An Endonuclease Activity of Venom Phosphodiesterase Specific for Single-stranded and Superhelical DNA*

A homogeneous preparation of venom phosphodiesterase from Crotalus adamanteus possesses an intrinsic endonuclease activity, specific for superhelical (form 1) and single-stranded DNA. The phosphodiesterase degrades single-stranded Ti DNA by endonucleolytic cleavages. Duplex T, DNA is hydrolyzed by the liberation of acid-soluble products simultaneously from the 3’ and 5’ termini but without demonstrable internal scissions in duplex regions. Since venom phosphodiesterase is known to hydrolyze oligonucleotides stepwise from the 3’ termini, the cleavage at the 5’ end of duplex T, DNA is ascribed to an endonuclease activity. Form I PM2 DNA is nicked to yield first relaxed circles and then linear DNA which is subsequently hydrolyzed only from the chain termini. The linear duplex DNA intermediates consist of a discrete series of fragments (11 are usually resolved on with initial molecular weights ranging (the to approximately x The cleavage closed circular

From the Laboratory of Enzymology, Roswell Park Memorial Institute, Buffalo, New York 14263 A homogeneous preparation of venom phosphodiesterase from Crotalus adamanteus possesses an intrinsic endonuclease activity, specific for superhelical (form 1) and singlestranded DNA. The phosphodiesterase degrades singlestranded Ti DNA by endonucleolytic cleavages. Duplex T, DNA is hydrolyzed by the liberation of acid-soluble products simultaneously from the 3' and 5' termini but without demonstrable internal scissions in duplex regions. Since venom phosphodiesterase is known to hydrolyze oligonucleotides stepwise from the 3' termini, the cleavage at the 5' end of duplex T, DNA is ascribed to an endonuclease activity. Form I PM2 DNA is nicked to yield first relaxed circles and then linear DNA which is subsequently hydrolyzed only from the chain termini. The linear duplex DNA intermediates consist of a discrete series of fragments (11 are usually resolved on agarose gels) with initial molecular weights ranging from 6.3 x 10" (the intact PM2 DNA size) to approximately 1 x 106. The cleavage of the form I molecule must, therefore, occur at a limited number of unique sites. The enzyme also cleaves nonsuperhelical, covalently closed circular PM2 DNA but at a 10" times slower rate. Both the endonuclease activity on form I DNA and the known exonuclease activity co-migrate on polyacrylamide gels, are optimally active at pH 9, are stimulated by small concentrations of Mg*+, and are similarly inactivated by heat, reducing agents; and EDTA.    analyzed for acid-soluble material and the molecular weight of the DNA (native and after alkali denaturation) was determined by agarose gel electrophoresis. Fig. 3 is a graph of the molecular weight of the alkali-denatured DNA uersus the per cent acid-soluble DNA. The experimental results (filled circles) are compared with the calculated change in molecular weight (solid line) for exclusive terminally directed hydrolysis of the 26 x 10" dalton duplex. Allowing for some uncertainty in the molecular weight values due to conformational and/or charge effects on single-stranded DNA during agarose gel electrophoresis (36), the agreement is good. In addition, for each experimental point, the recovery of the DNA on the gels was high and no lower molecular weight products were seen. Thus, nicks or double strand cleavages in internal regions G-1000 nucleotides from an end) of the duplex are rare or nonexistent under these conditions.
For reaction times at which the DNA was >50% acidsoluble, alkali treatment of the acid-precipitable DNA was required to obtain single-stranded molecules. Without prior denaturation, agarose gel electrophoresis of the products revealed duplex DNA of approximately 2 times the molecular weight of the single-stranded molecules (data not shown). These results are in agreement with mechanism B but not with mechanism A. According to mechanism A, if >50% of a duplex molecule is rendered acid-soluble, the acid-precipitable material is single-stranded (Fig. 2). Another experiment capable of distinguishing between the two mechanisms was performed. Duplex T7 [33P1DNA was digested with phosphodiesterase under the conditions described in Fig. 3 until 13% of the DNA was acid soluble. The reaction was stopped by lowering the pH to 5 and the solution was made 1 mM in L-serine and 0.1 mM in zinc acetate. Mung bean nuclease (371, a known single strand specific endonuclease (19, 20), was then added to a final concentration of 0.6 unit/ml and the per cent acid-soluble products was determined after incubation at 37" for 10 and 20 min. No increase in acidsoluble material was observed under conditions where, if the DNA were all single-stranded, mung bean nuclease would render it 100% acid-soluble. If phosphodiesterase were acting according to mechanism A, DNA which was 13% acid-soluble would be 26% acid-soluble after hydrolysis of the 5' singlestranded tails by mung bean nuclease. We conclude that phosphodiesterase digests linear duplex DNA simultaneously from the 3' and 5' termini ( Fig. 2, mechanism B).
The question mark in mechanism B, Fig. 2 ever, the experiment with mung bean nuclease shows that it is less than 1% of a T, DNA molecule or less than 200 nucleotides per strand. Starting with a substrate that is greater than 90% form I, the enzyme first produces form II and form III molecules. Within minutes, all of the form I DNA is converted to a series of discrete linear DNA fragments varying in size from approximately 6 x lo6 to 1 x lo6 daltons. A total of approximately 11 unique species is produced, with -50% of the linear DNA less than unit length. The molecular weight of each of these linear intermediates is subsequently reduced by -0.5 X 10" after 300 min of digestion. Analysis of gels (not shown) on which alkali-denatured (after phosphodiesterase digestion) samples were electrophoresed revealed a decrease in molecular weight of 0.4 x 10fi for the largest linear product in the same time period (Fig. 5). Thus, the linear substrate is degraded only from the termini as was the case, described above, with T, DNA.
In a separate experiment (enzyme concentration of 8 X 10e4 units/ml under standard reaction conditions), the relative amounts of forms I, II, and III PM2 DNA were monitored by agarose gel (1.4%) electrophoresis as a function of phosphodiesterase digestion. In the stages of the reaction covering 0 to -70% form III DNA, the data (not shown) indicate that the amount of form II DNA goes through a maximum (see also Fig. 4). There is a linear increase in the amount of form III DNA, but at a slower rate than form I DNA disappears. of form III DNA (see also the following paragraph). The initial rates of the I + II and II + III reactions were calculated from these data and are presented in Table II relative to the rate on form I' DNA (see below). For the purpose of calculation, it was assumed that each reaction involves only one cleavage per molecule, although -50% of each reaction proceeds by at least two cleavages per molecule to give ultimately linear DNA fragments of less than unit length. The cleavage rate on form II DNA was taken to be the rate of appearance of form III. The cleavage rate on form I DNA under standard reaction conditions at enzyme concentrations of 4 x 10m4 and 8 x low4 units/ml was also determined by the fluorescence assay (data not shown). This rate (Table  II) is in good agreement with that determined by the electrophoresis method.
Using agarose gel electrophoresis, the rate was also determined for the reaction of phosphodiesterase on form II DNA with -one single strand scission per molecule produced by the limited action of DNase I on form I DNA. This rate, also shown in Table II, is in agreement with that determined as described in the preceding paragraph by assuming that form Form I PM2 DNA was digested by phosphodiesterase as described in Fig. 4 except that the reaction lacked EDTA and NaCl. Aliquots were removed at selected intervals and the reaction terminated by the addition of EDTA. Samples containing 1.25 pg were electrophoresed on 0.7% agarose tube gels containing 0.5 kg/ml of ethidium bromide for 19 h at 22 V. Duplex DNA molecular weight markers were those described in Fig. 4. Gels were stained with "Stains-All" and scanned. This procedure revealed the same PM2 digestion products visualized in Fig. 4 plus a smaller fragment that was not seen by staining with ethidium. The molecular weight of each form III intermediate is shown as a function of incubation time and on the right side of the graph is the change between 5 and 300 min. In addition to the fragments shown, there is one that appears in the scans as a shoulder of the largest form III intermediate.
Since it was not well resolved, its molecular weight is not shown. The molecular weight for the largest linear species is inaccurate probably because for these gels the log of molecular weight is not a linear function of mobility above -5 x 106. This species is initially unit length since it co-migrates with form III PM2 DNA (gels not shown) produced from native PM2 DNA by the cleavage with nuclease Hpa II at a unique site.
II DNA was an intermediate in the formation of form III DNA.
Also shown in Table II is the cleavage rate on form I' DNA (nonsuperhelical, covalently closed circles) measured by the fluorescence assay under standard reaction conditions minus NaCl. Reaction rates determined at 0.4 and 0.98 unit/ml were in good agreement after normalization by enzyme concentration. The rate was 4 times less at 30" using form I' DNA that was prepared by the action of DNA ligase on form II DNA at the same temperature (data not shown). The fluorescence assays used in these experiments showed only a decrease in double-stranded DNA concentration indicating no accumulation of form II or III intermediates.
Agarose gels (not shown) also showed no intermediates.
These results are consistent with an initial endonucleolytic cleavage of form I' which converts the molecule to a more reactive form. The II += III and III -+ acid-soluble reactions are -lo3 and -lo5 times faster, respectively, than the rate of disappearance of I'.
All of these results are consistent with the mechanism depicted in Fig. 6. The endonuclease activity requires a form I or II substrate. Form III DNA, produced via a form II intermediate, is then hydrolyzed only from the molecular ends to the ultimate mononucleotide products. To explain the appearance of the numerous discrete linear fragments, it is necessary to postulate that the form I DNA sometimes experiences at least two simultaneous single strand cleavages which can occur at a number of unique sites. The same pattern of discrete fragments is observed over a range of variations in temperature, ionic strength, and enzyme concentration, but it is seen only when form I DNA is the initial substrate. In an experiment (not shown) with an initial substrate of form II DNA (produced by limited pancreatic  Both activities co-migrate with single protein band.

Effect of Mg2+
Both active with no added Mg2+ but both are stimulated by small amounts (<lo mM) and inhibited by larger amounts.

Effect of EDTA
Inactivates Inactivates the single protein band on polyacrylamide gels (Fig. 7), have similar pH dependence profiles, and are similarly inactivated by heat and reducing agents (Table III) Phosphodiesterase also possesses an endonuclease activity on form I' DNA, but the rate of hydrolysis is -lo6 times slower than the rate of release of acid-soluble products from singlestranded DNA. It is, therefore, reasonable to describe the endonuclease action as single strand specific (see also discussion below on super-helical DNA). It is possible that the low form I' DNA endonuclease activity is also intrinsic to phosphodiesterase. The single strand specific nuclease from mung bean is known to have intrinsic endonuclease activity on double-stranded linear DNA (20) and the ratio of single to double strand endonuclease rate is -lo4 under optimal conditions.
Although phosphodiesterase has a 3' to 5' polarity of action on single-stranded oligonucleotides (3, 41, we have shown that it hydrolyzes intact linear double-stranded DNA by liberation of acid-soluble products simultaneously from the 3' and 5' termini (mechanism B, Fig. 2). It can be inferred, therefore, that the action on the 5' termini is that of a single strand specific endonuclease, which recognizes a singlestranded 5' tail resulting from the release of mononucleotides from the 3' end. Some of the details of this mechanism are not yet clear. In particular, it would be useful to know the initial size of products released from the 5' end. Although the ultimate products of digestion by phosphodiesterase have always been reported to be 5'-mononucleotides, oligonucleotides, rather than mononucleotides, may be initially cleaved from the 5' ends after limited digestion of the 3' termini. This result is expected from a consideration of phosphodiesterase hydrolysis of single-stranded di-and trinucleotides bearing a terminal 3'-monophosphate.
Compared to digestion of 3'-hydroxyl-terminated substrates, these chains are very resistant and this resistance increases with decreasing chain length (4, 38). Similarly, the rate of attack on single-stranded cyclic oligonucleotides (up to 4 residues) is extremely low compared +o that on the linear analogue (6). Apparently, there is a minimum size required for endonuclease activity.
It can be inferred that the endonuclease activity of phosphodiesterase on PM2 DNA is due to single strand like regions in superhelical DNA, but the conclusion lacks ultimate proof. A number of chemical and enzymatic probes have detected such regions in form I DNA. These include formaldehyde (391, methyl mercury hydroxide (40), water-soluble carbodiimide (41, 421, the T4 gene 32 protein (43), and a number of single strand specific nucleases (44-52). The cleavage site of S, nuclease on form I SV40 DNA in high salt is in the same region as the gene 32 protein binding site and this region is most easily denatured at high pH (46). It is likely that the torsional stress produced by the superhelical turns results in the formation of kinks or bends in the double helix (27). The base pairing would be distorted at such sites, which are probably A + T rich regions of low thermodynamic stability. It is of interest to compare the phosphodiesterase activity on form I DNA with that of other single strand specific nucleases. The rate of cleavage on form I DNA is reported to be less than that on single-stranded DNA for phosphodiesterase (Table II) at the same rate. S, and the Pseudomonas nucleases also show endonuclease action on DNA with pre-existing nicks (49, 50).
The single strand specific nucleases do not have endonuclease activity on duplex linear DNA at a rate that is significant compared to rates on form I or single-stranded DNA. However, exonuclease-like action, which releases acid-soluble products from the ends of duplex linear DNA, has been reported for S, (561, Pseudomonas nuclease (501, and mung bean nuclease (20). It is possible that the terminally directed activity of these enzymes is due to the relative thermodynamic instability of the helix at the chain ends, but evidence for this conclusion has only been demonstrated for mung bean nuclease (20). Compared to other single strand specific endonucleases, phosphodiesterase seems to have a relatively high ratio of duplex DNA exonuclease activity to single-stranded or form I DNA endonuclease activity, but little quantitative data are available. Table II shows that phosphodiesterase cleaves -15 phosphodiester bonds at the termini of duplex T, DNA in the same amount of time that it takes to introduce a single nick in form I PM2 DNA. However, other experiments (not shown) indicate that this ratio of -15 to 1 is not constant but depends on reaction conditions. A ratio of -7 to 1 can be calculated for the action of Pseudomonas nuclease on linear and form I PM2 DNA from the data of Gray et al. (50).
Phosphodiesterase cleaves a large fraction of the form I DNA molecules at a minimum of two of a few specific sites resulting in the production of linear molecules of less than unit length. These multiple cleavages must be simultaneous since one nick removes the topological constraint required for superhelicity and only superhelical DNA substrates give the discrete, multiple fragment pattern.
Studies with S, (46,48,49) and N. crussu (52) nucleases have shown that superhelical SV40 and polyoma DNA are cleaved at one of a few specific sites, but 4X-174 is cleaved at one of many widely distributed sites. Pseudomonas (50) and mung bean (27) nucleases cleave PM2 form I at one of several possible sites, which for mung bean nuclease are not randomly distributed. Thus, the ability to cleave a superhelical DNA at two or more specific sites per molecule appears to distinguish phosphodiesterase from other single strand specific nucleases. However, it is not certain whether this is a unique property of phosphodiesterase or the PM2 DNA. Other single strand specific nucleases may also produce discrete multiple fragments with PM2 DNA under favorable reaction conditions. We have found (data not shown) that phosphodiesterase produces only unit length linear PM2 DNA when the Mg*+ concentration is lowered to 0.1 mM.
It is possible that the nuclease-sensitive regions in PM2 DNA correspond to the approximately eight specific A + T rich regions that are the early melting areas in form I DNA (571.
In order to account for the 11 discrete fragments we commonly 28. observe, there must be at least four separate sites of attack, but the exact number has not been determined. 29.