Elastomeric structures which are able to reversibly contract and extend are well known in nature, and are found for example in muscles, arteries, lungs, ligaments and skin. The contraction and extension results from alteration of the secondary structure of the molecules which make up the elastomeric structures, for example elastin. Considerable work has been performed to understand the nature of the change and the peptide sequence which undergoes the transition. Elastin for example, is composed of repeating hydrophobic monomers, typically (VPGVG)n, where n is as large as 11-13 in pig and chicken.

Synthetic elastin polymers have been studied extensively by Urry, and the results are reviewed by Urry in "Methods in Enzymology" (1982), Eds. Sidney Colowick, Nathan Kaplan, Academic Press, p673-716; Urry (1992), Prog. Biophys. Molec. Biol., 57, p23-57; and Urry, Angew. Chem. Int. Ed. Engl. (1993), 32, p819-841.

Synthetic elastin polymers, formed from the pentapeptide of elastin are soluble at temperatures below 24"C (when n=150), but precipitate at higher temperatures to form a dense viscoelastic phase called a coacervate. The process of coacervation is readily reversible. The binding of water molecules to hydrophobic parts of the protein is an exothermic process. Therefore, an increase in heat above a certain transition temperature leads to release of bound water molecules and its intra- and inter-molecular hydrophobic interactions increase.

In the case of elastin, heating disrupts the elongated, less ordered hydrophobic hydrated structure, and the polymer contracts intramolecularly. Contraction results from initially folding into repetitive type II turns, forming a P-spiral with approximately three VPGVG pentamers per turn, and then by forming an intermolecular twisted filament. This increase in order of the polypeptide with an increase in temperature is referred to as an inverse temperature transition and occurs at a specific temperature for specific molecules.

This temperature may however vary if other environmental parameters are altered.

The transition temperature, Tm, may be manipulated by altering the hydrophobicity of an elastomeric polymer by altering the polymer itself or its environment (Urry et al., 1991, J. Am. Chem. Soc., 113, p4346-4348; Luan et al., 1990, Biopolymers, 29, p1699-1706). Thus, for example, Trn may be varied by phosphorylation (Pattanaik et al., 1991, Biochem. Biophys. Res. Comm., 178, p539- 545), electrochemical or chemical reduction of a prosthetic group (Urry et al., 1992, Biochem. Biophys.

Res. Comm., 188, p611-617; Urry et al., 1995, Biochem.

Biophys. Res. Comm., 210, p1031-1039; Urry et al., 1994, Biochem. Biophys. Res. Comm., 204, p230-237), or photochemical reaction, by altering the pressure (Urry et al., 1993, Chemical Physics Letters, 201, p336-340), pH (Urry et al., 1988, Proc. Natl. Acad. Sci., 85, p3407-3411), concentration of the polymer or composition of the polymer (Urry et al., 1992, Biopolymers, 32, p1243-1250; Urry et al., 1986, Biochem. Biophys. Res.

Comm., 141, p749-755), or by adding salt (Luan et al., 1991, Biopolymers, 31, p465-475) or organic solutes, as generally described in Urry, 1992, Prog. Biophys. Molec.

Biol., 57, p23-57 and Urry, 1993, Angew. Chem. Int. Ed.

Engl., 32, p819-841.

As a result of the alteration of Tm, the temperature of a particular system may be maintained at a constant

and transition between the different states triggered by alteration of one of the above mentioned parameters such that the transition temperature is below (or above) the temperature of the system and contraction (or extension) occurs. Thus for example, a change in pH or pressure may be used to trigger or induce contraction or elongation of an elastomeric peptide.

Non-peptide polymers have also been identified which exhibit similar properties insofar as environmental parameters, including temperature, may induce a phase transition to form insoluble aggregates above the transition temperature. This also results from hydrophobic interactions (Feil et al., 1993, Macromolecules, 26, p2496-2500).

In particular, poly(N-isopropylacrylamide) (polyNIPAAm) polymers have been used which have various applications, including attachment to polypeptides to produce polypeptide-polymer conjugates. Such conjugates have been proposed for the separation of the polypeptides from experimental systems using the phase transition and thus resultant insolubility of the collapsed polymer attached to the polypeptide (e.g. in purification) or for drug delivery (Ding et al., 1996, Bioconjugate Chem., 7, pl21-125; Matsukata et al., 1996, Bioconjugate Chem., 7, p96-101; Chen & Hoffman, 1995, Macromol. Chem. Phys., 1995, 196, pl251-1259; Chen & Hoffman, 1995, Nature, 373, p49-52; Chen & Hoffman, 1993, Bioconjugate Chem., 4, p509-514 and Park & Hoffman, 1993, J. Biomater. Sci. Polymer Edn., 4(5), p493-504). Stayton et al., 1995, Nature, 378, p472-474 have conjugated poly(NIPAAm) to streptavidin which when contracted prevents biotin binding to the active site by collapse of the polymer onto the active site preventing access by the ligand.

Elastomeric peptides reduce in length by more than 50% on contraction. Elastin polymers are capable of lifting a weight of more than 1000 times their dry

weight, performing work and producing motion. The energy involved for transition within one monomer, VPGVG, is reported to be around 2.Okcal/mol (Chang & Urry, 1988, Chemical Physics Letter, 147, p395-400).

The ability of elastomeric peptides to transform heat/chemical energy into motion has been suggested to provide utility for these peptides as elastic biomolecular machines in drug delivery, biomolecular desalination and as mechanochemical engines etc. (Urry, 1993, Angew. Chem. Int. Ed. Engl., 32, p819-841; Urry, 1995, Scientific American, 272, p44-49). The utility of these molecules has however been limited to applications involving their elastic, barrier, template and prosthetic properties, using synthetic materials composed almost exclusively of elastomeric polypeptides.

Surprisingly it has now been found that elastomeric peptides may be incorporated into biomolecules and on transition of the peptide between the extended and contracted state, the tertiary structure of the biomolecule is altered thereby affecting the characteristics or properties of the biomolecule. This may occur by altering the secondary structure of relevant regions of the protein or by altering the spatial relationship of certain domains within the biomolecule to one another, or a combination of both events.

Thus, viewed from one aspect, the present invention provides a biomolecule containing a functional moiety and at least one inserted amino acid sequence comprising an elastomeric peptide which is contractable under appropriate changes in one or more environmental parameters, wherein the functional moiety exhibits altered properties or characteristics on induction of contraction or extension of the inserted sequence.

Alternatively viewed, this aspect of the invention provides a biomolecule containing a first part into which at least one amino acid sequence comprising an

elastomeric peptide which is contractable under appropriate changes in one or more environmental Parameters is inserted, wherein said first part exhibits altered properties or characteristics on induction of contraction or extension of the inserted sequence, or alternatively a modified biomolecule modified by insertion of at least one amino acid sequence comprising an elastomeric peptide which is contractable under appropriate changes in one or more environmental parameters, wherein said biomolecule exhibits altered properties or characteristics on induction of contraction or extension of the inserted sequence.

As mentioned previously only large elastomeric polymers have been contemplated as having any utility and these have been used in such applications as drug delivery or as prosthetic materials (e.g. vascular prostheses). This perceived limitation has necessarily restricted the applications for which elastomeric polymers were thought to be useful. For example, biomolecules such as proteins generally have limited regions into which elastomeric polymers could be inserted without disrupting the required secondary and tertiary arrangement of the protein. Thus it would not previously have been thought possible to introduce elastomeric peptides into such biomolecules without entirely altering the structure and/or function of the molecules.

It has however unexpectedly been found that small oligomers with as little as a single elastomeric peptide repeat unit (or contractile unit) still undergo extension and contraction in response to external stimuli in a similar manner to large polymers with multiple repeat units. For examples, elastomeric peptides comprising 1 to 10, e.g. 1 to 5, "contractile units" may be used, where each "contractile unit" comprises 3 to 6 amino acids. These small oligomers can thus be inserted into biomolecules without substantially

affecting the integrity of the biomolecules and also introduce an inducible switch mechanism into the molecules.

A "biomolecule" is intended to mean any molecule, or collection of molecules, e.g. a complex, which exhibits a biological function and is composed of naturally occurring constituents or analogs or derivatives thereof. Biomolecules as defined herein have at least two parts, the first of which contributes the biological function, as defined above and is referred to herein as the "functional moiety". This functionality may be conferred by one or more molecules, e.g. subunits of an enzyme. The second part is the inserted elastomeric peptide. Such functions include enzymatic, binding, structural, activatory and inhibitory roles. Thus, functional moieties of biomolecules of the invention include for example, enzymes, receptors, antibodies, co-factors, inhibitors, antigens presenting particular epitopes, regulatory proteins and their peptides (functional variants) or derivatives or analogs with such functions.

Biomolecules are not limited to polypeptides and include also for example, nucleic acid molecules, e.g.

DNA template sequences or primers, into which peptides with the desired elastomeric properties, have been inserted. Biomolecules may have more than one function and may be composed of distinct moieties, e.g. fusion proteins or a nucleic acid sequence attached to a polypeptide or a polypeptide attached to a small chemical component such as a hapten. As mentioned above, biomolecules may have more than 1 constituent molecule or component. These may or may not separably have biological functions, for example a protein which is composed of more than one polypeptide chain.

Analogs and derivatives of naturally occurring constituents include those which have similar properties, such as the use of non-naturally occurring

amino acids to derive a polypeptide.

Preferably naturally occurring molecules are the first parts of biomolecules of the invention, for example antibodies. Such molecules include their functional parts, regions or domains (functional variants) such as the variable domain or region of antibodies, or may be as small as peptides, for example an antigenic determinant. Derivatives and analogs of these naturally occurring products and their variants may be produced by alteration of the natural sequence, by the use of non-natural constituents, and may be produced by synthetic methods. The natural sequence may be altered by replacement of one or more of their constituent components, e.g. amino acids or nucleic acids, or by altering the relative arrangement of different domains, optionally with the addition of further sequences, e.g. the production of single chain antibodies. Preferably the functional moieties and biomolecules according to the invention comprise one or more polypeptides.

"Inserted" as used herein refers to the inclusion of the amino acid sequence as part of the biomolecule structure. Thus the "inserted" amino acid sequence may be attached at the terminal end of the molecule (or one of the molecules making up the biomolecule) , together with which it forms the biomolecule. Alternatively the insert may be attached within the molecule. When more than one molecule is present in the biomolecule, the amino acid insert may or may not be attached to the molecule contributing the functional moiety. Preferably the attachment is covalent. Especially preferably, the amino acid sequence is inserted between constituent components e.g. amino acids or nucleic acids, making up a chain forming at least part of the remainder of the biomolecule to form a contiguous chain.

It will be appreciated that in many cases to minimize disruption of the structure of the biomolecule,

the inserted amino acid sequence should replace a part of the molecule into which is inserted, for example replace part of the functional moiety. This aspect forms a preferred feature of the invention. For example, in the case of a polypeptide biomolecule, the inserted sequence can replace the same or a similar number of amino acid residues taken from the host polypeptide, thus keeping the number of amino acids approximately the same, and allowing normal spatial arrangement of the biomolecule even after insertion of the elastomeric peptide. This of course requires that the position of insertion is appropriately chosen and this will be considered in more detail below.

The "amino acid sequence" for insertion, as recited above, includes but does not necessarily consist solely of the elastomeric peptide. Thus flanking regions may be used on one or both sides of the peptide, for example to facilitate binding to the molecule into which the sequence is to be inserted, to provide a site of recognition for affinity purposes, to provide a cleavage susceptible region (e.g. protease or acid/alkaline sensitive), to mimic regions which the inserted sequence will replace or otherwise for convenience. Such flanking regions are preferably small, for example 1 to 100 amino acids in length, especially preferably 1 to 20 amino acids in length.

Especially preferably the flanking region at the N- terminal may be GVG or GGVG. Although not wishing to be bound by theory, it appears that this sequence may be able to nucleate folding of P-turns.

As used herein, "elastomeric peptide" refers to a peptide which exhibits inducible, reversible contraction from an extended state to a contracted state, by reduction in the distance between the ends of the peptide backbone of each contractile unit, on induction by a change in temperature, pH or pressure, e.g. contraction on increasing temperature. It should be

noted that whilst the "elastomeric peptide" exhibits reversible contraction, when in situ in a biomolecule, the freedom to exhibit reversible contraction may be limited leading to complete or partial irreversibility.

Preferably contraction achieves a more than 20% reduction in overall length. Especially preferably, >40k reduction is achieved. Each contractile unit (or repeat unit) refers to the minimum peptide sequence able to contract on induction, e.g. VPGVG, which may be linked to the same or different sequences.

"Environmental parameters" which may be changed to induce contraction or extension of the peptide include those parameters which have been found to cause the state transition of synthetic elastin polymers, such as a change in temperature, pressure, pH, the addition of salts or organic solutes, e.g. an altered analyte concentration, or by photochemical reaction or electrochemical or chemical reduction. The environment as referred to herein denotes the local environment of the biomolecule, ie. the physical parameters of the solution or other carrier into which it is placed.

These parameters may provide indicators of various changes in the local environment, e.g. changes in electrical current or colour due to a change in analyte concentration.

It will be appreciated that the particular change required to cause the induction will depend on the state of the other environmental parameters, owing to their interrelationship in affecting the hydrophobicity interactions of the peptide, and will be influenced by the choice of elastomeric peptide in the biomolecules of the invention.

The "altered properties or characteristics" of the functional moiety on induction of contraction or extension of the inserted peptide include reversible and irreversible changes in the structure or geometry of the functional moiety such that the biological activity

attributed to that region is affected. This includes elimination of the activity, increase or reduction of the activity or alteration of the activity such that it has different characteristics, e.g. exhibits different kinetic parameters for a particular enzymatic or binding reaction.

Induction of contraction or extension may also cause a previously unobserved activity of the functional moiety with a latent or suppressed biological activity to be initiated or enhanced. This may result through spatial modification of the interaction between different domains of the functional moiety (which may be on one or more separate molecules) , by modification of a single domain of the functional moiety to affect its interaction with other discrete molecules (e.g. binding partners), or by modification through alteration of the steric relationship of the functional moiety to the elastomeric peptide.

Irreversible changes may be caused by for example appropriate selection of the elastomeric peptide such that induction of the peptide introduces a P-turn which cannot be reversed to the extended state.

Whilst suitable elastomeric peptides for use in the invention vary widely and require only that they have the aforementioned inducible contractile features, preferably the peptides have the sequence: (XP/G(X),),, wherein XP/G(X), represents a contractile unit in which P/G is either proline or glycine, each X is any amino acid and may be the same or different, m is an integer from 1 to 10, n is an integer from 1 to 100 and each unit in the sequence may be the same or different.

Preferably each X is selected from glycine, valine, alanine, isoleucine, leucine, phenylalanine or tyrosine and derivatives thereof, e.g. P-alanine, or other

hydrophobic amino acid residues, preferably aliphatic hydrophobic amino acid residues. Especially preferably, X is glycine, valine or isoleucine. Thus, in a preferred aspect, the general sequence may be (XPGXG), (ie. (x)m= GXG) or (XPGX)n (ie. (X)m = GX). Particularly preferably, the sequence XP/G(X), is the sequence VPGVG (ie. (X)m is GVG), VPGV or IPGVG. It will be appreciated that since the elastomeric peptide employed will exhibit a distinct transition temperature of contraction depending on its sequence, that one or more X residues may be modified accordingly to provide a desired temperature of contraction, e.g. by the use of more hydrophilic amino acid residues.

Further suitable elastomeric peptides for use in the invention have the sequence: (PEXK) n or (XKEXPEKX) n (Respectively, derived from the sequence of the contractile muscle protein tint in and from an artificial polypeptide known to undergo cold-denaturation and which forms an helix on heating.), wherein P, E and K refer to proline, glutamic acid and lysine, respectively, and each X, which may be the same or different, and n are as defined above.

Preferably X in the latter sequences is leucine or valine. Especially preferably, the sequence PEXK is PEVK and the sequence XKEXPEKX is LKELPEKL.

Preferably m is 1 to 6, especially preferably 1 to 3. Preferably n=1 to 10, especially preferably 1 to 5.

Especially preferably the length of the elastomeric peptide is 5 to 25 amino acids in length, e.g. 5, 15 or 25. Particularly preferred is the use of an amino acid insert of 5 to 100 amino acids in length, particularly, 8 to 20 amino acids.

A biomolecule of the invention may be produced by

chemical synthesis, for example by manual or automated synthesis of an appropriate polypeptide chain or polypeptide/nucleic acid molecule. Synthesis may be performed in solution or on a solid support using suitable solid phases known in the art. Separate parts of the complete molecule, produced synthetically or naturally, may be joined by chemical reaction to provide a covalent link. Alternatively, when the biomolecule is entirely composed of a polypeptide chain, this may be produced by expression of a DNA sequence encoding said biomolecule. Nucleic acid molecules containing a sequence encoding for biomolecules of the invention form a further aspect of the invention. Vector DNAs containing such nucleic acid molecules form a further feature of the invention. These vectors may be used to express biomolecules of the invention.

The applications for biomolecules of the invention are extremely varied and are based on any application in which variation of a biological function is required and such applications form further aspects of the invention.

As the variation may be induced by, for example, a change in pressure, temperature or pH, an inducible trigger may be incorporated into even the most sensitive systems. The majority of such applications involve the introduction of an elastomeric peptide switch into polypeptides. The biological activity contributed by the functional moiety of such polypeptides may then be modified (either altering (e.g. changing specificity), increasing or decreasing the activity) on contraction or extension of the peptide.

As mentioned previously, it is preferable to insert the elastomeric peptide into the (one of the) molecules of the biomolecule at a site which does not affect the functionality of said molecule. Thus for example, when the insert is positioned within such molecules and forms a contiguous chain with them, the point of insertion is generally one which will not significantly affect the

tertiary structure of domains contributing to the functional moiety. In proteins, this may be a surface loop (Hawrani et al., 1994, Tibtech, 12, p207-211) or in the single chain antibody example described below, the linker region. Surface loops are often 2 to 12 amino acids in length and may be replaced by similar sized elastomeric peptides.

As described in the examples herein, the Fc-binding three-helix bundle Z domain based on the IgG-binding domain B of staphylococcal protein A (Nilsson et al., 1987, Protein Engng., 1, p107-113) has a surface region which may readily be replaced without significantly affecting the surface geometry or function of that molecule.

The polypeptide into which an elastomeric peptide is introduced may be, for example, an enzyme which on contraction or extension converts the enzyme into an active or inactive form, or more/less active form or alters its specificity for particular substrates, inhibitors, activators or co-factors. This may be caused by alteration of the secondary or tertiary structure of the enzyme or relevant domains of the enzyme or by influencing the relative arrangement of domains within the enzyme, or accessibility of domains of the enzyme, by other portions of the enzyme or by the contracted/extended elastomeric peptide itself.

The increasing availability of enzymes from various organisms with specific defined activities has led to the use of these catalysts as reagents in many in vitro and in vivo systems. Notably, methods of detection and analysis in the area of molecular biology require the use of a least one of the enzymes involved in DNA/RNA replication, transcription and/or translation. Precise control of the activity of these enzymes is generally achieved through precise knowledge of their pH, temperature, ionic strength and co-factor requirements and the consideration of other criteria essential for

their working. Not only is the ability to control activation of these enzymes important, but also the ability to inactivate the enzymes, generally reversibly.

The capacity to turn the activity of an enzyme on and off is often crucial to the correct functioning of a particular analytical or diagnostic assay.

As a consequence of the invention described herein, biological activity of enzymatic reactions may now be controlled by the switch mechanism such that these reactions may be turned on or off by altering the state of the inserted peptide. This avoids either removing the enzyme or introducing inhibitors or deactivating (usually irreversibly) in some other way. Examples include the use of polymerases in polymerase chain reactions (PCR) and related techniques in which their activity is required only once the reaction has reached a certain temperature after initiation of the reaction to reduce production of a heterogeneous product. This has generally been overcome by the use of the hot-start method in which the polymerase is only activated at increased temperature, either by addition of the polymerase or a vital component to the reaction mix once that temperature has been reached. The present invention provides a further alternative hot-start method.

Furthermore, there are several enzymes which are regulated by conformational changes involving hinge movements, e.g. lactate dehydrogenase, triose phosphate isomerase, triglyceryl lipase, adenylate kinase (Gerstein, M., et al., 1993, J. Mol. Biol., 229, p494- 501, and references therein) . Elastomeric peptides may be used to replace loops, hinges and joints which thereafter will respond on temperature or other environmental changes. Thermostability and/or thermolability may also be improved by the introduction of an elastomeric peptide into a protein which exhibits thermosensitivity by replacement of an appropriate

region.

The duration of enzymatic reactions may also be precisely controlled by induction on altering particular environmental parameters (e.g. temperature, pressure or pH) for only a strictly controlled period of time after the initial mixing. This has applications particularly in processes in which the parameters may readily be altered or which vary as a consequence of the process.

In this connection, biomolecules may be induced at certain temperatures thus providing a desirable activity, e.g. enzyme activity, which may be used in processes such as fermentation which exhibits changes in temperature.

Alternatively, if the functional moiety contributes a binding function, such as in the case of antibodies, preferably single chain antibodies, the binding may be turned on or off or altered reversibly, or irreversibly if so desired. Many applications arise from this possibility. For example, affinity columns may be created in which the biomolecule with a binding function is bound to a solid support. The sample from which the other partner of the binding pair is to be extracted is then brought into contact with the solid phase under conditions at which binding can occur. After washing or other steps, the bound analyte may be released by appropriately altering the binding functionality of the biomolecule by contraction/extension of the elastomeric peptide contained therein. The biomolecules may thus be used for purification and extraction purposes to provide one partner of a binding pair, which may be immobilized if desired.

Numerous solid supports suitable for this purpose are well known in the art and widely described in the literature and generally speaking, the solid support may be any of the well-known supports or matrices which are currently widely used or proposed for immobilization, separation etc. in chemical or biochemical procedures.

Thus for example, the solid support may take the form of particles, sheets, gels, filters, membranes, microfibre strips, tubes or plates, fibres or capillaries, made for example of a polymeric material e.g. agarose, cellulose, alginate, teflon, latex or polystyrene. Particulate materials, e.g. beads, are generally preferred.

Conveniently, the solid support may comprise magnetic particles, which permit the ready separation of immobilized material by magnetic aggregation.

Preferably such magnetic particles are superparamagnetic to avoid magnetic remanence and hence clumping, and advantageously are monodisperse to provide uniform kinetics and separation. The preparation of superparamagnetic monodisperse particles is described by Sintef in EP-A-106873. The monodisperse polymeric superparamagnetic beads sold as DYNABEADS by Dynal AS (Oslo, Norway) are exemplary of commercially available magnetic particles which may be used or modified for use according to the invention.

An example of a biomolecule in which the functional moiety has a binding function is a single chain antibody, as described in US 4946778 and 5260203 of Ladner et al.; W088/09344 of Creative Biomolecules Inc.; and W093/11161 and W094/12520 of Enzon Inc. in which variable regions of the light and heavy chain of antibodies are linked by one or more linker regions. In the normal configuration, these molecules adopt an arrangement such that the single chain is configured in 2 separate domains, linked by the linker region(s), wherein the domains interact together to form an equivalent binding site to that occurring in native antibodies.

The linker region(s) is preferably from 10 to 30 amino acids in length, especially preferably 12 to 15 residues may be replaced at least in part by an amino acid sequence with an elastomeric peptide as described herein to form a biomolecule of the invention. On

induction of contraction, the linker would be reduced in length such that the domain at either end were forced out of its normal configuration such that the binding site is altered, thereby preventing or altering antigen binding.

As an alternative, the linker could be replaced with an elastomeric peptide of greater relative length in the extended state such that the domains again do not interact appropriately. On induction of contraction, the peptide would shorten to the appropriate length and the domains would be brought into contact thus increasing affinity for the antigen. As a further alternative, if an elastomeric peptide of less than 10 amino acids is used to replace the linker, if this region is sufficiently short, contraction may result in the formation of diabodies, in which 2 single chain antibodies become dimerized thus forming an antigen binding site. The reversible production of diabodies in this manner and the diabodies thus formed are preferred aspects of the invention. It will therefore be seen that contraction and extension may be used to affect the functionality in either a positive or negative manner depending on the choice of the properties of the elastomeric peptide, e.g. length.

In a further example, an elastomeric peptide may be inserted into a leucine zipper. The peptide may be appropriately selected such that contraction or extension brings the separate domains into or out of contact.

Non-polypeptide biomolecules of the invention include nucleic acid molecules into which an amino acid sequence has been inserted.

Biomolecules of the invention may consist of an elastomeric peptide linking two independent domains such as ligand:receptor, inhibitor:enzyme or nucleic acid template:repressor pairs. Induction of contraction or extension may be used to bring the appropriate regions

of association of the pairs into contact or out of contact such that the consequences which ensue from the interaction may be initiated, terminated or modified.

Biomolecules of the invention may be used as biosensors to detect or monitor changes in the parameters of the biomolecules' local environment such as temperature or analyte concentration. For such purposes, the use of reversibly inducible biomolecules is preferred. In this case when such a parameter changes, the elastomeric peptide would respond appropriately and the biomolecule in which it was inserted chosen such that its altered characteristics could be detected, e.g. activation of enzymatic activity which produces a coloured product.

If modified antibodies (or other binding proteins) are produced, the binding to their antigens may be manipulated, for example by inducing disruption of the active site in response to a defined physical stimulus.

Such proteins have applications in purification and detection procedures, e.g in ELISA or other binding assays. Applications in high throughput screening are also applicable. Such assay methods and kits for performing them form further aspects of the invention.

Furthermore, biomolecules of the invention may be used to form libraries, e.g. expressing random peptides, such as bacteriophage or RNA translation-derived libraries. Each member of said library would thus comprise a fusion molecule in which an elastomeric peptide was associated with a peptide (or oligonucleotide) for expression. The properties of these fusion molecules could then be altered by contraction or extension of the elastomeric peptide, e.g. to aid panning during screening, e.g. by reversibly introducing specific binding properties into the library members.

In particular, the use of biomolecules of the invention in diagnostic assays forms a preferred aspect

of the invention.

Furthermore, biomolecules of the present invention may be used in therapeutic applications. Thus, for example, molecules which have advantageous therapeutic properties may be modified to create biomolecules of the invention, wherein at least a part of said biomolecules may be influenced by the environmental parameters of the context into which they are placed.

In this connection, biomolecules which have more than one functionality may be used. The inducible functionality may be the therapeutically desirable activity or may be an aspect of the biomolecule which may be varied to aid the therapeutic treatment, e.g may be used for inducible localization to particular areas, e.g. to localize therapeutic activity or to remove the biomolecules from a body part or fluid.

For example, biomolecules of the invention may be used for extracorporeal treatment of body fluid and may have their activity increased, decreased or altered appropriately by the fluid into which they are placed or by altering the conditions of said fluid, e.g. pressure or temperature. As a further example, biomolecules of the invention may be administered to a human or animal body locally or systemically and be subject to alterations in their functionality due to the local environmental parameters of the body, e.g. pH variation during passage through the gastrointestinal tract, or on temperature fluctuation, e.g. to 370C, or to higher temperatures at sites of infection. Alternatively, modification of the biomolecules' local environment may be altered to cause contraction or extension of the elastomeric peptide in the biomolecule and hence alteration of its functionality. Thus, for example, biomolecules (for example having an enzymatic activity which is absent until contraction occurs) may be applied, locally or systemically and activated by, for example, heating superficially or within the body for

treatment of the site at which the environmental parameters are altered.

It will be appreciated that biomolecules for medicinal use should be administered according to known techniques and dosing regimes specific to the therapeutic functionality of the biomolecule.

Thus, the present invention further provides biomolecules of the invention for use as a medicament.

Alternatively, the invention provides a method of treating a disease or condition of a human or animal body with an effective amount of a biomolecule of the invention.

It will be apparent to the skilled person that the applications of the above described invention are considerable and not restricted to the small number of examples described above.

The present invention will now be described by reference to the following non-limiting Examples, in which: Figure 1 shows CD wavelength scans of H-GGVG(VPGVG)-OH (peptide D) in 10 mM phosphate buffer pH 7; Figure 2 shows melting curves of Peptides E (open square, Ac-GVGVPGVG-NH2), G (open circle, Ac-GGVGVPGVG-NH2) and H (closed circle, Ac-GGVGVPGVG-OH) in 10 mM phosphate buffer pH 7.0; Figure 3 shows melting curves of Peptide D (H-GGVGVPGVG-OH) at different pH (open square, pH 7.0, 200nm; open circle, pH 4, 198nm; closed circle, pH 9.5, 200nm); Figure 4 shows melting curves of Peptide J (open circle) and Peptide N (closed circle) in 10 mM phosphate buffer pH 7.0; Figure 5 shows the reversibility of Ac-GVG(VPGVG)3IL-NH2 (peptide N) in 10 mM phosphate buffer pH 7 in which results are shown at 200C ( ) , or at 200C after incubation at + 700C (---);

Figure 6 shows the concentration dependence of peptide L at +15.5°C, inset shows variation of MRE with concentration at 199nm (open circles) and 210nm (closed circles) Figure 7 shows the concentration dependence of peptide L at +62.5°C, inset shows variation of MRE with concentration at 199nm (open circles) and 210nm (closed circles) Figure 8 shows the effect of SDS and TFE on peptide E; Figure 9 shows the effect of SDS and TFE on peptide L; Figure 10 shows the effect of increasing concentrations of TFE on peptide L a) at various concentrations; and b) at different TFE concentrations measured at 198nm (open circles), 206nm (closed circles) or 213nm (open squares) Figure 11 shows the effect of temperature and TFE on peptide L at a) 6.80C or b) 58.6°C; Figure 12 shows CD-wavelength scans of peptides D and L at different temperatures in 10 mM phosphate buffer pH 7.0; (a) Peptide D (49.1µM) was recorded at 1.1°C (open circles; 15.4°C (closed circles); 25.0°C (open squares); 34.50C (closed squares); 43.80C (open triangles); 53.00C (closed triangles); 63.00C (open inverted triangles); and 81.50C (closed inverted triangles); (b) Peptide L (20.7µM) was scanned at 1.3°C (open circles); 15.40C (closed circles); 34.0°C (open squares); 48.40C (closed squares; and 76.60C (open triangles) Figure 13 shows melting curves of the 8- or 9-mer peptides C (37.1 M), D (49.1 yM), E (38.5 yM), G (41.8 yM) and H (31.3 yM) and the longer peptides J (16.4 M), L (20.7 UM) and N (17.9 yM) scanned from 190-240/260 at different temperatures in 10 mM acetate or borate buffers (panel a), or in 10 mM phosphate buffer (panels b and c), in which the mean residual ellipticity at or close to 200 nm is plotted versus temperature and van't

Hoff fits are also shown; (a) Melting of the 9 mer peptides C (open circles, at 201 nm, pH 4.0), D (closed circles, at 200 nm, pH 9.5) and G (open squares, at 201 nm, pH 9.5); (b) Melting of the 8-9 mer peptides H (closed squares, 202 nm), E (open triangles, 199 nm) and D (closed triangles, 200 nm); (c) Melting of the 18-21 mer peptides N (open inverted triangles), J (closed inverted triangles) and L (open diamond) at 199 nm; Figure 14 shows the reversibility and concentration dependence at pH 7.0; (a) Scans of the 18-mer peptide L (91 M in 10 mM phosphate buffer; 0.5 mm cuvette; spacing 0.5 nm; 4 scans each with 1 sec integration time) at 20.60C before heating (continuous line) and after incubation at 700C for 30 minutes with rapid cooling back to 20.60C (dashed line). The concentration dependence of MRE is shown in (b) and (c). The concentration of peptide L was varied from 4.6 yM to 264 yM (b) using different quartz cuvettes (0.5 mm, 2 mm and 10 mm), and similarly the 8- mer peptide E was varied between 45 FM and 855 pM (c).

The variation of mean residual ellipticity for peptide L at 15.50C or peptide E at 15.60C (at 199 nm, open circle; and at 210 nm, closed circle) and for peptide L at 62.5"C or peptide E at 48.30C (at 199 nm, open triangle and at 210 nm, open square) is shown; Figure 15 shows fitting of the melting curve of peptide B in 10 mM phosphate buffer pH 7.0; the CD-data at 199 nm (open square) were fitted to a macroscopic, reversible, two-state model by optimising the linearity of the correlated van't Hoff plot. By using the fitted endpoints for the transition, [0] F and [0]U, from the initial model (panel (a), dashed line) the corresponding van't Hoff plot was constructed (panel b),open circle, dashed line) which gave a linear correlation coefficient (r) of -0.91. After fixing [B]F, the correlation coefficient for the van't Hoff plot improved to -0.98.

The data with the correlation line (closed circle, continuous line) and its corresponding fit for data of MRE against temperature with fixed [0] F (continuous line) are respectively shown in panel (b) and panel (a); Figure 16 shows the average melting temperatures for thermal transitions of 8- and 9-mer Elastin Peptides, in which the large numbers represent the average of TM'S for the 8-mers (peptide A, B, E and F) and 9- mers (peptide C, D, G and H) from Table 6. The 'low TM' value is the average of the lowest melting temperatures at pH 4.0 or pH 9.5 for peptide A and F (or C and H for the 9-mer) and at pH 4 and 9.5 for peptide E (G for the 9-mer) - the pH-value which gives an uncharged molecule. Similarly, the 'high TM' value represents the average of high TM'S for peptide A and F (C and H), and at both pH 4 and pH 9.5 for peptide B (D for the 9-mer) - the pH-value which gives a charged molecule. The smaller numbers in italics represent the difference between average Th-values; Figure 17 shows the effect of SDS and TFE on peptides E and L, in which (a) Peptide E (218 yM) and (b) peptide L (91 UM) were scanned in water (open circles), in 2 mM SDS (continuous line), in 25 mM SDS (constant length dashed line), and in 87 W (v/v) TFE (variable length dashed line) at 20.50C using 0.5 mm quartz cuvettes (4 scans; 0.5 nm spacing; 1 sec integration time).

(c) Peptide L (20.7 yM) was also tested at different concentrations of TFE at 15.70C in a 2 mm cuvette, and the variation of mean residual ellipticity versus TFE concentration at 198 nm (closed circle), at 206 nm (open square) and at 213 nm (closed square) is shown; Figure 18 shows Circular Dichroic wavelength scans of peptides 1-4 in phosphate buffer pH 7.0 in which the peptides were scanned once in 10 mM buffer at +6.40C

using a 0.2 cm stoppered cuvette with 5 s integration time (0.5 nm steps; slit: 2.0 nm). The concentrations were respectively, peptide 1 (21.4 M), peptide 2 (26.9 M), peptide 3 (29.7 pM) and peptide 4 (21.1 M); Figure 19 shows the concentration effect of trifluoroethanol (TFE) on the peptides, in which the effect of increasing concentrations of TFE in phosphate buffer on peptide 1 (a) and peptide 2 (b) is shown. The peptides were scanned once at +6.40C and the experimental conditions and peptide concentrations were the same as described in Figure 18. The effect on the mean residual ellipticity at 222 nm for peptide 1 (open circle), peptide 2 (closed circle), peptide 3 (open square) and peptide 4 (closed square) is illustrated in panel (c); Figure 20 shows the temperature effect on peptides in buffers and trifluoroethanol (TFE) in which panel (a) shows the CD-wavelength scans of peptide 1 at different temperatures (experimental conditions as described in Figure 18). The inset scans are of peptide 1 in 10 mM phosphate buffer pH 7.0 while the larger panel shows those for the peptide in 30 % (v/v) TFE in phosphate buffer. The melting of the peptides was followed at 222 nm using the average of three independent scans with an integration time of 5 s (in 0.2 cm cuvette, slit: 2.0), and in panel (b) the temperature effects on peptide 1 in different solvents are shown. Respectively in 20 k (v/v) TFE (open circle), 60 W (closed circle) TFE and 25 mM SDS (closed square) in phosphate buffer and for phosphate buffer alone (open square). The melting of peptide 1 (open triangle), peptide 2 (closed triangle), peptide 3 (open inverted triangle) and peptide 4 (closed inverted triangle) in 60 k (v/v) TFE are displayed in (c). The data in (b) and (c) were fitted to a two- state transition with fixed pre-transitional

endpoints (see methods) . The reversibility of melting was tested as is illustrated in panel (a) (dashed line); Figure 21 shows fitting of association kinetics to surface plasmon resonance binding data to establish an appropriate kinetic model, in which the panels show respectively the fitting of the association kinetics data of Peptide 1 (0.32 µM) at +150C to a one-component (A+B = AB) model (a) or a two-component (A+ B1 + B2 = ABl + AB2) model (b). The bold line is the fitting to the raw-data (open circle) with the correlated residuals for the fit (closed circle); Figure 22 shows the determination of kinetic parameters for peptides 1 and 2 binding Fc-IgG2a at +19.90C, in which the on-rates for peptide 1 binding the Fc-IgG2a were found by plotting ks1 (open circle) and ks2 (closed circle) for the lowest concentrations tested, panel (a) : (III), 2.5 µM; (IV) , 1.25 µM; (V), 0.63 µM; and (VI), 0.31 µM (inset). The on-rates were found from the slopes of linear fits of the above data. Only the fit of ks2 versus peptide concentration gave a significant on-rate with ks2 = 148,396 M-ls-l (linear correlation coefficient, (r) = - 0.98). Using the highest concentrations (I) 25 AM or (II) 5 pM also the off-rates were determined for about 200 seconds (large panel). In this case 5 µM gave the highest off-rates respectively koff1 = 2.4 x 10-3s-l and koff1 = 0.237 s-l. Similarly the on-rates for peptide 2 (b, inset) were found by plotting ks1 (open circle) and ks2 (closed circle) versus their respective concentrations of the analyte from (II), 773.6 µM; (III), 386.8 µM ; (IV), 200.8 µM ; (V), 193.4 µM; (VI), 150.63 µM; (VII), 100.5 µM; (VIII), 96.7 pM. kon1 was 74 M-1 switch correlation coefficient, (r) = 0.3, and kon2 was 502 M-l srnwith its (r) = 0.98. The koff-rates, 0.367 s-l and 8.9 x 10-6 s-l, were found from about 200 seconds of the

1.547 mM dissociation phase (I); Figure 23 shows the temperature dependency of the fast kinetic constants for peptide 1 and peptide 2 in which the respective kinetic constants of peptide 1 (open circle) and peptide 2 (closed circle) are shown in panels a, b and c. Panel a) shows the fast off- rates, panel b) the fast on-rates and panel c) illustrates their respective fast dissociation constants (= koff/kon) at different temperatures. The off-rates were fitted using BIAevaluation 2.1 and the on-rates was fitted using both BIAevaluation 2.1 and GRAFIT (Leatherbarrow, 1989, Grafit 3.01 edit.

Erithacus Software Ltd). Both the on-rates and off- rates display their fitted standard errors. The standard errors of the dissociation constant were manually calculated based on SE's for kon and koff; Figure 24 shows the temperature dependency of binding free energy (AAG) of peptide 2 relative to peptide 1, in which the binding free energy at different temperatures was calculated as described in the methods and the AAG (AAG = +RT ln[ KD (peptide 2 at TX) /KD (peptide 1 at Tx)]) versus different temperatures is displayed. The linear fit of the data using GRAFIT (Leatherbarrow, 1989, supra) gave a linear correlation coefficient (r) of -0.99, and a slope of -0.0570 + 0.007 kcal mol1 K1 (intercept 20.2 + 2.0); Figure 25 shows the temperature CD-wavelength scans of peptides 2 and 4 in buffers, in which Peptide 2 (163.4 pM) in Hepes buffered saline (HBS) added 0.005 W (v/v) surfactant P20, was scanned once from 198 nm to 260 nm (0.05 cm quartz cuvette with 5 s integration time; slit 2.0 nm; 0.5 nm step size) at different temperatures, as displayed in panel a).

Similarly peptide 2 (31.8 pM) and peptide 4 (24.1 pM) were scanned at different temperatures in 10 mM phosphate buffer pH 7 using a 0.2 cm cuvette as shown

for peptide 2 in panel b) and for peptide 4 in panel c). The bold dashed lines illustrate the reversibility of the scans after being heated up to 60-800C and then by rescanning the peptides at around 250C; Figure 26 shows melting of peptides 2-4 in phosphate buffer and HBS, in which the melting of the different peptides was followed by measuring the CD-ellipticity at 222 nm/260 nm (in 0.2 cm cuvette, slit: 2.0 nm) at different temperatures averaging three independent scans with an integration time of 5 s. Panel a) illustrates the melting of the peptide 2 (open circle, 31.8 M), peptide 3 (closed circle, 32.7 M) and peptide 4 (open triangle, 24.1 pM) in 10 mM phosphate buffer pH 7.0. Likewise in panel b) are the melting curves of peptide 2 (closed square, 27.0 pM) and peptide 4 (open square, 24.1 pM) in Hepes buffered saline pH 7.4 with 0.005 k surfactant P20; Figure 27 panel (a) shows the structure of the pDAP2 plasmid and panel (b) shows the structure of the pUC119His6mycXba; Figure 28 shows the sequence of the linker region for the 3D6 gene; Figure 29 shows the structure and position of the 3D6 gene and associated primers in plasmids. RBS = Ribosome binding site and ALK.PHOSPHAT = alkaline phosphatase.

Figure 30 shows the assembly of linker region (I); Figure 31 shows the assembly of linker region (II); Figure 32 shows the assembly of linker region (III); Figure 33 shows the structure of linker regions (I) and (III); Figure 34 shows the sequences of various oligonucleotides used in the present invention; Figure 35 shows the sequence of peptides 5 and 6 synthesised by solid phase peptide synthesis; Figure 36 shows CD-scans of peptide 6 in different

solvents. 1.5 mM of peptide was scanned in 0.01 mm quartz cuvette by 10 seconds integration time and 0.5 nm spacing at 250C; Figure 37 shows peptide 5 (1 yM) (panel A) analysed on BIAcore 2000 by binding to immobilised Fc-561 (3400 RU) at different temperatures. Flow: 30 yl/min of HBS with 0.005% P20. Panel B illustrates likewise the binding of 500 nM peptide 6 to Fc-561; and Figure 38 shows fast on- and off-rates (and dissociation constants) of peptides 5 and 6 at different temperatures (analysed as described in Example 3).

EXAMPLE 1: Design of short elastomeric peptides Short elastin peptides, from 5-mers to 25-mers, with flanking sequences of 3 to 6 amino acids were synthesised and analysed by Circular Dichroism (CD).

The transition temperature, Th, for folding of these peptides was followed by monitoring the decrease in CD amplitude at 197-200nm, and at 205-212nm.

ABBREVIATIONS CD, Circular Dichroism; ESP/MS, positive electrospray mass spectrometry; Fmoc, -9-fluorenylmethyloxy- carbonyl; Fmoc-PAL, 5-[4-(9- fluorenylmethyloxycarbonyl)- aminomethyl-3,5- dimethoxyphenoxy)] valeric acid; HOBT, 1- hydroxybenzotriazole hydrate; HPLC, high-performance liquid chromatography; PEG, polyethylene glycol; PS, polystyrene; TFA, trifluoroacetic acid; TFE, trifluoroethanol; SDS, sodium dodecyl sulphate, [8],, MRE, mean residual ellipticity.

MATERIALS AND METHODS Materials Peptide synthesis reagent grade of Fmoc amino acids and Fmoc-PAL-PEG-PS resin were purchased from Perspective Biosystems GmbH (Hamburg, Germany).

Trans-Androsterone and Dioxane were obtained from Fluka. Acetic Anhydride and Methanol were of peptide synthesis grade and were from Applied Biosystems (currently Perkin Elmer Applied Biosystems Division).

Peptide Synthesis and Purification Peptides A-O (see Table 1) were synthesised on Fmoc- PAL-PEG-PS resin (to give an amidated C-terminal peptide) or Fmoc-Gly-PEG-PS resin (Millipore, Watford, UK to give a free C-terminal end) using MilliGen/Biosearch 9050 PepSynthesizer (Millipore, Bedford, MA) with Solid-Phase Fmoc chemistry on a 0.1 mmol scale. The first amino acid (Gly) required more than 12 hours of coupling to Fmoc-PAL-PEG-PS due to slow coupling kinetics, but using standard chemistry (Fmoc/HOBT-activation) and procedures. The remaining amino acids were coupled using standard chemistry and protocols as described by the manufacturer, typically an 89-minute cycle per amino acid. Coupling efficiency was followed continuously by recording absorbance at 544 nm using Acid Violet 17 dye colour giving yields above 95%. After acetylation with acetic anhydride in DMF (for-N-terminal protected peptides) the resin was washed sequentially with dichloromethane, methanol, diethylether and then dried using N2.

Peptides were cleaved from the resin and deprotected by treatment with TFA containing 5% thioanisole, 3% 1,2-ethanedithiol and 2% anisole as scavengers for 4- 6 hours at 200C. The cleaved peptides were washed

adding 10-fold excess of petrol ether and then precipitated in excess of diethylether. The peptide solid was dissolved in deionised water and lyophilised.

The crude peptides from the TFA-cleavage were applied on an analytical HPLC (LKB, Bromma, Sweden). They were analysed using a gradient of 4.5 - 54% acetonitrile in 0.1% TFA over 91 minutes at +200C on a C18 column (Vydac, 250 x 5 mm, 10 jim pore).

Peptides A-O were detected at 214 nm and eluted (0.7ml/min) respectively with A-H: 18-20.5% acetonitrile, and I:33%, J:32%, K:31%, L:29%, M:31%, N:33%, 0:36% acetonitrile. Large scale purification using Waters HPLC (Water, UK) at 200C was performed using a shallower gradient elution, e.g. for peptides A-H a gradient of 13.5 - 22.5% acetonitrile in 0.1% TFA over 90 minutes was applied. Peptides were chromatographed at 10 ml/min from a preparative 240 x 24 mm Vydac C18 column (10 jim pore) with detection at 214 nm and then lyophilised.

The purified peptides were dissolved in 1:1 water/methanol with 1.0% acetic acid at a concentration of 10 ng/yl and analysed for purity and identity on ESP/MS (VG Autospec with VG Analytical Electrospray, Fison, UK). Samples were subject to quantitative and qualitative amino acid analysis (samples were hydrolysed in HCl at 1100C for 24 hours under N2-gas and then freeze-dried, dissolved in citrate buffer pH 2.2, and analysed on Pharmacia AlphaPlus II amino acid analyser).

Circular Dichroism Studies CD spectra were recorded on Jobin-Yvon Model CD6 Dichrograph Instrument (Instruments S.A. UK Ltd., Stanmore, UK) which was calibrated with either a

solution of iso-andosterone (ISO) in dioxane (1.25 mg/ml) or a solution of (+)-10-camphorsulfonic acid (CSA) in water (Johnson, 1990, Proteins: Struct., Funcc. Genet., 7, p205-214). ISO-calibration was performed at 304 nm (gives an intensity at 142.5 AA using a 1.0 cm quartz cuvette) according to manufacturer's instructions.

Quartz cuvettes from Hellma (0.5 to 2 mm stoppered) were used, and the CD instrument was temperature controlled using a water bath and constantly purged with N2 during analysis. Temperature CD scans were performed by cooling the samples to +10C and then stepwise increasing temperature allowing samples to equilibrate at each temperature for 5 minutes from +10 to +850C. The cuvette temperature was measured with a Fluke 51 K/J digital thermometer.

The scans were collected using a spectral acquisition spacing of 0.5 or 1.0 nm (with 2.0 nm bandwidth) with an integration time of 1 second from 190 up to 260 nm. Scans were processed on a computer and the average of 4-15 scans were smoothed using Sabitzky Golay algorithm in the Dichrograph Software version 1.1 (Jobin-Yvon/Instruments S.A., France). Buffers (10 mM) were filtered through a 0.22 mm filter and centrifuged at 13000 rpm to remove dust and air. All spectra were baseline corrected for buffer and for off-axis drift using ellipticity at 240/260nm.

Purified peptides were dissolved in deionized water and then diluted 44 times in 10 mM buffer, ensuring the peptide concentrations in the cuvette were always below 80 yM (to keep sample absorbance below 1.0 in all spectral regions) Stock peptide concentration was determined by quantitative amino acid analysis. Data points at

each wavelength were converted from peptide molar ellipticity to mean residue molar ellipticity ( [#], degree cm2/dmol) by dividing by the number of residues in the peptide.

A single wavelength from the wavelength scans (190-240/260 nm) was graphed versus different temperatures. The free energy for the transition into folded form at higher temperatures was fitted to a macroscopic, reversible, two-state model (1) using the observed CD-data, [#]obs, temperature, T, and fixed endpoints for the transition (high temperature folded form, [e]F, and low temperature unfolded form [#]U).

#GU-F = -RT ln [([#]obs - [#]U)/([#]F - [#]obs)] (1) The heat capacity was set to zero due to the low molecular weight of the peptides, and all the data was thus fitted to a van't Hoff plot using equations (2) and (3), or (2) and (4) for 18- 28 mer peptides.

Equation (4) takes into account the slope effect (m) and off-axis adjustment (off) for transitions which have a slope in the pre- and postfolding base lines.

#GU-F = #HU-F - T#SU-F (2) [e]"b" = ([e]U + [e]F x exp- (#GU-F /RT))/(1 + exp- (#GU-F/RT)) (3) [e]"b" = ([e]" + [#]F x exp- (#GU-F /RT))/(1 + exp- (#Gu-F/RT)) + m x T + off (4) The data was analysed with least squares fitting using Grafit software (Leatherbarrow, 1989, supra) and van't Hoff plots were constructed (plot of lnK versus 1/T where K = ([0]obs - [#]U)/([#]F ~ [e]obs)

utilising the data from CD, [e] obs t and the previously fitted endpoints for transition, [e]Uand [e]F. In some cases where the initial free fit gave a lower linear correlation coefficient ( r < -0.90 to - 0.931) for the van't Hoff plot, the data were refitted using the initial fitted values for endpoints and starting points and fixing either [G]u or [Q]F, or both [0]Uand [e]F. This resulted in correlation coefficients closer to -1. A new fit of AH and AS was found from equations (2)-(4) by fixing [e]Uand [e]F from the best van't Hoff plot. The transition temperature, T,, for all peptides was calculated using the relation Tm = SH/AS.

Computer simulations of peptides Models of high and low temperature forms of the elastin-like peptides were built by extracting phi and psi angles from molecular dynamics simulations (Wasserman & Salemme, 1990, Biopolymer, 29, p1613- 1631). All modelling and simulations were carried out on a Silicon Graphics Indy using Insight 95 and Discover 2.6.0 (Molecular Simulations Inc.).

RESULTS Design of Short Elastin Peptides There have been several reports by Urry and co- workers describing chemically or physically induced structural transitions for elastin polymers (Urry, 1993, Angew. Chem. Int. Ed. Engl., 32, p819-841).

However, none of these engineered switch models described have been tested on shorter elastin peptides. It has also been seen that the length and the presence of N- and C-terminal caps in elastin peptides affects the transition temperature for aggregation (McPherson, et al, 1992, Biotechnol.

Prog., 8, p347-352). A correlation between CD-

spectroscopy and thermoelasticity data has been reported (Urry, et al, 1985, Biochem. Biophys. Res.

Comm., 130(1), p50-57; Urry et al., 1986, Int. J.

Pept. Prot. Res., 28, p649-660).

In this study short elastin peptides were designed to test the effect of sequence variations on the conformational transitions observed. Of specific interest was the effect of N- and C-terminal acetylation and amidation, and the effect of an extra glycine added at the N-termini of 8-mers (Table 1, Peptides A-H). In addition, it was reasoned that a glycine at the N-terminal end would increase mobility and solubility of the peptide, along with the charges on the N- and C-terminals (Harpaz et al., 1994, Proc.

Natl. Acad. Sci., USA, 91, p311-315). In order to test if there was any correlation between the number of elastomeric (-VPGVG-) units present and the transition temperature for structural folding, we also synthesised an 18-mer peptide (3 VPGW units) and a 28-mer peptide(5 VPGVG units) (Table 1, Peptides E, L and 0). To measure the influence of hydrophobic sequences on the transition temperature for folding of an 18-mer elastin-like peptide extra leucine and isoleucine residues were introduced stepwise at the C-terminus (Table 1, Peptides L-N).

It was also believed that an extra glycine lying close to these leucine and isoleucine residues in the C-terminal end in the C-terminal end might influence the mobility of this hydrophobic patch (Table 1, Peptide I). The effect of charged residues adjacent to or within the elastomeric sequences were also studied. In one design an extra hydrophobic residue was placed close to a charged lysine at the N-terminal end (Table 1, Peptide J) while a glutamic acid residue was placed in the middle of the elastin

sequence (Table 1, Peptide K). All the above peptide designs were studied by CD at different temperatures and, where feasible, thermodynamic data were extracted.

Peptide Purification and Analysis The 15 short elastin sequences shown in Table 1 were synthesised and purified on HPLC using shallow gradient conditions. The elution of these peptides as a single peak on reverse-phase analytical HPLC indicated that they were pure, and the identity, purity and concentration of each peptide were verified determined by ESP/MS and amino acid analysis (Table 2).

CD Analysis of Pep tides CD spectra of the short and of the longer peptides in buffer at different temperatures showed a similar profile (Figure 1 - which shows the results obtained with peptide D). The global minimum was somewhat dependent on the buffer (Table 3), but was typically within the range 198-200 nm - as found for more unordered peptides(Woody, 1995, Methods Enzymol., 246, p34-71). However, for the short peptides at +10 C the minimum mean residual ellipticity was -7,000 to -10,000 deg cm2 dmole-l compared to -40,000 deg cm2 dmole-l for an ideal random coil. This is within the same range observed by Urry et al.(Urry et al., 1986, supra, Urry et al., 1985, supra) for the longer elastin polymers, reported to be between -5,000 and -16,000 deg cm2 dmole-l for the (VPGG)n and the (VPGVG)n polymers. The corresponding values for the longer peptides (I-O) were within the range -17,000 to -19,000. This suggests, somewhat surprisingly, that the shorter peptides have a more ordered structure at low temperatures than the longer peptides. These values are those expected from an

equimolar mixture of random coil and type II P-turns (Perczel et al., 1993, Int. J. Pept. Protein Res., 41, p223-236). Furthermore, the energy minimised structural model of the elastin polypentapeptide also incorporates a partially folded structure as an extended, low temperature form (Chang & Urry, 1988, supra; Wasserman & Salemme, 1990, supra). For all peptides the CD amplitude around 195-205 decreased with increasing temperature, with a smaller decrease in amplitude at 206-212 nm. A positive peak in this region is a characteristic of type II P-turns (Urry et al., 1986, supra; Urry et al., 1985, supra; Woody, 1995, supra).

For peptide K, which has the most heterogeneous composition (Table 1), there was an increase in CD amplitude at 210-230 nm with a local minimum at 222 nm (with an approximate isodichroic point at 208nm).

These were also found at higher temperatures for some short peptides with the N-terminal GGVG sequence.

See Table 1: Peptides C and G, pH 7, and Peptide D, pH 4. These spectra resemble class C-spectra associated with helical or type I/III P-turn structures. Class C-spectra have an exciton (n-n*) splitting with a positive band at 195 nm (often blue-shifted towards 180 nm due to solvation) and a negative band at 208 nm. There is also a strong negative n-n* transition around 220-224 nm. Only band intensities and ratios are different from the helix CD spectrum and therefore it is often difficult to distinguish helix from type I P-turns(Baldwin et al., 1994, Int. J. Pept. Protein Res., 43, p180-183; Perczel et al., 1993, supra; Woody, 1995, supra).

The melting curves for the different peptides (Figure 2-4) again surprisingly gave a similar curve to that

described for the longer elastin polymers (Urry et al., 1985, supra). For the polytetrapeptide (VPGG)n the transition range for peptides A-H in this study was 3-5,000 MRE units increasing to 5-10,000 for the longer peptides I-O (data not shown). This was lower than for the polypentapeptide (VPGVG)n which has a change of 16,000 MRE units for a transition.

After fitting the CD data to van't Hoff equation it was possible to extract data for AS and AH for the folding of the smaller peptides (Table 3). For some of the larger peptides the slope before and after transitions were sufficient large that a full thermodynamic analysis was statistically difficult.

For these the transition midpoint only was reported (Table 4). It was not possible to determine Tm for peptides K and 0. All the other peptides however had a positive AH and AS which implies an entropy-driven process. The similar AH and AS values for the 8-mer, 9-mer and 21-mer peptides (Tables 3 & 4) also indicated that each residue in the repeat is acting as an independent unit. On the other hand, as is shown in Table 5, AH and AS were much larger per residue for the shorter peptides compared to the 21-mer and the elastin polymer.

There were not any significant differences in Tm for 8- and 9-mers (Table 3). However, acetylation of a-amino and amidation of a-carboxy groups lowered the transition temperatures by 8-10° C for both 8- and 9-mers compared to free N- and C-termini (peptides E and G vs. peptides B and D). There were no differences in Tm at different pH values for the unprotected peptides B and D, indicating that the presence of either one or the other charge had a similar effect on the structural transition for the short peptides. In support of this, when one of the

termini were protected by amidation or acetylation, there was an increased Tm for the ionic form (compare peptide F with H and A with C in Table 3).

Increasing the length of the peptides from 8-mer to 18-mer without altering the composition, did not change the Tm by more than 3"C (Table 4; +32"C for peptide E vs. +35° C for peptide L).

There was no effect on the transition temperature by placing similar amino acid residues close to each other at one end of the peptides. Two glycines at the N-terminal of the shorter 8/9 mers had the same Tm as the peptide with one glycine, as also found for the leucine/isoleucine pair at the C-terminal end of the 18/20 mer peptides. However, addition of a hydrophobic residue near the C-terminus lowered the transition temperature for the peptide by 17-21"C.

For some peptides, it was not possible to detect an explicit transition temperature due to lack of end points for the transition, even though there was a similar decrease in CD amplitude around 200 nm. This includes the long peptide K with a more mixed composition, and the 28-mer VPG peptide 0. With peptide K the experiments suggest that this peptide was probably more unordered at lower temperatures than the other peptides due to a lower MRE value at +1.4"C (around -20,000 deg cm2 dmole-l). A transition temperature was obtained however for peptide J (+18"C) which has three residue insertions at the N- and C-termini. The effect of inserting lysine into peptide J was opposed by more hydrophobic residues giving a relatively low Tm.

A selection of the short and the longer peptides (peptides C, E, L and N) were tested for reversibility by incubation for 30 minutes at +700 C and then rapidly cooling to +200 C. All were found

to be fully reversible (Figure 5). Likewise, the concentration dependence of peptide L was analysed at two different temperatures, +15.5° C and +62.5° C, (Figures 6 and 7). The ellipticity at 199nm and at 210 nm were plotted from wavelength scans over the peptide concentration range 5-304.5 yM. At low temperature these were found to be independent of peptide concentration (Figure 6). At high temperatures (Figure 7) there was a more marked variation in ellipticity around 210 nm at concentrations below 100 yM.

Effect of my cellar and nonmicellar SDS The effects of the detergent SDS were examined on the short (Table 1, B,E and G) and the longer (Table 1, K, L, N and 0) peptides. For 8- and 9-mers micellar SDS (25 mM) induced an increase of MRE at 210 nm for both peptides, and for the 8-mers this was also observed in nonmicellar (2 mM) SDS (Figure 8). This is characteristic of a transition to a type II P-turn, which exhibits a positive peak at 206-210 nm, as observed for the longer polymers (Urry et al., 1986, supra; Urry et al., 1985, supra; Woody, 1995, supra). Surprisingly, however, with the exception of the peptide K, there was no effect of non-micellar SDS on the longer 18- to 28-mer peptides (peptides L, N and 0), as is illustrated for peptide L in Figure 9. Peptide K showed a decrease in ellipticity from 206-240 nm for both concentrations of SDS. This similar transition to a local minimum around 222 nm was also observed in micellar SDS for peptides L, N and 0. The increase in MRE around 200 nm and slow increase around 206-212 nm indicated that all peptides were becoming more ordered in the presence of SDS.

Effects of trifluoroethanol Induction of the type II P-turn was more pronounced in 87-92 W (v/v) TFE than in SDS for the 8-, 9-, 18- and 28-mer peptides (Figures 8 and 9). For peptide L the concentration of TFE was varied from 0-97 %, and at 40 W TFE rate of change of MRE at 198 nm, 206 nm and 213 nm shifted markedly, suggesting the presence of a structural transition (Figure 10). The effect of temperature and increasing concentrations of TFE on peptide L was also studied (Figure 11). Between 6.8° C and 58.6° C the proportion of type II P-turn decreased at high concentrations of TFE or was unchanged at low concentrations of TFE. At low concentrations of TFE the ellipticity at 200 nm increased at high temperature indicating a transition to more ordered structures.

Computer model of peptides Models of the shorter peptides based on phi and psi torsion angles taken from a MD simulation of the longer elastin peptide were build. They showed that the transition from random coil/Z-structure to a P-helix at higher temperatures was not only described by a contraction along the helix-axis but also an expansion within the monomer across the axis, giving a helix with a larger diameter and a shorter length than for the 28 residue peptide.

EXAMPLE 2: Further experiments were conducted using the peptides and methodology as described in Example 1.

RESULTS: CD analyses CD spectra of the 9-mer peptide D at various temperatures are shown in Figure 12a. For comparison, spectra of the 18-mer peptide L are shown in Figure 12b. All other short and longer peptides showed a similar CD profile (data not shown).

For all peptides the CD amplitudes at 195-205 nm decreased with increasing temperature, with a smaller decrease in amplitude at 206-212 nm. A positive peak in this latter region is characteristic of type II b- turns (Perczel et al., 1993, supra; Urry et al., 1985, supra; Woody, 1995, supra). A near isodichroic point around 218 nm was found for the short peptides in the reported transition. This was seen both for the effect of increasing concentrations of TFE (data not shown), which induces a type IIp-turn, and for the temperature effect described in Figures 12a and b. A near isodichroic point for melting of the larger elastin polymer has previously been detected by Urry and colleagues (1988a, J. Protein Chem., 7(1), 1-34) around 220 nm. However, some peptides showed little change in the position of minima/maxima when the temperature is varied and the spectroscopic changes in these are dominated by changes in intensity.

The transition curves of the short peptides D, E, H in phosphate buffer, peptides C, D and G in different buffers, and for the longer peptides J, L and N at pH 7 are shown in Figures 13 a-c. The start of the transition was initially fitted for each melting curve, although well defined transitions were obtained for the majority of peptides.

Thermodynamic analyses Reversibility was tested as described in Example 1.

All were found to be fully reversible (Figure 14a).

The concentration dependence was also tested on peptide L (18-mer) and on the most hydrophobic short peptide E (8-mer) at two different temperatures (Figures 14b, c). In all cases ellipticities were found, within the limit of experimental error, to be independent of peptide concentration over this temperature range. This observation shows that the temperature-induced formation of the turn is due to intramolecular interactions rather than intermolecular associations. The melting curves reported here were obtained at peptide concentrations 10-20 times lower than the maximum concentrations shown in Figure 14c (around 18 pM for the longer peptides and below 50 M for the shorter peptides) Data were fitted to van't Hoff plots and were linear.

Where a low signal-to-noise ratio in the CD- measurements resulted in a larger spread of data, a better fit was obtained by fixing one or both endpoints for the transition (Figure 15). The results using free [0]F and [0]U were AH = 13.7 + 6.6, AS = 0.045 + 0.021; whereas when modelled with fixed [O]Fthe following results were obtained: AH = 10.4 + 1.4, AS + 0.034 + 0.004.

Fitting of the CD data to the van't Hoff equation allowed calculation of AH, AS and values for the transitions of most peptides. However, due to the uncertainty of the start of the transition for some short peptides (Figure 13b) only those with well defined transition curves are reported (Table 16).

Melting temperatures for short peptides which had N-

and C-termini in the same state were well correlated (Table 7 and Figure 16). Acetylation of amino or amidation of a-carboxy groups (or both) lowered the transition temperatures by an average of 130C for both 8- and 9-mers compared to free N- and/or C- termini. However, there were no significant differences in melting temperatures for peptides where the terminal charges differed in sign (Table 7, compare peptide A and B at pH 4 versus B and F at pH 9.5 for the 8-mers, and peptide C and D at pH 4 versus D and H at pH 9.5 for the 9-mers), indicating that the presence of either one or the other charge had a similar effect on the structural transition for the short peptides. The melting temperatures for the peptides at pH 7.0 were lowered compared to those with 'high TM' values (Figure 16), and the effect was larger for the 9-mer peptides compared to the 8-mers.

This suggests that there may be an additive stabilising effect through N-C-terminal interactions - the lower melting temperatures found for some of the smaller 8-mers compared to the larger 18-mer may be explained by this. The presence of an extra glycine at the N-termini of the smaller peptides increased the melting temperatures by about 6" for both 'low TM' and 'high TM' peptides (Figure 16). An opposite effect was seen by the addition of a hydrophobic residue (or residues) near the C- terminus, which resulted in a lowering of the melting temperature by 20-23° C (compare peptide L with M and N, Table 6).

Effects of SDS and TFE Results of the effects of SDS and TFE (conducted as described in Example 1) are shown in Figure 17.

DISCUSSION OF THE RESULTS OF EXAMPLES 1 AND 2 For all the peptides studied the increase in MRE at 195-205 nm and at 206-212nm, a signature of type II P-turn formation, observed with increasing temperature supports the notion that the folding of these sequences is driven by hydrophobic collapse.

The exact structures formed by the elastin-like peptides studied here was dependent on both flanking sequences and the solvent conditions. At low temperature (-0" C) the short peptides showed evidence of partial structure, as has been seen with the larger polymers (Urry et al., 1986, supra; Urry et al., 1985, supra). In addition, most peptides showed a small increase in MRE at 210 nm, typical of type II P-turn formation in VPGVG polymers (Urry et al., 1985).

For peptide K, there was a pronounced increase in CD- amplitude at 210-230 nm, suggesting the formation of a type I of type III turn. This was also observed for this peptide in both micellar and non-micellar SDS. Some evidence of type I/III P-turns was seen in the CD-spectra of other peptides at high temperatures. This suggest that an equilibrium of different P-turn populations may exist for many of these sequences, although peptides having a GGVG N-terminal region are thought to preferentially adopt the type II' P-turn (Broch et al., 1996, Int. J.

Pept. Protein Res., 47, p394-404). Interconversion between type I and type II turns is thought to require +0.2 to -1.7 kcal/mol (Yang et al., 1996, J.

Mol. Biol., 259, p873-882, and references therein), a small enough barrier to be affected by flanking sequences or buffer compositions (TFE, SDS, etc.).

Alternatively, isomerisation of the Val-Pro bond to a cis-conformation in the shorter peptide would favour

type I or type III P-turns as found for a cyclic analogue of VPGVG studied in water (Khaled et al., 1981, Int. J. Pept. Protein Res., 17, p23-33; Renugopalakrishnan et al., 1978, J. Chem. Soc. Perkin Trans., 2, plll-119). However, whether or not type I, II or II P-turns are formed, elastomeric transitions are still possible (Bhandary et al., 1990, Int. J. Pept. Protein Res., 36, p122-127).

Previous crystallographic studies on the linear pentapeptide Boc-VPGVG-OMe (Ayato et al., 1980, in "Peptide Chemistry", Ed.: Okawa, K. p107-112) showed the absence of any p-turn structure while the structure of a cyclic decapeptide analogue of VPGVG exhibited a type III turn for one VPGVG unit followed by a type I turn for the second unit (Bhandary et al., 1990, supra).

Flanking regions are important to consider in order to specify a transition temperature for an elastin sequence grafted into other protein. A hydrophobic scale by Urry and co-workers(1993, Angew. Chem. Int.

Ed. Engl., 32, p819-841) only predicts Tm for residues within the polymer sequence. Here we have clearly demonstrated that, when suitable flanking sequences are present, the VPGVG monomer is able to form the identical type II turn structure (within the discrimination capacity of CD) to that formed by the elastin-like polymer (VPGVG)n. We have also confirmed that, when particular types of flanking residue are present, the transition temperature can be altered in the same direction as seen for the polymers (Urry, 1993, supra). For example, the addition of extra hydrophobic residues adjacent to the VPGVG turn sequence lowered the melting temperature by up to 200 C and an additional glycine increased TM by 60 C. It seems quite remarkably that

these large temperature effects could be produced by adding a single residue (compare the 8-mers versus 9- mers (Figure 16) and peptides L and M, Table 6), although when multiple additions were made the effect was not necessarily additive (compare peptides M and N, Table 6). Further, when a charged residue was inserted in the presence of these additional hydrophobic residues, the expected increase in melting temperature was not seen (compare peptides I and J, Table 6). This suggests that Tm for an elastomeric sequence flanked by different non-elastomeric residues may be found by averaging the balance of hydrophobic and hydrophilic residues at the flanking regions. The effect of charges is illustrated by the shorter peptides. For all peptides the charges at the N- and C-termini increased the transition temperature for the hydrophobic folding by as much as 200 C (compare peptides C and H at pH 4 and 9.5, Table 7). The thermodynamic effect of charged groups at the termini is likely to derive from an increased stability of the water shell surrounding the peptide.

In contrast Tm was lowered for peptides with charged termini in phosphate buffer pH 7.0 compared to pH 4 and pH 9.5. At pH 7 both termini will be charged, and a possible salt bridge formation between N- and C-terminus may stabilise the P-turn and lower the Tm for the transition. For the amidated and acetylated VPGVG peptides of different length (peptides E and L) transition temperatures varied from those observed for the much longer VPGVG polymer (Luan et al., 1990, Biopolymers, 29, p1699-l7O6; Urry et al., 1985, supra). Curiously, as also found from Table 5, the difference in free energy between two temperatures, 0° (unfolded state) and 60° C (folded state) is approximately -2.0 kcal/mol for a pentamer sequence

(AG = 5 x 2.0 + (5 x 0.0066) T). This is about the same transition energy per pentamer as found by a molecular dynamics simulation of 35 residues(Chang & Urry, 1988, supra). The important conclusion here is that the structural transition seems to be independent of peptide length and to some extent insensitive of flanking regions, making these peptides useful for protein engineering purposes.

The effect of SDS on short elastin-sequences supports the assumption that hydrophobic interactions are the driving force for stabilisation of elastin structure at higher temperatures. In this study, low concentrations of SDS (up to 25mM) induced turn formation (Figure 17a). One possible mechanism is that SDS may bridge the two valines spanning each side (cis-face) of the PG-turn. It may then disrupt the hydration shell around these valines and facilitate closer Val-Val interaction with an induction of a type II P-turn. In fact, the model based on MD simulation of elastin peptides Wasserman & Salemme (1990, supra) suggests that the distance between the two valines in the high temperature form is approximate 2A shorter than in the low temperature form. The effect of SDS on the longer peptides (more than one VPGVG monomer) is consistent with the hypothesis that the valines are more shielded there.

The relatively higher AH and AS for the shorter peptides compared to the longer elastin polymer (Table 5) supports this mechanism. As AH and AS are dependent on the amount of clathrate-like water molecules surrounding the hydrophobic moieties, larger nH and AS values imply that a larger percentage of the hydrophobic surface in these peptides is accessible to solvent.

The effect of increasing concentrations of TFE on

peptide L indicates that the change in ellipticity rate reflects or balances the internal hydrophobicity of the peptide. The strongest effect of TFE was observed below 40 W TFE where the hydrophobic interactions are dominant (Bodkin & Goodfellow, 1996, Biopolymers, 39, p43-50). The shorter distance between Val-groups in the contracted/type II P-turn form of the VPGVG monomer, implies that partial shielding of these hydrophobic residues from water, by minimising surface area, may be a driving force for induction of type II P-turn in elastin by TFE.

Bodkin & Goodfellow (1996, supra) have discussed this as a general model for TFE-induced folding of secondary structures. Since the total free energy is lower for the TFE/water mixture than for water alone, the preferred form will be where the Val groups are close in the TFE-mixture. Above 40 W TFE the electrostatic interactions dominate, reflected by the slower formation of type II P-turn structures.

However, the recent work of Cammers-Goodwin et al.

(1996, J. Am. Chem. Soc., 118, p3082-3090) suggests that TFE also destabilises the coil state, where amide groups are solvent exposed, by replacing the surrounding water molecules and increasing the cis-/trans- interconversion rate. This is proposed to maximise amide-amide hydrogen bonding resulting in increased secondary structure formation in order to lower the free energy. In our studies, the temperature effects on the stability of type II P-turn at high concentrations of TFE, and likewise on the more unordered structure at low concentrations, were divergent. At increased temperatures the MRE at 206 nm decreased both for the polymer (VPGVG)n and for peptide L in this study. There may be several factors affecting this temperature induced transitions at high concentrations of TFE. The type II P-turn may undergo melting or may be transformed

into other turns. This can be a consequence of a temperature dependent equilibrium constant for the different turn types, an increased cis-trans interconversion rate, destabilisation of hydrophobic Val-Val interactions due to more dominant backbone electrostatic forces, or due to TFE dependent decrease in CD amplitude around 206 nm for type II turns. At low concentrations of TFE, however, the decrease in CD amplitude around 200 nm may reflect the additive affect of TFE and temperature to disrupt the hydration shell surrounding Val groups in the peptide and increasing Val-Val interactions.

This study thus demonstrates that short, elastin-like peptides can undergo the identical structural transitions that one observed in neutral and modified polymers. In elastin polymers, (VPGVG),, the identical pentamer sequences all contribute to form a beta-spiral with type II turns (Pro2-Gly3 with a Val C=O H-N Val4 hydrogen bond). Intramolecular hydrophobic contacts are thought to drive the transition within this spiral (Urry, 1988a, supra).

However, the only stabilising hydrogen bonds reside within each of the monomers, VPGVG. Even so, it is plausible that formation of the spiral is a cooperative transition, driven by hydrophobic collapse and stabilised by intra-pentamer H-bonds.

We find no evidence for such a cooperative effect.

The AH and AS values for an 8-mer (1 VPGVG unit) and an 18-mer (3 VPGVG units) are the same indicating that a 'cooperative unit' does not exceed one repeat.

The contribution of configurational entropy to the total entropy on going from an 8-mer to an 18-mer can be shown to be minimal - the difference in conformational entropy, ASCOnf, between the number of states for the relaxed (764) and extended (58)

conformations of a single VPGVG peptide has been calculated to be 1.024 cal mol-l K1 per residue (one entropic unit per residue). The increased conformational entropy change between an 8-mer and an 18-mer would then be just 0.010 kcal mol~l- K1, assuming the entropy derives from internal chain dynamics without random chain networks (Urry, 1988a, supra). This shows that AS through the transition (Table 6) cannot be balanced by such small increases in ASconf as the chain length increases.

This leads to the possibility of engineering such sequences into globular proteins. The free energy involved for a transition of an 8- mer (peptide E) from 0° to 600 was -3.2 kcal/mol (Table 5; AG = 8 x 2.0 -T (8 x 0.0066)). This should be a sufficient amount of energy to change conformation of a structure and function in a globular protein. On the other hand it might be desirable to increase the numbers of monomers (VPGVG) in order to have a stronger and more acute force. For example, 3-5 monomers coupled in series as for a linker peptide, may use the energy of one single large plastic deformation to move different domains or subunits of a protein. Again such sequences may be used to replace surface turns inducing local structural transitions or by insertion into enzymes or binding proteins thus offering a novel method for modulating thermo-stability or -lability of proteins.

Table 1 : Sequences of short elastin peptides with protected or unprotected ends and longer pep tides with different composition. Differences between the peptides are highlighted Ac- acetyl, -NH2 - amide A: H-GVG(VPGVG)-NH2 E: Ac-GVG(VPGVG)-NH2 B: H-GVG(VPGVG)-OH F: Ac-GVG(VPGVG)-OH C: H-GGVG(VPGVG)-NH2 G: Ac-GGVG(VPGVG)-NH2 D: H-GGVG(VPGVG)-OH H: Ac-GGVG(VPGVG)-OH I: Ac-GVG (VPGVG) (VPGVG) (VPGVG) ILG-NH2 J: Ac-GKL (VPGVG) (VPGVG) (VPGVG) ILG-NH2 K: Ac-GKL (VPGVG) (VPGEG) (VPGVG) ILG-NH2 L: Ac-GVG (VPGVG) (VPGVG) (VPGVG) -NH2 M: Ac-GVG (VPGVG) (VPGVG) (VPGVG) L-NH2 N: Ac-GVG (VPGVG) (VPGVG) (VPGVG) IL-NH2 O: Ac-GVG (VPGVG) (VPGVG) (VPGVG) (VPGVG) (VPGVG) -NH2 Table 2: Characterisation of short elastin peptides<BR> MH+<BR> peptide [found (expected)] amino acid analysis [aa, found (expected)]<BR> A 639.4 (640.1) P, 0.94 (1); G, 4.04 (4); V, 2.44 (3)<BR> B 641.4 (640.4) P, 1.05 (1); G, 4.22 (4); V, 2.52 (3)<BR> C 696.4 (697.2) P, 0.92 (1); G, 5.31 (5); V, 2.67 (3)<BR> D 698.4 (697.4) P, 0.90 (1); G, 5.15 (5); V, 2.54 (3)<BR> E 682.4 (682.1) P, 0.95 (1); G, 4.29 (4); V, 2.63 (3)<BR> F 683.4 (683.4) P, 0.88 (1); G, 4.36 (4); V, 2.65 (3)<BR> G 738.4 (739.2) P, 1.01 (1); G, 5.20 (5); V, 2.66 (3)<BR> H 739.4 (740.2) P, 0.95 (1); G, 5.06 (5); V, 2.53 (3)<BR> I 1784.2 (1785.2) P, 3.74 (3); G, 8.24 (9); V, 5.68 (7); I, 1.13 (1);<BR> L, 1.27 (1)<BR> J 1869.3 (1870.3) P, 3.52 (3); G, 7.39 (8); V, 4.84 (6); I, 1.02 (1);<BR> L, 2.24 (2); K, 1.21 (1)<BR> K 1900.5 (1900.3) P, 3.25 (3); G, 7.98 (8); V, 4.32 (5); I, 0.88 (1);<BR> L, 1.97 (2); K, 1.09 (1); E, 1.17 (1)<BR> L 1500.7 (1501.8) P, 3.80 (3); G, 7.65 (8); V, 5.66 (7)<BR> M 1614.0 (1614.9) P, 3.43 (3); G, 6.96 (8); V, 5.06 (7); L, 1.24 (1)<BR> N 1728.6 (1728.1) P, 3.53 (3); G, 7.80 (8); V, 5.87 (7); I, 1.03 (1);<BR> L, 1.19 (1)<BR> O 2320.8 (2320.8) P, 5.82 (5); G, 12.22 (12); V, 9.13 (11) Table 3. Thermodynamic values for thermal transitions in 8- and 9-mer elastin<BR> peptidesa<BR> Tm (°C) #HU#F (kcal mol-1) #SU#F (kcal mol-1K-1)<BR> Peptide pH 4.0 pH 7.0 pH 9.5 pH 4.0 pH 7.0 pH 9.5 pH 4.0 pH 7.0 pH 9.5<BR> A 44 40 31 11.2#4.2 14.0#3.6 22.8#6.3 0.035#0.013 0.045#0.011 0.075#0.021<BR> B 43 39 41 13.7#3.6 16.3#3.8 13.2#4.5 0.043#0.012 0.052#0.012 0.042#0.014<BR> C 44 41 31 17.1#6.3 14.7#3.2 12.6#3.4 0.054#0.007 0.047#0.010 0.042#0.011<BR> D 38 36 38 16.3#3.9 13.9#1.5 15.9#2.1 0.052#0.012 0.045#0.005 0.051#0.007<BR> E 31 32 33 15.7#2.8 15.6#3.1 16.8#5.1 0.052#0.009 0.051#0.010 0.055#0.017<BR> F 33 37 40 16.8#3.3 13.4#4.3 10.2#2.4 0.055#0.011 0.043#0.014 0.033#0.008<BR> G 32 31 33 17.3#4.7 15.7#3.1 20.8#5.8 0.057#0.015 0.052#0.010 0.068#0.019<BR> H 30 39 44 15.8#6.9 16.3#2.4 17.0#3.1 0.052#0.023 0.052#0.008 0.054#0.010<BR> aThe MRE for each peptide was followed at one wavelength for each pH and plotted against temperature. Dependent of buffer, pH and type<BR> of peptide, this was ranging from 197-202 nm. Values were obtained by fitting data to van't Hoff equation with #Cp = 0 allowing #H and<BR> #S to float. Tm was found from the relation #H/#S.

Table 4. Thermodynamic Values for Thermal Transitions in 18- to 21-mer Elastin peptides at pH 7.0a Peptide Tm (°C) #HU#F #SU#F (kcal mol-1) (kcalmol-1K-1) I 16 19.5 + 25.9 0.068 + 0.088 J 18 25.2 + 5.2 0.083 + 0.018 L 35 M 14 16.0 + 13 0.055 + 0.043 N 18 a The data analysis was performed as described in Table 3 with exception that values were obtained by fitting data to van't Hoff equation with ACp = 0, specifying the slope and off-axis drift of the baselines, and allowing AH and AS to float.

Table 5. Comparison of AH and AS for Elastin peptides With Other Works.

Peptide AH AS (kcal mol-l) (cal mol-1K-1) per residue per residue 8-mer buffer average (E) 2.0 6.6 9-mer buffer average (G) 2.0 6.6 21-mer average (I + J) 1.1 3.6 50-mer Alanine peptidea 1.3 (-VPGVG-)n, n>l20b 0.24 0.8 a (Scholtz et al., 1991, Proc. Natl. Acad. Sci. USA, 88, p2854-2858), unfolding of helix to coil determined from differential scanning calorimetry (DSC) data (van't Hoff estimate) ; b(Luan et al., 1990, supra), measured by DSC.

Table 6. Thermodynamic values for thermal transitions in elastin peptidesa Peptide pH TM (°C) SHU-F ASU F (kcal mol-l) (kcal mol-1K-1) A 7.0 41 11.0 + 2.7 0.035 + 0.009 B 9.5 35 14.9 + 4.9 0.048 + 0.016 C 4.0 45 20.0 + 2.9 0.063 + 0.009 D 9.5 40 13.0 + 4.8 0.042 + 0.016 G 9.5 35 13.7 + 2.8 0.047 + 0.009 H 9.5 47 14.4 + 2.3 0.045 + 0.007 I 7.0 20 26.0 + 10.6 0.089 + 0.036 J 7.0 18 25.3 + 2.5 0.087 + 0.009 L 7.0 37 18.9 + 4.7 0.061 + 0.015 M 7.0 14 15.9 + 10.7 0.055 + 0.036 N 7.0 17 12.1 + 2.4 0.042 + 0.008 aPeptides A-H (8- and 9-mers) at their specific pH- values were chosen because these melting curves had a well defined transition (Figure 13a). The longer 18- to 21 mer peptides (I-N) were fitted with free endpoints, AHUF and ASU F only specifying the slope of the pre- and post-transition curves.

Table 7: Melting temperatures for transitions in short elastin peptidesa Peptide pH 4.0 pH 7.0 pH 9.5 A 35 41 30 B 40 34 35 C 45 37 24 D 39 27 40 E 22 21 20 F 23 34 39 G 33 26 35 H 26 36 47 a The MRE for each peptide was followed at one wavelength for each pH and plotted against temperature. Dependent of buffer, pH and type of peptide, this was ranging from 197-202 nm. Values were obtained by fitting data to van't Hoff equation with ACp = 0 allowing initially AH, AS and endpoints for transition to float. TM was found from the relation AH/AS.

EXAMPLE 3: Engineering elastic sequences into the primary structure of a minidomain of Protein A Protein A from Staphylococcus aureus is a protein which binds to the Fc-part of mammalian IgG. This three helix-bundle protein is found on the cell surface of bacteria and contains a tandem homologous binding domain, each of approximately 60 residues, designated as E, D, A, B, and C (Lofdahl et al., 1983, Proc. Natl.

Acad. Sci. USA, 80, p697-701; Uhlen, et al., 1984, J.

Biol. Chem., 259, p1695-1702.). Both the X-ray and the NMR (Gouda et al., 1992, Biochemistry, 31(40), p9665-9672) structures are available, and a synthetic protein A domain, the Z-domain, designed for high-level protein expression has been constructed. Recently, the Z-domain has been further minimised to 33 residues retaining most of its affinity towards the Fc-portion of IgG (Braisted & Wells, 1996, supra). The structure consists only of two helices connected by a single type I P-turn.

Peptides having a mutated protein A sequence with different turn regions (ie. in which wild-type sequence is replaced with elastomeric peptides) (Table 8 in which the amino acids in bold are those replaced to introduce the elastomeric peptides) were synthesised by solid phase peptide chemistry and analysed by circular dichroism as described in Example 1. The instrument was calibrated with iso-andosterone in dioxane at 304nm which gives an intensity at 142.5 AA using a 1.0cam quartz cuvette. Melting transitions of the peptides were followed at 222 nm by simultaneously correcting for off-axis drift at 260 nm. The average of three recordings at 222 nm and 260 nm, each of 5-10 seconds integration time, were recorded. The reversibility for the transition was tested by rapidly cooling to 240 C with scanning from 190 to 260 nm. Different solvents

were tested, and the effect on structure of different concentrations of trifluoroethanol was tested at different temperatures. The free energy for the melting of the peptides was found from a model describing a reversible two-state model and the data was fitted using Grafit software (see Example 1).

Surface Plasmon Resonance Spectroscopy Interaction kinetics of the peptides (analyte) with immobilised Fc (ligand) was monitored by surface plasmon resonance spectroscopy using BIAcoreXTX (Biosensor, Sweden). BIACoreX was cooled/heated to equilibrium before measurements at 100 C below or above room temperature (according to manufacturer's recommendations). Purified Fc of the IgG2a mouse monoclonal antibody 561 (aCD34), a gift from Dynal A/S (Norway), was immobilised on research grade CM-5 carboxymethylated sensor chips in 10 mM sodium acetate buffer pH 4.5 using the amine coupling kit with N-ethyl- N' -[(dimethylamino) propyl] carbodiimide (EDC), N- Hydroxysuccinimide (NHS) purchased from manufacturer.

The Fc-561 was immobilised at +100 C using 35 pL of a solution of 100 yg Fc-561/mL (flow of 5 AL/min), and unreacted groups on the surface were blocked with 1 M ethanolamine pH 8.5. After regeneration of the surface with 10 mM Glycine-HCl buffer pH 2.4 approximately 1000 RU of Fc-561 was immobilised. Human serum albumin (HSA, 30 pL of 50 jig HSA/mL, Novo Nordisk, Denmark) was immobilised under similar conditions as a control protein.

The analysis was performed with a flow at 30 pL/min of Hepes buffered saline (10 mM Hepes pH 7.4, 150 mM Nail, 3.4 mM EDTA and 0.005 W P20). Each peptide (analyte) was injected 4-10 x 100 pL. The binding of the analyte to Fc-561 was corrected by simultaneously subtracting the RU's at the HSA portion of the chip (due to bulk-

effects). The surface was regenerated prior to each injection by using 5 uL 10 mM Glycine-HCl buffer pH 2.4.

The data for the association constant was sampled by diluting the analyte in HBS with 0.005 W P20 to obtain several low concentrations, and the dissociation constant was determined from the highest concentration of the analyte to minimise rebinding. The on-rates of binding were followed over the first 200 seconds and likewise the dissociation phase (off-rate) in running buffer was recorded for another 200-300 seconds.

Evaluation of BIAcoreX data The changes in refractive index close to the gold film surface on the chip are correlated to changes in RU, and thus make it possible to monitor changes in binding and dissociation of the analyte. Each sensorgram represents a time dependent change in refractive index recorded in real time. The on-rate and off-rate constants, respectively kon and toff, were fitted by evaluating the sensorgram data using BIAevaluation 2.1 software (Pharmacia Biosensor). By nonlinear fitting of two exponentials, the two-component association phase was found from the equation (5) which is just a sum of two independent one-component events.

R = RA(1 - exp(-ks,t)) + RB(1 - exp(-ksB t)) (5) where R is the response at time, t, and RA and RB are steady state response levels for the two events. In order to determine the on-rates, the fitted values for ksA and ksB were plotted against analyte concentration C.

By using the relations ksA = konAC + koffA and ksB = konBC C + kOffB the kOn-rates for the parallel association were found from the slope of a linear fit of ksA or ksB versus C. The dissociation constants, or off-rates, were determined for the two component interaction as

R = RD exp(-kOffA t) + RE exp(-kOffB t) (6) Here are RD and RE the contribution of each event on the total response at the start of dissociation. However, only the fitted koffA and koffB were further used by specifying the start of dissociation, t = to.

Determination of the different on- and off-rates made it possible to calculate the dissociation constant, KD, KD = koff/kon. And likewise, the relative differences in free energy of binding, AAG between the different peptides and the wild-type peptide 1 were found over a temperature range as: AAG = +RTln[K0(peptide x) at Tx/KD(wild-type peptide 1) at Tx] (7) The R is the molar gas constant and Tx is the temperature (in Kelvin) for KD (based on fast on- and off-rates) and AAG calculations.

RESULTS The structure minimised IgG-binding peptide based on Protein A (Braisted & Wells, 1996, supra) was modified (Table 8, peptide 1, control). It has a sequence which can easily be made using peptide synthesis and it binds IgG with a good on-rate. The type I turn in this peptide, HDPNLN, was replaced with the elastin sequence GVPGVG. We omitted the GV portion at the N-terminal start of the turn because it was flanked by the similar small and hydrophobic residues AL (ALHDPNLN). To avoid affecting the folding of the elastin sequence at higher temperatures, the EE residues at the C-terminal end of the turn (HDPNLNEE) were replaced with QQ (Table 8, Peptide 2 (mutant 1) and Peptide 4 (mutant 2)).

Peptides were thus synthesised by Syntem (Nîmes, France) and were 95-98 W pure from reverse phase HPLC. The

molecular mass was verified by mass spectroscopy, and the composition by amino acid analysis (data not shown) Structural characterisation of pep tides by circular dichroism The peptides were scanned on far-W circular dichroism in phosphate buffer at a low temperature in order to study the effect of mutations in the turn-region (Figure 18). Only the wild-type peptide (Peptide 1, Table 8) had a detectable characteristic alpha-helical CD-spectra with minima around 208 and 222 nm and maximum around 190 nm. However, the relatively low CD-amplitude (MRE around -10,000 deg cm2 dmol-l) indicates that the spectra only represented a partial folded a-helix structure (Luo & Baldwin, 1997, Biochemistry, 36, p8413- 8421 and references herein) . More surprising was the observation that the mutants with the elastin turn were unstructured at the low temperature. However, peptide 3 (mutant 3) with a protected C-terminus, had a larger proportion of helicity (measured at 222 nm) than the other mutants.

TFE has a stabilising effect on a-helices and was thus tested on peptides 1-4. Figures 19a-c also confirms that TFE stabilised or induced a-helix structure in these peptides as is illustrated for peptide 2 (Figure l9b).

By increasing the concentration of TFE in phosphate buffer there was a CD-transition from a random coil to an a-helical spectra for peptide 2, with an isodichroic point around 204 nm. Similar spectra could be obtained for peptide 3 and 4. This shows that all the mutants peptides had the propensity of folding a-helix even though they were largely unordered in phosphate buffer.

The CD-spectra of the wild-type peptide in phosphate buffer only had a partial folded a-helix structure as was found by the scan of the peptide by increasing

concentrations of TFE (Figures l9a and c). At around 20 W (v/v) TFE the transition seemed to be complete, and likewise for peptides 2-4 the main inductive effect of TFE was seen up to 30 W. However increasing the concentration of TFE above 20 æ for peptide 1 decreased the CD-signal at 222 nm. The sigmoid profiles of these CD-curves with an inflection point around 20 W TFE, suggesting that cooperative interactions are involved with this transition.

At TFE concentrations up to 50 W the peptides which were the most helical were peptide 1(wild-type) > peptide 3 = peptide 4 > peptide 2. Micellar SDS (25 mM) also induced CD- spectra which resembled a-helices for both peptide 1 and 2 (data not shown). However, these spectra had a much larger 208/222 ratio than observed for the TFE- induced a-helical spectra (Figures 19a, b), typical for an incomplete a-helix induction or a transition to a 310- helical structure (Hungerford et al., 1996, Angew. Chem.

Int. Ed. Engl. 35, p326-329). The stability of the induced structures in TFE for peptides 1-4 at higher temperatures is shown in Figures 20a-c.

The melting of peptide 1 in 30 W TFE and test of reversibility are shown in Figure 20a. For peptide 1 and 2 the melting curves were approx. 95-100 W reversible, and for peptide 3 and 4 the curves were approx. 95 W reversible after cooling. The isodichroic point at 204 nm also indicated that melting of the a-helices in TFE for peptide 1 followed a two-state reversible transition.

Melting of the a-helices in peptide 1 in different solvent is illustrated in Figure 20b. There was a significant thermo stabilising effect by TFE on the secondary structure, and by increasing the concentration of TFE in phosphate buffer the slopes at TM of the melting curves were decreasing as a function of temperature (lowered cooperativity) . Melting of peptide

1 in micellar SDS had the shallowest slope intercepting the melting curve of peptide 1 in phosphate buffer. Note the destabilising effect on the a-helix structure by SDS on peptide 1 at lower temperatures and the stabilising effect at higher temperatures.

In order to quantitatively study the effect of the mutations in the turn region for the IgG-binding peptide, a-helix structure was induced in all peptides by using 30 and 60 W TFE. The melting of the peptides in 60 k TFE (Figure 20c) thus revealed the differences in a-helix stability for the different peptides.

Further analysis of the melting curves were performed by fitting the data to a reversible van't Hoff two-state transition. The fitting was based on the fitted endpoints for peptide 1 in either 30 W or 60 W TFE and fixing the endpoints for the other melting curves in TFE as these values. Table 9 displays the thermodynamic data for the different peptides. The most dramatic differences in TM for the different peptides were revealed in 30 W TFE. Compared to wild-type peptide 1, TM was lowered by 33-350 C for peptides 3 and 4 and by massive 480 C for peptide 2. These results confirmed that peptide 2 was the least stable peptide and wild-type peptide 1 was the most thermostable peptide.

Evaluation of pep tides by surface plasmon resonance The kinetics of the synthesised peptides binding to immobilised Fc of the mouse monoclonal IgG2a antibody 561 was studied using BIAcoreXTM (BIAcore, Sweden). The fitting of the data for the association phase for peptide 1 at +200 C is illustrated in Figures 21a, b. There was a better fit for peptide 1 to a two-component model.

This was also found for other peptides at all temperatures. However, at +100 C peptide 1 could also be fitted well to a one-component model. The two-component

model was chosen as the best appropriate model over the temperature range tested for both association and dissociation phases.

Figures 22a and b illustrate the typical sensorgrams recorded for peptide 1 and 2. The sensorgrams revealed a binding associated with relatively fast on- and off- rates. The on-rates were measured from the slope a linear fit of the ksA or ksB versus the lowest concentrations used (Figures 22a, b insets). The linear correlation coefficient (r) for the secondary fit of the fast on-rates was around 0.98. Although the fitting of heterogeneous kinetics was more appropriate for the peptides at all temperatures only the fast on-rates could be statistically determined. The slower on-rates had approximate 50-fold lower magnitude and much larger variability (up to 100 %, data not shown). However, both the fast and slow off-rates were statistically determined at lower temperatures, and the slow off-rates for peptides 1-4 at 100 C are shown in Table 10.

The kinetic constants for peptide 1 using the slow off- rates, either obtained from a fit to a one- or a two- component model, were very similar to those reported previously (Starovasnik et al., 1997, Proc. Natl. Acad.

Sci. USA, 94, plO080-10085) . The temperature dependence of the fast on- and off-rates and the dissociation constants for peptide 1 and peptide 2 is illustrated in Figures 23 a-c. The fast off-rates for peptide 1 and 2 were of same order. However, although the on-rates for peptide 1 were 100 fold higher than peptide 2 they responded with opposite effect on temperature. The on- rate for peptide 1 was halved by increasing temperature from 100 C to 300 C, and for peptide 2 it was doubled.

This increased on-rate for peptide 2 at higher temperatures stabilised its dissociation constant.

Surprisingly, peptide 3 and 4 had a much poorer binding to the ligand than peptide 2 although they had a larger a-helix propensity. Their fast on-rates at 100 C were respectively 211 + 113 and 289 + 90 s-l M-1, and their fast off-rates were 0.674 s-l (peptide 3) and 0.563 s- (peptide 4). This was about half of the on-rate and higher off-rate compared to peptide 2. Their rates at the other temperatures were not determined.

The difference in free energy of binding (tag) between peptide 2 relative to the wild-type peptide 1 was linear as is illustrated in Figure 24. In the temperature range 100 C to 300 C AAG was positive showing that the binding of peptide 1 to Fc-561 was stronger than for peptide 2.

However, in the same temperature range AAG decreased by 0.057 kcal moll K1, or the A(AAG) was lowered by 1.2 kcal mol-1. If this relation still is linear over wider temperature range this implies that peptide 2 will be a better binder to Fc-561 than wild-type peptide 1 above 354 K (810 C), and will a gain a free energy of binding with more than 4 kcal mol-1. For peptide 3 and 4 their AAG values relative to peptide 1 at 100 C were respectively +4.8 and +4.5 kcal mol-i (15-20 k more positive AG value than for peptide 2). The surprising result that the most structural unstable peptide 2 was a better binder to the ligand compared to peptide 3 and 4 and that its on-rate increased with temperature, was further analysed using CD.

The temperature dependence of peptides in buffers studied by CD The effect on temperature for peptide 1, 2 and 4 in HBS- buffer with 0.005 W P20 (BIAcore running buffer) was studied by CD at 222 nm, and for peptide 1 the CD- amplitude was decreasing by increasing temperature, as was found previously for melting of peptide 1 in TFE or phosphate buffer (Figure 20b). The wavelength scans for

peptides 2 and 4 in HBS with P20 and in 10 mM phosphate buffer (from 210-260 nm) revealed almost identical spectra in this region by increasing temperature. They had a lowered CD-amplitude below 210 nm and an increased CD-amplitude from 210-260 nm by increasing temperature, an opposite effect compared to peptide 1 (Figures 25a-c).

The temperature CD-spectra were almost 100 k reversible as illustrated in Figures 25a-c. The concentration dependence of the spectra at 15.50 C and 480 C in phosphate buffer was also tested for peptide 2 up to 620 pM and they were without any significant deviations at 222 nm (data not shown). The wild-type peptide (peptide 1) has also been shown to be very soluble and monomeric (Starovasnik et al., 1997, supra). The MRE at 222 nm was further followed for peptide 2-4 in 10 mM phosphate buffer and HBS by increasing temperature and their respective melting curves are shown in Figures 26a, b.

The sigmoid profiles indicate the cooperativity involved, and the data were fitted to a two-state reversible transition from unfolded to folded conformation. The melting of peptide 3 followed an unusual melting curve composed of increased and decreased MRE at 222 nm. Thus, only the thermodynamic data for peptides 2 and 4 are illustrated in Table 11.

The van't Hoff plots of peptides 2 and 4 in their respective buffers were linear with correlation coefficients on the average of -0.98. Note that peptide 4 had a longer pre-folding area and a larger TM than peptide 2 The energies involved in the transition were higher than for the melting of the peptides in TFE (Table 9).

DISCUSSION By increasing the hydrophobicity in the turn-region the probability for induction of structure in the elastin GVPGVG region should increase with increasing

temperature. The glutamic acid residues at position 25- 26 in the wild-type peptide were therefore both mutated to glutamine. In all, the following residues were altered; the wild-type type I P-turn was replaced with residues which potentially can form type II P-turn, residues which are helix-destabilising (Val, Gly) were introduced, and the charge of the phage display chosen Glu-Glu groups close to the turn were altered to Gln- Gln. Glutamine in the N-terminal position has a negative effect on helix-stability (Chakrabartty et al., 1993, Proc. Natl. Acad. Sci. USA, 90, p11332-11336). In other words the risk of destroying the 20 structure by introducing an elastin sequence into a so small peptide, chosen from phage display, was large.

However, the ambiguity of the elastin turn region has previously been shown. Some reports show that elastin sequences may form both type I and III turn in addition to the well characterised type II P-turn (Arad & Goodman, 1990, Biopolymers, 29, p1651-1668; Bhandary et al., 1990, supra). The apparent complete melting of the secondary structure at lower temperatures is also one of the characteristics of elastin peptides or polymers (Urry, 1988a, supra; 1988b, J. Protein Chem., 7, p81- 114; 1993, Angew. Chem. Int. Ed. Engl., 32, 819-841).

This entropically driven folding implies that the hydrophobic groups of elastin are hydrated and thus do not contribute to structure or stability at lower temperatures.

The elasticity of elastin is described by Urry (1988a, supra) as a librational entropy mechanism. The coupling of certain torsion angles in the elastin sequence induces a large-amplitude rocking motion in the sequence. This implies that the inserted elastin sequence in the IgG-peptide should have a large entropy due to internal chain dynamics, and the coupled change

of torsion angles in the elastin turn by temperature should induce structural changes in the whole sequence.

Indeed, the substitution of the type I turn in the short IgG-binding peptide by the more hydrophobic elastin turn completely melted the secondary structure of the wild-type peptide in phosphate buffer. However, even the wild-type peptide is unstable per se and was later on stabilised by a disulphur-bridge at the termini (Starovasnik et al., 1997, supra).

We needed to induce structure in the peptides by TFE in order to classify them. By using higher concentrations of TFE (> 30 W (v/v)) it was also possible to obtain improved melting curves with more melting data of the different peptides. However, the risk of destroying some of the existing helicity in the wild-type peptide was present as was seen for this peptide at concentrations above 20-25 t. There was a correlation between the cooperativity of the melting curves and their respective AS and SH-values at different concentrations of TFE. Melting curves at increasing concentrations of TFE had a lowered cooperativity and AH-values, as was also recently found by Luo & Baldwin (1997, supra). This also followed earlier observations that the slopes of melting curves at TM are related to AH (Applequist, 1963, J. Chem. Phys., 38, p934-941).

This implies that for peptide 1 in 25 mM SDS which had the shallowest slope (Figure 20b) should also have a very small AH value. The lowered AH may therefore also be linked with the hydrophobicity of the solvent and its interaction with the peptide and not only with the loss of H-bonded water-molecules associated with the peptide.

However, in addition to the study by Luo & Baldwin (1997, supra) on the effect on AH, we show that also AS was decreasing in a similar manner with increasing concentrations of TFE. This further supports their

suggestion that TFE is stabilising the a-helix by strengthening its intramolecular H-bonds. However, the increased entropy from the loss of peptide H-bonded water molecules was not seen, probably because they were replaced by more molecules of TFE and water, and with an additional lowered energy of the TFE-water mixture (Rico et al., 1986, Biopolymers, 25, p1031-1053).

Stabilising of a-helices in TFE offer a way to express the difference of intramolecular stability between mutants and wild-type a-helical peptides in which the helicity of the mutants is completely destroyed. This was shown for the mutant with the most hydrophobic turn (peptide 2) which had a lowered transition temperature by 480 C for melting of a-helix in 30 % TFE compared to the wild type peptide. The other peptides with elastin- turn but with the Glu-Glu sequence close to the turn (peptides 3 and 4) were more helical in TFE and their Th's were 'only' lowered by 33-350 C. There was a helix stabilising effect found from the Tx's of the a-helix with an amidated C-terminus - peptide 3 (Table 9). This has also previously been shown for a-helices (Fairman et al., 1989, Proteins: Struct., Funct. Genet., 5, pl-7).

The CD-amplitude for peptides 2-4 in phosphate buffer or in HBS with P20, was increasing in the 222 nm region by temperature, which is characteristic for the formation of an a-helix or type I p-turn (Perczel et al., 1993, supra; Woody, 1995, supra). However, the magnitude of the CD-signal at 222-nm for the melting curves in buffer was much lower than expected for a transition into an a-helix, and more similar to the lowered intensity from a type I (3-turn (Woody, 1995, supra).

The CD-signal for this transition was dominated by the larger number of residues adopting a random coil

conformation at higher temperatures, and thus does not shift into the typical a-helix spectra by increasing temperature (Perczel et al., 1993, supra). The stability of all the peptides in TFE showed that the a- helical content should decrease by increasing temperature, as was also found for peptide 1 in phosphate buffer. This was however not shown for peptides 2-4 in phosphate buffer which had an increased CD-amplitude at 222 nm in phosphate buffer by temperature. The cooperativity involved for the melting curves was larger than observed for the melting of a- helices. Peptide 4 which had a larger a-helix propensity in TFE than peptide 2, should have a lower TM for the transition at 222 nm than peptide 2. However, the opposite effect was found. The concentration independence at two different temperatures and the full reversibility involved in this transition suggest that this was an intramolecular transition. Thus, the effect at 222 nm for peptides 2-4 was probably due to a temperature induced type I turn.

This turn structure has also been found in elastin sequences (Arad & Goodman, 1990, supra; Bhandary et al., 1990, supra) although the -Pro-Gly- type II turn is between 0.2 - 1.7 kcal moll more stable than type I - turn and it is also the more common turn in elastin (Urry, 1993, Angew. Chem. Int. Ed. Engl., 32, p819-841; Yang et al., 1996, supra and references herein). Type I turn is the evolutionary, phage display selected turn between helix I and helix II in the structural minimised peptide (Starovasnik et al., 1997, supra). This type of turn may ensure that the helices are correctly aligned for binding although they are melted. The are also reports from nature about structural transitions induced by binding (Daughdrill et al., 1997, Nature Struc.

Biol., 4, p285-291; Uesugi et al., 1997, Science, 277, p1310-1313). A structural switch which induces

alignment of helices may help to increase the a-helicity later on upon binding to ligand. peptides with apparent no secondary structures from CD- spectra are commonly assumed to be a 'random coil' peptides. However, by a slightly perturbation in solvents or induction by binding the suppressed structure may be revealed (Waterhous & Johnson, 1994, Biochemistry, 33, p2121-2128). Thus random coils are not 'random' but do contain a large amount of hidden 20 structure which may or may not be induced by solvents/binding (Siligardi, 1996, Biopolymers). There is also a clear correlation between the effect of charges in the elastin peptides or polymers and their respective transition temperatures (see Examples 1 and 2; Urry, 1993, Angew. Chem. Int. Ed. Engl., 32, p819- 841), and the results in this work followed this trend.

Peptide 2 with the most hydrophobic turn region had the lowest TM for the temperature induced transition at 222 nm in phosphate buffer compared to peptide 3 and 4. The latter peptide has two more negative charges, Glu-Glu, close to the turn region relative to peptide 2 (Gln- Gln). According to that observed in elastin sequences, a more charged peptide will increase the probability of hydration and thus also increase the TM of hydrophobic interactions. The energies involved with this transition (Table 11) were also surprisingly similar to that found for short elastin peptides. For those (8 to 21 mer elastin peptides) the average AH and AS for a similar transition involving the formation of a type II turn were respectively 16.8 kcal mol-l and 0.056 kcal mol-- K-l (based on the results of Examples 1 and 2). And again the differences in AH and AS found here compared to those found previous for the melting of a-helices (Table 10) were larger, suggesting that hydrophobic interactions involving temperature induced transition

into type I turn was the most probable structure formed upon heating.

The transition from OOC to 600C of peptide 2 in HBS with P20 involved a free energy difference of 3.0 kcal mol-l.

This is the same energy cost which is required to form a turn (2-4 kcal mol-l) but is below this for the initiation of an helix (4 kcal mol1) (Yang et al., 1996, supra, and references herein) The activity of all peptides was studied by surface plasmon resonance by first deciding which kinetic model was appropriate to use. The better fit to a heterogeneous two-component binding kinetics model using BIAevaluation 2.1 for all peptides may be explained if either the analyte or ligand exists in several conformations. The ligand used in this study was a purified Fc of a monoclonal mouse IgG2a (561), but the peptides 2-4 in this study were all without any helical structures in HBS and likewise for peptide 1 the helicity decreased with temperature. However, it was possible to fit peptide 1 to a one-component model at the lowest temperature in which the stability of the 2" structure was at its local peak level. The latest observations that a 46-residue segment of protein A (B-domain) may fold through two different helix conformations which have different H-bond patterns (Guo et al., 1997, Proc. Natl. Acad. Sci.

USA, 94, p10161-10166), suggest that different intermediate conformations may also exist for the unfolded protein A derived 33-mer peptides in this study.

By increasing temperature the proportion of unfolded peptides (peptide 1, control) increased and made up a larger proportion of a heterogeneous population.

The increased on-rates for peptide 2 followed the CD-

melting curves in 10 mM phosphate buffer pH 7 or HBS with P20, displaying an increase in 20 structure with temperature (Figures 23a, b). Peptide 1 had a correlation between decreasing on-rates versus a decrease in 20 structure by rising temperature. This shows that the on-rates followed the 2" structure of the peptides. Previous reports have shown that there is a good correlation between alpha helix content and binding strength (Jansson et al., 1997, Biochemistry, 36, p4108- 4117, and references herein) . However, the off-rates for both peptides were quite similar (and large) and both slightly increased by elevating the temperature.

This indicates that there are similar intermolecular dissociation forces for both peptides. This further suggest that peptide 1 and 2 have gained similar structure after binding, or that the effect arises from structural changes in the immobilised ligand.

The temperature induced binding and structure responded similarly on the changing the charges on Glu residues close to the turn (peptide 2 (-GVPGVGQQ-) versus peptide 3 and 4 (-GVPGVGEE-)). Although peptide 3 and 4 had a larger propensity for forming alpha-helices than peptide 2 due to the helix- destabilising QQ-residues in peptide 2 (compare melting curves of peptide 2, 3 and 4 in 60 k TFE, Chakrabartty et al., 1993, supra), they do not have the better binding to the ligand. For peptide 2 the more hydrophobic groups in the turn region interact and induce a structure (type I 13turn) with a lower TM than peptide 4, which may affect the intramolecular proximity of the two melted helices or the position of the residues involved in binding. The binding energies involved (peptide 2 relative to peptide 1) with the negative slope of AAG by increasing temperatures arises from observation that the temperature induced folding of peptide 2 compared to

melting of peptide 1 are both entropically driven.

By engineering a peptide or protein which has a large enough negative slope AG/T, may be produced.

The introduction of an inducible switch mechanism into the A protein as described herein suggests that the wild-type structure is altered in such a manner that the initial stability is lost. By introducing elastin- switches into the structure which can fold into a stable structure at higher temperatures the activity and stability can be regained again. It is therefore appropriate to use the elastin sequence also between helices. Therefore, by changing the Protein A turn from the type I turn to the elastin (3-turn with the VPGVG helix breaking sequence, we initially melted the alpha helices and introduced a novel thermo induced switch back to type I turn. This may therefore offer a novel temperature regulation for protein A and also other helical proteins in which lowering of the on-rates at lower temperatures and increasing them at higher temperatures regulates binding.

Table 8. Sequences of minimised Protein A with elastic sequences inserted Noa Peptide 1 Peptide 2 Peptide 3* Peptide 4 (Protein A (Mutant 1) (Mutant 2*) (Mutant 3) Control) A5 PHE PHE PHE PHE A6 ASN ASN ASN ASN A7 MET MET MET MET A8 GLN GLN GLN GLN A9 GLN GLN GLN GLN A10 GLN GLN GLN GLN All ARG ARG ARG ARG A12 ARG ARG ARG ARG A13 PHE PHE PHE PHE A14 TYR TYR TYR TYR A15 GLU GLU GLU GLU A16 ALA ALA ALA ALA A17 LEU LEU LEU LEU A18 HIS GLY GLY GLY A19 ASP VAL VAL VAL A20 PRO PRO PRO PRO A21 ASN GLY GLY GLY A22 LEU VAL VAL VAL A23 ASN GLY GLY GLY A24 GLU GLN GLU GLU A25 GLU GLN GLU GLU A26 GLN GLN GLN GLN A27 ARG ARG ARG ARG A28 ASN ASN ASN ASN A29 ALA ALA ALA ALA A30 LYS LYS LYS LYS A31 ILE ILE ILE ILE A32 LYS LYS LYS LYS A33 SER SER SER SER A34 ILE ILE ILE ILE A35 ARG ARG ARG ARG A36 ASP ASP ASP ASP A37 ASP ASP ASP ASP No. 33 33 33 33 Res * This peptide has an amidated C-terminus, the other peptides have free N- and C-termini. a Amino acid number in the protein A sequence Table 9: Thermodynamic values for melting of peptides in tri fluoroethanola TM (°C) AH AS (kcal.mol-1) (kcal.mol-1.K-1) Peptide 1 Phosphate buffer 13 13.1 + 0.4 0.046 + 0.001 20 % TFE 48 10.7 + 0.7 0.033 + 0.002 30 k TFE 48 11.9 + 0.4 0.037 + 0.001 60 W TFE 44 9.1 + 0.1 0.0286 +0.0005 Peptide 2 30 W TFE 0 8.7 + 0.6 0.032 + 0.002 60 W TFE 9 6.8 + 0.2 0.0241 +0.0006 Peptide 3 30 k TFE 18 6.4 + 0.2 0.0219 +0.0007 60 W TFE 23 6.2 + 0.1 0.0211 +0.0005 Peptide 4 30 k TFE 8 9.7 + 1.0 0.035 + 0.004 60 W TFE 23 7.8 + 0.1 0.0263 +0.0005 a The Mean Residual Ellipticity was followed at 222 nm for each peptide and the fitted to a reversible van't Hoff equation with ACp =0 allowing AH, AS and the post- transitional endpoint to float. The pre-transitional endpoints were fixed as this value initially fitted for melting of peptide 1 in similar solution.

Table 10: Kinetic parameters for the interaction between peptides and Fc of the mouse<BR> monoclonal IgG2a 561 (αCD34) at +10° C measured by surface plasmon resonance a<BR> koff fast (s-1) Kd fast (µM-1) koff slow (s-1) Kd slow (µM-1)<BR> Peptide 1 0.109 # 0.004 0.69 # 0.03 0.0322 # 0.0004 0.203 # 0.004<BR> Peptide 1b 0.044 0.24<BR> Peptide 1c 0.041 0.185<BR> Peptide 2 0.39 # 0.08 953 # 207 0.0349 # 0.004 86 # 11<BR> Peptide 3 0.7 # 0.1 3194 # 1775 0.10 # 0.04 493 # 325<BR> Peptide 4 0.56 # 0.06 1948 # 638 0.0212 # 0.001 73 # 23<BR> a The fast and slow off-rates (koff) were determined from the a fit of heterogeneous binding<BR> kinetics of analytes to ligand (1000 RU) using a flow rate of 30 µL HBS with 0.005 % (v/v)<BR> surfactant P20 per minute, and by correcting for bulk effects using immobilised HSA. The<BR> dissociation constants (KD) were based on the fast on-rates (Figure 23b). Uncertainties<BR> are given as # S.E, and those for KD were manually calculated from the relation KD = koff/kon<BR> based on the S.E's for kon and koff.<BR> b Peptide 1 was fitted to a single homogeneous 1:1 binding at +10°C which gave an on-rate<BR> of 1.87 x 10-5 M-1 s-1 which was further used to calculate the KD value shown above.<BR> c Starovasnik et al. (1997, supra) for binding to monoclonal IgG1; kon is 2.2 x 10-5 M-1 s-1.<BR> <P>Temperatures for experiments in c is not given. The fit in c is based on an 1:1<BR> interaction between peptide and ligand.

Table 11: Thermodynamic data for thermal transition at 222 nm for peptides 2 and 4 in buffers" TM (°C) AH AS (kcal.mol-1) (kcal.mol-1.K-1) Peptide 2 in 26 19.0 + 1.6 0.063 + 0.005 phosphate buffer Peptide 2 in 31 15.4 + 1.9 0.051 + 0.006 HBS with P20 Peptide 4 in 45 18.5 + 1.5 0.058 + 0.005 phosphate buffer Peptide 4 in 39 13.8 + 1.9 0.044 + 0.006 HBS with P20 GVG(VPGVG) peptidesb 18.2 0.062 The melting data at 222 nm was fitted to an reversible van't Hoff equation allowing AS, AH and the endpoints for transition to float. The heat capacity was fixed at 0. b The values are the average of short GVG(VPGVG)n peptides (n=l and 3) in phosphate buffer pH 7.0.

EXAMPLE 4: Engineering of an anti-gp41 ScFv 3D6, with an elastic linker The sequence of the scFv 3D6 (anti HIV1-gp41, Engelhardt et al., 1994, BioTechniques 17, p44-46, genebank ref. U35316) was subcloned from the plasmid pDAP2 (Kerschbaumer et al., 1996, Immunotechnology 2, p145-150) into the plasmid pUC119His6mycXba (Griffiths et al., 1994, EMBO J. 13, p3245-3260). In the former plasmid (Figure 27a), the scFv is expressed as an alkaline phosphatase fusion protein with a pelB leader sequence for periplasmic expression and a His6-tag for purification via metal affinity chromatography. The latter plasmid (Figure 27b) also includes a c-myc tag for positive identification/purification using the antibody 9E10 (Hoogenboom et al., 1991, Nucleic Acids Res. 19, p4133-4137). Both plasmids are regulated by the LacZ promoter and carries an Ampicillin resistance gene, and the 3D6 mRNA and protein is synthesised by using 1 mM of the inducer isopropylthio-galactoside (IPTG).

The 3D6 scFv plasmid with SGGGGSGGRASGGGGS linker (Figure 28) is used as the starting point for engineering scFv's with elastin-based linkers as follows: Old scFv linker sequence: -SGGGGSGGRASGGGGS- New scFv linker sequences: -X[VPGVG]nY-, in which n = 1 to 10 and X, Y may be the same or different and may be any amino acid residues, preferably a small aliphatic amino acid residue such that flexibility of the linker is retained.

Specifically in this example we have designed the following linkers to be cloned into 3D6 scFv.

(I) -S[GVG(VPGVG)VPG]S- ll-mer linker (Il) -S[GVG(VPGVG)4VPG]S- 26-mer linker (III) -S[GVG(VPGVG)WPG]S- 41-mer linker The first linker region (I) was cloned as described in Figure 29a-c. Figure 29a shows the structure of 3D6 gene in the pDAP2 plasmid and in Figure 29b the gene has been subcloned into pUC119His6mycXba using the restriction sites HindIII and NotI. Futher PCR- mutagensis was carried out using primer combination LMB3(B) & VHLINK(F) and PHENCO(F) and VLLINK(B) as described in Figure 29c and 30. This step gave rise to new restriction sites being introduced (labelled * in Figure 30). The positions of primers that were used to clone, screen and sequence the 3D6-gene are illustrated in Figure 29c. The assembly of the 3D6 gene with the new linker construct is shown in Figure 30b. The initial ll-mer linker (I) was expanded by successive restrictions and ligation of annealed VPGVG-coding oligos (Figure 31 and Figure 32). The assembly of the 26-mer linker region (II) is shown in Figure 31 and was assembled by ligation of annealed oligos with two open restriction sites (labelled ** in Figure 31) into linker region (I) of the gene (shown as bold and underlined text in Figure 31). A new XmaI restriction site was introduced during this procedure and is labelled (*) in Figure 31. The assembly of the the 41-mer linker region (III) is shown in Figure 32 and was assembled by ligation of annealed oligos with two open restriction sites (labelled ** in Figure 32) into the restricted linker region (II) of the 3D6-gene in the plasmid (shown as bold and underlined text in Figure 32). The structure of the final 41-mer linker region (III) is illustrated in Figure 33b.

Subcloning of the 3D6-gene into pUC119His6mycXba.

The freeze dried pDAP2-3D6 plasmid was dissolved in sterile purified water and heat transformed (1-2 minutes at 42"C) to CaCl2 competent TOPP2 cells and screened on 2 x YT plates with 100 g/ml Ampicillin. Colonies were further grown in 4 ml LB with 100 g/ml Ampicillin over night at +37"C and 250 rmp. Samples of the bacterial culture were taken as glycerol stocks (stored in LB with 15 W Glycerol at -700C) and the pDAP2-3D6 plasmid was purified on Hybaid RecoveryTM Quick Prep Mini Kit (Hybaid, Middlesex, UK) or WizardTM Plus Minipreps DNA Purification Systems (Promega, Madison, WI). The plasmids (pDAP2-3D6 and pUC119His6mycXba) were identified on 1.0 W agarose gel. Both the pDAP2-3D6 and pUC119His6mycXba plasmids were separately restricted using 10U of Not I and 20U of HindIII (New England Biolabs, NEB) in buffer system 2 (NEB) with 100 g/ml BSA at 370C for 16 hours, and then heat inactivated for 20 minutes at 65 0C. The 3D6-gene and the open pUC119His6mycXba plasmid were isolated on 1 W (w/v) low melting point (LMP) agarose (Gibco) at +40C using a Tris-Acetic Acid-EDTA buffer system (TAE) pH 8.0, detected by a long wavelength (> 300 nm) W-radiation lamp and cut out of gel. The 3D6-gene was purified on MAGICTM- PCR Preps DNA Purification System (Promega, Madison, WI) and the open plasmid was purified using Qiaquick Gel Extraction Kit (Qiagen, Crawley, UK). The products were analysed on 1 % agarose gel and the 3D6 gene was ligated into open pUC119His6mycXba using 400U of T4 DNA Ligase (NEB) in provided T4 ligase buffer at 16"C over night. A control ligation with only open vector was performed. The ligation mixture (10 /21) was heat transformed to CaCl2 competent TG1 cells and incubated on 2xTY agar plates over night at +37"C.

Positive transformants were toothpicked and identified by PCR-screening in a reaction mixture of 20ul using 10 pmol of each of the primers LMB3(B) and Phenenco(F) (Fig. 34), 200 M of each nucleotide (Pharmacia or Bioline), and buffer with MgC12 and Taq-polymerase purchased from Bioline. The PCR-screening was performed on a thermocycler (MJ Research) with the programme: Initially 94"C for 5 minutes, then 30 repeating cycles of the sequence 94"C (30 seconds), 42°C (30 seconds) and 72°C (1 minute). The programmed ended with 720C for 5 minutes and cooling to 40C. PCR-products were analysed on 1 or 1.5 W agarose gel in Tris-Borax-EDTA buffer system (TBE), and positive colonies were grown in an over-night culture of 4 ml LB with 100 g/ml Ampicillin.

Glycerol stock were taken and plasmids were isolated was purified on Hybaid RecoveryTM Quick Prep Mini Kit (Hybaid, Middlesex, UK). The 3D6 gene from three positive colonies with the plasmid pUC119His6mycXba was sequenced in both directions using the primers LMB3(B) and Phenenco(F). An automatic fluorescent dye- terminator sequencing method (377 Perkin-Elmer sequencer) was used. One mutation at the N-terminus (the amino acid E to D) was detected on all three clones and some silent mutations.

Cloning of Linker Construct (I) The first linker construct (ll-mer; no. I) was inserted using PCR amplification with high fidelity Expand polymerase (Boehringer Mannheim) in a two reaction mixtures (Figures 29c and 30). Each contained the forward and back primer combination VHLINK(F) and LMB3(B) covering the VH-region of scFv with a new linker, or VLLINK(B) and PHENECO (F) for the VL-region of scFv with the new linker (Figure 29c). The primers (15 pmol of each), 200 yM of dNTP's (Pharmacia), 1 l of a 1:10 dilution of template (3D6-PUC119His6mycXba with old linker) and with the supplied Expand buffer were incubated on a thermocycler with the Expand programme: 94"C (2 minutes), 10 cycles of the sequence 94"C (15 seconds), 43"C (30 seconds) and 72"C (45 seconds), and then 15 cycles of the sequence 94°C (15 seconds), 43°C (30 seconds) and 72°C (45 seconds initially which extends 20 seconds per cycle). The programme finishes with 72°C for 5 minutes and cooling down to 40C. The PCR-products for VH and VL were analysed on 1.5 W agarose gel in TBE-buffer and revealed respectively one clear band for each product, VH (just above 500 bp) and for VL (just below 500 bp). The products were purified on MAGICTM PCR Preps DNA Purification System (Promega, Madison, WI). The PCR-product for VH should have a new linker with KpnI, EcoNI and XhoI sites. The VL should likewise have the sites KpnI, EcoNI and BamHI on the new linker. The sequence covering KpnI and EcoNI sites is overlapping in both PCR-products for VH and VL. The sites KpnI and EcoNI were each tested by restriction and T4 ligation using the respective NEB enzymes and protocols. The ligation products were analysed on 1.5 W agarose gel (TBE system) and revealed the KpnI gave a much stronger band around 1000 bp (3D6 scFv) than EcoNI.

Both PCR-products were thus restricted with 20U KpnI according to NEB's protocol. The product was purified on MAGIC>- PCR Preps DNA Purification System (Promega, Madison, WI) to remove the small digests (10-16 bp) and the VH and the VL were further ligated with T4 ligase over night at +160C as described previously. The ligation mixture was heat inactivated (65"C for 10 minutes), purified on MAGICTM- PCR Preps DNA Purification System and further restricted with 40U Hind III and 20U Not I as described previously. The digestion mixture after heat inactivation at 650C (20 minutes) was run on a LMP 1.5 k agarose gel (TAE-system) at +40C. The respective area corresponding to the MW of the gene was cut out of gel and P-agarase I digested at +40"C over night according to NEB procedure. The DNA was isopropanol precipitated at -200C for 2 hours, dried, resuspended in 20 l sterile purified water and analysed on 1.5 W agarose gel (TBE). Three bands around 1000 bp were detected, possibly from VH-VH, VL-VL and VH-VL assemblies. However, only VH-VL may be successfully inserted into the open vector. The three isolated bands were T4 ligated (NEB) into Hind 111/Not I restricted vector pUC119His6mycXba and heat transformed (90 seconds at 42"C) into competent TG1 cells. Colonies were PCR- screened and isolated as described previously. The isolated plasmids from the colonies were tested for the new sites XhoI and KpnI using their respective restriction procedure described by NEB. Both for XhoI and KpnI the plasmid was restricted. The plasmid was further sequenced back and forward as described previously and the new linker was positively identified.

Cloning of Linker Construct (II) The linker was assembled as described in Figure 31. The plasmid coding for the linker construct (I) was first restricted with 80 U XhoI in NEB 2 buffer and then further restricted with 40U KpnI in NEB 1 buffer. Both restrictions were performed with 100 g/ml BSA at +370C over night, and the restriction mixtures after each incubation were purified on Wizards DNA Clean-Up System.

The ss-oligos LINK1(F) and LINK1(B), both of 150 pmol, were separately kinased in T4 DNA ligase buffer with T4 polynucleotide kinase for 60 minutes at 370C. The kinase mixture was heat inactivated at 650C for 20 minutes and then annealed by first incubating a mixture of 25 pmol of each oligo for 5 minutes at 940C to completely dissociate the oligos. This oligos were then annealed slowly by cooling from 94"C to 20oC using a descending temperature slope of 1"C per minute on a thermocycler. The ds-oligos were freeze dried and redissolved in ice-cold sterile purified water and a 100-fold molar excess of ds-oligo was added to restricted vector and ligated over night at +160C using T4 DNA ligase (NEB). The ligation mixture was heat- inactivated (65"C, 10 minutes) and added to competent TG1 cells and grown over night at +370C on 2xTY plates with 100 g/ml Ampicillin. Positive colonies were identified by PCR-screening with either the primer combination LMB3(B) and Phenenco(F), or INT1(F) and INT1(B) which covers approximately 220 bp comprising the linker region (Figure 29c). The latter primer pair gives a better resolution of the increased size of the new linker. The sequence of the plasmid was also verified by XmaI digestion and sequencing in both directions.

Cloning of Linker Construct (III) The plasmid with the gene with the linker construct (II) was restricted with 40U XmaI and 40U KpnI in NEB 1 buffer with 100 g/ml BSA at +37"C over night, and the restriction mixture was purified on Wizard DNA Clean-Up System. The similar procedures for kinasing, annealing and cloning of the ss-oligos LINK2(F) and LINK2(B) into XmaI/KpnI restricted vector as for construct of linker (II) were followed (Figure 32). Clones were identified as described for Linker construct (II). The structure of linker region (III) is shown in Figure 33.

Expression and purification of 3D6 scFv The constructs can either be expressed as pure scFvs with c-myc and His6 tags in the vector pUC119His6mycXba, or they can be expressed as fusion proteins of alkaline phosphatase with only a His6 tag in the plasmid pDAP2 (by subcloning). The plasmid with the target gene was heat transformed to the Escherichia coli strain TG1, and the expression of the clone was achieved in a two-litre Erlenmeyer flask with 500 ml M9ZB medium with 1 W (w/v) glucose (added ampicillin 100 Ug/ml). The inoculated freshly transformed colony was grown to A600 of 1.0 to 2.0 after 12-18 hours at +370C (250 rpm) in a incubator shaker (New Brunswick Scientific Co., Inc., Edison, NJ).

Before induction the cells were centrifuged (4500 x 10 minutes at room temperature) and resuspended in M9ZB medium with 1 W glycerol prewarmed to 300C. Expression of 3D6 is induced by lowering the temperature to around +200C, adding 1 mM IPTG (final concentration) and by further incubation for 3-24 hours.

After exposure of the cells to osmotic shock, the cells were harvested by centrifugation (Sorvall, 5400 rpm, 4"C, 12 minutes) and the pellets were dissolved in osmotic shock lysis buffer (100 ml of 20 % sucrose, 30 mM Tris, 1 mM EDTA) and shaken vigorously for 10 minutes at room temperature before recentrifugation (8500 rpm, 4"C, 30 minutes). After similar treatment of the pellet in 100 ml 5 mM MgSO4 (40C, 10 minutes shaking), the periplasmic lysate was extracted and dialysed in a Spectropore tubing.

The expressed 3D6 scFv was purified from periplasmic lysate with metal affinity chromatography (Qiagen, Ni2 or Zn2+ column). The fractions were concentrated and analysed on 12 W SDS-polyacrylamide gels and Western electroblotted to a nitro-cellulose membrane. Detection and purity was carried out with the 9E10 mouse antibody (c-myc) and horseradish peroxidase conjugated goat-anti- mouse antibody (Sigma) diluted to 0.1 k (v/v) in 2 W (w/v) milk-powder PBS. The 3D6-alkaline phosphatase fusion protein was detected directly through the activity of the alkaline phosphatase part (enzyme assay). The concentration of 3D6 was spectrophotometrically determined at 280 nm using the method by Gill and von Hippel to calculate the

extinction coefficient #0.1 % from the primary sequence (Anal. Biochem., 1989, 182, p319-326).

The activities of the 3D6 scFv were analysed using the fusion protein GST-epi41 which has a gp41 epitope (short peptide) immobilised on the surface of GST (Glutathion Transferase). GST-epi41 was immobilised on through its primary amines onto a gold-chip and the binding to 3D6 scFv was analysed using Surface Plasmon Resonance (BIAcore 2000, Biasensor, Sweden) at different temperatures. The kOn-rates was determined from at least 5 different concentrations and the kOff-rates was determined by using the highest concentration of 3D6.

EXAMPLE 5: Engineering elastic sequences into the primary structure of a minidomain of Protein A containing a disulphur bridge Two more peptides were synthesised - peptides 5 and 6 (Figure 35) and studied as described in Example 3.

Peptides 5 and 6 behaved in a similar way to the peptides studied in Example 3. Peptide 6 with the elastic sequence increased its CD-amplitude at 222 nm by increasing temperature, and likewise in the control peptide the alpha helices melted with increasing temperature. In addition, by adding sodium sulphate Peptide 6 formed what is believed to be a type I P-turn as shown by CD-spectroscopy (Figure 36).

The peptides where further studied using surface plasmon resonance (BIAcore 2000) by immobilising 3370 RU of Fc-561 (as described previously) with HSA (3717 RU) on a control CM-5 chip surface using an amine coupling kit supplied by the manufacturer (BIAcore). Figure 37 illustrates the effect of temperature on the peptides 5 and 6. This is further illustrated in Figure 38 which also gives the

different kinetic parameters (similar to those described previously in Example 3 but showing a somewhat larger effect).

RESULTS The affinity of peptide 6 (with the VPGVG-turn) towards Fc- 561 increased with temperature approximately 20-fold between 50C to 370C. This is in contrast to peptide 5 without the VPGVG-turn (Figure 38c).

Peptide 6 behaved structurally like the peptides studied in Example 3 in that it formed a type I P-turn with increasing temperature.