rpoB Gene Sequencing for Identification of Corynebacterium Species

Abstract

The genus Corynebacterium is a heterogeneous group of species comprising human and animal pathogens and environmental bacteria. It is defined on the basis of several phenotypic characters and the results of DNA-DNA relatedness and, more recently, 16S rRNA gene sequencing. However, the 16S rRNA gene is not polymorphic enough to ensure reliable phylogenetic studies and needs to be completely sequenced for accurate identification. The almost complete rpoB sequences of 56 Corynebacterium species were determined by both PCR and genome walking methods. In all cases the percent similarities between different species were lower than those observed by 16S rRNA gene sequencing, even for those species with degrees of high similarity. Several clusters supported by high bootstrap values were identified. In order to propose a method for strain identification which does not require sequencing of the complete rpoB sequence (approximately 3,500 bp), we identified an area with a high degree of polymorphism, bordered by conserved sequences that can be used as universal primers for PCR amplification and sequencing. The sequence of this fragment (434 to 452 bp) allows accurate species identification and may be used in the future for routine sequence-based identification of Corynebacterium species.

The genus Corynebacterium is one of the largest genera in the coryneform group of bacteria (which consist of irregular gram-positive rods and aerobically growing, asporogenous, non-partially acid-fast bacteria). Originally, the genus Corynebacterium was created essentially to accommodate the diphtheria bacillus and some other species pathogenic for animals. Bergey's Manual of Systematic Bacteriology listed only 17 Corynebacterium species; however, 11 new species were defined between 1987 and 1995 (6), and another 32 new species were described between 1996 and 2003. From 2001 to 2003, up to 13 new species were validly published (http://www.bacterio.cict.fr/c/corynebacterium.html). At present, the genus Corynebacterium contains more than 60 species, the vast majority of which have been isolated from human or animal samples. Chemotaxonomically, this genus includes species that possess wall chemotype IV (arabinose, galactose and meso-diaminopimelic acid), short-chain mycolic acids (approximately 22 to 36 carbon atoms), and DNA G+C contents ranging from 51 to 63 mol% (5, 6). The narrower definition of the genus Corynebacterium has resulted in the transfer of several species (Clavibacter, Rhodococcus, and Turicella) to other genera. However, there is still some evidence of heterogeneity within the genus Corynebacterium. For example, Corynebacterium amycolatum and Corynebacterium kroppenstedtii lack mycolic acids (1, 2), Corynebacterium afermentans and Corynebacterium auris exhibit G+C contents of more than 65 mol% (6). The use of molecular genetic methods such as 16S rRNA gene (rDNA) sequence analysis has facilitated a much tighter circumscription of the genus, and the availability of comparative 16S rRNA gene sequence data with improved phenotypic data has resulted in much improved and more reliable species identification (14, 16). These improvements in taxonomy and means of detection, together with an increased interest in Corynebacterium as an opportunistic infectious agent in humans, have resulted in the delineation of a plethora of new Corynebacterium species from human sources in recent years (6, 8). However, the identification of Corynebacterium species is difficult because it often requires fastidious procedures, such as chromatography, or a high number of tests that are not available with commercial identification systems (5). The sequence of the 16S rRNA gene is the most widely used molecular marker to determine the phylogenetic relationships of bacteria. However, low intragenus polymorphism limits its usefulness for taxonomic analysis or identification to the species level. As an example, the species Corynebacterium pseudodiphtheriticum and Corynebacterium propinquum and the species Corynebacterium minutissimum and Corynebacterium aurimucosum have high 16S rDNA similarity values (99.3 and 98.7%, respectively). Moreover, from the perspective of automated systems for gene sequence based-identification, this low degree of polymorphism obligates sequencing of the complete 16S rRNA gene (approximately 1,500 bp) for accurate identification. Variable areas are spotted along the gene at positions 0 to 150, 300 to 400, 650 to 800, 850 to 950, and 1100 to 1250 (a total of 650 bp), with maximal variability ranging from 8 to 19% according to the region (Fig. 1).

FIG. 1.

Graphical representation of RSVs (y axis) in the rpoB and 16S rRNA gene sequences of Corynebacterium species studied by the use of windows of 50 nucleotides (the x axis indicates the nucleotide position). The hypervariable region bordered by conserved regions used for species identification with primers C2700F and C3130R is framed.

Among the universal genes that can be used for taxonomic analysis and gene sequence-based identification, the RNA polymerase beta subunit-encoding gene (rpoB) was used to study several unrelated genera, including Bartonella spp. (15), Staphylococcus spp. (3), members of the family Enterobacteriaceae (13), Bosea spp. and Afipia spp. (9), Mycobacterium spp. (10), and Legionella spp. (11). In the study described here, we investigated the usefulness of rpoB sequencing for the differentiation and identification of 56 Corynebacterium species and 2 related species, Rhodococcus equi and Turicella otitidis. As rpoB is a large gene (approximately 3,500 bp), we also determined regions of variability in the sequence bordered by conserved sequences with the objective of designing universal primers for amplification of a small but discriminative sequence for use in the routine identification of Corynebacterium species.

MATERIALS AND METHODS

Bacterial strains.

The bacterial stains used in this study are listed in Table 1. Most strains were obtained from the Collection de l'Institut Pasteur (CIP) and from the Culture Collection of the University of Göteborg, Göteborg, Sweden (CCUG). All strains were cultured on Columbia agar plates with 5% sheep blood (Trypticase soy agar; bioMérieux, Marcy-l'Etoile, France) and were incubated for 24 to 72 h at 30 to 37°C in a 5% CO₂ atmosphere.

TABLE 1.

Species for which rpoB sequences were determined, including GenBank access numbers and sizes of the sequences determined

Species	Strain	Genbank accession no.		Size (bp) of rpoB sequence determined
		16S rRNA gene	rpoB gene	Complete	Partial
Corynebacterium accolens	CIP 104783T	AJ439346	AY492242	3,282	446
Corynebacterium afermentans subsp. afermentans	CIP 103499T	X82054	AY492265	3,347	446
Corynebacterium afermentans subsp. lipophilum	CIP 103500T	X82055	AY492266	3,178	446
Corynebacterium ammoniagenes	CIP 101283T	X82056	AY492243	3,349	446
Corynebacterium amycolatum	CIP 103452T	X82057	AY492241	3,435	434
Corynebacterium argentoratense	CIP 104296T	X83955	AY492249	3,349	446
Corynebacterium aurimucosum	CCUG 47449T	AJ309207	AY492282	3,330	446
Corynebacterium auris	CIP 104632T	X81873	AY492234	3,357	446
Corynebacterium auriscanis	CIP 106629T	AJ243820	AY492244	3,346	452
Corynebacterium bovis	CIP 5480T	X82051	AY492236	3,450	452
Corynebacterium callunae	CIP 104277T	X82053	AY492245	3,340	446
Corynebacterium camporealensis	CIP 105508T	Y09569	AY492246	3,340	446
Corynebacterium capitovis	CIP 106739T	AJ297402	AY492247	3,350	446
Corynebacterium confusum	CIP 105403T	Y15886	AY492248	3,356	446
Corynebacterium coyleae	CIP 104919T	X96497	AY492250	3,314	446
Corynebacterium cystitidis	CIP 103424T	X82058	AY492251	3,340	446
Corynebacterium diphtheriae	CIP 100721T	X82059	AY492230	3,477	446
Corynebacterium durum	CIP 105490T	Z97069	AY492252	3,340	446
Corynebacterium efficiens	YS-314	AB055963	AP005215	3,480	446
Corynebacterium falsenii	CIP 105466T	Y13024	AY492253	3,330	452
Corynebacterium felinum	CIP 106740T	AJ401282	AY492254	3,334	446
Corynebacterium flavescens	CIP 69.5T	X82060	AY492255	3,303	446
Corynebacterium freneyi	CIP 106767T	AJ292762	AY492237	3,345	434
Corynebacterium glucuronolyticum	CIP 104577T	X86688	AY492256	3,328	434
Corynebacterium glutamicum	ATCC 13032	X80629	NC_003450	3,480	446
Corynebacterium imitans	CIP 105130T	Y09044	AY492259	3,333	446
Corynebacterium jeikeium	CIP 103337T	X82062	AY492231	3,463	452
Corynebacterium kroppenstedtii	CIP 105744T	Y10077	AY492274	3,349	452
Corynebacterium kutscheri	CIP 103423T	X82063	AY492257	3,168	446
Corynebacterium lipophiloflavum	CIP 105127T	Y09045	AY492260	3,340	446
Corynebacterium macginleyi	CIP 104099T	X80499	AY492276	3,173	446
Corynebacterium mastitidis	CIP 105509T	Y09806	AY492281	3,174	446
Corynebacterium matruchotii	CIP 81.82T	X82065	AY492238	3,338	446
Corynebacterium minutissimum	CIP 100652T	X84679	AY492235	3,358	446
Corynebacterium mucifaciens	CIP 105129T	Y11200	AY492261	3,330	446
Corynebacterium mycetoides	CIP 55.51T	X82066	AY492262	3,332	446
Corynebacterium phocae	CIP 105741T	Y10076	AY492277	3,180	446
Corynebacterium pilosum	CIP 103422T	X84246	AY492258	3,296	446
Corynebacterium propinquum	CIP 103792T	X81917	AY492279	3,179	446
Corynebacterium pseudodiphtheriticum	CIP 103420T	X81918	AY492232	3,477	446
Corynebacterium pseudotuberculosis	CIP 102968T	X81916	AY492239	3,447	446
Corynebacterium renale	CIP 103421T	X81909	AY492240	3,442	446
Corynebacterium riegelii	CIP 105310T	Y14651	AY492278	3,180	446
Corynebacterium seminale	CIP 104297T	X84375	AY492263	3,153	434
Corynebacterium simulans	CIP 106488T	AJ012837	AY492264	3,176	446
Corynebacterium singulare	CIP 105491T	Y10999	AY492280	3,180	446
Corynebacterium spheniscorum	CCUG 45512T	AJ429234	AY492283	3,283	446
Corynebacterium striatum	CIP 81.15T	X81910	AY492267	3,346	446
Corynebacterium sundsvallense	CIP 105936T	Y09655	AY492268	3,359	446
Corynebacterium terpenotabidum	CIP 105927T	AB004730	AY492269	3,286	452
Corynebacterium testudinoris	CCUG 41823T	AJ295841	AY492284	3,320	446
Corynebacterium thomssenii	CIP 105597T	AF010474	AY492270	3,352	446
Corynebacterium ulcerans	CIP 106504T	X81911	AY492271	3,176	446
Corynebacterium urealyticum	CIP 103524T	X81913	AY492275	3,172	452
Corynebacterium variabile	CIP 102112T	AJ222815	AY492272	3,343	452
Corynebacterium vitaeruminis	CIP 827T	X84680	AY492273	3,296	446
Corynebacterium xerosis	CIP 100653T	X81914	AY492233	3,447	434
Turicella otitidis	CIP 104075T	X73976	AY492287	3,250	446
Rhodococcus equi (formerly Corynebacterium hoagii)	CIP 81.17T	X82052	AY492285	3,357	449
Rhodococcus equi	CIP 5472T	AF490539	AY492286	3320	449

rpoB gene amplification and sequencing.

The sequence of the rpoB gene from Corynebacterium species and species most closely related to Corynebacterium species were aligned in order to produce a consensus sequence. The bacteria chosen were Corynebacterium glutamicum, Amycolatopsis mediterranei, and Mycobacterium smegmatis (GenBank accession numbers NC_003450, AF242549, and MSU24494, respectively). The consensus sequence was used to generate primers that were used in PCRs, for genome walking (17), and for sequencing. Additional primers were selected from ongoing base sequence determinations. All primers used in this study are summarized in Table 2.

TABLE 2.

Primers used for amplification and sequencing of the rpoB gene in this study

Primer name	Primer sequence (5′-3′)	Positiona	Tmb (°C)
C240F	GGAAGGAYGCATCTTGGCAGTCT	−13	68
C150F	GGYACGCCYGAGTGGC	133	56
C35F	GGAAGGACCCATCTTGGCAGT	−13	66
C41F	CAGTCTCCCGCCAGACCA	5	60
C445R	CATYGGGAARTCRCCGATGA	401	60
C40F	CAGTCTCCCGCCAGACCAA	5	62
C390F	ATCAAGTCYCAGACKGTYTTCATC	322	68
C390R	GATGAARACMGTCTGRGACTTGAT	322	68
C630F	GACCGCAAGCGYCGCCAG	621	64
C600f	TGGYTBGARTTYGACGT	574	50
C600r	ACGTCRAAYTCVARCCA	574	50
C640R	GGCTGRCGRCGCTTGCGGT	623	66
C890F	TACAAGRTCAACCGCAAG	883	52
C820R	GGRCGYTGCTTGCGGTAGA	772	62
C1050F	CGAYGACATYGACCACTT	1040	54
C1050R	GGTTRCCRAAGTGGTCRATGTC	1045	68
C1295F	CAGTTYMTGGACCAGAACAAC	1254	62
C1410F	GAGCGYATGACCACBCAGGA	1144	64
C1410R	TCCTGVGTGGTCATRCGCTC	1144	64
C1415F	CBCACTACGGMCGYATGTG	1373	62
C1740F	ACGATGCTAACCGTGCACTGAT	1739	66
C1740R	CCCATCAGTGCACGGTTAGCAT	1742	68
C1765R	GTGCTCSAGGAAYGGRATCA	1718	62
C1770F	TGATGGGYGCSAACATGCAG	1757	64
C1800f	ATGGGYGCSAACATGCAG	1759	56
C1800r	CTGCATGTTSGCRCCCAT	1759	56
C2160R	GRCCYTCCCAHGGCATGAA	2107	60
C2130F	GGARGGCCACAACTACGAGGA	2118	64
C2130R	GTGGCCYTCCCAHGGCATGAA	2107	68
C2350F	ACATCCTGGTCGGTAAGGTCAC	2339	68
C2350R	GTGACCTTACCGACCAGGATGT	2339	68
C2385F	CATCCTSGTSGGYAAGGTCA	2340	64
C2410R	ATGATCGCRTCCTCGTAGTTGTG	2125	68
C2410F	CACAACTACGAGGAYGCGATCAT	2125	68
C2470R	CGATCTCGTGCTCCTCGATGT	2192	66
C2590F	CARAAGCGCAAGATCCARGA	2563	60
C2625F	AGATCCARGAYGGCGAYAAG	2572	60
C3190F	ATGGAGGTGTGGGCAATGCAG	3154	66
C3190R	CTGCATTGCCCACACCTCCAT	3154	66
C3200r	CTGCATBGCCCACACCTCCAT	3154	68
C3215R	GCCTGCATBGCCCACACCT	3158	64
C3300F	GAAGGGCGADAAYATYCCGGAT	3264	66
C3300R	TCCGGRATRTTHTCGCCCTTCA	3263	66
C3350R	CCTTGAASGACTCHGGRATAC	3290	64
C3490R	CACGGGACAGGTTGATGCC	3430	62
C3630R	GAGMACCTCSACGTTSAGGCACA	3335	70
C3500R	TCGTCDCGBGACAGGTTGATG	3433	66
C2700F	CGWATGAACATYGGBCAGGT	2714	60
C3130R	TCCATYTCRCCRAARCGCTG	3140	62

^aThe position of a primer is relative to that of the Corynebacterium diphtheriae rpoB gene sequence (GenBank accession no. AY492230).

^bT_m, melting temperature.

Bacterial DNA was extracted from a heavy suspension of strains by using the QIAamp blood kit (Qiagen, Hilden, Germany), according to the recommendations of the manufacturer. All PCR mixtures contained 2.5 × 10⁻² U of Taq polymerase per μl; 1× Taq buffer; 1.8 mM MgCl₂ (Gibco BRL, Life Technologies, Cergy Pontoise, France); dATP, dCTP, dTTP, and dGTP (Boehringer Mannheim GmbH, Hilden, Germany) at concentrations of 200 μM each; and each primer (Eurogentec, Seraing, Belgium) at a concentration of 0.2 μM. The PCR mixtures were subjected to 35 cycles of denaturation at 94°C for 30 s, primer annealing for 30 s, and extension at 72°C for 2 min. Every amplification program began with a denaturation step of 95°C for 2 min and ended with a final elongation step of 72°C for 10 min. Determination of the complete sequences of the rpoB sequence ends was achieved by use of the sequences of both the 3′ and the 5′ ends of the gene and amplification by PCR with the Universal GenomeWalker kit (Clontech Laboratories, Palo Alto, Calif.). Briefly, genomic DNA was digested with EcoRV, DraI, PvuII, StuI, and ScaI. The DNA fragments were then ligated with a GenomeWalker adaptor, which had one blunt end and one end with a 5′ overhang. The ligation mixture with the adaptor and the genomic DNA fragments were used as templates for the PCR. This PCR was performed by use of an adaptor primer supplied by the manufacturer and specific primers to walk through the DNA sequence downstream. For the amplification, 1.5 U of Elongase (Boehringer Mannheim) was used with 10 pmol of each primer, each deoxynucleoside triphosphate at a concentration of 20 mM, 10 mM Tris-HCl, 50 mM KCl, 1.6 mM MgCl₂, and 5 μl of template with a final volume of 50 μl. Amplicons were purified for sequencing by use of a QIAquick spin PCR purification kit (Qiagen) by the protocol of the supplier. Sequencing reactions were carried out with the reagents of the ABI Prism 3100 DNA sequencer (dRhod.Terminator RR Mix; Perkin-Elmer Applied Biosystems) by the standard automated sequencer protocol.

Determination of discriminative partial sequences in 16S rRNA and rpoB genes.

In order to search for parts of sequences with high degrees of variability bordered by conserved regions, we used SVARAP software (Sequence Variability Analysis Program [http://ifr48.free.fr/recherche/jeu_cadre/jeu_rickettsie.html]) (9). After this analysis was done, the most polymorphic areas in rpoB were identified, and primers designed to be specific for the border conserved region were used for PCR amplification of this region. The PCR conditions that incorporated this consensus primer pair (C2700F-C3130R; Table 2) were the same as those described above. These primers were then used for amplification and sequencing of the hypervariable region for all the strains studied in this work.

rpoB sequence analysis.

The nucleotide sequences of the rpoB gene fragments obtained were processed into sequence data with Sequence Analysis Software (Applied Biosystems), and partial sequences were combined into a single consensus sequence with Sequence Assembler Software (Applied Biosystems). All GenBank accession numbers are listed in Table 1. Multiple-sequence alignments and percent similarities of the rpoB and 16S rRNA genes between the different species were obtained with the CLUSTAL W program (18) on the EMBL-EBI web server (http://www.ebi.ac.uk/clustalw/). Phylogenetic trees were obtained from DNA sequences by three different methods: the neighbor-joining, maximum-parsimony, and maximum-likelihood methods (4). Bootstrap replicates were performed in order to estimate the reliabilities of the nodes of the phylogenetic trees obtained. Bootstrap values were obtained from 1,000 trees generated randomly with the SEQBOOT program in the PHYLIP software package.

RESULTS

rpoB sequences of Corynebacterium species.

Almost complete rpoB gene sequences were determined for all strains. The rpoB sequence was more polymorphic than the 16S rDNA sequence. This higher degree of polymorphism was particularly evident for species not well differentiated by 16S rDNA sequence analysis (Table 3), as among the 11 pairs of species with 16S rRNA gene similarities ranging from 98.5 to 99.7%, the similarities of the rpoB gene ranged from 84.9 to 96.6%. The means for the similarities between the 16S rRNA gene and rpoB gene sequences among these 11 pairs were statistically significant. This higher degree of polymorphism was also significant when it was calculated on the basis of range site variability (RSV) (Fig. 1). RSVs of ≥10 were observed in the rpoB gene for 44 of 67 windows of 50 nucleotides (WOFN) and in the 16S rRNA gene for 5 of 27 WOFN (P < 0.001 by the chi-square test). Likewise, RSVs of ≥20 were observed in the rpoB gene for 13 of 67 WOFN and in the 16S rRNA gene for 0 of 27 WOFN (P = 0.008 by Fisher's exact test). The similarity between the two C. afermentans subspecies was 98.2% and, thus, was 1.6% above the highest level of similarity between two species.

TABLE 3.

Comparison of similarities of 16S rRNA and rpoB gene sequences between the two subspecies of C. afermentans and among the 11 pairs of closely related species, with statistical comparison of mean similarities

Species or subspecies compared	% Similarity
	16S rDNA	ropB gene
		Complete sequenceb	Partial sequencec
Closely related speciesa
C. diphtheriae-C. ulcerans	98.5	86	87.9
C. diphtheriae-C. pseudotuberculosis	98.5	84.9	87.9
C. ulcerans-C. pseudotuberculosis	99.7	93.6	93
C. pseudodiphtheriticum-C. propinquum	99.3	89.7	93.9
C. aurimucosum-C. singulare	99	94.2	93.9
C. aurimucosum-C. minutissimum	98.7	94.6	93.9
C. singulare-C. minutissimum	98.9	93.8	95.5
C. xerosis-C. freneyi	98.7	96.6	95.9
C. macginleyi-C. accolens	98.7	93.3	91.7
C. sundsvallens-C. thomssenii	98.9	90.4	91
C. mucifaciens-C. afermentans	98.5	94	92.4
Mean	98.85	91.91	92.45
C. afermentans subsp. afermentans- C. afermentans subsp. lipophilum	99.8	98.2	96.6

^aDetermined on the basis of ≥98.5% similarity of the 16S rRNA gene sequence.

^bP = 0.03 compared to the results for 16S rDNA (Student's t test).

^cP = 0.01 compared to the results for 16S rDNA (Student's t test).

Phylogenetic analysis.

On the basis of rpoB gene sequence analysis, phylogenetic analysis by the neighbor-joining, maximum-parsimony, and maximum-likelihood methods provided similar and reliable organizations for the four clusters supported by high bootstrap values (Fig. 2). On the contrary, only cluster 4 was evidenced when 16S rRNA gene sequence analysis was used (Fig. 3). The bootstrap values at the nodes were in all cases higher than those observed by 16S rRNA gene sequencing. Values ≥95% were observed for 14 of 55 nodes for the 16S rRNA gene, whereas values ≥95% were observed for 24 of 55 nodes for the rpoB gene (P = 0.004 by the chi-square test). For some species, such as Corynebacterium testudinoris, Corynebacterium renale, Corynebacterium seminale, and Corynebacterium glucuronolyticum, the phylogenetic position was more difficult to assess. The position of T. otitidis in a genus separate from Corynebacterium is also questionable. Study of the rpoB gene confirms that the genus Rhodococcus is different from Corynebacterium and that Corynebacterium hoagii is not another species but is R. equi and that C. seminale and C. glucuronolyticum are the same species (http://www.bacterio.cict.fr/c/corynebacterium.html).

FIG. 2.

Dendrogram representing the phylogenetic relationships of Corynebacterium species obtained by the neighbor-joining method. The tree was derived from the alignments of rpoB gene sequences. The support of each branch, as determined from 1,000 bootstrap samples, is indicated by the value at each node (in percent).

FIG. 3.

Dendrogram representing the phylogenetic relationships of Corynebacterium species obtained by the neighbor-joining method. The tree was derived from alignment of 16S rRNA gene sequences. The support of each branch, as determined from 1,000 bootstrap samples, is indicated by the value at each node (in percent).

Strain identification.

Four highly variable zones were determined by the use of SVARAP software (Fig. 1). These zones were between positions 1 and 450, 800 and 1100, 1400 and 1750, and 2750 and 3200. Attempts to design universal primers that amplify hypervariable areas were unsuccessful for the first three regions. We designed a consensus primer pair (C2700F-C3130R) that allowed the successful amplification of the region in all Corynebacterium species, R. equi, and T. otitidis between positions 2750 and 3200. The amplified fragment was from 434 to 452 bp, depending on the species. Interestingly, this region was the most variable one (Fig. 1). The similarities observed in the partial rpoB sequence were also significantly less than those observed in the 16S rRNA gene and ranged from 87.9 to 95.9% (Table 3). The similarity between the two C. afermentans subspecies was 96.6% and was thus 0.7% greater than the highest degree of similarity between two species.

DISCUSSION

The description of new bacterial species at present is based on the results of DNA-DNA hybridization and the description of phenotypic characteristics, so-called polyphasic classification data (7, 19). However, DNA-DNA hybridization is difficult to perform, expensive, technically complex, and labor-intensive. The scarcity of reproducible and distinguishable characters frequently limits phenotypic characterization and, thus, phenotype-based identification in routine clinical microbiology laboratories. The development of gene amplification and sequencing, especially that of 16S rRNA gene sequences, has simplified the taxonomy and identification of bacteria, particularly those lacking distinguishable phenotypic characteristics. However, the 16S rDNA sequences of Corynebacterium spp. are not variable enough to ensure confident results from phylogenetic studies based on high bootstrap values (Fig. 3) or to allow determination of a short sequence for accurate identification (Fig. 1). Our data, based on the rpoB sequences of these bacteria, confirm that this gene is significantly more polymorphic than the 16S rRNA gene, and we propose that it be used to replace or complement the 16S rRNA gene for phylogenetic studies of Corynebacterium. Deeply branching nodes were supported by high bootstrap values and allowed the identification of four clusters (Fig. 2). Even among species not resolved into clusters, some groups of bacteria were confidently identified, such as groups containing Corynebacterium diphtheriae, Corynebacterium pseudotuberculosis, Corynebacterium ulcerans, and Corynebacterium kutscheri.

The high similarity values for the 16S rRNA gene sequences observed among closely related Corynebacterium spp. indicate that the complete sequence should be determined for accurate sequence-based identification (Table 3). By using SVARAP software, we have designed universal primers for rpoB that allow amplification and sequencing of a 434- to 452-bp fragment polymorphic enough to ensure accurate identification of all Corynebacterium spp. The highest degree of similarity of this partial sequence between two species was 95.9%, whereas it was 99.7% for the complete 16S rRNA gene (Table 3), a sequence nearly four times longer. Moreover, the partial sequences of the rpoB genes of two subspecies of C. afermentans had a similarity of 96.6%, which was thus 0.7% above the limit of similarity between two different species. This difference was only 0.1% for the complete 16S rRNA gene sequence, rendering it impossible to distinguish a subspecies from a closely related species only on the basis of this sequence. This difference was even higher (1.6%) when the complete rpoB sequence was considered. From these data, the cutoff for the definition of species and subspecies in the genus Corynebacterium based on the complete rpoB sequence can be made on the basis of similarities of <96.6 and >98%, respectively. These cutoffs are in the same range as those observed for the genera Bartonella, Afipia, and Bosea (12, 9). However, the similarities of a large collection of different strains within particular species would have to be determined for validation of these cutoffs.