Type III-dependent translocation of the Xanthomonas AvrBs3 protein into the plant cell



Many plant pathogenic bacteria utilize a conserved type III secretion system (TTSS) to deliver effector proteins into the host tissue. Indirect evidence has suggested that at least some effector proteins are translocated from the bacterial cytoplasm into the plant cell. Using an immunocytochemical approach, we demonstrate that the type III effector AvrBs3 from Xanthomonas campestris pv. vesicatoria localizes to nuclei of infected pepper leaves. Importantly, AvrBs3 translocation was observed in situ in native tissues of susceptible and resistant plants. AvrBs3 was detected in the nucleus as soon as 4 h post infection, which was dependent on a functional TTSS and the putative translocator HrpF. N-terminal AvrBs3 deletion derivatives are no longer secreted by the TTSS in vitro and could not be detected inside the host cells, suggesting that the N-terminus of AvrBs3 is important for secretion. Deletion of the nuclear localization signals in the AvrBs3 C-terminus, which are required for the AvrBs3-mediated induction of the hypersensitive reaction in resistant pepper plants, abolished AvrBs3 localization to the nucleus. This is the first report on direct evidence for translocation of a native type III effector protein from a plant pathogenic bacterium into the host cell.


To interact with their host, many Gram-negative pathogenic bacteria have evolved a specialized type III protein secretion system (TTSS) that transports proteins across both bacterial membranes without cleavage of a classical signal peptide. It has been shown for a number of animal pathogenic bacteria that so-called effector proteins are even translocated directly into the host cell via the TTSS (Hueck, 1998). In phytopathogens such as Pseudomonas syringae, Xanthomonas subsp., Ralstonia solanacearum and Erwinia amylovora, the TTSS is encoded by hrp (hypersensitive reaction and pathogenicity) genes, nine of which are highly conserved in plant and animal pathogens (hrc genes, for hrp conserved; Bogdanove et al., 1996). Loss of hrp gene function results in a pleiotropic phenotype: hrp mutants are unable to grow in the plant, no longer cause disease symptoms and fail to induce the hypersensitive reaction (HR) in resistant host and non-host plants (Lindgren, 1997). The HR is a rapid local, programmed cell death that is induced upon recognition of the pathogen and is concomitant with the inhibition of pathogen growth within the infected plant tissue (Klement, 1982). Whereas the core components of TTSS are highly conserved among plant and animal pathogenic bacteria, their secreted substrates appear to be extremely diverse (Cornelis and Van Gijsegem, 2000). Secreted proteins of plant pathogens include proteins essential for the function of the secretion machinery (e.g. Hrp pilus subunit; Romantschuk et al., 2001), pathogenicity proteins, as well as triggers of the plant defence reaction, for example, harpins and avirulence (Avr) proteins (Kjemtrup et al., 2000). Avr proteins were originally defined by their HR-inducing activity in plants that express a corresponding disease resistance (R) gene. In the absence of the avr or the R gene or both, no recognition takes place and disease occurs (Bonas and Van den Ackerveken, 1999). However, it has become clear that the primary function of many if not all avr genes is in virulence, and that plants have probably evolved the capacity to recognize these pathogen-derived molecules.

The subject of this study is the effector AvrBs3 from Xanthomonas campestris pv. vesicatoria (Xcv), the causal agent of bacterial spot disease on pepper and tomato. Bacteria enter the leaves via stomata or wounds and colonize the intercellular spaces of the mesophyll of susceptible plants. Early disease symptoms probably correlate with the release of water and nutrients which benefit bacterial multiplication (Innes, 2001). The ability of Xcv to cause disease in susceptible plants and to induce the HR in resistant plants depends on the hrp gene cluster, which encodes a TTSS (Bonas et al., 1991; Fenselau et al., 1992; Fenselau and Bonas, 1995; Huguet and Bonas, 1997; Rossier et al., 2000). Type III-dependent secretion in vitro has recently been shown for several pathogenicity and Avr proteins from Xcv (Rossier et al., 1999; Mudgett et al., 2000; Rossier et al., 2000; Escolar et al., 2001). Among these proteins is AvrBs3, which triggers recognition of the bacterium by pepper plants containing the resistance gene Bs3 (Minsavage et al., 1990).

The AvrBs3 protein belongs to a large family of conserved proteins in species of Xanthomonas, which often play a role in both virulence and avirulence (Swarup et al., 1991; 1992; Bai et al., 2000; Yang et al., 2000; Marois et al., 2002). An intriguing structural feature of members of the AvrBs3 protein family is their central region, which is composed of a variable number of nearly identical 34 amino acid tandem repeats (Vivian and Arnold, 2000). The 17.5 repeats of the 122 kDa AvrBs3 protein are essential for recognition by the Bs3 resistance gene (Herbers et al., 1992) and induction of plant cell hypertrophy in susceptible plants (Marois et al., 2002). Interestingly, the C-terminal region of AvrBs3 and its homologues contains typical eukaryotic features: nuclear localization signals (NLSs) and an acidic transcription activation domain (AAD) (Yang and Gabriel, 1995; Van den Ackerveken et al., 1996; Zhu et al., 1998; Szurek et al., 2001). Transient expression of a protein fusion between the AvrBs3 C-terminus and β-glucuronidase (GUS) in onion epidermis cells resulted in nuclear GUS activity, indicating that the NLSs are functional (Van den Ackerveken et al., 1996). As AvrBs3 is secreted by the Xcv TTSS in vitro (Rossier et al., 1999) and the NLSs and AAD are essential for both HR elicitation and the induction of cell hypertrophy (Van den Ackerveken et al., 1996; Zhu et al., 1998; Szurek et al., 2001; Marois et al., 2002), AvrBs3 is likely to be translocated into the host cell through the TTSS.

Translocation of type III effector proteins into host cells has been demonstrated for animal pathogenic bacteria (Staskawicz et al., 2001), using several experimental approaches, for example, translational fusions to adenylate cyclase (cya) (Sory and Cornelis, 1994), subcellular fractionation and immunofluorescence microscopy (Rosqvist et al., 1994; Lee et al., 1998). In plants, there is currently only indirect evidence for type III effector ‘injection’, the strongest being the observation that a specific HR is induced when an avr gene (e.g. avrBs3) is expressed in resistant plant cells (Bonas and Van den Ackerveken, 1999). Furthermore, the Pseudomonas syringae type III effector AvrRpt2 is processed during infection of Arabidopsis thaliana, and this processing was shown to take place in the plant and to be TTSS-dependent (Mudgett and Staskawicz, 1999). Additional support for the Avr protein translocation hypothesis came from subcellular localization studies of P. syringae Avr proteins after their expression in plant cells. It was shown that N-myristoylation motifs, which are important for HR-triggering activity, determine localization of the Avr protein in the plant plasma membrane (Nimchuk et al., 2000; Shan et al., 2000). Moreover, several Avr proteins interact with intracellular plant proteins in yeast, in vitro and in vivo (Scofield et al., 1996; Tang et al., 1996; Leister and Katagiri, 2000; Szurek et al., 2001; Mackey et al., 2002).

In this study, we provide direct evidence for TTSS-dependent translocation of AvrBs3 from Xanthomonas into infected plant cells using immunocytochemistry. Detection of AvrBs3, which is targeted to the nucleus of pepper cells, was shown to require a functional TTSS, the potential translocator HrpF and N-terminal sequences in AvrBs3.


The N-terminus of AvrBs3 is essential for type III secretion but not for HR induction

In many type III effector proteins, signals for secretion/translocation have been mapped to the amino-terminal region (Sory et al., 1995; Schesser et al., 1996; Mudgett et al., 2000; Lloyd et al., 2001). To address whether sequences essential for type III secretion of AvrBs3 are also localized in the amino terminus, we deleted amino acids 3–27 (AvrBs3Δ1) and 2–153 (AvrBs3Δ2) from AvrBs3 respectively (Fig. 1A). The avrBs3 gene and its derivatives were expressed in Xcv strain 85* from the lac promoter in plasmid pDSK602, which is constitutive in Xcv (Murillo et al., 1994; Table 1). In total protein extracts, both truncated AvrBs3 derivatives were present although in lower amounts than the wild-type protein (Fig. 1B). To compensate for differences in protein stability, loading in Fig. 1B (lane 4) was adjusted to obtain similar signal intensities (data not shown for AvrBs3Δ2). However, only the wild-type AvrBs3 protein was detected in culture supernatants, indicating that sequences in the N-terminal region of AvrBs3 are required for secretion (Fig. 1B, lanes 5 and 8). In contrast, deletion of 83 amino acids encompassing the NLS region in the C-terminus, which abolishes AvrBs3 activity (Van den Ackerveken et al., 1996), did not affect AvrBs3 secretion (Fig. 1B, lane 7). As reported previously, secretion of AvrBs3 was abolished in the TTSS-mutant strain 85*ΔhrcV (Fig. 1B, lane 6), which lacks a conserved inner membrane protein of the core TTSS (Rossier et al., 1999). Infection of resistant pepper plants (ECW-30R) with strain 85* expressing either AvrBs3Δ1 or AvrBs3Δ2 failed to induce the HR (Table 2). However, expression of the truncated avrBs3 alleles in the plant cell using Agrobacterium tumefaciens-mediated gene transfer resulted in the HR induction (Table 2). This demonstrates that, once inside the plant cell, AvrBs3 activity, i.e. recognition by Bs3, does not require the 152 N-terminal amino acids. Taken together, these data suggest that N-terminal deletions of AvrBs3 abolish both its secretion and translocation via the TTSS.

Figure 1.

The N-terminal region of AvrBs3 is needed for type III secretion.

A. AvrBs3 wild-type and deletion constructs used in this study. Structural features of the 122 kDa AvrBs3 protein are depicted. The central region consists of 17.5 tandem, nearly identical 34-amino-acid repeats. Two functional nuclear localization signals (NLSs) and an acidic activation domain (AAD) are in the C-terminal region (see also Table 1 ).

B. Immunoblotting analysis of total protein extracts (lanes 1–4) and culture supernatants (lanes 5–8) of the following strains: 85*(pDS300F) (wild-type AvrBs3, lanes 1 and 5), 85*Δ hrcV (pDS300F) (TTSS mutant, lanes 2 and 6), 85*(pDS330F) (AvrBs3ΔNLS, lanes 3 and 7) and 85*(pDS351F) (AvrBs3Δ1, lanes 4 and 8). 3 µl aliquots of total protein extracts and 6 µl aliquots of supernatants were loaded. To compensate for AvrBs3Δ1 protein instability, 7 and 14 µl were loaded in lanes 4 and 8 respectively. Proteins were separated by SDS-PAGE (8% polyacrylamide) and transferred to nitrocellulose. The blot was reacted with an AvrBs3-specific polyclonal antibody, and reprobed with a specific antibody against the intracellular protein HrcN to ensure that no bacterial lysis had occurred.

Table 1. . Xanthomonas strains and plasmids used in this study.
DesignationRelevant characteristicsReference/source
Xcv strains
 85*Derivative of 85–10 carrying hrpG*; constitutive expression of hrp genes Wengelnik et al. (1999)
 85*ΔhrcV hrcV deletion mutant of 85* Rossier et al. (2000)
 85*ΔhrpF hrpF deletion mutant of 85* Büttner et al. (2002)
 82–8Pepper race 1; wild-type strain expressing AvrBs3 from its own promoter Minsavage et al. (1990)
 82*Derivative of 82–8 carrying hrpG* Wengelnik et al. (1999)
 82*ΔhrcV hrcV deletion mutant of 82* Rossier et al. (1999)
 pDS300F1996pDSK602 expressing AvrBs3 from the lac promoterVan den Ackerveken et al. (1996)
 pDS351FpDSK602 expressing AvrBs3 deleted in 25 N-terminal aaThis study
 pDS356FpDSK602 expressing AvrBs3 deleted in 152 N-terminal aaThis study
 pDS330FpDSK602 expressing AvrBs3 deleted in 83 C-terminal aa including the NLS region Van den Ackerveken et al. (1996)
Table 2. . Type III secretion and recognition of AvrBs3 derivatives.
AvrBs3 constructExpression in XcvExpression in planta
via Agrobacterium tumefaciens
in vitro secretion Plant phenotype on pepper Bs3aPlant phenotype on pepper Bs3a
  • a

    . HR, hypersensitive reaction (defence); WS, watersoaking (disease symptoms).

  • Identical results were obtained in four independent experiments.


Immunodetection of AvrBs3 in pepper leaf tissue after transient expression

Next, we tested whether AvrBs3 could be detected in plant cells using immunocytochemistry. As protein amounts delivered by Xcv are believed to be low, we first established immunocytochemistry protocols for pepper cells expressing avrBs3 via Agrobacterium-mediated gene transfer under control of the constitutive cauliflower mosaic virus 35S* promoter. This promoter contains a bacterial transcription terminator resulting in undetectable expression in bacteria (Van den Ackerveken et al., 1996). The subcellular localization of AvrBs3 was determined in serial transverse sections of susceptible pepper (ECW) leaf tissue expressing avrBs3, using an AvrBs3-specific polyclonal antibody and alkaline phosphatase-coupled secondary antibody (see Experimental procedures). DNA-specific staining with DAPI revealed that the AvrBs3 protein was detectable in the nucleus of some but not all mesophyll cells (Fig. 2A and A′). This is probably due to the fact that not all cells are transformed by Agrobacterium. Thus, the untransformed cells act as negative control for the immunostaining. No significant signals were observed in leaf sections of tissue infected with Agrobacterium carrying the empty vector (data not shown). As expected, deletion of the NLS region in the C-terminus of AvrBs3 (avrBs3ΔNLS) abolished its targeting to the nucleus and led to specific staining in the cytoplasm of infected pepper cells (Fig. 2B and B′). Similar data were obtained using fluorescein-isothiocyanate (FITC)-conjugated secondary antibody (data not shown).

Figure 2.

AvrBs3 localizes to the nucleus of transformed pepper cells. Leaves of pepper ECW were infected with Agrobacterium tumefaciens strains (A) GV3101(pVSF300) and (B) GV3101(pVSF330), carrying wild-type avrBs3 and avrBs3ΔNLS on the T-DNA respectively. After T-DNA transfer into the plant cell both genes are highly expressed under control of the constitutive 35S* promoter. Leaves were inoculated with bacterial suspensions of 1 × 10 9 cfu ml −1 and harvested 60 hpi for sectioning. Transverse leaf sections were incubated with anti-AvrBs3 antibody and alkaline phosphatase-coupled secondary antibody. The wild-type AvrBs3 protein is localized in nuclei of infected mesophyll cells (A), whereas after expression of AvrBs3ΔNLS (B) AvrBs3-specific staining is confined to the cytoplasm. (A′) and (B′) show the identical fields UV-illuminated for the DAPI-stained nuclei. Arrowheads and arrows indicate nuclei that are stained and not stained for AvrBs3 respectively. The bar represents 20 µm.

Importantly, immunoblotting analysis of pepper leaf tissue expressing avrBs3 or avrBs3ΔNLS after A. tumefaciens-mediated gene delivery showed similar protein levels indicating that the NLS deletion does not affect protein stability (Fig. 3A, lanes 1 and 2). No specific signal was detectable in total protein extracts prepared from leaves infiltrated with A. tumefaciens carrying the empty vector (Fig. 3A, lane 3). Altogether, these results demonstrate that avrBs3 expression in the plant results in targeting of the protein to the nucleus. Moreover, deletion of the NLS region, which previously was shown to abolish AvrBs3 recognition in resistant plants (Van den Ackerveken et al., 1996; Table 2), prevents nuclear localization of AvrBs3.

Figure 3.

Comparison of amounts of AvrBs3 and derivatives in infected pepper leaf extracts.

A. Agrobacterium tumefaciens -mediated expression of AvrBs3 in ECW pepper leaves. Leaves were inoculated with a 1 × 10 9  cfu ml −1 suspension of A. tumefaciens and harvested 60 hpi. 50 µg of total protein extracts were loaded in each lane, subjected to SDS-PAGE and Western blot analysis using anti-AvrBs3 polyclonal antibody. Lane 1, AvrBs3; lane 2, AvrBs3ΔNLS; lane 3, empty vector.

B. Expression of AvrBs3 and derivatives in ECW pepper leaves infected with Xcv . The following strains were used: lane 1, 85*(pDSK602) (empty vector); lane 2, 85*(pDS300F) (wild-type AvrBs3); lane 3, 85*Δ hrcV (pDS300F); lane 4, 85*(pDS351F) (AvrBs3Δ1); lane 5, 85*(pDS330F) (AvrBs3ΔNLS); lane 6, 85*Δ hrpF (pDS300F); lane 7, 82–8; lane 8, 82*; lane 9, 82*Δ hrcV . Pepper ECW leaves were inoculated with Xcv suspensions of 1 × 10 9  cfu ml −1 and harvested 12 hpi. 20 µg of total protein extracts were loaded in each lane, except for lane 4 (75 µg) and lanes 7–9 (100 µg), owing to lower protein expression level or stability. Similar results were obtained in three independent experiments.

In situ immunolocalization of AvrBs3 in infected leaf tissue

We then examined whether AvrBs3 could be detected inside the plant cell after delivery by Xcv. To maximize AvrBs3 transfer by the TTSS, AvrBs3 was expressed under control of the lac promoter in Xcv strain 85*(pDS300F). Strain 85* carries the hrpG* mutation that leads to the expression of a constitutively active master regulator of the TTSS, hrpG (Wengelnik et al., 1999). Type III secretion from strain 85*, however, is still dependent on plant inducing conditions (Rossier et al., 1999). AvrBs3 localization in pepper ECW was determined immunocytochemically using the AvrBs3-specific antibody on 3-µm-thick serial sections of leaf tissue, embedded 12 hpi. Consistent with the data shown in Fig. 2, AvrBs3 staining was confined to the nucleus of examined mesophyll cells (arrowheads in Fig. 4A). When pepper plants were infected with the TTSS mutant strain 85*ΔhrcV(pDS300F), no AvrBs3 signal was detectable (Fig. 4B). To rule out that lack of AvrBs3 detection in the case of the hrcV mutant was due to fewer bacteria or less protein in infected leaves, bacterial numbers and AvrBs3 amounts were examined. Pepper ECW leaf discs taken at 12 hpi showed similar bacterial numbers and AvrBs3 protein amounts for both 85*(pDS300F) and 85*ΔhrcV(pDS300F) (Fig. 3B, lanes 2 and 3). Thus, AvrBs3-specific staining of nuclei clearly depends on a functional TTSS. As anticipated from the in vitro secretion experiments presented above (Fig. 1B), no AvrBs3-specific signal was observed in infected tissue when the putative N-terminal secretion signal in AvrBs3 was deleted (Fig. 4C). Furthermore, deletion of the NLS region in AvrBs3 abolished nuclear localization of AvrBs3 (Fig. 4D). Although Western blot analysis of pepper leaf tissue infected with Xcv expressing AvrBs3 or AvrBs3ΔNLS showed similar protein levels (Fig. 3B, lanes 2 and 5), the AvrBs3ΔNLS protein was barely detectable in the cytoplasm of plant cells. This is probably due to the large dilution effect in this compartment rather than to protein instability and underlines the lower protein amounts in the plant cell when delivered by Xcv as compared with Agrobacterium-mediated expression (compare Figs 2B and 4D). In conclusion, lack of AvrBs3Δ1 detection in the nuclei is due to the inability to be secreted and not because of lower protein amount (see below), whereas lack of AvrBs3ΔNLS detection is probably due to the absence of nuclear targeting. Finally, we investigated whether AvrBs3 could also be detected in leaf cells of the resistant pepper line ECW-30R, which recognizes the AvrBs3 protein (Minsavage et al., 1990). As shown in Fig. 4E, presence of the Bs3 resistance gene did not impair AvrBs3 accumulation in the nucleus at 12 hpi, which is before the macroscopic appearance of the HR (24 hpi).

Figure 4.

AvrBs3 immunolocalization in sections of pepper leaves infected with Xcv . Leaves of pepper ECW (A–D) and pepper ECW-30R, which contains the resistance gene Bs3 (E) , were infected with the following Xcv strains: (A and E) 85*(pDS300F), which expresses wild-type AvrBs3; (B) 85*Δ hrcV (pDS300F); (C) 85*pDS351F, which expresses AvrBs3Δ1; and (D) 85*pDS330F, which expresses AvrBs3ΔNLS. AvrBs3 was visualized in situ 12 hpi using an AvrBs3-specific antibody and alkaline phosphatase-coupled secondary antibody. AvrBs3 is detected in the nuclei of mesophyll cells of susceptible (A) and resistant (E) pepper lines. No signal is detected in nuclei of susceptible plants infected with strains carrying a deletion in hrcV (B), in the N-terminus of AvrBs3 (C), or in the NLS region of AvrBs3 (D). (A′–E′) show the identical fields UV-illuminated for the DAPI-stained nuclei. The reduced DAPI fluorescence in AvrBs3-stained nuclei is due to the interference with the product of the alkaline phosphatase chromogenic reaction. Note that Xcv only infects mesophyll (arrowheads in A), but not epidermis (arrows in A′) cells. The spotty background staining visible in (A) was observed occasionally and was independent of the presence of the AvrBs3 protein. e, epidermis cell; m, mesophyll cell. The bar represents 20 µm.

The ‘nuclear targeting assay’ as a tool to characterize AvrBs3 translocation

Using immunocytochemical protocols, we were able to detect AvrBs3 inside the host plant cell in situ. However, this method is time-consuming and not amenable for quantitative analysis. We developed therefore a ‘nuclear targeting assay’, taking advantage of AvrBs3 localization to the plant nucleus. First, susceptible pepper leaves (ECW) were infiltrated with Agrobacterium strains carrying different constructs. Nuclei and other organelles prepared 60 hpi were fixed on microscopy glass slides (see Experimental procedures). Using the AvrBs3-specific antibody and alkaline phosphatase-coupled secondary antibody, followed by DAPI staining, the AvrBs3 protein was detected in plant nuclei after Agrobacterium-mediated expression of avrBs3 (data not shown). In control experiments using Agrobacterium containing the empty vector or avrBs3ΔNLS on the T-DNA, no staining was observed in plant cell nuclei. As described above the N-terminus of AvrBs3 is essential for type III secretion but not for AvrBs3 recognition in planta (see Fig. 1 and Table 2). As expected from the biological activity assays (Table 2), Agrobacterium-mediated expression of the N-terminal deletion mutant AvrBs3Δ1 in the plant led to strong staining in nuclei (data not shown), indicating that the N-terminus of AvrBs3 is dispensable for nuclear targeting. In conclusion, the establishment of the ‘nuclear targeting assay’ using Agrobacterium was successful and used to further analyse the translocation of AvrBs3 into the plant cell following natural infection with Xanthomonas. Nuclei and other organelles, prepared 12 hpi from susceptible pepper leaves (ECW) infected with Xanthomonas strain 85*(pDS300F), were fixed on microscopy glass slides. Using the AvrBs3-specific antibody, the AvrBs3 protein was detected in plant nuclei (Fig. 5A), which were identified by DAPI staining (Fig. 5A′) and phase contrast microscopy (Fig. 5A′′). In control experiments using the TTSS mutant strain ΔhrcV, AvrBs3 was not detectable in plant cell extracts (Fig. 5B). As anticipated from the data presented above, no AvrBs3 signal was observed when the putative N-terminal secretion signal (data not shown) or the NLS region (Fig. 5D) was deleted in AvrBs3. In an attempt to compensate for the apparently lower amount of AvrBs3Δ1 compared with the wild-type protein (Fig. 3B), leaves were inoculated with lower density (1.5 × 108 cfu ml−1) of the wild-type strain 85*(pDS300F). Although in this case the relative amounts of AvrBs3 and AvrBs3Δ1 per leaf disk were similar (data not shown), AvrBs3-stained nuclei were only observed with 85*(pDS300F). In addition to the TTSS mutant ΔhrcV, we tested Xcv strains carrying a deletion in hrpF. hrpF mutants still secrete proteins via the TTSS in vitro but no longer cause disease or induce the HR in the plant, suggesting that hrpF encodes the ‘translocator’ (Rossier et al., 2000; Büttner et al., 2002). As shown in Fig. 5C, no stained nuclei were obtained from pepper leaves infected with the mutant strain 85*ΔhrpF(pDS300F). Taken together with the results presented above (Fig. 4), detection of AvrBs3 in plant nuclei depends on both secretion from the bacteria and translocation into the plant cytosol.

Figure 5.

‘Nuclear targeting assay’: AvrBs3 detection is TTSS- and hrpF -dependent. Pepper ECW leaves were infected with the following Xcv strains: (A) 85*, (B) 85*Δ hrcV , (C) 85*Δ hrpF (all expressing AvrBs3 from pDS300F), (D) 85*(pDS330F) (expressing AvrBs3ΔNLS), and (E) 82–8, which expresses AvrBs3 from a native plasmid. Leaves were inoculated with bacterial suspensions of 1 × 10 9  cfu ml −1 and harvested 12 hpi. Total cell extracts were fixed and immunostained for AvrBs3 (see Experimental procedures ).

A–E. Bright-field images.

A′–E′ and A′′–E′′. The identical fields UV-illuminated for the DAPI-stained nuclei and phase contrast respectively. The nuclei are indicated by arrowheads. The bar represents 20 µm.

To study the time-course of AvrBs3 translocation, nuclei from pepper ECW leaves infected with strain 85*(pDS300F) were examined in 2 h-intervals from 0 to 12 hpi. The AvrBs3 protein was detected in plant nuclei as early as 4 hpi (not shown). The number of stained nuclei increased over time: 8% of the nuclei (80 out of 998 analysed) at 4 hpi and 63% of the nuclei (526 out of 835) at 12 hpi stained positive for AvrBs3.

Visualization of AvrBs3 delivered by a naturally occurring Xcv strain

The experiments described above were performed using Xcv strains harbouring a plasmid in which avrBs3 is under the control of the lac promoter. To address whether AvrBs3 is also detectable in plant nuclei after pepper infection with a strain expressing wild-type levels of AvrBs3, we used Xcv strain 82*, which expresses avrBs3 under the control of its own promoter (Table 1). The ‘nuclear targeting assay’ developed above was performed with leaf tissue of pepper ECW infected with Xcv strain 82*. Nuclei exhibited significant AvrBs3 staining 12 hpi but not at 10 hpi (data not shown). In contrast, no AvrBs3 signal was detected in nuclei isolated from leaves infected with the isogenic TTSS mutant 82*ΔhrcV (data not shown). All the strains analysed so far in this study express components of the TTSS constitutively. When we used the naturally occurring strain 82–8, AvrBs3 was clearly detected in the nuclei at 12 hpi (Fig. 5E and E′), although the protein expression level was lower than with Xcv 85*(pDS300F) (Fig. 3B, lanes 2 and 7). In conclusion, translocation of AvrBs3 expressed in the wild-type strain could be clearly demonstrated.


Using immunocytochemical detection protocols the type III effector AvrBs3 was demonstrated to be delivered by the plant pathogen Xcv into host plant cells in situ. Significantly, TTSS-dependent AvrBs3 translocation was observed in the native tissues of plants infected with Xanthomonas, whereas translocation studies in systems involving animal pathogenic bacteria used cultured HeLa cells or macrophages, but not the intact host organism (Rosqvist et al., 1994; Collazo et al., 1995; Sory et al., 1995; Wood et al., 1996). Moreover, we provide experimental evidence for AvrBs3 targeting to the plant nucleus. Nuclear localization of AvrBs3 confirmed our previous data showing that a fusion between the AvrBs3 C-terminal region and β-glucuronidase resulted in nuclear targeting of the reporter protein in onion epidermis cells (Van den Ackerveken et al., 1996).

We show here that translocation was not only dependent on a functional TTSS, but also on the HrpF protein, which is itself secreted but not required for type III-dependent secretion in vitro. As hrpF mutants are unable to grow in planta and do not provoke any plant reaction, it has been suggested that the HrpF protein plays a role after secretion, i.e. in translocation (Rossier et al., 2000). The data presented here corroborate this model. Whether HrpF interacts with the host plasma membrane is not known. However, it has recently been shown that HrpF has lipid-binding activity and forms a pore in lipid bilayers, supporting the function of HrpF as a translocon protein (Büttner et al., 2002).

Despite extensive efforts to demonstrate the translocation of effector proteins into the plant cell by the TTSS of a plant pathogenic bacterium, direct evidence was lacking until now. During the course of the preparation of this manuscript, type III-dependent translocation of the Xcv protein AvrBs2 was shown using cya fusions (Casper-Lindley et al., 2002). In our work, AvrBs3 accumulation in the nucleus was key to facilitate detection by immunocytochemistry. In the ‘nuclear targeting assay’, AvrBs3-staining was only observed in nuclei after freezing the samples, probably because this facilitates the penetration of antibodies into the nuclei. Moreover, the use of alkaline phosphatase-coupled secondary antibody appeared to be important, because AvrBs3 was undetectable in nuclei of Xcv-infected leaves using FITC conjugates here and in previous studies (Van den Ackerveken et al., 1996). Localization of AvrBs3 by immunofluorescence was only successful when the protein was overexpressed in the plant cell using Agrobacterium-mediated gene transfer (data not shown). Although bacteria could be identified by DAPI staining or phase contrast using the ‘nuclear targeting assay’ (data not shown), they were not stained for AvrBs3. Possible explanations are the structural differences between the bacterial and nuclear envelope and differential permeability for antibodies after treatment with Triton X-100.

In this study, we have established the ‘nuclear targeting assay’ that will enable us to analyse the translocation signal of AvrBs3. Type III translocation signals have been most extensively studied in Yersinia Yops (Yersinia outer proteins) (Cornelis, 2000). For several Yops, a minimal region of 15–17 N-terminal amino acids was found to be essential for secretion in vitro, whereas additional 50–70 amino acids were required for translocation into the eukaryotic cell (Plano et al., 2001). Interestingly, a putative translocation motif consisting of a proline residue surrounded by basic amino acids is conserved in the N-terminus of all avirulence proteins characterized so far in Xcv (Escolar et al., 2001). An additional potential application of the ‘nuclear targeting assay’ is the use of this method to demonstrate the translocation of other type III effectors fused to the AvrBs3 NLSs into the plant cell.

Visualization of AvrBs3 in the plant nucleus was dependent on the NLSs in the C-terminus, which is consistent with our earlier observation that the NLSs are required for AvrBs3 activity (Van den Ackerveken et al., 1996). Nuclear localization of AvrBs3 is probably due to its interaction with pepper importin α, a component of the nuclear import machinery, which was recently isolated from pepper by a yeast two-hybrid screen and which also interacts with AvrBs3 in vitro (Szurek et al., 2001). One of the key questions for future research is the biochemical function of AvrBs3 and the nature of the cellular target(s) of AvrBs3 in the plant. Intriguingly, AvrBs3 and its homologues contain an acidic transcription activation domain in the C-terminus, reinforcing the presumed function of members of this effector protein family in the nucleus (Zhu et al., 1998; Yang et al., 2000; Szurek et al., 2001). We therefore believe that AvrBs3 modulates transcription and thus induces changes in the plant metabolism, which contribute to disease development. The observation that AvrBs3 induces hypertrophy of mesophyll cells in susceptible plants (Marois et al., 2002) fits with this idea. Although there is now direct evidence for effector translocation from a plant pathogen into the host cell, it is still a mystery how the TTSS transfers proteins across the plant cell wall and plasma membrane. The finding of TTSS-specific pili that might serve as conduit (Brown et al., 2001; Jin and He, 2001; Jin et al., 2001) is a step towards a better understanding of this crucial process.

Experimental procedures

Bacterial strains, growth conditions and plasmids

Bacterial strains used in this study were Escherichia coli strain DH5α (Bethesda Research Laboratories), Agrobacterium tumefaciens strain GV3101 (Van Larebeke et al., 1974), and different Xanthomonas campestris pv. vesicatoria (Xcv) strains (see Table 1). E. coli cells were cultivated at 37°C in Luria–Bertani medium, Xcv at 30°C in NYG broth (Daniels et al., 1984), and A. tumefaciens at 30°C in YEB medium (Van den Ackerveken et al., 1996). Plasmids were introduced into E. coli by electroporation, and into Xcv and A. tumefaciens by conjugation, using pRK2013 as a helper plasmid in triparental mating (Figurski and Helinski, 1979).

Construction of plasmids

The N-terminal deletions in AvrBs3 were constructed as follows: for avrBs3Δ1, a linker (obtained with oligonucleotides 5′-GATCGTCAGGCCTATGCA-3′ and 5′-TAGGCCTGAC-3′) was ligated into pUS300F digested with BamHI–PstI (Van den Ackerveken et al., 1996), yielding pUS351F. pUS351F encodes AvrBs3Δ1 in which amino acid residues 3–27 are replaced by a short peptide (RQAS). AvrBs3Δ2 lacks amino acids 2–153: a 2886 basepair (bp) fragment, obtained by BstYI digestion of pUS300F, was ligated to the 3308 bp BamHI fragment from pUS300F, resulting in pUS356F. The avrBs3 derivatives were cloned into the expression vector pDSK602 (Murillo et al., 1994), resulting in pDS351F and pDS356F. For Agrobacterium-mediated gene transfer, avrBs3 derivatives were cloned into pBI1.4t (Mindrinos et al., 1994).

Plant material and plant inoculations

Pepper (Capsicum annuum) plants of cultivar Early Cal Wonder (ECW) and the near-isogenic line ECW-30R, which contains the resistance gene Bs3 (Minsavage et al., 1990), were grown in the greenhouse. Xcv and A. tumefaciens suspensions of 5 × 108 colony forming units (cfu) ml−1 in 10 mM MgCl2 were infiltrated into the intercellular space of leaves using a needle-less syringe as described (Bonas et al., 1989; Van den Ackerveken et al., 1996). To rule out that lack of hypersensitive reaction (HR) was due to lower protein amounts in the case of bacteria expressing AvrBs3Δ1 or AvrBs3Δ2 (Table 2), leaves of pepper ECW-30R were infiltrated with a higher inoculum (1 × 109 cfu ml−1). No HR was observed while bacteria expressing wild-type AvrBs3 were able to induce the HR when inoculated at 1 × 108 cfu ml−1. Macroscopic reactions were scored 2–5 days after inoculation. For immunocytochemistry, the inoculum concentration was approximately 109 cfu ml−1. Bacterial growth in planta was determined as described (Bonas et al., 1991).

Protein secretion experiments and immunoblot analysis

In vitro secretion experiments were performed as described ( Rossier et al., 1999 ). Total protein extracts and culture supernatants were concentrated 10- and 100-fold respectively. Western blots were reacted with affinity-purified AvrBs3-specific polyclonal antibody (diluted 1:15 000) ( Knoop et al., 1991 ) and with anti-rabbit horseradish peroxidase-conjugated secondary antibody, which was detected by enhanced chemiluminescence (Amersham Pharmacia Biotech). The membranes were reprobed with a specific antibody against the intracellular protein HrcN to ensure that no bacterial lysis had occurred ( Rossier et al., 1999 ). Protein extracts from infected leaves were prepared by grinding two leaf discs (0.8 cm in diameter) from different leaves in liquid nitrogen in 80 µl of protein extraction buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) to which 20 µl of 5× Laemmli buffer was added. Protein samples were separated by SDS-PAGE and subjected to immunoblot analysis.

Immunostaining of leaf sections

Leaf segments were fixed for 3 h with 3% paraformaldehyde in phosphate-buffered saline (PBS), washed, dehydrated and embedded in polyethylene glycol (Van Lammeren et al., 1985). Then, 3 µm serial sections were mounted on glass slides, incubated with 0.1 M NH4Cl, washed with PBS, and incubated in PBS-3% BSA and anti-AvsBs3 antibody (diluted1:3000). Antibody binding was visualized by sheep anti-rabbit IgG conjugated to alkaline phosphatase (Rochd) according to the manufacturer's instructions. To identify nuclei the samples were incubated with 0.1 µg ml−1 of 4′,6-diamidino-2-phenylindole (DAPI) and mounted in Citifluor-glycerol (Citifluor). Fluorescence images were obtained using a Zeiss (Jena) Axioplan 2 microscope equipped with a Fuji Dig HC-300Z digital camera (Fujifilm) using filter blocks for DAPI. Experiments were repeated three times with similar results. Approximately 100 sections were examined per assay for each tested strain.

‘Nuclear targeting assay’

Leaf discs (1 cm in diameter) were chopped on glass slides in 4% formaldehyde in PBS, pH 7, 0.05% Triton X-100. Samples were frozen in liquid nitrogen and treated as described above for leaf sections. In addition to fluorescence images, we used phase-contrast, enabling the visualization of both nuclei and chloroplasts. Experiments were repeated at least three times with similar results.


We are grateful to R. Kahmann, E. Marois and R. Koebnik for critically reading the manuscript, and to A. Fischer for technical advice. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (SFB 363) to U.B.