S02 Nature Vol. 271 9 February 1978 Restriction endonucleases: a new role in vivo? from Richard J. Roberts

FEw enzymes have been exploited as experiments by Chang and Cohen (Proc. magnitude of the bacterial population on thoroughly as the bacterial restriction natn. A cad. Sci. U.S.A. 74, 4811 ; 1977) the face of this Earth would still mean enzymes. In recent years they have been show that an £. coli strain carrying the that the phenomenon could be of con­ instrumental in dramatic advances in EcoRl restriction enzyme can indeed siderable importance to bacterial evo­ DNA sequence analysis, genetic en­ mediate genetic engineering in vivo. In a lution. Such a possibility has been raised gineering, and studies of gene structure series of elegant experiments, they have previously (Reanney Bact. Rev. 40, 552; and function. Yet surprisingly little is demonstrated that a variety of site­ 1976). Clearly, it is necessary that ex­ known of their biological role. It has specific recombination events can be periments be carried out under con­ become almost axiomatic to associate catalysed in vivo. ditions resembling those in vivo for if them with a defence mechanism by which Their experiments centred around the indeed bacteria routinely exchange genetic bacteria insulate themselves against in­ fact that a chloramphenicol (Cm) resis­ information in this way, such exchange vasion by unwelcome foreign DNA. This tance gene contains an EcoRI site lo­ would not be limited to DNA originating view has its origins in the genetic pheno­ cated so that cleavage will inactivate the from prokaryotes. We would be forced menon of host-controlled restriction and gene. Restoration of gene activity re­ to the conclusion that species barriers modification. Two enzymes mediate this quires precise rejoining at the EcoRl exist, not because of a physical barrier process: an endodeoxyribonuclease (re­ site. It was first shown that such precise to the exchange of genetic information, striction enzyme) which degrades DNA, rejoining could be accomplished in vh·o but rather because a functional barrier and a methylase (modification enzyme) by transfecting E. coli with two EcoRI prevails. If such is the case, it would which protects DNA against the action fragments, each containing a part of the surely provide the strongest argument of the restriction enzyme. Because both Cm resistance gene. Chloramphenicol­ against the strict regulation of recom­ enzymes coexist in the same cell, the resistant transformants were recovered. binant DNA experiments in l'itro. cell's own DNA is protected by the action Then a plasmid, pSC352, was construc­ This question of frequency has other of the methylase, whereas incoming ted in which the same Cm gene was profound implications. If the frequency foreign DNA, which lacks the proper inactivated by the insertion of an EcoRI can be raised in the laboratory, then it modification, is susceptible to destruct­ fragment. Transfection of pSC352 into may be possible to construct recom­ ion by the restriction enzyme. an £. coli strain (C600, pMB4) which binant DNAs in this way. Shotgun Two different types of restriction contained the EcoRI restriction enzyme experiments might merely require trans­ enzymes have been found and can be resulted in a few transformants ( 6 x 10 -s formation of the appropriate strain distinguished by their mode of cleavage. of all transformants) which had acquired with intact DNA. In this way, both the Although both recognise a specific se­ Cm resistance. In vitro analysis showed biochemistry and the present regulations quence within a DNA molecule, the that such transformants arose from pre­ could be circumvented. Of course, ethical Type I enzymes cleave at random sites cise excision of the inserted EcoRI questions remain and, presumably, the remote from that sequence and give fragment. In more extensive experi­ next round of guidelines for recom­ heterogeneous products, while the Type ments, they showed that EcoRI pro­ binant DNA research will take account II enzymes cleave at specific sites within moted site-specific recombination could of these new observations. It will be of or close to the recognition sequence. involve multiple and physically separate some interest to see how they are phrased. It is these latter enzymes that have en­ fragments of plasmid DNA. In each Perhaps all experiments involving re­ joyed such popularity as the molecular case, the plasmids used for transformation combination, whether in vivo or in vitro, scalpels of the biochemist. Perhaps the were prepared so as to be unmodified will fall under their jurisdiction; this best-known example is EcoRI from against the action of the EcoRI restriction could include the whole of classical Escherichia coli which recognises the enzyme. What if the sites are already genetics. Even the most ardent bureau­ sequence, G~"AA TTC, and cleaves at modified? To answer this question, Chang crats will find it impossible to regulate the site indicated by the arrow. It pro­ and Cohen propagated pSC352 in £. recombination in the environment. duces specific fragments which carry coli C600 (pMB4), where it is modified by Plants, insects, animals, do it all the time. short, single-stranded extensions at each the EcoRI methylase, and then observed The line may have to be drawn in the end allowing them to reanneal and be­ the frequency of excision of the inserted laboratory. There sex will require a come a substrate for a DNA ligase. fragment under normal growth con­ Memorandum of Understanding and Thus, any two EcoRI fragments can be ditions. Chloramphenicol-resistant clones Agreement, while elsewhere it remains joined, which has made possible the developed with a frequency of about unrestrained. U powerful techniques of genetic engineer­ J0-9. In this case, therefore, even the ing. presence of the correct modification But paradoxically, the very enzymes enzyme within the cell was insufficient which are thought to prevent the ex­ to prevent site-specific recombination. change of DNA in vivo are precisely Although the observed frequency of those which facilitate that exchange in these events in vivo is extremely low, their vitro. Could it be .that the restriction importance should not be underesti­ enzymes have a dual role in vivo and mated. They raise anew the question of catalyse both degradation and synthesis the biological role of restriction enzymes, A hundred years ago just as they do in vitro ? This possibility and have implications for the current THE New York Tribune gives an has been raised previously but until now heated controversy about in vitro re­ account of a public exhibition in that it had not received serious attention. combinant DNA experiments. Because city of Eddison 's Phonograph, which seems to have been verv successful. It seemed unlikely that bacteria could these observations were made in the The tones reproduced by the vibrating be as smart as molecular biologists, laboratory, it is difficult to assess their disk of the machine were so distinct but apparently they are. For recent relevance to processes which might occur that thev could be heard and under­ in the natural environment. Nevertheless, stood in ditferent portions of the Richard J. Roberts is a Senior Staff crowded room. even if they occur at the low frequency Investigator at the Cold Spring Harbor From Nature 17, Feb. 7, 291; 1878. Laboratory. observed by Chang and Cohen, the

0028-0836/78[0271-0502$01.00 Cl Macmillan Journa Is Ltd 1978