A pathogen-inducible endogenous siRNAin plant immunitySurekha Katiyar-Agarwal*, Rebekah Morgan*, Douglas Dahlbeck†, Omar Borsani‡, Andy Villegas, Jr.*, Jian-Kang Zhu‡,Brian J. Staskawicz†§, and Hailing Jin*§

Departments of *Plant Pathology and ‡Botany and Plant Sciences, Center for Plant Cell Biology and Institute for Integrative Genome Biology,University of California, Riverside, CA 92521; and †Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720

Contributed by Brian J. Staskawicz, September 19, 2006

RNA interference, mediated by small interfering RNAs (siRNAs), isa conserved regulatory process that has evolved as an antiviraldefense mechanism in plants and animals. It is not known whetherhost cells also use siRNAs as an antibacterial defense mechanism ineukaryotes. Here, we report the discovery of an endogenoussiRNA, nat-siRNAATGB2, that is specifically induced by the bacterialpathogen Pseudomonas syringae carrying effector avrRpt2. Wedemonstrate that the biogenesis of this siRNA requires DCL1, HYL1,HEN1, RDR6, NRPD1A, and SGS3. Its induction also depends on thecognate host disease resistance gene RPS2 and the NDR1 gene thatis required for RPS2-specified resistance. This siRNA contributes toRPS2-mediated race-specific disease resistance by repressing PPRL,a putative negative regulator of the RPS2 resistance pathway.

antibacterial defense � DCL1 � RDR6 � RPS2-specific

Endogenous small interfering RNAs (siRNAs) and microR-NAs (miRNAs) have emerged as important regulators of

eukaryotic gene expression by guiding mRNA cleavage, trans-lational inhibition, or chromatin modification (1, 2). In Arabi-dopsis, �100 miRNAs have been reported and shown to beimportant for plant development (3, 4) and abiotic stress toler-ance (5–7). One miRNA was recently shown to contribute tobasal defense against bacteria by regulating auxin signaling (8).In contrast to the relatively limited number of miRNAs, thou-sands of endogenous siRNAs have been sequenced (6, 9–11).However, their biological roles are largely unknown except forthe functions of transacting siRNAs (ta-siRNA) in plant devel-opment and hormone signaling (4) and the roles of somechromatin-associated siRNAs in DNA methylation and tran-scriptional gene silencing (4). Borsani et al. (12) recently dis-covered a new class of endogenous siRNAs derived from theoverlapping region of a pair of natural antisense transcripts(NATs) (12). These so-called nat-siRNAs regulate salt stressresponse in Arabidopsis (12). Despite large intergenic spaces, asignificant proportion of eukaryotic genomes are arranged asNATs (13). More than 1,000 pairs of NATs exist in Arabidopsis(14, 15). Our analysis of transcript profiles from an Arabidopsismicroarray database (16) has revealed that, in many cases, onetranscript of a NAT pair is specifically induced under certainabiotic or biotic conditions. The induced transcript may pair withthe existing antisense transcript and trigger the nat-siRNAformation, resulting in the silencing of the antisense transcript incis or other homologous loci in trans. NATs may serve as one ofthe major sources of endogenous siRNAs for gene regulation inresponse to different environmental conditions. This hypothesisis well supported by the presence of �100 nat-siRNAs in theMassively Parallel Signature Sequencing (MPSS) and Arabidop-sis Small RNA Project (ASRP) databases (9, 17). In this study,we identified a nat-siRNA that is specifically induced by thebacterial pathogen Pseudomonas syringae (Ps) carrying effectoravrRpt2 (18). We demonstrate that its induction depends on anovel biogenesis pathway that requires the cognate host diseaseresistance (R) gene RPS2 (19) and the NDR1 (20) gene that is

also required for RPS2-specified resistance. This siRNA re-presses a negative regulator of the RPS2 resistance pathway.

ResultsInduction of a nat-siRNA by Bacterial Pathogen Ps Carrying avrRpt2.Pathogen effectors can be recognized by R proteins and cantrigger a series of disease resistance responses, including acti-vating and repressing a large array of genes (21). To addresswhether endogenous siRNAs play a role in gene expressionreprogramming in R gene-mediated disease resistance, wesearched the small RNA databases (9, 17) and examined nat-siRNAs generated from NAT pairs that are potentially regulatedby bacterial pathogenesis. Excitingly, we discovered that a 22-ntnat-siRNA (ASRP1957), derived from the overlapping region ofa Rab2-like small GTP-binding protein gene (ATGB2,At4g35860) and a PPR (pentatricopeptide repeats) protein-likegene (PPRL, At4g35850), is strongly induced by Ps pathovartomato (Pst) carrying avirulence (avr) gene avrRpt2 but notavrRpm1, avrRps4, or avrPphB (Fig. 1A). We named it nat-siRNAATGB2. We used Pst strain DC3000 for all of theexperiments in this study.

The nat-siRNAATGB2 sequence is complementary to the 3�UTR region of the antisense gene PPRL and thus could poten-tially induce silencing of PPRL. We examined the expression ofPPRL as well as the sense gene ATGB2 upon Pst challenge. TheATGB2 transcript was strongly induced by both Pst (avrRpt2) andPst (avrRpm1) (Fig. 1B), whereas the PPRL mRNA was sub-stantially down regulated only by Pst (avrRpt2) infection wherethe nat-siRNAATGB2 was strongly induced (Fig. 1 A and C).The result suggests that down-regulation of PPRL depends onthe induction of the nat-siRNAATGB2 and the induction ofATGB2 alone is not sufficient for inducing nat-siRNAATGB2.To determine whether the induction of ATGB2 is necessary forinducing nat-siRNAATGB2, we obtained a T-DNA insertionline (Salk�083103) of ATGB2 from the Salk collection (22). Thehomozygous line is a partial knock-down mutant with T-DNAinserted in the 3rd intron (Fig. 1D). We detected less inductionof nat-siRNAATGB2 and less repression of PPRL mRNAexpression in this knock-down line than in the WT plants afterPst (avrRpt2) challenge (Fig. 1 D and E). These results suggestthat the induction of sense transcript ATGB2 is necessary but notsufficient for nat-siRNAATGB2 accumulation (Fig. 1 A, B, andD), implying that the induction of this siRNA is under multiple

Author contributions: H.J. designed research; S.K.-A., R.M., D.D., O.B., A.V., and H.J.performed research; J.-K.Z., B.J.S., and H.J. analyzed data; and S.K.-A. and H.J. wrote thepaper.

The authors declare no conflict of interest.

Abbreviations: miRNA, microRNA; ta-siRNA, transacting siRNA; NAT, natural antisensetranscript; Ps, Pseudomonas syringae; Pst, Ps pathovar tomato; avr, avirulence; DCL,DICER-like; RDR, RNA-dependent RNA polymerase; Dex, dexamethasone; hpi, h postinocu-lation; dpi, days postinoculation; PPR, pentatricopeptide repeat; PPRL, PPR protein-like;HR, hypersensitive responses.

§To whom correspondence may be addressed. E-mail: hailingj@ucr.edu or stask@nature.berkeley.edu.

© 2006 by The National Academy of Sciences of the USA

18002–18007 � PNAS � November 21, 2006 � vol. 103 � no. 47 www.pnas.org�cgi�doi�10.1073�pnas.0608258103

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0

layers of control and requires other factors associated with Pst(avrRpt2) infection.

Biogenesis of nat-siRNAATGB2. To define the components requiredfor its biogenesis, we examined nat-siRNAATGB2 in Pst (avr-Rpt2)-challenged small RNA biogenesis mutants and the corre-sponding WT ecotypes. The Arabidopsis genome has four DI-CER-like (DCL) proteins (23). Interestingly, the induction ofnat-siRNAATGB2 could be detected in dcl2-1, dcl3-1, anddcl4-2 mutants but not in the dcl1-9 mutant (Fig. 2A). The resultindicates that the miRNA biogenesis component DCL1 is re-quired for the formation of nat-siRNAATGB2. This observationdiffers from the biogenesis of the 24-nt nat-siRNASRO5 thatrequires DCL2, 21-nt nat-siRNAs that require both DCL1 andDCL2 (12), and ta-siRNAs that require DCL1 and DCL4 (24).We did not detect any other siRNAs generated from theoverlapping region of PPRL and ATGB2 or siRNA that iscomplementary to nat-siRNAATGB2 (data not shown). Thus,this nat-siRNA is generated from a specific site of the overlap-ping region of PPRL and ATGB2 transcripts and is strand-specific.

Mutations in the dsRNA-binding protein HYL1 and RNA-dependent RNA polymerase (RDR) 6 also totally blocked theaccumulation of nat-siRNAATGB2, whereas a mutation inRDR2 had no effect (Fig. 2 A). RDR6 is required for virus-induced gene silencing, transgene silencing, and ta-siRNA pro-duction (25). HYL1 has been indicated to interact with DCL1

(26) and affects the accumulation of several miRNAs (27) andta-siRNAs (24). The level of nat-siRNAATGB2 was reduced insgs3, the RNA methyltransferase mutant hen1, and the RNApolymerase IVa mutant nrpd1a (Fig. 2 A), which is similar to thatof salt-induced nat-siRNAs (12). These results suggest a biogen-esis pathway for nat-siRNAATGB2 in which nat-siRNAATGB2is processed by the DCL1-HYL1 complex, stabilized by HEN1-mediated methylation, and amplified by RDR6-, SGS3-, andRNA polymerase IVa-mediated reactions.

To confirm that the down-regulation of PPRL depends on theinduction of nat-siRNAATGB2, we examined the PPRL mRNAlevel in the mutants that failed to generate nat-siRNAATGB2upon Pst (avrRpt2) infection. The down-regulation of PPRL wasabolished in dcl1-9, hyl1, and rdr6 compared with WT Landsbergerecta (Ler), Nossen-0 (No) and C24, respectively (Fig. 2B). Thesgs3 and nrpd1a mutants, where the accumulation of nat-siRNAATGB2 is significantly reduced, also accumulate 2- to3-fold more PPRL mRNA than that in their correspondingcontrols (Fig. 2B). Thus, the down-regulation of PPRL is me-diated by nat-siRNAATGB2.

To test whether the overlapping region is sufficient for gen-erating nat-siRNAATGB2, the full-length or overlapping regionof ATGB2 was coexpressed with PPRL transiently in Nicotianabenthamiana leaves. Flag-tagged PPRL cDNA with its 3� UTRwas cloned into a binary vector driven by the CaMV 35Spromoter. Full-length (F) or only the overlapping region (O) ofATGB2 cDNA was cloned into the inducible expression vector

Fig. 1. A nat-siRNA is induced by Pst (avrRpt2). (A) Detection of the nat-siRNA by Northern blot analysis. The nat-siRNA sequence is shown under the panel.Small RNA was extracted from the leaves harvested at 15 hpi of Pst (2 � 107 cfu�ml) carrying various avr genes and an oligonucleotide probe complementaryto the siRNA was used. DNA probe was used for detecting U6 RNA for measuring the relative abundance (RA) (shown below). Ethidium bromide-stained tRNAis also shown as a loading control. (B and C) Relative expression levels of ATGB2 (1B) and PPRL (1C) as measured by real-time RT-PCR. The expression levels ofAtGB2 and PPRL were normalized to that of ubiquitin. (D) Northern blot analysis shows reduced ATGB2 and nat-siRNAATGB2 expression levels in Salk�083103homozygous line upon Pst (avrRpt2) (2 � 107 cfu�ml) challenge comparing to the WT Col-0. The levels of actin and U6 were used for quantification and loadingcontrols. (E) Relative mRNA level of PPRL in Salk�083103 homozygous line. The expression levels in untreated WT Col-0 were used as 100 % and standard deviationswere plotted from three replicates (B, C, and E).

Katiyar-Agarwal et al. PNAS � November 21, 2006 � vol. 103 � no. 47 � 18003

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0

pTA7002 (28). Substantial down-regulation of PPRL was ob-served at both the RNA and protein levels after induction ofeither full length (F) or only the overlapping region of ATGB2(O) by dexamethasone (Dex) (Fig. 2C), as was the induction ofnat-siRNAATGB2. The results suggest that the overlappingregion alone is sufficient to give rise to the nat-siRNA and toinduce antisense gene silencing.

Induction of nat-siRNAATGB2 Depends on RPS2 and NDR1. Pathogen-derived effectors are recognized directly or indirectly by specific

plant R proteins and trigger rapid race-specific resistance re-sponses. The effector avrRpt2 of Ps is specifically recognized bythe coiled-coil NBS-LRR type R protein RPS2 (19) and triggersa series of resistance responses, including the generation ofreactive oxygen species, reprogramming of gene expression, andinduction of hypersensitive responses (HR), which limit bacterialgrowth. The specific induction of nat-siRNAATGB2 by avrRpt2implies a functional involvement of the siRNA in RPS2-mediatedrace-specific disease resistance. To further understand the reg-ulation of nat-siRNAATGB2 by pathogen infection, we exam-

Fig. 3. Accumulation of nat-siRNAATGB2 is controlled by RPS2 and some components of the disease resistance signaling pathway. Northern blot analysis ofnat-siRNAATGB2 (A) and ATGB2 (B) was performed on Pst (avrRpt2)-treated defense-signaling mutants and WT Col-0 plants. U6 RNA was used for small RNAquantification. (C) Relative quantification of PPRL expression in defense signaling mutants by real-time RT-PCR analysis. PPRL expression level was normalized to thatof ubiquitin. The expression levels in untreated WT Col-0 were used as 100%. Standard deviations were plotted from three replicates.

Fig. 2. Accumulation of nat-siRNAATGB2 depends on DCL1, HYL1, and RDR6, and also requires HEN1, NRPD1a, and SGS3. (A) Northern blot analysis ofnat-siRNAATGB2 in various Pst (avrRpt2)-treated small RNA biogenesis mutants and their corresponding WT controls. MiR171 and U6 RNA was used as controls.U6 level was used for quantification. (B) Relative PPRL mRNA levels in sgs3, dcl1-9, hyl1, rdr6, nrpd1a, and their corresponding WT controls after Pst (avrRpt2)infection. The expression levels were normalized to that of ubiquitin. The expression level in untreated WT Col-0 was used as 100%. Standard deviations wereplotted from three replicates. (C) Transient coexpression of PPRL and ATGB2 in N. benthamiana. Agrobacterium GV3101 harboring PPRL was coinfiltrated withGV3101 carrying full-length (F) or only the overlapping region (O) of ATGB2 constructs into 3-week-old N. benthamiana leaves. The expression of AtGB2 wasinduced by Dex at 2 dpi, and tissue was harvested after 24 h of induction. The expression of PPRL was measured at both RNA and protein levels by semiquantitativeRT-PCR (Top) and Western blot analysis (Sigma anti-FLAG, 1:2,000 dilution, Middle), respectively. Actin was used as a control for RT-PCR. The Rubisco large subunitfrom a gel that was run in parallel was stained with Coomassie blue for Western blot loading control. nat-siRNAATGB2 is detected after the induction of eitherfull-length or overlapping region of ATGB2. (Bottom) tRNA was used as a loading control.

18004 � www.pnas.org�cgi�doi�10.1073�pnas.0608258103 Katiyar-Agarwal et al.

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0

ined its accumulation in various mutants of resistance signalingcomponents. Although the induction level of ATGB2 by avrRpt2was not substantially different in these mutants, the accumula-tion of nat-siRNAATGB2 differed considerably (Fig. 3 A and B).nat-siRNAATGB2 was not detected in rps2 (101C) and ndr1mutants, which indicates that nat-siRNAATGB2 induction re-quires the functional resistance protein RPS2 and NDR1, bothof which are required for avrRpt2-induced resistance (20). Mu-tations in other resistance signaling components, SCF ubiquitinligase complex component SGT1b, systemic acquired resistancesignaling component NPR1, and ethylene signaling componentEIN2 also reduced the level of nat-siRNAATGB2. Consistentwith the silencing of PPRL by nat-siRNAATGB2, the mutantswith no or reduced levels of nat-siRNAATGB2 showed no orless suppression of PPRL expression (Fig. 3C). These mutationshad no effect on the accumulation of miR173, which demon-strates a specific biogenesis regulation of nat-siRNAATGB2 bythe disease resistance signaling pathways. The jasmonic acid(JA) signaling mutant jar1, salicylic acid (SA) signaling mutantpad4, SA biosynthesis mutant eds16, and SGT1b homologueSGT1a had no effect on nat-siRNAATGB2 accumulation (Fig.3A). These results suggest that some components in basaldefense and ethylene signaling may interfere with nat-siRNAATGB2 regulation.

PPRL Acts as a Negative Regulator of RPS2 Resistance Pathway. Basedon the down-regulation of PPRL by nat-siRNAATGB2 in re-

sponse to Pst (avrRpt2) challenge, we hypothesized that PPRLmay negatively regulate RPS2-mediated resistance. PPRL is anatypical PPR protein with an unknown function and is localizedin mitochondria (29). To assess its function in disease resistance,we isolated T-DNA insertion lines of PPRL (Salk�013843 andSalk�071137) from the Salk collection (22) and also generatedPPRL cDNA-Flag (without UTR) overexpression lines (Fig. 4A)for loss- and gain-of-function studies. Complete knockout ofPPRL expression may lead to enhanced disease resistance toavrRpt2 because the PPRL gene is silenced after pathogeninfection in the WT resistance plants. No difference was ob-served in the growth of both virulent Pst (EV) and avirulent Pst(avrRpt2) between PPRL knockout lines and the WT control(data not shown). It is likely that the possible enhanced resis-tance was masked by the existing strong resistance to avrRpt2and, therefore, was difficult to score. However, when PPRL-Flagoverexpression plants were inoculated with a high concentration(1 � 107 cfu�ml) of Pst (avrRpt2), delayed HR was observed (Fig.4B) and the transgenic plants displayed considerably less elec-trolyte leakage at 24 h postinoculation (hpi) (Fig. 4C), whichindicates a reduced level of cell death in the overexpressionplants. Bacterial growth was measured on the plants infectedwith a low concentration (2 � 105 cfu�ml) of Pst carrying EV,avrRpm1, or avrRpt2. The Pst (avrRpt2) titer of the overexpres-sion line 32 containing a high level of PPRL was �6- to 8-foldhigher than that of the WT at 4 days postinoculation (dpi). Line

Fig. 4. Overexpression of PPRL attenuates RPS2-mediated resistance in Arabidopsis plants. (A) Western blot analysis of transgenic Arabidopsis plantsoverexpressing PPRL (Sigma anti-FLAG, 1:2,000 dilution). Shown is the Rubisco large subunit from a gel that was run in parallel and stained with Coomassie blue.Two lines with high (line 32) or low (line 33) expression level of PPRL were selected for phenotypic analysis. (B) PPRL overexpression line displays delayed HR.Picture of line 32 was taken at 16 hpi of Pst (avrRpt2) (2 � 107 cfu�ml). (C) PPRL overexpression lines exhibit reduced electrolyte leakage. Plants treated with 10mM MgCl2 and Pst (avrRpt2) (1 � 107 cfu�ml) were measured at 0 and 24 hpi. Error bars represent standard deviation of four replicates. Similar results wereobtained from two independent experiments. (D) PPRL overexpression lines display enhanced pathogen growth of Pst (avrRpt2). Bacterial growth was measuredat 0 and 4 dpi of Pst carrying EV, avrRpt2 or avrRpm1 (2 � 105 cfu�ml). Error bars represent standard deviation of five replicates. Similar results were obtainedin three independent experiments. (E) Pst (avrRpt2) accumulates to a higher level in rdr6 and hyl1, but not in dcl3, as compared with that in the correspondingWT C24, No, or Col-0, respectively. No difference was observed in the growth of Pst (avrRpm1) between the mutants rdr6, hyl1, or dcl3 and their correspondingWT control. Pathogen growth was measured at 0 and 4 dpi. Similar results were obtained in two independent experiments.

Katiyar-Agarwal et al. PNAS � November 21, 2006 � vol. 103 � no. 47 � 18005

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0

33, with a low level of PPRL overexpression, had about a 4- to5-fold increase in Pst (avrRpt2) bacterial growth than that in theWT (Fig. 4D). No difference was observed in the growth of Pst(EV) or Pst (avrRpm1) between PPRL overexpression line andWT control. These results show that overexpression of PPRLattenuates RPS2-mediated disease resistance and suggest thatPPRL may function as a negative regulator of the RPS2 pathway.

The induction of nat-siRNAATGB2 is blocked in dcl1-9, hyl1,and rdr6. Because the dcl1-9 mutation has strong pleiotropicphenotypes, we chose to examine bacterial growth in rdr6 andhyl1 mutants. An 8-fold increase of Pst (avrRpt2) bacterialgrowth was observed in rdr6 at 4 dpi compared with that in theWT C24 plants, whereas no significant difference in Pst (avr-Rpm1) growth was detected (Fig. 4E). Most strikingly, weobserved a complete loss of RPS2-mediated resistance in hyl1,whereas RPM1-mediated disease resistance was not affected(Fig. 4E). hyl1 may affect the biogenesis of an array of smallRNAs induced by Pst (avrRpt2), and the elimination of nat-siRNAATGB2 in hyl1 may contribute a portion of the observedpathogen susceptibility phenotype. As expected, we did notdetect any obvious difference in pathogen growth between Col-0WT control and dcl3-1 mutant (Fig. 4E), which affects theaccumulation only of siRNAs associated with chromatin modi-fication, but not nat-siRNAATGB2 (Fig. 2 A). No difference inpathogen growth was observed between the rdr6, hyl1, or dcl3mutants and their corresponding WT plants after Pst (avrRpm1)inoculation (Fig. 4E). Thus, RDR6 and HYL1 play critical rolesin RPS2-mediated resistance pathway by controlling the biogen-esis of nat-siRNAATGB2 and possibly of other endogenoussiRNAs.

DiscussionHere we identified an endogenous siRNA, nat-siRNAATGB2,which is specifically induced by Pst (avrRpt2). This nat-siRNA isproduced by a unique biogenesis pathway that requires DCL1,HYL1, HEN1, RDR6, SGS3, and RNA polymerase IVa (Fig. 5).Its formation not only requires the induction of the sensetranscript ATGB2, but also depends on the host resistance geneRPS2 and its resistance signaling components, including NDR1(Fig. 5). The biogenesis pathway of this 22-nt nat-siRNA suggestsan intricate regulation of endogenous siRNA formation. Thespecific induction of nat-siRNAATGB2 leads to the silencing ofthe antisense gene PPRL. Our results suggest that PPRL is anegative regulator of RPS2 signaling pathway and silencing ofPPRL by nat-siRNAATGB2 plays a positive role in diseaseresistance. More than 450 PPR proteins, characterized by thepresence of tandem pentatricopeptide repeats, exist in Arabi-dopsis and the majority of them have unknown functions (30). A

few studies point to an involvement of PPR proteins in post-transcriptional processes mainly in organelles, including RNAediting (31), mRNA silencing by cleavage (32) and translationalregulation (33), etc. The PPRL protein contains five atypicalPPR motifs and is mitochondrial localized (29, 30). Mitochon-drion is the major organelle involved in oxidative burst andhypersensitive responses in plant disease resistance and leads tolocal cell death. How these events are regulated and how thesignal is transduced are still largely unknown. A recent studyshows that a mitochondrial-localized PPR-containing proteininteracts with inhibitor of apoptosis proteins and regulatescaspase activity and programmed cell death in mammalian cells(34). We speculate that PPRL may regulate avrRpt2-triggeredoxidative burst, hypersensitive responses, or programmed celldeath, possibly through specific protein–RNA or protein–protein interactions. Future biochemical analysis on PPRL andidentification of its interaction proteins and RNAs will elucidatethe mechanism of its function in RPS2-mediated bacteriaresistance.

siRNA-mediated gene silencing plays an essential role inantiviral defense in both plant and animal systems (35, 36).However, these siRNAs generated from viral RNAs are ex-tragenomic in origin. Defense regulation mediated by endoge-nous small RNAs has been reported in only a few cases thus far,all of which involve only miRNAs. In mammals, miRNA-mediated antiviral defense has been reported (37), but thebiological roles of endogenous siRNAs have not been explored.In plants, miRNA miR393 regulates plant basal defense bytargeting auxin signaling components (8). A direct connectionbetween endogenous siRNAs and defense responses has notbeen reported previously in any organism. Our study hereprovides the first example of endogenous siRNAs that play a rolein bacterial disease resistance in Arabidopsis. Gene expressionprofiling studies indicate that the defense responses are medi-ated by activation and repression of a large array of genes, buthow the regulation of gene expression is achieved is largelyunknown. Our data suggest that endogenous siRNA-mediatedgene silencing may serve as one important mechanism for geneexpression reprogramming in plant defense responses. Ourfinding of induction of a nat-siRNA in responses to bacterialinfection opens up many new questions and provides newopportunities to elucidate the molecular events controlling plantdisease resistance.

Materials and MethodsPlant Material and Growth Conditions. Arabidopsis thaliana mutantsrdr2-1, dcl2-1, dcl3-1, and dcl4-2, were provided by Jim Carrington(Center for Genome Research and Biocomputing, Oregon StateUniversity, Corvallis). dcl1-9 and hen1-1 were a gift from XuemeiChen. sde1 (rdr6 in this study) and sde4�nrpd1a were provided byDavid Baulcombe (Sainsbury Laboratory, Norwich, U.K.). sgs3 wasa gift from Herve Vaucheret (Institut National de la RechercheAgronomique, Versailles, France). hyl1 was a gift from NinaFederoff (The Huck Institutes of Life Science, Pennsylvania StateUniversity, University Park). npr1 and pad4 were gifts from XinnianDong (Duke University, Durham, NC). sgt1a and sgt1b wereprovided by Jane Parker (Max-Planck-Institut fur Zuchtungsfors-chung, Cologne, Germany). ein2 was a gift from Athanasios The-ologis (Plant Gene Expression Center, Albany, CA). eds16 wasprovided by Mary Wildermuth (University of California, Berkeley).jar1 was provided by Linda Walling. These mutants were in theColumbia (Col-0), Landsberg erecta (Ler), Nossen-0 (No), or C24genetic backgrounds as indicated in the text and figures. The mutantrps2 (101C) has a stop codon at amino acid 235 of RPS2. Arabidopsisplants were grown at 23°C � 1°C at 12-h light�12-h dark photo-period. N. benthamiana plants were grown at 23°C � 1°C at 16-hlight�8-h dark photoperiod.

Fig. 5. Model for nat-siRNAATGB2 biogenesis and function. Components inred are required for nat-siRNAATGB2 formation. RISC, RNA-induced silencingcomplex.

18006 � www.pnas.org�cgi�doi�10.1073�pnas.0608258103 Katiyar-Agarwal et al.

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0

Plasmid DNA Constructs. For generating PPRL overexpression lines,full-length PPRL cDNA without 3� UTR was amplified withprimers 5�-CAC CAT GAA GTT CCT CAT GCA ATC CAT T-3�and 5�-ACG CCT ATT AGG TAA TGT CCC T-3� and cloned intothe plant expression GATEWAY destination vector p35SGATFHwith C-terminal Flag tag to avoid disruption of the signal peptideat the N terminus of the protein.

Isolation and Northern Blot Analysis of Small RNAs. Leaves harvestedat 15 hpi of Pst (2 � 107 cfu�ml) were used for RNA extraction andNorthern blot analysis of both high and low molecular weightRNAs. For enrichment of small RNAs, the total RNA was dissolvedin 4 M lithium chloride and precipitated. About 75–120 �g oflow-molecular-weight RNA was used and separated by 17% dena-turing polyacrylamide gel. The blots were probed and washed asdescribed (12).

Real-time RT-PCR was performed as in ref. 38. PPRL1 wasamplified with primers that locate outside of the overlappingregion: 5�-GCT TCA TCG CCG GAG GAA ATC-3� and5�-TTA ACC GAG CAC CCT TCA TCG T-3�. Transcript levelswere normalized to that of ubiquitin (5�-CGG AAA GAC CATTAC TCT GGA-3� and 5�-CAA GTG TGC GAC CAT CCTCAA-3�). Each experiment was repeated three times. Thecomparative Ct method was applied (ABI User Bulletin No. 2,Applied Biosystems, West Chester, PA).

Transient Expression Studies in N. benthamiana. A 564-bp overlap-ping region was amplified with the primers 5�-ACG CGT CGACAT GTG GAG CCA CCC GCA GTT CGA AAA ACG TACTCA AGG TGC AGC TGG AGG A-3� and 5�-GAA GGG GAACTA GTG TTA GTG ACG CGA ACA TAC AAT AAC TTGCG-3�. Full-length ATGB2 was amplified by using primers5�-ACG CGT CGA CAT GTG GAG CCA CCC GCA GTTCGA AAA ATC TTA CGA TTA TCT CTT CAA G-3� and5�-GAA GGG GAA CTA GTG TTA GTG A C GCG AAC ATACAA TAA CTT GCG-3�. A strep tag sequence was included inthe forward primers. The amplified products were cloned inXhoI and SpeI sites of the pTA7002 (28). Agrobacterium tume-

faciens strain GV3101 cells harboring PPRL or ATGB2 con-structs (OD600 � 1.0) were mixed at 1:1 ratio and coinfiltratedinto 3-week-old N. benthamiana leaves. The expression of full-length or overlapping region of ATGB2 was induced by infiltra-tion of 30 �M Dex at 48 hpi, and leaf tissue was collected at 24 hafter Dex induction.

Bacterial Growth Assays. Pst carrying EV (pVSP61) or avrRpt2,avrRpm1, avrRps4 and avrPphB were used to infect 4-week-oldArabidopsis leaves by infiltration at a concentration of �2 � 105

cfu�ml. The bacterial titer was measured at 0 and 4 dpi as in ref. 38.

HR Assay and Electrolyte Leakage Measurements. Leaves of 4-week-old Arabidopsis plants were infiltrated with 2 � 107 cfu�ml Pst(avrRpt2) for HR assay. Leaves were infiltrated with either 10mM MgCl2 (mock) or 1 � 107 cfu�ml Pst (avrRpt2) for electrolyteleakage assay. The leaf disks were washed in water for 50 min andthen transferred into 15 ml of water incubating for 16 h. Thetubes containing leaf disks and water were then autoclaved.Conductivity was measured before and after autoclave by an ECmeter (VWR Scientific, West Chester, PA). The percentage ofion leakage before and after autoclave was calculated andplotted. Four replicates were conducted in each treatment.

We thank Shou-Wei Ding and Xuemei Chen (University of California,Riverside, CA) and Jim Carrington for stimulating discussion on themanuscript and for providing seeds of various mutants; David Baul-combe, Nina Federoff, Herve Vaucheret, Xinnian Dong, Jane Parker,Athanasios Theologis, Mary Wildermuth, and Linda Walling (Universityof California, Riverside) for providing seeds of various genotypes;Thomas Girke for bioinformatics assistance; Julia Bailey-Serres (Uni-versity of California, Riverside) for binary plasmids; and James Borne-man for access to a real-time Icycler in his laboratory. This work wassupported by U.S. Department of Agriculture, State Agricultural Ex-periment Station Research Allocation Award PPA-7517H from theUniversity of California, Riverside (to H.J.), Department of EnergyGrant DE-FG02-88ER13917, and National Institutes of Health GrantsR01-FM069680-01 (to B.J.S.) and R01GM59138 and R01GM070795(to J.-K.Z.).

1. Baulcombe D (2005) Trends Biochem Sci 30:290–293.2. Sontheimer EJ, Carthew RW (2005) Cell 122:9–12.3. Jones-Rhoades M, Bartel D, Bartel B (2006) Annu Rev Plant Biol 57:19–53.4. Mallory AC, Vaucheret H (2006) Nat Genet 38:S31–S36.5. Jones-Rhoades MW, Bartel DP (2004) Mol Cell 14:787–799.6. Sunkar R, Zhu JK (2004) Plant Cell 16:2001–2019.7. Fujii H, Chiou TJ, Lin SI, Aung K, Zhu JK (2005) Curr Biol 15:2038–2043.8. Navarro L, Dunoyer P, Jay F, Arnold B, Dharmasiri N, Estelle M, Voinnet O,

Jones JD (2006) Science 312:436–439.9. Lu C, Tej SS, Luo S, Haudenschild CD, Meyers BC, Green PJ (2005) Science

309:1567–1569.10. Llave C, Kasschau KD, Rector MA, Carrington JC (2002) Plant Cell 14:1605–1619.11. Xie ZX, Johansen LK, Gustafson AM, Kasschau KD, Lellis AD, Zilberman D,

Jacobsen SE, Carrington JC (2004) PloS Biol 2:E104.12. Borsani O, Zhu J, Verslues P, Sunkar R, Zhu J (2005) Cell 123:1279–1291.13. Werner A, Berdal A (2005) Physiol Genomics 23:125–131.14. Wang X-J, Gaasterland T, Chua N-H (2005) Genome Biol 6:R30.15. Jen CH, Michalopoulos I, Westhead DR, Meyer P (2005) Genome Biology

6:R51.16. Zimmermann P, Hirsch-Hoffmann M, Hennig L, Gruissem W (2004) Plant

Physiol 136:2621–2632.17. Gustafson AM, Allen E, Givan S, Smith D, Carrington JC, Kasschau KD (2005)

Nucleic Acids Res 33:D637–D640.18. Whalen MC, Innes RW, Bent AF, Staskawicz BJ (1991) Plant Cell 3:49–59.19. Bent AF, Kunkel BN, Dahlbeck D, Brown KL, Schmidt R, Giraudat J, Leung

J, Staskawicz BJ (1994) Science 265:1856–1860.20. Century KS, Shapiro AD, Repetti PP, Dahlbeck D, Holub E, Staskawicz BJ

(1997) Science 278:1963–1965.21. Eulgem T (2005) Trends Plant Sci 10:71–78.

22. Alonso JM, Stepanova AN, Leisse TJ, Kim CJ, Chen H, Shinn P, StevensonDK, Zimmerman J, Barajas P, Cheuk R, et al. (2003) Science 301:653–657.

23. Schauer SE, Jacobsen SE, Meinke DW, Ray A (2002) Trends Plant Sci7:487–491.

24. Allen E, Xie ZX, Gustafson AM, Carrington JC (2005) Cell 121:207–221.25. Wassenegger M, Krczal G (2006) Trends Plant Sci 11:142–151.26. Hiraguri A, Itoh R, Kondo N, Nomura Y, Aizawa D, Murai Y, Koiwa H, Seki

M, Shinozaki K, Fukuhara T (2005) Plant Mol Biol 57:173–188.27. Han MH, Goud S, Song L, Fedoroff N (2004) Proc Natl Acad Sci USA

101:1093–1098.28. Aoyama T, Chua NH (1997) Plant J 11:605–612.29. Heazlewood JL, Tonti-Filippini JS, Gout AM, Day DA, Whelan J, Millar AH

(2004) Plant Cell 16:241–256.30. Lurin C, Andres C, Aubourg S, Bellaoui M, Bitton F, Bruyere C, Caboche M,

Debast C, Gualberto J, Hoffmann B, et al. (2004) Plant Cell 16:2089–2103.31. Shikanai T (2006) Cell Mol Life Sci 63:698–708.32. Wang ZH, Zou YJ, Li XY, Zhang QY, Chen L, Wu H, Su DH, Chen YL, Guo

JX, Luo D, et al. (2006) Plant Cell 18:676–687.33. Schmitz-Linneweber C, Williams-Carrier R, Barkan A (2005) Plant Cell

17:2791–2804.34. Verhagen AM, Kratina TK, Hawkins CJ, Silke J, Ekert PG, Vaux DL (June 23,

2006) Cell Death Differ, 10.1038�sj.cdd.4402001.35. Wang XH, Aliyari R, Li WX, Li HW, Kim K, Carthew R, Atkinson P, Ding SW

(2006) Science 312:452–454.36. Voinnet O (2005) Nat Rev Genet 6:206–220.37. Lecellier CH, Dunoyer P, Arar K, Lehmann-Che J, Eyquem S, Himber C, Saib

A, Voinnet O (2005) Science 308:557–560.38. Jin HL, Axtell MJ, Dahlbeck D, Ekwenna O, Zhang SQ, Staskawicz B, Baker

B (2002) Dev Cell 3:291–297.

Katiyar-Agarwal et al. PNAS � November 21, 2006 � vol. 103 � no. 47 � 18007

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Apr

il 21

, 202

0