Control of directionality in lambda site specific recombination

Abstract

The simple relation between the substrates and products of site-specific recombination raises questions about the control of directionality often observed in this class of DNA transactions. For bacteriophage lambda, viral integration and excision proceed by discrete pathways, and DNA substrates with the intrinsic property of recombining in only one direction can be constructed. These pathways display an asymmetric reliance on a complex array of protein binding sites, and they respond differently to changes in the concentrations of the relevant proteins. The Escherichia coli protein integration host factor (IHF) differentially affects integrative and excisive recombination, thereby influencing directionality. A four- to eightfold increase in intracellular IHF coincides with the transition from exponential to stationary phase; this provides a mechanism for growth phase-dependent regulation of recombination that makes the cellular physiology an intrinsic part of the recombination reaction.

Fig. 1

The DNA substrates, proteins, and protein binding sites in lambda site specific recombination: Circular phage DNA (attP,—), and linear bacterial DNA (attB,═) undergo integrative recombination to yield attL and attR. For attP and attB, the core-type Int sites and overlap regions (COC′ and BOB′, respectively) are indicated. For attL and attR, which can undergo excisive recombination to yield attP and attB, the top strand DNA sequence is shown (49). The protein binding sites for IHF (H), Xis (X), arm-type Int (P), and core-type Int (C or B) are numbered from left to right and marked with prime signs if present to the right of the overlap region. Although the regions underscored with symbols include all of the bases apparently important for specific protein recognition, some of the interior bases are not specified in the consensus sequences (8, 10,11,20, 50). The relative orientation of sites for a given protein is indicated by the symbols (IHF: H or H, arm-type Int: A or A core-type Int: c or ○, and Xis: ×▸). The sites of strand exchange in the top strand (↘) and omitted bottom strand (↖) demark the 7-bp overlap region (○). The resected attR sites (lower portion) are named for their outermost complete protein binding site as described in the text (Nde-B′ refers to an attR cut with Nde I prior to use in order to remove all DNA to the left of −99). The last base for which each resected attR matches the unresected att DNA is shown (↑). Most resected attR plasmids were constructed from attP plasmids by in vitro recombination. The construction of P1-P′3, P2-P′3, X2-P′3, H2-P′3, and C-P′3 has been described (18). H1-B′ was made by (i) digestion of P1-B′ with Dde I, (ii) filling in the staggered ends with the Klenow fragment of DNA polymerase, (iii) gel purification of the appropriate sized fragment on low-melting agarose, and then digestion with Bam HI. This was ligated to the large Eco RV-Bam HI fragment of pBR327. X1-B′ was made by cutting P1-B′ and pBR322 with Nde I and Bam HI. The appropriate fragments were then ligated. B-H′ (end point at +46), B-P′1 (+64), and B-P′3 (+100) were made by recombination in vitro with previously described resected attP plasmids (18). Constructions that are missing different extents of the same protein binding site are not shown separately because identical results were obtained within such sets.

Fig. 2

Excisive and integrative recombination with resected att sites. Excisive recombinations with supercoiled attR. and radioactive linear attL; and integrative recombination, with supercoiled attP and radioactive linear attB, were carried out in vitro (24). The identity of the attR and attP substrates used in the reactions is indicated at the top of each lane. attL and attB were linearized with Eco RI and labeled with [α-³²P]dATP and the Klenow fragment of DNA polymerase. Labeled DNA’s were 1.25 × 10⁻¹¹M in the 20-μl sample. Purified proteins were added to the DNA-buffer mix in the following order and concentrations (for excision and integration, respectively): IHF (2.5 nM and 10 nM), Int (62 nM and 125 nM), and Xis (150 nM and 0 nM). After 4 hours at 25°C, 10 μl of "loading" solution [1 percent SDS, 10 percent Ficoll, salmon sperm DNA (25 μg/ml)] were added to each sample and placed onto a 1.2 percent agarose gel. After electrophoresis for 16 hours at 40 V, the gels were dried and autoradiographed. The product for the reaction of H1-B′ with attL is smaller than for reactions involving the other attR DNA’s because it is on the smaller pBR327 backbone rather than pBR322.

Fig. 3

Kinetics of excisive recombination: Labeled attR substrates were recombined with supercoiled attL as described in Fig. 2, except that IHF, Int, and Xis were present at 1.25 nM, 25 nM, and 150 nM, respectively, and reactions were done at 20°C so that kinetics could be more accurately studied. Similar results are obtained at 25°C. Autoradiograms were analyzed with a Hoefer GS300 scanning densitometer attached to a Hewlett-Packard 3390A integrator. Percentages of recombination for each time point were normalized to the percentage obtained at 24 hours for each combination of att sites. The values for the efficiency at 24 hours were 39 percent (P1-B′), 41 percent (H1-B′), 43 percent (P2-B′), 48 percent (X1-B′), 25 percent (Nde-B′), and 12 percent (X2-B′).

Fig. 4

Int dependence of excision. The concentration of Int was varied in twofold increments under the recombination conditions described in Fig; 3, except that IHF and Xis were present at 5 nM and 150 nM, respectively. The abscissa is a logarithmic scale and the ordinate is the absolute percentage of each attR recombined (not the fraction of the final recombination as in Fig. 3).

Fig. 5

Xis dependence of excision. The concentration of Xis was varied in twofold increments (and plotted on a logarithmic scale) under the recombination conditions described in Fig. 3. IHF and Int were 5 nM and 60 nM, respectively.

Fig. 6

IHF dependence of excision. The concentration of IHF was varied in twofold increments (and plotted on a logarithmic scale) as described in Fig. 3. Int and Xis were 12 nM and 150 nM, respectively.

Fig. 7

Intracellular IHF as a function of growth phase. E. coli strain HN356 (9) was grown in rich medium, and optical density (OD) was measured at 650 nm. A double-headed arrow shows the last point during exponential phase growth. Cells (constant wet weight) were harvested (non-linearity of light scattering at high cell density was taken into account by diluting cells to determine the volume needed), centrifuged (SS34 rotor) at 4000 rev/min, and resuspended in 10 mM tris (pH 7.9) containing 1 mM EDTA. Cells were lysed on ice by addition of NaOH to 0.1M and vortexed for 1 minute. The cells were put on ice for 15 minutes and HCl was added to 0.2M; they were then vortexed for 1 minute. After sedimenting the cellular debris in a microfuge, the supernatant was neutralized with tris base and assayed for DNA binding activity in a solution with 5 × 10⁻¹³M labeled DNA containing IHF binding sites, unlabeled salmon sperm DNA (100 μg/ml), 100 mM NaCl, 10 mM MgCl₂, and 50 mM tris (pH 7.9). These samples were then subjected to electrophoresis on 5 percent polyacrylamide gels as described (24). The DNA binding patterns of the in vivo extracts were compared to samples containing known amounts of IHF. With this method, the yield of IHF from exponential phase cells is approximately the same as that obtained in the original IHF purification (9). The ratio between IHF present in the stationary (as compared to the exponential phase) varied between 3 and 10 in different experiments.