Dynamic Transcriptomic and Phosphoproteomic Analysis During Cell Wall Stress in Aspergillus nidulans

The transcriptome and phosphoproteome of the filamentous fungus Aspergillus nidulans was monitored after cell wall stress. Over 1800 genes and 700 phosphorylation sites showed dynamic expression and occupancy, respectively. These data guided selection of kinase deletion strains for testing for cell-wall related phenotypes. Kinase-substrate analysis revealed possible connections between CWIS, SIN, HOG, and calcium pathways. Fungi exposed to wall stress experience an increase in septation formation, which may function as a biological defense mechanism.


In Brief
The transcriptome and phosphoproteome of the filamentous fungus Aspergillus nidulans was monitored after cell wall stress. Over 1800 genes and 700 phosphorylation sites showed dynamic expression and occupancy, respectively. These data guided selection of kinase deletion strains for testing for cellwall related phenotypes. Kinasesubstrate analysis revealed possible connections between CWIS, SIN, HOG, and calcium pathways. Fungi exposed to wall stress experience an increase in septation formation, which may function as a biological defense mechanism.

Graphical Abstract
The cellular signaling network is a complex system that controls nearly all cellular processes underlying growth, metabolism, morphogenesis, and development (1,2). Generally, cell surface receptors are perturbed by environmental stimuli and transduce information to the nucleus via signaling proteins, which typically include G-proteins, kinases, and transcription factors. A signaling pathway of interest in fungi is the cell wall integrity signaling (CWIS) pathway. The CWIS is nominally responsible for wall maintenance and repair (3), which are critical as fungi rely on the wall for protection, shape, strength, and host invasion (4 -7). The CWIS is also of interest as cell walls are present in fungi, but absent in humans, thereby making the fungal cell wall an ideal target for antifungal therapeutics (8 -10).
The CWIS is activated in response to wall stress and/or damage (3) and is composed of a conserved set of proteins across a variety of fungal species including Aspergillus nidulans (11), A. fumigatus (12), A. oryzae (13), Saccharomyces cerevisiae (3), Schizosaccharomyces pombe (14), and Candida albicans (15). In the model fungus A. nidulans, the CWIS pathway is composed of cell surface receptors which sense cell wall stress (16). These sensors activate the guanine nucleotide exchange factor (GEF), Rom2, which acts on the GTPase RhoA to trigger activation of PkcA (16 -19). This subsequently activates a MAPK cascade, which includes BckA-MkkA-MpkA (11,20,21). MpkA, the final kinase in this cascade, activates the transcription factor RlmA which is responsible for controlling expression of ␣-1,3glucan synthase genes (11).
Notably, unlike S. cerevisiae, the transcription of many cell wall related genes (including ␤-1-3-glucan and chitin synthase genes) in A. nidulans is regulated in an MpkA-independent manner (11). In an effort to characterize the CWIS pathway, several studies have sought out this alternative signaling pathway to identify how ␤-1,3-glucan and chitin synthase genes are regulated in response to cell wall perturbation (22)(23)(24)(25)(26)(27). Connections between the CWIS pathway and several other pathways have also been identified. Examples include the unfolded protein response (UPR) pathway, calcium signaling, branching regulation, iron homeostasis, the high osmolarity-glycerol (HOG) pathway, and the cyclic AMP protein kinase A (PKA) pathway (17, 28 -30). These pathway networks are complex and highly interconnected, making it difficult to resolve network topography. We are interested in developing a better understanding how cell wall integrity signaling is transduced to the nucleus and identifying downstream effectors whose expression is controlled by this process.
The most common post translational modification (PTM) mediating signal transduction is phosphorylation (1). To capture the global phosphorylation-state of the cell, mass spectrometry (MS) has emerged as a powerful tool. In the past decade, major technological advances in MS have enabled researchers to accurately identify and quantify thousands of phosphorylation sites from a single run (31). However, although a single MS run provides an abundance of information, when studying signaling behavior it is important to consider the dynamics of protein phosphorylation. Phosphorylation events typically occur rapidly (within minutes), thus to capture signaling progression, multiple samples must be taken over a short time span (32). In addition, evaluating dynamic changes in phosphorylation can aid in determining which events are more likely to be direct interactions of kinases and substrates (33).
In this work, we study the dynamic response of the cell wall integrity signaling pathway to a wall perturbation. Micafungin, a ␤-glucan synthase inhibitor, is used as a cell wall perturbant as it is known to directly activate the CWIS pathway (11). We used quantitative, label-free, mass spectrometry (over a short period of time; 10 min) to assess dynamic changes in protein phosphorylation and transcriptomic analysis (over a longer period of time; 120 min) to assess changes in global geneexpression levels. Overall, we identified over 700 dynamic phosphorylation sites, which includes 15 sites on kinases and 16 on putative transcription factors. Over 1800 significantly dynamic genes were expressed upon micafungin treatment, 25 of which have previously been identified as cell-wall related. Our findings reveal that the coordinated response of the calcium, HOG, and Septation Initiation Network (SIN) pathways are involved in the response to cell wall stress. Furthermore, our results underscore the potential of this multi-omics approach to study signaling networks in general.
Western Blotting-After ϳ20 h of growth (mid-exponential growth phase), 20 ng/ml micafungin per 1g/kg dry cell weight (DCW) was added to the control strain culture. DCW was determined by removing ϳ25 ml of culture from the shake flask during growth. Amount of culture removed was determined by weight (i.e. "wet weight" kg culture). Fungal biomass was recovered from the liquid via vacuum filtration, and then dried at 100°C before determining "dry weight." Dry cell weight (DCW) is the concentration of biomass in culture, and was calculated as "dry weight" divided by the "wet weight." Twentyfive milliliters of culture was removed after 0, 1, 3, 5, 10, 15, and 30 min of micafungin exposure and mycelia were separated from broth and immediately frozen in liquid nitrogen. Frozen mycelia were crushed with mortar and pestle into a fine powder and mixed with TNE buffer at 1:1 (v/v) (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM EDTA). Samples were spun at 10rpm at 4°C for 10 min then centrifuged at 4°C and 10,000xG for 10 min and BCA assay (Pierce, Rockford, IL) was performed on the supernatant. One hundred micrograms of protein was exposed to SDS-page electrophoresis and transferred to a nitrocellulose membrane (Life technologies, Carlsbad, CA). Western blotting preparation steps were performed as described in Chelius et al., 2019. Two biological replicates were modeled using Michaelis-Menten kinetics.
Dynamic Micafungin Treatment, Experimental Design, and Phosphoproteomic Sample Preparation-Fungi were grown until mid-exponential phase (about 20 h) and 20 ng micafungin/ml culture per 1g/kg DCW was added. About 25 ml of culture was removed from the flask while the shaker was still shaking (ensuring a homogeneous sample) at 0, 30s, 1, 1.5, 2 ,2.5, 3, 3.5, 4, 4.5, 5, 7.5, and 10 min after micafungin exposure. Immediately after removing sample, the fungal mass separated from broth, frozen in liquid nitrogen, and stored at Ϫ80°C. This experiment was run with two biological replicates.
Samples harvested from shake flask culture were prepared for phosphoproteomic analysis following methods of Chelius et al., 2019. Briefly, frozen biomass was crushed by mortar and pestle with liquid nitrogen, mixed with 1:1 v/v TNE buffer, and the BCA assay (Pierce, Rockford, IL) was used to determine protein concentration. The Filter aided sample preparation (FASP) was used to digest protein (29) and then 500 g protein was incubated with Trypsin Gold (Promega, Madison, WI) (50 g protein: 1 g trypsin) overnight in a 37°C water bath. The samples were lyophilized to dryness and phosphopeptides were collected using the Pierce High-Select TiO 2 Phosphopeptide Enrichment Kit (Pierce, Rockford, IL) protocol. Samples were dried to completeness and stored at Ϫ80°C until mass spectrometry analysis.
LC-MS/MS and LFQ Analysis-Stored samples were reconstituted in 20 l of acetonitrile/water/formic acid 5/95/0.5 (v/v/v) and run in duplicate on a Dionex Ultimate 3000 nanoLC system coupled to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, San Jose, CA). Separation of peptides was performed on EASY spray C18 75 m ϫ 50 cm column for a 190 min gradient at flow rate of 200 nL/min using mobile phase A of 0.1% formic acid in water and mobile phase B of acetonitrile/water/formic acid 80/20/0.1 (v/v/v). Mass spectrometry data were collected in positive ionization mode using a data dependent acquisition method with a full MS scan for m/z range 350 -1500 in orbitrap at 120 K resolution. Consecutive MS/MS scans were performed in the ion trap by top-speed decision selection with a dynamic exclusion of 20 s. Precursor ions selected from the first MS scan were isolated with an isolation width of 1.6 m/z for collision induced dissociation (CID) energy and normalized collision energy (NCE) set to 30.
Samples from both biological replicates and two technical replicates for all timepoints (52 samples) were processed simultaneously with MaxQuant software version 1.5.3.17 (34). Database search was completed using Andromeda search engine (which is a part of Max-Quant) after quantification. MaxQuant analysis was performed as described in Chelius et al., 2019 to quantify peptide abundance. MaxQuant automatically quantified phospho-peptides using relative quantification of ion intensities which compares the abundance of the same peptide species across all runs. Parameters that were used include: fixed modification of carbamidomethyl (C), variable modifications of phosphorylation (STY), acetylation (N terminus), and oxidation (M), trypsin as the protease, and 2 missed cleavages. The mass tolerance for precursor ions was set 20 ppm in the first search and 4.5 ppm in the main search, and the fragment mass tolerance was set to 20 ppm. The FDR was set to 0.01. MS/MS spectra were searched against a target database consisting of 10,555 entries from the UniprotKB database (UP000000560). Combining a high number of timepoints (13) with several replicates (2 biological replicates with 2 technical replicates) ensures a sample size large enough to provide statistical power. Peak lists from all MS/MS spectra were extracted from MaxQuant result files using MS-viewer (35) and can be accessed at http://msviewer.ucsf.edu/prospector/cgi-bin/msform.cgi?formϭ msviewer (search key: yhspxctyvc).
Experimental Design and Statistical Rationale-Phosphoproteomic analysis was completed on two biological replicates each containing 13 timepoints. All samples were run in duplicate for a total of 52 mass spectrometry runs. The control is the first time point (0 min) which was taken immediately before micafungin addition (i.e. cell perturbation). To determine if the phosphorylation level is significantly dynamic, multivariate adaptive regression splines (MARS) model (36) was used following a method developed by Oliveira et al., 2018. This method ensures that our dynamic response is not background noise.
Phospho-site Filtering and Statistical Assessment-Perseus 1.6.1.1 software was used to filter phosphorylation sites (37). Sites were removed if there was a contaminant, reverse match, or the location probability was less than 0.75. Next, we applied filtering rules to the biological replicates separately. A phosphorylation must be present in both technical replicates of at least 10 of the 13 time points and present in either the initial time point (0 min) or the 30 s time point (our control) (33). Finally, these rules applied to both biological replicates resulted in a total of 925 highly confident phosphorylation sites. For the remainder of the analysis the intensity values were transformed into log 2-fold changes. A MARS model was fit to the fold changes from 26 (13 time points, 2 technical replicates) samples using ARE-SLab, a Matlab package (38). The arespredict function (with default settings) determined the quality of fit (mean square error; MSE) of the actual data and 10,000 random permutations of the fold change levels. The p value was calculated as the fraction of randomly permutated models with better fit than the actual model. If the p value was less than 0.05 the phosphorylation sight was deemed significantly responsive (Class I sites are significant in both biological replicates, Class II are significant in 1 biological replicate).
RNA Sequencing and Statistical Analysis-Fungi were grown and treated with micafungin identically to phosphoproteomic samples. Samples were taken at 0, 5,10,15,20,25,30,40,50,60,75,90, and 120 min for three biological replicates. Harvested mycelia were frozen with liquid nitrogen and ground into a fine powder by mortar and pestle. RNA extraction and preparation was completed as previously described (39). The samples were sent to the University of Nebraska Medical Center's Bioinformatics and Systems Biology Core for library preparation and RNA Sequencing (https://unmc.edu/bsbc). The raw samples were of single end reads of high quality in 4 separate lanes in FASTQ files. The following costume RNA-Sequencing pipeline was used: HISAT2 2.1.0 (40), HTSeq-Count 0.9.1(41), DESeq2 1.20.0 (42). The annotation file (Gene Transfer File) and whole genome of wildtype strain FGSC A4 FNA file was supplemented from the database Ensemble Fungi (43).
Like the phosphoproteomic data, genes were required to have reads in 2 out of 3 replicates in all 13 timepoints. The normalized reads were transformed into log 2-fold change values. Finally, the dynamic test (MARS model) described above was applied to all replicates. Significance threshold was set to p Ͻ 0.01.
Predictive Software -PHOSIDA, NetworKIN, STRING, GO and Cluster Analysis-Several software tools were used to further analyze the phosphorylation hits. PHOSIDA analyses all phosphorylation motifs to identify statistically overrepresented motifs (http://141.61. 102.18/phosida/index.aspx) (44,45). PHOSIDA parameters were set to the following: minimum score at 10 and minimum proportion of matching sites was 5% (46). Dynamically phosphorylated sites were analyzed by NetworKIN to predict which kinases target specific phosphorylation sites from the yeast proteome (http://networkin.info/) and phosphosite-kinase interactions with a score greater than 2 are considered significant (47). STRING database (http://string-db.org) was used to build a protein-protein interaction web based on both computational predictions and experimental observations (48). All Class I and Class II phospho-proteins were imported into the database and the interaction sources selected include text mining, experiments, databases, co-expression, and co-occurrences. The minimum required interaction score was 0.700 (or high confidence). The Aspergillus genome database (AspGD) Gene Ontology (GO) Term Finder tool (http://www.aspergillusgenome.org) was used to identify enriched GO terms from imported proteins (49). Here a significance threshold of p value Ͻ 0.01 was used. A heat map of significantly expressed genes was created using Perseus v 1.6.5.0 (http://www. perseus-framework.org). Average Log2Fold Change is depicted, the micafungin exposure time order was preserved, and the distance metric was set to Euclidean.
Apparent Cell Wall Strength-To measure apparent cell wall strength in shake flask culture mycelia were removed from shake flasks (ϳ5 ml) during exponential growth phase and subjected to particle size analysis using a Mastersizer 3000 instrument with a Hydro SM manual sample dispersion unit following published methods (29,50). The average mycelial size (S 90 ) for each biological replicate was calculated from at least 3 timepoints (between 17 and 24 h growth) using duplicate technical replicates (p value set at 0.05, nՆ3). In parallel with size analysis, dry cell weight measurements were taken to determine the specific growth rate of these strains.
Fungal Growth in Micafungin-To test whether these kinases may respond to cell wall perturbation, we grew kinase deletion mutants both in submerged culture and agar plates with micafungin. For plate growth experiments, spores were harvested from fresh fungal lawns in 10 ml of sterile diH 2 O and the spore concentration was determined. The spore solution was then diluted to 1 spore/l and 100 l of spore solution was spread onto both MAGV and MAGVϩmicafungin plates with 3 replicates. Plates were grown at 28°C and the number of developed colonies were counted after 3 days. The critical concentration of micafungin (i.e. highest concentration of micafungin with no visible effect, 0.007 g/ml) was determined by characterizing the susceptibility of the control strain, A1405 at varying concentrations of micafungin (0.001, 0.005, 0.007, 0.01, 0.015, 0.05, 0.1 g/ml). All deletion mutants were spread on MAGVϩmicafungin plates (with 0.007 g micafungin/ml) and the average number of developed colonies on MAGVϩmicafungin were normalized to the average number of developed colonies on MAGV plates. The significance was determined using a t test to compare the MAGV and MAGVϩmicafungin colony growth for each strain, p Ͻ 0.05, n ϭ 3.
Next, we determined the susceptibility of fungi to micafungin during shake flask culture. Both the control strain and kinase deletion mutants were grown in YGV media to at least 1g/kg DCW (to ensure ample amount of biomass for measurements), whereupon micafungin was added (20 ng/ml micafungin per 1g/kg DCW, same as -omic experiments). Now, DCW measurements were taken over the next 6 h (usually every 1.5-2 h) to determine fungal growth rate in micafungin. Three biological replicates were used for each strain, and samples were removed after micafungin treatment for qPCR analysis, described below.
Quantitative RT-PCR-Samples were removed from the submerged culture with micafungin experiment at 0, 60, 90, and 120 min of micafungin exposure. Mycelia were separated from broth and frozen in liquid nitrogen and stored at Ϫ80°C for up to 1 week. RNA extraction, purification, and cDNA conversion was completed as described in Chelius et al., 2019. Primers were designed using Primer3 (supplemental File S8) (51). Three technical replicates and two biological replicates of each time point were run using a BioRad C1000 Touch Thermal Cycler with CFX96Real-Time System. The target genes were quantified with respect to the reference gene, histone (H2B), and fold change was determined with respect to the zero time point following the ⌬⌬Ct method (52).
Septation Analysis by Coverslip Experiment-Coverslips with spores were prepared as described in Chelius et al., 2019. After 12 h of growth in YGV media at 28°C the first coverslip was removed from media. One drop of calcofluor white stain (18909, Sigma, Darmstadt, Germany) was added before mounting to a slide. The slide was viewed under an Olympus IX-81 inverted fluorescence microscope (Olympus, Tokyo, Japan) and 30 images were captured, for each of three biological replicates (90 images total). Each image contained at least one mycelium. Following this initial assessment, micafungin was added to media at a final concentration of 10 ng/ml. Each hour thereafter, a coverslip was removed, and images were taken until coverslips were overgrown, ϳ10 h later. The images are analyzed using Image J software (53) for their projected area and the number of branches and number of septa are counted for each mycelium. The growth and branching rates are calculated as described in Chelius et al., 2019 and significance was calculated based on three biological replicates (student's t test, p Ͻ 0.05).

RESULTS
Dynamic Time-scale-Eukaryotic signal transduction typically occurs over a relatively short time-scale (54). Studies in the yeast S. cerevisiae indicate phosphorylation events involved in signal transduction occur over a range from 5 s to 80 min, depending on the induced pathway (33,55,56). To determine an appropriate sampling time-scale for phosphoproteomic analysis in our study, we grew A. nidulans in shakeflask culture and added micafungin (␤-glucan synthase inhibitor) to induce the CWIS pathway. Total and phosphorylated MpkA were determined by Western blotting (Fig. 1, supplemental Fig. S1) and modeled using a simple saturation kinetic model (e.g. Michaelis-Menten kinetics). The half-saturation constant, k t , was determined to be 2 min, meaning the majority of the dynamic MpkA phosphorylation response occurs in the first 5 min after perturbation with micafungin. We thus chose to draw samples for phosphoproteomic analysis every 30s for the first 5 min, and then at 7.5 and 10 min (13 samples total). Changes in gene expression take a significantly longer period as shown by previously published qPCR data (11) involving micafungin induced expression changes. Based on these previous findings, we drew samples for transcriptomic analysis at 0, 5,10,15,20,25,30,40,50,60,75,90, and 120 min (13 samples total).
Statistically Dynamic Phosphorylation Sites-Because of recent advancements in mass spectrometry technologies, there has been a shift from static -omic studies to dynamic studies (33,(55)(56)(57)(58)(59)(60)(61)(62)(63). However, comparison of these dynamic studies reveals significant differences in experimental design as well as data processing and analysis, such that no clear best-practice currently exists. For example, although some studies use many time points with little to no replication (33), others utilize relatively few time points, but include replicates (64). In an attempt to combine these approaches, we designed our study to include many time points (13) with multiple replicates (2 technical and 2 biological) and used labelfree quantification (peak intensity) to survey phosphosite occupancy.
Although our technical replicates show good correlation, there is some discrepancy between biological replicates (Fig.  2). This occurs because biological replicates are for samples taken only 30 s apart, such that even small differences in behavior shift values between time points. Previous studies have not manifested this behavior, as they have used multiple biological replicates over fewer time points, typically over much longer time scales (e.g. hours) (60 -62).
To address this issue, we developed a data-analysis pipeline to identify, with high confidence, phosposites that were significantly dynamic (i.e. show a significant change in occupancy after micafungin addition (Fig. 3)). We note that in the 52 different mass spectrometry runs carried out for this study, a total of 5106 phosphorylation sites were observed. However, many of these sites appeared in relatively few runs. When stringent confidence tests are applied (Materials and Methods), 1156 and 1056 phosphosites were identified in biological replicates 1 and 2, respectively, with 925 of these phosphosites (i.e. between 80 and 90%) in common between the two replicates (supplemental File S1).
To determine whether these 925 phosphosites were significantly changing in response to micafungin perturbation, Log2 fold change of intensity was modeled with a multivariate adaptive regression splines (MARS) model (38,56). The quality of fit (mean square error, MSE) was used to determine if individual phosphosites were significantly changing occupancy with time (56). Fig. 4 shows MARS model fits for 12 phosphosites, all on protein kinases. Although there is some discrepancy between biological replicates, it is clear that nearly all the MARS trends (i.e. lines) for biological replicates are: (1) in the same direction and (2) show similar magnitude. Thus, this approach has allowed us to overcome the challenge related to comparing biological replicates taken during periods of rapid sampling.
Using this method, we narrowed the 925 P-sites, to 431 dynamically significant phosphorylation sites in both biological replicates (Class I) and 363 in one replicate (Class II; i.e., over 700 total). From the 431 Class I sites, 51 motifs were significantly overrepresented ( Fig. 5B and 5C, supplemental File S2). Further motif analysis was conducted using Networ-KIN (47). Given a protein sequence, NetworKIN predicts which kinase (based on the yeast kinome) is likely to have carried out this phosphorylation. From the 431 phospho-sites, Networ-KIN predicted with significant confidence that 28 different kinases phosphorylated 173 of these sites (45,65). Of note, NetworKIN predicted that the prominent CWIS pathway kinases PkcA and MpkA phosphorylated 22 and 5 phosphosites respectively, revealing putative, new signaling connections to the CWIS pathway (Fig. 5A).
RNA-sequencing Reveals CWIS Downstream Effectors-RNA-sequencing was carried out using shake flask cultures grown identically to the phosphoproteomic set, however FIG. 3. Sampling and processing pipeline. Fungi were exposed to micafungin and 13 samples for phosphoproteomics (0 -10m) and transcriptomics (0 -120m) were harvested. Three biological replicates (BRs) were used in RNA-sequencing whereas 2 were used in mass spectrometry analysis with 2 technical replicates (TRs). Overall 431 (Class I) phosphosites were found to be statistically dynamic whereas 1810 genes were dynamically expressed. the sampling time ranged from 0 to 120 min. From the 10,687 putative ORFs in A. nidulans (49), transcripts were identified from 9344 and of these, transcripts were present in all 13 time points and 2 of 3 replicates for 8,188 genes (supplemental File S3). From the 8188 genes, 1810 were dynamically significant (p Ͻ 0.01) using the MARS method (Fig. 2).  Table I) in their CWIS study. Expression profiles of these genes can be found in supplemental File S4. Of these, strict rules were used to parse out genes that were the most overexpressed (50 genes) and most under expressed (4 genes). Of the 54 most over and under expressed, 45 of these are uncharacterized genes (supplemental File S5). These uncharacterized genes may have yet unknown roles in cell wall maintenance and repair.
A heatmap of dynamically expressed genes reveals 6 gene expression clusters (Fig. 7). Gene ontology (GO) analysis of each cluster can be found in supplemental File S6. All dynamically expressed genes (1810) were searched for their GO biological process and only 805 had an associated GO term. The top 5 GO terms were transport, regulation of biological process, response to chemical, response to stress, and organelle organization (supplemental File S7). A GO term enrichment assessment revealed that the most significantly enriched group was cytoskeleton organization.
Cell Wall Strength of Kinase Deletion Mutants-In our phosphoproteomic data we observed 19 kinases and 3 phospha-tases to be significantly and dynamically phosphorylated (Table II). From this group, we tested each non-essential kinase for its involvement in cell wall maintenance and repair by growing kinase deletion mutants in shake-flask culture and measuring mycelial size-distribution using laser particle-size analysis. We developed this approach previously (50) and have shown (that for similar growth rates) smaller mycelia can result from either (1) compact (i.e. highly branched) morphology, or (2) hyphal fragmentation during shaking. If compact morphology (deduced via microscopy) is absent, smaller mycelia are a result of hyphae breaking in the flask, suggesting fungal strains with smaller mycelia (when compared with a control strain) have weaker cell walls. We used this approach previously to show an A. nidulans, mpkA deletion mutant has similar growth rate and morphology to a control strain but is approximately four times smaller implying it has significantly weaker cell walls (29). We used an identical approach here, measuring growth rate and average cell size for 12 different kinase deletion mutants (Fig. 8). Compared with the control, the growth rates of all mutants were similar (p Ͼ 0.05). However, 8 of the 12 kinase deletion mutants had significantly smaller mycelia than the control strain (p Ͻ 0.05). This implies  Susceptibility of Deletion Mutants to Micafungin-Following shake flask experiments, the same 12 kinase deletion mutants were grown on agar plates with added micafungin to test their susceptibility to a cell wall perturbant. As a member of the CWIS pathway, MpkA plays a role in regulating cell wall biosynthesis gene expression and the ⌬mpkA strain is known to be highly susceptible to micafungin (11). To test the relative impact of micafungin on the other deletion mutants we used a highly quantifiable plate assay to determine the amount of colonial growth on micafungin agar plates. Compared with traditional spot assays, this method enables us to accurately measure small differences between strains and use statistics to confidently report susceptibility. We found that ⌬mpkA, ⌬prkA, ⌬yak1, and ⌬nrc2 mutants had a significantly lower number of colonies as compared with the control (Fig. 9,   Table II). Work here suggests PrkA Yak1, and Nrc2 may be involved in regulating wall-repair related genes. Gene Expression of Mutants Exposed to Micafungin-Four kinase deletion mutants which showed reduced cell wall strength ( Fig. 8; ⌬hk-8 -4, ⌬stk19, ⌬prkA, ⌬mpkA) were grown in shake flasks in the presence and absence of micafungin. Micafungin was added at the same concentration as in the -omic experiments presented above. The growth rate before and after micafungin addition was measured, and ⌬mpkA, ⌬prkA, and ⌬stk19 all showed significantly slower growth in micafungin compared with the control (Fig. 10). For the ⌬mpkA mutant, growth rate in micafungin appears to be negative. This is likely because of a reduction in the culture dry cell weight as this strain has relatively weak walls (29) and mycelia fragment over the extended period in the flask. The other kinases (⌬hk-8-4, ⌬stk19, and ⌬prkA) have a non-negative growth rate, but grow more slowly than the control suggesting these gene products play a role in mediating cell wall repair, but are not critical to fungal survival in this environment.
To further assess the impact of these kinases (PrkA, Stk19, and Hk-8 -4) on CWIS signaling and wall repair, and at the same time explore their impact on as yet unidentified CWIS effectors, we used quantitative reverse transcription PCR (qRT-PCR). During culture of the kinase deletion mutants we measured the expression of four "fingerprint" genes (agnE, brlA, AN3339, and AN2116) during growth of the kinase deletions in the presence of micafungin (Fig. 11, supplemental File S8). We selected agnE, brlA, AN3339, and AN2116 as "fingerprint" genes as they were among the 50 most overexpressed genes in our transcriptomic data (AgnE (alpha-1,3-glucanase), BrlA (zinc finger transcription factor), AN3339 (uncharacterized), and AN2116 (uncharacterized)). Each of these genes showed a significant increase in expression during growth of the control stain in micafungin. For each gene, its expression reached a maximum at 60 min and then decreased. We note two phenomena regarding expression in the kinase mutants. First, agnE appears to be regulated in an MpkA-independent manner (i.e. ⌬mpkA pro- Last three column contain phenotype data, MF is micafungin. NA: not applicable because experiment was not run on that strain (usually because kinase is essential), Ϫ: statistically significantly lower value than the control strain, ϩ: either the same as or slightly higher than the control strain. file looks same as control). In contrast, it appears expression of brlA, AN3339 and AN2116 is mediated by MpkA, as they all show relatively flat profiles in the MpkA deletion mutant. Second, in the other kinase deletion strains (⌬hk-8 -4, ⌬stk19, and ⌬prkA) gene expression levels take longer to reach a maximum (90 -120 min). This suggests the kinases Hk-8 -4, Stk19, and PrkA are involved in the temporal regulation of wall-related genes, but they are not required for their de facto expression.
Phosphorylation of GAPs, GEFs, and Transcription Factors-Phosphoproteomic analysis also shows thirteen putative transcription factors, with 16 phosphorylation sites, are significantly changing in response to micafungin addition (Table III). Those with known function in A. nidulans include HapC, Hsf1, RtfA, SrrA, and StuA. StuA regulates conidiophore development (66), SrrA is a stress response regulator with ties to cell wall integrity signaling (67,68), RtfA is associated with secondary metabolism, morphological development (69), and minor effects on cell wall integrity in A. fumigatus (70), Hsf1 appears to regulate oxidative stress response (71), and HapC is a member of the AnCP/AnCF CCAATbinding complex (72).
From our phosphoproteomic data set we identified 11 phosho-sites belonging to 8 proteins with putative GTPase acti-vating protein (GAP) or guanine nucleotide exchange factor (GEF) functionality. Of these, 2 GEFs, HypB and RicA, have been verified, whereas all other hits are based on homology with S. cerevisiae (Table III) RicA has known connections with growth and development signaling (73) whereas HypB is likely to act in COPI-mediated vesicle formation (74).
CWIS Role in Septation-To test the role of cell wall perturbation on septation the control strain was grown on coverslips in both the presence (10 ng/ml) and absence of micafungin and stained with calcofluor white (Fig. 12) (additional images can be seen in supplemental Fig. S2). The growth rate decreases in the presence of micafungin, but the branching rate does not. This implies that branching is not a cellular process connected with cell wall stress signaling. Because septation is generally regarded as being controlled by hyphal size together with mitosis (75), when we look at the number of septa per area it is much greater for fungi grown in micafungin (Fig. 12D). Even though the fungi are growing slowly (compared with the control), they are undergoing septation more frequently per area. This demonstrates the role of septation in cell wall response to micafungin. Phosphorylation Network-To deduce possible proteinprotein interactions, we used STRING v11 to generate an interaction map (supplemental Fig. S3, Supplemental File. S9) showing connections between differentially phosphorylated

FIG. 8. Apparent cell wall strength of kinase deletion mutants.
Growth rates of strains in rich medium were calculated from the exponential growth phase using dry cell weight. No significant differences between the control and mutants are observed (A). Particle size analysis of shake flask samples over the growth period reveal that kinase deletion mutants, ⌬mpkA, ⌬npkA, ⌬hk-8 -4, ⌬srrB, ⌬nrc2, ⌬stk19, ⌬kin1, ⌬prkA, are significantly smaller than the control (B). A student's t test of mutants against the control was used with p Ͻ 0.05 for significance and nՆ3.
FIG. 9. Micafungin susceptibility of kinase deletion mutants. 100 fresh spores of the control strain were plated on MAGV plates with various concentrations of micafungin (0 -0.1 g/ml) in order to find the minimum inhibitory concentration (0.007 g/ml) (A). 100 spores of each kinase deletion mutant were grown on MAGV plates with (0.007 g/ml) and without micafungin and the number of colonies was counted. The percent growth is displayed and ⌬mpkA, ⌬yak1, ⌬nrc2, and ⌬prkA grew significantly less than the control strain (* p Ͻ 0.05, n ϭ 3) (B).
proteins (48). We see three clusters form including ribosome biogenesis, calcium and cell cycle signaling, and septation and actin regulation. Of these, the septation and actin regulation cluster included several kinases (PrkA and SidB) and phosphatases (PtcA and PtcB). Proteins with altered phosphorylation related to septation initiation and formation are shown in Table IV. DISCUSSION Our dynamic, multi-omic analysis provides insight into the signaling mechanisms and the downstream effectors involved with addressing a cell wall perturbation. Our data confidently identify statistically significant phosphorylation site occupancy changes and altered gene expression trends. We identify three main signaling pathways (septation and actin re- 10. Shake flask culture growth with and without micafungin addition. Four deletion mutant strains (⌬mpkA, ⌬hk-8 -4, ⌬stk19, and ⌬prkA) were grown up in YGV and after sufficient biomass was attained (1g/kg or greater), micafungin was added to culture at 20 ng/ml per 1g/kg DCW. During both growth phases (Ϯ micafungin) the growth rate was calculated using an exponential regression (A). Three biological replicates of each strain were grown and ⌬mpkA, ⌬stk19, ⌬prkA were found to have a significantly lower growth rate when compared with the control (*p Ͻ 0.05, ϩp Ͻ 0.1). There is no difference in growth rates without micafungin (B).
FIG. 11. qPCR analysis of most overexpressed genes. The control strain and four kinase deletion mutants were grown in the presence of micafungin. Samples were harvested at 0, 60, 90, and 120 min of exposure (dark to light bars) and RNA was extracted. These samples were probed for the expression of four genes, agnE (alpha-1,3-glacanase) (A), brlA (a C2H2 zinc finger transcription factor) (B), AN3339 (uncharacterized) (C), and AN2116 (uncharacterized) (D). The target genes were normalized with the reference gene histone, h2B, and the fold change was determined using zero time point following the ⌬⌬Ct method (Livak et al., 2001). When compared with the control, we see that ⌬mpkA does not alter its expression of brlA, AN3339, or AN2116 suggesting MpkA signaling is involved in the expression of these genes. The 3 other mutants do alter their expression, but it appears that their expression levels change around 90 min, while the control strain responds within 60 min. * indicate significant difference from the zero time point (p Ͻ .05, n ϭ 2). lated, calcineurin and calcium signaling, and the HOG pathway) with multiple points of crosstalk between the conserved CWIS (Fig. 13) as well as other, novel protein kinases and their downstream effectors.
Septation and Actin-related Functions Stimulated by Cell Wall Stress-In the filamentous fungus A. nidulans, septum formation is an actin-dependent process that results in localized deposition of new cell wall that partitions hyphae into individual cells. Septation necessitates the formation of contractile actin rings at locations determined by the distribution of nuclei (75). Ring formation is coupled to the completion of mitosis (75,76), and is also coordinated with growth to maintain appropriate ratios of cytoplasmic volume per nucleus (76). Our results show that septation increases in response to cell wall perturbation caused by micafungin, and our data imply potential mechanisms by which this could occur (Table  IV). For example, exposure to micafungin results in reduced phosphorylation of the cyclin-dependent kinase (CDK) NimX at residue Y15, which is known to promote septum formation in A. nidulans (77). In addition, we also observed altered phosphorylation of the serine/threonine kinase SidB, which is a component of the septation initiation network (SIN) required for septation in A. nidulans (78,79). Last, AN4693 is a homologue of S. cerevisiae Hof1 and S. pombe cdc15, both of which are key regulators of septation. Our results demon-strate that AN4693 is down-phosphorylated at multiple sites upon exposure to micafungin. In S. pombe, extensive dephosphorylation of Cdc15 enables interaction with multiple partner proteins that promote septum formation (80), including the SepA homologue Cdc12. Taken together, these observations suggest CWIS activation functions though the SIN and AN4693 to trigger septation. We propose that this is part of a stress response that maintains the structural integrity of damaged hyphae by confining the damage to specific regions of the mycelium.
In A. nidulans, polarized hyphal growth requires actin cables and actin patches; the former mediate localized exocytic delivery of vesicles at the immediate hyphal tip, whereas the latter are required for endocytosis at sub-apical sites (81)(82)(83). We identified several proteins implicated in actin patch formation that are dynamically phosphorylated because of CWIS activation (Table IV). These include regulators of the Arp2/3 complex such as PrkA, Sla1, Las17, Pan1, and Crn1. Because Arp2/3 catalyzes the formation of the branched actin filaments that form patches, these results suggest that activation of the CWIS pathway alters actin formation with subsequent effects on morphogenesis. For example, disassembly of actin patches could conceivably free up actin monomers that could be diverted toward formation of contractile actin rings to support increased septation (84). Similarly, our observation that the actin filament severing protein AN0837 is dynamically phosphorylated implies another mechanism for increasing the actin monomer pool. In addition, altered actin patch formation should impact endocytosis with resulting effects on hyphal extension. Indeed, SlaB, which localizes to actin patches and is required for endocytosis, also shows dynamic phosphorylation in response to CWIS activation.
Moreover, we observed multi-branched hyphae during micafungin exposure (Fig. 12) that most likely occurred because the extension rate of the tip slowed. The combined effect of cell wall damage and disorganization of the actin cytoskeleton may account for this. For example, the loss of actin filaments could disrupt the tip growth apparatus (85) and therefore trigger increase apical branching. We also know that a mutant (sepA) in which actin filament formation is severely reduced also displays apical branching (86).
A gene ontology (GO) analysis of the 1810 genes dynamically expressed in response to micafungin shows 7 of the top 10 GO terms are related to the cytoskeleton and actin organization (all categories p Ͻ 0.025). These include: "cytoskeleton organization," "actin-filament based process," "regulation of actin filament length," and "regulation of actin polymerization," for example (supplemental File S7). Of the 43 putative "cytoskeleton organization" genes, 18 of these have been verified (Table V). The genes budA, fimA, and actA all have implications in septum formation as its presence is transient at septation sites (82,87). Both ampA and fimA are involved in endocytosis (82,88). The Ras GTPase, gapA, has cortical localization and is believed to be involved in Ras  12. Growth in micafungin leads to changes in septation. Control strain was grown in rich medium for 12 h on coverslips. Micafungin was added to media (10 ng/ml) and images were taken every hour thereafter. Fungi grown in micafungin show more septa and branches (A) than fungi grown in rich medium (B). Both images were taken at 17 h total growth, white arrows point to septa. Images from this experiment were used to calculate the growth rates and branching rates for 3 biological replicates with and without micafungin (* p Ͻ 0.05) (C). When you plot septation per area, it is clear that fungi in micafungin continue to septate regardless of fungal size (D). signaling that polarizes the actin cytoskeleton (89). Microtubule (MT) dynamics and organization genes include migA, alpA, and clipA (90 -92). Calcium and Calcineurin Signaling Aids CWIS Pathway-In our data, calmodulin (CaM) and its associated regulatory protein CmkC were dynamically phosphorylated in response to micafungin. CaM reversibly binds Ca 2ϩ and is required for polarized growth and septation in A. nidulans (93,94). Transient CaM localization occurs at sites of hyphal growth, branch emergence, and septation, which implies that it performs a critical function in localized cell wall deposition (94). CaM has also been implicated in the activation of the CDK NimX via CmkB, which is phosphorylated and activated by CmkC (Fig. 13) (95). This provides another mechanism by which CaM could impact morphogenesis in response to cell wall stress.
Key effectors of calcium signaling that regulate cell growth and cell wall integrity in A. nidulans are the CaM-dependent phosphatase calcineurin (CnaA) and the transcription factor CrzA (96,97). Among proteins with roles related to calcium signaling, the genes encoding CnaA and CrzA were differen-tially expressed in response to micafungin (both increase expression for the first 20 min and then decrease). Accordingly, modulation of calcium signaling is likely to be an important output of CWIS activation that has broad effects on growth and morphogenesis. This is consistent with the observation that inhibition of calcineurin function exacerbates the effects of caspofungin exposure (98), and that CrzA regulates the expression of chitin synthases that are thought to reinforce damaged cell walls (99).
High-osmolarity Glycerol (HOG) Pathway Crosstalk-Several putative high-osmolarity glycerol (HOG) signaling proteins showed changing phosphorylation levels in response to micafungin. For example, SrrA underwent dynamic changes in phosphorylation, which is notable given that it has been previously implicated in cell wall assembly because srrA deletion mutants show significant resistance to micafungin (68). In addition, PtcB, which is yet uncharacterized in A. nidulans, was down-phosphorylated in the presence of micafungin and is a predicted target of HogA. In A. fumigatus, a ⌬ptcB strain is more susceptible to cell wall perturbing agents, has increased levels of chitin and ␤-1.3-glucan, and has higher MpkA phosphorylation levels (100). Additional connections between components of the HOG and CWIS signaling pathways have been described in both S. cerevisiae and S. pombe (Fig. 13) (101-104), thereby reinforcing the notion that pathway cross-talk might be an integral feature of the response to cell wall perturbation. In A. nidulans, Zhou et al. recently reported that hyperosmotic stress could trigger septation in a HogA-and SIN-dependent manner (105). This raises the intriguing possibility that a general function of stress-induced MAP kinase signaling in filamentous fungi might be activation of septum formation.
Kinases with Cell Wall Abnormalities-Eight kinase deletion strains, ⌬hk-8 -4, ⌬stk19, ⌬prkA, ⌬kin1, ⌬nrc2, ⌬npkA, ⌬yak1, and ⌬srrB exhibited phenotypes consistent with cell wall integrity signaling activity (Table II). These kinases are not members of the conserved CWIS BckA-MkkA-MpkA MAPK module (11) of which only the MAPKKK, BckA (S719, S1003) and MAPKK, MkkA (S169) were identified from phosphoproteomics. Both PrkA and Kin1 have been discussed above in regards to their roles in Arp2/3 regulation. As many of these kinases have not been characterized in A. nidulans, here we have identified their putative function in the CWIS network for the first time by combining phenotype and phosphoproteomic data.
Hk-8 -4 is a histidine kinase (HK), had weaker cell walls than the control strain (Fig. 8B), was phosphorylated at S314, apparently by Ste20 (based on NetworKIN analysis) which has been implicated in the SHO1 branch of the HOG pathway (106), and its deletion mutant had altered putative cell wall regulated gene expression trends from the control (Fig. 11).
The nrc2 deletion mutant was more susceptible to micafungin on solid media, had weaker cell walls, and was predicted to be auto phosphorylated at S433. Nrc2 is uncharacterized in A. nidulans, however its yeast homologue, Fpk1, is known to be involved in cell wall integrity signaling as it contributes (directly or indirectly) to Mpk1 (MpkA homologue) activation (110). NpkA (S74) is predicted to be phosphorylated by Snf1 and its deletion mutant (⌬npkA) had a weaker cell wall (this work) and showed aberrant septa formation (111).
The yak1 mutant showed a significant decrease in colony growth in the presence of micafungin (Fig. 9B). We found Yak1 was dynamically phosphorylated at Y502, a tyrosine predicted to control full kinase activity (112). SrrB (S645) was predicted to be phosphorylated by PkaB (NetworKIN). Not only this, but ⌬srrB shows sensitivity to osmotic stress (79) and had significantly smaller mycelia in shake flask culture (this work) suggesting it plays a role in CWIS (Fig. 8B).
Putative, Currently Unknown Genes Downstream of Micafungin Perturbation-Previous work identified two genes, agsA and agsB, that are downstream of the conserved MAPK signaling cascade, BckA-MkkA-MpkA (11). However, all other cell wall related genes were not dependent on this cascade. We show many interconnections between various signaling cascades which may account for this. Moreover, many genes that are not directly associated with cell wall repair are expressed at very high level in response to micafungin. Of the 50 most overexpressed genes, 45 of these are uncharacterized (Table VI; supplemental File S5). However, using computational predictions, 14 were predicted to be localized in the membrane (GO cellular component) and associated with transport functions (GO cellular function) (AN2321, AN2598, AN2781, AN2959, AN4106, AN4129, AN4173, AN7557, AN8084, AN8342, AN8595, AN8995, AN9219, and furG). Increased expression of membrane transport genes, in response to micafungin, may facilitate movement of cell wall material through the membrane to facilitate repair of damaged walls. These hits make up less than 30% of the highly expressed genes. We hypothesize, that these remaining genes are downstream effectors of the CWIS pathway and are involved in cell wall repair. AN2116 and AN3339 were probed by qPCR during micafungin exposure. They were highly expressed in the control strain after 60 min. In the ⌬mpkA strain, the expression of these genes changed little. This suggests AN2116 and AN3339 are dependent on MpkA. AN2116 (uncharacterized) is predicted to play a role in coenzyme binding and catalytic activity (49) whereas, AN3339 (uncharacterized), is predicted to have oxidoreductase activity, transferase activity, and zinc-binding activity (49). Because of the overwhelming increase in gene expres-sion of these 50 genes, we hypothesize they are novel downstream effectors of CWIS. CONCLUSION We have applied a multi-omic methodology for identifying signaling mechanisms in response to cell stimuli. Short timescale phosphoproteomic sampling, combined with longerscale transcriptomic sampling, have provided a global picture of the cellular response to fungal cell-wall stress. Our statistical approach in determining dynamic effects resulted in the identification of 431 (794 with Class II) phosphorylation sites and 1810 genes. Cell wall strength and micafungin susceptibility assays provided a relatively rapid means for phenotyping relevant kinase deletion mutants and showed hk-8 -4, stk19, prkA, npkA, nrc2, srrB, and kin1 deletion mutants have weaker cell walls than the control. Moreover, ⌬prkA, ⌬yak1 and ⌬nrc2 are more susceptible to micafungin on solid media, whereas ⌬prkA and ⌬stk19 are susceptible to micafungin in submerged culture. We found many signaling molecules related to Arp2/3 regulation, septation initiation, calmodulin and calcium signaling, and high-osmolarity are changing phosphorylation and most likely activity because of micafungin perturbation. We hypothesize septation formation is being initiated via NimX, SidB, and CaM signaling, whereas PrkA and Kin1 are regulating activity of Arp2/3 and endocytosis. We propose that septation formation is a defense mechanism of fungi under cell wall perturbation to prevent further damage to internal hyphal compartments. Many uncharacterized genes were significantly overexpressed, suggesting novel, downstream CWIS effectors. DATA AVAILABILITY RNA-Seq data has been deposited in the Gene Expression Omnibus (GSE136562) and mass spectrometry phosphoproteomics data in the Proteomics IDEntifications (PRIDE) repository, PXD015038 (113). MS/MS spectra can be viewed using MS-viewer (search key: yhspxctyvc; http://msviewer. ucsf.edu/prospector/cgi-bin/msform.cgi?formϭmsviewer).

Conflict of interest-Authors declare no competing interests.
Abbreviations-The abbreviations used are: BCA, bicinchoninic acid; FDR, false discovery rate; MAPK, mitogen activated protein kinase.