Mutation Research, 226 (1989) 1-8 1 Elsevier

MTRL 0192

Mutagen-induced recombination between stably integrated n e o gene fragments in CHO and EM9 cells

D. Hellgren, S. Sahl6n and B. Lambert

Department of Clinical Genetics, Karolinska Institute, S-104 01 Stockholm (Sweden)

(Accepted 13 December 1988)

Keywords." DNA rearrangement; Homologous recombination; Neomycin (neo) resistance gene; CHO cells; EM9; MMS; HN2; BPDE; MMC; X-ray; BrdUrd

Summary

The ability of mutagenic agents to induce homologous recombination was studied in a 'normal' Chinese hamster ovary (CHO) cell clone (CHO:5) and a cell clone (EM9"2) derived from the presumptive DNA repair-deficient mutant cell line EM9, which has a high spontaneous sister-chromatid exchange (SCE) level and shows hypersensitivity towards monofunctional alkylating agents and bromodeoxyuridine (BrdUrd). The 2 cell clones have been transfected with and allowed to incorporate stable genomic inserts of the vector plII-14gpt, which contains 2 tandemly arranged n e o gene fragments with a common 400-bp region of homology. Recombination between the truncated n e o genes gives rise to geneticin sulfate (G 418)-resistant revertants with a spontaneous frequency of about 10 -4 in both cell clones. In CHO:5 an increased frequency of revertants was obtained after treatment with methyl methanesulfonate (MMS) and mitomycin C (MMC), while HN2, benz[a]pyrene diolepoxide (BPDE) and X-irradiation gave negative results. EM9:2 showed about the same increase of revertants after treatment with MMS as CHO:5, but in a 10-fold lower dose range. HN2 as well as BrdUrd induced revertants in EM9:2. These results show that mutagenic agents (MMS, MMC, HN2, BrdUrd) can induce homologous recombination in this system. This effect does not seem to be an unspecific effect of DNA damage (no effect of X-ray and BPDE), or related to SCE induction in general (similar spontaneous and MMS-induced frequencies of revertants in CHO:5 and EM9:2). However, the positive effect of BrdUrd in EM9:2 and the difference between CHO:5 (negative) and EM9:2 (positive) with regard to HN2-induced revertants suggest that certain types of DNA damage are more recombinogenic in EM9 than in 'normal' CHO cells, which possibly reflects the specific mutation in the former cell line.

Correspondence: Dr. Dennis Hellgren, Department of Clinical Genetics, Karolinska Hospital, S-104 01 Stockholm (Sweden).

Abbreviations: MMS, methyl methanesulfonate; HNz, nitrogen mustard; SCE, sister-chromatid exchange; MMC, mitomycin C;

BPDE, benz[a]pyrene diolepoxide; DMSO, dimethyl sulfoxide; BrdUrd, bromodeoxyuridine; MPA, mycophenolic acid; G 418, geneticin sulfate; TK, thymidine kinase; CHO, Chinese hamster ovary; DMEM, Dulbecco's modification of Eagle's medium; FCS, fetal calf serum; neo, neomycin resistance gene.

0165-7992/89/$ 03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)

Recombination in mammalian cells includes a number of events by which pre-existing nucleotide sequences are rearranged through interaction be- tween regions of sequence homology within or be- tween chromatids or chromosomes. Gene mutation by insertion, deletion and gene conversion as well as chromosome translocation and sister-chromatid exchange (SCE) may result from such events. Recombination may also be involved in the un- covering of recessive mutations at heterozygous loci of tumor suppressor genes (Cavenee et al., 1986). In spite of these important consequences of recombination in somatic cells, little is known about its frequency, inducibility by mutagenic/car- cinogenic agents or the molecular mechanisms in- volved.

In bacteria and yeast at least some proteins in- volved in recombination play a role in DNA repair, for example recA and RAD 52 (Smith, 1988; Friedberg, 1988). Results from studies with recA

begin to yield information about interaction be- tween protein and DNA (West, 1988). Various aspects of recombination in mammalian cells have been studied during the recent years by using ex- trachromosomal plasmids or exogenous DNA se- quences stably integrated in the cellular genome

amp gpt H

pill- 14gpt

--5 >

X n e o - ] t4 B n e o - 2 B E

Fig. 1. Description o f pl l l-14gpt. The two neo gene fragments

are shown as neo-I and neo-2. The triangles show where t h e deletions have been introduced to mutate the genes. The distances between the 2 Hindl l l sites is about 4200 bp. The black box indicates regions of homology in neo-1 and neo-2. The ar- rows show the direction of transcription. For further details see Hellgren et al. (1988)..g~1, SV40 enhancer and promoter; ~ , E. coli gene gpt coding for mycophenolic acid resistance; ~ , amp, bacterial ampicillin resistance gene; LA, neo gene fragments; m , region of homology in the neo gene fragments. Restriction enzymes: H, Hindl l l ; M, Mbol l (not all sites are shown); B, BamHl; E, EcoRl.

(Kucherlapati, 1986; Letsou and Liskay, 1986). Mammalian cell mutants with defects of DNA

repair and recombination could be useful in these studies. One such cell mutant is the Chinese hamster ovary (CHO) cell line EM9, which shows a high spontaneous frequency of SCE, hypersen- sitivity towards bromodeoxyuridine (BrdUrd) and mono-functional alkylating agents and impaired ability to repair DNA double-strand breaks (Thompson et al., 1982).

We have previously shown (Hellgren et al., 1988) that spontaneous and induced recombinational events can be studied in CHO cells containing the stably integrated vector plII-14gpt (Fig. 1). The vector contains a selectable marker gene, gpt (to allow selection of stable transformations in myco- phenolic acid (MPA) medium) and 2 tandemly ar- ranged neo gene fragments which are able to recombine within a common 400-bp region. After DNA transfection and selection of MPA-resistant cell clones, stable revertants with the active neo

gene are recovered in medium containing geneticin sulfate (G 418). The associated change in the vector DNA is verified by restriction analysis. Our previous results showed that methyl methane- sulfonate (MMS) is able to induce a 5-fold increase in the frequency of revertants in 2 MPA-resistant clones CHO:I and CHO:5, whereas no effect was observed after treatment with HN2. To obtain fur- ther information about the possible relationship between DNA repair, SCE and recombination in this system, we have introduced pIII-14gpt in EM9 cells. Here we report on the frequency of G 418-resistant revertants in one MPA-selected EM9 clone, EM9:2, after treatment with MMS, HN2 or BrdUrd. We also present additional data on the ability of mitomycin C (MMC), benz[a]pyrene diolepoxide (BPDE) and X-ray to induce recom- bination in CHO:5 cells.

Material and methods

Chemicals

MMS, BrdUrd, and MMC were obtained from Sigma and HN2 (mustine hydrochloride) from Boots Company. BPDE (anti-7,8-dihydrodiol-

9,10-epoxide) was kindly provided by Dr. Bengt Jernstr6m, Department of Toxicology, Karolinska Institute. G 418 was obtained from Gibco.

Cell lines A 'wild-type' laboratory line of CHO cells,

CHO-3, and the mutant cell line EM9 were originally obtained from Dr. Siv Ljungqvist, In- stitute of Environmental Medicine, Karolinska In- stitute, Stockholm. CHO:5 was isolated by transfection of CHO-3 cells with plII-14gpt and selection in MPA medium as described (Hellgren et al., 1988).

The mutant phenotype of EM9 was confirmed by its high spontaneous SCE frequency and an in- creased sensitivity (reduced plating efficiency as compared to CHO-3 cells) in medium containing BrdUrd and MMS.

Cells were grown in Dulbecco's modified Eagle's medium (DMEM, Flow) with 10°70 fetal calf serum (FCS, Gibco) supplemented with 20 mM L- glutamine, 150 IU/ml penicillin and 150 ~g/ml streptomycin.

TABLE 1

VECTOR INTEGRATION

Transfectant Approximate Number o f Frequency of

cell clone number of integration revertants truncated n e o sites b × 10 4

gene a pairs

EM9/p l I l -14gp t / 1 n.d. 2 8500 c

2 1 4 0.9

3 1 3 24.6 4 0 d 2 0.07

CHO:5 e 1 3 1.2

a See Material and methods for details for estimation o f the

number of tandem fragments.

b Note that this is not necessarily the number of active sites, only that they contain n e o - h y b r i d i z i n g material. The high frequency is most probably caused by formation of

an active n e o gene during the transfection. Quantif icat ion shows that this clone does not contain an intact pair o f n e o gene fragments.

e Data from Hellgren et al. (1988).

DNA transfection and isolation and characteriza- tion o f MPA-resistant cell clones

The procedure used was as described previously for CHO cells (Hellgren et al., 1988). In brief, plasmid plII-14gpt (Fig. 1) was introduced into EM9 cells by calcium phosphate precipitation, followed by selection of stable transformants in medium containing 25 #g/ml MPA. After 12-14 days in selective medium individual clones were isolated and expanded in non-selective medium. In 4 of these independent clones (Table 1) integration of the vector was studied by restriction analysis on Southern blots. The number of integration sites was determined from the number of fragments recognized by a neo-specific probe in DNA digested by SacI (which has no cleavage site within the vector). This procedure identifies not only vec- tor inserts with an intact pair of neo gene fragments capable of recombining, but also other kinds of neo-containing vector fragments that may have rearranged during transfection and clonal outgrowth. The number of tandemly arranged neo gene pairs was determined by comparison between aprt and neo-specific hybridization in the same blots of BamHI- and BamHI/HindlII-digested DNA using semi-quantitative densitometry (fur- ther details in Hellgren et al., 1988). The frequency of spontaneous revertants in each of the MPA- resistant EM9 clones was determined by selection in medium containing G 418 (1200 tzg/ml) as described below.

Clone EM9:2 was found to have integrated neo- hybridizing vector sequences at 4 sites, and ap- peared to have retained the full size of tandemly ar- ranged neo gene fragments in only 1 of these sites (Table 1). In the G 418-resistant revertants of this clone (spontaneous as well as induced) rearrange- ment was found to occur in only 1 of these sites, and this site was always the same. This suggests that EM9:2 contains only 1 set of tandemly arrang- ed neo gene pairs which takes part in the type of recombination studied. The spontaneous frequen- cy of revertants was similar to that in the previously studied clone CHO:5 (Table 1). Thus, EM9:2 was considered a suitable clone for comparison with CHO:5, and only this clone was used in subsequent experiments.

Chemical treatment and selection o f revertants Cells were seeded in 75-cm 2 bottles at a density

106 cells per bottle. After 24 h, increasing concen- trations of freshly made solutions of chemicals to be tested for induction were added to the medium at the indicated concentrations. Treatment was ter- minated after 24 h, the cells were washed and in- cubated in fresh medium. The cells were then in- cubated for a further 24 h before they were tryp- sinized, counted and plated to select for revertants. 200,000 cells/plate were seeded in G 418-con- taining medium (1200 t~g/ml), and 200-500 cells/plate in medium without G 418 were used for determination of plating efficiency.

MMS, HN2 and MMC Were diluted to the ap- propriate concentrations with media just before addition of the compounds to the bottles. BPDE was dissolved in tetrahydrofurane, which was removed with nitrogen gas. The remaining dry BPDE was dissolved in dimethyl sulfoxide (DMSO) and immediately added to the bottles. BrdUrd was dissolved in sterile water and stored cold protected from light. After addition of BrdUrd to the culture medium, the bottles were wrapped in aluminum foil and subsequently handl- ed in dimmed light to reduce the risk of photolysis. The effect of the length of the expression time on the frequency of revertants was studied by transfer- ring cells directly after MMS exposure (50/~M) and 24 or 48 h later to selective medium. No differences

TABLE 2

INDUCTION OF G 418 RESISTANCE IN EM9:2 AND

CHO:5 WITH VARIOUS AGENTS a

Plating Dose b Frequency Induced

efficiency × 10 -5 frequency

× 10 -s

CHO:5

MMC (t~g/ml) 100 0.0 9.7 -

92.3 10.8 1.05

72.7 8.6 0.0

68.0 12.6 2.9 38.3 0.05 16.9 7.2

21.5 18.0 8.3

11.4 0.1 18.7 9.0

TABLE 2 (continued)

Plating Dose b Frequency

efficiency × 10 -5 Induced

frequency × 10 -5

X-ray (Gy)

100 0 6.8 -

76.7 2 6.4 0.0

64.6 5.8 0.0

45.4 4 5.5 0.0

33.9 6.8 0.0

25.5 7.6 0.8 16.8 7 3.9 0.0

BPDE (/~M) 100 0 12.3

85.6 0.15 17.0 4.7

68.9 17.3 5.0

43.5 0.4 17.6 5.3

25.3 12.2 0.0

9.4 0.6 8.0 0.0

EM9:2

MMS (ttM)

100 0 12.6

98.6 1 15.4 2.8

70.3 16.9 4.3

55.9 3 17.9 5.3

43.9 19.9 7.2

33.1 29.7 17.1

24.4 6 45.4 32.8

HN2 (/zM)

100 0 11.2

85.8 1 11.7 0.5

76.7 13.0 1.9

67.8 13.0 1.9

59.8 10.5 0.0

45.0 3 14.5 3.3

23.2 17.1 5.9

8.6 5 28.2 17.0

BrdUrd (#M)

100 0 8.3

96.7 0.05 10.3 2.0

84.2 8.8 0.5

74.4 12.2 3.9

65.8 1.0 11.9 3.6

51.7 13.7 5.4

32.5 16.8 8.5

14.8 10.0 17.6 9.3

a Each figure shows the mean of 2-5 experiments per dose and 3-8 plates per experiment.

b The effect of doses varied somewhat between experiments,

therefore only the lowest, middle and highest doses are in- dicated.

were found between the three time points in the number of induced revertants (results not shown).

X-irradiation The X-ray source was an AEG Telefunken MB

350/I with a 7-mm Be filter and a Pantak type HF350C power generator operated at the National Institute of Radiation Protection, Stockholm. Cells were exposed to X-ray radiation in 80-cm 2 bottles in DMEM supplemented with 5% FCS and standard concentrations of penicillin and strep- tomycin. Immediately after exposure, the medium was changed to DMEM containing 10070 FCS. After 24 h cells were plated for selection of rever- tants as described above.

50-

~ 3 o -

%2o- ~ 10-

"O

c- O-

20- C

g ~ 1 5 -

a l O -

"o

-lJ t-- O-

A

' 8 '0 ' 6'0 ' 4'0 ' 2'0 1'0 Percent surv]val

B

'8 '0 ' 6'0 ' 4~0 ' 2'0 1'0 Percent survival

Fig. 2. (A) Comparison between CHO:5 and EM9:2 with regard

to the induction of G 418-resistant revertants by MMS. Treat-

ment protocol as described in Material and methods. Note that

the MMS concentration used for EM9 clone 2 was 2-6 t~M, a dose 10 times lower than for clone 5. (B) Comparison between CHO:5 and EM9:2 with regard to the induction of G 418-resistant revertants by HNz. The concentrations of HN2 for both clones were 1-6 #M. The data for CHO'5 are taken from

Hellgren et al. (1988). O, EM9:2; O, CHO'5.

Results

The frequency of spontaneous G 418-resistant revertants in EM9:2 and CHO:5 remained at the same level (about 1 x 10 -4 , Table 1), and both cell lines retained their characteristic pattern of in- tegrated vector fragments on Southern blots during the whole experimental period. Thus, the vector se- quences appear to be stably integrated in the cellular genome.

As reported earlier (Hellgren et al., 1988), CHO:5 cells showed a dose-dependent increase in the frequency of G 418-resistant revertants after treatment with MMS in the dose range 10-50 #M. EM9:2 cells showed hypersensitivity to MMS, in- dicating that the mutant phenotype of the parent cell line EM9 was retained in this clone. When the concentration of MMS was reduced 10-fold to achieve the same survival in EM9:2 as in CHO:5, the 2 cell lines showed a very similar dose- dependent increase in the frequency of G 418-resistant revertants (Table 2, Fig. 2A).

EM9:2 was found to have about the same sen- sitivity towards HN2 as CHO:5, but in contrast to the latter cell line, HN2-treated EM9:2 cells showed an increased frequency of G 418-resistant rever- tants. The induced frequency was about one-third of that obtained after MMS treatment at similar survival rates (Table 2, Fig. 2B). EM9 is hypersen- sitive to the cell killing and SCE-inducing effects of

10- c- a -~ - -

x~ 6- ~ ..~

d g 4" "U ~ "

~ 2-

" 0

r'- O"

' 8 ' 0 ' 6'0 ' 4'0 ' 2'0 1'0 Percent survival

Fig. 3. Induction of G 418-resistant revertants by BrdUrd in EM9:2. The induced frequency of revertants (background fre- quency subtracted) has been plotted as a function of survival. Data from Table 2.

BrdUrd (Pinkel et al., 1985), which in normal cells acts as a weak mutagenic and clastogenic agent (San Sebastian et al., 1980). We found BrdUrd to be lethal to EM9:2 cells at much lower concentra- tions than CHO:5 cells. As shown in Fig. 3, BrdUrd induced a dose-dependent increase in the frequency of G 418-resistant revertants in EM9:2 cells over the range 0.05-10 I~M, indicating a weak but significant recombinogenic activity.

The very clear induction of revertants by MMS and the likewise clear absence of an effect by HN2 in CHO:5 prompted studies of other mutagenic agents in this cell line. MMC, BPDE and X-ray were chosen to cover a broad range of DNA lesions. Only MMC induced a significant increase in the frequency of G 418-resistant revertants (Table 2, Fig. 4). The effect was small and seeming- ly limited to a narrow dose range. BPDE and X-ray showed no dose-dependent effect in CHO:5 cells up to doses causing about 10% survival (Table 2). In EM9:2, a weak effect of X-ray was found in preliminary experiments, but this observation re- quires further study.

Discussion

Few reports in the literature concern the effect of mutagens on recombination in mammalian cells. In an assay based on recombination between overlap-

10" c-

"~ 8" >

..Q 6-

~ 2- ~ Ck

c O"

J

I i i i i i

80 60 4'0 2'0 Percent survival

i

10

Fig. 4. Induction of G 418 resistance by MMC in CHO:5. The induced frequency of revertants (background frequency sub-

tracted) has been plotted as a function of survival. Data from Table 2.

ping thymidine kinase (TK) gene fragments in a stably integrated vector construction Lin and Sternberg (1984) reported a dose-dependent in- crease in the number of TK ÷ revertants after treat- ment with MMC while the tumor-promoting agents mezerein and TPA lacked this effect. Wang et al. (1988) used a similar system, and showed that MMC, BPDE, MNNG and UV light but not X-rays were able to induce homologous recombination between duplicated TK genes in a chromosomally integrated plasmid. Mouse cells were used in both of these studies. While the reversion to TK + in the work of Wang et al. (1988) seemed mainly due to gene conversion, Lin and Sternberg (1984) reported at least some of the recombination events to in- volve unequal SCE.

Our assay system makes use of CHO ceils, and the n e o gene is the target for recombination. In spite of the differences between the systems our results agree with the previous studies in showing that recombination between stably integrated target sequences in mammalian cells can be induc- ed by mutagenic agents, and that there is a dif- ference in the recombinogenic activity between mutagens. This suggests that recombinational events in these systems are not primarily induced by DNA damage in general, but specifically related to certain types of DNA damage or DNA repair mechanisms. However, there are some differences between the systems and results that deserve com- ment. The negative result we have obtained for BPDE is in contrast to the results obtained by Wang et al. (1988). In their system most of the in- duced recombinants by BPDE were apparently due to gene conversion, and a mutation caused by an 8-bp insertion might be easier to change through a gene conversion event than deletions over 250 bp in size, as in our system.

The negative results obtained for X-ray (this study and Wang et al., 1988) might be related to the observation that ionizing radiation induces large deletions in some genes in CHO cells (Urlaub et al., 1986; Vrieling et al., 1985). Deletions larger than about 1500 bp would not be scored as recom- binogenic in the present system.

The differences between the ability of HN2

(negative) and MMC (positive) to induce an in-

creased frequency of revertants in CHO:5 ceils in

our system is surprising, since both agents give rise to cross-links and mono-adducts in DNA.

However, MMC is metabolized (Mason et aI., 1982) to an activated form, a semiquinone, which

has the possibility of forming radicals which in turn

can cause further damage to DNA. Thus, the ac-

tivity of MMC observed by us as well as Wang et al.

(1988) and Lin and Sternberg (1984) may be due to

the wide spectrum of DNA damage caused by this

agent.

The cell line EM9 has previously been shown to

be defect in double-strand break repair and to be hypersensitive to alkylating agents (Thompson et

al., 1982). BrdUrd induces an increased frequency

of SCE in EM9, even at low concentrations (Pinkel

et al., 1985). Results from these studies suggest that

EM9 had a lower ability to recombine sequences

extrachromosomally. The results from BrdUrd

treatment and from transfections indicate that EM9 is defective in a step involved in recombina-

tion (Hoy et al., 1987). Nevertheless, the ability of MMS to induce

revertants appears to be similar for CHO:5 and

EMg:2 when the 2 cell lines are compared on the

basis of survival, i.e., a 10-fold lower concentra-

tion was used for EM9:2. This suggests that the same type of damage is responsible for both cell

lethality and the observed recombinogenic activity

in the 2 cell lines. In the case of HN2, the damage

responsible for cell killing appears not to be related

to the specific genetic defect in EM9 since there is

no difference in survival after HN2 treatment in the

2 cell lines. In spite of this, HN2 induced revertants

in EM9:2 but not in CHO:5. The explanation for this result may be that HN2-induced DNA damage

interacts with a defective DNA repair system in the

mutant cell line EM9. The weak but significant recombinogenic activi-

ty of BrdUrd in EM9:2 together with the hypersen- sitivity to cell killing and SCE induction by BrdUrd suggests that the specific genetic defect in EM9 may be related to its impaired ability to repair certain types of BrdUrd-induced, SCE-promoting DNA damage. This observation indicates a possible

mechanistic relationship between (unequal) SCE

and recombination in the present system.

Acknowledgements

We thank Jan-Olov Grindborg at the National

Institute for Radiation Protection for kind help

with X-ray exposures. The research was sponsored

by the Swedish Medical Research Council (7905),

Marcus Borgstr6m Foundation, Nilsson-Ehle

foundation and Karolinska Institutet.

References

Cavenee, W.K., A. Koufos and M.F. Hansen (1986) Recessive mutant genes predisposing to human cancer, Mutation Res., 168, 3-14.

Friedberg, E.C. (1988) Deoxyribonucleic acid repair in the yeast Saccharomyces cerevisiae. Microbiol. Rev., 52, 70-102.

Hellgren, D., H. Luthman and B. Lambert (1988) Induced recombination between duplicated neo genes stably in- tegrated in the genome of CHO cells, Mutation Res., in press.

Hoy, C.A., J.C. Fuscoe and L.H. Thompson (1987) Recom- bination and ligation of transfected DNA in CHO mutant EM9, which has high levels of sister chromatid exchange, Mol. Cell. Biol., 7, 2007-2011.

Kucherlapati, R. (1986) Homologous recombination in mam- malian somatic cells, in: R. Kucherlapati (Ed.), Gene Transfer, Plenum Press, New York, pp. 363-381.

Letsou, A., and R.M. Liskay (1986) Intrachromosomal recom- bination in mammalian cells, in: R. Kucherlapati (Ed.), Gene Transfer, Plenum Press, New York, pp. 383-409.

Lin, F.-L., and N. Sternberg (1984) Homologous recombina- tion between overlapping thymidine kinase gene fragments stably integrated into a mouse cell genome, Mol. Cell. Biol., 4, 852-861.

Mason, R.P., W.G. Harrelson, B. Kalyanraman, C. Mottley, F.J. Peterson and J.L. Holtzman (1982) Free radical metabolites of chemical carcinogens, in: D.C.H. McBrien and T.F. Slater (Eds.), Free Radicals, Lipid Peroxidation and Cancer, Academic Press, New York, pp. 377-400.

Pinkel, D., L.H. Thompson, J.W. Gray and M. Vanderlaan (1985) Measurement of sister chromatid exchanges at very low bromodeoxyuridine substitution levels using a monoclonal antibody in Chinese hamster ovary cells, Cancer Res., 45, 5795-5798.

San Sebastian, J.R., J.P. O'Neill and A.W. Hsie (1980) Induc- tion of chromosome abberation, sister chromatid exchanges, and specific locus mutations in Chinese hamster ovary cells by 5-bromodeoxyuridine, Cytogenet. Cell Genet., 28, 47-54.

Smith, G.R. (1988) Homologous recombination in prokaryotes, Microbiol. Rev., 52, 1-28.

Thompson, L.H., K.W. Brookman, L.E. Dillehay, A.V. Car- rano, J.A. Mazrimas, C.L. Mooney and J.L. Minkler (1982) A CHO-cell strain having hypersensitivity to mutagens, a defect in DNA strand-break repair, and an extraordinary baseline frequency of sister-chromatid exchange, Mutation Res., 95,427-440.

Urlaub, G., P.J. Mitchell, E. Kas, L.A. Chasin, V.L. Funange, T.T. Myoda and J. Hamlin (1986) Effect of gamma rays at the dihydrofolate reductase locus: deletions and inversions, Somat. Cell Mol. Genet., 12, 555-566.

Vrieling, H., J.W.I.M. Simons, F. Arwert, A.T. Natarajan and A.A. van Zeeland (1985) Mutations induced by X-ray at the HPRT locus in cultured Chinese hamster ceils are mostly large deletions, Mutation Res., 144, 281-286.

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.

West, S.C. (1988) Protein-DNA interactions in genetic recom- bination, Trends Genet., 4, 8-13.

Communicated by K. Sankaranarayanan