Non-enzymatic covalent modifications of proteins: mechanisms, physiological consequences and clinical applications

Abstract

Given the complexity of the biosynthetic machinery and the delicate chemical composition of proteins, it is remarkable that cells manage to produce and maintain normally functioning proteins under most conditions. However, it is now well known that proteins are susceptible to various non-enzymatic covalent modifications (NECM) under physiological conditions. Such modifications can be of no or little importance to the protein or they can be absolutely detrimental. Often NECM are difficult to study due to the complex and technically demanding methods required to identify many of these modifications. Thus, the role of NECM has not yet been adequately resolved but recent research has allowed a better understanding of such modifications. The present review outlines the various forms of NECM that involve covalent modifications of proteins, and discusses their relevance, biological impact and potential applications in the study of protein turnover and diagnosis of disease.

Introduction

Proteins are subject to a variety of spontaneous non-enzymatic modifications that may affect their structure, function and stability. The occurrence of these alterations and their localization within a protein seem to be determined primarily by protein sequence and structure, the protein microenvironment and half-life of the protein. Non-enzymatic modifications are time-dependent and the effects of individual changes may be cumulative. The term ‘protein fatigue’ has been used to describe this phenomenon (Galletti et al., 1995) as an analogy to the so-called ‘metal fatigue’. Non-enzymatic modifications may be subdivided into two general forms (Visick and Clarke, 1995):

• Conformational changes (or conformational damage), i.e. a modification of the three-dimensional structure of the protein without compromising its chemical composition (Fig. 1a), and

• Covalent modifications (or covalent damage) where the primary structure of the protein or peptide in question is altered on the amino acid/peptide bond level, by cleavage and formation of covalent bonds (Fig. 1b).

A clear distinction from one type of modifications to another is not always obvious as covalent changes may cause conformational modifications and vice versa.

The present review focuses on the mechanisms resulting in non-enzymatic covalent protein modifications (NECM), and the impact of such reactions on protein function. An overview of the various forms of NECM is given in Fig. 1.

Amino acids and peptide bonds are prone to a variety of chemical modifications. Changes in the protein microenvironment such as changes in temperature, pH, or redox potential, etc. may induce or accelerate changes in the primary structure of a protein. Most types of covalent modifications are often restricted to specific susceptible residues or areas of proteins. Table 1 summarizes examples of non-enzymatic covalent modifications.

Due to the large number and heterogeneous nature of potential covalent modifications that can occur in proteins, only a broad description of the different modifications will be given with an emphasis on the physiological roles and consequences established from in vitro and in vivo experiments.

Oxidative damage of proteins is caused by the action of free radicals and various other oxidizing compounds on proteins (Scheme 1). These compounds include Nitric oxide (NO), peroxynitrate, H₂O₂ or hydroxyl, hydroperoxyl superoxide and lipid peroxyl radicals. A common term for these compounds is reactive oxygen species (ROS). ROS are formed by a number of different pathways. Some of these pathways are mediated by specific enzyme systems such as Nitric Oxide Synthetase, Cyclo-Oxygenase and Mono-Amine Oxidase B. Protein modifications induced by these enzyme systems are implicated in several pathological processes as well as in inflammation, tissue healing and other physiological processes. Other mechanisms of inducing oxidative damage are related to environmental effects such as ionizing radiation, reduction of metal ions such as Fe(II) or Cu(I), or chemical compounds. A clear distinction between enzyme catalyzed and non-enzyme catalyzed protein oxidation is difficult to make.

Highly reactive ROS radicals are produced by ionizing radiation or as side-products of metabolism (Farr and Komoga, 1991) and may in principle attack any amino acid (Stadtman, 1992). However, in most proteins certain ‘hot spots’ in the protein seem to be particular susceptible to oxidative damage. Compounds formed by oxidation of carbohydrates may react with proteins forming advanced glycation end products (AGEs). Furthermore, ROS-induced changes may induce the formation of covalent cross-links within and between protein molecules. Glycation and cross-linking reactions are reviewed separately in the following sections.

Oxidation of amino acid side-chains can give rise to a number of adducts. Many reactions results in the formation of protein carbonyl derivatives such as direct oxidation of arginine or proline to glutamic semialdehyde or lysine to α-amino-adipic-semialdehyde (Stadtman and Berlet, 1998). Other prevalent modifications comprise: oxidation of Met producing Met sulfoxide (Farr and Komoga, 1991); nitrosylation of tyrosine and tryptophan residues; formation of oxo, hydroxy, nitro and chloro derivatives of amino acids (Davies, 1987); transformation of His to Asx; and of Pro to Glu-adducts (Stadtman, 1992, Davies, 1987) (see Scheme 1). Furthermore, peptide linkages of the protein backbone may react with ROS resulting in cleavage of the peptide linkage and formation of N-α-ketoacyl peptides or N-pyrovyl peptides (Davies, 1987, Stadtman and Berlet, 1998).

Under normal conditions the oxidative potential of the protein micro-environment is under tight control of a number of balancing systems including antioxidants and free radical scavengers, peroxidases, catalase and superoxide dismutase. These systems can be viewed as protection mechanisms, more than repair systems. Actual repair mechanisms specific for oxidative damage are rare. They include the enzyme methionine sulfoxide reductase (MSR), which recognizes methionine sulfoxide in peptides and can convert it back to methionine (Stadtman et al., 1988) and heme oxygenase 1, which can repair some of the oxidative damages caused by elevated NO levels.

Oxidation may have a deleterious effect on protein function and stability. Many enzymes have been shown to loose their biological activity as a consequence of oxidation (Rattan, 1996). It has been shown that enzymes isolated from older animals show decreased temperature stability compared to enzymes from young animals where less oxidative damage has occurred. Protein oxidation may be implicated in the pathogenesis of several diseases (Smith et al., 1991). Oxidative damage of proteins often leads to the formation of protein carbonyl compounds as described above. Sensitive assays are available for quantification of such adducts which have been used as an indicator of the oxidative stress exposure of a given tissue or organism. Increased levels of protein carbonyls have been associated with rheumatoid arthritis, Alzheimer's disease, Parkinson's disease, muscular dystrophy and several other serious chronic and debilitating conditions. A direct link has been proposed between oxidative stress and free-radical mediated protein damage and the formation of senile plaques and neurofibrillar tangles in Alzheimer's disease. The exact role of the oxidized protein adducts in these diseases still needs further elucidation, but the accumulation of such compounds in tissues and cells may result in excessive and damaging aggregation, fragmentation and denaturation of proteins. Oxidation has been reported to alter the susceptibility of proteins towards proteolysis (Farr and Komoga, 1991, Davies, 1987). In some cases proteolytic degradation is increased by the presence of oxidized protein adducts, but in may cases the oxidation of proteins renders them more resistance to proteolysis resulting in the accumulation of oxidized proteins and enzyme forms with increasing age of the tissues and cells.

Incubation of proteins with sugar leads to a complex series of condensation, rearrangement and fragmentation reactions forming a variety of degradation products, collectively termed advanced glycation end products (AGEs) (Scheme 2). The term glycation is used to discriminate between these spontaneous reactions and the enzyme-catalyzed glycosylation that occur as a highly regulated post-translational processing of many proteins. The reaction schemes for formation of AGEs can be very complex, and are collectively termed Maillard reactions. AGEs are formed when an aldose group of a saccharide condenses with a reactive amino group, typically in lysine side chains of target proteins (Baynes, 1991). These compounds subsequently undergo further reactions, which are referred to as Amadori rearrangements. Other reaction schemes also occur, such as oxidation of Schiff base-linked carbohydrate moieties. Many AGEs are unstable and gradually undergo various dehydration and re-arrangements to produce reactive carbonyl containing compounds as well as covalent intra- and inter-molecular cross-links in the affected proteins. Thus, as a group, AGE reactions result in the formation of very complex and heterogeneous reaction products.

Physiological concentrations of monosaccharides such as ribose, fructose and glucose have been shown to induce AGE formation both in vitro and in vivo. These reactions are greatly increased by oxidative stress conditions, i.e. conditions which induce/increase the presence of ROS (see previous section). To stress the link between glycation and oxidation resulting in AGE formation, the processes has been referred to as glycoxidation.

No enzyme system has been described which can revert the AGE modifications of proteins. However, several ‘AGE receptors’ have been identified, which potentially could be involved in a repair pathway but the role of these receptors has not been fully elucidated (Sano et al., 1999). AGE-binding receptors are: scavenger receptor types I and II, the receptor for advanced glycation end-products (AGEr); oligosaccharyl transferase-48 (OST-48, AGE-R1); 80K-H phosphoprotein (AGE-R2); and galectin-3 (AGE-R3) (Thornalley, 1998). AGE receptors are found in monocytes, macrophages, endothelial cells, pericytes, podocytes, astrocytes and microglia. AGE-modified proteins also bind to lysozyme and lactoferrin. Interaction of AGE modified proteins with these receptors may serve diverse purposes including: endocytosis and disposal of the modified protein; initiation of tissue repair; and regulation of tissue turnover. The main pathways for dealing with AGE modification involve proteolytic degradation of the modified proteins. In this context it is noteworthy that AGE modification of proteins can serve to increase their resistance to proteolysis. Thus, proteins containing various AGE adducts can accumulate with age.

Proteins containing AGEs prepared in vitro have been demonstrated to have several potential pathogenic properties (Vlassara et al., 1994, Daniels and Hauser, 1992, Yan et al., 1994). AGEs have been reported to accumulate with age and are implicated in the aging process. This has been shown for instance in human articular cartilage (Bank et al., 1998) or in cells of the eye lens (Nagaraj et al., 1991). AGEs elicit a wide range of cell-mediated responses leading to vascular dysfunction, matrix expansion and athero- and glomerulosclerosis, as well as other pathological processes. A number of studies have implicated the accelerated AGE formation, which occurs in diabetic patients with secondary complications of the disease such as diabetic microangiopathy, rethinopathy and nephropathy (Sano et al., 1999, Nakamura et al., 1993). AGE modifications of proteins have also been associated with several neurological disorders such as Alzheimer's and Parkinson's disease. The accumulation of AGEs in cartilage has been suggested to contribute to the decrease in the resistance of cartilage to mechanical influence observed in elderly persons (Bank et al., 1998). The presence of AGE modified erythrocyte membrane proteins has been reported to result in impairment of red cell function. It is common to measure AGE modified hemoglobin and serum albumin in diabetic patients as an indicator of the glycation-induced secondary complications associated with diabetes (Iberg and Flückinger, 1986, Johnson and Baker, 1986).

AGE modified proteins can become more resistant to proteolytic degradation, and they have a propensity to aggregate, which can result in accumulation of large insoluble protein aggregates, such as the aggregation of neurofilament proteins found in Parkinson's and Alzheimer's disease. AGE modified proteins can also activate specific signaling mechanisms. Such activation has been reported to induce production of superoxide and nitric oxide by glial cells and may be considered part of a vicious cycle, which finally leads to neuronal cell death in the substantia nigra in Parkinson's disease.

Immunohistochemical analyses of human atherosclerotic lesions using a monoclonal anti-AGE antibody have demonstrated diffuse extra-cellular AGE-deposition as well as dense intracellular AGE-deposition in macrophage- and vascular smooth muscle cells. Here the deposition of AGE modified proteins may serve not only an indirect role in the formation of arthrosclerotic plaques, but also a direct role by interaction with AGE receptors (Chappey et al., 1997, Vlassara, 1996). Cellular responses mediated by binding to AGE receptors can result in altered gene expression and cytokine regulation affecting vascular function (Vlassara, 1996). AGE modified proteins have also been suggested as playing a role in regulation of the bone turnover process. An immuno-histochemical study using anti-AGE antibody revealed positive immuno-staining for AGEs in bone tissues from elderly subjects (Miyata et al., 1996). AGE-modified proteins were shown to stimulate monocytes and macrophages to secrete cytokines known to affect bone resorption, such as IL-1β, IL-6 and TNF-α. Thus, it has been suggested that AGEs enhance osteoclasts-mediated bone resorption. The modification of bone matrices with AGEs might, therefore, play a pathophysiological role not only in the remodeling of senescent bone matrix tissues, but also in dialysis-related amyloidosis or osteoporosis associated with diabetes and aging.

Covalent cross-links may form in proteins through a range of non-enzymatic pathways, and have been shown to accompany several types of NECM (Friedman, 1999, Davies, 1987). Generally the cross-linking reactions are closely linked to the protein oxidation and glycation reactions described in the two previous sections. A common side-product of glycation is pentosidine, which is formed by reaction of 3-deoxyglucosone or methylglyoxal with arginine side-chains. Also imidazolium cross-links such as glyoxal-lysine- and methylglyoxal-lysine-dimers formed by reaction of lysine side chain with various aldose adducts and bityrosine are common products of oxidation and glycation (Scheme 3).

The cross-links lysinoalanine (LAL), histidinoalanine (HAL), lanthionine, methyl-lysinoalanine, methyl-histidinoalanine and methyl-lanthionine, may form spontaneously within certain peptides and proteins (Scheme 4). A number of other covalent cross-links are likely to exist, due to the complex and heterogeneous nature of the chemical reactions involved in cross-link formation.

The sites of cross-link formation within proteins are limited, the number and distribution of such sites within a protein are determined by the amino acid composition, accessibility, conformation and chain mobility of affected proteins.

No repair systems are known.

Cross-links may form both inter- and intra-molecularly. Biological and physical properties of cross-link-containing proteins are probably strongly affected by the relative numbers of these cross-links. In most cases cross-linking is assumed to have damaging effects on protein function, but it may also affect protein half-life. The presence of cross-links has been shown to significantly decrease the susceptibility of peptides to proteolytic digestion. Thus, cross-links such as pentosidine, LAL and HAL appear to survive renal and hepatic proteolysis and are found in urine as free or peptide-associated cross-links (Fujimoto, 1986).

Glycation mediated intermolecular cross-links in the extracellular matrix have been shown to decrease the flexibility and permeability of tissues and reduce turnover. Cardiovascular tissue also contains a significant proportion of the fibrous connective tissue protein elastin, and its properties are similarly modified by glycation and oxidation-mediated cross-link formation. The nature of these glycation cross-links is now being unraveled and this knowledge is crucial in any attempt to inhibit formation of such reactions. LAL cross-links have been detected in bone, dentin, articular cartilage and lens protein (Shikata et al., 1985, Fujimoto and Roughley, 1984, Kanajama et al., 1986). In these tissues the content of LAL was very low in young subjects but increased with age. In cartilage, cross-link formation between components of the cartilage matrix has been shown to gradually reduce the flexibility and mechanical properties of the tissue (Monnier et al., 1996). This may suggest that NECM such as LAL formation may be involved in the cartilage deterioration seen in osteoarthritis and other diseases where cartilage is degraded.

Cross-link formation in proteins has also been implicated in the pathological processes of various neurological disorders. In Parkinson's disease, so-called Lewy bodies are composed of densely cross-linked intracellular protein deposits formed from cytoskeleton components. These complexes accumulate in presymptomatic stages of the disease, and it is supposed that they play an important role in the pathogenesis of the neurological destructions. Recent findings indicate that glycation-generated protein adducts are the major structural cross-linkers that cause the transformation of soluble neurofilament proteins to insoluble Lewy bodies (Goedert, 2001).

Amino acids with the exception of Gly can occur in two stereo-isomeric forms, designated d- and l-enantiomers. The stereo-specificity of the ribosomal biosynthetic system ensures that only l-amino acids are incorporated into mammalian proteins. However, racemization may occur spontaneously at a low rate causing an accumulation of d-amino acid enantiomers during aging of proteins. A large number of reports have demonstrated an age-dependent increase of d-amino acids in human tissues with a low metabolic turnover, such as dentin, bone, dermis, brain, cartilage, eye-lens, etc.

The relative racemization rate of the 19 amino acids varies between different proteins but has generally been found to be Asx>Glx>Ser>Ala>other amino acids. It was originally assumed that the racemization process of free and protein bound amino acids proceeded by direct proton abstraction creating a planar α-carbanion, followed by a rapid, random re-protonation leading to one or the other enantiomer, as illustrated by Scheme 5.

According to this hypothesis, the ease with which racemization occurs at various amino acids would be dependent on the electrophillic properties of their side-chains. Indeed, this relationship has been confirmed experimentally for free amino acids. However, the de-protonation/re-protonation model does not adequately describe racemization mechanism of peptide-bound amino acids. As an example, it does not explain the experimental observation that Asp residues racemize faster than Thr and Ser residues, whose side chains are more electrophillic than the aspartyl side chain.

The rapid racemization of Asx and Glx within proteins is believed to reflect the propensity of these amino acids for cyclic imide formation (Geiger and Clarke, 1987, Radkiewicz et al., 2001) (Scheme 6).

The involvement of an imide pathway in the racemization of Asx residues has been verified in several studies. The rate of succinimide racemization in the hexapeptide Val-Tyr-Pro-Asu-Gly-Ala (where Asu denotes a succinimide residue) was 117 000 times the rate found in the corresponding Asp-peptide (Geiger and Clarke, 1987). The enhanced susceptibility of the succinimide to racemize is due to the increased acidity of the α-carbon of this structure (Radkiewicz et al., 2001).

Imide formation also appears to be the central event in two other common spontaneous covalent reactions: deamidation and isomerization.

Deamidation refers to the hydrolysis of the side chain amide bond in an Asn or Gln residue to form Asp and Glu, respectively.

In isomerization, the peptide backbone is re-directed from the α-carboxyl group of an Asx or Glx residue through the side chain β or γ-carboxyl, creating a kink in the peptide backbone. For Asx, and presumably also Glx, all these reactions are presumed to take place through a imide intermediate formed by the nucleophilic attack of the of the side-chain carbonyl on the peptide bond nitrogen, (Scheme 6). The imide is highly susceptible to racemization leading to the formation of both racemised and isomerized residues: d-Asp; and d-isoAsp (d-Glu and d-isoGlu).

The reaction rates of deamidation, racemization and isomerization are strongly correlated to the ease with which imides are generated. Peptide dihedral angles of ψ=−120° and χ=+120°, are optimal for succinimide formation. The reaction is dependent on polypeptide conformation and the nature of the residues immediately preceding or following the Asx residue. Imide formation is facilitated when polypeptides are flexible (or possess the necessary conformation) and when neighboring residues have small (non-bulky) side-chains (Geiger and Clarke, 1987, Radkiewicz et al., 2001).

The process of in vivo-racemization is not restricted to Asx and Glx residues. Hence, d-enantiomers of Ser, Tyr, Ala and His have been shown to accumulate in various tissues during human aging (Dunlop and Neidle, 1997, Luthra et al., 1994, Cloos and Jensen, 2000). The accumulation of d-Ala observed in some proteins is probably due to the formation of racemic Ala via the intermediates dehydroalanine and pyruvic acid through degradation/re-arrangement of Ser(P), Ser and Cys residues (Cloos and Jensen, 2000).

There is no known cellular system for the identification and repair of an Asp or Glu residue formed by deamidation of Asn and Gln.

However, as mentioned previously, deamidation involves imide formation which also results in formation of d-enantiomeric or isomerized adducts (Scheme 5). These ‘unnatural’ adducts may serve to reveal that a protein modification has taken place and make enzymatic recognition and degradation possible.

The enzyme l-isoaspartyl/O-methyltransferase (IAMT, EC 2.1.1.77), catalyses the transfer of a methyl group from S-adenosyl-methionine to the α-carboxyl group of l-isoAsp residues in a variety of organisms (Lowenson and Clarke, 1995).

Methylation of a l-isoAsp peptide induces the reformation of a succinimide ring and has been shown to result in the conversion to the l-Asp form of the peptide (Brennan et al., 1994, McFadden and Clarke, 1987). Although the IAMT genes are well conserved from bacteria to humans, the mammalian methyl transferases have the additional ability to recognize d-Asp residues (but not d-isoAsp residues), permitting them to repair some racemized residues (Lowenson and Clarke, 1992). Presumably the racemization reaction, which proceeds slower than isomerization, may be less detrimental for bacteria, where the protein life span is shorter than in a mammalian cell. No repair systems are known for the handling of isomerized or racemised Glu residues.

The enzymes d-amino acid oxidase (d-AAO, EC 1.4.3.3) and d-aspartate oxidase (d-AspO, EC 1.4.3.1) catalyse the oxidative deamidation of free d-amino acids to their corresponding α-keto acids, which in turn can be specifically re-amidated to the l-form (D'Aniello et al., 1993). These enzymes are assumed to have a central role in the metabolism and elimination of d-amino acids accumulated during aging (D'Aniello et al., 1993).

Deamidation changes the charge of the affected protein. Racemization and isomerization alter the bonding angles of adjoining peptide bonds. A few racemized or isomerized amino acids may change the dipole moments of a polypeptide chain, thus potentially altering the structure of the whole protein (Hol et al., 1981). Such modifications may change protein structure, stability, properties and functions.

Mutations of the gene encoding IAMT in Eschericia coli results in mutants which are unusually susceptible to heat shock and survive poorly in a stationary phase, a stage with little or no protein synthesis (Li and Clarke, 1992). Moreover, IAMT deletion mutants have been shown to possess distinct phenotypes (Kim et al., 1997, Kim et al., 1997b). IAMT knockout mice exhibit growth retardation, and die of fatal seizures at an average age of 42 days, suggesting that the ability to repair l-isoAsp and d-Asp residues is essential for normal growth and central nervous system function (Kim et al., 1997, Kim et al., 1997b). Rats and chickens fed with d-Ala and d-Asp in amounts exceeding the capacity of the d-AAO and d-AspO enzymes exhibit growth retardation and suppression of protein synthesis (D'Aniello et al., 1993). These observations indicate that systemic accumulation of deamidated, isomerized, or racemized residues is detrimental.

Disulfide bonds of Cys residues and hydroxyl groups of Ser and Thr residues and their phosphorylated adducts Ser(P) and Thr(P) may undergo spontaneous hydroxide ion-induced β-elimination reactions to form dehydroalanine [or methyl-dehydroalanine in the case of Thr and Thr(P)] residues. Phosphorylated Thr and Ser residues may also dephosphorylate by spontaneous hydrolysis to yield a variety of adducts, as outlined in Scheme 7 (Cloos and Jensen, 2000, Shikata et al., 1985, Fujimoto and Roughley, 1984, Kanajama et al., 1986). Among the main products of these elimination-addition reactions are bifunctional cross-links, as lysinoalanine (LAL) and histidinoalanine (HAL) (Cloos and Jensen, 2000, Shikata et al., 1985), described in a previous section.

Evidence for such decomposition pathways have been reported in various mammalian tissues, including bone and dentin. Ser(P) has been shown to decrease in an age-dependent manner in human dentin, with a concomitant increase in Ala, LAL and HAL (Cloos and Jensen, 2000).

Protein kinases (PKs) may act as repair systems for the spontaneous dephosphorylation of Ser(P), Thr(P), Tyr(P) through hydrolysis. However, at present no repair or recognition system is known for adducts formed by the elimination reactions (dehydroalanine, pyruvic acid, cross-links, etc.).

Phosphorylation is one of the most important biological signal mechanisms controlling the activity of various enzymes, receptors, signaling molecules and other target proteins. Obviously a change in the phosphorylation status may directly affect the function of a protein. In addition, β-elimination and hydrolysis of phosphorylated residues alters the charge of affected proteins, influencing protein conformation and function indirectly. Some of the secondary decomposition pathways may cause hydrolysis of the peptide backbone, and formation of bi-functional cross-links, as LAL and HAL (Scheme 7). The effect of the presence of ‘unnatural’ residues, dehydroalanine, pyruvic acid and others in proteins has not received great attention. Hence, no examples are described where such modifications are directly involved in pathological processes, but in accordance with the observations made with other types of NECM, they are likely to affect function, structure and stability of the modified proteins.

Disulfide bonds linking the sulfohydryl groups of two cysteine residues may be reduced (under reducing conditions or by the action of reducing agents) or formed inappropriately under oxidizing conditions (or by oxidizing agents).

Protein disulfide isomerases, which catalyse disulfide bond formation, are important members of the protein-folding pathway (Bardwell and Beckwith, 1993). These proteins have the potential of acting as a repair system by re-oxidizing reduced disulfide bonds in damaged proteins. In analogy, some proteins (i.e. the cytoplasmic protein thioredoxin reductase) may reduce disulfides to counteract the effects of oxidative stress (Stein, 1993).

Disulfide bond formation is a covalent alteration that clearly influences the higher structures as well as stability of proteins. Incorrectly formed disulfide bonds, or disruption of essential disulfide bonds will change proteins structure and is likely to be detrimental to both the function and stability of the affected protein.

Peptide bonds involving Pro can exist in either the ‘normal’ native trans-configuration or the rare cis-conformation (Fisher et al., 1983). Pro isomerization denotes the conversion of proline trans-linkages to cis-linkages. Pro isomerization occurs spontaneously at a low rate after protein synthesis (Scheme 8).

Whether such a modification constitutes a covalent modification or a conformational one can be debated (Stein, 1993). However, due to the partial double-bond character of the peptide bond and the necessary disruption of this bond in the isomerization event, this modification may be classified as a covalent alteration.

Peptidyl-prolyl cis–trans isomerases (PPI), i.e. enzymes capable of catalyzing Pro isomerization, have been identified (Schmid et al., 1993). These proteins probably expedite the initial folding of nascent proteins but can also function in protein repair by the re-conversion of cis-linkages (Yaron and Naider, 1993, Schmid et al., 1993).

The presence of the altered cis-isomer has been shown to alter the physico-chemical characteristics of the affected peptide. Thus, Pro cis-isomers may delay protein folding/refolding, and have also been reported to prevent proteolysis (Yaron and Naider, 1993). Proline isomerization has not been implicated in pathological processes, but this issue probably needs further investigation to be properly resolved.