ABSTRACT
CDC coryneform group A-3 bacteria are rare human pathogens. In this study, six group A-3 isolates (two from blood, one from cerebrospinal fluid, and one each from homograft valve, lip wound, and pilonidal cyst) were compared to the type strains of phenotypically related organisms, Cellulomonas fimi, Cellulomonas hominis, Oerskovia turbata, and Sanguibacter suarezii, and characterized by phenotypic, chemotaxonomic, and genotypic studies. DNA-DNA reassociation analysis identified two genomic groups, and phylogenetic analysis of the 16S rRNA gene sequence identified the taxonomic positions of these groups to genus level. Two groups were defined, and both were more closely related to Cellulomonas species: one group of three strains, for which we propose the new species Cellulomonas denverensis sp. nov., with the type strain W6929 (ATCC BAA-788T or DSM 15764T), was related to C. hominis ATCC 51964T (98.5% 16S rRNA gene sequence similarity), and the second group of three strains was related to C. hominis ATCC 51964T (99.8 to 99.9% 16S rRNA gene sequence similarity). The definition of this new Cellulomonas species and the confirmation of three strains as C. hominis serve to further clarify the complex taxonomy of CDC coryneform group A-3 bacteria and will assist in our understanding of the epidemiology and clinical significance of these microorganisms.
Since 1965, the Centers for Disease Control and Prevention (CDC) Special Bacteriology Reference Laboratory (SRBL) has received eight unidentified gram-positive, rod-shaped bacteria that appeared on the basis of routine microbiologic tests to form a distinct group, which has been designated CDC coryneform group A-3. Group A-3 strains are motile, fermentative, pale yellow to yellow pigmented and have not been recognized as belonging to any established species. Six of these strains from human clinical sources were included in a 1977 study of Sottnek et al. (14). The increase in the interest in these bacteria as serious opportunistic pathogens of humans was stimulated when Hollis and Weaver (5) recognized and described many CDC coryneform groups among clinical isolates, including the coryneform group A-3. In a recent report, Funke et al. (4) assigned two of the coryneform group A-3 strains to the genus Cellulomonas as a new species, Cellulomonas hominis, on the basis of phenotypic characteristics and 16S rRNA gene sequencing. In this study, we present the phenotypic and genotypic comparison of the remaining six strains of coryneform group A-3 bacteria with C. hominis, as well as with the type strains of other related species and genera, Cellulomonas fimi, Oerskovia turbata, and Sanguibacter suarezii. From the results of our phenotypic and genotypic data, we propose a new clinically relevant species of Cellulomonas, Cellulomonas denverensis sp. nov., and establish its close association with C. hominis.
(This work was presented in part at the General Meeting of the American Society for Microbiology, Washington, D.C., 18 to 22 May 2003.)
MATERIALS AND METHODS
Strains.The clinical isolates used in this study were six strains of CDC group A-3 (Table 1). They were obtained from the collection of CDC's SBRL and Actinomycete Reference Laboratory. The reference strain of C. hominis (DMMZ CE39) was provided by Guido Funke, Department of Medical Microbiology and Hygiene, Gartner and Colleagues Laboratories, Weingarten, Germany. The type strains of C. fimi ATCC 484, C. hominis ATCC 51964, Oerskovia turbata ATCC 25835, and Sanguibacter suarezii ATCC 51766 were obtained from the American Type Culture Collection (Manassas, Va.). All strains were maintained at 4°C on Middlebrook 7H11 agar slants (Remel, Lenexa, Kans.) until use.
Isolates used in this studya
DNA reassociation studies.Harvesting and lysis of the bacterial cells were performed by previously described methods (7). DNA relatedness experiments utilized the hydroxyapatite method described previously (2). DNA was labeled in vitro with [32P]dCTP by using a nick translation kit (Invitrogen, Carlsbad, Calif.). The temperature used for optimal hybridization was 70°C, and the percent divergence was calculated to the nearest 0.5% (2).
16S rRNA gene sequencing.The 16S rRNA genes of strains C. denverensis W6929T (ATCC BAA-788T), W6124, and W6117 and C. hominis ATCC 51964T, DMMZ CE39, W7335, W7336 (ATCC BAA-786), and W7387 were analyzed as described by McNeil et al. (7). Related sequences were identified in a BLAST search against GenBank. Similarity searches were performed with Clustal W, and a distance matrix was created. In Treecon, the phylogenetic tree of aligned sequences was constructed with the neighbor-joining method and bootstrapped based on 1,000 replications (7).
CFA and quinone analysis.Bacteria were grown for 2 days on heart infusion agar with 5% rabbit blood at 37°C. Harvested cells were saponified, and the liberated fatty acids were methylated and analyzed by gas-liquid chromatography (16). Identification of fatty acids was performed using a commercially available software package (MIDI, Newark, Del.). Isoprenoid quinones were extracted from a 100-mg portion of lyophilized cells and analyzed by reverse-phase high-performance liquid chromatography (9). Cellular fatty acid (CFA) profiles were identified using a commercially available system (MIDI, Newark, Del.) utilized with a CDC library created using LGS software (16).
Whole-cell analyses.The methods used for whole-cell analyses for diaminopimelic acid and monosaccharides are those described by Berd (1).
G+C content in DNA.The G+C content was determined spectrophotometrically by thermal denaturation as previously described (6). Escherichia coli DNA was used as a control, and all samples were run at least three times.
Phenotypic characterization.All isolates were inoculated onto heart infusion agar with 5% rabbit blood (BBL, Becton Dickinson, Microbiology Systems, Cockeysville, Md.) and incubated at 25 and 35°C for 2 days for morphological studies. We used Gram staining to study microscopic morphology, and flagellum staining was used to demonstrate the presence and the site of attachment of flagella. The isolates were examined at low power (magnification, ×10) under a stereomicroscope for the presence of aerial and substrate hyphae. We conducted biochemical tests that are routinely used in the SBRL using previously described methods (16) except decomposition of casein at 25 and 35°C was done as described by Berd (1). Antimicrobial susceptibilities were determined by a previously described broth microdilution method with cation-supplemented Mueller-Hinton broth (10, 11). The antimicrobial agents tested were amikacin, amoxicillin-clavulanate, ampicillin, ceftriaxone, ciprofloxacin, clarithromycin, clindamycin, doxycycline, imipenem, minocycline, rifampin, trimethoprim-sulfamethoxazole, and vancomycin. The plates were incubated at 25 or 35°C for 48 h. Currently, there are no interpretive MIC breakpoints for isolates of this genus.
Nucleotide sequence accession numbers.The GenBank accession numbers for the 16S rRNA gene sequences of C. denverensis ATCC BAA-788T, W6124, and W6117 are AY501362, AY655726, and AY655727, respectively. The GenBank accession numbers for the 16S rRNA gene sequences of C. hominis W7387, ATCC BAA-786, W7335, DMMZ CE39, and ATCC 51964T are AY655732, AY655731, AY655730, AY655729, and AY655728, respectively.
RESULTS AND DISCUSSION
The six clinical strains (Table 1) studied were isolated from blood (n = 2), cerebrospinal fluid (n = 1), homograft valve, lip wound, and pilonidal cyst. These strains were received from reference laboratories in the United States and Canada. Except for C. denverensis ATCC BAA-788T, a blood isolate from a patient with endocarditis, and W6124, from a homograft valve associated with the same patient, each at the University of Colorado Medical Center (7, 12), little clinical information was available to determine the clinical significance of the remaining strains.
A species is defined as a group of strains that exhibits levels of DNA relatedness of 70% or more at the optimal reassociation temperature and whose related sequences exhibit 5% or less divergence (15). A strain is assigned to a particular species when the relatedness of its DNA to labeled DNA from the type strain of that species fulfills the species definition. The DNA relatedness results for the six coryneform A-3 isolates summarized in Table 2 were consistent with the results of 16S rRNA gene analysis (Fig. 1). Three strains were found to belong to a new species, Cellulomonas denverensis, and three strains were confirmed as C. hominis.
Phylogenetic tree based on 16S rRNA sequences showing the position of Cellulomonas denverensis strains (ATCC BAA-788T, W6124, and W6117) and Cellulomonas hominis strains (W7387, ATCC BAA-786, W7335, DMMZ CE39, and ATCC 51964T). The tree was rooted by using Sanguibacter suarezii as the outgroup. The scale bar represents a 2% difference in sequence.
DNA relatedness of representative strains in this study
When we labeled the DNA of the reference and type strains of the nearest phylogenetic neighbors, C. hominis ATCC BAA-786 and C. fimi ATCC 484T, we found the levels of relatedness to C. denverensis ATCC BAA-788T to be low (33 and 9%, respectively) (Table 2). The labeled DNA of C. denverensis ATCC BAA-788T was related to that of the other coryneform group A-3 isolates, with values ranging from 9 to 32%, with divergences ranging from 11 to 12%, confirming it as a new species. Labeled DNA of C. denverensis ATCC BAA-788T was related 100%, with a divergence of 1.0% to W6124 (a homograft valve isolate associated with the same patient) (7), and was related 83%, with a divergence of 2.0%, to W6117. The high levels of DNA relatedness demonstrated that these isolates belong to the same species, and the low levels of 16S rRNA gene sequence similarity to other taxa support the distinctness of these isolates (Fig. 1).
The labeled DNA of C. hominis ATCC 51964T (Table 2) was related 100%, with a divergence of 0.5%, to C. hominis DMMZ CE39 and was related 85%, with a divergence of 1.0%, to C. hominis ATCC BAA-786. In a reciprocal reaction, C. hominis ATCC BAA-786 was 98% related to C. hominis ATCC 51964T, with a divergence of 1.0%. We obtained high DNA-DNA relatedness values (92 and 95%) for the two remaining strains of C. hominis when compared with labeled C. hominis ATCC BAA-786. These relatedness experiments were consistent with the results of 16S rRNA gene sequence analysis since the percentages of similarity were close (ranging from 99.8 to 99.9%) (Fig. 1). In addition, the labeled ATCC BAA-786 did not show high reassociation levels with the type strain of C. denverensis (Table 2). These results indicate that these strains constitute a separate taxon.
Chemotaxonomic characteristics.Predominant fatty acids of Cellulomonas species, along with S. suarezii and O. turbata, are listed in Table 3. In contrast to the genetic differences of the six coryneform group A-3 strains, chemotaxonomic characteristics, CFA and quinone analyses, were not useful in separating C. denverensis from related species. These organisms shared a common cellular fatty acid profile characterized by predominant amounts of 14:0, i15:0, a15:0, i16:0, 16:0, and a17:0. In addition, these organisms shared similar respiratory quinone profiles characterized by major amounts of menaquinone-9 (MK-9) and minor amounts of MK-7 and MK-8.
Diagnostic chemotaxonomic properties of type strains and representative isolates in this study
Although all group A-3 isolates shared common cell wall structures, including the lack of diaminopimelic acid in the whole-cell wall analysis, this group was heterogeneous with respect to sugar composition (Table 3). These data suggest that examining whole-cell sugar composition may be of practical value in species identification: C. denverensis possesses mannose, rhamnose, and ribose, and C. hominis possesses mannose, fucose, and rhamnose.
Results of G+C content for representative coryneform A-3 strains and related species ranged from 68.0% for S. suarezii ATCC 51766T to 71.0 mol% for C. fimi ATCC 484T (Table 3). These results were consistent with those obtained with the denaturation method in other studies (3, 13). However, results for G+C content found for C. hominis in our study were significantly lower, at 70.0 mol%, than the results reported by Funke et al. (4). This discrepancy may have resulted from differences in methodologies, G+C content detected by high-performance liquid chromatography (8) versus detection determined with the denaturation method described by Mandel et al. (6).
Phenotypic analysis.Microscopic morphological studies showed that all isolates studied were gram-positive, pleomorphic bacilli, lacking spores and capsules. The isolates, under low-power stereomicroscopic examination, showed pale yellow to yellow, small-diameter (approximately 1-mm) colonies after incubation on heart infusion agar with 5% (vol/vol) rabbit blood for 2 days at 35°C. No substrate hyphae were seen. All colonies were smooth, convex, and entire edged. All isolates studied were motile with lateral and polar flagella, reduced nitrate, and hydrolyzed esculin. All coryneform group A-3 isolates were negative for gelatin liquefaction in 14 days and urease production and produced acid fermentatively from cellobiose, d-galactose, d-glucose, lactose, maltose, mannose, salicin, sucrose, d-trehalose, and d-xylose. None produced acid from adonitol, dulcitol, i-erythritol, i-myo-inositol, and d-mannitol. The biochemical differences among these isolates and type strains of related organisms are summarized in Table 4.
Differential characteristics of study isolates compared with type strains of related organismsa
The antimicrobial susceptibility results are given in Table 5. Clarithromycin, clindamycin, imipenem, minocycline, rifampin, and vancomycin appear to be active against the six coryneform group A-3 isolates and the reference strains of C. hominis, C. fimi, S. suarezii, and O. turbata. The MICs of ciprofloxacin were high against C. denverensis, and the MICs of amoxicillin-clavulanic acid, ampicillin, and ceftriaxone had intermediate values against isolates of C. denverensis. The MICs of amikacin were high for all isolates studied. The trimethoprim-sulfamethoxazole MICs ranged from 0.25/4.8 μg/ml for O. turbata ATCC 25835 to >8/152 μg/ml for all C. hominis isolates. The results of the MICs of the type and reference strains of C. hominis, as well as the type strains of C. fimi and O. turbata, with the agar dilution method have been reported (4). Except for resistance of O. turbata to rifampin, results were comparable to our results.
MICs of 12 antimicrobial agents for Cellulomonas denverensis, Cellulomonas hominis, and type strains of related species
The new species, Cellulomonas denverensis, a group of three strains, was phenotypically and phylogenetically (98.5% by 16S rRNA gene sequencing) most similar to C. hominis but differed by fermentation of d-sorbitol.
Cellulomonas denverensis sp. nov.(den.ver.en′sis, N.L. adj. denverensis of Denver, Colo., the city of origin of the type strain). Cells are short (1 μm), thin, gram-positive, nonsporulating rods that are motile by polar and lateral flagella. Colonies are circular, smooth, and convex, and are pale yellow in about 3 days. Cells are fermentative. Growth occurs at 35 and 45°C but not at 25°C. Cells are catalase positive. Esculin is decomposed, while casein, gelatin, and urea are not. Nitrate is reduced to nitrite. Acid is produced from cellobiose, d-galactose, d-glucose, lactose, maltose, mannose, l-rhamnose, salicin, d-sorbitol, sucrose, d-trehalose, and d-xylose, and sometimes from l-arabinose (2/3), glycerol (2/3), and melibiose (1/3). No acid is produced from adonitol, dulcitol, i-erythritol, i-myo-inositol, d-mannitol, melezitose, or raffinose. The diagnostic whole-cell sugars are mannose, ribose, and rhamnose. G+C content is 68.5 mol%. The GenBank accession numbers of the 16S rRNA gene sequences of ATCC BAA-788T, W6124, and W6117 are AY501362, AY655726, and AY655727, respectively. The type strain is W6929 (ATCC BAA-788 T or DSM 15764 T) from blood.
ACKNOWLEDGMENTS
We thank Deanna Jannat-Khah for assistance with the chemical analysis.
FOOTNOTES
- Received 25 June 2004.
- Returned for modification 3 September 2004.
- Accepted 20 December 2004.
