The Pseudomonas syringae Type III Effector HopF2 Suppresses Arabidopsis Stomatal Immunity

Pseudomonas syringae subverts plant immune signalling through injection of type III secreted effectors (T3SE) into host cells. The T3SE HopF2 can disable Arabidopsis immunity through Its ADP-ribosyltransferase activity. Proteomic analysis of HopF2 interacting proteins identified a protein complex containing ATPases required for regulating stomatal aperture, suggesting HopF2 may manipulate stomatal immunity. Here we report HopF2 can inhibit stomatal immunity independent of its ADP-ribosyltransferase activity. Transgenic expression of HopF2 in Arabidopsis inhibits stomatal closing in response to P. syringae and increases the virulence of surface inoculated P. syringae. Further, transgenic expression of HopF2 inhibits flg22 induced reactive oxygen species production. Intriguingly, ADP-ribosyltransferase activity is dispensable for inhibiting stomatal immunity and flg22 induced reactive oxygen species. Together, this implies HopF2 may be a bifunctional T3SE with ADP-ribosyltransferase activity required for inhibiting apoplastic immunity and an independent function required to inhibit stomatal immunity.


Introduction
Pseudomonas syringae is a prominent microbial inhabitant of freshwater ecosystems that has evolved to proliferate and causes disease in plants [1]. In association with plants P. syringae survives as an epiphyte on the leaf surface and subsequently enters plant tissues via wounds or natural openings such as stomata. Inside plant tissues P. syringae proliferates extracellularly in the apoplastic spaces of leaf or vascular tissues and severely compromises plant fitness. The plant innate immune system limits infection by preventing the invasion of the apoplastic space by epiphytes as well as by compromising the proliferation of the pathogen postinvasively. Both levels of plant immunity are activated by the recognition of microbe-associated molecules termed microbial-or pathogen-associated molecular patterns (MAMPs or PAMPs) by membrane-associated pattern recognition receptors (PRRs); a process termed PAMP-triggered immunity (PTI). Preinvasively, PTI activates a stomatal immune response resulting in stomatal closure which limits microbial ingress into the apoplastic space [2]. Post-invasively, PTI results in transcriptional reprogramming and the secretion of antimicrobials and cell wall associated compounds such as callose into the apoplastic space [3]. Both levels of PTI involve the production of reactive oxygen species (ROS) and the activation of MAP kinase signalling cascades [3,4]. Although these two levels of PTI are effective at preventing the invasion and proliferation of most microbes, pathogens such as P. syringae have evolved strategies to overcome both pre-and post-invasive PTI.
Many pathovars of P. syringae produce the phytotoxin coronatine which enables them to reopen stomata closed by PTI, thereby allowing the pathogen to invade the plant tissue [2]. As a result, P. syringae strains that have been deprived of coronatine production do not efficiently invade and proliferate in the apoplast when surface inoculated [2]. In addition to disabling stomatal immunity, coronatine also inhibits apoplastic immunity through suppression of salicylic acid mediated defenses [5,6]. Apoplastic virulence of P. syringae also heavily relies on the type III secretion system (T3SS) to inject effector proteins into host cells. A major function of type III secreted effectors (T3SEs) is the suppression of PTI by targeting PRR complexes and/or downstream signalling components [7,8].
Although the role of T3Es in apoplastic PTI suppression is well established, evidence of their ability to influence stomatal immunity is limited. Recognition of the T3SE AvrRpt2 by the Arabidopsis thaliana (hereafter Arabidopsis) RPS2 NB-LRR (NLR) protein activates effector-triggered immunity (ETI) that is accompanied by stomatal closure [2,9]. More recently, the P. syringae T3SE HopM1 has been demonstrated to suppress PAMP-induced stomatal closure when expressed transgenically in Arabidopsis demonstrating that T3SEs can influence stomatal immunity [10]. In addition, the T3SE HopX1 can promote the growth of COR-deficient P. syringae pv. tomato DC3118 (Pto DC3118) and also promote stomatal opening [11].
The HopF T3SE family is broadly distributed among pathovars of P. syringae. HopF1 from the bean pathogen P. syringae pv. phaseolicola 1449B is recognized in bean cultivars expressing the R1 resistance (NLR) protein [12,13]. The crystal structure of HopF1 revealed that it adopts a mushroom-like shape with ''head'' and ''stalk'' subdomains [14]. The head subdomain possesses limited structural similarity to ADP-ribosyltranferases such as Diphtheria toxin. This similarity was used to predict functional residues that were demonstrated to be required for virulence and avirulence functions of HopF1 on susceptible and resistant bean plants, respectively [14]. Equivalent residues have also been demonstrated to be required for the virulence functions of HopF2 from Pto DC3000 (hereafter HopF2) in tomato and Arabidopsis [15,16]. These virulence functions likely involve interaction and modification of the PRR-associated receptor-like kinase BAK1, MAP kinase kinases and the PTI/ETI regulator RIN4 [15][16][17]. HopF2 has been demonstrated to ribosylate MKK5 and RIN4 in vitro and requires RIN4 for apoplastic growth promotion of P. syringae in Arabidopsis [15,16]. HopF2 has also been demonstrated to suppress the ETI-response induced by the T3SE AvrRpt2 [16]. Importantly, all of the HopF2 activities described above require intact catalytic residues of the head-domain.
Here we demonstrate that HopF2 can suppress stomatal immunity in Arabidopsis. Unlike the functions ascribed to HopF2 thus far, suppression of stomatal immunity does not require catalytic residues of the head-domain. These results indicate that HopF2 has distinct molecular functions for the manipulation of apoplastic and stomatal immunity.

Bacterial strains, Plasmids and Plant Materials and Methods
Arabidopsis thaliana ecotype Col-0 were routinely grown with 9 h of light (130 microeinsteins m 22 s 21 ) and 15 h of darkness at 22˚C in Sunshine Professional Growing Mix LC1 (SunGro, Canada) soil supplemented with 20:20:20 fertilizer.

Pseudomonas syringae in planta Growth Assay and Measurements of Stomatal Aperture
Growth assay with Pto DC3000 or Pto DC3118 were performed on 4 week old Arabidopsis [26,27]. Surface inoculation was performed by dipping Arabidopsis grown in mesh enclosed pots into P. syringae suspensions according to Melotto et al. [2]. Bacterial in planta growth assay by syringe inoculation was performed according to Wilton et al. [16,28]. Stomatal apertures were measured according to Liu et al. for P. syringae induced closure and re-opening [29]. ABA-induced stomatal closure was analyzed on true leaves of two week old Arabidopsis leaves. Leaves were floated on stomata opening buffer (5 mM KCl, 50 uM CaCl2, 10 mM MES-KOH, pH 6.15) overnight and incubated in light (100 microeinsteins m 22 s 21 ) for four hours to induce uniform stomatal opening. A final concentration of 10 mM ABA was added to each well and incubated for 2 hours. Stomatal apertures were then measured according to Liu et al. [29].

Isolation of High Molecular Weight HopF2 Complexes
HopF2 complexes were isolated from 4 week old Arabidopsis. HopF2-HA transgenic Arabidopsis rosettes were snap frozen in liquid nitrogen then homogenized in of ice-cold extraction buffer (1 g FW: 2 ml EB) composed of 50 mM HEPES-KOH (pH 7.5), 50 mM NaCl, 2 mM EDTA, 1 mM DTT, 0.2% Triton-X 100, 0.1 mg/ml dextran and 1:100 (v/v) plant protease inhibitor (Sigma, P9599). Homogenates were clarified by centrifugation at 4˚C and 10,0006g for 10 min to remove cell debris and insoluble particulates, and supernatants reserved as clarified extract.
Pooled concentrated fractions (,10 ml) were incubated with 150 ul mMACS anti-HA resin for 3 hours, then immobilized on a paramagnetic column (Miltenyi, M column, 130-042-801) For immunoblot analysis of purifications, immobilized resin was washed with 10 bed volumes low salt wash (LSW, 50 mM HEPES-KOH (pH 7.5), 50 mM NaCl, 1 mM DTT) and 10 bed volumes no salt wash (NSW, 50 mM HEPES-KOH (pH 7.5), 1 mM DTT), then eluted over three bed volumes with 0.1 M NH 4 OH. Samples for proteomic analyses were washed three times with 10 bed volumes PBS to remove detergent and eluted over three bed volumes with 250 mg/ml HA peptide in PBS.

Proteomic Methods
Proteins in HA peptide eluates were precipitated by addition of trichloroacetic acid to a final concentration of 20% (w/v), incubated overnight at 4˚C and pelleted by centrifugation (30 min, 20,0006g). Pellets were washed once with 10% TCA, three times with ice cold acetone then solubilized in 50 mM ammonium bicarbonate and digested with trypsin. Spectra were developed using a Thermo Scientific LTQ-XL MS/MS. MS/MS spectra were analyzed and protein identities assigned using SeQuest against the Arabidopsis UniProt FASTA database [30]. Protein identities were verified by Petide/Protein Prophet using a probability score of 0.95 [31].

MAPK Phosphorylation Status
Arabidopsis seedlings were grown for 11 days on 0.5x MS salts with Gamborg's vitamins (M0404; Sigma) media solidified with 0.8% agar then transferred to six well plates (six seedlings per well) containing 3 ml of liquid medium containing 0.5x MS salts with 30 mM DEX (for +DEX treatment) or 0.5xMS salt (-DEX). Seedlings were gently shaken in 12 h light, 12 h dark cycle at day/night temperature regime of 22˚C/18˚for. After 24 h, flg22 peptide was added to a final concentration of 1 mM. Shaking was maintained for 20 minutes then seedlings were harvested and snap frozen in liquid nitrogen. Frozen tissue was homogenized in 100 ml of extraction buffer (100 mM HEPES, pH 7.5, 5 mM EDTA, 5 mM EGTA, 2 mM dithiothreitol, 10 mM Na 3 VO 4 , 10 mM NaF, 50 mM ß-glycerolphosphate, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol, 1% (w/v) polyvinylpolypyrrolidone). After centrifugation at 13,000 rpm for 30 min at 4˚C, supernatants were reserved as clarified extract. Protein concentrations of clarified extracts were determined using a Bradford assay (BIO-RAD, Hercules, CA, USA) and twenty mg of total protein was separated by SDS-PAGE and immunoblotted.

ROS Generation Assay
Measurement of ROS production was performed using a modified version of the luminol-based assay as described [35] A series of 20 leaf discs (diameter 4 mm) were taken from the leaves of 4 week-old Arabidopsis. Leaf disks were divided with 10 disks placed in each of two wells of a 96-well plate and washed with 200 ml of sterile water for 20 hours. The water was removed and replaced with 100 mM Tris-HCl (pH 8.0) containing 20 mg/ml horseradish peroxidase and 34 mg/ml luminol. Each matched pair of wells from a single plant was treated with either water or 2 mM flg22 peptide. The luminescence was measured for a 2 second interval every 2 minutes for a total of 60 minutes on an Infinite M1000Pro microplate reader (TECAN Group Ltd.). Each treatment was performed on three individual plants.

Analysis of HopF2 protein complexes
To determine if HopF2 targets high molecular weight (Mr) complexes, protein extracts from Arabidopsis carrying dexamethasone (DEX) inducible HopF2:HA transgene were resolved by gel filtration chromatography using fast protein liquid chromatography (FPLC, Fig. 1a). Fractions were collected from the void volume (V o ; 36 ml) and every second fraction was analyzed by immunoblot (IB; Fig. 1B) for the presence of HopF2:HA and RIN4, a known interactor of HopF2 [16]. Two peaks containing HopF2:HA were identified with one peak eluting at high molecular weight (Mr; Elution volume (Ve) 36-46 ml, solid arrow Fig. 1) and one eluting around the Mr of monomeric HopF2:HA (Ve 70.5-82.5, dashed arrow Fig. 1). This implies that two pools of HopF2:::HA are present -one associating with high Mr plant protein complexes and one as monomeric HopF2, making gel filtration a viable strategy for enriching HopF2/target protein complexes from HopF2 unbound to targets. Two HopF2 bands were detected in high Mr fractions; one corresponding to the predicted size of the full length protein (,25 kDa) and one lower Mr product (,15 kDa). Interestingly, HopF2 in the lower Mr fractions migrated as a doublet between 15 and 25 kDa suggesting that this form of HopF2 is distinct from those found in the high Mr fractions. RIN4 was found to co-elute with HopF2:HA in high Mr fractions, implying RIN4 exists as a member of a high Mr complex. The elution profile of RIN4 was independent of HopF2:HA expression since RIN4 elutes at the column V 0 in both untreated and DEX treated HopF2:HA transgenic Arabidopsis (S1 Figure).
Next we tested whether HopF2:HA and RIN4 interact in high Mr fractions. We performed immunoaffinity purification of HopF2:HA from concentrated pooled high Mr FPLC fractions ([pool]; Ve 36-46) using anti-(HA) IgG coupled to paramagnetic beads (mMACS a HA IP, Fig. 2). For both RIN4 and HopF2, immunoreactive bands co-migrating with bands from the mMACS a HA IP input Washing with low salt (LSW) or no salt (NSW) failed to elute either protein, suggesting a bona fide protein-protein interaction. In contrast, mMACS a HA IP performed on high Mr FPLC fractions from uninduced (-DEX) Arabidopsis showed no HopF2:HA reactive bands in input, washes or eluate and no RIN4 immunoreactive bands in mMACS a HA IP eluates despite overexposure of blots ( Fig. 2, lower panel). Therefore, isolation of RIN4 was dependent on the presence HopF2:HA, suggesting that HopF2 and RIN4 interact in high Mr complexes isolated from Arabidopsis.
Next we sought to determine what other proteins are present in the high Mr HopF2/RIN4 complex using proteomic methods. Eluates from mMACS a HA IP were analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) to determine the identities of proteins co-purified with HopF2:HA. Protein ID by LC-MS/MS of samples eluted with NH 4 OH was confounded by spectra dominated by non-specific contaminants, such as rubisco and light harvesting complex proteins. Therefore, to improve the purity of mMACS a HA IP for protein identification by LC-MS/MS elutions were attempted with HA peptide in phosphate buffered saline (PBS) [36,37]. Elutions performed with HA peptide in PBS (PBS) produced eluates enriched for HopF2:HA and amenable to proteomic identification (S2 Figure). Three mMACS a HA IP experiments from HopF2:HA induced (+ DEX) and three experiments from uninduced (-DEX) Arabidopsis were analyzed by LC-MS/MS. Proteins identities were assigned using SeQuest to search the UniProt Arabidopsis FASTA database (S1 Table). Proteins were considered as HopF2 complex members if they were identified with at least two unique peptides (p,0.05) in every + DEX HopF2 experiment and zero protein identities in any -DEX experiments (Table 1).
Our proteomic analysis identified ten proteins as members of a high Mr complex co-purified with HopF2:HA. All are membrane associated proteins, which is consistent with the predicted myristoylation of HopF2 and its membrane localization in planta [16,38]. Although RIN4 was not identified by our mass spectrometry analysis, AHA2 and ERD4 are known members of a RIN4 complex that mediates the closing of stomata in response to P. syringae [29]. AtHIR2 is an RPS2 interacting protein identified through proteomic methods [39,40]. AtHIR2 and AtREM1.3 are thought to play a role in the formation of plasma membrane subdomains in response to pathogen perception [40,41] while AtPDR8 and AHA2 become phosphorylated in flg22-treated Arabidopsis cell culture [42]. Both PIP2a and PIP3 show increased phosphorylated after ABA treatment [43]. It would therefore appear that HopF2 interacts with a membrane localized complex that is involved in the response to pathogens. In particular, the presence of AHA2 in the HopF2 complex led us to hypothesize that HopF2 can manipulate stomatal immunity.

HopF2 inhibits stomatal immunity
To investigate whether HopF2 can alter stomatal immunity we determined if HopF2 expression can alter the stomatal response to pathogens in Arabidopsis. Leaf discs prepared from Col-0 and transgenic Arabidopsis expressing HopF2:HA were floated in water or bacterial suspensions of Pto DC3000. Epidermal peels were produced and used to determine stomatal aperture after 1 and 3 hours (Fig. 3). While treatment with Pto DC3000 caused a significant two-fold reduction in stomatal aperture after one hour relative to the water treated control, Arabidopsis expressing HopF2:HA treated with Pto DC3000 did not show a reduction in stomatal aperture and instead had significantly wider stomatal apertures than the water control (Fig. 3a). Next, we monitored the stomatal response of transgenic Arabidopsis expressing mutant HopF2 D175A :HA, which lacks residues required for ADP-ribosyl transferase (ADP-RT) activity [15]. Interestingly, HopF2 D175A :HA expressing plants showed no difference between Pto DC3000-treated and water-treated controls (Fig. 3b). After three hours, stomatal aperture of Pto DC3000 treated plants increased 1.6 fold while plants expressing HopF2:HA and HopF2 D175A :HA continued to show no difference to water controls (Fig. 3). Therefore expression of HopF2 or HopF2 D175A results in stomata insensitive to PAMP induced stomatal closure, indicative of disabled stomatal immunity. This loss of stomatal immunity was not due to disruption of stomatal function since HopF2-expressing plants still closed stomata in response to ABA treatment (Fig. 3c).
HopF2 promotes the growth of coronatine-deficient P. syringae independently of D175 and RIN4 To further investigate whether HopF2 can compromise stomatal immunity we investigated whether transgenic expression of HopF2 in Arabidopsis can alter the growth of surface inoculated P. syringae (Fig. 4). The phytotoxin coronatine Counts are the number of unique peptides identified in each + DEX HopF2 experiment. Protein ID required P,0.01 (Peptide Prophet), minimum two unique peptides in each experiment and zero peptides in any negative control. None of these proteins were identified in any -DEX experiment. Bold text indicates proteins associated with RIN4 or RPS2 protein complexes [29,39]. *Proteins identified as differentially phosphorylated in response to flg22 [42]. **Proteins identified as differentially phosphorylated in response to ABA [43]. { Proteins enriched in plasma membrane subdomains after flg22 treatment [41].
doi:10.1371/journal.pone.0114921.t001  Pseudomonas syringae and Stomatal Immunity produced by Pto DC3000 disables stomatal immunity, preventing stomatal closure in response to bacterial perception [44]. Consequently, Pto DC3000 grows well when surface inoculated on Arabidopsis while the coronatine deficient P. syringae pv. tomato D3118 (Pto DC3118) grows poorly by comparison [2]. If HopF2 can manipulate stomatal immunity, transgenic expression of HopF2 should restore virulence of surface inoculated Pto DC3118. Therefore we carried out bacterial in planta growth assays by dip inoculation with Pto DC3000 and Pto DC3118 on +/2 DEX HopF2:HA-expressing Arabidopsis (Fig. 4a). After three days of growth in -DEX Arabidopsis, Pto DC3000 grew to significantly higher levels than Pto DC3118 (Fig. 4a). However, there was no significant difference in growth between Pto DC3000 and Pto DC3118 in + DEX Arabidopsis expressing HopF2:HA (Fig. 4a). Bacterial growth was also assayed two days after inoculation with identical trends, implying that differences in bacterial growth are not due to apoplastic proliferation late in the time course (S3 Figure). Interestingly, transgenic expression of HopF2 D175A :HA also increases the virulence of surface inoculated Pto DC3118 (Fig. 4b). Therefore, HopF2 can promote the virulence of coronatine-deficient P. syringae independently of the D175 catalytic residue.
To investigate if RIN4 is required for the ability of HopF2 to promote Pto DC3118 growth, DEX-inducible HopF2 transgenic plants were generated by crossing the DEX inducible transgene into rin4/rps2 mutant Arabidopsis plants [24]. Plants were surface inoculated with Pto DC3000 or Pto DC3118 and growth was measured +/2 HopF2 expression. In the absence of HopF2 expression Pto DC3000 grew 1.22 logs more than Pto DC3118 over three days. Whereas HopF2:HA expression promoted the growth of Pto DC3118 and Pto DC300 to similar levels. Expression of HopF2:HA also significantly increases the growth of Pto DC3118 in rin4/rps2 plants (white bars, Fig. 4c, t-test, p,0.05). Together, this demonstrates that HopF2 can promote the growth of the coronatine deficient strain Pto DC3118 independently of RIN4 (Fig. 4c).

The HopF2 catalytic residue D175 is differentially required to block PTI responses
The ADP-ribosyltransferase activity of HopF2 mediates its virulence function in disabling apoplastic immunity [15,16]. Further, HopF2 is known to ADPribosylate and disable MAP kinases downstream of flg22 perception, as well as RIN4 [15]. Consistent with this we were able to demonstrate that transgenic expression of HopF2:HA prevents phosphorylation of MPK3/6 after flg22 treatment in a catalytic residue-dependent manner since HopF2 D175A :HA did not prevent MPK3/6 phosphorylation (Fig. 5a). In addition, D175 was required for the ability of HopF2 to block flg22-induced callose deposition and promote the growth of the Pto DC3000DhrcC mutant which lacks a functional T3SS and is unable to suppress PTI (Fig. 5b, c). Perception of flg22 is also known to trigger an increase in the production of oxidative intermediates that help signal stomatal closure [10]. We therefore sought to determine if expression of HopF2 can alter the flg22 induced oxidative burst using a luminol-dependent chemiluminescence Values are means ¡ SEM from n512 leaves. Asterisk denotes statistically significant differences between day 3 samples as determined by a pairwise Student's t-test (p,0.05). Results are representative of 3 independent replicates. (d) Transgenic expression of HopF2 inhibits flg22 triggered ROS production. Four week old transgenic Arabidopsis expressing HopF2 were sprayed with 30 mM DEX (+DEX) or water (-DEX) 24 h prior to inoculation and leaf discs floated in 2 mM flg22. ROS production was quantified using luminol and horseradish peroxidase with luminescence recorded every two minutes for 60 min. Values are mean sums of total luminescence ¡ SEM from from n56 wells. Asterisk denotes statistically significant differences between DEX treatment samples as determined by a pairwise Student's t-test (P,0.05).
doi:10.1371/journal.pone.0114921.g005 assay. Transgenic HopF2:HA and HopF2 D175A :HA plants were induced with dexamethasone 24 hours before conducting the ROS assay. Application of flg22 strongly induced ROS production in Col-0 and transgenic plants without DEX treatment, whereas expression of either HopF2:HA or HopF2 D175A :HA blocked ROS production. Therefore, HopF2 can block the PTI-associated ROS burst independently of the catalytic residue D175 (Fig. 5d).

Discussion
In this study we demonstrated that HopF2 can block stomatal immunity independently of its ADP-ribosyltransferase activity [15]. This activity was independent of RIN4 and was not due to a general disruption of stomatal function since HopF2 did not block ABA-induced stomatal closure (Fig. 3c).
We used a proteomic approach to identify a HopF2 associated protein complex in Arabidopsis. This complex contained two RIN4-associated proteins (AHA2 and ERD4), and one RPS2-associated protein (AtHIR2), supporting our previous report that HopF2 targets RIN4 [16,29,39]. Our method failed to identify RIN4, RPM1 or RPS2 directly as members of this complex, although RIN4 could be detected by immunoblotting (Fig. 1). Although RPS2 was previously detected in purified RIN4 complexes, RPM1 was not [29]. Proteomic investigation of RPS2 interacting proteins also failed to reliably identify RIN4, despite being readily detectable by immunoblot [39]. The small size and high basic amino acid content of RIN4 may make it difficult to detect by standard proteomic methods. A failure to identify the RIN4 interacting NLR-proteins RPM1 or RPS2 might also be due to low NLR-protein expression levels relative to RIN4 or ATPase proteins and/or the presence of RIN4 complexes not associated with these NLR-proteins [29]. Our approach may also decrease the possibility of discovering NLR-proteins since our bait (HopF2) does not directly interact with these proteins, and requires RIN4 to ''bridge'' interactions with NLR-proteins in a tertiary complex. Nevertheless, our proteomic approach supports that HopF2 forms part of a high Mr RIN4containing complex in planta.
Two populations of HopF2 were identified from our size exclusion chromatography analysis. One population elutes in high Mr fractions and contained RIN4, whereas another population elutes in low Mr fractions corresponding to the size of monomeric HopF2 (Fig. 1). The high Mr HopF2 population mirrored the distribution of RIN4 and spanned molecular weights of 300 kDa to approximately 1000 kDa. This population may represent one complex or multiple large complexes. Since these complexes are likely to membrane associated, their size may reflect complexes held together by proteins and/or lipids that are maintained during the extraction procedure [41]. Directed binary interaction maps will help resolve which proteins from these complexes can directly interact versus those that may co-localize to the same membrane microdomains.
It is interesting to note that the HopF2 eluting in lower Mr fractions following size exclusion chromatography was smaller than HopF2 from high Mr fractions according to their SDS-PAGE migration rates (Fig. 1b). High Mr fractions contained HopF2 that migrated at the expected 25 kDa with a lower band at about 15 kDa, whereas HopF2 from low Mr fractions migrated as a doublet between 15 and 25 kDa, with no band corresponding to predicted full length HopF2. Since the HA tag used for immunoblotting is translationally fused to the C-terminus, the lower molecular weight HopF2 likely represents an N-terminal truncation. An N-terminal truncation of HopF2 would remove the myristoylation sequence which is required for its membrane localization [38]. As such, this may represent a population of HopF2 that is not membrane associated in planta thereby expanding its range of targets to include soluble proteins.
Our proteomics approach led us to hypothesize and demonstrate that HopF2 can block stomatal immunity. Transgenic expression of HopF2 can inhibit stomatal closing triggered by P. syringae (Fig. 3) and improves the virulence of surface inoculated coronatine deficient P. syringae strain Pto DC3118 (Fig. 4). It is possible that some virulence promotion of surface inoculated Pto DC3118 is mediated by HopF2's ability to inhibit apoplastic immunity (Fig. 5a-c) as opposed to increased stomatal invasion. However, HopF2's ability to promote the virulence of syringe inoculated P. syringae depends on the catalytic residue D175, as well as the host protein RIN4 [16]. Consistent with this, transgenic expression of HopF2 D175A does not increase the growth of syringe inoculated mutant P. syringae lacking a functional T3SS (Fig. 5c). In contrast, transgenic expression of HopF2 D175A or HopF2 in rin4/rps2 Arabidopsis promotes the virulence of surface inoculated Pto DC3118 (Fig. 4b, c). Therefore, HopF2 can inhibit stomatal immunity independent of the molecular mechanisms required for its inhibition of apoplastic immunity.
Two early PTI signalling events involved in stomatal immunity, MAP kinase activation and ROS production, were differentially inhibited by transgenic expression of HopF2 (Fig. 5). Inhibition of MAP kinase activation induced by the flagellar PAMP flg22 was dependent on the catalytic D175 residue, whereas inhibition of ROS production was not. MAPK activation and ROS production represent independent, early PTI signalling events induced by flg22 perception [45]. The ADP-RT dependent inhibition of MAP kinase signalling by HopF2 likely involves its direct targeting of MAPKKs or the PRR co-receptor BAK1 [15,17].
A salient question that remains is what protein is targeted by HopF2 to suppress stomatal immunity? It is unlikely to be RIN4 since HopF2 can promote Pto DC3118 growth in rps2/rin4 plants (Fig. 4c). It is also unlikely to be the H+-ATPases AHA1/2 since their disruption would likely alter ABA-responsiveness [46]. The ability of HopF2 to block ROS production independently of its ADP-RT activity would indicate that it targets a signalling component that influences ROS production but not MAPK activation in order to block stomatal immunity. Potential targets include the NADPH oxidase RBOHD, the RLCKs BSK1 and PBL1, and/or the aspartate oxidase FIN4 [47][48][49][50][51][52].
If HopF2 does target a protein involved in the early bifurcation of PRRsignaling in order to suppress stomatal immunity, what is its mode of action? The crystal structure of HopF1 revealed that it adopts a mushroom-like structure with a head subdomain that possess the ADP-RT catalytic residues and a ''stalk'' subdomain with no obvious structural similarity to known enzymes. Both subdomains contained surface-exposed residues that were important for both virulence and avirulence activities in bean [14]. HopF2 is predicted to adopt a similar structure to HopF1 Pph1449B (Phyre2, identity554%, confidence5100.0) [53]. Since the suppression of stomatal immunity by the HopF2 is independent of the ADP-RT catalytic activity of the ''head'' domain, it is possible that residues of the ''stalk'' are required for this function, in which case HopF2 would represent a bifunctional effector similar to the Pseudomonas aeruginosa T3SE ExoT. ExoT has an N-terminal GTPase-activating protein (GAP) domain and a C-terminal ADP-RT domain [54,55]. Although the predicted head and stalk subdomains of HopF2 would not be separable as C-and N-terminal fragments, it is possible that HopF2 has evolved a mushroom-like structure to accommodate two independent functions. An important next step to characterizing the ADP-RT-independent function will involve the identification of HopF2 residues required for its ability to block stomatal immunity. In addition, it remains to be determined whether HopF2 carries out this second function directly in guard cells or influences stomatal immunity from non-stomatal cells. There is evidence that the T3SS functions during epiphytic growth [56]. However, it remains to be determined whether T3SEs are injected into guard cells and whether they can function in guard cells, for example, when expressed from guard cell specific promoters.
Overall, our results support the hypothesis that HopF2 is a bifunctional T3SE. One function involves its ADP-RT activity that is required for its ability to block PTI-activated MAP kinase signaling and to promote post-invasive apoplastic growth in a RIN4-dependent manner [15,16]. A second function is ADP-RT independent and involves the suppression of PRR-triggered stomatal-immunity in a RIN4-independent manner. Therefore HopF2 has separable PTI suppressing functions and represents an appealing probe to study signal branching following PRR activation.
Supporting Information S1 Figure. RIN4 exists in a high Mr complex independent of HopF2 expression. Clarified extracts from induced (+DEX) and induced (-DEX) plants expressing HopF2:HA under the control of a dexamethasone inducible promoter were subjected to gel filtration chromatography on a Sephacryl S-300 HR 16/60 column. Every second fraction from the void volume was resolved by SDS-PAGE and immunoblotted with anti RIN4 immunosera and anti HA IgG. doi:10.1371/journal.pone.0114921.s001 (TIF) S2 Figure. HopF2 is successfully eluted with HA peptide in PBS. High molecular weight FPLC fractions containing HA immunoreactive (V e 536-46 ml) bands were pooled and concentrated for one hour in a 10,000 Mr cutoff concentrator. Concentrate was incubated with anti HA magnetic resin. HA resin was immobilized and complexes were eluted over three bed volumes of 250 mg/ml HA peptide in PBS. Samples were resolved by SDS-PAGE and immunoblotted with anti RIN4 immunosera and anti HA IgG.