|Home | About | Journals | Submit | Contact Us | Français|
Multilocus sequence typing with nine selected genes is shown to be a promising new tool for accurate identifications of Brevibacteriaceae at the species level. A developed microarray also allows intraspecific diversity investigations of Brevibacterium aurantiacum showing that 13% to 15% of the genes of strain ATCC 9174 were absent or divergent in strain BL2 or ATCC 9175.
Brevibacteriaceae play a major part in the cheese smear community (6, 11). The classification and typing of cheese-related Brevibacteriaceae have been based mainly on molecular methods such as amplified ribosomal DNA restriction enzyme analysis, pulsed-field gel electrophoresis, and ribotyping (8, 10, 12). Recently, the original Brevibacterium linens group was split into two species on the basis of their physiological and biochemical characteristics, the sugar and polyol composition of their teichoic acids, and their 16S rRNA sequence and DNA-DNA hybridization levels. One species remains B. linens and is represented by type strain ATCC 9172. The other, represented by type strain ATCC 9175, has been renamed Brevibacterium aurantiacum. Regarding this new classification, the taxonomic position of cheese-related isolates has to be revisited and potential relationships between phylogenetic affiliation and the potential occurrence of given metabolic characteristics redefined (7). The unfinished genome sequence of B. aurantiacum ATCC 9174 has recently been released by the Joint Genome Institute (http://genome.jgi-psf.org/draft_microbes/breli/breli.home.html). The development of focused phylogenetic approaches using multiple markers in conjunction with whole-genome screening techniques such as comparative genomic hybridization (CGH) has proven to be useful for the detailed characterization of pathogenic species, including food pathogens (3, 5, 9). However, only a few technological species have been investigated at an intraspecies level (2). Our intention was thus to develop modern tools to facilitate the typing of strains of technological interest, for which Brevibacteriaceae could be used as a case study.
Three cheese Brevibacterium sp. strains, BL2, CNRZ918, and ATCC 9174, and B. aurantiacum and B. linens type strains ATCC 9175 and ATCC 9172, respectively, (7), were analyzed both by 16S rRNA sequencing with the universal primers and by multilocus sequence typing (MLST). The 16S rRNA analysis showed the phylogenic relationship between these strains (Table (Table1).1). B. linens ATCC 9172 is an independent lineage. BL2 and ATCC 9174 were related to B. aurantiacum type strain ATCC 9175. Interestingly, strain CNRZ918 presents similarities to the B. aurantiacum lineage, but this strain appeared to be closely related to Brevibacterium antiquum, with 99% identity between their 16S rRNA sequences.
To extend the 16S rRNA phylogenetic organization of technological Brevibacterieceae, an MLST approach was used. Nine genes (cysN, glnA, gyrA, metY, metX, mgl, pheS, sahH, and tkt) were chosen from the conserved housekeeping genes involved in sulfur, carbon, or nitrogen metabolism or general cellular processes. For each gene, we designed primers (Table (Table1)1) within the conserved part surrounding a variable area after sequence alignment of these genes from B. aurantiacum ATCC 9174, Arthrobacter aurescens (TC1), and Arthrobacter sp. strain FB24. Among these genes, sahH was especially discriminating at both the interspecific and intraspecific levels, showing as much as 3% sequence divergence between B. aurantiacum ATCC 9174 and ATCC 9175 (Table (Table1).1). The discriminating ability of glnA and cysN was more questionable, while the gyrA, metY, metX, and tkt markers were discriminating only at an interspecific level (Table (Table1).1). After concatenation of the nine DNA sequences, the phylogenetic tree obtained shows three MLST clusters (Fig. (Fig.1).1). Cluster I contained B. aurantiacum ATCC 9175, ATCC 9174 and BL2. Cluster II corresponded to strain CNRZ918, and cluster III was composed of the B. linens ATCC 9172 type strain. Similarities of 92.3% and 86.5% were observed between the MLST sequence of ATCC 9174 and those of CNRZ918 and ATCC 9172, respectively. We can propose tkt as a marker for interspecies discrimination. According to 16S rRNA and MLST data, the original classification of cheese-related Brevibacteriaceae into two different species was also confirmed and extended. Indeed, the occurrence of B. antiquum had previously been restricted to permafrost, while this study constitutes the first report of a B. antiquum-related cheese isolate (7). Interestingly, cheese-related Brevibacteriaceae bacteria were of a polyphyletic origin, suggesting that adaptation to the cheese habitat could have occurred several times independently in various lineages of Brevibacteriaceae.
In order to extend our analysis of cheese-related Brevibacteriaceae, we performed a genome comparison by hybridization analysis with the five strains studied by MLST. For this purpose, we used a microarray based on the unfinished genome sequence of B. aurantiacum ATCC 9174 (see the supplemental material). The array was further hybridized with the DNAs from the three cheese-related Brevibacterium groups as defined by MLST. The CGH analysis was performed by using the Franck Picard model (13). Although a signal was obtained with DNAs from ATCC 9172 and CNRZ918, the log2 ratio was not significantly discriminating (see Fig. S1 in the supplemental material). The use of arrays should therefore be restricted to investigation of the genomic variability of B. aurantiacum species. Eighty-five percent and 87% of the probes gave a positive signal with B. aurantiacum ATCC 9175 and BL2, respectively. Among the 4,211 genes of ATCC 9174 for which probes were present, 3,308 were conserved in both ATCC 9175 and BL2. Five hundred fifty-seven and 650 genes were probably absent or highly divergent in BL2 and ATCC 9175 compared to ATCC 9174, respectively. Table Table22 summarizes the functional groups associated with the divergent parts of the B. aurantiacum genome and the mobile genetic element-related genes associated. We further focused our investigations on the variable parts of the genome with potential links with the adaptation of B. aurantiacum to cheese characteristics such as salt resistance, peptide transport, and amino acid- or sulfur-related metabolism. The presence or absence of these regions (see Table S1 in the supplemental material) was further checked by PCR when possible. A first deleted region spanning the region from BL614 to BL621 in the genome of strain ATCC 9174 was confirmed by PCR amplification. By sequencing these PCR fragments, we determined that the ends of the fragments deleted in strains ATCC 9175 and BL2 were conserved (Fig. (Fig.2B).2B). This observation suggests an insertion of a catabolic island in the genome of strain ATCC 9174. This chromosomal region encodes an oligopeptide-type ABC transporter (BL618 to BL621), two putative monooxygenases (BL615, BL617), and a reductase (BL616) that could be involved in sulfonate catabolism (Fig. (Fig.2A).2A). The CGH results also showed that five peptide or amino acid transporters are deleted in both the ATCC 9175 and BL2 genomes while an additional peptide transporter (BL1002 to BL1006) seems to diverge between ATCC 9174 and BL2 (see Table S1 in the supplemental material). The diversity among the peptide uptake systems could be of potential interest for biotechnological applications. Indeed, yeast (Geotrichum candidum) and bacteria present during the cheese-ripening process degrade casein and produce peptides (1, 4). These peptides can be further used by other microorganisms for volatile sulfur compound production. The microarray developed in this study offers opportunities for further studies aiming at evaluating the biodiversity of a higher number of B. aurantiacum strains of technological interest, coming from various cheese habitats.
The B. linens BL2 sequence data were produced by the U.S. Department of Energy Joint Genome Institute (http://genome.jgipsf.org/draft_microbes/breli/breli.home.html). We are grateful to Antoine Danchin for support and stimulating discussions. We thank Donald White for correcting the English in this report.
This work was supported by the EcoMet program (ANR-06-PNRA-014) funded by ANR (Agence Nationale de la Recherche) and the Centre National de la Recherche Scientifique (CNRS URA 2171). Marie-Pierre Forquin is grateful to ANR for a Ph.D. scholarship, and I.M.-V. is a full professor at the Université Paris 7.
Published ahead of print on 31 July 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.