The Unfolded Protein Response Regulates Pathogenic Development of Ustilago maydis by Rok1-Dependent Inhibition of Mating-Type Signaling

The unfolded protein response (UPR) is crucial for endoplasmic reticulum (ER) homeostasis and disease development in fungal pathogens. In the plant-pathogenic fungus Ustilago maydis, the UPR supports fungal proliferation in planta and effector secretion for plant defense suppression. In this study, we uncovered that UPR activity, which is normally restricted to the biotrophic stage in planta, inhibits mating and the formation of infectious filaments by Rok1-dependent dephosphorylation of the pheromone responsive mitogen-activated protein kinase (MAPK) Kpp2. This observation is relevant for understanding how the fungal virulence program is regulated by cellular physiology. UPR-mediated control of mating-type signaling pathways predicts that effector gene expression and the virulence potential are controlled by ER stress levels.


RESULTS
UPR activity inhibits the b mating-type-dependent regulatory network and upstream signaling pathways. To study the effects of a constitutive active UPR on pathogenic development, the intronless cib1 s mRNA under the control of its endogenous promoter was integrated into the ip locus of the solopathogenic strain SG200 in single (cib1 s ) or multiple (cib1 s(x) ) copies (18). In SG200, expression of the compatible bE1 and bW2 genes supersedes the need of a mating partner to trigger the pathogenic program and fulfill its life cycle (27). When SG200 is spotted on charcoal-containing solid medium, formation of b-dependent filaments is induced as visualized by fuzzy colonies (10). As reported previously (18), we confirmed that filament formation of SG200 on charcoal-containing solid medium is suppressed by an active UPR ( Fig. 2A), which was connected to a dose-dependent reduction of bE, bW, and rbf1 transcript levels (18). Since the b regulatory network is composed of hierarchically arranged transcription factors, we examined if the complete b regulatory network is affected by the UPR. To this end, we determined expression levels of the key transcriptional regulators downstream of Rbf1, namely, Biz1, Hdp1, and Hdp2, by quantitative reverse transcriptase PCR (qRT-PCR) analysis (10,28,29). Consistent with the results obtained for the b genes and rbf1, expression levels of all three transcription factors were significantly and dose-dependently reduced by cib1 s (Fig. 2B). This suggests that UPR activity leads to global suppression of the b regulatory network.
To address at which stage the UPR interferes with the mating-type signaling pathway, we investigated the effects of UPR activity on the pheromone response pathway acting upstream of bE/bW. As a read-out, we determined expression levels of the a mating-type locus genes mfa1 and pra1, encoding the pheromone precursor and the pheromone receptor, respectively, and of prf1 in SG200 (wild type [WT]) and the cib1 s -expressing derivative. In comparison to the WT, expression levels of all three genes were significantly reduced in cib1 s -expressing strains (Fig. 2C). Moreover, expression levels of two transcriptional regulators of Prf1, Hap2 and Rop1 (Fig. 2B), and of the Prf1-induced target genes UMAG_00306, UMAG_05348, and UMAG_06190 (8) were FIG 1 UPR activity is required for fungal proliferation in planta but interferes with b-dependent filament formation on the plant surface. Binding of the pheromone (Mfa2) to the compatible pheromone receptor (Pra1) induces mating of compatible haploid sporidia. The pheromone signal is transmitted by a PKA and a MAPK cascade, inducing G 2 cell cycle arrest and increasing expression of the a and b mating-type genes (mfa/pra and bE/bW). Cell-cell fusion results in formation of the infectious dikaryotic filament. The bE-bW heterodimer controls maintenance of the G 2 cell cycle arrest, elongation of the filament, and appressoria formation to penetrate the plant surface. After plant penetration, the UPR is activated, supporting the release of the cell cycle block by complex formation with Clp1. By contrast, premature UPR activity interferes with b-dependent filament formation and inhibits expression of the b mating-type genes. Since this effect is independent of the Clp1 interaction, an active UPR potentially also interacts with mating-type pathways upstream of bE and bW.
suppressed by the UPR as well (Fig. 2D). Interestingly, expression of the Prf1-repressed target gene UMAG_10838 (8) was dose-dependently induced by the UPR (Fig. 2D). This suggests that the UPR interferes with mating-type signaling upstream of the b heterodimer by negatively affecting Prf1-dependent gene regulation.
UPR activity impedes the morphological and transcriptional responses to pheromone. In the solopathogenic strain SG200, expression of a compatible pheromonepheromone receptor pair and compatible b genes leads to constitutive activation of the a and b mating-type pathways. Both pathways are connected by various feedback loops (7,8,12), complicating their functional analysis. We thus used the haploid FB1 and FB2 WT strains to specifically address the impact of the UPR on the pheromone response (controlled by the a mating-type locus). Similar to the results obtained in the SG200 background, mating assays between compatible FB1 and FB2 strains revealed reduced filament formation in crossings between cib1 s -expressing derivatives on charcoal-containing solid medium (see Fig. S1 in the supplemental material). When liquid cultures of FB1 (WT) and the cib1 s -expressing derivative were treated with synthetic a2 pheromone, conjugation tube formation was inhibited in cib1 s -expressing strains compared to that in the WT (Fig. 3A). Conjugation tube formation was significantly reduced from 32.9% (WT) to 17.3% in FB1 cib1 s and to 4.5% in FB1 cib1 s(x) strains (Fig. 3B). This demonstrates that the UPR inhibits the morphological response to pheromone in a dose-dependent manner. dependent filament formation. SG200 and derivatives constitutively expressing cib1 s or cib1 s(x) were spotted on potato dextrose solid medium supplemented with 1% charcoal. Photos were taken after 24 h at 28°C. White fuzzy colonies indicate the formation of filaments. (B to D) qRT-PCR analysis of gene expression in SG200 and derivatives expressing one (cib1 s ) or multiple (cib1 s(x) ) copies of cib1 s . eIF2b was used for normalization. Expression values represent the means from three biological replicates with two technical duplicates each. Error bars represent the standard errors of the means (SEMs). Statistical significance was calculated by one-way analysis of variance (ANOVA), followed by Bonferroni's multiplecomparison test. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001. Only comparisons that revealed significant differences are shown. (B) Expression of rbf1-regulated transcription factors biz1, hdp1, and hdp2 and regulators of prf1 expression, rop1 and hap2. (C) Expression of the pheromone genes mfa1, pra1, and the central transcription factor prf1. (D) Expression values of genes UMAG_00306, UMAG_05348, UMAG_06190, and UMAG_10838, which are differentially expressed upon MAPK-dependent phosphorylation of Prf1.
We next tested if suppression of the pheromone-induced morphological response correlates with an altered transcriptional response. qRT-PCR analysis revealed that pheromone-induced expression levels of mfa1, pra1, and prf1 were significantly lower in the cib1 s -expressing strain than in the WT (Fig. 3C). This suggests that the UPR inhibits both the morphological and the transcriptional responses to pheromone.
Since signal transduction within the pheromone response is mediated by PKA and MAPK pathways, we tested whether UPR activation affects the activity of the cAMPdependent PKA pathway. To specifically activate the PKA pathway, we treated FB1 (WT) and FB1 cib1 s cells with cAMP (6 mM) and monitored expression of mfa1 and pra1 genes. This revealed that cAMP-induced expression of mfa1 and pra1 was slightly lower in cib1 s -expressing strains than in FB1 (Fig. 3D). However, under these conditions, the inhibitory effect of an active UPR on mfa1 and pra1 expression was modest and less pronounced compared to that in pheromone-treated cells. Therefore, the strong impact of an active UPR on the pheromone response pathway is most likely not achieved via the cAMP-dependent PKA signaling pathway.
Genetic activation of the pheromone responsive MAPK pathway reveals differential inhibition of morphological and transcriptional responses by the UPR. To address a potential interaction between the UPR and the MAPK pathway, we geneti- Depicted values represent means from three independent experiments. Error bars represent the standard deviations (SDs). n, number of total cells counted. Statistical significance was calculated by one-way analysis of variance (ANOVA), followed by Bonferroni's multiple-comparison test. **, P Ͻ 0.0; ***, P Ͻ 0.001. (C) qRT-PCR analysis of mfa1, pra1, and prf1 gene expression in FB1 (WT) and derivatives expressing cib1 s after treatment with synthetic a2 pheromone. Cells were treated with synthetic a2 pheromone (2.5 g/ml) for 6 h at 28°C. eIF2b was used for normalization. Expression values represent the means from three biological replicates with two technical duplicates each. Error bars represent the SEMs. Statistical significance was calculated by one-way analysis of variance (ANOVA), followed by Bonferroni's multiplecomparison test. ***, P Ͻ 0.001. Comparisons between strains incubated under identical conditions with significant differences are shown. (D) qRT-PCR analysis of genes mfa1 and pra1 in FB1 (WT) and derivatives expressing cib1 s after treatment with cAMP. Cells were treated with 6 mM cAMP for 12 h at 28°C. eIF2b was used for normalization. Expression values represent the means from two biological replicates with two technical duplicates each. Error bars represent the SEMs.
UPR Inhibits Mating-Type Signaling in Ustilago maydis ® cally activated the pathway by expression of the active MAPK kinase (MAPKK) fuz7 DD in FB1 (WT) and FB1 cib1 s strains (6,30). Expression of Fuz7 DD by the arabinose-inducible crg1 promoter results in formation of filamentous cells, resembling conjugation tubes (6). Under these conditions, 86.8% of the WT cells and 95.9% of cib1 s -expressing cells reacted with the formation of conjugation tube-like structures ( Fig. 4A and B). Conversely, transcript levels of the pheromone genes mfa1 and pra1 and of prf1 were significantly less induced by Fuz7 DD in cib1 s -expressing cells than in the WT (Fig. 4C). Hence, the UPR inhibits the Fuz7 DD -induced gene regulation but not the morphological response. Importantly, transcriptional regulation of the mfa1 and pra1 genes is Prf1 dependent, whereas the morphological response is not (6,31). This indicates that the UPR negatively affects Prf1 activity, which is regulated by phosphorylation of the MAPK Kpp2.
Phosphorylation of the MAPK Kpp2 but not the interaction with the MAPKK Fuz7 is suppressed by UPR activity. Since the MAPK Kpp2 controls the morphological response to pheromone and Prf1 activity (6,32,33), we reasoned that the UPR might affect the MAPK cascade at the level of Kpp2. To test our hypothesis, we expressed a functional Kpp2-green fluorescent protein (GFP) fusion protein under the control of its , and a derivative expressing cib1 s . Cells were grown for 6 h at 28°C in CM-glucose (noninducing) or CM-arabinose (inducing) medium. eIF2b was used for normalization. Expression values represent the means from three biological replicates with two technical duplicates each. Error bars represent the SEMs. Statistical significance was calculated by one-way analysis of variance (ANOVA), followed by Bonferroni's multiple-comparison test. ***, P Ͻ 0.001. Comparisons between fuz7 DD and fuz7 DD cib s strains incubated under identical conditions with significant differences are shown. endogenous promoter from the endogenous genomic locus in WT (FB1 fuz7 DD ) and cib1 s -expressing (FB1 fuz7 DD cib1 s ) strains. Functionality of Kpp2-GFP was demonstrated previously (6). Since Kpp2 is essential for conjugation tube formation (6), we confirmed functionality of Kpp2-GFP by pheromone treatment (see Fig. S2). Western hybridization revealed that abundance of Kpp2-GFP was not altered by induced expression of fuz7 DD or cib1 s -mediated UPR activity (Fig. 5A). By contrast, phosphorylation of the Kpp2-GFP fusion protein was strongly increased upon induced expression of fuz7 DD in the WT but remained almost undetectable in cib1 s -expressing strains under identical conditions (Fig. 5A). This suggests that an active UPR negatively affects phosphorylation of Kpp2.
Fuz7 DD -mediated genetic activation of the MAPK pathway relies on an artificial expression system that bypasses feedback regulation within the MAPK pathway. To address UPR-dependent effects in a more "natural" setting, we monitored Kpp2 phosphorylation in response to pheromone treatment in FB1 (WT) and cib1 s -expressing strains. Consistently, expression of cib1 s did not affect the abundance but specifically reduced phosphorylation levels of Kpp2-GFP in response to pheromone (Fig. 5B), confirming our previous results obtained upon genetic activation of the MAPK pathway (Fig. 5A).
Signal transduction within MAPK cascades requires the physical interaction between pathway components that is supported by accessory or scaffold proteins to establish signaling modules and foster phosphotransfer between MAPK pathway components (34,35). To address if the UPR affects the interaction between Fuz7 and Kpp2, we ) and FB1 fuz7 DD cib1 s (cib1 s ) expressing a Kpp2-GFP fusion protein were grown in CM-glucose (noninducing) or CM-arabinose (inducing) liquid medium for 6 h at 28°C. Three independent transformants were analyzed for each strain. Kpp2-GFP was visualized using a GFP-specific antibody. Phosphorylation of Kpp2-GFP was visualized using the ␣-phospho44/42 antibody that specifically recognizes the phosphorylated TEY motif of MAPKs. The Ponceau S-stained membrane served as loading control. (B) Immunoblot analysis with the same strains as in panel A treated with synthetic a2 pheromone for 6 h at 28°C. Three independent transformants were analyzed for each strain. The same antibodies as in panel A were used to visualize Kpp2-GFP and MAPK phosphorylation. The Ponceau S-stained membrane served as loading control. (C) Coimmunoprecipitation analysis of Kpp2-GFP and Fuz7 DD -HA. Derivatives of strains FB1 fuz7 DD kpp2-GFP (WT) and FB1 fuz7 DD kpp2-GFP cib1 s (cib1 s ) containing a fuz7 DD -HA fusion under the control of the arabinoseinducible crg1 were grown in CM-arabinose (inducing) liquid medium for 6 h at 28°C. A strain expressing untagged Kpp2 served as negative control. Total protein extract was subjected to coimmunoprecipitation using GFP-specific beads. Fuz7 DD -HA was visualized using an HA-specific antibody. Kpp2-GFP was visualized using a GFP-specific antibody. The Ponceau S-stained membrane served as loading control.
UPR Inhibits Mating-Type Signaling in Ustilago maydis ® expressed functional Kpp2-GFP and Fuz7 DD -hemagglutinin (HA) fusion proteins in WT (FB1) and cib1 s -expressing strains and immunoprecipitated Kpp2-GFP by GFP-Trap. This revealed that Fuz7 DD -HA was coimmunoprecipitated to similar amounts in the WT and cib1 s -expressing strains. By contrast, Fuz7 DD -HA was not coimmunoprecipitated in the control strain expressing untagged Kpp2 (Fig. 5C). This indicates that the Kpp2-Fuz7 interaction is specific and not affected by an active UPR.
Deletion of rok1 restores Kpp2 phosphorylation, filament formation, and virulence of cib1 s -expressing strains. Phosphorylation levels of MAPKs are determined by relative activities of phosphorylating and dephosphorylating proteins (36,37). Previously, it was shown that the dual specificity phosphatase (DSP) Rok1 mediates dephosphorylation of Kpp2 and thereby counteracts activity of the pheromone response pathway (38). To test if the inhibitory effect of an active UPR is mediated by Rok1, we deleted rok1 in the WT (FB1 P crg :fuz7 DD ) and the cib1 s -expressing derivative (FB1 P crg :fuz7 DD P cib1 :cib1 s ). Consistent with our previous results, Kpp2-GFP abundance was not altered under any of the tested conditions. Strikingly, induced expression of fuz7 DD resulted in Kpp2-GFP phosphorylation in WT and cib1 s -expressing Δrok1 cells to similar levels (Fig. 6A). Hence, deletion of rok1 fully repressed the negative effect of an active UPR on Kpp2 phosphorylation. were grown in CM-glucose (noninducing) or CM-arabinose (inducing) liquid medium for 6 h at 28°C. Three independent transformants were analyzed for each strain. Protein extracts were subjected to immunoblot analysis, and Kpp2-GFP levels were visualized using a GFP-specific antibody. Phosphorylation of Kpp2-GFP was visualized with an ␣-phospho44/42 antibody specifically recognizing the phosphorylated TEY motif of MAPKs. The Ponceau S-stained membrane served as loading control. (B) Analysis of b-dependent filament formation and virulence of rok1 deletion strains. SG200, SG200cib1 s , and a derivative deleted for rok1 (cib1 s Δrok1) were spotted on potato dextrose solid medium supplemented with 1% charcoal. Photos were taken after 24 h at 28°C. White fuzzy colonies indicate the formation of filaments. For plant infection assay, strains were inoculated into 8-day-old maize seedlings. Disease symptoms were rated 8 days after inoculation (dpi) and grouped into categories as shown in the legend. n, number of inoculated plants. Photos were taken at 12 dpi and represent the most common infection symptom. Statistical significance was calculated using the Mann-Whitney test. ***, P Ͻ 0.001.
We previously demonstrated that constitutive UPR activity inhibits filament formation and virulence of the solopathogenic strain SG200 (18). These effects are reminiscent of the phenotype observed upon constitutive overexpression of rok1 (38). By contrast, rok1 deletion mutants show increased filament formation and are hypervirulent in plant infection assays (38). To test if deletion of rok1 restores not only phosphorylation of Kpp2 in cib1 s -expressing strains but also filament formation and virulence, we deleted rok1 in the SG200cib1 s strain. Strikingly, both reduced filament formation and virulence of strain SG200cib1 s were fully restored in the Δrok1 derivative (Fig. 6B). Interestingly, virulence of SG200cib1 s Δrok1 was not only restored to WT levels but was further increased in comparison to that in the WT control, resembling the hypervirulent phenotype of SG200 Δrok1 (38). Therefore, our data show that UPRdependent inhibition of mating-type signaling is mediated via the DSP Rok1.
UPR activity does not affect Rok1 transcript or protein levels. To investigate how the UPR regulates Rok1, we determined Rok1 transcript and protein levels in the WT (FB1 P crg :fuz7 DD ) and the cib1 s -expressing derivative. This revealed that rok1 transcript levels were only slightly reduced in cib1 s -expressing strains compared to that in the WT control under both fuz7 DD -repressing (glucose) and -inducing (arabinose) conditions (see Fig. S3A). To test for potential UPR-induced alterations of Rok1 protein levels, we expressed a Rok1-mCherry fusion protein under the control of the endogenous promoter from the endogenous genomic locus in WT and cib1 s -expressing strains. Since the fusion protein was not detectable by Western hybridization, Rok1-mCherry was enriched by immunoprecipitation. Total protein amounts of U. maydis protein extracts were adjusted and immunoprecipitation was performed with an increased amount of mCherry-Trap beads (Chromotek) to prevent saturation of the beads and enable a quantitative comparison between both strains. While Rok1-mCherry was not detectable under noninduced conditions, protein levels were similar in both strains in response to fuz7 DD -mediated genetic activation of the MAPK pathway (Fig. S3B). Taken together, these results suggest that the UPR regulates Rok1 on a posttranslational level, potentially by influencing Rok1 activity.
The MAPK Kpp2 is dispensable for biotrophic development in planta. During the life cycle of U. maydis, the UPR is specifically activated after plant penetration (18). Hence, the cross talk between the UPR and the MAPK pathway is supposedly involved in controlling biotrophic growth in planta. Kpp2 is important for the mating process and appressoria formation (32), whereas the partially redundant MAPK Kpp6 is expressed in a bE/bW-dependent manner, and its phosphorylation is required for penetration of the plant surface (39).
To address the potential function of Kpp2 after plant penetration, we used a conditional gene expression approach, allowing the stage-specific depletion of gene expression during development in planta. FB1 Δkpp2 and FB2 Δkpp2 strains were transformed with constructs expressing kpp2 under the control of the conditional promoters P UMAG_12184 and P UMAG_03597 . These promoters are active during the first 2 (P UMAG_12184 ) or 4 (P UMAG_03597 ) days after inoculation (the detailed description of our conditional gene expression approach will be published elsewhere) but show strongly reduced activity at subsequent stages of pathogenic growth in planta (40). We used compatible mixtures of FB1 and FB2 (WT) and derivatives thereof for plant infection experiments. Consistent with previous studies, virulence of compatible combinations of Δkpp2 strains was significantly impaired compared to that of the WT (Fig. 7) (32). Importantly, when kpp2 was expressed under the control of P UMAG_12184 or P UMAG_03597 promoters in the Δkpp2 background (FB1 Δkpp2 P UMAG_12184 :kpp2 ϫ FB2 Δkpp2 P UMAG_12184 :kpp2; FB1 Δkpp2 P UMAG_03597 :kpp2 ϫ FB2 Δkpp2 P UMAG_03597 :kpp2), virulence was fully restored (Fig. 7). Hence, kpp2 is required for the mating process and important for appressoria formation but dispensable for pathogenic development in planta. As we observed that Rok1 interacts with Kpp2 and Kpp6 in the yeast two-hybrid system (see Fig. S4) and that both MAP kinases are dephosphorylated by Rok1 (38), it is conceivable that the cross talk between UPR and the mating-type pathways might as well affect the function of both proteins (39).

DISCUSSION
In U. maydis, the UPR is connected to the control of biotrophic growth via the interaction between the Hac1-like UPR regulator Cib1 and Clp1. In this study, we show that the UPR affects pathogenic development of U. maydis on additional levels. While the plant-specific activation of the UPR fosters pathogenic growth in planta, premature UPR activation counteracts pathogenic development prior host plant penetration. Hence, unfit or stressed cells are less likely to undergo mating than nonstressed cells. We observed that UPR activity counteracts pheromone-dependent signaling by suppression of Kpp2 phosphorylation in a rok1-dependent manner, identifying Rok1 as a novel regulatory target of the UPR.
Gene expression analysis confirmed the global inhibitory effect of the UPR on the b-dependent transcription factor network (18). Moreover, constitutive UPR signaling suppressed the morphological and transcriptional responses to pheromone. In the human-pathogenic fungus Cryptococcus neoformans, deletion of IRE1, encoding the ER stress sensor kinase upstream of the Hac1-like transcription factor Hxl1, resulted in increased pheromone gene expression but defects in sexual mating (41). By contrast, in U. maydis, cib1 s -mediated inhibition of filamentous growth is not dependent on Ire1 (18). Sensing of compatible pheromone by haploid U. maydis cells results in activation of the cAMP-dependent PKA Adr1 and MAPK-mediated signaling pathway, finally leading to differential phosphorylation and activation of Prf1 by Adr1 and Kpp2. Kpp2 itself is activated by phosphorylation and regulates the transcriptional pheromone response via Prf1 and the morphological response in a Prf1-independent manner (6). After pheromone treatment, both responses are inhibited by an active UPR (Fig. 3), whereas only pheromone gene expression but not the formation of conjugation tubes was suppressed after genetic activation of the pheromone pathway. These findings are in line with previous results, illustrating that the morphological and the transcriptional response to pheromone bifurcate downstream of Kpp2 (6). Importantly, phosphoryla- , and derivatives of ⌬kpp2 strains expressing P UMAG_12184 :kpp2 or P UMAG_03597 :kpp2 were inoculated into 8-day-old maize seedlings. Disease symptoms were rated 8 days after inoculation and grouped into categories as shown in the legend. n, number of inoculated plants. Two independent clones were investigated for conditional kpp2 expression. Photos were taken at 8 dpi and represent the most common infection symptom. Statistical significance was calculated using the Mann-Whitney test. ***, P Ͻ 0.001. tion of Kpp2 was strongly reduced by the UPR under these conditions. Since strains expressing nonphosphorylatable Kpp2 AEF fail to form conjugation tubes upon pheromone treatment (6), our results imply that low levels of Kpp2 phosphorylation might be sufficient for conjugation tube formation but not for Kpp2-dependent Prf1 activation.
The interaction between the UPR and MAPK pathways is not limited to U. maydis. In the budding yeast Saccharomyces cerevisiae, nitrogen-induced pseudohyphal growth, which depends on MAPK and PKA pathways, is inhibited by an active UPR (42). However, a direct connection between the UPR and phosphorylation of a pheromonedependent MAPK has not been reported. In the human pathogens Aspergillus fumigatus and C. neoformans, a functional UPR is required for cell wall stress resistance, suggesting connections to the cell wall integrity (CWI) pathway (43)(44)(45). In budding yeast, activation of the CWI pathway triggers the UPR (46) and the UPR is augmented by delayed PKA signaling, leading to an extended secondary response to ER stress (47). These connections do not appear to be conserved in U. maydis: deletion of cib1 did not affect resistance to cell wall perturbing agents in U. maydis and UPR activity did not substantially affect expression of PKA subunits, the PKA-repressed regulator Hgl1 (48), or the PKA-dependent induction of mfa1 and pra1 genes.
To allow tight control of signaling, MAP kinase phosphatases (MKP) are regulated on multiple levels, including transcriptional and posttranslational control (49). Transcription of MSG5, the rok1 ortholog in S. cerevisiae, is induced by pheromone stimulation, placing Msg5p in a negative-feedback loop with the pheromone signaling cascade (50). Similarly, rok1 transcript levels in U. maydis were elevated upon activation of the MAPK module (38). However, since neither transcript nor Rok1 protein levels were altered by an active UPR, additional posttranslational mechanisms must exist regulating Rok1 activity under these conditions. Posttranslational modifications of MKPs such as phosphorylation and ubiquitination can alter subcellular localization, stability, and specificity of phosphatases. S. cerevisiae Msg5p is phosphorylated by its MAPK targets Fus3 and Slt2 as part of a regulatory feedback mechanism (50,51). In the hemibiotrophic blast fungus Magnaporthe oryzae (Pyricularia oryzae), the MKP Pmp1 dephosphorylates the MAPKs Pmk1 (Fus3/Kpp2 ortholog) and Mps1 upon phosphorylation at a conserved serine residue (52). Accordingly, two potential phosphorylation sites have been predicted for Rok1 (53), but their functionality has not been addressed, yet.
Deletion of rok1 results in hypervirulence of U. maydis and restored phosphorylation of Kpp2, filament formation, and virulence in strains with active UPR, indicating a UPR-dependent regulation of Rok1. Previous studies suggested that hypervirulence of rok1 deletion mutants is connected to increased activity of the pheromone response pathway and thus elevated expression of bE and bW genes (38). Consistently, increased virulence was also observed in strains overexpressing the pheromone receptor Pra2 (FB1 P otef :pra2) (54), even allowing to partially bypass the loss of the MAPK Crk1 that is crucial for full activity of the signaling pathway (55). This suggests that activity of the mating-type pathways correlates with the aggressiveness of U. maydis. As a biotrophic pathogen, successful completion of its life cycle is dependent on the living host plant. In this way, the UPR-dependent increased activity of Rok1 might support the coordinated and well-balanced growth of U. maydis inside the plant, ensuring host survival during fungal colonization.
Another important role of this regulatory connection for biotrophic growth in planta is predicted based on the influence on the b regulatory network. Expression of effector-encoding genes during pathogenic development is mainly controlled by the b heterodimer or b-regulated transcription factors, including Rbf1, Hdp2, and, potentially, Biz1 (10,28,29,40,56). Their proper folding, processing, and secretion, however, is dependent on a functional UPR (48,57,58). Deleterious hyperactivation of the UPR is suppressed by Clp1-dependent modulation of UPR gene expression (18,48). In parallel, our data predict that increased ER stress levels initiate Rok1-mediated negative feedback on these regulators. Consequently, effector gene expression is temporarily reduced to support recovery from excessive ER stress and achieve ER homeostasis. Importantly, transcription of the Pit2 effector, which mediates suppression of host cysteine proteases (59,60), is directly regulated by Cib1 and, in addition, by Hdp2 (29,58). As a consequence, pit2 expression is likely not to be affected by UPR-dependent inhibition of the mating-type regulatory network, which is in line with the pivotal function of Pit2 in U. maydis virulence (61).
Kpp2 is highly similar to the unconventional MAPK Kpp6 (39), and both MAPKs are targeted for dephosphorylation by Rok1 (38). While Kpp2 is important for mating, Kpp6 is not expressed or required at this stage. However, both MAPKs are partially redundant during appressoria formation (39). While the single deletion of either MAPK resulted in reduced pathogenicity, only the double deletion completely abolished virulence of U. maydis (38,39). Conditional gene expression revealed that Kpp2 is not required for later stages of biotrophic development in planta, suggesting that Kpp6 might be redundant to Kpp2 also at this stage. The physical interaction between Rok1 and both MAP kinases implies a similar and direct mode of action and suggests that the UPR affects phosphorylation levels of Kpp6 as well. Hence, we cannot exclude that Kpp6 is the functionally important Rok1 target with respect to biotrophic growth in planta. This assumption is supported by regulated kpp6 expression that is induced by the b heterodimer and Rbf1 (10,30). Expression levels peak after plant penetration (2 day postinoculation [dpi]) and remain high during subsequent stages in the plant. In comparison, expression of kpp2 is even slightly reduced after plant penetration (40).
In essence, the negative feedback of the UPR on the pheromone response pathway negatively regulates mating and pathogenic development prior to plant infection but likely supports biotrophic growth in planta (Fig. 8). We identified a UPR-regulated automatic gain control that is predicted to support efficient plant colonization, fungal biotrophy, and ER homeostasis during the interaction with the host plant.

MATERIALS AND METHODS
Strains and growth conditions. Escherichia coli TOP10 strain was used for cloning purposes and amplification of plasmid DNA. Ustilago maydis cells were grown at 28°C in yeast extract-peptone-sucrose (YEPS) light medium (62), complete medium (CM) (63), or yeast nitrogen base (YNB) medium (64,65). crg1-driven gene expression was induced as described previously (30). Yeast-two-hybrid (Y2H) analysis was performed according to the Matchmaker 3 manual (Clontech). S. cerevisiae cells (AH109) containing Y2H plasmids were grown in synthetic defined (SD) medium without Leu/Trp and spotted on SD solid medium without Leu/Trp or without Ade/His/Leu/Trp. U. maydis strains used in this study are listed in Table S1 in the supplemental material.
DNA and RNA procedures. Molecular methods were performed according to described protocols (66). DNA isolation from U. maydis and transformation procedures were performed according to the methods described in reference 67. For gene deletions, a PCR-based approach and the SfiI insertion cassette system were used (68,69). Linearized plasmid DNA or PCR amplified DNA was used for homologous integration into the genome. Correct integration was verified by Southern hybridization. Total RNA was extracted from exponentially growing cells in axenic culture using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. To check RNA integrity, total RNA was run on an agarose gel and visualized by ethidium bromide staining. Residual DNA was removed from RNA samples using the TurboDNA-free kit (Ambion/Life Technologies). cDNA was synthesized using the RevertAid First Strand cDNA Synthesis kit (Thermo Scientific). Primers used in this study are listed in Table S2.
Quantitative RT-PCR. qRT-PCR analyses were performed as described before (58). qRT-PCR experiments were performed with three independent biological replicates and two technical replicates each (if not stated otherwise) using the MESA Green qPCR Mastermix Plus (Eurogentec). qRT-PCR was performed using the CFX Connect real-time PCR detection system and analyzed with the CFX Manager Maestro software (Bio-Rad).
Plasmid construction. For the kpp2-GFP fusion, 1 kb downstream of the kpp2 stop codon and 1 kb of the kpp2 open reading frame (ORF) were PCR amplified from genomic DNA and ligated to the 3.7-kb SfiI GFP-Hyg r fragment of pUMa317 (70). The resulting ligation product was integrated into the pCR2.1 TOPO vector (Invitrogen), generating plasmid pCR2.1 pUMa317-kpp2. For generation of the fuz7 DD -3ϫHA fusion, the fuz7 DD gene was amplified from the p123P crg1 :fuz7 DD plasmid (6) adding a C-terminal HA tag and two NdeI restriction sites. The resulting 1.3 kb NdeI fuz7 DD -HA fragment was ligated to the 8.1-kb NdeI fragment of the p123P crg1 :fuz7 DD backbone resulting in plasmid p123fuz7 DD -HA. To generate the rok1-mCherry fusion, 1 kb downstream of the rok1 stop codon and 1 kb of the rok1 ORF were PCR amplified from genomic DNA and ligated to the 3.1-kb SfiI mCherry-G418 fragment of plasmid pUMa1723. The ligation product was integrated into the pCR2.1 TOPO vector (Invitrogen), generating plasmid pCR2.1 rok1-mCherry.
To generate the rok1 deletion construct, 1 kb upstream of the rok1 start codon and 1 kb downstream of the rok1 stop codon were PCR amplified and ligated to the 2.4-kb SfiI Phleo r fragment of vector pUMa263 (69).
To replace the carboxin resistance in vector P cib1 :cib1 s (18) by nourseothricin resistance, the 4.2-kb HindIII Nat r fragment of vector pNEB-Nat r (30) and the 4.1-kb HindIII fragment of vector P cib1 :cib1 s were ligated to generate the plasmid pNEB-NatR-cib1 s .
To obtain the promoter fusions/gene replacement constructs for plant infection, 1-kb flanking regions up-and downstream of the genes UMAG_12184 and UMAG_03597, the kpp2 ORF and the Hyg r of vector pUMa1442 were PCR amplified. Thereby, SfiI restriction sites were added to the kpp2 gene, a SfiI restriction site was added to the 3= end of the upstream flanking region, a SfiI (N terminal) and a BamHI (3= end, for UMAG_12184) or KpnI (3= end, for UMAG_03597) restriction site were added to the Hyg r , and a BamHI (5= end, for UMAG_12184) or KpnI (5= end, for UMAG_03597) restriction site was added for the downstream flanking region of the respective gene. The obtained fragments were ligated to obtain a LB-kpp2-Hyg r -RB fragment. The fragment was then integrated into the pCR2.1 TOPO vector (Invitrogen) generating plasmids pCR2.1 P UMAG_12184 :kpp2 (Hyg r ) and pCR2.1 P UMAG_03597 :kpp2 (Hyg r ). PCR-amplified constructs were transformed into the FB1 Δkpp2 and FB2 Δkpp2 strain background, replacing the conditionally expressed UMAG_12184 and UMAG_03597 genes as described in reference 71. Resistance marker cassettes used for integration were subsequently removed by FLP/FRT recombination to maintain the native genomic context.
For generation of the yeast-two-hybrid fusion proteins, the genes of interest were PCR amplified adding SfiI restriction sites and ligated to modified version of the vectors pGBKT7 and pGADT7 (Clontech) containing SfiI restriction sites.
Mating assay/filamentous growth assay. Mating assays were performed as described in reference 30. Cells were grown in CM-glucose medium overnight, and optical density at 600 nm (OD 600 ) was measured and adjusted to 0.2 in fresh CM-glucose medium. Cells were grown for 4 h at 28°C. OD 600 was then adjusted to 1.0 in CM-glucose medium. Compatible FB1 and FB2 strains and derivatives were mixed 1:1, and 5 l of the mixture was spotted on potato dextrose medium supplemented with 1% charcoal (PD-CC) (63). To test for filamentous growth of SG200, 5 l of SG200 and derivatives were spotted on PD-CC medium and incubated for 24 to 48 h at 28°C.
Pheromone and cAMP treatment. To test for conjugation tube formation of cells in liquid culture, cultures of U. maydis strains were grown in CM-glucose medium at 28°C to reach an OD 600 of 0.25.
Synthetic a2 pheromone (final concentration, 2.5 g/ml) or dimethyl sulfoxide (DMSO) as negative control was added, and 2 ml of cell suspension was incubated in a 15-ml reaction tube on a rotating wheel at 6 rpm for 6 h. Cells were then microscopically analyzed using an Axio Imager.M2 equipped with an AxioCam MRm camera (Zeiss). Images were processed using ImageJ. Cells were harvested by centrifugation and flash frozen in liquid nitrogen for subsequent mRNA isolation.
For cAMP treatment, U. maydis strains were grown in CM liquid medium supplemented with 1% glucose to reach an OD 600 of 0.2 and treated with 6 mM cAMP (final concentration) for 12 h at 28°C, harvested by centrifugation, and flash frozen in liquid nitrogen for subsequent mRNA isolation.
Protein procedures. For preparation of protein extracts, overnight cultures of U. maydis were grown in CM liquid medium supplemented with 1% glucose, adjusted to an OD 600 of 0.25 in CM supplemented with 1% glucose (noninducing) or arabinose (induction of P crg1 -controlled expression) and further incubated for 6 h at 28°C. Cells were then harvested by centrifugation and washed once with lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% Triton X-100) supplemented with 2ϫ Complete protease inhibitor (PI; Roche). Cell pellets were resuspended in 100 to 200 l lysis buffer plus 2ϫ PI, supplemented with glass beads and flash frozen in liquid nitrogen. Cell lysis was performed on a Vibrax shaker (IKA) at 2,200 rpm at 4°C for 30 min. The resulting suspension was cleared by centrifugation at 20,000 rcf at 4°C for 10 min. The supernatant was supplemented with 4ϫ Roti-load 1 buffer (Carl Roth), boiled for 5 min at 95°C, and subjected to SDS-PAGE.
For immunoprecipitation of proteins, overnight cultures of U. maydis were grown in CM supplemented with 1% glucose, adjusted to an OD 600 of 0.25 in CM supplemented with 1% glucose (noninducing) or arabinose (induction of P crg1 -controlled expression), and incubated for 6 h at 28°C. Cells were then harvested by centrifugation, washed once with IP lysis buffer (25 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% NP-40, 5% glycerol) supplemented with 2ϫ Complete protease inhibitor (PI; Roche). The pellet was frozen in liquid nitrogen and disrupted in a cell mill (Retsch MM400, 30 Hz, 2 min, 2ϫ). The disrupted cells were then resuspended in 750 l IP lysis buffer plus 2ϫ PI and centrifuged at 22,000 rcf for 30 min at 4°C. The supernatant was added to 25 l of agarose GFP-Trap or RFP-Trap beads (Chromotek) and incubated for 3 h at 4°C on a rotating wheel. After incubation, beads were washed 3 to 5 times with IP lysis buffer plus 1ϫ PI. Fifty microliters 2ϫ Roti Load 1 (Carl-Roth) was added to the beads and they boiled at 95°C for 5 min. Samples were run on a 10% (Kpp2-GFP/Fuz7 DD -HA) or 8% (Rok1-mCherry) SDS-PAGE and subsequently subjected to Western hybridization. All experiments were repeated at least three times.
Plant infection studies. For infection studies of maize (Zea mays), the solopathogenic strain SG200 and derivatives or FB1 and FB2 and their respective derivatives were grown to an OD 600 of 0.6 to 0.8 in YEPS light medium, adjusted to an OD 600 of 1.0 in water. Haploid strains (FB1, FB2, and derivatives) were mixed 1:1 with a compatible mating partner. Approximately 500 l of the resulting cell suspension was used to inoculate 8-day-old maize seedlings of the maize cultivar Early Golden Bantam. Plants were grown in a CLF Plant Climatics GroBank with a 14-h (28°C) day and 10-h (22°C) night cycle. Symptoms were scored according to the disease rating criteria reported in reference 27. Three independent clones were used for plant infection experiments or as otherwise indicated, and the average scores for each symptom are shown. Photos from infected leaves were taken and represent the most common infection symptom for the respective mutant.
Statistical analysis. Statistical significance was calculated by one-way analysis of variance (ANOVA), followed by Bonferroni's multiple-comparisons test. The statistical significance of plant infection phenotypes was calculated using the Mann-Whitney test as described previously (65). Results were considered significant if the P value was Ͻ0.05.
Data availability. Sequence data from this article can be found in the National Center for Biotechnology Information (NCBI) database under the following accession numbers: biz1, UMAG_02549, XP_011388956.