Polymorphism of collagen triple helix revealed by 19F NMR of model peptide [Pro-4(R)-hydroxyprolyl-Gly]3-[Pro-4(R)-fluoroprolyl-Gly]-[Pro-4(R)-hydroxyprolyl-Gly]3.

We have characterized various structures of (Pro-Hyp(R)-Gly)(3)-Pro-fPro(R)-Gly-(Pro-Hyp(R)-Gly)(3) in the process of cis-trans isomerization and helix-coil transition by exploiting the sole (19)F NMR probe in 4(R)-fluoroproline (fPro(R)). Around the transition temperature (T(m)), we detected a species with a triple helical structure distinct from the ordinary one concerning the alignment of three strands. The (19)F-(19)F exchange spectroscopy showed that this misaligned and that the ordinary triple helices were interchangeable only indirectly via an extended monomer strand with all-trans peptide bonds at Pro-fPro(R), Pro-Hyp(R), and Gly-Pro in the central segment. This finding demonstrates that the helix-coil transition of collagen peptides is not described with a simple two-state model. We thus elaborated a scheme for the transition mechanism of (Pro-Hyp(R)-Gly)(n) that the most extended monomer strand can be the sole source both to the misaligned and correctly folded triple-helices. The staggered ends could help misaligned triple helices to self-assemble to higher-order structures. We have also discussed the possible relationship between the misaligned triple helix accumulating maximally at T(m) and the kinetic hysteresis associated with the helix-coil transition of collagen.

The triple helical structure and stability of collagen model peptide with 4(S)-hydroxyprolyl-Pro-Gly units.

Extensive studies on the structure of collagen have revealed that the hydroxylation of Pro residues in a variety of model peptides with the typical (X-Y-Gly)(n) repeats (X and Y: Pro and its analogues) represents one of the major factors influencing the stability of triple helices. While(2S,4R)-hydroxyproline (Hyp) at the position Y stabilizes the triple helix, (2S,4S)-hydroxyproline (hyp) at the X-position destabilizes the helix as demonstrated that the triple helix of (hyp-Pro-Gly)(15) is less stable than that of (Pro-Pro-Gly)(15) and that a shorter peptide (hyp-Pro-Gly)(10) does not form the helix. To clarify the role of the hydroxyl group of Pro residues to play in the stabilization mechanism of the collagen triple helix, we synthesized and crystallized a model peptide (Pro-Hyp-Gly)(4) -(hyp-Pro-Gly)(2) -(Pro-Hyp-Gly)(4) and analyzed its structure by X-ray crystallography and CD spectroscopy. In the crystal, the main-chain of this peptide forms a typical collagen like triple helix. The majority of hyp residues take down pucker with exceptionally shallow angles probably to relieve steric hindrance, but the remainders protrude the hydroxyl group toward solvent with the less favorable up pucker to fit in a triple helix. There is no indication of the existence of an intra-molecular hydrogen bond between the hydroxyl moiety and the carbonyl oxygen of hyp supposed to destabilize the triple helix. We also compared the conformational energies of up and down packers of the pyrrolidine ring in Ac-hyp-NMe(2) by quantum mechanical calculations.

Structure and reaction mechanism of human nicotinamide phosphoribosyltransferase.

Nicotinamide (NM) phosphoribosyltransferase (NMPRTase) catalyzes the reaction of NM and 5'-phosphoribosyl-1'-pyrophosphate (PRPP) to form NM mononucleotide (NMN) and pyrophosphate (PPi) in the pathway of NAD-biosynthesis. Monitoring the (1)H and (31)P NMR spectra of the reaction mixture, we found that this reaction is reversible as dictated by the equilibrium constant K = [NMN][PPi]/([NM][PRPP]) = 0.14, which agreed well with the ratio of second-order rate constants for forward and backward reactions, K = 0.16. The crystal structures of this enzyme in the free form and bound to NM and PRPP at the resolution of 2.0-2.2 A were essentially identical to that of the complex with NMN, except for some variations that could facilitate the substitution reaction by fixing the nucleophile and the leaving group for the requisite inversion of configuration at the C1' carbon of the ribose ring. In the active site near the C1' atom of the bound PRPP or NMN, there was neither negatively charged group nor waterproof environment necessary to support the feasibility of a ribo-oxocarbocation intermediate inherent in the S(N)1 mechanism. The structures and catalytic mechanism thus revealed are also discussed in connection with the multiple biological functions of NMPRTase.

Effects of antibody affinity and antigen valence on molecular forms of immune complexes.

The effect of antibody affinity on molecular forms of immune complexes was investigated by measuring antigen-antibody interactions using surface plasmon resonance (SPR), electrospray ionization time-of-flight mass spectrometry under non-denaturing conditions (MS), analytical ultracentrifugation (AUC), and transmission electron microscopy (TEM). (4-Hydroxy-3-nitrophenyl)acetic acid (NP) of different valences was conjugated to bovine serum albumin (BSA) and these conjugates were used as antigens. In the interaction between N1G9, a low affinity antibody, and NP(7)-BSA, a 1:1 immune complex was detected as the major product and higher molecular weight complexes were not obtained by any of the methods employed. These results suggested that N1G9 predominantly formed an intramolecular divalent complex with NP(7)-BSA using the two Fab arms of an antibody. Although complexes of various sizes were detected by MS, AUC, and TEM in the interaction between C6, a high affinity antibody, and NP(7)-BSA, only 1:1 immune complexes were observed by SPR. These results showed that two NP(7)-BSA molecules cannot simultaneously bind to an antibody, irrespective of antibody affinity strength, when the Fc region is immobilized to a flexible dextran matrix on sensor chip but are able to do so with high affinity antibodies free in solution. The results also showed that the stoichiometry of the antigen-antibody interaction is altered by restricting the movement of the Fc region. Since immunoglobulins exist as antibodies in solution or as B cell receptors on the cell surface, it is suggested that interactions of B cell receptors with polyvalent antigens such as NP-BSA might be different from those of antibodies free in solution.

Organic ligand binding by a hydrophobic cavity in a designed tetrameric coiled-coil protein.

The design and characterization of a hydrophobic cavity in de novo designed proteins provides a wide range of information about the functions of de novo proteins. We designed a de novo tetrameric coiled-coil protein with a hydrophobic pocketlike cavity. Tetrameric coiled coils with hydrophobic cavities have previously been reported. By replacing one Leu residue at the a position with Ala, hydrophobic cavities that did not flatten out due to loose peptide chains were reliably created. To perform a detailed examination of the ligand-binding characteristics of the cavities, we originally designed two other coiled-coil proteins: AM2, with eight Ala substitutions at the adjacent a and d positions at the center of a bundled structure, and AM2W, with one Trp and seven Ala substitutions at the same positions. To increase the association of the helical peptides, each helical peptide was connected with flexible linkers, which resulted in a single peptide chain. These proteins exhibited CD spectra corresponding to superhelical structures, despite weakened hydrophobic packing. AM2W exhibited binding affinity for size-complementary organic compounds. The dissociation constants, K(d), of AM2W were 220 nM for adamantane, 81 microM for 1-adamantanol, and 294 microM for 1-adamantaneacetic acid, as measured by fluorescence titration analyses. Although it was contrary to expectations, AM2 did not exhibit any binding affinity, probably due to structural defects around the designed hydrophobic cavity. Interestingly, AM2W exhibited incremental structure stability through ligand binding. Plugging of structural defects with organic ligands would be expected to facilitate protein folding.

Characterization and crystallization of an IscU-type scaffold protein with bound [2Fe-2S] cluster from the hyperthermophile, aquifex aeolicus.

IscU plays a key role during iron-sulphur (Fe-S) cluster biosynthesis as a scaffold for the assembly of a nascent, highly labile Fe-S cluster. Here we report the characterization of an IscU-type protein (Aa IscU) from the hyperthermophilic bacterium Aquifex aeolicus. Unlike other known homologues of IscU, expression of Aa IscU in Escherichia coli has yielded an Fe-S cluster-containing holo-protein. Biochemical and spectroscopic studies of the wild-type Aa IscU and its Asp38-to-Ala substituted (D38A) variant molecule indicate that the holo-protein forms a trimer containing substoichiometric [2Fe-2S] cluster with its stability substantially increased by a D38A substitution. The [2Fe-2S] cluster was oxygen-labile and upon loss of the cluster, the resultant apo-form dissociated into a smaller species, a mixture of monomer and dimer with the dimer form predominating. Reddish-brown crystals of holo-Aa IscU-D38A were obtained under anaerobic conditions, that gave diffractions beyond 2.0 A resolution with synchrotron radiation. The crystal belongs to the space group P2(1)2(1)2 with unit-cell parameters a = 72.6, b = 122.3, c = 62.4 A, where the asymmetric unit contains three molecules of Aa IscU. Successful crystallization of holo-Aa IscU-D38A strongly suggests that the trimer association carrying substoichiometric [2Fe-2S] cluster represents a conformationally stable oligomeric state.

Role of Asn(2) and Glu(7) residues in the oxidative folding and on the conformation of the N-terminal loop of apamin.

The X-ray structure of [N-acetyl]-apamin has been solved at 0.95 A resolution. It consists of an 1-7 N-terminal loop stabilized by an Asn-beta-turn motif (2-5 residues) and a helical structure spanning the 9-18 residues tightly linked together by two disulfide bonds. However, neither this accurate X-ray nor the available solution structures allowed us to rationally explain the unusual downfield shifts observed for the Asn(2) and Glu(7) amide signals upon Glu(7) carboxylic group ionization. Thus, apamin and its [N-acetyl], [Glu(7)Gln], [Glu(7)Asp], and [Asn(2)Abu] analogues and submitted to NMR structural studies as a function of pH. We first demonstrated that the Glu(7) carboxylate group is responsible for the large downfield shifts of the Asn(2) and Glu(7) amide signals. Then, molecular dynamics (MD) simulations suggested unexpected interactions between the carboxylate group and the Asn(2) and Glu(7) amide protons as well as the N-terminal alpha-amino group, through subtle conformational changes that do not alter the global fold of apamin. In addition, a structural study of the [Asn(2)Abu] analogue, revealed an essential role of Asn(2) in the beta-turn stability and the cis/trans isomerization of the Ala(5)-Pro(6) amide bond. Interestingly, this proline isomerization was shown to also depend on the ionization state of the Glu(7) carboxyl group. However, neither destabilization of the beta-turn nor proline isomerization drastically altered the helical structure that contains the residues essential for binding. Altogether, the Asn(2) and Glu(7) residues appeared essential for the N-terminal loop conformation and thus for the selective formation of the native disulfide bonds but not for the activity.

Conformational analysis of human calcitonin in solution.

The solution conformation of human calcitonin in a mixture of 60% water and 40% trifluoroethanol has been determined by the combined use of 1H NMR spectroscopy and distance geometry calculations with a distributed computing technique. 1H NMR spectroscopy provided 195 distance constraints and 13 hydrogen bond constraints. The 20 best converged structures exhibit atomic rmsd of 0.43 A for the backbone atoms from the averaged coordinate position in the region of Asn3-Phe22. The conformation is characterized by a nearly amphiphilic alpha-helix domain that extends from Leu4 in the cyclic region to His20. There are no significant differences observed among the overall structures of a series of calcitonins obtained from ultimobranchial bodies, including those that possess 20- to 50-fold greater activity. Three aromatic amino acid residues, Tyr12, Phe16 and Phe19, form a hydrophobic surface of human calcitonin. Bulky side chains on the surface could interfere with the ligand-receptor interaction thereby causing its low activity, relative to those of other species.

Effect of hydration on the stability of the collagen-like triple-helical structure of [4(R)-hydroxyprolyl-4(R)-hydroxyprolylglycine]10.

X-ray analysis has been carried out on a crystal of the collagen model peptide (Hyp(R)-Hyp(R)-Gly)10 [where Hyp(R) is 4(R)-hydroxyproline] with 1.5 A resolution. The triple-helical structure of (Hyp(R)-Hyp(R)-Gly)10 has the same helical parameters and Rich and Crick II hydrogen bond patterns as those of other collagen model peptides. However, our full-length crystal structure revealed that almost all consecutive Hyp(R) residues take the up-up pucker in contrast to putative down-up puckering propensities of other collagen model peptides. The unique feature of thermodynamic parameters associated with the conformational transition of this peptide from triple helix to single coil is that both enthalpy and entropy changes of the transition are much smaller than those of other model peptides such as (Pro-Pro-Gly)10 and (Pro-Hyp(R)-Gly)10. To corroborate the precise structural information including main- and side-chain dihedral angles and intra- and interwater bridge networks, we estimated the degrees of hydration by comparing molecular volumes observed experimentally in solution to those calculated ones from the crystal structure. The results showed that the degree of hydration of (Hyp(R)-Hyp(R)-Gly)10 is comparable to that of (Pro-Hyp(R)-Gly)10 in the triple-helical state, but the former was more highly hydrated than (Pro-Hyp(R)-Gly)10 in the single-coil state. Because hydration reduces the enthalpy due to the formation of a hydrogen bond with a water molecule and diminishes the entropy due to the restriction of water molecules surrounding a peptide molecule, we concluded that the high thermal stability of (Hyp(R)-Hyp(R)-Gly)10 is able to be described by its high hydration in the single-coil state.

Collagen-like triple helix formation of synthetic (Pro-Pro-Gly)10 analogues: (4(S)-hydroxyprolyl-4(R)-hydroxyprolyl-Gly)10, (4(R)-hydroxyprolyl-4(R)-hydroxyprolyl-Gly)10 and (4(S)-fluoroprolyl-4(R)-fluoroprolyl-Gly)10.

For the rational design of a stable collagen triple helix according to the conventional rule that the pyrrolidine puckerings of Pro, 4-hydroxyproline (Hyp) and 4-fluoroproline (fPro) should be down at the X-position and up at the Y-position in the X-Y-Gly repeated sequence for enhancing the triple helix propensities of collagen model peptides, a series of peptides were prepared in which X- and Y-positions were altogether occupied by Hyp(R), Hyp(S), fPro(R) or fPro(S). Contrary to our presumption that inducing the X-Y residues to adopt a down-up conformation would result in an increase in the thermal stability of peptides, the triple helices of (Hyp(S)-Hyp(R)-Gly)(10) and (fPro(S)-fPro(R)-Gly)(10) were less stable than those of (Pro-Hyp(R)-Gly)(10) and (Pro-fPro(R)-Gly)(10), respectively. As reported by Bächinger's and Zagari's groups, (Hyp(R)-Hyp(R)-Gly)(10) which could have an up-up conformation unfavorable for the triple helix, formed a triple helix that has a high thermal stability close to that of (Pro-Hyp(R)-Gly)(10). These results clearly show that the empirical rule based on the conformational preference of pyrrolidine ring at each of X and Y residues should not be regarded as still valid, at least for predicting the stability of collagen models in which both X and Y residues have electronegative groups at the 4-position.

Different effects of 4-hydroxyproline and 4-fluoroproline on the stability of collagen triple helix.

Differential scanning calorimetry (DSC) analyses of a series of collagen model peptides suggest that 4-hydroxyproline (Hyp) and 4-fluoroproline (fPro) have different effects on the stability of the collagen triple helices according to the sequence of amino acids and stereochemistry at the 4 positions of these imino acids. The thermodynamic parameters indicate that the enhanced stabilities are classified into two different types: the enthalpy term is primarily responsible for the enhanced stability of the triple helix of (Pro-Hyp(R)-Gly)(10), whereas the entropy term dominates the enhanced stability of (Pro-fPro(R)-Gly)(10). The difference between the molecular volumes observed in solution and intrinsic molecular volumes calculated from the crystal structure indicates the different hydration states of these peptides. (Pro-Hyp(R)-Gly)(10) is highly hydrated compared to (Pro-Pro-Gly)(10), which contributes to the larger enthalpy. In contrast, the volume of (Pro-fPro(R)-Gly)(10) shows a smaller degree of hydration than that of (Pro-Pro-Gly)(10). The entropic cost of forming the triple helix of the fPro-containing peptides is compensated by a decrease in an ordered structure of water molecules surrounding the peptide molecule, although the contribution of enthalpy originating from the hydration is reduced. These arguments about the different contribution of entropic and enthalpic terms were successfully applied to interpret the stability of the triple helix of (fPro(S)-Pro-Gly)(10) as well.

High-resolution X-ray structure of the unexpectedly stable dimer of the [Lys(-2)-Arg(-1)-des(17-21)]endothelin-1 peptide.

Previous structural studies on the [Lys((-2))-Arg((-1))]endothelin-1 peptide (KR-ET-1), 540-fold less potent than ET-1, strongly suggested the presence of an intramolecular Arg(-1)-Asp(8) (R(-1)-D(8)) salt bridge that was also observed in the shorter [Lys((-2))-Arg((-1))-des(17-21)]endothelin-1 derivative (KR-CSH-ET). In addition, for these two analogues, we have shown that the Lys-Arg dipeptide, which belongs to the prosequence, significantly improves the formation of the native disulfide bonds (>or=96% instead of approximately 70% for ET-1). In contrast to what was inferred from NMR data, molecular dynamics simulations suggested that such an intramolecular salt bridge would be unstable. The KR-CSH-ET peptide has now been crystallized at pH 5.0 and its high-resolution structure determined ab initio at 1.13 A using direct methods. Unexpectedly, KR-CSH-ET was shown to be a head-to-tail symmetric dimer, and the overall interface involves two intermolecular R(-1)-D(8) salt bridges, a two-stranded antiparallel beta-sheet, and hydrophobic contacts. Molecular dynamics simulations carried out on this dimer clearly showed that the two intermolecular salt bridges were in this case very stable. Sedimentation equilibrium experiments unambiguously confirmed that KR-ET-1 and KR-CSH-ET also exist as dimers in solution at pH 5.0. On the basis of the new dimeric structure, previous NMR data were reinterpreted. Structure calculations were performed using 484 intramolecular and 38 intermolecular NMR-derived constraints. The solution and the X-ray structures of the dimer are very similar (mean rmsd of 0.85 A). Since the KR dipeptide at the N-terminus of KR-CSH-ET is present in the prosequence, it can be hypothesized that similar intermolecular salt bridges could be involved in the in vivo formation of the native disulfide bonds of ET-1. Therefore, it appears to be likely that the prosequence does assist the ET-1 folding in a chaperone-like manner before successive cleavages that yield the bioactive ET-1 hormone.

Hydrophobic core around tyrosine for human endothelin-1 investigated by photochemically induced dynamic nuclear polarization nuclear magnetic resonance and matrix-assisted laser desorption ionization time-of-flight mass spectrometry.

Human endothelin-1 (ET-1) is a potent cardiovascular bioactive peptide. Its activity is based on the C-terminal residues, e.g., Trp 21 in particular. Recently, we reported an NMR solution structure of ET-1, which has a C-terminal hydrophobic core around Tyr 13. This C-terminal conformation does not agree with a previously reported X-ray crystal structure. To clarify the discrepancy, we performed photo-CIDNP NMR in combination with MALDI-TOF MS. The photo-CIDNP results revealed that the Tyr 13 aromatic ring is concealed in a hydrophobic interaction. MALDI-TOF MS experiments showed this is an intramolecular interaction in monomeric form, which is also supported by sedimentation analysis and two-dimensional NMR cross-peak line shapes. Thus, we confirmed the intramolecular hydrophobic core around Tyr 13 in aqueous solution, which agrees with the solution structure. The C-terminal conformational discrepancy between the solution and crystal was caused by the intermolecular hydrogen bond between Tyr 13 of one molecule and Asp 8 of the other in a dimer-like formation of crystalline ET-1. On the other hand, we indicated that endothelin-3, another isoform of the endothelin, has an apparent self-association equilibrium under the same condition in which three tyrosines participate.

Alpha,alpha-disubstituted glycines bearing a large hydrocarbon ring: peptide self-assembly through hydrophobic recognition.

A method was developed for synthesizing alpha,alpha-disubstituted glycine residues bearing a large (more than 15-membered) hydrophobic ring. The ring-closing metathesis reactions of the dialkenylated malonate precursors proceed efficiently, particularly when long methylene chains tether both terminal olefin groups. Surprisingly, the amino groups of these alpha,alpha-disubstituted glycines are inert to conventional protective reactions (e.g., N-tert-butoxycarbonyl (Boc) protection: Boc(2)O/4-dimethylaminopyridine (DMAP)/CH(2)Cl(2); N-benzyloxycarbonyl (Z) protection: Z-Cl/DMAP/CH(2)Cl(2)). Curtius rearrangement of the carboxylic acid functionality of the malonate derivative after ring-closing metathesis leads to formation of an amine functionality and can be catalyzed by diphenylphosphoryl azide. However, only the intermediate isocyanates can be isolated, even in the presence of alcohols such as benzyl alcohol. The isocyanates obtained by Curtius rearrangement in an aprotic solvent (benzene) were isolated in high yields and treated with 9-fluorenylmethanol in a high-boiling-point solvent (toluene) under reflux to give the N-9-fluorenylmethoxycarbonyl (Fmoc)-protected aminomalonate derivatives in high yield. These hydrophobic amino acids can be incorporated into a peptide by Fmoc solid-phase peptide synthesis and the acid fluoride activation method. The stability of the monomeric alpha-helical structure of a 17-amino-acid peptide was enhanced by replacement of two alanine residues with two hydrophobic amino acid residues bearing a cyclic 18-membered ring. The results of sedimentation equilibrium studies suggested that the peptide assembles into hexamers in the presence of 100 mM NaCl.

Characterization of collagen model peptides containing 4-fluoroproline; (4(S)-fluoroproline-pro-gly)10 forms a triple helix, but (4(R)-fluoroproline-pro-gly)10 does not.

Collagen model peptide (Pro-Pro-Gly)10 has a triple helical structure and undergoes a thermal transition to a single random coil structure. The transition temperature of the analogous model peptides depends largely on amino acid substitution. Substitution of Pro by 4-hydroxyproline (Hyp) or 4-fluoroproline (fPro) has especially attracted attention because the position of substitution and chirality of the hydroxyl group or fluorine atom affect the transition temperatures. Here, we demonstrated that (4(S)-fPro-Pro-Gly)10 takes a triple helical structure, but (4(R)-fPro-Pro-Gly)10 exists in a single chain structure. This is not consistent with the case of Hyp substitution in our previous report where both (4(S)-Hyp-Pro-Gly)10 and (4(R)-Hyp-Pro-Gly)10 are in a single random coil state.

Interaction among silkworm ribosomal proteins P1, P2 and P0 required for functional protein binding to the GTPase-associated domain of 28S rRNA.

Acidic ribosomal phosphoproteins P0, P1 and P2 were isolated in soluble form from silkworm ribosomes and tested for their interactions with each other and with RNA fragments corresponding to the GTPase-associated domain of residues 1030-1127 (Escherichia coli numbering) in silkworm 28S rRNA in vitro. Mixing of P1 and P2 formed the P1-P2 heterodimer, as demonstrated by gel mobility shift and chemical crosslinking. This heterodimer, but neither P1 or P2 alone, tightly bound to P0 and formed a pentameric complex, presumably as P0(P1-P2)2, assumed from its molecular weight derived from sedimentation analysis. Complex formation strongly stimulated binding of P0 to the GTPase-associated RNA domain. The protein complex and eL12 (E.coli L11-type), which cross-bound to the E.coli equivalent RNA domain, were tested for their function by replacing with the E.coli counterparts L10.L7/L12 complex and L11 on the rRNA domain within the 50S subunits. Both P1 and P2, together with P0 and eL12, were required to activate ribosomes in polyphenylalanine synthesis dependent on eucaryotic elongation factors as well as eEF-2-dependent GTPase activity. The results suggest that formation of the P1-P2 heterodimer is required for subsequent formation of the P0(P1-P2)2 complex and its functional rRNA binding in silkworm ribosomes.