxonomic composition of saliva and dental plaque in dogs (Fig. 1 B and Table S5 ).Ho w e v er, as onl y two dogs were on a diet based on bones and raw food ( Table S1 ), this observation could be biased due to the small number of samples.
The only two factors that seemed to affect the oral microbiota of saliva and dental plaque microniches in dogs were represented by breed and age (Fig. 1 B and Table S5 ).In detail, to assess the impact of age on the biodiversity of the oral microbiota in dogs, the collected samples were divided into three different age groups, i.e. junior (9-24 months), adult (25-96 months), and senior ( > 97 months), as pr e viousl y described (Alessandri et al. 2019 ).These two determinants have been already indicated as main variables influencing the onset of oral pathologies in dogs with a preponderance of cases for small dog breeds and/or with increasing age (Oba et al. 2021b, Wallis et al. 2021a ,b ).Pr obabl y, the physiological and structur al differ ences of the or al cavity among dog breeds may play a role in modulating the taxonomic composition of saliva and dental plaque microbiota leading to a different susceptibility to the onset of oral pathologies based on breed.
Ov er all, these observ ations highlighted that certain bacterial species are typical of either dental plaques or saliva, suggesting clear differences in the bacterial composition of these two oral micr onic hes, r egardless of the clinical status, and that the taxonomic composition of the or al micr obiota of dogs is not influenced by factors other than breed and age.
Furthermore, to identify the potential impact of considered variables on biodiversity, as measured by the Shannon index, a mixed linear model a ppr oac h w as emplo y ed.Specifically, the lme4 mixed linear model (Bates et al. 2015 ), which included age, gender, breed, diet, dental cleaning, and type of sample as fixed variable, along with clinical status as a r andom effect, r e v ealed a statistically significant effect of certain fixed variables on Shannon index biodiv ersity.Specificall y, onl y those inter actions that sho w ed either a t -value > 2 or ← 2 wer e consider ed as statisticall y significant.Based on these cut-offs, among all e v aluated m ultiple-way interactions, only dental cleaning as well as the 2-way interaction involving gender and dental cleaning as variables appeared to have a significant effect on microbial biodiversity ( t -value of −2.516 and 2.729, r espectiv el y).These r esults wer e further inv estigated through a Redundancy Analysis (RDA) based on the species-level taxonomic profiles, follo w ed b y an ANOVA permutation test.In detail, with an ANOVA P -value of .002and an explained variance corresponding to 5.08% for the model incor por ating a ge, gender, breed, diet, dental cleaning, and type of sample independent variables, this analysis suggests that, although the considered variables may partially explain the observed diversity in taxonomic profiles, their impact is very limited.Moreover, an additional RDA analysis including only gender and dental cleaning metadata resulted in a P -value of .378that, exceeding the conventional significance le v el of .05,suggests that the differ ences observ ed in the data could likely be attributed to chance rather than to a systematic effect of the studied independent variables on the taxonomic composition of samples .T her efor e , o v er all, these anal yses indicate that, while an impact on the Shannon biodiversity was observed by the lme4 analysis, their combined impact on the variability of the taxonomic profiles is not significant.
In ad dition, be yond the assessment of αand β-di versity, to further identify possible specific bacterial biomarkers correlated with CGI and/or periodontitis, a taxonomic insight into the saliva and dental plaque microbiota of healthy dogs was first assessed, follo w ed b y the identification of possible bacterial and functional biomarkers for CGI and periodontitis.

Uncovering the salivary microbiota of healthy dogs
To explore the species-level bacterial composition of the saliva microbiota in dogs, the taxonomic profiles of the 30 salivary samples collected from healthy dogs, established through the METAnno-tatorX2 softw are, w ere exploited.In detail, to identify the most r epr esentativ e bacterial species of the canine salivary microbiota, only those bacterial taxa with a pr e v alence > 80% wer e consider ed, as pr e viousl y described (Alessandri et al. 2019(Alessandri et al. , 2020 ) ).Based on this criterion, 20 bacterial species were identified as highly conserved taxa in the saliva microbiota of healthy dogs (Fig. 2 A).Among the latter, the two most abundant species belonged to the genus P orph yromonas , including P. gulae (av er a ge r elativ e abundance of 15.62%) and P. cangingivalis (10.87%) (Fig. 2 A).Notably, members of this bacterial genus are widely reported as predominant commensals of the oral cavity of healthy dogs (Ruparell et al. 2020a, Oba et al. 2021a,b , 2022, Portilho et al. 2023 ).T hus , this data strengthens the ecological r ele v ance of P orph yromonas species in the salivary microbial ecosystem of dogs.In addition, although with an av er a ge r elativ e abundance consider abl y lo w er than the P orph yromonas species, other tw o classified taxa, i.e.Tannerella forsythia (0.43%) and Capnocytophaga cynodegmi (0.86%), were found to be shared by more than 80% of salivary samples from healthy dogs (Fig. 2 A).Inter estingl y, both species ar e gener all y described to be part of the oral cavity microbiota in dogs (Oh et al. 2015, Oba et al. 2021a,b , 2022 ), thus suggesting extensive coevolution between these taxa and the canine oral microbial ecosystem.
Ho w e v er, beyond these classified species, the other most pr e v alent bacterial taxa of the canine salivary samples matched with unclassifiable species belonging to 16 different bacterial genera (Fig. 2 A).Among the latter, the most abundant taxa (av er a ge r elative abundance > 1%) belonged to the genera Moraxella (average r elativ e abundance of 4.49%), P orph yromonas (3.54%), Haemophilus (1.97%), Treponem a (1.87%), Tannerella (1.34%), Actinomyces (1.31%), Corynebacterium (1.30%), and Campylobacter (1.10%).Of note, all these bacterial genera are commonly detected in the oral cavity of healthy dogs (Oba et al. 2021a,b , Alessandri et al. 2022 ), thus strengthening the ecological relevance of these taxa in the saliv ary micr obiota of dogs (Fig. 2 A).The only exception was r epr esented by the genus Haemophilus whose presence has not been detected in pr e vious studies.In this context, considering that the saliv ary micr obiota of dogs seemed to be lar gel y dominated by bacteria not yet identified and that this microbial community is dir ectl y involv ed in influencing the health condition of the host (Oh et al. 2015 , Davis andWeese 2022 ), the application of a culturebased a ppr oac h is necessary to isolate and c har acterize these dominant unclassifiable taxa to better understand their role in the assembly and development of the canine salivary microbiota.

Ev alua tion of the species-level biodiversity of the dental plaque microbiota of healthy dogs
In addition to saliva, the microbial community involved in the establishment and de v elopment of the dental plaque biofilm also plays a crucial role in the onset of pathological conditions, such as CGI or periodontitis (Gr av  Most pr e v alent bacterial species in the saliv ary and dental plaque micr obiota of healthy dogs.P anels displa y the a v er a ge r elativ e abundance of the most pr e v alent species (pr e v alence > 80%) for each saliva (panel A) and dental plaque samples (panel B).The column on the left reports the bacterial species, while each column of the heat map displays the relative abundance of each bacterial species per each sample reported on the top of the gr a ph.Columns on the right of eac h heat ma p show the av er a ge r elativ e abundance (A) and pr e v alence (P) of eac h bacterial species.2021 ).Ho w e v er, to identify possible bacterial biomarkers responsible for oral diseases, it is first necessary to define the species-le v el taxonomic composition of dental plaque in healthy dogs.
Taxonomic composition reconstruction revealed that the most pr e v alent (pr e v alence > 80%) and abundant microbial species (aver a ge r elativ e abundance > 5%) of the canine dental plaques in healthy dogs corresponded to P. gulae (av er a ge r elativ e abundance of 21.40%), P. cangingivalis (7.93%), and P orph yromonas spp.(5.26%) (Fig. 2 B), thus indicating the predominance of this bacterial genus not only in the canine salivary microbiota, but also in the dental plaque microbial ecosystem.An observation that leads to hypothesize that, pr obabl y, both these ecological micr onic hes of the canine oral cavity may act as a reservoir for these species (Fig. 2 ).Ho w e v er, with 33 bacterial species with a pr e v alence > 80%, the dental plaque microbiota displayed a higher number of highly pr e v alent species compar ed to saliv ary samples .T his finding suggests, as alr eady r eported for humans (Ren et al. 2017, Li et al. 2022 ), a lar ger "cor e" micr obial comm unity of the canine dental plaque micr oenvir onment, when compar ed to the saliv ary micr obiota.Specifically, beyond members of the genus P orph yromonas , D. orale as well as members of the genera Actinomyces , Campylobacter , Cardiobacterium , Corynebacterium , Desulfomicrobium , Moraxella , Prevotella , Tannerella , Tessaracoccus , and Treponema resulted to be the most abundant with an av er a ge r elativ e abundance > 1% (Fig. 2 B), denoting the ecological adaptation of these species to the canine oral ca vity en vironment.Furthermore , even if not included in the "core" dental canine plaque microbiota, C. canis and P orph yromonas gingivicanis displayed a high abundance (av er a ge r elativ e abundance of 9.29% and 2.82% and pr e v alence of 78.33% and 73.33%, r espectiv el y) in the dental biofilm of dogs.An observation that suggests the ecological r ele v ance of these two taxa, when present, in the assembly of the canine dental plaque microbiota.
Altogether, taxonomic profiles retrieved from saliva and dental plaque samples from healthy dogs evidenced our curr entl y limited knowledge of the host-associated bacterial species inhabiting these ecological niches despite the close relationship between dogs and humans and their putative implications on human health.In this context, the obtained data underlines the urgent need for r esearc h efforts aimed at isolating and c har acterizing this microbial dark matter.

Loca tion-rela ted differences in the taxonomic composition of dental plaque of healthy dogs
The different pH values as well as oxygen tension and mucosal surface that c har acterize the different ecological microniches of the canine oral cavity are just some of the main intrinsic factors that, together with external determinants such as exposure to food or mechanical insults, affect the local taxonomic composition of eac h or al micr onic he (Rupar ell et al. 2020a,b , Oba et al. 2021a ,b ).Ther efor e, to e v aluate whether also dental plaque localization may have an impact on bacterial biodiversity, taxonomic profiles obtained from dental plaque samples collected from the anterior (canines) and posterior (premolars) teeth of each dog were compared.
Inter estingl y, the nonpar ametric Mann-Whitney U-test for two independent samples corrected for multiple comparisons using the false discov ery r ate (FDR), r e v ealed that thr ee of the taxa pr eviously identified as dominant in the dental plaque microbiota of dogs , i.e .Camp ylobacter spp ., Corynebacterium spp ., and Brachymonas spp.exhibited significantl y differ ent r elativ e abundances between anterior and posterior teeth, suggesting that the location of the dental plaque biofilm plays a crucial role in influenc-ing the r elativ e abundances of some of the major microbial players of the dental plaque microbiota ( Table S7 ).In detail, Campylobacter spp.(av er a ge r elativ e abundance of 1.24% and 2.12% in the canine and premolar teeth, respectively) and Brachymonas spp.(0.20% and 0.68% in the canine and premolar teeth, respectively) displayed a significant increment in the dental plaque from premolars when compared to those from canines ( Table S7 ).On the other side , the a v er a ge r elativ e abundance of Corynebacterium spp.(3.88% and 1.54% in the canine and pr emolar teeth, r espectiv el y), sho w ed a statistically significant increase in the anterior tooth dental plaque compared to that of premolars ( Table S7 ).
In addition, Desulfobulbus spp.and Pasteurella dagmatis displayed not only a higher pr e v alence but also a significant av er a ge r elative abundance increments in the anterior and posterior dental plaques, r espectiv el y ( Table S7 ).
All together, these findings pointed out that biofilm localization, i.e. on anterior or posterior teeth, may have a role in modulating the r elativ e abundances of certain bacterial species that correspond to both "core" and minority bacterial species .T hus , anterior and posterior dental plaques have been considered separ atel y for the subsequent comparison with dogs affected by CGI and periodontitis.

Identification of community state types in the salivary and dental plaque microbiome of healthy dogs
Although the identification of a "cor e" micr obial ecosystem of the v arious micr onic hes of the or al micr obiota allo w ed the identification of the most pr e v alent and abundant bacterial species across a sample cohort, it does not investigate the presence of possible distinct motifs in the ov er all comm unity composition pr ofiles, thereby ignoring the presence of highly abundant and representative bacterial taxa in a subset of samples (Arumugam et al. 2011, Costea et al. 2018, Alessandri et al. 2022 ).In this context, to e v aluate the possible presence of different compositional patterns, also known as community state types (CSTs), among canine salivary and dental plaque samples of healthy dogs, the taxonomic profiles related to the 30 saliva and 60 dental plaque samples were distinctly used for the calculation of the Silhouette index ( Fig. S1 ), i.e. an unsupervised index, to define the ideal number of clusters to optimally subdivide samples according to their bacterial composition.Subsequently, based on this data, a hierarchical clustering analysis was performed to classify samples into the predicted number of clusters ( Fig. S2 ).Ho w e v er, to be considered as real CSTs, each cluster must include at least three samples.
The prediction of common motifs in the ov er all comm unity composition profiles from salivary samples, here referred as saliva community state types (s-CSTs), highlighted the separation of samples into two real s-CSTs.Specifically, a main cluster (s-CST3) containing 17 salivary samples was characterized by a high avera ge r elativ e abundance ( > 10%) of P. gulae and P. cangingivalis, while s-CST5, counting a total of six salivary samples, resulted to be dominated only by P. gulae ( Fig. S2 ).Interestingly, despite the clustering of samples into different groups, the cluster-characterizing taxa corresponded to the most abundant species of the salivary samples of dogs when considered all together, thus confirming the e volutionary ada ptation of these species to this specific micr onic he of the canine oral cavity.
On the other side, the prediction of the canine dental plaque community state types (c-dp-CSTs) allo w ed the identification of thr ee differ ent c-dp-CSTs ( Fig. S2 ).In depth insights into the latter r e v ealed that most of samples fell within c-dp-CST1 (18 samples), which was dominated by P. gulae (av er a ge r elativ e abundance of 26.79%), str engthening the centr al ecological r ole of this taxon in the microbial ecosystem of the canine oral cavity ( Fig. S2 ).Conv ersel y, while c-dp-CST2 was codominated by C. canis and P. gulae, c-dp-CST3 was c har acterized by a predominance of Corynebacterium spp.and Corynebacterium freiburgense ( Fig. S2 ).Of note, while some cluster-dominant species, including P. gulae and Corynebacterium spp., corresponded to "core" anterior dental plaque bacterial taxa (Fig. 2 ), corr obor ating their r ole as commensal micr oorganisms in anterior biofilms, the stratification of anterior dental plaque samples into clusters highlighted the cluster dominance of other two species not pr e viousl y identified as "core" member of the anterior dental plaques , i.e .C. canis and C. freiburgense ( Fig. S2 ), thus suggesting the ecological r ele v ance of these species in the assembly and development of the dental plaque microbiota in certain dogs.Not by chance, the latter taxa have been reported as typical microbial colonizers of the canine oral cavity (Funke et al. 2009, 2010, Santibanez et al. 2021, Thongma et al. 2023 ).
The assessment of the premolar teeth dp-CSTs (p-dp-CSTs) highlighted, instead, the subdivision of samples into four different clusters of which only two represented real CSTs comprising more than three samples ( Fig. S2 ).Specifically, while p-dp-CST1, counting four samples, was dominated by N. dumasiana (av er a ge r elative abundance of 15.97%), the main cluster comprising 22 samples , i.e .p-dp-CST4, sho w ed a predominance of P. gulae (29.62%).Furthermore, despite the reduced number of samples, dominant species were also identified for the other two clusters , i.e .C. canis for cluster 2 and P. cangingivalis together with P. gingivicanis for cluster 3 ( Fig. S2 ).
All together these data sho w ed that, bey ond the clear dominance of species of the genus P orph yromonas both in the saliva and dental plaque microbiota of dogs regardless of location, the species C. canis a ppear ed to be a possible dominant taxon only of the dental plaque biofilms, suggesting the latter species as a main colonizer of the dental plaque ecological micr onic hes.Finall y, the HCL analysis highlighted that also N. dumasiana and C. freiburgense may be considered as dominant bacterial players in certain posterior and anterior dental plaque samples, r espectiv el y, indicating their ecological r ele v ance in the formation of the dental plaque microbiota in dogs.

Prediction of putative CGI and/or periodontitis bacterial biomarkers in the oral microbiota of dogs fr om differ ent microniches
Reconstruction of a detailed ov ervie w of the healthy canine oral microbiota guided us in the assessment of possible microbial biomarkers involved in the onset of CGI and/or periodontitis.For this pur pose, saliv a and dental plaque samples fr om dogs with CGI and periodontitis were compared to those collected from healthy dogs by using the nonparametric Kruskal-Wallis test with the Bonferr oni corr ection.
Remarkabl y, Streptom yces spp.and Haemophilus spp., aboveidentified as "core" bacterial species of the salivary microbiota of dogs (Fig. 2 ), sho w ed a significantly higher abundance in dogs affected by periodontitis when compared to the healthy and CGIaffected ones, r espectiv el y (Fig. 3 A and Table S8 ), suggesting that an increase of this prevalent species may be considered as a biomarker of periodontitis.In contrast, the other 17 bacterial species that significantly differed among the clinical conditions corresponded to minority taxa, pointing at subdominant species as biologicall y r ele v ant mark ers in sali va (Fig. 3 and Table S8 ).In this regard, Mycoplasma edwardii , Fusobacterium russii , and Elizabethkingia spp.sho w ed a significant incr eased av er a ge r elativ e abundance in salivary samples from dogs with periodontitis when compared to the other two groups (Fig. 3 A and Table S8 ), while the av er a ge r elativ e abundances of Mycoplasma spp., Methanosarcina spp ., Clostridioides spp., and Salmonella spp .resulted to be significantly higher in the saliva of dogs with periodontitis with respect to the healthy ones (Fig. 3 A and Table S8 ).Ther efor e, the increased abundance of these species may be considered as potential periodontitis salivary bacterial biomarkers.In this context, while M. edwardii have been depicted as pathogen of the r espir atory tract in dogs (Jambhekar et al. 2019, Alves et al. 2023 ), various species belonging to the genus Elizabethkingia have been recognized as emerging pathogens responsible of various septicemia cases (Bordelo et al. 2016, Zajmi et al. 2022, Weese et al. 2023 ).At the same time, a member of the genus Fusobacterium with a high abundance in the human oral ca vity, i.e .Fusobacterium nucleatum , and in general other re presentati ves of this taxon, have been shown to produce several virulence factors, including adhesin, endotoxins, and serine proteases that not only allow this species to better survive in a hostile environment, but ultimately trigger the host immune response, causing both local and systemic inflammation (Ding andTan 2016 , de Andrade et al. 2019 ).
Ther efor e, it is possible to suggest that other less c har acterized species of the genus Fusobacterium , such as F. russii , may perform similar features in dogs .Furthermore , Muribaculum spp. was not only absent in salivary samples from healthy dogs, but its relative abundance was significantly higher in salivary samples collected from dogs with periodontitis, suggesting that the presence of members of this bacterial genus is closely associated with a periodontitis condition in salivary samples and its presence can be considered as a clear biomarker of this oral disease in dogs (Fig. 3 A and Table S8 ).On the other side, Bifidobacterium spp., Lachnoanaerobaculum spp ., Cellulomonas spp ., Georgenia spp ., Amycolatopsis spp ., and Treponema medium displayed a significantly higher av er a ge r elative abundance in the CGI salivary samples when compared to the healthy ones (Fig. 3 A and Table S8 ), thus electing these six species as microbial biomarker of a CGI condition.Not by chance, T. medium has been described as a micr oor ganism involv ed in the etiology of g ing ivitis and periodontitis in humans, corr obor ating the potential role as a CGI biomarker in dogs when examining the saliv ary micr obiota (Zeng et al. 2021, Oba et al. 2022 ).Ov er all, this data suggests that oral diseases are associated with a dysbiosis of the saliva microbial community corresponding to an increase of the av er a ge r elativ e abundance of certain taxa that are either "core" or "accessory" species in the salivary samples of healthy dogs.
Similarl y to saliv a, significant c hanges in the av er a ge r elativ e abundances of both highly prevalent and minority species were observed also for dental plaque samples among the three considered clinical conditions (Fig. 3 B and C and Table S8 ), thus suggesting that oral diseases are accompanied by a marked alteration of the oral equilibrium ecosystem not only in saliva but also in dental plaques.Specifically, the biofilms collected from the anterior teeth of dogs suffering from periodontitis displayed a significantly higher abundance of Actinomyces oricola and the "core" taxon Fusobacterium spp.coupled with Bacteroides pyogenes and Desulfobulbus oralis when compared to the CGI-affected and healthy ones, r espectiv el y (Fig. 3 B and Table S8 ).In addition, Methanobrevibacter oralis resulted to be significantly more abundant in the anterior dental plaques of dogs with periodontitis with respect to the other two clinical conditions (Fig. 3 B and Table S8 ).Ov er all, these results suggest that a high proportion of these taxa can be considered as marker of a periodontitis status in the anterior dental plaque biofilms.Notably, as abov e r e ported for sali va, mem-Figure 3. Putative periodontal disease microbial biomarkers.Panels A, B, and C display the box and whisker plots of the r elativ e abundances of those bacterial species that significantly differ among the three clinical conditions in saliva, anterior, and posterior dental plaques, respectively.For the posterior dental plaques, only those species with an av er a ge r elativ e abundance > 0.2% in at least one clinical status wer e r eported.The x -axis r eports the bacterial species, while the y -axis displays the r elativ e abundance.Boxes are determined by the 25th and 75th percentiles .T he whiskers are determined by the maximum and minimum values and correspond to the box extreme values.Lines inside the boxes represent the a verage , while the squar e corr esponds to the median.H, healthy; CGI, c hr onic g ing ival inflammation, and P, periodontitis.* , Krusal-Wallis P -value < .05;* * , Kruskal-Wallis P -value < .01;and Kruskal-Wallis P -value < .001.bers of the genus Fusobacterium are widely recognized as potential pathogen of the oral cavity (Goldstein et al. 2017, Crowley et al. 2024, Krieger et al. 2024 ), while members of the genus Actinom yces hav e been positiv el y corr elated with or al diseases in dogs (Oba et al. 2021 ), particularly A. oricola that was isolated for the first time from a human dental abscess (Hall et al. 2003 ).Similarly, the methane producer M. oralis and the sulphate-reducing D. oralis have been strictly associated with severe periodontal disease both in humans and dogs (Cross et al. 2018, Niemiec et al. 2021 ), while B .p yogenes ha v e been fr equentl y described as an animal pathogen (Fernandez Vecilla et al. 2023, Sadhwani et al. 2024 ).In addition to periodontitis biomarker, P orph yromonas macacae and Berge yella spp.w er e significantl y mor e abundant in CGI-affected dogs when compared to those with periodontitis and the two other two clinical conditions, r espectiv el y (Fig. 3 B and Table S8 ), suggesting the high abundance of these two taxa as possible bacterial biomarker of CGI in the anterior dental plaques of dogs.Finall y, Buc hananella spp.was the only bacterial taxon with a significantl y higher r elativ e abundance in healthy dogs when compared to those with periodontitis (Fig. 3 B and Table S8 ), indicating that members of this genus may be considered as potential biomarker of a healthy status.Ho w e v er, since v ery little is known about the genus Buchananella, further studies aimed at isolating and characterizing members of this taxon are necessary to dissect their role in the oral cavity microbiota of healthy dogs and to fully understand their potential biological functions.
Contr aril y fr om the anterior dental plaques in whic h onl y eight species differed among the clinical conditions, the Kruskal-Wallis statistics applied to dental plaque samples from premolars revealed 39 bacterial taxa with a significantly altered average relative abundance in the case of oral diseases (Fig. 3 C and Table S8 ).Specificall y, the r elativ e abundance of 34 of the latter was significantly higher in posterior dental plaque samples collected from dogs suffering from CGI (Fig. 3 C and Table S8 ), suggesting that the investigation of the posterior dental plaque microbiota may be used as an excellent indicator of a CGI status.Among the species with an av er a ge r elativ e abundance > 0.2% in at least one group, Brachymonas spp., Brevilactibacter spp., Buchananella spp., Pseudopropionibacterium spp., and Schaalia spp.wer e significantl y higher in CGI when compared to periodontitis (Fig. 3 C and Table S8 ), while Methanosarcina spp.and Streptomyces spp.sho w ed a significantly higher abundance in dogs with CGI when compared to the healthy ones and both the other two clinical conditions, respectiv el y (Fig. 3 C and Table S8 ).On the other side, the "core" species Cardiobacterium spp.coupled with Neisseria spp., recorded a significant increment in relative abundance in the biofilms of healthy dogs when compared to those with periodontitis (Fig. 3 C and Table S8 ), suggesting that a high proportion of these two taxa may correlate with a healthy oral cavity when examining the posterior dental plaques.Finally, as also observed for anterior dental plaques, also for the posterior biofilms, the abundance of M. oralis resulted to be significantly higher in dogs with periodontitis when compared to the healthy ones (Fig. 3 C and Table S8 ), r einforcing the corr elation between this taxon and a periodontitis status when considering dental plaque samples regardless of their location.
Ov er all, these r esults not onl y show that the thr ee differ ent clinical conditions are characterized by significant alterations in the r elativ e abundance of certain bacterial species, but also that these shifts in taxonomic composition in the evolution from a healthy status to periodontitis seem to be specific for each oral ecological micr onic he.

Prediction of oral microbial metabolic pathway signatures differentiating the dental plaque microbiome in healthy or periodontal conditions in dogs
Availability of shallow shotgun metagenomics data allo w ed also to explore the metabolic potential of the oral microbiota in health and periodontal disease in terms of enzymatic reaction profiles based on the MetaCyc database and the Enzymatic Commission classification.
Inter estingl y, while in saliv a samples the abundance of a total of 172 enzyme-encoding genes significantl y differ ed between healthy and periodontitis-affected dogs, only 41 and 69 enzymes resulted to be differentially present in the oral microbiota betw een the tw o clinical conditions from anterior and posterior dental plaque, r espectiv el y, as e videnced by the a pplication of the Mann-Whitney U-test with FDR correction ( Table S9 ).
Specifically, for all three microniches, among the various enzyme-encoding genes significantly more abundant in the or al micr obiota of healthy dogs when compar ed to that of periodontitis-affected dogs, se v er al genes a ppear ed to be involv ed in the biosynthetic pathways of v arious B gr oup vitamins, including biotin (B8), folate (B9), nicotinamide (B3), pantothenate (B5), riboflavin (B2), and thiamine (B1) (Fig. 4 and Table S9 ).In this context, since B group vitamins are essential micronutrients that participate in a plethora of metabolic, physiological, and regulatory processes comprising immune cell regulation to w ar ds an antiinflammatory condition and suppression of the colonization by pathogenic bacteria in the intestine (Ueland et al. 2017, Peterson et al. 2020, Uebanso et al. 2020 ), it can be suggested that the increased number of genes involved in B vitamin production may play a similar beneficial role also in the oral cavity of dogs.
Furthermore, in depth insight into the enzymatic reaction profiles fr om saliv a samples highlighted that two genes predicted to encode enzymes involved in the monobactam metabolism, a β-lactam antibiotic with a specific spectrum of activity against Gr am-negativ e aer obes, wer e significantl y enric hed in the saliv ary microbiota of healthy dogs when compared to that from dogs affected by periodontitis (Fig. 4 and Table S9 ) (Bush andBradford 2016 , De Angelis et al. 2020 ).In this context, the higher abundance of genes involved in the production of this antibiotic ma y pla y a role in modulating the taxonomic composition of the salivary microbiota to w ar d a healthy microbial ecosystem by targeting certain potential pathogens of the canine oral cavity.In addition, a lipo yl(octano yl) tr ansfer ase and a lipo yl synthase, tw o genes predicted to be involved in the lipoic acid biosynthetic pathway, are significantl y r educed in the saliv a of dogs affected by periodontitis (Fig. 4 and Table S9 ).In this context, since lipoic acid is an antioxidant that acts by regulating mechanisms of inflammation in se v er al c hr onic diseases by exerting anti-inflammatory effects (Poles et al. 2021, Sztolsztener et al. 2022 ), it can be suggested that the increased number of genes involved in lipoic acid production in the genomes of bacteria colonizing the saliva of healthy subjects may also play an anti-inflammatory role in the oral cavity.
The posterior dental plaques of healthy dogs, instead, were c har acterized by a statistically significant increment of the abundance of a gene predicted to encode a nitrous-oxide reductase, which is a central gene of the nitrate-nitrite-nitric oxide pathway, when compared to those affected by periodontitis (Fig. 4 and Table S9 ).Nitrate and nitrate-reducing bacteria have been proposed as potential prebiotics and pr obiotics, r espectiv el y, for human oral and systemic health, since the metabolite produced by nitr ate r eduction, i.e .nitrite , seems to pr e v ent or al disease and impr ov e systemic health (Takahashi 2015, Rosier et al. 2020a ,b ).Ther efor e, the significant increase in the abundance of a gene encoding for nitrous-oxide reductase in the dental plaque microbiome of healthy dogs can be considered as a microbial metabolic marker associated with healthy conditions.(P ac her et al. 2007, Sharma et al. 2007, Alves et al. 2021, Leclerc et al. 2021 ).In parallel, a gene coding a tryptophan synthase was significantly more abundant in the posterior dental plaque biofilm of healthy dogs when compared to those affected by periodontitis (Fig. 4 and Table S9 ).In this context, since tryptophan has been identified as an essential amino acid able to exert anti-inflammatory effects in the intestine, guaranteeing gut homeostasis, regulating intestinal barrier function, and enhancing intestinal barrier integrity (Gao et al. 2018, Scott et al. 2020, Fiore and Murray 2021 ), it can be suggested that the increased number of genes involved in tryptophan production in the genomes of bacteria colonizing the dental plaque of healthy subjects may also play an anti-inflammatory role in the oral cavity.
Ov er all, these data support the notion that periodontitis not only alters the taxonomic composition of the salivary and dental plaque microbiota in dogs, but also the genetic repertoire of this bacterial ecosystem.Indeed, the microbial community of healthy dogs seems to be c har acterized by a higher number of genes that may exert anti-inflammatory function and limit the pr olifer ation of pathogens when compared to the periodontitis-affected ones.Ho w e v er, although the shallow shotgun a ppr oac h pr ovides information on the functional potential of the oral microbiota of dogs allowing the assessment of those genes whose abundance differs between clinical conditions and, ther efor e, the identification of possible functional biomarkers, it does not allow to carry out an accurate and precise gene-bacterial species association, preventing fr om e v aluating whether the genes that significantl y differ ed between clinical conditions belong to the genome of those species above identified as microbial biomarkers (Lugli and Ventura 2022 ).In this context, the application of a deep shotgun approach is necessary to obtain a more accurate insight into the or al micr obiome of dogs allowing a precise and detailed association between functions and bacterial species.

Conclusions
Dental plaque and saliv a micr obial ecosystem ar e involv ed in the onset of oral pathologies, including CGI and periodontitis (Costalonga and Herzberg 2014, Yamashita and Takeshita 2017, Zhang et al. 2018, Bell et al. 2020 ).Specifically, the latter are widespread in dogs and, if not pr operl y tr eated, ma y ha ve a serious impact on canine health (Oba et al. 2021a,b , Wallis et al. 2021a ).In this context, to c har acterize the saliv a and dental plaque micr obiota of healthy dogs at the species-le v el, samples collected fr om 30 dogs wer e anal yzed allowing the identification of the most pr e v alent and abundant bacterial species of the salivary and dental plaque microbiota of healthy dogs .Furthermore , these analyses not only allo w ed to highlight the subdivision of samples into distinct recurrent bacterial compositions, but also underlined that the microbiota of the two micr onic hes ar e c har acterized by a high abundance of bacterial species not yet c har acterized, underling how both bacterial ecosystems ar e ric h r eserv oirs of not y et isolated and c har acterized species .At the same time , taxonomic differences in the microbiota of dental plaques collected from anterior or posterior teeth were observed, suggesting how the different intrinsic and extrinsic factors c har acterizing the micr onic he of the two different dental plaque locations may play a fundamental role in influencing their bacterial composition.Furthermore, by comparing the salivary and dental plaque microbiota of healthy dogs with those from dogs affected by CGI or periodontitis, bacterial and functional biomarkers associated with periodontal diseases were identified.Howe v er, since a large part of the identified microbial biomarkers corresponded to taxa not yet isolated and c har acterized, the a pplication of a cultur e-dependent a ppr oac h aimed at the isolation of these putativ e nov el species could help to pr ovide mor e accur ate information about their involvement in periodontal diseases by scrutinizing their genetic r epertoir e and associated metabolic functions and e v aluate whether the oral pathologies can be exclusiv el y attributable to a higher or lesser abundance of a certain species or, r ather, to str ain-specific genetic signatur e.In addition, a cultur ebased a ppr oac h combined with the whole genome sequencing of new isolated species would help to expand the r efer ence database allowing to obtain more accurate information regarding the composition of the oral microbiota of dogs at the species le v el.
Furthermor e, since a r educed number of samples from dogs with c hr onic g ing ivitis and periodontitis were collected, further investigations with a larger number of samples could provide higher statistical robustness to the identification of microbial and functional biomarkers related to periodontal diseases and, at the same time, would allow the application of machine learning models to effectiv el y pr edict the biomarkers of a specific clinical condition.In addition, the application of deep shotgun sequencing to saliva and dental plaque samples may be useful to ac hie v e mor e accur ate information about the taxonomic composition and functional features of the oral microbiota in healthy and diseased dogs, including the investigation of the presence of antibiotic r esistance genes.Furthermor e, since most or al disease biomarkers correspond to low abundance microbial taxa, a deep shotgun a ppr oac h may help to better understand the functional potential of these species and investigate the presence of potential genetic biomarkers possibly involved in canine oral diseases.Another limitation of the current study is r epr esented by the lack of data on saliva pH.Indeed, since it has been demonstrated that pH plays an important role in affecting the taxonomic composition of saliva (Iacopetti et al. 2017, Damian et al. 2018, Pasha et al. 2018 ), further studies aimed at corr elating saliv a pH and microbial composition ma y pro vide a better c har acterization of this oral microniche.Ho w ever, despite these limitations, the present study provides a detailed species-level characterization of the microbial community of saliva as well as anterior and posterior dental plaques in dogs coupled with the identification of both microbial and functional biomarkers of the two most widespread canine oral diseases.

Figure 1 .
Figure1.Evaluation of the bacterial diversity of saliva and dental plaque samples in dogs.Panel A reports the box and whisker plots of the calculated species richness (on the left) and Shannon index (on the right) based on the number of bacterial species observed in the three sample types (on the top) and by dividing sample types according to the clinical status (on the bottom).For each plot, the x -axis reports the different considered groups (H, healthy; CGI, c hr onic g ing ival inflammation; and P, periodontitis), while the y -axis shows the number of species.Boxes are determined by the 25th and 75th percentiles .T he whiskers ar e determined by the maxim um and minim um v alues and corr espond to the box extr eme v alues.Lines inside the bo xes re present the a verage , while crosses correspond to the median.Panel B shows the two-dimensional Bray-Curtis dissimilarity matrix-based PCoA.In the gr a ph, onl y those bacterial species with a significant P -value and an average relative abundance > 0.5% were reported.In addition, only those metadata that significantly explain the biodiversity of samples were displayed through rhombus.
Figure 2.Most pr e v alent bacterial species in the saliv ary and dental plaque micr obiota of healthy dogs.P anels displa y the a v er a ge r elativ e abundance of the most pr e v alent species (pr e v alence > 80%) for each saliva (panel A) and dental plaque samples (panel B).The column on the left reports the bacterial species, while each column of the heat map displays the relative abundance of each bacterial species per each sample reported on the top of the gr a ph.Columns on the right of eac h heat ma p show the av er a ge r elativ e abundance (A) and pr e v alence (P) of eac h bacterial species.

Figure 4 .
Figure 4. Putative functional biomarkers related to periodontitis.Bar plots indicate the av er a ge r elativ e abundance of eac h enzyme-encoding genes in healthy and periodontitis-affected dogs in saliva (A), anterior (B), and posterior (C) dental plaques.Only enzymes involved in the stim ulation/r egulation of the host immune system were reported.