Suppression of the MicroRNA Pathway by Bacterial Effector Proteins

See allHide authors and affiliations

Science  15 Aug 2008:
Vol. 321, Issue 5891, pp. 964-967
DOI: 10.1126/science.1159505


Plants and animals sense pathogen-associated molecular patterns (PAMPs) and in turn differentially regulate a subset of microRNAs (miRNAs). However, the extent to which the miRNA pathway contributes to innate immunity remains unknown. Here, we show that miRNA-deficient mutants of Arabidopsis partly restore growth of a type III secretion-defective mutant of Pseudomonas syringae. These mutants also sustained growth of nonpathogenic Pseudomonas fluorescens and Escherichia coli strains, implicating miRNAs as key components of plant basal defense. Accordingly, we have identified P. syringae effectors that suppress transcriptional activation of some PAMP-responsive miRNAs or miRNA biogenesis, stability, or activity. These results provide evidence that, like viruses, bacteria have evolved to suppress RNA silencing to cause disease.

In RNA silencing, double-stranded RNA (dsRNA) is processed into small RNAs (sRNAs) through the action of RNase-III– like Dicer enzymes. The sRNAs guide Argonaute (AGO)–containing RNA-induced silencing complexes (RISCs) to inhibit gene expression at the transcriptional or posttranscriptional levels (1). In the Arabidopsis thaliana microRNA (miRNA) pathway, miRNA precursors (pre-miRNAs) are excised from noncoding primary transcripts (pri-miRNAs) and processed into mature miRNA duplexes by Dicer-like 1 (DCL1). Upon HEN1-catalyzed 2′-O-methylation (2), one miRNA strand incorporates an AGO1-containing RISC to direct endonucleolytic cleavage or translational repression of target transcripts (1). DCL4 and DCL2 perform major defensive functions by processing viral-derived dsRNA into small interfering RNAs (siRNAs), which, like miRNAs, are loaded into AGO1-RISC. As a counterdefensive strategy, viruses deploy viral suppressors of RNA silencing, or VSRs (3). RNA silencing also contributes to resistance against bacterial pathogens (47), which elicit an innate immune response upon perception of pathogen-associated molecular patterns (PAMPs) by host-encoded pattern recognition receptors (PRRs). For example, the Arabidopsis miR393 is PAMP-responsive (4, 8) and contributes to resistance against virulent Pseudomonas syringae pv. tomato strain DC3000 (Pto DC3000) (4). Nonetheless, the full extent to which cellular sRNAs, including miRNAs, participate in PAMP-triggered immunity (PTI) in plants remains unknown.

To address this issue, Arabidopsis mutants defective for siRNA or miRNA accumulation were challenged with Pto DC3000 hrcC, amutant that lacks a functional type III secretion system required for effector-protein delivery into host cells (9). This bacterium elicits but cannot suppress PTI and, consequently, multiplies poorly on wild-type Col-0– and La-er–inoculated leaves (Fig. 1A and fig. S1). However, Pto DC3000 hrcC growth was specifically enhanced in the miRNA-deficient dcl1-9 and hen1-1 mutants (Fig. 1A), in which induction of the basal defense marker gene WRKY30 was also compromised (fig. S2A) (10).

Fig. 1.

The Arabidopsis miRNA pathway promotes basal and nonhost resistances to bacteria. (A) Six-week-old plants were inoculated by syringe infiltration using a Pto DC3000 hrcC concentration of 106 colony-forming units (CFUs) per milliliter. Error bars indicate SE of log-transformed data from five independent samples. Similar results were obtained in three independent experiments. (B to D) Plants were inoculated as in (A) but with Psp NPS3121 (B), P. fluorescens Pf-5 (C), or E. coli W3110 (D). Similar results were obtained in two independent experiments. (E) Plants were inoculated as in (A) and pictures were taken at 6 days after inoculation.

Because PTI is also a major component of nonhost resistance (10, 11), we challenged dcl1-9 and hen1-1 mutants with P. syringae pv. phaseolicola (Psp) strain NPS3121, which infects beans but not Arabidopsis. Both dcl1-9 and hen1-1 mutants sustained Psp NPS3121 growth (Fig. 1B) and displayed compromised WRKY30 induction (fig. S2B). Enhanced bacterial growth was also observed with the nonpathogenic Pseudomonas fluorescens Pf-5 and Escherichia coli W3110 strains (Fig. 1, C and D). Furthermore, the above nonvirulent bacteria all induced chlorosis and necrosis on miRNA-deficient mutants, resembling bacterial disease symptoms triggered by virulent Pto DC3000 (Fig. 1E and fig. S3, A to D). However, we cannot exclude the participation of other endogenous sRNAs in this process (6, 7), because the hen1-1 mutant, which is additionally impaired in the accumulation of many types of sRNAs (12), consistently displayed more disease symptoms than the dcl1-9 mutant did (fig. S3). Despite this possibility, our results indicate that the miRNA pathway is probably an essential component of plant basal defense. As a corollary, some bacterial effectors must have evolved to suppress host-miRNA functions to enable disease.

In principle, suppression of the miRNA pathway could affect miRNA transcription, biogenesis, or activity. To test the first possibility, we challenged wild-type plants with Pto DC3000 or Pto DC3000 hrcC and analyzed the levels of several pri-miRNAs. In virulent Pto DC3000–treated plants, induction of the PAMP-responsive pri-miR393a/b and pri-miR396b (4, 8) was appreciably suppressed, as was induction of WRKY30 and Flagellin Receptor Kinase 1 (FRK1) (13), which were used as internal controls (Fig. 2A). In contrast, the PAMP-insensitive pri-miR166a and pri-miR173 remained unaffected. We then used the previously described At-miR393a-p::eGFP and At-miR393b-p::eGFP transgenic lines, which report At-miR393a and At-miR393b transcriptional activity (4). Pto DC3000 hrcC caused an increase in enhanced green fluorescent protein (eGFP) mRNA levels in both transgenic lines, indicating the presence of PAMP-responsive elements upstream of At-miR393a and At-miR393b (Fig. 2B). However, this induction was compromised by Pto DC3000, as was induction of the FRK1 control (Fig. 2B). These results suggest that some Pto DC3000 effectors suppress PAMP-triggered transcriptional activation of At-miR393a and At-miR393b.

Fig. 2.

Transcriptional repression of PAMP-responsive miRNAs by Pto DC3000 and AvrPtoB. (A) Five-week-old plants were syringe-infiltrated with a concentration of 2 × 107 CFU/ml of either Pto DC3000 or Pto DC3000 hrcC and pri-miRNA expression was monitored by semiquantitative reverse transcription polymerase chain reaction (RT-PCR). NT, nontreated. Similar results were obtained in three independent experiments. (B) Five-week-old miR393a-p::eGFP and miR393b-p::eGFP transgenic lines were treated as in (A) for 6 hours. eGFP and FRK1 mRNA levels were analyzed by quantitative RT-PCR. Similar results were obtained in three independent experiments. (C) Semiquantitative RT-PCR analysis of pri-miRNAs. Agrobacterium-mediated transient expression was performed in 4-week-old efr leaves. Similar results were obtained in three independent experiments. GUS, β-glucuronidase. (D) Detection of AvrPtoB-3xHA and AvrPtoBF525A-3xHA by Western analysis. In the bottom panel, Coomassie staining shows equal protein loading. Transient expression was performed as in (C). (E) Transient expression was performed in 4-week-old At-miR393a-p::eGFP/efr and At-miR393b-p::eGFP/efr lines as in (C). Expression of eGFP was monitored by quantitative RT-PCR. (F) As in (E), except that 3 days after Agrobacterium-infiltration, plants were challenged with 1 μM active or inactive flg22 for 6 hours.

To test this hypothesis, we engineered constructs to deliver distinct Pto DC3000 effectors into leaves of the Arabidopsis efr mutant, which sustains efficient Agrobacterium-mediated transient transformation (14) without affecting pri-miRNA, mature miRNA, or miRNA target levels (fig. S4). Bacterial effector expression was confirmed (fig. S5A), and pri-miRNA levels were subsequently monitored. AvrPtoB, an effector with E3-ubiquitin ligase activity (15), down-regulated pri-miR393a and pri-miR393b accumulation without affecting the PAMP-insensitive pri-miR166a (Fig. 2C). This effect occurs, at least in part, at the transcriptional level because AvrPtoB delivery into At-miR393a-p::eGFP/efr and At-miR393b-p::eGFP/efr leaves inhibited both basal expression and PAMP-triggered induction of eGFP (Fig. 2, E and F, and fig. S5B). Similar effects were obtained with AvrPtoBF525A, a stable mutant in which the E3-ligase activity is abolished (Fig. 2, D to F, and fig. S5B). Therefore, AvrPtoB suppresses At-miR393a and At-miR393b transcription independently of its E3-ligase activity, as was previously shown for AvrPtoB-mediated suppression of PAMP-responsive FRK1 (16).

To assay for interference with miRNA biogenesis or stability, levels of several PAMP-insensitive and -sensitive miRNAs were monitored upon transient effector delivery into efr leaves. Three bacterial effectors consistently reduced the accumulation of unrelated miRNAs (Fig. 3A and figs. S5, A to C, and S6), among which AvrPto is a well-characterized Pto DC3000 effector with demonstrated virulence function (fig. S9) (17). Further molecular analysis revealed that AvrPto-mediated reduction in miRNA accumulation may occur, at least in part, at the posttranscriptional level, because AvrPto did not alter pri-miRNA transcript levels (fig. S7A). Accordingly, conditional AvrPto expression in stable transgenic lines stabilizes miR393 precursors and concomitantly decreases accumulation of mature miR393, with no or little effects on pri-miR393 transcript levels (Fig. 3, B and C, and fig. S7B). Therefore, AvrPto possibly interferes with the processing of some miRNA precursors, a phenomenon also observed during Pto DC3000 infection (Fig. 3C and fig. S8).

Fig. 3.

Suppression of miRNA accumulation or activity by bacterial effectors. (A) Four-week-old efr plants were syringe-infiltrated with A. tumefaciens carrying AvrPto, AvrPtoY89D, AvrPtoG2A, or GUS-intron constructs. Five days after infiltration, accumulation of miRNAs was assayed by Northern analysis. U6, the control, shows sRNA equal loading. Similar results were obtained in two independent experiments. (B) Same experiments as in (A) but in Dex::AvrPto and Dex::6xHis-AvrPto transgenic lines. miRNA accumulation was analyzed by Northern analysis 30 hours after dexamethasone (Dex) treatment (at a concentration of 30 μM). Similar results were obtained in two independent experiments. (C) Left panel shows the same experiment as in (B) but on miR393. Right panel shows miR393 levels 30 hours after inoculation with Pto DC3000 hrcC or Pto DC3000 at 108 CFU/ml. Inoculation was performed by syringe infiltration on 5-week-old plants. Similar results were obtained in five independent experiments. Arrows: miR393 precursors stabilization (pre-miR393 stabilization was detected by 24 hours after inoculation). (D) Northern analysis of SUL siRNA levels in SS plants overexpressing HopT1-1. (E) Four-week-old SS plant (left) and HopT1-1 overexpressing (SS) plants (middle and right). (F) Northern analysis of canonical miRNA levels in HopT1-1–overexpressing (SS) plants. (G) Quantitative RT-PCR analysis of SUL and miRNA target transcript levels in HopT1-1–overexpressing SS plants. (H) HopT1-1 and Pto DC3000 elevate CIP4 protein levels. Upper panels show CIP4 protein levels upon transient delivery of HopT1-1 (T1) or HopC1 (C1) in 4-week-old efr leaves. Western analysis was performed using an antibody to CIP4. Ponceau staining shows equal protein loading. Bottom panels show CIP4 protein levels after inoculation with Pto DC3000 or Pto DC3000 hrcC at 108 CFU/ml. Inoculation was performed by syringe infiltration on 5-week-old Col-0 plants. CIP4 protein levels were monitored 6 hours after inoculation by Western analysis. Coomassie staining shows equal protein loading. Similar results were obtained in two independent experiments.

AvrPto interacts with, and inhibits the kinase activity of, multiple transmembrane PRRs (18). Moreover, AvrPto strongly interacts with BAK1 (BRI1-associated receptor-kinase 1), a shared signaling partner of the brassinosteroid receptor BRI1 (brassinosteroid-insensitive 1) and the flagellin receptor FLS2 (flagellin sensing 2) (1921). The AvrPto-BAK1 interaction compromises the ligand-dependent FLS2-BAK1 association resulting in suppression of PTI (19). Accordingly, AvrPtoY89D, which is unable to interact with PRRs and BAK1 (18, 19), displays compromised virulence function (18). Transient delivery of AvrPtoY89D did not alter miRNA accumulation, nor did delivery of AvrPtoG2A, carrying a mutated myristoylation site that disrupts AvrPto host plasma membrane localization (Fig. 3A and fig. S5C) (16, 22). Conditional expression of an N-terminal histidine-tagged version of AvrPto (6xHis-AvrPto) displaying similarly compromised subcellular localization and virulence function gave comparable results (Fig. 3, B and C, and fig. S9). We conclude that AvrPto interferes with miRNA accumulation and that this interference is linked to its virulence function [supporting online material (SOM) text].

Finally, we tested whether Pto DC3000 effectors could also suppress miRNA activity. We used the SUC-SUL (SS) reporter line in which phloem-specific expression of an inverted-repeat transgene triggers non–cell-autonomous RNA interference (RNAi) of the endogenous SULPHUR (SUL) transcript, causing vein-centered chlorosis (23). Of the 21-nucleotide (nt) (DCL4-dependent) and 24-nt (DCL3-dependent) SUL siRNA species, only the former is required for RNAi in an AGO1-dependent manner (23). Stable transgenic lines overexpressing HopT1-1 (fig. S10) exhibited attenuated chlorosis (Fig. 3E) and accumulated higher SUL mRNA levels (Fig. 3G). However, the SUL siRNA levels remained unchanged (Fig. 3D), mimicking the reported effects of the ago1-12 mutation in SS (23). Also as in ago1-12, canonical miRNAs accumulated normally in HopT1-1–overexpressing lines, despite higher levels of miRNA target transcripts (Fig. 3, F and G, and fig. S11), suggesting that HopT1-1 probably suppresses slicing mediated by AGO1. Further transient overexpression of HopT1-1 in efr plants showed a dramatic increase in the protein, but not mRNA, levels of the miR834 target COP1-interacting protein 4 (CIP4) (Fig. 3H and fig. S12, A and B). Thus, HopT1-1 additionally, and perhaps predominantly, suppresses miRNA-directed translational inhibition, which is consistent with the involvement of AGO1 in this second process (24). Similarly, higher protein levels of CIP4 and of the copper/zinc superoxide dismutase 1 (CSD1-miR398 target) were detected in plants infected with virulent Pto DC3000 (Fig. 3H and fig. S12C), with no effect on CSD1, CIP4, and some other miRNA target transcript levels (fig. S13).

We show here that the miRNA pathway plays a major role in antibacterial basal defense and, accordingly, we have identified bacterial suppressors of RNA silencing, or BSRs. This finding provides a plausible explanation for the synergistic interactions observed in the field between some viral and bacterial phytopathogens. Consistent with this idea, we found that infection by Turnip Mosaic Virus (TuMV), which produces the P1-HcPro suppressor of siRNA and miRNA functions (25, 26), reduces basal and nonhost resistances to promote growth and diseaselike symptoms from nonvirulent Pto DC3000 hrcC and Psp NPS3121 bacteria (Fig. 4). It will now be important to elucidate how silencing factors are modified by VSRs and BSRs, and whether such modifications are sensed by specific resistance (R) proteins as postulated by the “guard hypothesis” (27).

Fig. 4.

TuMV infection enhances growth of and rescues symptoms induced by Pto DC3000 hrcC and Psp NPS3121. (A) Five-week-old Col-0 plants were infected for 7 days with TuMV and further challenged with Pto DC3000 hrcC at a concentration of 106 CFU/ml. The picture was taken 5 days after bacterial infiltration. (B) Same as in (A), but with Psp NPS3121. (C) Growth of Pto DC3000 hrcC in mock-inoculated or TuMV-infected Col-0 plants 6 days after bacterial injection. Error bars indicate SE of log-transformed data from five independent samples. Similar results were obtained in two independent experiments. (D) Same as in (C), but with Psp NPS3121.

The implication of the miRNA pathway in innate immunity is not specific to plants. For example, human MiR146 is induced by several microbial components (28). Because type III secretion systems are widespread across Gram-negative bacteria (29), the intriguing possibility emerges that human pathogenic bacteria also have evolved to suppress RNA silencing to cause disease.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S13


References and Notes

View Abstract

Navigate This Article