CA1326639C - Nucleic acid encoding tgf-.beta. and its uses - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE

Nucleic acid encoding TGF-.beta. has been isolated and cloned into vectors which are replicated in bacteria and expressed in eukaryotic cells. TGF-.beta. is recovered from transformed cultures for use in known therapeutic modalities. Nucleic acid encoding TGF-.beta.
is useful in diagnosis and identification of TGF-.beta. clones.

Description

32~3~
NUCLEIC ACID ENCODING TGF-~ AND ITS USES

Peptides which can induce the reverslble phenotypic transformation of mammalian cells in c-~lture hAve been gi~en the name transfonming growth factor (TGF)l~2. Type ~ TGF competes with epider~al growth factor (EGF) for binding to the same cell surface receptor3. A 50 amino acid TGF-~ species has been purlfied and shown to share sequence homology with EG ~. TGF-~ Ls synthesized by various transf3rmed c~ll lines3~5~6~91. The 50 amino acid TGF-~ is iniLtlally synth0si~ed as part of a 160 amino acid precursor molecule which undergoes N- and C-terminal proteolytic processing , to yield the mature peptide7~8. The detection of TGF-~ species with apparently higher molecular weightsl~2~9 might be due to ~ariable processLng of the 160 amino acid precursor92.
Type ~ TGF activity has be0n isolated from tumor cells as well as many normal tissues~ l, includin~ kldneyl2, placental3 and ! blood plateletsl4~15. TGF-~ is present in platelets, which also contain plat~let-derlved growth factor (PDGF) and an EGF-like peptidel6. Bovine TGF-~ has been demonstrated to accelerate wound healing in ratsl7 and to induce fetal RMM cells to undergo differentia~ion and synthesize cartilage-specific macromolecules80.
Treatment of NRK fibroblasts wil:h TGF-~ does, however, ~esult in an . increase ln the number of membrane receptors for EGF20. This ~; 25 observation is in agrsement wiLth the known abiliLty of TGF-~ to ~ greatly potentiate the activity of ~GF and TGF-~ on these :~ cellsl~ll. Moreover, TGF-~ alone can induce ARR- ~B fibroblasts to ~ form colonies in soft agar21. Eleva~ed l~vels of TGF-~ are i secreted by some transformed cells22. In addition to its ability to stimulate cell proli~eration, TGF-~ has been demonstrated ~o -l~ inhibit the anchorage-dependent growth of a variety of human cancer ~:~; cell~ linesl3. It is now thought tha~ TGF-~ may be identical or very similar to a growth inhibitor isolated from African grePn monkey (BSC-l~ cells24. Whether TGF-~ acts to stimulate or inhibit . 35 the growth of a particular cell line type appears to depend on many -2- ~32~
variables, including the physiological condition of the cell and the presence of additional growth factors.

Bo~ine TGF-~ has been purified to sequenceable grade. The first 15 amino-terminal resldues of the mature protein were found to bs Ala-Leu-Asp-Thr-A~n-Tyr-CMC-Phe-Ser-Ser-Thr-Gly-Lys-Asn-CMC-, whérein CMC is S-carboxymethyl cysteine representing cysteine or half-cystine residues.

Hu~an T~F-~ from human plaoenta and platelets has been p~rified to the same degree. Placental TGF-~ was reported to have the following amino terminal sequence: Ala-Leu-Asp-Thr-Asn-Tyr-CMC-Phe-(Ser-Ser)-Thr-Glu-Lys-Asn-CMC-Val~X-Gln-Leu-Tyr-Ile-Asp-Phe-X-(Lys~-Asp-Leu-Gly-, wherein X was undetermined and CMC is as defined above. Platelet TGF-~ was reported as the amino terminal sequence Ala-Leu-Asp-Thr-Asn-Tyr-X-Phe-Ser-, wherein CMC and X are as de~lned a~ove.

Human TGF-~ was reported to be composed of two polypeptide chains of very similar molecular ~eight (Mr 12 7 500) which are maintained in covalent association by disulfide bonds. The dlsulfide bonds wer2 considered likely to play an important role in c~nfPrring struc~ure on the TGF-,~ molecule.

t Several other factors have been described that are related to TGF-~ by limited amino acid sequ~nce homology. The inhibin A and B
1~ beta chains are related to TGF-~ by the placement of homologous! cysteins resldues ~nd othPr limited amino acid sequence homology, from which it has been inferred ~hat inhibin, or more accurately ~ 30 actlvin (dimers of ~he inhibin be~aA or ~eta~ chains), is j~ structurally rcla~ed to TGF-~. Inhibin represses the release of'~ ~ FSH from the pltultary, while activin enhances ~he release of FSH~l~B2. TGF-~ is not known to have this activity.

,' : ~

3 ~32~9 :
Mullerian inhibitory sub~tance has a C-terminal region which is homologous with T~F-~ and inhlbits the growth of Mullerian-derived tumors83,84 TGF-~ prepared by purification from biolo~ical materials presents a risk of contamination by infectious Pgents such ~s HTLV-III or hepatitis viruses. Accordingly, it ls an ob~ect of this invention to prepare TGF-~ from sources that do not present a risk of contamination.
1~
It is another ob~ect to prepare nucleic acid that will hybridize with DNA encoding biolsgically active TGF-~. When appropriately labelled, this nuclelc scid is useful in diagnostic assays for TGF-~ mRNA and in isolating DNA encoding TGF-~.
It i5 a ~urther ob~ect herein to prepare vectors containing j DNA tha~ encodes TGF-~, together with hos~ cell transformants ~hat express biologically active TGF-~.

In accordance with this invention, the foregoing objects are achieved by a method comprisirlg (a) constructing a vector which includes nucleic acid encoding TGF-~, (b) transforming hetexologous host eukaryotic c6~11 with the vector, (c) culturing the transformed cell, and (d) r~coverin~ TGF-~ from the culture.
Nucleic ~cid encoding two s~btypes of TGF-~ (TGF-~l and TGF-3~ is provided which~s u5e~ul in constructing the above vectors.
This nuclei~ acid or a nucleic acid cap~ble of hybr~dizing therewith also ~s~labelled and used in diagnostic assays ~or DNA or m~NA encoding TGF-~ or related proteins.

The preparation of TGF-~ derivatives by recombinant methods i5 made possible by knowledge of~the TGF-~ coding sequences disclosed herein. These~derivatives include silent and expressed mutants in ~ the nucleic acid encoding TGF~

-4- ~3~63~

Silent variants invol~e the substitution of one degenerate codon for another where both codons code ior the same amlno acid, bu~ ~hich substitutlon could e2ert a salutary effect on TGF-~ yield in recombinant culture, e.g. by modifying the secondary structure of TGF-fl mRNA, and which salutary substitution is identified by screening TGF-~ yield~ from transformants.

~, Expressed TGF-~ variants fall into one or more of three classes: deletions, substitutions or insertions. Deletions are characterized by the elimina~ion of amino acid residues without the insertion of a replacement residue. Deletional variants of TGF-~
l are useful in making TGF-~ fragments, for example, where it is desired to delete an immune epitope.
Substitution variants are those in which one amino acid ¦ residue has been replaced by another. Such variants are extremely difficult to make by methods other than recombinant synthesis, especially substitution~ targeted for the int0rior of the primary amino acid sequence. They are useful in modifylng the biological activity of TGF-~. Substitution variants include allelic forms of TGF-~ as well as TGF-~ subtypes.

TGF-~ is found as a disulfide linked dimer. Further variants include heterodimers of TGF-~ subtypes and heter~dimers of one TGF-chain hav~ng a native amino ~cid sequence disulfide bonded ~hrough the ordinary homodimer dlsulfide linka~s to a ~ predetermined variant of TGF-~. In this case, such variants .~
~l ~ incIude blologically actlve as well as biologically inactive TGF-~.
Insertional ~ariants are those in which one or more resldues are placed within the internal TGF-~ sequence Dr at either end thereof. Variants of this class include fusion proteins resulting from insertions at the carboxyl or amino terminal residues of TGF-~. TGF-~ fusions with bacterial or other immunogenic proteins are .~ :
1 ~ ~
ii:: :

', " ' ' ~ " -', ~ " ;, ,;

5 ~ 32~
use~ul for raising antlbodies against TGF-~ or its predetermined fragments.

Fig. la is a schematic diagram of the TGF-~l m~NA showing the boxed coding sequence. The 112 amino acid TGF-~l (dashed) is encoded by the 3' end of the coding sequence. The sequenced cDNA
inserts of ~Cl, 3.19, 3.32, 4.10, 4.33, 4.37 and 5.7b (de~cribed infra) and the genomic DNA sequence for the 3' untranslated re~ion are aligned above the diPgram.
Figs. lb(I) - lb(III~ (hereinafter referred to cDllectively as Fig. lb) dep~ct the ~equence and ded~ced amino acid sequence of the preTGF-~l cDNA, determined from the several overlappin~ cDNAs and the genomic 3' region. The 5' terminal region which could be folded into stable ha~rpin loops is underlined. The preTGF-~l cDNA
encodes a 390 amino acid protein, of which the C-terminal 112 amino acids (boxed) encode mature TGF~l. A hydrophobic domain found at the N-terminus of the precursor is overlined. An overlined Arg-Arg dipeptide precedes the protesl~ytic cleavage site for release of ~ 20 TGF-~l. Three potential N-~lycosylation sites in preTGF-~l are¦ overlined. The stop codon i8 followed by the underlined ~-C rich I sequence and a downstream TATA-Iike sequence.

;I Fig. 2 depicts a genomic fragment encoding a TGF-~l exon and its deduced amino acid sequence. Arrows show the mRNA processing sltes (lntro~-exon ~unctions). The res$due numbers correspond to ~7 Fig. lb.

Fig. 3 ls a comparison between the known N-terminal sequence of bovine TGF-~2 and the mature h~an TGF-~l and human and porcine TGF-~3 ~mino acid sequences. Unconserved substitutions in TGF-~
;!~ which ~onstltute the TGF-~2 and ~3 subtypes are designated by dots above the unconserved residues. The a~ino acid residue numbers shown in this Figure shall be used herein unless otherwise indicated.

. , ~

,:i -6- ~3~ 39 Figs. 4a - 4c show the cDNA sequences for human and porcine TGF-~3. The sequence for the hu~,an cDNA encodes a portion of the presequence region and all of the mature sequence. The human and porc$ne sequ0nces are ad;acent to the lines designated hu4 and 10~11.3, respe~tively. Gaps are introduced into the sequences in order to max~mize nucleotide homology. HomolDKous bases are designatfsd with an asterisk.

Fig. 5 depicts the amino acid sequences encoded by the cDNAs of Figs. 4a-4c. Homologous residues are designated wlth an ~ asterisk. Candidate translational start methionyl residues for the j porclne sequence are located at positions 43, B8 or 90 (boxed methionyl residues). The C-termlnal residue ior both the hu~,an and porcine sequences iB the seryl at 499 (porcine~ or 204 (human).

¦ TGF-~ has proven to be extremely difficult to synthesize in Il recombinant cell culture while retaining its growth-alter~ng 1 activity. As can be seen from Figs. lb, 3 and 5, the mature TGF- p aminG acid sequence contains ~ large number of cysteine residues9, ¦ at least some of which apparently are involved in interchain 3 crosslinking in forminjg the homodimeric TGF-~ which is recovered1 from natural sources. Furth~Prmore, TGF-~ is expressed as a i precursor molecule havlng a large amino terminal region not :, 25 containing the recognizable NH2-terminal signal peptids sequence typical cf most secreted proteins, even though TGF-j,B normally appears to some degree in the extracellular mediu~. Howiever, eukaryo~ic cells have been transformed to expr~ss heterologous TGF-~, notwithstanding the anticipated difficulty in properly processing the primary translation product in recombinant culture.

This in~ention is directed to recombinant synthesis of TGF-~, ~11 which is define~ a8 inclusive of blologically active preTGF-~
ha~ing thfs Fig. lb sequence, mature TGF-~, polypeptide fragments `1 35 thereof and lnsertion, substitution and~or deletion variants of .,, :

',~1 :

~7~ 1~2~3~
such preTGF-~ (including alleles or other TGF-~ subtypes than the TGF-~l subtype shown in Fig. lb~, mature TGF-~ or polypeptide fragments.

Biologically active TGF-~ is defined as being capable of inducing EGF-potentiated anchorage independent growth of target cell lines81 and/or growth lnhibition of neoplastic cell lines23.
Anchorage independent gro~th refers to the a~ility of TGF-~ and EGF
treated non-neoplastic target cells to form colonies in soft agar, a characterlstic Pscribed to transformation of the cells (hence the name trQnsfo~lng growth factor). This is not to qay that TGF-~
will "cause" cancer because it is now known that many normal cells express TGF-~. Paradoxically, TGF-~ also i9 known to lnhibit the growth of several normal cell types and various neoplastic cells such as A549.

~iologic~l activity ~or the p~rposes herein also generally includes the abili~y to cross-re~ct with antisera raised against native TGF-~. Native TGF-~ is that which is obtained from platelets or other natural sourc,es. Immunological cross-reactivity is ~ measure of a sin~le active epitope and not necessarily of active TGF-~ domains involved ln inducing anchorage-independent growth oP target cells. However, immunologically cross-reactive proteins ~E se are not biologically active as defined herein unless they &lso exert growth-afecting activity, i.e., all biologically active TGF-~ ~ariants promote growth in the defined assays, but not all immunologicPlly ¢ross-reactiv~ TGF-~s are biologically active. Of course, TGF-~ which is capable of induclng anchorage independent gro~th frequently will exhibit immunological j 30 cross-reactivity with antlsera raised against the native molecule as a corollary ~o maintenance of proper conformatlon.

The figure lb nucleotide sequence was obtained by an analysis of several overlapping cDNAs and gene fragments, leading to the determInation oP a continuous sequence corresponding to the TGF-~l : , .~ .

-8- ~32~39 precursor mRNA. According to Fig. lb an initiatDr ATG i5 loca~ed 841 nucleotides from the 5' end and establishes a coding sequence for a 390 residue polypeptide. Several areas ~ithin the cDNA
sequence have an exceptionally high G-C content. The initiator ATG
is flanked by two G-C rich areas of approximately 200 bp each. In addition, several regions of the cDNA, particularly the 5'-terminus, have regions with greater than 80 percent G-C content.
The location of these G-C rich regions coincides with the areas in which the many cD~A cloning artifacts occurred and where partial length cDNAs were obt~ined.

The S' untranslated region of the TGF-~l mRNA is 841 nucleotides long (assuming the ATG is located at nucleotide 842) and contains a long sequence consisting almost exclusively of 1 15 purines. The biological relevance of this exceptionally long 5' il untranslated region oP hlgh G-C content is unknown, but it is similar to the structural organizatlon of c-myc mRNA. However, there is no strikin~ sequence h~mology between these two sequences.
The long 5' untranslated region of c-myc has been hypothesized to have a functional signiflcance37. me G-C rich 5'-proximal part of the 5' untranslated sequence of human c-myc mRNA has several regions whlch could form stable hairpin loopq. Likewise, ~he first I 120 bp of the untranslated preTGF-~l c~NA can theoretically be ¦ folded lnto ha$rpin loop ~truct~res with a calculated stability of -91 kcal. The long 5' untranslated equence a~d the potentially stable hairpin loop structures could play a role in the mRN~
stability or in the regulation of transcriptiDn. Accordingly, this region o~n be deleted aDd substituted for by other 5' untranslated sequences, e . g. from viral proteins, in order to ldentify structures that may improve TGF-~l yields ~rom recombinant cell culture.

The stop codon preceding ba6e 2015 is im~ediately followed by a re~arkable, G-C rich ~equence of 75 nucleotides (underlined in Fig. lb). Th~s sequence consists of multiple repeats of CCGCC.

1 ~

9 ~32~3g The peculiar nature of this sequence is probably responsible for the fact that the 3' untranslated end of the ~RNA could not be cloned as a cDNA sequence, perhaps due to the inability of the E.
coli DNA polymerase I to use this sequence as a template for the ~econd strand cDNA synthesis. Repeat ~equences of a similar nature have been found in the promoter regions of the genes for human dihydrofola~e reductase38, hu~ian transferrin receptor, human adenosine deaminase39, and Herpes~irus thymidine kinase40. In the latter case, McKnight et al.40 have shown that these structural elements are of mA~ior importance for the transcription efficiency.
In addition, it has been shown that the promoter speclfic transcriptional factor Spl binds to such sequences in the SV40 early promoter region and in a related monkey promoter41~42. In all of these cases the G-C rich repeats are followed closely by the Goldberg-Hogness TATA sequence. In the case of preTGF-~l, howe~er, these sequences are located in the 3' untranslated region of the gene, but are interestingly also followed by a TATA-like sequence.
No e~idence that this region could function as a promoter is available. Ihe preTGF-~l gene sequence has the hexanucleotide ¦ 20 AATAAA about 500 nucleotides downstream from the stop codon. This sequence, ~hich usually precedes the site of polyadenylat~on by 11 to 30 bases32, probably ftmctions as the preTGF-~l mRNA
`l polyadenylation signal, since t:his would be in agreement with the size of preTGF-~ mRNA e~timsted from Northern hybridizations, and since 3' untranslated regions rarely contain intervening sequences.
Benoist et al.43 ha~e propo~ed a consensus sequence TTCACTGC which follows the A~TAAA closely and i~mediately precedes the polyA-tail.
I A similar ~equence, TTCA~GCC, follows the AATAAA equence in the 3' $~ ~ untrsnslated re~ion of the preTGF-~l mRNA, pro~iding further ; 30 support for~ the assign~ent of the polyadenylation site at posltion 2530 ~Fig. lb).

PreTGF-~l is a polypeptide of 390 amino aclds. Comparison of this sequence with ~he previously determlned NH2-terminus of mature TG~ shows that ~GF-~l constitutes the C-terminal 112 amiino acids .i ~ :

~ 32~3~

of preTGF-~l. The mature TGF-~l monomer is cleaved from the precursor at the Arg-Arg dlpeptide immediately preceding the mature TGF-~l ~H2-terminus. A similar dibasic cleavage site is located immediately upstream from the mature TGF-~3 amino terminus. Such proteolytic cleavage sites have been found in several other polypeptide precursor sequences, including preproenkephalin44~45, the calcitonin precursor46, and corticotropin-~-lipotropin precursor47. Determination of the hydrophobicity profile by the method of Kyte and Doolittle48 predicts that the Arg-Arg sequence is located within a hydrophilic region which would make it accessible to a trypsin-like peptidase. Post-translational cleavage of the precursor gives rise to the mature TGF-~ monomer.
The disposition of the presequence is not known but may give rise to other biologically active peptides. The TGF-~l and TGF-~3 precursors contain several palrs of basic residues ~Figs. lb and 5) , which could also undergo post-translation cleavage and give rise to I separa~e polypeptide entities. Mature TGF-~l contains two Arg-~ys s dipeptides which appaxsntly are not clea~ed. As shown ln Fig. lb, ¦ the preTGF-~l precursor contains three potential N-glycosylation sites, Aqn-X-Ser or Thr (Fig. lb). None of these are localized within mature TGF-~l. Accordingly, a method is provided whereby mature T~F-~ is purified frce of glycoproteins by adsorbing the glycoproteins on immobllized lectins and eluting TGF-~ with the unadsorbed fraction.
3 2g The ~equence for human TGF-~ ~as determined by direct ~mino j acid sequence analysis and by deductlon from the TGF-~ cD~A. The J~ sequence of the different TGF-~l peptides obtained by clostripain digestion is in agreement with the cDNA sequence, except for a few residue.s which presumably are due to incorrect amino acid assign~ent in sequencing. The 112 amino acid TGF-~ sequence co~tains ~9 cysteines~, whereas the rest of the precursor contains only two (Figs. lb and~ 5). Previous studies have shown that reduction of the TGF-~l dimer o$ 25 Xd results in the ~eneration of two polypeptide chains of 12.5 kdl5. Sequence analysis of the TGF-2~3~
amino-terminus and of the TGF-~l peptides obtained after clostripain digestion strongly su~gests that the TGF-~ dimer consists of two identical polypeptides. This homodimeric nature is also supported by the presence of only a single hybridizing DNA
fragment upon Southern hybridization of human genomic DNA with a TGF-~ exon probe. Chou-Fasman analysis50 Of the secondary structure shows ~hat the TGF-~ polypeptide has an extensive ~-sheet character with little, if any, ~-helicity. The region immediately preceding the basic dipeptide cleavage site is likely in an ~-helical configuration.

For purposes hereln, preTGF-~ is defined a~ the normal TGF-~
precursor depicted in Figs. lb and 5 as well as other precursor forms of TGF-~ in which the presequence is not that normally associated with TGF-~. These latter forms are to be considered insertlonal mutants of DNA encoding mature TGF-~. ~hese mutants ordinarily comprise a presequence which is heterologous ~o TGF-~ in I the orm of a fusion with mature TGF-~. The heterologous ¦ presequences preferably are obtained from other secreted proteins, ~ 20 for example pregrowth hormone, preproinsulin, viral envelope ,¦ proteins, interferons and ~yeast or bacterial presequences I reco~nized by mammallan host cells. The sequences for these I secretory leaders fire known, as are suitable sources for DNA
I encoding same if it ls not desired to synthesize the DNA 1n ~itro.
They are linked to DNA ~coding mature TGF-~ by restriction enzyme digestion of the DN~ containing ~he desired signal and the preTGF-~
D~A. Synthetic oligonucleotides are prepared in order to in~roduce unique restrlction sites (linkers) and, if necessary, DNA fragments j~ needed to complete a~y preseq~ence and mature TGF-~ coding regions remo~ed during restric~ion enzyme digestion. The synthesized 1; linkers and/or fragments then are ligated to the restriction enzyme J: digest fragments containing the substitute signal and TGF-~ coding region, ~nd lnserted in~o ~ cloning vector and the vector is used to transform bacterlal hosts. The mutant presequence thereafter is cloned in~o an express~on vector and used to transform host cells.

~, .

-12- ~ 3~
An illustrative example eMpl~ying a viral envelope protein presequence is described below.

Optlmally, DNA encoding the complete heterologous preseqllence is linked to the first codon of TGF-~ DNA. Alternati~ely, DNA
encoding the mature TGF-~ coding sequence is llgated to DNA
encoding the complete heterologous presequence plus a short portion, e.g. 21 to 45 base pairs, encoding the mature heterologous prote~n; this will result in the secretio~ of a fission which is useEul as an immun~gen or which can be cleaved to yield mature TGF-(for example, by insertion of a collagenase cleavage site between th~ N-terminus of TGF-~ and the C-terminu~i of the heterologous protein fragment). The objective of these constructions is to substitute a high efficiency secretory system for the natiYe TGF-fl secretory leader. However, it is by no means necessary to secrete TGF-~ in order to produce it in recombinant culture.

Other deletion-inserti~n ~utants include linking mature TGF-j~
specie~ to vir~l proteins expressed in large intracellular quanti~ies, e.g. retroviral coa-e proteins, large T antigen from SV40 and the like, or to immunoganic bacterial proteins or polypeptide~ such as chemotac~i.c polypeptldes, in particular the potent chemotactic tripeptide Met-Leu-Phe-.

Expressed ~ariants of preTGF-~, ma~ure TGF-~ or fragments there~f will ~xhibit amino acid sequences thi~t gradually depart from the Fi~s. lb or 5 s2quences as the nu~ber and scope of in~ertions, deletio~s and substitution~ inereases. This departure is measured as a reductlon in homology between preTGF-~ and the `I 30 variant. All protein~i or polypeptides that display TGF-~ ~nchorage indepe~dent gr~w~h-promoting biological activity are included i; ~ wi~hin ~h~ ~cope of this in~ention, regardless oP the degree o homology that ~hey s~ow to the Fig. 1 protein. The reason for this is that some regions of preTGF-~, e.g. the prssequence, are readily . ~ 35 mutated, or e~en completely deleted as in the case of mature TGF-~, ; :

-13- ~3~3~
and thus biological activity will be retained. On the other hand, deletion of the nine cysteine residues (and accompanylng di~ulfide linkages) in the mature TGF-~ molecule will have a substantial adverse impact on this biological activity and in all likelihood would completely ~brogate biolo~ical activity. In addition, a substitution mutant may exhlbit full TGF-~ growth-promoting activity and yet be less homologous if residues containing functionally similar amino acid side chains are substituted.
Functionally similar refers to dominant characteristics of the side chalns such as hydrophobic, basic, neutral or acidic, or the presence or ~sence of steric bul~. Thus the degree of homology that a given polypeptide bears to preTG~-~ is not the principal measure of its identity as TGF-~. However, as a general guide, proteins or polypeptides that share at least some biological activity with maturP TGF-~ fEom natural sources and which are substantially homologous with the Fig. lb sequence are to be considered as falling within the scope of the term TGF-~, e.g., ~ TGF-~ variants b~ing about :Erom 40 percent to 100 percent ,j homologous with preTGF-~ or any frag~ent thereof greater than about ¦ 20 20 r~sid~es, including varlants having an amino acid sequence greater than about 75 percent homologous with the mature TGF-~l sequence. With respect ~o neoplastic cell growth-inhibiting ~ activlty, TGF-~ excludes pclypeptides known heretofore to exert j such growth inhibitory activity, e.g. interf~rons, tumor necrosis factor and lymphotoxin, but otherwise need not necessarily have homologous regions with the Fig. lb sequences.
. : ' .
More narrow ~nd specific factors in establishing the identity of a polypeptide as TGF-~ ~re (a) the ability of antisera which ar~
i 30 capa~le of subRtantlally neutralizing the growth inhlblt~ry or the `j anchorage independent growth promoting activity of mature TGF-~
j also to substantially neutralize the activity of the polypeptide in ~ question, or (b) the abillty of the candidate polypeptide to I ~ compe~e with TGF fl for a TGF-~ cell surface receptor. HoweYer, it , 35 will be reco~nized th~t immunological identity and growth promoting . I .

..... . , . ... , ., .,, . .... . , . , . . " , , , , ~

~3~3~

identity are not necessarily coextensive. A neutralizing antibody for the mature TGF-~ of Fig. lb may not bind a candidate protein because the neutralizing antibody happens to not be directed to a site on TGF-~ that is critical for i~s growth promoting activity.
Instead, the antibody may bind an innocuous region and exert its neutralizing effect by steric hindrance. Therefore, a candidate protein mutated in this lnnocuous region might no longer bind the il neutralizing antibody, but it would nonetheless be TGF-p in terms of substantial homology and biological activity~
~ The TGF-~ residue~ whlch are subject to site-directed ,, mutagenesis for the preparation of variants which are likely to be i antagonists to biologlcally ctive TGF-~ are the cysteine residues, Argl8. Lysl9~ LeU20. TYr21. Ile22, Phe24, Leu2g, Gly29~ Trp3~, TrP32~ 33, Pro36, GlY38. Tyr3g, Asn42, rly46, Pro4g, Leu62, TYr65, Pr70~ Val79, Pro80, LeU83, Leug6, Ilegg, Valgo, Tyrgl, Tyrg2~ ~eU102~ ASnl05J Metl06. Ilel07 and Vall08~ Substitutions which are made at these residues generally will be non-conserved, i.e. the substituted residue (a~ differs su~stantially in ;¦ ~0 hydrophobicity, for example a hydrophobic residue (Val, Ile, Leu, I Phe or Met) substituted for a hydrophilic residue such as Arg, Lys, ¦ Trp or Asn, or a hydrophilic residue such as Thr, Ser, His, Gln, ' Asn, Lys, Asp, Glu or Trp substituted for a hydrophobic residue;
-l (b) differs substantially in its effect on polypeptide backbone , 25 orientation such as substitution of or for Pro or Gly by another -j residue; (c) differs substantially in electric charge, for example `J ~ substitution of a negatiYely charged residue such as Glu or Asp for ~¦~ a positively charged residue such as ~ys, ~is or Arg (and vice versa); or (d) differs substantislly in steric bulk, for example substitution of a bulky residue such a~ His, Trpj Phe or Tyr for j ~ one having a minor side chain, e.g. Ala, Gly or Ser ~and viceversa). Each of the foregolng target residues also ~s deleted (prefarably in pairs) or non-conserved residues (also preferably in pairs) are inserted ad~acent to the target residues. Ortinarily, ODly one residue at a ti=e i~ sub~ect to introduction of sequence .

. ' -15- ~2~
variatlon. The regions for investl&ation of site-directed variation are residues 105-112, 77-95, and 20-49.

Identification of antagonists is routine. The candidate is incubated together with an equimolar amount of TGF-~ otherwise detectable in the EGF-potentiated anchorage independent target cell growth assay~il, and the culture observed for failure to induce anchorage independent growth.

Antagonists that remain immunologically cross-reactive with nat~ve TGF-~ are useful in immunoassays as standards or, when labelled, as reagents in competitlve-type assays. Antagonists also are useful in therapy, e.~. oi TGF-~ dependent tumors.

~e ~ame residues targeted for site-directed variation in the generation of candldate antagonists also are targe~ed for the generation of agonists. Here, however, the variant residues which are substituted or inserted are conserved, i.e., members of each class, e.g. hydrophobic, electronegative and the like as described above, are substituted for one another or inserted adJacent to a ! member of the same class, again preierably in palrs.
1 .
Also within the scope of this invention are TGF-~ vsriants representing fusions of one TGF--~ allele or sub-type with another, for example a C-terminal domain from TGF~ having the homologous N-term~nal domain substltuted from TGF-~3, or substitution~l ~, ~ariants in which a domain from one allele or subtype is substituted for the homologous region from another allele or I subtype.
" 30 It Is importan~ to observe that characteristics suc~ as molecular weight and the like for the native or wild type mature ~ TGF-~ of ~ig. lb obtained from placenta or platelets are -,~ dascriptive only for the nat~ve species of TGF-~. The variants i 35 con~emplated herein may modify the characteristics of native TGF-~

~ 2 ~

considerably, and this in fact may be the objective of the mutagenesis as is more fully described below. While TGF-~ as defined herein includes native TGF-~l, other related biologically active polypeptides will fall ~ithin the definition. TGF-~ species like the in~ertion mutants, deletion mutants, or fusion proteins described above will bring the mutant outside of the ~olecular weight established for native TGF-~. For example, fusion protelns with mature TGF-~ or preTGF-~ itself wlll have a greater molecular weight than native, matuxe TGF-~, ~hile deletion mutants of mature TGF-~ will ha~e a lower molecular weight. Similarly, TGF-~ is engineered in order to introduce glycosylation sites, thereby resulting ln glycosylated TGF-~, or to substitute serine for cysteine at ~tes not critical for biological activity. Finally, post-translational processing of hu~an preTGF-~ in cell lines '~ 15 derived from nonprimate m = als may produce microheterogeneity in the amino terminal region of mature TGF-~, so that alanine will no lon~er be the amino terminal amino acid.

Note that the languags "capable~ of inducing anchorage independent growth in the deinition for biolo~ical activity means that the preTGF-~ or fragments thereof include polypeptides which can be converted, as by enzymatic dige~tion, to a polypept~de I fragment which exhi'bits the desired blological activity.
! Typically, inactive precursors will be fusion pro~eins in which l~ 25 mature TGF-~ is link~d by a peptide bond at its carboxyl terminus '~ to an insoluble or ~elatinous protein. The sequence at or wlthin ¦~ the re~ion o~ this peptide bond ls selected so as to be susceptible ;~ to proteolytic hydrolysig whereby TGF-~ is released, either i~ vivo for ~n situ generation or, as part of a manufacturing protocol, in I 30 vit~o.

',~ While TGF-~ ordinarily means human TGF-jB, TGF-~ from souroes ~; such as murine, porcine, equine or bovine is included within the definit~on of TGF-~ so long as it otherwise meets the standards described iabove for biological activity. TGF-~ is not species , ~ :
~( . .

i ::

-17- ~3~6~9 specific, e.g., murine and human TGF-~ are both effective in inducing anchorage independent growth of the same cell line.
Therefore, TGF-~ from one species can be used in therapy of another species. DNA encoding the TGF-~ of other specles is obtained by probing cDNA or genomic libraries rom such spec~es with labelled human preTGF ~ cDNA.

Derivatlves of TGF-~ are included within the scope of this in~ention. Derivative~ include glycosylated and covalent or ag~rega~ive conjugates with other TGF-~ molecules, dimers or unrelated chemical moieties. Covalent derivative~ are prepared by linkage of functionalit~es to groups which are found in the TGF-~
amino acid chains or at the N- or C-termini, by means known in the art. These derivati~es may, or example, include: aliphatic or acyl esters or amides of the carboxyl ter~inus, alkylamines or residues containing carboxyl side chains, e.g., con~ugates to alkylamines a~ aspartic acid residues; 0-acyl derivatives of hydroxyl group-containing residues and N-acyl derivatives of the ~ amino ~erminal ami~o acid or amino-group containing residues, e.g.
ji 20 con~ugates with fMet-Leu-Phe or immunogenic pro~eins. Derivatives of the acyl groups are selected from ~he group of alkyl-moieties ~including C3 to C10 ~or~al ~lkyl), thereby forming alkanoyl ' species, and carbocy~lic or heterocyclic compounds, thereby forming :1 aroyl species. The reacti~e groups preferably are difunctional compounds known ~ se for use in cross-llnking proteins to ~nsoluble matrices through reacti~e side groups.

~; Covalent or aggregative derivatives are useful as reagents in immunoassay or for affinity purification procedures. For example, TGF-~ is insolubil~zed by covalent bonding to eyanogen bromide-activated Sepharose by method~ known per se or sdsorbed to polyolefin surfsces (wlth or without glutaraldehyde cross-linking) for use in the ~ssay or purification of an~i-TGF-~ antibodies or cell surface receptors. TGF-~ ~lso is labelled with a de~ectable , 35 group , e . ~., radioiodinated by the chloramine T procedure, 1~
:, i; :

-18- 132~639 covalently bound to rare earth chelates or conjugated to another fluorescent moiety for use in diagniDstic assays, especially for diagnosis of TGF-~ levels in biological samples by competitiYe-type immunoassays.
TGF-~ variants generally are made by predetermined, i.e. site spec1flc, methods. The objective of site specific mutagenesis is to construct DNA that encodes TGF-~ as defined above, i.e., TGF-~
which exhibits biologlcal acti~ity.

While the mutation site is predetermined, it is unnecessary that the mutation ~E se be predetermined. FDr example, in order ! to optimize the performance of the mutan~s at a given position random mutagenesis is con~lucted at the target codon and the , 15 expressed TGF-~ vari~nts are ~creen0d for optimal activity.
¦ Techniques are will known for making ~.ubstitution variants at predetermined sites in DNA having a known sequence, for example M13 primer mutagenes~s.

TGF-,,B mutagenesis is conducted by making amino acid j insertions, usually on the ordler of about from 1 to 5 amino acid residues, ~Dr deletions of about from 1 to 10 residues.
; Substitutions, deletions, insirtions or any subcombination may be combined to arrive at a final construct. As noted abo~e, i ~ 25 insertions include amino or carboxyl-terminal fusions, e.,~, ~7ith a bydrophobic or immunogenic proteln. The mutations ~n the DNA
'~ encoding such mutations should not ultlmately place the sequence out of reading frame in an expression vector whereby the resulting protein is not biologically active TGF-,,3. Thè mutations also 3~ 30 prefer~bly will not create complemcntary regions that could produce translation-suppressing secondary mRNA structure.

Included herein are heterodim~rs of TGF-,,B. These typically ~ inClude dimers in ~hich one TGF-~ chain is from one subclass, e.g.
,~ ~ 35 TGF-~l, while the other is from another subclass, e.g. TGF-,B3.
i :

~L 3 ~

Similarly, TGF-~ amino acid sequence Yariants are produced a~
homodimers or heterodimers with oth2r ~mino acid sequence variants or with native TGF-~ 6equenceq. Heterodimers include dimers containing a first TGF-~ sequenc~ that is biologically active when present in a homodimer and a se~ond TGF-~ sequence that is not biologically active. Such heterodimers may be active as TGF-~
j antagonists. Heterodimers are readily produced by cotransforming host cells with DNA encoding the TGF-~ chains selected. A
proportion oi the TGF-~ so produced is expr ssed as the desired , 10 heterodim~r, and transformant~ are screened for those which produce the greatest proportion of heterodimer.

. DNA which encodes TGF-~ is obtained by chemical synthesis, by j screenin~ reverse transcripts of mRNA from placental or o~her cells or by screening genomic libraries from eukaryotic cells. This DNA
ne~d not use the codons set forth in Figs. lb or 4a-4c so long as ~he host ~ell recognizçs the codons which are used. DNA of thls sort is as easily manufactured ~a vltro as the DNA of Figs lb or ~ 4a-4c. Also useful herein is nucleic acid, either RNA or DNA, 1 20 which doss not encode TGF-~ thereof as defined herein but which nonetheless is capable of hybridizing with such DNA or RNA.
~¦ Noncoding but hybridizing nuc]Leic acid, while not used in the recombinant synthesis of TGF-~, is useful as an intermediate for 1 making labell~d prsbes in diagnostic assays for TGF-~ mRNA or ¦ 25 genomi~ ~NA in teqt cells.
:
Diagnostic nu~leic acid is cov~lently labelli~.d with a detec~abl~ substance such as a fluorescent group, a radioactive ~ atom or a chemiluminescent group by methods known 2~E se. It is ;~j 30 then used in con~entional Southern or Nort~ern hybridization ssays. Such assays are employed in identifying TGF-~ vectors and transformants as described in the Examples ~ , or for ~ ~itro diagnosls such as detection of TGF-~ mRNA in tissues iQS a measure of mltogenic activity.
j:J
',~

i ~32~3~

TGF-~ is synthesized herein in host cells transformed with vectors containlng DNA encodlng TGF-~. A vector is a replicable nucleic acid construct. Vectors are used either to ampllfy and/or to express ~NA which encodes TGF-~, or to amplify DNA that hybridizes with DNA or RNA encoding TGF-~. An expression ~ector ls a replicable DNA construct in which a DNA ~equence encoding TGF-~
; is vperably linked to sult~ble control sequences capable of effecting the expression of TGF-~ ~n a suitable host. Such control sequences include a transcript~onal promoter, an optional operator sequen~e to control transcription, a sequence encoding suitable ! mRNA ribosomal binding sites, and sequences which control ~ termlnation of transcription and translation. For expression of $ TGF-~ in eukaryotic cells the vector also should include DNA
~ encoding a selection gene. However, the selection gene can be ; 15 supplied by an unlinked plasmid in cotransformation.

¦ Vectors comprise plasmids, viruses (including phage), and integratable DNA fragments i.e., fragments that are integratable lnto the host genome by recombination. In the present specification, the vector is a plssmid in the sense that it is ~ cloned in bacterial cells, but it integrates into the host cell i~ genome upo~ cotransformation. However, all other forms oi vectors which serve an equi~alent function and whlch are, or become, known in the art are suitable for use herein. Suitable vectors will ~¦ 25 contain repllcon and control sequences which are derived from species co~patible with the intended expression host.

DNA regions axe operably linked when they are functionally i related to each other. For exampIe, DNA for a presequence orsecretory leader i9 operably linked to DNA for a polypeptide if it is expressed as a preprotein which participates in the secretion of the polypeptide; a pro~oter is operebly linked to a coding sequence '1 ~
if ~ controls the transcr~ption of the sequence; or a ribosome bindlng site i~ operably linked to a coding sequence if i~ is ~ ~ 35 ~ positioned so as to permit trenslation. Generally, operably linked !~

-21- ~2~
means eontiguous and, in the case of secretory leaders, contiguous ; and in reading phase.

Cultures of cells derived from multicellular organisms are the preferred host cells herein. In principle, any higher eukaryotic cell culture ~s workable, whether irom vertebrate or invertebrate culture. Howe~er, interest has been grea~est in lnvertebr~te cells, and propagat~on of vertebrate cells ln culture ~r se has become a routine procedure in recent years ~Tissue Culture, Academlc Press, Kruse and Patterson, editors (1973)]. Examples of useful host cell lines are V~RO and HeLa cells, Chinese hamster ovary (CHV) cell lines, and WI38, BHK, COS-7, 293 and MDCK cell lines.
,, ~ 15 Many eukaryotic cells are known to synthesize endogenous TGF-i ~. l'hus, many potential host cells synthesize TGF-~ of the host species. This TGF-~ therefore is present in the TGF-~ produced by I transcription and translation of the transfor~ing DNA. For 3 ex~mple, hamster TGF-~ is present in transformed CHO cells, andJ 20 thus is present ln cell extracts containing human TGF-~ harvested ¦ from such c~lls While this is not necessarily disadvantageous because animal and human TGF~ is active across species lines, it ~ is desirable to select a host/~e~ct~r system that secr~tes as little ¦ of the endogenous anim~l TGF-~ as possible. Thls is accompl~shed J 25 by (a) selectin~ host animal cell lines that synthesize low levels .j o~ ani~al TGF-~, (b) transforming the animal cell line with a vector for high efficiency TGF-~ secretion (described above) and recovering human TGF-~ from the culture mediu~ or (c) transforming a human cell line, whereby any endo~enous hTGF-~ that ls produced would not be a contaminant.

It may be desirable to use a host cell line that is 'I
differenti~ted to ~ynthesize 0ndcgçnous TGF-~. Examples include megakaryoblast, promegakaryocytic or basophilic megakaryocytic cell lines. If su~table cell lines are not available they m~y be ~ 1 ~
' -22- ~3~6~3~
produced by EBV immortalization of megakaryoblasts, promegakaryocytes or basophilic me~akaryocytes recoYered from mammalian bone marrow. The TGF-~ of the desired species is recovered from transformant cell cultures by immunoaffinity chromatography using antibodies specific for host TGF-~.

Expression vectors for such cells ordinarily include an origin of replication (for extrachromosomal amplification), a promoter located upstream from the TGF-~ coding sequences 9 along with an enhancer if desired, RNA splice si~e (if intron-containing TGF-~-encoding genomic DNA ls used), and a transcriptional termination sequence including a polyadenylation site located 3' to the TGF-~
sequence.

The transcriptional and translational control sequences in expression vectors to be used in transforming vertebrate cells preferably are provlded from ~iral sources. For e~ample7 commonly used promoters are derived from polyoma, AdeDo~irus 2, and most 3 preferably Simian Virus 40 (SY40). The early and lat0 promoters of SV40 are particularly useful because both are obtained easily from the virus as a fragment which also contains the SV40 viral origin of replication54. Smaller or larger SV40 fra~ments may also be used, provided the approximately 250 bp sequence extending from the Hind III site toward the Bgl I site located in the viral origin of replication is inoluded. Since TGF-~ sppears to be toxic to ; ma~malian cell transformants and thus may interfere with attempts : ; to amplify the ~ene, y~elds may be improved by the use of inducible pro~ot~rs, e.g. the metallothionein promoter' Drosophila heat shock promoter or mouse mammary tumor virus promoter. Further, it is ,A~ ' 30 also poss~ible to utiI~ze the~TGF-~ genomic promoter, control and/or si~nal sequences oormally associated with TGF-~, provided such control sequences are compatible with and recognized by the host cell. If TGF-~ untranslated regions are included in expression vectors, yields ~ay~ be improved by substituting A or T bases for G

?- ~: : .
, ~ ~

~32~

or C bases immedlately 5' to the start codon and deleting G-C rich domains in the 3' untranslated sequences of the cDNA.

An orlgin of replication may be provided either by construction of the vector to include an exogenous origin, such as may be derived from SV40 or other viral (e.g. Polyoma, Adeno~irus, VSV, or BPV) source, or may be provided by the host cell chromosomal replication mechanism. If the vector is integrated into the host cell chromosomP, the latter is often sufflcient.
Rather than using vectors which contain viral origins of replication, one can transform mam~alian cells by cotransformation with a selecta~le marker and the TGF-~ DNA. An example of a suitable selectable ~arker is dihydrofolate reductase (DHFR) or thymidine kinase. Such markers are prot~ins, generally enzymes that enable the identification of transformant cells, i.e., cells which had been competent to take up exogenous DNA. Generally, identification is by survival of transformants in culture medium ;, that is toxic or from which the cells cannot obtain critical i 20 nutrition without having taken llp the marker protein. In selecting a preferred host maMmalian cell for tranqfection by veotors which comprlse DNA sequences encoding both TGF-~ ~nd DHFR, it 1s ! appropriate to sele~t the host according ~o the type of DHFR
¦ ~ protein employed. If wild type DHFR protein is employed, it ispreferable to select a host cell which is deficient in DHFR, thus permitting the use of the DHFR codin~ sequenc~ as a marker for successful transfection in selective medlum which lacks hypoxanthine, glycine, And thymidine. An appropriate host cell in his case is the ~hinese ham~ter o~ary (CHO) cell line deficient ln DHFR activi~y, prepared and propagated as dsscribed by Urlaub and Chasin55.

On the other hand, if DNA encodin~ DHFR protein with low ~ bindin~ affinity for methotrexate (MTX) is used as the controlling "!~ 35 ~ sequence, it i6 not necessary to use DHFR reslstant cells. Because 3 ~

the mutant DHFR ls resistant to MTX, MTX containing medla can be used as a means of selection provided that the host cells are themsPlves MTX sensitive. Most eukaryotic cells which are capable of absorbing MTX appear to be methotrexate sensitive. One such useful cell line is a CHO line, CHO-Kl (ATCC No. CCL 61).
Alternatively, DHFR+ host cells are used by cotransforming the i cells with DNA encoding the neomycin resistance gene, DHFR and TGF-. The initlal transfections are screened for neomycin resistance, and resistant transformants then amplified on MTX.
I Other methods suitable for adaptation to the synthesis of TGF-, ~ in recombinant vert~brate cell culture are described in Gething 3 et al.56, Mantei et al.57, and Levinson et al.58~59.
TGF-~ is reco~ered from lysed, transPormed rells and insoluble cell debris separated by centr~fugation. Alternatively, the c1~1ture supernatants from transormed cells that secrete TGF-~ are simply separated from the cells by centrlfugation. Then the TGF-~
generally is purified by me~hods ~nown in the ~rtl5~16~17 using gel ~3j 20 filtration in the presence of acid ollowed by HPLC and el~tion on an acetonitrile grad~ent. ~owever, such methods are not j necessarily required to prepare a therapeutic product.

As a further or substitute purification step, cell lysates or supern~tants are heated or a priod and at a temperature suficient to denature ~nd precipi~te contaminant proteins but not GF-~; TGF-~ is a remarkably heat stable protein, perhaps as a result of ~xtensiYe disulfide bond formAtion. As a result, the ~ heat~ng shsuld be conducted in a medium that contains low amounts ,~ 30 of disulfide reagents 5uch as dithiothreitol or the like. Heating 3~ also i5 combined w~th acldification since TGF-~ is known to be s~able to lM acetic acid.

~ ;Natu~e, native TGF-~ is not glycosylated. Therefore it is 3~ ` 35 separated rom any residual cont~minant heat- anA acid-stable - 25 ~

glycoproteins by ad~Zorbing the glycoproteins Z~n lectin col~Zmns such as lentil lectin-linked sepharose. This step, less desirably, can go before the heat and acid treatment. TGF_ZBZ will elute with the unadsorbed fraction. The recombinant TGF-~Zis recovered from host 5 cells exp~essing endogenous TGF-3 (or undesired homodimers) by transforming the host cells with a TGF-ZBl variant which is glycosylated by the host. The sugar "tag" enables the recon~inant TGF-Z~ to be recovered free of endogenous TGF-~f by lectin affinity chromatography, elution of the glycosylated TGF-S and removal, if Zf 10 deZ~lired~ of the Zsugar residue3 by conventional enzymatic digeZstion.
,Z
, ii If high purity product is desired the crude or partially Z purified mixture thereafter is subjected to chromatofocusing.
.,Z :-I 15 TGF-f~ is prepared for administration by mixing TGF-Z~f at the desired degree of purity with phyZ~iologically acceptable carriers, i.e., carriers which are nontoxic to recipients at the dosages and concentrations employed. Ordinarily, this will entail cornbining TGF ~Zwith bufers, antioxidants such as ascorbic acid, low molecular 20 weight (less than about 10 resicZiues) polypeptides, proteins, amino acids, carbohydrates including glucoZse or dextrins, chelating agents such as EDTA, and other excipients. TGF-1,3 for use in therapeutic adminiZstration must be sterile. This is readily accomplished by filtration through sterile filtration (0.2 rnicron) membranes. TGF-ZB
25 ordinarily will be stored as an aqueous solution sinceZ it if~ highly Z stable to thermal and oxi,dative denaturation.

TGF-Z~f optionally is corf~ined with activating agents such as ~1 TGZF_fZX or EGF species as iZS dZelscribed further in Canadian Patent No.
j 30 I,223,198 granteZd JuneZ 23, 1987, and is administered in accord with f said application.
, .
¦ Various therapeutic indications for TGF-~I'f compositions are `Z known.
lZ

~1 :1 ..
,.f, 1 ` ~
f ,:

~32~

The first, and preferred, indication is topical application to incisions or exposed tissue for the promotion of wound healing.
There are no limltations as to the type of wound or other traumata that can be treated, and these include (but are not limited to):
S firs~, second and thlrd degree burns ~especially second and third degree); epidermal nnd internal surgical incisions, including those of cosmetic surgery; wounds, including lacerations, incisions, and penetrations; and epidermal ulcer~ including decubital ~bed-sores~, dlabetic, dental, hemophillac, and varicose. Doses such as those previously described for wound healin~l7 will be suitable as startin~ doses in these indicatlons.

TGF-~ compositions are applied to burns ln the form of a ~ sterile irrigant, preferably in combination with a physiological A 15 saline solution, or in the form of ointments or suspenslons, preferably in co~blnation with purified collagen. The compositions also may be impre~nated into transdermal pa~ches, plasters, and bandages, preferably in a llquid or semi-liquid form.
Automicrobial agents such as silver sulfadiazine should be lncluded in such articles or compos1tions. Debridement agents such as proteolytic en~ymes also can be included if they do not hydrolyzP
TGF-~ or a hydrolysis-resistan~ TGF-~ mutant is employed.
':
TGF-~ also is administered ~ystemically for the treatm2nt of wounds and similar traumata. SystPmic administration is useful i provided that there are no, or limited, undesirable side-effects, such as the ~timulation o neoplastic cellular growth in patients with cancer. TGF-~ compositions for systemic ~dministration prefer~bly nre formu~atad as sterile, lsotonic parenteral ~ 30 inJection~ or infusions.
I
i The ~mount of ~cti~ating agent (such ~s TGF-~, EGF or other ~! ~ growth factors) administered wlth TGF-~ depends directly upon the moun~ of TGF-~ present in the act~ated composit~ons as i: :
~` ' 7 ~ : :
'.,~
~i ~32~

administered to the recipient, the growth factors selected and the clinical status of the patient.

Initial dosing of TGF-~ should bs delivered to the ~herapeutic site in a concentration of about fro~ 0.1 to 150 ng/ml and thereafter adJusted in line with clinical exper~ence. Since TGF-~
compositions both provoXe and sustain cellular regeneration, a continual application or periodic reapplication of t~e compositions ~s indlcated. The cliniclan will be expected to modify the dosage in ac~ordance with clinical experience.

In order to simpli~y the Examples certain frequently occurring methods will be referenced by shorthand phrases.

Plasmids are designated by a low case p preceded and/or followed by capital letters ~nd/or numberg. The starting plasmlds herein are commer¢ially available, are publicly available on an unrestricted basis, or can be constructed from such available plasmids in accord with published procedures. In addition, other equivalent plasmids are kno~m in the art and ~111 be apparent tG
the ordinary artisan.

"Digestion" of DNA refers to catalytic cleavage Df ~he DNA
with an enzyme that acts only at certain locations in the DNA.
Such enzymes are called restrictLGn enzymes, and the site fDr which each is specific is ealled a restriction site. The varlous restriction enzymes used herein are commercially available and ~heir re~ction conditions, cofactors and other requlrements as established by ~he eDzy~e suppliers were used. Restriction enzymes commonly are designated by ~bbrevlatlons composed of a capital ` letter followed by other letters representing the microorganism rom which each restriction enzyme orlginally was obtained and then a nu~ber designating the partioular enzyme. Xn general, about 1 ~g of~plasmid or DNA ~rag~ent is u~ed with about 2 units of enzyme ln about 20 ~l of bufar solution. Appropriate buffers and substrate amounts for particular restriction enzymes are specified by the manufacturer. Incubation times of about 1 hour at 37 C are ordinarily used, but may vary in accordance with the supplier's -~ instructions. After incu~ation, protein is removed by extraction with phenol and chlorofor~, and the dige~ted nucleic acid is reco~ered from the aqueous fractlon by precipitation ~ith ethanol.
! Digestion with a restriction enzyme infrequently ~s followed with bacterial alkaline phosphatase hydrolysis of the terminal 5' phosphates to prevent the two restriction clsaved ends of a DNA
fra~ent from "circularizing" or forming a closed loop that would i~pede insertion of another DNA fra~ment at the restriction site.
'J Unless otherwise stated, digestion of plasmids is not followed by ~` 5' terminal dephosphoryla~ion. Procedures and reagen~s for dephosphorylation are conventional85.
"Recovery~ or "isolation~ of a given fragment of ~NA from a rsstriction dlge~t means separation of the digest on polyacrylamide or agarose gel by electrophoresls, identlf~cation of the ragment of interest by comparlson of its~ mobility ver~us that of marker DNA
fragments of known molecular ~7eight, removal of the gel section containing the d~sired fragment, and separation of the gel from DNA. This procedure is known generPlly~6~87.

; , "Southern Analysis" is a method by which the preqence of DNA
sequences in a digest or DNA-containing Gompositlon ls confirmed by ~¦ hybridization to a known, labelled oligonucleotide or DNA fragment.
For the purposes herein, unless otherwise provided, Southern analy6is shall mean separation of d~gests on 1 percent agarose, denaturation and transfer to nitrocellulose by the method of E.
~ 30 Southern68, and hybridizstLnn as describet by T. M~niatis et al.88.
: !~
`,; "Transformation" means introducing DNA in~o an organism so `l~ that the DNA is replicable, e~ther as an e~trachromosomal element or chromosomal integrant. Unless otherwise provided, the method ''.

.. ~ .

-2g-used herein for transformation of E. coli is the CaC12 method of Mandel et al.89 nLigation~ refer~ to the process Df forming phosphodiester bonds between two double stranded nucleic acid fragments (T.
Maniatis et al., Id., p. 146). Unless otherwise provided, ligation may be accomplished using known buffers and conditions with 10 units oi T4 DNA ligase (nllgase~ per 0.5 ~g of approximately equimolar a~ounts of the DNA frAgments to be ligated.
. 10 "Preparation" of DNA from transformants means isolating pla~id DNA fro~ microbial culture. Unless otherwise provid~d, the alkaline/SDS method of Maniatis et al., Id~, p. 90, ~ay be used.

i 15 "Oligonucleotides" are short length single or double s~randed polydeoxynucleotides which are chemically synthesized ~y known ~ methods and ~hen purified on po:Lyacrylamide gels.
,~
'~ E~9MPL~ 1 Purifi~tlon and Sequelce Analy~i8 0~ Human TGF-0 The known purificntion me~hod of Assoian e~ al.15 was æcaled up and modified to ob~ain enough homogeneously pure hu~an TGF-~l for amino acid sequencing. :250 units of hu~an platelets were extracted in a Waring blender with 1 litre of acid-ethanol.
Addition of 4 liters o~ ethsr gave rise to a p~ecipitate which was collect~d by ~acuum fil~ration ovr Whatman No. 1 paper. The precipitate was di~solved overnight in 50 ml of lM acetic acid and purlfied by gel fiitration on a Biogel P-60 colu~n (lOxlO0 cm), equilibrated in lM acet~c acid. The fractionq containing TGF-~l were ident~fied by analytical S~S-polyacryla~ide gel electrophoresis and bioassaylS. - Peak fractions were pooled, freeze-drisd a~d redissolved in 20 ml lN acetlc acid, 8M urea.
Subsequent gel filtration ~ver a Biogel P-60 column (5x90 cm) in lM
1~ acetic acid, 8M urea yielded about 50 percent pure TGF-~. These i~ 35 peak fractions were then diluted wit~ 1 volume of water and applied .,,: : : :

-30- 1 3 ~ ~ g ~
to a ssmipreparative RPP C18 (Synchropak)* HPLC column in 0.1 percent trifluoroacetic acid and eluted ~ith a 20-50 percent acetonitrile gradîent. The TGF-~l thus obtained was quantitated by amino acid analysis, showing a yield cf about 0.5 mg per preparation. Denaturing SDS-polyaorylamide gel electrophoresis was performed as described60. In agreement with previous work the non-reduced TGF-~l migrated ~s a 25 kD protein in a SDS-polyacrylamid~
gel, while reduction with ~-mercaptoethanol converted it into a 12.5 kD specles. This suggested that TGF-~ consists of two 12.5 kD
polypeptide chains linked by intermol~c~lar disulfide bridgesl5.

In order to obtain protein sequence information, the purified TGF-~l was reduced, alkylated and subjected to amino-terminal sequence analysis. 1.2 nmole of TGF-~ was dialyzed into 8M urea and reduced by incubati~n in O.LM Tris-HCl (pH ~.5), 10 mM
dithiothreitol, 8M urea. Subsequent al~ylation took place in the , presence of 50 mM iodoacetate at room temperature in the dark.
¦ This reaction was terminated after 30 min. by addition of an excess~-mercaptoethanol and dialysis. 0.7 nmole of this TGF-~l was used for the direct NH2-terminal sequenee analysis. 1.2 nmole of reduced and alkylated ~GF-~l was di~ested in O.75 M urea, 50 m~
~ NH4HC03, 5mM dithiothreitol for 24 hours ~ith 1 percent I clostxipalnl5. An additional 1 perc:ent of clostripain was added ¦ ater 12 hours reaction time. The rsaction products were separated on a Synchropak RPP C18 reYerse phase colu~n ~4.6 ~ 250 m~) with a 0-70 percent acetonitrile gradient i~ 0.1 percent ~trifluoroacetic acid. Sequence deterMination took place usin~ either an extensively modif~ed Beck~an 890C spinning cup sequencer61 or a vapor phase sequencer as described by Hewick et al 62 (Applied Biosystems, model 470A), with amino acid derivative identification by reversed phase HPLC on a Rainin Microsorb*C-8 column. The amino `, acid sequence ~f several peptides was determined. One of these ; frag~ent~ was the NH2-terminal segment, while another large peptide yielded a 37 amino acid sequence which overlapped the NH2-terminal 35 ~ sequeDce and established 60 residues of contiguous sequence.

' ~ *trade-mark J

~L 3 ~

Unmodified TGF-~l was also treated with CNBr. Cleavage at the methlonine residue re~ulted in the complete loss of bi~logisal activity, documenting that at least 2art of this C-terminal octapeptide is needed for biological activity (data not shown).
~,.
~XA~PL~ 2 ~, Isolat10~ of a TGF-~ ~on The ~pproach we followed for the initi~l identification of a . 10 nuclsotide sequence encod~ng TGF-~l adopted the ~long probe"
', strategy used previously for TGF-~7. Long oligonucleo~ides i~ de3igned on the basis of the partlal pr~tein sequence were used as hybridi~ation probes for the ldentificat~4n of a TGF-~l exon in a human genomic DNA library. The TGF-~l exon was then used as a probe for the isolation of TGF-~l cD~As.
.~ .
Two 44-base-long te~xyoligo~uoleotides, ~LPl and ~LP2, comple~entary to sequences eoding for ~mino cids 3 to 17 and 30 to 44, respectively, were chem~cal:Ly synthesized 63~64. The choice of nucleotlde sequence was based u~pon the codon bias ~b~erved in human mRNAs26. CpG dinucleotides, which ar~ relatively rare in vertebrate DNA27, were ~voided whenever possible. In addition, . sixteen 14-mers were sy~thesized which are complementary to all : possible codons for amino acids 13 to 17. These deoxyoligo-~:~ 25 nucleotldes and the corregpond$ng iamin~ acid sequence ~re shown ~; below.
''I
NH~-AlaLcuAspThrAsnTyrCysPheSerSerThrGluLysAsnCysCysSlalArgGluLeu'ryr . ~ 30 CT~TGGTTGATAACGAAGAGGAGGTGTCTCTTCTTGACGACGCA 5' TTTTTAAcAAcAcA

L~uGlyTrpLy~TrpIleHisGluProLysGlyTyrHisAlaAsnPheCysLeuGlyProCysProTyr ~, : ... 0. ... .. .. .. .. .. .. .. .. .. .. ..
~ ; 40 ACCTTCACCTAGGTACTCGGTTTCCCGATAGTACGGTTG M GAC 5' .j :
: ~ :

. : :

-32- ~3~3~ ~

The nucleotides marked with a dot are bases for which there is no ambiguity in the codon.

A hwman genomic DNA library28 was screened under low strlngency hybridization sonditions using 32P-labelled ~LP-l as probe. ~pproximately 7.5 x 105 recombinant phage from a human genomic fetal liver library28 were hybridized using low stringency condition~5 with the 32P-lab211ed 44~mer ~LP-l after replica plating onto nitrocellulose fil~ers66. D~A was prepared from 58 of the hybridizing phage and hybridized with the 32p labelled ~LP-l i and ~LP-2 oligonucleotides uslng the dot blot" analysis method67 and Southern hybridization68 of ~HI diges~ion mixtures. The two l~ 15 phage DNAs which hybridized with both oligonucleotides were ! digested and probed ~ith the pool of 32P-labelled 14-mexs again by Sou~hern hybridization. 14-~er hybridizations were performed at 37C in 6xSSC, 0.5 perc~nt NP40, 6mN EDTA, lX Denhardt's solution I and 5C ~g/ml salmon sperm D~A. Several wa~hes were performed at room te~perature in 6xSSC before autoradiography. DNA from phage ~58 hybridized wlth the oligonucleotides ~LP-l, ~LP-2 ~nd w~th the 1 14-mer pool. The sequences h~)ridizing to ~LP-2 and the 14-mers were localized within the same 4.2 kbp ~HI fragment, while probe ~LP-l hybridized to a 20 kbp ~mHI fragment. The hybridizing Ba~HI
, 25 fra~ment.s were subcloned into pBR322. The nucleotide sequence of~: smaller hybridizing fragments was determined by dideoxynucleotide chain termination method69 ~fter subcloning into M13 deri~atives70.

; The screéning of the genomic D~A library resulted in th0 ~ isola~ion of an exon coding the part of the TGF-~l cod~ng sequence startlng at mature residue 10. In order to obtai~ the ent~re TGF-1 coding sequence, t~is exon was used as a probe to screen a ~gtlO
based cDN~ library derived from human term placenta mRNA.

35~ `~

~A~P7~. 3 Isolation of TGF~ As Total ~NA was extracted71 from the different cell sources and the polyadenylated mRNA fraction was isolated by oligo(dT)-cellulose chrom~tography72. The cDNA was prepared73 by prlming wit~ dT12 18 or the deoxyoligonucleotide ACACGGGTTCAGGTAC. The double-stranded cDMA was treated with nuclease Sl (Miles Laboratories) followed by E. coli D~A polymerase I ~lenow frag~ent ,' (Boehringer Mannheim) ~nd subcloned into EcoRI cleaved ~gtlO as ~, 10 described74, except that as~mmetric ~coRI linkers75 were used, thus '~' avoiding the need for the EcoRI methylase treatment. The ~ recombinant phage were plated on E coli C600 Hfl74 and replica'~ plated onto nitrocellulose filters6~. These were hybridized wlth 32P-labelled76 restriction frag~ents of the E~ample 2 exon at 42C
in 50 percent formamide, 5x SSC, 50 ~M sodium phosphate pH 6.8, 0.1 ~, percent sodium pyrophosphate, 5x Denhardt's solu~ion, 50 ~g/mlI salmon sperm DNA and washed in 0.2x SSC, 0.1 percent SDS at the same temperature. Low stringency hybridization conditions65 were ,1 used in the case of the 32P-labelled deoxyoligonucleotides. The ~l 20 nucleotide sequen~e of the TGF-~l cDNA restriction fragments was , determlned by the dideoxyoligonucleotide chain term~nation method69 ', after subcloning into M13 phage. derivatives70. The cDNAs obtained i are schematically shown in Fig. la. ~Cl was isolated from a hum2n ~'~, plac~nta cDNA library using ~le genomic exon (Fig. 3) as prob~.
The screening of approximately 750,000 ollgo-dT primed placenta ~, cDNA clones resulted ~n the $sDlation of one TGF-~ cDNA (~C13 of ~bout 1,050 bp. ~he prev~ously determ$ned partial TGF-~l sequence , esta~liPhed ~he reading fra~e and revealed the sequence codlng for ,~ the complete TGF-~l polypeptide. This sequence begins with the,, 30 N~2-terminal alanine residue and is followed 112 codons later by a '` s$op codon 9 only 20 base pairs from the 3' end. The ~Cl EcoRIcDNA insert w~s uqed in turn to 3creen the A172 glioblastoma cDNA
library leading to the isolation of ~C3.19. Screenlng of a speciflcally primed HT1080 ~ibrosarcoma cDNA library with the 32p labelled ~pnI-~E~I &nd the up~tream EcoRI-~E~I fragment of the , ~3~3~

A~C3.19 cDNA insert yielded ~C4.10, 4.33 and 4.37. Another similar library was screaned with the ~C4.33 insert and a synthetic 40-mer correspondlng to nucleotides 1-40, leading to the ,~ isolation of ~C5.7b.

:'!' Since none of more than seventy TGF-~ cDNAs isolated from different oligo(dT)-primed cDNA 11braries contained mGre than a few nucleotides of 3' untranslated region, the 3' l~ntranslated sequence ~ was determined using cloned genomic DNA. Hybridization analysis i~ 10 showed that the 3' end of the ~Cl cDNA insert ~as present in the genomic DNA phage ~58. DN~ sequence analysis re~e~led the presence o an e~on codlng for the carboxy terminal part of TGF-~l, followed by the stop codon and the 3' untranslated end (Fig. lb).
~ An AATM A hexanucleotide sequence32 was encountered 500 bp i~ 15 downstream from the termination codon, thus permitting an ~ assignment of the putative polyadenylation site. Assuming this is S indeed the polyadenylAtion si~nal, the calculated size of TGF-~l ¦ mRNA i in close agreement with the 2.3 to 2.5 kb length determined ~i from the Northern hybridi,zation experiments (Example 4).
Addltlonal screening of oligo(dT)-primed placenta and HT1080 cDNA
libraries using the genomic DN~ probe for the 3' untranslated end did not identi~y a single hybridizing ~DNA phage.
:Y
, d E~oKPLE 4 Diagnn:~L~hrtho~d l~ing~TG -~ cDN~Pro~es Polyadenyl~ted RNA was recov~red from the hepatoma HEP-G2, Uilms tumor ~TuWi, ~lioblastoma A172, bladder carcinoma T24, squamous epidermoid carcinoma A431, mammary carcinoma MCF-7, nasopharyngeal c rcinoma RB, fibrosarcoma HT1080, Burkit~ lymphoma ~ B-lymphoolaits Daudi and Ra~i, T-lymphoblast Mol~-4. Peripheral blood l~mphocytes (PBLs) were prepared and mitogen-induced with ; staphylococcal en~erotoxin B and phorbol myristats as described53.
NA ~as ha~vested in this case ater 24 hours. 4 ~ig of polyadenylAted mRNA was electrophoresed into formaldehyde~l.2 p~rcent agarose gel29 and blotted onto nitroeellulose fil~ers30.

,. ~
. ~ , .,: :

The 32P-labelled76 EcoRI cDNA insert of ~Cl was usad as probe ; under high stringency conditions used above. Comparison ~ith the position of the 28S and 18S rRNA on the gel suggests a length of 2.3-2.5 ~b for the TGF-~ mRNA. In some cases a smaller mRNA
species may be present, although partial degradation of the mP~A
cannot be excluded.

TGF-~ mRNA was detectable in all human tumor c~ll lines including tumor cells of neuroectodermal origin, such as TuUi j 10 (Wilms Tumor) and A172 ~glioblastoma), and the carclnoma cell lines T24 bladder carcinoma, A431 (squamous epldermoid ¢arcinoma~, MCF-7 ~ma~mary carcinoma) and KB (nasoph ngeal carcinoma). HT1080, a fibrosarcoma derived cell line, which we had chosen s a source of mRNA for the cDNA cloning, contained relatively high le~els of TGF-~ mRNA. TGF-~ mRNA was not only present in cell lines derived from solid tumors of meso-, endo- and çctoblastic origin, but was also i detect~ble in tumor cell lines of hematopoletic orlgin, e.g. D~udi ~, (Burkitt lymphoma B-lymphoblast), Raji (Burkitt lymphoma B-lymphoblast), and Molt-4 (T-cell leukemia). The presence of TGF-~
mRNA is not restricted to tumor cells, since it is clearly detectable in placenta and peripheral blood lymphocyte (P~L) mRNA.
' Strikingly, the level of TGF-~ ~RNA is significantly elevated after j mitoganic stlmulation of PBLs. TGF-~ ~RNA ~as not detectable in I human li~er, yet was present in the HEP-G2 hepatoma cell line. In all cases, the TGF-~ ~RNA migrated as a species of an apparent length of 2.3 to 2.5 kbases. In ~ome cases a smaller mRNA species , of about 1.8 to 1.9 kb may be present, although ~his could be due to partial d~gr&dation of the mRNA.

~co~bi~lni_S~;pthesls of TGE-~
The plasmid used for recom~inant synthesis of TGF-~l was p ~ TE6. The followln~ prophetic metbod for maklng thi~ plasmd ls preferred over the more complsx method actually employed in its ~d ~ 35 constructio~.

:: :

p342E79 is di~ested with ~RI, blunted with ~ DNA
poly~erase I (Kl0n~w fragment) and the four dNTPs, digested wlth ~11 and Frag~ent 1 (c~ntalnln~ the Ampr gene of pBR322) rec~vered.

- p342E is ~imul~aneously digested wlth S~lI and ~i~dIII and the , HBsAg-encoding fragment i~ reco~ered as Frag~ent 2.

3 Finally, the SV40 geno~e i8 ~i~ultaneously dlgest.ed with HindIII ~nd inclI, and the 596 bp fragment c~n~aining ~he SY40 orig~n ~nd early promoter recovered as Fragment 3.
,.
~ Fragmen~s 1, 2 nnd 3 are li~ated in ~ three way ligatlon and j the li~ation mixture is transf~rmed into ~. ~Qli strain 2~4 (ATCC
J, 15 31446)~ The transformed culture i plated on ampicillin media pl~es and resistan~ colonies are selected. p342E-blunt ~as reco~ered ~r3m a transformant cDlony.

p3~2E bluDt i~ di~ested ~imultaneously wit~ dIII and ~_RI
and the larg0 vector fragment recovered. This fragment is ligated ~o a polylinker ha~in~ th~ follDwin~ sequence ~coRI
3 e e.ç 5 l ~ ~ I_S 5L~ ~ T _~ Q.~ A ~ T T C
HindIII ClaI Xb~I
and the llga~i~n mixture used t~ transform ~. coli ~CC 31446 BS
~i: described aboYe. pCVSV-HBs is recc~ered from an ampicillin-~: re~istant tr~nsforma~t.
~! ~ 30 s pCVSV-~B~ is digested with ~dlII a~d ~ç_RI simultane~u~ly J and the Yector frag~nt isolated (the 18 bp ~l~dIII-~_RI frag~ent ~ ~ill no~ appear i~ the gel due to its s~all size~.
.
,~ ~: 35 pgD-DHF~-Trunc ~European Paten~ No. 139,417B, a : . plasmid eont~inln~ DNA enooding ~he herpes simple~ gD protein, is " ~ s~ult2noDusly digested with ~I aDd ~dIlI and the approximately ~ 3 ~

760 bp fragment recovered which cont~ins DNA encoding the herpes simplex signal peptide and the coding region ~Eor the N-terminal part of the mature HSV-lgD protein. Plasmid pJ2.9 from European Patent Application 84.305909.8 can ~e used in the same fashion.
S
p~Cl (Fig. la) is digested with SmaI and BamHI, and the 480 bp fragment recovered. This fragment contains most of the sequence coding or preTGF-~l including the sequence coding for the N-terminu6 of mature TGF-~l through residue 314.
p~Cl is digested with ~mHI and EcoRI and the 270 bp fragment recovered. These two seperate digestl~ns of p~Cl aliquots were conducted because the Ba~HI-EcoRI p~Cl fragment contains a SmaI
site. The 270 bp fra~ment contains the sequence coding for the r~st of the TGF-~l Molecule and e~tends 20 bp beyond the stop codon.

The pCVSV-HBs vector fra,gment is ligated in a four way ligation with ~he foregoing 760, 270 and 480 bp fragments. The 1 20 resulting construction (pCVSVgD) thus contained a hybrid coding i sequence (Herpes simplex gD-l signal peptide and part of the gD-l envelope protein linked in ~'rame to the preTGF-~l precursor fragment) under the con~rol o~E ~he SV40 ~arly promoter. ~hls hybrid coding sequence ~s in turn ;Eollowed by the 3' untranslated sequence and the polyadenylation signal of the hepatitis surface tigen.

pCVSVgD ~s dl~ested with EcQRI, blunted with Klenow and the four dNTPs, ~nd thereafter digested with PstI. Two fragments are ~o obtained, with the fragmPnt includlng the hy~rid coding sequence ' and the SV40 promo~er (fragment A~ being recovered.

pCVSVgD ig digested with BamHI, blunted with Klenow and the four dNTPs, and thereafter dige~ted with ~I. Four fragMen~s are Dbtalned ~fter these dlgestions. The fragment containing the ., pBR322 origin and ~mpr gene (about 1900 bp) is recovered as Fragment B.
;~
Fr~gments A and B are ligated and the llgation mi~ture used to transform E. ~1i ATCC 31,446. Plasmld p~BTE6 is recovered from an , Ampr colony.

J Plasmid pMBTE6 was transfected into DHFR deficient CH0 cells55 i together with plasmid pFDll90. The latter plasmid encodes DHFR,thereby conferring methotrexate resistance on the transfected cells and allowing for selection of TGF-~ sxpressing transformants. Any DHFR- mammalian host cell is suitable for use. Al~ernatively, one can use any mammalian host cell, cvtransfDrm the host cell with a i plasmid encoding neomycin resi~tance, and ldentify transformants by their ability to grow in neomycln-containing medium.

J The transfected C~0 cells w~re selected by culturing in HGT-medium. The cells were allowed to grow to confluency in 15 cm diameter plateR. The cells thereafter were cultured in serum-free medium for 48 hours prior to harvest. The culture ~edium was decanted rom the plates and assayed in a ~oft agar assay for the presence of TGF-~ as described7~.
:1 ¦ 50 ml supe~natant mediu~ was lyophilized and dissolved in 700 ~1 4~N HCl-0.1 percent bovine serum albumin. 200 ~1 of this solution and o serial three-fold dilut~ons were assayed. The number of coloniss in soft agar with a diameter of ~89 ~m were coun~ed. The maximal response (plateau-value) obtsined in the presenoe of saturating levels of TGF-~l was abou~ 1500 per plate.
l~ 30 Less than 50 coloni~s were ~btained in the absence of TGF-~l. A
half ~aximal response was obtained with a 9-fold dilution of ~he ¦ ; sample derived from the cells transformed with & negative control plasmid. Galculations from the values obtalned by serial dilution ¦~ of the NBTE6 supernatant showed that the half maximal ~alue was 3~ 35 ob~ained at ~ 70-fold dilution.
~j ~
,,, : ~
i .

~, 326639 _3cZZ

The assays of serial dilutions show that cells transformed with MBTE6 ~nd pFDll synthesize about 8-10 times more TGF-~'l per ~ZZl of madium than do CH0 c211s transfected ~ith pFDll alone or with pFI'Zll together with a control plasmid (one which was similar to pMBTE6 except that the Herpes coding sequence is replaced by the sequence coding for the bacterial STII slgnal peptide), even prior to subcloning ~nd selection in MTX-containing medla, This ', additional ZZ*lount Df TGF-~' i8 hunZZan TGF-~
'6 19 ' Since biologlcally active TGF-~'l is found in the culture -` medium it is concluded ~hat CH0 cells cleave preTGF-~' in the same i, fashion as do human cells ln vivo to secrete mature native TGF~
i~ This conclusion is very much strengthened by the fact that the , 15 slope of the TGF-~'l concentration dilu~ion cur~e in the soft agar :J iS identical for both the endogenous natural TGF-~'l and the recombinant TGF-~'l, thus reflecting a similar if not identical affinity for ~he TGF-~' receptor, 3, E~MPLE 6 lg~ sgLs~ A F~ZZcoding TGF-Q3 , 1.5x106 plaqu0s from porcine ovarian cDNA~ llbrary was screened under low hybridizing conditions with the 32P-labelled E.
~ Ql~ ins~rt of ~'Cl (1050bp) in pH6.8 hybridizat~on buffer 3~ ~ 25 containi~g 5~SSC, 20~ formami de, 5xDenhardts, 0.1% Na-pyropho~phate, 0.05M NaP04, O.l~SDS and 50 ~g/ml salmon sper~ DNA
; Qt 42G~ overnight. Washes were in 2xSSC at 37C. Phage w~s purified from positively hybridizing plaques and their D~A inserts ; were sequenc2d.~ The ~pproximately 200 posit~'vely hybridiz~ng cDNA
~ insert'~ fell into three ~iasses: ~ P~rcine TGF-~'l (2 plaques), G-C
rich cDNAs which did not encode a TGF-~ polypeptide (6) and ~11.3, a gene~ ~fragment which contained DNA encoding porcine TGF-~3 ~; dowTZstrea~ from n~ature residue 10 ~2~3~

The labelled EcoRI insert of ~11.3 ~as used under high strin~ency conditions ~as above but 50% formamide, and with washes usin~ O.lxSSC at 42C) to rescreen lxlO~ plaques from a porcine ovarian cDNA ~ library. Of 20 positiYely hybrldizing plaques, one (~10) contained the entire porcine TGF-~3 sequence. The combined nucleotide and imputed amino acld sequences encoded by ~10 + 11.3 are shown in Figs. 4a-4c.

lx106 plaques from a hu~an ovarian cDNA library ~ere screened with labelled porcine cDNA. One positive plaque (~hu4) was identified. Ahu4 has the nucleotide and imputed amino acid sequence set forth in Figs. 4a-4c. Fi~. 5 is a compari30n of the amino acid sequences i~puted from the porcine and human TGF-~3 cDNAs. The ~andidate start codons for th~ porcine prscur~ors ~re boxed.

TGF-~3 is expressed in r~combinant cell cultu~e and recDvered therefrom in substantially the same way as TGF-~l, making allowances for departures in nuc:leotide and amino acid sequence as will be app~rent to those skilled in the art. Sinc~ the complete ¦ precursor for human TGF-~3 i9 not disclosed, in order to express human TGF-~3 it will be desirabl~ to reprobe genomic or cDNA
libraries for ~Na encoding the remaining ~-t~rminal precursor sequence, or ligate DNA encoding the available human s~quence (starting at: the codon for residue 3) wi~h the DNA encoding the porc~ne TGF-~3 precursor through regi~ue ~97 (numbered ~s shown in Fi~. 5), or prep~re DNA enc~ding a heterologou~ ~ammalian or Yiral signal fusion with DNA encoding mature human TGF-~3.

:

i~ :

` :

Claims (17)

1. A method for preparing mature TGF-.beta.3 comprising (a) constructing a vector which includes nucleic acid encoding preTGF-.beta.3 or mature TGF-.beta.3 having the amino acid sequence shown in Figure 5, or an allelic variant thereof, (b) transforming a host eukaryotic cell with the vector, (c) culturing the transformed cell, and (d) recovering from the culture mature TGF-.beta.3 having the amino acid sequence shown in Figure 5 or an allelic variant thereof.

2. The method of claim 1 wherein the eukaryotic cell is a Chinese hamster ovary cell line.

3. The method of claim 1 wherein the TGF-.beta.3 is preTGF-.beta.3.

4. The method of claim 3 wherein the pre TGF-.beta.3 is a fusion having a viral secretory signal sequence at its N-terminus and the sequence of mature TGF-.beta.3 at its C-terminus.

5. The method of claim 4 wherein the nucleic acid encoding the preTGF-.beta.3 is operably linked to a viral promoter.

6. The method of claim 4 wherein the nucleic acid encoding the preTGF-.beta.3 is operably linked to an inducible promoter.

7. The method of claim 1 wherein mature TGF-.beta.3 is recovered from the culture medium.

8. Isolated nucleic acid encoding preTGF-.beta.3 or mature TGF-.beta.3 having the amino acid sequence shown in Figure 5, or an allelic variant.

9. The nucleic acid of claim 8 that is DNA.

10. mRNA encoding preTGF-.beta.3 or mature TGF-.beta.3 having the amino acid sequence shown i Figure 5, or an allelic variant thereof, or a given species, which mRNA is cell-free and free of mRNA encoding other proteins of the given species.

11. The nucleic acid of claim 8 or 9 which is labelled with a detectable moiety.

12. The mRNA of claim 10 which is labelled with a detectable moiety.

13. A replicable vector comprising DNA that encodes preTGF-.beta.3 or mature TGF-.beta.3 having the amino acid sequence shown in Figure 5, or an allelic variant thereof.

14. The vector of claim 3 wherein the DNA is free of introns.

15. A host cell containing the vector of claim 13.

16. The cell of claim 15 which is a eukaryotic cell.

17. The cell of claim 15 which is a bacterial cell.