EcoRI Cleavage and Methylation of DNAs Containing Modified Pyrimidines in the Recognition Sequence*

The effects of substituents at position 5 in the pyrimidine ring of a variety of phage DNAs upon EcoRI endonuclease and methylase activities have been examined. The replacement of cytidine in DNA with glucosylated hydroxymethylcytidine confers resistance to cleavage by the EcoRI endonuclease. Substitution of thymidine in DNA by hydroxymethyluridine (a change in the methyl at position 5 of thymidine for a hydroxymethyl) lowers the maximal velocity of endonucleolytic cleavage Z&fold, but has no detecta-ble effect upon the K,,,. Substitution of thymidine in DNA by uridine (a change in the methyl at position 5 of thymidine for a hydrogen atom) has no effect upon either the maximal velocity or the K,,,. The effect of these modifications upon EcoRI methylase activity was markedly different. DNA containing glucosylated hydroxymethylcytidine is methylated as well as normal DNA. DNA containing uridine or hydroxymethyluridine, in place of thymidine, is much more poorly methylated than normal DNA. These different sensitivities of the EcoRI endonuclease and methylase to modifications

SB19 (ATCC 15575) was grown to an A,,,,,, of 1.0 in media containing, per liter, 10 g of Bacto-tryptone, 5 g of Difco yeast extract, and 10 g of NaCl supplemented with 0.14% glucose, 10 rnM MgSO,, and 100 pM MnCl,, and infected with PBS2 at a multiplicity of infection (m.o.i.1 of two. The culture was agitated at 37" for 3 additional h and then lysozyme was added to 20 pg/ml. After stirring for 90 min at 37" the culture was chilled and the phage purified as described below. Phage &-Phage be was from D. Roscoe via A. Price (18). B. subtilis 3610 (ATCC 6051) was grown at 37" in media containing, per liter, 10 g of Bacto-tryptone, 5 g of Difco yeast extract, 10 g of NaCl, and 0.9 g of glucose. Upon reaching an A,,,, of 0.3, the cells were infected with he at a multiplicity of infection of 0.1, and the culture agitated until it was clear, indicating complete lysis (approximately 5 h.1 (19). The culture was chilled, and the phage purified as described below. Phage &22 -Phage bm2 is a mutant of be that codes for thermosensitive dTTPase and dUTPase activities and hence contains DNA with hydroxymethyluridine and thymidine (20)  by the EcoRI endonuclease, each was incubated with sufficient enzyme to produce a limit digest, and the products were analyzed by electrophoresis through agarose slab gels (Fig. 2). Normal DNA (A, containing thymidine), and DNAs containing uridine (PBS&!), or hydroxymethyluridine (+e) were cleaved into fragments that migrated considerably faster than intact DNA. However, even with a large excess of EcoRI endonuclease, cleavage of DNA containing glucosylated hydroxymethylcytidine (T, DNA) was not detected. Although it is possible to determine the number of cleavage sites per genome for A DNA by simple examination of the limit digest (Fig. 21, it is not possible to do the same for the digests of PBS2, $e, or $m, DNAs. Digestion of each of these DNAs generates too many fragments of overlapping sizes to be distinguishable in agarose gels. An approximation of the number of cleavage sites can be made by determining the number average molecular weight of the DNA fragments and dividing this into the molecular weight of the DNA. Using the analysis described by Botchan et al. (321, a Kuhn distribution was fit to the densitometer tracing of the agarose gel-separated limit digest, and with these estimates of the average molecular weight, the numbers of cleavage sites for each DNA were calculated (Table I).
HmdCyd' DNA (T4) was detected, in agreement with the analysis by agarose gel electrophoresis described above. The lack of incorporation of 32P, into this DNA after incubation with EcoRI endonuclease rules out any significant level of host contamination.
E. coli DNA would be expected to be cleaved, and labeled.
DNAs containing hmdUrd in place of thymidine (4e and 4rns) were much poorer substrates than normal DNA (Fig. 3). Their initial rates of cleavage are one-twentieth that of DNA containing thymidine (A) or uridine (PBSB).
To obtain a better picture of the effect of modified nucleotides in the DNAs upon their susceptibility to digestion, each DNA was incubated with several different quantities of EcoRI endonuclease and the production of fragments was measured with the polynucleotide kinase exchange reaction (17). In Fig.  3, the initial velocity of digestion for each DNA is plotted as a function .of enzyme concentration.
No digestion of Glc-Although there might be a slight difference in the susceptibility of h DNA and PBS2 DNA, this result is uncertain because of a possible effect of substrate availability upon the initial rate of digestion. The concentration of DNA used in these studies was the same in each case (80 pg/ml). However, since PBS2 DNA contains approximately twice the number of EcoRI sites as h DNA on a weight basis (Table I) -To examine the dependence of the initial rate of EcoRI endonuclease digestion upon the concentration of cleavage sites, h DNA, $e DNA, 4rn, DNA, and PBS2 DNA were each incubated at different DNA concentrations with a constant amount of EcoRI endonuclease. The quantities of 5'-phosphoryl termini generated were measured using the polynucleotide kinase exchange assay. A sample time course digestion for each DNA is shown in Fig. 4. In preliminary experiments, we demonstrated that the exchange assay provides a reliable means to measure variable numbers of 5'-phosphoryl termini at varying DNA concentrations, since the extent of exchange of y-:"P with 5'-phosphoryl termini is independent of the DNA concentration over the concentration range used here (Fig. 5).
The initial rates of cleavage of 4e DNA and 4rn, DNA are plotted as a function of DNA concentration in Fig. 6u

Labeling of Modified DNAs by Polynucleotide
Kinase Exchange -In the previous experiments in which polynucleotide kinase exchange was employed to label the 5'-phosphoryl termini of EcoRI endonuclease-digested DNAs, an assumption was made that the extent of labeling was unaffected by the modifications at the pyrimidine ring within the DNA. Several studies using polynucleotide kinase phosphorylation support this assumption (34, 35), since fingerprinting of DNAs and RNAs labeled either by polynucleotide kinase phosphorylation in vitro or by in viva incorporation provide identical results. Furthermore, the actual nucleoside which accepts the phosphate from [@*]ATP is adenosine, rather than a modified nucleoside (36).
To provide further support for the notion that the differences in exchange described previously reflect the concentrations of 5'-phosphoryl termini, and not the reactivity of the substrate with polynucleotide kinase, experiments were performed to show that polynucleotide kinase labels PBSP, 4e, 4m,, and T, DNAs equally as well as X DNA. Equimolar DNA-phosphate mixtures of each DNA were mixed with h DNA and each was sheared by repeated passages through a syringe. The DNAs were then labeled with polynucleotide kinase, using the exchange reaction, and divided into two parts. One-half of each mixture was acid-precipitated, and the total amount of 32Pi exchanged into termini was measured (Table II). In each case, the polynucleotide kinase exchange reaction labeled the DNAs with modified bases equally as well as normal DNA. The other half of each sample was banded in CsCl and the extent of incorporation into each DNA measured after separation by its buoyant density. Again, all the modified DNAs were labeled as efficiently as the DNA containing thymidine (Fig. 7). Although the termini produced by shearing do not have the sequence of the EcoRI termini, these results argue that the polynucleotide kinase does not differentiate between DNA containing a pyrimidine other than thymine and normal DNA.
In summary, these and the previous experiments indicate that the EcoRI endonuclease differentiates among DNAs that contain modifications at position 5 of the thymine and cytosine rings. The substitution of the methyl group of thymidine by a  hydrogen atom does not dramatically affect either the K,,, or V max. Modification of cytosine to glucosylated hydroxymethylcytosine renders the DNA insusceptible to the endonuclease.

Methylation of Phage DNAs by EcoRI Methylase
-We examined the capacity of the methylase to act on DNAs containing modified pyrimidines, to determine whether the effect of each modification was the same upon methylase activity as in 100 mM TrisiHCl, pH 8.0, 10 rn~ EDTA, and 0.95 +vrS-adenosyl-I,llnethyl-:'Hlmethionine in a final volume of 150 ~1. Methvlase (from 0.04 to 8 pg) was added, and at various times 'aliquots "were withdrawn, acid precipitated and collected on GF/C filters as described under "Experimental Procedures." The T, DNA used in this experiment was purified by banding in CsCl, as described under "Experimental Procedures." Before this purification step. 2 ua of methvlase were required to achieve extensive methylation.'The-CsCl banding was required to remove an inhibitor of the EcoRI methvlase. The DNAs are (0-O) A, (X-X) T,, (A-A) be, (O=O) +m,, and (a--n) PBS2.
upon endonuclease activity.
To do this, the rates of methylation were measured for each DNA at a number of methylase concentrations.
The results are illustrated in Fig. 8. The striking result is that the susceptibility of the modified DNAs to methylation is not correlated with their susceptibility to cleavage. PBS2 DNA is not methylated, whereas T, DNA is methylated as well as h DNA.
The $e and 6rnr DNAs were methylated, but only at enzyme concentrations much higher than those needed to methylate normal DNA. We observed, as have Greene et al. (16) that the EcoRI methylase does not show a linear relationship between enzyme concentration and initial velocity. This was seen with h DNA and T, DNA. However, $e DNA and $m, DNA were both methylated at rates that are directly proportional to the amount of methylase added. It has been suggested that the methylase, which exists as a monomer in solution, needs to dimerize in order to modify its DNA substrate (12). With increasing enzyme concentrations, then, dimerization of the enzyme is enhanced. If this is so, the linearity of $e DNA and cbm, DNA can be explained, since at the very high concentrations of methylase needed to modify them, the enzyme is already maximally dimerized. Contrary to this observation with the methylase, the EcoRI endonuclease shows a linear response of initial velocity to all enzyme concentrations (seen here in Fig. 3 and observed by others for substrates like SV40 DNA (12)). This is consistent with the proposal that the EcoRI endonuclease acts catalytically as a dimer and is present as a dimer in solution (12).