The Increasing Impact of Activity-Based Protein Profiling in Plant Science

Abstract

The active proteome dictates plant physiology. Yet, active proteins are difficult to predict based on transcript or protein levels, because protein activities are regulated post-translationally in their microenvironments. Over the past 10 years, activity-based protein profiling (ABPP) is increasingly used in plant science. ABPP monitors the activities of hundreds of plant proteins using tagged chemical probes that react with the active site of proteins in a mechanism-dependent manner. Since labeling is covalent and irreversible, labeled proteins can be detected and identified on protein gels and by mass spectrometry using tagged fluorophores and/or biotin. Here, we discuss general concepts, approaches and practical considerations of ABPP, before we summarize the discoveries made using 40 validated probes representing 14 chemotypes that can monitor the active state of >4,500 plant proteins. These discoveries and new opportunities indicate that this emerging functional proteomic technology is a powerful discovery tool that will have an increasing impact on plant science.

Introduction

Plant scientists are challenged with the daunting task of annotating functions to the vast number of genes that are uncovered by current genomics efforts. Because proteins are regulated post-translationally in their microenvironments, it is impossible to predict the active proteome based on transcript or protein accumulation, or post-translational modifications. The large number of non-annotated proteins that act in concert calls for methods that study protein activities globally rather than individually.

Activity-based protein profiling (ABPP) is a powerful functional proteomics technology that is increasingly used in plant science. ABPP applications in non-plant systems have been used to show the selectivity of drugs and as a diagnosic tool to detect and image malignant cancers. These topics have been reviewed extensively elsewhere ( Serim et al. 2012 , Hunerdosse and Nomura 2014 , Niphakis and Cravatt 2014 , Willems et al. 2014 ). Here we summarize the approaches and discoveries made by this technology in plant science. We will discuss the general principles first, followed by the findings for the various classes of proteins that have been studied so far. Finally, we will conclude with remaining challenges and future prospects.

The ABPP Concept

ABPP takes advantage of small molecule probes that label the active site of enzymes in an activity-dependent manner ( Fig. 1 ). Labeling is covalent and irreversible, and facilitates the display of labeled proteins upon separation on protein gels, but also the purification and identification of the labeled proteins by mass spectrometry (MS). These approaches require that the probe carries a reporter tag, being either an affinity tag (e.g. biotin or desthiobiotin), a fluorescent tag (e.g. bodipy, rhodamine or cyanine dyes Cy2/3/5), a radioactive tag (e.g. ¹²⁵ I) or a chemical tag (e.g. azide or alkyne minitag) that can be coupled to an affinity or fluorescent tag using ‘click chemistry’ ( Fig. 1 A).

Fig. 1

Principle of ABPP. (A) Four types of probes and reporter tags. Mechanism-based probes are based on irreversible inhibitors that mimic a substrate by locking the enzymatic mechanism in the covalent intermediate state. Photoaffinity probes are based on reversible inhibitors equipped with a photoreactive group that cross-links within the inhibitor-binding site. Suicide substrate probes are based on suicide inhibitors that become reactive upon enzymatic activity and often react within the substrate-binding site. Unbiased reactive probes consist of a tuned reactive chemical warhead that preferentially reacts with hyper-reactive residues. (B) Concept of proteome labeling with chemical probes.

The reporter tag is linked to a reactive group that can have different chemistries (chemotypes). The reactive group is often combined with a binding group that provides specificity to the probe. These binding and reactive groups (warheads) target and react with the active site of specific enzymes, respectively. Because enzymes have different mechanisms and substrates, these probes do not label all enzymes, but are specific to certain enzyme classes that share the same mechanism and similar substrates. ABPP therefore displays only a subset of the active enzymes, depending on the choice of probe.

The term ‘activity-based’ can be confusing as the labeling is most often based on tagged inhibitors that irreversibly inactivate the enzyme. The ABPP readout is not based on substrate conversion, but on the availability and reactivity of the active site of enzymes. Importantly, enzyme activities are nearly always regulated by the exposure of the active site to its environment ( Kobe and Kemp 1999 ). Inhibitors often block the active site, and even allosteric regulation affects the exposure of the active site. Therefore, labeling of the active site reflects which enzymes are in an active state, irrespective of the presence of substrates ( Bachovchin et al. 2009 ).

The Choice of Probe

There are four types of chemical probes ( Fig. 1 A). Most activity-based probes are based on mechanism-based inhibitors ( Cravatt et al. 2008 , Johnson et al. 2010 , Niphakis and Cravatt 2014 ). Many enzymes catalyze reactions by going through a covalent intermediate with the substrate. Inhibitors of these enzymes often mimic substrates but lock the mechanism in the covalent intermediate state. Probes based on these covalent inhibitors are by definition mechanism-based inhibitors that label the active site in an activity-dependent manner.

In contrast, photoaffinity probes are based on reversible inhibitors that display the availability (not reactivity) of the active site ( Sumranjit and Chung 2013 ). The covalent attachment is facilitated by incorporating a photoreactive group that labels proximal residues in the enzyme when UV irradiated. For instance, a probe for matrix metalloproteases (MMPs) is based on the reversible marimastat inhibitor and equipped with a benzophenone photoreactive group ( Lenger et al. 2012 ). Although these probes are not mechanism based, they do read out the availability of the substrate-binding pocket and often reflect enzymatic activity.

A third type of probes are the suicide substrate probes. These probes act as substrate but their modification by the enzyme results in a hyper-reactive intermediate that quickly reacts with a nucleophile within the substrate-binding pocket. Labeling is still activity dependent but these probes do not label the active site but rather other nearby amino acid residues. An example of a suicide substrate probe is 2-ethynylnaphthalene, which is activated by Cyt P450 enzymes, resulting in a reactive ketene moiety that reacts within the enzyme ( Wright and Cravatt 2007 , Wright et al. 2009 ). A risk with suicide substrate probes is that they can diffuse out of the enzyme before they react with a protein ( Sellars et al. 2010 ).

The fourth class of probes are called reactivity probes. These probes consist of only a reactive group that lack an obvious binding group and label intrinsically hyper-reactive residues without selectivity for certain enzyme classes. For instance, iodo/chloroacetamide and sulfonyl fluoride (SF) probes react with hyper-reactive cysteine and tyrosine residues, respectively ( Weerapana et al. 2010 , Gu et al. 2013 ). Interestingly, because these probes first react with the most reactive amino acid residues, they preferentially react with active site residues when used at relatively low concentrations. These probes also label other sites in proteins that have an elevated reactivity, and these sites often have regulatory functions, e.g. sensing post-translational modifications such as oxidation and nitrosylation. Because reactivity probes lack a binding group, they label hyper-reactive sites in an otherwise unbiased manner, which may have powerful implications in the future because these probes also display novel sites having biological significance.

An important parameter of chemical probes is their selectivity. Many probes are highly selective to their target enzyme class, and the number of these target enzymes in a given proteome can vary between a few [e.g. proteasome or vacuolar processing enzymes (VPEs); Gu et al. (2010) , Kolodziejek et al. (2011) , Misas-Villamil et al. (2013) ] and a few hundred enzymes [e.g. serine hydrolases (SHs) or glycosidases; Kaschani et al. (2009a) , Chandrasekar et al. (2014) ]. Another group of probes react not only with the expected targets, but also at unexpected positions. AcATP probes, for example, also label outside known ATP-binding pockets ( Villamor et al. 2013 ).

How to Detect and Confirm Probe Targets

The best detection method depends on the number of targets and knowledge of the labeling sites ( Fig. 2 ). Mechanism-based probes that only label a few active proteins in a proteome are the easiest to work with because the detection only requires the separation on protein gels ( Fig. 2 A). Probes that label >50 proteins, however, cause very dense labeling profiles in which single proteins can no longer be distinguished. The analysis then requires separation on 2D gels or by liquid chromatography ( Fig. 2 B). The detection of labeled proteins is simple on protein gels, but more challenging using MS. Finally, reactivity probes and some other probes require the identification of the labeled site by sequencing labeled peptides ( Fig. 2 C).

Fig. 2

Detection of labeled proteins. (A) Proteomes labeled with fluorescent probes are separated on protein gels and scanned for fluorescence. (B) Biotinylated proteins are purified from proteomes labeled with biotinylated probes and either separated on protein gels and identified by in-gel trypsin digest and MS analysis (left); or digested on-bead by trypsin and analyzed by MS (right). (C) Alternatively, whole labeled proteomes are digested in-solution by trypsin and biotinylated peptides are purified and identified by MS.

The confirmation of labeling a protein is achieved in different ways. The detection of proteins that were expected from the used probe is a strong indication of a true target, but an additional confirmation should be included. The absence of the target protein in the ‘no probe’ control is an important check point, but not decisive. The availability of genetic tools to knock-out or knock-down a gene allows the confirmation by the absence of the signal in the activity profile ( Misas-Villamil et al. 2013 , Lu et al. 2015 ). Probing a pull-down of the labeled protein with a protein-specific antibody has also been used ( van der Hoorn et al. 2004 , Wang et al. 2008 ). Alternatively, labeling of a heterologously expressed protein and its labeling site mutant can be used for confirmation, e.g. of glutathione S -transferases (GSTs; Gu et al. 2013 ) and Avirulence protein Pseudomonas phaseolicola -B (AvrPphB; Lu et al. 2013 ). However, the most robust confirmation usually comes from the annotation of a fragmentation spectrum of a labeled peptide since this provides direct evidence of the conjugation of the probe to the protein active site. This confirmation has been achieved many times, e.g. for β-thioglucoside glucohydrolase TGG1 ( Chandrasekar et al. 2014 ) and ATP-binding proteins ( Villamor et al. 2013 ). However, this approach can be challenging because some conjugates hydrolyze non-enzymatically during work-up (e.g. conjugates with glycosidases; Chandrasekar et al. 2014 ). In addition, many labeled peptides escape detection above the set peptide quality threshold because the fragmentation spectra contain additional peaks that are not recognized. Manual annotation of fragmentation spectra is frequently needed to increase confidence in the labeling site.

ABPP has been applied to study biological systems in three ways ( Table 1 ). In ‘comparative ABPP’, different proteomes are labeled and compared to display differences between different biological conditions ( Fig. 3 ). The comparison will display differential labeling of single proteins. Comparison of different proteins within a single proteome indicates relative activity levels, but such a comparison cannot be accepted in cases where labeling is not saturated and different proteins have different intrinsic affinities for the probe. The advantage of comparative ABPP over traditional, substrate-based studies is that multiple enzymes can be monitored simultaneously in crude extracts and in living organisms, without the need to know the substrates. Comparative ABPP directly identifies active proteins, even if they act redundantly on the same substrate.

Fig. 3

Example of comparative ABPP by DIGE. Two different proteomes are both labeled with an alkyne-minitagged probe, and alkyne-labeled proteins are coupled to azide-tagged Cy3/Cy5 by click chemistry. Samples are mixed, separated on protein gels and scanned for fluorescence for both Cy3 and Cy5. Fluorescence images are overlaid and the mixed lane is quantified and plotted as in the histogram. The specific sample shows differential serine hydrolase activities in apoplastic proteomes of N. benthamiana upon infection with P. syringae , displayed by labeling with FP≡ (adapted from Hong and van der Hoorn 2014 ).

Table 1

Three ways to apply ABPP

	Comparative ABPP	Competitive ABBP	Convolution ABPP
Usage	To detect differential protein activities	To test a putative inhibitor(s)	To detect post-translational regulatory mechanisms
Approach	Label two (or more) extracts and compare the labeled proteomes in-gel or by gel-free proteomics.	Pre-incubate extract with putative inhibitor(s) and then add probe to label the non-inhibited enzymes.	Mix two proteomes (samples A and B) having differential activities and then label them (sample D), and compare this with proteomes that are first labeled and then mixed (sample C).
Outcome	Differential labeling indicates proteins that differentially accumulate or are regulated by activation or suppression mechanisms.	Suppressed labeling indicates that the respective protein is inactivated by the putative inhibitor. Putative inhibitors can also be non-covalent, reversible and allosteric.	Suppressed labeling in sample D compared with sample C indicates the presence of inhibitor or other suppressor mechanisms working in trans. Increased labeling indicates activation mechanisms that can act in trans.
Example	See Fig. 3	See Fig. 4	See Fig. 5

In contrast, ‘competitive ABPP’ is used to detect inhibitors. In this ABPP mode, proteomes are pre-incubated with putative inhibitors and then the probe is added to label the non-inhibited enzymes ( Fig. 4 ). Inhibitors can be reversible or irreversible and allosteric or competitive, based on proteins or small molecules. The suppression of labeling indicates that the protein is inhibited. This approach is frequently used, e.g. to show targets of phosphate-based agrochemicals in plant proteomes ( Kaschani et al. 2012a ) and to reveal the adaptation of a protease inhibitor from Phytophthora infestans for novel host proteases upon a host jump ( Dong et al. 2014 ). The advantage of competitive ABPP over traditional inhibition assays is that the target proteins do not need to be purified and that no (artificial) substrate needs to be identified.

Fig. 4

Example of competitive ABPP. A plant proteome is pre-incubated with putative inhibitors and then incubated with activity-based probes to label the non-inhibited enzymes. Labeled proteins are separated on protein gels and detected, in this case by fluorescence scanning. Suppression of labeling indicates that the specific enzyme is inhibited. This specific example shows the selective suppression of the three catalytic subunits of the proteasome using aldehyde-based inhibitors (adapted from Gu et al. 2010 ).

A third, new way of applying ABPP has been termed ‘convolution ABPP’ (B. Chandrasekar and R. Van der Hoorn, unpublished data). This method is basically a combination of comparative and competitive ABPP and is based on mixing samples before and after labeling. If mixing samples before incubation with the probe suppresses labeling when compared with mixing labeled samples, then there must be factors in one of the samples that suppresses labeling in the other sample. This ‘ in trans suppression’ of labeling leads to the discovery of novel inhibitors and other regulatory mechanisms. For instance, convolution ABPP revealed that there is a suppressor of a secreted plant β-galactosidase produced upon infection with Pseudomonas syringae ( Fig. 5 ; B. Chandrasekar and R. Van der Hoorn, unpublished data).

Fig. 5

Example of convolution ABPP. Samples A and B are mixed before or after labeling, resulting in samples D and C, respectively. If sample B contains excess inhibitor, then labeling is suppressed in sample D when compared with sample C. In this case, apoplastic proteomes from uninfected (sample A) or P. syringae -infected (sample B) N. benthamiana plants were used, revealing a suppression of a 50 kDa enzyme in the apoplast (adapted from B. Chandrasekar and R. Van der Hoorn, unpublished data).

Quantification of Labeling

Comparison of two or more samples in comparative, competitive and convolution ABPP requires a quantification of labeling. Quantification of fluorescence from 1D gels is sufficient in cases where the signals can be sufficiently separated, e.g. for the proteasome and VPEs ( Gu et al. 2010 , Misas-Villamil et al. 2013 ). However, signals in most activity profiles are caused by multiple enzymes that migrate at a similar molecular weight. Quantification by spectral counts on a tryptic digest from gel slices of purified labeled proteins has occasionally been used, but this is rarely reproducible and no longer recommended. Nevertheless, the trifunctional fluorophosphonate (TriFP) probe that carries both biotin and rhodamine has been used to show a correlation between spectral counts and fluorescence intensity for SHs ( Kaschani et al. 2009a ). An alternative, more robust way to quantify is to separate fluorescently labeled proteins on 2D gels, e.g. for the papain-like cysteine proteases (PLCPs; Shabab et al. 2008 ). A more precise pairwise comparison of two samples can be accomplished using proteomes that are labeled with different fluorophores. This approach, called ABPP with differential gel electrophoresis (ABPP-DIGE; Fig. 3 ), has been used to display differential SH activities in the apoplast upon infection by P. syringe ( Hong and van der Hoorn 2014 ). Ultimately, the quantitative comparison of labeled proteomes requires the use of more sophisticated proteomic technologies, such as label-free quantification, stable isotope labeling and isobaric tags (Bindschedlet and Cramer 2011).

ABPP is broadly applicable to any biological system from which >20 µg of proteome can be extracted per sample. MS analysis and an annotated genome sequence are not required in cases where the probe targets have been sufficiently described (e.g. proteasome and VPEs). An even broader application is possible by in vivo labeling because most fluorescent and minitagged probes are cell permeable. For instance, in vivo labeling of PLCPs increased the number of targets by 2-fold when compared with in vitro labeling ( Richau et al. 2012 ), possibly because each enzyme is located in its optimum microenvironment in a living cell. In contrast, the chosen buffer during labeling will dictate a snapshot of proteins that are active under that condition (e.g. pH, redox or salts). However, a practical limitation of in vivo labeling is that probes do not penetrate every cell equally well and that the efficiency of protein extraction after labeling frequently varies, which demands the inclusion of many biological replicates. In addition, in vivo labeling may affect the physiology of the organism because of the toxicity of probes, and this may affect the labeling profile. However, one can use non-toxic low probe concentrations and short labeling times (e.g. for the proteasome; Kolodziejek et al. 2011 ). Nevertheless, in vitro labeling on crude extracts or subproteomes (e.g. nuclei, membranes or secretome) is easier and often causes less variation between replicates.

Chemical probes have been introduced to study 11 classes of plant proteins. These probes are summarized in Fig. 6 and in the Probe Target Database ( www.plantchemetics.org ). In the next sections, the major protein classes for which ABPP has been established and applied in plants will be discussed.

Fig. 6

Probes used in plant science. The class of targeted enzymes is indicated on the left. Structures contain a reporter tag (purple), the position for nucleophilic attack by the enzyme (red circle) and specificity determinants (explained in green). The chemotype of the reactive group is printed in red. Probe names of each type are summarized on the right with affinity tag (B, biotin; D, desthiobiotin); fluorophore (B, bodipy; R, rhodamine); and minitag (≡, alkyne; N ₃ , azide).

Papain-Like Cysteine Proteases

PLCPs are secreted or vesicular proteases of 25–40 kDa with an active site cysteine residue. PLCPs prefer substrates that have basic or hydrophobic residues at the second amino acid position preceding the cleavage site (the so-called P2 position). Arabidopsis has 30 PLCPs that fall into nine conserved subfamilies that are found across the plant kingdom ( Richau et al. 2012 ). PLCPs are regulated by pH and endogenous inhibitors (e.g. cystatins and serpins) and their activation requires the removal of the autoinhibitory prodomain ( Kuroyanagi et al. 2005 , Gu et al. 2012, Lu et al. 2015 ).

PLCPs are nearly all inhibited by the natural epoxide-based inhibitor E-64, which contains P2 = Leu and an epoxide reactive group. The first activity-based probe to profile active PLCPs in plants was DCG-04, a biotinylated derivative of E-64 developed in the Bogyo laboratory ( Fig. 6 A; Greenbaum et al. 2000 ). Since its introduction into plant science ( van der Hoorn et al. 2004 ), this probe has been used by various research groups testified by 24 plant-related publications. DCG-04 profiling revealed that the Required for C. fulvum resistance-3 (Rcr3) protease of tomato is inhibited by Avirulence-2 (Avr2) of the fungal pathogen Cladosporium fulvum ( Rooney et al. 2005 ). The same competitive ABPP assay revealed that Rcr3 is also inhibited by cystatin-like Extracellular protease inhibitor of cysteine proteases (EpiCs) from the oomycete pathogen P. infestans ( Song et al. 2009 ), and by Venom allergen-like protein (Vap1) from the pathogenic nematode Globodera rostochiensis ( Lozano-Torres et al. 2012 ). Similarly, EpiCs and Avr2 also inhibit Phytophthora -inhibited protease-1 (Pip1), a close homolog of Rcr3 ( Tian et al. 2007 , Shabab et al. 2008 , Van Esse et al. 2008 ), whereas EpiCs preferentially inhibit yet another, less related protease, called C14 ( Kaschani et al. 2010 ). Potato C14 is also a target of Vap1 ( Lozano-Torres et al. 2014 ). Further studies using competitive ABPP for investigating natural variation in Rcr3 of wild tomato species revealed that some variant residues in Rcr3 interfere with Avr2 binding ( Shabab et al. 2008 . Hörger et al. 2012 ). Similar assays were used to show that EpiC1 of Phytophthora mirabilis has adapted to new host proteases upon its host jump onto Mirabilis jalapa ( Dong et al. 2014 ). DCG-04 assays have been instrumental in these studies since they allowed quick analysis of transiently expressed proteases without the need for purification.

Comparative ABPP with DCG-04 revealed increased PLCP activities during senescence in Arabidopsis ( van der Hoorn et al. 2004 ) and wheat ( Martínez et al. 2007 ). DCG-04 profiling of senescence-associated vesicles (SAVs) isolated from tobacco leaves demonstrated that these SAVs contain active PLCPs ( Carrión et al. 2013 ). In addition, comparative DCG-04 profiling revealed that the tomato apoplast accumulates increased PLCP activity upon salicylic acid (SA) treatment ( Shabab et al. 2008 ). Likewise, DCG-04 profiling showed that maize leaves also have increased PLCP activity upon SA treatment and these PLCPs are inhibited by Corn cystatin-9 (CC9), which is up-regulated upon infection by the fungal maize pathogen Ustilago maydis ( van der Linde et al. 2012 ). Interestingly, this maize pathogen also secretes the apoplastic effector Protein involved in tumors-2 (Pit2), which inhibits maize PLCPs and is essential for disease development ( Mueller et al. 2013 ).

DCG-04 profiling also demonstrated that cathepsin B (CTB) is activated upon secretion ( Gilroy et al. 2007 ), and that the Response to Desiccation 21 (RD21) protease is inhibited by AtSerpin1 ( Lampl et al. 2010 ). Interestingly, DCG-04 profiling showed three levels of post-translational regulation of RD21: prodomain removal, granulin domain removal and enzymatic latency (Gu et al. 2012). Furthermore, DCG-04 profiling demonstrated that VPEs are not required for PLCP activation (Gu et al. 2012) and indicated that metacaspase 9 (MC9) might activate some PLCPs in the absence of Xylem-specific cysteine peptidases 1 and 2 (XCP1 and XCP2; Bollhöner et al. 2012 ). DCG-04 labeling was also used to demonstrate that the β-lactone probe IS4 specifically reacts with the RD21 protease ( Wang et al. 2008 ), and that the proteasome probe MV151 also targets some PLCPs ( Gu et al. 2010 ). Finally, DCG-04 profiling was used to demonstrate that PIRIN2 stabilizes the xylem-specific XCP2 protease, which increases susceptibility to the vascular pathogen Ralstonia stoliniferum ( Zhang et al. 2014 ).

More probes for PLCPs have been introduced more recently. Fluorescent versions of E-64 significantly improve the resolution of the activity profiling and facilitated a more reliable quantification ( Richau et al. 2012 ). MV201 carries a bodipy fluorophore and azide minitag, and MV202 a bodipy fluorophore and biotin ( Fig. 6 A). MV201 labeling showed increased PLCP activities in the tomato apoplast upon challenge with virulent and avirulent strains of the fungal pathogen C. fulvum ( Sueldo et al. 2014 ). However, under some conditions (e.g. total leaf extract at pH > 7), MV201 causes background labeling, but this can be prevented by using MV202 and purifying MV202-labeled proteins before in-gel fluorescent scanning ( Lu et al. 2015 ).

Although most PLCPs seem to have a broad cleavage specificity, some PLCPs prefer more specific cleavage sites. To study these proteases, specific substrate-inspired probes have been developed. For instance, AvrPphB is a PLCP from the bacterial plant pathogen Pseudomonas phaseolicola that is injected into the host cell as a type-III effector, becomes myristoylated and cleaves AvrPphB Susceptible-1 (PBS1) and related kinases ( Shao et al. 2003 ). AvrPphB has a unique preference for P3 = Gly; P2 = Asp, and this knowledge was used to develop a rhodamine-tagged probe (FH11; Fig. 6 B) based on the tripeptide Gly–Asp–Ala, followed by the acyloxymethylketone (AOMK) reactive group. FH11 indeed selectively labels AvrPphB expressed in Escherichia coli and P. syringae. This study revealed that this protease is also active in bacteria but that the prodomain is required for translocation into the host, and prodomain removal is essential for AvrPphB activation ( Lu et al. 2013 ). Labeling of transiently expressed AvrPphB in Nicotiana benthamiana showed a membrane-localized AvrPphB activity, consistent with its myristoylation, but also that FH11 labels endogenous plant proteins. Subsequent studies revealed that FH11, and its bodipy derivative JOGDA1 ( Fig. 6 B), specifically label plant CTBs. CTBs form a distinct family of PLCPs, represented by three genes in the Arabidopsis genome. The absence of labeling in ctb triple mutants of Arabidopsis and suppressed JOGDA1 labeling in N. benthamiana upon CTB silencing demonstrates that JOGDA1 is highly specific for CTBs, implicating that plant CTBs accept acidic residues at the P2 position of the substrates ( Lu et al. 2015 ).

In addition, a specific probe (FY01) has been validated for aleurain-like proteases (ALPs), a PLCP subclass having two representatives in Arabidopsis. FY01 contains a vinyl sulfone (VS) reactive group with an N-terminal dipeptide and a bodipy fluorophore ( Fig. 6 C, Yuan et al. 2006 ). FY01 was originally developed to profile cathepsin C in mammals ( Yuan et al. 2006 ). Plants do not have a cathepsin C ortholog, but screening of Arabidopsis mutants for the absence of labeling revealed that FY01 preferentially labels ALPs ( Lu et al. 2015 ). Labeling is specific for ALPs at neutral pH. but FY01 can also label RD21 and other cysteine proteases under some conditions. In conclusion, PLCPs can be efficiently studied with a variety of broad-range and subfamily-specific activity-based probes.

The Proteasome

The 26S proteasome is a large proteolytic protein complex that resides in the cytoplasm and nucleus, and is responsible for the degradation of selected substrates. Substrate selection is usually mediated through selective ubiquitination by E3 ligases ( Kurepa and Smalle 2008 ). The 26S proteasome consists of the 19S regulatory particle (RP) and the 20S core protease (CP). The CP consists of 28 subunits that are assembled in two rings of seven β-subunits, flanked by two rings of seven α-subunits. The two β-subunit rings create the proteolytic chamber and three of the seven β-subunits have catalytic activity: β1 cleaves proteins after acidic residues, β2 after basic residues and β5 after hydrophobic residues. Arabidopsis frequently has two genes for each proteasome subunit: β1 is encoded by PBA1 , β2 by PBB1 and PBB2 and β5 by PBE1 and PBE2.

The proteasome is a crucial protease in plant physiology as it is involved in the selective degradation of regulatory components in nearly all phytohormone signaling cascades and in most other signaling events, including the circadian clock and light and developmental signaling ( Santner and Estelle 2010 ). Although substrates are selected by the ubiquitination machinery, the activity of the proteasome itself can also be regulated. However, traditional activity assays based on fluorogenic substrates require the proteasome to be purified, and that was only achieved from soft tissues. Activity-based probes for the proteasome have made it possible to study the activity of the proteasome in crude extracts and even in living cells without purification.

Proteasome activity probes MV151, BioVS and MVA178, based on Leu–Leu–Leu tripeptides carrying a VS reactive group, were developend in the Overkleeft laboratory and were the first proteasome probes to be introduced in plant science. MV151 carries a bodipy fluorophore, BioVS a biotin and MVA178 a bodipy fluorophore and an azide minitag ( Verdoes et al. 2006 ; Fig. 6 D). In vivo labeling of Arabidopsis cell cultures with MV178 followed by click chemistry revealed that MVA178 labels both PBE1 and PBE2 ( Kaschani et al. 2009b ), whereas labeling of BioVS on Arabidopsis leaf extracts identified PBA1, PBB1 and PBE1 ( Gu et al. 2010 ). The labeling profile of Arabidopsis leaf extracts consists of three signals, and MS analysis and the suppression by subunit-selective aldehyde inhibitors (e.g. MG132, MG115 and leupeptin) revealed that these are from PBB1/2 (β2, top), PBE1/2 (β5, middle) and PBA1 (β1, bottom).

Notably, VS-based proteasome probes also label some of the PLCPs, especially in vivo ( Kaschani et al. 2009b , Gu et al. 2010 ). However, labeled PLCPs can be distinguished because they migrate at a different apparent molecular weight compared with the labeled proteasome subunits, as confirmed by the selective suppression of labeling by the PLCP inhibitor E-64d and the proteasome inhibitor epoxomicin ( Kaschani et al. 2009b , Gu et al. 2010 ). Pull-down assays demonstrate that PLCPs, RD21, RD19A and RD19B are MVA178 targets in vivo ( Kaschani et al. 2009b ), but RD21 is the major target in vitro, as confirmed by rd21-1 mutant analysis ( Gu et al. 2010 ). The detection of RD19A/B by in vivo labeling and not in vitro labeling is consistent with in vivo labeling of RD19A/B by MV201, suggesting that their active state can only be detected in vivo ( Richau et al. 2012 ). Competition experiments revealed that MV151 suppresses DCG-04 labeling of RD21 but not of Arabidopsis aleurain-like protease (AALP), confirming its selectivity for some PLCPs ( Gu et al. 2010 ).

Importantly, MV151 labeling on leaf extracts from Arabidopsis plants revealed that proteasome activity increases during SA signaling. At the protein level, the proteasome is not more abundant, but the labeling intensity increases to 1.46-fold within 1 d after treatment with the SA analog benzothiadiazole (BTH). The activation occurs in the cytoplasmic portion of the proteasome and is dependent on the nuclear SA regulator Nonexpressor of PR genes-1 (NPR1) ( Gu et al. 2010 ). MV151 profiling also indicated an increased proteasome activity in roots of pea plants that were treated with two distinct herbicides that block amino acid synthesis ( Zulet et al. 2013 ). In contrast, the proteasome activity profile does not significantly change in tomato seedlings undergoing a synchronized cell death during the hypersensitive immune response (HR) ( Sueldo et al. 2014 ).

New-generation proteasome activity probes based on the selective proteasome inhibitor epoxomicin have been introduced that carry an Ile–Ile–Thr–Leu tetrapeptide followed by an epoxyketone reactive group ( Kolodziejek et al. 2011 ). MVB003 carries a bodipy fluorophore; MVB070 a bodipy and azide minitag; and MVB072 a bodipy and biotin ( Fig. 6 E). These probes also label all three catalytic β-subunits, although time course labeling showed that these epoxyketone-based proteasome probes preferentially target β5, followed by β2 and then β1 ( Kolodziejek et al. 2011 ). The probes quickly enter living cells and label the proteasome in the cytoplasm and nucleus, visualized by confocal microscopy and isolation of nuclei ( Kolodziejek et al. 2011 ). MVB070 labeling followed by click chemistry, pull-down and MS analysis confirmed that MVB070 preferentially labels β5 (both PBE1 and PBE2) in vivo ( Kolodziejek et al. 2011 ). Importantly, pre-incubation with epoxomicin suppresses in vivo MVB003 fluorescence, indicating that fluorescence represents the location of the labeled proteasome within the plant cell.

Epoxyketone-based proteasome probes have been used to study the target of syringolin A (SylA), a cyclic non-ribosomal peptide produced by the phytopathogen P. syringae pv . syringae. SylA represents a novel class of covalent proteasome inhibitors, termed syrbactins ( Groll et al. 2008 ). Interestingly, SylA preferentially inhibits the β2 and β5 subunits, which is explained by a steric hindrance in the β1-binding pocket seen in the SylA–proteasome crystal structure ( Groll et al. 2008 ). Indeed, a rhodamine-tagged SylA (RhSylA; Fig. 6 F) also preferentially labels β2 and β5 of the proteasome, which indicates that these subunits might represent the biologically relevant target of SylA ( Kolodziejek et al. 2011 ). Structure–function analysis using competitive MVB003 labeling with synthetic SylA derivatives revealed the selectivity for targeting β2 resides in the conformation of the dipeptide tail ( Kolodziejek et al. 2011 ). Surprisingly, pre-incubation of cell cultures with SylA suppressed MVB003 fluorescence in the nucleus, suggesting that SylA targets the nuclear proteasome. Nuclear targeting by SylA was confirmed using RhSylA, which specifically accumulates in the nucleus. Nuclear targeting might be caused by the fact that the nuclear proteasome is different from the cytoplasmic proteasome. The preference of SylA for the nuclear proteasome correlates with the crucial role that the nuclear proteasome plays in SA signaling, which can be efficiently blocked by SylA ( Misas-Villamil et al. 2013 ). SylA also suppresses labeling of the proteasome of N. benthamiana but at higher concentrations than needed to block SA signaling, consistent with its preferential targeting of the nuclear proteasome ( Misas-Villamil et al. 2013 ). In conclusion, in vivo imaging and profiling of the plant proteasome has been an instrumental tool to understand how SylA manipulates the host plant.

Vacuolar Processing Enzymes

VPEs (family C13) are also called legumains or asparagnyl endopeptidase (AEPs). These cysteine proteases belong to the same clan (clan CD) as caspases (family 14A) and metacaspases (family C14B), which means that they are evolutionarily and structurally related proteases. Despite their relationship, VPEs are located to the vacuole, whereas caspases are cytonuclear. All CD clan proteases select cleavage sites following a specific amino acid residue (P1), but these P1 specificities are different. VPEs prefer P1 = Asn and occasionally P1 = Asp, whereas caspases strictly select for P1 = Asp, and metacaspases tend to prefer P1 = Arg.

Arabidopsis has only four VPEs and they are implicated in processing seed storage proteins ( Gruis et al. 2002 , Shimada et al. 2003 ), activating vacuolar enzymes ( Rojo et al. 2004 ) and in developmental programmed cell death (PCD) in the seed coat ( Nakaune et al. 2005 ) and toxin-induced PCD in leaves ( Kuroyanagi et al. 2005 ). VPEs are also required for full virulence by the biotrophic powdery mildew pathogen ( Misas-Villamil et al. 2013 ) and for successful symbiosis with the fungus Piriformospora indica ( Qiang et al. 2012 ). In other plants, VPEs are required for tomato fruit development ( Ariizumi et al. 2011 ), hypersensitive cell death in N. benthamiana ( Hatsugai et al. 2004 ), and the circularization of sunflower trypsin inhibitor SFTI-1 ( Bernath-Levin et al. 2015 ).

The relevance of VPEs in various biological processes and their well-described substrate specificity has prompted the development of activity-based probes for VPEs. VPE probes AMS101 and bAMS101 carry a bodipy fluorophore or biotin reporter tag, respectively, and are based on Pro–Asn as the P2–P1 dipeptide, followed by an aza-epoxide reactive group ( Fig. 6 G; Misas-Villamil et al. 2013 ). The proline at the P2 position prohibits cross-reactivity to PLCPs because they prefer hydrophobic residues at this position. AMS101 labels all four Arabidopsis VPEs and has no additional targets in Arabidopsis, as no labeling was detected in vpe quadrupole mutants and only VPEs were detected in pull-down assays ( Misas-Villamil et al. 2013 ). In vitro labeling required reducing agent and acidic pH, and is blocked by the caspase-1 inhibitor Ac-YVAD-cmk. Interestingly, in vivo labeling with AMS101 in Arabidopsis cell cultures and seedlings revealed speckled fluorescence in the vacuole that is absent in vpe quadrupole mutants. Furthermore, AMS101 labeling on infected Arabidopsis plants revealed that VPE activity is up-regulated upon infection with Hyaloperonospora parasitica ( Misas-Villamil et al. 2013 ). Application of AMS101 labeling on tomato indicated that VPEs also accumulate extracellularly in the leaf apoplast and that their activity profile changes upon challenge with both virulent and avirulent isolates of the fungal pathogen C. fulvum ( Sueldo et al. 2014 ). AMS101 labeling of nematode-induced syncytium in Arabidopsis roots displayed an additional VPE isoform ( Hütten et al. 2015 ).

The challenging chemical synthesis of AMS101 prompted the synthesis of JOPD1, which is much easier to synthesize and carries a bodipy fluorophore and a Pro–Asp dipeptide, followed by an AOMK reactive group ( Fig. 6 H; Lu et al. 2015 ). Aspartate can be used at the P1 position because it becomes more neutral at acidic pH and is therefore often recognized by VPEs. JOPD1 labels VPEs similarly to and equally well as AMS101, and this has been used to study VPE activities in germinating seeds ( Lu et al. 2015 ) and to show that VPE activity depends on NHX-type sodium–proton antiporters ( Ashnest et al. 2015 ).

Serine Hydrolases

SHs are a large class of enzymes that carry an active site serine residue. Most of these enzymes are proteases, acyltransferases and esterases, including lipases ( Tripathi and Sowdhamini 2006 ). Most SHs share the α/β-hydrolase fold, but other SHs are evolutionarily unrelated. Arabidopsis has approximately 250 SHs that include subtilases (family S8, e.g. SDD1, ALE1 and TPP2), Prolyl-oligo peptidase-like proteases (POPLs; family S9), serine carboxypeptidase-like proteases (SCPLs; family S10, e.g. SNG1, SNG2 and BRS1), carboxyesterases (CXEs; e.g CXE12), methylesterases (MESs; e.g. MES2 and MES3) and several other enzymes. Some SCPLs catalyze acyltransferase reactions to produce secondary metabolites (e.g. synapoyl by SNG1; Lehfeldt et al. 2000 ).

The active site serine residues in SHs can be labeled with fluorophosphonate (FP)-based probes ( Fig. 6 I; Liu et al. 1999 ). RhFP carries a rhodamine fluorophore; FPpBio a biotin; TriFP a biotin and rhodamine; and FP ≡ an alkyne minitag. Since the warhead is simple and lacks a binding group, FP probes are rather reactive probes that label all serine residues that have an elevated reactivity. These hyper-reactive serine residues reside in the active sites of SHs, created by a catalytic dyad or triad. FP probes are extremely powerful probes that easily label > 50 different SHs in a single proteome.

MS analysis of the FPpBio-labeled proteome of the Arabidopsis leaf extract reproducibly identified 45 proteins that were absent in the ‘no probe’ control ( Kaschani et al. 2009a ). Notably, all these 45 proteins were annotated SHs, confirming the remarkable selectivity of FP probes. Some of the detected SHs were known, but the majority of these SHs had not been characterized before. Labeling of the acyltransferase Sinapoylglucose accumulator-1 (SNG1) was confirmed by labeling an extract from N. benthamiana transiently overexpressing SNG1 ( Kaschani et al. 2009a ).

Application of SH profiling with a trifunctional FP probe (TriFP; carrying FP, biotin and rhodamine) on resistant wild-type Arabidopsis plants and the susceptible pad3 mutant upon infection with the necrotrophic fungal pathogen Botrytis cinerea revealed dozens of differential SH activities ( Kaschani et al. 2009a ). These differentials included increased SNG1 labeling and reduced S -formylglutatione hydrolase (SFGH) labeling, and the appearance of Botrytis -derived cutinases and lipases in the pad3 mutant. RhFP labeling also displayed differential SH activities in nematode-induced syncytium in roots of Arabidopsis ( Hütten et al. 2015 ). TriFP has also been used to study the role of the carboxyesterase CXE12 in herbicide bioactivation ( Gershater et al. 2007 ).

The power of FP probes to label so many SHs is also a disadvantage because many signals overlap upon separation on protein gels. One way to simplify the labeled proteome is by using para -nitrophenol phosphonates (NPs), that have a bulky leaving group and therefore label fewer SHs ( Nickel et al. 2012 ). Indeed, the labeling profile of Arabidopsis leaves labeled with the trifunctional NP probe (TriNP), carrying biotin and rhodamine ( Fig. 6 J), is significantly simplified. These experiments indicated that CXE12 is the major target of NP probes. Indeed, transient overexpression of CXE12 in N. benthamiana confirms that this enzyme is labeled by TriNP ( Nickel et al. 2012 ), and the major signal disappears from Arabidopsis leaves upon labeling leaf extracts of cxe12 mutant plants ( Kaschani et al. 2012a ). CXE12 labeling is blocked upon pre-incubation with paraoxon, a chemical frequenctly used in agroindustry ( Nickel et al. 2012 ). Further studies of phosphate- and phosphonate-based agrochemicals revealed a remarkable selectivity of these chemicals in suppressing labeling of a subset of plant SHs ( Kaschani et al. 2012a ). This selectivity was confirmed using transiently expressed SH and indicates that agrochemicals are likely to inactivate diverse SHs in the plant. These experiments also indicate that there is scope for the further development of probes that target distinct SHs of this superfamily.

Comparative TriNP profiling and pull-down assays revealed that many SHs are activated at an early stage of the HR in synchronously dying tomato seedlings, before tissue collapse occurs ( Sueldo et al. 2014 ). HSR203 and other CXEs, and also subtilases P69B and P69C are among the activated enzymes, whereas labeling of an S10/SCPL is reduced before tissue collapse ( Sueldo et al. 2014 ). Similar drastic early changes in the activity profile were observed in the extracellular proteome of dying seedlings and in leaves of both resistant and susceptible tomato plants upon inoculation with the fungal pathogen C. fulvum ( Sueldo et al. 2014 ).

The large number of SHs that are labeled in a given proteome makes the comparison of the labeling profiles in two samples very challenging. To overcome this limitation, ABPP-DIGE has been introduced using alkyne-tagged FP (≡FP) probes that can be coupled to different commercially available azide-tagged fluorophores using Cu(I)-catalyzed click chemistry ( Fig. 3 ; Hong and van der Hoorn 2014 ). Mixing labeled proteomes pair wise allows the simultaneous separation on the protein gel and the detection of each labeled proteome by selecting specific excitation and emission wavelengths during fluorescence scanning. This technique displayed differential SH activities in the apoplast of N. benthamiana upon infection with P. syringae ( Hong and van der Hoorn 2014 ).

Matrix Metalloproteases

MMPs are M10 proteolytic enzymes that use a Zn ²⁺ cation for catalysis. These enzymes are usually secreted but remain anchored to the plasma membrane either by an integral transmembrane domain or by a glycosylphosphatidylinositol (GPI) anchor. Reverse genetics experiments has implicated plant MMPs in both symbiosis and immunity ( Flinn 2008 ). Arabidopsis has five MMPs.

There are no known irreversible inhibitors of MMPs, but MMPs can be blocked by hydroxamate-based reversible inhibitors, such as marimastat. To facilitate a covalent linkage that is required for the detection and purification of probe-labeled MMPs, marimastat has been equipped with a benzophenone photoreactive group ( Lenger et al. 2012 ). UV irradiation activates the benzophenone group, resulting in a covalent bond with the enzyme substrate-binding groove.

Labeling and pull-down experiments with the marimastat–benzophenone–biotin probe JL01 ( Fig. 6 K) resulted in labeling of At2-MMP, At4-MMP and At5-MMP, when transiently overexpressed in N. benthamiana ( Lenger et al. 2012 ). Labeling was suppressed upon pre-incubation with an excess of marimastat and in the absence of MMP overexpression ( Lenger et al. 2012 ). This proof-of-concept remains to be exploited by studying endogenous MMPs e.g. by pelleting membranes from labeled total extracts. In addition, minitagged photoaffinity probes for MMPs have been used on mammalian cell lysates ( Sieber et al. 2006 , Yu et al. 2013 ), but remain to be validated in plant science.

Retaining Glycosyl Hydrolases

Retaining glycosidases are a large superfamily of glycosyl hydrolases (GHs) that hydrolyze glycosidic bonds in between glycons and aglycons, without changing the stereochemistry of the released sugars. There are genes in the Arabidopsis genome encoding >250 retaining glycosidases that belong to 24 GH families. These enzymes have been classified based on the glycan they recognize, e.g. xylosidases, glucosidases and fucosidases. These enzymes play diverse roles in plants, ranging from cell wall and protein modification to hormone and metabolite regulation.

Retaining GHs hydrolyze substrates using two glutamic acid residues that reside on opposite sites in the substrate-binding groove. One glutamate acts as a nucleophile and the other as a base. Activity-based probes based on cyclophellitol aziridine lock the catalytic mechanism in the covalent intermediate state, resulting in an ester bond. The first cyclophellitol aziridine probes were designed to target glucosidases and were JJB70 and JJB111, which carry a bodipy or biotin, respectively ( Fig. 6 L; Kallemeijn et al. 2012 ). It was found, however, that these probes also label plant GHs that act on related glycans, such as xylosidases and galactosidases ( Chandrasekar et al. 2014 ). Labeling Arabidopsis leaf extracts causes strong signals derived from myrosinases TGG1 and TGG2, which were absent in tgg1 and tgg2 mutant plants. Another 18 GHs were detected by MS at lower intensity. Myrosinases cleave the S -glucosidic bond in glucosinolates to release toxic cyanide compounds that act against herbivorous insects ( Barth and Jander 2006 ). Manual annotation of the fragmentation spectrum of the labeled peptide of TGG1 confirmed labeling of the glutamate active site ( Chandrasekar et al. 2014 ). JJB70 also labels GHs in other plant species and in vivo.

MS analysis of JJB111-labeled proteins from the apoplastic proteome of N. benthamiana revealed the presence of 16 GHs, including xylosidases, glucosidases and galactosidases ( Chandrasekar et al. 2014 ). Interestingly, the labeling of some of these enzymes is reduced in the apoplast of plants infected with P. syringae (B. Chandrasekar and R. Van der Hoorn, unpublished data). Convolution ABPP indicates that there is an inhibitor in the apoplast of infected plants that can suppress galactosidase labeling in the apoplast of non-infected plants.

ATP-Binding Proteins

ATP hydrolysis is the engine of every cell. Plant genomes encode thousands of proteins that bind and hydrolyze ATP, including 1,100 protein kinases as well as metabolic kinases and ATP-driven transporters. Kinases transfer the γ phosphate of ATP onto other cellular components, and this terminal phosphate is stabilized by one or more lysine residue that is often conserved, e.g. in the P-loop of protein kinases.

Probes for ATP-binding proteins (BHAcATP and DBAcATP; Fig. 6 M; Patricelli et al. 2007 ) contain ATP linked to biotin or desthiobiotin via an acyl linker at the γ phosphate . Binding of these AcATP probes to ATP-binding pockets places the acyl group in close proximity to the lysine residues that stabilize the γ phosphate. These lysines react with the acyl group, resulting in a covalent amide bond between the probe and the protein. These probes therefore detect the availability of the ATP-binding pockets. Labeling is suppressed by ATP-mimicking inhibitors, but labeling does not necessarily report the enzymatic activity of the protein ( Patricelli et al. 2007 ).

Purification and separation of BHAcATP-labeled Arabidopsis leaf extracts on protein gels, followed by MS analysis revealed 112 labeled proteins ( Villamor et al. 2013 ). Most identified proteins were known to bind ATP, and these included several (receptor-like) protein kinases, ATP-driven transporters and chaperones. Interestingly, two receptor-like kinases were detected at about half of their predicted molecular weight, suggesting that they resulted from a receptor shedding event ( Villamor et al. 2013 ). Labeled peptides were detected, but only when the search included a modification with an oxidized biotin. Mapping the labeled peptides often confirmed that labeling occurred in the ATP-binding pocket in close proximity to the γ phosphate, but labeling also occurred outside known nucleotide-binding pockets.

Because AcATP probes also label surface-exposed lysine residues, the analysis requires purification and sequencing of labeled peptides rather than proteins. Desthiobiotinylated AcATP probes have been used for this analysis to ensure efficient peptide elution and detection, because desthiobiotin cannot be oxidized and has a lower affinity for streptavidin than biotin. Sequencing labeled peptides revealed 242 labeling sites, including 24 peptides from protein kinases such as receptor-like kinases, calcium-dependent protein kinases and mitogen-activated protein kinases ( Villamor et al. 2013 ). Most of the labeled peptides are derived from metabolic kinases, ATP-based transporters and chaperones, and further analysis using crystal structures revealed that these were nearly always labeled in the ATP-binding pocket. In conclusion, AcATP labeling is a powerful technique to monitor the availability of ATP-binding pockets, but it requires deep sequencing of labeled peptides to detect protein kinases.

N-Terminal Labeling by β-Lactone Probes

In an attempt to develop novel probes for cysteine and serine proteases, probes have been tested based on a reactive β-lactone in a dipeptide backbone with a biotin affinity tag (IS probes; Fig. 6 N; Wang et al. 2008 ). These β-lactone probes were based on β-lactone inhibitors of lipases and proteases and contained a cyclic threonine residue preceded by an amino acid. These probes, and especially IS4, caused intensive labeling of many proteins in Arabidopsis leaf extracts. MS analysis of the IS4-labeled proteins revealed that one of the most dominantly labeled proteins was the PSII subunit P (PsbP), a component of the oxygen-evolving complex of PII ( Wang et al. 2008 ).

Analysis of fragmentation spectra revealed that the IS4 probe was linked to the N-terminal alanine of mature PsbP through a peptide bond. Although puzzling at first, it was soon discovered that this fusion required a PLCP because labeling could be blocked by E-64 and other PLCP inhibitors. Screening of Arabidopsis mutants lacking individual PLCPs revealed that labeling is absent in the rd21-1 mutant, which lacks the abundant protease RD21 ( Wang et al. 2008 ). Addition of heterologously produced RD21 could restore labeling in leaf proteomes of rd21 mutant plants and in E-64-treated proteomes of wild-type plants. Since thioester intermediates are common for cysteine proteases when acting on natural peptide substrates, it was tested if RD21 could also ligate peptides to the N-terminus of other proteins. A biotinylated peptide based on the sequence of IS4 and the N-terminus of PsbP was indeed ligated onto other proteins in leaf extracts, and this activity was absent upon treatment with E-64 and in extracts of rd21-1 mutant plants, and was complemented by adding recombinant RD21. In conclusion, IS4 labeling does not reflect the activity of the labeled proteins, but this probe revealed an unexpected property of the RD21 protease—its ability to transligate peptides. The physiological role of this transpeptidation reaction remains to be elucidated.

Glutathione S -Transferases

SF probes were also generated in an attempt to develop new probes for serine proteases. SF probes are based on the serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF). Besides the SF reactive group, DS6B carries a biotin and DS6R a rhodamine ( Fig. 6 O; Gu et al. 2013 ). DS6B causes strong labeling in Arabidopsis leaf extracts, but MS analysis of DS6B-labeled proteins revealed no serine proteases, but a long list of unrelated proteins ( Gu et al. 2013 ). GSTs were consistently found among the labeled proteins, causing a strong 24 kDa signal in the labeling profile. Analysis of the peptide fragmentation data revealed that SF probes consistently labeled GSTs at tyrosine residues situated in the promiscuous substrate-binding pocket. Although the sequence of these substrate-binding sites is unconserved, the consistent labeling of a tyrosine in this pocket, irrespective of the GST superfamily, indicated that these probes highlight a hyper-reactive tyrosine residue that may have a functional role in GSTs. Indeed, SF probes only label Y111 in the model SjGST protein, and the Y111F mutant cannot be labelled and is inactive in the conjugation reaction. Thus, although DS6B did not display labeled serine proteases, it did highlight functionally relevant tyrosine residues in unrelated proteins, consistent with the concept of reactivity profiling. Intriguingly, SF probes also label tyrosine residues near active sites of other enzymes, such as aldo-ketoreductase AKR4C8 and alkenal-double-bond-reductase DBR1, indicating that these residues play important roles in these enzymes ( Gu et al. 2013 ).

Glutaraldehyde 3-Phosphate Dehydrogenases

Glutaraldehyde 3-phosphate dehydrogenases (GAPDHs) are important enzymes in glycolysis that carry an active site cysteine residue that performs nucleophilic attack on the aldehyde carbonyl. A selective probe for GAPDH was discovered by coincidence. This probe, Mrl ( Fig. 6 P), was synthesized during an attempt to produce the proteasome inhibitor SylA. Because this hydantoin by-product contained a Michael system, it was tested whether this molecule labels anything in a plant proteome ( Kaschani et al. 2012b ). Mrl-Bio carries a biotin; Mrl-Rh a rhodamine; and Mrl≡ an alkyne minitag. Interestingly, a single, pH-dependent signal was detected, and purification of Mrl-labeled proteins revealed that this signal is generated by two cytosolic variants of GAPDH: GAPC-1 and GAPC-2. Labeling was confirmed using heterologously expressed GAPC-1 and GAPC-2 proteins, and gapc-1 mutants of Arabidopsis. Labeling is suppressed by the glyceraldehyde 3-phosphate substrate and by oxidizing agents, consistent with active site labeling. Thus, unexpectedly, Mrl labels GAPDHs in an activity-dependent manner.

Aldehyde Dehydrogenases

Aldehyde dehydrogenases (ALDHs) are a large superfamily of enzymes that use nicotinamide cofactors to dehydrogenate aldehydes in various metabolites into a less toxic carboxylic acid. They carry an active site cysteine residue that performs a nucleophilic attack on the aldehyde carbonyl, while the cofactor donates hydrogen atoms. The active site cysteine of ALDHs can be labeled using a rhodamine-tagged chloroacetamide (CA) probe, CA-Rh ( Fig. 6 Q; Stiti et al. 2016). Heterologously expressed and purified Arabidopsis ALDH3H1 is labeled by CA-Rh only at the active site cysteine and labeling is suppressed by ALDH inhibitors and cysteine-modifying agents. Enzymatic activities measured under the same conditions revealed strong correlations between active site labeling and enzymatic activity, with one important exception: labeling of ALDH3H1 by CA-Rh is suppressed by the cofactor NAD ⁺ . Cofactors that do not bind ALDH3H1 do not suppress active site labeling, and the suppression of labeling by the corresponding cofactors was also detected for members of other ALDH superfamilies. Importantly, NAD(P) also strongly suppress Mrl-Rh labeling of GAPDH (see above), revealing that suppression of active site labeling by cofactors is common for unrelated ALDHs and detected by distinct probes. This phenomenon not only indicates an unexpected cofactor-dependent regulatory mechanism in ALDHs, but may also allow the identification of cofactors for large numbers of proteins, simply by monitoring the suppression of labeling by cofactors.

Conclusions and Prospects

The application of ABPP in plant science is only just beginning. The commercial availability of the probes, increased analytical power and validation of novel probes will increase their applications in diverse areas in various plant species. These projects have already uncovered unexpected regulatory mechanisms in plant physiology, ranging from germination and senescence, to immunity and disease caused by various pathogens ( Fig. 7 ). These are just the very first examples of discoveries to be made using ABPP in plant science. More applications will follow when ABPP is applied to study, for example, abiotic stress, symbiosis and industrial processes.

Fig. 7

Biological phenomena uncovered by applying ABPP in plant science. This figure summarizes the major discoveries in plant science made by ABPP so far, as discussed in this review. Plant species: At , Arabidopsis thaliana ; Mj , Mirabilis jalapa ; Nb , Nicotiana benthamiana ; Nt , Nicotiana tabacum (tobacco); Ps , Pisum sativum (pea); Sl , Solanum lycopersicum (tomato); St , Solanum tubersum (potato); Ta , Triticum aestivum (wheat); Zm , Zea mays (maize). References: [1] Lu et al. (2015) ; [2] Villamor et al. (2013) ; [3] Wang et al. (2008) ; [4] Van der Hoorn et al. (2004) ; Martinez et al. (2007); Carrion et al. (2013); [5] Zulet et al. (2013) ; [6] Kaschani et al. (2012a) ; [7] Gu et al. (2010) ; [8] Shabab et al. (2008) ; van der Linde et al. (2012) ; [9] Kaschani et al. (2009a) ; Sueldo et al. (2014) ; [10] Rooney et al. (2005); Shabab et al. (2008) ; Van Esse et al. (2008); Mueller et al. (2013); Hörger et al. (2013); [11] Dong et al. (2014) ; [12] Misas-Villamil et al. (2013) ; [13] Tian et al. (2007) ; Song et al. (2009) ; Kaschani et al. (2010) ; [14] (B. Chandrasekar and R. Van der Hoorn, unpublished data) (2016); [15] Zhang et al. (2014) ; [16] Lu et al. (2013) ; [17] Kolodziejek et al. (2011) ; Misas-Villamil et al. (2013) ; [18] Lozano-Torres et al. ( 2012 , 2014 ); [19] Hütten et al. (2015) .

Funding

Financial support was received from the Japanese Society for the Promotion of Science; the University of Oxford; the European Research Council [ERC consolidator grant ‘GreenProteases’].

Disclosures

The authors have no conflicts of interest to declare.

Abbreviations

Abbreviations
 
• ABPP

  activity-based protein profiling

 
• ALDH

  aldehyde dehydrogenase

 
• ALP

  aleurain-like protease

 
• Avr2

  Avirulence-2

 
• AvrPphB

  Avirulence protein Pseudomonas phaseolicola -B

 
• CTB

  cathepsin B

 
• DIGE

  differential gel electrophoresis

 
• EpiC

  Extracellular protease inhibitor of cysteine protease

 
• FP

  fluorophosphonate

 
• GAPDH

  glutaraldehyde 3-phosphate dehydrogenase

 
• GST

  glutathione S -transferase

 
• MMP

  matrix metalloprotease

 
• MS

  mass spectrometry

 
• NP

  para -nitrophenol phosphonate

 
• PLCP

  papain-like cysteine protease

 
• Rcr3

  Required for C. fulvum resistance-3

 
• SA

  salicylic acid

 
• SF

  sulfonyl fluoride

 
• SH

  serine hydrolase

 
• SylA

  syringolin A

 
• TriFP

  trifunctional fluorophosphonate

 
• TriNP

  trifunctional para -nitrophenol phosphonate

 
• VPE

  vacuolar processing enzyme

 
• VS

  vinyl sulfone