High-Frequency Intermolecular Homologous Recombination during Herpes Simplex Virus-Mediated Plasmid DNA Replication

Abstract

Homologous recombination is a prominent feature of herpes simplex virus (HSV) type 1 DNA replication. This has been demonstrated and traditionally studied in experimental settings where repeated sequences are present or are being introduced into a single molecule for subsequent genome isomerization. In the present study, we have designed a pair of unique HSV amplicon plasmids to examine in detail intermolecular homologous recombination (IM-HR) between these amplicon plasmids during HSV-mediated DNA replication. Our data show that IM-HR occurred at a very high frequency: up to 60% of the amplicon concatemers retrieved from virion particles underwent intermolecular homologous recombination. Such a high frequency of IM-HR required that both plasmids be replicated by HSV-mediated replication, as IM-HR events were not detected when either one or both plasmids were replicated by simian virus 40-mediated DNA replication, even with the presence of HSV infection. In addition, the majority of the homologous recombination events resulted in sequence replacement or targeted gene repair, while the minority resulted in sequence insertion. These findings imply that frequent intermolecular homologous recombination may contribute directly to HSV genome isomerization. In addition, HSV-mediated amplicon replication may be an attractive model for studying intermolecular homologous recombination mechanisms in general in a mammalian system. In this regard, the knowledge obtained from such a study may facilitate the development of better strategies for targeted gene correction for gene therapy purposes.

The genome of herpes simplex virus type 1 (HSV-1) is a 152-kb double-stranded linear DNA molecule. It consists of two covalently linked segments, designated long (L) and short (S). Each segment consists of unique sequences that are bracketed by inverted repeats (26). During HSV-1 infection, the linear viral genome becomes circularized after it reaches the nucleus (11, 21) and then serves as a template for DNA replication. The majority of the replicative intermediates are long concatemers that are thought to have been synthesized by a rolling-circle mechanism (2, 18, 19). The essential cis elements for viral DNA replication and encapsidation include the origin of replication and the packaging signal that resides in the a sequences (24). There are three internal origins of replication along the viral genome, one located in the long segment (OriL) and a diploid origin (OriS) in the repeated region bracketing the short segment (24).

The HSV amplicon, which contains these cis elements (i.e., the OriS replication origin and the a sequence), appears to be replicated in the same way as the viral genome when they are cointroduced into cells with an infectious HSV as a helper virus and can subsequently be packaged into viral particles (10). Consequently, HSV amplicons have been considered very useful reagents for studying the mechanism of HSV replication (16, 32).

A prominent feature of HSV-1 DNA replication is frequent homologous recombination. This has been traditionally studied in experimental settings where repeated sequences are either present or being introduced into a single molecule. For example, frequent homologous recombination through the repeat sequences in the viral genome leads to free inversion of the L and S segments relative to each other and the generation of four isomeric forms in equimolar proportions (7, 13). Introduction of extra inverted repeats into the viral genome can cause additional isomerization, while insertion of direct repeats can result in sequence deletion (22, 28). These phenomena are also observed in HSV amplicon constructs when they are amplified through an HSV-mediated DNA replication mechanism (6, 9, 23). Other studies have suggested that homologous recombination does not seem to require HSV DNA as the source of homology, as non-HSV DNA, such as the bacterial ampicillin gene, can also cause isomerization when it is cloned into an HSV amplicon as duplicate inverted repeats (32, 33).

Homologous recombination between different HSV molecules (called intermolecular homologous recombination [IM-HR]), on the other hand, has been less well studied. Early investigations on IM-HR were carried out by cointroducing two HSV genomes of different phenotypes into the same cell; IM-HR was identified by the occurrence of progeny virus that obtained both phenotypes (17). More recently, IM-HR has been investigated through restriction site polymorphism (5, 30). However, due to the complexity of the viral genome structure and the limitations of the detection methods used, the actual extent of the IM-HR has not been accurately estimated.

There are two possible outcomes of an IM-HR event, sequence replacement and sequence insertion (see reference 31 for a review). In sequence replacement, target sequences on one molecule are precisely replaced by sequences that are flanked by homologous regions that are present on the transfected molecule. This replacement occurs primarily by a double-crossover event during a homologous recombination event. Because of the precise nature of this type of recombination, there is no significant net change in the number of nucleotides in the target sequence. Sequence integration results from a single-crossover recombination event. Thus, in a sequence insertion, part of or the entire transfected molecule inserts into the target locus, resulting in a net increase in the number of nucleotides in the genomic sequence.

In the present report, we have studied IM-HR in a uniquely designed HSV amplicon system. We constructed two nearly identical amplicon plasmids and inactivated the ampicillin gene in one of the plasmids through a point mutation. Cotransfection of these two plasmids into cells followed by infection with a helper HSV resulted in an extremely high frequency of homologous recombination: up to 60% of the amplicon concatemers retrieved from virion particles had undergone IM-HR. This high frequency of IM-HR required that both plasmids be replicated by HSV-mediated replication machinery, as it did not occur when one or both plasmids were replicated by simian virus 40 (SV40)-mediated DNA replication even with the presence of HSV infection. The majority of the homologous recombination events resulted in sequence replacement or targeted gene repair, whereas a minority resulted in sequence insertion.

Together, these data demonstrate that IM-HR can occur at a high frequency when two homologous amplicon plasmids are replicated by an HSV-mediated replication mechanism. This finding may have significant implications for further understanding of HSV replication mechanisms and also in developing gene targeting strategies for gene therapy purposes.

MATERIALS AND METHODS

Cells and viruses.

Vero cells (African green monkey kidney fibroblasts), HEK-293 (human embryonal kidney) cells, and COS-1 cells were obtained from the American Type Culture Collection and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (FCS). The BHK cells (TK⁻, gH⁺) have been described previously (37) and were cultured in Glasgow modified Eagle's medium supplemented with 5% tryptose broth and 10% FCS. A clone of HSV-1 strain SC16, a well-characterized clinical isolate with a low passage history (14), was used as the helper virus for amplicon replication and passage throughout the experiments presented in this paper. The virus stocks were routinely grown and titrated in Vero cells and stored at −70°C until required.

Plasmid construction.

The construction of the basic amplicon plasmid pW7, which carries an HSV origin of replication (oriS), a packaging signal, and a copy of the ampicillin resistance gene, has been reported previously (29, 37). To construct pW1, the kanamycin resistance gene from pVAX1 (Introgen Therapeutics, Houston, Tex.) was cut out with BspMI and XcmI and ligated into the unique NheI site of pW7 through blunt-end ligation. Plasmid pW2 was constructed from pW1 and carries a frameshift inactivation mutation in the ampicillin gene (achieved by digesting pW1 with AseI, followed by filling in the ends and ligating the plasmid). Plasmid pW2-lacZ was constructed by blunt-end ligating the 800-bp AatII-AflIII fragment of pNeb193 (Clontech), which contains the β-galactosidase gene (lacZ), into the unique EcoRV site of pW2. To delete oriS from pW7 and pW2-lacZ, the plasmids were digested with NruI to excise the 1.1-kb oriS-containing fragment and ligated to create pW7-del and pW2-lacZ-del, respectively.

To clone the SV40 replication origin into pW7-del and pW2-lacZ-del, the SV40 origin was first cut out from pcDNA3.1(+)/Zeo (Invitrogen) through AseI digestion. The 401-bp SV40 ori-containing fragment was then cloned into the unique NruI site of either pW7-del or pW2-lacZ-del through blunt-end ligation to create pW7-SV40-ori and pW2-SV40-ori, respectively. Plasmid DNA was prepared by the alkaline lysis procedure and then purified through Qiagen ion-exchange columns in accordance with the manufacturer's protocols (Qiagen Inc., Valencia, Calif.).

Transfection, HSV infection, and DNA extraction.

All the plasmid DNAs used in the experiments were grown in Escherichia coli DH10B (dam⁺ dcm⁺), released by alkaline lysis of the bacterial culture, and purified with a Qiagen-Tip 500 column. Mammalian cells were seeded in six-well plates at around 4 × 10⁵/well and grown overnight until they reached 70 to 80% confluency. Cells were transfected with a total of 2 μg of plasmid DNA, using Lipofectamine reagent (Gibco-BRL) according to the manufacturer's instructions. For cotransfection, the plasmids were mixed at a 1:1 ratio before being added to Lipofectamine solution. The DNA was initially added to 100 μl of H₂O and was then mixed with 5 μl of Lipofectamine, also in 100 μl of H₂O. The mixture was left at room temperature for 30 min before it was added to 0.8 ml of serum-free medium. The medium containing the Lipofectamine formulated DNA was overlaid directly on the cells for 4 h before it was replaced with complete medium.

For the subsequent virus infection, cells were infected with 1 PFU/cell of SC16 16 h after DNA transfection and collected 24 h later after cytopathic effect (CPE) had reached ≈100%. For continual passage of amplicon stocks, half of the virus harvested from the cells was used to infect freshly seeded cells. DNA was extracted from either the culture supernatant or the cell pellets through different procedures according to our earlier publications (36, 37). To prepare DNA from viruses released into the culture supernatant, the virus particles were initially pelleted through high-speed centrifugation (27,000 × g for 2 h). Viral pellets were resuspended in TE (10 mM Tris, 1 mM EDTA, pH 8.0) and gently phenol extracted three times, and viral DNA was precipitated with ethanol.

For extracting extrachromosomal DNA from cell pellets, a modified Hirt's procedure was used (15). Briefly, cells were collected, washed once with phosphate-buffered saline (PBS), and then resuspended in 160 μl of buffer I (5 mM Tris [pH 7.7], 10 mM EDTA), followed by addition of 20 μl of proteinase K (10 mg/ml) and 200 μl of buffer II (5 mM Tris [pH 7.7], 10 mM EDTA, 1.2% sodium dodecyl sulfate [SDS]). The samples were incubated at 37°C for 15 min before 100 μl of 5 M NaCl was added to each tube. After overnight incubation at 4°C, the supernatant was collected by 45 min of centrifugation at 18,000 × g. The supernatants were extracted with phenol-chloroform twice before the DNA was precipitated with 2.5 volumes of ethanol. The DNA pellets were dissolved in a volume of 50 μl of H₂O.

Bacterial transformation and determination of recombination frequency.

E. coli DH10B cells were transformed with either undigested extracted DNA or DNA that had been digested with PacI and ligated. PacI was chosen as the enzyme to digest the amplicon concatemers because it cuts only the pW2-lacZ amplicon and its derivatives (pW2-lacZ-del and pW2-SV40-ori) but does not cut pW7 and its derivatives (pW7-del and pW7-SV40-ori). After PacI digestion, the enzyme was inactivated by heat treatment (20 min at 65°C), followed by phenol extraction and ethanol precipitation. The DNA was then ligated overnight before it was transformed into competent DH10B cells (ElectroMax; Gibco-BRL) through electroporation using an Electroporator 2510 (Eppendorf, Madison, Wis.).

Following electroporation, the E. coli cells were grown in 0.5 ml of SOC medium without antibiotics for 1 h. Equal volumes of the cells were then plated onto antibiotic selection plates (ampicillin or kanamycin) that also contained isopropylthio-β-d-galactoside (IPTG) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). Bacterial colonies were enumerated 18 h later. The blue colonies that appeared on ampicillin plates represented those that contained the recombined DNA; this was further confirmed by AseI digestion on miniprep DNA prepared from a subset of the colonies. The frequency of intermolecular homologous recombination events was determined by dividing the number of Amp^r blue colonies with the number of Kan^r blue colonies.

RESULTS

Experimental design and plasmid construction.

Due to the extreme complexity and large size of the HSV genome, we decided to use HSV amplicons to conduct our investigation of HSV-mediated IM-HR. Initially, we constructed two similar HSV amplicon plasmids, pW7 and pW2-lacZ, that had approximately 70% homology. The major differences between the two plasmids are illustrated in Fig. 1. First, pW2-lacZ contains two extra genes encoding kanamycin resistance (kanR) and β-galactosidase (lacZ). Second, the ampicillin resistance gene (ampR) in pW2-lacZ has been inactivated through a frameshift mutation that destroys the AseI site (see Materials and Methods for details). Both amplicon plasmids can be efficiently amplified and packaged into viral particles when cointroduced into BHK cells with a helper HSV (data not shown).

FIG. 1.

Experimental design and possible outcomes of IM-HR between pW7 and pW2-lacZ. The positions of the crucial components contained in the plasmids, including oriS, the a sequence that contains the HSV packaging signal (a seq), the ampicillin (AmpR) and kanamycin (KanR) resistance genes, and the lacZ gene are marked inside the circular plasmid maps. The positions of the mutated AseI site (∂-AseI) and the sites of other restriction enzymes that were used for plasmid characterizations are indicated by their names outside the plasmid maps. The expected three-progeny plasmid patterns from the two outcomes of homologous recombination are depicted on the bottom. Among them, pattern I is from sequence replacement, which results in the mutated AseI site in the ampR gene's being repaired, and AseI digestion will generate two fragments, of 4.8 and 2.9 kb. Both patterns II and III are generated from sequence integration, and AseI digestion of the recombined product will generate two DNA fragments, one of them being substantially larger than that generated by sequence replacement.

These two plasmids were cotransfected into mammalian cells together with a helper virus, and DNA isolated from the viral particles was then transformed into E. coli. The color and drug resistance properties of the bacterial colonies provided a phenotypic readout of the recombination events that had occurred in mammalian cells. In the absence of homologous recombination between the two plasmids, transformation with pW7 would be expected to produce ampicillin-resistant white bacterial colonies, whereas transformation of pW2-lacZ would produce kanamycin-resistant blue colonies on X-Gal-containing plates. In contrast, homologous recombination between pW7 and pW2-lacZ would be expected to restore the function of the mutated ampicillin resistance gene in pW2-lacZ; such recombinants could be identified by the appearance of ampicillin-resistant blue colonies.

As mentioned earlier, we predicted that there would be two consequences resulting from IM-HR between these two amplicon plasmids, sequence replacement or sequence insertion. Although both events could result in the generation of blue colonies with ampicillin resistance, these two types of homologous recombination events could be differentiated by digesting the plasmid DNA with the restriction enzyme AseI. A sequence replacement event that leads to targeted gene repair would restore the deleted AseI site in the ampicillin gene of pW2-lacZ (pattern I in Fig. 1). The recombination product, upon AseI digestion, would generate two DNA fragments of 4.8 and 2.9 kb (see Fig. 1). AseI digestion of DNA from a sequence insertion event (where part of or the entire sequence of pW7 is integrated into pW2-lacZ through homologous recombination) would still generate two DNA fragments, but depending on the position of insertion relative to the generated AseI site, one of the DNA fragments will be substantially larger than that generated by sequence replacement (patterns II and III in Fig. 1).

High frequency of IM-HR during HSV-mediated plasmid DNA replication.

Purified pW7 and pW2-lacZ DNAs were mixed at a 1:1 (wt/vol) ratio, and the DNA mixture was then transfected into BHK cells by using Lipofectamine. Sixteen hours after transfection, the cells were either infected with wild-type helper HSV strain SC16 or mock infected with medium only. Twenty-four hours after virus infection, extrachromosomal DNA was extracted from the cell pellets and digested with PacI, which cuts pW2-lacZ once but does not cut pW7. Digestion of PacI therefore released the monomeric unit of pW2-lacZ (with or without homologous recombination) from the amplicon concatemer. The digested DNA was ligated after the PacI enzyme was inactivated. Competent E. coli cells were then transformed with the DNA, and equal volumes were plated onto agar plates containing X-Gal plus either ampicillin or kanamycin, and the blue colonies on the plates were enumerated.

Table 1 summarizes the results of four independent experiments. The number of Kan^r blue colonies was considered to represent the total number of pW2-lacZ molecules that either had or had not undergone homologous recombination. In contrast, the number of Amp^r blue colonies was considered to represent the pW2-lacZ DNA that had regained ampicillin resistance either through repairing the mutated ampicillin gene (sequence replacement) or obtaining the nonmutated ampicillin gene from pW7 (sequence insertion). Based on these assumptions, the data showed that a high percentage of pW2-lacZ (ranging from 22 to 52%) had undergone IM-HR during a single round of HSV-mediated DNA amplification. In contrast, no Amp^r blue colonies were detected in the control experiments, in which the amplicon plasmids were not replicated in the cells due to the absence of HSV infection or in which the two amplicons were transfected separately and the cells were infected with HSV before the extracted DNA was mixed for PacI digestion and ligation. This result suggested that the high frequency of IM-HR between these two amplicon plasmids required HSV-mediated DNA replication. This result also excluded the possibility that the IM-HR events detected might have occurred in E. coli, as this would have resulted in a similar number of blue colonies on ampicillin plates regardless of whether the BHK cells were infected with HSV.

TABLE 1.

Frequent IM-HR during HSV-mediated amplicon replication

Expta	No. of Ampr blue colonies	No. of Kanr blue colonies	% IM-HR
1	448	2,002	22.4
2	654	1,260	51.9
3	56	151	37.1
4	924	2,114	43.7
Control 1	0	1,184b	<0.1
Control 2	0	958	<0.1

^aControl 1, no HSV infection after plasmid transfection. Control 2, pW7 and pW2-lacZ were transfected separately, and the cells were infected with HSV before the DNA was mixed for digestion and ligation.

^bValues were from experiment 1, and no blue colonies were detected in ampicillin plates from the controls in the other three experiments.

To verify that the blue colonies from ampicillin plates truly represented pW2-lacZ DNA that had undergone IM-HR, DNA was prepared from a fraction of these colonies and digested with AseI. As illustrated in Fig. 1, if the homologous recombination event resulted mainly in sequence replacement (targeted gene repair), the mutated AseI site in the ampicillin gene would be repaired (pattern I) and AseI digestion would produce two predicted DNA fragments of 4.8 and 2.9 kb. If the homologous recombination event resulted in sequence insertion, extra DNA sequence of pW7 would be integrated into pW2-lacZ. AseI digestion of these DNA molecules would still generate two DNA fragments, but the sizes of these two fragments would be different from those of fragments generated by sequence repair due to the insertion of extra DNA sequence from pW7. Depending on the position of the insertion relative to the generated AseI site, two general band patterns are predicted. If the insertion occurs in the 4.8-kb region, the 2.9-kb fragment will remain the same size and the 4.8-kb fragment will become larger (pattern II). In contrast, if the insertion occurs within the 2.9-kb AseI fragment (pattern III), the 4.8-kb fragment will remain the same size and the 2.9-kb fragment will show as a larger band.

Based on these predictions, 12 of 19 DNA samples (63%) appeared to have undergone IM-HR events resulting in gene repair (Fig. 2). The remaining seven samples (37%) displayed banding patterns consistent with sequence insertion events. For the latter class, two samples (no. 7 and 18) had the insertion in the 2.9-kb region (since the 4.8-kb band remained unchanged after AseI digestion), and the other five samples (no. 1, 4, 10, 13, and 15) had the insertion site in the 4.8-kb region (since the 2.9-kb band remained unchanged after AseI digestion). Together, these data suggest that the Amp^r blue colonies truly represented the recombined amplicon pW2-lacZ and therefore validate the assumption that the enumeration of blue colonies on ampicillin plates is an easy and convenient method for quantification of IM-HR during HSV-mediated DNA replication. These data also suggest that during homologous recombination, sequence replacement occurred at a higher frequency than sequence insertion.

FIG. 2.

Confirmation of the recombined amplicon DNA from IM-HR by AseI digestion. Nineteen blue colonies were randomly picked from ampicillin plates, and DNA was prepared by miniprep extraction. The DNA samples (lanes 1 to 19) as well as the plasmid DNA of pW7 (lane 20) and pW2-lacZ (lane 21) were digested with AseI and electrophoresed on a 0.8% agarose gel. DNA size markers (lanes M) are the 1-kb DNA ladder.

To further characterize these recovered plasmids, two isolates were randomly chosen from the DNA samples that represented each of the three types of recombination patterns (Fig. 1): pattern I, sequence replacement; pattern II, sequence insertion in the 4.8-kb AseI region; and pattern III, sequence insertion in the 2.9-kb AseI region. The DNA was digested with either a single enzyme or a combination of restriction enzymes. The expected fragments from these digestions, which remained constant with different entry sites of integration along the homologous region, are listed in Fig. 3A. The actual results from the gel electrophoresis are shown in Fig. 3B. All samples produced the predicted DNA fragment patterns, further confirming the conclusions drawn from the AseI digestion experiment. Together, these results confirm that during HSV-mediated amplicon DNA replication, homologous recombination between two different plasmids is a frequent event which results in either sequence replacement or sequence insertion.

FIG. 3.

Further confirmation of the different IM-HR patterns. Two samples were randomly chosen from each of the three patterns of IM-HR shown in Fig. 2 (pattern I, sequence replacement; pattern II, sequence insertion in the 2.9-kb AseI fragment; pattern III, sequence insertion in the 4.8-kb AseI fragment) and numbered from 1 to 6. The DNA was digested with the indicated restriction enzymes. (A) List of the expected DNA fragments and their sizes. (B) Actual results from agarose gel electrophoresis. DNA markers (lanes M) are the 1-kb ladder.

Comparison of IM-HR frequency from DNA samples of different preparations.

To determine if DNA obtained from different stages of amplicon propagation or from different fractions of samples would show a difference in either the frequency or pattern of IM-HR, we prepared DNA from both the cell pellets and culture supernatant either from the first round (transfection-infection) or after the amplicon stocks were passaged once (P1). To extract amplicon concatemers from the supernatant, the viral particles were initially spun down through high-speed centrifugation, and the DNA was then extracted from the pelleted virion particles (see Materials and Methods). The Hirt extraction procedure was used to extract the cell-associated extrachromosomal DNA from the cell pellets. The DNA, with or without PacI digestion and subsequent ligation, was then transformed into competent E. coli and spread onto agar plates containing either ampicillin or kanamycin. The results from this experiment are presented in Table 2.

TABLE 2.

IM-HR frequency and patterns in samples of different preparations^a

Sample prepn	No. of blue colonies		% IM-HR	Outcome (%) of IM-HR
	Ampr	Kanr		Sequence replacement	Sequence insertion
First round
Supernatant	482	1,470	32.8	NDb	ND
Cell pellet	396	1,288	30.7	12/19 (63)	7/19 (37)
Cell pellet, undigested DNA	56	160	35.0	13/15 (87)c	2/15 (13)
P1 cell pellet	754	1,218	61.9	26/28 (93)d	2/28 (7)

^aThe colony numbers represent average numbers in two repeated experiments.

^bND, not done.

^cP < 0.05.

^dP < 0.01 compared with first-round cell pellet.

During the first round of amplification, DNAs from both the supernatant and cell pellet fractions gave similar frequencies of IM-HR, 32.8 and 30.7%, respectively. However, the Hirt extraction of cell pellet DNA from cells of P1 gave a substantially higher percentage of IM-HR. Unfortunately, there was not enough DNA from the supernatant fraction of the P1 amplicon for this analysis. In addition, the uncut DNA from the cell pellet fraction of P1 produced almost the same percentage of IM-HR as the PacI-digested and ligated DNA, although there was a substantial reduction in the number of blue colonies on both ampicillin and kanamycin plates. These data demonstrate that a high frequency of IM-HR existed in DNA prepared from different fractions of samples that were obtained from different stages of amplicon propagation, with the highest frequency of IM-HR detected in amplicon samples that had been passaged once.

To determine the relative frequency of the three different patterns of IM-HR in each of the DNA preparations, blue colonies were randomly picked from ampicillin plates, and DNA was prepared and digested with AseI. The results again showed that all the blue colonies tested contained amplicon plasmid DNA that had undergone IM-HR. However, there was a significant difference in the frequency of individual types of IM-HR (Table 2). The DNA from both the P1 cell pellet fraction (PacI digested and ligated) and the uncut first-round cell pellet fraction had a much higher percentage of sequence replacement than the DNA from the first-round cell pellet fraction that had been digested with PacI and ligated; approximately 90% of colonies from these two groups were found to have had sequence replacement, in contrast to the 63% detected in the DNA of the first-round cell pellet. The reasons for this difference are not yet clear.

One assumption made in these experiments was that the DNA obtained from amplicons that had been passaged once represented replicated DNA. To confirm this, the DNA from the P1 cell pellet fraction was digested with either AseI alone or AseI together with DpnI, which digests any unreplicated plasmid DNA. The DNA was then ligated and transformed into E. coli. The number of blue colonies on ampicillin plates remained essentially unchanged regardless of whether the DNA had been digested with DpnI (data not shown), indicating that the DNA that had undergone IM-HR was indeed mostly newly synthesized.

Determination of IM-HR frequency in cells of different origins.

To determine if the observed high frequency of IM-HR during HSV-mediated amplicon replication could also occur in other types of cells, we repeated the experiment described in Table 1 in a panel of cells of different species and origins that sufficiently support HSV replication. The DNA used for bacterial transformation was from the first-round cell pellet fraction and had been digested with PacI and ligated. The results showed that the frequency of IM-HR was highest in BHK cells, whereas the IM-HR frequency in other cells tested was lower; around 15% of the amplicon DNA amplified in these cells underwent IM-HR (Table 3). Again, these results were confirmed by AseI digestion of DNA prepared from a fraction of the blue colonies present on ampicillin plates (data not shown). These results indicate that high-frequency IM-HR mediated by the HSV replication machinery is a universal phenomenon that occurs in a variety of mammalian cells.

TABLE 3.

Frequency of HSV-mediated IM-HR in different cell lines

Cells	No. of blue coloniesa		% IM-HR
	Ampr	Kanr
BHK	482	1,470	32.8
COS	99	644	15.4
Vero	133	844	15.7
293	102	569	17.9

^aValues are averages for three repeated experiments.

High-frequency IM-HR requires that both amplicon plasmids be amplified by the HSV replication machinery.

The data presented in Table 1 suggests that the HSV-mediated replication mechanism is required for the efficient generation of IM-HR. To determine if the replication of both amplicon plasmids is required or if replication from only one is sufficient, we deleted the HSV replication origin (oriS) from the amplicon plasmids to create pW7-del and pW2-lacZ-del. We then mixed an oriS-deleted amplicon plasmid with another unmodified oriS-containing plasmid for transfection into BHK cells. Deletion of oriS from either of the amplicon plasmids severely reduced the incidence of IM-HR (Table 4), indicating that the presence of the oriS sequence in both amplicon plasmids was essential for the high frequency of IM-HR. As deletion of oriS reduces the homologous region between these two plasmids by only 200 bp, such a small decrease in homology is unlikely to be responsible for the observed reduction in IM-HR incidence.

TABLE 4.

oriS is required in both amplicon plasmids for efficient IM-HR

Amplicon plasmids	No. of blue coloniesa		% IM-HR
	Ampr	Kanr
pW7 + pW2-lacZ	324	741	43.7
pW7-del + pW2-lacZ	2	700	0.3
pW7 + pW2-lacZ-del	8	112	7.1
pW7-del + pW2-lacZ-del	0	200	0

^aValues are averages for two repeated experiments.

Next we investigated whether high-frequency IM-HR could also occur during plasmid DNA replication driven by other viral machinery, either in the presence or absence of HSV superinfection. Among the DNA viruses, the SV40 genome is amplified through a theta replication mechanism. However, in the presence of HSV-1, replication of SV40 DNA can lead to predominant concatemeric replication (12, 20). This SV40 origin-dependent process is governed by the SV40 large T antigen and facilitated by HSV-encoded DNA replication proteins, such as DNA polymerase, single-strand DNA binding protein (SSB), and helicase-primase, in a mode specific for SV40 DNA replication (4). It has previously been shown that during this combined action of DNA replication, frequent homologous recombination within a single SV40 molecule can also occur when inverted repeats are introduced into the SV40 plasmid genome (3, 4). We therefore replaced the HSV oriS sequence in both pW7 and pW2-lacZ with the replication origin from SV40 to create pW7-SV40-ori and pW2-SV40-ori, respectively.

We initially tested these two SV40 ori-containing plasmids for their ability to be replicated in COS-1 cells and found that the large T antigen provided by this cell could amplify both plasmids more than 1,000-fold after they were transfected into the cells (data not shown). These two plasmids were then mixed and transfected into COS-1 cells in the presence or absence of HSV superinfection. The DNA was extracted and treated the same way as described for the experiment presented in Table 1 and then used to transform competent E. coli. The results in Table 5 show that the frequency of IM-HR between these two plasmids during SV40-mediated replication was very low compared with that of amplicon plasmids replicated by the HSV mechanism. In addition, HSV superinfection seemed to only marginally enhance the IM-HR frequency when two SV40 ori-containing plasmids were used. Overall, these results indicate that the generation of high-frequency IM-HR during HSV-mediated amplicon replication requires, at least in part, the participation of the HSV replication machinery.

TABLE 5.

IM-HR during SV40-mediated plasmid DNA replication

Amplicon plasmids	HSV infection	No. of blue coloniesa		% IM-HR
		Ampr	Kanr
pW7 + pW2-lacZ	+	10	70	14.3
pW7 + pW2-lacZ	−	0	539	0
pW7-SV40-ori + pW2-SV40-ori	+	2	127	1.6
pW7-SV40-ori + pW2-SV40-ori	−	2	326	0.6

^aValues are percent averages for two repeated experiments. IM-HR was calculated from the averaged numbers.

DISCUSSION

We have shown in this report that during HSV-mediated amplicon plasmid replication, up to 60% of the replicative intermediates have undergone intermolecular homologous recombination. This frequency may be an underestimate, considering that not 100% of cells receive both plasmids during Lipofectamine-mediated cotransfection. Therefore, the actual frequency of IM-HR could be even higher than these experiments have indicated. As nonhomologous recombination in mammalian cells is extremely rare (usually one in >10⁶) (8), we do not believe that nonhomologous recombination contributed significantly to the high frequency of recombination detected in our experiments. Analyses of the recombined DNA revealed that the majority of these IM-HR events resulted in sequence replacement, while the minority resulted in sequence insertion. One possible reason for the discrepancy in the frequency of these two outcomes is that the plasmids resulting from sequence insertion may contain more than one bacterial origin, which may be selected against during growth in E. coli.

Studies by our laboratory and others have shown that HSV replicative intermediates very frequently contain adjacent genomic units with L and S segments in different orientations (called concatemers with internal segment isomerization, or concatemer-ISI), from which all the four possible HSV isomers can be generated in equal molar ratios through random cleavage and packaging (1, 21, 25, 36). Subsequent studies have indicated that concatemer-ISI may be generated through IM-HR, as mixed infection with two viruses can generate replicative intermediates in which the two viral genomes are frequently interspersed along a single concatemer (27). However, due to the extremely large size of the molecules analyzed in those experiments, the mechanism for IM-HR could not be elucidated. The results presented in this work seems to indicate that these unique HSV concatemers may have been generated through the two mechanisms that we have elucidated in the present study, through either sequence replacement or sequence insertion. If this indeed is the case, it is highly likely that such frequent IM-HR may contribute directly to HSV genome isomerization.

It has been demonstrated that HSV-mediated DNA replication is specifically required for high-frequency homologous recombination between repeats in a single molecule (9, 33). Our data demonstrate that this is also the case during IM-HR. Our data also indicate that both amplicon plasmids need to be replicated by an HSV-mediated replication mechanism in order to generate high-frequency IM-HR. However, although high-frequency homologous recombination through the repeats in a single molecule has been demonstrated during SV40-mediated plasmid replication in the presence of HSV superinfection (3), this has been proven not to be the case for IM-HR. These results, together with our earlier work on HSV replication-mediated isomerization (32), suggest that HSV-1 replication-mediated homologous recombination, either intramolecular or intermolecular, is an integrated part of the viral DNA replication mechanism. However, our data suggest that inter- and intramolecular HR may employ slightly different mechanisms in different experimental settings.

Our analysis of IM-HR in different cell lines showed that the frequency of IM-HR between the two amplicons was substantially higher in BHK cells than in COS-1, Vero, or HEK-293 cells. Studies by Yao and Elias have also revealed that BHK cells are more efficient than other cells, such as BALB/c 3T3 cells, at supporting homologous recombination on repeated sequences in plasmid DNA (35). This suggests that in addition to viral proteins, certain cellular proteins may also be required for the efficient generation of IM-HR; these cellular proteins may be more abundantly expressed in BHK cells than in the other cells we have tested. Nonetheless, the frequency of IM-HR in COS-1 and Vero cells is still very significant (reaching 15%), indicating that IM-HR during HSV-mediated DNA replication is a specific event that is directly linked to the mechanism of HSV DNA replication itself. Thus, cellular factors may augment but do not dictate this process.

The detection of relatively abundant blue bacterial colonies, which are assumed to represent the circular products of IM-HR, in the absence of PacI digestion and ligation cannot be explained by the rolling-circle mechanism of replication. Two possibilities may have resulted in the generation of these circular molecules: they may have been generated from the concatemers through recombination, or they may have been generated from a circle-to-circle (theta) replication mechanism with the participation of frequent homologous recombination. A recent study in our laboratory showed that during lytic HSV infection, theta replication may indeed be a part of the viral replication strategies (Fu et al., submitted for publication). In addition, characterization of the amplicon DNA of IM-HR has identified a discrepancy between DNA extracted from different stages of amplicon preparation: the DNA extracted from cells harboring passaged amplicons has a twofold higher incidence of sequence replacement than the DNA extracted from cells of first-round transfection-infection. The reason for this is currently unclear. One possibility is that the DNA from cells of transfection-infection may contain a much higher proportion of products from circle-to-circle replication, and in those products, sequence insertion may be more frequent than that from rolling-circle replication.

It is expected that in addition to its possible role in the generation of HSV genome isomerization, high-frequency IM-HR may also play some other important roles during HSV DNA replication and the virus life cycle. One possibility is that the IM-HR may help maintain the integrity of the viral genome through repairing any mutated or damaged HSV DNA that may occur during the rapid replication of viral DNA. Also, from an evolutionary point of view, the high frequency of IM-HR may have helped the virus to quickly adapt to the changing environment of the host through free and rapid accumulation of mutations or extra DNA sequences from each individual viral genome. Alternatively, it is also possible that the same mechanism may have contributed to the diversity of the viral genome in the many different strains of virus that existed in nature.

The demonstration that oriS must be present in both amplicon plasmids to generate a high frequency of IM-HR indicates that the virus-encoded polypeptides (in particular, the viral proteins essential for origin-dependent DNA replication) may play a key role. Dissecting the genes or mechanisms involved in this high frequency of IM-HR may help in designing a better strategy for gene therapy for many inherited genetic disorders, such as cystic fibrosis and hemophilia, which are caused by defined mutations in single disease genes. Current gene therapy strategies that are considered for such diseases usually involve the introduction of an additional, mutation-free copy of the disease-causing gene into the cells of an appropriate tissue. However, there are inherent problems with such gene augmentation strategies, including inappropriate or unstable expression of the new gene, and safety concerns arising from random integration into the host genome. Repair of the mutated endogenous gene through homologous recombination would avoid these problems. The efficiency of current gene targeting approaches, however, is generally low, and this has prevented them from being widely considered for gene therapy (34). Further understanding of the mechanism of HSV-mediated high-frequency IM-HR, in particular, adapting the efficient targeted gene repair mechanism to popular gene delivery vectors, will greatly benefit this field. Also, our current understanding of homologous recombination, both the genetics and the biochemistry of the process, is primarily based on observations made in E. coli. HSV-mediated IM-HR of amplicons may serve as an attractive model for studying homologous recombination in a mammalian system, and the knowledge obtained from it will likely provide significant input into the gene therapy field.