In vitro site-specific integration of bacteriophage DNA catalyzed by a recombinase of the resolvase/invertase family

Abstract

The genome of the broad host range Streptomyces temperate phage, phiC31, is known to integrate into the host chromosome via an enzyme that is a member of the resolvase/invertase family of site-specific recombinases. The recombination properties of this novel integrase on the phage and Streptomyces ambofaciens attachment sites, attP and attB, respectively, were investigated in the heterologous host, Escherichia coli, and in an in vitro assay by using purified integrase. The products of attP/B recombination, i.e., attL and attR, were identical to those obtained after integration of the prophage in S. ambofaciens. In the in vitro assay only buffer, purified integrase, and DNAs encoding attP and attB were required. Recombination occurred irrespective of whether the substrates were supercoiled or linear. A mutant integrase containing an S12F mutation was completely defective in recombination both in E. coli and in vitro. No recombination was observed between attB/attB, attP/attP, attL/R, or any combination of attB or attP with attL or attR, suggesting that excision of the prophage (attL/R recombination) requires an additional phage- or Streptomyces-encoded factor. Recombination could occur intramolecularly to cause deletion between appropriately orientated attP and attB sites. The results show that directionality in phiC31 integrase is strictly controlled by nonidentical recombination sites with no requirement to form the topologically defined structures that are more typical of the resolvases/invertases.

Figure 1

Organization of the region encoding int in φC31 and its comparison with the attB sequence from S. ambofaciens (adapted from ref. 20). (a) Organization of attP and int from φC31 and restriction sites relevant to this study. The black box represents the int ORF reading left to right, and the green, vertical arrowhead represents the position of attP. The stem-loop icon shows the position of a putative rho-independent transcription terminator. (b) The sequences of the attP and the N-terminal region of int (green) and attB (blue). The crossover occurs within the boxed nucleotides (20). The extent of the 84-bp attP site used in recombination in E. coli (Fig. 2) is shown in bold. Underlined sequences in attP/int show the positions of the BsrI and EcoRI restriction sites. Horizontal arrows indicate the positions of inverted repeats in attP and attB. In the attP/int sequence, the start of translation, the N-terminal 18 residues and the position of the S12F mutation are indicated. The attL and attR sequences are shown in yellow and pink, respectively.

Figure 2

Recombination by φC31 integrase between attP and attB sites in E. coli DH5α recA⁻. (a) Restriction maps of plasmids used and expected recombinant products. E. coli containing pHS33 and pHS34 encode attP and attB, respectively, and express the φC31 int gene from the tac promoter located within the vector sequences; the vector in each case is a λ defective plasmid, pZMR100. The attB (i) and attP (ii) sites have been placed on compatible plasmids. (b) Restriction analysis by SphI of parental and recombinant products after extraction of plasmid DNA from E. coli. Bands containing recombination products attL and attR are indicated by arrows. In lane 6 the recombinant fragments migrate close together but can be resolved after a longer electrophoresis. In lane 8 the attL recombinant product and linearized pHS34 comigrate. Molecular weight markers (M) are provided by 0.5 μg of 1-kbp ladder (Life Technologies).

Figure 3

Recombination by φC31 integrase between attP and attB sites in vitro. (a) Restriction maps of plasmids used and expected recombinant products. Detection of the recombination products in b–d was by restriction with EcoRI followed by agarose gel electrophoresis. Parental and recombinant products are indicated. (b) A time course of recombination in vitro. (c) No effect on recombination by MgCl₂ and abolition of activity by the S12F mutation. Recombination reactions were incubated at 30°C for 16 hr before analysis. The smear of degraded DNA in the presence of the S12F mutant integrase is probably because of contaminating nucleases. (d) Recombination between linear substrates. Linearized pHS20 and pHS23 were prepared by cleaving with ScaI, and the fragments were purified before use as substrates in in vitro recombination. Recombination reactions were incubated for 16 hr before analysis. Molecular weight markers (M) are as in Fig. 2.

Figure 4

φC31 integrase only catalyzes attP/B recombination. Plasmids encoding attP (pHS20 or pHS22), attB (pHS21 or pHS23), attL (pHS50), or attR (pHS52) or attL and attR together (pHS55) were used as substrates for recombination with φC31 integrase in vitro. Recombinant products were obtained only in lane 2 that contained attP and attB as substrates, and the reaction is the same as that shown in Fig. 3. If recombination had occurred, the predicted recombinant products between attP/attP (lane 3) and attB/attB (lane 4) cut with NsiI and PstI and between attB/attL (lane 7) and attB/attR (lane 8) cut with BamHI would have been 6 kbp and 0.4 kbp, respectively. For attP/attL (lane 5) and for attP/attR (lane 6) recombination, the predicted products would have been 6.4 kbp and 0.4 kbp when cut with XhoI and ApaI, respectively. For attL/attR (lane 9) on different plasmids, the predicted product would have been 3.6 kbp when cut with SphI and SmaI, and when attL–attR (lane 10) were on the same plasmid the predicted recombinant product would have been 0.4 kbp detected by restriction with KpnI. Note that pHS52 cut with SphI and SmaI (lane 9) yielded a parental band of a size (3.4 kbp) similar to the predicted recombinant product in the pHS50/pHS52 reaction, but still no recombinant could be detected in gels that had undergone electrophoresis for a longer period. Note also that the 0.4-kbp SphI-SmaI fragment in pHS50 is a parentally derived fragment. Molecular weight markers (M) are as in Fig. 2.

Figure 5

Intramolecular recombination between attP and attB. (a) Restriction maps of pHS44 and the expected recombinant products. (b) Restriction analysis of parental and recombinant products after extraction of plasmid DNA from E. coli (lanes 1–3) or after recombination assays in vitro (lanes 4–7). BamHI and ScaI were used to analyze the plasmids from E. coli (lanes 1–3) and after in vitro recombination (lanes 7 and 8). In E. coli the recombinant product containing attL was barely visible (even though attR is abundant) because it does not contain an origin of replication and is lost. To detect the attR-containing circles obtained after recombination in vitro, BamHI alone was used (lanes 4 and 5), and this cut both the parental DNA and the attL-containing product. Molecular weight markers (M) are as in Fig. 2.