INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteroides spp. are gram-negative obligate anaerobes that comprise 20 to 30% of the normal microbiota of the human colon. Some Bacteroides species are opportunistic pathogens, and infections caused by them are becoming more difficult to treat successfully due to increasing antibiotic resistance in this genus. Bacteroides spp. have been shown to carry a plethora of self-transmissible and mobilizable elements, which are probably responsible for the spread of antibiotic resistance genes. Antibiotic resistance genes have been found on conjugative and mobilizable plasmids (21, 34, 43), conjugative transposons (CTns) and integrated elements that are mobilized by CTns (33, 35, 37). In particular, a family of CTns, exemplified by CTnDOT and CTnERL, appears to be playing an important role in transferring resistance genes among Bacteroides strains. CTns of this family not only transfer themselves but also mobilize coresident plasmids. In addition, proteins encoded on these CTns trigger in trans the excision and circularization of mobilizable integrated elements called NBUs, and they mobilize these circular forms to Bacteroides or Escherichia coli recipients (35, 49, 57). They may also mobilize other integrated elements that have been given transposon designations, such as Tn4399 (14), Tn4555 (51), and Tn5520 (60). To distinguish mobilizable elements that have been given a transposon designation from nonmobilizable Bacteroides transposons such as Tn4351 or Tn4551, we will designate them as MTns, e.g., MTn5520.

So far, the MTns of Bacteroides species seem to be falling into two distinct groups. MTn5520 (60), the smallest of the MTns (5 kbp) and the only other MTn besides NBU1 (48) to be sequenced completely, integrates almost randomly and does not duplicate the target site. By contrast, the NBU1 integration was highly site specific, at least in Bacteroides spp., and the target site was duplicated when NBU1 was inserted. We had noted that the integrase of NBU1, IntN1, was very different at the amino acid sequence level from the integrase of MTn5520, although both were distantly related to the phage lambda integrase (46, 60). Since NBU2 seems to have the same general integration features of NBU1, we wanted to identify the integrase of NBU2 and determine whether it was more closely related to that of NBU1 than to that of MTn5520.

Previously, we obtained the entire sequence of NBU1 (10.3 kbp) and identified its integrase gene (46). Subsequently, we have identified three other genes that appear to be essential for NBU1 excision (48). In earlier surveys of Bacteroides clinical isolates for NBU1-like elements, we had identified a second NBU-type element that appeared at first to be very closely related to NBU1. A 2.5-kbp region of NBU2 that contained the transfer origin (oriT) and mobilization gene (mobN2) was sequenced and found to be >85% identical to a similarly sized segment from NBU1 (19, 20, 44). However, results of further hybridization experiments suggested that outside this region, NBU2 might be quite different from NBU1. We report here the complete DNA sequence of NBU2 and the characteristics of its integrase gene. NBU1 carried no antibiotic resistance genes, but an MTn that seems to be related to NBU1, MTn4555, carries a cefoxitin resistance gene (28). Accordingly, we were interested in determining whether NBU2 carried any resistance genes.

Finally, we wanted to learn more about the distribution of NBU-type elements in Bacteroides strains. Virtually all work to date on CTns and MTns of Bacteroides species has focused on a small number of strains, most of which are clinical isolates. We have a collection of Bacteroides strains that includes a variety of clinical and community isolates, including strains of Bacteroides isolated before 1970, as well as strains isolated in recent years. We were interested in determining not only how prevalent NBUs are today but also whether carriage of these elements has changed with time. Information obtained by comparing NBU2 with NBU1 allowed us for the first time to design probes that would distinguish NBU1 from NBU2 and thus allow us to determine how widely each was distributed in Bacteroides strains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. The bacterial strains and plasmids used in this study are shown in Table 1. The E. coli strains were grown aerobically in Luria broth (LB) or plated on Luria agar (LA) plates. The Bacteroides strains labeled BT (e.g., BT4001 and BT4004) are derivatives of the B. thetaiotaomicron 5482 strain (Virginia Polytechnical Institute [VPI] Anaerobe Laboratory, Blacksburg, Va.). The source and time period of isolation of the Bacteroides strains used in the survey are described, along with the results of the NBU survey (see Table 4). The Bacteroides strains are grown anaerobically in prereduced Trypticase-yeast extract-glucose (TYG [15]) or on TYG agar plates incubated in BBL GasPak jars.

Preparation of plasmid and total cellular DNA. Plasmid preparations from either Bacteroides or E. coli strains were done using the alkaline lysis procedure (39). Total DNA was prepared by a modification of the method of Saito and Miura (32). A quick method for preparing total DNA from 2 ml of an overnight culture was used routinely. The cells were pelleted in a microfuge tube, washed one time in 0.5 ml of saline-EDTA (0.5 M NaCl, 0.1 M EDTA; pH 8). The cells were resuspended in 0.5 ml of saline-EDTA and frozen in a 80°C freezer until solid. Then, 0.5 ml of Tris-SDS (0.1 M Tris, 1% sodium dodecyl sulfate [SDS]; pH 9.0 to 9.3) was added to the frozen cells. The tubes were agitated constantly until the cells thawed and lysed completely. The lysate was mixed with 0.5 ml of phenol saturated with saline-EDTA, mixed well, and placed on ice for 20 min with occasional mixing. The mixture was centrifuged at 12,000 rpm in a microfuge for 10 min. The supernatant was removed and put into a fresh tube. Then, 0.8 ml of isopropanol was added, the tubes were inverted several times for complete mixing, and the DNA was allowed to precipitate at room temperature for at least 30 min. The tubes were centrifuged for 10 min at 12,000 rpm. The pellets were rinsed with 70% ethanol and then dried. The DNA was resuspended in 0.2 to 0.4 ml of TE (0.01M Tris, 0.001 M EDTA; pH 8) containing 50 µg of RNase per ml.

Cloning and sequencing of NBU2. NBU2 was induced to excise from the chromosome of B. thetaiotaomicron BT4104N3-1 by growing the strain in medium containing tetracycline. Exposure to tetracycline induces the regulatory functions on the conjugative transposon required to induce the excision of NBU2 (55). The circular intermediate of NBU2 was isolated by using a plasmid preparation procedure described previously (44) and then digested with PstI and cloned into the PstI site of pBR328 (11). This clone was stable enough to allow subcloning for sequencing and for other analysis (Table 1). Various regions were subcloned into pUC19 (62) and sequenced using the M13 universal and reverse primers. Primer walking and sequencing of PCR products across restriction sites was done to get the total sequence of both strands. The sequencing was performed by the University of Illinois Biotechnology Facility using the Applied Biosystems model 373A version 2.0 dye terminator sequencing system.

Southern blot and dot blot analysis. Southern blot analysis of restriction enzyme-digested DNA was performed as outlined in Sambrook et al. (39). The probes were made from isolated DNA fragments or purified PCR products labeled with fluorescein-dUTP using random primers according to the protocol in the Renaissance kit from NEN Life Sciences. The Southern blots were incubated with a chemiluminescence substrate for exposure to film as directed by the manufacturer. Dot blot analyses of bacterial strains were done by spotting 3 µl of a 2-ml (total) DNA preparation onto GeneScreen (NEN-Dupont) in a grid configuration. The spotted membranes were then treated as in colony or plaque hybridizations and were hybridized with NBU1- and NBU2-specific probes using the same protocol used for Southern blots. The NBU probes used for the dot blots were the 4.5-kbp HindIII fragment of NBU1 (C-terminal of IntegraseN1 to the N-terminal of MobN1), the PCR product of the prmN2, oriT'-mobN1 (1.6-kbp AvaI-PvuII), intN1-attN1 (2.4-kbp PstI-ScaI), intN2-attN2 (1.8-kbp PCR product), and a 1.3-kbp HincII-EcoRV NBU2 fragment containing the C-terminal end of mefE_N2 and the N-terminal end of linA_N2. Primers used to produce specific PCR products to make the probes or for cloning are shown in Table 2.

Construction of minimal integration vector, pEPIntN2. A mobilizable shuttle vector which could be used to follow the integration of cloned regions of NBU2 in either Bacteroides or E. coli recipients was constructed using the pir-dependent R6KoriV vector, pEP185.2 (Fig. 1) (24). The ermG gene from the conjugative transposon CTn7853 was PCR amplified (10) (Table 2) with PstI sites in the primers and then cloned into the PstI-compatible unique NsiI site of pEP185.2 to produce pEPE (Fig. 1). The sequences of the ends of the integrated form of NBU2 were used to design primers for PCR amplification of the joined ends of NBU2 (attN2) plus the adjacent ORF, intN2. The 1.8-kbp fragment was first cloned into the pGEM-T (Promega) PCR product cloning vector to form pGEM::IntN2 and was sequenced. The attN2-intN2 or IntN2 region was then isolated on a 1.8-kbp ApaI-SstII fragment and cloned into the corresponding sites on pEPE to form pEPIntN2 (Fig. 1). This vector can be mobilized out of the Pir⁺ S17-1 derivative, BW19851 (23), either to Bacteroides or to E. coli recipients, by filter matings to select for possible integration due to the NBU2 IntN2 region.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Construction of the integration vector, pEPIntN2. An insertional vector that could be used in either Bacteroides or in E. coli was constructed from the pir-dependent mobilizable pEP185.2 (24). The erythromycin-clindamycin resistance gene, ermG, from CTn7853 was PCR amplified with PstI sites encoded in the primers and cloned into the PstI-compatible unique NsiI site on pEP185.2 to form pEPE. pEPE, a Pir-dependent vector, has selectable markers for both E. coli and Bacteroides hosts. The NBU2 integration region, IntN2, consisting of the joined ends, attN2, from the circular form of NBU2 and the adjacent gene, intN2, was PCR amplified from the induced circular form of NBU2. The PCR product was first cloned into pGEM-T (Promega) and sequenced. The 1.8-kbp ApaI-SstII fragment was isolated from pGEM-T and cloned into the ApaI and SstII sites of pEPE to form the NBU2 integration vector, pEPIntN2.

Filter matings. The procedure for filter matings between E. coli donors and E. coli or Bacteroides recipients has been previously described (42, 59). BW19851(pEPIntN2) was filter mated with BT4001. BT4001 transconjugants containing insertions of pEPIntN2 in the chromosome were selected as gentamicin-resistant (Gen^r, 200 µg/ml) and erythromycin-resistant (Em^r, 10 µg/ml) isolates. Integration into the chromosome via the NBU2 ends was verified by Southern blot analysis using the 1.8-kbp ApaI-SstII fragment of NBU2 cloned in pEPIntN2 (Fig. 1) as the probe. In matings between BW19851(pEPIntN2) and the E. coli recipient EM24NR, the transconjugants were isolated as rifampin-resistant (Rif^r, 10 µg/ml), chloramphenicol-resistant (Cm^r, 20 µg/ml), trimethoprim-sensitive (Tp^s) isolates. Integration was verified by Southern blot analysis using the 1.8-kbp ApaI-SstII NBU2 attN2-intN2 fragment as the probe.

Cloning and sequencing of NBU2 target sites in Bacteroides and E. coli hosts. A left junction of NBU2 was obtained from BT4004N6 (Table 1) on a 5.0-kbp HindIII fragment, which was cloned into the HindIII site of pUC19 to produce pUC19::LJ₁, and was then sequenced. The sequence identified the crossover region on NBU2 attN2. In the process of cloning the ends of NBU2, we learned that BT4004N6 had two copies of NBU2. Both junctions of both integration sites were cloned simultaneously from the BT4001pEPIntN2 transconjugants, all of which contained only single insertions. First, the DNA from a BT4001pEPIntN2 transconjugant was digested with EcoRV, which does not cut within the integrated pEPIntN2 vector. The digested DNA was cleaned, diluted, and ligated. The ligation reaction was then used to transform competent BW19851 cells selecting for Cm^r. The Cm^r transformants obtained from the two observed insertion sites on the Southern blots contained the chromosomal sequences adjacent to both ends of the integrated NBU2 derivative cloned on pEPIntN2-J1 (site 1) or pEPIntN2-J2 (site 2). The R6KoriV-based vector has a copy number too low for good template preparation for sequencing. Therefore, the cloned chromosomal DNA between the NBU2 ends on the resultant J-clones were amplified by PCR using the NBU2 end primers (Table 2). The PCR products were cloned into the high-copy-number pGEM-T and sequenced using the M13 forward and reverse primers. The sequences obtained for the chromosomal junctions allowed primers to be designed (Table 2) to PCR amplify and sequence the Bacteroides target sites. A similar strategy was used to obtain target sites in E. coli.

DNA amplification and cloning. The primers used for the DNA amplifications are shown in Table 2. A 2 µl sample of genomic DNA or 10 to 100 ng of plasmid DNA was mixed with 200 ng of each primer in 100 µl of reaction buffer (1× Gibco-BRL PCR buffer, 1.5 mM MgCl₂, 0.2 mM deoxyribonucleoside triphosphate mixture) and amplified with Taq polymerase. Amplification was preceded by denaturation for 5 min at 95°C, followed by the addition of polymerase and 25 cycles of 95°C for 1 min, annealing at 50 to 55°C for 1 min, and extension at 72°C for 2 min. Then, 5 µl of the PCR product was checked on agarose gels for production, size, and concentration. The PCR products were gel purified and sequenced directly using the PCR primers, used to prepare probes for the dot blot hybridization survey, or cloned on the PCR cloning vector pGEM-T as specified by the manufacturer (Promega).

Expression of the mefE_N2 and linA_N2. The BT4004, BT4004N3, BT4004N6, and BT4004(pNLY-ML) strains (Table 1) were tested for their ability to grow in TYG broth containing erythromycin or lincomycin. Fresh overnight cultures of each strain were inoculated in triplicate into media containing 0, 3, 5, 10, 20, 30, or 40 µg of the respective antibiotic per ml. The cultures were incubated at 37°C, and growth was checked after 24 and 48 h.

Nucleotide sequence accession numbers. The GenBank accession number (L42370) for the prmN2-oriT-mobN2 region has been previously reported (20). The nucleotide sequences for the entire NBU2 and the Bacteroides targets, BT2-1 and BT2-2, containing the Ser-tRNA_UGA genes have been deposited in GenBank under accession numbers AF251288.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequence analysis of NBU2. The 2.5-kbp mobilization region of NBU2, which has high sequence identity to NBU1, had been previously sequenced and characterized by Li et al. (20). This region had been cloned from the excised circular form of NBU2 using an NBU1-derived hybridization probe to detect it. To determine the sequence of the rest of NBU2, we first cloned the entire circular intermediate from plasmid preparations of a tetracycline-induced B. thetaiotaomicron strain that contained both the conjugative transposon CTnERL and NBU2. Sequence analysis of NBU2 revealed that the element was 11,123 bp in size, slightly larger than NBU1 (10,276 bp [48]). The sizes of the predicted ORFs and the sizes of the proteins they could encode are shown in Table 3. In Fig. 2A the location of the ORFs are shown on the circular intermediate form of NBU2. Genes on the integrated form of NBU2 are compared to genes on integrated NBU1 in Fig. 2B. Comparisons of these genes at the nucleic acid and amino acid sequence level showed clearly that, outside of the prmN-oriT-mob region, NBU2 differed appreciably from NBU1. Like NBU1, most of the NBU2 genes were transcribed in the same direction, but there was little sequence similarity to the other NBU1 genes. Examination of the sequences at the edges of the highly conserved internal prmN-oriT-mob region revealed that the transition from nearly identical sequences to very dissimilar sequences was not an abrupt one, as might be expected if these genes are on a gene cassette (Fig. 3).

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Circular and integrated forms of NBU2. (A) Partial restriction map of the excised circular form of NBU2. The location and orientation of the possible ORFs derived from the NBU2 sequence are indicated. The region containing the joined ends of the NBU2 (attN2) is indicated. The attN2 is contained within a 1.2-kbp HincII fragment. (B) The integrated form of NBU2 is compared to the integrated form of NBU1. The double-headed arrows indicate the region of high sequence identity between NBU2 and NBU1 (see Fig. 3). The attN-left sequences of the integrated NBUs are the same sequence as the attN on the elements, and the attN-right sequences are the attBT sequences of the target sites. Both elements integrate site specifically into the 3' end of tRNA genes: Ser-tRNAUGA for NBU2 and Leu-tRNACAA for NBU1.

View larger version (45K): [in this window] [in a new window]   FIG. 3.   Sequence comparison of NBU2 and NBU1. The prmN-oriT-mobN regions of NBU1 and NBU2 have more than 85% identity. Outside of this mobilization region the sequence identity drops to <30%. Near the borders of the sequence identities both elements have inverted repeat sequences (underlined) of 11 to 12 bp indicated by arrows above the sequences for NBU1 (IR1) and dotted arrows below the sequences for NBU2 (IR2). The start and stop codons for the prmN (PrmN) and mobN (MobN) are indicated in boldface. The oriT nick sites as determined by sequence identity to MTn4555 (53) are at the end of the TAG codon ( Stop) of the prmN genes and are indicated by the arrows.

Localization of the integration region (attN2) on NBU2. DNA from strains of B. thetaiotaomicron that contained insertions of NBU2 (NBU2) were probed on Southern blots and the ends of the NBU2 were determined to be within the 1.2-kbp HincII fragment indicated in Fig. 2 that is also called attN2. The Southern blots and previous pulse field analysis of B. thetaiotaomicron strains revealed that NBU2 had two primary target sites (3), and some of our isolates contained copies of NBU2 in both sites (e.g., Fig. 4A). In Fig. 4A, the chromosomal DNAs from BT4004N3 (lane 1), BT4004N6 (lane 2), and BT4004N6 grown in tetracycline to induce the excision of NBU2 (lane 3) were digested with HincII. The Southern blot of the agarose gel was probed with the 1.2-kbp HincII NBU2 fragment (Fig. 2). BT4004N3 has one NBU2 (two junction bands) and BT4004N6 had two copies of NBU2 in two different sites (four junction bands). One of the NBU2 insertions in BT4004N6 is the same as the NBU2 in BT4004N3, as is evident from the fact that two junction bands of the same size appear in both lanes 1 and 2. The arrow in lane 3 indicates an additional 1.2-kbp HincII band that runs just above one of the junction bands. This band is formed when tetracycline stimulation of the coresident CTnERL leads to excision of NBU2. No 1.2-kbp band is observed if the cells are not grown in medium containing tetracycline (lanes 1 and 2). To determine the exact region on NBU2 where the crossovers occurred, a 5-kbp HindIII fragment containing one end of NBU2 in BT4004N6 was cloned and sequenced and compared to the sequence of the 1.2-kbp HincII fragment. The HindIII fragment was later determined to be the left junction of NBU2, as shown in Fig. 2B, and the chromosome from site 1. The single insertion of NBU2 in BT4004N3 (Fig. 4A, lane 1) was later shown to be in site 2. 

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Southern blot analysis of the NBU2 insertions in Bacteroides attB2 sites. (A) Southern blot of two B. thetaiotaomicron BT4001 transconjugants from BT ERL that received both CTnERL and NBU2. The DNAs of the two strains were digested with HincII, and Southern blots were probed with the labeled 1.2-kbp HincII fragment containing attN2 (Fig. 2). BT4004N3 in lane 1 has a single insertion (two junction bands), and BT4004N6 in lane 2 has two insertions of NBU2 (four junction bands). In lane 3, BT4004N6 was grown in tetracycline to induce the circular excised form of NBU2. The 1.2-kbp fragment from the excised circular form of NBU2 is indicated by the arrow. (B) Southern blot showing 6 of 10 independent insertions of the minielement vector, pEPIntN2, into the B. thetaiotaomicron BT4001 chromosome. The DNAs from the strains were digested with ApaI-SstII, and the Southern blot was probed with the labeled 1.8-kbp ApaI-SstII fragment containing the intN2-attN2 region of NBU2 cloned on pEPIntN2 (Fig. 1). The sequence obtained for the cloned targets sites (see Materials and Methods) indicated that the insertion site of pEPIntN2 in lanes 1, 2, and 4 were the same as the wild-type element in BT4004N3 (panel A, lane 1). In BT4004N6, one of the NBU2 copies was integrated into this site and the other copy of NBU2 was integrated into the site represented by lanes 3, 5, and 6 of panel B. The locations of the HindIII lambda size fragments in kilobase pairs are marked on the left of each panel.

View larger version (40K): [in this window] [in a new window]   FIG. 5.   C-terminal alignment of IntN2 and related members in the lambda family of site-specific integrases. The similarity of IntN2 to members of the lambda family of site-specific integrases was in the C-terminal end. The alignment of IntN2 and the integrases with the highest similarities is shown. Domain I and domain II in the C-terminal ends of this family of integrases contain conserved amino acids for the active sites of the integrases (2, 27). The conserved amino acids are labeled (#) and boxed for these active site domains. The total size of each integrase is shown at the end of its sequence as COOH-, with the number of amino acids in parentheses. The accession numbers for the integrases shown here are as follows: Bacteroides MTns NBU2 (L42370), NBU1 (L13840), and MTn5520 (AF038866) and cyanobacterial plasmid pDU1 (L23221), A. aeolicus plasmid, ece1, (C55205), L. lactis CTn5276 (L27649), bacteriophage P21 (P27077), and prophage E14 (M61865).

View larger version (11K): [in this window] [in a new window]   FIG. 6.   Unrooted tree of the NBU2 integrase and related integrases. The entire sequences of the lambda family of site-specific integrases, including the IntN2 of NBU2, that were aligned in Fig. 5 were grouped on a cladogram or unrooted tree to show their relationships using the Bioinformatics analyses conducted on BioNavigator.com provided by eBioinformatics. The tree showed three clusters: the three Bacteroides MTn integrases [NBU1 (IntN1), NBU2 (IntN2) and MTn5520 (IntBIP)], the two bacteriophage integrases from P21 and prophage E14, and the group containing the putative integrase-recombinases on plasmids from Nostoc (pDU1) and A. aeolicus plasmid ece1 (pAquifex) and the integrase from the L. lactis conjugative transposon CTn5276. Accession numbers are given in Fig. 5.

Use of a special vector to study integration and the targets sites of NBU2. An insertional shuttle vector, pEPE, was constructed to locate the integration gene of NBU2 and to clone chromosomal target sites (Fig. 1). This vector can be mobilized out of E. coli BW19851 to Bacteroides or E. coli recipients. In E. coli strains that contain the R6K pir gene the vector replicates but in Bacteroides recipients and E. coli recipients such as EM24NR, pEPE cannot replicate and transconjugants are only obtained if the vector contains regions that allow it to integrate (24, 47). Since we suspected that the first ORF at the right end of NBU2 was the integrase gene, we PCR amplified this gene together with the joined ends of the circular form and cloned it into pEPE to form pEPIntN2 (Fig. 1 and 2A).

4

5

3

4

1

9

View larger version (18K): [in this window] [in a new window]   FIG. 7.   NBU2 attN2 and chromosomal target sites. The attN2 region of NBU2, shown at the top of the figure, has two regions that have identity or similarity to its target sites. The crossovers occurred within or adjacent to a 13-bp sequence in highlighted capital letters that has sequence identity to the two attBT2 Bacteroides target sites. The second region of attN2 has a 14-bp sequence, boxed and highlighted, that has partial sequence identity to the attBT2 sites, and is located at the 5' end of a pair of IRs. The IRs are indicated by arrows and the bases which could pair in a stem and loop structure are capitalized. The two attBT2 sites are located at the 3' ends of Ser-tRNAUGA genes that differ from each other by only a single base pair. The crossovers occurred within or adjacent to the 13-bp sequence of identity to attN2. Both attBT2-1 and attBT2-2 also had 15-bp sequences with partial identity, boxed and highlighted, to the 14-bp sequence of attN2 that were located 5' to IRs. The sequence for one of the pEPintN2 insertion sites in E. coli (attBEc) is shown at the bottom. The insertion occurred in the 3' end of fecI. The crossover occurred adjacent to or within the triplet (CCT) at the beginning of a sequence with partial identity (8 of 13) to the 13-bp region on attN2. The crossover region for attBEc was followed by a set of IRs indicated by arrows which had no sequence identity to the NBU2 IR region.

6

7

1

5

1

Expression of the putative NBU2-encoded antibiotic resistance genes, mefE_N2 and linA_N2. The sequence of NBU2 revealed two ORFs whose derived amino acid sequences were related to those of known antibiotic resistance genes previously found in the gram-positive bacteria. The deduced amino acid sequence of mefE_N2 had 34% identity and 54% similarity to MefE of Streptococcus pneumoniae (58), a protein that is thought to be a macrolide pump. Although the sequence identity was low, it extended throughout the protein. This is the first mefE homolog seen in Bacteroides species. The deduced amino acid sequence of linA_N2, LinA(N2), had 50 to 52% identity and 70 to 72% similarity to the LinA' of Staphylococcus aureus (5) and LinA on pIP855 in S. haemolyticus (6). An alignment of the three resistances is shown in Fig. 8. LinA is an O-nucleotidyltransferase which inactives lincosamides including lincomycin and clindamycin (6). Clindamycin resistance is a clinical problem because clindamycin is still a drug of choice for treating anaerobic infections, including those caused by Bacteroides spp. This is also the first sighting of a linA type gene in Bacteroides species. The erythromycin and clindamycin resistances in the Bacteroides spp. have previously been associated exclusively with MLS (ribosome methylation)-type resistances, e.g., ermF and ermG. ermF and ermG have been found on transposons, conjugative and mobilizable plasmids, and conjugative transposons (10, 21, 33, 35). Neither of these genes, however, has yet been seen on a MTn.

View larger version (37K): [in this window] [in a new window]   FIG. 8.   Alignment of the NBU2 lincosamide resistance, LinAN2, with LinA' and LinA from Staphylococcus spp. The amino acid sequence of the lincosamide resistance on NBU2, LinA(N2), was aligned with the sequences of LinA' from S. aureus and LinA from pIP855 in S. haemolyticus (5, 6) using the BLASTP search program. LinA(N2) has 52% identity and 70 to 72% identity to LinA' and LinA. The regions of identity are indicated in boldface. The total number of amino acids for each protein is indicated at the ends of the respective sequences. The accession numbers for LinA' and LinA are J03947 and A25633, respectively.

Prevalence of NBU genes in community and clinical Bacteroides isolates. We have seen NBU-like elements in several Bacteroides clinical isolates and were interested in determining how widespread these elements were in Bacteroides species. To this end, we surveyed community and clinical isolates for the presence of NBU-type elements in general and for NBU1 and NBU2, separately. The probe used to detect NBU-type elements was the 4.5-kbp HindIII fragment of NBU1 that contains the highly conserved prmN1-oriT-mobN1 region. The probes used to detect NBU1 and NBU2 specifically were the integrase genes of these two elements. Since earlier data had shown that the highly conserved region might be modular, we also probed the strains separately with prmN1- and mobN1-specific probes to learn how often these genes were found together. Finally, we probed the strains with the mefE-linA region of NBU2. We were interested in answering three different questions. First, are NBUs widely distributed in Bacteroides species and, if so, are they as widely distributed in community as well as clinical isolates? Second, has the prevalence of NBUs in Bacteroides species changed over the past few decades? Finally, how prevalent are the mefE-linA homologs in Bacteroides species and are they always associated with NBU-positive strains? Some of the strains were isolated in the 1960s and 1970s by the Anaerobe Laboratory at VPI in Blacksburg, Va. (16), and some strains were isolated in the 1990s, so we were able to look at prevalence data from older and more recently isolated strains. A summary of the results is provided in Table 4.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The integrase of NBU2, intN2, has been identified. This gene and the joined ends of NBU2 were all that were required for integration. The NBU2 integrase, like that of the related mobilizable element, NBU1, appears to be a member of the family of phage lambda recombinases, in the sense that it has the conserved C-terminal catalytic amino acids that are preserved on all members of the lambda Int family (27). The finding of lamba-type integrases in the NBUs and other nonphage integrating elements suggests that this mode of site-specific integration is more general than was previously realized. NBU2 and NBU1 both have some other phage-like traits. For one thing, they integrate site specifically via a 13-bp att sequence that is identical to the att site in the NBU2 joined ends. For another, they integrate into the 3' ends of tRNA genes. The number of integrating elements that share one or more of these properties is growing and now includes not only bacteriophages and the NBUs but also integrative Streptomyces plasmids, sone pathogenicity islands, and integrative elements in Dichelobacter nodosus (4, 7, 31, 40, 61). The use of tRNA genes as targets for integration may increase the host range of an integrating element because tRNA genes are fairly highly conserved in different species. On the other hand, the use of tRNA genes as target sites could be considered as limiting their movement since such integration sites are unlikely to be found on plasmids or other self-transmissible elements. This limitation, however, does not seem to have prevented the NBUs from spreading extensively among different Bacteroides species in the human colon.

Transmissible elements that use tRNA genes as an integration sites could be hazardous to a bacterial recipient if the structure of the tRNA gene was disrupted. In the case of NBU2, the crossover event exchanges the IR at the 3' end of the tRNA with the IR found in the attN site. This change seems not to be deleterious for the recipients, but such a change could possibly influence the processing of the tRNA transcript (61). The importance of the IRs in the attN or the attBT sites for either the integration or the excision of the NBUs is not yet known.

At present, there is no information available about Bacteroides host factors that might be involved in NBU integration or excision. Previous work has shown that NBU integration is independent of RecA (10), but whether integration requires a host factor-like IHF remains to be determined. Integration of NBU2 did not require any CTn functions, in contrast to the excision of NBUs, which requires functions provided by a CTn. In E. coli, integration was much more random than in B. thetaiotaomicron. This could be due to lack or a suitable primary integration site, because the Ser-tRNA gene of E. coli is not identical to that of B. thetaiotaomicron. It is also possible, however, that the relative lack of NBU2 integration specificity in E. coli, together with the much lower frequency of integration, reflects the absence in E. coli of a host factor that aids accurate alignment of the att sequences and efficient formation of the protein-DNA contacts in Bacteroides species. Excision of NBUs is proving to be a much more complex process than integration. The integrase is required, along with at least three other proteins and the oriT region (48). Moreover, excision requires trans action of CTn regulatory proteins, RteA and RteB. Thus, although the integration process of NBUs may resemble that of lambdoid phages, the excision process seems much different.

Integrated elements carrying genes that cross-hybridize with genes on NBU1 and NBU2 are very widespread in Bacteroides species. The fact that their incidence seems to be increasing is a good indication of the efficiency of their transfer and their stability once acquired. Our results suggest that this group of integrated elements is likely to be a very heterogeneous group. Not only were NBU1 and NBU2 quite different outside the prmN-oriT-mob region, but the integrases of NBU1, NBU2, and a related element, MTn5520, had substantially different amino acid sequences (Fig. 6). The differences between the integrases is evident at the functional as well as the amino acid sequence level because the integrase of NBU1 did not substitute for the integrase of NBU2 in an integration assay. Why the prmN-oriT-mob region of NBU1 and NBU2 is so highly conserved at the sequence level, whereas the remainder of the elements are so different, has yet to be determined. This region could be on a cassette but, if so, it has been in the NBUs long enough for its edges to be obscured by mutation. Another possible explanation is that this region of NBU1 is very important for excision (48). Yet, so are the integrase and genes downstream of the integrase, which are quite different on NBU1 and NBU2. Still another possible explanation for the conservation of the prmN-oriT-mob region is that one or more proteins or DNA segments in this region interact with functions supplied by the CTns. This is most likely in the case of the Mob protein, which must interact with transfer functions on encoded on the CTn, which mediate the transfer of the NBU circular form. Both NBU1 and NBU2 also interact in some with the CTn regulatory proteins that trigger excision (RteA, RteB), so the conserved region might also be involved in that interaction. In our survey of Bacteroides isolates, we noted a few isolates that hybridized with prmN but not with mob. It will be interesting to determine whether the NBU-like elements in such isolates are capable of excision and transfer.