Role of Methylation in the Modification and Restriction of Chloroplast DNA in Chlamydomonas (Genetics/Maternal Inheritance/5-Methylcytosine) WILLIAM G

Proc. Natl. Acad. Sci. USA Vol. 76, No. 3, pp. 1390-1394, March 1979 Genetics

Role of methylation in the modification and restriction of chloroplast DNA in Chlamydomonas (genetics/maternal inheritance/5-methylcytosine) WILLIAM G. BURTON*, CONSTANCE T. GRABOWY, AND RUTH SAGERt Sidney Farber Cancer Institute, 44 Binney Street, Boston, Massachusetts 02115 Contributed by Ruth Sager, December 29, 1978

ABSTRACT The different metabolic paths followed by MATERIALS AND METHODS homologous chloroplast DNAs of maternal and paternal origins in zygotes of Chlamydomonas were examined by prelabeling Cells and Media. The strains used-the wild-type 21gr parental cells, before mating them, with [3H]adenine, [3Hlthy- (maternal), mating type plus (mt+), and the streptomycin-re- midine, and [3Hldeoxycytidine. Within 6 hr after mating, ma- sistant 5177D (paternal), mating type minus (mt-)-have been ternal chloroplast DNA was extensively methylated to 5-methylcytosine and its bouyant density decreased. Paternal chloro- described (10). They were grown in minimal medium (10) plast DNA was largely degraded. Some radioactivity from bubbled with 5% CO2. For radioisotope labeling experiments, deoxycytidine of maternal origin reappeared in thymine, and cultures were grown for six or seven doublings in the presence residual paternal DNA contained radioactivitv in a base ten- of the labeling compound to a density of 5 X 106 cells per ml. tatively identified as uracil. These results conirm and extend Radioisotope was added to 40 ml of culture medium: 2mCi (1 our previous findings and support our hypothesis that modifi- = cation (methylation) and restriction enzymes determine ma- Ci 3.7)X 1010 becquerels) of [3H]deoxycytidine; 500 ,uCi of ternal inheritance of chloroplast DNA and that the two parental [3H]thymidine; 250 ,uCi of ['4C]adenine; 1 mCi of [3H]adenine. DNAs have different metabolic fates within the zygote. For gamete induction and zygote formation, cells were washed once in nitrogen-free minimal medium diluted 1:5 at pH 8 and In 1972, it was postulated (1) that a modification-restriction resuspended in this medium at 5 X 106 cells per ml. Cells dou- mechanism analogous to that responsible for host-range re- bled twice in 24 hr, producing about 2 X 107 gametes per ml. striction of bacteriophage (2) was the molecular basis of ma- Equal numbers of gametes of the two mating types were mixed; ternal inheritance of chloroplast genes in Chlamydomonas. mating was monitored by phase-contrast microscopy and by This hypothesis was based upon the finding that chloroplast plating. Zygote formation reached >90% within 2-3 hr. The DNAs of male and female origins, indistinguishable in vege- yield of viable zygotes in each experiment and the percentage tative cells, follow different paths in the zygotes that form after of maternal and exceptional (i.e., biparental and paternal) zy- gametic fusion (fertilization). Chloroplast DNA from the female gotes was determined by plating at dilutions on agar and replica parent persists in the zygote but is shifted to a lighter bouyant plating to streptomycin agar as described (10). The frequency density, whereas that from the male is lost soon after zygote of exceptional zygotes averaged 0.5% ± 0.26 in 10 experiments formation (1). Thus, the density shift could result from enzy- and the yield of viable zygotes was >90%. matic modification of chloroplast DNA of female origin; and DNA Extraction and Fractionation. Isotopically labeled the loss of chloroplast DNA of male origin could result from gametes and zygotes were harvested by centrifugation and attack by a restriction enzyme upon unmodified DNA mole- washed once in NET buffer (0.15 M NaCl/0.1 M Na2EDTA/50 cules. mM Tris-HCl, pH 8) and resuspended in NET at 5 X 108 cells The modification-restriction hypothesis was further sup- per ml. The cells were then mixed with 1010 cells of unlabeled ported by studies in which maternal transmission of chloroplast 21gr vegetative cells at the same cell density and 90% broken genes was converted to a biparental or paternal pattern by in a French pressure cell [1200 psi (8.27 mPa) for gametes and UV-irradiation of female gametes before mating (3), by various To zygotes, 3000 psi for 6-hr zygotes, and 4000 psi for 24-hr drug pretreatments (4), or by a mutant nuclear gene, mat-i (5). zygotes]. Each sample was lysed with 2% sodium dodecyl sulfate Complementing the genetic studies, in studies of radioisotope plus 2% Sarkosyl at 50°C for 10 min and then frozen at -20°C. prelabeled chloroplast DNAs from zygotes after UV-irradiation Lysates were thawed, diluted 1:1 with NET, and digested with and in crosses with mat-i it was found (6, 7) that chloroplast Pronase (200 ,tg/ml) added three times at 1-hr intervals at 37°C DNA of male origin was preserved in parallel with the genetic and further incubated for 18 hr at 50°C. markers. These experiments demonstrate that the molecular Lysates were then extracted twice with equal volumes of pattern of inheritance of chloroplast DNA parallels that of distilled phenol saturated with 0.1 M Tris-HCl (pH 8) and once chloroplast genes in Chlamydomonas (reviewed in refs. 8 and with an equal volume of chloroform/isoamyl alcohol, 24:1 9). (vol/vol). Remaining CIA was removed by ether extraction. In this paper, we report that the bouyant density decrease Nucleic acids were precipitated with 2 vol of cold absolute of zygotic chloroplast DNA occurs in concert with the extensive ethanol at -20°C for 30 min. The precipitate was pelleted, methylation of cytosine residues to form 5-methylcytosine and dissolved in 15 ml of TE buffer (10 mM Tris/0.1 mM EDTA, that a low level of methylation is already present in the gametic pH 8), and dialyzed for 2 hr against TE. Samples were dialyzed chloroplast DNA of female origin and is greatly increased in in transmission is zygotes which the genetic pattern of maternal. Abbreviation: HPLC, high-pressure liquid chromatography. These results provide definite support for the modification- * Present address: Bethesda Research Laboratory, 411 Stonestreet restriction mechanism of non-Mendelian inheritance of chlo- Avenue, Rockville, MD 20850. roplast genes in Chlamydomonas. t To whom reprint requests should be addressed. 1390 Genetics: Burton et al. Proc. Natl. Acad. Sci. USA 76 (1979) 1391 against 0.15 M NaCI/0.015 M Na citrate, pH 7, at 370C and treated with RNase T1 (20 units/ml) and pancreatic RNase (200 ,qg/ml) for 18 hr. Samples were extracted with phenol as before until the interface was clear. The amount of DNA was then determined spectrophotometrically. One milligram of DNA 1.6 in TE buffer was mixed with CsCl to give a density of 1.706 1.2 g/ml. Gradients were run in a vertical Ti 65 rotor at 40,000 rpm for 18 hr at 20'C and fractionated to separate chloroplast and 0.8 nuclear DNAs; appropriate fractions were pooled and rerun. Chloroplast DNA fractions from the second CsCl gradient were 0.4 pooled on the basis of absorbance profiles and isotope incorporation and dialyzed against TE buffer to remove CsCI. Hydrolysis and Chromatographic Separation of Bases. DNA samples from the second CsCl gradient were extracted 0 with chloroform/isoamyl alcohol, ethanol precipitated, redis- x solved in 200,gl of 98% anhydrous formic acid, and heated to E 1800C in sealed pasteur pipettes for 25 min. The hydrolysates I'a were cooled to room temperature, dried, redissolved in 100,. a: of elution buffer (20mM ammonium carbonate, pH 9.98), and heated to 680C for 5 min. The bases were identified by the method of Singhal (11, 12). Each total hydrolysate was applied to a high-pressure liquid chromatography (HPLC) Aminex A-6 column operated at a flow rate of 0.25 ml/min. The eluate was monitored by an UV optical unit at 254 nm. Fractions (0.25 ml) were collected, 10 ml of Aquasol (New England Nuclear) was added to each vial, and the radioactivity was determined in a LS-9000 Beckman scintillation counter. The elution positions of the bases were determined with a set of standards: cytosine and 5-methylcytosine from Sigma; adenine, guanine, thymine, and uracil from Calbiochem. Recovery of bases was monitored with DNAs of known base composition: A DNA, maize nuclear DNA containing 5-methylcytosine (gift of L. Bogorad), and 4 8 12 16 20 24 Chlamydomonas nuclear and chloroplast DNAs. ['4C]Adenine, Fraction [3H]adenine, [3H]thymidine, [5-3H]deoxycytidine, and [G- FIG. 1. Bouyant density shift in zygotic chloroplast DNAs. DNA 3H]deoxycytidine were from New England Nuclear. was extracted from replicate cultures in which strain 21gr (female) was prelabeled with [3H]thymidine and fused with unlabeled 5177D RESULTS (male); zygotes were sampled immediately after fusion (Top), at 6 hr Bouyant Density Shift. In 1972 it was reported (1) that later (Middle), and at 24 hr later (Bottom). Carrier cells were added, and DNA was extracted and fractionated in two successive CsCl chloroplast DNA extracted from zygotes as early as 6 hr after gradients. Absorbance tracing of carrier DNA from second CsCl zygote formation has undergone a bouyant density shift to a gradient identifies position of chloroplast (C) and residual nuclear value approximately 5 mg/cm3 lighter than that of homologous (N) peaks and provides the basis for estimating bouyant density of DNAs from vegetative cells and from gametes of both mating [3H]thymidine-labeled maternal (0) and paternal (0) zygotic types. We confirmed this observation in the experiments shown chloroplast DNA, assuming linearity of the gradient. in Fig. 1, in which the chloroplast DNA of each parent was prelabeled by growth in medium containing [3H]thymidine. Each parent was then mated to unlabeled cells of the opposite sorbance profiles of chloroplast DNAs corresponded well with mating type in a pair of reciprocal crosses, and chloroplast DNA published values. The identities of the bases were further con- was extracted from zygotes immediately after mating (To) and firmed in experiments in which cells were prelabeled with after 6 and 24 hr. The data are from a second preparative CsCl [3H]adenine, [3H]thymidine, or [3H]deoxycytidine. gradient containing chloroplast DNA pooled from the first Results are given in Fig. 3 and Table 1 of two experiments gradient in which a small amount of nuclear DNA was present. in which parental cells were prelabeled with [3H]deoxycytidine By utilizing the absorbance peaks of nuclear and chloroplast and reciprocal crosses were made between labeled and unla- DNAs as reference points and assuming a linear gradient, we beled cells. Chloroplast DNAs were prepared from vegetative have estimated a bouyant density decrease of about 6-10 cells, from gametes, and from zygotes recovered 6 and 24 hr mg/cm3 in the tritiated chloroplast DNA of maternal origin. after cell fusion. The DNAs were hydrolyzed with formic acid The homologous DNA from the paternal parent was largely and the resulting bases were separated by HPCL on Aminex degraded within 6 hr after zygote formation. A-6. When the female parent (21gr) was labeled, a definite Presence of 5-Methylcytosine in Zygote Chloroplast DNA. peak in the position of 5-methylcytosine was seen in both the To account for the density shift, experiments were designed to 6- and 24-hr zygote preparations (Fig. 3 upper), amounting to look for altered bases in zygote chloroplast DNAs. Purified 22 and 27%; respectively, of the total radioactivity (Table 1). DNA preparations were hydrolyzed with formic acid and the In contrast, when the male parent (5177D) was prelabeled (Fig. bases were separated by HPLC. A typical profile based on the 3 lower), only a low level of incorporation was seen in the 5- absorbances of bases recovered after formic acid hydrolysis of methylcytosine position. This dramatic difference in labeling chloroplast DNA is shown in Fig. 2, together with an absorbance pattern represents definitive evidence that chloroplast DNA profile of a mixture of authentic bases run under the same of maternal origin is selectively methylated during the first 6 conditions. The molar ratios of bases computed from the ab- hr of zygote formation. No labeling was detected in the 5- 1392 Genetics: Burton et al. Proc. Nati. Acad. Sci. USA 76 (1979)

1400 = _o . 1200 " E C4 a '5c 1000 - bG N- 800 E z a a 600 V 0 0 z E 400 c0 a Ccm1 0 200 > I N ._-lC 0 x

S E a 10 15 20 25 30 35 ]'O Fraction FIG. 2. Elution profile of vegetative chloroplast DNA compared with standards. - -, Elution profile from HPLC Aminex A-6 column V- ofsix authentic bases: U, uracil; T, thymine; G, guanine; A, adenine; z C, cytosine; 5MeC, 5-methylcytosine. -, Elution profile from du- a plicate column separation of a formic acid hydrolysate ofvegetative 0 0 chloroplast DNA from strain 21gr, to which 5-methylcytosine (*) was E added as internal position marker. o C I methylcytosine peaks of chloroplast DNA from vegetative cells of either parent, but a small peak of radioactivity was found in the maternal gamete DNA. This 5-methylcytosine peak was Fraction also seen in large-scale preparations of unlabeled gamete FIG. 3. Chloroplast DNAs from gametes and zygotes in which DNA. one parent was prelabeled with deoxycytidine. One strain was pre- similar to those described above in 3 and labeled with [G-3H]deoxycytidine (50 MCi/ml), gametes were prepared Experiments Fig. and fused with unlabeled gametes ofthe other strain, and zygotes were Table 1 were carried out with unlabeled cells and with cells sampled at 6 and 24 hr. Chloroplast DNAs were purified and hydro- prelabeled either with [3H]thymidine (reciprocal crosses of lyzed with formic acid, and bases were separated on HPLC Aminex labeled X unlabeled cells) or with adenine (one parent labeled A-6 column. Positions of marker bases are shown: U, uracil; T, thymine; C, cytosine; 5MeC, 5-methylcytosine. Peak at left is void volume, containing [3H]deoxyribose. (Upper) Strain 21gr labeled. Table 1. Percentage distribution of radioactivity from (Lower) Strain 5177D labeled. [G-3Hldeoxycytidine-prelabeled cells present in chloroplast DNAs Total Distribution, % incorporation,* with [3H]adenine, the other with [14C]adenine). In the experi- U T C 5MeC dpm ments with unlabeled cells, performed without carrier DNA, the amount of 5-methylcytosine present in 6- and 24-hr zygotes Maternal parent prelabeled was estimated as representing 4-8% of the total absorbance Vegetative (estimated from absorbance tracings). In the experiments with cells (21gr) 38 8 54 9,861 prelabeled cells, label appeared only in thymidine in the thy- Gametes 22 13 63 2 17,049 midine labeling experiment (Fig. 1) and only in adenine and Zygotes at 6 hr 4 21 53 22 6,120 guanine in the adenine labeling experiment (data not Zygotes at 24 hr 13 42 18 27 5,254 shown). Nuclear DNAs were examined in the experiments using Paternal parent prelabeledt Vegetative unlabeled cells and those using cells labeled with [3H]adenine cells (5177D) 12 9 80 7,764 and [14C]adenine. No peaks were seen in the position of 5- Gametes 6 10 83 - 11,755 methylcytosine or uracil. Nuclear DNAs were unlabeled in the Zygotes at 6 hr 33 10 53 4 3,015 experiments with [3H]thymidine because the thymidine kinase Zygotes at 24 hr 43 6 45 5 3,421 responsible for phosphorylation is a chloroplast enzyme (13). Different Metabolic Paths of Maternal and Paternal Vegetative cells were grown with 1 mCi of [3H]deoxycytidine per Chloroplast DNAs. In addition to these anticipated findings, 40-ml culture; other cells were grown with 2 mCi/40-ml culture. U, uracil; T, thymine; C, cytosine; 5MeC, 5-methylcytosine. additional peaks of radioactivity were observed in other posi- * Excluding void volume. tions. When the maternal parent was prelabeled (Fig. 3 upper) t In this experiment, a peak between U and T-contained 4% of the the percentage of total label in cytosine decreased substantially counts in the 6- and 24-hr preparations. in the interval between 6 and 24 hr after mating (Table 1), Genetics: Burton et al. Proc. Natl. Acad. Sci. USA 76 (1979) 1393 much of it reappearing in thymine. When the paternal cells DNA. (iii) This position is not labeled when [5-3H]deoxycyti- were prelabeled (Fig. 3 lower), the relative incorporation into dine is used as prelabel. (Both the 4 and 5 positions of the py- thymine did not change but the radioactivity lost from cytosine rimidine ring are tritiated in [G-3H]deoxycytidine and conse- reappeared in the position of uracil, mostly within the first 6 quently the base remains radioactive after the 3H in the 5 po- hr after mating. sition has been replaced by an unlabeled methyl group.) As noted in Table 1, losses of total radioactivity occurred in The decrease in bouyant density in zygotic chloroplast DNA all zygote preparations compared with the gametic aliquots, of maternal origin (Fig. 1) of approximately 6-10 mg/cm3 can in part owing to the difficulty in cracking the zygote wall to be accounted for (24) by the extent of methylation we have release DNA. In addition, a differentially greater loss was seen estimated from the absorbance profiles of large-scale nonra- of paternal than of maternal label. This loss was less than that dioisotope-labeled preparations: 4-8% (molar basis) of the total seen with [3H]thymidine label (Fig. 1), suggesting that some chloroplast DNA. reincorporation from paternal into maternal DNA occurs with The results of reciprocal crosses between [3H]thymidine- deoxycytidine but not with thymidine. labeled and unlabeled gametes are important also in unam- This loss and redistribution of label provides further evidence biguously demonstrating the degradation of chloroplast DNA of the breakdown of the chloroplast DNA of male origin and of paternal origin. When adenine is used to prelabel chloroplast suggests that some turnover is occurring in the chloroplast DNA DNA, a significant precursor pool from the breakdown of ri- of female origin. Thus, the increase in radioactivity in the po- bosomal RNAs in gametes (25) becomes available for incor- sitions of thymine and uracil must reflect previously unde- poration into newly synthesized DNA and confuses the iden- scribed processes of degradation and resynthesis in which the tification of paternal DNA. This pool problem led to Chiang's precursor pools available within chloroplasts of paternal and erroneous conclusion that both parental chloroplast DNAs are maternal origins are different. If confirmed, the presence of physically conserved in zygotes (26). Previous experiments using uracil is an interesting clue that merits further investigation. either ['5N]adenine (1) or [3H]adenine (7) as prelabel established the loss of paternal DNA in 6- and 24-hr zygotes, but the results were less definitive than those shown here with [3H]thymi- DISCUSSION dine. This paper describes the different metabolic paths that chlo- In addition to methylation and degradation, other differences roplast DNAs of maternal and paternal origin follow in zygotes were found in the metabolic fates of chloroplast DNAs of ma- of Chlamydomonas during the first 6 hr after mating and ternal and parental origins. When [3H]deoxycytidine was used during the subsequent 18 hr. As previously described, pairs of as a prelabel of the maternal DNA, a large increase was seen gametes of opposite mating type fuse to form zygotes in which in the thymine peak, especially at 24 hr. This result suggests the chloroplasts remain unfused in separate compartments that, in the young zygotes, some synthesis or repair of chloro- within a common cytoplasm for some 5-6 hr (14). We now plast DNA of maternal origin is occurring, in which free report the different fates of homologous DNAs facilitated by deoxycytidine is converted to thymidine triphosphate and in- this temporary compartmentalization. During the first 6 hr, the corporated into DNA. chloroplast DNA of maternal origin is methylated to the extent No comparable increase in the thymine peak is seen in Fig. of 4-8% (molar basis) of 5-methylcytosine, and a concomitant 3 lower, in which the paternal parent had been prelabeled. On decrease is seen in bouyant density of this DNA. In the same the contrary, the poor recovery of total radioactivity indicated time interval, the homologous DNA of paternal origin is de- loss of paternal chloroplast DNA, but not as fully as with thy- graded as described (1, 3, 9) and as further documented here, midine. Thus, some recycling of deoxycytidine may occur, especially with thymidine labeling. In addition, the maternal leading to some incorporation of this paternal label into ma- chloroplast DNA undergoes further changes, mainly between ternal DNA. A similar explanation probably accounts also for 6 and 24 hr, suggesting some new synthesis or repair. the small 5-methylcytosine peak in Fig. 3 lower. These new findings provides strong evidence in support of In the residual paternal chloroplast DNA (Fig. 3 lower), there the hypothesis formulated earlier (1, 3, 15) that the maternal was a large incorporation into a peak tentatively identified as inheritance of chloroplast DNA in Chlamydomonas is deter- uracil from its position in the elution profile (Fig. 2). If no new mined at the molecular level by the action of DNA modification synthesis of paternal DNA is occurring, as shown by the thy- (methylation) and restriction enzymes, leading to modification midine incorporation experiment (Fig. 1), then the only source (i.e., methylation) of the maternal DNA and destruction of the of uracil in this residual DNA must be deamination of cytosine homologous paternal DNA. Additional support for this hy- residues on the macromolecule. To our knowledge, no enzyme pothesis comes from the isolation of a site-specific endonuclease with this activity has been described. We may speculate that (16) and of a methyl transferase (17) from Chlamydomonas. the accumulation of uracil is a signal to DNA degradation. A general role for modification-restriction systems in the Enzymes that remove uracil and repair DNA have been de- regulation of gene expression in eukaryotes has been proposed scribed (27). Perhaps Chlamydomonas contains a restriction (15), in part because of similarities between the chloroplast enzyme that degrades the DNA from which uracil has been genetics of Chlamydomonas and of Pelargonium (18, 19) and removed. in part because of parallels between these organelle systems and The findings reported in this paper have a direct bearing on the differential chromosome loss and inactivation seen in insects genetic data proposed to estimate the number of copies of and in mammals (15). A role for methylation of DNA in eu- maternal and paternal chloroplast DNAs present in zygotes (28, karyotic gene regulation has also been proposed by Scarano (20), 29). In particular, Wurtz et al. (29) reported that pretreatment by Holliday and Pugh (21), by Comings (22), and by Riggs of the maternal parent with FdUrd decreases the amount of (23). chloroplast DNA and leads to the retention of paternal chlo- Our identification of 5-methylcytosine is based on three lines roplast genes in zygotes. They proposed that these results rule of evidence. (i) The peak position in the elution profiles (frac- out our modification-restriction mechanism because "the tion 47 in Figs. 2 and 3) coincides with that of the authentic normal, maternal pattern of inheritance of chloroplast genes base. (ii) This position is labeled when [G-3H]deoxycytidine but should not be altered by FdUrd treatment" (29). not when [3H]thymidine or [3H]adenine is incorporated into It has been found (5) that low-dose UV-irradiation of ma- 1394 Genetics: Burton et al. Proc. Natl. Acad. Sci. USA 76 (1979) ternal gametes just before mating, followed by photoreactiva- 8. Sager, R. (1972) Cytoplasmic Genes and Organelles (Academic, tion to repair damage to the DNA, prevented loss of the paternal New York). chloroplast genes. Two effects of UV irradiation were postulated 9. Sager, R. (1977) Adv. Genet. 19,287-340. (5): the blocking of induction of DNA "processing" enzymes 10. Sager, R. & Ramanis, Z. (1976) Genetics 83,303-321. in the zygote at low doses comparable to phage induction; and 11. Singhal, R. P. (1972) Arch. Biochem. Biophys. 152,800-810. 12. Singhal, R. P. (1974) Sep. Purif. Methods 3,339-398. direct inactivating effects on the irradiated chloroplast DNA 13. Swinton, D. C. & Hanawalt, P. C. (1972) J. Cell Biol. 54,592- at high doses. It is probable that, like UV radiation (30), FdUrd 597. has additional effects on enzymes of DNA synthesis, processing, 14. Cavalier-Smith, T. (1970) Nature (London) 228,333-335. and degradation and thus that FdUrd interferes with some or 15. Sager, R. & Kitchin, R. (1975) Science 189,426-433. all of the processes described in this paper. 16. Burton, W., Roberts, R. J., Myers, P. A. & Sager, R. (1977) Proc. The methylation, destruction, and complex enzymatic pro- Natl. Acad. Sci. USA 74,2687-2691. cessing of zygotic chloroplast DNAs described in this paper are 17. Sano, H. & Sager, R. (1979) Fed. Proc. Fed. Am. Soc. Exp. Biol., predicted by our hypothesis but not by others (26, 28, 29). We in press. show here unequivocally that the DNAs of maternal and pa- 18. Sager, R. (1975) in Genetics and Biogenesis of Mitochondria and in Chloroplasts, eds. Birky, C. W., Jr., Perlman, P. S. & Byers, T. ternal origins undergo very different processing reactions J. (Ohio State Univ. Press, Columbus, OH), pp. 252-267. the zygote. The effects of FdUrd and UV pretreatments on 19. Tilney-Bassett, R. A. E. (1975) in Genetics and Biogenesis of these reactions can now be dissected directly at the molecular Mitochondria and Chloroplasts, eds. Birky, C. W., Jr., Perlman, level. P. S. & Byers, T. J. (Ohio State Univ. Press, Columbus, OH), pp. This work was supported by National Institutes of Health Grant 268-308. GM-22874. 20. Scarano, E. (1967) in Comprehensive Biochemistry, eds. Florkin, M. & Stotz, E. H. (Elsevier, New York), Vol. 28, p. 55. 1. Sager, R. & Lane, D. (1972) Proc. Natl. Acad. Sci. USA 69, 21. Holliday, R. & Pugh, J. E. (1975) Science 187, 226. 2401-2413. 22. Comings, D. E. (1972) Exp. Cell Res. 74,383-390. 2. Arber, W. & Linn, S. (1969) Annu. Rev. Biochem. 38, 467- 23. Riggs, A. D. (1975) Cytogenet. Cell Genet. 14,9-25. 500. 24. Kirk, J. T. 0. (1967) J. Mol. Biol. 28, 171-172. 3. Sager, R. & Ramanis, Z. (1967) Proc. Natl. Acad. Sci. USA 58, 25. Siersma, P. W. & Chiang, K. S. (1971) J. Mol. Biol. 58, 167- 931-937. 185. 4. Sager, R. & Ramanis, Z. (1973) Theor. Appl. Genet. 43, 101- 26. Chiang, K. S. (1968) Proc. Natl. Acad. Sci. USA 60, 194. 108. 27. Lindahl, T., Ljungquist, S., Siegert, W., Nyberg, B. & Sperens, 5. Sager, R. & Ramanis, Z. (1974) Proc. Natl. Acad. Sci. USA 71, B. (1977) J. Biol. Chem. 252, 3286-3294. 4698-4702. 28. Adams, G. M. W. (1978) Plasmid 1, 522-535. 6. Schlanger, G. & Sager, R. (1974) J. Cell Biol. 63, 301a. 29. Wurtz, E. A., Boynton, J. E. & Gillham, N. W. (1977) Proc. Natl. 7. Sager, R. (1977) in Cell Biology: A Comprehensive Treatise, eds. Acad. Sci. USA 74,4552-4556. Goldstein, L. & Prescott, D. (Academic, New York), pp. 279- 30. Westergaard, O. & Pearlman, R. E. (1969) Exp. Cell Res. 54, 317. 309-313.