Detailed analysis of the DNA recognition motifs of the Xanthomonas type III effectors AvrBs3 and AvrBs3Δrep16


*(fax +49 345 5527277; email


The Gram-negative phytopathogenic bacterium Xanthomonas campestris pv. vesicatoria (Xcv) employs a type III secretion system to translocate effector proteins into plant cells where they modulate host signaling pathways to the pathogen’s benefit. The effector protein AvrBs3 acts as a eukaryotic transcription factor and induces the expression of plant genes termed UPA (up-regulated by AvrBs3). Here, we describe 11 new UPA genes from bell pepper that are induced by AvrBs3 early after infection with Xcv. Sequence comparisons revealed the presence of a conserved AvrBs3-responsive element, the UPA box, in all UPA gene promoters analyzed. Analyses of UPA box mutant derivatives confirmed its importance for gene induction by AvrBs3. We show that DNA binding and gene activation were strictly correlated. DNase I footprint studies demonstrated that the UPA box corresponds to the center of the AvrBs3-protected DNA region. Type III delivery of AvrBs3 and mutant derivatives showed that some UPA genes are induced by the AvrBs3 deletion derivative AvrBs3Δrep16, which lacks four repeats. We show that AvrBs3Δrep16 recognizes a mutated UPA box with two nucleotide exchanges in positions that are not essential for binding and activation by AvrBs3.


Pathogenicity of most Gram-negative phytopathogenic bacteria depends on a conserved type III secretion system that translocates effector proteins into the host cell cytosol where they contribute to bacterial virulence, e.g. by suppression of host defense (Tampakaki et al., 2004; da Cunha et al., 2007). In resistant plants, single effectors, termed avirulence (Avr) proteins, are recognized on the basis of corresponding resistance (R ) genes. Specific recognition usually results in induction of a rapid, localized cell death reaction, the hypersensitive response (HR), that restricts further pathogen ingress (Kiraly et al., 2007).

While many type III effectors show enzymatic activities and thus interfere with signal transduction processes (Abramovitch et al., 2006; Janjusevic et al., 2006; Fu et al., 2007; Rosebrock et al., 2007; Zhang et al., 2007; Kay and Bonas, 2009), members of the large AvrBs3 effector family from Xanthomonas act as transcription factors that specifically induce the expression of plant genes (Marois et al., 2002; Gu et al., 2005; Yang et al., 2006; Kay et al., 2007; Römer et al., 2007; Sugio et al., 2007). Proteins of the AvrBs3 family consist of a central region of direct, nearly identical repeats, usually of 34 amino acids, with most differences at amino acid positions 12 and 13 of each repeat (Gürlebeck et al., 2006). The number and order of the repeats determine the specificity of recognition in resistant plants and the virulence function in susceptible plants (Gürlebeck et al., 2006). The C-terminal regions of AvrBs3 and homologs contain typical features of eukaryotic transcription factors, i.e. nuclear localization signals (NLSs) and an acidic activation domain (AAD), which are required for most avirulence activities and all virulence functions of AvrBs3 family members analyzed (Gürlebeck et al., 2006). AvrBs3 homologs are present only in Xanthomonas spp. and Ralstonia solanacearum (Gürlebeck et al., 2006), and comprise major virulence determinants. Examples are PthA from X. axonopodis pv. citri, which is required for citrus canker symptoms (Swarup et al., 1991), Avrb6 from X. campestris pv. malvacearum, which increases symptoms on cotton and bacterial release to the leaf surface (Yang et al., 1996), and AvrXa7 from X. oryzae pv. oryzae (Xoo) that increases lesion length and bacterial growth in rice (Bai et al., 2000; Yang and White, 2004).

AvrBs3 from Xanthomonas campestris pv. vesicatoria (Xcv), the founding member of this effector family, was isolated based on its avirulence activity in pepper plants (Capsicum annuum) containing the resistance gene Bs3 (Bonas et al., 1989). AvrBs3 causes hypertrophy, i.e. an enlargement of mesophyll cells, in susceptible pepper and other solanaceous plants (Marois et al., 2002; Kay et al., 2007). The hypertrophy probably supports bacterial release from infected plant tissue at later infection stages (Marois et al., 2002). Indeed, field studies revealed a positive influence of AvrBs3 on bacterial spread (Wichmann and Bergelson, 2004). Previously, we described that AvrBs3 acts as a transcription factor that induces UPA genes (up-regulated by AvrBs3), resulting in hypertrophy in susceptible plants and an HR in resistant plants (Marois et al., 2002; Kay et al., 2007; Römer et al., 2007). Interestingly, the AvrBs3 deletion derivative AvrBs3Δrep16, which lacks repeats 11–14, shows an alteration of recognition specificity, and induces an HR in Bs3-E (formerly bs3) pepper plants (Herbers et al., 1992; Römer et al., 2007). The NLSs and AAD in AvrBs3 are absolutely required for elicitation of both hypertrophy and the HR (Marois et al., 2002). The NLSs mediate the interaction with importin α and import of the effector into the nucleus (Van den Ackerveken et al., 1996; Szurek et al., 2001, 2002). AvrBs3-specific induction of plant gene expression depends on the AAD (Marois et al., 2002; Kay et al., 2007; Römer et al., 2007). Recent studies have demonstrated that development of hypertrophy and elicitation of the HR require induction of single direct target genes of AvrBs3, UPA20 and Bs3, respectively (Kay et al., 2007; Römer et al., 2007). Chromatin immunoprecipitation and electromobility shift assays (EMSA) revealed a direct interaction of AvrBs3 with the promoters of UPA20 and Bs3 (Kay et al., 2007; Römer et al., 2007). Interestingly, DNA binding of the effector is mediated by the central repeat region, which represents a novel DNA binding motif (Kay et al., 2007). Sequence comparisons of the UPA20, UPA10 and Bs3 promoters identified a conserved motif with the consensus nucleotide sequence TATATAAACCN2–3CC that was termed UPA box (Kay et al., 2007; Römer et al., 2007). Mutagenesis of the cytosine nucleotides revealed that the UPA box is required for AvrBs3 inducibility (Kay et al., 2007). However, more detailed analysis is necessary to define essential and non-essential bases in the UPA box.

Here, we describe expression analysis of 11 new UPA genes from pepper that were identified by suppression subtractive hybridization and are putative direct targets of AvrBs3. Sequence comparisons of promoter regions together with mutational analysis of the UPA20 promoter resulted in a refinement of the UPA box sequence. In addition, our analyses placed the UPA box in the center of the DNA region covered by AvrBs3, and identified nucleotides that are important for gene induction by the AvrBs3 deletion derivative AvrBs3Δrep16.


Identification of 11 new AvrBs3-induced genes

In a recent screening, cDNA fragments corresponding to AvrBs3-induced genes were isolated by suppression subtractive hybridization from susceptible pepper plants of cultivar ECW infected with Xcv strain 85-10 expressing avrBs3 or Xcv carrying the empty vector (Kay et al., 2007). In addition to UPA20, we isolated 19 additional pepper cDNA fragments that correspond to 11 putative direct targets of AvrBs3, i.e. their induction was independent of plant protein biosynthesis (UPA1425; Table 1). Predicted gene products include proteins with homology to enzymes (UPA14 and UPA15), a transcription factor of the growth regulating factor (GRF) family (UPA17), an RNA-binding protein of the polyC-binding protein (PCBP) family (UPA21), and a nodulin of the MtN3 family (UPA16).

Table 1.   New pepper genes up-regulated by AvrBs3 (UPA) independently of plant protein biosynthesis
GenePredicted gene product
Size (amino acids); mass (kDa)Homology or predicted functionaHomolog (plant)b; identity/coverage (amino acids)
  1. aBased on predicted or known functions of similar proteins identified by BLASTP.

  2. bMost homologous protein sequence obtained with BLASTP against the National Center for Biotechnology Information (NCBI) non-redundant database.

UPA14376; 40.8Carbohydrate kinase (PfkB family)CAO21458 (Vitis vinifera); 238/379 (62%)
UPA15528; 60.6Mannan synthaseACF33171 (Coffea canephora); 439/521 (84%)
UPA16301; 33.5Nodulin (MtN3 family)CAN71371 (V. vinifera); 153/259 (59%)
UPA17596; 64.2Transcription factor (GRF family)CAO61507 (V. vinifera); 377/634 (59%)
UPA18341; 37.5Hypothetical proteinABK92946 (Populus trichocarpa); 166/328 (50%)
UPA19276; 31.0Hypothetical proteinNP_198805 (Arabidopsis thaliana); 156/272 (57%)
UPA21433; 46.7RNA-binding protein (PCBP family)CAO69585 (V. vinifera); 235/372 (63%)
UPA22140; 15.3Light-induced protein (Lir1 family)CAO68684 (V. vinifera); 89/141 (63%)
UPA23287; 32.1Hypothetical proteinCAN80354 (V. vinifera); 161/322 (50%)
UPA24113; 12.7Hypothetical proteinCAO45116 (V. vinifera); 49/86 (56%)
UPA25136; 15.7Hypothetical proteinCAN62371 (V. vinifera); 71/130 (54%)

Kinetics and specificity of UPA gene induction

To analyze the kinetics of UPA1425 induction of time-course experiments in pepper cultivar ECW were performed. We also included UPA12 and UPA13, which were isolated in a first screen for AvrBs3-inducible pepper genes and encode putative transcription factors (Marois et al., 2002). As shown in Figure 1(a), all UPA genes are induced by AvrBs3 even in the presence of cycloheximide, and are therefore putative direct targets. Most genes were induced as early as 4–6 h post-infection (hpi) with Xcv expressing avrBs3 (Figure 1a). Generally, the AvrBs3-dependent induction of UPA genes appeared to be stronger in the presence of cycloheximide (Figure 1a,b). It has been reported that cycloheximide treatment increased the AvrBs3-mediated accumulation of UPA10 and UPA11 transcripts, implying that a negative feedback loop involving protein synthesis normally limits accumulation of these transcripts (Marois et al., 2002).

Figure 1.

 Time course of UPA gene induction by AvrBs3.
RT-PCR analysis of susceptible pepper plants of cultivar ECW inoculated with Xcv strain 85-10 expressing avrBs3 and containing empty vector respectively.
(a) Bacterial inoculum containing 50 μm cycloheximide.
(b) Bacteria inoculated without cycloheximide. Plant material was harvested at various time points (hours post-infiltration, hpi). EF1α was used as a control for equal cDNA amounts. The experiments were repeated once (a) and twice (b), respectively, with similar results.

The strongly induced genes UPA12, UPA14, UPA19, UPA21 and UPA25 were chosen for further analyses. To analyze the requirement for the AvrBs3 repeat region and the eukaryotic motifs in the C-terminus, we performed RT-PCR analyses of pepper plants infected with Xcv expressing avrBs3 and mutant derivatives thereof. The latter included AvrBs3Δrep16, which carries a deletion of four repeats (11–14) (Herbers et al., 1992). As shown in Figure 2, AvrBs3Δrep16 failed to induce UPA21 and UPA25, but UPA12, UPA14 and UPA19 were well induced. UPA19 was induced to a higher level by AvrBs3Δrep16 than by AvrBs3. In addition, the RT-PCR analyses revealed that the NLSs in the C-terminal region of AvrBs3 are absolutely required for induction of all UPA genes tested, demonstrating that the effector protein must enter the plant cell nucleus (Figure 2). Unexpectedly, the AAD is not always essential for gene induction. For instance, UPA14, UPA21 and UPA25 were also induced weakly by the AAD deletion derivative of AvrBs3 (Figure 2).

Figure 2.

 Specificity of UPA gene induction.
The importance of the NLSs, AAD and repeat region of AvrBs3 was tested by RT-PCR analysis. Leaves of pepper plants of cultivar ECW were inoculated with Xcv strain 85-10 containing an empty vector or expressing avrBs3, NLS and AAD mutants or avrBs3Δrep16. The AvrBs3 derivative AvrBs3Δrep16 lacks four repeats (numbers 11–14) (Herbers et al., 1992). Plant material was harvested at 8 hpi. EF1α was used as a control for equal cDNA amounts. The experiment was repeated twice with similar results.

The UPA box is a common motif in UPA gene promoters

Recently, a binding motif, the UPA box (consensus TATATAAACCN2–3CC), has been identified in the promoters of three AvrBs3-inducible genes (Kay et al., 2007; Römer et al., 2007). In these cases, the UPA box was located within 80 bp upstream of the transcriptional start sites that were used in the presence of AvrBs3 (Kay et al., 2007; Römer et al., 2007). 5′ RACE of UPA10 (Figure S1a) and UPA20 (Kay et al., 2007) revealed that basal expression of both UPA genes starts upstream of the transcription initiation site used in the presence of AvrBs3.

To determine whether the UPA box is present in other UPA gene promoters, we identified the transcription start sites and promoter regions of the weakly AvrBs3-induced gene UPA23 and the strongly induced genes UPA12, UPA14, UPA19, UPA21 and UPA25 in the presence of AvrBs3. DNA sequences were compared with the known promoter regions of UPA10, UPA20 (Kay et al., 2007) and Bs3 (Römer et al., 2007). As shown in Figure 3, the UPA box is present in all UPA gene promoters within 100 bp upstream of the transcription start site. Curiously, the UPA12 promoter contains two UPA box motifs (Figure S2a). Promoter analyses using various mutants and the GUS reporter showed that the proximal UPA box starting at position −46 in the UPA12 promoter does not significantly contribute to AvrBs3 inducibility, whereas the distal UPA box of the UPA12 promoter is essential (Figure S2b). Careful comparison of all UPA box sequences revealed a pyrimidine-rich sequence downstream of the second stretch of cytosine nucleotides that was not discovered previously (Figure 3). The refined core UPA box has the consensus sequence TATATAAACCTN2CCCTCT.

Figure 3.

 Sequence comparison of UPA gene promoters.
UPA box-containing promoter regions of UPA10 (Kay et al., 2007), UPA12 (Marois et al., 2002), UPA20 (Kay et al., 2007), Bs3 (Römer et al., 2007) and five new UPA genes were aligned by hand using BioEdit (Hall, 1999). Nucleotides that are present in at least five of the nine sequences are shaded in gray. The UPA box is indicated and the nucleotides defining it are numbered above. Numbers on the left refer to the transcriptional start site (+1) used in the presence of AvrBs3. The corresponding sequence logo was created using WebLogo (Crooks et al., 2004).

Mutational analysis of the UPA20 UPA box

In a first attempt to assess the importance of the UPA box for AvrBs3 responsiveness, five C→G exchanges were introduced into the UPA20 promoter, affecting both inducibility by AvrBs3 in planta and DNA binding of AvrBs3 in vitro (Kay et al., 2007). To analyze the UPA box in more detail, additional mutations were introduced into the UPA20 promoter (Table 2). GUS assays revealed that each patch of cytosine nucleotides, i.e. at UPA box positions 9 and 10 and 14–16, respectively, is required for induction by AvrBs3 (mutants ubm1-1 and ubm1-2) (Figure 4a). The TATA box-like sequence (positions 1–8) is also important, because mutation to thymine or adenine nucleotides (ubm2 and ubm3) abolished AvrBs3 inducibility (Figure 4a). Even a single A→T substitution (ubm2-1) reduced promoter activity, but did not abolish inducibility by AvrBs3 (Figure 4a). This is in accordance with the observation that single base-pair exchanges are often present in the TATATAAA part of the naturally occurring and analyzed UPA boxes (Figure 3). Disruption of the AT-rich region by a single T→G exchange (ubm4) reduced both basal and AvrBs3-induced promoter activity. This, together with the fact that mutants ubm2 and ubm3 have strongly reduced basal promoter activity, suggests that the AT-rich sequence in the UPA box serves as a TATA box for basal expression of UPA20 in the absence of AvrBs3. This is corroborated by the fact that, in the promoters of UPA20 and UPA10, this sequence is located 29 bp upstream of the respective transcription start site under non-inducing conditions (Kay et al., 2007) (Figure S1a), which fits well with the position of functional TATA boxes in plant promoters (Shahmuradov et al., 2003). In contrast to the mutations described above, substitutions of the last three nucleotides in the UPA box, i.e. from TTT to AAA or TCT (ubm5 and ubm6), had no strong effect on inducibility (Table 2 and Figure 4b).

Table 2. UPA20 promoter and UPA box mutant derivatives
UPA20 promoter derivative Sequencea
  1. a The UPA box nucleotide sequence is in bold and italic. Mutations generated in this study (except for ubm1) are shaded in grey.

  2. *(Kay et al., 2007)

Figure 4.

 Analysis of UPA box mutations in the UPA20 promoter.
(a, b) GUS activity as a measure of the AvrBs3 inducibility of various UPA20 promoter derivatives in N. benthamiana. Mutations (see Table 2) were introduced into an 838 bp UPA20 promoter fragment including the 5′ UTR (Kay et al., 2007) and cloned in front of a promoterless egfp::uidA cassette. Mutant ubm1 described previously (Kay et al., 2007) served as reference. GUS activity was determined 3 days after Agrobacterium-mediated co-delivery of the reporter constructs with silencing inhibitor p19 and empty T-DNA or avrBs3.
(c) Electromobility shift assay (EMSA) of biotin-labeled wild-type and mutant UPA box DNA fragments (36 bp) incubated with increasing amounts of GST::AvrBs3. Arrow, bound DNA; asterisk, free DNA. The uppermost signals correspond to the slots. All experiments were repeated twice with similar results.

Electromobility shift assays (EMSAs) demonstrated that a decrease in promoter activation correlated with a strong reduction in AvrBs3 binding to the respective UPA box mutants (Figure 4c). In addition, yeast one-hybrid (Y1H) studies confirmed that AvrBs3 strongly binds the UPA20 wild-type UPA box, but not ubm1 or ubm2 (Figure S3). This demonstrates the specific binding of AvrBs3 to the UPA box in an in vivo assay.

The strong effects of mutations in the C patches of the UPA box suggest that these nucleotides are crucial for recognition by AvrBs3. Alternatively, introduction of guanine nucleotides could disturb the functionality of the UPA box, which is usually embedded in a nucleotide sequence in which guanine nucleotides are under-represented (Figure 3). To test the latter possibility, individual cytosine nucleotides at UPA box positions 8, 9, 14, 15 and 16 (Figure 3) were permutated (mutants ubm1-3ubm1-17) (Table 2). The UPA20 promoter derivatives were cloned in front of a promoterless uidA gene (GUS) in binary vector pGWB3. This vector backbone shows weaker basal GUS activity compared to pBGWFS7 used previously, without changing the specificity of AvrBs3-mediated induction (Figure S4). pGWB3 is therefore more suitable for analyzing subtle activation differences. GUS assays revealed that, in three of five cases, a C→G exchange most strongly affected AvrBs3 inducibility (Figure 5). However, the mutant with a C→G exchange at position 9 of the UPA box (ubm1-5) retained more than 20% and the C→G mutant at position 14 (ubm1-11) more than 40% of wild-type inducibility. Therefore, the presence of guanine nucleotides in the UPA box sense strand per se does not completely abolish activation by AvrBs3. Interestingly, in most cases, a C→A exchange showed either a weak or no effect on promoter activation, whereas C→T exchanges affected inducibility more strongly (Figures 5 and S5). This finding is corroborated by the fact that both types of substitutions are present in naturally occurring UPA boxes, especially at UPA box position 9, where A instead of C is present in three of nine cases (Figure 3).

Figure 5.

 The effect of single base-pair exchanges in the C patches of the UPA box on AvrBs3-mediated inducibility.
The sequence of the UPA20 UPA box is shown, with the position of the cytosine nucleotides analyzed indicated at the top. Individual cytosine nucleotides were replaced by the nucleotides indicated on the left, and introduced into the 838 bp UPA20 promoter fragment (Kay et al., 2007) in front of a promoterless uidA reporter gene in pGWB3. The numbers refer to percentage GUS activity of the respective mutant in the presence of AvrBs3 compared to the wild-type UPA20 promoter (100%). Darker shading indicates stronger effects of mutations on AvrBs3-mediated inducibility. The data represent combined results of three to six independent experiments. A representative experiment is shown in Figure S5. Dashes indicate wild-type nucleotides as shown at the top.

The UPA box is sufficient for AvrBs3-specific activation

To determine whether the UPA box is not only necessary but also sufficient for activation by AvrBs3, the box was introduced into the well-known Bs4 resistance gene promoter from tomato, which has a low basal activity (Schornack et al., 2005). Two constructs were derived from a 302 bp Bs4 promoter fragment that includes the 5′ UTR, cloned into pGWB3 and tested for AvrBs3 responsiveness (Figure 6a). In the first construct, the UPA box was inserted at a site comparable to the position in the UPA20 promoter (-55 with respect to the Bs4 transcription start site; construct Bs4-U20b). In the second construct, Bs4-TATA-U20b, the predicted TATA box of the Bs4 promoter was converted into the UPA20 UPA box (Figure 6a). Mutant ubm1-1 (Bs4-ubm1-1, Bs4-TATA-ubm1-1) served as a control (Figure 6a). GUS assays revealed that insertion of the UPA box at position -55 rendered the Bs4 promoter strongly AvrBs3-inducible, in contrast to UPA box mutant ubm1-1 (Figure 6b). Conversion of the Bs4 TATA box into the UPA20 UPA box also resulted in AvrBs3 responsiveness, but the activity was much lower compared to construct Bs4-U20b (Figure 6b). One possible explanation is that lack of an additional TATA box downstream of the UPA box in Bs4-TATA-U20b might reduce promoter activity. However, mutation of the Bs4 promoter TATA box in construct Bs4-U20b by two nucleotide exchanges (TAGGTAA; Bs4-U20b-mutTATA) did not influence promoter responsiveness to AvrBs3 (Figure 6b). Taken together, our results demonstrate that the UPA box is not only necessary but also sufficient for activation of transcription by AvrBs3 in the absence of a TATA box. This is consistent with the fact that several UPA genes lack an obvious TATA box motif between the UPA box and the transcription start site.

Figure 6.

 The UPA box is sufficient for AvrBs3 inducibility.
(a) Sequence of the Bs4 promoter and 5′ UTR. The UPA20 UPA box and mutant ubm1-1 were inserted at position -55 relative to the Bs4 transcription start site (Schornack et al., 2005) to yield constructs Bs4-U20b and Bs4-ubm1-1, respectively. The Bs4 TATA box was mutated to yield Bs4-TATA-U20b and Bs4-TATA-ubm1-1 (shaded in gray).
(b) GUS activities as a measure of AvrBs3 inducibility of various Bs4 promoter derivatives in N. benthamiana. The mutations shown in (a) were introduced into the 302 bp Bs4 promoter fragment including the 5′ UTR (Schornack et al., 2005), and cloned in front of a promoterless uidA gene. GUS activity was determined 3 days after Agrobacterium-mediated co-delivery of the reporter constructs, silencing inhibitor p19 and empty T-DNA or avrBs3. The experiment was repeated twice with similar results.

5′ RACE revealed that the transcription initiation site of the Bs4 promoter derivatives depends on the position of the UPA box introduced (Figure S6). The start site for transcription from construct Bs4-TATA-U20b was 35 bp downstream of that for construct Bs4-U20b (Figure S6a,b0), i.e. there are similar distances between the UPA box and the respective transcription start sites for both constructs (46 bp versus 44 bp). The presence of an additional TATA box (in construct Bs4-U20b compared to Bs4-U20b-mutTATA) had no influence on start site selection (Figure S6b,c).

The AvrBs3 protein covers the UPA box and flanking sequences

To identify the DNA region that is bound by the AvrBs3 protein, DNase I footprint analyses were performed. For this, a fluorescently labeled 299 bp PCR product containing the UPA20 UPA box was incubated with various amounts of AvrBs3 protein and subjected to partial DNase I digestion. The DNA fragments were analyzed on a capillary sequencer (Zianni et al., 2006). BSA and the AvrBs3 derivative AvrBs3Δrep16, which fails to induce UPA20 (Kay et al., 2007) or to bind the UPA box in vitro (see below), served as negative controls. DNase I footprinting identified 36 nucleotides of the sense strand and 33 nucleotides of the antisense strand that were protected by AvrBs3. The core region, i.e. the region that was protected by the effector protein in all independent experiments and at lowest concentration of AvrBs3 protein, consisted of 30 and 28 nucleotides on the sense and antisense strands, respectively. These data indicate that the UPA box is located in the center of the DNA region bound by AvrBs3 (Figures 7, S7, S8 and S9).

Figure 7.

UPA20 promoter sequence protected by AvrBs3.
(a) DNase I footprinting revealed that 36 nucleotides of the sense strand and 33 nucleotides of the antisense strand of the UPA20 promoter are protected by AvrBs3. The DNA region shaded in gray represents the ‘core footprint’ which was protected by AvrBs3 in every experiment, even with low protein amounts (equal molarity of DNA and protein dimers). The UPA box is shown in bold. Numbers refer to the transcription start site used in the presence of AvrBs3. Data are combined results of four independent experiments.
(b, c) A representative DNase I footprint experiment. The sense (b) and antisense (c) strand sequences are given, with the UPA box underlined. The numbers above refer to nucleotide positions relative to the transcription start site (+1) in the presence of AvrBs3. Electropherograms show the combined protection patterns after digestion with DNase I following incubation with a 5× molar excess of His6::AvrBs3, His6::AvrBs3Δrep16 or BSA (calculated for protein dimers). The y axis shows the relative fluorescence intensity corresponding to (b) the 6-FAM-labeled sense strand and (c) the HEX-labeled antisense strand, on an arbitrary scale. The traces for the reactions with His6::AvrBs3 (green) and the negative controls His6::AvrBs3Δrep16 (red) and BSA (black) are superimposed. A reduction of peak height in the presence of AvrBs3 in comparison to the negative controls corresponds to protection. The protected region is indicated by black vertical lines. The data for all protein concentrations are shown in Figures S7 and S8.

AvrBs3Δrep16 recognizes a modified UPA box

Recently, it was shown that the AvrBs3 derivative AvrBs3Δrep16, but not wild-type AvrBs3, strongly induced the Bs3-E gene (Römer et al., 2007). AvrBs3Δrep16 targets a sequence in the Bs3-E promoter that contains a UPA box disrupted by a 13 bp insertion (Römer et al., 2007), and which therefore lacks the distal C patch that we found to be essential for recognition by AvrBs3 (Figure 4a). In contrast to Bs3-E, expression of UPA12, UPA14 and UPA19 was induced by both AvrBs3 wild-type and AvrBs3Δrep16 (Figure 2). To find out what determines AvrBs3Δrep16 recognition specificity, we inspected the UPA boxes in these promoters for polymorphisms compared to UPA genes that are not induced by AvrBs3Δrep16. As shown in Figure 8(a), all four AvrBs3Δrep16-inducible promoters contain a CT stretch directly downstream of UPA box position 9, in which at least two single thymine residues are surrounded by cytosines. This is in contrast to the promoters of UPA10, UPA20, UPA21, UPA25 and Bs3 (Figure 3), which are not induced by AvrBs3Δrep16 (Figure 2) (Marois et al., 2002; Kay et al., 2007). To analyze the importance of the CT stretch for AvrBs3Δrep16 inducibility, a GA→TC exchange was introduced into the UPA20 UPA box (ubm-r16; Table 2). This creates a CT stretch 3′ of the TATATAAA sequence without affecting base pairs that are crucial for recognition by AvrBs3. GUS assays showed that mutant ubm-r16 was clearly inducible by both AvrBs3 and AvrBs3Δrep16 (Figure 8b). The activation in planta correlated with strong binding of the proteins to the modified UPA box sequence both in vitro (EMSA; Figure 8c and Figure S10) and in vivo (Y1H; Figure S3). In conclusion, two nucleotides at UPA box position 12 and 13 that are dispensable for DNA binding and activation by the AvrBs3 wild-type protein are important for AvrBs3Δrep16 recognition specificity.

Figure 8.

 Analysis of the AvrBs3Δrep16 DNA recognition sequence.
(a) Promoter sequences of genes inducible by AvrBs3Δrep16. The UPA box is underlined and nucleotide positions are marked above. The UPA box in the Bs3-E promoter is truncated. CT stretches in which at least two thymines are surrounded by cytosines are shaded in gray.
(b) GUS assay to compare inducibility of the UPA20 promoter and mutant ubm-r16 by AvrBs3 and AvrBs3Δrep16, respectively. The ubm-r16 mutation (Table 2) was introduced into an 838 bp UPA20 promoter fragment including the 5′ UTR (Kay et al., 2007), and cloned in front of a uidA reporter gene in pGWB3. GUS activity was determined in N. benthamiana 3 days after Agrobacterium-mediated co-delivery of the reporter construct with silencing inhibitor p19 and empty T-DNA, avrBs3 or avrBs3Δrep16.
(c) EMSA with increasing molar amounts of GST::AvrBs3 or GST::AvrBs3Δrep16, incubated with biotin-labeled 36 bp UPA20 promoter fragments containing the UPA box (wt) or ubm-r16. See (e) for details.
(d) GUS assay to compare AvrBs3Δrep16 inducibility of the UPA20 promoter mutant ubm-r16 shown in (a) and derivatives carrying additional mutations (Table 2). GUS activity was determined in N. benthamiana 3 days after Agrobacterium-mediated co-delivery of the reporter constructs with silencing inhibitor p19 and empty T-DNA or avrBs3Δrep16.
(e) EMSA of the biotin-labeled 36 bp ubm-r16 fragment incubated with GST::AvrBs3Δrep16. Unlabeled DNA fragments containing additional mutations (Table 2) were used for competition. Arrow, bound DNA; asterisk, free DNA. The uppermost signals correspond to the slots. All experiments were repeated twice with similar results.

To check reciprocally whether nucleotides that are essential for AvrBs3-mediated inducibility are dispensable for promoter recognition by AvrBs3Δrep16, the base-pair exchanges in ubm-r16 were combined with mutations ubm1-1, ubm1-2, ubm2 and ubm3 (Table 2). GUS assays revealed that additional mutations in the first C patch (ubm-r16-1) or the TATA box-like sequence (ubm-r16-3 and ubm-r16-4) of the UPA box abolished reporter gene activation by AvrBs3Δrep16 (Figure 8d). In contrast, mutation of the second C patch (ubm-r16-2) strongly reduced, but did not completely abolish, AvrBs3Δrep16-mediated promoter activation. The in planta data correlated well with the in vitro binding affinity of AvrBs3Δrep16 for the mutant promoter fragments (EMSA; Figure 8e). Taken together, AvrBs3 and its derivative AvrBs3Δrep16, which differs from AvrBs3 by the deletion of only four repeats, have similar but distinct DNA recognition specificities.


Presumed roles of the new UPA genes

In this study, we analyzed 11 new UPA genes that are activated by AvrBs3 independently of plant protein synthesis and are therefore putative direct target genes of the type III effector. Some of these UPA genes might enhance the hypertrophy induced by AvrBs3 via activation of UPA20, a key regulator of cell enlargement (Kay et al., 2007). For example, UPA17 encodes a homolog of the GRF transcription factors that regulate cell enlargement in Arabidopsis (Kim et al., 2003). The putative glycosyl transferase UPA15, which is most similar to the mannan synthase ManS2 from coffee (Pre et al., 2008), could contribute to cell enlargement by synthesis of cell-wall polymers (Scheible and Pauly, 2004). Interestingly, CcManS2 is most strongly expressed in expanding coffee leaves (Pre et al., 2008). Future analyses will reveal to what extent these UPA genes contribute to AvrBs3-induced hypertrophy.

Some UPA genes may have hypertrophy-unrelated functions that contribute to the positive effect of AvrBs3 on dissemination of Xanthomonas under field conditions (Wichmann and Bergelson, 2004). One candidate is UPA16, a member of the MtN3 family from plants and animals. MtN3 is a Rhizobium-induced nodulin from Medicago truncatula (Gamas et al., 1996). Interestingly, an MtN3 homolog from rice, Os8N3, is a host disease-susceptibility gene induced by PthXo1. The latter is an AvrBs3-like effector from Xoo that is required for in planta growth of strains that express PthXo1 as a major virulence factor (Yang et al., 2006). However, the biochemical function of Os8N3 and the exact mode of gene activation by PthXo1 have not been described. Finally, the possibility should be taken into consideration that some UPA genes are ‘collateral targets’ of AvrBs3 that are not related to the virulence function of the effector.

The UPA box specifies AvrBs3 inducibility

Sequence comparison of known and five new UPA gene promoters led to refinement of the UPA box sequence to TATATAAACCTN2CCCTCT (Figure 3). Mutant analyses of the UPA20 promoter confirmed that the UPA box is required for DNA binding and gene activation by AvrBs3. For AvrXa7 from Xoo, a general affinity for AT-rich DNA in vitro has been reported (Yang et al., 2000); however, the ability to induce transcription from these sequences has not been analyzed. As many of the mutations introduced here do not change the AT content of the UPA20 promoter but nevertheless strongly affect AvrBs3 binding and activation, our data demonstrate that a specific DNA sequence rather than nucleotide composition is crucial for AvrBs3-mediated activation of target promoters. Even single nucleotide exchanges can drastically reduce AvrBs3-mediated inducibility (Figure 5).

Furthermore, we have proven by using a heterologous promoter that the UPA box is not only necessary but also sufficient for gene activation by AvrBs3 (Figure 6). DNase I footprinting revealed that the UPA20 promoter region bound by AvrBs3 contains the UPA box in its center, flanked by nucleotides that are not involved in binding specificity (Figure 7).

AvrBs3Δrep16 has different recognition specificity than AvrBs3

Here, we obtained first insights into the recognition specificity of the AvrBs3 derivative AvrBs3Δrep16 that was generated by deletion of four repeats, numbers 11–14 (Herbers et al., 1992). Two nucleotides in the central part of the UPA box that appear not to be required for gene induction by AvrBs3 are essential for activation by AvrBs3Δrep16. A GA→CT exchange at positions 12 and 13 of the UPA box strongly enhanced binding of AvrBs3Δrep16 to the UPA20 promoter both in vitro and in yeast, and promoter activation in planta, without affecting inducibility by AvrBs3 (Figures 8b,c and S3). On the other hand, some nucleotides that are indispensable for AvrBs3 recognition, such as the C patch at positions 14–16 of the UPA box, are not essential for AvrBs3Δrep16 binding and gene activation. These data are consistent with the fact that the Bs3-E R gene promoter lacks part of the second C patch (CTA instead of CCC; Figure 8a) but is nevertheless strongly bound and activated by AvrBs3Δrep16 (Römer et al., 2007). The data also explain why UPA19 is better induced by AvrBs3Δrep16 than by wild-type AvrBs3 (Figure 2). The UPA19 promoter contains four substitutions compared to the UPA box consensus (Figure 3). In particular, the C→T exchange in the second C patch (position 16) strongly decreases AvrBs3-mediated activation as indicated by the UPA20 promoter variants (Figure 5). However, this exchange might only weakly affect the inducibility by AvrBs3Δrep16. Hence, AvrBs3 and its deletion derivative specifically bind to different DNA sequences. We believe that AvrBs3 family members generally display different DNA recognition specificities that are determined by the amino acid sequence of the repeats, and thus target a large spectrum of plant genes. Although target genes of AvrBs3 homologs from Xoo are known in rice (Gu et al., 2005; Yang et al., 2006; Sugio et al., 2007), DNA binding motifs are unknown. Taking into account that AvrBs3 binds to target DNA via the repeat region (Kay et al., 2007), and that the repeat region determines the specificity of action, it is essential to study target promoters induced by various AvrBs3-like proteins to elucidate the exact mode of action of these type III effectors.

Mechanism of AvrBs3-mediated gene induction

The UPA boxes known so far are located in the proximal promoter region, i.e. within 100 bp upstream of the transcription start site used in the presence of AvrBs3 (Figure 3). Interestingly, non-inducible and inducible alleles of Xa27 and Os8N3 from rice display nucleotide polymorphisms within 100 bp upstream of the transcription start site. It is conceivable that these substitutions are responsible for the lack of activation of the non-inducible alleles by the AvrBs3 homologs AvrXa27 and PthXo1 from Xoo, respectively (Gu et al., 2005; Yang et al., 2006). This suggests that DNA recognition motifs of these AvrBs3 homologs are located in positions similar to the position of the UPA box. It might very well be that AvrBs3-like effectors generally bind close to the transcription start site and directly interact with components of the basal transcription machinery. One candidate interactor is the γ subunit of the general transcription factor IIA (TFIIAγ). In rice, this protein is encoded by the recessive R gene xa5 (Iyer and McCouch, 2004; Jiang et al., 2006), which is involved in recognition of Avrxa5 from Xoo, which is most likely an AvrBs3 family member (Hopkins et al., 1992; Bai et al., 2000). The xa5 protein differs from the product of the susceptible allele (Xa5) in only one amino acid (E39V) (Iyer and McCouch, 2004; Jiang et al., 2006), which might prevent a putative interaction between Avrxa5 and xa5, thus abolishing Avrxa5-induced transcription of disease susceptibility genes in xa5/xa5 rice lines.

Intriguingly, our data suggest that not all target promoters of AvrBs3 and probably of AvrBs3 homologs are activated in the same manner. Although AvrBs3 must enter the plant cell nucleus for induction of all UPA genes, the AAD in AvrBs3 is in some cases dispensable (Figure 2). Similarly, UPA10 is also weakly induced by AvrBs3ΔAAD (Marois et al., 2002). This suggests that AvrBs3 activates UPA gene expression via alternative mechanisms, for example by interaction with plant transcription factors or co-activators that can compensate for the missing AAD in AvrBs3. Another possibility is that binding of AvrBs3ΔAAD displaces a negative regulator. The identification of interacting plant proteins will shed more light on the mechanism of gene induction used by AvrBs3 and related type III effectors.

Experimental procedures

Plant material and inoculations

Pepper (Capsicum annuum) plants of cultivar ECW (Minsavage et al., 1990) and Nicotiana benthamiana plants were grown in the greenhouse as described previously (Kay et al., 2007). Leaves of 6–7-week old plants were inoculated with Xcv (OD600 = 0.4 in 10 mm MgCl2) or Agrobacterium tumefaciens (see below) using a needleless syringe. Inoculated plants were kept in a Percival growth chamber (Percival Scientific, as described previously (Kay et al., 2007).

Bacterial strains and growth conditions

Escherichia coli TOP10, DB3.1, BL21 and BL21(DE3) (Invitrogen, were cultivated at 37°C in lysogeny broth (LB) (Bertani, 1951), and Agrobacterium tumefaciens GV3101 (Van Larebeke et al., 1974) was cultivated at 30°C in YEB (yeast extract broth). Xcv strain 85-10 (Minsavage et al., 1990; Thieme et al., 2005) was grown on NYG (nutrient yeast glycerol) agar plates as described previously (Daniels et al., 1984).

RT-PCR analysis

For time-course experiments, leaves of pepper plants of the ECW cultivar were inoculated with Xcv strain 85-10 carrying pL6Δ622 (Kay et al., 2005), to express avrBs3 from its own promoter, or with pLAFR6 (empty vector) (Bonas et al., 1989). To block eukaryotic protein biosynthesis, bacterial suspensions used for inoculations contained 50 μm cycloheximide. For specificity tests, Xcv 85-10 derivatives containing plasmids pDSF340 (Marois et al., 2002), pDSF330 (Van den Ackerveken et al., 1996), pDSF341 (Szurek et al., 2001) or pDSF316 (Van den Ackerveken et al., 1996) were used to express AvrBs3, AvrBs3ΔNLS, AvrBs3ΔAAD or AvrBs3Δrep16, respectively, from the triple lacUV5 promoter in pDSK602 (Murillo et al., 1994). Empty pDSK602 served as a negative control. RT-PCR was performed as described previously (Kay et al., 2007) using gene-specific oligonucleotides (Table S1).

Isolation and analysis of cDNA and promoter sequences

Full-length sequences of UPA genes were obtained by a combined analysis of cDNA and genomic sequences. 5′ and 3′ RACE fragments were generated using the SMART RACE cDNA amplification kit (Takara,, cloned into pCR-Blunt II-TOPO® (Invitrogen) and sequenced using the dideoxy termination method (Sanger et al., 1977) and an ABI 3130xl genetic analyzer (Applied Biosystems, Sequences were processed using sequencher 4.1.2 software (GeneCodes Corporation, Homologous sequences were obtained from the National Center for Biotechnology Information (NCBI) non-redundant database using BLASTP (Altschul et al., 1990). The promoter of UPA12 was isolated from genomic DNA of the pepper cultivar ECW using a GenomeWalker™ kit (Takara) and inverse PCR. The promoters of UPA14, UPA19, UPA21, UPA23 and UPA25 were isolated from a genomic BAC library of pepper cultivar ECW-30R (Jordan et al., 2006). Sequences were obtained by BAC sequencing and aligned by hand using BioEdit (Hall, 1999). Sequence logos were created using WebLogo (Crooks et al., 2004).

In planta promoter studies using β-glucuronidase (GUS)

UPA20 UPA box mutations were introduced into construct F0R0 (Kay et al., 2007) by SOE-PCR (splicing by overlap extension) (Horton et al., 1989), and cloned into pENTR/D-TOPO® (Invitrogen). Similarly, UPA box motifs were inserted into a 302 bp Bs4 promoter fragment (Schornack et al., 2005). To convert the Bs4 TATA box into UPA box motifs, the Phusion® site-directed mutagenesis kit (Finnzymes Oy, was used. Similarly, mutations were introduced into a 232 bp UPA12 promoter fragment that was amplified from genomic DNA of the pepper cultivar ECW and cloned into pENTR/D-TOPO® (for oligonucleotides, see Table S1). Promoter fragments were recombined into pBGWFS7 (containing a promoterless egfp::uidA fusion) (Karimi et al., 2005) and pGWB3 (containing a promoterless uidA) (Nakagawa et al., 2007), respectively. For GUS assays, N. benthamiana plants were co-infected with three Agrobacterium strains harboring pBGWFS7 or pGWB3 derivatives, carrying pBin61:p19 (Voinnet et al., 2003) and containing pVB60 (empty vector) (Mindrinos et al., 1994), pVSF300 (35S:avrBs3) (Van den Ackerveken et al., 1996) or pVSF316 (35S:avrBs3Δrep16) (Van den Ackerveken et al., 1996). Each Agrobacterium strain was adjusted to OD600 = 1.0 in infiltration medium (10 mm MgCl2, 10 mm MES, 150 μm acetosyringone), and mixed with the other two in the ratio 1:1:1. GUS activities were determined at 3 days post-infiltration as described previously (Kay et al., 2007).

Electromobility shift assay (EMSA)

For expression of GST::AvrBs3 and GST::AvrBs3Δrep16, plasmids pGEX-2TKM::300 (Gürlebeck et al., 2005) and pGEX-2TKM::316 (Römer et al., 2007), respectively, were used. Protein purification and EMSA were performed as described previously (Kay et al., 2007) with the following modifications: the binding reactions contained 100−500 fmol GST fusion protein, 100 fmol 5′-biotin-labeled DNA and 0−10 pmol unlabeled competitor DNA (for oligonucleotides, see Table S1).

DNase I footprinting

pENTR356 (Gürlebeck et al., 2005) derivatives were used for cloning of avrBs3 and avrBs3Δrep16 with translational stop codons into pDest17 (N-terminal His6 tag) (Invitrogen) by Gateway recombination (Invitrogen), resulting in pDest17::300 and pDest17::316, respectively, which were transformed into E. coli BL21(DE3) (Invitrogen).

Proteins were purified under native conditions using Ni-NTA agarose (Qiagen, according to the QlAexpressionist manual. Protein concentration was determined by the Bradford assay (Bio-Rad, and Coomassie brilliant blue staining after SDS–PAGE.

DNase I footprinting was performed as described previously (Zianni et al., 2006) with the following modifications. Fluorescently labeled PCR product was generated using oligonucleotides 5′-(6-FAM)-FPU20F and 5′-(HEX)-FPU20R (Table S1) (Metabion GmbH,, plasmid pCRBluntII-TOPO::FPU20 (UPA20 promoter fragment from −213 to +86) as template, and Phusion™ DNA polymerase (Finnzymes) and was purified using a Illustra™ GFX™ PCR DNA and gel band purification kit (GE Healthcare, Labeled DNA (100 ng) was incubated in binding buffer [18.4 mm Tris, 40 mm KCl, 2.5 mm MgCl2, 0.1 mm CaCl2, 0.04 mm EDTA, 0.04% NP-40, 0.8 mm DTT, 2% glycerol, 40 ng/μl poly(dI·dC), pH 7.5] with His6::AvrBs3, His6::AvrBs3Δrep16 or BSA in up to a 10× molar excess calculated on the basis of a protein dimer. Protein samples were pre-incubated on ice for 10 min before adding labeled DNA, then incubated at room temperature for 20 min. DNase I (MBI Fermentas, treatment was performed with 0.005 units per 50 μl reaction for 5 min at room temperature, and stopped by 15 min incubation at 75°C. DNA fragments, purified as above, were eluted in 20 μl H2O. For detection, 5 μl of DNase I digestion products, 0.025 μl of GeneScan™-500LIZ® size standard (Applied Biosystems) and 11 μl Hi-Di™ formamide (Applied Biosystems) were subjected to electrophoresis on an ABI PRISM® 3130xl genetic analyzer (Applied Biosystems) using fragment analysis application, 15 sec injection time and a voltage of 1.6 kV. Spectral calibration was performed using a multi-capillary DS-33 (dye set G5) matrix standard kit (Applied Biosystems).

Plasmid pCRBluntII-TOPO::FPU20 was sequenced using oligos 5′-(6-FAM)-FPU20F and 5′-(HEX)-FPU20R and the Thermo Sequenase dye primer manual cycle sequencing kit (USB Inc., according to the manufacturer’s instructions. Reactions were diluted 1:5 in water, and 4 μl were added to 0.025 μl GeneScan™-500LIZ® size standard and 12 μl Hi-Di™ formamide. The samples were analyzed using the 3130xl genetic analyzer with the same parameters as for DNase I reactions. Electropherograms were aligned using genemapper software version 4.0 (Applied Biosystems).

Additional methods

The methods for yeast one-hybrid studies, yeast colony PCR and immunoblot analysis of yeast protein extracts are described in Appendix S1.


We are grateful to J. Boch and F. Thieme for critical reading of the manuscript. We thank A. Landgraf, C. Kretschmer and B. Rosinsky for excellent technical assistance. This work was funded by a grant from the Deutsche Forschungsgemeinschaft (SFB 648) to U.B.

The Genbank accession numbers for the UPA14UPA19 and UPA21UPA25 sequences are FJ997849, FJ997850, FJ997851, FJ997852, FJ997853, FJ997854, FJ997855, FJ997856, FJ997857, FJ997858 and FJ997859, respectively.