RgsD negatively controls development, toxigenesis, stress response, and virulence in Aspergillus fumigatus

The regulator of G protein signaling (RGS) domain proteins generally attenuate heterotrimeric G protein signaling, thereby fine-tune the duration and strength of signal transduction. In this study, we characterize the functions of RgsD, one of the six RGS domain proteins present in the human pathogenic fungus Aspergillus fumigatus. The deletion (Δ) of rgsD results in enhanced asexual sporulation coupled with increased mRNA levels of key developmental activators. Moreover, ΔrgsD leads to increased spore tolerance to UV and oxidative stress, which might be associated with the enhanced expression of melanin biosynthetic genes and increased amount of melanin. Yeast two-hybrid assays reveal that RgsD can interact with the three Gα proteins GpaB, GanA, and GpaA, showing the highest interaction potential with GpaB. Importantly, the ΔrgsD mutant shows elevated expression of genes in the cAMP-dependent protein kinase A (PKA) pathway and PKA catalytic activity. The ΔrgsD mutant also display increased gliotoxin production and elevated virulence toward Galleria mellonella wax moth larvae. Transcriptomic analyses using RNA-seq reveal the expression changes associated with the diverse phenotypic outcomes caused by ΔrgsD. Collectively, we conclude that RgsD attenuates cAMP-PKA signaling pathway and negatively regulates asexual development, toxigenesis, melanin production, and virulence in A. fumigatus.


RgsD negatively controls asexual development. Previous studies have revealed that the RGS proteins
FlbA, GprK, Rax1 (RgsB), and RgsC are positive regulators of asexual development as the conidia numbers were drastically decreased by the absence of any one of these RGS proteins [10][11][12]14 . To investigate functions of rgsD, we generated the rgsD null (ΔrgsD) mutant and complemented strains (C′). Although the ΔrgsD mutant demonstrated no significant change in radial growth, it formed very dark colonies with highly increased thallic density and significantly increased formation of mature conidiophores compared to wild type (WT) and C' strains ( Fig. 1A,C). Distinct from the other RGS mutants, the ΔrgsD mutant produced 4 fold more conidia than WT and C′ strains (Fig. 1B). Moreover, the ΔrgsD mutant produced a higher number of mature conidiophores (asexual developmental structures) than WT and C′ strains in liquid medium after 24 hours incubation (Fig. 1C). In WT and C' strains, mRNA levels of rgsD were high at late stage of development whereas those of other asexual developmental regulators increase at earlier stage of development in WT (Fig. 1D). As shown in Fig. 1E, in the ΔrgsD mutant, mRNA levels of the key asexual developmental regulators abaA, brlA, vosA, and wetA were significantly increased at all time points tested. Conversely, mRNA levels of nsdD and veA, known negative regulators of brlA expression and conidiation, were low in the ΔrgsD mutant compared to WT and C′ (Fig. 1E). These results suggest that RgsD negatively controls conidiation and expression of developmental activators, but positively regulates nsdD and veA. Morevoer, as rgsD mRNA accumulates later phases of conidiation, it is likely that the RgsD protein (but not mRNA) is present and acting at early time points of life cycle and development, and the RgsD protein is re-generated during conidiogenesis.
RgsD interacts with Gα protein in yeast. As the RgsD protein contains a clear RGS domain, it is conceivable that RgsD might physically interact with one or more of the three Gα proteins (GpaA, GpaB, and GanA) present in A. fumigatus. To test this interaction potential, we carried out the split-ubiquitin yeast two-hybrid interaction assays with RgsD and three Gα proteins. RgsD::Cub plasmid was constructed by fusing Cub to the C-terminus of the full-length RgsD cDNA. Individual Gα protein constructs were generated by fusing full-length GpaA, GpaB, and GanA cDNA with the mutated N-terminal half of ubiquitin (NubG), respectively. As shown in Fig. 2A,B, those yeast transformants co-expressing RgsD::Cub/NubG::GpaB and RgsD::Cub/NubG::GanA grew better than positive control on the medium lacking histidine or adenine, and showed high levels of ß-galactosidase enzyme activity. Somewhat unexpectedly, those yeast transformants co-expressing RgsD::Cub/ NubG::GpaA also grew on the medium lacking histidine or adenine, and showed ß-galactosidase enzyme activity equal to the positive control. These results indicate that RgsD may interact GpaB and GanA preferentially, but also has the ability to interact with GpaA.
RgsD downregulates cAMP-dependent PKA signaling pathway. To investigate the role of RgsD in governing spore germination, we analyzed the kinetics of germ tube emergence in three strains. Although the germination of the ΔrgsD conidia was started slightly earlier, there were no significant differences between the mutant and WT ( Supplementary Fig. S2). The components of the PKA signaling pathway in A. fumigatus, GpaB (Gα), AcyA (adenylyl cyclase), PkaC1 (the major catalytic subunit), and PksP (a polyketide synthase needed for spore pigmentation) are required for the proper growth, asexual sporulation, and conidial pigmentation [15][16][17] . Previous studies have demonstrated that the cAMP-PKA signaling pathway is attenuated by RgsA in A. nidulans 13,18,19 . To investigate the relationship between RgsD and cAMP-PKA signaling pathway, we analyzed mRNA levels of gpaB, acyA, pkaC1, and pksP mRNA. As shown in Fig. 3A, gpaB, acyA, pkaC1, and pksP mRNA levels were significantly higher in the ΔrgsD mutant than in WT and C′ strains. To investigate this further, we assessed PKA activity using a peptide substrate, kemptide, which is specifically recognized and phosphorylated by PKA. Proteins were extracted from each strain grown in MMG at indicated time and assayed for PKA activity. The phosphorylated negatively charged kemptide migrates to the anode and the signal is proportional to higher PKA activity in the protein extracts. While WT and C′ strains exhibited very little PKA activity in all tested protein extracts, the ΔrgsD mutant showed 2-fold higher PKA activity at 24 h than other samples (Fig. 3B,C). These  results indicate that RgsD is required for proper control of expression of gpaB, acyA, pkaC1, and pksP, and may negatively regulate a cAMP-PKA signaling pathway.

Repressive role of RgsD in stress response and melanin biosynthesis.
To test a potential role of RgsD in stress response, the ΔrgsD conidia were exposed to UV and H 2 O 2 , and the survival rates were determined. As shown in Fig. 4A, the ΔrgsD conidia were more resistant than those of WT and C′ strains against UV irradiation. While the relative survival rates of conidia of WT and C′ strains were about 20%, survival rate of the ΔrgsD conidia was about 40% at 10 mJ/cm 2 . The difference was clearer at higher dose (20 mJ/cm 2 ), where approximately 18% of the ΔrgsD conidia survived but only about 5% of conidia of WT and C′ strains remained alive (Fig. 4A). As shown in Fig. 4B, the conidia of the ΔrgsD mutant were also more resistant to H 2 O 2 . The survival rate of WT, mutant, and C′ conidia was about 65, 95, and 64%, respectively, at 2.5 mM H 2 O 2 . Noticeably, the mutant strain formed a strongly pigmented colony. As the proposed functions of melanin are protection against UV irradiation and scavenging reactive oxygen species, we examined mRNA levels of melanin biosynthetic genes, melanin content, and ultrastructure of conidia. The mRNA levels of arp1 and arp2 which function in dihydroxynaphthalene (DHN)-melanin pathway of fungi 20 were maximum at 12 hours and gradually decreased in the mutant, whereas those were highest at 24 hours and rapidly decreased in WT and C′ strain (Fig. 4B). Moreover, the amount of melanin in the ΔrgsD conidia was about 1.5-fold higher than that of WT and C′ strains conidia (Fig. 4C). We next examined the cell wall architecture of the WT, ΔrgsD, and C′ conidia with transmission RgsD negatively controls gliotoxin production. We showed that the key asexual developmental activator BrlA positively regulates gliotoxin (GT) production in A. fumigatus 22 . Moreover, we also showed that the deletion of RGSs including FlbA, GprK, Rax1, and RgsC resulted in lowered brlA mRNA levels and decreased GT production [10][11][12] . As the deletion of rgsD resulted in up-regulated brlA expression different from previously reported RGSs, we examined the effect of rgsD deletion in GT production. We assessed levels of GT in WT, ΔrgsD, and C' strains, and found that the ΔrgsD mutant produced about 5-fold more GT (Fig. 5A). We then checked mRNA levels of key GT biosynthetic genes by qRT-PCR, and found that mRNA levels of gliM, gliP, gliT, and gliZ were significantly (p < 0.01) higher in the ΔrgsD mutant than in WT and C′ strains (Fig. 5B).
Nullifying rgsD leads to enhanced virulence. We also tested the effects of RgsD on virulence using the wax moth larvae Galleria mellonella. Conidia (1 × 10 6 ) of WT, ΔrgsD, and C′ strains were inoculated to the larvae, and the survival rate of G. mellonella was recorded daily. The most rapid lethality occurred with the ΔrgsD mutant and this strain was significantly more virulent than the other challenge doses (p < 0.0001). As shown in Fig. 6A, larvae inoculated with conidia of either WT or C′ strains began to die at day 3 post-inoculation, with the number of survivors continuing to decrease over the course of the experiment. In the 7 days after infection, about 50% of larvae in both strains died. However, the mortality level of ΔrgsD strain was drastically higher than those of WT and C′ strains, all larvae tested died at 6 days after infection. We also observed that larvae infected with the mutant conidia darkened and shrank at 24 hours post inoculation. In contrast, larvae infected with WT and C′ strains remain in color and did not shrink at same time (Fig. 6B). Typically, infected G. mellonella larvae melanized by forming melanotic capsules surrounding pathogens 23,24 , and the observed darkening of the infected larvae may be due to the formation of these melanotic capsules. To better understand the fate of A. fumigatus inoculated into G. mellonella, infected larvae were fixed in neutral buffered formalin at 24 hours post infection and sectioned for histopathology. Figure 6C showed Periodic acid-Schiff (PAS) and Grocott methenamine-silver (GMS)-stained sections of uninfected (PBS) and infected larvae. The more fungal cells were observed in ΔrgsD conidia-infected larvae than those of WT and C′ conidia-infected larvae. No fungal cells were detected in the PBS-injected control larvae.   Transcriptome analysis. To investigate the roles of RgsD in A. fumigatus biology, we carried out RNA-sequencing analysis using the ΔrgsD and WT cells collected at 12 hours post asexual-developmental induction as described in the Methods section. Two biological replicates showed a high level of correlation (R = 0.93, Fig. 7A). 9,825 genes were differentially expressed and 7,503 genes expressed with slight fold change (−1 < log 2 FC < 1) (Fig. 7B). A total of 385 genes showed more than two-fold differential expression (p < 0.05), of which 168 genes exhibited higher transcript levels in ΔrgsD strain than in WT strain and 217 genes were down-regulated. Functional category analysis was carried out by determining Gene Ontology (GO) terms that were enriched in differentially expressed genes (DEGs). The top significant molecular function GO categories are "signal transducer activity", "transporter activity", "oxidoreductase activity", "cofactor binding activity", and "catalytic activity". The top significant cellular component GO categories are "fungal-type cell wall", "cell periphery", "intrinsic and integral component of membrane", and "extracellular region". The top significant biological process GO categories are "secondary metabolic process", "developmental process", and "pathogenesis". The most enriched molecular function, cellular component, and biological process GO categories are "catalytic activity", "intrinsic component of membrane", and "metabolic process", respectively (Fig. 7C). The top 20 DEGs with increased mRNA accumulation levels in ΔrgsD strain compared to WT are listed in Table 1. Notably, mRNA levels of the key regulator of asexual sporulation brlA and the melanin synthesis-related gene abr1 were above 8-fold higher in the ΔrgsD mutant than WT, consistent with our phenotypic data. Numerous genes involved in signal transduction, asexual development, cAMP-dependent PKA signaling pathway, and pigment biosynthesis were also up-regulated in the absence of rgsD (Supplementary Fig. S3)

B
Relative mRNA expression  (Table 2), including the MgtC/SapB family membrane protein which are homologous to a bacterial virulence factor but their function was still not known 25 .

Discussion
RGS proteins are GTPase accelerating proteins (GAPs) that act as negative regulators of multifunctional signaling in many fungi [10][11][12]14,26,27 . They function by enhancing the intrinsic GTPase activity leading to accelerated hydrolysis of GTP-Gα subunits of heterotrimeric G protein and inactivation of the trimers [28][29][30] . Since RGS proteins are pathologically and physiologically important regulators of the signaling pathways, elucidating the regulatory mechanisms of RGS proteins will provide an expanded basis for understanding G protein signaling and for controlling human pathogenic fungi. Among six potential RGS proteins identified in A. fumigatus, previous studies revealed that the RGS proteins FlbA, GprK, Rax1 (RgsB), and RgsC play redundant roles in regulating asexual sporulation, stress response, and virulence [10][11][12]14 . FlbA (Afu2g11180) regulates GpaA signaling and contributes to proper progression of asexual sporulation 10 . RgsB (Afu4g12640), the orthologue of Saccharomyces cerevisiae Rax1, also controls growth and asexual development of A. fumigatus. Vegetative growth and conidia number is highly reduced in the absence of rgsB. Moreover, it also modulates the content of trehalose and melanin in conidia, providing resistance against external oxidative stress 12 . Another RGS protein RgsC (Afu1g09040) is needed for proper growth, conidiation, stress response, and GT production. The deletion of rgsC results in impaired growth and asexual sporulation 14 . The GPCR-RGS hybrid protein GprK (Afu4g01350) down-regulates the PKA-mediated germination pathway and is responsible for normal asexual development. The ΔgprK mutant shows severely impaired asexual sporulation and reduced tolerance to oxidative stresses 11 .
Despite having a very simple domain structure ( Supplementary Fig. S1A), RgsD may play an opposite role in regulating signaling compared to previously reported RGSs. The deletion of rgsD causes high production of conidia (about 4 fold) coupled with increased expression of key developmental regulators, but reduced the expression of nsdD and veA (Fig. 1). NsdD is known as a major negative regulator of brlA expression and the ΔnsdD mutant produced more conidia than WT in A. fumigatus 31 . VeA also acts as a negative regulator in conidiation. The absence of veA results in highly increased production of conidiophores and accumulation of brlA mRNA 32 . These findings indicate that RgsD may function as a negative regulator of conidiation and some developmental regulatory genes.
With the split-ubiquitin system, it appears that RgsD can interact the three Gα proteins in A. fumigatus. However, the data clearly indicate that RgsD might preferentially interact with GpaB, then GanA and GpaA (Fig. 2). Previous studies showed that GpaB stimulates cAMP-PKA signaling 15 . We demonstrated that the rgsD mutant showed the higher mRNA levels of PKA signaling components and PKA activity, which implicates that RgsD serves as an important negative regulator of cAMP-PKA signaling (Fig. 3). In A. fumigatus, the cAMP-PKA signaling cascade composed G protein α subunit GpaB, adenylate cyclase AcyA, and protein kinase A, regulates growth, development, and secondary metabolites synthesis. Deletion of PKA catalytic subunit pkaC1 showed reduced conidiation, growth, and pigment formation 15 . PKA activity was not detected in ΔgpaB and ΔacyA strains and conidia of both strains were almost avirulent 33 . Moreover, elevated PKA activity led to high expression of polyketide synthase gene pkaP which is essential for the production of melanin and pathogenicity 15 . As expected, the ΔrgsD mutant was more resistant to UV irradiation and oxidative stress, and mRNA of arp1 and arp2 which encode scytalone dehydratase and hydroxynaphthalene reductase, respectively, accumulated more rapidly and at higher levels in the ΔrgsD mutant. Furthermore, the content of melanin in the ΔrgsD conidia was higher and the mutant conidia formed thicker electron dense melanized layer than those of WT and C′ strain (Fig. 4), indicating that RgsD downregulates cAMP-PKA signaling pathway and plays an important negative role in melanin biosynthesis that may modulate fungal response to external stresses. Synthesis and regulation of GT is done by the gli cluster composed of 13 genes in A. fumigatus 34,35 . We checked the GT content and mRNA levels of four genes, gliM (o-methyltransferase), gliP (dioxopiperazine synthase), gliT (GT oxidoreductase), and gliZ (zinc finger transcriptional regulator) in three strains. As shown in Fig. 5, both GT production and GT biosynthetic genes mRNA levels increased in the absence of rgsD, suggesting that RgsD serves as a negative regulator of GT synthesis.
It is well-known that both melanin and GT are important virulence determinants of A. fumigatus. DHN-melanin specifically interferes the functions of host phagocytes, leading to survival of the fungus within host cell 36 . GT inhibits the oxidative burst of neutrophil and increases the ability of the fungus to survive against neutrophil's attack 37 . As ΔrgsD strain produces higher amount of melanin and GT, it was conceivable that the mutant strain would be more virulent than WT and C′ strain. To confirm this, we performed virulence studies   with insect wax moth larvae. The ΔrgsD conidia were highly virulent to the insect and killed all the larvae within 6 days post inoculation. We readily observed that larvae infected with the mutant conidia darkened at 24 hours inoculation and it seemed to be a positive correlation between the degree of larvae darkening and the level of pathogenicity of the strain. Furthermore, higher density of fungal cell was found in the ΔrgsD mutant infected larvae gut (Fig. 6). From these results we conclude that RgsD modulates (attenuates) virulence likely by downregulating biosynthesis of melanin, GT, and potentially other virulence factors in A. fumigatus. Comparative transcriptomics of WT and the ΔrgsD mutant was used for the investigation of RgsD regulated target genes and signaling pathways. Notably, the C 2 H 2 type transcription factor BrlA and the conidial pigment biosynthesis oxidase Abr1 were significantly induced in mutant strain (Table 1). BrlA regulates diverse biological processes in A. fumigatus including asexual sporulation and GT production 22  The results of the RNA-seq analysis demonstrate the diversity in cellular processes especially, asexual sporulation and DHN-melanin synthesis regulated by RgsD in A. fumigatus. Collectively, we propose a genetic model depicting the regulatory roles of RgsD in A. fumigatus (Fig. 8). In the absence of RgsD, the cAMP-PKA signal transduction pathway may be consistently activated, which might lead to increased melanin and secondary metabolite production, tolerance to environmental stresses, and virulence.

Methods
Strains and culture conditions. Glucose minimal medium (MMG) and MMG with 0.1% yeast extract (MMY) with appropriate supplements were used for general culture of A. fumigatus strains 40 . For liquid submerged culture and phenotypic analyses on air-exposed culture were performed as described previously 11 . To examine secondary metabolite production, spores of relevant strains were inoculated 50 ml of liquid MMY and incubated at 250 rpm at 37 °C for 4 days.
Generation of the rgsD deletion mutant. The oligonucleotides used in this study are listed in Supplementary Table S1. The rgsD gene was deleted in A. fumigatus AF293.1 (pyrG1) strain 41 . The deletion construct generated employing double-joint PCR (DJ-PCR) 42 containing the A. nidulans selective marker (AnpyrG) with the 5′ and 3′ flanking regions of the rgsD gene was introduced into the recipient strain AF293.1 43 . The selective marker was amplified from A. nidulans FGSC4 genomic DNA with the primer pair oligo 109/oligo 110. The ΔrgsD mutant was isolated and confirmed by PCR, followed by restriction enzyme digestion ( Supplementary  Fig. S4) 42 . To complement rgsD null mutant, a single joint PCR (SJ-PCR) method was used 42 . The ORF of rgsD    (Table S1). For RNA-seq analyses, 12 hours-old culture of WT and mutant strains were harvested from solid MMY. Total RNA was extracted and submitted to eBiogen Inc. (Seoul, Korea) for library preparation and sequencing.

Phenotypic analyses.
To examine germination levels, conidia of WT and mutant were inoculated in 5 ml of liquid MMY and incubated at 37 °C. Levels of germination were examined every 2 hours after inoculation under a microscope. Conidia were collected in 0.5% Tween 80 from the entire colony and filtered through Miracloth (Calbiochem, CA), and counted using a hemacytometer. Hydrogen peroxide sensitivity of conidia was examined by incubating 1 ml of spore suspensions containing 10 5 conidia with varying concentrations (0, 1.25 or 2.50 mM) of H 2 O 2 and incubated for 30 min at room temperature. Each spore suspension was diluted with sterilized H 2 O, and conidia were inoculated into solid MMY. After incubation at 37 °C for 48 hours, colony numbers were counted and calculated as a survival ratio of the untreated control. UV tolerance test was carried out as follows. Fresh conidia were plated out on solid MMY plates (100 conidia per plate). The plates were irradiated immediately with UV using a UV cross-linker at indicated doses and incubated at 37 °C for 48 hours. The colony numbers were counted and calculated as a survival ratio of the untreated control. The production of gliotoxin (GT) was determined as described previously 45  Protein-protein interaction assay. Yeast two-hybrid interaction assays were performed using DUALhunter Kit (Dualsystem Biotech, Switzerland) according to the manufacturer's instruction. Briefly, full-length RgsD cDNA was cloned into yeast expression vector pDHB1 and cDNAs of GpaA, GpaB, and GanA were cloned into the pPR3-N vector. All cDNA sequences were confirmed by DNA sequencing. Cub and NubG fusion constructs were co-introduced into host yeast strain NMY51. pAI-Alg5 contains a wild type Nub serves as a positive control and pDL2-Alg5 serves as a negative control. Interaction was determined by the growth of yeast transformants on medium without histidine or adenine, and by measuring the β-galactosidase activity using Yeast ß-Galactosidase Assay Kit (Thermo Scientific, USA).

PKA assay. For determination of PKA activity, Pep Tag Non-Radioactive cAMP-Dependent Protein Kinase
Assay kit (Promega, USA) was applied according to the manufacturer's instructions. Homogenized mycelia of each strain were suspended in extraction buffer 33 and incubated on ice for 15 min. After centrifugation, 10 µl of supernatant adjusted to a protein concentration of 3 mg/ml was used in an assay for PKA activity. Insect infection model. The insect survival assay was performed as previously described with some modifications 47 . Briefly, sixth instar Galleria mellonella were infected by injecting the fresh conidia (1 × 10 5 ) into the last left pro-leg and incubated at 37 °C in the dark for the duration of the experiment. Larvae were checked daily for survival and Kaplan-Meier survival curves were analyzed using the Log-Rank (Mantel-Cox) test for significance (p < 0.0001).

RNA-seq experiment.
For control and test RNAs, the construction of library was performed using SENSE mRNA-Seq Library Prep Kit (Lexogen, Inc., Austria) according to the manufacturer's instructions. Briefly, each 2 µg total RNA are prepared and incubated with magnetic beads decorated with oligo-dT and then other RNAs except mRNA was removed. Library production is initiated by the random hybridization of starter/stopper heterodimers to the poly(A) RNA still bound to the magnetic beads. These starter/stopper heterodimers contain Illumina-compatible linker sequences. A single-tube reverse transcription and ligation reaction extends the starter to the next hybridized heterodimer, where the newly-synthesized cDNA insert is ligated to the stopper. Second strand synthesis is performed to release the library from the beads, and the library is then amplified. Barcodes were introduced when the library is amplified. High-throughput sequencing was performed as paired-end 100 sequencing using HiSeq. 2000 (Illumina, Inc., USA). RNA-Seq reads were mapped using TopHat software tool in order to obtain the alignment file. The alignment file was used for assembling transcripts, estimating their abundances and detecting differential expression of genes or isoforms using cufflinks. Gene classification was based on searches done by BioCarta (http://www.biocarta.com/), GenMAPP (http://www.genmapp.org/), DAVID (http://david.abcc.ncifcrf.gov/), and Medline databases (http://www.ncbi.nlm.nih.gov/).

Data Availability
The RNA-seq data are available from NCBI Gene Expression Omnibus (GEO) database (the accession number is GSE100101).