Using decoys to expand the recognition specificity of a plant disease resistance protein

See allHide authors and affiliations

Science  12 Feb 2016:
Vol. 351, Issue 6274, pp. 684-687
DOI: 10.1126/science.aad3436

Improving plant disease responses

Disease resistance in plants depends on genes that allow them to recognize when they are infected by a pathogen so that they can mount a timely defense response. Unfortunately, pathogens can often overcome endogenous disease resistance genes by evolving new virulence strategies that escape detection. Kim et al. modified the pathogen recognition systems in the model plant Arabidopsis thaliana to widen its reach. The approach should enable the development of crops with more durable disease resistance and hence reduce pesticide use and increase crop yields.

Science, this issue p. 684


Maintaining high crop yields in an environmentally sustainable manner requires the development of disease-resistant crop varieties. We describe a method to engineer disease resistance in plants by means of an endogenous disease resistance gene from Arabidopsis thaliana named RPS5, which encodes a nucleotide-binding leucine-rich repeat (NLR) protein. RPS5 is normally activated when a second host protein, PBS1, is cleaved by the pathogen-secreted protease AvrPphB. We show that the AvrPphB cleavage site within PBS1 can be substituted with cleavage sites for other pathogen proteases, which then enables RPS5 to be activated by these proteases, thereby conferring resistance to new pathogens. This “decoy” approach may be applicable to other NLR proteins and should enable engineering of resistance in plants to diseases for which we currently lack robust genetic resistance.

Intracellular receptors belonging to the nucleotide-binding leucine-rich repeat (NLR) family play central roles in both the human and plant innate immune systems (1, 2). In plants, their primary function is in pathogen detection, and this often involves the recognition of pathogen-derived virulence factors known as effector proteins. After detection of effector proteins, NLRs become activated, leading to the induction of numerous defense responses, including a localized cell death response termed the hypersensitive response (HR) that serves to prevent spread of infection. NLR proteins are highly specific with regard to the pathogen effectors that each can detect, with a single NLR protein capable of detecting only a limited number of effectors. Research on plant NLRs conducted over the past 20 years has focused on understanding the mechanistic basis of this specificity, with a long-term goal of being able to create new specificities. The ability to engineer novel specificities would enable the production of crop plants with genetically based resistance to diseases that currently must be controlled by environmentally damaging, and expensive, pesticides.

Although substantial progress has been made toward understanding how plant NLRs detect pathogen effectors, it has not yet been possible to engineer entirely new specificities. The successes to date have resulted in only minor expansion of specificities (35). Because the direct modification of NLR proteins has met with limited success, we have been pursuing an alternative strategy to activate the defense pathways regulated by NLR proteins. Many NLR proteins detect pathogen effector proteins indirectly by sensing modification of other host proteins that are themselves targeted by pathogen effectors. This indirect recognition mechanism can be likened to a mousetrap in which the NLR protein is the trap, the effector is the mouse, and the effector target is the bait. When the mouse (effector) nibbles on the bait (effector target), it triggers the trap (NLR protein) to undergo a large conformational change. We hypothesized that it would be possible to alter the type of mouse caught (detected) by altering the bait rather than the trap.

To pursue this strategy, we used the Arabidopsis NLR protein RESISTANCE TO PSEUDOMONAS SYRINGAE 5 (RPS5), which mediates detection of the effector protein AvrPphB from this pathogen (6). AvrPphB is a protease that targets a small family of protein kinases that function in basal immune signaling (7, 8). One of these kinases is PBS1. Cleavage of PBS1 by AvrPphB activates RPS5 (9). By analogy to the mousetrap model, PBS1 serves as the bait in this system. RPS5 and PBS1 form a preactivation complex, and when PBS1 is cleaved by AvrPphB, the resulting conformational change in PBS1 triggers RPS5 (10). This conformational change can be mimicked by insertion of three amino acids at the cleavage site; the kinase so altered can activate RPS5 in the absence of AvrPphB and in the absence of cleavage (10). Thus, activation of RPS5 does not require direct binding of RPS5 to the bacterial effector. RPS5 should be able to respond to any pathogen effector that can cause the requisite conformational change in the plant’s intermediary kinase, PBS1.

We tested this hypothesis by swapping the proteolytic target site in PBS1 normally cleaved by AvrPphB (Gly-Asp-Lys-Ser-His-Val-Ser) with another proteolytic site (Val-Pro-Lys-Phe-Gly-Asp-Trp) that would be recognized by a different P. syringae effector, AvrRpt2 (Fig. 1A). This proteolytic site is found in a target of AvrRpt2 named RPM1 INTERACTING PROTEIN 4 (RIN4) (11). Cleavage of RIN4 by AvrRpt2 normally triggers immune responses through a different NLR protein, RPS2 (12). We refer to this modified PBS1 as PBS1RCS2 because this sequence corresponds to RIN4 cleavage site 2 (11, 12). We transiently coexpressed PBS1RCS2 with RPS5 and AvrRpt2 in Nicotiana glutinosa and assessed RPS5 activation by monitoring tissue collapse resulting from induction of HR-associated cell death. AvrRpt2 induced a strong HR that was dependent on AvrRpt2 protease activity and on the modified PBS1 (Fig. 1B). This HR was also dependent on coexpression of RPS5, indicating that we had succeeded in switching the recognition specificity of RPS5 from AvrPphB to AvrRpt2. To quantify the HR, we measured electrolyte leakage, a proxy for cell death. Consistent with the macroscopic symptoms, PBS1RCS2 with AvrRpt2 induced as much electrolyte leakage as wild-type PBS1 cleaved by AvrPphB, whereas PBS1RCS2 with the mutation Cys122 → Ala (C122A) only weakly activated RPS5 (Fig. 1C). Immunoblot analysis confirmed that AvrRpt2 cleaved PBS1RCS2 at 4 hours after induction, whereas C122A or AvrPphB did not (Fig. 1D). Together, these data establish that PBS1RCS2 is a substrate for AvrRpt2 and that AvrRpt2-mediated cleavage activates RPS5, at least when transiently overexpressed in N. glutinosa.

Fig. 1 The specificity of the Arabidopsis RPS5 disease resistance protein can be altered by modifying the Arabidopsis PBS1 protein.

(A) Schematic representation of the PBS1RCS2 construct. Magenta indicates the kinase domain within PBS1; dark blue indicates the position of the kinase activation loop. Vertical bars indicate protease cleavage position. (B) Activation of RPS5 after cleavage of PBS1RCS2 by AvrRpt2 causes cell death in N. glutinosa leaves. The indicated constructs were transiently coexpressed in N. glutinosa by means of Agrobacterium tumefaciens infiltration. C122A indicates a protease-inactive form of AvrRpt2. (C) Activation of RPS5 after cleavage of PBS1RCS2 by AvrRpt2 causes electrolyte leakage in N. glutinosa leaves. Data are means ± SD (n = 4). R5 and P1 denote RPS5 and PBS1. (D) Cleavage of PBS1RCS2 by AvrRpt2. Total protein was isolated from duplicates of the leaves shown in (B) and analyzed by immunoblotting with the indicated antibodies. Transient gene expression was induced using dexamethasone (DEX). Asterisk indicates position of the PBS1 C-terminal cleavage product.

Encouraged by these results, we used the native PBS1 promoter to generate transgenic Arabidopsis plants expressing PBS1RCS2. For this experiment, we needed to use a genotype of Arabidopsis that does not normally activate a HR in response to AvrRpt2. We thus used an Arabidopsis mutant line that lacked both RIN4 and RPS2 (13). We assessed four independent transgenic lines expressing PBS1RCS2 for HR to infection by P. syringae expressing AvrRpt2. At 21 hours after inoculation with P. syringae strain DC3000(avrRpt2), two independent transgenic lines (#5 and #2) showed a visible HR, whereas the untransformed rin4rps2 mutant did not (Fig. 2A). In planta bacterial growth assays showed that growth of DC3000(avrRpt2) in lines #5 and #2 was less than in the parent rin4rps2 line by a factor of 100 to 200, whereas bacterial growth in lines #1 and #3 was reduced by a factor of 5 to 50 (Fig. 2B). However, these lines were equally susceptible to DC3000(C122A), indicating that AvrRpt2 protease activity is required to activate resistance. Both induction of HR and restriction of bacterial growth correlated with expression levels of PBS1RCS2 (Fig. 2C). At 12 hours after inoculation, we detected a cleavage product of PBS1RCS2 in transgenic line #5 after inoculation with DC3000(avrRpt2), but not with DC3000 lacking avrRpt2 [DC3000(empty vector); Fig. 2D]. Thus, cleavage of PBS1RCS2 by AvrRpt2 activates RPS5 in Arabidopsis. These transgenic plants, which still contain a wild-type copy of PBS1, also displayed HR 21 hours after injection with DC3000(avrPphB) (Fig. 2E), demonstrating that the native recognition specificity of RPS5 was retained. Together these results suggest that RPS5-mediated disease resistance can be activated by two different protease effector proteins in the PBS1RCS2 transgenic plants, and that the recognition specificity of RPS5 can be expanded by addition of new “decoy” variations of PBS1.

Fig. 2 Transgenic expression of PBS1RCS2 in Arabidopsis confers resistance to P. syringae expressing avrRpt2.

(A) Recognition of AvrRpt2 by transgenic Arabidopsis plants expressing PBS1RCS2. The indicated Arabidopsis lines were inoculated with P. syringae DC3000(avrRpt2) (top row) or DC3000(C122A) (bottom row) in the left half of each leaf and scored for HR. Col-0 indicates wild-type Arabidopsis, which recognizes AvrRpt2 using RPS2; rin4rps2 indicates the double mutant parent used to generate the transgenic lines. (B) PBS1RCS2 confers resistance to DC3000(avrRpt2) in Arabidopsis. Bacterial growth was measured in the indicated plant lines shown in (A). Data are shown as mean colony-forming units (cfu) cm−2 ± SD (n = 4). Asterisks indicate statistically significant differences from growth observed in the rin4rps2 parent for a given strain (**P < 0.01, *P < 0.05; Tukey’s honest significant difference test). (C) Resistance to DC3000(avrRpt2) correlates with expression of PBS1RCS2. Proteins from transgenic lines shown in (A) were immunoprecipitated with anti-HA agarose and immunoblotted. (D) PBS1RCS2 expressed in transgenic Arabidopsis is cleaved by AvrRpt2 delivered by DC3000. The asterisk indicates the C-terminal PBS1RCS2 cleavage product. (E) PBS1RCS2 transgenic plants, which contain a wild-type copy of PBS1, recognize both AvrRpt2 and AvrPphB. PBS1RCS2 transgenic plants displayed a visible HR 21 hours after injection with P. syringae strain DC3000(avrPphB).

To test whether this approach could be extended to recognize pathogens beyond P. syringae, we created a PBS1 variant that can be cleaved by the NIa protease of tobacco etch virus (TEV), which was chosen because the recognition sequence of TEV NIa protease is well characterized (14). We replaced the AvrPphB cleavage site in PBS1 with a TEV protease cleavage site, generating PBS1TCS (Fig. 3A). TEV is a positive-stranded RNA virus that encodes a polyprotein that must be posttranslationally processed by its embedded NIa protease. This protease is essential for viral replication; thus, disease resistance that is triggered by its enzymatic activity should be highly durable, as it would be extremely difficult for the virus to simultaneously change the specificity of its protease and the protease cleavage sites embedded within its polyprotein. Because all potyviruses depend on endogenous proteases for replication, this general approach could enable engineering durable resistance to many different plant viruses of economic importance.

Fig. 3 Replacement of the AvrPphB cleavage site in PBS1 with a TEV NIa protease cleavage site enables HR activation by TEV protease.

(A) Schematic representation of the PBS1TCS construct. (B) Cleavage of PBS1TCS by TEV protease activates HR in N. benthamiana leaves when coexpressed with RPS5. The indicated constructs were transiently coexpressed in N. benthamiana. (C) Electrolyte leakage analysis of N. benthamiana leaf disks coexpressing RPS5, PBS1TCS, and TEV protease. Data are means ± SD (n = 4). (D) Cleavage of PBS1TCS by TEV protease. PBS1TCS-HA or PBS1-HA was transiently coexpressed with TEV protease-Myc (TEV) or AvrPphB-Myc in N. benthamiana and proteins analyzed by immunoblotting. Asterisk indicates position of C-terminal PBS1 cleavage product. PBS1TCS is rapidly degraded after cleavage, hence the faint signal in lane 1.

We transiently coexpressed PBS1TCS with TEV NIa protease and RPS5 in N. benthamiana. Cell death was induced when RPS5 was coexpressed with PBS1TCS and TEV protease, but was not induced when either PBS1TCS or TEV protease was excluded (Fig. 3B). Quantification of cell death, assessed by electrolyte leakage, showed that PBS1TCS and TEV protease induced cell death equivalent to wild-type PBS1 and AvrPphB (Fig. 3C). Immunoblot analysis confirmed that TEV protease cleaved the target PBS1TCS at 6 hours after induction, whereas AvrPphB did not (Fig. 3D). Also, TEV protease did not cleave wild-type PBS1.

These results establish that PBS1 can be engineered to function as a target for proteases from two very different classes of pathogen: viruses and bacteria. To determine whether cleavage of modified PBS1 can productively activate RPS5 and initiate an effective immune response against viruses, we created a PBS1 decoy containing a consensus cleavage site for the NIa protease from turnip mosaic virus (PBS1TuMV; fig. S1). This virus was selected because it is more virulent on Arabidopsis than is TEV, but belongs to same family of viruses (Potyviridae) (15). Transient expression assays in N. benthamiana confirmed that PBS1TuMV is cleaved by TuMV NIa protease and activates RPS5 (fig. S1). We generated transgenic Arabidopsis plants expressing PBS1TuMV under the native PBS1 promoter and terminator sequences, and then infected them with a TuMV derivative that contained a fusion between the viral 6K2 protein and green fluorescent protein (GFP) (16). This 6K2-GFP fusion protein enabled us to visualize the spread of the virus through plants by means of ultraviolet light. At 11 days after inoculation, TuMV/6K2-GFP had spread throughout the rosette leaves and newly emerging leaves of wild-type (nontransgenic) Arabidopsis (Fig. 4, A and B). In three PBS1TuMV transgenic Arabidopsis lines, most of the newly emerging leaves became chlorotic and died (Fig. 4, A and B, and fig. S2). This cell death correlated with reduced spread of the GFP fluorescence. To quantify whether the transgenic plants carried less virus, we collected entire rosettes, isolated total protein, and then assessed viral protein quantities by immunoblot (Fig. 4C). Consistent with the observed reduction in GFP fluorescence, we observed a sharp reduction in GFP in transgenic lines that displayed severe necrosis (Fig. 4C). We also analyzed three lines that displayed a moderate necrosis phenotype and one that displayed susceptible mosaic symptoms. The quantity of 6K2-GFP was inversely correlated with the quantity of PBS1TuMV (Fig. 4C), indicating that the level of defense activation is proportional to the level of PBS1TuMV expression. These data indicate that RPS5 can be activated by cleavage of engineered PBS1 by the cognate TuMV NIa protease and that this activation limits virus accumulation, presumably through cell death.

Fig. 4 Transgenic Arabidopsis expressing PBS1 containing a TuMV NIa protease cleavage site accumulate less virus.

TuMV induces trailing necrosis in transgenic Arabidopsis expressing PBS1TuMV. (A) Wild-type and transgenic plants under white light 11 days after agro-infection with TuMV/6K2:GFP. (B) The same plants under UV light. Note the extensive green fluorescence in the wild-type plant, indicating systemic viral infection. The transgenic plant shown is from line #1 and is representative of three independent lines showing severe necrosis. (C) Virus accumulation in PBS1TuMV transgenic plants is inversely correlated with PBS1TuMV expression. Whole rosettes of five to six plants from each line were harvested at 15 days after agro-infection, tissue combined, protein extracted, and virus content compared using an anti-GFP immunoblot. The Rubisco (ribulose-1,5-bisphosphate carboxylase-oxygenase) band is shown as a loading control. PBS1TuMV levels were assessed using an anti-hemagglutinin (HA) immunoblot after anti-HA immunoprecipitation. Lines 1 to 3 were severely necrotic, lines 4 to 6 moderately necrotic, and line 7 fully susceptible.

The cell death phenotype we observed in the PBS1TuMV transgenic plants is similar to a phenomenon previously described in plant-virus interactions called “trailing necrosis” or “lethal systemic necrosis” (17). In severe cases, trailing necrosis can lead to plant death due to destruction of the apical meristem, which is what we observed in our severely necrotic lines (fig. S2). Trailing necrosis is usually associated with partially compromised disease resistance responses, which can be caused by an imperfect match between an NLR protein and the virus being detected (3, 17). We speculate that in our system, the viral protease must attain high levels before it encounters PBS1, which is tethered to the plasma membrane (18). Potyvirus NIa proteases accumulate in the nucleus (NIa stands for “nuclear inclusion antigen”), with only a small portion found in the cytoplasm (19). We thus think it is likely that RPS5 is being activated at a late stage of infection, which could account for the spread of virus to adjacent cells prior to cell death.

To be an effective virus resistance mechanism, the RPS5-PBS1 system will require modifications that enable more rapid activation of RPS5. Given the observed correlation between PBS1 protein expression and levels of resistance (Figs. 2 and 4), increasing the expression level of PBS1 may be sufficient. More likely, however, it will be necessary to relocate RPS5 and PBS1 to sites of viral replication, which would enable early encounters between the viral protease and PBS1. This could be accomplished by replacing the N-terminal acylation motifs of RPS5 and PBS1, which direct PBS1 and RPS5 to the plasma membrane (18, 20), with the 6K2 domain of TuMV, which localizes to viral replication complexes (21).

Although further optimization is needed, the above experiments demonstrate that swapping protease cleavage sites in PBS1 can change the specificity of the RPS5 immune response pathway. This system should thus be usable to engineer resistance to any pathogen that uses a protease as part of its infection process, provided that the protease targets a defined recognition sequence of seven or fewer amino acids and localizes to the host cell cytoplasm early in the infection cycle. To assure durability, the protease should also be essential for infectivity of the pathogen. Pathogens known to secrete or express proteases during host infection include viruses, bacteria, oomycetes, fungi, and nematodes (2227), hence this strategy may be broadly applicable.

Using such a decoy approach to engineer resistance in crop plants might not require transfer of the Arabidopsis RPS5 gene if crop plants already possess the ability to detect AvrPphB by a similar mechanism. Indeed, we have determined that most varieties of soybean display a HR in response to P. syringae expressing AvrPphB (28), as do some varieties of barley (22), indicating that these crops species possess an NLR gene functionally analogous to RPS5. In addition, PBS1 is highly conserved among crop plants, including soybean and barley (29, 30). It may thus be possible to engineer soybean, barley, and other crop plants to detect various pathogen proteases by using genome editing techniques to modify endogenous PBS1 genes. More generally, similar decoy approaches may be applicable to other NLR proteins that use an indirect recognition mechanism to detect pathogen effectors.

Supplementary Materials

Materials and Methods

Figs. S1 and S2

Table S1

References (3134)

References and Notes

Acknowledgments: We thank S. Pottinger for assistance with statistical analyses. T-DNA insertion mutants of Arabidopsis were obtained from the Arabidopsis Biological Resource Center at Ohio State. pCambiaTuMV/6K2:GFP was kindly provided by A. Laliberte. Supported by National Institute of General Medical Sciences grant R01 GM046451, NSF Plant Genome Research Program grant IOS-1339348, and the Indiana University Office of the Vice Provost for Research Faculty Research Support Program. A patent application has been submitted covering the RPS5-PBS1 protease recognition system (U.S. Patent App. 14/427,753). Author contributions: S.H.K., D.Q., T.A., M.H., and R.W.I. designed and performed the experiments; R.W.I. and S.H.K. wrote the manuscript with editing provided by D.Q., T.A., and M.H. Supplement contains additional data.

Stay Connected to Science

Navigate This Article