Proc. Natl. Acad. Sci. USAVol. 83, pp. 6465-6469, September 1986Cell Biology

Human erythropoietin gene: High level expression in stablytransfected mammalian cells and chromosome localization

(genomic screening/mammalian cdl expression/chromosome sorting/gene amplification)

JERRY S. POWELL*, KATHLEEN L. BERKNERt, ROGER V. LEBO*, AND JOHN W. ADAMSON**University of Washington, Seattle, WA 98195; tZymogenetics, Inc., 2121 North 35th Street, Seattle, WA 98103; and tHoward Hughes Medical Institute andUniversity of California, San Francisco, CA 94143

Communicated by Eloise R. Giblett, May 15, 1986

ABSTRACT The glycoprotein hormone erythropoietinplays a major role in regulating erythropoiesis and deficienciesof erythropoietin result in anemia. Detailed studies of thehormone and attempts at replacement therapy have beendifficult due to the scarcity of purified material. We used acloned human erythropoietin gene to develop stably transfectedmammalian cell lines that secrete large amounts ofthe hormonewith potent biological activity. These cell lines were producedby cotransfection ofmammalian cells with a plasmid containinga selectable marker and plasmid constructions containing adoned human erythropoietin gene inserted next to a strongpromoter. The protein secreted by these cells stimulated theproliferation and differentiation of erythroid progenitor cellsand, with increased selection, several of these cell lines secreteup to 80 mg of the protein per liter of supernatant. Hybrid-ization analysis ofDNA from human chromosomes isolated byhigh resolution dual laser sorting provides evidence that thegene for human erythropoietin is located on human chromo-some 7.

Normal production of erythrocytes in man requires thesecretion of erythropoietin by the kidney, apparently as themature protein (1). In the steady state, the hormone circulatesin the blood at a concentration of 10-18 milliunits (128-230pg) per ml, and with the stimulus of severe tissue hypoxia thelevels may increase as much as 1000-fold (2). The elevatedhormone levels trigger proliferation and differentiation of apopulation of receptive progenitor cells in the bone marrow,stimulate hemoglobin synthesis in maturing erythroid cells,and accelerate release of erythrocytes from the marrow intocirculation, thus increasing the erythrocyte mass and ame-liorating the hypoxic conditions. Patients with deficiencies ofthis hormone, such as those with chronic renal failure, oftensuffer severe anemia.

Erythropoietin is a glycoprotein of 34-38 kDa with ==40%oof its molecular size provided by carbohydrate. At least onedisulfide bridge is required for activity (3, 4). However, littleis known about the conformation of this hormone, and thedetails of its synthesis are not well understood. The isolationsof genomic and cDNA clones provide opportunities toanalyze the control of erythropoietin production as well as toprovide sufficient quantities ofmaterial both for further studyand for replacement therapy (5-7). Here we report the use ofa genomic clone of human erythropoietin to develop stablytransfected mammalian cell lines that secrete high levels ofactive erythropoietin. Furthermore, using restriction frag-ments of the cloned gene as probes, we have mapped the genefor human erythropoietin to chromosome 7.

MATERIALS AND METHODSIsolation of Genomic Clones. A human genomic library in

bacteriophage X (8) was screened using low stringencyhybridization conditions and mixtures of oligonucleotideprobes (9). Oligonucleotide mixtures were prepared using anApplied Biosystems (Foster City, CA) synthesizer and end-labeled using [32P]ATP and T4 polynucleotide kinase. Thesynthetic oligonucleotides were designed to correspond toportions of the amino-terminal amino acid sequence ofH2N-Ala-Pro-?-Arg-Leu-Ile-Leu-Asp-Ser-Arg-Val-Leu-Glu-Arg-Tyr-Leu-Leu-Glu-Ala-Lys-Glu-Ala-Glu-?-Ile-Thr-Asp-Gly-Gly-Ala obtained by Yanagawa et al. (10) for the humanprotein purified from urine of patients with aplastic anemia.To reduce the degeneracy of the codons for the amino acidsequence of this region the codon usage rules of Grantham etal. (11) and Jaye et al. (12) were employed. These rules takeinto account the relatively rare occurrence of CpG dinu-cleotides in DNA of vertebrates and avoid, where appropri-ate, potential A-G mismatch pairings. At amino acid position24 an asparagine was placed as most likely (10). For the aminoacids Glu-Ala-Lys-Glu-Ala-Glu-Asn, 2 pools of72 sequenceseach were synthesized to correspond to the predictedcodons. Thus, one pool was TT(C/T)TC(A/G/T)GC(C/T)-TC(C/T)TT(A/G/T)GCTTC for the 20-nucleotide probe andthe second pool replaced a T with a C at position 18. For theamino acids Glu-Asn-lle-Thr-Asp-Gly, one pool of sequences[AGC TCC TCC ATC AGT ATT ATT T(C/T)] was con-structed for the 23-nucleotide probe. Plaques that hybridizedto the oligonucleotide probes were rescreened at lowerdensity until pure. [32P]ATP was from ICN; enzymes werefrom New England Biolabs or Bethesda Research Labora-tories. After EcoRI restriction enzyme digestion of thepositive clones, insert DNA was gel-purified and ligated bystandard techniques into plasmid pUC13 that had beendigested with EcoRI. DNA sequence was determined bydideoxynucleotide chain termination (13) using dATP[a-355]and the 17-mer universal primer or, in selected regions,specific oligonucleotide primers.

Construction of Expression Plaids Carrying ErythropoietinGene Sequences. The plasmid expression vector pDll wasderived from a previously described plasmid (14) and containedthe simian virus 40 enhancer sequences and origin ofreplicationas well as the adenovirus type 2 major late promoter andtripartite leader sequences (see Fig. 2). The Apa I fragment ofhuman erythropoietin genomic sequences (Fig. 1) was gel-purified, and single-stranded ends were filled in by treatmentwith T4 polymerase. BamHI linkers were ligated to bothblunt ends, and the fragment was inserted into the expressionplasmid at the BamHI restriction site to form pDll-Ep.Recombinant plasmids were cloned in Escherichia coliHB101 and purified by isopycnic centrifugation in cesium

Abbreviation: bp, base pair(s).

6465

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.

Dow

nloa

ded

by g

uest

on

May

3, 2

021

Proc. Natl. Acad. Sci. USA 83 (1986)

O"b.-OM 0 .4" ""

4 nm X xzY°e n

I I 1\1 1110 500

FIG. 1. Schematic representation of the Apa I restriction fragment that contains the human erythropoietin gene sequences.

chloride. The expression plasmid pDll-Ep is approximately6500 base pairs (bp) in length. The construction was con-firmed by restriction mapping and partial dideoxynucleotidesequencing.

Transfection of Mammalian Cells. Mammalian cell lines,COS-7 (monkey kidney) and BHK (baby hamster kidney),were maintained in Dulbecco's modified essential mediumcontaining 10% (vol/vol) fetal calf serum. Cells were pas-saged and, when cultures were 50-70% confluent, cells weretransfected by the calcium phosphate method (15). Fortransient expression of cells in a 100-mm culture dish, a totalof 20 ug of DNA was used as follows: 10 ,ug of plasmidcontaining the erythropoietin gene and 10 ,ug of carriersalmon sperm DNA. After 48 hr the supernatant was col-lected, centrifuged at 400 x g for 10 min to remove cells anddebris, and frozen at -20°C. The cells were harvestedseparately. Control experiments for these transfection assaysincluded supernatants from nontransfected cells and parallelcultures of cells transfected with plasmids containing DNAencoding other proteins, including bacterial chloramphenicolacetyltransferase and human coagulation protein, factor IX.For transfections to establish stable cell lines producing

high levels oferythropoietin, either COS-7 orBHK cells werecotransfected with the pD11-Ep plasmid and pDHFRr, aplasmid containing a cDNA for dihydrofolate reductase, in asimilar mammalian expression vector. The transfection pro-cedure was modified so that 5 ,ug of pDll-Ep plasmid, 5 ,ugofpDHFRr plasmid, and 10 ,ug of carrier DNA were cotrans-fected. After additional incubation for 18-24 hr, variousconcentrations of methotrexate (10 nM to 1 mM) were addedto the cultures. After incubation for several more days, viablecolonies resistant to methotrexate were isolated, passaged,and screened for the presence of erythropoietin bioactivity inthe supernatant. To amplify the expression of the transcrip-tional unit containing the erythropoietin gene and thedihydrofolate reductase (DHFR) gene, cell lines 'secretinghigh levels oferythropoietin were passaged several times intoincreasingly higher concentrations of methotrexate (16). Celllines were considered stable if erythropoietin productionremained high for more than 15 passages in the absence ofmethotrexate selective pressure.

Assays for Erythropoietin. The in vitro assay for erythro-poietin biological activity was based on the formation oferythroid colonies (from erythroid colony-forming cells) incultures of mouse bone marrow cells in plasma clot (17). Thesensitivity of this assay is about 5 milliunits per ml. Theerythropoietin used as the standard for assay was a partiallypurified preparation from plasma from anemic sheep (Con-naught, Step 3 erythropoietin, Lot 3026). Supernatants wereassayed from passaged cell lines grown for 24 hr in freshmedium without methotrexate. The supernatant was diluted1:200 with medium, and amounts between 1 and 10 IlI wereadded per ml of assay culture containing 2 x 105 cells, 10%(vol/vol) bovine citrated plasma, 20% (vol/vol) fetal calfserum, 1% bovine serum albumin, and 1.6% (wt/vol) beefembryo extract (GIBCO). After incubation for 36-48 hr, theplasma clots were fixed on microscope slides, stained withbenzidine for hemoglobin, and erythroid colonies werecounted. In the absence ofadded erythropoietin, no erythroidcolony-forming cell-derived colonies were detected. Optimalerythroid colony growth (100-150 erythroid colony forming

cells detected per 2 x 104 marrow cells) was observedroutinely with 50 milliunits (0.64 ng) of erythropoietin per mlof culture. Selected cell lines were also assayed for in vivoerythropoietin activity in exhypoxic polycythemic mice (18).In addition, supernatants from selected cell lines were as-sayed for immunologically reactive erythropoietin by com-petitive radioimmunoassay using a polyvalent anti-humanerythropoietin antiserum (19).Chromosome Mapping. Human DNA was treated with

several restriction endonucleases and electrophoresed in a1% agarose gel. Southern transfer to a nitrocellulose filterwas followed by hybridization using high stringency condi-tions with nick-translated 32P-labeled DNA restiction frag-ments derived from the human erythropoietin gene. Severalrestriction fragments were tested to select fragments provid-ing optimal hybridization signals and to avoid Alu repetitivesequences located in the second intron of the humanerythropoietin gene. Autoradiography was performed over-night and for 10 days. The following two restriction fragmentswere chosen for chromosomal localization studies: a 916-bpSma I fragment from the 5' end ofthe gene and an 854-bpXmnI-Apa I fragment from the 3' end (Fig. 1).Chromosome suspensions were prepared from a lympho-

cyte cell line in Tris/spermine buffer and stained with theDIPI-chromomycin A3 stain pair (20). Thirty thousand chro-mosomes of each type were sorted directly onto a single spotof a nitrocellulose filter using a high resolution dual laserchromosome sorter (21). Thus, 21 unique fractions of humanchromosomes bound to nitrocellulose filters were isolated(20). The filter-bound chromosomal DNA was denatured,neutralized, hybridized, washed, and autoradiographed usingstandard conditions (21). Both the 5' fragment probe and the3' fragment probe were hybridized independently to chro-mosomal filter sets.

RESULTS AND DISCUSSIONIsolation of Genomic Sequences. Approximately 4.8 X 106

bacteriophage were screened by hybridization of replicatenitrocellulose filters. Three different clones remained posi-tive through plaque purification, and the DNA insert wascharacterized by restriction mapping and partial dideoxynu-cleotide sequencing. Two of the three clones containedapparently complete information for the erythropoietin gene.The restriction map and sequence for the Apa I fragment ofthese clones were essentially the same as those published byJacobs et al. (5) for the gene for human erythropoietin. Thelow frequency ofphage isolates containing the erythropoietingene in this library, one in 2 x 106 bacteriophage, may beattributed to the loss of sequences during amplificationprocedures used to maintain this library. Southern blothybridization of total human genomic DNA with the Apa Ifragment of other restriction fragments of the erythropoietingene indicated only a single hybridizing band with no addi-tional regions of highly homologous DNA. This finding isconsistent with the suggestion that erythropoietin exists as a

single copy in the human genome.Expression ofErythropoietin Gene in Mammalian Cels. The

structure of the expression plasmid pDll-Ep is shown in Fig.2. The 2426-bp Apa I restriction fragment of the erythropoi-etin gene was treated with T4 DNA polymerase to produce

_ 2

c oE .0

x 2

I I1000

0 0.0 a

I I

1500 2000 2500 bp

- m-ff

m. m

6466 Cell Biology: Powell et al.

Dow

nloa

ded

by g

uest

on

May

3, 2

021

Proc. Natl. Acad. Sci. USA 83 (1986) 6467

RI B

/ o-l E fMLP L)-3 5'ss 3ssi

B

Ep

1450 3926

FIG. 2. Diagram of the plasmid expression vector pD11-Ep. The 2426-bp Apa I fragment containing the human erythropoietin gene wastreated with T4 DNA polymerase to produce blunt ends. BamHI linkers were added to the fragment, and the fragment was ligated into the uniquecloning site of the plasmid vector (B, BamHI restriction site). The plasmid pD11 contains 350 bp of the adenovirus left terminus (0-1), the originand enhancer sequences from simian virus 40 (E), the adenovirus major late promoter (MLP), the adenovirus type 2 tripartite leader (L1-3),and third leader 5' splice site (5' ss), an immunoglobulin 3' splice site (3' ss), and the late simian virus 40 polyadenylylation signal (pA) in theEcoRI (RI) restriction site of pML (24). The plasmid pD11-Ep is approximately 6500 bp in length.

blunt ends, and BamHI linkers were added to the fragment.The construct was inserted into the uniqueBamHI restrictionsite of the plasmid vector to direct transcription of theerythropoietin gene from a strong promoter. The plasmidpDll contains the adenovirus major late promoter andtripartite leader sequences that enhance translation. Theinserted Apa I restriction fragment contained 58 bp of5'-untranslated sequences followed by sequences coding fora putative 27-amino acid signal peptide, the mature protein,four intervening sequences, and 22 bp of 3' noncoding DNAsequence. This expression vector construction was chosen tooptimize production of erythropoietin protein. We used thecomplete erythropoietin gene for expression to include po-tential regulatory or enhancing sequences located in intronsthat might contribute to erythropoietin gene expression orprotein modification and secretion.The results of transfections with the erythropoietin gene

alone for transient expression are shown in Table 1. Thelevels of erythropoietin secreted into the supernatant ofeither the COS-7 or BHK mammalian cell lines were -80times higher than those reported for transient expression ofa cDNA coding for erythropoietin (5, 6). The reasons for thishigher expression are not clear but may relate to the use ofthe erythropoietin gene with intervening sequences ratherthan erythropoietin cDNA or to our use of a plasmid vectorthat is more suitable for expression. None of the controlcultures, mock transfections, or cultured cells transfectedwith other genes had detectable erythropoietin activity.To establish stable cell lines producing large amounts of

erythropoietin protein, mammalian cell lines were cotrans-fected with the pD11-Ep plasmid and pDHFRr plasmidcontaining a cDNA for dihydrofolate reductase. Aftertransfection, the culture medium was changed to includeseveral different concentrations of methotrexate. Cells thatincorporated the dihydrofolate reductase gene would beviable in the selective medium. Thus colonies resistant to themethotrexate selection were isolated and passaged. Approx-imately half of the methotrexate-resistant colonies thatwere assayed secreted detectable erythropoietin activity.Amounts of erythropoietin activity in the cell pellets couldnot be determined due to the presence of significant inhibitorsof the bioassay in the cell extracts. Consequently our resultsdo not analyze the intracellular levels of erythropoietinprotein but, rather, the amount of erythropoietin proteinproduced and secreted into the supernatant by the cell lines(Table 2).

Table 1. Secretion of recombinant human erythropoietin bymammalian cells transfected with plasmid pD11-Ep fortransient expression assay

Erythropoietin per ml of culture

Mammalian Activity, unitscells Protein, pZg (in vitro bioassay)BHK 3.4 0.2 270 ± 16COS-7 3.2 0.4 255 ± 32

Data are from three experiments for each cell type.

The recombinant erythropoietin protein secreted into thesupernatant of transfected cell lines was biologically activeand large amounts ofhormone were secreted (up to 7000 unitsper ml). If the recombinant erythropoietin has a specificactivity equivalent to that of natural erythropoietin (78,000units per mg of protein), the biological assay corresponds toapproximately 80 ug of erythropoietin protein per ml. Theconcentration of erythropoietin protein from selected celllines also was assayed by a competitive radioimmunoassayusing a polyvalent rabbit anti-human erythropoietin antise-rum. The amount of protein measured by the radioimmu-noassay was equivalent to the protein level estimated by thebiological assay. These data indicate that the transfected celllines expressed and secreted erythropoietin protein that wasgreater than 98% active.The recombinant erythropoietin produced by the trans-

fected cells was further characterized to demonstrate thatthese cells were secreting authentic hormone. Supernatantsfrom several cell lines had potent in vivo biological activitywhen assayed in the exhypoxic polycythemic mouse. Inexperiments using partially purified native erythropoietin, ithas been noted that neuraminidase treatment completelyabrogated erythropoietin activity when assayed in the intactanimal (22). The loss of activity presumably was due toincreased clearance by the liver of the desialated hormonesince neuraminidase-treated erythropoietin remained fullyactive in vitro. The observation of potent in vivo biologicalactivity is important primarily because it partially confirmsthat the transfected mammalian cell lines appropriately addcarbohydrate including the terminal sialic' acids to theerythropoietin protein during posttranslational modification.In separate experiments, the activity of erythropoietin in thein vitro biological assay was neutralized by a neutralizinganti-human erythropoietin antibody added to the culture.While these data suggest that the recombinant erythropoietinproduced by the transfected cell line is very similar to nativeerythropoietin, protein sequence determination and analysesof the carbohydrate components will be necessary to deter-mine precisely the fidelity of expression ofthe erythropoietingene and of processing the erythropoietin protein by the celllines.The erythropoietin secreted into the supernatant of repre-

sentative transfected cell lines was also assayed for prolifer-

Table 2. Expression of recombinant erythropoietin in stablytransfected mammalian cell lines

Erythropoietin per ml ofsupernatant

Activity, unitsCell line Protein, ,ug (in vitro bioassay)F 1.1 12.4 970F 3.4 32.0 2500F6.1 79.6 6210S 1.2 6.4 500S 2.4 64.2 5000S 5.2 82.1 6400

Cell Biology: Powell et al.

Dow

nloa

ded

by g

uest

on

May

3, 2

021

6468 Cell Biology: Powell et al.

1

10-12

13

Proc. Natl. Acad. Sci. USA 83 (1986)

2 3

9 9-2

18 19

4 5

14,15 17

20 22

8

14

y

6 7 &

16 15

21 x

FIG. 3. Autoradiogram of sorted human chromosomes on sets of nitrocellulose filters hybridized with the 915-bp 5' probe bound by SmaI restriction sites. The 854-bp 3' probe Xmn I-Apa I gave an identical autoradiogram with an independently sorted filter set. Twelve 25-mmcircular nitrocellulose filters with two chromosome spots each are displayed with the chromosome numbers on the edge of each filter adjacentto each sorted chromosome spot. Chromosome 7 produced an intense hybridization signal, mapping the gene to that chromosome.

ative effects on other bone marrow progenitor cells. Recom-binant erythropoietin was assayed for its effect on a varietyof progenitors from mouse and human marrow includingerythroid colony-forming cells, erythroid burst-forming cells,granulocyte-macrophage precursors, and mixed-cell colony-forming cells (23). Erythroid progenitor cells exhibited aproliferation response to recombinant erythropoietin thatwas parallel to the dose-response relationship found withnatural erythropoietin. Neither granulocyte-macrophageprecursors nor mixed-cell colony-forming cells exhibited anyproliferative response to the recombinant erythropoietin atconcentrations up to 10 units/ml of assay cell culture.Chromosome Mapping. High resolution dual laser sorting

offluorescent chromosomes was employed to sort and isolate21 fractions of human chromosomes bound to nitrocellulosefilters. The autoradiograph obtained after hybridization withthe nick-translated 32P-labeled 5'-restriction fragment isshown in Fig. 3. Even after autoradiography for 1 week, onlythe dot blot identified as chromosome 7 DNA exhibited anyhybridizing signal. A similar autoradiograph was obtainedwhen the 3'-restriction fragment was hybridized indepen-dently. To confirm that the gene for human erythropoietinwas located on chromosome 7, we subsequently hybridizedthe 3' probe to nitrocellulose filters of chromosomes sortedfrom the human fibroblast cell line GM44 that carries areciprocal translocation of the distal portion of chromosome7 (p21.2) exchanged for a portion ofchromosome 10 (qll.21).In these experiments, the probe identified the sorted deriv-ative chromosome 7 and not the translocated portion onderivative chromosome 10. Again, no other chromosomeDNA hybridized labeled probe, even after extensiveautoradiography. We have confirmed that our probes hybrid-ize with DNA from sorted chromosome 7 by several controlexperiments. A probe for epidermal growth factor receptor(kindly provided by Michael Rosenfeld), which we mapped tochromosome 7 by dot-blot analysis, had been localized tochromosome 7 by somatic cell hybridization (25). Thisradiolabeled probe hybridized specifically to the spot forchromosome 7 in the filter set used to map erythropoietin. Todate, this technique has localized 24 genes, and each assign-ment has been confirmed by other methods.

In the present experiments, we have shown that biologi-cally active human erythropoietin can be expressed at highlevels (up to 80 mg protein per liter of supernatant) fromstably transfected mammalian cell lines and that the gene forhuman erythropoietin is located on chromosome 7. Thus thebasis is established for an abundant source of purified humanerythropoietin for further biochemical and clinical studies.

We thank Barry D. Bruce for sorting the fluorescent chromo-somes; Mei-Chi Cheung for hybridization of the chromosome prep-arations; Dr. Patrick Chou of the Department of Chemistry forpreparation of the synthetic oligonucleotides; Nancy Lin and CarylCampbell for biological assays; Dr. Joan Egrie for the radio-immunoassay determinations; Dr. Jaime Caro for performing in vivobioassays; and Drs. Earl Davie and Kotoku Kurachi for insightfuldiscussion. We also thank Carmen Nott for preparation of themanuscript. This work was supported in part by Research Grants HL16919, AM 19410, and CA 31615 from the National Institutes ofHealth. J.S.P. is the recipient of Clinical Investigator Award AM01418 from the National Institute of Arthritis, Diabetes, and Diges-tive and Kidney Diseases.

1. Jacobson, L. 0., Goldwasser, E., Fried, W. & Plzak, L. (1957)Nature (London) 179, 633-634.

2. Garcia, J. F., Ebbe, S., Hollander, L., Cutting, H. O., Miller,M. & Cronkite, E. P. (1982) J. Lab. Clin. Med. 99, 624-635.

3. Miyake, T., Kung, C. K. & Goldwasser, E. (1977) J. Biol.Chem. 252, 5558-5564.

4. Sytkowski, A. J. (1980) Biochem. Biophys. Res. Commun. 96,143-149.

5. Jacobs, K., Shoemaker, C., Rudersdorf, R., Neill, S., Kauf-man, J., Musfon, A., Seehra, J., Jones, S., Hewick, R.,Fritsch, E., Kawakita, M., Shimizu, T. & Miyake, T. (1985)Nature (London) 313, 806-810.

6. Lin, F.-K., Suggs, S., Lin, C.-H., Browne, J. K., Smalling,R., Egrie, J. C., Chen, K. K., Fox, G. M., Martin, F.,Stabinsky, Z., Badrawi, S. M., Lai, P.-H. & Goldwasser, E.(1985) Proc. NatI. Acad. Sci. USA 82, 7580-7584.

7. Powell, J. S., Segal, G. M., Berkner, K. L. & Adamson, J. W.(1985) Blood 66, 756 (abstr.).

8. Lawn, R. M., Fritsch, E. F., Parker, R. C., Blake, G. &Maniatis, T. (1978) Cell 15, 1157-1174.

9. Deryneck, R., Roberts, A. B., Winkler, M. E., Chen, E. Y. &Goeddel, D. V. (1984) Cell 38, 287-297.

10. Yanagawa, S., Kirade, K., Hideki, O., Sasaki, R., Chiba, H.,Ueda, M. & Masaaki, G. S. (1984) J. Biol. Chem. 259,2707-2710.

11. Grantham, R., Gautier, C., Gouy, M., Jacobzone, M. &Mercier, R. (1981) Nucleic Acids Res. 9, 43-59.

12. Jaye, M., de la Salle, H., Schamber, F., Balland, A., Kohli,V., Findeli, A., Tolstoshev, P. & Lecocq, J. (1983) NucleicAcids Res. 11, 2325-2335.

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

14. Berkner, K. L. & Sharp, P. (1985) Nucleic Acids Res. 13,841-857.

15. Graham, F. L. & van der Eb, A. J. (1973) Virology 52,456-467.

16. Simonsen, C. C. & Levinson, A. D. (1983) Proc. Natl. Acad.Sci. USA 80, 2495-2499.

Dow

nloa

ded

by g

uest

on

May

3, 2

021

Cell Biology: Powell et al.

17. Adamson, J. W., Torok-Storb, B. & Lin, N. (1978) Blood Cells4, 89-103.

18. Cotes, P. M. & Bangham, D. R. (1961) Nature (London) 191,1065-1087.

19. Egrie, J. C., Brown, J., Lai, P. & Lin, F. K. (1985) Prog. Clin.Biol. Res. 191, 339-350.

20. Lebo, R. V., Gorin, F., Fletterick, R. J., Kao, F., Cheung,M. C., Bruce, B. D. & Kan, Y. W. (1984) Science 225, 57-59.

Proc. Nadl. Acad. Sci. USA 83 (1986) 6469

21. Lebo, R. V. & Bastian, A. M. (1982) Cytometry 3, 213-219.22. Goldwasser, E. & Kung, C. K. H. (1972) J. Biol. Chem. 247,

5159-5160.23. Powell, J. S., Fialkow, P. J. & Adamson, J. W. (1982) J. Cell.

Physiol. Suppl. 1, 79-85.24. Lusky, M. & Batchan, M. (1981) Nature (London) 293, 79-81.25. Shimizu, N., Behzadian, M. A. & Shimuzu, Y. (1980) Proc.

Natl. Acad. Sci. USA 77, 3600-3604.

Dow

nloa

ded

by g

uest

on

May

3, 2

021