The Pseudomonas type III effector HopQ1 activates cytokinin signaling and interferes with plant innate immunity

Authors


Summary

  • We characterized the molecular function of the Pseudomonas syringae pv. tomato DC3000 (Pto) effector HopQ1.
  • In silico studies suggest that HopQ1 might possess nucleoside hydrolase activity based on the presence of a characteristic aspartate motif. Transgenic Arabidopsis lines expressing HopQ1 or HopQ1 aspartate mutant variants were characterized with respect to flagellin triggered immunity, phenotype and changes in phytohormone content by high-performance liquid chromatography-MS (HPLC-MS).
  • We found that HopQ1, but not its aspartate mutants, suppressed all tested immunity marker assays. Suppression of immunity was the result of a lack of the flagellin receptor FLS2, whose gene expression was abolished by HopQ1 in a promoter-dependent manner. Furthermore, HopQ1 induced cytokinin signaling in Arabidopsis and the elevation in cytokinin signaling appears to be responsible for the attenuation of FLS2 expression.
  • We conclude that HopQ1 can activate cytokinin signaling and that moderate activation of cytokinin signaling leads to suppression of FLS2 accumulation and thus defense signaling.

Introduction

Most bacterial plant pathogens remain extracellular throughout their life cycle. They enter the leaf through natural openings such as wounds and stomata. Once inside the plant, they multiply in the apoplast and form dense colonies attached to the cell walls (Soylu et al., 2005). However, the presence of microbes in the apoplast is betrayed by microbe-associated molecular patterns (MAMPs), which are essential to microbes but absent from the host (Boller & He, 2009). MAMPs are perceived by membrane-localized pattern recognition receptors (PRRs). Activation of PRRs leads to a whole range of immune responses, including an increase in cytosolic Ca2+, the generation of reactive oxygen species (ROS), activation of a mitogen-activated protein kinase (MAPK) cascade and transcriptional changes. These responses lead collectively to pattern-triggered immunity (PTI) (Chisholm et al., 2006). To overcome PTI, phytopathogens secrete effector proteins that interfere with plant innate immunity at several levels (Bozkurt et al., 2011; Hann et al., 2011; de Jonge et al., 2011; Feng & Zhou, 2012). If the plant immune response is successfully suppressed by effector proteins, disease occurs. However, in some cases, the presence of effectors is detected by resistance (R) proteins, which leads to a very strong immune response typically associated with programmed cell death and is often called effector-triggered immunity (ETI) (Jones & Dangl, 2006). In contrast to PTI, which is typically effective against a broad range of microbes, ETI is usually induced in a race- and cultivar-specific manner in which a particular effector is specifically recognized by a corresponding R gene.

A well-studied PRR that was first characterized in the model plant Arabidopsis thaliana is the FLAGELLIN-SENSING 2 (FLS2) receptor for bacterial flagellin or its active epitope flg22 (Boller & Felix, 2009). FLS2 is a leucine-rich repeat receptor kinase (LRR-RK), which rapidly associates with another receptor kinase called BAK1 (for BRI1-ASSOCIATED KINASE) in a ligand-dependent manner (Chinchilla et al., 2007; Heese et al., 2007). Signal transduction is initiated rapidly upon complex formation, leading to PTI. BAK1 was originally identified as an interactor of the plant growth hormone receptor BRASSINOSTEROID (BR) INSENSITIVE 1 (BRI1) and is now known to be a coreceptor of several other LRR-RKs involved in PTI (Li et al., 2002; Chinchilla et al., 2007; Heese et al., 2007; Wang et al., 2008; Schulze et al., 2010; Roux et al., 2011). Besides this potential link of PTI to BR signaling, evidence is accumulating suggesting intimate links between innate immunity and phytohormone signaling (Robert-Seilaniantz et al., 2007; Boutrot et al., 2010; Choi et al., 2011a). One hormone class is the cytokinins (CKs), which are derived from aminopurines and regulate a plethora of physiological and developmental processes (Sakakibara, 2006). Various plant-associated microbes produce CKs, causing the formation of galls, green islands or disruption to primary carbon metabolism (Costacurta & Vanderleyden, 1995; Ortiz-Castro et al., 2009). Increases in endogenous CK production enhances resistance against the bacterial pathogen Pseudomonas syringae strains pv. tomato (Pto) and pv. tabaci (Pta) in Arabidopsis and Nicotiana benthamiana, respectively (Choi et al., 2011a,b; Grosskinsky et al., 2011; Argueso et al., 2012). However, CK treatments can also lead to increased susceptibility of wheat (Triticum aestivum) to powdery mildew (Blumeria graminins) infection when applied at low concentrations (0.25–1.5 μM) (Babosha, 2009). Overall, besides physiological changes mediated by CK signaling, different CK concentrations also affect resistance in different plant hosts and to a variety of pathogens.

As for PTI responses per se, microbes have evolved mechanisms to counteract host-induced changes in phytohormone concentrations associated with immunity. For example, bacterial phytopathogens such as Pto secrete effectors into the host cell by a needle-like structure known as the type III secretion system (TTSS, or T3S) to suppress PTI and modify hormone signaling (Hann & Rathjen, 2011; Feng & Zhou, 2012). Pto delivers c. 30 type III effectors, one of which is HopQ1. HopQ1 is widely conserved across different pathogenic bacterial genera, including Xanthomonales and Ralstoniales. HopQ1 is a determinant of Pto DC3000 host range and is recognized by an as yet unidentified R protein in N. benthamiana leading to ETI (Wei et al., 2007; Ferrante et al., 2009). Two very recent reports also show that HopQ1 interacts with 14-3-3 proteins in tomato and bean, and is phosphorylated at its N-terminus (Giska et al., 2013; Li et al., 2013b). Mutations in the phosphorylation site affect the virulence of Pto DC3000, suggesting that HopQ1 interaction with 14-3-3 proteins might be necessary for its function. In addition, it has recently been published that HopQ1 promotes bacterial virulence dependent on its putative nucleoside hydrolase (NH) domain, in both tomato and Arabidopsis (Li et al., 2013a). At the same time, putative catalytic residues of the NH domain were dispensable for ETI induction in tobacco plants. However, a role in PTI suppression has not been clearly demonstrated for HopQ1.

We addressed this question by generating transgenic Arabidopsis plants expressing HopQ1 from the β-estradiol-inducible XVE promoter. HopQ1 contains a conserved D101XXXDXDD motif, which is similar to the conserved aspartate motif present in NHs, representing the nucleoside binding site. This atypical aspartate motif is highly conserved among a wide range of HopQ1 homologs present in different families of phytopathogens, including Xanthomonales and Ralstoniales. We produced Arabidopsis transgenic plants expressing HopQ1 mutant variants with different substitutions in the aspartate motif. To test the requirement of the NH domain for HopQ1, we made two mutant genes encoding HopQ1-D101A and HopQ1-D107/108A proteins, and expressed them inducibly in transgenic Arabidopsis plants as for HopQ1. Here we show that, in planta, HopQ1, but not the aspartate substitution mutants, suppressed flg22-induced PTI responses such as ROS generation and MAPK activation. This phenomenon occurred through reduced expression of FLS2 mRNA and a concomitant decrease in FLS2 protein accumulation, in a promoter-dependent manner. In addition, HopQ1 expression resulted in elevated concentrations of endogenous CKs as well as induction of CK signaling. Results obtained by prolonged exogenous application of CKs further suggest that attenuation of FLS2 mRNA levels was mediated by activation of CK signaling. We propose that HopQ1 acts as a CK-activating enzyme by cleaving the aminopurine-ribose precursor molecules of CK, and that induction of CK signaling leads to suppression of PTI via down-regulation of FLS2.

Materials and Methods

Strains, plasmids and primers

Bacterial strains and plasmids used for this study are listed in the Supporting Information, Table S1. Plasmids were constructed according to standard methods (GATEWAY®; Invitrogen) using pDONR207 as a donor plasmid and either pMDC7 or pGWB417 as acceptor plasmids (Table S1); corresponding PCR primers are listed in Table S2.

Cytokinins

The CK derivative N6-isopentenyladenosine-5′-monophosphate (IPRMP) sodium salt was produced by OlChemIm Ltd (Olomouc, Czech Republic), and the CK derivatives trans-zeatin (tZ), N6-isopentenyladenosine (iP) and trans-zeatinriboside were obtained from Sigma.

Construction of HopQ1-D101A and HopQ1-D107/108A mutants

HopQ1-D101A and HopQ1-D107/108A substitution mutants were generated with GeneArt® Site-Directed Mutagenesis Protocol (Invitrogen) using mutagenic oligonucleotide primers D101A and D107/108A as described in Table S2.

Expression of HopQ1, HopQ1-D101A and HopQ1-D107/108A in Arabidopsis

Plasmids (pMDC7-HopQ1, pMDC7-HopQ1(D101A), pMDC7-HopQ1(D107/108A) (Zuo et al., 2000) were transformed into chemically competent Agrobacterium tumefaciens strain GV3101 and stably transformed into Arabidopsis Col-0 plants by dip inoculation. To induce expression, plant tissues were treated with 10 μM β-estradiol solved in 100% ethanol.

Expression of HopQ1, HopQ1-D101A and HopQ1-D107/108A in N. benthamiana

Plasmids (pGWB417-HopQ1-myc, pGWB417-HopQ1(D101A)-myc, pGWB417-HopQ1(D107/108A)-myc and the empty vector pGWB413-GUS; see Table S1) were transformed into chemically competent A. tumefaciens strain GV3101 by heat shock. Leaves from 4-wk-old Nicotiana benthamiana Domin plants were infiltrated with bacteria (OD600 = 0.2) resuspended in infiltration buffer (10 mM MgCl2, 10 mM MES, pH 5.6). Expression of HopQ1 was detected by immunoblot analysis with anti-myc antibodies at a 1 : 5000 dilution (Sigma). Expression of At FLS2 was detected by immunoblot analysis with polyclonal anti-FLS2 (Chinchilla et al., 2007; Gimenez-Ibanez et al., 2009). Blots were developed with CDPstar reagents (New England Biolabs, Ipswich, MA, USA). Staining of N. benthamiana leaves for cell death detection was performed with Trypan Blue as described previously (Hann & Rathjen, 2007).

Measurement of ROS generation

Leaf disks were floated on water overnight and ROS released by the leaf tissue were measured using a luminol-based chemiluminescent assay (Hann & Rathjen, 2007). ROS were elicited with 1 μM flg22 peptide (QRLSTGSRINSAKDDAAGLQIA) or 100 μg ml−1 chitin (crab shell chitin; Sigma) in all experiments. Mock treatments without flg22 were performed with the BSA/NaCl solution (1% w/v BSA, 100 mM NaCl) used to solubilize flg22. Luminescence was measured over a time period of 30 min using a luminometer (MicroLumat LB96P; EG&G Berthold, Bad Wildbad, Germany). Data from 12 leaf disks derived from four independent infiltrations were statistically analyzed by one-way ANOVA considering  0.05 as significantly different.

MAPK activation assay

Leaf strips pretreated with 10 μM β-estradiol or an equivalent volume of ethanol were induced with 1 μM flg22 peptide or mock-treated with BSA/NaCl for 15 min. Samples were subjected to immunoblot analysis using the anti-p42/44-phospho-ERK antibody (cell signaling). Blots were developed using CDP-star technology (NEB).

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Leaf disks of Col-0 or transgenic lines expressing HopQ1, HopQ1(D101A) or HopQ1(D107/108A) were floated overnight in water and subsequently treated with 1 μM flg22 or mock-treated with BSA/NaCl solution for 30 min before freezing them in liquid nitrogen. Total RNA was extracted using the NucleoSpin RNA Plant extraction kit (Machery-Nagel, Oensingen, Switzerland). The absence of genomic DNA was checked by PCR amplification of the housekeeping ubiquitin gene using 1 μg of RNA (ubiquitin amplification crosses an exon/intron boundary). First-strand cDNA was synthesized from 1 μg of RNA using AMV reverse transcriptase (Promega) and an oligo(dT) primer (Microsynth, Balgach, Switzerland), according to the manufacturer's instructions. For quantitative PCR, 5 μl of a 1 : 100 μl dilution of cDNA were combined with SYBR master mix. PCRs were performed in triplicate with the 7500 Real Time PCR system (Applied Biosystems, Zug, Switzerland). Data were collected and analyzed with the respective ABI analyzing program. The ubiquitin RNA was analyzed as an internal control and used to normalize the values for transcript abundance. Primers for the genes NbEF1α, ubiquitin, AtFLS2, AtCERK1, AtARR5, AtARR6, AtARR2 and NbArr5 are listed in Table S2. Data derived from three biological repeats were statistically analyzed by ANOVA (one-way ANOVA) considering  0.05 as significantly different.

Phytohormone analysis

Phytohormones were extracted by methanol/formic acid/water (15/1/4 v/v/v) from leaf tissues of 6-wk-old plants frozen in liquid nitrogen and lyophilized before analyses and purified using the dual-mode solid-phase method (Dobrev & Kaminek, 2002). Hormonal analysis and quantification was performed by high-performance liquid chromatography (HPLC; Ultimate 3000; Dionex, Reinach, Switzerland) coupled to hybrid triple quadruple/linear ion trap mass spectrometer (3200 Q TRAP; Applied Biosystems) using a multilevel calibration graph with [2H]-labeled internal standards (Dobrev & Vankova, 2012; Djilianov et al., 2013). Determination of phytohormones was done in four biological samples (each injected twice) per control and any of each of the transgenic lines and was repeated twice. Results are presented as means ± SE. The statistical significance of differences between the means for individual phytohormone forms was determined by Student's t-test with < 0.05.

In vitro NH assay

Purified HopQ1 (5 μg) or its mutant derivatives were incubated for 15 min in 20 μl reaction buffer (50 mM HEPES, 100 mM NaCl, 10 mM immidazole pH 8.0 containing 1 mM dithiothreitol and 5 mM CaCl2) supplemented with various concentrations of N6-(2-isopentenyl)adenine-9-riboside-5′-monophosphate at 25°C. The reaction was performed at pH 8 for 15 min at room temperature and terminated by the addition of NaOH to a final concentration of 0.1 N, which results in a pH of c. 13, a value where the absorption differential is most pronounced (Dawson et al., 1968).The release of the ribose was measured as change in absorption at A = 280 nm, as previously described (Shi et al., 1999).

Accession numbers

The accession numbers were as follows: HopQ1, NC_004578.1; At FLS2, At5 g46330; At ARR5, At3 g48100; At ARR2, ARR3 AT1G59940 [Correction added after online publication 29 October 2013 to correct an error in the text. ‘At4g16110’ has been corrected to ‘ARR3 AT1G59940’.]; At ARR6, At5 g62920; Nb Fls2, EF417987.

Results

In silico analysis of HopQ1PtoDC3000

In this study we investigated the effects of the Pto DC3000 effector HopQ1 on PTI. Consistent with previous reports, we found in homology searches with HopQ1 that the C-terminus of HopQ1 has significant homology with a group of NHs called the inosine/uridine-preferring NHs (Giska et al., 2013; Li et al., 2013b). This was most evident when applying the Conserved Domain Architecture Retrieval Tool (CDART) algorithm (http://www.ncbi.nlm.nih.gov/Structure/lexington/lexington.cgi) or the Protein Homology/ analogY Recognition Engine (PHYRE) (www.sbg.bio.ic.ac.uk/~phyre/). The aspartate motif DXDXXXDD, a hallmark of NH activity (Jung et al., 2009), is mostly conserved in the HopQ1 NH domain (Fig. S1a). However, the second aspartate of the DXDXXXDD motif is shifted by two amino acids (aa) (D101XXXDXDD) towards the C-terminus. Interestingly, this change in the putative nucleoside binding site is broadly conserved among HopQ1 homologs from diverse bacterial phytopathogens, such as Xanthomonas and Ralstonia. In addition, the aspartate motif of other characterized members of the NH group is present at their respective N-termini (Glockner et al., 2003; Versees & Steyaert, 2003), whereas it is preceded by an N-terminal domain (amino acids 1–100) in HopQ1 and homologous effectors. The NH domain itself is poorly conserved between HopQ1 and characterized members of the NH family (c. 10% aa identity/16% aa similarity), while this domain is highly conserved among HopQ1 family members, with up to 70% aa identity and 80% conservation (Fig. S1b). Taken together, the bioinformatic analysis suggests that HopQ1 could act as a (nucleobase) hydrolase, but prediction of the specific substrate and its mode of action is not possible.

HopQ1 suppresses flg22-induced PTI responses in Arabidopsis

To test a role for the HopQ1 NH domain in PTI suppression, we produced transgenic Arabidopsis Col-0 lines expressing HopQ1 and lines expressing HopQ1-D101A or HopQ1-D107/108A from the β-estradiol-inducible XVE promoter (Fig. 1a; Zuo et al., 2000). We searched for suppression of flg22-triggered immune responses in these lines. One such response is a rapid burst of ROS shortly after MAMP treatment which is a typical hallmark of PTI. As expected, flg22 treatment of Col-0 wildtype (WT) control plants resulted in the characteristic burst of ROS, both in the presence and in the absence of β-estradiol. Thus β-estradiol does not interfere with ROS production (Fig. 1b). Induction of HopQ1 expression 24 h before MAMP treatment fully suppressed flg22-triggered ROS production. Conversely, treatment with flg22 still induced ROS in plants expressing either HopQ1 aspartate mutant, albeit less so than in Col-0 control tissue. Next, we analyzed the levels of the dually phosphorylated MAPKs, MPK6 and MPK3, in response to flg22, which represent the activated form of MAPKs and are also induced by flg22. As expected, high amounts of dually phosphorylated MPK6 and MPK3 were found in the Col-0 control tissue and MAPK activation was not affected by β-estradiol treatment (Fig. 1c). By contrast, activation of MPK6 and MPK3 was strongly reduced in the HopQ1-expressing leaf tissue. Conversely, lines expressing HopQ1-D101A or HopQ1-D107/108A accumulated the active forms of MPK3/6 strongly. Overall, these results suggest that HopQ1 suppresses flg22-triggered early defense responses, depending on its conserved aspartate motif within the NH domain.

Figure 1.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and pattern-triggered immunity (PTI) marker assays on Arabidopsis leaf tissue (6 wk old) harboring HopQ1, HopQ1-D101A or HopQ1-D107/108A controlled by the XVE promoter (T2 lines) in the presence or absence of induction by β-estradiol for 24 h. (a) qRT-PCR of HopQ1 (#66), HopQ1-D101A or HopQ1-D107/108A expression. In this qRT-PCR, the expression was normalized to the housekeeping gene ubiquitin and relative expression to ubiquitin is presented. Error bars indicate ± standard error of the mean of three biological replicates. (b) Reactive oxygen species (ROS) burst induced by 500 nM flg22 in Col-0 or Col-0 harboring HopQ1, HopQ1-D101A (D101A) or HopQ1-D107/108A (D107/108A). (c) Western blot of a mitogen-activated protein kinase (MAPK) assay where dually phosphorylated and thus activated MPK3 and MPK6 were detected using monoclonal anti-pERK1-2 antibodies (cell signaling). Control-induced tissue is marked by a minus (−) and β-estradiol-induced tissue by a plus (+). Each tissue sample was subsequently treated with either a negative control (‘c’) or 500 nM flg22 (‘F’) for 15 min. Ponceau red staining is included as a loading control. This experiment was repeated twice with similar results.

HopQ1 represses FLS2 accumulation

As both ROS and MAPK activation are very early responses induced within min of flg22 application, we decided to test the integrity of the FLS2 receptor. To do this, we induced HopQ1 expression as well as HopQ1-D101A and HopQ1-D107/108A with β-estradiol for 24 h before analyzing the leaf tissue for FLS2 mRNA via qRT-PCR. Interestingly, the level of FLS2 mRNA was dramatically reduced in the presence of HopQ1 with respect to negative controls, but not in the presence of either aspartate mutant (Fig. 2a). Consistently, the FLS2 glycoprotein was undetectable in the protein extracts of these samples, even when we enriched the sample for glycoslated proteins by Concavalin A (ConA) precipitation (Fig. 2b). We conclude that HopQ1 interferes with FLS2 expression by an as yet unknown mechanism that is dependent on an intact aspartate motif within the putative NH domain.

Figure 2.

Suppression of FLS2 accumulation in Arabidopsis leaf tissue harboring HopQ1 controlled by the XVE promoter (T2 lines) in the presence or absence of induction by β-estradiol for 24 h. (a) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) of FLS2 in Col-0 and HopQ1, HopQ1-D101A or HopQ1-D107/108A expressing transgenic Col-0 lines. In this qRT-PCR the expression was normalized to the housekeeping gene ubiquitin and is expressed as percentage change to the uninduced control sample. The bars show the average of three biological repeats, with error bars indicating ± SEM. Statistics were performed on each experiment using Student's t-test; *, P < 0.05; ***, < 0.01. (b) Western blot developed with anti-FLS2 polyclonal antibodies after ConA precipitation of 250 mg tissue extracts to enrich those in glycosylated proteins, including FLS2 for enhanced FLS2 detection. The samples are indicated as follows: T, total; W, wash; ConA, Concavalin A precipitate.

FLS2 repression by HopQ1 is promoter dependent

We next investigated whether attenuation of FLS2 expression by HopQ1 is mediated by the FLS2 promoter. Arabidopsis plants expressing FLS2 under the control of the 35S promoter were silenced for FLS2 expression (Robatzek et al., 2007). Thus we decided to use the N. benthamiana transient gene expression system to perform this experiment. However, HopQ1 triggers ETI in N. benthamiana via an as yet unidentified resistance protein (Wei et al., 2007). This ETI is associated with a very weak cell death phenotype that is not visible until 72 h post-infiltration with Agrobacterium (Li et al., 2013b).We performed trypan blue cell death staining in tissue transiently expressing an empty vector (EV) control construct, HopQ1 or the aspartate mutants HopQ1-D101A and D107/108A under the control of the 35S promoter, in order to exclude the possibility of microscopic cell death interfering with our analysis. We did not detect any signs of cell death within 2 d of transformation (Fig. S2a), but there was emergence of cell death in leaves transformed with HopQ1 as well as both aspartate mutants at 5 d post-infiltration (Fig. S2b). These results are compatible with the working hypothesis that HopQ1 triggers a slow ETI independently of its putative NH activity. In addition, these results enabled us to test PTI suppression and FLS2 attenuation by HopQ1 in N. benthamiana within the first 2 d of expression. To do this, we coexpressed 35S : HopQ1 or an EV control with Pro AtFLS2 FLS2, Pro 35S : FLS2 or the β-glucuronidase reporter GUS under the control of the FLS2 promoter, Pro AtFLS2 : GUS fusion, for 36 h, and analyzed the tissue by qRT-PCR for changes in FLS2 and GUS expression, respectively (Fig. 3). Consistent with previous results, and as described in Fig. S2(a), no cell death emanating from HopQ1 expression was seen at this time (Li et al., 2013b). The mRNA levels for both AtFLS2 FLS2and Pro AtFLS2 : GUS were strongly reduced in HopQ1-myc-expressing N. benthamiana leaves compared with EV control tissue when expressed from the At FLS2 promoter (Fig. 3a,b). However, FLS2 expression was not affected by HopQ1-myc when expressed from the 35S promoter (Fig. 3b). As expected, comparable effects were observed on the amount of protein, where FLS2 and GUS did not accumulate when expressed from the At FLS2 promoter in the presence of HopQ1 (Fig. 3d,f), whereas accumulation of these proteins was unchanged by HopQ1 when they were expressed from the 35S promoter (Fig. 3e). Thus, HopQ1 seems to directly or indirectly target PRR transcription in a promoter-specific manner.

Figure 3.

HopQ1 attenuates At FLS2 mRNA and protein accumulation in a promoter-dependent manner. Constructs with Pro FLS2 : FLS2, Pro 35S : FLS2 or Pro FLS2 : GUS were coexpressed with HopQ1 or empty vector (EV) for 2 d in mature Nicotiana benthamiana leaves. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed using the housekeeping gene EF1a as a reference (a–c). Values represent averages of three independent biological replicates, each consisting of three dependent biological replicates with ± SEM indicated by the error bars. Statistics were performed on each experiment using ANOVA and significant differences; *, P < 0.01 (a–c). (a) Coexpression of Pro FLS2 : FLS2 depicting FLS2 expression. (b) Coexpression of Pro 35S : AtFLS2 depicting FLS2 expression. (c) Coexpression of Pro FLS2 : GUS depicting GUS expression. (d–f) Western blots of HopQ1 and Pro FLS2 : FLS2 coexpression (d), HopQ1 and Pro 35S : FLS2 coexpression (e) or HopQ1 and Pro FLS2 : GUS coexpression (f) for 2 d in N. benthamiana leaf tissue. HopQ1 was detected using monoclonal anti-myc antibodies and FLS2 was detected using polyclonal anti-FLS2 antibodies (d–f). These experiments were repeated twice with similar results.

HopQ1 exhibits nuclear-cytoplasmic localization

These results prompted us to investigate the subcellular localization of HopQ1. We fused HopQ1 genetically to a C-terminal cyan fluorescent protein (CFP) tag and monitored its accumulation by confocal laser scanning microscopy after transient expression from the 35S promoter. Consistent with previous reports, HopQ1-CFP localized to the cytosol and possibly also the nucleus in N. benthamiana leaves (Fig. S3a; Li et al., 2013a). The HopQ1-CFP fusion protein was functional as a suppressor of MAMP-induced ROS production, and it accumulated as a full-length HopQ1-CFP product detectable on Western blots (Fig. S3b,c). According to these results, we cannot exclude that HopQ1 functions directly as a transcriptional repressor of FLS2 and possibly other genes, but other mechanisms of suppression are also conceivable.

HopQ1 expression in young seedlings leads to developmental defects, especially in the root

We observed that prolonged and early expression of HopQ1 in transgenic Arabidopsis plants resulted in dramatic developmental defects especially on the root system (Figs 4, S4), while induction of HopQ1 expression in leaves had no effect (data not shown). Roots expressing HopQ1 were severely shortened, deficient in lateral root formation and did not produce normal root hairs (Figs 4, middle panel, S4). In some cases, apical dominance was lost and anthocyanins accumulated in the apical meristem and associated leaves and/or cotyledons while the cotyledon was bleached for Chl (Fig. 4). These developmental defects were caused only by HopQ1 (middle panel) and not by the aspartate mutants (upper panel). In addition, the developmental defects were most evident when transgenic plants were treated with β-estradiol at the cotyledon stage (Fig. 4), whereas the developmental changes in 1-wk-old β-estradiol treated seedlings were very subtle and restricted mainly to the roots (Fig. S4a,b, right panel), with reduced root length and largely missing lateral roots. Such phenotypic changes are typical for disturbance of hormone signaling networks (Peng et al., 2009) and we thus hypothesized that HopQ1 expression interferes with one or more hormone signaling pathways, especially those active during early development. To obtain more indications on possible hormonal targets, we searched for known hormone mutants in the Arabidopsis hormone database (http://ahd.cbi.pku.edu.cn/) using the observed phenotypic changes (such as short primary root, no lateral root, purple cotyledon) as search terms. We found that the observed phenotype largely resembles that of the known CK mutant pga22, a gain-of-function mutant of the isopentenyl transferase 8 (IPT8). IPTs catalyze the rate-limiting step in CK biosynthesis, andCK signaling is thus up-regulated in the pga22 mutant. CKs are also attractive HopQ1 targets, because they are modified aminopurines which are produced from aminopurin-monophosphate-riboside precursors.

Figure 4.

Phenotype of Arabidopsis Col-0 wildtype plants and Col-0 plants harboring HopQ1, HopQ1-D101A or HopQ1-D107/108A controlled by the XVE promoter (T2 lines) in the presence or absence of induction by β-estradiol for 10 d in 24 h light chambers. Four-day-old seedlings (cotyledon stage) grown on MS plates in continuous light were transferred to liquid MS medium substituted with an ethanol control or 10 μM β-estradiol and grown for another 10 d in continuous light.

HopQ1 expression leads to changes in endogenous CK concentrations, but does not significantly affect auxin or salicylic acid (SA) concentrations

To test whether CK contents or other phytohormones are changed in HopQ1-expressing tissue, we analyzed the in vivo concentrations of several plant hormones in Col-0 as well as transgenic HopQ1 lines treated with β-estradiol or a negative mock control. Ethylene, a known regulator of FLS2 transcription (Boutrot et al., 2010; Mersmann et al., 2010), could not be detected with this method, but significant effects of HopQ1 expression on steady-state concentrations of ethylene were not observed using gas chromatography (data not shown). By contrast, we observed no significant changes in auxin concentrations (Fig. S5a) or the immunity-associated hormone SA acid contents (Fig. S5b) upon HopQ1 expression as compared with mock-treated control tissue. Interestingly, HopQ1 expression significantly induced the levels of different CK types, namely N6-(2-isopentenyl)adenine (iP), trans-zeatin (tZ), cis-zeatin (cZ) and dihydrozeatin (DHZ), present in Arabidopsis (Figs 5a,left panel, S6a–c). CKs are formed from adenine nucleotides carrying an isoprene-derived side chain at the N6 residue of the purine ring (Sakakibara, 2006). A phosphoribohydrolase activity converting inactive CK nucleotides in a single-step pathway is required to release the CKs’ active free base forms, which then can interact with the CK receptors AHK2, AHK3 and AHK4 (Fig. 5b; Romanov et al., 2006; Stolz et al., 2011; Wulfetange et al., 2011). In Arabidopsis, this phosphoribohydrolase activity is provided by a seven-member family of so-called lonely guy (LOG) enzymes (Kuroha et al., 2009; Tokunaga et al., 2012). The profile of the levels of iP and its derivatives was similarly changed in HopQ1 expressing tissue when compared to the levels reported for overexpression of individual LOGs (Fig. 5), suggesting that HopQ1 might possess similar enzymatic activity (Kuroha et al., 2009). While the hormonally active iP was significantly increased, the LOG substrate and iP precursor, iP-9-riboside-5′-monophosphate (iPRMP), was significantly decreased in the presence of HopQ1. Conversely, as observed for LOG overexpressing lines, the glycosylated versions and thus deactivated forms of iP such as iP-7-glucoside (iP7G) and iP-9-glucoside (iP9G) also accumulated to significantly higher levels upon HopQ1 expression (Fig. 5a), suggesting a tight control of CK signaling. We also tested whether changes in CK concentrations can be observed during natural infections with Pto DC3000 in comparison to the Pto DC3000 ΔHopQ1 mutant. For this, we infiltrated Arabidopsis leaves with the respective strains at 10−3 colony forming units (CFU) ml–1 culture and harvested the tissue 24 h later. Consistent with the results obtained with the transgenic Arabidopsis lines expressing HopQ1 under an inducible promoter, we observed that leaves infected with the Pto DC3000 HopQ1 mutant produced less active CKs than tissue infected with the WT strain (Figs 5c, S6d). Significant differences were not observed for the CK precursor iPRMP and we noticed that the absolute concentrations of the different CKs varied compared with those observed in the transgenic lines. A possible explanation for the observed differences could be changes resulting from the difference in plant age or altered environmental conditions.

Figure 5.

Changes in N6-(2-isopentenyl)adenine (iP) and its derivatives in Arabidopsis Col-0 plants harboring HopQ1 under the β-estradiol-inducible XVE promoter. (a) Leaf tissue of 6-wk-old Col-0 or the transgenic HopQ1 line was cut into strips and treated with either β-estradiol (HopQ1) or an ethanol control (control) for 48 h. The tissue was frozen and lyophilized before extraction and analysis by high-performance liquid chromatography-MS (HPLC-MS). The results are presented in pmol g–1 DW. Abbreviations are as follows (Kamínek et al., 2000): iP, N6-(2-isopentenyl)adenine; iP7G, N6-(2-isopentenyl)adenine-7-glucoside; iP9G, N6-(2-isopentenyl)adenine-9-glucoside; iPRMP, N6-(2-isopentenyl)adenine-9-riboside-5′-monophosphate. The asterisks indicate statistically significant differences (P < 0.05) as determined by Student's t-test. The error bars indicate ± SEM. (b) Simplified scheme to illustrate the role of each of the cytokinin (CK) derivatives in CK signaling. LOG, lonely guy enzymes. (c) Determination of N6-(2-isopentenyl)adenine (iP) and derivatives by high-performance liquid chromatography-MS (HPLC-MS) in Arabidopsis leaf tissue of 8-wk-old plants infected with Pseudomonas syringae pv. tomato DC3000 (Pto) or Pto ΔC3000 ΔHopQ1 at 10−3 CFU ml−1 for 24 h.

HopQ1 expression causes up-regulation of CK-responsive genes

Next, we tested if the changes in CK profiles were reflected in up-regulation of A-type CK response genes, which act as negative regulators of the CK signaling pathway and are typically induced by CKs. Induction of HopQ1 expression resulted in significant up-regulation of ARR5 (Fig. 6a), ARR6 (Fig. 6b) and ARR3 [Correction added after online publication 29 October 2013 to correct an error in the text. ‘ARR2’ has been corrected to ‘ARR3’.] (Fig. 6c), while β-estradiol alone had no effect when applied to Col-0 plants. Likewise, ARR5 was not induced by the expression of the HopQ1 aspartate mutants D101A or D107/108A (Fig. 6a).

Figure 6.

Induction of cytokinin (CK) responsive genes in HopQ1-expressing transgenic Col-0 lines of Arabidopsis. (a) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) of ARR5 in Col-0 and HopQ1, HopQ1-D101A- or HopQ1-D107/108A-expressing transgenic Col-0 lines (10 μM β-estradiol for 24 h). (b) qRT-PCR of ARR6 in Col-0 and HopQ1-expressing transgenic Col-0 lines (10 μM β-estradiol for 24 h). (c) qRT-PCR of ARR3 in Col-0 and HopQ1-expressing transgenic Col-0 lines (10 μM β-estradiol for 24 h). In all qRT-PCRs, the expression was normalized to the housekeeping gene ubiquitin and is expressed as percentage to the uninduced control sample. The bars show the average of three biological repeats with error bars indicating ± SEM. Statistics were performed on each experiment using Student's t-test and significant differences (*, P < 0.05). [Corrections made to figure and legend after online publication 29 October 2013 to correct an error in the text. ‘ARR2’ has been corrected to ‘ARR3’.]

We also tested whether ARR5 induction is specific for HopQ1 or generally occurs in the presence of bacterial type III effectors. Therefore we transiently transformed N. benthamiana leaf tissue with an EV construct, 35S : HopQ1 or 35S : AvrPto, and monitored the expression level of the putative NbARR5 gene. In addition, we included EV leaf tissue treated with 100 nM exogenously applied tZ as a control to show that the putative NbARR5 transcript is indeed induced by CK perception. We found that NbARR5 was induced by exogenously applied tZ as well as HopQ1 expression for 36 h, but not by the EV construct or AvrPto (Fig. S7). Overall, these results are consistent with the hypothesis that HopQ1 may mimic LOG activities to manipulate CK signaling.

HopQ1-containing extracts purified from E. coli lead to CK precursor hydrolysis in vitro

We tested enzymatic activities of HopQ1 as well as the HopQ1-D107/108A mutant in vitro. Therefore we expressed both proteins in E. coli as N-terminal Glutathione-S-Transferase (GST) fusions for simple purification by immunoaffinity (Fig. S8a). Unfortunately, HopQ1 and the aspartate mutant were mostly insoluble and only small amounts of protein could be obtained in the soluble fraction. Furthermore, additional purification steps were not possible because of the limited amount of protein extract available. Nevertheless, we assessed the elution fractions of HopQ1 and the HopQ1-D107/108A mutant for hydrolytic activity towards iPRMP as well as adenosine-monophosphate (AMP) in a colorimetric assay, making use of the difference in optical absorbance at 280 nm between the substrate and the nucleobase product (Parkin et al., 1991; Shi et al., 1999). While AMP was a poor substrate (data not shown), iPRMP was hydrolyzed by HopQ1 and, to a much lesser, extent by the HopQ1-D107/108A mutant (Fig. S8b). The respective Km value was not determined, because this spectrometric assay is not very sensitive and not suitable for robust Km determinations, especially in light of the relatively crude HopQ1 elution fractions. However, we believe that these data provide further evidence that HopQ1 is indeed able to cleave CK precursor molecules both in vivo and in vitro.

CK treatment suppresses FLS2 accumulation

We next asked if manipulation of CK signaling by HopQ1 is responsible for the lack of FLS2 mRNA (Fig. 2). In this case, exogenous application of CKs should also attenuate FLS2 transcription. Thus, we searched publicly available microarray databases (www.geneinvestigator.com) for a connection between CK signaling and PTI components at the level of gene transcription. We found that exogenous application of CKs, such as spraying of 20 μM tZ onto aerial plant tissues, led to modest attenuation of FLS2 transcript accumulation at 1 h after treatment (c. 50% reduction; www.geneinvestigator.com; experiment ID: AT-00110). In fact, we could detect a clear reduction in FLS2 expression by qRT-PCR after 5 h of tZ application in Arabidopsis Col-0 leaf tissue (Fig. 7a), whereas At ARR5 was moderately up-regulated in the same conditions (Fig. 7b). This was also reflected by the amounts of FLS2 protein, which were strongly reduced upon tZ treatment (Fig. 7c). Consistent with a decrease in FLS2, flg22-triggered ROS production and MAPK activation were also reduced when leaf tissue was pretreated with tZ for 5 h (Fig. 7d and e, respectively). Thus, CK signaling seems to impair FLS2 expression.

Figure 7.

Prolonged external application of low trans-zeatin (tZ) concentrations leads to pattern-triggered immunity (PTI) suppression and undermines FLS2 accumulation in Arabidopsis. (a) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) of FLS2 in Col-0 lines treated with 100 nM tZ or a negative control for 5 h. (b) qRT-PCR of ARR5 in Col-0 lines treated with 100 nM tZ or a negative control. In all qRT-PCRs, the expression was normalized to the housekeeping gene ubiquitin and is expressed as percentage change to the uninduced control sample. The bars show the average of three biological repeats, with error bars indicating ± SEM. Statistics were performed on each experiment using Student's t-test, and significant differences (P < 0.05) are marked by an asterisk. (c) Western blot with anti-FLS2 antibodies to illustrate FLS2 disappearance upon treatment with 100 nM tZ for 5 h. (d) Reactive oxygen species (ROS) generation in Arabidopsis leaf disks pretreated with different concentrations of tZ for 4 h followed by induction with 500 nM flg22 and immediate ROS measurements. (e) Western blot showing the amount of phosphorylated MPK6 (6) and MPK3 (3) in Arabidopsis leaf tissue pretreated with 100 nM tZ for 4 h followed by 15 min induction with either a negative control or 500 nM flg22. Equal loading is shown by Ponceau staining (P). All experiments were repeated at least twice.

Low doses of exogenously applied CKs suppress PTI responses and lead to increased susceptibility to the Pto DC3000 hrcC mutant

The failure of FLS2 to accumulate in tissue expressing high levels of CKs implies that antibacterial immunity would be compromised in such tissues. HopQ1 deletions from Pto DC3000 have no virulence defect in Arabidopsis, suggesting that redundancy in the effector repertoire masks the specific contribution of HopQ1 (Wei et al., 2007). This would also be expected from any effector targeting FLS2 for its virulence function, as several effectors of Pto DC3000 were previously shown to target FLS2 (Gohre et al., 2008; Shan et al., 2008; Xiang et al., 2008; Zong et al., 2008). Thus we decided to exploit the N. benthamiana model system to investigate whether increased CK concentrations can promote bacterial growth. First, we confirmed that exogenous application of CKs can dampen flg22-triggered PTI responses, namely ROS and MAPK activation, in this system (Fig. S9a and b, respectively). We found that ROS production by flg22 was decreased in a dose-dependent manner when leaf tissue was pretreated with trans-zeatin-9-riboside (tZR) for 5 h (Fig. S9a). We also performed time course experiments, pretreating the tissue for different amounts of time with tZR before measuring flg22-induced ROS production, and found that ROS was significantly suppressed after 4–16 h of pretreatment (Fig. S9c). Likewise, flg22-mediated MAPK activation was inhibited by pretreatments with tZ. Thus, the N. benthamiana system seems suitable to address the hypothesis that increases in CK signaling can promote bacterial growth. However, previous reports showed that application of high concentrations (1–18 μM) of CKs lead to increased resistance to Pto DC3000 and Pta in Arabidopsis and N. benthamiana as a result of induction of SA signaling or phytoalexin production, respectively (Choi et al., 2011b; Grosskinsky et al., 2011). By contrast, low concentrations of CKs (0.25–1.5 μM) can be beneficial for microbial growth, as shown for fungal (Erysiphe graminis) and oomycete (Hyaloperonospora arabidopsis) pathogens on wheat seedlings and Arabidopsis, respectively (Babosha, 2009; Argueso et al., 2012). Thus, we chose two different concentrations of tZR, namely 10 nM and 10 μM, for this experiment. Interestingly, we found that growth of the type III secretion system-deficient Pto DC3000 hrcC mutant, which is unable to secrete effectors into the host cell, was increased in N. benthamiana leaves when low doses of tZR were co-infiltrated (c. 0.5loge; Fig. 8). By contrast, coinfiltration of high concentrations of tZR (10 μM) led to increased resistance, consistent with previous reports (Grosskinsky et al., 2011). Thus, finely balanced CK concentrations might be important for microbial success and plant defense.

Figure 8.

Low concentrations of exogenously applied cytokinins (CKs) lead to increased susceptibility to the Pseudomonas syringae pv. tomato (Pto) DC3000 hrcC mutant in Nicotiana benthamiana. Bacterial growth is shown for Pto hrcC coinfiltrated with 10 nM trans-zeatin-riboside (tZR), 10 μM tZR or a negative control solution. Each bar represents the average of four biological repeats and the error bars represent ± SEM. Statistics were performed on each experiment using one-way ANOVA, and significant differences (P < 0.01) are marked by respective letters. This experiment was repeated twice with similar results.

Discussion

HopQ1 is a widely conserved bacterial type III effector with a central sequence of c. 275 amino acids that contains an NH domain. This domain contains the characteristic DXDXXXDD motif at the N-terminus, although the second aspartate is shifted by two amino acids. Interestingly, the HopQ1 gene is present in the important phytopathogenic bacterial genera Pseudomonas, Xanthomonas and Ralstonia, despite the fact that these are not closely related. This unusual D101XXXDXDD motif is common to all these HopQ1 homologs, suggesting that a novel function might have evolved in these bacterial strains. This indicates that the HopQ1 effectors have been acquired by various pathogens in order to fulfill the same function in the host plant. Here, we provide evidence that HopQ1 can act as a CK-activating enzyme, and that the conserved aspartate motif is essential for this activity. CKs are active in plants as free nucleobases; the corresponding nucleosides and nucleotides are CK precursors, but they are mostly inactive as hormones (Romanov et al., 2006). A group of genes called LOG (lonely guy) are essential to CK production and encode highly specific hydrolases, converting CK nucleotides in a single-step pathway directly into active CK bases (Kurakawa et al., 2007; Kuroha et al., 2009; Tokunaga et al., 2012). Our results suggest that HopQ1 mimics LOG enzymes, cleaving CK nucleotide precursors, and thus acting as a CK-activating enzyme (Kurakawa et al., 2007; Kuroha et al., 2009; Tokunaga et al., 2012).

Previous data showed that FLS2 was slightly but significantly down-regulated by tZ treatment of aerial Arabidopsis tissues, supporting a model in which moderate up-regulation of the CK concentrations leads to PTI suppression. Consistently, we found that HopQ1 induces the expression of the A-type CK response genes ARR5, ARR6 and ARR3 [Correction added after online publication 29 October 2013 to correct an error in the text. ‘ARR2’ has been corrected to ‘ARR3’.], which is – at least for ARR5 – analogous to the effects of exogenous CKs (Gajdosova et al., 2011), and that CK treatments suppress FLS2 expression. The biological relevance of subtle changes of CK concentrations was supported by our finding that low concentrations of exogenously applied CK increased the susceptibility of N. benthamiana plants to Pto DC3000 hrcC mutant bacteria. Thus, our results indicate that CK signaling is up-regulated by HopQ1 in a finely-tuned way, which can contribute to PTI suppression by reducing expression of FLS2 (Fig. S10). Conversely, when CK signaling is massively induced, SA concentrations are also increased, leading to enhanced resistance (Fig. S10). However, we found no induction of endogenous SA concentrations by HopQ1, and the capacity of HopQ1 to induce CK signaling is limited by the availability of CK precursor molecules. Whether CKs directly target PRR expression, act via known transcriptional regulators induced by CKs, or interact with other hormone signaling networks cannot be answered at present. However, we did not find significant changes in ethylene, auxin or SA concentrations, all of which could possibly interfere with FLS2 transcription (Navarro et al., 2006; Boutrot et al., 2010; Mersmann et al., 2010). Still, promoter analysis of FLS2 will be important to find the transcriptional repressor involved.

A role for the HopQ1 N-terminus in bacterial virulence was demonstrated recently (Giska et al., 2013; Li et al., 2013b). The N-terminal region of HopQ1 interacts with 14-3-3 proteins in a phosphorylation-dependent manner. Mutation of the phosphorylated serine decreased bacterial growth of Pto DC3000 on resistant tomato cv 76R, suggesting that interaction with 14-3-3 proteins plays a role in weak suppression of ETI, rather than PTI (Li et al., 2013b). However, in competitive index assays in bean leaves, Pto DC3000 carrying the mutated HopQ1 allele had a disadvantage compared with the WT strain (Giska et al., 2013). Thus, interaction with 14-3-3 proteins might also affect PTI signaling in an as yet unknown manner.

Many plant pathogens produce CKs or induce CK biosynthesis. Among these are bacterial pathogens such as A. tumefaciens, Pseudomonas savastoni and Pseudomonas solanacearum (Akiyoshi et al., 1987). Although P. syringae has not been reported to induce or produce CKs, CK-induced genes such as ARRs are up-regulated by Pto DC3000 in Ler (c. 4.5-fold for ARR5, e.g. www.genevestigator.com) and Col-0 (c. twofold for ARR5; www.genevestigator.com) backgrounds. These pathogens can benefit from CK elevation by means other than PTI suppression. For example, CKs counteract mesophyll cell senescence, and pathogenic fungi, as well as some bacteria, are thought to secrete CKs to increase the sink strength of the colonized host tissue (‘green island effect’) (Walters & McRoberts, 2006; Walters et al., 2008). To our knowledge, Pto DC3000 does not produce CKs itself. Thus, CKs produced by HopQ1 might have roles in addition to diminishing FLS2, for example, by increasing nutrient availability, or by otherwise manipulating the host tissue in favor of the pathogen, and the observed enhanced growth of the bacteria in the presence of low doses of CKs may also be related to such effects. Likewise, HopQ1 can promote growth of Pto DC3000 when inducibly expressed in Arabidopsis (Li et al., 2013a). In this case, HopQ1 was expressed as a C-terminally tagged fusion protein, and although bacterial growth was promoted in its presence, the authors could not observe suppression of flg22-triggered ROS or MAPK activation. We deliberately avoided fusion tags, as previous experiments performed with N. benthamiana indicated that C-terminal fusions may interfere with at least some of the functions of HopQ1, specifically ETI in N. benthamiana (data not shown). Alternatively, the timing in HopQ1 accumulation may not have been sufficient to allow manipulation of FLS2 expression via CK signaling.

FLS2 is a major virulence target for several effectors, including the Pto DC3000 effectors AvrPto and AvrPtoB, underlying its importance for innate immunity in plants (Gohre et al., 2008; Shan et al., 2008; Xiang et al., 2008; Zong et al., 2008; Ntoukakis et al., 2009). AvrPto and AvrPtoB target FLS2 directly and at the protein level, thereby shutting off immunity immediately. How long suppression of immunity by AvrPto and AvrPtoB is maintained is not clear, but there is a possibility, at least in the case of AvrPtoB, that the effector is deactivated by the plant (Ntoukakis et al., 2009). The functional redundancy in effector repertoires is evident, but often functional redundancy overlaps only partly. Possibly, this could be the case for Pto DC3000 effectors targeting FLS2. While AvrPto and AvrPtoB appear to be responsible for a fast and efficient turning off of FLS2 signaling, HopQ1 might promote long-term dampening of FLS2 signaling.

To conclude, CKs appear to play a major role in plant–microbe interactions. Microbes interfering with plant CK signaling are widespread and include organisms of diverse genera. Some of these microbes produce CKs themselves; others produce IPTs to drastically increase the production of CKs in planta (generally gall-forming pathogens) (Choi et al., 2011a). We now provide evidence that HopQ1 may subtly increase CK concentrations, possibly by mimicking LOG function. The elevation of CK concentrations in turn leads to reduced FLS2 accumulation and thus dampened PTI responses. Our current data suggest that dampening of FLS2 through HopQ1 may not be immediate (observed only 5 h after CK treatment). Therefore, we propose that HopQ1 might be dispensable for early disease establishment but is rather involved in a long-lasting control of PRR expression during the course of infection.

Acknowledgements

We thank L. Steinle and S. Gimenez-Ibanez for technical support, N.H. Chua for the pMDC7 vector, A. Collmer for the Pto DC3000ΔHopQ1 strain, and B. Müller for valuable advice on CK signaling. This work was supported by grants from the Swiss Systems-X initiative and the Swiss National Research Foundation to T.B. Part of this research was supported by the Czech Science Foundation (grant 13-26798S).

Ancillary