Phylogenomic and molecular demarcation of the core members of the polyphyletic pasteurellaceae genera actinobacillus, haemophilus, and pasteurella.

The genera Actinobacillus, Haemophilus, and Pasteurella exhibit extensive polyphyletic branching in phylogenetic trees and do not represent coherent clusters of species. In this study, we have utilized molecular signatures identified through comparative genomic analyses in conjunction with genome based and multilocus sequence based phylogenetic analyses to clarify the phylogenetic and taxonomic boundary of these genera. We have identified large clusters of Actinobacillus, Haemophilus, and Pasteurella species which represent the "sensu stricto" members of these genera. We have identified 3, 7, and 6 conserved signature indels (CSIs), which are specifically shared by sensu stricto members of Actinobacillus, Haemophilus, and Pasteurella, respectively. We have also identified two different sets of CSIs that are unique characteristics of the pathogen containing genera Aggregatibacter and Mannheimia, respectively. It is now possible to demarcate the genera Actinobacillus sensu stricto, Haemophilus sensu stricto, and Pasteurella sensu stricto on the basis of discrete molecular signatures. The other members of the genera Actinobacillus, Haemophilus, and Pasteurella that do not fall within the "sensu stricto" clades and do not contain these molecular signatures should be reclassified as other genera. The CSIs identified here also provide useful diagnostic targets for the identification of current and novel members of the indicated genera.

1. Introduction

The family Pasteurellaceae, the single constituent family of the order Pasteurellales, represents a diverse group of commensal and pathogenic bacteria within the class Gammaproteobacteria. The family currently contains 19 genera, some of which are particularly important human and animal pathogens [1, 2]. The genera Haemophilus contains species responsible for human bacteremia, pneumonia, acute bacterial meningitis, and the sexually transmitted disease chancroid [3-5]; Aggregatibacter species have been implicated in juvenile periodontitis [6]; members of the genera Mannheimia, Pasteurella, and Actinobacillus have been implicated in the causation of shipping fever in cattle, fowl cholera, and pleuropneumonia in pigs, respectively [7-9].

The family Pasteurellaceae was originally proposed as a higher level taxonomic grouping of the related pathogenic genera Actinobacillus, Haemophilus, and Pasteurella [10]. Classification of organisms into these three genera was primarily based on DNA G-C content and a handful of phenotypic traits [11]. The phenotypic traits were later found not to be characteristic of any single genus [12]. Consequently, the genera Actinobacillus, Haemophilus, and Pasteurella each exhibit extensive polyphyly in subsequent 16S rRNA based phylogenies [12, 13]. Additional studies based on individual or concatenated gene sets and DNA-DNA/rRNA-DNA hybridization also support the presence of extensive polyphyly within the genera Actinobacillus, Haemophilus, and Pasteurella [14-20].

Extensive work has been undertaken to amend the classification of the genera Actinobacillus, Haemophilus, and Pasteurella [1, 5, 9, 18]. New genera have been created to house phylogenetically coherent clusters of Actinobacillus, Haemophilus, and Pasteurella. The species [Actinobacillus] actinomycetemcomitans, [Haemophilus] aphrophilus, [Haemophilus] paraphrophilus, and [Haemophilus] segnis have been transferred to the genus Aggregatibacter [21]; the species [Haemophilus] paragallinarum, [Pasteurella] gallinarum, [Pasteurella] avium, and [Pasteurella] volantium have been transferred to the genus Avibacterium [22]; the species [Haemophilus] somnus and [Haemophilus] agni have been transferred to the genus Histophilus [23]; and the species [Pasteurella] haemolytica and [Pasteurella] granulomatis have been transferred to the genus Mannheimia [7]. Additionally, some individual species within the genera Actinobacillus, Haemophilus, and Pasteurella that do not cluster with other members of their genus in phylogenetic trees have been moved or proposed to be moved to novel or neighbouring genera (namely, the transfer of the species [Haemophilus] pleuropneumoniae to the genus Actinobacillus [24], the transfer of the species [Pasteurella] anatis to the genus Gallibacterium [25], the transfer of the species [Pasteurella] trehalosi to the genus Bibersteinia [26], the transfer of the species [Pasteurella] ureae to the genus Actinobacillus [27], and the proposed transfer of the species [Haemophilus] ducreyi to a novel genus [28]). However, despite these changes, the classification of the genera Actinobacillus, Haemophilus, and Pasteurella is still problematic and each genus continues to contain members which exhibit polyphyletic branching [5, 17–20].

Multiple studies have attempted to define a core group of species which cluster around the nomenclatural type species of Actinobacillus, Haemophilus, or Pasteurella as the only true members of these genera (i.e., sensu stricto) [13, 15–17, 29–31], but the taxonomy and phylogeny of these bacteria continue to remain inconclusive [20, 32, 33]. Several methods have been employed for the demarcation of these genera; however, no simple method or criterion is available that can clearly delimit these genera. It has been suggested that genome based studies may provide reliable means of clarifying the evolutionary relationships of these bacteria [33].

Since the availability of the first complete genome sequence of the Haemophilus influenzae [34], a large number of genomes for the members of the family Pasteurellaceae have become available in public databases [35, 36]. The availability of these genomes provides us with an opportunity to complete comprehensive genome scale phylogenetic analyses of the family Pasteurellaceae. These genome sequences have also been utilized to carry out comparative genomic analyses to identify molecular signatures (namely, conserved signature indels (CSIs) in various proteins), commonly shared by all or closely related subsets of species within the family Pasteurellaceae. On the basis of the molecular signatures identified from comparative analyses of Pasteurellaceae genomes in conjunction with core genome based and multilocus sequence based phylogenetic analyses, we have identified sensu stricto clades of Actinobacillus, Haemophilus, and Pasteurella that are supported by 3, 7, and 6 unique molecular signatures, respectively. We also report sets of molecular signatures that are unique characteristics of the pathogen containing genera Aggregatibacter and Mannheimia.

2. Methods

2.1. Multilocus Sequence Analysis

Multilocus sequence analysis was completed for members of the family Pasteurellaceae using widely available nucleotide sequences of the 16S rDNA, infB (translation initiation factor IF-2), recN (DNA repair protein), and rpoB (DNA-directed RNA polymerase subunit beta) genes which have been used, individually or as part of a set, in a number of previous phylogenetic analyses of the family Pasteurellaceae [15–17, 29, 30]. Gene sequences for these four genes were obtained for 52 Pasteurellaceae strains, representing a large majority of the known Pasteurellaceae species, and 2 members of Vibrio cholerae from the NCBI nucleotide database [37]. Species which were missing one of these four genes or which did not have a gene sequence that was at least 50% of the length of the full gene were excluded from the analysis. The four genes were individually aligned using MUSCLE [38] and manually concatenated to create a combined dataset that contained 10 183 nucleotide long alignments. A maximum-likelihood tree based on 100 bootstrap replicates of this alignment was constructed using MEGA 6.0 [39] while employing maximum composite likelihood substitution model.

2.2. Pasteurellaceae Core Genome Phylogenetic Tree

A phylogenetic tree of 76 Pasteurellaceae strains, rooted using 7 members of the family Vibrionaceae, based on the core genome of the family Pasteurellaceae was created for this study. The core set of Pasteurellaceae proteins were identified using the UCLUST algorithm [55] to identify widely distributed protein families with at least 30% sequence identity and 50% sequence length. Proteins families which were present in less than 50% of the input genomes were excluded from further analysis. Potentially paralogous sequences (additional proteins from the same organism in a single protein family) within the remaining protein families were also excluded from further analysis. Each protein family was individually aligned using MAFFT 7 [56]. Aligned amino acid positions which contained gaps in more than 50% of organisms were excluded from further analysis. The remaining amino acid positions were concatenated to create a combined dataset that contained 128 080 amino acid long alignments. An approximately maximum-likelihood tree based on this alignment was constructed using FastTree 2 [57] while employing the Whelan and Goldman substitution model [58].

2.3. Identification of Molecular Signatures (CSIs) for Different Genera of the Family Pasteurellaceae

The detailed outline of the process of identifying CSIs has been recently published [59]. In brief, Blastp searches were performed on all proteins from the genome of Haemophilus influenzae F3047 [47]. Ten to fifteen high scoring homologues that were present in Haemophilus, other Pasteurellaceae, and Gammaproteobacteria species were retrieved, and their multiple sequence alignments were constructed using Clustal X 1.83 [60]. The alignments were visually inspected to identify any conserved inserts or deletions (indels) that are restricted to the particular clades of the family Pasteurellaceae, which are flanked on each side by at least 5-6 identical/conserved residues in the neighbouring 30–40 amino acids. The selected sequences containing the indels and their flanking conserved regions were further evaluated by detailed Blastp searches to determine species distribution and group specificity. The results of these Blast searches were processed using Sig_Create and Seq_Style to construct signature files [59]. Due to space constraints, the sequence alignment files presented here contain sequence information for a limited number of species within the order Pasteurellaceae and a representative selection of outgroup species. However, in each case, all members of the order and outgroups exhibited similar sequence characteristics to the representatives.

3. Results and Discussion

3.1. Phylogenetic Analysis of the Pasteurellaceae

Elucidating an accurate phylogeny of the members of the family Pasteurellaceae has been a long standing challenge in Pasteurellaceae research [10–12, 18, 19]. Early 16S rRNA based studies revealed that the established taxonomy of the family Pasteurellaceae was not consistent with their genetically inferred phylogeny [12, 14]. This has led to a long series of taxonomic revisions within the family Pasteurellaceae, a process which is still taking place today [7, 18, 22, 28]. However, it was subsequently discovered that phylogenetic trees of Pasteurellaceae species based on different genes did not completely agree with each other [15, 16, 31]. In particular, phylogenetic trees based on the 16S rRNA gene, often considered the gold standard in bacterial taxonomy and phylogeny [61, 62], disagreed with highly robust multilocus sequence and concatenated protein sequence based phylogenetic trees [9, 17, 19, 20, 53, 63].

Phylogenetic trees based on concatenated sequences for a large number of unlinked and conserved loci are more reliable and robust than phylogenetic trees based on any single gene or protein [64, 65]. Due to a rapid increase in the availability of genomic sequence data, we are now able to complete genome scale phylogenetic analyses of the family Pasteurellaceae which cover a vast majority of the diversity within the family. In this work we have produced a phylogenetic tree for 74 genome sequenced members of the family Pasteurellaceae based on 128 080 aligned amino acid positions (Figure 1(a)). The branching patterns of the core genome phylogenetic tree produced in this work largely agree with a previous genome based phylogenetic tree produced for a limited number of Pasteurellaceae species [19] and a concatenated protein based phylogenetic tree of the family Pasteurellaceae produced by our lab in a previous study [20]. Additionally, we have also produced a multilocus sequence based phylogenetic tree using the 16S rDNA, infB, recN, and rpoB genes which are commonly used in the phylogenetic analysis of the family Pasteurellaceae (Figure 1(b)) [15–17, 29, 30]. This tree also showed broadly similar branching patterns to past multilocus sequence based phylogenetic trees [17, 18] and to our core genome based phylogenetic tree. Both our core genome based and multilocus sequence based phylogenetic trees provide evidence for a division of the Pasteurellaceae into at least two higher taxonomic groups (families) which are broadly similar to the two clades of Pasteurellales identified in our previous work [20]. A similar division of the family Pasteurellaceae into two or more large groups is seen in many other robust multilocus or concatenated protein based phylogenetic trees [17, 19, 53, 63]; however, this division is not readily apparent in phylogenies based on the 16S rRNA gene [9, 66].

Figure 1

(a) A maximum-likelihood core genome phylogenetic tree of sequenced members of the family Pasteurellaceae. (b) A maximum-likelihood phylogenetic tree based on concatenated nucleotide sequence alignments of the 16S rDNA, infB, recN, and rpoB genes. Both trees are rooted using members of the Vibrionaceae (not shown). Nodes with >80% bootstrap support are indicated by diamond shaped symbols at the node. Clusters of species representing Actinobacillus sensu stricto, Haemophilus sensu stricto, Pasteurella sensu stricto, Aggregatibacter, and Mannheimia are indicated by brackets. Members of the genera Actinobacillus, Haemophilus, and Pasteurella which do not fall into their respective “sensu stricto” clades are indicated by the presence of square brackets around their generic name (ex. [Pasteurella] pneumotropica).

A majority of the known genera within the family Pasteurellaceae form well-defined and coherent clusters in phylogenetic trees (Figure 1) [9, 17, 19, 20, 66]. The genera Actinobacillus, Haemophilus, and Pasteurella, which were described before the advent of genetic characterization, exhibit polyphyletic branching in all gene and protein based phylogenetic trees, including the core genome based and multilocus sequence based phylogenetic trees created in this work (Figure 1). However, there are large clusters of Actinobacillus, Haemophilus, and Pasteurella species identifiable in the phylogenetic trees which represent the core or “sensu stricto” members of each genera. The clusters of species that represent Actinobacillus sensu stricto, Haemophilus sensu stricto, and Pasteurella sensu stricto are indicated in Figure 1. Members of each genera which fall outside of the sensu stricto clusters, indicated in our phylogenetic trees by the presence of square brackets around their genus name (e.g., [Pasteurella] pneumotropica), are only distantly related to the sensu stricto members of their genus and will require reclassification in order to make their taxonomy and phylogeny concordant.

3.2. The Usefulness of Conserved Signature Indels as Phylogenetic and Taxonomic Markers

Whole genome sequences are a rich resource for the discovery of molecular signatures which are unique to a group of organisms [67-69]. One useful class of shared molecular signatures are conserved signature indels (CSIs), which are insertions/deletions uniquely present in protein sequences from a group of evolutionarily related organisms [59, 70, 71]. The unique, shared presence of multiple CSIs by a group of related species is most parsimoniously explained by the occurrence of the genetic changes that resulted in these CSIs in a common ancestor of the group, followed by vertical transmission of these CSIs to various descendant species [59, 71–73]. Hence, these CSIs represent molecular synapomorphies (markers of common evolutionary decent) which can be used to identify and demarcate specific bacterial groups in molecular terms and for understanding their interrelationships independently of phylogenetic trees [59, 70–72]. CSIs have recently been used to propose important taxonomic changes for a number of bacterial groups (namely, Aquificae, Spirochaetes, Thermotogae, Xanthomonadales, and Borrelia) at different taxonomic ranks [69, 74–77]. In the present work, we have completed comprehensive comparative analysis of Pasteurellaceae genomes (Table 1) in order to identify CSIs that are primarily restricted to the different genera within the family Pasteurellaceae. We have identified 3, 7, and 6 unique molecular signatures which are shared by Actinobacillus sensu stricto, Haemophilus sensu stricto, and Pasteurella sensu stricto, respectively. Information regarding these CSIs and their evolutionary significances is discussed below.

Table 1

Genome characteristics of the sequenced Pasteurellaceae included in our analyses.

Organism name	BioProject	Size (Mb)	Proteins	G-C (%)	References
Actinobacillus pleuropneumoniae L20	CP000569	2.27	2013	41.3	[40]
Actinobacillus pleuropneumoniae serovar 3 str. JL03	CP000687	2.24	2036	41.2	[41]
Actinobacillus pleuropneumoniae serovar7 str. AP76	CP001091	2.35	2142	41.2	STHHb
Actinobacillus ureae ATCC 25976a	AEVG0	2.30	2475	—	BCMg
Actinobacillus minor 202a	ACFT0	2.13	2050	39.3	McGill University
Actinobacillus minor NM305a	ACQL0	2.43	2411	39.3	McGill University
Actinobacillus succinogenes 130Z	CP000746	2.32	2079	44.9	Joint Genome Institute
Aggregatibacter aphrophilus NJ8700	CP001607	2.31	2219	42.2	[19, 42]
Aggregatibacter actinomycetemcomitans D11S-1	CP001733	2.20	2280	44.3	[43]
Aggregatibacter actinomycetemcomitans D7S-1	CP003496	2.31	2250	44.3	[44]
Aggregatibacter segnis ATCC 33393a	AEPS0	1.99	1956	—	BCMg
Gallibacterium anatis UMN179	CP002667	2.69	2500	39.9	[45]
Haemophilus aegyptius ATCC 11116a	AFBC0	1.92	2020	—	BCMg
Haemophilus ducreyi 35000HP	AE017143	1.70	1717	38.2	Ohio State University
Haemophilus haemolyticus M21621a	AFQQ0	2.09	1894	—	[46]
Haemophilus influenzae 10810	FQ312006	1.98	1903	38.1	WTSIh
Haemophilus influenzae F3031	FQ670178	1.99	1770	38.2	[47]
Haemophilus influenzae F3047	FQ670204	2.01	1786	38.2	[47]
Haemophilus influenzae 22.1-21a	AAZD0	1.89	2224	38.0	[48]
Haemophilus influenzae 3655	AAZF0	1.88	1929	38.0	[48]
Haemophilus influenzae 6P18H1a	ABWW0	1.91	1893	38.2	CGS, ASRIe
Haemophilus influenzae 7P49H1a	ABWV0	1.83	1752	37.9	CGS, ASRIe
Haemophilus influenzae NT127a	ACSL0	1.87	1809	38.0	BIGSPc
Haemophilus influenzae PittAAa	AAZG0	1.88	1981	38.1	[48]
Haemophilus influenzae PittIIa	AAZI0	1.95	2028	38.0	[48]
Haemophilus influenzae PittHHa	AAZH0	1.84	1977	38.0	[48]
Haemophilus influenzae R3021a	AAZJ0	1.88	2307	37.9	[48]
Haemophilus influenzae RdAWa	ACSM0	1.80	1718	38.0	BIGSPc
Haemophilus influenzae 86-028NP	CP000057	1.91	1792	38.2	[49]
Haemophilus influenzae PittEE	CP000671	1.81	1613	38.0	[48]
Haemophilus influenzae PittGG	CP000672	1.89	1661	38.0	[48]
Haemophilus influenzae Rd KW20	L42023	1.83	1657	38.2	[34]
Haemophilus influenzae R2846	CP002276	1.82	1636	38.0	UW-SBRId
Haemophilus influenzae R2866	CP002277	1.93	1795	38.1	UW-SBRId
Haemophilus parainfluenzae ATCC 33392 a	AEWU0	2.11	2010	—	BCMg
Haemophilus parainfluenzae T3T1	FQ312002	2.09	1975	39.6	WTSIh
Haemophilus parasuis 29755a	ABKM0	2.22	2244	39.8	Iowa State University
Haemophilus parasuis SH0165	CP001321	2.27	2021	40.0	[50]
Haemophilus pittmaniae HK 85a	AFUV0	2.18	2390	—	J. Craig Venter Institute
Haemophilus sputorum CCUG13788a	AFNK0	2.14	2073	—	Aarhus University Hospital
Haemophilus parahaemolyticus HK385a	AJSW0	1.81	1764	—	J. Craig Venter Institute
Haemophilus paraphrohaemolyticus HK411a	AJMU0	2.02	2025	—	J. Craig Venter Institute
Haemophilus sp. oral taxon 851 str.F0397a	AGRK0	1.84	1809	—	GCG-WUf
Histophilus somni 2336	CP000947	2.26	1980	37.4	Joint Genome Institute
Histophilus somni 129PT	CP000436	2.01	1798	37.2	[51]
Mannheimia succiniciproducens MBEL55E	AE016827	2.31	2370	42.5	[52]
Mannheimia haemolytica PHL213a	AASA0	2.57	2695	41.1	[53]
Pasteurella multocida subsp. multocida str. Pm70	AE004439	2.26	2012	40.4	[54]
Pasteurella dagmatis ATCC 43325a	ACZR0	2.25	2053	37.4	BCMg

aThe genomes of these species/strains are currently under scaffolds/contigs status.

bStiftung Tieraerztliche Hochschule Hannover (STHH).

cThe Broad Institute Genome Sequencing Platform (BIGSP).

dUniversity of Washington; Seattle Biomedical Research Institute (UW-SBRI).

eCenter for Genomic Sciences, Allegheny-Singer Research Institute (CGS, ASRI).

fGenome Sequencing Center (GSC) at Washington University (WashU) School of Medicine.

gBaylor College of Medicine (BCM).

hWellcome Trust Sanger Institute (WTSI).

3.3. Molecular Signatures Specific for Actinobacillus sensu stricto

The genus Actinobacillus was originally defined as a group of growth factor independent host-associated rods which shared phenotypic or biochemical similarity with Actinobacillus lignieresii, the type species of the genus [24, 78]. However, the original classification scheme for the genus Actinobacillus led to the inclusion of a highly heterogeneous and polyphyletic grouping of species within the genus [12-14]. An assemblage of Actinobacillus species closely related to Actinobacillus lignieresii has been recognized as Actinobacillus sensu stricto (i.e., the core members of the genus Actinobacillus) in both our phylogenetic analysis (Figure 1) and past phylogenetic analyses [12–14, 17]. Differentiation of Actinobacillus sensu stricto from other Actinobacillus species and the modern criteria for placing novel species within the genus Actinobacillus sensu stricto is heavily reliant on genetic and genomic criteria, namely, DNA-DNA hybridization values, 16S rRNA sequence similarity, and other single gene sequence comparisons [13, 18]. There are currently no known discrete characteristics which are unique to Actinobacillus that define the genus. In this work, we have completed a comprehensive comparative analysis of Pasteurellaceae genomes in order to identify unique, defining molecular signatures for different genera within the family Pasteurellaceae. We have identified 3 CSIs which are unique, defining molecular signatures for the sequenced members of Actinobacillus sensu stricto (namely, Actinobacillus capsulatus, A. pleuropneumoniae, A. suis, and A. ureae). An example of a CSI specific for Actinobacillus sensu stricto is shown in Figure 2. The CSI consists of a 1-amino-acid insertion in a conserved region of a 3′-nucleotidase which is present in all sequenced members of Actinobacillus sensu stricto and absent in all other sequenced Gammaproteobacteria. Sequence information for 2 other CSIs which are also unique characteristics of the Actinobacillus sensu stricto clade is presented in Supplemental Figures 1-2 available online at http://dx.doi.org/10.1155/2015/198560 and their characteristics are briefly summarized in Table 2(A).

Figure 2

A partial sequence alignment of a 3′-nucleotidase showing a 1-amino-acid insertion identified in all members of Actinobacillus sensu stricto. This insertion was not found in the homologues from any member of the genus Actinobacillus that was not part of the “sensu stricto” clade or any other member of the Gammaproteobacteria. Sequence information for a representative subset of the family Pasteurellaceae and the class Gammaproteobacteria is shown, but unless otherwise indicated, similar CSIs were detected in all members of the indicated group and not detected in any other bacterial species in the top 250 BLAST hits. The dashes (-) in the alignments indicate identity with the residue in the top sequence. GenBank identification (GI) numbers for each sequence are indicated in the second column. Sequence information for other CSIs specific to Actinobacillus sensu stricto are presented in Supplemental Figures  1-2 and their characteristics are summarized in Table 2(A).

Table 2

Conserved signature indels specific for genera within the family Pasteurellaceae.

Protein name	Gene name	GenBank identifier	Figure number	Indel size	Indel positiona
(A) CSIs specific for Actinobacillus sensu stricto
3′-nucleotidase	surE	126208128	Figure 2	1 aa ins	367–402
GTP pyrophosphokinase	relA	126207889	Sup. Figure 1	1 aa ins	368–412
Anaerobic glycerol-3-phosphate dehydrogenase subunit	glpA	491834528	Sup. Figure 2	1 aa ins	359–400
(B) CSIs specific for Haemophilus sensu stricto
Biotin-protein ligase	birA	144979005	Figure 3	6 aa del	138–178
Aspartate ammonia-lyase	aspA	145630289	Sup. Figure 3	1 aa ins	34–75
NAD(P) transhydrogenase subunit alpha	pntA	145631394	Sup. Figure 4	1 aa del	352–378
Fumarate reductase subunit C	frdC	301169552	Sup. Figure 5	3 aa ins	31–89
Hypothetical tRNA/rRNA methyltransferase	—	145636352	Sup. Figure 6	1 aa del	17–58
Gamma-glutamyl kinase	proB	145629980	Sup. Figure 7	1 aa ins	197–253
ACP phosphodiesterase	acpD	68250119	Sup. Figure 8	2 aa del	119–159
(C) CSIs specific for Pasteurella sensu stricto
Menaquinone-specific isochorismate synthase	menF	386834899	Figure 4	4 aa ins	29–86
tRNA s(4)U8 sulfurtransferase	thiI	15602400	Sup. Figure 9	2 aa del	412–446
FKBP-type peptidyl-prolyl cis-trans isomerase	slyD	378775595	Sup. Figure 10	2 aa del	151–188
Aspartate-semialdehyde dehydrogenase	asd	383311492	Sup. Figure 11	1 aa del	173–245
Lactate permease family transporter	lldP	492154065	Sup. Figure 12	2 aa ins	390–427
Cell division protein ftsA	ftsA	492155843	Sup. Figure 13	1 aa ins	357–387
(D) CSIs specific for Aggregatibacter
nhaC family sodium:proton antiporter	nhaC	493769836	Figure 5(a)	3 aa ins	396–437
Outer membrane protein	omp	261866907	Sup. Figure 14	4 aa del	25–64
Multidrug transporter murJ	murJ	365966332	Sup. Figure 15	1 aa del	190–220
NADH dehydrogenase	nuoE	387120244	Sup. Figure 16	1 aa ins	372–412
(E) CSIs specific for Mannheimia
Methyl-galactoside ABC transporter substrate-binding protein	—	472335016	Figure 5(b)	1 aa del	33–73
UDP-N-acetylmuramoylalanyl-D-glutamate–2,6-diaminopimelate ligase	murE	472333011	Sup. Figure 17	2 aa del	418–473
Glutathione-regulated potassium-efflux protein	kefC	472333189	Sup. Figure 18	1 aa ins	504–531
Glycerol-3-phosphate acyltransferase	plsB	472334521	Sup. Figure 19	2 aa del	214–252

3.4. Molecular Signatures Specific for Haemophilus sensu stricto

The classification of novel species into the genus Haemophilus was initially based on phenotypic and biochemical properties, most importantly, the dependence of growth on the presence of factor V and factor X in blood [13, 78, 79]. As with Actinobacillus, the classification of Haemophilus on the basis of phenotypic and biochemical properties has led to the genus containing an extremely heterogeneous group of species [12–14, 32]. Species from the genus Haemophilus have undergone a number of transfers and reclassifications [21–24, 28]. However, the genus remains highly polyphyletic (Figure 1) [17, 19, 28]. The core members of the genus Haemophilus (namely, Haemophilus sensu stricto) consist of Haemophilus influenzae, H. aegyptius, and H. haemolyticus based on 16S rRNA sequence analysis [12–14, 32]. However, phylogenetic analysis based on DNA-DNA hybridization and multilocus sequence analysis suggests that H. parainfluenzae and H. pittmaniae are also members of Haemophilus sensu stricto [30, 80]. Phylogenetic analysis of rpoB, infB, and concatenated gene sets also suggest that [Pasteurella] pneumotropica and related isolates are closely related to Haemophilus sensu stricto [15, 16].

Our comparative analysis of Pasteurellaceae genomes has led to the identification of 7 CSIs that are unique characteristics of Haemophilus sensu stricto which consists of Haemophilus influenzae, H. aegyptius, H. haemolyticus, H. parainfluenzae, H. pittmaniae, and [Pasteurella] pneumotropica (Figure 1). One example of a CSI specific for the members of Haemophilus sensu stricto, shown in Figure 3, consists of a 4-amino-acid deletion in a biotin-protein ligase which is uniquely found in homologs from Haemophilus sensu stricto and absent in all other sequenced Gammaproteobacteria. Sequence information for 6 additional CSIs which are also unique characteristics of Haemophilus sensu stricto is presented in Supplemental Figures  3–8 and their characteristics are briefly summarized in Table 2(B). These CSIs and our phylogenetic trees (Figure 1) suggest that Haemophilus influenzae, H. aegyptius, H. haemolyticus, H. parainfluenzae, H. pittmaniae, and [Pasteurella] pneumotropica share a close evolutionary relationship and should all be considered members of Haemophilus sensu stricto. Additionally, these results also suggest that [Pasteurella] pneumotropica is incorrectly classified as a member of the genus Pasteurella and should be reclassified as “Haemophilus pneumotropica.”

Figure 3

A partial sequence alignment of 1,4-dihydroxy-2-naphthoate octaprenyltransferase showing a 2-amino-acid insertion identified in all members of Haemophilus sensu stricto. This insertion was not found in the homologues from any member of the genus Haemophilus that was not part of the “sensu stricto” clade or any other member of the Gammaproteobacteria. Sequence information for other CSIs specific to Haemophilus sensu stricto is presented in Supplemental Figures  3–8 and their characteristics are summarized in Table 2(B).

3.5. Molecular Signatures Specific for Pasteurella sensu stricto

The genus Pasteurella is highly heterogeneous and polyphyletic (Figure 1) [13]. Similar to the members of Actinobacillus, bacterial isolates were originally classified as members of the genus Pasteurella based on growth factor independent growth and phenotypic or biochemical similarity to Pasteurella multocida, the type species of the genus [78, 81]. The monophyletic clusters of Pasteurella species that branch with Pasteurella multocida are considered the core members of the genus (namely, Pasteurella sensu stricto) [9, 13, 16, 17]. Our comparative analysis of Pasteurellaceae genomes has led to the identification of 6 CSIs which are unique characteristics for the sequenced members of Pasteurella sensu stricto (namely, Pasteurella multocida and P. dagmatis). An example of a CSI uniquely found in the sequenced members of Pasteurella sensu stricto, consisting of a 4-amino-acid insertion in a conserved region of Menaquinone-specific isochorismate synthase, is shown in Figure 4. This CSI is only found in the sequenced members of Pasteurella sensu stricto and is absent from all other sequenced Gammaproteobacteria. Partial sequence alignments for 5 additional CSIs which are also unique characteristics of Pasteurella sensu stricto are presented in Supplemental Figures 9–13 and their characteristics are briefly summarized in Table 2(C).

Figure 4

A partial sequence alignment of Menaquinone-specific isochorismate synthase showing a 4-amino-acid insertion identified in all members of Pasteurella sensu stricto. This insertion was not found in the homologues from any member of the genus Pasteurella that was not part of the “sensu stricto” clade or any other member of the Gammaproteobacteria. Sequence information for other CSIs specific to Pasteurella sensu stricto is presented in Supplemental Figures  9–13 and their characteristics are summarized in Table 2(C).

3.6. Molecular Signatures Specific for the Genera Aggregatibacter or Mannheimia

The genus Aggregatibacter was proposed as a novel taxonomic classification for a monophyletic cluster of Actinobacillus and Haemophilus species which branched distinctly from the “sensu stricto” members of their respective clades [21]. Similarly, the genus Mannheimia was proposed as a novel classification for the Pasteurella Haemolytica complex which did not branch with Pasteurella sensu stricto in phylogenetic trees [7]. Currently other than branching in phylogenetic trees or relatedness in DNA-DNA hybridization studies, the members of the genera Aggregatibacter or Mannheimia do not share any single unique or defining biochemical or molecular characteristic that can differentiate them from all other bacteria [5, 82].

In this study we have identified 4 CSIs that are unique molecular characteristics shared by all sequenced species of the genus Aggregatibacter and another 4 CSIs which are uniquely found in all sequenced members of the genus Mannheimia. Examples of CSIs specific to the sequenced members of Aggregatibacter and Mannheimia are shown in Figure 5. A partial sequence alignment of a nhaC family sodium:proton antiporter containing a 3-amino-acid insertion specific for all sequenced species of the genus Aggregatibacter is shown in Figure 5(a) and a partial sequence alignment of a methyl-galactoside ABC transporter substrate-binding protein containing a 1-amino-acid deletion specific for all sequenced species of the genus Mannheimia is shown in Figure 5(b). In each case, the identified CSIs were only found in the sequenced members of the genera Aggregatibacter or Mannheimia and were absent from all other sequenced Gammaproteobacteria. Partial sequence alignments additional CSIs specific for the genera Aggregatibacter or Mannheimia are provided in Supplemental Figures 14–19 and their characteristics are summarized in Tables 2(D)-2(E). These CSIs are the first discrete molecular characteristics which are unique for the genera Aggregatibacter and Mannheimia and support their observed monophyly in phylogenetic trees. Additionally, these CSIs could be useful targets for the development of PCR based diagnostic assays for the genera Aggregatibacter and Mannheimia which amplify the CSI containing DNA segment using the conserved flanking regions of the CSIs [83, 84].

Figure 5

A partial sequence alignment of (a) a nhaC family sodium:proton antiporter containing a 3-amino-acid insertion specific for all sequenced species of the genus Aggregatibacter (b) a methyl-galactoside ABC transporter substrate-binding protein containing a 1-amino-acid deletion specific for all sequenced species of the genus Mannheimia. In each case, the identified CSIs were only found in the sequenced members of the genera Aggregatibacter or Mannheimia and were absent from all other sequenced Gammaproteobacteria. Sequence information for other CSIs specific to Aggregatibacter or Mannheimia is presented in Supplemental Figures  14–19 and their characteristics are summarized in Tables 2(D) and 2(E).

4. Conclusion

The genera Actinobacillus, Haemophilus, and Pasteurella, within the family Pasteurellaceae, are known to exhibit extensive polyphyletic branching. We have utilized molecular signatures and phylogenetic analyses to clarify the taxonomic boundary of these genera. We have been able to identify large clusters of Actinobacillus, Haemophilus, and Pasteurella species which represent the “sensu stricto” members of these genera. We have identified 3, 7, and 6 unique molecular signatures which are specifically shared by the members of the genera Actinobacillus sensu stricto, Haemophilus sensu stricto, and Pasteurella sensu stricto, respectively. The group specificity of the molecular signatures we have identified in this work is summarized in Figure 6 and their characteristics are briefly summarized in Table 2. Our comparative genomic analyses have not come across any CSIs that were unique characteristics of all sequenced members of the genera Actinobacillus, Haemophilus, or Pasteurella as currently defined, suggesting that the members of these genera that do not fall into the “sensu stricto” clusters should not be considered members of their respective genus.

Figure 6

A summary diagram depicting the distribution of identified CSIs for genera within the family Pasteurellaceae.

Examinations of phenotypic and biochemical characteristics do not provide a reliable means of assigning a novel isolate to the genera Actinobacillus, Haemophilus, and Pasteurella [18]. However, based upon the CSIs described in this work, it is now possible to demarcate the genera Actinobacillus sensu stricto, Haemophilus sensu stricto, and Pasteurella sensu stricto on the basis of the presence or absence of unique molecular signatures. It is important to note that the current analysis of CSIs is limited to the currently available genomic sequence data and may show slight variance as additional bacterial genomes are sequenced. However, earlier work on CSIs for other groups of bacteria provides evidence that the identified CSIs have strong predictive value and will likely be found in other members of these groups as more species are sequenced and novel species are isolated [74, 77, 85, 86]. The conserved nature of the sequence regions that contain these CSIs, in conjunction with their strong predictive value, makes CSIs promising targets for the development of highly specific diagnostic assays for Actinobacillus sensu stricto, Haemophilus sensu stricto, Pasteurella sensu stricto, Aggregatibacter, and Mannheimia [83, 84]. Additionally, further analysis of these genus specific CSIs should lead to the discovery of their functional role in their respective organisms and may provide important insights into novel distinguishing features of these groups of organisms.

Partial sequence alignments of the CSI containing regions that were unique characteristics of the genera Actinobacillus sensu stricto, Haemophilus sensu stricto, Pasteurella sensu stricto, Aggregatibacter, or Mannheimia.