Report

A Bacterial Virulence Protein Suppresses Host Innate Immunity to Cause Plant Disease

See allHide authors and affiliations

Science  14 Jul 2006:
Vol. 313, Issue 5784, pp. 220-223
DOI: 10.1126/science.1129523

Abstract

Plants have evolved a powerful immune system to defend against infection by most microbial organisms. However, successful pathogens, such as Pseudomonas syringae, have developed countermeasures and inject virulence proteins into the host plant cell to suppress immunity and cause devastating diseases. Despite intensive research efforts, the molecular targets of bacterial virulence proteins that are important for plant disease development have remained obscure. Here, we show that a conserved P. syringae virulence protein, HopM1, targets an immunity-associated protein, AtMIN7, in Arabidopsis thaliana. HopM1 mediates the destruction of AtMIN7 via the host proteasome. Our results illustrate a strategy by which a bacterial pathogen exploits the host proteasome to subvert host immunity and causes infection in plants.

Many plant and human pathogenic bacteria rely on an essential virulence system—the type III secretion system—to inject virulence effector proteins into the host cell to cause infection (13). Recent research has documented the ability of effector proteins of mammalian pathogenic bacteria to modulate host cytoskeleton dynamics, membrane composition, vesicle trafficking, and host immunity (4). In contrast, very little is known about the molecular mechanisms by which bacterial effector proteins induce disease in plants. Emerging evidence suggests that a major function of these effector proteins is to suppress host immune responses in susceptible plants (511). However, the mechanisms by which effector proteins subvert host immune responses are poorly understood at the molecular level.

Pseudomonas syringae infects a wide range of economically important plant species. All of the examined P. syringae strains contain a common genomic pathogenicity island, which is composed of type III secretion–associated hrp/hrc genes, an exchangeable effector locus, and a conserved effector locus (12). A partial deletion of the conserved effector locus in the ΔCEL mutant of Pst DC3000 resulted in a notable reduction of the bacterial population and the complete elimination of disease symptoms (necrosis and chlorosis) in infected tomato and Arabidopsis plants (12, 13). The severe virulence defect in the ΔCEL mutant bacteria is caused by the deletion of the functionally redundant effector genes hopM1 (formerly hopPtoM) and avrE (13). pORF43, a plasmid expressing only HopM1 and its cognate chaperone ShcM, is sufficient to fully complement the virulence defect of the ΔCEL mutant in Arabidopsis (13).

HopM1 is a 712–amino acid protein that is translocated into the host cell (14). In this study, we investigated whether HopM1, expressed inside the host cell, could restore the virulence of the ΔCEL mutant. Plant-expressed full-length HopM1 almost fully complemented the virulence defect of the ΔCEL mutant (Fig. 1A). Moreover, the complementation was specific to the ΔCEL mutant because the multiplication of the type III secretion–defective hrcC mutant (15), which does not secrete any effector proteins, was increased only slightly (Fig. 1A). Subcellular fractionation experiments revealed that HopM1 was enriched in the endomembrane fraction in the transgenic plants (Fig. 1B). Taken together, these results suggest that HopM1 acts in a host's endomembrane compartment or compartments to promote bacterial pathogenesis.

Fig. 1.

Analysis of HopM1 transgenic Arabidopsis plants. Bacterial multiplication in leaves of wild-type Arabidopsis plants (Col-0 gl1) or transgenic plants expressing full-length HopM1 (A), or in leaves of Col-0 gl1 and transgenic plants expressing deletion derivatives of HopM1 (C). Plants were sprayed with dexamethasone 24 hours before bacterial inoculation [1 × 106 colony-forming units (CFUs) per milliliter]. Bacterial populations were determined at day 3 after inoculation. Error bars indicate SD. (B) Immunoblot analysis of HopM1, plasma membrane–localized H+ adenosine triphosphatase, and Golgi-localized xyloglucan xylosyltransferase AtXT1 after total leaf proteins of HopM1 transgenic Arabidopsis plants were separated into the indicated subcellular fractions. TM, total membrane; S, soluble fraction; PM, plasma membrane; EM, endomembrane.

To define the regions important for the virulence function of HopM1, we produced transgenic Arabidopsis plants expressing a series of C- and N-terminally truncated derivatives of HopM1. HopM1101–712, which lacks the first 100 amino acids, partially restored bacterial multiplication and the chlorotic symptom of the ΔCEL mutant (Fig. 1C and fig. S1). None of the other 11 truncated derivatives could complement the virulence defect of the ΔCEL mutant (Fig. 1C and fig. S1). However, further analysis showed that the N terminus of HopM1 (HopM11–200 and HopM11–300) exerted a dominant negative effect on the function of full-length HopM1 delivered from the ΔCEL mutant (pORF43) (Fig. 1C and fig. S1). Necrosis, chlorosis, and bacterial multiplication were significantly reduced in these plants, compared with those in Col-0 gl1 or transgenic plants expressing other HopM1 truncated derivatives (Fig. 1C and fig. S1). The dominant negative effect was specific to HopM1 because HopM11–200 and HopM11–300 plants were still susceptible to Pst DC3000, which produces AvrE, in addition to HopM1 (Fig. 1C and fig. S1). These results suggest that the N-terminal 200 to 300 amino acids of HopM1 can function as an independent domain in vivo, interfering with the virulence function of the full-length HopM1 delivered from bacteria.

The dominant negative effect in a cellular process is often caused by unproductive protein-protein interactions (16, 17). We therefore reasoned that HopM11–200 and HopM11–300 may compete with full-length HopM1 for interaction with host proteins. To test this hypothesis, we performed yeast two-hybrid (Y2H) screens of an Arabidopsis cDNA library using HopM11–300 and full-length HopM1 as baits. We did not recover any interactors with full-length HopM1, but we obtained 21 strong interactors of HopM11–300, which were named AtMIN (Arabidopsis thaliana HopM interactors). AtMIN proteins interacted not only with HopM11–300 (Fig. 2A) but also with HopM11–200 (fig. S2A). Furthermore, the interaction between HopM11–300 and AtMIN proteins could also be observed in plant cells, based on transient expression experiments in Nicotiana benthamiana after conducting pull-down assays (Fig. 2B for AtMIN7).

Fig. 2.

Physical interaction between HopM1 and AtMIN proteins and HopM1-dependent destabilization of AtMIN proteins. (A) Y2H assay of physical interaction between HopM11–300 expressed from pGILDA (Clontech) and AtMIN proteins expressed from pB42AD (Clontech), shown for AtMIN2, 7, 10, and 12. Yeast colonies were grown on complete minimal medium containing galactose and 5-bromo-4-chloro-3-indoxyl-β-d-galactopyranoside. Blue indicates interaction and white indicates no interaction. +, positive control strain containing pLexA-p53 and pB42AD-T. (B) Immunoblot analysis of the physical interaction between AtMIN7-HA and 6xHis-HopM1–300 (1) or between AtMIN7-HA and 6xHis-HopM301–712 (2) in N. benthamiana leaves with a protein pull-down assay (27). AtMIN-HA and 6xHis-HopM1 proteins were detected with the HA and 6xHis epitope antibodies, respectively. AtMIN7-HA was pulled down with HopM1–300 but not with 6xHis-HopM301–712. (C) Western blot and reverse transcription polymerase chain reaction analyses of HopM1-dependent destabilization of AtMIN7 in Arabidopsis plants. Leaves of Col-0 gl1 plants were infiltrated with water or 1 × 108 CFUs per milliliter of ΔCEL mutant bacteria with or without pORF43 and harvested 10 hours later. The endogenous AtMIN7 protein—detected with the use of a rabbit polyclonal AtMIN7 antibody—was absent in leaves infiltrated with ΔCEL mutant bacteria (pORF43) that produce HopM1; however, the AtMIN7 transcript level was not reduced. (D) Proteasome inhibitors (MG132 and epoxomicin) blocked the HopM1-mediated destabilization of AtMIN7 in N. benthamiana leaves, whereas a cocktail of inhibitors of serine-, cysteine-, aspartic-, and metallo-proteases did not. AtMIN7::HA and 6xHis::HopM11–712 proteins were detected with HA and 6xHis epitope antibodies, respectively. (E) Detection of polyubiquitination of AtMIN7 in plants. AtMIN7::HA and 6xHis-HopM1–712 proteins were transiently coexpressed in 1% dimethyl sulfoxide (-)- or in MG132-treated N. benthamiana leaves. AtMIN7 was immunoprecipitated (IP) with the use of an epitope antibody to HA (27). Ubiquitinated proteins were detected by Western blot (WB) with a polyclonal ubiquitin (Ub) antibody (Sigma). HopM1 induced the polyubiquitination of AtMIN7::HA in vivo.

The failure to isolate interacting host proteins by means of full-length HopM1 was unexpected, but subsequent immunoblot analysis of yeast cells expressing each of the 21 AtMIN proteins revealed a notable result: eight AtMIN proteins (AtMIN2, 3, 4, 6, 7, 9, 10, and 11) (table S1) either disappeared or were present in much smaller amounts in yeast cells expressing full-length HopM1, compared with yeast cells expressing HopM11–300 (AtMIN2, 7, and 10) (fig. S2B). This result explains why these AtMIN proteins could not be isolated when full-length HopM1 was used in Y2H screening. The amounts of the remaining AtMIN proteins, most of which are predicted to be chloroplast or mitochondrial proteins, did not differ between the two yeast strains (AtMIN12 in fig. S2B). Because HopM1 is localized in the plant endomembranes (Fig. 1B) and does not contain any organelle-targeting signature sequences, we reasoned that organelle-targeted AtMINs identified in Y2H experiments would not be physiological targets.

The HopM1-dependent destabilization of AtMIN proteins was observed not only in yeast cells but also in N. benthamiana leaves transiently expressing HopM1 and AtMIN proteins (fig. S2C) and, most important, in Arabidopsis leaves during bacterial infection [see Fig. 2C for AtMIN7, which is a low-abundance ∼200-kilodalton (kD) protein in Arabidopsis]. Reduction of the AtMIN protein level (for example, AtMIN7) was not accompanied by a corresponding reduction of the AtMIN transcript level (Fig. 2C). Consistent with the specific dominant negative effects of HopM11–200 and HopM11–300 on HopM1, but not on AvrE (Fig. 1C and fig. S1), AvrE did not destabilize AtMIN proteins (AtMIN7 in fig. S2D). Subcellular fractionation analysis of AtMIN10-hemagglutinin (HA) transgenic plants, in which AtMIN10-HA is localized in both soluble and membrane fractions, showed that membrane-associated AtMIN10-HA was preferably destabilized by HopM1 during bacterial infection (fig. S2E). Collectively, these results suggest that HopM1 acts in the host endomembranes to destabilize AtMIN proteins by means of a posttranscriptional mechanism.

The HopM1-dependent destabilization of AtMIN proteins (such as AtMIN7) in N. benthamiana leaves was not affected by a cocktail of inhibitors of serine-, cysteine-, aspartic-, and metallo-proteases and aminopeptidases (Fig. 2D). However, proteasome inhibitors (e.g., MG132 and epoxomicin) completely blocked HopM1-mediated destabilization of AtMIN7 (Fig. 2D). Furthermore, immunoblot analysis of total leaf extract revealed that transiently expressed HopM1 greatly enhanced the protein ubiquitination in N. benthamiana leaves, as evidenced by the increased accumulation of the characteristic ubiqituin smear of >200 kD (Fig. 2E). Nonfunctional HopM11–300 and HopM1301–712 did not enhance protein ubiquitination (fig. S3). The polyubiquitinated AtMIN7-HA protein could be precipitated with either the HA epitope antibody or the polyclonal AtMIN7 antibody (Fig. 2E and fig. S3). Together with the observed physical interaction between the N terminus of HopM1 and the AtMIN7 protein in vivo (Fig. 2B), these results suggest a mechanism for HopM1 action: HopM1 recruits AtMIN7, or an AtMIN7-containing complex, via its N terminus and promotes the subsequent ubiquitination and degradation of AtMIN7 via the host proteasome.

To address the critical question of whether destabilization of a specific AtMIN protein or proteins is necessary for HopM1-mediated promotion of Pst DC3000 pathogenesis in Arabidopsis, we analyzed Arabidopsis SALK lines (18) carrying transferred DNA (T-DNA) insertions in each of the AtMIN genes listed in table S1. When infected with the ΔCEL mutant, all of the AtMIN knockout (KO) lines, except for the AtMIN7 KO line, restricted the growth of the ΔCEL mutant in a manner similar to the wild-type Col-0 plants. AtMIN7 KO plants (fig. S4) infected by the ΔCEL mutant showed increased bacterial multiplication and enhanced chlorotic and necrotic disease symptoms, compared with Col-0 plants (Fig. 3, A and B). AtMIN7 KO plants remained resistant to the nonpathogenic hrcC mutant bacteria and responded to Pst DC3000 similarly to wild-type Col-0 plants (Fig. 3, A and B). This result demonstrates that the increased susceptibility in AtMIN7 KO plants is specific to ΔCEL mutant bacteria, mirroring the results shown in Fig. 1A, and is therefore biologically relevant to the virulence function of HopM1. Thus, a host-target mutation specifically increased the virulence of a plant-pathogen mutant lacking the cognate effector protein.

Fig. 3.

Analysis of AtMIN7 knockout (KO) plants. (A) Growth of Pst DC3000, the ΔCEL mutant, and the hrcC mutant in AtMIN7 KO plants or in Col-0 plants. Plants were inoculated by dipping with 1 × 108 CFUs per milliliter of bacteria. Bacterial populations were determined at day 4. Two independent T-DNA insertion lines (fig. S4) were analyzed with similar results; results from line #1 are shown here. (B) Disease symptoms (chlorosis and necrosis) in Col-0 plants and AtMIN7 KO plants at day 4. (C) Effect of BFA treatment on bacterial multiplication.

AtMIN7 encodes one of the eight members of the Arabidopsis adenosine diphosphate (ADP) ribosylation factor (ARF) guanine nucleotide exchange factor (GEF) protein family (19) (fig. S5A). HopM1 did not act on all Arabidopsis ARF GEFs. For example, HopM1 interacted strongly with and destabilized only AtMIN7, but not At1g13980 or At4g35380 in yeast (fig. S5, B and C).

The ARF GEF proteins are key components of the vesicle trafficking system in eukaryotic cells and are the primary molecular targets of Brefeldin A (BFA), a well-known inhibitor of vesicle trafficking (20, 21). If the virulence function of HopM1 (i.e., destabilization of AtMIN7) is to inhibit host vesicle traffic, we reasoned that BFA treatment might substitute for HopM1 and restore the virulence of this bacterial mutant. Indeed, BFA treatment significantly enhanced the virulence (both bacterial multiplication and disease symptoms) of the ΔCEL mutant in wild-type Col-0 gl1 plants (Fig. 3C). Notably, the restoration of bacterial virulence by BFA was also specific to the ΔCEL mutant, because there were no significant differences in bacterial multiplication or disease symptoms caused by Pst DC3000 or the hrcC mutant in Col-0 gl1 plants treated with water or BFA (Fig. 3C). The restoration of the virulence of the ΔCEL mutant in BFA-treated leaves was even more complete than that in the AtMIN7 KO plants, suggesting that BFA and HopM1 target other host components in addition to AtMIN7.

The HopM1-mediated destruction of AtMIN7 and the ability of BFA to restore the virulence of the ΔCEL mutant suggest that HopM1 may be involved in the inhibition of a host vesicle trafficking pathway. Accelerated vesicle traffic is associated with a polarized cell wall–associated defense in plants (22, 23), and our previous study showed that a major function of HopM1 is the suppression of this defense (13). To assess the requirement of AtMIN7 for the cell wall–associated defense, we examined the callose deposition (a cellular marker of this defense) in the leaves of Col-0 and AtMIN7 KO plants infected by Pst DC3000 or the ΔCEL mutant. Col-0 leaves accumulated a high number of polarized callose deposits in response to the ΔCEL mutant, whereas Pst DC3000 suppressed callose deposition in Col-0 leaves (13) (Fig. 4). The leaves of AtMIN7 KO plants were reduced in the ability to mount an active polarized callose response to the ΔCEL mutant (i.e., the remaining callose deposition was mostly not polarized), whereas their response to Pst DC3000 was similar to that of Col-0 plants (Fig. 4). This result and the increased susceptibility of AtMIN7 KO plants to the ΔCEL mutant (Fig. 3) are both consistent with an active role of AtMIN7 in the host immune response.

Fig. 4.

Callose deposition in leaves of Col-0 and AtMIN7 KO plants. Arabidopsis Col-0 and AtMIN7 KO leaves (line #1) were stained to show callose deposition 7 hours after inoculation with 1 × 108 CFUs per milliliter of DC3000 and ΔCEL mutant bacteria. Average numbers of callose depositions per field of view (0.9 mm2) are presented with standard deviations displayed as error bars.

The HopM1-dependent destabilization of a host ARF GEF family protein via the host proteasome is interesting in light of the recent findings that a P. syringae effector protein, AvrPtoB, has intrinsic E3 ligase activity (24) and that vesicle trafficking and extracellular secretion play important roles in plant immune response (23, 25). HopM1 does not show any sequence similarity to AvrPtoB, nor does it contain any known structure motifs present in various components of the ubiquitination/proteasome system, including various types of E3 ligases. Therefore, HopM1 probably functions as an adaptor protein that targets AtMIN7 to the host ubiquitination/proteasome system.

The majority of plant pathogenic bacteria, including P. syringae, are extracellullar pathogens, living outside the plant cell wall. Our results suggest that P. syringae has evolved a mechanism to eliminate a component of a putative vesicle traffic pathway as an effective strategy of suppressing the extracellular cell wall–associated host defense (fig. S6). The intracellular human pathogen Salmonella enterica also uses effector proteins to interfere with host vesicle trafficking, although in this case the purpose is for the biogenesis and maintenance of a specialized membrane-bound compartment in which bacteria live (4, 26). Despite the difference, our results suggest that modulation of host vesicle trafficking serves a common final purpose for plant and human pathogens, creating a host environment favorable for bacterial survival and multiplication.

Supporting Online Material

www.sciencemag.org/cgi/content/full/313/5784/220/DC1

Materials and Methods

Figs. S1 to S6

Table S1

References

References and Notes

View Abstract

Navigate This Article