The Search for Homology Does Not Limit the Rate of Extrachromosomal Homologous Recombination in Mammalian Cells

Mouse LTK(-) cells were transfected with a pair of defective Herpes simplex virus thymidine kinase (tk) genes. One tk gene had an 8-bp insertion mutation while the second gene had a 100-bp inversion. Extrachromosomal homologous recombination leading to the reconstruction of a functional tk gene was monitored by selecting for tk positive cells using medium supplemented with hypoxanthine/aminopterin/thymidine. To assess whether the search for homology may be a rate-limiting step of recombination, we asked whether the presence of an excess number of copies of a tk gene possessing both the insertion and inversion mutations could inhibit recombination between the singly mutated tk genes. Effective competitive inhibition would require that homology searching (homologous pairing) occur rapidly and efficiently. We cotransfected plasmid constructs containing the singly mutated genes in the presence or absence of competitor sequences in various combinations of linear or circular forms. We observed effective inhibition by the competitor DNA in six of the seven combinations studied. A lack of inhibition was observed only when the insertion mutant gene was cleaved within the insertion mutation and cotransfected with the two other molecules in circular form. Additional experiments suggested that homologous interactions between two DNA sequences may compete in trans with recombination between two other sequences. We conclude that homology searching is not a rate-limiting step of extrachromosomal recombination in mammalian cells. Additionally, we speculate that a limiting factor is involved in a recombination step following homologous pairing and has a high affinity for DNA termini.     

Dependence of intrachromosomal recombination in mammalian cells on uninterrupted homology.

Recombination between a 360-base-pair (bp) segment of a wild-type thymidine kinase gene (tk) from each of three different strains (F, MP, and 101) of herpes simplex virus type one and a complete herpes simplex virus type 1 (strain F) tk gene containing an 8-bp insertion mutation was studied. The pairs of tk sequences resided as closely linked repeats within the genome of mouse LTK- cells. The frequency of recombination between sequences exhibiting 232 bp of uninterrupted homology and containing no mismatches other than the insertion mutation was comparable to the frequency of recombination between two sequences exhibiting four additional nucleotide mismatches distributed in such a way to preserve the 232-bp stretch of contiguous homology. In contrast, the placement of only two single-nucleotide mismatches (in addition to the insertion mutation) in such a manner to reduce the longest uninterrupted homology to 134 bp resulted in a 20-fold reduction in recombination. We conclude that the rate of intrachromosomal recombination in mammalian cells is determined by the amount of uninterrupted homology available and not by the total number of mismatches within a given interval of DNA. Furthermore, efficient recombination appears to require between 134 and 232 bp of uninterrupted homology; single-nucleotide heterologies are most likely sufficient to disrupt the minimal efficient recombination target. We also observed that if recombination was allowed to initiate within sequences exhibiting perfect homology, the event could propagate through and terminate within adjacent sequences exhibiting 19% base pair mismatch. We interpret this to mean that heterology exerts most of its impact on early rather than late steps of intrachromosomal recombination in mammalian cells.     

Model for homologous recombination during transfer of DNA into mouse L cells: role for DNA ends in the recombination process.

We have constructed phage lambda and plasmid DNA substrates (lambda tk2 and ptk2) that contain two defective herpesvirus thymidine kinase (tk) genes that can be used to detect homologous recombination during the transfer of DNA into mouse L cells deficient in thymidine kinase activity. The recombination event reconstructs a wild-type tk gene and is scored because it converts Tk- cells to Tk+. Using this system, we have shown that (i) both intramolecular and intermolecular homologous recombination can be detected after gene transfer; (ii) the degree of recombination decreases with decreasing tk gene homology; and (iii) the efficiency of recombination can be stimulated 10- to 100-fold by cutting the tk2 DNA with restriction enzymes at appropriate sites relative to the recombining sequences. Based on the substrate requirements for these recombination events, we propose a model to explain how recombination might occur in mammalian cells. The essential features of the model are that the cut restriction site ends are substrates for cellular exonucleases that degrade DNA strands. This process exposes complementary strands of the two defective tk genes, which then pair. Removal of unpaired DNA at the junction between the paired and unpaired regions permits a gap repair process to reconstruct an intact gene.     

Gene conversion associated with site-specific recombination in yeast plasmid pSR1.

A circular DNA plasmid, pSR1, isolated from Zygosaccharomyces rouxii has a pair of inverted repeats consisting of completely homologous 959-base pair (bp) sequences. Intramolecular recombination occurs frequently at the inverted repeats in cells of Saccharomyces cerevisiae, as well as in Z. rouxii, and is catalyzed by a protein encoded by the R gene of its own genome. The recombination is, however, independent of the RAD52 gene of the host genome. A site for initiation of the intramolecular recombination in the S. cerevisiae host was delimited into, at most, a 58-bp region in the inverted repeats by using mutant plasmids created by linker insertion. The 58-bp region contains a pair with 14-bp dyad symmetry separated by a 3-bp spacer sequence. The recombination initiated at this site was accompanied by a high frequency of gene conversion (3 to 50% of the plasmid clones examined). Heterogeneity created by the linker insertion or by a deletion (at most 153 bp so far tested) at any place on the inverted repeats was converted to a homologous combination by the gene conversion, even in the rad52-1 mutant host. A mechanism implying branch migration coupled with DNA replication is discussed.     

Repair of double-stranded DNA breaks by homologous DNA fragments during transfer of DNA into mouse L cells.

To test the validity of various models for recombination between extrachromosomal DNAs in mammalian cells, we measured recombination between a plasmid containing a herpesvirus thymidine kinase (tk) gene with an internal BamHI linker insertion mutation (ptkB8) and a tk gene deleted at both ends (tk delta 3' delta 5'). The two DNAs shared 885 base pairs of perfect tk homology except for the interruption at the linker insertion site. Recombination events that restored the mutated insertion site to wild type were monitored by the generation of hypoxanthine-aminopterine-thymidine-resistant colonies after cotransformation of Ltk- cells with the two DNAs. We found that cleavage of the ptkB8 DNA at the linker insertion site was essential for gene restoration. If the tk delta 3' delta 5' DNA was ligated into mp10 vector DNA, then recombination with the cleaved ptkB8 DNA was inefficient. In contrast, if it was excised from that vector by cleavage at flanking restriction sites, then recombination was stimulated about 150-fold. Using restriction site polymorphisms, we showed that most of the recombination events leading to restoration of the tk gene with the excised tk delta 3' delta 5' fragment involved three double-strand duplexes: two ptkB8 DNAs and one tk delta 3' delta 5' fragment. These results are much more readily explained by the single-strand annealing model of recombination than by the double-strand break repair model, and they suggest that the deficiency of the latter pathway for extrachromosomal mammalian recombination may be due, at least in part, to the obligate tripartite nature of the reaction. Finally, we measured the effect of DNA homology on the efficiency of the ptkB8-tk delta 3' delta 5' reaction. Our results showed a near-linear relationship between the efficiency of recombination and the amount of homology flanking either side of the linker insertion site. Moreover, we could detect thymidine kinase-positive transformants with as little as 10 base pairs of homology.     

Induction of recombination between homologous and diverged DNAs by double-strand gaps and breaks and role of mismatch repair.

Sequence homology is expected to influence recombination. To further understand mechanisms of recombination and the impact of reduced homology, we examined recombination during transformation between plasmid-borne DNA flanking a double-strand break (DSB) or gap and its chromosomal homolog. Previous reports have concentrated on spontaneous recombination or initiation by undefined lesions. Sequence divergence of approximately 16% reduced transformation frequencies by at least 10-fold. Gene conversion patterns associated with double-strand gap repair of episomal plasmids or with plasmid integration were analyzed by restriction endonuclease mapping and DNA sequencing. For episomal plasmids carrying homeologous DNA, at least one input end was always preserved beyond 10 bp, whereas for plasmids carrying homologous DNA, both input ends were converted beyond 80 bp in 60% of the transformants. The system allowed the recovery of transformants carrying mixtures of recombinant molecules that might arise if heteroduplex DNA--a presumed recombination intermediate--escapes mismatch repair. Gene conversion involving homologous DNAs frequently involved DNA mismatch repair, directed to a broken strand. A mutation in the PMS1 mismatch repair gene significantly increased the fraction of transformants carrying a mixture of plasmids for homologous DNAs, indicating that PMS1 can participate in DSB-initiated recombination. Since nearly all transformants involving homeologous DNAs carried a single recombinant plasmid in both Pms+ and Pms- strains, stable heteroduplex DNA appears less likely than for homologous DNAs. Regardless of homology, gene conversion does not appear to occur by nucleolytic expansion of a DSB to a gap prior to recombination. The results with homeologous DNAs are consistent with a recombinational repair model that we propose does not require the formation of stable heteroduplex DNA but instead involves other homology-dependent interactions that allow recombination-dependent DNA synthesis.     

糖多孢红霉菌同源片段长度与染色体重组率关系的研究

为了探索同源片段长度与糖多孢红霉菌染色体同源重组率的关系,化学合成或用重叠PCR合成带有突变位点、在突变位点两侧长度为（26bp+27bp）、（500bp+576bp）和（1908bp+1749bp）的同源序列,克隆于糖多孢红霉菌同源重组载体pWHM3后,分别构建了pWHM1113、 pWHM1116和 pWHM1119质粒。以PEG介导转化糖多孢红霉菌A226原生质体,3个质粒分别获得每皿30个、69个和170个转化子,但pWHM1113质粒不能与染色体有效整合,pWHM1116质粒与染色体整合率为转化子的2%,而pWHM1119质粒与染色体整合率达到转化子的19%。 pWHM1116和 pWHM1119质粒均可进行有效的染色体二次重组,将突变位位点引入染色体。因此,同源片段长度为（500bp+576bp）或更长时,可与糖多孢红霉菌染色体进行有效的单重组和双重组。     

Homology dependence of targeted recombination at the Chinese hamster APRT locus.

Using simple linear fragments of the Chinese hamster adenine phosphoribosyltransferase (APRT) gene as targeting vectors, we have investigated the homology dependence of targeted recombination at the endogenous APRT locus in Chinese hamster ovary (CHO) cells. We have examined the effects of varying either the overall length of targeting sequence homology or the length of 5' or 3' flanking homology on both the frequency of targeted homologous recombination and the types of recombination events that are obtained. We find an exponential (logarithmic) relationship between length of APRT targeting homology and the frequency of targeted recombination at the CHO APRT locus, with the frequency of targeted recombination dependent upon both the overall length of targeting homology and the length of homology flanking each side of the target gene deletion. Although most of the APRT+ recombinants analyzed reflect simple targeted replacement or conversion of the target gene deletion, a significant fraction appear to have arisen by target gene-templated extension and correction of the targeting fragment sequences. APRT fragments with limited targeting homology flanking one side of the target gene deletion yield proportionately fewer target gene conversion events and proportionately more templated extension and vector correction events than do fragments with more substantial flanking homology.     

Genetic exchange between homeologous sequences in mammalian chromosomes is averted by local homology requirements for initiation and resolution of recombination

We examined the mechanism by which recombination between imperfectly matched sequences (homeologous recombination) is suppressed in mammalian chromosomes. DNA substrates were constructed, each containing a thymidine kinase (tk) gene disrupted by insertion of an XhoI linker and referred to as a "recipient" gene. Each substrate also contained one of several "donor" tk sequences that could potentially correct the recipient gene via recombination. Each donor sequence either was perfectly homologous to the recipient gene or contained homeologous sequence sharing only 80% identity with the recipient gene. Mouse Ltk(-) fibroblasts were stably transfected with the various substrates and tk(+) segregants produced via intrachromosomal recombination were recovered. We observed exclusion of homeologous sequence from gene conversion tracts when homeologous sequence was positioned adjacent to homologous sequence in the donor but not when homeologous sequence was surrounded by homology in the donor. Our results support a model in which homeologous recombination in mammalian chromosomes is suppressed by a nondestructive dismantling of mismatched heteroduplex DNA (hDNA) intermediates. We suggest that mammalian cells do not dismantle mismatched hDNA by responding to mismatches in hDNA per se but rather rejection of mismatched hDNA appears to be driven by a requirement for localized homology for resolution of recombination.     

Transformation depending on intermolecular homologous recombination is stimulated by UV damage in transfected DNA

DNA-mediated gene transfer (DMGT) was performed in DNA repair-proficient and UV-hypersensitive, repair-deficient Chinese hamster ovary (CHO) cell lines using the UV-irradiated thymidine kinase gene from herpes simplex virus (HSV-TK). Transformation frequencies in repair-deficient CHO cell lines declined relative to repair-proficient cells with increasing UV damage in transfected DNA; approximately 3-fold higher UV fluence was required to inactivate 50% of irradiated HSV-TK plasmid molecules in repair-proficient cells. In cotransfection experiments performed with pairs of HSV-TK plasmids containing linker insertion mutations in TK coding sequences, moderate UV damage in plasmid DNA enhanced the yield of TK+ transformants resulting from homologous recombination between HSV-TK sequences up to 4-fold. These results suggest that UV damage in DNA can stimulate transformation of mammalian cells dependent on intermolecular DNA homology.     

Foreign DNA introduced by calcium phosphate is integrated into repetitive DNA elements of the mouse L cell genome.

We investigated the sites of integration of exogenous DNA fragments introduced by DNA-mediated gene transfer. Mouse Ltk- cells were transformed with the herpes simplex virus thymidine kinase gene and pBR322 DNA by the calcium phosphate precipitation method. Some of the integrated exogenous DNA sequences were recovered from the stable tk+ transformants in the form of plasmids that were capable of propagation in bacteria. Four plasmids derived from two cloned cell lines were analyzed in detail by nucleotide sequencing and hybridization techniques. These plasmids contained a total of seven cellular-exogenous DNA junctions. In all cases, there was no sequence homology between the exogenous and cellular DNA sequences adjacent to the joining sites, and no specific exogenous or cellular sequences occurred at the junctions. Rearrangement or deletion of Ltk- DNA was always associated with the integration of exogenous DNA. All of the assignable cellular sequences at the junctions were repetitive sequences. Two of these sequences were from the MIF-1 repetitive sequence family, and a third consisted of a 40-base pair simple copolymer of alternating deoxyadenosine-deoxythymidine. Our results suggest that repetitive sequences are relatively favorable sites for the integration of exogenous DNA.     

Suppression of intrachromosomal gene conversion in mammalian cells by small degrees of sequence divergence

Pairs of closely linked defective herpes simplex virus (HSV) thymidine kinase (tk) gene sequences exhibiting various nucleotide heterologies were introduced into the genome of mouse Ltk- cells. Recombination events were recovered by selecting for the correction of a 16-bp insertion mutation in one of the tk sequences. We had previously shown that when two tk sequences shared a region of 232 bp of homology, interruption of the homology by two single nucleotide heterologies placed 19 bp apart reduced recombination nearly 20-fold. We now report that either one of the nucleotide heterologies alone reduces recombination only about 2.5-fold, indicating that the original pair of single nucleotide heterologies acted synergistically to inhibit recombination. We tested a variety of pairs of single nucleotide heterologies and determined that they reduced recombination from 7- to 175-fold. Substrates potentially leading to G-G or C-C mispairs in presumptive heteroduplex DNA (hDNA) intermediates displayed a particularly low rate of recombination. Additional experiments suggested that increased sequence divergence causes a shortening of gene conversion tracts. Collectively, our results suggest that suppression of recombination between diverged sequences is mediated via processing of a mispaired hDNA intermediate.     

Effect of limited homology on gene conversion in a Saccharomyces cerevisiae plasmid recombination system.

Plasmids containing heteroallelic copies of the Saccharomyces cerevisiae HIS3 gene undergo intramolecular gene conversion in mitotically dividing S. cerevisiae cells. We have used this plasmid system to determine the minimum amount of homology required for gene conversion, to examine how conversion tract lengths are affected by limited homology, and to analyze the role of flanking DNA sequences on the pattern of exchange. Plasmids with homologous sequences greater than 2 kilobases have mitotic exchange rates as high as 2 x 10(-3) events per cell per generation. As the homology is reduced, the exchange rate decreases dramatically. A plasmid with 26 base pairs (bp) of homology undergoes gene conversion at a rate of approximately 1 x 10(-10) events per cell per generation. These studies have also shown that an 8-bp insertion mutation 13 bp from a border between homologous and nonhomologous sequences undergoes conversion, but that a similar 8-bp insertion 5 bp from a border does not. Examination of independent conversion events which occurred in plasmids with heteroallelic copies of the HIS3 gene shows that markers within 280 bp of a border between homologous and nonhomologous sequences undergo conversion less frequently than the same markers within a more extensive homologous sequence. Thus, proximity to a border between homologous and nonhomologous sequences shortens the conversion tract length.     

Characterization of double-strand break-induced recombination: homology requirements and single-stranded DNA formation.

In the yeast Saccharomyces cerevisiae, a double-strand chromosome break created by the HO endonuclease is frequently repaired in mitotically growing cells by recombination between flanking homologous regions, producing a deletion. We showed that single-stranded regions were formed on both sides of the double-strand break prior to the formation of the product. The kinetics of the single-stranded DNA were monitored in strains with the recombination-deficient mutations rad52 and rad50 as well as in the wild-type strain. In rad50 mutants, single-stranded DNA was generated at a slower rate than in the wild type, whereas rad52 mutants generated single-stranded DNA at a faster rate. Product formation was largely blocked in the rad52 mutant. In the rad50 rad52 double mutant, the effects were superimposed in that the exonucleolytic activity was slowed but product formation was blocked. rad50 appears to act before or at the same stage as rad52. We constructed strains containing two ura3 segments on one side of the HO cut site and one ura3 region on the other side to characterize how flanking repeats find each other. Deletions formed preterentially between the homologous regions closest to the double-strand break. By varying the size of the middle ura3 segment, we determined that recombination initiated by a double-strand break requires a minimum homologous length between 63 and 89 bp. In these competition experiments, the frequency of recombination was dependent on the length of homology in an approximately linear manner.     

Intermolecular recombination between DNAs introduced into mouse L cells is mediated by a nonconservative pathway that leads to crossover products.

We describe experiments designed to measure the efficiency of intermolecular recombination between mutant herpesvirus thymidine kinase (tk) genes introduced into mouse L cells. Recombinants were scored as stable transformants containing a functional tk gene. The two recombination substrates used were ptkB8, a pBR322-based plasmid containing a mutant tk gene, with a BamHI linker in an SphI restriction site that is centrally located within the gene, and mp10tk delta 3' delta 5', an mp10 vector with a tk gene deleted at both the 3' and 5' ends. The only homology shared by the two DNAs is 885 base pairs within the tk gene. To determine whether the double-strand break repair model that has been used to explain recombination in yeast cells (J. W. Szostak, T. L. Orr-Weaver, R. J. Rothstein, and F. W. Stahl, Cell 33:25-35, 1983) can account for recombination during the introduction of these DNAs into mammalian cells, we transformed cells with BamHI-linearized ptkB8 and supercoiled mp10tk delta 3' delta 5' replicative-form DNA. These two DNAs should recombine efficiently according to that model and should generate gene conversion products. In this reaction, the supercoiled DNA acts as the donor of information to repair the cleaved tk gene. Our results indicated that the efficiency of this reaction was very low (less than 10 transformants were obtained per 0.1 microgram of each DNA used in the reaction per 10(6) cells). In contrast, if BamHI-cleaved ptkB8 DNA was cotransformed into cells along with a circular DNA molecule containing a tk gene deleted only at its 3' end or only at its 5' end (mp10tk delta 3' or mp10tk delta 5'), then the efficiency of recombination could be more than 4 orders of magnitude higher than it was with circular mp10tk delta 3' delta 5' DNA. Recombination frequencies were highest when the tk delta 3' or tk delta 5' DNA used was cleaved at the tk deletion junction. Southern analyses of DNA from TK+ transformants generated with BamHI-cleaved ptkB8 and BamHI-cleaved mp10tk delta 3' DNAs indicated that recombination was almost always associated with the reassortment of markers flanking the reconstructed tk DNA. Together, these results are more consistent with the nonconservative single-strand annealing model for recombination that we proposed several years ago (F.-L. Lin, K. Sperle, and N. Sternberg, Mol. Cell. Biol. 4:1020-1034, 1984) than they are with the double-strand break repair model.     

An Examination of the Effects of Double-Strand Breaks on Extrachromosomal Recombination in Mammalian Cells

We studied the effects of double-strand breaks on intramolecular extrachromosomal homologous recombination in mammalian cells. Pairs of defective herpes thymidine kinase (tk) sequences were introduced into mouse Ltk- cells on a DNA molecule that also contained a neo gene under control of the SV40 early promoter/enhancer. With the majority of the constructs used, gene conversions or double crossovers, but not single crossovers, were recoverable. DNA was linearized with various restriction enzymes prior to transfection. Recombination events producing a functional tk gene were monitored by selecting for tk-positive colonies. For double-strand breaks placed outside of the region of homology, maximal recombination frequencies were measured when a break placed the two tk sequences downstream from the SV40 early promoter/enhancer. We observed no relationship between recombination frequency and either the distance between a break and the tk sequences or the distance between the tk sequences. The quantitative effects of the breaks appeared to depend on the degree of homology between the tk sequences. We also observed that inverted repeats recombined as efficiently as direct repeats. The data indicated that the breaks influenced recombination indirectly, perhaps by affecting the binding of a factor(s) to the SV40 promoter region which in turn stimulated or inhibited recombination of the tk sequences. Taken together, we believe that our results provide strong evidence for the existence of a pathway for extrachromosomal homologous recombination in mammalian cells that is distinct from single-strand annealing. We discuss the possibility that intrachromosomal and extrachromosomal recombination have mechanisms in common.     

Differential Requirements of Singleplex and Multiplex Recombineering of Large DNA Constructs

Recombineering is an in vivo genetic engineering technique involving homologous recombination mediated by phage recombination proteins. The use of recombineering methodology is not limited by size and sequence constraints and therefore has enabled the streamlined construction of bacterial strains and multi-component plasmids. Recombineering applications commonly utilize singleplex strategies and the parameters are extensively tested. However, singleplex recombineering is not suitable for the modification of several loci in genome recoding and strain engineering exercises, which requires a multiplex recombineering design. Defining the main parameters affecting multiplex efficiency especially the insertion of multiple large genes is necessary to enable efficient large-scale modification of the genome. Here, we have tested different recombineering operational parameters of the lambda phage Red recombination system and compared singleplex and multiplex recombineering of large gene sized DNA cassettes. We have found that optimal multiplex recombination required long homology lengths in excess of 120 bp. However, efficient multiplexing was possible with only 60 bp of homology. Multiplex recombination was more limited by lower amounts of DNA than singleplex recombineering and was greatly enhanced by use of phosphorothioate protection of DNA. Exploring the mechanism of multiplexing revealed that efficient recombination required co-selection of an antibiotic marker and the presence of all three Red proteins. Building on these results, we substantially increased multiplex efficiency using an ExoVII deletion strain. Our findings elucidate key differences between singleplex and multiplex recombineering and provide important clues for further improving multiplex recombination efficiency.     

Fine-resolution analysis of products of intrachromosomal homeologous recombination in mammalian cells.

Mouse Ltk- cell lines that contained a herpes simplex virus type 1 (HSV-1) thymidine kinase (tk) gene with a 16-bp insertion mutation linked to either a defective HSV-2 tk gene or a hybrid tk sequence comprised of HSV-1 and HSV-2 tk sequences were constructed. HSV-1 and HSV-2 tk genes have 81% nucleotide identity and hence are homeologous. Correction of the insertion mutant HSV-1 tk gene via recombination with the hybrid tk sequence required an exchange between homeologous tk sequences, although recombination could initiate within a region of significant sequence identity. Seven cell lines containing linked HSV-1 and HSV-1-HSV-2 hybrid tk sequences gave rise to tk+ segregants at an average rate of 10(-8) events per cell division. DNA sequencing revealed that each recombinant from these lines displayed an apparent gene conversion which involved an accurate transfer of an uninterrupted block of information between homeologous tk sequences. Conversion tract lengths ranged from 35 to >330 bp. In contrast, cell lines containing linked HSV-1 and HSV-2 tk sequences with no significant stretches of sequence identity had an overall rate of homeologous recombination of <10(-9). One such cell line produced homeologous recombinants at a rate of 10(-8). Strikingly, all homeologous recombinants from this latter cell line were due to crossovers between the HSV-1 and HSV-2 tk genes. Our results, which provide the first detailed analysis of homeologous recombination within a mammalian genome, suggest that rearrangements in mammalian genomes are regulated by the degree of sequence divergence located at the site of recombination initiation.     

Nucleotide sequence analysis of human hypoxanthine phosphoribosyltransferase (HPRT) gene deletions.

We have determined the nucleotide sequences of 10 intragenic human HPRT gene deletion junctions isolated from thioguanine-resistant PSV811 Werner syndrome fibroblasts or from HL60 myeloid leukemia cells. Deletion junctions were located by fine structure blot hybridization mapping and then amplified with flanking oligonucleotide primer pairs for DNA sequence analysis. The junction region sequences from these 10 HPRT mutants contained 13 deletions ranging in size from 57 bp to 19.3 kb. Three DNA inversions of 711, 368, and 20 bp were associated with tandem deletions in two mutants. Each mutant contained the deletion of one or more HPRT exon, thus explaining the thioguanine-resistant cellular phenotype. Deletion junction and donor nucleotide sequence alignments suggest that all of these HPRT gene rearrangements were generated by the nonhomologous recombination of donor DNA duplexes that share little nucleotide sequence identity. This result is surprising, given the potential for homologous recombination between copies of repeated DNA sequences that constitute approximately a third of the human HPRT locus. No difference in deletion structure or complexity was observed between deletions isolated from Werner syndrome or from HL60 mutants. This suggests that the Werner syndrome deletion mutator uses deletion mutagenesis pathway(s) that are similar or identical to those used in other human somatic cells.