CA2190289A1 - Dna construct for effecting homologous recombination and uses thereof - Google Patents

Abstract

The invention relates to constructs comprising: a) a targeting sequence; b) a regulatory sequence; c) an exon; and d) an unpaired splice-donor site. The invention further relates to a method of producing protein in vitro or in vivo comprising the homologous recombination of a construct as described above within a cell. The homologously recombinant cell is then maintained under conditions which will permit transcription and translation, resulting in protein expression. The present invention further relates to homologously recombinant cells, including primary, secondary, or immortalized vertebrate cells, methods of making the cells, methods of homologous recombination to produce fusion genes, methods of altering gene expression in the cells, and methods of making a protein in a cell employing the constructs of the invention.

Description

W0 95/31560 r~ IJ
DNA construct for effectlng homologous recombinat~on and uses thereof c Backqxol~n~l o~ the Invention Current approaches to treating disease by administer-ing therapeutic proteins include n vitro prof~ i r n of therapeutic proteins for conventional pharmaceutical delivery (e.g. intravenous, subcutaneous, or intramuscular injection) and, more recently, gene therapy.
Proteins o~ therapeutic interest are generally pro-duced by introducing exogenous DNA encoding the protein of therapeutic interest into appropriate cells For example, exogenous DNA encoding a desired therapeutic protein is introduced into cells, such as immortalized cells in a vector, such as a plasmid, from which the encoded protein is expressed. Further, it has been suggested that endoge-nous cellular genes ~ and their expression may be modif ied by gene targeting. See for example, IJ.S. Patent No.
5,371,071, WO 91/06666, W 91/06667 ~IDd wo 90/11394.

21 qO2~q WO 95/31560 ~ f~4!;
~ resently-available approaches to gene therapy make use of ;nfP~t;ollq vectors, such as retroviral vectors, which include the y-enetic material to be expressed. Such approaches have limitations, such as the potential of 5 generating replication-competent virus during vector prr,rll-rt jnn; re~ ' in~tinn between the tl~r~rPut;c virus and ~ndoy~.-uus retroviral genomes, pntf~nt;~l1y generating infectious agents with novel cell sper;f;rit;~ host ranges, or increased virulence and cytotoxicity; indepen-10 dent integration into large numbers of cells, increasingthe risk of a tumorigenic inaertional event; limited cloning capacity in the retrovirus (which restricts thera-peutic applicability) and short-lived n v vo expression o~ the product of interest. A better approach to provid-15 ing gene products, particularly one which avoids thelimitations and risks ~Rsor;~ted with presently available methods, would be valuable.
of the Tnvention The present invention relates to improved methods for 20 both the ' y~Q pro~llrtinn of therapeutic proteins and for the prr,~-lrt;nn and delivery of therapeutic proteins by gene therapy. In the present method, expression of a desired targeted gene in a cell (i . e ., a desired endoge-nous cellular~gene) is altered by the intro~lllrt;nn, by 25 homologous ., n~t;nn into the cellular genome at a preselected site, of DNA which includes at least a regula-tory se~uence, an exon and a splice donor site. These C are i~troduced into the chromosomal (genomic) DNA in such a manner that this, in effect, results in 30 production of a new transcription unit (in which the regulatory seSIuence, the exon and the splice donor site present in the DNA construct are operatively linked to the endogenous gene) . As a result of intrn~ rt;nn of these WO 9513156~ 2 1 9 ~ 2 ~ ~ P~~ ~T ~4r .

-ntc into the chromosomal DNA, the expression of the desired endogenous gene is altered.
Altered gene expression, as used herein, ~n~ -qSF'A
activating (or causing to be expressed) a gene which is 5 normally silent (unexpressed) in the cell as r1ht~;nF.~, increasing expression of a gene which is not e~Lessed at phyAi~ ;r~1ly significant levels in the cell as obtained, changing the pattern of re~l~t;r~n or induction such that it is different than occurs in the cell as 10 obtained, and reducing (including .o1 im;n~t;n~) expression of a gene which is e~Lesséd in the cell as obtained.
The present invention further relates to DNA con-struct6 useful in the method of altering expression of a target gene. The DNA constructs comprise: (a) a targeting 15 se~uence; (b) a regulatory sequence; (c) an exon; and (d) an unpaired splice-donor site. The targeting sequence in the D~A construct directs the integration of elements (a) - (d) into a target gene in a cell such that the elements (b) - (d) are operatively linked to sf'q~-~'n~-'A of

2 0 the ~ A target gene . In another ' ' , the DNA constructs comprise: (a) a targeting sequence, (b) a regulatory setauence, (c) an exon, (d) a splice-donor site, (e) an intron, and (f ) a splice-acceptor site, wherein the targeting se~uence directs the integration of elements 25 (a) - (f) such that the elements of (b) - (f) are opera-tively linked to the ~ gene. The targeting aequence is 1- 1 O~O-lA to the prPR--l .ortei site in the cellular ~hLI _ 1 DNA with which ~ 1 C~O--A recombina-tion is to occur. In the construct, the exon is generally 30 3' of the regulatory sequence and the splice-donor site is

3 ' of the exon.
The following serves to illu8trate two f.mho~l; A of the present invention, in which the sequences upstream of the human erythropoietin h (EPO) gene are altered to allow 35 expression of hEP0 in primary, secondary, or immortalized .

W09~131~60 ~1 9 ~ ~9 1 ~I/L ~ ~ ?45 cells which do not express EP0 in detectable quantities in their untransfected state as obtained. In: ' '; l, the targeting construct contains two targeting sequences.
The f irst targeting sequence is homologous to sequences 5 ' 5 of the second targeting sequence, and both sequences are u~LLeil", of the hEP0 coding region. The targeting con-struct also contains a regulatory region (the mMT-l pro-moter) an exon (human growth hormone (hGH) ) exon l) and an unpaired splice-donor site. The product of homologous 10 r~ ' ;n~t;on with this targeting construct i8 shown in Figure l.
In: ' '; t 2, the targeting construct also con-tains two targeting sequences . The f irst targeting se -quence is homologous to sequences within the endogenous 15 hEP0 regulatory region, and the second targeting sequence is homologous to hEP0 intron l. The targeting construct also contains a regulatory region ~the mMT-l promotor), an exon (hGH exon l) and an unpaired splice-donor site. The product of homologous r~ ' ;n~t;nn with this targeting 20 construct is shown in Figure 2.
In these two embodiments, the products of the target-ing events are chimeric transcription units which generate a mature mR~ in which the f irst exon of the hGH gene is positioned upstream of hEP0 exons 2-5. The product of 25 transcription, splicing, and tr~n~l~t;~n is a protein in which amino acids 1-4 of the hEP0 signal peptide are replaced with amino acid reaidues 1-3 of hGH. The two ~mhod; ~ differ with respect to both the relative positions of the regulatory se~lut~ s of the targeting 30 construct that are inserted and the specific pattern of splicing that needs to occur to produce the f inal, pro-cessed transcript.
The invention further relates to a method of pro-ducing protein in vit,ro or n Y vo through introduction of 35 a construct as described above into host cell ~

WO95/31~60 2~90289 r ~
.

DNA by homologous rr ' ;nAtir~n to produce a h~ o~QIlRly recom.~inant cell . The hQmQlogously recnm~; n 7nt cell is then m-intA;n,orl under conditions which will permit tran-scription, translation and secretion, resulting in produc-5 tion of the protein of interest.
The present invention relates to trAnRfect~ cells, such as transfected primary or secondary cells (i.e., non-immortalized cells) and transfected immortalized cells, useful for producing proteins, particularly th~r.7r~ tic lO proteins, methods of making such cells, methods of using the cells for i~ y~ protein pro~ lrt;nn, and methods of gene therapy. Cells of the present invention are of vertebrate origin, particularly of l i An origin, and even more particularly of human origin. Cells produced by 15 the method of the present invention contain DNA which encodes a therapeutic product, DNA which i8 itself a therapeutic product and/or DNA which causes the transfected cells to express a gene at a higher level or with a pattern of regulation or ;nr7-l~ tif,n that is differ-20 ent than occurs in the ~:uLL 'l~'"'r7;n~ nontransfected cell.
The present invention also relates to methods bywhich cells, such as primary, ~e~ Ly, and immortalized cells, are transfected to include f~Y~ n~uq genetic mate-rial, methods of producing clonal cell gtrains or heterog-25 enous cell strains, and methods of; ;7;n~ animals orproducing Ant;hor'.i~R in i ; ~d animals, using the transfected primary, secondary, or immortalized cells.
The present invention relates particularly to a method of gene targeting or h~ l 03~l-q recom.bination in 30 eukaryotic cells, such as cells of fungal, plant or ani-mal, e.g., vertebrate, particularly I l iAn, and even more particularly, human origin. That is, it relates to a method of introducing DNA into primary, secondary, or immortalized cells of vertebrate origin through homologous 35 re ' ;nAt;~7n, such that the DNA is introduced into genom-WO 95/31~60 ~ I q Q ~ 8 9 ~ ~ ", ~ 5 .

ic DNA of the primary, secondary, or immortalized cells ata preselected site. The targeting 6er~uences used are selected with reference to the site into which the DNA in the targeting DNA construct is to be inserted. The pres-5 ent invention further relates to hnmnlrrjollqly recombinantprimary, secondary, or immortalized cells, referred to as homologously Ll__ ' ;nAnt ~R) primary, secondary or immor-talized cells, produced by the present method and to uses of the HR primary, secondary, or immortalized cells.
In one ~mho~l; t of the present invention in which expression of a gene is altered, the gene is activated.
That is, a gene present in primary, secondary, or immor-talized cells of vertebrate origin, which is normally not expressed in the cells as obtained, is activated and, as a 15 result, the encoded protein is expressed. In this embodi-ment, homologous le~ ' inAt jnn is used to replace, dis-able, or disrupt the regulatory region normally associated with the gene in cells as nhtA;n~d through the insertion of a regulatory sequence which causes the gene to be 20 expressed at levels higher than evident in the uuLL~ ~uulld-ing nontrAnqfert~l cell.
In one ~mhorii , the activated gene can be further amplified by the inrlllqinn of an _lif;Ahle selectable marker gene which has the ULuueLLy that cells rnntA;n;
25 amplified copies of the selectable marker gene can be selected for by culturing the cells in the presence of the appropriate selectable agent. The activated ell-lo~ uu~
gene is ~ 7; f; ed in tandem with the ~ l; f i Ahl f' select-able marker gene. Cells rnntA;n;n, many copies of the 30 activated f-n.lnJ~nmlq gene are useful for n vitro protein production and gene therapy.
Gene targeting and amplification as ~l;qrloq~ in the present invention are particularly useful for activating the expression of genes which f orm transcription units 35 which are s~ff;r;~ntly large that they are difficult to W095/31560 2 1 9 0289 ~ s .

i601ate and express, or for activating genes for which the entire protein coding region is unavailable or has not been cloned.
In a further embodiment, expression of a gene which 5 is expressed in a cell as obtained is ~nh~nned or caused to display a pattern of regulation or induction that is different than evident in the corr~Rpnn~1n~ nontransfected cell. In another ' '; -, expression of a gene which is expressed in a cell as nht:~;n~l is reduced (i.e., 10 lessened or ~1; m; n~t~ ) . The present invention also de-scribes a method by which homologous recombination is used to convert a gene into a cDNA copy, devoid of introns, for transfer into yeast or bacteria for i ~itrQ protein production .
Transfected cells of the present invention are useful in a number of applications in humans and animals. In one , the cells can be implanted into a human or an animal for protein delivery in the human or animal. For example, hGH, hEP0, human insulinotropin, and other pro-teins can be delivered systemically or locally in humans for therapeutic benefits. In addition, transfected non-human cells producing growth hormone, erythropoietin, insulinotropin and other proteins of non-human origin may be produced.
Barrier devices, which contain transfected cells which express a therapeutic product and through which the therapeutic product is freely permeable, can be used to retain cells in a fixed position ' v vo or to protect and isolate the cells from the host ~ 8 immune system . Barrier devices are particularly u8eful and allow tr~ncferterl immortalized cells, transfected xenogeneic cells, or transfected ~11 o~f~n~; C cells to be implanted for treatment of human or animal conditions or for agricultural uses ~e.g., bovine growth hormone for dairy pro~ rt;nn).
Barrier devices also allow convenient short-term (i.e., WO 95J31560 2 1 9 Q 2 ~ 9 1 .,. ~ '4~
.

transient) therapy by providing ready access to the cells for removal when the treatment regimen is to be halted or any reason. In addition, transfected xenogeneic and allogeneic cells may be used in the absence of barrier 5 devices for short-term gene therapy, such that the gene product produced by the cells will be delivered ~ vivo until the cells are rejected by the host's immune system.
Transfected cells of the present invention are also useful for ~lir;t;nJr antibody production or for; i7;nrJ
10 humans and animals against pAt~ng~n; r agents . T l Ant~o~
transfected cells can be used to deliver; ; 7; nrJ anti-gens that result in st;m~llAtinn of the host's cellular and humoral immune responses. These immune responses can be designed for protection of the host from future infectious 15 agents ti.e., for vA~r;n~t;r~n), to stimulate and augment the disease-ighting capabilities directed against an ongoing ;nfert;nn, or to produce ~nt;ho~;~c ~;re~ted against the antigen produced ~ vivo by the transfected cells that can be useful for therapeutic or diagnostic 20 purposes. Removable barrier devices rnntAininJ the cells can be used to allow a simple means of trrm;n~t;nrJ expo-sure to the antigen. Alternatively, the use of cells that will ultimately be rejected (~rnn~.cn~ir or allogeneic transfected cells) can be used to limit exposure to the 25 antigen, since antigen production will cease when the cells have been rejected.
The methods of the present invention can be used to produce primary, secondary, or immortalized cells produc-ing a wide variety of therAr~lltirAlly useful products, 30 including ~but not limited to): hormones, cytokines, antigens, Ant;hnri;~, enzymes, clotting factors, transport proteins, receptors, regulatory proteins, structural proteins, transcription factors, ribozymes or anti-sense ~NA. Additionally, the methods of the present invention 35 can be used to produce cells which produce non-naturally W095/31560 2 1 ~ 9 .
g occurring ribozymes, proteins, or nucleic acids which are useful for i vitro production of a therapeutic product or f or gene therapy .
3rief Descril:1tion of the ~rawincs Figure 1 is a schematic diagram of a 6trategy for transcriptionally activating the hEpO gene; thick lines, mouse metalloth~nnP;n I promoter; stippled box, 5' un-tr~ncl~te~ region of hGH; solid box, hGH exon 1; striped box, 10 bp splice-donor sequence from hEPO intron 1;
cross-hatched box, 5' untrAncl~tpr~ region of hEPO; open numbered boxe6, hEPO coding sequences; diagonally-stripped box, hEPO 3' untranslated sequence6; HIII, HindIII site.
Figure 2 is a schematic diagram of a strategy for transcriptionally activating the hEpO gene; thick lines, mouse met~lloth;nnp;n I promoter; stippled box, 5' un-tr~nRl Atp~l region of hGH; solid box, hGH exon l; open numoered boxes, hEPO coding sequences; rl;:-~nn~l1y-stripped box, hEPO 3' llnt~n~l~ted sequences; HIII, HindIII site.
Figure 3 is a schematic le~JLese~l~ation of plasmid pXGH5, which includes the hGH gene under the control of the mouse metalloth;~nP;n promoter.
Figure 4 is a 8, ` ; c reprP~Pnt~t; nn of plasmid pE3neoEPO. The positions of the human erythr~pnjpt;n gene and the neomycin phosphotranferase gene (neo) and ampicillin (amp) resistance genes are ;nf~lr~tP~l~ Arrows indicate the directions of transcription of the various genes. pmMT1 denotes the mouse metalloth;nnp;n promoter (driving hEPO expression) and pTK denotes the Herpes Simplex Virus thymidine kinase promoter (driving nêo 3 0 expression) . The dotted regions of the map mark the positions of sequences derived from the human hypoxan-thine-guanine phogphoribosyl transferase (HPRT) locus.
The relative pogitions of restriction endonuclease recog-nition sites are indicated.

Wo 95l3 l560 I ~ 5 21 q~289 ~

Figure 5 iB a schematic representation of plasmid pcDÆO, which includes the neo coding region (BamHI-BglII
fragment) from plasmid pSV2neo inserted into the BamHI
site of plasmid pcD; the Amp-R and pBR3220ri seSIuences 5 from pBR322; and the polyA, 16S splice junctions and early promoter regions f rom SV4 o .
Figure 6 is a srh ~; c representation of plasmid pREPO4 .
Figure 7 is a graphic repro~ontAt;r,~ of erythropoie-lO tin expression in a targeted human cell line subjected tostepwise selection in 0 . 02, O . 05, O .1, O . 2 and 0 . 4 IlM
methotrexate .
Figure 8 is a srh ~ i c representation of plasmid pREPO15. Fragments derived from genomic hEPO sequences 15 are indicated by filled boxes. The region between BamHI
(3537) and BgIII'/HindIII' .~ L~ to sequences at positions 1-4008 in Genbank entry HUMERPALU. The region between BgIII'/HindIII' (11463) ~oLLes~ lds to DNA se-s~uences at positions 4009-5169 in Genbank entry HUMERPALU.
20 The region between HindIII (11463) and XhoI (624) contains sequence coLLe~ llding to positions 7-624 of Genbank entry ~IUMERPA. CMV promoter sequences are shown as an open box and contains seq~lonro from nucleotides 546-2105 of Genbank sequence ~S5MI~P. The neo gene is shown as an open box 25 with an arrow. The thymidine kinase (tk) promoter driving the neo gene is shown as a hatched box. pBSIISR* sequenc-es ;nr1ll~1;nr the amp gene are ;n~l;r~to~ by a thin line.
Figure 9A presents restriction enzyme maps and sche-matic representations of the products observed upon 30 digestion of the endu~el~Jus hEPO gene (top) and the acti-vated hEPO gene after hrn~r~lo~oll~ re: ' ;nAt;rn with the targeting fragment from pREPO15 (bottom).
Figure 9B presents the results of restriction enzyme digestion and Southern hybridization analysis of untreated WO95/31560 ~ ,lIL_ ' 15 ~XF) and targetea ~T1) human fibrobla6t clone HF342-15 see Example 7 ) .
Figure 10 i8 a ~qr~ tiC representation of plasmid pREPO1~. FL _ tq derived from genomic hEP0 ser~uences are indicated by filled boxes. The region between BamXI
~3537) and ClaI ~7554) corresponds to ser~uences at posi-tions 1-400~ in Genbank entry XUMERPALU. The region between ATG ~12246) and HindIII ~13426) corresponds to DNA
seriuence at positions 4009-5169~in Genbank entry XDMERPLAU. The region between XindIII ~13426) and XhoI
(624) r~ntA;nq sequence corresponding to positions 7-624 of Genbank entry X~MERPA. CMV promoter ser~uences are shown ae an open box and contains ser~uence from nucleo-tides 546-2015 o~ Genbank sequence XS5MIEP. The dihydrofolate r~ lrt~qe (dhfr) transcription unit is shown as a stippled box with an arrow. The neo gene is shown as an open box with an arrow. The tk promoter driving the neo gene is shown as a hatched box. pBSIISK+ seriuences inrl~l;n~ the amp gene are in~i r~ted by a thin line.
Figure 11 is a schematic illustration of a construct of the invention for activating and amplifying an intronless gene, the a-interferon gene, where the con-struct comprises a first targeting sequence (1), an ampli-fiable marker gene (AM), a selectable marker gene (SM), a regulatory sequence, a CAP site, a splice-donor site ~SD), an intron ~thin lines), a splice-acceptor site ~SA) and a second targeting ser~uence (2). The black box represents coding DNA and the stippled boxes represent untr~nql ~t~
ser~uences .
Figure 12 is a schematic illustration of a construct of the invention for activating and amplifying an endoge-nous gene wherein the f irst exon contributes to the signal peptide, the human GM- CSF gene, where the construct com-prises a first targeting ser~uence ~1), an amplifiable marker gene ~AM), a selectable marker gene ~SM), a regula-W095/31~C0 ~21 9028q Y~ J., irt~

tory sequence, a CAP site, a splice-donor site (SD), and a second targeting sequence (2 ) The black boxes represent coding DNA and the stippled boxes represent untranslated seguences .
Figure 13 is a Ar~ ;^ illustration of a construct of the invention for activating and amplifying an endoge-nous gene wherein the first exon contributes to the signal peptide, the human G-CSF gene, where the construct com-prises a first targeting sequence (1), an: ,l;fiAhle marker gene (AM), a selectable marker gene (SM), a regula-tory sequence, a CAP site, a splice-donor site (SD), and a second targeting sequence (2). The black boxes represent coding DNA and the stippled boxes represent untranslated seSIuences .
Figure 14 is a schematic illustration of a construct of the invention for activating and amplifying an endoge-nous gene wherein the first exon is non-coding, the human FSH,~ gene, where the construct comprises a first targeting sequence (1), an amplifiable marker gene (AM), a selecta-ble marker gene (SM), a regulatory s-qu-n~ a CAP site, a splice-donor site (SD), and a second targeting seguence (2). The black boxes represent coding DNA and the stippled boxes repregent untran81ated 5_~ _n^~R
Det^;led Descrivtion of the Invention The invention is based upon the discovery that the regulation or activity of _n~ln~_nnll_ genes of interest in a cell can be altered by inserting into the cell genome, at a preselected site, through homologous re~ '-;nAtinn, DNA constructs comprising: (a) a targeting se5~uence; (b) a regulatory seguence; (c) an exon and (d) an unpaired splice-donor site, wherein the targeting seguence directs the integration of elements (a) - (d) such that the ele-ments (b) - (d) are operatively linked to the endogenous gene. In another: ' :'; , the DNA constructs comprise:

WO95131560 21 q~289 ~ L CE~45 (a) a targeting sequence, ~b) a regulatory seguence, (c) an exon, (d) a splice-donor site, (e) an intron, and (f) a splice-acceptor site, wherein the targeting sequence directs the integration of elements (a) - (f) such that 5 the elements of (b) - (f ) are operatively linked to the first exon of the endogenous gene. The targeting seguen-ces used are selected with reference to the site into which the DNA is to be inserted. In both; ' '; ~c the targeting event is used to create a new transcription lO unit, which is a fusion product of RPrluPnr~C introduced by the targeting DNA constructs and the ~ nJ-~ R cellular gene. As discussed herein, for example, the formation of the new transcription unit allows transcriptionally silent genes (genes not expressed in a cell prior to transfec-15 tion) to be activated in host cells by introducing intothe host cell ' s genome DNA constructs of the present invention. As also ~1; Rr--RRPd herein, the expression of an lR gene which is ~ ,L~ed in a cell as obtained can be altered in that it is increased, reduced, including 20 Pl;min:~t~d~ or the pattern of re~llRtinn or ;n~-lrt;nn may be changed through use of the method and DNA constructs of the present invention.
The present invention as set forth above, relates to a method of gene or DNA targeting in cells of eukary-otic 25 origin, such as of fungal, plant or animal, such as, vertebrate, particularly l; :~n ~ and even more particu-larly human origin. That is, it relates to a method of introducing DNA into a cell, such as primary, 3e~ c.Ly, or immortalized cells of vertebrate origin, through homol-33 ogous L~ ' in~tion or targeting of the DNA, which isintroduced intD genomic DNA of the cells at a preselected site. It is particularly related to homologous re- ' ln~-tion in which the transcription and/or translation prod-ucts of ~nrin~nn~c genes are modified through the use of 35 DNA constructs comprising a targeting se~luence, a regula-21 qO289 WO 95131560 _l/L_ '~G'` 15 .

tory sequence, an exon and a splice-donor site. The pres-ent invention further relates to homologougly rPcrmh;n~nt cells produced by the present method and to uses of the homologously recombinant cells.
The present invention also relates to a method of activating a gene which is present in primary cells, secondary cells or immortalized cells of vertebrate ori-gin, but is normally not expressed in the cells. Homolo-gous re~ ' ln~t jnn or targeting is used to introduce into the cell's genome se~uences which causes the gene to be expressed in the rPr;riPnt cell. In a further ' _ ';
expression of a gene in a cell i6 Pn~ nred or the pattern of regulation or induction of a gene is altered, through intro~l~rt; nn of the DNA construct . As a result, the encoded product is expressed at levels higher than evident in the corrPcpnn~;n~ nontransfected cell. The present method and DNA constructs are also useful to produce cells in which expression of a desired product is less in the transfected cell than in the corrpcpnnrl; nJ nontransfected cell. That is, in the tr~nqferte~l cell, less protein (including no protein) is produced than in the cells as obtained .
In another: ' '; , a normally silent gene encod-ing a desired product is activated in a transfected, primary, secondary, or immortalized cell and ~ ; PC~
This ' '; - is a method of introducing, by I l~rj~-lc re ' in:lt;nn with genomic DNA, DNA se~uences which are not normally functionally linked to the ~ J ~ c gene and (1) which, when inserted into the host genome at or near the ~lldu~ uus gene, serve to alter (e.g., activate) the expression of the Pn~ln,Pnnuc gene, and further (2) allow for sPlert;nn of cells in which the activated endog-enous gene is amplified. Alternatively, expression of a gene normally expressed in the cell as obtained is en-hanced and the gene is amplified.

WO 9S/31560 ~ t 9 ~ 2 ~ q ~ t5 .

The following is a description of the DNA constructs of the present invention, methods in which they are used to produce transfected cells, transfected cells and uses of these cells.
The DNA Construct The DNA construct of the present invention includes at least the following ~ _ q: a targeting sequence;
a regulatory s~r~ nre; an exon and an unpaired splice-donor site. In the construct, the exon is 3~ of the 10 regulatory sequence and the unpaired splice-donor site is 3 ' of the exon. In addition, there can be multiple exons and/or introns preceding (5 ' to) the exon fla~ked by the unpaired splice-donor site. As described herein, there ~Lé~U~:l-Lly are additional construct c, ^ntq, such as a 15 sPl PrtAhl e markers or: ,1; f; ~hl e markers .
The DNA in the construct may be ref erred to as exoge-nous. The term "exogenous" is defined herein as DNA which is illLLuduced into a cell by the method of the present invention, such as with the DNA constructs defined herein.
20 E~Luyeuuus DNA can possess serluPncPq i~Gnt;r~l to or dif-ferent from the ellduyellOus DNA present in the cell prior to trans f ection .
The Tarqetinq Ser~ nrF~ or Ser~uences The targeting sequence or sequences are DNA sequences 25 which permit ]pj;t;r-te ~ loqo~q ~e~__ ' ;n~t;nn into the genome of the selected cell rnnt:~;ninq the gene of inter-est. Targeting sequences are, generally, DNA sequences which are homologous to (i.e., ;~ nt;rzll or sufficiently similar to cellular DNA such that the targeting sequence 30 and cellular DNA can undergo homologous r,o ' in~t;nn) DNA
sequences normally pregent in the genome of the cells as obtained (e.g., coding or nnnrorl;nq DNA, lying upstream of the transcriptional start site, within, or downstream of W09513~!i60 ~ 1 q O 2 ~ 9 r~ 15 .

the tran6criptional ætop site of a gene of interest, or sequence6 present in the genome through a previous modif i -cation). The targeting seguence or sequences used are selected with reference to the site into which the DNA in S the DNA construct is to be inserted.
One or more targeting 5rr~ nrF~ can be employed. For example, a circular }~lasmid or DNA fragment preferably employs a single targeting sequence. A linear plasmid or DNA fragment preferably em~ploys two targeting sequences.
10 The targeting sequence or sequences can, ;nflf~r~n~l~ntly, be within the gene of interest (such as, the sequences of an exon and/or intron), ; ';~t~ly adjacent to the gene of interest (i.e., with no additional nucleotides between the targeting s~oqu~nre and the coding region of the gene of 15 interest), upstream gene of interest (such as the seguenc-es of the U~ Le~ non-coding region or ~ ln~_.,...,c~ promot-er sequences), or upstream of and at a distance from the gene (such as, sequences u~"LL~ of the ~ ln~ pro-moter). The targeting sequence or sequences can include 20 those regions of the targeted gene presently known or sequenced and/or regions further upstream which are struc-turally uncharacterized but can be mapped using restric-tion enzymes and determined by one skilled in the art.
As taught herein, gene targeting can be used to 25 insert a regulatory sequence isolated from a different gene, assembled from c isolated from difference cellular and/or viral sources, or synth~;7ed as a novel regulatory sequence by genetic engineering methods within, ';~t~ly adjacent to, upstream, or at a substantial 3û distance from an t:ild~yel~ s cellular gene. Alternatively or additionally, 8~q~ nrP~ which affect the structure or stability of the RNA or protein produced can be replaced, removed, added, or otherwise rn~;f;~d by targeting. For example, RNA stability elements, splice sites, and/or 35 leader sequences of RNA molecules can be modified to 21 9g~28~
Wo95/31560 -17- r~ ?IJ
improve or alter the function, stability, and/or tran61at-ability of an RNA molecule. Protein sequences may also be altered, such as signal sequences, propeptide s~ nt~
active sites, and/or structural sequences for ,onh~nr;n~ or modifying transport, secretion, or functional properties of a protein. According to this method, introduction of the exogenous DNA results in the alteration of the normal expresaion propertiea of a gene and/or the atructural propertiea of a protein or RNA.
10 The Requlatorv Sequenee The regulatory ae~auence of the DNA eonstruct can be comprised of one or more promoters (æuch as a eonstitutive or inducible promoter), enhancers, seaffold-att~' regions or matrix attaehment sites, negative regulatory 15 elements, transcription faetor binding sites, or eombina-tions of said sequenees.
The regulatory sequenee ean eontain an; nrl~lr; hl e promoter, with the result that eells as ~Ludu~ d or as introduced into an individual do not express the product 20 but can be induced to do so (i.e., expression is induced after the transfected cells are produced but before im-rl;lnt;~t;on or after; ~ tnt;on). DNA encoding the desired product can, of course, be introduced into cells in such a manner that it is ~uLe:s~ed upon introduction 2~ (e.g., under a constitutive promoter). The regulatory setauence can be ;~nl~t~d from cellular or viral genomes, (such regulatory sequences include those that regulate the expression of SV40 early or late genes, adenovirus major late genes, the mouse metallothionein- I gene, the elonga-30 tion factor-la gene, cytomegalovirus genes, collagen genes, actin genes,; glnh~ll;n genes or the HMG-CoA
reductase gene). The regulatory se~auence preferably co~-tains transcription factor binding sites, such aa a TATA
Box, CCAAT Bo~c, AP1, Spl or NF-I~B binding aites.

WO95/31560 ~ 7 9 ~ ".,~ ~ r~4~
.

Additional DNA Construct Elements The DNA construct further comprises one or more exon6 . An exon i8 def ined herein as a DNA sequence which is copied into RNA and is present in a mature mRNA mole-5 cule. The exons can, optionally, contain DNA which encodesone or more amino acid6 and/or partially encodes an amino acid (i.e., one or two bases of a codon). Alternatively, the exon contains DNA which corresponds to a 5' non-coding region. Where the exogenous exon or exons encode one or 10 more amino acids and/or a portion of an amino acid, the DNA construct is rlr~i~n~ such that, upon transcription and splicing, the reading frame is in-frame with the second exon or coding region of the targeted gene. As used herein, in-frame means that the encoding sequences of 15 a first exon and a second exon, when fused, join together nucleotides in a manner that does not change the appropri-ate reading frame of the portion of the mRNA derived from the second exon.
Where the f irst exon of the targeted gene contains 20 the sequence ATG to initiate tr~n~ inn, the ~ ..R
exon of the construct preferably contains an ATG and, if required, one or more nucleotides such that the resulting coding region of the mRNA ;nrll-~;nrj the second and subse-quent exons of the targeted gene is in-frame. Examples of 25 such targeted genes in which the first exon contains an ATG include the genes encoding hEPO, hGH, human colony stimulating factor-granulocyte/macrophage (hGM-CSF), and human colony stimulating factor-granulocyte (hG-CSF).
A splice-donor site is a sequence which directs the 30 splicing of one exon to another exon. Typically, the first exon lies 5~ of the second exon, and the splice-donor site overlapping and flanking the first exon on its 3~ side r~rn~n;7~ a splice-acceptor site flanking the second exon on the 5 ~ side of the secoIld exon. Splice-35 donor sites have a characteristic consensus sequence W095/315C0 21 ~ 9 r~

es~nted as: (A/C) AG GURAGU (where R denotes a purinenucleotide) with the GU in the fourth and fifth positions, bei~g required (Jackson, I.J., Nucleic Acid6 .Research 19:
3715-3798 (1991) ) . The first three bases of the splice-5 donor consensus site are the last three bases of the exon.Splice-donor sites are f~ln~;nn~ly defined by their ability to effect the appropriate reaction within the mRNA
6plicing pathway.
An unpaired splice-donor site is defined herein as a 10 splice-donor site which is present in a targeting con-struct and is not ~ n;~d in the construct by a splice-acceptor site positioned 3 ' to the unpaired splice-donor site. The unpaired splice-donor site results in splicing to an endogenous splice-acceptor site.
A splice-acceptor site in a sequence which, like a splice-donor site, directs the splicing of one exon to another exon. Acting in conjunction with a splice-donor site, the splicing apparatus uses a splice-acceptor site to effect the removal of an intron. Splice-acceptor sites 20 have a characteristic se5~uence lt:yLe:s~l-Led as: rYYYYYYYYY-NYAG, where Y denotes any pyrimidine and N denotes any nucleotide (Jackson, T,~J,, Nucleic Acids 1i!esearch 19:3715-3798 (1991) ) .
An intron is def ined as a sequence of one or more 25 nucleotides lying between two exons and which is removed, by splicing, from a ~L~ u- ~ 0~ RNA le~ in the forma-tion of an mRNA, 1 ec~
The regulatory sequence is, for example, operatively linked to an ATG start codon, which initiate8 tr~nRl~t;nn.
30 Optionally, a CAP site (a specific mRNA initiation site which is ~R~or;~ted with and l~;l;7e~ by the regulatory region) is operatively linked to the regulatory se~auence and the ATG start codon. Alternatively, the CAP site associated with and llt;1;7P~ by the regulatory se~uence is 35 not ;nnlll~ in the targeting construct, and the trans-Wo 95131560 ~ 1 ~ O ~ ~7 , ~".,~ " ~^ ~5 .

criptional apparatu8 will def ine a new CAP site . For most genes, a CAP 8ite is usually found approximately Z5 nucle-otides 3' of the TATA box. In one embodiment, the splice-donor site is placed; ~ ly adjacent to the ATG, for 5 example, where the presence of one or more nucleotides is not re~Iuired f or the exogenous exon to be in- f rame with the second exon of the targeted gene. Preferably, DNA
encoding one or more amino acids or portions of an amino acid in-frame with the coding seguence of the targeted 10 gene, is placed i ~;~t~oly adjacent to the ATG on its 3' side . In such an ~ ; r- ', the splice-donor site is placed; ~ t~ y adjacent to the encoding DNA on its 3 ' side .
Operatively linked or flln~ir~n~lly placed i5 defined 15 as a conf iguration in which the exogenous regulatory se5Iuence, exon, splice-donor site and, optionally, a se-quence and splice-acceptor site are appropriately targeted at a position relative to an endogenous gene such that the regulatory element directs the production of a primary RNA
20 transcript which initiates at a CAP site ~optionally included in the targeting construct) and includes seS~uen-ces corr~rlp~n~lin~ to the exon and splice-donor site of the targeting construct, DNA lying upstream of the endogenous gene~s regulatory region (if pregent), the .~ ., ..,e 25 gene's regulatory region (if present), the el~doy~ uus genes 5 ' nontranscribed region (if pre8ent), and exons and introns (if present) of the ~I~dGye luus gene. In an opera-tively linked configuration the splice-donor site of the targeting ,uuDLluc:L directs a splicing event to a splice-30 acceptor site flanking one of the exons of the endogenousgene, such that a desired protein can be produced from the fully spliced mature transcript. In one: ' '; , the splice-acceptor site is ~ d~y~lluus, such that the splicing event is directed to an t:l~duye:lluus exon, for example, of 35 the endogenous gene. In another embodiment where the WO95/31560 21 90289 ,~ S;~^ l5 .

splice-acceptor site is included in the targeting con-struct, the splicing event removes the intron introduced by the targeting construct.
The encoding DNA (e.g., in exon l of the targeting 5 construct) employed can optionally encode one or more amino acids, and/or a portion of an amino acid, which are the same a6 those of the ~ , q protein . The enco-ding DNA ser~uence employed herein can, for example, corre-spond to the f irst exon of the gene of interest . The -nr~7;n~ DNA can alternatively encode one or more amino acids or a portion of an amino acid different from the first exon of the protein of interest. Such an; ' -'; t is of particular interest where the amino acids of the first exon of the protein of interest are not critical to the activity or activities of the protein. For example, when fusions to the endogenous h73PO gene are constructed, sequences encoding the first exon of hG;I can be employed.
In this example, fusion of hG.~ exon l to h73PO exon 2 results in the ft-rr-7-;-7n of a hybrid signal peptide which is fl7nrt;nn;71 In related constructs, any exon of human or non-human origin in which the encoded amino acids do not prevent the function of the hybrid signal peptide can be used. In a related: ' _'; , this technique can also be employed to correct a, 77-;nn found in a target gene.
Where the desired product is a fusion protein of the c:l.doy~ ,us protein and encoding se~r,uences in the targeting construct, the ~ Jye~ U5 encoding DNA incorporated into the cells by the present method include6 DNA which encodes one or more exona or a se~auence of cDNA ~orr~pnn~7;nr to a tr;7n~1;77-;nn or transcription product which is to be fused to the product of the ~:llduyt~ us targeted gene. In this embodiment, targeting is used to prepare chimeric or multifunctional proteins which corr~7ine structural, enzy-matic, or ligand or receptor binding properties from two or more proteins into one polypeptide. For example, the .

21 qO289 WO 95/31~60 exogenous DNA can encode an anchor to the membrane for the targeted protein or a signal peptide to provide or improve cellular 6ecretion, leader sequences, enzymatic regions, tr~n~ e domain regions, co-factor binding regions or 5 other functional regions. Examples of proteins which are not normally 6ecreted, but which could be fused to a signal protein to provide gecretion include dopa-~P~rhn~r-ylase, transcriptional regulatory protein6, ~-galactosi-dase and tyrosine hydroxylase.
Where the first exon of the targeted gene corre6ponds to a non-coding region (for example, the first exon of the follicle-stimulating hormone beta (FSH~) gene, an exoge-nous ATG is not required and, preferably, is omitted.
The DNA of the construct can be obtained from sources 1~ in which it occurs in nature or can be produced, using genetic engineering techniques or synthetic ~ u~es~es.
The Tar~eted Gene ;~n~ Resl~lt-nn Product The DNA construct, when transfected into cells, such as primary, secondary or immortalized cells, can control 20 the expression of a desired product for example, the active or, functional portion of the protein or RNA. The product can be, for example, a hormone, a cytokine, an antigen, an antibody, an enzyme, a clotting factor, a transport protein, a receptor, a regulatory protein, a 2~ structural protein, a transcription factor, an anti-sense RNA, or a ribozyme. Additionally, the product can be a protein or a nucleic acid which does not occur in nature (i.e., a ~usion protein or nucleic acid).
The method as described herein can produce one or 30 more t~r~r~llt-c products, such as erythropoietin, calci-tonin, growth hormone, insulin, insulinotropin, insulin-like growth factors, parathyroid hormone, interferon ~, ~
and interferon ,B, nerve growth factors, FSH~, TGF-,~, tumor necrosis factor, glucagon, bone growth factPr-2, bone .

Wo 95/31560 ~ 2 ~ 9~289 growth factor-7, TSH-~, interleukin 1, interleukin 2, interleukin 3, interleukin 6, interleukin 11, interleukin 12, CSF-granulocyte, CSF-macrophage, CSF-granulocyte/
macrophage,; nn~lobulins, catalytic ~nt;h~;es, protein 5 kinase C, glucocerebrosidase, superoxide dismutase, tissue rlArY~;nngen activator, urokinase, antithrombin III, DNAse, ~Y-galactosidase, tyrosine hydroYylase, blood clotting factors V, blood clotting factor VII, blood clotting factor VIII, blood clotting factor IX, blood clotting lO factor X, blood clotting factor XIII, ~rol irorrotein E or apolipoprotein A-I, globins, low density lipoprotein receptor, II,-2 receptor, I~-2 ~ntArJoniRts, alpha-1 anti-trypsin, immune response modifiers, and soluble CD4.
Selectable Markers and I l i f ication The ;.lPnt;f;CAt;nn of the targeting event can be facilitated by the use of one or more RPl e~t~hl e marker genes . These markers can be ; nrl l~Pd in the targting construct or be present on different constructs. Select-able markers can be divided into two categories: posi-20 tively selectable and negatively selectable ~in other words, markers for either positive sPlPct;nn or negative selection) . In positive #Pl P~t; nn, cells expressing the positively selectable marker are capable of surviving treatment with a gelective agent (guch ag neo, Y:~ntl~;nP-25 guanine phosphoribo~yl trangfera8e (gpt), dhfr, A-lPnnR;nP
lP~m; n:~Re (ada), puromycin (pac), 1-yy ~ in ~hyg), CAD
which encodes carbamyl r~rRph~te synthase, aspartate lase~ and dihydro-orotase rJlu~m;np synthPt~e (GS), multidrug resistance 1 (mdrl) and histidine D
30 (hisD), allowing for the selection of cells in which the targeting construct integrated into the host cell genome.
- In negative selection, cells expressing the negatively selectable marker are destroyed in the presence of the selective agent. The ;~lPnt;f;r~t;nn of the targeting WO 95/31560 ~ Q 2 8 ~ 4' event can be iacilitated by the use of one or more marker genes exhibiting the property of negative selection, such that the negatively selectable marker is linked to the exogenous DNA, but configured such that the negatively 5 6electable marker flanks the targeting qPquPn~e, and such that a correct homologous recombination event with se-quencefi in the host cell genome does not result in the stable integration of the negatively selectable marker (Mansour, S.L. et j~., Nature 336:348-352 (1988) ) . Mark ers useful for this purpose include the Herpes Simplex Virus thymidine kinase (TK) gene or the bacterial gpt gene .
A variety of selectable markers can be incorjporated into primary, secondary or immortalized cells. For exam-15 ple, a selectable marker which confers a selectable pheno-type such as drug resistance, nutritional auxotrophy, resistance to a cytotoxic agent or expression of a surface protein, can be used. Selectable marker genes which can be used include neo, gpt, dhfr, ada, pac, hyg, Ci~D, GS, 20 mdrl and hisD. The sPlp~t~hlp phenotype conferred makes it possible to identify and isolate recipient cells.
r l;fiF~hle geneg encoding gelectable markers (e.g., ada, GS, dhfr and the multiflln~t;t~ni7l CaD gene) have the added characteristic that they enable the selection of 25 cells cr~nti~;nin~ amplified copies of the gelectable marker inserted into the genome. This feature provides a mecha-nism for significantly increasing the copy number of an adjacent or linked gene for which i l;fir~t;rm is desir-able. Mutated versions of these Eequences showing im-3 o proved selection properties and other ,l; f ; i7hl e sequenc-es can also be used.
The order of I c in the DNA construct can vary. Where the construct is a circular plasmid, the order of elements in the resulting structure can be:
35 targeting sequence - plasmid DNA ~compriged of c~ lPnr-Pc Wo 95/31560 ~ 5 2 1 ~0~

used for the selection and/or replication of the targeting plasmid in a microbial or other suitable host~ - select-able marker(s) - regulatory sequence - exon - splice-donor site. Preferably, the plasmid rnnt~;n;ng the targeting 5 sequence and exogenous DNA elements is cleaved with a restriction enzyme that cuts one or more times within the targeting sequence to create a linear or gapped molecule prior to intrnflllntinn into a recipient cell, such that the free D~A ends increase the frequency of the desired homol-10 ogous r~ ' - n~tion event as described herein. In addi-tion, the free DNA ends may be treated with an F-Ynn~ qe to create protruding 5~ or 3' ov~h~n~;ng single-stranded DNA ends to increase the frequency of the desired homolo-gous re~ ` ;n~t;nn event. In this ' -~fl; , homologous 15 ~ ~ '~;n~tion between the targeting sequence and the cellular target will result in two copies of the targeting sequences, flanking the elements cnn~;nefl within the introduced plasmid.
Where the construct is linear, the order can be, for 20 example: a first targeting sequence - sF~ ct~hl ~ marker -regulatory sequence - an exon - a 3plice-donor site - a second targeting sequence or, in the alternative, a first targeting sequence - regulatory sequence - an exon - a splice-donor site - DNA encoding a selectable marker - a 25 second targeting 8~ n~ e. Cells that stably integrate the construct will survive treatment with the selective agent; a subset of the stably transfected cells will be ~- lOgOllqly L- ' -n:~nt cellg. The homologously recombi-nant cells can be ;fl~n~;f;er~ by a variety of techniques, 30 including PCR, Southern hybr;fl;7~tinn and phenotypic screening .
In another; ' -~;r ', the order of the construct can be: a f irst targeting sequence - selectable marker regulatory sequence - an exon - a splice-donor site - an 35 intron - a splice-acceptor site - a second targeting sequence.

21 90~8q WO 95/3 1560 2 6 r c I, ~, s c Alternatively, the order of ~ nn~ntF: in the DNA
construct can be, for example: a first targeting se auence - selectable marker l - regulatory se~uence - an exon - a splice-donor site - a second targeting sequence - select-able marker 2, or, alternatively, a first targeting se-cluence - regulatory se~uence - an exon - a splice-donor site - selectable marker l - a second targeting se~uence -selectable marker 2. In this: ' '; selectable marker 2 displays the property of negative selection That is, the gene product of selectable marker 2 can be selected against by growth in an appropriate media fnrr~ ,t;nn r~nntA;n;ng an agent (typically a drug or metabolite ana-log) which kills cells expressing selectable marker 2.
Recombination between the targeting sequences flanking selectable marker l with homologous sequences in the host cell genome results in the targeted integration of select-able marker l, while selectable marker 2 is not integrat-ed. Such ,~ ' ;nAt;o~ events generate cells which are stably transfected with ~l.QrtAhle marker l but not stably transfected with s,olet~t~hle marker 2, and such cells can be selected for by growth in the media cnntA;n;n~ the selective agent which selects for 8~ol ectAh~ e marker l and the selective agent which selects against selectable marker 2.
The DNA construct also can include a positively selectable marker that allows for the s,ole~t;nn of cells ~nntA;n;n~ amplified copies of that marker. The amplifi-cation of such a marker results in the co-amplification of flAnk;ng DNA seS~uences. In this: ' '; -, the order of construct . _ -ntR is, for example: a first targeting sequence - an: ~1; f; Ahl e positively selectable marker - a second selectable marker (optional) - regulatory se~auence - an exon - a splice-donor site - a second tar-geting DNA seSIuence.

21 qO2~9 WO95/31560 i~~ 15 --2~--In this: ' ir t, the activated gene can be further amplified by the inclusion of a selprt-~hle marker gene which has the property that cells rnnt~;n;n~ amplified copies of the selectable marker gene can be selected for 5 by culturing the cells in the presence of the appropriate selectable agent . The activated Pnrl~Pn~ uC gene will be amplified in tandem with the amplified selectable marker gene . Cells crnt~; ni n~ many copies of the activated endogenous gene may produce very high levels of the de-10 sired protein and are useful for ~.a vitro protein produc-tion and gene therapy.
In any: ' - '; t, the gelectable and 1; f; ~hle marker genes do not have to lie immediately adj acent to each other.
Optionally, the DNA construct can include a bacterial origin of replication and bacterial antibiotic resistance markers or other selectable markers, which allow for large-scale plasmid pror~ti-~n in b~rtpr;~ or any other suitable cloning/host system. A DNA construct which 20 i nr~ PA DNA encoding a ~Pl ~rt, Ahle marker, along with additional se~u~.lces, such as a promoter, and splice junctions, can be used to confer a splect~hle phenotype upon transfected cells (e.g., plasmid pcDNBO, schematical-ly represented in Fiqure 4 ) . Such a DNA construct can be 25 co-transfected into primary or secondary cells, along with a targeting DNA sequence, using methods described herein.
Tr;~n~fection :~lnrl Homoloqous Recombination According to the present method, the construct is introduced into the cell, such as a primary, secondary, or 3 0 immortalized cell, ag a single DNA construct, or as sepa -rate DNA sequences which become incorporated into the C~lL- 1 or nuclear DNA of a transfected cell The targeting DNA construct, including the targeting sequences, regulatory sequence, an exon, a splice-donor WO 95/31560 ~ S '~4~
-28- 21 9~289 site and selectable marker gene(s), can be introduced into cells on a single DNA construct or on separate constructs.
The total length of the DNA construct will vary according to the number of on~ntc (targeting sequences, regula-5 tory sequences, exons, selectable marker gene, and otherelements, for example) and the length of eaeh. The entire construct length will g~nPr~l l y be at least about 200 nucleotides. Further, the DNA can be iilLluduced as lin-ear, double-stranded (with or without single-stranded 10 regions at one or both ends), single-stranded, or circular .
Any of the construct types of the ~1; crlo~c~ invention is then introduced into the cell to obtain a transf ected cell. The transfected cell is --;nt~;nP~ under conditions 15 which permit homologous re~ ' ;n~t;nn, as is known in the art (Capecehi, M.R., Science 244:1288-1292 ~19a9)). When the homologously l. ' ;n~nt eell is --;nt~;n~l under eonditions sllff;~;~nt for transeription of the DNA, the regulatory region introdueed by the targeting eonstruet, 20 as in the case of a promoter, will activate transcription.
The DNA construets may be ; ntr~ into cells by a variety of physical or chemical methods, ;nt lll~1;n~ elec-troporation, microinjection, microprojectile ~ ' , , calcium rhncrh~t~ precipitation, and liposome-, poly-25 brene-, or DEAE dextran-r '; ~ted transfection. Alter-natively, infectious vectors, such as retroviral, herpes, adenovirus, adenovirus-associated, mumps and poliovirus vectors, can be used to introduce the DNA.
Optionally, the targeting DNA can be introduced into 3 0 a cell in two or more separate DNA ~, _ c In the event two _ragments are used, the two fragments share DNA
sequence homology (overlap) at the 3' end of one fragment and the 5~ end of the other, while one carries a first targeting sequence and the other carries a second target-35 ing sequence. IJpon introduction into a cell, the two WO95/31560 ~ 8q I~J~ 5,~t~

CL~. tc can underqo homologous recombination to form a 6ingle f ragment with the f irst and second targeting se-quences flanking the region of overlap between the two original fl _ ~. The product fragment is then in a 5 form suitable or ~ ollc recombination with the cellular targe~ sequences. More than two f _ R can be used, de6igned such that they will undergo homologous re ' ;nAti~n with each other to ultimately form a product suitable for I -lO~Oll~ re- ' ;nAtinn with the cellular 10 target sequences as described above.
The Homoloqouslv Recombinant Cells The targeting event results in the insertion of the regulatory sequence of the targeting construct, placing the endogenous gene under their control (for example, by 15 insertion of either a promoter or an enhancer, or both, upstream of the elldoy~ uus gene or regulatory region).
Optionally, the targeting event can simultaneously result in the deletion of the ~ ",c regulatory element, such as the deletion of a tissue-Rreci ~i ~ negative regulatory 20 element. The targeting event can replace an existing element; for example, a tissue- specific enhancer can be replaced by an en~ancer that has broader or different cell-type spe~ ;ty than the naturally-occurring ele-ments, or digplayg a pattern of regulation or ;n~ll,t;r,n 25 that is different from the C~LL- ~l.."..l;n~ nontransfected cell. In this '; - t the naturally or~rr;n~ sequenc-es are deleted and new se~r~rn~ r~ are added. Alternative-ly, the endoye..uus regulatory elements are not removed or replaced but are disrupted of disabled by the targeting 30 event, such as by targeting the exogenous sPqn~n-~eR within the t:llduye~ ,us regulatory elements.
Af ter the DNA is introduced into the cell, the cell is ~-;nt~;n~d under condltions appropriate for homologous recombination to occur between the genomic DNA and a WO95~11560 ~ a28~ Q4' .

portion of the introduced DNA, as is known in the art (Capecchi, M.~., Science 244:1288-1292 (1989)).
~omologous recombination between the genomic DNA and the introduced DNA results in a homologously ~rt_ ' inAnt 5 cell, such as a ~ungal, plant or animal, and particularly, primary, secondary, or immortalized human or other mamma-lian cell in which sprl~l~onrDc which alter the expression of an ~n~lnrj~nr,~lc gene are operatively linked to an ~ J~ C
gene eneoding a product, producing a new transcription 10 unit with expression and/or eoding potential that is different from that of the endogenous gene. Particularly, the invention ;nr~ c a homologously r~ ~ ;nAnt eell comprising regulatory se~uences and an exon, flanked by a splice-donor site, which are introduced at a predetermined 15 site by a targeting DNA construet, and are operatively linked to the seeond exon of an el~doye~ us gene. Option-ally, there may be multiple ~Yn~nmlC exons (eoding or non-coding) and introns operatively linked to any exon of the ~nrlnJ~nmlc gene. The regulting 1 ~loJollcly reeombi-20 nant eeIls are eultured under conditions which seleet foramplifieation, if appropriate, of the DNA e~coding the 1 i fi Ahl e marker and the novel transcriptional unit .
With or without amplification, cells produced by this method can be cultured under eonditions, as are known in 25 the art, suitable for the expression of the protein, thereby produeing the protein n vitro, or the eells ean be used for vivo delivery of a th~rAre~lt; r protein ( i . e ., gene therapy) .
As used herein, the term primary eell includes cells 30 present in a suspension of cells isolated from a verte-brate tissue source (prior to their being plated, i.e., attached to a tigsue culture gubstrate such as a dish or flask), cells present in an eYplant derived from tissue, both of the previous types of cell8 plated for the first 35 time, and cell sugpensions derived from these plated W095/31i560 2 1 ~ ~89 cells. The term secondary cell or cell straln refers to cells at all subsequent steps in culturing. That is, the first time a plated primary cell i8 removed from the culture substrate and replated (passaged1, it is referred 5 to herein as a secondary cell, as are all cells in subse-quent passages. Secondary cells are cell strains which consist of secondary cells which have been r~c8p~rl one or more times. A cell strain consists of secondary cells that: l) have been p~s~ed one or more times; 2) exhibit 10 a finite number of mean pop~ t;nn tlnuhl;nrJ~ in culture;
3) exhibit the properties of contact-inhibited, anchorage dependent growth (anchorage-dependence does not apply to cells that are propagated in suspension culture); and 4) are not immortalized.
Immortalized cells are cell lines (as opposed to cell strains with the ~1.QC; jr~ni~t;nn "strain" Leselved for primary and secondary cells), a critical feature of which is that they exhibit an apparently unlimited lifespan in culture.
Cells selected for the subject method can fall into 20 four types or categories: l) cells which do not, as ob-tained, make or contain the protein or product (such as a protein that is not normally e~essed by the cell or a fusion protein not normally found in nature), 2) cells which make or contain the protein or product but in quan-25 tities other than that desired (such as, in quantitiesless than the phys;nloJriri~lly normal lower level for the cell as it is obtained), 3 ) cells which make the protein or product at physiologically normal levels for the cell as it is obtained, but are to be i _ Pd or ~nhilnrPrl in 30 their content or prr"lllrt;nn~ and 4) cells in which it i5 desirable to change the pattern of regulation or induction of a gene encoding a protein.
Primary, secondary and immortalized cells to be transfected by the present method can be obtained from a 3~ variety of tissues and include all cell types which can be WO 9513156~ 4' ~-;ntA;n~f~ in culture. For example, primary and secondary cells which can be transfected by the present method include fibroblasts, keratinocytes, epithelial cell6 (e.g., mammary epithelial cells, intestinal epithelial 5 cells), endothelial cells, glial cells, neural cells, formed elements of the blood (e.g., lymphocytes, bone marrow cells), muscle cells and ~Le~:uL~u~ o~ these somat-ic cell types. Where the ~ ln~ollqly ~. ;n~nt cellS
are to be used in gene therapy, primary cells are prefera-10 bly obtained from the individual to whom the transfectedprimary or secondary cells are administered. However, primary cells can be obtained from a donor (other than the recipient) of the same species.
Homologously recombinant immortalized cells can also 15 be produced by the present method and used for either protein prn~ t t;nn or gene therapy. Examples of immortal-ized human cell lines useful for protein production or gene therapy by the present method include, but are not limited to, HT1080 cells (ATCC CC~ 121), He~a cells and 20 derivatives of HeLa cells (ATCC CCI, 2, 2.1 and 2.2), MCF-7 breast cancer cells (ATCC BTH 22), K-562 lellk~m~A cells (ATCC CC~ 243), XB carcinoma cells (ATCC CCL 17), 2780AD
ovarian carcinoma cells (Van der Blick, A.M. et ~., r~n~r Res. 48 5927-5932 ~1988), Raji cells ~ATCC CC~ 86), 25 Jurkat cells ~ATCC TIB 152), Namalwa cells ~ATCC CRI.
1432), HL-60 cells ~ATCC CCJJ 240), Daudi cells ~ATCC CCI, 213), RPMI 8226 cells ~ATCC CCI, 155), U-937 cells ~ATCC
CRL 1593), Bowes Melanoma cells ~ATCC CR~ 9607), WI-38VA13 subline 2R4 cells ~ATCC CL~ 75.1), and MOLT-4 cells ~ATCC
30 CRL 1582), as well as heterohybridoma cells produced by fusion of human cells and cells of another species.
Secondary human f;hrnhlAct strains, such as WI-38 ~ATCC
CC~ 75) and MRC-5 ~ATCC CC~ 171) may be used. In addi-tion, primary, secondary, or immortalized human cells, as 35 well as pr~mary, secondary, or immortalized cells from WO95131560 21 qo289 r~ Q1~

other species which display the properties of gene ampli-fication in vitro can be used for ~n vitro protein produc-tion or gene therapy.
Method of Convert; n~ a Gene into a cDNA CoPy The present invention also relates to a method by which h~ 1O~Q;1C re ' 1n~t;nn is used to convert a gene into a cDNA copy (a gene copy devoid o introns). The cDNA copy can be transferred into yeast or bacteria for Yitro protein pro~ t; nn, or the cDNA copy can be inserted into a, li~n cell for ;Ln Yitro or ~L}LY~vo protein pro~rtion. If the cDNA is to be transferred to microbial cells, two DNA constructs nnnt~;n;n~ targeting sequences are introduced by homologous " - ' ;n~t;nn, one construct upstream of and one construct downstream of a human gene ~ncnrl;ng a therapeutic protein. For example, the 6equenc-es introduced upstream include DNA sequences h~ - lo~o~1~ to genomic DNA sequences at or upstream of the DNA encoding the first amino acid of a mature, ~ ucessed therapeutic protein; a retroviral long term repeat (LTR); sequences encoding a marker for selection in microbial cells; a regulatory element that functions in microbial cells; and DNA encoding a leader peptide that promotes secretion from microbial cells with a splice-donor site. The sequences introduced upstream are introduced near to and upstream of genomic DNA encoding the first amino acid of a mature, processed therapeutic protein. The 8~ nr~R introduced downstream include DNA 5.~q--~n~ ~A 1- ~1Q~O1~C to genomic DNA
sequences at or downstream of the DNA encoding the last amino acid of a mature, processed protein; a microbial 3 o transcriptional termination sequence; gequences capable of directing DNA replication in microbial cells; and a retro-viral LTR. The gequenceg; nt~n~l11t ed downgtream are intro-duced adj acent to and downstream of the DNA encoding the stop codon of the mature, processed therapeutic protein.

WO 95131560 ~ 2 8 9 I~ IJ
~34--After the two DNA constructs are introduced into cells, the resulting cells are ~~;nt~;nl~-l under conditions appro-priate for homologous re~ ' ;n~t;nn between the introduced DNA and genomic DNA, thereby producing homologously recom-5 binant cells. Optionally, one or both of the DNA con-structs can encode one or more markers for either positive or negative selection of cells rnnt~;n;n~ the DNA con-struct, and a selection step can be added to the method after one or both of the DNA constructs have been intro-10 duced into the cells. Alternatively, the se~ue,~.es encod-ing the marker for :selection in microbial cells and the sequences capable of directing DNA replication in microbi-al cells can both be present in either the upstream or the downstream targeting construct, or the marker f or selec-15 tion in microbial cells can be present in the downstreamtargeting construct and the seS~uences capable of directing DNA replication in microbial cells can be present in the upstream targeting construct . The ' - lo~ouql y recombi-nant cells are then cultured under conditions appropriate 20 for ~TR directed transcription, processing and reverse transcription of the RNA product of the gene encoding the therapeutic protein. The product of reverse transcription is a DNA construct comprising an intronless DNA copy encoding the therapeutic protein, operatively linked to 25 DNA sequences comprising the two exogenous DNA constructs described above. The intronless DNA construct produced by the present method is then introduced into a microbial cell. The microbial cell is then cultured under condi-tions appropriate for expression and secretion of the 3 o therapeutic protein .
In Vivo Protein Production Homologously r~c ' ;n~nt cells of the present inven-tion are useful, as populations of homologously recombi-nant cell lines, as pnr~ t;nnc of homologou~ly recombi-WO 95/31560 2 1 9 0 2 8 9 ~ !45 nant primary or secondary cells, homologously rf-~ ;n~nt clonal cell strains or lines, homologously L~ ' ;n:~nt heterogenous cell strains or lines, and as cell mixtures in which at least one representative cell of one of the 5 four preceding categories of homologously L~ ' in~lnt cells is present. Such cells may be used in a delivery system for treating an individual with an abnormal or undesirable rnnt~; t; nn which responds to delivery of a therapeutic product, which is either: 1) a therapeutic 10 protein (e.g., a protein which is absent, ulldeL~Ludu~ d relative to the individual's physiologic needs, defective or ;n~ff;r;/~ntly or illc-~L~.~Liately ~tt;li7~'~ in the indi-vidual; a protein with novel functions, such as enzymatic or transport fllnrt;nn~) or 2) a th~r~relltic nucleic acid 15 (e . g ., RNA which inhibits gene expression or has intrinsic enzymatic activity). In the method of the present inven-tion of providing a t~.r~rF~llt; r protein or nucleic acid, homologously re~ ' ;n~nt primary cells, clonal cell strains or heterogenous cell strains are administered to 20 an individual in whom the abnormal or u~desirable co~di-tion is to be treated or prevented, in sl-ff;r;~nt quantity and by an appropriate route, to express or make available the protein or ~Yr~nrllc DNA at physiologically relevant levels. A physiologically relevant level is one which 25either approximates the level at which the product is normally ~L~-du.:ed in the body or results in; .,~ of the abnormal or undesirable condition. According to an ' ~'; of the invention described herein, the homo-logously re ' ;n~nt immortalized cell lines to be admin-30 istered can be enclosed in one or more semipermeable barrier devices . The pt:LI -h; l; ty properties of the device are such that the cells are prevented from leaving the device upon ;rnrl~nt~tion into an animal, but the therapeutic product is freely permeable and can leave the 35 barrier device and enter the local space surrounding the W095~l560 ~1 ~0~ r~l" '~~ ~45 implant or enter the systemic circulation. For example, hGH, hEP0, human insulinotropin, hGM-CSF, hG-CSF, human ~-interferon, or human FSH~ can be delivered systemically in humans for therapeutic benefits.
Barrier devices are particularly useful and allow homologously re~ ;n~nt immortalized cells, ~ lor~o1l~ly rel ' ;n~nt ceils from another species (homologously ;nAnt xenogeneic cellg), or cells from a nonhisto-t; h; l; ty-matched donor (homologously re~ '; n:!lnt allogeneic cells) to be i ,1 ~nted for treatment of human or animal conditions or for agricultural uses (i.e., meat and dairy production). Barrier devices also allow conve-nient short-term (i.e., transient) therapy by providing ready access to the cells for removal when the treatment regimen is to be halted for any reason.
A number of synthetic, semisynthetic, or natural filtration membranes can be used for this purpose, includ-ing, but not limited to, cellulose, cellulose acetate, nitrocellulose, polysulfone, polyvinylidene difluoride, polyvinyl chloride polymers and polymers of polyvinyl chloride derivatives. Barrier devices can be utilized to allow primary, secondary, or immortalized cells from another species to be used for gene therapy in humans.
In Vitro Protein Pro~rtirn M lr~roll~ly r~: ' ;nAnt cells from human or non-human species according to this invention can also be used for n vitro protein pro~l11rt;r~n. The cells are r~;nt~;nPr~
under conditions, as are known in the art, which result in expression of the protein. Proteins expressed using the methods described may be purified from cell lysates or cell s1lr~r"~t~nt~ in order to purify the desired protein.
Proteins made according to this method include therapeutic protei~s which can be delivered to a human or non-human animal by conventional pharmaceutical routes as is known WO95/31560 2 1 ~ ~ ~ 8 ~

in the art (e.g., oral, intravenous, intL ~c~ r, intra-nasal or s~hcut~n~n~) . Such proteins include hGH, hEPO, and human insulinotropin, hGM-CSF, hG-CSF, FSH~ or ~-;nt~rforon These cell6 can be immortalized, primary, or 5 secondary cells. The u6e of cells frbm other species may be desirable in cases where the non-human cells are advan-tageous for protein production purpose6 where the non-human protein is tho~reutir~lly or commercially useful, for example, the use of cells derived from salmon for the lO production of salmon calcitonin, the use of cellæ derived from pigs for~the pro~illrtinn of porcine insulin, and the use of bovine cells for the pro~ rt;nn of bovine growth hormone .
Advanta~es lS The r-thn-lnlr~ies, DNA constructs, cells, and resul-ting proteins of the invention herein possess versatility and many other advantages over processes currently em-ployed within the art in gene targeting. The ability to activate an Pn~ln~onmlc gene by positioning an oYn~onm 20 regulatory se5~uence at various positions ranging from ';~toly adjacent to the gene of interest (directly fused to the normal gene's transcribed region) to 30 kilobase pairs or further upstream of the transcribed region of an ~lldoy~ll.,us gene, or within an intron of an 25 endogenous gene, is adv~nt~gonuc for gene expression in cells. For example, it can be employed to pn~it;nn the regulatory element U~ L~::O,III or downstream of regions that normally silence or negatively regulate a gene. The positioning of a regulatory element upstream or downstream 30 of such a region can override such dominant negative effects that normally inhibit transcription. In addition, regions of DNA that normally inhibit transcription or have an otherwise detrimental ef f ect on the expression of a WO 95131560 . ~
~1 ~02gq ~

gene may be deleted using the targeting constructs, des-cribed herein.
Additionally, since promoter function is known to depend strongly on the local environment, a wide range of 5 positions may be explored in order to find those local environments optimal for fllnrt;~n However, since, ATG
start codons are found frequently within 1; ~n DNA
(approximately one occurrence per 48 base pairs~, tran-scription cannot simply initiate at any position upstream 10 of a gene and produce a transcript rnnt~;n;nrJ a long leader seguence preceding the correct ATG start codon, since the frequent UC-,UL ' ellCe of ATG codons in such a leader se~uence will prevent translation of the correct gene product and render the message useless. Thus, the 15 incorporation of an ~ u~ exon, a splice-donor site, and, optionally, an intron and a splice-acceptor site into targeting constructs comprising a regulatory region allows gene expression to be optimized by identifying the optimal site for regulatory region function, without the limita-20 tion imposed by needing to avoid inappropriate ATG startcodons in the mRNA produced. This provides significantly increased f lexibility in the rl ~: of the construct and makes it possible to activate a wider range of genes.
The DNA constructs of the present invention are also 25 useful, for example, in processes for making fusion pro-teinS encoded by rP: ~ 1n:1nt, or exogenous, setauences and q sequences.
Gene targeting and amplification as ~l;qrlo5e~l above are particularly useful for altering on the expression of 30 genes which form transcription units which are sufficient-ly large that they are diffisult to isolate and express, or for turning on genes for which the entire protein coding region is unavailable or has not been cloned.
Thus, the DNA constructs described above are useful for 35 operatively linking exogenous regulatory elements to WO 95/31~60 2 t 9 ~ 2 8 9 1 ~1, . t,~ l5 endogenous genes in a way that preci6ely defines the tran6criptional unit, provides flexibility in the relative positioning of exogeneous regulatory elements and endoge-nous genes ~llt;r-t.oly, enables a highly controlled system 5 for nhtAin-nr, and regulating expression of genes of thera-peutic interest.
, nl AnAtion of the rx le6 As described herein, ~rrl ;rAntq have ' LLe.ted that DNA can be introduced into cells, such as primary, 10 secondary or immortalized vertebrate cells and integrated into the genome of the transf ected cells by homologous L~ _ ' ;n::lt jnn. They have further demonstrated that the exogenous DNA has the desired function in the homologously L~ ~ ;nAnt (~R) cells and that correctly targeted cells 15 can be ; fl~nt; f; ed on the basis of a detectable phenotype conferred by the properly targeted DNA.
z~ppl ;rAntq describe construction of a plasmid useful for targeting to a particular locus tthe ~PRT locus) in the human genome and selection based upon a drug resistant 20 phenotype (Example la). This plasmid is designated pE3Neo and its integration into the cellular genome at the ~PRT
locus produces cells which have an hprt~, 6-TG resistant phenotype and are also G418 resistant. As described, they have shown that pE3~eo functions properly in gene target-2~ ing in an estAhl; Rh~d human fibroblast cell line (Examplelb), by demonstrating lorAl;7At;nn of the DNA introduced into estAhl; qh~d cells within exon 3 of the ~PRT gene.
In addition, ~rr~; rAntq ~ LL.~te gene targeting in primary and secondary human skin f ibroblasts using pE3Neo 30 (Example lc). The subject application further demon-strates that ';~;r~t;nn of DNA termini enhances target-ing of DNA into genomic DNA (Examples lc and le).
Applicants also describe methods by which a gene can be inserted at a preselected site in the genome of a cell, WO 95131560 ~ ~^ l5 such as a primary, secondary, or immortalized cell by gene targeting (Example ld) .
In addition, the present invention relates to a method of protein pro~ rt;nn using transfected cells. The 5 method involves transfecting cells, such as primary cells, secondary cells or immortalized cells, with exogenous DNA
which encodes a therapeutic product or with DNA which is auf f icient to target to an , ,.~ c gene which encodes a therapeutic product. For example, Examples lg, lh, lj, 10 lk, 2, 3, 4 and 6 - 9 describe protein production by targeting of a selected ~n~lnrJ~nmlq gene with DNA sequence elements which will alter the expression of the elliuyélluuS
gene .
~rpl;rAntq also describe DNA constructs and methods 15 for amplifying an ~lldoy~:lluus cellular gene that has been activated by gene targeting (Bxamples 3, 6, 8 and 9).
r _l~q lf-lh, 2, 4 and 6 illustrate ~ q in which the normal regulatory ser~uences upstream of the human EPO gene are altered to allow expression of hBPû in 20 primary or secondary fibroblast strains which do not express EPO in detectable quantities in their untrans-fected state. In one: ' '; the product of targeting leaves the normal EPO protein intact, but under the con-trol of the mouse metallothionein promoter. Bxamples li 25 and 1~ demonstrate the use of similar targeting constructs to activate the ~::,ldo~ -uus growth hormone gene in primary or secondary human f ibroblast6 . In other embodiments ~qrr;hG~ for activating EPO expression in human fibro-blasts, the products of targeting events are chimeric 30 transcription units, in which the first exon of the human growth hormone gene is positioned upstream of EPO exons 2-5. ~he product of transcription (controlled by the mouse metallothionein promoter), splicing, and tr~nal ~t; nn is a protein in which amino acids 1-4 of the hBPO signal pep-35 tide are replaced with amino acid residues 1-3 of hGH.

WO95/31560 21 9028~ r .~
The chimeric portion of this protein, the signal peptide, is removed prior to secretion from cells. Example 5 describes targeting constructs and methods for producing cells which will convert a gene (with introns) into an 5 expressible cDNA copy of that gene (without introns) and the recovery of such expressible cDNA molecules in micro-bial (e.g., yeast or bacterial) cell6. Example 6 de-scribes construction of a targeting vector, designated pREPO4 for dual selection and selection of cells in which 10 the dhfr gene is amplified. Plasmid pREP04 has been used to amplify the human EP0 (hEPO) locus in HT1080 cells (an immortalized human cell line) af ter activation of the f~n~rr~nrllc hEPO gene by homologous ,~ ' ;n~tinn As described, stepwise selection in methotrexate-rr,nt:q;n;nr 15 media resulted in a 70-fold increa6e in hEP0 production in cells resistant to 0 . 4 ~M methotrexate .
r ~ c 7 and 8 degcribe methods for inserting a regulatory s~r~ nre upstream of the normal EP0 ~ r and methods for EP0 pror;llrt;nn using such a construct. In addition, Example 8 describes the ~l;f;r~t;r~n of a targeted EPO gene produced by the method of Example 7.
Example 9 describes methods for targeting the human ct-interferon, GM-CSF, G-CSF, and FSH~ genes to create cells useful for in protein pro-l-lrt;--n.
The Examples provide methods for activating or for activating and amplifying endogenous genes by gene target-ing which do not reSluire ~n;r~ ;rn or other uses of the target genes ' protein coding regions . Using the methods and DNA constructs or plasmids taught herein or r '; ~; c~-3 0 tions thereof which are apparent to one of ordinary skill in the art, gene expression can be altered in cellæ that have properties desirable for 'n vitro protein production (e.g., pharm-re~t;ra) or vivo protein delivery methods ~e.g. gene therapy). Figures 5 and 6 illustrate two strategies for transcriptionally activating the hEP0 gene.
.

WO 95/315611 2 7 ~ O ~ 8 9 ~ ;r 1~;

Using the methods and DNA constructs or plasmids taught herein or modifications thereDf which are apparent to one of ordinary skill in the art, exogenous DNA which encodes a therapeutic product (e.g., protein, ribozyme, 5 anti-sense RNA) can be inserted at preselected sites in the genome of vertebrate (e.g., liAn, both human and nr~nl ) primary or secondary cells.
The present invention will now be illustrated by the following examples, which are not int~n~ to be limiting 10 in any way.

WO 95~1~60 1 ~ll~,, 5: 5~45 21 9~289 EX~MPLES
EXAMPLT; 1. PROD~CTION OF TR~N~ L~ r~r,T STRAINS 3Y GENE
T~Rr.~TTNr.
Gene targeting occurs when transfecting DNA either 5 integrates into or partially replaces ~:hL~ - 1 DNA
se~uences through a hn~-lo~o~ ; n:~nt event . While such events can occur in the couree of any given transfec-tion experiment, they are usually masked by a vast excess of events in which plasmid DNA integrates by nonhomolo-10 gous, or illegitimate, r~ ' inAt;nn a . C~N~R~TION OF A WN~il~U~_ ~ IJS13Fi;JL FOR SELECTION OF
('.T.'r~ T~Rr.T'TING EVENTS IN HU~N CEI-I.S
One approach to selecting the targeted events is by genetic selection for the loss of a gene function due to 15 the integration of transfecting DNA. The human HPRT locus encodes the enzyme hyp~ nthin~-phclD-uhul ibosyl transfer-ase. hprt~ cells can be selected for by growth in medium c~nt~;n;n~ the rl~rleosi~ analog 6-thioguanine (6-TG):
cells with the wild-type (HPRT+) allele are killed by 20 6-TG, while cells with mutant ~hprt~) alleles can survive.
Cells harboring targeted events which disrupt HPRT gene function are therefore sf~ ct~hle in 6-TG medium.
To construct a plasmid for targeting to the HPRT
locus, the 6.9 kb HindIII fragment extending from posi-25 tions 11,960-18,869 in the HPRT sequence (Genebank name ~u.~ ; Edwards, A. et al., Genomics :593-608 (1990) ) and including exong 2 and 3 of the HPRT gene, is subcloned into the HindIII site of pUC12. The resulting clone is cleaved at the unique XhoI site in exon 3 of the HPRT gene 30 ~r~qS ' and the 1.1 kb SalI-XhoI fragment t~r,nt:l;n;n~ the neo gene from pMClNeo (Stratagene) is inserted, disrupting the coding sequence of exon 3. One orientation, with the direction of neo transcription oppoeite that of HPRT
transcription was chosen and designated pE3Neo The WD95/31560 21 9 02 8q ~ 45 replacement of the normal HPRT eYon 3 with the neo-disrup-ted version will result in an hprt~, 6-TG resistant pheno-type. Such cells will also be G418 resistant.
b. GENE T~r-~TING IN AN EsTA~r rc~ H~AN FIBROBLAST
r~T ~, LINE
As a demonstration of targeting in immortalized cell lines, and to establish that pE3Neo functions properly in gene targeting, the human fibrosarcoma cell line HTl080 - (ATCC CCL 121) was transfected with pE3Neo by electropora-tion.
HT1080 cells were ---;nt~inPrl in HaT (hyrAY:lnthinP/
aminopterin/Y~nthin~) 5~ p1S Pd DMEM with 159~ calf serum (Hyclone) prior to electroporation. Two days before electroporation, the cells are switched to the same medium without aminopterin. ~Yr~npnti~lly growing cells were tryp~ini 7Pd and diluted in DMEM/15~ calf serum, centri-fuged, and rPs~pPntlP~ in PBS (phosphate buffered 6aline) at a final cell volume of 13.3 million cells per ml.
pE3Neo is digested with HindIII, separating the 8 kb HPRT-neo fragment from the pUC12 ha~khnnp~ purified by phenol extraction and ethanol precipitation, and resus-pended at a concentration of 600 ~g/ml. 50 ~Ll (30 llg) was added to the electroporation cuvette (0.4 cm electrode gap; Bio-Rad Laboratories), along with 750 1ll of the cell 5--RpPn~; O n (10 million cellg) . Electroporation was at 450 volts, 250 ILFarads (Bio-Rad Gene Pulser; Bio-Rad Laborato-ries). The contents of the cuvette were immediately added to DMEM with 15~ calf serum to yield a cell suspension of 1 million cells per 25 ml media. 25 ml of the treated cell S~pPn~ n was plated onto 150 mm diameter tissue culture dishes and ;n~-l-h~tF-d at 37C, 5~ CO2. 24 hrs later, a G418 solution was added directly to the plates to yield a final concentration of 800 ~Lg/ml G418. Five days later the media was replaced with DMEM/15~ calf serum/

WO 95/31560 2 1 ~ ~ 2 ~

800 /lg/ml G418. Nine days after electroporatiorl, the media was replaced with DMEM/159~ calf serum/800 llg/ml G418 and 10 ILM 6-thioguanine. Colonies resistant to G418 and 6-TG were picked using cloning cylinders 14-16 days after 5 the dual sPlect;rn was initiated.
The results of f ive representative targeting experi -ments in ~T1080 cells are shown in Table 1.
TA~3,7JE 1 Number of Number of G418r Transfection Treated Cells 6-TGr Clones 1 x 107 32 2 1 x 107 28 3 1 x 107 24

4 1 x 107 32

5 1 x 107 66 For transfection 5, control plates rlp~irnpd to deter-mine the overall yield of G418r color,ies ;n~;r~te~l that 33,700 G418r colonies could be generated from the initial 1 x 107 treated cells. Thus, the ratio of targeted to non-targeted ever~ts is 66/33, 700, or 1 to 510 . In the five experiments ' ;nPrl, targeted events arise at a frequency of 3.6 x 106, or 0.00036~ of treated cells.
Restriction enzyme and Southern hybr;~;7~t;rn experi-ments using probes derived from the neo and HPRT genes localized the neo gene to the ~PRT locus at the predicted site within ~PRT exon 3.
.

WO95131560 2l ~028~ r~ r~4s C. G~l~ TMGI~TING IN PRIMARY AND ~EC6)NT~RY HUMAN SKIN
FTRRnR~ ,~,CTS
pE3Neo is digested with ~indIII, separating the 8 kb XPRT-neo fragment from the pUC12 h~rkhrn~, and purified by phenol extraction and ethanol precipitation. DNA was resuspended at 2 mg/ml. Three million secondary human foreskin fibroblasts cells in a volume of 0 . 5 ml were electroporated at 250 volti7 and 960 ~uFarads, with 100 ~g of HindIII pE3Neo (50 ~ul). Three separate transfections were peLL~ -1, for a total of 9 million treated cells.
Cells are processed and selected f or G418 resistance .
500,000 cells per 150 mm culture dish were plated for G418 selection. Af ter 10 days under selection, the culture medium is replaced with human fibroblast nutrient medium r-1nt~;nin~ 400 l~g/ml G418 and lO IlM 6-TG. Selection with the two drug combination is rrn~;n~ rl for 10 additional days . Plates are scanned microscopically to l r,r~l; 7e human fibroblast colonies resistant to both drugs. The fraction of G418r t-TGr colonies is 4 per 9 million treat-ed cells . These colonies constitute 0 . 0001% (or 1 in a million) of all cells capable of forming colonies. Con-trol plates ~ ; rJnf~d to determine the overall yield o~
G418r colonies ;n~;r~ted that 2,850 G418r colonies could be generated from the initial 9 x 106 treated cells.
Thus, the ratio of targeted to non-targeted events is 4/2,850, or 1 to 712. Restriction enzyme and Southern hybr~ ;rn experiments using probes derived from the neo and HPRT geneg were used to localize the neo gene to the HPRT locus at the predicted gite within HPRT exon 3 and demonstrate that targeting had occurred in these four clonal cell strains. Colonies resistant to both drugs have also been isolated by transf ecting primary cells (1/3 . o x 107) .
The results of geveral pE3Neo targeting f~pF~r;r 35 are summarized in Table 2. XindIII digested pE3Neo was WO95131560 2 1 ~ ~ 2 ~ 9 either transfected directly or treated with exonuclea~e III to generate 5 ' single-stranded overhangs prior to transfection (see Example lc) . DNA preparations with single-stranded regions ranging from 175 to 93Q base pairs 5 in length were tested. Using pE3neo digested with HindIII
alone, 1/799 G418-resistant colonies were ;~l~nt;fied by restriction enzyme and Southern hybri~li7~;cm analysis as having a targeted insertion of the neo gene at the HPRT
locus (a total of Z4 targeted clones were; ~o] ~te~
10 Targeting was maximally St; l~te~l (apprn~;r-t~1y 10-fold St; l~t;nn) when overhangs of 175 bp were used, with 1/80 G418r cnlnn;-~c displaying restriction LL _ 1: that are diagnostic for targeting at HPRT (a total o 9 targeted clones were isolated). Thus, using the cnnrl1t;nn~ and 15 I~ ' inzlnt DNA constructs described here, targeting is readily obserYed in normal human f ibroblasts and the overall targeting fre auency (the number of targeted clones divided by the total number of clones stably transf ected to G418-resistance) can be st; l ~trd by transfection with 20 targeting c;~ LLu~:Ls cnnt~;n;n~ single-stranded overhang-ing tails, by the method as described in Example le.
TAB~E 2 TARGETING TO THE HPRT ~OC~S IN HUMAN FIBROBIIASTS
pE3neo Number of ~umber Targeted Total Number of Treatment E~neriments Per G418r ColonY Tarqeted Clone HindIII digest 6 l/799 24 175 bp overhang l l/80 9 350 bp ~V~L;1~1~ 3 l/117 20 930 bp overhang 1 1/144 2~ 9rJ289 WO 95/3l560 ~ J~5~ 5 .

d. ~NFRATION OF A ~ KU~l FOR T~R~TFn INSERTION OF A
~FNF OF rr~Rz~pEurIc ~ INT0 THE H~MAN o.~n~qE
AND ITS USE IN GENE TMGETING
A variant of pE3Neo, in which a gene of therapeutic 5 interest is inserted within the HPRT coding region, adj a-cent to or near the neo gene, can be used to target a gene of therapeutic interest to a specific position in a recip-ient primary or secondary cell genome. Such a variant of pE3Neo can be constructed for targeting the hGH gene to0 the HPRT locus.
pXGH~ (schematically presented in Figure 3) is di-gested with EcoRI and the 4.1 kb fragment Co~ ;n;n~ the hGH gene and linked mouse metallothionein (mMT) promoter is isolated . The EcoRI overhangs are f illed in with the 15 Klenow fragment from E. coli DNA polymerase. Separately, pE3Neo is digested with XhoI, which cuts at the junction of the neo fragment and HPRT exon 3 (the 3 ' junction of the insertion into exon 3 ) . The XhoI ovf~rh~n~; n~ ends of the l;n~ ;7ed plasmid are filled in with the Klenow 20 fragment from ~. coli DNA polymerase, and the resulting f ragment is ligated to the 4 .1 ko blunt -ended hGH-mMT
fragment. Bacterial colonies derived from the ligation mixture are screened by restriction enzyme analysis ~or a single copy insertion of the hGH-mMT fragment and one 25 orientation, the hGH gene transcribed in the same direc-tion as the neo gene, is chosen and designated pE3Neo/hGH.
pE3Neo/hGH is digested with HindIII, releasing the 12.1 kb fragment cnntr~;n;ns HPRT, neo and mMT-hGH sequences.
Digested DNA is treated and transfected into primary or 30 secondary human fibroblasts as described in Example lc.
G41~r TGr colonies are selected and analyzed for targeted insertion of the mMT-hGH and neo sequences into the HPRT
gene as described in Example lc. Individual colonies are assayed for hGH expression using a commercially available ; n~qs;ly ~Nichols Institute).

WO 9S/31560 2 1 9 ~ 2 8 ~ s .

Secondary human fibroblasts were transfected with pE3Neo/hGH and thioguanine-resistant colonies were ana-lyzed for stable hGH expression and by restriction enzyme and Southern hybridization analysis. of thirteen TGr 5 colonies analyzed, eight colonies were ;rlont;f;od with an insertion of the hGH gene into the ~ nJ~ "R HPRT locus.
All eight strains stably expressed sign;f;~nt 9uantities of hGH, with an average expression level of 22 7 ~g/106 cells/24 hours. Alternatively, plasmid pE3neoEPO, Figure 10 4, may be used to target EPO to the human HPRT locus.
The use of homologous re~ ' ;n~tinn to target a gene of therapeutic interest to a specific position in a cell's genomic DNA can be o~nrlod upon and made more useful for producing products for therapeutic purposes ~e.g., pharma-15 ceutics, gene therapy) by the insertion of a gene throughwhich cells rnnt~;n;n~: l;fio~ copies of the gene can be selected for by exposure of the cells to an appropriate drug selection regimen. For example, pE3neo/hGH ~Example ld) can be modi~ied by inserting the dhfr, ada, or CAD
20 gene at a position; ';:~toly adjacent to the hGH or ~eo genes in pE3neo/hGH. Primary, secondary, or immortalized cells are transfected with such a plasmid and correctly targeted events are i~ nt; f ie~ . These cells are further treated with increasing ~ ce..LL~ions of drugs appropri-25 ate for the 8olo~t;nn of cellg cnnt~;ntn~ l;fied genes~for dhfr, the selective agent is methotrexate, for CAD
the selective agent is N- ~rhnRrhnn~r etyl) -L-aspartate ~PA~,), and for ada the selective agent is an adenine nucleoside ~e.g., ~l~nnR;no) In thig manner the ir,tegra-30 tion of the gene of therapeutic interest will be coampli-f ied along with the gene f or which ~ l; f; ~ copies are selected. Thus,- the genetic on~;noor;n~ of cells to produce genes for therapeutic uses can be readily con-trolled by proRole~t;n~ the site at which the targeting WO 95/31560 ~ C '~4''~

construct integrates and at which the amplified copies re6ide in the amplified cells.
e. ~qODIFICATION OF DNA TERMINI TO E~HANCE TARGETING
Several lines of evidence suggest that 3 ' -overhanging 5 ends are involved in certain homologous re~ ` ;n~t;~n pathways of E. 5~Q~, bacteriophage, S. cerevisiae and Xenopus laevis. In Xenopus laevis oocytes, molecules with 3~-uv .i,..".~;ng ends of several hundred base pairs in length underwent L- - ' in~t;nn with similarly treated molecules much more rapidly af ter microinj ection than molecules with very short UV~LilOl~y~ (4 bp) ~l~nf~rz~t~ri by restriction enzyme digestion. In yeast, the generation of 3 ~ -ovPrh~nsin~ ends several hundred base pairs in length appears to be a rate limiting step in meiotic r~- ; nct; -on. No evidence _or an involvement of 3'-ov~rh~n~;n~ ends in rf~ ' ;n~t;nn in human cells has been reported, and in no case have r ';fieci DNA substrates of any sort been shown to promote targeting (one form of homologous recom-~;n~tinn) in any specie8. The experiment described in the following example and Example lc suggests that 5'-over-hanging ends are efective for sti l~t;n~ targeting in primary, secondary and immortalized human $ibroblasts.
There have been no reports on the ~onh:ln~ of targeting by modifying the ends of the transfecting DNA
molecules . This example serves to illustrate that modif i -cation of the ends of linear DNA molecules, by conversion of the molecules~ termini from a double-stranded form to a single-stranded form, can stimulate targeting into the genome o primary and secondary human f ibroblasts .
1100 llg of plasmid pE3Neo (Example la) is digested with HindIII This DNA can be used directly after phenol extraction and ethanol precipitation, or the 8 kb HindIII
f1~, , nnt~;n;n~ only HPRT and the neo gene can be separated away from the pUC12 vector Af~ nt~c by gel W0 95L~1560 r~ 6?45 ) electrophoresi6. ExoIII digestion of the ~indIII digested DNA re6ults in extensive exonucleolytic digestion at each end, initiating at each free 3' end, and leaving 5~-ovorh~njinrj ends. The extent of ~7cnnllrl,oolytic action 5 and, hence, the length of the resulting 5'-overhangs, can be controlled by varying the time of ExoIII digestion.
ExoIII digestion of 100 l~g of HindIII digested pE3Neo is carried out according to the supplier~s r~ '-' condi-tions, for times of 10 sec, 1 min, 1.5 min, 2 min, 2.5 min, 3 min, 3.5 min, 4 min, 4.5 min, and 5 min. To moni-tor the extent of digestion an aliquot from each time point, crn~;n;n~ 1 ~g of ExoIII treated DNA, is treated with mung bean nuclease (Promega), under r~nnrlit;rne recom-mended by the supplier, and the samples fr~rt;r~n~t~-l by 15 gel electrophoresis. The difference in size between non-treated, ~indIII digested pE3Neo and the same mole-cules treated with ExoIII and mung bean nuclease is mea-sured. This size difference divided by two gives the average length o~ the 5 ' -overhang at each end of the 20 molecule. Using the time points described above and digestion at 30, the 5'-ovt:Ll.al.y~ produced should range from 100 to 1, 000 bases .
60 llg of ExoIII treated DNA (total XindIII digest of pE3Neo) from each time point is purified and el~:LL~ L-25 ated into primary, secondary, or immortalized human fibro-blasts under the conditiong described in Example lc. The degree to which targeting is ~.nh~nre~ by each ExoIII
treated preparation is rillAnt;~;e~ by counting the number of G418r 6-TGr colonies and comparing these numbers to 30 targeting with HindIII digested pE3Neo that was not treat-ed with ExoIII. :
The effect of 3'-ov~rh~n~;n~ ends can also be riuanti-fied using an analogous system. In this case ~indIII
digested pE3Neo i6 treated with bacteriophage T7 gene 6 35 exonuclease (United States Biorhpm; r~l e) for varying time WO 95/31560 ~ l ~ 0 2 ~

intervals under the supplier' s L- '^'l conditions .
Det~m; nAt lon of :the extent of digestion ~average length of 3~-overhang produced per end) and electroporation conditions are as de6cribed for ExoIII treated DNA. The 5 degree to which targeting is enhanced by each T7 gene 6 exonuclease treated preparation is ~Ant;r;~d by counting the number of G418r 6-TGr colonies and comparing these numbers to targeting with HindIII digested pE3Neo that was not treated with T7 gene 6 exonuclease.
Other methods for generating 5' and 3' .,v ,l,,.l, ~;n~
ends are possible, for example, ~l~nAt~lrAt;on and :~nnF~Al ;ns of two linear molecules that partially overlap with each other will generate a mixture of molecules, each molecule having 3~-~Ve:llldny~ at both ends or 5'-over_angs at both ends, as well as reannealed fragments indistinguishable f rom the starting linear molecules . The length of the ~V~:l;ld~ly~ is determined by the length of DNA that is not in common between the two DNA ~L ~
f. WN~U~'l'lON OF TARGETING PIASMIDS FOR PT~CING THE
HUMAN ERYTHROPOIETIN GENE llNDER THE t~ONTROI, OF THE
MOUSE META~LOTHIO~T~ PROMOTER IN PRIMARY. ~EcnNnARY
AND IMMoRTATlT~n HUMAN FIBROBI~STS
The following serves to illustrate one: ` ~'; ' of the present invention, in which the normal positive and negative regulatory sequences upstream of the human eryth-ropni~t; n (hEPO) gene are altered to allow expression of human erythropoietin in primary, secondary or immortalized human f ibroblasts, which do not express hEPO i~ signif i-cant quantities as obtained.
3 0 A region lying exclusively upstream of the human EPO
coding region can be amplified by PCR. Three sets of primers useful for this purpose were designed after analy-sis of the rl-hl; ~h~d human EPO sequence [Gen-hank designa-WO95/31560 2 1 ~ ~239 tion ~UMERPA; ~in, F-K., et al., Proc. Natl. Acad. Sci., SA 82:7580-7584 ~1985)] . These primer pairs can ampli~y fragments of 609, 603, or 590 bp.

~l~qERPA
Primer Coordinate Sequence Fragment Size Fl 2 , 2 0 5 ' AL~ L l ~ AC
( SEQ ID NO 1 ) R2 610 , 595 5 ' si~l.:~l~GCGAC 609 bp ( SEQ ID NO 2 ) F2 8 ~ 24 5 ' l~iG~ t'An ( SEQ ID NO 3 ) R2 610 > 595 5 ~ l~GCCAC 603 bp F3 21 _ 40 5 ~ rr~ T~ ~AaCTC
( SEQ ID NO 4 ) R2 610 > 595 5' ~w~ AGCGAC 590 bp The three LL _ ~ C overlap substantially and are interchangeable for the present purposes. The 609 bp LL. _ , ~rt~n~ from -623 to -14 relative to the trAn~l At; nn start site (HUMERPA nucleotide positions 2 to 5 610), is ligated at both ends with ClaI linkers. The resulting ClaI-linked fragment is digested with ClaI and inserted into the ClaI site of pBluescriptIISK/+ (Strata-gene), with the orientation such that ~MERPA nucleotide position 610 is adjacent to the SalI site in the plasmid 10 polylinker). This plasmid, p5~EPO, can be cleaved, sepa-rately, at the unique FspI or SfiI sites in the human EPO
upstream ~L _ t (HUMERPA nucleotide positions 150 and WO 95/31560 2 1 ~ 0 2 8 ~ , ~"1,~ 'q45 405, respectively) and ligated to the mouse metallothion-ein promoter. Typically, the 1. 8 kb EcoRI-BglII from the mMT-I gene [rnnt~inin~ no mMT coding s~rlllPnr~c; Hamer, D.H. and Walling M., J. Mol. A~Pl. Gen. 1 j273 238 ~1982);
this fragment can al60 be isolated by known methods from mouse genomic DNA using PCR primers ~cirjno~ from analysis of mMT sequences available from Genbank; i.e., MUSMTI, MUSMTIP, MU~ L1~ I] is made blunt-ended by known methods and ligated with SfiI digested ~also made blunt-ended) or lo FspI digested p5~EPO. The ori~nt~t;nnc of resulting clones are analyzed and those in which the former mMT BglII site is proximal to the SalI site in the plasmid polylinker are used for targeting primary and secondary human fibro-blasts . This or; ~nt~tinn directs mMT transcription to-wards H~MERPA nucleotide position 610 in the f inal con-struct. The resulting plasmids are ~ jsnAt~d p5'EPO-mMTF
and p5 ' EPO-mMTS for the mMT insertions in the FspI and Sf iI sites, respectively .
Additional upstream sequences are useful in cases where it is ~ hl ~ to modify, delete and/or replace negative regulatory elements or enhancers that lie up-stream of the initial target sequence. In the case of EPO, a negative regulatory element that inhibits EPO
expression in extrahepatic and extrarenal tissues [Semen-za, ~ . et al., Mol. Cell. Biol. 10:930-938 ~1990)] can be deleted. A series of ~1 et; nn~ within the 6 kb frag-ment are ~. e~aLed. The deleted regions can be replaced with an enhancer with broad host-cell activity [e.g. an enhancer from the Cytomegalovirus ~CMV) ] .
The orientation of the 609 bp 5'EPO fragment in the pBluescriptIIsR/+ vector was chosen since the H~MERPA
sequences are preceded on their 5' end by a samHI ~distal) and HindIII site (proximal)~ Thus, a 6 kb BamHI-HindIII
fragment normally lying upstream of the 609 bp fragment 35 [Semenza, G. ~. et al., Mol. Cell. Biol. 10:930-938 WO95/31560 21 ~0289 r~. c ~ ,r l5 (1990) ] can be isolated from genomic DD~A by known methods.
For example, a bacteriophage, cosmid, or yeast artificial chromosome library could be screened with the 609 bp PCR
amplified fragment as a probe. The desired clone will have a 6 kb BamHI-HindIII fragment and its identity can be cnnf; 9 by comparing its restriction map from a restric-tion map around the human EPO gene determined by known methods. Alternatively, constructing a restriction map of the human genome upstream of the EPO gene using the 609 bp fL _ as a probe can identify enzymes which generate a LL _ originating between Ht~MERPA coordinates 2 and 609 and Pl~tPnfl;n~ past the upstream BamHI site; this LL _ can be isolated by gel ele.:LLu~huLesis from the appropri-ate digest of human genomic DNA and ligated into a bacte-rial or yeast cloni~g vector. The correct clone will hybridize to the 609 bp 5'EPO probe and contain a 6 kb BamHI-HindIII fragment. The; ~nl At~prl 6 kb fragment is inserted in the proper oriPnt~tinn into p5'EPO, p5'EPO-mMTF, or p5'EPO-mMTS lsuch that the HindIII site is adja-cent to HUMERPA nl~clAnt;~G position 2). Additional up-stream sPSrlPn~ P~ can be ;~nlAtPfl by known methods, using C1IL~ walking techni~ues or by ;~nlAt;nn of yeast artificial ~:;IL~ ~ A hybridizing to the 609 bp 5'EPO
probe .
The cloning strategies described above allow se~[uenc-es upstream of EPO to be r ';fiPcl n y'tro for AuL,se~ue-lt targeted transfection of primary, 9e~ulldaLy or immortal-ized human f ibroblasts . The strategies describe simple insertions of the mMT promoter, as well as deletion of the negative regulatory region, and deletion of the negative regulatory region and rPrl A~ ' with an enhancer with broad host-cell activity.

Wo9S~1560 ~ 1 ~ 0 ~ r~"., ~ iq45 g. ACTIVATING THE H~LAN RPO r~N~ AND IS~T~TION OF TAR-GETr'n PRTI~ARY. SErr~NnAR'f AND IM~QRTArlTz~n HU~N
FIBRQRr~CTS BY St'Rr~UNING
For targeting, the plasmids are cut with restriction enzymes that free the insert away from the plasmid back-bone . In the case of p5 ' EPO-mMTS, HindIII and SaII diges-tion releases a targeting fragment of 2.4 kb, comprised of the 1. 8 kb mMT ~ rr f lanked on the 5 ' and 3 ' sides by 405 bp and 204 base pairs, respectively, of DNA for tar-10 geting this construct to the regulatory region of the human EPO gene . This DNA or the 2 . 4 kb targeting fragment alone is purified by phenol extraction and ethanol precip-itation and transfected into primary or secondary human fibroblasts under the conditions described in Example lc.
15 Transfected cells are plated onto 150 mm dishes in human fibroblast nutrient medium. 48 hours later the cells are plated into 24 well dishes at a density of 10, 000 cells/cm' [approximately 20,000 cells per well; if target-ing occurs at a rate of 1 event per 106 clonable cells 20 (Example lc, then about 50 wells would need to be assayed to isolate a single expressing colony]. Cells in which the transfecting DNA has targeted to the homologous region upstream of the human EPQ gene will express hEPO under the control of the mMT promoter. After 10 days, whole well 25 supernatants are agsayed for EPQ expression using a com-mercially available; --Qq:ly kit (Amgen). Clones from wells displaying hEPO synthesiG are isolated using known methods, typically by assaying fractions of the heteroge-nous populations of cells separated into individual wells 30 or plates, asgaying fractions of these positive wells, and repeating as needed, ultimately iqnl~t;n~ the targeted colony by gcreening 96-well microtiter plates seeded at one cell per well. DNA from entire plate lysates can also be analyzed by PCR for amplification of a fragment using a 35 mMT sper;flr primer in conjunction with a primer lying WO95131560 2 1 9~)2~39 r ~ tr~15 upstream of HUMERPA nucleotide position 1. This primer pair should amplify a DNA fragment of a size precisely predicted based on the DNA sequence. Positive plates are tryrs;n; 7~d and replated at successively lower dilutions, 5 and the DNA pr,o~ArAtinn and PCR steps repeated as needed to isolate targeted cells.
The targeting schemes herein described can also be used to activate hGH expression in immortalized human cells (for example, HT1080 cells (ATCC CC~ 121), Xe~a 10 cells and derivatives of HeI,a cells (ATCC CCL2, 2.1 and 2.2), MCF-7 breast cancer cells (ATCC HBT 22), K-562 leukemia cells (ATCC CC~ 232), RB carcinoma cells (ATCC
CC~ 17), 2780AD ovarian carcinoma cells (Van der Blick, A.M. et al., Cancer Res 48:5927-5932 (1988), Raji cells (ATCC CCL 86), Jurkat cells (ATCC TIB 152), Namalwa cells (ATCC CR~ 1432), H~-60 cells ~ATCC CCL 29~0), Daudi cells ~ATCC CC~ 213), RPMI 8226 cells (ATCC CCL 155), U-937 cells (ATCC CR~ 1593), Bowes Melanoma cells (ATCC CR~
9607), WI-38VA13 subline 2R4 cells (ATCC CL~ 75.1), MO~T-4 cells (ATCC CR~ 1582), and varous heterohybridoma cells) for the purposes of producing hGH for conventional pharma-ceutic delivery.
h. ACTIVATING THE HUMAN EPO GENE AND IS~lT~TIoN OF TAR-GET~ PRIMARY, sFmt~NnARy AND IMrlmRT~T T~En HUM~N
FTRRRT ~qTS BY A POSITIVE OR A COMBINED POSITIVE/
NEGATIVE ~q~T ~-'TION SYST~M
The strategy for constructing p5'EPO-mMTF, p5'EPO-mMTS, and derivatives of such with the additional upstream

6 kb BamHI-HindIII fragment can be followed with the addi-tional step of inserting the neo gene adjacent to the mMT
promoter. In addition, a negative selection marker, for example, gpt [from pMSG (Pharmacia) or another suitable source], can be inserted adj acent to the HUM~RPA sequences in the pBluescriptIISR/+ polylinker. In the former case, WO~5/3156û 21 ~ 02~ r ~ 4~

G418r colonies are isolated and screened by PCR amplifica-tion or re6triction enzyme and Southern hybr;~1;7Pt;nn analy6is o~ DNA prepared from pools of colonies to identi-fy targeted colonie8. In the latter case, G418r colonies are placed in medium rnnt~;n;n~ 6-th;n~ nth;n~ to select against the integration of the gpt gene [Besnard, C. ~
al., Mol. Cell; Biol. 7:4139-4141 (1987)]. In addition, the HSV-TK gene can be placed on the opposite side of the insert as gpt, allowing Eelection for ~eo and against both gpt and TK by growing cells in human fibroblast nutrient medium rnnt~;n;ns 400 ~g/ml G418, 100 IIM 6-th;nY~nth;nP, and 25 I~g/ml gancyclovir. The double negative selection should provide a nearly absolute selection for true tar-geted events and Southern blot analysis provides an ulti-mate ~`nn f; rr- t ~ nn .
The targeting schemes herein described can also be used to activate hEPO expression in immortalized human cells (for example, HTl080 cells ~ATCC CCL 121), HeLa cells and derivatives of He~a cells (ATCC CCL2, 2.1 and 2.2), MC~-7 breast cancer cells (ATCC ~BT 22), K-562 kPm; a cells ~ATCC CC~ 232), KB carcinoma cells (ATCC
CCL 17), 2780AD ovarian carcinoma cells (Van der Blick, A.M. et al~"`;~nr~r Res, 48:5927-5932 (1988), Raji cells (ATCC CCL 86), Jurkat cells (ATCC TIB 152), Namalwa cells (ATCC CRL 1432), i~-60 cells (ATCC CCL 24~0), Daudi cells (ATCC CCL 213), RPMI 8226 cells (ATCC CC~ 155), U-937 cells (ATCC CRL 1593), Bowes Melanoma cells (ATCC CRL
9607), WI-38VA13 subline 2R4 cells (ATCC CLI. 75.1), MOLT-4 cells (ATCC CRL 1582), and various heterohybridoma cells) for the purposes of producing h~PO for conventional phar-maceutic delivery.
.

WO 95/31560 P~
21 93~89 i. WN~l~UULlON OF TARGETING pT,~qMTnq FOR pT,~rTNG T~7~
HUMAN GROWTH HORMON-E r~T~N~ TJNDER THE CONTRO~ OF T~T~`
MOUSE MET~TlrloTTTToN-EIN PROMOTER TN PRIMARY. SECONDARY
OR IMMnRTZ~T T7T'n HUMAN FJR~nRT.~qTS
The following example serves to illustrate one em-hC~ t of the present invention, in which the normal regulatory sequences upstream of the human growth hormone gene are altered to allow expression of human growth hormone in primary, secondary or immortalized human fibro-blasts.
Targeting molecules similar to those described in Example lf for targeting to the EPO gene regulatory region are generated using cloned DNA LL _ tc derived from the 5 ' end of the human growth hormone N gene . An approxi-mately 1. 8 kb fragment spanning HUMGHCSA (Genbank Entry) nucleotide positions 3787-5432 (the positions of two EcoMI
sites which generate a convenient sized fragment for cloning or for diagnostic digestion of sllhrl r~nPq involving this fragment) is: 1 i fiP1 by PCR primers designed by analysis of the T~Tn~rT~rq~ sequence in this region. This region extends f rom the middle of hGH gene N intron 1 to an upstream position approximately 1. 4 kb 5 ' to the trans-l;~tinn;ll 8tart site. pUC12 is dige8ted with EcoRI and BamHI, treated with EClenow to generate blunt ends, and recircularized under dilute conditions, resulting in r' Ar-; ~lq which have lost the EcoRI and Bam~I sites . This plasmid is designated pUC12XEB. HindIII linkers are ligated onto the, ~ ; P~ hGH fragment and the resulting fragment is digested with XindIII and ligated to HindIII
digested pUC12XEB. The resulting plasmid, plJC12XEB-5'hGH, i8 digested with EcoRI and BamHI, to remove a O . 5 kb fragment lying i -; ~tely upstream of the hGH transcrip-tional initiation site. The digeated DNA is ligated to the 1.8 kb EcoRI-BglII from the mMT-I gene [r~nt~;nin~ no mMT coding se~uences; ~amer, D.H. and Walling, M., T. Mol.

WO 95131560 2 1 ~ O ~ . C~ 4~

AT~T~1 Gen. L:273-288 (1982); the fragment can also be isolated by known methods from mouse genomic DNA using PCR
primers designed from analysis of mMT sequences available from Genbank; i.e., MUSMTI, MUSMTIP, MUSMTIPRM]. This plasmid p5 ' hGH-mMT has the mMT promoter f lanked on both sides by upstream hGH sequences.
The cloning strategies described above allow sequenc-es upstream of hGH to be modified ~ vitro for subsequent targeted transfection of primary, secondary or immortal-ized human fibroblasts. The strategy described a simple insertion of the mMT promoter. Other strategies can be envisioned, for example, in which an enhancer with broad host-cell spe~-;f;~-;ty is inserted upstream of the inserted mMT sequence.
j. ACTIVATING THE HUM~N hGH G~NE AND ISOI,ATION OF TAR-~TT'r) PRIM~RY. ~coNn~Ry AND IMMnRTATlT5T~n HUM1~N
FTT~Rt)T~T~2~Ts BY SrRT`~NTNr.
For targeting, the plasmids are cut with restriction enzymes that free the insert away from the plasmid back-bone. In the case of p5'hGH-mMT, HindIII digestion re-leases a targeting f~ of 2 . 9 kb, comprised of the 1.8 kb mMT promoter flanked on the 5' e~d 3' sides by DNA
f or targeting this construct to the regulatory region of the hGE gene . This DNA or the 2 . 9 kb targeting fragment alone is purified by phenol extraction and ethanol precip-itation and transfected into primary or se-,u~ r human fibroblasts under the cnnfi;ti~nc described in Example 11.
Transfected cells are plated onto 150 mm dishes in human fibroblast nutrient medium. 48 hours later the cells are plated into 24 well dishes at a density of 10,000 cells/cm1 [approximately 20,000 cells per well; if target-ing occurs at a rate of 1 event per 1o6 clonable cells (Example lc), then about 50 wells would need to be assayed to isolate a single expressing colony]. t:ells in which the WO 95/31560 ~ l 9 Q ~ 8 ~ r~l" ' "~

transf ecting DNA has targeted to the homologous region upstream of hGH will express hGH under the control of the mM~ promoter. After lO days, whole well supernatants are assayed for hGH expression using a commercially available ; ~say kit (Nichols). Clones from wells displaying hGH synthesis are isolated using known methods, typically by assaying fractions of the heterogenous pQpl~lation~ Of cells separated into individual wells or plates, assaying fractions of these positive wells, and repeating as need-ed, ultimately isolated the targeted colony by screening 96-well microtiter plates seeded at one cell per well.
DNA from entire plate lysates can also be analyzed by PCR
for amplification of a fragment using a mMT specific primer in conjunction with a primer lying downstream of XUMGHCSA nucleotide position 5,~32. This primer pair should amplify a DNA fragment of a size precisely predict-ed based on the DNA se~auence. Positive plates are tryp-sinized and replated at succe6sively lower ~ lti~n~, and the DNA preparation and PCR steps repeated as needed to isolate targeted cells.
The targeting schemes herein described can also be used to activate hGX expression in immortalized human cells (for example, HT1080 cells (ATCC CCL 121), He~a cells and derivatives of HeLa cells (ATCC CCL2, Z.1 and 2.2), MCF-7 breast cancer cells (ATCC HBT 22), K-562 leukemia cells (ATCC CCL 232 ), KB carcinoma cells (ATCC
CCL 17), 2780AD ovarian carcinoma cells (Van der Blick, A.M. et al., r~nc.o~ Res. 48:5927-5932 (1988), Raji cells (ATCC CCL 86), Jurkat cells (ATCC TIB 152), Namalwa cells (ATCC CR~ 1432), H~-60 cells (ATCC CC~ 240), Daudi cells (ATCC CC~ 213), RPMI 8226 cells (ATCC CC~ 155), U-937 cells (ATCC CRL 1593), Bowes Melanoma cells (ATCC CRL
9607), WI-38VA13 subline 2R4 cells (ATCC CL~ 75.1), MO~T-4 cells (ATCC CR~ 1582), and various heterohybridoma cells) W095/31560 2 1 ~ ~ P~l/l e. ~rl5 f or the purposes of producing hGH f or conventional pharma -ceutic delivery.
k. ACTIVATTNG TM~ HUMAN hGH GENE AND ISOI,ATION OF TAR-6TRTrn PRT~RY. SECONDARY AND IMMORTA~IZED HUM~N
FTRR~RT ~TS BY A POSITIVE OR A t:OMBINED POSITIVE/
NT~ TIVE ~T rCTION SYSTEM
The strategy f or constructing p5 ' hGH-mMT can be followed with the additional step of inserting the neo gene adjacent to the m~T promoter. In addition, a nega-tive selection marker, for example, gpt [from pMSG (Phar-macia) or another suitable source], can be inserted adja-cent to the HUMGHCSA se~uences in the pUC12 poly-linker, In the former case, G4l8r colonies are ~ qn1 ~tP~ and screened by PCR amplification or restriction enzyme and So~lth~rn hybri~l;7~tinn analygig of DNA ~Le~aLed from pools of colonies to identify targeted colonies. In the latter case, G418~ colonies are placed in medium cont~in;n~
thin~z~nthin~ to 8elect againgt the integration of the gpt gene (Besnard, C. et al., Mol. Cell. Biol. 7: 4139-4141 (1987) ] . In addition, the HSV-TK gene can be placed on the opposite side of the ingert a5 gpt, allowing s~olect;c~n for neo and againgt both gpt and TK by gro~ing cells in human fibroblast nutrient medium cnnt~'nln~ 400 llg/ml G418, 100 IlM 6-thin~nthin~o, and 25 ~g/ml gancyclovir.
The double negative s~le~t;nn should provide a nearly absolute gelection for true targeted event8. .~ollth~orn hybri~i7~t;nn analysis is rnnf;rr-tnry The targeting schemes herein described can also be used to activate hGH expression in immortalized human cells (for example, HT1080 cells (ATCC CCL 121), ~e~a cells and derivatives of HeLa cells (ATCC CCL2, 2.1 and 2.2), MCF-7 brea8t cancer cellg (ATCC HBT 22), K-562 leukemia cells (ATCC CCL 232), KB carcinoma cells (ATCC
CC~ 17), 2780AD ovarian carcinoma cells (Van der Blick, WO 95/31~60 ~ 4 A.M. et al., Cancer Res 48 :5927-5932 (1988), Raji cells (ATCC CCL 86), Jurkat cells (ATCC TIB 152), Namalwa cells (ATCC CRL 1432), HL-60 cells (ATCC CCL 240), Daudi cells (ATCC CCL 213), RPMI 8226 cells (ATCC CCL 155), U-937 5 cells (ATCC CRL 1593 ), Bowes Melanoma cells (ATCC CRL
9607), WI-38VA13 subline 2R4 cells (ATCC CLL 75.1), MOLT-4 cells (ATCC CRL 1582), and various heterohybridoma cells) or the purposes o producing hGH f or coIlventional pharma-ceutic delivery.
The targeting constructs described in Examples lf and li, and used in Examples lg, lh, lj and lk can be 'if;Prl to include an l;f;i~hle gf~lectiqh~e marker (e.g., ada, dhfr, or CAD) which is useful for selecting cells in which the activated ~ld~ us gene, and the i _ l; f j iqhle select-15 able marker, are ~ l; f; .~ Such cells, expressing or capable of expressing the ~ y~ u~ gene encoding a therapeutic product can be used to produce proteins (e . g ., hGH and hEPO) for conventional phar--re~tir delivery or f or gene therapy .
20 l. TRANSFECTION OF PRIMARY AND ~qT~ Nn~Ry FIBROBLASTS
WITH ~i~.)ti~;NUU:i l NA AND A SELECT~TiT ~ MARKER GENE BY
::~T.T~'TROp~)T~1~TION
~ Yrnn~nt;i~ly growing or early stationary phase fibroblasts are trvpsinized and rinsed from the plastic 25 surface with nutrient medium. An aliquot of the cell s~rPn~inn is removed for counting, and the, ;n;n~
cells are subjected to centrifugation. The s--r~rniqtiqnt is aspirated and the pellet is r~ rPn~<~d in 5 ml of elec-troporation buffer (20 mM HEPES pH 7.3, 137 mM NaCl, 5 mM
3 0 KCl, 0 . 7 mM Na2HPO~, 6 mM dextrose) . The cells are recen-trifuged, the g--r~rni~ti~nt aspirated, and the cells resus-pended in electroporation buffer cnntiq;n;ng 1 mg/ml acety-lated bovine serum albumin. The final cell suspension Wo 9S/31S60 2 1 9 0 2 8 9 r~l" s o~45 contains approximately 3 x 106 cells/ml. Electroporation should be performed immediately following resuspension.
Supercoiled plasmid DNA is added to a sterile cuvette with a 0 .4 cm electrode gap (Bio-Rad. ) The final DNA
5 concentration is generally at least 120 ~Lg/ml. 0.5 ml of the cell s~p~n~inn (cnnt~;n;n~ approximately 1.5 x 106 cells) is then added to the cuvette, and the cell suspen-sion and DNA solutions are gently mixed. Electroporation is performed with a Gene-Pulser apparatus ~Bio-Rad).
Capacitance and voltage are set at 960 ~LF and 250-300 V, respectively. As voltage increases, cell survival de-creases, but the percentage of survivi~g cells that stably incorporate the intrQduced DNA into their genome increases dr--~t;r~lly. Given these parameters, a pulse time of apprn~ir-tGly 14-20 mSec should be observed.
Electroporated cells are ~~;ntA;n~d at room tempera-ture for apprn~;r-t~ly 5 min, and the co~tents of the cuvette are then gently removed with a sterile transfer pipette. The cells are added directly to lO ml of pre-warmed nutrient media (as above with 15% calf serum) in a 10 cm dish and ;nr~h~t~d as described above. The follow-ing day, the media is aspirated and replaced with 10 ml of fresh media and incubated for a further 16-24 hours.
Subculture of cells to determine cloning Pff;r;~nry and to select for G418-resistant colonies is performed the fol-lowing day. Cells are trypsinized, counted and plated;
typically, fibroblasts are plated at 103 cells/lO cm dish for the ~ t~rm;nAt;nn of cloning ~ff;r;~nry and at 1-2 x lO' cells/lO cm dish for G418 selectior,.
Xuman fibroblasts are selected for G418 resistance in medium consigting of 300-400 llg/ml G418 (~pn~t;r;n~ disul-fate salt with a potency of approximately 5036; Gibco) in fibroblasts nutrient media (with 15~ calf serum). Cloning efficiency is determined in the absence of G418. The plated cellg are incubated for 12-14 days, at which time WO95/31~60 21 90289 ~ s ,~ l5 .

colonies are fixed with formalin, stained with crystal violet and counted (for cloning efficiency plated) or isolated using cloning cylinders (for G418 plates).
Electroporation and selection of rab~it fibroblasts is 5 performed ~c~ntiz~71y as described for human fibroblasts, with the exception of the selection conditions used.
Rabbit fi~roblasts are selected for G418 resistance in medium ~-~nt:~;n;n~ 1 gm/ml G418.
Fibroblasts were; col AtF~ from freshly excised human foreskins. Cultures were seeded at 50, 000 cells/cm in DMEM + lO~ calf serum. When cultures became confluent, fibroblasts were harvested by trypsinization and trans-fected by electroporation. Electroporation conditions were evaluated by transfection with the plasmid pcDNE0 (Figure 5). A L~L~senLative electroporation experiment using near optimal conditions (60 I~g of plasmid pcDNEO at an electroporation voltage of 250 volts and a capacitance setting of 960 ~Farads) resulted in one G418 colony per 588 treated cells (0.17~ of all cells treated), or one G418 colony per 71 clonable cells (1.49~) .
When nine separate el~LLu~uLclLion experiments at near optimal conditions (60 ~g of plasmid pcDNEO at an electroporation voltage of 300 volts and a capacitance setting of 960 ~Farads) were performed, an average of one G418 colony per l, 899 treated cells (0 . 05g,~) was observed, with a range of l/882 to 1/7, 500 treated cells . This corresponds to an average of one G418 colony per 3 8 clon-able cells (2 . 6~
Low passage primary human f ibroblasts were converted to hGH expressing cells by co-transfection with plasmids;
pcDNE0 and pXGH5. Typically, 60 ~g of an ~ lAr mix-ture of the two plasmids were tran6fected at near optimal conditions (electroporation voltage of 300 volts and a capacitance setting of 960 I~Farads). The results of such WO 95131560 2 1 9 0 2 8 q ~ 5~45 an experiment re6ulted in one G418 colony per 14, 705 treated cell6.
hGH expression data for these and other cells isolat-ed under ;~Pnt;n~ transfection conditions are summarized 5 below. Ultimately, 98~ of all G418r colonies could be PYr~n~1~1 to generate mass cultures.
Number of G418r Clones Analyzed 154 Number of G418r/hGH
lO Bxpressing Clones 65 Average hGH Bxpression ~evel 2.3 ~Lg hGH/106 Cells/24 hr Maximum hGH Bxpression Level 23 . O /lg hGH/106 Cells/24 hr Stable transfectants also have been generated by electroporation of primary or se~ d~ ~ y human f ibroblasts with pXGH301, a DNA con6truct in which the neo and hGX
genes are present on the same plasmid molecule. pXGH301 was constructed by a two-step procedure. The SaII-ClaI
~ _ from pBR322 (positions 23-651 in psR322) was ; Anl atf~l and inserted into SaII-ClaI digested pcDNBO, introducing a BamHI site upstream of the SV40 early pro-moter region of pcDNBO. This plasmid, pBN~O was digested with BamHi and the 2.1 kb fragment rnn~!;n;n J the neo gene under the control of the SV40 early promoter, was ;Anl~ted and inserted into Bam~I digested pXGH5. A plasmid with a single insertion of tke 2.1 kb BamHI fragment was ;An1~t~d in which neo and hGH are transcribed in the same direction relative to each other. This plasmid was designated pXGH301. For e~ample, 1.5 x 1o6 cells were electroporated with 60 llg pXGH301 at 300 volts and 960 IlLFarads. G418 resistant colonies were isolated from transfected second-ary fibroblasts at a frequency of 652 G418 resistant ~ ~ 3~8q WO95/31560 1~"~ ~ ^l5 colonies per 1.5 x 10 treated cells (1 per 2299 treated cells) . Approximately 59~ of these colonies ex~ress hGH.
R~MP~E 2. wN~lKu~:LlON OF TARGETING P~ASMIDS WHICH r~R-SU~T IN rTm~RRTc TR~N~f~r~TpTIoN UNITS IN WHICH
HUMAN GROWTH HORMON_ AND ERYTHROPOIETIN SE-OrTRNt~r~ ARE FUS_D --The following serves to illustrate two further em-bs~l; ~q of the present invention, in which the normal regulatory se~uences U~LL~I~III of the human E3PO gene are 10 altered to allow expression of hEPO in primary or second-ary f ibroblast strains which do not express hEPO in de-t,ort~hl ~ quantities in their untransfected state as ob-tained. In these ~omhor~; ' q, the products of the target-ing events are chimeric transcription units in which the 15 f irst exon of the human growth hormone gene i8 ~ q- i~; nn~
u~sLLc:alll of hEPO exons 2-5. The product of transcription, splicing and tr~nql ~t; (~n is a protein in which amino acids 1-4 of the hEPO signal peptide are replaced with amino acid residues 1-3 of hGH. The two ~ `-'; q differ with 20 respect to both the relative positions of the foreign regulatory sequences that are inserted and the specif ic pattern of splicing that needs to occur to produce the f inal, processed transcript .
Plasmid pXEPO-10 is designed to replace exon 1 of 25 h_PO with exon 1 of hGH by gene targeting to the endoge-nous hEPO gene on human C1IL~ ~ ~ 7. Plasmid pXEPO-10 is constructed as follows. First, the inte~ t~ plasmid pT163 is constructed by inserting the 6 kb HindIII-BamHI
fragment (see Example lf) lying upstream of the h_PO
30 coding region into HindIII-Bam~HI digested pBluescriptII
SR+ (Stratagene, ~aJolla, CA). The product of this liga-tion is digested with XhoI and HindIII and ligated to the 1.1 kb HindIII-XhoI ~L _ ' from pMClneoPolyA rThomas, K.
R. and Capecchi, M. R. Cel1 51: 503-512 (1987) available WO 9513~60 1' ~ 45 2lsa~

from Strategene, LaJolla, CA] to create pT163. Oligo-nucleotides 13 .1 - 13 . 4 are utilized in polymera6e chain reactions to generate a fusion fragment in which the mouse meta11Othinn~o;n 1 (mMT-I) promoter - hG~ exon l sequences 5 are additionally fused to hEP0 intron 1 sequences. ~irst, oligonucleotides 13 .1 and 13 . 3 are used to amplify the approximately 0.73 kb mMT-I promoter - hGH exon 1 fragment from pXG~5 (Figure 5) . Next, n~ i~nn-lrleotides 13.2 and 13.4 are used to amplify the approximately 0.57 kb frag-lO ment comprised ~l~ ~ n~ntly of hEP0 intron 1 from humangenomic DNA. Finally, the two ~ ~lif;~-i fragments are mixed and further amplified with oligonucleotides 13.1 and 13.4 to generate the final fusion fragment (fusion frag-ment 3 ) f lanked by a Sal I site at the 5 ' side of the mMT- I
15 moiety and an XhoI site at the 3 ' side of the ~EP0 intron 1 se~uence. Fusion fragment 3 is digested with XhoI and SalI and ligated to XhoI digested pT163. The ligation mixture i5 trangformed into E. coli and a clone rnnt;l;n;ng a single insert of fusion fragment 3 in which the XhoI
20 site is r~n~rat~d at the 3 ' side of hEPO intron 1 se-uences is i~nt;f;~d and designated pXEPO-10.
13 .1 5 ' ~ ~C GGTACCTTGG TTTTTA~AAC C
SalI KpnI
(SEQ ID N0 5) 13 . 2 5 ' Cr~aC~Gar~ ATGG~ GTGAGTACTC G~ CG
( SEQ ID N0 6 ) 13 . 3 5 ' CGCCCAGCCC GCGAGTACTC ACCTGTAGCC A~ Lu~:~G~:lA GG
(SEQ ID N0 7) 13 . 4 5 ' ~ ~G r~ o-~ TAGCCAGGCT G
XhoI
(SEQ ID N0 8) The non-boldface region of oligo 13.1 is identi-cal to the mMT-I promoter, with the natural KpnI

W095/31!i60 21 9a2~9 ~ "`15 .

6ite as its 5 ~ boundary. The hnl tlfAre type denotes a SalI site tail to convert the 5 ~ boun-dary to a SalI site. The boldface region of oligos 13.2 and 13.3 denote hGH sequences, while the non-boldface regions are lntron 1 seguences from the hEPO gene. The non-boldface region of oligo 13.4 is ~rnt;rAl to the last 25 bases of hEPO intron 1. The boldface region includes an XhoI site tail to convert the 3 ' boundary of the ; ,1; fie~i f~ to an XhoI site.
Plasmid pXEPO-ll is ~ ; gn~d to place, by gene tar-getinq, the mMT- I promoter and exon 1 o~ hGH upstream of the h~PO structural gene and promoter region at the endog-enous hEPO locu6 on human C~1L~ ~~ 7. Plasmid pXEPO-ll is constructed as follows. ~ nnl~rl~otides 13.1 and 13.5 - 13 . 7 are llt; 1; 7e~ in polymerase chain reactions to generate a fusion fragment in which the mouse metallo-thionein I (mMT- I ) promoter - hGH exon 1 sequences are additionally fused to hEPO se~nr~A from -1 to -630 relative to the hEPO coding region. First, oligonucleo-tides 13 .1 and I3 . 6 are used to amplify the approximately 0.75 kb mMT-I promoter - hGH exon 1 fragment from pXGH5 ~Figure 5) . Next, nl irJrnllrleotides 13.5 and 13.7 are used to amplify, from human genomic DNA, the approximately 0.65 kb frA; ' comprised ~L~ ~ nAntly of hEPO sequences from -1 to -620 relative to the hEPO coding region. Both oligos 13 . 5 and 13 . 6 contain a 10 bp linker ser~ nre located at the hGH i~tron 1 - hEPO promoter region, which corresponds to the natural h~PO intron 1 splice-donor site. Finally, the two ~l;f;ed LL _ ~ are mixed and further amplified with nl ;innllrl~r-~tide8 13.1 and 13 .7 to generate the final fusion fragment ~fusion fragment 6) flanked by a SalI gite at the 5~ side of the mMT-I moiety and an XhoI site at the 3 ~ side of the hEPO promoter W09S~31560 2 1 9 02 8 9 region. Fusion fragment 6 is digested with XhoI and SalI
and ligated to XhoI digested pT163. The ligation mixture is transformed into E. coli and a clone ~nto;nin,A a single insert of fusion fragment 6 in which the XhoI site 5 is regenerated at the 3 ' side of hEPO promoter seguences i5 ;rl~nt;f;~ and designated pXEPO-11.
13.5 5~ GA._AGCTCAC rT~AAGGA~ TG~ r~ TGAGTACTC
~CTGG GCTTCCAGAC CCAG (SEQ ID NO 9 HindIII

13 . 6 5 ' ~ i Al~r~rrrz~r~ GCTTGAGTAC TCACCTGSAG
HindIII
G,ATTGCCGC T~GGTGAGCT GTC (SEQ ID NO lO
13 . 7 5 L~L~;L~ AG ~:L~ JC~ w~ , CCTC
XhoI
(S~Q ID NO ll~
The boldface regions of oligos 13.5 and 13.6 denote hGH se~auences . The it=l; C; ~ rl regions ~ULLe~Ulld to the first 10 base pairs of hEPO
intron 1. The L~ ;n-lf.r of the oligos corre-spond to hEPO sequences from -620 to -597 rela-tive to the hEPO coding region. The non-bold-face region of oligo 13.7 is i~l~nt;f~l to bas-es -1 to -24 relative to the hEPO coding region.
The boldf ace region includes an XhoI site tail to convert the 3 ' boundary of the amplif ied LL _ to an XhoI site.
Plasmid pXEPO-lD can be used for gene targeting by digestion with BamHI and XhoI to release the 7.3 kb frag-30 ment C, nt=;n;n,A the mMT-I/hGH fusion flanked on both sides by hEPO seguences. This fragment (targeting LL~I ' 1) WO 95/31560 2 1 ~ (~ 2 8 9 I ~ I/L _ ~ ~r 15 .

contains no hEPO coding sequences, having only ~eguences lying between -620 and approximately -6620 upstream of the hEPO coding region and hEPO intron 1 seguences to direct targeting to the human EPO locus. Targeting LL _ 1 is 5 transfected into primary or secondary human skin fibro-blasts using conditions similar to those described in Example lc. G418-resistant colonies are picked into individual wells of 96-well plates and screened for EPO
expression by an E~ISA assay ~R&D Systems, ~;nnP~rol;~
10 MN). Cells in which the transfecting DNA integrates randomly into the human genome cannot produce EPO. Cells in which the transfecting DNA has undergone homologous rec ` ;n~tion with the . ~ R hEP0 intron 1 and hEPO
u~lLLealll se~lPnrPC contain a chimeric gene in which the 15 mMT-I promoter and non-transcribed sequences and the hGH
5' untr~n~1~ted sequences and hGH exon 1 replace the normal hEPO promoter and hEPO exon 1 ( see Figure 1~ . Non-hEPO ~Pr~ nrPR in targeting fragment 1 are joined to hEPO
5Prl~nrP~ down-stream of hEPO intron 1. The rPrl ~
20 of the normal hEPO regulatory region with the mMT-I pro-moter will activate the EPO gene in fibroblasts, which do not normally express hEPO . The rPr~ R~ ' of hEPO exon 1 with hGH exon 1 results in a protein in which the first 4 amino acids o~ the hEPO signal peptide are replaced with 25 amino acids 1-3 of hGH, creating a fllnrt;nn~l, chimeric signal peptide which is removed by post-tr~n~l~t;nn pro-cessing from the mature protein and is secreted from the expressing cells.
Plasmid pXEPO-ll can be used for gene targeting by 30 digestion with BamHI and XhoI to release the 7.4 kb ~rag-ment rnnt~;n;nJ the mMT-I/hGH fusion flanked on both sides by hEPO seguences. This fragment (targeting CL ~ t 2) contains no hEPO codi~g sequences, having only sequences lying between -1 and approximately -6620 upstream of the 35 hEPO coding region to direct targeting to the human EPO

WO 95/3156D 2 1 q 5 2 8 9 r~ ;r 1~

locus. Targeting fragment 2 is transfected into primary or secondary human skin fibroblasts u6ing conditions similar to those described in Example lg. G418-resistant colonies are picked into individual wells of 96-well 5 plates and screened for EPO expression by an ELISA assay (R&D Systems, Mi nn~.~rnl i c, MN) . Cells in which the trans-fecting DNA integrates randomly into the human genome cannot produce EP0. Cells in which the transfecting DNA
has undergone homologous re: ' in:~t;nrl with the endogenous 10 hEPO promoter and upstream seqllPnrPA contain a chimeric gene in which the mM~- I promoter and non- transcribed se~l~nnPc, hGH 5' untr~nAlF~tprl se~uences and hGh exon 1, and a 10 base pair linker comprised of the first 10 base6 of hEPO intron 1 are inserted at the HindIII site lying at 15 position -620 relative to the hEPO coding region (see Figure 2~. The lnn~l;7a~inn of the mMT-I promoter up-stream of the normally silent hEP0 promoter will direct the synthesis, in primary or sec~,l,dcl~y skin fibroblasts, of a message reading (5' to 3') non-translated metallo-20 thinnP;n and hGH se5~uences, hGH exon 1, 10 bases of DNA;~iPnt;c;-l to the first 10 base pairs of hEP0 intron 1, and the normal _EPO promoter and hEPO exon 1 (-620 to +13 relative to the hEP0 coding sequence). The 10 base pair linker se~luence from hEP0 intron 1 acts as a splice-donor 25 site to fuse hGH exon 1 to the next downstream splice acceptor site, that lying; ';~t~ly upstream of hEPO
exon 2. Procesging of the resulting transcript will therefore splice out the hEPO promoter, exon 1, and intron 1 se~uences. The r~rl~ of hEP0 exon 1 with hG~ exon 30 1 results in a protein in which the first 4 amino acids of the hEP0 signal peptide are replaced with amino acids 1-3 of hG~, creating a functional, chimeric signal peptide which is removed by po8t-tr~ncl at; on processing from the .
mature protein and is secreted from the expressing cells.

W O 95/3156 0 2 1 9 0 ~ ~ ~ I, ., ~ 5 ~ ~ 5 .

A 3eries of con6tructs related to pXEPO-10 and pXEPO-11 can be constructed, using known methods. In these constructs, the relative positions of the mMT-I promoter and hGH seguences, as well as the position at which the 5 mMT- I/hGH seguences are inserted into hEPO upstream se-~auences, are varied to create alternative chimeric tran-scription units that facilitate gene targeting, result in more efficient expression of the fusion transcripts, or have other desirable properties. Such constructs will 10 give similar results, such that an hGH-hEPO fusion gene is placed under the control of an PY~ nr~u~ ~romoter by gene targeting to the normal hEPO locus. For example, the 6 kb HindIII-BaA,lHI fragment upstream of the hEPO gene (See Example lf ) has numerous restriction enzyme recognition 15 sequences that can be l~t; 1 7F-~ as sites for insertion of the neo gene and the mMT-I promoter/hGH fusion ~ _ One such site, a BglII site lying approximately 1.3 kb upstream of the HindIII site, is uniS~ue in this region and can be used for insertion of one or more selectable mark-2 0 ers and a regulatory region derived f rom another gene thatwill serve to activate hEPO expression in primary, second-ary, or immortalized human cells.
First, the int~Ll ';~to plasmid pT164 is constructed by inserting the 6 kb HindIII-BamHI fragment (Example lf ) 25 lying upstream of the hEPO coding region into HindIII-samHI ~igested pBluescriptII SK+ (Str~ no, La~olla, CA). Plasmid pMClneoPolyA [Thomas, K.R. and Capecchi, M.R. ~LL 51:503-512 (1987); available from Stratagene, La.Jolla, CA] is digegted with BamHI and XhoI, made blunt-30 ended by treatment with the Klenow fragment of E. coli DNApolymerase, and the resulting 1.1 kb fragment is puri~ied.
pT164 is digested with BglII and made blunt-ended by treatment with the Klenow f ragment of E. col i D~A polymer-ase. The two preceding blunt-ended ~L _ ' ' ~ are ligated 35 together and transformed into competent E. coli. Clones WO 95/31560 ~ 1 ~ () 2 ~ q4~
.

with a single insert of the 1.1 kb neo fragment are iso-lated and analyzed by restriction e~zyme analysis to identify those in which the BglII site recreated by the fusion of the blunt XhoI and BglII sites is localized 5 1.3 kb away from the unique HindIII site present in plas-mid pT164. The resulting plasmid, pT165, can now be cleaved at the unique BglII site flanking the 5' side of the neo transcription unit.
n~ l eotide6 13 . 8 and 13 . 9 are llt; 1; 7ed in poly-10 merase chain reactions to generate a fragment in which themouse metalloth;nn~o;n I (mMT-I) promoter - hGI; exon 1 sequences are additionally fused to a 10 base pair frag-ment comprising a splice-donor site. The splice-donor site chosen corresponds to the natural hEPO intron 1 15 splice-donor site, although a larger number of splice-donor site~ or c~nqPnql~q splice-donor sites can be used.
The oligonucleotides (13 . 8 and 13 . 9) are used to amplify the approximately 0 . 73 kb mMT- I promoter - hG~ exon 1 r, ' from pXG~5 (Figure 5). The I ,l;f;Ptl fragment 20 (I L~I ' 7) is digested with BglII and ligated to BglII
digested pT165. The 1 ;~Atinn mixture is transformed into E. coli and a clone, r-1ntA;n;ns a single insert of frag-ment 7 in which the KpnI site in the mMT-I promoter is adj acent to the 5 ' end of the neo gene and the mMT- I
25 promoter is oriented such that transcription is directed towards the unique ~indIII site, i3 identified and desig-nated pXEPO-12.
13 . 8 5 ' ~ TCT GGTAC~TTGG TTTTTAAAl~C CAGCCTGGAG
BglII KpnI
(SEQ ID NO 12) The non-~ f~ e region of oligo 13 . 8 is identi-cal to the mMT-I promoter, with the natural KpnI
site as its 5' boundary. The boldface type -WO95~31560 2 1 9 0 2 3 9 r~
.

denotes a BglII site tail to convert the 5' bUUlldaL~ to a BglII site.
13 . 9 5 ~ TTTTAGATCT GAGTACTCAC CTGTAGCCAT TG~rG~T~"C
BglII
~SEQ ~D NO 13 ) The bn7 r7f~re region of oligos 13 . 9 denote hGH
sequences. The it~l r; 7D~7 region CULLC:D~U~1d~ to the first 10 base pairs of hEPO intron 1. The lln~7Prl inP~7 BglII site is added for plasmid con-struction purposes.
Plasmid pXEPO-12 can be used for gene targeting by digestion with BamHI and HindIII to release the 7 . 9 kb LL _ rnnt::l;n;n~ the neo gene and the mMT-I/hGH fusion flanked on both sided by hEPO seqUPnrPR. This fragment 15 (targeting fragment 3) contains no hEPO coding sequences, having only seguences lying between approximately -620 and approximately -6620 upstream of the hEPO coding regioTl to direct targeting upstream of the human EPO locus. Target-ing f1--_ 3 ls transfected into primary, se~ulldal~, or 20 immortalized human skin fibroblasts using conditions similar to those described in F _ le~ lb and lc. G418-resistant colonies are picked into individual wells of 96-well plates and screened for EPO expression by an E~ISA
assay (R~D Systemg, r~linnP~rnl; c M~) . Cells in which the 25 transfecting DNA integrates randomly into the human genome cannot produce hEPO. Cells in which the transfecting DNA
has undergone ~ 10~JOllC rP_ ' 'n~t;nn with the Pn~ln~Pnnllc hEPO promoter and u~L~c:alll CP~DnrPC contain a chimeric gene in which the mMT-I promoter and non-transcribed 30 sequences, hGH 5~ untr~ncl~tpd sequences, and hGH exon 1, and a 10 base pair linker comprised of the first 10 bases of hEPO in~ron 1 are inserted at the BglII site lying at 02~9 Wo 95~1560 1 ~ Is .

position approximately -1920 relative to the hEPO coding region. The lr~r~li,atirn of the mMT-I promoter upstream of the normally silent hEPO promoter will direct the synthesis, in primary, secondary, or immortalized human fibroblasts (or other human cells), of a message reading:
(5 ' to 3 ' ) nontranslated metallothionein and hGH sequenc-es, hGH exon 1, 10 bases of DNA identical to the first 10 base pairs of hEPO intron 1, and hEPO upstream reyion and hEPO exon 1 (from approximately -1920 to +13 relative to the EPO coding sequence). The 10 base pair linker se-quence from hEPO intron 1 acts as a splice-donor site to fuse hGH exon 1 to a downstream splice acceptor site, that lying; ~;~tF~ly upstream of hEPO exon 2. Processing of the resulting transcript will therefore splice out the hEPO u~LLea", sequences, promoter region, exon 1, and intron 1 sequences. When using pXEPO-10, -11 and -12, post-transcriptional processing of the message can be improved by using 'n vitro mutagenesis to ~ lmin~te splice acceptor sites lying in hEPO upstream ~esrlrnreR between the mMT-I promoter and hEPO exon 1, which reduce level of productive splicing events needed create the desired message. The r~rl ~ of hEPO exon 1 with hGH exon 1 results in a protein in which the ~irst 4 amino acids of the hEPO signal peptide are replaced with amino acids 1-3 of hGH, creating a fllnr~ir,n~l, chimeric signal peptide which is removed by post-trAn~l~t;rn processing from the mature protein and is secreted from the expressing cells.
~MPI,R 3. T~Rt~T~n MoDIFIc~TloN OF ~ N~ ~:C UPsTR~z~M
AND AMPI,IFICA~ION OF T~T~ TARGET~n ~.'~F
3 0 Human cells in which the hEPO gene has been activated by the methods previously described can be induced to amplify the neo/mMT-l/EPO transcription unit if the tar-.
geting plasmid contains a marker gene that can conf er resistance to a high level of a cytotoxic agent by the WO 9!j131560 ~ q r~ C~

~h~ - of gene amplif ication. Selectable marker genes such as dihydrofolate reductase (dhfr, selective agent is methotrexate), the multifllnrtinn~l caD gene [encoding carbamyl phosphate synthase, aspartate transcarbamylase, and dihydro-orotase; selective agent is N- ~rhnsFhnnn-acetyl) -~-aspartate ~PA~A) ], glutamine synth~t~e; selec-tive agent is ~~th;on;n~ s~ hn~;m;n~ ~MSX), and ~lPnns m;n~e ~ada; gelective agent ig an adenine n--rl~os;tl~), have been do~ e~, among other genes, to be ~ l; f; Ahl e o in immortalized human cell lines (Wright, J.A. et al.
P~oc. N~tl. Acad. Sci. USA 87:1791-1795 ~1990); Cockett, M.I. f~ al. Bio/Technolor~v 8:662-667 ~1990) ) . In these studies, gene amplification has been do_ ed to occur in a number of immortalized human cell lines. HT1080, He~a, MCF-7 breast cancer cells, K-562 l~l-kPm;~ cells, KB
carcinoma cells, or 2780AD ovarian carcinoma cells, among other cells, display _l;f;r~t;nn under appropriate selection conditions.
Plasmids pXEP0-10 and pXEP0-11 can be modified by the insertion of a normal or mutant dhfr gene into the unique XindIII sites of these plasmids. After transfection of ~T1080 cells with the appropriate DNA, selection for G418-resistance ~conferred by the neo gene), and ;~-nt;fication of cells in which the hEPO gene has been activated by gene targeting of the neo, dhfr, and mMT-1 sequences to the correct position upstream of the hEPO gene, these cells can be exposed to stepwise s~l ~c t; nn in methotrexate ~MTX) in order to select for: _l;f;r~t;nn of dhfr and co-ampli-fication of the linked neo, mMT-1, and hEP0 sequences ~Kaufman, R.J. Techniaue 2:221-236 ~1990) ) . A stepwise selection scheme in which cells are first exposed to low levels of MTX ~0.01 to 0.08 IlM), followed by successive exFosure to in. ~ Al increases in MTX concentrations up to 250 ~M MTX or higher is employed. ~inear in.:L~ ~1 35 steps of 0 . 04 to 0 . 08 I~M MTX and successive 2-fold in-W095/31~60 21 90~89 P~
.

crease6 in MTX concentration will be effective in select-ing for amplified transfected cell lines, although a variety of relatively shallow increments will also be effective. Amplification i5 monitored by increases in 5 ~ gene copy number and confirmed by measuring ~ vitro hEPO expression. By this strategy, substantial over-expression of hEPO can be attained by targeted r ~;f;r~A~-tion of seguences lying rn~ letPl y outside of the hEPO
coding region.
Constructs similar to those f1Pcrr; hP~a~ (Examples lf, lh, li, lk, 2 and 7) to activate hGH expression in human cells can also be further n a;fie~l to include the dhfr gene for the purpose of nhtA;n~n,r cells that YveLe,~L~as the hG~ gene by gene targeting to non-coding se~lpnrpc and 15 subseguent amplification.
T~XA~p~E 4. TARGETIN~ AND ACTIV~TION OF THE HYM~N T'PO
LOC~TS IN AN IMMt~RTAT,T~T~n H~AN FIBROBL~ST T,T~
The targeti~g construct pXEPO-13 was made to test the hypothesis that the ~-~dOye~ JUs hEPO gene could be activat-20 ed in a human fibroblast cell. First, plasmid pT22.1 wasconstructed, cnntA;n;n~ 63 bp of genomic hEPO seguence upstream of the first codon of the hEPO gene fused to the mouse metallo~l~;nnP;n-l promoter (mMT-I) . Oligonucleo-tides 22.1 to 22.4 were used in PCR to fuse mMT-I and hEPo 25 _eguences. The properties of these primers are as fol-lows: 22.1 is a 21 base nl;r,nn~lclp~tide ~ loJo~-q to a segment of the mMT-I promoter heJr;nn;n~ 28 bp upstream of the mMT-I KpnI site; 22.2 and 22.3 are 58 nucleotide complementary primers which define the fusion of hEPO and 30 mMT-I seriuences such that the fusion contains 28 bp of hEPO seguence beginning 35 bases upstream of the first codon of the hEPO gene, and mMT-I se~uences beginning at base 29 of oligonucleotide 22.2, comprising the natural BglII site of mMT-I and P~Pn~a~;nJr 30 bases into mMT-I

WO95/31560 ~1 q(~a9 1~.111 15 .

sequence; 22 . 4 i6 21 nucleotides in length and is homolo-gous to hEPO sequences b~;nn;n~ 725 bp downstream of the first codon of the hEPO gene. These primers were used to amplify a 1.4 kb DNA fL _ comprising a fusion of mMT-I
and hEPO sequences as described above. The resulting fragment was digested with KpnI (the PCR fragment con-tained two KpnI sites: a single natural KpnI site in the mMT- I promoter region and a single natural KpnI Eite in the hEPO sequence), and purified. The plasmid pXEPO1 was also digested with KpnI, releasing a 1.4 kb fragment and a 6.4 kb fragment. The 6.4 kb fragment was purified and ligated to the 1. 4 kb KpnI PCR fusion LLC~ - . The resulting construct was called pT22.1. A second interme-diate, pT22 . 2, was constructed by ligating the approxi -mately 6 kb HindIII-Bam~I fragment lying upstream of the hEPO structural gene (see Example lf) to BamHI and HindIII
digested pBSIISK+ (Stratagene, LaJolla, Q). A third ;nt, -';~t~, pT22.3, wag congtructed by first excising a 1.1 kb XhoI/BamHI fragment from pMCINEOpolyA (Stratagene,, I,aJolla, Q) r~nt~;n;n~ the neomycin pho~ oLLc~ Lerase gene . The ~ was then made blunt - ended with the Klenow f ragment of DNA polymerase I (New England Biolabs ) .
This - _ was then ligated to the HincII slte of pBSIISK+ (similarly made blunt with DNA polymerase I) to produce pT22.3. A fourth inteL, ';~te, pT22.4, was made by purifying a 1.1 kb XhoI/HindIII ~---_ comprising the neo gene from pT22.3 and ligating this fragment to XhoI
and HindIII digested pT22 . 2 . pT22 . 4 thus contains the neo gene adjacent to the HindIII side of the BamHI-HindIII
upstream hEPO fragment. Finally, pXEPO-13 was generated by first excising a 2 . O kb EcoRI/AccI fragment from pT22 . -1. The EcoRI gite of this fragment defines the 5' bound-ary of the mMT-I promoter, while the AccI site of this ~L _ - lies within hEPO exon 5. Thus, the AccI/EcoRI
,35 fragment contains a nearly complete hEPO expression unit, 2~ 9rj2~
Wo 9513156D ~ ''` l5 .

missing only a part of exon ~ and the ~atural polyadenyla-tion site. This 2.0 kb EcoRIjAccI fragment was purified, made blunt-ended by treatment with the Klenow fragment of DNA polymerase I, and ligated to XhoI digested, blunt-5 ended, pT22 . 4 .
HT1080 cells were transfected with PvuI-BamHI digest-ed pXEPO-13. pXEPO-13 digested in this way generates three LL _ C; a 1 kb vector fragment ;n~ 1in~ a por-tion of the amp gene, a 1.7 kb fragment of L~ ;n;n~
10 vector sequences and an approximately 9 kb fragment con-taining hEPO, neo and mMT-I ~e~u~:u~es. This approximately 9 kb BamHI/PvuI fragment cnnt~;nPd the following sequences in order from the BamHI site: an approximately 5 . 2 kb of upstream hEPO genomic sequence, the 1.1 kb neo transcrip-15 tion unit, the 0.7 kb mMT-I promoter and the 2.0 kb frag-ment cnnt~;n;n~ hEPO coding sequence truncated within exon 5. 45,ug of pEXPO-13 digested in this way was used in an electroporation of 12 million cells (electroporation conditions were described in Example lb). This electro-20 poration was repeated a total of eight times, resulting inelectroporation of a total of 96 million cells. Cells were mixed with media to provide a cell density of 1 million cells per ml and 1 ml aliquots were dispensed into a total of 96, 150mm tissue culture plates (Falcon) each 25 r~ nt~;nin~ a minimum of 35 ml of DMEM/15~ calf serum. The following day, the media was aspirated and replaced with fresh medium ~nnt:l;n;n~ 0.8 mg/ml G418 (Gibco). After 10 days of incubation, the media of each plate was sampled for hEPO by ELISA analysis (R ~ D Systems). Six of the 96 30 plates rnnt~;nf~d at least 10 mU/ml hEPO. One of these plates, number 18, was selected for purification of hEPO
expressing colonies. Each of the 96, 150 mm plates con-tained approximately 600 G418 resistant colonies (an estimated total of 57, 600 G418 resistant colonies on all 35 96 plates). The approximately 600 colonies on plate WO 9~/31560 2 1 9 ~ 2 8 q T ,n l5 .

number 18 were tryps~ n; 7e~1 and replated at 50 cells/ml into 364 well plates (Sterilin). After one week of incu-bation, single colonies were visible at approximately 10 colonies per large well of the 364 well plates (these plates are comprised of 16 small wells within each of the 24 large wells). Each well was screened for hEPO expre6-sion at this time . Two of the large wells cnntA; nPrl media with at least 20 mU/ml hEPO. Well number A2 was found to contain 15 colonies distributed among the 16 small wells.
The contents of each of these small wells were tryp8ln;7Pr and transferred to 16 individual wells of a 96 well plate.
following 7 days of ;nrllh~A~t;nn the media from each of these wells was sampled for hEPO ELISA analysis. Only a single well, well number 10, rnntA;nPd hEPO. This cell strain was designated HT165-18A2-10 and was PYrAntlPd in culture for guantitative hEPO analysis, RNA isolation and DNA isolation. Quantitative measurement of hEPO produc-tion resulted in a value of 2,500 milliunits/million cells/24 hours.
A 0.2 kb DNA probe Pytpnri;n~ from the AccI site in hEPO exon 5 to the BglII site in the 3 ' untrAnRl ~t~d region was used to probe RNA; ~lnl AtP~ from HT165-18A2-10 cells. The targeting construct, pXEPO-13, truncated at the AccI site in eYon 5 does not contain these AccI/BglII
se~uences and, therefore, is diagnostic for targeting at the hEPO locus. Only cell strains that have re~ ' inPd in a homologous manner with natural hEPO EPqnPnrP~ would produce an hEPO mRNA cnntA;n;n~ 9PT~Pn~-e homologous to the AccI/BglII s~ue~ s. HT165-18A2-10 was found to express an mRNA of the predicted size hybridizing with the 32-P
labeled AccI/BglII hEPO probe on Northern blots. ~estric-tion enzyme and Southern blot analysis confirmed that the neo gene and m~T- I promoter were targeted to one of the two hEPO alleles in HT165-18A2-10 cells.

WO 95/31560 2 1 ~ 0 2 8 9 ~ ,r l;
.

These re6ults demonstrate that homologous recombina-tion can be used to target a regulatory region to a gene that is normally silent in human ~ibrobla6ts, resulting in the functional activation oE that gene.
22.1 5~ r~rrT~h~T GAT~ cl~i G (SEQ ID NO 14) 22 . 2 5 ' ~ iC~ iL r.~rr~r~rrr. r,~ r ATCTGGTGAA
GCTGGAGCTA CGGAGTAA ( SEQ ID NO 15 ) 2 2 . 3 5 ' TTACTCCGT~ GCTCCAGCTT CACCAGATCT AG~;~
~L~ AC CCGGCGCG (SEQ ID NO 16) 22.4 5' GTCTCACCGT GATATTCTCG G (SEQ ID NO 17) Mpr,~ 5 . ~vL~u~ ~ lON OF INTRONr~q~ r~mr~c Gene targeting can also be used to produce a pro-cessed gene, devoid of introns, for transfer into yeast or bacteria for gene expression and Ln y~ protein produc-15 tion. For example, hGH can by produced in yeast by theapproach ~lP~r; hPd below.
Two separate targeting constructs are gPnPr~tPfl.
Targeting construct 1 (TC1) includes a retroviral LTR
se~uence, for example the LTR from the Moloney Murine 20 r.P~.kPm; ~ Virus (MoMLV), a marker for selection in human cells (e.g., the neo gene from Tn5), a marker for selec-tion in yeast (e . g ., the yeast URA3 gene), a regulatory region capable of directing gene expression in yeast (e.g., the GAI~4 promoter), and optionally, a sequence 25 that, when fuged to the hGH gene, will allow secretion of hGH from yeagt cells (leader seguence). The vector can also include a DNA sequence that permits retroviral pack-aging in human cells. The construct is organized such that the above sequences are flanked, on both sides, by 21 qO289 hGH genomic ser~uences which, upon homologous recombination with genomic hGH gene N sequences, will integrate the P~nrPnnUA 8e,r,uence8 in TC1 immediately upstream of hGH
gene N codon 1 (corrpqpnnr~;nJ to amino acid position 1 in 5 the mature, processed protein). The order of DNA sequenc-es upon integration is: hGX upstream and regulatory se-quences, neo gene, ~TR, ~JRA3 gene, GAIJ4 promoter, yeast leader sequence, hGX SPr~lPnrPR ;nrlllri;n ~ and downstream of amino acid 1 of the mature protein. Targeting Construct 2 10 (TC2) ;nrlll~iPA sequences suff;~iPnt for plasmid replica-tion in yeast (e.g., 2-micron circle or ARS sequences), a yeast transcriptional term;n~t;nn sequence, a viral LTR, and a marker gene ~or sPlf~t;nn in human cells (e.g., the h:lrtPr;~l gpt gene). The construct is organized such that 15 the above sequences are flanked on both sides by hGH
genomic SPrlllPnr~A which, upon i ]r~ lc re ' ;n~t;nn with genomic hGH gene N sequences, will integrate the uc sequences in TC2; ';;-tPly downstream of the hGH gene N stop codon. The order of DNA sequences upon 20 integration is: hGH exon 5 sequences, yeast transcription tPrm;n~t;nn 8equence8, yeast plasmid replication sequenc-es, I,TR, gpt gene, hGH 3 ~ non-tr;lnAl :~tl~ri 8PrlllPnrF.A.
~ inear rL - A derived from TC1 and TC2 are sequen-tially targeted to their respective positions flanking the 25 hGH gene. After superinfection of these cells with helper retroviruS, LTR directed transcription through this region will result in an RNA with LTR sequences on both ends.
Splicing of this RNA will generate a molecule in which the normal hGX introns are removed. Reverse transcription of 30 the processed transcript will result in the ~ lct;nn of double-stranded DNA copies of the processed hGX ~usion gene. DNA is isolated from the doubly-targeted, retro-virally- inf ected cellg, and digested with an enzyme that cleaves the trangcription unit once within the ~TR. The 3 5 digested material is ligated under conditions that promote WO 95/31560 2 1 9 0 2 8 ~

circularization, introduced into yeast cells, and the cells are 6ubsequently exposed to selection for the URA3 gene Only cells which have taken up the URA3 gene (linked to the sequences introduced by TC1 and TC2 and the processed hGH gene) can grow. These cells contain a plasmid which will express the hGX protein upon 5~1 ~ctnce ;n~1lrt;nn and gecrete the hGH protein from cells by virtue of the fused yeast leader peptide sequence which is cleaved away upon secretion to produce the mature, biolog-ically active, hGH molecule.
Expression in bacterial cells is ~ hP-l by simply replacing, in TC1 and TC2, the ~ ~;ll;n-resis-tance gene from pBR322 for the yeast ~A3 gene, the tac promoter ~deSoer e~ al., Proc. Natl . Acad. Sci . 80 :21-25 (1983) ) for the yeast GAL4 promoter, a bacterial leader sequence for the yeast leader gequence, the p~3R322 origin of rPrl; r~t; nn for the 2-micron circle or ARS sequence, and a bacterial transcriptional termination (e.g., trpA
transcription terminator; Christie, G.E. et al., P~oc.
Natl . Acad. Sci . 78 :4180-4184 (1981) ) sequence for the yeast transcriptional tf~rm;n~tio~ sequence. Similarly, hEPO can be expressed in yeast and bacteria by simply replacing the hGH targeting sequences with hEPO targeting sequences, such that the yeast or bacterial leader se-quence is positioned; '; ~t~l y upstream of hEPO codon l (corresponding to amino acid position 1 in the mature processed protein).
EXAMPLE 6. ACTIVATION AND AMP~IFICATION OF THE EPO GENE
IN AN Ir~ORT~T T~Rn HUM/~N (~T'T~T. LINE
3 0 Incorporation of a dhfr expression unit into the unique HindIII site of pXEPO-13 (see Example 4) results in a new targeting vector capable of dual selection and selection of cells in which the dhfr gene is amplified.
The single XindIII site in pXEPO-13 defines the junction WO95131560 21 9~28~ r~ '`15 of the neo gene and genomic sequence naturally residing upstream of the human EPO gene. Placement of a dhfr gene at this site provides a construct with the neo and dhfr genes surrounded by DNA sequence derived from the natural 5 hEPO locus. Like pXEPO-13, derivatives with the dhfr gene inserted are useful to target to the hEPO locus by homolo-gous re: ' ini~t;nn. Such a construct ~lP~i~ni~t~ pREPO4, is Le~L~8t:11Led in Figure 6. The plasmid includes exons 1-4 and part of exon 5 of the human EPO gene, as well as the 10 HindIII-BamHI fragment lying U~LL~ of the hEPO coding region. pSVe, pTK and pmMT-I correspond to the promoters from the SV40 early region, the Herpes Simplex Virus (HSV) thymidine kinase (TK) gene and the mouse metallothionein-I
gene. It was produced as follows: ~lindIII-digested 15 pXEPO- 13 was purif ied and made blunt with the Klenow LL _ ~ of DNA polymerase I. To obtain a dhfr expression unit, the plasmid construct pF8CIS9080 (Eaton et al., ~3iochemistrv j~: 8343-8347 ~1986) ) was digested with EcoRI
and SalI. A 2 Kb E~ _ cnnti~;n;ns the dhfr expression 20 unit was purified from this digest and made blunt with Klenow LL _ of DNA polymerase I. This dhfr-cnnti~;n;
LL _ was then ligated to the blunted ~indIII site of pXEPO-13 . An aliquot of this 1 igat;~n was t~
into ~ coli and plated on i ir; 11; n 8o1 ~rt; f~n plates .
25 Following an overnight ;nrllhi~ti~m at 37C, individual h~rtPr; i~1 colonieg were observed, picked and grown.
~;n;r1;~r~ L~ nFi were made from these cultures and the resulting D~A was then subjected to restriction enzyme digestion with the enzymes BglI+HindIII, and SfiI in order 30 to r~tPrm;n~ the or;~nti~t;~n of the inserted dhfr frag-ments. Plasmid DNA from one of these preparations was found to contain such a 2 Kb insertion of the dhfr frag-ment. The trangcription oriPnti~t;~n of the dhfr expres-sion unit in this plasmid was found to be opposite that of WO95/31560 2l 90~8~ r~ 45 the adjacent neo gene. Thi6 i8 the construct designated pREPO4 .
Plasmid pREPO4 was used to amplify the hEPO locus in cells subse~uent to activation of the /-n-lnrJ~nnllR hEPO gene 5 by homologous rc- ~ ;nAt;nn Gene activation with this construct allows s~lert;nn for ;nrr~ARe~9 D~FR expression by the use of the drug methotrexate (MTX). Typically, increased D~FR expression would occur by an increase in copy number through DNA _ l; f; cat i on . The net result 10 would be co-i _ 1; f; r~tion of the activated hEPO gene along with dhfr se~auences. Co- l;f;rAt;nn of the activated EPO 10CUB should result in increased EPO expression.
Targeting experiments were perfo med in E!T1080 cells with pREPO4 . hEPO expregging line EITREPQ-52 wa8 ; nnl At~
15 This line was analyzed S~uantitatively for EPO production and by r~ollth~-rn and Northern blot. This strain was found to be targeted with a single copy of &fr/neo/mMT-1 se-quences. Expression levels nhti~;n~l under 0.8 mg/ml G418 g~ rt;nn were apprn~;r-t~ly 1300 mtJ/million cells/day.
20 Because the targeted EPO locus rnntA; nf.r; a &fr expression unit, it was possible to select for increased expression of DHFR with the Ant;folAte drug, MTX. This strain was therefore subjected to stepwise s~le~t;nn in 0.02, 0.05, O .1, 0 . 2 and o . 4 ~LM MTX . Results of initial 8~1 ect; nn Of 25 this strain are shown in Table 4 and Figure 7.

WO95t31560 21 902~9 .C.,~ 5 mU/
Million Cells/
Cell T.in~ ~X(uM) 24h 52C20-5- ~01 0 . 01 1744 52C20-5- . 02 0 . 02 11643 52C20-5-0.05 0.05 24449 52-3-5-0.10 0.1 37019 52-3-2-0.20 0.2 67867 52-3-2-0.4B 0.4 99919 Selection with elevated levels of MTX was successful in increasing hEPO expression in li~e H~rREPO-52, with a 70-fold increase in EPO rro~letinn seen in the eell line resistant to 0.4 UM MTX. ~'nnf;rr~~;nn of i l;f;ri~;nn Of 5 the hEPO locus was ~ l; AhC~d by Sr~ut71~rn blot analysis in MTX-resistant eell lines, which revealed an approxi-mately 10-fold increase in the copy number of the activat-ed hEPO locus relative to the parental ~untargeted) hEPO
allele .
0 EX~MP~ 7: PRODUCTION OF AN hFPO FUSION GENE BY INSERTION
O~ THE CMV PROMOTER 1. 8 KB UPSTREAM OF THE GE-NOMIC hEPO CODING REGION
Construction of ti~rqet;nr Plasmid PREPO15:
pREPO15 was constructed by first fusing the CMV
15 promoter to hGH exon 1 by PCR l; f; rilt; nn . A 1. 6 kb ~ _ ' was i ~l;f;ed from hGH expression construct WO 95/31560 2 1 9 Q 2 PJ 9 ~ 5 pXGH308, which has the CMV promoter region be~;nn;n~ at nucleotide 546 and ending at nucleotide 2105 of Qer~bank sequence HS5MIEP fused to the hGH ~e~Pn~Pq beginning at nucleotide 5225 and ending at nucleotide 7322 o Genbank 5 sequence H~MGHCSA, using oligonucleotides 20 and 35.
Oligo 20 (35 bp, SEQ ID N0: 18~, hybridized to the CMV
promoter at -614 relative to the cap site (in Genbank sequence HEHCMVP1), and inr~ pr; a SalI site at its 5' e~d. Oligo 35 (42 bp, SEQ ID NO: 19), AnnPAled to the CMV
10 promoter at +966 and the adjacent hGH exon 1, and included the first 10 base pairs of hEPO intron l(cnnt7;n;n~ a portion of the splice-donor site) and a HindIII site at its 5' end. The resulting PCR fragment was digested with HindIII and SalI and gel-purified. ~lasmid pT163 ~Example 15 2 ) was digested with XhoI and HindIII and the approxi-mately 1.1 kb fragment cnntA;n;n~ the neo expression unit was gel-purified. The 1.6 kb CMV promoter/hGH exon 1/splice-donor site E _ and the 1.2 kb neo Eragment were ligated together and inserted into the HindIII site 20 of pBSIISK+ (Stratagene, Inc.). The resulting ;nt~ ';-ate plasmid (degignated pBNCHS) cnnt~inpd a neo expression unit in a trangcriptional orientation opposite to that of the CMV promoter/hGH exon 1/splice-donor site fragme~t).
A second inte:L~ te, pREPO~AU;n~TTT~ was uu..~LLuc Led by 25 first digegting pREP05 with HindIII. This released two ~L _ ' R of 1.9 kb and 8.7 kb, and the 8.7 Rb fragment rnnt:: ;n;n~ EP0 targeting ge~uei~c~g was gel purified and circularized by self-l;~At;nn. The resulting plasmid, pREPO5~TT;nrlTII, cnntA;nPd only non-coding genomic DNA
30 sequences normally residing upstream of the hEP0 gene.
This included gequence from -5786 to -1 relative to EPO
exon 1. The 2.8 kb ~L__ ' cnntAin;n~ neo, the CMV
promoter, hGH exon 1, and the splice-donor site was ex-cised from pBNCHS with HindIII and gel-purified. This 35 fragment was made blunt with the Rlenow fragment of DNA
.

Wo 95~1560 ~ r~ ,3~ 4;
-as-polymerase I (New England Biolabs, Inc. ) and ligated to BglII-digested and blunt-ended pREPO5~HindIII. BglII cuts at a position -1779 bp upstream of hEPO exon 1 in pREPO5~HindIII. The resulting construct, pREPO15 ~Figure 8), rnntA;nr~ EPO u~La~lll sequences from -5786 to -1779 relative to the hEP0 coding region, the neo expression unit, the CMV promoter, hGH exon 1, a splice-donor site, and sequences from -1778 to -1 bp upstream of the hEPO
coding region, with the various elements Al ~ 7, in the order listed, 5 ' to 3 ' relative to nllrl ~.nt; t~ s~ nre of the hEP0 upstream region. For transfection of human cells, pREPO15 was digested with Not I and PvuI to liber-ate an 8 . 6 kb targeting fragment. The targeting fragment ~nntPI;n.orl firgt and gecond targeting sequences of 4.0 kb and 1. 8 kb, respectively, with homology to DNA upstream of the hEPO gene.
Cell culture. transfection. and identification of EP0 e~ressinq tarqeted clones:
All cells were -~;ntA;n-~-l at 37C, 596 CO, and 98 humidity in DMEM rnntA;n;n~ 10~; calf serum (DMEM/10, HyClone TAhorAtnrieg). Transfection of Flecondary human foreskin fibroblasts was pe.L~ ' by electroporating 12 x 106 cells in PBS (GIBCO) with 100 ~Lg of DNA at 250 volts and 960 ~LF. The treated cells were seeded at 1 x 106 cells per 150 mm plate. The following day, the media was changed to DMEM/10 rnntA;n;n~ 0.8 mg/ml G418 (GIBC0) .
Selection ~-u~:~eded for 14 days, at which time the media was sampled for EP0 prr~ll-rt;r~n All colonies on plates exhibiting si,n;f;rAnt hEP0 levels (~ 5 mU/ml) as deter-mined by an EPO ELISA (Genzyme Inc. ) were; qo~ AteC~ with sterile glass cloning cylinders (Bellco) and transferred to individual wells of a 96 well plate. Following incuba-tion for 1-2 days, these wells were sampled for hEP0 pro-duction by ELISA. Resulting hEP0-producing cell strains WO 95~31560 ~ ~ 9 ~
so were ~ n~fl in culture for freezing, nucleic acid isola-tion, and s~ nt;fi;3tinn Of EP0 p~n~ll,t;nn.
Transfection of HT1080 cells (ATCC CCL 121) was performed by treating 12x 106 cells in PBS (GIBCO) with 45 ~g of DNA at 450 volts and 250 ~F. Growth and ;~lontifir~-tion of clones O.:c:u-Lèd as for secondary human foreskin fibroblasts described above . T~ol ;Itinn of hEP0 producing clonal cell lines o..uLLed by limiting ~ t;nn. This was performed by first plating colonies harvested from the 10 initial selection plates iR pools of 10-15 colonies per well of a 24 well plate. hEP0 producing pools were then plated at cell ~i~.n~it;~ resulting in ~ 1 colony per well of a 96 well plate. Individual clones were t~ nfl/~d for further analysis as described for human foreskin fibro-15 blasts abovê.
Characterization of EP0 ex~re~sinq clones:
pREPO15 is devoid of any hEPO coding sequence. Upontargeting of the ReO/CMV promoter/hG~ exon l/splice-donor ~L _ upstream of hEP0 exon 1, hEP0 expression occurs 20 by transcriptional initiation from the CMV promoter, producing a primary transcript that i~cludes CMV sequenc-es, hGH exon 1 and the splice-donor site, 1. 8 kb of up-strearn hEPO ~eSrl~n~oR, and the normal hEP0 exons, introns, and 3 ~ untr~n~l ~t~i sequences . Splicing of this tran-25 script would occur from the splice-donor site adjacent to hG~I exon 1 to the next dowRstream splice-acceptor site, which is located adjaceRt to hEPO exon 2. Effectively, this results in a new intron consisting of genomic se-quence upstream of the hEPO gene, the normal hEPO promot-3 0 er, hEP0 exon 1, and hEP0 intron 1. In the mature tran-script, hGH exon 1 would replace hEPO exon 1. hEPO exon 1 encodes only the first four and one-third amino acids of the 26 amino acid signal peptide, which is cleaved off of the precursor protein prior to secretion from the cell.

WO95~1560 21 ~Q2aq ~lIL~ 4~
.

hGH exon 1 encodes the first three and one-third amino acids of the hGH signal peptide, which also is cleaved off of the precursor protein prior to secretion from the cell.
TrAnAlAt; nn of the message in which hGH exon 1 replaces hEPO exon 1 would therefore result in a protein in which the signal peptide is a chimera of hGH and hEPO sequence.
Removal~of the signal peptide by the normal post-transla-tional cleavage event will produce a mature hEPO molecule whose primary sequence is indistinguishable from the normal product.
Transfection of pREPO15 into human f;hrnhlAAts re-sulted in EPO expression by these cells. Table 5 shows the results of targeting experiments with pREPO15 in human fibroblasts and HT1080 cells. The targeting frequency in normal human fibroblasts was found to be 1/264 G418r rn1nn;PA, and the targeting frequency with HT1080 cells was found to be 1/450 G418r colonieg . hEPO pro~ t; nn levels from each of these cell strains was quantif ied . An hEPO producer nhtA;n~d from transfection of human fibro-blasts was found to be secreting 7,679 mU/ 106 cells/ day ~Table 5). An activated hEPO cell line from HT1080 cells was producing 12,582 mU/106 cells/ day (Table 5). These results ;n~;rAted that activation of the hEPO locus was efficient and caused hEPO to be ~ e1 constituitively at relatively high levels. Restriction enzyme and South-ern hybr;-l;7Atinn analysis was used to confirm that tar-geting events had OC~,UL I e:d at the EPO locus.
~o--tll~rn blot analygig of the human f;hrnhlAAt and HT1080 clones that were targeted with pREPO15 was per-formed. Figure 9A shows the restriction map of the parental and targeted hEPO locus, and Figure 9B shows the results of restriction enzyme and Southern hybr;~;7At;on analysis of a targeted human f ibroblast clone .
BglII/Eco~I and Bam~I digests revealed 5. 9 and 6 . 6 kb 35 ~L, ~A, respectively, as a result of a targeting event WO95~31560 -92- ;~ 1 9 a 2 ~ .'û15 at the hEP0 locus ~lanes Tl). Both of these fragments resulted from the insertion of 2.7 kb of DNA cnnt~;ninrJ
the neo gene and CM~ promoter se~uence6. Since only one of the two hEPO alleles were targeted, rL _ q Qf 4.3 kb 5 ~BglII/EcoRI) or 10.6 kb ~BamHI) rPfl~rt;nJr the unaltered hEPO locu8 were seen in these strains and in parental DNA
~lanes HF) . These results confirm that a ~ ~lor,o -~1 L~ ' n:lt;r~n event had occurred at the hEP0 locus result-ing in the pro~l~rtiC~n of a novel tr~nRr~;rt;~n unit which lO directed the production of human erythropoietin.
Oli~onucleotide Seauence 20 5 ' ~ aG Trr~rr~r~T TGATTATTGA CTAGT
~ SEQ ID N0: l 3 ) 35 5 ' TTTTAAGCTT GAGTACTCAC rTt~T~r~rr~T GGTGGATCCC GT
~SEQ ID NO: l9) WO 95/31~60 2 1 ~ ~ 2 ~ 9 . I, ~ ~, A l5 .

Table 5. Transfection of pREPO15 and Activation of hEPO
Expression in Human Cells hEPO 'hEP0 Tr~nsieY-P Cell~ Colonles WLth 7rpo p~r G418r ed Cel; oel 3~2 ~te~ Treate~ ~ Colony hr) Human Fibro- 3.3 x 10~ 264 1 1/26q l/3.3 X 107 7679 bla3t6 HT1080 3.1 x lO' Z700 6 1/~150 1/5.2 x 10~ 12,582Cell3 estimated by ~o-lnt;n~ colonies on 2 plates, averaging the results and extr~rol ~t i n~ to the total number of plates b medium from plates with G418r colonies was sampled for EPO
E~IgA analysis and those exhibiting hEPO levels greater than 5 m'J/ml were counted as EPO activation events C cluantitative hEPO pro~llrt;nn was ~-t~-7~ n~ from human fibroblast strain, XF342-15 or XT1080 cell line, ~MP, E 8: ~vvu~llON AND AMP~IFICATION OF AN h'~PO EUSION
BY 1~5~;~L l~.)N OF THE CMV PROMOTE'~ 1. 8 '~3 ~PSTREAM OP TXE GENOMIC hEPO CODING REGION
Construction of tarqetinq ~,~lasmid ~REPO18:
pREPO18 (Fiqure 10) was ~:Ull~LLlL-;Led by insertion of a dhfr expression unit at the ClaI site located at the 5 ' end of the neo gene of pREPO15. To obtain a dhfr ex-pression unit, the plasmid construct pF8CIS9080 [Eaton et al ., Biochew ~ ~trv 25: 8343-8347 (1986) ] was digested with EcoRI and SalI. A 2 cb fragment ~nnt~;n;n~ the dhfr expression unit was purified from this digest and made blunt by treatment with the Klenow fragment of DNA poly-merase I. A ClaI linker (New England Biolabs) was then llgated to the blunted dhfr fragment. The products of WO 9~131560 2 l 9 0 2 8 9 I CI,~,.. ^ l5 .

this ligation were then digested with ClaI ligated to ClaI
digested pREPO15. An aliquot of this ligation was trans-formed into ~. coli and plated on: ;rill;n ~}Plect;nn plates. Bacterial colonies were analyzed by restriction 5 enzyme digestion to 61Ptprm;np the nr;Pnt~t;nn of the inserted dhfr fragment. One plasmid with dhfr in a transcriptional ori~nt~t i n~ opposite that of the neo gene was designated pREPO18(-). A second plasmid with dhfr in the same transcriptional oriPnt~t; nn as that of the neo 10 gene was ~PR jrjn~tP~ pREPOlB (+) Cell culture transfection, ~n~ identification of EPO
PYnressin~ tar~eted clones:
All cells were ~-;nt~;nPd at 37C, 5% CO2, and 98%
humidity in DMEM rnnt~;n;nrJ 10% calf serum (DMEM/10, 15 ~yClone Laboratories). Transfection of HT1080 cells ~ATCC, CC~ 121) occurred by treating 12x 106 cells in PBS
~GIBCO) with 45 ~Lg of DNA at 450 volts and 250 IlF. The treated cells were seeded at 1 x 106 cells per 150 mm plate. The following day, the media was changed to DMEM/10 cnnt~;n;n, 0.8 mg/ml G418 ~GIBCO) . Selection proceeded for 14 days, at which time the media was sampled for hEPO pro~ rt;nn. Plates IYh;h;t;n~ significant hEPO
pro~ rt; nn levels ~ ~ 5 mU/ml ) as ~ptprTn; np~ by an hEPO
E~ISA ~Genzyme Inc.) were tryps;n;7P~l and the cells were re-plated for clone isolation. T~ol~t;rn of h~pO produc-ing clonal cell lines occurred by limiting tl;lllt;nn, by first plating clones in pools of 10-1~ c~lnn;e~ per well of a 24 well plate, and next plating cells from hEPO pro-ducing pools at cell densities re6ulting in less than 1 colony per well of a 96 well plate. Individual clones were PYr~n~lP~ in culture for freezing, nucleic acid isola-tion and quantif ication of hEPO production .

2 1 902~
WO95131560 I~ IIL)..._.~ '~4r IsolatiQn of cells c~nt=;n;n~ am31ified dhfr seauen~es bY
methotrPYRte select; nn Targeted G418r cell lines producing hEPO following transfection with pREPO18 were plated at various cell densities for selection in methotrexate (MTX). As new clones emerged following 8P1 ect; nn at one MTX rnnrPntr~-tion, they were assayed for hEPO production and re-plated at various cell ~lPnci t1 PR in a higher concentration of MTX
(usually double the previous cnnnpntr~t; nn) . This process was repeated until the desired hEPO pro~rt; nn level was reached. At each step of MTX-resistance, DNA and RNA was ; Rnl =tP.1 for re8pective southprn and northern blot analy-sis .
mh=racterization of EPO exoressin~ clones:
pREPO18, with two different oriPntAt;nnR of dhfr, was transfected into HT1080 cells. Prior to transfection, pREPOla (+) and pREPO18 (-) were digested with XbaI, releas-ing a 7.9 kb targeting LL - rnntR;n;n~ in the follow-ing order, a 2.1 kb region of genomic DNA upstream of hEPO
exon 1 (from -3891 to -1779 relative to the hEPO ATG start codon), a 2 kb region cnntA;n;n~ the dhfr gene, a 1.1 kb region cnntR;n;n~ the neo gene, a 1.5 kb region contain-ing the CMV promoter fused to hGH exon l, 10 bp of hEPO
intron 1 (rnntR;n;ns a splice-donor site), followed by a 1.1 kb region of genomic DNA u~LLe:clul of hEPO exon 1 (from -1778 to -678 relative to the EPO ATG start codon).
Transfection and targeting frP~lpn~;pc from two experi-ments are shown in Table 6. Five primary G418r clones were; Rol=tPd from these experiments. These were in culture for quantitative analysis of hEPO expression (Table 7). As pREPO18 rnnt=;np~l the dhfr gene, it is possible to select for cells rnntR;n;n~ amplified copies of the targeting construct using MTX as described in Example 6. G418r clones r nnf; -~ to be targeted to the WO95131~60 2 ~ 9 02 8 9 r~ c ~l.

hEPO locus by restriction enzyme and Southern hybridiza-tion analysis were sub; ected to stepwise selection in MTX
as ~e~-r; hed .
Table 6- Targeting of pR13PO18 in HT1080 cells hEP0 Plates E,.~ Primary DNA Cell6 G418r With hEP0 /G418' Clo~es Collstruct Dig~st Treated Colorlies r Colo ly Analyzed pREP018 X~aI 36 x 10' 16,980 39 1/435 t ~
pREP018 XbaI 36 x 10' 19,290 41 1/470 4 (+~
Table 7. hEPO pr~ ti~n in HT1080 Cell lines targeted with pREpO18 Cell Line Co~truct hl3PO mU/lO' Cell~/24 hr 18B3_147 pREPO18 (+) 24759 18i33-181 pREPO18 (+) 20831 18B3-145 pRE~PO18 (+) 17586 18i33-168 pREPO18 (+) 5293 18A3-119 pREPO18 (-) 2881 2 l 9D2~9 WO 95/31560 I ~ C ~
.

r~ MpI~E 9: ACTIVATION AND AMPI,IFICATION OF ~ NLO~ ,.JU~i KF'I-:K(~N, GM-CSF. G-CSF AND FSHB f~ IN
RT~r,T~n H~N C~S
A wide variety of Dn~ln~nmlR cellular genes can be 5 activated and: _ l; f i ~rl using the methods and DNA con-structs of the invention. The following ~lDRrr; hDR a general strategy for activating and amplifying the human -interferon (leukocyte interferon), GM-CSF (colony stimu-lating factor-granulocyte/macrophage~, G-CSF (colony 10 8t; lAt;n~ factor-granulocyte) and FSH,B (fnll;rl~ stimu-lating hormone beta subunit) genes.
~-int~rferon The human cr-interferon gene (Genbank seSr~Dnre HUMIFNAA) encodes a 188 amino acid ~Le~ uL~Ol protein 15 cnntA;nin~ a 23 amino acid signal peptide. The gene contains no intron6 . Figure 11 g ' t; cally illustrates one strategy for activating the ct-interferon gene. The targeting ~DLLu~:L is ~DR;~n~d to include a first target-ing sequence 1 1 n~ollR to RD~lDnr~.R U~DLL - of the gene, 20 an amplifiable marker gene, a sDlDctAhle marker gene, a regulatory region, a CAP site, a splice-donor site, an intron, a splice acceptor site, and a second targeting sequence ~:uLL-~lJ~,tl;n~ to sequences downstream of the f irst targeting SDq~Dnre . The second targeting sequence 25 should not extend further u~DLL.- than to poQ;t;nn -107 relative to the normal start codon in order to avoid undesired ATG start codons.
In this strategy the f irst and second targeting Dec~u~ces are; ';~tDly adjacent to each other in the 30 normal target gene, but this is not required (see below).

7`-rl;fi~hle marker genes and QDlert~hle marker genes suitable for selection are described herein. The amplifi-able marker gene and SDl ectAhl e marker gene may be the same gene, their pogitions may be reversed, and one or Wo 9~l31560 2 1 9 0 2 ~ 9 ~ 45 _9~,_ both may be situated in the intron of the targeting con-struct. A selectable marker gene il3 optional and the ampl; ~i Ahl ~ marke~ gene is only required when amplifica-tion is desired. The incorporation of a specific CAP 3ite 5 is optional. Optinn~l ly, exon sequences from another gene can be ; n~ 3 ' to the splice-acceptor site and 5 ' to the second targeting S~1onr~ in the targeting construct.
The regulatory region, CAP site, splice-donor site, intron, and splice acceptor site can be ;RolAtf~d as a 10 complete unit from the human ~lnn~:lt;nn factor-l~r (EF-lc~;
Genbank sequence HUME~lA) gene or the cytomegalovirus (CMV; Genbank sequence HEHCMVPl) ; ~; Ate early region, or the ~ - q can be assembled from appropriate compo-nents; ~ol Atec~ from different genes .
Genomic DNA c uLL~ ; n~ to the upstream region of the ~Y-interferon gene for use as targeting sequences and assembly of the targeting ~ LLu~ L can be performed using recombinant DNA methods known by those skilled in the art.
As described herein, a number of selectable and amplifi-20 able markers can be used in the targeting ~ LU~_L8, and the activation and I l;f;rAt~nn can be ~ffert~d in a large number of cell-types. Transfection of primary, secondary, or immortalized human cells and ;qolAt;on of h~ lo~ollqly L~ ~ nAnt cells expre8ging ~-interferon can 25 be A~ ,l;qhP~l using the methods described in Example 4, using an ELISA assay for human a-interferon (Biosource Tnt~rnAt; nnAl, Camarillo, CA) . Alternatively, homo-logously LeC nAnt cellg may be i~l~nt;f;~ by PCR
screening as described in Example lg and lj. The isola-30 tion of cells rnntA;n;n~: ,l;f;e-~ copies of the amplifi-able marker gene and the activated ~-interferon locus is perf ormed as described in Example 6 .
In the homologously r-~: ;nAnt cells, an mRNA pre-cursor is produced which includes the exogenous exon, 35 splice-donor site, intron, splice-acceptor site, second WO 9S/31560 2 1 ~ 3 2 8 ~
.

targeting sequence, and human ~-interferon coding region and 3 ' untrAnCl At~i qf~qu~n~q ~Figure 11) . Splicing of this message will ~n~Ate a f~mrt;nnAl mRNA which can be - translated to produce human IY-interferon.
The size of the intron and thus the position of the regulatory region relative to the coding region of the gene may be varied to opt;m;7~ the function of the regula-tory region. Multiple exons may be present in the target-ing construct. In A~,l;tinn, the second targeting s~oqu~n~e does not need to lie; -~;At~-ly adjacent to or near the first targeting qPq~ nr.o in the normal gene, such that portions of the gene~s normal u~L,~ region are deleted upon homologous r,o: ' inA~;nn GM- CSF
The human GM-CSF gene (Genbank SeSr~Pnre ~MGMCSFG) encodes a 144 amino acid ~Le~:uL~ protein cnntA;n;ns a 17 amino acid signal peptide . ~he gene contains f our exons and t_ree introns, and the N-terminal 50 amino acids of the ple~:uL r-u . are encoded in the f irst exon . Figure 12 20 5~ h ; ~-Al ly illustrates a strategy for activating the GM-CSF gene. In this strategy the targeting construct is qi~n~d to include a first targeting s~ n~e 1 lo~o~q to se~uences u~ LL ~.. of the gene, an l; f; Ahl e marker gene, a q~lectAhle marker gene, a regulatory region, a C7~P
25 site, an exon which encodes an amino acid sequence which is ;~nt;rAl or fl~nrtinnAlly equivalent to that of the first 50 amino acids of GM-CSF, a splice-donor site, and a second targeting se~auence ~:uLl-q~ ;n~ to q~ nn~q downstream of the first targeting sequence. By this 30 strategy, ~ lo~ollqly ~. ;nAnt cells produce an mR~A
precursor which ~uLLe~u--ds to the ~ ~''y-11'"'4 exon and splice-donor site, the gecond targeting ses~uence, any sequences between the second targeting se~uence and the start codon of the GM-CSF gene, and the exons, introns, . , 2 ~ ~Q289 WO9~131560 ~.,I/~J,. l ~q;J~
.
,-100 -and 3 ' untrAncl At/~d region of the GM-CSF gene (Figure 11) .
Splicing of this message results in the fusion of the exogenous exon to exon 2 of the f~n~lnr,~nnus GM-CSF gene which, when translated, will produce GM-CSF.
In this strategy the f irst and second targeting sequences are; ';At~ly adjaaent in the normal target gene, but this is not required (see below). ~ l;f;Ahle marker genes and s~l ectAhl e marker genes suitable for s~lectinn are described herein. The amplifiable marker gene and selectable marker gene can be the same gene or their positions can be reversed. A selectable marker gene is optional and the; l; f; Ahl ~ marker gene is only re-quired when lifirAtinn is degired. The selectable marker and/or l;f;Ahle marker can be positioned between the splice-donor site and the second targeting sequence in the targeting construct . The incu ~ ~, . cLLion of a specif ic CAP site is optional. The regulatory region, CAP 8ite, and splice-donor site can be; Rnl At~C~ as a complete unit from the human elongation factor-l~ (EF-1~; Genbank se-quence E~UMEFlA) gene or the cytomegalovirus (CMV; Genbank ser,urnr~- ~EE;CMVP1) ; '; Ate early region, or the compo-nents can be assembled from an appropriate _ ; col At-~tl from different geneg (8uch as the mMT-I promoter and CAP site, and exon 1 and a gplice donor site from the hGH or hEPO genes.
Other approaches can be employed, for example, the first and second targeting sequences can cn~r~qpnnd to sequences in the first intron of the GM-CSF gene. Alter-natively, a targeting construct similar to that described for the ~-inter_eron can be used, in which the targeting construct ig designed to include a first targeting se-quence homologous to sequences upstream of the GM-CSF
gene, an amplifiable marker gene, a selectable marker gene, a regulatory region, a CAP site, a splice-donor site, an intron, a splice acceptor 8ite, and a second WO95~31560 ~ 289 ~ S~ q4;

targeting sequence corresponding to ser~uences downstream of the ~irst targeting se~luence.
In any case the second targeting seS~uence does not need to lie i ~ tPly adjacent to or near the first targeting seS~uence in the normal gene, such that portions of the gene ' 8 normal upstream region are deleted upon h, lorJollc re. ' n~ti~nn. In addition, multiple non-coding or coding exons can be present in the targeting construct. Genomic DNA ~uLL~7~ ;n~ to the upstream or lO intron regions of the human GM-CSF gene for use as target-ing sequences and assembly of the targeting ~ ~JllD~U~ ~ can be performed using .. ' ;n~nt DNA methods known by those skilled in the art. As described herein, a number of selectable and: , l; f; ~hle markers can be used in the 15 targeting constructs, and the activation can be effected in a large number of cell-types. Transfection of primary, se~_~,l.d-Ly~ or immortalized human cells and ;col~t;nn of homologously l~ ' n~nt cells expressing GM-CSF can be a: _ 1; chPrl using the methods described in Example 4, 20 using an ~3~ISA assay for human GM-CSF (R&D Systems, Minne-apolis, MN). Alternatively, 1 lo~ol-cly ,, n:~nt cells may be ; ~iPnt; f; prl by PCR scree~ing as described above. The ;col~t;nn of cells rnnt~;n;n, amplified copies of the . ~ l; f; ~hle marker gene and the activated GM-CSF
2~ locus is performed as described above.
G-CSF
The human G-CSF gene (Genbank 6equence ~UMGCSFG) encodes 204-207 amino acid ~Lt:~:uLDOl protein rnnt~;n;n,r a 30 amino acid signal peptide. The gene rnnt~;nc five 30 exons and four introns. The first exon encodes 13 amino acids of the signal peptide . Figure 13 5rh t; cally illustrates a gtrategy for activating the G-CSF gene. The targeting construct is designed to include a f irst target -ing seg~uence homologoug to ~cpslupnrp~ upstream of the gene, WO 9~131~60 , ~ ' 'Ofird li~
21 9~q an i ,l;f;Ah~e marker gene, a 8PlPrtAhl~- marker gene, a regulatory region, a CAP site, an exon which encodes an amino acid sec~uence which is ;~lont;ri~l or functionally ecauivalent to that oi the f irst 13 amino acids of the 5 G-CSF signal peptide, a splice- donor site, and a second targeting secluence iul L. _1-' .".l:n~ to serl--PnrPC ~'.c - ~ ecu of the first targeting sec~uence. By this strategy, homo-logously rP~ nAnt cells produce an mRNA ~)Le-UL~:lUL which corresponds to the PYnc3Pnmlc exon and splice-donor site, 10 the second targeting beqnPnnP ~ any spqnpnrpq between the second targeting secluence and the start codon of the G-CSF
gene, and the exons, introns, and 3~ llnt~AnclAted region of the G-CSF gene (Figure 13). Splicing of this message results in the fusion of the ~ c exon to exon 2 of 15 the enduye~,ous G-CSF gene which, when trAnCl AtPd, will produce G-CSF. The ability to fllnt-t;nnAlly s~hst;t~t~ the f irst 13 amino acids of the normal G-CSF signal peptide with those present in the e~wyel~uub exon allows one to make ' f;~At;nnq in the signal peptide, and hence the 20 secretory properties of the protein ~l-,ducied.
In this strategy the _irst and second targeting ~qP,~ n, Pc are i ~;AtPly adjacent in the normal target gene, but this is not rec~uired. The second targeting secluence does not need to lie; ';AtPly adjacent to or 2S near the first targeting secIuence in the normal gene, such that portions of the gene' 8 normal upstream region are deleted upon ~ ~lg~o~ q L ~ ' n A t; nn . The i , l; f i Ahl e marker gene and ~qplectAhlp marker gene can be the same gene or their positions can be reversed. A sPlpc~ti~hle 30 marker gene is optional and t~e: _l;f;AhlP marker gene is only rec~uired when _l;f;nAt;nn is desired. The select-able marker and/or l;f;AhlP marker can be poc;t;nnP~l between the splice-donor site and the second targeting -secluence in the targeting construct . The incorpnrat; nn of 35 a spen; f; n CAP site is optional . The r~gulatory region, .

W0 95131560 ~ Z 8 9 r~ 5 .

CAP site, and splice-donor site can be i Rnl i~ted as a complete unit from the human elongation factor-la (EF-la;
Genbank seriuence E}UMEF3 A) gene or the cytomegalovirus (CMV; Genbank ser~uence HEHCMVPl) i ~; Ate early region, 5 or the ~ ,u~.ellLs can be i~Ar ' l~fl from an appropriate . ~ isolated from different genes (such as the mMT-I
~L~ ~r and CAP 8ite, and exon 1 and a splice donor site from the hGH or EPO genes. Multiple ~Ynj~nn~lR exons, coding or non-coding, can be used in the targeting con-10 struct 80 long as an ATG start codon which, upon splicing,will be in-frame with the mature protein, is ;nrl~flod in one of the exons.
Other approaches may be employed, for example, the f irst and second targeting se~Iuences can culLc::,uul.d to 15 seruences in the first intron of the G-CSF gene. Alterna-tively, a targeting construct similar to that described ~or the a-interferon can be used, in which the targeting construct is fl~q;~nf~fl to include a first targeting se-quence ~ I19~JOllR to 8e~r~uence8 upgtream of the G-CSF gene, 20 an l;f;i~hle marker gene, a sf~ ctiqhle marker gene, a regulatory region, a CAP site, a splice-donor site, an intron, a splice acceptor site, and a second targeting 8e~r~ nre ~:ULL- ~.. fl;nrJ to 8eqn~nr~q downstream of the first targeting seriuence.
Genomic DNA CULL - '-L" 1 i n~ to the upstream or intron regions of the human G-CSF gene for use as targeting seriuences and assembly of the targeting construct can be performed using L~ ;nilnt DNA methods known by those skilled in the art. A8 described herein, a number of selectable and i ~l;f;Ahle markers can be used in the targeting constructs, and the activation can be effected in a large number of cell-types. Transfection of primary, secondary, or immortalized human cells and isolation of lor,ol~qly re~ ' ;ni~nt cellg expressing &-CSF can be i,: 1; Rh~ u8ing the methodg described in Example 4, . .

W095/3156U ~ P~l/_9'/l'~4' .

using an ~LISA assay for human G-CSF ~R&D Systems, Minne-apolis, MN). Alternatively, homologously l~ ;nAnt cells may be ; ~iPnt i f; oc~ by PCR gcreening as ~ocrr; hP~l above. The ;qnlAtinn of cells cnntAin;ng: ,l;f;ed copies 5 of the amplif iable marker gene and the activated ~-interferon locus iæ performed as described above.
FSHB
The human PSHB gene (Genbank seo~uence HUMFSX1) en-codes a 129 amino acid ~ e:eULQC1L protein ~nnt~;n;n~ a 16 10 amino acid signal peptide. The gene contains three exons and two introns, with the first exon being a non-coding exon. The activation of FSHB can be accomplished by a number of strategies. One strategy is shown in Figure 14.
In this strategy, a targeting construct is designed to 15 include a first targeting sequence 1- logo~q to 8T'n'`
upstream of the gene, an _l;f;Ah1e marker gene, a selec-table marker gene, a regulatory region, a CAP site, an exon, a splice-donor site, and a second targeting sequence corroqpnnrl;n~ to soquonroq downstream of the first target-20 ing sequence. By this strategy, homologously re ~ ;nAntcells produce an mR~A precursor which ~ULLe~ 11d8 to the PY~onn~R exon and 8plice-donor site, the second targeting sequence, any sequences between the second targeting sequence and the start codon of the FSHB gene, and the 25 exons, introns, and 3~ untrAnq~Ato~l regions of the FSHB
gene (Figure 14). Splicing of this message results in the fusion of the PYn~Qnn~q exon to exon 2 of the ~ J~
FSHB gene which, when trAnqlAto~, can produce FSHB. In this strategy the f irst and second targeting sequences are 30 ; '~Atoly adjacent in the normal target gene, but this i6 not reguired (see below).
Other approaches can be employed, for example, the f irst and second targeting sequences can correspond to sequences in the first intron of the FSHB gene. Alterna-.

WO95~1560 ~1 9~89 1~11. ^15 .

tively, a targeting construct similar to that described for the ~-interferon can be used. In this strategy, the targeting construct is drRir,nPd to include a first target-ing seriuence homologous to sequences upstream of the FSH~
gene, an amplifiable marker gene, a selectable marker gene, a regulatory region, a C~P site, a splice-donor site, an intron, a splice acceptor site, and a second targeting sequence corrrRpnn~l;n~ to se~u~ s ~' LLea", o~ the f irst targeting sequence . The second targeting s~-r~ n~re should not extend further upstream than to posi-tion -40 relative to the normal FSH~ tr~ncrr;rt;nn~l start site in order to avoid undesired ATG start codons. In the homologously r~. ' ;n~nt cells, an mRNA ~Le~uLDL~L is produced which includes the ~n~nnl~R exon, splice-donor site, intron, splice-acceptor site, second targeting ser~uence, and human FSH,B coding exons, intron and 3 ' un-tr~nRl ~ted sf-~qll~nr~R. Splicing of this message will generate a fllnrt;nn~l mRNA which can be tr~nRlAt~l to produce human PSH,B. The size of the intron and thus the position of the regulatory region relative to the coding region of the gene can be varied to optimize the function of the regulatory region.
In any activation strategy, the second targeting ser~uence does not need to lie; ';~t~ly adjacent to or near the first targeting gerluence in the normal gene, such that portions of the gene~ 8 normal upstream region are deleted upon ~ IO~O11R re n~t;nn. Furthermore, one targeting sequence can be upstream of the gene and one may be within an exon or intron of the FSH,B gene.
The: _l;f;Ahle marker gene and sel~rt~hle marker gene can be the same gene, their positions can be reversed, and one or both can be situated in the intron of the targeting construct . r ~ l; f i ~hl e marker genes and selectable marker genes suitable f or selection are de-35 scribed herein. A selectable marker gene is optional and W095/31560 r~l" ~'Ci'i4~
~1 9~q the l;fiAhle marker gene is only reciuired when amplifi-cation is desired. The incorporation of a specific ~A~
site is optional. Optionally, exon se~uences from another gene can be included 3 ' to the splice-acceptor site and 5 ' 5 to the second targeting seciuence in the targeting con-struct. The regulatory region, CAP site, exon, splice-donor site, intron, and splice acceptor site can be isolated as a complete unit from the human _1~^,nrJ~t;nn factor-lcY (EF-1~; Genbank seyuence HUMEFlA) gene or the 10 cytomegalovirus (CMV; Genbank se~ Anre ID3H MV~1) immediate early region, or the ~_ ^ can be Al - ' 1 ed from appropriate ~ ^nts iqo1 Ated from different genes . In any case, the ~Yrjonmlq exon can be the same or different from the first exon of the normal FSH,B gene, and multiple 15 exons can be present in the targeting construct.
Genomic DNA corrraprn~l;n; to the u~,LLealll region of the FSH,~ gene for use as targeting se~ =^nr-^q and assembly of the targeting construct can be per C ~1 using recombi-nant DNA methods known by those skilled in the art. As 20 described herein, a number of s~le~AtAh]e and: 1;f;Ahle markers can be used in the targeting constructs, and the activation can be f-ffe~^t~-l in a large number of cell-ty~es. If degirable, the product of the activated FSH~ gene can be produced in a cell type that expresses 25 the human glycoprotein cY-gubunit, the product of which forms a heterodimer with the product of the FSH,~ gene.
This may be a naturally occurring cell strain or line.
Alternatively, the human glycoprotein ~-subunit gene (Genbank se~uence H~MG~IYCA:L) can be co-e_pressed with the 30 product of the FSH,~ gene, with such co-expression accom-plished by expression of the human glycoprotein ~-subunit gene or cDNA under the control of a suitable promoter, or by activation of the human glycoprotein ~-subunit gene through the methodg described herein. Transfection of 35 primary, secondary, or immortalized human cells and isola-WO95/31560 ~ 2 8 9 ~ . PJ5 tion of ~ lrrjQllqly rr~ ' ;n~nt cellg expressing FSE,~ can be ~r l; q~Pd using the methods r~f~qrr; hPrl above using an E~ISA assay f or human FSH,B (Accurate Chemical and Scien-tific, Westbury, NY). Alternatively, I lo~o~lqly recom-5 binant cells may be i~ n~; fi Pd by PCR screening as de-scribed above. The ;qol~;on of cells rnn~A;n;nJ ampli-fied copies of the; ,lif;~hle marker gene and the acti-vated cY-interferon locus is performed as described above.
Eauiv~l erlts Those skilled in the art will recognize, or be able to ascertain using not more than routine exper; ~;r,n, many equivalents to the specific '-'; q of the inven-tion described herein . Such equivalents are ; n~pnripd to be ~ , R=~d by the ~o11Ow~ng c1a1~i.

Claims (153)

-108-

1. A DNA construct capable of altering the expression of a targeted gene when inserted into chromosonal DNA of a cell comprising:
(a) a targeting sequence;
(b) a regulatory sequence;
(c) an exon; and (d) an unpaired splice-donor site.

2. The DNA construct of Claim 1 wherein the exon com-prises a CAP site.

3. The DNA construct of Claim 2 wherein the exon further comprises the nucleotide sequence ATG.

4. The DNA construct of Claim 3 wherein the exon further comprises encoding DNA which is in-frame with the targeted gene.

5. The DNA construct of Claim 4 wherein the encoding DNA
of the exon is the same as the encoding DNA of the first exon of the targeted gene.

6. The DNA construct of Claim 4 wherein the encoding DNA
of the exon is different from the encoding DNA of the first exon of the targeted gene.

7. The DNA construct of Claim 4 wherein the targeting sequence is homologous to a sequence within the targeted gene.

8. The DNA construct of Claim 4 wherein the targeting sequence is homologous to a sequence upstream of the coding region of the targeted gene.

9. The DNA construct of Claim 4 wherein the targeting sequence is homologous to a sequence upstream of the endogenous regulatory sequence of the targeted gene.

10. The DNA construct of Claim 4 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence within the targeted gene.

11. The DNA construct of Claim 4 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence upstream of the coding region of the targeted gene.

12. The DNA construct of Claim 4 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence upstream of the endogenous regula-tory sequence of the targeted gene.

13. The DNA construct of Claim 4 wherein the targeted gene encodes a therapeutic protein.

14. The DNA construct of Claim 4 wherein the targeted gene encodes a hormone, a cytokine, an antigen, an antibody, an enzyme, a clotting factor, a transport protein, a receptor, a regulatory protein, a struc-tural protein or a transcription factor.

15. The DNA construct of Claim 4 wherein the targeted gene encodes a protein selected from the group con-sisting of erythropoietin, calcitonin, growth hor-mone, insulin, insulinotropin, insulin-like growth factors, parathyroid hormone, .beta.-interferon, .gamma.-inter -feron, nerve growth factors, FSH.beta., TGF-.beta., tumor necrosis factor, glucagon, bone growth factor-2, bone growth factor-7, TSH-.beta., interleukin 1, interleukin 2, interleukin 3, interleukin 6, interleukin 11, inter-leukin 12, CSF-granulocyte, CSF-macrophage, CSF-granulocyte/ macrophage, immunoglobulins, catalytic antibodies, protein kinage C, glucocerebrosidase, superoxide dismutase, tissue plasminogen activator, urokinase, antithrombin III, DNAse, .alpha.-galactosidase, tyrosine hydroxylase, blood clotting factor V, blood clotting factor VII, blood clotting factor VIII, blood clotting factor IX, blood clotting factor X, blood clotting factor XIII, apolipoprotein E or apolipoprotein A-I, globins, low density lipoprotein receptor, IL-2 receptor, IL,-2 antagonists, alpha-1 antitrypsin, immune response modifiers, and soluble CD4.

16. The DNA construct of Claim 15 wherein the targeted gene encodes growth hormone, FSH.beta., G-CSF or GM-CSF.

17. The DNA construct of Claim 15 wherein the targeted gene encodes erythropoietin.

18. The DNA construct of Claim 17 wherein the encoding DNA of the exon is the same as the encoding DNA of the first exon of erythropoietin.

19. The DNA construct of Claim 17 wherein the encoding DNA of the exon is different from the encoding DNA of the first exon of erythropoietin.

20. The DNA construct of Claim 19 wherein the encoding DNA of the exon is the same as the encoding DNA of the first exon of human growth hormone.

21. The DNA construct of Claim 1 wherein the regulatory sequence is a promoter, an enhancer, a scaffold-attachment region or a transcription factor binding site.

22. The DNA construct of Claim 21 wherein the regulatory sequence is a promoter.

23. The DNA construct of Claim 22 further comprising an additional regulatory sequence.

24. The DNA construct of Claim 22 wherein the construct further comprises an enhancer.

25. The DNA construct of Claim 24 further comprising one or more selectable markers.

26. The DNA construct of Claim 25 further comprising an amplifiable marker gene.

27. The DNA construct of Claim 21 wherein the regulatory sequence is a regulatory sequence of the mouse metallothionein-I gene, a regulatory sequence of an SV-40 gene, a regulatory sequence of a cytomegalo-virus gene, a regulatory sequence of a collagen gene, a regulatory sequence of an actin gene, a regulatory sequence of an immunoglobulin gene, a regulatory sequence of the HMG-CoA reductase gene or a regulato-ry sequence of the EF-1.alpha. gene.

28. A method of making a homologously recombinant cell wherein the expression of a targeted gene is altered, comprising the steps of:
(a) transfecting a cell with a DNA construct, the DNA construct comprising:

(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon; and (iv) an unpaired splice-donor site, thereby producing a transfected cell; and (b) maintaining the transfected cell under condi-tions appropriate for homologous recombination.

29. The method of Claim 28 wherein the exon comprises a CAP site.

30. The method of Claim 29 wherein the exon comprises the nucleotide sequence ATG.

31. The method of Claim 30 wherein the exon further comprises encoding DNA in-frame with the targeted gene.

32. The method of Claim 31 wherein the encoding DNA of the exon is the same as the encoding DNA of the first exon of erythropoietin.

33. The method of Claim 31 wherein the encoding DNA of the exon is different from the encoding DNA of the first exon of erythropoietin.

34. The method of Claim 31 wherein the targeting sequence is homologous to a sequence within the targeted gene.

35. The method of Claim 31 wherein the targeting sequence is homologous to a sequence upstream of the coding region of the targeted gene.

36. The method of Claim 31 wherein the targeting sequence is homologous to a sequence upstream of the endoge-nous regulatory sequence of the targeted gene.

37. The method of Claim 31 wherein the construct further comprises a second targeting sequence homologous to a sequence within the targeted gene.

38. The method of Claim 31 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the coding region of the target-ed gene.

39. The method of Claim 31 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the endogenous regulatory se-quence of the targeted gene.

40. The method of Claim 31 wherein the cell is a human cell.

41. The method of Claim 23 wherein the targeted gene encodes erythropoietin.

42. The method of Claim 31 wherein the encoding DNA is the same as the encoding DNA of the first exon of erythropoietin.

43. The method of Claim 31 wherein the encoding DNA is different from the encoding DNA of the first exon of erythropoietin.

44. The method of Claim 31 wherein the encoding DNA of the exon is the same as the encoding DNA of the first exon of human growth hormone.

45. The method of Claim 28 further comprising the step of:
(c) maintaining a homologously recombinant cell from step (b) under conditions appropriate for pro-duction of a protein.

46. The method of Claim 45 in which the gene whose ex-pression is altered is the erythropoietin gene.

47. Erythropoietin produced by the method of Claim 45.

48. A fusion protein containing amino acids encoded by exons from the DNA construct and amino acids encoded by an endogenous gene produced by the method of Claim 45.

49. A fusion protein of Claim 48 wherein the endogenous gene is erythropoietin.

50. A fusion protein of Claim 49 comprising amino acids 1-3 of human growth hormone and amino acids 6-165 of human erythropoietin.

51. A homologously recombinant cell produced by the method of Claim 28.

52. A homologously recombinant produced by the method of Claim 29.

53. A homologously recombinant cell produced by the method of Claim 30.

54. A homologously recombinant cell produced by the method of Claim 31.

55. A homologously recombinant cell produced by the method of Claim 32.

56. A homologously recombinant cell produced by the method of Claim 33.

57. A homologously recombinant cell produced by the method of Claim 40.

58. A homologously recombinant cell produced by the method of Claim 41.

59. A homologously recombinant cell produced by the method of Claim 42.

60. A homologously recombinant cell produced by the method of Claim 44.

61. A homologously recombinant cell comprising an exoge-nous regulatory sequence, an exogenous exon and a splice-donor site, operatively linked to the second exon of an endogenous gene.

62. The homologously recombinant cell of Claim 61 wherein the exogenous exon comprises a CAP site.

63. The homologously recombinant cell of Claim 62 wherein the exogenous exon further comprises the nucleotide sequence ATG.

64. The homologously recombinant cell of Claim 63 wherein the exogenous exon further comprises encoding DNA in-frame with the targeted endogenous gene.

65. The homologously recombinant cell of Claim 64 wherein the encoding DNA is the same as the encoding DNA of the first exon of the targeted gene.

66. The homologously recombinant cell of Claim 64 wherein the encoding DNA is different from the encoding DNA
of the first exon of the targeted gene.

67. The homologously recombinant cell of Claim 64 wherein the exogenous regulatory sequence, exogenous exon and splice-donor site are upstream of the coding region of the targeted gene.

68. The homologously recombinant cell of Claim 67 wherein the exogenous regulatory sequence, exogenous exon and splice-donor site are upstream of the endogenous regulatory sequence of the targeted gene.

69. The homologously recombinant cell of Claim 61 wherein the endogenous regulatory sequence is deleted.

70. The homologously recombinant cell of Claim 69 wherein the first endogenous exon is deleted.

71. The homologously recombinant cell of Claim 64 wherein the targeted gene encodes a hormone, a cytokine, an antigen, an antibody, an enzyme, a clotting factor, a transport protein, a receptor, a regulatory protein, a structural protein or a transcription factor.

72. The homologously recombinant cell of Claim 64 wherein the targeted gene encodes a protein selected from the group consisting of erythropoietin, calcitonin, growth hormone, insulin, insulinotropin, insulin-like growth factors, parathyroid hormone, .beta.-inter-feron, .gamma.-interferon, nerve growth factors, FSH.beta., TGF-.beta., tumor necrosis factor, glucagon, bone growth factor-2, bone growth factor-7, TSH-.beta., interleukin 1, interleukin 2, interleukin 3, interleukin 6, interleukin 11, interleukin 12, CSF-granulocyte, CSF-macrophage, CSF-granulocyte/macrophage, immunoglobulins, catalytic antibodies, protein kinase C, glucocerebrosidase, superoxide dismutase, tissue plasminogen activator, urokinage, antithrombin III, DNAEe, .alpha.-galactosidase, tyrosine hydroxylase, blood clotting factor V, blood clotting factor VII, blood clotting factor VIII, blood clotting factor IX, blood clotting factor X, blood clotting factor XIII, apoli-poprotein E or apoliporotein A-I, globins, low density lipoprotein receptor, IL-2 receptor, IL-2 antogonists, alpha-1 antitrypsin, immune response modifiers, and soluble CD4.

73. The homologously recombinant cell of Claim 61 wherein the cell is a eukaryote.

74. The homologously recombinant cell of Claim 73 wherein the cell is of fungal, plant or animal origin.

75. The homologously recombinant cell of Claim 74 wherein the cell is of vertebrate origin.

76. The homologously recombinant cell of Claim 75 wherein the cell is a primary or secondary mammalian cell.

77. The homologously recombinant cell of Claim 75 wherein the cell is a primary or secondary human cell.

78. The homologously recombinant cell of Claim 75 wherein the cell is an immortalized mammalian cell.

79. The homologously recombinant cell of Claim 75 wherein the cell is an immortalized human cell.

80. The homologously recombinant cell of Claim 75 wherein the cell is selected from the group consisting of:
HT1080 cells, HeLa cells and derivatives of HeLa cells, MCF-7 breast cancer cells, K-562 leukemia cells, KB carcinoma cells, 2780AD ovarian carcinoma cells, Raji cells, Jurkat cells, Namalwa cells, HL-60 cells, Daudi cells, RPMI 8226 cells, U-937 cells, Bowes Melanoma cells, WI-38VA13 subline 2R4 cells, and MOLT-4 cells.

81. The homologously recombinant cell of Claim 80 wherein the targeted gene encodes erythropoietin.

82. The homologously recombinant cell of Claim 81 capable of expressing erythropoietin.

83. The homologously recombinant cell of Claim 82 wherein the encoding DNA is the same as the encoding DNA of the first exon of erythropoietin.

84. The homologously recombinant cell of Claim 81 wherein the encoding DNA is different from the encoding DNA
of the first exon of erythropoietin.

85. The homologously recombinant cell of Claim 84 wherein the encoding DNA is the same as the encoding DNA of the first exon of human growth hormone.

86. The homologously recombinant cell of Claim 61 capable of expressing a fusion protein comprising amino acids encoded by the exogenous exon and amino acids encoded by the endogenous gene.

87. A fusion protein of Claim 86 wherein the endogenous gene is erythropoietin.

88. A fusion protein of Claim 87 comprising amino acids 1-3 of human growth hormone and amino acids 6-165 of human erythropoietin.

89. The homologously recombinant cell of Claim 66 wherein the regulatory sequence is a promoter, an enhancer, a scaffold-attachment region or a transcription factor binding site.

90. The homologously recombinant cell of Claim 89 wherein the exogenous regulatory sequence is a promoter.

91. The homologously recombinant cell of Claim 89 wherein the exogenous regulatory sequence is a regulatory sequence of the mouse metallothionein-I gene, a regulatory sequence of an SV-40 gene, a regulatory sequence of a cytomegalovirus gene, a regulatory sequence of a collagen gene, a regulatory sequence of an actin gene, a regulatory sequence of an immuno-globulin gene, a regulatory sequence of the HMG-CoA
reductase gene or a regulatory sequence of the EF-1.alpha.
gene.

92. A method of altering the expression of a gene in a cell , comprising the steps of:
(a) transfecting a cell with a DNA construct, the DNA construct comprising:
(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon; and (iv) an unpaired splice-donor site, thereby producing a transfected cell;

(b) maintaining the transfected cell under condi-tions appropriate for homologous recombination, thereby producing a homologously recombinant cell; and (c) maintaining the homologously recombinant cell under conditions appropriate for expression of the gene.

93. The method of Claim 92 wherein the exon comprises the nucleotide sequence ATG.

94. The method of Claim 92 wherein the exon further comprises a CAP site.

95. The method of Claim 94 wherein the exon further comprises encoding DNA which is in-frame with the targeted gene.

96. The method of Claim 95 wherein the encoding DNA is the same as the encoding DNA of the first exon of the targeted gene.

97. The method of Claim 96 wherein the targeted gene is the erythropoietin gene.

98. The method of Claim 96 wherein the encoding DNA is different from the encoding DNA of the first exon of the targeted gene.

99. The method of Claim 93 wherein the targeted gene is the erythropoietin gene.

100. The method of Claim 98 wherein the targeting sequence is homologous to a sequence within the targeted gene.

101. The method of Claim 98 wherein the targeting sequence is homologous to a sequence upstream of the coding region of the targeted gene.

102. The method of Claim 98 wherein the targeting sequence is homologous to a sequence upstream of the endoge-nous regulatory sequence for the targeted gene.

103. The method of Claim 98 wherein the construct further comprises a second targeting sequence homologous to a sequence within the targeted gene.

104. The method of Claim 98 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the coding region of the target-ed gene.

105. The method of Claim 98 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the endogenous regulatory se-quence for the targeted gene.

106. The method of Claim 92 further comprising the step of:
(c) maintaining a homologously recombinant cell under conditions appropriate for production of a protein.

107. The method of Claim 106 in which the gene whose ex-pression is altered is the erythropoietin gene.

108. Erythropoietin produced by the method of Claim 107.

109. A fusion protein produced by the method of Claim 106.

110. A fusion protein of Claim 109 wherein the endogenous gene is erythropoietin.

111. A fusion protein of Claim 110 comprising amino acids 1-3 of human growth hormone and amino acids 6-165 of human erythropoietin.

112. A method of making a protein by altering the expres-sion of a gene in a cell, comprising the steps of:
(a) transfecting a cell with a DNA construct, the DNA construct comprising:
(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon; and (iv) an unpaired splice-donor site, thereby producing a transfected cell;
(b) maintaining the transfected cell under condi-tions appropriate for homologous recombination, thereby producing a homologously recombinant cell; and (C) maintaining the homologously recombinant cell under conditions appropriate for production of the protein.

113. The method of Claim 112 wherein the exon comprises a CAP site.

114. The method of Claim 113 wherein the exon comprises the nucleotide sequence ATG.

115. The method of Claim 114 wherein the exon further comprises encoding DNA which is in-frame with the targeted endogenous gene.

116. The method of Claim 115 wherein the encoding DNA is the same as the encoding DNA of the first exon of the targeted gene.

117. The method of Claim 116 wherein the encoding DNA is different from the encoding DNA of the first exon of the targeted gene.

118. The method of Claim 117 wherein the targeting se-quence is homologous to a sequence within the target-ed gene.

119. The method of Claim 117 wherein the targeting se-quence is homologous to a sequence upstream of the coding region of the targeted gene.

120. The method of Claim 117 wherein the targeting se-quence is homologous to a sequence upstream of the endogenous regulatory sequence for the targeted gene.

121. The method of Claim 117 wherein the construct further comprises a second targeting sequence homologous to a sequence within the targeted gene.

122. The method of Claim 117 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the coding region of the target-ed gene.

123. The method of Claim 117 wherein the construct further comprises a second targeting sequence homologous to a sequence upstream of the endogenous regulatory se-quence for the targeted gene.

124. An erythropoietin produced by the method of Claim 112

125. The erythropoietin of Claim 124 wherein the cell is of human origin.

126. A protein produced by the method of Claim 112.

127. The protein of Claim 126 which is a fusion protein.

128. The fusion protein of Claim 127 wherein the endoge-nous gene is the erythroroietin gene.

129. The fusion protein of Claim 128 comprising amino acids 1-3 of human growth hormone and amino acids 6-165 of human erythropoietin.

130. The DNA plasmid pREPO18.

131. A DNA construct capable of altering the expression of a targeted gene when inserted into the chromosomal DNA of a cell, comprising:
(a) a targeting sequence;
(b) a regulatory sequence;
(c) an exon;
(d) a splice-donor site;
(e) an intron; and (f) a splice-acceptor site.

132. The DNA construct of Claim 131 wherein the targeting sequence is homologous to a sequence within the targeted gene.

133. The DNA construct of Claim 131 wherein the targeting sequence is homologous to a sequence upstream of the coding region of the targeted gene.

134 The DNA construct of Claim 131 wherein the targeting sequence is homologous to a sequence upstream of the endogenous regulatory sequence of the targeted gene.

135. The DNA construct of Claim 131 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence within the targeted gene.

136. The DNA construct of Claim 131 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence upstream of the coding region of the targeted gene.

137. The DNA construct of Claim 131 wherein the construct further comprises a second targeting sequence homolo-gous to a sequence upstream of the endogenous regula-tory sequence of the targeted gene.

138. A homologously recombinant cell comprising a regula-tory sequence, an exon, a splice-donor site, an intron and a splice-acceptor site introduced by homologous recombination upstream of the coding region of a targeted gene.

139. The homologously recombinant cell of Claim 138 where-in the targeted gene is the .alpha.-interferon gene.

140. The homologously recombinant cell of Claim 138 where-in the targeted gene is the erythropoietin gene.

141. A homologously recombinant cell comprising the dhfr gene, the neo gene, the CMV promoter, hGH exon 1 and an unpaired splice-donor site targeted to a position upstream of the endogenous erythropoietin regulatory region.

142. The homologously recombinant cell of Claim 141 pro-duced by the integration of DNA from pREPO18.

143. A method of making a homologously recombinant cell wherein the expression of a targeted gene is altered, comprising the steps of:
(a) transfecting a cell with a DNA construct, the construct comprising:
(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon;
(iv) a splice-donor site;
(v) an intron; and (vi) a splice-acceptor site;
wherein the targeting sequence directs the inte-gration of elements (b) - (f) upstream such that they are operatively linked to the first exon of a targeted gene; and (b) maintaining the transfected cell under condi-tions appropriate for homologous recombination.

144. A homologously recombinant cell produced by the method of Claim 143.

145. A method of altering the expression of a gene in a cell, comprising the steps of:
(a) transfecting a cell with a DNA construct, the construct comprising:
(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon;
(iv) a splice-donor site;
(v) an intron; and (vi) a splice-acceptor site;

wherein the targeting sequence directs the inte-gration of elements (b)-(f) upstream such that they are operatively linked to the first exon of a targeted gene;
(b) maintaining the transfected cell under condi-tions appropriate for homologous recombination;
and (c) maintaining the homologously recombinant cell under conditions appropriate for expression of the gene.

146. A method of making a protein by altering the expres-sion of a gene in a cell, comprising the steps of:
(a) transfecting a cell with a DNA construct, the construct comprising:
(i) a targeting sequence;
(ii) a regulatory sequence;
(iii) an exon;
(iv) a splice-donor site;
(v) an intron; and (vi) a splice-acceptor site;
wherein the targeting sequence directs the inte-gration of elements (b)-(f) upstream such that they are operatively linked to the first exon of a targeted gene;
(b) maintaining the transfected cell under condi-tions appropriate for homologous recombination;
and (c) maintaining the homologously recombinant cell under conditions appropriate for expression of the protein.

147. The method of Claim 146 wherein the targeted gene is the .alpha.-interferon gene or the erythropoietin gene.

148. DNA sequences located between about 5 kilobases and 30 kilobases upstream of the ATG of the erythropoie-tin gene.

149. A method for targeting the erythropoietin gene in a mammalian cell compriging transfecting the cell with a construct comprising a DNA sequence homologous to a sequence upstream of the sequence ATG of the erythro-poietin gene.

150. The method of Claim 149 wherein the construct com-prising a DNA seguence homologous to a sequence located between about 5 kilobases and 30 kilobases upstream of the sequence ATG of the erythropoietin gene.

151. The method of Claim 150 wherein the mammalian cell is a human cell.

152. A method for targeting the erythropoietin gene in a mammalian cell comprising transfecting the cell with a construct comprising a DNA sequence homologous to a sequence within the erythropoietin gene.

153. The method of Claim 152 wherein the mammalian cell is a human cell.