Target Joining of Duplicated Insertion Sequence IS21 Is Assisted by IstB Protein In Vitro

Abstract

Tandemly repeated insertion sequence IS21, located on a suicide plasmid, promoted replicon fusion with bacteriophage λ in vitro in the presence of ATP. This reaction was catalyzed in a cell extract containing the 45-kDa IstA protein (cointegrase) and the 30-kDa IstB helper protein of IS21 after both proteins had been overproduced in Escherichia coli. Without IstB, replicon fusion was inefficient and did not produce the 4-bp target duplications typical of IS21.

Insertion sequence IS21 occurs on the broad-host-range plasmid R68 and contains two genes, istAB, in an operon (22, 29). The istA gene encodes, in frame, a 46-kDa and a 45-kDa protein, termed transposase and cointegrase, respectively (25). Transposase essentially catalyzes transposition of single IS21 elements. Cointegrase rarely carries out this reaction but is highly effective in another type of transpositional recombination, i.e., replicon fusion between plasmids carrying an IS21 tandem duplication [(IS21)₂] and target replicons (25). Formally, the products of this replicon fusion reaction are cointegrates. However, they are formed by a cut-and-paste transpositional mechanism (20) rather than by replicative transposition, which, for example, underlies the move of bacteriophage Mu or transposon Tn3 (18, 26). The 30-kDa helper protein IstB contains an ATP-GTP binding motif (7), as do the transposition helper proteins MuB of phage Mu and TnsC of transposon Tn7 (3, 18). In the absence of IstB, IS21 transposition cannot be detected and (IS21)₂-mediated cointegration occurs at very low frequencies (21, 25).

Under optimal conditions in Escherichia coli, the frequency of (IS21)₂-dependent replicon fusion approaches 10⁻¹, as measured by a mating-out assay (25). This high frequency is made possible on the one hand by an IS21 tandem with an optimal 4-bp spacer (5′-TATA-3′) between the two IS21 elements and on the other hand by the overexpression of cointegrase and IstB in trans by using the expression vector pJF118EH (25). When the overexpressed istAB gene products are supplied in trans, a fragment carrying the reactive IS21-IS21 junction suffices to give replicon fusion (21, 25). Encouraged by the high in vivo activity of cointegration, we have now set up an in vitro system. Previously, we have shown that an istA gene product (probably the cointegrase) can cleave the IS21-IS21 junction at the inner 3′ ends of the IS21 elements in vitro (21). This cleavage, which does not require IstB, results in a staggered cut, exposing 3′-OH groups of the terminal nucleotides (A) in each inverted repeat (see the IS21-IS21 junction of pME3940 [Fig. 1]). We have now obtained evidence that strand transfer from the IS21-IS21 junction of (IS21)₂ to λ DNA can occur in vitro.

FIG. 1.

In vitro cointegration assay. The incubation mixture contained concatemerized λgtWES.λB DNA (●, cos site), the pVS1-derived suicide plasmid pME3940 carrying the reactive IS21-IS21 junction on a 1.3-kb fragment from pME3918 (25), and the supF (amber suppressor) gene on a 0.21-kb fragment, as well as IS21 transposition proteins in the cell extract. The DrdI cleavage site (indicated by vertical arrows) at the artificial IS21-IS21 junction may mimick the cuts made by IstA (21). The assay of λgtWES.λB::pME3940 cointegrates is explained in the text. IS21L′ and IS21R′, truncated left and right IS21 elements, respectively; IR, inverted repeat; bla, ampicillin resistance gene.

Components of the in vitro cointegration system.

The principle of the in vitro cointegration system shown in Fig. 1 is inspired by in vitro systems devised for murine leukemia virus, the Ty element of Saccharomyces cerevisiae, and Tn3 (1, 4, 9). The istA and istB genes were overexpressed in E. coli ED8767 (metB recA56 hsdS supE supF) (23), using the previously described inducible tac promoter constructs pME3902, pME3913, pME3910, pME3944, and pME3945 (Table 1) (25). Cells were grown aerobically at 37°C in L broth (23) containing 100 μg of ampicillin per ml to a density of about 5 × 10⁸ per ml, induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 2 h, harvested by centrifugation, and resuspended in 25 mM HEPES–1 mM EDTA, pH 7.5, to a density of approximately 10¹² per ml. Cells were placed on ice; KCl and dithiothreitol were added to final concentrations of 100 and 2 mM, respectively. Cells were shaken gently at 0 to 4°C and treated successively with 250-μg/ml lysozyme for 20 min and 10 mM MgCl₂ for 30 min. Thereafter, cells were frozen in liquid N₂ and lysed by thawing on ice. Cell debris was removed by centrifugation in an Eppendorf 5415C centrifuge at 14,000 rpm for 30 min, and the cell extract was stored in 10-μl aliquots (quick frozen in liquid N₂) at −70°C. Under these conditions, extracts were stable for several months.

TABLE 1.

In vitro replicon fusion between the (IS21)₂ suicide plasmid pME3940 and concatemeric λgtWES.λB

Expression construct used for extract preparationa	IS21 gene(s) overexpressedb	Cointegration rate,c h−1(n)
pJF118EH		<2 × 10−7 (1)
pME3902	istA (P45, P46), istB	3 × 10−4 (3)
pME3913	istA (P45), istB	2 × 10−4 (2)
pME3910	istA (P46), istB	2 × 10−5 (2)
pME3944	istA (P45, P46)	2 × 10−6 (2)
pME3945	istB	<3 × 10−7 (2)
pME3944 + pME3945	istA (P45, P46) + istB	2 × 10−2 (3)

^aAll recombinant plasmids were derived from the expression vector pJF118EH (25); they were harbored by E. coli ED8767. 

^bistA (P46) encodes transposase; istA (P45) encodes cointegrase. 

^cThe cointegration (replicon fusion) rate was calculated by determining the number of PFU formed on E. coli CES200 (sup⁰) per PFU formed on E. coli LE392 (supF) per h of incubation. This value was divided by 0.27, the fraction of λgtWES.λB that tolerates an insertion, and multiplied by 0.1, a correction factor that compensates for the different packaging efficiencies of the parental and recombinant phages (1). Mean values were calculated for the data from n independent experiments. 

The donor plasmid pME3940 (Fig. 1 and 2) was constructed essentially as follows. The 6.8-kb Pseudomonas vector pME290, which is derived from pVS1 and cannot replicate in E. coli (10, 11), was used to clone the IS21-IS21 junction region with the 4-bp spacer 5′-TATA-3′ from pME3918 (25) and the supF gene from pBRG1310 (19) in Pseudomonas aeruginosa PAO25 (6). Plasmid pME3940 was purified from strain PAO25 by centrifugation in a CsCl-ethidium bromide density gradient (10).

FIG. 2.

Analysis of pME3940 insertions in λgtWES.λB. (a) The λ map and genotype symbols are according to Leder et al. (12) and Hendrix et al. (8), and the orientation of pME3940 (indicated by a flag) is defined below. rep, replication; sta, segregational stability; bla, ampicillin resistance gene. (b) The insertion sites were sequenced by the method of Manfioletti and Schneider (15), using 15-mer primers specific for either IS21 end. IstA/IstB, transposition proteins provided by pME3902; IstA, cointegrase and transposase provided by pME3944; IstB, helper protein provided by pME3945; IstA + IstB, transposition proteins provided by a mixture of ED8767/pME3944 and ED8767/pME3945 extracts; IstA(P45)/IstB, cointegrase and helper protein provided by pME3913. The asterisk indicates that there was a 14-bp deletion at the insertion site. Underlined nucleotides in boldface indicate the target duplications.

Concatemerized λgtWES.λB DNA served as the target. Phage λgtWES.λB was propagated on E. coli LE392 (supF58 supE44 hsdR514 galK2 galT22 metB1 trpR55 lacY1) (12, 23). Phage DNA isolated by standard methods (23) was purified on a Qiagen midicolumn according to the instructions of the supplier and ligated at 200 μg/ml (1).

The in vitro reaction was carried out in a 20-μl volume containing 14 μl of reaction buffer (25 mM HEPES [pH 7.5], 10 mM MgCl₂, 50 mM KCl, 1 mM dithiothreitol, 50 μg of bovine serum albumin/ml, and 5% [wt/vol] polyvinylpyrrolidone K90, amended with 1.8 mM ATP and 143 μM each dATP, dCTP, dGTP and dTTP [21]), 4 to 5 μl of DNA (0.2 μg of pME3940 plus 1 μg of concatemeric λgtWES.λB), and 1 to 2 μl of a cell extract containing overexpressed istAB gene products (protein concentration, 20 to 30 mg/ml). This mixture was incubated on ice for 10 min and at 30°C for 60 to 120 min. The reaction was stopped by mixing a 4-μl aliquot with a commercial λ packaging extract (Gigapack II Gold; Stratagene). Packaging was performed as described by the supplier. The phage lysate thus obtained was titrated (23) on two maltose-grown indicator strains of E. coli, LE392 and CES200 (sbcB15 recB21 recC22 hsdR hisG4 thi-1 leuB6 Δ[gpt-proA]62 argE3 thr-34::Tn10 lacY1 galK2 ara-14 kdgK51 mtl-1 xyl-5 rfbD1 tsx-33 rpsL31) (4). Phage λgtWES.λB, which contains three amber mutations, makes plaques on the amber suppressor strain LE392. On the sup⁰ strain CES200, λgtWES.λB will form plaques only when it contains a pME3940 insertion providing the amber suppressor supF (Fig. 1). Thus, the frequencies of λgtWES.λB::pME3940 cointegrates can be determined (see footnote b of Table 1).

Roles of the istAB gene products in replicon fusion in vitro.

Incubation with a cell extract prepared from E. coli ED8767 carrying pME3902 (expressing cointegrase, transposase, and IstB) produced 3 × 10⁻⁴ cointegrates per h (Table 1). A 2-h incubation raised the cointegration frequency to 10⁻³. In these experiments, typically several thousands of plaques were obtained per microgram of DNA on the sup⁰ host CES200. A vector (pJF118EH) control did not yield cointegrates (Table 1), indicating that illegitimate recombination did not interfere. When IstB was present, cointegrase (encoded by pME3913) gave a higher cointegration rate than did transposase (encoded by pME3910) (Table 1), suggesting that most of the cointegrates formed in a cointegrase-transposase mixture are the result of cointegrase activity. This situation has previously been observed in vivo (25). However, the in vitro cointegration rates measured should be interpreted cautiously since the amounts of the istAB gene products in the cell extracts are not known. (In Coomassie blue-stained gels, the IstA and IstB proteins were not conspicuous after expression in E. coli carrying istAB constructs.) Importantly, the pME3940 insertions (no. 1 to 7 [Fig. 2]) that were obtained from several pME3902-dependent reactions all occurred at different sites in λgtWES.λB; in each case, the junction sequence 5′-TATA-3′ (identifiable as a DrdI restriction site [Fig. 1]) was lost from the donor pME3940, and each insertion was flanked by a 4-bp target duplication (Fig. 2). IS21 is known to generate target duplications of 4 bp normally and 5 bp exceptionally (23, 25). An extract prepared from strain ED8767/pME3913, which contained cointegrase and IstB but no transposase, also allowed the formation of regular 4-bp target duplications (no. 13 and 14 [Fig. 2]).

In the absence of IstB, the istA construct pME3944 gave a low cointegration rate (Table 1) and none of the four pME3940 insertions analyzed (no. 8 to 11 [Fig. 2]) was flanked by a 4-bp direct repeat. Addition of an extract of E. coli cells carrying the istB expression plasmid pME3945 sharply increased the cointegration rate (Table 1) and restored the typical 4-bp target duplication in the example checked (no. 12 [Fig. 2]). In this extract, the IstB protein was clearly visible in Coomassie blue-stained gels, representing a few percent of the total protein (data not shown). An extract containing IstB, but not IstA, was inactive in cointegration assays (Table 1). These results suggest that IstB facilitates intermolecular strand transfer in vitro in the presence of cointegrase. IstB may also be involved in the appropriate alignment of target DNA with a putative donor-cointegrase complex, since the formation of regular target duplications requires IstB.

When ATP and the four deoxynucleoside triphosphates (dNTPs) were not included in the reaction buffer, no cointegrate formation was detected. A control showed that the dNTPs inhibited nucleolytic degradation of the linear λ concatemers by the E. coli extract; it is not clear whether the dNTPs are involved in the cointegration reaction. ATP stimulated cointegration about 1,000-fold (data not shown). A role for ATP is plausible considering the presence of an ATP-GTP binding motif in IstB (7).

At 4°C, the cointegration activity of ED8767/pME3913 extracts decayed rapidly and attempts to stabilize this activity by the addition of compounds known to stabilize proteins were not successful. Moreover, overexpressed istA gene products had a marked tendency to form insoluble aggregates. The IS21 cointegration system, therefore, has not reached the sophistication of some other in vitro transposition systems based on purified components (2, 5, 13, 14, 17, 24, 27, 28).

In conclusion, IstB of IS21 and MuB of phage Mu may have comparable functions. MuB, a DNA binding protein with ATPase activity, stimulates intermolecular strand transfer (16). In the absence of MuB, the MuA transposase can carry out slow intramolecular strand transfer but hardly any intermolecular strand transfer. MuB is important for the selection of proper target DNA sites and may serve as a scaffold which directs the assembly of the transposition complex (18, 24). In our in vitro system, IstB stimulated the intermolecular joining reaction at least 100-fold. The cointegrates formed in the absence of IstB may be the products of residual, imprecise strand transfer by the IstA protein(s). Alternatively, the cleaved IS21-IS21 donor (21) may be joined to the target by illegitimate recombination brought about by some proteins in the E. coli cell extract. Irrespective of this mechanistic aspect, target capture is an important function of IstB.