Crystal structures of two mutants that have implications for the folding of bovine pancreatic ribonuclease A

Harnessing protein folding neural networks for peptide-protein docking

Highly accurate protein structure predictions by deep neural networks such as AlphaFold2 and RoseTTAFold have tremendous impact on structural biology and beyond. Here, we show that, although these deep learning approaches have originally been developed for the in silico folding of protein monomers, AlphaFold2 also enables quick and accurate modeling of peptide-protein interactions. Our simple implementation of AlphaFold2 generates peptide-protein complex models without requiring multiple sequence alignment information for the peptide partner, and can handle binding-induced conformational changes of the receptor. We explore what AlphaFold2 has memorized and learned, and describe specific examples that highlight differences compared to state-of-the-art peptide docking protocol PIPER-FlexPepDock. These results show that AlphaFold2 holds great promise for providing structural insight into a wide range of peptide-protein complexes, serving as a starting point for the detailed characterization and manipulation of these interactions.

Modeling and Structure Determination of Homo-Oligomeric Proteins: An Overview of Challenges and Current Approaches

Protein homo-oligomerization is a very common phenomenon, and approximately half of proteins form homo-oligomeric assemblies composed of identical subunits. The vast majority of such assemblies possess internal symmetry which can be either exploited to help or poses challenges during structure determination. Moreover, aspects of symmetry are critical in the modeling of protein homo-oligomers either by docking or by homology-based approaches. Here, we first provide a brief overview of the nature of protein homo-oligomerization. Next, we describe how the symmetry of homo-oligomers is addressed by crystallographic and non-crystallographic symmetry operations, and how biologically relevant intermolecular interactions can be deciphered from the ordered array of molecules within protein crystals. Additionally, we describe the most important aspects of protein homo-oligomerization in structure determination by NMR. Finally, we give an overview of approaches aimed at modeling homo-oligomers using computational methods that specifically address their internal symmetry and allow the incorporation of other experimental data as spatial restraints to achieve higher model reliability.

Consequences of proline-to-alanine substitutions for the stability and refolding of onconase

Peptidyl-prolyl isomerization reactions can make for rate-limiting steps in protein folding due to their high activation energy. Onconase, an unusually stable ribonuclease A homologue from the Northern leopard frog, contains four trans proline residues in its native state. During the refolding from its guanidine hydrochloride unfolded state, which includes the formation of a folding intermediate, the slowest of the three phases has earlier been attributed to a cis-to-trans peptidyl-prolyl isomerization reaction. We thus substituted all four proline residues individually by alanine and investigated the effect of the amino acid substitutions on the folding and stability of the onconase variants. All onconase variants proved to adopt a tertiary structure comparable with that of the wild-type protein. Although the slow phase was not eliminated for any of the variants, the P43A substitution resulted in an increase in the rate constant of the fast folding phase, i.e. a faster formation of the folding intermediate. This variant also exhibits a significant increase in thermodynamic stability. As residue 43 belongs to those residues that are protected from hydrogen exchange with the solvent in the folding intermediate, the increase in the rate constant and stability of the P43A variant emphasizes the importance of the intermediate for the folding of onconase.

Application of Drug-Perturbed Essential Dynamics/Molecular Dynamics (ED/MD) to Virtual Screening and Rational Drug Design

We present here the first application of a new algorithm, essential dynamics/molecular dynamics (ED/MD), to the field of small molecule docking. The method uses a previously existing molecular dynamics (MD) ensemble of a protein or protein-drug complex to generate, with a very small computational cost, perturbed ensembles which represent ligand-induced binding site flexibility in a more accurate way than the original trajectory. The use of these perturbed ensembles in a standard docking program leads to superior performance than the same docking procedure using the crystal structure or ensembles obtained from conventional MD simulations as templates. The simplicity and accuracy of the method opens up the possibility of introducing protein flexibility in high-throughput docking experiments.

Chain termini cross-talk in the swapping process of bovine pancreatic ribonuclease

3D domain swapping is the process by which two or more protein molecules exchange part of their structure to form intertwined dimers or higher oligomers. Bovine pancreatic ribonuclease (RNase A) is able to swap the N-terminal α-helix (residues 1-13) and/or the C-terminal β-strand (residues 116-124), thus forming a variety of oligomers, including two different dimers. Cis-trans isomerization of the Asn113-Pro114 peptide group was observed when the protein formed the C-terminal swapped dimer. To study the effect of the substitution of Pro114 on the swapping process of RNase A, we have prepared and characterized the P114A monomeric and dimeric variants of the enzyme. In contrast with previous reports, the crystal structure and NMR data on the monomer reveals a mixed cis-trans conformation for the Asn113-Ala114 peptide group, whereas the X-ray structure of the C-terminal swapped dimer of the variant is very close to that of the corresponding dimer of RNase A. The mutation at the C-terminus affects the capability of the N-terminal α-helix to swap and the stability of both dimeric forms. The present results underscore the importance of the hydration shell in determining the cross-talk between the chain termini in the swapping process of RNase A.

On the information expressed in enzyme structure: more lessons from ribonuclease A

Brownian computations were directed at Ribonuclease A (RNase A) and variants in folded states so as to quantify information of the statistical type at the atom/covalent bond level. This advanced the research reported in this journal last year on the information properties of enzyme primary structure. Brownian computation data are illustrated for a sixteen-member library. The results identify signature traits that distinguish the folded wild type (WT) molecule from variants. The distinctions are explainable in terms of correlated information and dispersion energy. The Brownian tools used for this study can be directed at other protein families (e.g., kinases, isomerases, etc.) in rapid screening, QSAR, and design applications.

X-ray crystallographic studies of RNase A variants engineered at the most destabilizing positions of the main hydrophobic core: further insight into protein stability

To investigate the structural origin of decreased pressure and temperature stability, the crystal structure of bovine pancreatic ribonuclease A variants V47A, V54A, V57A, I81A, I106A, and V108A was solved at 1.4-2.0 A resolution and compared with the structure of wild-type protein. The introduced mutations had only minor influence on the global structure of ribonuclease A. The structural changes had individual character that depends on the localization of mutated residue, however, they seemed to expand from mutation site to the rest of the structure. Several different parameters have been evaluated to find correlation with decrease of free energy of unfolding DeltaDeltaG(T), and the most significant correlation was found for main cavity volume change. Analysis of the difference distance matrices revealed that the ribonuclease A molecule is organized into five relatively rigid subdomains with individual response to mutation. This behavior consistent with results of unfolding experiments is an intrinsic feature of ribonuclease A that might be surviving remnants of folding intermediates and reflects the dynamic nature of the molecule.

On the information expressed in enzyme primary structure: lessons from Ribonuclease A

The information expressed in an enzyme's primary structure is investigated. Brownian computations are directed at Ribonuclease A (RNase A) so as to quantify the information at the atom/covalent bond level. The information content and distribution are crucial because the primary structure underpins the molecule's chemical functions. Brownian computation data are illustrated for the native protein, mutants, and sequence isomers. The results identify signature features of the active protein on new information grounds. The same tools offer rapid screening of proteins and polypeptides whereby several examples are illustrated.

Disulfide-linked protein folding pathways

Determining the mechanism by which proteins attain their native structure is an important but difficult problem in basic biology. The study of protein folding is difficult because it involves the identification and characterization of folding intermediates that are only very transiently present. Disulfide bond formation is thermodynamically linked to protein folding. The availability of thiol trapping reagents and the relatively slow kinetics of disulfide bond formation have facilitated the isolation, purification, and characterization of disulfide-linked folding intermediates. As a result, the folding pathways of several disulfide-rich proteins are among the best known of any protein. This review discusses disulfide bond formation and its relationship to protein folding in vitro and in vivo.

The structure of a two-disulfide intermediate assists in elucidating the oxidative folding pathway of a cyclic cystine knot protein

We have determined the three-dimensional structure of a two-disulfide intermediate (Cys(8)-Cys(20), Cys(14)-Cys(26)) on the oxidative folding pathway of the cyclotide MCoTI-II. Cyclotides have a range of bioactivities and, because of their exceptional stability, have been proposed as potential molecular scaffolds for drug design applications. The three-dimensional structure of the stable two-disulfide intermediate shows for the most part identical secondary and tertiary structure to the native state. The only exception is a flexible loop, which is collapsed onto the protein core in the native state, whereas in the intermediate it is more loosely associated with the remainder of the protein. The results suggest that the native fold of the peptide does not represent the free energy minimum in the absence of the Cys(1)-Cys(18) disulfide bridge and that although there is not a large energy barrier, the peptide must transiently adopt an energetically unfavorable state before the final disulfide can form.

Genetic selection for critical residues in ribonucleases

Homologous mammalian proteins were subjected to an exhaustive search for residues that are critical to their structure/function. Error-prone polymerase chain reactions were used to generate random mutations in the genes of bovine pancreatic ribonuclease (RNase A) and human angiogenin, and a genetic selection based on the intrinsic cytotoxicity of ribonucleolytic activity was used to isolate inactive variants. Twenty-three of the 124 residues in RNase A were found to be intolerant to substitution with at least one particular amino acid. Twenty-nine of the 123 residues in angiogenin were likewise intolerant. In both RNase A and angiogenin, only six residues appeared to be wholly intolerant to substitution: two histidine residues involved in general acid/base catalysis and four cysteine residues that form two disulfide bonds. With few exceptions, the remaining critical residues were buried in the hydrophobic core of the proteins. Most of these residues were found to tolerate only conservative substitutions. The importance of a particular residue as revealed by this genetic selection correlated with its sequence conservation, though several non-conserved residues were found to be critical for protein structure/function. Despite voluminous research on RNase A, the importance of many residues identified herein was unknown, and those can now serve as targets for future work. Moreover, a comparison of the critical residues in RNase A and human angiogenin, which share only 35% amino acid sequence identity, provides a unique perspective on the molecular evolution of the RNase A superfamily, as well as an impetus for applying this methodology to other ribonucleases.

Pressure-jump-induced kinetics reveals a hydration dependent folding/unfolding mechanism of ribonuclease A

Pressure-jump (p-jump)-induced relaxation kinetics was used to explore the energy landscape of protein folding/unfolding of Y115W, a fluorescent variant of ribonuclease A. Pressure-jumps of 40 MPa amplitude (5 ms dead-time) were conducted both to higher (unfolding) and to lower (folding) pressure, in the range from 100 to 500 MPa, between 30 and 50 degrees C. Significant deviations from the expected symmetrical protein relaxation kinetics were observed. Whereas downward p-jumps resulted always in single exponential kinetics, the kinetics induced by upward p-jumps were biphasic in the low pressure range and monophasic at higher pressures. The relative amplitude of the slow phase decreased as a function of both pressure and temperature. At 50 degrees C, only the fast phase remained. These results can be interpreted within the framework of a two-dimensional energy surface containing a pressure- and temperature-dependent barrier between two unfolded states differing in the isomeric state of the Asn-113-Pro-114 bond. Analysis of the activation volume of the fast kinetic phase revealed a temperature-dependent shift of the unfolding transition state to a larger volume. The observed compensation of this effect by glycerol offers an explanation for its protein stabilizing effect.

Homodimerization of the G protein SRbeta in the nucleotide-free state involves proline cis/trans isomerization in the switch II region

Protein translocation across and insertion into membranes is essential to all life forms. Signal peptide-bearing nascent polypeptide chains emerging from the ribosome are first sampled by the signal-recognition particle (SRP), then targeted to the membrane via the SRP receptor (SR), and, finally, transferred to the protein-conducting channel. In eukaryotes, this process is tightly controlled by the concerted action of three G proteins, the 54-kD subunit of SRP and the alpha- and beta-subunits of SR. We have determined the 2.2-A crystal structure of the nucleotide-free SRbeta domain. Unexpectedly, the structure is a homodimer with a highly intertwined interface made up of residues from the switch regions of the G domain. The remodeling of the switch regions does not resemble any of the known G protein switch mechanisms. Biochemical analysis confirms homodimerization in vitro, which is incompatible with SRalpha binding. The switch mechanism involves cis/trans isomerization of a strictly conserved proline, potentially implying a new layer of regulation of cotranslational transport.

Topological Determinants of Protein Domain Swapping

Protein domain swapping has been repeatedly observed in a variety of proteins and is believed to result from destabilization due to mutations or changes in environment. Based on results from our studies and others, we propose that structures of the domain-swapped proteins are mainly determined by their native topologies. We performed molecular dynamics simulations of seven different proteins, known to undergo domain swapping experimentally, under mildly denaturing conditions and found in all cases that the domain-swapped structures can be recapitulated by using protein topology in a simple protein model. Our studies further indicated that, in many cases, domain swapping occurs at positions around which the protein tends to unfold prior to complete unfolding. This, in turn, enabled prediction of protein structural elements that are responsible for domain swapping. In particular, two distinct domain-swapped dimer conformations of the focal adhesion targeting domain of focal adhesion kinase were predicted computationally and were supported experimentally by data obtained from NMR analyses.

The crystal structure of the cis-proline to glycine variant (P114G) of ribonuclease A

Replacement of a cis-proline by glycine at position 114 in ribonuclease A leads to a large decrease in thermal stability and simplifies the refolding kinetics. A crystallographic approach was used to determine whether the decrease in thermal stability results from the presence of a cis glycine peptide bond, or from a localized structural rearrangement caused by the isomerization of the mutated cis 114 peptide bond. The structure was solved at 2.0 A resolution and refined to an R-factor of 19.5% and an R(free) of 21.9%. The overall conformation of the protein was similar to that of wild-type ribonuclease A; however, there was a large localized rearrangement of the mutated loop (residues 110-117-a 9.3 A shift of the Calpha atom of residue 114). The peptide bond before Gly114 is in the trans configuration. Interestingly, a large anomalous difference density was found near residue 114, and was attributed to a bound cesium ion present in the crystallization experiment. The trans isomeric configuration of the peptide bond in the folded state of this mutant is consistent with the refolding kinetics previously reported, and the associated protein conformational change provides an explanation for the decreased thermal stability.

Dissimilarity in the Reductive Unfolding Pathways of Two Ribonuclease Homologues

Using DTT(red) as the reducing agent, the kinetics of the reductive unfolding of onconase, a frog ribonuclease, has been examined. An intermediate containing three disulfides, Ir, that is formed rapidly in the reductive pathway, is more resistant to further reduction than the parent molecule, indicating that the remaining disulfides in onconase are less accessible to DTT(red). Disulfide-bond mapping of Ir indicated that it is a single species lacking the (30-75) disulfide bond. The reductive unfolding pattern of onconase is consistent with an analysis of the exposed surface area of the cysteine sulfur atoms in the (30-75) disulfide bond, which reveals that these atoms are about four- and sevenfold, respectively, more exposed than those in the next two maximally exposed disulfides. By contrast, in the reductive unfolding of the homologue, RNase A, there are two intermediates, arising from the reduction of the (40-95) and (65-72) disulfide bonds, which takes place in parallel, and on a much longer time-scale, compared to the initial reduction of onconase; this behavior is consistent with the almost equally exposed surface areas of the cysteine sulfur atoms that form the (40-95) and (65-72) disulfide bonds in RNase A and the fourfold more exposed cysteine sulfur atoms of the (30-75) disulfide bond in onconase. Analysis and in silico mutation of the residues around the (40-95) disulfide bond in RNase A, which is analogous to the (30-75) disulfide bond of onconase, reveal that the side-chain of tyrosine 92 of RNase A, a highly conserved residue among mammalian pancreatic ribonucleases, lies atop the (40-95) disulfide bond, resulting in a shielding of the corresponding sulfur atoms from the solvent; such burial of the (30-75) sulfur atoms is absent from onconase, due to the replacement of Tyr92 by Arg73, which is situated away from the (30-75) disulfide bond and into the solvent, resulting in the large exposed surface-area of the cysteine sulfur atoms forming this bond. Removal of Tyr92 from RNase A resulted in the relatively rapid reduction of the mutant to form a single intermediate (des [40-95] Y92A), i.e. it resulted in an onconase-like reductive unfolding behavior. The reduction of the P93A mutant of RNase A proceeds through a single intermediate, the des [40-95] P93A species, as in onconase. Although mutation of Pro93 to Ala does not increase the exposed surface area of the (40-95) cysteine sulfur atoms, structural analysis of the mutant reveals that there is greater flexibility in the (40-95) disulfide bond compared to the (65-72) disulfide bond that may make the (40-95) disulfide bond much easier to expose, consistent with the reductive unfolding pathway and kinetics of P93A. Mutation of Tyr92 to Phe92 in RNase A has no effect on its reductive unfolding pathway, suggesting that the hydrogen bond between the hydroxyl group of Tyr92 and the carbonyl group of Lys37 has no impact on the local unfolding free energy required to expose the (40-95) disulfide bond. Thus, these data shed light on the differences between the reductive unfolding pathways of the two homologous proteins and provide a structural basis for the origin of this difference.

Oxidative Folding of Amaranthus α-Amylase Inhibitor

Oxidative folding is the fusion of native disulfide bond formation with conformational folding. This complex process is guided by two types of interactions: first, covalent interactions between cysteine residues, which transform into native disulfide bridges, and second, non-covalent interactions giving rise to secondary and tertiary protein structure. The aim of this work is to understand both types of interactions in the oxidative folding of Amaranthus alpha-amylase inhibitor (AAI) by providing information both at the level of individual disulfide species and at the level of amino acid residue conformation. The cystine-knot disulfides of AAI protein are stabilized in an interdependent manner, and the oxidative folding is characterized by a high heterogeneity of one-, two-, and three-disulfide intermediates. The formation of the most abundant species, the main folding intermediate, is favored over other species even in the absence of non-covalent sequential preferences. Time-resolved NMR and photochemically induced dynamic nuclear polarization spectroscopies were used to follow the oxidative folding at the level of amino acid residue conformation. Because this is the first time that a complete oxidative folding process has been monitored with these two techniques, their results were compared with those obtained at the level of an individual disulfide species. The techniques proved to be valuable for the study of conformational developments and aromatic accessibility changes along oxidative folding pathways. A detailed picture of the oxidative folding of AAI provides a model study that combines different biochemical and biophysical techniques for a fuller understanding of a complex process.

Structural determinants of oxidative folding in proteins

A method for determining the kinetic fate of structured disulfide species (i.e., whether they are preferentially oxidized or reshuffle back to an unstructured disulfide species) is introduced. The method relies on the sensitivity of unstructured disulfide species to low concentrations of reducing agents. Because a structured des species that preferentially reshuffles generally first rearranges to an unstructured species, a small concentration of reduced DTT (e.g., 260 microM) suffices to distinguish on-pathway intermediates from dead-end species. We apply this method to the oxidative folding of bovine pancreatic ribonuclease A (RNase A) and show that des[40-95] and des[65-72] are productive intermediates, whereas des[26-84] and des[58-110] are metastable dead-end species that preferentially reshuffle. The key factor in determining the kinetic fate of these des species is the relative accessibility of both their thiol groups and disulfide bonds. Productive intermediates tend to be disulfide-secure, meaning that their structural fluctuations preferentially expose their thiol groups, while keeping their disulfide bonds buried. By contrast, dead-end species tend to be disulfide-insecure, in that their structural fluctuations expose their disulfide bonds in concert with their thiol groups. This distinction leads to four generic types of oxidative folding pathways. We combine these results with those of earlier studies to suggest a general three-stage model of oxidative folding of RNase A and other single-domain proteins with multiple disulfide bonds.

Oxidative folding of proteins

The oxidative folding of proteins is reviewed and illustrated with bovine pancreatic ribonuclease A (RNase A). The mutual effects of conformational folding and disulfide bond regeneration are emphasized, particularly the "locking in" of native disulfide bonds by stable tertiary structure in disulfide intermediates. Two types of structured metastable disulfide species are discerned, depending on the relative protection of their disulfide bonds and thiol groups. Four generic pathways for oxidative folding are identified and characterized.

The residual pro-part of cathepsin C fulfills the criteria required for an intramolecular chaperone in folding and stabilizing the human proenzyme

The 13.5 kDa N-terminal part of the propeptide remains associated with mature cathepsin C after proteolytic activation and excision of the activation peptide. This residual pro-part, isolated from the recombinant enzyme, folds spontaneously and rapidly to a stable, compact monomer with secondary structure and stable tertiary interactions. Folding and unfolding kinetics of the residual pro-part with intact disulfides are complex, and accumulation of transient intermediates is observed. The cleaved form of the pro-part isolated from natural human cathepsin C also folds, suggesting that the intact form comprises two folding domains. The linkages of the two disulfide bridges have been established as 30-118 and 54-136 for the native enzyme. The native disulfide bonds can be re-formed from the fully reduced and denatured state by oxidative refolding, resulting in a domain that is spectroscopically indistinguishable from the original refolded residual pro-part. Both disulfides are solvent-exposed and can be reduced in the absence of denaturant. The reduced form retains most or all of the native tertiary structure and is only approximately 2 kcal.mol(-1) less stable than the oxidized form. It folds fast relative to the rate of biosynthesis, to the same conformation as the oxidized form. Folding and disulfide formation are sequential. These results indicate that the proenzyme folds sequentially in vivo and that the residual pro-part constitutes a rapidly and independently folding domain that stabilizes the mature enzyme. It thus fulfills the criteria required of an intramolecular chaperone. It may also be involved in stabilizing the tetrameric structure of the mature enzyme.

Disulfide bonds and protein folding

The applications of disulfide-bond chemistry to studies of protein folding, structure, and stability are reviewed and illustrated with bovine pancreatic ribonuclease A (RNase A). After surveying the general properties and advantages of disulfide-bond studies, we illustrate the mechanism of reductive unfolding with RNase A, and discuss its application to probing structural fluctuations in folded proteins. The oxidative folding of RNase A is then described, focusing on the role of structure formation in the regeneration of the native disulfide bonds. The development of structure and conformational order in the disulfide intermediates during oxidative folding is characterized. Partially folded disulfide species are not observed, indicating that disulfide-coupled folding is highly cooperative. Contrary to the predictions of "rugged funnel" models of protein folding, misfolded disulfide species are also not observed despite the potentially stabilizing effect of many nonnative disulfide bonds. The mechanism of regenerating the native disulfide bonds suggests an analogous scenario for conformational folding. Finally, engineered covalent cross-links may be used to assay for the association of protein segments in the folding transition state, as illustrated with RNase A.

Solution NMR evidence for a cis Tyr-Ala peptide group in the structure of [Pro93Ala] bovine pancreatic ribonuclease A

Proline peptide group isomerization can result in kinetic barriers in protein folding. In particular, the cis proline peptide conformation at Tyr92-Pro93 of bovine pancreatic ribonuclease A (RNase A) has been proposed to be crucial for chain folding initiation. Mutation of this proline-93 to alanine results in an RNase A molecule, P93A, that exhibits unfolding/refolding kinetics consistent with a cis Tyr92-Ala93 peptide group conformation in the folded structure (Dodge RW, Scheraga HA, 1996, Biochemistry 35:1548-1559). Here, we describe the analysis of backbone proton resonance assignments for P93A together with nuclear Overhauser effect data that provide spectroscopic evidence for a type VI beta-bend conformation with a cis Tyr92-Ala93 peptide group in the folded structure. This is in contrast to the reported X-ray crystal structure of [Pro93Gly]-RNase A (Schultz LW, Hargraves SR, Klink TA, Raines RT, 1998, Protein Sci 7:1620-1625), in which Tyr92-Gly93 forms a type-II beta-bend with a trans peptide group conformation. While a glycine residue at position 93 accommodates a type-II bend (with a positive value of phi93), RNase A molecules with either proline or alanine residues at this position appear to require a cis peptide group with a type-VI beta-bend for proper folding. These results support the view that a cis Pro93 conformation is crucial for proper folding of wild-type RNase A.

Cis peptide bonds in proteins: residues involved, their conformations, interactions and locations

An analysis of a non-redundant set of protein structures from the Brookhaven Protein Data Bank has been carried out to find out the residue preference, local conformation, hydrogen bonding and other stabilizing interactions involving cis peptide bonds. This has led to a reclassification of turns mediated by cis peptides, and their average geometrical parameters have been evaluated. The interdependence of the side and main-chain torsion angles of proline rings provided an explanation why such rings in cis peptides are found to have the DOWN puckering. A comparison of cis peptides containing proline and non-proline residues show differences in conformation, location in the secondary structure and in relation to the centre of the molecule, and relative accessibilities of residues. Relevance of the results in mutation studies and the cis-trans isomerization during protein folding is discussed.

Non-proline cis peptide bonds in proteins

In a non-redundant set of 571 proteins from the Brookhaven Protein Data Base, a total of 43 non-proline cis peptide bonds were identified. Average geometrical parameters of the well-defined cis peptide bonds in proteins determined at high resolution show that some parameters, most notably the bond angle at the amide bond nitrogen, deviate significantly from the corresponding one in the trans conformation. Since the same feature was observed in cis amide bonds in small molecule structures found in the Cambridge Structural Data Base, a new set of parameters for the refinement of protein structures containing non-Pro cis peptide bonds is proposed.A striking preference was observed for main-chain dihedral angles of the residues involved in cis peptide bonds. All residues N-terminal and most residues C-terminal to a non-Pro cis peptide bond (except Gly) are located in the beta-region of a phi/psi plot. Also, all of the few C-terminal residues (except Gly) located in the alpha-region of the phi/psi plot constitute the start of an alpha-helix in the respective structure. In the majority of cases, an intimate side-chain/side-chain interaction was observed between the flanking residues, often involving aromatic side-chains. Interestingly, most of the cases found occur in functionally important regions such as close to the active site of proteins. It is intriguing that many of the proteins containing non-proline cis peptide bonds are carbohydrate-binding or processing proteins. The occurrence of these unusual peptide bonds is significantly more frequent in structures determined at high resolution than in structures determined at medium and low resolution, suggesting that these bonds may be more abundant than previously thought. On the basis of our experience with the structure determination of coagulation factor XIII, we developed an algorithm for the identification of possibly overlooked cis peptide bonds that exploits the deviations of geometrical parameters from ideality. A few likely candidates based on our algorithm have been identified and are discussed.