UC Davis UC Davis Previously Published Works Title Comparative Genomics of the Genus Porphyromonas Identifies Adaptations for Heme Synthesis within the Prevalent Canine Oral Species Porphyromonas cangingivalis. Permalink https://escholarship.org/uc/item/4zx1h3rp Journal Genome biology and evolution, 7(12) ISSN 1759-6653 Authors O'Flynn, Ciaran Deusch, Oliver Darling, Aaron E et al. Publication Date 2015-11-13 DOI 10.1093/gbe/evv220 Peer reviewed eScholarship.org Powered by the California Digital Library University of California GBE Comparative Genomics of the Genus Porphyromonas Identifies Adaptations for Heme Synthesis within the Prevalent Canine Oral Species Porphyromonas cangingivalis Ciaran O’Flynn1,*, Oliver Deusch1, Aaron E. Darling2, Jonathan A. Eisen3,4,5, Corrin Wallis1,IanJ.Davis1,and Stephen J. Harris1 1The WALTHAM Centre for Pet Nutrition, Waltham-on-the-Wolds, United Kingdom 2The ithree Institute, University of Technology Sydney, Ultimo, New South Wales, Australia 3Department of Evolution and Ecology, University of California, Davis 4Department of Medical Microbiology and Immunology, University of California, Davis 5UC Davis Genome Center, University of California, Davis *Corresponding author: E-mail: ciaran.ofl[email protected]. Accepted: November 6, 2015 Abstract Porphyromonads play an important role in human periodontal disease and recently have been shown to be highly prevalent in canine mouths. Porphyromonas cangingivalis is the most prevalent canine oral bacterial species in both plaque from healthy gingiva and plaque from dogs with early periodontitis. The ability of P. cangingivalis to flourish in the different environmental conditions char- acterized by these two states suggests a degree of metabolic flexibility. To characterize the genes responsible for this, the genomes of 32 isolates (including 18 newly sequenced and assembled) from 18 Porphyromonad species from dogs, humans, and other mammals were compared. Phylogenetic trees inferred using core genes largely matched previous findings; however, comparative genomic analysis identified several genes and pathways relating to heme synthesis that were present in P. cangingivalis but not in other Porphyromonads. Porphyromonas cangingivalis has a complete protoporphyrin IX synthesis pathway potentially allowing it to syn- thesize its own heme unlike pathogenic Porphyromonads such as Porphyromonas gingivalis that acquire heme predominantly from blood. Other pathway differences such as the ability to synthesize siroheme and vitamin B12 point to enhanced metabolic flexibility for P. cangingivalis, which may underlie its prevalence in the canine oral cavity. Key words: periodontal disease, comparative genomics, Porphyromonas, canine, plaque. Introduction health and periodontitis. For instance, Porphyromonas gulae The genus Porphyromonas is considered one of the most im- which is closely related to P. gingivalis (~98% sequence iden- portant in human periodontitis and a large amount of re- tity) based on 16S rDNA sequencing data (Fournier et al. 2001) search has been undertaken into the role of Porphyromonas is the second most common Porphyromonad measured by gingivalis in particular (Lamont et al. 1995; Griffen et al. 1998; 16S rDNA sequencing of plaque from dogs with early peri- Holt et al. 1999; Holt and Ebersole 2005). The central role of P. odontitis (Davis et al. 2013). In addition, Porphyromonas can- gingivalis in the progression of human periodontitis has led to gingivalis showed the highest relative abundance of 16S rDNA its classification as a “keystone pathogen” (Darveau et al. sequences of any canine oral species, with an average of 2012) which influences the composition of the oral micro- 10.5% in plaque from healthy gums and 4.6% in plaque biome even when present at low levels. Recent surveys of from periodontitis samples and reached as high as 25% rela- the canine oral microbiota (Dewhirst et al. 2012; Davis et al. tive abundance in both health and disease samples (Davis IJ, 2013) have hinted at a role for Porphyromonads in canine oral unpublished data). The ability of P. cangingivalis to ß The Author(s) 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Genome Biol. Evol. 7(12):3397–3413. doi:10.1093/gbe/evv220 Advance Access publication November 13, 2015 3397 O’Flynn et al. GBE predominate in both health and disease environments sug- Identification of Homologous Clusters gests that it is both metabolically flexible enough to colonize Homologous clusters used in the subsequent analyses of phy- in health and also able to compete against other logeny and gene content were identified by using a Markov Porphyromonas spp. in a disease environment (Davis et al. clustering approach. Predicted protein sequences from RAST 2013). A comparison of the genes present in the were first filtered to remove those shorter than 50 amino acids P. cangingivalis genome compared with those found in in length. A reciprocal Basic Local Alignment Search Tool other Porphyromonads therefore has the potential to identify (BLAST) was performed, BLASTall (v2.2.25) (-b 10000 -v genes that support that metabolic flexibility. The genomes of 10000 -p blastp -e 1e-5) (Altschul et al. 1990). The BLAST various Porphyromonads from dog, human, and other mam- results were filtered to remove spurious hits (percentage iden- mals were compared to look at their relatedness and for tity 25, alignment length 40, and bit score 45), then differences in gene content. The ultimate aim is to better further filtered so that only queries with reciprocal matches understand the role of the Porphyromonads found in were retained. Markov cluster algorithms were applied to canine plaque and generate information to assist the group the sequences, MCL (v12.135) ((mcxload –stream- future development of treatments to reduce periodontitis mirror –stream-neg-log10 -stream-tf ‘ceil(200)’) and (mcl -I in dogs. 1.4)) (Li et al. 2003). The resultant clusters were then con- verted into counts and binary matrices using in house Perl scripts (supplementary tables S1 and S2, Supplementary Material online) and annotated using the consensus annota- Materials and Methods tion within the cluster. Sequences that failed to meet the Genomes Used above criteria or failed to form clusters were excluded from downstream analyses. This study used 32 Porphyromonas genomes including repre- sentatives from all publicly available Porphyromonas species Phylogenetic Analysis (table 1). Of the 32 genomes, 18 were of canine oral isolates previously sequenced and assembled by the authors (Coil et al. To investigate the phylogenetic relationship between 2015). The remaining 14 were complete or partially assembled Porphyromonas species, core sequence and gene content genomes downloaded from National Center for methods were applied. For the supermatrix method, single Biotechnology Information (NCBI) (Sayers et al. 2009). copy core sequences were extracted from the set of homol- Species characteristics (assembly statistics, host species, host ogous clusters. The sequences from each cluster were sites, and disease associations) (table 1)forthecanineisolates aligned using “muscle” (v3.8.31) (default settings) (Edgar were acquired from previous studies (Sayers et al. 2009; Davis 2004) and then concatenated for each strain. The conse- et al. 2013), for noncanine strains from NCBI or individual quent concatenated alignment was filtered for low-quality genome release papers (Hirasawa and Takada 1994; sites using Gblocks (v.0.91b) (maximum contiguous noncon- Summanen et al. 2005, 2009; Sakamoto and Ohkuma served positions 8, minimum block length 10) (Castresana 2013; Sakamoto et al. 2013). In addition to the 2000). The filtered alignments were used to construct a max- Porphyromonas genomes, four further Bacteroidales genomes imum-likelihood (ML) tree using FastTree (v2.1.7) (using a Whelan and Goldman matrix and optimized using were used as outgroups in the phylogenetic analyses. These Gamma20 likelihood for 1,000 bootstraps) (Price et al. were Bacteroides fragilis, Paludibacter propionicigenes, 2010). Tannerella forsythia, and the canine isolate Porphyromonadaceae [G-1] sp. COT-184. Gene Content Clustering The binary presence/absence matrix of homologous clusters computed by MCL was used to calculate the Jaccard dis- Functional Analysis of Genomes tance between each strain, using the “dist.binary” function The genomes for all strains were uploaded to RAST (v2.0) from the R (v3.0.2) package “ade4” (v1.6-2) (Dray and (Aziz et al. 2008; Overbeek et al. 2014) and annotated with Dufour 2007). The resultant distance matrix was used to the ClassicRAST scheme with automatic error fixing and create a hierarchical clustering of the strains. The trees backfilling of gaps. Annotations were then manually cu- from both the supermatrix and gene content methods rated to reduce the impact of partial/misassemblies or erro- were plotted using FigTree (v1.4.1) (unpublished) and an- neous gene annotations; this was done using a combination notated using Inkscape (v0.48.4) (unpublished). The gene of visual inspection, homology searches, and multiple se- content of strains is also represented as a heatmap calcu- quence alignments. The subsequent amino acid sequences