The Potato Nucleotide-binding Leucine-rich Repeat (NLR) Immune Receptor Rx1 Is a Pathogen-dependent DNA-deforming Protein

Background: Direct targets for plant NLR proteins in immune signaling are largely unknown. Results: The Rx1 NLR protein of potato binds and distorts DNA following pathogen perception, resulting in immune activation. Conclusion: DNA is a direct signaling target for a plant NLR immune receptor. Significance: Plant NLR receptors might regulate immune transcriptional responses by directly interacting with plant chromatin.

Recently, the NB subdomain of rice Os2g_25900 and NB-ARC domains of maize pollen-signaling protein (PSiP) and Arabidopsis Rpm1 were demonstrated to possess a nucleotide phosphatase activity compatible with the switch model (14).
Activation of animal NLRs typically triggers NB domain-mediated self-association, resulting in the formation of a cytoplasmic signaling scaffold on which partners are activated due to their induced proximity (15). For plant NLRs, such partners have not been identified, and a pivotal yet unanswered question concerns the nature of the downstream signaling component(s) and how these are activated by NLR proteins in their "on" state. The identity of the specific NLR subdomain that transduces a signal to such a downstream component is also unresolved. Whereas for Rx1, the NB subdomain of Rx1 induces cell death, the N-terminal TIR domains of L6 and RPS4 or the coiled-coil domain of MLA10 suffices to trigger cell death, suggesting that the signaling domain might vary for different NLRs or that they act as heterodimers (11, 16 -18). The location of the NLR signaling event is also the subject of increased scrutiny. Several NLR proteins, including N, Mla10, and Rx1 have a dynamic nuclear-cytoplasmic distribution, whereas RRS1-R is restricted to the nucleus, dependent upon the presence of the PopP2 immune elicitor (19 -23). Genetic screens for compromised NLR-mediated resistance identified genes encoding components of the nuclear pore complex (24), indicating involvement of nuclear transport in immune signaling. More direct proof for nuclear activity is the observed nuclear localization for barley MLA1 and MLA10, Arabidopsis RPS4 and SNC1, and the tobacco N protein (22,(25)(26)(27). Redirection of nucleus resident MLA10, N, RPS4, and SNC1 to the cytoplasm compromises their ability to activate immune signaling, suggesting a nuclear signaling target (19,22,26,28). The potato Rx1 protein, which confers PVX resistance, localizes to both cytoplasm and nucleus (23). The Rx1 N terminus interacts with a member of the RanGAP2 family that controls nuclear-cytoplasmic trafficking through the nuclear pore (29). Together, these studies indicate that nuclear-cytoplasmic trafficking and compartmentalization are essential for NLR protein function and suggest distinct activities in different cellular compartments. Recent studies on Arabidopsis RPS4 and barley Mla10 (25,30) have shown that induction of cell death is associated with cytoplasmic localization, whereas nuclear localization of RPS4 is associated with local resistance responses. The presence of a WRKY DNA-binding domain in RRS1-R (21) and the association of Mla10 with both Myb and WRKY transcription factors (31) have led to the hypothesis that plant NLRs regulate transcription in the immune response (32). This notion is further supported by interactions between an SPL transcription factor and the tobacco N NLR protein, the interaction between the SNC1 NLR protein of Arabidopsis and the TPR1 transcriptional corepressor, and the presence of BED DNA-binding domains in many plant NLRs (27,(33)(34)(35).
Based on these observations, signaling from plant NLRs can be viewed from two perspectives that are not necessarily mutually exclusive. In the first perspective, activated NLRs may act as platforms from which signaling proteins promoting immune responses are permitted to function. Alternatively, NLRs may themselves have an additional biochemical activity, indepen-dent of their ATPase activity, required for direct activation of plant immunity. In support of the latter model, we here demonstrate that the Rx1 NLR protein of potato is able to bind DNA in vitro and in situ and that its in vitro activity consists of bending and melting DNA. We further demonstrate that the interaction of Rx1 with DNA as observed in situ only occurs after its genuine activation by the coat protein of PVX virus.

Experimental Procedures
Structural Modeling-Protein fold searches using the Phyre 2 protein homology/analogy recognition engine version 2.0 (36) were undertaken using amino acids 143-488 of Rx1, using both normal and intensive modeling modes. Similar structural homology was also detected using the SAM-T08, Hidden Markov Model-based protein structure prediction server (37). All superpositions were performed using the SSM algorithm in Coot (38). Models of Rx1 based on Cdc6/Orc1 (PDB accession number 2V1U) were made using Chainsaw within the CCP4 package (39), and sequence alignments were generated by the Phyre 2 server. Side chain packing and energy minimization was performed using GalaxyRefine (40). Figures were generated using the PyMOL molecular graphics system (41).
Protein Expression and Purification-Protein corresponding to the NB-ARC domain of PSiP (amino acids 178 -505; PSiP-NBARC) was generated as described previously (14).
Circular Dichroism-80 M protein was dialyzed into double-distilled H 2 O at 4°C. The baseline CD spectra of blank sample (double-distilled H 2 O) and 1.7 M protein were measured using a J-810 spectropolarimeter (Jasco) at 180 -300 nm (20 nm min Ϫ1 ). The averaged data of replicate blank spectra were subtracted from the protein spectra, and the data were normalized to zero at 250 nm. The corrected CD spectra from 190 -240 nm were analyzed using CDPro (46). The protein database generating the lowest root mean square deviation was used as the best approximation for secondary structure content.
Electrophoretic Mobility Shift Assays-The oligonucleotides used for quantitative EMSA are derived from a series of oligonucleotides that enables a comparison of relative DNA binding affinity to varying DNA topologies independent of DNA sequence (47). The oligonucleotides sequences were 5Ј-TGG GTC AAC GTG GGC AAA GAT GTC CTA GCA ATG TAA TCG TCT ATG ACG TT-3Ј (SS1; DNA sense strand), 5Ј-AAC GTC ATA GAC GAT TAC ATT GCT AGG ACA TCT TTG CCC ACG TTG ACC CA-3Ј (SS2; DNA antisense strand), and 5Ј-UGG GUC AAC GUG GGC AAA GAU GUC CUA GCA AUG UAA UCG UCU AUG ACG UU-3Ј (RNA sense strand) (47). Oligonucleotides were end-labeled with 10 Ci of [␥-32 P] ATP using T4 polynucleotide kinase, and unincorporated nucleotides were removed using Micro Bio-Spin columns (Bio-Rad). Protein and 0.15 nM nucleic acids (oligonucleotide 1 ss-DNA, annealed oligonucleotide 1 and oligonucleotide 2 ds-DNA, and ssRNA) were incubated in 20 mM Tris-HCl, pH 8.0, 60 mM NaCl (unless otherwise stated), 2 mM EDTA, 1 mM DTT, 10% (v/v) glycerol, 0.1 mg/ml BSA for 20 min on ice. Quantitative EMSAs were separated on a native 7% (w/v) polyacrylamide gel. Experiments to assess the role of nucleotides in DNA binding used binding reactions and gels supplemented with 10 mM ZnCl 2 and nucleotide. Polyacrylamide gels were dried and analyzed by autoradiography. EMSAs using unlabeled virion DNA were separated using 0.8% (w/v) Tris acetate-EDTA-agarose gels and stained with ethidium bromide. All reported values for K d represent apparent K d due to the potential for dissociation of protein-DNA complexes during electrophoresis. Curves were fitted by non-linear regression in GraphPad Prism version 6.0.
Construction of DNA Structures-DNA substrates corresponding to double-stranded branched structures (F12-ds/ds), branch structures with two single-stranded arms (F12-ss/ss), and branch structures with one double-stranded and one single-stranded arm (F12-ds/ss) were made by annealing synthetic oligonucleotides from a series that enables the comparison of relative DNA binding affinity to varying DNA topologies independent of DNA sequence (47). Oligonucleotide sequences were 5Ј-GAC GCT GCC GAA TTC TGG CTT GCT AGG ACA TCT TTG CCC ACG TTG ACC C-3Ј (SS3), 5Ј-GCC AGA ATT CGG CAG CGT C-3Ј (LAG), and 5Ј-AAC GTC ATA GAC GAT TAC A-3Ј (LEAD). SS3 was end-labeled with 10 Ci of [␥-32 P]ATP using T4 polynucleotide kinase, and unincorporated nucleotides were removed using Micro Bio-Spin columns (Bio-Rad). SS3, SS1, LAG, and LEAD were annealed to make F12-ds/ds. SS3, SS1, and LEAD were annealed to make F12-ds/ss. SS3 and SS1 were annealed to make F12-ss/ss. SS3 and the corresponding antisense oligonucleotide (5Ј-GGG TCA ACG TGG GCA AAG ATG TCC TAG CAA GCC AGA ATT CGG CAG CGT C-3Ј) were annealed to make a linear dsDNA control (dsF12) and SS3 used as linear ssDNA control (ssF12). Annealing synthetic oligonucleotides with a defined sequence mismatch made DNA substrates corresponding to linear DNA containing bubbles of defined length. Oligonucleo  CAT TGT AGG TAA GCT TAG TAG CGG TAA AGT TAG  ACC AAA-3Ј (P6), and 5Ј-TAA TAA CCT GAT TCA GCG  AAT GAC CGA TAA CGT CCA CTT GAG CGG TAA AGT  TAG ACC AAA-3Ј (P7). P1 was end-labeled with 10 Ci of [␥-32 P]ATP using T4 polynucleotide kinase, and unincorporated nucleotides were removed using Micro Bio-Spin columns. P1 and P2 were annealed to make linear dsDNA (dsP1). P1 and P5 were annealed to make linear dsDNA with a 5-nucleotide bubble (dsP1-5). P1 and P6 were annealed to make linear dsDNA with a 13-nucleotide bubble (dsP1-13). P1 and P7 were annealed to make linear dsDNA with a 20-nucleotide bubble (dsP1-20). P1 was used on its own as a ssDNA control (ssP1). All substrates were gel-purified on a native 10% (w/v) polyacrylamide gel. EMSA was performed as described above.
Time-resolved FRET in Vitro-Synthetic oligonucleotides, unlabeled or end-labeled with fluorescein or tetramethylrhodamine, were purchased from Eurofins MWG. The oligonucleotides used were 5Ј-TGG GTC AAC GTG GGC AAA GA-3Ј (sense strand) and 5Ј-TCT TTG CCC ACG TTG ACC CA-3Ј (antisense strand). Strands were annealed by heating to 90°C for 3 min in 10 mM Tris, pH 8.0, 1 mM EDTA before cooling to room temperature. Measurements used 1.5 M protein with 50 nM DNA in the presence of 60 mM NaCl and were incubated for 10 min at room temperature before analysis. Time-resolved FRET was assessed using the time-correlated, single photon counting technique. The excitation source was a Picoquant pulsed diode laser LDH-P-C-485 (excitation wavelength 485 nm, 70-ps pulse full width at half-maximum at 20 MHz). Fluorescence was detected using an avalanche photodiode (Id Quantique 100-50) linked to a Becker and Hickl SPC 130 timecorrelated, single photon counting module. An instrument response function of ϳ200 ps was measured from Rayleigh scattered light. Fluorescence decays were collected for both donor-and donor-acceptor-labeled double-stranded DNA with or without protein using band pass filter detection of the donor emission and at magic angle polarization. Data were analyzed by the Grinvald-Steinberg method (48) to obtain the fluorescence lifetime for the donor and acceptor ( DA )-and donor only ( D )-labeled oligonucleotides. The data were fitted to a sum of exponentials using an iterative least squares reconvolution procedure with the optical/electrical excitation profile to produce a biexponential decay containing two lifetimes. This profile was obtained from a slide covered with silica LUDOX particles, which provides an instant scatter of the excitation pulse. This data-fitting method provided more accuracy in the determination of shorter lifetimes than calculating a single average lifetime. Donor-acceptor distances (R) were calculated using the equation, E ϭ R 0 6 /(R 0 6 ϩ R 6 ), and a calculated Förster distance (R 0 ) of 49.99 Å. The total length of the oligonucleotide Rx1 Is a DNA-deforming Protein with linkers and fluorescent dyes, at maximum extension, was calculated as 81.1 Å. P 1 Nuclease Sensitivity-Oligonucleotides for P 1 nuclease sensitivity were 5Ј-CTC AAT ACA ATT GTC TCT GTG TAA  ATT TCC TAC GTT TCA TCT GAA AAT CTA GCT ATT  AGA GCT TGG TTT A-3Ј (sense strand) and 5Ј-TAA ACC  AAG CTC TAA TAG CTA GAT TTT CAG ATG AAA CGT  AGG AAA TTT ACA CAG AGA CAA TTG TAT TGA G-3Ј (antisense strand) and represent the C3/mORB dual site sequence at oriC2 of S. solfataricus (49). The sense strand oligonucleotide was end-labeled with 10 Ci of [␥-32 P]ATP as described above, and sense and antisense oligonucleotides were annealed as required. Reactions were performed in 20-l volumes containing 20 mM Tris acetate, pH 7.5, 10 mM magnesium acetate, 100 mM NaCl, 0.15 nM oligonucleotide, and 1.5 M protein. Protein was allowed to bind for 10 min at 37°C. P 1 nuclease was added to a final concentration of 0.01-0.1 units l Ϫ1 and incubated for a further 20 -60 min at 37°C. Reactions were stopped with 5 l of 100 mM Tris-HCl, pH 8.0, 2.5% (w/v) SDS, 100 mM EDTA, 10 units l Ϫ1 proteinase K. 5 l of loading buffer (97.5% (v/v) formamide, 10 mM EDTA, 0.3% (w/v), 0.3% bromphenol blue) was added, and reactions were electrophoresed on a 15% (w/v) polyacrylamide gel with 8 M urea. Polyacrylamide gels were dried and analyzed by autoradiography.
Time-resolved FRET in Situ-A. tumefaciens strain GV3101 (pMP90) was transformed with constructs pK7WGF2 (GFP negative control), pK7WGF2-H2B (GFP-H2B positive control), pBIN35S-NBARC-GFP, pBIN35S-CC-NBARC-GFP, pBIN35S-NBARC-LRR-GFP, pBIN35S-GFP-LRR, pBIN35S-CC-GFP, pBIN35S-Rx1-GFP, pBIN35S-CP105, or pBIN35S-CP106 and grown to A 600 nm 0.8 in YEB medium supplemented with 20 M acetosyringone and 10 mM MES, pH 5.6. Cells were washed three times in infiltration medium (10 mM MES, pH 5.6, 2% (w/v) sucrose, 20 M acetosyringone) and infiltrated at A 600 nm 0.4 into 4 -5-week-old N. benthamiana leaves. Leaves were harvested after 72 h and prior to any observed cell death in a compatible immune interaction, and the agroinfiltrated region was infiltrated with 10 g/ml LDS 751 (Molecular Probes, Inc.). For experiments with CP105 and CP106, the elicitor-encoding A. tumefaciens culture was infiltrated into preinfiltrated sectors after 48 h (24 h before harvest). Leaves were fixed for 4 h at room temperature in 4% (w/v) paraformaldehyde in phosphate-buffered saline (PBS). Fixative was quenched for 30 min at room temperature in 125 mM glycine, and leaves were washed in PBS at 4°C before mounting. A modified Zeiss Axiovert inverted epifluorescence microscope was used for time-resolved fluorescence microscopy. The overall excitation/detection of the fluorescence was performed using the time-correlated single photon counting technique. The excitation source was a Picoquant pulsed diode laser LDH-P-C-440 (excitation wavelength 440 nm, 70-ps pulse full width at half-maximum at 20 MHz). The objective lens (Zeiss ϫ100 oil immersion Ph3) focused the excitation light on the sample material. The emission was detected using suitable band pass/ long pass filters for GFP and LDS 751 fluorescence, respectively. Fluorescence was detected with a photon counting module (Id Quantique 100-50) in a single photon counting mode. Data fitting was performed as for time-resolved FRET in vitro.
The relative orientation of the GFP tag does not affect Rx1 function (23); nor does it affect the ability to observe energy transfer. All data are reported for the analysis of GFP lifetimes because LDS 751 emission is influenced by photobleaching and variability in concentration.
Statistical Analysis-Error bars represent the S.E. with the number of replicates as indicated in the figure legends. Statistical comparisons (p values) for data that pass a test for normality (D'Agostino and Pearson omnibus normality test and Shapiro-Wilk normality test) were obtained from one-way ANOVA with the indicated post hoc test. Statistical comparisons (p values) for data that do not pass a test for normality were obtained from a Kruskal-Wallis test with a post hoc multiple comparisons test. p values in statistical comparisons are indicated in the figures through letters and indicate compared data sets as described in the figure legends.

Results
Plant NLRs Are Structurally Related to Cdc6/Orc1 Family Proteins-The Rx1 gene, introgressed in potato from the wild species Solanum andigena, confers resistance to PVX upon recognition of its coat protein (50,51). The Rx1 protein is a member of the CC-NB-LRR class of plant NLR proteins that consists of an N-terminal CC domain, a central NB-ARC domain, and a C-terminal LRR domain. The NB domain, containing a central ␤-sheet flanked by ␣-helices, is flanked by two ARC subdomains. ARC1 forms a four-helix bundle, and ARC2 adopts a winged helix fold characteristic of DNA-binding transcription factors (52). We hypothesized that an investigation of proteins structurally related to the Rx1 NB-ARC domain could provide insight into NLR biochemistry. Amino acids 143-488, encompassing the NB-ARC domain, were analyzed using the Phyre 2 protein fold recognition engine and expected matches with the pro-apoptotic proteins CED-4 (PDB code 2A5Y) and Apaf-1 (PDB code 1Z6T) were recovered to 100% confidence (10,52). In agreement with earlier structural studies (12), high scoring matches (Ͼ99.4% confidence) were obtained with the Cdc6/ Orc1 proteins of Pyrobaculum aerophilum (PDB 1FNN) and of Aeropyrum pernix in complex with DNA (PDB 2V1U). These proteins are members of a family of proteins involved in origin recognition and DNA replication in archaea and eukaryotes (45,49,53,54). NB subdomain and tandem ARC domain residues (ARC1 and ARC2) of Rx1 are conserved between Cdc6/ Orc1 of A. pernix and Rx1 (35.0% similarity and 12.7% identity between amino acids 134 -479 of Rx1 and amino acids 13-382 of PDB entry 2V1U) (Fig. 1A).
Both the N-terminal NB and C-terminal ARC domain-like regions of Cdc6/Orc1 contact DNA, inducing deformation of the double helix (45,49). The modeled tertiary structure of Rx1 (Fig. 1B, left) was related to Cdc6/Orc1 bound to DNA (PDB code 2V1U) (Fig. 1B, center) but differed from Cdc6/Orc1 in the DNA-unbound state (PDB code 1FNN) (Fig. 1B, right). An overlay demonstrated that the difference between the modeled tertiary structure of Rx1 and Cdc6/Orc1 in the DNA unbound state (PDB code 1FNN) was due to rotation of amino acids 279 -388 of the C-terminal Cdc6/Orc1 ARC-like domain (Fig.  1C, left and center). Amino acids 279 -388 of the Cdc6/Orc1 C-terminal ARC-like domain can be excised from PDB entry 1FNN and directly superimposed onto the Rx1 ARC2 domain to demonstrate how this rotation has occurred in the absence of any global structural change (Fig. 1C, right). Cdc6/Orc1 forms part of a larger family of structural homologues that includes the AAAϩ ATPase SSO1545 from Sulfolobus, RuvB from Thermus, Orc2 from Aeropyrum, mammalian Apaf-1, CED-4 from Caenorhabditis elegans, and NLRC4 from mouse (9,35,(55)(56)(57)(58)(59)(60)(61) of an NB domain that is coupled via its neighboring ARC1 domain to a C-terminal ARC2 domain with varying orientations. For example, the individual domains of the closed form of mouse NLRC4 (PDB code 4KXF) can be extracted and superimposed onto Cdc6/Orc1 in the DNA-bound state (PDB 2V1U), although their actual orientation does not support a DNA binding activity. The modeled structural relationship with Cdc6/Orc1 suggests the intriguing possibility that Rx1 might also interact directly with DNA. We therefore investigated whether Rx1 is a DNA-binding protein.
Rx1 Binds Nucleic Acids in Vitro-A possible direct Rx1-DNA interaction was investigated through in vitro experiments. EMSA using nucleic acid fragments of Ͼ5 kb derived from circular bacteriophage X174 (62) represents a straightforward methodology to qualitatively assess interactions between a protein and either ssDNA or dsDNA with identical sequences. EMSAs were therefore performed using recombinant wild-type Rx1 protein (Rx1(1-489) WT ), consisting of the CC-NB-ARC region but lacking the LRR domain ( Fig. 2A). EMSA experiments performed with the Rx1(1-489) WT protein showed an association with both ssDNA and dsDNA, producing a small upward shift in the migration of the nucleic acid that is fully consistent with similar EMSA experiments using unrelated DNA-binding proteins (Fig. 3A) (63). No mobility shift was observed with a control protein (BSA) that has a similar mass and isoelectric point as Rx1(1-489) WT .
The K176R mutation in the P-loop of Rx1 abolishes its ability to mount an immune response in the presence of the viral coat protein (23). The Rx1(1-489) K176R loss-of-function mutant exhibited a barely detectable binding to DNA as compared with wild type Rx1 protein under these conditions (Fig. 3A). This difference is unlikely to be due to misfolding of the mutant because comparison of Rx1(1-489) WT and Rx1(1-489) K176R by circular dichroism (CD) reveals a generally similar secondary structure composition (Fig. 2, B and C). The CDSSTR method for secondary structure fraction prediction gave similar estimates for secondary structure content for both Rx1(1-489) WT (64). Hence, subtle structural changes rather than an improperly folded protein probably explain differences in DNA binding between Rx1(1-489) WT and Rx1(1-489) K176R .
The Rx1-DNA interaction was relatively stable because it could be visualized after gel electrophoresis (Fig. 3A). Nevertheless, although EMSA using circular bacteriophage X174 DNA is a well established method to qualitatively assess protein-DNA interactions, it does not enable robust quantification of the affinity of a protein for nucleic acids. EMSA with small synthetic oligonucleotides is a standard method to quantify protein-nucleic acid interactions (65). Furthermore, the high molecular weight of X174 DNA and consequent small band shifts were not suited to further analysis. We therefore quantified the affinity of Rx1(1-489) for various nucleic acids by EMSA using 32 P-labeled synthetic oligonucleotides whose sequences were unrelated to that of bacteriophage X174 DNA and which should provide more robust band shifts on EMSA due to their lower molecular weights (Fig. 3A). Rx1(1-489) WT showed broadly similar apparent affinities (K d app ) for dsDNA   OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 and ssRNA but exhibited a significantly higher apparent affinity for ssDNA (Fig. 3, B-D, and Table 1). The affinity of Rx1(1-489) WT for dsDNA is within the submicromolar range and is of a similar magnitude as both eukaryotic and prokaryotic Cdc6/ Orc1 proteins (66,67). The apparent affinity of the Rx1(1-489) K176R mutant for ssDNA, dsDNA, and ssRNA was lower than the apparent affinity of wild type Rx1 in each case, which corresponds to the observed lower affinity established using the X174 DNA (Akaike information criterion, p Ͼ 0.99). To  exclude the possibility that the observed nucleic acid binding was an artifact of the recombinant protein, we purified fulllength Rx1 protein from plants. The protein was purified using a C-terminal 4-fold Strep-tag (Rx1-4Strep) from agroinfiltrated N. benthamiana leaves. The amount of purified Rx1 protein obtained was limited but sufficient to demonstrate that plantderived Rx1-4Strep is also able to bind to ssDNA in vitro (Fig.  3E). Plant-derived Rx1-4Strep DNA binding was weaker than that of bacterially derived protein, which could be due to the fact that the majority of the full-length Rx1 is presumably in the autoinhibited off-state. Only a small fraction is proposed to be spontaneously active and thought to be responsible for the weak HR phenotype observed when Rx1 is overexpressed in the absence of the CP elicitor (68). In addition, the possibility cannot be excluded that the tag has impacted folding of a portion of the plant-expressed Rx1 protein.

Rx1 Is a DNA-deforming Protein
The NLR NB-ARC Domain Binds Nucleic Acids in Vitro-Despite the structural relationship between the Rx1 NB-ARC domains and Cdc6/Orc1 proteins, it is formally possible that the data of Fig. 3 can be explained by an interaction between nucleic acids and the N-terminal CC domain of Rx1(1-489) rather than its NB-ARC domain. We were unable to produce truncated Rx1 fragments encompassing solely the NB or NB-ARC domains. We therefore examined another plant NLR protein to assess whether the NB-ARC domain alone is able to bind nucleic acids and whether DNA binding is unique to Rx1 or represents a common property of at least a subset of plant NLRs. The NLR subdomains of the orphan NLR of the monocot Zea mays were chosen because both the NB and NB-ARC domains can be produced as soluble recombinant protein (14). We compared ssDNA binding of the NB subdomain of PSiP alone (PSiP-NB) with that of the complete NB-ARC domain of PSiP (PSiP-NB-ARC) (Fig. 3F). Although both fragments bound, the PSiP-NB-ARC domains bound ssDNA with a considerably higher affinity than the PSiP-NB domain alone (Table  1). Together, these data demonstrate that the NB-ARC domain is sufficient for nucleic acid binding in Rx1 and PSiP, that DNA binding is a property of at least a subset of plant NLR proteins, and that both the NB and the ARC subdomains contribute to the DNA interaction.
Rx1 Deforms DNA-In the "switch" model for plant NLR activation, binding of ATP to the NB-ARC domain establishes the "on" state, whereas hydrolysis of ATP to ADP restores the "off" state (9). An intact P-loop is essential for nucleotide binding, and mutations in this motif typically result in loss-of-function alleles (9). We therefore investigated the relationship between P-loop-dependent ATPase activity and DNA binding. We detected no ATPase activity in Rx1(1-489) WT , possibly indicating the absence of a catalytic water molecule, as observed previously for the STAND ATPase Ced-4 (69). Neither ATP nor ADP had any discriminatory influence on Rx1(1-489) binding to dsDNA (Fig. 4A). We therefore investigated whether Rx1 has activities at DNA other than binding that are affected by the type of nucleotide (ATP/ADP) bound. The Cdc6/Orc1 family proteins ORC1 of A. pernix and the Orc1-1/Orc1-3 heterodimer of S. solfataricus substantially deform origin DNA by bending it with angles of 35 and 20°, respectively, thereby inducing localized melting of the double helix (45,49,70). We therefore examined whether Rx1(1-489) can deform DNA in a similar fashion and whether this process is nucleotide type-dependent. To measure DNA bending, time-resolved FRET was used because it allows measurements of distances between fluorophores. This method offers considerable advantages over steady-state FRET because the fluorescence lifetime represents an intrinsic property of the fluorophore and is independent of concentration, photobleaching, or light scattering (71). We monitored DNA deformation using time-resolved FRET with  OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 24953
dual end-labeled dsDNA (72). Upon FRET, the fluorescence lifetime of the donor fluorophore decreases; therefore, we deconvoluted the fluorescence donor emission for its constituent lifetime components. We hypothesized that following DNA bending, we would observe a shortened donor lifetime due to energy transfer to the acceptor. DNA bending was assessed under conditions to saturate binding of Rx1(1-489) WT or Rx1(1-489) K176R to dsDNA. Bending was evident as a decrease in the contribution of a 4.1-ns component (indicative of unperturbed donor emission; fluorescein fluorescence) to and the appearance of a 129-ps component during the total fluorescence decay of FRET donor emission (Fig. 4B, the 129-ps component is marked with an arrow). The 129-ps lifetime component, attributed to energy transfer, was only observed for the Rx1(1-489) WT protein and Rx1(1-489) K176R protein and not in controls without protein or with BSA except for a minor contribution with the latter in the presence of ATP (Fig. 4C). The 129-ps lifetime corresponds to a calculated donor-acceptor distance of 29 Å (assuming isotropic orientations) and therefore an overall bend angle of 42°around a presumed oligonucleotide midpoint.
Next, it was investigated whether nucleotides had an influence on the observed DNA bending. Notably ATP, but not ADP, strongly increased the contribution of the 129-ps lifetime to the overall time-resolved data. This increase was only observed for the Rx1(1-489) WT protein and not for the Rx1(1-489) K176R mutant, indicating that DNA bending requires an intact P-loop capable of binding nucleotides. The distinct response of the Rx1(1-489) WT protein following incubation with either ADP or ATP provides additional support for a correct native fold of the nucleotide-binding pocket in the recombinant protein. The absence of any change in the value of the shortened lifetime (129 ps) shows that the calculated donoracceptor distance is constant. Because the relative proportion of the 129 ps lifetime to the total fluorescence signal increases in the presence of ATP and Rx1(1-489) WT , we can conclude that ATP binding enhances the pool of protein-DNA complexes in the bent state but not the bending angle.
Time-resolved FRET is a well validated method to examine intramolecular distances and therefore DNA topology, but it does not provide further information on other DNA distortions associated with changes in topology. To examine whether Rx1(1-489) can induce local DNA melting, as has been observed for Orc1, we explored a non-fluorescence-based methodology. P 1 nuclease has been used previously as a tool to examine local DNA distortion using the Orc1 protein of A. pernix (70). We therefore examined the sensitivity of dsDNA oligonucleotides to the ssDNA-specific P 1 nuclease in the presence of Rx1(1-489) (73). As expected, ssDNA was significantly degraded by P 1 nuclease (positive control), whereas dsDNA, in the presence of BSA (negative control), was largely resistant to P 1 nuclease activity (Fig. 5). Although dsDNA was more sensitive to P 1 nuclease in the presence of Rx1(1-489), the mutant Rx1(1-489) K176R did not induce local DNA melting because no increased DNA degradation was observed. Thus, although in the absence of nucleotides, Rx1(1-489) K176R bends dsDNA to a similar magnitude as Rx1(1-489) WT , its failure to melt DNA might be a manifestation of subtle changes to DNA binding evidenced through the decrease in binding affinity (Fig. 3,  B-D). These experiments were insufficiently sensitive to examine the further influence of nucleotides on NLR-mediated DNA melting. The P 1 sensitivity of dsDNA in the presence of Orc1-1/Orc1-3 was indistinguishable from that of dsDNA in the presence of Rx1(1-489) WT , supporting the interpretation that plant NLRs can cause local dsDNA melting. In conclusion, Rx1 is able to both bend DNA and provoke local DNA melting, and this bending activity requires an intact P-loop and is stimulated by the presence of ATP.
Rx1 Preferentially Binds Specific DNA Topologies in Vitro-We sought independent experimental support for Rx1-mediated distortion of DNA. We hypothesized that if Rx1 distorts linear DNA, then the free energy of Rx1 binding to DNA structures that resemble the distorted state would be favored (with a corresponding increase in affinity). Indeed, Rx1 bound branched double-stranded DNA with a significantly higher affinity than control linear double-stranded DNA of similar sequence (Fig. 6A; compare dsF12 with F12-ds/ds). The branched double-stranded DNA represents a non-natural DNA and is a control to demonstrate a preference for Rx1(1-489) binding to a branched topology. When comparing binding affinities for naturally occurring branched topologies, we noted a higher affinity for branched structures with one dsDNA and two ssDNA arms (e.g. similar to a transcription bubble) compared with structures with one or two duplex arms (e.g. resembling a DNA replication fork) (Fig. 6A, compare F12-ds/ss with F12-ss/ss). Consistent with our model of local DNA melting, Rx1(1-489) showed a higher affinity for small DNA bubbles compared with linear dsDNA (Fig. 6B). This increased affinity was not due to the increased affinity for ssDNA because affinity did not correlate with increasing DNA bubble size. Although these data cannot reveal the exact nature of the distorted DNA state on DNA bending (Fig. 4) and melting (Fig. 5), analysis of the relative affinities does demonstrate that Rx1 shows an increased affinity for specific DNA structures, and the DNA distortion we observed in the presence of Rx1 is probably a genuine response following its activation.
Rx1 DNA Binding Is Specifically Activated by Its Cognate Elicitor in Vivo-DNA binding, bending, and melting is a new aspect of plant NLR biochemistry. To validate DNA as a downstream target for NLR signaling and link this biochemical activity to its function in plant cells, we tested whether Rx1 is able to interact with DNA in vivo. To investigate the possibility of a direct interaction with genomic DNA inside the cell, we studied Rx1-DNA interactions in the nucleus using Forster resonance energy transfer-fluorescence lifetime imaging microscopy (FRET-FLIM). FRET-FLIM has been used previously to demonstrate transcription factor binding to DNA in response to environmental signals (74).
N. benthamiana was infiltrated with A. tumefaciens carrying constructs encoding either GFP (negative control), a protein consisting of A. thaliana histone H2B fused to GFP (GFP-H2B; positive control), or discrete domains of Rx1 fused to GFP. Previous work has similarly utilized H2B-GFP and naked GFP as controls for DNA binding in paraformaldehyde-fixed preparations (74). The constituent fluorescence lifetimes for the GFP tag were examined in leaves counterstained with LDS 751. LDS 751 is a cell-permeable nucleic acid stain with an excitation maximum, when bound to DNA, that overlaps with the GFP emission spectrum. GFP showed two distinct lifetimes at ϳ0.5 and 1.5 ns (Fig. 7A). Because energy transfer from donor (GFP) to acceptor (LDS 751) decreases the fluorescent lifetime, we hypothesized that the shorter lifetime for GFP is representative of energy transfer consistent with an interaction between the fluorophores. Notably, such a decrease in the GFP fluorescence lifetime by time-correlated single-photon counting is independent of protein expression levels, quenching, photobleaching, or fluctuations in the excitation source. A decrease in lifetime can therefore specifically be attributed to quenching of the excited state of the GFP and represents strong evidence for energy transfer from GFP to LDS 751 and thus a direct protein-DNA interaction. Consistent with this interpretation, a significant decrease in the ratio of the yields of the GFP fluorescence lifetimes was observed for the DNA-binding protein GFP-H2B (Fig. 8A). To demonstrate that energy transfer to LDS 751, and not to surrounding proteins, explains the data, we confirmed that the decrease in the fluorescence GFP lifetime ratio indicative of DNA binding was correlated with an increase in LDS 751 emission that does not arise from the excitation source (Fig.  7B). Although the exact stoichiometry of GFP and LDS 751 levels is not known in each experiment, the finding that the ratio of fluorescence emission for GFP/LDS 751 is significantly decreased for H2B compared with the negative control is strong evidence that the reduction in GFP lifetimes is due to energy transfer to LDS 751 and not to an alternative molecule. As predicted, Rx1-GFP fusions containing the NB-ARC domain (NB-ARC-GFP, CC-NB-ARC-GFP, and NB-ARC-LRR-GFP) showed a significant decrease in the ratio of GFP lifetime yields, consistent with its observed DNA binding activity in vitro, whereas the LRR (GFP-LRR) domain did not. Surprisingly, the CC domain alone (CC-GFP) also showed a decrease in the ratio of GFP lifetime yields. The Rx1 CC domain has been shown previously to associate with a high molecular weight complex in the nucleus (23), and our findings indicate that this complex probably contains genomic DNA. Taken together, these data demonstrate that the CC and CC-NB-ARC Rx1 domains can form a stable interaction with DNA in situ. The FRET-FLIM methodology used is independent of expression levels of the various constructs. However, the methodology can be sensitive to cleavage of the GFP tag of even a small percentage of   OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 expressed protein. Fortunately, cleavage of the GFP tag can be resolved because it yields a high rather than the observed low ratio of fluorescence lifetimes, allowing us to conclude that the positive results for DNA binding in situ are not attributable to tag cleavage.

Rx1 Is a DNA-deforming Protein
Next, we investigated whether the full-length Rx1 molecule (Rx1-GFP) also binds to DNA in the plant cell. Notably, a fulllength Rx1-GFP fusion showed no binding to DNA as compared with the negative control (Fig. 8B). This implies that the inactive full-length Rx1 protein adopts a structure refractory to interacting with DNA. To test whether there is a relationship between DNA-binding and Rx1 activation and subsequent immune signaling, we next co-expressed Rx1 with the PVX coat protein elicitor, which is known to trigger immunity (23). Fulllength Rx1-GFP was found to bind to DNA only in the presence of the wild type (avirulent) coat protein (CP106) and not in the presence of a mutant (virulent) elicitor (CP105) that is unable to activate Rx1 (Fig. 8B). These data show that DNA binding in vivo by Rx1 only occurs upon perception of its cognate elicitor.
To test whether DNA binding in situ requires elicitor recognition in the cytoplasm, we investigated DNA binding of Rx1-GFP fused to either an NLS or an NES. These chimeric tags have been demonstrated previously to constrain Rx1 to the nucleus or cytoplasm, respectively (23). This experiment addresses two questions. 1) Is enforced nuclear accumulation of a GFP fusion protein sufficient to confer DNA binding? 2) At what subcellular localization can Rx1 be activated by the coat protein to permit DNA binding? GFP-NLS-Rx1 did not bind DNA in the presence or absence of CP106, demonstrating that enforced Rx1 accumulation in the nucleus is insufficient to drive DNA binding and that DNA binding requires CP106 recognition in the cytoplasm, consistent with previous findings (23) (Fig. 8C). GFP-NES-Rx1 also did not bind DNA in the presence or absence of CP106, demonstrating that the DNA binding signal is dependent on the ability of the cytosolic Rx1 protein to gain access to genomic DNA regardless of exposure to CP106. GFP-NES-Rx1 and GFP-NLS-Rx1 are not sensitive to cleavage of GFP, thus excluding the possibility that the observed absence of DNA binding is due to sensitivity to proteolysis (23). Taken together, the data support a model where Rx1 binding to DNA is a specific nuclear event in immune activation subsequent to coat protein detection in the cytosol.
We further investigated binding of Rx1-GFP to DNA upon activation of another immune receptor to exclude the possibility that Rx1 binds DNA as a nonspecific consequence of defense activation. We co-infiltrated N. benthamiana with A. tumefaciens carrying constructs encoding Rx1-GFP, the Pto kinase of tomato, and the AvrPto effector. The Pto kinase activates an immune response in N. benthamiana upon binding the AvrPto effector of Pseudomonas syringae pv. tomato (75-77). Rx1-GFP FIGURE 8. Binding of Rx1 protein domains to DNA in situ. A, the ratio of the long (Ͼ1.0 ns) to short (Ͻ0.5 ns) GFP lifetimes for GFP fusion constructs representing varying Rx1 subdomains produced in N. benthamiana leaves using agroinfiltration (n Ͼ 6; *, p Ͻ 0.05 compared with GFP by one-way ANOVA with post hoc Dunnett test). B, ratio of the long (Ͼ1.0 ns) to short (Ͻ0.5 ns) GFP lifetimes for Rx1-GFP full-length constructs alone and upon co-expression with virulent (CP105) and avirulent alleles (CP106) of the PVX CP (n Ͼ 6; *, p Ͻ 0.05 compared with GFP by one-way ANOVA with post hoc Dunnett test). C, ratio of the long (Ͼ1.0 ns) to short (Ͻ0.5 ns) GFP lifetimes for GFP-NLS-Rx1 and GFP-NES-Rx1 full-length constructs alone and upon co-expression with the avirulent allele of PVX, CP106 (n ϭ 4; *, p Ͻ 0.05 compared with GFP by one-way ANOVA with post hoc Dunnett test). D, ratio of the long (Ͼ1.0 ns) to short (Ͻ0.5 ns) GFP lifetimes for Rx1-GFP full-length construct alone and upon co-expression with either the avirulent allele of PVX, CP106, or the Pto kinase and AvrPto (n ϭ 12-20; *, p Ͻ 0.05 compared with GFP by one-way ANOVA with post hoc Dunnett test). Error bars, S.E.

Rx1 Is a DNA-deforming Protein
did not bind DNA when Pto was activated by AvrPto, indicating that Rx1-GFP DNA binding is not a generic response following defense activation (Fig. 8D). Because the role of Rx1 in immunity is dependent upon both its activation by the viral coat protein in the cytoplasm and its DNA binding activity in the nucleus, our findings therefore provide the first evidence for a direct molecular target between activation of a plant NLR and subsequent cellular immune responses.

Discussion
The molecular mechanism underlying the function of activated NLR proteins in plant immunity is a crucial, but still unanswered, question. Existing in vitro, in vivo, and bioinformatics data pinpoint the NB-ARC domain as a central switch in regulating NLR activity. We here propose that the NB-ARC domain also possesses an intrinsic DNA binding activity, and we demonstrate that its DNA binding activity is associated with the cellular immune response. The Rx1 protein is observed to bind and deform dsDNA in vitro and to bind cellular DNA in response to activation following elicitor perception. Importantly, although the described biochemistry for Rx1 is novel for a NLR protein, DNA distortion is a well characterized feature of other proteins that interact with DNA through non-sequencespecific interactions, including TATA box-binding protein (78,79), integration host factor (80), and the HMG box (81). Rx1 biochemistry is therefore consistent with the activity of known DNA-binding proteins.
Our observation that Rx1 can interact with DNA in response to immune activation might provide a rationale for its nuclear localization. For example, a P-loop mutant in Rx1 can potentially establish a correlation between DNA binding and immunity. The K176R P-loop mutant of Rx1 is defective in triggering immunity (23) to PVX. We show that this mutant is also defective in nucleotide-dependent DNA bending and DNA melting in vitro. This finding represents a potential link between the ability of Rx1 to distort DNA in vitro and the ability to trigger immunity in planta. Equivalent mutations in the NB domain of Cdc6 have been used to investigate the activity of Cdc6 at dsDNA (82,83).
In vivo activation of Rx1 by the PVX coat protein induces the plant immune response (84) (Fig. 8B). We found that Rx1 only bound nuclear DNA following recognition of the CP106 coat protein and not the CP105 variant, which is unable to trigger Rx1 signaling. These data show that only properly activated Rx1 has the ability to interact with DNA in situ. In addition, only cytosolic recognition of CP106, followed by translocation of activated Rx1 to the nucleus, results in full activation of immunity (23). We demonstrate that, even in the presence of the CP106 coat protein, no DNA binding occurs when Rx1 is artificially retained in either cytosol or nucleus (Fig. 8C). This finding presents a potential link between the known spatial requirements for Rx1-mediated immune activation and the DNA binding observed in situ. Such a translocation mechanism might be analogous to that of WHIRLY1, an immune activated transcriptional regulator that translocates to the nucleus and is involved in defense gene expression (85). In vitro, full-length (hence mostly inactive) Rx1 purified from N. benthamiana did interact with DNA, albeit less strongly than the CC-NBARC form produced in E. coli (Rx1(1-489) WT ) (Fig. 3E), which is free of the autoinhibitory constraint posed by the LRR domain (86). DNA binding in vitro with full-length Rx1 occurred under conditions where relatively high protein concentrations can be assayed. Presumably, Rx1 levels in vivo are too low to observe DNA binding in its non-activated state (Fig. 8B). The observed DNA binding by full-length Rx1 in situ is not a generic consequence of plant immunity because activation of immunity through another immune receptor (Pto/AvrPto) did not induce Rx1 DNA binding. We therefore propose that DNA binding by Rx1 upon PVX coat protein perception is an essential, specific, and early step in the cellular immune response.
The Rx1 NBARC domains share remarkable biochemical properties with the Cdc6/Orc1 family DNA-binding proteins. Rx1 was observed to bind both ssDNA and dsDNA similar to ORC of Saccharomyces cerevisiae (87). The Cdc6/Orc1 homology with NLR proteins and the DNA binding characteristics of the separate PSiP NB and NB-ARC domains accord with multiple contacts with DNA across both NB and ARC domains. Hence, single point mutations are unlikely to abolish DNA binding, and, consistent with previous observations (67), we have not identified point mutations that ablate DNA binding. Eukaryotic ORCs lack DNA sequence specificity in vitro but show higher affinity for specific DNA topologies (88,89). Consistent with this, Rx1 shows higher affinity for branched and melted DNA topologies than for dsDNA. The bend angle introduced into DNA by Rx1(1-489) is also of a magnitude similar to that observed in crystal structures of ORC1 from A. pernix (45). Analysis of A. pernix ORC2 revealed a considerable conformational flexibility stabilized by ATP (54). In this context, it is interesting to note that although the bend angle is identical for both wild type and mutant Rx1 proteins in the absence of nucleotide or presence of ADP, the population of DNA in the bent state was more prevalent in Rx1(1-489) WT supplemented with ATP. The Rx1 activated state is therefore specifically linked to DNA distortion.
The activity of Rx1 on DNA provides biochemical evidence that Rx1 might act as a transcriptional regulator through its NB-ARC domain. DNA binding by a NLR is a signaling event because the NB-ARC domain is not involved in recognition specificity. Pathogen recognition by NLRs is typically determined by the LRR, often in conjunction with integrated effector targets (90,91). A key process in transcriptional activation is the distortion of DNA to enable the formation of the transcription preinitiation complex (92)(93)(94)(95). In the cell, Rx1 protein might have sequence-specific DNA binding conferred by interacting protein partners, whereas the NB-ARC domain distorts DNA to a state that activates or represses transcription, depending upon the locus (96). The region encompassing the CC domain, which might interact with DNA via an accessory protein (e.g. a transcription factor), could confer this sequence specificity. The identification of such a binding partner that can confer sequence specificity to the Rx1-DNA interaction represents a significant challenge for the future.
In summary, we have identified a conserved DNA binding and distorting activity in the NB-ARC domain of the Rx1 protein in vitro and link Rx1 activation following elicitor recognition to nuclear DNA binding in situ. Rx1 induces cellular immune responses after viral coat protein recognition. We hypothesize that a function for Rx1 is to manipulate DNA into an "immune competent" state. The precise nature and role of this Rx1 protein-DNA immune competent state can now be addressed in future studies.