First Description of Curtobacterium spp. Isolated from Human Clinical Specimens

MATERIALS AND METHODS

RESULTS

The clinical strains 2384, 3426, and 3430 grew heavily in pure culture and a moderate leukocyte reaction was observed in direct Gram stains of the clinical material suggesting a disease association in these three cases. For strains 1594 and 2340, a disease association was not so clear as these two strains were grown in mixed culture although they had been the predominant microorganisms.

All 15 strains studied grew as nonhemolytic, creamy, yellow- or orange-pigmented colonies of about 1 to 1.5 mm in diameter after 24 to 48 h of incubation. Colonies of C. pusillum tended to exhibit glistening and mucoid colonies whereas colonies of C. plantarum were significantly larger (>2 mm) than the colonies of all other strains. C. albidum, C. flaccumfaciens (all four pathovars except pathovar flaccumfaciens), C. herbarum, as well as strains 2340 and 2384 showed better growth at 30°C than at 37°C whereas similar growth at 30°C and at 37°C was observed for C. citreum, C. luteum, C. plantarum, C. pusillum, and strains 1594, 3426, and 3430. However, all strains included in the present study were able to grow at 37°C.

Gram stains of all strains showed relatively small coryneform bacteria (curtus, shortened) except for the C. plantarum strain which was a gram-negative rod. Most strains were motile and exhibited an oxidative metabolism except C. plantarum which was fermentative. Because of the discrepant results of the Gram stain and the oxidation/fermentation test of the C. plantarum type strain further biochemical identification was performed and resulted in the identification of this particular strain as Pantoea sp. Hence this strain was excluded from further analyses.

The remaining 14 strains were all catalase positive, did not reduce nitrate, exhibited no urease activity, but all very strongly hydrolyzed esculin. Acid was produced within 4 days from glucose by all strains except C. citreum and C. pusillum. Acid production from maltose was positive for C. flaccumfaciens, C. herbarum, C. pusillum, as well as strains 2340, 2384, 3426, and 3430. Sucrose was not acidified by C. citreum, C. albidum, and C. pusillum but all other strains. Acid production from mannitol was positive for C. flaccumfaciens pathovar flaccumfaciens, C. flaccumfaciens pathovar poinsettiae, C. herbarum, and strains 3426 and 3430 whereas xylose was acidified by all strains included in the present study. It is emphasized that the acid production was only very weak in comparison to other oxidative coryneform bacteria. All strains exhibited activity of the following enzymes: esterase, esterase lipase, leucine arylamidase, acid phosphatase (except C. herbarum and strain 3426), α-galactosidase (except C. pusillum and strains 2340 and 3430), β-galactosidase, α-glucosidase, β-glucosidase, and α-mannosidase. Activities of lipase (C₁₄), trypsin, and β-glucuronidase were not detected in any of the strains tested.

Table 2 outlines the antimicrobial susceptibility patterns of the nine Curtobacterium reference strains and the five clinical isolates. The 50% and 90% MICs of β-lactams were mostly greater than 1 μg/ml. Significantly, the MICs for macrolides (except azithromycin) and rifampin were ≤0.03 μg/ml for all strains tested. The MICs for amikacin and gentamicin were lower than for netilmicin and tobramycin as were the MICs for doxycycline and minocycline in comparison to tetracycline. The MICs of teicoplanin and vancomycin were lower than 2 μg/ml for every strain examined.

Cellular fatty acid analysis revealed that C_15:0ai and C_17:0ai presented more than 75% of all cellular fatty acids in every strain tested except two (Table 3). Much smaller amounts of C_15:0i, C_16:0i, and C_16:0 were also detected. The cellular fatty acid patterns of the strains studied were very similar with the exceptions of C. pusillum and strain 3430. For these particular two strains the Sherlock system named the major peak feature 7, consisting of C_{18:1ω7cis/ω9cis/ω12trans} which could not be separated by the system.

Analysis of the peptidoglycan structure of the five clinical isolates demonstrated ornithine as the diamino acid and that the interpeptide bridge consisted of ornithine alone. The acyl type of the peptidoglycan was found to be acetyl. This combination of chemotaxonomic features is found in the genus Curtobacterium only (Table 4). Therefore, the five clinical isolates were unambiguously identified as Curtobacterium species.

The data of the 16S rRNA gene sequencing of the five clinical strains are given in Table 5. The homologies of the 16S rRNA genes between the clinical strains and their closest phylogenetic neighbors, curtobacteria, were always greater than 99.1%, unambiguously demonstrating that the strains are true members of the genus Curtobacterium. With no mismatches to C. flaccumfaciens, strains 2384 and 3426 are most likely representatives of this species.

DISCUSSION

With a polyphasic approach the five clinical strains were identified as Curtobacterium species. Definitive identification on the species level was not attempted since the very few Curtobacterium strains described in the literature did not allow the creation of a reliable database for phenotypic differentiation on the species level.

Curtobacteria had never been described before as being recovered from human clinical specimens. Since curtobacteria have so far not been isolated from blood cultures or from other normally sterile body sites (except for strain 3430 in the present study) it is suggested that they act as colonizers rather than as invasive pathogens. In contrast, strains belonging to the species C. flaccumfaciens are well-established plant pathogens. For epidemiological purposes this species had been divided into four different pathovars whereas C. citreum, C. albidum, and C. luteum are not known to cause any disease on rice from which they were primarily isolated (14), and C. herbarum was isolated from grass (1). Other strains belonging to genera which are primarily associated with plants and can cause disease in those but may also colonize or infect humans include Pseudomonas, Stenotrophomonas, and Burkholderia.

Some curtobacteria with an optimal growth temperature of 30°C may not be isolated from human clinical specimens if the plates are routinely incubated at 37°C and read after 24 h only. Curtobacteria might be underdiagnosed because they are presently not included in the databases of commercial identification systems (11, 12) and chemotaxonomic investigations are necessary for complete identification of the organisms. However, it is the authors′ experience that curtobacteria belong to the least frequently encountered yellow- or orange-pigmented coryneform bacteria (see Table 4) in clinical specimens. For example, the oxidative Microbacterium/Aureobacterium strains are 20 to 25 times more frequently detected in clinical specimens than curtobacteria (G. Funke, unpublished observation).

Apart from chemotaxonomic investigations simple biochemical tests like rapidity of acid production from carbohydrates (microbacteria usually within 2 days whereas in curtobacteria it may take up to 1 week or even longer) may serve in the identification of yellow- or orange-pigmented coryneform bacteria. In fact, an orange pigment in coryneform bacteria other than curtobacteria is rarely seen (e.g., M. arborescens, M. imperiale, M. schleiferi, and M. testaceum exhibit an orange pigment). It should also be noted that all curtobacteria tested until now are nitrate reductase and urease negative but strongly hydrolyze esculin (see Table 4).

C. pusillum is the only established Curtobacterium species which was not primarily isolated from plants. It is also very unusual in its characteristic that the majority of cellular fatty acids are identified as C_{18:1ω7cis/ω9cis/ω12trans} by the Sherlock system and as ω-cyclohexyl undecanoic acid by another independent system (17). This discrepancy can be resolved by determining the precise cellular fatty acid structure by mass spectrometry. The authors are not aware of any gram-positive genus which includes both ω-cyclohexyl-containing and non-ω-cyclohexyl-containing bacteria. Some ω-cyclohexyl-containing Bacillus species had recently been transferred into a separate genus, Alicyclobacillus (20).

C. plantarum (strain ATCC 49174) is certainly not a member of the genus Curtobacterium but a Pantoea strain. Since strain ATCC 49174 is the only strain that has ever been deposited (Phyllis Pienta, personal communication) the present paper raises the pursuit for a true C. plantarum strain as described by Dunleavy (3).

In summary, Curtobacterium strains are rarely isolated from clinical samples but clinical microbiologists should be aware of the possible appearance of these bacteria in material from humans although their pathogenicity is considered rather low since many people are probably exposed to curtobacteria every day. The tests outlined should facilitate the diagnosis of Curtobacterium spp. in the routine clinical laboratory. Finally, as with other recently described yellow-pigmented coryneform bacteria, it is expected that once a genus appears in the clinical microbiology literature, other workers will also find strains belonging to this particular genus in their specimens.