The alphabet of intrinsic disorder

Abstract

A significant fraction of every proteome is occupied by biologically active proteins that do not form unique three-dimensional structures. These intrinsically disordered proteins (IDPs) and IDP regions (IDPRs) have essential biological functions and are characterized by extensive structural plasticity. Such structural and functional behavior is encoded in the amino acid sequences of IDPs/IDPRs, which are enriched in disorder-promoting residues and depleted in order-promoting residues. In fact, amino acid residues can be arranged according to their disorder-promoting tendency to form an alphabet of intrinsic disorder that defines the structural complexity and diversity of IDPs/IDPRs. This review is the first in a series of publications dedicated to the roles that different amino acid residues play in defining the phenomenon of protein intrinsic disorder. We start with proline because data suggests that of the 20 common amino acid residues, this one is the most disorder-promoting.

Introduction

Intrinsically disordered proteins (IDPs) and intrinsically disordered protein regions (IDPRs) have recently become a hot topic in molecular and structural biology.^1,2 Computational analyses show that about 10–20% of full-length eukaryotic proteins are IDPs and that 25–40% of all protein residues are classified as IDPRs.^3-7 Furthermore, more than half of IDPs experimentally characterized by NMR are in fact IDPRs.⁸ Despite the fact that IDPs/IDPRs do not form regular, three dimensional structures on their own,⁹ they are nevertheless associated with various important cellular roles^10-24 and implicated in a number of prominent human diseases.^14,25-33 The unique structural properties of IDPs/IDPRs require new methods for their analyses³⁴ and new concepts for understanding their functions.^10,11,15

Structural and functional properties of a protein are encoded by the alphabet of the 20 naturally occurring amino acids. Therefore, to understand the unique structural and functional properties of IDPs/IDPRs it is necessary to determine how their amino acid sequences differ from ordered proteins. A number of research groups, including ours, have interrogated this problem using computational methods and determined that the amino acid compositions of IDPs and IDPRs are biased in relation to ordered proteins.^5,9,11,35-37 (add ref. 0) Based on these studies, the concept of “order-promoting” (cysteine, tryptophan, tyrosine, isoleucine, phenylalanine, valine, leucine, histidine, threonine, asparagine) and “disorder-promoting” residues (aspartic acid, methionine, lysine, arginine, serine, glutamine, proline, glutamic acid) has been proposed.³⁸ From a physico-chemical point of view, the majority of order-promoting residues are non-polar and commonly found within the hydrophobic cores of ordered proteins, whereas the majority of disorder-promoting residues are polar, often charged, and commonly found on the surfaces of ordered proteins. This notion is consistent with our current understanding of the highly dynamic structures of IDPs/IDPRs that do not form stable hydrophobic cores and probably expose most of their amino acids to the solvent.^5,11 Important exceptions to the just stated polar or charged tendencies are prolines, which are the most disorder-promoting residues³⁹ despite the non-polar nature of their side chains.

The differences in composition between ordered and disordered proteins are coupled to distinct evolutionary patterns, with IDPs and IDPRs typically displaying higher global mutation rates than ordered proteins.⁴⁰ Despite this, some IDP residues, such as aromatic amino acids (tryptophans, tyrosines, and phenylalanines), leucines and prolines are well-conserved.⁴¹ With the exception of prolines, all other conserved residues are generally less abundant in IDPs than in ordered proteins. Conserved aromatic and hydrophobic IDP residues are frequently found in protein segments with molecular recognition features (MoRFs)^42,43 and in the pre-structured motifs (PreSMos).⁸ MoRFs are short IDPRs that of 10-fold upon binding to other proteins, as well as to DNA. MoRFs determine the functions of many IDPs because they define specific protein-protein interaction surfaces, which likely explain their higher degree of evolutionary conservation.

Figure 1 and Table 1 show the statistics of amino acid compositions of proteins in four standard data sets, Swiss-Prot,⁴⁵ PDB Select 25,⁴⁶ surface residues³⁷ and DisProt,⁴⁴ where Figure 1A recapitulates Table 1 in a graphical form, and Figure 1B shows the compositional differences between the structured and disordered data sets. The Swiss-Prot database (UniProtKB/Swiss-Prot) was chosen because it contains sequence and functional information on ~550,000 proteins from all kingdoms of life and therefore represents the unbiased distribution of amino acids throughout nature.³⁷ PDB Select 25⁴⁶ contains a representative set of PDB entries with less than 25% of sequence identity. This database was chosen because of its bias toward “structural” proteins that are likely to crystalize.³⁷ Surface residues were determined with the Molecular Surface Package and a number of PDB structures of monomeric proteins that were found suitable for studying biological activities associated with protein surface properties, such as protein binding, for example.³⁷ Finally, the DisProt⁴⁴ database comprises entries of proteins and protein regions that had been experimentally verified to be intrinsically disordered.³⁷Figure 1A and Table 1 show that average proline contents in these four data sets are 4.83 ± 0.03%, 4.57 ± 0.05%, 5.6 ± 0.1% and 8.1 ± 0.6%, respectively (cprofiler.org/help.html).³⁷ Hence, IDPs contain, on average, 1.7- to 1.8-times more prolines than proteins in UniProt, or PDB Select 25, respectively. Furthermore, the overall proline content in IDPs is 1.4-times higher than on surfaces of folded proteins.

Figure 1. Amino acid determinants defining structural and functional differences between the ordered and intrinsically disordered proteins. (A) Amino acid compositions of several data sets discussed in the text (DisProt,⁴⁴ UniProt,⁴⁵ PDB Select 25⁴⁶ and surface residues³⁷). (B) Fractional difference in the amino acid composition (compositional profile) between the typical IDPs from the DisProt database⁴⁴ and a set of completely ordered proteins⁴⁶ calculated for each amino acid residue. The fractional difference was evaluated as (C_DisProt-C_PDB)/C_PDB, where C_DisProt is the content of a given amino acid in a DisProt databse, and C_PDB is the corresponding content in the data set of fully ordered proteins. Positive bars correspond to residues found more abundantly in IDPs, whereas negative bars show residues, in which IDPs are depleted. Amino acid types were ranked according to their decreasing disorder-promoting potential.³⁶

Table 1. Amino acid compositions of the standard data sets (modified from ref. 37).

Residuea	Disorder propensityb	SwissProtc	PDB S25d	Surface residuese	DisProtf
Cys (C)	0.000	1.50 ± 0.02	1.74 ± 0.05	0.78 ± 0.04	0.80 ± 0.08
Trp (W)	0.004	1.13 ± 0.01	1.44 ± 0.03	1.33 ± 0.05	0.67 ± 0.06
Ile (I)	0.090	5.90 ± 0.04	5.61 ± 0.06	2.77 ± 0.07	3.24 ± 0.13
Tyr (Y)	0.113	3.03 ± 0.02	3.50 ± 0.04	3.58 ± 0.08	2.13 ± 0.15
Phe (F)	0.117	3.96 ± 0.03	3.98 ± 0.04	2.38 ± 0.05	2.44 ± 0.13
Leu (L)	0.195	9.65 ± 0.04	8.68 ± 0.08	5.11 ± 0.08	6.22 ± 0.25
His (H)	0.259	2.29 ± 0.02	2.41 ± 0.04	2.60 ± 0.06	1.93 ± 0.11
Val (V)	0.263	6.73 ± 0.03	6.72 ± 0.06	4.01 ± 0.06	5.41 ± 0.44
Asn (N)	0.285	4.13 ± 0.04	4.58 ± 0.06	6.23 ± 0.15	3.82 ± 0.27
Met (M)	0.291	2.38 ± 0.02	2.22 ± 0.04	1.13 ± 0.04	1.87 ± 0.10
Arg (R)	0.394	5.40 ± 0.04	4.93 ± 0.06	6.56 ± 0.13	4.82 ± 0.23
Thr (T)	0.401	5.41 ± 0.02	5.63 ± 0.05	6.08 ± 0.11	5.56 ± 0.24
Asp (D)	0.407	5.35 ± 0.03	5.83 ± 0.05	8.18 ± 0.10	5.80 ± 0.30
Gly (G)	0.437	6.96 ± 0.04	7.16 ± 0.07	7.06 ± 0.11	7.41 ± 0.40
Ala (A)	0.450	7.89 ± 0.05	7.70 ± 0.08	6.03 ± 0.13	8.10 ± 0.35
Lys (K)	0.588	5.92 ± 0.05	6.37 ± 0.08	9.75 ± 0.16	7.85 ± 0.45
Gln (Q)	0.665	3.95 ± 0.03	3.95 ± 0.05	5.21 ± 0.09	5.27 ± 0.37
Ser (S)	0.713	6.83 ± 0.04	6.19 ± 0.06	6.87 ± 0.13	8.65 ± 0.43
Glu (E)	0.781	6.67 ± 0.04	6.65 ± 0.07	8.70 ± 0.17	9.89 ± 0.61
Pro (P)	1.000	4.83 ± 0.03	4.57 ± 0.05	5.63 ± 0.10	8.11 ± 0.63

^a Residues are arranged according to their decreasing intrinsic disorder propensity; ^bDisorder propensity is calculated based on the fractional difference in the amino acid compositions between the disordered and ordered proteins obtained by renormalizing these values to lie between 0 and 1; ^cSwissProt 51 is closest to the distribution of amino acids in nature among the four data sets;^45dPDB Select 25 is a subset of proteins from the Protein Data Bank with less than 25% sequence identity, biased toward the composition of proteins amenable to crystallization studies;^46eSurface residues determined by the Molecular Surface Package over a sample of PDB structures of monomeric proteins suitable for protein surface analysis; ^fDisProt 3.4 comprised of a set of experimentally determined disordered regions.⁴⁴

Figure 1B shows that proline exhibits the largest fractional change between structured and disordered proteins, and the fractional changes for the various residues provide the basis for estimating the disorder propensities given in Table 1 (see Table 1, footnote b). Indeed, the disorder propensities here yield the same P, E and S ranking for the most disorder-promoting residues as obtained in a previous study,³⁹ while the remaining amino acids show some alterations in the ranking compared with the previous study, especially for amino acids with similar disorder propensity values. Of course such estimates depend on both the methods used and the sets of proteins in the databases, which were both significantly different in the previous study³⁹ as compared with this one. Overall, the disorder propensity ranking between the two studies differ in detail but these differences are not significant.

This article starts a series of publications on the alphabet of intrinsic disorder, which is dedicated to exploring the amino acid determinants of intrinsic protein disorder. Here, we review the functions of prolines in IDPs/IDPRs and provide compelling evidence for proline-specific biological activities that may provide explanations for their high levels of abundance and conservation in disordered proteins and protein regions.

Structural Properties of Prolines

Chemical structure of prolines

Among the 20 natural amino acids, proline is unique in that it is the only imino acid; that is, the proline backbone nitrogen is bound to two alkyl carbons and lacks the usual proton (see Fig. 2). Proline’s distinctive cyclic structure renders the backbone conformation more rigid than in any other amino acid. Hence, proline peptide bonds exhibit structural features that differ substantially from other residues, also because they do not contain backbone amide hydrogen atoms at physiological pH and therefore do not form stabilizing hydrogen bonds in α-helices, or β-sheets. In consequence, prolines are rarely found as integral parts of secondary structure elements,^47,48 but rather at the ends of a-helices, or in protein loop regions.⁴⁹ Their characteristic backbone angle properties and unique structural properties in proteins and polypeptides (see below) also give rise to atypical Ramachandran plot features.^50-53 Prolines sample restricted areas of the Ramachandran space, which are primarily defined by their backbone pyrrolidine constraints.⁵⁴ They also exert pronounced effects on the backbone geometries of residues preceding them, i.e., pre-prolines.⁵⁵

Figure 2. Chemical structure of peptide fragments in trans (A) and (C) and cis conformation (B) and (D); (C) and (D) show a proline-containing fragment. The red arrows point out the steric hindrances between the C_α of the residue (−1) with the H_amide (A) or the C_α of the residue (0) (B) for the non-proline-containing peptides, and between the C_α of the residue (-1) with the C_δ (C) or the C_α of the proline (D). Ramachandran plots of non-proline, non-glycine, non-isoleucine, non-valine residues (E) and proline residues (F) result from the analysis of 1.5 million residues in 8,000 protein chains with resolution < 2 Å and backbone B-factors < 30.The contours separate the “outlier,” “allowed” and “favored” regions of the Ramachandran plots. The Ramachandran plots were adapted from commons.wikimedia.org/wiki/User:Dcrjsr. The β-strand (β), α-helix (α), α-L-helix (α_L), poly-proline II (PPII) regions of the Ramachandran plots are indicated and we show a representation of a model poly-proline II helix.

Cis-trans isomerization

Although most amino acids form peptide bonds that are in their trans-isomer conformations (> 99.5%),^56,57 Xaa-Pro peptide bonds populate both cis- and trans-states. Xaa-Pro trans isomers are indeed less favored because of relatively high steric conflicts between Xaa-Cα atoms and Pro-Cδ’s (see Fig. 2). The energy differences between proline cis/trans conformers are less pronounced than in other amino acids, which, in connection with a high energy barrier between the two isomers (~20 kcal/mol)^58,59 results in slow cis/trans interconversion rates (10⁻³ s⁻¹).⁵⁶ Hence, on average, ordered proteins contain 5–10% cis-conformers of the Xaa-Pro peptide bonds, whereas the occurrence of cis-isoforms of usual amide bonds in proteins is typically below 0.5%.^56,57 The cis-isomer content is influenced by the nature of the surrounding residues and by the types of surrounding secondary structure.^60-62 Despite these similar energy levels in disordered peptides, prolines in natively folded proteins tend to display exclusive cis-, or trans-conformations, which are primarily established via the protein fold and the resulting specific interactions with residues close in space.^63,64

Within protein Xaa-Pro motifs, C_α(Xaa)/C_α(Pro) distances of trans-proline conformations are on average 1.5 Å larger than for cis proline isomers;^65,66 however, these effects are not systematic and strongly influenced by the nature of Xaa. In most folded proteins, isomer-specific structural changes are local, and vanish at a distance of 2–3 residues from the proline of interest. More extended conformational rearrangements have only been observed for a few cases.⁶⁷ From a local point of view the effects that proline cis/trans isomers induce in polypeptide chains are important. Cis-isoforms result in turn-like structures, whereas trans-isoforms favor locally extended conformations (see Fig. 2). In protein folding cis/trans isomerization plays an important role and often functions as the rate limiting step in the overall folding process.⁶⁴ Important cellular enzymes such as peptidyl-prolyl isomerases (PPIases) accelerate proline isomerization processes and thereby enhance the kinetic rates with which thermodynamic equilibrium states are reached. The relationships between PPIases and IDPs will be discussed in more detail, later in the article. One aspect that we want to stress is that proline cis-trans characteristics and behaviors of IDPs are similar to those of peptides. IDPs display cis population averages of ~5–10% and, therefore, IDPs with 10 or more prolines have high probabilities for multiple cis conformations. This creates substantial diversity in population conformers that sample a vast conformational space.

On the hydrophobicity of the proline residue

In the initial hydrophobicity scale development, the backbone was considered to be constant for all of the amino acids, and thus only the side chain was considered to be contributing to the values of the scale.⁶⁸ However, with regard to residue hydrophobicity, the proline imine brings the backbone into play. That is, upon burying a typical amino acid residue, the backbone has both hydrogen bond donors and acceptors, leading to helices, sheets, turns or other structures in which the backbone hydrogen bonding potential is self-satisfied. For proline, on the other hand, the backbone has hydrogen bond acceptors but no donors, and for this reason it is costly from an energetic point of view to sequester the proline backbone from the solvent. The consequences of this donor / acceptor imbalance in the backbone are that, compared with valine, the other amino acid with a side chain containing 3 aliphatic carbons, proline is less frequently buried and more frequently on protein surfaces (Table 1; Fig. 1B). In this regard, the solubility of the individual amino acids is generally inversely correlated with hydrophobicity, yet proline is by far the most soluble of the amino acids at neutral pH,⁶⁹ and furthermore, polyproline is much more water soluble than polyglycine, polyalanine and polyleucine due to polyproline’s lack of an NH group.⁷⁰ Thus, despite its hydrophobic side chain, the proline residue is very hydrophilic.

Prolines in IDPs/IDPRs: Structural and Functional Roles

The polyproline type II helix as a unique binding interface

The unusual chemistry of prolines imposes several constraints on neighboring residues and proline-rich motifs (PRMs) have high propensities for adopting non-classical conformations such as the polyproline type II (PPII) helix.^71-73 PPII helices are left-handed, extended structures that contain three residues per turn and no internal hydrogen bonding. They are surprisingly abundant structural scaffolds in virtually every proteome. Even ordered proteins contain short PPII stretches, and PPII backbone dihedral angles (−75°, 150°) are frequently observed in amino acids other than prolines.^74,75 In PPII helices, side-chain and backbone carbonyls are solvent-exposed and often engage in intermolecular hydrogen bonds, thereby mediating generic intermolecular recognition events of rather low ligand specificities. In turn, a great number of proline-recognition domains (PRDs) interact with PRMs and PPII helices, among which SH3 and WW domains are probably the most well-known examples. The giant human protein titin, with a total of 34,000 amino acids, contains ~550 SH3 binding motifs, of which ~100 are found in PRMs.^76-79

PPII-mediated interactions regulate diverse sets of particular cellular functions.^72,80,81 A statistical analysis on 74 scaffolding proteins for example, has revealed that this class of proteins contained predicted degrees of disorder (i.e., 49.7% by IUPred, 63.36% by VSL2 and 47.82% by FoldIndex⁸²) that were comparable to highly disordered classes of proteins, such as transcription factors¹⁴ and RNA chaperones.⁸³ Furthermore, 26 of the most disordered scaffolding proteins contained average proline contents of 11.2 ± 0.4%, which appears to predispose PRM-proteins to function as hubs in protein-protein interaction networks.^84-91 PRMs, or polyproline regions (PPRs) are also found in the proteomes of several viruses, such as hepatitis E (HEV), rubivirus and cutthroat virus (CTV).⁹² Although the functional significance of PPRs in viruses remains poorly understood, they appear to mediate interactions of viral proteins with cellular host factors to modulate viral replication efficiencies.⁹³ A recent study further demonstrated that sequence variabilities in viral PPRs play important roles in adaptation and in specifying the range of host cells.⁹² PPRs of HEV genotypes 3 and 4, for example, indicating viral variants of zoonotic origins that can infect humans and animals, are twice as heterogeneous than PPRs in the HEV genotype 1 variant, which is purely anthropotropic and can infect humans only.⁹²

Also, in these PRM-containing binding regions, proline not only is involved in maintaining an open conformational state compatible with binding, it is also the most important residue that contacts the partner protein. An analysis of short linear motifs (SLiMs, also termed Eukaryotic Linear Motifs, ELMs) showed that Pro is the residue most significantly enriched in sites that determine binding specificity of the motif (restricted sites, RSs).⁹⁴

PRMs and IDP conformations

Based on the high levels of PPII sequence conservations in folded proteins, it has been suggested that these structural elements constituted a separate class of secondary structure elements,⁷⁵ with two major functions: To promote super-secondary structures, such as PPII/α-helical interactions, and to form inter-domain linkers.⁷⁵ In IDPs, the unique propensities of PPII structures in rigidifying polypeptide backbone conformations is thought to spatially separate functionally important protein regions.¹³ An example for such a separation function is provided by the human oncoprotein and transcription factor p53 that contains two PRMs in PPII-type conformations. One separating the intrinsically disordered N-terminal transactivation domain (NTAD) of p53 from its folded DNA-binding domain (DBD), the other one within the NTAD separating a helical pre-structured segment and two pre-structured turns³² that mediate distinct protein-protein interactions.^33,95-97 Similarly, two transactivation domains within the C-terminus of herpes simplex virus protein 16 (VP16) are separated by a conserved PRM (₄₅₂PGPGFTPHDSAP₄₆₄).^98,99 In both cases, spatial positioning via PRMs likely regulates independent transcription activation processes that rely on different interactions with the RNA polymerase II machinery.^98,100 By analogy, two helical segments within the C-terminal portion of human securin, potentially mediating the interactions with separase,¹⁰¹ are separated by a PRM (₁₆₂PPSPVKMPSPP₁₇₃), whereas a PRM in the human transcription factor FoxA3 (₂₅₀PPQPPPPAPEP₂₆₀) separates its DNA- from its histone-binding domain.

Whereas PRMs often induce extended conformations, many IDPs are usually more compact than chemically denatured proteins of comparable lengths,^16,33 whose conformational behaviors still cannot be described as random coils.¹⁰² Because most IDPs are not restricted to stable three-dimensional architectures, to seamlessly vary their degrees of global compactions is thought to constitute an important functional IDP feature.^103,104 Therefore, the ability of PRMs to elongate and stiffen polypeptide chains has to be discussed in this context. For example, proline-rich salivary proteins possess significantly higher radii of gyration than are expected for unfolded polypeptides of similar lengths.¹⁰⁵ It has been proposed that organized PPII helices in these proteins result in larger collisional cross sections that facilitate their interactions with tannins,¹⁰⁶ which form the basis of the sensory perception of astringency.¹⁰⁷

Extending IDP structures via PRM-mediated effects may not necessarily be restricted to long proline sequences alone. In fact, a strong correlation between the number of prolines in an IDP and its radius of gyration has been established.¹⁰⁸ Such expansions have been attributed to the unique properties of Xaa-Pro peptide bonds to adopt backbone dihedral angles that correspond to extended conformations. However, prolines can also promote β-turn conformations, which elicit various degrees of polypeptide chain compactions.^109,110 The degree of compaction can moreover be tuned by cis/trans equilibria.¹¹¹ In line with these observations, mutating proline residues in a short, disordered elastin-like peptide has been shown to induce a stepwise expansion.¹¹² In contrast, the overall stiffness of four disordered peptides were reported to be more correlated with their PPII contents than their proline counts, whereas the intrinsic capacities for hairpin structures strongly correlated with the numbers of glycines and prolines.¹¹³ Therefore, the possible role(s) of prolines in compacting, or expanding IDPs conformations would depend on the context. While increasing the number of prolines in PPII conformations appears to rigidify IDPs, a high-abundance of prolines in combination with favorable glycine contents, or with selective positioning of charged and/or hydrophobic residues, gives rise to preferred hairpin conformations that result in more collapsed structures.¹¹⁴

Prolines as secondary structure-breakers

Because of their unique chemical and structural properties, and because of their negative influence on classical secondary structure, it is tantalizing to speculate that proline positions in folded, but also in intrinsically disordered proteins, had been evolutionary selected, as well as conserved, for their unique capacities to modulate the structural propensities of neighboring protein residues. In folded proteins, a preference for prolines at helix-capping positions had been recognized very early on.¹¹⁵ Depending on the data set, or the methods for defining secondary-structures, prolines in N- or C-cap positions preferentially occur between N_cap-1 and N_cap+2 and between C_cap and C_cap+3, respectively.^116-120 In these instances, high proline frequencies do not relate to helix stabilization effects, but more likely function as border elements that confine existing secondary structures to certain lengths.^121,122 In IDPs, proline positions may have been evolutionarily conserved to ensure that protein regions with residual structural propensities, such as MoRFs for example, retain their partially folded states in a balanced manner. Recent findings support this notion by showing that prolines at positions that flank partially folded IDP segments (PreSMos) occur more frequently³³ and display higher levels of positional conservation, than elsewhere in these proteins.⁹⁴ In essence, this notion represents an extension of the “proline bracket” concept,^123,124 according to which prolines in segments flanking protein interaction sites negatively modulate the propagation of α-helices and β-strands. Such effects may preserve various degrees of conformational IDP plasticity, which may eventually steer different binding behaviors in protein-protein interactions.

Prolines and prevention of amyloid-like aggregation

As mentioned earlier, positional proline effects in IDPs may preserve levels of disorder in regions with residual structural propensities. This, in turn, may also reduce the likelihood for spontaneous IDP aggregation, which is often cytotoxic, results in cell death and produces several devastating disease phenotypes.¹²⁵ In fact, many different IDP aggregation processes proceed via intermediate conformations that harbor folded aggregation cores, which progressively expand into highly ordered macromolecular assemblies such as amyloids fibrils, for example. In folded proteins, uncontrolled association events via existing secondary structure elements are often prevented by combinations of dedicated structural features that “protect” aggregation-prone entities such as peripheral β-strands. These include “covering” interactions with loop- or helical-segments, β-strand distortions via inward-pointing, charged residues, incorporation of prolines, β-bulges, or glycine-promoted bends and twists, or via formations of continuous β-sheets to yield β-barrels.¹²⁶ Therefore, prolines at the domain boundaries are often highly conserved and mutating them usually promotes aggregation.^125,127 In depth analyses of various protein segments that display high propensities for β-aggregation have shown that β-breaking prolines, together with charged amino acids such as lysines, arginines, glutamates and aspartates, are specifically enriched at these positions and thought to serve as anti-aggregation “gatekeepers.”¹²⁸

Elastomeric proteins

Elastomeric proteins exemplify another important aspect of the “usage” of prolines for specific biological functions. These proteins display remarkable propensities for elastic recoiling behaviors and undergo innumerous reversible deformations in the course of their lifetimes, which are directly related to their specific biological functions in tissues and other biomaterials.¹²⁹ In all vertebrates, elastomeric proteins constitute the building blocks of blood vessels; in insects, they give rise to specialized structures such as a spider’s silk; in arthropods they make up the intrinsic energy storage apparatus that enables jumping. Some of these proteins are IDPs that have evolved to aggregate in a controlled manner to form dedicated, rubber-like structures that are able to be stretched under extreme physical circumstances and to recoil by itself later.¹²⁹ Although these elastomeric proteins can spontaneously organize themselves into elastomeric protein complexes, they are surprisingly resistant to forming β-rich amyloid structures.¹²⁵ Despite their sequence and functional diversities, all elastomeric proteins and IDPs contain unusually high proline and glycine contents,¹³⁰ which clearly separates elastomeric proteins from amyloidogenic proteins and peptides (Fig. 3).¹³⁰ Prolines in these structures, together with glycines, prevent the formation of long, stable amyloid structures, whereas their relatively high hydrophobicities promote aggregation-like behaviors such as recoiling. Thus, amino-acid compositions of elastomeric proteins depend on a fine balance between polypeptide hydrophobicity and high proline and glycine contents.^125,130

Figure 3. A two-dimensional plot correlating proline and glycine content for a wide variety of elastomeric and amyloidogenic peptides. Elastomeric proteins are characterized by high GP content and are located in the upper-right part of this plot. Contrarily, amyloidogenic peptides are characterized by low PG content and therefore are located in the left bottom corner of the plot. The coexistence region (shaded in gray) contains P and G compositions consistent with both amyloidogenic and elastomeric properties. Elastomeric proteins, including the domains of elastin, major ampullate spindroin (MaSp) 2, flagelliform silk, the elastic domains of mussel byssus thread, and abductin, appear above a composition threshold (upper dashed line). Amyloidogenic sequences are primarily found below the PG-threshold, along with rigid lizard egg shells, tubulliform silk (TuSp1), a protective silk for spider eggs, and aciniform silk (AcSp), used for wrapping prey. The coexistence region contains amyloid-like peptides as well as the elastomeric adhesive produced by the frog Notaden bennetti, the PEVK domains of titin, wheat glutenin protein, and the strongest spider silks, namely MaSp1 and minor ampullate spindroin (MiSp). Figure reproduced from ref. 130 Abbreviations: AcSp, aciniform silk; MaSp, major ampullate spindroin; MiSp, minor ampullate spindroin; TuSp1, tubulliform silk.

Proline-Directed Post-translational Modifications

Post-translational protein modifications (PTMs) range from enzymatic cleavage reactions of peptide bonds to covalent additions of particular chemical groups, lipids, carbohydrates or even entire proteins onto selected subsets of amino acid side chains. PTMs extend the range of amino acid structures and properties and greatly diversify the functional space of virtually every proteome.¹³¹ With regard to our subject, strong correlations between predicted, and experimentally verified protein disorder and the occurrence of PTMs exist,²⁶ the most common among which are phosphorylation,^132,133 ubiquitination,¹³⁴ acetylation,¹³⁵ methylation^136,137 and glycosylation¹³⁸ reactions. These PTMs are typically involved in the regulation and control of various signaling and recognition processes (for example see ref. 139). Although direct post-translational modifications of proline residues only have a limited range of functions, prolines play important roles in the regulation of the occurrences of other PTMs.

Proline PTMs

Annotated lists of experimentally verified PTMs, in Swiss-Prot and other databases, clearly indicate that prolines are primarily subject to post-translational hydroxylation (selene.princeton.edu/PTMCuration/),¹⁴⁰ which can occur on Cβ ((2S,3S)-3-hydroxyproline) or Cγ ((2S,4R)-4-hydroxyproline) positions. These nonreversible conversions of prolines to (2S,4R)-4-hydroxyprolines (Hyps) are catalyzed by prolyl 4-hydroxylase enzymes and surprisingly, represent the most common PTM in humans.¹⁴¹ In fact, Hyps are more abundant in animals than seven of the most “common” amino-acid types: Cys, Gln, His, Met, Phe, Trp and Tyr.¹⁴² The best known roles for Hyp’s are in stabilizing collagen triple helices.¹⁴¹ Proline hydroxylation enhances the stability of trans-isoforms of Xaa-Pro peptide bonds relative to cis-isoforms.¹⁴¹ Since proline trans-isoforms already constitute the major conformations in IDPs (~90%), hydroxylation is not thought to play additional important roles in their conformational behaviors. Apart from their roles in collagen-like coiled-coil structures, Hyp’s are also found in many other connective tissue proteins, in proteins with collagen-like domains, as well as in the (partially) disordered proteins elastin, conotoxin and argonaute 2.¹⁴¹

The best example for Pro-hydroxylation generating a signal for regulation is hypoxia-inducible transcription factor 1α (HIF-1α). At low oxygen conditions (hypoxia), HIF-1α activates transcription by recruiting the general coactivator CBP/p300 via interaction with its TAZ1 domain. Upon elevation of oxygen level, Pro564 of HIF-1α becomes hydroxylated, it binds to the ubiquitin ligase von Hippel-Lindau factor and undergoes ubiquitination that targets the protein for degradation.¹⁴³

Proline-directed limited proteolysis

Structural disorder and the extended structure ensured by Pro residue(s) are also involved in directing the action of proteases in limited proteolysis. Due to being an irreversible modification, limited proteolysis is a serious and tightly regulated signaling decision by the cell. For example, calpain, the intracellular protease only cleaves specific substrates if activated by calcium and released by its tight inhibitor, calpastatin, and shows a strong preference for regions of local structural disorder dominated by Pro residues (Tompa et al. J Biol Chem 2004; 279: 20775-85).¹⁴⁴ Actually, Pro is depleted around the scissile bond (positions P2, P1 and P1’), but is highly significantly enriched in flanking regions (positions P4, P3 and P2’ to P6’).¹⁴⁴

Roles of prolines in protein phosphorylation

Many serine/threonine kinases modify substrate sites that constitute integral or distal parts of kinase consensus motifs.^145,146 Within these consensus motifs, proline residues often define substrate site specificities. Examples include many proline-directed protein kinases, such as cyclin-dependent kinases (CDKs),^147,148 the mitogen-activated family of protein kinases (MAPKs),^149,150 extracellular signal-regulated kinases (ERKs), stress-activated protein kinases/c-Jun-N-terminal kinases (SAPKs/JNKs), p38 kinases, glycogen synthase kinase-3 (GSK3) and Polo-like kinases (PLKs),¹⁵¹ all of which require prolines at positions +1 with respect to the sites of modification. These kinases play important roles in diverse cellular processes, such as cell-cycle progression, sensing of metabolic states, regulating of cellular growth, mediating of intracellular signaling, as well as executing deterministic cell response behaviors. In turn, mutations of proline-directed kinase consensus- and phosphorylation-sites are often involved in different forms of cancer and in neurodegenerative disorders.^152-157 A second, less stringent proline −2 position has recently been identified as a supplementary specificity determinant for some proline +1-directed kinases.^158,159

Whereas many prolines positively regulate kinase activities, by targeting them to their phosphorylation sites, proline residues within kinase consensus motifs can also weaken kinase activities, especially when they occur at positions −1 and −2, relative to the PTM sites,¹⁵⁹ or even at positions +1.^160-163 In other phosphorylation reactions, prolines play important roles in serving as specific kinase docking sites that are distal from actual phosphorylation sites but key to recruiting kinases to substrate proteins.^164-166 In addition to kinases, the enzymatic properties of phosphatases are also modulated by prolines, either in the vicinities of phospho-sites¹⁶⁷ or at distal docking sites.^168,169 Finally, prolines that are close to modified substrate residues may critically influence PTM-mediated protein-protein interactions. It has been shown that phosphorylated serines, or threonines, followed by a proline, are more specifically recognized by subsets of 14–3-3 proteins¹⁷⁰ or by Group IV WW domains.^171,172

Roles of prolines in protein glycosylation

Glycosyltransferases are classes of enzymes that transfer sugar moieties onto proteins and they are strongly influenced by the presence of prolines in their substrate proteins. N-glycosylation of asparagines within the Asn-Xaa-Ser/Thr motif has been found to have a very low penetrance when the Xaa residue is a proline or when prolines are present at the +1 positions. In contrast, N-glycosylation is greatly enhanced when prolines are present at the −2 positions.^173,174

O-glycosylation preferentially occurs in protein regions with high proline contents^138,175 and particularly high proline frequencies have been reported for positions −1 and +3 relative to O-glycosylation sites.¹⁷⁶ Both phosphorylation and glycosylation do not affect proline cis-conformer contents of phospho-Ser/Thr/Tyr-Pro motifs^177-179 and of glyco-Ser-Pro motifs,¹⁸⁰ respectively.

Roles of proline isomerases in PTM establishments

As mentioned previously, proline cis/trans isomerization reactions play important roles in protein folding and refolding processes, via the establishment of rather long-lived kinetic intermediates. Therefore, classes of cellular enzymes, so-called peptidyl-prolyl isomerases (PPIases), specifically enhance proline cis/trans isomerization without affecting their thermodynamic equilibrium states.¹⁸¹ PPIases are evolutionarily conserved and often characterized as foldases, or annotated as catalytic structural chaperones.¹⁸² Due to their inherent differences in stereochemistry, proline cis/trans isomers can also define different functional states of proteins.¹⁸³ In these cases, PPIase activity drastically impacts protein function, as has been shown for the folded SH2 domain of the interleukin-2 inducible T-cell kinase (Itk)^184-188 and the PHD-BRD tandem domain of the MLL1 protein.^189-204 In both cases, proline cis/trans isomerization leads to large inter-domain conformational changes that subsequently affect protein-protein interaction behaviors.

Enhanced proline cis/trans isomerization in the presence of PPIases, leads to rapid sequestration of binding-competent protein states, which shifts the global population equilibrium toward the structure with which the more abundant binding partner interacts.^184,189,191 Therefore, without changing protein free energies of cis/trans isomers, PPIases are capable of promoting new cis/trans distributions via additional factors that form complexes with, and thereby stabilize, individual isomer states. Because many IDPs are PPIase substrates,^192-194 enzyme-controlled proline cis/trans isomerization processes provide intricate extensions to the long list of possible proline functions in IDPs. For example, proline isomerization controls switching of the adaptor protein Crk between two conformations: an auto-inhibitory state is stabilized by intramolecular association of two, tandem SH3 domains via a flexible linker IDPR containing a cis-proline isomer and a non-inhibited, activated conformation results from the promoted interconversion of this proline into its trans form. In turn, this particular cis/trans isomerization is targeted by the PPIase cyclophilin A.¹⁹¹

Among other PPIase enzymes, the phospho-dependent Pin1 [protein interacting with NIMA (never in mitosis A)-1] enzyme is of special interest. Pin1 functions in phospho-dependent signaling by catalyzing cis/trans interconversions of pSer/pThr-Pro peptide bonds in their phosphorylated states.¹⁵¹ Structurally, Pin1 consists of an N-terminal phospho-recognition WW domain and a C-terminal, catalytic PPIase domain.¹⁹⁵ Whereas cis/trans population ratios in these Ser/Thr-Pro motifs are not affected by phosphorylation in a peptide/IDP context, cis/trans isomerization rates are severely reduced when the motif is modified.^177-179 In folded proteins, the protein fold and amino acids that surround these Ser/Thr-Pro sites often stabilize, or de-stabilize one of the isomers. Enzymes such as Pin1 establish faster inter-conversion rates upon phosphorylation, which enables a 2-way control over the protein’s function:^151,196-198 One way is regulation via phosphorylation, processed by a kinase or removed by a phosphatase, and a second way is control via isomerization, accelerated by a non-phospho-dependent PPIase or by the phospho-dependent PPIase Pin1.

Could similar 2-way controls be utilized by IDPs? A limitation is that Ser/Thr-Pro cis/trans thermodynamic equilibrium is not greatly affected by protein phosphorylation but is substantially affected in folded proteins. A supplementary IDP protein partner is thus required for the emergence of a function of the phospho-dependent cis/trans isomerization. For example, 2-way control like that discussed above has been observed for the pSer7-Pro8 motif within the intrinsically disordered, C-terminal domain (CTD) of RNA polymerase II, whose phosphorylation status correlates with transcriptional activity. Only the cis-isomer of the modified peptide motif serves as a substrate for the Ssu72 phosphatase.^199,200 Hence, Ssu72-mediated dephosphorylation of the CTD pSer7-Pro8 sequence occurred much faster when Pin1 was present and proline cis/trans isomerization has been identified as the rate-limiting step in Ser7 dephosphorylation.

Another interesting example is afforded by pSer62 of the c-Myc oncoprotein, a key regulator of cell growth that is stabilized by Ser62 phosphorylation. Dephosphorylation by PP2A only occurs when Thr58-Pro59 is phosphorylated and Pin1 is present. Therefore, pSer62 dephosphorylation may similarly require Pro59 to be in the cis isomer state.²⁰¹ Analogous relations between the Alzheimer disease-associated protein Tau, Pin1 and PP2 have been observed.²⁰² Based on these examples, it is evident that PPIase activities represent important supplementary levels of regulatory controls in many cellular processes, although, in some cases, it remains unclear whether Pin1 binding, or catalysis, constitutes is the mechanism of action.^203,204

Conclusions

Examples presented in this review show that there are multiple, distinct mechanisms by which proline regulates IDP and IDPR structure and function. The unique chemical properties of proline define its role as a modulator of secondary structural elements, but also its propensity to promote specific structural motifs such as the polyproline type II helix. In turn, these features appear to be especially important in regulating a multitude of functional IDP and IDPR properties that include their aggregation propensities. In addition, nature seems to have taken full advantage of the slow proline cis/trans isomerization characteristics in a number of biological processes that, altogether, extend the impressive functional range of this unique imino acid.

Glossary

Abbreviations:

intrinsically disordered proteins

IDPs

intrinsically disordered protein regions

IDPRs

proline-rich motifs

PRMs

proline-recognition domains

PRDs

polyproline type II

PPII

hydroxyproline

Hyp

protein interacting with NIMA

Pin1

never in mitosis A

NIMA

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.