Thirty years of resistance: Zig-zag through the plant immune system

Abstract Understanding the plant immune system is crucial for using genetics to protect crops from diseases. Plants resist pathogens via a two-tiered innate immune detection-and-response system. The first plant Resistance (R) gene was cloned in 1992 . Since then, many cell-surface pattern recognition receptors (PRRs) have been identified, and R genes that encode intracellular nucleotide-binding leucine-rich repeat receptors (NLRs) have been cloned. Here, we provide a list of characterized PRRs and NLRs. In addition to immune receptors, many components of immune signaling networks were discovered over the last 30 years. We review the signaling pathways, physiological responses, and molecular regulation of both PRR- and NLR-mediated immunity. Recent studies have reinforced the importance of interactions between the two immune systems. We provide an overview of interactions between PRR- and NLR-mediated immunity, highlighting challenges and perspectives for future research.

The alphabet soup digested: nomenclatures applied to the plant immune system PTI was originally an abbreviation for "PAMP-triggered immunity", mediated by PRRs such as Arabidopsis thaliana Flagellin-Sensing 2 (FLS2). ETI is an acronym for "effectortriggered immunity," which is mostly mediated by NLRs (Jones and Dangl, 2006), but can also involve RLP-mediated detection of apoplastic effectors (Jones et al., 1994). While the terms PTI and ETI are frequently used in the literature, there are limitations to their use in describing specific immune responses (Thomma et al., 2011). For example, the apoplastic effector Avr4 from the tomato (Solanum lycopersicum) leaf mold pathogen Cladosporium fulvum binds to fungal chitin to retard cell wall degradation by host chitinases and thus the release of N-acetyl glucosamine oligomers that activate defense (Joosten et al., 1994;van den Burg et al., 2006). Avr4 is recognized by the tomato cell-surface RLP Cf-4 ( Thomas et al., 1997). Thus, while immunity activated by some PRRs can be classified as PTI, others can be classified as ETI, since cell-surface receptors can recognize both PAMPs and apoplastic effectors (Thomma et al., 2011). Other terms have been introduced to classify immune responses based on receptors, such as PRR-mediated immunity and NLR-mediated immunity (Lacaze and Joly, 2020). Immune responses are best defined by the location of recognition by the initiating protein, such as extracellularly triggered immunity and intracellularly triggered immunity (van der Burgh and Joosten, 2019), or surface-receptor-mediated immunity and intracellular-receptor-mediated immunity (van der Burgh and Joosten, 2019;. Each of these terms has its own advantages and should be used with caution ( Figure 1A). In this review, we try to minimize the overuse of these acronyms and emphasize immune responses triggered by the corresponding receptors.

Structural and evolutionary overview of PRR proteins
Plant PRR proteins are either RLPs or RLKs. RLKs consist of an extracellular domain, a transmembrane domain, and cytoplasmic kinase domain. RLPs lack a cytoplasmic kinase domain, and both require co-receptors to transduce immune signals. PRRs are localized to the PM via a transmembrane a-helix or a glycophosphatidylinositol (GPI) anchor (Boutrot and Zipfel, 2017). Both RLPs and RLKs perceive ligands via a range of extracellular domains. These include leucine-rich repeat (LRR), lectin, malectin, lysin motif (LysM), and epidermal growth factor-like domains (Boutrot and Zipfel, 2017).
RLKs are found in Plasmodium, plants, and animals but not fungi (Shiu and Bleecker, 2003). Conceivably, RLKs were present in the common ancestors of these organisms but were later lost in the fungi. Plant RLKs underwent remarkable expansion and constitute 60% of the kinases in the Arabidopsis genome (Shiu and Bleecker, 2003). Arabidopsis RLKs can be classified into 44 subfamilies based on their kinase domains (Shiu and Bleecker, 2003). The LRR-RLKs represent the largest subfamily of RLKs and are the best characterized RLKs in plants. A phylogenetic study of 33 plant species concluded that the average number of LRR-RLKs in angiosperms is approximately 250 per species (Dufayard et al., 2017; Figure 1B). LRR-RLKs are further classified into 20 subgroups, with subgroup XII constituting genes involved in pathogen recognition, such as FLS2, EFR, and Xa21 (Dufayard et al., 2017). Interestingly, the gene number in the LRR-RLK subgroup XII is highly variable across plant species, indicating that these genes underwent either expansion or contraction in particular lineages (Dufayard et al., 2017;Ngou et al., 2022). Similarly, the LRR-RLPs represent the largest subfamily of RLPs in plants, and the size of this gene family is also highly variable across plant species (Ngou et al., 2022; Figure 1B).

Structural and evolutionary overview of NLR proteins
NLRs are grouped into three classes according to their Nterminal domains: coiled-coil (CC) NLRs (CNLs), Toll/ Interleukin-1 receptor/Resistance (TIR) protein NLRs (TNLs), and RPW8-like CC domain (RPW8) NLRs (RNLs). Both CNLs and RNLs contain N-terminal CC-domains. Plant NLRs carry a nucleotide-binding (NB) domain shared by APAF-1, various plant R proteins and CED-4 (together, the NB-ARC domain), and LRR domains at their C-termini. These domains vary between NLRs, and additional noncanonical domains can be integrated into some NLRs (also known as NLRintegrated domains, or NLR-IDs; Sarris et al., 2016). The functions of these domains also vary among NLRs. The LRR domain is involved in direct or indirect recognition of effectors (Krasileva et al., 2010;Ma et al., 2020a;Martin et al., 2020). The NB-ARC domain exhibits ATP binding activity and acts as a switch for NLR activation (Wang et al., 2019b). The CC, TIR, and RPW8 domains function as signaling domains to downstream responses upon NLR activation (Adachi et al., 2019a;Bi et al., 2021;Duxbury et al., 2021;Jacob et al., 2021). Some CC-domains are involved in effector recognition and interact directly with effectors (Avr-Pik) as well as a "guardee" protein (such as RIN4), which is a target of pathogen effectors (Lukasik and Takken, 2009;Kanzaki et al., 2012). The a-helices in both the CC and RPW8 domains were recently shown to form cation channels required for defense signaling Jacob et al., 2021). TIR domains can also self-associate or associate with the TIR domains from paired TNLs, which is crucial for their activation (Williams et al., 2014;Duxbury et al., 2020). TIR domains, upon oligomerization, exhibit NADase activity, which leads to the production of variant-cyclic-ADP-ribose (v-cADPR; Horsefield et al., 2019;Wan et al., 2019a). TIR domains also exhibit 2 0 ,3 0 -cAMP/cGMP synthetase activity . These small molecules produced by TIR domains likely function in signaling. The ID domain in NLR-IDs functions as a decoy, which enables the NLR to detect effectors targeting proteins with homology to the ID (van der Hoorn and Kamoun, 2008;Sarris et al., 2016;Baggs et al., 2017).
NLR genes are present in the genomes of all land plants (Gao et al., 2018). CNLs, TNLs, and RNLs are present in basal angiosperm species such as Amborella and Nymphaea (Baggs et al., 2020;Liu et al., 2021). However, TNLs are absent from most monocot genomes, indicating that gene loss likely occurred before monocots diverged from dicots (Tarr and Alexander, 2009). The loss of TNLs was also accompanied by the loss of TNL-signaling components, such as ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1), PHYTOA LEXIN DEFICIENT 4 (PAD4), and SENESCENCE-ASSOCIATED GENE 101 (SAG101; Baggs et al., 2020;Liu et al., 2021). The loss of these signaling components may have driven the contraction of TNLs in some angiosperm lineages, or vice versa . Similar to the LRR-RLK-XII and LRR-RLP, the number of NLRs (or NB-ARC containing proteins) is also highly variable across the angiosperms (Baggs et al., 2020;Liu et al., 2021). Furthermore, the LRR-RLK-XII, LRR-RLP, and NLR gene families have undergone lineage-specific co-expansion or co-contraction (Ngou et al., 2022;Figure 1B). The cause of these concerted expansions and/or contractions is currently unclear but has been proposed to be linked to pathogen pressure and ecological specialization (Plomion et al., 2018;Baggs et al., 2020;Liu et al., 2021;Ngou et al., 2022).

PRRs involved in pathogen recognition
PRRs recognize PAMPs/MAMPs/HAMPs from bacteria, fungi, oomycetes, parasitic plants, and herbivores. Some PRRs also recognize self-molecules, such as DAMPs and other plant endogenous peptides (phytocytokines; Hou et al., 2021). Some PRRs are not involved in direct ligand recognition but function as PRR co-receptors and negative regulators of immune signaling. There are more than 60 characterized immunity-related PRRs with known elicitors, and we attempt here to list those PRRs with known elicitors that are involved in pathogen recognition ( Figure 2). Due to space limitations, some PRR gene names are abbreviated: the full gene names can be found in Supplemental Data Set 1.

PRRs involved in the recognition of parasitic plants
In addition to eH 2 O 2 , AtCARD1 has also been shown to perceive the self-derived quinone compound 2,6-dimethoxy-1,4benzoquinone (DMBQ; Laohavisit et al., 2020). Perception of DMBQ induces AtCARD1-dependent immune responses. On the other hand, the parasitic plant Phtheirospermum japonicum perceives DMBQ via AtCARD1 homologs PjCADL1/2/3, which leads to development of haustoria for parasitic infection (Laohavisit et al., 2020). Thus, CARD1 is involved in both immunity (for nonparasitic plants) and parasitic plant infection. Plants also perceive PAMPs from parasitic plants to restrict infection. The tomato LRR-RLP SlCuRe1 perceives the peptide Crip21 from the parasitic plant Cuscuta spp. (Hegenauer et al., 2020). Crip21 is derived from a Cuscuta glycine-rich cell wall protein. Activation of SlCuRe1 by Crip21 elicits cell death and defense responses in tomato (Hegenauer et al., 2020; Figure 2E).

PRRs involved in viral recognition
While some PRRs, such as AtNIK1, have been shown to be required for viral resistance, no PRR has been reported to directly perceive viral particles (Zorzatto et al., 2015). However, the Arabidopsis PRR co-receptor bak1 loss-offunction mutant exhibits enhanced susceptibility to multiple viruses (Kørner et al., 2013). In addition, exogenous application of double-stranded RNAs and viral coat protein (CP) elicits PTI responses in plants ( Allan et al., 2001;

PRRs involved in the recognition of animals
In addition to eNAD + , AtLecRK-1.8 and AtLecRK-1.1 are involved in the perception of Pieris brassicae (cabbage moth) eggs (Gouhier-Darimont et al., 2019;Groux et al., 2021). The ligand from P. brassicae eggs that activates AtLecRK-1.8 remains to be identified and characterized. The Arabidopsis LRR-RLK AtNILR1 is involved in the perception of Heterodera schachtii (sugarbeet nematode) extracts, and nilr1 mutants are hypersusceptible to nematode infection (Mendy et al., 2017). The cowpea (Vigna unguiculata) LRR-RLP VuINR was shown to perceive inceptin, a proteolytic fragment of chloroplastic ATP synthase from the oral secretions of Lepidopteran herbivores (a HAMP; Steinbrenner et al., 2019). Whether PRRs can perceive ligands directly from herbivores remains to be determined ( Figure 2G).

NLRs involved in pathogen recognition
Sensor NLRs are involved in the recognition of effectors from viruses, bacteria, fungi, oomycetes, parasitic plants, and herbivores. Some NLRs act as helpers or co-receptors to transduce immune signals from sensor NLRs following effector recognition . Currently, there are more than 140 characterized NLRs with known recognized effectors (Kourelis and Kamoun, 2020). Here, we summarize a list of NLRs involved in effector recognition (Figure 3; Supplemental Data Set 2).

Apparent absence of NLRs involved in self-recognition in plants
In mammals, DAMPs can be indirectly recognized the intracellular NOD-, LRR-, and pyrin domain-containing protein 3inflammasome in macrophages (Swanson et al., 2019). However, no plant NLRs have been reported to detect selfmolecules so far ( Figure 3D).

NLRs involved in the recognition of parasitic plants
Virus-induced silencing of the CNL VuRSG3-301 from V. unguiculata leads to enhanced susceptibility to the parasitic plant Striga gesnerioides race 3 (Li and Timko, 2009). The effector recognized by VuRSG3-301 has not yet been identified ( Figure 3E).

The PRR signaling pathway
The extracellular domains of plant PRRs perceive diverse ligands (Boutrot and Zipfel, 2017). Binding of ligands leads to heterodimeric receptor complex formation between PRRs and their co-receptors, such as BAK1 and CERK1 (Miya et al., 2007;Ma et al., 2016;Hohmann et al., 2017). On the other hand, RLPs constitutively interact with SOBIR1 and recruit BAK1 upon ligand recognition (Liebrand et al., 2013;Albert et al., 2015). In Arabidopsis, the bacterial flagellin peptide flg22 is perceived by the LRR-RLK FLS2 (Felix et al., 1999;Chinchilla et al., 2006). Flg22 acts as a "molecular glue" and interacts with and brings together the extracellular LRR domains of FLS2 and BAK1 (Sun et al., 2013;Hohmann et al., 2017). Heterodimeric complex formation between the LRR domains of FLS2 and BAK1 brings their cytoplasmic kinase domains into close proximity, which leads to a series of auto-and trans-phosphorylation events Cao et al., 2013;Sun et al., 2013). This activated receptor complex then phosphorylates RLCKs Liang and Zhou, 2018). RLCK subfamily VII members (collectively known as RLCK-VIIs) were first shown to be important for surface receptor-mediated immunity in tomato and tobacco and to be required for Cf-4 and Cf-9 to confer fungal resistance (Rowland et al., 2005). In Arabidopsis, RLCKs play particularly important roles during PRR-mediated immunity (Lu et al., 2010;Lin et al., 2014;Liang and Zhou, 2018;Rao et al., 2018). BAK1 associates with and phosphorylates the RLCK-VII BIK1 at the Try243 and Try250 residues (Lu et al., 2010;Lin et al., 2014).

Signaling pathway of singleton NLRs
NLR-mediated immunity is triggered by the detection of effectors through intracellular NLRs. NLRs detect effectors either via direct interactions with effectors, guarding effector targets, or guarding decoy proteins (Van der Biezen and Dangl and Jones, 2001;van der Hoorn and Kamoun, 2008). In Arabidopsis, CNLs and TNLs act as sensor NLRs that recognize effectors, while RNLs act as helper NLRs to transduce immune signals (Feehan et al., 2020). While the majority of sensor NLRs in Arabidopsis require helper NLRs to mediate immunity, some CNLs mediate immune responses alone. These are known as singleton NLRs, such as ZAR1 and RPM1 (Adachi et al., 2019b). ZAR1 recognizes a range of effectors by monitoring pseudokinases such as RKS1 and PBL2, which mimic authentic RLCK targets of effectors (Wang et al., 2019a). The bacterial effector AvrAC from X. campestris uridylylates the RLCK PBL2. The ZAR1/ RKS1 heterodimer associates with uridylylated PBL2 (PBL2 UMP ), which leads to conformational changes in the heterodimer. ADP in the NB-ARC domain in ZAR1 is ejected Figure 4 Plant immune signaling pathways. A, PRR signaling pathway. Ligand perception by PRRs activates multiple kinases, which leads to calcium influx to the cytosol, ROS production, transcriptional reprogramming, and callose deposition. B, Singleton NLR signaling pathway. The ZAR1/RKS1 heterodimer detects the effector AvrAC via association with uridylylated PBL2 by AvrAC. This leads to the activation and oligomerization of ZAR1. The ZAR1 resistosome localizes to the PM and triggers calcium influx, which leads to the HR and cell rupture. C, Helper-NLR-dependent sensor NLR signaling pathway. Recognition of ATR1 by the TNL RPP1 leads to oligomerization and the induced proximity of TIR domains. The TIR domain exhibits NADase activity and produces v-cADPR, which might activate EP-proteins and the helper NLRs (RNLs). Following TNL activation, EP-proteins and RNLs associate with each other and activate downstream immune responses, likely via cation channel activity from the helper NLRs. Timeline on the right indicates the order and duration of each signaling event following ligand/effector perception. Numbers indicate the corresponding signaling events in the figure on the left. Note that the activation of ETI is usually preceded by PTI activation, and the strength and duration of each event vary and are dependent on the PRRs/NLRs that are activated. and replaced by ATP (Wang et al., 2019b). This results in the oligomerization of ZAR1/RKS1/PBL2 UMP oligomers into pentameric resistosomes (Wang et al., 2019a) that localize to the PM to trigger downstream immune responses (Wang et al., 2019a;Bi et al., 2021).
ZAR1 resistosomes were recently shown to exhibit cation channel activity . The N-terminal a-helices in ZAR1 form a funnel-shaped structure with a negatively charged carboxylate ring, which allows cations to pass through into the cytosol. Co-expression of ZAR1 with RKS1, PBL2, and AvrAC in plant protoplasts results in cytosolic calcium influx, ROS accumulation, and the perturbation of chloroplasts and vacuoles . Robust ROS accumulation during ZAR1 activation is likely caused by the activation of multiple downstream signaling components, such as the NADPH oxidases, since the CPKs are activated by cytosolic calcium influx . In addition, multiple CPKs and RbohD have been shown to be phosphorylated during RPS2 activation Kadota et al., 2019). Defense-related transcription factors are also likely activated by cytosolic calcium influx (Boudsocq et al., 2010;Gao et al., 2013). The perturbation of chloroplasts and vacuoles is quickly followed by the loss of PM integrity and cellular rupture  Figure 4B). How these processes are regulated by immune signaling components and their relationships to transcriptional reprogramming are currently unclear.
The signaling pathway of helper-NLRdependent sensor NLRs The majority of sensor NLRs requires helper NLRs to mediate immunity. In solanaceous plants, the NB-LRR REQUIRED FOR HR-ASSOCIATED CELL DEATH proteins (collectively known as NRCs) are required for immunity and hypersensitive cell death response (HR) mediated by multiple sensor NLRs . Interestingly, the N-terminal CC domain in ZAR1 contains a "MADA motif" that is also present in NRCs (Adachi et al., 2019a). This suggests that perhaps NRCs also form cation channels with a-helices following activation. In Arabidopsis, ADR1s and NRG1s are required for resistance and HR mediated by some CNLs and many TNLs (Bonardi et al., 2011;Castel et al., 2019a;Wu et al., 2019;Saile et al., 2020). Following effector recognition, TNLs also oligomerize into resistosomes to mediate resistance (Ma et al., 2020a;Martin et al., 2020). The Arabidopsis RPP1 recognizes the Hpa effector ATR1, and N. benthamiana ROQ1 recognizes the Xanthomonas effector XopQ. These effectors are recognized by the LRR and post-LRR domain, which likely leads to conformational changes and oligomerization of these TNLs into tetrameric resistosomes (Ma et al., 2020a;Martin et al., 2020).
The TIR domains of TNLs are brought into close proximity following oligomerization, activating NADase activity and producing v-cADPR (Horsefield et al., 2019;Wan et al., 2019a;Duxbury et al., 2020;Ma et al., 2020a;Martin et al., 2020). TIR domains also exhibit 2 0 ,3 0 -cAMP/cGMP synthetase activity by hydrolyzing RNA or DNA . v-cADPR and 2 0 ,3 0 -cAMP/cGMP are proposed to be signaling molecules that activate downstream signaling components (Horsefield et al., 2019;Wan et al., 2019a;Yu et al., 2021). Following the activation of TNLs, the EP-domain containing proteins (EP-proteins) SAG101 and EDS1 associate with NRG1 . Similarly, the activation of TNLs also leads to the association of the EP-proteins PAD4 and EDS1 with ADR1 (Wu et al., 2021b). These associations lead to the activation of these signaling components, which in turn activate downstream immune responses, such as defense-related gene expression and HR (Lapin et al., 2019;Sun et al., 2021). The RNLs ADR1 and NRG1 were also recently shown to function as calcium channels to activate immunity (Jacob et al., 2021). It is conceivable that the association and activation of helper RNLs and EP-proteins induces calcium influx and triggers downstream immune responses ( Figure 4C).

Physiological responses induced by CNLs alone
Activation of the Arabidopsis CNL RPS2 in the absence of PTI leads to the phosphorylation of RbohD (in Ser343/347), CPKs, and MAPKs Tsuda et al., 2013;Kadota et al., 2019;Ngou et al., 2021a;Yuan et al., 2021). RPS2-induced RbohD phosphorylation and ROS production are dependent on BAK1/BKK1 and BIK1 (Yuan et al., 2021). However, it is currently unclear whether BAK1/BKK1 and BIK1 are directly or indirectly activated by CNLs. While the ZAR1 resistosome directly triggers calcium influx, other Purple shading represents unclear responses that cannot be predicted. Asterisks indicate inoculation with the bacterial pathogen P. syringae pv. maculicola (Psm) leads to NHP accumulation (Wang et al., 2018c;Liu et al., 2020). (Right) PRR and NLR signaling network. Activation of PRRs (red) and NLRs (blue) lead to the activation of downstream signaling components (orange) and physiological responses (yellow), which result in resistance against pathogens (pink). Note that the activation of physiological responses can vary between immune receptors and are dependent on specific PRRs/NLRs. calcium channels may also be activated by CNLs . The activation of RPM1, RPS2, and RPS5 leads to MAPK activation and the HR (Ngou et al., 2021a). In addition, the activation of many CNLs leads to the upregulation of SA-and NHP-biosynthesis genes (Jacob et al., 2018;Ngou et al., 2021a). Thus, ET, SA, and NHP are likely to be produced during CNL activation ( Figure 5).

Physiological responses induced by the co-activation of PRRs and NLRs
Co-activation of PRRs and NLRs ("PTI + ETI") leads to the robust activation of BIK1, RbohD, and MPK3 Su et al., 2018;Ngou et al., 2021a;Yuan et al., 2021). This results in stronger calcium influx, ROS accumulation, and callose deposition compared to PTI or ETI alone (Ngou et al., 2021a;Yuan et al., 2021). In addition, "PTI + ETI" leads to stronger accumulation of SA and NHP compared to PTI alone, which is likely due to the stronger expression of SA-and NHP-biosynthesis genes during ETI (Wang et al., 2018c;Castel et al., 2019a;Liu et al., 2020;Figure 5).

Regulation of PRR-mediated immunity
The PRR-signaling pathway is tightly regulated as the excessive activation of PRRs leads to autoimmunity and growth inhibition (Navarro et al., 2006;Albrecht et al., 2012;Huot et al., 2014).

Regulation of PRRs
Both the transcript and protein levels of PRRs are regulated by multiple mechanisms. For example, the expression of FLS2 is regulated by the microRNA miR172b (Zou et al., 2018). The expression of FLS2 is also upregulated by ET (Boutrot et al., 2010). U-BOX DOMAIN-CONTAINING PROTEIN 12 (PUB12) and PUB13 mediate the polyubiquitination of FLS2, which leads to the endocytosis and degradation of this protein (Lu et al., 2011). Cf-4 also undergoes endocytosis upon Avr4 recognition (PostMa et al., 2016). The activation of PRRs and their co-receptors must also be regulated. BAK1-INTERACTING RECEPTOR (BIR)-LIKE KINASE 1 is an RLK that associates with and sequesters BAK1 to prevent the auto-activation of BAK1-associated PRRs (Gao et al., 2009;Ma et al., 2017;Hohmann et al., 2018). Following PAMP perception, the peptide RAPID ALKALINIZATION FACTOR 23 (RALF23) is perceived by a PRR complex composed of the CrRLK1L FERONIA (FER) and the LORELEI-LIKE-GPI ANCHORED PROTEIN 1. The perception of RALF23 by FER negatively regulates the formation of the FLS2-BAK1 complex (Stegmann et al., 2017;Xiao et al., 2019). FER regulates PM nanodomain organization to modulate PRR signaling (Gronnier et al., 2020). In addition, the phosphorylation status of PRRs is regulated by multiple protein phosphatases. In Arabidopsis, POLTERGEIST-LIKE 4 (PLL4) and PLL5 associate with EFR and negatively regulate elf18-induced responses . PROTEIN PHOSPHATASE 2A negatively regulates the phosphorylation status of BAK1  Figure 6).

Regulation of PRR-signaling components
In addition to PRRs, downstream signaling components are also regulated to prevent prolonged activation. As a central signaling component in the PRR-signaling pathway, the Arabidopsis RLCK BIK1 is regulated by multiple mechanisms. EXTRA-LARGE G PROTEIN 2 (XLG2) functions with other heterotrimeric G proteins to attenuate proteasomemediated degradation of BIK1 (Liang et al., 2016). The turnover of BIK1 is regulated by CPK28, PUB4/25/26, and the E3 ubiquitin ligases RING-H2 FINGER A3A/B (Monaghan et al., 2014;Wang et al., 2018a;Derkacheva et al., 2020;Ma et al., 2020b). The phosphorylation status of BIK1 is also negatively regulated by the protein phosphatase PP2C38 (Couto et al., 2016). In addition to RLCKs, other PRR-signaling components must also be regulated. RbohD is ubiquitinated by the E3 ubiquitin ligase PIRE (PBL13 interacting RING domain E3 ligase), which leads to proteasome-mediated degradation . PHAGOCYTOSIS OXIDASE/ BEM1P (PB1) DOMAIN-CONTAINING PROTEIN negatively regulates ROS production by controlling the localization of RbohD (Goto et al., 2020). The PP2C phosphatases PP2C5 and AP2C1 negatively regulate the phosphorylation of MPK3 and MPK6 (Brock et al., 2010; Figure 6).

Regulation of NLR-mediated immunity
Similar to PRRs, the prolonged activation of NLRs also leads to autoimmunity. Thus, the regulation of both NLRs and downstream signaling components is important to prevent autoimmunity.
The localization of the ZAR1 resistosome to the PM is required for ZAR1-mediated resistance (Wang et al., 2019a;Bi et al., 2021). In addition, the Arabidopsis importin-a nuclear transport receptor protein IMP-a3/MOS6 is required for SUPPRESSOR OF SNC1-mediated immunity (Lüdke et al., 2021). Thus, the localization of NLRs is important and is likely regulated by proteins involved in trafficking ( Figure 6).

Regulation of NLR-signaling components
The correct localization of helper NLRs is likely important for signaling. For example, the helper NLR NRC4 accumulates at the extra-haustorial membrane following P. infestans infection (Duggan et al., 2021). In addition, the balanced activity of both cytosolic-and nuclear-EDS1 is required for full immunity (Garc ıa et al., 2010). Thus, the localization of helper NLRs and NLR-signaling components is important for defense. The activity of NLR signaling components is also negatively regulated. The Arabidopsis RNL NRG1C functions as a negative regulator in NLR-mediated immunity; overexpressing NRG1C compromised TNL-mediated HR and resistance (Wu et al., 2021a). In addition, an atypical member of the NRC family, NRCX, negatively regulates other NRC members to modulate immunity (Adachi et al., 2021). Posttranslational modifications (PTMs) are important for the functions of both PRRs and NLRs. For example, the phosphorylation of the C-terminus of the TNL RRS1-R is crucial for its recognition of the effector PopP2 (Guo et al., 2020). It is currently unclear whether PTMs are important for the activation and/or stability of NLR-signaling components. Perhaps, EP-proteins and helper NLRs must also undergo PTMs in order to function properly. The additional regulation of NLR-signaling components pre-NLR activation and postNLR activation remains to be investigated ( Figure 6).
Phosphorylation of SGT1 by MAPKs is required for NLR activation, implying that NLRs are regulated by SGT1 following PTI-induced MAPK activation (Hoser et al., 2013;Yu et al., 2020). The R. solanacearum effector RipAC prevents MAPK-mediated phosphorylation of SGT1, which suppresses NLR-mediated immunity (Yu et al., 2020). Two effectors were recently shown to suppress NRC-mediated HR. The P. infestans effector AVRcap1b and the cyst nematode effector SPRYSEC15 can suppress autoimmunity induced by autoactive alleles of NRC2 and NRC3 (Derevnina et al., 2021). Suppression of NRC2 and NRC3 by AVRcap1b is dependent on the membrane trafficking-associated protein TARGET OF MYB 1-LIKE PROTEIN 9A (NbTOL9a; Derevnina et al., 2021). AVRcap1b suppresses NRC2 and NRC3 by directly interacting with their NB-ARC domains (Derevnina et al., 2021). Another Phytophthora effector (from Phytophthora capsici), PcAvh103, suppresses immunity by promoting the disassociation of the EDS1-PAD4 complex . More studies are needed to identify pathogen effectors that target the NLR signaling pathway.

The interactions between PTI and ETI
While PRR-and NLR-mediated immunity has been extensively studied for the last 20 years, it has not been clear how or if these defense mechanisms interact. NLR-mediated immunity is mostly activated in the presence of microbes or PAMPs. Most studies on NLR-mediated immunity have involved transient expression-based comparisons between PTI and "PTI + ETI." The activation of NLRs in the absence of PTI has not been extensively studied until recently. There have been multiple reports on the different interactions between these two immune systems. Here, we describe three situations in which PTI and ETI interact with each other.

Interdependency of signaling components between PRRs and NLRs
PRR co-receptors, RLCKs, NADPH oxidases, calcium channels, CPKs, and MAPKs are considered to be canonical PRRsignaling components, while EP proteins and helper NLRs are considered to be canonical NLR-signaling components. However, recent studies indicated that PRR-mediated resistance is dependent on canonical NLR-signaling components and vice versa (Ngou et al., 2021a;Pruitt et al., 2021;Tian et al., 2021;Yuan et al., 2021; Figure 7B). As mentioned, flg22-and nlp20-induced resistance is partially dependent on EDS1, PAD4, SAG101, ADR1s, and NRG1s (Pruitt et al., 2021;Tian et al., 2021). Pruitt et al. (2021) proposed that EP-proteins and helper NLRs are activated by RLPs through interactions between RLP co-receptors (SOBIR1), EPproteins, and helper NLRs, although it remains to be determined whether EP-proteins play a primary or secondary role in RLP defense signaling. Another report, however, suggested that the activation of PRRs leads to increased expression of multiple NLRs and other TIR-domain-containing proteins, promoting downstream signaling (Tian et al., 2021). These two hypotheses are not mutually exclusive, and the exact mechanisms by which PRR-mediated immunity involves NLR-signaling components remain to be determined.
NLR-mediated immunity is also dependent on PRRs and multiple PRR-signaling components. In Arabidopsis, RPS2-, RPS5-, and RRS1/RPS4-mediated resistance is dependent on BAK1 and BKK1 (Ngou et al., 2021a;Yuan et al., 2021). RPS2-mediated resistance is also dependent on BIK1 and RbohD (Kadota et al., 2019;Yuan et al., 2021). Both RPM1and RPS2-mediated resistance and the HR are dependent on CPK1/2/5/6 ( Gao et al., 2013). The activation of MPK3 and MPK6 is also required for the HR and resistance mediated by multiple NLRs including RPM1, RPS2, RPS5, and RRS1/ RPS4 (Su et al., 2018). One of the proposed key mechanisms by which ETI halts pathogen infection is to potentiate and restore PTI from turnover and the action of pathogen effectors (Ngou et al., 2021a;Yuan et al., 2021). As a result, PRRs and PRR-signaling components are required for NLRmediated resistance. The molecular mechanisms by which ETI potentiates PTI will be discussed in the next section.

Mutual potentiation between PRR-and NLR-mediated immunity
Activation of the TNLs RRS1/RPS4 and RPP4 using an estradiol-inducible recognized effector (ETI without PTI) did not trigger the HR. The presence of PAMPs/MAMPs restored the HR induced by these TNLs (Ngou et al., 2020(Ngou et al., , 2021a. Similarly, the HR induced by the CNLs RPM1, RPS2, and RPS5 was also potentiated by the activation of PRRs (Ngou et al., 2021a). In addition, the HR and resistance induced by RPS2 are compromised in PRR mutants (Ma et al., 2012;Yuan et al., 2021). There are a few possible mechanisms by which PRRs potentiate NLR-induced immunity. First, the activation of PRRs could induce the expression of NLRs and NLR-signaling components (Navarro et al., 2004;Bonardi et al., 2011;Brendolise et al., 2018;Jung et al., 2020). A recent transcriptomics study suggested that the activation of different PRRs induces highly overlapping transcriptional changes (Bjornson et al., 2021). Indeed, the activation of six distinct PRRs led to the upregulation of genes encoding most TNLs, CNLs, EP-proteins, and helper NLRs in Arabidopsis (Bjornson et al., 2021; Figure 7C; Supplemental Data Set 3). The increased abundance of these proteins might therefore "prime" the activation of NLRs upon effector recognition. Second, the activation of PRRs might prime NLR-mediated immunity via PTMs. Upon PAMP perception, SGT1 is phosphorylated by MAPKs, which is important for the stability of NLRs (Yu et al., 2020). In addition, nonsensemediated decay of NLR transcripts is inhibited upon PAMP recognition (Jung et al., 2020). Thus, the stability of NLRs can be affected by both transcriptional and posttranscriptional modifications activated by PTI. Conceivably, EP proteins and helper NLRs might also be primed via PTMs induced by PTI. Flg22 treatment led to reduced polyubiquitination levels of EDS1 (Grubb et al., 2021;Ma et al., 2021). Whether and how PTI primes NLR-signaling components remain to be investigated.
The activation of NLRs potentiates PAMP-induced cellular responses, such as ROS production, callose deposition, and defense-related gene expression (Ngou et al., 2021a). The activation of multiple PRR signaling components, such as BIK1, RbohD, and MPK3, is also potentiated by ETI (Ngou et al., 2021a;Yuan et al., 2021). ETI induces the transcript and protein accumulation of SOBIR1, BAK1, BIK1, RbohD, and MPK3 (Ngou et al., 2021a). Transcriptomic analysis confirmed that multiple PRR signaling components are also upregulated upon the activation of RRS1/RPS4. These include CPK1/2/5/6, XLG2, and the calcium channels OSCA1.3, CNGC19/20, GLR2.7/2.8/2.9 (Ngou et al., 2021a; Figure 7C; Supplemental Data Set 4). Interestingly, the transcript levels of BIK1, MPK3, and RbohD are only transiently upregulated during ETI. However, the protein levels of these genes remain upregulated for an extensive period of time (Ngou et al., 2021a). This implies that PTMs or other posttranscriptional mechanisms might also influence the stability of PRR-signaling components during ETI. The protein abundance of PRR signaling components, such as BAK1, BIK1, and RbohD, is tightly regulated by multiple processes Figure 7 Interactions between PRR-and NLR-mediated immunity. A, NLRs guarding the PRR-signaling pathway. Multiple PRR-signaling components are suppressed by effectors. NLRs guard these signaling components and reverse susceptibility triggered by these effectors. Question marks indicate unidentified effectors or NLRs. B, Tabular summary of signaling components required for PRR-and NLR-mediated immunity. Green shading represents confirmed requirement from publications. Gray shading indicates predicted requirement. Purple shading represents unclear requirement that cannot be predicted. C, Mechanisms involved in the mutual potentiation between PRR-and NLR-mediated immunity. Transcriptomic data were obtained from previously published data (Bjornson et al., 2021;Ngou et al., 2021a). Numbers indicate the corresponding mechanisms to potentiate PRR-or NLR-mediated immunity to achieve robust resistance against pathogens.
( Figure 6). How ETI regulates or affects these processes remains unclear. In addition, calcium influx induced by NLRs might contribute to the potentiation of PTI through CPKs Jacob et al., 2021;Ngou et al., 2021b). To summarize, PTI and ETI mutually potentiate each other through multiple mechanisms to induce robust immunity against pathogens ( Figure 7C).

Historic overview of research in PTI and future challenges
Researchers identified the first PRR-encoding gene, Cf-9, back in 1994 (Jones et al., 1994). Multiple PRR genes, such as Xa21, Cf-2, Cf-4, FLS2, EFR, and RLP23, were subsequently identified and used as models to study PTI (Song et al., 1995;Dixon et al., 1996;Thomas et al., 1997;Gómez-Gómez and Boller, 2000;Zipfel et al., 2006). Researchers then explored PRR-induced physiological responses and identified multiple signaling components. The activation of MAPKs by cell-surface receptors were reported back in 1997 (Ligterink et al., 1997) and was verified for Cf-genes 2 years later (Romeis et al., 1999). In tobacco (N. tabacum), the perception of PAMPs leads to the activation of wounding-induced protein kinase (WIPK) and SA-induced protein kinase (SIPK; Zhang and Klessig, 1998;Yang et al., 2001). WIPKs and SIPKs are orthologs of the subsequently identified Arabidopsis MPK3 and MPK6, respectively (Asai et al., 2002). Accumulation of ROS and callose deposition during infection were also reported in 1997 (Thordal-Christensen et al., 1997), and for Cf-initiated responses (Piedras et al., 1998). Researchers identified the human Rbohs in Arabidopsis and showed that two of these (RbohD and RbohF) are required for ROS production during infection (Torres et al., 1998(Torres et al., , 2002. It was unclear how these signaling components were activated by PRRs until the identification of the PRR coreceptors and RLCKs. BAK1 was identified as a co-receptor essential for FLS2-mediated resistance in (Chinchilla et al., 2007. In the same year, CERK1 was also shown to be essential for chitin-mediated immunity (Miya et al., 2007). In 2013, SOBIR1 was identified as a co-receptor of RLPs, and the structure of the FLS2/BAK1 receptor complex was also defined (Liebrand et al., 2013;Sun et al., 2013). In 2018, a genome-wide analysis of Arabidopsis LRR-RLKs interactions was reported, further supporting the theory that PRRs interact with each other to modulate and transduce signals (Smakowska-Luzan et al., 2018). Tomato ACIK1 was the first RLCK shown to be an essential signaling component in PRRmediated immunity (Rowland et al., 2005). The Arabidopsis ortholog BIK1 was subsequently shown to be a central PRRsignaling component (Lu et al., 2010;Zhang et al., 2010a). RbohD, MAPKKKs, and multiple calcium channels were shown to be phosphorylated by RLCKs, which leads to downstream immune responses (Boudsocq et al., 2010;Kadota et al., 2014;Yamada et al., 2016;Bi et al., 2018;Tian et al., 2019;Thor et al., 2020; Figure 8A).
More than 60 immunity-related PRRs with known ligands have now been identified. Arabidopsis EFR has been introduced into multiple plant species, such as tomato, rice, orange, and apple, providing broadspectrum resistance to many bacteria (Lacombe et al., 2010;Schwessinger et al., 2015;Mitre et al., 2021;Piazza et al., 2021). Therefore, the identification of novel PRRs that recognize PAMPs or other elicitors would provide resources to engineer diseaseresistant crops. Other challenges in PRR biology include trying to understand how PRRs activate downstream signaling components and physiological responses, how these processes are regulated and suppressed by effectors, and how resistance against pathogens is achieved ( Figure 8B).

Historic overview of research in ETI and future challenges
Arabidopsis RPS2 and the tobacco N gene were the first reported NLR genes (Bent et al., 1994;Mindrinos et al., 1994;Whitham et al., 1994). Multiple NLRs, including RPM1 and L6, were subsequently identified (Grant et al., 1995;Lawrence et al., 1995). Understanding how NLRs detect effectors has led to multiple models. The guard hypothesis was proposed to explain how the protein kinase Pto confers Prf-dependent recognition of AvrPto (Van der Biezen and . Many other examples have emerged that are consistent with this hypothesis, such as the requirement of the protease Rcr3 for Cf-2-mediated resistance (Van der Biezen and Dangl and Jones, 2001;Krüger et al., 2002). The decoy model was then proposed, which is further supported by the discovery of integrated decoy domains in NLRs (van der Hoorn and Kamoun, 2008;Cesari et al., 2014;Le Roux et al., 2015;Sarris et al., 2015Sarris et al., , 2016. The discovery of NRCs led to the concept of NLR networks (Gabriëls et al., 2007;Wu et al., 2017aWu et al., , 2018. Following the identification of multiple NLRs, researchers identified multiple genetic components required for NLR-mediated immunity. These include EDS1, NDR1, PAD4, RPW8, SGT1, RAR1, HSP90, SAG101, NRG1s, and ADR1s (Parker et al., 1996;Century et al., 1997;Zhou et al., 1998;Falk et al., 1999;Xiao et al., 2001;Azevedo et al., 2002;Takahashi et al., 2003;Feys et al., 2005;Peart et al., 2005;Bonardi et al., 2011). EDS1 was later shown to co-function with SAG101 and PAD4 to mediate HR and resistance during ETI (Feys et al., 2001(Feys et al., , 2005Wagner et al., 2013;Sun et al., 2021;Wu et al., 2021b). Similarly, ADR1 and NRG1 have been shown to function downstream of multiple sensor NLRs to mediate the HR and resistance (Castel et al., 2019a;Wu et al., 2019;Saile et al., 2020). How sensor NLRs activate these signaling components is currently under investigation. v-cADPR produced by TIR domains might contribute to the activation of EPproteins and helper NLRs (Horsefield et al., 2019;Wan et al., 2019aWan et al., , 2019b. NLRs were shown to oligomerize and trigger cytosolic calcium influx following effector recognition (Grant et al., 2000;Mestre and Baulcombe, 2006). The discovery of the structures of multiple NLR resistosomes proved that the oligomerization of NLRs is required for resistance, likely through the formation of cation channels (Wang et al., 2019a;Ma et al., 2020a;Martin et al., 2020;Bi et al., 2021;Figure 8 Historic overview of PTI and ETI and future challenges. A, Discoveries in PTI (left) and ETI (right) in the past 30 years. Bar charts represent the number of "plant biology" publications that mentioned "pattern-trigger immunity" (red) and "effector-triggered immunity" (blue). Data obtained from Dimensions (https://www.dimensions.ai/). B, Future challenges and outlook in plant immunity research. Jacob et al., 2021). However, oligomerization of TIR domains imposed by an NLRC4 scaffold is sufficient to activate defense (Duxbury et al., 2020; Figure 8A).
More than 140 NLRs with known recognized effectors have been identified (Kourelis and Kamoun, 2020). Crossspecies transfer of NLR "stacks" provides durable resistance against pathogens (Jones et al., 2003;Mukhtar, 2013;Ghislain et al., 2019;Luo et al., 2021;Witek et al., 2021). Identification of novel NLRs will provide resources to engineer crop resistance against multiple pathogens. Current challenges in NLR biology include understanding how NLRs activate downstream signaling components, how these signaling components then trigger immune responses, how these processes are regulated and suppressed by effectors, and how NLRs and PRRs co-function to achieve resistance against pathogens ( Figure 8B).

Conclusion and perspectives
Plants respond to pathogens using a two-tier innate immune system activated by both cell-surface and intracellular immune receptors. The perception of PAMPs/MAMPs/ DAMPs/HAMPs on the cell surface leads to PRR-mediated immunity, and the recognition of effectors leads to intracellular NLR-mediated immunity. The first plant Resistance (R) gene, Hm1, was cloned back in 1992 (Johal and Briggs, 1992). Many immune receptors have been identified since 1994, when the first PRR and NLRs were identified. Tremendous efforts have been made to understand the PRR-and NLR-signaling pathways. PRRs and NLRs utilize some overlapping but also unique signaling components to activate each of their downstream physiological responses, which thwart pathogen proliferation. Both signaling pathways are tightly regulated to prevent autoimmunity, while being suppressed by pathogen effectors. Recent studies have shown that PRR-and NLR-mediated immunity can be mutually potentiated and are dependent on each other. Great opportunities for novel discoveries remain in addressing the following challenges in the research of plant immunity: (1) identifying novel immune receptors; (2) understanding the signaling pathways and physiological responses triggered by both cell-surface and intracellular immune receptors; (3) understanding how immunity is intrinsically regulated and manipulated by external biotic and/or abiotic factors; (4) understanding the vastly diverse mechanisms by which plants resist pathogen infections; and (5) understanding how different immune systems function synergistically during infections. These challenges overlap with some of the "top 10 unanswered questions in molecular plant-microbe interactions" (Harris et al., 2020) and will shape our understanding of plant immunity in the coming decades ( Figure 8B).

Supplemental data
The following materials are available in the online version of this article.