Two TAL effectors of Xanthomonas citri promote pustule formation by directly repressing the expression of GRAS transcription factor in citrus

Abstract

Citrus bacterial canker (CBC), caused by Xanthomonas citri subsp. citri (Xcc), poses a significant threat to the citrus industry. Xcc employs the transcription activator-like effector (TALE) PthA4 to target the major susceptibility (S) gene CsLOB1 in citrus, promoting host susceptibility to bacterial canker. However, the contribution of other Xcc TALEs, aside from PthA4, to virulence remains underexplored. In this study, we characterized two PthA1 variants, designated PthA5 and PthA6, which facilitate Xcc infection in susceptible citrus species by promoting the formation of hypertrophy and hyperplasia symptoms. Both PthA5 and PthA6 bind directly to effector-binding elements (EBEs) in the promoter of CsGRAS9, suppressing its expression. CsGRAS9 negatively regulates Xcc growth in citrus and contributes to CBC resistance. Notably, natural variations in the EBEs of the FhGRAS9 promoter, a homolog of CsGRAS9 in Hong Kong kumquat, prevent Xcc from affecting FhGRAS9 expression. Using the PTG/Cas9 system, we generated proCsGRAS9-edited sweet orange lines #18–2 and #23, which contain 86-bp and 62-bp deletions in the EBE regions of the CsGRAS9 promoter. These mutant lines showed enhanced CsGRAS9 expression and increased resistance to CBC during Xcc infection. Several GA-related genes and CsTAC1, regulated by CsGRAS9, were also identified. This is the first report that TALEs act as repressors of a resistance gene to confer host susceptibility.

Core

The natural variations at EBE regions of GRAS9s promoter affect the binding affinity of TALE proteins (PthA4, PthA5, and PthA6), resulting in the differential expression of GRAS9s, which is induced in Hong Kong kumquat during Xcc infection. Mutations generated with 86-bp and 62-bp deletions in the EBEs of the CsGRAS9 promoter by the PTG/Cas9 system resulted in an increased canker resistance in sweet oranges, providing a potential resource for the disease-resistant genetic breeding of cultivated citrus.

Gene and accession numbers

Sequences were obtained from the National Center for Biotechnology Information database (NCBI), Arabidopsis Information Resource (TAIR) (https://www.arabidopsis.org/), Rice Genome Annotation Project (http://rice.uga.edu/), Sol Genomics Network (https://solgenomics.net/), and Citrus Pan-genome to Breeding Database (CPBD) (http://citrus.hzau.edu.cn/index.php). The accession numbers are listed in Supplementary Table S2 and Table S3.

Introduction

Citrus bacterial canker (CBC), caused by Xanthomonas citri subsp. citri (Xcc), is a severe disease that significantly affects citrus production and results in substantial economic losses (Stover et al. 2014). Xcc strains (pathotype A) have an extensive geographic distribution and infect various hosts. Citrus has a diverse collection of germplasm resources, including grapefruit (Citrus paradisi), sweet orange (Citrus sinensis), lemon (Citrus limon), and pummelo (Citrus grandis), all of which are susceptible to citrus canker (Duan et al. 2022). In addition, the Xcc pathotypes A* and A^w have a more limited host range and infect alemow (Citrus macrophylla) and Mexican lime (Citrus aurantifolia), but not grapefruit (Citrus paradisi) (Patané et al. 2019). Ongoing efforts to breed profitable citrus cultivars that are resistant to disease remain a crucial task. During infection, bacteria compress intercellular spaces, leading to cell hydration and swelling (An et al. 2020). Xanthomonas species promote xanthan gum synthesis, which enhances water absorption through capillary action in the xylem (Shahbaz et al. 2022). Canker symptoms first appear as water-soaked lesions on the abaxial surface of leaves, which then develop into eruptive lesions caused by rupture and splitting of the outer leaf layer, ultimately leading to leaf loss. Kumquats (Fortunella spp.) exhibit resistance to Xcc through early leaf abscission and development of a hypersensitive response (HR)-like phenotype (Teper et al. 2020). Swelling, division, and proliferation of host cells after infection contribute to the water-soaked appearance (Brunings et al. 2003, Zou et al. 2021). Further research is needed to clarify the ongoing arms race between plant defense mechanisms and pathogens striving to evade or overcome these defenses.

Xanthomonas spp. secrete transcription activator-like effectors (TALEs) into plant cells through the type III secretion system (T3SS) to manipulate bacterial virulence in hosts (Ji et al. 2016). TALE proteins are potent and essential factors that directly bind to specific DNA bases in promoters through interactions with the 12th and 13th repeat variable diresidues (RVDs) located in the central region of polymorphic repeats (CRRs) (White et al. 2009). The virulence of XccA strains mainly depends on the interaction between PthA4 and the effector binding element (EBE) region in the promoter of the canker susceptibility (S) gene CsLOB1 (Hu et al. 2014, Li et al. 2014). In plant–pathogen interactions, TALEs reprogram host cells to promote hypertrophy and hyperplasia, leading to morphological and physiological changes that enhance favor pathogen fitness, proliferation, and spread, or trigger disease resistance (Timilsina et al. 2020). The activation of CsLOB1 expression is essential for pustule formation, cell wall modifications, and cell expansion (Zou et al. 2021, Li et al. 2014). XccA strains also contain three additional homologs of PthA: PthA1, PthA2, and PthA3 (Yan et al. 2012). In the citrus variety ‘Pera’, PthA1 induces the expression of the S gene DIOX, promoting the formation of necrotic pustules and facilitating pathogen adaptation (Abe et al. 2016). However, the precise mechanisms through which PthA1 and its derivatives promote necrotic pustule formation and pathogen adaptability in hosts remain unclear and ambiguous. X. citri strains use horizontal gene transfer, which is mediated by plasmid diversity arising from human activities and plant transportation, to circumvent host resistance and adapt to environmental changes (Pruvost et al. 2021). TALE variations arise through recombination, replication, insertions/deletions (InDels), and single nucleotide polymorphisms (SNPs) in the CRR region, driving a co-evolutionary arms race between pathogens and host resistance (Teper et al. 2021, Ruh et al. 2017, Ferreira et al. 2015). In response to pathogen infection, plants trigger a type of programmed cell death (PCD) known as HR. The short PthA4^AT from X. citri A^T strains, which contains 7.5 repeats in the RVD region, elicits HR and resistance in C. limon and C. sinensis, suggesting that its function extends beyond “classical” TALEs (Roeschlin et al. 2019). Another TALE, Tal2_Xss-V2-18, with 25.5 repeats in the RVD region, is a critical component for the full virulence of X. citri pv. malvacearum (Xcm) strains in cotton (Haq et al. 2020). In our previous study, we isolated two highly pathogenic strains, Xcc003 and Xcc086, from infected grapefruit and kumquat leaves, respectively (Ye et al. 2013). However, the contribution of the long avrBs3/pthA variants in these strains to bacterial virulence remains unknown.

Repetitive regions of TAL effectors specifically target EBEs in gene promoters to regulate gene expression and facilitate disease progression. Xu et al. (2019) used the CRISPR system to generate broad-spectrum resistance in rice by introducing gene-edited mutations in the EBE regions of the OsSWEET13 promoter (Xu et al. 2019). Similarly, mutations in the EBE_PthA4 regions of the CsLOB1 promoter in citrus plants reduced canker symptoms by using polycistronic tRNA–gRNA/Cas9 (PTG/Cas9) and Cas12a-mediated genome editing systems (Su et al. 2023; Tang et al. 2021a). However, the relationship between EBE mutations in target promoters and citrus resistance remains insufficiently understood, especially in citrus plants. Primitive and wild citrus germplasms suppress the development of canker symptoms following Xcc infection, supporting the use of natural genetic resources in breeding for improved canker resistance (Fu et al. 2020). Recently, Tang et al. (2021b) reported natural variation in the AbLOB1 promoter in Atalantia buxifolia (Chinese box orange, a primitive Chinese variety) and demonstrated that mutations in the AbLOB1 promoter led to distinct LOB1 transcription levels compared with C. sinensis, altering citrus canker resistance (Tang et al. 2021b). Thus, both natural variations and genome editing targeting EBE regions have been used to generate broad-spectrum resistant plants (Su et al. 2023). However, natural variation in promoters and similar disease resistance mechanisms in resistant Hong Kong kumquat (Fortunella hindsii) have not been reported. The susceptibility or resistance genes associated with citrus canker in Hong Kong kumquat remain to be validated, and the target gene directly regulated by TALEs is yet to be identified.

GRAS proteins are plant-specific transcription factors that play a pivotal role in regulating stress response, gibberellin (GA) signaling, and plant development (Hirsch et al. 2009). In Arabidopsis, miR171c targets GRAS proteins to regulate various biological processes, including the branching of the shoot and maintenance of the shoot meristem (Wang et al. 2010). In addition, SCARECROW-LIKE (SCL) proteins function pleiotropically in rice growth and disease resistance. After the inoculation of rice by Magnaporthe oryzae, the expression and transcriptional activity of OsSCL7 were increased, resulting in the elevated expression of defense-related genes and enhanced rice resistance to M. oryzae (Lu et al. 2022). miR171b inhibits the transcription of SCL6-IIs to enhance defense responses against M. oryzae and increase resistance to rice blast (Li et al. 2022a, b). Moreover, CMLs and GRASs contribute to disease resistance in sesame genotypes resistant to Macrophomina phaseolina (Yan et al. 2021). CsGRAS9 (Cs2g22130) has been classified into the AtSHR subfamily (Zhang et al. 2019). Our previous study demonstrated that the loss of function of TAL effectors induced Cs2g22130 expression during Xcc infection (unpublished data). Recently, Teper et al. (2020) reported that PthA4 downregulates the expression of CsGRAS9 at 4 days post infection of Xcc. These findings suggest that CsGRAS9 is involved in the response to Xcc infection and could serve as a potential target for TAL effectors. However, the characterization of CsGRAS9 function in response to biotic stress and the role of GRAS9s proteins in citrus species with varying levels of canker resistance have not been adequately explored. Plant responses to biotic stress involve regulation of phytohormone levels, telomerase activity, and senescence (Kalinina et al. 2018). For example, the suppression of the AtSCL32 homolog SlGRAS26 in tomato impairs GA biosynthesis by downregulating the expression of the GA metabolism genes GA3ox1 and GA3ox2 and the GA biosynthesis genes CPS and KAO while activating the GA inactivation pathway (Zhou et al. 2018). IDD proteins function as transcriptional coactivators with the DELLA protein REPRESSOR OF GA1-3 (RGA) to promote the expression of SCL3 and regulate the GA signaling pathway (Jaiswal et al. 2022). Plants have evolved mechanisms to maintain telomere length by regulating telomerase activity, thereby preventing chromosome degradation, senescence, and eventual cell death due to telomere shortening (Procházková Schrumpfová et al. 2016). The telomerase activator TAC1 may play a role in plant defense mechanisms in olive trees (Olea europaea L. subsp. europaea var. europaea), which affects auxin signaling (Ramírez-Tejero et al. 2021). In highly resistant cultivars, TAC1 expression is suppressed in roots, whereas in extremely susceptible cultivars, its expression is induced. Cs5g06630/TAC1 is orthologous to Arabidopsis thaliana TAC1, which enhances auxin responses and induces telomerase activity in plant leaves (Ren et al. 2007). However, the downstream mechanisms regulated by CsGRAS9 that affect citrus canker susceptibility remain unclear.

The present study examined the mechanism of TALEs action in pathogen dispersal and pathological hypertrophy in citrus hosts to breed varieties resistant to Xcc infection. We characterized two avrBs3/pthA variants, pthA5 and pthA6, which promote bacterial pathogenesis and symptom development in hosts. CsGRAS9 expression was suppressed in response to PthA4, PthA5, and PthA6 during Xcc infection. We found that PthA5 and PthA6 directly bind to EBEs in the CsGRAS9 promoter, thereby suppressing gene expression in grapefruit. We also identified the transcription factor FhGRAS9 in the resistant F. hindsii ‘Hong Kong kumquat’. In Hong Kong kumquat, in addition to the FhGRAS9 promoter genotype that aligns with the susceptible grapefruit CsGRAS9 promoter (named FhGRAS9-P1), we detected a natural variation (a 558-bp deletion) in the FhGRAS9 promoter (named FhGRAS9-P2). Moreover, we determined that natural variations in the promoters of GRAS9s lead to differential gene expression in the two citrus germplasms, which exhibited contrasting levels of canker resistance. Overexpression of CsGRAS9 conferred tolerance to citrus canker in grapefruit. We also generated CsGRAS9 promoter-edited C. sinensis cv. ‘Anliu’ lines using the PTG/Cas9 system, demonstrating that editing the PthA4, PthA5, and PthA6 EBEs in the CsGRAS9 promoter region enhanced resistance to Xcc in sweet orange. These results elucidate the molecular mechanism of CsGRAS9 in citrus plants during Xcc infection, providing a potential resource for the disease-resistant genetic breeding of cultivated citrus.

Results

PthA1 variants PthA5 and PthA6 promote the canker symptom formation

Our previous report highlighted significant variability in the size of the tal genes from different Chinese Xcc strains (Ye et al. 2013). Among these, strains Xcc086 and Xcc003 were shown to contain unique tal genes that differ from the pthA4 gene and tal genes found in other Chinese Xcc strains (Ye et al. 2013). Given the mode of action of TALE proteins, these variants are expected to play specific roles during bacterial pathogenesis. The pathogenicity of Xcc003 and Xcc086 was confirmed in susceptible grapefruit ‘Paradise’ and resistant kumquat (Fig. 1A). To define the unique tal genes of Xcc086 and Xcc003, we cloned a 3.6-kb band from Xcc086 (corresponding to the second hybridization signal band), and a 4.4-kb band from Xcc003 (corresponding to the first hybridization signal band), from the BamHI-digested plasmid DNA of Xcc086 and Xcc003 according to our previous Southern blot assays (Ye et al. 2013). The BamHI bands, containing the central RVD repetitions, were sequenced (Supplementary Fig. S1). The 3.6-kb band of Xcc086 contains a tal gene encoding a TALE with 19.5 RVD repeats, and the 4.4-kb BamHI band of Xcc003 corresponds to a tal gene encoding a TALE with 27.5 RVD repeats (Fig. 1B and Supplementary Fig. S1). We name these TALE proteins PthA5 and PthA6, respectively. Alignment analysis of the RVDs revealed that the region of PthA5 from the 1st to the 15th and the last two repeats was identical to all the repeats of PthA1 (Fig. 1B). The regions of PthA6 resembled those of PthA5 and PthA1, but contained 8 and 11 additional RVDs compared to PthA5 and PthA1, respectively (Fig. 1B). Notably, the 21nd to 26th extra RVDs of PthA6 matched the 9th to 14th central repeats of PthA5 and PthA1, based on the arrangement of the RVD repeats (Fig. 1B). These analyses indicate that PthA5 and PthA6 belong to the members of the PthA1 group. Additionally, repeat-based phylogenetic lineages of PthA5, PthA6, and 262 TALEs obtained from 77 fully sequenced X. citri strains in the NCBI database were assessed using DisTAL. On the phylogenetic tree, PthA5 and PthA6 clustered together with PthA1, PthAW1, Apl3, TAL4_{Xc03-1638-1-1}, and TALEs from Xcc and Xcm strains isolated from diverse geographic regions, suggesting the common ancestral origin (Supplementary Fig. S2A and S2C). TAL4_{Xc03-1638-1-1} was identified in X. citri strain Xc-03–1638-1-1 and was conserved among other X. citri strains, including LJ207–7, LL074–4, LH276, and LH201. PthA5 also clustered with TAL1 of Xcc LJ207-7 strain (from the Southwest Indian Ocean islands), which is encoded by a plasmid and is closely related to TAL1 of TX160197 in the PthAW1 and AW pathotypes (Supplementary Fig. S2A). Interestingly, PthA6 was most closely related to TAL1 of the Korean strain X. citri pv. glycines (Xcg) 1018, which is encoded on the chromosome. As TALE genes are genetically unstable, the adaptability of Xanthomonas species is enhanced by recombination events and RVD polymorphisms that arise under selective pressure during evolution (Schandry et al. 2018). Based on the clustering in repeat-based phylogeny, we speculate that recombination events likely occurred among ancestral PthA homologs from Xcc, Xcg and Xcm strains, leading to the variations in repeat numbers seen in PthA5 and PthA6.

To assess whether PthA5 and PthA6 play roles in Xcc pathogenicity, we constructed two pHZY derivative plasmids expressing PthA5 and PthA6, and introduced these recombinant plasmids into Xcc049E, a mutant strain of Xcc049 with deletions of all tal genes, generating Xcc049E/pthA5 and Xcc049E/pthA6. The expression of PthA5 and PthA6 in the transformants was confirmed by western blotting (Supplementary Fig. S3). Bacterial titers of Xcc049E/pthA5 and Xcc049E/pthA6 were inoculated into grapefruit via infiltration to evaluate bacterial virulence and pathogenicity. Both Xcc049E/pthA5 and Xcc049E/pthA6 strains complemented Xcc049E's function to cause typical canker symptoms including hyperplasia and hypertrophy (Fig. 1C). The removal of TALEs effectively eliminated citrus cankers in grapefruit, and we observed that PthA5 and PthA6 positively contributed to disease symptom development. Microscopic observations further confirmed that PthA5 and PthA6 promoted canker symptom formation. Compared to leaves infected with Xcc049E/EV, the spongy parenchyma cells in grapefruit leaves infected with Xcc049E/pthA5 and Xcc049E/pthA6 were significantly enlarged (2–3 times), and the palisade cells infected with Xcc049E/pthA5 were wider and more cratered at 21 days post-inoculation (dpi)(Fig. 1C). Furthermore, bacterial growth of Xcc049E/pthA5 and Xcc049E/pthA6 was significantly higher than Xcc049E at 2 and 4 dpi, though slightly lower than Xcc049E/pthA4 (Fig. 1D). These results indicate that PthA5 and PthA6 function as virulence factors, inducing canker symptom formation in hosts.

Both Xcc003 and Xcc086 strains have the capacity to induce CsLOB1 transcriptional activity due to the presence of pthA4 (Ye et al. 2013) . In contrast, after inoculation with Xcc049E strains containing pthA5 and pthA6, no significant differences in CsLOB1 expression were observed compared to Xcc049E alone (Supplementary Fig. S4). To further investigate the molecular mechanisms of PthA5 and PthA6, we used FuncTAL for functional analysis based on DNA-binding affinity. The analysis revealed that PthA5 and PthA6 clustered with PthA1 (Supplementary Fig. S2B). As shown in Supplementary Fig. S5, CsLOB2 and CsDIOX were induced in response to PthA5, consistent with previous findings that PthA1 enhances the expression of CsLOB2 and CsDIOX (Abe et al. 2016). TALgetter also predicted that the binding sites of PthA1 and PthA5 overlap in the promoter of CsDIOX, reinforcing the hypothesis that PthA1 and PthA5 activate transcription in the host by binding to the EBE site (Supplementary Fig. S6). Compared to individual infection with Xcc049E/pthA4, the simultaneous presence of pthA4 and either pthA5 or pthA6 elicited more severe water-soaked canker symptoms and promoted bacterial growth in grapefruit (Fig. 1E-F). In agreement with previous observations, PthA5 functions similarly to PthA1, playing an additive role in PthA4-elicited pustule formation and acting as a pathogenicity factor by activating DIOX expression.

PthA5 and PthA6 directly bind to the CsGRAS9 promoter to inhibit its transcription

To investigate the molecular mechanisms of PthA5 and PthA6 in canker formation, we used TALE target prediction tools (AnnoTALE, Target Finder and Talvez) to identify putative EBEs for PthA5 and PthA6 by scanning the sweet orange (C. sinensis) promoterome (Supplementary Fig. S7 and Tables S5-6). The predicted EBEs for PthA5 and PthA6 were primarily located in the promoter region of Cs2g22130, which encodes a putative GRAS family member, CsGRAS9 (Fig. 2A). Additionally, we analyzed the potential targets of PthA1 and PthA4 in the CsGRAS9 promoter and observed that the candidate EBE regions displayed features of TATA-like elements (Fig. 2A and Supplementary Tables S7-8), as previously reported (Pereira et al. 2014). To gain further insights into the function of PthA5, we generated a structural model using Alphafold 2, which revealed an uninterrupted right-handed superhelical molecule. PthA5 comprised 19.5 RVD repeats (residues 300 to 908), including a 0.5 repeat at the C-terminus formed by non-conserved amino acids. The 19.5 RVD repeats created a complete helical turn, with the innermost spiral formed by the RVD loops, exhibiting a nearly identical conformation (Fig. 2B). Protein-DNA docking analysis showed a binary complex that the DNA molecule was positioned along the axis within PthA5 superhelical assembly. This structure incorporated the 19.5 RVD repeats and the 27-base pair (bp) DNA-binding element through hydrogen bonds.

We further performed a competitive electrophoretic mobility shift assay (EMSA) to determine whether the PthA4, PthA5, and PthA6 proteins could bind to the predicted EBEs in the CsGRAS9 promoter. His-tagged fusion proteins PthA6-His, PthA5-His, and PthA4-His were purified, and a 33 bp double-stranded DNA fragment (ProCsGRAS9) containing EBEs for PthA4, PthA5, and PthA6 was used as the probe for EMSA (Supplementary Fig. S8 and Fig. 2A). The EMSA results showed that PthA4, PthA5, and PthA6 proteins bound to ProCsGRAS9, indicating that the presence of EBEs enabled the specific binding of these TALEs (Fig. 2C-E). Additionally, the binding affinities of these TALEs to the CsGRAS9 promoter were estimated using the ‘Target Finder’ tool from TAL Effector Nucleotide Targeter 2.0 and the ‘predict and intersect Targets’ feature of AnnoTALE. The analysis demonstrated that PthA1 had the strongest affinity for EBE regions in the CsGRAS9 promoter, particularly for EBEs G-M and 25–28 (Supplementary Fig. S7B-C) . PthA5 also showed strong affinity for the EBEs, though slightly less than PthA1. In contrast, PthA6 and PthA4 exhibited weaker affinities for the CsGRAS9 promoter compared to PthA1 and PthA5. Taken together, these findings suggest that among the three effectors, PthA5 exhibits the strongest binding activity to the EBEs in the CsGRAS9 promoter and plays a pivotal role in regulating CsGRAS9 expression.

In general, Xanthomonas TALEs (e.g., AvrBs3, PthXo2, and PthA4) target the EBE regions of promoters containing TATA-box motifs (Wang et al. 2017). We cloned and sequenced the CsGRAS9 promoter from grapefruit and analyzed it using PlantCARE. Several core TATA-box/AT-TATA-box motifs were identified in the CsGRAS9 promoter, overlapping with the EBE regions for PthA4, PthA5, and PthA6 (Fig. 3A and Supplementary Fig. S9). A palindromic sequence was also presented in these EBE regions (Supplementary Fig. S7A). To further confirm the interaction of PthA5 and PthA6 with the CsGRAS9 promoter in vivo, we fused a 2k-bp fragment of the CsGRAS9 promoter to the pLacZ reporter vector. As shown in Fig. 2F, PthA5 and PthA6 positively regulated β-galactosidase (LacZ) reporter gene expression, demonstrating that PthA5 and PthA6 directly bound to the CsGRAS9 promoter in vivo.

Gene suppression of CsGRAS9 was observed in Xcc049E/pthA4-, Xcc049E/pthA5-, and Xcc049E/pthA6-infected grapefruit leaves, but not in Xcc049E/EV-infected leaves (Fig. 3B). To confirm that PthA5 inhibits gene expression by binding to the EBEs in the CsGRAS9 promoter, we used Agrobacterium-mediated transient β-glucuronidase (GUS) and luciferase (LUC) expression assays in N. benthamiana. The full-length CsGRAS9 promoter was fused with pCAMBIA1381::GUS and pGreenII0800-LUC vector to construct pCsGRAS9::GUS and pCsGRAS9::LUC, respectively, where the reporter genes were driven by the CsGRAS9 promoter (Fig. 3C). As expected, in leaves co-transformed with pCsGRAS9::GUS and PthA5 or PthA6, we observed a significant reduction in GUS expression, indicating that PthA5 and PthA6 can inhibit CsGRAS9 expression (Fig. 3D). Histochemical staining also showed that CsGRAS9 expression was suppressed by PthA5 and PthA6. Similar results were obtained using a transient dual-luciferase assay in N. benthamiana leaves, where both PthA5 and PthA6 significantly decreased luciferase activity and signal driven by the CsGRAS9 promoter compared to the pHB empty vector (Fig. 3E-H). These findings indicate that PthA5 and PthA6 directly and negatively regulate CsGRAS9 expression.

Natural variation in citrus GRAS9 promoters and importance of variations in response to Xcc infection

To verify the role of GRAS9s in response to Xcc infection, we explored the development of canker symptoms in canker-resistant Hong Kong kumquat (F. hindsii). When inoculated with both PthA4 and either PthA5 or PthA6, we observed significantly more severe canker symptoms in Hong Kong kumquat compared to those infected with Xcc049E/pthA4 alone (Fig. 4A). Increased bacterial growth in Hong Kong kumquat was also noted with the presence of PthA5 or PthA6 (Supplementary Fig. S10A). The GRAS isoforms from sweet orange and wild tetraploid Hong Kong kumquat were designated CsGRAS9 and FhGRAS9, respectively. To explore the relationship between resistance and promoter polymorphisms, we amplified and sequenced the core 740-bp regions of the CsGRAS9 and FhGRAS9 promoters containing EBE regions (Fig. 4B). Sequence alignment revealed a 558-bp deletion (InDel) in the promoter region of FhGRAS9 (Supplementary Fig. S10B). Different promoter genotypes were identified in Hong Kong kumquat, including a 182-bp fragment (designated pFhGRAS9-P2) and a 740-bp fragment identical to the CsGRAS9 promoter (designated pCsGRAS9/FhGRAS9-P1) (Supplementary Fig. S10B-C). The 182-bp pFhGRAS9-P2 lacked several AT-TATA/TATA-box and EBE regions for TALEs but contained a cis-element similar to the SA-inducible activation sequence-1 (as−1). Additionally, the EMSA probe ProCsGRAS9, which contains EBEs for PthA4, PthA5, and PthA6, was also detected in the FhGRAS9-P1 promoter.

Two truncated plasmids, pFhGRAS9/CsGRAS9-P1 (referred to as pCsGRAS9-P1) and pFhGRAS9-P2, were constructed using the pLacZ reporter vector. To determine whether the deletion variant in FhGRAS9 promoter is crucial for interaction with PthA5 and PthA6, we conducted yeast one-hybrid (Y1H) assays in EGY48 strains. The results showed that PthA5 and PthA6 could not target the FhGRAS9-P2 promoter to induce LacZ reporter gene expression but were able to directly bind to the FhGRAS9/CsGRAS9-P1promoter in vivo (Fig. 4C). Next, we used Agrobacterium-mediated transient GUS expression assays in N. benthamiana to assess the promoter activities of pCsGRAS9-P1 and pFhGRAS9-P2. The 740-bp pCsGRAS9-P1 and 182-bp pFhGRAS9-P2 promoters were fused with the GUS reporter vector pCAMBIA1381 to construct pCsGRAS9-P1::GUS and pFhGRAS9-P2::GUS, respectively. As shown in Fig. 4D, the relative GUS activity of pCsGRAS9-P1 significantly decreased to 84.10% and 84.48% after co-transformation with PthA5 or PthA6, respectively. This result is consistent with findings that PthA5 and PthA6 inhibit CsGRAS9 expression (Fig. 3D). Conversely, when pFhGRAS9-P2::GUS was co-infiltrated with pHB-PthA5 or pHB-PthA6 into N. benthamiana, GUS expression increased by 127.91% and 123.14%, respectively, compared to cells co-transformed with the pHB empty vector (Fig. 4E). Histochemical staining also confirmed that PthA5 and PthA6 suppressed the promoter activity of FhGRAS9/CsGRAS9-P1 but not that of FhGRAS9-P2 (Fig. 4D and E). Consistent with the quantitative analysis of GUS activity, expression analysis indicated that PthA4, PthA5, and PthA6 induced the FhGRAS9 expression in Hong Kong kumquat leaves (Fig. 4F). These findings suggest that natural variation in the FhGRAS9 promoter disrupts its interaction with TALEs, leading to increased FhGRAS9 expression.

Citrus germplasm, which includes wild species resistant to citrus canker, provides an abundant genetic resource for unraveling the molecular mechanisms of plant-microbe interactions (Duan et al. 2022). The genome of 14 citrus species, including C. grandis, C. hongheensis, C. australasica, C. medica, C. ichangensis, C. linwuensis, C. mangshanensis, C. clementina, C. sinensis, C. reticulata, F. hindsii, Murraya paniculata, Poncirus trifoliata, and A. buxifolia, are available in the Citrus genome database (http://citrus.hzau.edu.cn/orange/). To investigate whether natural variations in the GRAS9s promoters contribute to differential canker symptoms in citrus plants, we retrieved the 2000 bp GRAS9 promoter sequences from the Citrus genome database. Alignment of these promoters revealed that InDel and SNP variations were concentrated in the EBE regions, causing mismatches between RVDs and EBEs (Fig. 4G). Specifically, 26-bp, 12-bp, and 10-bp deletions were detected in the GRAS9 promoters of C. medica, M. paniculata, C. australasica, and C. linwuensis, respectively. Phylogenetic analysis of the GRAS9 promoters showed that the TALE-targeted EBE regions in C. sinensis clustered with those in susceptible species like C. clementina and C. reticulata (Fig. 4G). C. medica has been identified as strongly resistant to Xcc (Fu et al. 2020), while orange jessamine (M. paniculate) is a non-host of Xcc, displaying only faint chlorosis as the most severe symptom at 26 dpi (Ference et al. 2020). To validate the binding efficiency of PthA4, PthA5, and PthA6 with the C. medica and M. paniculate GRAS9 promoters, we performed EMSA assays using two 35 bp double-stranded DNA fragments (ProCmGRAS9 and ProMpGRAS9) containing sequences aligned with EBEs for TALEs (Fig. 4H and I). EMSA results confirmed that natural deletion variations in the EBE regions of the CmGRAS9 and MpGRAS9 promoters slightly affected the binding efficiency of PthA4, PthA5, and PthA6 (Fig. 4H and I). These findings suggest that resistance to citrus canker is associated with natural variations in the EBE regions of the GRAS9s promoters, which influence the expression patterns of GRAS9 under Xcc infection.

TALEs suppress disease resistance to citrus canker through regulating CsGRAS9 expression

Considering that TALEs inhibited CsGRAS9 homologs expression in susceptible citrus plants, but not in kumquat, we further examined whether GRAS9 expression attributed to citrus canker resistance. First, we obtained the coding sequence (CDS) of CsGRAS9 from grapefruit leaves. Sequence analysis revealed that the CsGRAS9 transcript was 1345 bp, encoding a 49.7 kDa protein with 448 amino acids. Among the GRAS family proteins in citrus, CsGRAS9 is most closely related to CsGRAS3 (Zhang et al. 2019). Phylogenetic analysis of CsGRAS9 and its orthologs in various citrus species, as well as tomato, rice, Arabidopsis, and Tamarix hispida, was conducted by aligning amino acid sequences. CsGRAS9 clustered in the same clade as known plant development-related GRAS proteins, including Arabidopsis AtSCL32, tomato SlGRAS38 and SIGRAS26, Tamarix hispida ThSCL32, and rice OsSCL32-2 (Fig. 5A). Multiple sequence alignment further revealed high amino acid identity among citrus GRAS9 proteins (Supplementary Fig. S11), indicating evolutionary conservation across citrus species. CsGRAS9 shared the highest homology with GRAS9 from C. clementina and C. reticulata. Moreover, the MEME suite (http://meme-suite.org/tools/meme) identified 20 conserved motifs in GRAS9 proteins of citrus species and their orthologs in other plants. In general, GRAS9 proteins within the same clade contain similar motifs. Motifs 1–14 and motifs 16–17 were conserved in citrus GRAS9 proteins, while three motifs (motifs 1, 3, and 11) and four motifs (motifs 2, 9, 11, and 13) were also found at the N-terminal and C-terminal regions of SlGRAS26, SlGRAS38, CsGRAS3, ThSCL32, and OsSCL32-2 (Supplementary Fig. S12). To assess whether CsGRAS9 has transcriptional activity, CsGRAS9 and CsLOB1 were cloned into the pGBKT7 vector. The BD-CsGRAS9 and BD-CsLOB1 constructs, along with the empty vector BD, were individually co-transformed with pGADT7 into yeast strain AH109. Yeast cells carrying BD-CsLOB1 and BD served as positive and negative controls, respectively. Yeast cells carrying BD-CsGRAS9 grew and turned blue on SD/−Trp/−Leu/X-α-Gal and SD/-Trp-Leu-His/X-α-Gal selection medium, indicating that CsGRAS9 possesses transcriptional activity (Fig. 5B). Subcellular localization of CsGRAS9 was determined by transfecting a CsGRAS9-GFP fusion construct into N. benthamiana leaves. Fluorescence signals in N. benthamiana overexpressing CsGRAS9-GFP were detected in the nucleus and cytosol (Fig. 5C).

Fig. 1

Effect of PthA5 and PthA6 in canker formation. A Grapefruit (Citrus paradisi) and kumquat (Fortunella margarita) leaves were infiltrated with Xcc003 and Xcc086 strains (OD₆₀₀ = 1.0). Canker symptoms were recorded at 21 days post-inoculation (dpi). B Alignment of repeat-variable diresidues (RVDs) of PthA6, PthA5, PthA1, and PthA4. PthA6 (from Xcc003) and PthA5 (from Xcc086) were compared with PthA1 and PthA4 from X. citri strains. Black lines (-) indicate missing RVDs. Different RVDs are highlighted in gray and colored, corresponding to nucleotide association and DNA base specificities. Green: NI recognizes adenine (A); Red: NG recognizes thymine (T); Blue: HD recognizes cytosine (C); Orange: N* recognizes C and T; Purple: NS recognizes ACT and guanine (G). C Grapefruit leaves were infiltrated with Xcc suspensions (OD₆₀₀ = 1.0) of tal-free deletion mutant Xcc049E/EV and complemented strains Xcc049E/pthA5 or Xcc049E/pthA6. Canker symptoms were evaluated at 21 dpi. Microscopic observation of thin cross-sections in the circled regions. D Effect of PthA5 and PthA6 on bacterial growth in grapefruit. Grapefruit leaves were infiltrated with Xcc suspensions (OD₆₀₀ = 0.8). Bacterial growth assays of Xcc strains were monitored at 0, 1, 2, 3, 4, 8, and 12 dpi. Bacterial growth is presented in mean colony-forming units (CFU)/cm², indicating PthA5 and PthA6 enhance bacterial growth in citrus host. Error bars represented standard deviations of three biological replicates. Asterisks denote statistically significant differences in bacterial-inoculated leaves, as determined by two-way ANOVA with Dunnett’s test: **p < 0.01, ***p < 0.001, and ****p < 0.0001. E Grapefruit leaves were infiltrated with Xcc suspensions (OD₆₀₀ = 0.6) of Xcc049E/pthA5, Xcc049E/pthA6, and mixed suspensions of Xcc049E/pthA4 and Xcc049E/pthA5 or Xcc049E/pthA6 at a 1:1 ratio. Xcc049E/EV and Xcc049E/pthA4 were used as the negative and positive controls, respectively. Citrus canker symptoms were evaluated at 16 dpi. F Bacterial growth of Xcc strains in grapefruit leaves was monitored at 25 dpi. Error bars represent standard deviation of three biological replicates. Asterisks denote statistically significant differences, as determined by one-way ANOVA with Dunnett’s test: *p < 0.05, ****p < 0.0001

Fig. 2

PthA4, PthA5, and PthA6 directly bind to the promoter of CsGRAS9. A Composition of the effector binding elements (EBEs) in the CsGRAS9 promoter. Specific nucleotides in the predicted EBEs that match individual RVD sequences of PthA5, PthA6, PthA1 and PthA4 from X. citri are shown above. AnnoTALE’s "Predict and Intersect Targets" tool, incorporating TALENT 2.0, was used to efficiently scan the host promoterome and identify target EBE regions for TALEs. Comprehensive analysis of the putative target EBEs in the CsGRAS9 promoter for PthA1, PthA4, PthA5, and PthA6 was performed using Target Finder and AnnoTALE’s "Predict and Intersect Targets" tool. The EBEs in the CsGRAS9 promoter are listed in Supplementary Tables S5-8. ( +) indicates the plus strand of the DNA sequence. B Prediction of the multichain PthA5–EBE_proCsGRAS9 complex. C-E Confirmation of the binding capacity of PthA6 (C), PthA5 (D), and PthA4 (E) to the EBE regions in the CsGRAS9 promoter using electrophoretic mobility shift assays (EMSA). The probe (ProCsGRAS9) is a Cy5-labeled fragment generated from the EBE region of the CsGRAS9 promoter. Competitive EMSAs showed that purified PthA6-His, PthA5-His, and PthA4-His proteins bound to the Cy5-labeled probes and cold probes (unlabeled competitors). Cold probes acted as competitors at increasing concentrations of 1 × , 100 × , 400 × , and 800 × . The presence or absence of proteins and probes is indicated by " + " or "–". Black arrows denote specific shifts and free probes (below). F Yeast one-hybrid (Y1H) assays of PthA5 and PthA6 binding to CsGRAS9 promoter. The coding sequences of PthA5 and PthA6 were fused to pB42AD, and the CsGRAS9 promoter was fused to the pLacZ vector. Co-transformations of pB42AD-PthA5/pB42AD-PthA6 and pLacZ - proGRAS9 were carried out in yeast strain EGY48. Empty vector pLacZ and pLacZ - proGRAS9 were introduced into EGY48 cells carrying pB42AD as negative controls, while the transformants with pB42AD-PthA5/pB42AD-PthA6 and empty vector pLacZ were also used as additional negative controls

Fig. 3

PthA5 and PthA6 regulate the transcription of CsGRAS9. A An illustration of the 2-kb promoter region of CsGRAS9 used for yeast one-hybrid (Y1H), glucuronidase (GUS), and luciferase (LUC) assays. Red blocks represent core promoter TATA-box/AT-TATA-box motifs, and blue blocks represent predicted effector binding element (EBE) region. The same components are highlighted with the same color across experiments. B Relative expression of CsGRAS9 in the leaves of grapefruit measured at 24- and 48-h post-inoculation (OD₆₀₀ = 1.0). CsEf1a was used as a constitutive reference gene. Three biological replicates are presented as mean ± SEM values. Different letters indicate significant differences (p < 0.05) analyzed by two-way ANOVA with Tukey's test. C Schematic diagrams of plasmids used in the transient expression assays in N. benthamiana. The coding sequences (CDS) of PthA5 and PthA6 were cloned into pHB vector as the effector. A 2000-bp promoter fragment of CsGRAS9 containing the EBE regions was fused to pCAMBIA1381::GUS and pGreenII0800-LUC vectors to drive the expression of the GUS and LUC reporter genes, respectively. D GUS staining and activity assays performed in N. benthamiana leaves by infiltration with the labeled effector and reporter combinations. Empty pHB was co-transformed with vector pCsGRAS9::GUS as the negative control. Results are presented as mean ± SEM values for three biological replicates. Asterisks indicate statistically significant differences using Tukey's test (****p < 0.0001). E–H Effector constructs pHB-PthA6, pHB-PthA5 and reporter vector pCsGRAS9::LUC were used in transient expression assays to assess luciferase activity. LUC fluorescence images of N. benthamiana leaves by infiltration with the pHB-PthA6 (E), pHB-PthA5 (G) and reporter combinations were captured 72 h post-infiltration. Relative luciferase activity of CsGRAS9 promoter was determined by the ratios of LUC to REN (F and H). The results show that both PthA6 (F) and PthA5 (H) repressed CsGRAS9 transcriptional activity. Mean ± SEM values are shown for three biological replicates. Asterisks denote statistically significant differences with Student’s t-test (****p < 0.0001, **p < 0.01)

Fig. 4

Sequence divergence of CsGRAS9 in citrus cultivars. A Leaves of Hong Kong kumquat (F. hindsii, Shan Jin Gan) were inoculated with Xcc049E/pthA4 (control) and a mixed suspension (OD₆₀₀ = 1.0) of Xcc049E/pthA4 + pthA5 and Xcc049E/pthA4 + pthA6 at a 1:1 ratio. Canker symptoms were observed 16 days post-inoculation. B Illustration of the 740-bp promoter region of CsGRAS9 and the 182-bp promoter region of FhGRAS9-P2 used for Y1H, GUS, and LUC assays. Pink blocks represent core promoter TATA-box/AT-TATA-box motifs, blue blocks represent predicted EBE regions and green blocks represent as-1 motifs. C Schematic of plasmids used in Y1H assays. The CDS of pthA5 and pthA6 were cloned into pB42AD vector as the effector, respectively. Truncated promoter fragments of CsGRAS9 (CsGRAS9-P1/FhGRAS9-P1) and FhGRAS9 (FhGRAS9-P2) were fused with pLacZi reporter genes. Constructs were co-transformed into yeast EGY48 strains, grown on SD/-Leu/-Trp medium, and evaluated on SD/-Leu/-Trp/X-gal plates. Negative controls were co-transformants with empty vectors pB42AD or pLacZi. D-E GUS staining and expression assays in N. benthamiana leaves transformed with labeled effector and reporter constructs via transient expression assays. The CDS of pthA5 and pthA6 were cloned into the pHB vector. Truncated promoter fragments (740-bp CsGRAS9-P1 in D and 182-bp FhGRAS9-P2 in E) were fused with pCAMBIA1381::GUS. Negative controls included empty pHB co-transformed with vectors pCsGRAS9-P1::GUS or pFhGRAS9-P2::GUS. Three biological replicates were presented as mean ± SEM values. Asterisks indicate statistically significant differences using one-way ANOVA followed by Tukey's test (****p < 0.0001, ***p < 0.001, **p < 0.01). F Relative expression of FhGRAS9 in Hong Kong kumquat (Shan Jin Gan) leaves 48 h post-inoculation (OD₆₀₀ = 1.0). CsEf1a was used as a reference gene. Mean ± SEM values from three biological replicates are shown. Asterisks indicate significant differences between bacterial-inoculated leaves using one-way ANOVA with Tukey's test (*p < 0.05, **p < 0.01). G Alignment of the allelic variations of the EBE regions of GRAS9 promoters in citrus species. Sequences show the 26-bp and a 12-bp deletion in the CsGRAS9 homologs of M. paniculata, C. australasica and C. medica. Conserved nucleotides are in blue, while nucleotide variations are highlighted in black or represented by black dots for missing nucleotides. H-I Binding of purified His-tagged proteins PthA6-His, PthA5-His, and PthA4-His to EBE regions in the promoters of GRAS9 homologs in the M. paniculata, C. australasica and C. medica. EMSA assays were performed with Cy5-labeled probes for ProCmGRAS9 (H) and ProMpGRAS9 (I) and cold competitors. Cold probes were used as competitors with 1 × or 100 × excess over Cy5-labeled probes. Specific shifts and free probes are indicated by arrows. Symbol “ + ” or “–” indicates the presence or absence of protein and probes. Results are from three independent experiments

Fig. 5

Subcellular localization, transcriptional activities and function of CsGRAS9. A Phylogenetic tree of CsGRAS9, its closest homolog (CsGRAS3), homologous proteins in Citrus spp., and orthologous proteins in Arabidopsis, rice, and tomato, constructed using MEGA X. Branch lengths represent the number of substitutions per site. B Analysis of CsGRAS9 transcriptional activity in yeast. The CDS of CsLOB1 and CsGRAS9 were fused to pGBKT7 vector, respectively. The pGBKT7-CsGRAS9 construct was co-transformed with pGADT7 into yeast strain AH109. Yeast cells expressing BD-CsGRAS9 were spotted on SD/-Trp/-Leu, SD/-Trp/-Leu/-His media, and selection media containing X-α-gal. pGBKT7-CsLOB1 and pGADT7 served as positive and negative controls, respectively. C Subcellular localization of CsGRAS9 in N. benthamiana leaves. CsGRAS9 was fused with YFP in the pHB vector (35S::CsGRAS9-YFP). Fluorescence (yellow for YFP, blue for DAPI) was observed using a confocal microscope. Scale Bars: 50 and 75 μm. D CsGRAS9 specifically activated by dTALEs. The region targeted by dGRAS9 in the CsGRAS9 promoter is highlighted in yellow. Red nucleotides represent the translation start site. E Grapefruit leaves were infiltrated with Xcc suspensions (OD₆₀₀ = 0.6) of Xcc049E/EV and Xcc049E/dGRAS9. Citrus canker symptoms and bacterial growth were evaluated at 16 days post-inoculation. Error bars represent the standard deviation of three biological replicates. Asterisks indicate statistically significant differences (****p < 0.0001) using Student’s t-test. F Relative expression of CsGRAS9 in grapefruit leaves was measured 48 h post-inoculation (OD₆₀₀ = 1.0). CsEf1a was used as the internal control. Data from three biological replicates are presented as mean ± SEM values. Asterisks denote significant differences using Student’s t-test (*p < 0.05, **p < 0.01, ****p < 0.0001). G Grapefruit leaves were inoculated using a pinprick method with Xcc suspensions (OD₆₀₀ = 1.0) of mixed suspensions Xcc003 + Xcc049E/EV, Xcc003 + Xcc049E/dGRAS9, Xcc086 + Xcc049E/EV, and Xcc086 + Xcc049E/dGRAS9 in a 1:1 ratio. Positive controls included inoculations with Xcc003 + Xcc049E/EV and Xcc086 + Xcc049E/EV suspensions. Citrus canker symptoms were assessed 8 days after inoculation. H-I Grapefruit leaves were infiltrated with Xcc suspensions (OD₆₀₀ = 1.0). Xcc049E/EV and Xcc049E/dGRAS9 inoculations were performed, alongside Xcc003 + Xcc049E/EV (H) and Xcc086 + Xcc049E/EV (I) inoculations, to evaluate citrus canker severity. Symptoms were assessed 12 days post-inoculation. J Bacterial growth of Xcc strains in grapefruit leaves was monitored at 1, 4, 7, 10 days post-inoculation. Data are presented in mean colony-forming units (CFU)/cm². Error bars represent standard deviations from three biological replicates. Different letters denote statistically significant differences using two-way ANOVA with Tukey's test (p < 0.05)

A designer transcription activator-like effector (dTALE), designated dGRAS9, was constructed to target the 17-bp EBE of the CsGRAS9 promoter located 212 bp upstream of the ATG start codon (Fig. 5D). The EBE targeted by dGRAS9 is positioned 1 bp upstream of the predicted ATTATA and TATA-box in the CsGRAS9 promoter. Expression of the dGRAS9 protein in Xcc049E strains harboring dGRAS9 (Xcc049E/dGRAS9) was confirmed by western blotting using an anti-flag antibody (DYKDDDDK), with the dGRAS9 protein appearing at the expected size of 120.2 kDa (Supplementary Fig. S13). In grapefruit leaves, bacterial growth was reduced in Xcc049E/dGRAS9 -infected plants compared to Xcc049E/EV, within 18 dpi, suggesting that CsGRAS9 plays a role in plant resistance (Fig. 5E). Additionally, Xcc049E/dGRAS9 activated CsGRAS9 expression within 48 h post-inoculation (Fig. 5F).

To further investigate whether CsGRAS9 contributes to citrus canker resistance, we observed canker symptom development in grapefruit leaves inoculated via the pinprick and infiltration methods using mixed bacterial suspensions of Xcc003/EV or Xcc086/EV combined with Xcc049E/dGRAS9 at a 1:1 ratio. Inoculation with Xcc003/EV + Xcc049E/dGRAS9 and Xcc086/EV + Xcc049E/dGRAS9 suppressed canker symptom development compared with the controls (Fig. 5G-I and Supplementary Fig. S14). Consistent with these findings, reduced bacterial growth in grapefruit leaves, induced by CsGRAS9 expression, confirmed that CsGRAS9 plays a role in citrus canker resistance (Fig. 5J).

Evaluation of CsGRAS9 mutants

Natural variations in the promoters of GRAS9 genes are associated with their expression and resistance to Xcc infection in citrus germplasms. To explore the role of CsGRAS9 in canker resistance, several mutations in the CsGRAS9 promoter (proCsGRAS9) were generated using the PTG/Cas9 system (Fig. 6A). One sgRNA was specifically designed to target the EBE regions of the CsGRAS9 promoter (Fig. 6B). The recombinant vector pCAMBIA1300-pYAO-Cas9/pGRAS9 was introduced into ‘Anliu’ sweet orange epicotyl via Agrobacterium-mediated transformation. As a result, 19 eGFP-positive T0 lines were successfully regenerated and survived in the greenhouse. Among these, three lines were identified as wild-type (Fig. 6A). Hi-Tom sequencing of the 19 regenerated lines tentatively revealed that 9 of them were homozygous mutants. Sweet orange has two alleles, each with conserved EBEs in the promoter regions. Sanger monoclonal sequencing confirmed the mutant genotypes in both alleles of the edited lines. Sequence chromatograms at the target sites showed that the most frequent mutation was a 1 bp insertion. Line #23 was identified as a homozygous mutant (−62/−62), featuring a 62 bp deletion in the EBE regions of the CsGRAS9 promoter (Fig. 6A). Among the two chimeric proCsGRAS9-edited mutants, line #18-2 exhibited three mutation types (−86/+1/+1), while line #9 had four mutation types (+1/−86/−62/WT) at the target site (Fig. 6A). Additionally, line #13, a biallelic proGRAS9 mutant (−7/−9), had 7 bp and 9 bp deletions upstream of the EBE regions for TALEs (Supplementary Fig. S15A). To assess potential off-target effects, two possible off-target sites containing up to four mismatches were identified in the targeted CsGRAS9 promoter using CRISPR P v2.0. Sanger sequencing of the off-target regions in edited lines (#9, #13, #18-2, and #23) revealed no off-target mutations (Supplementary Fig. S16).

To further validate the role of natural variations in promoter activity, CsGRAS9 promoters from three proGRAS9 lines (#18-2, #23, and #9) were cloned into the pCAMBIA1381::GUS vector. Constructs pCsGRAS9 Type I (−62 bp)::GUS, pCsGRAS9 Type III (−86 bp)::GUS, and pCsGRAS9 Type I (+1 bp)::GUS were co-infiltrated with PthA5 or PthA6 into N. benthamiana leaves. The results showed higher GUS activity for the pCsGRAS9 Type I (−62 bp) and Type III (−86 bp) promoters, while decreased GUS activity was observed for the pCsGRAS9 Type I (+1 bp) promoter, which was consistent with wild-type levels (Fig. 6C and D, and Supplementary Fig. S17). These findings indicated that gene-edited variations in the CsGRAS9 promoter can regulate CsGRAS9 transcription by disrupting the TALE-EBE interactions. Overexpression of CsGRAS9 suppressed the development of canker symptoms (Fig. 5). As expected, typical canker symptoms were observed in wild-type ‘Anliu’, while only slight lesions appeared in the proGRAS9 lines (#9, #18-2, and #23) following Xcc infection via infiltration (Fig. 6E and Supplementary Fig. S18). Correspondingly, bacterial titers of Xcc were significantly reduced in the proGRAS9 lines (#9, #18-2, and #23) compared to those in wild-type ‘Anliu’ (Fig. 6F). Interestingly, canker symptoms in the proGRAS9 line #13 were similar to those in the wild-type (Supplementary Fig. S15B), which is consistent with previous studies that typically excluded chimeric-edited lines due to suboptimal editing types (Yu et al. 2024). Despite this, stable CsGRAS9 expression and canker resistance were observed in the chimeric-edited line #9 (including the wild-type segment) (Supplementary Fig. S19). PthA4 and PthA5 downregulated CsGRAS9 expression during Xcc infection (Fig. 3B). However, mutations in the EBE regions of the CsGRAS9 promoter prevented the suppression of CsGRAS9 by Xcc049E/pthA4 and Xcc049E/pthA5 (Fig. 6G). Additionally, the expression of CsLOB1 was significantly lower in the proGRAS9 mutants (#9, #18-2, and #23) than in the wild type (Fig. 6H). These alterations in CsGRAS9 and CsLOB1 expression help explain the reduced canker symptoms observed in proGRAS9-edited lines.

CsGRAS9 regulates the expression of CsTAC1 and GA-related genes

To explore the biochemical functions of CsGRAS9 and map its molecular interaction network, we used STRING to identify protein-protein interactions (PPIs) (Supplementary Fig. S20A). We identified nine proteins interacting with CsGRAS9, including NAC domain-containing protein 33 (SMB/NAC33, orange1.1t03221), Dof zinc finger protein DOF4.6 (DOF4.6, Cs5g01740), Calmodulin-like protein 30 (CML30, Cs5g28300), TELOMERASE ACTIVATOR1 (TAC1, Cs5g06630), LOB DOMAIN-CONTAINING PROTEIN 29 (ASL16/LBD29, Cs5g30050), LOB DOMAIN-CONTAINING PROTEIN 16 (LBD16, Cs5g30060), BOLA-LIKE PROTEIN 1 (BolA1, orange1.1t02810) and two DELLA proteins, Scarecrow-like protein 22 (SCL26/GRAS23, Cs6g09040), and MYCORRHIZ A-INDUCED GRAS (MIG1, orange1.1t01684). Quantitative real-time PCR (qRT-PCR) assays showed that CsGRAS9 activated the expression of DOF4.6 and CML30, while repressing the expression of TAC1 (Supplementary Fig. S20B and C). To validate the interaction between CsGRAS9 and CsTAC1, we performed a yeast two-hybrid (Y2H) analysis. Yeast strains co-transformed with AD-CsGRAS9 and BK-CsTAC1 grew on selective medium (SD/-Trp/-Leu/-His/-Ade) and exhibited α-galactosidase activity, confirming that CsGRAS9 interacted with CsTAC1 that is the orthologous to AtTAC1, which responses to auxin signaling and induce telomerase activity in Arabidopsis thaliana (Supplementary Fig. S20D). Studies have shown that DELLA proteins act as repressors of the gibberellin (GA) signaling pathway (Neves et al. 2023). Infection by X. campestris pv. vesicatoria (Xcv) triggers defense responses by inducing the expression of GRAS subfamily members, such as SlGRAS4, SlGRAS13, and SlGRAS6. To investigate the role of CsGRAS9 in the regulation of GA biosynthesis and metabolism during Xcc049E/dGRAS9 infection, we assessed the expression of key GA-related genes. The expression of the GA biosynthetic enzyme GA3 oxidase genes (GA3ox2 and GA3ox3, encoded by Cs4g20350 and Cs4g17110, respectively) and GA20 oxidase 1 (GA20ox1, encoded by Cs1g09880) were upregulated in CsGRAS9-overexpressing leaves (Supplementary Fig. S21A-C). In contrast, the expression of GA methyltransferase GAMT2 (Cs9g03520) was reduced by approximately 2-fold (Supplementary Fig. S21D). These findings suggest that CsGRAS9 may contribute to citrus canker resistance by modulating the expression of CsTAC1 and GA-related genes, potentially influencing the GA signaling pathway.

Fig. 6

CsGRAS9 mutants of Citrus sinensis cv. 'Anliu' sweet orange generated by genome editing of the promoter of CsGRAS9. A Chromatograms of PCR products from wild-type (WT) C. sinensis cv. 'Anliu' sweet orange serve as the negative control. Colony sequencing chromatograms from the T0 proCsGRAS9-edited C. sinensis cv. 'Anliu' line #18–2 show three types of insertions and deletions at the target site: Type I (+ 1 bp insertion), Type II (+ 1 bp different insertion), and Type III (−86 bp deletion). Representative chromatograms are shown. In line #23, mutations in both alleles of the CsGRAS9 promoter are shown, with a 62 bp deletion as the only type of mutation. Colony sequencing chromatograms from line #9 show four types of mutations: Type I (+ 1 bp insertion), Type II (−62 bp deletion), Type III (−86 bp deletion), and Type IV (WT). EBE regions are highlighted in red. “–” indicates deletions, “ + ” indicates insertions. The underlined black (CCT) indicates PAM. B Schematic representation of the CRISPR construct containing the CsGRAS9 gRNA. The purple block indicates the sgRNA scaffold, and red nucleotides were chosen for the gRNA in the EBE region of the CsGRAS9 promoter; the protospacer-adjacent motif (PAM) is underlined. C-D GUS staining and expression in N. benthamiana leaves transformed with effector and reporter combinations via transient expression assays. The CDS regions of pthA5 and pthA6 were cloned into the pHB vector as effectors. Truncated promoter fragments of the 604-bp Type III (−86 bp) promoter from T0 proCsGRAS9 line #18–2 were fused to pCAMBIA1381::GUS. Effector construct pHB-pthA5 and reporter vector T0 proCsGRAS9 line #18–2 Type III (−86 bp)::GUS were used in GUS assays (C). Asterisks indicate statistically significant differences (****p < 0.0001) using Student’s t-test. Truncated promoter fragments of the 628-bp Type I (−62 bp) promoter from T0 proCsGRAS9 line #23 were fused to pCAMBIA1381::GUS. Effector constructs pHB-pthA5 or pHB-pthA6 and reporter vector T0 proCsGRAS9 line #23 Type I (−62 bp)::GUS were used for GUS assays (D). Asterisks indicate statistically significant differences by one-way ANOVA with Tukey’s test (**p < 0.01, ***p < 0.001). E Leaves of WT and T0 proCsGRAS9-edited C. sinensis cv. 'Anliu' lines (#9, #18–2, and #23) were infiltrated with Xcc suspensions (OD₆₀₀ = 1.0) of Xcc049E/EV, Xcc049E/pthA4, and mixed suspensions of Xcc049E/pthA4 + pthA5 and Xcc049E/pthA4 + pthA6 in a 1:1 ratio. Citrus canker symptoms were evaluated 21 days after inoculation. F Bacterial growth of Xcc strains in the leaves of WT and proCsGRAS9 mutants of 'Anliu' sweet orange was monitored 17 days post-inoculation. Error bars represent standard deviations of three biological replicates. Different letters indicate statistically significant differences by two-way ANOVA with Tukey’s test (p < 0.05). G-H Relative expression of CsGRAS9 (G) and CsLOB1 (H) in leaves of WT and T0 proCsGRAS9-edited C. sinensis cv. 'Anliu' lines (#9, #18–2, and #23) measured 48 h post-inoculation (OD₆₀₀ = 1.0). CsEf1a was used as a constitutive standard. Different letters indicate statistically significant differences by two-way ANOVA with Tukey’s test (p < 0.05)

Discussion

Xanthomonas TALEs undergo evolutionary changes in response to selective pressures exerted by their hosts (Gochez et al. 2018). The evolutionary dynamics of the LJ207-7 strain indicated that genetically monomorphic bacteria adapt to environmental changes and diverse hosts through the horizontal gene transfer of TALEs (Ruh et al. 2017; Richard et al. 2017). To further investigate this adaptive mechanism, in the present study, we sequenced the complete pthA5 and pthA6 genes to determine their target specificities and shed light on pathogen population behavior and host aggressiveness (Fig. 1 and Supplementary Figs. 1–3). These findings support the hypothesis that recombination and horizontal gene transfer drive convergent evolution among TALEs in Xanthomonas strains (Gochez et al. 2018). Understanding the target specificity of TALEs and mechanisms underlying the regulation of host susceptibility or resistance is critical for breeding citrus plants with stable resistance to pathogens.

FuncTAL and DisTAL categorize Xanthomonas diversity on the basis of functional binding specificities and ecological diversity, as determined by RVD alignment (Pérez-Quintero et al. 2015). In the present study, FuncTAL analysis revealed that PthA5 and PthA6 share functional similarities with PthA1 (Supplementary Fig. S2). After infection with Xcc049E/pthA5, the expression of CsLOB2 and CsDIOX increased slightly (Supplementary Fig. 5), consistent with their upregulation in response to PthA1 (Abe et al. 2016). Moreover, PthA5 and PthA6 promoted the formation of water-soaked lesions and bacterial colonization in grapefruit leaves, independent of CsLOB1 activation (Supplementary Fig. 4). Typical hypertrophy and hyperplasia were observed after inoculation with Xcc049E/pthA5 and Xcc049E/pthA6, highlighting the need to further investigate mechanisms underlying the actions of PthA5 and PthA6. In grapefruit, a susceptible genotype, the expression level of CsGRAS9 was suppressed following Xcc inoculation, whereas that of FhGRAS9, the homolog in Hong Kong kumquat, was expressed (Fig. 4F). This differential response suggests a key role of GRAS9 homologs in citrus resistance mechanisms.

Xanthomonas species enhance bacterial virulence and promote symptom formation by regulating the expression of TALE-targeted genes (Timilsina et al. 2020). Several TALEs, including PthA4, PthAW, PthA∗, PthB, and PthC consistently recognize CsLOB1 during plant–pathogen interactions, following a well-characterized mechanism (Teper et al. 2020). Deletion events within the repeat arrays of TALEs do not disrupt their conserved functions (Ji et al. 2016; Read et al. 2016). Our findings indicate that both PthA5 and PthA1 may bind to more than four EBEs in the CsGRAS9 promoter, overlapping with EBEs for PthA4 and PthA6 (Fig. 2A). Protein-DNA docking, which is widely used to predict DNA-binding site architecture, revealed the interactions of TALEs with their target DNA (Deng et al. 2012). The crystal structures of Xanthomonas TALEs bound to DNA have established a foundation for exploring TALE functions with high modularity (Li et al. 2023). Structural analysis confirmed that the recognition region of PthA5, composed of 17 direct repeats, specifically binds to the sense strand of the CsGRAS9 promoter’s double-stranded DNA along the major groove (Fig. 2B). For example, residues N300 and I301 in PthA5 repeat 1 recognize adenine (A) through van der Waals interactions. These physical interactions between PthA5 or PthA6 and the EBE regions of the CsGRAS9 promoter effectively repress CsGRAS9 transcription, thereby promoting bacterial pathogenicity. Palindromic sequences in the EBE region might interfere with transcription initiation (Ueda et al. 2020; Luo et al. 2022). However, additional data are required to confirm this mechanism. Moreover, CsNCED1 (Cs5g14370) and CsABA2 (Cs6g19380) were identified as potential target genes of PthA5 (Supplementary Table S5), indicating the complexity of the relationship between TALEs and host gene regulation. These results demonstrate that the interaction between TALEs and host genes is more complex and multifaceted than previously anticipated.

GRAS family members play crucial roles in plant growth and development as well as in various physiological processes, including GA and phytochrome signaling, responses to abiotic and biotic stresses, and chlorophyll biosynthesis (Lu et al. 2022, Neves et al. 2023). For example, GhSCL13-2A enhanced resistance to Verticillium dahliae in cotton by regulating jasmonic acid (JA) and salicylic acid (SA) signaling pathways and reactive oxygen species (ROS) accumulation (Chen et al. 2023). Similarly, MeDELLAs improve the resistance of cassava to bacterial blight caused by Xanthomonas axonopodis pv. manihotis (Xam) infection (Li et al. 2018). Moreover, grapevine (Vitis vinifera L.) VviSHR5 expression was induced by Botrytis cinerea infection (Grimplet et al. 2016). In rice, three homologs of CsGRAS9 (OsGRAS8, OsGRAS39, and OsSHR1) confer resistance to bacterial leaf blight (BLB) caused by Xanthomonas oryzae pv. oryzae (Xoo) and sheath blight caused by Rhizoctonia solani (Dutta et al. 2021). In the present study, phylogenetic analysis revealed that CsGRAS9 was closely related to SCL32 subfamily members (AtSCL32, SlGRAS38, SlGRAS26, ThSCL32, and OsSCL32-2). In Tamarix hispida, ThSCL32 regulates ROS accumulation by enhancing antioxidant enzyme activity (Lei et al. 2023). To investigate the role of CsGRAS9 in canker resistance, we induced gene expression of CsGRAS9 by Xcc049E/dGRAS9 (Fig. 5). Overexpression of CsGRAS9 led to delayed pustule formation, decreased development of canker symptoms, and repression of CsLOB1 expression. These findings indicate that CsGRAS9 promotes resistance to citrus canker.

In response to Pseudomonas syringae pv. actinidiae infection, resistant plant varieties exhibit increased expression of resistance-related genes, whereas susceptible varieties display decreased expression (Zhao et al. 2024). The protein structures of citrus GRAS9 homologs were found to be remarkably conserved (Supplementary Fig. 11). F. hindsii demonstrates durable resistance to canker disease through PCD and ROS accumulation while also inducing defense-related gene expression (Wu et al. 2018; Zhu et al. 2019). In the present study, we identified transcriptional differences between CsGRAS9 and FhGRAS9 during Xcc infection in citrus plants (Fig. 4). In addition, we detected a natural variation involving a 558-bp deletion in the FhGRAS9 promoter (Supplementary Fig. 10B). PthA5 and PthA6 significantly inhibited the transcriptional activity of the CsGRAS9-P1 promoter but did not affect the FhGRAS9-P2 promoter. This finding is consistent with the observed differences in CsGRAS9 and FhGRAS9 expression levels following Xcc infection (Fig. 4D–F). However, the presence of a genotype matching the CsGRAS9 promoter suggests that multiple TALEs, including PthA4, PthA5, and PthA6, collaboratively regulate FhGRAS9 expression, resulting in enhanced citrus canker symptoms. Similarly, natural variation in the EBE regions of the AbLOB1 promoter in A. buxifolia (primitive citrus) affects AbLOB1 expression (Tang et al. 2021b). On the basis of these findings, we speculate that mutations in promoter regions stabilize FhGRAS9 expression and increase canker resistance in F. hindsii. In addition, Citron C-05 (C. medica) exhibits resistance to citrus canker by restricting Xcc proliferation (Fu et al. 2020), and orange jessamine (M. paniculata) shows faint chlorosis with low Xcc populations (Ference et al. 2020). Natural genetic variation improves growth-defense trade-offs, reflecting the adaptation of plants to diverse stress pressures (He et al. 2022). In the present study, we identified multiple sequence variations in the GRAS9 promoters of orange jessamine and citron but not in susceptible citrus biotypes. These insertions and deletions caused frameshifts that slightly affected the binding of TALE to the EBEs of CmGRAS9 and MpGRAS9 (Fig. 4H–I). Future experiments are required to validate the hypothesis that natural mutations of GRAS9s promoter in C. medica and M. paniculate could evade the inhibitory effects of TALEs on GRAS9s gene expression by obstructing TALEs from binding to the EBE regions in plants. These findings suggest that natural variations in the GRAS9 promoters among citrus germplasms confer resistance to different Xcc races.

CRISPR/Cas9 systems have been widely used to elucidate gene function by enabling efficient genetic modifications in fruit tree species, such as ‘Anliu’ sweet orange and grape (Tang et al. 2021b, Ren et al. 2021). Additionally, genome editing has been applied to modify the EBE regions in gene promoters, creating plants resistant to bacterial and fungal pathogens, including citrus canker (Su et al. 2023), BLB (Xu et al. 2019), and powdery mildew (Li et al. 2022a, b). In this study, we employed the PTG/Cas9 system to modify the EBE regions in the CsGRAS9 promoter (Fig. 6B). Multiple homozygous and biallelic (chimeric) sweet orange lines with targeted mutations in the EBE regions were generated (Fig. 6A). Consistent with previous studies, the predominant mutation types in citrus were 1-bp A/T base insertions or deletions (Xu et al. 2022). No off-target mutations were observed in the CsGRAS9-edited lines (Supplementary Fig. 16). We also assessed CsGRAS9 expression in wild-type and proGRAS9-edited lines (#9, #18-2, and #23) after inoculation with the control and the tal-free strain Xcc049E/EV. As shown in Supplementary Fig. S19, a slight change in CsGRAS9 expression was observed in chimeric-edited line #18-2, which exhibited an 86-bp deletion. Furthermore, both alleles of the homozygous line #23 contained the same 86-bp deletion, significantly reducing CsGRAS9 expression. In line with this, CsGRAS9 expression in these lines was altered by Xcc049E/EV infection, correlating with increased resistance to citrus canker. Editing the promoter region effectively abolished the inhibition of CsGRAS9 by Xcc049E/pthA4 and Xcc049E/pthA5 (Fig. 6G). Modifying the EBEs in CsGRAS9 promoter delayed canker symptom development and decreased Xcc growth in host, consistent with the observed symptoms under Xcc049E/dGRAS9 infection (Fig. 6F). These results suggest that CRISPR/Cas9 system is an efficient tool for studying the Xcc/citrus pathosystem and offers a sustainable approach to managing citrus canker. Furthermore, our findings support the hypothesis that mutations in the EBE regions of CsGRAS9 promoter enhance resistance to citrus canker.

Previous studies have shown that GRAS proteins play a key role in disease resistance by regulating GA biosynthesis and signaling pathways (Zhou et al. 2018). For example, XopD_Xcc8004 interfered with DELLA degradation, thereby suppressing GA signaling and promoting disease tolerance (Tan et al. 2014). In rice, overexpression of OsGA20ox3 leads to increased GA accumulation and heightens the susceptibility of wild-type plants to Xoo and M. oryzae (Qin et al. 2013). Our research demonstrated that overexpression of CsGRAS9 in grapefruit elevated the transcript levels of GA biosynthetic genes (GA20ox1, GA3ox2 and GA3ox3) while reducing the expression of GA methyltransferase 2 (GAMT2) (Supplementary Fig. S21A-D). These findings suggest that CsGRAS9 promotes the expression of GA biosynthetic genes, potentially revealing an adaptive mechanism that fine-tunes GA signaling to balance defense responses in citrus. In summary, our findings allowed us to put forward a model of how TALEs promote bacterial virulence and how CsGRAS9 regulates citrus canker resistance (Supplementary Fig. S21E). In susceptible species, TALEs (PthA5, PthA6, and PthA4) interact with the EBE regions of the CsGRAS9 promoter, repressing CsGRAS9 transcription and promoting susceptibility. In resistant species, where the EBE regions of the CsGRAS9 promoter are absent or mutant, TALEs cannot recognize the CsGRAS9 promoter, thus preventing transcriptional repression. CsGRAS9 enhances resistance to citrus canker and regulates the expression of GA-related genes.

Building on these results, we present evidence that GRAS9 expression plays a crucial role in controlling canker resistance. In citrus, natural variations and gene-edited mutations in the EBE regions of the GRAS9 promoters disrupt the EBEproGRAS9-TALE complex, leading to transcriptional activation of GRAS9 and enhanced canker resistance. The successful application of the PTG/Cas9 system, combined with the identification of promoter variations, provides a valuable genetic resource and a reliable platform for future molecular breeding efforts aimed at improving citrus resistance.

Methods

Plant materials and bacterial strains

Citrus plants, including wild-type Hong Kong kumquat (Fortunella hindsii), grapefruit (Cirus paradisi L.), ‘Anliu’ sweet orange (C. sinensis), and proCsGRAS9-edited seedlings, were cultivated in a greenhouse under controlled conditions of 28°C, 60% relative humidity (RH), and a 16:8 h light/dark photoperiod. Xanthomonas citri strains were grown on nutrient agar (NA) plates or in nutrient broth (NB) medium at 28°C, while E. coli strains were cultured in Luria-Bertani (LB) medium at 37°C. Appropriate antibiotics were added to bacterial growth media, including kanamycin (50 μg/ml), spectinomycin (100 μg/ml), and ampicillin (100 μg/ml). Strains and plasmids used in this study are listed in Supplementary Table S1.

Sequencing of tal genes

Following the manufacturer's protocol and previous studies, the EZ-Tn5™ < KAN-2 > Insertion Kit was used to generate mutants by inserting the Tn5 transposon into the repeat regions of the pthA5 and pthA6 genes. (Haq et al. 2020). Mutants were selected using SphI and BamHI restriction enzymes, and the complete sequences of pthA5 and pthA6 were confirmed by sequencing with primer pairs TALE-SF/SR and Tn5-NF/NR. The primers used are listed in Supplementary Table S4.

Phylogenetic tree construction and host target prediction

We obtained 77 complete genome sequences of X. citri strains from the NCBI database. AnnoTALE v1.4.1, was used to predict TALEs in each X. citri strain and to acquire the RVD sequence of TALEs (Grau et al. 2016). TALEs were grouped into classes based on RVDs that showed possible functional and evolutionary relationships (Grau et al. 2016). A disTAL module of the QueTAL suite was used to align and classify all X. citri strain TALEs based on their tandem repeat region sequences (Pérez-Quintero et al. 2015). FuncTAL was used to phylogenetically classify TALEs based on their DNA-binding specificities (Pérez-Quintero et al. 2015). Phylogenetic trees were constructed using the iTOL web server v5 (Letunic et al. 2021).

Design and assembly of dGRAS9

Artificial dTALE (dGRAS9) was constructed by TALEN targeter 2.0 (https://tale-nt.cac.cornell.edu/node/add/talen) as previously described (Haq et al. 2020, Xu et al. 2024). The repeat regions of dGRAS9 targeted the 17-bp promoter region (AACTTCTTAAGTTTGAG) of the citrus gene CsGRAS9 (Cs2g22130). dGRAS9 was cloned into the plasmid pUC57 with the SphI site, replaced with the SphI fragment of pthXo1 to generate pZY-dGRAS9, and then fused with pHM1 at the HindIII site to construct pHZY-dGRAS9 (Table S1). Primers used in this study are listed in Supplementary Table S4.

Expression Analysis of TALEs by Western Blotting

The strain Xcc049E was used to assay the function of the tale genes and dGRAS9 in citrus canker formation (Li et al. 2014).Empty vector pHZY, along with plasmids pHZY-pthA5, pHZY-pthA6, pHZY-pthA4, and pHZY-dGRAS9, were introduced into Xcc049E cells via electroporation (2.5 kV, 4.5 ms). The expression of constructed plasmids was confirmed by Western blotting using anti-FLAG antibodies. Primers used in these experiments are listed in Supplementary Table S4.

Pathogenicity and bacterial growth assays

For leaf inoculation, Xcc bacterial suspensions were diluted with double-distilled water and administered to citrus leaves using 1-mL needleless syringes (Fu et al. 2020). In addition, healthy citrus leaves were punctured with a 0.6-mm pin, followed by application of 6 μL of Xcc cultures to each puncture. The severity of Xcc infection was evaluated at 8 days post-inoculation (dpi), and canker lesion development was photographed at 8, 12, 16, 20, 21, and 27 dpi. All experiments were conducted in triplicate with three biological replicates.

Bacterial growth in the leaves was quantified by serial dilution of bacterial suspensions extracted from three leaf discs with visible lesions. The discs were homogenized in tubes containing 1 mL of double-distilled water and steel balls, and the bacterial suspensions were plated on nutrient-rich media with the appropriate antibiotic. Each treatment was performed in triplicate at different time points (Conforte et al. 2019).

Microscopic analysis

Xcc-infected grapefruit leaves were cut into 4 mm × 4 mm square pieces using a sharp stainless-steel scalpel and immediately placed into fixative solution (FAA) at 4°C overnight. The samples were rinsed, dehydrated, and resin-infiltrated following the protocol outlined by Long et al (Long et al. 2021). Blocked samples were manually trimmed, stained with toluidine blue, and observed under a light microscope. Leaf thickness and cell size were quantified using CaseViewer software.

TALgetter predictions in citrus plants and promoter sequence analysis

The promoter sequences of citrus genes, including a 2000 bp fragment upstream from the translation start site (ATG) and the amino acid sequences of CsGRAS9 homologs, were retrieved from the Citrus Pan-genome to Breeding Database (CPBD) (http://citrus.hzau.edu.cn). Binding sites recognized by TALEs PthA1, PthA4, PthA5, and PthA6 were predicted in the CsGRAS9 promoter using the Target Finder tool (https://tale-nt.cac.cornell.edu/node/add/talef-off) and AnnoTALE (http://jstacs.de/index.php/PrediTALE). The Alphafold2 platform was used to predict the 3D structure of the Xanthomonas PthA5 protein using deep neural networks, and protein-DNA complexes were modeled using DP-DOCK and visualized in PyMol (http://www.pymol.org/pymol) (Gao et al. 2009). Cis-acting elements in the promoter regions of GRAS9 homologs were identified using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) (Lescot et al. 2002). Phylogenetic trees of CsGRAS9 homologs were constructed using the MEGA X program after aligning the promoter sequences.

RNA extraction and Quantitative real-time PCR

Citrus leaves were collected at 1, 2, and 7 dpi. Total RNA was extracted from both infected and uninfected citrus leaves, and first-strand cDNA synthesis was performed using the FastKing RT Kit with gDNase (TIANGEN, China). qRT-PCR was carried out on an ABI 7500 real-time system using the TransStart Top Green qPCR SuperMix (TRANS, China). The relative expression levels were normalized to CsEf1a as an internal control, and the comparative 2^−ΔΔCt method was used to determine fold changes in gene expression. The RT-qPCR primers are listed in Supplemental Table S4.

Electrophoretic mobility shift assay

A 35-bp promoter fragment corresponding to the predicted TALE binding region upstream of CsGRAS9, CmGRAS9, and MpGRAS9 was synthesized and labeled with Cy5. BamHI fragments of pthA5, pthA6, and pthA4 were cloned into pET28a vector (containing a His tag). Recombinant PthA5-His, PthA6-His, and PthA4-His proteins were expressed in E. coli BL21 strains induced with 0.5 mM IPTG at 16°C overnight. After cell lysis, the supernatant was passed through a Ni-NTA Sefinose Resin column (BBI, C600791, USA). Various concentrations of unlabeled CsGRAS9, CmGRAS9, and MpGRAS9 probes were used for competitive binding assays. For the EMSA, Cy5-labeled probes were incubated with PthA5-His, PthA6-His, or PthA4-His proteins, EMSA Binding Buffer, and poly dI-dC at 25°C for 25 minutes. The DNA-protein mixtures were resolved on 10% native polyacrylamide gel at 4°C for 1.5 hours at 100V in the dark (Xu et al. 2024). The fluorescent Cy5-labeled DNA was visualized using the Amersham Typhoon RGB imager (GE, Sweden).

Yeast one-hybrid assay

The CsGRAS9 and FhGRAS9 were cloned into the pLacZ vector (bait). Multiple cloning sites (MCS) with XbaI, EcoRI, BamHI, XhoI, and PstI restriction sites was inserted into the pBluescriptII SK(−) vector, which was digested with XbaI and PstI. BamHI fragments of pthA5 and pthA6 were incorporated into the pB vector containing EcoRI and BamHI sites. The EcoRI-BamHI-cut fragments of pthA5 and pthA6 were then recombined into the pB42AD vector, producing pB42AD-PthA5 and pB42AD-PthA6 (prey). Yeast strain EGY48 was transformed with the bait and prey plasmids, grown on synthetic dropout (SD) medium lacking tryptophan (Trp) and uracil (Ura), and assayed on SD/-Trp-Ura plates with X-gal. Negative controls included yeast cells transformed with empty pLacZ or pB42AD vectors. Primers are listed in Supplementary Table S4.

GUS activity assays

BamHI fragments of pthA5, pthA6, and pthA4 replaced avrXa10 in the pHB-avrXa10 vector through BamHI sites, serving as activators for transient expression in N. benthamiana. Reporter constructs (pCsGRAS9::GUS, pCsGRAS9-P1::GUS, and pFhGRAS9-P2::GUS) were generated by cloning GRAS9 homolog promoters into the pCAMBIA1381::GUS vector digested with BamHI and PstI. The activator and reporter constructs were transformed into Agrobacterium tumefaciens strain GV3101 and grown on YEB medium containing kanamycin and rifampicin. Agroinfiltration was performed on N. benthamiana leaves following standard protocols (Xu et al. 2024). Leaf discs (10 mm in diameter) were collected at 2–3 dpi, stained with GUS buffer at 37°C overnight, and washed in graded ethanol to remove chlorophyll. GUS activity was quantified using 4-methylumbelliferyl-β-glucuronide. Primers are listed in Supplementary Table S4.

Dual-luciferase reporter (dual-LUC) assay

The pCsGRAS9::LUC reporter was constructed by inserting the CsGRAS9 promoter into the pGreenII0800-LUC vector, digested with HindIII and BamHI. The constructs were transformed into A. tumefaciens strain GV3101 and agroinfiltrated into N. benthamiana for dual-LUC assays (Long et al. 2021). Firefly (LUC) and Renilla (REN) luminescence ratios were determined using the Dual-Luciferase Reporter Gene Assay Kit (Shanghai, China) and the Dual-Glo® Luciferase Assay System (Promega). Fluorescence was observed using an imaging system. Primers are listed in Supplementary Table S4.

Identification of GRAS9 proteins in citrus and other species

Citrus GRAS protein sequences were obtained from the Citrus Pan-genome to Breeding Database (CPBD) (http://citrus.hzau.edu.cn) and the NCBI database. Multi-sequence alignments of citrus GRAS proteins and conserved GRAS9 members were performed using MEGA 7.0, and phylogenetic trees were constructed from full-length amino acid sequences. The MEME suite (version 5.5.5) was employed to identify 20 conserved motifs in citrus GRAS9 proteins and predict protein functions (Dutta et al. 2021). Visualization was performed using TBtools software.

Transcriptional activation activity assay

The coding sequences (CDS) of CsLOB1 (positive control) and CsGRAS9 were inserted into pGBKT7 to fuse with the GAL4 DNA-binding domain. Yeast strain AH109 was co-transformed with these vectors. Positive clones were selected on SD/-Leu/-Trp medium and assayed for transcriptional activation on SD/-Leu/-Trp/-His medium with or without X-α-gal (Long et al. 2021). Primers are listed in Supplementary Table S4.

Subcellular Localization

The full-length CDS of CsLOB1 and CsGRAS9 were cloned into the pHB vector to generate YFP fusion constructs. pHB-CsLOB1 was used as a positive control, while the empty pHB vector served as a negative control. The vectors were transformed into A. tumefaciens strain EHA105, which was then infiltrated into N. benthamiana leaves. The nucleus was stained with DAPI (excitation at 405 nm), and YFP fluorescence signals (excitation at 514 nm, emission at 524–580 nm) were detected two days post-infiltration using a Leica TCS SPS-II confocal laser scanning microscope (Carl Zeiss SAS, Jena, Germany). Primers are listed in Supplementary Table S4.

Generation of citrus mutants

The 23-bp gRNA sequence (target site: CCTTTCGCCTCTATATATATACA in the EBE region of CsGRAS9 promoter) was designed using the online CRISPR-P2.0 design tool website (http://cbi.hzau.edu.cn/CRISPR2/). The gRNA-CsGRAS9 fragment was formed by annealing and was immediately cloned into the AtU6-26-sgRNA-SK vector that was predigested with BsaI. The SpeI fragment of the AtU6-26-CsGRAS9-sgRNA cassette was ligated into the pCAMBIA1300-pYAO-Cas9 vector containing Cas9 and GFP proteins. The construct was transferred into A. tumefaciens EHA105 and then transformed into ‘Anliu’ sweet orange epicotyls, as previously described (Tang et al. 2021a, b; Yan et al. 2015).

Genotyping assay of CRISPR-edited mutations and Off-target analysis

Genomic DNA was extracted from CsGRAS9 promoter-edited seedlings. Specific primers (Supplementary Table S4) were used to amplify the targetsites for mutation detection by Sanger sequencing. Off-target sites were screened using CRISPR-P2.0, and PCR products from potential off-target regions were sequenced for off-target analysis. Primers are listed in Supplementary Table S4.

PPI and Yeast two-hybrid assays

The CsGRAS9 protein sequence was submitted to the STRING database (https://string-db.org/) to predict functional protein-protein interaction partners based on the C. sinensis genome (Szklarczyk et al. 2023). The interaction network was visualized using Cytoscape. The CDS of CsGRAS9 was cloned into pGADT7 to create pGADT7-CsGRAS9, and CsTAC1 was cloned into pGBKT7 to generate pGBKT7-CsTAC1. These vectors were co-transformed into yeast strain AH109, and transformed positive clones were selected on SD/-Leu/-Trp and SD/-Leu/-Trp/-His medium with or without X-α-gal to assess protein-protein interactions. Primers are listed in Supplementary Table S4.