TEXT

Top Abstract Text References

The genus Gemella currently comprises the species G. haemolysans, G. morbillorum, G. bergeriae, and G. sanguinis (2-4). While there is limited information on the natural occurrence of G. bergeriae and G. sanguinis, both G. haemolysans and G. morbillorum are part of the human resident microbiota of several mucosal surfaces (1, 4). Both species have been detected in the oropharynx, and G. morbillorum has, in addition, been detected in the gastrointestinal tract. The clinical significance of Gemella species is not entirely clear, though all four species have been isolated from blood of patients with subacute endocarditis. In addition, G. haemolysans and G. morbillorum have been cultured from cerebrospinal fluid of patients with meningitis and from infected eyes, G. haemolysans has been cultured from urine, and G. morbillorum has been cultured from bone and synovial fluid in a patient with septic arthritis (1-4).

Identification of Gemella isolates is notoriously difficult because of the easy decolorization in the Gram stain, the low growth rate, and the still not well defined biochemical characteristics of the individual species. This problem has been enhanced by the recognition of several related genera of gram-positive cocci, mainly based on distinct 16S rRNA sequences.

Members of the genus Gemella share many physiologic and biochemical properties with the viridans group streptococci, including the range of infections that they cause (4). Several commensal Streptococcus species, i.e., Streptococcus sanguis, Streptococcus oralis, Streptococcus mitis biovar 1, and important mucosal pathogens such as Neisseria meningitidis, Neisseria gonorrhoeae, Haemophilus influenzae, and Streptococcus pneumoniae have evolved specific immunoglobulin A1 (IgA1) proteases. By cleaving human IgA1 in the hinge region into monomeric Fab fragments that are capable of antigen binding but devoid of antibody effector functions, these proteases allow bacteria to evade the protective functions of the principal adaptive immune factor at mucosal surfaces (6, 10). A previous comprehensive screening of bacteria revealed IgA1-cleaving activity in two strains of G. haemolysans (7). However, apart from this preliminary observation, the occurrence and nature of this enzyme activity in members of the genus Gemella as presently defined and in more recently described, related taxa have never been examined.

Twenty-two Gemella strains originally isolated from humans were included in the study. Of these 19 were obtained from the Culture Collection of the University of Göteborg, Göteborg, Sweden: G. haemolysans strains CCUG 411, CCUG 4815, CCUG 28134, CCUG 33602, CCUG 37278, and CCUG 37985^T; G. sanguinis strains CCUG 37820^T, CCUG 24073, CCUG 37821, and CCUG 37970; G. morbillorum strains CCUG 15561, CCUG 18164^T, CCUG 18165, CCUG 32414, and CCUG 32957B; G. bergeriae strains CCUG 31456, CCUG 37817^T, CCUG 37818, and CCUG 37968. An additional three dental plaque isolates of G. haemolysans were from our own collection (SK940, SK912, and SK891). The identity of the final three isolates was verified by 16S rRNA gene sequence analysis (9). In addition, reference strains of the related catalase-negative, gram-positive cocci Globicatella sanguinis (strains CCUG 33367 and CCUG 36563), Helcococcus kunzii (strain CCUG 32213^T), and Facklamia hominis (strain CCUG 36813^T), were examined. S. sanguis strain SK1, Streptococcus pneumoniae strain 1416-93, and isolated IgA1 proteases of H. influenzae strains HK393 (cleavage type 1) and HK224 (cleavage type 2) served as references of different IgA1 protease cleavage specificities.

All strains of gram-positive cocci were propagated on 5% horse blood agar incubated in air plus 5% CO₂ and were biochemically characterized using the API-ZYM system (API bioMérieux, Marcy l'Etoile, France) supplemented with tests for pyrrolidonyl aminopeptidase activity, acetoin production (Voges-Proskauer [VP] test), and neuraminidase activity. Acid production from raffinose, mannitol, sorbitol, sucrose, lactose, maltose, and trehalose was examined in fluid medium. All tests were performed as previously described (8). IgA1 protease activity was detected by suspending several colonies of each strain in 100 µl of substrate solution containing 2.1 µg of myeloma IgA1 or IgA2m(1) in phosphate-buffered saline (PBS). After incubation for 20 h at 37°C bacteria were removed by centrifugation at 10,000 × G for 10 min, after which the supernatant was boiled for 5 min with reducing sample buffer before electrophoresis on a sodium dodecyl sulfate (SDS)-polyacrylamide gradient (4 to 20%) gel. Protein bands were visualized using 0.1% (wt/vol) Coomassie blue.

IgA1 protease activity was exclusively detected among the nine reference strains of G. haemolysans, including the type strain. Cleavage of IgA2m(1) was not observed for any of the bacterial strains. As illustrated in Fig. 1, SDS-polyacrylamide gel electrophoresis showed Fc and Fd fragments generated by the IgA1 protease of G. haemolysans that were indistinguishable from those produced by the S. sanguis IgA1 protease, suggesting the site of cleavage in the IgA1 hinge region is the peptide bond Pro₂₂₇-Thr₂₂₈, which is the target for the streptococcal enzyme. The similarity between IgA1 proteases of these species is further supported by the demonstrated inhibition of the G. haemolysans IgA1 protease by the metal chelator EDTA, which corresponds to a metallo-type proteinase (5).

View larger version (31K): [in this window] [in a new window]   FIG. 1.   SDS-polyacrylamide gel electrophoresis of human IgA1 and IgA2 myeloma proteins before and after incubation for 24 h with G. haemolysans or reference IgA1 proteases. Lane 1, molecular weight standards (in thousands); lane 2, intact IgA1 control; lane 3, IgA1 incubated with G. haemolysans cells; lane 4, PBS incubated with G. haemolysans cells; lane 5, IgA1 incubated with S. pneumoniae cells; lane 6, PBS incubated with S. pneumoniae cells; lane 7, IgA1 incubated with S. sanguis cells; lane 8, IgA1 incubated with cleavage type 1 IgA1 protease from H. influenzae HK393; lane 9, IgA1 incubated with cleavage type 2 IgA1 protease from H. influenzae HK224; lane 10, IgA1 incubated with G. haemolysans cells in the presence of 100 mM EDTA; lane 11, IgA1 incubated with S. pneumoniae cells in the presence of 100 mM EDTA; lane 12, IgA2 incubated with G. haemolysans cells; lane 13, IgA2 incubated with S. pneumoniae cells and 100 mM EDTA; lane 14, intact IgA2; lane 15, molecular weight standards. Lane 3 demonstrates that G. haemolysans cleaves IgA1 to yield Fc and Fd fragments identical in size to those released by S. sanguis (lane 7) and distinct from those released by S. pneumoniae (lane 5) and the two cleavage types of H. influenzae (lanes 8 and 9). Note that Fd fragments released by the protease activity of G. haemolysans and S. sanguis are close to the size of IgA1 light chains (L chain). The activity is inhibited by EDTA (lane 10). The distinct size of Fc fragments released from IgA1 by S. pneumoniae (lane 5), although cleaving the same peptide bond as S. sanguis, is due to glycosidase activities releasing carbohydrate side chains of the heavy chain (8). None of the bacteria cleave IgA2 (lanes 12 and 13).

All strains of the four Gemella species were alpha-hemolytic, produced acid in glucose broth, and were positive for esterase lipase C8, naphthol-AS-BI-phosphohydrolase, and pyrrolidonyl aminopeptidase activity, with the exception of G. haemolysans strain SK891. None of the isolates were positive for valine arylamidase, trypsin, -galactosidase, -galactosidase, -glucoronidase, -glucosidase, -glucosidase, N-acetyl--glucosaminidase, -mannosidase, -fucosidase, or neuraminidase. Other biochemical activities are summarized in Table 1. As illustrated by the table, G. haemolysans could be distinguished from all other species by the ability to cleave IgA1. Most of the biochemical results shown in Table 1 are in agreement with those of other studies (1, 4). However, some reactions are at variance with those reported by Collins et al. (2, 3), in spite of the fact that the majority of strains are common to the two studies. While eight of nine strains of G. haemolysans examined by us were positive in the VP test in agreement with results reported by Berger (1), Collins and coworkers (3) reported this reaction to be negative. Interestingly, the one negative exception in our study was the type strain. Likewise, Collins et al. reported no acid production from sorbitol by G. bergeriae, while three of four strains were positive in our study, and no acid production by G. sanguinis from raffinose, while three of four strains were positive in our study. These discrepant results are likely to be due to different methods (conventional tube fermentation tests in this study in contrast to dehydrated substrates in commercial kits in the study reported by Collins et al. [2, 3]) and emphasize the problems associated with phenotypic differentiation of members of this group of bacteria. Thus, the test for IgA1 protease activity is a valuable differential test for G. haemolysans and G. morbillorum as it is for differentiation of some of species of the mitis group of streptococci (8).

Although phylogenetically related, members of the genus Gemella constitute a distinct branch separate from the genus Streptococcus (12). The finding that their IgA1 proteases are metallo-proteinases and appear to cleave the same peptide bond shows that they belong to the same family of IgA1 proteases. Although genetic studies are required to elucidate this question, it is conceivable that the gene encoding IgA1 protease in G. haemolysans shares a common ancestor with the mutually related iga genes of S. sanguis, S. oralis, S. mitis, and S. pneumoniae (11). Although IgA1 protease enables bacteria to evade the adherence-inhibitory activity of secretory IgA in vitro and is assumed to constitute an important virulence determinant in bacterial meningitis (6, 10), direct proof is still missing due to the lack of a suitable animal model. Why the activity has been conserved in G. haemolysans along with the mentioned Streptococcus species is an intriguing question.