Complementation in vitro by mutant restriction enzymes from Escherichia coli K.

Abstract Two mutant strains of Escherichia coli K, K-18 lacking restriction activity (endonucleolytic cleavage of foreign DNA), and the other, K-19 lacking both restriction and modification activities (specific DNA methylation that protects against the homologous restriction activity), complement in vivo to yield a wild type phenotype. Although neither mutant extract alone binds unmodified DNA, the wild type extract and the mixture of mutant extracts do. Retention of the DNA-enzyme complex on membrane filters was used as an assay to purify the two mutant restriction enzymes. Both of these sediment like the wild type enzyme. Complementation in vitro by the two mutant enzymes could be demonstrated by specific DNA binding, cleavage of the unmodified DNA, and restriction-dependent ATP hydrolysis. All of these activities are absent in the individual mutant endonucleases but present in the wild type restriction endonuclease from E. coli K. However, both mutant enzymes show an activity that hydrolyzes ATP which is different from that of the wild type enzyme since it does not require unmodified DNA, but is dependent on the presence of S-adenosylmethionine.

From the Department of Microbiology, Biozentrum of the University of Basel, Basel, Switzerland SUMMARY Two mutant strains of Escherichia coli K, K-18 lacking restriction activity (endonucleolytic cleavage of foreign DNA), and the other, K-19 lacking both restriction and modification activities (specific DNA methylation that protects against the homologous restriction activity), complement in uiuo to yield a wild type phenotype. Although neither mutant extract alone binds unmodified DNA, the wild type extract and the mixture of mutant extracts do. Retention of the DNAenzyme complex on membrane filters was used as an assay to purify the two mutant restriction enzymes. Both of these sediment like the wild type enzyme. Complementation in vitro by the two mutant enzymes could be demonstrated by specific DNA binding, cleavage of the unmodified DNA, and restriction-dependent ATP hydrolysis. All of these activities are absent in the individual mutant endonucleases but present in the wild type restriction endonuclease from E. coli K. However, both mutant enzymes show an activity that hydrolyzes ATP which is different from that of the wild type enzyme since it does not require unmodified DNA, but is dependent on the presence of S-adenosylmethionine.

Restriction
endonucleases are enzymes that cleave doublestranded DNA synthesized in other bacterial strains. They arc involved in the phenomenon of host-controlled restriction and modification in which the acceptance or destruction of an entering DNA molecule is dependent on the restriction specificity of the host cell and on certain nonheritable characteristics acquired by the DNA in the cell in which it was replicated.
This host-controlled modification appears to be due to specific methylation at certain nucleotide sequences on the DNA which renders it resistant to attack by the corresponding restriction endonuclease from the same cell (1, 2).
The restriction enzymes from the related Escherichia coli strains K and B1 cleave foreign DNA in the presence of Ado-* This project was supported by Grant 3.716.72 from the Swiss National Foundation for Scientific Research.
The K restriction enzyme also possesses a specific methylase activity that transfers the methyl groups from Ado-Diet to unmodified X DNA in t,he absence of ATP and Mg2+. The methylated DNA is protected against degradation by the restriction activity of the same enzyme (7). In addition, K-and B-specific restriction endonucleases also have an ATPase activity associated with restriction (2,8,9).
Mutants of E. coli K and B lacking restriction activity can have two kinds of phenotype: defective only in restriction (r-m+) or defective in both restriction and modification (r-m-).
Onestep and two-&p r-m-mutants can be obtained. Genetic experiments with such mutants have led to a three-gene model for restriction and modification in E. coli. In its simplest formulation, one gene, hsdR, would be primarily responsible for restriction; another one, hsdS, for the site specificity of the recognition reaction; and a third one, hsd.11, for modification (10, 11). The results of Hubacek and Glover suggest that hsdkf is also required for restriction (12). An oligomeric protein composed of t,he hsdR, h&S', and hsdJl gene products would therefore have both restriction and modification activities, whereas one composed of the hsdS and hsdM gene products would only have modification activity. This model is in agreement with the subunit structure of the purified enzymes (1, 6, 13). The restriction endonuclease from E. coli K has a molecular weight of approximately 400,000 and is composed of subunits of three different sizes with approximate molecular weights of 135,000, 62,000, and 52,000. This enzyme can recognize specific nucleotide sequences and, depending on the cofactors and incubation conditions, can proceed to cleave or methylate the unmodified DNA (I).
The strains K-18 and K-19 have been shown to complement in viuo for restriction.* The present study was undertaken in order to determine which subunits are responsible for nucleotide sequence recognition, endonucleolytic activity, DNA methylation, and ATPase activity.
In this paper, the purification and properties of the defective restriction enzymes from the mutant strains K-18 with a rK-mK+ phenotype and a presumed hsdRKgenotype, and K-19 with a rK-mK-phenotype and presumed hsd&-genotype are described.
That K-19 has a hsd&-genotype is indicated by complementation experiments with a F' carrying rg+mg+.
We  Additions of enzyme extracts were as follows: 55 rg of K, 52.5/*g of K-18, and 43 rg of K-19. These were all 35y0 to 55% ammonium sulfate fractions.
After incubation for 2 min at 30", the reactions were terminated by the addition of 0.03 ml of 0.5 M EDTA, pH 8. The mixtures were then filtered through nitrocellulose filters. the enzyme is added to a reaction mixture containing differentially labeled modified and unmodified X DNAs, Ado-Met, ATP, and ,\I$+.
After a brief incubation it is passed through a nitrocellulose filter, and restrict,ion activity is detected as specific binding of unmodified X.0 DNA to the filter (17). These complementation results were similar to those reported previously by Linn and Arber (4) from infectivity experiments with extracts from rg-mg+ and rg-mg-strains.
In the experiment summarized in Table I, 100 c(g of calf thymus DNA were added to each reaction mixture to suppress the nonspecific binding of other proteins to the X DNAs.
The extract. from E. coli K was able to bind twice as much unmodified X.0 DNA as X.K DNA.
Although neither of the two mutant extracts specifically bound unmodified X DNA, when combined they bound 5 times more X.0 I)X\'h than X.K DNA. 111 both the wild type estract and the complementing mixture of K-18 and K-19 extracts, the specific binding was dependent on the presence of Ado-1lrt, and Al']'. However, these results should only be considered qualitative since the presence of calf thymus DNA also suppresses to a certain extent the specific binding of the restriction enzyme to the unmodified X DNA.
From our unpublished results there is evidence that calf thymus DNA may possess restriction sites for the E. co2i K restriction endonuclease. These results indicated that mixing together the two mutant extracts produced an activity similar in properties and requirements to that of the restriction endonuclease from E. coli K.
Purijcalion of Jlutant Restriction Enzymes-Hy using partially purified preparations of one mutant enzyme and the filter binding assay, the complementing restriction enzyme could be purified. The purified mutant enzyme could in its turn be used to purify the other one. Unless otherwise indicated, the activity of a mutant restriction endonuclease is operationally defined as its ability to specifically bind unmodified X DNA in the presence of an excess of the complementing enzyme. Table II gives a summary of the two purifications, the details appear under "Xethods." The endonuclease R.K-18 purifies in a manner identical with that of the wild type restriction enzyme and has been purified On the other hand, the endonuclease R.K-19 does not bind to phosphocellulose, and only weakly to hydrosylapatite, and in these respects is very different to the wild type and K-18 proteins.
This has resulted in its being purified only 400.fold and analysis on polyacrylamide gels shows the presence of a number of contaminating proteins. 130th mutant restriction endonucleases sediment similarly to the wild type enzyme in glycerol gradients and thus are not enzyme fragments.
The glycerol gradient fractions were stored at -40" and were stable for 6 months.
In all of the following experiments, the glycerol gradient fraction was used routinely unless otherwise indicated. DNA Binding-The ability of the purified enzymes to bind unmodified DNA was measured in the experiment shown in Table III.
The wild type enzyme specifically bound almost The reaction mixtures were the same as 7 rg of endonuclease R.K-19. The incubation was carried out at those described in Table I  After incubation at 37' for 18 hours X [32P]DNA (1950 cpm) and 25Obg of calf thymus DNA were added, the DNA was precipitated by addition of 1 ml ice-cold 10% trichloroacetic acid-O.01 M sodium pyrophosphate-1 M NaCl. The samples were left in ice for 30 min before centrifuging at 6000 X 9 for 15 min. The supernatants were discarded and the pellets were dissolved in 0.4 ml 0.5 M NHIOH with vigorous agitation.
The solution was extensively dialyzed against 0.01 M Tris (pH 8.0)-0.01 M NaCl-2 X 1OF M EDTA and reprecipitated as described above. Each precipitate was collected on a glass fiber filter and washed with 10 ml of cold trichloroacetic acid, sodium pyrophosphate, and NaCl followed by 10 ml methanol.
The filters were then counted in Aquasol.
cofactors and passed through a membrane filter, specific retention of almost all of the X.0 DNA was observed. The cofactor requirements for specific DNA binding by the complementing mixture of endonucleases R.K-18 and R.K-19 was identical with that of the wild type restriction enzyme (i.e. Ado-Met, ATP, and Mg2+).
As is the case with the wild type enzyme, a certain amount of nonspecific binding was observed if ATP was omitted.
Restriction Endonuclease Activity-The wild type restriction endonuclease readily cleaves unmodified X.0 DNA by making a limited number of double-stranded scissions in the presence of Ado-Met, ATP, and Mg2+. It has no effect on X.K DNA (3).
In the experiment shown on Fig. 1, the effect of the three en donucleases on the sedimentation rate of a mixture of X.K [aH]-DNA and X.0 [a*P]DNA was determined on neutral sucrose gradients. As was expected, the wild type endonuclease specifically degraded unmodified X.0 DNA while neither of the two mutant endonucleases had any effect on it. When the endonucleases R.K-I8 and R.K-19 were present together with all of the cofactors, the unmodified X.0 DNA was specifically cleaved. In the absence of either Ado-Met or ATP, no reaction was observed. Roth in its specific double-stranded cleavage of unmodified DNA as well as in its requirements for Ado-Met and ATP in this reaction, the combined mutant endonucleases resembled the wild type enzyme. all of the unmodified X.0 DNA in a complete system, but no such effect was observed with the endonucleases R.K-18 or R.K-19 by themselves. However, when the two mutant endonucleases were mixed together along with the DNAs and the

DNA Modification
Jfethylase Activity-It has been demonstrated that the restriction endonuclease from E. coli K has a DNA modification methylase activity which transfers an average of 8 methyl groups from Ado-Met to unmodified X.0 DNA.
Incubations carried out with larger amounts of enzyme can lead to a maximum incorporation of 10 to 15 methyl groups per X molecule. This reaction requires only Ado-Met, and the in vitro modified DNA is rendered resistant to cleavage by the same enzyme in the complete restriction system (7). The number of methyl groups incorporated in vitro agrees well with the figure The two mutant restriction enzymes were tested for DNA methylase activity, and compared with that of the wild type enzyme. Table IV shows that when modified X.K DNA was used as the substrate, neither the wild type endonuclease nor the mutant endonucleases nor the combined mutant endonucleases had any activity.
On the other hand, when unmodified X.0 DNA was used as the methyl acceptor, 6.6 methyl groups were incorporated per phage genome in the presence of endonuclease R.K. The enzyme from strain K-18 which possesses a functional modification system incorporated 5.1 methyl groups per DNA molecule, while the enzyme from strain K-19 which is unable to either restrict or modify showed a value of 0.7 methyl group per DNA molecule.
When both mutant enzymes were combined, the value remained essentially the same as for either the wild type or the endonuclease R.K-18 by itself. The results obtained fit the expectations arising from the genetic data, i.e. the wild type and r-m+ strains should modify to the same extent, but no complementation should be observed with the K-18 and K-19 strains since the limiting element would be the subunit responsible for site recognition.
ATP Ilydrolysis-An unusual feature of the K restriction endonuclease is an ATPase activity that is associated with it. This activity requires unmodified DNA and the same cofactors as the restriction reaction, cleaving the ATP to yield equimoiar amounts of ADP and inorganic phosphate.
This ATPase has two curious properties: it continues to hydrolyze ATP for periods of up to 2 hours, long after the nucleolytic reaction has gone to completion, and the number of ATP molecules split is greatly in excess of the number of double-stranded breaks (approximately lo5 ATP molecules per h genome with an average of 5 breaks per genome) (8).
In order to study the ATPase activity of the two mutant restriction endonucleases, it was necessary to purify the endonuclease R.K-19 by polyacrylamide gel electrophoresis in order to remove a contaminating ATPase unrelated to restriction. Experiment 1 of Table V shows the ATPase activities of the mutant enzymes alone. R.K-18 shows an activity which is dependent on Ado-Met only. Presence of unmodified DNA does not make any difference.
R.K-19 still contains a low residual contaminating ATPase activity even after further purification on polyacrylamide gels. In the complete reaction system more than a a-fold increase in this activity is observed.
That this activity is Ado-Met-dependent is further suggested by theobservation that when DN.4 is removed from the reaction mixture a further increase in the activity is obtained.
Experiment 2 of Table V shows the ATPase activity of the combined mutant endonucleases.
Equivalent DNA binding units of each enzyme were used in this experiment.
In the reactions with the combined endonucleases R.K-18 and R.K-19, the same amount of each enzyme was added. This results in an enzyme concentration double that of the incubations with each enzyme alone. As was shown in the previous experiment each of the mutant enzymes had an ATPase activity which was about one-fourth of that of the wild type enzyme and did not require DNA and is dependent on the presence of Ado-Met. This makes it highly unlikely that the activities in question are due simply to contamination.
When the combined mutant enzymes w-ere assayed for ATPase, the value obtained was dependent on the presence of unmodified DNA and Ado-Met as is the case for the wild type activity.
Binding of Ado-Met to Restriction Enzyme-No clear-cut role has been found for Ado-Met in the restriction reaction and it is not known whether it is consumed in the reaction or acts as an allosteric effector. We have looked at the fate of Ado-Met when it was incubated with the various restriction endonucleases. Fig.  2 shows the results of experiments in which the wild type endonuclease or one of the mutant enzymes was incubated for 2 min. with [methyLaH]Ado-Met and was then put through a Sephadex G-50 column.
Given the high molecular weight of the enzymes, they should emerge in the void volume while the Ado-Met should be included.
Aliquots of the fractions from each column were counted for radioactivity, and the samples in the void volume were assayed for DNA binding in the presence or absence of additional Ado-Met. Fig. 2A shows the effect of incubation with heat-inactivated and native endonuclease R.K.
Only a small amount of radioactivity emerged in the void volume in the case of the inactivated enzyme and no enzyme activity was observed. If native restriction enzyme was used, a tritium peak was observed in the void volume which coincided with the enzyme activity as measured in the presence of all cofactors, Ado-Met included.
How- The column was run with Buffer C at a flow rate of 12 Ci per mmole). The amount of eniyme added to each sample was ml per hour at 4" and fractions of 0.25 ml were collected.
Aliquots 37 pg of endonuclease R.K, 15 pg of endonuclease R.K-18, and 45 of 0.05 ml from the first 20 fractions were assayed for DNA binding pg of endonuclease R.K-19. The samples were incubated at 30" in the presence and absence of Ado-Met.
Samples of 0.05 ml were for 2 min and 0.01 ml 0.5 M EDTA (pH 8.0) was then added. Each also counted for tritium in 3 ml of Aquasol. ever, if Ado-Met was omitted in the enzyme assays, 50% of the activity could still be detected indicating that the Ado-Met bound to the enzyme appears to be sufficient to allow specific binding to unmodified DNA without further addition of Ado-Met. The enzyme complex carrying the tritium label was concentrated, sodium dodecyl sulfate was added to a final concentration of 0.4%, and the sample was analyzed by thin layer chromatography.
All of the tritium was present as Ado-Met indicating that the methyl group of the Ado-Met was neither covalent,ly bound to the enzyme nor had undergone any chemical change.
Similar experiments were done with the mutant enzymes.
The results are shown in Fig. 2B. As mentioned previously, the activity of the mutant restriction enzymes is measured in terms of its ability to bind specifically unmodified DNA in the presence of the complementing enzyme. With the endonuclease R.K-18, a tritium peak emerged in the void volume along with the enzyme activity (measured with the complete assay system). The percentage of DNA binding was higher in the case of endonuclease R.K-18 because a lower concentration of unmodified DNA was present in the assays. If Ado-Met was omitted from This indicated that the mutant enzyme from strain K-18 was able to bind Ado-Met and use it in conjunction with the complementing endonuclease R.K-19 to bind unmodified DNA.
In the case of endonuclease R.K-19, a tritium peak appeared in the void volume along with the enzyme activity.
If the Ado-Met was omitted in the assays, no enzyme activity was detected. It must be stressed however, that the Ado-Met binding observed may very well be due to one of the contaminating proteins and not to the mutant enzyme itself. What is clear is that even if the mutant enzyme from K-19 does bind Ado-Met, it is unable to use it in combination with the endonuclease R.K-18. DISCUSSION The wild type and mutant restriction enzymes of E. coli K have been compared by studying a number of different reactions: the specific binding and endonucleolytic cleavage of unmodified X DNA, DNA methylation, and the hydrolysis of ATP during the restriction reaction.
The restriction endonuclease R.K-18 (from the rK-mKf strain) purifies in a manner identical with that of the wild type enzyme. On the other hand, the enzyme from the rK-mK-strain K-19 which presumably lacks the site recognition function is quite different insofar as it is unable to bind to phosphocellulose and is weakly bound to hydroxylapatite.
All three enzymes have the same size of approximately 12 S. Neither of the two mutant endonucleases bound to unmodified X.0 DNA. This is unexpected since the endonuclease R.K-18 should have an intact recognition subunit.
These results would indicate that both the restriction and recognition subunits are necessary for formation of the specific complex that is bound to the nitrocellulose filters. At the same time, it suggests that the specific binding seen with the wild type endonuclease is associated with the cleavage reaction rather than with just recognition of the host specificity sites on the DNA.
The endonuclease and methylase activities of the two mutant enzymes are in accordance with the in tivo observations; neither of the two can cleave unmodified DNA, and the endonuclease R.K-18 but not the endonuclease R.K-19 can methylate unmodified DNA.
The hydrolysis of ATP is an unexplained property of the restriction endonuclease but it is known that it requires the endonucleolytic reaction to take place. Two hypotheses have been advanced to explain this ATPase: (a) ATP hydrolysis may provide the activation energy for some conformational change required for the specific recognition of unmodified DNA; or (b) the cleavage of unmodified DNA yields an altered enzyme that catalyzes the ATI' hydrolysis.
It is interesting to note that the ATPase characteristic of the wild type enzyme is absent from both mutant enzymes, neither of which can restrict. These enzymes however do have an ATPase activity which is dependent solely on the presence of Ado-Met.
Complementation in vitro can be readily demonstrated with the mutant enzymes from strains I<-18 and K-19. When both proteins are present together, there is specific binding to unmodified DNA, double-stranded scission of the DNA, and ATP breakdown-all of these showing the same requirements as the wild type enzyme. None of these results allow us to distinguish between complementation in vitro due to subunit exchange between the two mutant proteins and a reaction consisting of at least two steps each one catalyzed by one of the mutant enzymes. Several experiments have been done in an effort to isolate wild type enzyme by chromatography on phosphocellulose columns of a mixture of endonucleases R.K-18 and R.K-19 that had been incubated uith or without cofactors under a variety of conditions. No wild type activity was observed. Attempts to isolate a DNA intermediate after incubation with one of the mutant enzymes and then incubating it with the complementing enzyme have yielded ambiguous results. Further experiments are in progress with these enzymes and those from other mutants to elucidate this matter.