Broader functions of TIR domains in Arabidopsis immunity

Significance Plant diseases are a major cause of yield loss. We identified SADR1, a TIR-containing immune receptor and found that it participates in immune signaling and in a pathogen containment mechanism sufficient to prevent the spread of a virulent pathogen. SADR1 acts partially independently of EDS1, the receptor for TIR-derived signaling molecules. Inhibition of TIR enzymatic activity with nicotinamide revealed a broader function of TIR domains in immunity. TIR domains can act independently of EDS1 to potentiate immune signaling triggered by other immune receptors.

TIR domains are NAD-degrading enzymes that function during immune signaling in prokaryotes, plants, and animals. In plants, most TIR domains are incorporated into intracellular immune receptors termed TNLs. In Arabidopsis, TIR-derived small molecules bind and activate EDS1 heterodimers, which in turn activate RNLs, a class of cation channel-forming immune receptors. RNL activation drives cytoplasmic Ca 2+ influx, transcriptional reprogramming, pathogen resistance, and host cell death. We screened for mutants that suppress an RNL activation mimic allele and identified a TNL, SADR1. Despite being required for the function of an autoactivated RNL, SADR1 is not required for defense signaling triggered by other tested TNLs. SADR1 is required for defense signaling initiated by some transmembrane pattern recognition receptors and contributes to the unbridled spread of cell death in lesion simulating disease 1. Together with RNLs, SADR1 regulates defense gene expression at infection site borders, likely in a non-cell autonomous manner. RNL mutants that cannot sustain this pattern of gene expression are unable to prevent disease spread beyond localized infection sites, suggesting that this pattern corresponds to a pathogen containment mechanism. SADR1 potentiates RNL-driven immune signaling not only through the activation of EDS1 but also partially independently of EDS1. We studied EDS1-independent TIR function using nicotinamide, an NADase inhibitor. Nicotinamide decreased defense induction from transmembrane pattern recognition receptors and decreased calcium influx, pathogen growth restriction, and host cell death following intracellular immune receptor activation. We demonstrate that TIR domains can potentiate calcium influx and defense and are thus broadly required for Arabidopsis immunity.
In a forward genetic screen, we sought to identify genes required for immunity and cell death activation by RNLs. We found that the TNL Suppressor of ADR1-L2 1 (SADR1) is required for the phenotypes driven by ADR1-L2 autoactivity but is dispensable for other TNL functions. We found that SADR1 regulates defense triggered by the activation of a plasma membrane pattern recognition receptor and the "runaway cell death" phenotype in the Arabidopsis mutant lesion simulating disease 1 (lsd1) (15). Because these responses involve the perception of extracellular signals, we investigated the requirement for SADR1 and RNLs in the spatial regulation of defense. Virulence effectors delivered to the plant cell from the pathogen Pseudomonas syringae pv. tomato (Pst) DC3000 increase transcriptional defense responses around the infection site. This pattern of host gene expression requires RNLs and SADR1. The loss of defense gene expression on the infection border is associated with the systemic spread of Pst DC3000. We found that SADR1 is required for the residual ADR1-L2 autoactivity in the absence of EDS1. These results indicate that SADR1 functions downstream of ADR1-L2 activation partially independently of EDS1 and is thus distinct from the canonical TNL-EDS1-RNL pathway. We tested the requirement for EDS1-independent TIR function in plant immunity using a pharmacological inhibitor of TIRdependent NADase enzymatic function. We discovered that TIR function is generally required to potentiate immune responses triggered by a plasma membrane pattern recognition receptor, RNLs, and CNLs. Importantly, inhibition of TIR function decreased Ca 2+ influx resulting from RNLs and CNLs, suggesting that TIR function can generally potentiate Ca 2+ influx in the context of immune signaling.

Results
To identify signaling components downstream of ADR1-L2, we screened for mutants able to suppress the autoimmunity-associated stunted growth phenotype of an Arabidopsis transgenic line expressing the activation mimic mutant ADR1-L2 D484V from the native promoter, hereafter ADR1-L2 DV [(16), see Materials and Methods for full genotype]. This mutation in the MHD motif is commonly used to mimic NLR activation and can be suppressed in cis by P-loop mutations (17)(18)(19)(20). ADR1-L2 DV expressing plants exhibit hallmarks of autoimmune signaling: stunted growth, ectopic cell death activation, ectopic salicylic acid accumulation, and induction of defense gene expression, including ADR1-L2 itself [ Fig. 1 and (16)].
Identification of ADR1-L2 D484V Suppressors. We identified phenotypically suppressed mutants and performed bulk segregant analysis using suppressed plants from segregating backcrossed F2 populations. Two mutants, 26.6 and 30.4 had mutations in the same gene, located in a genomic region cosegregating with the suppression phenotype in F2 plants ( Fig. 1 A and B). Whole genome resequencing of 32 additional suppressed M3 mutants allowed the identification of three additional mutant alleles of the same gene, 12.6, 17.4, and 30.3 (Fig. 1B). Overall, 5 out of the 39 suppressor mutants identified were affected in this gene (the others will be described elsewhere), which we consequently named Suppressor of ADR1-L2 1 (SADR1, AT4G36150). RNA sequencing showed that mutations 26-6 and 30-4 suppressed the vast majority of ADR1-L2 DV-driven gene expression changes (SI Appendix, Fig. S1 and Table S1). We created a sadr1-c1 loss-of-function allele in ADR1-L2 DV expressing plants using CRISPR-Cas9 (c.87_88insA, leading to a frameshift after Q29; Materials and Methods). The sadr1-c1 mutation suppressed the ADR1-L2 DV stunted growth phenotype and constitutive expression of Pathogenesis-Related 1 (PR1). Importantly, ADR1-L2 DV mRNA and protein levels were reverted to wild-type levels ( Fig. 1 C-F). This demonstrates that SADR1 is required for ADR1-L2 DV self-amplification, an important feature of RNL signaling (16,21).
Functional Characterization of SADR1. We characterized the function of SADR1 in defense. We generated a sadr1-c2 loss-offunction mutant (c.39_88del leading to a frameshift after V12) with CRISPR-Cas9 in the wild-type Col-0 background (Fig. 1B). We observed that SADR1 was not required for basal resistance to the virulent pathogen Pst DC3000 or to the avirulent strains Pst DC3000 AvrRpt2 and Pst DC3000 AvrRps4, which activate the CNL RPS2 and the TNL pair RPS4-RRS1 immune receptors, respectively (Materials and Methods). As a control, we showed that the RNL defective helperless quintuple mutant [adr1 adr1-l1 adr1-l2 nrg1.1 nrg1.2 (11,13)] was indeed more susceptible to infection than Col-0 in each of these situations (SI Appendix, Fig. S4, A-C).
We investigated the possibility that loss of SADR1 could be affecting only ADR1-L2 and thus be compensated by redundant RNL signaling. We compared adr1 adr1-l1 and adr1 adr1-l1 sadr1-c2 to adr1 adr1-l1 adr1-l2 mutants during TNLdriven immunity following challenge with Pst DC3000 AvrRps4 (SI Appendix, Fig. S4D). ADR1s were required for full bacterial growth restriction in these conditions, as seen with the 10-fold increase in pathogen growth in adr1 adr1-l1 adr1-l2 compared to Col-0. ADR1-L2 RNL function to "help" RPS2, which was required for approximately half of this growth restriction, was not affected by sadr1-c2 loss of function (SI Appendix, Fig. S4D). Resistance to Hyaloperonospora arabidopsidis isolate Cala2, which activates the TNL RPP2, was also not affected by sadr1-c2 (SI Appendix, Fig. S4 E and F). In addition, SADR1 was not required for the autoactive phenotype of the snc1 TNL allele [SI Appendix, Fig. S4 G and H (22)]. These results indicate that SADR1 is not required for RNL-driven defense against virulent or avirulent bacteria, at least for the TNL functions we measured.
ADR1-L1 and ADR1-L2 are also required for "runaway cell death", the superoxide-driven self-perpetuating cell death observed in lesion simulating disease 1 (lsd1) (15,25). To test whether SADR1 mediates runaway cell death, we treated 4-wk-old plants with BTH (benzothiadiazole), a salicylic acid analog (27) which triggers runaway cell death in lsd1. After 2 wk, we measured fresh and dry weight of the BTH-treated plants to estimate the extent of cell death induction (Materials and Methods). As expected, lsd1 displayed extensive lesions covering most or all of the plant, which resulted in a very low fresh:dry weight ratio in lsd1 compared to the suppressed lsd1 adr1-l2 phenotype [ Fig. 2 B and C; (25)]. lsd1 sadr1-c2 exhibited an intermediate phenotype, indicating that SADR1 contributes positively to runaway cell death ( Fig. 2 B and C). Overall, SADR1 is not required for TNL signaling but is involved in PAMP signaling and lsd1 runaway cell death.

Spatial Regulation of Defense by RNLs and SADR1 Prevents
the Systemic Propagation of Pst DC3000. The lsd1 runaway cell death phenotype is non-cell autonomous because the induction of self-perpetuating cell death depends on the proximity and perception of a dead or dying cell (15). PRR signaling also involves non-autonomous relay of defense gene activation in neighboring cells (28). Similarly, damage-associated molecular patterns (DAMPs) trigger calcium-dependent defense responses in surrounding tissue (29). Consistent with this, expression of PR1 occurs in the area surrounding the cells undergoing cell death during NLR-mediated immune responses (30,31). Using reporter plants (31) expressing YFP NLS under the control of the PR1 promoter, we reproduced and extended these observations. We used mCherry-tagged bacteria and observed that inoculation with either Pst DC3000 EV (virulent) or Pst DC3000 AvrRpt2 or Pst DC3000 AvrRps4 (activating the CNL RPS2 or the TNL RPS4, respectively) induced a pattern of PR1 expression at the border of the infection site (SI Appendix, Fig. S5). This pattern did not result from inhibition of PR1 expression in the infection zone by coronatine (SI Appendix, Fig. S5), a pathogen-derived phytotoxin and jasmonic acid mimic known to antagonize SA signaling and inhibit PR1 expression (32). The pattern of PR1 expression was not observed in plants challenged with Pst DC3000 hrcC-, which cannot deliver virulence effectors, consistent with effector-dependent defense inhibition (SI Appendix, Fig. S5). At 6 hpi, PR1 promoter activity appeared to be enhanced in NLRactivating inoculations compared to Pst DC3000 hrcC-, suggesting that NLR signaling increased defense around the infection site (SI Appendix, Fig. S5A). At 24 hpi, bacterial growth led to a visible mCherry signal, largely nonoverlapping with YFP-positive areas defining PR1 expression (SI Appendix, Fig. S5B). Overall, cells expressing PR1 are likely not subjected to effector-driven defense inhibition and are spatially separated from the bacteria. These results suggest that NLR signaling relays defense gene activation in areas devoid of type III effectors, possibly through DAMP activation or reactive oxygen-based signaling (15,29).
We next investigated whether RNLs and SADR1 were involved in spatial regulation of defense. We infected Col-0 pPR1:YFP NLS reporter plants mutated in SADR1 or RNLs with a high-concentration inoculum (OD 600 = 0.2) of Pst DC3000 AvrRps4 mCherry to activate the TNL RPS4. Fluorescence observation and qPCR quantification of PR1 mRNA indicated that both SADR1 and RNLs (ADR1s in particular) regulate defense at the borders of infection sites (Fig. 2 D and E). These results suggest that SADR1 and RNLs mediate defense gene expression at the borders of infection sites.
Activating defense around the infection area could serve to prevent the systemic propagation of pathogens. To test this hypothesis, we inoculated half-leaves with Pst DC3000 EV and isolated noninfiltrated tissues from the same leaves after 10 d ( Fig. 3; Materials and Methods). We found that only a very low level of Pst DC3000 propagates to the noninfiltrated side of the leaf in Col-0 (Fig. 3B). Similar levels were found in sadr1-c2 and nrg1.1 nrg1.2 mutants. However, we observed a dramatic increase in the spread of disease symptoms and bacterial growth in noninfiltrated tissues in adr1 adr1-l1 adr1-l2 and helperless plants at 10 d after inoculation. ADR1 defective plants exhibited systemic disease symptoms after four weeks, including reduced growth, anthocyanin accumulation, and systemic lesions (Fig. 3D). Overall, RNLs induce defense gene expression at the infection site borders, and ADR1s in particular are required to limit systemic Pst DC3000 propagation and disease spread from a localized infection event. SADR1 and NRG1s contribute to defense induction but are not required for this disease resistance mechanism. These results suggest that the selective activation of ADR1s (4, 5) is most relevant for this phenotype. SADR1 Functions Partially Independently of EDS1. SADR1 is required for ADR1-L2 DV activation mimic phenotypes, but it is not required for either RPS4, RPP2, or the snc1 autoactivity TNL phenotypes (SI Appendix, Fig. S4). TNLs regulate immunity by activating ADR and NRG RNLs through the selective TIR ligandbound forms of the EDS1-PAD4 or EDS1-SAG101 heteromeric complexes, respectively (4,5). We sought to understand why SADR1 TNL activity would be required in a context where an RNL, ADR1-L2, is already active. We hypothesized that SADR1 could be amplifying the defense signal initiated by ADR1-L2 DV in a positive feedback loop, as evidenced by the expression data in Fig. 1. Different eds1 loss-of-function alleles differentially affect the ADR1-L2 DV autoimmune phenotype (16, 33). We repeated these observations but with the "clean" CRISPR deletion eds1-12 allele (34). We found that eds1-12 only partially suppresses the ADR1-L2 DV stunted growth phenotype (Fig. 4 A and B). In contrast, the sadr1-c1 allele fully suppressed ADR1-L2 DV-driven stunted growth and defense priming (Fig. 4). These results define an EDS1-independent potentiation of ADR1-L2 DV activity by SADR1 that is retained in ADR1-L2 DV eds1-12 plants (where ADR1-L2 DV is expressed from the native promoter; see SI Appendix, Materials and Methods). We next questioned whether this potentiation of ADR1-L2 DV activity was specific to SADR1. Introgression of the autoactive snc1 or CHilling Sensitive 2 [chs2-1 (35)] TNLs into the fully suppressed ADR1-L2 DV sadr1-c1 background also restored some ADR1-L2 DV activity, suggesting that potentiation of ADR1-L2 DV can also be provided by at least this additional autoactive TNLs (SI Appendix, Fig. S6). Therefore, SADR1 can potentiate ADR1-L2 DV activity independently of EDS1, and this function may be shared by other active TIR-containing proteins.

Nicotinamide Inhibits TIR Enzymatic Function in planta.
Arabidopsis possesses a large repertoire of TIR-containing proteins (2), many of which are transcriptionally induced during the initial steps of defense signaling. TNL overexpression is often sufficient to trigger immune responses (24). Consequently, evaluating the contribution of potential EDS1-independent TIR signaling to defense would benefit from an inhibitor of TIR enzymatic function. The first step of plant TIR enzymatic pathway involves cleavage of NAD + into nicotinamide (NAM) and ADPR (36). Interestingly, high concentrations of NAM (50 mM) inhibit the NADase activity of the mammalian CD38 NADase and have been deployed in Arabidopsis tissues to inhibit cADPR accumulation, suggesting that NAM could inhibit plant TIR NADase activity (37,38).
We looked for a readily measurable bioindicator of TIR enzymatic activity in planta because it is difficult to detect the TIR-derived molecules that are the signaling ligands for EDS1-dependent heteromers (4,5). 2′/3′-RA [2′-O-β-D-ribofuranosyladenosine or 3′-O-β-D-ribofuranosyladenosine] nucleotide metabolites similar to 2′/3′cADPR but lacking the pyrophosphate groups accumulate during TIR-dependent plant immune responses in planta (7,39,40). We transiently expressed in N. benthamiana leaves either active full-length TNLs or TIR domains fused with SARM1 oligomerization domain [the SAM domain, which enhances TIR activation; (36)]. We used the corresponding TNLs or TIR domains rendered inactive by mutation of their respective catalytic glutamic acid residues as negative controls (SI Appendix, Fig. S7; Materials and Methods). SADR1 TIR-SAM was able to trigger cell death in N. benthamiana, while both SADR1-P1 TIR1-SAM and TIR2-SAM could not (SI Appendix, Fig. S7 A and B). However, SADR1-P1 TIR1 and TIR2 protein accumulation could not be detected; thus, it is unclear if these TIR domains are functional or not (SI Appendix, Fig. S7D). SADR1-driven cell death required EDS1 and the conserved catalytic glutamic acid (SI Appendix, Fig. S7 A and C). We found that active TNLs or TIR-SAM domain fusions reliably induced the accumulation of 2′/3′-RA [SI Appendix, Fig. S7E; (39)]. This accumulation was dependent on the conserved catalytic glutamic acid in all cases. Interestingly, a SADR1 TIR-SAM domain fusion induced a very small and inconsistent accumulation of 2′/3′-RA, suggesting it may act differently than RPP1, RPS4, or BdTIR (SI Appendix, Fig. S7E). However, overall, 2′/3′-RA is a reliable and readily measured bioindicator of TIR enzymatic activity.

TIR Enzymatic Function Is Broadly Required for Defense.
We therefore used NAM treatment to evaluate the contribution of TIR enzymatic function to defense. TIR activity and subsequent EDS1-dependent immune signaling contribute not only to ETI activated by TNL receptors but also to basal defense and consequent growth restriction of Pst DC3000 EV. This is because basal defense responses include "weak" ETI, at least some of which is likely to be TNL and EDS1 dependent (43,44). Also, as noted above, defense responses include transcriptional upregulation of many TIR domain-encoding genes which could boost immunity via production of TIR enzymatic products to functional levels (1,24,45,46).
We treated Col-0, sadr1-c2, and helperless plants with NLP20, with or without coinoculation of 50 mM NAM to test general TIR function. NAM inhibited PR1 induction following NLP20 treatment in Col-0 and suppressed it in sadr1-c2, which we demonstrated above is required for full NLP20-dependent signaling (Fig. 5A). We then tested the impact of 50 mM NAM on defense against virulent and avirulent Pst DC3000. NAM significantly inhibited resistance against Pst DC3000 EV and Pst DC3000 AvrRps4 in Col-0 but not in eds1, consistent with NAM inhibiting an EDS1-dependent defense pathway (Fig. 5 B and C). We noted that NAM cotreatment inhibited pathogen growth in otherwise hypersusceptible eds1 plants. We therefore tested the impact of NAM on bacterial growth in minimal (MS) or rich culture medium (LB) and found that NAM was also bacteriostatic, potentially explaining why NAM had a negative impact on bacterial growth in eds1 plants (SI Appendix, Fig. S9). These results collectively indicate that the impact of NAM on plant defense as measured by bacterial growth is likely to be underestimated. Overall, NAM inhibits EDS1-dependent defenses.
We used RNA-seq to characterize the effect of NAM treatment, and thus overall TIR activity, on defense. We infected Col-0 and eds1 plants with Pst DC3000 EV to trigger basal defense or Pst DC3000 AvrRps4 to activate TNL RPS4, with or without 50 mM NAM, and we identified genes inhibited by NAM treatment (SI Appendix, Figs. S10 and S11, Table S2 and Materials and Methods). We observed that a large number of infection-regulated genes are affected by NAM (SI Appendix, Fig. S10A). NAM treatment alone regulated mostly genes related to "stress" or "response to chemical" (SI Appendix, Fig. S11). More than half of the genes regulated by NAM treatment alone were also regulated by infection (SI Appendix, Fig. S10B). We defined genes regulated by infection with each strain and then subdivided these genes into NAM-sensitive (genes differentially expressed in Col-0 without NAM but not in the presence of NAM), EDS1-dependent (genes differentially expressed in Col-0 but not in eds1), and TNL RPS4dependent genes (genes differentially expressed in Col-0 infected with Pst DC3000 AvrRps4 but not with Pst DC3000 EV). We found that 85% and 67% of NAM-sensitive genes were also either EDS1 dependent or specifically RPS4 regulated, respectively, during Pst DC3000 AvrRps4 infection (SI Appendix, Table S2). NAM fully inhibited 20% of genes up-regulated during Pst DC3000 AvrRps4 infection but up to 41% of strictly RPS4-dependent genes (SI Appendix, Table S2). We conclude from these analyses that NAM predominantly affects TIR-dependent transcriptional outputs.
We hypothesized that general activation of TNLs and TIR domain proteins could also contribute to CNL-dependent immune responses. We tested the impact of NAM on CNL-dependent signaling. RPM1 and ZAR1 activate defense in response to AvrRpm1 and HopZ1a, respectively, in a Ca 2+ -dependent manner and independently of EDS1 or RNLs (9,47). 50 mM NAM inhibited RPM1-or ZAR1-dependent growth restriction of Pst DC3000 AvrRpm1 and Pst DC3000 HopZ1a, respectively, in Col-0 and in mutant plants lacking the cognate CNLs (Fig. 5 D and E). In addition, NAM treatment delayed cell death induction by the TNL RPS4 and the CNL RPM1 but not by the CNL ZAR1 (Fig. 5 F-H). Overall, inhibition of TIR enzymatic activity by 50 mM NAM inhibits NLP20 signaling, basal defense against virulent bacteria, and both RNL-dependent immune responses triggered by TNLs and RNL-independent immune responses triggered by CNLs. These data collectively argue for a broad role for TIR activity in defense responses.

Discussion
SADR1 is a newly defined TNL that is required downstream of RNL activation for full NLP20 response (Fig. 2). Interestingly, SADR1 is up-regulated by NLP20 but not by flg22 treatment (24), suggesting SADR1 functions with specific PRRs. RNLs are required for some PAMP responses and are located in close subcellular proximity to PRRs (23)(24)(25). It is tempting to speculate that RNLs and SADR1 survey PRR activation to boost defense. SADR1 is also involved in lsd1 runaway cell death (Fig. 2). Additionally, SADR1 and RNLs are required for PR1 expression at the border of a bacterial inoculation site during RPS4-dependent ETI (Fig. 2). This requirement likely can be extended to other ETI contexts as SADR1 and RNLs also regulate SA signaling during PTI [ Fig. 2 A and 5 A (23-25)]. NLP20 signaling, lsd1 runaway cell death, and PR1 expression each involves a spatial component (23). We hypothesize that this pattern could be explained by non-autonomous signaling from pathogen-engaged to nonengaged cells as in lsd1 (15) or by a gradient of effector-triggered defense inhibition in neighboring cells. The cells expressing PR1 during the response to Pst DC3000 AvrRpt2 are not dead at the time of observation, contrary to the cells directly in contact with the bacteria, and therefore are likely not activating RPS2 (31). In addition, infection with Hpa Emwa1, which triggers the TNL RPP4, leads to strong PR1 expression in a layer of cells bordering the cells in direct contact with the oomycete (30). NLR signaling in one dying cell would result in the leakage of immunogenic molecules, DAMPs, or reactive oxygen, activating immunity in nearby naive cells. It is worth recalling that ADR1s RNLs are required for non-autonomous feedforward cell death signaling as ADR1-L2 mutation suppresses runaway cell death (16). A non-autonomous immune signaling mechanism was implicated in the context of cell damage, but a requirement for RNLs was not tested (28,29). Thus, a role in potentiating non-autonomous signaling from pathogen-engaged to nonengaged cells provides a harmonious explanation of SADR1 function.
We hypothesized that SADR1-mediated regulation of immune response could define a pathogen containment strategy. We found that Pst DC3000 EV could infect plants systemically in the absence of ADR1s (Fig. 3). Consistent with these results, Pst DC3000 can propagate systemically in N. benthamiana only in the absence of TNL activation (50). RNL mutants lacking ADR1s exhibited systemic symptoms (growth retardation, anthocyanin accumulation, and lesions), whereas Col-0 did not (Fig. 3). Therefore, RNLs and consequent Ca 2+ signaling limit disease from localized Pst DC3000 EV infection events. These results highlight the importance of bacterial containment as a disease resistance mechanism. We note that Pst DC3000 triggers basal defense which involves weak ETI and upregulation of TIR domain proteins and is thus at least partially TIR dependent (24,43,44,51).
We noted that SADR1 is required for ADR1-L2 DV overaccumulation, suggesting that SADR1 is required for an as yet unidentified process ultimately affecting ADR1-L2 transcription and protein accumulation (Fig. 1 D, E, and F). However, it is difficult to distinguish if the lower accumulation of ADR1-L2 DV is the cause or the consequence of sadr1-c1 suppression of ADR1-L2 DV phenotypes. Any ADR1-L2 DV suppressor mutation would likely lead to a reduction in ADR1-L2 mRNA and protein levels since ADR1-L2 DV triggers its own expression (16). SADR1 also functions downstream or at the level of Ca 2+ influx (48). A positive amplification loop was postulated in TIR domain signaling (3), and RNLs are known to trigger self-amplification through an SA-based positive feedback loop (16). Our collective data demonstrate that SADR1 is a required component of this feedback loop.
We observed that SADR1, but not EDS1, was fully required for ADR1-L2 DV autoactivity (Fig. 4). This surprising finding suggests that SADR1 function is at least partially independent of EDS1 (Figs. 1, 2, and 4 and SI Appendix, Fig. S1). SADR1 is, however, dispensable for other TNL functions (RPS4, RPP2, and snc1) and is not required for ADR1-L2 function when the TNL RPS4 is activated (SI Appendix, Fig. S4). In addition, autoactive snc1 or chs2-1 partially restore ADR1-L2 DV activity in the suppressed ADR1-L2 DV sadr1-c1 background (SI Appendix, Figs. S4 and S6). These results collectively suggest that the mechanism underlying SADR1 function downstream of RNLs may not be specific to SADR1 and may be shared by multiple TIR domain proteins. Future studies will address the regulatory mechanism governing SADR1 and SADR1-P1 activation, the function of SADR1-P, and downstream signaling. Interestingly, during the writing of this manuscript, it was found that overexpression of SADR1 by AITF1 (ARS3 interacting transcription factor 1) leads to EDS1-and RNL-dependent autoimmunity (52). It is unclear if SADR1 is upstream or downstream EDS1/RNLs in this overexpression context. ADR1-L2 is functionally redundant with the Ca 2+ -permeable channel ADR1 and possesses an N-terminal motif required for ion flux in ADR1 and NRG1.1 (11,48). Activation of a Ca 2+ channel should be associated with pleiotropic defects as Ca 2+ also regulates growth and development. However, we observed that sadr1-c1 fully suppressed the stunted growth and defense gene expression activation mimic syndrome induced by ADR1-L2 DV (Figs. 1 and 3 and SI Appendix, Fig. S1). Therefore, SADR1 is likely to regulate ADR1-L2 DV activity at the level of Ca 2+ influx. Consistent with this hypothesis, inhibition of TIR NADase function with NAM decreases [Ca 2+ ] cyt levels in the context of both coupled TNL-RNL signaling and CNL signaling (Fig. 6). We cannot rule out the possibility that NAM inhibits other NADases like poly(ADP-ribose) polymerases or sirtuins. However, there are no other plant NADases known to regulate Ca 2+ flux and cell death. In fact, TIRs are the only known plant NADases with ADPR cyclase activity which is linked to regulation of Ca 2+ (53,54). TIR enzymatic function may positively influence [Ca 2+ ] cyt levels by activating Ca 2+ influx mechanisms (including RNLs themselves), by inhibiting Ca 2+ sequestration, or both. Interestingly, cADPR can regulate [Ca 2+ ] cyt levels in animals by the regulation of ryanodine receptors, a class of Ca 2+ channels involved in the calcium-induced calcium release mechanism (55). Although plants do not possess ryanodine receptors, cADPR can also regulate [Ca 2+ ] cyt levels in plants (56)(57)(58). TIR domains are the only known proteins with ADPR cyclase activity in plants (53). Investigating the potentially varied mechanisms by which TIRs regulate [Ca 2+ ] cyt is key to fully understanding the plant immune system.

Materials and Methods
A detailed description of Materials and Methods used in this study can be found in SI Appendix.
Plant Material and Growth Conditions. Plants were grown in short-day conditions (8- RNL-free helperless (59), pPR1:YFP NLS (31), rps2-101C (60), rpm1-3 (61), snc1 (22), chs2-1 (35), and zar1-3 (62) mutants have been described. The sadr1-c1 and sadr1-c2 mutations were introduced with CRISPR-Cas9 into pADR1L-2:ADR1-L2 D484V adr1-l2-4 and Col-0, respectively. Pathogen Infection Assays. P. syringae pv tomato DC3000 syringe infiltrations were performed as previously described (11). Plants were covered with a humidity dome for at least 30 min prior to the start of the experiment to facilitate infiltration. Bacteria grown overnight on solid King's B (KB) medium at room temperature, resuspended into 1 mL of 10 mM MgCl 2 , and diluted to the appropriate optical density 600 nm (OD 600 ) in 10 mM MgCl 2 . When using nicotinamide (NAM, Sigma-Aldrich N0636), the dried NAM was directly added to the infiltration solution to a final concentration of 50 mM right before infiltration to limit the potential toxicity of NAM. To determine pathogen sensitivity, four leaves from four plants were infiltrated with a 1-mL insulin syringe, left to dry for 2 h, and covered with a humidity dome for 24 h. After 3 d, four samples consisting of four 0.5-cm −2 leaf discs from four different plants were gathered and ground in 1 mL distilled water. Samples were serially diluted in water, and 5 µL was spotted on KB medium supplemented with the appropriate antibiotics. For half-leaf pathogen propagation assays, plants were infiltrated with Pst DC3000 at OD 600 = 0.001. Only half-leaf was infiltrated, and humidity domes were kept for 48 h after infiltration. After 10 d, infiltrated leaves were gathered and surface-sterilized for 1 min in 70% ethanol, and the noninfiltrated part of the leaves (starting from half a mm away from the midvein) was dissected using a sterile razor blade and dried with Kimtech wipes. Samples consisting of four half-leaves from four plants were weighed and ground in 1 mL water, serially diluted, and spotted on KB supplemented with rifampicin. For dip-inoculation assays, 14-d-old plants grown through a mesh in 3-inch round pots were dipped in solutions of bacteria and Silwet L77 0.02% in 10 mM MgCl 2 . Samples consisting of three to five plantlets were weighed and ground in 1 mL water, serially diluted, and spotted on KB supplemented with rifampicin.
The impact of NAM on hypersensitive cell death was studied by infiltrating half-leaves with a saturating solution of avirulent Pst DC3000 (OD 600 = 0.2) and observing the samples with UV lamps at 6 (AvrRpm1) or 20 (AvrRps4 and HopZ1a) hours after infection. Cell death was evidenced by high green autofluorescence and loss of red chlorophyll fluorescence.
Hyaloperonospora arabidopsidis infection assays were performed as described (11). Hpa isolate Cala2 was propagated on eds1-12 mutants for 3 wk prior to infection. Plants were grown in 3-inch round pots for 11 d before being sprayed with approximately 1 mL of an Hpa spore solution at 50,000 spores per mL. Plants were covered with a humidity dome, and spores were counted after 7 d. Plants were carefully placed in a 2-mL Eppendorf tube containing 1 mL water and vigorously vortexed. Spores were counted with a hemacytometer. Approximately 10 plants were used for trypan blue staining as previously described (63). Plants were placed in lactophenol-trypan blue (10 mL lactic acid, 10 mL glycerol, 10 g phenol, and 10 mg trypan blue dissolved in 10 mL distilled water and diluted 1:2 in ethanol right before use) at 60 °C for at least an hour and then destained in chloral hydrate overnight or as required. Observations were performed on the Leica DMi8 (Leica Microsystems).
Data, Materials, and Software Availability. RNA sequencing data have been deposited in [GEO] (GSE225075). All study data are included in the article and/ or SI Appendix.