|Home | About | Journals | Submit | Contact Us | Français|
Supercoiled DNA is the relevant substrate for a large number of DNA transactions and has additionally been found to be a favorable form for delivering DNA and protein-DNA complexes to cells. We report here a facile method for stoichiometrically incorporating several different modifications at multiple, specific, and widely spaced sites in supercoiled DNA. The method is based upon generating an appropriately gapped circular DNA, starting from single-strand circular DNA from two phagemids with oppositely oriented origins of replication. The gapped circular DNA is annealed with labeled and unlabeled synthetic oligonucleotides to make a multiply nicked circle, which is covalently sealed and supercoiled. The method is efficient, robust and can be readily scaled up to produce large quantities of labeled supercoiled DNA for biochemical and structural studies. We have applied this method to generate dye-labeled supercoiled DNA with heteroduplex bubbles for a Förster resonance energy transfer (FRET) analysis of supercoiled Holliday junction intermediates in the λ integrative recombination reaction. We found that a higher-order structure revealed by FRET in the supercoiled Holliday junction intermediate is preserved in the linear recombination product. We suggest that in addition to studies on recombination complexes, these methods will be generally useful in other reactions and systems involving supercoiled DNA.
Supercoiled DNA is the relevant substrate for a large number of DNA reactions, including replication initiation and elongation, transcription initiation and elongation, recombination, repair, nucleosome assembly and dynamics, regulation of gene expression, long-range enhancer effects, DNA compaction, localized melting and cruciform extrusion (1). It has additionally been found to be a favorable form for delivering DNA and protein–DNA complexes to cells (2). We report here a method that should greatly facilitate analysis of pathways involving DNA with this topology. It is a relatively facile method for stoichiometrically incorporating several different DNA modifications (we are using fluorescent dye labels and bubbles containing mismatched DNA pairs) at multiple specific sites in supercoiled DNA. The method is based upon generating an appropriately gapped circular DNA that is annealed with labeled and unlabeled synthetic oligonucleotides to make a multiply nicked circle. The circular DNA is covalently sealed and supercoiled, either by ligase in the presence of ethidium bromide, or by incubation with DNA gyrase. The method enables the stoichiometric placement of many different DNA modifications at widely spaced specific sites; it is efficient, robust and can readily be scaled up.
Our interest in developing this method grew out of a long-standing desire to study the mechanisms and structures responsible for bacteriophage λ integrative recombination—a reaction that requires supercoiled DNA substrate. Validation of the method, and several examples of its application, will be based upon this recombination reaction.
When the temperate bacteriophage λ infects an Escherichia coli cell under conditions unfavorable for bacterial growth or phage production, it initiates a lysogenic response that represses most of the viral genome and promotes insertion of its circular chromosome into the E. coli chromosome via an integrative recombination reaction. This occurs between DNA ‘attachment’ sites called attP and attB on the viral and bacterial chromosomes, respectively. As a result, the integrated λ DNA is flanked by hybrid attachments sites, called attL and attR. In response to an improvement in the physiological state of the bacterial host, or a threat to the host’s survival, λ phage DNA is excised from the chromosome by an excisive recombination between attL and attR that regenerates attP and attB on their respective circular chromosomes (3). Despite the X-ray crystal structures, genetics and biochemistry describing the individual proteins and their interactions with DNA, we still lack a comprehensive view of the λ site-specific recombination pathway. The situation is particularly unsatisfying for the integrative reaction, which has been difficult to study because it requires a supercoiled attP DNA substrate (4).
The integrative and excisive recombination reactions are carried out by the phage-encoded λ integrase protein (Int) by a mechanism that is common to a large family of tyrosine recombinases with diverse biological functions such as chromosome segregation, chromosome copy number control, gene expression, conjugative transposition, gene dissemination and viral integration and excision. The hallmark of this large family is a multistep reaction pathway in which a recombinase tetramer executes a sequential pair of isoenergetic strand exchanges at four ‘core-type’ Int binding sites, that first generate and then resolve a four-way DNA junction (Holliday intermediate) (5–8).
The λ reaction differs from that of other family members, such as Cre and Flp, in having four different recombination sites, an additional DNA-binding domain on Int, and a complex dependency upon accessory proteins that confers directionality and regulation. λ Int and its cousins comprise a subgroup of heterobivalent recombinases that bind and bridge two different families of DNA sequences (9–12). The small amino-terminal domain of Int (N-domain) binds with high affinity to five ‘arm-type’ binding sites distributed throughout the att sites and distant from the sites of strand exchange (the core-type sites). Int bridging between the arm and core sites is assisted by DNA bends induced by accessory proteins, the host-encoded IHF and the λ-encoded Xis, bound to specific sites located between the arm and core sites (Figure 1). The integration and excision reactions utilize two overlapping subsets of 16 protein binding sites to form the integrative and excisive complexes (‘intasomes’) that confer directionality and regulation on the recombination reaction (4). The reactions are further distinguished from each other by the unique requirement for supercoiling in attP DNA.
Although we shall focus on integrative recombination, the method described here for generating labeled supercoiled DNA is very general, and should be readily applicable to studying other questions in different systems.
Int and IHF proteins were purified as described previously (13,14). Oligonucleotides were synthesized and polyacrylamide gel electrophoresis (PAGE) or high-performance liquid chromatography (HPLC) purified by Eurofins MWG Operon. Sequences of the oligonucleotides are listed in the Supplementary Tables S1, S2, S4 and S6 in the Supplemental Data. The donor dye, fluorescein and the acceptor dye, tetramethyl rhodamine (TAMRA), were attached to C5 of thymine via a 6C linker and incorporated during oligonucleotide synthesis as their respective phosphoramidites.
The dye labeling efficiency of the labeled oligonucleotides was determined from the CE and HPLC data provided by Eurofins MWG Operon. We also analyzed each fluorescently labeled oligonucleotide by 7M urea 20% PAGE to determine the percentage of fluorescent dye labeling and the overall purity of the oligonucleotides. Approximately 10pmol of each oligonucleotide was loaded on the gel mixed with 2X formamide loading buffer (98% formamide; 1mM EDTA pH 8; 0.0025% Xylene cyanol). Gels were run for 5h at 25V/cm, and then stained with SyBr gold nucleic acid gel stain (Invitrogen). Gels were scanned on Typhoon laser scanner (Amersham biosciences) and analyzed by Multi Gauge software (Fujifilm). Unmodified oligonucleotides were used as controls, which run slightly faster in the gel than the oligonucleotides modified with fluorescent dyes. We did not detect any unmodified oligonucleotide in these gels. The labeling efficiency, as determined by each of the methods, was >99%.
Single-strand phage were prepared by infecting cells carrying the appropriate pBluescript phagemid with the helper phage VCSM13, according to the protocol provided by Stratagene and scaling up to 1–3l volumes. Following extraction of the DNA with phenol–chloroform and precipitation with ethanol, the DNA was further purified by passage through a Promega Wizard SV Gel and polymerase chain reaction (PCR) Clean up column. The final yield of single-strand circular phage DNA was ~1–2mg per liter of growth medium.
The attP DNA sequence from plasmid pWR1 (15) was copied by PCR using primers #1 and 2 (Supplementary Table S1), and cloned between the KpnI and SacI sites of phagemids pBluescript SK (+) and (–) (Stratagene). By incorporating the appropriate sequences into the 5′ adapter sequence of primer #1, an Nb.BbvCI site (CCTCAGC) was included at the KpnI junction of the product clones. The sequences of all of the primers (Supplementary Tables S1 and S2) and details of this and other cloning steps are provided in Supplementary Data. The phagemids pMM12 and pMM30 were derived from pBluescript SK (+) and (–), respectively.
Phagemid pMM13 was derived from pMM12 by changing the two T:A base pairs at the left end of the overlap region, at positions −2 and −1 (Figure 4), to A:T, using the QuikChange® II Site-Directed Mutagenesis Kit (Stratagene) and primers #3 and 4. Phagemid pMM32, for the production of gapped attP top-strand, was made by PCR by using primers #5 and 6, with pMM30 as a template, as described in the ‘Results’ section and Supplementary Data.
A family tree for the mutated attB phagemids used to trap the Holliday junction intermediate complex is shown in Supplementary Data (Supplementary Table S3). It starts with phagemids pMM50 and pMM60, which are derived from the pBluescript SK (+) and (–), respectively. These phagemids contain attB inserts with a mutated overlap sequence: the TT bases at positions −2 and −1 are replaced by AA, the T at position 2 is replaced by C, and the C at position 4 is replaced by A (Figure 4). Construction of the inserts and the introduction of additional restriction sites are described in Supplementary Data. All mutagenesis steps were performed with QuikChange® II Site-Directed Mutagenesis Kit (Stratagene).
Circular single-strand (SS) DNA from phagemid pMM32 was annealed with a five-fold excess of a 45 base oligonucleotide (#17, Supplementary Table S4) complementary to the PstI–HindIII sites by heating to 80°C for 7min in 10mM Tris (pH 8) and 50mM NaCl, and cooling to 4°C, in steps of 0.2°C/20s from 80°C to 40°C, and 0.2°C/10s from 40°C to 4°C in a gradient thermocycler (Eppendorf). This was digested with HindIII and PstI (NEB) (2U of enzymes per 1μg of SS DNA) at 37°C for 2h. A mixture containing this linearized DNA, the complementary circular DNA and a 5-fold molar excess of each of the three oligonucleotides was annealed in 75mM NaCl as described above. The nicks were sealed by incubation with T4 DNA ligase (NEB) (4U of enzyme per 1μg of DNA) at 16°C for 1h. Following incubation with E. coli DNA Gyrase (TopoGEN, Inc.; 0.5U of enzyme per 1μg of total DNA) for 2h at 37°C, the reaction was extracted three times with an equal volume of phenol/chloroform/iso-amyl alcohol (25:24:1), twice with chloroform/iso-amyl alcohol (24:1), and the covalently closed supercoiled molecules were purified by electrophoresis overnight at 2V/cm on 1% agarose gels. The supercoiled attP DNA was cut from the gel and purified with a Wizard® SV Gel and PCR Clean up system (Promega).
The heteroduplex bubble attP substrate was made using the same general protocol with the following DNAs: single-strand DNA from pMM13, single-strand DNA from pMM32 cleaved with PstI and HindIII, and three oligonucleotides; #18, 19 and 20 (Supplementary Table S4) for donor only substrate, or with oligonucleotides #18, 19 and 21, for donor plus acceptor substrate. AttP substrates for the wild-type recombination were prepared with same material, except that we used single-strand from pMM12, instead of pMM13 (Supplementary Table S5).
To prepare the long attB substrate for HJ trapping, circular single-strand DNA from pMM66 was annealed with the 30 base oligonucleotide (#22, Supplementary Table S6), complementary to the region around the NheI site at position 2711, and digested with NheI (2U of enzyme per 1μg of DNA, at 37°C for 2h). The linearized single-strand DNA was annealed to circular single-strand DNA from pMM55, and the annealed product was linearized with HindIII to produce a nicked linear 2.9kb attB. Phenol extraction, gel electrophoresis, and purification of the eluted linear attB were carried out as described above. The attB was dephosphorylated with Antarctic phosphatase (NEB) and labeled with 32P at the 5′ termini using T4 polynucleotide kinase (NEB).
Short attB substrate for wild-type recombination was prepared by annealing top-strand oligonucleotide #23 (Supplementary Table S6), which was 5′ end-labeled with 32P using T4 polynucleotide kinase (NEB), to the bottom-strand oligonucleotide #24 (Supplementary Table S6). Oligonucleotides #23 and 24 were annealed in the 1:1 ratio in 10mM TRIS (pH 8) and 50mM NaCl. The mixture was put in the boiling water bath, and cooled to room temperature overnight.
Integrative recombination reactions were performed by incubating supercoiled attP and radioactively labeled attB with 60nM Int and 60nM IHF in 25mM Tris, pH 8, 5mM EDTA, pH 8, 0.5mg/ml bovine serum albumin (BSA), 2.5mM DTT, 75mM NaCl and 6mM spermidine for 1h on 25°C. Wild-type reactions were resolved by electrophoresis in 1% agarose/0.4% Synergel (Diversified Biotech), while Holliday junction intermediate trapping reactions were resolved in a 1% agarose gel, at 5V/cm for 3–3.5h, respectively. Reactions designed to generate attL and attR recombinant products contained 20nM supercoiled attP and 40nM short attB substrate. Reactions designed to generate Holliday junction intermediates contained 20nM supercoiled attP bubble substrate and 20nM radioactively labeled attB bubble/heterology substrate, as described in the text.
Förster resonance energy transfer (FRET) was determined by measuring the extent of donor fluorescence quenching in a donor plus acceptor reaction relative to the fluorescence in a donor only reaction. In the former, attP substrates were labeled with fluorescein at position P+18, near the C′ core site, and with TAMRA at position P+63, between the P′1 and P′2 sites. In the latter, attP substrates were labeled only with fluorescein at position P+18 (see Figure 1).
Recombination and Holliday junction reactions (50μl each) were prepared according to the protocols for attP by attB recombination described above and incubated at 25°C for 1h. One-sixth volume of loading buffer was added to each reaction (15% ficoll in 6X TAE buffer) and either four (7, 9.5, 12.5 and 16μl) or five (4, 7, 10, 13 and 16μl) aliquots of each reaction was electrophoresed for 3–3.5h at 5V/cm on a 1% agarose/0.4% Synergel gel for wild type recombination or a 1% agarose gel for Holliday junction intermediates. The gels were scanned on a Typhoon laser scanner (Amersham Biosciences) with the blue laser (488nm), and a 520-nm emission filter to measure the fluorescence of the fluorescein in complexes with and without TAMRA. The amount of the recombinant product or HJ complex was determined after drying the gel, exposing it on an imaging plate (Fujifilm), and scanning the imaging plate using BAS 2500 system (Fujifilm). The fluorescence intensity and the amount of 32P were determined for each band using Multigauge software (Fujifilm).
The efficiency of energy transfer was determined by measuring the extent of donor quenching in the HJ complexes with both donor and acceptor dyes, as previously described (16), based on following equation:
where is the fluorescence intensity of the donor in the Holliday junction band j, labeled with donor (D) only; is the fluorescence intensity of the Holliday junction band i, labeled with both D and acceptor (A) dyes; and are the total amounts of the Holliday junction complex labeled with D only and with A plus D, respectively, as determined from the phosphorimager scan of the gel; r is the efficiency of DNA labeling with acceptor; and E is the efficiency of energy transfer.
The fluorescence intensities of the donor plus acceptor complexes () are corrected for the differences in the total amounts of the complex, between donor-only and donor plus acceptor complexes, using the ratio of , determined from the radioactivity scans. An average value for () was determined as described in Supplementary Data. The standard deviation for each point in this averaging process was less than ±0.05. All of the values in the plot were normalized to 1 by dividing by the highest donor intensity of the donor-only complex bands. The line was constructed from data collected in three or four independent experiments. In a plot of () (y-axis), versus the fluorescence intensity of the donor only complex, (x-axis), the slope of the line equals . As described above and in the ‘Results’ section, we have determined that the oligonucleotides are substituted with TAMRA to >99%. As a complete complement of oligonucleotides is required to obtain supercoiled DNA (data not shown), r=1 in all of the TAMRA-labeled supercoiled DNA substrates.
Phagemids, in addition to their own plasmid origin of replication, contain a phage origin of replication, such that infection of a host cell by an appropriate helper phage leads to production of phage particles containing a circular DNA corresponding to one strand of the phagemid (17,18). The choice of which DNA strand is replicated and packaged is determined by the orientation of the viral origin of replication within the phagemid (19,20). We used pBluescript SK±phagemids (Stratagene), which contain a 454 bp intergenic region from filamentous phage f1 (M13 related) that includes the 307-bp phage origin of replication oriented in one direction (+) or the other (–) (21). Our procedure starts by cloning the DNA region of interest in two phagemids that differ only in the orientation of their respective origins of phage DNA replication (see Figure 2A for schematic summary). One of them, phagemid A(+), is used directly to make circular single-strand phage DNA that will comprise the ‘template’ strand of a gapped circular DNA molecule.
DNA from the other phagemid, A(–), is used to direct a series of PCR reactions leading to a family of deletion phagemids, that will ultimately give rise to the ‘non-template’ strands, which will flank gaps of different sizes in the region of interest. A PCR primer that will ultimately define the right side of a gap has a 5′ adapter sequence (not complementary to the template) encoding a BamHI site followed by a PstI site and then DNA sequence complementary to the template on the right side of the gap (primer #5; Supplementary Table S1). A primer that will ultimately define the left side of a gap has a 5′ adapter sequence encoding a BamHI site followed by a HindIII site and then DNA sequence complementary to the template on the left side of the gap (primer #6; Supplementary Table S1). The linear products of PCR amplification are cut with BamHI, circularized by ligating the annealed overhangs, and used to generate the B(–) family of phagemids.
Circular single-strand DNA made from the B(–) family of phagemids is linearized by annealing it with a 45 base oligonucleotide complementary to the HindIII–PstI region and cleaving with those two enzymes. The resulting linear single-strand DNA is complementary to the circular single-strand template DNA made from Phagemid A(+), except for the DNA sequence that was deleted in the PCR reactions. Using HindIII and PstI sites enables placement of the gap boundaries between any A and any G in the region of interest. Different 5′-and 3′-overhang restriction sites can be used to obtain bases other than A or G at the gap boundaries.
The gap resulting from annealing the linear deletion strand and the full-length circular strand is filled in by annealing with the appropriate labeled and unlabeled synthetic oligonucleotides (38–51 bases long) to generate a nicked circular molecule that is then sealed by DNA ligase. Negative supercoiling can be introduced either by the presence of ethidium bromide during ligation or by incubation with DNA gyrase (Figure 2). The final supercoiled molecule will have a 454-nt bubble corresponding to the identical origins of replication.
We cloned a 371-bp DNA fragment containing attP and some flanking sequences [from pWR1 (15)] between the KpnI and SacI sites of pBluescript II SK (+) and pBluescript II SK (–), creating phagemids pMM12 and pMM30, respectively. DNA derived from pMM12 phage particles [A(+) DNA in Figure 2] was used directly for making one strand of attP DNA, referred to as the ‘bottom’ strand, in conformance with previous nomenclature for the att site DNAs (22). DNA derived from pMM30 phagemid [A(–) DNA in Figure 2] was used as the template for making a linear PCR product deleted for a portion of the attP region. In these experiments, the primer specifying the left side of the deletion had 3′-terminal sequences complementary to positions −26 to −38 of the attP region. The primer specifying the right side of the deletion had 3′ terminal sequences complementary to sequences immediately adjacent to the right side of the attP region. Both primers had 5′-terminal adapter sequences, as described above and illustrated in Figure 2A. The resulting linear PCR fragment was circularized and used to generate phagemid pMM32. This will be the source of the top strand [B(–) in Figure 2], which, when annealed to the bottom strand derived from pMM12, will produce a top-strand deletion of 136 bases (extending downstream from the H1 site in the P arm and spanning the entire P′ arm) (Figure 1).
To construct the supercoiled attP, the circular SS DNA from pMM32 phage was linearized by digestion with PstI and HindIII (after annealing with a complementary 45 base synthetic oligonucleotide) and annealed with the circular SS pMM12 template strand to generate a 136- base gap. The gap was filled in by annealing with a 5-fold excess of three complementary synthetic oligonucleotides, 51, 38 and 47 bases in length. Following incubation with ligase and DNA gyrase the supercoiled attP was purified by gel electrophoresis.
In anticipation of the FRET experiments described below, one of the synthetic oligonucleotides used to fill in the 136-base gap contained a donor dye (dT-fluorescein) at position +18 and another oligonucleotide contained an acceptor dye (dT-TAMRA) at position +63 (see ‘Materials and Methods’ section). The dye-labeling efficiency of the fluorescently labeled oligonucleotides was determined to be >98–99% as determined by three different methods: capillary electrophoresis and HPLC data provided by the manufacturer (Eurofins MWG Operon); and by the ratio of labeled to unlabeled oligonucleotides separated by gel electrophoresis in 20% polyacrylamide, 7M urea. The labeling efficiency of acceptor dye-labeled oligonucleotides was also checked by determining the ratio of the absorbances of the dye and the oligonucleotide, which also showed that dT-TAMRA labeling efficiency is >98–99% (see ‘Materials and Methods’ section).
The yield of covalently closed circular attP is ~60% (Figure 2B, lane 5). We have also made molecules in which the top strand deletion comprises the entire attP region: when this 263 base gap is filled in by annealing with 5nt the yield of closed circular attP is approximately 40% (data not shown). The conversion of covalently closed circles to supercoiled molecules by incubation with E. coli DNA gyrase is very efficient (Figure 2B, lane 7), while the final yield of the supercoiled molecules, from preparative scale agarose gels, is ~15–20%. At the scale of these experiments, we routinely recover at least 10μg of labeled supercoiled attP DNA. Purification and recovery of large quantities of the supercoiled product by CsCl2 density gradient centrifugation should be much more efficient than gel electrophoresis, making this method very easy to scale up. The supercoiled dye-labeled attP was tested for recombination with a 32P-labeled 60-bp oligonucleotide encoding the attB partner, and was found to be comparable to supercoiled plasmid DNA in kinetics (data not shown) and dependence upon Int concentration (Figure 3).
The Holliday recombination intermediate is formed by a DNA strand swap on the left side of a seven base pair ‘overlap’ region that must be identical between the recombination partners. It is resolved to fully duplex recombination products by a DNA strand swap on the right side of the overlap region (5,23–25). Thus, in the 7-bp overlap region of the recombination products, each DNA strand originates from a different partner (Figure 4). A sequence difference between the overlap regions of two partners blocks recombination (24). On the other hand, partners carrying unpaired bubbles within their overlap region will recombine as long as the swapped strands recreate a paired duplex in the recombinant products (26). Reversal of this top-strand exchange would result in reformation of the unpaired bases and is thus not favored. We have combined these two observations to efficiently trap Holliday recombination intermediates (Figure 4) [see also ref (27)]. The appropriate unpaired bubbles on the left side of the attP and attB overlap region promote, and prevent reversal of, the first (top) strand swap. A sequence difference between the partners on the right side of the overlap region prevents execution of the second (bottom) strand swap and thus prevents resolution of the Holliday junction (23,28).
To generate the phagemid for producing the bottom strand of a heteroduplex bubble attP, we changed the two T:A base pairs at the left boundary of the overlap region in pMM12, at positions −2 and −1, to A:T base pairs, to create pMM13. Phage particles from pMM13 yield the bottom strand of attP with an overlap sequence of 3′-TTATATG-5′ (instead of the wild-type sequence 3′-AAATATG-5′). The heteroduplex bubble in attP is generated during the annealing with top-strand synthetic oligonucleotides, because the oligonucleotide complementary to the overlap region has the wild-type top-strand sequence, 5′-TTTATAC-3′. Additionally, some of the top-strand synthetic oligonucleotides contained a dT-fluorescein and/or a dT-TAMRA fluorescent dye at positions +18 and +63, respectively, as described above (Figure 1).
Preliminary experiments indicated that during agarose gel electrophoresis good separation of a Holliday junction intermediate from other reaction components required a long attB partner. We therefore cloned the 42bp of attB sequence and flanking DNA between the KpnI and SacI sites of a pair of pBluescript II SK (+) and (–) phagemid vectors to generate pMM50 and pMM60, respectively. We introduced additional restriction sites for NheI and HindIII in those two vectors, leading to the creation of pMM55 and pMM66 phagemids (see ‘Materials and Methods’ section, and Supplementary Data). Annealing the single strands from pMM66 and pMM55 generates the AA left-side 2-bp bubble in attB (complementary to the TT bubble generated in attP). Additionally, the sequence on the right side of the attB overlap region, while perfectly paired, differs from the attP overlap at positions 2 and 4 (Figure 4).
To make the attB recombination partner, we linearized circular SS DNA from pMM66 by digesting with NheI (after annealing a 30-base oligo complementary to the target NheI site at position 2711 of the AttB plasmid sequence). This linear DNA was then annealed with circular SS DNA from pMM55 and the resulting nicked circle was linearized by digestion with HindIII and purified by agarose gel electrophoresis.
Recombination between the supercoiled attP and linear attB partners described above does not yield the linear recombination product seen with wild-type substrates. Rather, we observe a new product with properties of the anticipated supercoiled Holliday junction intermediate. The fluorescent dye from the attP partner and the 32P label from the attB partner are both present in the new product (Figure 5, lane 3). When the new product is recovered from the native agarose gel, deproteinized and re-electrophoresed, no linear recombination product is observed, and if it is digested with the restriction enzyme KpnI, it does not generate the fragments expected from a linear recombination product (data not shown). If a 60-bp attB is used in place of the large attB, the 32P label from the attB that was incorporated into the new product has the same mobility as supercoiled attP, as would be expected for a supercoiled Holliday junction with two very short linear arms (data not shown). Finally, when the putative Holliday junction product is digested with the nicking enzyme Nb.BbvCI (which has a single target in the attP DNA), the mobility is reduced, as expected for the conversion from a supercoiled to a relaxed circular molecule (Figure 5, lane 4). We conclude that the substrates designed to trap supercoiled Holliday junction intermediates of integrative recombination are working as predicted.
Current views of the λ site-specific recombination pathway predict the collaboration of accessory DNA bending proteins and the bivalent DNA binding of λ integrase to bring some of the distal arm-type sites into close proximity with the core-type Int binding sites (where strand exchange takes place) (4). However, we have only limited notions of which higher-order structures form, change, or dissolve during the course of a recombination reaction. The ability to position multiple DNA modifications at different loci within supercoiled DNA opens the door on several experimental approaches to addressing these questions for the integrative pathway. In the following experiments, we have used FRET to probe and compare an element of higher-order structure in the supercoiled integrative Holliday junction and the linear recombination product.
In the absence of any higher-order structure, the linear distance in base pairs between the fluorescein donor dye at position +18 (adjacent to the C′ core-type site) and the acceptor TAMRA dye at position +63 (between the P′1 and P′2 arm-type sites) greatly exceeds the maximum distance (~100Å) over which energy can be transferred by FRET. The long linear attB partner for trapping Holliday junctions was constructed and labeled with 32P at the 5′ termini, as described above and in ‘Materials and Methods’ section. In addition to this ‘donor plus acceptor’ reaction, a parallel ‘donor only’ reaction was carried out under the same conditions with a heteroduplex bubble attP containing only the fluorescein donor dye at position +18.
Following incubation with Int and IHF, different amounts of each reaction were loaded in adjacent lanes and fractionated by native agarose gel electrophoresis. The gels were scanned in a Typhoon laser scanner with the Blue laser (488nm) and a 520-nm emission filter. The amount of Holliday junction complex in each lane was quantified from a Phosphor-Imager scan of the 32P-labeled attB. The efficiency of energy transfer () was determined from the extent of quenching of the donor fluorescence (520nm) in the complexes labeled with donor plus acceptor compared with the complexes labeled with donor only. The comparison was facilitated by using the same 32P-labeled attB to form both sets of Holliday junction complexes, thus enabling normalization of the fluorescence of each band to the amount of complex in that band.
A typical gel scanned for donor fluorescence (left panel) and radioactivity (right panel) is shown in Figure 6A. The quantified data from four experiments was combined to generate the plot in Figure 6C, left panel, as described in ‘Materials and Methods’ section. From the slope of the line, which equals , the average FRET efficiency from these four experiments is 23%.
We then asked whether the higher-order structure reflected by this FRET in the supercoiled Holliday junction complex was lost or retained during the progression to a linear recombinant product. We constructed a supercoiled attP substrate without the heteroduplex bubble but labeled with the donor and acceptor dyes in the same positions as the above experiment. This was recombined with a duplex linear attB that had the same overlap region sequence as the attP partner, thus leading to a complete attL plus attR linear recombinant product. These reactions were fractionated by native agarose gel electrophoresis (Figure 6B) and the average FRET efficiency in the recombinant product, determined from three experiments, as described above, was found to be 18% (Figure 6C, right panel).
The close similarity between the FRET efficiencies of the two dye pairs, identically positioned in the supercoiled Holliday junction and the linear recombinant product strongly suggests that the portion of the higher-order structure identified by this FRET is preserved in both complexes.
The majority of previous approaches to incorporating labels into covalently closed circular DNA can be grouped into two categories. The first group is based on the initial approach of Zoller and Smith (29) using a short synthetic oligonucleotide, complementary to the circular single-strand DNA, as a primer for DNA polymerase. A number of modifications and enhancements, such as using different polymerases, the addition of different ensembles of accessory proteins, and different reaction conditions, have been devised to reduce the limitations of low yields and/or displacement of the labeled oligonucleotide primer by polymerase elongating around the circle (30,31). The second group consists of various methods for constructing gapped molecules, which were then filled in with a single oligonucleotide. Circular ssDNA was used as the source of one strand and an appropriate duplex DNA (made either by restriction endonuclease digestion or by PCR) as a source of the second strand (32).
The method we have described here for incorporating a large number of synthetic oligonucleotides, any or all of which can carry a variety of different labels and/or DNA modifications, is well suited for analyses involving FRET. It is likely to be advantageous in many other types of analyses, as well. In the experiments reported here, we used three oligonucleotides to fill in a gap of 136bp but we have also made supercoiled attP substrates using five oligonucleotides to fill in a 263-bp gap with acceptable overall yields. In those cases where there are no constraints on the DNA sequences at one or both boundaries of the gap, the method can be simplified by using only one restriction enzyme to cut the ssDNA.
One of the features of the gapped molecule approach that is essential for our studies is the assurance that substrates are homogeneous and completely labeled. Our method is also well suited for experiments in which multiple labels must be positioned and/or repositioned throughout the course of a study. This method can be readily scaled up to produce large quantities of labeled supercoiled DNA for biochemical and structural studies, in which case it might be advantageous to replace the final gel electrophoresis purification of supercoiled substrate with CsCl2 density gradient centrifugation.
The FRET experiments reported here indicate the feasibility of this approach to studying the integrative recombination reaction, and, by extension, many other reactions in a wide range of systems that depend upon or utilize supercoiled DNA. The specific placement of labels and DNA modifications in supercoiled substrates should be particularly helpful in studies on the initiation and elongation steps of both replication and transcription, nucleosome positioning and many pathways of recombination (33–36). Biophysical studies of DNA, including the internal dynamics of supercoiled DNA are also likely beneficiaries of such substrates (37).
A potentially practical application of site-specifically modified supercoiled DNA is in the area of non-viral gene delivery into eukaryotic cells and non-viral gene therapy (2). Among the modifications used to improve gene delivery to the nucleus are DNA–peptide conjugates, called nuclear localization sequences (NLS) (38). However, most of the methods used thus far do not allow controlled addition of NLS peptide molecules to supercoiled DNA (39). This could be important, since a single NLS attached to a DNA molecule is sufficient for nuclear entry, while several peptides might inhibit entry (40). Additionally, too many chemical modifications of the delivery DNA (and/or in the wrong place) could inhibit transcription (41). [For a recent review see ref. (42).]
According to the Förster theory (43), the efficiency of resonance energy transfer is related to the distance between the donor and acceptor, , by the equation: , where is the distance between the donor and acceptor that gives 50% FRET (the Förster distance). There are many factors complicating accurate determination of (44); however, in the experiments reported here, the donor and acceptor dyes are in the same immediate environment (with respect to DNA sequence and protein neighbors) such that, whatever the precise value of , it should be the same in both the Holliday junction and the recombinant product. In other words, the ratio of the two FRET energies will be a measure of the ratio of the physical distances: . The values (Figure 6 and text above) for (from the Holliday junction complex), and (from the recombinant product complex), are 0.23 and 0.18, respectively, yielding 0.95 for the approximate ratio of the distances observed in the Holliday junction and product complexes.
These data strongly indicate that the portion of the higher-order structure revealed by FRET in the supercoiled Holliday junction intermediate is preserved in the linear recombination product. This had been an open question and the results raise additional questions about how many other features of the Holliday junction intermediate are preserved in the products and which of these preserved and/or altered features contributes to the directionality and regulation of the recombination reaction. Fortunately, the experiments reported here also open the door to answering many of the questions that have been raised in regard to the integrative reaction. It is not unreasonable to expect these methods, which are efficient, robust and readily scaled up, will also be useful in other reactions and systems involving supercoiled DNA.
Supplementary Data are available at NAR Online.
Funding for open access charge: The National Institutes of Health (grant numbers GM33928 and GM62723 to A.L.).
Conflict of interest statement. None declared.
We dedicate this paper to the memory of Jo-Anne Nelson, a generous colleague and a true friend. We thank past members of the Landy laboratory, Marta Radman-Livaja, Xingmin Sun, Jeffrey Mumm, and Mohammad Rashel, and current lab members Gurunathan Laxmikanthan and Wenjun Tong for helpful advice at various stages of this project. We thank Christine Lank and Jo-Anne Nelson (deceased) for expert technical assistance and Joan Boyles for help in manuscript preparation.