BACKGROUND OF THE INVENTION Amyloid fibrils are highly ordered protein aggregates associated with a range of diseases including Alzheimer's disease, type II diabetes, systemic amyloidosis and transmissible spongiform encephalopathies (Tan 1994; Ghetti 1996; Kelly 1997; Cohen 1998). Fibrils may be observed by electron microscopy, where they appear as long unbranched structures with a diameter of 7 to 12 nm, typically about l Onm.

When examined by Fourier-transform infrared (FT-IR) spectroscopy and X-ray fibre diffraction, amyloid fibrils show a cross-p diffraction pattern, indicative of (3- structure content.

We have previously found that proteins unrelated to amyloid diseases have been found to convert under mild denaturation conditions in vitro into fibril structures and display the same characteristics as disease-associated fibrils. Thus it has been concluded that the ability to form fibrils is a common property of all polypeptide chains (Guijarro, 1998; Chiti, 1999).

The formation of aggregates, in particular fibrils, by a protein can have advantages or disadvantages dependent on the role or application of the protein.

Aggregation may be advantageous, for example for use of the aggregate as a plastic, in electronics, for catalysis, or as a slow-release form of proteins, particularly if the released proteins are pharmaceutically active. However, the formation of aggregates may be disadvantageous for many reasons, for example as a symptom or cause of amyloid diseases, or when used at concentrations or under conditions desirable for physiological activity, therapeutic administration or industrial application. It is therefore highly desirable to modify proteins such that the tendency of the protein to

aggregate is tailored to the role or application in hand.

Aggregation occurs when proteins in their native (globular) state denature (break intramolecular bonds), causing their polypeptide main chain and hydrophobic sidechains to become exposed. This provides an opportunity for hydrogen bonds and other interactions to form between denatured protein molecules (formation of intermolecular bonds) resulting in the formation of an aggregate. Amyloid diseases are associated with conditions which destabilise the native state (Chiti, 1999) and thus favour aggregation (Dobson, 1999). In particular, it has been shown that mutant proteins associated with amyloidoses of lysozyme (Booth 1997; Canet 1999), immunoglobulin light chain (Hurle, 1994) and transthyretin (McCutchen, 1995) have changes in amino acid sequence causing a change to the overall stability of the protein.

Wildtype protein activity and regulation can be dependent on the stability of the native state of the wildtype. Proteins have evolved to act optimally at physiological conditions. Under such conditions, the native state is only marginally stable relative to denatured states and thus wildtype proteins exist at a dynamic equilibrium as populations of proteins with native states, partially denatured states and fully denatured states (Dobson, 1999). Proteins with altered native stability may disturb this equilibrium, and therefore, exhibit aberrant activity or regulation compared to the wildtype protein.

SUMMARY OF THE INVENTION Against this background, the inventors have made the surprising discovery that the tendency of a polypeptide to aggregate can be modified without substantially altering the global folding of the native state. This means that modified polypeptides can be produced with the dual advantages of (i) a tailored tendency for aggregation (as suited to the role or application in hand) and (ii) wildtype-like native state stability, activity and regulation.

Accordingly the invention provides A method of designing a modified polypeptide having a desired altered tendency to aggregate compared to the

unmodified polypeptide, which method comprises: analysing the amino acid sequence of a predetermined polypeptide to determine the propensity of the polypeptide to form local structure ; comparing the propensity to form local structure of a modified polypeptide to the propensity to form local structure of an unmodified polypeptide; determining thereby whether the modified polypeptide has an altered tendency to aggregate in the denatured state relative to the unmodified polypeptide; and producing a selected modified polypeptide having the desired altered tendency to aggregate.

The invention also provides a method of producing a modified polypeptide having an altered tendency to aggregate compared to the unmodified polypeptide, which method comprises: (i) introducing at least one amino acid modification into a predetermined polypeptide sequence such that said modified polypeptide has an altered propensity to form local structure in the denatured state relative to the unmodified polypeptide, and optionally (ii) recovering the modified polypeptide, and/or optionally (iii) allowing the modified polypeptides to form an aggregate.

The invention also provides a method of producing a polypeptide having an altered tendency to aggregate, and in particular to form non-covalent aggregates which method comprises introducing amino acid sequence modifications into the polypeptide such that: (i) the denatured state has an altered propensity to form local structure, relative to the unmodified polypeptide, and (ii) the native state exhibits a global native folding substantially equivalent to that of the unmodified polypeptide, and (iii) recovering the modified polypeptide and/or optionally (iv) allowing the modified polypeptide to form an aggregate.

The invention also provides: -a method of producing a polynucleotide that encodes a polypeptide of the invention, its complement, or a fragment of either sequence, which method comprises either :

(a) modifying a polynucleotide that encodes the unmodified polypeptide such that its modified nucleotide sequence encodes an modified polypeptide of the invention or (b) synthesising a polynucleotide that encodes an modified polypeptide of the invention.

-A method of producing a vector for use in the expression of a polypeptide of the invention or the cloning of a polynucleotide of the invention, which method comprises inserting a polynucleotide of the invention into a vector.

-A method of generating a cell, cell culture or multicellular organism that comprises or is transformed with a polynucleotide of the invention or a vector of the invention, which method comprises: (a) transforming a cell with a polynucleotide of the invention or a vector of the invention, and optionally; (b) culturing a cell obtained by (a) under conditions that permit the selective growth of the transformants ; and optionally; (c) culturing a cell obtained by (a) or (b) under conditions suitable to allow the production of a multicellular organism.

-A method of producing a polypeptide of the invention, which method comprises either: (a) maintaining a cell, culture, or organism of the invention under conditions suitable to allow the production of the polypeptide of the invention, and isolating the polypeptide; or (b) chemically synthesising a polypeptide of the invention.

-A method for the production of an aggregate comprising contacting polypeptides of the invention under conditions that permit nucleation.

-A polypeptide, polynucleotide, vector, cell, cell culture, organism and aggregate of the invention.

-Use of polypeptide amino acid sequence modifications that cause the denatured state of the polypeptide to have an altered propensity to form local structure relative to the unmodified polypeptide, and the native state exhibits a global native folding substantially equivalent to that of the unmodified polypeptide to alter

the tendency of the polypeptide to aggregate.

-A pharmaceutical composition comprising a polypeptide of the invention, a polynucleotide of the invention, a vector of the invention, a cell, cell culture or organism of the invention or an aggregate of the invention and a pharmaceutically acceptable carrier.

-A polypeptide of the invention, a polynucleotide of the invention, a vector of the invention, a cell, cell culture or organism of the invention or an aggregate of the invention for use in a method of treatment or diagnosis of the human or animal body.

-Use of a polypeptide of the invention, a polynucleotide of the invention, a vector as defined of the invention, a cell, cell culture or organism of the invention or an aggregate of the invention in the manufacture of a medicament for use in the treatment of an amyloid or related disease.

-An aggregate of the invention for use in a method of treatment or diagnosis of the human or animal body, wherein the rate of polypeptide release from the aggregate is modified relative to the rate of unmodified polypeptide release from an aggregate that does not comprise a polypeptide of the invention.

-Use of an aggregate of the invention as an in vitro slow-release form of a polypeptide of the invention, wherein the rate of polypeptide release from the aggregate is modified relative to the rate of unmodified polypeptide release from an aggregate that does not comprise a polypeptide of the invention.

-A method of treating a disease in a patient comprising administering an agent of the invention.

-A polypeptide or aggregate obtained by a method of the invention.

-A method of assessing the tendency of a polypeptide in the denatured state to form aggregates comprising : analysing the amino acid sequence of the polypeptide to determine the propensity of the polypeptide to form local structure; comparing the propensity to form local structure of the polypeptide to the propensity to form local structure of a predetermined polypeptide sequence ; and determining thereby whether the polypeptide has an altered tendency to aggregate in the denatured state compared to the predetermined polypeptide.

-A method of designing a polypeptide having a desired tendency to aggregate comprising: analysing the amino acid sequence of the polypeptide to determine the propensity of the polypeptide to form local structure; introducing a modification to the amino acid sequence and assessing the effect of the modification on the propensity of the polypeptide to form local structure; and designing thereby a modified polypeptide having a desired tendency to form aggregates.

-A method of assessing the risk or susceptibility of an individual to an amyloid or aggregate associated disease comprising identifying the amino acid sequence of a polypeptide expressed in said individual associated with an amyloid or aggregate disease; assessing the tendency of the polypeptide to aggregate according to a method of the invention.

DESCRIPTION OF THE FIGURES Figures 1. Thermal denaturation analysis at pH 3.0 of WT-ADA2h and M1, M2 and DM variants. (a) WT, heating (open circles), WT cooling (open squares), DM heating and cooling (filled circles, the heating and cooling data are superposable). The WT protein was monitored at 214 nm whereas the DM variant was monitored at 222 nm. (b-e) Far-UV CD spectra of WT-ADA2h and variants, at 25 °C (open circles), 95 °C (open squares) and at 25 °C (filled circles) after cooling (b) WT; (c) DM ; (d) M1 ; (e) M2. (f) Amide I band in the FTIR spectra of WT- ADA2h samples before (solid line) and after (dashed line) thermal denaturation. The second derivatives of the spectra reveal peaks centred at 1622 and 1649 cm~'for the sample before heating, and 1615 and 1685 cl-'cafter heating. The band at 1615 cm'' is indicative of (3-strands whereas the one at 1685 cm-'is commonly associated with a splitting in the amide I band due to antiparallel inter-strand interactions (Fabian, 1993 a; Krimm and Bandekar, 1986).

Calculations of the residual helical content at 95 ° C were carried out from CD spectra as indicated in materials and methods with the following results: WT retains 45% of the native helical structure (spectra were collected in a very dilute sample that was subjected to a very fast temperature change to prevent aggregation during data

collection), MI retains 63%, M2 retains 61% and DM retains 68%. Therefore the residual helical structure at 95 ° C for DM is about 50% greater that the amount retained by WT.

Figure 2. Urea renaturation of WT-ADA2h and the DM variant at pH 3.0.

(a) Far-UV CD spectra of WT-ADA2h. (b) Far-W CD spectra of DM-ADA2h.

Circles refer to samples prepared under the initial conditions. Squares refer to the polypeptide after incubating the samples of 4 M urea, and then diluting to 1M urea.

Fig. 3-AGADIR prediction run over the whole sequence of PI3-SH3 in order to identify regions of the protein with a higher intrinsic propensity to adopt a helical conformation.

Fig. 4-Sequence of PI3-SH3. The secondary structure elements (s = strand, 3 = 3, 0 helix) are indicated after analysis of the structure with the DSSP package.

Also regions with an intrinsic helical propensity higher than 2% indicated (according to AGADIR predictions).

Fig. 5-AGADIR prediction showing the intrinsic helical propensities exhibited by the wild type and E17R/D23R mutant sequences.

Fig. 6-Prediction of helical propensities of human PrP helix C residues 198- 227 evaluated using AGADIR. The wild type sequence and those corresponding to the mutant peptides E200K, R208H and V210I are included.

Fig. 7-Far-UV CD spectra of the four Helix C peptides at pH 7.0. Buffer was sodium phosphate 20 mM, p-mercapto ethanol 20 mM.

Fig. 8-TFE titration of the four Helix C peptides monitored by measuring the ellipticity at 222 nm. In all the cases the buffer was 20 mM sodium phosphate and 20 mM p-mercapto ethanol.

Fig. 9-Far-UV CD spectra of Helix C peptides at pH 2.0. Buffer was 20 mM phosphoric acid equilibrated at pH 2.0 by adding HC 1 and 20 mM TCEP. A) Peptide concentration 20 M. B) Peptide concentration 60 M.

Fig. 10-Far-UV CD spectra of Helix C peptides at pH 2.0 after two months of incubation at room temperature. Peptide concentration was 60 pM.

Fig. 11-Time course of the aggregation of wild-type AcP monitored by ThT fluorescence (A), Congo red absorbance (B) and far-UV CD (C). AcP was incubated

under conditions favouring aggregation (25% vol/vol TFE, 50 mM acetate buffer, pH 5.5,25°C) and aliquots were withdrawn at regular time intervals as described in the Materials and Methods section. The three techniques yield very similar kinetic traces and analysis of these produces rate values that are identical within experimental error.

Fig. 12-Time course of the increase of ThT fluorescence upon aggregation for some representative mutants. The mutants shown are I75V (filled circles) (panel A), V20A (empty circles) and Y98Q (filled circles) (panel B) E29D (empty circles) and A30G (filled circles) (panel C). The continuous lines through the data are the best fits to single exponential functions. In each panel the best fit to the wild-type data is represented as a dashed line and is shown for comparison. Experimental conditions are as described in Figure 11.

Fig. 13-Bar graph showing the change in the rate of aggregation resulting from mutation of residues at various positions. Each bar refers to a variant with an amino acid replacement at the position shown on the x axis. Rate data, reported on the y axis, are expressed as the natural logarithm of the ratio between the rate for the mutant and that for the wild-type protein. A value of zero implies that the mutant aggregates at a rate identical to that of the wild-type protein. Values higher and lower than zero involve aggregation rates of the mutants faster and slower than the wild-type protein, respectively. The error bars displayed in the figure for each mutant is given by [(2#vmut)2 + (2#vwt)2] where #vmut and #vwt correspond to the standard errors of ln (vmut) and ln (vwt), respectively (the standard error is defined as the standard deviation divided by the root mean square of the number of measurements). This operation allows the experimental errors of vmut and vwt to be propagated in order that there is a 95% confidence that the real value of 1n (vmut/vw » falls within the reported experimental error (Taylor, 1982). Hence, values with experimental error ranges higher or lower than zero represent significant changes of rate of the mutant with respect to the wild-type.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to a method of assessing the tendency of a polypeptide to form aggregates comprising: analysing the amino acid sequence of the polypeptide to determine the propensity of the polypeptide to form local structure; comparing the propensity to form local structure of the polypeptide to the propensity to form local structure of a predetermined polypeptide sequence; and determining thereby whether the polypeptide has an altered tendency to aggregate compared to the predetermined polypeptide. The present invention thus relates to a method of predicting the tendency of a variant polypeptide to aggregate, particularly when denatured or partially denatured, when compared to a wild type polypeptide.

The method of the present invention can be used in the rational design of polypeptides which have an altered tendency to aggregate for example to design and produce a modified polypeptide with an enhanced or reduced tendency to aggregate compared to an unmodified protein. The method of the invention may also be used to compare the tendency to aggregate of a first and second polypeptide based on their propensities to form local structure and can be used to select the first or second polypeptide for production based on a selected tendency to aggregate. to We describe below the modifications in the polypeptide which are the subject of design.

A. Polypeptides of the Invention The present invention provides modified polypeptides with an altered tendency to aggregate. Polypeptide as used herein, preferably relates to any polypeptide molecule and to any fragment thereof at least 41 amino acids in length.

Preferably the polypeptide fragment will be at least 45 amino acids in length, more preferably at least 50,70,90,100 or more amino acids in length. Yet more preferably the fragment will have at most 20,10 or 5 amino acids deleted from the original sequence. Most preferably the polypeptide fragment will retain the biological activity of the protein from which it was derived. In this context, retention of a biological activity means that the fragment displays substantially the same desirable property as the wild-type protein, for example, fragments of an enzyme will typically retain substantially the same enzymatic activity, fragments of binding proteins will typically retain substantially the same binding specificity, and fragments of antigenic

proteins will typically retain substantially the same antigenic properties. Any polypeptide may be modified. Typically the polypeptide will be naturally occurring, but may be non-unnatural in sequence, such as that found in mutant organisms, or when produced recombinantly by transgenic or in techniques. Alternatively the polypeptide may be synthetic, that is, produced by an entirely chemical process.

B. Modifications The present invention relates in particular to a method of producing polypeptide having an altered tendency to aggregate and in particular to form non- covalent or non-crosslinked aggregates. The aggregates may be fibrils or amorphous/non-specific aggregates such as inclusion bodies. In a preferred aspect the invention provides a method which involves the analysis of the secondary structure of the polypeptide with a view to identifying modifications which will have an altered tendency for aggregation. Thus, identification of a desirable modification as described below may form part of the design process of the invention. In particular the method may comprise as a first step, analysis of the secondary structure of the polypeptide and identification of modifications which effect the propensity of the polypeptide to form local structure, and subsequently the production of a polypeptide so modified.

The modifications made to polypeptides of the invention result in changes to the primary structure, that is, the sequence of amino acids in the polypeptide mainchain. The modifications cause the denatured, or partially denatured form of the polypeptide to have an altered tendency to form local structure, particularly a-helical structure, or to form intermolecular interactions compared to the unmodified polypeptide. The reasoning for this alteration is based on the finding that, whilst it is essential for the native state of a polypeptide to be disrupted in order to allow aggregates to form, once denaturation has occurred the tendency to aggregate will depend on the intrinsic properties of the denatured state, that is, the equilibrium between the propensity to denature or to bind other denatured polypeptides. Thus the tendency of polypeptides to aggregate can be altered by changing the propensity of the denatured or partially denatured state to form intramolecular bonds instead of

intermolecular bonds. The inventors have demonstrated that the balance between the propensity to form intra-, or inter-, molecular bonds, or interactions such as hydrophobic or electrostatic interactions, can be changed by altering the formation of local structures, in particular by altering the propensity to form a-helical structures compared to the propensity to form a p-pleated sheet structures in the denatured state, and that this leads to an altered tendency to form aggregates such as fibrils.

Thus the modifications of the invention result in a polypeptide with a primary sequence that, in the denatured or partially denatured state, has an altered propensity tendency to form regions of local secondary structure, such as a-helical and/or 0- pleated sheet secondary structure. In one embodiment, the modifications made result in a polypeptide with an increased propensity to form a-helical secondary structure, and thus a decreased tendency to form aggregates (such as fibrils), relative to the unmodified polypeptide. In another embodiment, the modifications made result in a polypeptide with an decreased propensity to form a-helical secondary structure, and thus a increased tendency to form aggregates (such as fibrils), relative to the unmodified polypeptide. Therefore, polypeptides of the invention may be polypeptides obtained by this method of modification. Furthermore, aggregates which are obtained from polypeptides of the invention may be aggregates of the invention.

Preferably, the aggregates which are formed in accordance with the invention, or the aggregate having a reduced tendency to be formed by a polypeptide of the invention are non-covalent or non-cross linked aggregates. In particular, covalent aggregates may be formed by disulphide bridges being formed between two cysteine residues present in polypeptide monomers. Although a polypeptide may include intra-molecular disulphide cysteine bonds, it is also possible that such disulphide bonds may be produced during production of a protein or following denaturation of a protein. The present invention is primarily concerned with aggregates which are non-covalent aggregates, that is, aggregates which are not necessarily formed through disulphide bonds but which are formed through adoption of a high (3-structure content.

In certain embodiments, it will be particularly desirable to reduce the

tendency of the polypeptide in quetion to form aggregates form denatured or partially denatured states B 1. Selecting Amino Acid Residues for Modification The selection of amino acids for modification is dependent on the desired outcome of the modification, and a skilled person will be familiar with the methods used to choose which residues to change in any given situation.

If it is desirable to reduce the tendency of a polypeptide to form aggregates, such as fibrils, then the amino acid sequence should be modified to increase the propensity of the denatured polypeptide to form a-helical secondary structure, either native-like or non native like a-helical structure. This may be achieved in two, mutually compatible, ways. Firstly, regions that form a-helical secondary structure in the native state can be identified (techniques for identifying a-helical secondary structure are discussed below) and modified to include amino acids that increase the propensity of the denatured state to reform a-helices. Secondly, regions that do not form a-helical secondary structure in the native state can be identified and modified to include amino acids that increase the propensity of the denatured state to form a- helices.

Preferably mutations are made by substitution. However, they may also be made by other means, for example, by deletion of one or more amino acids, such as 1, 2,3 or 4 amino acids, inversion, and/or insertion of one or more amino acids, such as, 1, 2,3,4 or 5 amino acids. As highlighted above, preferably, the alteration of the tendency of the polypeptide to form aggregates is an alteration in the tendency to form non-covalent aggregates. Thus, in a preferred aspect of the invention, the modifications to the amino acid sequence do not involve modification of a cysteine residue which is involved in the formation of disulphide bonds associated with formation of covalent aggregates. Preferably, the amino acid residue selected for modification is not a cysteine residue. In one aspect of the invention the substitution or modification does not involve removal of a negative charge such as a conservative substitution of Asp to Asn or Glu to Gln.

As outlined below, the amino acid modifications preferably lead to the polypeptide exhibiting a global native folding or structure-topology which is

substantially equivalent to that of the unmodified polypeptide. Preferably, a modified polypeptide exhibits a global stability in the native state that is substantially equivalent to the global stability of the unmodified polypeptide in the native state.

Thus, it is preferred to retain similar structural features in the mutated polypeptide to those displayed by the wild type protein. Such structural features include: the number of secondary structure elements and/or their location in the protein sequence; the secondary structure content or amount of each secondary structure element in the native state; and the final tridimensional tertiary structure of the polypeptide (i. e. arrangement of the in space of the secondary structure elements and the interactions between them).

The extent to which these structural features are retained in a mutant polypeptide of the invention can be assessed by standard biophysical and/or biochemical techniques. In particular, one or more of the following measurement techniques may be performed. It is preferred to use a combination of several, eg 2,3, 4,5 or 6 of the following measurements to analyse the secondary structure of the full length protein: (1) Circular dichroism spectra of near UV and far UV regions in the absence and presence of chemical denaturants and/or at different temperatures ; (2) Fluorescence signals from aromatic residues in the absence and presence of chemical denaturants and or different temperatures, and also fluorescent signal arising from specific probes ; (3) Fourier-transform infrared spectra of native and denatured proteins; (4) mono-and/or multi-dimensional Nuclear Magnetic Resonance (5) X-ray diffraction analysis; (6) Analytical ultracentrifugation, and/or size-exclusion chromatography, and/or light scattering and/or NMR pulse-field gradient experiments to establish the size of the native protein.

Additionally, analysis of fragments of native polypeptides and polypeptides of the invention can be performed using the above-mentioned techniques.

Preferably, the modified polypeptide maintains intact at least one of its

biological functions such as enzymatic activity, ligand-binding or other biological activity. However, the substitutions do not necessarily need to be conservative modifications and/or charge replacement. The amino acid substitution is one which affects the secondary structure intrinsic propensity of the polypeptide sequence. For example, if changes are made far away in the 3-D structure from the active site of the protein (i. e. if during the design process the region selected to mutate is far from the active site), then a no-conservative alteration, eg substitution, may not be disruptive to the properties of the polypeptide. On the other hand, if the design of the mutant requires a change in a region of the protein close to or directly involved in the active site, conservative alterations are likely to be more appropriate and should desirably be attempted first, before non-conservative changes are investigated.

The design of mutations that increase local interactions such as a-helical propensities, can be carried out as follows: (i) The target sites for mutagenesis are preferably selected in an a-helix as those fully solvent-exposed positions in which the side chain does not interact with residues not included in the a-helix, and are not required for activity. The intention is to guarantee that the protein packing is maintained, that the mutations do not eliminate medium or long-range interactions and that the activity of the protein is maintained. It is preferred to select fully solvent exposed residues for mutation. ii) The mutations are designed to enhance a-helix propensity as indicated by a helix/coil transition algorithm tuned for heteropolypeptides, for example, AGADIR (Munoz and Serrano, 1997; Munoz and Serrano, 1994; Munoz, 1994a; Munoz, 1994b). The algorithm is modified to predict the stability for a particular sequence in an a-helical conformation instead of helical content for a peptide. The template sequence used typically corresponds to the X-ray a-helix plus about 4 residues, for example 4,5, or 6, on each side to prevent end effects. Sequences differing by the residues at the positions targeted for mutagenesis can be evaluated by the programme. The skilled person will appreciate that mutations which reduce local interactions such as a-helical propensities can also be designed by this method by selecting residues in which the side chain interacts with residues not included in the a-helix. Similar techniques may be applied to other local structure, for example, ß-

pleated sheets.

If it is desirable to increase the tendency of a polypeptide to form aggregates, such as fibrils, then the amino acid sequence should be modified to decrease the propensity of the denatured polypeptide to form a-helical secondary structure. This may be achieved by selecting regions that form a-helical secondary structure in the native state and modifying them to include amino acids that decrease the propensity of the denatured state to reform a-helices.

Similarly, modifications to the wild-type amino acid sequence could be made which alter the propensity of the polypeptide to form local structure, and alter formation of intermolecular interactions compared to the unmodified polypeptide which leads to an altered tendency to aggregate and in particular to form non- covalent aggregates. In particular, one or more amino acid residues may be substituted with another having a decreased hydrophobicity, for example, as defined by Roseman, 1998. Alternatively, the substitution may lead to a decrease in the intrinsic propensity to form P-structure, as defined for example, by the method of Chou and Fasman, 1978 or Street and Mayo, 1999. Such a substitution decreases the tendency of the polypeptide to form aggregates. On the other hand, substitutions which increase the hydrophobicity or which increase the propensity to ß- structure increase the tendency of the polypeptide to form aggregates and in particular non-covalent aggregates. Techniques for predicting the propensity to form P-sheet structure may similarly be carried out using modelling by computer-run algorithms.

The selection of target sites for mutagenesis may be carried out, for example, by identifying those regions of the polypeptide which have an intrinsic helical propensity. For example, non-helical polypeptides could be examined using suitable modelling software such as AGADIR to identify sequences with intrinsic helical propensity. Even if the propensity to form helical structure is low, mutagenesis may be used to enhance the intrinsic helical propensity and thus reduce the tendecy to aggregate.

In the alternative, for a selected polypeptide a set of mutants may be prepared, in which conservative substitutions are made. Subsequently, the rate of aggregation

can be monitored to identify regions in the polypeptide which play a more important role in aggregation. Suitable mutations in these regions can then be identified, for example, to reduce the hydrophobicity or propensity to form a-helical structure to reduce the tendency to form non-covalent aggregates.

In an alternative aspect of the invention, mutations are made to increase the hydrophobicity, reduce the tendency to form a-helices or increase the tendency to form (3-sheet structure to increase the production of aggregates.

B2. Predicting Secondary Structure Techniques for predicting regions of secondary structure such as a-helical secondary structure from the amino acid sequence of polypeptides are known in the art and typically employ modelling by computer-run algorithms, such as AGADIR (Munoz and Serrano, 1997), and the Zimm-Bragg and Lifson-Roig formalisms.

These techniques are increasingly reliable, since the understanding of a-helical secondary structure is virtually complete (Munoz and Serrano, 1997).

The mutation or modification of the present invention are selected such that, in the native state the folding, structure topology and preferably global stability is substantially equivalent to that of the unmodified polypeptide. That is to say that, for example, the native folding of the polypeptide and in particular the formation of a- helices or p-sheets associated with the native polypeptide are substantially unaltered compared to an unmodified polypeptide. In this way, the mutated polypeptide may be produced and undergo normal folding such that, for example, at least one of the biological activities of the protein is maintained. In particular, a modified polypeptide will retain a biological function such as enzyme activity, ligand-binding or antigen production of that of the wild-type protein. In addition, preferably the folded mutated or modified polypeptide maintains essentially the same global stability of the unmodified polypeptide such that the modified polypeptide does not show an enhanced tendency to denature compared to the unmodified polypeptide.

Preferably, an modified polypeptide will demonstrate an altered tendency to form aggregates under denaturation. Thus, if a polypeptide of the invention is placed under denaturing conditions, preferably, the tendency to aggregate is altered compared to the wild-type. In a preferred embodiment, a denatured polypeptide

according to the invention has a reduced tendency to aggregate compared to the wild- type.

C. Further Modifications of the Amino Acid Sequence The amino acid sequence of the polypeptide may be further modified for any reason. Typically the modification may aid purification by the addition of histidine residues or a T7 tag, or assist in identification or purification or by the addition of a signal sequence to promote their secretion from a cell where the polypeptide does not naturally contain such a sequence. The sequence modification may change the activity of the polypeptide, for example, by changing amino acids that contribute to the activity of an active site, ligand binding site, or regulatory region. The polypeptide may be modified by fusion with other amino acid sequences. Fusion may, therefore, result in additional amino acid sequence at one or both terminal regions of the polypeptide, that is, at the amino-terminal or carboxy-terminal regions, or at an internal region of the polypeptide. The polypeptide may additionally or alternatively be modified by deletion. The deletion may remove sequences from one or both terminal region of the polypeptide, that is, from the amino-terminal or carboxy-terminal regions, or may remove sequences from one or more internal regions of the polypeptide. Modification by deletion and addition can therefore encompass substitution, whereby any number of amino acid residues, individually, or as contiguous sequences, can be replaced by different amino acids. Additional and/or deletion modifications typically involve, for example, 1 amino acid, typically 2 or more, such as 5,10,20,50,100 or more amino acids.

D. Production of Mutant Polypeptide with Modified Amino Acid Sequence Polypeptides designed to encompass alterations as defined above can be produced by any method of polypeptide synthesis known in the art. Typically polypeptides of the invention are produced by chemical synthesis or recombinant in vitro or in vivo expression.

Polypeptides may be chemically synthesised using various solid-phase techniques (e. g. Roberge et al. 1995) and automated synthesis may be achieved, for example, using the ABI 431 A Peptide Synthesizer (Perkin Elmer).

Recombinant in vivo or in vitro production of polypeptides can be achieved by the expression of a polynucleotide that comprises a sequence that encodes a polypeptide of the invention.

D. 1. Polynucleotides A further embodiment of the invention provides a polynucleotide which comprises a sequence that encodes a polypeptide of the invention. The polynucleotide sequence may be designed with reference to the degeneracy of the genetic code and in light of the preferred codon-usage for any particular organism in which the polynucleotide might be expressed. The polynucleotide may be DNA or RNA, and may be single or double stranded, that is comprising a polynucleotide of the invention and its complement. They thus consist essentially of DNA or RNA encoding the amino acid sequence of the invention.

The polynucleotides may include within them synthetic or modified nucleotides. A number of different types of modification to polynucleotides are known in the art. For the purposes of the present invention, it is to be understood that the polynucleotides described herein may be modified by any method available in the art. Such modifications may be carried out, for example, in order to change the in vivo activity or lifespan of polynucleotides of the invention.

Polynucleotides of the invention may be used to produce a primer, e. g. for use in PCR (polymerase chain reaction), or alternative amplification reaction (for example to facilitate amplification or site directed mutagenesis). Such primers and other fragments will be at least 15, preferably at least 20, for example at least 25,30 or 40 nucleotides in length, and are also encompassed by the term polynucleotides of the invention as used herein.

Polynucleotides such as a DNA polynucleotide and primers according to the invention and the unmodified forms thereof may be produced recombinantly, synthetically, or by any means available to those of skill in the art. They may also be cloned by standard techniques. Thus polynucleotides may be cloned into any vector available in the art. The polynucleotides are typically provided in isolated and/or purified form.

In general, short polynucleotides of the invention, e. g. primers, will be

produced by synthetic means, involving a step wise manufacture of the desired nucleic acid sequence one nucleotide at a time. Techniques for accomplishing this using automated techniques are readily available in the art.

Longer polynucleotides of the invention and of unmodified forms may be produced by combining short polynucleotides using standard techniques, for example by ligation. They may also be produced by recombinant means, for example using PCR cloning techniques. This will involve making a pair of primers (e. g. of about 15-30 nucleotides) to a region of the gene which it is desired to clone and bringing the primers into contact with a target polynucleotide. For the unmodified form, the target polynucleotide used is typically obtained from a cell in the form of genomic DNA (to allow cloning of the whole gene, typically including introns and promoter regions) or mRNA, or cDNA prepared therefrom. For production of polynucleotides of the invention, small quantities of the polynucleotide, produced by any means, may be used as the target polynucleotide in a PCR amplification reaction. Amplification is performed under suitable conditions to bring about selective amplification.

Following amplification of the desired region, the amplifie fragment may be isolating (e. g. by purifying the reaction mixture on an agarose gel) and recovered.

The primers may be designed to contain suitable restriction enzyme recognition sites so that the amplified DNA can be cloned into a suitable cloning vector.

Although in general the techniques mentioned herein are well known in the art, reference may be made in particular to Sambrook et al, 1989.

Polynucleotides of the invention may be obtained by site directed mutagenesis of a polynucleotide comprising the unmodified sequence. This technique may be performed in any manner. Typically the technique may be performed by PCR. Such techniques typically include the use of a primer that comprises a predominantly identical nucleotide sequence to a region of the unmodified sequence in which mutation is desired, other than changes at the nucleotide residues appropriate to bring about the desired alteration. Subsequent use of this primer in a PCR amplification reaction will thus introduce the desired changes to the nucleotide sequence of the PCR product. The desired site for site directed mutagenesis is typically within the coding sequence of the gene, and thus the PCR

product may be a truncated form of the target polynucleotide. A full length modified polynucleotide of the invention may be generated by combining the amplification product with other polynucleotides that have unmodified or modified sequences (generated by any technique). Combination may be performed by any technique known in the art, e. g. ligation. This technique may also be useful where for example silent codon changes are required to optimise codon preferences for a particular host cell in which the polynucleotide sequences are being expressed. Other sequence changes may be desired in order to introduce restriction enzyme recognition sites, or to further alter or modify the property or function of the polypeptide encoded by the polynucleotide.

The modified polynucleotide generated may be tested for the desired sequence by its sequencing. This may be performed, for example, by bringing a sample containing the putative modified polynucleotide, as target, into contact with a probe comprising a polynucleotide or primer of the invention under hybridizing conditions and determining the sequence by, for example the Sanger dideoxy chain termination method (see Sambrook et al, 1989).

Such a method generally comprises elongating, in the presence of suitable reagents, the primer by synthesis of a strand complementary to the target polynucleotide and selectively terminating the elongation reaction at one or more of an A, C, G or T/U residue; allowing strand elongation and termination reaction to occur; separating out according to size the elongated products to determine the sequence of the nucleotides at which selective termination has occurred. Suitable reagents include a DNA polymerase enzyme, the deoxynucleotides dATP, dCTP, dGTP and dTTP, a buffer and ATP. Dideoxynucleotides are used for selective termination.

D. 2. Vectors Polynucleotides of the invention can be incorporated into any vector available in the art. A vector of the invention consists essentially of a polynucleotide of the invention, therefore. Usually the vector will be a recombinant replicable vector.

The vector may be used to replicate the nucleic acid in a compatible host cell. Thus in a further embodiment, the invention provides a method of making polynucleotides

of the invention by introducing a polynucleotide of the invention into a replicable vector, introducing the vector into a compatible host cell, and growing the host cell under conditions which bring about replication of the vector. The vector may be recovered from the host cell. Suitable host cells are described below in connection with expression vectors.

D. 2.1. Expression Vectors Preferably, a polynucleotide of the invention in a vector is operably linked to a control sequence which is capable of providing for the expression of the coding sequence by the host cell, i. e. the vector is an expression vector. Such expression vectors can be used to express polypeptides of the invention.

The term"operably linked"refers to a juxtaposition wherein the components described are in a relationship permitting them to function in their intended manner.

A control sequence"operably linked"to a coding sequence is ligated in such a way that expression of the coding sequence is achieved under conditions compatible with the control sequences.

Such vectors may be transformed into a suitable host cell as described above to provide for expression of a polypeptide or polypeptide fragment of the invention.

Thus, in a further aspect the invention provides a process for preparing polypeptide or polypeptide fragment according to the invention, which process comprises cultivating a host cell transformed or transfected with an expression vector as described above under conditions to provide for expression of the polypeptide or fragment, and recovering the expressed polypeptide or fragment.

The vectors may be for example, plasmid, virus or phage vectors provided with an origin of replication, optionally a promoter for the expression of the said polynucleotide and optionally a regulator of the promoter. The vectors may contain one or more selectable marker genes, for example, an ampicillin resistance gene in the case of a bacterial plasmid, a neomycin resistance gene for a mammalian vector, or a kanomycin resistance gene for a plant vector. Vectors may be used in vitro, for example for the production of RNA or used to transfect or transform a host cell. The vector may also be adapted to be used in vitro, for example in a method of gene therapy.

A further embodiment of the invention provides host cells transformed or transfected with the vectors for the replication and expression of polynucleotides of the invention. The cells will be chosen to be compatible with the said vector and may for example be bacterial, yeast, plant, insect, or mammalian.

D. 3. Expression in host cells Expression vectors of the invention may be introduced into host cells using conventional techniques including calcium phosphate precipitation, DEAE-dextran transfection, electroporation, particle bombardment or Agrobacterium tumefaciens- mediated techniques. Expression from the host cell may be transient. Stable host cell transformation may be achieved by integration of a polynucleotide of the invention, or a fragment thereof, into a genome of the host cell. Typically the transformed genome is nuclear, although transformation of other genomes may be desired, for example the mitochondrial genome of eukaryotic cells, or a plastidic genome of plant cells. Alternatively, stable transformation may be achieved using replicable autonomous vectors. The expression vector may contain a selectable marker and/or such a selectable marker may be co-transfected with the expression vector and stable transfected cells may be selected.

Suitable cells include cells in which the abovementioned vectors may be expressed. These include microbial cells such as bacteria such as E coli, mammalian cells such as CHO cells, COS7 cells, P388 cells, HepG2 cells, KB cells, EL4 cells or Hela cells, insect cells, yeast such as Saccharomyces or plant cells, typically of crop plants such as wheat, maize or oil-seed rape. Baculovirus or vaccinia expression systems may be used.

Cell culture will take place under standard conditions. Commercially available cultural media for cell culture are widely available and can be used in accordance with manufacturers instructions.

D. 4. Recoverv Polypeptides and aggregates of the invention expressed in host cells may be recovered by any technique known in the art. This may lead to isolation and purification of the polypeptide or aggregate. In some cases, the polypeptide may aggregate within the host cell. Aggregates can typically be isolated by lysing the

host cell, isolating the aggregate by centrifugation, washing the thus isolated aggregate and optionally allowing the aggregate to dissociate, and individual polypeptides to denature. Typically an isolated or purified polypeptide will account for at least 10% to 100% dry mass of the polypeptide present in the sample, more preferably at least 40% or 50%, even more preferably at least 60% or 70% yet more preferably at least 80%, 90%, 95%. Most preferably a purified polypeptide will account for at least 99% by dry mass of the polypeptide present in a sample.

E. Aggregates The aggregated of the present invention may comprise non-naturally occurring polypeptides. The polypeptides may be, for example, polypeptides which have been chemically modified such as polypeptides which have been glycosylated or polypeptides which comprise a modified amino acid residue, a pharmaceutically active compound, a metal or a functional group such as a thiol group which is capable of binding one or more reactants.

Preferably the aggregates formed are fibrils. The fibrils of the present invention are preferably long, and/or straight and/or unbranching. Most preferably they are long, straight and unbranching. The diameter of the fibrils is generally from 1 to 20 nm, preferably from 5 to 15 nm and more preferably from 7 to 12 nm. The diameter of the fibrils may be varied by selecting suitable polypeptides.

Fibrils of the present invention may comprise a hollow core which may be useful in a variety of applications.

E. 1. Production of Aggregates Aggregates of the invention may be produced by any method known in the art. Aggregation typically occurs by nucleation. Nucleation occurs when intermolecular bonds form between polypeptides in a partially or fully denatured state. The process of nucleation can, therefore, be brought about in any situation by contacting polypeptides under suitable conditions. For the purposes of nucleation, a suitable condition is any condition able to generate partially or fully denatured polypeptide molecules, although conditions must not favour denaturation to the extent that intermolecular bonds are prevented from forming. The optimal nucleation conditions are different for each polypeptide. Important parameters for nucleation

typically include variations in solvents, polypeptide concentration, salt, ligands, temperature and pH. A skilled person will be able to determine suitable conditions for any given polypeptide. Typically nucleation can be caused by incubation in urea, preferably at 3 to 6 to 7M concentration, for example 4 to 6 M, preferably around 4 to 5 M, preferably for at least 1 hour, followed by dilution, preferably to at most 1M, and preferably followed by further incubation for at least 1 hour.

In some cases, it may be advantageous to alter the optimal nucleation conditions for any given polypeptide. For example, if aggregation is disadvantageous then it may be desirable to produce a polypeptide that is more resistant to aggregate formation. Thus polypeptides of the invention with an increased propensity to form local structure, preferably a-helical secondary structure, relative to a polypeptide with an unmodified sequence may be used to overcome disadvantageous aggregation.

Alternatively, nucleation and aggregate formation, particularly fibril formation, may be beneficial. Polypeptides of the invention with a decreased propensity to form local structure, preferably a-helical secondary structure, relative to a polypeptide with an unmodified sequence may be used to promote nucleation and aggregation. Thus polypeptides of the invention may be tailored to increase aggregation under any conditions, such as to improve rates of production of aggregates, such as fibrils, or to allow production in conditions unsuitable for nucleation of the unmodified polypeptide.

E. 2. Measuring Aggregate Formation In order to ascertain optimal nucleation and aggregate formation conditions, or to test whether amino acid sequence modifications have caused the predicted effect to the tendency to aggregate, it is necessary to measure aggregate formation.

Aggregate formation, such as fibril formation can be measured by any technique known in the art. Techniques typically used include circular dichroism, sedimentation analysis, Thioflavin-T and Congo red binding assays, polypeptidease resistance assays, Fourier-transform infrared spectroscopy, electron microscopy and X-ray diffraction, as exemplified below.

F. Uses and Examples of Polypeptides of the Invention

F. 1. Use of Polypeptides of the Invention Polypeptides of the invention may be used for any application. Typically they will be used in medical and industrial applications in which aggregate formation, such as fibril formation, is advantageous or disadvantageous.

F. l. 1. Medical and Industrial Uses of Polypeptides of the Invention Polypeptides made for use in medical or industrial applications typically experience a range of conditions. The conditions experienced by the produced polypeptide, such as those in the reaction vessel (in the case of synthetic methods) or at intracellular, extracellular or in vitro locations (for recombinant expression methods), during isolation, modification, storage, administration or use, may favour disadvantageous nucleation and the formation of aggregates, such as fibrils, or may be unsuitable for the desired aggregate formation.

Thus polypeptides of the invention can be used to overcome disadvantageous production of aggregates, such as fibrils. For example, polypeptides of the invention may be used to prevent aggregation at the high polypeptide concentrations sometimes generated during synthetic or recombinant production, or required for therapeutic benefit or industrial applicability. Polypeptides of the invention may be used in conditions where unmodified polypeptides could not, such as in conditions where the choice of solvent, temperature, pH, or salinity would promote undesirable aggregate formation of polypeptides the unmodified sequence. Any polypeptide can be used in this manner. Examples of pharmaceutically useful polypeptides include calcitonin, insulin and cytokinin. For example, polypeptides of the invention with a reduced tendency to aggregate could be used in industrial processes such as washing powders, or in pharmaceutical or diagnostic processes.

Polypeptides of the invention may also be used in applications where nucleation and aggregation may be beneficial. For example, aggregates, such as fibrils, produced by recombinantly expressed polypeptides can form inclusion bodies which represent a form of polypeptide that can be conveniently isolated, typically by centrifugation. Polypeptides of the invention may be generated that have a greater tendency to form aggregate, such as fibrils, in intracellular, extracellular, of in vitro conditions and thus facilitate isolation. Aggregates have many potential applications

but the production of them can be a slow process. Polypeptides of the invention may be used to improve rates of formation of aggregates, such as fibrils, and thus increase the efficiency of production. It may be convenient, for example, for use in drug delivery, for polypeptides of the invention to be bound to a molecule, for instance, a pharmaceutically active one. In one embodiment of this, it may be desirable for the thus modified polypeptide to form an aggregate such as a fibril, for example, for use as a slow-release mechanism (see section F. 1.2 below). Polypeptides of unmodified sequence may only form aggregates in conditions that are undesirably harsh for the bound molecule. Thus polypeptides of the invention may provide a suitable vehicle for a bound molecule, since the polypeptides may be tailored to aggregate under suitably mild conditions such that the bound molecule is substantially unaffected, that is, that, on release, the molecule is able to perform the function for which it was required.

F. 1. 2. Use of Fibrils as a Slow-Release Mechanism Once formed, aggregates, such as fibrils, are able to dissociate to release polypeptide molecules. This can be particularly useful, for example, where such aggregates can be introduced as a slow release mechanism. This may be useful when the released polypeptide is usefulper se for any application, for example as a therapeutic agent. Such polypeptides can include pharmaceutically or therapeutically active polypeptides, for example, hormones, vaccines, polypeptides suitable for the treatment of cancer or involved in blood clotting or fibrinolysis, and polypeptides useful in industrial processes, for example in biological washing powders and the like.

The constituent polypeptides of aggregates may also be useful as"vehicles" or"carriers"for a slow-release mechanism of other molecules. For example, a molecule bound to polypeptides in an aggregate may be inactive until the polypeptide molecule to which it is bound dissociates from the aggregate. Following release of the polypeptide from the aggregate, the molecule may remain attached to the released polypeptide, or the bond may be broken, by passive or active means. Any molecule capable of being bound to a polypeptide of the invention, by covalent or any other type of bond, can be used. Molecules can be useful for any application, such as a

drug for use in the treatment of the human or animal body, or for use in the manufacture of a medicament for use in the treatment of an disease such as cancer.

However, the rate of dissociation of a polypeptide from its fibril may not be suitable to the application in hand. Therefore, polypeptides and fibrils of the invention further provide a method of modifying the rate of release from such slow-release mechanisms.

F. 1. 5. Antibodies The invention also provides monoclonal or polyclonal antibodies which are specific to the polypeptides and/or aggregates of the invention. Typically these will be able todiscriminate between modified polypeptides of the invention and unmodified polypeptide. Such monoclonal antibodies may be prepared by conventional hybridoma technology using polypeptides of the invention as immunogens. Polyclonal antibodies may also be prepared by conventional means which comprise inoculating a host animal, for example a rat or a rabbit, with a polypeptide of the invention and recovering immune serum. In order that such antibodies may be made, polypeptides may be haptenised to another polypeptide for use as immunogens in animals or humans. For the purposes of this invention, the term"antibody"includes antibody fragments such as Fv, F (ab') and F (ab') 2 fragments, as well as single chain antibodies. Furthermore, the antibodies and fragments thereof may be humanise antibodies, e. g. as described in EP-A-239400.

The antibodies may be used for detecting polypeptides of the invention present in biological samples by a method which comprises: (a) providing an antibody of the invention ; (b) incubating a biological sample with said antibody under conditions which allow for the formation of an antibody-antigen complex; and (c) determining whether antibody-antigen complex comprising said antibody is formed.

Suitable samples include extracts from any tissue or cell type. Typically the tissue will be brain, liver, kidney, spleen, heart or the lymph system.

Antibodies of the invention may be bound to a solid support and/or packaged into kits in a suitable container along with suitable reagents, controls, instructions

and the like.

F. 1. 6 Examples of Polypeptides of the Invention Some preferred polypeptides of the invention are growth factors for example cytokines and neurotrophic factors. Cytokines are preferred, and interleukins, especially IL-4 are particularly preferred. Other interleukins such as IL-6 and IL-12 may also be used according to the invention.

F. 2. Use of Polynucleotides of the Invention F. 2.1. Diagnosis Polynucleotides and primers of the invention can be used for looking at polymorphism in genes leading to changes in the tendency of their polypeptide products to aggregate, especially as fibrils. Individuals with disease states such as Alzheimer's disease, type II diabetes, systemic amyloidosis, transmissible spongiform encephalopathies, Huntington's disease, Parkinson's disease, cystic fibrosis, emphysema and the like can be examined for gene polymorphism and this can be examined for correlation with disease status. Thus the polynucleotides or probes of the invention may conveniently be packaged in the form of a test kit in a suitable container. In such kits the probes may be bound to a solid support where the assay format for which the kit is designed requires such binding. The kit may also contain suitable reagents for treating the sample to be probed, hybridizing the probe to nucleic acid in the sample, control reagents, instructions, and the like.

F. 2.2 Gene therapy Polynucleotides of the invention may inserted into an appropriate polynucleotide vector and used for gene therapy. Thus polynucleotides of the invention also comprise those suitable for gene therapy. Vectors suitable for gene therapy may be of any kind, although are typically viral. Preferred viral vectors comprise polynucleotide sequences derived from retroviruses, lentiviruses, adenoviruses and adeno-associated viruses. Gene therapy may be performed on in vivo, or ex vivo cells, the later being optionally transplanted into a host organism.

Such techniques allow the expression of polynucleotides of the invention in all cell types, preferably somatic cells. Polynucleotides preferred for expression in gene therapy include those encoding polypeptides responsible for amyloid fibril and prion

related diseases, more preferrably polypeptides responsible for Alzheimer's disease, type II diabetes, systemic amyloidosis, scrapie, transmissible spongiform encephalopathies, Parkinson's disease, Huntington's disease, cystic fibrosis or emphysema and the like.

F. 2. 3. Generation of Transgenic Organisms Host cells of the invention (see above) can be used to generate stably transformed first generation and subsequent progeny organisms comprising at least one polynucleotide of the invention. Propagation of a single transformed cell by techniques known in the art allows genetically stable transgenic cultures to be generated. Commercially available cultural media for cell culture are widely available and can be used in accordance with manufacturers'instructions. These may be subsequently used for the mass production of an expressed polypeptide of the invention. Alternatively, transformed cells of the invention, when derived from a multicellular organism can be used to generate a transgenic multicellular organism.

Any multicelluar organism may be generated, although plants and animals are preferred. More preferably, non-human transgenic animals are generated.

Particularly preferred animals include cows, sheep, goats, horses, pigs, chickens, turkeys and deer. The most preferred transgenic animals include those transformed by a polynucleotide that encodes a polypeptide of the invention related to an amyloidogenic or prion disease such as scrapie or bovine spongiform encephalopathy. Transgenic organisms of the invention can thus be bred to provide stable transgenic progeny. Thus by generating transgenic animals of the invention it may be possible to develop animals lines with stable resistance to such diseases.

F. 2.4. Administration Any of the polypeptides, polynucleotides, vectors, cells, cell cultures, organisms, antibodies or aggregates discussed above in any form or in association with any other agent discussed above is included in the term'agent'below. An effective non-toxic amount of such a agent may be given to a human or non-human patient in need thereof. Some non-human species that may be preferred are listed in F. 2.3. above. The condition of a patient suffering from a disease can therefore be improved by administration of such an agent. The agent may be administered

prophylactically to an individual who does not have a disease in order to prevent the individual developing the disease.

Thus the invention provides the agent for use in a method of treating the human or animal body by therapy. The invention provides the use of the agent in the manufacture of a medicament for treating the disease. Thus the invention provides a method of treating an individual comprising administering the agent to the individual.

The agent is typically administered by any standard technique used for administration, such as by injection.

Typically after the initial administration of the agent, the same or a different agent of the invention can be given. In one embodiment the subject is given 1,2,3 or more separate administrations, each of which is separated by at least 12 hours, 1 day, 2, days, 7 days, 14 days, 1 month or more.

The agent may be in the form of a pharmaceutical composition which comprises the agent and a pharmaceutically acceptable carrier or diluent. Suitable carriers and diluents include isotonic saline solutions, for example phosphate- buffered saline. Typically the composition is formulated for parenteral, intravenous, intramuscular, subcutaneous, transdermal, intradermal, oral, intranasal, intravaginal, or intrarectal administration.

The dose of vaccination may be determined according to various parameters, especially according to the substance used; the age, weight and condition of the patient to be treated; the route of administration; and the required regimen. A physician will be able to determine the required route of administration and dosage for any particular patient. A suitable dose may however be from I Ogg to lOg, for example from 100 pg to lg of the agent. These values may represent the total amount administered in the complete treatment regimen or may represent each separate administration in the regimen.

In the case of agents which are polynucleotides transfection agents may also be administered to enhance the uptake of the polynucleotides by cells. Examples of suitable transfection agents include cationic agents (for example calcium phosphate and DEAE-dextran) and lipofectants (for example lipofectam and

transfectarnTM).

When the agent is a polynucleotide which is in the form of a viral vector the amount of virus administered is in the range of from 104 to 1012 pfu, preferably from 107 to 1010 pfu (for example for adenoviral vectors), more preferably about 108 pfu for herpes viral vectors. A pox virus vector may also be used (e.g. vaccinia virus), typically at any of the above dosages. When injected, typically 1-2 ml of virus in a pharmaceutically acceptable suitable carrier or diluent is administered.

As outlined above, the analysis of the intrinsic propensity of a polypeptide to form local structure can be used in the design of modified polypeptides which have an altered tendency to form aggregates and in particular to form non-covalently linked aggregates. The same analysis may also be used to assess the tendency of a polypeptide to form non-covalent aggregates. Such an analysis may not only be used in the design of polypeptides which have an altered tendency to aggregate, but may be also be used to analyse the general tendency of a polypeptide to aggregate, for example, to establish whether such a polypeptide might be suitable for production in a particular recombinant system under selected conditions. For example, should the analysis of the propensity of the polypeptide to form local structure demonstrate that the polypeptide in question had a high propensity to form a-helical structure, the tendency of the polypeptide to form aggregates when expressed in vitro can be assessed. The analysis can also be used to choose between first and second polypeptides, for example known variant proteins to select the polypeptide having a desired or selected tendency to aggregate.

The analysis of the propensity of a polynucleotide to form local structure such as a-helical structure may also be used in the diagnosis or assessment of the likelihood of an individual to suffer from a disease associated with the formation of amyloid fibrils or aggregates. For example, an analysis may be carried out of those proteins known to be associated with amyloidosis or formation of aggregates in disease conditions such as amyloidosis or Alzheimers. Comparison of the propensity of a polypeptide to form aggregates, and comparison of that propensity with polypeptides found in non diseased and diseased individuals may provide

indication of whether an individual is at risk of subsequently suffering from amyloidosis or other disease associated with aggregate formation. In particular, the amino acid sequence of a protein associated with amyloid disease could be identified for an individual. Should a variant polypeptide be identified which shows a decrease propensity to form a-helical structure, such an individual may be considered to be at increased susceptibility or risk of an amyloid or aggregate associated disease. Such individuals could then be treated with a view to reducing the tendency to form aggregates. Alternatively, such a variant polypeptide could be analysed as part of a diagnosis, to confirm that a variant polypeptide expressed by that individual had an increased tendency to form aggregates and thus, for example, to assist in the diagnosis of that individual, for example as suffering from CJD or Alzheimers disease associated with aggregate or amyloid formation.

EXAMPLES Examples 1 to 4 describe the Inventors'studies of Interleukin 4 (IL-4) a pleiotropic cytokine produced mainly by T helper lymphocytes type 2 (TH2). In this work two possible strategies to improve the refolding yield of recombinant IL-4 are presented based on the elimination of surface hydrophobic residues and on the local stabilisation of secondary structure elements. It is shown that the rational stabilisation of an a-helix using the helix/coil transition algorithm, AGADIR (Munoz and Serrano, 1997; Lacroix, 1998), increases the stability of IL-4 against thermal and GdnHCl denaturation but, more importantly, it doubles the in refolding yield of the recombinant IL-4 protein.

Example 1: Design of the mutations Using AGADIRl s-2 (Lacroix, 1998) three different types of mutations have been designed that increase the average helical content up to 20% and the stability of the a-helix by about 1.8 Kcal/mol. The N-capping box motif present in the wild type protein (T-X-X-Q) has been replaced by a better counterpart (S-X-X-E). Better helix formers (Chakrabartty and Baldwin, 1995, Munoz, 1994d) have been introduced into the sequence of the helix: Gln71 and Phe73, were mutated into Ala, and H74 was substituted by Asn. The reason why Asn was preferred over Ala to replace H74 is that, introduction of yet another Ala at this position would result in a hydrophobic cluster at the N-terminus comprising five Ala residues that could reduce the

solubility of the protein and produce aggregation. Finally, two favourable electrostatic pairs (E-X-X-R and E-X-X-X-H) have been introduced. The mutant bearing the stabilised helix C will be herein designated IL-4BChelix.

Example 2: Peptide structural characterisation In order to demonstrate that the predictions made by the algorithm were correct, two peptides corresponding to the wild-type and mutant sequences of helix C were synthesised. Circular Dichroism and NMR analysis show a clear increase in the helical content of the mutant peptide. The values predicted by AGADIRls-2 for the helical content of the wild-type (IL-4WT) and mutant peptide are in agreement with the experimental results.

Example 3: Thermodynamic and structural characterisation of the proteins The two IL-4 mutants, IL-4W91S and IL-4BChelix, were overexpressed and purified to homogeneity, displaying NMR spectra that are similar to that of the wild- type protein. In order to investigate the effect of the designed mutations on the stability of the protein, equilibrium thermal and GdnHCl denaturation studies were carried out. IL-4 is a highly stable protein that cannot be denatured by temperature in the 0 to 100°C range. Therefore, in order to observe a complete unfolding transition, the thermal denaturation experiments were carried out in the presence of 2 M GdnHCl. Under these conditions, the whole transition can be observed, although the end of the curve corresponding to the fully denatured state is missing.

Interestingly, the WT protein undergoes cold denaturation below 20°C. At higher GdnHCl concentrations the proteins are fully denatured at high temperature, but they are not fully folded at low temperatures. As a result it is not possible to obtain accurate thermodynamic data from these curves, although the temperature at which half of the protein is denatured (Tm) can be inferred. The Tm values indicated that IL-4W91S is the most stable protein, followed by IL-4BChelix and IL-4WT.

From the GdnHCl denaturation curves, it is possible to obtain reliable thermodynamic data. In agreement with the temperature denaturation studies, we find that IL-4W91S is the most stable protein (1.4 kcal/mol more stable than the WT), followed by IL-4BChelix (0.5 kcal/mol more stable than the WT).

Example 4: Refolding vield and activity assays Two experiments were performed to test the refolding yields of IL-4WT and

the mutant proteins. The proteins were overexpressed in 1L shaking flasks, as described under materials and methods. In another two experiments, IL-4WT and the two mutant proteins were isolated from cells obtained after low cell density fermentation on the 10L and 100L scale (see Materials and Methods). In the four experiments, the refolding yield for IL-4BChelix is approximately two-fold that of IL-4WT, while the Kd for IL-4Ra remains identical to that of the wild-type protein.

Preliminary data on the oxidative folding of IL4 suggest that the observed increase in the refolding yield of IL-4BChelix, results from the destabilisation of a non-native isoform which accumulates during the refolding of the wild-type protein. On the other hand, replacement of Trp91 by Ser does not improve the refolding of the protein.

DISCUSSION OF EXAMPLES 1 TO 4 This work has been aimed at improving the refolding yield of recombinant EL-4 without interfering with the binding to IL-4Ra. The main goal was to design a mutant that would refold at higher yield than the WT, while retaining its biological activity. Such an IL-4 mutant could be used as a scaffold into which other mutations would be introduced. This strategy would, for example, make the large-scale production of IL-4 and IL-4 antagonist mutants, economically more affordable.

Thermodynamic analysis of the oxidized proteins The stability of the two mutants in the oxidized form of IL-4, as well as their binding to IL-4Ra were determined and compared to the wild-type protein. An unfolding free energy of 4.3 kcal/mol was obtained for IL-4WT.

Removal of solvent exposed hydrophobic residues should stabilise the target protein through the"inverse hydrophobic effect" (Pakula and Sauer, 1990). In the Examples presented, substitution of Trp91 by Ser produces a large stabilising effect (1.4 kcal/mol) and induces a shift of 8°C in the Tm of the wild-type protein. The observed stabilisation arises from an increase in the m value. This increase in m value might arise from the destabilisation of a folding intermediate, the denatured state, or both (for a comprehensive review see Shortle, 1996). A similar effect (increase in AG of 1.4 kcal/mol and increase in m value from 1.6 to 1.9), was found in the chemotactic protein, CheY, when a solvent exposed Phe residue (F14), was replaced by Asn (Munoz, 1994d). In this case the change in slope was due to a

relative destabilisation of a folding intermediate, which could only be detected kinetically. The equilibrium denaturation curves of IL-4 have been fitted assuming a three-state model. Significant improvement in the fitting of the curves was not detected, as was the case for CheY. Therefore, at present we cannot assign the change in m to a destabilisation of an equilibrium folding intermediate present in the folding of the oxidized enzyme.

As far as helix stabilisation is concerned, it has been shown whilst an increase in protein stability is observed (and in some cases could produce thermostable proteins) (Villegas, 1996), the increase in stability is less than expected. This is due to a simultaneous stabilisation of the denatured state under native conditions. Thus, the results obtained in IL-4 are in line with that expected. The mutations introduced into helix C, stabilise the protein to a lesser extent (0.5 kcal/mol) and induce a smaller shift (6°C) in the Tm of IL-4WT. In this case, the stabilisation effect comes mainly from an increase in the GdnGCl concentration necessary to denature half of the protein, (GdnGCI),,,.

Improvements of the refolding yield for the reduced IL-4 protein Two of the mutants were found to have a reduced refolding yield of IL- 4W91 S similar to that of the wild-type protein, while there is a significant improvement in the case of the IL-4BChelix (approximately two-fold). This change in the refolding yield has been found in four different experiments in which the bacteria were grown in flasks, 10L and 100L fermentors, thus are not the results of particular expression conditions but rather reflect a general improvement in the refolding process of the reduced protein. HPLC analysis of the refolding process for the three proteins shows that in the case of the IL-4BChelix mutant a major peak corresponding to a non-productive folding intermediate is absent. This could indicate that stabilisation of a-helix C relatively destabilises a folding intermediate leading to the formation of a non-native disulfide bridge.

The rational design strategies described in this work could provide a useful means to optimise the sequences of this helix in different polypeptides, particularly cytokines, for foldability in vitro or in heterologous hosts. The ability to produce cytokines in large amounts, would also make possible the isotopic labeling of members of the family whose three-dimensional structure is not yet known with 15N,

and particularly with"C (Markley and Kainosho, 1993), at a reasonable cost.

Labeling strategies would also allow the investigation of the dynamic properties of these proteins, which may shed some light on the flexibility of the functional epitopes and on the mechanisms used for receptor binding.

Biological implications Biochemical and structural studies are often hampered by the limited quantities of protein available. The ability to produce proteins in large amounts is also the first prerequisite to the development of proteins with clinical significance into commercial pharmaceutic drugs. IL-4 is one of the main players in the allergic response and antagonists of the protein produced by mutating residues involved in binding to a low affinity component of the dimeric receptor system are of high therapeutic interest. However, like most cytokines, IL-4 and its variants refold in vitro in very low yields, making the production of the protein laborious and expensive.

Two-fold improvement in the in vitro refolding yield of 1L-4 have been introduced by enhancing the helical propensity of helix C, using rational design. The stabilisation of this indicates that the optimisation of a-helices results in an increase in protein stability and in the acceleration of the folding reaction. We suggest that the stabilisation of a-helices in proteins may represent a general strategy to improve in vitro refolding yields.

METHODS FOR EXAMPLES 1 TO 4 AGADIR design of the helix stabilising mutations. The design of mutations that enhance helical propensities, which has been carried out based on the following principles: i) The target sites for mutagenesis are selected in an a-helix as those fully solvent exposed positions in which the side chain does not interact with residues not included in the a-helix, and are not required for activity. The intention is to guarantee that the protein packing is maintained, that the mutations do not eliminate medium or long-range interactions and that the activity of the protein is maintained. ii) The mutations are designed to enhance a-helix propensity as indicated by a helix/coil transition algorithm tuned for heteropolypeptides, like AGADIR (Munoz and Serrano, 1997; Lacroix, 1998). Essentially the algorithm is modified to predict instead of helical content for a peptide the stability for a

particular sequence in an a-helical conformation. The template sequence used corresponds to the X-ray a-helix plus four residues on each side to prevent end effects. Several sequences differing by the residues at the positions targeted for mutagenesis are evaluated by the programme.

Cloning. The IL-4 gene has been obtained by PCR from a previously described plasmid R'5prC109/IL4 (Kruse, 1991). NocI and BamHI restriction sites were introduced at the 5'and 3'ends, respectively, using the following oligonucleotides: Oligo5': CTG GAG ACT GCC ATG GAT CAC AAG TGC GAT ; Oligo3' : A CGC GGC TCC TTA TCA GCT CGA ACA. The gene coding for the wild type IL-4 and for the mutant proteins was inserted into PBAT4 (Peranen, 1996).

Due to the introduction of a Ncol site at the 5'end, wild type IL-4 and the mutant proteins were expressed with an additional amino acid (Asp) at the N-terminus.

Site-Directed mutagenesis. The mutants IL-4W91 S and IL-4BChelix were obtained by PCR (Ho, 1989) using oligo 5'and oligo 3'as flanking sequences, and the following mutagenic primers: IL-4W91Sa : AGG AAC CTC AGT GGC CTG GCG GGC TTG. IL-4W91Sb : CAA GCC CGC CAG GCC ACT GAG GTT CCT.

IL-4BChelixa : CTG GGT GCG AGT GCA GCA GAA GCA AAC AGG CAC AAG C. IL-4BChelixb : G CTT GTG CCT GTT TGC TTC TGC TGC ACT CGC ACC CAG.

Protein expression. For shake flask fermentations Escherichia coli AD494 (DE3) (Derman, 1993) was transformed with the plasmids. 1L flasks with LB medium were inoculated with a single colony and incubated on a shaker at 37°C.

After an OD600 of 0.5 was reached IPTG was added resulting in a final concentration of 0.16 mM. The culture was incubated overnight and cells harvested by centrifugation.

In order to produce larger amounts of protein, fermentations in a 10 and 100 L bioreactor were performed. For these fermentations the plasmids were transformed into E. coli W3110 (DE3). The cells were grown by batch fermentation at 37°C in a complex medium (30 g/L soya peptone, 20 g/L yeast extract, 20 g/L glycerol, 5 g/l KH2PO4, 1 g/l MgSO4, 100 mg/l ampicillin) to an OD6oo of 3. 0. At this stage expression of the recombinant protein was induced by the addition of 0.4 mM IPTG. After an induction phase of 4 hours cells were harvested by

centrifugation.

Protein purification. The harvested cells were resuspended in 25 mM Tris- HC 1 pH 8.0. Cell disruption was performed enzymatically by the addition of lysozyme (1 mg/g cell dry weight) and incubation at room temperature for 30 min.

Released inclusion bodies were harvested by centrifugation at 8000 g (30 min). The pellets were washed four times by resuspension in 0.1 M Tris-HCl pH 8/1 mM EDTA/0.1% zwittergent and centrifugation (8000 g, 15 min). Washed inclusion bodies were solubilised in 8 M GnHCl/0. 1 M Tris-HCl, pH 9. SH-groups were chemically modified to S-S03 groups by the addition of excess sulfite and tetrathionate as described by Kella and Rurella (1985) and Kella (1988). Refolding was carried out at a protein concentration of 200-300 mg/L, by cross-flow ultrafiltration against five volumes of 50 mM Na2BO4 pH7, lmM EDTA, 0.4 mM L-cysteine, 0.6 mM L-arginine. In-process analysis was performed by RP4-HPLC on a Vydac C4 resin, applying a linear gradient of Trifluoroacetic acid-Acetonitrile.

Peptide synthesis. The peptide were synthesised on polyoxyethylene- polystyrene graft resin in a continuous flow instrument. Peptide chain assembly was performed using Fmoc chemistry (Carpino, 1972) and in situ activation of amino acid building blocks by PyBOP (Coste, 1990). The peptide was purified by reversed phase HPLC and characterised by laser desorption mass spectrometry (MALI).

Circular Dichroism. The far UV CD spectra of the peptides were recorded, on a Jasco-710 instrument, at 278 K, in a cuvette with a 2mm path. Measurements were made every 0.1 nm, with a response time of 2s and a bandwidth of I nm, at a scan speed of 50 nm/min. The peptide concentration, calculated from the absorbance at 280 nm (Pace, 1995), was 25 uM. The helical percentage was calculated from the mean residue ellipticity at 222 nm, taking into account the peptide length (Chen, 1974), according to the following: % Helix = 100 #obs222nm/(39 500(1-2.57/n) eq.1 wherein n is the number of residues in the peptide and 9°222nm is the ellipticity of the peptide at 222 nm.

Thermal denaturation. The thermal denaturation was measured on a Jasco- 710 instrument, by monitoring the change in the CD signal at 222 nm over a

temperature range of 6-95°C in a cuvette with a 2mm path. Measurements were made in 0.5 degrees increments, with a response time of 2s and a bandwidth of 1nm, at a temperature slope of 50 ° C/h. The protein concentration calculated from the absorbance at 280 nm was 10 M. The measurements were done in 25mM Na2HPO4/2M GdnHCl pH 6.5.

Chemical denaturation. GdnHCl denaturation and renaturation experiments were carried out at 25°C, in 50 mM Na2HPO4, pH 7.0. The protein concentration used was 7 µM for IL-4 wild type and for IL-4W91 S, and 5 uM for IL-4BChelix.

The Jasco automatic titration system was used to mix the denaturant and the protein.

Unfolding and refolding of the protein were monitored following the change in the CD signal at 222 nm. The ellipticity readings were normalised to fraction unfolded using the standard equation: Funf=(#-#N)/(#D-#N) eq. 2 where 0 is the ellipticity value at a certain concentration of denaturant, and ON and AD stand for the ellipticities of the fully native and fully unfolded species at each denaturant concentration, and were calculated from the linear regression of the pre and post-unfolding baselines. Assuming a two state model, the equilibrium constant for denaturation, at each denaturant concentration, can be calculated using the equation below: KD= (F,, rF) l (F F e.3 Where FN and FU represent, respectively, the fraction of fully native and fully unfolded protein obtained by the linear fitting of the baselines preceding and following the transition region. It has been found experimentally that the free energy of unfolding in the presence of GdnHCl is linearly related to the concentration of denaturant (Pace, 1986): <BR> <BR> <BR> #GD=#GDH2O - m [GdnHCl]<BR> <BR> <BR> <BR> <BR> <BR> eq. 4 The proportionality constant m reflets the cooperativity of the transition and is believed to be related to the difference in surface exposed to the solvent between the native and the denaturated states. When all these dependencies are taken into

account, the change is ellipticity as a function of the concentration of denaturant, can be fitted to the following equation: Funf=FN + a[GdnHCl] + {[FU + b[GdnHCL]) {exp((m[GdnHCL]-#GDH2O)/RT) /(1+exp((m[GdnHCl]-#GDH2O)/RT))}} eq.5 in which the dependence of the intrinsic ellipticity upon denaturant concentration, in both the native and the denatured states, is taken into account by the terms of a [GdnHCI] and b [GdnHCl], respectively (linear approximation (Santoro and Bolen, 1988)).

NMR Spectroscopy. NMR samples of EL-4 and the mutant proteins were prepared by dissolving the lyophilised protein in 45 mM deuterated sodium acetate (NaOAcd4) pH 5. 3 10 % D2O, 0. 1 mM TSP, 0.2% NaN3 to give a protein concentration of 500 I1M. The spectra were recorded at 303 K. The wild type and mutant peptides of helix C were analysed in 25 mM Na2HP04 pH 6.0,10% D20 and 0.1 mM TSP, 0.2% NaN3 at a peptide concentration of 1.5 mM, at 295 K. All spectra were acquired on Bruker DRX 500 and 600 spectrometers. Standard pulse sequences and phase cycling were employed to record two-dimensional (2D) TOCSY (total correlation spectroscopy) and NOESY (nuclear Overhauser enhanced spectroscopy) spectra.

Binding assays. The binding affinities were measured by surface plasmon resonance using a BIAcore 2000 (Pharmacia Biosensor). A recombinant extracellular domain of the receptor a-chain ([C 182A, Q206C] IL4-BP) was immobilised at a biosensor CM5 to a density of 1500 to 2000 pg/mm2, as described by Shen (1996). Ligand binding was analysed at 25°C by perfusion with HBS buffer (10 mM Hepes, pH 7.4/150 mM NaCl/3. 4 mM EDTA/0.005% surfactant P20) plus 0.5 M NaCl at a flow rate of 50 Ill/min.

EXAMPLES 5 AND 6 Examples 5 and 6 describe the inventors'studies of an 81-residue polypeptide, the activation domain of human procarboxypeptidase A2, ADA2h. This domain has two a-helices packed against a four-standed (i-sheet (Catasus, 1995; Garcia Saez, 1997; Aloy, 1998). The protein has been found to fold at a neutral pH in a two-state manner through a compact transition state, possessing some secondary

structure and a rudimentry hydrophobic core (Villegas 1995, Villegas, 1998). The inventors have compared the wild-type polypeptide (WT-ADA2h) with three mutants in which one (Helix 1 in Ml-ADA2h and Helix 2 in M2-ADA2h) or both a-helices (DM-ADA2h) have been stabilised by several amino acid replacements (Villegas, 1997). The stabilising mutations on the solvent-exposed face of ADA2h a-helices are the following: a-helix 1 (Ml-ADA2h) : N25/K, Q28/E, Q32/K, E33/K (Villegas, 1996). a-helix 2 (M2-ADA2h) : Q60/E, V64/A, S68/A, Q69/H (Villegas, 1996). The double mutant (DM-ADA2h) contains the substitutions corresponding to both M1 and M2 (Villegas, 1997).

Example 5: Comparing tendency for aggregation in wild type and mutant polypeptides At pH 3.0 the three mutants Ml, M2 and DM show fully reversible unfolding transitions (in a concentration range lOM to 3mM) and at high temperatures reveal a CD spectrum typical of a denatured polypeptide (Fig. la & I c-1 e). The WT polypeptide, however, undergoes a transition that is not reversible (Fig. la). The CD spectra at 95 ° C, and after cooling to 25 ° C, show that the polypeptide has converted to a conformation with extensive (3-sheet structure (Fig. lb). Analytical centrifugation analysis indicates that the sample is highly aggregated (40% of the sample showed a Sw, 20 of 35 S, thus indicating a Mr higher than 106 Da, corresponding to an aggregate that is on average more than a 100 mer). The conversion of a-helical structure to P-sheet for the WT-ADA2h polypeptide upon thermal denaturation is corroborated by FTIR spectroscopy. Before heating, the amide I band shows two main components at 1622 and 1649 cm 'respectively (fig.

If), attributable to (3-sheet and a-helical structure respectively (Surewicz, 1993), and consistent with the native state of the polypeptide. After incubating the sample at 90°C two new bands appear, centred at 1615 and 1685 cmi'respectively, replacing the original bands (fig. If). This pattern is normally associated with aggregated species with (3-pleated structures (Fraser, 1992; Gasset, 1992; Fabian, 1993b). The band at 1615 cni 1 is indicative of (3-sheet whereas the one at 1685 cm''is associated with a splitting in the amide I band due to antiparallel inter-strand interactions (Krimm and Banddes, 1986; Fabian, 1993a). The polypeptide after heating also produces a clear shift in the Congo red absorbance spectrum from 486 to 500 nm,

together with an increase in absorption, and exhibits thioflavine-T binding (Table 1).

All these properties are consistent with the formation of amyloid deposits by the WT-ADA2h polypeptide.

The behaviour of the WT and DM polypeptides was also examined following incubation in the presence of a range of urea concentrations (4 to 7 M). At 7 M urea the CD spectra of both polypeptides are indicative of highly unfolded species (decrease in ellipticity at 222 nm). When samples of both WT and DM-ADA2h in 7M urea were diluted rapidly to a concentration of 1M urea, the proteins refolded to their native states. Incubation of the WT polypeptide in 4 M urea for lh, followed by dilution to 1M urea and incubation for lh, however resulted in a CD spectrum indicative of extensive (3-sheet structure (Fig. 2a). No such behaviour is found with the DM polypeptide (Fig. 2b) which recovered its native spectrum after the same successive equilibration steps that promote the appearance of ß-sheet structure in WT-ADA2h. In accordance with this, DM-ADA2h was found to be fully digested by pepsin after the successive equilibration steps in urea, as well as after thermal denaturation. The WT polypeptide, however, showed a reduced susceptibility to digestion in samples where an increase in (3-sheet structure had been detected (Table 2). Indeed, a clear correlation exists between the transition to (3-structure revealed by CD and FT-IR and an increased resistance to proteolysis.

Example 6: Aggregate analysis Analysis of the aggregated WT polypeptide by electron microscopy shows clear evidence for fibrils that are long, unbranched, narrow (diameter 30-100 A) and apparently quite flexible. These typically form tight networks of intertwined structures. Samples observed after long periods of incubation, 48h, at 90°C feature longer and more regular fibrils than those incubated for short times lh and show more clearly a ribbon-like pattern twisting at irregular intervals. The structural changes resulting from longer periods of incubation at high temperatures can be explained as a slow reorganisation of the protofilaments which make up the fibrils.

Fibrils are observable in aggregates formed from solutions containing WT-ADA2h concentrations as low as 20 ßM. Fibrils with characteristics similar to those shown with a shorter incubation time were also observed in samples of WT-ADA2h subjected to chemical denaturation, a result consistent with appearance of resistance

to proteolysis under similar conditions. As expected, fibrils could not be detected from any sample of the DM mutant subjected to similar conditions of thermal and chemical denaturation.

Fibrils from thermally denatured preparations of WT-ADA2h were also characterised by X-ray fibre diffraction, and found to have the distinct cross-p-ray diffraction pattern, characteristic of amyloid fibrils. (Blake and Serpell, 1996; Sunde, 1997). Reflections are observed at 4.7 A (corresponding to the inter-strand distance in the direction of the fibril axis) and 9.3 A (corresponding to the distance between p-sheets in the direction perpendicular to the fibril axis). Clear anisotropy was observed in the sharp 9.3 A diffraction (the equatorial axis is the vertical axis in the display). Another faint reflection was observed at 3.1 A, with the anisotropy and sharpness of the 9.3 A reflection, probably arising from a harmonic. A fourth weak reflection is observed at 3.8 A with no apparent anisotropy. Reflections of this type have been observed previously in studies of amyloid fibrils form a variety of sources (Sunde, 1997).

ADA2h, therefore, constitutes an example of a polypeptide not associated with any known disease that is able to aggregate in the form of amyloid fibrils.

Remarkably, however, the mutations in helices 1 and 2 of ADA2h decrease substantially to propensity of this polypeptide to aggregate. That cannot be correlated with the overall stability of the native state as aggregation was initiated under conditions where both WT and mutant polypeptide are at least partially denatured. Thus it is shown that the decreased propensity to aggregate is associated with the increased a-helical propensities of the modified polypeptides. The stabilisation of a-helical structures favours local intra-chain interactions relative to inter-chain ones in the denatured state, and therefore hinders aggregation of the polypeptide.

The present study suggests, therefore, that mutations that stabilise regions of the secondary structure of ADA2h, rather than simply the overall global fold, inhibit its aggregation into amyloid fibrils. It is clearly essential for the native state of a protein to be disrupted in order to allow ordered aggregates to form. Thus, it is entirely reasonable that strong correlation will exist between mutations that destabilise a protein structure and its tendency for form fibrils. But the propensity

for the aggregation process to take place must also depend on the intrinsic properties of the denatured state that results from the disruption of the native fold. Mutations that change the properties of the denatured state are not expected to correlate in a simple manner with the structure or even the stability of the native protein, as is the case in the mutations shown here. This conclusion suggests an explanation for the existence of disease-related amyloidogenic mutants which do not simply cause a decrease in protein stability.

It is well established that mutations to native proteins can increase amyloidogenic behaviour. The present results with ADA2h show for the first time that aggregation of a native protein into amyloid fibrils can be reduced by modifications that maintain unchanged the native function, structure, and even stability of the protein.

Gene therapy could be used to engineer into an organism modified proteins with a. reduced tendency to aggregate, or to produce transgenic animal strains that are resistant to prion-diseases. In addition, polypeptides with industrial or medical importance, but having a high propensity to aggregate, such as insulin and calcitonin (Sluzky, 1992; Arvinte, 1993; Moriarty, 1998), could perhaps be engineered to be less aggregation prone by altering their secondary-structure propensities using a similar approach to that used with the ADA2h protein.

Table 1. Thioflavine-T binding of ADA2h samples, before and after thermal denaturation at pH 3.0. All the samples were prepared in 25 mM glycine, pH 3.0 and incubated for 30 minutes at the indicated temperatures.

Sample conditions Wild Type a Double Mutant 25°C, 500 vit 3.1-0.8 90°C, 500 I1M 520.1-0.9 90°C, 200 go 288.7 (721.8) b- 90°C, 100 mu 111.5 (557.4) 2.3 90°C, 50 M 64.3 (642.5)

a Results show the increase in fluorescence (arbitrary units) after subtracting the Thioflavine-T contribution alone (45.2 units).

Data in brackets represent the value normalised for the same concentration.

Table 2. Resistance in proteolysis of the WT-ADA2h aggregates. The data are given as the percentage of the intensity in the chromatographic peak corresponding to the native polypeptide remaining after treatment.

Temperature denatured Urea denatured untreated Severe 5. 0% 0. 7% 0% proteolysisa 39.7% 7.4% 1.5% Mild proteolysis' ° Severe conditions. Ratio ADA2h: pepsin 100: 1 (W: W), 2 h digestion, 20°C b Mild conditions. Ratio ADA2h : pepsin 400: 1 (W: W), 15 min digestion, 0°C MATERIALS AND METHODS FOR EXAMPLES 5 AND 6 ADA2h expression and purification. The WT-ADA2h and its stabilised mutants were expressed and purified as previously reported (Villegas, 1996 ; Villegas, 1997, Villegas, 1998). All of the recombinant proteins were examined by MALDI-TOF-MS and found to have the molecular weight anticipated from the sequence.

Circular Dichroism. CD spectra of 20,80,160 and 200 M polypeptide samples in 50 mM sodium phosphate (pH 7.0) or 25 mM glycine (pH 3.0) were recorded using a JASCO-710 spectropolarimeter, at 278,298 and 368 °K in a 2.0 or 0.2 mm quartz cuvette. Measurements were averaged for 30 scans recorded at 50 nm min-'. Thermally induced unfolding of 20 LM protein samples was monitored in the temperature range of 278-368 K at a heating rate of 50°C h-'by following their ellipticity at 222 nm or 214nm. Calculations of residual helical structure at 95 °C

were carried out by measuring [0] at 220 nm (Chen et al., 1974). To prevent artefacts due to protein aggregation, wild-type protein spectra were collected at a concentration of 5llM, and sample was heated from 25 to 95°C at a rate of 150°C h-l.

Sedimentation analysis. Sedimentation experiments were performed in a Beckman XLA analytical ultracentrifuge at 3000g with 20 and 200 FM polypeptide samples in a buffer solution at pH 3.0 containing 25 mM glycine. Samples were heated at a rate of 50°C from 5 to 95°C and then left at 95°C for 10 min. A 200 M polypeptide sample in 50 mM sodium phosphate at pH 7.0 was used as negative control.

Thioflavine-T and Congo red binding assays. Samples of WT-ADA2h and DM-ADA2h (at concentrations ranging from 20 to 500 jj. M) were incubated for 30 min at 90 °C before the assay. A 2.5 mM Thioflavine-T stock solution was freshly prepared in 10 mM potassium phosphate, 150 mM NaCl, pH 7.0, and passed through a 0.2 pm filter before use. Typically 10 1 of sample was diluted in the reaction buffer (10 mM potassium phosphate, 150 mM NaCI, pH 7.0) containing 65 M Thioflavine-T (1 ml final volume). Samples were taken into and out of the pipette several times to facilitate fibril dispersion and to disrupt large aggregates.

Data were collected in a Perkin-Elmer LS 50B luminescence spectrometer using a 440 nm (slit width 5 nm) excitation wavelength and 482 nm (slit width 10 nm) emission wavelength. Fluorescence values were obtained after 3-5 min to ensure thermal equilibrium had been achieved. Samples were continuously stirred to prevent signal oscillation due to the presence of large fibrillar aggregates and signals were averaged for 60 s to increase the signal-to-noise ratio. Congo red binding assays were performed according to Klunk et al. (Klunk, 1989a; Klunk, 1989b).

Aliquots of 10 1 of sample were diluted in 10 mM potassium phosphate, 150 mM NaCI, pH 7.0 containing 5 µM Congo red (1 ml final volume). Congo red solutions were prepared just before its use and passed through a 0.2 llm filter. Absorption spectra of samples in the reaction solution were collected together with negative controls (dye in the absence of polypeptide and polypeptide samples in the absence of dye) to subtract the signal associated with the absorption of the dye and the scattering contribution to the signal.

Proteinase resistance. 20 I1M samples of WT and DM-ADA2h in 25 mM

glycine (pH 3.0) were incubated at 4 M urea, then diluted to give an urea concentration of 1M and dialysed before incubation with proteinase. Samples in the same buffer were also incubated for 10 minutes at 95 °C before proteolysis. These samples together with untreated samples of the same polypeptides were digested in the presence of pepsin at two different ADA2h : pepsine ratios (100: 1 and 400: 1) for 2 h at 20 °C or 15 min at 0 °C respectively. Digested samples were then analysed by RP-HPLC in a Vydac C4 column (214TP54,5 urn particle size, 300 A pore, 1.0 x 25 cm) with a linear gradient from 10 to 52 % of acetonitrile. Detection was carried out at 214 nm in a Waters 994 model.

Fourier-transform infrared spectroscopy. Infrared spectra were recorded in a Bio-Rad FTS 175C FT-IR spectrometer equipped with a liquid N2-cooled MCT detector, and purged with a continuous flow of N2 gas. 5001lM WT-ADA2h samples were prepared in 2H2O, glycine 25 mM, p2H 3.0 (electrode reading was corrected for isotope effects), and spectra were collected at 25 ° C before and after incubating the sample at 90 °C for 30 min. Polypeptide solutions were placed between a pair of CaF2 windows separated by a 12 Fm Mylar spacer. For each sample 256 interferograms were collected at a spectral resolution of 2 cm~'. Spectra were collected under identical conditions for the buffer solution in the absence of polypeptide and subtracted from the spectra of the polypeptide samples. Second derivatives of the Amide I band spectra were produced to determine the wavenumbers of the difference spectral components.

Electron microscopy. Samples were applied to Formvar-coated nickel grids (400 mesh), negatively stained with 2% uranyl acetate (w/v), and viewed in a JEOL JEM1010 transmission electron microscope, operating at 80 kV.

X-ray diffraction. Fibril suspensions were washed with Microcon-100 ultrafiltration tubes (Amicon), to eliminate salts and buffers that could interfere with the X-ray measurements. Samples were prepared by air-drying salt-depleted ADA2h-WT fibril preparations between two wax-filled capillary ends. The capillaries were separated slowly while drying, to favour fibril orientation along the stretching axis. A small stalk of fibrils protruding from one end of the capillaries was obtained. The sample was aligned in a X-ray beam, and diffraction images were collected in a Cu Ka rotating anode equipped with a 180 or 300 MAR-Research

image plate (MAR Research, Hamburg, Germany) during 20-30 min. Images were analysed by using IPDISP and MarView software.

Example 7: Rational design of a new form of PI3-SH3 unable to aggregate into amyloid fibrils We selected a domain previously studied, the PI3-SH3 domain (SH3 domain of bovine phosphatidylinositol-3'-kinase) for rational design of new polypeptide variants, to engineer a non-helical protein having modified aggregation properties.

There are two major motivations to study this protein: 1) its native structure is all- beta (2 p-sheets of 3 and 2 strands respectively arranged in a 0-barrel feature) and 2) the availability of important data concerning the amyloidogenic properties of the domain and the structure of amyloid fibrils formed by PI3-SH3.

In order to decide which residues of the protein should be modified we ran a prediction with the software AGADIR (Munoz and Serrano, 1997; Lacroix et al., 1998) to identify particular sequence regions with a certain intrinsic helical propensity. The analysis yielded three major regions with a certain propensity (though rather low in all three of them) (fig. 3).

The first region was chosen given its higher intrinsic propensity to facilitate the design of a new variant with the required characteristics.

The second requirement for the design was to identify residues in the mentioned region whose side chains are fully solvent-exposed and that do not participate directly in the p-scaffold (fig. 4). Several proposed mutants fulfilling these requirements were analysed using the software AGADIR. We focussed in a first stage on those that involved 1, 2 and 3 substitutions in order to minimise major effects on the native global structure and/or stability. With the same goal in mind the introduction of highly hydrophobic residues was avoided as much as possible. After running several series of predictions with AGADIR we chose a mutant with two point substitutions in residues 17 and 23 (fig. 4) in which the intrinsic helical propensity seemed much more stabilised. The mutant of choice therefore was E17R, D23R, that is substitution of native glutamic acid in position 14 and aspartic acid in position 23 by arginine in both cases. Predictions on the intrinsic helical propensity of the fragment modified indicate an increase for the 17R/23R mutant when compared to the wild type protein (fig. 5).

The mutant was elaborated following standard recombinant DNA protocols, and the protein was expressed. In order to be able to compare the results obtained with the modified protein a similar batch of wild type protein was expressed and purified at the same time under identical conditions.

Samples of protein at a concentration of 5 mg/mL where incubated under standard conditions of amyloidogenesis (Guijarro et al., 1998; Jimenez et al., 1999; Chamberlain et al., 2000) at pH 2.0 and 37°C. Under these conditions formation of amyloid fibrils was expected for the wild type protein during the first 3 or 4 days.

Aliquots of both samples were analysed after 4 weeks of incubation and analysed by electron microscopy, thioflavine-T binding and ultracentrifugation.

Analysis of the samples by EM and thioflavine-T binding clearly indicated that there was no aggregation in the sample containing the E17R/D23R mutant, whereas the one prepared with the wild type protein exhibited characteristic fibrils (table 3). Quantification of the amount of protein present in the supernatant fraction clearly corroborates the absence of aggregation in the sample with the mutant protein.

Table 3. Aggregation of PI3-SH3 wild type and mutant form after 4 weeks of incubation at 37°C and pH 2.0 Thioflavine-T binding (a. u.)' % protein in supernatant2 Wild type 163.3 42.3 E17R/D23R mutant 10.9 97.4 control 9.6 96.9 'Signal of the dye alone with no protein was subtracted from all the samples 2% of protein in supernatant after centrifugation at 300000 g for 1 hour.

In order to discard any relationship between the behaviour of the protein and its stability we analysed the denaturation profile of both wild type and mutant proteins in the presence of guanidine. The thermodynamic parameters obtained from such analysis clearly show that the mutant protein is somewhat destabilised compared to the wild type (table 4). Therefore any effect on the aggregation properties of the protein does not come from a stabilisation of the native structure, but from an

alteration of the intrinsic propensities of the polypeptide chain. On the other hand, from the denaturation pattern (specially the co-operativity coefficient m) indicates that the mutated protein is a compact one with a defined tertiary structure.

Table 4. Thermodynamic parameters for GndHCl denatured PI3-SH3 both wild type protein and E17R/D23R mutant. dGH2o (Kcal mol-1) meq (Kcal M 1) c ( Wild type 3. 13 0. 55 2.50 0. 38 1.25 E17RJD23R mutant 2. 09 0. 65 2.61 0. 44 0.80 As a conclusion we showed that it is possible to rationally engineer a non- helical protein using the same methodology that was used before with the ADA2h domain and IL-4 protein in order to prevent it from aggregating. Even though the mutant form displays a lower thermodynamic stability (due to the increase of non- native local propensities), it is a compact protein that keeps the main structural features of the native protein but lacks the aggregation capabilities of the original polypeptide. Therefore we can conclude that the design of a new polypeptide unable to aggregate into amyloid fibrils has been a success, by using the same methodology already employed with the ADA2h and IL-4 polypeptides. Moreover it seems that native helical structure is not required in the original polypeptide in order to apply the methodology.

Example 8 We investigated the helical propensity and aggregation properties of wild type human prion helix-3 and four Creutzfeldt Jakob disease-related mutants: E200K, R208H, V201I, Q217R.

Circular dichroism. CD spectra of helix C peptides were recorded using a JASCO-720 spectropolarimeter. Sample concentration was 20-100 M, and spectra were collected at 278 °K in a 1.0 mm or 0.1 mm quartz cuvette. Measurements were averaged for 5 scans and recorded at 10 nm min-'. 20 mM Sodium phosphate, 20 mM P-mercaptoethanol pH 7.0 was used as buffer for neutral pH measurements.

Samples were incubated at pH 2.0 by resuspending the peptides in 20 mM

phosphoric acid previously equilibrated with HC at pH 2.0 and 20 mM TCEP.

Electron microscopy. Samples were applied to Formvar-coated nickel grids (400 mesh), negatively stained with 2% uranyl acetate (w/v), and viewed in a JEOL JEM1010 transmission electron microscope, operating at 80 kV.

Secondary structure propensity prediction was carried out for all the PrP disease-related mutations described so far in Helix C. The protein is being reported to maintain largely intact its stability in all of these mutants, with the exception of the substitution Q217R for which a G°fold 8. 9 1.7 kJ mol'1 has been reported (Liemann and Glockshuber 1999). However the helical propensities for two mutants E200K and R208H are about half of that predicted for the wild type protein whereas those of mutants V210I and Q217R are somewhat increased (Table 5). Interestingly these two mutations correspond to residues that are clearly solvent-exposed and therefore cannot participate in long-range interactions inside the protein.

To further test the involvement of secondary structure propensities in explaining some of these CJD-related PrP mutants we have analysed the properties of a peptide expanding the hPrP helix C and the two CJD-related mutants for which their secondary structure propensity was predicted to be decreased, E200K and R208H, together with another one corresponding to the mutation V210I (figure 6, table 5).

The peptide expands the residues 198-227 of the hPrP sequence.

Table 5. AG°fold and helical propensities of CJD and GSS-related mutants in prion Helix C Mutation AAGofid Global helical Helix C local Helix C a- (mutant-wt) propensity a-helical helical reported (kJ (%) b propensity content at pH mol- $ (%) b 7. 0 (%) WT-4. 59 8. 60 11.4 E200K 0.6 : L 2. 4 3.70 4.40 7.5 R208H 6.0 2. 5 3.52 3.35 6.2 V210I 1.1 2. 6 5.11 10.99 11.8 Q217R 8.9i 1.7 5.33 11.82-

^AG°fold (mutant-wild type) as reported using mPrP. bcalculated using AGADIR Mutation AAG Helix C local Helix C a-Helix C a- (mutant-wt) a-helical helical helical reported (kJ propensity content at pH content at pH mol'1) ° (%) b 7. 0 (%) 2.0 (%) WT-8. 60 11.4 34.2 E200K 0.6 2. 4 4.40 7.5 22.2 R208H 6.0 2. 5 3.35 6.2 16.7 V210I 1. 1 2. 6 10.99 11.8 36.2 Q217R 8.91.7 11.82-- When resuspended in phosphate buffer at pH 7.0 all the peptides showed a mainly disordered conformation when analysed by CD (figure 7). However both peptides corresponding to wild-type and V210I mutant sequences display a higher helical content than that detected in the mutants E200K and R208H (see also table 5).

TFE titration of the four peptides confirmed the same trend regarding helical propensities (figure 8), and proved the theoretical predictions performed using AGADIR as true (figure 6).

Ithas been reported recently that recombinant prion protein can be converted in vitro into fibrils by either incubating the protein at low pH and reducing conditions or low pH and 1M guanidine (Jackson, Hosszu et al. 1999; Swietnicki, Morillas et al.

2000). To explore the effect of low pH on the conformation and properties of the helix C, aliquots of the four peptides were incubated at pH 2.0 in the presence of TCEP as a reducing agent. The spectra of the four peptides showed variations with concentration most likely due to aggregation (figure 9A-B), however, below 20 RM their CD features seemed to remain constant. At this concentration the four peptides exhibit a similar trend to that observed at pH 7.0. Nevertheless the wild-type and V2 1, 01 peptides exhibit a much higher helical content (figure 9A). On the contrary

the far-UV CD spectra of the mutants E200K and R208H still show a low a-helical content in comparison (figure 9A).

Samples of peptides at both pH 7.0 and 2.0 where incubated for several weeks to evaluate the presence of aggregation. Whereas none of the peptides incubated at pH 7.0 showed any changes in their far-UV spectra (data not shown), all those incubated at pH 2.0 developed clear p-pleated CD features, with a minimum centered at 214- 216 nm (figure 10). To test the nature of the aggregates in all the four peptides, samples were analysed by EM. Even though fibrillar structures are present in all peptide samples, several differences in the quality of the fibrils can be observed between them. In this way both wild type and V201I mutant show highly flexible and disordered fibrils, whereas E200K and R208H tend to form very long and straight and helical structures, suggesting a higher degree of compactness.

Example 9: A strategy to identify the regions of a protein sequence which are important in aggregation and in the processes bv which amyloid fibrils form Mutants of a protein were produced having conservative amino acid replacements distributed throughout the whole protein sequence. A mutational study was carried out on human muscle acylphosphatase (AcP), a 98 residue protein that has a simple fold with a pappapp topology, lacks complicating factors such as disulfide bridges or cofactors, and that has been shown to be capable of forming amyloid fibrils in vitro.

Choice of mutations 34 mutant proteins, with single amino acid substitutions distributed throughout the sequence of AcP, were expressed and purified. As with other globular proteins, the conformational stability of AcP is very sensitive to amino acid substitutions especially when the mutated residues are buried in the hydrophobic core. We therefore chose conservative mutations in which each residue is replaced by another of similar character but producing small cavities or insertions (e. g. Phe to Leu, Ile to Val, Ser to Thr or Glu to Asp replacements). The mutants produced were S5T, V9A, Y11F, V13A, V17A, V20A, F22L, Y25A, E29D, A30G, I33V, G34A, V36A, V39A, T42A, G45A, V47A, V51A, P54A, M61A, W64A, L65V, K67A, P71A, I75V, I78S, E83D, I86V, S87T, L89A, Y91Q, S92T, F94L and Y98Q. This set of mutants allows the polypeptide chain to be probed approximately every three residues,

allowing the involvement of the different regions of the polypeptide chain in the process of aggregation to be assessed.

Strategy to measure the rate of aggregation Amino acid replacements that destabilise the native fold of a globular protein are likely to favour aggregation because they cause unfolded or partially folded conformations of the protein to be populated. It is from such states that aggregation processes are thought to be initiated. For this reason the rate of aggregation of AcP was studied in the presence of 25% (vol/vol) trifluoroethanol (TFE), conditions under which the various AcP variants have been shown to be fully denatured and where the aggregates have been found to develop relatively slowly. The study of aggregation under these conditions allows the contribution of the conformational stability of the native protein to the aggregation process to be eliminated. Thus, the effects of the amino acid replacements on the intrinsic rate of aggregation of the polypeptide chain can be studied and compared directly.

Aggregation was initiated by incubating solutions of AcP, at a concentration of 0. 4 mg ml-', in 25% (vol/vol) TFE, 50 mM acetate buffer, pH 5.5,25 °C, for prolonged periods of time. At regular time intervals aliquots were withdrawn from the sample to quantitate protein aggregates using far-UV circular dichroism (CD), or optical tests in the presence of Congo red and of thioflavine T (ThT), as described in the Materials and Methods section. By the time of the first acquisition of experimental data, following addition of TFE, AcP is completely denatured. Within two hours AcP converts into relatively ordered aggregates observable as granules or very short fibrils by electron microscopy, possessing extensive ß-sheet structure, as revealed by CD and Fourier transform infrared spectroscopy (FT-1R), and with the ability to bind specific dyes such as Congo red and ThT. Figure II shows the time courses, for wild-type AcP, of the increase in ThT fluorescence, of the increase in absorbance at 540 nm arising from Congo red and of the decrease in the ellipticity at 216 nm arising from (3-sheet structure, following initiation of aggregation. The time dependent changes of all three spectroscopic parameters are closely similar to each other. There is no detectable lag time arising from slow nucleation processes in the aggregation kinetics studied here, allowing a straightforward measurements of the kinetics to be made. The aggregation rate, measured as described in theMaterial

and Methods section, is identical, within experimental error, for the three sets of measurements; the rates of AcP aggregation as monitored by changes of ThT fluorescence, Congo red absorbance and CD ellipticity are 5.43 0.30%, i 0.55% and 5.09 0.79% per minute (the percentages are defined relative to the overall changes in the spectroscopic values from time zero to apparent equilibrium).

This procedure enables the relatively rapid analysis of aggregation behaviour for a large number of samples provided the aggregates are sufficiently small in size to detect reproducibly using spectroscopic techniques. These straightforward experiments, therefore, allow the early stages of the aggregation process to be monitored prior to the development of mature amyloid fibrils after some days. It has been shown, however, that such early aggregates are ordered, possess many elements of the ultrastructure of amyloid fibrils, and that their formation correlates well with the subsequent development of amyloid fibrils.

One issue that needs to be considered carefully when studying processes as complex as aggregation and amyloid formation is the reproducibility of the kinetic behaviour and of the kinetic parameters resulting fro such experiments. To address this issue, the time course of ThT fluorescence resulting from aggregation was measured seven times for wild-type AcP. The resulting kinetic traces are highly reproducible and yield values of 5.43% and 0.79% min-'for the mean aggregation rate and its standard deviation, respectively. This value for the standard deviation was used to assess the significance of the differences between the rates of aggregation determined for the various mutants relative to the wild-type protein.

Effect of mutations on the rate of aggregation The time courses of the change of ThT fluorescence induced by aggregation were measured for all the 34 mutants of AcP. Figure 12 shows the results obtained with various representative mutants as compared with those obtained with the wild- type protein (data for the latter are represented in all three panels as a dashed line).

Most of the mutants exhibit aggregation kinetics similar, within experimental error, to those of the wild-type protein, as is shown here for the I75V variant (Figure 12A).

But several mutants were found to aggregate significantly more slowly (Figure 12B) or more rapidly (Figure 12C) than the wild-type protein. In order to confirm that these differences were reproducible and significant, mutants with rates different from

wild-type AcP were analysed at least twice.

Figure 13 summarises the results obtained for all AcP variants. Each bar refers to the aggregation rate of the AcP variant carrying the mutation at the position reported on the x-axis.. The rate data are reported as natural logarithms of the ratio between the value for the mutant and that for the wild-type protein (ln (Vmu/VR, t)). A value of zero implies no difference between the rate of aggregation of the mutant and that of the wild-type protein, whereas values above and below zero correspond to aggregation rates higher and lower that the wild-type, respectively. The analysis of the propagation of the experimental error associated with vm,,, and vwt, described in the legend of Figure 13 and performed according to statistical principles, helps assess whether a difference between the rate of a mutant and that of the wild-type protein is significant or not. It is evident from the figure that most of the mutants of AcP aggregate at a rate which is within experimental error equal to that of the wild-type protein. This is the case for all 22 mutations studied here in the region of the sequence 1-15 and 32-86. By contrast, all six mutations in the region 16-31 (namely V17A, V20A, F22L, Y25A, E29D and A30G) and five of the six mutations in the C- terminal region 87-98 (S87T, L89A, Y91Q, S92T and Y98Q) lead to an aggregation rate significantly different from that of the wild-type protein. The sensitivity of the polypeptide chain to amino acid replacements within these regions indicates that these stretches play a critical role in the rate-determining step of the aggregation of AcP.

The study present here suggests that only the regions of the sequence encompassing residues 16-31 and 87-98 are significant in determining the rate of the early stages of amyloid formation, i. e. the conversion of the ensemble of conformations present in the denatures state to the initial aggregates. The question naturally arises as to what type of information can be obtained from these results concerning the mechanism of aggregation of AcP. The mutational approach used here provides information on those regions of the protein which play critical roles in the process of aggregation, rather than identifying those regions that are involved in the characteristic ß-sheet structure or the resulting aggregates. From the spectroscopic investigations performed using CD and FT-IR, it is clear that a significant proportion of the AcP sequence is in a P-sheet conformation in the

amyloid fibrils and even in the initially formed aggregates. Regions other than those found to be important in the kinetic process examined here are likely to be part of the (3-sheet structure of the aggregates, they are simply less relevant in the rate- determining steps for their development. The results therefore indicate that the regions of the sequence 16-31 and 87-98 are involved in the intermolecular association of different AcP molecules during growth of the aggregates or in the stabilisation of any amyloidogenic species that form prior to self-assembly of the polypeptide chains. Such specific information regarding the rate-determining steps of the process is important for the design of strategies aimed at retarding and inhibiting the process of amyloid formation.

Hydrophobicity and p-sheet propensity determine aggregation kinetics Analysis of the particular amino acid substitutions that cause the aggregation rate to be altered provides important details concerning the aggregation process. All the mutations retarding aggregation (V17A, V20A, F22L, Y25A, L89A, Y91Q and Y98Q) involve substitution of one residue with another having a decreased hydrophobicity (as defined by Roseman, 1988) and a decreased intrinsic propensity to form (3-structure (as defined by either method of Chou and Fasman, 1978 or that of Street and Mayo, 1999). Two accelerating mutations (S87Y and S92T) involve addition of an extra methyl group at the position of the mutation and favour (3-sheet structure. The two accelerating mutations within the regionl6-31 (E29D and A30G) alter the hydrophobicity of the protein to a small extent, but more dramatically they decrease considerably the a-helical propensity of the polypeptide chain (Munoz and Serrano, 1994a, b). Hydrophobicity and a-helical propensity within key regions emerge therefore as the determinants of the aggregation process for AcP. This implies that the intermolecular association of AcP molecules determining the rate of the aggregation process takes advantage of both hydrophobic interactions and the formation of p-sheet structure.

Materials and Methods for Example 9 Mutants production and purification Site-directed mutagenesis and purification of AcP mutants was carried out as described previously (Taddei et al., 1996b). DNA sequencing and electrospray mass spectrometry were used to confirm that the desired mutations were present in each

case. All proteins used in this study have the cysteine at position 21 replaced by a serine residue. For simplicity, the protein carrying only the C21S mutation is referred to as wild-type AcP and the variants used in this study, all carrying the C21 S substitution in addition to the mutation of interest, are referred to as single point mutants. Thioflavine T, Congo red and trifluoroethanol were all purchased from Sigma-Aldrich.

Aggregation kinetics probed by thioflavine T fluorescence Aggregation of AcP and its variants was initiated by incubating the protein at a concentration of 0.4 mg ml-'in 25% (vol/vol) TFE, 50 mM acetate buffer, pH 5.5,25 °C. The development of protein aggregates was monitored mixing aliquots of 62 ul of this solution, withdrawn at regular time intervals, with 438 ul of 25 uM ThT, 25 mM phosphate buffer, pH 6.0, and measuring the fluorescence at 480 nm (using an excitation wavelength of 445 nm). This test is based on the increase of ThT fluorescence on binding to ordered aggregates (LeVine in, 1995). Fluorescence measurements were carried out using a Shimadzu RF-5000 spectrofluorimeter thermostated at 25 °C. As the rate of aggregation was found to be substantially dependent on the concentration of protein, the protein concentrations of the stock solutions of the various AcP variants were checked immediately before use with W absorption using an E2so value of 1.49 ml mg-l cm-I. The e280value for AcP mutants with tyrosine or tryptophan residues replaced with others was calculated as described previously (Gill and von Hippel, 1989).

Aggregation kinetics probed by Congo red binding The development of aggregates by wild-type AcP was also probed by measuring the increase in absorbance of Congo red at 540 nm (Klunk et al., 1989). The aggregation process was initiated as described above. Aliquots of 62 pi ère withdrawn at regular time intervals and mixed with 438 VI of a solution containing 10 FM Congo red, 150 mM NaCl, 5 mM phosphate buffer, pH 7.4. The absorbance of such mixtures was measured at 540 nm after 2-3 minutes of equilibration.

Aggregation kinetics probed by CD The time-resolved development of aggregates was followed under the conditions described above by measuring the decrease of 216 nm ellipticity using a thermostatted Jasco J720 spectropolarimeter and a cuvette of 1 mm path length.

Determination of the rate of aggregation Since the study presented here consists of a comparative analysis of the aggregation processes of various AcP variants, it is important to quantify the rate of aggregation in each case. We determine the rate of aggregation (v) immediately after the nucleation delay, i. e. at the stage when the aggregation rate is maximal. For the aggregation process of AcP this corresponds to the time of the first acquired data points. The first data points of a plot of ThT fluorescence versus time were therefore fitted in a linear manner in order to determine the increment of fluorescence per minute (AF) when the aggregation rate is maximal. This increment was converted into a percentage of the total fluorescence change using =100. AF/ (F-FJ where v is the aggregation rate, F,, q is the final fluorescence obtained at the end of the observed kinetics and Fo is the initial fluorescence extrapolated to time zero. The use of this equation allows the rates of aggregation of different mutants to be compared directly (particularly when the initial or final fluorescence values are different). It also facilitates the comparison of rate measurements obtained with different spectroscopic probes.

REFERENCES Aloy et al, (1998) Biol Chem 379 (2), 149-55.

Arvinte et al (1993) J. Biol. Chem 268,6415-22 Blake & Serpell (1996) Structure 4 (8), 989-98.

Booth et al, (1997). Nature 385 (6619), 787-93.

Canetetal, (1999). Biochemistry 38, 6419-6427 Carpino and Han (1972) J. Organ. Chem. 37,3404-3409.

Catasus etal, (1995) J. BiolChem 270 (12), 6651-7.

Chakrabartty and Baldwin (1995), Adv. Protein Chem. 46,141-176.

Chen, et. al. (1974) Biochem. 13,3350-3359.

Chitietal, (1999). Proc Natl Acad Sci USA 96 (7), 3590-3594.

Chou, P. Y. and Fasman, G. D. (1978) Adv. Enzymol. Relat. AreasMol. Biol., 47, 45-

148.

Cohen et al, (1998). Annu Rev Biochem 67,793-819.

Coste, et. al. (1990) TetrahedronLetf. 31, 205-208.

Derman, et. al. (1993) Science 262,1744-1747.

Dobson (1999) TIBS, 24: 329-332 Dobson et al, (1999). Curr Opin StructBiol 9, 89-91.

Fabian el al, (-1993a) Applied Spectroscopy 47 (9), 1513-1518.

Fabian et al, (1993b) Biochem. Biophys. Res. Comm. 191 (1), 232-239.

Fraser et al, (1992) Biochemistry 31 (44), 10716-23.

Garcia Saez et al, (1997) EMBO J 16 (23), 6906-13.

Gasset et al, (1993) Proc Natl Acad Sci USA 90 (1), 1-5.

Ghetti et al (1996) Brain Pathol. 6 (2), 127-45.

Gill, S. C. andvonHippel, P. H. (1989) Anal. Biochem., 182,319-326.

Guijarro et al (1998). Proc Natl Acad Sci USA 95 (8), 4224-8.

Ho, et. al. (1989) Gene 77,51-59.

Hurle et al (1994). Proc Natl Acad Sci USA 91 (12), 5446-50.

Jackson, G. S., L. L. Hosszu, et al. (1999). Science 283 (5409): 1935-7.

Kella, et. al. (1988) J. Prot. Chem. 7,535-548.

Kella and Kinsella (1985), J. Biochem. Biophys. Methods 11,152-263.

Kelly (1997) Structure 5 (5), 595-600.

Klunk etal, (1989b) J. Histochem Cytochem 37 (8), 1293-7.

Klunk et al, (1989a) JHistochem Cytochem 37 (8), 1273-81.

Krimm & Bandekar (1986) Adv Protein Chem 38,181-364.

Kruse, et. al. (1991) FEBSLett. 286,58-60.

Lacroix, et. al. (1998) J. Mol. Biol. 284,173-191.

LeVine in, H. (1995) Amyloid : Int. J. Exp. Clin. Invest., 2,1-6.

Liemann, S. and R. Glockshuber (1999). Biochemistry 38: 3258-3267.

Litvinovich et al (1998) J. Mol Biol 280 (2), 245-58.

Markley and Kainosho (1993), NMR ofMacromolcules : A practical Approach (ed.

Roberts GCK, New York) 101-148.

McCutchen et al (1995) Biochemistry 34 (41), 13527-36.

Moriarty et al (1998) Biochem. Biophys. Res. Commun. 245,344-8

Munoz and Serrano (1994) Nat. Struct. Biol. 1,399-409.

Munoz and Serrano (1997) Biopoly 41,495-509.

Munoz (1994a) J. Mol. Biol. 245,2 Munoz (1994b) J. Mol. Biol. 245,297-308 Munoz, ef. al. (1994c) Biochem. 33,5858-5866.

Munoz (1994d) Protein. Struct. Funct. and Gen. 20,4.

Munoz and Serrano (1997) Biopolymers 41,495-509 Nikolva et al, (1998) Proc Natl acad Sci USA 95, 14675-14680.

Pace (1986) Methods Enzymol. 131,226.

Pace, et. al. (1995) Protein Science 4, 2411-2423.

Pakula and Sauer (1990) Nature 344, 363-4.

Peranen, et. al. (1996) Anal. Biochem. 236,371-373.

Roberge et al, (1995) Science 269,202-204 Roseman, M. A. (1988) J. Mol Biol., 200,513-22.

Sambrook et al (1989) Molecular Cloning : A LaboratoryManual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY Santoro and Bolen (1988) Biochem. 27,8063-8068.

Shen, et. al. (1996) Eur. J. Biochem. 240,252-261.

Shortle (1996) FASEB J. 10,27-34.

Sluzky et al (1992) Biotechnol. Bioeng. 40,895-903.

Street, A. G. and Mayo, S. L. (1999) Proc. Natl. Acad Sci. USA, 96,9074-9076.

Sunde et al, (1997) JMol Biol 273 (3), 729-39.

Surewicz et al, (1993) Biochemistry 32 (2), 389-94.

Swietnicki, W., M. Morillas, et al. (2000). Biochemistry 39 (2) : 424-31.

Taddei, N., et al. (1996) Biochemistry, 35,7077-7083.

Tan et al, (1994). Histopathology 25 (5), 403-14.

Villegas et al, (1997) Fod Des 2(1), 23-33.

Villegas et al, (1995) Biochemistry 34 (46), 15105-10.

Villegas et al, (1996) FoldDes 1, 29-34.

Villegas et al, (1998) JMB 283, 1027-1036.