Diverse pathogens secrete effector proteins into plant cells to manipulate host cellular processes. Oomycete pathogens contain large complements of predicted effector genes defined by an RXLR host cell entry motif. The genome of Hyaloperonospora arabidopsidis (Hpa, downy mildew of Arabidopsis) contains at least 134 candidate RXLR effector genes. Only a small subset of these genes is conserved in related oomycetes from the Phytophthora genus. Here, we describe a comparative functional characterization of the Hpa RXLR effector gene HaRxL96 and a homologous gene, PsAvh163, from the Glycine max (soybean) pathogen Phytophthora sojae. HaRxL96 and PsAvh163 are induced during the early stages of infection and carry a functional RXLR motif that is sufficient for protein uptake into plant cells. Both effectors can suppress immune responses in soybean. HaRxL96 suppresses immunity in Nicotiana benthamiana, whereas PsAvh163 induces an HR-like cell death response in Nicotiana that is dependent on RAR1 and Hsp90.1. Transgenic Arabidopsis plants expressing HaRxL96 or PsAvh163 exhibit elevated susceptibility to virulent and avirulent Hpa, as well as decreased callose deposition in response to non-pathogenic Pseudomonas syringae. Both effectors interfere with defense marker gene induction, but do not affect salicylic acid biosynthesis. Together, these experiments demonstrate that evolutionarily conserved effectors from different oomycete species can suppress immunity in plant species that are divergent from the source pathogen’s host.
Plant pathogens have evolved from diverse kingdoms of life and pose a perennial threat to agriculture. Some of the most destructive pathogens are oomycetes, which include over 800 species of downy mildew pathogens in 17 genera (Peronosporaceae), along with over 100 species in the Phytophthora genus (Erwin and Ribiero, 1996; Clark and Spencer-Phillips, 2000). Hyaloperonospora arabidopsidis (Hpa) is a naturally occurring downy mildew pathogen of Arabidopsis thaliana (Holub and Beynon, 1996; Slusarenko and Schlaich, 2003; Holub, 2008). Hpa and other downy mildews are obligate biotrophs, which obtain nutrients exclusively from living plant cells and cannot be cultured apart from their hosts (Clark and Spencer-Phillips, 2000; Gisi, 2002; Coates and Beynon, 2010). Contrastingly, Phytophthora species employ a hemi-biotrophic lifestyle, in which an initial phase of biotrophic growth is followed by a transition to necrotrophy, during which host tissue is destroyed (Judelson and Blanco, 2005). Oomycete diseases are difficult to control because of the pathogens’ high evolutionary potential to overcome plant immunity and fungicides (Tyler, 2007).
Successful pathogens must evade or suppress multiple layers of innate immune surveillance. One level is comprised of extracellular receptors that recognize microbe-associated molecular patterns (MAMPs), and activate defense responses that include the production of reactive oxygen species (ROS), induction of defense genes and deposition of callose in papillae at sites of infection (Jones and Dangl, 2006). This suite of responses comprises pattern-triggered immunity (PTI) (Katagiri and Tsuda, 2010). A second level is mediated by resistance (R) proteins that directly or indirectly recognize pathogen effector proteins inside the host cell. This recognition triggers a rapid and robust immune response that often includes programmed cell death (PCD), termed the hypersensitive response (HR) (Dodds and Rathjen, 2010). This suite of responses, which shares considerable overlap with PTI, is termed effector-triggered immunity (ETI) (Chisholm et al., 2006). Together, PTI and ETI enable most plants to resist most pathogens. However, many oomycete pathogens have adapted to overcome these immune responses in specific host plant species. The molecular basis of these adaptations is a critically important but poorly understood aspect of plant–oomycete interactions.
Many plant pathogens use effector proteins to subvert plant immunity. Bacterial species secrete 20–40 effectors into the cytoplasm of plant cells through a type-III secretion system (TTSS) (Grant et al., 2006). Bacterial effectors have been studied in detail, and have been shown to suppress PTI and/or ETI through a variety of mechanisms (Hann et al., 2010). In contrast, the understanding of effector functions from oomycetes and fungi is in its infancy (de Jonge et al., 2011; Koeck et al., 2011).
Only a handful of RXLR effectors have been characterized in detail. For example, P. sojae Avr1b increases the virulence of P. sojae when overexpressed and suppresses cell death, triggered by multiple elicitors, in Glycine max (soybean) and Nicotiana benthamiana (Dou et al., 2008a,b). The virulence and avirulence functions of Avr1b are dependent on so-called W, Y and L motifs that are also found in a large proportion of RXLR effectors (Dou et al., 2008a,b; Jiang et al., 2008). Mutations in these motifs abolished the recognition of Avr1b by the cognate R gene (Rps1b), and abolished the ability to suppress Bax-induced cell death (Dou et al., 2008a,b). Phytophthora infestans Avr3a perturbs the plant E3 ubiquitin ligase CMPG1, which is a positive regulator of immunity (Bos et al., 2010; Gilroy et al., 2011a,b). P. sojae Avr3b is an NADH and ADP-ribose pyrophosphorylase that modulates plant immunity and is essential for full virulence (Dong et al., 2011a,b).
The genome of Hpa isolate Emoy2 was recently sequenced (Baxter et al., 2010). Bioinformatic surveys revealed at least 134 candidate RXLR proteins. Very little conservation exists between pools of effector genes maintained by oomycete pathogens (Baxter et al., 2010). Only 30% of the Hpa effectors have >20% identity with their best Phytophthora match, and only 5% share >40% identity with their best match (Baxter et al., 2010). This small subset of conserved effectors is of particular interest as it may contain ‘core’ pathogenicity genes that target cellular processes conserved between diverse plant species.
Here, we describe functional comparisons of one pair of computationally predicted conserved effector genes from Hpa and P. sojae: HaRxL96 and PsAvh163, respectively (Dou et al., 2008a,b). These genes are expressed early in infection and have functional RXLR motifs, confirming that they are bona fide effectors. Both effectors suppress or induce plant immunity in diverse plant species, which is suggestive of interaction with conserved host targets.
HaRxL96 and PsAvh163 share conserved functional domains
HaRxL96 and PsAvh163 encode proteins of 415 and 504 amino acids, respectively, that share 26% identity and 43% similarity with each other. Both proteins contain a predicted signal peptide at the N terminus, followed by a predicted host-targeting sequence (HTS) consisting of an RXLR-like motif and a short stretch of acidic amino acids (Figure S1). Both effectors contain multiple copies of degenerate W, Y and L motifs (Figure 1; Dou et al., 2008a). No other functional domains or subcellular localization signals were apparent in the predicted sequences of these effectors.
To further examine the conservation of these genes, we searched for homologs within and outside of the Phytophthora genus. Genomes of P. infestans and Phytophthora ramorum contain homologs in which the SP, HTS, and W, Y and L motifs are apparent at conserved positions (e.g. PITG_10341 from P. infestans; Figure S1). No homologs were evident in the genomes of Pythium ultimum, Saprolegnia parasitica or Albugo labachii, indicating that the HaRxL96/PsAvh163 lineage evolved after the divergence of the last common ancestor of Phytophthora and the downy mildew pathogens (Thines and Kamoun, 2010).
Allele sequences reveal differing levels of polymorphism for HaRxL96 and PsAvh163
To examine the sequence diversity of HaRxL96, we amplified and sequenced alleles of HaRxL96 from five Hpa isolates (Emoy2, Cala2, Emco5, Hiks1 and Noco2). Alleles of PsAvh163 were obtained from genome sequences of four P. sojae isolates (Wang et al., 2011) that encompass most of the genetic diversity of P. sojae (Förster et al., 1994). The HaRxL96 nucleotide sequences were completely identical in all of the Hpa isolates. Contrastingly, the PsAvh163 alleles exhibit a high degree of diversity (Figure S2). The alleles from P6497 (genotype I) and P7076 (genotype II) are identical, whereas the nucleotide sequences from P7064 (genotype IV) and P7074 (genotype III) differ from the genotype I/II reference sequence by 75 and 43 non-synonymous substitutions, and only two and three synonymous substitutions, respectively. The amino acid substitutions are unevenly distributed over the length of the protein. For example, only one substitution occurs in the N-terminal 60 amino acids that span the signal peptide and RXLR-EER regions. By contrast, at least one substitution occurs at seven of the 18 positions in the first W motif. The dN/dS ratios over the entire length of the P7064 and P7074 alleles, compared with the P6497 allele, are 6.4 and 3.8, indicative of diversifying selection (P < 1.0E–08, 6.6E–04, respectively). This diversity contrasts strongly with the absence of variability, even at third codon positions, among HaRxL96 alleles.
HaRxL96 and PsAvh163 are induced during infection
Effector genes are often induced during early stages of infection (Haas et al., 2009; Wang et al., 2011). We monitored HaRxL96 transcript levels in planta following infection of the susceptible Arabidopsis ecotype Oystese (Oy-0) by virulent Hpa isolate Emoy2 using quantitative RT-PCR. HaRxL96 was upregulated during the first 12 h after inoculation. Its expression declined during subsequent stages of the interaction (Figure 2). A similar pattern of induction was described for PsAvh163 during the interaction between P. sojae and soybean, and has been termed ‘immediate–early, low’ (Wang et al., 2011). Thus, both effectors display similar patterns of expression, suggestive of a function during the initial stages of infection.
HaRxL96 and PsAvh163 contain functional host targeting motifs
To further validate that HaRxL96 and PsAvh163 encode functional effector proteins, we tested whether their predicted host targeting sequences (HTSs) could deliver a chimeric fusion to the interior of soybean cells, using a quantitative bombardment assay that employs a double-barreled gene gun (Kale and Tyler, 2011). We constructed chimeric genes encoding proteins with the signal peptide from soybean Pathogenesis-Related 1a protein, followed by the predicted HTSs of HaRxL96 or PsAvh163 and the C-terminal avirulence domain (CTD) of P. sojae Avr1b (Figure 3a). We co-bombarded soybean leaves with each chimeric construct, along with a plasmid containing the β-glucuronidase (GUS) gene to measure cell viability (Figure S3b, c). Bombardments with either HaRxL96HTS-Avr1bCTD or PsAvh163HTS-Avr1bCTD resulted in reduced numbers of GUS-expressing foci, suggesting that an HR response was triggered in Rps1b soybean leaves but not in rps1b soybean leaves (Figure 3). Furthermore, mutation of the RXLR motif to AAAA abolished this reduction in GUS expression, suggesting that cell entry of the secreted fusion protein was impaired (Figure 3). These results suggest that the HTS of both HaRxL96 and PsAvh163 can mediate uptake by plant cells, further supporting that the HaRxL96 and PsAvh163 genes encode functional effector proteins.
PsAvh163 triggers a hypersensitive response in N. benthamiana
To test for potential avirulence activity of both effectors, we used transient assays to express each gene inside plant cells, and then screened for a macroscopic cell death response. For screens of Arabidopsis ecotypes, we used Pseudomonas syringae DC3000(ΔCEL) to deliver HaRxL96 or PsAvh163(P6497) via the TTSS, as described by Sohn et al. (2007). DC3000(ΔCEL) was transformed with ‘effector detector vector’ expression constructs in which the leader from Pseudomonas syringae AvrRps4 was fused to HaRxL96 or PsAvh163, beginning from the predicted signal peptide cleavage site. The AvrRPS4 leader directs the fusion through the TTSS and is then removed by an endogenous plant protease (Sohn et al., 2007). The processed effector could then induce an HR response if it is recognized by the plant. The ΔCEL mutant was chosen because it does not trigger strong disease symptoms that could be mistaken for a weak HR. We screened 48 Arabidopsis ecotypes for a macroscopic HR in response to Pseudomonas syringae with pEDV-Ha96 or pEDV-PsAvh163, injected at a concentration of OD600 = 0.01. Although we reproduced a previously reported HR response in control experiments with the Hpa avirulence gene Atr1 (Sohn et al., 2007), we did not observe a robust HR to HaRxL96 or PsAvh163 in any of the ecotypes (Table S1).
As a complement to the assays in Arabidopsis and soybean, we screened N. benthamiana for an HR response using Agrobacterium tumefaciens-mediated transient expression (ATTA) of 35S:HaRxL96 and 35S:PsAvh163. HaRxL96 did not trigger cell death, but the PsAvh163 alleles from P6497, P7064 and P7074 triggered a rapid cell death response within 24 h (Figure 4a), consistent with observations reported recently (Wang et al., 2011). The three alleles of PsAvh163 also triggered strong cell death in Nicotiana tabacum (Figure 4b).
To learn more about the nature of this cell death response, we used virus-induced gene silencing (VIGS) with the Tobacco Rattle Virus (TRV) system to knock down expression of genes that are associated with R protein functionality. PsAvh163-dependent cell death was abolished in plants that were silenced for Hsp90.1 and SGT1, but not for Hsp90.2, Hsp90.3 or RAR1 (Figure 4c). The requirement for Hsp90.1 and SGT1 suggests that PsAvh163-induced cell death results from ETI rather than non-specific toxicity of PsAvh163.
HaRxL96 suppresses programmed cell death in N. benthamiana
Effectors from bacteria and oomycetes are often able to suppress PCD triggered by avirulence proteins, cell death regulators and chemical treatments (Alfano et al., 2009; Cabral et al., 2011; Wang et al., 2011). Suppression of PCD is a potentially important virulence mechanism for biotrophic pathogens that extract nutrients from living plant cells. Previous examinations of oomycete effector function have used an assay for the suppression of cell death in N. benthamiana triggered by the elicitor Infestans 1 (INF1) from Phytophthora infestans (Bos et al., 2006; Wang et al., 2011). We transiently expressed HaRxL96 using ATTA, delivered INF1 using ATTA 48 h later and scored infiltration sites for macroscopic cell death 5 days later. These experiments indicated that HaRxL96 could suppress INF1-induced cell death (Figure S4). We also observed that HaRxL96 could suppress the cell death triggered by PsAvh163 (Figure S4).
HaRxL96 and PsAvh163 suppress effector-triggered programmed cell death in soybean
Previously, we demonstrated that HaRxL96 and PsAvh163 could suppress PCD triggered in soybean by mouse Bax using a transient assay (Dou et al., 2008a,b). To further investigate the ability of HaRxL96 and PsAvh163 to suppress PCD in soybean, we tested for suppression of PCD triggered by the P. sojae avirulence protein Avr4/6 in the presence of the corresponding R genes Rps4 or Rps6 (Dou et al., 2010). The double-barreled particle bombardment assay was employed to test whether cell death was reduced in soybean co-bombarded with a DNA mixture that carried HaRxL96 or PsAvh163 plus Avr4/6, and GUS, compared with an equimolar DNA mixture consisting of empty vector, plus Avr4/6 and GUS. We observed a 50% decrease of GUS-expressing cells in Rps4 or Rps6 tissue bombarded with Avr4/6 plus EV, relative to control samples bombarded with GUS but not with Avr4/6, similar to reports by Dou et al. (2010). When HaRxL96 or PsAvh163 was included with Avr4/6, we observed a significant increase in cell viability after bombardment, indicating that HaRxL96 and PsAvh163 could suppress Avr4/6-triggered cell death (Figure 5). This suggests that PsAvh163 and HaRxL96 share the ability to suppress ETI in soybean.
HaRxL96 and PsAvh163 suppress effector-triggered immunity and basal resistance against Hpa in Arabidopsis
We generated stably transformed Arabidopsis Col-0 lines that express either HaRxL96 or PsAvh163 under the control of the CaMV 35S promoter. We bred multiple, independent lines to homozygosity and determined transgene mRNA abundance with quantitative, real-time PCR (Figure S5). We then tested lines expressing the transgene at different levels for increased susceptibility to the virulent Hpa isolate Emco5. We measured Emco5 growth by counting sporangiophores (Figure 6a), and with a quantitative PCR assay that we recently developed (Figure S6; Anderson and McDowell, in preparation). Both assays demonstrate that Col:HaRxL96 and Col:PsAvh163 transgenic plants exhibit enhanced susceptibility to virulent Hpa, compared with wild-type Col.
Col-0 lines expressing HaRxL96 or PsAvh163 also exhibited decreased resistance against avirulent Hpa isolate Emoy2. Resistance to Emoy2 in Col-0 is provided by RPP4, a Toll/interleukin-1 receptor, nucleotide binding, leucine-rich repeat protein (TIR-NB-LRR) (van der Biezen et al., 2002). Transgenic lines expressing each effector exhibited reduced resistance to Hpa Emoy2, compared with wild-type plants (Figure 6b).
HaRxL96 and PsAvh163 transgenes suppress callose deposition induced by non-pathogenic Pseudomonas syringae in Arabidopsis
One well-studied component of the Arabidopsis PTI response is the formation of callose at the site of infection (Luna et al., 2011). We tested whether HaRxL96 and PsAvh163 could suppress the callose response to Pseudomonas syringae DC3000(ΔHrcC), which cannot form a functional TTSS, and therefore cannot suppress PTI. We infiltrated DC3000(ΔHrcC) into transgenic and control lines, and stained with aniline blue at 16 h post-inoculation (hpi) to visualize the callose. Col:HaRxL96 and Col:PsAvh163 transgenic plants had an attenuated response, exhibiting up to 50% less callose production compared with wild-type plants (Figure 7). These results suggest that HaRxL96 and PsAvh163 can interfere with Arabidopsis PTI deployed against Pseudomonas syringae.
HaRxL96 and PsAvh163 transgenes suppress induction of salicylic acid-responsive defense genes downstream or independently of salicylic acid
To further study the molecular basis for the suppression of immune responses in Col:HaRxL96 and Col:PsAvh163, we used quantitative real-time PCR (qPCR) to measure transcript abundance of four defense marker genes: Mitogen-Activated Protein Kinase 3 (MPK3), Pathogenesis-Related 1 (PR1), Accelerated Cell Death 6 (ACD6) and Wall-Associated Kinase 1 (WAK1). These genes are induced during the RPP4-mediated resistance response to Hpa (Eulgem et al., 2007). These four genes are also transiently suppressed in a compatible interaction between Hpa and Arabidopsis, suggesting that they could be direct or indirect targets of Hpa effectors (Eulgem et al., 2007). We inoculated Col:HaRxL96 and Col:PsAvh163 with avirulent Hpa Emoy2, and measured transcript abundance from all four genes. As shown in Figures 8 and S7, the induction of all four genes was attenuated at early time points during the interaction in Col:HaRxL96 and Col:PsAvh163, compared with wild-type, resistant Col. The transcript abundance from ACD6 and MPK3 was lower than in the compatible interaction between Oy-1 and Emoy2.
The suppression of PR-1 suggests that HaRxL96 and PsAvh163 directly or indirectly interfere with the activation of SA-responsive defense genes. To explore this inference further, we tested whether SA biosynthesis was suppressed in the transgenic plants. We observed that SA biosynthesis was not attenuated by HaRxL96 or PsAvh163. The kinetics and magnitude of SA accumulation were not significantly different in wild-type and transgenic plants (Figure S8). These results suggest that both effector genes impinge on the immune network at a point downstream or independent of SA biosynthesis.
The availability of genome sequences has opened the door to identify and compare the functions of effectors conserved among distantly related oomycete pathogens. The utility of a comparative functional approach has been validated by studies of effectors from bacteria (e.g. the Conserved Effector Locus) and fungi (e.g. the LysM effector Ecp6) (DebRoy et al., 2004; Nomura et al., 2006; de Jonge et al., 2010). Recent genome-level comparisons indicate that oomycetes are under strong selective pressure to diversify their RXLR weaponry (Win et al., 2007; Jiang et al., 2008; Haas et al., 2009; Baxter et al., 2010; Raffaele et al., 2010). Thus, it is plausible that the most conserved RXLR genes might play important ‘core’ roles in oomycete pathogenicity. Another rationale for focusing on conserved effectors is that comparative studies of conserved effectors across species can provide insight into effector gene evolution. Finally, conserved effectors can be used in ‘effector-directed’ screens for new resistance genes (Vleeshouwers et al., 2008, 2010; Oh et al., 2009). Based on these rationales, we are studying HaRxL genes that have clear homologs in Phytophthora genomes (Baxter et al., 2010). Our hypothesis is that conserved effectors manipulate the same or similar targets in the pathogens’ respective hosts.
Homologs of HaRxL96 are evident in the genomes of Phytophthora species but not in other oomycetes, indicating that this lineage emerged after the divergence of the last common ancestor of the Peronosporales. Although HaRxL96 and PsAvh163 are not highly conserved, important functional motifs are apparent, including the signal peptide, the RXLR and EER motifs in the host targeting region, and core residues in the W, Y and L domains. HaRxL96 and PsAvh163 appear to be evolving under different modes. HaRxL96 DNA sequences are completely invariant amongst the five Hpa isolates that we surveyed. This absence of divergence, even at third codon positions, suggests a very conservative mode of evolution or a recent selective sweep. By contrast, the four alleles of PsAvh163 exhibit the highest degree of allelic diversity among all of the predicted RXLR genes in P. sojae, and are under positive selection for sequence diversification (Wang et al., 2011). This divergence could reflect selection pressure to avoid a resistance gene in soybean. It will be of interest to compare the functions of different allelic forms and to survey allelic diversity of HaRxL96 and PsAvh163 orthologs in other oomycetes.
One experimental challenge, relevant to any effector predicted from bioinformatic criteria, is to confirm that the predicted effector is a bona fide effector. Figure 2 and data from Wang (2011) demonstrate that HaRxL96 and PsAvh163 genes are induced in planta, with an immediate–early transcript induction pattern that suggests a role in the preparation of host tissue for colonization. We also demonstrated that the predicted host cell targeting sequences were sufficient for translocation of the P. sojae Avr1b avirulence domain into soybean cells, and that the RXLR motif in both proteins was necessary for this translocation. The ability of the HaRxL96 HTS to enter soybean (a non-host of Hpa) is not surprising, given previous demonstrations that RXLR-mediated entry is host non-specific (Bhattacharjee et al., 2006; Grouffaud et al., 2008). In sum, both genes are transcribed in a pattern consistent with a role in early infection, and both proteins are capable of entering plant cells, suggesting that HaRxL96 and PsAvh163 encode bona fide effectors.
Additional support for effector functionality was provided by experiments demonstrating that HaRxL96 and PsAvh163 can suppress or induce immunity in soybean, Nicotiana and Arabidopsis. As Hpa cannot be genetically manipulated because of its obligate lifestyle, our assays used the expression of individual effectors inside plant cells, either transiently or as stably integrated transgenes. This approach has provided important insights into the function of effectors from bacteria (Munkvold and Martin, 2009; Cabral et al., 2011). We screened several plant species for macroscopic HR-like responses against HaRxL96 and PsAvh163. Screens of 43 Arabidopsis ecotypes using the effector detector vector in Pseudomonas syringae DC3000(ΔCel) did not reveal any consistent HR responses against either protein. However, PsAvh163 induces a cell death response in N. benthamiana and N. tabacum. Wang et al. (2011) recently described a similar result for PsAvh163 and several other PsAvh genes (Wang et al., 2011). In principle, this response could reflect cellular toxicity of the effector proteins. However, VIGS experiments demonstrated that the cell death response requires components that have been previously associated with ETI (SGT1 and HSP90). Additionally, PsAvh163-dependent cell death can be suppressed by HaRXL96. These attributes suggest that PsAvh163 is recognized by a resistance gene in the Nicotiana genus. All three alleles of PsAvh163 could induce the HR response, despite their divergence, suggesting that recognition is indirect (i.e. triggered by the interaction of PsAvh163 with a guarded target or decoy). The R gene responsible for this recognition may have evolved to recognize a homolog of PsAvh163 in a Phytophthora pathogen of Nicotiana species (e.g. Phytophthora nicotianae). The cell death induced in Nicotiana by PsAvh163 will be very useful for future studies, which can exploit ATTA, VIGS and other tools in Nicotiana (Goodin et al., 2008). Additionally, PsAvh163 could be used as a probe to identify resistance genes against P. sojae or other oomycetes.
Our assays to test for the suppression of PTI and ETI indicate that both effectors are capable of suppressing immunity triggered by diverse stimuli. For example, HaRXL96 and PsAvh163 can enhance growth of virulent Hpa and partially suppress RPP4-dependent resistance. Moreover, HaRXL96 and PsAvh163 can suppress PTI triggered by Pseudomonas syringae DC3000(ΔHrcC). These results suggest that both proteins target a central component of the resistance network that contributes to resistance against oomycetes and bacteria. The ability of HaRXL96 and PsAvh163 to suppress cell wall-based defenses, PCD and defense gene expression suggest that both proteins target a regulatory node that influences multiple aspects of plant immunity.
The defense marker genes tested in this study have been shown previously to respond to RPP4-dependent ETI. These genes have been previously demonstrated to be transiently suppressed early in the interaction with virulent Hpa, consistent with active suppression by the pathogen (Eulgem et al., 2007). Our assays suggest that HaRXL96 and PsAvh163 are both sufficient to interfere with the induction of these genes. However, these experiments are not sufficient to conclude whether these two effectors are required for the suppression of these genes; it is possible that other effectors also contribute to the suppression. We do not know whether the suppression is direct or indirect. However, the SA assays suggest that the mechanism of suppression does not involve suppression of SA biosynthesis, which would indicate that these effector genes must target a network node that acts downstream or independently of SA accumulation.
Our experiments demonstrate that both effectors can function in divergent plant species: HaRxL96 suppresses PCD triggered by oomycete elicitors in soybean, and PsAvh163 can suppress PTI and ETI responses in Arabidopsis. Additionally, PsAvh163 and HaRxL96 can trigger and suppress immunity in Nicotiana, respectively. Thus, the biological activity of these proteins is not restricted to the source pathogens’ host. This is reminiscent of effectors from Pseudomonas syringae that exhibit biological activity in closely related solanaceous plants (e.g. tomato and Nicotiana), as well as in more distantly related Arabidopsis and even in yeast (e.g. AvrPtoB; Abramovitch et al., 2003; Jamir et al., 2004; de Torres et al., 2006). A total of 107 of 169 RXLR effectors from P. sojae can suppress PCD triggered by Bax in N. benthamiana when expressed via ATTA. A subset of 43 genes from these 107 were tested further, and exhibited the capacity to suppress cell death triggered by INF1 and/or Avr proteins (Wang et al., 2011). By contrast, only 13 of 57 Hpa effectors can suppress immunity in Brassica rapa (turnip, closely related to Arabidopsis) when delivered from Pseudomonas syringae (Fabro et al., 2011). The experiments with HaRXL96 and PsAvh163 suggest that some Hpa effectors, and their P. sojae homologs, function across a wide taxonomic range. The molecular basis of this apparently broad functionality bears further investigation.
HaRxL96 and PsAvh163 have the same apparent spectrum of activity in Arabidopsis (impairment of basal defense, ETI and defense gene activation in response to Hpa, as well as callose deposition in response to non-pathogenic Pseudomonas syringae mutants). Additionally, both proteins suppress PCD triggered by multiple elicitors in soybean. These concordances are consistent with our hypothesis that the effectors target the same or similar proteins in their respective hosts, and that the targets are sufficiently conserved to enable the effectors to interact with those targets in diverse species. Future studies should test this hypothesis by identifying the subcellular sites of action and the targets of these effectors in Arabidopsis and soybean.
Construction of expression plasmids
HaRxL96 was amplified from genomic DNA isolated from Arabidopsis Oy-1 tissue colonized by Hpa Emoy2, using the primer sets Ha96NOSP and Ha96S (with stop codon) or Ha96NS (without stop codon). PsAvh163 was amplified from genomic DNA isolated from the P. sojae isolate P6497 (Race 2), P7064 (Race 7), P7074 (Race 17) or P7076 (Race 19), with primer sets Ps163 NOSP and Ps163S (stop codon) or Ps163NS (without stop codon). The PsAvh163 P6497 allele was used for all experiments described in this article, except as depicted in Figure 4. For all cloned open reading frames (ORFs), the 5′ primer began with the codon immediately downstream of the signal peptide cleavage site predicted by signalp. PCR amplicons were cloned in pENTR D/TOPO and shuttled into expression plasmids using LR recombinase (Invitrogen, http://www.invitrogen.com). For plant expression studies, HaRxL96 and PsAvh163 alleles were shuttled from pENTR D/TOPO into pB2GW7. For host targeting analysis, clones were created by overlapping PCR extensions. The soybean PR1a signal peptide was cloned from soybean genomic DNA with primers Pr1aF and Pr1aR. The Avr1b-1 CTD was cloned from genomic DNA with primers Avr1b CTDF and Avr1b CTDR. The RXLR host-targeting segments included sequences from the predicted signal peptide cleavage sites to 60 bp beyond the last codon in the predicted -EER motifs. The HTS was amplified from HaRxL96 using the primer set Ha96OPEF and Ha96OPER, and Ps163OPEF and Ps163OPER for Ps163. Three PCR products, including the Pr1A signal peptide, effector host targeting region and the Avr1B CTD were used as a template to generate a fusion product using the Pr1AF and Avr1bCTDR primers. The fusion products were digested and ligated in Xma1 and Kpn1 sites of a modified pUC19 plasmid with a CaMV 35S promoter (Dou et al., 2008a,b). All PCR products and resultant clones were confirmed by sequencing. Primer sequences are provided in Table S2.
Plant growth, assays with Hyaloperonospora arabidopsidis and generation of transgenic Arabidopsis plants
Arabidopsis, soybean and N. benthamiana plants were grown in Sunshine Mix #1 with 16 h of light and 8 h of dark at 22˚C. Arabidopsis plants for pathogen assays were grown under 8 h of light at 22˚C and 16 h of dark at 20˚C. The Hpa isolates Emoy2 and Emco5 were propagated and maintained on Oy-1 and Ws-0 Arabidopsis plants, respectively (McDowell et al., 2011). Conidial suspensions of 5 × 104 spores ml−1 were applied with a Preval spray unit, and the plants were kept under short-day conditions. Hpa disease assays were performed as described by McDowell et al. (2011). Transgenic Arabidopsis Col-0 plants were generated by floral dipping (Clough and Bent, 1998). BASTA-resistant plants were selected, transgene presence was confirmed by PCR and transcript abundance was determined with Q-PCR. Lines with single transgene loci were identified by segregation in the T2 generation, and homozygous lines were identified by progeny testing in the T3 generation. All experiments were performed on non-segregating T3 or T4 populations.
RNA isolation, reverse-transcriptase PCR and real-time PCR
Tissue infected with Hpa Emoy2 was harvested at the indicated time points and RNA was extracted with TriSure reagent (Bioline, http://www.bioline.com). cDNA was synthesized using an OmniScript cDNA synthesis kit (Qiagen, http://www.qiagen.com). Forty PCR cycles were used to amplify effector targets from cDNA templates. For real-time PCR, 25-μl samples were prepared by mixing 1 μl of cDNA template with 12.5 μl of Sybr Green Mastermix (Applied Biosystems, http://www.appliedbiosystems.com) with the appropriate primers and water. Real-time PCR reactions were preformed on an ABI 7300 device (Applied Biosystems) and fold change was calculated relative to the 0 HPI time point.
Real-time PCR assay for growth of Hyaloperonospora arabidopsidis
Five individuals from each sample were pooled in extraction buffer (200 mm Tris, pH 7.5, 25 mm EDTA, pH 7.5, 250 mm NaCl, 0.5% SDS) and genomic DNA (gDNA) was extracted using a bead beater. gDNA samples were quantified and diluted to a final concentration of 10 ng μl−1. Then 25-μl samples were prepared by mixing 5 μl of gDNA sample with 12.5 μl of Sybr Green Mastermix (ABI, http://www.appliedbiosystems.com), along with primers and water. The primer sets AtActin Fwd/AtActin Rev were used for AtActin, and HaAct Fwd/HaAct Rev were used for HpaActin (Brouwer et al., 2003). PCR reactions were preformed on an ABI 7500 device. Cycle threshold (Ct) values were determined using ABI software. Relative abundance compared with AtActin was calculated as
Transient assays in soybean
Unifoliate primary leaves from 2-week-old soybean plants were bombarded using a modified Bio-Rad PDS1000 gene gun (Bio-Rad, http://www.bio-rad.com) with a double-barrel attachment (Kale and Tyler, 2011). Mixtures of the effector, control and elicitor plasmids were prepared as described by Dou et al. (2008a,b). For Avr4/6 suppression assays, 115 μg of the effector plasmid, 50 μg of Avr4/6 plasmid and 50 μg of GUS plasmid were combined. For Bax suppression assays, 115 μg of the effector plasmid, 15 μg of Bax plasmid and 50 μg of GUS plasmid were combined. The control GUS sample contained 115 μg of empty vector and 50 μg of GUS plasmid DNA. The elicitor samples were mixed as follows: Avr4/6 – 30 ng of Avr4/6 plasmid, 50 μg GUS plasmid and 70 μg empty vector; Bax – 15 μg Bax plasmid, 50 μg GUS plasmid and 85 μg empty vector. After bombardment, detached soybean leaves were incubated in Petri dishes with moistened Whatman filter paper at 22/20°C (8 h light/16 h dark). Leaves were stained with X-Gluc and cleared with 70% ethanol for 2 days. GUS-expressing tissue patches were counted using a dissecting microscope.
Transient assays and VIGS in Nicotiana benthamiana
Agrobacterium tumefaciens GV3101 was grown overnight in LB with the appropriate antibiotics, harvested by centrifugation, resuspended in 10 mm MgCl2, 10 mm 2-(N-morpholine)-ethanesulphonic acid (MES) and 200 mm acetosyringone, incubated at room temperature for 1–3 h and infiltrated using needless syringes on the abaxial side of 3–5-week-old leaves. VIGS was preformed as described by Liu et al. (2004). Plants silenced for the target genes were challenged in the upper leaves with Agrobacterium tumefaciens GV3101 containing PsAvh163 (P6497), 2 weeks after the initial inoculation with the VIGS constructs. Cell death was visually scored 5 days later.
Liquid cultures of Pseudomonas syringae were grown overnight in LB at 28˚C, 200 rpm. Cultures were centrifuged at 1500 g for 10 m at 4°C and resuspended in 10 mm magnesium sulfate to an OD600 of 0.1. Four-week-old transgenic and wild-type plants were infiltrated using needless syringes. Challenged leaf tissue was harvested at 16 hpi, stained with aniline blue and cleared as described by Sohn et al. (2007). Leaves were imaged with a Zeiss Axio Imager M1 and DAPI filter set. Callose was quantified with Quantify One (Bio-Rad).
Quantification of salicylic acid
Levels of salicylic acid (SA) were quantified as described by Schmelz et al. (2004). Briefly, 200–300 mg of frozen tissue was homogenized in FastPrep® tubes (MP Biomedicals, http://www.mpbio.com) containing 1 g of ceramic beads, 100 ng of internal standard and 300 μl of H2O : 1-propanol : HCl (1:2:0.01). After homogenization the tissue was partitioned by adding 1 ml of methylene dichloride (MeCl2). The organic phase was derivatized to methyl esters using trimethylsilyldiazomethane. After 30 min of derivatization the reaction was stopped by the addition of 5 μl of 2 m acetic acid in MeCl2. Vapor phase extraction was than performed by applying a stream of N2 over the sample and through a collection of absorbent Super Q filter papers. The samples were then analyzed by GC/CI-MS. A DB-35MS (30 m × 0.25 mm × 0.25 μm; Agilent, http://www.home.agilent.com) GC column was used with the following temperature program: held at 70°C for 1 min after injection, followed by a temperature gradient of 15°C min−1 to 300°C (for 7 min), with helium as the carrier gas (0.7 ml min−1). Quantity estimates of SA (retention time, 8.44 min; ion, 153) were based on the isotopically labeled internal standard 2H6-SA (retention time, 8.43 min; ion, 157). This method does not discriminate between methyl esters and free acids.
We thank Jonathan Jones for pEDV6, Guido van den Ackerveken for INF1, Brian Staskawicz for N. tabacum seed, Savithiramma Dinesh-Kumar for VIGS clones, and Shiv Kale and Daolong Dou for Avr4/6 clones and assistance with bombardment assays. We thank Richard Duong and Dan Deegan for excellent technical assistance, and the anonymous reviewers for their constructive comments. This work was supported by grants to J.M.M. and B.M.T. from the National Science Foundation (IOS-0744875) and the US Department of Agriculture–Agriculture and Food Research Initiative (2009-03008 and 2011-68004). The authors have no conflicts of interest to declare.