TALE‐carrying bacterial pathogens trap host nuclear import receptors for facilitation of infection of rice

Summary Many plant‐pathogenic Xanthomonas rely on the secretion of virulence transcription activator‐like effector (TALE) proteins into plant cells to activate plant susceptibility genes to cause disease. The process is dependent on the binding of TALEs to specific elements of host target gene promoters in the plant nucleus. However, it is unclear how TALEs, after injection into host cells, are transferred from the plant cytoplasm into the plant nucleus, which is the key step of successful pathogen infection. Here, we show that the host plant cytoplasm/nuclear shuttle proteins OsImpα1a and OsImpα1b are key components for infection by the TALE‐carrying bacterial pathogens Xanthomonas oryzae pv. oryzae (Xoo) and Xanthomonas oryzae pv. oryzicola (Xoc), the causal agents of bacterial leaf blight and bacterial leaf streak, respectively, in rice. Direct interaction between the second nuclear localization signal of TALEs of Xoo or Xoc and OsImpα1a or OsImpα1b is required for the transportation of TALEs into the nucleus. Conversely, suppression of the expression of OsImpα1a and OsImpα1b genes attenuates the shuttling of TALEs from the cytoplasm into the nucleus and the induction of susceptibility genes, thus improving the broad‐spectrum disease resistance of rice to Xoo and Xoc. These results provide an applicable strategy for the improvement of resistance to TALE‐carrying pathogens in rice by moderate suppression of the expression of plant nuclear import receptor proteins.


I N T RO D U C T I O N
Xanthomonas, which is a large group of Gram-negative bacterial plant pathogens, consists of almost 30 species and causes diseases on at least 124 monocotyledonous and 268 dicotyledonous plants (Ryan et al., 2011). Many Xanthomonas bacteria secrete transcription activator-like effectors (TALEs), the major virulence and determinant factors in host-pathogen interactions, into plant cells, where they act as transcriptional activators to reprogram the host plant transcriptome. TALEs function as eukaryotic transcription factors, which activate the transcription of host susceptibility genes to cause susceptibility of plants or induce the expression of host resistance genes to cause resistance of plants, and rely on their injection into the plant nucleus and their binding to the effector binding element (EBE) of host target gene promoters (Cox et al., 2017;Zhang et al., 2015). TALEs are highly conserved, with homologues across Xanthomonas species sharing greater than 90% amino acid identity and almost identical structural features. In general, TALEs typically consist of the following: an N-terminal type III secretion signal that guides the translocation of TALEs from the bacterium into the host plant cytoplasm through the type III secretion system (T3SS); a central repeat region (RR), which is the main hallmark of TALEs, that specifically binds to EBE of the host gene promoter; a transcription factor binding (TFB) motif that hijacks the host plant basal transcription factor IIA γ subunit to fulfil the binding of TALEs to EBE; three short nuclear localization signals (NLSs) that guide the translocation of TALEs from the plant cytoplasm into the plant nucleus; and a highly conserved acidic activation domain (AD) that allows TALEs to activate gene transcription in plant cells (Doyle et al., 2013;Huang et al., 2017;Yuan et al., 2016;Zhang et al., 2015).
Xanthomonas oryzae pv. oryzae (Xoo) causes bacterial leaf blight of rice and Xanthomonas oryzae pv. oryzicola (Xoc) causes bacterial leaf streak of rice, both of which are devastating bacterial diseases. Massive efforts have been undertaken to decipher Xoo and Xoc TALE biology. Numerous susceptibility genes in rice targeted by TALEs have been forecasted and validated. TALEs pthXo1, pthXo2 and pthXo3/AvrXa7/Tal5/TalC/ TalF of Xoo induce the expression of Xa13/Os8N3/SWEET11, Xa25/Os12N3/SWEET13 and Xa41/Os11N3/SWEET14, respectively, which are three members of the rice MtN3/saliva/SWEET family and function as sucrose transporters (Antony et al., 2010;Blanvillain-Baufumé et al., 2017;Chu et al., 2006;Hutin et al., 2015;Liu et al., 2011;Streubel et al., 2013;Tran et al., 2018;Yang et al., 2006;Zhou et al., 2015). TALEs pthXo6, TalB and pthXo7 of Xoo up-regulate the transcription factors OsTFX1, OsERF#123 and OsTFIIAγ1, respectively (Sugio et al., 2007;Tran et al., 2018). TALEs Tal9a of Xoo and Tal1c of Xoc trigger the induction of OsHen1, which encodes a protein with a predicted methyltransferase domain involved in micro-RNA (miRNA) maturation (Moscou and Bogdanove, 2009). TALE Tal2g of Xoc activates OsSULTR3;6, which encodes a sulfate transporter (Cernadas et al., 2014). All of these susceptibility genes can benefit Xoo or Xoc multiplication and disease development after activation by TALEs, although they encode different types of protein.  TALEs, respectively, based on genome sequence (Booher et al., 2015;Cernadas et al., 2014). In theory, each TALE specifically binds to the sole EBE of the host target gene promoter to activate its transcription. During this process, the Xooand Xocderived TALEs, after injection into plant cells, must be actively transported into the plant nucleus to bind to the target susceptibility gene promoter to reprogram the transcriptome of rice to cause disease. However, the underlying mechanism of how TALEs are transferred from the plant cytoplasm into the plant nucleus is still unclear.
So far, multiple pathways of nucleocytoplasmic transport have been identified, each likely to be involved in carrying a specific group of proteins (Goldfarb et al., 2004;Peters, 2006), the best characterized of which is the import of proteins containing a classical NLS that consists of either a short stretch of three to five basic amino acids or two basic domains separated by a spacer. The NLS-containing proteins are recognized and bound in the cytoplasm by the NLS receptor, and then transported into the nucleus. In rice, importin proteins OsImpα1a and OsImpα1b, which function as nucleocytoplasmic transporters, selectively bind and transport proteins containing T-NLS (a monopartite-type NLS with typical amino acid residues CTPPKKKRKV) and O 2 -NLS (a bipartite-type NLS with typical amino acid residues MPTEERVRKRKESNRESARRSRYRKAAHLKC), but not R-NLS (a yeast Matα-2-type NLS with typical amino acid residues CYMISEALRKAIGKR) (Chang et al., 2012(Chang et al., , 2014Goldfarb et al., 2004;Jiang et al., 1998Jiang et al., , 2001. Here, we reveal that rice OsImpα1a and OsImpα1b coordinately transfer the TALEs of Xoo and Xoc from the plant cytoplasm into the nucleus by selectively binding the bacterial pathogen-derived NLS of TALEs, which features a short stretch of five amino acids rich in arginine and lysine residues (RKRSR). We further demonstrate that OsImpα1a and OsImpα1b (cytoplasm/nucleus shuttle proteins)-mediated transportation of TALEs is vital for Xooand Xoc-triggered induction of targeting of susceptibility genes. Our results suggest that modification of the host nuclear import receptor genes OsImpα1a and OsImpα1b of rice by moderate suppression may provide a universally applicable strategy to improve plant resistance to the TALE-carrying bacterial pathogens Xoo and Xoc.

Rice OsImpα1a interacts with Xoo pthXo1
Like other Xoo TALEs, pthXo1 also typically contains an amino-terminal translocation signal (TS), a central RR, a newly identified TFB motif, three NLSs and a carboxyl-terminal transcription AD. Our previous studies have demonstrated that the complete pthXo1 possesses auto-activation activity because of the presence of the TS or AD domain; the truncated pthXo1 containing RR-TFB-NLS does not show auto-activation transcription activity in yeast (Yuan et al., 2016). To screen for pthXo1-interacting host rice proteins, the RR-TFB-NLS fragment of pthXo1 was used as a bait to trap proteins putatively interacting with pthXo1 by yeast two-hybrid assay. The prey cDNA library was prepared from mRNA isolated from leaves of rice variety IR24 after inoculation with Xoo strain PXO99. Several candidate pthXo1-interacting proteins were identified from approximately 5 × 10 5 independent colonies screened. Among the putative interacting proteins, the cDNA for the rice importin α1a gene (hereafter designated as OsImpα1a) showed the strongest interaction with the RR-TFB-NLS fragment of pthXo1. Rice OsImpα1a encodes a protein of 526 amino acids and has high amino acid sequence similarity with its homologues in rice, such as OsImpα1b and OsImpα2 (Jiang et al., 1998(Jiang et al., , 2001. A phylogenetic tree of plant importin α proteins, obtained from Xanthomonas-invaded host plants, showed that OsImpα1a grouped with Arabidopsis AtImpα1 and AtImpα2, pepper CaImpα2 and citrus CsImpα2 (Fig. S1, see Supporting Information); CaImpα2 has been shown to interact with Xanthomonas campestris pv. vesicatoria (Xcv) TALE AvrBs3 in yeast (Szurek et al., 2001).

pthXo1 interacts with OsImpα1a and OsImpα1b through NLS2
To narrow down which domain of pthXo1 interacts with OsImpα1a, we generated a series of pthXo1 constructs in which truncated coding sequences of pthXo1 were translationally fused with the DNA-binding domain of GAL4. These constructs included the entire RR to NLSs (RR-TFB-NLS), the TFB motif with NLSs (TFB-NLS), TFB, the three NLSs (NLS1/2/3), the first and second NLSs (NLS1/2) and the three different NLSs (NLS1, NLS2, NLS3) (Fig. 1A). The resulting constructs were co-transformed into yeast cells with the complete coding sequence of OsImpα1a fused with the DNA AD of GAL4, and empty vectors as the negative controls. The second NLS (NLS2) was the only domain to interact with OsImpα1a in the yeast two-hybrid assay (Fig. 1B). In contrast, RR, TFB, NLS1 and NLS3 did not exhibit interaction with OsImpα1a in yeast (Fig. S2A, see Supporting Information). Further, we found that NLS2 of pthXo1 also interacted with OsImpα1b, but not with OsImpα2, in yeast two-hybrid assay (Fig. S2B,C). These results demonstrate that the second NLS of pthXo1 is necessary for the interaction with the two rice nuclear import receptor proteins OsImpα1a and OsImpα1b.

pthXo1 interacts with OsImpα1a and OsImpα1b in vivo
To validate the yeast two-hybrid results, the interaction between pthXo1 and OsImpα1a or OsImpα1b was verified by in vivo assay. We first performed a firefly split-luciferase complementation assay in Nicotiana benthamiana leaves to investigate the interaction. During this assay, pthXo1 was fused with the N-terminus (nLUC) of firefly luciferase; OsImpα1a, OsImpα1b and OsImpα2 were fused with the C-terminus (cLUC) of firefly luciferase. We transiently co-expressed pthXo1-nLUC and OsImpα1a-cLUC, pthXo1-nLUC and OsImpα1b-cLUC, pthXo1-nLUC and OsImpα2-cLUC, as well as pthXo1-nLUC and cLUC as negative controls, in N. benthamiana leaves. A strong fluorescence signal was observed in leaves that co-expressed pthXo1-nLUC and OsImpα1a-cLUC, or pthXo1-nLUC and OsImpα1b-cLUC, but not in pthXo1-nLUC and OsImpα2-cLUC, or negative controls ( Fig. 1C), suggesting that pthXo1 interacted with OsImpα1a and OsImpα1b, but not OsImpα2, in planta.
We then performed co-immunoprecipitation (Co-IP) assays using pthXo1-myc and OsImpα1a-FLAG, OsImpα1b-FLAG or OsImpα2-FLAG co-expressed transiently in N. benthamiana leaves. The OsImpα1a-FLAG, OsImpα1b-FLAG and OsImpα2-FLAG proteins were immunoprecipitated using anti-FLAG-conjugated agarose. Immunoblots were washed and probed with anti-myc antibodies. The pthXo1-myc protein was pulled down by OsImpα1a-FLAG and OsImpα1b-FLAG. In contrast, OsImpα2-FLAG could not pull down pthXo1-myc protein in the same conditions (Fig. 1D). Taken together, these results demonstrate that Xoo TALE pthXo1, via the second NLS, physically associates with host rice OsImpα1a and OsImpα1b.

NLS2 of TALE is essential for Xoo virulence
We confirmed that Xoo TALE pthXo1 interacts with OsImpα1a and OsImpα1b in vitro and in vivo. We narrowed down the core The NLS2 of pthXo1 is required for the interaction with OsImpα1a analysed by yeast two-hybrid assay. The interactions were assessed by the growth of yeast cells on synthetic defined premixed (SD) medium lacking (-) leucine (L), tryptophan (W), histidine (H) and adenine (A). Vector, empty vector as control. (C) Firefly luciferase (LUC) complementation imaging assay. Nicotiana benthamiana leaves were co-infiltrated with agrobacterial strains containing different pairs of constructs. LUC images were captured using a cooled charge coupled device (CCD) imaging apparatus. The grey shape is one N. benthamiana leaf. The black circle indicates an interaction between two proteins. (D) Detection of interactions between Xoo pthXo1 and rice OsImpα1a or OsImpα1b in planta by co-immunoprecipitation (Co-IP). The protein-protein interaction assays were performed in N. benthamiana leaf cells. Proteins before (Input) and after immunoprecipitation (IP) were detected with anti-myc and anti-FLAG antibodies. interaction region of pthXo1 to NLS2. When aligning the NLS2 of TALEs from genome sequenced Xoo and Xoc strains, we found that all of the TALEs of different Xoo and Xoc strains exclusively have the same amino acid residues of their NLS2 (RKRSR). To assess which amino acid residues of RKRSR in NLS2 are essential for TALE function or Xoo virulence, we first produced pthXo1 derivatives with the five conserved amino acid residues of NLS2 substituted by alanine, individually. Of the NLS2 forms of pthXo1 tested, those with RKRSR and RKRAR interacted with OsImpα1a, but those with AKRSR, RARSR, RKASR and RKRSA did not, in yeast two-hybrid assay ( Fig. 2A). The same interaction was also observed between OsImpα1b and pthXo1 with NLS2 containing RKRSR or RKRAR, but not with the other mutations ( Fig. S3, see Supporting Information). We then validated the interaction in vivo. An in planta Co-IP assay strongly demonstrated that the substitution of the NLS2 residues of pthXo1 completely destroyed its interaction with OsImpα1a ( Fig. 2B).
To learn whether the conserved amino acid residues in NLS2 of TALE are directly responsible for the expression of TALE-induced host plant susceptibility genes and virulence on host rice, we generated NLS2 substituted with alanine in pthXo1. The constructs were re-introduced into Xoo strain PH, which is an engineered TALE-free strain with the genetic background of PXO99 A (Ji et al., 2016). Simultaneously, pthXo1(RKRSR) was introduced into PH as a control. After inoculation of rice variety IR24 with these Xoo strains, the expression of Xa13, the corresponding target gene of pthXo1, was activated by the presence of pthXo1 with RKRSR or RKRAR in NLS2, but not by versions of pthXo1 with AKRSR, RARSR, RKASR or RKRSA in NLS2 (Fig. 2C). The lesion length in these experiments was correlated with the up-regulated expression level of susceptibility gene Xa13 (Fig. 2D). In summary, these data show that the conserved amino acid residues of NLS2 of TALEs play key roles in the transcription of TALE-induced host plant susceptibility genes and Xoo virulence on host rice.

Localization of OsImpα1a and OsImpα1b
To investigate the intracellular location of OsImpα1a and OsImpα1b, we transiently expressed OsImpα1a-FLAG and OsImpα1b-FLAG fusion constructs, as well as OsNMD3-FLAG as a nucleocytoplasmiclocalized protein control (Shi et al., 2014), into N. benthamiana leaves. We performed subcellular fractionation analyses, followed by immunoblotting using FLAG antibody, accompanied by histone H3 antibody and phosphoenolpyruvate carboxylase (PEPC) antibody, which have been used as nuclear and cytosolic markers, respectively. The OsImpα1a-FLAG and OsImpα1b-FLAG proteins were detected in total protein extracts, the nucleus-enriched fraction and the nucleus-depleted fraction, in accordance with the nucleus and cytoplasm shuttle protein OsNMD3 (Fig. 3A). Taken together, these data indicate that OsImpα1a and OsImpα1b are located in the nucleus and cytoplasm, suggesting their roles as cytoplasm/nuclear shuttle proteins to transfer proteins between the cytoplasm and nucleus.

Expression patterns of OsImpα1a and OsImpα1b
To determine the expression patterns of OsImpα1a and OsImpα1b, we sampled the five tissues (root, stem, sheath, leaf and panicle) of IR24 at the booting stage, and assessed the transcript levels of the two genes in these tissues. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analyses showed that OsImpα1a and OsImpα1b were constitutively expressed in all of these tissues, with OsImpα1a and OsImpα1b having relatively higher transcription levels in the panicle compared with other tissues. OsImpα1a showed slightly higher expression than OsImpα1b in the different tissues (Fig. 3B). We then tested the transcript accumulation of OsImpα1a and OsImpα1b in response to bacterial pathogen infection. OsImpα1a and OsImpα1b accumulation was rapidly and markedly induced after Xoo infection in the leaf tissue of IR24, which is susceptible to Xoo strain PXO99 (Fig. 3C). In addition, OsImpα1a and OsImpα1b transcription was significantly activated after Xoc infection (Fig. S4, see Supporting Information), suggesting that OsImpα1a and OsImpα1b transcription occurs in response to bacterial pathogen infection.

OsImpα1a and OsImpα1b influence resistance to bacterial pathogens
To determine whether OsImpα1a and OsImpα1b play a role in rice-bacterial pathogen interaction, OsImpα1aand OsImpα1bsuppressing plants were generated using an RNA interference (RNAi) strategy. Because OsImpα1a and OsImpα1b showed similar expression patterns in diverse tissues and after bacterial pathogen infection, and had high nucleotide identity at 77% at the mRNA level (Fig. S5, see Supporting Information), it was difficult to choose a gene-specific fragment for the RNAi construct to target individual genes; therefore, we chose the common region of OsImpα1a and OsImpα1b to generate the RNAi construct to simultaneously suppress these two homologous genes. After the OsImpα1a/1b-RNAi construct had been transformed into IR24, which is susceptible to a large number of Xoo and Xoc strains, 11 independent transgenic plants were obtained. The expression of OsImpα1a and OsImpα1b was significantly reduced in 10 of the 11 plants. All of the transgenic plants were inoculated with Xoo strain PXO99 at the booting stage. The transgenic plants with lower OsImpα1a and OsImpα1b expression levels showed remarkably enhanced resistance to PXO99 with lesion lengths of 4.5 ± 2.5 cm to 8.8 ± 2.7 cm, compared with the wild-type lesion length of 21.8 ± 1.1 cm (Fig. 4). Increased resistance to Xoo was associated with reduced OsImpα1a and OsImpα1b expression, which was further confirmed in two T 1 families. T 1 families from two independent OsImpα1a/1b-RNAi T 0 plants with reduced expression of OsImpα1a and OsImpα1b were inoculated with Xoo at the booting stage. Some of the transgenic plants showed significantly enhanced resistance compared with the wild-type, and this resistance was associated with suppressed expression of OsImpα1a and OsImpα1b (Fig. S6, see Supporting Information). The reduction in Xoo-related disease symptoms in OsImpα1a/1b-RNAi plants was significantly correlated with the reduced expression of OsImpα1a and OsImpα1b in the OsImpα1a/1b-RNAi7 T 1 family (r = 0.989 for OsImpα1a, r = 0.992 for OsImpα1b, n = 8, α < 0.01) and the OsImpα1a/1b-RNAi8 T 1 family (r = 0.981 for OsImpα1a, r = 0.968 for OsImpα1b, n = 10, α < 0.01). Moreover, the Xoo growth rates in the leaves of OsImpα1a/1b-RNAi plants were observably lower than those in the wild-type (Fig. 5A). Furthermore, to assess whether OsImpα1a/1b-RNAi plants enhance resistance to other Xoo strains, we inoculated OsImpα1a/1b-RNAi7 and OsImpα1a/1b-RNAi8 T 2 families with 13 Xoo strains containing seven Philippine strains, four Chinese strains, one Japanese strain and one Korean strain, which are commonly used to test for broad-spectrum resistance to bacterial leaf blight. We found that OsImpα1a/1b-RNAi plants showed broad-spectrum resistance to the different Xoo strains (Fig. 5B).
In addition, we examined the resistance of OsImpα1a/1b-RNAi plants to the bacterial leaf streak pathogen Xoc. The RNAi lines and wild-type were inoculated with different Xoc strains at the tillering stage. The Xoc population was significantly lower in OsImpα1a/1b-RNAi plants than in the wild-type (Fig. 5C). The OsImpα1a/1b-RNAi plants exhibited clearly shorter lesion length than the wild-type after inoculation with different Xoc strains Fig. 4 Modulation of OsImpα1a and OsImpα1b expression influences rice response to Xanthomonas oryzae pv. oryzae (Xoo) infection. Analysis of the response of OsImpα1a/1b-RNAi T 0 plants to Xoo strain PXO99. Data represent the mean (five to eight leaves from one plant for lesion length) ± standard deviation (SD). Asterisks indicate a significant difference between transgenic plants and wild-type (WT) IR24 at **P < 0.01. (Fig. 5D). Taken together, these results suggest that OsImpα1a and OsImpα1b negatively regulate rice resistance to the bacterial pathogens Xoo and Xoc.

Activation of TALE-targeted susceptibility genes is inhibited in OsImpα1a/1b-RNAi plants
The bacterial pathogens Xoo and Xoc cause disease mainly through the targeting and activation of host rice susceptibility genes by their virulence TALEs. To monitor the expression pattern of rice TALE-targeted susceptibility genes on Xoo and Xoc infection in plants, we inoculated OsImpα1a/1b-RNAi plants with Xoo strain PXO99 at the booting stage and Xoc strain RS105 at the tillering stage. qRT-PCR assays showed that the induced expression of the known susceptibility genes Xa13, OsTFIIAγ1 and OsTFX1, each of which is targeted by a different TALE of Xoo PXO99, was significantly more suppressed in OsImpα1a/1b-RNAi plants than in wild-type plants (Fig. 6). Similarly, suppression of OsImpα1a and OsImpα1b in rice increased resistance to Xoc, which was associated with significantly hindered Xoc-activated expression of the rice susceptibility gene OsSULTR3;6, which is targeted by a TALE Tal2g of Xoc RS105, compared with wildtype plants (Fig. 6). These results demonstrate that the induction of TALE-targeted susceptibility genes is largely inhibited in OsImpα1aand OsImpα1b-suppressing plants after bacterial pathogen infection.

Virulence of TALE-free and T3SS-free strains on OsImpα1a/1b-RNAi plants
To assess whether the ability of OsImpα1a and OsImpα1b to confer resistance to Xoo and Xoc is exclusively associated with the virulence factors of bacterial pathogens, TALEs, but not non-TALEs, we inoculated OsImpα1a/1b-RNAi plants at the booting stage with Xoo strain PXO99 A (TALE -), also named PH (Ji et al., 2016). The OsImpα1a/1b-RNAi plants showed similar lesion lengths to the wild-type (Fig. 7A), and the Xoo population was indistinguishable between leaves of OsImpα1a/1b-RNAi plants and the wild-type at 4 and 8 days after infection (Figs 7B, S7A, see Supporting Information). In addition, we inoculated OsImpα1a/1b-RNAi plants and the wild-type at the booting stage with Xoo strain PXO99 A ΔhrcU, which is a T3SS-free strain lacking a functional T3SS and is unable to deliver TALEs and non-TALEs into plant cells (Guo et al., 2012). We found that there were similar Xoo populations in the leaves of OsImpα1a/1b-RNAi plants and the wild-type on the different days assessed after infection (Figs 7C, S7B). Furthermore, we simultaneously inoculated OsImpα1a/1b-RNAi plants and the wild-type at the tillering stage with Xoc strain RS105ΔhrcV, which is a T3SS-free strain lacking a functional T3SS (Guo et al., 2012). Similar Xoc populations were observed in the leaves of OsImpα1a/1b-RNAi plants and the wild-type on the different days evaluated after infection (Figs 7D, S7C). In brief, these data suggest that the broadspectrum resistance of OsImpα1a/1b-RNAi plants to Xoo and Xoc relies on the existence of TALEs in bacterial pathogens.

Mutational analysis of OsImpα1a/1b NLS binding sites
A previous study has validated that OsImpα1a has two separate NLS binding sites, the major site and the minor site, both of which recognize positively charged amino acid clusters in NLSs (Chang et al., 2012). The residues D188 and E388 of OsImpα1a are essential for plant-specific NLS binding to the major site and the minor site, respectively (Chang et al., 2012). To investigate whether these two core residues of OsImpα1a are responsible for bacterial pathogen-derived NLS binding, we produced OsImpα1a derivatives with the key residues D188 and E388 substituted by positively or negatively charged amino acid residues. These Fig. 6 Effect of OsImpα1a and OsImpα1b on the expression of the disease susceptibility genes Xa13, OsTFIIAγ1 and OsTFX1 after Xanthomonas oryzae pv. oryzae (Xoo) infection, and OsSULTR3;6 after Xanthomonas oryzae pv. oryzicola (Xoc) infection. Plants were inoculated with Xoo strain PXO99 [harbouring the transcription activator-like effector (TALEs) pthXo1, pthXo6 and pthXo7] at the booting stage, or Xoc strain RS105 (harbouring TALE Tal2g) at the tillering stage. Asterisks indicate a significant difference between transgenic plants and wild-type (WT) IR24 at **P < 0.01. derivatives of OsImpα1a were co-transformed into yeast with Xoo-specific NLS to assess binding activity. Yeast two-hybrid assays showed that mutation in the major site D188 with positively charged amino acid residues (D188E) or negatively charged amino acid residues (D188H, D188K, D188R), and mutation in the minor site E388 with positively charged amino acid residues (E388D) or negatively charged amino acid residues (E388H, E388K, E388R), did not attenuate the interaction between OsImpα1a and NLS2 of pthXo1 (Fig. S8A, see Supporting Information). In addition, the residues D194 and E394 of OsImpα1b, the key amino acids for NLS binding to the major site and the minor site, respectively, were assessed for interaction with Xoo-derived NLS. Similarly, mutation in the major site D194 and in the minor site E394 did not influence OsImpα1b binding to Xoo-specific NLS (Fig. S8B).
In conclusion, the mutations in the major site and minor site of rice OsImpα1a and OsImpα1b do not have a noticeable impact on the binding affinity for Xooor Xoc-derived NLS.

D I SC U SS I O N
Previously, we have revealed that TALE-carrying Xoo and Xoc hijack the host basal transcription factor IIA γ subunit (TFIIAγ) in the plant nucleus to cause disease in rice (Yuan et al., 2016). Here, we further demonstrate that Xoo and Xoc capture host plant cytoplasm/nuclear shuttle proteins OsImpα1a and OsImpα1b for successful transfer of their virulence TALEs from the plant cytoplasm into the plant nucleus (Fig. 8). TALEs then interact with TFIIAγ, and bind and activate target susceptibility genes to cause disease in rice. This inference is supported by the following evidence. First, Xoo and Xoc employ their NLSs, specifically conserved NLS2, to interact with the host cytoplasm/ nuclear shuttle proteins OsImpα1a and OsImpα1b. Second, transcriptional suppression of OsImpα1a and OsImpα1b enhances the resistance to diverse Xoo and Xoc strains, which is associated with the attenuated induction of susceptibility genes. Third, the efficient transportation of Xoo and Xoc TALEs into the plant nucleus is vital for bacterial pathogen invasion of rice.
The TALEs of isolated Xoo and Xoc strains with different geographical distributions all exclusively contain highly conserved NLS2, which is composed of five amino acid residues and is rich in arginine and lysine residues (RKRSR) (Booher et al., 2015;Cernadas et al., 2014). Although there are different amino acid residues of bacterial pathogen-derived NLS2 and plant nuclear proteins contained T-NLS and O 2 -NLS, which two have been proved selectively binding by both OsImpα1a and OsImpα1b (Jiang et al., 2001), TALEs of Xoo and Xoc mimic host nuclear proteins harbouring similar NLS for trapping by cytoplasm/nuclear shuttle proteins OsImpα1a and OsImpα1b, with the probably evidence that NLS2 of TALEs features with positively charged amino acids residues, lysine and arginine, which is accordance with T-NLS and O 2 -NLS. The exception to the transfer of plant-derived proteins from the cytoplasm into the nucleus, OsImpα1a, can directly bind to NLSs of Agrobacterium tumefaciens virulence protein VirD2, which forms a protein-nucleic acid supercomplex with T-DNA, facilitating T-DNA nuclear import (Chang et al., 2014). Moreover, the orthologues in pepper, CaImpα1 and CaImpα2, interact with TALE AvrBs3 of Xcv (Szurek et al., 2001). Thus, bacterial pathogens Xoo and Xoc, like Xcv, have evolved the simplest NLSs to mimic plant-derived nucleus-localized proteins for the facilitation of the transportation of virulence TALEs into the plant nucleus for the infection of host rice.
In rice, at least three nucleus importin α proteins have been characterized with nucleus trafficking capacity. We found that OsImpα1a and OsImpα1b, but not OsImpα2, interact with NLS2 of TALEs in in vitro and in vivo assays. The possible mechanism is that these three cytoplasm/nuclear shuttle proteins are responsible for the transportation of different nucleus-localized proteins by binding different types of NLS (Goldfarb et al., 2004). In order to identify whether there is functional redundancy of OsImpα1a and OsImpα1b for TALE transportation, we used the clustered regularly interspersed short palindromic repeats (CRISPR)/Cas system to separately knock out these two genes, with both homozygotes of the osImpα1a and osIm-pα1b mutants being developmentally lethal. An RNAi strategy was used to suppress the transcript of these two genes, and OsImpα1a/1b-RNAi plants showed slightly shorter flag leaf and panicle, fewer grains per panicle, decreased 1000-grain weight and significantly less seed setting than wild-type plants (Table S1, see Supporting Information). These data indicate that these two cytoplasm/nuclear shuttle proteins, OsImpα1a and OsImpα1b, probably play comparable and pivotal roles in the transportation of NLS-containing proteins, which are essential for plant growth and development. Whether these two proteins play greater roles than their homologue OsImpα2 should be investigated further.
Rice importins α1a and α1b have been validated as components of the NLS receptor in plant cells, and transfer distinct groups of nuclear proteins (Jiang et al., 1998(Jiang et al., , 2001. Here, we found that these two importin proteins also carry bacterial pathogen-derived nuclear proteins into the plant nucleus. Our in vitro and in vivo assays showed that OsImpα1a and OsImpα1b associated with TALEs are dependent on the presence of NLS2, but not NLS1 or NLS3, which is in accordance with the analysis of TALE AvrBs3 of Xcv, where deletion of NLS1 or NLS3 of AvrBs3 is not sufficient to abolish the induction of the AvrBs3-activated hypersensitive response (HR) on Bs3-containing resistant pepper plants (Szurek et al., 2001). NLS1, NLS2 and NLS3 of Xanthomonas are rich in positively charged amino acid residues. However, Xanthomonas species selectively use their NLS2, not NLS1 or NLS3, of TALEs to trap host plant cytoplasm/nuclear shuttle proteins. The functions of NLS1 and NLS3 during the process of Xanthomonas TALE-triggered susceptibility in plants requires further study.
Both OsImpα1a and OsImpα1b were induced in rice leaves in response to infection with the bacterial pathogens Xoo or Xoc, whereas their homologous genes in pepper, Caimpα1 and Caimpα2, were constitutively expressed independent of infection with Xcv (Szurek et al., 2001). This diversity could be caused by differences between the respective host plants, rice and pepper. Plants with suppressed OsImpα1a and OsImpα1b transcription showed broad-spectrum disease resistance to Xoo and Xoc, accompanied by attenuated induction of susceptibility genes. The reason that TALEs targeting susceptibility genes could not be immediately up-regulated after Xoo and Xoc infection is a result of the inefficient transportation of TALEs into the plant nucleus. However, OsImpα1a/1b-RNAi plants showed a similar lesion length and bacterial population to the wild-type after inoculation with TALE-free and T3SS-free bacterial strains. These results further indicate that suppression of OsImpα1aor OsImpα1b-mediated resistance to Xoo or Xoc is solely associated with TALEs, the major virulence factors of bacterial pathogens.
In conclusion, the present results and a previous report (Szurek et al., 2001) suggest that the TALE-carrying bacterial pathogens, Xoo, Xoc and Xcv, use the same mechanism, i.e. the trapping of host plant nuclear import receptor proteins, to transfer their virulence TALEs from the plant cytoplasm into the nucleus to activate susceptibility genes to cause disease in rice and pepper. The TALE-carrying genus Xanthomonas infects a wide range of plants (Jacques et al., 2016), and the virulence TALEs contain an identical NLS2. The plants simultaneously carry nuclear import receptors with high amino acid sequence similarity (Wiermer et al., 2007). Thus, moderate suppression of expression of plant nuclear import receptor proteins may provide an applicable strategy to improve disease resistance to TALE-carrying bacterial pathogens in other plants.

E X PE R I M E N TA L PROC E D U R E S Plant and bacterial materials
Rice (Oryza sativa ssp. Xian) IR24 is susceptible to Xoo and Xoc, and was used in this study. Plants were grown during a normal rice-growing season under natural field conditions.
The reference Xoo strain PXO99 A and its mutation PH [PXO99 A (TALE -)], with deletion of all 19 TALE genes, and Xoc strain RS105 have commonly been used in studies of rice resistance to bacterial leaf blight disease and bacterial leaf streak disease, respectively (Ji et al., 2016), and were used in this study. Xoo strain PXO99 was used for pathogen inoculation. PXO99 A , a 5-azacytidine-resistant mutant of PXO99, was used for genetic manipulation and pathogen inoculation. 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, 100 μg/mL; rifampicin, 75 μg/mL; kanamycin, 25 μg/mL.

Site-directed mutation
Site mutation of plant OsImpα1a and OsImpα1b genes and Xoo pthXo1 gene was performed using the GeneTailor Site-Directed Mutagenesis System (Invitrogen Life Technologies, Carlsbad, CA, USA), as described previously (Yuan et al., 2009). The sitemutated genes were confirmed by Sanger sequencing.

Vector construction and plant transformation
The TALE pthXo1 and its variations were cloned into the pHM1 vector to produce pHM1-pthXo1, and then transferred into Xoo strain PH following a published method (Ji et al., 2016). The NLS2 region of pthXo1 was replaced with its site-directed mutations by Gibson assembly (Gibson et al., 2009), following confirmation by Sanger sequencing.
The full-length cDNAs of OsImpα1a, OsImpα1b and OsNMD3 were ligated into the pU1301-3FLAG vector. The recombinant vectors were introduced into A. tumefaciens strain GV3101. Agrobacterium-mediated transformation was performed by infiltration into N. benthamiana leaves using a needleless syringe (Yuan et al., 2016).
To construct the RNAi vector, the gene-specific and highsimilarity fragment of OsImpα1a and OsImpα1b genes was amplified and inserted into the pDS1301 vector (Yuan et al., 2010). The recombinant vector was introduced into A. tumefaciens strain EHA105. Agrobacterium-mediated transformation was performed using calli derived from mature embryos of rice variety IR24, according to a published protocol .

Protein-protein interaction
To study the interaction between Xanthomonas TALEs and rice OsImpα1a, OsImpα1b and OsImpα2 in yeast two-hybrid assays, the different domains of TALE pthXo1 and its variations were amplified using gene-specific primers (Yuan et al., 2016); the amplified DNA segments were then ligated into the pGBKT7 vector. Rice OsImpα1a, OsImpα1b, OsImpα2 and their variations were amplified using gene-specific primers (Table S2, see Supporting  Information), and the amplified DNA segments were ligated into the pGADT7-Rec vector. The recombinant pGBKT7 and pGADT7 plasmids were then co-transformed into yeast strain AH109 for interaction analyses (Yuan et al., 2010). The yeast clones were restreak on synthetic defined premixed (SD) medium lacking leucine (L) and tryptophan (W) (−LW) and selective SD medium lacking L, W, histidine (H) and adenine (A) (−LWHA).
To study the interaction in planta, Co-IP assays were performed (Yuan et al., 2016). The DNA segments of pthXo1 and its variations were ligated into the pU1301-9myc vector; the DNA segments of OsImpα1a, OsImpα1b and OsImpα2 were ligated into the pU1301-9myc vector (Yuan et al., 2010). The recombinant constructs were introduced into A. tumefaciens strain GV3101 by electroporation. Agrobacterium-mediated transformation was performed by infiltration into N. benthamiana leaves using a needleless syringe. Co-IP assays were carried out using anti-FLAG antibody (F7425, Sigma, Sigma-Aldrich, St. Louis, Missouri, USA) and anti-myc antibody (AB103, Tiangen, Beijing, China), as described previously (Yuan et al., 2016). Each Co-IP assay was repeated at least twice.

Split-luciferase complementation assay
To construct split-luciferase complementation assay vector, the open reading frame of pthXo1 was inserted into the vector pCAMBIA-35S-nLUC to generate construct pthXo1-nLUC; the open reading frames of OsImpα1a, OsImpα1b and OsImpα2 were inserted into the vector pCAMBIA-35S-cLUC to generate constructs OsImpα1a-cLUC, OsImpα1b-cLUC and OsImpα2-cLUC, respectively. The recombinant constructs were introduced into A. tumefaciens strain GV3101. Equal amounts of Agrobacterium cultures for nLUC and cLUC constructs were mixed and co-transformed into fully expanded leaves of N. benthamiana. After 2 days, 1 mm of precooled luciferin was sprayed onto the leaves, and the samples were incubated in the dark for 5 min. LUC images were captured using a cooled charge coupled device (CCD) imaging apparatus (Chen et al., 2008).

Pathogen inoculation
To evaluate rice bacterial blight disease, five to seven uppermost fully expanded leaves of each plant were inoculated with different Xoo strains at an optical density at 600 nm (OD 600nm ) = 0.5 by the leaf clipping method (Yuan et al., 2016) at the booting (panicle development) stage. Xoo strains included Philippine strains PXO61, PXO86, PXO71, PXO112, PXO99, PXO347 and PXO341, Chinese strains Zhe173, KS-1-21, YN11 and FuJ23, Japanese strain T7174 and Korean strain KACC10331. Disease was scored by measuring the lesion length at 14 days after inoculation. The bacterial growth rate in rice leaves was determined by counting the number of colony-forming units (Yuan et al., 2016).
To evaluate rice bacterial streak disease, five to eight fully expanded leaves were inoculated with Xoc strains with OD 600nm = 0.5 by the penetration method at the tillering stage (Yuan et al., 2016). Xoc strains included Chinese strains RH3, RS85, RS105, JSB2-24 and HNB8-47. The disease was scored by measuring the lesion length at 14 days after inoculation. The bacterial growth rate in rice leaves was determined by counting the number of colony-forming units (Yuan et al., 2016).
For the measurement of the growth in planta of Xoo strains PXO99 A (TALE -) and PXO99 A ΔhrcU, and Xoc strain RS105ΔhrcV, these pathogens were infiltrated into the intercellular spaces of fully expanded rice leaves with needleless syringes at three different locations per leaf. The bacterial growth rate in rice leaves was determined by counting the number of colony-forming units (Yuan et al., 2016).

Gene expression analysis
For gene expression analysis, real-time qRT-PCR was performed using SYBR Premix Ex Taq (Takara, Dalian, China) in an ABI 7500 Real-Time PCR System (Applied Biosystems, Foster, California, USA). In brief, 2-cm rice leaf fragments near the bacterial infection sites were collected for RNA isolation. Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, California, USA). An aliquot (5 μg) of total RNA was treated with RNase-free DNase I (Invitrogen) to remove potentially contaminating DNA, and first-strand cDNA was reverse transcribed from total RNA with oligo(dT) 18 primer using M-MLV reverse transcriptase (Promega, Madison, Wisconsin, USA) according to the manufacturer's instructions. qRT-PCR was conducted using gene-specific primers (Table S3, see Supporting Information). The expression level of the rice actin gene was used to standardize the RNA sample as an internal control. The expression level relative to that of controls was assessed. Each qRT-PCR assay was repeated at least twice with a similar result, with each repetition having three replicates.

Sequence analysis
Multiple sequence alignments of amino acid sequences were generated using ClustalW in MEGA X with the default parameters. The sequence alignments obtained were used as input for the neighbour-joining analysis in MEGA X to construct the phylogenetic tree. For phylogenetic tree construction, a bootstrap method with 1000 replications was used for test of phylogeny, with Poisson model and pairwise deletion during gaps/missing date treatment.

Statistical analysis
Differences between samples were analysed for statistical significance using SPSS software and Student's t-test (two-tailed). The correlation analysis between disease symptom and gene expression level was analysed using the Pearson correlation coefficient analysis in SPSS (IBM SPSS Statistics, Version 19.0, IBM Corp., released 2010, Armonk, New York).