Procollagens

[0002] The present invention concerns novel molecules, in particular novel procollagen molecules, together with collagen molecules, fibrils and fibres comprising a non-natural combination of collagen α-chains, DNA encoding same, expression hosts transformed or transfected with same, transgenic animals and methods of producing a non-natural collagen.

[0003] Collagen (also known as processed procollagen molecule and triple helical processed procollagen monomeric molecule) (for general reviews see Kadler. K. 1995, Protein Profile, “Extracellular Matrix 1: fibril-forming coliagens”, 2: 491-619, Avad, S. et al., 1994, The Extracellular Matrix Facts Book, Academic Press, London, ISBN 0-12-068910-3 and references therein) is a major structural protein in animals where it occurs in the extracellular matrix (ECM) of connective tissues, mostly in the form of fibrils (also known as polymeric collagen). The collagen fibrils (polymeric collagen) are the major source of mechanical strength of connective tissues, providing a substratum for cell attachment and a scaffold for dynamic molecular interactions. The family of collagens comprises complex multidomain proteins comprising three collagen α-chains wound into a triple helix. At least twenty genetically-distinct collagen types have been described to date and they can be classified into subgroups on the basis of gene homology and function of the encoded protein. Fibril-forming collagens (types I, II, III, V and XI; see Table 1) are synthesized as soluble procollagens (also known as proα chains, procollagen α-chains and monomer chains) and comprises a C-propeptide, a Gly-X-Y repeat containing region (which in the case of monomer chains of fibril-forming collagens comprise an uninterrupted collagen α-chain) and an N-propeptide. The proα chains trimerise to form unprocessed procollagen molecules (also known as monomeric procollagen molecules and trimerised proα chains), assembling into fibrillar structures upon enzymic cleavage of their N- and C-terminal propeptide domains (the N- and C-propeptides) (see FIG. 1).

[0004] Although the genes encoding the proα chains are remarkably similar, relatively little is known about the processes which control the folding and trimerization of the proα chains (Dion, A. S. and Myers, J. C., 1987, J. Molec. Biol., 193: 127-143), and only a restricted range of collagens is formed. For example, skin fibroblasts synthesise co-incidentally the six highly homologous proα chains (proα1(I), proα1(III), proα1(V), proα2(I), proα2(V) and proα3(V)). Despite the great number of possible combinations of the six proα chains, only specific combinations of collagen chains occur—these are those resulting in types I, III and V collagen. Type I collagen exists as a heterotrimer and assembles with the stoichiometry of two proα1(I) chains and one proα2(I) chain ([proα1(I)]₂proα2(I)). Homotrimers of proα2(I) have not been detected and hence the inclusion of this chain in a trimer is dependent upon its association with proα1(I) chains. Type III collagens comprise a homotrimer ([proα1(III)]₃), and the constituent chains do not assemble with either of the Type I collagen proα chains. Type V collagen displays heterogeneity with regard to chain composition, forming both homo-([proα3(V)]₃) and hetero-trimers ([proα1(V)]₂proα2(V) and [proα1(V) proα2(V) proα3(V)]).

[0005] The C-propeptide is known to be implicated in the assembly of the monomer chains into trimerised proα chains (unprocessed procollagen) prior to cleavage of the N- and C-propeptides and formation of collagen in fibril-forming proα chains. The assembly of the three monomer chains into trimerised proα chains is initiated by association of the C-propeptides. This association can be divided into two stages: an initial recognition event between the proα chains which determines chain selection and then a registration event which leads to correct alignment and folding of the triple helix. Comparison (FIG. 2) of the amino acid sequences of the C-propeptides of proα1(I), proα2(I) and proα1(III) proα chains, which assemble to form collagen types I and III, demonstrates the striking level of sequence similarity between these proα chains yet, despite the homology, they invariably assemble and fold in a collagen type-specific manner.

[0006] It has now been found that the C-propeptides, and more particularly certain sequences within them, are not only necessary but are also sufficient to determine the type-specific assembly of the moieties to which they are attached because of the presence of these certain sequences, the C-propeptides are capable of autonomously directing the assembly of the attached moieties, which in particular may be an alien collagen α-chain. The present inventors have isolated and characterised a region of the C-propeptide which defines the chain selection event but which does not affect the subsequent folding. This has allowed the synthesis of novel proα chains which have formed novel trimerised proα chains and collagen. Now that the chain selection interactions between the proα chains can be controlled, a vast range of novel trimeric molecules, in particular collagens, may be synthesised at will using existing and novel proα chains and C-propeptides. These new molecules may possess selected biological and physical properties and have a wide range of uses. For example, novel collagens may be used in industries which use collagen either as a product or as part of a process. Such collagens and uses may include for example: novel gelatins for use in food, food processing and photography; novel finings for clearing yeast during the brewing process, novel gelatins for the food packaging industry; novel polymers for the manufacture of textiles; novel glues for use in construction, building and manufacturing; novel coatings for tablets; novel glues for use with the human or animal body: novel collagens for use as body implants; novel collagens and procollagens as adjuvants; novel collagens and procollagens as molecular carriers for drugs and pharmaceuticals; and as modulators of collagen fibril formation for use in, for example, wound healing and fibrosis.

[0007] According to the present invention there is provided a molecule comprising at least a first moiety having the activity of a procollagen C-propeptide and a second moiety selected from any one of the group of an alien collagen α-chain and non-collagen materials, the first moiety being attached to the second moiety.

[0008] The molecule may be able to bind to other similar molecules. It may trimerise with other similar molecules.

[0009] The first moiety will generally be attached to the C-terminal end of the second moiety, although intervening amino acid residues may be present.

[0010] The first moiety may comprise a proα chain C-propeptide or a partially modified form thereof or an analogue thereof, and when forming the C-terminal region of a proα chain, may allow the molecule to bind to other similar molecules. The C-propeptide region of a proα chain may be the C-terminal fragment resulting from C-proteinase cleavage of a proα chain. The C-proteinase may cleave between the residues G and D or A and D or an analogue thereof in the sequence FAPYYGD (residues 376-382 of SEQ ID NO: 2), YYRAD (residues 1-5 of SEQ ID NO: 14) or FYRAD (residues 284-288 of SEQ ID NO: 1) (FIG. 2) or an analogue thereof.

[0011] Modifications to molecules may include the addition, deletion or substitution of residues. Substitutions may be conservative substitutions. Modified molecules may have substantially the same properties and characteristics as the molecules form which they are derived. Modified molecules may be homologues of the molecules from which they are derived. They may for example have at least 40% homology, for example 50%, 60%, 70%, 80%, 90% or 95% homology.

[0012] The present inventors have isolated and identified (see “Experimental” section) a site—the recognition site—in the procollagen C-propeptide which contains a sequence which is necessary and sufficient to determine the type-specific assembly of the moieties to which it is attached. The recognition site is defined as being the part of the C-propeptide containing the sequence (the recognition sequence) which, in an alignment plot of the C-propeptide against other C-propeptides, corresponds to the sequence in the region between the junction points B and G (FIG. 2). Alignment plots may be done using the PILE-UP program on SEQNET at the Daresbury Laboratories, UK. An existing proα chain which has been substituted at the recognition sequence and as a result has different properties or characteristics is considered to be a molecule comprising a proα chain C-propeptide and an alien collagen α-chain since the C-propeptide is novel, all collagen α-chains therefore being alien to it.

[0013] As can be seen from FIG. 2, the recognition sequences contain a region of homologous amino acids. Substitution to the conserved residues (see “Experimental” section below) has not disrupted chain selection nor has it prevented the formation of a correctly aligned helix, and so it appears that the conserved residues are not involved in chain selection. Hence the recognition sequence, although comprising a continuous sequence of about 23 amino acids, may be considered to have the chain selection properties contained within a discontinuous variable sequence. For example in the recognition sequence of alpha 1(III) (SEQ ID NO: 6) the discontinuous variable sequence may be considered to comprise residues 1-12 and 21-23.

[0014] The C-propeptide and/or the recognition sequence may be that of a fibrillar proα chain. More generally, the C-propeptide may be an existing C-propeptide, for example a C-propeptide found in nature, or it may be a partially modified form of or an analogue (i.e. possessing substantially the same properties and characteristics but having a different sequence) of an existing proα chain C-propeptide, or it may comprise a novel C-propeptide (i.e. a C-propeptide having significantly different properties or characteristics to other C-propeptides) and may for example have different binding kinetics or α-chain selection properties.

[0015] The existing C-propeptide may be selected from any one of the group of the proα1(I), proα2(I), proα1(II), proα1(III), proα1(V), proα2(V), proα3(V), proα1(XI), proα2(XI), and proα3(XI) proα chain C-propeptides or a partially modified form thereof or an analogue thereof. Partially modified forms of procollagen C-propeptides include the recognition sequences of C-propeptides, for example those identified in FIG. 3 of the accompanying drawings in relation to the C-propeptides from which they are derived. In some embodiments of the invention, such modified forms may be the only, or substantially the only, elements derived from a C-propeptide, in other words, no other C-propeptide-derived sequences need be present. However, this will not always be the case, as the invention also encompasses the presence of other parts of the C-propeptide including, of course, the balance of it.

[0016] The C-propeptide may comprise an existing C-propeptide or a partially modified form thereof or an analogue thereof substituted at the recognition site. The C-propeptide may be substituted at the recognition site with the recognition sequence of an existing C-propeptide, for example that of a different C-propeptide. It may for example be substituted at the recognition site with the recognition sequence of the C-propeptide of any one of the group of proα1(III), proα1(I), proα2(I), proα1(II), proα1(V), proα2(V), proα1(XI), proα2(XI) and proα3(XI) proα chains. It may be substituted at the recognition site with a recognition sequence having the sequence of any one of the group of SEQ ID NOS: 6-13. The recognition sequence of a C-propeptide which has been modified for example by addition, deletion or substitution of amino acid residues yet which has substantially the same properties and/or characteristics is considered to be essentially that of an existing C-propeptide. The recognition sequence may generally be at least 40% homologous, or even at least 50, 60, 70, 80, 90 or 95% homologous to the sequence from which it was derived.

[0017] Such a substitution at the recognition site may significantly affect the properties or characteristics of the C-propeptide.

[0018] Alternatively, the recognition sequence may be novel. Such a novel recognition sequence may for example give the first moiety novel binding kinetics or specificity for a novel first moiety or a novel set of first moieties.

[0019] The second moiety is a molecular component which may be anything bound to the first moiety. This may include, for example, alien collagen α-chain molecules, or other proteins or fragments of proteins, such as antibodies or antigen binding fragments thereof, or combinations thereof. Proteins constituting or contributing to the second moiety may be glycosylated or otherwise post-translationally modified. By “alien collagen α-chain” is meant a collagen α-chain which does not form part of a proα chain with the C-propeptide in nature; collagen α-chains comprise a triple helical forming domain, and an N-propeptide may also be present. Other collagen α-chains from the same species, as well as those from different species, may be used. Included as collagen α-chains which do not form part of a proα chain with the C-propeptide in nature are partially modified forms and analogues of existing collagen α-chains which form part of a proα chain with the C-propeptide in nature and which do not significantly affect the relevant properties or characteristics of the procollagen molecule, such as binding specificity. Partially modified forms and analogues of collagen α-chains may, for example, have additions, deletions or substitutions which do not significantly affect the relevant properties or characteristics of the C-propeptide or collagen α-chain.

[0020] By means of the invention, therefore, novel collagens may be produced. Such novel collagens have combinations of α-chains which are not seen in nature because of the assembly-directing effect of the natural C-propeptides. The invention allows the protein engineer to construct novel collagens having a non-natural combination of α-chains. The invention therefore extends to a procollagen molecule comprising a non-natural combination of α-chains. Non-natural pro-collagen homotrimers and heterotrimers, including all the possible trimers not mentioned in Table 1, are within the scope of the invention.

[0021] The second moiety may comprise at least a collagen α-chain. A collagen α-chain may be selected from any one of the group of proα1(I) chain, proα2(I) chain, proα1(II) chain, proα1(III) chain, proα1(V) chain, proα2(V) chain, proα3(V) chain, proα (XI) chain, proα2(XI) chain, and proα3(XI) chain collagen α-chains.

[0022] The second moiety may also comprise a proα chain N-propeptide. An N-propeptide may be selected from any one of the group of the proα1(I), proα2(I), proα1(II), proα1(III), proα1(V), proα2(V), proα3(V), proα1(XI), proα2(XI), and proα3(XI) proαchain N-propeptides.

[0023] The second moiety may comprise a collagen Δ-chain and N-propeptide which are naturally associated (for example those of the proα2(I) chain), or it may comprise a non-natural combination of collagen α-chain and N-propeptide. Depending upon the host organism in which it may be desired to express molecules of the invention, the N-terminal propeptide may be replaced or adapted to facilitate secretion or other handling or processing in the expression system.

[0024] The molecule may comprises a first moiety having the activity of the proα1(III) C-propeptide attached to a second moiety comprising the collagen α-chain and N-propeptide of the proα2(I) chain. The molecule may have the sequence of SEQ ID NO: 4.

[0025] In the natural formation of a collagen molecule in vivo, the N- and C-propeptides are cleaved off the procollagen molecule to yield a collagen molecule during the formation of polymeric collagen. Consequently, the invention includes within its scope a collagen molecule comprising a non-natural combination of α-chains. Non-natural collagen homotrimers and heterotrimers, including all the possible collagen trimers not mentioned in Table 1, are within the scope of the invention. (If for any reason it is desired to have a non-natural collagen molecule with a C-propeptide but not an N-propeptide, or vice versa, the enzymes responsible for processing in the chosen expression system may be manipulated or selected accordingly or the sequence of the molecule modified to make it susceptible to enzymatic processing as appropriate.)

[0026] Collagen molecules naturally self-assemble into collagen fibrils, which in turn aggregate to form a collagen fibre. Collagen fibrils and collagen fibres comprising collagen molecules as described above are therefore also contemplated by the invention.

[0027] Molecules of the first aspect of the invention may be prepared by any convenient method, including peptide ligation and complete synthesis. It is preferred however, that the molecules be prepared by expression from a recombinant DNA system. For this purpose, and according to a second aspect of the invention, there is provided a DNA molecule, which may be in recombinant or isolated form, encoding a molecule as described above (particularly a non-natural procollagen α-chain).

[0028] Recombinant DNA in accordance with the invention may be in the form of a vector. The vector may for example be a plasmid, cosmid or phage. Vectors will frequently include one or more selectable markers to enable selection of cells transfected (or transformed: the terms are used interchangeably in this specification) with them and. preferably, to enable selection of cells harbouring vectors incorporating heterologous DNA. Appropriate start and stop signals may be present. The vector may be an expression vector having regulatory sequences to drive expression. Vectors not including regulatory sequences are useful as cloning vectors; and, of course, expression vectors may also be useful as cloning vectors.

[0029] Cloning vectors can be introduced into E. colior another suitable host which facilitate their manipulation. According to another aspect of the invention, there is therefore provided a host cell transfected or transformed with DNA as described above. Such host cells may be prokaryotic or eukaryotic. Eukaryotic hosts may include yeasts, insect and mammalian cell lines. Expression hosts may be stably transformed. Unstable and cell-free expression systems may be used in appropriate circumstances, but it is unlikely that they will be favoured, at the present state of technology, for bulk production.

[0030] DNA of the invention may also be in the form of a transgene construct designed for expression in a transgenic plant or, preferably, animal. In principle, the invention is applicable to all animals, including birds such as domestic fowl, amphibian species and fish species. The protein may be harvested from body fluids or other body products (such as eggs, where appropriate). In practice, it will be to (non-human) mammals, particularly placental mammals, that the greatest commercially useful applicability is presently envisaged, as expression in the mammary gland, with subsequent optional recovery of the expression product from the milk, is a proven and preferred technology. It is with ungulates, particularly economically important ungulates such as cattle, sheep, goats, water buffalo, camels and pigs that the invention is likely to be most useful. The generation and usefulness of such mammalian transgenic mammary expression systems is both generally, and in certain instances specifically, disclosed in WO-A-8800239 and WO-9005188. The β-lactoglobulin promoter is especially preferred for use in transgenic mammary expression systems. WO-A-9416570 purports to disclose the production of human recombinant collagen in the milk of transgenic animals but contains no experimental details of such production having taken place.

[0031] Expression hosts, particularly transgenic animals, may contain other exogenous DNA to facilitate the expression, assembly, secretion and other aspects of the biosynthesis of molecules of the invention. For example, expression hosts may co-express prolyl 4-hydroxylase, which is a post-translational enzyme important in the natural biosynthesis of procollagens, as disclosed in WO-9307889.

[0032] DNA, particularly cDNA, encoding natural procollagen chains is known and available in the art. For example, WO-A-9307889, WO-A-9416570 and the references cited in both of them give details. Such DNA forms a convenient starting point for DNA of the present invention, which may be prepared by recombinant techniques from it. While in general terms DNA encoding a C-propeptide (or a minimal essential region from it) may simply be ligated to DNA encoding an alien collagen triple helical domain (usually attached to DNA encoding the corresponding N-propeptide), in practice it is useful to use PCR-based techniques to effect the precise ligation. For example, PCR products flanking the junction region between the C-propeptide and the triple helical domain may be prepared and combined; an overlap extension reaction can then be carried out to yield a PCR product which is a hybrid between DNA encoding the C-propeptide of one procollagen chain and DNA encoding the triple helical domain (and the N-propeptide, usually) of another procollagen chain.

[0033] The invention is in principle capable of accommodating the use of synthetic DNA sequences, cDNAs, full genomic sequences and “minigenes”, which is to say partial genomic sequences containing some, but not all, of the introns present in the full length gene. Because of the large number of introns present in collagen genes in general, though, experimental practicalities will usually favour the use of cDNAs or, in some circumstances, minigenes.

[0034] DNA in accordance with the invention can in principle be prepared by any convenient method involving coupling together successive nucleotides, and/or ligating oligo- and/or poly-nucleotides, including in vitro processes, but recombinant DNA technology forms the method of choice.

[0035] Molecules of the invention may be useful in a method of treatment or diagnosis of the human or animal body. The invention therefore extends to molecules as described above for use in medicine.

[0036] The molecule may be for use as an adhesive or implant. It may be for use in promoting the healing of wounds or fibrotic disorders with reduced scarring. It may be for use in promoting the healing of chronic wounds. By “wounds or fibrotic disorders” is meant any condition which may result in the formation of scar tissue. In particular, this includes the healing of skin wounds, the repair of tendon damage, the healing of crush injuries, the healing of central nervous system (CNS) injuries, conditions which result in the formation of scar tissue in the CNS, scar tissue formation resulting from strokes, and tissue adhesion, for example, as a result of injury or surgery (this may apply to e.g. tendon healing and abdominal strictures and adhesions). Examples of fibrotic disorders include pulmonary fibrosis, glomerulonephritis, cirrhosis of the liver, and proliferative vitreoretinopathy.

[0037] For example in the inhibition of fibrosis a novel collagen molecule or proαchain may be applied to a site of wounding or fibrosis, the novel collagen (or proα chain) inhibiting collagen fibril formation and thus fibrosis. The novel collagen or proα chain may for example have a shortened α-chain.

[0038] DNA of the invention may be useful, in appropriate constructs, in a method of gene therapy. It may be for use in the treatment of Osteogenesis Imperfecta (OI), Ehlers-Danlos Syndrome (EDS), Stickler Syndrome, Spondvloepiphyseal dysplasia, Hypochondrogenesis or Aortic Aneunrsms. Mutations within collagen genes are the cause of most forms of OI, some forms of EDS and of some chondrodysplasias. In most cases the devastating effects of the disease are due to substitutions of glycine within the triple helical domain—the Gly-X-Y repeat containing regionαfor amino acids with bulkier side chains.

[0039] This results in triple helix folding being prevented or delayed with the consequence that there is a drastic reduction in the secretion of trimerised proα chains. The malfolded proteins may be retained within the cell, probably within the ER (endoplasmic reticulum), where they are degraded. As the folding of the C-propeptide is not affected by these mutations within the triple helical domain. C-propeptides from wild-type as well as mutant chains associate and may be retained within the cell. The retention and degradation of wild-type chains due to their interaction with mutant chains amplifies the effect of the mutation and has been termed “procollagen suicide”. The massive loss of protein due to this phenomenon may explain the dominant lethal effects of such mutations. By engineering proα chains having altered chain selectivity, proα chains maybe produced which do not associate with the mutant chains, and will therefore fold and be secreted normally. Such engineered proα chains may contain the wild-type collagen α-chain, thereby making up for the deficit caused by the mutant collagen α-chain. Expressed protein may in some circumstances also be useful in the treatment of diseases and conditions which could be addressed at a more fundamental level by gene therapy.

[0040] The invention may also be useful in photography, brewing, foodstuffs, textiles or adhesives.

[0041] Also provided according to the present invention is a method of treatment or diagnosis of the human or animal body comprising the use of a molecule according to the present invention.

[0042] The invention will be further apparent from the examples, which comprise the following Figures and description of experiments, of which:

[0043] FIG. 1 shows the initial stages in the intra-cellular folding, assembly and modification of procollagen. As can be seen, co-translational translocation and signal peptide cleavage occurs at stage number 1. Intra-molecular disulphide bond formation then takes place as well as N-linked glycosylation, proline isomerisation and proline hydroxylation at stage 2. There then follows at stage 3 type-specific assembly of the proα chains by trimerisation and inter-molecular disulphide bond formation. Finally, at stage 4, triple helix formation proceeds in a carboxy-to-amino-direction to give trimerised proαchains;

[0044] FIG. 2 shows an alignment plot made using the PILE-UP program on SEQNET at the Daresbury Laboratories, ULK using default settings, of the C-propeptides of proα chains of type I and type III procollagen. Alpha 1(I) is SEQ ID NO: 14; alpha 2(I) is residues number 284-534 of SEQ ID NO: 1; and alpha 1(III) is residues number 379-626 of SEQ ID NO: 2. C-proteinase cleavage sites (marked CP) are between A and D (alpha 1(I)), A and D (alpha 2(I)) and G and D (alpha 1(III)). Junction points A, F, B, C and G are as shown. Numbers indicate conserved cysteine residues. # indicates identical amino acids and˜indicates amino acids with conserved side chains;

[0045] FIG. 3 shows recognition sequences for proα1(I), proα2(I), proα1(II), proα1(III), proα1(V), proα2(V), proα1(XI) and proα2(XI) proα chains having SEQ ID NOs: 7, 8, 9, 6, 10, 11, 12, and 13 respectively and which were identified by alignment plots of C-propeptides against other C-propeptides (specifically, those of FIG. 2) as corresponding to the sequences in the regions between junction points B and G of FIG. 2;

[0046] FIG. 4 shows an SDS-PAGE gel of translated procoliagen constructs. Lanes are as follows: 1—molecular weight markers; 2 and 6—proα1(III)Δ1; 3 and 7—proα2(I)Δ1; 4 and 8—proα2(I):(III)CP; 5 and 9—proα (III):(I)CP;

[0047] FIG. 5 shows an SDS-PAGE gel of translated procollagen constructs. Lanes are as follows: 1—molecular weight markers; 2 and 7—A-join; 3 and 8—F-join; 4 and 9—B-join; 5 and 10—C-join; 6 and 11—recip-C-join;

[0048] FIG. 6 shows translated procollagens in the presence (Lanes 3, 5 and 7) and absence (Lanes 2, 4 and 6) of α,α′-dipyridyl. Lanes are as follows: 2 and 3—proα2(I):(III)CP; 4 and 5—BGR; 6 and 7—proα1(III):(I)CP;

[0049] FIG. 7 shows an SDS-PAGE gel of translated procollagen constructs under reducing (lanes 1-3) and non-reducing (lanes 4-6) conditions. Lanes are as follows: l and 4—BGR^S-C; 2 and 5—BGR; 3 and 6—BGR^L-M; and

[0050] FIG. 8 shows an SDS-PAGE gel of translated procollagen constructs. Lanes are as follows: 1—molecular weight markers; 2—BGR^S-C; 3—BGR; 4—BGR^L-M.

[0051] A cDNA clone coding for a truncated proα2(I) chain (designated proα2(I)Δ1; SEQ ID NO: 1; nucleic acid coding sequence—SEQ ID NO: 19) was constructed from two partial cDNA clones, pHf1131 and pHp2010 (Kuivaniemi, H. et al., 1988, Biochem. J., 252: 633-640) which were sub-cloned into the EcoRI site of pBluescript SK⁺. A 3.4 kb fragment PstI fragment containing the vector and the 5′-terminal 0.5 kb of the gene was isolated from pHp2010 and ligated to a 1.4 kb PstI fragment derived from pHf1131 encoding the 3′ terminus. The resultant recombinant, proα2(I)Δ1, has a 2.2 kb deletion in the coding sequence (Lees and Bulleid, 1994, J. Biol. Chem., 269: 24354-24360).

[0052] This construct was analysed using a semi-permeabilised (SP) HT1080 cell system as described by Wilson et al. (1995, Biochem J. 307: 679-687). Semi-permeabilised cells were prepared from HT1080 cells. Confluent HT1080 cells from a 75 cm²flask were rinsed once with PBS (phosphate buffered saline), then incubated with 2 ml of PBS containing 2.5 mg/ml trypsin for 3 minutes at room temperature. The flask was transferred to ice where 8 ml of ice-cold KHM (110 mM KOAc. 20 mM Hepes, pH 7.2, 2 mM MgOAc) was added containing 100 μg/ml soyabean trypsin inhibitor and the cells released from the plate. Cells were pelleted at 12,000 rpm for 3 minutes and resuspended in 6 ml of KHM containing 40 μg/ml digitonin (diluted from a 40 mg/ml stock in DMSO (dimethyl sulfoxide)) and incubated on ice for 5 minutes. To terminate permeabilisation 8 ml of KHM was added and cells were pelleted and resuspended in 50 mM Hepes, pH 7.2, 90 mM KOAc. After 10 minutes the cells were pelleted and resuspended in 100 μl of KHM (approximately 2×10⁶cells). Endogenous mRNA was removed by adding CaCl₂to 1 mM and Staphylococcal nuclease to 10 μg/ml and incubating at room temperature for 12 minutes. The reaction was terminated by the addition of EGTA to 4 mM, and pelleting the cells. Semi-permeabilised cells were resuspended in 100 μl of KHM. RNA was translated using a rabbit reticulocyte lysate (FlexiLysate, Promega, Southampton, U.K.) for 1 hour at 30° C. The translation reaction (25 μl) contained 16 μl reticulocyte lysate, 1 μl 1 mM amino acids (minus methionine), 15 μCi L-[³⁵S] methionine, 1 μl transcribed RNA and semi-permeabilised cells (approx. 1×10⁵). After translation, N-ethylmaleimide was added to a final concentration of 20 mM. The formation of disuiphide bonds was verified by comparative gel electrophoresis on 7.5% polyacrylamide gel of translation products run under reducing and non-reducing conditions.

[0053] When analysed using this cell-free system the translation product from proα2(I)Δ1 mRNA did not self-associate to form homotrimers indicating that it does not contain the information necessary for the initial recognition event (FIG. 4, lanes 3 and 7).

[0054] A cDNA clone coding for a truncated proα1(III) chain (designated proα1(III)Δ1; SEQ ID NO: 2; nucleic acid coding sequence—SEQ ID NO: 20) was constructed from a full-length type III procollagen cDNA which was constructed from two partial cDNAs, pS413 and pS31 (Ala-Kokko, et al., 1989, Biochem. J., 260: 509-516). Each cDNA was subcloned into the EcoRI site of pBluescript SK⁻. A 4.7 kb Sal I (restriction enzyme) fragment containing the vector and the 5′ terminal 1.8 kb was isolated from pS413 and ligated to a 3.6 kb Sal I fragment derived from pS31 to produce proα1(III). An internal 2.5 kb XhoI fragment was excised from proα1(III) and the parental plasmid re-ligated to create proα1(III)Δ1 (Lees and Bulleid, 1994, J. Biol. Chem., 269: 24354-24360).

[0055] The translation product from proα1(III)Δ1 mRNA was able to assemble to form a homotrimer as judged by its ability to form inter-chain disulphide bonded dimers and trimers when translated in a semi-permeabilised cell-free translation system. This demonstrated that it contained the information required for self-assembly (FIG. 4, lanes 2 and 6).

[0056] Hybrid cDNA clones were prepared which contained sequences derived from proα1(III)Δ1 and proα2(I)Δ1. The C-proteinase cleavage site of proα1(III)Δ1 was, for these experiments, taken to be between Ala 377 and Pro 378 of SEQ ID NO: 2, instead of between Gly 381 and Asp 382 of SEQ ID NO: 2 as shown in FIG. 2. In the first of these constructs the coding sequence for the C-propeptide from the proα1(III)Δ1 chain was replaced by that for the C-propeptide from the proα2(I) chain, the resulting chimera being designated proα1(III):(I)CP (SEQ ID NO: 3). This construct failed to self-associate when translated in the cell-free system (FIG. 4, lanes 5 and 9). A reciprocal construct was made where the C-propeptide from the proα2(I)Δ1 was replaced with the C-propeptide from the proα1(III) chain with the resulting chimera designated proα2(I):(III)CP (SEQ ID NO: 4).

[0057] This construct was able to self associate to form dimers and homotrimers (FIG. 4, lanes 4 and 8), demonstrating directly for the first time that all the information required for selective association resides within the C-propeptide. The construct proα1(III):(I)CP was prepared as described below. Other constructs were produced using the same approach and published sequences.

[0058] The hybrid cDNA clones were prepared using a PCR-based approach. For the construction of proα1(III):(I)CP, a PCR product was prepared from proα1(III)Δ1 with primers, one of which (SEQ ID NO: 15; JL-35) hybridised within the triple helical domain whilst the other (SEQ ID NO: 16; JL-32) hybridised with 21 nucleotides upstream from the junction point at the C-propeptide. This primer also contained an overlap of 21 nucleotides which were complimentary to the first 21 nucleotides of the C-propeptide of proα2(I)Δ1. This gave a 0.25 Kb PCR product.

[0059] A second PCR product was prepared from proα2(I)Δ1 with primers, one of which (SEQ ID NO: 17; JL-3IKpn) hybridised downstream from the stop codon for translation within the 3′-non-translated region. This primer also contained a KpnI site. The other primer (SEQ ID NO: 18; JL-36) hybridised with the first 18 nucleotides of the C-propeptide of proα2(I)Δ1. This gave a 0.85 Kb PCR product.

[0060] The two PCR products were combined and a third PCR (an overlap extension) was carried out with primers JL-35 (SEQ ID NO: 15) and JL-31Kpn (SEQ ID NO: 17) to yield a 1.1 Kb product. This was cut with XhoI and KpnI and subcloned into XhoI and KpnI cut proα1(III)Δ1 to yield proα1(III):(I)CP.

[0061] A variety of hybrid constructs were then prepared in which parts of the proα2(I)Δ1 C-propeptide sequence was replaced with the corresponding region from the proα1(III) C-propeptide. The various regions are outlined in FIG. 2 with the junction points designated as A, F, B, C, and G. So for example the A-join molecule contains all of the proα2(I)Δ1 sequence up to but not including the A site (i.e. . . . DY) with all of the sequence carboxy-to this site (i.e. ET . . . ) being derived from the corresponding region from the C-propeptide of the proα1(III) chain. Pros 1(III) and proα2(I) C-propeptides differ in their complement of cysteine residues (and hence in their ability to form disulphide bonds), with proα2(I) lacking the Cys 2 residue (FIG. 2), instead having a serine residue. In order to ease analysis under non-reducing conditions (see below) the F, B and C constructs contained a serine to cysteine mutation at the Cys 2 site of the proα2(I) chain. To ensure that this mutation played no role in chain selection, a similarly mutated construct (proα2(I):(III) BGR^S-C; also referred to as BGR^S-C; see below) was back-mutated. The back-mutated construct (i.e. proα2(I):(III) BGR^S-C—also referred to as BGR) had the same chain selectivity as its parent molecule (proα2(I):(III) BGR^S-C) (see below).

[0062] The A-join, F-join and B-join chimeras all assembled to form homotrimers when translated in the cell-free system (FIG. 5, lanes 7, 8 and 9). However, the C-join molecule did not assemble (FIG. 5, lane 10) suggesting that the recognition site for assembly was contained within the sequence carboxy-terminal to the B-site and amino-terminal to the C-site. The possibility that the lack of assembly of the C-join molecule was due to this site being within the recognition sequence for assembly could not be ruled out. Evidence that the recognition site had been disrupted was obtained when the reciprocal construct was made. This construct contained the proα1(III)Δ1 chain up to the C-site with the rest of the C-propeptide being derived from the proα2(I) C-propeptide. No assembly occurred from this construct (recip-C-join) illustrating that the recognition site had been disrupted (FIG. 5, lane 11).

[0063] The next construct made contained all of the proα2(I)Δ1 sequence apart from a short stretch of 23 amino acids between the B-site and the G-site (SEQ ID NO: 6) which were replaced with the corresponding region from the C-propeptide of proα1(III). This construct (designated BGR; SEQ ID NO: 5) was altered by site directed mutagenesis of cysteine for serine at the Cys 2 site of the proα2(I) part of the molecule (i.e. cysteine was substituted for the serine 335 residue of SEQ ID NO: 5). The resultant molecule (designated proα2(I):(III) BGR^S-Cwas shown to assemble to form inter-chain disulphide bonded homotrimers when translated in the cell-free system (FIG. 7, lane 4), demonstrating that this short stretch of 23 amino acids contains all the information to drive homotrimer formation.

[0064] To verify that the serine to cysteine mutation did not affect chain selection, a back mutation was made (i.e. to give proα2(I):(III) BGR) and homotrimer formation analysed. As expected, no inter-chain disulphide bonded trimers were detected (FIG. 7, lane 5) as this molecule does not contain the cysteine residue required for inter-chain disulphide bond formation.

[0065] To determine whether a stable triple helix was formed after translation of the chimeric procollagens, a simple protease protection assay was carried out. This involved treating the translation products with a combination of proteolytic enzymes (trypsin, chymotrypsin and pepsin). Isolated SP-cells following translation were resuspended in 0.5 M acetic acid in 1% (v/v) Triton X-100 and incubated with pepsin (100 mg/ml) for 2 hours at 20° C. Digestions were stopped by neutralisation with 1 M Tris base and proteins precipitated with ethanol at a final concentration of 27% (v/v). Precipitated protein from pepsin digests were resuspended in 50 mM Tris-HCl, pH 7.4 containing 0.15 M NaCl, 10 mM EDTA (ethylenediaminetetra-acetic acid) and 1% Triton X-100. Chymotrypsin and trypsin were added to a final concentration of 250 μg/ml and 100 μg/ml respectively and samples incubated at room temperature for 2 minutes. The digestion was stopped by the addition of soyabean trypsin inhibitor to a final concentration of 500 μg/ml and 5 volumes of boiling SDS-PAGE (sodium dodecyl sulphate-polyacrylamide gel electrophoresis) sample buffer and boiling the samples for 3 minutes. The results are shown in FIGS. 6 and 8. The formation of a stable triple helix is characterised by the appearance of a protease resistant fragment (corresponding to the triple helical domain) after digestion. Only the products of translation of the proα2(I):(III)CP (FIG. 6, lane 2) and the BGR constructs (FIG. 6, lane 4; FIG. 8, lanes 2 and 3) generated a protease resistant fragment which were only formed when α,α′-dipyridyl (an inhibitor of prolyl 4-hydroxylase) was not present during the translation (FIG. 6). As proline hydroxylation is necessary for formation of a thermally stable triple helix, this demonstrates that a correctly folded triple helix was formed with these constructs.

[0066] This also demonstrates that although BGR was not able to form trimers stabilised by inter-chain disulphide bonds, it was able to trimerise to form a correctly aligned triple helix.

[0067] Analysis of the B-G motif from the proα1(III) and proα2(I) chains (FIG. 3) shows that of the residues in the recognition sequences (FIG. 3; SEQ ID NOs: 6 and 8 respectively), residues 13-20 are identical with the exception of residue 17—Leu (L) in proα1(III) and Met (M) in proα2(I). In order to determine the role played by these residues in the chain selection process, site directed mutagenesis was used to substitute Met for Leu in the proα1(III) recognition sequence in the proα2(I):(III) BGR^S-Cconstruct (designated proα2(I):(III) BGR^L-M—also referred to as BGR^L-M), i.e. residue Leu 425 of SEQ ID NO: 5 was substituted for Met, and residue Ser 335 was substituted for Cys.

[0068] Chain assembly of proα2(I):(III) BGR^L-Mwas performed as described above and the electrophoretic mobility of the chains analysed. Under non-reducing conditions this construct formed inter-chain disulphide bonds (FIG. 7, lane 6), and formed protease-resistant triple helical domains (FIG. 8, lane 4). The substitution of Leu for Met did not, therefore, disrupt the process of chain selection nor did it prevent the formation of a correctly aligned helix.