Characteristics of the Conjugative Transfer System of the IncM Plasmid pCTX-M3 and Identification of Its Putative Regulators

Horizontal gene transfer is responsible for rapid changes in bacterial genomes, and the conjugative transfer of plasmids has a great impact on the plasticity of bacteria. Here, we present a deletion analysis of the conjugative transfer system genes of the pCTX-M3 plasmid of the IncM group, which is responsible for the dissemination of antibiotic resistance genes in Enterobacteriaceae. We found that the deletion of either of the orf35 and orf36 genes, which are dispensable for conjugative transfer, increased the plasmid mobilization efficiency. Real-time quantitative PCR (RT-qPCR) analysis suggested the involvement of orf35 and orf36 in regulating the expression of transfer genes. We also revised the host range of pCTX-M3 by showing that its conjugative transfer system has a much broader host range than its replicon.

IMPORTANCE Horizontal gene transfer is responsible for rapid changes in bacterial genomes, and the conjugative transfer of plasmids has a great impact on the plasticity of bacteria. Here, we present a deletion analysis of the conjugative transfer system genes of the pCTX-M3 plasmid of the IncM group, which is responsible for the dissemination of antibiotic resistance genes in Enterobacteriaceae. We found that the deletion of either of the orf35 and orf36 genes, which are dispensable for conjugative transfer, increased the plasmid mobilization efficiency. Real-time quantitative PCR (RT-qPCR) analysis suggested the involvement of orf35 and orf36 in regulating the expression of transfer genes. We also revised the host range of pCTX-M3 by showing that its conjugative transfer system has a much broader host range than its replicon.
KEYWORDS IncM group, conjugative transfer, plasmid analysis, plasmid mobilization C onjugative transfer is a prevalent phenomenon among bacteria; this phenomenon is crucial for horizontal gene transfer in the biosphere and is a major contributor to the rapid variability of bacterial genomes. In the process of conjugative transfer, DNA (a conjugative plasmid, a conjugative transposon, or an integrative conjugative element [ICE]) is transferred from a donor to a recipient cell after physical contact between the cells is established. The process may be regarded as a combination of DNA processing functions coupled to a type IV secretion system (T4SS; also known as mating pair formation [Mpf]) by a dedicated protein (coupling protein [CP]). The DNA process-ing functions are provided by the DNA transfer and replication (Dtr) system, also called the relaxosome complex (1). The T4SSs of Gram-negative bacteria are classified into two large phylogenetic groups, namely, IVA and IVB: the Agrobacterium tumefaciens VirB/D4 secretion system and the conjugation systems of the IncF and IncP plasmids are classified as type IVA (T4ASS) (2,3), whereas type IVB (T4BSS) is represented by secretion systems found in Legionella pneumophila (Dot/Icm) and in other important pathogens (4,5). The majority of the Dot/Icm proteins share homology with the constituents of the conjugation system of the R64 plasmid of the IncI1 group. Despite an increasing amount of information becoming available in recent years on the organization and regulation of T4BSSs, they are still less thoroughly characterized than T4ASSs.
The canonical T4SS is represented by the Agrobacterium tumefaciens VirB/D4 secretion system responsible for transfer DNA (T-DNA) transfer to plant cells during infection. This T4SS consists of 11 proteins (VirB1 to VirB11) and the coupling protein VirD4 (for review, see references 5 to 8). The translocation channel comprises the VirB3, B6, B7, B8, B9, and B10 proteins. In the translocation channel, three components form the core channel complex in the outer membrane (OM), also called outer membrane complex (OMC): VirB9, the pore-forming protein; VirB7, a small lipoprotein; and VirB10, the protein spanning both the inner membrane (IM) and the outer membrane (OM). Interactions of the OMC with the inner membrane complex (IMC), which comprises the VirB3, VirB6, and VirB8 proteins, and with the ATPases VirB4 and VirB11 result in the formation of a pore. The extracellular structure important for the establishment of contact between mating cells, namely, the T-pilus, is composed of the major subunit VirB2 and the minor component VirB5 localized at the tip. VirB3, the least-characterized Mpf component, is also necessary for T-pilus assembly. Finally, VirB1 shows homology to a lytic transglycosylase that cleaves peptidoglycan (3,9). The system is energized by three cytoplasmic ATPases: VirB4, VirB11, and the coupling protein VirD4. Of all of the Vir proteins listed above, VirB4 is the only component present in every T4SS described so far (10). The universal presence of VirB4 enabled all known Mpfs of both Gramnegative and Gram-positive bacteria and archaea to be divided into eight groups on the basis of VirB4 phylogeny (11). The A. tumefaciens VirB/D4 system and the conjugation system of the IncP plasmids are now classified in the MPF T group (11). The conjugation system of the F plasmid, one of the most well-known plasmids, belongs to the MPF F group (11). In addition, the IncI1 plasmid R64 codes for TraU, which is a distant VirB4 homologue, and constitutes the prototype of the MPF I group. The R64 conjugative transfer system is encoded by 22 transfer genes, namely, traE-traY, three trbA-C genes, and the nuc gene, 16 of which have been shown to be indispensable for conjugation (12). The homology of R64 to the VirB/D4 system of A. tumefaciens is rather low. However, TraO displays homology to VirB10 (13), and TraM is distantly homologous to VirB8, TraJ to the VirB11 ATPase, and TraQ and TraR to the pilin subunit VirB2 (11).
Another plasmid encoding an MPF I conjugation system that displays homology to the T4BSS systems of the IncI1 plasmids is pCTX-M3 (accession no. AF550415), a member of the IncM incompatibility group (14). Plasmid pCTX-M3 was isolated from a clinical Citrobacter freundii strain in Poland in 1996 as a vector of the bla CTX-M-3 gene (14,15). It is noteworthy that members of the IncM group are closely related not only to each other but also to plasmids of the IncL group (16), with which they were earlier classified jointly as the IncL/M group (17). IncL and IncM plasmids are widespread in Enterobacteriaceae (18) and are responsible for the dissemination of antibiotic resistance genes. These genes include bla CTX-M-3 , which encodes an extended spectrum ␤-lactamase, bla NDM-1 , which encodes a metallo-␤-lactamase, bla OXA-48 , which encodes a carbapenem-hydrolyzing enzyme (19,20), bla  , which codes for the Klebsiella pneumoniae carbapenemase (21), bla IMP-4 , which codes for imipenemase (22), and the aminoglycoside resistance gene armA (23).
The pCTX-M3 plasmid can be transferred by approximately 10% of cells in an Escherichia coli donor population under optimal conditions (14). Bacteria bearing pCTX-M3 can also conjugate in liquid culture; however, in contrast to IncI1 plasmids, which require type IV pili for conjugation in liquid media, pCTX-M3 does not encode additional pili (14,24,25). The conjugative transfer genes of pCTX-M3 are localized in two separate regions with predicted operon structures, namely, tra and trb (14), and these genes do not exhibit substantial sequence similarity to genes with ascribed functions available in public databases apart from those encoded by IncI1 and IncL/ IncM plasmids (Fig. 1). The tra and trb genes of IncL/IncM plasmids such as pCTX-M3 encode proteins that exhibit 39% to 65% similarity to those encoded by IncI1 plasmids (such as R64 and ColIb-P9) ( Table 1); however, a number of genes from each system do not have counterparts in the other system (14). The pCTX-M3 plasmid can mobilize plasmids that contain the heterologous oriT ColIb-P9 (from the IncI1 plasmid ColIb-P9), and plasmids with oriT pCTX-M3 can be mobilized by a ColIb-P9-derived plasmid (14); both of these mobilizations occur at low frequencies.
Here, we present a deletion analysis of the tra and trb genes potentially involved in pCTX-M3 conjugative transfer. We found that the deletion of either of the orf35 and orf36 genes, both of which are dispensable for pCTX-M3 transfer, increases the mobilization efficiency of oriT pCTX-M3 -bearing plasmids into E. coli and A. tumefaciens. The deletion of these genes also affected the transcription of other conjugative transfer genes. In addition, we verified the host range of the pCTX-M3 conjugation system and found that the host range of its replicon, reported previously to comprise Alpha-, Beta-, and Gammaproteobacteria (26), is in fact much narrower than previously believed and is limited to Enterobacteriaceae.

RESULTS AND DISCUSSION
Organization of the regions coding for the conjugative transfer systems of the pCTX-M3 and R64 plasmids. In pCTX-M3, both the tra and trb regions coding for the conjugative transfer system display extensive conservation of synteny with the conjugation system genes of the IncI1 plasmids R64 and ColIb-P9 ( Fig. 1) (14). However, there are certain differences. pCTX-M3 has no homologues of the traEFG, traST, and traV genes. Neither traEFG nor traS is required for the conjugative transfer of R64, whereas traT and traV are indispensable (12). Moreover, in pCTX-M3, the nuc gene, which encodes a nuclease, is located at a distance from the tra region. In addition, the single orf38 located between traR and traU replaces traST, while orf36, which is found only in IncL and IncM plasmids, separates traL and traM. Furthermore, the trbN gene of pCTX-M3, which encodes a putative lytic transglycosylase, has no homologues in R64. The homologue of orf46 is also absent in R64. However, the major difference between these plasmids concerns the position of their oriT regions with the nikAB genes. In R64, oriT lies far from the tra genes in a tail-to-tail orientation with respect to the trb operon. In pCTX-M3, the oriT region, along with the nikAB genes, is situated immediately upstream of the tra genes, and the nikAB and tra genes are predicted to constitute a single operon ( Fig. 1) (14).
Identification of genes necessary for conjugation. A systematic deletion analysis was performed for the pCTX-M3 genes in the tra and trb regions. For this purpose, a collection of 27 derivatives with a deletion in each of the genes in the tra and trb regions, as well as orf35 from the pCTX-M3 leading region and orf46, which is located downstream of trbC, was constructed (see Table S1 in the supplemental material) by replacing a given gene with the cat gene (27). To avoid a difference in expression levels depending on the position of the gene in the operon, the cat gene with its own promoter sequence was inserted in the opposite orientation to that of the tra or trb genes. For each pCTX-M3 deletion derivative, the frequency of conjugative transfer (conjugation efficiency) from E. coli BW25113 cells to the recipient E. coli JE2571Rif r cells, both in liquid and on filters, was determined ( Fig. 2A). Liquid mating occurred at a lower efficiency than filter mating; therefore, further conjugative transfer analysis of the pCTX-M3 derivatives was performed using filter mating only. All the plasmids displayed a diminished or completely inhibited conjugative transfer ability except for pCTX-M3orf35::cat, pCTX-M3orf36::cat, and pCTX-M3orf46::cat; these plasmids showed the same conjugation efficiency as the parental pCTX-M3 plasmid, which was approximately 10 Ϫ1 for filter mating (Fig. 2B). We therefore concluded that orf35, orf36, and orf46 are dispensable for conjugative transfer. Complementation of the deleted tra and trb genes. To exclude the possibility that the reduced conjugative transfer of the pCTX-M3 deletion derivatives was caused by a polar effect, each mutated plasmid was complemented with an appropriate gene cloned into pMT5 or pAL3 under the control of the P lac promoter (Table S1). For the majority of the pCTX-M3 derivatives, the complementation fully or at least partially restored the conjugation efficiency (10 Ϫ4 to 10 Ϫ1 per donor cell). However, for the nikB, traM, traN, traU, traW, and traY deletion derivatives, the presence of the complementing gene rescued the conjugation efficiency only to levels less than 10 Ϫ5 ; for traP and pri, the complementation had no detectable effect (Fig. 2B). Although several independent traP and pri deletion mutants were investigated, none regained the conjugative transfer ability upon complementation with pMT5traP or pALpri, respectively.
In R64, the disruption of nikB, which encodes nickase (relaxase), abolished conjugative transfer; however, nikB can be complemented well (28). Furthermore, the disruption of traM, traN, traW, and traY was detrimental to the transfer of R64, but these genes were able to be complemented by plasmids expressing the relevant gene to the wild-type (WT) transfer efficiency or, for traY, 10ϫ lower than the WT transfer efficiency (12). In R64, disruptions of traU are detrimental to transfer. Similar to the effect of limited complementation of pCTX-M3traU::cat, the transfer efficiency of R64 that expressed a traU gene disrupted close to its start site but that was complemented by a plasmid bearing traU was 9 ϫ 10 Ϫ7 (12). Taking into account the differences in complementation between the respective mutants of pCTX-M3 and R64, the nikB, traM, traN, traW, traY, pri, traP, and traU mutants of pCTX-M3 were analyzed further. First, the expression of the complementing genes was verified by real-time quantitative PCR (RT-qPCR). The transcript levels of all these genes were at least 10-fold higher than those found in the strain bearing pCTX-M3 (Fig. 3A).
Given that a polar effect of the deletion of those genes was not strictly excluded, the transcription of genes located directly downstream of the target genes was quantified by RT-qPCR (Fig. 3B). Indeed, the transcript levels of each of the downstream genes analyzed, except for those of the traN::cat and traY::cat derivatives, were lower than those found in the strain bearing the intact pCTX-M3 (Fig. 3B). The complementation of the analyzed deletions with the appropriate genes restored the expression of the downstream genes to the levels observed in the strain with pCTX-M3 for the pri, traM, and traP deletion mutants and nearly restored expression for the traU deletion mutant, but complementation had only a small effect on the nikB::cat and traW::cat mutants (Fig. 3B). In the case of the traY::cat mutant, where no decrease in the expression of the downstream gene was found, the complementation had no effect on the transcript level of the downstream gene.
The results so far showed that five deletion mutants of pCTX-M3 behaved in an unexpected manner: traP and pri, which could not be complemented; traY, where the complementation was very poor; and nikB and traW, where the downstream genes, traH and traX, respectively, showed low expression in addition to poor complementation. To determine whether the behavior of these genes was due to the deletion of the genes in question or was connected with the presence of the cat gene in their loci, the cat gene was eliminated from the mutants with the use of the pCP20-encoded FLP recombinase. The ability of the constructed plasmids (pCTX-M3ΔtraP, pCTX-M3Δpri, pCTX-M3ΔtraW, pCTX-M3ΔtraY, and pCTX-M3ΔnikB) to undergo conjugative transfer was tested after complementation with a plasmid bearing the corresponding gene (Fig.  4). All the pCTX-M3 derivatives lacking both the gene of interest and the cat gene were transferred by conjugation when the missing gene was delivered in trans: the transfer efficiencies of the ΔnikB, ΔtraP, ΔtraW, and ΔtraY mutated plasmids were greater than 10 Ϫ3 transconjugants per donor cell (Fig. 4). Therefore, we conclude that the lack of complementation of the previously described plasmids with mutated nikB, traP, traW, and traY was connected to the presence of the cat gene in the loci of the affected genes and was corrected by cat elimination.
However, the transfer efficiency of pCTX-M3Δpri after complementation was low (approximately 2 ϫ 10 Ϫ7 ). The results obtained for the pri deletion in pCTX-M3 differ from those for the sog deletion in R64, where the deletion resulted in only a small drop in the transfer efficiency, from 10 Ϫ2 to 10 Ϫ3 (29). The sog gene encodes two proteins, the SogL primase (1,255 amino acids [aa]), which generates RNA primers for plasmid replication, and the shorter SogS (844 aa), which is a product of translational reinitiation within the sog reading frame (30). It was speculated that Sog proteins create a complex with DNA, coating the transferred single DNA strand to protect and stabilize it (31). Both proteins are transported from the donor to the recipient cell during conjugation, and the transport has been shown to rely on the pil genes encoding thin pili (32), which are absent in pCTX-M3 (14). The pri gene of pCTX-M3 also has the potential to encode two proteins, a primase (1,070 aa) and a putative 689-aa protein comprising the C-terminal moiety of the primase. The high level of the pri transcript (Fig. 3A) might result in the production of a large amount of these two proteins, which can coat the single-stranded DNA (ssDNA) of the transferred plasmid and block the transporter.
However, the reason for the lack of complementation of the pri-deficient pCTX-M3 is highly speculative, and the pCTX-M3 pri gene therefore needs further study.
The results presented above demonstrate that though the disruptions in the majority of the tra and trb genes of the pCTX-M3 plasmid, which were performed via cat insertion, do not prevent the expression of a functional conjugative transfer system, in some cases, the deletion of the inserted cat gene was required. In these cases, the observed decreased transcript levels and lower conjugation efficiencies of the mutated plasmids even in the presence of the appropriate complementing genes suggest a defective regulation of gene expression. The mechanism of the regulation of the pCTX-M3 tra and trb operons is unknown and needs further research.
Putative roles of the tra and trb genes of pCTX-M3. (i) Putative components of the T4CP subcomplex. The analysis of the trb region of pCTX-M3 showed that the deletion of either trbA, trbB, or trbC abolished conjugation. In R64, the trbA and trbC genes were found to be indispensable for conjugative transfer, while trbB was required for a high transfer efficiency (33). TrbC pCTX-M3 , with its ATPase Walker motifs A and B, is homologous to TrbC R64 and to DotL (IcmO) of L. pneumophila, which acts as a type IV coupling protein (T4CP) ( Table 1). In the Dot/Icm system, DotL forms the T4CP subcomplex along with other proteins that have no CP function: two inner membrane proteins, namely, DotM (IcmP) and DotN (IcmJ), and the secretion adapter proteins IcmS and IcmW (34). The DotM homologue of pCTX-M3 is encoded by the trbA gene. Homologues of TrbB pCTX-M3 , except for the TrbB proteins encoded by IncI1 plasmids, are putative disulfide bond isomerases. Therefore, the trbC gene of pCTX-M3 is likely to encode the coupling protein, while trbA codes for an element of the T4CP subcomplex, whose other components are to be characterized.
(ii) Putative components of the transmembrane subcomplex. The traN gene of both R64 (12) and pCTX-M3 encodes a homologue of DotH (IcmK) of L. pneumophila. The proper localization of DotH in the OM is assisted by the lipoproteins DotC and DotD, which together form a pore similar to that formed by the VirB7/VirB9 proteins of A. tumefaciens (13). DotC, DotD, and DotH, along with the IM proteins DotF (IcmG) and DotG (IcmE), have been found to form the core transmembrane subcomplex that bridges the IM and the OM in L. pneumophila (13,35). A homologue of L. pneumophila DotC, encoded by traI, is indispensable for the conjugative transfer of pCTX-M3, while the disruption of traI R64 led only to a reduction in the transfer efficiency (12). In pCTX-M3 and R64, TraH is a homologue of DotD of L. pneumophila; however, in contrast to the deletion of traH R64 (12), the disruption of traH pCTX-M3 abolishes the transfer of pCTX-M3. The putative localization of TraH pCTX-M3 in the cell membrane is supported by the presence of a predicted signal peptide and a lipid attachment motif in its sequence (Table 1). Distant homologues of DotF of L. pneumophila are TraP pCTX-M3 and TraP R64 , which are necessary for the conjugative transfer of both plasmids. In the TraP pCTX-M3 sequence, a single transmembrane helix was found, suggesting IM localization ( Table  1). DotG of L. pneumophila shares homology with TraO R64 (13) and TraO pCTX-M3 (Table  1) homologues of VirB10 of A. tumefaciens. Interestingly, the deletion of the traO gene abolished the conjugative transfer of pCTX-M3, while in R64, traO deletion only reduced the transfer efficiency (12). Therefore, we propose that TraH, TraI, TraN, TraO, and TraP of both pCTX-M3 and R64 are components of the core transmembrane subcomplex. The different consequences of the deletion of traH, traI, or traO on the conjugative transfer of pCTX-M3 and R64 raise the possibility that the compositions of the core transmembrane subcomplexes encoded by these two plasmids are also different.
(iii) Putative functions of other pCTX-M3-encoded proteins. The nikA and nikB genes encode components of the nickase complex (an auxiliary protein and a relaxase, respectively). The deletion of these genes completely abolishes the conjugative transfer of pCTX-M3 and R64 (36). The traJ gene encodes an ATPase homologous to VirB11 of A. tumefaciens and DotB of L. pneumophila (37) ( Table 1). Its deletion abolishes the conjugative transfer of pCTX-M3, while in R64, traJ deletion only reduced the transfer efficiency (12). In turn, TraK R64 and TraK pCTX-M3 ( Table 1) are homologues of the IcmT protein of L. pneumophila, whose function is unknown (37). TraM R64 and TraM pCTX-M3 display homology with DotI (IcmL) of L. pneumophila and with VirB8 of A. tumefaciens (11,38). The traU gene encodes a putative ATPase that is a homologue of DotO (IcmB) of L. pneumophila (37,39). The encoded protein is also distantly homologous to VirB4 (11), which is involved in pilus assembly (40) and is essential for the virulence of Agrobacterium (41).
The traR and traQ genes, which are indispensable for the conjugative transfer of pCTX-M3 and R64 (12), code for proteins distantly homologous to each other (Table 1) and that belong to the VirB2 family, which forms the major T-pilus subunit in the A. tumefaciens VirB/D4 system (11). The product of traY is a distant homologue of DotA of L. pneumophila (37) and, together with ExcA, builds the entry exclusion system of R64 (42) and of pCTX-M3 (17). The putative functions of the other proteins encoded in the tra and trb regions of pCTX-M3 remain unknown (Table 1).
To further analyze the effects of orf35 and orf36 on plasmid mobilization, the deletions of these genes were complemented with appropriate plasmids, namely, pALorf35 and pALorf36, respectively. The pABBoriT plasmid was mobilized to an E. coli recipient from E. coli DH5␣ donors. The helper plasmids pCS, pC35S, and pC36S, which were derived from pCTX-M3, pCTX-M3orf35::cat, and pCTX-M3orf36::cat, respectively, and lack kanamycin resistance, were generated for use with the Km r pABBoriT plasmid ( Table 2).
In the presence of pC35S and the complementing pALorf35 plasmid, the mobilization efficiency of pABB20oriT was slightly reduced relative to that observed in the presence of pC35S and the pAL3 vector (Fig. 6A), but this reduction occurred only when freshly obtained transformants were used as donors and the experiment was performed at 28°C. It is worth noting that the growth of E. coli bearing both pCTX-M3 and pALorf35 was disturbed, while the presence of pALorf35 alone did not affect cell growth. This effect can result from the possible deregulation of the tra genes controlled by the product of the orf35 gene, especially when expressed from the two coresident plasmids.
The complementation of the orf36 mutation in pC36S by pALorf36 decreased the pABB20oriT mobilization efficiency even below the level obtained with pCS as the helper plasmid (Fig. 6B). Interestingly, in the strain bearing both pCS and pALorf36, the mobilization efficiency of pABB20oriT was reduced.
To address the question of the role of orf35 and orf36 in conjugative transfer, the transcript levels of the nikA, nikB, and traH genes, the first three genes of the tra operon, were determined in E. coli strains bearing either pCTX-M3orf35::cat or pCTX-M3orf36::cat and were compared with those in the control strain bearing intact pCTX-M3 ( Fig. 6C and D). In the strain bearing pCTX-M3orf35::cat, the transcript levels of all three genes were elevated, approximately 40-, 23-, and 80-fold for nikA, nikB, and traH, respectively, relative to those in the control strain. In the strain bearing pCTX-M3orf36::cat, the levels of the nikA and nikB transcripts were unchanged, while the traH transcript was approximately 120-fold more abundant than in the control strain. We propose that the pCTX-M3 tra operon, which encodes both the nickase complex and the T4SS, is subject to orf35-dependent repression. The effect of derepression of the tra operon in pCTX-M3orf35::cat cells would be visible for mobilizable multicopy plasmids, while the conjugation ability of the low-copy-number pCTX-M3orf35::cat would not benefit from the derepression of tra due to the limited number of accessible oriT-bearing plasmid molecules.
Similarly, the deletion of orf36, which is unique to IncL and IncM plasmids (14) and is dispensable for conjugation, increased the mobilization efficiency into E. coli and upregulated traH but not nikA or nikB. Moreover, the presence of additional copies of orf36 significantly impaired mobilization even in donors bearing the native pCTX-M3 conjugative transfer region. We propose that the expression of traH and probably also that of the downstream genes encoding the T4SS are additionally regulated by the orf36 product in a manner independent of nikAB transcription. The mechanism underlying the Orf35-and Orf36-dependent regulation is currently unknown and deserves further study, especially given that these predicted proteins do not contain known DNA-binding motifs (Table 1). Host ranges of the replicon and the conjugative transfer system of pCTX-M3. Earlier studies (26) have demonstrated that pCTX-M3 can be transferred to A. tumefaciens via conjugation. Unexpectedly, despite a number of attempts, we were unable to transfer pCTX-M3 from E. coli to A. tumefaciens by mating (Fig. 5B). However, our mobilization experiments demonstrated that the conjugation system of pCTX-M3 is highly efficient in transferring the mobilizable broad-host-range (oriV pBBR1 ) plasmid pToriT, which contains oriT pCTX-M3 , into A. tumefaciens (10 Ϫ4 transconjugants per donor after 30 min of mating).
The finding that pCTX-M3 is not transferred into A. tumefaciens is inconsistent with previous results (26) showing that the conjugative transfer efficiency of the entire pCTX-M3 into Alpha-, Beta-, and Gammaproteobacteria was on the order of 10 Ϫ5 transconjugants per donor cell after 24 h of mating. To investigate this discrepancy, we performed a 24-h mating experiment using E. coli DH5␣(pCTX-M3, pToriT) as the donor and E. coli, A. tumefaciens, Ralstonia eutropha, and Pseudomonas putida as recipients. In such a system, the transfer of pCTX-M3 during mating reflects the host range of both its conjugation system and the IncM replicon, while the transfer of pToriT, the mobilizable broad-host-range plasmid containing oriT pCTX-M3 , reflects the host range of the conjugation system of pCTX-M3 only. Transconjugants carrying pCTX-M3 were selected on plates containing gentamicin and rifampin, while those with pToriT were selected on plates with tetracycline and rifampin (A. tumefaciens, R. eutropha, and E. coli) or with kanamycin and rifampin (A. tumefaciens and P. putida). As shown in Fig. 7, the pCTX-M3 conjugation system is highly efficient in transferring pToriT into A. tumefaciens and R. eutropha (2 ϫ 10 Ϫ3 and 3 ϫ 10 Ϫ4 transconjugants per donor cell, respectively) and, with a lower efficiency, also into P. putida (3 ϫ 10 Ϫ6 per donor cell). In contrast, pCTX-M3 transconjugants were obtained only in the E. coli recipient. Thus, pCTX-M3 itself, when transferred, cannot be established in A. tumefaciens, R. eutropha, or P. putida. These results indicate that the host ranges of the pCTX-M3 conjugative transfer system and its replicon differ markedly; the former shows a broad range comprising Alpha-, Beta-, and Gammaproteobacteria, and the latter is restricted to Enterobacteriaceae. A similar observation concerning the differences between the host ranges of the conjugation system and the replicon has been reported previously for the narrow-hostrange mobilizable Klebsiella pneumoniae plasmids pIGMS31 and pIGMS32 (43). These plasmids replicate only in Gammaproteobacteria, but their mobilization systems enable the conjugative transfer of a heterologous replicon into several Alphaproteobacteria hosts by the RK2 (IncP1␣) conjugation system (43). In addition, recently, different host ranges of the conjugative and the replicative systems have been shown for the self-transferable P. putida plasmid NAH7 of the IncP-9 group (with the MPF T system) (44).
Concluding remarks. Although the conjugation system of pCTX-M3 belongs to the MPF I group, it differs from the one encoded by IncI1 plasmids. Therefore, it would be valuable to reevaluate the mobilization host range of MPF I systems. It has been shown that ssDNA transiently generated during conjugative transfer triggers the SOS response in recipient cells unless the plasmid codes for an anti-SOS factor (45). As a consequence, homologous recombination and integron integrase genes are induced, leading to DNA rearrangements (45). Therefore, plasmids with an Mpf host range broader than their replicon host range, such as pCTX-M3, which does not code for an anti-SOS factor but does bear an integron, may have a greater impact on the adaptability of bacterial populations than previously appreciated.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. The strains used in this work are listed in Table  2. E. coli DH5␣ was used as the host strain for DNA cloning. In the mating experiments, E. coli strain BW25113 or, where stated, strain DH5␣ bearing pCTX-M3 and its derivatives, was used as the donor, and E. coli strain JE2571Rif r was used as the recipient. In transspecies matings, Pseudomonas putida, Ralstonia eutropha, or Agrobacterium tumefaciens was used as the recipient. The bacteria were cultured with agitation in LB medium (Biocorp, Warsaw, Poland) or on agar-solidified LB plates (46) at either 37°C (E. coli and P. putida) or 28°C (A. tumefaciens and R. eutropha). When required, antibiotics were added to the medium at the following final concentrations (g/ml): ampicillin, 100; chloramphenicol, 20; gentamicin, 50; kanamycin, 50; rifampin, 100; tetracycline, 20 or 6 (for pToriT selection).
Cloning and DNA manipulation. Plasmid DNA was isolated by alkaline lysis using Plasmid Mini or Plasmid Midi kits (A&A Biotechnology, Gdań sk, Poland) according to the manufacturer's instructions. DNA cloning was performed according to standard protocols (46). All the enzymes used for cloning were from MBI Fermentas/Thermo Scientific (Vilnius, Lithuania).
Plasmid constructions. The plasmids that were constructed and used in this study are listed in Table  2 and Table S1 in the supplemental material. pCTX-M3 derivatives with deletions of genes from the tra and trb regions were constructed through lambda Red-mediated recombination (27). First, the BW25113(pKD46) strain was electrotransformed with pCTX-M3 and was selected on LB agar plates with gentamicin at 28°C. PCR products comprising the cat gene sequence with extensions homologous to the gene to be replaced on the pKD3 plasmid template were obtained using the primers listed in Table S2. Then, BW25113(pCTX-M3, pKD46) cells were electrotransformed with these DpnI-treated PCR products and were selected on LB agar plates with chloramphenicol at 37°C (to avoid the propagation of pKD46, which shows temperature-sensitive replication). Single colonies were isolated by the streak plate method at 37°C, and the correct integration of the cat gene into the target gene was verified by PCR (35 cycles) with the primer pairs listed in Table S3 and S4. The integration of cat in the four longest genes (nikB, pri, traU, and traY) was further analyzed by multiplex PCR with three primers (see Table S5): (i) catReVer, which anneals to the cat gene, (ii) a primer located upstream of the deleted gene (nikAF, priUVer, traUsU, or traYsU, respectively), and (iii) a primer that anneals to the gene to be deleted (nikBDVer, priDVer, traUDVer, or traYDVer, respectively). The primers were designed so that the expected products were smaller than 1 kb and enabled discrimination between the native pCTX-M3 and the appropriate mutant plasmid. All mutated plasmids were verified by sequencing with the catU142 primer. The loss of pKD46 was checked by multiplex PCR with the repKD46F and repKD46R primers, which were designed to amplify the repA101 (thermosensitive replication) gene fragment, and with the TEMfor and TnTEMrev primers (for amplification of the bla TEM-1 gene, which is present in both pCTX-M3 and pKD46) as an internal PCR control (see Table S4). The constructed plasmid derivatives are listed in Table S1.
The cat gene was eliminated from six pCTX-M3 derivatives, namely, pCTX-M3nikB::cat, pCTX-M3pri:: cat, pCTX-M3traP::cat, pCTX-M3traU::cat, pCTX-M3traW::cat, and pCTX-M3traY::cat, using pCP20, which encodes FLP recombinase, as a helper plasmid. In this process, the strain DH5␣(pCP20) was electrotransformed with the appropriate pCTX-M3 derivative, and transformants were selected on LB agar with gentamicin at 28°C. Then, the transformants were streaked in parallel on LB with both chloramphenicol and gentamicin and were grown at 37°C. After colony purification, the clones that were chloramphenicol sensitive and gentamicin resistant were verified by PCR with the primers listed in Tables S3 and S5. The loss of pCP20 was verified by PCR with the repKD46F and repKD46R primers. The plasmids were then introduced into the BW25113 strain.
To construct the plasmids carrying individual genes from the tra and trb regions for use in the complementation experiments (Table S1), specific genes were amplified by PCR using Pfu DNA polymerase (Thermo Fisher Scientific, Waltham, MA) with the primers listed in Table S6 and were cloned into the pMT5 or pAL3 vectors, as described in Table S1. Genes which, probably due to the harmful effects of high-level expression, were not able to be cloned into the multicopy plasmid pMT5 (oriV pMB1 ), were Conjugation System of the pCTX-M3 Plasmid Journal of Bacteriology cloned into the low-copy-number vector pAL3 (oriV P15A ). Only the pALpri plasmid was constructed without a PCR amplification step, as indicated in Table S1. The cloned genes were verified by sequencing (primers listed in Tables S3 and S4). The expression of the complementing genes cloned into the pAL3 and pMT5 vectors is driven by the lactose operon promoter (P lac ). Plasmids bearing oriT pCTX-M3 were obtained by cloning the oriT sequence into appropriate plasmids, as described in Table 2.
The plasmids pCS, pC35S, and pC36S (Table 2) were obtained by the digestion of pCTX-M3, pCTX-M3orf35::cat, and pCTX-M3orf36::cat, respectively, with SalI and the recircularization of the largest DNA fragment. Thus, these plasmids are devoid of all resistance genes except the bla TEM-1 and bla CTX-M-3 genes present in pCTX-M3.
PCR conditions. PCR was performed in a Veriti thermal cycler (Applied Biosystems, Foster City, CA) using DreamTaq or Pfu DNA polymerase with the supplied buffers (Thermo Fisher Scientific), deoxynucleoside triphosphate (dNTP) mixture, template DNA (purified DNA or bacterial colonies), and the appropriate primer pairs listed in Tables S2 to S6, according to the manufacturer's recommendations.
DNA sequencing. The sequencing was performed in the DNA Sequencing and Oligonucleotide Synthesis Laboratory at the Institute of Biochemistry and Biophysics, Polish Academy of Sciences, using a dye terminator sequencing kit and an automated sequencer (ABI 377; PerkinElmer, Waltham, MA).
Real-time quantitative PCR. RNA was isolated from cells of BW25113 strains bearing a specific pCTX-M3 deletion derivative alone or in combination with a plasmid carrying an appropriate complementing gene in the late exponential phase of growth (optical density at 600 nm [OD 600 ] of 0.8 to 1) using a GeneJET RNA purification kit (Thermo Fisher Scientific) according to the manufacturer's protocol. RNA quality and integrity were checked by agarose gel electrophoresis, and the concentration was estimated using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). Three biological replicates were analyzed.
Reverse transcription was performed with random hexamer primers using the Maxima First Strand cDNA synthesis kit for RT-qPCR with dsDNase (Thermo Fisher Scientific). The specific qPCR primers used to amplify the reference (the repA gene encoding the replication initiator protein of the pCTX-M3 plasmid) and target genes are listed in Table S7. Real-time PCR was carried out using Real-Time 2ϫHS-PCR master mix SYBR (A&A Biotechnology) in a final volume of 10 l in the LightCycler 480 system (Roche Life Sciences, Penzberg, Germany) with an initial denaturation at 95°C for 5 min followed by 40 cycles of amplification (95°C for 10 s and 50°C for 10 s). The relative gene expression in the deletion strains was calculated and normalized to the value obtained for a strain carrying the native pCTX-M3 plasmid (47).
Plasmid conjugative transfer. Cultures of the donor and recipient strains (approximately 10 8 CFU ml Ϫ1 ) were grown in LB to stationary phase; the cultures were then washed twice with LB medium, and the donor strain was resuspended in a volume of LB equal to the initial volume of the culture, while the recipient strain was resuspended in one-fourth of the initial culture volume. Then, 0.5 ml of the donor and recipient suspensions was mixed and filtered through a sterile Millipore HA 0.45-m filter (Millipore, Billerica, MA). The filter was incubated on an LB plate for 30 min at 37°C (E. coli) or 28°C (A. tumefaciens and E. coli). For the pCTX-M3 host range tests, the filter was incubated on an LB plate for 24 h at 37°C (P. putida and E. coli) or 28°C (A. tumefaciens and R. eutropha). Bacteria were washed from the filter with 1 ml of a sterile 0.85% NaCl solution. Conjugation was stopped by vigorously vortexing the mating mixture for 30 s and then placing it on ice. Serial dilutions of the mixture of donor, recipient, and transconjugant cells were plated on LB agar supplemented with the appropriate selection antibiotics. As a control, the donor and recipient cells were plated on LB supplemented with antibiotics for transconjugant selection. Mating in liquid was performed as described above but without the use of the filter: the mating mix was incubated for 30 min at 37°C, and conjugation was stopped by vortexing for 30 s. The conjugative transfer efficiency is equivalent to the number of transconjugants per donor cell.
Statistical analysis. Data concerning the plasmid conjugative transfer frequencies are presented as the means Ϯ standard deviations (SDs). The differences between the mobilization efficiencies of pToriT by the different plasmids (Fig. 5) were tested for statistical significance using the t test (Prism 6; GraphPad Software, Inc., La Jolla, CA). P values of Ͻ0.05 were considered statistically significant.