Molecular Mechanism for Genetic Recombination (Protein-Nucleic Acid Symmetry/Protein Binding/Polarity/Gene Conversion) HENRY M
Proc. Nat. Acad. Sci. USA Vol. 69, No. 9, pp. 2483-2487, September 1972
Molecular Mechanism for Genetic Recombination (protein-nucleic acid symmetry/protein binding/polarity/gene conversion) HENRY M. SOBELL Department of Chemistry, The University of Rochester, Rochester, New York 14627; and Department of Radiation Biology and Biophysics, The University of Rochester, School of Medicine and Dentistry, Rochester, New York 14620 Communicated by E. W. Montroll, June 8, 1972
ABSTRACT Symmetry considerations of protein- The binding of actinomycin to DNA and its specificity nucleic acid interaction suggest the existence of an alter- in inhibiting the RNA polymerase reaction suggest a primi- nate branched configuration for DNA induced by binding specific structural proteins to symmetricallyarranged poly- tive repressor-operator character for this complex that may nucleotide base sequences. The concept that such se- quences exist at the ends of genes or operons leads to a mo- lecular model for genetic recombination in eukaryotic cells. A B NUCLEASE SPECIFiCITY ACTINOMYCIN SPECIFICITY 3' 5' The molecular mechanism underlying genetic exchange in 5' 3' 5' 3' eukaryotic organisms is a subject that has received wide at- tention in recent years (1). It is generally thought that genetic I recombination begins with the pairing of two parental DNA duplexes on the nucleotide level (synapsis) to give a hybrid GC)3 *
A A B B 5'63 563' 56 3' 5' 3'
AT
T A T A TA T A G C G C C G C G C G A AI C G A T 'GC T T A T T A I ACC GC G A GA A T C G C C G A T A T A T GA CT T A T A CG GC G C I-,;-- -1 A T X~~ Vs\ \ A T AT A T H F G C .. D G C AT A T 3' . . T A 36 GCT A T TA AY TtK A TH T A T A A I G C G C TTATAGCCC G C A AG C G AACCG~~~~~ C G C G C G AT I(TAACGGT CCGTTA) FA A T T A GAT1 IATTGCCA TGGC AATI C T A G C A T C G C G C A GC G C C G 56 C G C G A T A T A T AT A T A T A_ A3' T A _T G G H A 3' B 3 TA 3 56 S' 3' GH '3 D DGD 5-A A T TA TA A T CG A T T A G C A 2T T A GC T A GC 3*3 * TAACGGTAC GTAC C G T T A *5'B CG T A C G 5'A- AATTGCCATGGcCATGGCAAT * 3' AA ~C G G C C GCT A C T A G c C G G TA T E TAACGGT AC C GGTA CCGT TA E TAG G ATTGCCA TG CA TGGCAAT FC G ^ t T GTA G G A AA C 3***TCGTAAT T T T A *A S GAT GC C G G C T S'* AT - A A T * *3 A T G C GT A T C G T A A T G C AT AT A T C G A T T A TA AT * T A AT * C t A C3'5 6A 3' 5' c 3t D C T 3'
FIG. 2. A model for genetic recombination. Homologous chromosomes A-C and B-D possess specific regions (perhaps placed every cistron or operon length along the chromosome) capable of forming Gierer-like structures in the presence of a specific recombination structural protein. Regions such as E-G and H-F form single-stranded, denatured loops outside the immediate environment of the protein (which senses only the symmetry-related nucleic acid structure shown) and are, therefore, susceptible to nuclease attack (shown by the arrows). When complementary loops G, H are nicked and opened, Watson-Crick base pairing occurs; this is followed by extensive propa- gation of the heteroduplex (shown in the lower two figures). The final structural intermediate is shown in the center of Fig. 3.
prove to have more general meaning with regard the recogni- that can be recognized by a tetrameric protein having identi- tion of naturally occurring operators by repressors. If a re- cal subunits related by 4-fold symmetry (222 symmetry is also pressor molecule has identical subunits related by two-fold possible; however, one must relax the requirement that the symmetry when it binds to DNA-the 2-fold axis coinciding nucleic acid structure possesses exact 222 symmetry. See with the dyad axis on DNA-then a necessary consequence refs. 13 and 14). is that the base sequence in the operator have 2-fold sym- Although no structural evidence is yet available concerning metry. This would be true regardless of precisely which the arrangement of subunits in the lac or the X repressors, groove (or grooves) the repressor binds. This general principle both proteins appear to bind DNA in tetrameric form (23, 24). for dimer recognition may be extended to include tetramer The precise symmetry relating subunits of these proteins will recognition in the following way (see Fig. 1C). One postulates eventually be revealed by x-ray crystallography; however, a tandem genetic duplication of the DNA sequence involved strong genetic evidence already points to the existence of two- in dimer recognition, followed by a hydrogen-bonding re- fold symmetry in the lac operator genetic map, with higher- arrangement (22). This generates a clover leaf-like structure order subdivision possible (25, 26). These findings, along with Downloaded by guest on September 28, 2021 Proc. Nat. Acad. Sci. USA 69 (1972) Molecular Mechanism for Genetic Recombination 2485 A 3. A B 5! 3' 5' G ~~~TA GH ~~AA 5- AGT TTGC A A T T A T C T A A T C G T A A T G C GC C G AT CG G C T A CG GC TA AT T A G C T A A T -- TAGTC G CC G C GGGTACCGTTA,::: B s. G CC G G C CG GC C G A T T A A T A T T A A T TA A T TA CG GC C G GC CG G C A T TA A T AT TA I +LIGASE -LIGASE E T AF GCAH cMMEEMEMMMEMMMM~gG' E G A F G C CG G C 'bb AT TA A TT TA A T A TA A T T A T A A T TA G C T A A T A G C TA A T 3,- **TAAC G T TA * - *5' G C C G * D C GG C A ATTG C AAT * *3' C GG C 3'-- *TAACGGTAC GTACCGTTA . AT T A 5 AT TGCCA A TCATGGCAAT ..* . .*5'3 C G T A A T '. * TGG GC C G C G I A GC CG G C A T G C CG GC A T G A T T A A C A T T A AT T A T TA A T AT A T TA t A 35. *' T A T s5 A T T A 3S 53 G 5 C 5' C D 1 G C T 3, HOLLIDAY (1964) BROKER & LEHMAN (1971)
FIG. 3. A model for genetic recombination (continued). The central structural intermediate can give rise to the Broker-Lehman hybrid DNA structure (28) in the absence of polynucleotide ligase, or to the Holliday hybrid DNA structure (5) in the presence of polynucleotide ligase. The Holliday structure possesses 2-fold symmetry. It can, therefore, be recognized by a nuclease(s) possessing two-fold symmetry that would be able to simultaneously nick strands of the same polarity at homologous sites (independent of base sequence). This could give rise to reciprocal recombination involving either single- or double-strand exchange, depending- on which strands were cut and joined. Another feature of this heteroduplex structure that results from its 2-fold symmetry is its ability to migrate with zipper-like action along parental DNA molecules without unwinding difficulties. This would allow recombination to occur randomly throughout the genome and could give rise to polarity effects observed in gene conversion.
the observation that lac repressor binds d(A-T) polymer selec- Details of the model are as follows. Symmetrically ar- tively with high affinity (27), directly support the Gierer-like ranged polynucleotide base sequences on homologous chro- operator structure shown in Fig. 1C. Definitive evidence, how- mosomes (spaced every operon length or so along the chromo- ever, must await the nucleotide-sequence data of the lac and X some) are first converted to their Gierer structures in the operators, and these studies are in progress (personal com- presence of a specific structural protein(s), which, for sim- munication from W. Gilbert and M. Ptashne). plicity, we shall call a recombination protein. Typically, a symmetrically arranged sequence contains central sequences The recombination model [E,G] and [H,F] that do not possess symmetry. Therefore, The concept that regions of DNA possessing symmetrically the Gierer structure contains loops of single-stranded DNA arranged polynucleotide base sequences can exist either as that are susceptible (either randomly, or with specificity) to Watson-Crick structures or Gierer-like structures (the latter attack by nuclease. When complementary loops are nicked, being induced by a specific structural protein or proteins), i.e., [G,H], homologous Gierer structures can then come to- leads to a simple but powerful model to explain genetic re- gether through base pairing; this is followed by propagation combination. This is shown schematically in Figs. 2 and 3. of the hybrid DNA (the details of which are shown in the The model involves the following features. (i) Chromo- lower figures in Fig. 2). One then arrives at the central inter- somes pair due to the formation of Gierer-like structures, which mediate shown in Fig. 3. In the absence of ligase, this inter- are induced by a specific recombination structural protein(s); mediate can become the Broker-Lehman structure (28). In (ii) a Holliday (5) hybrid DNA structure is formed, and this the presence of ligase, one can form the Holliday hybrid can migrate in either direction along the parental DNA DNA structure (5). This results from base pairing of sticky molecules; (iii) reciprocal recombination results from the ends [H,G], followed by sealing of nicks in the polynucleotide action of a nuclease(s) possessing 2-fold symmetry. chain by ligase. The Holliday structure is a particularly in- Downloaded by guest on September 28, 2021 2486 Genetics: Sobell Proc. Nat. Acad. Sci. USA 69 (1972)
+ + 16 17
$ + 17 2 + + 1~, + + 16 17 B 16 17 1 I> aim0 2 NN 4-19-1 -2 16 17 1-2 -!-. 19-1-2 A In I _ * EXONUCLEASE ACTIVITY 1fit -16 --- DNA POLYMERASE REPAIR 2-16 (-) + + 16 17 + \ 1-2-16 ( 1)I 2 + + 19-1-2 -16 4-19-1-2-16 -17 16-17 2-16-17 A 1-2-16-17 19-1-2 -16-17 4-19-1-2-16-17
4 19 1 2 16 17 4 19 2 16 17
arg4 locus arg, locus
FIG. 4. An explanation for polarity in gene conversion. A. Schematic figure showing Holliday hybrid DNA structure migrating from right to left encountering mutant allele 17 in the arg4 locus (9). triggering endonuclease attack. In addition, the nuclease has exo- nucleolytic activity and jumps on to either of the two nicked strands. Subsequent migration to the left causes the right half of the migra- tory duplex to fall apart, this fixing the starting point at which hybrid DNA begins in the recombinant molecules. Further migration, followed by nicking of homologous strands, results in reciprocal exchange, either of the single- or the double-strand type. The double- strand type is associated with exchange of flanking markers and is shown in this figure. Diploid cross is+ +26 17 B. Polarity in co- conversion is interpreted as arising from heteroduplex regions of variable size in one recombinant molecule with subsequent excision and repair of the complete heteroallelic area. Coconversion events of the type 17-16-2-1 are more frequent than 17-16-2, which, in turn, are than 17-16. This may reflect the diminishing probability of forming hybrid DNA regions with progressively smaller size. more frequent conversions Coconversionatc7naesfar morentsevalentevents of the type 16at-162-1, 162,21.andTh21 aree infreuerationinfrequent arentorrabsent in the diploidid1ncross +f2nd+ and single-site type 16^2, ++ 16 +gle-sitelconvsented.1 at 17 are far more prevalent than at 16, 2, or 1. These observations are readily interpretable in terms of the model presented.
teresting structure in that it possesses 2-fold symmetry. It conversion as documented by extensive studies of unselected can, therefore, be recognized by a nuclease(s) possessing 2-fold tetrads in yeast by Hurst, Fogel, and Mortimer (9). symmetry that would simultaneously act to nick strands of In Fig. 4A and B, we envision a fixed-length migratory hy- the same polarity at homologous sites (independent of base brid DNA structure (containing, perhaps, 50-100 nucleotide sequence). This would give rise to reciprocal recombination pairs) moving from right to left and encountering the first involving either single- or double-strand exchange, depending mutant allele 17 in the arg4 locus. This triggers endonuclease on which strands are cut and joined (5). Another feature of attack, nicking homologous strands (30). In addition, we the heteroduplex structure (which results from its 2-fold sym- postulate this nuclease to have exonucleolytic activity [analo- metry) is its ability to migrate with a zipper-like action in gous to the nuclease coded by the rec B and C cistrons in either direction along parental DNA molecules without un- Escherichia coli (31)], degrading either of the two nicked winding difficulties. This would allow genetic recombination strands. Subsequent migration of the hybrid DNA structure to occur throughout the genome and, in addition, may explain to the left causes the right half of the migratory duplex to fall polarity effects observed in gene conversion. apart, this fixing the starting point at which hybrid DNA be- gins in the final recombinant molecules. Further migration, Polarity in gene conversion followed by nicking of homologous strands, results in recipro- Implicit in this model for genetic recombination is the idea cal exchange, either of the single- or double-strand type (5). that synapsis between homologous chromosomes occurs be- Due to exonucleolytic action, only one of two recombinant tween genes and not within genes (29). This results in the molecules contain hybrid DNA. DNA polymerase activity formation of hybrid DNA, which can then migrate into the repairs the other duplex either before or immediately after the structural gene area. It will now be shown that such a migra- final recombination event, this giving rise to homoallelic and tory heteroduplex structure can give rise to hybrid DNA of heteroallelic duplexes. Polarity in coconversion is readily in- different classes in recombinant molecules, this leading to an ex- terpreted as arising from heteroduplex regions of variable size planation of polarity effects in single-site conversion and co- (containing, for example, alleles 17-16-2-1, 17-16-2, 17-16, or Downloaded by guest on September 28, 2021 Proc. Nat. Acad. Sci. USA 69 (1972) Molecular Mechanism for Genetic Recombination 2487
17) in one recombinant molecule with subsequent excision 6. Whitehouse, H. L. K. & Hastings, P. J. (1965) Genet. Res. 6, 27-92. and repair of the complete heteroallelic area. The model 7. Holliday, R. (1968) in Replication and Recombination of readily explains why coconversion events of the type, 16-2-1, Genetic Material, eds. Peacock, E. J. & Brock, R. D. (Aus- 16-2, and 2-1 are infrequent or absent in the diploid cross tralian Academy of Sciences, Canberra), pp. 157-174. 1 2 8. Fogel, S., Hurst, D. D. & Mortimer, R. K. (1971) Stadler + + and why single-site conversions at 17 are far more Symposia 1 and 2, 89-110. + + 16 17 9. Hurst, D. D., Fogel, S. & Mortimer, R. K. (1972) Proc. Nat. prevalent than at 16, 2, or 1. Implicit in the model is the pre- Acad. Sci. USA 69, 101-105. diction that the exchange of flanking markers should ac- 10. Lissouba, P., Mousseau, J., Rizet, G. & Rossignol, J. L. company gene conversion events in about 50% of the time, (1962) Advan. Genet. 11, 343-380. 767-782. this is in agreement with the yeast gene conversion data 11. Kitani, Y. & Olive, L. S. (1967) Genetics 57, and 12. Case, M. E. & Giles, N. H. (1964) Genetics 49, 529-540. (9). An extensive discussion of these and other points will 13. Sobell, H. M., Advan. Genet. 17, in press. follow (13). 14. Sobell, H. M., in Progress in Nucleic Acid Research and Molecular Biology, eds. Davidson, J. N. & Cohn, W. E. Synaptinemal complex (Academic Press, Inc., New York), 13, in press. electron microscopy have revealed the presence of 15. Sobell, H. M., Jain, S. C., Sakore, T. D. & Nordman, C. E. Studies by (1971) Nature 231, 200-205. a characteristic structure, the synaptinemal complex, which 16. Sobell, H. M., Jain, S. C., Sakore, T. D., Ponticello, G. & is seen in meiotic cells of a wide range of organisms at the Nordman, C. E. (1971) Cold Spring Harbor Symp. Quant. time of pachytene pairing (32-34). The complex lies between Biol. 36, 263-270. paired homologous chromosomes and is parallel to them, and 17. Jain, S. C. & Sobell, H. M., J. Mol. Biol., in press. H. M. & S. J. Mol. Biol., in press. a element 18. Sobell, Jain, C., consists of a ribbon-like axis that contains central 19. Bernardi, G. (1968) Advan. Enzymol. 31, 1-49. and two lateral elements. It has been suggested that this 20. Kelly, T. J. & Smith, H. 0. (1970) J. Mol. Biol. 51, 393-409. central element consists of fine lateral DNA loops that extend 21. Meselson, M. & Yuan, R. (1968) Nature 217, 1110-1114. between the two chromatids and undergo close homologous 22. Gierer, A. (1966) Nature 212, 1480-1481. pairing. We speculate that, among other things, this central 23. Gilbert, W. & Muller-Hill, B. (1967) Proc. Nat. Acad. Sci. USA 58, 2415-2421. element contains the Holliday hybrid DNA structure, the 24. Chadwick, P., Pirrotta, V., Steinberg, R., Hopkins, N. & central intermediate in intragenic, and, perhaps, more im- Ptashne, M. (1970) Cold Spring Harbor Symp. Quant. Biol. portant for higher organisms (35), intergenic recombination. 35, 283-294. 25. Smith, T. F. & Sadler, J. R. (1971) J. Mol. Biol. 59, 273-305. I thank Ernst Caspari, Allan Campbell, Christopher Lawrence, 26. Sadler, J. R. & Smith, T. F. (1971) J. Mol. Biol. 62, 139-169. Robin Holliday, Charles A. Thomas, Jr., Thierry Boon, and 27. Lin, S. Y. & Riggs, A. D. (1970) Nature 228, 1184-1186. Helen Eberle for stimulating discussions. This work has been 28. Broker, T. R. & Lehman, I. R. (1971) J. Mol. Biol. 60, 131- supported in part by grants from the National Institutes of 149. Health, the American Cancer Society, and the Atomic Energy 29. Whitehouse, H. L. K. (1966) Nature 211, 708-713. Commission. This paper has been assigned report no. UR-3490-178 30. Fincham, J. R. S. & Holliday, R. (1970) Mol. Gen. Genet. at the Atomic Energy Project, the University of Rochester. 109,309-322. 1. Emerson, S. (1969) in Genetic Organization, eds. Caspari, E. & 31. Goldmark, P. J. & Linn, S. (1972) J. Biol. Chem. 247, 1849- Ravin, A. W. (Academic Press, Inc., New York), Vol. 1, pp. 1860. 267-360. 32. Westergaard, M. & von Wettstein, D. (1966) C. R. Lab. 2. Fincham, J. R. S. & Day, P. R. (1971) Fungal Genetics Carlsberg 35, 261-286. (Blackwell Scientific Publications, Oxford and Edinburgh), 33. Lu, B. C. (1967) J. Cell Sci. 2, 529-536. 3rd Ed. 34. Moses, M. J. & Coleman, J. (1964) in The Role of Chromo- 3. Mitchell, M. B. (1955) Proc. Nat. Acad. Sci. USA 41, 215- somes in Development, ed. Locke, M. (Academic Press, Inc., 220. New York), pp. 11-49. 4. Case, M. E. & Giles, N. H. (1958) Cold Spring Harbor 35. Thomas, C. A., Jr. (1970) in The Neurosciences: Second Symp. Quant. Biol. 23, 119-135. Study Program, ed. Schmitt, F. 0. (The Rockefeller Uni- 5. Holliday, R. (1964) Genet. Res. 5, 282-304. versity Press, New York), pp. 973-998. Downloaded by guest on September 28, 2021