Addition of transcription activator-like effector binding sites to a pathogen strain-specific rice bacterial blight resistance gene makes it effective against additional strains and against bacterial leaf streak
Aaron W. Hummel,
Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA 50011, USA
Author for correspondence: Adam J. Bogdanove Tel: +1 515 294 3421 Email: firstname.lastname@example.org
•Xanthomonas transcription activator-like (TAL) effectors promote disease in plants by binding to and activating host susceptibility genes. Plants counter with TAL effector-activated executor resistance genes, which cause host cell death and block disease progression. We asked whether the functional specificity of an executor gene could be broadened by adding different TAL effector binding elements (EBEs) to it.
•We added six EBEs to the rice Xa27 gene, which confers resistance to strains of the bacterial blight pathogen Xanthomonasoryzae pv. oryzae (Xoo) that deliver the TAL effector AvrXa27. The EBEs correspond to three other effectors from Xoo strain PXO99A and three from strain BLS256 of the bacterial leaf streak pathogen Xanthomonasoryzae pv. oryzicola (Xoc).
•Stable integration into rice produced healthy lines exhibiting gene activation by each TAL effector, and resistance to PXO99A, a PXO99A derivative lacking AvrXa27, and BLS256, as well as two other Xoo and 10 Xoc strains virulent toward wildtype Xa27 plants. Transcripts initiated primarily at a common site. Sequences in the EBEs were found to occur nonrandomly in rice promoters, suggesting an overlap with endogenous regulatory sequences.
•Thus, executor gene specificity can be broadened by adding EBEs, but caution is warranted because of the possible coincident introduction of endogenous regulatory elements.
Many crops, including rice, wheat, cotton, citrus, tomato, cassava, banana, soybean, sugarcane, and others, suffer losses as a result of infection by pathogenic members of the bacterial genus Xanthomonas. Rice (Oryza sativa), a staple for more than half the world’s population, is host to the xylem pathogen Xanthomonasoryzae pv. oryzae (Xoo), which causes bacterial blight, and the leaf mesophyll pathogen, Xanthomonasoryzae pv. oryzicola (Xoc), which causes bacterial leaf streak. These diseases reduce yields by up to 50 and 30%, respectively (Niño-Liu et al., 2006). Like several Xanthomonas species, Xoo relies on transcription activator-like (TAL) effectors to render the host susceptible to colonization (Bai et al., 2000; Yang & White, 2004; Yang et al., 2006; Sugio et al., 2007; Antony et al., 2010). Xoc also deploys TAL effectors, which, though less well studied, are presumed to contribute similarly to virulence (Makino et al., 2006; Bogdanove et al., 2011).
Named for features shared with eukaryotic transcription factors (Yang et al., 2006), TAL effectors are secreted into host cells via the bacterial type III secretion system (T3SS), then directed to the host cell nucleus by C-terminal nuclear localization signals (Yang & Gabriel, 1995; Van den Ackerveken et al., 1996; Szurek et al., 2001, 2002). There they bind to cognate effector binding elements (EBEs) in specific host gene promoters to activate transcription of those genes using a C-terminal, acidic activation domain (Zhu et al., 1998, 1999; Kay et al., 2007; Römer et al., 2007; Boch et al., 2009; Moscou & Bogdanove, 2009). Binding specificity is dictated by a variable number of central, 33–35 amino acid repeats. In each repeat, a pair of variable residues at positions 12 and 13 (together called the repeat-variable diresidue (RVD)) preferentially associates with a different nucleotide to define the length and sequence of the EBE (Boch et al., 2009; Moscou & Bogdanove, 2009). With this modular protein-DNA recognition mechanism, the pathogen can activate multiple susceptibility (S) genes in the host by deploying different TAL effectors. Several bacterial blight S genes in rice that correspond to Xoo TAL effectors important for virulence have been identified (Yang et al., 2006; Sugio et al., 2007; Antony et al., 2010), and several more S gene candidates for bacterial blight and bacterial leaf streak have been predicted computationally based on RVD sequences of uncharacterized Xoo and Xoc TAL effectors (Moscou & Bogdanove, 2009).
Plants counter TAL effector-wielding pathogens with S gene mimics that cause host cell death and block disease progression when transcriptionally activated. Such normally silent, EBE-controlled ‘executor’ resistance genes (Bogdanove et al., 2010) include the pepper (Capsicum annuum) Bs3 gene, which provides resistance against strains of the bacterial spot pathogen Xanthomonascampestris pv. vesicatoria that express the TAL effector AvrBs3 (Bonas et al., 1989; Römer et al., 2007), and the rice bacterial blight resistance gene Xa27, which is activated by the Xoo TAL effector AvrXa27 (Gu et al., 2005). In addition to executor genes, alleles of major S genes that are immune to activation by the corresponding TAL effector, such as the rice xa13 and xa25 bacterial blight resistance genes, provide another form of defense (Yang et al., 2006; Liu et al., 2011). A third type of resistance directed at TAL effectors, again exemplified by a rice bacterial blight resistance gene, xa5, is a polymorphism in the gamma subunit of general transcription factor TFIIA. A single amino acid substitution in the protein is thought to impair the ability of TAL effectors to recruit the transcriptional machinery to activate target genes (Iyer & McCouch, 2004; Sugio et al., 2007; Gu et al., 2009). Each of these types of resistance, however, is subject to defeat by adaptation of the pathogen TAL effector inventory. For example, executor genes can be defeated by mutation or loss of the corresponding TAL effector, provided the TAL effector is dispensable for virulence, as appears to be the case for AvrXa27 (Tian & Yin, 2009). S gene promoter polymorphisms that confer resistance can be overcome by TAL effectors that target the new sequence or an alternative S gene (Antony et al., 2010). Also, the xa5 gene is rendered ineffective by a strain with two apparent adaptations, a TAL effector that activates the corresponding S gene particularly strongly such that the reduction in activity caused by xa5 might be inconsequential, and a TAL effector that induces expression of a TFIIA gamma paralog that may substitute for the allele found in susceptible plants (Yang et al., 2006; Sugio et al., 2007; B. Yang, unpublished). Furthermore, the latter two types of resistance are genetically recessive, rendering them less easily bred into elite hybrids. A genetically dominant, broad-spectrum, and durable form of resistance effective against pathogens that deploy TAL effectors would be beneficial, but none has been identified.
In an Agrobacterium-mediated transient expression assay in Nicotiana benthamiana, Römer et al. (2009a) showed that amending the Bs3 gene promoter with the AvrXa27 EBE (which they call a ‘UPT box’, which stands for up-regulated by TAL effector) and an EBE matching an AvrBs3 variant called AvrBs3Δrep16 rendered the promoter responsive to all three TAL effectors. This pioneering study suggested that broad-spectrum and potentially durable resistance might be achieved by stable integration of an executor gene engineered in this way to respond to TAL effectors from multiple pathogen strains or even different pathogens.
In previous studies, stable integration of Xa27 into rice under conditions in which it was expressed constitutively resulted in reduced tillering, delayed flowering, and stiff, early-senescing leaves; nonetheless, the expression of the gene indeed conferred resistance to several Xoo strains normally virulent on wildtype Xa27 lines, and partial resistance to a strain of Xoc (Gu et al., 2005; Tian & Yin, 2009). We therefore chose Xa27 to test the notion suggested by Römer et al. (2009a) that functional specificity of an executor gene could be broadened, without deleterious effects associated with constitutive expression, by making its promoter responsive to several distinct TAL effectors.
We added to the Xa27 promoter EBEs corresponding to three additional TAL effectors each from the Xoo strain PXO99A, which harbors AvrXa27, and the Xoc strain BLS256, which does not. Stable integration of this construct into rice produced healthy lines exhibiting gene activation by each of the TAL effectors, and resistance to PXO99A, a PXO99A derivative lacking AvrXa27, and BLS256, as well as two other Xoo and 10 Xoc strains from a diverse collection virulent toward wildtype Xa27 plants. Our results establish the efficacy of executor gene promoter engineering for broader specificity. They also demonstrate that a rice gene for bacterial blight resistance can be readily modified to provide complete protection from bacterial leaf streak as well, a disease for which no major gene resistance in rice has been identified.
Materials and Methods
Modification of Xa27
A 2.4 kb sequence including the Xa27 gene (AY986492.1) was PCR-amplified from rice cultivar IRBB27 (Gu et al., 2005) with primers P250 (CACCTGCAGCTGAACCAAACAGTTTTAGC) and P251 (GGGCCCCACTTACTTATTTATTTATTTTTGCTGAC) using the Phire Hotstart polymerase (Thermo Fisher Scientific, Waltham, MA, USA) and cloned into the Gateway entry vector pENTR/D-TOPO (Life Technologies, Grand Island, NY, USA). To modify the promoter, custom oligonucleotides (Integrated DNA Technologies, Coralville, IA, USA) were synthesized to generate double-stranded DNA fragments with appropriate overhangs that were then cloned into the unique SpeI site 100 bp upstream of the Xa27 coding sequence and 14 bp upstream of the AvrXa27 EBE. Oligonucleotides were designed such that EBEs are flanked by the weak consensus context of TAT at the 5′ end and CCC at the 3′ end (Moscou & Bogdanove, 2009) and such that EBEs for TAL effectors from Xoo alternate with those for TAL effectors from Xoc. The EBE and control constructs (Fig. 1) were transferred to the binary vector pTF101.1gw1 (Plant Transformation Facility, Iowa State University, http://www.agron.iastate.edu/ptf/), which carries the bar gene for resistance to the herbicide phosphinothricin, using the Gateway LR II Clonase enzyme kit (Life Technologies).
Plant transformation and growth
Rice (Oryza sativa L. cv Kitaake) was transformed by the Iowa State University Plant Transformation Facility using Agrobacterium tumefaciens gene transfer in callus tissue, as previously described (http://www.agron.iastate.edu/ptf/protocol/Rice.PDF). Herbicide-resistant lines were advanced to the T2 generation for characterization. Presence of the transgene was assessed using the Kapa3G Plant PCR Kit (Kapa Biosystems, Woburn, MA, USA) with primers P664 (GGCATTCTTCCTTTCTTCAGC) and P352 (GGAGGCAGCTTCTTGGGTGTCTCAG). Transgenic and IRBB27 rice plants were grown in a growth chamber under cycles of 12 h, 28°C : 12 h, 25°C; light : dark. Fertilizer for acid-loving plants (30-10-10; Earl May, Shenandoah, IA, USA) and iron chelate micronutrient (Becker Underwood, Ames, IA, USA) were applied by watering twice a week at rates of 0.185 and 0.595 g l−1, respectively.
TAL effector clones
Clones encoding PthXo1, PthXo6, Tal9a, AvrXa27, and AvrXa10 were provided by B. Yang (Iowa State University), and clones encoding Tal2g, Tal4a, Tal4c, and Tal1c were from C. Schmidt (our laboratory). The SphI fragment of each TAL effector gene, encoding its central repeat region, was transferred into a functionally equivalent tal1c gene backbone (missing its own SphI fragment) in plasmid pCS466, a derivative of the pCR8/GW/TOPO TA entry vector (Life Technologies, Grand Island, NY, USA). Each effector construct was then moved via the Gateway LR II Clonase enzyme kit (Life Technologies, Grand Island, NY, USA) into pKEB31, a derivative of the broad host range plasmid pDD62 (Mudgett et al., 2000) in which the nptII gene is replaced by the tetR gene of pBR322. pKEB31 places the effector gene under the control of the lac promoter, which is constitutive in Xanthomonas.
Isolation and transformation of Xag EB08 and cultivation of Xanthomonas strains
Xanthomonas axonopodis pv. glycines (Xag) strain EB08 was isolated from diseased soybean (Glycine max L. Merr.) leaves that had been stored at −20°C after being collected in Iowa in 2008. Leaf tissue was ground in sterile water, an aliquot of supernatant was cultured on glucose yeast extract (GYE) agar amended with 150 μg ml−1 cyclohexamide, and a yellow, mucoid colony was isolated. The isolate was confirmed as the causal agent of bacterial pustule of soybean by reinoculation to healthy soybean plants. Xag EB08 was transformed by electroporation as previously described (Tsuge et al., 2001) and cultured on GYE agar amended with 10 μg ml−1 tetracycline for plasmid selection. X. oryzae strains were cultured on GYE.
Plant inoculations, virulence assays, and qPCR
Fresh (24–48 h) bacterial cultures were scraped from GYE plates and suspended in 10 mM MgCl2 to an optical density at 600 nm (OD600) = 0.5 as inoculum for virulence assays, and OD600 = 0.9 as inoculum for quantitative real-time reverse-transcriptase PCR (qPCR) assays. For Xoc virulence assays, 6-wk old plants were inoculated by infiltration of leaves with inoculum using a needleless syringe (Schaad et al., 1996), and the length of water-soaked lesions was measured at 10 d. For Xoo, leaf tips of 6-wk-old plants were clipped with scissors dipped in bacterial suspension (Kauffman et al., 1973), and the length of leaf curl was measured at 14 d. For qPCR, bacteria were inoculated with a needleless syringe into the youngest, fully expanded leaf on 4-wk-old plants. Inoculated tissue was flash-frozen after 48 h, and total RNA was extracted with the RNeasy Mini Plant Kit (Qiagen, Hilden, Germany). qPCR was performed on an iCycler Thermo Cycler (Bio-Rad, Hercules, CA, USA) with 100 ng total RNA as template for cDNA synthesis and PCR amplification using the iScript One-Step RT-PCR kit with SYBR Green (Bio-Rad). Gene-specific primers P772 (CCGTCATCCTCATGCACATGCTCACCAC) and P771 (CACGGAGGAGAACTAGAGAGACCAGAGAC) were used for amplification of Xa27 cDNA, and P787 (CCGGTGGATCTTCATGCTTACCTGG) and P788 (CGACGAGTCTTCTGGCGAAACTGC) were used for amplification of Actin-6 (EU215044) cDNA for normalization. A minimum of three independent biological replicates, each with three qPCR technical replicates, were tested for each treatment. The method (Livak & Schmittgen, 2001) was used to quantify expression of Xa27 transcript for each treatment relative to mock-inoculated tissue.
cDNA for 5′-RACE was produced from 1 μg total RNA with the SMARTer RACE cDNA amplification kit (Takara Bio, Inc., Otsu, Shiga, Japan) using the manufacturer’s protocol, from the same samples prepared for qPCR. Xa27 mRNA was amplified with the Universal forward primer (Takara Bio, Inc., Otsu, Shiga, Japan) and P771. For sequencing, PCR products were cloned using the pCR8/GW/TOPO TA entry vector system (Life Technologies, Grand Island, NY, USA).
Analysis of EBE sequence representation in rice promoters
Rice promoters (defined as the sequences from 400 to 50 bases upstream of all translational start sites) were extracted from the MSU Rice Genome Annotation Project database (http://rice.plantbiology.msu.edu/), version 6.1, and concatenated into a single ‘promoterome’. The promoterome was then randomized 5000 times. Next, all six-nucleotide sequences of each EBE were extracted using a sliding window. For each six-nucleotide sequence, the numbers of exact matches in the promoterome and in the 5000 randomized promoteromes were determined. To test if each six-nucleotide sequence occurred at higher- or lower-than-expected frequencies in the actual promoterome, P-values were estimated by dividing by 5000 the number of randomized promoteromes that contained more (or fewer) exact matches. To determine whether the observed numbers of six-nucleotide sequences that occurred at lower- or higher-than-expected frequencies were significant, the EBEs were randomized 100 times and the analysis was repeated for each randomization. P-values were estimated by dividing by 100 the number of EBE randomizations with more six-nucleotide sequences showing higher- or lower-than-expected frequencies than those from the real EBEs.
Stable integration of EBE-amended Xa27 constructs into rice
We amplified Xa27 from rice variety IRBB27 (Gu et al., 2005) and amended its promoter with EBEs from target genes of three TAL effectors important for virulence of the Xoo strain PXO99A and EBEs found in rice for three TAL effectors of the Xoc strain BLS256, whose roles in virulence are as yet uncharacterized (Table S1). These include EBEs from rice genes 8N3/Xa13 (Os08g42350), TFX1 (Os09g29820), and Hen1 (Os07g06970) that correspond to virulence factors PthXo1, PthXo6, and Tal9a, respectively, from PXO99A (Yang et al., 2006; Sugio et al., 2007; Salzberg et al., 2008; Römer et al., 2010; B. Yang, unpublished), and EBEs from rice genes Os03g37840 (a putative potassium transporter), Os06g37080 (a putative l-ascorbate oxidase precursor) and Os05g34600 (a putative NAC (which stands for NAM (no apical meristem), ATAF, CUC (cup-shaped cotyledon)) transcription factor), predicted to be bound by TAL effectors Tal4a, Tal4c and Tal2g, respectively, of BLS256 (Moscou & Bogdanove, 2009; Bogdanove et al., 2011; E. L. Doyle & A. J. Bogdanove, unpublished). The TAL effectors are reciprocally unique to each strain. The EBEs were inserted upstream of the native AvrXa27 binding site. To maximize space for simultaneous binding of multiple effectors from the same pathogen while keeping the length of the overall modification short, sites for Xoo were alternated with those for Xoc and further separated by 6 bp (Fig. 1 and Supporting Information, Fig. S1). The nopaline synthase terminator was placed upstream of the Xa27 promoter to prevent read-through expression of the transgene should it insert downstream of an active endogenous promoter. This construct, containing all six EBEs, was designated as UXO (up-regulated by X. oryzae). Three other constructs were also made (Figs 1, S1). In the first, termed RAN (for randomized EBEs), sequences of all six EBEs were internally randomized, thereby maintaining the nucleotide composition and length of the UXO construct, but destroying the binding sites. In the second, termed XOO, the Xoo EBEs were kept intact but the Xoc EBEs were randomized as in the RAN construct. In the third, called XOC, the inverse was done to retain only the Xoc EBEs. The native AvrXa27 EBE was unchanged in all except the XOC construct, where, in order to generate strictly Xoc-specific resistance, it was converted to the corresponding sequence of the nonfunctional xa27 allele, which is not activated by AvrXa27 (Gu et al., 2005). None of the constructs contains any sequence that would be predicted to bind a PXO99A or BLS256 TAL effector other than those for which the constructs were designed.
All four constructs were stably integrated into the readily transformed, short-season japonica rice cv. Kitaake. The xa27 allele in this variety is nonfunctional and identical in sequence to that (Os06g39810) in the reference japonica genome, Nipponbare. All recovered events were advanced for characterization. Of three lines from independent transformation events with the UXO construct that were advanced to the T2 generation, lines 1 and 3 were found by PCR amplification to contain the UXO construct, and were analyzed further. Similarly, two independent lines for each of the RAN and XOC constructs were confirmed by PCR and characterized. Only a single line with the XOO construct survived to the T2 generation, and this was also characterized. Finally, two events resulting from transformation with the wild type Xa27 gene were advanced to the T2 generation. One, termed HRB, was found to contain the herbicide resistance cassette but not the modified Xa27 gene. In the second, called AZY, neither the herbicide resistance cassette nor the Xa27 gene could be detected at T2. These two lines were retained as controls. Plants from all lines advanced to T2 were morphologically and developmentally indistinguishable from untransformed plants with the exception of the XOC lines, which were slightly reduced in stature. In leaves of the UXO and XOO lines, basal levels of expression of Xa27 were similar to that in IRBB27. In the RAN lines and the one tested XOC line, basal expression was moderately elevated relative to that in IRBB27 (Table S2).
EBE-amended Xa27 is activated by Xoo and Xoc independent of AvrXa27
To examine responsiveness of the UXO construct to the corresponding TAL effectors, we first analyzed the relative fold-change in Xa27 mRNA 48 h after syringe inoculation of T2 leaves with Xoo PXO99A, the avrXa27-deficient mutant derivative ME1 (Gu et al., 2005), Xoc BLS256, and T3SS-deficient mutants of PXO99A and BLS256. qPCR revealed significant (P < 0.05) up-regulation of the transgene in the UXO-1 and UXO-3 lines when inoculated with PXO99A, ME1, or BLS256 relative to mock-inoculated tissue, but not when inoculated with either of the T3SS mutants (Fig. 2). By contrast, the gene in the RAN-1 and RAN-2 lines was not significantly up-regulated by any of the bacterial strains, unexpectedly including PXO99A, in spite of the AvrXa27 EBE present in the RAN construct. Inoculation of PXO99A to the nontransgenic variety IRBB27, which carries the wildtype Xa27 gene, resulted in the expected activation, indicating that the failure of the RAN lines to respond was not the result of a defect in the PXO99A isolate.
EBE-amended Xa27 is specifically activated by each TAL effector
To assess the function of each EBE in the UXO construct, we measured the fold-change in Xa27 mRNA in leaves of T2 plants of the UXO-3 line 48 h after syringe infiltration with transformants of strain EB08 of the soybean pathogen Xag expressing each of the corresponding effectors individually. This strain on its own produces no visible response in rice. Xa27 in the UXO-3 plants was significantly (P < 0.05) up-regulated by all seven TAL effectors, that is (in the order of their EBEs) Tal4c, PthXo1, Tal2g, PthXo6, Tal4a, Tal9a, and AvrXa27, when compared with mock-inoculated tissue (Fig. 3). The Xoo TAL effector AvrXa10 (Hopkins et al., 1992), and the Xoc TAL effector Tal1c (Bogdanove et al., 2011), for which UXO-3 contains no EBEs, as well as Tal1c lacking its central repeat region, caused no significant up-regulation (Fig. 3). Fold induction by the six targeted TAL effectors (Fig. 3) positively correlated with proximity of the corresponding EBE to an apparent TATA box immediately upstream of the native AvrXa27 EBE (Fig. 1; Spearman’s rank correlation coefficient = 0.8857, P < 0.01). The same pattern of induction of Xa27 by the individual TAL effectors was observed in the UXO-1 line (Fig. S2).
All six EBEs drive transcription of Xa27 primarily from a shared start site
Previous studies reported start sites for TAL effector-induced transcription at positions distinct from the start site for basal level expression; these were generally located 42–54 nucleotides downstream of the 3′ end of the EBE (Kay et al., 2007, 2009; Römer et al., 2009a; Römer et al., 2009b; Antony et al., 2010). To determine whether the position of the EBE for each TAL effector activating the UXO construct similarly defines the transcriptional start site (TSS), 5′ rapid amplification of cDNA ends (5′-RACE) was conducted on RNA collected for the UXO-3 gene expression assays described earlier (Fig. 4). Based on the 12–29 transcript sequences obtained per sample, despite the distinct positions of their corresponding EBEs, PthXo1, PthXo6, Tal9a, Tal4a, Tal4c, and Tal2g each initiated transcription primarily at a shared site, identical to that of the majority of transcript sequences obtained from basal expression in the mock and AvrXa10-treated negative controls. The site, 60 bp upstream of the start codon of the Xa27 coding sequence, resides 27 bp downstream of the native, putative TATA box that immediately precedes the AvrXa27 EBE, which positions it only 10 bp downstream of the 3′ end of that EBE. Consistent with the observations to date that TAL effector-driven transcription initiates a minimum of 42 bp from the 3′ end of an EBE, none of the AvrXa27-generated transcripts initiated at that common site. Rather, they initiated further downstream, nearly all of them at a distance of 46 bp from the 3′ end of the AvrXa27 EBE, 24 bp upstream of the Xa27 start codon. Xoo PXO99A- and Xoc BLS256-inoculated plants each exhibited a greater diversity of TSSs than was generated by any of the individually delivered effectors, including some TSSs not observed following individual TAL effector delivery. Conversely, however, some sites that were observed upon delivery of the TAL effectors individually were not detected following inoculation with the pathogen strains. The most common TSS in PXO99A-inoculated plants was the shared site at 60 bp upstream of the Xa27 start codon. This was the second most common for BLS256-inoculated plants, the first being a unique site upstream, at −149 relative to the Xa27 coding sequence. Other apparent TSS locations for PXO99A- and BLS256-inoculated plants ranged as far upstream as −165 relative to the Xa27 translational start, with no discernible correlation to EBE positions.
EBE-amended Xa27 confers resistance to a broader spectrum of Xoo strains and to all tested Xoc strains
To assess the effectiveness of adding EBEs to Xa27 to broaden its functional specificity, we first carried out quantitative virulence assays of ME1 and BLS256 in T2 plants of the UXO, XOO, XOC, and RAN lines, as well as the HRB and AZY controls and IRBB27 (Fig. 5a,b). The extent of bacterial blight symptoms (length of leaf curling) following leaf clip inoculation (Kauffman et al., 1973) with ME1 was significantly (P < 0.01) and markedly reduced specifically in the UXO lines and the XOO line relative to the RAN, XOC, and HRB lines, and IRBB27. The single AZY line, for unknown reasons, showed an intermediate mean leaf curl length. Bacterial leaf streak lesions in the UXO and XOC lines following syringe inoculation (Schaad et al., 1996) with BLS256 were significantly (P < 0.01) and markedly reduced relative to all other lines. The patterns of resistance and susceptibility to ME1 and BLS256 in the EBE-amended and control lines correlate strictly with the patterns of Xa27 activation described earlier (Fig. 2), and demonstrate that the amendment resulted in a specific broadening of the resistance spectrum of Xa27 to include both bacterial blight and bacterial leaf streak.
The distribution of the TAL effectors corresponding to the six EBEs in the UXO construct and of AvrXa27 in field populations of Xoo and Xoc is not yet known. Nonetheless, we sought to determine the effectiveness of the UXO construct against geographically diverse strains of both pathovars. Seven Xoo strains reported to be virulent (1947 from Africa, C4 and ZHE173 from China, K202 from Korea, and 2 from Thailand) or weakly virulent (JW89011 from Korea and PXO71 from the Philippines) on IRBB27 (Gu et al., 2004) were inoculated to UXO-1 and, for comparison, to RAN-1 by clip inoculation (Fig. 5c). C4 and ZHE173 caused significantly (P < 0.05) and markedly reduced leaf curl lengths on UXO-1 relative to RAN-1. JW89011 caused significantly (P < 0.05) but less dramatically reduced leaf curl lengths on UXO-1. The remaining Xoo strains were not significantly less virulent on UXO-1 than on RAN-1. Ten geographically diverse strains of Xoc, each virulent toward wildtype Xa27 plants (data not shown), were inoculated to UXO-1 and RAN-1 by syringe infiltration. The strains differed in their virulence on RAN-1, but each showed significantly (P < 0.01) and markedly reduced relative virulence on UXO-1 (Fig. 5d). Thus, EBE amendment to the Xa27 promoter broadened its functional specificity to include a diversity of Xoo and Xoc strains.
The UXO EBEs contain sequences apparently under selection in rice promoters
To assess whether the EBEs added to the Xa27 promoter in the UXO construct might contain cis regulatory elements, a condition that could prevent TAL effector-mediated activation or cause TAL effector-independent activation under certain environmental conditions or during development, we examined representation of the EBEs in the rice ‘promoterome’, the collection of sequences from 400 to 50 bases upstream of all annotated translational start sites. We chose this sequence space based on the observation that the EBEs in the UXO construct in their natural contexts occur from positions −362 (Tal4a) to −86 (AvrXa27) relative to the translational start site. We reasoned that any portions of EBE sequences corresponding to cis elements in the promoterome would be under selection and therefore occur at nonrandom frequencies. For every possible six-nucleotide fragment of each EBE in the UXO construct, we determined the number of instances of that sequence in the promoterome and in each of 5000 randomized promoteromes. Of the 97 six-nucleotide sequences contained in the EBEs, 40 occurred in the promoterome at lower frequencies than expected based on observed frequencies in the randomized promoteromes, and 54 occurred at higher-than-expected frequencies (P < 0.05, Fig. 6). To distinguish whether the large number of six-nucleotide sequences with nonrandom frequencies was a function of the EBE sequences or instead an artifact of the EBE nucleotide composition, we conducted the same test with six-nucleotide sequences from each of 100 different internal randomizations of the EBEs. Based on the numbers of six-nucleotide sequences derived from these randomizations that occurred at higher- or lower-than-expected frequencies in the promoterome, we determined that the number of six-nucleotide sequences from the actual EBEs that occurred nonrandomly in the promoterome was significantly higher than expected to occur by chance (P < 0.01), meaning that those frequencies are indeed a function of the EBE sequences, and suggesting selection. Taken together, the data indicate a strong likelihood that the six EBEs added to the UXO gene configuration contain or overlap sequences likely to be cis regulatory elements.
A resistance gene that recognizes an effector that otherwise contributes strongly to virulence can be relatively durable owing to the fitness cost to the pathogen of losing or modifying that effector to evade detection (Leach et al., 2001). A striking example is bacterial blight resistance mediated by the rice Xa7 gene, which is triggered by AvrXa7, a major virulence factor in Xoo (Hopkins et al., 1992; Bai et al., 2000; Vera Cruz et al., 2000; Ponciano et al., 2003). However, a mutation that uncouples the virulence and avirulence (resistance-triggering) properties of an effector could result in defeat of the resistance gene. Also, because most resistance genes exhibit pathogen race specificity linked to recognition of a single effector, evolution or introduction of pathogen strains that use alternative effectors for virulence can render a resistance gene ineffective. Against pathogens that rely on TAL effectors for virulence, an executor gene engineered to respond to multiple TAL effectors important for virulence and conserved across a pathogen population could provide durable and broad-spectrum protection.
Here, we established the feasibility of such an approach by adding EBEs to the promoter of the Xa27 gene for resistance to rice bacterial blight and demonstrating consequent broadening of its functional specificity to include Xoo strains virulent toward wildtype Xa27-containing plants as well as strains of the bacterial leaf streak pathogen Xoc, for which no simply inherited resistance genes had been identified in rice. Without altering the coding sequence and by adding < 200 bp of recombinant DNA to the gene promoter, EBE amendment resulted in a single gene with an effective recognition spectrum similar to that achieved by the arduous and time-consuming process of pyramiding multiple resistance genes. Our selection of EBEs was based on available data from one strain each of Xoo and Xoc. Systematic studies to catalog the diversity of TAL effectors in Xanthomonas field populations should enable the rational modification of executor genes to provide broad and durable resistance on a geographically specific basis.
Importantly, each of the targeted TAL effectors individually activated the transgene in the UXO construct. Also, the observed patterns of resistance and susceptibility to Xoo ME1 and Xoc BLS256 across all transgenic lines were as expected based on the corresponding EBE and TAL effector content in each interaction, and they correlated directly with the patterns of Xa27 activation. We used ME1 in addition to PXO99A for the gene expression assays and in place of PXO99A for the virulence assays to distinguish resistance as a result of the EBEs added to Xa27 from resistance mediated by the native AvrXa27 EBE. It should be noted that in addition to its deficiency in avrXa27, ME1 is also disrupted in pthXo6, which neighbors avrXa27; although AvrXa27 makes no measurable contribution to virulence, pthXo6 does (Sugio et al., 2007). The virulence reduction as a result of the disruption of pthXo6 in ME1 is slight, however, and did not mask the resistance conferred by the EBE-amended Xa27 constructs. The disruption of pthXo6 in ME1 did, though, limit the effectors in that strain expected to interact with EBEs in the UXO promoter to PthXo1 and Tal9a. Together, the results strongly suggest that activation of the transgene by the wildtype Xoo and Xoc strains is mediated by at least one, and possibly each, of the corresponding TAL effectors in those strains.
Each of the TAL effectors we chose from PXO99A plays a role in virulence. None of the ones we chose from BLS256 has been characterized with respect to virulence, but each has a well-matched EBE in a putative target gene promoter in rice. The set of EBEs matching these effectors expanded the resistance spectrum of Xa27 to include not only the PXO99A mutant lacking AvrXa27, and BLS256, but two of seven additional, geographically diverse Xoo strains and each of 10 additional, diverse Xoc strains tested. The wide resistance spectrum conferred by these EBEs against the Xoc strains and the narrower spectrum against Xoo strains suggest conservation of one or more of the BLS256 TAL effectors in the Xoc strains, and poor conservation of the PXO99A effectors in that group. This may reflect greater overall TAL effector diversity among Xoo strains, potentially as a result of diversifying selection exerted by the nearly 30 known bacterial blight resistance genes in rice (Niño-Liu et al., 2006). As noted, no simply inherited genes for bacterial leaf streak resistance have been identified in rice (the only identified source of complete resistance is the Rxo1 gene from maize (Zhao et al., 2005)), and this might explain the apparently lesser TAL effector diversity across Xoc strains. To conclude with confidence whether a difference in TAL effector diversity truly exists between Xoo and Xoc, however, a more comprehensive and direct inventory across a broad collection of strains would be necessary.
In light of the absence of major, native genes for resistance to bacterial leaf streak in rice, it is particularly promising that Xa27 was fully effective against Xoc. The only executor gene in rice cloned to date, Xa27 encodes a 113-amino-acid product that has no similarity to functionally characterized proteins, but contains an amino-terminal signal-anchor-like sequence that mediates export to the apoplast and is required for bacterial blight resistance (Wu et al., 2008). Xoo colonizes rice xylem vessels, interacting with cells in the surrounding xylem parenchyma to cause bacterial blight. Xoc multiplies in the mesophyll parenchyma. The mechanism by which Xa27 confers resistance to bacterial blight is unknown. Its activation causes cell death, but it is as yet unclear whether the death is programmed or a result of toxicity of the Xa27 protein. Its effectiveness against bacterial leaf streak, demonstrated here, indicates that the mechanism is general and not restricted to cells in the xylem parenchyma.
As noted earlier, weak constitutive expression of Xa27, though deleterious to the plant, conferred partial resistance to Xoc strain L8 (Tian & Yin, 2009). The complete resistance we observed is likely a result of stronger, localized expression in response to one or more of the targeted TAL effectors delivered by Xoc. Fold induction of Xa27 in the UXO construct by each of the targeted Xoo and Xoc TAL effectors individually correlated positively with proximity of the EBE to a putative TATA box downstream and to the common start site of transcription. Although differences in expression, delivery, or affinity of the TAL effectors may have contributed to this pattern, the strength and significance of the correlation of EBE position to fold induction suggest that there may be a limit to the number of EBEs that can be effectively added to a promoter, as those farthest upstream might drive expression only weakly or not at all. We also observed, unexpectedly, that PXO99A failed to activate Xa27 in the RAN lines, despite its ability to do so in IRBB27. The RAN construct maintains the native AvrXa27 EBE, differing from the UXO construct only in its internal randomization of the added EBEs. Thus, the randomized sequences in some way prevent activation by AvrXa27, or, less likely, in both lines activation is blocked as a result of the positions at which the construct integrated. Although the TAL effector-DNA binding code enables prediction of TAL effector binding sites with relative confidence, these observations highlight the fact that we still know little of the contextual requirements for functional EBEs, with respect to both promoter sequences and chromosomal location and chromatin status. Until more is known, successful placement of EBEs in a promoter may require some trial and error.
Another complexity in EBE amendment is the possibility of interaction among different TAL effectors binding to the same promoter. We sought to minimize this by spacing EBEs 6 bp apart and alternating those for Xoo with those for Xoc. Nonetheless, our 5′ RACE results comparing TSSs detected following individual TAL effector delivery with those detected following inoculation with the corresponding Xoo or Xoc strain provided evidence for such interaction. Some TSSs detected following inoculation with the pathogen strain were not detected on delivery of any of the corresponding individual effectors and, to a lesser extent, vice versa. Thus, although the collection of transcript sequences may not have been saturating, the data suggest that both cooperative interaction to generate novel TSSs as well as interference that blocks initiation from some sites take place.
Previous analyses showed that TAL effector-initiated transcription occurred between 42 and 54 nucleotides downstream of the 3′ end of the EBE, even when the EBE was moved out of context (Kay et al., 2007, 2009; Römer et al., 2007, 2009a, 2009b; Antony et al., 2010). Our results differ. Most transcripts driven by each of the six nonnative TAL effector–EBE interactions on the Xa27 promoter in the UXO construct were initiated at a site shared by all interactions, rather than at sites corresponding to the relative positions of the EBEs. Because this was the same site of transcript initiation observed in mock-inoculated leaves and leaves treated with the nontargeted negative control AvrXa10, it appears that in some configurations, and perhaps depending on the EBE, TAL effector-induced transcript initiation can default to the primary site used for basal expression. The location of the common TSS at 27 bp downstream of a putative TATA box suggests that presence of such an element might influence the location of TAL effector-driven transcript initiation. Some of the TAL effectors reported to dictate the TSS display a TATA box-like sequence within their EBE, which might explain that ability, but not all of them do. Indeed, the AvrXa27 EBE contains no TATA-like sequence, yet drives transcription from a site downstream of the common TSS, perhaps because it is too close to that TSS. In all, it appears that the site of TAL effector-driven transcript initiation is influenced by position of the EBE and by other promoter features in a yet poorly understood way.
We discovered that the EBEs we chose for the UXO construct contain sequences apparently under selection in rice promoters, suggesting coincidence with endogenous regulatory elements. These might include elements for gene activation in response to a particular environmental or developmental cue. Introduction of an EBE with such an element into the promoter of an executor gene could lead to TAL effector-independent cell death on that cue, with potentially disastrous consequences. We examined sequences from only six EBEs, so our findings may not be generalizable. However, TAL effectors might typically be expected to target endogenous regulatory elements in the host genome, by chance because of their localization in promoters, but also because such regulatory elements would likely be relatively immutable, and thereby confer selective advantage on any corresponding TAL effector.
Although the lines included in this study were healthy and fertile under controlled growth conditions, we encountered difficulty retrieving viable, stably transformed lines for some constructs we made. In addition to the UXO construct, in which the EBEs are exactly those found in the rice genome and contain some nucleotides that do not match the RVD at that position in the corresponding TAL effector, we also made a construct containing analogous, perfectly matched EBEs. Plants with this construct were unhealthy, and no lines survived to T2. Also, we were unable to obtain plants transformed with wildtype Xa27. The latter observation, consistent with previous findings (Wu et al., 2008), suggests that the unmodified Xa27 promoter can drive expression of the executor gene and consequent cell death when outside of its native genomic context. Similarly, even with the terminator upstream, transformation events using the XOO construct only yielded one healthy line. Thus, in addition to the possibility that EBEs added to a promoter may contain regulatory elements that could drive TAL effector-independent activation, synthetic sequences, sequences out of context, and position effects might also be problematic.
In sum, though the results presented here establish the effectiveness of EBE amendment to expand the spectrum of pathogen genotypes against which an executor gene functions, multiple alternative constructs may be necessary to obtain effective, stable transformants, and all lines should undergo testing under a variety of growth conditions and field environments before distribution or commercialization. Analysis of the representation of EBE sequences in the host promoterome might be a useful preliminary to selection of EBEs, in order to exclude those that contain sequences apparently under selection. Although beyond the scope of the work we have presented here, it might also be useful to determine whether any sequences that are over- or under-represented are found in promoters of genes of a particular functional class, in order to better predict conditions or processes in which the sequence might act as a regulatory element. Taking advantage of the degeneracy in the TAL effector-DNA binding code to generate nonnative EBE sequences may enable the rational design of a promoter recognized by a desired suite of TAL effectors without the inclusion of endogenous cis elements. Finally, making the promoter modifications in situ using engineered nucleases such as TALENs (Bogdanove & Voytas, 2011) may help to guard against position effects on expression.
The authors thank T. Li, B. Yang, J. Baller, S. Whitham, and S. Kamoun for helpful discussion, E. Braun for infected soybean leaves, B. Yang and C. Schmidt for TAL effector clones, Z. Yin for IRBB27, F. White for ME1, B. Yang for several Xoo strains, C. Song, R. Sonti, C. Vera-Cruz, and V. Verdier for Xoc strains, and M. Moscou, K. Thilges, and M. Peterson for technical assistance. This work was supported by the NSF (award 0820831).