Plant proteases in the control of the hypersensitive response.

Highlight : This review emphasizes the crucial functions of plant proteases during the regulation of defence-related hypersensitive cell death. Abstract The hypersensitive response (HR) is a plant defence reaction triggered by activation of immune receptors upon pathogen recognition. It results in rapid cell death at the attempted invasion site, confining the pathogen and sending signals to distal parts of the plant that can in turn activate defences for subsequent attacks. HR cell death is a highly controlled phenomenon, requiring the concerted action of diverse plant proteases and regulatory mechanisms to keep it efficient yet confined. understanding of the mechanisms leading to HR, although our knowledge about the pathways that regulate this form of programmed cell death still remains scattered. In this review, we explore current knowledge on plant proteases as HR regulators. Proteases are key regulatory enzymes that not only serve degradative purposes, but also have very important signalling roles. In animals, caspases have been shown to be the major regulators and executioners of programmed cell death. Plants caspases, and instead, programmed cell death is carried out by caspase-like and other protease activities owing to different protease classes. Here, we will summarise the mechanistic roles of plant proteases whose role in HR regulation is relatively well understood, which includes members of the cysteine, threonine and serine protease families.


Introduction
Due to their lack of physical mobility, plants must defend themselves against rapidly evolving pathogens. Unlike animals, plants do not possess an adaptive immune system with mobile defender cells, thus they rely on each cells´ innate immunity for effective defence responses (Jones and Dangl, 2006). In what is known as gene-for-gene interactions, plant resistance (R) gene products, such as surface-localised and intracellular nucleotide-binding leucine rich repeat (NB-LRR) immune receptors, perceive avirulent (avr) pathogen-derived gene products, also known as effector proteins, often leading to a form of confined programmed cell death, known as the hypersensitive response (HR).
The first reports on HR date back to the beginning of the 20 th century when H.
Marshall Ward described a variable discoloration of leaves that turned from yellow to brown/black when infected with the leaf rust Puccina dispersa (Ward, 1902). Additional studies at the time on the plant pathosystems Chrysanthemum-Uredo (Puccinia) chrysanthemi and wheat-Puccinia glumarum (leaf yellow rust), also claimed such cell death phenomenon upon pathogen infection (Gibson, 1904;Marryat, 1907). Nevertheless, it was not until 1915 that the term "hypersensitiveness" was intuitively used by Elvin C. Stackman to convey an "abnormal rapid cell death" in cereal crops when attacked by black stem rust fungus (Puccina graminis) (Stakman, 1915). Since the plant exhibited hypersensitiveness at the fungal entry sites, the fungus was unable to develop normally, thus the plant was deemed resistant. This phenotypic definition of HR has remained partially unchanged over the years, though in certain pathosystems we know now that HR is often uncoupled from resistance (Coll et al., 2011).
Defining HR cell death has not been an easy task due to its mixed morphological and biochemical features, partly resembling several other forms of cell death in both plants and mammals. In mammals, up to 12 types of programmed cell death (PCD, also known as regulated cell death) modalities have been described so far (Galluzzi et al., 2018). Amongst them, apoptosis is the best characterised. Apoptosis is a non-inflammatory process mainly regulated by caspases, in which the following features are observed: A c c e p t e d M a n u s c r i p t 3 cytoplasmic shrinkage, chromatin condensation, nuclear fragmentation and plasma membrane blebbing, which ultimately lead to the formation of intact vesicles (apoptotic bodies) that are engulfed and digested by phagocytes (Galluzzi et al., 2018). Whilst certain features such as cytoplasmic shrinkage and chromatin condensation are also observed during HR, other events like phagocytosis of apoptotic bodies after cellular demise do not occur in plants (Table 1). As a result, resemblance of specific aspects of HR to apoptosis is not sufficient to consider HR as an apoptotic-like cell death. On the other hand, HRpresents the majority of morphological features of plant regulated necrosis cell death wherein mitochondrial swelling, shrinkage of the protoplast and early rupture of the plasma membrane is observed. These features can also be found in other types of mammalian PCD such as pyroptosis or necroptosis (Table 1) ( Galluzzi et al., 2018). However, characteristics reminiscent of plant vacuolar cell death such as enlargement of the vacuole and tonoplast rupture are also exhibited during HR (van Doorn et al., 2011). When considering the cytological features of HR, it is also of great importance to consider the nature of the invading pathogen. For instance, vacuolar rupture can be an effective measure to restrict viruses that proliferate in the host cytoplasm (Hatsugai et al., 2004).
By contrast, fusion of the tonoplast with the plasma membrane allows discharge of antimicrobial compounds to the intercellular space where bacterial pathogens tend to proliferate (Hatsugai et al., 2009). In summary, HR is an atypical and confined plant cell death modality that occurs at the site of successful recognition of pathogens, and generally displays the following hallmarks: cytoplasmic shrinkage, mitochondrial swelling, chromatin condensation, chloroplast and plasma membrane disruption and vacuolisation (Table 1).
Interestingly, necrotrophic pathogens such as the fungus Cochliobolus victoriae can hijack the HR machinery through the delivery of toxins that target the plant cell in order to kill it and feed on cell remnants (Lorang et al., 2012). In the course of this cell death, expected features that resemble necrosis (protoplast shrinkage and a mitochondrial permeability transition) are displayed, though membrane and tonoplast integrity is maintained (Curtis and Wolpert, 2004).
Despite its early discovery more than a century ago, a thorough understanding of the mechanisms regulating HR is lacking. In the last few decades, a growing A c c e p t e d M a n u s c r i p t 4 body of evidence indicates that plant proteases are involved in pathogen perception and induction of effective local and systemic defence responses, that are often accompanied by an HR-related cell death confined to the attempted pathogen ingress site (Bozkurt et al., 2011;Coll et al., 2010;Rooney et al., 2005) Proteases are ubiquitous enzymes required for the correct functioning of living cells. Operating at the post-translational level, proteases catalyse irreversible hydrolytic reactions wherein peptide bonds of target substrates are cleaved, giving rise to new protein products. (van der Hoorn, 2008). Whilst traditionally believed to act solely as destructing enzymes, we currently know that proteases can also influence the activity of other proteins, regulate protein fate and localisation, modulate protein-protein interactions, and contribute to processing of cellular information through signal transduction (Turk, 2006).
Based on the MEROPs database, an integrated information resource of proteases (http://merops.sanger.ac.uk), there are five mechanistic classes of proteases found on living organisms depending on the catalytic residue involved in the cleavage of the substrate peptide bond, namely cysteine, aspartate, threonine, serine and metalloproteases (Rawlings et al., 2018). In the case of cysteine, threonine and serine proteases, the orchestrated action of a catalytic triad comprised of a nucleophile (Cys, Thr or Ser), a base (usually His) and in certain cases an acid (Asp), allows cleavage of the peptide bond (Lopez-Otin and Bond, 2008). A second classification in the MEROPS database discriminates between clan types (usually denoted by a letter) where proteases fall into distinct categories depending on their protein tertiary structure. Within each clan, a third and final subdivision is made into distinct families of proteases based on their evolutionary relationships (Rawlings et al., 2018).
As occurs in animals, plant proteases are directly implicated in the regulation of host responses to pathogen infection, including PCD. This review is intended to underscore the crucial functions of the distinct classes of plant proteases in the regulation of HR. Due to space limitations, we will only emphasise proteases whose mechanistic roles in HR regulation are well understood, which involve members of the cysteine, threonine and serine protease class.

Cysteine proteases: PLCPs, Metacaspases and VPEs
In mammals, apoptosis requires an evolutionarily conserved group of cysteine proteases termed caspases. Since particular characteristics are shared between animal apoptosis and the defence-related hypersensitive cell death in plants, it was reasoned in the past that a certain level of conservation on the molecular components involved in PCD should be present across kingdoms (del Pozo and Lam, 1998). Nevertheless, although certain structurally unrelated plant proteases have been shown to exhibit caspase-like activities in the course of defence-related HR (Chichkova et al., 2004;del Pozo and Lam, 1998;Hatsugai et al., 2004), no caspase homologues are found within plant genomes.
Plant genomes encode approximately 140 cysteine proteases which fall into 5 distinct clans. In the context of plant-pathogen interactions, the CA clan, comprising proteases with a papain-like fold named papain-like cysteine proteases (PLCPs), and the CD clan, comprising proteases with a caspase-like fold, have been well documented (Misas-Villamil et al., 2016). Biochemical tools such as specific protease inhibitors as well as activity-based probes have been pivotal in the discovery of many cysteine proteases implicated in plant defence by monitoring their protease activity (van der Hoorn and Kaiser, 2012). Here, we will discuss the role in HR of 3 PLCPs (Cathepsin B, Rcr3, Pip1), 3 Metacaspases (AtMC1, AtMC2 and AtMC4) and the Vacuolar Processing Enzymes (VPEs).

PLCPs
PLCPs are released as a preproteases bearing a signal peptide at the Nterminal end, an autoinhibitory domain or prodomain and the catalytic domain  Interestingly, rcr3 mutant lines do not exhibit higher susceptibility compared with tomato lines missing the Cf-2 gene cluster, implying that Rcr3 inhibition does not contribute to virulence (Dixon et al., 2000). Moreover, Pip1 accumulates to higher levels compared to Rcr3 in the apoplast upon SA treatment or diverse pathogen infections. Hence, in agreement with the decoy model, it can be hypothesised that Rcr3 evolved as a decoy to perceive effector-mediated perturbations and that the original operational target of Avr2 is Pip1 (Shabab et al., 2008;van der Hoorn and Kamoun, 2008). In addition to C. fulvum, the oomycete, Phytophthora infestans, and the root parasitic nematode Globodera rostochiensis are also able to inhibit Rcr3 via secretion of apoplastic effectors EPIC1 and EPIC2B, and Gr-VAP1, respectively (Lozano-Torres et al., 2012;Song et al., 2009). However, whilst the weak interaction between EPICs and Rcr3 is not sufficient to trigger HR, complex formation between the allergen-like effector Gr-VAP1 and Rcr3 is sensed by the guardian Cf-2 which ultimately induce HR at the attempted site of infection (  VPEs play a role during compatible interactions that is independent of cell death (Misas-Villamil et al., 2013). Collectively, it can be concluded that the role of VPEs in host pathogen-triggered PCD also appears to be dependent on the pathosystem.

Metacaspases
Along with VPEs in the CD clan and belonging to the C14 family, are metacaspases (Rawlings et al., 2018). Metacaspases are an ancient group of cysteine proteases found in protozoa, fungi, plants and bacteria, predominantly known for their pivotal roles in PCD of non-metazoan organisms (Minina et al., 2017). From an evolutionary standpoint, metacaspases are distantly related to animal caspases, though bioinformatic analysis predict close structural homology to animal caspases at the catalytic domain which harbours a M a n u s c r i p t 9 caspase-like His-Cys catalytic dyad and a caspase-hemoglobinase fold (Tsiatsiani et al., 2011). With regards to their biochemical features, metacaspases are quite distinct compared to caspases, owing to their lack of aspartate specificity and their preference for substrate cleavage after Arg or Lys residues (Gonzalez et al., 2007;Vercammen et al., 2004;Watanabe and Lam, 2005). Metacaspases are classified into type I and type II based on their domain architecture. In plants, type I metacaspases bear an N-terminal prodomain extension that is absent in type II metacaspases. Type II metacaspases, however, possess an extended linker region between catalytic subunits and the C-terminus (Tsiatsiani et al., 2011). Evidence for a direct role of metacaspases in HR came mainly from studies in Arabidopsis wherein an upregulation of type I metacaspase 1 (AtMC1) was initially reported upon pathogen infection (Zimmermann et al., 2004). Thereafter genetic analysis of AtMC1 function through knockout mutants revealed that AtMC1 is required for the HR-like runaway cell death phenotype of the lesion mimic mutant lesion stimulating disease 1 (lsd1) (Coll et al., 2010). In parallel, atmc1 plants exhibit suppression of HR triggered by infection with an avirulent strain of Pst DC3000 (AvrRpm1) (see Figure 1) or an avirulent strain of the oomycete Hyaloperonospora arabidopsidis (Hpa) ( Table 2). Of note, pathogen growth is unaffected in atmc1 plants, providing another example of HR uncoupled from disease resistance.
Interestingly, AtMC2, a closely related type I metacaspase in Arabidopsis, genetically serves the opposite function of AtMC1 by negatively regulating HR as AtMC2 overexpression phenocopies the abrogated HR phenotype of atmc1 mutant plants, whereas atmc2 mutants show exacerbated HR (Figure 1 and  In parallel to type I metacaspases, the constitutively expressed Arabidopsis type II metacaspase, AtMC4, has been found to contribute to the HR-like cell death response triggered by fungal mycotoxin FB1 and avirulent Pseudomonas syringae pv. maculicola ES4326 carrying AvrRpt2 (Pma AvrRpt2) (Figure 1 and

Threonine proteases: PBA1 subunit of the proteasome
The ubiquitin-proteasome system (UPS) is a protein degradation system that has long been known for its role in many fundamental cellular processes,

Subtilisin-like proteases: saspases and phytaspases
Given the importance of caspases in animal PCD processes, over the past few decades the hunt for caspase-like proteases in plants has been unceasingly Subtilases are serine proteases, belonging to the SB clan and S8A family, which rely on the catalytic triad: aspartate, histidine and serine for execution of their catalytic activity (Rawlings et al., 2018). In the context of HR, saspases and phytaspases represent two examples of serine proteases that might play indirect and direct roles, respectively, in the regulation of cell death upon biotic insults (Chichkova et al., 2010;Coffeen and Wolpert, 2004).

Saspases
The necrotrophic fungus Cochliobolus victoriae, the causative agent of Victoria blight of oats (Avena sativa), produces the host selective toxin victorin. In a gene-specific manner, victorin triggers a form of cell death reminiscent of HR.
Proteolysis of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), an in vitro substrate of victorin, was demonstrated to be inhibited by caspasespecific and general inhibitors of cysteine and serine proteases (Navarre and Wolpert, 1999). In parallel, purification of two specific caspase-like activities of victorin-treated A. sativa protein extracts followed by substrate cleavage assays of caspase-like synthetic tretrapeptides, suggested the existence of a proteolytic signalling cascade upstream of Rubisco cleavage (Coffeen and Wolpert, 2004).
Purification of active proteolytic enzymes followed by N-terminal sequencing revealed two peptidases that shared extensive homology to diverse plant subtilases, particularly, rice subtilisin-like serine proteases. As a result, the term "saspases" was intuitively coined owing to its serine catalytic residue and their "aspase" activity (Coffeen and Wolpert, 2004). Alike animal caspases, saspases appear to serve a processing enzymatic function rather than a degradative one, owing to their low activity towards general protease substrates, including Rubisco. Collectively, it appears likely that, upon victorin sensitivity, saspases may be constituents of a proteolytic cascade that leads to Rubisco cleavage and PCD. Interestingly, saspases localise to the extracellular fluid at the early A c c e p t e d M a n u s c r i p t 13 stages of victorin-induced PCD, in what appears to be a tightly regulated secretion event rather than a consequence of PCD (Figure 1). As a result, the subcellular localisation of saspases and rubisco makes it unlikely that saspases cleave rubisco directly (Coffeen and Wolpert, 2004;Vartapetian et al., 2011).
Unfortunately, besides the intriguing biochemical data on saspases, no direct genetic evidence for their involvement in HR has been identified to date.

Phytaspases
An alternative approach to scout around for caspase-like proteases was based on the previous knowledge that the Agrobacterium tumefaciens-encoded protein VirD2 is cleaved by human caspase 3 at D 400 within a TATD motif (Chichkova et al., 2004). VirD2 from A. tumefaciens harbours a nuclearlocalisation signal (NLS) and assists in the transfer of single-stranded DNA fragments (T-DNA) into the genome of the plant (Tinland et al., 1995). Since the NLS of Vir2D is essential for successful nuclear uptake of foreign DNA, Chichkova and colleagues hypothesised the existence of a plant protease capable of cleaving VirD2 in a caspase 3-like manner (Chichkova et al., 2004).
In order to test such hypothesis, Vir2D was utilised as a substrate to detect a "plant caspase" that operates in the course of a TMV infection in Nicotiana tabacum plants carrying an N resistance gene. In this manner, they elucidated a caspase 3-like activity that was exclusively present in plants undergoing TMVinduced PCD (Chichkova et al., 2004). Purification of the protein responsible for the activity in tobacco and rice followed by mass spectrophotometer analysis suggested that the protein was a subtilisin-like protease of the S8 family, which was thereafter named, phytaspase (Chichkova et al., 2010). This enzyme is comprised of a signal peptide, a prodomain and a protease-associated domain within its peptidase domain (Vartapetian et al., 2011).  With regards to their role in HR and defence, several lines of evidence suggest that phytaspases are required for TMV-triggered HR in tobacco plants harbouring an N resistance gene (Chichkova et al., 2010). Transgenic tobacco plants overproducing phytaspases exhibit enhanced HR upon TMV infection. By contrast, impairment in phytaspase production, in phytaspase-silenced plants, results in attenuated HR triggered by TMV (Figure 1 and Table 2). Notably, this latter phenotype can be restored by heterologous expression of rice wild-type phytaspase but not by its catalytically inactive mutant (Chichkova et al., 2010).
In this pathosystem, HR triggered by phytaspases appear to serve a protective function, owing to the fact that as opposed to phytaspase-silenced plants, which tend to accumulate high levels of TMV, phytaspase-overproducing plants reduce TMV accumulation compared to wild-type control plants (Chichkova et al., 2010).
As opposed to animal caspases that retain an intracellular localisation, phytaspases seem to be constitutively synthesised as zymogens and processed into prodomain-less mature forms which are secreted to the apoplast (Figure 1).
Intriguingly, upon viral infection, phytaspases shuttle their subcellular localisation into the cytoplasm where they may cleave intracellular substrates required to induce HR, pointing towards a spatial regulation of their activity (Chichkova et al., 2010). This re-entry into the cytoplasm would be needed to explain the observed VirD2 cleavage by N. tabacum phytaspase. Such protease redistribution is exclusive to phythaspases since other apoplastic proteases that exhibit caspase-like activities, such as CathB, are confined in the apoplast throughout the entire course of a viral infection (Gilroy et al., 2007;Vartapetian et al., 2011).

Concluding remarks
Now that more than a century since the HR discovery went by, we are still far from understanding how this type of PCD is mechanistically carried out.
However, research in the last decade has significantly contributed to a better understanding of the proteases involved in this phenomenon, largely due to the successful efforts of the expanding plant protease community, which has These caspase-like activities in plants are not only involved in HR, but also in processes not related to cell death.. In this review we report only the function of a few proteases in HR belonging to the cysteine, threonine and serine protease families, which are the best characterized. Still, for most of them we do not know the substrates nor their upstream regulators. In addition, several other plant proteases that have been directly or indirectly linked to HR await characterisation. Our current knowledge of HR execution consists of scattered protease activities in different cell compartments of different plant species infected by different pathogens. These separated pieces of the HR puzzle will hopefully come into place during the next years, thanks to the coalescing efforts that the plant protease community is putting together. M a n u s c r i p t 23  A c c e p t e d M a n u s c r i p t