Proc. NatL Acad. Sci. USAVol. 79, pp. 3498-3502, June 1982Biochemistry

Bacteriophage A DNA packaging: Scanning for the terminalcohesive end site during packaging

(terminase)

MICHAEL FEISS AND WILLIAM WIDNERDepartment of Microbiology, University of Iowa, Iowa City, Iowa 52242

Communicated by Franklin W. Stahl, March 11, 1982

ABSTRACT Bacteriophage A packages the DNA of the re-lated phage 21 poorly [Hohn, B. (1975)J. Mol. Biol 98, 93-106].To understand the nature of the packaging defect, the interactionof the cohesive end site (cos) specific for phage 21 (coss2l) withphage A terminase has been investigated. The ability of A ter-minase to cleave cos42l was studied in vitro; A terminase cleavedcos421 only 1% as well as it cleaved the phage A cohesive end site(cosA). In vitro packaging experiments showed that the A and 21packaging specificities observed in vivo are also found in vitro. Thecos cleavage reaction was modified so that competition experi-ments could be performed; these experiments showed that cosO21was unable to bind A terminase, thus identifying the nature of thedefect. Previous work [Feiss, M., Fisher, R. A., Siegele, D. A.,Nichols, B. P. & Donelson, J. E. (1979) Virology 92, 56-67] hasshown that the base pairs giving A or 21 packaging specificity areat the left end of the chromosome, outside the 22-base-pair sym-metry region that includes the annealed cohesive ends. Therefore,terminase binding to cos requires interactions with base pairs tothe Nul side of the cohesive end symmetry segment. The evidencesupports the proposition that cos consists ofadjacent sites for bind-ing ofterminase and for nicking by terminase. Because coSO21 canbe cut by A terminase to terminate DNA packaging, it is proposedthat the terminase that binds and nicks at the initial cos site isbrought into contact with the terminal cos site by the packagingprocess. Terminase recognizes and nicks the cohesive end se-quence of the terminal cos without requiring the binding site.

The phage A chromosome is a linear double-stranded DNAmolecule ca. 49 kilobase pairs (kbp) in length (1). At the 5' endsare single-stranded segments, 12 bases long, that are comple-mentary; they are called cohesive ends (2). The presence of thecohesive ends results in cyclization of the DNA molecule afterinjection. The resulting nicks are then sealed by host ligase. Atearly times after infection, replication produces progeny ringsthat are not a substrate for packaging; at later times, rolling cir-cle replication produces the linear multichromosomal lengthsofDNA, called concatemers, that are the substrate for the pack-aging process (reviewed in ref. 3).DNA packaging for phage A involves specific recognition of

A DNA from a pool that also includes bacterial DNA. The firstidentified step in DNA packaging is likely to involve specificinteractions; this step is the formation ofcomplex I between thephage A DNA and terminase, the product of phage A genes Aand Nul (4, 5). The presence of a specific A DNA site(s) is in-dicated by the ability oftransducing variant chromosomes to bepackaged as long as a DNA segment including the site of theannealed cohesive ends is present (see ref. 1). The maximumsize of the segment is delimited by studies showing that DNAmolecules are packaged even if they contain gross rearrange-

ments 200 bp from the Nul gene (left) end and 100 bp from theR gene (right) end of the chromosome (1, 6). The site of actionof A terminase is called the cohesive end site (cos) (7, 8). Com-plex I then interacts with the preformed protein shell, or pro-head, to form complex II (5). Terminase plays a role in proheadbinding, in that proheads apparently bind to complex I but notto A DNA or terminase alone. The prohead binding function ofterminase may be by direct interaction or by indirect activationofaDNA site for prohead binding. DNA packaging follows com-plex II formation, and, in turn, is followed by cleavage of theDNA by terminase to generate cohesive ends (4, 5, 9). Thus,three functions ofterminase can be identified; cos binding, pro-head binding, and cos nicking. cos nicking will refer specificallyto the introduction ofstaggered nicks to generate cohesive ends.cos cleavage will refer to the in vitro reaction, described below,in which A terminase cleaves a DNA substrate at cos: cos cleav-age includes the cos binding and cos nicking functions ofterminase.

Because phage A terminase cuts the DNA to generate cohe-sive ends, the site of the annealed cohesive ends is likely partof cos. Sequence determinations show a 22-bp segment withstrong rotational symmetry (10). The segment is strictly sym-metrical for 10 bp, and the positions of the nicks produced byterminase and an additional 10 bp have purine-to-pyrimidinesymmetry. By analogy with type II restriction endonucleases,it is likely that the symmetry-is offunctional significance as partof cos.We have been studying the packaging specificities of phage

A and the related lambdoid phage 21. Heteroduplex DNA mol-ecules between A and 21 show considerable homology in thehead gene region, so these two viruses are probably descendedfrom a common ancestor. In spite of the relatedness, each viruspackages the DNA of the other with an efficiency of 1% or less(4, 11). Hence phage 21 is an interesting natural variant for pack-aging specificity. We found that a hybrid, A-21 hybrid phage19, which has all A sequences except for the 21 head gene regionat the left chromosome end, has phage 21 packaging specificity.Thus, the specificity determinant(s) reside at the left chromo-some end. Sequence determination of the DNA revealed thatthe 22-bp symmetry segments of A and the A-21 hybrid 19 areidentical, indicating that base-pair differences in the chromo-some left end outside the 22-bp symmetry segment must beinvolved in recognition. A puzzling observation was that pack-aging specificity occurs at the cos site from which packaging isinitiated but not at the cos site at which packaging is terminated.

In the study presented here, we have investigated the in-teraction of phage A terminase with cosq521 and found a defectin the interaction. The cos site of A-21 hybrid phage 19 will be

Abbreviations: cos, cohesive end site, specificities noted for phage A

or phage 21 as in coso21; kbp, kilobase pair(s).

3498

The publication costs ofthis article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

Proc. Nati. Acad. Sci. USA 79 (1982) 3499

2/

100/0

H3

6.0

lq

H3

-~~~~~ ~1.3(6.3)

Bg soI I

0/K00 97.0

Sq so

1.0/97.3 \97.0

pMF517 097.3 100/0 9.0

RI 30

Pts

363 652

pBR322

FIG. 1. cos plasmids. The plasmids consist ofphage-derived DNA segments inserted into pBR322(wavy circle at bottom). Plasmid pBW3 carriescos42l; pDS15, pMF1, and pMF517 carry cosA; pMF2lacks a cos site. Inserts are oriented as inserted intopBR322. Restriction endonuclease sites shown are:HinduII (H3), BamHI (Ba), Bgl II (Bg), Pst I (Pt),EcoRI (RI) and Sal I (Sa). The pBW3 and pDS15 seg-ments are inserted into the H3 site of pBR322, theremaining segments are inserted into the Ba site;insertions into either site inactivate the tetracycline-resistance determinant of pBR322. DNA segmentsare derived from: phage A or 4480 (-), phage 21() and E. coli (o). cos sites are noted by a small cir-cle; the specificity of each is given above. Positionson the inserts are in percentage of A+ units (100%= 49 kbp), with 0 at the left (Nul gene) end and 100at the right (R gene) end. The length of the pDS15insert from cos to the right end is given in parenthe-ses. The pBW3 insert is derived from the A-21 hybridphage 19 (13), and the pDS15 insert is derived fromthe cos duplication of Aa4.5061.1 (17, 18). The insertsfor the remaining plasmids are from A charon 4 (19);the 97.0-100 segment is 480 DNA, and the rest is ADNA; pMF2 was derived from pMF1 by Bgl E diges-tion to delete cosA. Positions of restriction endonu-clease sites on pBR322 (total length, 4,362 bp or 8.9%of A+) are given in base pairs. Lengths are drawnapproximately to scale.

referred to as cos421 because A-21 hybrid 19 has the packagingspecificity of phage 21 (11).

MATERIALS AND METHODSStrains. E. coli strains were R594, a supo strain (12); MF611,

a recAl sup' strain (13); MF902, a derivative ofMF611 carryingmultiple, tandem A' prophages; NS428, which is N100 (recAsup') carrying AAamll b2 red3 c1857 Sam7 (14); and MF871,W3350 (sup') carrying AcI857 Sam7 ga1805 (15). Phages wereAcI857 (16) and A-21 hybrid cd857 (11).

Plasmids. cos plasmids used are diagrammed in Fig. 1. Con-struction of pDS15 is described in detail elsewhere (17); theothers were constructed and characterized in like manner. Thecomposite plasmids pGM2, pBW5, and pBW6 (see Fig. 3 Up-per) were prepared by the method of Miller and Feiss (17),which makes use of the Kaiser-Hogness (20) helper-mediatedtransformation system. In this system, free DNA containing a

phage A cohesive end is taken up by A-infected cells. Cohesiveend annealing occurs between free DNA and the helper chro-mosome. When cos plasmid DNA, cleaved in vitro at cos withphage A terminase, is used, the resulting transformants (am-picillin resistant) carry a composite plasmid with the helperchromosome joined to the plasmid DNA by cohesive end an-

nealing. To prepare the composite plasmids pGM2 and pBW6,cleaved pDS15 plasmid was used, and cleaved pBW3 was usedto prepare pBW5. The helper was AcI857 for the preparationof pGM2 and pBW5 and A-21 hybrid 19 cd857 for the prepa-ration ofpBW6. Transformation was performed as described byMiller and Feiss (17) using strain MF902. Plasmid DNA was

ARstrktionEndonuciose

isolated by the method of Clewell (21). For the composite plas-mids, a small scale (40 ml) version of the method was used (17).

cos Cleavage Reaction. The reaction is outlined in Fig. 2;conditions are those ofBecker and Gold (22). The 70-A.l reactionmixture contained 20 A.l of buffer A (22) at double strength, 41l of buffer M2 [6 mM Tris-HCl, pH 7.4/60 mM spermidine/18 mM MgCl2/15 mM ATP/28 mM 2-mercaptoethanol], 0.75,ug of linear cos plasmid DNA in 6 1.l of 10 mM Tris-HCl, pH8/0.1 mM EDTA, 20 pl of sonic extract of uninfected MF611cells, and 20 1.l of partially purified terminase. Incubation was

for 15 min at 20TC, and the nucleic acids were isolated (22). Theamount of cos cleavage was determined by analyzing agaroseelectrophoresis gels of the reaction products by ethidium bro-mide staining or by Southern transfer. For staining, the gel wasilluminated with UV light and photographed. Densitometertracings of the films were taken and integrated (in the linearresponse range). For transfer, DNA in gels was transferred tonitrocellulose by the method of Southern (23). Hybridizationof 32P-labeled plasmid probes was for 18 hr at 65TC as describedby Southern. Autoradiograms were analyzed by densitometryas described for staining. The 32P-labeled probes were preparedby nick translation with the method of Maniatis et aL (23). Ter-minase was purified from a sonic extract of induced MF871cells. The purification steps, polyethyleneimine precipitationand (NH4)2SO4 precipitation followed the procedures of Blatt-ner et aL (24).

Complete Packaging Reaction (15, 24). The reaction mixturecontained 20 1.l of buffer A at double strength, 4 1.l of buf-fer M2, 5 1.l (0.75 Ag) of DNA in 10 mM Tris-HCl, pH 8/0.1

Terminose+

FIG. 2. cos cleavage assay. A cos plasmid is first cleaved with a restriction enzyme and then used as a substrate in the cos cleavage reaction.

113pBW39

H3pDS15

pFI

pMF2

100/0

soLIL3

11.3

Iw It .

Biochemistry: Feiss and Widner

Gl

R,;

3500 Biochemistry: Feiss and Widner -

mM EDTA, 20 /.d of sonic extract of induced cells of NS428(source of host factors and proheads), and 20 1ul of terminase.After 15 min at 20'C, 150 A.l of freeze-thaw lysozyme extractofinduced NS428 cells was added [source ofproducts ofA genesD, W, and FII and of tails (15)]. The reaction was terminatedafter an additional 15 min of incubation in the case of cos cleav-age or after 60 min in the case of packaging assays.

RESULTSPackaging Specificity During in Vitro DNA Packaging. To

draw conclusions about cos-terminase interactions from in vitrostudies, it seemed important to verify that the packaging spec-ificity that is observed in vivo also occurs in vitro. Phage A chro-mosomes are packaged from a concatemer-that is, from be-tween two cos sites (see ref. 25). The packaging is polarized ina Nul-A-to-R direction (7, 18); so, for any chromosome beingpackaged, there is an initial cos site at the Nul-A end (at whichpackaging is initiated) and a terminal cos site at the R end. Aprevious study showed that phage A or phage 21 specificity wasonly found at the initial cos site (11). Bacteriophage A will onlyefficiently package chromosomes that have cosA in the initialposition but will efficiently package chromosomes with eithercosA or cos421 at the terminal position.

Therefore, we wished to ask whether this same specificityoccurred in the in vitro packaging system. Accordingly, com-posite plasmids containing a packagable phage A chromosomewere constructed for use as DNA substrates in in vitro pack-aging reactions. These plasmids had cosA or cos421 at the initialor terminal position as diagrammed in Fig. 3.

The plasmids were isolated and used in the complete pack-aging system; the results are shown in Fig. 3. From the first twoplasmids (pGM2 and pBW5) phage chromosomes were pack-aged with the same efficiency by the A packaging system, andfew were packaged from the third (pBW6). The efficiency ofpackaging (phage yield per number of input composite plasmidmolecules) was 3.3 x 10-4 for pGM2 and pBW5 and 4.4 X 10-7for pBW6. Thus, the presence of cos421 at the initial positionlowered packaging efficiency by a factor of 1,000. In agreementwith the previous in vivo results, packaging specificity is foundonly at the initial cos site (specificity is also observed whenmature DNA is used; A-21 hybrid phage 19 DNA is packagedless efficiently by a factor of 1,000 than A DNA) (unpublisheddata).A Defect in the A Terminase-cos421 Interaction. To ask

whether phage A terminase can interact with cos421, we usedthe in vitro cos cleavage reaction of Becker and Gold (22). Thecos cleavage reaction is a prohead-independent reaction inwhich a linear DNA substrate containing a cos site is cleavedby terminase to yield two DNA fragments that are detected byagarose gel electrophoresis. The DNA substrates used are plas-mids containing the cohesive end site of phage A (pDS15, thecosA plasmid) or A-21 hybrid phage 19 (pBW3, the cos421 plas-mid). The plasmids were made linear by cleavage with a re-striction enzyme before use as a terminase substrate. The re-sults of a cos cleavage experiment are shown in Fig. 4.

Cleavage of the cosA plasmid was readily observed by ethid-ium bromide staining, whereas no cleavage (<1%) ofthe cos421plasmid was detectable by ethidium bromide staining (Fig. 4,lanes A and C, respectively). To quantitate the amount ofcleav-age, the DNA in the gel was transferred to nitrocellulose filtersand hybridized to 32P-labeled plasmid DNA according to themethod of Southern (23). The amount of hybridization was de-termined by autoradiography and densitometer tracings of thefilms. For the cosA plasmid, 16% of the substrate moleculeswere cleaved (Fig. 4, lane B). For the cos421 plasmid, pro-

oh'

\cI

2 x 106I

'V

LU

1.5 x 106

1 X 106

5 x 10I

0 0.1 0.2 0.3 0.4 0.5

DNA, pg

FIG. 3. (Upper) Composite plasmids for in vitro packaging. Theseplasmids are derived from the cos plasmids shown in Fig. 1. Each con-tains a AcI857 (pGM2 and pBW5) or a A-21 hybrid 19 cI857 (pBW6)chromosome inserted at cos (pGM2 and pBW6 have pDS15; pBW5 haspBW3). Symbols are as in Fig. 1. Because of the A-to-R polarity ofpackaging of phage A chromosomes, the cos site of initiation of pack-agingon these plasmids canbe defined (arrow). Initial cos sites are cosAfor pGM2 and pBW5, cos421 for pBW6. Terminal cos sites are cosA forpGM2 and pBW6 and cos4b21 for pBW5. (Lower) In vitro packaging ofADNA from composite plasmid DNAs. The total yield of phage deter-mined on R594 and produced by a given amount of DNA is plotted.0, pGM2; o, pBW5; A, pBW6.

Proc. Natl. Acad. Sci. USA 79 (1982)

Proc. NatL Acad. Sci. USA 79 (1982) 3501

A B C D E

FIG. 4. cos cleavage reaction with linear (BamEH cleaved) cos plas-mid DNA as substrate. Lanes: A and C, ethidium bromide staining forreactions with the cosA plasmid (pDS15) and cos42l plasmid (pBW3)substrates, respectively [the position of the cleavage products ofpDS15(7.8 and 3.4 kbp) are noted at lane A (Pm)]; B, D, and E, hybridizationsof the DNA in the gel to 32P-labeled probes after Southern transfer tonitrocellulose filters. Lane B is pDS15 (lane A) hybridized to a 32P-la-beled pDS15 probe, autoradiographed for 1 hr. Lane D is pBW3 (laneC) hybridized to 32P-labeled pBW3 DNA, autoradiographed 1 hr. LaneE is pBW3 (lane C) with autoradiography for 48 hr. Lane E shows thata small amount of the 3.0-kbp fragment of pBW3 (4) is generated bycos cleavage (the 7.8-kbp fragment peak is obscured by the substratepeak).

longed exposure revealed a small amount ofcleavage, estimatedto be 0.14% (Fig. 4, lane E). The concentration of A terminasewas limiting in this reaction. Cleavage of the cosA plasmid wassharply reduced by a 1:2 dilution ofterminase and undetectablein a reaction with 1:4 diluted terminase, using ethidium bro-mide staining as an assay. Thus, cleavage of cos421 was lessefficient by a factor of 100 than was cleavage of cosA. Thesame results were obtained in experiments with the completepackaging system with proheads present (not shown).The Defect Is in Phage A Terminase Binding to cos#2l. The

defect in the A terminase-cos42l interactions could be due toan inability of A terminase to bind or nick cos421 (or both). Weasked about a cos binding defect with competition experiments,which are possible because the cos cleavage reaction is saturatedwith cosA plasmid DNA under our conditions (unpublisheddata). Cleavage of the cosA plasmid, pMF517, was followed inreactions with no competitor or in the presence of an excess ofa competitor plasmid carrying cosA (pMF1), cos421 (pBW3),or no cos site (pMF2).The pMF517 plasmid was cut with Pst I prior to the cos cleav-

age reaction; further cutting at cos generates fragments of ca.2 kbp and 4 kbp. The competitor plasmids were sufficientlylarge to allow separation of the pMF517 cleavage fragments bygel electrophoresis. Competitor plasmids were added in 10-foldweight excess, giving a 5-fold excess of competitor cos sites topMF517 cos sites. In Fig. 5, lane B shows the cleavage ofpMF517 in the absence ofcompetitor (34% ofthe cos sites werecleaved, as estimated by the ethidium bromide staining); laneC shows the result ofa reaction in the presence ofthe cosA com-petitor (cleavage ofpMF517 was less than 3%; this strong com-

A B C D E

FIG. 5. Phage A terminase is unable to bind cos421. Lanes A, un-reacted, Pst I-cleaved pMF517 DNA substrate; B, a cos cleavage re-action with the DNA (0.75 ,ag) as substrate, showing the ca. 2- and 4-kbp reaction product fragments (P.) and a remaining band of the linearsubstrate; C, D, and E, reactions in which 7.5 ,ug of competitor plasmidDNA was also present. The competitors contained cosA (pMFl, lane C),cos421 (pBW3, lane D), or no cos site (pMF2, lane E). The competitorplasmids are about twice as large as pMF517, so a 5-fold excess wasadded in terms of cos sites. The competitor plasmids were predomi-nantly in supercoiled form when added to the reaction, but consider-able nuclease activity converts much of the competitor DNA to otherforms during the reaction. The remainingpMFl andpBW3 supercoiledcompetitor DNA migrates with the linear pMF517 substrate mole-cules. Estimates of extent of cleavage were made by densitometry ofthe staining of the reaction products and by assuming that recoveryof DNA from the reactions was the same as for the reaction withoutcompetitor (lane B). The pMF2 plasmid was a dimer and, hence, mi-grated slower than linear pMF517. The band below linear pMF517 inlane E is a product of pMF2 that migrates with the rate of supercoiledmonomers and appears in reactions without A terminase.

petition is due to the presence of cosA); lane E shows a reactionwith pMF2 (derived from pMF1 and lacking cosA), and no in-hibition ofcleavage was seen (76% cleavage estimated); and laneD shows a reaction with pBW3 (the cos42l plasmid as com-petitor), and no competition was detected (41% cleavage). Thevariation in estimation of the amount of cleavage for the un-competed control and for the pMF2 and pBW3 competitionsis largely due to variable recovery from the phenol extraction.In a competition in which recovery could be followed (using[32P]pMF517), the corrected values for cleavages were: uncom-peted, 15%; pSF1, <3%; pBW3, 15%; and pMF2, 17.5%. Con-trols showed the dependence of the pMF517 cleavage on ter-minase and that pMF1 was cleaved when used as a competitor.The possibility that terminase binds poorly to cosA and

cos421 and strongly to the cohesive ends generated by cleavageofcosA is eliminated by work on a mutation ofcos that abolishesnicking but not binding (unpublished data).

In sum, cos421 is unable to compete with cosA for A ter-minase in a cos cleavage reaction. Phage A terminase is unableto bind cos421 normally.

Biochemistry: Feiss and Widner

3502 Biochemistry: Feiss and Widner

DISCUSSIONWe have shown that the in vitro packaging system has the samespecificity as found in vivo for bacteriophages A and 21 cos sites:specificity at an initial cos site, lack of specificity at a terminalcos site. Cleavage of cos421 by phage A terminase proceeds toan extent ca. 1% of the cleavage of cosA. Competition experi-ments show A terminase binding to cos421 is defective. Theresults have implications for several aspects ofA DNA packaging.

Terminal Cos Site Cleavage. If cos421 cannot bind phage Aterminase, then how is cos421 cleaved by A terminase whenit is in the terminal position? We propose that terminase isbrought into contact with the terminal cos site by packaging.In the situation in which the first cos site has A specificity andthe terminal cos site is cos421, the only available A terminasebinding site is at the initial cos. Terminase that binds the initialcos site must cleave the terminal cos site and be brought intocontact with the terminal cos site by packaging. One possibilityfor how terminase bound at the initial cos also would cleave theterminal cos site is that a multimeric form ofterminase may bindthe initial cos site with one active site and scan for the terminalcos site with a separate active site as packaging proceeds. Fig.6 shows this model for terminal cos site scanning. Terminasedoes not require, at the terminal cos site, the base pairs thatare needed for terminase binding at the initial cos site. In thenormal case, where both cos sites are of phage A specificity,what happens when a terminal cos site already has a terminasemolecule bound to it? Packaging of the first chromosome mightpause until packaging of the next chromosome results in cos

cleavage, or subunit dissociation might occur to result in theproper multimeric form of terminase, or the second terminasemight simply be displaced.

Each terminase molecule that binds an initial cos site mustbe able to cleave the terminal cos site. Feiss et al (11) con-

structed lysogens of tandem prophages in which the initial cos

site was cosA and the terminal cos site was cosA or cos421. Witha heteroimmune phage A helper, the yields of packaged pro-

phages were the same regardless of the specificity of the ter-minal cos site. Thus, chromosomes are packaged with equalefficiency whether or not the terminal cos site can bind ter-minase. An advantage to the virus for this mechanism is that theterminase bound to the initial cos site is sufficient to packagethe chromosome; the situation of failing to cut the terminal cos

FIG. 6. Model for terminal cos site scanning. Terminase is posi-tioned so that the DNA being packaged can be scanned until the ter-minal cos site is recognized and cut. Placement of terminase at theconnector is speculative.

site because a required terminase molecule has failed to binddoes not arise.

coo Site Structure. The cos421 site is defective in phage Aterminase binding, but is not defective as a nicking substrate(as a terminal cos site) for A terminase. The cohesive end sym-metry region cannot be the major determinant for terminasebinding because phages A and 21 are identical for this sequence.Where might sequences for terminase binding lie? Hohn (4)argued that there might be a terminase binding site at the leftchromosome end near to but separate from the cohesive endsymmetry segment. She proposed this because (i) linear matureDNA can be packaged in vitro even though it does not have acomplete cos segment, (ii) packaging is polarized in an A-to-Rdirection, and (iii) A and 21 have identical cohesive ends anddiffer in packaging specificity. Our results fit her proposal ofseparate binding and nicking sites within cos. Another cos vari-ant also supports this model for cos. In work to be presentedelsewhere, a mutant cos site deleted for the cohesive end se-quence has been found to be defective for nicking but able tobind terminase. cos contains distinct nicking and binding sitesfor terminase.

We thank Barbara Hohn and Don Court for their interest and sug-gestions. We appreciate the help of Marcia Reeve with the manuscript.This work was supported by National Institutes of Health Grant AI-12851.

1. Szybalski, E. H. & Szybalski, W. (1979) Gene 7, 217-270.2. Wu, R. & Taylor, E. (1971) J. Mol. Biol 57, 441-511.3. Skalka, A. M. (1977) Curr. Top. Microbiol Immunol 78, 201-237.4. Hohn, B. (1975) J . Mol Biol 98, 93-106.5. Becker, A., Marko, M. & Gold, M. (1977) Virology 78, 291-305.6. Weisberg, R., Sternberg, N. & Gallay, E. (1979) Virology 95,

99-106.7. Emmons, S. W. (1974)J. Mol Biol 83, 511-525.8. Feiss, M. & Campbell, A. (1974) J. Mol. Biol 83, 527-540.9. Kaiser, A. D., Syvanen, M. & Masuda, T. (1975)J. Mol Biol 91,

175-186.10. Weigel, P. H., Englund, P. T., Murray, K. & Old, R. W. (1973)

Proc. Natl Acad. Sci. USA 70, 1151-1155.11. Feiss, M., Fisher, R. A., Siegele, D. A., Nichols, B. P. & Don-

elson, J. E. (1979) Virology 92, 56-67.12. Campbell, A. M. (1961) Virology 14, 22-32.13. Feiss, M., Fisher, R. A., Crayton, M. A. & Egner, C. (1977) Vi-

rology 77, 281-293.14. Sternberg, N., Tiemeier, D. & Enquist, L. (1977) Gene 1,

255-280.15. Becker, A., Murialdo, H. & Gold, M. (1977) Virology 78,

277-290.16. Sussman, R. & Jacob, F. (1962) C. R. Hebd. Seances Acad. Sci.

254, 1517-1519.17. Miller, G. & Feiss, M. (1981) Virology 109, 379-390.18. Feiss, M. & Bublitz, A. (1975)J. Mol Biol 94, 583-594.19. Blattner, F. R., Williams, B. G., Blechl, A. D., Denniston-

Thompson, K., Faber, H. E., Furlong, L.-A., Gurnwald, D. J.,Kiefer, D. O., Moore, D. D., Schumm, J. W., Sheldon, E. L.& Smithies, O. (1977) Science 196, 161-169.

20. Kaiser, A. D. & Hogness, D. S. (1960)J. Mol. Biol 2, 392-415.21. Clewell, D. B. (1972)J. Bacteriol 110, 667-676.22. Becker, A. & Gold, M. (1978) Proc. Nati Acad. Sci. USA 75,

4199-4203.23. Southern, E. (1979) Methods Enzymol. 68, 152-176.24. Blattner, F. R., Blechl, A. E., Denniston-Thompson, K., Faber,

H. E., Richards, J. E., Slighton, J. L., Tucker, P. W. & Smith-ies, 0. (1978) Sciences 202, 1279-1284.

25. Hohn, T. & Katsura, I. (1977) Curr. Top. Microbiol Immunol 78,69-110.

Proc. Natl. Acad. Sci. USA 79 (1982)