The design of the protein examined in this study is illustrated in Figure 1. Terminally blocked oligoalanine is nucleated over ^DPro₆-Gly₇ and ^DPro₆-^LAsp₇ dipeptide structures as Type II’ β-hairpin. The sequence positions 3 and 10 were mutated stereochemically and conformationally as D residues; this furnishes the desired boat- shaped fold. Ala residues are mutated to possible other protein residues, except Pro and Gly. Thus minimum energy sequences specific for the desired fold are generated computationally. Sequences were optimized with β-sheet favoring residues compatible with targeted fold. Polar residues were planned over solvent-accessible base of the fold and neutral-aromatic residues were planned in solvent-sequestered molecular cleft. The two sequences H1 and H2 are differ in having Gly₇ or ^LAsp₇ as the second corner residue in their β-turns. Hydrogen bonds, ion-pairs, hydrophobic contacts, p-p, and cation-p interactions, recruited for possibility of locking the heteropolypeptide as a protein, are listed in Table 1. Neutral-aromatic and cation-aromatic residues provided in molecular pocket of the fold are for possible interaction with pNPP (p-nitophenyl phosphate) as the binding ligand based on p-p, cation-p, and ion-pair interactions. H1 and H2 were synthesized and evaluated for folding, and ligand binding function. A1 is oligoalanine analogous to H1. In addressing ^DPro₆-^LAsp₇ dipeptide for possible role in folding of H1 and A1, the structure is mutated to ^LPro₆-^LAla₇ in the oligoalanine analog A2.

A1:Ac-Ala₁-Ala₂-^DAla₃-Ala₄-Ala₅-^DPro₆-Asp₇-Ala₈-Ala₉-^DAla₁₀-Ala₁₁-Ala₁₂-NH₂

H1:Ac-Arg₁-Glu₂-^DAla₃-Ser₄-Tyr₅-^DPro₆-Asp₇-His₈-Asn₉-^DThr₁₀-Lys₁₁-Trp₁₂-NH₂

H2:Ac-Arg₁-Glu₂-^DAla₃-Ser₄-Tyr₅-^DPro₆-Gly₇-His₈-Asn₉-^DThr₁₀-Lys₁₁-Trp₁₂-NH₂

A2:Ac-Ala₁-Ala₂-^DAla₃-Ala₄-Ala₅-^LPro₆-Ala₇-Ala₈-Ala₉-^DAla₁₀-Ala₁₁-Ala₁₂-NH₂

Figure 1. Stepwise design involving stereochemical mutation of β-hairpin as boat-shaped fold and inverse optimization of H1 as receptor for p-ni- trophenyl phosphate.

H1 and H2 were synthesized by solid-phase peptide synthesis. Requisite ion peaks appearing in MALDI mass spectra, as noted in Figure S1 (Supplementary Material), characterize the structures. The peptides display ¹H NMR resonances broadly in accordance with their structures as is noted in Figure S2 (Supplementary Material). In concentration regime of the NMR experiment (0.25 - 2.5 mM), peptides were soluble and manifested no noteworthy changes in chemical shifts or line widths on tenfold dilution (results not shown). The structures thus appear to be freely soluble contrasted with poly-L hairpins, which tend to aggregate. The stereochemical modification of hairpin by bending the structure appears to have suppressed tendency of natural hairpin to aggregate to a diminished aqueous solubility.

CD spectra recorded for peptide H1 Figure 2. A maximum of ellipticity at ~198 nm and a coupled minimum at ~208 nm suggests ordering of the peptides as β-hairpins. Another minimum of ellipticity at ~215 coupled with a maximum at ~228 nm appears and could be coupled exciton due to involvement of aromatic residues in p-p interaction. As is noted in Figure 1, the boat-shaped fold has aromatic side chains clustered in its molecular cleft, which may promote aromatic-aromatic interactions within or between strands of the bent hairpin. The bend in artificial hairpin has provided for the interactions that are atypical for canonical poly-L β-hairpin. Ordering of the structures was tested in a thermal unfolding experiment, which was monitored with CD. Ellipticity at 228 nm diminishes with increase of temperature and recovers fully on cooling, as is noted in the inset of Figure 2. This suggest that Peptide H1 may fold and unfold like proteins.

H1 and H2 were tested for binding with p-nitrophenyl phosphate (pNPP). On titration with pNPP, both peptides manifest strong quenching of fluorescence as is noted in Figure 3 and Figure S3 of Supplementary material. Analysis of the data with Stern-Volmer equation gave the binding constants given in Table 2. Peptides were evaluated for ligand binding with AutoDock. The evaluation was undertaken with central member of the largest cluster in H1 populating equilibrium modeled with MD, as we shall discuss. The result of the calculation is in Table 2. The calculated binding energy is in reasonable agreement with the experimentally determined value. Aromatic side chains in molecular cleft provide a specific site for ligand binding conforming to quenching of fluorescence on ligand binding. On combined evidence of reversible folding, and ligand binding, H1 and

H2 are justified as protein models.

H1, A1, and A2 were submitted to molecular dynamics in a solvent bath of water to sample their conformational ensemble. Gromos 96 force field modified for accommodation of D residues is implemented at 298 K using SPC-water model as the solvent. Trajectories were monitored in conformational phase space of polypeptide structure, which was mapped to discrete microstates using 0.15 nm RMSD cut-off over backbone atoms to dis-

tinguish conformationally specific clusters. The clustering was implemented with Duara et al. algorithm [41] ; the central member of each cluster was taken to model a specific microstate, viz., as a discrete fold populating equilibrium. Time evolution of population in microstates during MD is compared in Figure 4. H1 equilibrates early and saturates as an ensemble of 6 microstates. A1 and A2 evolve slowly and do not manifest a robust asymptote in evolution of microstates even after 250 ns of MD. However, assuming equilibria to be reasonably approximated, the simulations were concluded at this point in time. Compared to 6 microstates in H1, A1 and A2 are noted to populate in, respectively, 523 and 983 microstates (see Table 3). Clearly, sequence length is critical for the ability to compute equilibria. Side chains and β-turn fold H1. β-turn in ^DPro₆-^LAsp₇ structure also restricts conformational diversity in A1; mutated to ^LPro₆-^LAla₇ structure, A2 is nearly two-fold greater in density of microstates than A1. According to mole fraction in minima of energy, 0.96 in H1, free energy change in ordering of the structure in its ensemble is −7.8 kJ∙M⁻¹. Likewise, 0.13 and 0.21 in mole fraction, A1 and A2 have their energy minima ordered with free energy change of, respectively, 4.7 kJ∙M⁻¹ and 3.3 kJ∙M⁻¹. Relative to A1, the energy minima of H1 manifest net free energy change of −12.5 kJ∙M⁻¹; this reflects sequence contribution in ordering of H1.

H1, A1, and A2 were assessed in distribution of Rg over the conformers populating macrostates, in percentage occupancy of specific f, y basins, and in percentage occurrence of specific main chain hydrogen bonds, short (SR), medium (MR), and long ranged (LR). The data are summarized in Table 4 and Figure S4 of Supplementary Material. Appearnece of Rg maxima of entire macrostate at 0.54 nm for H1 and at 0.52 for A1; and maxima of first microstate at 0.55 nm for H1 and at 0.52 nm for A1, suggest more compaction of A1 than H1. Macrostate in H1 is ~92 % in β + PPII basin and ~6% in a basin, while in A1 it changes to ~80% occupancy in β + PPII basin and ~10% in a basin, as is illustrated in Figures S5 of Supplementary Material. According to residue-level basin occupancies (Figure S6 of Supplementary Material), Pro₆ and Asp₇ are locked in Type II’ β-turn and only marginally dispersed in A1. ^LPro₆ and ^LAla₇ in A2 are considerably dispersed and accordingly, stereochemical effect of D-proline and interactions of Asp side chain contribute in ordering both A1 and H1.

According to the data in Table 4, H1 is locked with 8.7 hydrogen bonds per molecule of which 5.0 are with in backbone, 2.8 are with backbone-sidechain (mc-sc), and 0.9 within different side chains. A1 lacks hydrogen bonding groups in Ala side chains and has mc-mc hydrogen bonds diminished to 3.4 from 5.0 in H1. In H1, ~61% of mc-mc hydrogen bonds are LR type (n-n ≥ 6), implying an ordered β-sheet structure, while ~32% are MR type (n-n ± 3), conforming to its β-turn structure. In its conformational dispersal, A1 diminishes in LR hydrogen bonds to 55% from ~61% in H1, and increases in MR hydrogen bonds to ~35% from ~32% in H1.

Interactions ordering H1 were assessed by monitoring distribution of Rg over specific side chain pairs. Specific pairs with Rg distribution peaking at ≤0.3 nm, noted in Figure 5, are strongly associated. Specifically, Trp₁₂- Tyr₅, His₈-Tyr₅, and Trp₁₂-His₈ are closely interacting, and could be involved in p-p interaction, and Lys₁₁-Glu₂ and Ser₄-Asn₉ are closely interacting, and could be involved in mutual hydrogen bonds. The appearance of peak at 0.25 nm in the Rg distribution of ^DAla₃-^DThr₁₀ indicate the presence of hydrogen bond between ^DThr₁₀ side chain hydroxyl and ^DAla₃ main chain carbonyl. Arg₁-Tyr₅, a case of potential cation-p interaction, and Asp₉-His₈, a case of potential hydrogen bond, are bimodal in Rg distribution (Figure 5), which implies that the side chains only interact transiently. Arg₁-Trp₁₂ pair, a possible case for p-p and cation-p interaction, has Rg distribution peaking at 0.65 nm; clearly the side chains do not interact.

H1 was assessed in its interactions with solvent. The radial distribution of water oxygen against specific backbone and C_β atoms are presented in Figure 6 and against other specific side chain atoms are presented in Figure S7 of Supplementary material. The overall spatial distribution plots of water oxygen are presented in Figure S8 of Supplementary material. As evidenced in reduction of radial distribution function (RDF) magni-

^aHydrogen bonds are short (SR; i → i ± 2), medium (MR; i → i ± 3 + i → i ± 4) and long ranged (LR; i → i ± 5 + i → i ± ³6 according to sequence separation between donor and acceptor residue. ^bBasin definitions are, α: ^L/Df = −/+ 20 to −/+ 100, ^L/Dy = −/+ 20 to −/+ 80; β: ^L/Df = −/+ 90 to −/+ 170, ^L/Dy = +/− 80 to +/− 180; PPII: ^L/Df = −/+ 30 to −/+ 90, ^L/Dy = +/− 80 to +/− 170. mc-mc indicate main chain-main chain; sc-sc indicate side chain-side chain; mc-sc indicate main chain-side chain hydrogen bonds whereas p-p indicate the total hydrogen bonds observed in the entire peptide.

tudes relative to equilibrium solvent density of 1, H1 and A1 have specific backbone atoms sequestered from solvent possibly in intramolecular hydrogen bonds. The greater density of main chain-main chain hydrogen bonds in H1 than in A1 and A2 support the observation of radial distribution. RDF peaks of water oxygen against oxygen and nitrogen atoms of specific side chains display maxima typically at ~0.3 nm, indicating the strong solvation of polar atoms of side chains. RDF maxima of water oxygen are diminished against β carbons of aromatic and aliphatic side chains, conformed to solvent sequestered nature of non polar side chain groups and their participation in p-p interaction and hydrophobic clustering.

Figure 7 shows ribbon representation of top five microstates in each ensemble. The individual populations, shown in parenthesis, add up to ~99% in H1 and ~40% in A1 and A2. The microstates in H1 and A1 have the general appearance of bent hairpins, which are well locked in H1 and floppy in A1. From f, y plots, shown alongside, H1 and A1 are noted to have ^DPro and ^LAsp locked in Type II’ β-turn. Remarkably, many microstates

in A2 also appear to be in hairpin-like folds. Presumably ^LPro induces chain reversal to promote the folds. An effect contributing specific folds may involve L, D, L segments ordering to a curved morphology due to ^Lβ, ^Dβ, ^Lβ conformation, in practically all the microstates, according to f, y plots. Identical in curvature, the strand sections are well poised for mutual antiparallel β-sheet hydrogen bonds in curved hairpins. Indeed, A2 is comparable to A1 in the number of backbone hydrogen bonds at ~3.5 (see Table 4), and is significantly higher in LR hydrogen bonds, at ~70%, implying more extensive β-sheet structure, compared to 60% in H1 and only

55% in A1. Thus the identical curved morphology of alternating L, D, L structures may constrain conformational diversity in H1, A1, and A2 due to intramolecular main chain hydrogen bonds between strands facilitated by chain reversal in the models. H1 is well locked in its first microstate with aromatic interactions, interactions of several cross-strand side chains, and in intramolecular main chain hydrogen bonds. Aromatics are clustered in all the microstates of H1; polydispersity of microstates involves considerable fraying of Arg as N-terminal residue. Due to a local twist in main chain, Asp₇ is pushed out of a basin uniquely in microstate 2 of the ensemble.

Peptides were modeled either with the in-house software package CAPM (Computer Aided Peptide Modeling), capable of handling D-amino acid effectively. In-house program PDB make was used for generation of PDB coordinates of CAPM modeled structure. Sequences were designed with help of in-house sequence optimization program IDeAS, capable of handling D-amino acid effectively.

Molecular dynamics were performed with gromos-96 43A1 force field in GROMACS 3.3.3 [42] [43] in a periodic box of with water as explicit solvent. The simulation was performed under NVT condition [38] . We used 1.4 nm cut off for Non-bonded list with 0.8 nm shift and used 2 fs as integration step. Initial velocities were drawn from Maxwellian distribution. Temperature was coupled to an external bath with relaxation time constant of 0.1 ps. Bond lengths were constrained with SHAKE [44] to geometric accuracy 10⁻⁴. Electrostatics were treated with Particle Mesh Ewald [45] [46] for charged system whereas SHIFT was used for neutral peptides. We implemented a 1.4 nm coulomb cutoff, 0.12 nm fourier spacing, and 4 as an interpolation order. Peptides constrained to the center of the periodic cubic box were surrounded by solvent water to 1 atm density at 298 K. We first minimize the energy of solute followed by minimization of solvent energies while restraining solute, and finally both were energy minimized after removing restraint. We started the molecular dynamics simulation and sampled the trajectory at 10 ps intervals. We have discarded initial 3 ns trajectory as pre-equilibration period, before analyzing the data.

We used Daura et al. Algorithm for clustering peptide conformers in cartesian space with ≤0.15 nm RMSD cutoff over backbone atoms. This out put different microstates in their diminishing population, viz., diminishing thermodynamic stability. We calculated the free energy of first microstate (most populous) using equation DG = −RT ln K, where K = p₁/p_total − p₁, where is R gas constant, and T is temperature, and p₁ is the population of the first microstate. We considered the most populated first microstate as ordered state and evaluated its stability with respect to the remaining microstate considered as unordered state.

We calculated the radial distribution and spatial distribution of specific solvent atoms around peptide for first microstate with g_rdf and g_spatial functions in GROMACS.

For docking the ligand to peptide, we used in build flexible docking algorithm of AutoDock 4.0 [47] . The representative structure of first microstate was obtained by clustering the three aromatic residues over entire trajectory. This structure was used as a receptor structure for docing with ligand. Using genetic algorithm with RMSD tolerance of 2 Å, structurally distinct conformational clusters of the ligand were ranked in terms of increasing energy. The observed the lowest energy of peptide-ligand complex were reported as the binding energy.

Pepetides were synthesized by solid phase peptide synthesis using Fmoc chemistry on Rink Amide AM resin with HOBt/DIC as coupling reagents [48] . Each coupling, monitored with Kaiser and chloranil tests were periodically performed to check the coupling of each amino acid. 30% (v/v) piperidine-DMF was used for deprotection in each step of synthesis. Acetylation of N-terminus were achieved with Ac₂O:DIPEA:DMF in 1:2:20 ratio. Reagents K (82.5% TFA/5% dry-phenol/5% thioanisole/2.5% ethandithiol/5% water) was used for simultaneous deprotection of side chain and final cleavage of peptide chain from resin. The finally cleaved peptides were precipitated in anhydrous diethyl ether. The precipitated peptides were then lyophilized in 1:4 H₂O:^tBuOH solution, resultant peptides were stored in freeze. Peptide purity was assessed with HPLC over RP-C18 (10 mM, 10 mm × 250 mm; Merck) eluting with CH₃CN\H₂O (0.1%TFA) 0% - 100% gradients.

Mass spectra were recorded either by MALDI-TOF (Matrix Assisted Laser Desorption Ionization-Time of Flight) mode on AXIMA-CFR Kratos instrument.

¹H NMR spectra were recorded on 800 MHz Bruker instrument at 298 K in 90% H₂O/10% D₂O in citrate buffer at pH ~3 with 2.5 mM and 0.25 mM concentrations of peptides. Solvent was suppressed with pre-saturation or WATERGATE sequence, as provided in Bruker softwares.

Far-UV Circular Dichroism (CD) spectrum were recorded on JASCO J-810 CD instrument at 298 K in 0.2 cm path length quartz cell. Using 2 nm bandwidth and scanning speed 100 nm/min with 1.0 s time constant in 1 nm steps, we record five scans and averaged them. Each spectra was corrdcted for solvent absorbance. We finally report the values in molar residue ellipticity [q_MRW] by converting the observed values in millidegrees using well reported equation.

Fluorescence spectra were recorded on a Perkin Elmer LS-55 spectrofluorimeter. We collectec the data at 298 K in 1 mL cell by exiting the sample at 280 nm and recording the emission in the wavelength range of 300 - 500 nm range, with 5 nm excitation and emission slits width. A scan rate of 100 mm/min with 1 nm steps were used. We kept the fixed concentrations of peptide 20 mM and varied the substrate (pNPP and pNPA) concentration in the range of 0 - 400 mM. All experiments were performed in 20 mM Tris-HCl buffer at ~7.5 pH. We calculated Stern-Volmer constant (K_SV) for the external quencher i.e. pNPP using the following biomolecular quenching equation.

where I₀ = fluorescence intensity in the absence of external quencher, I = fluorescence intensity in the presence of quencher, Q = concentration of the quencher, and K_SV = Stern-Volmer constant calculated from the slope of line. The emission maximum intensities of tryptophan were fit as a function of pNPP concentration to the described 1:1 binding isotherm and K_d and hence binding energy were estimated.