Cryptic Plasmids of Mycobacterium Avium: Tn552 to the Rescue

Molecular Microbiology (2002) 43(1), 173–186

Cryptic plasmids of Mycobacterium avium: Tn552 to the rescue

Carolyn Kirby,1 Al Waring,2 Thomas J. Griffin IV,3 with other members of the M. avium complex (MAC), is Joseph O. Falkinham III,4 Nigel D. F. Grindley3 an opportunistic pathogen that causes a variety of and Keith M. Derbyshire1,2* diseases in animals and humans (Inderlied et al., 1993). 1Department of Biomedical Sciences, State University of MAC isolates from both clinical and environmental New York at Albany, NY, USA. sources often contain naturally occurring plasmids of 2Division of Infectious Disease, Wadsworth Center, varying size (10 to >100 kb) (Crawford and Falkinham, NYS Department of Health, NY, USA. 1990; Falkinham and Crawford, 1994; Pashley and 3Department of Molecular Biophysics and Biochemistry, Stoker, 2000). Southern hybridization studies on crude Yale University, New Haven, CT, USA. extracts of plasmid DNA from independent isolates 4Department of Biology, Virginia Polytechnic Institute, have indicated that some of these plasmids are related Blacksburg, VA, USA. (Jucker and Falkinham, 1990). Based on these studies, plasmids from M. avium, Mycobacterium intracellulare and Mycobacterium scrofulaceum isolates have been Summary assigned to at least four related groups (Jucker, 1991). Plasmids have been described in almost all bacterial The widespread occurrence of related plasmids from species analysed and have proven to be essential different isolates suggests that the plasmids have the genetic tools. In many bacteria these extrachro- ability to transfer between mycobacterial species in the mosomal DNAs are cryptic with no known markers environment. In addition, multiple plasmids are often or function, which makes their characterization and found in the same isolate, showing that these plasmids genetic exploitation extremely difficult. Here we can stably coexist and, therefore, that they represent describe a system that will allow the rescue of any cir- different plasmid incompatibility groups. Little is known cular DNA (plasmid or phage) using an in vitro trans- about these plasmids, their mechanism of replication and position system to deliver both a selectable marker coexistence, the genes that they encode or whether the (kanamycin) and an Escherichia coli plasmid origin of plasmids play any role in virulence. Various reports have replication. In this study, we demonstrate the rescue of speculated that they may contribute to virulence and also four cryptic plasmids from the opportunistic pathogen may encode resistance to metals, restriction modification Mycobacterium avium. To evaluate the host range of systems and metabolic enzymes (Crawford et al., 1981; the rescued plasmids, we have examined their ability Meissner and Falkinham, 1984; Erardi et al., 1987). to be propagated in Mycobacterium smegmatis and However, the inability to cure plasmids from these iso- Mycobacterium bovis BCG, and their compatibility lates, as well as the lack of selectable markers and an with other mycobacterial plasmids. In addition, we use understanding of the genetic makeup of these MAC a library of transposon insertions to sequence one plasmids, have prevented both rigorous analysis of plasmid, pVT2, and to begin a genetic analysis of their genetic composition and confirmation of plasmid- plasmid genes. Using this approach, we identified a associated functions (Falkinham and Crawford, 1994; putative conjugative relaxase, suggesting this myco- Pashley and Stoker, 2000). bacterial plasmid is transferable, and three genes Plasmids are important tools for the genetic analysis of required for plasmid establishment and replication. any organism. They provide a means to clone, comple- ment and express host genes, and to introduce selected genes into different genetic backgrounds. The mycobac- Introduction teria lack fully characterized, compatible plasmid systems Mycobacterium avium is a naturally occurring, slow- to facilitate genetic manipulation. Most mycobacterial growing mycobacterium. Mycobacterium avium, along plasmids are based on the low copy number vector, pAL5000 (Labidi et al., 1985; Pashley and Stoker, 2000). More recently, a second family of plasmids, the pMSC262 Accepted 27 September, 2001. *For correspondence. David Axelrod or pLR7 family, has been described (Guilhot et al., 1999; Institute, NYS Department of Health, PO Box 22002, Albany, NY 12201–2002, USA; E-mail [email protected]; Tel. Pashley and Stoker, 2000). These plasmids (pMSC262, (+1) 518 473 6079; Fax (+1) 518 486 7971. pLR7, pJAZ38, pMF1 and pCLP) are all compatible with

© 2002 US Government 174 C. Kirby et al. pAL5000 (Qin et al., 1994; Gavigan et al., 1997; Bachrach The in vitro transposition reactions were electroporated et al., 2000; Picardeau et al., 2000). Except for one into E. coli and kanamycin-resistant (Kmr) transformants member, pCLP, these plasmids have not been completely selected. Restriction analysis of plasmid DNA from these sequenced or characterized, and only regions associated transformants revealed that a significant proportion of with replication have been identified (Le Dantec et al., these were circular transposons, in which Tn552 had 2001). Thus, there is still a clear demand for additional, inserted into itself close to one end, but outside regions well-characterized mycobacterial plasmids. The develop- necessary for replication and kanamycin resistance. ment and characterization of additional plasmid vectors These transposon circles are most probably a conse- will not only provide significant new insight into plasmid quence of the low amounts of target DNA in the reaction biology in this species, but it will also facilitate, inevitably, and are not normally seen under standard reaction the genetic analysis of all mycobacteria, including the conditions (Griffin et al., 1999). To enrich for insertions major human pathogen Mycobacterium tuberculosis. into the M. avium plasmids, all the colonies from a A major hurdle in characterizing plasmids from organ- single electroporation were pooled and plasmid DNA isms for which there exist only a limited number of genetic was isolated. This DNA was then subjected to gel elec- tools is the inability to purify plasmid DNA and to intro- trophoresis, without digestion, to separate large plasmid duce it into more genetically amenable organisms such DNA from the smaller transposon DNA circles. The larger as Escherichia coli. Here we describe a novel way to circular species were purified from the agarose gel and rescue cryptic plasmids from crude DNA extracts using retransformed into E. coli. This enrichment step ensured an in vitro transposition system. A derivative of Tn552, that all the plasmid DNAs screened were plasmids Tn552kan∑ori, has been constructed which contains a rescued from the M. avium extract rather than transposon selectable marker and an E. coli plasmid origin of repli- circles. cation (Griffin et al., 1999). In vitro transposition of this element into a plasmid DNA confers on the plasmid the Restriction and Southern analysis of rescued ability to replicate, and be selected, in E. coli. We show plasmid DNAs that this system is very robust and can be used to rescue different coexisting plasmids from a single extract. A Plasmid DNAs of Kmr transformants from each extract library of transposon insertions was also used to deter- were analysed by SalI restriction digestion. Analysis of the mine the DNA sequence of one of the plasmids, and restriction patterns suggested that at least four plasmid allowed regions required for plasmid replication to be types had been rescued from the crude extracts, which is genetically defined. entirely consistent with the number of plasmids originally observed in each isolate by Jucker and Falkinham (1990). All the plasmids rescued from the MD1 extract had a Results similar profile (Fig. 1, left); consistent with rescue of a single plasmid and with the prediction that MD1 contained Rescue of cryptic plasmids only one plasmid, pVT2. Single-fragment differences in Crude DNA extracts from two M. avium isolates were individual plasmid profiles are a result of insertion of the used as a substrate in an in vitro transposition reaction 2 kb Tn552 transposon. Plasmids with three different using Tn552kan∑ori. One isolate, MD1, was thought to restriction patterns were rescued from MD22, consistent contain a single plasmid, pVT2 (Jucker and Falkinham, with this strain carrying more than one plasmid (Fig. 1, 1990). The second strain, MD22, had been predicted right). The three profiles were classed as types A, B and to contain up to four different plasmids based on Sou- C, with 26, 14 and 8 plasmids assigned to each class thern hybridization analyses (Jucker and Falkinham, respectively. Southern hybridization confirmed that all 1990). Samples of these DNA extracts were examined by plasmids within a group show complete cross-hybridiza- agarose gel electrophoresis and ethidium bromide stain- tion and, therefore, are independent insertions into the ing, which identified a potential plasmid species in the same plasmid (data not shown). There were also several MD1 extract, in addition to chromosomal DNA (data plasmids that had unique profiles, which did not belong to not shown). By contrast, only chromosomal DNA was any of the four classes, and they have not been studied detected in the MD22 extract. This may be a consequence further. of the difficulty in purifying plasmid DNA from this species, To confirm the assignment of these plasmids to differ- and/or that these samples had been stored for 14 years ent classes, a representative plasmid from each group and may have degraded. However, it emphasizes the was used for Southern analyses. A lack of strong sensitivity of the in vitro transposition system, as it cross-hybridization between the plasmids would indicate enabled rescue of at least three distinct plasmids from that they were not closely related. Each plasmid was MD22 (see below). labelled and used as a probe against duplicate filters. The

Fig. 2. Southern analysis of M. avium plasmid groups. Plasmids from each group were digested with SalI and probed with a member of each group. The plasmid used as a probe is shown in Fig. 1. Restriction analysis of rescued plasmids identifies four bold type above the panel. One restriction fragment in each distinct plasmid types. Representative plasmids of each type are plasmid will cross-hybridize with the probe, as it contains the shown digested with SalI. Plasmid fragment sizes are shown on the Tn552 insertion. The restriction patterns correspond to those shown left of each panel. The Tn552kan∑ori insertion causes an increase in the left-hand lane of each panel in Figure1. in size of one fragment by 2 kb; this novel fragment is not included in the fragment sizes shown at the left. analysis shows that plasmid types A, B and C are unre- undergone any gross rearrangement and, more impor- lated, but that the group A plasmid and pVT2 show some tantly, that it was the same as that originally introduced cross-hybridization (Fig. 2). Note that in the analysis at into BCG (data not shown). least one fragment cross-hybridizes with the probe as a Plasmid DNAs were also electroporated into the fast result of the presence of the Tn552 insertion. The lack of growing species M. smegmatis. However, transformants significant cross-hybridization between plasmid types A, containing pVT2 and plasmids from groups A and C had B and C is consistent with their ability to co-exist in MD22. an unusual phenotype: colonies took up to 14 days Although the pVT2 and type A plasmids exhibit some to appear on selective media instead of the normal cross-hybridization, they have different properties and are 3–4 days. The delayed growth phenotype varied be- compatible with one another (see below). tween the plasmid classes; group A transformants took 7–8 days to grow, whereas pVT2 and C transformants took 12–14 days. In each case, it appears that the initial Rescued plasmids can be introduced into M. bovis BCG establishment of the plasmid delays growth. When trans- and M. smegmatis formants were restreaked onto fresh selective media they To confirm that these plasmids were of mycobacterial formed colonies within a normal timespan (2–3 days). The origin, plasmids were introduced into the slow-growing presence of the Kmr gene was demonstrated by poly- mycobacterium M. bovis BCG by electroporation. This merase chain reaction (PCR), and we have rescued plas- species was chosen, rather than M. avium, as efficient mids from these transformants by electroduction into E. and reproducible electroporation conditions have been coli, confirming that Kmr was plasmid-encoded. In addi- established for this organism (Wards and Collins, 1996). tion, restriction analysis of these rescued plasmids con- Two different plasmid derivatives of each type were used firms that they are identical to the parental DNA. However, to ensure that the Tn552 insertion was not in a gene we have noted that electroduction efficiencies are less essential for plasmid replication in mycobacteria. All four reproducible with these plasmids compared with type B plasmid types were successfully transformed into M. plasmids and the pMD30 control. We suspect that this is bovis BCG with transformation efficiencies similar to, or related to the slow-growth phenotype. greater than, the pAL5000-based control plasmid pMD30 (>103 transformants per mg of plasmid DNA). Representa- Isolation of an ept mutant of mc 2155 that allows tive transformants were selected and their plasmid DNA transformation of M. avium plasmids was electroduced back into E. coli. Restriction analysis of this plasmid DNA confirmed that the plasmid had not We show later that pVT2 is related to the pMSC262 family

© 2002 US Government, Molecular Microbiology, 43, 173–186 176 C. Kirby et al. of mycobacterial plasmids and note that derivatives of at least two members of this family, pLR7 and pMSC262, are apparently unable to replicate in M. smegmatis (Goto et al., 1991; Beggs et al., 1995). Given our observation of this delayed growth phenotype, it is possible that these two plasmids may be able to transform M. smegmatis, but in those initial studies colony growth was not monitored for such an extended time. We have also tested the hypothesis that these delayed-growth transformants are host mutants that have acquired a mutation that allows plasmid establishment. This mutation would be functionally equivalent to the efficient plasmid transformation mutation (ept) originally described in the generation of mc2155 (Snapper et al., 1990), but would extend the range of plasmids capable of replicating in M. smegmatis. The class A plasmids are unstable in the absence of selection in M. smegmatis (see below). Therefore, a plasmid-free segregant was purified from a class A plasmid transformant and used to make electrocompetent Fig. 3. Plasmid stability and compatibility. Each graph shows the extent of plasmid retention when cells are grown in the absence of cells. Representatives of each of the M. avium plasmids antibiotic selection. Kmr colonies containing the plasmids indicated and a control plasmid, pMD30, were electroporated into were grown and samples taken at the time indicated. Cells were the cured strain and the parent strain mc2155. The colony plated on media with and without Km to determine the percentage of cells retaining the plasmid. Each graph shows the results of one growth rate of pMD30 transformants was not noticeably experiment; however, as can be seen from each graph, plasmid affected, although the transformation efficiency was retention was reproducibly in the 50–100% range for each slightly higher (two- to threefold). However, transformants independent series of experiments, except in M. smegmatis (graph D). of all rescued plasmids grew significantly faster, confirm- A. Plasmid stability in M. bovis, BCG. ing that the cured strain had acquired an ept type muta- B. Plasmid stability and compatibility between pVT2 and plasmid r tion in a new gene that we will designate ept2. Most types A, B and C and pBP10, a derivative of pMF1. A Hyg derivative of pVT2, pCK5, was used in these experiments to dramatically, colonies containing plasmid types pVT2 and monitor compatibility between pVT2 and the other plasmids, which A were detectable within 4 days, compared with 7 and were Kmr. 12 days, respectively, in mc2155. Type C transformants C. Plasmid stability and compatibility between each M. avium plasmid and a Hygr pAL5000-derived plasmid, pLP961. were detectable after 5–6 days of incubation, compared D. Plasmid stability in M. smegmatis. with 12 days in mc2155. Transformants containing type B Key to symbols: , pVT2, or its Hygr derivative pCK5 (in graph B); plasmids also grew faster; colonies were detected after , type A plasmid; , type B plasmid; , type C plasmid; in graph B, open box with cross, pBP10; in graph D, open box with dashed 3 days, rather than the 4 days when they were introduced line, pMD30. In graphs B and C, plasmid retention indicates those into mc2155. The overall transformation efficiency for cells carrying both plasmids as determined by plating cells on pVT2, A and C plasmids was also significantly (at least media containing kanamycin and hygromycin. 50-fold) more efficient into the ept2 mutant strain than transformation into mc2155. To develop these plasmids as vectors for use in M. avium and other slow-growing species such as M. tuber- Plasmid stability and compatibility culosis, it was important to establish that these plasmids Each of the plasmids was stable in M. bovis BCG in the can co-exist with other mycobacterial plasmids. It is prob- absence of antibiotic selection. Plasmid transformants able that all three plasmids isolated from strain MD22 are were cultured without selection for 21 days (approxi- compatible; thus, we have focused on the compatibility mately 20 generations). During this time, samples were of pVT2 with these three plasmids, with pAL5000, the taken to monitor cell growth and plasmid retention by most commonly used mycobacterial vector (Snapper plating on appropriate selective media. All plasmids were et al., 1990), and with pMF1, a plasmid belonging to the retained with high efficiency (Fig. 3A). The presence of pMSC262 family (Bachrach et al. 2000). each plasmid in cells from the final plating was confirmed The Kmr gene of a pVT2::Tn552 derivative was by PCR amplification of the Kmr genes and by electro- replaced with a gene encoding hygromycin resistance duction into, and purification of, plasmid DNA from E. coli, (Hygr) to generate pCK5. pCK5 was electroporated into followed by restriction analysis. BCG and then this strain was used as a recipient for trans-

© 2002 US Government, Molecular Microbiology, 43, 173–186 Rescue of cryptic mycobacterial plasmids with Tn552 177 formation by Kmr plasmid types A, B and C. In all three sequencing, without the need to subclone smaller DNA cases Kmr, Hygr transformants were obtained, demon- fragments into plasmid vectors. Eighty-two independent strating that these plasmids can coexist. To assess pVT2::Tn552kan∑ori derivatives were used for sequence their compatibility, we measured plasmid stability in the determination. In addition, a second generation of in vitro absence of antibiotic selection, as described above. After insertions using an ampicillin-resistant Tn552 derivative, 21 days (approximately 20 generations) the majority of Tn552bla, were used to expand the library. Five Kmr the cells retained both plasmids (Fig. 3B). The presence pVT2::Tn552kan∑ori derivatives from the first library of each plasmid was confirmed by PCR amplification and were used as targets for in vitro transposition using the by electroduction. This demonstrates that plasmid types 1 kb Tn552bla transposon. This generated five inde- A, B, C and pVT2 must belong to different incompatibility pendent insertion libraries that could be used for both groups. This is also consistent with the assignment of sequence and functional analyses. From these second- these plasmids to different groups based on restriction generation libraries, 39 plasmids were used for sequence analysis and lack of homology as detected by Southern determination. hybridization (Figs 1 and 2). Similarly, pCK5 and a Kmr Using primers specific for the transposon, a total of 229 derivative of pMF1, pBP10 (Bachrach et al., 2000), were sequence reads were compiled to generate a 12 868 bp shown to co-exist stably in BCG without antibiotic selec- contig (GenBank accession no. AY056023). The complete tion (Fig. 3B), and therefore. they must belong to differ- sequence of both strands was determined and each base ent incompatibility groups. This was of particular interest, was sequenced on average 6.6 times. Restriction digest as we later show that pVT2 belongs to the pMSC262 patterns of pVT2 were consistent with those predicted class of plasmids, and shares significant sequence simi- from the sequence, confirming the accuracy of the larity with pMF1 in the replication region. sequence assembly. The GC content of pVT2 is 66.4%, A Hygr derivative of pAL5000, pLP961 (Parsons et al., consistent with its mycobacterial origin. Although we 1998), was introduced into BCG, and this strain was then define the sequence as 12 868 bp in length, there was a used as a recipient for electroporation with pVT2 and small segment (54 bp) that was very difficult to sequence plasmid types A, B and C. In all four cases, Kmr/Hygr on either strand (plasmid co-ordinates 12 846–32). The transformants were obtained, demonstrating that these sequence of this region was eventually determined by plasmids can coexist with pAL5000. By contrast, two PCR amplification of this segment of DNA, followed by differently marked derivatives of pAL5000 could not sequence analysis. High-resolution restriction mapping be co-transformed into BCG (data not shown). Kmr/Hygr of the plasmid DNA across this region was consistent transformants were then cultured in the absence of selec- with the length of the sequence predicted by sequence tion to monitor plasmid retention. For each plasmid com- analysis confirming the fidelity of the PCR reaction (data bination, more than 50% of the cells were resistant to both not shown). We have since heard of other examples of hygromycin and kanamycin after 21 days of growth mycobacterial DNA being difficult to sequence and con- (Fig. 3C). As before, PCR and electroduction were used sisting of poly-G tracts (G. Hatfull, personal communi- to confirm the presence of each plasmid type in the cation). However, we did not detect a poly-G tract or colonies plated after 21 days of growth. an obvious secondary structure in this region that In contrast to the stability of the plasmids in BCG, might impede the DNA polymerase used in sequencing plasmid types A and B were relatively unstable in mc2155 reactions. in the absence of antibiotic selection. Only 1% of the cells Seven open reading frames (orfs) that are probably retained these plasmids after 48 h of growth (approxi- expressed were identified by sequence analysis and mately 24 generations). To rule out the possibility that database comparisons (Fig. 4 and Table 1). Two of these instability was caused by a mutant phenotype associated orfs encode proteins (Orfs) that share extensive similar- with the Tn552 insertion, two other insertion derivatives of ity with hypothetical proteins of unknown function. orf3 each plasmid were examined and shown to be as unsta- would encode a protein of 257 amino acids with signifi- ble as those in Fig. 3D (data not shown). Both pVT2 and cant amino acid identity to similar-sized proteins from type C plasmids were as stable as observed in BCG several species including: Streptomyces (54% identity, (Fig. 3D). 67% similarity), Corynebacterium glutamicum (52%, 68%) and Aquifex aeolicus (49%, 66%). Orf4 is similar to several hypothetical proteins and lipoproteins. The most DNA sequence analysis closely related are probable lipoproteins from M. tuber- pVT2 was selected for a more detailed genetic analysis culosis H37Rv (43% identity, 60% similarity) and from M. by taking advantage of the library of Tn552 insertions to leprae (41%, 60%). determine its entire DNA sequence. The transposon inser- Four Orfs share extensive sequence identity with tions provided a complete set of priming sites for DNA proteins of known function: these will be discussed more

Fig. 4. Genetic map of pVT2. Arrows below indicate ORFs, their locations in basepairs (bp), and protein lengths in amino acids (aa). The probable function of the Orf’s based on sequence similarity to known proteins is shown above each arrow. The approximately 4 kb region that is shaded was shown to be essential for mycobacterial plasmid establishment and maintenance (see Results). The extent of DNA regions showing high degrees of similarity to two mycobacterial plasmids, pLR7 and pMSC262, are shown below the replication region as solid lines. Restriction sites used for restriction analysis and cloning are indicated by symbols: , SalI; , BamH I; , HindIII; , EcoRI.

Table 1. pVT2 proteins and their homologues.

Size % Identity/ Reference Protein (aa) Homologue Organism Function % similarity E-value /accession no.

Orf1 391 pLR7 RepA M. avium Putative replication protein 67/74 e–100 Beggs et al. (1995) pJAZ38 RepA M. fortuitum Putative replication protein 50/58 1e–67 Gavigan et al. (1997) pMSC262 RepA M. fortuitum Putative replication protein 54/57 2e–57 Qin et al. (1994)

Orf2 942 pCG4 TraA Corynebacterium Transfer protein homologue 34/45 e–113 AF164956 glutamicum pTIC58 TraA Agrobacterium Probable conjugative 25/40 2e–17 Farrand et al. (1996) transfer protein R100 DNA Helicase I E. coli Conjugative helicase 24/38 3e–15 Yoshioka et al. (1990) R388 TrwC Helicase E. coli Conjugative helicase 22/34 2e–14 Llosa et al. (1994)

Orf3 257 Hypothetical protein Streptomyces Unknown 54/67 2e–54 Redenbach et al. (1996) coelicolor A3(2) Hypothetical protein Aquifex aeolicus Unknown 49/66 1e–47 Deckert et al. (1998) Hypothetical protein Corynebacterium Unknown 52/68 6e–47 Wehrmann et al. (1994) glutamicum

Orf4 415 Probable LppS M. tuberculosis Probable lipoprotein 43/60 7e–72 Cole et al. (1998) H37RV Hypothetical protein M. leprae Unknown 41/60 2e–70 Cole et al. (2001) Putative lipoprotein Streptomyces Putative lipoprotein 30/48 2e–39 Redenbach et al. (1996) coelicolor A3(2) Possible lipoprotein M. leprae Possible lipoprotein 32/47 6e–39 Cole et al. (2001)

Orf5 408 Invertase Rhodococcus Putative DNA invertase 53/70 4e–50 Takai et al. (2000) Resolvase Xanthomonas Resolvase 49/64 5e–41 Liu et al. (1998) campestris Tn552 Resolvase Staphyloccus Resolvase 47/63 3e–39 Rowland and Dyke (1990) aureus

Orf7 207 Putative ParA Corynebacterium Putative partitioning protein 32/42 1e–12 AAG00274 glutamicum Partitioning protein Biﬁdobacterium Putative partitioning protein 29/43 7e–11 O´Riordan and Fitzgerald breve (1999) pCLP Par protein M. celatum Putative partitioning protein 25/39 0.13 Le Dantec et al. (2001)

The pVT2 Orf and predicted amino acid homologies are shown. Values for percentage identity, percentage similarity and E-values were taken from BLAST alignments between the two sequences (NCBI). An E (Expect) value of one represents the chance that in the available database you might expect to see one match with a similar score simply by chance.

© 2002 US Government, Molecular Microbiology, 43, 173–186 Rescue of cryptic mycobacterial plasmids with Tn552 179 fully below along with their genetic analysis. Briefly, the product of orf1, which we now will refer to as RepA, shares significant similarity to a set of proteins associated with mycobacterial plasmid replication (Table 1). For example, RepA is 67% identical to a replication- associated protein encoded by pLR7 (Beggs et al., 1995), 54% identical to an Orf in pMSC262 (Qin et al., 1994) and 29% identical to Orf1 of pMF1(Bachrach et al., 2000). We show below that RepA is essential for replication in pVT2. Orf2 shares significant similarity over its entire length to conjugative DNA relaxases and, most notably, to TraI Fig. 5. Defining the minimal replication region of pVT2. A relaxases encoded by plasmids F (30% identity, 43% schematic map of pVT2 is shown, with orfs numbered according to similarity), R100 (24%, 34%) and TrwC of R388 (22%, Fig. 4 and Table 1. Vertical arrows below the genetic map indicate 34%) (Bradshaw et al., 1990;; Yoshioka et al., 1990; Llosa Tn552 insertions that allow replication in BCG. The vertical arrows above the map indicate the sites of Tn552 insertion that prevent et al., 1994). Orf5 belongs to the resolvase/DNA inver- transformation of BCG. Plasmid clones that were used to define the tase (or serine recombinase) family of site-specific re- minimal region are shown below the genetic map of pVT2. Each combinases (Hatfull and Grindley, 1988; Leschziner et al., plasmid contained the regions of pVT2 indicated. They were created using unique restriction sites within Tn552 insertions at 1995). It shares up to 53% identity over its first 190 amino 9924, 10 858, 10 889 and 12 743. The end-point of the clones was acids to individual members of this family, and this either at a HindIII site (, at 2430) or a BamHI site (, at 3476) in includes all residues shown to be critical for catalysis. the plasmid. The ability of these clones to replicate (Rep) in BCG is indicated by a + or - on the right. However, it is significantly longer (by >200 aa) than typical family members. Orf7 shows significant similarity to putative plasmid partition proteins (32% identity to a ParA 9.5 and 4.4 kb. The insertions that prevented or allowed homologue in Corynebacterium jeikeium). Although initial plasmid transformation are shown in Fig. 5. These inser- BLAST searches detected only significant sequence simi- tion profiles define a region between co-ordinates 10 980 larity with putative partition proteins, subsequent reitera- and 2120 as essential for successful transformation into tive PSI-BLAST searches identified significant homologies BCG. Eight insertions within this region abolished trans- to characterized partition proteins encoded by two plas- formation. Three insertions (at 1297, 1673 and 1900) are mids (Altschul et al., 1997; Gallie and Kado, 1987). No located in repA, confirming its role in plasmid replication. homologues to Orf6 were identified in the database. In addition, insertions in orf6 and orf7 (at 11215 and However, orf6 is likely to be co-transcribed with orf7, and 11879) prevented transformation of BCG, providing we show below that it plays a role in plasmid establish- genetic evidence for a role in plasmid establishment ment or maintenance. or maintenance. PSI-blast analysis had suggested that Orf7 was similar to characterized partition proteins, but it detected no homologies to Orf6, although the organiza- Identification of the minimal replication origin of pVT2 tion of orf6 and orf7 suggests they would be transcribed We have taken advantage of the library of insertions as an operon. Two other insertions that prevented trans- to define regions of pVT2 that are essential for replication formation of BCG (at 556 and 12 743) did not appear in BCG. The initial pool of Kmr transposon insertions to be within orfs. One of these (at 12 743) was inserted was electroporated into BCG and Kmr transformants at the centre of a 12 bp inverted repeat, upstream of selected. Plasmid DNA from 16 individual transformants orf7. The second (at 556) was inserted upstream of a was rescued by electroduction into E. coli to allow plasmid DNA sequence that is highly conserved between mem- purification and analysis. The sites of transposon inser- bers of the pMSC262 family (Fig. 6), suggesting that tions in these plasmids were then determined by DNA these insertions have disrupted two important sites nec- sequence analysis as they define regions of the plasmid essary for origin function. One insertion within the repli- not required for replication in BCG. These insertions were cation region (at co-ordinate 48) did not prevent plasmid distributed throughout the plasmid between co-ordinates transformation. 4.4–10.0 kb, implying that sequences outside this segment A series of deletion derivatives was also constructed to were required for plasmid establishment and replication confirm the essential role of this region in replication, and in BCG. to demonstrate that no other segments of the plasmid To define the boundaries of critical sequences more were required. Again, these deletions took advantage accurately, we selected pVT2::Tn552kan∑ori derivatives of the transposon library by using unique restriction sites with insertions at regular intervals between co-ordinates in the transposon for cloning and deleting segments of

Fig. 6. DNA homology between pVT2 and other mycobacterial plasmids within the replication region. A region from 218 to 95 nts upstream of the start codon (at 921) of orf1 (repA) of pVT2 is shown. Background highlighted letters indicate identical nucleotides found among the plasmids. In the most highly conserved region, only conservation of 5 or 6 nucleotides is shown. In addition, the conserved nucleotides found only between the more closely related pVT2, pLR7 and pMSC262 are shaded. A putative 12 bp inverted repeat found in pVT2, pLR7 and pMSC262 is indicated.

pVT2. These derivatives, shown in Fig. 5, define the outer was over 14 years old and no plasmid DNA could be boundaries of the replication region at co-ordinates detected by agarose gel electrophoresis and ethidium- 10 889 and 2430 kb. pCK3 (10 889–3476) was also bromide staining. This suggests that the system is robust grown in culture without selection to monitor its stability. enough to rescue relatively rare circular species from a A total of 85% of the cells retained the plasmid after biological sample. Although we have used plasmids as a approximately 20 generations (21 days of growth), de- target in this assay, it could be equally well applied to monstrating that no other segment of pVT2 is necessary bacteriophage, which is probably the most environmen- for transformation and stable replication in BCG. tally abundant form of DNA. Propagation and analysis of these naturally occurring cryptic phages is almost impossible without prior knowledge of their bacterial host. Insertions in the putative pVT2 recombinase do not Thus the ability to rescue a whole-phage genome without interfere with plasmid replication or maintenance the requirement for phage propagation would be a very Orf5 shows strong conservation of catalytic residues powerful tool in examining this large biomass (Hendrix (including the serine nucleophile) and other motifs found et al., 1999). In addition, this rescue system need not be in the serine recombinase family of site-specific recombi- limited to bacterial DNA, as the rescue of eukaryotic nases, suggesting that this protein may be active (Hatfull viruses may have both important research and clinical and Grindley, 1988). One function of this recombinase applications. could be to resolve plasmid multimers into monomers and A second benefit of the system is that a library of inde- thus ensure correct plasmid segregation at cell division. pendent insertions is obtained in the target DNA. This To test this hypothesis, we examined the stability of two provides a wonderful resource from which both se- pVT2 derivatives in which Tn552 was inserted into orf5. quencing and genetic analyses can be carried out. We One of these insertions was within the N-terminal region, demonstrated this application by determining the entire which shares extensive amino acid identity with other sequence of pVT2 and by beginning a functional charac- recombinases, and a second insertion was within the non- terization of genes required for its replication. related C-terminal domain. Each construct was as stable as other pVT2 derivatives containing insertions in regions Defining replication functions of pVT2 not required for plasmid replication or maintenance (data not shown). In addition, as described above, pCK3 (Fig. 5) Our genetic analysis identified three orfs as essential for does not encode the recombinase and is stable in BCG. plasmid replication. Indirect evidence for their involve- ment in replication was also obtained by searching for protein homologues in the database. One of the orfs, Discussion repA, encoded a protein that was closely related to Here we have described the successful application of an several other proteins that are associated with mycobac- in vitro transposition system to rescue four cryptic plas- terial plasmid replication origins. These plasmids, pLR7, mids from M. avium. The three plasmids isolated from pJAZ38, pMSC262, pMF1 and pCLP, have been assigned MD22, types A, B, and C, have not been characterized to one family, the pMSC262 family (Pashley and Stoker, previously. We designate these plasmids pKMD302, 303, 2000), which is distinct from the most commonly used and 304 respectively. In principle, the Tn552 system could mycobacterial plasmid, pAL5000. Sequence analysis of a be applied to the rescue of any circular DNA. The ability minimal replication region for each of these plasmids has to rescue at least three plasmids from the MD22 DNA shown that they share both conserved non-coding DNA extract was particularly remarkable, as this DNA extract sequences and an orf (Qin et al., 1994; Beggs et al., 1995;

Gavigan et al., 1997; Bachrach et al., 2000; Picardeau orf6 and orf7 were identified by transposon mutagene- et al., 2000). A direct role for the conserved protein in sis and deletion analysis as being required for successful replication has not been proven but, given its conser- establishment and maintenance of the plasmid. Although vation and association with an origin region, it is most we have not ruled out that the insertions in orf7 affect probably a replication protein. The pVT2 RepA protein is plasmid establishment by having polar effects on expres- similar both in size and in sequence (up to 67% identity) sion of orf6, this seems unlikely, given the similarities to each of these proteins. In addition, we showed that detected between Orf7 and putative ParA proteins from a insertions within repA prevent transformation in BCG number of plasmids. No homologues were detected to (Fig. 5A) and M. smegmatis (data not shown), thus pro- orf6, but its location would be consistent with it being a viding genetic evidence RepA has a direct role in plasmid parB homologue, as they are often co-expressed with replication. parA genes and are often quite diverse (Gerdes et al., An extensive region of sequence similarity between 2000). Despite the apparent similarity with ParA, our data pLR7, pMSC262 and PVT2 includes both the repA gene suggest that Orf6 and Orf7 may also play a role in plasmid (up to 88% identity over 1300 nt, Fig. 4), and a region of establishment or replication, as mutation or deletion of DNA immediately upstream of the repA gene, which is their genes prevented transformation. Defects in parti- also found in other members of the pMSC262 family tioning generally result in decreased stability. In fact, a (Fig. 6). As suggested by others, the location of this region ParA homologue in pCLP, a linear mycobacterial plasmid and its high conservation make it an ideal candidate that has weak sequence similarity to Orf7, resulted in a for sequences playing a critical role in DNA replication 10-fold reduction in plasmid stability when deleted (Bachrach et al., 2000; Picardeau et al., 2000). In addi- (Picardeau et al., 2000; Le Dantec et al., 2001). A more tion, two Tn552 insertion mutants (at 12 743 and 566) detailed molecular analysis of this locus should elucidate demonstrated that sequences further upstream of the the role of these genes in replication and partitioning repA gene, in regions not conserved (or not sequenced) functions. between plasmids, were also required for transformation into BCG. One of these would disrupt a 10 bp inverted Host range of mycobacterial plasmids repeat sequence, which may be a binding site for a replication-associated protein. Unfortunately, the lack of One of the most important steps in developing a genetic complete plasmid sequences for all but pCLP (Le Dantec system in mycobacteria was the isolation of a trans- et al., 2001) prevents further comparison of the more formable mutant of M. smegmatis (Snapper et al., 1990). extensively defined pVT2 replication region. Although The basis of the Ept phenotype has not been defined, but we have compared the sequence of pCLP with the repli- it is generally assumed that this has no plasmid specificity. cation region of pVT2, we detect no significant extended Our ability to overcome the delayed growth phenotype of DNA homologies beyond those seen with the other plas- transformants of M. smegmatis suggests that we have mids (shown in Fig. 6). Perhaps this is not surprising, as identified another host gene (ept2) that specifically regu- pCLP is a linear plasmid with invertron terminal structures lates plasmid establishment in M. smegmatis. Alterna- at each end and thus will have evolved specialized func- tively, the original ept allele may have acquired a second tions to ensure replication of the plasmid ends (Picardeau mutation that allows a broader range of plasmids to trans- and Vincent, 1998). form M. smegmatis. Interestingly, the effect of the muta- The homologies detected in the replication region place tion was specific to the plasmids rescued from the M. pVT2 within the pMSC262 family. However, our compati- avium isolates and not to the pAL5000 plasmids tested. bility studies showed that pVT2 is compatible with at least It had been reported previously that derivatives of one member of this family, pMF1. In addition, earlier work pMSC262 and pLR7 did not transform M. smegmatis had also placed pVT2 and pLR7 into different plasmid (Goto et al., 1991; Beggs et al., 1995). Although we did groups based on cross hybridization studies and on their not specifically test these plasmids in our ept2 strain, the ability to coexist in M. avium isolates (Jucker and close relationship between pVT2 and these plasmids sug- Falkinham, 1990). Thus, although these plasmids are gests that they too would be capable of transformation of closely related, they are compatible. Presumably, pVT2 M. smegmatis. The identification of these ept genes initiates replication by a similar mechanism, but encodes should provide further insight into this phenotype and different functions that regulate plasmid compatibility. plasmid replication in mycobacteria. Examination of incompatibility between other members of the pMSC262 family is also hampered by the fact that pVT2 encodes a large serine recombinase most studies have involved working with subclones of the plasmids and thus determinants of incompatibility fea- orf5 encodes a site-specific recombinase. It shares tures may have been deleted. extensive identity through the first 190 aa to the

© 2002 US Government, Molecular Microbiology, 43, 173–186 182 C. Kirby et al. resolvase/invertase family of serine recombinases, but it ing TraI, are also present in the C-terminal half of the puta- is more than 200 aa longer. Most strikingly, the critical tive relaxase (Gorbalenya and Koonin, 1993; Hall and active site serine (residue 10 in the resolvase family) and Matson, 1999). Structural studies have shown that these other highly conserved motifs are also present in the motifs are clustered to form an ATP binding pocket and pVT2 recombinase (Hatfull and Grindley, 1988). These DNA binding site, which are essential for helicase activity include a predicted helix–turn–helix motif with the con- (Subramanya et al., 1996; Korolev et al., 1997). served DNA binding/recognition sequence (RSTVY at The obvious implication of this homology is that pVT2 residues 175–9), which is found in the same location as is a conjugative plasmid. We find this an attractive hypoth- in other serine recombinases. The conservation of these esis, as it is entirely consistent with the widespread oc- residues strongly suggests that this protein is active and currence of pVT2 in M. avium, M. intracellulare and M. performs a similar function in mycobacteria. The large C- scrofulaceum isolates, both clinical and environmental terminal extension of Orf5 places it within a subfamily of (Jucker and Falkinham, 1990). Conjugative transfer of large serine recombinases, which all encode an additional chromosomal DNA has been described previously extended C-terminal domain (Liu et al., 1998;Thorpe and between strains of M. smegmatis (Parsons et al., 1998), Smith, 1998; Lyras and Rood, 2000; Wang et al., 2000). but not in the slow-growing mycobacteria. The ability of The C-terminal domain of Orf5 does not appear to be slow-growing mycobacteria to exchange information by related to the C-terminal domain of these other serine conjugation would dramatically expand their genetic pool recombinases or to other proteins in the database (data and could potentially facilitate the spread of antibiotic not shown). This C-terminal region may play a role in reg- resistance amongst mycobacterial species. One caveat to ulating recombination activity, perhaps by interacting with this hypothesis is that no other conjugation-like proteins other cellular- or plasmid-encoded proteins, or by binding were detected in the pVT2 coding sequence. Presumably, to additional DNA sites to co-ordinate synaptic complex these proteins might be supplied either by the host or from formation. other coexisting plasmids. Perhaps the latter include Serine recombinases act on a pair of defined DNA sites those plasmids isolated from strain MD22. We are cur- to generate a variety of DNA rearrangements. Intermole- rently investigating these possibilities by characterizing cular recombination between these sites results in inte- the MD22 plasmids and by trying to establish plasmid gration, while intramolecular recombination can result in DNA transfer in BCG and M. avium. We note that an alter- excision (resolution) or inversion (Hatfull and Grindley, native role for the relaxase could be in plasmid replica- 1988). At this time, we do not know which of these activ- tion. However, this seems improbable, as insertions in this ities is mediated by the pVT2 recombinase, although it gene did not prevent replication in BCG, although we appears not to be involved in the resolution of plasmid cannot rule out the suppression of such insertion mutants multimers, as recombinase-defective derivatives of pVT2 by host-encoded helicases. were stable. pVT2 encodes a conjugative relaxase Experimental procedures Perhaps one of the most exciting revelations of the Bacterial strains and growth media sequence analysis was the identification of a large ORF Mycobacterium bovis BCG was the fifth passage BCG- (942 amino aids) with significant amino acid similarity to Pasteur strain originally obtained from the Pasteur Institute the DNA relaxases of conjugative plasmids. Homologies (K. McDonough, personal communication). BCG cultures were found throughout the length of the protein to a were grown in Middlebrook 7H9 broth supplemented with variety of putative conjugative helicases, but the most 10% v/v ADC (Difco) and 0.05% Tween 80. On solid media, notable were those to the characterized relaxases of plas- BCG was grown on Middlebrook 7H10 agar containing ADC and cyclohexamide (100 mgml–1). The M. smegmatis strain mids F, R100 and R388. These proteins are essential for used was mc2155. This is a highly transformable derivative plasmid transfer; they are responsible for introducing the of M. smegmatis strain mc26 (Snapper et al., 1990). Myco- specific nick at oriT and then they facilitate unwinding bacterium smegmatis was grown in trypticase soy broth of plasmid DNA to initiate transfer (Lanka and Wilkins, containing 0.05% Tween 80. For growth on solid media, TSA 1995; Firth et al., 1996). The N-terminal portion of the trypticase soy agar plates were used. pVT2 protein contains sequence motifs common to con- The E. coli strain used for high-efficiency electroporations jugative relaxases, which have been shown to be essen- was DH10B (Grant et al., 1990). XLI-blue (Bullock et al., 1987) was used as a recipient for electroducing plasmid DNA tial for oriT nicking activity in the plasmid RP4; (Pansegrau from mycobacteria. E. coli was grown in Luria–Bertani (LB) and Lanka, 1991; Pansegrau et al., 1994; Byrd and broth or, for growth on solid media, LB agar. SOC broth Matson, 1997). The seven conserved helicase motifs that (Sambrook et al., 1989) was used for the outgrowth of cells are common to the superfamily of DNA helicases, includ- after electroporation or electroduction.

Antibiotics were used at the following concentrations for and Collins, 1996; Parsons et al., 1998;). When determining E. coli: kanamycin sulphate, 50 mgml–1; tetracycline, 10 mg the ability of a plasmid to transform these species, electro- ml–1; ampicillin, 100 mgml–1; and hygromycin B, 100 mgml–1. porations were always performed multiple times, and in par- For selection in mycobacteria, antibiotics were used as allel with positive controls. Electroporation of in vitro follows: kanamycin sulphate, 10 mgml–1; and hygromycin B, transposition reactions and ligations were performed under 100 mgml–1. standard conditions into DH10B. Plasmid DNA was electroduced into E. coli XL1-blue from M. smegmatis or BCG as described previously (Baulard et al., 1992). Tetracycline was Plasmids included in the media to select for the recipient. pAL5000-based shuttle vectors used included pMD30, which r contains a pUC origin of replication and the IS903 Km gene PCR (Donnelly-Wu et al., 1993) and pLP961, a pACYC184 shuttle vector encoding hygromycin resistance (Parsons et al., PCR was used to conﬁrm the presence of the plasmid- 1998). pBP10 is a Kmr shuttle vector derivative of the encoded IS903 Kmr gene and the Hygr gene in mycobacter- Mycobacterium fortuitum plasmid pMF1 and belongs to the ial cells. PCR reactions were performed using a colony PCR pMSC262 family of mycobacterial plasmids (Bachrach et al., method (Trower, 1996). Mycobacterial colonies were boiled 2000). pKD207 is a Kmr pACYC184 derivative that contains in 20 ml of sterile distilled water for 10 min. Then, 1 ml of the the Kmr gene from pKD100 (Derbyshire et al., 1987), cloned boiled mycobacterial colony was used in the PCR reaction. into the unique XbaI site of pACYC184. The two primers used to detect the Kmr gene were OKD251 pCK1 and pCK2 were made using pVT2::Tn552 deriva- (5¢-ACGTCTTGCTCGAGGCCGCG-3¢) and OKD252 tives containing insertions at bp 9924 and 10 858 respectively (5¢-TCTGCGATTCCGACTCGTCC-3¢). The primers used (Figs 4 and 5). A unique HindIII site at the end of the trans- to amplify the Hygr gene were as follows: OKD369 poson was used to subclone a DNA fragment extending from (5¢-CGGAACGACCAGGAATTCTG-3¢) and OKD370 (5¢- the point of insertion to a HindIII site in pVT2 (2430). HindIII TCATCCGGCTCATCACCAGGTAG-3¢). fragments of 5.457 kb and 4 4523 kb from each derivative were ligated into the Kmr vector, pKD207 to generate pCK1 and pCK2. Two Tn552 insertions, at 10 889 and 12 743, were Southern analysis used to generate pCK3 and pCK4 (Figs 4 and 5). These two Probes were labelled with [ -32P]-dCTP (3000 Ci mmol–1) insertions are in the opposite orientation to those described a using a Random priming DNA labelling system according to above. Tn552kan∑ori contains a single BamHI site 71 bp from manufacturer’s instructions (Gibco BRL). Following restric- its right hand end and pVT2 contains a BamHI site at 3476. tion digestion, plasmid DNA was separated by agarose gel The two pVT2 derivatives were digested to completion with electrophoresis, and transferred onto a nitrocellulose mem- BamHI and the kan∑ori-containing fragment recircularized to brane using the capillary blotting method. generate pCK3 and pCK4. Therefore, these plasmids contain the kan∑ori segment of Tn552 and the adjacent segment of pVT2 up to co-ordinate 3476. Stability experiments

All of the stability experiments in BCG were performed using In vitro transposition freshly transformed BCG colonies grown on appropriate selective plates. Individual colonies were grown in Middle- Two DNA extracts were used for the in vitro transposition brook 7H9 broth without antibiotic. Cultures were grown for reaction, which were originally isolated from clinical, non- 5–7 days to a cell density of approximately 108–109 cells ml–1 AIDS related M. avium isolates (Jucker and Falkinham, (OD 2.5–3.5). At this time, cells were sonicated to disperse 1990). Strain MD1 carries a single 12.9 kb plasmid, pVT2, clumps, diluted (1:100) into fresh broth for further growth, and whereas MD22 carries multiple co-existing plasmids. Crude also plated on solid media. This was repeated at each of DNA extracts from MD1 and MD22 were used directly in an three serial transfers. Cells were plated on media without in vitro transposition reaction as described previously (Griffin antibiotic to determine the total number of viable cells, and et al., 1999). The transposon used in the reaction was Tn552 on media containing the appropriate antibiotic selection (Km, kan∑ori. This transposon contains the IS903 Kmr gene and Hyg or Km/Hyg) to determine the percentages of cells con- the E. coli replication origin from pBR322 (Griffin et al, 1999). taining one or both plasmids. In all stability and transforma- Five Tn552kan•ori insertion derivatives of pVT2 were used tion experiments, the presence of the plasmid in individual for the second generation libraries. In these libraries, a Tn552 colonies was confirmed using PCR to amplify the Kmr and derivative encoding ampicillin resistance, Tn552bla, was Hygr genes (see above), and electroduction followed by used to select for insertions. Tn552bla consists of the bla restriction analysis. gene (pBR322 co-ordinates 3289–4278) flanked by BsrGI A similar protocol was adapted to determine the stability of restriction sites and Tn552 ends. the M. avium plasmids in M. smegmatis (mc2155), but over a shorter time period (3 days), involving daily serial transfers. Electroporation and electroduction The presence of the plasmid in colonies grown on selective media was confirmed by PCR amplification of the Kmr gene. Electroporations of plasmid DNA into M. smegmatis and M. Group B plasmids were electroduced into E. coli to confirm bovis BCG were performed as described previously (Wards the restriction pattern of the plasmid. However, electroduc-

© 2002 US Government, Molecular Microbiology, 43, 173–186 184 C. Kirby et al. tion of plasmid types A, C and pVT2 was not always XL1-blue: a high efficiency plasmid transforming recA successful. In the latter cases, PCR was used to confirm the Escherichia coli strain with beta-galactosidase expression. presence of the plasmid. Biotechniques 5: 376–379. Byrd, D.R., and Matson, S.W. (1997) Nicking by transe- sterification: the reaction catalysed by a relaxase. Mol Sequencing of pVT2 Microbiol 25: 1011–1022. Cole, S.T., Brosch, R., Parkhill, J., Garnier, T., Churcher, Plasmid DNA was prepared for sequencing by alkaline lysis C., Harris, D., et al. (1998) Deciphering the biology of (Sambrook et al., 1989). Sequence was determined in both Mycobacterium tuberculosis from the complete genome directions from the transposon inserts using left and right sequence. Nature 393: 537–544. end-specific primers. For Tn552kan∑ori the primers were Cole, S.T., Eiglmeier, K., Parkhill, J., James, K.D., Thomson, oKD20 (5¢-GATATATTTTTATCTTGTGC-3¢) and oKD303 (5¢- N.R., Wheeler, P.R. et al. (2001) Massive gene decay in CTGACGCTCAGTGGAACG-3¢), and for Tn552bla they were the leprosy bacillus. Nature 409: 1007–1011. oKD304 (5¢-GCACATTTCCCCGAAAAGTG-3¢) and oKD305 Corpet, F. (1988) Multiple sequence alignment with hierar- (5¢-CACTGATTAAGCATTGGTAACTG-3¢). Sequencing was chical clustering. Nucleic Acids Res 16: 10881–10890. performed at the Wadsworth Center Molecular Genetics Core Crawford, J.T., and Falkinham, J.O. (1990) Plasmids of the facility. Sequence reads were aligned using SEQUENCHER Mycobacterium avium complex. In Molecular Biology of the (Gene Codes Corporation). Mycobacteria. McFadden, J. (ed.). San Diego: Academic Sequence analysis was performed using the GCG pack- Press, pp. 97–120. age (Genetics Computer Group), SEQUENCHER (Gene Codes Crawford, J.T., Cave, M.D., and Bates, J.H. (1981) Evidence Corporation), DNA STRIDER (CEA) and BLAST (Altschul et al., for plasmid-mediated restriction-modification in Mycobac- 1997). Sequence alignments were performed using the pro- terium avium intracellulare. J Gen Micro 127: 333–338. grams CLUSTAL W (Thompson et al., 1994) and MULTALIGN Le Dantec, C., Winter, N., Gicquel, B., Vincent, V., and (Corpet, 1988). Picardeau, M. (2001) Genomic sequence, transcrip- tional analysis of a, 23-kilobase mycobacterial linear plasmid: evidence for horizontal transfer and identification Acknowledgements of plasmid maintenance systems. J Bacteriol 183: We thank Drs Victoria Derbyshire, Linda Parsons, Norma 2151–2164. Tavakoli and Jun Wang for their insightful comments Deckert, G., Warren, P.V., Gaasterland, T., Young, W.G., on the manuscript. We gratefully acknowledge Dr M. Jucker Lenox, A.L., Graham, D.E., et al. (1998) The complete for providing the M. avium plasmid extracts, Dr K.G. genome of the hyperthermophilic bacterium Aquifex aeoli- Papavinasasundaram for supplying pBP10 and the cus. Nature 392: 353–358. Wadsworth Center Molecular Genetics Core facility for DNA Derbyshire, K.M., Hwang, L., and Grindley, N.D. (1987) sequencing and oligonucleotide synthesis. This work was Genetic analysis of the interaction of the insertion supported by the National Institutes of Health; grant AI42308 sequence IS903 transposase with its terminal inverted to K.M.D. and grant GM28470 to N.D.F.G. repeats. Proc Natl Acad Sci USA 84: 8049–8053. Donnelly-Wu, M.K., Jacobs, W.R., Jr, and Hatfull, G.F. (1993) Superinfection immunity of mycobacteriophage L5: appli- References cations for genetic transformation of mycobacteria. Mol Microbiol 7: 407–417. Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Erardi, F.X., Failla, M.L., Falkinham, J.O., III. (1987) Plasmid- Z., Miller, W., and Lipman, D.J. (1997) Gapped BLAST and encoded copper resistance and precipitation by Mycobac- PSI-BLAST: a new generation of protein database search terium scrofulaceum. Applied & Environ Microbiol 53: programs. Nucleic Acids Res 25: 3389–3402. 1951–1954. Bachrach, G., Colston, M.J., Bercovier, H., Bar-Nir, D., Falkinham, J.O., and Crawford, J.T. (1994) Plasmids. Anderson, C., and Papavinasasundaram, K.G. (2000) A In Tuberculosis. Bloom, B.R. (ed.). Washington, DC: new single-copy mycobacterial plasmid, pMF1, from American Society for Microbiology, pp. 185–198. Mycobacterium fortuitum which is compatible with the Farrand, S.K., Hwang, I., and Cook, D.M. (1996) The tra pAL5000 replicon. Microbiology 146: 297–303. region of the nopaline-type Ti plasmid is a chimera with Baulard, A., Jourdan, C., Mercenier, A., and Locht, C. (1992) elements related to the transfer systems of RSF1010, Rapid mycobacterial plasmid analysis by electroduction RP4 and F. J Bacteriol 178: 4233–4247. between Mycobacterium spp and Escherichia coli. Nucleic Firth, N., Ippen-Ihler, K., and Skurray, R.A. (1996) Structure Acids Res20: 4105. and function of the F Factor and mechanism of conjuga- Beggs, M.L., Crawford, J.T., and Eisenach, K.D. (1995) Iso- tion. In Escherichia Coli and Salmonella Cellular and Mol- lation and sequencing of the replication region of Mycobac- ecular Biology, Vol 2. Neidhardt, F.C. (ed.). Washington, terium avium plasmid pLR7. J Bacteriol 177: 4836–4840. DC: American Society for Microbiology, pp. 2377–2401. Bradshaw, H.D., Jr, Traxler, B.A., Minkley, E.G., Jr, Nester, Gallie, D.R., and Kado, C.I. (1987) Agrobacterium tumefa- E.W., and Gordon, M.P. (1990) Nucleotide sequence of the ciens pTAR parA promoter region involved in autoregula- traI (Helicase I) gene from the sex factor F. J Bacteriol 172: tion, incompatibility and plasmid partitioning. J Mol Biol 4127–4131. 193: 465–478. Bullock, W.W., Fernandez, J.M., and Short, J.M. (1987) Gavigan, J.A., Ainsa, J.A., Perez, E., Otal, I., and

Martin, C. (1997) Isolation by genetic labeling of a new the DNA cleavage and rejoining reactions. Mol Microbiol mycobacterial plasmid, pJAZ38, from Mycobacterium 15: 865–870. fortuitum. J Bacteriol 179: 4115–4122. Liu, C.C., Huhne, R., Tu, J., Lorbach, E., and Droge, P. (1998) Gerdes, K., Moller-Jensen, J., and R.J. (2000) Plasmid and The resolvase encoded by Xanthomonas campestris trans- chromosome partitioning: surprises from phylogeny. Mol posable element ISXc5 constitutes a new subfamily closely Microbiol 37: 455–466. related to DNA invertases. Genes Cells 3: 221–233. Gorbalenya, A.E., and Koonin, E.V. (1993) Helicases: amino Llosa, M., Bolland, S., and de la Cruz, F. (1994) Genetic acid sequence comparisons and structure-function rela- organization of the conjugal DNA processing region of tionships. Current Opinion Structural Biol 3: 419–429. the Inc W plasmid R388. J Mol Biol 235: 448–464. Goto, Y., Taniguchi, H., Udou, T., Mizuguchi, Y., and Lyras, D., and Rood, J.I. (2000) Transposition of Tn4451 Tokunaga, T. (1991) Development of a new host vector and Tn4453 involves a circular intermediate that forms a system in mycobacteria. FEMS Microbiol Letters 67: 277– promoter for the large resolvase, TnpX. Mol Microbiol 38: 282. 588–601. Grant, S.G.N., Jessee, J., Bloom, F.R., and Hanahan, D. Meissner, P.S., and Falkinham, J.O., III. (1984) Plasmid- (1990) Differential plasmid rescue from transgenic mouse encoded mercuric reductase in Mycobacterium scrofu- DNAs Into Escherichia coli methylation-restriction mutants. laceum. J Bacteriol 157: 669–672. Proc Natl Acad Sci USA 87: 4645–4649. O’Riordan, K., and Fitzgerald, G.F. (1999) Molecular charac- Griffin, T.J., Parsons, L.M., Leschziner, A.E., DeVost, J., terisation of a 5.75-kb cryptic plasmid from Bifidobacterium Derbyshire, K.M., and Grindley, N.D.F. (1999) In vitro trans- breve NCFB 2258 and determination of mode of replica- position of Tn552: a tool for DNA sequencing and mutation. FEMS Microbiol Lett 174: 285–294. gensis. Nucleic Acids Res 27: 3859–3865. Pansegrau, W., and Lanka, E. (1991) Common sequence Guilhot, C., Jackson, M., and Gicquel, B. (1999) Mobile genetic motifs in DNA relaxases and nick regions from a variety of elements and plasmids in mycobacteria. In Mycobacteria, DNA transfer systems. Nucleic Acids Res 19: 3455. Molecular Biology and Virulence. Ratledge, C., and Dale, J. Pansegrau, W., Schroder, W., and Lanka, E. (1994) (eds). Oxford: Blackwell Science, pp. 17–37. Concerted action of three distinct domains in the DNA Hall, M.C., and Matson, S.W. (1999) Helicase motifs: the cleaving-joining reaction catalyzed by relaxase (TraI) of engine that powers DNA unwinding. Mol Microbiol 34: conjugative plasmid RP4. J Biol Chem 269: 2782– 867–877. 2789. Hatfull, G.F., and Grindley, N.D.F. (1988) The resolvases Parsons, L.M., Jankowski, C.S., and Derbyshire, K.M. (1998) and DNA invertases: a family of enzymes active in site Conjugal transfer of chromosomal DNA in Mycobacteria specific recombination. In Genetic Recombination. Kucher- smegmatis. Mol Microbiol 28: 571–582. lapati, R., and Smith, G. (eds). Washington, DC: American Pashley, C., and Stoker, N.G. (2000) Plasmids in mycobac- Society for Microbiology, pp 357–396. teria. In Molecular Genetics of Mycobacteria. Hatfull, G.F., Hendrix, R.W., Smith, M.C., Burns, R.N., Ford, M.E., and and Jacobs, W.R., Jr. (eds). Washington, DC: American Hatfull, G.F. (1999) Evolutionary relationships among Society for Microbiology, pp 551–567. diverse bacteriophages and prophages: all the world’s Picardeau, M., and Vincent, V. (1998) Mycobacterial linear a phage. Proc Natl Acad Sci USA 96: 2192–2197. plasmids have an invertron-like structure related to other Inderlied, C.B., Kemper C.A., and Bermudez, L.E.M. (1993) linear replicons in actinomycetes. Microbiology 144: The Mycobacterium avium complex. Clin Microbiol 1981–1988. Reviews 6: 266–310. Picardeau, M., Le Dante, C., and Vincent, V. (2000) Analysis Jucker, M.T. (1991) Relationship of plasmids in Mycobac- of the internal replication region of a mycobacterial linear terium avium, Mycobacterium intracellulare and Mycobac- plasmid. Microbiology 146: 305–313. terium scrofulaceum. Blacksburg, VA: Virginia Polytechnic Qin, M., Taniguchi, H., and Mizuguchi, Y. (1994) Analysis Institute and State University. of the replication region of a mycobacterial plasmid, Jucker, M.T., and Falkinham, J.O. (1990) Epidemiology of pMSC262. J Bacteriol 176: 419–425. infection by nontuberculous mycobacteria. Am Rev Respir Redenbach, M., Kieser, H.M., Denapaite, D., Eichner, A., Dis 142: 858–862. Cullum, J., Kinashi, H., and Hopwood, D.A. (1996) A set of Korolev, S., Hsieh, J., Gauss, G.H., Lohman, T.M., and ordered cosmids and a detailed genetic and physical map Waksman, G. (1997) Major domain swiveling revealed by for the 8 Mb Streptomyces coelicolor A3 (2) chromosome. the crystal structures of complexes of E. coli Rep helicase Mol Microbiol 21: 77–96. bound to single-stranded DNA and ADP. Cell 90: 635– Rowland, S.J., and Dyke, K.G. (1990) Tn552, a novel 647. transposable element from Staphylococcus aureus. Mol Labidi, A., David, H.L., and Roulland-Dussoix, D. (1985) Microbiol 4: 961–975. Restriction endonuclease mapping and cloning of Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecu- Mycobacterium fortuitium var fortuitum plasmid pAL5000. lar Cloning: a Laboratory Manual. Cold Spring Harbor, NY: Ann Institute Pasteur 136: 209–215. Cold Spring Harbor Laboratory Press. Lanka, E., and Wilkins, B.M. (1995) DNA processing reac- Snapper, S.B., Melton, R.E., Mustafa, S., Kieser, T., and tions in bacterial conjugation. Ann Rev Biochem 64: Jacobs, W.R., Jr. (1990) Isolation and characterization of 141–169. efficient plasmid transformation mutants of Mycobacterium Leschziner, A.E., Boocock, M.R., and Grindley, N.D.F. (1995) smegmatis. Mol Microbiol 4: 1911–1919. The tyrosine-6 hydroxyl of gd resolvase is not required for Subramanya, H.S., Bird, L.E., Brannigan, J.A., and Wigley,

D.B. (1996) Crystal structure of a DExx box DNA helicase. Wang, H., Roberts, A.P., Lyras, D., Rood, J.I., Wilks, M., and Nature 38: 379–383. Mullany, P. (2000) Characterization of the ends and target Takai, S., Hines, S.A., Sekizaki, T., Nicholson, V.M., sites of the novel conjugative transposon Tn5397 from Alperin, D.A., Osaki, M., et al. (2000) DNA sequence and Clostridium difficile: excision and circularization is medi- comparison of virulence plasmids from Rhodococcus equi ated by the large resolvase, TndX. J Bacteriol 182: ATCC 33701 and 103. Infect Immun 68: 6840–6847. 3775–3783. Thompson, J.D., Higgins, D.G., and Gibson, T.J. (1994) Wards, B.J., and Collins, D.M. (1996) Electroporation at CLUSTAL W: improving the sensitivity of progressive elevated temperatures substantially improves transfor- multiple sequence alignment through sequence weighting, mation efficiency of slow-growing mycobacteria. FEMS position-specific gap penalties and weight matrix choice. Microbiol Letters 145: 101–105. Nucleic Acids Res 22: 4673–4680. Wehrmann, A., Eggeling, L., and Sahm, H. (1994) Analysis Thorpe, H.M., and Smith, M.C. (1998) In vitro site-specific of different DNA fragments of Corynebacterium glutam- integration of bacteriophage DNA catalyzed by a recombi- icum complementing dapE of Escherichia coli. Microbiol- nase of the resolvase/invertase family. Proc Natl Acad Sci ogy 140: 3349–3356. USA 95: 5505–5510. Yoshioka, Y., Fujita, Y., and Ohtsube, E. (1990) Nucleotide Trower, M.K. (1996) A rapid PCR-based colony screening sequence of the promoter distal region of the tra operon protocol for cloned inserts. Methods Mol Biol 58: 329– of plasmid R100, including traI (DNA helicase I) and 333. traD genes. J Mol Biol 214: 39–53.