The intrinsic disorder alphabet. III. Dual personality of serine

Abstract

Proteins are natural polypeptides consisting of 20 major amino acid residues, content and order of which in a given amino acid sequence defines the ability of a related protein to fold into unique functional state or to stay intrinsically disordered. Amino acid sequences code for both foldable (ordered) proteins/domains and for intrinsically disordered proteins (IDPs) and IDP regions (IDPRs), but these sequence codes are dramatically different. This difference starts with a very general property of the corresponding amino acid sequences, namely, their compositions. IDPs/IDPRs are enriched in specific disorder-promoting residues, whereas amino acid sequences of ordered proteins/domains typically contain more order-promoting residues. Therefore, the relative abundances of various amino acids in ordered and disordered proteins can be used to scale amino acids according to their disorder promoting potentials. This review continues a series of publications on the roles of different amino acids in defining the phenomenon of protein intrinsic disorder and represents serine, which is the third most disorder-promoting residue. Similar to previous publications, this review represents some physico-chemical properties of serine and the roles of this residue in structures and functions of ordered proteins, describes major posttranslational modifications tailored to serine, and finally gives an overview of roles of serine in structure and functions of intrinsically disordered proteins.

Introduction

Proteins are the major building blocks of life. Although for the long time it was believed that any protein-based function should rely on unique protein structure, the modern understanding of the biologically active proteins is extended by consideration of the intrinsically disordered proteins (IDPs) and hybrid proteins containing both ordered domains and intrinsically disordered protein regions (IDPRs).^1,2 These proteins play a number of important roles in various cellular processes.^3-17 They are highly abundant in nature,^18-22 have intriguing structural and functional properties,²³ and are commonly involved in the pathogenesis of various diseases.^7,24-29 IDPs/IDPRs are promiscuous binders, being known to interact in various protein-protein, protein-nucleic acid, and protein-small molecule interactions. Many IDPs/IDPRs are able to undergo at least partial disorder-to-order transition caused by their interactions with binding partners.^{3-6,9,17,30-36} In molecular recognition events, some IDPs/IDPRs are known to interact with several unrelated binding partners; i.e., are engaged in so-called one-to-many binding, and are able to fold differently at binding to different binding partners.^16,30,37 The binding modes attainable by IDPs/IDPRs range from the formation of highly stable complexes, to action as dynamic, sensitive, controllable, and regulatable “on-off” switches in signaling networks, to interactions associated with constant “bound-unbound” transitions, to highly specific interactions without very high affinity.^4,14 Although some partial folding during the IDP/IDPR-based interactions is very common,^31,32 many other IDPs/IDPRs are involved in the formation of “fuzzy complexes,” where an IDP/IDPR keeps a certain amount of disorder in its bound state.^35,38-40 Often, not the entire disordered chain is involved in the partner binding. Instead IDPs/IDPRs might contain specific interacting regions, which are only loosely structured in their unbound forms, but gain specific structure at binding.^30,31,41,42 As a rule, IDPs/IDPRs return to their intrinsically disordered state after the completion of a particular function. All this constitutes an important arsenal of the unique physiological properties of IDPs/IDPRs that determines their ability to exert different functions in different cellular contests according to a specific conformational state.²³

Although amino acid sequences code for both ordered proteins/domains and IDPs/IDPRs, the corresponding sequence codes are dramatically different.^4,20,43 The aforementioned differences between amino acid sequence codes for order and disorder are many, with the mot general being amino acid compositions. In fact, in comparison to ordered proteins/domains, IDPs/IDPRs are globally depleted in order-promoting residues (cysteine, tryptophan, isoleucine, tyrosine, phenylalanine, leucine, histidine, valine, asparagine, and methionine) being instead enriched in disorder-promoting residues (lysine, glutamine, serine, glutamic acid, and proline).^{4,20,23,43-45}

Serine is the third most common disorder-promoting residues. Figure 1 and Table 1 represent the result of a statistical analysis of the amino acid content of proteins in 4 standard datasets, such as DisProt,⁴⁶ UniProt,⁴⁷ PDB Select 25,⁴⁸ and surface residues.⁴⁵ The mean contents of serine in these data sets are 8.65 ± 0.43%, 6.83 ± 0.04%, 6.19 ± 0.06%, and 6.87 ± 0.13%, respectively (http://cprofiler.org/help.html).⁴⁵ In other words, the content of serine residues in IDPs/IDPRs exceeds those of the averaged proteins in UniProt, the ordered proteins in PDB, and the surfaces of ordered proteins by the factor of 1.27, 1.40, and 1.26, respectively.

Figure 1.

Amino acid determinants defining structural and functional differences between the ordered and intrinsically disordered proteins. Fractional difference in the amino acid composition (compositional profile) between the typical IDPs from the DisProt database⁴⁶ and a set of completely ordered proteins⁴⁸ 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 dataset 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.⁴⁴

Table 1.

Amino acid compositions of the standard data sets (modified from ref.⁴⁵)

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

^aResidues 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;

^cSwissProt 51 is the set closest to the distribution of amino acids in nature among the 4 datasets;⁴⁷

^dPDB 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;⁴⁸

^eSurface 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.⁴⁶

This article continues a series of publications on the intrinsic disorder alphabet. The goal of this series is the exploration of the amino acid determinants of protein intrinsic disorder. Some functions of serine in IDPs/IDPRs and in ordered proteins/domains are reviewed below to illustrate the panoply of serine-specific functions in disordered proteins and regions.

Structural properties of serine

Serine (Ser, S, see Fig. 2) is one of the 20 major amino acids found in proteins. Being encoded by the standard genetic code, serine is unique among the amino acids in that it has 6 codons, from 2 distinct groups: the TCN group (TCA, TCC, TCG, and TCT) and the AGY group (AGC and AGT).⁴⁹ Since there is a noticeable bias toward TCN codons in conserved ancient essential proteins, it is believed that evolutionary the TCN group predates the AGY group.⁵⁰ Serine has a chemical formula HO₂CCH(NH₂)CH₂OH and is considered as a polar residue due to the presence of the hydroxyl group. The molecular mass of this nutritionally nonessential (dispensable) amino acid is of 105.09 Da, whereas the molecular mass of Ser residue is 87.09 Da. Serine is a water soluble (up to 250 mg/ml), uncharged amino acid that has a pK_a of the carboxyl and amino groups of 2.21 and 9.15, respectively. The volume of serine residue is of 73 ų and its surface is of 115 Ų (http://www.rfdn.org/bioinfo/APDbase.php). The side chain of serine contains one methylene group and the hydroxyl functional group (see Fig. 2), and serine is one of the 2 biological amino acids with a hydroxyl substituted side chain. Therefore, although serine has a non-charged side chain, it is a polar amino acid that often serves as the nucleophilic amino acid at the active sites of several enzymes. Serine residue has a non-polar surface of 56 Ų, and the estimated hydrophobic effect associated with the burial of this residue is 1.40 kcal/mol, whereas the estimated hydrophobic effect associated with the burial of side chain of this residue is 0.2 kcal/mol.⁵¹ In ordered proteins, serine residues are predominantly located on protein surface so that they have access to the solvent. In fact, 70% (375 of 533) of serine residues in analyzed structures of 9 folded proteins are classified as exposed since they have solvent exposed areas of >30 Ų, and 30% of serines in folded proteins possess solvent exposed areas of <10 Ų and therefore are buried.⁵²

Figure 2.

Chemical structure of L-serine.

Some evolutionary properties of serine

According to François Jacob, protein evolution is much more similar to the tinkering process than to the free design,⁵³ since “nature does not create a new protein function from a blank canvas nor with unlimited resources, but instead evolves through innovation with existing proteins.”⁵⁴ Although this “evolutionary tinkering” hypothesis suggested that the probability of mutation is equal for all amino acid residues, recent study based on the analysis of the reconstructed ancestral sequences combined with the computational evaluation of the mutational probabilities of all the amino acid residues revealed that the mutational activities of different residues are highly divergent.⁵⁴ This analysis revealed that serine evolves very fast by mutating very often and serving as a more likely target of mutations. Therefore, serine is the fastest-evolving amino acid residue that serves as a mutational hot spot being characterized by the highest mutability and high targetability.⁵⁴ The authors argued that serine is very close in mutational space to most other amino acid residues, being, in most cases, only one nucleotide mutation away. Also, it less likely that the amino acid substitution to serine will be rejected by selection since serine does not possess a bulky or charged side chain, being characterized by moderate physico-chemical properties (see below).⁵⁴ Despite the fact that serine clearly serves as fast-evolving mutational hub, the mutational activity of this residue in the phosphorylation sites is dramatically suppressed, indicating that, in special circumstances, the mutational properties of amino acid residues are modulated by the cells to prevent functionally important serine residues from evolving as fast as they would under normal circumstances.⁵⁴

It has been also emphasized that serine is one of the several residues (together with cystein, metionine, histidine, and phenylalanine) that are being accrued significantly in the human proteome.^55-57 Finally, based on the analysis of frequencies of mutations in the 1000 Genomes data and in databases collecting information on inherited diseases associated with variants, an important conclusion has been made that serine is involved in proportionally more disease-related mutations than polymorphisms.⁵⁷

Biological significance of free serine

Metabolic significance of L-serine

Although L-serine is typically classified as one of the nutritionally nonessential (dispensable) amino acids, it is clearly indispensible metabolically due to its multifactorial involvement in a number of essential cellular processes.⁵⁸ For example, serine, that can be synthesized in the body from an intermediate of the glycolytic pathway, 3-phosphoglycerate, serves as the major source of the one-carbon units required for the methylation reactions that take place via the generation of S-adenosylmethionine.⁵⁸ Serine serves as a building block for amino acids (e.g. glycine, cysteine), taurine, phospholipids, and is important for the synthesis of purines, pyrimidines, tryptophan (in bacteria), and heme.⁵⁹ For example, serine hydroxymethyltransferase is responsible for the conversion of serine to glycine (retro-aldol cleavage) and simultaneous conversion of tetrahydrofolate to N⁵,N¹⁰-methylene-tetrahydrofolate (hydrolysis).⁵⁹ Since the retro-aldol cleavage reaction is reversible, the serine hydroxymethyltransferase controls the interconversion of serine and glycine. Furthermore, the N⁵,N¹⁰-methylene-tetrahydrofolate generated during the tetrahydrofolate hydrolysis serves as a precursor for the 5-methyl-tetrahydrofolate, which is an intermediate in the homocysteine methyltransferase (methionine synthase)-catalyzed methylation of homocysteine to methionine.⁵⁸ Also, serine is involved in the “transsulfuration pathway” leading to cysteine synthesis.⁵⁸ Here, cystathionine β-synthase first catalyzes the cystathionine synthesis from serine and homocysteine, and then cystathionine γ-lyase split cystathionine into cysteine and α-ketobutyrate.⁵⁸ Therefore, there is an intricate metabolic relationship between serine metabolism, methionine, glycine, homocysteine, cystathionine, and cysteine.⁵⁹

Because of this interlink, a decreased synthesis of serine is known to be accompanied by the decreased concentrations of serine, glycine, and 5-methyltetrahydrofolate.⁵⁹ Decreased levels of serine production in humans are associated with the serine deficiency disorders originating from the inborn errors of serine metabolism (e.g., caused by the inherited deficiency in the 3-phosphoglycerate dehydrogenase (3PGDH), an enzyme playing a crucial role at the initial stages of the de novo serine biosynthesis from 3-phosphoglycerate). These serine deficiency disorders are manifested by the major neurological symptoms such as congenital microcephaly, seizures, psychomotor retardation or polyneuropathy.⁵⁹ On the other hand, an increased activity of 3PGDH was reported in human colon carcinoma, rat sarcoma, and rat hepatoma cell lines during the proliferative phase.^58,60–62 More recently, the markedly higher expression of the 3PGDH gene was found in several breast cancer cell lines and in estrogen receptor-negative breast cancers.^58,63

Curiously, it has been shown that L-serine can also act as a glia-derived trophic factor that strongly supports the survival and differentiation of neurons.⁶⁴ Another illustration of the indispensable biological role of L-serine is the fact that this amino acid serves as the only source for the formation of D-serine, which is known to play a critical role as a neuromodulator in the brain.⁵⁸

Biological roles of D-serine

The fact that mammalian brain tissues contain substantial quantities of free D-serine was reported for the first time in 1992.⁶⁵ D-serine concentrations in the mammalian CNS and peripheral tissues are unexpectedly high, non-uniformly distributed, vary with age,⁶⁶ typically D-serine concentrations amount one-third of L-serine concentrations,⁶⁷ but sometimes even exceed the concentrations of the L-enantiomer.⁶⁸ Free D-serine, being localized predominantly to the forebrain, shows highest levels in the cerebral cortex, hippocampus, and striatum, followed by the limbic forebrain, diencephalon, and midbrain and is present at low levels in the pons, medulla, cerebellum, and spinal cord.^68–72 The exceptionally high levels of free endogenous D-serine are found in the brain of mammals throughout their life, whereas peripheral tissues, blood serum, saliva, urine, and retina contain low or trace levels of this D-amino acid.⁷³ Concentrations of free D-serine in human serum were suggested to correlate with damaged renal function.⁷⁴ In the brain, D-serine is selectively localized to a subtype of glial cells known as protoplasmic type II astrocytes that enclose nerve terminals and are particularly enriched in the gray matter of the cerebral cortex, hippocampus, anterior olfactory nucleus, olfactory tubercle, and amygdala.^75-77

Since one of the major pathways for the acquiring of D-serine is the serine racemase (SR)-catalyzed biosynthesis from L-serine, D-serine concentrations are highly correlated with the L-serine and glycine concentrations.^66,78 The concentration of free D-serine in the brain depends on the activity of 2 enzymes: CR, which is responsible for both the synthesis and degradation of D-serine and D-amino acid oxidase (DAO) catalyzing D-serine degradation.⁷⁹ Distribution of free D-serine in the brain correlates with the regional distribution of the N-methyl-D-aspartate receptors (NMDArs), suggesting their functional relationship.⁶⁶ NMDArs are ionotropic receptor channels known to play an important role in glutamatergic synaptic transmission. They require simultaneous binding of glutamate molecule and co-agonist glycine at 2 different sites for activation. D-serine is known to act as a selective and potent agonist for the “glycine site” of NMDArs.⁶⁶ Therefore, the NMDAr activity can be regulated by endogenous D-serine. Binding of glycine or D-serine to the “glycine site” of the NMDAr allows glutamate to open the ion channel, but also it decreases desensitization of the receptor.⁶⁶ The biological importance D-serine is further supported by the fact that the enhanced stimulation of NMDArs was shown to be related to a number of neuropsychiatric disorders, such as schizophrenia and bipolar disorder, and to several acute and chronic neurodegenerative diseases, such as chronic pain, polyneuropathies, amyotrophic lateral sclerosis (ALS), stroke, epilepsy, Huntington's disease (HD), Alzheimer disease (AD), and Parkinson disease (PD).^66,80

Serine in structures of ordered proteins

Serine is a short-chain, hydroxyl group-containing residue. High chemical reactivity of this residue defines its versatile roles as active site residue, as well as an important site of several posttranslational modifications (e.g., phosphorylation and O-glycosylation). The hydroxyl group of serine can serve both as a donor or an acceptor of a hydrogen bond, and this residue is commonly involved in hydrogen bond formation with solvent or backbone groups.⁸¹ Curiously, due to its short side chain and the ability to form a hydrogen bond with the protein backbone, serine is effectively mimicking proline. Therefore, serine can be found both on the protein surface and within the interior of a protein. Also, based on the analysis of the structural plasticity (or conformational flexibility) of various amino acid residues defined as the number of conformations exhibited by each residue in a set of known protein structures, it has been concluded that in addition to glycine and alanine, the list of highly flexible residues includes serine, threonine, aspartate, and lysine. Such high conformational plasticity of serine, threonine, aspartate, and lysine was explained by the involvement of these residues in strong side-chain interactions with other residues inside proteins.^82,83

Hydrogen bonds between donor and acceptor groups in the protein backbone are the major determinants of the protein secondary structure. However, due to their spatial orientation relative to the backbone and the ability to affect backbone motions, different amino acids can favor or disfavor the formation of different secondary structure elements, such as α-helices, β-sheets, β-turns, and loops. Among the residues known to favor α-helices are alanine, cysteine, leucine, methionine, glutamic acid, glutamine, histidine, and lysine; β-structure is favored by valine, isoleucine, phenylalanine, tyrosine, tryptophan, and threonine; whereas β-turns are the place for the abundant presence of serine, glycine, proline, asparagine, and uncharged aspartic acid. Importantly, serine is one of the most unfavorable residues in the middle parts of α-helices due to its ability of making a second shared hydrogen bond to the peptide >C=O group of the residue n-3, thereby causing noticeable destabilization of an α-helix.⁸⁴

One should keep in mind however, that although being positioned at the middle of an α-helix serine serve as helical destabilize, this residue might favorably contribute to the α-helix stability when found at the helical ends. This is due to the fact that α-helices found in peptides and proteins are not homogeneous entities, since although all the main chain hydrogen bonding groups within the helix body are satisfied by the secondary structure formation, each end of the helix has 4 unsatisfied groups that are either hydrogen bonded to solvent or to the side chain of the so-called cap residue.⁸⁴ In other words, these first >N‒H groups and last 4 >C=O groups of an α-helix differ from the remaining residues in this helix by being unable to make intra-helical hydrogen bonds and therefore are involved in hydrogen bonding with alternative partners.^85-87 Physico-chemical and statistical analysis suggested that certain residues are more preferable at the C- and N-termini of an α-helix, serving as the helical C- and N-caps.⁸⁶ A global analysis of proteins with known structure revealed the presence of at least 7 distinct capping motifs. Three of these motifs were found at the helix N-terminus and 4 – at the C-terminus.⁸⁷ One of the most common N-capping motifs is the Ser-X-X-Glu.^87–89 Thermodynamic analysis of this Ser-X-X-Glu motif from the GCN4 leucine zipper dimer revealed that helix capping might play a significant role in protein folding since the free energy of helix stabilization associated with the hydrogen-bonding and hydrophobic interactions in this capping structure is −1.2 kcal/mol.⁸⁸ Since the hydroxyl group of serine can serve as an acceptor of hydrogen bond from the free NH-groups in the first turn of an α-helix,⁸⁴ this residue is commonly found in N-caps of α-helices, where serine side chain adopts the gauche^– rotamer and forms a hydrogen bond to the >N‒H group of the residue N3 in the N to C direction.⁹⁰ Based on their N-cap reference, the amino acids were divided into two groups, the good N-cap residues (Asn, Asp, Ser, Thr, and Cys) and the bad N-cap residues (Gly, Ala, Val, Leu, Ile, Phe, Tyr, Trp, Pro, Cystine, Met, Lys, Arg, His, Glu, and Gln).⁹⁰ It was pointed out that the good N-cap residues Ser, Thr, Asp, and Asn define the presence of strong structural constraints at a helix N-cap are greater than structural constraints found at any position yet examined in a protein.⁹⁰ These good N-cap residues are characterized by a restricted range of allowed ψ values when at the N-cap.⁹⁰

Curiously, serine was found in some C-capping motifs too.⁹¹ It was pointed out that the short side chain of serine does not contain enough non-polar atoms to be engaged in strong hydrophobic interactions and these relatively weak hydrophobic interactions are frequently compensated for by a local hydrogen bond of this side-chain back onto the helix (typically with one of the 4 nearest neighboring residues along the sequence).⁹¹ Finally, a systematic analysis of α-helical propensity of a series of dodecapeptides containing alanine, asparagine, aspartate, glutamine, glutamate, and serine at the N-terminus and arginine, lysine, and alanine at the C-terminus, it was concluded that the α-helix-stabilizing abilities of these residues can be ranged as follows: aspartate > asparagine > serine > glutamate > glutamine > alanine at the N-terminus and arginine > lysine > alanine at the C-terminus.⁹²

As it was already mentioned, serine is a perfect residue for tight turns, where it can make hydrogen bonds to the neighboring backbone >N‒H and >C=O groups.⁸¹ However, the most common location of serine in ordered proteins is within the solvent-exposed parts of turns and loops, where this residue interacts with solvent.⁸¹ Also, serine is abundantly found within the disordered regions of ordered proteins primarily due to its low hydrophobicity, low restrictions in backbone conformations, and the ability to readily interact with solvent.⁸¹

It was emphasized that when located in the active site of several enzymes (see below), catalytic serine often serves as a nucleophile.⁹³ Structurally, this nucleophile is always located in a very sharp turn at the tip of the strand-turn-helix motif, called the “nucleophile elbow” or “serine elbow.” This nucleophile elbow is positioned so that it can be easily approached by the substrate, as well as by the hydrolytic water molecule.⁹³ The nucleophile elbow is identified by the consensus sequence Sm-X-Nu-X-Sm (Sm = small residue, X = any residue and Nu = nucleophile). The important consequences of the tightness of this functional strand-turn-helix motif are the presence of noticeable steric restrictions on residues located in the proximity of the nucleophilic amino acid residue and the need for the nucleophilic amino acid residue to adopt energetically unfavorable main chain torsion angles.⁹³ As a result, in Ramachandran plot, such nucleophiles are located within the area with unfavorable angles.⁹⁴

Finally, being a polar residue, serine is expected to positively contribute to protein solubility. This hypothesis was validated by the analysis of the solubility-changing substitutions in proteins, which revealed that serine, together with glutamic acid and aspartic acid, contributes more favorably to protein solubility than other hydrophilic residues (asparagine, glutamine, threonine, lysine, and arginine).⁹⁵

Serine and functions of ordered proteins

Serine in the active sites of serine hydrolases

Serine hydrolases constitute one of the largest known enzyme family with a very broad range of biological activities.⁹⁶ In the human proteome the serine hydrolase superfamily comprises almost 200 enzymes. These proteins are present in all organisms, being active in different cellular compartments. The class of serine hydrolases includes serine proteases or serine endopeptidases involved in various processes including the coagulation cascade.⁹⁷ There are also amidases responsible for the metabolism of endogenous signaling molecules,⁹⁸ penicillin-binding proteins responsible for antibiotic sensitivity,⁹⁹ and carboxylesterases related to the hydrolytic biotransformation of a vast number of structurally diverse drugs.¹⁰⁰ Some other serine hydrolases are intracellular and extracellular lipases catalyzing the hydrolysis of lipids,¹⁰¹ thioesterases including fatty acid synthase¹⁰² and acyl-CoA thiesterase,¹⁰³ cholinesterases hydrolyzing the neurotransmitter acetylcholine into choline and acetic acid,¹⁰⁴ and phospholipases catalyzing hydrolysis of phospholipids into fatty acids and other lipophilic entities.¹⁰⁵ This list would be incomplete without mentioning of protein and glycan hydrolases, such as protein phosphatase methylesterase 1,¹⁰⁶ acyloxyacyl hydrolase,¹⁰⁷ and sialic acid acetylesterase,¹⁰⁸ and several peptidases, such as prolylendopeptidase,¹⁰⁹ dipeptidyl peptidase 4,¹¹⁰ and fibroblast activation protein.¹¹¹

Since serine hydrolases are many, and since they have an enormous variety of functions and biological roles, being able to handle a wide array of structurally diverse substrates, the detailed analysis of structures and functions of these important proteins is definitely outside the scopes of this review. However, all of them, despite their immense structural and functional variability, have one very important feature in common, a feature that serves as a defining hallmark of this protein superfamily. This common trait is the presence in their active sites of a nucleophilic serine that is used for the hydrolysis of various substrates. The very peculiar feature of this nucleophilic serine is that it is typically activated by a proton relay involving an acidic residue (glutamic or aspartic acid) and a basic residue (typically histidine). The corresponding functional motifs are known as catalytic triads. Serine-containing catalytic triads can be grouped into Ser/His/Asp, Ser/His/His, Ser/His/Glu, Ser/Glu/Asp, Ser/N_term, and Ser/cisSer/Lys.^112-114 Besides the aforementioned triads, active sites of some serine hydrolases contain catalytic dyads, such as Ser/Lys and Ser/His.^113-115

Serine and structure of IDPs/IDPRs

Although catalytic functions of many ordered proteins/domains clearly require the presence of serine residues in their active sites, and also serine residues provide crucial means for the specific regulation of these proteins/domains via various posttranslational modifications, such functionally/regulationally important serine residues, being strategically placed within the corresponding protein structures, are not too many. From the structural viewpoint, serine is not included to the list of residues favoring β-structure (valine, isoleucine, phenylalanine, tyrosine, tryptophan, and threonine) or α-helices (alanine, cysteine, leucine, methionine, glutamic acid, glutamine, histidine, and lysine), being, actually, one of the most unfavorable residues in the middle parts of α-helices. The two logical places for this residue to be needed for structure of ordered proteins are tight turns (which are indeed favored by serine, glycine, proline, asparagine, and uncharged aspartic acid) and helical caps. On the other hand, it is unlikely that protein (or protein region) containing excessive amount of serines would be able to spontaneously fold.

In agreement with this hypothesis, computational analysis of 46 extracellular proteins involved in complex carbohydrate degradation from Microbulbifer degradans that contain polyserine linkers (PSLs) revealed that these PSLs were not predicted to have a regular secondary structure being instead extended ‘loopy’ regions.¹¹⁶ This situation is further complicated by the ability of serines to be variously modified by numerous PTMs. For examples, the aforementioned heavily glycosylated Ser/Thr(/Pro)-rich domains serving as the protease-resistant spacers in some cell surface glycoproteins were proposed to have a ‘bottle brush’-like structure.¹¹⁷ Furthermore, despite the fact that serine is one of the polar residues polyserine is not water-soluble at chain lengths greater than 20 residues.¹¹⁸

Serine and functions of IDPs/IDPRs

Serine as a part of the PEST motifs

Many proteins involved in signaling pathways that control cell growth, differentiation, stress responses and physiological cell death contain specific degradation signals known as PEST motifs, since the sequences are enriched in proline (P), glutamic acid (E), serine (S), and threonine (T).^119-122 These PEST motifs are crucial for controlled degradation of regulatory proteins, since these motifs define cellular instability of proteins carrying them by directing these proteins either to the ubiquitin-proteasome degradation or to the calpain cleavage.^121,122

As typical for important regulatory regions in proteins, PEST-containing sequences are solvent exposed, conformationally flexible, and not visible in X-ray structures.¹¹⁹ In agreement with these observations, comprehensive computational analysis revealed that these PEST motifs, being disordered by themselves, are most frequently found within the IDPRs.¹²¹

Serine in entropic bristle domains

Some highly mobile protein regions were assigned the entropic chain activities that can be attributed to their entropic bristle domains (EBDs). Contrarily to the structurally stable domains of ordered proteins, the intrinsically disordered EBD is defined by a time-averaged occupancy of space by a polypeptide chain and therefore represents a time-average 3D region defined by the thermally driven motion of a certain protein region.¹²³ Such EBDs can exclude lager molecules while allowing small molecules and water to move freely through it.¹²³

Obviously, to serve as an EBD, a given fragment of a protein should be highly disordered and have specific amino acid composition. One of the illustrative examples of biologically active EBDs is given by the side-arms of neurofilament (NF) proteins.¹²⁴ In fact, analysis of the amino acid sequence of the porcine NF medium polypeptide revealed that this protein has several peculiar features,¹²⁵ including the presence of a highly disordered C-terminal tail (∼500 residues) with unique amino acid composition characterized by a high content of lysines (17.6%), glutamic acids (26.8%), valines (8.6%), and serines (7.5%).¹²⁵ Although disordered state of this EBD is supported by the high level of charged residues, the abundance of serines should be noted too. In human NF-M, there are 89 serines (9.7%), 39 of which are located within the C-terminal tail (the last 504 residues). Similarly, human NF-H (a polypeptide comprising 1,026 residues) has 88 serines, 57 of which are found within the 613 residues-long C-terminal tail of this protein. In human NF-L that has 543 residues, there are 57 serine residues, 16 of which are located within the C-terminal tails of this protein (the last 147 residues). In addition to neurofilament polypeptides, EBDs were found in microtubule-associated protein 2 (MAP2)¹²⁶ and NuMa.¹²⁷ Analysis of the amino acid compositions of these proteins revealed that they follow the trend established by NFs and contain significant amount of serine residues (186 out of 1,827 residues in human MAP2 are serines, and there are 162 serines in the 2,115 residues-long human NuMa). The high abundance of serine residues in EBDs is not too surprising since serine is not only one of the hydrophilic residues, but is also characterized by a very high conformational variability.⁸²

In biotechnology, EBDs were proposed to be used as protein solubility enhancers, and the translational fusion of a wide variety of natural and artificial EBDs represents an effective solubilizing tool.¹²⁸ Among the more successful solubilizers were artificial EBDs containing the most disorder-promoting residues (Glu, Pro, Gln, and Ser) in the proportion Glu:Pro:Gln:Ser = 2:2:1:1; i.e., sequences containing >16% serine residues, as well as EBDs with a Asp:Glu:Pro:Gln:Ser:Gly = 1:2:2:1:2:1 composition (i.e., EBDs containing >22% serine residues), where a larger subset of disorder-promoting residues was added to avoid potential issues with expression problems associated with high levels of sequence redundancy.¹²⁸

The aforementioned intrinsically disordered solubilizers, entropic bristles, are a member of a family of protein tags with a wide spectrum of biotechnological applications. This important subject was covered in a recent comprehensive review published in this journal, where a detailed description of various protein tags, both ordered and intrinsically disordered, was provided.¹²⁹ The authors emphasized that the disorder content in these tags can range from 6 to 100%, and that less structure often means more function. They also showed that the protein tags with the most disorder can be used not only for the increase in the solubility of target proteins as the aforementioned entropic bristles do, but also can serve as important tools for extending half-life of proteins and for characterizing their biological properties, e.g., binding.¹²⁹ An incomplete list of the disordered serine-containing tags shown to increase the half-life of a protein of interest inside the cell includes an unstructured recombinant polypeptide of 864 amino acids with the PESTAG composition, called XTEN,^130,131 serine-glycine-containing HAP¹³² and proline-alanine-serine-containing PAS tags,¹³³ as well as tags containing PESAK, PASTDH, and PASTD repeats.^134,135

Some functions of serine-rich proteins

To get a clue on what proteins with high serine content can be used for biologically, functions of several proteins with polyserine regions are outlined below. Although for a long time, polyserine sequences have been ignored as the non-functional result of transcription slippage, because these and other homopolymer sequences are typically encoded by trinucleotide repeats at the DNA level,⁴⁹ data overviewed below clearly shows that polyserine regions not only serve as flexible linkers, but have a rather diverse set of functions. Considered proteins either possess overall high serine content or have functionally important polyserine tracks (PSTs), polyserine linkers (PSLs), or other serine-rich regions. Notes below are arranged not by protein function or disorder contents, but according to the serine content in the corresponding proteins.

Polyserine linkers (PSLs) in bacterial proteins

Although domain likers composed predominantly of serine are rare, the analysis of Microbulbifer degradans proteome revealed the presence of 46 extracellular proteins containing polyserine linkers (PSLs) separating their carbohydrate-binding or catalytic domains from other binding domains.¹¹⁶ The common feature of these 46 M. degradans proteins and 65 PSL-containing bacterial proteins found in sequence depositories is their involvement in complex carbohydrate degradation. Based on these observations, it has been concluded that flexible, disordered nature of PSLs in these proteins enhances their functionality expanding the potential substrate target area available to the enzyme after a carbohydrate-binding module makes contact with a polymer or enhancing the substrate availability to an enzyme anchored to a bacterial outer membrane.¹¹⁶

Abundance and functions of proteins with long PSTs in human proteome

The search for human proteins containing PSTs of 10 residues or longer revealed 59 such proteins, containing a total of 66 serine homopolymers.⁴⁹ Functional analysis of these proteins reveled that many of them are involved signaling, transcription regulation, DNA binding, or protein–protein interactions.⁴⁹

Long serine homopolymer of Dictyostelid amoebae

The single-celled eukaryote Dictyostelium discoideum was recently shown to contain one of the longest serine-based homopolymers consisting of 306 tandem serine repeats.¹³⁶ The authors emphasized that this gene has a paralog with 132 serine repeats and several orthologs with high serine repeat numbers, in various other Dictyostelid species, suggesting that this very long polyserine module is functional.¹³⁶

Fibroins

Fibroins are insoluble structural proteins present in silk created by spiders and the larvae of many insects. Structurally, fibroin represents a supramolecular ensemble organized as layers of antiparallel β-sheets, tight packing of which defines the known rigid structure and tensile strength of silk. The major component of the fibroin is the recurrent amino acid sequence (Gly-Ser-Gly-Ala-Gly-Ala)_n. Although this amino acid sequence feature of fibroins is rather conserved, some noticeable exceptions are known. Spiders can produce up to 7 different types of the protein-based silks/glues that have diverse physical properties.¹³⁷ One of these 7 silks/glues is the egg case fibroin, tubuliform spidroin 1 (TuSp1). Analysis of the TuSp1 from the black widow spider, Latrodectus hesperus, revealed that this protein is composed of highly homogeneous repeats of 184 amino acids.¹³⁷ This protein has multiple polyserine blocks and short polyalanine stretches. The overall serine content in the L. hesperus TuSp1 is very high 25.9%, and this high value is characteristic for TuSp1s from other organisms.

Profilaggrin

Human profilaggrin is a 4061 residue-long protein which is predicted to be almost 100% disordered. Filaggrin (FLG) and its precursor proFLG have a highly repetitive structure and a unique amino acid composition. For example, proFLG is characterized by very high serine content (24.1%). Among major biological functions of proFLG are its role as a precursor of FLG, a precursor of the natural moisturizing factor (NMF), a protein aggregating keratin intermediate filaments and promoting disulfide-bond formation among the intermediate filaments during terminal differentiation of mammalian epidermis, as well as multiple functions in skin homeostasis and barrier formation.¹³⁸ ProFLG becomes heavily phosphorylated immediately after translation in the granular layer of epidermis and oral mucosa.¹³⁹ This mature phosphorylated form of proFLG is insoluble and is stored in specific membrane-less keratohyalin granules.¹³⁸ Therefore it is likely that the overly abundant serine residues in intrinsically disordered proFLG are used as easily accessible phosphorylation sites.

SRm160/300 splicing coactivator

The SRm160/300 splicing coactivator is important for the processing of a subset of constitutively spliced pre-mRNAs where it plays crucial role by promoting critical interactions between splicing factors bound to pre-mRNA, including snRNPs and SR family proteins.¹⁴⁰ Being a component of the active spliceosome, SRm160/300 splicing coactivator is found in a pre-mRNA splicing complex with SFRS4, SFRS5, SNRP70, SNRPA1, SRRM1 and SRRM2. Sequence analysis revealed that similar to SRm160 (which has a serine content of 16.7%), the SRm300 (with very high serine content of 23.3%) contains domains rich in alternating serine and arginine residues, several unique repeated motifs rich in serine, arginine, and proline residues, and 2 very long (>30 consecutive residues) polyserine tracts.¹⁴⁰

Mucin-16 and other human mucins

Mucin-16 or ovarian cancer-related tumor marker CA125 is a member of mucin family of proteins providing protective and lubricating barrier against particles and infectious agents at mucosal surfaces. Mucin-16 is a large protein (22,152 residues) with a strong bias toward serine (14.8%) and threonine (15.9%) residues. It has 3 serine-rich domains (residues 1638–3055, 3961–4385, and 7085–10,395) that contain 21.4%, 22.4% and 22.1% serines. It is predicted to be mostly disordered (with the amount of predicted disordered residues exceeding 75%). This protein is heavily glycosylated. It was already pointed out that mucins contain ∼80% carbohydrate and are characterized by the presence of numerous serine-based O-glycosylation sites.¹⁴¹ Therefore, even though not all mucins have officially defined serine-rich domains, it was not too surprising to find that the overall serine content in these proteins is high, systematically exceeding 11%, with some mucins having significantly large serine proportions (e.g., human mucin-12 (22.7%), mucin-17 (21.6%), mucin-3A (19.5%), mucin-22 (18.5%), etc.).

Gp40 protein of Cryptosporidium parvum

Cryptosporidium parvum is an intestinal apicomplexan parasite causing cryptosporidiosis, the diarrheal disease with the worldwide distribution.¹⁴² One of the proteins mediating sporozoite attachment and invasion of host cells is the 40 KDa glycoprotein (gp40), which is present in oocysts and sporozoites and is known to be shed from the parasite during invasion.¹⁴³ The total serine content of gp40 is of 19.2%. Furthermore, it has been shown that the N-terminal region of gp40 (which is predicted to be highly disordered) contains a polyserine domain consisting of 19 contiguous serine residues that modulates the ability of the parasite to attach to and invade the host cells.¹⁴³

Multifunctional cellulase CelAB

The endosymbiont Teredinibacter turnerae T7902^T encodes a multifunctional cellulase CelAB, which binds cellulose and chitin, degrades multiple complex polysaccharides, and displays 2 catalytic activities, cellobiohydrolase and β-1,4(3) endoglucanase.¹⁴⁴ This polyfunctionality of CelAB is determined by the presence of 2 catalytic and 2 carbohydrate-binding domains.¹⁴⁴ Each domain of CelAB is separated by polyserine linker regions of 38, 55, and 54 residues in length, containing 32, 40, and 43 serine residues, respectively.¹⁴⁴ Furthermore, the overal serine content of this protein is of 18.4%. It is possible that these polyserine linkers act similarly to the aforementioned PSLs in serve the M. degradans proteins,¹¹⁶ expanding the potential substrate target area available to the catalytic domains of CelAB after carbohydrate-binding domains binding to cellulose or chitin.

Human positive cofactor 4

A positive cofactor 4 (PC4) (which is also known as coactivator p15) is a member of the upstream stimulatory activity (USA)-derived cofactors that act complementary to the transcription factor IID (TFIID) machinery and are needed for the activator-dependent transcription in mammalian cells.¹⁴⁵ PC4 is a 127 residue-long protein with serine-rich regions near the N-terminus,¹⁴⁵ and the overall serine content of this protein is high (16.5%). PC4/p15 is able to bind both single-stranded and double-stranded DNA, and this DNA-binding activity is inhibited by the casein kinase II-catalyzed phosphorylation.¹⁴⁶ Besides DNA binding, this protein is involved in interaction with the components of the TFIID complex.¹⁴⁶ Using nuclear magnetic resonance (NMR) analysis of the full-length PC4 and its N- and C-terminal domains it has been concluded that although the C-terminal core domain (PC4ctd) is structured, the PC4 N-terminal domain (PC4ntd) is highly flexible and mostly unstructured.¹⁴⁷ Furthermore, inter- and intramolecular interactions were shown to promote some local conformational changes in both the structured and unstructured subdomains.¹⁴⁷

Human bHLHb5

Basic helix-loop-helix protein 5 (bHLHb5) is a member of the bHLHb family of transcription factors that are implicated in many aspects of neural development, including cell growth, differentiation, and cell migration.¹⁴⁸ Being a transcription factor, bHLHb5 is predicted to be highly disordered and has high serine content (15.2%). Although human protein has a profound serine-rich region (residues 208–238) that incorporate a polyserine stretch (residues 226–234), only serine was found at the corresponding position in the mouse Bhlhb5 and hamster BETA3.¹⁴⁸

Hox-D13

The HOX proteins constitute a highly conserved family of transcription factors encoded by HOX genes. Although these proteins are key regulators of embryonic development, they continue to be expressed throughout postnatal life.¹⁴⁹ In human and mice, there are 39 human HOX genes which are located in 4 clusters (A-D) on different chromosomes and are assumed to have arisen from a primordial homeobox gene by duplication and divergence.¹⁴⁹ In the genital tubercle (GT), HOXA13 and HOXD13 are the most highly expressed HOX genes.¹⁵⁰ Mutations in HOXA13 and HOXD13 genes are known to cause syndromes affecting genitourinary development and defects of limb formation, such as synpolydactyly and hand-foot-genital syndrome.¹⁴⁹ The products of the HOXA13 and HOXD13 genes, the Hox-A13 and Hox-D13 proteins, are similarly distributed in the GT suggesting that they may function in a redundant manner.¹⁵⁰ As the majority of other transcription factors,^151–153 both Hox-A13 and Hox-D13 are predicted to be highly disordered. The characteristic feature of the Hox-D13 proteins from different species ranging from amphibians (e.g., red spotted newt Notophthalamus viridescens) to marsupials (e.g., opossum and tammar), to mammals (e.g., human and mice) is the presence of high content of serine residues (>14%), with these residues being assembled into one or 2 polyserine regions.¹⁵⁴

SRrp37

Analysis of proteins interacting with pNO40, a ribosomal 60S core subunit, by yeast 2-hybrid identified a novel SR-related protein, referred to as SRrp37.¹⁵⁵ This protein serves as a splicing regulator located in the nuclear speckles and nucleoli¹⁵⁵ and is predicted to be highly disordered. SRrp37 is involved in modulation of alternative pre-mRNA splicing with either 5' distal site activation or preferential use of 3' proximal site, and in case of infection by Herpes simplex virus (HSVI), this protein may act as a splicing inhibitor of HSVI pre-mRNA.¹⁵⁵ The characteristic features of SRrp37 are high serine (13.5%) and arginine contents (12.8%) and the presence of 3 motifs, a serine-arginine (SR) dipeptide enriched region (190–220), a polyserine stretch (residues 258–282), and a potential nucleolar localization signal comprising a long array of basic amino acids (residues 286–312).¹⁵⁵

Scaffolding protein piccolo

Piccolo is a 5065 residue-long scaffold protein, which is predicted to be highly disordered (it contains ∼85% disordered residues). High level of disorder is a characteristic feature of scaffold proteins that helps them to organize and maintain stable and dynamic protein complexes.¹⁵⁶ Piccolo contains 12.4% serine residues and has 2 serine-rich regions (residues 4215–4278 and 4699–4765). The major function of piccolo is to serve as a scaffolding protein responsible for the organization of synaptic active zones, where it interacts with the cytomatrix at the active zone (CAZ)-associated structural protein (CAST) and other CAZ proteins, such as RIMs, Munc13s, Bassoon, and participates in synaptic vesicle trafficking.^157,158

Adenomatous polyposis coli (APC) and its mutation cluster region (APC-MCR)

APC (adenomatous polyposis coli) is another large scaffold protein (2843 residues) which is massively enriched in serine residues that account for 15.3% of the entire sequence. APC serves as one of the central hubs involved in the spatiotemporal regulation of key Wnt signaling components.^159,160 Wnt pathway involves 5 major proteins, Axin, CKI-α, GSK-3β, APC, and β-catenin, all of which were shown to contain several intrinsically disordered regions, which facilitate protein-protein interactions, post-translational modifications, and signaling.¹⁶¹ APC, which is predicted to be massively disordered (in fact, there are no globular domains in the 2000 amino acid long C-terminal region of APC¹⁶²), is known to act as flexible concentrator that brings together all other proteins involved in the Wnt-pathway.¹⁶¹ Furthermore, intrinsically disordered APC helps the collection of β-catenin from cytoplasm, facilitates the β-catenin delivery to the binding sites on Axin, and controls the final detachment of β-catenin from Axin.¹⁶¹ APC is known to be mutated in around 80% of somatic colorectal cancer patients.¹⁶³ The most frequently mutated regions of APC is known as mutation cluster region (APC-MCR, residues 1200–1550), which contains 19.1% serine residues and was shown to be highly disordered by several biophysical techniques.¹⁶⁴

WNK1 Kinase

WNK [with no lysine (K)] protein kinases influence ion balance and are present in all multicellular and many unicellular organisms.¹⁶⁵ Besides a protein kinase domain WNKs have no other currently known folded domains, but contain an autoinhibitory domain suppressing protein kinase activity, several 2 coiled-coil regions, and many protein-interaction motifs.¹⁶⁵ An illustrative member of this interesting kinase family is a serine-rich WNK1 kinase, that contains 13% serines which are preferentially found within the 2000 residues-long, intrinsically disordered, serine-rich region. Although in resting cells, the majority of WNK1 is localized on cytoplasmic puncta, during cell division, this protein localizes to mitotic spindles.¹⁶⁵

Lysyl-5 hydroxylase Jmjd6

In vertebrates, the hydroxylation of lysine residues in proteins involved in pre-mRNA splicing is catalyzed by an Fe(II)- and 2OG (2-oxoglutarate)-dependent oxygenase, jumonji-domain-containing protein 6 (Jmjd6),^166–168 which is highly conserved throughout animal evolution and plays an important role in embryonic development.¹⁶⁹ Loss of this protein function is associated with severe developmental defects and embryonic lethality.¹⁷⁰ Jmjd6 has a total serine content of 11.2% and possesses several functional domains and conserved functional motifs, such as catalytic JmjC domain (residues Pro¹⁴¹–Arg³⁰⁵), 5 nuclear localization sequences (NLSs, residues 6–10, 91–95, 141–145, 167–170, and 373–378), an AT-hook motif (residues Lys³⁰⁰–Ser³⁰⁹), a SUMOylation site (Leu³¹⁶–Glu³¹⁹),^171,172 and a polyserine region comprising 16 serine residues interrupted by 4 aspartate residues (Ser³⁴⁰–Ser³⁵⁹).¹⁶⁹ Although the crystal structure for the human Jmjd6 has been resolved, only residues 1 to 334 of this protein were well defined, whereas the C-terminal serine-rich region (residues 335 to 403) was shown to be completely disordered.¹⁷³ Several alternatively spliced variants of Jmjd6 were reported that lack the polyserine region,¹⁷² and functional analysis revealed that the alternatively spliced Jmjd6 without the polyserine domain was localized to the nucleolus, where it was involved in interaction with nucleolar proteins.¹⁶⁹ These observations suggested that the Jmjd6 polyserine region might be related to the control of the nuclear/ nucleolar shuttling of this important protein.¹⁶⁹

Polyserine tracts in vitellogenins: Insulated connectors

The female-specific, ancient egg-yolk precursor proteins known as vitellogenins (Vg) are found in most oviparous species.¹⁷⁴ Although the major functions of Vgs are related to the lipid and ion transport during oogenesis, the honeybee (Apis mellifera) protein (AmVg) is also known to affect social behavior and life-span plasticity of this insect.¹⁷⁵ Although the overall serine content of insect Vgs does not exceed that of an average IDP (∼8.5%), an important feature of insect Vgs is the presence of a repetitive and highly variable segment known as polyserine tract (PST) or polyserine linker (PSL), which connects conserved N-terminal and α-helical domains.¹⁷⁵ The careful structural analysis of several PSTs from various Apocrita species revealed that highly pronounced sequence variation in this region is translated in noticeable structural variations.¹⁷⁵ In fact, structurally, these PSTs ranged from highly extended and disordered conformations (e.g., the polyserine regions of honeybee Apis mellifera, bumblebee Bombus ignitus, and wasp Pimpla nipponica) to a more compact and helical (but still disordered) conformation as in the polyserine region of the jewel wasp Nasonia vitripennis.¹⁷⁵ Furthermore, a casein kinase II (CKII) recognition motif was found in PSTs of insect Vgs and these regions were shown to undergo multiple phosphorylation by the CKII kinase.¹⁷⁵ Since phosphorylated polyserine region was unexpectedly resistant to trypsin/chymotrypsin digestion, it has been concluded that the multi-site phosphorylation might shield the PST against unspecific cleavage.¹⁷⁵

Dehydrin ERD14

The dehydrin family belongs to a class of Late Embryogenesis Abundant (LEA) proteins in plants that includes stress proteins accumulating in response to dehydration-related environmental stresses. Although the dehydrins’ expression in somatic cells is very low under normal conditions, it increases dramatically when dehydration is promoted by water stress, high salinity or cold. Members of the dehydrin family were shown to be intrinsically disordered both computationally¹²⁸ and by experiment.^176,177 One of the well-studied members of this family is the ERD14 protein (early response to dehydration 14) of Arabidopsis thaliana. Comprehensive structural analysis by NMR techniques revealed that although ERD14 is fully disordered under near native conditions, this protein has several short regions of somewhat restricted motion and 5–25% helical propensity.¹⁷⁷ In agreement with its experimentally validated intrinsically disordered nature, amino acid sequence of ERD14 is highly enriched in glutamate (21.1%) and lysine residues (18.4%). Although the overall serine content of EDR14 is not too high (8.4%), it was pointed out that its polyserine domain (residues 76–90) serves as a primary target for the multi-site phosphorylation that activates calcium binding activityof this protein.¹⁷⁸

Titin

Being the largest vertebrate protein, titin (in human - 34,350 residues) represents an indispensable component of the vertebrate striated muscles. Among numerous cellular functions of titin (also known as connectin) are contribution to the fine balance of forces between the 2 halves of the sarcomere which is crucial for the elasticity of muscle cells, as well as participation in chromosome condensation and chromosome segregation during mitosis of non-muscle cells. The ability of titin to reversibly extend relies on a set of PEVK segments, rich in proline (P), glutamate (E), valine (V), and lysine (K) residues. Although the overall fraction of serine residues in human titin is not too high (7.2%), this protein has a serine-rich domain (residues 34102–34244) with the serine content of 30.6%.

The H^±-ATPase fromSaccharomyces cerevisiae

An electrogenic proton pump, the plasma membrane H⁺-ATPase from Saccharomyces cerevisiae, is involved in regulation of the intracellular pH and generation of the electrochemical proton gradient needed for the cell growth and development.¹⁷⁹ Although the overall serine content of yeast H⁺-ATPase is relatively low (6.4%), the N-terminal tail of this protein contains a serine-rich cluster of 11 serine residues in the first 17 amino acids, including a stretch of 8 consecutive serine residues. Mutational and deletion analysis revealed that although this polyserine cluster was not important for the ATP hydrolysis, it definitely played an important role in the overall stability of this protein, since mutants with 8 or more serine residues converted to alanine showed significantly decreased abundance in the plasma membrane under stress conditions caused by the accelerated degradation rates.¹⁷⁹

ICP4 of herpes simplex virus (HSV)

In herpes simplex virus type 1 (HSV-1), which is a member of the Alphaherpesvirinae, ICP4 is a large (175-kDa), homodimeric, DNA-binding phosphoprotein primarily found in the nucleus of the infected cell.^180-182 As many other viral proteins,¹⁸³ ICP4 is characterized by a diverse functionality, being required for the transcriptional activation of most of the essential early (E) and late (L) genes,^184-188 acting as an expression repressor of latency-associated transcript (LAT),^189-191 long/short transcripts (L/ST),^192,193 and even serving as a regulator of own expression.^{182,184,194,195} DNA binding, dimerization, nuclear localization, and transcriptional activation of ICP4 are ascribed to discrete functional domains.^{182,194,196-198} Among those functional ICP4 domains, transcriptional activation is attributed to a large C-terminal domain (residues 775–1298) and to a relatively short serine-rich region near the ICP4 N-terminus (residues 143 to 210).^182,197 The transcription activation of a protein with deleted transactivation C-terminal domain was noticeably altered,^196,199-201 and subsequent deletion of the serine-rich N-terminal domain completely abrogated the ICP4 transactivation function.^197,202,203

The overall serine content of ICP4 is relatively low (5.9%). However, among the important features of the serine-rich N-terminal domain are the presence of the polyserine tract (aa 184 to 198) and consensus sites for the cellular kinases protein kinase A (PKA), protein kinase C (PKC), and casein kinase II (CKII).^204,205 Furthermore, this region is subjected for the in vitro autophosphorylation.¹⁸² Being poorly conserved among the alphaherpesviruses, this region always contain similar stretches of polyserine residues.¹⁸² Although for a long time, the functional significance of the polyserine motif was not known, based on the analysis of the viral growth in tissue culture and in neuronal and non-neuronal cells in vivo revealed that the polyserine tract supports an activity of ICP4 specifically required for growth in the sensory ganglia.¹⁸²

E-cadherin

Calcium-dependent cell adhesion protein E-cadherin is an important player in the epithelial cell-cell adhesion that determines integrity and organization of any tissue, and loss of cell adhesion activates the detachment-induced apoptosis, anoikis.^206,207 Being synthesized as a preproprotein, mature E-cadherin (residues 155–882 in human protein, UniProt ID: P12830) is a single-pass membrane protein with a conserved cytoplasmic domain (residues 731–882) and a divergent extracellular region (residues 155–709) composed of 5 cadherin domains separated by interdomain Ca²⁺ binding sites.²⁰⁸ Although the overall serine content of E-cadherin is not too high (just 4.8%), these residues are not uniformly spread through the sequence, and protein has a short serine-rich region (residues 828–851) located within the disordered C-terminal tail (residues 740–882). This serine-rich region is required for the O-GlcNAcylation of E-cadherin that plays a role in inhibition of the E-cadherin trafficking to the plasma membrane and accelerates anoikis.²⁰⁹ This is because the O-GlcNAcylation leads to the retaining of E-cadherin in the endoplasmic reticulum and blocks the E-cadherin interaction with the type I gamma phosphatidylinositol phosphate kinase (PIPKIγ), a protein required for recruitment of E-cadherin to adhesion sites.²⁰⁹

Epstein-Barr nuclear antigen-1 (EBNA1)

The Epstein-Barr nuclear antigen 1 (EBNA1) protein of Epstein-Barr virus (EBV) is critical for the EBV latent infection due to the involvement of this protein in the replication and segregation of the viral genome in proliferating host cells, and due to its roles in the transactivation of the expression of other EBV latency genes.²¹⁰ The N-terminal 479 residues of the 641 residue-long EBNA1 are predicted to be disordered. This region has a strong compositional bias, being heavily enriched in Ala and Gly residues. This Ala/Gly-rich region (residues 87–325) is followed by the Gly/Arg-rich region (residues 328–380) and by the Ser-rich module (residues 383–393).

Although the overall serine content of EBNA1 is low (4.2 %), serines definitely play crucial roles in function of this protein. Binding of EBNA1 protein to casein kinase 2β-subunit (CK2β) recruits CK2 to promyelocytic leukemia nuclear bodies (PML NBs), leads to the increased phosphorylation of PML proteins, and triggers degradation of these proteins. This EBNA1-induced degradation of PML proteins in multiple carcinoma cell lines results in the loss of the PML NBs that would otherwise promote apoptosis and suppress viral lytic infection.^211-213 Interaction of CK2β with EBNA1 is mediated via a novel CK2β binding pocket, is primed by phosphorylation of EBNA1 on Ser393 located within the polyserine region, and affects the ability of EBNA1 to induce PML degradation.²¹⁴

Serine-based posttranslational modifications of proteins

At the later stages of protein biosynthesis, after their transcription, many proteins undergo intensive chemical alterations, collectively known as protein posttranslational modifications (PTMs), which range from the proteolytic cleavage to the covalent attachment of diverse functional groups or even proteins to specific side chains of a target protein. PTMs change local physical and chemical properties of a polypeptide chain, such as charge, hydrophobicity, and flexibility. These chemical modifications of a polypeptide chain after its biosynthesis extends the range of amino acid structures and properties, and consequently diversifies structures and functions of proteins. Although DNA typically encodes 20 primary amino acids, proteins contain more than 140 different residues, because of various PTMs. Altogether, as many as 400 post-translational modifications of proteins are known to occur physiologically.^215,216

PTMs play a number of fundamental roles in regulating the folding of proteins, their targeting to specific subcellular compartments, their interaction with ligands or other proteins, and their functional state, such as catalytic activity in the case of enzymes or the signaling function of proteins involved in signal transduction pathways. Functions of some proteins rely on several different types of posttranslational modifications.

Although all amino acid side chains are known to undergo chemical diversification due to various PTMs, most often protein PTMs are found at side chains that can act as either strong (Cys, Met, Ser, Thr, Tyr, Lys, His, Arg, Asp, and Glu) or weak (Asn and Gln) nucleophiles, whereas the remaining residues (Pro, Gly, Leu, Ile, Val, Ala, Trp, and Phe) are rarely involved in covalent modifications of their side chains. Different residues are subjected to different types of PTMs. Among PTMs frequently affecting serine are phosphorylation, O-glycosylation, N-acetylation, O-acetylation, phosphopantetheinylation, autocleavage, N-ADP-ribosylation, amidation, N-decanoylation, O-octanoylation, O-palmitoylation, and sulfation. More rare serine-based PTMs found in bacteria are serine conversion to D-alanine and to the thioether amino acid lanthionine. Some of the illustrative examples of these serine-targeting PTMs are considered below.

Phosphorylation

It is estimated that functions of one-third of eukaryotic proteins are controlled via phosphorylation/dephosphorylation cycles that originate from carefully regulated protein kinase and phosphatase activities.²¹⁷ Phosphorylation modulates the activity of numerous proteins involved in signal transduction, and regulates the binding affinity of transcription factors to their coactivators and DNA thereby altering gene expression, cell growth and differentiation.²¹⁸ Phosphorylation sites are frequently located within functionally important protein domains. For example, the majority of phosphorylation sites of Mdm2 are located in its p53- and p14-ARF-binding regions,²¹⁹ whereas ubiquitin-mediated degradation of many proteins is controlled by the phosphorylation of their PEST motifs.¹¹⁹

Eukaryotic protein kinases constitute one of the largest protein families. For example, yeast kinome includes 119 kinases, there are 1019 kinase- and 300 phosphatase-coding genes in Arabidopsis thaliana, there are 540 kinases in mouse kinome, whereas human genome contains ∼520 genes encoding kinases and more than 150 genes encoding phosphatases. Although kinases/phosphatases are very common proteins, in any given eukaryotic proteome, the number of kinases and phosphatases is noticeably smaller than the number of their potential substrates. In fact, on average, each eukaryotic protein kinase serves ∼20 substrates, whereas each human phosphatase is expected to dephosphorylate ∼65 clients.³⁶

Although kinases/phosphatases are known to modify mostly 3 residues (serine, threonine, and tyrosine), it seems that the major phosphorylation target is serine. In fact, according to the PhosphoBase,²²⁰ of the 1700 phosphorylated residues found in 400 proteins, 1135 were serines.²²¹ Curiously, numerous bioinformatics studies clearly linked intrinsic disorder with phosphorylation, since the vast majority of this PTM affect protein sites located within the disordered regions.^221,222 In agreement with these computational studies, there are several well-characterized examples of phosphorylatable IDPs. An illustrative example of such experimentally validated IDPs targeted to phosphorylation is given by the mutation cluster region (MCR) of the signaling hub adenomatous polyposis coli (APC), which was shown to be was susceptible to proteolysis, lacked α-helical secondary structure, did not display thermal unfolding transition, displayed an extended conformation, and was accessible for phosphorylation by CK1ε in vitro.¹⁶⁴

O-glycosylation

More than half of all proteins are glycosylated, and the molecular volume occupied by an oligosaccharide in a glycoprotein is frequently as large as that of the protein domain to which it is attached.¹¹⁷ O-glycosylation is the PTM process of the carbohydrate addition to the hydroxylated amino acids of proteins.²²³ O-glycosylation represents the addition of heterogeneous in size and structure oligosaccharides to a protein.²²³ O-linked glycosylation is normally initiated in the Golgi apparatus, most commonly by a N-acetyl galactosaminyltransferase that transfers a N-acetylgalactosamine (GalNAc) residue to the side chain of a serine or a threonine residue.¹¹⁷ Importantly, this event of the O-glycosylation initiation is dependent on the primary, secondary, and tertiary structure of the glycoprotein.²²⁴ Subsequent events of a stepwise enzymatic elongation by specific transferases yield several core structures, which are further elongated or modified by sialylation, sulfatation, acetylation, fucosylation, and polylactosamine-extension.¹¹⁷

Among the well-studied O-glycosylation targets in animals is a family of mucins, which are the structural components of the mucus gels that protect the respiratory, gastrointestinal, and reproductive tracts.¹⁴¹ Mucins contain ∼80% carbohydrate on a mass basis and have a high intrinsic viscosity due to their large size (250 – 2,000 kDa) and extreme hydrophilicity caused by the presence of numerous and heterogeneous oligosaccharides.¹⁴¹

Not only mucins, but many other cytoplasmic and nuclear proteins can be reversibly glycosylated with an O-linked GlcNAc, and this O-glycosylation might function in a similar manner as phosphorylation.¹¹⁷ Among the O-glycosylated proteins are low-density lipoprotein receptor,²²⁵ the epidermal growth factor (EGF) domains of different proteins, such as the bovine blood coagulation factors VII and IX, human coagulation factors VII and IX, human and bovine protein Z, tissue plasminogen activator, thrombospondin, urokinase, factors VII, IX, and XII, and murine fetal antigen 1,²²⁶ the neurotrophin receptor (nerve growth factor receptor, NGFR), IgA1, heparin- and gelatin-binding horse seminal plasma protein-1 (HSP-1), as well as many other biologically important proteins.¹¹⁷

It has been pointed out that since O-linked oligosaccharide chains are typically clustered over short peptide stretches in Ser/Thr(/Pro)-rich domains of several proteins, they can serve as multivalent carbohydrate antigenic or functional determinants for antibody recognition, mammalian cell adhesion, and microorganism binding.²²⁴ Furthermore, these heavily glycosylated Ser/Thr(/Pro)-rich domains may serve as protease-resistant spacers in cell surface glycoproteins.¹¹⁷ Curiously, sections of proteins with a high proline-content and many sugars attached to serines or threonines may assume a ‘bottle brush’-like structure.¹¹⁷

Phosphopantetheinylation

Phosphopantetheinylation is a crucial posttranslational modification that is absolutely necessary for the action of the polyketide synthases (PKSs) producing a large class of secondary metabolites known as polyketides²²⁷ and multifunctional non-ribosomal peptide synthetases (NRPSs) responsible for the biosynthesis of various peptide-based compounds.²²⁸ For example, NRPSs activate amino acid monomers first as aminoacyl-AMP mixed anhydride species then as covalent amino acyl-thioester intermediates, where the thiol group is provided by a 4′-phosphopantetheinyl group posttranslationally attached to a specific serine by a phosphopantetheinyl transferase (PPTase).²²⁷

Two illustrative examples of NRPSs are surfactin synthetase responsible for biosynthesis of the lipoheptapeptide antibiotic surfactin by Bacillus subtilis²²⁹ and the cyclosporin A producing fungal cyclosporin synthetase, which is a very large single chain enzyme containing more than 15,000 residues and 11 amino acid activation modules, each equipped with the adenylation domains and adjacent consensus serine sites marking condensation, epimerization, N-methylation, and peptidyl carrier protein (PCP) domains.²³⁰

Fugal iterative type I PKSs are complex enzymes responsible for the biosynthesis of various important polyketide compounds, such as statins, melanin pigments, and toxic aflatoxins.²³¹ These type I PKSs are large machines consisting of multiple catalytic domains, such as starter acyltransferase (SAT), ketosynthase (KS), acyltransferase (AT), product template (PT), tandem acyl carrier proteins (ACPs) and Claisen cyclase (CLC) domains.^231,232 On the contrary, type III PKSs (which, in plants, are responsible for biosynthesis of flavonoids and related compounds) are simple homodimers of ∼42 kDa.²³³

Autocleavage: Inteins and autocatalytic enzyme-like proteins

Inteins

Inteins are self-splicing elements that catalyze their own removal from the host protein and the formation of a peptide bond between their flanking protein regions (exteins) through a posttranslational process of protein splicing.^234-237 There are 4 coupled nucleophilic displacements in the protein splicing mechanism: 1) an N→O(S) acyl shift of Ser/Cys at the intein N-terminus; 2) a transesterification reaction to form a branched intermediate with 2 N termini; 3) cyclization of the intein C-terminal Asn/Gln to release the intein; and 4) an O(S)→N acyl shift of Ser/Thr/Cys to form a native peptide bond between the exteins.²³⁸ As it follows from this description, nucleophilic nature of serine place this residue at the center of the series of transformations leading to protein self-splicing.²³⁹ In fact, the splice junctions of all inteins are closely related, with 3 types of amino acids being required at precise positions at the junctions: the last residue of the intein must be asparagine, and a hydroxyl or thiol residue (serine, threonine, cysteine) must be at the C-terminal side of each of the 2 junctions.²³⁹

One of the first reported examples of inteins is the generation of 2 enzymes via protein splicing from a single Saccharomyces cerevisiae VMA1 gene.²³⁶ Here, the self-splicing of the nascent 120-kDa translational product of VMA1 excises out the 50-kDa VMA1-derived endonuclease (VDE or VMA1 intein) and splices the N- and C-terminal exteins to form the mature Vma1p (a catalytic 70-kDa subunit of the vacuolar H⁺-ATPase).²³⁶

Autocatalytic enzyme-like proteins

Inteins can be considered as members of a family of autocatalytic enzyme-like proteins that perform chemical reactions using the same strategies as enzymes, but do not act on multiple substrates.²³⁸ Other examples of autoprocessing include RecA-assisted autocleavage of LexA or lambda repressor, autocleavage of N-terminal nucleophile amidohydrolase precursors (glycosylasparaginase, penicillin acylase, proteasome, etc.), processing of Hedgehog embryonic signaling proteins, activation of plasma proteins, and formation of pyruvoyl enzymes.^240,241 The common feature of these processes, including protein splicing, is that all of them are initiated by a nucleophilic attack by the activated side-chain thiol of cysteine or hydroxyl group of serine or threonine at an amide bond.²³⁸

N-acetylation

N^α-terminal acetylation (Nt-acetylation) is a cotranslational or posttranslational modification found in many eukaryotic proteins by which the acetyl group is added to the α-amino group of their N-terminal residues. In fact, a majority of eukaryotic proteins, for example, more than 85% of human proteins and 68% of yeast proteins are cotranslationally Nt-acetylated.²⁴² Several archaeal and prokaryotic proteins are also modified by N-terminal acetylation.

Since nascent eukaryotic proteins contain N-terminal Met, which, being retained in the mature protein and being followed by the “acetylation-permissive” residues, is usually acetylated at its α-amino group.^243,244 If the residue at position 2 of a nascent protein has a small enough side chain, then the N-terminal Met is cleaved off by Met-aminopeptidases, resulting in N-terminal Ala, Val, Ser, Thr, Cys, Gly, or Pro.²⁴⁵ Similarly to the N-terminal Met, these N-terminal residues (with the exception of Pro) are often N^α-terminally acetylated.^246–248 This is because the major N-terminal acetyltransferase (NAT), NatA, is able to act on subclasses of proteins with Ser-, Ala-, Thr-, Gly-, Cys-, and Val- N-termini.²⁴⁸ Since these amino acids are more frequently expressed in the N-termini of eukaryotic proteins, NatA is the major NAT corresponding to the whole number of its potential substrates.²⁴⁸ Furthermore, another eukaryotic NAT, NatD, is specifically responsible for the acetylation of proteins with the Ser-Gly-Gly- or Ser-Gly-Arg- N-termini, which are strongly resembling the N-termini of the human histones H2A and H4.²⁴⁹

Although for a long time the functional importance of this most common cotranslational modification was elusive, recent experiments with the yeast Saccharomyces cerevisiae revealed that the Nt-acetylated Met residue, as well as Nt-acetylated Ser, Ala, Val, Thr, and Cys residues, could act as a degradation signal (degron), targeted by the Doa10 ubiquitin ligase.²⁴⁴

O-acetylation

Traditionally, lysine residues were considered as the primary target of acetylation in proteins. However, recently, the ability of a bacterial effector from Yersinia YopJ to serve as an acetyltransferase responsible for acetylation of serine and threonine residues in the host cells was discovered.^250,251 YopJ is known to inhibit mitogen-activated protein kinase (MAPK) and the nuclear factor kappaB (NFkappaB) signaling pathways playing an important role in innate immune response by preventing activation of the family of MAPK kinases (MAPKK). Here, YopJ was shown to block phosphorylation of MAPKK6 by acetylation of the critical serine and threonine residues in the activation loop of this important kinase.²⁵⁰ In a subsequent study, another important target of the YopJ acetylation was found, the MAP kinase kinase MEK2, where acetylation of the serine residues in the activation loop was shown to prevent the phosphorylation of these serine residues needed for the MEK2 activation and downstream signal propagation.²⁵¹ Also, YopJ was able to acetylate a threonine residue in the activation loop of both the α and β subunits of the IKK kinase from the NF-kappaB pathway.²⁵¹ Based on these observations, it has been concluded that the fact that serine and threonine can be not only phosphorylated and O-glycosylated but also (reversibly) acetylated adds a new level of complexity to the signaling machinery.²⁵²

ADP-ribosylation

Although arginine, cysteine, asparagine, and modified histidine (diphthamide) residues in proteins are well-established targets of the ADP-ribosylation (which is the transfer of the ADP-ribose moiety from NAD⁺ to specific amino acid residues on substrate proteins) via the action of various protein-mono-ADP-ribosyltransferases, the capability of formation of an acetal linkage between ADP-ribose and the hydroxyl group of a protein acceptor such as serine, threonine, tyrosine, hydroxyproline, or hydroxylysine residues has been demonstrated,²⁵³ indicating that these residues can be ADP-ribosylated. In agreement with this hypothesis, it has been established that an endoplasmic-reticulum-associated mono-ADP-ribosyltransferase PARP16/ARTD15 interacts with karyopherin-β1, and modifies neither acidic nor basic residues, but threonine or serine residues of this protein.^254,255

O-palmitoylation

Palmitoylation is the covalent attachment of a palmitoyl group derived from the palmitic acid (which is a 16-carbon fatty acid) to cysteine (S-palmitoylation), serine or threonine residues (O-palmitoylation). This type of PTM is rather common among various cellular signaling proteins, since it mediates the interaction of proteins with membranes and other proteins and can control the biological activity of a protein.²⁵⁶ In S-palmitoylation, the palmitate attachment occurs through a labile thioester bond and is readily reversible in cells,²⁵⁷ whereas O-palmitoylation involves a serine -OH group to form an oxyester linkage²⁵⁶ via the process catalyzed by the membrane-bound O-acyltransferases.²⁵⁸ An illustrative example of a protein with a crucial O-palmitoylation is Wnt protein, acylation of which at a conserved serine residue Ser209 plays an important role in the Wnt secretion, since protein defective in acylation at Ser209 is not secreted from cells, being retained in the endoplasmic reticulum (ER).²⁵⁸ Another example is histone H4, which has been recently shown to be O-palmitoylated at Ser47 by action of the acyl-CoA:lysophosphatidylcholine acyltransferase (Lpcat1) in a calcium-regulated manner.²⁵⁹

O-decanoylation and O-octanoylation

Serine residues of some proteins or functional peptides are acylated with 8- or 10-carbon fatty acids, octanoate (C8) or decanoate (C10), respectively. An illustrative example of a polypeptide with these PTMs is ghrelin, which is an appetite-regulating hormone (“hunger hormone”) secreted by the food-deprived stomach.²⁶⁰ The aforementioned forms of ghrelin acylation are known to target Ser3 residue of this hormone.²⁶¹ Fatty acid modifications of ghrelin are essential for the ghrelin-induced growth hormone release from the pituitary and also for the appetite stimulation.²⁶⁰ These acyl-modification of ghrelin are catalyzed by the ghrelin-O-acyl transferase.²⁶⁰

O-sulfonation

O-sulfonation of proteins on serine and threonine residues is one of the PTMs discovered via the use of the on-line high-performance liquid chromatography (HPLC) tandem electrospray mass spectrometry applied to proteins isolated by SDS-PAGE.²⁶² Although O-sulfonation is the major type of cellular sulfonation and although different biopolymers (such as polysaccharides, steroids, catecholamines, and proteins) are known to be commonly subjected to O-sulfonation catalyzed by several cytosolic and membrane-bound sulfotransferases, the primary mode of action of these sulfotransferases on proteins was the catalysis of the direct protein sulfonation on the O₄ position of tyrosine residues.²⁶³ Since O-sulfonation on tyrosine residues is one of the last PTMS affecting proteins transiting the trans-Golgi, this modification is present on many secreted and plasma membrane proteins of all metazoan species.²⁶³ In addition to the tyrosine-targeted O-sulfonation, serine and threonine residues were shown to be subjected to this type of PTM in several proteins of diverse class and function isolated from eukaryotes spanning the range from a unicellular parasite to humans.²⁶²

Serine-based PTMs in lantabiotics

Analysis of the lacticin 3147, which is a 2-component bacteriocin (i.e., a member of a large group of proteinaceous compounds produced by Gram-positive bacteria, which display an antimicrobial activity directed primarily against other Gram-positive organisms) generated by the Lactococcus lactis subspecies lactis DPC3147, revealed the presence of several specific PTMs.²⁶⁴ Among these PTMs is selective dehydration of the hydroxy amino acids (Ser and Thr) to form α,β-unsaturated residues dehydroalanine (Dha) and dehydrobutyrine (Dhb), respectively.²⁶⁵ Subsequently, Dha and Dhb can undergo Michael addition reactions with the thiol group of specific cysteine residues to form lanthionine (Lan) and β-methyl-lanthionine (MeLan) residues, which are specific to lantabiotics.^264,266 Another unusual posttranslationally modified residue described for the lacticin 3147 is D-alanine, which is produced not due to a simple isomerase conversion from L-alanine, but as a result of a 2-step α-carbon stereoinversion reaction, where serine is initially dehydrated to Dha and then further modified by some stereospecific hydrogenating enzyme or enzyme system to produce D-alanine.^264,267 Therefore, this posttranslational serine-D-alanine conversion represents the unique case of the incorporation of a d-amino acid in a ribosomally synthesized prokaryotic peptide.²⁶⁷

Amidation

Secreted proteins can undergo the C-terminal peptidyl-serine amidation, where the hydrolysis and oxidation of an interior Ser-Gly peptide results in the formation of a C-terminal peptidyl-serine amide. Here, glycine donates the “amine group” to the neighboring C-terminal residue generating an amide of the recipient amino acid, effectively removing the reactive carboxy group from the C-terminal amino acid. Although in mediating this amide formation, the only known donor amino acid is glycine, the recipient residue could be any amino acid with the exception of aspartate.²⁶⁸

Yin-Yang sites

Sections above illustrate the ability of serine to undergo multiple types of post-translational modifications (PTMs). Although these different serine-targeting PTMs are discussed individually, one should keep in mind that different PTMs may compete for the occupancy of a single residue. Obviously, this between-PTM competition is not unique to serine. However, one of the best studied classes of such PTM competitive sites is related to the cases which involve the competition between phosphorylation and O-glycosylation (Yin-Yang sites). Competition for the aforementioned Yin-Yang sites is determined by the remarkable similarities of global functional outcomes caused by these 2 PTMs. In fact, similar to phosphorylation, O-GlcNAcylation is known to be a rapidly cycling posttranslational modification that is directly involved in the regulation of many cellular processes.²⁶⁹ The phenomenon of the between-PTM competition was briefly discussed in the 'O-acetylation section', and several illustrative examples of proteins with such Yin-Yang sites are represented below. The murine estrogen receptor-β (mER-β) is one of the best studied cases, where there is an intricate functional competition between the O-GlcNAcylation and O-phosphorylation for Ser₁₆.²⁷⁰ It has been shown that reciprocal occupancy of this residue by either O-phosphate or O-GlcNAc plays a crucial role in modulation of the degradation and activity of mER-β.²⁷⁰

Obviously, the cross-talk between O-GlcNAcylation and O-phosphorylation is not limited to the mER-β, being instead a very common phenomenon. In fact, when phosphorylation site occupancy of 711 sites was determined after the global increase in O-GlcNAcylation, almost all actively cycling phosphorylation sites were noticeably affected, suggesting that an interplay between these 2 PTMs may arise due to the steric competition for occupancy at the same or proximal sites and by the ability of each modification to regulate the enzymatic machinery of other PTM.²⁷¹ A recent comprehensive review emphasized that this extensive interplay between the O-GlcNAcylation and O-phosphorylation is crucial for functionality of many proteins.²⁶⁹ Also, most anecdotal examples suggest that Yin-Yang sites happen to occur in disordered regions. This conclusion is supported by the resent statistical association study which clearly showed that although many PTM sites were systematically found within the disordered regions, target sites modified by more than one type of PTM showed even stronger preferences toward disordered regions than their single-PTM counterparts.²²²

Polyserine side of the polyglutamine disorders

A family neurodegenerative diseases including Huntington's disease, spino-cerebellar ataxia (SCA) 1, 2, 6, 7 and 3, spinobulbar muscular atrophy, and dentatorubal-pallidoluysian atrophy are autosomal dominant neurodegenerative conditions caused by the abnormal expansion of a (CAG)_n repeat encoding a polyglutamine tract within the target protein (e.g., N-terminus of huntingtin).²⁷² In addition to the polyglutamine expansion some human diseases and malformations (e.g., autosomal dominant oculopharyngeal muscular dystrophy (OPMD), synpolydactyly type II; cleidocranial dysplasia; holoprosencephaly (HPE5); hand-foot-genital syndrome; Blepharophimosis, ptosis, and epicanthus inversus; X-linked mental retardation with isolated growth hormone deficiency; X-linked infantile spasm syndrome; Partington syndrome; X-linked lissencephaly with ambiguous genitalia; congenital central hypoventilation syndrome/Ondine curse; and mental retardation X-linked 36 and 54²⁷³) are known to be caused by the polyalanine expansion mutations resulting from the abnormal expansion of the (GCG)_n, (GCA)_n, (GCT)_n, and/or (GCC)_n repeats all encoding polyalanine tracks.^273-276

An intricate link was found between the polyglutamine and polyalanine expansion diseases, where the +1 frameshift of the original SCA3 gene was shown to result in shifting from the polyglutamine (CAG)_n to the polyalanine (GCA)_n-encoding frame, thereby defining the transition from the polyglutamine to the polyalanine expansion mutations.^277,278 Based on these observations, it was hypothesized that since one of the serine codons is AGC, there is a chance that the +2 frameshift of the original polyglutamine-encoding frame (CAG)_n would promote a transition to the polyserine-encoding frame (GCA)_n.²⁷⁹ This hypothesis was confirmed by finding both polyalanine- and polyserine-containing huntingtin mutants in human postmortem Huntington's disease brains and in a transgenic mouse model of Huntington's disease.²⁷⁹ On a more general level, it seems that the aforementioned mechanism of homopolymeric conversion is rather common. In fact, recent studies showed that RNA transcripts with expanded CAG repeats can be translated in the complete absence of a starting ATG codon. This novel translation mechanism known as Repeat Associated Non-ATG translation (RAN-translation) occurs across expanded CAG repeats in all reading frames (CAG, AGC, and GCA) to produce homopolymeric proteins of long polyglutamine, polyserine, and polyalanine tracts.^280,281 Furthermore, the possibility for RAN-translation is not limited to the expanded CAG repeats since the expanded CTG tracts expressing CUG transcripts also have RAN-translation at all 3 frames (CUG, UGC, and GCU), thereby generating polyleucine, polycysteine, and polyalanine.²⁸⁰

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported in part by a grant from Russian Science Foundation RSCF № 14–24–00131