Abstract
The Xo1 locus in the heirloom rice variety Carolina Gold Select confers resistance to bacterial leaf streak and bacterial blight, caused by Xanthomonas oryzae pvs. oryzicola and oryzae, respectively. Resistance is triggered by pathogen-delivered transcription activator-like effectors (TALEs) independent of their ability to activate transcription, and is suppressed by variants called truncTALEs common among Asian strains. By transformation of the susceptible variety Nipponbare, we show that one of 14 nucleotide-binding, leucine-rich repeat (NLR) protein genes at the locus, with a zfBED domain, is the Xo1 gene. Analyses of published transcriptomes revealed that the Xo1-mediated response is similar to those of NLR resistance genes Pia and Rxo1 and distinct from that associated with induction of the executor resistance gene Xa23, and that a truncTALE dampens or abolishes activation of defense-associated genes by Xo1. In Nicotiana benthamiana leaves, fluorescently-tagged Xo1 protein, like TALEs and truncTALEs, localized to the nucleus. And, endogenous Xo1 specifically co-immunoprecipitated from rice leaves with a pathogen-delivered, epitope-tagged truncTALE. These observations suggest that suppression of Xo1-function by truncTALEs occurs through direct or indirect physical interaction. They further suggest that effector co-immunoprecipitation may be effective for identifying or characterizing other resistance genes.
Bacterial leaf streak of rice, caused by Xanthomonas oryzae pv. oryzicola (Xoc), is an increasing threat to production in many parts of the world, especially in Africa. Bacterial blight of rice, caused by X. oryzae pv. oryzae (Xoo) has long been a major constraint in Asia and is becoming prevalent in Africa. The purified American heirloom rice variety Carolina Gold Select (hereafter Carolina Gold; McClung and Fjellstrom, 2010) is resistant to all tested African strains of Xoc and some tested strains of Xoo (Read et al., 2016). Using an African strain of Xoc, the resistance was mapped to chromosome 4 and designated as Xo1 (Triplett et al., 2016). Both Xoc and Xoo deploy multiple type III-secreted transcription activator-like effectors (TALEs) during infection. TALEs enter the plant nucleus and bind to promoters, each with different sequence specificity, to transcriptionally activate effector-specific target genes (Perez-Quintero and Szurek, 2019). Some of these genes, called susceptibility genes, contribute to disease development (Hutin et al., 2015). In some host genotypes, a TALE may activate a so-called executor resistance gene, leading to host cell death that stops the infection (Bogdanove et al., 2010). Most of the cloned resistance genes for bacterial blight are in fact executor genes (Zhang et al., 2015). Xo1 is different. It mediates resistance in response to TALEs non-specifically, independent of their ability to activate transcription (Triplett et al., 2016). Also, unlike executor genes, Xo1 function is suppressed by a variant class of these effectors known as truncTALEs (also called iTALEs), which nuclear localize (Ji et al., 2016) but do not bind DNA (Read et al., 2016).
Xo1 maps to a region that in the reference rice genome (cv. Nipponbare) contains seven nucleotide-binding, leucine-rich repeat protein genes (“NLR” genes) (Triplett et al., 2016). NLR genes are the largest class of plant disease resistance genes. NLR proteins recognize specific, corresponding pathogen effector proteins typically through direct or indirect protein-protein interactions, and mediate downstream defense signaling that leads to expression of defense genes and a programmed localized cell death, the hypersensitive reaction (HR) (Lolle et al., 2020). Recently, by whole genome sequencing, we determined that the Xo1 locus in Carolina Gold comprises 14 NLR genes. We identified one of these, Xo111, as a strong candidate based on its structural similarity to the previously cloned and only known NLR resistance gene for bacterial blight, Xa1 (Read et al., 2020). Xa1, originally identified in the rice variety Kogyoku, maps to the same location (Yoshimura et al., 1998) and behaves similarly to Xo1: it responds to TALEs non-specifically (and thus confers resistance also to bacterial leaf streak), and its activity is suppressed by truncTALEs (Ji et al., 2016). Xo111 and Xa1 are members of a small subfamily of NLR genes that encode an unusual N-terminal domain comprising a zinc finger BED motif (Read et al., 2020).
To ascertain whether Xo111 is the gene responsible for Xo1 resistance, we generated transgenic Nipponbare plants expressing it. We amplified the genomic Xo111 coding sequence (5,882 bp) as well as the 993 bp promoter region upstream of the start codon and cloned them together into a binary vector with a 35S terminator to generate plasmid pAR902. Agrobacterium tumefaciens strain EHA101 carrying pAR902 was used for the transformation, which was performed by the Cornell University Plant Transformation Facility. After rooting, regenerants from two independent events were moved to soil and maintained in a growth chamber. These T0 plants were inoculated by syringe infiltration with African Xoc strain CFBP7331, which has no truncTALE of its own, carrying either an empty vector (EV) or the plasmid-borne truncTALE gene tal2h (p2h) from the Asian Xoc strain BLS256 (Read et al., 2016). Inoculum was confirmed on untransformed Nipponbare and Carolina Gold plants (Fig. S1). Plants from both Xo1 events displayed resistance to the strain with the EV, manifesting as HR (necrosis) and lack of water-soaking, and this was suppressed by Tal2h (Fig. 1), demonstrating that Xo111 is the Xo1 gene.
NLR protein activation is characteristically followed by a suite of responses that includes massive transcriptional reprogramming leading both to HR and to activation of a large number of defense-associated genes (Cui et al., 2015). To gain insight into the nature of Xo1-mediated resistance, we compared the global profile of differentially expressed genes during Xo1-mediated defense to those of two other NLR genes in rice, and to the profile associated with an executor gene. We used our previously reported RNAseq data from Carolina Gold plants inoculated with CFBP7331(EV) or mock inoculum (Read et al., 2020), data for the NLR gene Pia for resistance to the rice blast pathogen Magnaporthe oryzae (Tanabe et al., 2014), data from rice resistant to bacterial leaf streak due to transgenic expression of the maize NLR gene Rxo1 (Xie et al., 2007; Zhou et al., 2010), and data for the transcriptomic response associated with induction of the executor resistance gene Xa23 by an Xoo strain with the corresponding TALE (Tariq et al., 2018). Differentially expressed genes (log2-fold change >1 or <-1; p-value >0.05) in the comparison between pathogen-inoculated and mock-inoculated plants were compared across the four datasets. The total number of DEGs ranged from 10,050 for Xo1 to 628 for Xa23 (Fig. 2A, Table S1). For each resistance gene, there were a number of DEGs found only in the pathogen to mock comparison for that dataset, and this was highest for Xo1 (7,121 genes) (Fig. 2A, Table S1). These DEGs may be specific to the rice-genotype and pathogen combinations assayed, or they may be due to differences in the expression assay (RNAseq vs. microarray), annotation, or timepoints used. Overall, the DEG profile for Xo1 was more similar to those of Pia and Rxo1 than to the profile for Xa23 (Fig. 2A). This was even more apparent when the expression of 340 rice genes associated with plant defense response (gene ontology group 0006952) was examined (Fig. 2B). The Xo1 profile comprised the largest number of plant defense DEGs (99), and these included 16 of 26 total defense DEGs for Rxo1, 26 of 46 for Pia, and 8 of only 14 for Xa23 (Fig. 2B and Table S2).
We also compared DEGs relative to mock in Carolina Gold plants inoculated with CFBP7331(EV) and Carolina Gold plants inoculated with CFBP7331(p2h) (Read et al., 2020), to gain insight into how Xo1-mediated resistance is overcome by a pathogen delivering a truncTALE. In contrast to the 99 defense response genes differentially expressed in response to CFBP7331(EV), only 18 defense genes were differentially expressed in response to CFBP7331(p2h) (Fig. 2C). Of these 18 genes, 7 were differentially expressed only in the response to the strain with tal2h, 4 up and 3 down. Of the remaining 11, 3 were up and 2 were down in both responses, but each less so in the response to the strain with tal2h. The other 6 moved in opposite directions entirely, up in the absence but repressed in the presence of tal2h, relative to mock. This expression profile during suppression of Xo1-mediated resistance is consistent with Tal2h functioning early in the defense cascade. Interestingly, the Xoc susceptibility gene OsSULTR3;6 is strongly induced by both CFBP7331(EV) and CFBP7331(p2h), indicating that TALE function is not compromised by Xo1 or by Tal2h.
The observation that Xo1 reprograms transcription in a manner consistent with other rice NLR proteins upon recognition of the cognate pathogen effector and that reprogramming by Xo1 is essentially blocked by Tal2h led us to explore whether Xo1 localizes to the same subcellular location as TALEs and truncTALEs. Some, but not all, NLR proteins nuclear localize (Shen et al., 2007; Wirthmueller et al., 2007; Caplan et al., 2008; Cheng et al., 2009), and we previously identified putative nuclear localization signals (NLSs) in Xo111 (Read et al., 2020). We generated expression constructs for a green fluorescent protein (GFP) fusion to the N-terminus of Xo1 as well as an N-terminal monomeric red fluorescent protein (mRFP) fusion both to a TALE (Tal1c of Xoc BLS256) and to Tal2h. These constructs were delivered into Nicotiana benthamiana leaves using A. tumefaciens strain GV3101, and the leaves imaged with a Zeiss 710 confocal microscope (Fig. 3). GFP-Xo1 in the absence of either effector but with free mRFP localized to foci that appeared to be nuclei. Co-expression with mRFP-Tal1c or with mRFP-Tal2h confirmed that these foci were nuclei.
The localization of Xo1, the TALE, and the truncTALE to the nucleus when transiently expressed in N. benthamiana led us to pursue the hypothesis that Xo1 physically interacts with one or both of these proteins in the native context. We generated plasmid constructs that add a 3x FLAG tag to the C-terminus of Tal1c or Tal2h (Tal1c-FLAG and Tal2h-FLAG) and introduced them individually into the TALE-deficient X. oryzae strain X11-5A (Triplett et al., 2011) for co-immunoprecipitation from inoculated Carolina Gold leaves (Fig. 4). We included also a plasmid for expression of a second, untagged TALE (Tal3c from BLS256) and a plasmid for untagged Tal2h. By pairing the X11-5A transformants with each other or with the untransformed control strain, we were able to probe for Carolina Gold proteins interacting with the tagged TALE or truncTALE, and for interactions of these proteins with each other or with the second TALE. Select combinations were inoculated to Nipponbare leaves for comparison. Inoculation was done by syringe infiltration, in 30-40 contiguous spots on each side of the leaf midrib. For each co-inoculation, tissue was harvested at 48 hours and ground in liquid N2, then soluble extract was incubated with anti-FLAG agarose beads and washed to immunopurify the tagged and interacting proteins. Immunoprecipitates were eluted, and an aliquot of each was subjected to western blotting with anti-TALE antibody (Fig. S2). The remainders were then resolved on a 4-20% SDS-PAGE and eluates from gel slices containing proteins between approximately 60 and 300 kDa (Fig. S3) were digested and the peptides analyzed by mass spectrometry. Proteins were considered present in a sample if at least three peptides mapped uniquely to any of the pertinent annotated genomes searched: the X. oryzae strain X11-5A genome (Triplett et al., 2011) plus the TALE(s) or TruncTALE being expressed, the Nipponbare genome (MSU 7; Kawahara et al., 2013), and the Carolina Gold genome (Read et al., 2020). For the Carolina Gold genome, we re-annotated using the RNAseq data from CFBP7331(EV), CFBP7331(p2h), and mock-inoculated plants cited earlier. We carried out the experiment twice.
In the western blot for each experiment (Fig. S2), we detected the tagged TALE or truncTALE in each corresponding sample, with the exception of a Tal1c-FLAG/Tal3c/Nipponbare sample in the first experiment. No Tal3c or untagged Tal2h was detected in any sample. The mass spectrometry confirmed these observations, suggesting that neither TALEs with truncTALEs nor TALEs with other TALEs interact appreciably (Fig. 4). Xo1 was consistently detected in the Carolina Gold/Tal2h-FLAG samples, irrespective of any co-delivered Tal1c or Tal3c, and not in any other sample (Fig. 4). No other protein consistently co-purified with Tal2h-FLAG or Tal1c-FLAG in either Carolina Gold or Nipponbare samples (Dataset S1).
In summary, we have shown that 1) an NLR protein gene at the Xo1 locus, harboring an integrated zfBED domain, is Xo1; 2) the Xo1-mediated response is similar to those mediated by two other NLR resistance genes and distinct from that associated with TALE-specific transcriptional activation of an executor resistance gene; 3) a truncTALE abolishes or dampens activation of defense-associated genes by Xo1; 4) the Xo1 protein, like TALEs and truncTALEs, localizes to the nucleus, and 5) Xo1 specifically co-immunoprecipitates from rice leaves with a pathogen-delivered, epitope-tagged truncTALE. Thus, Xo1 is an allele or paralog of Xa1, and suppression of Xo1 function by a truncTALE is likely the result of physical interaction between the resistance protein and the effector.
Whether the interaction is direct or indirect is not certain, but that fact that no other protein was detected consistently that co-immunoprecipitated with Tal2h and Xo1 suggests the interaction is direct. It is tempting to speculate that TALEs trigger Xo1-mediated resistance also by direct interaction with the protein and that truncTALEs function by disrupting the association. This is consistent with the results of our comparative analysis of the Xo1 DEG profile during TALE-triggered HR, which showed it to be a typical NLR protein-mediated response and thus plausibly the result of direct interaction with the TALE. And it is consistent with the Xo1 DEG profile during suppression by Tal2h, which suggested that Tal2h functions early in the defense cascade, perhaps by blocking TALE recognition. While tagged Tal1c did not detectably pull down Xo1, it is possible that the interaction might be weak, or transient, or that any complex of the proteins in the plant cells had begun to degrade with the developing HR at the 48 hour time point sampled. An alternative hypothesis is that Tal2h interacts with TALEs and masks them from the resistance protein, but both our co-immunoprecipitation results and the fact that Tal2h did not impact TALE activation of the OsSULTR3;6 susceptibility suggest that this is not the case.
The results presented constitute an important step toward understanding how Xo1 works, and how its function can be suppressed by the pathogen. An immediate next step might be to determine the portion(s) of Xo1 involved in its interaction with Tal2h. Our previous comparison of the motifs present in Xo111, Xa1, and the closest Nipponbare homolog (Nb-xo15, which is expressed) revealed that the zfBED and CC domains are identical and the NB-ARC domains nearly so (Read et al., 2020). In contrast, the leucine rich repeat domain of Nb-xo15 differs markedly from those of Xo1 and Xa1, which, with the exception of an additional repeat in Xa1, are very similar. Thus, the LRR may be the determinative interacting domain. Supporting this hypothesis, differences in the LRR determine the pathogen race specificities of some flax rust resistance genes (Ellis et al., 1999). More broadly, the ability of tagged Tal2h to pull down Xo1 suggests that effector co-immunoprecipitation may be an effective approach to characterizing pathogen recognition mechanisms of other resistance proteins, or for identifying a resistance gene de novo.
Author contributions
AR, MH, FR, and AB conceived and designed the study; AR, MH, and FR carried out the experiments; AR, MH, FR, MM, and AB analyzed data; AR, MH, and AB wrote the manuscript.
Supplemental files
1. Supplemental text and figures
Materials and methods
Fig. S1. Confirmation of CFBP7331(EV) and CFBP(p2h) inoculum on
Nipponbare and Carolina Gold plants
Fig. S2. Western blot of immunoprecipitates using anti-TALE antibody
Fig. S3. SDS-PAGE of immunoprecipitates and size range excised for mass spectrometry
Supplemental references
2. Supplemental tables
Table S1. DEGs in Fig. 2A (all DEGS)
Table S2. DEGs in Fig. 2B (GO:0006952 DEGs)
Table S3. DEGs in Fig. 2C (GO:0006952 in disease)
3. Dataset S1
Mass spectrometry data
Funding
National Science Foundation (IOS-1444511 to AB)
National Institute of Food and Agriculture, U.S. Department of Agriculture (2018-67011-28025 to AR)
Gatsby Charitable Foundation (to MM)
Acknowledgments
The authors thank M. Carter and B. Szurek for critical reading of the manuscript, Matthew Willmann and the Plant Transformation Facility of Cornell’s School of Integrative Plant Science for carrying out the rice transformation, Sandra Harrington and Susan McCouch for assistance growing the regenerants, and Ruchika Bhawal and Elizabeth Anderson at the Proteomics Facility of the Biotechnology Resource Center at the Cornell University’s Institute of Biotechnology (BRC) for conducting the mass spectrometry. Confocal microscopy was carried out at the BRC’s Imaging Facility. This work was supported by the Plant Genome Research Program of the National Science Foundation (IOS-1444511 to AB), the National Institute of Food and Agriculture of the U.S. Department of Agriculture (2018-67011-28025 to AR), and the Gatsby Charitable Foundation (to MM). We also acknowledge support from the National Institutes of Health to the Proteomics Facility for the Orbitrap Fusion mass spectrometer (shared instrumentation grant 1S10 OD017992-01) and to the Imaging Facility for the Zeiss LSM 710 confocal microscope (shared instrumentation grant S10RR025502).