New beginnings and new ends: methods for large-scale characterization of protein termini and their use in plant biology.

Dynamic regulation of protein function and abundance plays an important role in virtually every aspect of plant life. Diversifying mechanisms at the RNA and protein level result in many protein molecules with distinct sequence and modification, termed proteoforms, arising from a single gene. Distinct protein termini define proteoforms arising from translation of alternative transcripts, use of alternative translation initiation sites, and different co- and post-translational modifications of the protein termini. Also site-specific proteolytic processing by endo- and exoproteases generates truncated proteoforms, defined by distinct protease-generated neo-N- and neo-C-termini, that may exhibit altered activity, function, and localization compared with their precursor proteins. In eukaryotes, the N-degron pathway targets cytosolic proteins, exposing destabilizing N-terminal amino acids and/or destabilizing N-terminal modifications for proteasomal degradation. This enables rapid and selective removal not only of unfolded proteins, but also of substrate proteoforms generated by proteolytic processing or changes in N-terminal modifications. Here we summarize current protocols enabling proteome-wide analysis of protein termini, which have provided important new insights into N-terminal modifications and protein stability determinants, protein maturation pathways, and protease-substrate relationships in plants.

Introduction

Variation at the protein level underpins the complexity, differentiation, and resilience of biological systems. Diversifying mechanisms result in multiple protein molecules with distinct sequences and/or chemical modifications, termed proteoforms, arising from a single gene (Smith and Kelleher, 2013). Many proteoforms can be distinguished by their unique termini (Fig. 1), including proteoforms generated by alternative RNA splicing (Cheng and Tu, 2018; Laloum ; Szakonyi and Duque, 2018) and alternative translation initiation (Willems ). On the protein level, proteolytic processing by a large range of endo- and exoproteases generates distinct proteoforms that are defined by their new protease-generated neo-N- and neo-C-termini (Huesgen and Overall, 2012; Klein ). Also allelic variants can exhibit point mutations or sequence variations leading to differential splicing, new translation starts (Xu ), and altered proteolytic processing (Fig. 1). Finally, site-specific co- and post-translational protein modifications create a dazzling space of potential proteoforms that may differ from each other in subcellular localizations, interactions, activity, and function (Smith and Kelleher, 2013).

Fig. 1.

Possible origins of diverse proteoforms derived from one gene. A multitude of mechanisms can influence the appearance of a specific proteoform at the transcriptional (top panel), translational (middle panel), and co-/post-translational level (bottom panel). Profiling N-termini and C-termini enables the differentiation of these various proteoforms. Alternative translation initiation (AUG.1 and AUG.2) will lead to an alternative N-terminus (N1.1 and N1.2), as does N-terminal methionine excision (NME). Further, NME allows for subsequent N-terminal modifications, such as N-terminal acetylation (orange) or N-terminal myristoylation (purple), increasing the number of proteoforms and their specific N-termini. Proteolytic processing, either endo- or exoproteolytic, represents an additional important mechanism for the introduction of new proteoforms on the post-translational level which can be assessed by either N- or C-termini profiling. Numbering of N- and C-termini corresponds to exon number from which the specific proteoform has derived. Subnumbering represents alternative N- or C-termini originating from the same exon.

Dynamic control of proteoform stability and abundance assures protein quality control and regulates many cellular processes. An important determinant of proteoform stability in vivo is the identity and modification of its N-terminal amino acid, which is summarized by the N-end rule (Bachmair ; Gonda ; Tobias ; Potuschak ; Graciet ; Holman ; Dissmeyer ; Dissmeyer, 2019; Varshavsky, 2019). In the cytosol of eukaryotic cells, proteins exposing destabilizing N-termini can be recognized by E3 ubiquitin ligases called N-recognins, and targeted for rapid degradation by the ubiquitin–proteasome system. PROTEOLYSIS (PRT) 1 and 6 are the currently known N-recognins that target substrates initiated with primary destabilizing residues for degradation (Faden , 2019; Naumann ; Dong ; Reichman and Dissmeyer, 2017; Mot ; Vicente ). In this context, N-terminal arginylation has been recognized as a post-translational modification that targets proteins for degradation via the N-end rule ubiquitin–proteasome pathway (Dissmeyer, 2017, 2019; White ; Dissmeyer ), but also the autophagy–lysosomal pathway (Kim ; Cha-Molstad , 2018). In animals, N-terminal arginylation is more and more considered to generate a bimodal degron that operates in both autophagic and proteasomal proteolysis (Yoo ). N-terminal arginylation requires the activity of specific arginyl-tRNA protein arginyltransferases (or short arginyltransferases/arginine transfer enzymes, ATEs). Arabidopsis harbors two genes encoding putative ATEs, of which at least ATE1 can transfer an Arg to N-termini, exposing the secondary destabilizing acidic, negatively charged amino acids Asp, Glu, and dioxygenated or trioxygenated Cys (White ). Besides the chemical state of the N-terminal amino acid residue, the efficiency of the arginyl transfer also strongly depends on the identity and properties of the subsequent residues (Wadas ). The immediate result of this modification is the transformation of N-termini with secondary destabilizing residues into N-termini exposing the primary destabilizing Arg, enabling their rapid degradation (White ; Dissmeyer ). Also N-terminal acetylation is a dynamically regulated, ubiquitous protein modification in eukaryotes that affects protein–protein interactions, subcellular localization, protein folding, stability, and degradation via the Ac/N-end rule pathway (Pesaresi, 2003; Zybailov ; Varshavsky, 2011; Bienvenut ; Gibbs, 2015; Rathore ; Linster and Wirtz, 2018). In plants, N-terminal acetylation by N-terminal acetyltransferases (NATs) occurs co- or post-translationally in the cytosol, the Golgi apparatus, and the chloroplasts of plants, and may be dynamically regulated by phytohormones (Linster ; Linster and Wirtz, 2018). Since acetylated N-termini can be recognized as N-degrons, rapid and correct folding of the acetylated N-terminus into the protein’s inner structure is important for its stability. In contrast, slow or incorrect folding will expose the acetylated N-terminus, leading to its recognition and degradation (Varshavsky, 2011). N-terminal acetylation contributes to diverse aspects of a plant’s life, such as flowering time, immune response, or drought stress response (Kapos ; Linster ; Xu ). However, both acetylation and arginylation also occur midchain at internal Lys or Asp and Glu residues, respectively, distant from the N-terminus (Eriste ; Wang ; Hoernstein ).

MS-based proteomics has become the method of choice for unbiased proteome analysis, enabling proteome-wide protein identification, quantification, and profiling of their modifications and interactions (Aebersold and Mann, 2016). In standard bottom-up workflows, proteins are first proteolytically digested into predictable peptides more amenable to chromatographic separation and analysis by tandem MS (MS/MS). Peptide fragmentation spectra are recorded and matched to theoretical spectra, computationally predicted from a sequence database, constrained by the sequence specificity of the protease used for digestion. However, N- and C-terminal peptides constitute only a very minor fraction of the extraordinarily complex peptide mixture generated by proteome digest and are therefore rarely identified. Furthermore, neo-termini generated by endogenous proteolytic processing match the sequence specificity of the digestion protease only on one side and are not considered during standard search spectra to sequence matching (Niedermaier and Huesgen, 2019). Hence, selective enrichment and dedicated data analysis are necessary for comprehensive profiling of N- and C-termini and their modifications (Klein ).

Three principal strategies achieve proteome-wide enrichment of protein termini. First, the most basic strategy uses proteases with different sequence specificity to generate N- or C-terminal peptides containing a predictable number of primary amines, namely α-amines at peptide N-termini and ε-amines in Lys residue side chains, that differ from the remaining digest-generated peptides. Such a differential amine content results in differential charge states at low pH that can be used for the separation of target N- or C-terminal peptides from the majority of other peptides by strong cation exchange (SCX) chromatography (Dormeyer ). Secondly, positive selection workflows directly capture and enrich target terminal peptides while discarding all other digest-generated peptides (Fig. 2A). This is typically achieved by modification of unblocked, free α-amines or carboxyl groups with a purification tag prior to digestion (Table 1) or by affinity enrichment with antibodies against specific N-terminal modifications. Thirdly, negative selection strategies take the opposite approach and achieve enrichment by selective depletion of undesired internal peptides (Fig. 2B; Table 2). In these workflows, primary amines must be chemically modified on the protein level, enabling stable isotope labeling by amine reactive isotope labeling reagents such as formaldehyde (Boersema ), isobaric tags for relative and absolute quantification (Ross ), or tandem mass tags (TMTs, Thompson ) for reliable comparative analysis. Proteome digest with specific proteases then exposes new reactive α-amines at non-target peptides that are used for selective depletion while modified terminal peptides remain inert. Negative selection thus simultaneously enriches both naturally modified and in vitro modified (in vivo free) terminal peptides, facilitating discovery and comprehensive profiling of endogenous protein terminal modifications. Here, we review the different methodological solutions for proteome-wide enrichment of N- and C-termini enrichment and discuss recent and emerging applications in plant biology.

Fig. 2.

Basic workflow for the enrichment of protein N-terminal peptides by positive or negative selection. Proteolytic processing creates a new N-terminus (Nt) and a new C-terminus (Ct), also termed neo-termini (given in red). In positive selection strategies, neo-N-termini are selectively tagged with an affinity tag (blue oval) while for negative selection (bright green) the neo-N-termini are blocked by dimethylation (red triangle) prior to digestion. After digestion, positive selection strategies utilize an affinity capture of the added tag. Undesired digestion-generated peptides, as well as naturally blocked N-terminal peptides (e.g. acetylated N-termini, orange) and C-terminal peptides are washed out. The captured N-terminal peptides are subsequently released from the affinity column. Negative selection strategies modify all unblocked N-termini, exposing primary amines before digest. Undesired digestion-generated peptides are then tagged and/or captured for depletion. The desired neo-N-terminal peptides and naturally blocked N-terminal peptides are collected in the flow through.

Table 1.

Protocols for enrichment of N- or C-termini by positive selection

Name	Target peptides and enrichment principle	Specific requirements	Starting materiala	Advantages and disadvantages	Referencesb
Subtiligase biotinylation	Unblocked and protease-generated neo-N-termini; tagging by enzymatic biotinylation	Subtiligase, TEV protease, biotinylated peptide esters	30–300 mg	+ Direct enrichment of neo-N-termini – sequence preference of subtiligase – high amount starting material – not compatible with chemical stable isotope labeling – subtiligase not commercially available	Mahrus et al. (2008); Wiita et al. (2014); Weeks and Wells (2018)
ProC-TEL	Unblocked and protease-generated neo-C-termini; tagging by enzymatic biotinylation	Carboxypeptidase Y, biotinylated peptide ester	200–300 µg per condition	+ Direct enrichment of neo-C-termini + compatible with chemical stable isotope labeling – tricky reaction condition poses challenge for efficient labeling	Xu et al. (2011); Duan et al. (2016); Duan and Xu (2017)
Chemical biotinylation of N-termini	Unblocked and protease-generated neo-N-termini; tagging by chemical biotinylation	NHS-SS-biotin, neutravidin resin	1–10 mg	+ Selective tagging of neo N-termini – high amount starting material – not compatible with chemical stable isotope labeling – challenging differential α- and ε- amine labeling	Timmer et al. (2007); Timmer and Salvesen (2011)
Chemical biotinylation of C-termini	Unblocked and protease-generated neo-C-termini; biotinylation using oxazolone chemistry	Synthetic Arg-NHNH-biotin-peptide	10 µg per condition	+ Selective tagging of neo-N- termini + low amount of starting material – not compatible with chemical stable isotope labeling	Liu et al. (2015)
N-CLAP	Unblocked and protease-generated neo-N-termini; tagging by chemical biotinylation	NHS-SS-biotin, neutravidin resin	2 mg	+ Selective tagging of neo N-termini + selective deprotection of α-amine using well-established chemistry – peptides one residue shorter than true terminus – not compatible with chemical stable isotope labeling	G. Xu et al. (2009); Xu and Jaffrey (2010)

Starting material refers to the amount of extracted and purified protein/proteome.

References in italics are step-by-step protocols.

Table 2.

Protocols for enrichment of N- or C-termini by negative selection

Name	Target peptides and enrichment principle	Specific requirements	Starting materiala	Advantages and disadvantages	Referencesb
COFRADIC	Modified and neo-N- and C-termini; enrichment by diagonal reverse phase chromatography	Capillary HPLC system with fraction collector	1–3 mg	+ proven in multiple publications + versatile enrichment of N- and C-termini – labor intensive – extensive MS time	Gevaert et al. (2003) Staes et al. (2008, 2011); van Damme et al. (2010); Tsiatsiani et al. (2013); Tam et al. (2015); Willems et al. (2017)
ChaFRADIC	Modified and neo-N-termini; enrichment by diagonal SCX chromatography	Capillary HPLC system with fraction collector	50–200 µg	+ proven in multiple publications + less complex than COFRADIC + low amount of starting material – labor intensive	Venne et al., 2013); Carrie et al. (2015); Venne et al. (2015)
ChaFRAtip	Modified and neo-N-termini; enrichment by tip-based diagonal SCX chromatography	SCX beads	4.3 µg per condition	+ lowest amount of starting material – labor intensive – prone to handling errors	Shema et al. (2018)
TAILS	Modified and neo-N-termini; other peptides removed by covalent binding to polymer and ultrafiltration	HPG-ALD polymer, size exclusion filter	0.1–1 mg per condition	+ robustness proven by independent application in multiple laboratories – requires moderately expensive patented aldehyde polymer	Kleifeld et al. (2010); Kleifeld et al. (2011); Köhler et al. (2015a, b); Rowland et al., (2015); H. Zhang et al. (2015, 2018); Demir et al. (2018)
C-TAILS	Modified and neo-C-termini, other peptides removed with polymer and ultrafiltration Without amine blocking prior to C-terminal amidation	PAA polymer, labeling reagents, size exclusion filter	1.5–2 mg per condition 80 µg per condition	– labor- and loss-intensive	Schilling et al., (2010); (Schilling et al. (2011); Solis and Overall (2018); Y. Zhang et al. (2015, 2018)
Biotinylation of internal peptides	Modified and neo-N-termini, other peptides removed by covalent binding to NHS-activated Sepharose	NHS-biotin, streptavidin column, amine-scavenging beads	50 µg	+ simple procedure using standard chemicals – non-specific losses of N-termini on Sepharose beads	McDonald et al. (2005)
NHS bead capture	Modified and neo-N-termini, other peptides removed with NHS-activated Sepharose	NHS-activated Sepharose	50 µg	+ simple procedure using standard chemicals – non-specific losses of N-termini on Sepharose beads	McDonald and Beynon (2006)
NRich	Modified and neo-N-termini, other peptides removed with NHS-activated Sepharose and ultrafiltration	NHS-activated Sepharose beads	5 mg per condition	+ simple procedure – high amount of starting material – non-specific interactions of N-termini with bead material	Yeom et al. (2017)
Charge reversal	Modified and neo-N-termini, other peptides removed by disulfonate modification + SCX chromatography	SCX-packed tips or HPLC system with SCX columns and fraction collector	1.5–3 mg per condition	+ flexible format allows enrichment from gel slices	Lai et al. (2015); Lai and Schilling (2017)
HYTANE	Modified and neo-N-termini, other peptides removed by hydrophobic tagging C18 trap depletion	C18 trap columns	100 µg per condition	+ well-established chemicals – removal of excess aldehyde can be challenging	Chen et al. (2016a)
PTAG	Modified and neo-N-termini, other peptides removed by addition of phosphotag and TiO2 depletion	TiO2 affinity columns or beads	100 µg per condition	+ Chemistry used for depletion established in many laboratories – depletion sensitive to sample amount and reaction conditions	Mommen et al. (2012)
C-PTAG	Modified and neo-C-termini, other peptides removed by addition of phosphotag and TiO2 depletion	TiO2 affinity columns or beads	200 µg per condition	+ TiO2 chemistry used for depletion well established – limited to LysC for digestion – challenging differential α-amine labeling	Chen et al. (2016b)
StagAu	Modified and neo-N-termini, other peptides removed by sulfhydryl coupling to gold particles	Custom-made gold-covered graphene particles	10 µg per sample	+ low starting material – gold nanoparticles not readily available	Li et al. (2016)

Starting material refers to amount of extracted and purified protein/proteome

References printed in italics are step-by-step protocols, and those in bold list applications in plants.

Charge-based enrichment of protein C-termini and modified N-termini

Endogenously modified protein N-termini can be enriched by exploiting the differential primary amine content (α-amine plus basic residues Arg, Lys, and His) in peptides generated by digestion with trypsin, which cleaves after Arg and Lys residues. Modified N-terminal peptides (e.g. acetylated and propionylated) lack a free α-amine, while C-terminal peptides lack a basic residues after tryptic digest. C-terminal and modified N-terminal peptides thus mostly carry a single charge at pH 2.7, allowing separation from the majority of doubly charged tryptic peptides by SCX chromatography (Dormeyer ).

Charge-based enrichment is also used in stable-isotope protein N-terminal acetylation quantification (SILProNAQ), a dedicated protocol for the enrichment, identification, and quantification of acetylated N-termini (Bienvenut ). Free α-amines are modified by stable isotope-marked (D3)-acetyl using the reagent N-acetoxy-[2H3]succinimide to distinguish between in vivo acetylated and in vitro deutero-acetylated termini, respectively. O-Acetylation of Ser, Thr, and Tyr residues occurring as a side reaction is reversed by the addition of hydroxylamine before tryptic digest and enrichment of acetylated N-termini by SCX chromatography. A customized SILProNAQ data analysis pipeline, including the EnCOUNTer tool, automatically parses database search results, quantifies matching in vivo and in vitro acetylated N-termini, and calculates the degree of endogenous acetylation (Bienvenut ).

However, both charge-based protocols achieve only incomplete enrichment and are limited to selected digestion proteases. This excludes identification of the subset of termini with cleavage sites that result in too long, too short, or generally unfavorable peptides (Niedermaier and Huesgen, 2019). Internal peptides with a high content of acidic residues may co-elute with acetylated peptides, and N-terminal peptides with a high number of basic residues, for example by missed cleavages or containing His, may elute later within the bulk of internal peptides (Gorman and Shiell, 1993).

Enzymatic tagging of N- or C-terminal peptides for positive selection

A pioneering method for positive selection of N-termini utilizes subtiligase, a peptide ligase rationally engineered from the protease subtilisin BPN' (Chang ), to ligate a synthetic peptide ester bearing a biotin tag and a cleavage site for the Tobacco etch virus (TEV) protease to the α-amine of naturally unmodified and protease-generated neo-N-termini (Mahrus ). After proteome digestion, tagged N-terminal peptides are captured by immobilized avidin, subjected to stringent washes, and released by TEV protease cleavage (Mahrus ). Diverse synthetic peptide esters have been created for various purposes, including more water-soluble esters to push the equilibrium towards the labeled reaction (Yoshihara ). However, typically only 10–15% of α-amines are labeled, presumably due to poor accessibility of N-termini under native conditions (Wiita ). Subtiligase exhibits a preference for small amino acids at the P1' position and large hydrophobic or aromatic amino acids at the P2' position, and thus introduces a bias in N-terminal labeling (positions follow the Schechter and Berger nomenclature with protease substrate positions towards the N-terminus of the hydrolyzed peptide bond denoted P1, P2, P3, etc. and positions towards the C-terminus P1', P2', P3', etc. Schechter and Berger, 1967). To address these issues, stabiligase, a subtiligase variant tolerating denaturing conditions (Chang ), and a subtiligase mutant library with altered P1' and P2' selectivity have been created (Weeks and Wells, 2018). Rational selection of a combination of subtiligase variants now allows effective ligation of almost any protein N-terminal sequence (Weeks and Wells, 2018).

Enzyme-mediated biotinylation has also been used to enrich C-terminal peptides by positive selection. Profiling protein C-termini by enzymatic labeling (ProC-TEL, Xu ) utilizes the known transpeptidase activity of carboxypeptidase Y (CPY) at pH >11, where the hydrolytic activity of CPY is inhibited. First, all carboxyl groups are esterified by methanolic hydrochloric acid, followed by selective deprotection of side chain esters. CPY is then employed as a transpeptidase to selectively add a biotin tag to protein C-termini. After digestion, tagged C-terminal peptides are captured by immobilized avidin, purified by stringent washes, and eluted with 50% acetonitrile, 0.1% trifluoroacetic acid (TFA) (Duan ).

Selective chemical tagging of N- or C-terminal peptides for positive selection

Selective biotinylation of free protein N-termini can also be achieved by chemical modification (Timmer ). First, cysteine residues are blocked by iodoacetamide (IAA) and lysine residues converted to homoarginine by reaction with O-methylisourea under high pH conditions before α-amine labeling with sulfo-NHS-SS-biotin. Proteins are digested and tagged N-terminal peptides are captured with immobilized streptavidin and released after washes by reduction of the disulfide linker with DTT.

An alternative protocol for enrichment of N-terminal peptides by chemical tagging is N-terminomics by chemical labeling of the α-amine of proteins (N-CLAP, G. Xu ). Here, the Edman sequencing reagent phenyl isothiocyanate is used to modify all amines. Subsequent acidification with TFA causes an intramolecular cyclization and cleavage of the peptide bond between the first and second amino acid, but not at modified ε-amines. The resulting unblocked α-amines are tagged with NHS-SS-biotin prior to proteome digestion. N-terminal peptides are captured by immobilized avidin and released after washes by reduction of the disulfide linker. The N-terminal peptides identified by MS/MS analysis are one amino acid shorter than naturally occurring N-termini, which needs to be considered during computational data analysis and interpretation.

A method for selective chemical biotinylation of protein C-termini has also been developed (Liu , 2015). This method employs the established oxazolone chemistry (Liu ) to conjugate Arg–Arg dipeptide-linked biotin to protein C-termini. After digest, biotinylated C-terminal peptides were captured with streptavidin beads and released by boiling in 50% acetonitrile and 1% TFA before identification by MS.

Enrichment and identification/detection of modified N-termini with specific antibodies

Antibodies provide another approach to the enrichment of modified protein N-termini by positive selection (Fig. 3). The first class of N-terminal modifications to be detected and/or enriched were arginylated proteins (Table 3). Anti-N-arginyl- antibodies can be used in two ways (Fig. 3). First, they are applied in pipelines to positively enrich N-terminally arginylated proteins by capture and at the same time to deplete non-arginylated protein from the sample (Fig. 3A–F). The protocols make use of two peptides initiated with Arg–Asp or Arg–Glu, respectively, to generate ‘pan-arginylation’ antibodies (Table 3). Their random peptide backbone is unrelated to any known sequence but consists of a highly immunogenic stretch of five bulky, charged residues (composed of the amino acids Arg, Asp, Glu, and Lys), followed by seven small uncharged residues with predicted low immunogenicity (of the group of Asn, Gln, His, Ser, and Val) (Wong ; Kashina, 2015). Therefore, these antibodies should generally recognize proteins initiating with either of these two arginylated residues rather than specific targets. To reduce background from byproduct antibodies reactive against shorter versions of the peptide, for example lacking the N-terminal Arg, the specific antibodies are negatively selected against ‘scrambled’ synthetic peptides with similar amino acid content but different sequence (Fig. 3A–D) (Wong ; T. Xu ; Kashina, 2015). Challenges when working with N-terminal arginylation are potential cleavage of Arg due to the proteases used in sample preparation and identification in bottom-up proteomics, as standard database search parameters render peptides bearing an additional Arg as modification invisible. For identification, variable modification of N-terminal Asp, Glu, and Cys residues by arginylation (Ebhardt, 2015; White ) or dimethyl-arginylation (Hoernstein ) needs to be considered.

Fig. 3.

Enrichment and detection via N-terminally specific antibodies. Antibodies produced against N-terminal modifications can be used for enrichment and/or detection of modified proteins in various protocols and assays. (A) Antibodies are raised against peptides mimicking the N-termini of the desired target proteins or, as in the case of the ‘pan-arginylation’ antibodies, a random, highly immunogenic sequence. (B) Antisera are derived from different, subsequent bleedings to acquire a range of differentially performing antibodies and select ideal reagents. The resulting antisera need to be purified against unspecific (C) and specific peptides (D) lacking or showing the N-terminal modification (arginylation/additional Arg residue at the N-terminus of the natural protein sequence, acetyl- or formyl-groups). (E) Positive selection of N-terminally modified proteins or peptides by capture with the specific antibodies and wash-out of non-target proteins/peptides, followed by elution for identification by MS (F). The same antibodies can be utilized in immunological assays such as western blots (G) or immunocytochemistry (not shown). Red, unspecific antibody populations; green, specific antibody populations; R, Arg; AcM, acetylated Met; fM, formylated Met; C, Cys included at the C-terminus of antigenic peptide to increase antigenicity.

Table 3.

Antibodies raised against N-terminal arginylated sequences used for enrichment or immunological detection.

Target	Use	Application	Sequence	Reference
Generic	Enrichment	Mass spectrometry, IP-MS	(R)D/EHKHANQHMSVC	Wong et al. (2007); T. Xu et al. (2009); Hoernstein et al. (2016)
Generic	Enrichment	Mass spectrometry, IP-MS	(R)DHKH	Saha and Kashina (2011)
Generic	Enrichment	Mass spectrometry, IP-MS	(R)D/EHKHANQHMSVC	Hoernstein et al. (2016)
Arg-CRT (calreticulin; (different epitope, see below)	Detection	Western blot, immunocytochemistry	(R)DPAIYFK	Decca et al. (2007)
Arg-BRCA1 (different epitope, see below)	Detection	Western blot	(R)DVEIQGHTSFC	Piatkov et al. (2012)
Arg-β actin	Detection	Dot blot (verification), western blot	(R)DDIAAL	Saha et al. (2012)
Arg-BiP (ER chaperone BiP, also known as GRP78 and HSPA5, heat shock 70 kDa protein 5)	Detection	Dot blot (verification), western blot, immunocytochemistry	(R) EEEDKKEDVGC	Cha-Molstad et al. (2015b); Jiang et al. (2016); Yoo et al. (2018)
Arg-CRT (calreticulin)	Detection	Western blot	(R)EPAVYFKEQ	Cha-Molstad et al. (2015b); Yoo et al. (2018)
Arg-PDI (protein disulfide isomerase)	Detection	Western blot	(R)DAPEEEDHVL	Cha-Molstad et al. (2015); Yoo et al. (2018)
Arg-CDC6	Detection	Western blot	(R)DEPTFKASPPK	Yoo et al. (2018)
Arg-BRCA1	Detection	Western blot	(R)DGEIKEDTSFA	Yoo et al. (2018)

The same antibodies enable identification and quantification of modified N-termini by classical dot or western blot and immunocytochemistry (Fig. 3G). For this, a much broader portfolio of different and highly specific antibodies was raised against peptides according to the sequences of N-termini of putative ATE target proteins with an additional Arg residue at the N-terminus (Table 3). Thus, peptides were mimicking N-terminally arginylated target proteoforms. Also these antibodies are generated by classical immunization, negatively selected using the non-arginylated peptides to eliminate non-specific IgGs, followed by positive affinity purification using the arginylated peptides (Piatkov ; Cha-Molstad , b). These antibodies have in the past exclusively been used to qualitatively determine the presence of targets against which they were raised. Similarly, antibodies specific for N-terminal acetylated proteoforms were raised against specific targets with the help of acetylated peptides (Table 4; Shemorry ). Specific antibodies have also been raised to detect selected protease-generated neo-N-termini and their modified (arginylated) versions. These must be raised individually for each target proteoform and were shown to be highly specific for the cleaved (and modified) versus uncleaved sites. This is usually calibrated by using either recombinant protein with or without the N-terminal modification or the uncleaved versus the cleaved form or by synthetic peptides in dot blots (Klecker and Dissmeyer, 2016). Ratios of the amounts of proteolysis product to precursor can be experimentally determined by comparing signals derived from an neo-N-terminus-specific antibody versus a generic one that detects the target irrespective of cleavage and/or modification. However, these highly specific antibodies are not yet applicable for general enrichment (Huesgen ).

Table 4.

Antibodies raised against N-terminally acetylated or formylated sequences used for immunological detection

Target	Use	Application	Sequence	Reference
MATα2	Detection	Dot blot (verification), western blot	Ac-MNKIPIKDLLNC	Hwang et al. (2010)
Cog1	Detection	Dot blot (verification), western blot	Ac-MDEVLPLFRDS	Shemorry et al. (2013)
Generic (MD-D2-eK-ha)	Detection	Dot blot (verification), western blot	f-MDIAIGTYQEK	Piatkov et al. (2015); Kim et al. (2018)

Negative selection of N- or C-terminal peptides by diagonal chromatography

A pioneering negative selection strategy, combined fractional diagonal chromatography (COFRADIC), enriches N-terminal peptides with two consecutive reversed phase (RP) chromatography runs (Gevaert ). First, primary amines are blocked by acetylation prior to proteome digestion, which may include differential stable isotope labels to distinguish between endogenously and in vitro acetylated N-termini. A first chromatography run separates digested peptides into >12 fractions. Each fraction is treated with 2,4,6-trinitrobenzenesulfonic acid (TNBS) to modify unblocked α-amines, resulting in increased hydrophobicity of digest-generated peptides. Each fraction is then subjected to a second chromatography run, where TNBS-modified peptides shift towards later retention times while N-terminal peptides are collected at unchanged retention times. Further refinements of the COFRADIC protocol included the pre-enrichment of acetylated N-terminal peptides by SCX chromatography and enzymatic recovery of digest-generated peptide α-amines that were blocked by spontaneous pyroglutamate formation from peptide N-terminal Glu and Gln residues (Staes ). Based on the depletion of internal peptides by SCX, a version of COFRADIC was developed that allows subsequent separation of N- and C-terminal peptides, increasing the identification for both types of terminal peptides (van Damme ). In short, after removing the majority of internal peptides by SCX chromatography, the N- and C-terminal peptides are separated and fractionated via a first RP chromatography run, after which each fraction contains a subset of N- and C-terminal peptides. Unblocked α-amines of the C-terminal peptides are then butyrylated to increase their hydrophobicity. Each fraction is then subjected to a second RP chromatography separation, where acetylated N-terminal peptides will elute at the same retention time as in the first run while the butyrylated C-terminal peptides will display a shift towards later elution times.

In a variation of the diagonal chromatography, charge-based fractional diagonal chromatography (ChaFRADIC) employs two consecutive SCX chromatography steps for enrichment of N-terminal peptides (Venne , 2015). N-termini are blocked by stable isotope labeling reagents prior to digestion and SCX fractionation into five fractions of peptides carrying defined charges (+1, +2, +3, +4, and >+4). Next, α-amines of digestion-generated peptides are deutero-acetylated, reducing the charge state by one. Each fraction is again separated by SCX chromatography, where N-terminal peptides retain their retention time while the modified digestion-generated peptides shift to earlier fractions. A simplified protocol replaced separation on SCX chromatography columns with disposable pipette tips containing SCX beads (Shema ). Termed ChaFRAtip, primary amines are stable isotopes labeled before digestion. Peptides are desalted, loaded on self-packed pipette tips containing SCX beads, and manually fractionated by stepwise elution with an increasing salt concentration. Peptides with digest-generated α-amines in each fraction are modified by acetylation, reducing their charge by one, and separated from the inert N-terminal peptides by a second, two-step elution from an SCX tip. The reduced scale enabled successful N-termini enrichment from as little as 4.3 µg per condition, or a total of 40 µg, with similar performance and reproducibility to HPLC-based ChaFRADIC analysis of a 5-fold larger sample amount.

Negative selection of terminal peptides using polymer-based depletion of non-target peptides

Terminal amine isotope labeling of substrates (TAILS) enriches N-terminal peptides by covalent binding of undesired internal and C-terminal peptides to a high molecular weight, hyper-branched polyglycerol (HPG) polymer functionalized with aldehyde groups (Kleifeld ). Primary amines are blocked by reductive dimethylation with formaldehyde isotopes or other amine-reactive stable isotope labeling reagents such as iTRAQ (Prudova ) or TMTs, enabling multiplexing of up to 10 samples (Klein ). Labeled proteomes are combined and excess reagents removed by protein precipitation before proteolytic digestion. Unblocked, digest-generated peptides with free α-amines are then captured by the aldehyde functionalized HPG polymer using cyanoborohydride. The polymer with bound peptides is removed by ultrafiltration, leaving only N-terminal peptides in the flow through for subsequent MS-based identification.

C-terminal amine-based isotope labeling of substrates (C-TAILS) follows a similar strategy to select for C-terminal peptides (Schilling ). However, the C-TAILS requires more sample preparation steps than its N-terminal counterpart. To avoid cross-reactivity, protein primary amines are first protected by reductive dimethylation before modification of the carboxyl groups at acidic residues and protein C-termini by EDC [1-ethyl-3-(3-dimethylaminopropyl)carbodiimide]-mediated condensation of ethanolamine. Proteome digestion is followed by a second labeling step of the primary amines, including the protease-generated α-amines of the C-terminal peptides. Finally, C-terminal carboxyl groups of digest-generated peptides are covalently linked to the primary amines of high molecular weight linear polyallylamine polymer (PAA) with EDC. Further optimization of the C-TAILS protocol improved amidation conditions, evaluated several alternative carboxyl and amine blocking reagents, and protected amines by acetylation instead of dimethylation (Y. Zhang ). Reportedly, amine protection prior to amidation can be omitted, saving a loss-intensive precipitation step and allowing the use of Lys-specific digestion enzymes (Y. Zhang ).

Alternative methods for negative selection of N-terminal peptides

A variety of different chemical reactions have been applied to exploit the digest-generated α-amine functionality for depletion of digest-generated peptides (Table 2). As a prerequisite, all free α-amines are modified on the protein level by acetylation (McDonald and Beynon, 2006; Yeom ), propionylation (Yeom ), or dimethylation (Mommen ; Lai ; Chen ; Li ). A pioneering protocol modified α-amines with NHS-biotin for subsequent depletion with immobilized streptavidin (McDonald ). In an optimized version, the same group directly captured the digest-generated peptides with NHS-activated Sepharose (McDonald and Beynon, 2006). A further modified protocol termed NRich combined internal peptide depletion via NHS-activated Sepharose with the popular filter-aided sample preparation (FASP) (Wiśniewski, 2017; Yeom ). The charge reversal approach adds two covalently linked disulfonate groups to digest-generated α-amines, resulting in a strong negative charge that enables their depletion by SCX chromatography (Lai ). This procedure can be performed in tip-based format for limited samples such as excised gel slices. Hydrophobic tagging-assisted N-termini enrichment (HYTANE) modifies digest-liberated α-amines with hexadecanal (Chen ). This increases peptide hydrophobicity and enables depletion using C18 reverse phase trap columns, where hexadecanal-alkylated peptides remain bound even during washes with 80% acetonitrile. In STagAu, α-amines of digest-generated peptides are modified with sulfhydryl groups using Traut’s Reagent (Li ). This enables efficient depletion by the tight interaction of sulfhydryl groups with gold-coated nanoparticles (Li ). The phospho-tag (PTAG) approach modifies digest-generated α-amines with glyceraldhyde-3-phosphate, followed by depletion using TiO2 affinity that is well established for the enrichment of phosphorylated peptides (Mommen ).

The PTAG strategy was further adapted for the enrichment of C-terminal peptides (Chen ). Carboxyl groups are amidated by EDC-mediated reaction with methylamine, followed by LysC digestion. The resulting C-terminal peptides thus do not possess a Lys residue, whereas internal and N-terminal peptides carry a C-terminal Lys. In the next steps, α-amines are selectively protected by dimethylation at acidic conditions before selective attachment of the PTAG to Lys ε-amines of the LysC-digested peptides and depletion of the phosphorylated peptides by TiO2 affinity. However, a major limitation of the method is the strict requirement for LysC as digestion enzyme, which prevents identification of C-termini where no Lys residue is suitably placed to result in MS-identifiable peptides.

A snapshot of the plant terminome landscape

Termini-centric studies are still rare in plant sciences despite the variety of different techniques and their wide application and impact in medical sciences (Demir ; Klein ). However, emerging applications are providing new and broader insights into protein synthesis, maturation, modification, and degradation in various aspects of plant life (Fig. 4).

Fig. 4.

Characterization of cellular processes in plants by N-terminome profiling. N-terminome profiling has been applied to elucidate diverse cellular processes in plants. (A) N-termini as protein stability determinants in the N-end rule pathway for targeted degradation by the 26S proteasome. (B) The stromal terminome and maturation of nucleus-encoded plastidial proteins by the stromal processing peptidase (SSP) and potential additional proteases. (C) The alternative initiation of translation within the same mRNA resulting in two different proteoforms. (D) N-terminal protein modifications such as co-translational N-terminal initiator Met excision (NME) and Nα acetylation (NAA). (E) Identification of METACASPASE9 (MC9) protease substrates. (F) Proteolytic maturation of nucleus-encoded mitochondrial proteins after their import by the proteases ICP55 and OCT1. Ub, ubiquitin; E2/E3, E2/E3 ubiquitin ligases; TP, transit peptide; M, methionine; Ac, acetyl group.

An extensive 6-plex iTRAQ ChaFRADIC analysis of Arabidopsis seedlings employing multiple proteases for digestion identified 2791 N-terminal peptides of 2249 unique N-termini from 1270 proteins (Venne ). The data experimentally confirmed sequence determinants for N-terminal initiator Met excision (NME) and many transit, signal, and propeptide cleavage events annotated in UniProt (https://www.uniprot.org). Furthermore, a majority of 1793 unique N-termini matched no annotation, indicating endogenous proteolytic processing. Many of these were ‘ragged’ termini, for example termini from the same protein separated by only one residue. In agreement with the N-end rule, the non-destabilizing residues (S, A, T, G, and V) accounted for the vast majority of N-terminal residues of protease-generated neo-N-termini, with a similar frequency distribution to NME-processed N-termini matching to position 2 of their gene models.

A similar study characterized the N-terminome of Arabidopsis chloroplasts (Rowland ). TAILS N-terminome analysis of whole leaf and isolated chloroplast proteomes identified a total of 894 N-termini from 577 proteins, including 544 N-termini from 250 chloroplast proteins. Many of the nuclear-encoded proteins were observed to be ragged with two to three different N-termini, suggesting that these proteins undergo further processing directly after chloroplast import. Interestingly, for all 16 nucleus-encoded proteins with known dual targeting to chloroplasts, mitochondria, and/or the cytosol, only one N-terminus representing the chloroplast-localized form was observed. However, this may result from a bias towards the more abundant plastid proteoforms of these proteins in the investigated material. For 126 plastid proteins, only the mature N-terminus was observed, while for 100 proteins multiple N-termini were identified, including (i) N-termini located before the transit peptide cleavage site including precursor proteoforms; (ii) N-termini in close proximity to the predicted transit peptide cleavage site representing the mature plastid proteoform; and (iii) N termini reflecting degradation products. An analysis of the N-terminal residue occurrence found an over-representation of stabilizing residues (A, V, T, and S) for nuclear-encoded proteins. A similar distribution was found for the plastid-encoded proteins, with additional frequent occurrence of intact initiator Met. Destabilizing residues were significantly under-represented as N-terminal residues, providing indirect evidence for the existence of a plastid version of the N-end rule.

N-terminomics for identification of alternative translation initiation sites

Identification of bottom-up-based proteomics data depends greatly on well-annotated protein models, which in turn rely on correct identification of the protein translation initiation sites. N-terminome analysis is therefore ideally suited to obtain experimental evidence for predicted protein models (Hartmann and Armengaud, 2014). A novel proteogenomics pipeline combined computational genome analysis with ribosome profiling and N-termini profiling by COFRADIC to identify novel translation initiation sites (TISs) in Arabidopsis (Willems ). In the first stage, N-terminal peptides were enriched from Arabidopsis cell culture after parallel digest with complementary proteases to increase the coverage depth. Peptides were identified from a six frame translation Arabidopsis database with three different search engines. In the second stage, unmatched MS/MS spectra were searched against a customized database consisting of N-terminal peptide sequences derived from translation ribosome profiling data, computationally predicted gene models, and a six frame translation of the genomic data. Using strict thresholds criteria, experimental evidence for translation at 117 novel TIS locations was identified, of which 50 mapped to intergenic regions, 44 partly overlapped with existing protein models, 2 were located in pseudogenes, and 21 were found in transposable elements. Furthermore, 23 N-termini matched alternative translation initiation sites identified by ribosome profiling, providing experimental evidence for the physiological relevance of such alternative ribosome-binding sites. Importantly, the results obtained from Arabidopsis further enabled prediction of novel protein-coding genes in other species, with both well-annotated and poorly annotated genomes.

Profiling of N-terminal protein modifications

A landmark study compared protein N-terminal modifications such as NME, acetylation, and N-myristoylation in human and Arabidopsis cell cultures. Comparative SILProNAQ analysis identified 1072 N-terminal peptides from 1007 Arabidopsis proteins, and 717 N-terminal peptides from 715 human proteins (Bienvenut ). In both data sets, ~70% of the proteins were subject to NME and 90% of proteins were N-terminally acetylated in the steady state. Only five and three myristoylated N-termini were observed in the Arabidopsis and human sample, respectively. The vast majority of acetylated peptides represented annotated expected N-termini. In Arabidopsis, 148 N-terminal peptides of plastid-located proteins were post-translationally acetylated after signal cleavage, indicating the existence of a plastidial Nα-acetyl-transferase (NAT). Three years later, the same group identified NAA70 as the first plastid-located NAT (Dinh ). Recombinant NAA70 was able to acetylate both protein termini starting with the initiator Met and NME-processed termini in vitro, with the sequence specificity matching the known acetylated termini of plastid proteins.

SILProNAQ was also instrumental for the identification of a critical role for the cytosolic NatA complex in drought stress tolerance in Arabidopsis (Linster ). As in human or yeast, the NatA complex contains two subunits, the catalytic subunit NAA10 and the auxillary subunit NAA15. Comparison of the global acetylation profiles of wild-type Arabidopsis and artificial miRNA (ami) lines of NAA10 and NAA15, respectively, revealed that both NAA10 and NAA15 are necessary for NatA activity and responsible for the acetylation of almost half of the soluble proteins in the leaves. Interestingly, both amiNaa10 and amiNaa15 lines displayed drought stress tolerance, which could be traced to closure of stomatal apertures and altered root morphology. The steady-state transcriptomes of both mutant lines highly correlated with the transcriptome of drought-stressed wild-type plants. Abscisic acid, an important phytohormone in drought stress signaling, induced rapid degradation of NatA, resulting in an increase of free, unacetylated N-termini in drought-stressed wild-type plants.

A very recent study mapped N-terminal myristoylation in Arabidopsis and human cells in vitro on a proteome-wide scale and using SILProNAQ in vivo (Castrec ). First, the substrate preference of human N-myristoyltransferase (NMT) beyond the known requirement for N-terminal Gly was deduced from the crystal structure with co-crystalized substrate peptides and used to predict potential NMT substrates in the human and, based on the high NMT sequence similarity, also in Arabidopsis proteomes. A subset of several hundred termini including both predicted substrates and non-substrates were validated in vitro with recombinant protein. Hundreds of protein starting with Gly termini were thus shown to be myristoylated in both organisms in vitro, while SILProNAQ identified myristoylation of 46 human proteins and 75 Arabidopsis proteins in vivo, predominantly in membrane-enriched fractions. Interestingly, myristoylation and acetylation appear to compete for common substrates, providing a novel co-translational regulatory mechanism possibly affecting subcellular localization.

N-terminome profiling reveals conserved multi-step proteolytic maturation of nuclear-encoded proteins imported in plastids and mitochondria

Plant cells harbor two organelles of endosymbiotic origin, mitochondria and plastids, that rely on the import of most of their protein complement by N-terminal targeting sequences that are proteolytically removed after import (Ghifari ; Nakai, 2018). Detailed biochemical studies allowed the prediction of transit peptide sequences and their cleavage sites, but experimental evidence remained scarce for all but the best-studied proteins.

In a landmark study, mature protein N-termini in yeast mitochondria were determined by COFRADIC (Vögtle ). This and a subsequent ChaFRADIC study (Venne ) further demonstrated that a subset of imported proteins are subsequently processed by the aminopeptidase ICP55 and the octapeptidase OCT1 after transit peptide cleavage. Interestingly, two alternatively spliced ICP55 transcripts were found in Arabidopsis, giving rise to ICP55 proteoforms with distinct N-terminal sequences and different subcellular localization in vivo (Carrie ). ICP55.1 was found in the mitochondria, while ICP55.2 localized to the nucleus. Δicp55, Δoct1-1, and Δoct1-2 mutants isolated by duplex dimethyl ChaFRADIC mitochondrial proteome analysis were compared with the wild type. Stringent selection criteria based on differentially accumulating N-terminal peptides identified 88 mitochondrial proteins as putative ICP55 substrates. Alignment of the putative ICP55 cleavage sites revealed a conserved consensus motif shared with the yeast homolog, namely preferred cleavage of destabilizing hydrophobic residues (F, Y, I, and L). For OCT1, only seven putative substrates were identified which did not reveal a consensus cleavage site.

Similarly, N-terminome profiling was used to determine mature N-termini of nuclear-encoded proteins imported into the complex plastids of the diatom Thalassiosira pseudonana (Huesgen ). Complex plastids are derived by secondary endosymbiosis of a photosynthetic eukaryote and therefore surrounded by four membranes that must be crossed by nuclear-encoded plastid proteins (Sheiner and Striepen, 2013). Using the TAILS protocol, 1401 distinct N-terminal sequences from 976 proteins were identified from crude soluble and membrane protein fractions, revealing extensive N-terminal acetylation of both cytosolic and plastid-located proteins (Huesgen ). Positional and functional annotation combined with manual gene model validation identified 63 mature N-termini of nuclear-encoded plastid-imported proteins that allowed deduction and characterization of the bipartite transit peptide sequences. Alignment of the cleavage sites revealed a consensus motif for signal peptide cleavage and evidence for Met aminopeptidase processing after import. Sequence comparisons further suggested that the identified features and sequence characteristics were shared by other eukaryotes with complex plastids. The data set further helped to validate a novel algorithm for predicting proteins imported into complex plastids (Gruber ).

A similar study determined the composition and cleavage site of transit peptides for protein import into the cyanelles of the glaucophyte Cyanophora paradoxa (Köhler a). Cyanelles possess a peptidoglycan wall between the inner and outer envelope membrane and are therefore believed to resemble the ancestral status of chloroplasts of higher plants. N-terminal peptides were enriched from total cell extracts and isolated cyanelles using the TAILS procedure. In total, 303 N-terminal peptides were identified, including termini of 123 nuclear-encoded cyanelle proteins. Using stringent filter criteria, 77 of these were identified as likely to be mature N-termini of nuclear-encoded cyanelle proteins. Transit peptides deduced for these C. paradoxa proteins revealed a unique phenylalanine within the first 10 amino acids, but otherwise showed a similar composition, architecture, and cleavage site motif to 136 Arabidopsis transit peptides identified using the same approach (Köhler b). In both organisms, subsets of plastid-imported proteins appeared subjected to further N-terminal aminopeptidase processing

The Arabidopsis study further investigated the consequences of impaired plastid protein import in the tic56-1 and ppi2 mutant lines lacking critical subunits of the translocase supercomplexes in the inner and outer chloroplast envelope, respectively (Köhler b). Both mutants showed albino phenotypes and required an external carbon source for growth. TAILS N-terminome analysis identified N-terminal peptides of 348, 481, and 280 proteins from the wild type, tic56-1, and ppi2, respectively. Despite the severe phenotype, only two and four N-termini from nuclear-encoded plastid proteins matched unprocessed proteoforms with intact transit peptides in tic56-1 and ppi2, whereas the majority of plastid protein termini indicated unaltered proteolytic processing to the mature, imported proteoforms. A great overlap of imported proteins was observed in both mutant lines, although two different sites of import were affected, and the proteins showed no significant enrichment of specific functional categories. This further suggested the existence of redundant or alternative translocase complexes, as import of a subset of plastid proteins appeared unaffected.

Global analysis of N-terminal protein stability determinants

In the Arg/N-end rule pathway, basic N-terminal residues are targeted for proteasomal degradation by the Arabidopsis E3 ligase PROTEOLYSIS 6 (PRT6). Acidic residues and oxidized Cys may be N-terminally arginylated by the arginyltransferases ATE1 and 2, effectively converting N-termini with these residues to PRT6 substrates (Dissmeyer ; Dissmeyer, 2019). Triplex dimethyl TAILS N-terminome profiling of prt6 and ate1/2 identified 1465 unique N-termini, of which 864 were acetylated and 601 dimethylated, corresponding to free N-termini in vivo (H. Zhang ). Acetylated peptides predominantly exposed stabilizing residues Ala and Ser and mapped to the expected translation start site with and without subsequent NME and chloroplast transit peptide cleavage sites. The majority of the protease-generated neo-N-termini also exposed stabilizing residues (Ala, Ser, Thr, and Val), but a moderate number of primary (Phe, Leu, and Ile) and secondary (Asp and Glu) destabilizing residues were also observed. However, none of these N terminal peptides accumulated significantly in the N-end rule mutants. Quantitative root proteome analysis found only 17 out of almost 3300 quantified proteins significantly up-regulated in both lines prt6 and ate1/2. Most of these reflected transcriptional changes, including genes controlled by the known Arg/N-end rule substrates, the group VII ethylene response factors (ERFVII). This suggested that the Arg/N-end rule pathway targets only a small number of substrates under standard growth conditions.

In a follow-up study, the same group assessed the impact of PRT6 in etiolated Arabidopsis seedlings (H .Zhang ). Six-plex TMT TAILS analysis of the wild type and prt6 loss-of-function mutants with two complementary digestion enzymes resulted in a rich N-terminome data set of 5004 unique N-terminal peptides from 2396 protein groups. The N-termini of several proteases including RD21A were down-regulated, while the N-termini of 45 protein groups including cruciferins were significantly increased in the mutant line, most of which were directly or indirectly regulated by ERFVII transcription factors. Activity-based protein profiling confirmed reduced RD21 activity and increased cathepsin B activity, indicating that PRT6 affects protease activity and mobilization of seed storage proteins in a complex manner.

Profiling of N-terminal arginylation by IP-MS

The previously mentioned ‘pan-arginylation’ antibodies were used to enrich arginylated proteins by immunoprecipitation from mouse tissue extracts, followed by MS-based identification (IP-MS). Here, 43 diverse proteins with important physiological roles were arginylated in vivo at highly specific sites. These appeared to be located both at the N-terminus of Asp- and Glu-initiated proteins and at the side chains of midchain Glu residues (Wong ; T. Xu ). In the moss Physcomitrella patens, several candidate substrates of ATE1 (Schuessele ) were found by an IP-MS approach using the described ‘pan-arginylation’ antibodies (Wong ; T. Xu ). The identified arginylated peptides belonged to an uncharacterized protein and a putative AAA-type ATPase. Arginylation in the ABC transporter family protein PpABCB20 was found to be likely by using a less stringent search algorithm. Experimental evidence for arginylation occurring at an N-terminal Asp was reported for the acylamino-acid-releasing enzyme PpAARE (Hoernstein ). However, it remains unclear how such an N-terminal Asp residue could be generated because it is in position 2, preceded by the initiator Met that should be retained under these circumstances.

Identification of N-terminal arginylation by immunodetection

Prominent examples of post-translationally arginylated, proteolytic proteoforms (protein cleavage products) in the animal field are the endoplasmic reticulum (ER) chaperone BiP (also known as GRP78 and HSPA5, heat shock 70 kDa protein 5) (Cha-Molstad ; Jiang ; Yoo ), CRT (calreticulin) (Decca ; Cha-Molstad ; Yoo ), BRCA1 (Piatkov ; Yoo ), β actin (Saha ), PDI (protein disulfide isomerase) (Cha-Molstad ; Yoo ), and CDC6 (Yoo ). In these studies, important roles of N-terminal arginylation in autophagy and recognition of N-terminally arginylated targets by the N-recognin p62 were shown (Cha-Molstad , , 2016, 2017, 2018).

There are other examples for such antibodies raised against a plant-derived arginylated epitope, however, not in the context of N-terminal arginylation (Havé ). The strength of the antibody signals can also be related to in vitro arginylation assays where recombinant ATEs transfer Arg to short acceptor peptides initiated with Asp, Glut, or Cys sulfonic acid (Klecker and Dissmeyer, 2016; White ). Here, the ideal antibody dilutions needs to be precisely determined using control arginylated versus non-arginylated peptides.

Antibodies detecting co-translationally modified Met

N-terminal acetylation of cellular proteins creates specific degradation signals termed Ac/N-degrons that can be targeted by the Ac/N-end rule pathway (Hwang ; Gibbs, 2015). For example, the N-terminally acetylated Met residue of MATα2 was found to act as a degradation signal. It is, similar to N-terminally acetylated Leu, Gly, Ala, Val, Pro, Ser, Thr, and Cys, recognized by the Doa10 ubiquitin ligase in yeast (Hwang ). Then, Cog1, a subunit of the Golgi-associated COG complex in yeast, is an example for such a target and an example where an antibody specific for N-terminal acetylated proteins was used to determine levels of N-terminally modified protein in comparison with the non-acetylated proteoform (Shemorry ). The use of anti-acetylation antibodies for enrichment of N-terminally acetylated proteins is not common, probably mainly due to the large fraction of the proteome that is actually undergoing this co-translational modification. The lysine acetylome can be treated differently as it can indeed be characterized and quantified by enrichment with antibodies against acetylated lysine residues followed by MS (Choudhary ; Guan ; Finkemeier ; Hartl ; Lassowskat ; Ree ). In contrast to the clear situation in yeast, a recent study showed that the N-terminal acetylation can have antagonistic effects on the stability of two related proteoforms (Xu ). The Nod like receptor (NLR) protein SUPPRESSOR OF NPR1, CONSTITUTIVE 1 (SNC1), has two distinct proteoforms due to alternative translation which are selectively acetylated either by NATA or by NATB, respectively. The acetylation by NATA leads to the degradation of SNC1 while the acetylation by NATB enhances SNC’s stability (Xu ).

Very recently, antibodies specific for N-terminal formylation were also produced (Kim ). The yeast formyltransferase Fmt1 is imported into mitochondria and, nonetheless, produces N-terminal formylated proteins in the cytosol. These can be targets of the novel, so-called fMet/N-end rule pathway that aids degradation of N-terminal formylated proteins.

N-terminomics for unbiased identification of plant protease substrates

Plants contain hundreds of proteases with mostly unknown substrates and function (van der Hoorn, 2008; Demir ; Paulus and van der Hoorn, 2019). In a pioneering study, COFRADIC was used for complementary in vitro and in vivo approaches to identify physiological substrates of Arabidopsis metacaspase 9 (MC9) (Tsiatsiani ). For the in vitro approach, a proteome extracted from 2-day-old seedlings of a mutant line lacking MC9 was incubated with active recombinant MC9 (rMC9). COFRADIC identified 3050 N-terminal peptides from 1138 proteins, of which 332 N-terminal peptides were significantly more abundant after rMC9 treatment. Analysis of these 332 cleavage sites confirmed the known MC9 preference for cleavage after basic residues at the P1 site. Additionally, a strong preference for acidic residues at the P1' site was revealed. In the in vivo approaches, proteome samples of 2-day-old mc9 loss-of-fuction seedlings were compared with the wild type or with plants overexpressing MC9 under control of the 35S promoter. COFRADIC analysis revealed 3781 N-terminal peptides from 1705 proteins in the comparison of mc9 and the wild type, and 2879 N-terminal peptides from 1407 proteins in the comparison of mc9 with the MC9 overexpressor. In total, 99 cleavage events matching the MC9 sequence specificity in 74 candidate substrate proteins were observed in at least two of the three experiments. Randomly picked candidate cleavage events were confirmed by rMC9 cleavage of synthetic peptides representing the cleavage site-spanning region and with recombinant proteins produced by in vitro transcription and translation. From those experiments, the phosphoenolpyruvate carboxykinase 1 (PEPCK1) emerged as a highly likely physiological substrate of MC9. Further functional validation revealed that MC9 activated PEPCK1 in vivo, leading to impaired gluconeogenesis and shorter hypocotyls in dark-grown mc9 mutants that resembled the phenotype of pepck1 seedlings.

Similarly, COFRADIC was used to define the cleavage specificity and identify candidate substrate of the quality control proteases HhoA, HhoB, and HtrA in the model cyanobacterium Synechocystis sp. PCC 6803 (Tam ). COFRADIC analysis of Synechocystis proteome extracts incubated with recombinant HhoA, HhoB, and HtrA revealed similar broad specificity profiles. Combined 2D-difference gel electrophoresis (DIGE) proteome and COFRADIC N-terminome analysis showed that inactivation of each single protease had a similar impact, with differential expression and processing of enzymes involved in major metabolic pathways. Taken together, the in vivo and in vitro data sets provide evidence for RbcS as a physiological HhoA substrate, PsbO as a HhoB and HtrA substrate, and Pbp8 as a HtrA substrate.

Conclusions

Protein termini provide important information about the functional state of the proteome. A multitude of N-termini enrichment techniques have been developed, but are still rarely applied in plant sciences. Positive selection procedures enable targeted selection of one group of either unmodified or specifically modified N- or C-termini, but often require more sophisticated reagents. Negative selection strategies are often based on established chemical modifications routinely used in many laboratories, and enrich all N- or C-terminal peptides simultaneously, and are therefore ideally suited for unbiased profiling of protein termini and their endogenous modifications. Only a few proteomes have been studied by more than one technique, which generally showed low overlap between the results of different protocols (Chen ). However, inherent biases could not yet be identified as the observed low overlap may also be attributed to stochastic differences due to low proteome coverage or differences in sample amounts or digest enzymes and efficiency.

Most MS-based termini enrichment techniques still require expensive reagents and considerable technical experience that may limit their appeal to non-expert laboratories. Non-specific losses incurred during clean-up and depletion steps, such as by adsorption to plastic wear, filters, and/or bead material, pose challenges for sensitivity and reproducibility that practically exclude analysis of microscale samples such as microdissected tissues. Ongoing efforts in technology development strive to improve sensitivity, robustness, and ease of use of both experimental and data annotation workflows. In addition, termini-centric MS approaches face the challenge that proteoform identification and quantification are based on single peptides. Replicate experiments with stringent statistical analysis should be performed, and termini identifications that are based on single, spurious spectrum matches must be excluded. Nevertheless, single peptide quantification is inherently more variable than protein quantification by standard shotgun proteomics approaches, where data from several peptides are averaged for each protein. However, targeted MS and antibody-based assays can be developed to enable more accurate quantification and sensitive monitoring of known proteoforms (Huesgen ; Savickas and auf dem Keller, 2017). Such targeted assays with more accurate quantification will also enable more sensitive detection of small but meaningful changes between different genotypes or samples, especially in larger sample series. Finally, identification of proteolytically cleaved proteoforms also enables individual quantification of the processed and unprocessed forms (Fahlman ). This in turn allows calculation of the degree of processing, similar to site occupancy for other post-translational modifications (Arsova ). This may reveal threshold amounts for cellular switches triggering specific irreversible physiological responses or commitment to cell fate decisions.

In summary, studies of plant protein termini have already addressed a broad range of biological problems. These included plant protease substrate identification, profiling of protein terminal modifications, characterization of N-terminal protein stability determinants in various subcellular compartments, protein maturation in organelles of endosymbiotic origin in diverse organisms, and identification of alternative translation initiation sites (Fig. 4). Taken together, these studies have already revealed striking similarities in the N-terminal identity, modification, and maturation, and highlighted the extraordinary abundance of proteolytic proteoforms across evolutionarily distant organisms. With continuous improvement and more widespread applications, we expect that more studies of the plant terminomes will emerge in the near future and provide unprecedented insights particularly into proteolytic processes in plants.