Indica rice (O. sativa ssp. indica) IR24 and IRBB5 are near-isogenic lines. IR24 carries OsTFIIAγ5 gene and is susceptible to Xoo and Xoc. IRBB5, with the genetic background of IR24, carries recessive disease resistance gene xa5 encoding a mutated OsTFIIAγ5 (OsTFIIAγ5^{V 39E}) and is resistant to Xoo and Xoc. Plants were grown during a normal rice-growing season under natural field conditions. The wild Oryza species used in this study were from our collection (Yuan et al., 2014). Sweet orange (Citrus sinensis L. Osbeck cv. Newhall) plants were grown in a greenhouse at 25 to 30°C. Pepper (Capsicum annuum L. cv. Hua 50) and tomato (Solanum lycopersicum L. cv. Ailsa Craig) plants were germinated and grown in a growth chamber under a photosynthetic photo flux density of approximately 120 μmol photons m^-2 s^-1 with a 16-h light/8-h dark photoperiod cycle at 25°C and a relative humidity of approximately 60%.

The Philippine Xoo strains PXO99 and PXO341 have commonly been used in studies of rice resistance to bacterial blight disease (Li et al., 2012). The Xoc strain RH3 is also frequently used in research (Yuan et al., 2016). Xcc strain X02-007 (Fu et al., 2012) and Xcv strain 23-1 (Sun et al., 1999) were also previously used in research. All the Xanthomonas strains were grown at 28°C on nutrient agar medium. When genetic manipulation of bacteria was undertaken, antibiotics were used at the following final concentrations as required: ampicillin at 100 μg ml^-1, rifampicin at 75 μg ml^-1, and kanamycin at 25 μg ml^-1.

For stable transformation of rice, Agrobacterium-mediated transformation was performed using calli derived from mature embryos of indica rice line IRBB5 according to previously reported methods (Ge et al., 2006). For construction of P_OsTFIIAγ5:OsTFIIAγ5 derivative vectors, the 2055-bp promoter, 321-bp full-length cDNA and 1229-bp terminator of OsTFIIAγ5 were amplified with primers listed in Supplementary Table 1 and inserted into vector pCAMBIA1301 in order. For construction of P_Ubi:OsTFIIAγ5 derivative vectors, 321-bp full-length cDNA of OsTFIIAγ5 and its derivatives were amplified with primers listed in Supplementary Table 1 and inserted into vector pU1301 (Yuan et al., 2010).

For transiently suppressing the target gene in sweet orange, Xcc-facilitated agroinfiltration was performed. In brief, a 162-bp cDNA fragment of CsTFIIAγ was amplified with primers listed in Supplementary Table 1 and inserted into vector pDS1301, and the construct was then transformed into Agrobacterium tumefaciens strain GV3101 (Yuan et al., 2010). Sweet orange young leaves were inoculated with Xcc [5 × 10⁸ colony-forming unit (cfu)/ml] by the penetration method using a needleless syringe. Eight hours later, the same inoculated leaf areas were subjected to agroinfiltration with Agrobacterium carrying the recombinant construct (Hu et al., 2014; Jia and Wang, 2014). The Xcc-facilitated agroinfiltration plants were grown in a greenhouse at 25 to 30°C with a 16-h light/8-h dark photoperiod cycle.

For transiently suppressing target gene in pepper and tomato, virus-induced gene silencing (VIGS) was performed. In brief, the DNA segment of the target gene or green fluorescence protein (GFP) gene was inserted into the tobacco rattle virus (TRV) vector pTRV2 as described previously (Liu et al., 2002a). The recombinant vectors were introduced into GV3101. Agrobacterium culture was infiltrated into the cotyledons of germinating pepper plants or the first two true leaves of the four-leaf stage tomato plants using a 1-ml needleless syringe (Liu et al., 2002b; Chung et al., 2004). The Agrobacterium-infiltrated plants were first kept in a growth chamber at 16°C for 1 day and then grown in a chamber at 25°C with a 16-h light/8-h dark photoperiod cycle.

To study the interaction between TALE and host proteins in yeast two-hybrid assays (Yuan et al., 2010), the TFB motif of TALE genes and plant TFIIAγ genes were amplified using PCR primers listed in Supplementary Table 1. The amplified bacterial DNA segments were ligated into pGBKT7 vector, and the amplified plant DNA segments were ligated into pGADT7 Rec vector. The pGBKT7 and pGADT7 plasmids were then transformed into yeast strain AH109 (Yuan et al., 2010). The yeast clones were scribed on the synthetic defined premixes (SD) medium lacking leucine (L) and tryptophan (W) (-LW) and selective SD medium lacking L, W, histidine (H), and adenine (A) (-LWHA). The interactions of examined proteins were assessed by growth of yeast cells on selective medium and by examination of β-D-galactopyranoside (X-α-gal) activity and β-galactosidase (LacZ) activity as described previously (Yuan et al., 2010).

To study the interaction between TALE and plant proteins in planta, co-immunoprecipitation (CoIP) assays were performed (Yuan et al., 2016). The DNA segments of TFB motifs of TALEs were ligated into the pU1031-9myc vector, and the DNA segments of plant genes were ligated into the pU1301-3FLAG vector (Yuan et al., 2010). The recombinant vectors were introduced into GV3101. Agrobacterium-mediated transformation was performed by infiltrating into Nicotiana benthamiana leaves using a needleless syringe. CoIP assays were conducted using anti-FLAG antibody (Sigma-Aldrich, St. Louis, MO, United States) and anti-myc antibody (Tiangen, Beijing, China) as described previously (Yuan et al., 2016). Each CoIP assay was repeated at least twice. The original western blotting images are provided in Supplementary Figure 13.

Mutation of plant genes was performed using the GeneTailor Site-Directed Mutagenesis System (Invitrogen Life Technologies, Carlsbad, CA, United States) as described previously (Yuan et al., 2011). The mutagenic primers are listed in Supplementary Table 1.

To evaluate rice bacterial blight disease, rice plants were inoculated with Xoo by the leaf-clipping method at the booting (panicle development) stage (Chen et al., 2002). Disease was scored by measuring the lesion length at 14 days after inoculation.

To evaluate rice bacterial streak disease, rice plants were inoculated with Xoc strains by the needle stab method at the tillering stage (Yuan et al., 2016). The disease was scored by measuring the lesion length at 14 days after inoculation.

To evaluate citrus bacterial canker disease, sweet orange leaves were inoculated with Xcc at a concentration of 5 × 10⁸ cfu/ml in 10 mM MgCl₂ using a needleless syringe. The bacterial growth rate in sweet orange leaves was measured by counting the cfu as described previously (Hu et al., 2014).

To evaluate bacterial spot disease of pepper and tomato, 4-week-old VIGS-treated plants were sprayed with Xcv at a concentration of 5 × 10⁵ cfu/ml in 10 mM MgCl₂. The bacterial growth rate in pepper or tomato leaves was measured by counting the cfu as described previously (Bonas et al., 1991).

For gene expression analysis, 2-cm rice leaf fragments near the bacterial infection sites or citrus, pepper and tomato leaf tissues next to the infiltration sites were collected for RNA isolation. Quantitative reverse transcription-PCR (qRT-PCR) was conducted using gene-specific primers (Supplementary Table 2) as described previously (Yuan et al., 2010). The expression level of rice, pepper or tomato actin gene or citrus EF1a gene was used to standardize the RNA sample of rice, pepper, tomato or citrus, respectively. The expression level relative to that of controls was assessed. Each qRT-PCR assay was repeated at least twice with similar result, with each repetition having three replicates.

Differences between samples were analyzed for statistical significance by using the SPSS software and the Student’s t-test (two tailed).

Xcc causes citrus canker. Xcc strain X02-007 carries TALE genes, which contained TFB motifs of TALEs that showed high sequence similarity with the motifs of Xoo and Xoc TALEs (Supplementary Figure 1; Yuan et al., 2016). The TFB motifs of Xcc TALEs interacted with sweet orange TFIIAγ (CsTFIIAγ) both in yeast and in planta (Figures 1A,B). To investigate whether CsTFIIAγ is required for TALE-dependent induction of the host S gene during Xcc invasion, we suppressed CsTFIIAγ gene expression in the leaves of sweet orange by RNA interference (RNAi). The plants with suppressed expression of CsTFIIAγ appeared markedly less diseased, with less cell division and proliferation or hyperplasia after Xcc inoculation (Figures 1C–E) and significantly lower (P < 0.01) Xcc growth rate compared to control plants (Figure 1F). Citrus CsLOB1 is a S gene to Xcc (Hu et al., 2014; Li et al., 2014). Consistent with reduced disease symptoms, the CsTFIIAγ-RNAi leaves showed significantly suppressed Xcc-induced expression of CsLOB1 compared with control (Figure 1G).

Further, different citrus species, such as hongkong kumquat, grapefruit and trifoliate orange, have a TFIIAγ protein sequence identical to CsTFIIAγ in sweet orange, although the gene sequences differ for a few nucleotides (Supplementary Figure 2A). All the citrus species were similarly susceptible to Xcc (Supplementary Figure 2B). These results suggest that, like OsTFIIAγ5, CsTFIIAγ is also hijacked by TALE-carrying bacteria, with the host becoming infected via TALE-induced expression of the host S gene.

Rice plants carrying recessive xa5, which encodes the mutated OsTFIIAγ5^{V 39E}, have broad-spectrum resistance to TALE-carrying bacteria and no effect on rice yield and development (Yuan et al., 2016). To analyze whether OsTFIIAγ5^{V 39E} is the only type of mutation that confers disease resistance in rice and whether mutations similar to OsTFIIAγ5^{V 39E} can confer disease resistance in other plant species, we first examined the sequences of OsTFIIAγ5 coding region in different Oryza species.

The Oryza genus consists of two cultivated rice species, Asian cultivated rice (O. sativa, AA genome) and African cultivated rice (O. glaberrima, AA genome), and wild (undomesticated) rice species (AA, BB, CC, BBCC, CCDD, EE, FF, GG, HHJJ, KKLL genomes) (Ge et al., 1999; Ammiraju et al., 2010). We sequenced the OsTFIIAγ5 locus in 15 different Oryza genomes, including nine accessions of AA genome (O. sativa indica, O. sativa japonica, O. rufipogon, O. nivara, O. glaberrima, O. longistaminata, O. meridionalis, O. gluaepatula, and O. barthii), one accession of BB genome (O. punctata), two accessions of CC genome (O. officinalis and O. rhizomatis), one accession of EE genome (O. australiensis), one accession of FF genome (O. brachyantha), and one accession of GG genome (O. meyeriana). Sequence alignment analysis showed that different Oryza species contained the same sequences of the OsTFIIAγ5 coding region and encoded the identical OsTFIIAγ5 protein, which is consistent with its molecular function as a basal transcription factor.

We searched the putative OsTFIIAγ5 variant from database OryzaGenome¹, which contains genotype information of 463 accessions of wild rice O. rufipogon, which is evolutionally closely related to cultivated rice (Ohyanagi et al., 2016). Thirty-four single nucleotide polymorphisms (SNPs) were distributed in the OsTFIIAγ5 gene region, all within the introns and none in the exons. Thus the 463 O. rufipogon accessions encode the identical OsTFIIAγ5 protein. Moreover, we also searched the putative OsTFIIAγ5 variants from Rice SNP-Seek Database², which contain genotype information of 3024 O. sativa varieties from the International Rice GenBank Collection Information System (IRGCIS³). There were 93 SNPs distributed in the 6260-bp region of OsTFIIAγ5 gene, and 105 rice varieties contained SNPs in the coding region causing amino acid mutation. Of these 105 rice varieties, 104 have a mutated OsTFIIAγ5 gene encoding OsTFIIAγ5^{V 39E}, identical to the protein encoded by recessive disease resistance gene xa5. One variety has a different type of mutated OsTFIIAγ5 gene, which encodes a protein with valine changed to aspartic acid (D) at the 39th residue (OsTFIIAγ5^{V 39D}) (Supplementary Table 3). Eighty-five of the 104 rice varieties carrying the mutated OsTFIIAγ5^{V 39E} gene and one carrying the mutated OsTFIIAγ5^{V 39D} gene belong to the Aus group, which is mainly from South Asia, and the other 19 carrying the mutated OsTFIIAγ5^{V 39E} gene belong to Indica II, Indica IB, Indx and Admix groups, mainly from Southeast Asia (Huang et al., 2011). The distribution of OsTFIIAγ5^{V 39E} resistance allele in these groups likely reflects the high disease pressure in these regions, where Xoo is a devastating disease. The results suggest that the OsTFIIAγ5 gene mutation probably appeared in cultivated rice.

Amino acid sequence alignment of TFIIAγs from different species showed that they share a high sequence similarity, with the 39th amino acid residue having three types, V, leucine (L), and E (Yuan et al., 2016). The V type occurs in plants (O. sativa, Arabidopsis thaliana, Citrus sinensis, Capsicum annuum, and Solanum lycopersicum). The L type occurs in animals (Rattus norvegicus, Homo sapiens, Danio rerio, and Drosophila melanogaster). The E type was only detected in rice OsTFIIAγ5 locus (OsTFIIAγ5^{V 39E}), which is recessively inherited. In addition, we identified a D type in OsTFIIAγ5 locus (OsTFIIAγ5^{V 39D}) from rice germplasm (Supplementary Table 3). To determine whether the 39th amino acid of OsTFIIAγ5 is vital for TALE-dependent induction of rice genes, we first produced four OsTFIIAγ5 derivatives at the 39th residue, substituting V with alanine (A), D, L or glutamine (Q); V, A and L are hydrophobic amino acids, and D and Q are acidic hydrophilic amino acids. The TFB motifs of TALE PthXo1 from Xoo and TALE Tal3c from Xoc interacted as strongly with mutants OsTFIIAγ5^{V 39A}, OsTFIIAγ5^{V 39D}, OsTFIIAγ5^{V 39L}, and OsTFIIAγ5^{V 39Q} as with wild-type OsTFIIAγ5 in yeast cells (Supplementary Figure 3). The PthXo1 TFB interacted weakly and Tal3c TFB did not interact with OsTFIIAγ5^{V 39E} (E being an acidic hydrophilic amino acid) as reported previously (Yuan et al., 2016).

We then generated OsTFIIAγ5 transgenic plants with the four different amino acid substitutions (A, D, L, and Q) at the 39th residue driven by the native promoter of OsTFIIAγ5 (P_OsTFIIAγ5) in the xa5 (OsTFIIAγ5^{V 39E}) resistance allele background (IRBB5). The P_OsTFIIAγ5:OsTFIIAγ5^{V 39A}, P_OsTFIIAγ5:OsTFIIAγ5^{V 39D}, P_OsTFIIAγ5:OsTFIIAγ5^{V 39L}, and P_OsTFIIAγ5:OsTFIIAγ5^{V 39Q} plants showed susceptibility to Xoo strains PXO99 and PXO341, also susceptibility to Xoc strain RH3 compared to the wild-type plant (Figure 2A). The susceptibility of these transgenic plants to Xoo was associated with more efficiently induced expression of known rice S genes (Xa13/Os8N3/SWEET11, OsTFIIAγ1, and OsTFX1) to Xoo (Figure 2B; Chu et al., 2006; Sugio et al., 2007). The susceptibility of these transgenic plants to Xoc was associated with more efficiently induced expression of the known rice S gene (OsSULTR3;6) to Xoc (Figure 2B; Cernadas et al., 2014). We also generated OsTFIIAγ5 overexpressing plants with the same mutations listed above driven by maize ubiquitin promoter (P_Ubi). These transgenic plants also showed increased susceptibility to Xoo and Xoc, which was associated with increased induced expression of rice susceptibility genes, compared to wild-type plants (Supplementary Figures 4, 5). These results suggest that the 39th amino acid residue of OsTFIIAγ5 is essential for TALE-dependent induction of rice S genes, and it seems that only the V to E mutation at this residue (OsTFIIAγ5^{V 39E}) affects the interaction between OsTFIIAγ5 and TFB of TALEs.

The 39th amino acid residue of citrus CsTFIIAγ is also V, and CsTFIIAγ and OsTFIIAγ5 share 93/89% sequence similarity/identity (Supplementary Figure 6; Yuan et al., 2016). Furthermore, the TFB motifs of Xcc TALEs have high sequence homology with TFBs of Xoo and Xoc TALEs (Supplementary Figure 1). To investigate whether the OsTFIIAγ5^{V 39E} type mutation in CsTFIIAγ (CsTFIIAγ^{V 39E}) could prevent TALE-carrying bacteria from infecting plants, we first examined the interactions between the TALE TFB motifs of different Xanthomonas bacteria and CsTFIIAγ^{V 39E}. The TFB motifs of Xoo, Xoc, and Xcc TALEs could not interact with CsTFIIAγ^{V 39E} (Supplementary Figures 7A, 8A), although these TFBs interacted strongly with both OsTFIIAγ5 and CsTFIIAγ in yeast cells and in planta (Figures 1, 3 and Supplementary Figure 7B). The only exception was the TFB motif of Xoo Tal7a. It interacted with CsTFIIAγ^{V 39E} (Supplementary Figure 7A), which is consistent with this TFB also interacting with OsTFIIAγ5^{V 39E} (Yuan et al., 2016).

Furthermore, we overexpressed CsTFIIAγ in rice IRBB5 carrying the recessive xa5 gene. The transgenic plants displayed susceptibility to Xoo strains PXO99 and PXO341 (Figure 4A). The susceptibility was associated with overexpression of CsTFIIAγ, which was further confirmed in two independent T1 families (Supplementary Figure 9). The induced expression of known S genes Xa13/Os8N3/SWEET11, OsTFIIAγ1, and OsTFX1, each of which is targeted by a different TALE, was significantly higher (P < 0.01) in transgenic plants than in wild-type plants (Figure 4B). Similarly, overexpressing CsTFIIAγ in rice increased susceptibility to Xoc, which was associated with significantly promoted Xoc-induced expression of rice S gene OsSULTR3;6 compared to wild-type plants (Supplementary Figure 10). However, the transgenic plants with overexpression of CsTFIIAγ^{V 39E} showed similar lesion length compared to wild-type plants upon inoculation with Xoo strain PXO99 (Supplementary Figure 11). These results suggest that CsTFIIAγ can also facilitate the infection of Xoo and Xoc in rice by helping TALE-induced activation of rice S genes, and the 39th amino acid residue of CsTFIIAγ is also essential for TALE-dependent induction of host S genes.

We also examined the relationship between pepper (Capsicum annuum L. cv. Hua 50) CaTFIIAγ or tomato (Solanum lycopersicum L. cv. Ailsa Craig) SlTFIIAγ and Xcv, which causes bacterial spot disease in the two crops. CaTFIIAγ and SlTFIIAγ have identical sequences, and Xcv strain 23-1 carries TALE genes (Yuan et al., 2016). The TFB motifs of Xcv 23-1 TALEs, which featured high sequence similarity with the TFB motifs of Xoo, Xoc, and Xcc TALEs (Supplementary Figure 1), interacted with CaTFIIAγ/SlTFIIAγ both in yeast and in planta (Figures 5A,B), but they did not interact with Ca/SlTFIIAγ^{V 39E} (Supplementary Figures 8B, 12). Transiently suppressing CaTFIIAγ or SlTFIIAγ by VIGS enhanced the resistance of pepper and tomato to Xcv. The pepper plants with suppressed expression of CaTFIIAγ (TRV2:CaTFIIAγ) and tomato plants with suppressed expression of SlTFIIAγ (TRV2:SlTFIIAγ) showed markedly reduced disease symptoms and significantly lower (P < 0.01) Xcv growth rates compared to corresponding control plants expressing TRV2:GFP (Figures 5C,D). These results suggest that TALE-containing Xcv causes diseases in pepper and tomato and may also need the help of host TFIIAγs.