Int-B13, an Unusual Site-Specific Recombinase of the Bacteriophage P4 Integrase Family, Is Responsible for Chromosomal Insertion of the 105-Kilobase clc Element ofPseudomonas sp. Strain B13

Escherichia coli DH5α (31) was used for routine cloning experiments with plasmids. E. coliBL21(DE3)(pLysS) (35) was used as a host strain for inducible protein overexpression of pET3c-derived plasmids (29). E. coli cultures were grown at 37°C on Luria-Bertani medium (31), which was supplemented with the following antibiotics when appropriate: ampicillin, 100 μg/ml; chloramphenicol, 10 μg/ml; and tetracycline, 20 μg/ml. For induction of int-B13 in E. coli BL21, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to the medium at a 1 mM concentration.

The plasmid constructs used in this study are listed in Table 1. The plasmids pUC18Not (16), pUC28 (4), and pACYC184 (New England Biolabs, Beverly, Mass.) were used as general cloning vehicles. Plasmid pET3c (29) is an ATG vector derived from pBR322 which contains the φ₁₀ promoter, ribosome binding site, and terminator optimized for T7-directed protein expression. The linearized vector pGEM-T Easy (Promega Corporation, Madison, Wis.), containing single 3′-thymidine overhangs, was used for cloning of DNAs amplified by PCR.

For overexpression of the integrase gene int-B13 inE. coli, a translational fusion of the gene was constructed by using the ATG triplet in the NdeI site located downstream of the φ₁₀ promoter and the ribosome binding site in pET3c as the start codon. The start of theint-B13 gene was taken as position 262 (Fig.1). First, a 379-bp PCR product was generated with primers RR330 and RR331 (Table2) and subsequently digested withNdeI and SfiI to create ligation termini. In a three-point ligation, the resulting DNA fragment (NdeI/SfiI) and a fragment from pRR165 with the remainder of the int-B13 open reading frame (ORF) (SfiI/BamHI) were cloned into pET3c (BamHI and NdeI digested) to produce pRR169. The PCR-derived part of pRR169 was sequenced and confirmed to be identical to the original nucleotide sequence. Plasmid pRR169ΔNot is identical to pRR169 except for a frameshift mutation in the uniqueNotI site within the int-B13 coding sequence. The frameshift mutation was introduced by digestion of pRR169 withNotI, filling in of the 3′ recessed ends, and religation. The presence of 4 additional nucleotides (nt) was confirmed by sequencing of this region of pRR169ΔNot.

Plasmid pRR171 contained the right end of the clc element with the int-B13 gene plus the attB1 site ofP. putida F1 cloned in pACYC184. Plasmid pRR172 was constructed by replacing the 780-bp AatI-EcoRI fragment in pRR171 with a 200-bp AatI-EcoRI fragment from pRR146, thereby combining int-B13 andattP as on the original circular form of the clcelement. Plasmid pRR172ΔNot contained a frameshift mutation inint-B13 as described above for pRR169ΔNot.

E. coli BL21(DE3)(pLysS) harboring plasmid pRR169 or pRR169ΔNot was grown in Luria-Bertani medium to an optical density at 540 nm of 0.45 to 0.55. Subsequently, cells were induced by the addition of 1 mM IPTG, and the culture was grown for another 90 min. Bacterial cells (1 ml) were harvested by centrifugation, resuspended in 50 μl of protein loading buffer (19), and boiled for 5 min. After 1 min of centrifugation (at 15,000 × g), samples of 5 to 10 μl were used directly for analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), which was performed according to the method of Laemmli (19).

Plasmid DNA isolations (from E. coli DH5α), transformations, and other DNA manipulations were carried out according to established procedures (31). Total DNA was isolated with the Easy-DNA kit (Invitrogen, Carlsbad, Calif.) or by the method of Marmur (20). Restriction enzymes and other DNA-modifying enzymes were purchased from Amersham Life Science (Little Chalfont, Buckinghamshire, United Kingdom) and used according to the specifications of the manufacturer.

PCRs were performed withTaq polymerase according to the descriptions of the supplier (Life Technologies, Basel, Switzerland). PCR primers used in this study (Table 2) were purchased from MWG Biotech (Ebersberg, Germany) or Microsynth (Balgach, Switzerland). The method referred to as colony PCR was performed as follows. One bacterial colony from an agar plate was transferred with a sterile toothpick into a 0.5-ml PCR tube containing 100 μl of distilled sterile water. The sample was heated to 98°C for 6 min in order to lyse the bacterial cells and release their DNA. For DNA amplification by PCR, 1 μl of this solution was added to a PCR mixture with a total volume of 50 μl.

Double-stranded template sequencing was performed on plasmids with the Thermo Sequenase fluorescently labelled primer cycle kit with 7-deaza-dGTP (Amersham Life Science). Primers labelled with the fluorescent dye IRD-800 at the 5′ end were purchased from MWG Biotech. An automated DNA sequencer, model 4000L (LI-COR Inc., Lincoln, Nebr.), was used for sequencing. Computer analysis of the DNA and amino acid sequences was done with DNASTAR software (DNASTAR Inc., Madison, Wis.). Comparisons of our own sequence data with published sequences in GenBank were performed with the BLAST software via the Internet (http://www.ncbi.nlm.nih.gov/BLAST/ [2]).

The nucleotide sequence presented in this article (Fig. 1) has been deposited in GenBank under accession no. AJ004950 .

Previously, we have reported chromosomal integration of a 105-kb genetic element (named the clc element) at two sites in P. putida F1 (25). The two integration sites in F1 were both identified as glycine tRNA structural genes, and each integration appeared to occur at the 3′ end of the targetgly-tRNA. This observation suggested that a site-specific recombinase was responsible for the chromosomal integrations. Near the right end of the clc element (insert of plasmid pRR108 [Table 1]), an ORF was identified by sequencing and its predicted amino acid sequence was homologous to those of site-specific recombinases of the bacteriophage P4 integrase subfamily. The nucleotide sequence of this ORF (tentatively named int-B13) had a length of 1,971 bp, corresponding to a coding capacity of 657 amino acids (aa) and a molecular mass of 74 kDa. To confirm the presence and actual size of the int-B13-encoded polypeptide, the gene was cloned into pET3c and overexpressed in E. coli BL21(DE3)(pLysS). The DNA sequence of the int-B13ORF predicted two possible translational starts close to another (Fig.1, nt 262 and 304). Plasmid pRR169 contained the int-B13gene under the transcriptional control of the T7 promoter and carried a translational fusion from the ATG start codon at nt 262. When anE. coli BL21 culture with pRR169 was induced with IPTG and the crude extract was separated by SDS-PAGE, a protein band of the expected size (74 kDa) for Int-B13 was obtained (Fig.2). However, at this resolution it was not possible to discern whether the ATG at nt 262 or the second ATG codon at nt 304 was actually used (or whether both were used). To confirm the origin of this protein band from the int-B13gene, a frameshift mutation was introduced in the uniqueNotI restriction site at nt 1158 in the sequence ofint-B13 (Fig. 1B). Induction of E. coli BL21 carrying this plasmid (pRR169ΔNot) produced a protein of 56 kDa on an SDS-PAGE gel (Fig. 2), which was the expected size for the truncated Int-B13. This confirmed the coding capacity of the int-B13ORF.

A clustal alignment (MegAlign, DNASTAR software) of Int-B13’s amino acid sequence with the seven most similar integrases is shown in Fig. 3A. The protein sequence with the highest similarity to Int-B13 is the integrase of retronphage φR73 (37) with 37% identity and 57% homology in an alignment covering 407 aa. The conserved residues His-396, Arg-399, and Tyr-433 (integrase family positions) found in all site-specific recombinases of the integrase (Int) family, with the exception of pSAM2 Int (3, 8), were also present in the Int-B13 sequence (His-365, Arg-368, and Tyr-401 [Fig. 3A]). The conserved Tyr-342 of λ Int (corresponding to Tyr-401 for Int-B13) has been shown to be the residue forming a covalent bond between the Int protein and the DNA at the attachment site(s) during recombination (23). In accordance with the findings of Abremski and Hoess (1), a second conserved arginine residue was also found in Int-B13, Arg-261 (Fig. 3A). The Int-B13 protein, however, was considerably longer than the phage P4-related integrases, which are all between 385 and 440 aa in length. The C-terminal 220 aa of Int-B13 had high similarity to only one other polypeptide in GenBank, encoded by an ORF present near thepah gene cluster for naphthalene degradation found inPseudomonas aeruginosa PaK1 (Fig. 3B). No function has been assigned to the predicted protein from this ORF. Interestingly, theP. aeruginosa ORF could be part of a larger ORF, since it is located at the end of the DNA fragment sequenced. The translated sequence upstream of the ORF also matched the Int-B13 protein sequence very well (Fig. 3B).

Site-specific recombinases like the integrases of bacteriophages P4 and φR73 mediate recombination between a phage attachment site, attP, and a bacterial chromosomal attachment site, attB (9). Likewise, the junction sequences between a prophage (or another type of integrated element) and chromosomal DNA are termed attL(left-end junction) and attR (right-end junction). During integration, the actual recombination event involving strand exchange occurs within short sequences identical to both attP andattB, the att core or identity segment. Since the putative chromosomal attachment sites attB1-F1 of P. putida F1 (formerly INT1) and attB2-F1 (INT2) had been identified previously (25), we now determined theattP site of the clc element itself. The precise sequence of the attP site would be evident only from a circular form with a direct junction between the left and right ends, which in the integrated form are 105 kb apart. This junction could be amplified from total DNA of Pseudomonas sp. strain B13 by using PCR with primers RR316a (left end) and RR319 (right end) (Table2). As expected, the DNA sequence (insert of pRR146) of the amplified fragment contained the left- and right-end sequences and an 18-bp segment identical to the 3′ end of glyV (Fig.4). The 18-bp segment seemed to be the core sequence of the clc element’sattP, since only this segment was 100% identical to part of the chromosomal attB sites.

The PCR was used to amplify the junctions between chromosomal DNA and the left end of the integrated clcelement (attL) in different recipients. The aim was to analyze whether glyV would be used as the target for Int-B13-mediated integration in hosts other than the previously analyzed P. putida F1 (25). The exact junction would be evident only from the nucleotide sequence at the left end, where the remainder of the glyV gene from attB is found. The bacteria analyzed for this purpose werePseudomonas sp. strain B13 itself, Ralstonia sp. strain S11 (26), and B. cepacia JH230 (34). First, we determined by Southern hybridizations ofNaeI- and SphI-digested total DNAs what fragment sizes were to be expected from inverse PCR (iPCR) and what the copy number of the element in each strain was. From these hybridizations it was evident that the B13 genome contained two separate copies of theclc element, whereas Ralstonia sp. strain S11 andB. cepacia JH230 both contained only one integrated element (results not shown). The left junction of one of the integration sites of the clc element in strain B13 was amplified by iPCR. For this purpose, total DNA was digested with NaeI, religated, and subjected to PCR with the primers RR315 and RR316. Cloning and sequencing revealed the 3′ end of a glycine tRNA gene directly adjacent to the left end of the clc element, as inP. putida F1 (Fig. 4). A second, complete Gly-tRNA gene was detected 99 bp to the left of the integration site in strain B13.

The iPCR approach used to amplify attL from strain B13 did not yield any product for Ralstonia sp. strain S11 andB. cepacia JH230. Instead, for strain S11 we successfully used a linker-mediated PCR. Total DNA of this strain was digested withSphI, ligated to a linker (i.e., SphI-digested pUC28 DNA), and subjected to PCR with the primers RR316a and RR332. Sequencing of the cloned PCR product again revealed the 18-bp 3′ end ofglyV adjacent to the left end of the clc element (Fig. 4). Downstream of the partial glyV sequence, another putative structural tRNA with 77% identity to cysT fromStreptomyces lividans was found. For B. cepaciaJH230, an inverse nested PCR was needed for amplification of the left junction. Then total DNA from strain JH230 was digested withNaeI, religated, and subjected to two rounds of PCR, first with primers RR315 and RR316 and then with primers RR316a and RR327. This procedure resulted in a PCR product whose junction sequence again contained the 3′ end of glyV (Fig. 4). Also here, a putative cysteine tRNA gene (86% similarity to cysT fromE. coli K-12) was located downstream of the integration site.

Next, we examined if the clc element would always integrate in the same manner in one recipient strain. Six independently obtained transconjugants of P. putida F1 from matings with strain B13 (25) were analyzed for the left junction at theattB1 site. Total DNA from the transconjugants was amplified in the PCR with primers RR301 and RR316. The cloned PCR products were sequenced and found to be identical (results not shown). In all transconjugants, the left end of the clc element had joined the target glycine tRNA gene exactly 18 bp from its 3′ end.

The characterized attachment sites were all identical with respect to the last 18 bp (3′ portion of glyV) adjacent to the exact left end of the clc element. This indicated that integrations occurred with high accuracy at this position on the 3′ end of the Gly-tRNA gene. In addition to the 18-bp identity segment found in all attachment sites, some other conserved sequences were also observed. The attP site of the clc element andattB1 had a sequence identity of 83% in a 92-bp overlap including the 18-bp segment. The clc element’sattP site and the attB1 and attB2target sites from P. putida F1 had similar AT-rich inverted repeat sequences close to the target glyV gene (25), and the attL site of Ralstoniasp. strain S11 also contained an inverted repeat sequence (Fig. 4). Such structures are commonly found adjacent to (clusters of) tRNA structural genes, but their function or role in the attachment recognition is unknown.

To demonstrate that theint-B13 gene product was sufficient to promote site-specific recombination between attP and attB, we looked for integrase activity in E. coli DH5α, a strain which by itself is deficient in recombination. We cloned theint-B13 gene plus the attP sequence on one plasmid (as on the circular form of the clc element) and looked at integration into a cloned attB site, present on a second plasmid. Surprisingly, the attP region could not be combined with the int-B13 gene on a high-copy ColE1-based replicon, but only on the low-copy vector pACYC184 (i.e., pRR172 [Fig.5]). Even then, E. coliDH5α cells harboring plasmid pRR172 had a reduced growth rate. These observations indicated that some kind of detrimental activity was associated with the construct. E. coli DH5α cotransformed with the plasmids pRR123 (containing attB1) and pRR172 (attP plusint-B13) showed the formation of a chimeric plasmid in some instances. Restriction enzyme analysis of this chimeric plasmid indicated an integrative recombination between the attP site of pRR172 and the attB1 site of pRR123 (results not shown). DNA sequencing of the chimeric plasmid revealed that the sites of recombination were in fact identical to those previously characterized in the F1 transconjugant RR221 (25). The chimeric plasmid contained one attR site with a complete copy ofglyV and one attL site with only the 18-bp 3′ end of glyV (Fig. 5).

Plasmid isolations from E. coli DH5α (pRR123 plus pRR172) cultivated in the presence of ampicillin and tetracycline usually showed the presence of the two original plasmids, whereas the chimeric plasmid was observed only occasionally. To determine the integration in a more sensitive and statistically reliable manner, we used PCR on individual transformants. Colonies of E. coli DH5α cotransformed with pRR123 plus pRR172 were subjected to colony PCR with the primers RR303 and RR319 (Table 2) (Fig. 5), resulting in an 852-bp PCR product specific for the chimeric plasmid (Fig. 5). From each of three independent cotransformations of pRR123 plus pRR172, 10 randomly chosen colonies were analyzed by PCR. This resulted in 66.7% of the colonies being positive for the chimeric plasmid form (Table3). As controls, we used cotransformations of (i) a plasmid with a frameshift inint-B13 but otherwise identical to pRR172 (pRR172ΔNot) and a plasmid with the attB1 site (pRR123) and (ii) pRR123 and a plasmid containing an intact int-B13 but without theattP site (pRR171). In these cotransformations, only 6.7% of the colonies were positive for the chimeric form (Table 3). The results indicated that activity of an intact int-B13gene was needed for precise integration into attB and thatattP (but not attR) was required for this integration to occur. The few positives obtained with pRR171 and pRR172ΔNot could have been PCR artifacts, resulting from hybridizations between DNA transcripts from pRR123 and those from the second plasmid. Since the regions attP andattB1-F1 contain nearly identical sequence segments of 92 bp, transcripts terminating in either region could in the next cycle of annealing hybridize to the wrong template. We cannot exclude the possibility that some integrase activity resulted from pRR171 or pRR172ΔNot, but a chimeric plasmid could not be isolated when either of these plasmids was cotransformed with pRR123.