Proc. Nati. Acad. Sci. USAVol. 84, pp. 4171-4175, June 1987Genetics

Bacteriophage A cloning system for the construction of directionalcDNA libraries

(human lymphocyte cDNA library/bifunctional oligonucleotide linker/y interferon/IacZ insertions)

P. SCOTT MEISSNER, WILLIAM P. SISK, AND MICHAEL L. BERMAN*Molecular Biology and Recombinant DNA Laboratory, Bionetics Research, 1330-A Piccard Drive, Rockville, MD 20850

Communicated by Jonathan Beckwith, February 24, 1987 (received for review October 28, 1986)

ABSTRACT We have developed a bacteriophage A cloningvector, XORF8, that can be used for the construction of cDNAlibraries. The wild-type A genome contains five BamHI, fiveEcoRI, and seven Hindu restriction sites that have all beenremoved from the genome of AORF8. Sites for these endonu-cleases are present within the multiple cloning site of AORF8.We report a method for preparing cDNAs that can be clonedin a single orientation in our phage vector. The method utilizesthe synthesis of double-stranded cDNA, including priming offirst-strand synthesis by oligo(dT). After completion of second-strand synthesis, a bifunctional oligodeoxynucleotide linker isligated to the cDNA fragments. This linker, which contains aBamHI restriction site, will create a HindM restriction sitewhen ligated to the 3' end of cDNA fragments. Subsequenttreatment of methylated cDNA with Hindm and BamHIendonucleases allows these fragments to be cloned directionallyinto AORF8. To demonstrate the utility of this cloning system,we prepared a library from 5 ,ug of mRNA isolated fromphytohemagglutinin-stimulated human peripheral blood lym-phocytes. The primary library contained 2 X 108 plaque-forming phage, at least 80% of which contain inserts. A portionof the library was examined for the presence of y-interferon-related clones to verify the method had generated a library thatwas representative of phytohemagglutinin-stimulated periph-eral blood lymphocytes. This simple and efficient cDNA cloningsystem significantly reduces the amount of RNA and effortrequired for the preparation of large directionally clonedlibraries.

Biological research has benefited enormously from the studyof genes isolated by cDNA cloning techniques. Since theirintroduction (1, 2) these techniques have been refined toallow for the isolation of eukaryotic genes expressed at lowfrequencies. Oligonucleotide probes or antibodies to eukary-otic gene products are widely used for screening bacteriallibraries (3-7). Despite advances, the construction of cDNAlibraries still remains an arduous task, particularly if onewishes to construct a directional cDNA library. In directionalcloning the vector and fragments to be cloned share anoriented pair of specific endonuclease restriction sites at theirtermini. Such an approach offers efficient cDNA capture andpredictable insert orientation. Methods for constructing di-rectional libraries in plasmid vectors exist (3-5). However,bacteriophage X vectors can greatly simplify the tasks ofconstructing and screening large cDNA libraries (6, 7). Wereport the use of a phage vector for the construction ofdirectional libraries that retains many of the useful features ofexisting bacterial expression vectors.A directional library constructed in phage X offers the

following advantages relative to the nondirectional librariesconstructed with existing phage vectors: (i) The library would

contain a small number of nonrecombinant phage, and few,if any, phage would contain cDNA fragments inserted in theincorrect orientation. (ii) The library would be more efficientin the capture of cDNA fragments due to the high vector/insert molar ratio. (iii) After cleaving the vector withendonucleases at the cloning site, subsequent treatment withphosphatase to ensure high insertion rates would be unnec-essary. (iv) DNA inserts can be cloned in a known orientationrelative to vector-encoded expression signals.The cloning vector XORF8 possesses a multiple cloning site

(MCS) containing single recognition sequences for the re-striction enzymes BamHI, EcoRI, and HindIII (P.S.M., J. V.Knowlton, and M.L.B., unpublished data). To use this vectorfor directional cloning, a method for preparing cDNA frag-ments with appropriate cohesive termini was needed. Themethods available were either not applicable to phage (3) orinvolved steps such as treatment with S1 nuclease that tendedto produce truncated cDNAs (5, 8). To overcome the limi-tations ofthe existing methods, we developed a procedure forthe preparation of cDNA fragments that can be cloneddirectionally into our phage vector. The procedure is asfollows: (i) Double-stranded (ds) cDNA is prepared usingoligo(dT) as the primer for first-strand synthesis (9). (it) dscDNA is treated with specific methylases to protect internalBamHI and HindIII sites. (iii) The cDNA is ligated to abifunctional oligodeoxynucleotide linker that contains aBamHI site and will form an additional HindIII site whenligated to the 3' end of ds cDNA. (iv) Linkers are thendigested with HindIII and BamHI, and the cDNA is ligatedinto XORF8 that has been cleaved with HindIII and BamHI.To evaluate the directional vector and cloning technique,

we prepared a library from phytohemagglutinin (PHA)-stimulated human peripheral blood lymphocytes (PBLs).Phage from the library were analyzed for the presence ofcDNA inserts and for the presence of the expected BamHIand HindIII sites created by the bifunctional linker. Inaddition, to verify that our system was capable of efficientlycapturing cDNAs, we screened a portion of the library for-interferon (IFN-y), a gene expressed at low levels in

PHA-stimulated peripheral blood lymphocytes.

MATERIALS AND METHODSMaterials. All bacterial strains are Escherichia coli K12

derivatives. Bacterial strain MC1061, F- hsdR mcrB ara-D139 A(araABC-leu)7679 galU galK A(lac)X74 rpsL thi; orMBM7014.5, F- hsdR2 mcrBl zjj202::TnJO araD139araCU25am A(argF-lac)U169 trpam maliam supFrpsL retAthi, were used as hosts for the primary cDNA libraries. ThehsdR2 and mcrBl alleles in MBM7014.5 were introduced byPlvir cotransduction with the zj202::TnJO insertion fromstrain ER1351, a generous gift of Elisabeth A. Raleigh (New

Abbreviations: PBL, peripheral blood lymphocyte; IFN-y, y inter-feron; ds, double stranded; PHA, phytohemagglutinin.*To whom reprint requests should be addressed.

4171

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

Proc. Natl. Acad. Sci. USA 84 (1987)

England Biolabs). Culture conditions and standard mediumwere as described (10). TBA medium was prepared byautoclaving 10 g oftryptone/5 g ofNaCl/4 g of agarose/1 literof water. DNase and RNase were from Sigma; all otherenzymes were from New England Biolabs.RNA Isolation. Ficoll/Hypaque-separated normal human

peripheral blood mononuclear leukocytes were incubated ata density of 106 cells per ml in serum-free RPMI 1640 mediumcontaining PHA at 5 ,ug/ml for 24 hr at 370C in a CO2humidified atmosphere (11). Cells were harvested directlyinto lysis buffer [5 M guanidine thiocyanate, 50mM Tris HCl(pH 7.6), 8% (vol/vol) 2-mercaptoethanol] and disrupted for1 min with a Polytron tissue homogenizer. RNA was isolated,and poly(A) mRNA was prepared by oligo(dT) chromatog-raphy as described (12).

Synthesis of cDNA. Beginning with 5 ug of mRNA, cDNAwas prepared by the method of Gubler and Hoffman (9) usinga kit supplied by Amersham. All reactions were carried outas described by the supplier. The yields from first- andsecond-strand synthesis were determined by including 5 ,uCiof [1',2',5'-3H]dCTP (67 Ci/mmol; 1 Ci = 37 GBq) and 50 ,uCiof [a-32P]dCTP (3000 Ci/mmol) in the first- and second-strandreactions, respectively. Yields were quantitated by measur-ing the incorporation of radioactivity into trichloroaceticacid-insoluble material. Based on the percentage of incorpo-rated radioactivity, 500 ng of cDNA was made in thefirst-strand reaction and essentially all of this material wasconverted to ds cDNA during second-strand synthesis.Following treatment with T4 DNA polymerase, the reactionwas phenol/chloroform, 1:1 (vol/vol), extracted and ethanolprecipitated. Unincorporated nucleotides were removed bysequential precipitations from 2 M ammonium acetate (3).The cDNA was rinsed with 70% (vol/vol) ethanol, dried, andresuspended in 20 .ul of sterile water. The addition of carriertRNA or glycogen to the cDNA precipitation resulted ininhibition of subsequent ligation reactions. Control reactionscarried out in the presence of these carriers show no suchinhibition. Consequently, we did not add exogenous carrierto the ds cDNA precipitations.

Methylation of cDNA. The cDNA was treated with site-specific methylases to protect internal restriction enzymerecognition sites from cleavage in subsequent digestions (seeResults). The cDNA (500 ng) was treated with BamHI andAlu I methylases essentially according to directions suppliedby the manufacturer (New England Biolabs). The methyla-tion reactions were incubated with 5 units of each enzyme for1 hr at 37°C in 25 mM Tris HCl, pH 7.5/25 mM NaCl/10 mMEDTA/5 mM 2-mercaptoethanol/80 ,uM S-adenosylmethio-nine in a total volume of 50 ,ul. The methylated cDNA wasphenol/chloroform extracted, ethanol precipitated, rinsed,dried, and resuspended in 10 ,ul of water.

Addition of Linker. The linker was prepared on a Systecmodel 1450A automated DNA synthesizer, and 100 pmol (500ng) was labeled with 50 ,Ci of [-32P]dATP (3000 Ci/mmol)using 10 units of polynucleotide kinase in 50 Al of 50 mMTris HC1, pH 7.5/10mM MgCl2/1 mM dithiothreitol. After 30min at 37°C, unlabeled ATP (final concentration, 0.5 mM) and10 additional units of polynucleotide kinase were added, andthe reaction was continued for an additional 30 min at 37°C.The reaction was terminated by heating for 5 min at 65°C, andthe linkers were stored at -20°C. The presence of 32p in theoligonucleotide linker made it possible to follow subsequentreactions with the cDNA by polyacrylamide gel electropho-resis and autoradiography (12).

Linkers (50 ng) were ligated to cDNA (500 ng) using 400units of T4 DNA ligase in 50 mM Tris HCl, pH 7.5/10 mMMgCl2/5 mM dithiothreitol/0.2 mM ATP. The ligation wasincubated at room temperature for 2 hr in 20 ul and wasterminated by heating to 650C for 10 min.

Endonuclease treatment of the linker reaction was accom-plished by adding 5/l of 10x HindIII buffer (0.5 M NaCl/0.5M Tris-HCl, pH 8.0/100 mM MgCl2) and 20 units of HindIllenzyme in 50/l. After 90 min at 370C, 0.5 ,ul of 5M NaCl and20 units of BamHI enzyme were added. Incubation wascontinued for 60 min at 370C, and the reaction was terminatedby heating to 650C for 10 min.

Unligated linkers and small cDNA fragments were re-moved by passing the reaction over a Sepharose-CL4B(Pharmacia) column (prepared in a 1-ml disposable plasticpipette), equilibrated in running buffer (1x running buffer =0.3 M NaCl/10 mM Tris-HCl, pH 7.5/1 mM EDTA). TheHindIII and BamHI reaction mixture (50,ul) was adjusted bythe addition of 5 Al of 10x running buffer and 5 Al of loadingsolution [95% (vol/vol) glycerol/0.1% bromophenol blue]before loading on the column. The fractions containing thecDNA fragments were pooled, ethanol precipitated, dried,and resuspended in 10 /L4 of water. Analysis of the cDNAfraction by polyacrylamide gel electrophoresis and autoradi-ography showed that this method eliminated digested linkersand small DNA fragments (12).

Preparation of XORF8. Lysates of XORF8 were preparedusing standard techniques (10). Phage were banded in cesiumchloride, and DNA was extracted with formamide (13).Twenty micrograms of phage DNA were treated with 100units of HindIII for 16 hr at 37°C in 100 /l. After phenol/chloroform extraction and ethanol precipitation, the DNAwas digested with 100 units ofBamHI for 8 hr at 37°C in 100/al. The buffers used were those recommended by thesupplier (New England Biolabs). The DNA was extractedand precipitated as before and resuspended in sterile water ata concentration of 200 ,ug/ml.

Ligation of Vector to cDNA. A typical ligation reaction ofthe cDNA (=50 ng) to BamHI-HindIII cleaved XORF8 (500ng) included T4 DNA ligase (400 units) in 10 /l4 of ligationbuffer (50 mM Tris HCl, pH 7.5/10 mM MgCl2/5 mMdithiothreitol/l mM ATP). After a 2-hr incubation at roomtemperature, the ligation reaction was stored at 4°C over-night, without treatment at 65°C. One microliter of theligation reaction was removed and packaged according todirections provided by the supplier of the phage X packagingextracts (Stratagene Cloning Systems, San Diego, CA). As acontrol to determine the background level of nonrecombi-nants, identical ligation and packaging reactions were per-formed in the absence of cDNA.

Analysis of the Library. To analyze recombinants for thepresence of cDNA inserts, high-titer lysates (=1010 plaque-forming units/ml) were prepared from single plaques (10).Lawns ofMBM7014 were prepared by combining 0.5 ml of afresh saturated culture of cells with 5 ml of liquified TBAmedia. The mixture was poured over a plate containing TBAmedia and allowed to solidify. Approximately 108 plaque-forming units were spread over the plate using 1 ml of 10 mMMgSO4 as diluent. The plate lysates were incubated at 37°Cfor =8 hr, or 30°C for =14 hr and then cooled to 4°C. Thetop-agarose layer was scraped into a 15-ml polypropylenetube (Falcon 2059) containing 50 ul of chloroform, Vortexmixed, and then centrifuged at 3000 x g for 15 min. Theclarified aqueous phase (=0.5 ml) was removed, transferredto a 1.5-ml Microfuge tube, and incubated at room temper-ature for 30 min with 5 ,ul of DNase I (1 mg/ml of 0.3 Msodium acetate, pH 5) and 5 ul of RNase A (1 mg/ml ofwater). Phage DNA was extracted with Tris-saturated phenol(12) followed by two additional phenol/chloroform extrac-tions and precipitated from 2 M ammonium acetate. Asingle-plate lysate typically yielded several micrograms ofDNA. For sequencing directly from X DNA (14), the mini-prep DNA was passed over a Sepharose-CL4B column toremove any remaining RNA.

4172 Genetics: Meissner et al.

Proc. Natl. Acad. Sci. USA 84 (1987) 4173

Screening for IFN-y Recombinants. Approximately 105recombinants were plated on an mcrB host and transferred tonitrocellulose filters as described (10). Prehybridization,hybridization, and nick-translation were performed usingstandard procedures (10, 12). Filters were probed for IFN-yusing a nick-translated 1.2-kilobase (kb) DNA fragmentencoding the entire IFN-y structural gene (a generous gift ofR. Dijkema, Organon International, Oss, The Netherlands).Following hybridization, filters were washed twice at roomtemperature for 15 min in 2x SSC/0.1% NaDodSO4 ( x SSC= 0.15 M NaCl/0.015 M sodium citrate, pH 7.0), and twiceat 55°C for 15 min in 0.1 x SSC/0.1% NaDodSO4. Filters wereair dried and exposed for 12 hr at -70°C on Kodak RP5 x-rayfilm using a Dupont Cronex intensifying screen.

Subcloning and Sequence Analysis. Phage clone X-yIFN9was digested with BamHI and HindIII, the 1200-base-paircDNA insert was gel purified and subsequently ligated intoBamHI/HindIII cleaved M13mpl8 and M13mpl9. Single-stranded template DNA was prepared (12) and sequenced bythe dideoxy-chain-terminating method (15). The 5'- and3'-insert junctions were determined using the M13mpl8vector and M13mpl9 vector, respectively. The sequencingreaction mixtures were resolved on a 10%o polyacrylamide/8M urea gel. The sequencing primer (NEB 1200) and sequenc-ing reagents were from New England Biolabs.

RESULTSWe have developed a bacteriophage X cloning vector,XORF8, that can be used for the construction of cDNAexpression libraries (Fig. 1). Cloning in the vector can beachieved with DNA fragments containing EcoRI-BamHI,EcoRI-HindIII, or BamHI-HindIII cohesive termini.The multiple cloning site of XORF8 used for insertion of

foreign DNA is located within the lacZ gene. Due to the

AB

J att int N cI OP Q SRB B B B

* II II I I

EH HEH E HH E H E

imm 21intam nin5,'HBE'_

,-' MCS lacPO

f lacY lacZ lacI bla

4 ,H_go--

HindIII-BamHI-E~coRI

FIG. 1. Diagram of bacteriophage cloning vector XORF8. (Top)Genetic and physical maps of wild-type phage X showing thepositions of the naturally occurring BamHI (B), HindIII (H), andEcoRI (E) restriction enzyme sites. Uppercase letters above the maprefer to the position of some of the phage genes. The positions of theattachment site (att) and the integrase gene (int) are indicated.(Middle) Genetic and physical map ofXORF8. The solid bar refers towild-type X sequences, spaces in the line represent deletions of thewild-type genome. The open box denotes the cloning region; the solidbox refers to imm2l control region; the nin5 deletion is shown abovethe diagram; intam represents an amber mutation at codon 300 of theintegrase gene. The size of this vector is 43 kb (88% of the wild-typeX), and the cloning capacity is -6 kb. (Bottom) Expanded view ofcloning region ofXORF8. The solid, vertical bar refers to the multiplecloning site (MCS); lacPO indicates the lac promoter-operatorregion; lacI, lacZ, and lacY are the intact structural genes of the lacoperon; bla refers to the ,B-lactamase gene from pBR322. The arrowsbelow the diagram indicate the directions of transcription. The originof the MCS is from pUC8/Ml3mp8 (16) by homologous recombina-tion.

synthesis of 8-galactosidase from lacZ, XORF8 producesblue plaques on a A(lacZ) host cultured with the chromogenicindicator 5-bromo-4-chloro-3-indolyl 8-D-galactoside. Con-sequently, cDNA fragments inserted into the multiple cloningsite produce recombinant phage (colorless plaques) that caneasily be distinguished from nonrecombinants (blue plaques).The results of vector DNA ligated in the absence or presenceof cDNA are presented in Table 1.Our method for cDNA synthesis requires the protection of

internal HindIII and BamHI sites prior to the addition of theDNA linkers. Because HindIII methylase is not commercial-ly available, we used Alu I methylase to protect internalHindIII sites. Treatment of HindIII recognition sequenceswith Alu I methylase is known to completely protect thesesequences from cleavage by HindIII (17). However, DNAtreated with Alu I methylase is subject to a restriction systemencoded by the mcrB locus ofE. coli (18). To prevent possiblehost restriction of Alu I-methylated DNA, we used E. coliwith an mcrB mutation for our primary library. However,when we ligated unmethylated XORF8 DNA to methylatedcDNA, the resulting phage displayed no differences in theirplating efficiencies on mcrB+ versus mcrB- hosts.As seen in Fig. 2, we developed a method that generates

cDNA that can be cloned in a specific orientation into theBamHI-HindIII sites of XORF8. A library was constructedfrom 5 ,ug ofmRNA isolated from PHA-stimulated PBLs. Toexamine the library for the presence of cDNA inserts, 10colorless plaques were chosen at random. The phage DNAwas codigested with BamHI and HindIll, and the enzymereaction mixtures were resolved on a 1% agarose gel. Theanalysis revealed that 8 ofthe isolates contained inserts in thesize range of 0.5-5 kb, whereas inserts in the remaining twowere not detected (data not shown).To confirm that the cloning system had generated a library

that was truly representative of PHA-stimulated PBLs, weexamined i05 plaques for the presence of clones containingIFN-y sequences. Six clones hybridized to the nick-translat-ed probe, and DNA prepared from three of the clones wasdigested with BamHI and HindIII to examine the size of theinserts. The expected size of a full-length IFN-y cDNA isabout 1.2 kb (19). One of the clones (X-yIFN9) had aBamHI-HindIII insert of 1.2 kb, while the other two hadsmaller BamHI-HindIII inserts.DNA sequence analysis of the 5' vector-cDNA junction of

the two smaller clones was obtained by sequencing directlyfrom phage DNA (14). The data showed that these partialcDNA clones initiated within the X-IFN gene at nucleotides493 and 444, respectively (data not shown). To determine thesequence of the 5'- and 3'-vector-cDNA junctions of thefull-length clone X-yIFN9, the 1.2-kb BamHI-HindIII insertwas subcloned into M13mpl8 and M13mpl9. Analysis of thesequence showed that the expected linker sequences were

Table 1. LacZ phenotype of XORF8 plaques from phage armsligated in the presence and absence of cDNA

pfu/,ug of vector DNADNA Ligase Exp. LacZI LacZ-

XORF8 - 1 103 (>99) <1 (<1)2 5 x 102 (>99) <1 (<1)

XORF8 + 1 107 (>99) <1(<1)2 3.3 x 105 (>99) <1 (<1)

XORF8/cDNA + 1 8 x 106 (25) 3.2 x 10' (75)2 1.5 x 104 (60) 10 (40)

No LacZ- plaques were observed in the absence of added cDNA.The library screened for IFN-y is from experiment 1. Experiment 2,using another source of cDNA, is shown for comparison. Values inparentheses are % of total plaques. pfu, plaque-forming units.

Genetics: Meissner et al.

Proc. Natl. Acad. Sci. USA 84 (1987)

PNNN /7 AAAAAAOHHoNNN / TTTTTTP

BamHII '

PGCTTGGATCCAAGCOHHOCGAACCTAGGTTCGP

BamHIm I

BamHI BamHII

GCTTGGATCCAAGC GCTTGGATCCAAGC GCTTGGATCCAAGC NNNCGAACCTAGGTTCG CGAACCTAGGTTCG CGAACCTAGGTTCG NNN

HindIII BamHI(I

BamHIm

BamHIm I

-----~AAAAAA GCTTGGATCCAAGC GCTTGGATCCAAGC GCTTGGATCCAAGC-----TTTTTT CGAACCTAGGTTCG CGAACCTAGGTTCG CGAACCTAGGTTCG

'BamHI HindII'

PGATCCAAGCNNN /- AAAAAOHHoGTTCGNNN 1/- TTTTTTCGAP

FIG. 2. Simplified method for pre-paring cDNA fragments with distinctcohesive termini. (A) Blunt-endedcDNA prepared using oligo(dT) as aprimer for first-strand synthesis. P,the 5' phosphate group; OH, the 3'hydroxyl group. (B) Sequence of the14-mer bifunctional linker; BamHI in-dicates the internal endonuclease rec-ognition site. (C) Ligation of the link-er in B to the cDNA results in theformation ofconcatamers oflinkers atthe 5' and 3' ends of the cDNA. Notethat a single HindIII site is created atthe 3' end of the cDNA. (D) Digestionof the cDNA shown in C with HindIIIandBamHI results in the formation ofcDNA fragments containing cohesivetermini (BamHI and HindIII) that canbe ligated into BamHI/HindIIIcleaved vector.

present, and the cDNA insert began at nucleotide 2 of thepublished sequence of IFN-y (Fig. 3).

DISCUSSIONBacteriophage X cloning vectors have properties that makethem excellent tools for the construction of cDNA libraries(6). However, one limitation of existing phage vectors is thatcDNA fragments are inserted randomly. In addition, thebackground level of nonrecombinant phage in such librariescan be high in proportion to total phage (7). These limitationscan substantially increase the number of phage that must beanalyzed to detect a gene of interest, and they can be aparticular problem in vectors designed for expressing cDNAinserts in E. coli. However, these problems can be avoidedby using XORF8 because it can accept cDNA fragments in a5'-3' orientation. We also designed a bifunctional linker tofacilitate the preparation of cDNA fragments that can beligated directionally into the vector. These developmentsincrease the ease and efficiency of constructing large direc-tional cDNA libraries.Phage vector designs that include cloning sites within

functional genes, e.g., lacZ, provide convenient phenotypesfor determining the efficiency of the system (6, 7). The lacZinsertion vector Xgt11 uses the naturally occurring EcoRI siteat codon 1006 of lacZ. Insertions (or deletions) that alter thisregion of lacZ invariably result in the synthesis of inactive13-galactosidase monomer. The various cloning sites in thevector XORF8 are located at codons 5-10 of lacZ. LigationofcDNA fragments into these sites results in the alteration ofthe lacZ phenotypes of recombinant phage plaques. Incontrast to Xgtll, there is little specificity for particularamino acid codons in the vicinity of the cloning sites withinXORF8. Studies of N-terminal ,3-galactosidase hybrids haveshown that the first critical amino acids occur at positions23-25 in f-galactosidase (20). Therefore, any nonpolar in-sertion in XORF8 that places a ribosome binding site and aninitiation codon in-frame with lacZ will encode an active,-galactosidase. In practice, a high percentage of LacZ+insertions at the 5' end of lacZ may be of this type (21-23).

Therefore, we suggest that a significant proportion of theLacZ+ XORF8 phage may carry inserts. This idea is support-ed by the observation that one of the three IFN-y clonesforms LacZ+ plaques on indicator plates.The LacZ phenotype is also useful in assessing the quality

of the vectorDNA preparation. Theoretically, if the vector iscleaved to completion with two enzymes that do not generatehomologous single-stranded regions, there should be noviable phage produced by a ligation in the absence of addedcDNA. In fact, a certain proportion of phage are cleaved byonly one of the two endonucleases, hence producing LacZ+nonrecombinant phage when ligated. However, these phagecontribute a relatively insignificant background of non-recombinants to the library (Table 1).The use of Alu methylase to protect HindIII sites internal

to the cDNA required plating the primary library on an mcrBhost. Although we could demonstrate a thousand-fold re-striction ofAlu methylated wild-type X (7 HindIII and 143 AluI sites) on an mcrB+ compared to an mcrB- host, we saw nosignificant difference in the efficiency of plating on these twohosts of our primary methylated cDNA library. This may bethe result of the frequency of Alu I sites or the size of thecDNA inserts. However, we have no adequate explanationfor this observation. Finally, a number of unrelated cDNAclones that have internal HindIII sites have been isolatedfrom our library. This demonstrates that the Alu methylase isprotecting the cDNA as expected.The design of the cDNA linker takes advantage of the

oligo(dT) primer at the 3' end ofcDNA to create aHindIII site(Fig. 2). Two types of cDNAs will not be cloned by thismethod. The first type are those cDNAs from which thepoly(A) tail has been lost during synthesis. The second, andmore frequent, type are cDNAs that have a 5'-terminalsequence ofpTTNNN, where N stands for any nucleotide. AHindIII site will be formed at both the 5' and 3' ends of thesecDNAs when linkers are added. Statistically this classcomprises 6.25% ofthe cDNA products. However, due to thelarge number of cDNA clones usually obtained using ourdirectional method, we do not consider these serious prac-tical problems.

A

B

C

D

4174 Genetics: Meissner et al.

Proc. Natl. Acad. Sci. USA 84 (1987) 4175

A B

A G C T A G C T

AATTCGGT -

CTTCAAC1Tr

AGTGATCAAG

AACATAGTTC -

GACTAGTCCA-

GGTTTCCTGAA

TTGACTAGAAAG

A AGA AA

A

T

GC

C_

A-

T

AG-

A -A-

*T

G-A .

_ A

e C-XITE A -mG -

GG

C

AT

G

G j

synthesis of eukaryotic peptides is controlled by the lacexpression signals. This feature is especially desirable whenattempting to isolate genes using antibody probes (7). Anadditional benefit of the cloning system described here is itsoverall simplicity. Beginning with small amounts of mRNA,we have found that a large directional expression library canbe constructed easily within 2 days. This cDNA cloningsystem represents a significant increase in the ease andefficiency with which large directional libraries can beprepared.

AAATGAGTTA

AA1TTTAAC

AGATATTGAT

CATATATATT

- ACACiTMATT

TGAAC1T

- G-T

T

-TT

T-T

T-T

T-T

-cTT

-A_GA

G

C

FIG. 3. DNA sequence of X-yIFN9 cDNA insert ends. (A)Sequence of the 5' vector-cDNA junction. Vector and BamHI-linkerregion precede the first cloned IFN-y nucleotide (TI'). This nucleo-tide corresponds to nucleotide 2 of the published sequence (19). Thefirst 93 nucleotides of the cDNA insert are labeled. (B) Complemen-tary strand sequence of the 3' vector-cDNA junction. Vector,HindIII cloning site-linker region, and the poly(A) region adjacent tothe terminal nucleotide (To) of the yIFN9 cDNA insert are shown.This clone contains two additional nucleotides (*) not found in thepublished sequence. The terminal 76 nucleotides of the cDNA cloneare labeled.

Another useful feature of XORF8 is the location of thecloning sites within the lacZ gene of the intact lac operon.Consequently, when cDNA-encoded open reading framesegments are inserted in frame with the lacZ sequences,

We thank M. Sveda for preparation ofRNA and many thoughtfuldiscussions; J. Ransom for culture of PBLs; M. Hendricks fortechnical assistance in screening the library; and L. Burdette forcareful preparation of the manuscript.

1. Rougeon, F. & Mach, B. (1976) Proc. Natl. Acad. Sci. USA73, 3418-3422.

2. Efstratiadis, A., Kafatos, F. C., Maxam, A. M. & Maniatis, T.(1976) Cell 7, 279-287.

3. Okayama, H. & Berg, P. (1982) Mol. Cell. Biol. 2, 161-170.4. Alexander, D. C., McKnight, T. D. & Williams, B. G. (1984)

Gene 31, 79-89.5. Helfman, D. M., Feramisco, J. R., Fiddes, J. C., Thomas,

G. P. & Hughes, S. H. (1983) Proc. Natl. Acad. Sci. USA 80,31-35.

6. Huynh, T. V., Young, R. A. & Davis, R. W. (1985) in DNACloning, A Practical Approach, ed. Glover, D. (IRL, Oxford),Vol. I, pp. 49-78.

7. Young, R. A. & Davis, R. W. (1983) Proc. Natl. Acad. Sci.USA 80, 1194-1198.

8. Kraus, J. P., Williamson, C. L., Firgaira, F. A., Yang-Feng,T. L., Munke, M., Francke, U. & Rosen, L. E. (1986) Proc.Natl. Acad. Sci. USA 83, 2047-2051.

9. Gubler, U. & Hoffman, B. J. (1983) Gene 25, 253-269.10. Silhavy, T. J., Berman, M. L. & Enquist, L. W. (1984) Exper-

iments with Gene Fusions (Cold Spring Harbor Laboratory,Cold Spring Harbor, NY).

11. Ransom, J. H., Evans, C. H., McCabe, R. P., Pomato, N.,Heinbaugh, J. A., Chin, M. & Hanna, M. G., Jr. (1985) CancerRes. 45, 851-862.

12. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Labora-tory, Cold Spring Harbor, NY).

13. Davis, R. W., Botstein, D. & Roth, J. R. (1980) AdvancedBacterial Genetics, A Manual for Genetic Engineering (ColdSpring Harbor Laboratory, Cold Spring Harbor, NY).

14. Zagursky, R. J., Baumeister, K., Lomax, N. & Berman, M. L.(1985) Gene Anal. Tech. 2, 89-94.

15. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.Acad. Sci. USA 74, 5463-5467.

16. Vieira, J. & Messing, J. (1982) Gene 19, 259-268.17. McCleland, M. & Nelson, M. (1986) Nucleic Acids Res. 13,

r201-r238.18. Raleigh, E. A. & Wilson, J. (1986) Proc. Natl. Acad. Sci. USA

83, 9070-9074.19. Gray, P. W., Leung, D. W., Pennica, D., Yelverton, E.,

Najarian, R., Simonsen, C. C., Derynck, R., Sherwood, P. J.,Wallace, D. M., Berger, S. L., Levinson, A. D. & Goeddel,D. V. (1982) Nature (London) 295, 503-508.

20. Celada, R. & Zabin, I. (1979) Biochemistry 18, 404-406.21. Gray, M. R., Colot, H. V., Guarente, L. & Rosbash, M. (1982)

Proc. Natl. Acad. Sci. USA 79, 6598-6602.22. Weinstock, G. M., ApRhys, C., Berman, M. L., Hampar, B.,

Jackson, D., Silhavy, T. J., Wiessemann, J. & Zweig, M.(1983) Proc. Natl. Acad. Sci. USA 80, 4432-4436.

23. Koenen, M., Ruther, U. & Muller-Hill, B. (1982) EMBO J. 1,509-512.

Genetics: Meissner et al.