Genetic Classification and Distinguishing of Staphylococcus Species Based on Different Partial gap, 16S rRNA, hsp60, rpoB, sodA, and tuf Gene Sequences

Abstract

The analysis of 16S rRNA gene sequences has been the technique generally used to study the evolution and taxonomy of staphylococci. However, the results of this method do not correspond to the results of polyphasic taxonomy, and the related species cannot always be distinguished from each other. Thus, new phylogenetic markers for Staphylococcus spp. are needed. We partially sequenced the gap gene (∼931 bp), which encodes the glyceraldehyde-3-phosphate dehydrogenase, for 27 Staphylococcus species. The partial sequences had 24.3 to 96% interspecies homology and were useful in the identification of staphylococcal species (F. Layer, B. Ghebremedhin, W. König, and B. König, J. Microbiol. Methods 70:542-549, 2007). The DNA sequence similarities of the partial staphylococcal gap sequences were found to be lower than those of 16S rRNA (∼97%), rpoB (∼86%), hsp60 (∼82%), and sodA (∼78%). Phylogenetically derived trees revealed four statistically supported groups: S. hyicus/S. intermedius, S. sciuri, S. haemolyticus/S. simulans, and S. aureus/epidermidis. The branching of S. auricularis, S. cohnii subsp. cohnii, and the heterogeneous S. saprophyticus group, comprising S. saprophyticus subsp. saprophyticus and S. equorum subsp. equorum, was not reliable. Thus, the phylogenetic analysis based on the gap gene sequences revealed similarities between the dendrograms based on other gene sequences (e.g., the S. hyicus/S. intermedius and S. sciuri groups) as well as differences, e.g., the grouping of S. arlettae and S. kloosii in the gap-based tree. From our results, we propose the partial sequencing of the gap gene as an alternative molecular tool for the taxonomical analysis of Staphylococcus species and for decreasing the possibility of misidentification.

The genus Staphylococcus comprises 42 validly described species and subspecies of gram-positive, catalase-positive cocci (1, 21, 30), 10 of which contain subdivisions with subspecies designations (6, 10, 27, 30). Staphylococci, including S. aureus, generally are opportunistic pathogens or commensals on host skin. However, they may act as pathogens if they gain entry into the host tissue through a trauma to the cutaneous barrier, inoculation by needles, the implantation of medical devices, or in cases in which the microbial community is disturbed or in immunocompromised individuals (17-19). Thus, the accurate species identification of S. aureus as well as that of the other staphylococcal species in microbial communities is highly desirable to permit a more precise determination of the host-pathogen relationships of staphylococci (13, 15). The precise identification of these bacteria to the species level is quite laborious. Various molecular DNA-based methods for the identification of Staphylococcus species have been developed. These methods typically require the use of several species-specific PCR primers, hybridization probes, or multiple restriction enzymes and usually are not designed to differentiate all known species simultaneously. 16S rRNA gene sequencing and PCR-restriction fragment length polymorphism (PCR-RFLP) analysis have been described for Staphylococcus species identification (2-4), but these methods do not differentiate between Staphylococcus lentus and Staphylococcus sciuri. PCR-RFLP analysis of the 23S rRNA gene with two restriction enzymes is able to discriminate between Staphylococcus species (23), but the interpretation of the results is complicated by intervening sequences (9). More recently, amplified fragment length polymorphism fingerprinting has proven to be useful for Staphylococcus species identification, but the method is time-consuming and expensive (34). Whole-genome DNA-DNA hybridization analysis (31) allows species identification, but the method is not suitable for routine use.

The use of nucleic acid targets, with their high sensitivity and specificity, provides an alternative technique for the accurate identification and classification of Staphylococcus species. Besides the 16S rRNA gene (2-4), the 16S-23S rRNA intergenic spacer region (23), and the heat shock protein 60 (hsp60) gene (11, 12), other gene sequences have been used in genetic studies: the femA gene (35), the sodA gene (28), the tuf gene (24), the rpoB gene (5, 26), and the gap gene (36, 37).

In our study, we assessed the usefulness of the ∼931-bp partial sequence of gap for the studied staphylococci (n = 27) in species differentiation and for interfering interspecies phylogenetic relationships. These are among the most commonly occurring species of greater clinical significance and are preferentially novobiocin-sensitive staphylococci: e.g., S. aureus, S. epidermidis, S. warneri, S. haemolyticus, and S. lugdunensis. The other 15 species that were not subjects of this study are rarely associated with infections in humans; e.g., S. pasteuri, S. vitulinus, and S. saccharolyticus. The gap gene encodes a 42-kDa transferrin-binding protein (Tpn) located within the cell wall of the staphylococci. Tpn is a member of the newly emerging family of multifunctional cell wall-associated glyceraldehyde-3-phosphate dehydrogenases, which is well known for its glycolytic function of converting d-glyceraldeyde-3-phosphate to 1,3-bisphosphoglycerate. gap commonly has been considered a constitutive housekeeping gene (8, 25).

Yugueros and coworkers published the sequences of the gap genes of 12 staphylococcal species relevant for humans (36). We extended these studies and sequenced the ∼931-bp sequence encoding a partial region of the gap gene from a total of 27 different staphylococcal species (22). We consider these species to be among the clinically significant species, as do other groups (2-5, 12, 28).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

All of the staphylococcal strains were grown on blood agar and incubated at 37°C for 18 to 24 h. Reference strains were selected from the German Collection of Microorganisms and Cell Cultures (DSMZ), the Czech Collection of Microorganisms (CCM), and the American Type Culture Collection (ATCC), and they included the following: Staphylococcus arlettae DSM 20672^T, S. aureus ATCC 29213^T, S. carnosus subsp. carnosus DSM 20501^T, S. cohnii subsp. cohnii DSM 20260^T, S. delphini DSM 20771^T, S. epidermidis DSM 20044^T (CCM 2124^T), S. equorum subsp. equorum DSM 20674^T, S. hyicus DSM 20459^T, S. intermedius DSM 20373^T (CCM 5739^T), S. kloosii DSM20676^T, S. lugdunensis DSM 4804^T (ATCC 43809^T), S. warneri DSM 20316 (CCM 2730^T), S. capitis subsp. capitis CCM 2734^T, S. caprae CCM 3573^T, S. chromogenes CCM 3387^T, S. gallinarum CCM 3572^T, S. haemolyticus CCM 1798^T, S. hominis subsp. hominis CCM 2732^T, S. lentus CCM 3472^T, S. muscae CCM 4175^T, S. saprophyticus subsp. saprophyticus CCM 883^T, S. sciuri CCM 3473^T, S. simulans CCM 2705^T, S. xylosus CCM 2725^T, S. auricularis ATCC 33753^T, S. felis ATCC 49168^T, and S. schleiferi subsp. schleiferi ATCC 43808^T.

Isolation of genomic DNA.

Chromosomal DNA was isolated from overnight cultures grown on blood agar at 37°C. Genomic DNA was extracted by using the Qiagen DNA extraction kit according to the manufacturer's suggestions (Hilden, Germany), with the modification that 20 μl of lysostaphin (1 mg/ml; Sigma) and 20 μl lysozyme (100 mg/ml; Qiagen) were added at the cell lysis step. The concentration of the DNA was assessed spectrophotometrically.

DNA sequencing.

Consensus gap PCR primers (Table 1) were used as previously described (22). Gap1-for and Gap2-rev were used to amplify the ∼931-bp fragment as described before (37), and the PCR products were purified with the Qiagen gel extraction kit (Hilden, Germany). Partial reverse and forward sequencing of the ∼931-bp fragment was obtained by using the consensus primers at 3.25 pmol (Table 1). Sequencing reactions were carried out with the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems) according to the manufacturer's instructions and using the previously described sequencing protocols (22). The results were processed into sequence data with sequence analysis software (Applied Biosystems), and partial sequences were combined into a single consensus sequence with assembler software (Applied Biosystems) (22). The gene sequences other than that of gap were obtained from GenBank (Table 2).

TABLE 1.

Primers used for sequencing the gap gene

Primer	Sequence (5′→3′)	Positiona
Gap1-forb	ATGGTTTTGGTAGAATTGGTCGTTTA	22-47
Gap3-revc	G(ACT)TTT(AGCT)A(CT)TTCTT(AGT) (AT)CC(AG)TT(ACT)AC(AGT)C	220-196
Gap4-forc	GA(CT)GT(AGCT)GT(AGCT)(CT)T(AT)GAATGTAC(AT)GG	270-292
Gap5-revc	GTT(AT)GT(AT)GTACA(AGT)GA(ACT)GCACC(AT)G	462-440
Gap6-forc	GAAGG(ACT)(CT)T(ACT)ATGAC(AGT)AC(AT)AT(CT)CA(CT)G	511-535
Gap7-revc	GAACC(AT)GT(AT)GC(AT)AC(AT)GG (ACT)ACACGTTG	723-698
Gap8-forc	GAA(ACT)CATT(CT)GGTTACA(AC) (ACT)GA(AT)GA(CT)G	809-834
Gap2-revb	GACATTTCGTTATCATACCAAGCTG	956-932

^aPosition relative to the S. aureus gap sequence.

^bPrimer sequences from Yugueros et al. (36).

^cDesigned primer sequences from this study.

TABLE 2.

Partial gene sequences from GenBank used for phylogenetic tree

Staphylococcus strain	GenBank sequence for:
	16S rRNA	hsp60	rpoB	sodA	tuf
S. arlettae	AY688029	AF053580	AF325874	AJ343894
S. aureus	AY688034	AF060191	AF325894	AY485191	AF274739
S. auricularis	AY688030	AF242278	AF325889	AJ343937	AF298797
S. capitis subsp. capitis	AY688039	AF036322	AF325885	AJ343940	AF298798
S. caprae	AY688036	AF053574	AF325896	AJ343898
S. carnosus subsp. carnosus	AY688041	AF242279	AF325880	AJ343899
S. chromogenes	AY688044	AF242280	AF325892	AJ343945
S. cohnii subsp. cohnii	AY688046	AF053582	AF325893	AJ343902
S. delphini	AY688050	AF019774		AJ343905
S. epidermidis	AY688053	AF029245	AF325872	AJ343906	AF298800
S. equorum subsp. equorum	AY688054	AF242281	AF325882	AY878697
S. felis	AY688057	AF242282	AF325878	AJ343908
S. gallinarum	AY688059	AF053579	AF325890	AJ343909
S. haemolyticus	AY688062	U92809	AF325888	AJ343910	AF298801
S. hominis subsp. hominis	AY688064	AF053572	AF325875	AJ343911
S. hyicus	AY688066	AF019778	AF325876	AJ343913
S. intermedius	AY688070	AY123723	AF325869	AJ343914
S. kloosii	AY688072	AF053575	AF325891	AJ343915
S. lentus	AY688073	AF053586	AY036973	AY485195
S. lugdunensis	AY688076	AF053570	AF325870	AJ343917	AF298803
S. muscae	AY688079	AF242285	AF325884	AJ343919
S. saprophyticus subsp. saprophyticus	AY688089	AF053578	AF325873	AJ343954	AF298804
S. schleiferi subsp. schleiferi	AY688093	AF053585	AF325886	AJ343955
S. sciuri	AB212276	AY820255	AY820256	AY820257	AY763434
S. simulans	AY688101	AF053584	AF325877	AJ343956	AF298805
S. warneri	AY688106	AF053569	AF325887	AJ343958	AF298806
S. xylosus	AY688107	AF053573	AF325883	AJ343959	AY763438

Phylogenetic analysis.

Sequences were aligned manually in Sequencher 3.0 (Gene Codes Corporation) to edit the sequences, if necessary, and to note which regions were to be excluded for the phylogenetic analysis. Multiple-sequence alignments and topology predictions were done with DNASISMAX, version 2.0.5 (2003) (Hitachi Software Engineering, Japan). Phylogenetic trees were generated with the neighbor-joining algorithm by using DNASISMAX. All trees were resampled with 1,000 bootstrap replications to ensure the robustness of the data (7). The phylogenetic analyses were displayed with the TreeView drawtree program, version 1.6.6 (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html). The DNA sequence similarity analysis was performed with BioEdit, version 7.0.1 (14).

RESULTS

PCR, sequencing of the gap gene, and sequence similarity for the staphylococcal species.

The amplification of the partial gap gene for all of the Staphylococcus species yielded a single product of nearly 931 bp. The GenBank accession numbers are DQ321674 to DQ321700.

The sequence similarity of the gap sequences ranged from 24.3 to 96% (Table 3). The species S. lentus and S. sciuri revealed a sequence similarity of 93% according to the gap gene sequences, whereas the other gene-based similarities ranged between 88 and 98% (88% for hsp60, 88.9% for sodA, and 98% for 16S rRNA). The species S. capitis and S. caprae revealed a sequence similarity of 27% according to the gap gene sequences, whereas the other gene-based similarities ranged between 94 and 99% (94% for sodA, 95% for 16S rRNA, and 99% for hsp60). The species S. carnosus and S. simulans revealed a sequence similarity of 95% according to the gap gene sequences, whereas the other gene-based similarities ranged between 86 and 98% (87 to 88% for rpoB and 96% for 16S rRNA).

TABLE 3.

DNA sequence identity matrix based on comparisons of the gap gene sequences of the Staphylococcus species

Taxon (gap gene no.)	% Identity with gap gene no.a:
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27
S. arlettae (1)	ID
S. aureus (2)	75	ID
S. auricularis (3)	85	74	ID
S. capitis (4)	75	90	75	ID
S. caprae (5)	24	27	24	27	ID
S. carnosus (6)	29	27	27	27	45	ID
S. chromogenes (7)	84	74	83	72	24	29	ID
S. cohnii (8)	88	75	87	75	26	28	83	ID
S. delphini (9)	85	74	84	72	25	29	88	85	ID
S. epidermidis (10)	74	89	75	91	27	27	72	76	74	ID
S. equorum (11)	90	74	86	74	25	28	84	89	84	74	ID
S. felis (12)	47	46	47	46	25	27	47	48	48	47	46	ID
S. gallinarum (13)	89	74	87	74	25	28	84	88	84	74	89	47	ID
S. haemolyticus (14)	30	28	28	28	44	85	30	29	31	28	29	27	29	ID
S. hominis (15)	86	75	83	75	25	28	85	85	84	75	85	46	85	29	ID
S. hyicus (16)	84	74	84	72	24	27	91	84	88	72	84	48	84	29	83	ID
S. intermedius (17)	86	74	84	73	25	29	88	85	96	74	84	49	84	31	84	87	ID
S. kloosii (18)	92	75	86	75	25	28	84	89	85	74	89	48	90	30	85	85	85	ID
S. lentus (19)	81	70	80	71	26	27	81	81	82	71	83	48	83	28	80	81	83	82	ID
S. lugdunensis (20)	74	88	75	88	26	26	72	74	73	87	73	46	74	27	77	72	73	74	69	ID
S. muscae (21)	29	27	28	27	44	83	30	29	30	27	28	27	28	85	28	28	30	29	28	26	ID
S. saprophyticus (22)	89	74	86	73	24	28	85	90	84	73	91	47	90	29	85	84	84	89	82	72	29	ID
S. schleiferi (23)	30	27	28	27	43	83	29	29	30	28	29	26	28	86	29	28	29	30	27	27	88	29	ID
S. sciuri (24)	82	69	80	70	26	27	80	81	82	70	83	48	83	29	81	80	83	82	93	69	28	82	27	ID
S. simulans (25)	29	27	27	27	44	95	29	28	29	27	28	28	28	86	28	27	29	28	28	26	84	28	83	28	ID
S. warneri (26)	87	77	85	78	24	27	84	86	84	76	88	47	89	29	87	83	85	89	83	75	28	86	29	83	28	ID
S. xylosus (27)	29	27	29	26	41	81	30	28	30	27	28	27	28	84	29	28	30	29	27	25	80	29	80	27	82	27	ID

^aID, identical.

The species S. chromogenes and S. hyicus revealed a sequence similarity of 91% according to the gap gene sequences, whereas the other gene-based similarities ranged between 77.6 and 98% (77.6% for sodA and 98% for 16S rRNA).

Staphylococcus phylogeny derived from gap sequences.

The Staphylococcus species were divided into three clades (Fig. 1) with significant bootstrap values (>90%); the first contained the S. hyicus/S. intermedius group, comprising S. hyicus, S. chromogenes, S. delphini, and S. intermedius. The second clade contained two groups, the S. sciuri group, comprising S. sciuri and S. lentus, and the S. haemolyticus/S. simulans group, comprising S. haemolyticus, S. xylosus, S. muscae, S. simulans, S. schleiferi subsp. schleiferi, S. carnosus subsp. carnosus, S. caprae, and S. felis. The third clade contained only one group, the S. aureus/S. epidermidis group, comprising S. aureus, S. hominis subsp. hominis, S. warneri, S. epidermidis, S. capitis subsp. capitis, and S. lugdunensis. The branching of S. auricularis, S. cohnii, and the heterogeneous S. saprophyticus group, comprising S. saprophyticus subsp. saprophyticus, S. equorum subsp. equorum, S. gallinarum, S. arlettae, and S. kloosii, was not reliable (bootstrap values of 75, 75, and 29%, respectively).

FIG. 1.

Neighbor-joining tree based on the 931- to 933-bp gap sequences and 16S rRNA, rpoB, sodA, and tuf gene sequences showing the phylogenetic relationships among the staphylococcal species selected for this study. The value on each branch is the percent occurrence of the branching order in bootstrapped trees (7).

For the gap gene comparison between S. sciuri and S. lentus, a sequence similarity value of 82% (Table 3) was determined, as the position of these species in the phylogenetic tree is supported by a bootstrap value of 100% (see below).

Comparative phylogeny based on various gene sequence-derived trees.

The gap-derived sequence similarity analysis for the staphylococcal species is given in Table 3. According to the staphylococcal gene sequences of gap, rpoB, and sodA, the species S. felis and S. muscae grouped within the same cluster, whereas in the 16S rRNA- and hsp60-derived trees these two species did not show a close relationship.

The species S. hyicus and S. chromogenes clustered together according to the 16S rRNA, hsp60, and rpoB sequence analysis, but the clustering by sodA analysis was less close. The grouping of S. hyicus and S. chromogenes was more affirmed by the gap gene than by the other genes as well.

The grouping of S. arlettae and S. kloosii in the gap-based tree (with bootstrap values of 98%) was fairly supported by hsp60-, rpoB-, sodA-, and 16S rRNA-derived trees (bootstrap values of <43%).

S. hominis subsp. hominis and S. lugdunensis clustered into the same group according to gap (bootstrap value of 95%), tuf (bootstrap value of 44%), and hsp60 (bootstrap value of 29%) analyses, whereas this was not observed for 16S rRNA, rpoB, and sodA analyses.

The grouping of S. delphini and S. intermedius was supported by gap, 16S rRNA, hsp60, and sodA analyses. The phylogenetic relationship of S. chromogenes and S. hyicus in the gap- and rpoB-derived trees (for each bootstrap value of 100%) was supported by low bootstrap values in hsp60-, 16S rRNA-, and sodA-derived trees. The close relationship of S. carnosus subsp. carnosus and S. simulans in gap (bootstrap value of 100%), hsp60, sodA (for each bootstrap value of 98%), and rpoB (bootstrap value of 79%) analyses was confirmed as well. S. delphini and S. intermedius clustered together according to gap, 16S rRNA, sodA, and hsp60 analyses, with bootstrap values of 100% each.

DISCUSSION

Several molecular targets have been exploited for the molecular identification of Staphylococcus species. Because a large amount of rrs sequence data is available in a public database, it is not surprising that the 16S rRNA gene has been an obvious choice. Gene sequence-based identification of bacteria at the species level may require resolving the whole gene, yet in some cases, phylogenetically closely related bacterial species cannot be differentiated from each other. Although the comparison of the 16S rRNA gene sequences has been useful in phylogenetic studies at the genus level, its use has been questioned in studies at the species level. In this regard, the 16S rRNA sequence similarity has been shown to be very high, 90 to 99%, in 29 Staphylococcus species (20). S. caprae and S. capitis cannot be distinguished by their 16S rRNA gene sequences (34). Similarly, some Staphylococcus taxa have the same 16S rRNA gene sequences in variable regions V1, V3, V7, and V9, with identical sequences occurring in, e.g., S. vitulinus, S. saccharolyticus, S. capitis subsp. urealyticus, S. caprae, the two subspecies of S. aureus, and the two subspecies of S. cohnii (34). Other gene sequences have been used empirically in attempts to classify the Staphylococcus species, namely, the hsp60 gene, the sodA gene, the rpoB gene, and the tuf gene. With regard to the hsp60 gene, it should be noted that the cloned partial hsp60 gene DNA sequences of nine isolates of S. aureus showed a mean variability of only 2% (33). Also, cross-hybridization occurred in cloned partial hsp60 genes between the DNAs from S. intermedius and S. delphini (12). The gap sequences were less conserved compared to the above-mentioned genes (sequence similarities, 24 to 96%). Therefore, the gap gene is rather more discriminative, as shown for S. caprae and S. capitis, which were clearly distinguished from each other, in contrast to results from 16S rRNA gene analyses. For the sodA gene, a pairwise comparison of these sequences revealed a mean sequence similarity of 81.5%, which was lower than that calculated for the rrs sequences of staphylococci (98%). The rpoB, hsp60, and tuf partial sequences showed interspecific similarity values of 71.6 to 93.6, 74 to 93, and 86 to 97%, respectively.

The gap-based tree indicates the divergence of the selected staphylococcal species, which was well supported for most of the strains studied. We compared the gap-derived tree (Fig. 1) to those inferred from sequences of the 16S rRNA, rpoB, sodA, hsp60, and tuf genes of Staphylococcus species available from GenBank (Table 2). Earlier studies on the taxonomy of Staphylococcus species based on DNA-DNA reassociation indicated that in this genus there were eight distinct species groups, represented by S. epidermidis, S. saprophyticus, S. simulans, S. intermedius, S. hyicus, S. sciuri, S. auricularis, and S. aureus (17, 18). The same groups were identified in a study using hsp60 (20) and the sodA gene sequence analysis (28). The phylogenetic tree generated from rpoB sequences revealed nine clusters, including an additional S. haemolyticus group. The 16S rRNA sequence-derived trees with 38 taxa of the genus Staphylococcus identified 11 genogroups (S. epidermidis, S. saprophyticus, S. simulans, S. carnosus, S. hyicus/S. intermedius, S. sciuri, S. auricularis, S. warneri, S. haemolyticus, S. lugdunensis, and S. aureus) (32, 33). However, the bootstrap values for most of the nodes of the distinct clusters were low. With the gap sequences and a bootstrap value of >90%, the Staphylococcus species were divided into four well-supported clusters: the S. sciuri group, the S. hyicus/S. intermedius group, the S. haemolyticus/S. simulans group, and the S. aureus/S. epidermidis group.

In hsp60-, 16S rRNA-, sodA-, and rpoB-generated trees, the S. sciuri group was formed by the two members S. sciuri and S. lentus with bootstrap values of >97%. All of the species from the S. sciuri group, including S. vitulinus (not included in this study), differ from the other Staphylococcus species in several remarkable features. They are novobiocin resistant and oxidase positive, and they all share the same characteristic pattern of amino acid substitution in their hsp60 proteins (12, 20). The close relationship between S. sciuri and S. lentus was reproduced in the results from our tree analysis based on gap sequences. Thus, members of the S. sciuri group form a constant cluster in hsp60, 16S rRNA, sodA, rpoB, and gap gene trees.

The S. hyicus/intermedius group, as defined by 16S rRNA sequence analysis and confirmed by rpoB and hsp60 sequence analysis, includes S. hyicus, S. chromogenes, S. muscae, S. intermedius, S. delphini, S. schleiferi subsp. schleiferi, and S. felis. The S. intermedius group, as defined by sodA sequence analysis, consisted of S. intermedius and S. delphini (bootstrap value of 100%). S. schleiferi subsp. schleiferi and S. felis were not included in this cluster. Moreover, the related species S. hyicus, S. muscae, and S. chromogenes did not cluster to form an S. hyicus species subgroup. The gap-derived data support the idea that S. chromogenes and the non-S. aureus coagulase-positive staphylococci, such as S. intermedius, S. delphini, and S. hyicus, belong to the S. hyicus/S. intermedius species group. In our study, S. schleiferi subsp. schleiferi and S. muscae were outside of the S. hyicus/S. intermedius group as well. Based on the gap-derived data, S. schleiferi subsp. schleiferi and S. muscae could be grouped into the S. haemolyticus/S. simulans group. According to the 16S rRNA gene sequence, the phylogenetic classification of S. felis fell into the clade of S. hyicus/S. intermedius. However, based on DNA-DNA reassociation studies, S. felis was thought to be related to members of the S. simulans group (17, 18). In our gap-derived data, S. felis is outside of the S. hyicus/S. intermedius group and is closely related to members of the S. simulans group.

In contrast to the 16S rRNA sequence-based phylogenetic tree but in accordance with the sodA, hsp60, and rpoB gene sequences, we did not divide S. simulans and S. carnosus into two closely related groups by using the gap gene-derived sequences.

The S. saprophyticus group, as defined by 16S rRNA sequence analysis, includes the novobiocin-resistant and oxidase-negative species S. saprophyticus subsp. saprophyticus, S. arlettae, S. kloosii, S. cohnii, S. gallinarum, S. equorum subsp. equorum, and S. xylosus. The rpoB-derived data indicated that S. cohnii is outside of the S. saprophyticus group. From the sodA-derived data, one could conclude that the monophyly of this clade is uncertain, since it is associated with a bootstrap value of only 68%. In our study, S. cohnii and S. xylosus are clearly outside of the S. saprophyticus group. S. cohnii belongs to the S. saprophyticus group according to the 16S rRNA and hsp60 trees. Based on gap-derived sequences, the branching of S. auricularis, S. cohnii, and the heterogeneous S. saprophyticus group, comprising S. saprophyticus subsp. saprophyticus, S. equorum subsp. equorum, S. gallinarum, S. arlettae, and S. kloosii, was not reliable (98, 70, 43, and 36%, respectively).

Based on the 16S rRNA data, the S. epidermidis species group was divided into five cluster groups, as described by Kloos (16): S. lugdunensis, S. haemolyticus, S. warneri, S. epidermidis, and S. aureus (33). The S. epidermidis cluster, composed of S. epidermidis, S. capitis, S. caprae, and S. saccharolyticus (not included in our study), constitutes a monophyletic clade supported by a high bootstrap value (97%) on the basis of 16S rRNA sequence analysis (20). In the rpoB study, S. caprae and S. capitis appeared to be in the S. haemolyticus group. Similarly to the S. saprophyticus group, the S. epidermidis group did not form a clearly distinct lineage in the sodA-based study (bootstrap value of 38.9%). Similar results were obtained in our study using gap-based sequences. Moreover, in our study, S. caprae showed no close relationship to S. epidermidis or S. capitis. On the other hand, the association of S. warneri with the S. epidermidis group was inferred from our data as well as from the rpoB tree analysis.

Obviously, the determination of the sequences of several genes is an important tool for pathogen identification and phylogenetic studies. Although each gene-derived tree will differ from the others and will have different levels of statistical support, it has been found that groupings obtained with two different sequences with bootstrap values of >90% are stable and reliable (29). The data we present show that the sequence analysis of a 931-bp gap gene fragment is a suitable molecular method for the identification of Staphylococcus isolates at the species level.

The gap gene sequence-based relationships of the staphylococcal species obtained were in accordance with phylogenetic trees published previously (37). In support of the gap-derived tree results, the tuf gene sequence-derived tree indicates that S. warneri is associated with the S. epidermidis group, which includes S. capitis. This is in agreement with results described earlier for sequencing assays targeting the sodA gene (28). The tuf gene-derived data often showed more intraspecies sequence divergence than the 16S rRNA-derived data. Apparently, the 16S rRNA gene is more highly conserved than the tuf gene. A pairwise comparison of the tuf gene sequences revealed that their mean identity (92.6%) is lower than the mean identity (95.9%) of 16S rRNA gene sequences. These results indicate that the tuf gene constitutes a more discriminatory target gene than the 16S rRNA gene to differentiate closely related Staphylococcus species.

The phylogenetic analysis of the staphylococcal gap sequences yields an evolutionary tree having a topology similar to that of the tree constructed with the 16S rRNA sequences, although minor differences were observed (Fig. 1).

We have determined the gap sequences of 27 Staphylococcus type strains and demonstrated the usefulness of the gap GenBank database for distinguishing the staphylococcal species and giving an approach for interpreting the phylogenetic relationship of the staphylococci. This method consists of a PCR carried out with a single pair of degenerate oligonucleotides for the amplification of a staphylococcal partial gap gene sequence and the direct sequencing of the amplicon. The sequencing also can be performed with the two respective PCR primers instead of the eight primers, as was done in this study, since the sequencer is adjusted for sequence analysis. In our case, the sequencer also was used for terminal RFLP analysis, so that we had troubles with the sequencing of fragments of >800 bp to obtain a confident sequence properly. The methodology might be useful in reference laboratories for the characterization of strains that could not be assigned to a species on the basis of their conventional phenotypic reaction and can stand on its own more effectively than 16S rRNA analysis, as this is a highly conserved gene and has limited discriminatory power compared to that of the gap gene, especially in closely related staphylococcal species. Shortening the region of interest within the gap gene sequence was not possible due to the scattering of the divergent regions throughout the whole sequence. However, gap sequencing did not allow the detection of intraspecies polymorphism among the studied Staphylococcus species; e.g., for Staphylococcus epidermidis, the sequence similarity among different isolates of this species was more than 99%. This also was shown by the gap-based terminal RFLP analysis of S. epidermidis isolates in our previous publication (22).