RecBCD is composed of the products of recB, recC and recD, which lie between the argA and thyA intergenic regions of E. coli (Figure 1a). ptr, which encodes protease III, is located between recC and recB. The recB termination codon overlaps the initiation codon of the downstream recD by one nucleotide. Therefore, recB and recD form an operon with a major external promoter for both genes and with a minor internal promoter for recD. The major promoter to the left of recB directs the transcription of the 3543 and 1824 nucleotides of the recB and recD coding sequences, respectively, whereas the weaker promoter to the left of recD appears to direct the transcription of recD alone. To generate a genetic fusion between recB and recD, the mutations were targeted to this overlap site. Two fusion forms were generated: 1 a deletion fusion mutation and 2 a substitution fusion mutation. For the former, two nucleotides (TA) of the recB termination codon at the overlap site were deleted; this removes the termination codon of recB but does not yield a frameshift mutation. For the latter, three codons encoding two asparagine and one glycine residues, which has the potential of forming equivalent hydrogen bonds and facilitates a beta turn respectively, were substituted at the same site. This substitution leads to a short bridge between recB and recD.

Recombineering is the name given to the method that produces recombinants by the Red-recombination function of lambda phage. Short linear homologous DNA strands (e.g. 20–70 bases) are required. Recombination can occur between single-stranded or double-stranded DNAs 34 by the exo, beta and gamma proteins of lambda. Single-strand pairing is formed between a synthetic oligonucleotide containing a mutation and single-stranded chromosomal DNA arising from replication fork. This requires beta, an annealing protein, and a 35-nucleotide flanking homology on each end. To increase the frequency of recombination, the oligonucleotide should correspond to the lagging strand at the replication fork and the direction of replication should be counter-clockwise 35. In this study, the deletion and substitution fusion mutations were generated by single-strand pairing. Appropriate oligonucleotides were created and synthesized, then transformed into an E. coli strain containing lambda prophage (Table 1). Chromosomal DNAs isolated from the transformed cells were then amplified with appropriate primers. Six candidates with DNAs expected to encode RecBD fusion proteins were found for both mutants by colony-screening, and two were isolated by transduction (Figure 1b). They were named V2825 and V2826.

Purified transductants were subjected to the phage spot test on LB plate with λ red^-gam^-χ⁺, λ red^-gam^-χ⁰, λ red⁺gam⁺, T4, T4 gene 2^- P1 and P2 phages (Table 2). The sizes of plaques were evaluated 17 (Table 3): 1 λ red^-gam^-χ⁰ formed small plaques, λ red^-gam^-χ⁺ formed medium plaques, and λ red⁺gam⁺ formed large plaques. In recBCD⁺ cells, λ red^-gam^-χ⁰ forms small plaques and λ red^-gam^-χ⁺forms large ones, whereas in recBC mutants, χ⁰ and χ⁺ phages forms plaques of the same size. Our result indicates that V2825 and V2826 have recombination-proficient phenotypes. 2 There were many plaques in the P1 spot. P1 lysogeny requires RecBCD recombination function, so fusion mutants are recombination-proficient 3. λ red^-gam^-χ⁺produced larger plaques than λ red^-gam^-χ⁰; this result suggests that there is a Chi hotspot activity. 4 T4 formed plaques, but T4 lacking gene 2 protein did not. Gene 2 protein protects T4 DNA ends from RecBCD exonuclease degradation. In the absence of gene 2, T4 cannot form plaques. This result indicates that the recB-recD fusion mutants have exonuclease activity. 5 Many plaques were found in the P2 spot. P2 phage growth requires the exonuclease activity of RecBCD. Our result suggests that fusion mutants have exonuclease activity.

The recombination proficiency of haploid alleles in the recB-recD fusion mutants was measured in lambda vegetative crosses 1617. Table 4 shows the proficiency of recombination by each mutant with recBCD⁺, recB null mutant and the recD mutant. Both the deletion and substitution fusion mutants gave high recombination proficiencies, similar to that of recBCD⁺ cells. recB null mutants were found to be recombination-deficient and recD mutants were recombination-proficient, as expected.

Chi activity was measured in hotspot crosses to determine whether the fusion mutants interact with Chi 1617 (Table 4). The ratio of frequency of recombination in a genetic interval with or without a Chi site gives Chi hotspot activity. Both deletion and substitution fusion mutants produced high Chi hotspot activities similar to that of wild-type cells (approximately 6.0), indicating that more exchange occurs in an interval with a Chi site than that an interval without one. recBCD null mutants and recD mutants produced no Chi hotspot activity (approximately 1.0).

The phenotypes of the recB-recD fusion mutants on plasmids were tested by spot tests 17 (Table 5). λ red^-gam^-χ⁺produced larger plaques than λ red^-gam^-χ⁰. T4 formed plaques, but T4 gene 2^- did not. P1 and P2 produced plaques. It was concluded that recB-recD fusion mutants on plasmids are recombination-proficient and they have Chi activity and exonuclease activity.

The recombination proficiency of recB-recD fusion mutants on plasmids was tested by conjugational cross 17 (Table 6). His⁺ recombinants in the plasmid derivatives of recBD2728 and recBD2729 fusion mutants increased to the wild-type value. These results show that the fusion mutants are Rec⁺.

We wanted to determine whether the construction of a recB-recD fusion leads to an active RecBD fusion polypeptide. For this purpose, we performed native and denaturing polyacrylamide gel electrophoresis of crude extracts and blotted them. RecBD fusion polypeptides were detected on SDS-PAGE probed with anti-RecB and anti-RecD (Figures 2a, b), and RecC polypeptide was detected when the gels were probed with anti-RecC (Figures 2c, d). In the native gel, less heterotrimer was found in the two fusion mutants than in purified RecBCD, but neither intact RecB polypeptide nor RecBC was detected in cell extracts of either fusion mutant (Figure 2d). Unexpectedly, a polypeptide with size similar to RecD was observed (Figure 2b).

Because the fusion peptides were found on gels and they were recombination-proficient, we wondered whether the fusion mutants have enzyme activities. We first tested exonuclease activity in them. RecBCD degrades linear DNA to acid-soluble nucleotides by two activities, dsDNA and ssDNA exonucleases, both of which are ATP-dependent. We measured only ATP-dependent dsDNA exonuclease activity in this work 36; in both fusion mutants, it was approximately half that in wild type (Table 7). This result is consistent with the qualitative assays for exonuclease activity with T4, T4 gene 2 and P2 (Table 5).

Since the fusion enzymes are recombination-proficient and have Chi genetic activity, we wanted to demonstrate DNA unwinding and Chi cutting activities, which are measured together on linear dsDNA labeled with ³²P at the 5' end of a single strand containing a Chi site 24. In the presence of excess magnesium, wild-type RecBCD binds to dsDNA ends and degrades the 3'-ended strand until Chi site, so few if any ssDNA fragments are found. After Chi, the enzyme cuts ssDNA a few nucleotides to the 3' side of Chi and forms a Chi fragment (Fig 3, lanes 5–8) [see Additional file 1, lanes 3–5]. When we tested the fusion enzymes for this activity, the deletion fusion mutant enzyme had unwinding and exonuclease activities, so it did not produce a full-length ssDNA fragment but could cut near Chi and form Chi fragments (Figure 3, lanes 9–12). Similarly, the substitution fusion mutant enzyme unwound dsDNA and degraded the 3'-end strand, so it did not form an ssDNA fragment but formed Chi fragments (Figure 3, lanes 13–16). These observations are consistent with the phage test results.

The viability of the fusion mutant cells was tested by their sensitivity to DNA damaging agents 3738. Deletion and substitution fusion mutants were just as resistant to UV radiation as the recBCD wild type strain when compared to pBR322 (Figure 4). Both fusion mutants showed similar ability to the recBCD wild type strain in reducing DNA damage induced by Mitomycin C (Figure 5), but plasmids were sensitive. These data show that recB-recD fusions do not inhibit DNA repair.

The phenotypes of the fusion enzymes were demonstrated by genetic and enzymatic assays (Table 8). The deletion fusion enzyme is recombination proficient and resistant to DNA damaging agents (Table 3, 4, 5, 6, Figures 4, 5). In contrast to RecBC, this mutant enzyme has nuclease and Chi genetic activities (Table 7, Figure 3). It can unwind dsDNA as wild type enzyme does and recognizes the Chi sequence because it forms Chi fragments (Figure 4). Similarly, we found that the substitution fusion enzyme is recombination proficient and resistant to DNA damaging agents (Table 3, 4, 5, 6, Figures 4, 5). This mutant enzyme has Chi activity, nuclease and helicase activities, and also recognizes Chi (Table 7, Figure 3). Our findings indicate that both fusion enzymes can bind to dsDNA ends and unwind double strands. During unwinding they can recognize the Chi sequence and form a 3' overhang. After interaction with Chi, wild type RecBCD enzyme loads RecA protein on to the free 3' end 24, which allows pairing and strand exchange with homologues 15, so this activity is required for homologous recombination 33. We did not measure RecA loading activity in this study, but the genetic measure of recombination proficiency and enzymatic measure of Chi recognition 32 afford evidence for such activity.

In our mutations we removed the stop codon of recB, so translation continued to the end of the recD coding sequence and produced a RecBD fusion heterodimer, which might affect the folding of the enzyme and consequently its functions. Folding is a rapid process initiated by local elements that assemble to yield the final native fold 3940. Our protein analysis indicated not only the presence of RecBD fusion polypeptide, as we expected (Figures 2a–c), but also the absence of intact RecB monomer and RecBC dimer (Figure 2d). This result is the most important evidence for a RecBD fusion heterodimer. An intact RecC that we found is essential for assembly and association into holoenzyme. However, we observed that the amount of heterotrimer was lower than in wild type. One possible explanation is that the assembly of the fusion protein to form the heterotrimer was affected because an oligomeric protein is formed by consecutive folding and association of the constituent polypeptides. A more significant finding than the reduction of heterotrimer level is RecD with wild type length. This indicates that the RecD polypeptide is expressed by its minor promoter, which may affect the assembly process. In E. coli, recB and recD form an operon with two promoters; the major promoter is external and directs the transcription of both genes, whereas the minor one is internal and appears to direct the transcription of recD. A similar promoter composition is found in other E. coli operons such as glnALG and rpsU-dnaG-rpoD; the minor promoter allows downstream genes to be transcribed under certain conditions 6. However, it is not yet clear when the weaker promoter for recD governs, so we did not take it into consideration at the outset. Nevertheless, wild type length RecD could bind to Rec(BD fusion)-RecC assembly and could make Rec(BD fusion)-RecC-RecD enzyme that might be responsible for the functionality (Figure 6).

During our study the three-dimensional structure of RecBCD was discovered. It was found that the distance between 1174 of RecB polypeptide and 2 of RecD polypeptide is about 60.8 Å (Wigley and colleagues, personal communication) [see Additional file 2], which is longer than was thought. This distance corresponds to about 17 amino acids. However, in our study, three amino acids were substituted between 1180 of RecB and 1 of RecD, so it corresponds to total 10 amino acids (about 35 Å). Therefore at least seven more amino acids may be required to fuse recB and recD. Furthermore, this long distance between C-terminal of RecB and N-terminal of RecD may cause a functional enzyme which may be assembled by Rec(BD fusion)-RecC-Rec(BD fusion), in which RecB and RecD come from two different polypeptides.

Genetic studies showed that there is little or no Chi hotspot activity in frameshift or single amino acid mutations of RecC(recC*, recC343 respectively) 171819. Therefore it was proposed that hotspot Chi is recognized by RecC as the 3' single strand DNA during unwinding 5181920. It was suggested that this recognition initiates a signal, which causes a change in RecD polypeptide to form 3'overhang to the left of Chi for RecA loading 20. Based on our observations, it can be suggested that this signaling in both RecBDC fusion enzymes would retain in response to Chi.

The results presented here suggest that RecBD fusion enzymes have wild type phenotype and fully active. However, considering the distance between C-terminal of RecB and N-terminal of RecD in the three-dimensional structure, and the presence of wild type-length RecD, unusual subunit associations such as Rec(BD fusion)-RecC-RecD and Rec(BD fusion)-RecC-Rec(BD fusion) may be formed and may be responsible for the activity. To rule out this possibility, purification of those discrete molecules and then their extensive biochemical assays are required. For this purpose, purification methods, SDS-PAGE analysis and MALDI-TOF MS can be performed 4142.

Bacterial strains and plasmids are shown in Table 1, and phages are listed in Table 2 with their genotypes and sources. recBCD⁺ means recB⁺, recC⁺ and recD⁺. Rec⁺ indicates a strain with recBCD⁺ and recA⁺.

Tryptone broth, minimal medium (OMBG), phage suspension medium (SM) and LB (Lurial bertani) agar have been described 17. BBL agar contains trypticase (BBL-microbiology systems, Maryland); BBL-YE contains 0.1% yeast extract. Terrific broth contains yeast extract, trypticase and glycerol.

Oligonucleotides were purchased from Integrated DNA Technologies Inc (IDT, Coralville, IL, USA), the chromosomal DNA isolation kit from Promega (Madison, WI, USA), the protein assay kit and non-fat dried milk from BioRad (USA), electrophoresis equipment for DNA from Ellard Instrumentation (USA) and for protein from BioRad and Invitrogen (USA); and chemicals were purchased from Sigma Chemical Company (USA) and Fisher Scientific (USA).

For recombineering, two oligonucleotides were designed. The first, containing a TA deletion at the recB stop codon, 5'-TACTCTACAAACGGCCATACTGGGACCTCCTCCGCTACTTTAACGTTTTCGTTAATGACCTTCGACACCT-3', was used for the deletion mutation. The second, containing an aac-ggt-aac substitution at the overlap site, 5'-CCTACTCTACAAACGGCCATACTGGGACCTCCTCCGCTTGCCATTGACTTTAACGTTTTCGTTAATGACCTTCGACACC-3', was used for the substitution mutation. Oligonucleotides corresponded to the lagging strand at the replication fork and contained a 35 nucleotide flanking homology on each side 3435. Recombineering was performed on an E. coli strain (HME63) lacking the mismatch repair system but containing lambda prophage. A fresh culture of HME63 at 32°C was divided into two flasks. Half the culture was not induced but was kept for future use. The other half was induced by incubating at 42°C for 15 min, then on ice for 15 min. The culture was centrifuged at 4600 g and the pellet was dispersed in 1 ml cold water; 30 ml cold water was added and the suspension was centrifuged. The pellet was again dispersed in 1 ml cold water and centrifuged, and the pellet was dispersed in 200 μl cold water. Forty-five μl of induced cells was transferred into three electroporation cuvettes. Fifteen pmol of fusion oligo and 15 pmol of gal oligo were added to the second and third cuvettes, respectively. Forty-five μl of uninduced cells was transferred to two electroporation cuvettes. Fifteen pmol of gal oligo was added to the second cuvette. Electroporation was performed at 1.8 kV. LB (1 ml) was added to each cuvette. The cells were diluted with SM and spread on LB plates.

A large chromosomal fragment (1.7 kb) containing fusion mutations was amplified. Two primers were used; the forward was relative to 1046 bp to the mutation site, and the reverse was relative to 656 bp (IDT). Transformed cells were grown overnight at 42°C. Fifty patch-grid plates were prepared from these fresh colonies. The cells were lysed at 96°C for 8 min. The chromosomal DNA was amplified in a reaction mixture containing 0.2 μM of each primer, 1.5 mM MgCl₂, 0.2 mM dNTPs and 1 U Taq polymerase (Invitrogen, San Diago). The conditions of amplification were as follows: 94°C for 15 s, 55°C for 30 s, and 72°C for 2 min (30 cycles). PCR products were digested by BsaXI for the deletion mutation, and by NlaIII for the substitution mutation (New England BioLabs). As a result, approximately 1750 colonies were screened by colony PCR. A total of 5 candidates were detected for the deletion mutation and 1 candidate for the substitution mutation. Those candidates were purified and sequenced (Big Dye Cycle Sequencing Protocol, PE Biosystems).

Mutagenesis was generated in an E. coli strain (thy⁺arg⁺recBCD⁺) containing lambda prophage. To remove lambda phage, the mutants were transferred to another strain (thy^-arg^-Δ recBCD) by P1 transduction and then thy⁺arg⁺recombinants were selected. P1-mediated transduction was performed by the plate stock method 17. A cell culture grown at 32°C overnight was mixed with P1 phage grown on the donor strain at an MOI of 0.5 and incubated at 37°C for adsorption. The cells were grown overnight, harvested and centrifuged. The supernatant was the P1 phage stock containing the mutant strain. Phages were mixed with fresh E. coli cells (thy^-arg^-Δ recBCD) at 37°C for adsorption. Sodium citrate was added to stop phage adsorption. Transductants were selected on minimal medium containing histidine and methionine, then purified. The phenotypes of the mutant candidates were tested by a phage sensitivity test on LB plate 17. Fresh cultures of fusion mutants in TB were mixed with BBL-top agar, and poured on to BBL agar plates. Seven phages (10⁴ pfu/ml) were placed on the spots. Recombination frequency and Chi genetic activity were tested by λ red^-gam^-χ⁺, λ red^-gam^-χ⁰, λ red⁺gam⁺ and P₁ phages: λ red^-gam^-χ⁺and λ red^-gam^-χ⁰ produce plaques in Rec⁺cells. The sizes of the plaques in those spots reflect Chi activity. P1 needs Rec⁺ cells to grow. Exonuclease activity was tested by T4, T4 gene 2^- and P₂ phages. Gene 2 protein binds to the ends of T4 DNA and protects T4 DNA from the exonuclease activity of RecBCD after injection into the cell. P2 requires RecBCD exonuclease activity to grow. Recombination frequency and Chi genetic activity were assayed quantitatively by hotspot crosses as described by Schultz 17. First, the stocks of lambda phages 1081, 1082, 1083 and 1084 were prepared. Cross 1 was made between parents 1081(c⁺) and 1082 (cI857), and Cross2 was made between parents 1083(c⁺) and 1084 (cI857). A fusion mutant culture (100 μl) was infected with 5 phages per cell. The mixture was incubated at 37°C for phage adsorption. The total phage count was titered on a (sup^-) strain on BBL whereas the recombinant phage count was titered on a (sup⁺) strain on BBL-YE plates. Recombination frequency (%J⁺R⁺) was determined by the percentage ratio of recombinant phage count to total phage count on the same interval for two crosses. Chi activity was determined by counting the turbid plaques on a (sup^-) strain on BBL plates and clear plaques on a (sup⁺) strain on BBL-YE plates. The square root of ratio of the turbid plaques to clear plaques in Cross 1 to Cross 2 gives the Chi activity.

A 18.3 kb BamH1 fragment containing either wild type recBCD or recB-recD fusion mutants was ligated into the BamHI site of pBR322 by T4 DNA ligase (New England BioLabs), and transformed into V2831 thy^--arg^- cells to select thy⁺-arg⁺ cells. Transformants were confirmed by sequencing. The orientation of the plasmid was tested by Pst1 digestion.

Crude cell extracts were prepared by a modified Eichler-Lehman method 36. Isolated colonies of purified transformants were grown in terrific broth. Fresh culture was grown to OD₆₅₀ >2.0, centrifuged and washed. The pellet was resuspended in 50 mM Tris pH 7.5, 10% sucrose and 1 mM EDTA. Aliquots were stored at -70°C in liquid nitrogen. Cells were thawed, their volume was measured, and 100 mM PMSF, 50 mM DTT and 500 mM EDTA were added to reach the final concentration. Lysozyme (10 mg/ml) was added to the mixture and incubated on ice for 5 min. NaCl (5 M) was added, incubated on ice and centrifuged for 30 min. The supernatant was removed and the protein concentrations of the extracts were determined (BioRad). Native gels (5% polyacrylamide, 37.5:1 acrylamide:bis in 50 mM MOPS-KOH, pH 7.0) and SDS gels (3–8% of polyacrylamide in 50 mM Tris-acetate buffer, pH 8.25, or 4–12% of polyacrylamide in 1.25 M Bis-Tris buffer, pH 6.5–6,8) were poured in gel cassettes 20. The gels were prerun for 1 h, and samples were mixed with loading solution and run at 100 V overnight at 4°C. Proteins were transferred to PVDF membranes (Immobilon-P, Millipore) and probed with mouse monoclonal antibodies specific for RecB, RecD or RecC. The antibodies were visualized with horseradish peroxidase-linked antimouse IgG and a Phototope-HRP detection kit (New England Biolabs).

The recombination frequency of the fusion mutants on plasmids was tested by Hfr crosses. The donor V1306 strain (his⁺ rpsL⁺) was mixed with the recipient recB-recD fusion mutants (hisG4 rpsL31) as described by Schultz 17. The ratio of donor to recipient was 1:10. Mating cells were incubated in a 37°C water-bath for 8 min. Diluted mating cells were grown for 20 min and were plated on selective minimal medium (OMBG) containing methionine and streptomycin to select His⁺ str^R recombinants. Diluted recipient cells (10^-5 dilutions) were plated on LB+amp. The number of His exconjugants per donor corrected by viability of the recipient strains is called the recombination frequency.

The viability of recB-recD fusion mutant cells on plasmids was determined by two qualitative assays. In both, small colony formation means low cell viability, and no growth means sensitivity to that agent. In the UV sensitivity assay, fresh overnight cultures were grown to mid-log phase and exposed to UV (10 erg/m²/s) for 5–20 s 37. Cells were incubated at 37°C in the dark overnight, and colony growth was observed on the following day. In the Mitomycin C(MC) sensitivity assay 38, plasmid alleles of the fusion mutants were streaked on LB agar plates containing 0, 0.06, 0.12, 0.25 or 0.50 μg Mitomycin C, incubated at 37°C overnight, and checked for colony growth.

DNA unwinding, Chi cutting and dsDNA exonuclease activities were determined in crude extracts of the fusion mutants. ATP-dependent dsDNA exonuclease activity was measured according to Eichler-Lehman 36. Crude extracts were mixed with [³H] T7DNA, 25 μM ATP, BSA, 500 mM Tris HCl pH 8.5, 100 mM MgCl₂ and 10 mM DDT and incubated at 37°C for 20 min. Carrier DNA (0.2 mg/ml) and 5% TCA were added and the mixture was incubated on ice and centrifuged. The supernatant, containing acid-soluble nucleotides, was counted using a Beckman Coulter Scintillation Counter. One unit of enzyme was defined an amount that solubilizes 5 nmol of dsDNA at 37°C in 20 min.

Chi cutting and DNA unwinding activities were measured together 24. The substrate, plasmid DNA χ⁺F, was made linear with HindIII (New England BioLabs), incubated with shrimp alkaline phosphatase (SAP) (Invitrogen) at 37°C for 30 min, and labeled with γ³²P at the 5'ends by T4 polynucleotide kinase (Invitrogen) at 25°C for 30 min. Free nucleotides were removed with an SR-200 mini column. DNA substrate (4.5 nM) was mixed with the indicated amount of cell-free extract in 20 mM MOPS pH 7.5, 3.5 mM MgCl₂, 5 mM ATP, 1 mM DTT, and 1.6 μM SSB (Promega). After 2 min, ATP-γS and M13 DNA were added. Reaction products were separated on a 1% agarose gel and analyzed by autoradiography.