Corynebacterium Species Isolated from Bone and Joint Infections Identified by 16S rRNA Gene Sequence Analysis

Abstract

By the use of 16S rRNA gene sequence analysis we identified 28 of 31 Corynebacterium spp. isolated from bone and joint infections, including species never before isolated in such infections. Phenotypic analysis led to the correct identification of 8 of 31. 16S rRNA gene sequence analysis appears to be a good technique for identification of clinical strains of Corynebacterium spp.

Staphylococcal infections account for most bone and joint infections (3-5, 12). However, coryneform bacilli have been found in up to 10% of cases (24, 25). Besides surgical and conventional treatments (including long-term antibiotic therapy) (3, 4, 12), accurate identification of bacterial isolates has so far been an essential task for clinical microbiology laboratories (6, 7, 23, 24). In addition to the difficulties encountered in the determination of the clinical significance of Corynebacterium species, phenotypic identification of these bacteria remains routinely problematic. Thus, at this time it is difficult to evaluate particular bacterial species that have a specific potential to produce orthopedic-material-associated infections or to know what is needed for specific treatments (such as different durations of antibiotic therapy or the need of material removal). Moreover, recent identification of several new taxa and the increasing diversity of coryneform bacterial strains encountered in clinical specimens render phenotypic identification more difficult (1, 11, 13, 26, 27). Consequently, genotypic identification as an alternative or complementary method for bacterial taxonomy and for identification of new species (including Corynebacterium spp.) has emerged during the last few years (16, 19, 22). In the present study, we determined almost the complete 16S rRNA gene sequence for selected 31 isolates belonging to the genus Corynebacterium recovered from patients suffering from bone and joint infections.

A total of 31 patients with clinical, biological, and radiological evidences of either acute or chronic joint or bone infection with or without the presence of an orthopedic implant (orthopedic implants included prostheses and osteosynthetic plates) were included in this study (Table 1). Samples were classified as superficial samples or deep samples. Superficial samples were those collected from patients with a fistula. Pus was swabbed or needle aspirated. Deep samples were collected by needle aspiration (when a liquid collection was present) or by surgical biopsy (taken from infected tissues other than fistulas). Isolated Corynebacterium spp. were considered pathogenic when at least one of the following criteria was met: (i) in cases of superficial samples, isolation at least three times in samples taken at three different times and the presence of polymorphonuclear neutrophils on Gram staining, (ii) isolation from a deep sample, and (iii) in any kind of sample, the presence of gram-positive bacilli within polymorphonuclear neutrophils on Gram staining.

TABLE 1.

Characteristics of 31 patients included in this study, with results of phenotypic and genotypic identification of the Corynebacterium isolates

Patient	Sexa	Age (yr)	Infection	Samplec	Associated bacteria	Phenotypic identification	16S rRNA identification	Similarity (%)
1	M	63	Tibia osteitis	NA	None	Corynebacterium sp.	C. jeikeium	99
2	M	40	Foot osteitis	NA	None	Corynebacterium sp.	C. jeikeium	99.5
3	M	35	Foot osteitis	NA	None	Corynebacterium sp.	C. aurimucosum	99.5
4	M	81	Pelvis osteitis	F	Staphylococcus aureus, Esche- richia coli, Pseudomonas fluorescens	C. striatum	C. striatum	99.7
5	F	80	Hip prosthesis	NA	None	Corynebacterium sp.	C. aurimucosum	99.7
6	M	72	Knee prosthesis	B	None	Corynebacterium sp.	C. amycolatum	99.3
7	M	21	Elbow prosthesis	NA	None	Corynebacterium sp.	C. amycolatum	99
8	M	72	Hip prosthesis	F	None	Corynebacterium sp.	C. jeikeium	99.5
9	F	52	Osteitis	F	None	Corynebacterium sp.	C. nigricans/C. aurimucosum	99.6/99.3
10	F	68	Knee osteitis	NA	None	C. amycolatum	C. amycolatum	99.4
11	F	76	Foot osteitis	NA	Acinetobacter baumannii	C. striatum	C. striatum	99.9
12	F	68	Ankle osteitis	F	None	Corynebacterium sp.	C. amycolatum	99.2
13	M	71	Foot osteitis	NA	None	Corynebacterium sp.	C. pseudodiphteriticum	99.8
14	M	30	Tibia OSMRb	NA	None	C. jeikeium	C. jeikeium	99.6
15	F	58	Ankle osteitis	NA	Streptococcus sp.	Corynebacterium sp.	C. propinquum	99.5
16	F	25	Calcaneum osteitis	B	None	C. striatum	C. striatum	99.7
17	F	30	Foot osteitis	NA	Staphylococcus aureus	Corynebacterium sp.	C. striatum	99.6
18	M	78	Tibia osteitis	NA	None	Corynebacterium sp.	C. striatum	99.6
19	F	76	Hip prosthesis	NA	None	Corynebacterium sp.	C. aurimucosum	99.5
20	M	70	Tibia osteitis	NA	Acinetobacter baumannii	C. macginleyi	C. striatum	99.9
21	F	32	Pelvis osteitis	NA	None	Corynebacterium sp.	C. aurimucosum	99.6
22	M	82	Hip prosthesis	NA	Staphylococcus epidermidis	Corynebacterium sp.	C. nigricans/C. aurimucosum	99.8/99.6
23	F	74	Foot osteitis	NA	Enterobacter cloacae, Proteus mirabilis, Staphylococcus epidermidis	C. striatum/C. amycolatum	C. aurimucosum	100
24	M	54	Pelvis osteitis	NA	None	Corynebacterium sp.	C. nigricans/C. aurimucosum	99.7/99.5
25	F	80	Hip prosthesis	NA	None	Corynebacterium sp.	C. striatum	99.6
26	F	62	Ankle OSMR	NA	None	Corynebacterium sp.	C. aurimucosum	99.2
27	M	42	Foot osteitis	A	None	C. jeikeium	C. jeikeium	99.4
28	M	58	Tibia osteitis	NA	None	Corynebacterium sp.	C. jeikeium	99
29	F	77	Knee prosthesis	B	None	C. striatum	C. striatum	99.8
30	M	54	Foot osteitis	NA	Staphylococcus aureus, Enterococcus faecalis	Corynebacterium sp.	C. striatum	99.6
31	F	83	Osteitis	NA	None	C. pseudodiphteriticum	C. pseudodiphteriticum	100

^aM, male; F, female.

^bOSMR, osteosynthetic material removal.

^cNA, needle aspirate; F, fistula sample; B, biopsy.

Identification was performed by Gram staining and catalase activity determination and by using an API CORYNE system (version 2.0) (BioMerieux) (9, 10). 16S rRNA gene determination was performed as previously described (14). The sequences determined were compared with those available in the GenBank database with BlastN software (http://www.ncbi.nlm.nih.gov/BLAST/).

The results of the 16S rDNA sequence analyses were in accordance with the phenotypic identification given by an API Coryne system in 8 of 31 strains (Table 1). Two strains, C. aurimucosum and C. striatum, were misidentified. Among the remaining 21 unidentified strains, 18 were identified by the genomic approach. For three strains, the best match was obtained with C. nigricans (1, 20). The second-best match was C. aurimucosum. Our three isolates did not produce black pigment. On the basis of 16S rDNA sequence analysis, overlapping of positions for C. aurimucosum and C. nigricans strains has been observed (21). Further studies (including DNA-DNA hybridizations and large collections of strains) will have to be done to ensure that this black pigmentation is a constant feature for the characterization of C. nigricans species. Traditional phenotypic identification of Corynebacterium isolates is difficult and time consuming; when phenotypic methods are used to identify these isolates, interpretation of test results can involve substantial subjective judgment (11). Most of the systems used (such as the API system) need to combine these phenotypic systems with individual tests (10, 11). Performing additional tests is not well adapted to routine work in large clinical microbiology laboratories. Variations in results occurring with variations in sizes of inoculum, media used for culture, lipid requirements, and unusual phenotypes of some isolates can sometimes lead to unreliable results when tests are performed by microbiologists who are not expert in the field (1, 18). The clinical significance of most Corynebacterium isolates remains frequently questionable; thus, their identification to the species level (with the exception of highly pathogenic species such as C. diphteriae) may not appear useful in routine use. It is likely that identification of coryneform bacteria to the species level would be useful for distinguishing sources of contamination, colonization, or infection, however, thereby determining the method of clinical intervention.

The presence of C. amycolatum in infections following prosthetic joint infection and open fractures has been previously reported (25). Similarly, the presence of C. striatum in infections following prosthetic joint infection and open fractures (25) and vertebral osteomyelitis (8) has been previously reported. Likewise, the presence of C. jeikeium in infections following prosthetic joint infection and open fractures (25), total knee arthroplasty infection (28), and osteomyelitis without a foreign body (2) was also previously reported. In this study, C. aurimucosum represented the second most frequently isolated species after C. striatum. As its identification is difficult and its description is recent (27), its pathogenic implication in human infections might be underestimated at this time. C. propinquum and C. pseudodiphteriticum, two phylogenetically related species, have been mostly recovered from the respiratory tract. In this study, C. pseudodiphteriticum was isolated from two deep samples. It was previously isolated from one bone infection (25). As eight of our isolates were part of mixed floras (Table 1), their role in infection remains a question.

Classification and identification of coryneform bacteria has been significantly improved by incorporating 16S rRNA gene sequence analysis (16, 19), but this gene lacks polymorphism for accurate identification of corynebacteria (16, 17; A. Khamis, D. Raoult, and B. La Scola, submitted for publication). A novel approach (taken on the basis of the delineation of cutoff for sequence similarity of genes more polymorphic than 16S rDNA) has been recently proposed for species definition in genera (such as Bartonella and Corynebacterium) not well identified by 16S rRNA gene comparison, but this approach needs validation when large collections of isolates are tested (15; Khamis et al., submitted). In this study, however, 16S rRNA gene sequence comparison was effective for identification of 28 of 31 coryneform isolates. The usefulness of this approach is hampered by the poor quality of some sequences deposited in databases that may cause misidentification when interpreted by inexperienced staff. The low level of polymorphism of the 16S rRNA gene for the determination of complete sequences for accurate identification that implies should be avoided in many cases by using partial rpoB gene sequencing (Khamis et al., submitted).