Transcription activator-like type III effector AvrXa27 depends on OsTFIIAγ5 for the activation of Xa27 transcription in rice that triggers disease resistance to Xanthomonas oryzae pv. oryzae




The transcription activator-like (TAL) type III effector AvrXa27 from Xanthomonas oryzae pv. oryzae (Xoo) strain PXO99A activates the transcription of the host resistance gene Xa27, which results in disease resistance to bacterial blight (BB) in rice. In this study, we show that AvrXa27-activated Xa27 transcription requires host general transcription factor OsTFIIAγ5. The V39E substitution in OsTFIIAγ5, encoded by the recessive resistance gene xa5 in rice, greatly attenuates this activation in xa5 and Xa27 double homozygotes on inoculation with Xa27-incompatible strains. The xa5 gene also causes attenuation in the induction of Xa27 by AvrXa27 expressed in rice. The xa5-mediated attenuation of Xa27-mediated resistance to PXO99A is recessive. Intriguingly, xa5-mediated resistance to xa5-incompatible strains is also down-regulated in the xa5 and Xa27 double homozygotes. In addition, AvrXa27 expressed in planta shows weak virulence activity in the xa5 genetic background and causes enhanced susceptibility of the plants to BB inoculation. The results suggest that TAL effectors target host general transcription factors to directly manipulate the host transcriptional machinery for virulence and/or avirulence. The identification of xa5-mediated attenuation of Xa27-mediated resistance to Xoo provides a guideline for breeding resistance to BB when pyramiding xa5 with other resistance genes.


Gram-negative pathogenic bacteria rely on the type III secretion system (TTSS) to secrete multiple effector proteins into host cells to modulate host cell functions for the benefit of the invasion process (He et al., 2004). Xanthomonas oryzae pv. oryzae (Xoo), the causal agent of bacterial blight (BB) disease in rice, harbours multiple genes for type III effectors of the transcription activator-like (TAL) effector family (Gu et al., 2005; Sugio et al., 2007; Yang et al., 2006). TAL effectors are remarkably similar. Each possesses central near-perfect repeats of a 34-amino-acid sequence that varies in repeat number, imperfect heptad leucine zipper (LZ) repeats, three highly conserved C-terminal nuclear localization signals (NLSs) and C-terminal acidic transcription activator-like domains (AADs). Studies have shown that the AADs, NLSs and repeat regions are important for both virulence and avirulence specificities (Szurek et al., 2001; Yang and White, 2004; Yang et al., 2000). The structural features and the high sequence conservation among family members suggest that TAL effectors can manipulate the transcriptional machinery of plant cells by similar mechanisms. Indeed, the activation of several host genes by the TAL effectors from Xoo and Xanthomonas campestris pv. vesicatoria (Xcv) indicates that these TAL effectors mimic the actions of host nuclear components and activate disease-promoting genes (Kay et al., 2007; Sugio et al., 2007; Yang et al., 2006). However, in order to monitor and counteract such disease-promoting strategies, plants have been reported to acquire unique immune-sensing mechanisms by evolving or co-evolving plant adaptations to mimic the activity of the TAL effectors as virulence factors (Gu et al., 2005; Romer et al., 2007). In this case, TAL effectors are known as avirulence proteins which elicit host recognition and resistance responses.

The Xa27 gene in rice recognizes Xoo strains that harbour the TAL effector AvrXa27 (Gu et al., 2005). Unlike other typical resistance (R) genes, nucleotide differences in the promoter sequences between resistant Xa27 and susceptible xa27 alleles determine the recognition specificity. Transcriptional activation of the Xa27 resistance allele occurs only in the presence of AvrXa27 from Xoo (Gu et al., 2005) or AvrXa27 heterologously expressed in planta (Tian and Yin, 2009). NLS and AAD of AvrXa27 are required for effector-specific recognition by Xa27. The Xa27 protein itself provides non-specific and dosage-dependent resistance to Xoo strains and Xanthomonas oryzae pv. oryzicola (Xoc) (Gu et al., 2005; Tian and Yin, 2009; Wu et al., 2008). Several polymorphic DNA elements were identified in the Xa27 and xa27 promoters (Gu et al., 2005); however, none has been experimentally proven to be the specific binding site of AvrXa27 or to be required for the induction of the Xa27 gene.

The recessive xa5 allele on chromosome 5 provides race-specific resistance to Xoo. Map-based cloning and genetic transformation have revealed that xa5 is a V39E substitution variant of OsTFIIAγ5 (Xa5) (Iyer and McCouch, 2004; Jiang et al., 2006). OsTFIIAγ5 is one of the two small subunits of general transcription factor (GTF) IIA (TFIIAγ) in rice. Another TFIIAγ gene in rice is on chromosome 1 (OsTFIIAγ1) (Jiang et al., 2006). Interestingly, the critical mutation for xa5-mediated resistance in xa5 does not affect its function as a GTF for plant development. In addition, there was no differential expression of either the xa5 or Xa5 allele in response to infection (Iyer and McCouch, 2004; Jiang et al., 2006). As the dominant Xa5 allele is susceptible to Xoo, it has been proposed that Xa5 may be a nuclear target of Xanthomonas spp. that interacts with one or more Xoo TAL effectors, thus activating disease-promoting genes; however, the resistant protein xa5 in the xa5 homozygote prevents the activation of the disease-promoting genes because there is no interaction between a Xoo effector and xa5, or such interaction is non-productive (Iyer-Pascuzzi et al., 2008). In this report, we examined the hypothesis by studying the function of OsTFIIAγ5 that is involved in the transcription of host R gene Xa27 activated by TAL effector AvrXa27. Our results prove that AvrXa27 is dependent on the wild-type OsTFIIAγ5 to activate Xa27 transcription.


AvrXa27-activated Xa27 transcription is attenuated in the xa5 mutant background

To investigate whether the xa5 allele of OsTFIIAγ5 in IRBB5 has any effect on Xa27-mediated resistance to BB, we crossed IRBB5 with IRBB27 and selected xa5 and Xa27 double homozygotes from F2 progeny. For clarity, we refer to these double homozygotes as C1 (Table 1). We performed BB inoculation on C1 as well as parental lines with Xoo strain PXO99A, which harbours AvrXa27 (Gu et al., 2005). IRBB27 was completely resistant to PXO99A, whereas IRBB5 was fully susceptible (Fig. 1A; Table 1). C1 was also susceptible to PXO99A, although its BB lesions were shorter than those on IRBB5 plants (Fig. 1A; Table 1). Northern blot analysis showed that Xa27 in IRBB27 was induced by PXO99A, however, its expression in inoculated C1 plants was attenuated significantly (Fig. 1B). As Xa27 protein-mediated resistance to Xoo is non-specific and dosage-dependent (Gu et al., 2005; Tian and Yin, 2009; Wu et al., 2008), low levels of Xa27 transcription in C1 produced insufficient Xa27 proteins to provide complete resistance to PXO99A. Both Xa5 and xa5 were expressed constitutively in rice leaves of non-inoculated and inoculated plants (Fig. 1B), which is consistent with previous reports (Iyer and McCouch, 2004; Jiang et al., 2006). These results demonstrate that a single mutation in OsTFIIAγ5 attenuates the induction of Xa27 by AvrXa27 in C1.

Table 1.  Evaluation of bacterial blight (BB) resistance of single, double and triple lines to Xanthomonas oryzae pv. oryzae (Xoo) strains*.
Line (abbreviation in text)GenotypeLesion length of BB and disease score
  • *

    Six-week-old plants were inoculated with Xoo strains. For each strain, at least 16 leaves from four individual plants were inoculated.

  • The lesion length (LL) is the average of all infected leaves. The standard deviation of the mean is indicated. For score: R, resistant, 0 cm < LL ≤ 3.0 cm; MR, moderately resistant, 3.0 cm < LL ≤ 6.0 cm; MS, moderately susceptible, 6.0 cm < LL ≤ 9.0 cm; S, susceptible, LL > 9.0 cm.

IRBB5xa5/xa5, xa27/xa2728.8 ± 4.5 (S)0.2 ± 0.1 (R)0.2 ± 0.1 (R)
IRBB27Xa5/Xa5, Xa27/Xa270.3 ± 0.1 (R)0.3 ± 0.1 (R)30.9 ± 4.5 (S)
IRBB27 + IRBB5 (C1)xa5/xa5, Xa27/Xa2722.2 ± 6.4 (S)4.9 ± 1.6 (MR)11.4 ± 5.4 (S)
IRBB27 + IRBB5 hetero. (C1h1)Xa5/xa5, Xa27/Xa270.1 ± 0.0 (R)NDND
IRBB27 hetero. + IRBB5 hetero. (C1h2)Xa5/xa5, Xa27/xa270.1 ± 0.0 (R)NDND
IR24Xa5/Xa5, xa27/xa2731.8 ± 2.3 (S)27.9 ± 2.1 (S)31.9 ± 2.6 (S)
24L24Xa5/Xa5, xa27/xa27, avrXa27/avrXa2730.8 ± 2.7 (S)28.0 ± 1.6 (S)30.0 ± 3.5 (S)
IRBB27 + 24L24 (C2)Xa5/Xa5, Xa27/Xa27, avrXa27/avrXa270.2 ± 0.1 (R)0.3 ± 0.1 (R)0.1 ± 0.0 (R)
IRBB27 + 24L24 + IRBB5 (C3)xa5/xa5, Xa27/Xa27, avrXa27/avrXa2722.2 ± 6.8 (S)4.7 ± 0.3 (MR)14.6 ± 3.3 (S)
24L24 + IRBB5 (C4)xa5/xa5, xa27/xa27, avrXa27/avrXa2728.8 ± 2.6 (S)4.8 ± 0.8 (MR)4.4 ± 2.5 (MR)
Figure 1.

Phenotypes of bacterial blight (BB) and gene expression in IRBB5, IRBB27 and C1 after inoculation with Xanthomonas oryzae pv. oryzae (Xoo) strains. (A) Phenotypes of BB on IRBB5 (L1, L5 and L7), IRBB27 (L2, L4 and L8) and C1 (L3, L6 and L9) plants at 14 days post-inoculation with Xoo strains PXO99A (L1–L3), AXO1947 (L4–L6) and PXO86 (L7–L9). (B–D) Expression of Xa27 and OsTFIIAγ5 alleles in IRBB5, IRBB27 and C1 plants at 0 or 1 dpi with PXO99A (B), AXO1947 (C) or PXO86 (D).

To verify this finding and rule out any possibility that other factors from PXO99A may have caused this attenuation, we introduced the avrXa27 gene in transgenic rice line 24L24 into IRBB27 and C1, and produced double homozygotes of Xa27 and avrXa27 genes (C2) and triple homozygotes of xa5, Xa27 and avrXa27 genes (C3) (Table 1). Our previous study has shown that the avrXa27 gene in rice has a similar function and specificity in the induction of Xa27 as in bacteria (Tian and Yin, 2009). Western blot analysis with polyclonal anti-AvrXa27 antibodies further confirmed the presence of AvrXa27 proteins in these rice lines (Fig. 2). Xa27 in C2 was constitutively activated by the expression of AvrXa27 in rice (Fig. 3). The constitutive activation of Xa27 in C2 conferred complete resistance to all Xoo strains tested (Fig. 4A; Table 1). However, the induction of Xa27 by AvrXa27 in C3 was significantly attenuated (Fig. 3). BB inoculation with AXO1947 or PXO86 did not have any significant effect on Xa27 transcription in C3, and both inoculated and non-inoculated plants contained trace levels of Xa27 transcripts (Fig. 4B,C). Disease evaluation indicated that C3 was susceptible to PXO99A and AXO1947, and moderately resistant to PXO86 (Fig. 4A; Table 1).

Figure 2.

Expression of AvrXa27 proteins in rice lines. AvrXa27 expressed in rice was detected by Western blot analysis with anti-AvrXa27 polyclonal antibodies. Rice lines IRBB5, IRBB27 and IR24 are non-transgenic lines, whereas rice lines C2, C3 and C4 carry the avrXa27 gene from transgenic rice line L24 (Tian and Yin, 2009). The molecular mass values of standard protein markers (Fermentas, #SM0671) are shown in kilodaltons (kDa). The position of AvrXa27 is indicated.

Figure 3.

Expression of Xa27, avrXa27 and OsTFIIAγ5 in IRBB27, 24L24, IRBB5, C2 and C3.

Figure 4.

Phenotypes of bacterial blight (BB) and gene expression in IRBB27, 24L24, IRBB5, C2 and C3 after inoculation with Xanthomonas oryzae pv. oryzae (Xoo) strains. (A) Phenotypes of BB on IRBB27 (L1 and L6), 24L24 (L2 and L7), IRBB5 (L3 and L8), C2 (L4 and L9) and C3 (L5 and L10) plants at 14 days post-inoculation (dpi) with Xoo strains AXO1947 (L1–L5) and PXO86 (L6–L10). (B, C) Expression of Xa27 gene, avrXa27 gene and OsTFIIAγ5 alleles in IRBB27, 24L24, IRBB5, C2 and C3 plants at 0 or 1 dpi with Xoo strains AXO1947 (B) and PXO86 (C).

The above results clearly indicate that OsTFIIAγ5 is required for AvrXa27 to activate Xa27 transcription. During the selection of C1 from the IRBB5 × IRBB27 cross, we obtained 13 F2 plants with heterozygous alleles at the xa5 locus. Four of the 13 plants (C1h1, Table 1) were Xa27 homozygotes, whereas the remaining nine plants (C1h2, Table 1) were Xa27 heterozygotes. BB inoculation demonstrated that all of these plants were completely resistant to PXO99A (Table 1). This result suggests that the attenuation of Xa27-mediated resistance to PXO99A resulting from xa5 mutation is recessive, which resembles the inheritance of xa5-mediated race-specific resistance to xa5-incompatible Xoo strains (Iyer-Pascuzzi et al., 2008).

xa5-mediated resistance to xa5-incompatible strains is down-regulated in xa5 and Xa27 double homozygotes

To further study whether xa5-mediated recessive resistance to xa5-incompatible Xoo strains is affected in C1, we inoculated C1, IRBB5 and IRBB27 with AXO1947. AXO1947 was compatible to IRBB27, but incompatible to IRBB5 (Fig. 1A). Intriguingly, C1 was susceptible to AXO1947, although it carries homozygous xa5 genes (Fig. 1A; Table 1). Both IRBB5 and IRBB27 are near-isogenic lines in the IR24 genetic background (Gu et al., 2004; Ogawa et al., 1988). Theoretically, the genetic difference between IRBB5 and C1 is the Xa27 locus. IRBB5 harbours the xa27/xa27 alleles, whereas C1 carries the Xa27/Xa27 alleles. It seems that the down-regulation of xa5-mediated recessive resistance to AXO1947 results from the Xa27 gene in C1. To check whether this down-regulation is limited to AXO1947, we inoculated C1 with Xoo strain PXO86. PXO86 has been reported to be an incompatible strain to both IRBB5 and IRBB27 and may harbour both Avrxa5 and AvrXa27 (Gu et al., 2004, 2005; Hopkins et al., 1992). The inoculation of PXO86 on C1 produced a moderately rather than completely resistant phenotype (Fig. 1A; Table 1). Northern blot analysis showed that only a trace level of Xa27 transcripts was detected in C1 after inoculation with PXO86 (Fig. 1D), suggesting that Xa27-mediated resistance in C1 is abolished or severely impaired when inoculated with PXO86. Therefore, the resistance of C1 to PXO86 is mainly attributed to xa5-mediated resistance. However, because of the down-regulation of xa5-mediated resistance in C1, it collectively conferred partial resistance to PXO86 (Fig. 1A; Table 1).

AvrXa27 shows weak virulence activity in an xa5 genetic background

AXO1947 inoculation on C3 produced slightly longer BB lesions than those on C1 (Table 1). It seems that Xa27 and AvrXa27 in C3 have an additive effect on the down-regulation of xa5-mediated resistance to AXO1947. However, this additive effect is not observed on C3 when inoculated with PXO86 (Table 1). To further verify whether AvrXa27 contributes virulence function to bacterial infection on rice in the xa5 genetic background, we generated rice line C4 by introducing the avrXa27 gene into an IRBB5 background, and inoculated C4 and IRBB5 with AXO1947 (Table 1). As mentioned above, IRBB5 conferred complete resistance to AXO1947; however, C4 provided moderate resistance to the strain (Fig. 5). The results indicate that AvrXa27 expressed in planta shows weak virulence activity in an xa5 genetic background and causes enhanced susceptibility of C4 to BB inoculation.

Figure 5.

Phenotypes of bacterial blight (BB) and gene expression in IRBB5, 24L24 and C4 after inoculation with Xanthomonas oryzae pv. oryzae (Xoo) strain AXO1947. (A) Phenotypes of BB on IRBB5 (L1), 24L24 (L2) and C4 (L3) plants at 14 days post-inoculation (dpi) with Xoo strain AXO1947. (B) Expression of avrXa27 and OsTFIIAγ5 alleles in IRBB5, 24L24 and C4 plants at 0 or 1 dpi with Xoo strain AXO1947.


TFIIA is a GTF in eukaryotes that interacts with the TFIID–promoter complex required for transcription initiation by RNA polymerase II (Hampsey, 1998; Orphanides et al., 1996). TFIIA has been reported to bind to transcriptional activators (Kobayashi et al., 1995; Kraemer et al., 2006). As mentioned in the Introduction, OsTFIIAγ5 may be a nuclear target of TAL effectors for virulence during bacterial infection (Fig. 6A). The V39E substitution in xa5 could be a molecular adaptation evolved in rice to prevent OsTFIIAγ5 from being targeted by TAL effectors, or to make targeting inefficient (Fig. 6B). Similarly, Xa27-mediated resistance to Xoo is another method of plant molecular adaptation to pathogenic bacteria, in which rice has co-evolved the Xa27 promoter to adapt to the activity of AvrXa27 as a transcriptional activator (Fig. 6C). This molecular adaptation requires AvrXa27 to interact with OsTFIIAγ5 for Xa27 transcription (Fig. 6C). It is apparent that the two molecular adaptations conflict when performing their functions in a single plant (Fig. 6C). xa5-mediated resistance relies on xa5, whereas AvrXa27 requires Xa5 or OsTFIIAγ5 to activate Xa27. We have shown in this study that the V39E substitution in xa5 attenuates AvrXa27-activated Xa27 transcription. The identification of this attenuation not only verifies the hypothesis of OsTFIIAγ5 as a target for TAL effectors, but also provides guidelines for breeding resistance to BB when pyramiding xa5 with other R genes, such as Xa27.

Figure 6.

Model for transcription activator-like (TAL) effector-activated host gene transcription. Model shows that TAL effectors from Xanthomonas oryzae pv. oryzae (Xoo) target OsTFIIAγ5 of the pre-initiation complex in the activation of host disease-promoting genes for virulence (A) or Xa27 for resistance (C). The xa5 mutant of OsTFIIAγ5 weakens this targeting and attenuates TAL effector-activated host gene transcription (B, D). The curved bold arrows indicate strong interaction, whereas the curved broken arrows indicate weak interaction. The product of avrxa5 has not been clearly characterized and it could be either a TAL effector involved in xa5 targeting or a gene product affecting this targeting. The pre-initiation complex of eukaryotic gene transcription was drawn based on Yudkovsky et al. (2000). A, TFIIA (OsTFIIAγ5 or Xa5 in this study); B, TFIIB; D, TFIID; E, TFIIE; F, TFIIF; H, TFIIH; Pol II, RNA polymerase II; S, host disease-promoting gene; TATA, TATA box of class II promoter; Xa27, bacterial blight (BB) R gene Xa27; xa5, product of xa5 allele of OsTFIIAγ5 gene.

Both Xa27- and xa5-mediated resistances to their incompatible Xoo strains were down-regulated in C1. However, the mechanism of down-regulation of xa5-mediated resistance to AXO1947 and PXO86 is unknown. Genetic data indicate that the Xa27 gene in C1 may cause this down-regulation. However, the down-regulation is unlikely to be a result of Xa27 proteins or Xa27 transcripts, as Xa27 remains silent in C1 plants during AXO1947 infection. AXO1947 has been shown to be an incompatible strain to IRBB5, but is highly compatible to IRBB27 (Gu et al., 2004, 2005). According to the gene-for-gene concept (Flor, 1971), AXO1947 should harbour Avrxa5. The avrxa5 gene was originally identified in PXO86 (Hopkins et al., 1992) and may encode a TAL effector (Bai et al., 2000). However, little is known about the gene and its product. Understanding the function of Avrxa5 is essential to the future study of the mechanism of xa5-mediated recessive resistance to Xoo, as well as the down-regulation of this pathway in C1.

We could not detect any virulence activity of AvrXa27 in Xoo or heterologously expressed in planta in bacterial infection on susceptible rice varieties (Tian and Yin, 2009). However, in this study, we detected weak virulence activity of AvrXa27 in bacterial infection on rice in the xa5 genetic background. Hence, TAL effectors may have multiple targets in the plant host. For instance, AvrBs3 can target upa genes to promote disease (Kay et al., 2007) or R gene Bs3 to trigger resistance (Romer et al., 2007). AvrXa27 may also be able to interact with xa5 to activate other yet-to-be-identified host genes, which results in enhanced susceptibility to Xoo.


Plant materials, genotyping and growth conditions

Rice (Oryza sativa) line IRBB5 harbouring xa5 in the IR24 background was obtained from the International Rice Research Institute (Los Baños, Philippines) (Ogawa et al., 1988). IRBB27 is a near-isogenic line of Xa27 in the IR24 background (Gu et al., 2004). 24L24 is a transgenic rice line in the IR24 genetic background with the avrXa27 gene introduced from transgenic line L24 (Tian and Yin, 2009). The genotype of the Xa27 gene was determined by a restriction fragment length polymorphism (RFLP) marker derived from the Xa27 promoter at positions −415 to −2315 bp and restriction enzymes SacI and SpeI (Gu et al., 2005). The genotype of the OsTFIIAγ5 gene was determined with a cleaved amplified polymorphic sequence (CAPS) marker using the primer pair xa5_5F/xa5_4R (Anjali et al., 2007). The homozygotes of the avrXa27 gene were screened by selection for hygromycin resistance and genetic analysis. All rice lines and their genotypes are summarized in Table 1. All rice plants, including those inoculated with Xoo strains, were grown in a glasshouse at temperatures ranging from 32 °C during the day to 26 °C at night, a relative humidity averaging 85% and a photoperiod of 12–13 h.

BB inoculation

Xoo strains were grown on PSA medium (10 g/L peptone, 10 g/L sucrose, 1 g/L glutamic acid, 16 g/L bacto-agar, pH 7.0) with appropriate antibiotics for 2 days. The bacterial cells of Xoo were suspended in sterile water at an optical density at 600 nm (OD600) of 0.5. BB inoculation was carried out on two youngest fully expanded leaves on each tiller of 6-week-old rice plants using the leaf-clipping method (Kauffman et al., 1973), or on fully expanded leaves of 4-week-old plants using the infiltration method with a needleless syringe. The disease symptoms were scored at 14 days post-inoculation (dpi) according to the criteria described previously (Gu et al., 2004).

RNA extraction and Northern blot analysis

Northern blot analysis was carried out according to standard procedures. Total RNA was isolated from rice leaves 1 day after infiltration with Xoo using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany), and mRNA was purified with an mRNA Midi Kit (Qiagen). Approximately 5 µg of mRNA were used for each lane in Northern blot analysis for the Xa27 and OsTFIIAγ5 genes, and about 30 µg of total RNA was used for the avrXa27 gene. RNA loading was assessed by hybridizing RNA blots with a probe from the rice Ubiquitin gene 2 (Ubi). The probe for avrXa27 was the 3234-bp BamHI fragment from the gene (Gu et al., 2005). The probe for the Xa27 gene was the full-length cDNA of the gene (Gu et al., 2005). The probe for the OsTFIIAγ5 gene was a reverse transcriptase-polymerase chain reaction (RT-PCR) product of the Xa5 allele that was amplified from IRBB27 using the primer pairs 5′-TAGGCGCACGCATCGCCCATG-3′ and 5′-AGGCAAACCCTGATCTCGCAG-3′. The RT-PCR product was cloned into pGEM-T easy vector (Promega, Madison, Wisconsin, USA) and verified by DNA sequencing. Probes were labelled with 32P-dCTP (GE Healthcare, Little Chalfont, Buckinghamshire, UK).

Western blot analysis

About 20 µg of total proteins from rice leaves were separated on a 10% sodium dodecylsulphate-polyacrylamide gel, followed by blotting onto nitrocellulose membranes. AvrXa27 proteins were detected with anti-AvrXa27 polyclonal antibodies and horseradish peroxidase-coupled secondary antibody (Bio-Rad, Hercules, California, USA). The anti-AvrXa27 polyclonal antibodies were generated in rabbits using a truncated AvrXa27 with deletion of the first 152 amino acids.


We wish to thank N. R. Sackville Hamilton and Darshan Brar for providing the IRBB5 germplasm, Frank F. White and Bing Yang for useful discussions and Li Fang Chan for critical reading of the manuscript. This work was supported by intramural research funds from Temasek Life Sciences Laboratory (Z.Y.). K. Gu is supported by a postdoctoral fellowship from Singapore Millennium Foundation.