Lateral organ boundaries 1 is a disease susceptibility gene for citrus bacterial canker disease

Significance

Citrus bacterial canker, which is caused by several species in the genus Xanthomonas, is a severe disease with worldwide distribution affecting all the commercially important citrus species and cultivars. The mechanisms of canker development, involving erumpent pustule formation and bacterial growth, are not known. Our findings suggest that virulence determinants in several pathogens activate a single host disease susceptibility (S) gene that has a critical contribution to bacterial growth and host pustule development. The S gene represents an excellent candidate for control measures for the citrus bacterial canker.

Abstract

Citrus bacterial canker (CBC) disease occurs worldwide and incurs considerable costs both from control measures and yield losses. Bacteria that cause CBC require one of six known type III transcription activator-like (TAL) effector genes for the characteristic pustule formation at the site of infection. Here, we show that Xanthomonas citri subspecies citri strain Xcc306, with the type III TAL effector gene pthA4 or with the distinct yet biologically equivalent gene pthAw from strain XccA^w, induces two host genes, CsLOB1 and CsSWEET1, in a TAL effector-dependent manner. CsLOB1 is a member of the Lateral Organ Boundaries (LOB) gene family of transcription factors, and CsSWEET1 is a homolog of the SWEET sugar transporter and rice disease susceptibility gene. Both TAL effectors drive expression of CsLOB1 and CsSWEET1 promoter reporter gene fusions when coexpressed in citrus or Nicotiana benthamiana. Artificially designed TAL effectors directed to sequences in the CsLOB1 promoter region, but not the CsSWEET1 promoter, promoted pustule formation and higher bacterial leaf populations. Three additional distinct TAL effector genes, pthA*, pthB, and pthC, also direct pustule formation and expression of CsLOB1. Unlike pthA4 and pthAw, pthB and pthC do not promote the expression of CsSWEET1. CsLOB1 expression was associated with the expression of genes associated with cell expansion. The results indicate that CBC-inciting species of Xanthomonas exploit a single host disease susceptibility gene by altering the expression of an otherwise developmentally regulated gene using any one of a diverse set of TAL effector genes in the pathogen populations.

Citrus bacterial canker (CBC) is a severe disease with worldwide distribution affecting all of the commercially important citrus species and cultivars. The disease is caused by two species of bacteria in the genus Xanthomonas. The most widespread species is Xanthomonas citri subspecies (ssp.) citri (Xcc) and was originally identified in Asia. The disease is believed to have subsequently spread from Southeast Asia to other citrus growing regions. Strains of Xcc are further distinguished according to their host ranges. Type A strains of Xcc cause disease on most species of citrus, whereas type A^w and type A* strains are restricted to Key lime (Citrus aurantiifolia) (1–3). A second genetically distinct species, Xanthomonas fuscans ssp. aurantifolii (Xfa), is grouped into type B and type C. Outbreaks occur sporadically. In Florida, for example, the last extensive outbreak involving type A strains occurred in 1995, triggering an ultimately unsuccessful eradication program that ended in 2006, costing an estimated $1 billion, and stimulated renewed efforts for more effective and economical control methods (4).

Genomic resources exist for citrus species, including draft genome sequences of several species and extensive expression sequence tags (ESTs) of mRNAs from different developmental and disease states (5, 6). A variety of transcription profiling studies of diseased hosts with citrus bacterial canker have been conducted, comparing susceptible and resistant host reactions (7, 8). Genes involved in host defense, cell-wall remodeling, vesicle trafficking, and cell division genes were identified that may be involved in disease development.

Equally good genomic resources are available for the bacterial pathogens. Genome sequences are available for representative strains of Xcc type A, Xcc type A^w, Xfa type B, and Xfa type C (9–11). Both Xcc and Xfa contain type III secretion systems (T3SS), although the contributions, biochemically and functionally, of the individual substrate effectors are unknown. Xcc-A and Xcc-A^w contain 24 and 30 putative type 3 secretion (T3S) effectors, respectively, whereas Xfa-B and Xfa-C contain 27 and 26 T3S effectors, respectively (10). Loss of T3SS function in Xcc results in complete loss of disease symptoms and reduced bacterial populations in host tissue. However, phenotypic effects have only been observed upon mutation of several individual effector genes (12–14). The host range restriction of type A^w strains to Key lime has been attributed to the presence of T3S effector AvrGf1, whereas the characteristic symptom of pustule formation in citrus bacterial canker is dependent on different members of the AvrBs3/PthA family of T3SS effectors, known collectively as transcription activator-like (TAL) effectors (3, 15).

TAL effectors have been shown to direct the induction of specific disease susceptibility (S) and resistance (R) host genes during infection (16). TAL effectors bind to plant DNA elements within the promoter regions via a series of amino acid repeats in the coding central portion (17–19). PthA from Xcc was the first TAL effector to be associated with a distinct virulence function in infections, controlling both pustule formation and the level of bacterial leaf populations (20). Transient expression of pthA inside the host cells has been reported to induce CBC-like symptoms in excised leaf tissue (21). One target of the TAL effector AvrBs3 from Xanthomonas campestris pathovar (pv.) vesicatoria is upa20, which encodes a bHLH family transcriptional factor and acts as a regulator of cell enlargement in Nicotiana benthamiana (22). In rice, two major S genes, Os8N3 (OsSWEET11) and Os11N3 (OsSWEET14), are targets of the TAL effectors PthXo1 and AvrXa7, respectively, from the bacterial blight pathogen X. oryzae pv. oryzae, and strains that depend on either PthXo1 or AvrXa7 for full virulence and cannot induce either Os8N3 or Os11N3 due to host mutations or suppression of host gene expression are weakly virulent. Os8N3 and Os11N3 products are not related to upa20, and both are closely related members of a family of sugar transporters (23–25). Different TAL effectors can induce the same gene in the host. In pepper, upa20 is also the target of AvrHah1 from Xanthomonas gardneri (26, 27). In rice, Os11N3 is induced by any of three TAL effectors from Xanthomonas oryzae pv. oryzae, AvrXa7 and PthXo3, or TalC (24, 28). An S gene with minor effects on susceptibility in bacterial blight of rice encodes a bZIP transcription factor and is the target of yet another TAL effector, PthXo6 (29).

Representative strains of the five different types responsible for citrus canker, A, A*, A^w, B, and C, contain at least one pthA homolog, which are designated pthA, pthA*, pthAw, pthB, and pthC, respectively, and essential for pustule formation on citrus (30). Although closely related, each gene has a unique repetitive central domain. Xcc strain 306 contains four TAL effector genes, of which pthA4 is known to be required for pustule formation (12). Hypothetically, TAL effectors of Xcc and Xfa induce one or more host genes that result in pustule formation. Here, we combined transcription profiling of host responses to strains of Xcc that vary in TAL effector gene content, TAL effector binding element (EBE) prediction, and artificial TAL effectors designed to identify S genes of citrus.

Results

Experimental Design.

To identify the targets of TAL effectors that are involved in citrus canker, a strategy was devised to identify and test candidate target genes whose expression was dependent on the presence of representative pth genes from Xcc (Fig. 1A). In brief, near isogenic strains were constructed from a TAL effector mutant strain that was incapable of pustule formation. Expression profiles were then conducted on host tissue after inoculation with either wild-type or complemented strain and compared with the pustule-defective parent strain. Candidate S genes were selected from the common TAL effector-dependent expressed genes based on the fold increase in expression, presence of a candidate EBE in the promoter regions, and relatedness to known S genes. The candidate genes were then subjected to complementation by artificially designed TAL effectors (dTALes) that were either optimized by the consensus TAL effector binding codes or targeted to novel promoter sequences. Optimization and novel EBE targeting allow for resolution of collaterally induced or so-called “off-target” genes and the intended host S genes.

Fig. 1.

TAL effectors PthA4 and PthAw are required and sufficient for pustule formation in sweet orange and grapefruit. (A) Experimental design scheme for this study. (B) Loss of pthA4 eliminated pustule formation in sweet orange cultivar (cv.) Valencia (Left) and grapefruit cv. Duncan (Right), which was restored by PthA4 and PthAw. Panels: 1, inoculations with wild-type Xcc306; 2, pthA4 deletion mutant Xcc306∆pthA4; 3, water (mock) inoculation; 4, triple-deletion mutant Xcc306∆pthA1∆pthA2∆pthA3 (with intact pthA4); 5, Xcc306∆pthA4::pthA4; 6, Xcc306∆pthA4::pthAw. The leaves were photographed 5 d after infiltration.

Deletions of each individual TAL effector gene of Xcc306 as well as a triple genes mutant, Xcc306∆pthA1pthA2pthA3, were constructed (Fig. S1A). Previous work had shown that the loss of pthA4 resulted in loss of pustule formation (12). The triple mutant was constructed to determine if pthA4 alone was sufficient for pustule formation. As previously observed, only the strain with a deletion of pthA4 (Xcc306∆pthA4) showed loss of pustule formation on sweet orange and grapefruit (Fig. 1B). The triple mutant, lacking pthA1-3 and retaining pthA4, showed no change in pustule-forming ability (Fig. 1B). Two near-isogenic strains of Xcc306∆pthA4 were constructed containing pthA4 (Xcc306∆pthA4::pthA4) or pthAw (Xcc306∆pthA4::pthAw). Both complemented strains showed symptoms similar to the wild-type Xcc306, indicating that pthA4 was required and sufficient for pustule formation on sweet orange and grapefruit (Fig. 1B).

CsLOB1 and CsSWEET1 Are Candidate Targets of TAL Effectors PthA4 and PthAw.

Microarray analyses using the Affymetrix GeneChip Citrus Array were performed on host mRNA following inoculation with the mutant strain Xcc306∆pthA4, the parental strain with pthA4 (Xcc306) and Xcc306∆pthA4::pthAw. Young leaf tissues of sweet orange and grapefruit were infiltrated with Xcc306 (containing pthA4), Xcc306∆pthA4::pthAw, and Xcc306∆pthA4, and samples were collected at 120 h postinfection. The genes that showed significantly higher (adjusted P ≤ 0.01) expression levels in tissue infiltrated with wild-type Xcc306 or Xcc306∆pthA4::pthAw in comparison with tissue infiltrated with Xcc306∆pthA4 were selected as potential candidate host S genes (Table 1). The promoter regions of the most highly up-regulated genes in either sweet orange or grapefruit were scanned for probable PthA4 or PthAw binding elements (Materials and Methods). The EBEs were predicted based on the repeat variable diresidues (RVDs) and TAL code (Fig. S2A). Two genes, one represented by probe sets Cit.37210.1.S1_at and Cit.35190.1.S1_at and the second by set Cit.3027.1.S1_s_at, had candidate EBEs and were characterized further.

Table 1.

Combined top 10-fold induced genes for PthA4 and PthAw

Affymetrix ID	LFC, Cs	LFC, Cp	DNA	EBE	Annotation
	PthA4	PthAw
Cit.28626.1.S1_s_at	9.357	—	CV710534	No	β-expansin 6
Cit.9528.1.S1_x_at	8.176	—	CX641267	No	β-expansin 2
Cit.5370.1.S1_s_at	8.088	3.164	CX642883	No	Invertase inhibitor
Cit.20041.1.S1_at	7.587	3.402	CB250345	No	No hit
Cit.37210.1.S1_at	7.164	3.434	BQ623314	Yes	LOB domain
Cit.35754.1.S1_at	6.973	3.296	CB250305	No	Polygalacturonase-like
Cit.7877.1.S1_at	6.728	—	CX667721	No	Expansin B2
Cit.9020.1.S1_s_at	6.42	—	CX305834	No	Lipid binding
Cit.35190.1.S1_at*	6.342	3.450	CK932995	Yes	LOB domain
Cit.2392.1.S1_at	6.305		CF831790	No	Acidic cellulase
Cit.3027.1.S1_s_at	—	4.376	CX048987	Yes	Nodulin MtN3
Cit.15355.1.S1_at	—	3.413	CB291618	No	Oxidoreductase
Cit.18912.1.S1_x_at	—	3.073	CX301535	No	Germin-like
Cit.11963.1.S1_at	—	2.993	CF829030	No	Proline-rich PRP1
Cit.35756.1.S1_at	—	2.935	CB250319	No	Endopolygalacturonase

Cs, sweet orange (C. sinensis); Cp, grapefruit (C. paradise); LFC, log₂ fold-change.

^*Represents the same gene as Cit.37210.1.S1_at.

The gene represented by probes Cit.37210.1.S1_at and Cit.35190.1.S1_at contained promoter proximal sequence very close to the canonical PthA4 binding element (EBE_PthA4) located 92 bp upstream of predicted transcription start site, which was based on EST sequences from both sweet orange and grapefruit (Table 1; Fig. S2B). Another gene, which is represented by Cit.3027.1.S1_s_at, contained two candidate EBEs. The first one starts 43 bp upstream of the predicted transcription start site and coincides with putative TATAA box, which was labeled site A; the second one, site B, was found at 85 bp upstream of the start site and was similar to the canonical PthAw EBE (Table 1; Fig. S2C). The expression of both genes was observed to be elevated as determined by quantitative RT-PCR (qRT-PCR) analysis of mRNA from tissue infected either with Xcc306 or Xcc306∆pthA4::pthAw in comparison with mRNA from tissue infected with Xcc306∆pthA4 in sweet orange (Fig. 2) and in grapefruit (Fig. S3). A time course of 12, 24, and 48 h after inoculation in sweet orange indicated that expression of both genes reached high levels by 24 h after the infiltration (Fig. 2). Cit.37210.1.S1_at and Cit.35190.1.S1_at represent a gene encoding a member of the Lateral Organ Boundaries (LOB) domain family of transcription factors and was designated as CsLOB1 (Fig. S4A). The most closely related homologs in Arabidopsis are AtLBD1 and AtLBD11 (Fig. S4B). Cit.3027.1.S1_s_at represents a homolog to the TAL effectors targeted S genes OsSWEET11 and OsSWEET14 in rice, and was designated as CsSWEET1. CsSWEET1 product is most closely related to members of Clade I that includes AtSWEET1 of Arabidopsis (Fig. S4C). Measurements of sugar transport by CsSWEET1 in the HEK293T cells indicated that the transporter could mediate both glucose and sucrose transport activity (Fig. 3). In the assay, entrance of the sugar into the cell interferes with the fluorescence of the particular sensor—in this case, either FLIPsuc90µ∆1V or FLIPglu600µD13V.

Fig. 2.

CsSWEET1 and CsLOB1 are induced by PthA4. The expression level of CsSWEET1 and CsLOB1 reached peak levels at 24 h postinoculation of Xcc306 on sweet orange. Xcc306∆pthA4 did not induce either gene. Total RNA was isolated at 6, 12, 24, and 48 h after inoculation. The expression was normalized to housekeeping gene EF1α. Data represent the mean ± SD; different lowercase letters represent significant differences (P ≤ 0.01) using ANOVA analysis and Tukey test.

Fig. 3.

Identification of CsSWEET1 substrates through HEK293T cell–FRET sensor uptake assay. Sucrose/glucose transport activity for CsSWEET1 was measured by coexpression with cytosolic FRET sucrose sensor FLIPsuc90μΔ1V (A) and cytosolic glucose sensor FLIPglu600μD13V (B) in HEK293T cells. A drop in intensity ratios reflects uptake of the indicated sugar and loss of FRET fluorescence. Individual cells were analyzed by quantitative ratio imaging of CFP and Venus emission (acquisition interval 10 s). HEK293T cells transfected with sensor only (control, light blue) or with the sensor and the Arabidopsis SWEET12 (suc) or SWEET1 (glc; blue) as positive controls. CsSWEET1 shows sucrose and glucose influx (red). Bars are 1 SD unit.

CsLOB1 and CsSWEET1 Promoters Direct TAL Effector-Dependent Expression.

The respective promoters of CsSWEET1 and CsLOB1 were fused to the uidA [β-glucuronidase (GUS)] reporter gene and expressed transiently by Agrobacterium-mediated transfer. Truncated and versions with alterations in the predicted EBEs were tested in coinoculation assays with Xcc in citrus leaves (Fig. 4A). The wild-type promoter fragment of CsSWEET1 directed GUS activity when coinoculated with the wild-type strain Xcc306 and the complemented strain Xcc306∆pthA4::pthA4, whereas no GUS activity was observed when coinfiltrated with strain Xcc306∆pthA4 (Fig. 4B; CsSWPwt). Coinoculations with the truncated, substituted, and deleted versions of CsSWEET1 promoter and Xcc306 resulted in little or no GUS activity, indicating noncanonical structure to the candidate EBEs of CsSWEET1 (Fig. 4; CsSWPT, CsSWPM1, and CsSWPD, respectively). Wild-type, truncated, and substituted versions of CsLOB1 promoter were activated to the same approximate level (Fig. 4; CsLOBPT and CsLOBPM1). The deletion within the predicted EBE and TATAA box of CsLOB1 (Fig. 4A; CsLOBPD) resulted in the loss of PthA4-mediated expression (Fig. 4B).

Fig. 4.

PthA4 drives expression of CsSWEET1 and CsLOB1 promoter/uidA fusion genes. (A) Promoter constructs used in GUS transient expression assay. The predicted TAL EBEs are underlined. PD, deleted promoter; PM, mutated promoter; PT, truncated promoter; Pwt, wild-type promoter; and //, truncation. Base mutations are in lowercase letters, and red font represents putative TATAA box. Fragments including 5′ UTR and ∼100-bp coding sequences of the genes were fused to the ATG of the uidA coding sequence. (B) Transient GUS activity associated with CsSWEET1 and CsLOB1 promoters after inoculation with Xcc306 and derivative strains in sweet orange. Xanthomonas were inoculated 5 h after the inoculation with A. tumefaciens containing the GUS reporter constructs as indicated in A. N, empty vector without promoter fragment; + and −, with PthA4 and without PthA4, respectively; ^a, inoculation with Xcc306∆pthA4::pthA4. GUS activity was assayed 5 d after inoculation. SD values were calculated from three technical replicates of one experiment. The experiment was repeated twice with similar results.

The results for the alterations to CsSWEET1 and CsLOB1 promoters indicated that, although remarkably similar, the promoters of the respective genes with respect to the candidate EBEs have important differences. Based on the results with the truncated version CsLOBPT, the candidate EBE_PthA4 for CsLOB1 is contained within the region of the TATAA box (Fig. 4A, construct 8). Additional base substitutions and insertions were created and tested within the truncated version of CsLOB1 to further corroborate the function of this region as an EBE_PthA4 (Fig. 4A, constructs 8–12). Promoter variant CsLOBPM3, which has a substitution of GG for CC at the eighth and ninth positions in the EBE, had a severe effect on PthA4-dependent promoter activity, whereas the lone substitution of T (CsLOBPM3) at position 8 had little effect on activity. A single nucleotide insertion at position 11 (CsLOBPins) in EBE_PthA4 resulted in loss of GUS activity (Fig. 4B, constructs 9–11). By contrast, the promoter of another highly up-regulated gene, Cit.7877.1.S1_at, was not able to be induced by PthA4 (Fig. 4B, construct 13). Placement of the EBE_PthA4 from CsLOB1 at 20 bp upstream of putative TATAA box in the Cit. 7877 promoter resulted in PthA4-dependent expression of Cit. 7877.1.S1_at (Fig. 4B, construct 14). Agrobacterium tumefaciens-mediated transient ectopic expression of pthA4 or pthAw was also able to activate the same CsLOB1 or CsSWEET1 promoter patterns in N. benthamiana, respectively (Fig. S5).

Artificial dTALes Targeting CsLOB1 Induce Pustule Formation.

Artificial dTALe genes using pthA4 as a backbone sequence were designed with repeats specifically targeting unique sequences within promoters of CsSWEET1 and CsLOB1, respectively, using optimized repeat variable di-amino acid (RVD) residues (Fig. 5A). The genes were designated dCsLOB1.1, dCsLOB1.2, dCsSWEET1.1, and dCsSWEET1.2, introduced into Xcc306∆pthA4, and tested for activity on citrus leaves. Xcc306∆pthA4 with either dCsLOB1.1 or dCsLOB1.2 induced CsLOB1 expression, but did not induce CsSWEET1, whereas Xcc306∆pthA4 with dCsSWEET1.1 or dCsSWEET1.2 induced the expression of CsSWEET1 but not CsLOB1 expression (Fig. 5B). In parallel, the Xcc306∆pthA4 with the individual dTALe genes were tested for the ability to induce the promoter uidA reporter genes by quantitative and qualitative transient GUS assays in citrus and N. benthamiana. Both dTALes targeting CsLOB1-directed expression of the CsLOB1 promoter, but not CsSWEET1 promoter, and, conversely, both dTALes targeting CsSWEET1 drove expression of the CsSWEET1 promoter reporter genes and not the CsLOB1 promoter fusion in citrus leaves (Fig. 5 C and D). The dTALe-complemented strains of Xcc306∆pthA4 were infiltrated into sweet orange and grapefruit to determine what effect the artificial effectors would have on the disease phenotype. Only inoculations of Xcc306∆pthA4 with either dTALe targeting CsLOB1 resulted in pustule formation, whereas Xcc306∆pthA4 with dCsSWEET1.1 or dCsSWEET1.2 resulted in weak disease symptoms, which were similar to the response of Xcc306∆pthA4 alone (Fig. 6A). Histological analysis of tissue infected with Xcc306∆pthA4::dCsLOB1.1 revealed excessive cell division and proliferation (hyperplasia) similar to tissue with Xcc306∆pthA4::pthAw (Fig. 6B) and in contrast to inoculated tissue with Xcc306∆pthA4. The bacterial leaf populations were significantly higher in sweet orange leaves inoculated with Xcc306∆pthA4::dCsLOB1.1 compared with Xcc306∆pthA4 but lower than Xcc306∆pthA4::pthA4 (Fig. 7).

Fig. 5.

dTALes-mediated induction of CsLOB1 or CsSWEET1. (A) RVDs of dTALes and the corresponding targeted EBE sequences in the host genome. dCsSWEET1.1 targets EBE_PthA4 in CsSWEET1 promoter but with a 3′ extension, whereas dCsSWEET1.2 targets a sequence 13 bp upstream of the predicted EBE_PthA4. dCsLOB1.1 targets a sequence 33 bp downstream of EBE_PthA4 in CsLOB1 promoter, whereas dCsLOB1.2 is the optimized dTALe for EBE_PthA4 in CsLOB1 promoter (exact consensus match). (B) Artificial dTALes induced expression of the corresponding targeted genes. The dTALes genes were introduced into Xcc306∆pthA4, and qRT-PCR analysis of host mRNA was conducted 48 h after the inoculation. Data represent the mean ± SD with three replications. (C) GUS activity assay using dCsSWEET1.1, dCsLOB1.1, and dCsLOB1.2 complementing Xcc306∆pthA4 strains. Agrobacterium and Xanthomonas were coinfiltrated into leaf tissue of sweet orange, and assays were conducted at 5 d after the infiltrations. Black columns indicate A. tumefaciens with CsLOBPwt::GUS constructs, gray columns indicate A. tumefaciens with CsSWPwt::GUS constructs. Inoculation with Xcc306∆pthA4::pthA4 was used as a positive control. (D) GUS staining assay in N. benthamiana leaves upon ectopic expression of either dCsSWEET1.2 or dCsLOB1.2, respectively, using the CaMV35S promoter to drive expression. Agrobacterium harboring 35S:dCsSWEET1.2 or 35S:dCsLOB1.2 was coinfiltrated with Agrobacterium containing CsSWPwt or CsLOBPwt promoter/uidA constructs as indicated in Fig. 4A.

Fig. 6.

dTALes targeting CsLOB1 promoter when expressed in Xcc306∆pthA4 restore pustule formation in citrus. (A) Lesion symptoms after inoculation with strains containing natural or artificial TAL effector genes. (Left) Sweet orange; (Right) grapefruit. Leaves were inoculated with a bacterial concentration of 5 × 10⁸ cfu/mL and photographed at 5 d after infiltration. Panels: 1, Xcc306; 2, Xcc306∆pthA4 (mutant); 3, Xcc306∆pthA4::dCsLOB1.1; 4, Xcc306∆pthA4::pthA4; 5, Xcc306∆pthA4::pthAw; 6, Xcc306∆pthA4::dCsSWEET1.1. The right table indicates presence or absence of the pustule symptoms with Xcc306∆pthA4 containing the gene for the indicated effector or Xcc306∆pthA4 alone. In pustule column, −, no pustule; +, pustule formation observed at 5 d. (B) Thin cross-section images of grapefruit leaves 5 d after inoculation with Xcc306∆pthA4 (Left), Xcc306∆pthA4::dCsLOB1.1 (Upper Right), and Xcc306∆pthA4:pthAw (Lower Right).

Fig. 7.

dCsLOB1.1 enhances growth of Xcc306∆pthA4 in sweet orange. Xcc306∆pthA4 and Xcc306∆pthA4::dCsSWEET1.1 have reduced bacterial leaf population compared with Xcc306∆pthA4::pthA4 and Xcc306∆pthA4::dCsLOB1.1. Leaves were inoculated at the concentration of 5 × 10⁵ cfu /mL, and the population was measured at the time points indicated. Error bars represent 1 SD. Significance between strains was assessed at final time point at P < 0.01 by using Tukey–Kramer HSD test for post-ANOVA analysis. Values at 15 dpi with the same letter do not differ at the significance level of P < 0.01. The experiment was repeated twice with similar results.

CsLOB1 Is Target of Alternate TAL Effectors Involved in CBC.

The TAL effector genes pthA*, pthB, and pthC, which were previously shown to be associated with pustule formation (30), were tested for the ability to induce pustule formation in Xcc306∆pthA4. The complementing strains of Xcc306∆pthA4 with each respective gene were inoculated on sweet orange and grapefruit host to assess the ability to induce CsLOB1 and CsSWEET1. The three genes, similar to pthA4 and pthAw, could confer pustule formation in Xcc306∆pthA4 (Fig. S6A). PthA4, PthAw, and PthA* led to induction of both CsSWEET1 and CsLOB1 in both species, whereas PthB and PthC could only direct the expression of CsLOB1, but not CsSWEET1, in both species (Fig. 8).

Fig. 8.

Multiple TAL effectors associated with pustule formation in CBC induce CsLOB1 and/or CsSWEET1 in sweet orange and grapefruit. Black columns represent the expression values of CsSWEET1, and white columns indicate expression values of CsLOB1. RNA was prepared 48 h after inoculation. Strains with genes for PthB and PthC did not induce CsSWEET1 in either species. Data represent the mean ± SD of three replications.

The EBEs of all of the five TAL effectors were predicted on the basis of their contained RVDs and the DNA binding specificity (Fig. S2A). The predicted EBEs of both PthB and PthC were located six bases upstream of EBE_PthA4. The predicted EBEs of PthAw and PthA* are located at the same position as that of PthA4 (Fig. 9A). Xcc306∆pthA4::pthB and Xcc306∆pthA4::pthC strains could only direct expression of the wild-type CsLOB1 promoter but not the truncated versions (Fig. 9B), which are missing three bases of the predicted EBEs for PthB and PthC. PthB and PthC also did not activate uidA expression with the CsSWEET1 promoter reporter gene. Bacterial leaf populations of Xcc306∆pthA4::pthB in sweet orange were the same as Xcc306∆pthA4::pthA4, and higher than the mutant Xcc306∆pthA4 by 9 d after infiltration (Fig. S6B).

Fig. 9.

PthB and PthC drive CsLOB1 promoter, but not CsSWEET1 promoter, expression. (A) Consensus EBEs of PthA4, PthAw, PthA*, PthB, and PthC (in gray) and the corresponding nucleotide sequences are depicted in the CsLOB1 and CsSWEET1 promoters from sweet orange. Mismatches between predicted EBE and CsLOB1 promoter are indicated in bold red font and underlined; different bases in CsSWEET1 promoter compared with that of CsLOB1 are in green font. (B) GUS transient assays in sweet orange with the coinoculation of Xcc306 or derivative strains and A. tumefaciens harboring promoter/uidA fusion genes listed in Fig. 4A. Each set of columns is labeled with the specific TAL effector produced by the corresponding gene in strain Xcc306∆pthA4. Data bars represent the mean ± SD with three technical replicates of one experiment. The experiment was repeated twice with similar results. Columns for CsSWPwt with the same lowercase letters do not differ from each other at the significance level of P < 0.05 using the Tukey test.

In general, the induction level of CsLOB1 in grapefruit was lower than that in sweet orange. A comparison of the promoters for CsLOB1 and CsSWEET1 in sweet orange and grapefruit revealed several nucleotide differences. However, the respective genes from each species have identical sequences within the predicted EBEs (Fig. S7).

CsLOB1 Is Associated with Cell Wall-Related Gene Expression.

Based on expression analyses of inoculation with Xcc306, a high proportion of host genes are associated with cell wall metabolism (Fig. S8A). Six of the top 10 induced genes by PthA4 in sweet orange, for example, are predicted to be involved in cell wall metabolism (Table 1), and 12% of the genes induced greater than 16-fold after inoculation of sweet orange with Xcc306 wild type compared with the deletion mutant Xcc306∆pthA4 are categorized as involved in cell wall metabolism (Fig. S8A). However, the expression of the host genes may be due to off-target TAL effector-mediated expression. To examine the association of CsLOB1 expression and the expression of cell wall-related genes in more detail, a select group of genes relating to expansion and wall metabolism were chosen from the most induced genes during infection by strains carrying pthA4 and tested for induction by qRT-PCR in the presence of dCsLOB1.1. The rationale for the approach was that PthA4 and dCsLOB1.1 have different EBEs and, consequently, are unlikely to share the same set of off-target genes, and coinduced genes may be associated with CsLOB1 expression. The approach revealed that genes for pectate lyase, extension, α-expansin, and cellulose, which were highly up-regulated by PthA4, were all found to be up-regulated with dCsLOB1.1 (Fig. 10). PthB inoculation in grapefruit was also accompanied by elevated expression of cell wall-associated genes (Fig. S8B). Expression of six genes was measured 36 h after inoculation with X. citri to determine if expression was sensitive to protein translation inhibitor cycloheximide (CHX) treatment. Transcription of TAL effector-targeted genes has been shown experimentally to be CHX-insensitive, because the transcription factor is synthesized in the bacterium and no new host translation is required (31). Addition of CHX with inoculation led to the inhibition of Cit.7877, 39387, 20509, and 2392 transcript, whereas expression of CsLOB1 and CsSWEET1 was expressed at high levels (Fig. 11A). The same four genes were also found to be elevated upon transient expression of 35S::CsLOB1, whereas CsSWEET1, whose expression is not hypothesized to be controlled by CsLOB1, was not elevated (Fig. 11B). However, the transient overexpression of CsLOB1 alone did not result in the formation of an observable pustule phenotype.

Fig. 10.

Cell wall-related genes are coinduced by synthetic TAL effectors that target CsLOB1. Representative cell expansion or wall-related metabolism genes in sweet orange were activated by Xcc306∆pthA4::dCsLOB1.1 in comparison with Xcc306∆pthA4. Quantitative RT-PCR was conducted on host mRNA using gene-specific primers at 48 h after the infiltration. Expression values were normalized to housekeeping gene EF1α. The probe sets are labeled and annotated as follows: Cit.8700, Cit.8700.1.S1_at, extension; Cit.30858, Cit.30858.1.S1_at, expansin; Cit. 20509, Cit.20509.1.S1_at, pectate lyase; Cit. 2392, Cit.2392.1.S1_at, acidic cellulose; Cit. 3027, Cit.3027.1.S1_s_at, CsSWEET1; Cit. 37210, Cit.37210.1.S1_at, CsLOB1; Cit. 7877, Cit.7877.1.S1_at, expansin; Cit. 39387, Cit.39387.1.S1_at, pectate lyase.

Fig. 11.

Cell wall-related gene expression is sensitive to cycloheximide (CHX) and associated with transient CsLOB1 expression. (A) Semiquantitative RT-PCR at 36 h after the infiltration of mutant Xcc306∆pthA4 and wild-type Xcc306 in the presence (+) or absence (−) of CHX. (B) Quantitative RT-PCR assays were conducted 6 d after infiltration of Agrobacterium with 35S::CsLOB1 or empty vector. Expression values were calculated in relation to water infiltration. The representative cell wall-related genes were a subset of genes used in Fig. 10. Error bars represent 1 SD. Values between treatments were normalized to the housekeeping gene EF1α.

Discussion

Investigations into the TAL effector-mediated effects on host gene expression revealed the remarkable probing by Xanthomonas species for vulnerabilities in host physiology using TAL effectors. Based on the results, we propose that pustule formation involves cooption of a single host S gene CsLOB1 in two citrus species, sweet orange and grapefruit, via any one of the PthA homologs. The results also have historical interest in that PthA of X. citri ssp. citri was one of the first T3S effectors demonstrated for being essential for virulence and the first TAL effector determined to be essential for pustule formation in CBC and with disease symptoms (20). Although PthA itself was not tested specifically in this study, the effector has the same predicted target site as PthA4 (18). Subsequently, a variety of TAL effector genes have been discovered that are required for pustule formation symptoms of CBC (30). Genes for the TAL effectors PthA4, PthAw, PthA*, PthB, and PthC from genetically diverse Xanthomonas strains that cause CBC restore pustule formation to the impaired strain Xcc306∆pthA4. On the basis of microarray and qRT-PCR expression analyses, all PthA variants were associated with an increase in CsLOB1 expression upon infection. To further substantiate the claim and distinguish targeted from collateral and possible off-target gene inductions, dTALes were designed and targeted to unique or optimal binding sites within the CsLOB1 promoter with the rationale that effectors targeting alternate sites are unlikely to have the same off-target sites. Only dTALes targeting CsLOB1 restored pustule formation and enhanced bacterial growth when expressed in bacteria. The dTALes targeting another gene that was predicted to have an EBE—namely, CsSWEET1, did lead to CsSWEET1 expression but did not lead to pustule formation or enhanced bacterial growth. Finally, promoter reporter assays also demonstrated that CsLOB1 was indeed expressed in a TAL effector-dependent manner in both citrus and N. benthamiana expression assays, and CsLOB1 expression was less sensitive to cycloheximide inhibition in the presence of PthA4. The predicted EBEs in CsLOB1 promoter meet the general prediction requirements, and, for the most part, the results of experimental tests of EBE function for CsLOB1 were consistent with predictions (32). The truncated version of the target site in CsLOB1, which eliminated the upstream sequences, was also functional both in citrus and Nicotiana. Changes in the EBE box in some instances dramatically altered expression. Noncanonical substitutions in the proximal 5′ half of the binding site had severe effects for PthA4-mediated expression compared with changes in the distal sequence of TTT, and a single base insertion, which throws the distal part out of register, also eliminated effector-mediated expression for CsLOB1. Changes in the respective TATAA boxes for CsLOB1 and CsSWEET1 eliminated expression, as might be expected for the predicted TATAA boxes for each gene. When the predicted EBE for CsLOB1 was added to a non–EBE-containing promoter (Cit.7877.1.S1_at), the gene acquired PthA4-mediated expression ability.

The predicted EBE for PthB and PthC overlaps and starts 6 bp upstream of the EBE_PthA4. The arrangement is similar to the Os11N3 promoter in rice, which contains EBE_PthXo3 in front of the EBE_AvrXa7 (24). The difference between PthXo3 and AvrXa7 has been postulated to result from the avoidance of triggering incompatibility in rice lines with the R gene Xa7. No TAL effector-dependent dominant R genes have been identified in citrus. Alternatively, PthB and PthC were isolated from X. fuscans ssp. aurantifolii, and differences may simply reflect convergence to a functional EBE. A more appropriate comparison may be to TalC, which targets an upstream site in the Os11N3 promoter (28). TalC arose in strains of X. oryzae pv. oryzae, which are limited to West Africa. TalC may represent an independent convergence to the rice S gene Os11N3 and not represent an adaptation to Xa7, which has not been deployed in West Africa.

CsLOB1 represents the third disease complex in which the natural target of a TAL effector has been identified, and, in each host species involved, a unique class of host gene was identified. With the exception of several rice S genes, the evidence linking gene induction to disease susceptibility is correlative, and, ultimately, an understanding of CsLOB1 function in CBC, and other TAL-dependent S gene products in other disease complexes, will come from genetic and molecular/biochemical analyses of the gene and gene product. In rice, members of the SWEET gene family function as S genes for bacterial blight, and a variety of promoter sequence polymorphisms has been identified in resistant genotypes or engineered that interfere with specific TAL-effector induction of individual S genes (29, 33, 34). CsLOB1 is a member of the plant-specific LOB family of transcription factors. A target of TAL effector AvrBs3 in pepper is upa20, an auxin-responsive gene for a transcription factor in the large bHLH family (22). No promoter polymorphisms have been identified in citrus or pepper cultivars, and no engineered alterations have been constructed.

Further insight of CsLOB1 involvement will come from characterization of the normal function of the gene. The plant-specific LOB domain family is composed of a conserved DNA-binding Cys repeat motif (CX₂CX₆CX₃C), an invariant glycine residue, and a coiled-coil Leu zipper-like motif (LX₆LX₃LX₆L), the latter of which often functions in protein–protein interactions (35). Although the specific functions of CsLOB1 are unknown, previous studies have revealed that LOB domain proteins are involved in the regulation of lateral organ development, anthocyanin and nitrogen metabolism, and are responsive to phytohormones and environmental stimuli such as auxin, cytokinin, gibberellin, brassinosteroid, and salinity or glucose (36, 37). One member of the LOB domain family, AtLBD18, was reported to bind with the promoter of EXPANSIN14, a gene involved in cell wall loosening (38). Recently, the Arabidopsis LOB family protein LBD20 was proposed as a host S gene for the fungal pathogen Fusarium oxysporum, functioning in the jasmonate signaling pathway (39). AtLBD20 was induced by F. oxysporum, and the overexpression of LBD20 was correlated with increased susceptibility to infection and reduced the expression of JA-regulated genes VEGETATIVE STORAGE PROTEIN2 (VSP2) and THIONIN2.1 (Thi2.1). Other LOB domain family genes were also detected to be responsive to fungal and root pathogens from public Arabidopsis array data (40). Here, we showed that CsLOB1 expression is associated with expression of numerous cell wall-related enzymes, indicating a possible function in cell wall biochemistry. Expression of four of the genes was shown to be sensitive to cycloheximide inhibition, whereas expression of the TAL-dependent genes CsLOB1 and CsSWEET1 were not sensitive. Furthermore, transient expression of CsLOB1 led to elevated levels of the four genes, whereas CsSWEET1 remained below detection; however, transient expression of CsLOB1 did not result in observable pustule formation, possibly as a result of the low percentage of Agrobacterium-transfected cells. Further functional analyses of CsLOB1 and associated genes will be required to establish a direct causal relationship to the virulence of Xcc and expression of host genes other than CsLOB1. Pustule formation in CBC involves both hyperplasia and hypertrophy of cells (41, 42). It is interesting in this regard that an interaction between an LOB and bHLH transcription factor was reported for Arabidopsis (43). Although correlative, transient expression of upa20, a member of the bHLH family, has been reported to induce cell hypertrophy, one of the phenotypes of pepper infection by Xanthomonas axonopodis pv. vesicatoria containing avrBs3 (22). Future insights into CsLOB1 and UPA20 may reveal targeting a common pathway at different control points for diseases involving X. axonopodis pv. vesicatoria, X. citri ssp. citri, and X. fuscans ssp. aurantifolii.

The results for the candidate EBEs in CsSWEET1 were more complex. Only the longer promoter construct supported TAL effector-mediated induction. Truncation of the CsSWEET1 promoter to include only the TATAA box region could not support expression either in citrus or N. benthamiana. At the same time, a change similar to the one conducted for CsLOB1 in the distal TTT sequence of EBE site A also resulted in loss of expression in transient assays, and both naturally occurring effectors PthB and PthC directed CsLOB1 but not CsSWEET1 expression. Binding to the CsSWEET1 promoter may, therefore, be weak, and expression may be due to multiple binding sites, including ones upstream or more complex interactions. No evidence was found supporting a major function for the SWEET gene CsSWEET1 in CBC, although a function in CBC could not be ruled out with certainty. The gene is a member of another transporter clade from which no S gene in rice has been identified. Rice-susceptible SWEET genes OsSWEET11and OsSWEET14, which are in clade III, preferentially mediated efflux of sucrose over glucose, whereas CsSWEET1 is in clade I with AtSWEET1 that has been shown to predominantly transport several hexoses (44). All of the clade III OsSWEETs but not the other SWEET paralogs in rice can, potentially, condition host susceptibility and X. oryzae pv. oryzae virulence (45). Why, in rice, the type of transporter is important for susceptibility is unknown. However, CsSWEET1 shows TAL effector-dependent expression for PthA4, PthAw, and PthA*, and full complementation of bacterial growth in citrus leaves was not attained with the dCsLOB1.1 despite apparent robust expression of CsLOB1. The question remains as to whether expression of CsSWEET1 or other host genes represents more complex virulence adaptations on the part of the bacteria and TAL effector-mediated expression.

The EBEs of TAL effector have been proposed as effective tools to control the disease. By combining the natural EBEs of multiple TAL effectors from three distinct R genes or adding artificial EBEs of corresponding TAL effectors into one complex R gene promoter, the engineered R gene was induced by these effectors and conferred broader spectrum disease resistance (46, 47). Although no major R genes have yet been described in citrus, the well-characterized avr gene AvrGf1 may reflect the presence of a potential R gene in grapefruit and sweet orange (3). Alternatively, we may engineer the EBEs identified in CsLOB1 to drive the expression of AvrGf1 and transform into grapefruit or sweet orange; the AvrGf1 will be ectopically activated in plant when it encounters most of the canker-causing xanthomonads; and the encoded AvrGf1 can be recognized by the potential R protein and trigger a hypersensitive response (48). Another compelling approach using TAL effector-targeted S gene to control disease is reported by Li et al. (34), who mutated the EBEs of Os11N3 in rice by transcription activator-like effector nuclease-based cleavage and gained transgenic rice lines conferring resistance to X. oryzae pv. oryzae that contained TAL effectors AvrXa7 and PthXo3; CsLOB1 is a good candidate for this approach in that it is targeted by several TAL effectors and only one single mutation in EBEs region is required to obtain broad spectrum plant resistance to most kinds of citrus canker.

Materials and Methods

Plant Material, Bacterial Strains, and Plasmids.

Growth conditions of plants and bacteria are described in SI Materials and Methods. The plasmids and bacterial strains used in this study are list in Table S1.

Mutagenesis of pthAs in Xcc306.

The site-directed gene deletion process was described in SI Materials and Methods.

Bacterial Growth in Planta.

For the population of bacteria strains in citrus plants, one leaf disk with 1 cm² of the inoculated area was taken and macerated in sterile tap water; after serial dilutions, 50 µL were plated on nutrient agar medium and incubated at 28 °C for 3 d. The colony counts were calculated to determine the internal populations. Each experiment was repeated three times.

Microarray Analyses and TAL Effectors Target Search.

The microarray was conducted and analyzed as described in SI Materials and Methods. Genes with a P ≤ 0.01 were considered as differentially expressed genes at a statistically significant level. The 1,000-bp upstream sequences of selected genes were obtained from Phytozome (www.phytozome.org/citrus.php), and the regions were scanned by Target Finder using RVD sequences of PthA4 and PthAw (49).

Designer TALe Construction.

Four types of repeats encoding the RVDs NI, NN, NG, and HD that correspond to the respective nucleotide A, G, T, and C were used to assemble the repeat domains of the artificial dTALes. The description of library repeats and protocols involving Golden Gate cloning strategy were as described previously (50).

FRET Analysis in HEK293T Cells.

HEK293T cells were cotransfected with the sensors FLIPsuc90μΔ1V plus AtSWEET12 (as positive control for sucrose uptake) and FLIPglu600μD13V plus AtSWEET1 (as positive control for glucose uptake) or CsSWEET1 in six-well plates, and perfusion experiments were performed as described previously (51). HEK293T/FLIPsuc90m∆1V cells were perfused with medium, followed by a pulse of 10 mM sucrose, whereas HEK293T/FLIPglu600μD13V cells were perfused with medium, followed by a pulse of 2.5–5–20 mM glucose.

Quantitative, Semiquantitative RT-PCR Analyses and GUS Assays.

Citrus leaves were syringe-infiltrated with bacterial suspensions at 5 × 10⁸ cfu/mL. For cycloheximide treatment, the Xcc306 bacterial suspensions containing 100 µM cycloheximide were used, and the leaf tissues were harvested 36 h after inoculation. Total RNA was extracted by using TRIzol Reagent (Ambion) following the manufacturer’s instruction. The RNA was subjected to DNase I treatment and first-strand cDNA synthesis by using the ProtoScript AMV First Strand cDNA Synthesis Kit (NEB); two-step real-time PCR was performed using Real Master Mix SYBR Rox (5 PRIME). The gene-specific primer sequences are listed in Table S2. The elongation factor gene EF1α was used as endogenous control. The 2^−∆∆Ct method was used for relative quantification. The quantitative and qualitative GUS assays were described in SI Materials and Methods.