Complete Nucleotide Sequence of the Cohesive Ends of Bacteriophage P2 Deoxyribonucleic Acid*

Abstract At the ends of bacteriophage P2 DNA, the 5'-terminated strands are 19 nucleotides longer than the 3'-terminated strands. The complete nucleotide sequence of the two cohesive ends of P2 DNA has been determined. This was accomplished by repair synthesis, using Escherichia coli DNA polymerase and labeled deoxyribonucleoside triphosphates with or without a ribonucleoside triphosphate, followed by partial enzymatic digestions of the labeled segments and sequence analysis of the isolated oligonucleotides. Starting from the 5' end of one cohesive end, the 19 nucleotides are in the sequence dpGpTpGpCpTpTpTpCpCpCpCpGpCpCpTpCpGpCpC. The sequence from the other cohesive ends is exactly complementary to this one. The underlined bases differ from the sequence of the corresponding cohesive end of bacteriophage 186 DNA. Thus, DNA molecules from two related phages can form mixed dimers and help each other in infectivity assay even though the cohesive end sequences are not identical. The functional and thermodynamic significance of these findings is discussed.


Ithaca,
New York 14850 SUMMARY At the ends of bacteriophage P2 DNA, the S'-terminated strands are 19 nucleotides longer than the 3'-terminated strands.
The complete nucleotide sequence of the two cohesive ends of P2 DNA has been determined.
This was accomplished by repair synthesis, using Escherichia coIi DNA polymerase and labeled deoxyribonucleoside triphosphates with or without a ribonucleoside triphosphate, followed by partial enzymatic digestions of the labeled segments and sequence analysis of the isolated oligonucleotides. Starting from the 5' end of one cohesive end, the 19 nucleotides are in the sequence dpGpTpGpCpTpTpTpCpCpCpCpGp-CpCpT'pCpGpCpC.
The sequence from the other cohesive ends is exactly complementary to this one. The underlined bases differ from the sequence of the corresponding cohesive end of bacteriophage 186 DNA. Thus, DNA molecules from two related phages can form mixed dimers and help each other in infectivity assay even though the cohesive end sequences are not identical.
The functional and thermodynamic significance of these findings is discussed.
We have previously reported the complete sequence of the lefthand cohesive end of bacteriophage 186 DNA (3). P2 DNA belongs to the same family as 186 DNA on the basis of (a) formation of mixed dimers between the 2 DNA molecules (4); (b) the ability of I DNA molecule to help in the infectivity assay by the other phage (4); and (c) the extensive hybridization homology between them (5). Furthermore, P2 and 186 have similar morphology (6, 7), and depend on the Escherichia coli rep gene product for DNA replication (8). In order to understand the morphogenesis and evolution of these phages, we have carried out sequence analysis of the P2 cohesive ends, using two independent methods.
The first method is the same as published previously (3,9,10). In this method, the l>NA was labeled with one Y'labeled and all four 31-i-labeled deoxynucleoside triphosphatcs. The labeled T)NA was then digested partially with micrococcal nuclease and the oligonucleotides were separated by two-dimensional electrophoresis (I 1) and sequenced.
In the second method, we used the conditions for the incorporation of one ribonucleotide in the presence of two or three deoxyribormcleotides, DNA polymerase I and MI?+ (12, 13) for repairing the cohesive ends. The cohesive ends of the I>NA were then cleaved with a specific ribonuclease.
When rGTP,i dCTI', and dTT1' were used to repair the cohcsivr ends, ribonuclease Tr was used for the cleavage of the partially repaired cohesive ends. When rCT1' dGTI', and / dAT1' were used to label the cohesive ends, pancieatic RNase was used for digestion.
The sequence analysis of the cleavage products of the ribo-substituted I)NA confirmed the sequence derived by the first method.
In this communication, we wish to report the complete sequence of the cohesive ends of 1'2 I)NA which is found to be homologous but not identical with 186 1)NA. Partial sequences for the cohesive ends of 1 2 DNA have been reported by Murray and ,vurray (14) who used a terminal labeling method.
Thcsc authors have determined the scqucnce of a 9.mer and a 12.mer from the two 5' cuds, and they have deduced the complete sequence by assuming that the two cohesive ekids are complementary and are 19 nucleotides in length, even though the actual length of the cohesive ends can not be obtained with their method.

MATERIALS ANI) METHODS
DNA-Phage I?2 was grown and purified according to previously published procedures (15,16). DNA was extracted by phenol and further purified by sucrose gradient sedimentation as described elsewhere (17).
Enzymes-Human semen phosphatase was purified according to the unpublished procedure of Doctors M. Singer and I,. Heppel (the details are given in Ref. 17). Purified spleen phosphodiesterase which is free of phosphatase activity was a gift of Dr. G. Bernardi, Institut de Biologie Moleculaire, Universite Paris VII. Ribonuclease TX was from Calbiochem.
r Where there is no ambiguity, the symbols A, G, T, and C are used throughout to stand for deoxynucleosides. The prefix d is used in some cases for emphasis. The prefix r is used for all ribonucleotides.
3H out on a small scale trial run with two levels of the enzyme before each large scale incubation (3,9). The degradation products were separated by one-or two-dimensional ionophoresis (11) or homochromatography (22,23 showed the presence of one ApG, two GpG, and one CpG sequence (Experiment 2). When dGT1; and d'i'TP were present 1 pi' and 4 ;G residues were incorporated. Nearest neighbor analysis * indicated the presence of one AiG, two G;G, and one TpG sequences (Experiment 3). Again, in analogy to 186 DNA it was assumed that the sequence of nucleotides added to the right-hand end was G-G-C-G-OH and the sequence at the left-hand end was G-T-G-011 ( Fig. 1). Although the possibility existed that the sequence at the right-hand end could be G-G-T-G and at the left-hand cwd be G-(-G, it was ruled out by the complete sequcrm determination of the cohesive ends of P2 DNA. When d&Tl', dA '1'1',and d'i'T1' were present (Table I,Experiment 4), no incorporation of pA residue occurred and 3' end group analysis showed a 1 : 1 ratio of Gp :G, supporting the sequence as shown for Experiment 3. When d e:TP, dCT!?, and dkT1' were present * .
(Experiment 5), 4 pC, 5 pA, and 11 pG residues were incorporated. As will be shown later, the entire length of the right-hand cohesive end of P2 1)NA was repaired in this experiment.
Since the second nucleotide to be added at the left-hand end (pT) was not present, repair synthesis presumably did not take place beyond the addition of a bG residue at this end, and 19 nucleotides were added to the right-hand end.
3' End Group and Nearest Neighbor Analyses of P1 DNA Completely Labeled at Cohesive Ends--By using one labeled and three nonradioactive deoxynucleoside triphosphates, it was found that 14 pG, 14 pC, 5 pT, and 5 pA residues were incorporat.ed into both cohesive ends of 1'2 DNA to give the total number of 38.
Thus, each cohesive end is 19 nucleotides long, similar to the cohesive end of 186 DNA. In order to determine the 3' ends of the completely labeled cohesive ends, all four tritium-labeled deoxynucleoside triphosphates were used for incorporation.
Analysis for its 3' ends showed that P2 DNA has only C at its 3' ends after repair and the ratio of Cp:C was 6: 1. Since there were 14 pC residues iricorporated into the cohesive ends, this could only be explained if there were 2 C residues at the two 3' ends and 12 pC residues at the internal regions of completely repaired cohesive ends. This was confirmed by the determination of the 5' ends of native t The number of residues incorporated was obtained after the plateau was reached under standard conditions of incubation as described previously (9).
$ Labeled DNA was digested with micrococcal nuclease and spleen phosphodiesterase to 3'-nucleotides and nucleosides followed by separation by two-dimensional paper chromatography as described previously (9). Nucleotides within brackets represent the original 3' end nucleotide present in the native molecule of P2 DNA before repair synthesis. Symbol * over p residue denotes 32P label and the symbol . over the nucleoside moiety denotes 3H label. P2 DNA using the procedures as described previously (28). It was found that both 5' ends of 1'2 DNA were pG. 1 he method used for the determination of 16 possible dinucleotide sequences at the repaired cohesive ends of I'2 DNA was the same as described by Wu and Taylor (10). The results summarized in Table II suggest that the rmcleotide sequences at the cohesive ends of P2 DNA are complementary and of opposite polarity. For example, there were two GpA to match two TpC sequences and one GpT to match one ApC. There were three ApG but only two CpT sequences which can be explained as follows.
It has been established (see Table I, Experiment lb) that ;G was the first nuclcotide to be added to both ends and one of them was next to the pA at the natural 3' end, thus giving one extra ApG sequence. For the same reason, there was one extra GpG over CpC sequence. There could be either 1 or 2 molecules of TpG per molecule of XDNA on the basis of the experimental value of 1.4. However, since only 1 molecule of CpA was found, there must be only 1 TpG.
The results are expressed as the number of indicated dinucleotide sequences in the repaired ends of one 1'2 DNA molecule. The calculation was based on a knowledge of the total number of the particular 32P-labeled nucleotide residues incorporated into the cohesive ends of 1 molecule of DNA.
The values are the average of three sets of experiments.
For the explanation of symbol *, see Table I  ;k, pA, pG, p?' (Fig. 2) and ;G, pA, pdl, and pi' (Fig. 3) after limited digestion with different amounts of micrococcal nuclease. The structure of oligonucleotide 1 was established as follows.
Venom phosphodiesterase digestion (Analysis V) showed that A was at the 5' end and this was a pentamer with the composition A(G, C , A, C) Spleen phosphodiesterase digestion (analysis So) indicated that C was at the 3' end, thus A(G,C,A)C-OH could be deduced; since one AiC and one GiC sequences were present, the sequence ApG;CpAiC-OH was deduced. The alternative to have a phosphate at its 3' end.
The condition for semen phosphatase treatment is given under "Materials and Methods." § L and K. refer to the arbitrary assignment of left-hand and right-hand cohesive ends of P2 DNA (see Fig. 9), respectively. All nucleotides contain tritium; the dots on the nucleosides are omitted in the structure given here; the symbol * is as defined in Table  I.
* * sequence ApApCpGpC-OH was ruled out in an experiment where ;G was used in place of ;C (Fig. 3), and a AiG rather than a C;G was found.
Since this oligomer was terminated with a C-OH at the 3' end, this must belong to the 3' end of one of the cohesive ends to complement the G at the 5' end.

Oligonucleotide
2 was found to be a hexamer, ApApGiCpAGC-OH, due to the presence of an extra A residue compared to oligonucleotide 1. Similar analysis established the structure of oligo- Y, position of nucleotide 4 as AiGpCp which is included in the sequence of oligonucleotides 1 to 2. Oligonucleotide 7 was a tetramer with a C at the 3' end, and a C at the 5' end. In addition, there were one G;C and one C;C sequences and hence the sequence deduced was CpGiCiC-OH.
The alternative structure CiCpG*pC-OH was ruled out by the structure analysis of a related oligonucleotide (not shown) Tp*CpGGC-OH.
Occasionally we have observed that the repair synthesis stopped in some DNA molecules (30% of all molecules in this experiment) at the nucleotide penultimate to the 3' end of the completely repaired cohesive ends as in the case of TiCp-G;C-OH (L15 to 18). The sequence of oligonucleotide 10 was established unequivocally as T;CgCjCiCp. Oligonucleotide 12 had a T at its 5' end and a composition of T(Cs,G) could be deduced from the results of Analysis V. Spleen phosphodiesterase digestion after semen phosphatase treatment showed a C at its 3' end; a sequence of T(CI,G)C could thus be deduced. The structure of oligonucleotide 15 which is identical with oligonucleotide 13, but obtained from a DNA digest labeled with pA, ;G, PC, and pi', was * shown to be TpGpCp.
This also rules out the alternative possibility of TiGpCpGi for oligonucleotide 14 because the 3' POa group in TpGpCp was not labeled with 32P in oligonucleotide 15.

Sequence Analysis of Pyrimidine
Oligonucleotides from Labeled Cohesive Ends of P2 DNA- Fig.  4a shows the two-dimensional fractionation of the phosphorylated pyrimidine oligonucleotides obtained as described under "Materials and Methods." Nearest neighbor, 3' and 5' end group analyses of these oligonucleotides (after dephosphorylation) are given in Table IV. The analysis of Spot 2 shows the presence of one CiT, the presence of a C at both the 3' end and 5' end and 1 additional Cp residue; the sequence C(CiT)C-OH can be deduced. The alternative sequence C(TpC)C-OH could not be ruled out from this data. This was ruled out, as will be discussed later, by an independent observation using a synthetic octanucleotide as primer for the repair synthesis of the cohesive end of P2 DNA.
The analysis of Spot 3 (Table IV, Experiments a and b) showed the presence of a total of 5 C and 3 T residues. Both 3' and 5' ends were b Y B FIG. 4 (left). For labeling the pyrimidine cluster, 4 pmoles of P2 DNA were used. The labeled DNA was denurinated as described elsewhere (3,20)  found to be C and there were one CiT and two T;T sequences. The analysis of Spot 4 which was identical with Spot 3 but obtained from a DNA labeled with id and pi' showed the presence of three C&Z and one TiC sequences; the sequence C(T-T-T, C-C-C)C-OH could be deduced for Spot 3 (and 4). The exact structure of Spot 3 (and hence 4) was established by isolation of the degradation products after partial spleen phosphodiesterase (Table V, Experiment a) and partial venom phosphodiesterase (Experiment b). Partial spleen phosphodiesterase digestion of the pyrimidine cluster showed a successive removal of Cp, Tp, and Tp residues from its 5' end. This established the location of the two TGT sequences. This information together with the 5'-nucleoside analysis of oligonuoleotides 2, 3, and 4 established the sequence of the octanucleotide as C-T-T-T-C-C-C-C-OH.
This sequence was confirmed in Experiment b. This finding was consistent with the fact that a synthetic octanucleotide with the sequence, C-T-T-T-C-C-C-C-OH could serve as primer for the DNA polymerase I-catalyzed repair syn-by one-dimensional ionophoresis (pH 1.9) at 2200 volts for 14 hours. thesis. The sequence beyond the 3' end of this octanucleotide primer was established as G-C-C-T-C-OH. As mentioned above, this ruled out an alternative possibility of sequence for Spot 2 in Table IV and established its sequence as C-C-T-C.

Sequence Analysis of Ribonucleases T, and A Digestion Products from Ribonucleotide-substituted
Cohesive Ends of Pb DNA-In order to isolate more fragments from the cohesive ends of P2 DNA for a better overlap with the previously determined sequences, the original observation of Berg et al. was used (12). The cohesive ends of I2 DNA were repaired in the presence of one ribonucleoside triphosphate, two deoxyribonucleoside triphosphates, DNA polymerase, and Mn*+ ions. The conditions used for incorporation were essentially the same as described previously (24) and are given under "Materials and Methods." The reason for using only two deoxynucleoside triphosphates instead of three, in the presence of one ribonucleoside triphosphate, is the following.
The incorporation data shown in Table   I, Experiment 5 when only dGTP, dCTP, and dATP were present (or when dGTP, dCTP, and dTTP were present) only one of the The pyrimidine oligonucleotides were separated by ionophoresis as described under "Materials and Methods" from P2 DNA labeled with I;? and pd (Spots 1 to S in Fig. 4a) or CC and pi' (Spot 4 in Fig. 4b) in the presence of nonradioactive pA and pG. Spols I, 2, and S from Fig. 4a were dephosphorylated with semen phosphatase as described under "Materials and Methods" before they were analyzed in Experiment a or b. Spot 4 was identical with dephosphorylated Spot S except that the DNA used to isolate it was labeled with CC and p'i'. Results after complete spleen phosphodiesterase digestion for 3' end and nearest neighbor analysis are shown in Experiment a. Results after complete venom phosphodiesterase digestion for 5' end and composition analyses are shown in Experiment b. All nucleosides are tritium-labeled and dot.s on nucleosides are omitted here for simplicity, only the location of 32P is shown by an asterisk. (Pu) CT-(Pu) C as 5' end C as 5' end C as 5' end two cohesive ends was completely repaired. After the repair with the substitution of the ribonucleotide, the labeled DNA was digested with the appropriate ribonuclease and the products separated by one-dimensional homochromatography.
The results are shown in Fig. 5 (a and 5). The analysis of Spots 1 and 2 from Fig. 5a is given in Table VI. Spot 1 was a nonanucleotide containing 3 T, 5 C, and 1 rG residues; rG must be at the 3' end since this oligonucleotide was the product of ribonuclease Ti digestion of labeled DNA.
The 5' terminus of this oligonucleotide was a C (Experiment c, Spot 1). Since this oligomer contained one T;C, three C;C, one rG;C sequences, the sequence Cp(Ts, TjC;CjCjC)rGj can be deduced. 1 he same oligomer obtained from a DNA labeled with ;'I?, PC, and rpG indicated the presence of one ClT and two T;T sequences (Experiment b, Spot 1) also in the oligomer. Thus, the sequence Ci(TiTiT ,pCpCpCpC)-rGp can be deduced for the oligomer in Spot 1 (Experiment b). The sequence within the parentheses was determined after partial spleen phosphodiesterase digestion of the oligomer 1 (Experiment a, Spot 1) as shown in Fig. 6. From the characteristic shifts in mobilities of the nine spots in Fig. 6, and from the nearest neighbor data shown in Table VI, the sequence CpTpTpTjC~C&$-CrGi can be deduced for oligomer 1 in Fig. 5a. The presence of a rG$Z sequence in oligomer 1 indicates the presence of 6203 CpTpTpT;CiCECGCrG;C sequence in one of the cohesive ends of I'2 DNA.
The analysis of oligomer 2 in Fig. 5a showed the presence of 1 T, 3 C, and 1 rG residues. Oligomer 2 also contained one Tic, * * one CpC, and one rGpC sequences. It had a C as the 5' end, thus the sequence C(iCpTiC)rGG could be deduced. The two possible structures C"pCpT;CrG; and CpT&iCrG; could not be distinguished by these analyses alone. Hut by using a synthetic octanucleotide2 CpTpTpTpCpCpCpC-OH as primer for DNA polymerase-catalyzed repair synthesis, a pentanucleotide sequence GpCpCpTpC was determined beyond the 3' end of the primer. 2 It also established that ohgomcr 1 with the sequence rGpCpTpTpT~C~&$CrG~(C) was at the 5' end of oligomer 2 * * * with the sequence rGpCpCpTpCprGp(C).
The analyses of Spots 1 and 3 from Fig. 5b are given in Table  VII.
Spot 1 was a nonanucleotide containing 5 G, 3 A, and 1 rC residues. The 5' end of this oligomer was a G and it contained one GiA, two .$A, and one rC;A sequences, thus the * * sequence Gp(G4,Aa)rCp; the rCp was placed at the 3' terminus since this oligomer is the product of ribonuclease A digestion. The 3' terminus was 3Wlabeled as shown by the presence of one rCjA sequence. The complete sequence of this oligomer was established by partial digestion by spleen phosphodiesterase followed by fractionation by two-dimensional (ionophoresis-homochromatography) system as shown in Fig. 7. From the characteristic mobility shifts of the oligonucleotides, the presence of 4 G and 1 A from the 5' end of this oligomer was established. Thus the sequence GpGpGpG;AiAjApGrCp could be established unambiguously for this oligomer, and the presence of the sequence * * * * rCpGpGpGpGpApApApGrCpA in one of the cohesive ends was indicated.
The rCp at the 5' end is expected since the nonamer was released after ribonuclease A digestion of the ribo-c-substituted P2 DNA.
The ;A at the 3' end came from nearest neighbor transfer of Y from ;A to rC. The analysis of Spot 3 from Fig. 5b showed the presence of 3 pG residues and one G;A sequence (Table VII). A rCp residue must bc at the 3' end of this oligomer for the reason stated for Spot 1 but this was not 3'.terminally labeled since nonradioactive rCTP was used for incorporation and there was no rCpA sequence present. The 5' end of this oligomer was a G and thus, the sequence G(jApGpG)prCp could be deduced. The complete sequence of this oligonucleotide was obtained by partial digestion with spleen phosphodiesterase followed by the two-dimensional fingerprinting method (22, 27) as shown in Fig. 8.
From the characteristic mobility shifts of the ;G-and GA-labeled oligonucleotides, the sequence GlAiGiGprCi could be deduced for Spot 3 in Fig. 5b, and the presence of the sequence rCpGpAp-GpGprCp in one of the cohesive ends was indicated.
Alignment of Oligonucleotides to Give Complete Sequence-The sequences of the radioactive oligonucleotides, which arc complementary to the cohesive ends of P2 DNA, are derived from three types of experiments: (a) partial incorporation, (b) complete re- The pyrimidine cluster used in this experiment, identical with tate and 0.3 pg of purified venom phosphodiesterase. The incuba-Spot 4 mentioned under Table IV, was isolated from P2 DNA tion was for 17 hours. For partial venom phosphodiesterase labeled with $6 and p'i' in the presence of nonradioactive pA and digestion (1':xperiment b), 120 nmoles of the pyrimidine cluster PG. Spots 1 to 4 under li:xperiment a were obtained by partial with 7,500 cpm of 32P and 22,700 cpm of 3H were used. The maspleen phosphodiesterase digestion and Spots 1 and 2 under I':x-terial was divided into two parts and digested with two levels periment b were obtained by partial venom phosphodiesterase (0.08 and 0.16 rg) of the enzyme for 30 min at 37". The incubation digestion of the pyrimidine cluster. For partial spleen phos-mixture contained 40 mM Tris-acetate (pH 8.4) and 4 mrvr magphodiesterase digestion (P:xperiment a), 72 nmoles of pyrimidine nesium acetate. Nearest neighbor and 3' end group analyses of cluster with 3300 cpm of 32P and 13,600 cpm of 3H were used. The these products were carried out by spleen phosphodiesterase material was divided into two parts and digested with two levels of digestion.
Incubation was carried out at 37" for 30 min. The carriers.
The incubation was for 17 hours at 37". The nucleoproducts were separated by one-dimensional ionophoresis at pII tides and nucleosides were separated and analyzed as described 3.5 ('2LW.l volts for 8.5 hours).
The composition of the partial earlier (9). All nucleosides are tritium-labeled, but the dots on degradation product and its 5' end were determined by complete nucleosides are omitted here for simplicity.
Only the location of venom phosphodiesterase digestion. The incubation mixture 32P is shown by asterisk. contained 20 mM Tris-acetate (pH 8.5) and 2 mM magnesium ace- followed by partial digestion with micrococcal nuclease, and (c) ribo-substitution experiment (repair with ribo and deoxynucleotides) followed by digestion with ribonuclease. Of these experiments, the third type produced the largest amounts of sequence information.
Therefore, ribo-substitution experiments should be carried out whenever possible. Table I and Fig. 1 show that the sequeuce of nucleotides incorporated into one cohesive end of I:2 DNA during partial incorporation is G-G-C-G and the sequence at the other end is G-T-G (Possibility I). An alternative possibility (II) that the sequences G-G-T-G and G-C-G incorporated at the two ends could not be ruled out.
Experiment 5 in Table I indicated that 11 pG, 5 pA, and 4 pC residues were incorporated when only dCTP, dGTP, and dATP were present in the DNA polymerase I-catalyzed repair synthesis. If the repair synthesis starts with G-T-G (Possibility I) as the sequence at one end (see above), then the incorporation would stop at this end with 1 pG due to the absence of dTTP.
The rest of the 19 nucleotides will be incorporated at the other end. This is consistent with the length of the cohesive end obtained from the data given in Table II. On the other hand, if the repair synthesis starts with G-G-T-G (Possibility II) at one end, then 18 nucleotides would have been incorporated at the other end instead of 19 nucleotides.
The possible sequence G-G-T-G was also ruled out from the experimental evidence discussed below.
The repair synthesis in the prcsetice of dAT;j, dGTP, rCTl', and IVW+ which was catalyzed by DNA polymerase 1 should also take place essentially at one cohesive end. Therefore, the two fragments rC-G-G-G-G-A-A-A-G-rCpA and rC-G-A-G-G-rCp obtained in Fig. 5b should originate from one cohesive end (R end in Fig. 9). The sequence, rC-G-G-G-G-A-A-A-G-Cp could be joined by overlapping four nucleotides with the sequence of 3'. terminal oligonucleotide 2 in Table III, A-A-G-C-A-C-OH; thus the sequence rC-G-G-G-G-A-A-A-G-C-A-C-OH could be deduced to be present at the R end (see Fig. 9). The sequence rC-G-A-G-G-rCp which is also present at this end could only be at the 5' cud of the above dodecanucleotide sequence; thus a I7-nucleotide sequence rC-G-A-G-G-rCpG-G-G-G-A-A-A-G-C-A-C-OH can be derived by overlappiug a rCp. The sequence established by partial incorporation from the starting point of repair synthesis, namely G-G-C-G (Possibility I) or G-C-G (Possibility II) should also come from this end. Only Possibility I, G-G-C-G, can give rise to the correct length of 19 nucleotides for this end. In fact, 6205 only the sequence G-G-C-G-A-G-G-C-G-G-G-G-A-A-A-G-C-A-C-OH for the R end is consistent with all the data. Once the sequence determined by partial incorporation techniques at the R end is established to be G-G-C-G, the sequence at the L end must be G-T-G (see Figs. 1 and 9). This is also consistent with the occurrence of fragment G-T-G-C (oligonucleotide 14 in Table  III) in the micrococcal nuclease digest of labeled I'2 DNA.  7 (center).
Fingerprint of a partial spleen phosphodiesterase digest for Spot 1 from Fig. 56.
An aliquot of Spot 1 from the experiment described under Fig. 5b was digested as described under Fig. 6. The conditions for digestion and for the fractionation of the digest have been described elsewhere (23). FIG. 8 (right).
Fingerprint of a partial spleen phosphodiesterase digest of Spot 3 in Fig. 5b.
Spot 3 corresponding to the one in Fig. 5b

6206
The sequence C-G-C-C-OH for oligonucleotide 7 in Table III come from the 3' terminus of the cohesive end L. By overlapping three nucleotides with the 16.mer for the I, end mentioned earlier, the 17.mer sequcmcc G-C-T-T-T-C-C-C-C-G-CC-T-CG-C-C-OH can be established.
This could be joined to G-T-G-C by overlapping two nucleotides to give G-T-G-C:-'1.
-T-T-(:-C-C-C-G-C-C-T-C-G-C:-C:-OII which is the complete sequence for the cohesive elld L. This sequence is also consistent with that of all the fragments analyzed in Table 1 II. The correctness of the sequences for the two cohesive ends is supported by the total nurnber of tlucleotides incorporated at the two cohesive ends, and also their nearest Ilcighbor analysis shown in Table I I. Also, the scquenccs at the two cohesive ends are exactly complementary in agreement with the data in l'ablc I I. II represent sequences obtained from ribo-substitution experiments.

Differences and Similarities
between Cohesive End Sequences-The studies reported here were started as a part of our investigatiou of the role of cohesive ends in the formation of mixed dimers between 2 DNA molecules and in the helper-mediated infectivity assay. Two members of the same group of bacteriophages, such as X and 480 or I'2 and 186 are able to form mixed dimers with carh other and function as helpers.
ISut the members of different groups, such as X and 186 are unable to perform these two functions (6,29). Scqucnces at the cohesive ends of X and 186 have been shonn to be completely different (3,10). The studies on the scquencc tlctc~rmination of 480 and 1'2 started 2 years ago in our laboratory to find out whether their sequences are identical or similar to X and 186, respcctivcly.
This would define also the criterion for the formation of mixed dimers between 2 I>NA molcculcs, alld for one phago to serve as a helper in the infectivity assay by the other.
l'revioua stud& established that the sequenccs at the cohesive rnds of X and 480 11NA are exactly identical (14,30). Results from this report showed that, out of the 19 uucleotides at each cohesive end of P2 and 186 I)NA, two nuclcotides arc different.
In addition, our results point out the exact location of the differences between 186 and P2 1)NA as shown in Fig. 10. It also gives the exact length of the homology between the 2 DNA molecules as 13 nucleotides.
Our previous studies (31, 32) on the hybridization of oligonucleotides to native DNA indicated that a stable duplex can be formed between a I)NA and a nine-nucleotide long oligomer. Therefore, a 13nucleotide homology between P2 and 186 should be more than sufficient to form a stable duplex. This is indeed the case from the ability of the 2 I>NA molecules to form mixed dimers with each other. These studies also show that identical sequences at the cohesive ends of 2 E)KA molecules are not required to be able 10. Comparison of the sequences at one of the cohesive ends of P2 and 186 UNA.
The differences are pointed by arrows.
to form mixed dimers and to function in helper-dependent infectivity of DNA molecules. Thermodynamic Considerations-The thermal stability of endto-end DNA aggregates has been studied for PZ-related phages 1 '2, 186, and 299 (4,33).
The first results indicated that the base sequences at the cohesive ends of P2 and 186 are similar but not identical.
An expanded study by Wang et al. (34) gave a more precise conclusion.
These authors proposed that 186 cohesive ends have at least one internal AT pair, replacing a GC pair in the 1'2 cohesive ends. The possibility of another mismatched base pair was not eliminated.
A second mismatch between 186 and P2 cohesive ends has now been found by two sequencing methods (this report and Ref. 14). From these considerations, one can see that sequence analysis can resolve problems which thermodynamic: analysis carmot reach.