Minimal standards for describing new species belonging to the families Campylobacteraceae and Helicobacteraceae: Campylobacter, Arcobacter,

Ongoing changes in taxonomic methods, and in the rapid development of the taxonomic structure of species assigned to the Epsilonproteobacteria have lead the International Committee of Systematic Bacteriology Subcommittee on the Taxonomy of Campylobacter and Related Bacteria to discuss significant updates to previous minimal standards for describing new species of Campylobacteraceae and Helicobacteraceae. This paper is the result of these discussions and proposes minimum requirements for the description of new species belonging to the families Campylobacteraceae and Helicobacteraceae, thus including species in Campylobacter, Arcobacter, Helicobacter, and Wolinella. The core underlying principle remains the use of appropriate phenotypic and genotypic methods to characterise strains sufficiently so as to effectively and unambiguously determine their taxonomic position in these families, and provide adequate means by which the new taxon can be distinguished from extant species and subspecies. This polyphasic taxonomic approach demands the use of appropriate reference data for comparison to ensure the novelty of proposed new taxa, and the recommended study of at least five strains to enable species diversity to be assessed. Methodological approaches for phenotypic and genotypic (including whole-genome sequence comparisons) characterisation are recommended. INTRODUCTION The class Epsilonproteobacteria is a phylogenetically-distinct lineage within the Proteobacteria [1] and currently contains 16 genera (Tables S1 and S2, available with the online version of this article). The genus Campylobacter was the first of these established [2], and the realisation that certain species were important human and animal pathogens prompted many further studies investigating the wider ecology and distribution of similar organisms. Bacteria found in cases of human gastritis and initially classified as Campylobacter spp. were later reclassified into a separate but related genus, Helicobacter [3]. Improvements in isolation, detection and taxonomic characterizationmethods, together with continued interest in the significance and distribution of such bacteria have resulted in the present status of the Epsilonproteobacteria – a highly diverse group of organisms containing over 100 taxa. Within the class, phylogenetic subgroups can be identified. Of these, most taxa in the families Campylobacteraceae (namely Campylobacter, Arcobacter and Sulfurospirillum) and Helicobacteraceae (namely Helicobacter and Wolinella) appear more closely related to each other than to free-living Epsilonproteobacteria such as the Nautiliaceae [4]. Their known or potential significance as pathogens, and advances in cultivation, make Campylobacter and Helicobacter the most populous genera, with many taxa sharing the same ecological niche. Their close phenotypic similarity to each other is illustrated by many Arcobacter, Sulfurospirillum and Helicobacter spp. having originally been described as ‘Campylobacter’ species (reviewed in [5]). In every year since 1988, at least one novel species belonging to at least one of these genera has been described. On occasion, descriptions have been controversial [6–9], and on several occasions the ICSP Subcommittee on the Taxonomy of Campylobacter and Related Bacteria have remarked on the sometimes questionable quality of particular species descriptions that were observed [10–12]. Furthermore, taxonomic methods have undergone a ‘sea change’ with wholeAuthor affiliations: Department of Wine, Food and Molecular Biosciences, Lincoln University, PO Box 85084, Lincoln, New Zealand; U.S. Department of Agriculture, Produce Safety and Microbiology Research Unit, Agricultural Research Service, Albany, CA, USA; Department of Veterinary Public Health, Faculty of Sciences, Ghent University, K. L. Ledeganckstraat 35, B-9000 Ghent, Belgium; Department of Biochemistry and Microbiology, Laboratory of Microbiology, Faculty of Sciences, Ghent University, K. L. Ledeganckstraat 35, B-9000 Ghent, Belgium; Department of Comparative Medicine, Massachusetts Institute of Technology, 77, Massachusetts Avenue, Cambiridge, MA 02139, USA. *Correspondence: Stephen L. W. On, stephen.on@lincoln.ac.nz


INTRODUCTION
The class Epsilonproteobacteria is a phylogenetically-distinct lineage within the Proteobacteria [1] and currently contains 16 genera (Tables S1 and S2, available with the online version of this article).The genus Campylobacter was the first of these established [2], and the realisation that certain species were important human and animal pathogens prompted many further studies investigating the wider ecology and distribution of similar organisms.Bacteria found in cases of human gastritis and initially classified as Campylobacter spp.were later reclassified into a separate but related genus, Helicobacter [3].Improvements in isolation, detection and taxonomic characterization methods, together with continued interest in the significance and distribution of such bacteria have resulted in the present status of the Epsilonproteobacteriaa highly diverse group of organisms containing over 100 taxa.
Within the class, phylogenetic subgroups can be identified.Of these, most taxa in the families Campylobacteraceae (namely Campylobacter, Arcobacter and Sulfurospirillum) and Helicobacteraceae (namely Helicobacter and Wolinella) appear more closely related to each other than to free-living Epsilonproteobacteria such as the Nautiliaceae [4].Their known or potential significance as pathogens, and advances in cultivation, make Campylobacter and Helicobacter the most populous genera, with many taxa sharing the same ecological niche.Their close phenotypic similarity to each other is illustrated by many Arcobacter, Sulfurospirillum and Helicobacter spp.having originally been described as 'Campylobacter' species (reviewed in [5]).
In every year since 1988, at least one novel species belonging to at least one of these genera has been described.On occasion, descriptions have been controversial [6][7][8][9], and on several occasions the ICSP Subcommittee on the Taxonomy of Campylobacter and Related Bacteria have remarked on the sometimes questionable quality of particular species descriptions that were observed [10][11][12].Furthermore, taxonomic methods have undergone a 'sea change' with whole- This paper accounts for these substantive changes and expands and updates previously-published minimal standards for describing new species of Campylobacteraceae [13] and Helicobacter [14].Table S1 lists the current, validlydescribed species names in the genera covered by this proposal with their type strains and accompanying 16S rRNA gene sequence accession number.Members of the genus Sulfurospirillum are exclusively free-living and, from an ecological perspective, have more in common with, for example, members of the Nautilaceae (cf.Table S2), for which minimal standards will be proposed in due course.Intriguingly, phylogenetic analyses based on both 16S rRNA gene sequences (Fig. 1), and concatenated protein sequences derived from a subset of 289 conserved single-copy-number genes reveal a close relationship between Sulfurospirillum and Campylobacter [15], despite these taxa inhabiting radically different ecological niches.Such observations highlight the complexities of the Epsilonproteobacteria group as a whole, and support the need for a polyphasic taxonomic approach for accurate classification.

GENERAL FEATURES OF CAMPYLOBACTERACEAE
Gram-negative, curved, spiral or occasionally straight rodshaped cells, 0.2-5 µm long.Cells may become more spherical after time or with exposure to environmental stress.Non-spore forming.Most species are motile by use of one to two unsheathed polar flagella, but a few species are aflagellate.Optimum growth temperatures range from 25 C (notably Arcobacter spp.) to 42 C (some Campylobacter spp.).All Campylobacter spp.can be cultured at 37 C under appropriate conditions.The optimal growth atmosphere for Campylobacteraceae is microaerobictraditionally, such atmospheres are described as comprising 3-8 % O 2 , with commonly available Gaspak systems providing the balance with 10 % CO 2 and the remainder as N 2 .However, some species require H 2 for growth and studies have demonstrated all Campylobacter spp.can grow in an atmosphere of 3 % O 2 , 7 % H 2 , 10 % CO 2 and 80% N 2 .Most Arcobacter spp.demonstrate the ability to grow under aerobic conditions (A. anaerophilus is an exception as an obligate anaerobe) [16], but all arcobacters to date can grow at 20-30 C. Most species produce oxidase (undetected in C. gracilis and in some strains of C. showae [17]) and, in conventional laboratory testing for oxidation and/or fermentation of glucose, (e.g.[18]), do not demonstrate the ability to ferment or oxidise carbohydrates.Campylobacter and some Arcobacter spp.may be associated with the oral environment and/or the enteric or reproductive tracts of host animals; some Arcobacter spp.are free-living [4,5,16].

GENERAL FEATURES OF HELICOBACTERACEAE
Gram-negative, spiral, helical, curved, or fusiform rods of width 0.3-0.6 µm and length 1-5 µm.Cells may become more spherical after time or with exposure to environmental stress.Non-spore forming.Most species are motile by means of single or multiple flagella.In most species the flagella are sheathed.Optimum temperatures of growth range from 37 to 42 C and all known species can be cultured at 37 C.They are usually microaerophilic and best cultured in atmospheres containing H 2 .Oxidase producing with most strains producing catalase.Gastric Helicobacter spp.described to date predominantly produce copious amounts of urease.Using common laboratory methods (e.g.[18]), Helicobacteraceae do not demonstrate the ability to ferment or oxidise carbohydrates.Helicobacter and Wolinella species described thus far have been associated with gastric, enteric, reproductive and/or hepatic environments of a variety of host animal species [1,3,7,14,19,20]; none are known to be free-living.

General comments
The description of new species belonging to Campylobacteraceae or Helicobacteraceae should be based on characteristics necessary for assigning the new taxon to the genus, and on characteristics serving to differentiate the new taxon from existing taxa of the genus.In practice, this will require a polyphasic taxonomic approach utilizing both genotypic and phenotypic methods, especially since there is no single phenotypic characteristic that readily enables the assignation of strains to the genera Campylobacter, Arcobacter, Helicobacter or Wolinella.Furthermore, species such as C. coli and H. pullorum may be found in the same environments (presently chickens and humans) and share many phenotypic characters, rendering them difficult to discriminate using conventional phenotyping [5].Thus, a phylogenetic assignation based on, at least, comparative analysis of 16S rRNA gene sequences is mandatory to appropriately assign strains to genus level [5,21].Corresponding sequences from type strains of relevant, validly-named Epsilonproteobacterial taxa should be included in sequence comparisons to appropriately designate phylogeny and thus genus placement.Additional gene sequence comparisons, such as atpA [22], rpoB [23] or groEL (Hsp60) [24], may provide useful information for strain classification and a finer resolution of taxonomic identity.
The description should ideally be based on not fewer than five isolates from different sources, or five distinct genotypes (i.e.representing distinguishable, individual strains) from the same or similar sources.The most useful taxonomic descriptions involve studies where species heterogeneity can be adequately assessed.
For critical comparisons with other species, controls consisting of type or reference strains of the appropriate taxa must be tested.For all phenotypic test procedures, the inoculum size, composition of the gaseous atmosphere, temperature and period of incubation, and composition of the basal growth medium should be stated.The use of standardized, well-described tests and methods is recommended to facilitate comparison [25][26][27][28][29].For descriptions at the species level, the use of advanced phenotypic methods, such as and including MALDI-TOF MS analysis, should be regarded as supportive data but not the primary means by which species are delineated at the phenotypic level.Present  studies, although limited to relatively few taxa, indicate this approach has promise [30][31][32][33] with some caveats [34].
Putative new species of uncultured organisms for which molecular sequence data (such as 16S rRNA sequence) is available may qualify for assignment to the provisional taxonomic status Candidatus [35] in accordance with the proposals of Murray and Stackebrandt [36], as exemplified by the initial description of C. hominis prior to its culture [37].

Cell morphology
The reaction of cells in the Gram-staining procedure must be stated.The shape, size, and spiral wavelength (where appropriate) of bacterial cells should be reported.The tendency to undergo transformation to coccoid forms on exposure to air or in older cultures should be noted and the time taken for cells to change their appearance provided.The number and arrangement of flagella should be determined by electron microscopy, as well as the presence or absence of flagellar sheaths and periplasmic fibers.

Motility
Cells should be observed by microscopic examination of wet mounts or hanging drop preparations of young cultures in buffered saline or broth.

Growth conditions
Factors affecting growth should be tested under conditions that are near optimal unless stated otherwise.
(1) Colony morphology.The size, shape, and colour of colonies should be described for optimal growth conditions on solid media.The type of nutrient agar base medium used (brain-heart infusion [BHI], Mueller-Hinton etc.) and the manufacturer from which it was obtained should be stated.The presence of swarming on solid media should be noted.When cultured on blood-containing agar media, the percentage and species of blood (e.g.horse, sheep, cattle) should be stated and any haemolytic activity described.(2) Temperature range.The time of incubation and ability to grow in specified broth or agar media from standardized inocula at various temperatures should be reported.The following temperatures should be used: 25, 37 and 42 C. (3) Gaseous requirements.The ability of the strains to grow under aerobic, microaerobic and anaerobic conditions should be reported.The oxygen and hydrogen content must be specified for microaerobic conditions.The means by which such conditions were produced (e.g., gas replacement method, commercial gas-generating sachet [state manufacturer], incubator) must be given.

Biochemical properties
Results for the following tests are required: (i) oxidase activity, by use of any conventional method [28]; (ii) catalase activity, with percentage of reagent solution and time of observation given; (iii) nitrate reduction, preferably by the plate method of Cook [38]; (iv) indoxyl acetate hydrolysis, preferably using a disc method [28] with percentage of reagent solution, volume of impregnation and time of observation given; (v) urease activity, using a rapid method [39]; (vi) alkaline phosphatase activity [40], with time of observation given; (vii) hippurate hydrolysis [41]; and (viii) selenite reduction [28].Growth on media containing the following compounds using standardized methods [26,27] should be determined: 2.0 % NaCl, 1 % glycine and 0.04 % triphenyl tetrazolium chloride (TTC): the ability of strains to reduce the latter should also be recorded when growth is observed.
Resistance to antimicrobial agents Susceptibility to nalidixic acid (30 µg) and cephalothin (30 µg), should be determined either by disc-diffusion or plate MIC tests.For diffusion assays, the absence of a clear zone of inhibition should be recorded as resistance; for susceptible strains, the inhibition zone sizes should be stated.The type of base medium used should be stated.Mueller-Hinton agar with added 10 % horseor sheepblood is recommended.Standardised procedures should be employed [26,27,42].Brucella Agar and other media with bisulfite have been shown to inhibit the growth of H. pylori and therefore should be avoided [43].

Phylogenetic analyses
The essentially complete (greater than 1450 bases) 16S rRNA sequence must be determined for the type strain, and ideally at least four additional independent isolates of the putative new species or subspecies.Specific methods for sequencing 16S rRNA from Helicobacter species, including intervening sequences (IVS), have been described [44].
Intervening sequences in the 16S rRNA gene should be fully sequenced and the 16S rRNA gene sequence, including any IVS, deposited in nucleic acid databases as a single sequence.Deposition of a complete genome sequence accomplishes this goal de facto.The phylogenetic position of representative strains of the putative new taxon must be determined by comparative sequence analysis of the 16S rRNA macromolecule [21].Phylogenetic tree construction should demonstrate that the novel sequence clusters with those of all validly-named taxa of the appropriate genus within the Campylobacteraceae or Helicobacteraceae.A full description of alignment and treeing methods, including software, algorithms, treatment of gap penalties, and treatment of IVS sequences should be included.An exemplar tree is presented in Fig. 1.The 16S rRNA gene is the most widely used phylogenetic marker but there are known problems with the accurate phylogenetic placement of certain species within the Campylobacteraceae or Helicobacteraceae [7,[45][46][47].The use of additional phylogenetic markers such as atpA [22], rpoB [23] or groEL [24], may provide useful data to support the phylogenetic position of the new taxon.Multi-Locus Sequence Analysis [48] or rMLST [49] may provide a more robust description of the taxonomic position of a given strain.Taxonomic studies of Campylobacteraceae or Helicobacteraceae species using MLSA and rMLST are as yet uncommon, although MLSA has been applied to Arcobacter species [32,50].Ultimately, these approaches can be expected to be replaced by whole genome sequence based phylogenomic analyses (e.g.[15,51,52]).As with 16S rRNA gene sequence comparisons, representatives of all extant validly-published named taxa in the genus in which the proposed novel organism is placed must be included.

Genomic analyses
A complete genome sequence of the proposed type strain of a novel species or subspecies is a definitive means to determine the guanosine-plus-cytosine (% G+C) content of the DNA; however, classical methods (melting temperature or enzymatic) may also be used [53].Reference DNA such as Escherichia coli ATCC 11775 T (G+C, 51 mol%) or H. pylori ATCC 700392 (G+C, 39 mol%) should be analysed at the same time and its estimated G+C content (moles percent) expressed relative to the reference DNA should be reported.
The current taxonomic definition of a species requires determination of the whole-genomic similarity, whereby at or around 70 % DNA-DNA relatedness indicates strains are sufficiently related to be assigned to the same species [54][55][56].Methodological caveats notwithstanding (discussed below), novel species should demonstrate genomic relatedness to extant species at values discernibly lower than this level.The proposed type strain for any new taxon must be used in these comparisons.
Whole-genome relatedness can be tested in several ways.Classical DNA-DNA hybridization experiments have been used for many years (e.g.[57][58][59][60]).It is recognized that experiments on Epsilonproteobacteria can be difficult and that different methods can result in differing estimates of the degree of relatedness between strains [61,62].Numerical comparison of whole-cell protein profiles [5,6,60], and high-resolution amplified fragment length polymorphism (AFLP)-based fingerprints has also been shown to be effective in accurately determining genetic relationships between strains [63,64].It is essential that analyses are performed against a database of strain profiles of sufficient number and quality that represent all taxa with validly described species names, for results to be meaningful.
Where whole-genome sequences of proposed new species are available, in silico analyses that mimic conventional DNA-DNA hybridisations can be used in comparisons with type strain genomes of validly-published extant species,        Alkaline phosphatase Resistance to: Nalidixic acid (30 mg)      4. Exemplar results of whole-genome comparisons for well-characterized strains of Campylobacter, Arcobacter and Helicobacter using Average Nucleotide Identity (ANI) [65] and Genome Blast Distance Phylogeny (GBDP) [66] Genomes were accessed from public databases and ANI and GBDP values calculated using online tools (http://enve-omics.ce.gatech.edu/ani/and http://ggdc.dsmz.derespectively) during December 2016.For ANI, values !95 were proposed to mimic conventional DDH results of !70 % [65].For GBDP, values are proposed equivalent to conventional DDH [66].Genome designations are GenBank reference numbers.The identity of non-type strain genomes was further verified by determining the phylogenetic relationship of their 16S rRNA gene sequences with those derived from type strains (Fig. 1  with which a close phylogenetic relationship has been indicated, to determine interspecific genomic relatedness.Computational approaches described include Average Nucleotide Identity (ANI: [65]) and Genome Blast Distance Phylogeny (GBDP: [66]).Each algorithm is available online (presently http://enve-omics.ce.gatech.edu/ani/for ANI: and http://ggdc.dsmz.defor GBDP).For new species proposals, we recommend both these analyses be presented where whole-genome sequences are used as the source of the genomic data.Complete or draft sequences of appropriate comparator type strains should be used as the basis for comparison.Table 4 lists comparisons between selected complete genomes of well-characterised members of the families Campylobacteraceae and Helicobacteraceae, using the default parameters suggested online for ANI and GBDP algorithms as exemplar output.These data identify some discordance between certain comparator taxa among almost all outputs and presently-accepted classifications; Formula 3 used for GBDP analysis performed optimally on this data set (Table 4).Nonetheless, a continued validation of Formula 3 for GBDP analysis is prudent.

Description of subspecies
There is precedent for the description of genetically-defined subspecies among the Campylobacteraceae in particular (notably for C. jejuni, C. lari, C. fetus and C. hyointestinalis) [31,[67][68][69].Most of these subspecies (apart from C. fetus subsp.fetus and C. fetus subsp.venerealis, where the definition is historically based on differing disease aetiologies) [5,31] are characterized by a high level of infraspecific similarity and distinctive phenotype and often ecotype.This definition aligns with the concept described by Wayne et al. [54], whereby 'subspecies designations can be used for genetically-close organisms that diverge in phenotype'.Subspecies should therefore exhibit DNA-DNA relatedness to the type strain of the species approximating or exceeding 70 %, with strains belonging to a given subspecies demonstrating a higher degree of genomic similarity, and clear differential characteristics in genotype, phenotype and/or ecotype between differing subspecies, as with the examples listed above.

Ecology
The natural habitat(s) of the proposed species should be detailed as much as possible, to include location(s), host species (if applicable), site of isolation, pathogenicity and clinical features (if appropriate).

Closing remarks
Minimal standards for describing new prokaryotic species aim to provide clear guidelines to the scientific community to assist in the delineation of novel taxa in a robust and unambiguous manner.This can only help the community at large who may be required to rapidly recognize emerging threats (or benefits) to public, plant, animal or environmental health.The standards described herein consider (i) previous recommendations for these taxa [8,13,14]; (ii) current, and indeed previous recommendations for characterisation of prokaryotes for taxonomic purposes [54][55][56], and (iii) the emerging discipline of genomic taxonomy, which is in a relative state of infancy and indeed flux [70][71][72][73].At present, the continued need for a polyphasic taxonomic approach to achieve stable and robust classifications remains critical.These standards support that need.

Fig. 1 .
Fig.1.Phylogenetic relationships between 16S rRNA genes of type strains of the taxa encompassed in this study inferred by neighbor-joining tree using the Kimura 2-parameter distance estimation method, with bootstrapping based on 500 replicates.

Table 1 .
Summary phenotypic data for extant Arcobacter spp.