DNA isolation analysis.

Plasmid DNA was extracted by the alkali lysis method (37) or by using Qiagen columns (Qiagen, Inc.). The CsCl-ethidium bromide density gradient centrifugation technique was used to identify plasmids in ETBF strain 86-5443-2-2. Cosmids 5G11 and 11F11 containing bft-2 were restriction mapped by Southern blot walking according to the conditions described previously (9). Purification of DNA fragments and extraction from gel slices were performed with Qiaex II gel extraction kit (Qiagen, Inc.).

Colony blot hybridizations.

Colony blots of B. fragilis strains were prepared by the technique described previously (9). Briefly, B. fragilis organisms grown overnight on BHC agar were transferred to Whatman 541 filters. The filters were microwave processed in alkali solution (0.5 M NaOH, 1.5 M NaCl) followed by neutralization in 2 M ammonium acetate. The filters were hybridized with either fragment probes or specific oligoprobes. The fragment probes were labeled with [α-³²P]dATP by random priming (Random-Prime DNA-Labeling Kit; Boehringer GmbH, Mannheim, Germany), hybridized in 50% formamide–5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% SDS–1 mM EDTA–1× Denhardt’s solution at 37°C, and washed with 5× SSC–0.1% SDS at 65°C. Synthetic oligonucleotide probes were end labeled with [γ-³²P]dATP by using T4 polynucleotide kinase (37), hybridized in 6× SSC–5× Denhardt’s solution–1 mM EDTA (pH 8) for 16 h at 56°C and washed twice in 1× SSC–1% SDS at 56°C for 10 min.

Nucleotide sequence analysis.

Recombinant plasmids and PCR products were sequenced by the fluorescent dideoxy terminator method of cycle sequencing (20) on a PE/ABd 373a automated DNA sequencer at the DNA Analysis Facility of Johns Hopkins University. Sequences were compiled with Sequencher software (Gene Codes Corp., Ann Arbor, Mich.). DNA and amino acid sequences were analyzed by using programs developed by the Genetic Computer Group at the University of Wisconsin (4), and a sequence homology search was performed by using the National Center for Biotechnology Information (NCBI) BLAST server (1).

Expression of MPII in E. coli.

The T7 promoter polymerase expression system (Novagen pET-26b[+]) was used to express metalloprotease MPII in E. coli. Primers PrimA and PrimB.1 (sequences are described in Table ​Table2)2) generated a PCR product of the DNA region encoding the proprotein and mature protein of MPII and containing a BamHI and XhoI at the 5′ and 3′ ends of the product, respectively. Vector pET-26b(+) carries an N-terminal pelB signal sequence for potential periplasmic location. Cloning of the PCR product in the BamHI and XhoI sites of the pET-26b(+) vector generated a fusion protein containing pelB signal sequence-MPII proprotein-mature protein, whose expression was under the control of the T7 promoter. As recommended by the manufacturer, the mpII construction was first cloned into a nonexpression host (E. coli HB101) lacking the T7 RNA polymerase. Once established in a nonexpression host, plasmids were transferred into the expression host (E. coli BL21) containing a copy of the T7 RNA polymerase gene, which permitted expression of MPII. Expression of MPII in E. coli was determined by Western blot with anti-ETBF 86-5443-2-2 antibodies.

TABLE 2

Primers and oligonucleotide probes used in thisstudy

Primer or oligoprobe	Sequence (5′ to 3′)
Primers
P1T7	GCTGGTAGACTACCTGAGTAAGGAGTC
P1T7-1	GCTTCCGTACCCAGGTATCTCTCCATA
P1T3	TTCAACCTGATCGATCCGGAAGATCCG
P1T3-1	GGTAGTGCTTATGTCCCTGCAACCCTA
PrimA	GCATGGATCCATGTGCTGACGATCTTCTT
PrimB.1	CGCGCTCGAGTTTTTGTATGCATTCCAAC
Confir1	CAGCTTCATTAGAACATGCAGCTAATAA
Confir2	CTGTCCATAACTGTATTCCATTTGATAC
Confir3	CACCGACAGAACATATCCACTCAACTG
Confir4	GGTGTTGCAAGAGAGGCTCAAACACAT
OlirteB1	GACTACGCCGAAGTTAATACCG
OlirteB2	TCGTGCAGGCGGTACAGAAGAT
Oligoprobes
MT086	GGTGCTAGGCATGCGGATGATC
MTVPI	GGCGCTGAGCATACGGATAATT
K469	CAGATGCAGGATGCGGCGAACT
Olibft3	CGAGGTGTCTGCTCTTGAATC

Open in a separate window

Activity of MPII on HT29/C1 cells and E-cadherin.

The cell-free supernatant and the supernatants of whole-cell lysate of strain F2 (E. coli expressing MPII) were tested for biological activity on HT29/C1 cells and E-cadherin. Activity on HT29/C1 cells was determined at 3 or 24 h as described previously (26, 49). Activity of MPII on E-cadherin was determined as described previously (51). Briefly, HT29/C1 cells were treated with sterile MPII-containing filtrates for 16 h at 37°C. Cleavage of E-cadherin was determined by Western blot by using antibodies to the cytoplasmic domain of E-cadherin (a gift of James Nelson, Stanford University, Palo Alto, Calif.) as the primary antibody (1:1,000 dilution) and anti-rabbit immunoglobulin G-horseradish peroxidase as the secondary antibody (1:10,000 dilution). Purified BFT from ETBF strain 86-5443-2-2 was used as a positive control in both assays. Negative controls included medium alone and crude supernatant filtrates of the NTBF strain 077225-2 and E. coli BL21 containing plasmid pET26.

Bacterial matings.

Thymidine-dependent mutant strains 086Thy⁻3, I-1356-1Thy⁻1, I-1345Thy⁻2, and K518Thy⁻2 derived from ETBF 86-5443-2-2 (contains bft-2), ETBF I-1356-1 (contains bft-1), NTBF I-1345 (pattern III; see Results), and NTBF K518 (pattern II; see Results), respectively, were used as donor strains. All donor strains were resistant to tetracycline (Tc^r) and sensitive to rifampin (Rif^s). The presence of the tetracycline-inducible transposon (Tc^r element) in each Tc^r strain was confirmed by PCR by using a set of primers derived from the regulatory gene rteB (OlirteB1 and OlirteB2 [sequences in Table ​Table2])2]) (46). NTBF TM4000 (Tc^s Rif^r) was used as the recipient strain. B. thetaiotaomicron BT4107 (a thymidine-dependent donor containing the Tc^rEm^r-DOT element) and B. thetaiotaomicron BT4001 (a rifampin-resistant recipient) strains were used as bacterial mating controls. Conjugation matings were done by using the filter mating procedure as described by Shoemaker et al. (42). Briefly, exponential cultures of donor and recipient bacteria were centrifuged, washed, and resuspended in fresh medium. Donor and recipient cells were mixed in a final ratio of 2:1, and 20 μl of this mix was applied to nitrocellulose filters placed on BHC medium containing thymidine (100 μg/ml). The filter matings were incubated overnight at 35°C, and transconjugants were selected on BHC medium containing rifampin (20 μg/ml) and tetracycline (3 μg/ml) and lacking thymidine to avoid the growth of donor Rif^r spontaneous mutants. Transconjugants were confirmed to contain the Tc^r element by PCR by using primers OlirteB1 and OlirteB2 (Table ​(Table2).2). Donor strains were grown in medium containing 1 μg of tetracycline per ml before mating to induce factors in the Tc^r element. The frequency of transfer of the Tc^r element was determined by dividing the number of transconjugants by the number of recipients.

PCR conditions.

PCR primers used in this study are listed in Table ​Table2.2. PCRs were performed with Taq polymerase (1.5 U) in 50 μl containing either chromosomal DNA (20 to 50 ng) or plasmid (5 to 10 ng), primers (25 pmol), deoxynucleoside triphosphates (200 μM), and MgCl₂ (1.5 M). Amplification used a hot start (94°C for 5 min), followed by denaturation at 94°C for 1 min, annealing at 66°C for 2 min, and extension at 72°C for 1 min. The amplification cycle was repeated either 29 times for determination of the integration site of BfPAI or 19 times for generating mpII for the cloning experiments and was followed by a final extension at 72°C for 5 min. The amplified products were directly sequenced or cloned into the TA cloning vector (Invitrogen, San Diego, Calif.).

Data analysis.

Data were analyzed by the chi-square formula; a P value of <0.05 was considered statistically significant.

Nucleotide sequence of the specific 6-kb region.

The colony blot hybridization results indicate that an ~6-kb region flanking bft is present exclusively in all ETBF strains (n = 113). To examine this region for the presence of additional potential virulence genes, in addition to the bft-2 sequence already reported (9), we determined the sequence of 3,045 bp downstream of bft-2 and 5,799 bp upstream of bft-2 (total sequenced region, 10,038 bp) (Fig. ​(Fig.2a).2a). Analysis of the specific 6-kb region revealed that, in addition to bft-2, this region contains two other ORFs (ORF-A and ORF-B), which are 375 and 1,180 bp in length, respectively. ORF-A is transcribed in the opposite direction from bft-2, and its stop codon is located 270 bp downstream of stop codon of bft-2 (Fig. ​(Fig.2a).2a). The predicted amino acid sequence of the protein encoded by ORF-A showed no significant similarity to any sequence in the database. ORF-B also transcribes in the opposite direction of bft-2, with the start codon located 1.6 kb upstream of the start codon of the bft-2 gene (Fig. ​(Fig.2a).2a). The predicted protein encoded by ORF-B is 44.36 kDa with a predicted lipoprotein signal peptide (18 amino acids) and a zinc-binding metalloprotease motif similar to BFT-1 and BFT-2. However, this predicted protein is only 29 and 28% identical to BFT-1 and BFT-2, respectively, and contains an ATP/GTP-binding site motif A (48) not found in BFT-1 or BFT-2. The predicted protein encoded by ORF-B was termed metalloprotease II (MPII) in accordance with the results of Moncrief et al. (24).

Open in a separate window

FIG. 2

Partial restriction map of the region flanking bft in ETBF 86-5443-2-2. The thicker bar (both solid and striped) shows the region sequenced, and the striped bar shows the BfPAI. (a) ORFs detected in the BfPAI and flanking regions. Arrows show the locations of the ORFs and the direction of their transcription. (b) Relative positions of the primers and the expected PCR products to determine the integration site of the BfPAI. Only the restriction sites important for the construction of the probes (Fig. ​(Fig.1)1) are shown in the figure.

To correlate the presence of bft-1/bft-2 and mpII, a specific oligoprobe derived from mpII sequence (Olibft3) was designed. In colony blot hybridization, Olibft3 hybridized with all 113 ETBF strains containing either bft-1 or bft-2, but none of 191 NTBF strains. These results strongly suggest that bft-1/bft-2 and mpII colocalize in ETBF strains.

The G+C content of the specific 6-kb region (35%) differs substantially from that reported for the B. fragilis chromosome (42%) (45). This result and the presence of bft-2 and mpII in the 6-kb locus, which is unique to ETBF strains, suggest that this element is a pathogenicity island or islet (herein designated BfPAI for B. fragilis pathogenicity island).

Nucleotide sequence of the DNA region flanking the BfPAI.

Analysis of the sequence of the left end flanking the BfPAI revealed two ORFs of 540 and 948 bp (Fig. ​(Fig.2a),2a), which encode predicted proteins of 316 and 180 amino acids residues with calculated molecular masses of 21.37 and 37.24 kDa, respectively. A search in the GenBank and EMBL databases for related sequences revealed that these two ORFs, as well as the region immediately upstream of the ORFs, have the same organization as the mobilization region of the 5-nitroimidazole and clindamycin resistance plasmids of B. vulgatus (pIP417) and B. fragilis (pBFTM10), respectively (12, 47). The mobilization regions of both antibiotic resistance plasmids consist of two genes, named btgA and btgB in the clindamycin resistance plasmid, and mobA and mobB in the 5-nitroimidazole resistance plasmid. BtgA/MobA and BtgB/MobB proteins share 49 and 70% identity, respectively. Both genes are organized in an operon, and immediately upstream of the operon there is a putative origin of transfer (oriT). By using the NCBI BLAST server (1), the highest homology score for the predicted 316-amino-acid protein was found with the mobilization protein MobB [P(N)4.5e-15 by BLAST algorithm], with 41% of identity between amino acids 42 and 84, 46% identity between amino acids 176 and 203, 33% identity between amino acids 214 and 234, 46% identity between amino acids 17 and 31, and 38% identity between amino acids 139 and 151. The predicted 180-amino-acid protein shared 25 and 40% identity and similarity, respectively, with a limited region (102 amino acids) of MobA (with both the Gapped BLAST and PSI-BLAST searches). Due to the sequence similarity of the predicted proteins encoded by the ORF of 948 bp (316 amino acids) to MobB and the ORF of 540 bp (180 amino acids) to MobA and the analogous gene arrangement, we designated these ORFs bfmB (B. fragilis mobilization gene B) and bfmA (B. fragilis mobilization gene A), respectively.

Figure ​Figure3A3A shows the comparison of this region upstream of the BfPAI and the mobilization region of plasmid pBFTM10. Similar to btgB/mobB and btgA/mobA, the ATG initiation codon of bfmB overlaps the final codon of bfmA, suggesting that both genes are transcribed together, and the region immediately upstream to these genes revealed several features consistent with the origin of transfer (oriT) of conjugative or mobilizable plasmids (Fig. ​(Fig.3B).3B). This region is ca. 200 bp long and has a higher A+T content than flanking regions (72 versus 56%). The region contains three inverted repeat (IR) sequences (IR1, IR2, and IR3) with lengths of 7, 13, and 9 bp, respectively (Fig. ​(Fig.3B).3B). None of these three IR sequences have similarity with those found in the putative oriT from plasmid pIP417 or pBFTM10. The IR1 and IR3 elements have perfect right (R)- and left (L)-arm inverted nucleotide sequences. Only 11 of the 13 bp of IR2R and IR2L are identical. IR1(R)-IR1(L) and IR3(R)-IR3(L) are separated by 3 and 8 bp, respectively. Approximately 165 bp upstream of the start codon of bfmA a putative nick site was identified according to the consensus sequence established by Pansegrau and Lanka (31) from the IncP plasmid (RK2/RP4) (Fig. ​(Fig.3C).3C). There was no sequence similarity with the proposed nick region of plasmids pIP417 and pBFTM10.

Open in a separate window

FIG. 3

Putative mobilization region of the BfPAI. (a) Comparison between the putative mobilization region of BfPAI and the mobilization region of the clindamycin resistance plasmid pBFTM10. Arrows show the direction of gene transcription. (b) Nucleotide sequence (5′ to 3′) of the three IRs. (c) Nick site present in the putative oriT gene of the BfPAI.

The right end flanking the BfPAI contained an ORF (designated bfmC) of 1,602 bp which transcribes in the same direction as bft-2 but in the opposite direction of bfmA and bfmB (Fig. ​(Fig.2a).2a). The bfmC gene yields a predicted protein product (BfmC) of 534 residues with a molecular mass of 61.0 kDa. By using the BLAST search program, BfmC shares the highest homology score with a conjugal transfer protein TrsK of conjugative bacteriocin-producing plasmid pMRC01 from Lactococcus lactis [P(N)5e-09] (5) sharing 22% identity and 41% similarity over 395 amino acids. BfmC also shares significant similarity with mobilization proteins TraG of Helicobacter pylori [P(N)1.5e-07], TrsK of plasmid pGO1 from Staphylococcus aureus [P(N)3e-04], TraD of plasmids F and R100 from E. coli [P(N)0.003], and TrbC of Salmonella typhimurium [P(N)0.092] (14, 25, 29, 52). BfmC contains a perfect ATP/GTP-binding motif A (48) at the amino-terminal region; the same motif appears to be present in TrsK encoded by plasmids pMRCO1, as well as TraD encoded by plasmids F and R100. Alignment of these proteins in the region surrounding this motif is shown in Fig. ​Fig.4.4. The organization of the left and right ends flanking the BfPAI suggests that the BfPAI may have integrated into a plasmid. However, several different types of plasmid preparation procedures performed on ETBF strain 86-5443-2-2 failed to reproducibly reveal detectable plasmid DNA (data not shown).

Open in a separate window

FIG. 4

Alignment of the region surrounding the ATP/GTP-binding site motif A in BfmC, TrsK (pMRCO1), TrsK (pGO1), TraD (R100), and TraD (F). Amino acids similar to those of the BfmC sequence are in capital letters and boldface. The ATP/GTP-binding site motif A in BfmC is underlined.

Analysis of the G+C content of the left-end (47%) and right-end (50%) sequences flanking the BfPAI revealed that the G+C content of these regions is significantly different from that of the BfPAI (Fig. ​(Fig.5).5). Interestingly, the G+C content of the regions immediately upstream of the bfmA/bfmB operon, bft, mpII, and bfmC all decrease sharply (asterisks in Fig. ​Fig.5).5). Similar results were found by Smith in the DNA sequences of transposon Tn4555 and plasmid pBI143 (45a). This may be due to the fact that promoter regions often have higher A+T content to facilitate separation of DNA for the initiation of transcription. The differing G+C contents of the BfPAI, the flanking region, and the B. fragilis chromosome suggest that the BfPAI is contained in yet another foreign genetic element.

Open in a separate window

FIG. 5

G+C content across the BfPAI and its flanking regions. Arrows show the location of genes and the direction of their transcription. Asterisks show the sharp declines in G+C content immediately upstream of the designated genes.

Determination of the integration site of the BfPAI.

Based on the results of colony blot hybridization (Fig. ​(Fig.1)1) and the above sequence analysis, we hypothesized that ETBF strains could be generated by the integration of the BfPAI in certain NTBF strains (hybridization pattern III). Since the exact left and right borders of the BfPAI were contained in the DNA region spanning probes A and D, respectively (Fig. ​(Fig.1),1), two primers (P1T7 and P1T3 sequences in Table ​Table2)2) for these regions were designed (Fig. ​(Fig.2b)2b) and used in PCR experiments to amplify DNA from one strain of each pattern of hybridization. As expected, PCR yielded a product (ca. 1.6 kb) only in the NTBF strain with patterns III (strain I-1345) and IV (strain LM20) but not in any strain with the other patterns of hybridization (Fig. ​(Fig.6A).6A). NTBF strains of patterns II and V did not hybridize with probes A and/or D, which contain primers P1T7 and/or P1T3, respectively (Fig. ​(Fig.1),1), and in ETBF strains (pattern I) primers P1T7 and P1T3 flank a ca. 7-kb region (Fig. ​(Fig.2b),2b), a fragment too large to be amplified by Taq DNA polymerase under the conditions tested. To determine whether the integration of ETBF is site specific, we examined five additional NTBF strains with hybridization pattern III. All strains yielded products of 1.6 kb similar to strain I-1345 (data not shown). In addition, we designed two new primers (P1T7-1 and P1T3-1) containing sequences for the region in the BfPAI spanning probes G and B, respectively (Fig. ​(Fig.2b).2b). When P1T7 and P1T7-1 were used as primers, PCR experiments with six ETBF strains yielded the same predicted product (1.3 kb) (Fig. ​(Fig.6B).6B). Similarly, the six ETBF strains yielded the predicted 1.1-kb product when P1T3 and P1T3-1 were used as primers (Fig. ​(Fig.6C).6C). All of these results indicate that the integration of the BfPAI is in the same small region and in the same orientation in the cases examined and suggest that the integration of the BfPAI is site specific and unidirectional.

Open in a separate window

FIG. 6

PCR analysis of the left and right junctions of BfPAI. (A) Screening for the integration site by using primers P1T7 and P1T3 (see Fig. ​Fig.2b).2b). PCR yielded a product (1.6 kb) only in NTBF strains with pattern III (strain I-1345) and IV (strain LM20). Lanes: 1, 1-kb ladder molecular weight marker; 2, ETBF 86-5443-2-2 (pattern I, positive control); 3, NTBF 077225-2 (pattern II); 4, NTBF I-1345 (pattern III); 5, NTBF LM-20 (pattern IV); 6, NTBF LM-12 (pattern V). (B) Screening for the left junction by using primers P1T7 and P1T7-1 (see Fig. ​Fig.2b).2b). PCR yielded a similar-sized fragment (1.3 kb) in six ETBF strains (pattern I) and NTBF LM12 (pattern V). Lanes: 1, 1-kb ladder molecular weight marker; 2, ETBF 86-5443-2-2 (pattern I); 3, ETBF VPI 13784 (pattern I); 4, ETBF J-38-1 (pattern I); 5, ETBF DS-64 (pattern I); 6, ETBF DS-233 (pattern I); 7, ETBF DS-49 (pattern I); 8, NTBF 077225-2 (pattern II); 9, NTBF I-1345 (pattern III); 10, NTBF LM-20 (pattern IV); 11, NTBF LM-12 (pattern V). (C) Screening for the right junction by using primers P1T3-1 and P1T3 (see Fig. ​Fig.2b).2b). PCR yielded a similar-sized fragment (1.1 kb) in all six ETBF strains (pattern I). Lanes are as described for panel B.

To determine the actual integration site and the exact borders of the BfPAI, we sequenced the 1.6-kb PCR product obtained from NTBF strain I-1345 by using primers P1T7 and P1T3. Comparison of this sequence and the appropriate sequence of ETBF strain 86-5443-2-2 defined the integration site and the putative left and right borders of the BfPAI (Fig. ​(Fig.7).7). Integration of BfPAI in ETBF strains is predicted to lie between the stop codons of bfmB and bfmC. ETBF strains lack a span of 16 bp at the integration site which is present in pattern III NTBF strain I-1345. According to this alignment, the BfPAI is calculated to be 6,036 bp in length. Analysis of the borders of the BfPAI revealed nearly perfect 12-bp direct repeats present 45 and 40 bp from the left and right ends of the BfPAI, respectively (data not shown).

Open in a separate window

FIG. 7

Comparison of the nucleotide-amino acid sequences of the region flanking the BfPAI of ETBF strain 86-5443-2-2 with the equivalent region of NTBF strain I-1345. Nucleotide sequences belonging to bfmB and bfmC are given in capital letters. Arrows show direction of gene transcription, and stop codons are indicated by asterisks. The 16-bp sequence lost in ETBF is given in lowercase letters.

Sequence comparison of 86-5443-2-2 and VPI 13784 BfPAIs.

Moncrief et al. (24) recently reported the sequence of the pathogenicity island (islet) from an ETBF strain containing bft subtype 1 (strain VPI 13784). Comparison of the BfPAI sequence from strain 86-5443-2-2 and the reported sequence from ETBF strain VPI 13784 reveals that the two sequences are not identical. A summary of this comparison is shown in Table ​Table5.5. Overall, both sequences share 97% identity. However, the BfPAI region containing the mpII gene is more conserved than the region containing bft (99 versus 95% identity, respectively). Similarly, the predicted proteins encoded by mpII from ETBF 86-5443-2-2 and VPI 13784 are 99% identical; whereas BFT-1 from VPI 13784 and BFT-2 from 86-5443-2-2 are 92% identical (9).

TABLE 5

Sequence comparison of the BfPAI from strains 86-5443-2-2 and VPI13784

Nucleotide sequence or coding region	Value
Nucleotide sequence
Lengtha	6,036 bp (86-5443-2-2), 6,032 bp (VPI 13784)
Overall % identity	97
Overall number of gaps	33
% Identity of region surrounding bft (bp 1 to 3,232)	95
No. of gaps in region surrounding bft (bp 1 to 3,232)	33
% Identity of region surrounding mpII (bp 3,233 to 6,036)	99
No. of gaps in region surrounding mpII (bp 3,233 to 6,036)	0
% Identity of partial direct repeats at the ends	100
Coding region
% Identity of BFT	92
% Identity of MPII	99
ORF-A	Not present in VPI 13784
ORF1	Not present in 86-5443-2-2

Open in a separate window

^aDefined by using 16 bp as the target sequence for integration of the BfPAI (see Fig. ​Fig.77). 

Near the left end of the bft gene from strain VPI 13784, Kling et al. identified an ORF (ORF1) of 261 nucleotides that encodes a predicted protein with homology to a snake cytotoxin (16). This ORF1 was not identified in the BfPAI from strain 86-5443-2-2. Deletion of nucleotide 16 and the addition of one nucleotide between bases 47 and 48 and bases 142 and 143 of ORF1 changes the frame three times in strain 86-5443-2-2. To determine whether ORF1 was specific for ETBF strains containing bft-1, we amplified by PCR and sequenced the region spanning ORF1 from a strain containing bft-1 (strain J38-1) and a strain containing bft-2 (strain 20793-3). We found that the nucleotide sequence of strains J38-1 and 20793-3 were identical to the appropriate sequences of strains VPI 13784 and 86-5443-2-2, respectively. Conversely, we also found that ORF-A detected in BfPAI from strain 86-5443-2-2 (Fig. ​(Fig.2a)2a) was not present in strain VPI 13784. The deletion of nucleotide 60 and the addition of one nucleotide between bases 172 and 173 of ORF-A change the frame twice in strain VPI 13784. Sequence analysis revealed that ORF-A was not present in either strain J38-1 or 20793-3. However, the region spanning this ORF in strain 20793-3 shared 99% similarity with that from 86-5443-2-2 but only 95% similarity with that of VPI 13784, whereas the same region from strain J38-1 shares 99% similarity with that of strain VPI 13784 and 95% similarity with that of strain 86-5443-2-2. The role of the proteins encoded by ORF1 and ORF-A in the virulence of strains VPI 13784 and 86-5443-2-2, respectively, if any, remains to be determined.

The near-direct repeat sequences present at the borders of both BfPAIs are identical. We found that the insertion site of the BfPAI in strain 86-5443-2-2 is similar to that of strain VPI 13784 and that the precise location of the BfPAI is between the stop codons of bfmB and bfmC (Fig. ​(Fig.7).7). Our results indicate that the first nucleotide (G) of the 17 bp found in the NTBF strains (but not in the ETBF strains) reported by Moncrief et al. (24) is part of the bfmC stop codon, suggesting that 16 bp may be the target sequence for the integration site of the BfPAI and that the size of the VPI 13784 BfPAI is 6,032 bp. However, we cannot rule out the possibility that the last base of the bfmC codon is part of the target sequence for the integration of the BfPAI.

Mobilization of the BfPAI.

Analysis of the nucleotide sequence of the DNA region flanking the BfPAI suggested that bft-1/bft-2 and mpII are in a genetic element that may be mobilized from ETBF to NTBF strains. Transfer of Bacteroides plasmids pIP417 and pBFTM10 is linked to the presence of the Bacteroides tetracycline resistance transposons (termed Tc^r elements) (12, 36, 47). Tc^r elements are large chromosomal self-transmissible elements (ca. 65 to 150 kb) that also facilitate the mobilization of other elements, such as coresident plasmids, other transposons, and discrete unlinked 10- to 12-kb segments of chromosomal DNA nonreplicating Bacteroides units (NBUs) (36). Interestingly, the transfer efficiency of Tc^r elements is enhanced by pretreatment of donor cells with concentrations up to 1 μg of tetracycline per ml. Under this condition, Tc^r elements may be transferred in a frequency of 10⁻³ to 10⁻⁷, and ca. 10% of transconjugants will contain Tc^r elements which cotransferred with plasmids, other transposons, or NBU elements (36). To determine whether Tc^r elements can mobilize the transfer of BfPAI, we performed mating experiments between ETBF strain 086Thy⁻3 (a thymidine-dependent mutant strain derived from ETBF 86-5443-2-2 which contains an inducible Tc^r transposon [see Materials and Methods]) and NTBF strain TM4000 (tetracycline sensitive). However, under different conditions of induction, we failed to detect tetracycline-resistant transconjugants, indicating that the Tc^r element of ETBF 086Ty⁻3 was not mobilized. When we used the mating positive controls, B. thetaiotaomicron strains BT4107 (donor) and BT4001 (recipient), transfer of Tc^r element was observed at a frequency of 1.3 × 10⁻⁵ to 2.4 × 10⁻⁶. These results suggested that the Tc^r element is not conjugative in strain 086Thy⁻3, or, alternatively, that BFT and/or MPII may inhibit the mobilization of the Tc^r element. To test these possibilities, we selected ETBF strain I-1356-1Thy-1 (containing bft subtype 1) as donor strains, as well as NTBF strains K518Thy⁻2 (pattern II) and I-1345Thy⁻2 (pattern III). No transfer of Tc^r element was observed under any of the conditions of induction used when ETBF I-1356-1Thy-1 or NTBF K518Thy⁻2 were used as the donor strains. However, the Tc^r element was transferred at a frequency of 1.5 × 10⁻⁵ to 1.0 × 10⁻⁶ when NTBF I-1345Thy⁻2 was used as a donor strain. To determine whether the region flanking BfPAI was cotransferred with the Tc^r element in strain I-1345Thy⁻2, we performed colony blot of 600 Tc^r transconjugants by using the region spanning bfmB as a probe. No transfer of the region flanking BfPAI was detected in any of the 600 tetracycline-resistant transconjugants probed.

We thank Dwight Derr for tissue culture assistance; Jeffrey C. Smith, James Nataro, and Davis Karaolis for review of the manuscript; Nadja B. Shoemaker for the gifts of B. thetaiotaomicron strains and suggestions about the mobilization experiments; Jeffrey C. Smith for the parsimony analysis of bfmB; and Lyle L. Myers, R. Bradley Sack, Jay Reuben, Tracy D. Wilkins, and Peter Escheverria for the gifts of B. fragilis strains.

This work was supported by National Research Service Award AI09863-01 (A.A.F.) and National Institutes of Health Award DK45496 (C.L.S.).