For the "Inactivating Effect." Materials and Methods.-Transformation: DNA Donor Strain: 60 009 (=SB 19 of Romig) = Prototroph

VOL. 52, 1964 GENETICS: FREESE AND FREESE 1289

8 Matthaei, J. H., and M. W. Nirenberg, these PROCEEDINGS, 47, 1580 (1961). 9 Wood, W. B., and P. Berg, these PROCEEDINGS, 48, 94 (1962). 10 De Moss, J. A., and G. D. Novelli, Biochim. Biophys. Acta, 18, 592 (1955). 11 Nathans, D., and F. Lipmann, these PROCEEDINGS, 47, 497 (1961). 12Raacke, I. D., S. Matsushita, and J. Fiala, unpublished. 13 Webster, G., and S. L. Whitman, Biochim. Biophys. Acta, 68, 653 (1963). 14 Levinthal, C., A. Keynon, and A. Higa, these PROCEEDINGS, 48, 1631 (1962). 15 Revel, M., and H. H. Hiatt, these PROCEEDINGS, 51, 810 (1964). 16 Weisberger, A. S., S. Armentrout, and S. Wolfe, these PROCEEDINGS, 50, 86 (1963). 17 Hardesty, B., J. J. Hutton, R. Arlinghaus, and R. Schweet, these PROCEEDINGS, 50, 1078 (1963). 18 Watson, J. D., Science, 140, 17 (1963). 1" Allen, D. W., and P. C. Zamecnik, Biochim. Biophys. Acta, 55, 865 (1962). 20 Cammarano, P., G. Guidice, and G. D. Novelli, Biochem. Biophys. Res. Commun., 12, 498 (1963). 2"Lederberg, S., B. Rotman, and V. Lederberg, Biochem. Biophys. Res. Commun., 12, 324 (1963).

TWO SEPARABLE EFFECTS OF HYDROXYLAMINE ON TRANSFORMING DNA ]3 ELISABETH BAUTZ FREESE AND ERNST FREESE LABORATORY OF MOLECULAR BIOLOGY, NATIONAL INSTITUTE OF NEUROLOGICAL DISEASES AND BLINDNESS, NATIONAL INSTITUTES OF HEALTH Communicated by M. R. Irwin, October 1, 1964 In earlier experiments we observed a dual effect of hydroxylamine (HA) on phage T4. At high concentrations of HA (about 1 M) phages were readily mutated under conditions of little inactivation, whereas at lower concentrations (10-2 M) phages were rapidly inactivated without any induction of mutations.1 The mutagenic effect was shown to result from the specific reaction of HA with cytosine.' The highly inactivating effect, however, remained partially unexplained. It could be exclusively caused by the inactivation of phage tail protein, which has been shown to occur,3 or it could be due to a combination of two reactions, one on the phage tail and the other on phage DNA. In order to find out whether DNA was actually subject to a strong inactivation by low HA concentrations, we used the transformation system of B. subtilis, which allows one to measure both inactivation and mutation directly on the same DNA. The present paper will show that HA indeed exerts two different effects on DNA, a predominantly mutagenic effect at high HA concentrations (about 1 M) and a predominantly inactivating effect at low HA concentrations (102 to 104 M). From experiments with HA analogues it was furthermore possible to decide which group of HA is responsible for the "mutagenic effect" and which one is required for the "inactivating effect." Materials and Methods.-Transformation: DNA donor strain: 60 009 (=SB 19 of Romig) = prototroph. Recipient strain: 60 087 (= T, of Anagnostopoulos) = tryptophan . Transforma- tion to tryptophan independence and production of fluorescent mutants were assayed as described previously.4,' Each time-dependent curve was determined with the same batch of transformable bacteria. Downloaded by guest on October 1, 2021 1290 GENETICS: FREESE AND FREESE PROC. N. A. S.

Chemicals: HA X HCl was bought from Eastman; 0-methyl-HA X HCl (= methoxyamine) and N-methyl-HA X HCl were obtained from the Aldrich Chemical Co., Milwaukee, Wis. The purity of the latter was checked by paper chromatography in 1:1 n-butanol: 6 M HCl on Whatman #1 paper. The paper was sprayed by a 0.5% solution of picryl chloride in ethanol and then exposed to ammonia vapor. The Rf values are 0.50 for HA and 0.69 for N-methyl-HA. By over- loading the paper it could be established that the impurity of HA in N-methyl-HA was less than 1%. pK values of HA derivatives were measured at a concentration of 10 mg/ml in a radiometer recording pH meter. Sodium hyponitrite was prepared by Alfa Inorganics, Inc., Beverly, Mass., and showed a pre- cise three-electron reduction with potassium permanganate. The UV spectra of dilute solutions agreed with results described in the literature.6 The absorption maxima were observed for 2450 A at pH 12, 2280 A at pH 9.9, and 2080 A at pH 3.5. The solution was stable at 250C (pH 3.5) but not after prolonged treatment at 750C. Treatment of DNA: DNA (130 ,jg/ml) was diluted tenfold in the ice-cold reaction mixture, mixed, and a control sample diluted 50-fold into ice-cold stopping mixture (0.05 M Tris pH 7.5 + 1 M NaCl + 10% acetone). The reaction mixture contained, in addition to the HA compound, 0.02 M Na-phosphate (at pH > 6) or Na-phosphate + succinate (at pH < 6), and NaCI to give the desired Na+ concentration wherever that has been especially noted. The reaction tube was then placed (about 30 sec after mixing) in a water bath at the desired temperature, usually 750C. At different times samples were diluted 50-fold into the stopping mixture. The DNA was then further diluted (1:10) in the transformation experiment. Results.-Induction of mutations: We discuss the mutagenic effect of HA before tackling inactivation, because the mutation results can be simply explained in terms of a single reaction mechanism. The induction of mutations was measured by the method of linked mutations described earlier.4 The recipient strain of B. subtilis was a mutant lacking the tryptophan synthetase activity; it could grow on tryptophan but not on indole. The prototroph donor DNA was treated by HA under various conditions and added at suboptimal (but constant) concentration to bacteria in the transformable

t06 o 2M)

424H

4OUS m ota

O 2 3 5 6 7 o os S FIG. 1.-Induction of fluores- FIG. 2.-Rate of mutation induction at cent mutants at different HA different concentrations of HA and its concentrations. pH 6.2, 750C. derivatives. Temperature 75°. n = HA, * = 2 MHA; as = 1 MHA- pH4.2- = HA, 3M Na, pH6.2- A- o = 1 M HA + 3 M Na+; HA, pA 6.2; *-N-methyl-HA, pH 6.2; * = 0.5 M HA; o = 0.5 M * = carboxymethoxyamine, pH 6.2; 0 = HA + 3 M Na+; A 0.1 M HA, 0-methyl-HA, pH 6.2, Downloaded by guest on October 1, 2021 VOL. 52, 1964 GENETICS: FREESE AND FREESE 1291

state. When these bacteria were plated on H suboptimal concentrations of indole, the fre- N-OH hydroxylanine quency of fluorescent mutants among trypto- H phan-independent transformants could be di- H rectly counted on the plates. \-OH N-thyl-hydroxylanino During the treatment of DNA with HA, the / frequency of fluorescent mutants increased 3 linearly with time, as shown in Figure 1. H When the rate of mutation induction (% mu- N-O-CH3 0-methyl-hydroxylmine= tants induced/hr) was plotted against the con- / mothoxyamine centration of HA, a linear curve was obtained FIG. 3.-Structure of HA and some (Fig. 2). These two results indicate that each derivatives. mutation was established by a single chemical event. Figure 2 also shows that the rate of mutation induction increased with decreasing pH,4 but did not depend on the concentration of sodium ions. At HA concentrations lower than 0.1 M no induction of mutations could be detected. For example, the mutation rate was less than 0.1 per cent per hour at 10-5 M HA and pH 6.2. Both N-methyl- and O-methyl-HA also induced mutations whose frequency increased linearly with time and analogue concentration (Fig. 2). The structures of these HA derivatives are shown in Figure 3. Inactivation: The inactivation of transforming DNA was measured by the decrease in the number of tryptophan-independent transformants with the time of DNA treatment by HA or its analogues. The logarithm of the surviving fraction (B/Bo, with B = titer of transformants at time t and Bo = titer at time 0) decreased linearly with the time of treatment, for HA concentrations up to 1 M, and for different temperatures, pH, or ionic strength. An example is shown in Figure 4. Only at 2 M HA did the rate of inactivation increase slightly with the time of treatment. We define here the inactivation rate k by

k = d In Bo (t = time of treatment in hours).

At 1 M HA the inactivation rates rapidly increased with the temperature, according to the Arrhenius equation, as can be seen from Figure 5. In order to obtain suffi- cient inactivation within a few hours of treatment, all of the following experiments were performed at 750C. Although the time dependence of inactivation seemed to indicate that a single chemical reaction caused the inactivation at each HA concentration (except at 2 M HA), the concentration dependence of k revealed the existence of two inactivating effects (see Fig. 6). Instead of a linear decrease of the inactivation rate with decreasing HA concentration (which would give a 450 slope in Fig. 6), the value of k actually increased at pH 6.2, when the HA concentration decreased from 1 M. This was especially pronounced when the reaction mixture contained 3 M Na+ ions (see Fig. 6). At very high HA concentrations the curve bent upward again, obviously because the mutagenic effect, which increased with the HA concentration, exhibited a significant lethal effect. (It should be obvious that mutagenic reactions can be lethal if they occur in genes vital for the growth of the organism; the transforming pieces carrying the tryptophan markers apparently contain such vital genes.) At Downloaded by guest on October 1, 2021 1292 GENETICS: FREESE AND FREESE PROC. N. A. S.

I0~~~~~~~~'~~

3l-) I 2 0 i

HOURS FIG. 4.- Inactivation of transforming DNA at different HA concentrations. pH 6.2, 75°. U = 1 M; /\ = O.5M; A = 10-2M; O = 1O-3M; * = 10-4 A; 0 = 2 XK10-4MOf; * = 10-5M; Ho = 10-6 M; O = 1O-7 M.

pH 4.2 the mutagenic effect was much stronger, as shown before, so that no dip in the inactivation rates was observed at high concentrations (Fig. 6). Most surprising was the extended concentration range for which the inactivation rates remained essentially constant until they eventually declined. At low HA concentrations (10-2 M) the inactivation rates did not change much with the pH in the range between 4 and 7, but they increased significantly for higher pH values (Fig. 6). In contrast, at high HA concentrations (1 M) the inactivation rates increased significantly when the pH was decreased from 7.5 to 4.2, indicating that this inactivation was caused by the mutagenic effect described in the previous as-65-ss-45* Art- section. The inactivating effect of low HA concentrations10- was not, or very slightly, mutagenic. It is obvious that the predominantly inactivating and the predominantly mutagenic effects were caused by two different chemical reactions. Inactivation by HA derivatives: The existence of two ,00\. ~~~different reactions of HA with transforming DNA could \ ~~~~bedemonstrated most clearly by the use of HA deriva- X \ ~~~~tives. Whereas N-methyl-HA showed essentially the 0 same concentration dependence of inactivation as did z i\ ~ ~HA, the inactivating effect of 0-methyl-HA decreased .0,0 \ rapidly with the concentration (Fig. 6). Since the OH z \ ~~~~~group of this HA derivative is esterified, the predomi- OOS~~~ X ~nantly inactivating effect of HA or its derivatives apparently requires the presence of a free OH group. In contrast, the predominantly mutagenic reaction seems °°|X930' 3 '32'33to involve the nitrogen of HA or its derivatives, for l>TXIO}both N- and 0-methyl-HA induced mutations (Fig. 2). s ~~~Inagreement with this conclusion is the finding that FIG. 5.-ITemperature do another esterified HA derivative, carboxymethoxy- pendence of inactivation rates forn1eMnHA,tpHn6.2. amine, also inactivated DNA at a slow rate (Fig. 6) Downloaded by guest on October 1, 2021 VOL. 52, 1964 GENETICS: FREESE AND FREESE 1293

1001III0PI9I 1

A 10~~~~~~~~~~~~~~~~

N oIi HA .pH62 , *~~~~~~~~~~~~~~~~~~pH75

0.01 lo-7 10-6 lo-5 1o04 1o03 1c-2 lo-o 0.5 1 2 5 10 MOLAR CONCENTRATION FIG. 6.-Concentration dependence of inactivation rates at 75°. A = HA, pH 4.2; 0 = HA, pH 6.2; * = HA, pH 7.5; o = HA, pH 9.0; A = HA, pH 6.2 + 3 M Na+; 0 = N-methyl-HA, pH 6.2; I = 0-methyl-HA, pH 6.2; * = carboxymethoxyamine, pH 6.2. but showed a significant mutagenic effect (Fig. 2). The observed increase of the inactivation rate with the concentration of 0-methyl-HA is probably mainly caused by the mutagenic effect of this compound. However, 0-methyl-HA and other esters of HA are apparently slowly hydrolyzed, especially at high temperatures and at pH values below their pK values, giving rise to HA itself. This is indicated by the finding that 0-methyl-HA formed many gas bubbles at pH 4.2, which is below its pK value of 4.6. Although this effect was not observed at pH 6.2, it should occur to a small extent and may be responsible for the fact that the inactivation rates decreased more slowly than the concentration of 0-methyl-HA (Fig. 6). The effect of sodium hyponitrite; HA can react with itself to form nitroxyl (:NOH or HNO) by the reaction 2NH20H = NOH + NH3 + H20. Nitroxyl can subsequently react either with another HA to form N2 + H20 or with another nitroxyl to form hyponitrous acid, HONNOH. Thus either nitroxyl or hyponitrous acid might be responsible for the lethal effect at low HA concentrations. Since hyponitrous acid has been proposed as a possible agent responsible for chromosome breakage,7 we tested the trans form of hyponitrous acid for its effect of DNA. (This compound had all properties described for it in the literature, as described above.) The results depicted in Figure 7 show that no inactivation of transforming DNA was observed with sodium hyponitrite, irrespective of the concentration and the temperature used. Furthermore, no mutagenic effect was observed under any of these conditions. Concentration dependence of the HA reaction with cytosine: HA reacts with cytosine, its nucleosides, and nucleotides by converting the C-C double bond into a single bond, and thus eliminates the UY absorption of cytosine at its maximum.2 In order to determine whether this reaction has an unexpected concentration dependence, we measured the initial rate r d ODo dt OD of the OD decrease of deoxycytidine monophosphate (dCMP) at different con- Downloaded by guest on October 1, 2021 1294 GENETICS: FREESE AND FREESE PROC. N. A. S.

l0*~~~~~~~~~~M~0 01.101.lo,0o. / ,

oOUIS 500C OCOF HYNNOMOSIINE FIG. 7.-Effect of different concentra- FIG. 8.-Concentration tions of hyponitrous acid on transform- dependence of the reac- ing DNA, at pH 6.2 and different term- tion rate of dCMP with peratures. At 750C: a = 10-1, A = HA, measured by the 10-2, 0 = 10-3, 0 = 10-4 M; at 250C: decrease of ODo at pH A = 10-1, A = 10-2, = 10-3M. 6.2,750.

centrations of HA. dCMP was added (at a final OD280 = 0.7) to a reaction mixture containing, in addition to HA, 0.02 M phosphate, and 1 M NaCl, at pH 6.2 and 750C. The OD decrease at 280 mjs was recorded. As Figure 7 shows, r increased nearly linearly with the HA concentration. (We do not understand the deviation from linearity, which is not caused by a salt effect, but this deviation is unimportant in the present context.) Di8cussion.-HA exerts two biological effects on transforming DNA, a predominantly mutagenic reaction, observed only at HA concentrations >0.1 M, and a predominantly (or exclusively) inactivating reaction, observed at HA concentrations down to as low as 10- M. By the use of HA derivatives it has been established that the free NOH group of HA or its derivatives is required for the inactivating reaction. In contrast, the mutagenic reaction occurs apparently with any HA compound; it is caused by the reaction of the HA nitrogen with the reactive base (cytosine). Different biological effects can therefore be related to different chemical reactions. The mutagenic effect is apparently caused by the reaction of HA with cytosine giving rise to base pair transitions from GC -- AT;2 the same appears likely for the mutagenic effect of the HA derivatives. There are two sites on cytosine to which HA can attach: the C-C double bond, which can be converted into a single bond, and the amino group, which can be replaced by a hydroxylamine group. The reaction with the C-C double bond can be easily observed by the decrease in the extinction coefficient of cytosine. It increases approximately linearly with the concentration of HA as does the mutagenic reaction. But the chemical and mutagenic rates differ greatly: when deoxycytidine monophosphate is exposed to 1 M HA at pH 6.2 and 750C, the C-C double bonds of cytosine are converted into single bonds at the initial rate of 90 per cent per hour (see Fig. 7). In contrast, transforming DNA is mutated under the same conditions at the rate of 1.1 per cent fluorescent mutants per hour. Earlier experiments have shown2 that mutants induced by HA carry point mutations, as if they were caused by the alteration of a single cytosine base in DNA. The number of possible mutagenic sites in the fluorescent region may be of the order of 103 cytosine residues. The reaction rate of HA with the C-C double bond of dCMP would therefore be of the order of 101 times larger than the rate at which an alteration of cytosine in DNA leads to a mutation. The difference in rates could be attributed to three factors: (1) Only a Downloaded by guest on October 1, 2021 VOL. 52, 1964 GENETICS: FREESE AND FREESE 1295

fraction of the chemically altered cytosines gives rise to mutations. (2) Double- stranded transforming DNA reacts with HA much less efficiently than denatured DNA; at 750 this difference amounts to a factor of about 100.5 (3) Even in single- stranded DNA the pK of cytosine may be significantly lower than that of the free nucleotide. It is thus clear that a comparison of the chemical and mutation rates of HA alone does not permit one to decide which of the two reactions with cytosine is mainly responsible for the induction of mutations. This decision will be possible only if the relative rates of the two chemical reactions can be drastically altered (e.g., by change of temperature, pH, or the use of HA derivatives) and the concomitant change in the mutagenic rate can be measured. The esterified derivatives of HA may be especially useful for the induction of specific mutations (involving only changes of cytosine) since they do not exhibit the strong inactivating effect. One can estimate the pH at which the mutagenic reaction should be maximal, if one neglects the effect of low pH on the opening of loops in DNA which also increase the reaction rate with HA.4 The rate of mutation induction by HA increases with decreasing pH,4 and would be expected to decrease again at very low pH (e.g., the chemical reaction with the C-C double bond shows this decrease8). The mutagenic reaction rate should therefore be proportional to either fHA X fc+ or fHA+ X fc (f = Mol-fraction of HA or cytosine, respectively). One can easily show that in both cases the rate would be maximal when

PHinax = '/2 (pKHA + pKC). Since the pKc of cytosine in DNA is lower than 4,9 the maximal reaction rate should be obtained at a pH lower than 5. The same equation can be used to calculate the pHmax of HA derivatives since their pK values are known; they have been reported for infinite dilution'0 and found by us to be slightly higher at concentrations of 10 mg/ml: N-methyl-HA (6.1), 0-methyl-HA (4.75), carboxymethoxyamine (4.4). The mutation rate of the esters of HA should therefore increase with decreasing pH even faster than that of HA, when compared to the rates at pH 6.2 (see Fig. 2). The practically useful pH is limited by the finding that the esters of HA seem to hydrolyze when the pH decreases below their pK value. The strong inactivating effect, found at lower concentrations of HA, does not seem to be caused by the reaction of HA with the C-C double bond of cytosine, because the rate of that reaction decreases with increasing pH,8 whereas the rate of inactivation increases with the pH (at least up to pH 10.8). The mechanism of the inactivating reaction is not known. It seems to cause, at least occasionally, the breakage of DNA strands, since HA treatment has been reported to reduce both the viscosity and the sedimentation constant of DNA. "' This effect is further cor- related to the finding that HA'2 and some of its analogues,7 having a free NOH group, induce chromosome breaks, some of which result in large chromosomal alterations in tissue cultures of hamster cells. Possibly, even the first cytological observation of a chemical induction of large chromosomal alterations'3 has already been con- cerned with the same chemical reaction, since (ethyl)urethane might be converted into the HA derivative hydroxy-urethane inside the cell. "I Downloaded by guest on October 1, 2021 1296 GENETICS: FREESE AND FREESE PROC. N. A. S.

The inactivating effect may be caused by either one of the following molecules that can be derived from HA or its derivatives of the form RHNOH. (1) The neutral species of HA, or possibly the negatively charged ion HA- (pK for its formation >12). (2) Nitroxyl (:NOH) which is produced (together with NH3) by the reaction of two HA's and is eliminated (to form N2) by the reaction with another HA. The elimination reaction essentially determines the equilibrium concentration of nitroxyl since there is much less free ammonia than HA in the medium; the concentration of nitroxyl should therefore increase linearly with that of HA. (3) Hyponitrous acid, which is formed by the condensation of two nitroxyls; its equilibrium concentration should increase with the square of the HA concentration. Our experiments have shown that at low concentrations of HA (<10-4 M), the inactivation rate increases approximately linearly with the HA concentration, and certainly not with the square of it (see Fig. 6). It seems therefore unlikely that either the cis or the trans form of hyponitrous acid is responsible for the inactivation effect (the trans form has been excluded directly). We conclude that the inactivation may be caused by one of the three species HA, HA-, or nitroxyl. Most puzzling is the fact that with increasing HA concentration the inactivation rate levels off and, under certain conditions, even decreases later. The leveling off seems to result from a saturation of DNA with the reactive species of HA. The attached molecules could subsequently inactivate DNA at a rate that would no longer depend on the HA concentration. The attachment itself would be readily reversible since dilution (in a medium containing acetone) immediately stops the inactivation of DNA. In order to understand the decrease of the inactivation rate at high HA concentrations, one would have to assume that HA interfered in some way either with the attachment of the reactive species or with the subsequent inactivating reaction. Several HA derivatives are strong carcinogens. They probably act by causing chromosomal breaks and large alterations' that entail abnormal growth, rather than by the point mutagenic action of HA. This is highly probable, since the concentration of these compounds inside cells is probably very low. It should now be possible to prove this contention by comparing the carcinogenic effect of HA derivatives, having a free NOH group, with that of esters of HA, which should induce point mutations but not large alterations (provided they are not hydrolyzed in the cell).

We wish to thank Mrs. Cynthia L. McAlister for technical assistance, and Mr. Lawrence S. Phillips for the measurement of pK values. I Freese, E., E. B. Freese, and E. Bautz, J. Mol. Biol., 3, 133 (1961). 2For a review see Freese, E., in Molecular Genetics, ed. H. Taylor (New York: Academic Press, 1963), pt. I, p. 207. 3 Kozloff, L. M., M. Lute, and K. Henderson, J. Biol. Chem., 228, 511 (1957). 4 Freese, E., and H. B. Strack, these PROCEEDINGS, 48, 1796 (1962). 5 Strack, H. B., E. B. Freese, and E. Freese, Mutation Research, 1, 10 (1964). 6Hughes, M. N., and G. Stedman, J. Am. Chem. Soc., 85, 1239 (1963). 7 Borenfreund, E., M. Krim, and A. Bendich, J. Natl. Cancer Inst., 32, 667 (1964). 8 Zillig, W., D. W. Verwoerd, and H. Koalhage, in Acides Ribonucleiques et Polyphosphates Structure Synthhse et Fonctions (Colloques Internationaux, Centre National de la Recherche Scientifique, 1962), vol. 106, p. 229. Downloaded by guest on October 1, 2021 VOL. 52, 1964 BIOCHEMISTRY: THOMAS AND MACHATTIE 1297

9 For example, Jordan, D. O., The Chemistry of Nucleic Acids (London: Butterworths, 1960), p. 171. "° Bissot, T. C., R. W. Parry, and D. H. Campbell, J. Am. Chem. Soc., 79, 796 (1957). Bendich, A., E. Borenfreund, G. C. Korngold, and M. Krim, Federation Proc., 22, 582 (1963). 1Hsu, T. C., and C. F. Somers, these PROCEEDINGS, 47, 396 (1961); Somers, C. F., and T. C. Hsu, these PROCEEDINGS, 48, 937 (1962). 3 Oehlkers, F., Z. Vererbungslehre, 81, 313 (1943); Oehlkers, F., Heredity, 6, suppl., 95 (1952).

CIRCULAR T2 DNA MOLECULES BY C. A. THOMAS, JR., AND L. A. MACHATTIE DEPARTMENT OF BIOPHYSICS, JOHNS HOPKINS UNIVERSITY Communicated by A. D. Hershey, September 30, 1964 Each T2 bacteriophage particle yields a single linear duplex DNA molecule, an example of which may be seen in the electron micrograph (Fig. 2B). Although each molecule contains the same genetic message, the order of the nucleotides appears to be different from molecule to molecule. Renaturation experiments indicate that those sequences that are found near the ends of some molecules are found near the middles of others. ' This is precisely what would be expected if each linear molecule had a nucleotide sequence which was a different circular permutation of a common basic sequence. (A collection of linear molecules with circularly permuted sequences can be generated by making a single random break in each of a collection of identical circular molecules.) Other bacteriophage DNA molecules such as T5 (ref. 1) and X (ref. 2) are known to be "unique" in that most of them have the same nucleotide sequence. The unusual situation in regard to T2 or the related T4 is probably the physical basis for the circular genetic map observed in this phage. 3 If these molecules are circular permutations of each other, and if they consist mainly of two continuous polynucleotide chains as shown previously,4 then denaturation followed by reannealing as depicted in Figure 1 should lead to the formation of artificial circular molecules. Experiments.-DNA molecules from T2 bacteriophage were extracted with phenol and purified by chromatography5 as described previously.4 These molecules were diluted tenfold into 0.20 M NaOH at a final concentration of 1.25 'y/ml. After one minm 1/,o vol of 3.0 M NaCl, 0.30 M Na citrate, was added and the solution dialyzed for 10 hr against 0.30 M NaCl, 0.03 M Na citrate. This solution was then heated at 65°C for 40 min and cooled to 4°C. Visualization in the electron micro- scope was accomplished by the method of Kleinschmidt6 as previously described.7 The present procedure renders duplex DNA clearly visible, but single polynucleotide chains are not seen. An aliquot of the same solution receiving all treatments except that of denaturation by NaOH served as a control. In addition to the experiments done on T2 whole molecules, T2 half molecules (41% relative length or 54 million molecular weight), and T5 whole molecules were given the same experimental and control treatments. Results.-Grids prepared from the solutions of unbroken T2 molecules that had been treated with NaOH followed by reannealing at 650 showed many closed circles, Downloaded by guest on October 1, 2021