Redox Regulation, Rather than Stress-Induced Phosphorylation, of a Hog1 Mitogen-Activated Protein Kinase Modulates Its Nitrosative-Stress-Specific Outputs

ABSTRACT In all eukaryotic kingdoms, mitogen-activated protein kinases (MAPKs) play critical roles in cellular responses to environmental cues. These MAPKs are activated by phosphorylation at highly conserved threonine and tyrosine residues in response to specific inputs, leading to their accumulation in the nucleus and the activation of their downstream targets. A specific MAP kinase can regulate different downstream targets depending on the nature of the input signal, thereby raising a key question: what defines the stress-specific outputs of MAP kinases? We find that the Hog1 MAPK contributes to nitrosative-stress resistance in Candida albicans even though it displays minimal stress-induced phosphorylation under these conditions. We show that Hog1 becomes oxidized in response to nitrosative stress, accumulates in the nucleus, and regulates the nitrosative stress-induced transcriptome. Mutation of specific cysteine residues revealed that C156 and C161 function together to promote stress resistance, Hog1-mediated nitrosative-stress-induced gene expression, resistance to phagocytic killing, and C. albicans virulence. We propose that the oxidation of Hog1, rather than its phosphorylation, contributes to the nitrosative-stress-specific responses of this MAP kinase.

IMPORTANCE Mitogen-activated protein kinases play key roles in the responses of eukaryotic cells to extracellular signals and are critical for environmental-stress resistance. The widely accepted paradigm is that MAP kinases are activated by phosphorylation, which then triggers their nuclear accumulation and the activation of target proteins and genes that promote cellular adaptation. Our data suggest that alternative forms of posttranslational modification can modulate MAP kinase functionality in Candida albicans. We demonstrate that Hog1 is not significantly phosphorylated in response to nitrosative stress, yet it displays nuclear accumulation and contributes to the global transcriptional response to this stress, as well as promoting nitrosative-stress resistance. Instead, nitrosative stress triggers changes in the redox status of Hog1. We also show that specific Hog1 cysteine residues influence its activation of stress genes. Therefore, alternative posttranslational modifications appear to regulate the stress-specific outputs of MAP kinases.
KEYWORDS Candida albicans, MAP kinase signaling, nitrosative stress T o survive, all organisms must respond to environmental change. To achieve this, they activate specific signaling pathways that trigger adaptation to the external challenges, many of which are perceived as an environmental stress. Mitogen-activated protein kinase (MAPK) modules play central roles in many environmental responses in animals, plants, fungi, and protists (1)(2)(3)(4)(5).
The initial description of the Hog1 MAP kinase, which is critical for osmo-adaptation in Saccharomyces cerevisiae (1), was followed by the discovery of its mammalian homologues p38 and JNK (c-Jun N-terminal kinase) (2,3). Since then, the molecular and system analysis of fungal MAP kinase modules has contributed greatly to the current understanding of these modules and to stress-signaling networks in general (see, e.g., references 6 to 9). The generally accepted paradigm is that signal perception leads to the activation of a MAP kinase kinase kinase (MAPKKK), which phosphorylates and activates its MAP kinase kinase (MAPKK), which in turn phosphorylates and activates the MAPK and promotes its nuclear accumulation. This holds for the pathogenic yeast Candida albicans, where the Pbs2 MAPKK phosphorylates the Hog1 MAPK at the threonine and tyrosine residues in its highly conserved TGY motif in response to a variety of environmental stresses that include osmotic, oxidative, and heavy metal stresses (10,11). Hog1 phosphorylation leads to its accumulation in the nucleus, the activation of target genes, and the phosphorylation of target proteins, which together contribute to stress adaptation, stress protection, and virulence (12)(13)(14)(15).
The paradigm of MAPK phosphorylation and activation is now generally accepted to the extent that Hog1 phosphorylation has become a standard and quantifiable proxy for Hog1 activation (see, e.g., reference 7). However, although C. albicans Hog1 is phosphorylated in response to diverse stresses, including osmotic, oxidative, and heavy metal stresses (10), it activates different gene targets in response to these stresses (13). For example, Hog1 plays a significant role in the induction of genes encoding cation exporters and glycerol biosynthetic enzymes in response to osmotic stress and drug transporter and sulfur amino acid biosynthetic genes in response to heavy metal stress. However, Hog1 plays a relatively minor role in the induction of many oxidative-stressresponsive genes following exposure to hydrogen peroxide despite being required for resistance to this oxidative stress, suggesting greater emphasis on posttranscriptional regulation by Hog1 under these conditions (13). Therefore, mechanisms in addition to TGY phosphorylation must control the stress-specific outputs of Hog1. Here, we provide evidence for this in the context of nitrosative stress, which contributes to the antifungal activity of the myeloid cells that are central to our innate immune defenses (16). We show that Hog1 displays minimal levels of phosphorylation following exposure to nitrosative stress yet contributes to the resistance of C. albicans to this stress. We also demonstrate that Hog1 becomes oxidized in response to nitrosative stress, that this MAP kinase drives different transcriptional outputs in response to nitrosative and oxidative stresses, and that specific cysteine residues in Hog1 affect these nitrosativeand oxidative-stress outputs. Therefore, modifications at non-TGY residues influence the stress-specific outputs of a MAPK in vivo.

RESULTS
Hog1 contributes to nitrosative-stress resistance. C. albicans exists as a relatively harmless commensal in most healthy individuals and thus has evolved to evade or resist attack by our immune defenses. Macrophages and neutrophils exploit a battery of reactive oxidative and nitrosative species (ROS and RNS) in their attempts to kill invading microbes (16). However, compared to nonpathogenic fungi, C. albicans displays relatively high levels of resistance to these chemical species (17,18). The resistance of C. albicans to oxidative stress is mediated mainly by AP-1 (Cap1) and Hog1 signaling (10,(19)(20)(21)(22)(23), and accordingly, Cap1 and Hog1 promote fungal survival following phagocytic attack (22,24,25). The resistance to nitrosative stresses is dependent on the transcription factor Cta4 (26), which drives the transcriptional response to nitrosative stress and the activation of YHB1, which encodes a conserved flavohemoglobin, nitric oxide dioxygenase, that detoxifies nitric oxide (NO) (27,28).
Previously, Hog1 was not thought to contribute significantly to the nitrosative-stress response in C. albicans, in part because the inactivation of this MAP kinase does not appear to block the nitrosative-stress-mediated induction of a YHB1-RLUC reporter (24). However, we observed that C. albicans hog1⌬ cells are sensitive to NO and nitrite ( Fig. 1A; see also Table S1 in the supplemental material), suggesting that Hog1 contributes to nitrosative-stress resistance. This contribution appears to be dependent on phosphorylation at the TGY motif because mutation of these threonine and tyrosine phosphorylation sites in Hog1 to nonphosphorylatable residues (T174A, Y176F) (15) renders the resulting hog1 AF cells sensitive to nitrosative stress (Fig. 1A). Yet, compared to the strong phosphorylation observed in response to osmotic stress (1 M NaCl for 10 min), Hog1 displayed minimal levels of phosphorylation following exposure to 5 mM sodium nitrite or to 2.5 or 6 mM dipropylenetriamine (DPTA)-NONOate (Fig. 1B), which represent medium-to-high levels of this NO donor (29). Despite this minimal phosphorylation, Hog1-yellow fluorescent protein (YFP) accumulated in the nucleus in response to a nitrosative stress ( Fig. 1C and S1). We conclude that Hog1 accumulates in the nucleus in response to nitrosative stress and contributes to nitrosative-stress resistance but that Hog1 is not strongly phosphorylated in response to this stress.
Hog1 contributes to the transcriptional response to nitrosative stress. Given that this MAP kinase accumulated in the nucleus after exposure to nitrosative stress, we wondered whether Hog1 contributes to the transcriptional response to this stress. To test this, we performed transcript profiling on C. albicans wild-type and hog1⌬ cells exposed to 0 or 2.5 mM DPTA-NONOate by RNA sequencing. In wild-type cells, 321 genes consistently displayed a Ն2-fold increase in expression in response to this stress (Table S2). Hromatka and coworkers (28) described fewer nitrosative-stress-induced genes in their microarray analysis of the transcriptional response to nitrosative stress in C. albicans. This is to be expected given (i) the increased depth of coverage and higher dynamic range observed for RNA sequencing studies (30) and (ii) the lower dose that they examined in their study (1.0 versus 2.5 mM DPTA-NONOate). Of the 40 genes that were induced over 3-fold by nitrosative stress in their microarray study (see Table 2 in reference 28, 80% were upregulated in response to nitrosative stress in our study (Table S3). YHB1 was among the most strongly regulated genes in both studies (Table S3). Furthermore, both studies found that, in addition to YHB1, genes involved in oxido-reduction, the oxidative-stress response, and carbohydrate metabolism were induced in response to nitrosative stress ( Fig. 2A; Table S3).
We observed that the induction of a significant proportion of the nitrosative-stressinduced genes (64%) was attenuated in hog1⌬ cells ( Fig. 2A; Table S4) indicating that, despite the low levels of Hog1 phosphorylation in response to nitrosative stress (Fig. 1B), this MAP kinase makes a significant contribution to the transcriptional response to nitrosative stress in C. albicans. We compared the 207 Hog1-dependent nitrosative-stress genes identified by RNA sequencing in this study with the Hog1dependent oxidative-and osmotic-stress genes identified in our earlier microarray study (13). Fewer Hog1-dependent stress genes were observed in this microarray study than in our current RNA sequencing study (Table S4), no doubt because of the higher dynamic range of RNA sequencing (30). Nevertheless, only half of the Hog1-dependent oxidative-stress-induced genes and one-fifth of the Hog1-dependent osmotic-stressinduced genes observed on the microarrays also displayed Hog1-dependent induction in response to nitrosative stress (Table S4). Forty-three Hog1-dependent nitrosativestress genes retained some degree of nitrosative-stress activation in hog1Δ cells (Table S5). Of these 43 genes, 19 are induced in response to oxidative stress and 3 are upregulated by osmotic stress (Table S5). This suggests partial overlap between the Hog1 outputs for these different stress conditions.
As stated, genes involved in oxido-reduction and the oxidative-stress response were induced in response to nitrosative stress ( Fig. 2A). This was to be expected because NO inhibits mitochondrial oxidative phosphorylation, thereby generating superoxide, peroxynitrite, and hydrogen peroxide (31,32). Consistent with this, we also observed that exposure to nitrosative stress led to an increase in the intracellular levels of ROS in the wild type. This was also the case in hog1Δ, cta4Δ, and cap1Δ cells (Fig. 2B), which lack key regulators of the oxidative-and nitrosative-stress responses (10,13,19,21,26). Interestingly, the loss of Cta4 led to a 7.2-fold increase in ROS after the nitrosative stress, whereas the other strains displayed a 3.7-fold increase, largely because basal ROS levels were lower in cta4Δ cells (Fig. 2B). Most importantly, the data suggest that the nitrosative-stress-mediated induction of oxidative-stress genes might be triggered by the elevation of ROS in these strains. In addition, nitrosative stress causes molecular damage, such as protein S-nitrosylation. Accordingly, we observed the induction of genes involved in glutathione synthesis and recycling (GCS1, GTT1, GST2) and the glutaredoxin and thioredoxin systems (GRX2, GPX3, TRR1) in response to nitrosative stress (Table S2). Therefore, we tested the relative contributions of Hog1, Cta4, and Cap1 to the induction of classic nitrosative (YHB1)-and oxidative (TRR1)-stress genes. Interestingly, all three regulators contributed significantly to the nitrosative-stressmediated induction of both YHB1 and TRR1 (Fig. 2C). This suggests that nitrosative stress generates intracellular ROS, which leads to Cap1 activation, which then contributes to the transcriptional response to the nitrosative stress together with Cta4. We also conclude that Hog1 makes a significant contribution to the activation of some oxidative-stress genes, as well as nitrosative-stress genes, in response to nitrosative stress. This was unexpected because Hog1 is not required for the induction of most oxidative-stress genes in response to oxidative stress (13) (Table S4). These observations are consistent with the idea that Hog1 mediates different outputs in response to nitrosative and oxidative stresses.
Hog1 is oxidized in response to nitrosative stress. Hog1 displays minimal levels of phosphorylation in response to nitrosative stresses compared with the strong phosphorylation observed with osmotic stress (Fig. 1B), yet Hog1 contributes to the transcriptional response to nitrosative stress (Fig. 2). Might Hog1 be activated by some other means in response to this stress? In mammalian systems, S-nitrosylation or oxidation of redox-sensitive cysteine residues is emerging as an important signaling mechanism following nitrosative and oxidative stresses (33)(34)(35)(36), and the mammalian  Table S1) and hog1⌬ cells (JC50) exposed to 0 or 2.5 mM DPTA-NONOate for 10 min. In wild-type cells, 321 transcripts displayed statistically significant (Ͼ2-fold) increases in level in response to the nitrosative stress. Of these, 205 transcripts (64%) were considered to be Hog1-dependent (red nodes) because they were not induced Ն2-fold by nitrosative stress in hog1⌬ cells. The remaining 116 transcripts were considered to be Hog1 independent because they were still induced Ͼ2-fold in hog1⌬ cells. Using Cytoscape, GO term analysis was performed on these gene subsets, and the outputs were displayed as a gene function network in which the red nodes represent Hog1-dependent genes, the dark grey nodes represent Hog1independent genes, and the central hubs are colored according to the functional category. For example, blue hubs relate to stress, green hubs relate to metabolism, and the light-gray hub is for genes of unknown function. (B) The levels of intracellular ROS were assayed in C. albicans strains exposed to 0 or 2.5 mM DPTA-NONOate by performing cytometry on DHE-stained cells. Wild type (WT), gray; hog1Δ, red; cta4Δ, green; cap1Δ, yellow. DHE intensity is presented on a log scale, and the mean fluorescent intensity for each cytometry profile is shown. The data shown are representative of three independent experiments. (C) The fold induction of classical nitrosative-stress (YHB1) and oxidative-stress (TRR1) transcripts is shown following exposure of the same C. albicans strains to 2.5 mM DPTA-NONOate for 30 min. Using qRT-PCR, transcript levels were measured relative to the internal ACT1 mRNA control and normalized to the level of that transcript in the absence of the stress. Means and standard deviations are shown for three independent replicate experiments. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001. Hog1 Promotes Nitrosative-Stress Adaptation ® MAP kinases JNK1 and p38 are negatively regulated by S-nitrosylation and cysteine oxidation (37)(38)(39). Also, oxidative-stress-induced disulfide bond formation between redox-sensitive cysteine residues in the Hog1 orthologue Sty1 positively regulates the transcriptional response to oxidative stress in Schizosaccharomyces pombe (40). The oxidation of redox-sensitive thiol groups to the unstable sulfenic form often triggers the formation of the more stable disulfide bond. In addition, the S-nitrosylation of cysteine residues is often the precursor of disulfide bond formation (41,42). Therefore, we probed for nitrosative-stress-induced changes in disulfide bond formation within Hog1 by treating C. albicans protein extracts sequentially with N-ethylmaleimide (NEM), dithiothreitol (DTT), and 4-acetamido-4=-maleimidylstilbene-2,2=-disulfonic acid (AMS) (23). Both NEM and AMS specifically alkylate sulfhydryl groups on reduced cysteine residues, but while NEM alkylation has a minimal effect on a protein's mass, AMS alkylation adds 0.6 kDa per modified cysteine (43). First, we treated extracts with NEM to block reduced cysteines (Fig. 3A). Then, these extracts were treated with DTT to reduce any disulfides, and the released cysteine residues were then alkylated with AMS  (Table S1) was analyzed by Western blotting of AMS-and NEM-alkylated extracts with the anti-Hog1 antibody. As shown in the cartoon, proteins were first alkylated with NEM (Ͻ0.1 kDa) to block free thiols. Disulfide bonds were then reduced, and the newly exposed thiols were alkylated with AMS (~0.5 kDa). Control extracts that were not alkylated with AMS show that, in unstressed C. albicans cells, no disulfide bonds exist in Hog1. The increase in Hog1 mass following DPTA-NONOate treatment indicates the formation of a new disulfide bond(s) in these cells. (B) Comparisons of parts of the amino acid sequences of C. albicans (Ca) and S. cerevisiae (Sc) Hog1, S. pombe (Sp) Sty1, and Homo sapiens (Hs) p38, showing the juxtapositions of C156, C161, and C271 (pink) relative to the TGY phosphorylation motif (blue) in evolutionarily conserved stress-activated protein kinases. *, identical residue; :, conserved residue. (C) Comparison of the predicted structural model of C. albicans Hog1 (green) against the human p38 (gold) model (PDB accession number 1R39) indicates that their structures are very similar, with the same overall fold. The threonine and tyrosine residues in the TGY motif are highlighted in blue (stick representation), and the relevant cysteine residues are highlighted in red. A slightly less exposed positioning of these residues is predicted in Hog1. (D) Cartoon and surface representation of the predicted structural modeling of C. albicans Hog1 shows the positions of the targeted cysteines (red-dot spheres, C156, C161, and C271), relative to the TGY phosphorylation motif (stick representation, blue). N-term, N terminus.
Herrero-de-Dios et al. ® (Fig. 3A). Therefore, AMS-dependent increases in the molecular mass of a protein, as detected by a reduced mobility in SDS-PAGE, indicate the presence of disulfide bonds in that protein.
Western blotting of NEM-DTT-AMS-treated C. albicans extracts revealed that, unlike with Sty1 in S. pombe (40), oxidative stress does not induce disulfide bond formation in Hog1 (Fig. 3A). This was first observed using 5 mM H 2 O 2 but was also the case even when C. albicans cells were exposed to a high dose of H 2 O 2 (30 mM) (Fig. 3A). In contrast, such high levels of H 2 O 2 rapidly induce oxidation of Sty1 in S. pombe (40). Interestingly, however, there was a clear induction of disulfide bond formation in Hog1 after C. albicans cells were exposed to 2.5 mM DPTA-NONOate. This nitrosative stress caused an AMS-dependent increase in the molecular mass of Hog1 (Fig. 3A), indicating that the stress had led to disulfide bond formation in this MAP kinase. In principle, this disulfide bond formation might represent S-nitrosylation, as S-nitrosylation at cysteine residues can lead to disulfide bond formation (41,42).
In S. pombe Sty1, the cysteine residues C153 and C158 form a disulfide in response to oxidative stress (40). These residues lie in a highly evolutionarily conserved region of this MAP kinase (Fig. 3B), and therefore, we targeted the corresponding residues in C. albicans Hog1 for analysis (C156 and C161). In addition, we performed proteomics to test whether other Hog1 residues might be subject to modification in response to nitrosative stress. Briefly, C. albicans cells expressing tandem-affinity purification (TAP)tagged Hog1 (JC310) (Table S1) were treated with 0 or 2.5 mM DPTA-NONOate, Hog1 was purified from protein extracts by tandem-affinity purification, and the Hog1 tryptic peptides were subjected to mass spectrometry to detect irreversible nitrosylation or oxidation events (see Materials and Methods). C156-and C161-containing tryptic peptides were observed (DLKPSNILINENCDLK; ICDFGLAR), but no irreversible oxidation events, such as cysteine hyperoxidation, were detected on these peptides. (Reversible oxidation events, such as disulfide bond formation, would not have been detected because the sample preparation included reduction with DTT and S-alkylation with iodoacetamide.) However, we did detect hyperoxidation of Hog1 C271 to a sulfonic acid residue (ϪSOHO 2 ) in cells exposed to the nitrosative stress (Fig. S2). Therefore, we also targeted C271 for analysis. Unlike C156 and C161, C271 is not conserved in the Hog1 orthologues from S. cerevisiae, S. pombe, and humans (Fig. 3B).
C156, C161, and C271 exert differential effects on Hog1 outputs. Mutated versions of C. albicans Hog1 were created by converting the cysteines at C156, C161, and C271 to serines, and these mutations were confirmed by resequencing. We performed structural modeling of C. albicans Hog1 using Phyre2 and compared it to the available crystal structure of a closely related protein, human p38, suggesting a conserved overall fold and very similar structures (Fig. 3C). The model revealed the likely juxtapositions of these three cysteine residues and the TGY motif within the Hog1 structure (Fig. 3D). Also, our modeling of the Hog1 C156S , Hog1 C161S , and Hog1 C271S proteins suggested that the C156S, C161S, and C271S mutations are unlikely to cause major perturbations in the overall structure (Fig. S3).
We sequentially treated samples with NEM, DTT, and AMS and then subjected them to gel electrophoresis to test whether these mutations affect the redox state of Hog1 (Fig. 4A). This revealed that the single C156S and C161S mutations promoted disulfide bond formation in Hog1 even in the absence of stress, as a significant fraction of Hog1 displayed an AMS-dependent decrease in Hog1 mobility following NEM-DTT-AMS treatment. However, neither mutation blocked the nitrosative-stress-induced changes in Hog1's redox state (Fig. 4A). Also, neither the C156S nor the C161S mutation affected the stress sensitivity of C. albicans when the phenotypes of the single HOG1 C156S and HOG1 C161S mutants were compared with those of wild-type and hog1 control strains following osmotic (1 M NaCl), oxidative (5 mM H 2 O 2 ), or nitrosative (5 mM NaNO 2 and 6 mM DPTA-NONOate) stress (Fig. 4B). Moreover, these mutations did not affect the phosphorylation dynamics of Hog1 in response to nitrosative or oxidative stress (Fig. S4). Therefore, individually, these mutations do not appear to affect Hog1 func-  (Table S1). For the C156/161S columns, the wild-type controls were from the same blot as that showing the lanes with (ϩ) and without (Ϫ) DPTA-NONOate and treatment with NEM, DTT, and AMS in Fig. 3A. (B) Impact of C156S, C161S, and C271S mutations upon the nitrosative-, oxidative-, and osmoticstress sensitivity of C. albicans. Strains were spotted in 10-fold serial dilutions onto plates containing no stress (control), 1 M NaCl, 5 mM H 2 O 2 , 5 mM NaNO 2 , or 2.5 mM DPTA-NONOate. All figure panels for the wild type, null mutant, and C156S and C161S single and double mutants were derived from the same set of plates. The C271 mutant, which displayed no obvious stress phenotype, was compared to the wild-type control in a separate experiment. (C) Impact of C156S, C161S, and C271S mutations upon the phosphorylation dynamics of Hog1 phosphorylation following exposure of cells to 2.5 mM DPTA-NONOate or 5 mM H 2 O 2 . Western blotting was performed on cell extracts prepared at the times indicated (in minutes). Phosphorylated Hog1 (P-Hog1) was detected by probing the blots with phospho-p38 antibody, and for loading controls, the blots were reprobed for total Hog1 and actin. The results are indicative of three independent replicate experiments, and the data for the individual HOG1 C156S and HOG1 C161S mutants are shown in Fig. S4. (D) Impact of C156S, C161S, and C271S mutations upon the Hog1-mediated induction of nitrosative (YHB1)-and oxidative (TRR1)-stress genes in response to 5 mM H 2 O 2 or 2.5 mM DPTA-NONOate for 10 min. The strains are as described for panel A. The fold induction of the YHB1 and TRR1 transcripts was measured by qRT-PCR, relative to the internal ACT1 mRNA control, and by normalizing to the levels of these transcripts in the absence of the stress. Means and standard deviations are shown for three independent replicate experiments. *, P Ͻ 0.05. tionality. The C156S C161S double mutation resulted in changes in Hog1's redox state similar to those produced by the corresponding single mutations (Fig. 4A) and exerted only minor effects upon Hog1 phosphorylation dynamics, with Hog1 C156S C161S displaying protracted phosphorylation in response to oxidative stress (Fig. 4C). However, the HOG1 C156S C161S double mutation attenuated the resistance of C. albicans to osmotic, oxidative, and nitrosative stresses, albeit not as strongly as a hog1 null mutation (Fig. 4A). It has been reported that the catalytic activity of the p38 MAP kinase remains unaffected by a C162S mutation but that this mutation influences the structure of its docking site for substrates and activators (64). Therefore, it is conceivable that, while single C156S and C161S mutations do not significantly affect Hog1 function, the double mutation subtly alters the structure of Hog1 and hence its functionality, thereby attenuating the stress resistance of C. albicans.
The C271S mutation did not block disulfide bond formation in Hog1 in the absence of stress (Fig. 4A) but did exert a dramatic effect on Hog1 phosphorylation (Fig. 4C). Compared to the wild-type control, Hog1 C271S displayed increased phosphorylation levels after nitrosative stress and strong phosphorylation in response to oxidative stress (Fig. 4C). These HOG1 C271S cells also displayed wild-type levels of resistance to nitrosative and oxidative stresses (Fig. 4B). Clearly, the C156S C161S double and C271 single mutations impose different effects upon Hog1 functionality.
To test this further, we examined the impact of these mutations on Hog1 target genes by quantitative reverse transcription-PCR (qRT-PCR). Hog1 contributes to the activation of the nitrosative-stress gene YHB1 following exposure to a nitrosative stress ( Fig. 2C; Table S4). As expected (27,28), YHB1 was strongly induced by DPTA-NONOate in wild-type cells (Fig. 4D). This induction was blocked in the HOG1 C156S C161S mutant but not in HOG1 C271S cells (Fig. 4D). YHB1 displayed limited induction in response to H 2 O 2 . The activation of YHB1 in response to oxidative stress was not affected significantly by the HOG1 C156S C161S and HOG1 C271S mutations (Fig. 4D).
We also tested the effects of the HOG1 C¡S mutants on the induction of the thioredoxin reductase (TRR1) gene. As expected (13,14), the expression of this classic oxidative-stress gene was induced by H 2 O 2 (Fig. 4D). TRR1 activation in response to oxidative stress was not affected in HOG1 C156S C161S or HOG1 C271S cells (Fig. 4D). TRR1 was also induced by DPTA-NONOate in wild-type cells (Fig. 2C and 4D). Interestingly, this induction was blocked in the HOG1 C156S C161S mutant but not in HOG1 C271S cells (Fig. 4D). Taken together, the data indicate that the HOG1 C271S mutation does not affect TRR1 or YHB1 activation in response to nitrosative or oxidative stress. In contrast, the HOG1 C156S C161S mutations block the induction of both genes specifically in response to nitrosative stress, but not oxidative stress.
We conclude that cysteine residues C156, C161, and C271 differentially affect Hog1 functionality. Although none of these cysteine residues contribute to nitrosative-stressinduced oxidation, the C271S mutation triggers aberrantly strong Hog1 phosphorylation in response to oxidative and nitrosative stresses. In addition, although the single C156S and C161S mutations exert no obvious effects upon Hog1 functionality, the double mutations compromise Hog1 function, attenuating stress resistance and inhibiting gene induction specifically in response to nitrosative stress.
Hog1 C156S, C161S, and C271S mutations differentially affect C. albicans virulence. Myeloid cells use combinations of ROS and RNS in their attempts to kill fungal cells and prevent infection (16). C. albicans attempts to resist this killing by mounting robust oxidative-and nitrosative-stress responses (44), and Hog1 contributes to this resistance to phagocytic killing (22,24,25). Therefore, we tested the ability of the HOG1 C¡S mutants to survive exposure to human polymorphonuclear granulocytes (a 1:10 ratio of fungal cells to polymorphonuclear granulocytes [PMNs] for 2 h). The single HOG1 C156S and HOG1 C161S mutants did not attenuate survival, but the HOG1 C156S C161S double mutant was as sensitive as the hog1⌬ mutant to phagocytic killing (Fig. 5A). In contrast, HOG1 C271S cells did not display a significant decrease in survival (Fig. 5A). Taken together, the data suggest that wild-type levels of stress resistance are required for resistance to phagocytic killing.
We then tested the impact of the Hog1 C¡S mutations upon the virulence of C. albicans during systemic infection. We used the invertebrate wax moth (Galleria mellonella) model, which is a well-established proxy for systemic infection of the mammalian host (45,46) and for the impact of Hog1 on systemic infection (47). As expected (12,15), the inactivation of Hog1 significantly attenuated the virulence of C. albicans (Fig. 5B, compare the HOG1 and hog1 cells), and the heterozygous null mutant displayed an intermediate level of virulence (Fig. 5B, HOG1/hog1Δ cells). Compared to this HOG1/hog1⌬ control, HOG1 C156S , HOG1 C161S , and HOG1 C271S cells did not display a significant reduction in virulence. In contrast, the virulence of HOG1 C156S C161S cells was significantly reduced. These observations were consistent with those from the phagocytic killing assays (above), reinforcing the view that Hog1 functionality and stress resistance are important for the virulence of C. albicans.

DISCUSSION
Our observations show clearly that the Hog1 MAP kinase contributes to nitrosativestress adaptation and resistance in the major fungal pathogen of humans, C. albicans. The inactivation of Hog1 confers sensitivity to RNS (Fig. 1A) and compromises the global transcriptional response to this type of stress (Fig. 2). This extends the breadth of roles already ascribed to this MAP kinase, which include contributions to osmotic, oxidative, heavy metal, cell wall, and antimicrobial peptide stress resistance, adaptation  to antifungal drugs and quorum-sensing molecules, morphogenetic regulation, and virulence (10,12,13,15,(48)(49)(50)(51).
The breadth of these roles, together with the observation that Hog1's outputs differ depending on the nature of the input signal (13), raised an important question about the current paradigm for MAP kinase activation which centers on the phosphorylation of these regulators. What defines the stress specificity of Hog1 outputs? This paradigm of MAP kinase activation appears to require significant tuning. First, our data indicate that strong Hog1 phosphorylation is not essential for accumulation of this MAP kinase in the nucleus (Fig. 1) or for its activation of gene targets in response to specific input signals (Fig. 2). We note that while strong Hog1 phosphorylation is not required for nitrosative-stress gene induction, at least basal phosphorylation of this MAP kinase is essential for nitrosative-stress adaptation, because a nonphosphorylatable hog1 AF mutant is sensitive to nitrosative stress (Fig. 1A). We were unable to detect any obvious effects of nitrosative stress upon the physical interaction between Hog1 and its export factor Crm1 by coimmunoprecipitation (see Fig. S5 in the supplemental material). Nevertheless, it is conceivable that while Hog1 phosphorylation might be required for entry into the nucleus, additional forms of posttranslational modification might promote the retention of this MAP kinase in the nucleus.
Second, and most significantly, we show that a MAP kinase can be oxidized in response to nitrosative stress. Several findings support this view. First, Hog1 forms disulfides following exposure to nitrosative stress (Fig. 3A). Second, trioxidation of residue C271 following nitrosative stress was observed directly by proteomic analyses of Hog1 tryptic peptides (Fig. S2). The conservative C271S mutation was predicted to not affect the tertiary structure of Hog1 significantly (Fig. S3), it did not affect Hog1 levels ( Fig. 4C and S4 and 4C), and it did not compromise the role of Hog1 in stress adaptation, because HOG1 C271S cells were as resistant as wild-type cells to nitrosative, oxidative, and osmotic stresses (Fig. 4B). However, dramatic changes in Hog1 phosphorylation dynamics were observed in HOG1 C271S cells (Fig. 4C), but these did not appear to affect the induction of Hog1 target genes (Fig. 4D). We note that, unlike the situation in S. cerevisiae and Cryptococcus neoformans (52,53), hyperphosphorylation of Hog1 is not lethal in C. albicans (65).
Third, we show that specific, evolutionarily conserved cysteine residues (C156 and C161) influence the nitrosative-stress outputs of Hog1. The induction of key stress genes (YHB1 and TRR1) was compromised in response to nitrosative, but not oxidative, stress in HOG1 C156S C161S cells but not HOG1 C271S cells (Fig. 4D). Taken together, the data suggest that alternative Hog1 posttranslational modifications, other than TGY phosphorylation, contribute to the regulation of stress-specific outputs for this MAP kinase.
In the general context of MAP kinase signaling, our data indicate that the redox state of a fungal MAP kinase is altered in response to nitrosative stress and that specific cysteine residues in this MAP kinase promote the stress specificity of its outputs. Posttranslation modifications other than phosphorylation have been shown to modulate the activity and specificity of other MAP kinases. For example, the transcriptional outputs of Sty1 in S. pombe are altered by oxidation in response to oxidative stress (40), and JNK1 transcription is downregulated in macrophages by S-nitrosylation (38). Extending to other types of regulatory network, ROS-mediated oxidation activates AP-1-like transcription factors in evolutionarily divergent yeasts (19,23,54), and the oxidation of NF-B downregulates its transcriptional activity during anti-Candida inflammatory responses (55). In the context of fungal infection, our data indicate that Hog1 cysteine residues moderate outputs that affect the resistance of C. albicans to phagocytic killing (Fig. 5A) and its ability to cause systemic infection (Fig. 5B). We conclude that stress-specific posttranslational modifications, other than the canonical TGY phosphorylation, occur on the Hog1 stress-activated protein kinase (SAPK) of a major human pathogen and propose this as a key mechanism to drive the stressspecific outputs of this SAPK.

MATERIALS AND METHODS
Strain construction and culture conditions. Strains used in this study are listed in Table S1 in the supplemental material. C. albicans was routinely grown at 30°C and 200 rpm in yeast extract-peptonedextrose (YPD) (56) or in Tris-buffered YPD medium (YPDT; pH 7.4) as described previously (29). Overnight cultures were diluted into fresh YPD or YPDT to an optical density at 600 nm (OD 600 ) of 0.2 and incubated until they reached an OD 600 of 0.8, whereupon they were subjected to the appropriate treatment. Nitrosative stress was imposed with 5 mM NaNO 2 , 5 mM NaNO 2 plus 25 mM succinic acid, or DPTA-NONOate at the concentrations specified in the figures. Oxidative stress was applied with 5 mM hydrogen peroxide (H 2 O 2 ) and osmotic stress with 1 M NaCl (29).
For stress sensitivity tests, C. albicans strains were grown to an OD 600 of 0.8 and then 10-fold serial dilutions were spotted onto YPD or YPDT plates containing the compounds indicated in the figures. Plates were incubated at 30°C for 48 h.
The HOG1 C¡S mutants were constructed by performing site-directed mutagenesis with the QuikChange site-directed mutagenesis kit by following the manufacturer's specifications (Agilent, Stockport, United Kingdom). The mutagenesis was performed on the plasmid CIp-C-Hog1HM (10) with the primers listed in Table S1. All of the plasmids generated (CIp-C-Hog1 WT [where "WT" indicates "wild type"], CIp-C-Hog1 C156S , CIp-C-Hog1 C161S , CIp-C-Hog1 C156S C161S , CIp-C-Hog1 C271S ) were linearized by digestion with BlpI and integrated into the wild-type HOG1 allele in JC36 (hog1⌬/HOG1) (Table S1) to generate the HOG1 WT , HOG1 C156S , HOG1 C161S , HOG1 C156S C161S , and HOG1 C271 strains (Table S1). The HOG1 open reading frame in each mutant was fully resequenced to confirm the presence of the appropriate mutation and the absence of any secondary mutations.
Transcript profiling and analysis. C. albicans wild-type (RM1000ϩCIp20) ( Table S1) and hog1⌬ (JC50) cells were grown in YPDT to an OD 600 of 0.8, exposed to 0 or 2.5 mM DPTA-NONOate for 10 min, and then immediately harvested and frozen at Ϫ80°C. RNA was extracted from these cells by using TRIzol reagent (Invitrogen, Paisley, UK) and FastPrep-24 (MP Biomedicals, Luton, UK) according to the manufacturer's instructions. The RNA was treated with DNase (Turbo DNase; Ambion, Banchory, UK) and assessed using an Agilent 2100 Bioanalyzer with RNA 6000 nanokits (Agilent). Samples with an RNA integrity number (RIN) of Ͼ8 were used for RNA sequencing.
RNA sequencing was performed using an Ion Torrent Proton sequencer. Raw fastq files were processed through FastQC (v. 10.1), Trim Galore (v. 3.1), Samtools (v. 1.19), STAR aligner (v. 2.4), and HTSeq (v. 5.4). Genome alignments were generated against the C_albicans_SC5314_version_A21-s02-m09-r08 chromosome file from the Candida Genome Database (http://www.candidagenome.org) (57). Gene expression analysis was performed using Partek Genomics Suite software, v. 6.6, using a log 2 data transformation. Gene ontology (GO) term analysis was performed through the Candida Genome Database GO Term Finder and the Cytoscape v. 3 Clue GO plug-in (58). Network construction was performed with cytoscape v.3 freeware (59). A statistical comparison among GO term enrichment percentages was performed with GraphPad Prism (v. 6) using Student's t test for two-tailed data. Data from three independent biological replicates for each condition are available at EBI (https://www.ebi.ac.uk/ arrayexpress/) under accession number E-MTAB-5990.
The levels of specific transcripts were measured by qRT-PCR with a Roche Light Cycler 480 II, in triplicate, using independent biological replicates (60). The primers are described in Table S1. The data were expressed relative to the internal ACT1 mRNA control and then normalized against the transcript levels in unstressed cells. Results were statistically analyzed using one-way analysis of variance (ANOVA) and multiple-comparison and post-Dunnett's test.
Hog1 localization. C. albicans cells expressing YFP-tagged Hog1 (JC63) (Table S1) were grown to exponential phase in YPD or YPDT, exposed to the relevant stress, collected after 10 min, fixed in 3.7% paraformaldehyde, and prepared as described previously (10). Nuclei were localized by staining with DAPI (4=,6-diamidino-2-phenylindole), and YFP fluorescence was captured using a DeltaVision Core microscope (Applied Precision, Issaquah, WA). Data from one of three independent experiments are shown; all displayed similar effects. Quantification was performed using ImageJ 1.48 software. Results were analyzed statistically using one-way ANOVA and Dunnett's multiple-comparison test on at least 10 cells per condition.
Hog1 phosphorylation and redox state. To examine Hog1 phosphorylation, exponentially growing C. albicans cells were exposed to the stated stress, protein extracts were prepared at the times indicated in the figures, and these extracts were subjected to Western blotting using previously described protocols (10). Phosphorylated Hog1 was detected using a phospho-p38 antibody (Thr180/Tyr182 number 9211; Cell Signalling Technology, Leiden, The Netherlands), total Hog1 was examined with an anti-Hog1 antibody (y-215, sc-9079; Santa Cruz Biotechnology, Heidelberg, Germany), and actin was analyzed with an antiactin antibody (A5060; Sigma-Aldrich, Dorset, UK). The secondary antibody was horseradish peroxidase-linked anti-rabbit IgG (New England Biolabs, Hitchin, UK), which was detected using the ECL SuperSignal West Femto system (Thermo Fisher Scientific). Hog1 phosphorylation was quantified using ImageJ 1.48 software. The data shown are from one of three independent experiments, all of which showed similar effects.
To assay Hog1 redox state, lysates were prepared from 5 ϫ 10 7 C. albicans cells in trichloroacetic acid (TCA) essentially as described previously (23,40). Briefly, cells were harvested in buffer (100 mM Tris-HCl, pH 8.0, 1% SDS, 1 mM EDTA) containing 20% TCA and snap-frozen. Protein extracts obtained using FastPrep-24 (MP Biomedicals) were washed with acetone, resuspended in the same buffer containing 10% TCA, and washed with acetone. Pellets were treated with 25 mM NEM (Sigma-Aldrich) for 20 min at 25°C in TCA buffer to block free thiols. Proteins were then reprecipitated in TCA, washed in acetone, and then incubated with 50 mM DTT for 1 h at 37°C to reduce existing disulfides. To alkylate the newly Herrero-de-Dios et al. ® exposed thiols, the extracts were reprecipitated with TCA, washed in acetone, and incubated with 25 mM AMS in TCA buffer at 25°C for 30 min and then at 37°C for 5 min. Protein yields were quantified using a BCA protein assay kit (Thermo Fisher Scientific). The alkylated extracts were then subjected to SDS-PAGE, using a loading dye lacking ␤-mercaptoethanol, and then to Western blotting with rabbit anti-Hog1 antibodies (y-215, sc-9079; Santa Cruz Biotechnology). The data are representative of three independent experiments, all of which showed similar effects.
Proteomic analysis of Hog1 tryptic peptides. C. albicans JC310 cells expressing TAP-tagged Hog1 and its control with untagged Hog1 (RM1000ϩClp20) (Table S1) were grown to an OD 600 of 0.8 in YPDT, exposed to 0 or 2.5 mM DPTA-NONOate for 10 min, harvested, and washed in chilled distilled water. Cell extracts were prepared in 25 mM Tris-HCl, pH 7.5, 15 mM EGTA, 15 mM MgCl 2 , 0.1% NP-40, 1 mM DTT, 0.1 mM NaF, and 1 mM phenylmethylsulfonyl fluoride (PMSF), supplemented with protease inhibitor tablets (cOmplete, EDTA-free protease inhibitor cocktail; Roche, Welwyn Garden City, UK), and frozen in liquid nitrogen. TAP-tagged Hog1 was purified using Sepharose-IgG beads (GE Healthcare, Little Chalfont, UK), followed by tobacco etch virus (TEV) protease cleavage and purification with calmodulin affinity resin (Agilent Genomics) as described previously (61,62). Samples were precipitated using a ReadyPrep two-dimensional cleanup kit (Bio-Rad Laboratories Ltd., Watford, UK) and resuspended in 100 l 50 mM ammonium bicarbonate. Protein digestion was carried out in solution with 25 l porcine trypsin (20 g/ml, sequencing grade; Promega UK, Southampton, UK) by using the NPC protocol published by PRIME-XS (http://www.primexs.eu/protocols/Public-Documents/04---Protocols/PRIME-XS-Protocol-NPC-In-Solution-Digestion.pdf/). Peptides were desalted on ZipTip -C 18  1% formic acid in UHQ water at 300 nl/min. The entire LC program was 69 min long, and MS/MS data were acquired over 5 to 65 min. The Q Exactive Plus was operated in positive polarity using a "top 10" full MS/data-dependent MS/MS method. The full MS scan parameters were a resolution of 70,000, an AGC target of 3e6, a maximum injection time of 50 ms, and a scan range of 375 to 1,750 m/z. The data-dependent-MS/MS parameters were an underfill ratio of 4%; a charge exclusion of "unassigned," 1, 6 to 8, and Ͼ8; a peptide match of "preferred"; exclude isotopes set to "on"; and a dynamic exclusion of 40 s. The raw MS data files were processed using Proteome Discoverer v1.4 (Thermo Fisher Scientific) and a workflow incorporating Mascot Server v1.3 (Matrix Science, Inc., London, UK). The raw MS data files were processed using Proteome Discoverer v1.4 (Thermo Fisher Scientific) and a workflow incorporating Mascot Server v2.5 (Matrix Science, Inc., London, UK). The Mascot database was installed from the file ϽC_albicans_SC5314_A22_current_orf_trans_all.fastaϾ downloaded from the Candida Genome Database (http://www.candidagenome.org; date stamp, 19 February 2016). The Mascot database search parameters were as follows: a peptide tolerance of 10 ppm, a fragment ion tolerance of 20 milli-mass units, maximum missed cleavages of 1, and the dynamic modifications carbamidomethyl (C), oxidation (M), oxidation (C), dioxidation (C), trioxidation (C), and nitrosyl (C).
Structural modeling of Hog1. Structural models of wild-type and mutant versions of C. albicans Hog1 were calculated using the structural predictions tool Phyre2 Web portal (63). Appropriate sequence files, including the relevant mutated residues, were input, and an intensive search was carried out using all other default parameters. In each case, the highest-ranking model, based on confidence and percentage identity as calculated by Phyre2, was then used to assess overall structural features. Models were superimposed and analyzed in PyMOL (PyMOL Molecular Graphics System, version 1.8; Schrödinger, LLC). Several structures of the human homologue p38 are available in the Protein Data Bank (PDB), mostly with bound inhibitors and/or ligands. Only two apo structures are available, and the highestquality model (PDB accession number 1R39) was chosen for structural comparison (64). Intracellular ROS levels. Exponentially growing C. albicans cells were treated with 20 M dihydroethidium (DHE; Thermo Fisher Scientific) and incubated at 30°C for 1 h in darkness. Cells were washed twice with phosphate-buffered saline (PBS) and subjected to fluorescence-activated cell sorting using a BD LSR II flow cytometer (BD Biosciences, Oxford, UK). The data presented, which were analyzed using FlowJo software version 10.0.8 (FlowJo, LLC), are from one of three independent experiments, all of which showed similar effects. The emission of red fluorescence by DHE (612 nm) was dependent on the presence of cells and not mediated by DPTA-NONOate alone.
Phagocytic-killing assays. Polymorphonuclear granulocytes (PMNs) were isolated from the blood of healthy donors, obtained according to protocols approved by the College Ethics Review Board, University of Aberdeen (protocol number CERB/2012/11/676). PMNs were isolated with Hitopaque-1119 and Hitopaque-1077 (Sigma-Aldrich) according to the manufacturer's instructions. PMNs were suspended in RPMI 1640 (Sigma-Aldrich) containing 10% heat-inactivated fetal bovine serum (hiFBS) and adjusted to 10 6 cells/ml. Exponential C. albicans cells were washed three times in PBS, resuspended in 1 ml RPMI 1640, 10% hiFBS, and adjusted to 10 5 cells/ml. C. albicans cells were mixed with PMNs in a ratio of 1:10 (yeast cells to phagocytes) and incubated for 2 h at 37°C in RPMI 1640, 10% hiFBS, 5% CO 2 . The PMNs were then lysed with 0.25% SDS and treated with DNase I (Invitrogen), and C. albicans cell viability was determined via counting CFU. Data from three independent replicate experiments, each of which Hog1 Promotes Nitrosative-Stress Adaptation ® included 2 technical replicates, were analyzed in PraphPad Prism 6 using one-way ANOVA and Tukey's multiple-comparison test.
Virulence assays. The levels of virulence of C. albicans strains were compared using the Galleria mellonella model of systemic infection. Yeast cells grown in YPD medium at 30°C were harvested, washed, and resuspended PBS. Cells (10 5 ) were inoculated into the last proleg of a wax moth larva (20 larvae per C. albicans strain). Control groups of larvae received no injection or PBS alone. Larvae were incubated at 37°C, and survival was monitored for 7 days. The data presented, which were analyzed using PraphPad Prism 6 and compared using the log-rank (Mantel-Cox) test, are from one of two independent experiments, all of which showed similar effects.
Data availability. The RNA sequencing data set is available in EBI (https://www.ebi.ac.uk/ arrayexpress/) under accession number E-MTAB-5990. Other data that support the findings of this study are available from the corresponding author upon reasonable request.