Research Article

Damage on plants activates Ca2+-dependent metacaspases for release of immunomodulatory peptides

See allHide authors and affiliations

Science  22 Mar 2019:
Vol. 363, Issue 6433, eaar7486
DOI: 10.1126/science.aar7486

Rapid response to tissue damage

Damaged plants are susceptible to microbial attack. In response to physical damage, plants proactively generate signal peptides to activate their immune systems. Hander et al. examined wound responses in the model plant Arabidopsis thaliana. They identified a metacaspase that releases an immunomodulatory signal peptide from its precursor form within 30 seconds of the damage. The metacaspase itself was activated by a burst of calcium released by tissue damage.

Science, this issue p. eaar7486

Structured Abstract


Cellular damage caused by wounding triggers signals to alert the surrounding tissue. As a universal process in all multicellular organisms, these signals activate the immune system to prevent infection and promote tissue regeneration, eventually leading to wound healing, but they have to be secured because aberrant immune stimulation can negatively affect health and growth. Plants, as sessile organisms, are regularly subject to chewing or sucking insects and physical damage inflicted by metazoans or exposure to the environment. In plants, short protein fragments or peptides can have immunomodulatory functions, such as those derived from the plant elicitor peptide (Pep) gene family. Peps are part of precursor proteins and have been proposed to act as wound signals that bind and activate the extracellular Pep receptors (PEPRs) to initiate an immune-like response. How Peps are produced and released upon wounding and how far this response extends from the harm site remain unclear.


Proteases can generate peptides from precursor proteins, but the genome of the model plant, Arabidopsis thaliana, encodes approximately 600 different proteases. To understand the molecular mechanisms that control the immune-activating signals, it is essential to identify the immune response-contributing proteases. Furthermore, wound formation has to be observed with great speed and precision to delineate the extent of the overall process. Pep1 was studied here as a representative member of the gene family.


Pep1 generation was detected within 30 s after the injury, peaking at 5 min and lasting up to 1 hour in two wound model experiments on Arabidopsis seedlings: pinching with forceps or mechanical disruption of tissue integrity through grinding. Screening with various protease inhibitors revealed the involvement of metacaspases in Pep1 formation. Metacaspases are cysteine proteases, conserved in plants, fungi, protists, and bacteria that were historically named after caspases because of a certain degree of structural homology. However, they differ in activity and substrate specificity. Metacaspases require low-millimolar amounts of calcium (Ca2+) for in vitro activity and cleave their substrate proteins after the amino acids arginine and lysine. Physical damage activated the abundant METACASPASE4 (MC4) in both wound experiments. MC4 released Pep1 from its protein precursor PRECURSOR OF PEP1 (PROPEP1) by cleaving it behind a conserved arginine. A mutant lacking MC4 was unable to produce Pep1 in leaf tissue, whereas certain redundancy occurred with other metacaspase (or metacaspase-like) activities in root tissue. Inside the plant cell, PROPEP1 is attached to the cytosolic side of the vacuolar membrane. When undisturbed, cytosolic Ca2+ concentrations ([Ca2+]cyt) are too low to activate metacaspases that need unusually high [Ca2+] to function. When Arabidopsis root epidermis cells were damaged by means of multiphoton laser ablation, Pep1 release from the vacuolar membrane occurred only in the cytosol of directly hit cells, as observed with confocal microscopy. Loss of plasma membrane integrity in these cells led to a high and prolonged influx of extracellular [Ca2+] into the cytosol, sufficient to activate metacaspases, leading to the cleavage and release of Pep1 from PROPEP1. Application of exogenous PROPEP1 protein fragments, longer than the native Pep1, reduced root growth—a known negative effect of Pep1 overload—irrespective of MC4 cleavage. Accordingly, PROPEP1 cleavage to overcome retention of Pep1 at the vacuolar membrane seems more important than obtaining the mature Pep1 size.


Plants exploit highly conserved mechanisms—such as Ca2+-partitioning across intact membranes and maturation of immunomodulatory peptides by proteases—to rapidly trigger defense responses against tissue damage. Pep1 is released by activation of MC4 upon prolonged high levels of [Ca2+]cyt that occur only in directly damaged cells, and this response is safeguarded by subcellular retention of PROPEP1 at the vacuolar membrane in the absence of damage. Metacaspases, together with Peps and PEPRs, now emerge as potential targets for breeding and improving crop immunity.

Metacaspases activate defense responses upon wounding.

Undisturbed root epidermal cells contain inactive zymogen MC4 (zMC4) in the cytosol and PROPEP1 attached to the vacuolar membrane. Damage induces a prolonged influx of Ca2+ in the cytosol, triggering MC4 to cleave PROPEP1 and to release Pep1. Pep1 can then diffuse to neighboring cells and bind to the receptor, PEPR, to activate a defense response (bottom, light green cells). Pep1 release occurs only in cells that lose plasma membrane integrity (bottom, red cells).


Physical damage to cells leads to the release of immunomodulatory peptides to elicit a wound defense response in the surrounding tissue. In Arabidopsis thaliana, the plant elicitor peptide 1 (Pep1) is processed from its protein precursor, PRECURSOR OF PEP1 (PROPEP1). We demonstrate that upon damage, both at the tissue and single-cell levels, the cysteine protease METACASPASE4 (MC4) is instantly and spatiotemporally activated by binding high levels of Ca2+ and is necessary and sufficient for Pep1 maturation. Cytosol-localized PROPEP1 and MC4 react only after loss of plasma membrane integrity and prolonged extracellular Ca2+ entry. Our results reveal that a robust mechanism consisting of conserved molecular components links the intracellular and Ca2+-dependent activation of a specific cysteine protease with the maturation of damage-induced wound defense signals.

Physical trauma of plants can be provoked by, for example, insect chewing, animal feeding and trampling, or weather damage, creating entry sites for various pathogens. From damaged plant cells, a multitude of cellular constituents that act as signaling factors are released to alert and activate a defense response in the surrounding tissues against invading pathogens. These signaling factors, or damage-associated molecular patterns (DAMPs), include extracellular adenosine 5′-triphosphate (ATP) and DNA, cell wall–derived molecules, peptides, and glutamate (1, 2).

In Arabidopsis thaliana, the plant elicitor peptides (Peps) were proposed to act as DAMPs on the basis of their intracellular origin and ability to elicit defense responses when applied as synthetic peptides (3). The eight Arabidopsis Peps are embedded in the C terminus of their respective precursor proteins, the PROPEPs (4). After the supposed proteolytic maturation of PROPEPS, the Peps are perceived by two plasma membrane–localized leucine-rich repeat receptor-like kinases (LRR-RLKs) named PEP RECEPTOR 1 (PEPR1) and PEPR2 (5, 6). Upon Pep perception, PEPRs interact with the coreceptor BRI1-ASSOCIATED KINASE1 (BAK1) (6, 7) to induce a typical innate immunity–like response, leading to an oxidative burst, ethylene (a gaseous plant stress-related hormone) production, and pathogen defense–related gene expression (8). In the absence of both PEPRs, the pathogen and wound defense responses are reduced, increasing pathogen sensitivity and insect herbivory (912). Although the role of Peps as wound-immunomodulatory peptides is conserved across diverse plant families, including in the monocot Zea mays (maize) (13), the mechanism by which Peps are released from their precursors has been elusive. We demonstrate that the cysteine protease METACASPASE4 (MC4) is necessary and sufficient for cleavage and release of Pep1 from PROPEP1 in an immediate, spatiotemporally controlled, Ca2+-dependent manner.

Damage induces instant PROPEP1 processing

To assess the Pep maturation mechanism, we established two experimental setups in which the PROPEP cleavage kinetics after plant tissue damage were monitored and perturbed. Whole seedlings of transgenic plant lines expressing PROPEP1-YFP (yellow fluorescent protein) fusion proteins (8) were damaged with serrated forceps or by thawing of snap-frozen and crushed tissue (Fig. 1A). In both experimental setups, we evaluated the PROPEP1-YFP cleavage kinetics through immunoblot analysis. Prompt handling of the samples (Materials and methods) was essential to prevent cleavage during or after protein extraction. Upon thawing of the frozen tissue samples, the PROPEP1-YFP band was processed into a band with the approximate molecular weight of a Pep1-YFP fusion protein (PEP1-YFP), peaking at 5 min, followed by general protein degradation (Fig. 1, C and D, and fig. S1A). Accumulation of PEP1-YFP was evident after 30 s (Fig. 1D and fig. S1A). In the forceps-damaged samples, PEP1-YFP cleavage products accumulated more evenly over time compared with thawed tissue powder (Fig. 1B).

Fig. 1 Rapid PROPEP1 processing to Pep1 in damaged plants.

(A) Visual representation of the wound model experiments. (B and C) PROPEP1-YFP (top arrowhead in the immunoblot) is processed to PEP1-YFP (bottom arrowhead) after incubation of (B) forceps-wounded seedlings or (C) crushed tissue powder of whole seedlings at room temperature for the indicated time (minutes). Ponceau S stain of the rubisco large subunit (rbcL) indicates protein load. (D) Immunoblot band intensity quantification of PROPEP1-YFP and PEP1-YFP over time in crushed tissue powder. Time points indicate the mean of five biological replicates ± standard error on the mean (SEM). (E). PROPEP1-YFP and PEP1-YFP band intensity, 5 min after transfer of crushed tissue powder to room temperature in the presence of water or the indicated protease inhibitor. Mean band intensities are shown for three biological replicates + SEM. Statistics are described in the Materials and methods.

Metacaspase activity correlates with Pep1 cleavage specificity

To identify the protease activities necessary to process PROPEP1, we used a pharmacological approach in which whole seedlings were infiltrated with a range of protease inhibitors and ion-chelators 10 min before the damage. Whereas the majority of inhibitors did not affect PROPEP1-YFP cleavage (fig. S1C), PEP1-YFP fragment formation was 96% reduced on average by infiltration of either the ion chelators EGTA and EDTA or the metacaspase inhibitor Z-VRPR-fmk (Fig. 1E and fig. S1C) (14).

Metacaspases are a family of cysteine-dependent proteases that together with the metazoan paracaspases, such as mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1), were suggested to behave as caspase-like proteins (15). This view has been challenged because unlike caspases that cleave C-terminally to aspartic acid, metacaspases cleave their protein substrates C-terminally to the basic amino acids arginine (R) and lysine (K) embedded in a double basic (R/K)x(R/K) cleavage signature (16). Deviation from this pattern occurs, for example, for MC9 substrates (17, 18), but the cleavage takes place after R or K. PROPEP1 also lacks a double basic motif. However, mutation of the conserved arginine (R69 in Arabidopsis PROPEP1) that precedes the demonstrated or predicted mature Peps (Fig. 2A) (3, 19) to glutamate (E) or alanine (A) blocked the degradation of the PROPEP1-YFP protein to PEP1-YFP in vivo (Fig. 2, B and C). An R-specific cleavage occurred because cleavage was impaired similarly by a mutation into the homotypic amino acid K (Fig. 2B).

Fig. 2 Activation of MC4 upon damage and cleavage of PROPEP1 behind R69 to release Pep1 in vitro and in vivo.

(A) Multiple sequence alignment of the eight Arabidopsis PROPEPs. Previously identified Peps (with mass spectrometry) are underlined in red. Arginine 69, R69. (B) Replacement of PROPEP1 (PP1) R69 with A, K, or E impaired forceps-induced accumulation of PEP1-YFP in vivo. R69A mutation seemed to destabilize PROPEP1-YFP, leading to its accelerated degradation after wounding. (C) R69E mutation impaired PEP1-YFP accumulation at least up to 1 hour after damage in both wound model experiments. Extracts from Arabidopsis lines expressing YFP and PEP1-YFP were loaded to indicate their respective band heights. (D) Anti-MC4 immunoblots of both wound model experiments for the indicated time points. The top bands marked with (<) and (>) are unspecific (more information on anti-MC4 specificity is provided in fig. S1, E and F). (E) In vitro TNT-protease assay of PROPEP1 (PP1) and mutant versions incubated with increasing amounts of recombinant MC4 (rMC4) or its inactive mutant rMC4C/A (active-site cysteine mutated to alanine). The arrow and arrowhead represent PROPEP1 fused to glutathione S-transferase (PROPEP1-GST) and PEP1-GST, respectively. At increased rMC4 concentrations extended cleavage occurs at a downstream site, potentially the lysine (K)–rich region between R69 and the conserved Pep motif SSG-(R/K)x1-G-x2-N, resulting in a close-by lower band indicated with a dash. The band marked with an asterisk is most probably an alternative initiation product from a downstream methionine (M59) because it cleaved in the wild type as well but not in the mutant versions R69A and R69E. (F) Relative band intensity profiles of the images in (E) overlaid for the different rMC4 concentrations per construct. The black trace is the rMC4 C/A lane, and light gray is the 500 nM rMC4 lane. (G) Intensity profile overlay relative to the first peak for the indicated concentrations and constructs. (H to K) Quantification of mean immunoblot band intensities for three biological replicates + SEM. (H) PROPEP1-YFP, (I) PEP1-YFP (arrowhead), (J) MC4 p20*, and (K) MC4 p20 subunit (arrowhead), 5 min after damage by forceps. Statistics are described in Materials and methods. rbcL indicates protein load.

Damage activates MC4

The Arabidopsis genome encodes nine metacaspases, classified according to their domain organization and biochemical characteristics. Type I metacaspases—MC1, MC2, and MC3—contain a proline/glutamine–rich repeat domain and a zinc finger motif in their N-terminal prodomain. MC1 and MC2 are involved in a plant-specific programmed cell death (PCD), which is typical for a hypersensitive response (20), but their activation mechanism and proteolytic requirements remain unknown (21). Type II metacaspases, MC4 to MC9, lack the N-terminal prodomain, and most require Ca2+ concentrations in the low millimolar range and a pH optimum of 7.5 for activation in vitro and autocatalytic cleavage to p20 and p10 domains (designated according to the caspase nomenclature for their approximate molecular mass in kilodaltons). Crystal structures for type-I metacaspases from Trypanosoma brucei and Saccharomyces cerevisiae revealed that Ca2+ binding to four aspartate residues induced a conformational shift and was necessary for auto-activation (22, 23). On the basis of homology to type I metacaspase sequences, point mutation and domain swap experiments revealed similar conserved aspartate residues important for the activity in tomato (Solanum lycopersicum), tobacco (Nicotiana tababum), and Arabidopsis type II metacaspases (2426). By contrast, MC9 does not depend on Ca2+ and functions optimally at pH 5.5 (16, 27). Considering the ubiquitous expression pattern of Arabidopsis PROPEP1 and MC4 (8, 28), which would be a fitting prerequisite for a general wound-response regulator, we wondered whether damage activates MC4. With a MC4-specific antibody that recognizes both the zymogen (zMC4) and the activated p20 domain (fig. S1, E and F) (27, 28), autocatalytic MC4 processing was monitored during damage (Fig. 2D; quantified in Fig. 2, I and K). A MC4-derived band at an approximate size of 30 kDa, which we designated p20*, decreased in intensity (Fig. 2, D and I), most probably because the p20-p10 linker region C-terminal of p20* is trimmed, resulting in a p20 size of ~26 kDa. This observation is in agreement with N-terminal peptides (Table 1) (29) and endogenous peptides from peptidomics (table S1) identified by means of mass spectrometry and indicative of cleavage events upstream of the initial autocatalytic cleavage at K255 (27). Previously, trimmed p20 protein fragments were also identified in vitro and found to be active against a chemical test substrate, biotin-FPR-cmk (27). Therefore, we used the differential accumulation of the p20* and p20 bands in time as a proxy for MC4 activation. MC4 was activated with kinetics similar to those of PROPEP1 cleavage (fig. S1B). Both events were blocked by the addition of EDTA and EGTA (fig. S1D). Z-VRPR-fmk did not inhibit p20 accumulation (fig. S1D) but inhibited PROPEP1 cleavage (fig. S1C), indicating a sequence of events: Ca2+ binds zMC4 to induce autocatalytic cleavage [p20 accumulation is inhibited by EDTA and EGTA (fig. S1D)], and subsequently, active MC4 can cleave substrates, such as PROPEP1, or inhibitory substrate analogs, such as Z-VRPR-fmk.

Table 1 N-terminal peptides in between the active site cysteine (position 139) and initial autocatalytic cleavage site K225 of MC4.

PTM viewer (29) and public available proteomics data were queried for MC4 N-terminal peptides upstream of K225 (peptide in position 226 was also identified). P4-P1, four amino acids preceding the putative cleavage site; Peptide, N-terminally labeled peptides (Nt) or endogenous peptides; Workflow, protocol used to isolate the peptides: combined fractional diagonal chromatography (COFRADIC), terminal amine isotopic labeling of substrates (TAILS), charge-based fractional diagonal chromatography (ChaFRADIC).

View this table:

MC4 cleaves PROPEP1 at R69

To assay direct MC4-dependent maturation of Pep1, we tested in vitro whether purified recombinant His-tagged MC4 could process a PROPEP1–glutathione S-transferase (GST) fusion protein. MC4 efficiently cleaved PROPEP1-GST within the nanomolar range (Fig. 2E). Similarly to the in vivo situation, mutated PROPEP1R69A/E-GST cleavage sites were not cleaved by MC4 in vitro (Fig. 2E). The homotypic R69K mutation was cleaved, albeit less efficiently: The PROPEP1-GST band decreased, and PEP1-GST increased 50%, whereas when R69 is mutated to K, the corresponding bands decreased and increased 30% at a concentration of 62.5 nM recombinant MC4 (rMC4) (Fig. 2G). Despite the occurrence of extra cleavage sites downstream of R69, based on the size and intensity of the cleavage products (Fig. 2, F and G), R69 of PROPEP1 remains the preferred MC4 cleavage site, both in vivo and in vitro.

Other PROPEP1-YFP cleavage products, such as the double band just below PEP1-YFP, can be distinguished by immunoblots (Fig. 1, B and C). Anti-GFP immunoprecipitation combined with mass spectrometry revealed these bands (fig. S2, A and F, bands 4 and 5) as cleavage products downstream of R69 (table S2). The bands are mostly present under steady-state conditions (time point 0 in the PROPEP1-YFP immunoblots) and become more abundant after damage (fig. S2G). At least during forceps-induced damage, accumulation of band 4 might depend on MC4 activity because it accumulated less in a MC4-deficient mutant mc4−/− (fig. S2G) but is not necessarily a direct target of MC4 and because in crushed seedlings, it accumulated irrespective of MC4 (fig. S2C). MC4 may initially cleave at R69, followed by downstream proteolysis of PEP1-YFP by MC4 (Fig. 2, E to G) or other undefined proteases. Furthermore, different cleavages occur in crushed seedlings when compared with forceps-treated seedlings (fig. S2C), such as a peptide (fig. S2, A and F, band 2 and 7) that was detected with endogenous peptidomics (table S1) and can be distinguished by a human influenza hemagglutinin (HA) tag C-terminally to YFP (fig. S2, B and E).

In mc4−/−, the PEP1-YFP and PROPEP1-YFP levels did not increase and concurrently decrease, respectively, after 5 min (quantified in Fig. 2, H and J) and at least up to 1 hour after damage (fig. S2, C and G). As an additional control, PEP1-YFP maturation was not affected in a MC9-deficient mutant (fig. S2, D and G). Our results demonstrate that upon physical damage, PROPEP1 is cleaved in vitro and in vivo by Ca2+-dependent MC4 activity at R69 into the mature Pep1 immunomodulatory signaling peptide.

Wounding triggers distinct Ca2+ patterns in root epidermal cells

Ca2+ concentrations in the low millimolar range are needed to activate MC4 in vitro (16). We showed that the Ca2+ chelators EGTA or EDTA impaired MC4 activation and PROPEP1 cleavage in vivo after tissue damage. Ca2+ functions as a secondary messenger in signal transduction networks in all eukaryotic organisms, and many stimuli can trigger increases in cytosolic Ca2+ concentrations or so-called [Ca2+]cyt spikes (30). However, cleavage of PROPEP after every [Ca2+]cyt spike seems counterproductive. In an effort to link the spatiotemporal dynamics of [Ca2+]cyt fluxes, metacaspase activation, and Pep maturation, we focused on roots. Pep1 triggers a stronger defense gene expression in the root than typical bacterial or fungal elicitors (31). Although roots have not often been studied for wounding studies, they are constantly in contact with beneficial, but also potentially harmful, microorganisms as well as with invasive organisms, such as nematodes, and thus are relevant for tissue damage investigations. Furthermore, Pep1 maturation occurs similarly in root tissues (fig. S3, A, B, E, and F) as in leaf tissues (fig. S3, C and G) and whole seedlings (Fig. 1, B and C), but in contrast to leaf tissues, MC4 is not exclusively responsible for Pep1 maturation, because PEP1-YFP accumulation is only halved in mc4−/− root tissues (fig. S3D). Because Z-VRPR-fmk reduces cleavage better than mc4−/− does (fig. S3H), residual Pep1 maturation in roots is most probably due to redundancy with other (Ca2+-dependent type-II) metacaspases or with undefined proteases with similar inhibition characteristics.

Subtle cell damage was inflicted by means of multiphoton laser wounding to obtain the highest possible spatial and temporal resolution during imaging [Ca2+]cyt with a Yellow Cameleon Ca2+ probe (YC3.60-NES) in epidermal transition-zone root cells (Fig. 3A), revealing detailed spatiotemporal [Ca2+]cyt differences after damage (Fig. 3B and movies S1 and S2). [Ca2+]cyt responses could be grouped according to their duration and intensity in four zones in and surrounding the wound site (Fig. 3, D and E). Propidium iodide (PI) entry into the cell, as a proxy for plasma membrane integrity loss, marked the damaged cells of zone 1 (Fig. 3B) that accumulated long-lasting high [Ca2+]cyt (Fig. 3E). Zones 2 to 4 surrounding the damaged cells displayed [Ca2+]cyt spikes that returned to resting [Ca2+]cyt. The cells closest to the wound experienced the largest transient spikes, and a [Ca2+]cyt “wave” traveled outward beyond zone 4 with an average speed of 1.63 ± 0.4 × 10−6 m/s (zone 4 also peaks slightly later than zones 2 and 3). Similar [Ca2+]cyt dynamics were observed in Arabidopsis leaves after mechanical damage (32, 33) and in laser-ablated embryonic cells or larvae of Drosophila melanogaster (34, 35). Owing to the subtle damage inflicted by the multiphoton laser, we could obtain a similar or higher cellular resolution and enable the imaging of cells inside the wound (zone 1). [Ca2+]cyt peaks were reduced in zones 2 to 4 by addition of 1 mM EGTA (Fig. 3E and movie S2) and increasing concentrations of the alternative Ca2+ chelator BAPTA [1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid] (fig. S4, A and B), hinting at an extracellular source of Ca2+. Zone-1 cells showed a residual but considerable [Ca2+]cyt increase in the presence of 1 mM EGTA (Fig. 3E) or 0.5 mM BAPTA (fig. S4A). This might be indicative of Ca2+ release from internal stores that is chelated when EGTA or BAPTA enters the damaged cells or of such substantial (extracellular or intracellular) amounts that are not immediately chelated by either compound.

Fig. 3 Correlation of local [Ca2+]cyt fluxes and PROPEP1 processing in laser-damaged root cells.

(A) Arabidopsis root tip with the epidermal cells of the transition zone that were targeted by laser (filled in red). (B and C) Confocal microscopy time series of laser wounding (site indicated with an asterisk) of the YC3.60-NES Ca2+ probe and overlaid with PI stain (B), and PROPEP1-YFP delocalization from the tonoplast (sharp edges) to the cytosol (more diffuse signal) over time (C). Damaged cells either collapsed immediately after delocalization (>) or remained relatively intact after delocalization (arrow). The pound sign (#) indicates intracellular PI accumulation without PROPEP1-YFP delocalization. (D) Hypothetical zones in a laser-inflicted root wound. Color gradient outside the zones symbolizes the spread of the [Ca2+]cyt flux. (E) [Ca2+]cyt flux (top) and accumulation of PI (bottom) over time for the zones colored according to (D) in pure water or in the presence of 1 mM EGTA. (F) Quantification of YFP delocalization in cells from zone 1 and zones 2, 3, and 4 combined in pure water or in the presence of 1 mM EGTA or 50 μM Z-VRPR-fmk. Curves are the mean of N cells +SD in (E) and + SEM in (F). Gray vertical bars spanning 7 s in (E) and (F) indicate the average lag time between the laser shot and imaging start. Scale bar, 20 μm.

Pep1 maturation correlates with Ca2+ patterns

During imaging of PROPEP1-YFP that under resting conditions localizes to the cytosolic face of the vacuolar membrane (tonoplast) (Fig. 3C and fig. S5, A and D) (8), the YFP fluorescence delocalized to the cytosol only in damaged cells accumulating PI (zone 1) during laser wounding (Fig. 3, C and F, and movies S3 and S6). Two types of cellular events occurred in zone-1 cells: (i) The vacuole almost immediately collapsed after delocalization, and (ii) the vacuole and thereby the whole cell stay relatively intact after delocalization (Fig. 3C). Occasionally, cells accumulated some PI without YFP delocalization [Fig. 3C, pound sign (#)], hinting at a certain threshold of membrane integrity loss for PROPEP1-YFP processing. This suggestion was also apparent from the BAPTA dilution experiment (fig. S4), in which addition of 0.25 mM BAPTA lowered but did not alter the shape of the [Ca2+]cyt transient (fig. S4A), whereas YFP delocalization was inhibited (fig. S4C). Furthermore, addition of EGTA or Z-VRPR-fmk reduced YFP delocalization in the epidermal cells of the transition (Fig. 3F, fig. S5A, and movies S4 and S5) and maturation zones (fig. S5, C and D). These results underscore the localized heterogeneity of [Ca2+]cyt fluxes in zones 2 to 4 and the generation of passive near-millimolar [Ca2+]cyt transients in zone-1 damaged cells. Furthermore, PROPEP1 processing, evidenced by YFP delocalization, correlates with the [Ca2+]cyt abundance in zone-1 cells and is localized to damaged cells only. Consequently, the number of wounded cells could convert a digital cellular output (generation or no generation of Pep1) into an analog response, proportional to the damage.

Subcellular retention constrains Pep1 signaling

PROPEP1, which lacks a canonical transmembrane domain, is seemingly constrained at the tonoplast through its N terminus. Indeed, a double-labeled fluorophore construct with mCitrine fused to the N terminus of PROPEP1 and red fluorescent protein (RFP) fused to its C terminus localized to the tonoplast but only released the RFP moiety to the cytosol upon damage (Fig. 4A and movie S7). Furthermore, PEP1-YFP was present and PROPEP1-YFP was absent in the soluble phase, when seedling extracts were centrifuged to remove solid and membrane fractions (Fig. 4E). A GST-PEP1 fusion protein can bind the extracellular domains of PEPR1 in vitro (7), so the N terminus of Pep1 does not need to be free to interact with its receptor per se. Additionally, we found that exogenous application of purified N-terminally truncated PROPEP1 (Δ39-PP1) (fig. S6C) triggers root growth inhibition—a known negative effect of Pep1 application (5)—irrespective of cleavage by rMC4 (Fig. 4, B and C).

Fig. 4 Control, conservation, and hypothetical model of Pep1 release.

(A) A mCitrine-PROPEP1-RFP fusion protein in control (top) and laser-damaged root epidermal cells (bottom). Scale bar, 20 μm. (B) Arabidopsis seedlings 5 days after transfer to medium containing 25 nM of a N-terminally truncated PROPEP1 protein (Δ39-PP1) treated with active rMC4 or inactive rMC4C/A. The double mutant deficient for the PEP receptors (pepr1/pepr2) served as negative control. (C) Quantification of root length in (B). Values are the average of three biological repeats with 45 seedlings in total and plotted as mean + SEM. Statistics are described in Materials and methods. Blue, control; gray, ΔN-PROPEP1 + rMC4C/A; and yellow, ΔN-PROPEP1 + rMC4. (D) TNT-protease assays of GST-PROPEP fusion proteins (Arabidopsis PP1 to PP8) and the tomato PROPEP1 ortholog (PPT). All assays were repeated at least twice and were run together with PP1 as a positive control. An interpretation of the results is provided in fig. S6, A and B. (E) Immunoblots for PROPEP1-YFP (anti-GFP), the aquaporin TONOPLAST INTRINSIC 1-1 (anti-TIP1-1), and actin (anti-actin) of total protein extracts from seedlings at 0 and 5 min of forceps-induced damage, a soluble fraction after centrifugation at 20,000g (20K), 100,000g (100K), and the pellet fraction from 100K. (F) A hypothetical model for damage-induced Pep1 maturation.

The prevailing concept that proteases and substrates are sequestered in separate subcellular compartments and only mix upon cellular disintegration does not apply here. The vacuole, in which proteases are stored, changes shape, rounding up, and becomes immobile during PROPEP1-YFP delocalization, although it stays intact until after PROPEP1-YFP delocalization, when the vacuole can burst (Figs. 3C and 4A, fig. S5, and movies S3 and S6).

The Pep1 signal is probably short-lived in cells in which rapid vacuolar collapse can cause further “trimming” of the Pep1 peptide (Fig. 3C), as mirrored in the grinding damage investigated in Fig. 1, C and D. Truncated versions of synthetic Pep1 (similar to fig. S2, bands 4 and 5) had previously been shown to lose their activity (36). Conversely, prolonged Pep1 accumulation upon forceps application (Fig. 1B) is most probably caused by damaged cells with stable vacuoles (Fig. 3C) or by the continuous, unsynchronized loss of membrane integrity during wounding (movie S6). Taken together, MC4-dependent Pep1 maturation seems to mainly facilitate release from the tonoplast membrane to promote Pep1 motility into the surrounding tissue. This passage needs to be guarded because Pep1 overload negatively affects plant growth (Fig. 4, B and C) (5).

Conservation of Pep maturation

A family member of the subtilisin-like serine proteases or subtilases, called phytaspase (37), processes the wound signaling peptide prosystemin (38). Other subtilases are also relevant for development-related peptide hormone–processing events (39, 40). Whereas systemin is specific to the Solanaceae, PROPEP and metacaspase genes are found throughout the Angiosperms (41). We found that MC4 is able to cleave other Arabidopsis PROPEPs and the tomato ortholog of PROPEP1 in vitro as well (Fig. 4D and fig. S6, A and B), suggesting that metacaspase function in Pep maturation is conserved across plants.

Model and future perspectives

Taken together, a model emerges in which MC4 and PROPEP1 both remain inactive in the cytosol until a loss of plasma membrane integrity in the damaged zone-1 cells leads to a prolonged increase in intracellular [Ca2+] (Fig. 4F). Ca2+ mainly originates from the extracellular space and potentially from internal stores, such as the vacuole (Fig. 4F), and binds zMC4 to initiate autocatalytic cleavage, whereafter active MC4 can cleave PROPEP1. Pep1 is released from the tonoplast to the cytosol, from where it can passively diffuse or be potentially actively secreted through the compromised plasma membrane to bind the extracellular domains of the BAK1-PEPR1/2 receptor kinase complex and to signal the surrounding intact cells of zones 2 to 4 to activate a defense response (Fig. 4F).

Sequence diversity of PROPEPs and PEPRs implies that PROPEPs evolve quickly. Although PEPR signaling seems to be preserved, its loss might go unnoticed in unstressed plants (41). This multicomponent system may have been deactivated or even lost during plant breeding programs. Indeed, PROPEPs, PEPRs, and metacaspases may be useful targets for marker-assisted breeding or CRISPR-Cas mutagenesis to improve disease resistance of high-performance crop varieties. The ameliorative effect of Pep-mediated PEPR signaling against a variety of diseases caused by microbes and herbivores (912) should become useful as global warming leads to increased crop losses due to insect pests (42).

Material and methods

Plant material and treatments

For the preparation of sterile seedlings, A. thaliana (L.) Heynh. seeds were surface sterilized with 70% ethanol and plated on half-strength Murashige and Skoog (½MS) medium supplemented with 1% (w/v) sucrose and 0.5% (w/v) Phytagel (Sigma-Aldrich), stratified for at least 2 days at 4°C, and then germinated at 21°C under continuous light (MLR-350 Plant growth chamber; Sanyo). After 5 days, individual seedlings were transferred to liquid ½MS medium with 1% (w/v) sucrose and grown for an additional 9 days. Seedlings treated with the protease inhibitors antipain (100 μM), chymostatin (100 μM), pepstatin A (1 μM), PMSF (1 mM), E64 (10 μM), 1,10-phenanthroline (20 mM), Z-VRPR-fmk (50 μM), EDTA (1 mM), EGTA (1 mM), and protease inhibitor cocktail 1:100 (MFCD00677817) (Sigma-Aldrich) were vacuum infiltrated with the individual solutions 3 times for 2 min each, and incubated at room temperature (RT) for an additional 10 min before freezing. Because the cleavage event happens so fast, preinfiltration was necessary to deliver the inhibitors as close and as quickly as possible to the presumed proteolysis site (the cytoplasm) before damage ensued. In the case of cell-impermeable inhibitors, such as EGTA and EDTA, they are at least absorbed in the intercellular space and apoplast from where they can most probably enter the cell instantly upon damage.

For the in vivo wounding treatment, 8 seedlings were pooled, squeezed 5 times with serrated forceps, and incubated at RT for the indicated amount of time before freezing in liquid nitrogen and subsequent immunoblot analysis. Treated and untreated seedlings were frozen in liquid nitrogen and ground to powder with a mortar and pestle under constant supply of liquid nitrogen, because the use of automated homogenizers led to thawing and immediate PEP detection in the immunoblot.

To evaluate the potential difference of PROPEP1 processing in wild type (WT) and metacaspase mutants and between leaf (whole rosette) and root tissues, seedlings were grown for 14 days at 21°C under continuous light on ½MS medium supplemented with 0.7% (w/v) agar without sucrose. Roots were separated from leaf tissues with a scalpel, thereby minimizing unwanted tissue damage prior to the forceps treatment or grinding in liquid nitrogen. For inhibitor treatments of crushed root and leaf tissues, dimethyl sulfoxide (DMSO), 5 mM EGTA, and 50 μM Z-VRPR-fmk were not preinfiltrated, but added to the tissue in deionized water prior to thawing at RT. Slightly elevated cleavages in samples treated with EGTA and Z-VRPR-fmk (fig. S3H) could be due to a minor delay in absorption of the chemicals in the crushed tissue.

Tissue powder from liquid nitrogen-ground seedlings or tissues were stored at –80°C and for any application, other than incubation of ground tissue at RT, was immediately supplemented with 3× SDS loading buffer [0.5 M Tris, pH 6.8, 15% (w/v) glycerol, 0.3 M DTT, 5% (w/v) SDS, and bromophenol blue] preheated at 70°C. Transgenic Arabidopsis lines (Columbia-0 accession) expressing PROPEP1-YFP, PEP1-YFP, and YFP had been described previously (8).

Cloning of constructs and generation of transgenic Arabidopsis lines

Mutated PROPEP sequences were prepared by site-specific mutagenesis of the original coding sequence in the plasmid pEarley101 (3, 7). For the mCitrine-PROPEP1-RFP construct, mCitrine was ligated at a ScaI site to the N terminus of PROPEP1 in pDONR207 and cloned by Gateway into the destination vector pB7RWG2 for an in-frame fusion of RFP to the C-terminal end. For comparison of the native and R69-mutated PROPEP1 sequences, Arabidopsis (Col-0) was transformed by Agrobacterium tumefaciens by means of the floral dip method, whereas for comparison to METACASPASE 4-deficient plants, a homozygous mc4−/− mutant in the Arabidopsis Landsberg erecta accession (Ler-0), CSHL_GT7237 and a corresponding wild-type Ler-0 line were transformed with the pEarley101-PROPEP1 construct by the floral dip method and for comparison to MC9-deficient plants, a mc9−/− mutant (Col-0; GABI_540H06) was used. These lines were selected for an expected single-insertion genetic segregation of the pEarley101-PROPEP1 construct at a 3:1 ratio on selective ½MS medium containing 10 mg/L of glufosinate-ammonium (Sigma-Aldrich). Equal amounts of PROPEP1-YFP fusion proteins between the selected Ler-0 and mc4−/− lines and the Col-0 and mc9−/− lines were confirmed by immunoblots with anti-GFP (data not shown). To re-confirm the genetic background of the different lines, a minimum of 10 seedlings per line were genotyped with primers spanning the MC4 and MC9 genomic loci and T-DNA-specific primers (fig. S2D and table S3).

Immunoblotting and band percentage quantification

Ground tissue was immediately supplemented with approximately an equal volume of 3× SDS loading buffer preheated at 70°C, heated for 5 min, and centrifuged for 5 min at 16,000g to remove cellular debris. Proteins were separated in 10% precast SDS polyacrylamide gels (Genscript) for PROPEP1-YFP and PEP1-YFP separation, or in 4-20% precast gradient SDS polyacrylamide gels (Genscript) for MC4 subunit separation. Analysis was done by semi-dry Western blotting to PVDF membranes and incubated with anti-GFP antibodies (mouse, 1:1000; #11814460001 Roche) or anti-MC4 (rabbit, 1:15000) (21). Only non-overexposed immunoblots were analyzed. Bands were quantified for PROPEP1-YFP and PEP1-YFP as a percentage of the total signal per lane either with ImageJ (43) (Fig. 1D and E) or with Image Lab software (Bio-Rad) (Fig. 2, H to K, and figs. S2G and S3, D and H) and described as band percentage (band %) in all graphs. Briefly, each lane on a given immunoblot was represented as a pixel-intensity profile (lane profile) and peaks corresponding to the different bands were marked. The relative area under the curve for each peak corresponded to the band %. Quantification of a given band within a lane alleviated the need for comparison to a reference band or to the loading control (for band quantification across different lanes).

In vitro TNT-protease assay

Unmodified and mutated PROPEP-coding sequences (CDSs) were cloned by the Gateway method to pDEST15 (N-terminal GST tag) or pDEST24 (C-terminal GST tag). Addition of a GST tag was necessary to increase the size and amount of incorporated 35S-methionine and improve visualization of the fused protein product. Recombinant MC4 (rMC4) and mutated inactive rMC4C/A (alanine substitution of active-site cysteine at position 139) fused to a His-tag were expressed and purified from Escherichia coli as previously described (16) and stored in 50% (w/v) glycerol, 25 mM HEPES (pH 7.5). TNT-protease assays were carried out as described (44). Briefly, PROPEP CDSs were in vitro transcribed and translated (TNT coupled transcription/translation system, Promega) in the presence of radiolabeled 35S-methionine and aliquots of this reaction were subsequently mixed and incubated with the indicated amounts of rMC4 or the inactive rMC4C/A protease for 30 min at 30°C in optimal rMC4 reaction buffer [50 mM HEPES, pH 7.5, 150 mM NaCl, 10% (w/v) glycerol, 50 mM CaCl2, and 10 mM DTT]. Proteolysis was stopped by the addition of Laemmli buffer supplemented with 50 mM EGTA to avoid aberrant SDS-PAGE electrophoresis due to high levels of Ca2+ ions in the samples. Samples were separated by SDS-PAGE and visualized with storage phosphor screens.

Laser wounding and microscopy

Seedlings were grown upright on ½MS plates with 0.7% (w/v) and without sucrose after 48 hours of stratification and 7 to 10 days of growth under a 16-hour light/8-hour dark regime. Slides were prepared in two ways for microscopy. At first, seedlings were transferred to microscopy slides in 150 μl of 0.01 mg/ml PI-containing deionized water and, when indicated, 50 μM Z-Val-Arg-Pro-DL-Arg-fluoromethylketone trifluoroacetate (Z-VRPR-fmk; Bachem), 1 mM EGTA, or a range of concentrations of BAPTA dissolved in DMSO. Microscopy slides were taped at one end as a spacer to avoid squeezing and damaging of the root tip after transfer, whereafter the roots were carefully covered with standard glass coverslips, allowing a qualitative study. To overcome the problem that roots tend to move out of focus with this method and to improve quantitation, roots were mounted in custom-made sample holders as described (45). Precise focal regions were wounded with a Ti:Sa laser (MaiTai DeepSee multiphoton laser; SpectraPhysics) at an excitation wavelength of 900 nm at 70% power and for variable durations of 300 to 7000 ms. Confocal images were acquired on a Zeiss LSM780 confocal microscope with a Plan-Apochromat 40×/1.4 oil immersion objective, argon laser at an excitation wavelength of 514 nm, and respective emission regions for yellow fluorescent protein (YFP) and PI. Intracellular calcium concentrations were measured ratiometrically with a Yellow Cameleon 3.60 probe fused to a nuclear export signal (YC3.60-NES) as described (33). PI was excited at a wavelength of 561 nm, as to not cross-excite the YFP moiety of YC3.6-NES.

For the mCitrine-PROPEP1-RFP construct, 7- to 9-day-old seedlings were imaged on a Leica SP5-II-Matrix confocal microscope, with mCitrine and mRFP excitation at 514 nm and 561 nm and emission in the 530-560 nm and 550-750 nm range, respectively. Images were acquired with a 63× water-immersion objective, 2 Hybrid Detectors (HyD), and a bright-field scan. Wounding was achieved by slightly squeezing the roots with forceps prior to imaging (Fig. 4A). The mCitrine-PROPEP1-RFP line was also imaged after laser wounding on the Zeiss LSM780 setup (movie S7).

Microscopy image analysis for the quantification of calcium transients (NES-YC3.60) and YFP delocalization (PROPEP1-YFP) during laser wounding

Images were analyzed with custom-made scripts for ImageJ (43) (supplementary materials, YFP_deloc.groovy). First, the images were registered on a single channel (for example, the PI channel) by means of an elastic registration algorithm implementation available in ImageJ (bUnwarpJ), whereafter the others channels were transformed accordingly. Second, the scripts allowed the selection of multiple regions of interest (ROIs), corresponding to the various cells encompassing the wound region. Third, several intensity features (Mean, Median, and Maximum Intensities) were extracted from the different channels per ROI over time (images were taken every second). In addition, for the YC3.60-NES probe, the ratio of the Venus channel over the CFP channel (R) divided by the average intensity of 10 s before laser wounding (R0) was used as a proxy for free cytosolic calcium concentration (R/R0 [Ca2+]cyt).

For the PROPEP1-YFP probe, the tonoplast-to-cytosol delocalization of the YFP signal can be viewed as relatively few pixels with a high YFP signal intensity (tonoplast) moving in time to relatively more pixels with a lower YFP signal intensity (cytosol) in a given ROI (a visual explanation is provided in fig. S5B). For this quantification we divided the pixels in the upper third of a pixel intensity distribution by the lower two-thirds in that distribution per ROI over time. Because not all the cells express the PROPEP1-YFP probe equally, the pixel intensity distribution was normalized beforehand to the highest value in the first image in time as follows, with H, histogram of pixel intensities in the image, B, number of Bin in the Histogram H, and b, index of the Bin holding the maximum intensity value of the first time point:H[i] = number of pixel with intensity in the i – tier bin (1 ≤ iB)ratio=Σi=R+1bH[i]Σi=1RH[i]R=[23×b]The PI channel was quantified as PI signal intensity divided by the average intensity of 10 s before laser wounding (I/I0).

Ultracentrifugation for the biochemical separation of soluble and insoluble (membrane) fractions

Sterile Arabidopsis seedlings expressing the PROPEP1-YFP fusion protein were grown for 14 days at 21°C in continuous light on ½MS medium supplemented with 0.7% (w/v) agar (without sucrose). Seedlings were frozen in liquid nitrogen and ground to powder with a mortar and pestle under constant supply of liquid nitrogen. Approximately 1 g of tissue powder was divided in two 15-ml tubes. One tube was thawed and left for 5 min at RT, while the other was mixed immediately with ice-cold extraction buffer [10 mM HEPES, pH 7.5, 10 mM KCl, 2 mM MgCl2, 10 mM EGTA, 1 mM EDTA, and protease inhibitor cocktail (cOmplete ULTRA Tablets EDTA-free, #05892791001 Roche)]. For 500 mg of tissue powder, after addition of 5 ml of extraction buffer, the sample was vortexed and transferred to an ultracentrifugation tube (Ultra-clear tube, 13×51 mm, #344057 Beckman) and centrifuged (SW55 Ti swinging-bucket rotor, Beckman Coulter) for 10 min at 20,000g (20K) at 4°C. The supernatant was transferred to a new tube and centrifuged for an additional hour at 100,000g (100K) at 4°C. The 100K pellet was solubilized with 3× SDS loading buffer. Both 20K and 100K supernatant fractions were precipitated with acetone and the protein pellets were solubilized with 3× SDS loading buffer. A control sample taken before centrifugation containing total tissue powder was extracted in 3× SDS loading buffer. Samples were separated on NuPAGE 4-12% Bis-Tris Protein Gels, 1.0 mm, 10-well (Thermo Fisher Scientific) and transferred to PVDF membranes by semi-dry Western blotting (Trans-Blot Turbo Transfer System and transfer packs, #1704157, Bio-Rad). The following antibodies were used according to the manufacturers’ instructions: anti-GFP (mouse, 1:1000 dilution; #11814460001 Roche), anti-actin (rabbit, 1:1000 dilution; #AS132640 Agrisera), anti-TIP1-1 (rabbit, 1:1000 dilution; #AS09482, Agrisera) and their corresponding secondary antibodies, anti-mouse IgG (1:10000 dilution, #NA931V, GE healthcare) and anti-rabbit IgG (1:10000 dilution, #NA934V, GE healthcare).

Immunoprecipitation and in-gel digest of PROPEP1 fusion protein bands to determine peptide coverage by mass spectrometry

Total seedling tissue powder was stored at –70°C. Immunoprecipitation was carried out according to the manufacturer’s protocol (GFP-trap_MA, Chromotek) with some adjustments. Approximately 1 g of seedling powder was homogenized in 1.5 ml of extraction buffer [50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, and Complete protease inhibitor cocktail (Roche)]. The tissue lysate was centrifuged for 10 min at 20,000g and 120 μl of GFP-trap_MA beads slurry was added to the supernatant and incubated for 2 hours on a rotating wheel at 4°C. The beads were washed 3 times with 1 ml of wash buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.5% NP-40) and the protein was eluted with 30 μl of 2× Laemmli sample buffer. The sample was loaded on 1.0-mm, 10-well, NuPAGE 4-12% Bis-Tris protein gels (Thermo Fisher Scientific) and protein bands were visualized with a mass spectrometry-compatible Pierce Silver Stain Kit (Thermo Fisher Scientific). Bands corresponding to the relative sizes of PROPEP1-YFP and lower molecular mass-processed forms, were cut out of the gel and prepared by a standard in-gel digestion protocol for mass spectrometry analysis. Proteins were digested overnight in the gel bands soaked with either trypsin (at 37°C) or chymotrypsin (at 25°C) in 50 mM ammonium bicarbonate, 10% (v/v) acetonitrile (ACN). The following morning, samples were acidified with 0.5% trifluoroacetic acid (TFA) final volume and dried in a Speed-vac.

The immunoprecipitation experiment was done twice and samples were analyzed by liquid chromatography-tandem mass spectrometry on both a Q-Exactive HF and Orbitrap Elite mass spectrometer (Thermo Fisher Scientific). Generated tandem mass spectra from raw data files were extracted in a MGF file format with RawConverter. MGF files were searched against a concatenated target-decoy database of the representative Araport11 proteome supplemented with the PROPEP1-YFP protein sequence with the SearchGUI toolkit (version 3.3.3). SearchGUI was configured to run the tandem mass spectra with the identification search engines X!Tandem, MS-GF+, MyriMatch, Comet, and OMSSA. Non-default tandem mass spectrum identification settings were semi-specific digestion for either trypsin or chymotrypsin (both without P rule) and variable modifications included protein N-terminal acetylation, methionine oxidation, N-terminal pyroglutamate, and cysteine propionamidation. Fragment ion tolerance was set to 0.5 Da and 0.01 Da for the Orbitrap Velos and Q-Exactive HF spectrometers, respectively without fixed modifications, because no cysteine alkylation was performed. Search identification output files were processed by the companion tool PeptideShaker (version 1.16.25) and all default reports (.txt) and identification files (.mzid) were exported. Peptides corresponding to PROPEP1-YFP have been summarized (table S2). The full mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD010816.

Endogenously generated peptide extraction and mass spectrometry analysis (peptidomics)

Sterile Arabidopsis seedlings expressing the PROPEP1-YFP fusion protein were grown on vertical ½MS plates without sucrose in two densely seeded rows (two plates) for 12 days with a 16-hour light/8-hour dark regime at 21°C. Seeds had been stratified for 2 days at 4°C. Roots were separated from the rosette with scissors and the tissue was immediately ground in liquid nitrogen. Root tissue was unfrozen for 5 min at RT and proteins were extracted by sonication in 400 μl of urea lysis buffer [20 mM HEPES, pH 8.0, 8 M urea, 1 mM sodium orthovanadate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 10 mM EGTA, 5 mM EDTA, and cOmplete ULTRA protease inhibitor cocktail tablets (Roche)]. The lysates were cleared by centrifugation at 20,000g at 15°C for 15 min and transferred to new tubes. Samples were acidified with 0.5% TFA and again cleared by centrifugation. Native peptides were enriched by reversed-phase chromatography on a SampliQ C18 column (Agilent Technologies). Columns were pre-wetted with 70% ACN and 0.1% TFA and elution was done with 50% ACN and 0.1% TFA whereafter the eluate was dried by Speed-vac. No trypsin or any other protease was used to preserve the structure of the endogenously generated peptides.

Samples were dissolved in 2% ACN and 0.1% TFA and subjected to mass spectrometry on a Q Exactive Orbitrap mass spectrometer operated as previously described (46). From the tandem mass spectrometry data, Mascot Generic Files (mgf) were created with the Mascot Distiller software (version, Matrix Science). Peak lists were then searched with the Mascot search engine and the Mascot Daemon interface (version 2.5.1, Matrix Science). Spectra were searched against the TAIR10 database concatenated with the PROPEP1-YFP protein sequence. Variable modifications were set to pyroglutamate formation of the amino-terminal glutamine and methionine oxidation. Mass tolerance on precursor ions was set to ±10 ppm (with Mascot’s C13 option set to 1) and on fragment ions set to 20 mmu. The instrument setting was on ESI-quadruple (QUAD). As similarly to trypsin, metacaspase cleaves C-terminally to arginine and lysine, the enzyme was set to trypsin/P, allowing eight missed cleavages, and the cleavage was also allowed when lysine or arginine were followed by proline. Only peptides that were ranked first and scored above the threshold score, set at 99% confidence, were withheld. Peptides matching the above criteria have been summarized (table S1). The full mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD005740.

GST-TEV-Δ39-PP1, GST, rMC4, and rMC4C/A protein purifications

A N-terminally truncated version of PROPEP1 lacking the first 39 amino acids (starting at a serine at position 40) was subcloned from full-length PROPEP1 and an upstream cleavable tag for the Tobacco Etch Virus protease (TEVp) was added at the N-terminal part by means of iProof polymerase (Bio-Rad) with the Gateway-compatible primers attB1-TEVrs-S40PROPEP1-FW and attb2-PROPEP1_s-RV (containing a stop codon) (primer list is provided in table S3). Truncation was necessary to improve protein purification from E. coli, because a full-length GST-TEV-PROPEP1 fusion protein was difficult to obtain in soluble form. The PCR fragment was assembled into pDONR221 by BP reaction and correct insertion was confirmed by Sanger sequencing. The resulting entry clone was N-terminally fused to GST by LR reaction in pDEST15 and designated GST-TEV-Δ39-PP1. The construct was transfected into BL21 (DE3) E. coli cells and protein expression was induced at OD600 = 0.8 at 28°C overnight with 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). E. coli cells were pelleted and sonicated in GST extraction buffer (50 mM HEPES-KOH, pH 7.5, and 300 mM NaCl) on ice. Proteins were purified by gravity flow with glutathione Sepharose resin (#17075601) and PD10 columns according to the manufacturer’s protocol at 4 °C (GE Healthcare). Columns were washed with GST extraction buffer and the protein was eluted with 50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, and 10 mM reduced glutathione (#78259, Thermo Fisher Scientific). Free GST protein (as control for the root growth experiment) was purified under the same conditions. Purified GST and GST-TEV-Δ39-PP1 proteins were stored at -70°C in elution buffer supplemented with 10% (v/v) glycerol.

E. coli cells for the purification of TEVp (BL21 DE3 pRK793 plasmid), rMC4, and rMC4C/A (BL21 DE3 (pLysE) pDEST17-MC4 and pDEST17-MC4C/A plasmid, respectively) were grown to OD600 = 0.6 and protein expression was induced overnight with 0.2 mM IPTG at 20°C. Cultures were pelleted and sonicated in HIS extraction buffer (50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, and 20 mM imidazole). Protein was bound to Nickel Sepharose 6 Fast Flow agarose beads (#17531801, GE Healthcare) by gravity flow at 4°C, washed with 50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, and 40 mM imidazole, and eluted with 50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, and 400 mM imidazole. Proteins were subjected to size exclusion chromatography and eluted in 50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, and 10% (v/v) glycerol.

Root growth inhibition analysis

Protein concentrations of GST-TEV-Δ39-PP1, GST, TEVp, rMC4, and rMC4C/A were determined by A280 on a NanoDrop spectrophotometer. The GST-TEV-Δ39-PP1 protein was incubated with TEVp at a concentration ratio of 1:100 overnight at 30°C and subsequently with rMC4 or rMC4C/A at a concentration ratio of 1:50 for 1 hour at 30°C. As a control treatment, GST was similarly incubated with TEVp and rMC4C/A. The reaction buffer consisted of 50 mM HEPES-KOH, pH 7.5, 120 mM NaCl, 10% (v/v) glycerol, 10 mM DTT, and 50 mM CaCl2.

Seeds of Arabidopsis pepr1pepr2, mc4−/−, and their respective WT background accessions (Col-0 and Ler-0) were stratified at 4°C for 3 days, grown for another 3 days on vertical plates containing ½MS media with 0.7% (w/v) agar (without sucrose), and then transferred to plates containing 25 nM of each treatment mixed in the solid ½MS media. Treatments were done on the same plate per biological repeat and were changed from top to bottom on the 4-well plates (Nunc Nunclon 4-Well, #167063, Thermo Fisher Scientific) to avoid a positional effect on the root growth between treatments over the biological repeats. Root pictures were taken 5 days after transfer to the treatment and root length (millimeters per pixel) was measured with ImageJ (43).

R visualization

Peptide coverage plots for the PROPEP1-YFP fusion protein and fragments were generated by plotting the number of validated peptide spectrum matches (PSMs) per protein position (fig. S2F). From the exported default peptides reported, peptides with a confidence > 0.9 were used to cumulate the number of validated PSMs for each position of the peptide. Per sample, the number of validated PSMs was displayed with a heat-color scale ranging from 2 PSMs (gray) to the maximum of validated PSMs of that sample (red).


A one-way analysis of variance (ANOVA) was applied to the log-transformed PEP1 and PROPEP1 intensities with inhibitor as main effect for the data presented (Fig. 1E and fig. S3H). This transformation was necessary to stabilize the variance. The interest was in the difference in outcome of each inhibitor compared to the water treatment. P value adjustment was done with the Dunnett’s method. R software (, R version 3.5.1) was applied for the analysis with the lm function. Post-hoc analysis was done with the emmeans package and adjusted P values were displayed on the bar charts in case of significance.

A two-way ANOVA was applied to the log-transformed PEP1, PROPEP1, p20, and p20* intensities with genotype and time as fixed effects and their interaction term for the data presented (Fig. 2, H to K, and fig. S3D). The log transformation was necessary to stabilize the variance. The interest was in the difference in outcome over time between both genotypes. R software (, R version 3.5.1) was applied for the analysis with the lm function. Post-linear analysis was done by means of the capabilities of the emmeans function and adjusted P values were displayed on the bar charts in case of significance.

A two-way ANOVA was applied to the root length data with genotype and treatment as fixed effects and their interaction term for the data presented (Fig. 4C). Tukey’s multiple comparison test was used to query significant differences between treatment in a given genotype. GraphPad Prism version 7.03 was applied for the analysis.

Supplementary Materials

Figs. S1 to S6

Tables S1 to S3

Movies S1 to S7

Microscopy Data Analysis Script

References and Notes

Acknowledgments: We thank E. Lam (Rutgers University, New Jersey, USA), E. Russinova (VIB-UGent Center for Plant Systems Biology, Ghent, Belgium), and A. Costa (University of Milan, Milan, Italy) for kindly providing the antibody to MC4, the mutant pepr1pepr2 seeds, and the YC3.60-NES seeds, respectively; V. Storme (VIB-UGent Center for Plant Systems Biology, Ghent, Belgium) for advice on statistics; the VIB BioImaging Core for excellent technical support; and M. De Cock for help in preparing the manuscript. Funding: This work was supported by the Swiss National Science Foundation (grant 31003A_127563 to T.B.), the Research Foundation-Flanders (grant G.0C37.14N to K.G. and FWO14/PDO/166 to S.S.), the Ghent University Special Research Fund (grant 01J11311 to F.V.B.), and a CLEM grant from Minister Ingrid Lieten (Belgium) for the acquisition of the Zeiss LSM780 microscope. Author contributions: T.H., A.D.F.-F, S.B., and S.S. conceived and designed the analysis. T.H., A.D.F.-F, R.P.K., H.S., D.R., A.S., J.N., P.Y., R.P., A.G., B.P., and S.S. performed the experimental work. T.H. and S.S. wrote the manuscript. T.B., K.G., F.V.B., and S.B. revised the manuscript and were involved in the discussion of the work. Competing interests: The authors declare no competing interests. Data and material availability: All data to support the conclusions of this manuscript are included in the main text and supplementary materials. The full mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD010816 and PXD005740. All materials are available on request, including chemical compounds as supplies permit, subject to a standard materials transfer agreement.
View Abstract

Stay Connected to Science

Navigate This Article