Engineering of peptide b-sheet nanotapes† Amalia Aggeli, Mark Bell, Neville Boden,* Je N. Keen, Tom C. B. McLeish, Irina Nyrkova, Sheena E. Radford and Alexander Semenov Centre for Self OrganisingMolecular Systems, School of Chemistry, T he University of L eeds, L eeds, UK L S2 9JT A set of principles are outlined for the design of short oligopeptides which will self-assemble in appropriate solvents into long, semi-flexible, polymeric b-sheet nanotapes.Their validity is demonstrated by experimental studies of an 11-residue peptide (DN1) which forms nanotapes in water, and a 24-residue peptide (K24) which forms nanotapes in non-aqueous solvents such as methanol. Circular dichroism (CD) spectroscopy studies of the self-assembly behaviour in very dilute solutions (mM) reveal a simple transition from a random coil-to-b-sheet conformation in the case of DN1, but a more complex situation for K24.Association of DN1 is very weak up to a concentration of 40 mM at which there is a sudden increase in the fraction of peptide in the b-sheet structure, indicative of an apparent ‘critical tape concentration’. This is shown to arise from a two-step self-assembly process: the first step being a transition from a random coil to an extended b-strand conformation, and the second the addition of this b-strand to a growing b-sheet.Both peptides are shown to gel their solvents at concentrations above 2×10-3 volume fraction: these gels are stable up to the boiling point of the solvents. Rheology measurements on gels of the 24-residue peptide in 2-chloroethanol reveal that the tapes form an entangled network with a mesh size of 10–100 nm for peptide volume fractions 0.03–0.003; the persistence length of the tape is 13 nm or greater, indicative of a moderately rigid polymer; the tapes are about a single molecule in thickness.The mechanical properties of the gels in many respects are comparable to those of natural biopolymers such as gelatin, actin, amylose and agarose.Self-assembly of complementary molecular components the networks in the gels formed at higher concentrations is obtained by rheological measurements. through hydrogen bonding is emerging as a novel synthetic route to linear polymeric structures. Polymer tapes have been synthesised1–3 from nucleoside-like components (triaminopyri- Peptide Design midines/triazine barbiturates) and shown to form gels in certain organic solvents.4 These are essentially self-assembling ladder The objective is to design a peptide of minimal complexity, polymers.5 Biaxial polymers are of special interest, since they which will self-assemble into elongated, antiparallel b-sheet are expected to have quite dierent physics from classical, tapes in a particular solvent. This requires a large negative uniaxial linear polymers.6 Whilst the polymer will bend freely free energy change for the transformation of a monomeric in one plane, it will be relatively rigid in the perpendicular helical/random coil peptide from solution to the end of a direction.However, at long enough length scales, out of plane growing tape (Scheme 1).fluctuations are expected to lead to disc-like objects and, at It is believed that a peptide must have a minimum of six even longer length scales, to three-dimensional coils. residues to form stable b-sheet structures.8,9 The relative Consequently, solutions of tape-like polymers are predicted to stability of a b-sheet as compared to helix or random coil is form mesophases at high enough concentrations.Self-assemb- believed to stem, not from the dierences in hydrogen bonding ling tape-like polymers can be expected, therefore, to exhibit energies, but rather from the forces between the side-chains of far more complex behaviour. neighbouring amino acids and the solvation energies of these We have been exploring an alternative generic route to tape- side-chains.These interactions must be sucient to fully like polymers. This exploits the propensity of peptide chains compensate for the loss of translational and conformational to self-assemble via intermolecular or intramolecular hydrogen entropies of the peptide molecule as it is ‘frozen’ into the bonding into b-sheet structures. Such structures exist widely relatively rigid b-sheet organisation.It is also essential to as short b-sheets or barrels in proteins, and also as extended incorporate an element of molecular recognition into the side- b-sheets in silk.7 chain interactions. This is to ensure that the peptides arrange Here, we demonstrate that oligopeptides of minimal com- themselves into the requisite antiparallel tape-like structures plexity, can be designed to self-assemble in solution to form in preference to an interdigitated two-dimensional b-sheet.It long, semi-flexible, polymeric b-sheet tapes, a single molecule is also essential that the medium acts as a good solvent for in thickness. At volume fraction concentrations 0.001–0.005, the polymer tape. the tapes become entangled to form gels with viscoelastic These considerations lead us to the following working properties in some ways analogous to, but in other ways criteria for the rational design of peptides to produce b-sheet distinct from, those observed for gels of classical synthetic tapes in solution: (i ) cross-strand attractive forces (hydro- polymers. The shape and dimensions of the polymers are phobic, electrostatic, hydrogen-bonding) between side-chains, established by transmission electron microscopy, the confor- (ii) lateral recognition between adjacent b-strands to constrain mation of the peptides and their self-assembly (secondary and their self-assembly to one dimension, and avoid heterogeneous tertiary structures) are monitored in dilute solutions by circular aggregated b-sheet structures, and (iii ) strong adhesion of dichroism spectropolarimetry, and in semi-dilute solutions by solvent to the surface of the tapes to control solubility.FTIR spectroscopy, whilst information about the properties of The work presented in this paper focuses on the behaviour of two distinctly dierent peptides K24 and DN1, which have been designed to self-assemble into b-sheet tapes in, respectively, moderately polar (non-aqueous) and highly polar (aque- † A preliminary account of this work has been published in Nature, 1997, 386, 259.ous) media. J. Mater. Chem., 1997, 7(7), 1135–1145 1135Scheme 1 Schematic representation of the self-assembly of six-residue peptide molecules to form a growing intermolecular antiparallel b-sheet tape. Hydrogen bonding between backbones of adjacent peptide b-strands is indicated.The design of the 11-residue peptide DN1, CH3CO-Gln- Gln-Arg-Phe-Gln-Trp-Gln-Phe-Glu-Gln-Gln-NH2 [Fig. 1(b)], is not based on any native protein. Rather it has been rationally designed to form b-sheet polymer tapes in water. The (-CH2-)2 moieties of the six glutamine residues are expected to provide attractive intermolecular hydrophobic interactions between side-chains.The residues Phe4, Trp6 and Phe8 are also hydrophobic, but they are also expected to provide intermolecular recognition by p–p interactions.12,13 Arg3 and Glu9 provide an additional degree of recognition via their strong coulombic attraction,12 and favour an antiparallel alignment of the strands. Gln, Arg and Glu side-chains make one surface of the b-sheet more hydrophilic than the other.Chemically blocked termini were used to avoid edge-to-edge coulombic attractions between tapes. (a) (b) Fig. 1 Space-filling models of (a) K24 and (b) DN1 peptides in an extended b-strand conformation. Colour code: white=hydrogen, red= Materials and Experimental Methods oxygen, blue=nitrogen, yellow=sulfur, black=carbon.Peptide synthesis Standard automated solid phase methods were employed for The primary structure of the 24-residue peptide K24, NH2- Lys-Leu-Glu-Ala-Leu-Tyr-Val-Leu-Gly-Phe-Phe-Gly-Phe-Phe- the synthesis of the peptides. The synthesis of K24 was carried out as described in ref. 11 for K27. Solid-phase synthesis of Thr-Leu-Gly-Ile-Met-Leu-Ser-Tyr-Ile-Arg-COOH, [Fig. 1(a)] is related to the single transmembrane domain of the IsK peptide DN1 was also performed using Fmoc-chemistry (Fmoc: fluoren-9-ylmethoxycarbonyl). The peptide was protein.10 Its longer 27-residue version, K27, is known to readily form b-sheet structures in lipid bilayers,11 suggesting assembled on PEG–PS (polyethylene glycol–polystyrene) resin, incorporating a linker to generate the C-terminal amide upon these peptides would be good candidates for formation of bsheet tapes in amphiphilic solvents such as methanol and 2- cleavage of the peptide from the resin.Fmoc-amino acids were C-terminally activated using Hbtu [2-(1H-benzotriazol-l- chloroethanol. Furthermore, the amphiphilicity along the K24 molecule (polar side-chains predominantly near the peptide yl)-1,1,3,3-tetramethyluronium hexafluorophosphate] with DIPEA (diisopropylethylamine). The resin-bound peptide was termini, and apolar side-chains predominantly in the middle of the peptide chain) is expected to favour the alignment of N-terminally acetylated by reaction with 0.3 M acetic anhydride–0.03 M pyridine in DMF (dimethylformamide) the peptide b-strands with respect to each other, which will lead to the one-dimensional propagation of the b-sheet.Such (10 min at room temp.). Cleavage of the peptide from the resin and deprotection of amino acid side-chains were achieved by an intermolecular arrangement would thus allow the establishment of several kinds of favourable contacts along adjacent incubation for 1 h in TFA (trifluoroacetic acid), containing 1% (w/v) phenol, 2% (v/v) water, 4% (v/v) ethane-1,2-dithiol peptide b-strands: for example, intermolecular interactions between polar side-chains (such as Lys1, Glu3 and Arg24) and 2% (v/v) anisole.The peptide was precipitated with diethyl ether, centrifuged and washed a further 4 times with diethyl near the termini, between non-polar side-chains (such as Leu5, Val7, Leu8, Leu16, Ile18 and Leu20) in the central region of ether.The diethyl ether was evaporated and the peptide dissolved in water for purification by reversed-phase HPLC, the peptide chain, as well as specific interactions such as p–p interactions among the four aromatic phenylalanine rings at which was carried out using water–acetonitrile gradient in the presence of 0.1% TFA. Mass spectrometry has shown that the positions 10, 11, 13 and 14. 1136 J. Mater. Chem., 1997, 7(7), 1135–1145molecular masses are as expected (K24: m/z 2800 and DN1: m/z 1593). K24 was shown by 19F NMR and FTIR to contain ca. 4 mols of residual TFA per 1 mol peptide. Less TFA (ca. 2 mol TFA/mol peptide) is present in DN1, on the basis of FTIR. Peptides were stored in their lyophilised state. The gelation properties and self-assembly behaviour of 0.007 v/v K24 solutions as monitored by FTIR are independent of the method used to prepare the samples, i.e.starting from either a predominantly helical or a mixture of helical–b-sheet conformations in the original lyophilised sample. Experimental techniques FTIR spectroscopy. Spectra were averages of four scans, recorded with a resolution of 4 cm-1, at 20°C in an FTIR liquid cell equipped with CaF2 crystals and a 50 mm Teflon spacer, using a Perkin-Elmer 1760X FTIR spectrometer.After subtraction of the solvent spectrum, the component peaks of the peptide amide I band were obtained by second derivative analysis and peak-fitting of the absorption spectra, using a home-written program, consisting of iterative adjustment of the relative heights, widths and ratios of Lorenzian–Gaussian functions of the lineshapes of the individual components, to obtain the best fit between the spectrum calculated by the program and the experimental one.CD spectroscopy. Measurements were obtained with a Jasco J-715 spectropolarimeter using 1 mm and 1 cm quartz cuvettes at 20°C. Spectra were recorded with a step resolution of 1 nm, a scan speed of 50 nm min-1, a sensitivity of 50 millidegrees and a response time of 1 s. Each spectrum was the average of 4–10 scans.The peptide concentration in solution was determined by amino acid analysis and ninhydrin assay. Transmission electron microscopy (TEM). Fig. 2(a) was obtained using a 0.001 v/v (500 mM) K24 gel in methanol after dilution to 25 mM peptide concentration.The copper EM grids were of mesh size 300 (corresponding to grids with 300 bars per inch) and were coated with carbon films. The carbon films were glow-discharged in order to build static charges. A droplet of the peptide solution was deposited on a clean surface. Each grid was then deposited on top of the sample droplet for 1 min, so that the peptide network was adsorbed onto the surface of the carbon film.Excess sample was drained o the grid. The grid was then introduced on top of a droplet of uranyl acetate (UA) negative staining solution in water (4 g UA per 100 ml water) for 20 s and excess of the staining solution was drained o the grid. Fig. 2(c) was obtained with a starting solution of 0.001 v/v (500 mM) K24 in deionised water.The opalescent solution was centrifuged to remove the largest particles. A droplet of the supernatant was used to prepare a grid in the same way as above. The specimens were allowed to air-dry. Fig. 2(a) was obtained with a Phillips CM10 TEM at 100 kV accelerating voltage, set at 105 000±5% magnification, whilst Fig. 2(b) and (c) were obtained with a Jeol 100S TEM operating at 100 kV and set at 50000±5% magnification.The pictures were enlarged during printing to attain the total sample magni- fication which appears in the individual figure legends. Care was taken to obtain a photographic record of the specimens as Fig. 2 TEM micrographs of b-sheet structures of K24 stained with quickly as possible after sample insertion, in order to minimise uranyl acetate and adsorbed on carbon-coated, glow-discharged grids: artefacts which can be caused by long exposure of the sample (a) network of a 0.001 v/v (500 mM) K24 gel in methanol, following to high vacuum and the intense electron beam.twentyfold dilution; (b) network of a 0.001 v/v K24 gel in methanol, without prior dilution; (c) insoluble peptide in water. Total magnifi- cation: ×262 000.Rheology. A Rheometrics Dynamic Analyser II, with 25 mm diameter parallel plate geometry, was employed. The thickness of the material between the plates was between 0.5 and 0.8 mm. ethanol. Its high boiling point compared to the other gelfavouring solvents minimises the solvent evaporation rate and Gels were prepared 3 d prior to the measurements. The following precautions were taken to minimise the problem of solvent changes of peptide concentration. Secondly, the oven, which surrounds the sample, was kept closed during data collection.evaporation from the periphery of the sample between the two parallel plates. First, most gels were prepared in 2-chloro- Thirdly, two containers full of solvent were kept near the J. Mater. Chem., 1997, 7(7), 1135–1145 1137sample, inside the oven, to saturate the atmosphere with Characterisation of secondary structure solvent vapour.Furthermore, the air flow system (which is Infra-red spectroscopy. FTIR spectroscopy was used to probe used to control sample temperature) was turned o, and the the conformation of the peptide chains as well as their supra- experiments were carried out at ambient temperature.molecular organisation. Peak-fitting analysis of the amide I Temperature fluctuations of the sample were closely monitored. band of all K24 gel samples studied (which correspond to gels produced in more than 20 dierent solvents), reveals a single major component centred at 1624.5±0.5 cm-1 [Fig. 3(d) and (e), red spectra] with a half-height bandwidth Dn� equal to Results 17–19 cm-1.An amide II band centred at ca. 1530 cm-1 is General observations also observed (not shown). These features are characteristic of a stable homogeneous intermolecular b-sheet structure.14 When K24 is added to solvents such as ethanol, methanol or It was estimated from the integrated intensities of the 2-chloroethanol, at a peptide volume fraction of ca. 0.002 v/v corresponding IR components that ca. 90% of the peptide (ca. 3 mg ml-1), the dry peptide particles initially swell, and adopts b-she structure, such as those obtained in methanol, eventually coalesce to form a transparent, homogeneous gel. propanol–D2O (90510 v/v) and 2-chloroethanol, and the This procedure takes place over several hours depending on remainder adopts a mixture of random coil and helical confor- peptide concentration, solvent and temperature.The gels are mations, as inferred from the weak components at ca. 1645 stable for months at room temperature. They are optically and 1655 cm-1. isotropic, but become optically birefringent when sheared. Theoretical as well as experimental evidence has shown that These properties are indicative of the presence of a network of the ratio of the integrated intensities of the weak and strong long polymer molecules in solution.IR components at ca. 1696 and 1625 cm-1 respectively, is The apparent viscosity of the K24 gel sample is dependent equal to 0.09–0.19, if the b-sheet consists of 100% antiparallel on the peptide concentration. For example, a 0.0011 v/v K24 strands.14 Peak-fitting analysis of spectra obtained with K24 solution in methanol (1.5 mg ml-1) is fluid, whilst a 0.0074 v/v gels revealed that this ratio is between 0.075 and 0.11, which peptide solution (10 mg ml-1) is a solid-like transparent gel.is indicative of predominantly antiparallel b-sheets. The anti- Gels with peptide concentration below 0.02 v/v peptide in parallel arrangement of peptide strands allows the establish- methanol are transparent, whilst in more concentrated solu- ment of stronger interpeptidic hydrogen bonds, leading to the tions, particles of insoluble peptide aggregates are dispersed in formation of more stable b-sheets than the alternative parallel the gel.In another experiment, a 0.007 v/v K24 rigid gel in arrangement.7 Other factors, such as specific complementary methanol was diluted tenfold: after gentle shaking for a few interactions between oppositely charged groups on the N- and seconds to ensure homogeneity, a fluid solution was obtained. C-termini of adjacent peptides, could further contribute to the This reversibility of gelation is a characteristic signature of an stabilisation of the antiparallel b-sheet.entangled polymer network. Furthermore, K24 gels in 2- Polarised IR spectra of K24 gels spread on a CaF2 plate chloroethanol were found to be stable when incubated for 2 h showed that the CNO stretching vibration of the peptide in a thermostatted water bath at various temperatures from backbone at 1625 cm-1 is 13% more intense in the direction 40 to 90°C (which is close to the boiling point of the solvent).parallel to the direction of shear as compared to the direction Similarly, the gels of DN1 in water were found to be stable up perpendicular to the shear. Assuming that the long axes of the to the boiling point of water. polymers are partially orientated on the direction of shear, this result suggests that the long axis of the peptide b-strand is perpendicular to the long axis of the b-sheet polymer.Transmission electron microscopy studies A typical micrograph, obtained with a 0.001 v/v K24 gel in methanol after twentyfold dilution, is shown in Fig. 2(a). A network consisting of long polymers randomly distributed on the grid can be seen. Use of a higher peptide concentration results in higher density of polymers on the grid [Fig. 2(b)].In some areas, the polymers form complex entanglements or bundles. In a few cases, they are seen to bend and twist around each other. In areas where their density is lower, individual polymers with straight or wavy edges can be observed. Regions where individual polymers can clearly be observed were chosen to measure polymer dimensions. More than twenty measurements were made from micrographs of several K24 gel samples.The width was measured to be between 6.6 and 8.1 nm. Overlap of polymers makes it dicult to identify their ends, and to obtain an estimate of their length. Pieces of polymer between points where they meet/cross each other were found to be as long as 120 nm. K24 does not form a gel in water. Rather, it precipitates out of solution and forms fine, white, solid particles.Such particles were observed with TEM, in order to compare them with the structures in the gel. The micrograph in Fig. 2(c) is typical of such a sample. Individual polymers, scattered on the grid are observed. Their width has values between 7.6 and 8.5 nm, and Fig. 3 FTIR amide I bands of 0.004 v/v (2 mM) K24 solutions in: their length is between 38 and 76 nm, i.e.they are shorter than (a) HFIP, (b) 753 HFIP–methanol, (c) 151 HFIP–methanol, (d) meth- the ones in gel forming solvents. Moreover, they do not appear anol, (e) 83517 propanol–D2O, ( f) 154 propanol–D2O and (g) propa- to interact with each other to form bundles or networks as in nol. Colour code: red=gel, green=fluid solution, black=insoluble the case of the gel structure.Rather, a large number of polymers peptide. The band at ca. 1675 cm-1 is due to residual TFA, and it is particularly intense in polar solvents. aggregate to form numerous thick amorphous structures. 1138 J. Mater. Chem., 1997, 7(7), 1135–1145Fig. 4 Peak-fitted FT-IR amide I¾ band of 0.003 v/v (5 mg ml-1 ) DN1 in D2O. The outer black lines are the experimental and fitted spectra, respectively. Component peaks are shown in dierent colours.Assignment of component peaks (ref. 35) from right to left: (i) 1604 cm-1=arginine side chain; (ii) 1618.6 cm-1 (maximum of amide I¾)=b-sheet; (iii) 1633.9 cm-1=glutamine side-chain; (iv) 1645.5 cm-1=residual H2O; (v) 1673.5 cm-1=residual TFA; (vi) 1683.3 cm-1=antiparallel b-sheet.The half-height width of the major b-sheet component at 1618.5 cm-1 is 19.5 cm-1, whilst that of the weak peak indicating antiparallel b-sheet is 10 cm-1. In the case of DN1, a self-supporting, thermostable gel (up to at least 90°C) is produced at peptide concentrations above 0.01 v/v (ca. 15mgml-1) in water. DN1 adopts a b-strand configuration in 0.003 v/v solutions or above in D2O, as revealed by its IR spectrum (Fig. 4). The major component of Fig. 5 (a) Far-UV circular dichroism spectra of K24 in methanol as a function of peptide concentration: [h] is the mean residue molar the amide I¾ band is centred at 1619 cm-1 and has half-height ellipticity; (b) plot of [h] at 216 nm as a function of peptide bandwidth of 20 cm-1. This main peak is shifted 5 cm-1 to concentration lower wavenumbers compared to that on the spectrum of K24, due to the exchange of peptide amide protons by deuterium in the D2O solvent. A weak peak at ca. 1683 cm-1 is indicative sity of the positive band at 195 nm is indicative of the presence of a significant amount of antiparallel b-strand arrangements of a strong negative band in the region of 197 nm, characteristic in the b-sheet.of a disordered conformation. The positive band at 195 nm The absence of amide I bands (typically at 1660–1680 cm-1) and the negative band at 216 nm, which increase in intensity corresponding to turns for both peptides indicates that they with peptide concentration, are indicative of a transition to a do not adopt a b-hairpin structure but rather a straight b- b-sheet conformation at higher concentrations.The change in strand configuration. The simplest arrangement of the peptide, the shape of the spectra at ca. 10mM, but which nonetheless therefore, is an antiparallel arrangement of b-strands aligned have a minimum at 216 nm, characteristic of a b-sheet struc- perpendicular to the tape long axis, so that they grow in one ture, is likely to arise from a change in the stacking of aromatic dimension to form tapes (Scheme 1).The polarised IR results side-chains responsible for the negative band at 203 nm. The support such an arrangement. Given that the average separa- plot of the mean residue molar ellipticity in Fig. 5(b) is tion between adjacent residues in a b-strand is 0.335 nm, the indicative of the progression of the transition with concen- length of a 24 residue b-strand is 7.7 nm.This length falls into tration, but due to complexities in the nature of the spectra it the range of measured values for the width of the polymers by is dicult to relate this quantity to the absolute fraction of TEM (6.6–8.1 nm). This observation supports further the peptide in the b-sheet state. presence of tape-like polymers, consisting of extended peptide strands with the strand axis being normal to the polymer’s DN1.Similarly, experiments have been carried out on a long axis, whilst the direction of hydrogen bonding is parallel series of solutions of DN1 in water with concentrations ranging to the polymer axis. In this way, K24 peptides could form from 10 mM (15.94 mg ml-1) to 450 mM (717.08 mg ml-1) pre- tape-like structures, whose width corresponds to the length of pared by dilution of a stock solution of known concentration.a K24 b-strand and thickness to that of a single b-sheet The stock solution was birefringent after storage for 2 d. (see below). Fig. 6(a) shows the far-UV CD spectra of DN1 in water as a function of peptide concentration. At peptide concentrations Self-assembly in dilute solutions up to 40 mM, the CD spectra exhibit a distinct minimum at 200 nm.This is at a higher wavelength than the value of K24. A series of solutions of K24 in methanol were prepared with concentrations ranging from 1–20 mM by taking known 197 nm typically found with random coil structures, and is believed to stem from the presence of tryptophan and phenyl- amounts of a 2.5 mM stock solution of peptide in HFIP, evaporating to dryness, and dissolution of the resulting peptide alanine side chains which absorb strongly in this region.30 At higher peptide concentrations, a positive band develops in the film in the required amount of methanol.The far-UV CD spectra are shown in Fig. 5(a) as a function of peptide concen- place of the negative band at 200 nm, and a new minimum develops at ca. 224 nm which is attributed to b-sheet structures. tration. At peptide concentrations up to 5.2 mM, the spectra exhibit a distinctive isodichroic point at 198 nm indicative of Again this peak is at a higher wavelength than the typical value for b-sheet proteins of 217 nm, presumably due to CD a two-state conformational transition.The variation in inten- J. Mater. Chem., 1997, 7(7), 1135–1145 1139where df and elink are defined above, nlink is the eective volume of a peptide–peptide bond (it characterises how freely a bstrand can move around its neighbour), and n0 is the intrinsic volume per strand. In order that tapes are stable at some concentration, elink<df<0. The minimisation of the free energy [eqn.(2)] under condition eqn. (1) gives the formulae (3) and (4) for the fractional composition, N1 nlink V =EL , E�eelink-df (3) Nm nlink V =ALm, A�eelink (m2), (4) with ln L being the Lagrange multiplier (0<L<1). The latter can be found using the condition eqn. (1): it is determined indirectly by the total concentration of peptides, eqn. (5). N nlink V =EL+AL2C 2-L (1-L )2D (5) Finally, the fraction of peptides in tapes (with m2), i.e.the b-sheet fraction, is given by eqn. (6), wtape� 1 N .2 m=2 mNm=1-G1+A E L (2-L ) (1-L )2 H-1 (6) with L being determined by the total concentration as in eqn. (5). Thus, one can use eqns. (5) and (6) to fit the wtape versus peptide concentration data in Fig. 6(b) using coecients A and E as fitting parameters.In doing this we assumed that the Fig. 6 (a) Far-UV circular dichroism spectra of DN1 in H2O as a measured [h] are a linear function of wtape. Note that the fitted function of peptide concentration; (b) plot of fraction of peptide in b- coecients crucially depend on the value chosen for the link sheet states (wtape) as a function of peptide concentration. The continu- volume nlink: A and E being proportional to nlink.The best fit ous line represents the fit of the experimental data to eqns. (5) and (6) which is represented by the continuous line in Fig. 6(b) was using A/nlink ca. 1.125 mM and E/nlink ca. 45 mM. achieved for: A/nlink ca. 1.125 mM; E/nlink ca. 45 mM. The estimations for the energies entering eqn. (3) and (4) signals from the aromatic side chains in the same region.The depend on the choice of nlink. For nlink=1 A° 3, we obtain A ca. isodichroic point at 210 nm is indicative of a simple two state 6.75×10-10 and E ca. 2.7×10-8, and hence: elink random coil<b-sheet equilibrium. The mean residue molar ca. -21.1; elink-df#-17.4; df#-3.7 (in units of kBT ), ellipticity at 224 nm has been taken as a measure of the whilst for nlink=0.5 A° 3, we obtain A#8.4×10-11 and fractional concentration of peptide in the b-sheet state.E#3.4×10-9, therefore: Fig. 6(b) shows a slow growth of the fraction of peptide in b-sheet structures up to a concentration of 40 mM, at which it elink ca. -23.2 elink-df ca. -19.5 df ca. -3.7 (in units increases suddenly. This behaviour is more complex than of kBT ). expected for a simple one-dimensional association of peptides.Thus, the values of elink and df are only weakly dependent on In fact, the sudden jump in the fraction of peptide in b-sheet the choice of nlink. In fact, df is necessarily insensitive to it. structures is indicative of a critical concentration in the aggregation process, reminiscent of the aggregation of surfactants into sphere-like micelles in aqueous solution.It arises here Properties of semi-dilute solutions from the fact that before a peptide monomer can add to a Gelation behaviour. The conformational state and gelation growing tape (see Scheme 1), it must first undergo a transition properties of both K24 and DN1 peptides have been explored from, in this case, a random coil to an extended b-strand in a variety of solvents.Our observations for K24 are rep- conformation. The change in free energy involved df (the resented in Fig. 7, as a plot of the solvent polarity, er, as a dierence between the free energies of a coil and a b-strand), function of its hydrogen-bonding ability, a.15 though significantly smaller than the free energy change, elink, Thermostable, self-supporting, mostly transparent gels are on breaking a peptide–peptide bond in a tape, suppresses found to be stable in solutions of K24 in moderately polar aggregation at low peptide concentrations.To demonstrate solvents, such as methanol, with er in the range 25–68, a<1.5, that this is the case and to obtain values for df and elink we and peptide concentrations equal to or higher than 0.002–0.004 have developed and applied an appropriate theoretical model.v/v (region of phase diagram with red dots). For example, We consider a solution at equilibrium containing tapes 100% propanol (er=20.1) produces an insoluble peptide pre- having a distribution of sizes such that there are Nm tapes of cipitate; mixtures of propanol–D2O, with volume fraction of aggregation number m.If the total number of peptide molecules D2O 0.1–0.7 and er 26–62, produce gels whose apparent in a given volume V of solution is N, then eqn. (1) holds. rigidity decreases in mixtures of high volume fraction of D2O; .2 i=1 iNi=N (1) and mixtures of propanol–D2O, with volume fraction of D2O 0.8–1 and er 71–80, produce again progressively more insoluble The total free energy of the solution is given by eqn.(2), peptide solutions in parallel to increased D2O content. Studies in methanol–water mixtures as well as in a wide range of other mixed solvents and pure solvents also gave similar results. F=N1(df)+ .2 m=2 Nm(m-1)Celink-lnAnlink n0 BD+.2 i=1 Ni lnANin0 Ve B In less polar solvents (15<er<25) than the gel-favouring ones, the peptide is not suciently soluble to form gels (black (2) 1140 J.Mater. Chem., 1997, 7(7), 1135–1145of peptide solution in HFIP; maximum and half-height bandwidth of principal component of amide I are 1655 cm-1 and 30 cm-1, respectively, maximum of amide II at 1545 cm-1], and by CD (minima of negative ellipticity at 208 and 222 nm). A dramatic decrease in the content of the b-sheet is eected by increase of the a-value (and hence of the hydrogen-bond donor strength) of the solvent from methanol to HFIP (Fig. 8).The peptide is also completely soluble in such solvents as benzyl alcohol, o-nitrotoluene, 1,2-dichlorobenzene and benzene (greenes with numbers 27, 26, 25 and 23, respectively, in Fig. 7). The solubilising eect of these solvents is comparable to that of solvents with high a-value.It seems that aromatic groups in the solvent can interact favourably with the six aromatic rings on the peptide and thus compete with peptide– peptide interactions, causing some destabilisation of the bsheet structure (for example only 60% of the peptide is in b-sheet structure in benzyl alcohol). Control of the self-assembly process of the b-sheet polymers, Fig. 7 A correlation plot between the macroscopic properties of 0.004 as well as of the interactions between polymers, provides ways v/v (2 mM) K24 solutions in various solvents, the relative permittivity of tuning the mechanical properties of the peptide gel. One er of the solvent and its ability to act as hydrogen bond donor, a. The way that the gel-to-fluid transition can be brought about is by plot was constructed from observations of the IR spectra and mechan- varying the hydrogen bond donor strength (a) of the solvent.ical properties of the samples. The dotted lines show the boundaries This can be achieved, for example, in the case of the K24 of the gel region. Each dot represents a solvent with the corresponding er and a values. Red=gel, green=transparent fluid solution, black= methanol system quite simply by adding HFIP.The CD insoluble peptide. The a and er values of mixed solvents were estimated spectra in Fig. 9 show the sequential changes in the confor- assuming linear relationship between the two component solvents. The mation brought about by 10% increments of HFIP: spectra number next to each dot refers to a particular solvent: (1) dimethylfor- (i) (10%) to spectra (x) (100%) HFIP. Fig. 9(a) shows the mamide (DMF), (2) 753 DMF–formamide, (3) 753 DMF–methanol, conversion from the b-sheet (characteristic minima at 216 nm) (4) 357 DMF–methanol, (5) 951 propanol–formamide, (6) 151 propa- to the a-helix (characteristic minima at 208 and 222 nm) nol–formamide, (7) 951 propanol–water, (8) 83517 propanol–water, (9) 451 propanol–water, (10) 753 propanol–water, (11) 352 propanol– conformation [curves (i) to (vii)].The isodichroic point at water, (12) 151 propanol–water, (13) 253 propanol–water, (14) 357 202 nm is confirmation of the simple two state nature of this propanol–water, (15) methanol, (16) 951 methanol–HFIP, (17) 451 transition. The accompanying change in secondary structure methanol–HFIP, (18) 753 methanol–HFIP, (19) ethanol, (20) 2- is illustrated in Fig. 10. Fig. 9(b) shows a second isodichroic chloroethanol, (21) glycerol, (22) ethylene glycol, (23) benzene, point associated with a subsequent helix-to-random coil trans- (24) toluene, (25) 1,2-dichlorobenzene, (26) ortho-nitrotoluene, ition [curves (viii), (ix) and (x)]. (27) benzyl alcohol, (28) 151 HFIP–CH2Cl2 , (29) 159 methanol–HFIP, (30) HFIP, (31) formamide, (32) 357 DMF–formamide, (33) water, DN1 shows similar behaviour to K24, apart from the fact (34) 159 propanol–water, (35) 154 propanol–water, (36) 154 methanol– that its gel region is shifted upwards towards regions of higher water, (37) 151 HFIP–water, (38) 151 TFE–water, (39) 151 polarity (higher er).For example, it produces a self-supporting, methanol–HFIP, (40)357 methanol–HFIP, (41) TFE, (42) acetonitrile, thermostable gel, up to at least 90°C, and at concentrations (43) acrylonitrile, (44) acetone, (45) THF, (46) diethyl ether, of 0.009 v/v (ca. 5 mg ml-1) or above in water, by self- (47) hexane, (48) cyclohexane, (49) cyclohexene, (50) dichloromethane, assembling into b-sheet polymers.In solvents with er equal to (51) chloroform, (52) 151 methanol–CH2Cl2, (53) propanol, (54) 151 TFE–CH2Cl2, (55) hexanol, (56) butanol. or less than ca. 33, or in the very polar solvent formamide (er=109), the peptide is precipitated from solution. Similarly to the behaviour of K24, in solvents with high a-value such as circles near the bottom of Fig. 7).The fraction of b-structure in these solvents is still as high as in the gel-favouring solvents, i.e. equal to ca. 0.7–0.9. For example, the peptide which is insoluble in pure propanol (er=20.1), has an IR amide I band in this solvent almost identical to the one obtained with its gels [Fig. 3(g), estimated percentage of peptide in b-structure: 90%]. A similar situation pertains in solvents with er>68 (black circles near the top of Fig. 7). For example, the amide I band in the FTIR spectrum of a peptide solution in 20% v/v propanol–80% v/v D2O is shown in Fig. 3( f ). The amide I¾ band is centred at 1620 cm-1 and the estimated percentage of peptide in b-sheet conformation is equal to 78%. Consequently, gels are obtained in such solvents which prevent stacking and precipitation of the b-sheets.The average polarity of a K24 molecule, which was estimated to correspond to er ca. 26, may be a major contributor in determining the optimal solvent polarity required for solvation of tapes and gelation. In solvents such as 1,1,1,3,3,3-hexafluoroisopropyl alcohol (HFIP) or solvent mixtures with a1.5, clear, low-viscosity solutions are obtained (green circles in Fig. 7), even at peptide Fig. 8 Plot of the fraction of K24 peptide in helical conformation in concentration higher than 0.02 v/v. These simple Newtonian ca. 0.002 v/v (1 mM) peptide solutions, as a function of the hydrogen fluid solutions contain peptide molecules in the monomeric bonding strength (a) of the solvent. The fraction of helical conformation state. This is manifested by the absence of b-sheet and the was estimated by calculating the ratios of the intensities of the FT-IR presence of mixtures of helical and random coil peptide peaks at 1655 cm-1 (characteristic of helices) and the sum of the intensities at 1655 and 1625 cm-1 (characteristic of b-sheet).conformations, as shown both by FTIR [Fig. 3(a): spectrum J. Mater. Chem., 1997, 7(7), 1135–1145 1141Fig. 11 Typical mechanical spectrum of a 0.02 v/v (9 mM) K24 gel in 2-chloroethanol at 24.8°C, obtained with small oscillatory shear in the linear viscoelastic region (c=1%) rheological measurements on solutions of K24 in methanol and 2-chloroethanol. A typical mechanical spectrum of a K24 gel in 2-chloroethanol (0.019 volume fraction) is shown in Fig. 11. This was measured using small-strain oscillatory shear experiments (strain: 1–10%).Gels with peptide volume fraction as low as 0.006 were also studied and found to behave in a similar way. The elastic modulus G¾ is an order of magnitude larger than the viscous modulus G indicative of an elastic rather than a viscous material. G¾ and G are seen to be very weakly dependent on frequency v of the oscillatory shear for frequen- Fig. 9 Circular dichroism spectra of a 27 mM solution of K24 in HFIP– cies 10-2–102 rad s-1, implying that the relaxation time t of methanol mixtures: (a) spectra (i) (10% HFIP) to (vii) (70% HFIP); the network is very long and not reflected in the measurements. (b) spectra (viii) (80% HFIP) to (x) (100% HFIP) By comparison, for networks of linear synthetic polymers, t is equal to the time trept required for a polymer strand to reptate out of its entanglement.The longer the polymer, the larger the number of entanglements it participates in, and the longer the network relaxation time. For self-assembling polymers, the relaxation time of the network is equal to the geometric mean of two separate terms, namely the relaxation time trept due to overcoming entanglements by reptation of the polymers, and the relaxation time tbreak due to the probability of a selfassembled polymer to break.For example, in the case of wormlike surfactant micelles,22 G¾ and G cross at ca. 1 rad s-1, indicative of much shorter relaxation times compared to the peptide gel network. The long relaxation time of the peptide network is also consistent with the observation that a 0.019 volume fraction K24 gel was resistant to flow for several hours, following inversion of the sample tube. The mechanical spectrum in Fig. 11 supports the presence of stable peptide tapes which are either long and entangled or which form stable chemical crosslinks with each other. Calculations based on the rheological data to be presented below favour the presence of tape entanglements rather than chemical cross-links. Using rubber-like elasticity theory,16,17 we have been able to extract information about the mesh size of the peptide network, as well as the persistence length and the thickness of Fig. 10 Relationship between the mean residue molar ellipticity [h] at the tapes. The magnitude of G¾N0 (plateau elastic modulus) in 208 nm of the CD spectra of K24 solutions in methanol, as a function the linear viscoelastic region (Fig. 11) is related to j, the of the volume fraction of HFIP in these solutions. The more negative average distance between two nearest entanglements in space the value of the ellipticity, the higher the fraction of peptide in helical [Fig. 12(a)]: eqn. (7), conformation. Peptide concentration: ca. 0.00006 v/v (ca. 81 mg ml-1). G¾N0=gNntkBT (7) where gN is a numerical factor not far from unity, nt is the HFIP, the peptide DN1 is fully soluble, and the fraction of peptide in b-sheet is low. density of network tapes (mol of tapes per cm3), kB is the Boltzmann’s constant, and T the absolute temperature. j is in eect an upper limit to the mesh size of the network. As there Mechanical properties.To gain further insight into the structure and the dynamics of the gels, we have carried out is only one entanglement in volume j3, thus the density ne of 1142 J. Mater. Chem., 1997, 7(7), 1135–1145on the same tape is equal to or larger than j [Fig. 12(a)]. The polymer segment between two nearest crosslinks can be represented as a Kuhn chain consisting of a series of Ne segments each having a persistence length l.As a response to shear, the polymer segments between crosslinks straighten out. The strain at which the tapes are fully extended corresponds to cyield. Further increase in strain causes the tapes to break and the elastic stress to relax. When the applied strain is cyield, then the chain is fully uncoiled and the distance between the same entanglement points is L e=Nel.By definition, the shear strain Fig. 12 (a) Representation of a network tape, defined as the segment c is equal to the ratio of the length under stress and the length of the random walk tape between nearest crosslink points on the same at equilibrium, eqn. (11). tape. The crosses are individual entanglement points.For simplicity, no other network strands are shown. (b) Entanglement point of a network of polymer chains. Four network chains are involved with it, cyield# L e Re # Nel Ne1/2l #Ne1/2 (11) thus f=4. Ne was found to be 5.3±2.0 for a 0.019 volume fraction gel entanglements is given by eqn. (8) and approximately three times bigger for a 0.006 volume fraction peptide gel. From Fig. 12(a), we see that Rej. Thus, ne=1 j3 (8) since Re#Ne1/2l, we can calculate an upper limit to the persistence length of the tapes using eqn. (12). ne is related to nt by eqn. (9), l j Ne1/2 (12) nt=f2 ne= f 2j3 (9) This procedure yields l(12.7±3) nm, indicative of a moderwhere f is the number of tapes attributed to each entanglement ately rigid polymer, consistent with the intrinsic rigidity of b- [Fig. 12(b)]. sheet structures. Eqn. (7), (8) and (9) can be combined to yield eqn. (10). Since, the density of network tapes nt is related to their individual volumes Vt via nt=wp/Vt, where wp is the volume fraction of polymer in solution and Vt=L e wt=Nelwt, we can j=AgN fkT 2G3N0B1/3 (10) get from eqn. (7) an expression of the thickness of a network tape eqn. (13). Assuming that f is equal to or greater than 4, and using the magnitudes of G3N0 derived from the mechanical spectra, t= wpgNkT G3N0Nelw (13) we estimate the lower limit of j to be in the range (29–43)±13% nm, for gels with peptide volume fraction in the range of ca. 0.019–0.006, i.e. the mesh size decreases as the Using the following values for a 0.019 K24 volume fraction peptide volume fraction increases.gel: wp=0.019±0.001, gN=(1±0.2), G3N0=(328.8±111) Pa, The stress growth versus strain curve in Fig. 13 was obtained Ne=5.3±2.1, l(12.7±3) nm, T=297.8 K, and w= at a constant shear rate cÿ of 100% s-1, with a 0.019 K24 (7.35±0.70) nm, we deduce an upper limit of 0.7 nm for the volume fraction gel in 2-chloroethanol. The linear region thickness t of a tape. This value is consistent with tapes a implies that stress is proportional to strain up to a strain cyield single molecule thick (0.5<tb-sheet1.2 nm).This result is also of (230±45)%. Similar experiments carried out with a 0.006 in agreement with the notion of an entangled network of tapes. K24 volume fraction gel gave cyield ca. 415% independent of the shear rate cÿ, 50–100% s-1. These observations as well as Discussion the long network relaxation time evident in the mechanical spectra show that there is no significant network relaxation Structure and properties of tapes and networks going on during these measurements due to polymer The FTIR and CD results for K24 indicate that this peptide reptation or self-assembly.adopts an extended b-strand conformation in a pleated b-sheet At equilibrium, the distance Re between nearest crosslinks supramolecular aggregate.The electron micrographs reveal that the aggregates are elongated polymers rather than extended two-dimensional b-sheets. The analysis of the rheological measurements is also consistent with these observations and suggests that the aggregates are nanotapes a single molecule in thickness, i.e.thickness equal to 0.5–1 nm depending on the packing of the side-chains (Fig. 14). The estimated length of a 24 residue b-strand is 7.7 nm which falls into the range of the tape width (6.6–8.1 nm) measured by TEM. The rheological measurements have also provided an estimate of a lower limit of 13 nm for the persistence length of the tape, which is indicative of a moderately rigid polymer, consistent with the intrinsic rigidity of b-sheet structures.The tapes in the gels are of the order of microns in length. Whilst an extensive study of DN1 has yet to be made, the IR, CD and gelation studies suggest that it has similar structure and properties to K24. The initial concentration for gelation is of the order 0.002 volume fraction (1 mM K24). At lower concentrations, the Fig. 13 Stress–strain curve obtained with steady shearing of a 0.02 v/v polymer tapes must exist individually. The mesh size of the (9 mM) K24 gel in 2-chloroethanol at 24.8°C, at constant shear rate cÿ=100% s-1 entangled network in the gels decreases with increasing concen- J. Mater. Chem., 1997, 7(7), 1135–1145 1143gel than in the other biopolymer gels.Thus, our peptide gels are less brittle, and stronger than classical biopolymer gels. The thermal stability of biopolymer gels is shown in Table 1. The peptide gel is more stable at high temperature than most biopolymer gels, apart from alginate polysaccharide gel, which also possesses high thermal stability. This thermal stability of the peptide gels is indicative of a network of strong polymer chains.We note that other linear biological peptides, such as leucinerich (LRR) fragments of the drosophila Toll protein,23 a 28- Fig. 14 Schematic representation of an entangled network of K24 btapes in a gel, and a single self-assembled peptide nanotape. Each residue fragment of the b-amyloid protein24 and peptides vertical line of the tape represents the long axis of a peptide in a b- modelled on conserved domains of desmin,25 as well as syn- strand conformation (length of the b-strand=7.7 nm).The mesh thetic peptides incorporating non-natural chemical groups26–28 dimensions of the network correspond to gels with peptide volume have been reported to self-assemble into b-sheet polymers and fractions from 0.03 to 0.003. gel solvents. However, our aim has been to demonstrate the potential for production of b-sheet nanotapes by peptide design.tration: 10–100 nm for K24 gels with volume fraction It is also interesting that the peptide tapes show remarkable 0.03–0.003 (Fig. 14). similarity in structural terms, to protein fibrils formed in vitro The thermostability of the tapes is remarkable and presum- and in vivo from several dierent proteins includingthe polyglu- ably stems from the extensive cross-strand side chain–side tamine-containing proteins responsible for Huntington’s dis- chain interactions and the extensive network of intermolecular ease29 and the more complex structure of amyloid fibres.31 For backbone hydrogen bonds.This property, combined with the example, amyloid fibrils are rigid, non-branching structures, intrinsic chemical stability of the peptide bond, suggests that 10–15 nm in diameter, each consisting of two to five filaments these new polymers are quite robust.with cross b-structure, arranged in a twisted ribbon pattern. Thus, the peptide tapes may oer a simple model system for Comparison with other biopolymers studies of the mechanism of fibrillogenesis, which is of crucial The peptide gels (particularly the aqueous ones) in general importance in many physiological and pathological processes should be biodegradable and biocompatible. In this respect in biology.they can be compared with natural biopolymer gels18 such as gelatin, actin, amylose and agarose. The elastic (G¾) and dissi- Design and self-assembly behaviour pative (G) moduli of K24 gels are also quite similar with the moduli of most of these biopolymer gels (Table 1).Self-assembly mechanism. The results for the self-assembly behaviour of K24, depicted in Fig. 5(b), are of only qualitative The mechanical spectrum of the peptide gel in Fig. 11 is flat over the frequency range 10-1–102 rad s-1. Thus G¾ and G significance due to the complexity of the interpretation of the far-UV CD spectra [Fig. 5(a)]. However, the behaviour of are insensitive to the shear rate v, which indicates that the dominant viscoelastic relaxations of the network are at lower DN1 whose CD spectra [Fig. 6(a)] are interpreted in terms of a simple transition from a random coil to a b-sheet structure frequencies than we have measured (i.e. the relaxation time t of the network is long).This is indicative of either very long are quite similar. They are interesting in that the behaviour is more complex than a simple one-dimensional self-assembly and stable polymers which are highly entangled, or of strong, non-covalent chemical crosslinks between highly stable poly- process. This contrasts with the random coil to b-sheet transition of the Alzheimer’s b-amyloid fragment which can be mers.The latter mechanism is responsible for the shear rate insensitivity of G¾ and G of gelatin and agarose gels (Table 1). described by a simple non-cooperative one-dimensional aggregation model.32 From Fig. 6(b) we see that there is a simple b-Tape stacking would be an obvious crosslinking mechanism of the peptide polymers.However, the rheology measurements linear association of DN1 up to a concentration of 40 mM, above which there is a sudden increase in the fraction of show that the tapes are only one molecule thick, which implies that the best candidates for the crosslink points of the peptide peptide present in b-sheets, suggesting the requirement of a ‘critical tape concentration’. The behaviour is reminiscent of network are topological entanglements of the self-assembled tapes (Fig. 14). the aggregation of surfactants into spherical micelles in aqueous solution.We have shown that it has its origin in the requirement The stress response to strain for a 0.02 v/v peptide gel remains linear up to 230% strain (Fig. 13), which is much that before a peptide monomer can add to a growing tape it must first undergo a transition from, in this case, a random higher than for conventional biopolymer gels which typically break at strains of 50% (Table 1).The dierences in cyield coil to an extended b-strand conformation. The change in free energy df for this conversion, though significantly smaller than values indicate that the polymer segments between crosslinks are more flexible (i.e.contain more Kuhn segments) in the K24 the free energy change elink on breaking a peptide–peptide Table 1 Comparison of K24 with other aqueous polymer gels G¾ plateau/ temperature gel G¾/Pa G/Pa rad s-1 cyield/% stability/°C 2.5%w K24 in 2-chloroethanol 330 30 10-1–102 230 >90 2.2%w gelatin in water 40 — 10-2–102 <50 35 0.7%w pectin (polysaccharide) in water 100 ca. 2 10-3–10 <50 2%w amylose in water 700 200 1%w agarose (marine polysaccharide) in water 3500 300 10-2–102 %50 ca. 80 Alginate (marine polysaccharide) in water >100 Carageenan (marine polysaccharide) in water 20–50 2%w xanthane (microbial polysaccharide) in water %50 Worm-like micellar aqueous gel 1000 1–102 Entangled synthetic polymer network 100 1144 J. Mater. Chem., 1997, 7(7), 1135–1145bond in a b-sheet tape, is quite significant and destabilises References small aggregates. 1 J.-M. Lehn, M. Mascal, A. DeCian and J. Fischer, J. Chem. Soc., To observe the corresponding transition for K24 will necessi- Chem. Commun., 1990, 479. tate shifting the ‘critical tape concentration’ to higher concen- 2 J. A. Zerkowski, C. T. Seto, D. A. Wierda and G.M. Whitesides, trations so as to be detectable by CD spectroscopy. This could J. Am. Chem. Soc., 1990, 112, 9025. 3 J. A. Zerkowski and G. M. Whitesides, J. Am. Chem. Soc., 1994, be achieved by adding HFIP to the solution. This and other 116, 4298. experiments, aimed at controlling the pH and ionic strength 4 K. Hanabusa, T. Miki, Y. Tanaguchi, T. Koyama and H. Shirai, of the solution, are in progress.J. Chem. Soc., Chem. Commun., 1993, 1382. 5 G. J. Vroege and H. N. W. Lekkerkerker, Rep. Prog. Phys., 1992, 55, 1241. Peptide Design. The results from our studies of K24 and 6 I. A. Nyrkova, A. M. Semenov, J. F. Joanny and A. R. Khokhlov, DN1 suggest a working hypothesis for the design of peptides J. Phys. II, 1996, 6, 1411. which will assemble into b-sheet tapes in various solvents, and 7 T.E. Creighton, Proteins: Structures and molecular properties, we are currently involved in a programme of research to Freeman, New York, 1993, 2nd edn. rigorously test its validity. The results to date do however 8 M. T. Krejchi, E. D. T. Atkins, A. J. Waddon, M. J. Fournier, establish the viability of engineering nanotapes by peptide T.L. Mason and D. A. Tirell, Science, 1994, 265, 1427. 9 D. G. Osterman and E. T. Kaiser, J. Cell. Biochem., 1985, 29, 57. design. This opens up opportunities for producing materials 10 T. Takumi, H. Ohkubo and S. Nakanishi, Science, 1988, 242, 1042. with fascinating properties and applications. 11 A. Aggeli, N. Boden, Y. L. Cheng, J. B. C. Findlay, P. F. Knowles, The potential to vary the length of a peptide molecule (i.e.P. Kovatchev and P. J. H. Turnbull, Biochemistry, 1996, 35, 16 213. the tape width), the nature of its side-chains (natural and non- 12 K. C. Smith and L. Regan, Science, 1995, 270, 980. natural), and its environment, provides the means to control 13 G. D. Fasman, Prediction of Protein Structure and the Principles of the energetics and dynamics of the self-assembly process and Protein Conformation, Plenum Press, New York, 1989. 14 Yu. N. Chirgadze and N. A. Nevskaya, Biopolymers, 1976, 15, 607 consequently, the physical properties of the polymer solutions. and 627. There is also the unique opportunity to vary the structure on 15 M. J. Kamlet, J. L. M. Abboud, M. H. Abraham and R. W. Taft, opposing faces and edges of a tape.In this way tape–tape, J. Org. Chem., 1983, 48, 2877. tape–surface, and tape–ligand interactions can be controlled. 16 M. Doi and S. F. Edwards, T he T heory of Polymer Dynamics, These properties, together with the high temperature stability, Clarendon Press, Oxford, 1986. biocompatibility and biodegradability make these materials 17 J. D. Ferry, V iscoelastic Properties of Polymers, Wiley, New York, 1970.attractive for a wide range of applications. 18 A. H. Clark and S. B. Ross-Murphy, Adv. Polym. Sci., 1987, 83, 57. 19 H. McEvoy, S. B. Ross-Murphy and A. H. Clark, Polymer, 1985, Prospects for materials chemistry 26, 1483. 20 A. H. Clark, M. Watase, K. Nishinari and S. B. Ross-Murphy, It is interesting to compare the self-assembly of peptide nano- Macromolecules, 1989, 22, 346. tapes with other studies which have attempted to exploit the 21 S. B. Ross-Murphy and K. P. Shatwell, Biorheology, 1993, 30, 217. intrinsic self assembly behaviour of polypeptides as a route to 22 T. M. Clausen, P. K. Vinson, J. R. Minter, H. T. Davis, Y. Talmon and W. G. Miller, J. Phys. Chem., 1992, 96, 474. supramolecular materials. Tirell and co-workers8 have recently 23 D. A. Kirschner, H. Inouye, L. K. Duy, A. Sinclair, M. Lind and produced novel macromolecular solids by controlling the D. J. Selkoe, Proc. Natl. Acad. Sci. USA, 1987, 84, 6953. crystallisation of rationally designed polypeptides. In another 24 N. Geisler, T. Heimburg, J. Schuneman and K. Weber, J. Str. Biol., study, Ghadiri and co-workers33 have designed a ring-like 1993, 110, 205. peptide molecule, consisting of eight amino acid residues, 25 K. Hanabusa, Y. Naka, T. Koyama and H. Shirai, J. Chem. Soc., which self-assembles into long nanotubules. In another study, Chem. Commun., 1994, 2683. 26 H. T. Stock, N. J. Turner and R. McCague, J. Chem. Soc., Chem. Zhang and co-workers,34 have demonstrated that biologically Commun., 1995, 2063. derived linear oligopeptides can self-assemble into extensive b- 27 R. Vegners, I. Shestakova, I. Kalvinsh, R. M. Ezzell and sheets which condense into macroscopic membranes. Thus, a P. A. Janmey, J. Pept. Sci., 1995, 1, 371. peptide supramolecular materials chemistry appears to be 28 A. H. Clark and S. B. Ross-Murphy, Adv. Polym. Sci., 1987, 83, 57. emerging. 29 Yu. N. Chirgadze, B. V. Shestopalov and S. Yu. Venyaminov, Our work has as its focus the engineering of polymeric b- Biopolymers, 1973, 12, 1337. 30 S. Brahms and J. Brahms, J. Mol. Biol., 1980, 138, 149. sheet nanotapes. We believe it should be possible to design 31 C. Blake and L. Serpell, Structure, 1996, 4, 989. and synthesise peptides to generate an entire hierarchy of 32 E. Terzi, G. Ho�lzemann and J. Seelig, Biochemistry, 1994, 33, 1345. responsive self-assembled polymer architectures, including 33 M. R. Ghadiri, J. R. Granja, R. A. Milligan, D. E. McRee and homo- and hetero-tapes, dendrimers and cross-linked N. Kzazanovich, Nature, 1993, 366, 324. networks. 34 S. Zhang, T. Holmes, C. Lockshin and A. Rich, Proc. Natl. Acad. Sci. USA, 1993, 90, 3334. 35 W. K. Surewicz and H. H. Mantsch, Biochem. Biophys. Acta, 1988, This research was supported by the EPSRC, the Wellcome 952, 115. Trust and Schlumberger Cambridge Research. We also thank Dr P. McPhie for assistance with the electron microscopy and Mrs J. L. Johnson for assistance with peptide synthesis. Paper 7/01088E; Received 17th February, 1997 J. Mater. Chem., 1997, 7(7), 1135–1145 11