Mismatch Repair of Heteroduplex DNA Intermediates of ...

MOLECULAR AND CELLULAR BIOLOGY, Jan. 1994, p. 400-406

Vol. 14, No. 1

0270-7306/94/$04.00+0 Copyright © 1994, American Society for Microbiology

Mismatch Repair of Heteroduplex DNA Intermediates of Extrachromosomal Recombination in Mammalian Cells WIN PING DENG AND JAC A. NICKOLOFF* Department of Cancer Biology, Harvard University School of Public Health, Boston, Massachusetts 02115 Received 22 June 1993/Returned for modification 15 September 1993/Accepted 8 October 1993

Previous work indicated that extrachromosomal recombination in mammalian cells could be explained by the single-strand annealing (SSA) model. This model predicts that extrachromosomal recombination leads to nonconservative crossover products and that heteroduplex DNA (hDNA) is formed by annealing of complementary single strands. Mismatched bases in hDNA may subsequently be repaired to wild-type or mutant sequences, or they may remain unrepaired and segregate following DNA replication. We describe a system to examine the formation and mismatch repair of hDNA in recombination intermediates. Our results are consistent with extrachromosomal recombination occurring via SSA and producing crossover recombinant products. As predicted by the SSA model, hDNA was present in double-strand break-induced recombination intermediates. By placing either silent or frameshift mutations in the predicted hDNA region, we have shown that mismatches are efficiently repaired prior to DNA replication.

Investigations of homologous recombination by using extrachromosomal or intrachromosomal systems in mammalian cells are useful for elucidating the mechanisms of genetic rearrangement. Extrachromosomal homologous recombination may involve either reciprocal or nonreciprocal events (reviewed in reference 8). The nonreciprocal transfer of information from one duplex to a homologous duplex elsewhere in the genome, termed gene conversion, may occur by two mechanisms, including the repair of double-strand gaps enlarged from double-strand breaks (DSBs) (49) and the repair of mismatched base pairs present in heteroduplex DNA (hDNA) in recombination intermediates (19, 29). Alternate sources of mismatches include DNA metabolic events such as DNA biosynthetic errors or spontaneous deamination of 5-methylcytosine in m5C. G base pairs (reviewed in references 14 and 17). Unrepaired mismatches are expected to segregate during DNA replication and are manifested in yeast cells as postmeiotic segregation events (41). DSBs stimulate homologous recombination in yeast cells (22, 23, 36, 38, 42; reviewed in reference 48) and mammalian cells (9-11, 21, 24-28, 37, 46, 50, 52; reviewed in reference 8). DNA-damaging agents, such as UV light, X rays, and chemicals, also stimulate recombination (4-6, 13, 16, 31, 33, 43, 51, 53). On the basis of these observations, several recombination models were proposed to explain the recombinogenic effects of DSBs. The DSB repair model (49) was proposed to explain the results of yeast transformation experiments and DSB-induced gene conversion (39, 40). The DSB repair model predicts interactions between DNA ends of a double-strand gap enlarged from a DSB with a second unbroken homologous duplex in the genome. DNA synthesis using the unbroken homolog as a template repairs the gap and effects gene conversion. A second model, termed single-strand annealing (SSA), proposed by Lin et al. (25, 27, 28) explains features of extrachromosomal recombination in mammalian cells, including the efficient recombination between DNAs broken in

nonhomologous regions and the predominance of nonconservative crossover products (1, 11, 25, 27, 28, 45, 52). Although there have been reports of pathways distinct from SSA in mammalian cells (3, 55), most extrachromosomal recombination can be explained by this model. In this report, we describe our studies of extrachromosomal recombination between heteroallelic neomycin (neo) genes on shuttle vectors in Chinese hamster ovary (CHO) cells. Our system permits an examination of the formation and repair of mismatches in hDNA recombination intermediates. At least 80% of mismatches present in hDNA were repaired prior to DNA replication, and in most crosses tested, a strong repair bias was observed. MATERIALS AND METHODS

Plasmid DNA constructions and preparation. Plasmids constructed and prepared by standard procedures (44) and are shown in Fig. 1. Plasmid SVBss was described previously (37). Plasmid SVBss-XE, containing two mutations in neo, was constructed by inserting a 10-bp XhoI linker into the EagI site of SVBss. The 2.1-kbp HindIII-BamHI fragment of pSV2neo (47) was inserted into the HindIII and BamHI sites of a pUC19 derivative missing the EcoRI site (pUC19-RI), creating pneoAn. Silent mutations, converting the neo EagI site into a PstI site, were introduced into pSV2neo, SVBss, and pneoAn by site-directed mutagenesis (12) to create plasmids pSV2neo-PE, SVBss-PE, and pneoAn-PE, respectively. Cell culture, electroporation, and recombination assays. CHO cells (strain Klc) were cultured as described previously (37). Plasmid DNAs were linearized with appropriate restriction enzymes and purified by passage through a Sepharose CL-6B spin column (Pharmacia) prior to electroporation, which was performed essentially as described previously (37). Five micrograms of each DNA plus 4 x 106 cells in a volume of 0.8 ml was shocked with 300 V at 960 ,uF with a Bio-Rad Gene Pulser. Cells were then transferred to 20 ml of prewarmed growth medium. The live cell count was determined by plating dilutions of shocked cells in nonselective medium, and the remaining cells were inoculated into two 100-mm-diameter dishes for selection 24 h later with were

* Corresponding author. Mailing address: Department of Cancer Biology, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115. Phone: (617) 432-1184. Fax: (617) 432-0107.

400

VOL. 14, 1994

HETERODUPLEX DNA MISMATCH REPAIR

(

X(E6

sv

DSB

401

R(B)

BamI

BamHI SVBss

peA

ISB

A

An

Eagi

. A

Hindil

Hind ag

B

Eagl SV

SVBss-XE SV

SVBss-PE

X(E) X

R(B) *

P(E)

R(B)

pneoAn-

I

XXE) P

pneoAn-PE

FIG. 1. Structures of recombination substrates. SVBss is a derivative of pSV2neo missing the EcoRI site and carrying an 10-bp EcoRI linker frameshift mutation (black bar) at the BssHII site in neo [termed R(B) (37)]. The SV40 promoter is indicated by SV. pneoAn carries a 2,150-bp HindIII-BamHI fragment with neo and polyadenylation signals from pSV2neo in pUC19. The neo regions of derivatives of SVBss and pneoAn are diagrammed below. These derivatives have a 10-bp X7zoI linker frameshift mutation at EagI, termed X(E), or a silent mutation with EagI converted to PstI, termed P(E) (hatched bar).

G418 (500 p,g/ml, 50% active; GIBCO). Cell viability following electroporation was about 50% of untreated cells. Colonies were harvested or stained with 1% crystal violet in 70% ethanol after 10 days in selective medium. Recombination frequencies (or transfection frequencies, when only one plasmid was introduced) were calculated as the ratio of G418-resistant (G418r) colonies to the total number of live cells plated in G418 medium. Values are reported as averages + standard deviations. (Averages + ranges are given when only two determinations were made.) Genomic DNA preparation and Southern hybridization analysis. Genomic DNA was prepared from expanded G418r colonies and analyzed by Southern hybridization as described previously (13, 37). Independence of recombinant products was assured since no cell divisions were allowed between electroporation and plating, and only one G418' colony from each plate was chosen for analysis.

RESULTS Experimental design. Mismatch repair in hDNA intermediates of extrachromosomal recombination was investigated with two types of plasmid recombination substrates (Fig. 1). One type carried a neo allele regulated by the simian virus 40 (SV40) promoter and was inactivated by a linker frameshift mutation at a 3' position (BssHII site). The second type also carried inactive neo alleles, as these lacked promoters. Additional mutations, introduced into neo in both types of plasmids at a 5' position (EagI site), consisted of either silent mutations or inactivating linker frameshift mutations. Both linker and silent mutations created new restriction sites that could be detected by restriction mapping. To monitor recombination, pairs of plasmids with heteroallelic neo genes were mixed and electroporated into CHO cells. Integration of a recombinant neo+ gene driven by the SV40 promoter can produce a G418r transfectant. Recombination was stimulated by DSBs introduced into restriction sites in recombination substrates prior to electroporation-mediated gene transfer. DSBs introduced at appropriate sites were expected to stimulate recombination as predicted by the SSA model (25). In this model, diagrammed

X(E)

(:1-~==

R(B)

Eagi

...I

(..S

X(EX

v

M

7

Eagi

"I ...)

FIG. 2. Extrachromosomal recombination as predicted by the SSA model. Symbols are as described in the legend to Fig. 1. Shown is a cross between SVBss-XE and pneoAn. Single strands are degraded from DSBs (A), exposing complementary strands which anneal (B) to produce the crossover product with hDNA between the DSBs and flanked by single-strand breaks in opposite strands (triangles).

in Fig. 2, single-stranded regions are created by strandspecific exonuclease digestion proceeding from broken ends. This exposes complementary regions in the substrate molecules, which subsequently anneal to form an apparent crossover product. The SSA model predicts that hDNA will be created upon strand annealing in the region bounded by the positions of the DSBs. We reasoned that mutations located within the hDNA region would produce mismatches that may subsequently be processed by repair enzymes to either wild-type or mutant sequences, or they may escape repair. Unrepaired mismatches in integrated plasmids are expected to segregate during the next cell division. If most mismatches are repaired and repair is not biased towards wild-type or mutant sequences, then a frameshift mutation in hDNA will reduce the yield of G418' recombinants by 50% compared with the yield without such a mutation. If mismatch repair is inefficient or absent, a frameshift mutation in hDNA would not reduce the yield of G418' recombinants, since segregation of mutant and wild-type sequences occurs before selection pressure is applied. In this case, one daughter cell would always receive wild-type information. Reductions by more than 50% are an indication of efficient repair biased toward mutant information. Reductions by less than 50% may result from biased repair toward wild-type sequences, inefficient repair, or a combination of these factors. Finally, in crosses with silent mutations in hDNA, all products (mutant, wild type, or mixed [indicating unrepaired segregants or the integration of two products repaired in opposite directions]) can be recovered and identified by restriction mapping. Phenotypic characterization of mutations in recombination substrates. Plasmid SVBss (Fig. 1) was introduced into cells alone to determine the reversion frequency for the R(B) frameshift mutation. Electroporation and G418 selection were performed as described in Materials and Methods. In four determinations, G418' revertants were recovered at an average frequency of 3.0 x 10-5 + 0.6 x 10-5 when SVBss was cleaved at the R(B) mutation, a value about 10-fold lower than recombination values (see below). No revertants were recovered when SVBss was cleaved at other sites. Reversion of the X(E) frameshift mutation was not tested in CHO cells since this mutation was always paired with the

DENG AND NICKOLOFF

402

MOL. CELL. BIOL.

requenc *SD (10-5) Percent of _lreediab

Cass R(B)

SV

1_ SV X(E) 2

_

SV

P(E) VA

_

R(B) _

R(B)

SV

SV

_z2

X(E6

gx 2

urxv

24 ±3

36 :2

100

X(E) R(B)

*_~~

]

0

0

0

37 ±15

i ±18

1X

R(B)

3 SV

I

gm

R(B) SV

4

X(E) E

12

±7

23

±6

30

33±12

53

±13

100

32 i7

54

R(B)

SV

5 R(B) =

SV

6

X(E6

16 ±8

FIG. 3. Extrachromosomal homologouis recombination between SVBss and pneoAn derivatives. Symbol s are as described in the legend to Fig. 1. Six crosses between pS'V2neo and pneoAn derivatives are shown. DSBs were introduceed at sites connected by diagonal lines. For crosses 1 and 2, DS'Bs were present at R(B) mutations and TthlllI sites; for crosses 3 to 6, DSBs were present at MscI sites in pSV2neo derivatives and1 at BglII sites in pneoAn derivatives. Structures of predicted recomibination intermediates are diagrammed in the next column. Reconnbination frequencies are given for two independent sets of experinnents, with three determinations per set for crosses 1 and 2 ancI four determinations for crosses 3 to 6. The detection level in theste experiments is about 0.1 x 10'. In the final column, the average r f 100, are compared with for crosses 1, 3, and 5, assigned a value cecombination average frequencies for crosses 2, 4, and 6, respectively,

fomprequcies

R(B) mutation. However, the X(E) niutation in pneoAn-XE eliminated kanamycin resistance in i Escherichia coli and reduced recombination in CHO cells (see below), indicating that the X(E) mutation inactivates ne?0. The phenotypic silence of the P(E,) mutation in neo was confirmed in E. coli and in CHO cel[ls. E. coli transformed with pneoAn-PE or pSV2neo-PE wa s resistant to kanamycin. In CHO cells, duplicate transfec tion experiments using 0.5 ,g of pSV2neo or pSV2neo-PE yielded G418' transfectants at similar frequencies (3.2 x 10-4 ± 0.2 x 10-4 and 3.6 X 10-4 + 0.1 X 10-4 , respectively). Extrachromosomal recombinants are formed by a crossover event. To determine whether DSB-induced recombination proceeds as predicted by the SSA model, we monitored extrachromosomal homologous recombination between mutant neo genes by using crosses 1 and 2 as diagrammed in Fig. 3. In cross 1, DSBs were introduced into the linker mutation of SVBss and at a TthlllI site in pneoAn (connected by diagonal lines in Fig. 3). The recombination intermediate is predicted to carry the 5' part of neo driven by the SV40 promoter (white boxes) linked to the 3' part of neo carried by pneoAn (shaded), with hDNA including the SVBss mutation. Cross 2 was identical to cross 1 except for the presence of the X(E) frameshift mutation 5' to the hDNA region. As predicted for an SSA-mediated crossover event, the X(E) mutation prevented the recovery of functional G418r recombinants. Genetic evidence for hDNA formation and mismatch repair.

Four crosses were performed to examine hDNA in recombination intermediates (3 to 6 in Fig. 3). These crosses are similar to crosses 1 and 2 described above except that DSBs were introduced into recombination substrates at positions flanking 5' frameshift or silent mutations. These DSB and mutation arrangements produce recombination intermediates that have 5' mutations within hDNA. The resulting mismatches may be repaired to either wild-type or mutant sequences or remain unrepaired. Since there is no selection pressure for or against silent mutations, all repaired and unrepaired mismatches produce functional genes, and as expected, recombination frequencies for crosses 3 and 5 were similar to that for cross 1. As discussed above, the presence of a frameshift mutation in hDNA may or may not reduce recombination frequencies. We-found that the X(E) mutation in cross 4 reduced recombination frequencies to about 30% of cross 3 frequencies. Despite the variability among independent determinations, these reductions were significant: P(t) < 0.05 for both experiments 1 and 2. These results indicate that mismatch repair in hDNA is efficient and that repair often favors conversion to mutant sequences (i.e., the mutant top strand, donated by the pSV2neo derivative, is often used as a repair template). When the 5' mutations were donated by the pneoAn derivative (crosses 5 and 6), mutant bases in mismatched rgo regons were present on the bottom strand. In this case (cross 6), recombination frequencies were also significantly reduced by the X(E) frameshift mutation, to 54% of the level seen with a silent mutation in cross 5 [P(t) < 0.1 for experiment 1; P(t) < 0.05 for experiment 2]. As above, the observed reduction indicates that mismatch repair is efficient. The smaller reduction in recombination for cross 6 relative to cross 5 compared with the reduction for cross 4 relative to cross 3 provides additional evidence that the top strand is usually used as a repair template. Physical mapping of recombinant products. We further explored the mechanism of mismatch correction in hDNA by examining independent recombinant products of crosses 3 and 5, predicted to yield a silent mutation in hDNA. Silent mutations consisted of two single-base changes in the wildtype EagI site (both of which occur in third positions of the neo reading frame), converting the EagI site to a PstI site. Two single-base mismatches separated by two G. C base pairs are predicted in hDNA, shown here with the mutant sequence above the wild-type sequence: 5' T G C A *

*

3' C C G G Mismatch repair to wild type yields a product sensitive to EagI (EagIs) and resistant to PstI (Pstlr), and vice versa for repair to mutant sequences (Eagjr and PstPI). Since the two mismatched base pairs must be repaired in the same direction to form an EagI or a PstI site, repair of only one mismatch or repair of both in opposite directions would yield a site resistant to both enzymes. We analyzed 16 independent G418r products from cross 3 and 15 from cross 5. Genomic DNA from each recombinant was digested with NdeI, which releases a 2,970-bp fragment detectable by Southern hybridization with a probe containing SV40 promoter DNA. A second sample of each DNA was digested with NdeI and EagI. Products repaired to wild type were EagIP and released a hybridizing fragment of 764 bp, whereas EagIr products yielded the same pattern in both NdeI and NdeI-EagI digestions. We probed with SV40


VOL. 14, 1994

A Cross 3

Ndel

P(E)

7

403

NdeI

2206

I-// E E

Probe SV /

Cross 5

P(E)

B

Cross 3

1

2

3

5

4

H

7

6

Cross 5 8

9

P

10 1gm

5-6.6 -4.4

2.97-

*

" _

b

*

-_

to

-2.3

**

O

4*

aa.1 40

-2.0

i

0.76-

o

R

R

R

R

R

R

S

M

S

S

FIG. 4. Southern analysis of extrachromosomal recombination products. (A) Restriction maps of cross 3 and 5 recombination intermediates (see Fig. 3); fragment sizes are in base pairs. (B) Southern hybridization analysis of genomic DNA from 10 independent G418' recombinants for each cross. Each pair of lanes shows DNA from a single isolate digested with NdeI (left) or NdeI and EagI (right). Lanes marked P have parental CHO DNA. DNAs were hybridized with a 574-bp NdeI-HindIII fragment from pSV2neo (marked "Probe" in panel A) labeled with "P. EagIr and EagIs products yield 2,970- and 764-bp hybridizing fragments, respectively. EagI genotypes of each isolate are listed below. R, Eaglr; S, EagP; M, mixed genotype. Molecular sizes are given in kilobase pairs.

promoter DNA instead of neo DNA because functional neo genes would be linked to the SV40 promoter, whereas both

functional and nonfunctional neo genes (perhaps integrated at different chromosomal sites) would hybridize to neo, complicating the analysis. The SV40 promoter probe hybridized to many sites in the CHO genome, probably because of sequence conservation among elements that function as promoters in mammalian cells. However, these extra hybridization signals did not interfere with the analysis of the recombinant neo genes. Representative results of this analysis are shown in Fig. 4, and the results are summarized in Fig. 5. Consistent with the genetic results described above, the majority of the recombinant products had repaired the mismatched base pairs in hDNA. Only 20% exhibited a mixed mutant/wild-type pattern. This represents a maximum value for unrepaired products, since a mixed pattern may also arise if two products repaired in opposite directions integrate into chromosomal DNA. Cross 3 products were usually Eaglr, whereas cross 5 products were usually Eagls. These results clearly show that the top strand, donated by the pSV2neo derivative, was most often used as a template for repair. Since EagIr sites may also be Pstjr (as a result of

partial repair or repair of the two mismatches in opposite directions), we assayed the EagJr products for sensitivity to PstI and found that all EagJr products were PstP (representative data are shown in Fig. 6). Thus, the pair of mismatches created by the silent P(E) mutation are usually corrected in the same direction. Identical mismatches are present in hDNA for crosses 3 and 5, T-C and A-G, although the specific bases are present in opposite strands in the two crosses. Repair of cross 3 intermediates yielded primarily T-A and A-T base pairs, but cross 5 yielded primarily G-C and C-G base pairs. The repair of identical mismatches to opposite base pairs indicates that repair bias is strand

A

Ndel

PstI sv

Ndel

PstI

\1 39/

764

.1

--L-e

-

Probe

B

Pst5

1

2

3

4

5

-w

.0.

...

Mixed pstr ,,--F 6 7 8 9 P I,

4.* '.0 0 1.

Cross

hDNA Intermediates 175

3

SV

E 70 bp

1

3

E

1

of, I I

-0.9 -0.76

MiXed

I

bp

P(E)

31 1 12

SV

5

# AO

Products

-2.0

11

3

P(E) FIG. 5. Summary of product distributions. Symbols are as described in the legend to Fig. 3. Recombination intermediates for crosses 3 and 5 are shown, and the distribution of EagIr and EagIs recombinants is listed. Distances between strand breaks in each strand and the mismatches are shown.

FIG. 6. Southern analysis of EagF products. (A) Restriction map of recombination intermediates showing positions of NdeI and PstI sites. Symbols are as described in the legend to Fig. 1; fragment sizes are in base pairs. (B) Southern hybridization analysis of genomic DNA from EagIr, EagIs, and mixed recombinants; lane P has parental CHO DNA. DNAs were digested with NdeI and PstI and hybridized to an SV40 promoter probe as described in the legend to Fig. 4. PstlP products yield a 764-bp fragment; Pstlr

products yield a 903-bp fragment.

404

DENG AND NICKOLOFF

dependent and independent of mismatch type (at least for these particular mispairs; see below). Strand-dependent repair may be a result of single-strand breaks present in recombination intermediates (Fig. 2). DISCUSSION In this study, we set out to examine a previously untested prediction of the SSA model involving the formation of hDNA in homologous recombination intermediates. This model predicts that hDNA is formed by annealing complementary strands exposed by a strand-specific exonuclease acting on DSBs and that hDNA will be located in the region bounded by the DSBs. We have confirmed this prediction by placing mutations in the predicted hDNA region and determining the fate of the mismatches formed in hDNA by using genetic and physical assays. Our system depends on the isolation of G418' recombinants, which requires not only a productive homologous recombination event between heteroallelic neo genes but also a nonhomologous recombination event between the plasmid fusion product and chromosomal DNA (integration). Previous work in our laboratory with related plasmid substrates has indicated that the homologous interaction probably precedes the nonhomologous interaction (37). Thus, it is likely that hDNA is produced in extrachromosomal elements prior to integration. Because recombinants are not analyzed until 10 days after transfection, we cannot determine whether hDNA mismatch repair is occurring before or after plasmid integration. However, if repair occurs after integration, it must occur before integrated DNA is replicated. Previously we showed that elec-

troporated plasmids integrate randomly into CHO chromosomes (34). Control transfections with two plasmids carrying different selectable markers indicate that under the conditions used here, only 19% of cells integrating one marker also successfully integrated the other (35). These results are consistent with our earlier recombination studies (37), which indicated that most G418' products carry single copies of recombinant plasmids. Mismatch repair studies have demonstrated that different mismatches are repaired with different efficiencies and that repair may be biased with respect to strands or mismatch type (see reference 14 for a review). Strand bias has been reported in cases in which a single-strand break is present adjacent to a mismatch, as in E. coli GATC-methyl-directed mismatch repair, which initiates from a single-strand break introduced by the MutH protein following recognition of hemimethylated GATC sites (30). In mammalian cells, strand break-dependent strand bias of mismatch repair was suggested by Hare and Taylor (15) as a mechanism for repairing replication errors. On the basis of their studies with replication-proficient SV40 substrates in monkey CV1 cells, these authors suggested that strand breaks arising during discontinuous replication may direct repair in favor of the template strand. However, a contrary conclusion was reached by Heywood and Burke (18), who used replicationproficient SV40 shuttle vectors in monkey COS cells, as repair was not biased by single-strand breaks created in vitro adjacent to mismatched bases. To explain this difference, Heywood and Burke (18) suggested that strand break-dependent strand bias may require the binding of a single-strand endonuclease to mismatches before it cleaves DNA, or that a protein bound to the mismatch interacts with the endonu-

clease. We observed efficient repair of each of the mismatches created in the present study, including 14-base loops and two single-base mismatches separated by two base

MOL. CELL. BIOL.

pairs. In the latter case, repair was biased, favoring the strand donated by the pSV2neo derivative. Since the distance between the break on the top strand is 175 bp from the mismatched bases but only 70 bp from the break on the bottom strand (Fig. 5), it is possible that the more proximal nick on the bottom strand is biasing repair in favor of the top strand, in agreement with the results of Hare and Taylor (15). We found that loops formed by 14-base linker insertions in hDNA were efficiently repaired. These are palindromic and may form stem-loop structures of five G. C base pairs and a four-base loop. Previous studies have shown that longer (30 to 40-base) palindromic mismatches in hDNA are not repaired efficiently in yeast (32) or mammalian (7) cells and that repair efficiencies are inversely proportional to stemloop stabilities (32). Thus, the efficient repair of the 14-base palindromic loops may reflect their lower stability relative to 30- to 40-base palindromes. Alternatively, repair of palindromic loops in hDNA may be enhanced by adjacent strand breaks. However, in contrast to the strong strand bias for single-base mismatch repair, less bias was observed for loop repair. A significant degree of repair favored the strand containing the loop (i.e., the mutant strand), contrasting with the results of two studies with monkey cells. One study showed a twofold bias against loops 25 to 247 bases in length, and apparently, bias was independent of strand breaks (54). The second study showed a fourfold bias against longer loops when two different loops are juxtaposed (2). These contrasting results may be due to one or more of the following differences in assay systems: cell types, substrate replication proficiencies, loop structures, loop lengths, and proximity of single-strand nicks. Finally, consistent with the suggestion by Holmes et al. (20), the differential recognition of loops and single base-pair mismatches in our system argues that mismatches are recognized during repair, rather than repair being dependent only on the recognition of nicks. The system described here offers a convenient means to examine mismatch repair in hDNA recombination intermediates created in vivo, with both hDNA extent and mismatch type under simple experimental control. We are currently investigating nick-directed repair bias by varying distances between nicks and mismatches, and loop repair bias by creating phenotypically silent loops in hDNA. ACKNOWLEDGMENTS We acknowledge the expert technical assistance provided by Douglas Sweetser. Helpful comments from John Little, Leona Samson, F. Andrew Ray, Elizabeth Miller, Heather Hough, Danielle Gioioso, and Richard Reynolds are greatly appreciated. This research was supported by grant CA 54079 to J.A.N. from the National Institutes of Health. REFERENCES 1. Anderson, R., and S. L. Eliason. 1986. Recombination of homologous DNA fragments transfected into mammalian cells occurs predominantly by terminal pairing. Mol. Cell. Biol. 6:3246-

3252.

2. Ayares, D., D. Ganea, L. Chekuri, C. R. Campbell, and R. Kucherlapati. 1987. Repair of single-stranded nicks, gaps, and loops in mammalian cells. Mol. Cell. Biol. 7:1656-1662. 3. Ayares, D., J. Spencer, F. Schwartz, B. Morse, and R Kucheriapati. 1985. Homologous recombination between autonomously replicating plasmids in mammalian cells. Genetics 111:375-388. 4. Bhattacharyya, N. P., V. M. Maher, and J. J. McCormick. 1989. Ability of structurally related polycyclic aromatic carcinogens to induce homologous recombination between duplicated chromosomal sequences in mouse L cells. Mutat. Res. 211:205-214.

VOL. 14, 1994

5. Bhattacharyya, N. P., V. M. Maher, and J. J. McCormick. 1990. Effect of nucleotide excision repair in human cells on intrachromosomal homologous recombination induced by UV and 1-nitrosopyrene. Mol. Cell. Biol. 10:3945-3951. 6. Bhattacharyya, N. P., V. M. Maher, and J. J. McCormick 1990. Intrachromosomal homologous recombination in human cells which differ in nucleotide excision-repair capacity. Mutat. Res. 234:31-41. 7. Bollag, R. J., D. R. Elwood, E. D. Tobin, A. R. Godwin, and R. M. Liskay. 1992. Formation of heteroduplex DNA during mammalian intrachromosomal gene conversion. Mol. Cell. Biol. 12:1546-1552. 8. Bollag, R. J., A. S. Waldman, and R. M. Liskay. 1989. Homologous recombination in mammalian cells. Annu. Rev. Genet. 23:199-225. 9. Brenner, D. A., A. C. Smigocki, and R. D. Camerini-Otero. 1985. Effect of insertions, deletions, and double-strand breaks on homologous recombination in mouse L cells. Mol. Cell. Biol. 5:684-691. 10. Brenner, D. A., A. C. Smigocki, and R. D. Camerini.Otero. 1986. Double-strand gap repair results in homologous recombination in mouse L cells. Proc. Natl. Acad. Sci. USA 83:17621766. 11. Chakrabarti, S., and M. M. Seidman. 1986. Intramolecular recombination between transfected repeated sequences in mammalian cells is nonconservative. Mol. Cell. Biol. 6:2520-2526. 12. Deng, W. P., and J. A. Nickoloff. 1992. Site-directed mutagenesis of virtually any plasmid by eliminating a unique site. Anal. Biochem. 200:81-88. 13. Deng, W. P., and J. A. Nickoloff. 1994. Preferential repair of UV damage in highly transcribed DNA diminishes UV-induced intrachromosomal recombination in mammalian cells. Mol. Cell. Biol. 14:391-399. 14. Grilley, M., J. Holmes, B. Yashar, and P. Modrich. 1990. Mechanisms of DNA-mismatch correction. Mutat. Res. 236: 253-267. 15. Hare, J. T., and J. H. Taylor. 1985. One role for DNA methylation in vertebrate cells is strand discrimination in mismatch repair. Proc. Natl. Acad. Sci. USA 82:7350-7354. 16. Heligren, D., H. Luthman, and B. Lambert. 1989. Induced recombination between duplicated neo genes stably integrated in the genome of CHO cells. Mutat. Res. 210:197-206. 17. Heywood, L. A., and J. F. Burke. 1990. Mismatch repair in mammalian cells. Bioessays 12:473-477. 18. Heywood, L. A., and J. F. Burke. 1990. Repair of single nucleotide DNA mismatches transfected into mammalian cells can occur by short-patch excision. Mutat. Res. 236:59-66. 19. Hoiday, R. 1964. A mechanism for gene conversion in fungi. Genet. Res. (Cambridge) 5:282-304. 20. Holmes, J., Jr., S. Clark, and P. Modrich. 1990. Strand-specific mismatch correction in nuclear extracts of human and Drosophila melanogaster cell lines. Proc. Natl. Acad. Sci. USA 87:5837-5841. 21. Jasin, M., J. de Villiers, F. Weber, and W. Schaffner. 1985. High frequency of homologous recombination in mammalian cells between endogenous and introduced SV40 genomes. Cell 43: 695-703. 22. Klar, A. J. S., J. N. Strathern, and J. A. Abraham. 1984. Involvement of double-strand chromosomal breaks for matingtype switching in Saccharomyces cerevisiae. Cold Spring Harbor Symp. Quant. Biol. 49:77-88. 23. Kolodkin, A. L., A. J. S. Klar, and F. W. Stahl. 1986. Doublestrand breaks can initiate meiotic recombination in S. cerevisiae. Cell 46:733-740. 24. Kucherlapati, R S., E. M. Eves, K.-Y. Song, B. S. Morse, and 0. Smithies. 1984. Homologous recombination between plasmids in mammalian cells can be enhanced by treatment of input DNA. Proc. Natl. Acad. Sci. USA 81:3153-3157. 25. Un, F.-L., K. Sperle, and N. Sternberg. 1984. Model for homologous recombination during transfer of DNA into mouse L cells: role for DNA ends in the recombination process. Mol. Cell. Biol. 4:1020-1034. 26. Un, F.-L., K. Sperle, and N. Stemnberg. 1987. Extrachromo-


405

somal recombination in mammalian cells as studied with singleand double-stranded DNA substrates. Mol. Cell. Biol. 7:129140. 27. Lin, F.-L., K. Sperle, and N. Sternberg. 1990. Intermolecular recombination between DNAs introduced into mouse L cells is mediated by a nonconservative pathway that leads to crossover products. Mol. Cell. Biol. 10:103-112. 28. Un, F.-L., K. Sperle, and N. Sternberg. 1990. Repair of doublestranded breaks by homologous fragments during transfer of DNA into mouse L cells. Mol. Cell. Biol. 10:113-119. 29. Meselson, M., and C. M. Radding. 1975. A general model for genetic recombination. Proc. Natl. Acad. Sci. USA 72:358-361. 30. Modrich, P. 1991. Mechanisms and biological effects of mismatch repair. Annu. Rev. Genet. 25:229-253. 31. Mudgett, J. S., and W. D. Taylor. 1990. Recombination between irradiated shuttle vector DNA and chromosomal DNA in African green monkey kidney cells. Mol. Cell. Biol. 10:37-46. 32. Nag, D. K., M. A. White, and T. D. Petes. 1989. Palindromic sequences in heteroduplex DNA inhibit mismatch repair in yeast. Nature (London) 340:318-320. 33. Nairn, R. S., R. M. Humphrey, and G. M. Adair. 1988. Transformation depending on intermolecular homologous recombination is stimulated by UV damage in transfected DNA. Mutat. Res. 208:137-141. 34. Nickoloff, J. A. 1992. Transcription enhances intrachromosomal homologous recombination in mammalian cells. Mol. Cell. Biol. 12:5311-5318. 35. Nickoloff, J. A. Unpublished data. 36. Nickoloff, J. A., E. Y. C. Chen, and F. Heffron. 1986. A 24-base-pair sequence from the MAT locus stimulates intergenic recombination in yeast. Proc. Natl. Acad. Sci. USA 83:78317835. 37. Nickoloff, J. A., and R. J. Reynolds. 1990. Transcription stimulates homologous recombination in mammalian cells. Mol. Cell. Biol. 10:4837-4845. 38. Nickoloff, J. A., J. D. Singer, M. F. Hoekstra, and F. Heffron. 1989. Double-strand breaks stimulate alternative mechanisms of recombination repair. J. Mol. Biol. 207:527-541. 39. Orr-Weaver, T. L., and J. W. Szostak. 1983. Yeast recombination: the association between double-strand gap repair and crossing over. Proc. Natl. Acad. Sci. USA 80:4417-4421. 40. Orr-Weaver, T. L., J. W. Szostak, and R. J. Rothstein. 1981. Yeast transformation: a model system for the study of recombination. Proc. Natl. Acad. Sci. USA 78:6354-6358. 41. Petes, T. D., R. E. Malone, and L. S. Symington. 1991. Recombination in yeast, p. 407-521. In E. W. Jones, J. R. Pringle, and J. R. Broach (ed.), The molecular and cellular biology of the yeast Saccharomyces: genome dynamics, protein synthesis, and energetics, vol. 1. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 42. Ray, A., I. Siddiqi, A. L. Kolodkin, and F. W. Stahl. 1988. Intrachromosomal gene conversion induced by a DNA doublestrand break in Saccharomyces cerevisiae. J. Mol. Biol. 201: 247-260. 43. Resnick, M. A. 1979. The induction of molecular and genetic recombination in eukaryotic cells. Adv. Radiat. Biol. 8:175-217. 44. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 45. Seidman, M. M. 1987. Intermolecular homologous recombination between transfected sequences in mammalian cells is primarily nonconservative. Mol. Cell. Biol. 7:3561-3565. 46. Song, K.-Y., L. Chekuri, S. Rauth, S. Ehrlich, and R. Kucherlapati. 1985. Effect of double-strand breaks on homologous recombination in mammalian cells and extracts. Mol. Cell. Biol. 5:3331-3336. 47. Southern, P. J., and P. Berg. 1982. Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early region promoter. J. Mol. Appl. Genet. 1:327-341. 48. Stahl, F. W. 1986. Roles of double-strand breaks in generalized genetic recombination. Prog. Nucleic Acid Res. 33:169-194. 49. Szostak, J. W., T. L. Orr-Weaver, R. J. Rothstein, and F. W.

406

DENG AND NICKOLOFF

Stahl. 1983. The double-strand break repair model for recombination. Cell 33:25-35. 50. Thomas, K. R., K R. Folger, and M. R. Capecchi. 1986. High frequency targeting of genes to specific sites in the mammalian genome. Cell 44:419-428. i1. Tsujimura, T., V. M. Maher, A. R. Godwin, R. M. Uiskay, and J. J. McCormick. 1990. Frequency of intrachromosomal homologous recombination induced by UV radiation in normally repairing and excision repair-deficient human cells. Proc. Natl. Acad. Sci. USA 87:1566-1570. Wake, C. T., F. Vernaleone, and J. H. Wilson. 1985. Topological requirements for homologous recombination among DNA mol-

MOL. CELL. BIOL. ecules transfected into mammalian cells. Mol. Cell. Biol. 5:2080-2089. 53. Wang, Y., V. M. Maher, R. M. Liskay, and J. J. McCormick. 1988. Carcinogens can induce homologous recombination between duplicated chromosomal sequences in mouse L cells. Mol. Cell. Biol. 8:196-202. 54. Weiss, U., and J. H. Wilson. 1987. Repair of single-stranded loops in heteroduplex DNA transfected into mammalian cells. Proc. Natl. Acad. Sci. USA 84:1619-1623. 55. Yang, D., and A. S. Waldman. 1992. An examination of the effects of double-strand breaks on extrachromosomal recombination in mammalian cells. Genetics 132:1081-1093.