Capture of a Recombination Activating Sequence from Mammalian Cells

Home , Chi site, Gene conversion, RecBCD

P Olson and R Dornburg The Dorrance H Hamilton Laboratories, Center for Human Virology, Division of Infectious Diseases, Jefferson Medical College, Thomas Jefferson University, Jefferson Alumni Hall, 1020 Locust Street, Rm 329, Philadelphia, PA 19107, USA

We have developed a genetic trap for identifying revealed a putative recombination signal (CCCACCC). sequences that promote homologous DNA recombination. When this heptamer or an abbreviated form (CCCACC) The trap employs a retroviral vector that normally disables were reinserted into the vector, they stimulated vector itself after one round of replication. Insertion of deﬁned repair and other DNA rearrangements. Mutant forms of DNA sequences into the vector induced the repair of a 300 these oligomers (eg CCCAACC or CCWACWS) did not. base pair deletion, which restored its ability to replicate. Our data suggest that the recombination events occurred Tests of random sequence libraries made in the vector within 48 h after transfection.

Keywords: recombination; retroviral vector; vector stability; gene conversion; gene therapy

Introduction scripts are still made from the intact left LTR, but reverse transcription copies the deletion to both LTRs, disabling Recognition of cis-acting DNA signals occupies a central the daughter provirus.15–17 We found that the SIN vectors role in both site-specific and general recombination path- that could escape this programmed disablement did so 1–6 ways. Signals in site-specific pathways define the by recombinationally repairing the U3-deleted LTR. DNA points at which the exchange will occur. General signals sequence analysis of reconstituted LTRs indicated that do not promote defined rearrangements but rather create the U3 repair was error-prone, and occasionally resulted hotspots by increasing the frequency of homologous in only partial repair of the deletion. LTR sequences in recombination in their vicinity. The best studied general the helper cell genome and in the left LTR served as tem- signal is E. coli’s crossover hotspot instigator (chi), which plate to reconstitute the LTR deletion.13 Remarkably, such 7–12 functions at the inception of the recombination. Chi recombinogenic SINs were constructed directly from acts as a biochemical switch for the RecBCD exonuclease– stable precursor vectors by sequence insertion. The two helicase complex. Beginning at a double-strand break, the stimulatory sequences studied were the neomycin resist- RecBCD complex processes exonucleolytically inward ance gene and the murine leukemia virus (MLV) encap- from the lesion site until it encounters a chi site. Chi cata- sidation sequence, psi (χ).18,19 In this study, we tested ran- lyses the release of the RecD subunit, attenuating the dom sequences for their ability to promote such complex’s exonuclease activity without affecting its hel- recombination. These investigations led to the identifi- icase activity. The region unwound after chi is coated by cation of a heptamer (CCCACCC), which is frequently RecA monomers during invading strand preparation. found near hotspots of mammalian recombination. Although hotspots exist in eukaryotes, specific signals, mostly nuclease sites, have only been demonstrated in yeast. The recombination hotspots observed in higher Results eukaryotes typically lie near regions of highly repetitive sequences and are generally observed in the context of Experimental system 3–5 meiosis. The existence of ubiquitous, general somatic To analyze the unusually high frequency of retroviral 5 recombination signals has been proposed, but nothing vector reconstitution and to identify specific sequences comparable to chi has yet been functionally identified. that trigger this process, we developed a retroviral vector We had previously seen evidence for general recombi- derived ‘recombination activator trap’ (Figure 1). This nation signals while investigating a group of self-inactiv- trap employs the retroviral vector pRD17 which self-inac- ating (SIN) retroviral vectors that were able to escape tivates after one round of retroviral replication as 13,14 their programmed inactivation efficiently. Normally, expected and the retroviral packaging line C3A2.20–22 SIN vectors are restricted to one round of replication by C3A2 cells contain multiple copies of the SNV LTR to a deletion that removes U3 from the provirus’s right long express retroviral Gag-pol and Env proteins. The pRD17 terminal repeat (LTR, Figure 1). Vector genomic tran- vector contains a selectable marker gene, which confers resistance to hygromycin B and has been used exten- sively to study cDNA gene formation by retroviruses.20 Correspondence: R Dornburg This vector has been shown to be completely stable and Received 1 February 1999; accepted 15 July 1999 no vector rearrangements have been observed. By Recombination activating sequences P Olson and R Dornburg 1820

Figure 1 Recombination activator trap. A retroviral vector system has been used to isolate recombination activating DNA sequences from random sequences cloned into a self-inactivating (SIN) retroviral vector (pRD17, top). pRD17 contains the hygromycin B phosphotransferase gene under the control of an internal promoter, allowing hygromycin selection of transfected or infected cells. In Step 1, plasmid DNAs containing pRD17 SIN vectors are transfected into the retroviral packaging line C3A2, which contains multiple copies of the SNV-LTR. Virus is harvested from such cells and fresh C3A2 helper cells are infected (Step 2). If recombination (triggered by a speciﬁc DNA recombination activator, RA) in transfected Step 1 cells reconstitutes the deletion in the right LTR, a provirus with complete LTRs will be formed in Step 2 helper cells. RNA transcribed from this provirus can be packaged and transduced to fresh target cells (Step 3). Vectors without a recombination activator will copy the deleted right LTR to both ends of the provirus and, hence, will replicate only once. Thus, selection of cells infected with virus harvested from Step 2 helper cells, can lead to the isolation of vector proviruses containing a recombination activating sequence. This RA can be recovered by PCR (Figure 2).

creating DNA libraries in the vector, we now use it as a jected to the experimental protocol (Figure 1), about 200 trap for identifying recombination signals (Figure 1, top). hygromycin-resistant colonies were detected after the The insertion of a specific recombination activator (RA) second round of retroviral replication. When a control was expected to trigger recombination leading to a vector with two full-length LTRs was subjected to this repaired vector LTR using LTR templates in the C3A2 experimental protocol, about 106 hygromycin-resistant helper cell. Thus, subjecting such vectors to two rounds colonies were observed at step 3 (data not shown). These of retroviral replication enabled the identification of vec- data suggest that only a rather small number of the ran- tor species that underwent recombination (Figure 1), dom inserts was able to trigger vector repair. Further- because vectors without RAs replicate only once. more, a smaller library containing about 500 random A retroviral vector library was constructed by inserting sequences failed to produce any retroviral vectors with size-selected (400 bases or smaller) Sau3Al-digested D17 reconstituted LTRs. Latter data indicate that the small chromosomal DNA into the BamHI site of pRD17 library did not carry recombinogenic DNA sequences. (Figure 1). The restriction enzymes Sau3Al and BamHl To determine whether vectors with reconstituted LTRs create identical 39 overhangs after cutting the target had an insert with a common sequence motif, the DNA. Thus, they can be easily ligated with each other.23 chromosomal DNAs of hygromycin-resistant colonies After transformation of the ligation products into E. coli, were isolated. Inserts from vectors with reconstituted the number and size of inserts was determined by analyz- LTRs were recovered by PCR amplification using primers ing plasmid DNAs recovered from 20 bacterial clones. specific for flanking vector sequences (Figure 2). PCR The rest of the bacterial colonies which contained more products were cloned and sequenced (Figure 2 and data than 5000 unique inserts were pooled and plasmid DNAs not shown). were isolated. When this plasmid vector library was sub- Surprisingly, three vectors’ DNAs (recovered from Recombination activating sequences P Olson and R Dornburg 1821

Figure 2 Identification of recombination activators. The retroviral vector pRD17 is shown at the top. Inserts from proviruses with reconstituted LTRs were recovered PCR amplification using primers (indicated by the arrows) specific for regions flanking the BamHI cloning site. The sequence at the top is the original plasmid sequence of pRD17. The sequences below (pRA2, pRA4 and pRA9) were recovered from DNAs isolated from Step 3 cell clones. A dot indicates sequence homology; a dash indicates a missing bp. The box indicates the inserts found in clones pRA2, pRA4 and pRA9. three separate transfections) contained only a short four nucleotide insert (CCCA) which created a symmetric sequence element, CCCACCC (Figure 2). In each case, the Table 1 Reinsertion of the CCCACCC motif into pRD17 stimu- lates LTR repair clone had apparently undergone recombination, which deleted the remainder of the insert because the BamHI Oligo Sequence Step 2 titer overhang duplication necessarily created during cloning name was missing. Furthermore, two of these three clones had several mutations in the flanking DNA regions. A similar pRD17 GATC (no insert) ,1 c.f.u./ml observation was also made previously when reconsti- RA GATCCCCACCCGATC 200 c.f.u./ml tuted LTRs were analyzed by DNA sequencing.14 Fur- RA-m1 GATCCCCAACCGATC ,1 c.f.u./ml thermore, these data demonstrate that the amplified RA-m2 GATCCCACCAGCGATC 400 c.f.u./ml DNA originated from different, individual vector DNAs. RA-m3 GATCCCAACAGCGATC ,1 c.f.u./ml , To test whether this seven base pair motif by itself was RA-m4 GATCCCAACACCGATC 1 c.f.u./ml RA-m5 GATCCCTACACCGATC ,1 c.f.u./ml sufficient to induce vector recombination, a synthetic oli- RA-m6 GATCCCTACAGCGATC ,1 c.f.u./ml gonucleotide was cloned into the BamHl site of pRD17. RA-m7 GATCCCTACTCCGATC ,1 c.f.u./ml The insertion of the CCCACCC motif into the BamHl site RA-m8 GATCCCTACTGCGATC ,1 c.f.u./ml creates the sequence GATCCCCACCCGATC (termed RA-m9 GATCCCAACTCCGATC ,1 c.f.u./ml RA, Table 1). To prove sequence specificity further, 11 RA-m10 GATCCCAACAGCGATC ,1 c.f.u./ml , mutant forms of the original heptamer were tested in par- RA-m11 GATCCCAACTGCGATC 1 c.f.u./ml allel experiments. The oligonucleotide CCCAACC (termed RA-ml), contains one base pair substitution. It The sequence motif CCCACCC, and mutant forms of this motif were inserted into the BamHI cloning site (underlined) of pRD17 creates the mutant sequence GATCCCCAACCGATC (Figure 1) and subjected to the experimental protocol outlined (Table 1). Insertion of the heptamer CCACCAGC (RA- in Figure 1. Virus titers at step 3 are shown. m2) creates the sequence GATCCCACCAGCGATC Recombination activating sequences P Olson and R Dornburg 1822 (Table 1) and represents an abbreviated version of the mid stability was analyzed in five individual bacterial CCCACCC motif. Moreover, it alters the location of the clones for each plasmid. No plasmid rearrangements putative recombination activating motif relative to the were observed (data not shown). These data confirm that GATC sequences of the BamHl cloning site (Table 1). these rearrangements occurred in the C3A2 packaging Thus, this would also give an indication, whether the cells and not in the bacteria after recovery from the GATC motif is part of the RA sequence. The oligonucleot- C3A2 cells. ides CCAACAGC (RA-m3), or CCWACWSC (W = Aor = T, S G or C) are variations of the CCACCAGC octamer. Discussion CCWACWSC represents eight different variations of the putative recombination activator (RA-m4 to RA-m11, A retroviral vector-based genetic trap has been used to Table 1). After cloning of these oligonucleotides into identify specific sequences which activate DNA recombi- pRD17 and transfection into E. coli single bacterial clones nation. This trap led to the isolation of a short motif, were isolated and plasmids were analyzed by agarose gel CCCACC(C). When reinserted into our assay system, this electrophoresis and DNA sequencing. The insertion of sequence was able to stimulate recombination, whereas such oligonucleotides did not cause plasmid rearrange- mutant forms of this sequence did not. Moreover, a small ments in transformed DH5α bacteria and the correct library of 500 random sequences did not lead to the vec- insertion of all oligonucleotides was verified by DNA tor instability indicating that only a small number of spe- sequence analysis. cific sequences is involved in this process. Plasmid DNAs Retroviral vector plasmids containing such carrying the CCCACCC motif undergo recombination oligonucleotides were again subjected to two rounds of within 42 h after transfection. Thus, recombination retroviral replication as described above. We found that mediated by this motif is independent of retroviral vectors containing the CCCACCC motif or the CCCACC replication. motif efficiently triggered recombination leading to vec- Of note, the insertion of the CCCACCC heptamer into tors with reconstituted LTRs. About 200 to 400 hygromy- the pRD17 vector and testing in the experimental system cin-resistant colonies were obtained after the second outlined in Figure 1 resulted in only 200 hygror colonies round of retroviral replication (Table 1). However, no at step 3. This titer was as high as that obtained with a hygromycin-resistant colonies were detected with vectors random sequence library, which contained only a small that contained any of the other 10 mutant forms of the number of recombination activating sequences. However, CCCACC(C) motif (Table 1). These data indicate that as shown in Figure 3, insertion of the isolated CCCACCC both the CCCACCC motif and an abbreviated form heptamer into the vector resulted in the generation of a (CCCACC) can specifically enhance DNA recombination restriction fragment length polymorphism after transfec- in mammalian D17 C3A2 cells. Furthermore, it appears tion into the helper cells, which most certainly also unlikely that sequences of the BamHI restriction enzyme destroyed the retroviral vector structure. Thus, only a recognition sites are directly involved, because we would small fraction of the rearranged plasmid was still enco- expect that the insertion of a base pair would impair the ding a complete functional retroviral vector genome, function of a specific sequence, which is most certainly which could be packaged into virions and transduce the recognized by a protein or protein complex. hygro gene into fresh target cells. This certainly accounts for the relatively low vector virus titer at step 3. The CCCACCC motif activates DNA recombination The CCCACCC sequence is also found in the 39 non- within 42 h after transfection coding region of the neomycin (G418) resistance gene One concern when studying recombination in a retroviral which triggered recombination with unusually high fre- system is whether the observed alterations are occurring quency in our system,13,14 which may explain our earlier on the RNA (during reverse transcription) or on the DNA results. Furthermore, it is located near two major translo- level (after transfection). To exclude the possibility that cation breakpoints and has been postulated to stimulate the LTR repair was occurring during reverse transcrip- recombination within certain transcribed regions.26–29 tion,24,25 vector plasmid DNAs containing the CCCACCC Moreover, recently it has been shown that this motif is motif or its mutant RA-ml form were recovered 42 h after contained within a 15 bp DNA region that binds a protein transfection. Unintegrated plasmid DNA was recovered complex involved in recombinogenic reorganization of from step 1 C3A2 packaging cells and transformed into the Epstein–Barr virus genome.30,31 Here, we show in a the recombination-deficient E. coli host strain SOLR. Indi- functional assay system that this sequence is indeed vidual clones, which represent individual recombination directly involved in recombination events. events, were selected, plasmid DNAs were isolated, sub- Sequence analysis of PCR products from recombined jected to restriction analysis and compared with their DNAs revealed gross rearrangements surrounding the parental plasmid DNA. We found that 20 out of 22 vector cloning site of the vector. For example, all vectors, which DNAs that contained the CCCACCC motif had under- only contained the CCCACCC motif (eg RA2, RA4, or gone major rearrangements, which resulted in plasmids RA9, Figure 2), lacked the remains of the second BamHI with different restriction patterns. Rearrangements were site, which was expected to flank the original sequence observed in only three of 22 vector clones that contained inserted into pRD17 (Figure 2). In addition, all clones had the mutant form (CCCAACC) of this motif (Figure 3 and either other short deletions, one base substitutions, or one data not shown). These data show that the recombination bp insertions (Figure 2). Similar observations have been is independent of retrovirus replication and further prove made earlier in our laboratory, when DNA sequences of that the CCCACCC motif is a sequence-specific DNA reconstituted LTRs in SIN vectors were analyzed.13 The recombination activator. In control experiments, SOLR number of such mutations was far too high to be bacteria were transformed directly with plasmids pRD17, accounted for by errors during reverse transcription32–34 PRD17 + CCCACCC and pRD17 + CCCAACC and plas- or during PCR amplification.35 These data indicate that Recombination activating sequences P Olson and R Dornburg 1823

Figure 3 Agarose gel electrophoresis of plasmid DNAs recovered from C3A2 helper cells 42 h after transfection and digested with the restriction enzyme RsaI. M: marker DNA; S: pRD17 plasmid standard, digested with RsaI. the recombination mechanism in D17 C3A2 cells is to the identification of the first general recombination sig- highly mutagenic. nal (CCCACC(C)). The trap also recovered a small num- It is noteworthy, that one recombined vector with a ber of other CA-rich sequences. Although most of these partially reconstituted LTR analyzed in our previous sequences were not characterized further, some of them studies had an 8 bp insertion (TGGGTGGG) near the 59 also contained the CCCACCC motif and/or resembled end of the reconstituted LTR. This insertion was located other sequences found near recombinationally active directly adjacent to the partially reconstituted U3 region regions (data not shown). and this sequence is not present anywhere in the SNV Specific recombinogenic DNA sequences such as the wild-type LTR.13 It represents the complementary motif identified here may allow the isolation and identi- sequence of the CCCACCC motif. This observation fication of proteins involved in this process. Furthermore, further indicates the involvement of this motif in they may be useful to develop gene transfer tools for the recombination. repair of defective genes or the knock-out of pathogenic All experiments described here have been performed genes by homologous recombination. In particular, recent in C3A2 retroviral packaging lines, which have been findings indicate that short hairpin RNA–DNA hybrid derived from dog D17 osteosarcoma cells. We observed oligomers can be used to repair genes.6,36–38 It would be that the recombination frequency with plasmid DNAs (as interesting to test whether the CCCACC motif would described in Figure 3, eg pRD17 + CCCACCC) was further increase the efficiency of this process. Further- higher in C3A2 cells than in the parental D17 cells (data more, the knowledge of specific recombination activators not shown). Thus, C3A2 cells, which had been kept in will help to avoid problems of vector instability of gene tissue culture since 1981, appear to express higher levels transfer vectors. of proteins involved in DNA recombination than D17 cells at a low passage number. Thus, at this point it remains unknown, whether all mammalian cells express Materials and methods similar levels of proteins triggering high efficiency recombination of DNAs, which contain the RA sequence Nomenclature described here. All vectors were derived from spleen necrosis virus To the best of our knowledge, the experimental system (SNV), a member of the avian reticuloendotheliosis virus described here is the first functional assay system to (REV) family.15 Plasmid names are prefixed with the let- identify and test specific recombinogenic DNA sequences ter p (eg pRD18) to distinguish them from virus (eg in a mammalian system. Similar systems may also be RD18) encoded by the same construct. The gene encoding applied in cell lines of other organisms. The application hygromycin B phosphotransferase is referred to as of the SNV vector system in dog cells described here led hygro.39 The antibiotic resistance phenotype it confers is Recombination activating sequences P Olson and R Dornburg 1824 Hygror. The neomycin (G418) resistance gene is neo and Sequencing its phenotype is Neor.19 All sequencing was performed using the Sequenase V2.0 (USB, Cleveland, OH, USA) dideoxy chain termination Plasmid constructions kit following the protocol recommended by the supplier. All plasmids were constructed by standard cloning pro- cedures using DSHα bacteria.23 The vector pJD214HY PCR amplification µ contains all regulatory sequences required for normal Amplifications were performed in 100 l reactions using retrovirus replication. It expresses the hygro gene from Taq DNA polymerase (Promega, Madison, WI, USA) or the left LTR promoter. Vectors of the pRD line are self- Vent DNA polymerase (New England Biolabs, Beverly, inactivating (SIN); most of the U3 region of the right LTR MA, USA) in buffers supplied by the vendor. Twenty- µ has been deleted. The selectable gene (hygro) and all other five amplification cycles were performed on 10 g of gen- non-retrovirus sequences (eg internal promoters and omic DNA. Reaction cocktails included 50 pmol of each µ internal polyadenylation signal) are in the opposite orien- primer, 400 m dNTPs, 3% BSA, 2 mm MgSO4 and 5 units tation as vector transcription. pRD33 and pRD48 were of enzyme. Cycle parameters included a 4 min pre-melt ° ° derived from pRD18, which has been described pre- at 94 C, a final extension at 72 C for 5 min, and a core ° ° ° viously.20–22 pRD33 was made by inserting the neo gene cycling program (94 C, 1 min; 56 C, 30 s; 72 C, 2 min). in the BamHI site between the vector’s two internal pro- After amplification, fragments were resolved on a 1% moters. pRD48 contains the murine leukemia virus agarose gel and GeneClean (BI0101, Gaithersburg, MD, (MLV) encapsidation sequence (PstI–BalI fragment, map USA) purified. Any additional enzyme modifications units 0.563 to 0.747) instead.20–22 Each vector’s insert is in required for cloning the amplified fragment were perfor- the same orientation as the hygro gene. The pRD17 vector, med before electrophoresis. which contains the CCCACCC motif was made in the following way: two complementary oligonucleotides Library construction were constructed which contained GATC at their 59 ends The pRD17.GEN-A and pRD17.GEN-B libraries contain (eg 59-GATCCCCACCC-39 and 59-GATCGGGTGGG-39). D17 cell genomic DNA cloned into a pRD17 vector back- Annealing of these oligonucleotides yields a short bone. D17 DNA to be used for library construction was double-stranded oligonucleotide with GATC overhangs, digested with Sau3Al and size-selected by HPLC using a which was cloned into the BamHI site of pRD17. Plasmids WATERS (Milford, MA, USA) GenPack column. Frac- containing the other oligonucleotides were made in tions containing fragments smaller than 400 bp were a similar way. Plasmids were sequenced to verify the pooled and the DNA was ligated into pRD17’s unique insertion of one oligonucleotide. BamHI site (Figure 1). Transformation was performed using E. coli DH5α cells. Cell aliquots were plated for iso- Cells lation of individual clones and determination of library complexity. Less than 500 clones were recovered from the D17 dog cells and D17 C3A2 dog helper cells were grown GEN-A library and more than 5000 from the GEN-B as previously described.13,14 The C3A2 helper cells were library. Library complexity was gauged by analyzing the derived from D17 cells and supply all retrovirus proteins plasmid DNA of 20 random clones from each library and essential for virus replication without production of repli- examining the size range of inserts by restriction map- cation-competent helper virus.15,40 Selection of Hygror ping. More than 80% of clones examined had detectable cell colonies was performed in medium containing inserts (size .50 bp) and no redundancies were observed. hygromycin B at 80 µg/ml.

Bacteria Acknowledgements All cloning experiments were performed in DSHα cells − We would like to thank Hugh Fisher for his help in pre- (RecA ). Experiments with plasmids recovered after paring this manuscript. We would like to thank Dr transfection into C3A2 cells were performed with SOLR Howard Passmore (Rutgers University) for his critical recombination-negative bacteria, which were purchased and helpful comments on this paper. We also would like from Stratagene, La Jolla, CA, USA. The phenotype of to thank Dr Ren Sun and Dr G Miller (Yale University) these bacteria is: e14− (McrA−), ∆ (mcrcb-hsdSMR- − − − − for communication before publication. This work was mrr)171, sbcC , recB , recJ , uvrC (a complete descrip- sponsored by grants from the March of Dimes Birth tion is available by the supplier). Defect Foundation and the New Jersey Commission on Cancer Research. Transfections and infections Transfections were performed by the polybrene-dimethyl sulfoxide method.41 Virus titers were determined on D17 References cells as described previously and are expressed in terms 1 Sadowski PD. Site-specific genetic recombination: hops, flips, of colony-forming units per milliliter (c.f.u./ml) of tissue and flops. FASEB J 1993; 7: 760–767. culture supernatant medium.13,14 In all experiments, 2 Alt FW et al. VDJ recombination. Immunol Today 1992; 13: 306– virus was harvested from cells after selection for drug 314. resistance. 3 Shiroishi T et al. Hotspots of homologous recombination in mouse meiosis. Adv Biophys 1995; 31: 119–132. 4 Lafuse WP. Molecular biology of murine MHC class II genes. DNA isolation Crit Rev Immunol 1991; 11: 167–194. Chromosomal DNAs were isolated by proteinase K 5 Smith GR. Hotspots of homologous recombination. Experientia digestion and phenol–chloroform extractions.23 1994; 50: 234–241. Recombination activating sequences P Olson and R Dornburg 1825 6 Cole-Strauss A, Noe A, Kmiec EB. Recombinational repair of 25 Hu W-S, Temin HM. Retroviral recombination and reverse tran- genetic mutations. Antisense Nucleic Acid Drug Dev 1997; 7: scription. Science 1990; 250: 1227–1233. 211–216. 26 Gauwerky CE et al. Evolution of B-cell malignancy: pre-B-cell 7 Cheng KC, Smith GR. Distribution of Chi-stimuated recombi- leukemia resulting from MYC activation in a B-cell neoplasm national exchanges and heteroduplex endpoints in phage with a rearranged BCL2 gene. Proc Natl Acad Sci USA 1988; 85: lambda. Genetics 1989; 123: 5–17. 8548–8552. 8 Cheng KC, Smith GR. Cutting of chi-like sequences by the 27 Seite P et al. Molecular analysis of a variant 18;22 translocation RecBCD enzyme of Escherichia coli. J Mol Biol 1987; 194: 747–750. in a case of lymphocytic lymphoma. Genes Chromosom Cancer 9 Smith GR et al. Roles of RecBC enzyme and chi sites in homolo- 1993; 6: 39–44. gous recombination. Cold Spring Harb Symp Quant Biol 1984; 49: 28 Gauwerky CE et al. Activation of MYC in a masked t(8;17) trans- 485–495. location results in an aggressive B-cell leukemia. Proc Natl Acad 10 Cheng KC, Smith GR. Recombinational hotspot activity of Chi- Sci USA 1989; 86: 8867–8870. like sequences. J Mol Biol 1984; 180: 371–377. 29 Denny CT, Hollis GF, Magrath IT, Kirsch IR. Burkitt lymphoma 11 Kowalczykowski C et al. Biochemistry of homologous recombi- cell line carrying a variant translocation creates new DNA at the nation in Escherichia coli. Microbiol Rev 1994; 58: 401–465. breakpoint and violates the hierarchy of immunoglobulin gene 12 Myers RS, Stahl FW. Chi and the RecBC D enzyme of Escherichia rearrangement. Mol Cell Biol 1985; 5: 3199–3207. coli. Annu Rev Genet 1994; 28: 49–70. 30 Spain TA, Sun R, Miller G. The locus of Epstein–Barr virus ter- 13 Olson P, Temin HM, Dornburg R. Unusually high frequency of minal repeat processing is bound with enhanced affinity by Sp1 reconstitution of long terminal repeats in U3-minus retrovirus and Sp3. Virology 1997; 237: 137–147. vectors by DNA recombination or gene conversion. J Virol 1992; 31 Sun R, Spain TA, Lin SF, Miller G. Spl binds to the precise locus 66: 1336–1343. of end processing within the terminal repeats of Epstein–Barr 14 Olson P, Nelson S, Dornburg R. Improved self-inactivating virus DNA. J Virol 1997; 71: 6136–6143. retroviral vectors derived from spleen necrosis virus. J Virol 32 Dougherty JP, Temin HM. Determination of the rate of base-pair 1994; 68: 7060–7066. substitution and insertion mutations in retrovirus replication. J 15 Dornburg R. Reticuloendotheliosis viruses and derived vectors. Virol 1988; 62: 2817–2822. 33 Dougherty JP, Temin HM. High mutation rate of a spleen Gene Therapy 1995; 2: 301–310. necrosis virus-based retrovirus vector. Mol Cell Biol 1986; 168: 16 Varmus HE, Brown P. Retroviruses. In: Berg DE, Howe MM 4387–4395. (eds). Mobile DNA. American Society for Microbiology: Wash- 34 Pathak VK, Temin HM. Broad spectrum of in vivo forward ington DC, 1988, pp 53–108. mutations, hypermutations, and mutational hotspots in a retro- 17 Temin HM. Retrovirus vectors for gene transfer: efficient inte- viral shuttle vector after a single replication cycle: deletions and gration into and expression of exogenous DNA in vertebrate cell deletions with insertions. Proc Natl Acad Sci USA 1990; 87: genomes. In: Kucherlapati R (ed). Gene Transfer. Plenum Press: 6024–6028. New York, 1986, pp 144–187. 35 Eckert KA, Kunkel TA. High fidelity DNA synthesis by the 18 Jorgensen RA, Thothstein SJ, Reznikof W. A restriction enzyme Thermus aquaticus DNA polymerase. Nucleic Acids Res 1990; 18: cleavage map of TN5 and location of a region encoding neo- 3739–3744. mycin resistance. Mol Gen Genet 1979; 177: 65–72. 36 Cole-Strauss A et al. Correction of the mutation responsible for 19 Shinnick TM, Lerner RA, Sutcliffe JG. Nucleotide sequence of sickle cell anemia by an RNA–DNA oligonucleotide (see Moloney murine leukemia virus. Nature 1981; 293: 543–548. comments). Science 1996; 273: 1386–1389. 20 Dornburg R, Temin HM. Presence of a retroviral encapsidation 37 Gruenert DC. Cystic fibrosis therapy – where we are and how sequence in nonretroviral RNA increases the efficiency of forma- we got there. West J Med 1996; 164: 361–362. tion of cDNA genes. J Virol 1990; 64: 886–889. 38 Yoon K, Cole-Strauss A, Kmiec EB. Targeted gene correction of 21 Dornburg R, Temin HM. cDNA genes formed after infection episomal DNA in mammalian cells mediated by a chimeric with retroviral vector particles lack the hallmarks of natural pro- RNA. DNA oligonucleotide. Proc Natl Acad Sci USA 1996; 93: cessed pseudogenes. Mol Cell Biol 1990; 10: 68–74. 2071–2076. 22 Dornburg R, Temin HM. Retroviral vector system for the study 39 Gritz L, Davies J. Plasmid encoded hygromycin B phospho- of cDNA gene formation. Mol Cell Biol 1988; 8: 2328–2334. transferase gene and its expression in Escherichi coli and Sac- 23 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: a Labora- charomyces cerevisiae. Gene 1983; 25: 179–188. tory Manual. Cold Spring Harbor Press: Cold Spring Harbor 40 Watanabe S, Temin HM. Construction of a helper cell line for NY, 1995. avian reticuloendotheliosis virus cloning vectors. Mol Cell Biol 24 Hu W-S, Temin HM. Genetic consequences of packaging two 1983; 3: 2241–2249. RNA genomes in one retroviral particle: pseudodiploidy and 41 Kawai S, Nishizawa M. New procedure for DNA transfection high rate of genetic recombination. Proc Natl Acad Sci USA 1990; with polycation and dimethyl sulfoxide. Mol Cell Biol 1984; 4: 87: 1556–1560. 1172–1174.