Structure–function analyses of coiled-coil immune receptors define a hydrophobic module for improving plant virus resistance

Abstract Plant immunity relies on nucleotide-binding oligomerization domain (NOD)-like receptors (NLRs) that detect microbial patterns released by pathogens, and activate localized cell death to prevent the spread of pathogens. Tsw is the only identified resistance (R) gene encoding an NLR, conferring resistance to tomato spotted wilt orthotospovirus (TSWV) in pepper species (Capsicum, Solanaceae). However, molecular and cellular mechanisms of Tsw-mediated resistance are still elusive. Here, we analysed the structural and cellular functional features of Tsw protein, and defined a hydrophobic module to improve NLR-mediated virus resistance. The plasma membrane associated N-terminal 137 amino acid in the coiled-coil (CC) domain of Tsw is the minimum fragment sufficient to trigger cell death in Nicotiana benthamiana plants. Transient and transgenic expression assays in plants indicated that the amino acids of the hydrophobic groove (134th–137th amino acid) in the CC domain is critical for its full function and can be modified for enhanced disease resistance. Based on the structural features of Tsw, a super-hydrophobic funnel-like mutant, TswY137W, was identified to confer higher resistance to TSWV in a SGT1 (Suppressor of G-two allele of Skp1)-dependent manner. The same point mutation in a tomato Tsw-like NLR protein also improved resistance to pathogens, suggesting a feasible way of structure-assisted improvement of NLRs.


Introduction
Lack of robust genetic resistance to infectious microorganisms in crops is one of the major reasons for the rapid spread of many microbial diseases that severely reduce crop yield and quality annually. We have enough understanding of the genetic networks controlling plant immune systems now, and this knowledge can be applied to engineer pathogen-resistant plants. Plants employ multiple layers of defence, including disease resistance (R) genes, RNA silencing, and defensive phytohormone signalling pathways, to prevent the spread of pathogenic diseases and plant death (Zhou and Zhang, 2020;Ye et al., 2021). R genes typically encode nucleotide-binding leucine-rich repeat (NLR) proteins on plant cell surfaces or intracellular regions to perceive microbial pathogen-derived molecules (effectors), executing immune signalling by initiating the activation of programmed cell death (Tang et al., 2017;Kourelis and van der Hoorn, 2018;Van de Weyer et al., 2019;Zhou and Zhang, 2020).
The recent integration of the combined knowledge generated from structural, cellular, and biochemical analyses of NLR-mediated plant immunity has demonstrated that both the indirect and direct recognition of pathogens can trigger the oligomerization of plant NLRs into active complexes. Recent findings on Arabidopsis thaliana ZAR1 (HOPZ-ACTIVATED RESISTANCE1) have demonstrated that the activated resistosome could result in an array of immune signals including the influx of calcium (Ca 2+ ), which triggers immune responses that often culminate in host cell death to block pathogen survival (Wang et al., 2019a, Ma et al., 2020, Bi et al., 2021. In particular, the amino acid situated in the channel pore of N-terminal ZAR1 CC (coiled-coil) domain determines its ion channel activity. In fact, the CC domain of NLRs is usually considered as the signalling execution domain of NLRs; isolated CC domains of several NLRs are reported to induce cell death (Horsefield et al., 2019;Baudin et al., 2020). Moreover, the CC domain is also associated with the indirect surveillance of pathogen effectors, and could directly bind downstream signalling molecules (Kourelis and van der Hoorn, 2018). These modular architectures of plant NLRs offer a flexible platform for developing diverse strategies to perceive pathogen effectors, and to enhance the strength of the immune response. A pioneering study has shown that the protease cleavage site of PBS1 (Avrpphb susceptible1), which activates the NLR protein RPS5 (Resistance to Pseudomonas syringae 5) when cleaved, can be genetically modified to detect protease effectors from a wide range of pathogens. This strategy is expected to be effective to expand the pathogen recognition spectrum in the engineered plant (Kim et al., 2016). Beside this conceptive strategy, we hypothesized that it might be possible to develop more effective immunity against pathogens by structure-function analysis, which could assist improvements of NLRs for crops lacking known disease resistance genes.
Tospoviruses (Bunyaviridae), mainly the tomato spotted wilt orthotospovirus (TSWV), represent a constant threat to plants, through infecting about 800 plant species from over 50 families, and causing substantial crop losses worldwide (Oliver and Whitfield, 2016). TSWV mainly transmits among plants through their insect vector thrips, and could also be seed transmissible (Wu et al., 2019;Wang et al., 2022). Natural genetic sources of germplasms and resistance to tospoviruses for crop breeding are very limited, and new breeding and biotechnological approaches to obtain resistant crop cultivars are in high demand. Several groups, including ours, have demonstrated that a major defence phytohormone jasmonate plays critical roles in resistance against replication of tospoviruses and viral trans-plant transmission (Abe et al., 2008;Wu et al., 2019). Beside phytohormone-mediated host defence, so far only two CC-type NLR (CNL) resistance proteins against tospoviruses have been identified (Boiteux, 1995;Brommonschenkel et al., 2000;Spassova et al., 2001;Kim et al., 2017). Tomato Sw-5b is a CNL protein with one extra N-terminal domain, which coordinates with the CC domain to regulate its autoinhibition and activation Zhu et al., 2019). The movement protein (NSm) of TSWV is recognized by tomato Sw-5b (De Oliveira et al., 2016;Zhu et al., 2017). Another CNL protein Tsw confers pepper resistance by detecting the effector protein Non-structural protein (NSs) of TSWV, but the induced resistance shows low efficacy, a narrow pathogen spectrum, and temperature sensitivity, and has been frequently overcome by rapidly evolving pathogens in the pepper Capsicum (Turina et al., 2016;Kim et al., 2017;Chung et al., 2018;de Ronde et al., 2019;Yoon et al., 2021). Meanwhile, tospoviruses have evolved to suppress host defences to promote the population of the insect vector, thereby expanding the disease pandemic (Wu et al., 2019;Wu and Ye, 2020). Considering that the economic impact of tospoviruses has much increased during the last three decades, it is necessary to reinforce the efforts to characterize new/alternative resistance genes, and search for improvements on known NLR-involved immunity against tospoviruses.
In this study, we functionally characterized Tsw as a CNL with a 137 aa plasma membrane-associated CC domain that acts as an executor for R-gene-mediated resistance signalling. More importantly, we identified a hydrophobic module in the CC domain for improving disease resistance with NLRs. We verified this strategy on two Solanaceae CNL proteins [pepper Tsw and its paralog in tomato (Solanum lycopersicum)] in another Solanaceae plant, Nicotiana benthamiana. A simultaneous substitution of Tyr 135 or Tyr 137 to tryptophan in the hydrophobic groove of NLRs improved the disease resistance efficacy both in N. benthamiana and pepper. This module of NLRs could be further exploited in disease resistance breeding programs via genome editing in the future.

Plant materials and plasmids
Nicotiana benthamiana plants were grown in growth chambers at 23 °C with a 12 h light (80 μmol m -2 s -1 )/12 h dark cycle. The transgenic N. benthamiana plants harbouring Tsw and Tsw Y137W were generated by transforming them with pMDC32-Tsw-3xFlag and pMDC32-Tsw Y137W -3×Flag constructs, respectively. In detail, TswY137W was obtained by mutating the nucleotides TAT409-411 to TGG409-411 using mutant primers listed in Supplementary Table S1, through PCR. Subsequently, the amplified Tsw and Tsw Y137W gene coding sequences were ligated to pMDC32 empty vector using the KpnI and SpeI restriction sites. A widely used chilli pepper cultivar (Capsicum annuum L. 'GuoFu 208') in Northern China was used for the disease resistance assay. For constructing the expression vectors for the full length or domains of the NLRs, codon-optimized Tsw and tomato Tsw paralog coding sequences were artificially synthesized (GENEWIZ,USA) and cloned into the plant binary vector pGDGm, which contains the GFP tag on the C-terminus of genes (Goodin et al., 2002;Curtis and Grossniklaus, 2003). Site-directed mutants were generated using specific primers carrying the desired mutations, as described previously (Li et al., 2014). The entry vector pENTR-3C in the Gateway system (Invitrogen, A10464, USA) and destination vectors pBA-DC-6Myc, pH7WG2Y under the control of a CaMV 35S promoter were used for 35S:CC-Myc and 35S:YFP-CC constructs. Primers used in this study are listed in Supplementary Table S1.

Virus inoculation
The leaves of TSWV-infected N. benthamiana plants were kindly provided by Xiaorong Tao (Nanjing Agriculture University, China) initially and maintained in our laboratory by mechanical inoculation to N. benthamiana. These were used for the virus infection assay. Briefly, 0.1 g virus (isolate TSWV-YN) infected-leaves were ground to a powder in liquid nitrogen, and then dissolved in 5 ml of 0.05 M phosphate buffer (pH 7.0). Subsequently, 150 μl of virus containing supernatant was mechanically inoculated onto N. benthamiana leaves, as described previously (Wu et al., 2019).

Agroinfiltration, transient expression, and western blot analysis
All expression vectors were transformed into Agrobacterium tumefaciens strain EHA105. Agrobacterium carrying the binary vectors were cultivated overnight and diluted, then infiltrated into the abaxial sides of N. benthamiana leaves. The cell death symptoms were observed at 3 days (CC domain) or 7 days (full length NLR) post-inoculation (dpi) and photographed under UV/white light. To analyse protein expression, treated leaves were harvested and tested by immunoblotting, as described previously . Total proteins were extracted from infiltrated leaves, separated using 10% SDS-PAGE, and transferred to a polyvinylidene difluoride (PVDF) membrane (Merck Millipore, IPVH00010, Germany). Finally, pGDGm-CC series plasmids which encode the GFP epitope-tagged proteins, were detected using an anti-GFP monoclonal antibody with 1:5000 dilution ratio (TransGen Biotech, China).
For virus inoculation, Agrobacterium carrying the Tsw-expressing vectors was infiltrated into N. benthamiana or pepper leaves, followed by mechanical inoculation with TSWV at 2 dpi. Viral abundance in N. benthamiana or pepper plants was detected by RT-PCR and with antibodies against NSs and Ncp, at 7 dpi or 14 dpi, respectively.

Confocal microscopy
A. tumefaciens containing the pGDGm vector with CC and CC mutant were transiently expressed in 6-8-week-old N. benthamiana using agroinfiltration. Protein sub-cellular localization was detected at 2 dpi. For plasmolysis assay, agroinfiltrated N. benthamiana leaves were treated with 5% NaCl for 5-10 min to trigger cell plasmolysis, before observation. Confocal imaging was observed and recorded with a Leica SP8 laser scanning confocal microscope (Leica, Germany).

Membrane fractionation
Plasma membrane partitioning was performed with Minute TM Plasma Membrane Protein Isolation Kit, according to the manufacturer's protocol (Invent, SM-005-P, USA). The anti-H + -ATPase monoclonal antibody (Agrisera, Sweden) was used to detect the plasma membrane marker protein, H + -ATPase.

Co-immunoprecipitation assay (Co-IP)
A. tumefaciens containing the pGDGm-CC and empty vector pGDGm plasmids were co-infiltrated with A. tumefaciens containing the 35S:CC-Myc plasmids into N. benthamiana leaves individually. After 3 d, ~1 g of leaf tissue was collected and ground to a powder in liquid nitrogen, as described previously (Wu et al., 2019). Total proteins were extracted in 1 ml of extraction buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, 10% glycerol, 10 mM DTT, one tablet of protease inhibitor cocktail per100 ml; Sigma-Aldrich, USA). Then the protein extracts were incubated with 25 μl of GFP-trap beads (Lablead, China, 5500041921) for 3 h at 4 °C. The beads were then washed three times with extraction buffer and resuspended in 2×SDS buffer before using for immunoblotting analysis.

Blue native-polyacrylamide gel electrophoresis (BN-PAGE)
BN-PAGE was performed using the NativePAGE™ Novex® Bis-Tris Gel System (Thermo fisher Scientific, USA) according to the manufacturer's protocol. CC-GFP and the empty vector were transiently expressed in N. benthamiana leaves individually using Agrobacteria. Plant samples were collected 36 h post inoculation and proteins were extracted for subsequent native PAGE analysis.

RT-quantitative PCR
Total RNA from host and virus was extracted from N. benthamiana or pepper plant leaves using the Plant Mini Kit (Qiagen, 74904, Germany), and 2 µg of total RNA from each sample was reverse transcribed into cDNA using the TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, AT311-03, China), according to the manufacturer's instructions. RT-qPCR for both host mRNA and virus titre was performed on the CFX 96 system (Bio-Rad) using Thunderbird SYBR qPCR mix (TOYOBO, QPS-201), as described previously (Wu et al., 2019). The primers used for mRNA detection of target genes by real-time PCR are listed in Supplementary Table S1. Data were normalized to EF1α expression by the cycle threshold (CT) 2-ΔΔCT method. All experiments were repeated three times. Values are means ±SD; asterisks in RT-qPCR data panels indicate significant differences (Student's t-test, *P<0.05; **P<0.01).

Virus-induced gene silencing (VIGS)
For tobacco rattle virus (TRV)-based gene silencing, fragments of NbSGT1 (Suppressor of G-two allele of Skp1), NbRAR1 (Required for Mla12 resistance), and NbHSP90 (Heat shock protein 90) were cloned from cDNA of N. benthamiana leaves and constructed into a pTRV2 vector (Qu et al., 2012). Subsequently, pTRV::SGT1, pTRV::RAR1, and pTRV::HSP90 plasmids were transformed into A. tumefaciens EHA105 strain. The indicated A. tumefaciens strains were co-infiltrated with Agrobacterium carrying a pTRV1 plasmid into 3-week-old N. benthamiana seedlings. Co-inoculation of pTRV1 and pTRV2 combination was used as a negative control. pTRV::PDS, the VIGS vector for silencing of N. benthamiana PHYTOENE DESATURASE (PDS) gene was used as positive control (Ye et al., 2009;Qu et al., 2012). Two weeks post-treatment, the gene knockdown efficiency of silenced plants was measured by RT-qPCR. At least six silenced plants per treatment were selected to inoculate with Agrobacterium carrying 35S:CC-GFP plasmid. Later, cell death symptoms were observed and photographed under UV/white light. The experiment was repeated at least twice with similar results.

Ion leakage assay
The ion leakage of treated leaf samples was measured according to a previous report with some modifications (Zhu et al., 2017). Briefly, five leaf discs with a diameter of 5 mm were collected and floated on 5 ml of double-distilled water at 25 °C for 3 h, then the sample conductivity was measured and recorded as value A. Isolated leaf discs were treated at 95 °C for 25 min, and the value B of the corresponding samples were further measured when the solution cooled to 25 °C. The final conductivity was expressed as a percentage of the ion leakage=(value A/value B)×100.

Disease index
Plant disease index is a standard method to quantify the disease incidence and severity for a population of plants (Cardoso et al., 2010;H. Wang et al., 2015). We categorized the severities of TSWV-caused disease on N. benthamiana plants into six grades: 0, 1, 2, 3, 4, and 5. The grade 0 point indicates no disease symptom on the plant and grade 5 indicates dying plants after infection. Grade 1 to 4 is rated by disease symptom severity ( Supplementary Fig. S1). The corresponding disease index was calculated as follows: Disease index=∑ (number of plants in each grade × each grade score)/ (total number of plants investigated × the maximum grade score) × 100.

Data analysis
Differences in gene expression were determined by Student's t-tests. A significant level of 0.05 was used for statistical analysis. One-way analysis of variance (ANOVA) followed by Duncan's multiple range test were also used for data analysis. All statistical tests were carried out with GraphPad Prism. The relative fluorescence intensity of the cell death was estimated by ImageJ. In detail, the fluorescence channel of cell death images was split to measure the means of grey value, which indicated the fluorescence intensity (FI). Then the value was normalized to relative protein expression level (PE) observed in western blots. PE was estimated by ImageJ. The final calculating formula is as follows: The relative fluorescence intensity=[FI (sample) -FI (background)]/ [PE (sample)/ PE (loading)].

The coiled-coil domain of Tsw triggers cell death
Testing the potential for cell death-mediated disease resistance is the easiest method widely applied in plant NLR functional studies. To fully understand the mechanism of pepper NLR protein Tsw-triggered plant tospovirus defence, we first examined if Tsw possesses a single signalling domain to induce cell death and pathogen resistance. Tsw is a 2116 aa NLR protein initially identified from Capsicum chinense accessions (Kim et al., 2017). Through sequence alignment with well-characterized NLRs, three domains were identified from Tsw protein, a coiled-coil (CC) domain (1-169 aa), a central nucleotidebinding site (NBS) domain (170-491 aa) and a COOH-terminal leucine-rich repeat (LRR) domain (492-2116 aa; Fig.  1A). Accordingly, we tested the capacity of each domain to induce visible cell death in the presence and absence of the viral effector NSs when transiently expressed in the heterologous host Nicotiana benthamiana. As seen in Fig. 1B, the CC domain (1-169 aa) was sufficient to trigger cell death in N. benthamiana leaves to degrees comparable to those seen with full-length Tsw in the presence of the viral effector NSs. A western blot analysis confirmed the correct expression of GFP fused with each domain of Tsw (Fig. 1C). Interestingly, the CC domain also has the activity to trigger cell death alone in the absence of NSs ( Supplementary Fig. S2). The co-expression of either NBS or LRR domains of Tsw strongly abolished the deathinducing activity of the Tsw CC domain ( Supplementary Fig.  S3A), which could be explained by a direct inter-domain interaction ( Supplementary Fig. S3B). These results indicated that the CC domain is the main signalling executor of Tsw to induce cell death.
The length of the CC domain and N/C-terminal tags may affect CC-triggered cell death. A series of deletion analysis further defined the minimal cell death-inducing fragment of Tsw as 1-137 aa (CC 1-137 ). This fragment was even more potent to induce cell death than that of full-length Tsw. Importantly, the cell death induction ability by CC 1-137 was fully abolished when Y 137 was deleted within the hydrophobic rich motif F135-C136-Y137 (Fig. 1D). There was no obvious difference in protein accumulation among various mutants (Fig.  1E). We noticed that certain residues located in the hydrophobic groove of other NLRs also play key roles in their cell death activity, like Ile136 in ZAR1 (Cesari et al., 2016;Wang et al., 2019a). Based on the different performance of CC 1-137 (autoactive) and CC 1-136 (inactive) in cell death assays (Fig.  1D), we introduced several point mutations at Tyr137 within the CC domain (1-169 aa fragment) to test their efficacy in inducing cell death. Intriguingly, substitution of Tyr137 with a non-polar tryptophan residue (Y137W) strongly enhanced the capacity of the CC domain to induce cell death without changing CC protein accumulation ( Fig. 1F-H), as indicated by the enhanced electrical conductivity of ions in the solution, reflecting that the degree of membrane damage was intensified (Fig. 1G). These results support a critical role of the hydrophobic groove, especially the Tyr 137 residue, in inducing cell death for the Tsw protein.

Coiled-coiled domain localized on the plasma membrane
The sub-cellular localization of NLR proteins are pivotal for their proper functioning and full activity (Qi et al., 2012;Kawano et al., 2014;Cesari et al., 2016;Wang et al., 2019a). We next set out to characterize the sub-cellular localization of Tsw and its derivatives. We expressed the CC-GFP fusion protein in leaf cells of N. benthamiana. Confocal microscopy results showed that the fluorescence signal of CC-GFP fusion protein was fully co-localized with the plasma membrane marker CD3-1007-RFP (Nelson et al., 2007), while GFP alone had a partial fluorescence signal in the nucleus ( Fig. 2A). Meanwhile, plasmolysis analysis further suggested the plasma membrane localization of CC-GFP (Fig. 2B). The fluorescence signals in the cytoplasm and plasma membrane of N. benthamiana cells were indistinguishable. To confirm this membrane association of Tsw CC, we performed a cellular compartment fraction assay on the GFP and CC-GFP expressing leaf cells of N. benthamiana. As expected, CC-GFP was exclusively detected in the plasma membrane fraction, and GFP control was mostly detected in soluble fractions, which further confirmed the plasma membrane localization of the CC domain (Fig. 2C).
N/C-terminal tags may affect CC-mediated cell death. We observed that the N-terminal amino acid sequence is essential for Tsw CC function, and the N-terminal GFP fusion, but not the C-terminal fusion affects cell death-induction ability of Tsw CC ( Supplementary Fig. S4). We further verified whether the N-terminal region of the CC domain is involved in its activity by constructing several truncations of the CC domain, according to its predicted protein secondary structure. Through transient expression in N. benthamiana leaves, we found that its plasma membrane anchoring activity is greatly dampened, proving that the N-terminal region of CC is impaired. The sub-cellular locations of CC 30-169 -GFP and CC 41-169 -GFP were partially changed to the nucleus (Fig. 2D). The same results were observed in a membrane fractionation assay, and the above-described N-terminal deletion mutants of CC were partially soluble (Fig. 2E). Accordingly, these N-terminal truncated CC mutants abolished cell death induction in N. benthamiana (Fig. 2D). Interestingly, an N-terminal fragment (1-25 aa) of CC could partially re-localize GFP to the plasma membrane, as indicated by the membrane fractionation assay results of GFP and CC 1-25 -GFP proteins (Supplementary Fig.  S5). These results suggested that the N-terminal region of the CC domain is essential for its association with the plasma membrane, and the ability to trigger cell death.

A hydrophobic module of Tsw is associated with its cell death induction
To decipher the underlying mechanism of this single signalling CC domain, we next analysed the oligomerization state of the CC domain in vivo using protein extracts from infiltrated N. benthamiana leaves. The CC domain could form oligomers, as evidenced by co-immunoprecipitation (co-IP) and BN-PAGE (Fig. 3A, B). We then analysed the structure of the Tsw CC do-main by protein homology modelling using the PHYRE web server. Homology modelling results showed that the structure of the Tsw CC domain aligned well with that of the ZAR1 CC domain (Fig. 3C; Supplementary Fig. S6). The above results showed that hydrophobic amino acid Tyr137 is critical for its full function in cell death induction, and the CC Y137W mutant has enhanced cell death activity (Fig. 1F). We observed that the hydrophobic module is near the central pore with a putative ion permeation path, and therefore we next queried the relationship between the structure formed by hydrophobic amino acids and Tsw CC-induced cell death. We analysed the modelled oligomeric structure of Tsw, and found that the residues around Trp137 in the Tsw CC Y137W domain form a more compact hydrophobic groove than that in wild type Tsw (Fig. 3D). Considering the increased cell death caused by the CC Y137W domain in planta (Fig. 1F), we assumed that the change from Tyr to Trp causes a superhydrophobic funnel-like structure that further stabilizes its oligomerization. To further verify this hypothesis, we analysed other residues within the hydrophobic region of the CC domain (132-137 aa). Notably, the oligomerized Tsw CC structure indicated that the mutation of Val 134 to Trp 134 confers the 134 th residue to have extensive contact with its neighbouring residues (Fig. 3D, second panel). In contrast, substitutions with other residues within the hydrophobic region (e.g. Phe135) lead this residue to face outside of the funnel-like structure (Fig. 3D, third panel). The follow-up transient expression assay and ion leakage data showed that CC V134W indeed enhanced cell death, though a similar protein level accumulated with the WT CC control, consistent with the predicted Tsw oligomer structure (Fig. 3E-G). These results suggested that the strategy of structure-assisted improvements on the CC domain for NLR proteins against pathogens is feasible. Moreover, the amino acids in the hydrophobic groove affect the activity of the CC domain irrelevant of its localization, since the loss-of-function mutants of CC still localize on the plasma membrane ( Supplementary Fig. S7).

Tyrosine to tryptophan substitution in the hydrophobic groove of Tsw enhanced its resistance to TSWV
To test whether the hydrophobic residue substitution was functional in the full-length Tsw with enhanced immunity against TSWV, we generated a construct harbouring full-length Tsw Y137W -GFP and transiently expressed this Coomassie brilliant blue (CBB) stained bands of the large subunit of rubisco (rbcL) were used as a loading control. The CC domain and its mutants bands were quantified by ImageJ software, the wide type CC was normalized to 1.00. (F) Cell death phenotypes (at 2 dpi) of CC Y137 point mutants (as GFP fusions) transiently infiltrated in N. benthamiana leaves. The numbers in parentheses represent the relative fluorescence intensity normalized to the relative protein expression level. The relative fluorescence intensity of cell death was estimated by ImageJ. Scale bars=1 cm. (G) Quantification of CC Y137 point mutant-induced cell death in N. benthamiana leaves, via ion leakage assay. Electrolyte leakage of leaf disks were measured at 24 h and 48 h after agroinfiltration (n=4; lower case letters indicate significant differences between samples; P<0.05, one-way ANOVA followed by Duncan's multiple range test). (H) Accumulation of proteins detected by immunoblotting with α-GFP antibody. The CC and CC Y137 point mutant bands were quantified by ImageJ software; the wide type CC was normalized to 1.00 (relative values are indicated). Stained bands of rbcL were used as a loading control. In (B), (D), and (F), the agroinfiltrated leaves were illuminated under normal light (white) or a long-wavelength UV lamp.   gain-of-function mutant or WT Tsw control in N. benthamiana leaves, followed by mechanical inoculation with TSWV-YN at 2 d post-infiltration. The disease incidence of plants infected with TSWV was investigated, and the disease index recorded, to describe the average disease severity on the TSWV-infected plants ( Supplementary Fig. S1). Plants expressing Tsw Y137W -GFP exhibited greater resistance to TSWV infection than those expressing Tsw-GFP, as indicated by their much weaker disease symptoms and the dramatically reduced disease index (Fig. 4A, B). Consistent with results of cell death induction, the expression of Tsw Y137W -GFP confers plants with much stronger disease resistance, as indicated by much lower levels of viral RNA and proteins in newly emerged leaves than those expressing Tsw-GFP (Fig. 4C, D). To confirm this enhanced virus resistance observed in the model plant N. benthamiana, we transiently expressed WT Tsw and the gain-of-function mutant Tsw Y137W in a pepper cultivar sensitive to TSWV infection. Consistently enhanced resistance to TSWV infection was also observed on pepper plants expressing Tsw Y137W than those expressing Tsw (Fig. 4E, F). These data confirmed that Tsw Y137W confers improved resistance to TSWV infection in different plant species.
Moreover, to further test the enhanced resistance of Tsw Y137W to TSWV, we generated several transgenic N. benthamiana plants stably transformed with Tsw-Flag and Tsw Y137W -Flag constructs. Dozens of T 0 transgenic N. benthamiana plants were generated either expressing Tsw-Flag or Tsw Y137W -Flag proteins. Molecular characterization including gene expression and protein accumulation of Tsw-Flag and Tsw Y137W -Flag was performed to identify transgenic plants with similar expression level of Tsw-Flag (lines #1 and #14) and Tsw Y137W -Flag (lines #1 and #10; Supplementary Fig. S8). Each transgenic line of Tsw-Flag and Tsw Y137W -Flag was further tested to ascertain that T 1 progeny transgenic plants were resistant to the TSWV-YN strain. Milder symptoms were observed in T 1 plants ectopically expressing 35S:Tsw-Flag, and no obvious disease symptoms were seen on 35S:Tsw Y137W -Flag plants Fig. 4G; in contrast, typical leaf curling caused by TSWV infection was observed in systemic leaves of mock plants at 7 dpi. Notably, 35S:Tsw Y137W -Flag transformants exhibited enhanced resistance compared with 35S:Tsw-Flag transformants, even though Tsw showed comparable expression levels ( Fig. 4H; Supplementary Fig. S8). The following RT-qPCR and western blot analysis confirmed the lower virus titre and accumulation of viral Ncp and NSs proteins in infected 35S:Tsw Y137W transformants, than in 35S:Tsw transformants (Fig. 4I, J). These results demonstrated that the hydrophobic groove of Tsw can be engineered to improve the pepper NLR Tsw when expressed in a heterologous plant like N. benthamiana.

The hydrophobic groove conserved in an autonomous NLR clade in seed plants
Tsw was recently classified as a member of the well-conserved autonomous NLR (ANL) clade, which is present in all tested seed plants (Lee et al., 2021). To determine how common the strategy of hydrophobic groove-mediated immunity enhancement might be, we aligned the protein sequences of CC domains from members of the ANL clade from taxonomically distant plants, including Solanaceae, Rosaceae, and Chenopodiaceae. Indeed, the hydrophobic groove showed a highly conserved pattern, with a conserved aromatic amino acid (Tyr or Phe) at the position corresponding to Tsw Tyr137, while the Val134 exhibited a slightly changeable pattern (Fig. 5A). We randomly selected the Solanum lycopersicum ANL member SlANL-1 to verify our above result. Accordingly, we generated one loss-offunction substitution (CC Y135E -GFP) and one gain-of-function substitution (CC Y135W -GFP), and transiently infiltrated the resulting constructs in N. benthamiana leaves for functional verification. The loss-of-function substitution (CC Y135E -GFP) failed to induce cell death, while the gain-of-function substitution (CC Y135W -GFP) exhibited a more severe cell death phenotype than the control CC Y135 -GFP, accompanied by higher ion conductivity (Fig. 5B, C). A western blot analysis confirmed the correct expression of GFP-fused CC proteins of SlANL-1 (Fig. 5D). These results indicated that NLRs from ANL members could also benefit from hydrophobic groove improvement, regardless of their natural host.
photographed 3 weeks post-inoculation. White arrows point to typical symptoms; red arrows indicate inoculated leaves. Scale bars=1 cm. (C) Relative TSWV titre in N. benthamiana at 7 dpi, as determined by RT-qPCR of Ncp expression. Values are means ±SD; n=6, *P<0.05, **P<0.01, by Student's t-test. (D) TSWV abundance in plants at 7 dpi, as detected with antibodies against NSs or Ncp. Viral NSs or Ncp bands were quantified by ImageJ software, the viral accumulation in EV transiently expressed plants was normalized to 1.00. (E) The empty vector (EV), Tsw-Flag, or Tsw Y137W -Flag were transiently infiltrated into pepper leaves, followed by inoculation with TSWV-YN on infiltrated leaves 2 d later. Systemic symptoms of TSWV-inoculated pepper were photographed 14 dpi. Scale bars=1 cm. (F) TSWV abundance in pepper at 14 dpi, viral NSs or Ncp were detected with antibodies against NSs or Ncp, and were quantified by ImageJ software and the values obtained are indicated; the viral accumulation in EV transiently expressed plants was normalized to 1.00. (G) Tsw-Flag and Tsw Y137W -Flag transgenic lines were inoculated with TSWV. Systemic symptoms of TSWV-inoculated N. benthamiana were photographed 7 dpi. White arrows point to typical symptoms. Scale bars=1 cm. (H) Disease index of TSWV-infected N. benthamiana plants was investigated within 10 dpi, and the individual disease index was calculated (n=4; lower case letters indicate significant difference between samples; P<0.05, ANOVA followed by Duncan's multiple range test). (I) Relative TSWV titre in N. benthamiana at 7 dpi was determined by RT-qPCR of Ncp expression. Values are means ±SD, n=6; P<0.05, ANOVA followed by Duncan's multiple range test. (J) TSWV abundance in transgenic plants at 7 dpi, antibodies against NSs and Ncp were used. Viral NSs or Ncp bands were quantified by ImageJ software and the values obtained are indicated; the viral accumulation in non-transgenic plants was normalized to 1.00.

Host SGT1-dependent cell death induction for Tsw coiled-coil domain
In plants, molecular chaperone components SGT1, RAR1, and HSP90 are essential for plant resistance mediated by a number of NLR proteins (Botër et al., 2007). To gain insight into the cell death signalling mechanism of Tsw, we tested whether SGT1, RAR1, and HSP90 participated in Tsw CC-induced virus resistance. We conducted cell death assays on these plant cells with gene knock down expression by TRV-based VIGS.
The RT-qPCR results demonstrated that NbSGT1, NbRAR1, as well as NbHSP90 transcripts were dramatically decreased, compared with those of the vector control plants (Fig. 6A). The Tsw CC-induced cell death was dramatically compromised in SGT1-silenced plants (Fig. 6B). This conclusion was further supported by results from the ion leakage assay at 48 h (Fig. 6C). Moreover, we also measured the accumulation of Tsw CC domain in HSP90, SGT1, and RAR1-silenced plants (Bhattarai et al., 2007;Zhang et al., 2010); western blot analysis confirmed the correct expression of GFP-fused CC proteins (Fig. 6D). Sub-cellular compartment distribution of CC-GFP protein showed no obvious differences between WT and these genetically defective plants ( Supplementary Fig. S9), suggesting that HSP90, SGT1, and RAR1 are not necessary for the steady state levels and sub-cellular localization of the Tsw CC domain.

Discussion
Once activated, NLRs mediate disease resistance and induce cell death simultaneously. However, the precise connection between disease resistance and cell death is largely unknown. Here, we discovered that the hydrophobic groove in CC and its two variants (CC Y135W -GFP and CC Y135E -GFP) in N. benthamiana leaves. The agroinfiltrated leaves were illuminated under normal (white) light and a long-wavelength UV lamp at 2 days post-inoculation. The fluorescence intensity of cell death was estimated by ImageJ. Scale bars=1 cm. (C) Ion leakage assay of leaf disks from SIANL-1 CC mutants expressed in N. benthamiana was measured at 24 h and 48 h after agroinfiltration (n=4; lower case letters indicate significant difference between samples; P<0.05, one-way ANOVA followed by Duncan's multiple range test). (D) Protein accumulation was measured by immunoblotting with GFP antibody. The CC mutant bands were quantified by ImageJ software and the obtained values are shown; the intensity of CC Y135 band was normalized to 1.00. Coomassie blue stained the large subunit of rubisco (rbcL) that was used as a loading control.
domain can be explored to enhance plant NLR-mediated resistance. These results provide new targets to optimize plant immunity mediated by Tsw and other ANL proteins.

Coiled-coil domain individually induces cell death in planta
Each plant genome encodes hundreds of NLR proteins. Most NLRs possess an N-terminal signalling domain that induces visible cell death in heterologous plants, including TIR (Tollinterleukin-1 receptor) domains from RBA1 (Response to the bacterial type III effector protein HopBA1), RPS4 (Resistance to Pseudomonas syringae 4), L6 (flax polymorphic L locus protein 6), and CC domains from ZAR1, MLA10 (polymorphic barley mildew A 10) , Sr33 (Stem rust resistance gene 33), Sr50 (Stem rust resistance gene 50), NRG1 (N requirement gene 1), RPM1 (Resistance to Pseudomonas maculicola 1), and RP1 (Resistance to P. sorgbi 1; Qi et al., 2012;G.F. Wang et al., 2015;Casey et al., 2016;Nishimura et al., 2017;Zhang et al., 2017;Horsefield et al., 2019;Baudin et al., 2020;Bai et al., 2012). Recently, the N-terminal amphipathic helix in the CC domain of ZAR1 was shown to form a funnel-shaped structure that is required for plasma membrane-associated cell death (Wang et al., 2019a, b). A few functionally important motifs of the CC domain have been identified from CNLs, such as MADA and α1 helix (Adachi et al., 2019;Lee et al., 2021). Tsw is a singleton and autonomous NLR, carrying an autoactive CC domain in pepper (Fig. 1B), with a minimal unit to induce cell death in planta (Fig. 1D). Our data strongly support the concept that Tsw CC domain has a plasma membrane-anchoring ability, consistent with the predicted structural model of Tsw ( Fig.  2A-C). The importance of plasma membrane localization has been addressed in several CNLs (Zhou et al., 2015;   . Electrolyte leakage of leaf disks from N. benthamiana were measured at 24 h and 48 h after agroinfiltration (n=4; lower case letters indicate significant difference between samples; P<0.05, one-way ANOVA followed by Duncan's multiple range test). (D) Protein accumulation of CC-GFP determined by western blot. Total soluble protein was detected by monoclonal antibody against GFP. The CC bands were quantified by ImageJ software and the intensity of the bands are indicated; the CC accumulation in pTRV::00 infected plants was normalized to 1.00. Coomassie-stained rbcL was used as a loading control. El Kasmi et al., 2017;Wang et al., 2020). Data presented here also strongly suggested that the plasma membrane of Tsw has significant roles in immunity against TSWV (Fig. 2D, E), such as the Ca 2+ influx induction of the CC domain, as evidenced by the Tsw CC domain-mediated cell death being dependent on Ca 2+ (Supplementary Fig. S10).

Hydrophobic module regulates coiled-coil-mediated cell death
The hydrophobic region of the CC domain affects the formation of ZAR1 resistosome, and thereby NLR-induced immunity (Wang et al., 2019a). The Phe 135 , Ile 136 , and Thr 137 residues are centred at the ZAR1 oligomerization interface, which make extensive hydrophobic contacts with their neighbouring residues, and the ZAR1 I136E mutant compromised its cell death induction activity. Consistently, the corresponding Val 134 , Phe 135 , Cys 136 , and Tyr 137 residues of Tsw CC play an important role in cell death, and a mutant of Tyr 137 can potentially increase the Tsw-mediated virus resistance to TSWV in N. benthamiana (Fig. 4). We confirmed the self-association ability of Tsw CC (Fig. 3A, B). Unlike the Ile 136 of ZAR1, there are no amino acid residues near Tyr 137 that provide hydrogen bonds. Interestingly, Tyr 137 in every monomer of Tsw is close to each other and could form strong hydrophobic interactions to stabilize the funnel-shaped structure. The substitution of Tyr 137 to Glu 137 lost a hydrophobic environment and resulted in a loss-of-function in triggering cell death (Fig. 1F). As visible from the predicted structure of Tsw, hydrophobic interaction is enhanced by the substitution of Tyr 137 to Trp 137 mutant, leading to a more stable funnel-shaped structure (Fig.  3D). According to this principle, we substituted Val 134 (hydrophobic force enhanced) and Phe 135 (hydrophobic force weakened) with Trp. The cell death phenotypes further proved the concept that the hydrophobic groove of NLRs is a good target for improvement, and the structure-assisted method could be very useful (Fig. 3D, E).
The hydrophobic region (122-134 aa) of the Sr33 cluster is important for self-association and auto-activity of the CC domain (Cesari et al., 2016). There are some similarities among these different NLRs. The region (134-137 aa) of Tsw CC domain is also enriched in hydrophobic residues and important for its auto-activity ( Supplementary Fig. S7). Tsw was classified in the ANLs cluster, and the R protein in this novel cluster shared several features, such as carrying an autoactive CC domain and localizing to the plasma membrane (Lee et al., 2021). Consistently, the gain-of-function mutant for R protein improvement was further validated in the tomato Tsw paralogous R protein (Fig. 5B); Tsw-mediated resistance to TSWV was specific, as Tsw transgenic plants showed no obvious effect when infected by turnip mosaic virus (TuMV), another RNA virus Supplementary Fig. S11). It would be valuable to investigate whether the key function of amino acids in the hydrophobic groove is also a common feature of NLRs in this cluster.
As one of the two resistance genes against TSWV, Tsw is not very potent because of its narrow disease resistance spectrum, and it could be overcome with temperature and rapidly evolving pathogens (Turina et al., 2016). We observed that transgenic expression of Tsw in N. benthamiana and transient expression in pepper (C. annuum) could only induce moderate resistance. Several reasons may explain this observation. One reason could be the low expression level in both test systems. Several reports have shown that R gene overexpression results in enhanced disease resistance, and positively correlates with their expression levels (Oldroyd and Staskawicz, 1998;Stokes et al., 2002;Tang et al., 1999). Moreover, some newly emerging TSWV isolates were able to break the Tsw resistance gene-mediated plant defence, including a tomato isolate (TSWV-YN18) and a tobacco isolate (TSWV-YN53; Jiang et al., 2017). TSWV NSs have been found to be the key determinant for breaking resistance in Tsw-carrying pepper plants (Margaria et al., 2007). The amino acid sequence of NSs in our study (TSWV-YN) has a very high sequence identity with that of TSWV-YN18 and TSWV-YN53 (99.14% and 99.36%, respectively). The TSWV-YN isolate used in our study might be a resistance-breaking (RB) strain. Nevertheless, at a similar low expression level, the Tsw Y137W mutant enhanced resistance to TSWV in tobacco and pepper plants under the same Tsw construct background, which suggests that it is a feasible way of structure-assisted improvements of NLR against pathogens.
Artificial evolution of R genes in crops will provide more valuable resources for breeding. Whether the module shown in the current study is workable in fields needs further research on TSWV-sensitive gene editing of crops. Thus, it is necessary to carry out base-specific gene editing to improve resistance in pepper varieties that contain the Tsw gene. To develop transgenic plants in pepper is a challenging task, compared with that in other Solanaceae species, because of the lack of an effective transformation system for pepper and its low regeneration rate under in vitro conditions (Kothari et al., 2010;Turina et al., 2016). This limitation needs to be overcome in future research.

Molecular regulation of Tsw-mediated cell death
Precise artificial R gene evolution requires a detailed understanding of activation of NLRs. Plant NLRs recognize pathogens both indirectly and directly to trigger immunogenic signals (Zhou and Zhang, 2020). This is the first molecular analysis of Tsw in TSWV immunity. Tsw is classified as a singleton NLR which is supposed to function as a single unit to confer full resistance (Lee et al., 2021). Interestingly, we found that both the NBS and LRR (NBS-GFP and LRR-GFP) domains in Tsw could attenuate autoactivation of the CC domain. We confirmed the interaction between the NBS and CC domains in a co-IP assay ( Supplementary Fig. S3B). These results indicated that Tsw could self-regulate, and its activation may be due to CC domain exposure, with the NBS domain inhibiting its activation.
In addition, epigenetic regulation, temperature, and phytohormones are also reported to associate with NLR-mediated plant immunity (Chung et al., 2018;Li et al., 2020;Yang et al., 2020). Amongst these, as one of the important environmental factors, temperature has a modulating effect on many biological processes, especially to antiviral immunity (Qu et al., 2005;Wang et al., 2009). One of the drawbacks of some R genes is temperature sensitivity; they only provide resistance at a certain temperature range (Wang et al., 2009;Lim et al., 2014). Tsw-mediated resistance is reduced at 28 °C, and does not provide resistance at a temperature higher than 32 °C (de Ronde et al., 2019). Mutations in the R gene sequence against tobacco mosaic virus provide improved resistance at high temperature (Whitham et al., 1996), although the underlying mechanism is still elusive. Whether the enhanced TSWV resistance conferred by Tsw mutants (Tsw Y135W and Tsw Y137W ) could perform better under higher environmental temperature conditions still needs to be tested in the future.

Effect of SGT1 on coiled-coil/Tsw function
The precise function of SGT1 is still a mystery as it is involved in several unrelated processes. Our data presented here provide some clues that Tsw conducts the resistance response via SGT1 (Fig. 6). Molecular chaperone components SGT1, RAR1, as well as HSP90 form a complex to contribute to lots of R genemediated resistance (Bhattarai et al., 2007;Botër et al., 2007;Shirasu, 2009;Zhang et al., 2010). Several studies indicated that the steady state level of R protein depends on the chaperone activity of SGT1 and HSP90 (Takahashi et al., 2003;Azevedo et al., 2006); the chaperone function of SGT1 is required for R gene function in several plant-microbe interactions (Wang et al., 2010). For example, SGT1 is reported to play a positive role in R gene-induced cell death. It is required by Sw-5b-mediated TSWV resistance, and Mi-1-mediated pathogen and pest resistance (Bhattarai et al., 2007;Kadota et al., 2010;Chen et al., 2021). The interaction between SGT1 and other chaperones also contribute to the accumulation of NLR proteins and defence signalling (Kadota et al., 2008). In this study, our results showed that CC protein accumulation and subcellular localization were unaffected in SGT1-silenced plants ( Fig. 6D; Supplementary Fig. S9). Therefore, we hypothesize that SGT1 may not directly influence the function of Tsw-CC. Instead, SGT1 may act downstream of CC-mediated defence signal transduction. More research is required to underpin the underlying mechanisms.
Since it serves as a critical immune regulator, it is not surprising that SGT1 is often targeted by pathogens of plants and animals (Bhavsar et al., 2013;Yu et al., 2020). Despite that SGT1 plays a positive role in host and non-host resistance in N. benthamiana, the infection of some RNA viruses such as TSWV and potato virus X has been shown to be impaired in SGT1-silenced N. benthamiana, which is explained by the blocked function of viral movement protein Qian et al., 2018).
In conclusion, we have developed more robust R proteins in a structure-assisted manner by optimizing the hydrophobic groove of ANL-type NLRs. Function-and structure-based reprogramming of R genes using biotechnological applications, such as genome editing provide valuable resources for breeding disease-resistant crops, especially those lacking robust and defined genetic resistance. Further genome editing on Tsw-harbouring pepper species, and overexpressing the Tsw Y137W mutant in the genomes of non-pepper crops, will expand the feasibility and usefulness of this strategy.

Supplementary data
The following supplementary data are available at JXB online. Fig. S1. Symptoms scale of TSWV-infected N. benthamiana. Fig. S2. Cell death phenotypes of codon-optimized synthetic Tsw (Tsw-GFP) or domain truncations (CC-GFP, NBS-GFP, and LRR-GFP) expressed in N. benthamiana leaves. Fig. S3. Both NBS and LRR domains of Tsw attenuate the autoactivation of the CC domain. Fig. S4. N-terminal free of fusion is critical for cell death induction of Tsw. Fig. S5. CC 1-25 fragment partially re-localizes cytoplasmic GFP protein onto the plasma membrane. Fig. S6. Structure-based sequence alignment of Tsw CC and ZAR1 CC. Fig. S7. The hydrophobic groove in Tsw CC domain is critical for its autoactivation. Fig. S8. Detection of Tsw and TSW Y137W transgenic N. benthamiana plants. Fig. S9. Tsw CC domain is localized to the plasma membrane in SGT1-, RAR1-, and HSP90-silenced N. benthamiana plants.
Fig. S10. Cell death induced by Tsw CC is Ca 2+ -dependent. Fig. S11. Effect of Tsw transgenic plants on TuMV-GFP infection. Table S1. DNA primers used in this study.

Conflict of interest
JY and XW are inventors on a patent related to this work, filed by the Chinese Academy of Sciences (no. 202111079090.4, filed September 15, 2021).

Funding
The study was supported by the National key research and development program, China (2019YFC1200503), CAS Projects for Young Scientists in Basic Research (YSBR-080), National Natural Science Foundation of China (32125032 and 31830073), and the Second Tibetan Plateau Scientific Expedition and Research Program (STEP) (2021QZKK0100).

Data availability
All data supporting the findings of this study are available within the paper and within its supplementary data published online.