Evaluation of amplified rDNA restriction analysis (ARDRA) for the identification of Mycoplasma species

Background Mycoplasmas are present worldwide in a large number of animal hosts. Due to their small genome and parasitic lifestyle, Mycoplasma spp. require complex isolation media. Nevertheless, already over 100 different species have been identified and characterized and their number increases as more hosts are sampled. We studied the applicability of amplified rDNA restriction analysis (ARDRA) for the identification of all 116 acknowledged Mycoplasma species and subspecies. Methods Based upon available 16S rDNA sequences, we calculated and compared theoretical ARDRA profiles. To check the validity of these theoretically calculated profiles, we performed ARDRA on 60 strains of 27 different species and subspecies of the genus Mycoplasma. Results In silico digestion with the restriction endonuclease AluI (AG^CT) was found to be most discriminative and generated from 3 to 13 fragments depending on the Mycoplasma species. Although 73 Mycoplasma species could be differentiated using AluI, other species gave undistinguishable patterns. For these, an additional restriction digestion, typically with BfaI (C^TAG) or HpyF10VI (GCNNNNN^NNGC), was needed for a final identification. All in vitro obtained restriction profiles were in accordance with the calculated fragments based on only one 16S rDNA sequence, except for two isolates of M. columbinum and two isolates of the M. mycoides cluster, for which correct ARDRA profiles were only obtained if the sequences of both rrn operons were taken into account. Conclusion Theoretically, restriction digestion of the amplified rDNA was found to enable differentiation of all described Mycoplasma species and this could be confirmed by application of ARDRA on a total of 27 species and subspecies.


Background
Mycoplasmas are phylogenetically related to gram-positive bacteria with low GC-content and belong to the class of the Mollicutes. They form a unique group of bacteria that lack a cell-wall and that contain sterols in their cytoplasmatic membrane. They are of great importance, since several species are pathogenic to animals or humans, whereas species of other mollicute genera also infect plants and insects [1]. In addition, a series of mycoplasmas cause trouble in the laboratory, because they infect cell cultures. Already over 100 species have been described, and their number, as well as the number of different hosts is still increasing.
A correct identification of mycoplasmas, mostly performed after a fastidious initial isolation, may be achieved by various methods. Original tools to identify mycoplasmas were mainly based on biochemical and serological differentiation, varying from simple precipitation tests [2], to ELISA [3,4], immunofluorescence [5], or Western blot analysis [6]. These techniques are being replaced by faster DNA-based tools [7]. Many of these methods are based on the 16S rDNA sequence for various reasons. First, the 16S rDNA has been sequenced for all recognized Mycoplasma spp. and is required when describing a new species [8]. Secondly, the 16S rDNA sequences have lower intraspecific variability than most protein encoding genes, hence their use in the construction of phylogenetic topologies [9]. Recently, denaturing gradient gel electrophoresis of amplified 16S rDNA was shown to be useful to differentiate most Mycoplasma spp. [10]. In another approach, correct identification of related Mycoplasma spp. was based on differences of the 16S-23S intergenic spacer (ITS) region. Both size variation [11] as sequence differences [11,12] of the ITS were successfully used to differentiate related species. Compared to the 16S rDNA sequence, ITS sequences may vary more between strains of the same species due to a lower selection pressure [13], although reports of very highly conserved ITS regions are known as well [14].
Amplified rDNA restriction analysis (ARDRA) has already been used for the identification of some avian species [15][16][17] as well as for pathogenic mycoplasmas in cats [18]. Restriction analysis with PstI of an amplified 16S rDNA fragment was also shown useful to differentiate M. capricolum subsp. capripneumoniae from the other species belonging to the mycoides-cluster [19]. The potential and power of ARDRA to identify members of the Mollicutes was already put forward [20], but was never worked out in detail for a large number of species. In this study, we investigated the value of ARDRA to identify all (to date) recognized Mycoplasma spp.

Isolates
A total of 60 strains, belonging to 27 different Mycoplasma species and subspecies, were used during this study ( All isolates were previously identified using biochemical tests and growth precipitation tests with absorbed rabbit antisera [2]. Whenever discrepancies existed between the obtained ARDRA-profiles and the serological results, the 16S rDNA was sequenced for an exact identification [25].

DNA extraction
DNA of growing cultures was extracted using a phenolchloroform extraction described previously [26] or using alkaline lysis. For alkaline lysis, the cultures were centrifuged (2', 10000 g) and resuspended in 50 µl lysis buffer (0.25% SDS in 0.05 N NaOH). After 5' at 95°C, 300 µl water was added and the bacterial debris was centrifuged (2', 10000 g). One µl of the supernatant was used as template for amplification of the 16S rDNA.

Restriction digestion
For all 60 strains, 10 µl of the 16S rDNA PCR product was digested with 5 U of restriction enzyme AluI (Fermentas, Lithuania; sequence: AG^CT) and the associated Y + /Tango restriction buffer (Fermentas) in a total volume of 20 µl for 2 hours at 37°C. For a final identification, the amplified 16S rDNA of some strains were digested in addition with BfaI (New England Biolabs, USA; sequence: C^TAG) or HpyF10VI (Fermentas; sequence: GCNNNNN^NNGC). The restriction fragments were separated on a 3% Nusieve 3:1 agar (Tebu-Bio, France) for 2 hours at 130 V and visualized using a GeneGenius gel documentation system (Westburg, The Netherlands). A 50-bp ladder was used as a DNA marker (Fermentas).

Sequences &in silico ARDRA-profiles
ARDRA-profiles were calculated for all Mycoplasma spp. as acknowledged by the International Committee on Systematics of Prokaryotes (ICPS) to date. The 16S rDNA sequences were downloaded from Genbank (accession numbers are indicated in Figure 1). A consensus sequence was constructed and used for species for which more than one sequence was available. The M. orale 16S rDNA sequence was determined and submitted [Genbank:AY796060], since the only available sequence contained numerous ambiguities. For the members of the M. mycoides-cluster -for which differences between rrnA and rrnB have been published [27] -both sequences were used. For some Mycoplasma spp. only a partial sequence of the 16S rDNA was available. For these sequences, nucleotides were added to the 5' and/or 3' ends to generate fragments of expected length. These lengths and the choice of the nucleotides added were based on a 16S rDNA consensus sequence obtained by alignment of the complete Mycoplasma 16S rDNA sequences available in Genbank using Clustal W. The restriction sites and the exact size of the ARDRA fragments were calculated using Vector NTI Advance V9.0 (Invitrogen) and BioNumerics V3.5 (Applied-Maths, Belgium).
By way of illustration, a dendrogram, based on ARDRA patterns, was constructed using the Unweighted Pair Group Method with Arithmetic Means (UPGMA) using 1% tolerance (i.e. bands that differ about 7 nucleotides or less are considered identical) and taking only fragments from 80 to 800 nucleotides into account.

Results
For all Mycoplasma spp., the theoretical AluI,BfaI and HpyF10VI restriction patterns were calculated [see Additional file 1] and are represented in Figure 1, 2, 3. For a number of species, ARDRA was carried out in the laboratory to confirm the in silico obtained results and to check the validity of the technique for identification. ARDRA profiles obtained with AluI and BfaI are shown in Figure 4 and Figure 5, respectively. For a further verification of the technique and for the remaining 9 species that could not be identified with AluI or BfaI alone, ARDRA was also performed with HpyF10VI ( Figure 6, 7). Theoretical ARDRA patterns after in silico digestion with AluI for all currently recognized Mycoplasma spp  In case of the very related members of the mycoides-cluster, the differentiation is more complicated and a whole series of restrictions are needed. Based on the occurrence of different restriction sites, it is however theoretically possible to correctly identify these species as well, using only commercially available restriction endonucleases ( Table 2).

Discussion
Identification of mycoplasmas still largely relies on serological tests, but owing to the limited availability of quality-controlled sera, the high number of species, the serological cross-reaction between related species and the great variability in the surface antigens of different strains technique less favorable for routine diagnosis. In this study, we showed that theoretically all Mycoplasma spp. are distinguishable using ARDRA. The in silico determined discriminative power was confirmed in the laboratory and even closely related Mycoplasma spp. could be identified correctly, as exemplified by the restriction with AluI and BfaI of M. agalactiae and M. bovis.
We used universal primers to amplify the entire 16S rDNA to obtain a maximum discriminatory power. Working with universal primers implies that interference from other bacteria is to be expected when starting from clinical samples [29], especially when mycoplasmas are not abundantly present. The use of mycoplasma-specific primers binding to internal regions of the 16S rRNA genes may be helpful and result in a higher specificity as was already proposed by others [20,30]. However, care must be taken since the discriminatory power will decrease if primers are chosen in such a way that less restriction sites are present in the amplification products. Besides, most differences between the two operons will not lead to altered restriction sites and will not influence the ARDRA patterns. In case a mutation is located within one of both restriction recognition sites, as was shown in particular for M. columbinum, restriction will most likely yield an unknown ARDRA profile, rather than lead to a false identification. Moreover, this aberrant pattern can be included in the identification scheme.  a Two values indicate differences between rrnA and rrnB, based on the Genbank accession numbers indicated in 1.
criteculi and M. collis), it was calculated that restriction analysis with a single additional enzyme would result in different restriction patterns and therefore to a correct identification.

Conclusion
Restriction digestion with AluI of the amplified 16S rDNA can be used to differentiate between 73 of the 116 described Mycoplasma species and subspecies. An additional restriction with BfaI or HpyF10VI enables the identification of another 31 species and subspecies. Also the remaining 12 species can be differentiated, with the use of additonal enzymes, although other techniques may be preferred for some members of the M. mycoides-cluster.
The simplicity and the general applicability of ARDRA make it possible to implement this technique in most laboratories with basic molecular biology equipment.