Five phylogenetically close rice SWEET genes confer TAL effector-mediated susceptibility to Xanthomonas oryzae pv. oryzae



  • Bacterial plant-pathogenic Xanthomonas strains translocate transcription activator-like (TAL) effectors into plant cells to function as specific transcription factors. Only a few plant target genes of TAL effectors have been identified, so far. Three plant SWEET genes encoding putative sugar transporters are known to be induced by TAL effectors from rice-pathogenic Xanthomonas oryzae pv. oryzae (Xoo).
  • We predict and validate that expression of OsSWEET14 is induced by a novel TAL effector, Tal5, from an African Xoo strain. Artificial TAL effectors (ArtTALs) were constructed to individually target 20 SWEET orthologs in rice. They were used as designer virulence factors to study which rice SWEET genes can support Xoo virulence.
  • The Tal5 target box differs from those of the already known TAL effectors TalC, AvrXa7 and PthXo3, which also induce expression of OsSWEET14, suggesting evolutionary convergence on key targets. ArtTALs efficiently complemented an Xoo talC mutant, demonstrating that specific induction of OsSWEET14 is the key target of TalC. ArtTALs that specifically target individual members of the rice SWEET family revealed three known and two novel SWEET genes to support bacterial virulence.
  • Our results demonstrate that five phylogenetically close SWEET proteins, which presumably act as sucrose transporters, can support Xoo virulence.


Plant pathogens threaten the production of food crops worldwide. The bacterial pathogen Xanthomonas oryzae pv. oryzae (Xoo) causes bacterial leaf blight (BLB), one of the most devastating diseases of rice (Nino-Liu et al., 2006), which is a primary food plant for half of the world's population. Virulence of Xoo relies on injection of a cocktail of effector proteins into plant cells via a specialized type III secretion system (White & Yang, 2009). TAL (transcription activator-like) effectors constitute a major Xoo effector family as many as 19 genes per strain (Salzberg et al., 2008). TAL effectors employ tandem, nearly identical 34-amino-acid repeats to directly bind specific DNA sequences in promoter regions. Guided by a repeat variable diresidue (RVD) specific for one of the four DNA bases, or combinations thereof, each repeat binds one base pair in a contiguous sequence (Boch et al., 2009; Moscou & Bogdanove, 2009; Boch & Bonas, 2010). The DNA-binding domain, together with a C-terminal activation domain, enables TAL effectors to function as specific transcription factors that induce expression of target plant genes to the benefit of the pathogen (Boch & Bonas, 2010). The modular TAL effector architecture also allows one to artificially rearrange repeats to generate artificial TAL effectors with novel, predictable DNA specificities (Boch et al., 2009). Artificial TAL effectors (also called designer TAL effectors) have been used to analyze the specificities and efficiencies of RVDs (Boch et al., 2009; Christian et al., 2012; Cong et al., 2012; Streubel et al., 2012) and induce target gene expression in different organisms (Morbitzer et al., 2010; Geißler et al., 2011; Miller et al., 2011; Zhang et al., 2011; Bultmann et al., 2012; Li et al., 2013; Maeder et al., 2013; Perez-Pinera et al., 2013). Consequently, TAL effectors have become widely used in biotechnology as specific activators, repressors, and nucleases in many different organisms (Bogdanove & Voytas, 2011; Mussolino & Cathomen, 2012).

Several TAL effectors contribute significantly to bacterial virulence (Boch & Bonas, 2010), but in general little is known about their biological virulence targets. Because virulence of most Xoo strains mainly relied on the presence of individual TAL effectors and thus the ability to induce specific host genes, these targets have been termed susceptibility genes (White & Yang, 2009). The rice SWEET/nodulin-3 gene family holds the best-studied examples of TAL effector virulence targets so far. Two SWEET genes, Os8N3 and Os11N3 (also called OsSWEET11 and OsSWEET14, respectively), have been identified as targets of four different TAL effectors from various Xoo strains (Chu et al., 2006a; Yang et al., 2006; Antony et al., 2010; Chen et al., 2012). A third SWEET gene, Xa25, was also shown to be induced after infection with Xoo strain PXO339 (Liu et al., 2011), and it is possible that Xa25 is induced by an hitherto not identified TAL effector. The SWEET proteins form a heterogeneous family, which is divided into at least four clades in plants (Chen et al., 2010). SWEETs are membrane proteins that are involved in diverse functions, such as pollen development (Guan et al., 2008), senescence (Quirino et al., 1999), and microbe–plant interactions (Gamas et al., 1996; Chu et al., 2006a,b; Yang et al., 2006; Antony et al., 2010; Liu et al., 2011). Arabidopsis and rice genomes include 17 and 22 paralogs, respectively. By contrast, chordates and arthropods contain only one and Caenorhabditis elegans seven SWEET genes, respectively. This suggests that SWEETs may have undergone higher diversification in plants to potentially accompany the evolution of the vascular system (Baker et al., 2012). Recent data demonstrated that some members of the SWEET family facilitate sucrose and glucose efflux from the phloem parenchyma cells to the apoplast (Chen et al., 2010, 2012). This gave rise to the hypothesis that the pathogen-mediated induction of certain SWEET genes might increase sugar availability in the apoplast, and therefore plays a role in Xoo nutrition. Members of the SWEET gene family in Arabidopsis are also induced after infection with Pseudomonas syringae strain DC3000 (Chen et al., 2010), which indicates that this plant gene family has a general function for pathogen proliferation in the host.

Here we describe that a novel TAL effector, Tal5 from Xoo, also targets the rice OsSWEET14 gene. This observation provides compelling evidence that evolution of TAL effectors can converge on the same or highly related virulence targets. The importance of SWEETs for susceptibility towards Xoo prompted us to systematically study the role of this gene family for Xoo virulence. We constructed a series of artificial TAL effectors, which were designed to induce expression of individual SWEET genes in rice, and analyzed the ability of these TAL effectors to function as artificial virulence factors for Xoo. Our findings demonstrated that only the particular subgroup of rice SWEETs that are predicted to be sucrose transporters can be exploited as Xoo virulence targets. They also indicated that diverse SWEET family members probably fulfill different physiological functions. Finally, our approach demonstrates that artificial TAL effectors are suitable tools to study if candidate host genes are bona fide virulence targets for Xanthomonas.

Materials and Methods

Bacterial strains and growth conditions

The bacterial strains used in this study were Escherichia coli DH5α (Stratagene, La Jolla, CA, USA), Agrobacterium tumefaciens GV3101 (Van Larebeke et al., 1974) and Xoo strains BAI3 (Gonzalez et al., 2007), BAI3ΔtalC (Yu et al., 2011) and PXO99A (Hopkins et al., 1992). E. coli cells were cultivated at 37°C in lysogenic broth (LB) medium, Xoo strains at 28°C in PSA medium (10 g peptone, 10 g sucrose, 1 g glutamic acid, 16 g agar, l–1 H2O) and A. tumefaciens GV3101 at 28°C in YEB medium. Plasmids were introduced into E. coli and A. tumefaciens strains by electroporation and into Xoo by conjugation using pRK2013 as a helper plasmid in triparental mating (Figurski & Helinski, 1979). Rifampicin- and gentamicin-resistant clones were selected upon plating on PSA medium and one isolate was chosen for further experiments. Antibiotics were added to the medium at the following final concentrations: rifampicin, 100 μg ml−1; gentamicin, 20 μg ml−1.

Plant material and plant inoculations

Experiments were performed under glasshouse conditions under cycles of 12 h of light at 28°C and 80% relative humidity (RH) and 12 h of dark at 25°C and 70% RH. Oryza sativa ssp. japonica cv Nipponbare and the near-isogenic O. sativa ssp. indica lines IR24 and IRBB13 were used for virulence assays, growth curves and quantitative reverse transcription polymerase chain reaction (qRT-PCR). Leaves of 3-wk-old plants were infiltrated with a bacterial suspension with an optical density at 600 nm (OD600) of 0.5 using a needleless syringe, as previously described (Reimers & Leach, 1991), and symptoms (water-soaked lesions) were scored 6 d postinoculation (dpi). Leaf-clip inoculation was performed on 4- to 5-wk-old rice plants using a bacterial suspension with an OD600 of 0.2 (Kauffman et al., 1973), and sizes of lesions were measured 15 dpi. Statistical significance of the results was assessed using the Tukey honest significant difference test for post-ANOVA pairwise comparisons, set at 5% (< 0.05).

Nicotiana benthamiana plants were grown under 16 h of light, 40–60% RH, at 23 : 19°C, day : night in the growth chamber. Leaves of 4- to 6-wk-old plants were inoculated with A. tumefaciens strains using a needleless syringe.

β-Glucuronidase (GUS) reporter constructs and GUS assays

β-Glucuronidase assays from plant samples were essentially performed as described by Boch et al. (2009). Briefly, coding regions of talC and tal5 were inserted into pGWB2 under control of a 35S promoter (Nakagawa et al., 2007). A PCR-amplified 341 bp fragment containing the OsSWEET14 promoter region was cloned into pENTR/D-TOPO (Life Technologies GmbH, Darmstadt, Germany) and then fused to the uidA reporter gene by LR recombination into pGWB3 (Nakagawa et al., 2007; Yu et al., 2011).

To assay reporter activity, A. tumefaciens strains delivering TAL effector constructs and GUS reporter constructs were resuspended in infiltration medium (0.1 M 2-(N-morpholino) ethanesulfonic acid (MES), 1 M MgCl2, 0.1 M acetosyringone), mixed in equal amounts and inoculated into N. benthamiana leaves at a total OD600 of 0.8. Two days postinfiltration, leaf discs were sampled and GUS activities were quantified using 4-methyl-umbelliferyl-β-D-glucuronide (MUG). Protein concentrations were determined using Bradford assays. Data were compiled from triplicate samples originating from different plants.

RNA isolation and qRT- PCR

Leaves of 3-wk-old rice plants of the varieties Nipponbare, IR24 or IRBB13 were infiltrated with water or with the different Xoo strains using an OD600 of 0.5. At 1 dpi, leaf segments from three plants were ground into a fine powder using the Qiagen TissueLyser system (30 rps for 30 s). Rice total RNA was isolated using the Qiagen RNeasy kit following the manufacturer's recommendations (Qiagen). cDNA was generated from 2 μg RNA using the Fermentas first-strand cDNA synthesis kit following the manufacturer's recommendations (Thermo Fisher Scientific Inc., Waltham, MA, USA). Real-time PCR was performed using the iCycler (Bio-Rad, München, Germany) and 10 μl (50%) ABsolute qPCR SYBR Green Fluorescein Mix (ABgene Limited, Hamburg, Germany), 1 pmol forward, 1 pmol reverse oligonucleotide primer and 12.5 ng template cDNA. For each gene, a minimum of two independent biological replicates were analyzed, each with two technical qPCR replicates. The specificity of the primer pairs was checked by a melting curve and the PCR product was documented by agarose gel electrophoresis. The amplification efficiency for each primer pair was analyzed using a standard curve plot of a dilution series. cDNA amounts were normalized using actin as a reference gene. All RT primer sequences and corresponding amplification efficiencies are provided in Supporting Information, Table S1.

In planta growth assays

Leaves of 3-wk-old IR24 or IRBB13 plants were inoculated with a bacterial suspension of Xoo strains using an OD600 of 0.2. A 1 cm2 infected leaf segment was collected 6 d after infiltration, and ground into a fine powder using the Qiagen TissueLyser system (30 rps for 30 s). Ground material was resuspended in 1 ml sterile water, and 5 μl drops of a dilution series were spotted as triplicates onto selective PSA plates containing rifampicin and gentamicin. This experiment was performed three times.

Expression analysis by Western blotting

The X. oryzae pv. oryzae BAI3ΔtalC strains carrying a plasmid encoding an artificial or native TAL effector, or empty vector, were grown in liquid PSA medium supplemented with 20 μg ml−1 gentamicin at 30°C. Cells of 1 ml of a bacterial suspension at an OD600 of 0.2 were harvested and the TAL effector expression was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting using an anti-Flag antibody (Sigma-Aldrich).


Tal5 from Xoo strain MAI1 is a novel TAL effector that targets OsSWEET14

We analyzed the TAL effector repertoire of an African Xoo isolate from Mali, MAI1, which harbors a relatively small set of only eight TAL genes (Gonzalez et al., 2007). We screened a MAI1 genomic DNA cosmid library for clones containing TALs and subcloned eight BamHI fragments of different sizes containing putative TAL genes. Sequencing revealed that one of them, named tal5, encodes a new TAL effector with a unique order of 17.5 DNA-binding repeats (Fig. 1a).

Figure 1.

Tal5 is a novel transcription activator-like (TAL) effector predicted to target OsSWEET14. (a) Hypervariable amino acids at positions 12 and 13 (repeat variable diresidue) of the 17.5 repeats of Tal5. A single amino acid code was used for each amino residue (* represents a missing 13th residue). (b) The 370 bp sequence of the OsSWEET14 promoter region is depicted by the TalC box in bold (Yu et al., 2011), the AvrXa7 box underlined (Römer et al., 2010) and the candidate Tal5 box highlighted in gray. The artificial TAL effector (ArtTAL) boxes for ArtTAL14-1 and ArtTAL14-2 are indicated by dotted and dashed frames, respectively. The 5′ terminal nucleotide of each target box is indicated in white font. The predicted start codon of OsSWEET14 is indicated in bold italic font.

Using a program to scan rice promoters for potential TAL effector binding sites (Pérez-Quintero et al., 2013), we predicted which candidate rice genes are targeted by Tal5 (Table S2). Interestingly, the top score was obtained for a binding site located within the promoter region of the gene Os11 g31190 (hereafter OsSWEET14), which is a member of the SWEET/nodulin-3 gene family (Baker et al., 2012). Remarkably, OsSWEET14 is a host susceptibility gene that was previously reported as a target of the major virulence TAL effectors AvrXa7, PthXo3 and TalC from Xoo strains PXO86, JXO1A and BAI3, respectively (Chu et al., 2006; Antony et al., 2010; Yu et al., 2011). The target boxes for TalC and AvrXa7/PthXo3 are upstream of and overlapping the predicted OsSWEET14 TATA box, respectively. By contrast, the predicted Tal5 box is located downstream of the TATA box (Fig. 1b).

To test if OsSWEET14 is a direct virulence target of Tal5, we performed two complementary assays. First, we used a talC deletion mutant of Xoo BAI3, BAI3ΔtalC, which is impaired in OsSWEET14 induction, resulting in fewer symptoms and reduced proliferation of the bacteria in planta compared with the parental strain (Yu et al., 2011). Leaves of the susceptible rice variety IR24 were infiltrated with derivatives of BAI3ΔtalC containing the empty cloning vector or plasmids that encode talC or tal5 (Methods S1). Water and wildtype bacteria of strain BAI3 served as negative and positive controls, respectively. qRT-PCR analyses showed that OsSWEET14 was induced by both TalC and Tal5, but not upon inoculation of rice leaves with BAI3ΔtalC carrying the empty vector (Fig. 2b). Second, we used Xoo PXO99A, which is not able to cause disease on xa13 rice lines (e.g. IRBB13) as a result of a deletion in the PthXo1 target box which prevents PthXo1-mediated induction of the susceptibility gene OsSWEET11 (Yang et al., 2006). Similar to the previous experiment, only Xoo PXO99A containing tal5 or talC induced OsSWEET14 expression in xa13 rice lines (Fig. 2b).

Figure 2.

OsSWEET14 is induced by Tal5 and artificial transcription activator-like (ArtTAL) effectors. (a) Schematic representation of the OsSWEET14 promoter region and associated natural TAL and ArtTAL effectors next to their target boxes. The coding sequence is represented by an open arrow and the TATA box is indicated. (b) OsSWEET14 expression levels. Leaves of Oryza sativa IR24 variety were infiltrated with water (Mock), Xanthomonas oryzae pv. oryzae (Xoo) strain BAI3, and Xoo BAI3ΔtalC derivatives carrying an empty vector (ev) or plasmids containing the TAL effector genes tal5, talC, artTAL14-1 or artTAL14-2. Rice IRBB13 leaves were infiltrated with water or with Xoo strain PXO99A derivatives carrying the same constructs as used for BAI3ΔtalC. OsSWEET14 transcript abundances were determined 1 d postinoculation (dpi) by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The bars represent a minimum of two biological samples taken from a collection of three leaves. The fold change was calculated in comparison to leaves treated with BAI3ΔtalC (ev) or PXO99A (ev). Error bars represent + SD. Actin was used as a reference gene to normalize cDNA amounts. (c) The OsSWEET14 promoter is recognized by Tal5 and ArtTALs. A 341 bp fragment of the OsSWEET14 promoter region was cloned into a GUS reporter vector and codelivered via Agrobacterium tumefaciens into Nicotiana  benthamiana leaves, along with 35S-driven talC, tal5, artTAL14-1, artTAL14-2 or hax3 (error bars indicate ± SD; = 3). MU, 4-methyl-umbelliferone. The Hax3-box cloned in front of the minimal pBs4 promoter served as the specificity control.

To analyze the functionality of the putative target box for Tal5, a 341 bp fragment containing the OsSWEET14 promoter region was cloned in front of a promoterless uidA reporter gene (Yu et al., 2011). The reporter construct and tal5 under control of a constitutive promoter were cotransformed into N. benthamiana using Agrobacterium. GUS assays showed that the OsSWEET14 promoter fragment was recognized by Tal5, thus leading to reporter gene activity (Fig. 2c).

Finally, to analyze whether or not Tal5-mediated OsSWEET14 induction affects infection, IR24 and IRBB13 rice lines were challenged with BAI3ΔtalC and PXO99A derivatives, respectively, and disease symptoms were scored upon leaf infiltration and leaf clipping. As shown in Fig. 3(a), tal5 partially complemented the BAI3ΔtalC mutant strain. Similarly, introduction of tal5 into PXO99A resulted in increased aggressiveness in IRBB13 lines (Fig. 3b). Altogether, our data indicate that Tal5 can function as a virulence factor in Xoo by up-regulating OsSWEET14, and potentially facilitates propagation of Xoo strain MAI1 in rice leaves.

Figure 3.

Tal5 and artificial transcription activator-like (TAL) effectors act as virulence factors. Phenotypes of Xanthomonas oryzae pv. oryzae (Xoo) strains on Oryza sativa leaves. (a) Leaves of IR24 plants were inoculated with strains BAI3 or BAI3ΔtalC carrying empty vector (ev) or TAL effectors talC, tal5, artTAL14-1, or artTAL14-2. (b) Leaves of IRBB13 plants were inoculated with strain PXO99A carrying the same constructs as used for BAI3ΔtalC. (a, b) Leaves were photographed 6 d postinoculation (dpi). Lesion length was measured at 15 d after leaf-clip inoculation. Data are the mean of eight measurements. Error bars represent ± SD. Bars with same letters are not statistically different based on a Tukey's honest significant difference test (α = 0.05). (c) Bacterial growth in IR24 rice leaves. Bacterial counts in leaf segments infiltrated with Xoo strains BAI3 or BAI3ΔtalC carrying empty vector (ev) or TAL effectors talC, tal5, artTAL14-1, or artTAL14-2 were analyzed 6 dpi. Error bars represent SD (= 3). An asterisk indicates a significant difference in Student's t-test at < 0.01 when comparing samples with BAI3ΔtalC (ev).

Artificial TAL effectors targeting OsSWEET14 effectively substitute for the virulence contributions of TalC and PthXo1

Although TAL effectors have a high target site specificity, they nevertheless induce the expression of a significant number of genes in host cells (Marois et al., 2002; Kay et al., 2009). While some of them are directly induced, others are up-regulated indirectly. Collectively, most of the (directly and indirectly) induced genes probably represent collateral targets that are not related to the infection process. Therefore, it is difficult to know which of these genes is responsible for the virulence effect of a given TAL effector and thus represents a bona fide host susceptibility gene.

We aimed to develop a functional test to determine if OsSWEET14 is indeed a valid virulence target for Xoo. Artificial TAL effectors can be constructed with a designed repeat array and a tailored DNA-binding specificity to induce expression of specified target genes (Morbitzer et al., 2010; Bogdanove & Voytas, 2011). We constructed two artificial TAL effectors (ArtTAL14-1 and ArtTAL14-2) that target sequences in the OsSWEET14 promoter and that are different from those that are recognized by the four natural TAL effectors (Methods S2; Figs 1b, 2a), for more information on the design of the artificial TAL effectors, see Methods S1 and S2. The rationale was that both sets, ArtTALs and natural TAL effectors, will induce expression of OsSWEET14, but it is unlikely that they have common collateral targets. Indeed, the prediction of target sites in rice for the ArtTALs, Tal5, and TalC supported the fact that they probably share no target genes besides OsSWEET14 (Table S3). We numbered the artificial TAL effectors according to their target OsSWEET paralog (i.e. ArtTAL14 for OsSWEET14).

As envisaged, both ArtTALs induced expression of OsSWEET14 when delivered by Xoo BAI3ΔtalC and PXO99A, respectively (Fig. 2b). In addition, both ArtTALs also recognized the OsSWEET14 promoter in transient GUS reporter studies in N. benthamiana (Fig. 2c). We then tested whether the ArtTALs contribute to the virulence of Xoo, similar to the natural TAL effectors targeting OsSWEET14. To this end, Xoo BAI3ΔtalC derivatives, complemented with plasmid-borne copies of artTAL14-1 or artTAL14-2, were infiltrated into leaves of the rice line IR24. Strikingly, both strains caused typical disease symptoms (water-soaked lesions), basically identical to Xoo strains BAI3 and BAI3ΔtalC containing talC or tal5 (Fig. 3a). Quantitative leaf-clipping assays using these strains showed that both ArtTALs partially restored virulence of the Xoo BAI3ΔtalC strain, reaching levels similar to those observed upon complementation with talC. Similarly, both artTAL genes conferred virulence to the Xoo strain PXO99A when assayed by leaf infiltration or leaf clipping of otherwise resistant xa13 rice plants (Fig. 3b).

Next, we quantified bacterial populations upon leaf infection with Xoo strains BAI3 and BAI3ΔtalC derivatives (Fig. 3c). At 6 dpi, the sizes of bacterial populations of all BAI3ΔtalC strains harboring natural or artificial TALs were significantly larger than those of BAI3ΔtalC carrying the empty vector.

Together, these data show that the artificial TAL effectors were functionally equivalent to natural ones and further demonstrate that induction of OsSWEET14 is sufficient to confer virulence to the Xoo strains tested. These data also emphasize that induction of OsSWEET14 can fully compensate the lack of OsSWEET11 expression in the PXO99A-IRBB13 combination. Together with previous studies (Antony et al., 2010), these findings strongly suggest that these two SWEETs have a similar function in respect of Xoo virulence.

A series of artificial TAL effectors induce expression of individual SWEET genes and unmask new potential virulence targets of Xoo

To investigate whether additional SWEET genes may act as virulence targets of Xoo, we adopted the same strategy as used for OsSWEET14 (Figs 2, 3) for the remaining 21 OsSWEET rice paralogs (Table 1). Generally, two ArtTALs (named according to the OsSWEET paralog) were designed for each rice SWEET gene, targeting specific sequences in the corresponding promoter regions, with the exception of OsSWEET7d and OsSWEET7e, for which no specific ArtTALs could be generated (Table 1). We predicted the top 20 possible target sequences for each ArtTAL and found that they target almost exclusively nonoverlapping sets of genes. Next, we evaluated the capacity of each of these ArtTALs to confer virulence to the Xoo strain BAI3ΔtalC. Only ArtTALs targeting OsSWEET11 to OsSWEET15 led to the production of typical water-soaked lesions (Fig. 4a). Strikingly, no disease symptoms were observed upon infiltration of plants with Xoo strains producing ArtTALs targeting any of the other SWEET paralogs (Table 1), as exemplified for OsSWEET1a, OsSWEET6b and OsSWEET7a in Fig. 4(a).

Table 1. Artificial transcription activator-like (ArtTAL) effector repeat variable diresidue (RVDs), target sequences and in planta activity
Gene IDaGene namebArtTALcRVD SequencedTarget boxePosition from ATGfWater soakinggqRT-PCRh
  1. a

    MSU Version 7.

  2. b

    Based on the nomenclature and clade classification from Chen et al. (2010).

  3. c

    For each gene two ArtTAL effectors binding to distinct target sequences were designed.

  4. d

    Amino acids in one letter code.

  5. e

    The 5′ terminal nucleotide of each target box is indicated in bold.

  6. f

    Position of ArtTAL binding box including the initial T counted from ATG (not included) in bp.

  7. g

    Leaves of Oryza sativa cultivar IR24 were infiltrated using a blunt syringe and scored at 6 d postinoculation (dpi). No, partial, and water-soaked lesion are indicated as (−), (+/−) and (+), respectively.

  8. h

    OsSWEET expression analysis using quantitative reverse transcription polymerase chain reaction (qRT-PCR). Induction, +; no induction, −; no amplification product, na; not determined, nd.

Figure 4.

Artificial transcription activator-like (TAL) effectors that specifically induce clade-III OsSWEET genes confer virulence to Xanthomonas oryzae pv. oryzae (Xoo). (a) Phenotypes of Xoo strains on rice leaves. Oryza sativa cultivars IR24 were inoculated with water (Mock), Xoo BAI3, or Xoo BAI3ΔtalC carrying empty vector (ev), talC or artificial transcription activator-like (artTAL) effectors targeting the five clade-III OsSWEET11 to OsSWEET15 genes, as well as OsSWEET1a, OsSWEET6b and OsSWEET7a. (b) Transcript abundances of clade-III OsSWEET genes SWEET11 to SWEET15 were determined 1 d postinoculation by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The bars represent a minimum of two biological samples taken from three leaves. The fold change was calculated in comparison to BAI3ΔtalC (ev). Error bars represent + SD. Actin was used as a reference gene to normalize cDNA amounts.

To exclude the possibility that lack of symptom formation was the result of a failure of SWEET gene induction, we checked whether or not the ArtTAL proteins are stably produced in Xoo and the target genes in rice are induced. Indeed, Western blot analysis indicated that all ArtTALs accumulated at significant levels and showed protein integrity in Xoo (Fig. S1). Quantitative RT-PCR experiments showed that most SWEET genes were induced in an ArtTAL-dependent manner, except for OsSWEET2c, OsSWEET5 and OsSWEET7c (Fig. S2). OsSWEET6a and OsSWEET7b were not analyzed, because no specific amplification products could be detected, although several pairs of primers have been tested (Table 1).

Interestingly, all five rice SWEET paralogs whose activation was associated with the development of typical disease symptoms belong to clade III of the SWEET gene family, which comprises OsSWEET11 (Os8N3/Xa13), OsSWEET12, OsSWEET13 (Xa25), OsSWEET14, and OsSWEET15. OsSWEET11 and OsSWEET14 are known major Xoo susceptibility genes of rice (Yang et al., 2006; Antony et al., 2010), which confirms the suitability of our strategy. Furthermore, we evaluated the basal and induced OsSWEET gene expression levels (Table S4), which suggest that clade III SWEETs are generally not induced at higher levels than nonclade III ones. However, we cannot rule out the possibility that a few of them act as susceptibility genes when induced at higher levels and/or under different conditions.

To demonstrate that the observed gain of virulence phenotypes can be solely attributed to the induction of the targeted SWEET genes, we evaluated the induction profile of each of the five clade-III paralogs by qRT-PCR during infection of plants with Xoo strains delivering different ArtTALs. Among the five candidate susceptibility genes we identified, only OsSWEET14 was activated upon infiltration of rice leaves with Xoo wildtype strain BAI3. None of them was induced upon infiltration with water or Xoo strain BAI3ΔtalC (Fig. 4b). In addition, only the expected target SWEET gene was induced upon infection of Xoo strains delivering the cognate ArtTAL, indicating that the observed phenotypes are not the result of induction of collateral susceptibility genes.

Finally, we quantified bacterial populations in planta in order to further demonstrate the function of the clade-III SWEET genes as virulence targets of Xoo. As shown in Fig. 5, BAI3ΔtalC derivatives carrying ArtTALs inducing any of the five clade-III SWEET genes grew to significantly larger population sizes than BAI3ΔtalC or BAI3ΔtalC carrying ArtTALs that target the other SWEET clade members.

Figure 5.

Artificial transcription activator-like (TAL) effectors that induce clade-III OsSWEET genes restore bacterial growth in rice leaves. Oryza sativa IR24 leaves were infiltrated with Xanthomonas oryzae pv. oryzae (Xoo) BAI3, Xoo BAI3ΔtalC carrying empty vector (ev), or artTAL11-2, artTAL12-2, artTAL13-2, artTAL14-2, artTAL15-1, artTAL1a-1, artTAL6b-1, and artTAL7a-1, which target OsSWEET11, OsSWEET12, OsSWEET13, OsSWEET14, OsSWEET15, OsSWEET1a, OsSWEET6b and OsSWEET7a, respectively. Bacterial populations were scored 6 d postinfiltration. Data are the arithmetic means from three plants and three independent experiments. Error bars represent ± SD. *Significant difference in Student's t-test at P < 0.01 when comparing samples with BAI3ΔtalC (ev); **, no differences were detected at P < 0.01.


Modern agriculture relies on effective control of pathogens. Better knowledge about host molecular determinants underlying plant diseases is thus essential to develop new strategies leading to plant resistance. However, only a few host genes have been characterized until now that are critical to promote bacterial diseases. These include members of the SWEET/nodulin-3 family, which were recently identified as essential alternative Xoo rice virulence targets, also called susceptibility genes (White & Yang, 2009). Current models suggest that bacteria hijack individual SWEETs that function as sugar transporters, leading to the accumulation of nutrients in the apoplast where Xoo multiplies. Previous studies identified OsSWEET11 and OsSWEET14 as susceptibility genes, as their induction by cognate native TAL effectors provides virulence to strains carrying them (Talbot, 2010).

Here, we further corroborate the key role of OsSWEET14 in bacterial virulence as a direct target of natural Xanthomonas TAL effectors. Tal5, a novel major virulence TAL effector from a Malian strain of Xoo, recognizes a distinct sequence within the OsSWEET14 promoter sequence, which differs from those that are recognized by strains from the Philippines and Burkina Faso. This functional convergence of virulence factors illustrates the evolutionary pressure that is imposed on Xoo strains of different geographic origins and genetic lineages to evolve TAL genes that induce this particular SWEET gene.

In a recent report, Li et al. (2013) identified OsSWEET12 as a third potential susceptibility gene. Upon infection of rice with Xoo mutant strains defective for OsSWEET11 activation, the authors showed that complementation with two independent ArtTALs activating OsSWEET12 leads to a gain of virulence. We significantly extend this study by providing additional important experiments. We analyzed bioinformatically that the ArtTALs target no other common genes besides OsSWEET14 (Table S2). We verified the ArtTALs' stability (Fig. S1) and assessed their binding specificity to their target promoter using GUS reporter assays (Fig. 2c). Importantly, we demonstrate that ArtTAL-mediated OsSWEET activation is not the result of collateral expression of other clade III paralogs (Fig. 4). Finally, we extended this strategy to the full SWEET gene family, systematically addressing the function of 20 of the 22 OsSWEET paralogs. These studies highlight the functionality of ArtTALs as useful tools to investigate the function of host genes during the plant–pathogen interaction. In total, we generated 36 ArtTALs, 26 of which (72%) were functional, and induced their target plant genes upon delivery by Xanthomonas. Few ArtTALs were not functional in our study, which might be because of nonoptimal choice of the target region or a noninducible status of the target gene. Overall, our findings emphasize that ArtTALs are suitable tools to induce host genes of choice and that bacterial delivery is a highly efficient means to do so. In summary, we confirm that OsSWEET11, OsSWEET12, and OsSWEET14 are essential bona fide host virulence targets of Xoo; we formally validate the function of OsSWEET13/Xa25 as a fourth susceptibility gene, as up to this point only indirect evidence has been available (Liu et al., 2011); we identify OsSWEET15 as a novel potential susceptibility gene which can support the development of disease symptoms triggered by Xoo; and, in total, we analyzed 20 OsSWEET paralogs, only five of which could be demonstrated to function as susceptibility genes.

Our data suggest that, in addition to the three already identified SWEET genes (OsSWEET11, OsSWEET12 and OsSWEET14), two more (OsSWEET13 and OsSWEET15) may be exploited by Xoo, presumably to satisfy the pathogen's nutritional needs. OsSWEET13 is activated by Xoo strain PXO339, but the corresponding TAL effector(s) remain(s) to be identified (Liu et al., 2011). By contrast, no native TAL effectors have been reported so far for OsSWEET12 and OsSWEET15. Interestingly, screening GenBank for TAL effectors predicted to recognize DNA boxes within their promoter indicated that the TAL protein AAW76267 of Xoo strain KACC10331 (Lee et al., 2005) and Tal7b/Tal8b of Xoo strain PXO99A (Salzberg et al., 2008) may target OsSWEET12 and OsSWEET15, respectively (Grau et al., 2013). Although transcriptomic data confirmed an increase in OsSWEET15 expression in plants challenged with PXO99A as compared with water (Grau et al., 2013), OsSWEET15 induction might not be sufficient for growth on IRBB13 (Fig. 3b), indicating that a threshold expression level might be required to support infection. Indeed, the target box for Tal7b/Tal8b appears to be suboptimal, as a G is found at the 5′ end of the predicted DNA box, which is known to be less efficient for gene activation (Boch et al., 2009).

Our study highlighted the fact that five OsSWEET genes may be manipulated by Xoo to favor disease. Interestingly, these five SWEET genes, OsSWEET11 to OsSWEET15, all belong to phylogenetic clade III (Fig. 6). Several A. thaliana SWEETs, as well as rice OsSWEET11 and OsSWEET14, are both low-affinity glucose and sucrose transporters (Chen et al., 2010, 2012). OsSWEET11 and OsSWEET14 are bidirectional sugar transporters and differ in their transport direction from the SUT/SUC high-affinity H+ sugar symporters of plants (Kühn & Grof, 2010; Chen et al., 2012). Therefore, increased expression of SWEETs alone might lead to a net efflux of sugars from the plant cells. Under physiological conditions, SWEET sugar transporters facilitate flux of sucrose from the phloem parenchyma cells into the apoplast, whereas SUT/SUC transporters load sucrose from the apoplast into the phloem companion cells and sieve elements (Kühn & Grof, 2010; Braun, 2012). Our data imply that all clade-III OsSWEETs share a function that can support multiplication of Xoo and that this function is absent from the other SWEET paralogs in rice. It is tempting to speculate that this common function is the export of a particular sugar, possibly sucrose.

Figure 6.

Phylogenetic tree of OsSWEET proteins. Amino acid sequences were obtained from UniProt (; Bairoch et al., 2005) and aligned using MUSCLE (; Edgar, 2004) with default parameters. The sequence of OsSWEET7a was corrected, as UniProt reports an erroneous tandem duplication of 54 amino acids at the N terminus. Alignments were analyzed by the neighbor-joining method using MEGA5.10 (; Tamura et al., 2011), assuming a Poisson substitution model, uniform rates among sites, and pairwise deletions for missing data and gaps. The numbers at the nodes represent bootstrap percentage values based on 1000 replications. Proteins of clade I, blue squares; clade II, red triangles; clade III, green dots; clade IV, purple circle.

At present, it is unclear whether Xoo relies on SWEETs only at certain infection stages. In a study involving Xoo strain BAI3, only vascular spreading, but not local growth of Xoo, is prevented in the absence of SWEET induction (Yu et al., 2011). Curiously, the locally restricted rice pathogen X. oryzae pv. oryzicola is also able to benefit from TAL effectors that mediate SWEET gene induction, although no native SWEET-targeting TAL effectors have been reported in Xoc (Verdier et al., 2012). Other pathogens, such as P. syringae pv. tomato DC3000 and the fungal pathogens Golovinomyces cichoracearum and Botrytis cinerea, also induce the expression of SWEET genes in Arabidopsis, including those that facilitate sugar transport (Chen et al., 2010, 2012). This suggests that different pathogens exploit plant sugar transporters by different means to gain access to carbohydrate nutrients.

The hallmark of our approach relies on its great potential to discover as yet unknown plant susceptibility genes. ArtTALs perfectly mimic the action of natural TAL effectors and will therefore enable the major plant genes controlling bacterial proliferation in any TAL-mediated plant disease to be pinpointed, such as Citrus canker and cassava bacterial blight (Duan et al., 1999; Rao et al., 2002). In addition, searches for unresponsive alleles will offer new resources to achieve sustainable plant resistance by marker-assisted breeding, thus circumventing transgenic approaches. Strategies aimed at producing rice lines with improved and broad resistance against BLB already incorporate the concept of gain of ‘resistance by defective susceptibility’, as successfully done by pyramiding the unresponsive allele of OsSWEET11, xa13 (Rao et al., 2002).

In addition to OsSWEET11, hijacking of OsSWEET14 by Xoo appears to be a widespread strategy to provoke disease. As reported by Li et al. (2012), 32 out of 40 Xoo strains collected worldwide were impaired for virulence when inoculated on TALEN-edited OsSWEET14 rice lines mutated in the AvrXa7/PthXo3 box. Our results on Tal5 and ArtTAL14s increase the number of experimentally proven native and artificial TAL effectors activating OsSWEET14 to six, which recognize five different target boxes (Fig. 1). This finding illustrates the impressive potential of TAL effectors to induce genes by binding to differently located target sequences. Therefore, it will be essential to take the natural diversity of TAL effector repertoires in Xoo populations into account before exploiting or generating unresponsive susceptibility target gene variants.


We thank C. Kretschmer for technical assistance, and U. Bonas for continued support. We are grateful to Y. Yu, M. Soto-Suárez and V. Verdier for providing the MAI1 DNA cosmid library. This project was supported by an EMBO junior grant to J.S. (ASTF no: 355-2011), a grant from the Deutsche Forschungsgemeinschaft (SPP1212) to J.B., a grant from the Agence Nationale de la Recherche (ANR-2010-GENM-013) to R.K., and a grant from the Fonds pour la formation à la Recherche dans l'Industrie et dans l'Agriculture (093604) to C.P. We are grateful to Alvaro Pérez-Quintero and Sébastien Cunnac for bioinformatic prediction of TAL targets and advice on statistical analyses, respectively.