Homologous and nonhomologous recombination in monkey cells

Presence of replicating virus in recombinant hepadnavirus stocks results from recombination and can be eliminated by the use of a packaging cell line

Mutant hepatitis B viruses are useful tools to study the viral life cycle and viral pathogenesis. Furthermore, recombinant hepatitis B viruses are candidate vectors for liver-directed gene therapy. Because wild-type viruses present in recombinant or mutant virus stocks may falsify experimental results and are detrimental for a viral vector, we investigated whether and to what extent wild-type virus is present in recombinant virus stocks and where it originates from. We took advantage of the duck model of hepatitis B virus infection which allows very sensitive detection of replication-competent viruses by infection of primary duck hepatocytes or of ducklings in vivo. Recombinant hepatitis B virus stocks contained significant amounts of wild-type viruses, which were most probably generated by homologous recombination between plasmids containing homologous viral sequences. In addition, replication-competent viral genomes were reconstituted from plasmids which contained replication-deficient but redundant viral sequences. Using a stable cell line for packaging of deficient viral genomes, no wild-type virus was detected, neither by infection of primary hepatocytes nor in vivo.

d(GA x TC)(n) microsatellite DNA sequences enhance homologous DNA recombination in SV40 minichromosomes

The genomic distribution of the abundant eukaryotic d(GA x TC)(n) DNA microsatellite suggests that it could contribute to DNA recombination. Here, it is shown that this type of microsatellite DNA sequence enhances DNA recombination in SV40 minichromosomes, the rate of homologous DNA recombination increasing by as much as two orders of magnitude in the presence of a d(GA x TC)(22) sequence. This effect depends on the region of the SV40 genome at which the d(GA x TC)(22) sequence is cloned. It is high when the sequence is located proximal to the SV40 control region but no effect is observed when located 3.5 kb away from the SV40 ori. These results indicate that the recombination potential of d(GA x TC)(n) sequences is likely linked to DNA replication and/or transcription. The potential contribution of the structural properties of d(GA x TC)(n) sequences to this effect is discussed.

Herpes simplex virus type 1 DNA replication is specifically required for high-frequency homologous recombination between repeated sequences

Using an assay for recombination that measures deletion of a beta-galactosidase gene positioned between two directly repeated 350-bp sequences in plasmids transiently maintained in COS cells, we have found that replication from a simian virus 40 origin produces a high frequency of nonhomologous recombination. In contrast, plasmids replicating from a herpesvirus origin (oris) in COS cells superinfected with herpes simplex virus type 1 (HSV-1) show high levels of homologous recombination between the repeats and an enhanced recombinogenicity of the HSV-1 a sequence that is not seen during simian virus 40 replication. When the same assay was used to study recombination between 120- to 150-bp repeats in uninfected Vero cells, the level of recombination was extremely low or undetectable (< 0.03%), consistent with the fact that these repeats are smaller than the minimal efficient processing sequence for homologous recombination in mammalian cells. Recombination between these short repeats was easily measurable (0.5 to 0.8%) following HSV-1 infection, suggesting that there is an alteration of the recombination machinery. The frequency of recombination between repeats of the Uc-DR1 region, previously identified as the only segment of the HSV-1 a sequence indispensable for enhanced a-sequence recombination, was not significantly higher than that measured for other short sequences.

Intramolecular recombination in polyomavirus DNA is controlled by promoter elements

We show here that intramolecular homologous recombination in polyomavirus (Py) DNA depends upon discrete sequence elements of the viral regulatory region which are believed to regulate transcription initiation and exert little or no cis-control over replication. Either deleting the viral early promoter (EP) or inverting the viral late promoter (LP) strongly impairs viral DNA recombination under conditions allowing viral DNA replication to proceed undisturbed. These findings suggest that bi-directional transcription proceeding from the intergenic region favors intramolecular recombination.

Premature strand transfer by the HIV-1 reverse transcriptase during strong-stop DNA synthesis

Reverse transcription of retroviral genomes starts near the 5' end of the viral RNA by use of an associated tRNA primer. According to the current model of reverse transcription, the initial cDNA product, termed minus-strand strong-stop DNA, 'jumps' to a repeated sequence (R region) at the 3' end of the RNA template. The human retroviruses have relatively long R regions (97-247 nucleotides) when compared to murine and avian viruses (16-68 nucleotides). This suggests that the full complement of the R region is not required for strand transfer and that partial cDNA copies of the 5' R can prematurely jump to the 3' R. To test this hypothesis, we generated mutants of the human immunodeficiency virus with R region changes and analyzed whether 5' or 3' R sequences were inherited by the progeny. We found that in most cases, 5' R-encoded sequences are dominant, which is consistent with the model of reverse transcription. Using a selection protocol, however, we were also able to identify progeny viruses with R sequences derived from the original 3' R element. These results suggest that partial strong stop cDNAs can be transferred with R region homologies much shorter than 97 nucleotides.

Sequence conversion during postreplicative adenovirus overlap recombination

Sequence conversion efficiently transfers genetic information in high yield during postreplicative adenovirus overlap recombination. This process is intrinsically nonreciprocal, depends on adenovirus-specific strand-displacement replication by both partner molecules, and requires that complementary sequences on displaced strands must exceed a minimal length to form a heteroduplex intermediate.

Homologous recombination between plasmid DNA molecules in maize protoplasts

The requirements for homologous recombination between plasmid DNA molecules have been studied using the PEG (polyethylene glycol)-mediated transformation system of maize (Zea mays L.) protoplasts coupled with the transient expression assay for beta-glucuronidase (GUS). Two plasmids were introduced into maize protoplasts; one plasmid (pB x 26) contained a genomic clone of the Adh1 maize gene; the other plasmid (piGUS) was a promoterless construction containing part of intron A of the Adh1 gene fused to the gusA coding sequence. Thus, the two vectors shared an effective homologous region consisting of a 459 bp (HindIII-PvuII) fragment of the Adh1 intron A sequence. An active gusA fusion gene would result upon homologous recombination between the plasmids within the intron A sequence, and indeed GUS activity was observed in extracts following co-transformation of maize protoplasts with the two plasmids. The presence of recombinant DNA molecules in protoplast DNA isolated 1 day after co-transformation was verified using polymerase chain reactions (PCR) and Southern blots. For efficient homologous recombination, both plasmids had to be linearized. The recombination reaction was induced by restriction of the plasmid molecules either inside the effective homologous region or at the borders of the intron sequence. However, the presence of even small, terminal, nonhomologous sequences at the 3' end of the pB x 26 fragment inhibited the recombination reaction. Also, both ends of the linearized piGUS DNA molecules were involved in the recombination reaction. The results revealed some features of homologous recombination reactions occurring in plant cells which cannot be accommodated by mechanisms postulated for similar reactions in animal system and in lower eukaryotes.

Genetic exchange between endogenous and exogenous LINE-1 repetitive elements in mouse cells

The repetitive LINE (L1) elements of the mouse, which are present at about 10(5) copies per genome and share over 80% of sequence homology, were examined for their ability to undergo genetic exchange with exogenous L1 sequences. The exogenous L1 sequences, carried by a shuttle vector, consisted of an internal fragment from L1Md-A2, a previously described member of the L1 family of the mouse. Using an assay that does not require the reconstitution of a selectable marker we found that this vector, in either circular or linear form, acquired DNA sequences from endogenous L1 elements at a frequency of 10(-3) to 10(-4) per rescued vector. Physical analysis of the acquired L1 sequences revealed that distinct endogenous L1 elements acted as donors and that different subfamilies participated. These results demonstrate that L1 elements are readily capable of genetic exchange. Apart from gene conversion events, the acquisition of L1 sequences outside the region of homology suggested that a second mechanism was also involved in the genetic exchange. A model which accounts for this mechanism is presented and its potential implication on the rearrangement of L1 elements is discussed.

Alternative homologous and nonhomologous products arising from intramolecular recombination

RmI, a chimeric DNA molecule containing polyomavirus (Py) and mouse sequences, generates unit-length Py DNA via intramolecular recombination between two directly repeated viral sequences of 182 base pairs (S repeats). To analyze the contribution of the S repeats in this process, we produced mutants of RmI carrying deletions in either one or both S repeats and tested them for their ability to recombine in mouse 3T6 cells. Mutant DNAs were found to yield unit-length Py DNA as long as they carried a minimal internal homology of 40 to 50 base pairs. Unlike RmI itself, however, the mutants also gave rise to nonhomologous recombination products. These results suggest that when the generation of homologous products is hampered by a limiting homology, nonhomologous products may arise instead of homologous ones. Therefore, the initial step(s) in the mechanisms yielding the two kinds of products could be identical.

Homologous recombination in a mammalian plasmid

Bovine papillomavirus (BPV) shuttle vectors replicate as a circular plasmid in mouse cell nuclei without impairing host cell viability. We used these vectors to analyze homologous recombination in mammalian cells. When several BPV-based plasmids carrying direct repeats were introduced into C127 cells, we detected many recombinant plasmid molecules that have lost the sequence between the repeats. Many recombinant type molecules as well as parental type molecules were detected in all the cell clones isolated for analysis. Sequencing after rescue of the plasmid in Escherichia coli showed that most of the recombinants were from accurate homologous recombination. When the repeats on the plasmid were in inverted orientation, no crossing-over type products were detected. We discuss possible mechanisms that explain these features.

Efficient duck hepatitis B virus production by an avian liver tumor cell line

Duck hepatitis B virus (DHBV) is produced in small amounts following transfection of human hepatoma or hepatoblastoma cell lines with cloned viral DNA. In a search for better hosts for DHBV replication, two avian liver cell lines were investigated. One of these cell lines, LMH, produced 5 to 10 times more DNA replicative intermediates and 10 to 20 times more infectious DHBV than did either of the two human cell lines, HuH-7 and Hep G2. Utilization of cell lines in genetic analyses of virus replication is often dependent upon obtaining efficient complementation between cotransfected viral genomes. We assayed transcomplementation of a viral polymerase (pol) gene mutant, which is rather inefficient in transfected human cells, and found that viral DNA synthesis was at least 20 times more efficient following cotransfection of LMH cells than in similarly transfected HuH-7 cells. Recombination, a potential interpretation problem in complementation assays, occurred at low levels in the cotransfected cultures but was substantially reduced or eliminated by creation of an LMH subline stably expressing the viral polymerase. This cell line, pol-7, supported the replication of DHBV pol mutants at ca. 10 to 15% of the level of virus replication obtained following transfection with wild-type viral DNA. By transcomplementation of a pol gene mutant in LMH cells, we were able to produce sufficient virus with the mutant genome to investigate the role of polymerase in covalently closed circular DNA amplification. Our results substantiate the hypothesis that covalently closed circular DNA is synthesized by the viral reverse transcriptase.

Intrachromosomal recombination mediated by papovavirus large T antigens

To investigate the mechanism by which the large T antigen (T-Ag) of polyomavirus and simian virus 40 can promote recombination in mammalian cells, we analyzed homologous recombination events occurring between two defective copies of the polyomavirus middle T (pmt) oncogene lying in close proximity on the same chromosome in a rat cell line. Reconstitution of a functional pmt gene by spontaneous recombination occurred at a rate of about 2 x 10(-7) per cell generation. Introduction of the polyomavirus large T (plt) oncogene into the cell line by DNA transfection promoted recombination very efficiently, with rates in the range of 10(-1) to 10(-2) per cell generation. Recombination was independent of any amplification of viral sequences and could even be promoted by the large T-Ag from simian virus 40, which cannot activate polyomavirus DNA replication. To explain the role of large T-Ag, we propose a novel mechanism of nonconservative recombination involving slipped-strand mispairing between the two viral repeats followed by gap repair synthesis.

Preferred crossover sites on polyomavirus DNA

RmI is a hybrid replicon consisting of polyomavirus (Py) and mouse sequences that yields unit-length polyomavirus DNA via recombination between two directly repeated viral sequences of 182 base pairs (S repeats). To define the contribution of the S repeats in this intramolecular recombination, we derived from RmI a series of replicons containing the original S repeats as well as additional direct viral repeats which were 1 to 2 kilobases in length (L repeats). After mouse 3T6 cells were transfected with these constructs, recombination products that displayed the physical properties of homologous recombinants were detected. The structures of these recombinants indicated that whereas repeat length influences the likelihood of recombination, crossover occurs preferentially near the S repeats, provided that one of them is proximal to the viral origin of replication. This finding suggests that recombination near the S repeats depends on a process initiated near the viral origin of replication.

Short DNA fragments induce site specific recombination in mammalian cells

A defective hprt gene was corrected by homologous recombination in a lymphocyte cell line deficient in Hypoxanthine-phosphoribosyl-transferase activity (hprt). In a novel approach, only a fragment of a cDNA clone of the functional hprt gene was used to induce homologous recombination. The mutation that was corrected corresponds to a single base change in exon III of the hprt gene. Two transfection methods, electroporation and the previously unreported use of polyoma capsids containing only short DNA fragments, were able to induce the recombinational event. After transfection cells with a functional hprt gene were selected and homologous recombination events were identified using polymerase chain reaction. Double stranded fragments and both coding and non-coding single stranded fragments resulted in conversion to a functional gene. Analysis of the resulting hprt positive cells revealed that most cells had undergone a simple replacement reaction. Interestingly, however, some cells had lost an intron adjacent to the site of mutation. Potential mechanisms for this phenomenon, including the possible involvement of RNA in DNA repair, are discussed.

Homologous plasmid recombination is elevated in immortally transformed cells

The levels of intramolecular plasmid recombination, following transfection of a plasmid substrate for homologous recombination into normal and immortally transformed cells, have been examined by two independent assays. In the first assay, recovered plasmid was tested for DNA rearrangements which regenerate a functional neomycin resistance gene from two overlapping fragments. Following transformation of bacteria, frequencies of recombinationlike events were determined from the ratio of neomycin-resistant (recombinant) colonies to ampicillin-resistant colonies (indicating total plasmid recovery). Such events, yielding predominantly deletions between the directly repeated sequences, were substantially more frequent in five immortal cell lines than in any of three normal diploid cell strains tested. Effects of plasmid replication or interaction with T antigen and of bacterially mediated rejoining of linear molecules generated in mammalian cells were excluded by appropriate controls. The second assay used limited coamplification of a control segment of plasmid DNA, and of the predicted recombinant DNA region, primed by two sets of flanking oligonucleotides. Each amplified band was quantitated by reference to a near-linear standard curve generated concurrently, and recombination frequencies were determined from the ratio of recombinant/control DNA regions. The results confirmed that recombinant DNA structures were generated within human cells at direct repeats in the transfected plasmid and were markedly more abundant in an immortal cell line than in the diploid normal cells from which that line was derived.

High recombination rate of an Epstein-Barr virus-simian virus 40 hybrid shuttle vector in human cells

The stability of an Epstein-Barr virus (EBV)-simian virus 40 (SV40) hybrid shuttle vector, the p205-GTI plasmid, was analyzed in human cells during EBV- or SV40-type replication mode. When the p205-GTI plasmid was maintained as an episomal EBV vector in the human 293 cell line, no rearrangement was detected. To induce the SV40 replication mode, cells containing the episomal p205-GTI plasmid were either transfected with vectors carrying the T antigen gene or infected with SV40. Surprisingly, we observed both production and amplification of different classes of recombinant molecules. Particular types of modifications were found in most of the recombinants. The most striking rearrangement was a duplication of the promoter and enhancer regions of SV40 which was inserted in the thymidine kinase (TK) promoter. This recombination process involved a few bases of homology, and one of the recombination junctions implicated the GC boxes which constitute the essential components of the TK and SV40 early promoters. Our results suggest that a combination of a low level of base homology and a specific DNA sequence function (promoter and enhancer sites) leads to a very high level of recombinational activity during T-antigen-dependent plasmid replication.

Homologous plasmid recombination is elevated in immortally transformed cells

The levels of intramolecular plasmid recombination, following transfection of a plasmid substrate for homologous recombination into normal and immortally transformed cells, have been examined by two independent assays. In the first assay, recovered plasmid was tested for DNA rearrangements which regenerate a functional neomycin resistance gene from two overlapping fragments. Following transformation of bacteria, frequencies of recombinationlike events were determined from the ratio of neomycin-resistant (recombinant) colonies to ampicillin-resistant colonies (indicating total plasmid recovery). Such events, yielding predominantly deletions between the directly repeated sequences, were substantially more frequent in five immortal cell lines than in any of three normal diploid cell strains tested. Effects of plasmid replication or interaction with T antigen and of bacterially mediated rejoining of linear molecules generated in mammalian cells were excluded by appropriate controls. The second assay used limited coamplification of a control segment of plasmid DNA, and of the predicted recombinant DNA region, primed by two sets of flanking oligonucleotides. Each amplified band was quantitated by reference to a near-linear standard curve generated concurrently, and recombination frequencies were determined from the ratio of recombinant/control DNA regions. The results confirmed that recombinant DNA structures were generated within human cells at direct repeats in the transfected plasmid and were markedly more abundant in an immortal cell line than in the diploid normal cells from which that line was derived.

Targeted correction of a major histocompatibility class II E alpha gene by DNA microinjected into mouse eggs

DNA molecules containing the 5' end of a functional major histocompatibility class II E alpha gene were injected into mouse eggs bearing E alpha genes with 630-base-pair (bp) deletions encompassing the promoter and first exon. The deletion was corrected by homologous recombination in 1 of about 500 transgenic mice that incorporated the injected DNA. The corrected E alpha gene was transmitted to progeny, which were bred to homozygosity. Southern blot analysis, polymerase chain reaction amplification of the DNA spanning the deletion, and sequence analysis revealed that the corrected allele resembles the wild-type E alpha gene. At sites of single-base-pair polymorphisms, there was apparently random conversion to either the donor or recipient sequence. In addition, many point mutations were introduced. mRNAs were produced from the corrected allele in a tissue-specific manner, but their sizes were different from the wild-type allele, and they did not produce detectable E alpha protein. This experiment demonstrates the feasibility of targeting foreign DNA to a gene that is completely inactive in fertilized mouse eggs.

Cre-stimulated recombination at loxP-containing DNA sequences placed into the mammalian genome

The cre gene of coliphage P1 encodes a 38 kDa protein which efficiently promotes both intra- and intermolecular recombination at specific 34 bp sites called loxP. To demonstrate that the Cre protein can promote DNA recombination at loxP sites resident on a mammalian chromosome, a mouse cell line was constructed containing two directly repeated loxP sites flanking a 2.5 kb yeast DNA fragment and inserted between the SV40 promoter and the neo structural gene to disrupt expression of the neo gene. Expression of the cre gene in this cell line results in excision of the intervening yeast DNA and thus permits sufficient expression of the neo gene to allow cell growth in high concentrations of G418. Southern analysis indicated that Cre-mediated excision occurred at the loxP sites. In the absence of the cre gene such excisive events are quite rare. Cre-mediated recombination should thus be quite useful in effecting a variety of genomic rearrangements in eukaryotic cells.

Positive genetic selection for gene disruption in mammalian cells by homologous recombination

Efficient modification of genes in mammalian cells by homologous recombination has not been possible because of the high frequency of nonhomologous recombination. An efficient method for targeted gene disruption has been developed. Cells with substitution of exogenous sequences into a chromosomal locus were enriched, by a factor of 100, using a positive genetic selection that specifically selects for homologous recombination at the targeted site. The selection is based on the conditional expression of a dominant selectable marker by virtue of in-frame gene fusion with the target gene. The dominant selectable marker was derived by modification of the Escherichia coli neo gene so that it retains significant activity in mammalian cells after in-frame fusion with heterologous coding sequences. In the example presented here, homologous recombinants were efficiently recovered from a pool in which the targeted gene was disrupted in 1 per 10,000 cells incorporating exogenous DNA.

Compilation of DNA strand exchange sites for non-homologous recombination in somatic cells

DNA sequences of 496 somatic cell illegitimate crossing over regions were compiled and analyzed. Sites for non-homologous recombination on linear DNAs transfected into mammalian cells (Transfected Linear DNAs; TLD) were analyzed separately from the remaining illegitimate recombination regions (IRR). Trinucleotides that are preferentially cleaved by rat liver topoisomerase I in vitro (CAT, CTY, GTY, RAT where R = purine, Y = pyrimidine) were present in the 10 base pair (bp) vicinity of the cross-over sites in 92% of IRR and 93% of TLD. Multiple repeats of these trinucleotides have been observed in 39% of IRR and 38% of TLD. Runs of five or more contiguous purines (or pyrimidines on the complementary strand) were found in 26% of IRR and 14% of TLD. Adenine-Thymine rich regions of five or more bases were found in 14% of IRR and 21% of TLD. Alternating purine-pyrimidine tracks longer than four nucleotides in length were found in 11% of IRR, though only in 4% of TLD. I discuss the possible biological significance of these findings and present an appendix containing the sequences in the 10 bp vicinity of the non-homologous recombination sites analyzed.

Joining of nonhomologous DNA double strand breaks in vitro

Extracts of Xenopus laevis eggs can efficiently join ends of duplex DNA that differ in structure and sequence. This was analysed by recircularisation of linear plasmid DNA molecules with dissimilar termini, generated by successive cuts with two different restriction enzymes within the pSP65 polylinker. Use of various enzymes provided blunt ended or 4 nucleotides long 3' and 5' protruding single strand (PSS) termini which were successfully joined in vitro in any tested combination. Sequence analysis of numerous junctions from cloned reaction products of 7 terminus combinations reveal: apart from very rare base exchanges and single nucleotide insertions less than 10% deletions (1 to 18 nucleotides long) were detected. Blunt/PSS or 3'PSS/5'PSS terminus pairs undergo simple "blunt end" joining which preserves PSS ends by fill-in. In contrast, equally polar 3'PSS/3'PSS or 5'PSS/5'PSS terminus pairs are joined by a complex mode: PSS ends overlap by a defined number of nucleotides, set by matching basepairs. Even one basematch suffices to define the setting. This then determines the final mismatch repair and fill-in pattern. We propose that yet unknown terminal DNA-binding proteins stabilize the energetically highly unfavorable configuration of single matching basepairs and help to support defined overlap structures.

Homologous recombination of SV40 DNA in COS7 cells occurs with high frequency in a gene dose independent fashion

Homologous recombination between microinjected SV40 DNA fragments and endogenous SV40 DNA in COS7 cells was analysed by immunofluorescence staining and DNA blotting. Time course experiments revealed that recombination between the transferred (trans) DNA and the chromosomal DNA occurred about 8 hours after microinjection with high efficiency in a gene dose independent fashion. Deletions of up to 1018 basepairs (bp) within the early or the late SV40 region were efficiently repaired after the transfer of linear but not of circular DNA molecules. A 22 bp homology between the trans DNA and the endogenous DNA was sufficient to initiate recombination but 14 nonhomologous bp at one open end of the SV40 DNA fragments hindered gap repair.

Intermolecular homologous recombination between transfected sequences in mammalian cells is primarily nonconservative

Intermolecular recombination in mammalian cells was studied by coinfecting African green monkey cells in culture with two shuttle vector plasmids, each carrying an incomplete but overlapping portion of the gene for neomycin resistance. The region of homology between the two plasmids was about 0.6 kilobases. Recombination between the homology regions could reconstruct the neomycin resistance gene, which was monitored by analysis of progeny plasmids in bacteria. The individual plasmids carried additional markers which, in combination with restriction analysis, allowed the determination of the frequency of formation of the heterodimeric plasmid which would be formed in a conservative recombination reaction between the homologous sequences. Reconstruction of the neomycin resistance gene was readily observed, but only 1 to 2% of the neomycin resistance plasmids had the structure of the conservative heterodimer. Treatment of the plasmids which enhanced the frequency of the neomycin resistance gene reconstruction reaction did not significantly increase the relative frequency of conservative product plasmids. The results support nonconservative models for recombination of these sequences.

Intermolecular homologous recombination between transfected sequences in mammalian cells is primarily nonconservative

Intermolecular recombination in mammalian cells was studied by coinfecting African green monkey cells in culture with two shuttle vector plasmids, each carrying an incomplete but overlapping portion of the gene for neomycin resistance. The region of homology between the two plasmids was about 0.6 kilobases. Recombination between the homology regions could reconstruct the neomycin resistance gene, which was monitored by analysis of progeny plasmids in bacteria. The individual plasmids carried additional markers which, in combination with restriction analysis, allowed the determination of the frequency of formation of the heterodimeric plasmid which would be formed in a conservative recombination reaction between the homologous sequences. Reconstruction of the neomycin resistance gene was readily observed, but only 1 to 2% of the neomycin resistance plasmids had the structure of the conservative heterodimer. Treatment of the plasmids which enhanced the frequency of the neomycin resistance gene reconstruction reaction did not significantly increase the relative frequency of conservative product plasmids. The results support nonconservative models for recombination of these sequences.

Intermolecular recombination assay for mammalian cells that produces recombinants carrying both homologous and nonhomologous junctions

We present an intermolecular recombination assay for mammalian cells that does not involve the reconstitution of a selectable marker. It is based on the generation of a shuttle vector by recombination between a bacterial and a mammalian vector. The recombinants can thus be amplified in mammalian cells, isolated by plasmid rescue in an Escherichia coli RecA- host, and identified by in situ hybridization, by using mammalian vector sequences as probes. Since both parental molecules can share defined lengths of homology, this assay permits a direct comparison between homologous and nonhomologous intermolecular recombination. Our results indicate that the dominant intermolecular recombination mechanism is a nonhomologous one. The relative frequency of homologous to nonhomologous recombination was influenced by the length of shared homology between parental molecules and the replicative state of the parental molecules, but not by the introduction of double-strand breaks per se. Finally, almost all of the recombinants with a homologous junction did not have the reciprocal homologous junction but instead had a nonhomologous one. We propose a model to account for the generation of these recombinants.

Homologous recombination between coinjected DNA sequences peaks in early to mid-S phase

We have examined the effect of cell cycle position on homologous recombination between plasmid molecules coinjected into synchronized rat fibroblasts. Recombination activity was found to be low in G1 and to rise 10- to 15-fold, peaking in early to mid-S phase.

Gene recombination in X-ray-sensitive hamster cells

Recombination was measured in Chinese hamster ovary (CHO-K1) cells and in the X-ray-sensitive mutants xrs1 and xrs7, which show a defect in DNA double-strand break repair. To assay recombination, pairs of derivatives of the plasmid pSV2gpt were constructed with nonoverlapping deletions in the gpt gene region and cotransferred into the different cell types. Recombination efficiencies, measured as the transformation frequency with a pair of deletion plasmids relative to that with the complete pSV2gpt plasmid, were about 6% in both CHO-K1 and the xrs mutants for plasmids linearized at a site outside the gpt gene. However, these efficiencies were substantially enhanced by the introduction of a double-strand break into the homologous region of the gpt gene in one of a pair of deletion plasmids before cotransfer. This enhancement was apparently only about half as great for the xrs cells as for CHO-K1, but variation in the data was considerable. A much larger difference between CHO-K1 and the xrs mutants was found when the DNA concentration dependence of transformation was explored. While the transformation frequency of CHO-K1 increased linearly with DNA concentration, no such increase occurred with the xrs mutants irrespective of whether complete plasmids or pairs of deletion plasmids were transferred. The fraction of cells taking up DNA, assayed autoradiographically, was similar in all cell types. Therefore we suggest that while homologous recombination of plasmid molecules may not be substantially reduced in the xrs mutants,processes involved in the stable integration of plasmid DNA into genomic DNA are significantly impaired.

Homologous recombination between coinjected DNA sequences peaks in early to mid-S phase

We have examined the effect of cell cycle position on homologous recombination between plasmid molecules coinjected into synchronized rat fibroblasts. Recombination activity was found to be low in G1 and to rise 10- to 15-fold, peaking in early to mid-S phase.

Intermolecular recombination assay for mammalian cells that produces recombinants carrying both homologous and nonhomologous junctions

We present an intermolecular recombination assay for mammalian cells that does not involve the reconstitution of a selectable marker. It is based on the generation of a shuttle vector by recombination between a bacterial and a mammalian vector. The recombinants can thus be amplified in mammalian cells, isolated by plasmid rescue in an Escherichia coli RecA- host, and identified by in situ hybridization, by using mammalian vector sequences as probes. Since both parental molecules can share defined lengths of homology, this assay permits a direct comparison between homologous and nonhomologous intermolecular recombination. Our results indicate that the dominant intermolecular recombination mechanism is a nonhomologous one. The relative frequency of homologous to nonhomologous recombination was influenced by the length of shared homology between parental molecules and the replicative state of the parental molecules, but not by the introduction of double-strand breaks per se. Finally, almost all of the recombinants with a homologous junction did not have the reciprocal homologous junction but instead had a nonhomologous one. We propose a model to account for the generation of these recombinants.

Reconstruction of human alpha thalassemia-2 genotypes in monkey cells

The human adult alpha globin genes, alpha 2 and alpha 1, are contained within two tandemly arranged duplication units. Each unit spans 4 kb of DNA, and contains three homology blocks (X, Y, Z) separated by non-homologous sequences. Segmental DNA recombination processes between the two units have resulted in high frequencies of two types of deletions in certain human populations, each deletion removing one alpha globin gene from chromosome 16, (alpha-thalassemia 2). In order to study the molecular mechanisms of alpha-thalassemia 2, and of homologous DNA recombination in general in mammalian cells, we have reconstructed these two alpha-thalassemia 2 genotypes in monkey cells. The two duplication units have been cloned in an SV40 origin-containing vector, and transfected into COS 7 cells. Newly replicated plasmid DNA was isolated and analyzed by Southern blot hybridization. Homologous DNA recombination occurs with high frequencies (10-20% per kb of homology), and this generates both types of alpha-thalassemia 2 deletions on the episomes in the monkey cells. Removal of the 5' end of either one, or both, of the X blocks prior to DNA transfection affects the relative frequencies of the two alpha-thalassemia 2 genotypes in a novel way. We consider and discuss these results in terms of several alternative models. Our data suggest the existence of hot spot(s) for initiation of homologous DNA recombination, or recombination promoting element(s), in a specific region of the human adult alpha globin locus. A DNA sequence that defines the boundaries of the two duplication units, and has been implicated in the initiation of gene conversion of the two X blocks, is contained within this region.

Gene recombination in X-ray-sensitive hamster cells

Recombination was measured in Chinese hamster ovary (CHO-K1) cells and in the X-ray-sensitive mutants xrs1 and xrs7, which show a defect in DNA double-strand break repair. To assay recombination, pairs of derivatives of the plasmid pSV2gpt were constructed with nonoverlapping deletions in the gpt gene region and cotransferred into the different cell types. Recombination efficiencies, measured as the transformation frequency with a pair of deletion plasmids relative to that with the complete pSV2gpt plasmid, were about 6% in both CHO-K1 and the xrs mutants for plasmids linearized at a site outside the gpt gene. However, these efficiencies were substantially enhanced by the introduction of a double-strand break into the homologous region of the gpt gene in one of a pair of deletion plasmids before cotransfer. This enhancement was apparently only about half as great for the xrs cells as for CHO-K1, but variation in the data was considerable. A much larger difference between CHO-K1 and the xrs mutants was found when the DNA concentration dependence of transformation was explored. While the transformation frequency of CHO-K1 increased linearly with DNA concentration, no such increase occurred with the xrs mutants irrespective of whether complete plasmids or pairs of deletion plasmids were transferred. The fraction of cells taking up DNA, assayed autoradiographically, was similar in all cell types. Therefore we suggest that while homologous recombination of plasmid molecules may not be substantially reduced in the xrs mutants,processes involved in the stable integration of plasmid DNA into genomic DNA are significantly impaired.

Resolution of a polyomavirus-mouse hybrid replicon: viral function required for recombination

RmI, a circular chimera made of the polyomavirus (Py) genome with an insertion of mouse DNA (Ins), effectively undergoes intramolecular recombination in normal mouse cells, as indicated by the conversion of cloned RmI (RmIc) into unit-length Py DNA in transfected cultures. To follow the fate of the cellular component of RmI after recombination, the origin of simian virus 40 (SV40) DNA was inserted into the Ins region of RmIc, generating a new molecular species designated SV-RmIc. The recombination of SV-RmIc in simian cells synthesizing SV40 large T antigen gave rise to a molecule containing the SV40 origin, the reciprocal of unit-length Py DNA. However, SV-RmIc failed to yield unit-length Py DNA in murine cells unless Py large T antigen was provided in trans. In murine cells synthesizing SV40 large T antigen, the only detectable product from SV-RmIc contained only Py sequences, but was heterogeneous in size. These results and others also reported here strongly suggest that Py large T antigen plays a direct role in the resolution of RmI in murine cells.

Resolution of a polyomavirus-mouse hybrid replicon: release of genomic viral DNA

RmI is a circular chimera containing 1.03 copies of polyomavirus DNA and 1,628 base pairs of mouse DNA, joined through direct and inverted repeat sequences. It is excised from the chromosome of a transformed cell via a site-specific recombination event that is dependent on the activation of the viral gene coding for large T antigen. RmI is shown here to be highly infectious for normal mouse cells. This infectivity reflects the ability of RmI to effectively yield unit-length viral DNA via intramolecular recombination. The effectiveness with which infectious viral DNA is produced from RmI is consistent with the idea that the underlying recombination event is site specific, rather than homologous or illegitimate.

Effects of poly[d(pGpT).d(pApC)] and poly[d(pCpG).d(pCpG)] repeats on homologous recombination in somatic cells

Sequencing studies have shown that in somatic cells alternating runs of purines and pyrimidines are frequently associated with recombination crossover points. To test whether such sequences actually promote recombination, we have examined the effects of poly[d(pGpT).d(pApC)] and poly[d(pCpG).d(pCpG)] repeats on a homologous recombination event. The parental molecule used in this study, pSVLD, is capable of generating wild-type simian virus 40 DNA via recombination across two 751-base-pair regions of homology and has been described previously (Miller et al., Proc. Natl. Acad. Sci. USA 81:7534-7538, 1984). Single inserts of either a poly[d(pGpT).d(pApC)] repeat or a poly[d(pCpG).d(pCpG)] repeat were positioned adjacent to one region of homology in such a way that the recombination product, wild-type simian virus 40 DNA, could be formed only by recombination within the homologies and not by recombination across the alternating purine-pyrimidine repeats. We have found that upon transfection of test DNAs into simian cells, a poly[d(pCpG).d(pCpG)] repeat enhanced homologous recombination 10- to 15-fold, whereas a poly[d(pGpT).d(pApC)] repeat had less effect. These results are discussed in terms of the features of these repeats that might be responsible for promoting homologous recombination.

Recombination of DNAs in Xenopus oocytes based on short homologous overlaps

Linear molecules of pBR322 and closely related plasmid DNAs were injected into Xenopus oocyte nuclei. Such molecules were degraded unless their ends were recombined. Non-homologous ends were joined rarely, if at all, but measurable recombination was supported by homologous sequences of less than 10 base pairs (bp). The efficiency of recombination increased as the length and degree of homology improved, in the range of about 8-20 bp. The homologous sequences had to be very close to the original molecular ends (within about 20 bp); internal homologies, even when they included better matches, were never used. These observations are best accommodated by a model of recombination which envisions exonucleolytic resection to expose homologous sequences, followed by annealing of single-stranded tails, tidying up and sealing of the new joint. Some of the recombined plasmids had novel tetracycline resistance genes; their properties give some insight into the function of the tet gene product.

Transcriptional activities of mammalian genomes at sites of recombination with foreign DNA

The nucleotide sequences of several sites of recombination between adenovirus DNA and hamster, mouse, or human cell DNAs were determined. These sites of recombination had been cloned from adenovirus type 2 (Ad2)- or type 12 (Ad12)-transformed cells, from Ad12-induced tumor cells, or from a symmetric recombinant between Ad12 DNA and human cell DNA. One important precondition for the generation of recombinants between host and foreign DNAs might be the establishment of a chromatin configuration that permits access of foreign DNA and of the recombination machinery to cellular DNA. Such favorable chromatin structures might arise during cellular DNA replication or transcription or both. As a first approach toward investigating these more complex problems of foreign DNA insertion, we determined transcriptional activities of cellular DNA sequences at viral junction sites. The sites of linkage investigated in this study with respect to their transcriptional activities were those previously cloned and sequenced (W. Doerfler, R. Gahlmann, S. Stabel, R. Deuring, U. Lichtenberg, M. Schulz, D. Eick, and R. Leisten, Curr. Top. Microbiol. Immunol. 109:193-228, 1983). In addition, a site from cell line HA12/7 which is described in this paper was also analyzed. The results presented demonstrate that the cellular DNA sequences involved in linkage to viral DNA at five completely different sites in DNA from three different species are transcribed into RNAs even in cells which have not been transformed or infected by adenovirus. Some of these RNAs were cytoplasmic and were not poly(A)+. Human cell DNA sequences at the junction to Ad12 DNA in SYREC2 DNA were transcribed into poly(A)+ cytoplasmic RNA which could be translated in vitro. These results are consistent with the notion that at least some of the cellular DNA sequences at sites of insertion of adenovirus (foreign) DNA are transcriptionally active and thus provide an opportunity for recombination.

Intramolecular recombination between transfected repeated sequences in mammalian cells is nonconservative

When plasmids carrying a fragmented gene with segments present as direct repeats are introduced into mammalian cells, recombination or gene conversion between the repeated sequences can reconstruct the gene. Intramolecular recombination leads to the deletion of the intervening sequences and the loss of one copy of the repeat. This process is known to be stimulated by double-strand breaks. Two current models for recombination in eucaryotic cells propose that the reaction is initiated by double-strand breaks, but differ in their predictions as to the fate of the intervening sequences. One model suggests that these sequences are always lost, while the other indicates that the reaction will be conservative as a function of the position of the double-strand break. We have constructed a plasmid in which two overlapping portions of the simian virus 40 early region, which contains the origin and T-antigen gene, are present as direct repeats separated by sequences containing a plasmid with a simian virus 40 origin of replication. Recombination across the repeated segments could produce a plasmid with an origin of replication and/or a plasmid with a gene for a functional T-antigen which would drive the replication of both. Introduction of this construction into African green monkey kidney cells, without coinfection, establishes a condition in which the products of the recombination or gene conversion can be interpreted unambiguously. We find that the majority of the reconstruction reactions are nonconservative.

Efficient homologous recombination of linear DNA substrates after injection into Xenopus laevis oocytes

When DNA molecules are injected into Xenopus oocyte nuclei, they can recombine with each other. With bacteriophage lambda DNAs, it was shown that this recombination is stimulated greatly by introduction of double-strand breaks into the substrates and is dependent on homologous overlaps in the recombination interval. With plasmid DNAs it was shown that little or no recombination occurs between circular molecules but both intra- and intermolecular events take place very efficiently with linear molecules. As with the lambda substrates, homology was required to support recombination; no simple joining of ends was observed. Blockage of DNA ends with nonhomologous sequences interfered with recombination, indicating that ends are used directly to initiate homologous interactions. These observations are combined to evaluate possible models of recombination in the oocytes. Because each oocyte is capable of recombining nanogram quantities of linear DNA, this system offers exceptional opportunities for detailed molecular analysis of the recombination process in a higher organism.

Effects of poly[d(pGpT).d(pApC)] and poly[d(pCpG).d(pCpG)] repeats on homologous recombination in somatic cells

Sequencing studies have shown that in somatic cells alternating runs of purines and pyrimidines are frequently associated with recombination crossover points. To test whether such sequences actually promote recombination, we have examined the effects of poly[d(pGpT).d(pApC)] and poly[d(pCpG).d(pCpG)] repeats on a homologous recombination event. The parental molecule used in this study, pSVLD, is capable of generating wild-type simian virus 40 DNA via recombination across two 751-base-pair regions of homology and has been described previously (Miller et al., Proc. Natl. Acad. Sci. USA 81:7534-7538, 1984). Single inserts of either a poly[d(pGpT).d(pApC)] repeat or a poly[d(pCpG).d(pCpG)] repeat were positioned adjacent to one region of homology in such a way that the recombination product, wild-type simian virus 40 DNA, could be formed only by recombination within the homologies and not by recombination across the alternating purine-pyrimidine repeats. We have found that upon transfection of test DNAs into simian cells, a poly[d(pCpG).d(pCpG)] repeat enhanced homologous recombination 10- to 15-fold, whereas a poly[d(pGpT).d(pApC)] repeat had less effect. These results are discussed in terms of the features of these repeats that might be responsible for promoting homologous recombination.

Topological requirements for homologous recombination among DNA molecules transfected into mammalian cells

Cultured animal cells rearrange foreign DNA very efficiently by homologous recombination. The individual steps that constitute the mechanism(s) of homologous recombination in transfected DNA are as yet undefined. In this study, we examined the topological requirements by using the genome of simian virus 40 (SV40) as a probe. By assaying homologous recombination between defective SV40 genomes after transfection into CV1 monkey cells, we showed that linear molecules are preferred substrates for homologous exchanges, exchanges are distributed around the SV40 genome, and the frequency of exchange is not diminished significantly by the presence of short stretches of non-SV40 DNA at the ends. These observations are considered in relation to current models of homologous recombination in mammalian cells, and a new model is proposed. The function of somatic cell recombination is discussed.

Recombination and deletion of sequences in shuttle vector plasmids in mammalian cells

Shuttle vector plasmids were constructed with directly repeated sequences flanking a marker gene. African green monkey kidney (AGMK) cells were infected with the constructions, and after a period of replication, the progeny plasmids were recovered and introduced into bacteria. Those colonies with plasmids that had lost the marker gene were identified, and the individual plasmids were purified and characterized by restriction enzyme digestion. Recombination between the repeated elements generated a plasmid with a precise deletion and a characteristic restriction pattern, which distinguished the recombined molecules from those with other defects in the marker gene. Recombination among the following different sequences was measured in this assay: (i) the simian virus 40 origin and enhancer region, (ii) the AGMK Alu sequence, and (iii) a sequence from plasmid pBR322. Similar frequencies of recombination among these sequences were found. Recombination occurred more frequently in Cos1 cells than in CV1 cells. In these experiments, the plasmid population with defective marker genes consisted of the recombined molecules and of the spontaneous deletion-insertion mutants described earlier. The frequency of the latter class was unaffected by the presence of the option for recombination represented by the direct repeats. Both recombination and deletion-insertion mutagenesis were stimulated by double-strand cleavage between the repeated sequences and adjacent to the marker, and the frequency of the deletion-insertion mutants in this experiment was again independent of the presence of the direct repeats. We concluded that although recombination and deletion-insertion mutagenesis were both stimulated by double-strand cleavage, the molecules which underwent the two types of change were drawn from separate pools.

Mechanisms of nonhomologous recombination in mammalian cells

The primary mechanism of nonhomologous recombination in transfected DNA involves breakage followed by end joining. To probe the joining step in more detail, linear simian virus 40 genomes with mismatched ends were transfected into cultured monkey cells, and individual viable recombinants were analyzed. The transfected genomes carried mismatched ends as a result of cleavage with two restriction enzymes, the recognition sites of which are located in the intron of the gene encoding the T antigen. Because the T antigen gene was split by this cleavage, the transfected genomes were inert until activated by cell-mediated end joining. Clonal descendants of the original recombinants were isolated from 122 plaques and were grouped into four classes based on the electrophoretic mobility of the junction fragment. The structures of representative junctions were determined by nucleotide sequencing. The spectrum of nonhomologous junctions analyzed here along with a large number of previously reported junctions suggest that there are two mechanisms for the linkage of DNA molecules: (i) direct ligation of ends and (ii) repair synthesis primed by terminal homologies of a few nucleotides. A paired-priming model of nonhomologous recombination is discussed.

Intramolecular recombination between transfected repeated sequences in mammalian cells is nonconservative

When plasmids carrying a fragmented gene with segments present as direct repeats are introduced into mammalian cells, recombination or gene conversion between the repeated sequences can reconstruct the gene. Intramolecular recombination leads to the deletion of the intervening sequences and the loss of one copy of the repeat. This process is known to be stimulated by double-strand breaks. Two current models for recombination in eucaryotic cells propose that the reaction is initiated by double-strand breaks, but differ in their predictions as to the fate of the intervening sequences. One model suggests that these sequences are always lost, while the other indicates that the reaction will be conservative as a function of the position of the double-strand break. We have constructed a plasmid in which two overlapping portions of the simian virus 40 early region, which contains the origin and T-antigen gene, are present as direct repeats separated by sequences containing a plasmid with a simian virus 40 origin of replication. Recombination across the repeated segments could produce a plasmid with an origin of replication and/or a plasmid with a gene for a functional T-antigen which would drive the replication of both. Introduction of this construction into African green monkey kidney cells, without coinfection, establishes a condition in which the products of the recombination or gene conversion can be interpreted unambiguously. We find that the majority of the reconstruction reactions are nonconservative.

Efficient homologous recombination of linear DNA substrates after injection into Xenopus laevis oocytes

When DNA molecules are injected into Xenopus oocyte nuclei, they can recombine with each other. With bacteriophage lambda DNAs, it was shown that this recombination is stimulated greatly by introduction of double-strand breaks into the substrates and is dependent on homologous overlaps in the recombination interval. With plasmid DNAs it was shown that little or no recombination occurs between circular molecules but both intra- and intermolecular events take place very efficiently with linear molecules. As with the lambda substrates, homology was required to support recombination; no simple joining of ends was observed. Blockage of DNA ends with nonhomologous sequences interfered with recombination, indicating that ends are used directly to initiate homologous interactions. These observations are combined to evaluate possible models of recombination in the oocytes. Because each oocyte is capable of recombining nanogram quantities of linear DNA, this system offers exceptional opportunities for detailed molecular analysis of the recombination process in a higher organism.

Double-strand gap repair results in homologous recombination in mouse L cells

Previous studies have demonstrated that the presence of double-strand breaks or double-strand gaps increases the frequency of homologous recombination between two cotransferred DNAs when they are introduced into cultured mammalian cells. Here we demonstrate that the repair of these double-strand gaps is a major mechanism for homologous recombination between exogenous DNAs. In particular, when a plasmid DNA containing a 104-base-pair (bp) gap in its tk gene (herpes simplex virus gene for thymidine kinase) undergoes recombination in mouse L cells to generate an intact gene, the majority of events result from direct repair of the double-strand gap using a cotransferred DNA as the template. We analyzed the recombination events by comparing the frequency of tk+ colonies, Southern blotting of cloned tk+ cell lines, and cloning recombined functional tk genes by plasmid rescue. In addition, by creating double-strand breaks within or adjacent to heterologous insertions in a mutant tk gene, we estimate that the L cell can generate a double-strand gap of between 152 and 248 bp and then can repair the gap to create a functional tk gene. We conclude that double-strand breaks and double-strand gaps are recombinogenic in transferred plasmid DNAs because they serve as intermediates in homologous recombination by double-strand gap repair, a nonreciprocal exchange of DNA or gene conversion event.

Mechanisms of nonhomologous recombination in mammalian cells

The primary mechanism of nonhomologous recombination in transfected DNA involves breakage followed by end joining. To probe the joining step in more detail, linear simian virus 40 genomes with mismatched ends were transfected into cultured monkey cells, and individual viable recombinants were analyzed. The transfected genomes carried mismatched ends as a result of cleavage with two restriction enzymes, the recognition sites of which are located in the intron of the gene encoding the T antigen. Because the T antigen gene was split by this cleavage, the transfected genomes were inert until activated by cell-mediated end joining. Clonal descendants of the original recombinants were isolated from 122 plaques and were grouped into four classes based on the electrophoretic mobility of the junction fragment. The structures of representative junctions were determined by nucleotide sequencing. The spectrum of nonhomologous junctions analyzed here along with a large number of previously reported junctions suggest that there are two mechanisms for the linkage of DNA molecules: (i) direct ligation of ends and (ii) repair synthesis primed by terminal homologies of a few nucleotides. A paired-priming model of nonhomologous recombination is discussed.

Recombination between poly[d(GT).d(CA)] sequences in simian virus 40-infected cultured cells

CVI cells were transfected with oversized simian virus 40 (SV40) genomes that could be reduced to packageable size by alternative homologous recombination pathways involving either two polydeoxyguanylic-thymidylic acid X polydeoxycytidylic-adenylic acid (poly[d(GT).d(CA)]; abbreviated hereafter as poly(GT)] tracts or two tracts of homologous SV40 sequence. Plaque-forming viruses rescued by this procedure were found to contain genomes formed by homologous and nonhomologous recombination events. Half of the viable viral DNA molecules recovered were the result of recombination between two tracts of poly(GT). Approximately 20% of the rescued viral genomes were produced by homologous recombination between tracts of SV40 DNA. Nonhomologous recombination involving SV40 sequences was also a major pathway of deletion, producing ca. 30% of the viral plaques. Tracts of poly(GT) generated by recombination were variable in length, suggesting that recombination between poly(GT) tracts was usually unequal. On a per-nucleotide basis, poly(GT) recombination occurred eight times more frequently than did recombination between homologous SV40 DNA. This eightfold difference is the maximum recombinatory enhancement attributable to poly(GT) sequences. Although DNA sequence analysis showed that tracts of poly(GT) generated by recombination retained the alternating G-T repeat motif throughout their length, the contribution of the nonhomologous pathway to poly(GT) recombination cannot be ruled out, and the relative proclivity of a given length of d(GT).d(CA) sequence to undergo homologous recombination is probably less than eight times greater than that of an SV40 sequence of the same length.

Recombination and deletion of sequences in shuttle vector plasmids in mammalian cells

Shuttle vector plasmids were constructed with directly repeated sequences flanking a marker gene. African green monkey kidney (AGMK) cells were infected with the constructions, and after a period of replication, the progeny plasmids were recovered and introduced into bacteria. Those colonies with plasmids that had lost the marker gene were identified, and the individual plasmids were purified and characterized by restriction enzyme digestion. Recombination between the repeated elements generated a plasmid with a precise deletion and a characteristic restriction pattern, which distinguished the recombined molecules from those with other defects in the marker gene. Recombination among the following different sequences was measured in this assay: (i) the simian virus 40 origin and enhancer region, (ii) the AGMK Alu sequence, and (iii) a sequence from plasmid pBR322. Similar frequencies of recombination among these sequences were found. Recombination occurred more frequently in Cos1 cells than in CV1 cells. In these experiments, the plasmid population with defective marker genes consisted of the recombined molecules and of the spontaneous deletion-insertion mutants described earlier. The frequency of the latter class was unaffected by the presence of the option for recombination represented by the direct repeats. Both recombination and deletion-insertion mutagenesis were stimulated by double-strand cleavage between the repeated sequences and adjacent to the marker, and the frequency of the deletion-insertion mutants in this experiment was again independent of the presence of the direct repeats. We concluded that although recombination and deletion-insertion mutagenesis were both stimulated by double-strand cleavage, the molecules which underwent the two types of change were drawn from separate pools.

Topological requirements for homologous recombination among DNA molecules transfected into mammalian cells

Cultured animal cells rearrange foreign DNA very efficiently by homologous recombination. The individual steps that constitute the mechanism(s) of homologous recombination in transfected DNA are as yet undefined. In this study, we examined the topological requirements by using the genome of simian virus 40 (SV40) as a probe. By assaying homologous recombination between defective SV40 genomes after transfection into CV1 monkey cells, we showed that linear molecules are preferred substrates for homologous exchanges, exchanges are distributed around the SV40 genome, and the frequency of exchange is not diminished significantly by the presence of short stretches of non-SV40 DNA at the ends. These observations are considered in relation to current models of homologous recombination in mammalian cells, and a new model is proposed. The function of somatic cell recombination is discussed.