The carboxy-terminus, a key regulator of protein function

Abstract

All proteins end with a carboxyl terminus that has unique biophysical properties and is often disordered. Although there are examples of important C-termini functions, a more global role for the C-terminus is not yet established. In this review, we summarize research on C-termini, a unique region in proteins that cells exploit. Alternative splicing and proteolysis increase the diversity of proteins and peptides in cells with unique C-termini. The C-termini of proteins contain minimotifs, short peptides with an encoded function generally characterized as binding, posttranslational modifications, and trafficking. Many of these activities are specific to minimotifs on the C-terminus. Approximately 13% of C-termini in the human proteome have a known minimotif, and the majority, if not all of the remaining termini have conserved motifs inferring a function that remains to be discovered. C-termini, their predictions, and their functions are collated in the C-terminome, Proteus, and Terminus Oriented Protein Function INferred Database (TopFIND) database/web systems. Many C-termini are well conserved, and some have a known role in health and disease. We envision that this summary of C-termini will guide future investigation of their biochemical and physiological significance.

C-termini background

The carboxyl (C-) termini of proteins are often disordered and solvent exposed (Lobanov et al. 2010). This conformational flexibility of the C-terminus enables induced fit interactions through one or more C-terminal minimotifs. Unlike longer and more complex protein domains with stable 3D folds, shorter minimotifs can have secondary structures that drive weak, transient, and dynamic protein interactions (Neduva and Russell 2005; Sargeant et al. 2012). Minimotifs (also known as short linear motifs [SLiMs]) are much shorter (2–15 residues) functional components of proteins. Fundamentally, minimotifs are the targets of protein domains for protein–protein interactions, are modified by co- and post-translational modifications (PTMs), and are necessary and sufficient for trafficking of proteins to specific sub-cellular localizations (Balla et al. 2006; Rajasekaran et al. 2009; Vyas et al. 2009; Fortelny et al. 2015). The C-terminal minimotifs take advantage of the additional specificity of molecular recognition through the charged and polar backbone carboxyl moiety.

We have previously cataloged 3593 minimotifs in the human proteome that are anchored at the C-terminus and amalgamated in our C-terminome database/web-application (Sharma et al. 2016). For these experimentally verified C-terminal minimotifs, 82% are PTMs, 18% bind protein domains, and less than 1% are for protein trafficking (Sharma et al. 2016). For instance, the C-terminus of soluble proteins retained in the endoplasmic reticulum contains a KDEL > minimotif (letters are for the single letter amino acid code; “>” indicates the C-terminus of the proteins) (Cabrera et al. 2003; Sharma et al. 2016). Type II transmembrane proteins targeted to the plasma membrane often contains a CaaX > minimotif that enzymes recognize and prenylate the Cys thiol in the minimotif, “a” is an aliphatic amino acid, and “x” represents redundancy for any of the 20 amino acids (Sharma et al. 2016; Wang and Casey 2016). Lipidation modifications, like prenylation anchor proteins to membranes by insertion of the hydrophobic fatty acid chains into the lipid bilayer. Similarly, PDZ domains, bind to other proteins containing [DE]x[LV]>, x[ST[x[LV]>, and xϕxϕ> minimotifs at their C-terminus, where “ɸ” represents hydrophobic residues and residues within brackets indicate positional degeneracy (Beuming et al. 2005; Sharma et al. 2016). There are approximately 500 proteins with at least one PDZ domain and approximately 3,000 instances of C-terminal PDZ minimotifs in the human proteome (Letunic and Bork 2018). Such a large number of C-terminal minimotif instances for one type of minimotif: protein domain interaction supports a generalizable use of C-terminal minimotifs for regulating cellular processes.

Early studies on the C-termini of proteins examined their role in protein translation termination, folding, and enzymatic activities. Several reviews have focused on these aspects of protein C-termini. Marino et al. and Tanco et al. reviewed the contemporary proteomics techniques to study C-terminal modifications (Marino et al. 2015; Tanco et al. 2015), and Carugo covered the biophysical properties of both the N- and the C-protein termini (Carugo 2017). However, while Chung et al. summarized the molecular functions of protein C-termini (Chung et al. 2002), there is no recent comprehensive review of the C-terminal region of proteins and C-terminal minimotifs in the last 15 years.

Herein, we cover the biophysical properties of the C-terminus; different cellular mechanisms that produce a staggering number of C-termini, the C-terminal sequence patterns, their conservation, and evolution. We next, introduce the concept of minimotifs and subsequently, detail the molecular function of the C-terminal minimotifs. We also address the biochemical and bioinformatics approaches that have identified C-terminal sequences, sequence patterns, and functions. We also cover the databases dedicated to C-termini and their role in disease.

Biophysical properties of the carboxyl terminus

Because C-termini are polar, like charged amino acids, they are generally solvent exposed and available for binding to and modification by enzymes. To assess if C-terminal residues are solvent-exposed, Jacob and Unger calculated the distance of the terminal residues from the center of mass for 425 small monomeric proteins. The C-terminal residues were more distant from the COM than other residues including the charged residues (Jacob and Unger 2007). On average, ~87% of the C-terminal residues were solvent exposed, consistent with most solved proteins structures in the PDB. Given the polarity of the C-terminus, its solvent exposure is generally consistent with thermodynamics constraints (Jacob and Unger 2007).

While C-termini are on the surface, the last 5–10 amino acids lack significant electron density in structures determined by X-ray diffraction or inter-residue NOEs in structures determined by NMR. These disordered regions, generally are flexible with no stable structure, have a high net charge, low sequence complexity, and contain fewer aromatic residues (Lobanov et al. 2010). Several lines of evidence support disordered carboxyl termini. First, the C-terminal amino acid has fewer constraints, thus making it more flexible than other residues (Lobanov et al. 2010). Second, 30% of the disordered residues in proteins with solved structures are in the last 10 amino acids of the C-terminus (Lobanov et al. 2010). We and others have previously presented an in-depth discussion of the heterotypic nature and classification of disorder, an important consideration for C-terminal motifs (Sargeant et al. 2012; van der Lee et al. 2014). Because it is disordered, the C-terminus can assume different conformations a concept referred to as molecular recognition features or elements (MoRFs or MoREs) in intrinsically disordered proteins (Mohan et al. 2006; Hsu et al. 2013).

The disordered C-terminus binds to other proteins with an induced fit. Therefore, the MoRFs minimotifs with several different structures in the same protein can engage binding distinct partners with several different structural modes of interaction. Several disordered minimotifs assume α-helices or β-strands upon binding a target, whereas other disordered regions assume irregular secondary structures (Lobanov et al. 2010). Although disordered minimotifs can be present anywhere in a protein, they are particularly prevalent at the C-terminus. Notably, crystallizing a protein with its full C-terminus is difficult unless it is bound to its target (Laio and Micheletti 2006). Therefore, crystallographers often truncate the proteins at the C-terminus to reduce disorder, rendering the protein more prone to crystallization (Carugo 2017).

Despite the disorderedness, the C-terminus has local contacts and more often adopts an α-helix rather than a b-turn or coiled-coiled secondary structure (Laio and Micheletti 2006). Laio and Micheletti evaluated 373 monomeric proteins and 85 multidomain proteins from Protein Data Bank. By calculating the average gyration radius of the last 15 amino acids from the C-terminus, and the average number of contacts that the residues in any protein sequence make to the terminal residues, the authors found the C-terminal residues to be more locally organized and compact when compared to the N-terminus. The motif at the C-terminus is similar to an α-helix pattern (Laio and Micheletti 2006). An additional study identified a 1.3-fold preference for α-helices in the C-terminus over the N-terminus (Krishna and Englander 2005).

Although not relevant to minimotifs, there are examples where C-termini are also important for protein folding and are good epitopes for raising peptide-based antibodies (Asami-Odaka et al. 2005; Krishna and Englander 2005).

Cellular processes increasing the number of carboxyl termini

Approximately, 22,000 protein-encoding human genes are transcribed into mRNAs are subsequently translated into proteins with synthesis progressing from the amino to carboxyl terminus (Figures 1 and 2). There are several cellular processes that can produce many proteins with alternative C-termini from the same gene: splicing of pre-mRNA, translation termination by suppressor tRNAs, proteolytic processing of immature and mature proteins, degradation of C-termini by exonucleases, and PTMs. These processes create multiple C-terminal minimotifs in proteins produced from the same gene.

Figure 1.

Cellular processes creating a diverse C-terminal end. A protein-coding gene with three exons is transcribed into a pre-mRNA, alternative spliced, and each transcript is translated into a unique protein. Proteases cleave each protein and PTMs are attached creating new protein species.

Figure 2.

Estimate of factors contributing to numbers of C-terminal ends in the human proteome. The human genome has approximately 22,000 protein-coding genes. Each gene produces five alternatively spliced transcripts, with approximately 110,000 transcripts encoding the same number of proteins. Assuming each protein is proteolyzed during its life producing eight proteins then there are 1.8 M proteins with c-termini. With approximately 10 PTMs per protein, there are 17.6 M unique possible proteins present in a cell assuming one modification per protein (Jensen 2004; de Klerk and ‘t Hoen 2015; Tress et al. 2017). If we consider all PTM permutations, there could be nearly a quadrillion unique protein molecules, far exceeding the estimated 10 billion proteins per cell.

Alternative splicing of a pre-mRNA at differing intron-exon junctions produces multiple spliced mature mRNA molecules from the same gene. Approximately, 95% of the human genes are alternative spliced (Stamm et al. 2005; Wang et al. 2008). RNA-Seq analysis suggests that most genes produce, on average >10 unique transcripts (Tress et al. 2017). Approximately, 5% of the protein-coding human genes are more extensively spliced, expressing about 100 putative transcripts. An analysis of 1421 splice variants from 404 genes with minimotifs showed that minimotifs are enriched in splice variants (Weatheritt et al. 2012). Both PTM and binding minimotifs were identified in these splice variants supporting a well-defined role for molecular functions in C-termini. Based on these observations, we estimate about 220,000 unique C-termini from alternative splicing.

Another process increasing the number and diversity of the C-termini is alternative termination of protein translation by suppressor tRNAs. Suppressor tRNAs are tRNAs having mutations in anticodons. These mutations allow suppressor tRNAs to bind stop codons, add an amino acid and continue to elongate polypeptides in lieu of termination. Thus, this process extends the C-terminus of proteins. Additionally, deficiency of eukaryotic release factors (eRF1 and eRF3) generates alternative C-termini from premature translational termination through peptidyl-tRNA hydrolysis. It is clear that in some human cells premature translational termination and alternative termination of protein translation produces new C-termini (Janzen and Geballe 2004). Stop codon tRNAs can encode other amino acids express proteins with a different C-termini (Lee et al. 1990; Swart et al. 2016). The extent of this phenomenon at the human proteome level remains to be determined.

Precursor proteins are frequently proteolyzed creating new C-termini in mature proteins. This is prevalent in peptide hormones, precursor proteins, and zymogens. Exopeptidases and endopeptidases can produce different C-termini from the same protein substrate. Where exopeptidases trim proteins, one amino acid at a time from either protein termini, endopeptidases cleave proteins somewhere in the middle of their sequence. Both proteases produce new C-termini with minimotifs that change a protein’s function, stability, or location. It is challenging to biochemically identify the C-terminus of proteins biochemically as the C-terminal is relatively less reactive than the amino terminus (Marino et al. 2015). Nevertheless, the TopFIND group improvised MS techniques, such as C-terminal amine-based isotope labeling of substrates (C-TAILS) and combined fractional diagonal chromatography to identify 130,182 neo C-termini (Bonetto et al. 1997; Sechi and Chait 2000; Colaert et al. 2013; Nika et al. 2013; Fortelny et al. 2015; Zhang et al. 2018). Proteomic analysis of Escherichia coli, Mus musculus, and Homo sapiens identified that >10% of C-termini are derived from proteolysis, thus we estimate that endoproteolysis produces at least 2000 new human C-termini (Lange and Overall 2013). Although the large-scale impact of the exoprotease-generated C-terminal ends is not known, we extrapolate from a small set of proteins. Exoproteolysis produces an average of eight new C-terminal ends per protein for Fibrinogen-a C3f, C4a, and IYIH4 (Villanueva et al. 2006). Therefore, as many as 176,000 C-terminal ends may be present and available for C-terminal minimotif-driven interactions.

Nearly, all proteins have multiple PTMs. C-termini with PTMs can be considered chemically distinct entities, even though the sequence does not change. PTMs are added to growing polypeptides during the protein translation. An example is that Ser/Thr glycosylation is critical for protein folding quality control in the endoplasmic reticulum. PTMs on C-termini regulate the activity and localize proteins in a signal-dependent spatiotemporal manner. Although there are approximately 470 types of PTMs, only 1% of PTMs are thought to occur in vivo (Lobanov et al. 2010). Thus, PTM validation needs to be rigorous. The C-terminome database has approximately 3000 instances of PTMs at carboxyl termini.

Clearly, many types of transcriptional, translational, and post-translational events increase the number and diversity of C-termini and cells may have a total of 100,000–500,000 unique C-termini at a unique time. Many of these ends are likely to have functions.

Sequence conservation and composition of carboxyl-termini

Analysis of 1000 genomes project data revealed that 83% of the known C-terminal minimotifs are fixed in the human population with much of the variation in the PTM C-terminal minimotifs. Moreover, a global analysis done on the human proteome indicated that the C-terminal minimotifs are more constrained than the entire genome and protein-coding regions, indicating that the C-terminal minimotifs are likely to have important functional significance (Sharma et al. 2016).

The determinants of protein interactions and PTMs can be represented as sequence consensus motifs with a combination of fixed and degenerate positions. To predict and evaluate new motifs it is important to know whether the carboxyl termini in proteomes have altered amino acid composition or positional biases. Berezovsky et al. analyzed the last 10 amino acids of 1000 s of proteins: 1918 Escherichia coli, 3303 Saccharomyces cerevisiae, and 3243 Homo sapiens sequences (Berezovsky et al. 1997). In all three species, the positively charged K and R amino acids were favored in the two positions juxtaposed to the terminus. G residues were least common at the C-terminus (Berezovsky et al. 1997). However, a more recent, proteome-wide analysis indicated that the overall amino acid composition in the C-terminus (last 10 amino acids) is similar to that of the human proteome (Sharma et al. 2016). Certain residues, such as R, C, Q, H, K, P, and S, notably charged and polar amino acids, were slightly over-represented in the C-terminus (Sharma et al. 2016).

There are several biological processes that can contribute to bias in the amino acid composition of C-termini. Since many globular proteins begin to fold co-translationally, the C-terminus may contribute to folding and structural stability. In Escherichia coli, the last and the penultimate residue from the C-terminus of a growing polypeptide cooperatively enhance translation termination (Björnsson et al. 1996). The ordered last residues (an α-helix or β-strand) and K, R, and Q residues at the penultimate position correlated with an efficient translation termination for UGA stop codon. The last residue at the C-terminus may be important for the efficiency of translation termination and thus gene expression. In Escherichia coli, efficiently terminated genes have a K, whereas inefficiently terminated proteins have Thr or Pro as the last residue (Björnsson 1996).

Mutation of stop codons or a frameshift mutation that causes read-through of a stop codon introduces new C-termini. This is pervasive, as identified in human, mouse, rat, yeast, and Drosophila strains (Lobanov et al. 2010). In yeast new C-termini, are not yet fixed in the population indicating that they are likely still under selection (Giacomelli et al. 2007). Whether or not the alternative termini identified in these studies are of functional significance, will require further investigation.

Introduction to minimotifs

The term “motif” first described restriction enzyme recognition sites in nucleic acids in the literature (Smith and Wilcox 1970). One of the first functional protein motifs and the corresponding concept of a consensus sequence were identified for spore endoproteolysis of a protease recognition site in 1980s (Setlow et al. 1980). In 1987, the first trafficking motif, KDEL > was discovered and later identified as a receptor binding motif (Munro and Pelham 1987; Vitale and Denecke 1999). However, for many years, varied nomenclature of minimotifs existed in the literature. The conceptual term loosely included both structural and sequence motifs (Hodgman 1989; Thornton and Gardner 1989; Seto et al. 1990; Sheridan and Venkataraghavan 1992; Bork and Koonin 1996). In 1990, Tim Hunt unified the minimotif concept, calling these functional elements “short linear motifs” (Dice 1990). Eventually, it was the minimotif miner (MnM) and eukaryotic linear motif (ELM) resource groups that formalized a minimotif definition and model (Balla et al. 2006; Rajasekaran et al. 2009; Vyas et al. 2009; Dinkel et al. 2012; Mi et al. 2012; Lyon et al. 2018).

A minimotif has four attributes: a source protein, a target, an activity, and a secondary structure of the minimotif sequence (Vyas et al. 2009; Sargeant et al. 2012). A source is a protein that contains a minimotif sequence, having one or more known activities (binding, modification, and trafficking) and requires another molecule, a target, for its activity. Although structure is apparent in minimotif interaction with its targets, it is more complex often driven through induced-fit disorder to secondary structure transitions. Minimotifs are present in both disordered regions of proteins and also can occupy any of the 32 protein secondary structures (Vyas et al. 2009; Sargeant et al. 2012). An analysis of a nonrandom sample of 241,990 minimotifs in the MnM3 database quantified a distribution of 28% of the minimotifs in the intrinsically disordered regions, 27% in the folded regions, and 45% in the hybrid regions of proteins (Sargeant et al. 2012).

Minimotifs are modeled as consensus sequences, position specific-scoring matrices (PSSMs), and instances. Instances are contiguous amino acid sequences in a peptide or protein with that have an established molecular function. Sets of minimotif instances in several proteins can be aligned to obtain consensus sequences or PSSMs where degenerate substitutions at certain positions are modeled, generally by frequency. Consensus sequences and PSSMs infer new minimotifs instances that can be tested for prediction accuracy.

The majority of the known molecular functions and sequence specificity of minimotifs comes from the low-throughput experimentation examining a small number of instances. SPOT arrays, phage display, and affinity MS/MS, because of their throughput, are tools to more broadly assess minimotif specificity (Sundell and Ivarsson 2014; Liu 2017). Large numbers of PTM minimotifs are also discovered through affinity mass-spectrometry studies. Despite these advancements, determinants of minimotif specificity are incomplete. One likely explanation is that secondary structure is an essential component of specificity. However, because minimotif interactions are generally transient, reversible and have weak affinities (K_d = 1–10 μM), it may be difficult to resolve the complete specificity of minimotifs through high-throughput approaches that do not assess the relative strengths of the interactions (Neduva and Russell 2005).

C-terminal minimotifs

Minimotifs can be located at the C- or N-terminus, or at any position throughout a protein. The termini have unique chemistries and some minimotifs are specific to the termini. More than 3500 C-terminal minimotifs were identified from 73,532 human minimotifs from the MnM3 database (Sharma et al. 2016).

C-terminal minimotifs can have sequences with a consensus residue exactly on the C-terminus (anchored) or a short distance away from the last residue of the C-terminus (non-anchored). There are several examples of non-anchored C-terminal minimotifs. The FCYENEV minimotif in the Kir2.1 channel is both necessary and sufficient to export of Kir2.1 from the ER to Golgi complex (Ma et al. 2001; Stockklausner et al. 2001). Many G-protein-coupled receptors (GPCRs) contain a conserved hydrophobic minimotif, FxxxFxxxF, near its C-terminus (Bermak et al. 2001). DRiP78, an ER resident protein, binds to this minimotif and is essential for trafficking D1 GPCRs. The VxxSL motif in Kv1 channels is required for its trafficking to the cell surface (Levitan and Takimoto 2000).

Types of C-terminal minimotifs

Binding C-terminal minimotifs

At least 650 C-terminal minimotifs have a binding activity for protein interactions often driving large macromolecular complexes and cell signaling. Minimotif-domain interactions complement other modes of protein interactions, such as domain-domain interactions. At least 63 unique protein domains bind to specific minimotifs and several key examples are shown in Table 1 (Stein et al. 2011; Dinkel et al. 2012). Given a large number of minimotifs and their generally low complexity when compared to protein domains, many minimotif interactions are promiscuous engaging multiple target domains. However, the physical separation of proteins in different sub-cellular compartments, different phases of the cell cycle, and differential tissue-specific expression impart some specificity to minimotif interactions.

Table 1.

Examples of the C-terminal minimotifs and protein domain binding.

Domains	# H. sapiensa	# All species	Minimotif sequence	#Minimotifs	Reference
PDZ	666	333,040	Class I: x[S/T[x[L/V]>; Class II: xfs and protein domain	2849	Hung and Sheng (2002)
TPR	330	1,309,452	EEVD>	22	Brinker et al. (2002)
ZNF_UBD	11	–	GG>	–	Reyes-Turcu et al. (2006)
14_3_3	22	6,285	YDI>; YTV>	4	Fuglsang et al. (2003)

^aData from SMART databases.

14–3-3 proteins binding minimotifs.

The 14–3-3 family of proteins interact with phospho-proteins where the Ser and Thr residues within the minimotif sequence are phosphorylated (Coblitz et al. 2006). 14–3-3 domains can interact with both C-terminal and internal consensus sequences. In plants, 14–3-3 domains bind the YTV > motif in the Hþ-ATPase, present in the plasma membrane activating an auto-inhibited ATPase (Fuglsang et al. 2003). This instance of a 14–3-3 minimotif does not require phosphorylation to drive the interaction, although Ser/Thr phosphorylation is required for most 14–3-3 domains (Fuglsang et al. 2003; Madeira et al. 2015).

PDZ domain-binding minimotifs.

Proteins with PDZ domains function in signal transduction, trafficking, and are scaffolds for large molecule complexes (Kim and Sheng 2004; Lee and Zheng 2010). PDZ domains (80–100 residues) have only one binding site that binds to proteins with a C-terminal minimotif; however, there are exceptions for binding non-C-terminal motifs (Wong et al. 2003; Zhang et al. 2009). PDZ minimotifs are classified based on the consensus recognition patterns: class I: x[S/T[x[L/V]>, class II: xϕxϕ>, and class III: xDxV> (Hung and Sheng 2002). 591 C-terminal human minimotifs interact with PDZ domains (Sharma et al. 2016). In Homo sapiens, 546 proteins in the SMART database contain PDZ domains (Letunic and Bork 2018). Upon binding to the PDZ domain, the C-terminal minimotif assumes a b-strand conformation and extends a b-sheet in the PDZ domain, an example of a secondary structure determinant (Laskowski et al. 2005; Lee and Zheng 2010). Minimotifs for phosphorylation of Ser, Thr, and Tyr residues in, or juxtaposed to a PDZ binding motif, regulate binding of C-terminal minimotifs to PDZ domains. However, there is no consensus as to whether phosphorylation inhibits or enhances binding (Lee and Zheng 2010). PDZ-binding minimotifs in the C-terminus are well-conserved among vertebrates and invertebrates (Chimura et al. 2011).

Transducin-like enhancer (TLE) binding minimotifs.

There are 57 TLE of split domains in co-repressors bind to and repressor proteins that inactivate promoters (Fisher et al. 1996; Aronson et al. 1997). There are 11 human proteins that have the WRP[YW] TLE consensus binding motif. The core binding factor acute myeloid leukemia (AML1/CBF) a-subunit contains a C-terminal WRPY > minimotif that binds to transcriptional corepressors with TLE domains. Together, TLE domain proteins and their binding partners are required for development. TLE1 inhibits the trans-activation of T cell receptor enhancers (Levanon et al. 1998). Other C-terminal sequences in the AML1 are also required for these interactions (Fisher et al. 1996; Aronson et al. 1997). Hairy family factors contain a WRPW > minimotif, which is both necessary and sufficient for the binding of Groucho to the Hairy-related transcription factors, which contains a TLE domain (Chen and Courey 2000).

Tandem tetratricopeptide repeat (TPR) binding minimotifs.

TPRs are domains that interact with minimotifs in multiprotein complexes (Zeytuni and Zarivach 2012). TPRs are 34 amino acid helix-turn-helix structural motif with 1–16 repeat in proteins (Cortajarena and Regan 2006). Although there is no known general consensus minimotif sequence for TPRs, the TPRs in Hop proteins bind to the EEVD > minimotifs in the Hsp70 and Hsp90 chaperones. These minimotifs bind in an extended coil conformation (Brinker et al. 2002; Zeytuni and Zarivach 2012). There are 273 TPR domain-containing proteins and 22 EEVD > minimotifs in the human proteome (Sharma et al. 2016).

Ubiquitin-binding (UBP) minimotifs.

Isopeptidase T contains a zinc-finger ubiquitin-binding domain (ZnF UBP) that binds the GG > minimotif of ubiquitin (Reyes-Turcu et al. 2006). Although many UBP domains exist, the general specificity for GG > remains to be determined, as this minimotif may bind exclusively to ZnF UBP (Hurley et al. 2006).

Trafficking C-terminal minimotifs

Proteins trafficking to specific cellular compartments are important for correct function. Several trafficking signals are located at the C-terminus of proteins and masking of such signals by fusing an epitope tag often leads to their mislocalization. There are 44 trafficking C-terminal minimotif consensi and instances that are most often necessary and sufficient for localization (Sharma et al. 2016). Many proteins destined to be in the secretory pathway and endomembrane system contains minimotif signals at their C-termini (Yofe et al. 2016). Aberrant mislocalization to the wrong subcellular compartment can lead to disease. Standard techniques to determine protein localization, and thereby the trafficking determinants include immunostaining, confocal microscopy, Fösters resonance energy transfer (FRET) microscopy, fluorescence life-time imaging microscopy, and various ultracentrifugation fractionation approaches (Teesalu et al. 2009). Several examples of trafficking C-terminal minimotifs are summarized below:

Endoplasmic reticulum (ER) retention and retrieval minimotifs.

Protein sorting between the ER and Golgi-apparatus is mostly dependent upon C-terminal minimotifs (Teasdale and Jackson 1996). Soluble proteins synthesized in the ER lumen contain an ER retention minimotif, KDEL> (Gomez et al. 2000). Type I transmembrane proteins contain dilysine retrieval minimotifs, KKx(1,3), KxKx>, and KxKxx> (Jackson et al. 1990; Gomez et al. 2000). These Lys residues are critical for the functioning and cannot be replaced with other charged residues such as Arg and His (Jackson et al. 1990).

Internalization minimotifs.

The C-end rule (also known as C-endR) peptides have sequence (R/K)xx(R/K)> (Teesalu et al. 2009). These peptides are tissue penetrating internalization peptides that bind to Neuropilin-1 (NRP-1), a cell surface receptor that is essential for angiogenesis and nervous system development (Teesalu et al. 2009; Zanuy et al. 2013). A screen for proteins that bind C-endR peptide minimotifs identified NRP-1 (Teesalu et al. 2009).

Peroxisome targeting minimotif.

Peroxisomes are small organelles that are present in all eukaryotic cells and contain hydrolytic enzymes. Peroxisomes contain Pex5, a soluble import receptor expressed on peroxisomal membranes. Pex5 binds to proteins with the SKL > minimotif on the proteins that are imported into the peroxisome lumen. The SKL > minimotif is in 34 proteins and is called the PTS1 signal. Another peroxisome import signal, PTS2 has a minimotif on the N-terminus of imported proteins (Smith and Aitchison 2013).

Post-translation modification of C-terminal minimotifs

Translation is only the start of protein a proteins life. Co- and PTMs are the 100 s of covalent attachments of molecules to polypeptides changing their physiochemical properties. Broadly, there are two main types of PTMs: 1. The exoproteolysis of polypeptides removes small stretches of residues from either termini. Endoproteolysis of a polypeptide hydrolyzes an internal peptide bond producing two smaller polypeptide chains. 2. The reversible or the irreversible addition or removal of small chemical moieties, isomerizations, or covalent attachment of other proteins. PTMs can occur at multiple sites in a protein and/or multiple PTMs can occur sequentially in a protein regulating functional outputs. There are approximately 469 different types of modifications in eukaryotes (Khoury et al. 2011; UniProt Consortium 2015). The majority of known C-terminal minimotifs are PTM minimotifs. The major experimentally verified C-terminal modifications include phosphorylation, acetylation, ubiquitylation, proteolysis, amidation, and lipidation with another 18 PTMs occurring less frequently in a limited set of C-terminal minimotifs (Sharma et al. 2016).

Amidation.

The C-terminal a-amidation of bioactive neuropeptides and hormones catalyzed by peptidylglycine a-amidating mono-oxygenase (PAM) is one of the most common PTMs and is the 4th most common in Swiss-Prot (Marino et al. 2015). Amidated peptides are produced by sequential degradation of a C-terminal Gly residue. Peptidylglycine a-hydroxylating monooxygenase (PHM) and peptidyl-a-hydroxyglycine a-amidating lyase (PAL), the two catalytic domains of PAM amidates peptides. The amide group neutralizes the negative charge of the carboxyl end, thus changing its chemical properties (Marino et al. 2015). In addition to many bioactive peptides, the C-terminus of the heavy chain of many monoclonal antibody-based drugs is amidated, although its significance is not yet known (Tsubaki et al. 2013). Since cells contain exopeptidases, the C-terminus of the cellpenetrating peptides is often amidated, which increases their half-life (Soleymani-Goloujeh et al. 2017).

Crotonylation.

Crotonylation is the addition of crotonyl group to Lys side chains, regulating gene expression by modifying histones in promoters or enhancer nucleosomes (Tan et al. 2011; Ju and He 2017; Wei et al. 2017). Although not exclusive to the C-terminus, KxxK, and PxK minimotifs have a higher propensity for crotonylation (Wei et al. 2017). Histone proteins (H2A1H, H2A1B, H2A1, H2A1D, H2A2A, H2A3, and H2A1C) with conserved C-termini are crotonylated (Tan et al. 2011; Sharma et al. 2016).

Glycosylation and glycation.

Glycosylation is an enzyme-catalyzed covalent addition of sugars, both co- and post-translationally (Spiro 2002). Glycation, on the other hand, is the non-enzymatic reaction of adding sugars. Sugars are added primarily on the newly synthesized unfolded polyproteins resulting in the proper folding of proteins. Glycosylation usually occurs on the membrane-bound and soluble proteins of the secretory pathway. Most often, glycosylation occurs in the ER proteins, but several examples of proteins synthesized in the cytosol also exist. Several types of glycosylation exist based on the position of linkage, with N-linked and O-linked glycosylation being the most common. N-glycosylation forms a covalent amide link-age to the side chain of Asn in Nx[ST] minimotifs and O-glycosylation occurs on Ser and Thr residue with no clear specificity. While most glycosylation occurs throughout proteins, there are at least ten N-glycosylation, three O-glycosylation, and two glycation minimo-tif instances in the C-terminus (Sharma et al. 2016).

Lipidation.

Lipidation is the addition of different types of fatty acyl or polyisoprenyl groups moieties to either termini. Lipids are attached to the side chains of Cys, Ser, and Lys and target proteins to a lipid bilayer. Cholesterol esterification and glycosylphosphatidylinositol anchorage are two other types of C-terminal lipid modifications on extracellular membrane-bound proteins (Jiang et al. 2018). The two sub-types of prenylation modification are farnesylation (three isoprene groups) and geranylgeranylation (four isoprene groups) on the Cys residues near the C-termini. A total of 45 prenylation instances (six farnesylation and two geranylgeranylations) are noted at the C-termini (Jiang et al. 2018).

The substrates of farnesyltransferase and geranylgeranyl transferase contain Caax > minimotif. In Ras proteins and lamin B, after prenylation of C in Caax>, the aax > sequence gets removed and another enzyme carboxymethylates the prenylated Cys C-terminus. Rab geranylgeranyl transferase adds two geranylgeranyl groups at the CC > or CxC > minimotifs of Rab proteins (Sharma et al. 2016). N-terminal lipidation modification is also known to occur at Gly in the consensus sequence < MGxxx[S/T] (Jiang et al. 2018).

Methylation and demethylation.

There are two types of biological protein methylation: 1. methylation of the C-terminal protein carboxyl group; and 2. that on side chains of internal residues such as Arg or Lys. Carboxyl methylation neutralizes the charge of the C-terminus (Kawata et al. 1990; Favre et al. 1994; Ghomashchi et al. 1995; Choy et al. 1999; Brahms et al. 2000; Kowluru 2000; Ahmad et al. 2002; Winter-Vann et al. 2003; Ong et al. 2004; Zhang et al. 2008; Ren et al. 2010). Carboxymethylation is most often attached to the terminal Cys carboxylate during C-terminal protein lipidation. One of the classical examples is the methylation of phosphoprotein phosphatase 2 A which along with protein phosphatase 1 (PP1) accounts for more than 90% of the phosphatase activity in mammalian cells (Tolstykh et al. 2000). The TPDYFL > minimotif is methylated at the Leu carboxyl. While side chain methylation can occur throughout a protein, there are 43 methylation instances (mono-and di-) of methylation of these amino acids in the C-terminus. Methylated Arg or Lys residues drive protein-protein interactions.

Phosphorylation and dephosphorylation.

Phosphorylation, the most predominant form of PTM, is the reversible covalent attachment of phosphate groups, on Ser, Thr, and Tyr residues, and in the case of prokaryotes on His and Asp. The interplay of kinases phosphorylating proteins and phosphatases removing the phosphate moieties act as an “on” and “off” switch in signaling pathways and almost all cellular processes. Of 2096 C-terminal minimotifs are phosphorylated with ~73% modifications at Ser and the rest on Thr and Tyr (Karve and Cheema 2011). Given that Ser is the most common residue in the proteome, it is likely that the phosphoproteome studies are biased in predicting Ser to be more commonly phosphorylated residue when compared to Thr and Tyr (Sharma et al. 2016).

Proteolysis.

Exo- and endo-peptidases catalyze an irreversible hydrolysis of peptide bonds. There are ~130,000 C-terminal instances of proteins reported due to limited proteolysis (Fortelny et al. 2015, p. 0; Sharma et al. 2016). The solvent-exposed C-terminus is often prone to proteolytic cleavage. This may be because the C-terminus often does not assume a stable structure. The C-terminus is often removed from proteins to enhance their crystallization and also often does not produce a signal in electron density maps for protein structure determination by X-ray crystallography nor NOEs in structures determined by nuclear magnetic resonance spectrometry (Laio and Micheletti 2006). Dedicated sub-fields of proteomics termed as degradomics or terminomics have emerged to study proteolysis. The current methods for identifying peptides produced by proteolysis were recently reviewed (Rogers and Overall 2013).

Prohormones are expressed in an inactive form. Likewise, many enzymes, including proteases are expressed as inactive zymogens. Both are activated upon endoproteolytic processing, releasing the bioactive molecule and producing new C-termini. Carboxypeptidases hydrolyze a peptide bond at the C-terminus of proteins, and have a role in blood clotting, wound healing, and many other cellular processes (Wallace et al. 1982; Lyons and Fricker 2010; Tanco et al. 2015).

In addition to prohormones, the ubiquitin/SUMO/NEDD family members are expressed in an inactive form having an extended C-terminal tail. Many proteases cleave ubiquitin unmasking a GG motif, then recognized for attachment to Lys residues (Grou et al. 2015). The SENP (sentrin/SUMO-specific protease) family of proteases cleave the inactive form of small ubiquitinlike modifier (SUMO) exposing a di-glycine motif recognized by SUMO-E1 ligases (Reyes-Turcu et al. 2006). Similarly, NEDP1 or UCHL3 proteases cleave the C-terminus of Nedd8, another ubiquitin-like protein, activation it for recognition by Nedd8 E1 ligases (Zheng and Shabek 2017). The ligases bind the functionally active forms of ubiquitin and SUMO catalyzing formation of an isopeptide bond with the Lys of their target substrates. Some additional examples are listed in Table 2.

Table 2.

Examples proteins activated upon proteolysis.

Precursor protein	Mature protein	Protease	Function	PubMed
Precursor-small ubiquitin like modifier (SUMO)	SUMO-1, SUMO-2, and SUMO-3	SENP (sentrin/SUMO-specific protease) family of proteins	Mark for degradation, localization, etc.	van der Veen and Ploegh (2012)
Precursor polyubiquitin protein	Ubiquitin	Many proteases including UCHL3, USP9X, USP7, and USP5	Mark for degradation, localization, etc.	Kimura and Tanaka (2010); Grou et al. (2015)
Neural precursor cell expressed developmentally down-regulated protein 8 (precursor-Nedd8)	Nedd8	NEDP1 (DEN, SEN8); UCHL3	Cell cycle progression and survival	Wada et al. (1998); van der Veen and Ploegh (2012)
Fau with ribosomal S30	Ubiquitin-like protein FUBI	–	ERK-MAPK pathway and LPS/interferon-induced apoptosis	Watanabe et al. (2013)
Ubiquitin-fold modifier (UFM-1) precursor	Ufm1	UfSP1 and UfSP2	Nuclear receptors mediated transcription and cellular response to ER stress	van der Veen and Ploegh (2012); Yoo et al. (2014)
Interferon-stimulated protein 15 precursor	ISG15	UBP43	Immunity	Jeon et al. (2010); Rahnefeld et al. (2014)
Proinsulin	Insulin	PC1 and PC2	Metabolism	Orci et al. (1986)
Angiotensinogen	Angiotensin I and Angiotensin II	Renin and Angiotensin-converting enzyme	Increase vasopressin production in CNS, Adhesion and aggregation of platelets, etc.	Velez et al. (2012)
Pro-caspase 8a	Caspase p30	Not known yet	Sensitization of cells toward apoptosis	Hoffmann et al. (2009)

Sulfation.

Only a few proteins in the secretory pathway are sulfated on tyrosine as recently reviewed (Kehoe and Bertozzi 2000; Stone et al. 2009; Yang et al. 2015). Tyrosine sulfation drives protein–protein interactions, primarily of several extracellular and secreted proteins (Kehoe and Bertozzi 2000; Yang et al. 2015). For example, cholecystokinin (CCKN) contains two C-terminal sulfated tyrosines (Vishnuvardhan and Beinfeld 2000). The inhibition of sulfation in precursor CCKN had no effect on the processing of precursor CCKN but increased the amount of CCKN secretion in endocrine cells. Therefore, it seems that tyrosine sulfation renders proteins more stable and soluble. However, the implications of such mutations in vivo still need to be established. The C-terminal sulfated tyrosines are also present in the fibrin binding domain of fibronectin (Liu and Suiko 1987; Meh et al. 2001).

Ubiquitylation and sumoylation.

Ubiquitylation is a reversible modification that targets proteins for degradation in different proteolytic structures, such as proteasome, vacuoles, and lysosomes; it is also initiates cell signaling (Komander and Rape 2012; Herhaus and Dikic 2015). In addition, new functions of ubiquitin addition in transcription, translation, and DNA repair are reported recently (Weissman 2001). Ubiquitin is an 8.5 kDa protein that is conjugated to Lys on other proteins. Polyubiquitin chains are built by sequential ubiquitylation of several Lys residues in ubiquitin. Subtle differences in the Lys residue positions that are ubiquitylated, types of ubiquitin chains (mono-versus multi-) and intra-cellular location of ubiquitylated proteins determine the fate of a protein. Like ubiquitin modification, sumoylation also regulates proteins in different cellular processes. SUMO proteins conjugate to the C-terminal Lys residues. Both ubiquitylation and sumoylation minimotifs are located throughout proteins with several C-terminal minimotifs including sumoylated Lys residues in Axin1 and p53 and an additional 215 instances of ubiquitylated C-terminal minimotifs (Gostissa et al. 1999; Muller et al. 2000; Rui et al. 2002; Li et al. 2006). Although the frequency of Lys at the C-terminus is higher than in the overall proteome (Sharma et al. 2016), based on current studies it is unlikely that all C-terminal Lys residues are ubiquitylated and sumoylated and that those modifications target proteins for degradation.

Identifying new carboxyl terminal minimotif activities

Despite significant bioinformatics advancements, computational methods to predict functional C-terminal sequence patterns with high sensitivity and specificity are rather limited. Therefore, the identification of C-terminal minimotifs thus far largely relies on low-throughput experiments to discover new or validate predicted C-terminal functions. PTM minimotifs are identified primarily by mass-spectrometry. Binding minimotifs for specific protein domains are identified through spot peptide array, phage display libraries, and other low-throughput assays. Proteome-wide studies to identify trafficking minimotifs do not yet exist.

Analysis of the C-termini of proteins has identified many over-represented C-terminal sequences and patterns in proteomes from a variety of organisms (Attwood et al. 1997; Gostissa et al. 1999; Gatto and Berg 2003; Prakash et al. 2004; Bahir and Linial 2005, 2006; Austin et al. 2007; Lam et al. 2010; Fortelny et al. 2015; Sharma et al. 2016). In 2003, analysis of six diverse proteomes identified numerous overrepresented C-terminal tripeptide pattern sequences (Gatto and Berg 2003). In 2007, sequences for seven eukaryotic proteomes were analyzed and identified several known true positives and several new anchored C-terminal tripeptide minimotif that were statistically enriched (Austin et al. 2007). In ProTeUS, a set of the over-represented unanchored C-terminal patterns was identified (Bahir and Linial 2005). Although these studies give insight into the potential for modular functions for C-termini, they examined a small fraction of human proteome. In 2015, a comprehensive analysis of the human proteome identified overrepresented sequence patterns up to 10 amino acids long with up to five degenerate positions (Sharma et al. 2016). Published molecular functions for 13% of human C-termini are known and additional functional inferences were based on rodent proteome minimotifs and matches to C-terminal consensus sequences (Sharma et al. 2016). The C-terminome datbase of known C-terminal minimotfs can serve as a benchmark and new sequence patterns are a foundation to explore and discover functions and targets for new C-terminal minimotifs.

C-termini databases

Known C-terminal minimotifs can be specifically identified from larger minimotif databases, such as in MnM and ELMs (Lyon et al. 2015; Gouw et al. 2018). Likewise, potential C-terminal minimotifs can be predicted from comprehensive peptide signatures (CoPS), and MOTIPS, utilities built to identify the signature sequences throughout a given protein sequence (Prakash et al. 2004; Lam et al. 2010). More recently, several databases have emerged that are dedicated to the C-terminus such as the C-terminome, TopFIND, and ProTEUS (Table 3) (Sharma et al. 2016). Although not specifically designed to the C-terminus, many specialized databases focus on specific molecular functions, such as O-Glycbase, PHOSIDA, Phosho3D, PhosphoSite-Plus, dbPTM, Human Histone modification databases, MEROPS, CutDB, and TOPPR for proteolytic sites; PDZBase and Binding DB for binding minimotifs (Gupta et al. 1999; Beuming et al. 2005; Igarashi et al. 2007; Liu et al. 2007; Zanzoni et al. 2007; Gnad et al. 2011; Colaert et al. 2013; Lu et al. 2013; Rawlings et al. 2014; Hornbeck et al. 2015; El Kennani et al. 2017).

Table 3.

List of minimotifs based databases.

Web-applications	C-term only	Species	Type of dataset	# of functions	Reference
C-terminome	Yes	Homo sapiens	C-terminal minimotifs and Proteins	3546 verified and ⁓9 million predicted	Sharma et al. (2016)
TopFIND3.0	No	5 species	Neo C-termini as from mass-spectrometry	128 modifications, 130,182 proteolytic ends	Fortelny et al. (2015)
ProTeus	Yes	None specified	Signature groups in Swiss-Prot	Unknown (all predicted)	Gnad et al. (2012)
ELM	No	All	Short linear motifs anywhere in proteins	>275 motif classes and >300 motif instances	Dinkel et al. (2012)
MnM	No	All	Minimotifs anywhere in proteins	1,060,436 instances	Lyon et al. (2018)
DILIMOT	No	None specified	Discovery-based tool to identify motifs in a protein sequence based on homology	None specified	Neduva and Russell (2006)
Proteolysis
MEROPS	No	Many	Peptidase, peptidase substrates	92,216 cleaved products; 908,326 peptidases	Rawlings et al. (2014)
CutDB	No	All	Proteolytic events, both actual and predicted	3070 proteolytic events; 470 proteases	Igarashi et al. (2007)
TOPPR	No	human and mouse	Published proteolytically processed sites in human and mouse proteins based on MS/MS	Proteolysis sites	Colaert et al. (2013)
Binding
PDZBase	Yes	None specified	PDZ-domain interaction with protein s	300 interactions	Beuming et al. (2005)
Binding DB	No	None specified	Protein: drug molecules interactions	1,391,403 binding data, for 7225 protein targets and 621,060 small molecules	Liu et al. (2007)
Post-translational modification0073
SwissProt	No	All	PTM, protein sequences	556,196 entries on protein sequences	Boeckmann et al. (2003)
O-Glycbase	No	All	O-linked glycosylation sites	242 glycoprotein entries	Gupta et al. (1999)
PHOSIDA	No	9 species	Phosphorylation, acetylation and N-glycosylation site database	more than 60,000 sites	Gnad et al. (2011)
Phospho3D	No	All	3D structures for phosphorylation site database	5314 phosphorylation sites in 1805 sequences	Zanzoni et al. (2007)
PhosphoSite-Plus	No	Mammal	PTMs	330,000 non-redundant	Hornbeck et al. (2015)
dbPTM	No	None specified	Integrated resource from 14 public databases on PTMs	610,037 experimental and 546,911 putative PTM sites	Lu et al. (2013)
Human Histone Modification Database	No	Human	Histone modification database.	43 modification types	Zhang et al. (2010)

The C-terminome database contains the known experimentally verified C-terminal minimotifs, inferences based on the rodent C-terminal minimotifs, predictions of molecular functions for C-terminal sequences based on consensus sequences, and bioinformatic predictions of C-terminal sequences and patterns that are overrepresented in the human proteome when compared to that of random proteomes (Sharma et al. 2016). C-terminome can be searched or browsed for any human protein or C-terminal minimotifs.

The Terminus Oriented Protein Function Inferred Database (TopFIND) includes some published PTMs literature and 130,182 proteolytically derived neo-C-terminal ends (Fortelny et al. 2015, p. 0). The neo-C-terminal ends arise by in vivo proteolytic processing are identified through a mass-spectrometry-based approach, C-TAILS. TopFIND version 3.0 includes the data on five diverse species. The ProTEUS database has both predicted anchored and non-anchored C-terminal sequences allowing one position of degeneracy in the last 10 amino acids of proteins (Bahir and Linial 2005; Gnad et al. 2012). ProTEUS recognizes the terminal minimotifs based on a small set of approximately 100 known functional C-terminal minimotifs.

Carboxyl termini in diseases and medicine

The carboxyl end of proteins and C-terminal minimotifs are involved in several cellular processes including translation termination, protein folding, protein turnover, protein trafficking, and cellular signaling (Birrane et al. 2003; Chong et al. 2005; Millhouse and Manley 2005; Coblitz et al. 2006; Chinnadurai 2007; Wang and Malbon 2012; Lübkemeier et al. 2013; Sucic et al. 2013; Krishnan et al. 2014; Sun et al. 2014; Liu et al. 2015; Patil et al. 2015; Ryan et al. 2016; Wu et al. 2016). Consequently, the deletion of the C-terminus could compromise any of these processes by losing a C-terminal motif or creating a new one (Table 4). For instance, the 104-amino acid C-terminal truncation in the human ether-a-go-go-related gene (HERG) channels causes long QT syndrome type 2 (LQT2), a cardiac disorder (Kupershmidt et al. 2002). Because of this truncation, a gain-of-function endoplasmic reticulum retention signal RGR > is exposed reducing the trafficking of HERG to the cell surface where it functions. Several other examples are provided.

Table 4.

C-terminal minimotifs in human disease.

Protein	Minimotif	General function	Disease	Reference
HERG channelsa	RGR>	Traffic	Long QT syndrome type 2	Kupershmidt et al. (2002)
Rhodopsin	VAPA>	Traffic	Vision loss	Deretic et al. (2005)
MNK	LL>	Traffic	Menkes disease	Petris et al. (1998)
NKCC2	LL>	Traffic	Bartter syndrome	Zaarour et al. (2012)
Bile salt export pump	YYKLV>	Traffic	Primary familial intrahepatic cholestasis type 2	Lam et al. (2012)
SANS gene	TEL>	Bind	Usher syndrome type I	Reiners et al. (2006)
Claudin-16	TRV>	Bind	Nephrocalcinosis	Müller et al. (2003)
NaV1.5 channel	SIV>	Bind	Cardiac disease	Shy et al. (2013)

^aGain of function.

Apolipoprotein E (Apo E) binds amyloid b peptides of plaques and forms neurofibrillary tangles with hyperphosphorylated tau. Amyloid plaques and neurofibrillary tangles are the hallmarks of Alzheimer’s disease (AD), a progressive neurodegenerative disorder and ApoE are a risk factor for late-onset disease. Chymotrypsin-like serine proteases preferentially cleave the E4 haplotype of ApoE over the E3 haplotype. The resulting ApoE4 fragment (d272–299) is neurotoxic in the transgenic mice leading to neuronal pathology in AD (Harris et al. 2003). The deletion of the C-terminus of ApoE4 unmasks a lipid-binding function in ApoE (244–272) and also interacts with amyloid beta peptides.

The binding of calmodulin to the C-terminus of aquaporin-O (residues 225–263 of a 263-residue protein) inhibits water permeability leading to the progression of cataract and vision impairment (Schey et al. 1999; Gold et al. 2012). Protein kinase A phosphorylates Ser-235 of aquaporin-O preventing its interaction with CaM. Thus, a mutation that eliminates Ser-235 can be detrimental to the vision (Lin et al. 2007).

Missense mutations can disrupt a C-terminal minimotif critical for protein function leading to diseases (Table 4). For instance, mutations in the VAPA > trafficking minimotif in Rhodopsin cause autosomal dominant retinitis pigmentosa (Deretic et al. 1998, 2005). Rhodopsin interacts with ADP-ribosylation factor 4 (ARF-4), a rhodopsin transport carrier, through its VAPA > minimotif. VAPA > is required for Rhodopsin trafficking to the photoreceptor rod outer segments. Some other examples are provided.

Several examples of mutations in a PDZ domainbinding minimotif in diseases exist. For instance, frameshift mutations in the SANS gene cause Usher syndrome type I, an autosomal recessive genetic neurodegenerative condition with hearing and vision loss (Verpy et al. 2000; Weil et al. 2003). SANS protein contains a PDZ domain-binding motif that interacts with the PDZ domain of the harmonin and scaffold protein (Reiners et al. 2006). Compromise of similar protein networks based on PDZ domain binding in photoreceptor cells may also lead to Usher syndrome. Similarly, NaV1.5 channel proteins contain an SxV > PDZ binding minimotif, that binds the PDZ domains of syntrophins and SAP97 in cardiomyocytes (Shy et al. 2014). The mutation of this motif mislocalizes the NaV1.5 channel and causes cardiac disease. A mutation in the PDZ binding motif of Claudin-16 abrogates its interaction with the PDZ domain of ZO-1, a scaffolding protein. This mutation mislocalizes Claudin-16 to lysosomes instead of tight junctions in kidney epithelial cells causes familial hypomagnesemia, hypercalciuria, and nephrocalcinosis (Müller et al. 2003) (Table 3).

Mutations in some C-terminal di-leucine trafficking minimotifs cause Bartter’s syndrome (Zaarour et al. 2012). Di-leucine minimotifs are evolutionarily conserved in solute carrier family 12 (SLC12A) members. Electroneutral cation-chloride co-transporter family members are therapeutic drug targets. One example of this family of proteins is Na-K-2Cl co-transporter (NKCC2) proteins. NKCC2 proteins can cause type I Bartter’s syndrome, a life-threatening disease caused by the abnormal electrolyte absorption. The conserved C-terminal di-leucine motifs in NKCC2 gets glycosylated, aid in protein maturation, and act as ER export signals for proteins. However, the LL mutants of NKCC2 get trapped in ER (Zaarour et al. 2012). An additional ¹⁰⁴⁸LI¹⁰⁴⁹ motif was also identified in NKCC2 that disrupts ER exit. Similarly, a defect in dileucine minimotif of MNK can cause Menkes disease, an X-linked recessive disorder, and disrupting copper homeostasis. A mutation in the ¹⁴⁸⁷LL¹⁴⁸⁸ minimotif in the transmembrane copper-transporting P-type ATPase (MNK) protein blocks trafficking of MNK from the plasma membrane to the trans-Golgi network (Petris et al. 1998). MNK is a Cu^þ2 importer in the secretory pathway.

Several instances of C-terminal peptides (some modified) are at the stage of preclinical testing for new therapeutics (Schiöth et al. 2006; Tsubaki et al. 2013). For example, although the molecular mechanism of activation is not clear, it is known that the KPV > peptide of α-melanocyte stimulating hormone (α-MSH) contains both strong inflammatory properties and have antipyretic effects (Richards and Lipton 1984; Schiöth et al. 2006; Luger and Brzoska 2007).

Conclusions

Cellular processes, such as alternative splicing and proteolysis increase the number of C-terminal ends in the proteome. The negatively charged, often disordered and flexible C-termini have C-terminal minimotifs with a bind, traffic, or PTM function (Sharma et al. 2016). Given that 13% of the human proteome has known C-terminal motifs, the C-termini may have a generalizable function in all proteins. The further study of C-terminal minimotifs will most certainly enhance our understanding of hereditary disease etiology. Mutations in C-terminal minimotifs causes vision loss, Menkes disease, Usher syndrome, and other diseases (Petris et al. 1998; Verpy et al. 2000; Deretic et al. 2005; Puckerin et al. 2016). There are now several web systems for C-termini, the most comprehensive for function being the C-terminome (Sharma et al. 2016).