A LaeA- and BrlA-dependent cellular network governs tissue-specific secondary metabolism in the human pathogen Aspergillus fumigatus

Biosynthesis of many ecologically important secondary metabolites (SMs) in filamentous fungi is controlled by several global transcriptional regulators, like the chromatin modifier LaeA, and tied to both development and vegetative growth. In Aspergillus molds, asexual development is regulated by the BrlA>AbaA>WetA transcriptional cascade. To elucidate BrlA pathway involvement in SM regulation, we examined the transcriptional and metabolic profiles of ΔbrlA, ΔabaA, ΔwetA and wild-type strains of the human pathogen Aspergillus fumigatus. We find that BrlA, in addition to regulating production of developmental SMs, regulates vegetative SMs and the SrbA-regulated hypoxia stress response in a concordant fashion to LaeA. We further show that the transcriptional and metabolic equivalence of ΔbrlA and ΔlaeA is mediated by a LaeA requirement preventing heterochromatic marks in the brlA promoter. These results provide a framework for the cellular network regulating not only fungal SMs but diverse cellular processes linked to virulence of this pathogen.


Introduction 35
Filamentous fungi produce a remarkable diversity of specialized secondary metabolites 36 (SMs), which are small molecules that play diverse ecological roles in fungal defense, 37 communication, and virulence (1). In fungi, SMs are typically produced by pathways organized 38 into contiguous biosynthetic gene clusters (BGCs), an organization atypical of metabolic 39 pathways in most other eukaryotes (2). The transcription of these BGCs is often controlled by 40 both cluster-specific transcription factors as well as globally-acting transcriptional regulators. 41 These global regulators respond to a variety of environmental signals including pH, temperature, 42 light, and nutrient sources to transcriptionally regulate BGCs, and are typically well conserved in 43 filamentous fungi (3). 44 45 Many of the environmental signals that regulate SM production in Aspergillus fungi, 46 including temperature, pH, and carbon or nitrogen sources, also trigger the onset of asexual and 47 sexual development (4). At the cellular level, this coupling between SM production and 48 development is orchestrated in part by the velvet protein complex, which is composed of two 49 velvet domain proteins, VeA and VelB, and the methyltransferase LaeA (5). Although the precise 50 mechanism by which the velvet complex regulates the two processes is unknown, LaeA regulates 51 transcription epigenetically through heterochromatin reorganization of target DNA (6, 7). The 52 result of this coupling of SM and development is that several SMs show tissue specificity, i.e., 53 they are localized or produced only in certain tissues. For example, the SMs DHN melanin, 54 fumigaclavines, endocrocin, trypacidin, and fumiquinazolines, appear to be specifically produced 55 in the asexual spores of Aspergillus fumigatus (8)(9)(10)(11)(12). Importantly, several of these asexual spore 56 (conidial) metabolites, all , are part of the pathogenic arsenal of this human 57 pathogen (reviewed in 14). 58 59 network governing tissue-specific secondary metabolism as well as of diverse cellular processes 86 in filamentous fungi. 87 88

Results and Discussion 89
Genome-wide transcriptional impact of BrlA, AbaA, and WetA 90 To examine the genome-wide regulatory roles of the three central regulators of asexual 91 development, we performed RNA sequencing on A. fumigatus wild-type (WT), ΔbrlA, ΔabaA, and 92 ΔwetA mutant strains grown on minimal media in conditions known to induce the production of 93 both vegetative growth-specific and asexual development-specific SMs. 6,738 of the 9,784 genes 94 in the genome of the A. fumigatus Af293 strain were differentially expressed in the ΔbrlA versus 95 WT comparison (3,358 over-expressed and 3,380 under-expressed) ( Table 1, Table S1).  are reported to be asexual development-specific, we found that 27 / 33 (82%) of BGCs are 119 differentially expressed in one or more of the three comparisons examined (Figure 2A), 120 suggesting a much broader governance of SM production by these transcriptional regulators of 121 asexual development. Like the trend observed with genome wide transcriptional impact of these 122 regulators, we find BrlA to be a major contributor to changes in BGC expression, regulating all 123 but one of the differentially expressed BGCs (26 / 27; 96%), followed by WetA (15 / 27; 45%), and 124 AbaA (11 / 27; 34%) ( Figure 2A). Of the 27 differentially expressed BGCs, nine were regulated 125 by all three transcriptional regulators, ten showed BrlA-specific regulation, one showed WetA-126 specific regulation, two showed joint regulation by both BrlA and AbaA, and five showed joint 127 regulated by BrlA and WetA ( Figure 2B). These results suggest that, unlike BrlA, WetA (with the 128 exception of one BGC) and AbaA do not independently regulate their BGC targets. 129

130
The nine BGCs that are jointly regulated by BrlA, AbaA, and WetA include the BGCs for 131 ferricrocin, DHN melanin, fumigaclavines, endocrocin, gliotoxin, fumiquinazolines, and 132 pyripyropenes, as well as the unknown non-ribosomal peptide synthetase (NRPS) cluster 24 and 133 the unknown polyketide synthase (PKS) cluster 31 (Figure 2). Four of the five known asexual 134 development-specific BGCs (endocrocin, fumigaclavine, fumiquinazoline, and DHN melanin) are 135 example, production of ferricrocin, fumigaclavines, endocrocin, gliotoxin, fumiquinazolines, and 162 pyripyropene A is completely abolished or significantly reduced in the ∆brlA mutant culture 163 (Figure 4), mirroring the under-expression of their BGCs in the ∆brlA versus WT comparison 164 (Figure 2). These compounds are also significantly reduced in the ∆abaA mutant culture and 165 correlate with the gene expression patterns of their BGCs in the ∆abaA versus WT comparison, 166 albeit to a lesser degree when compared to that observed in ∆brlA (Figure 4; Figure 2). 167

168
In contrast to ∆brlA and ∆abaA, the correlation between ∆wetA gene expression and 169 metabolite profiles was much lower. For example, we observed a significant increase of several 170 SMs, such as the fumigaclavines and endocrocin, in the ∆wetA mutant even though their 171 corresponding BGCs are under-expressed in the ∆wetA versus WT comparison (Figure 3; Figure  172 2). Endocrocin is also produced as an early shunt product redundantly by the trypacidin BGC in 173 the strain of A. fumigatus used in this study (10), and thus could be attributed to that BGC as well. 174 Although we did not detect the final product trypacidin, minute amounts of the trypacidin precursor 175 questin were produced in the wildtype fungus and, to much lesser degree, in the ∆wetA mutant 176 (Figure 4). The high metabolite production levels in ∆wetA in spite of low gene expression levels 177 could be attributed to the compromised cell wall of this developmental mutant (25) that resulted 178 in increase in SM extraction efficiency compared to other test strains of the fungus. 179 180 Assessment of metabolites from both fungal tissue and growth supernatant (secreted) 181 showed that a major fraction of extracted SMs, including all the known conidial-associated SMs 182 (fumigaclavines, endocrocin, fumiquinazolines, and questin), the intracellular siderophore, 183 ferricrocin, and pseurotin A accumulate in the fungal tissue, whereas that of other SMs, such as 184 the fumitremorgins, terezine D, fumagillin, pyripyropene A, and helvolic acid, are secreted into the 185 growth supernatant ( Figure S1, Table S4). This may be a reflection of the chemical properties of 186 these metabolites and their ability to diffuse or be actively released to the outside of the cell. 9 188 Ten of the 27 differentially regulated BGCs are under BrlA-specific control ( Figure 2B). 189 Except for the BGC of the extracellular siderophore fusarinine C, the SM products of the remaining 190 nine BrlA-specific BGCs have yet to be characterized. Both BrlA and AbaA jointly govern 191 expression of the fumisoquin BGC and the fumitremorgin BGC ( Figure 2B). As the A. fumigatus 192 Af293 strain used in this study is reported to harbor a point mutation in ftmD (Afu8g00200) that 193 renders it incapable of producing the terminal product, fumitremorgin C (26), metabolomic 194 analysis on the fumitremorgin BGC was performed using an early pathway precursor, 195 brevianamide F. Total production of brevianamide F is significantly increased in cultures of all 196 three transcriptional regulator mutants (Figure 4). Surprisingly, we also detect a significant Our gene expression data indicate that WetA positively regulates its sole specific target, 207 the iron-coordinating hexadehydroastechrome (HAS) BGC. As metabolite detection of the iron 208 coordination complex of HAS is challenging, we used the monomeric unit of this complex, terezine 209 D, in our metabolite profiling of this BGC. In contrast to the gene expression data, which show 210 that the BGC is under-expressed in ∆wetA versus WT, we observed that terezine D production is 211 increased in the ∆wetA mutant (Figure 4). Even though the HAS BGC does not appear to be 212 transcriptionally regulated by BrlA or AbaA, we still observed a decrease of terezine D production 213 in both mutants (Figure 4). This could be related to other cellular processes as HAS is a 214 tryptophan derived metabolite dependent on iron-homeostasis (27), with genes in both networks 215 regulated by the BrlA cascade. BrlA and WetA jointly govern the helvolic acid BGC, the 216 fumagillin/pseurotin supercluster, and three unknown BGCs (Figure 2). Compared to WT levels, 217 production of both fumagillin and helvolic acid is increased in the ∆brlA mutant, unchanged in the 218 ∆wetA mutant, and substantially increased in the ∆abaA mutant (Figure 4). addition to these spore-associated BGCs and SMs, BrlA also appears to regulate BGCs and SMs, 229 such as helvolic acid and fumisoquin, which are not known to be associated with specialized 230 developmental tissues but rather with vegetative growth, suggesting that BrlA regulation of 231 secondary metabolism extends beyond asexual development. 232 233

LaeA regulation of secondary metabolism is extensively mediated through BrlA 234
LaeA, a member of the fungal-specific velvet protein complex, is known to regulate 235 secondary metabolism in many agriculturally and medically important filamentous fungi (28). 236 Given the surprising global changes in BGC expression in the ∆brlA mutant as well as the aberrant 237 conidia phenotype previously observed in the ∆laeA mutant (29), we further assessed the genetic 238 relationship between these two global regulators and their governance on secondary metabolism.
Global transcriptome comparison between the LaeA and BrlA regulons in A. fumigatus shows 240 striking concordance in BGC regulation, with 13 / 16 of the LaeA-regulated BGCs as determined 241 by microarray-based transcriptome analysis (13) also regulated by BrlA (Table S5). These include 242 the BGCs responsible for the production of DHN melanin, fumigaclavines, endocrocin, helvolic 243 acid, fumisoquins, gliotoxin, fumiquinazolines, fumitremorgins, fumagillin/pseurotin, and 244 pyripyropenes, and three uncharacterized BGCs (cluster 24, a NRPS-based cluster upstream of 245 the gliotoxin cluster, cluster 15, a PKS-based BGC and cluster 2, a nidulanin-like BGC) (Table  246   S5 there is a substantial decrease of a modification correlating with euchromatin (H3K4me3) in the 266 ∆laeA strain, while the heterochromatic mark H3K9-me3 is greatly enriched ( Figure 6C). Thus, 267 as with BGC regulation, it appears that LaeA epigenetically regulates brlA by impeding 268 heterochromatin formation on the brlA promoter ( Figure 6). Based on these results, we infer that 269 LaeA regulation of secondary metabolism is significantly mediated through its epistatic effect on  (Figure 1). Among these genes are the hypoxia 296 regulators srbA (Afu2g01260) and srbB (Afu4g03460) ( Table S1). Both transcription factors 297 contribute to virulence and are critical for regulation of iron uptake, heme biosynthesis and 298 ergosterol synthesis in A. fumigatus (32). Previous work has determined that SrbA is a DNA-299 binding protein that binds upstream of 97 genes in A. fumigatus CEA10, 91 of which have 300 orthologs in the Af293 strain used in this study (32). 69 / 91 (76%) of these genes are under-301 expressed in ΔbrlA versus WT ( Table 2, Table S6). In contrast, the percentages of SrbA-302 regulated genes were substantially smaller in both ΔabaA vs WT (14 / 91 genes or 15%) or ΔwetA 303 vs WT (13 / 91 genes or 14%). Among the genes co-regulated by BrlA and SrbA are those in the 304 ergosterol biosynthetic pathway, including the first enzyme in the pathway, Erg1 (Afu5g07780), 305 both 14- sterol demethylases (Erg11A/Cyp51A; Afu4g06890 and Erg11B/Cyp51B; 306 Afu7g03740), Erg5 (Afu1g03950) and both C4-sterol methyl oxidases (Erg25A; Afu8g02440 and 307 erg25B; Afu4g04820) (33, 34). The nitrate assimilation genes niiA (Afu1g12840) and niaD 308 perhaps as a member of the velvet protein complex, acts through epigenetic and epistatic 339 regulation of key 'cellular switches', with BrlA representing one of these switches (Figure 7). BrlA 340 is a known transcription factor that was first identified as a regulator of conidiophore development 341 in A. nidulans (36,37). BrlA has also been characterized in several Aspergillus and Penicillium 342 species, and its regulatory functions have always associated with sporulation and frequently with 343 secondary metabolism (16, Table S1) are highly regulated by BrlA. Thus it appears that, 355 minimally, these nine BGCs are induced by LaeA-mediated BrlA activation. However, not all 356 BGCs were similarly regulated by LaeA and BrlA, suggesting that they may require LaeA 357 activation through other or additional 'cellular switches', that they may be solely (positively or 358 negatively) regulated by BrlA, or that they may be regulated through LaeA-and BrlA-independent 359 cascades. Since both LaeA and BrlA are present in other fungal genera, including Penicillium and 360 Talaromyces, and the fact that the secondary metabolites produced by organisms in these genera 361 are distinct from those produced by A. fumigatus, it will be of future interest to address how 362 conserved global molecular circuitry are rewired to control species-specific processes such as 363 secondary metabolism (42). 364 Finally, our work shows that BrlA is the likely mediator of many of the known LaeA cellular 366 cascades, including several associated with A. fumigatus virulence, substantially expanding the 367 diversity of cellular processes that appear to be regulated by BrlA. For example, both proteins are 368 critical for activation of members of the aromatic amino acid and sulfur/methionine pathways, 369 which play a role in virulence of this pathogen (43, 44). We also find that BrlA is a key regulator 370 of hypoxia regulated genes, likely through its regulation of SrbA and SrbB, the two key 371 transcription factors critical for hypoxia adaptation in A. fumigatus (32,45). SrbA is also important 372 in azole resistance through its regulation of the ergosterol biosynthetic pathway (46), and our work 373 uncovers a direct signaling pathway from LaeA to BrlA to SrbA/B to ergosterol gene expression 374 which may reveal new avenues to study the expanding threat of antifungal resistance in 375 Aspergillus species (47). 376 377

Fungal strains and growth conditions 379
All strains used in this study are listed in Table 3 injected into the UHPLC-MS system, separated using an Agilent Zorbax Eclipse XDB-C18 column 460 (2.1 x 150 mm, 1.8 µM particle diameter), and ran using 0.05% formic acid in acetonitrile as the 461 organic phase and 0.05% formic acid in water as the aqueous phase at a flow rate of 0.2 mL/min. 462 The solvent gradient starts at 20% organic for 2 mins, followed by a linear increase to 60% organic 463 over 10 minutes, a linear increase to 100% organic over 1 min, and a final holding at 100% organic 464 for 5 mins totaling to 18 minutes of runtime and data collection. The XDB-C18 column was 465 equilibrated at 20% organic for 5 mins in between each sample injection throughout the entire 466 sequence. Antibodies used for ChIP were: rabbit polyclonal to histone H3 acetyl K9, Abcam, ab10812, rabbit 495 polyclonal to histone H3 trimethyl K4, Upstate, 07-473, rabbit polyclonal to histone H3 acetyl K9, 496 Abcam, ab8898, and rabbit polyclonal to C-terminus histone H3 antibody, ab1791. Two 497 micrograms of antibody were used per reaction of 200 mg total protein. Amplification and 498 detection of precipitated DNA in real-time qPCR was performed with iQ™ SYBR® Green 499 Supermix #170-8880 (Bio-Rad, Cat#170-8880) following the manufacturer's instructions using 500 primers AF brlA(p) F qPCR (CGTACGGGTGTAAGTCTGATC) and AF brlA(p) R qPCR 501 (CTCTGTATCTTCTAGTTCAATGG). Relative amounts of DNA were calculated by dividing the 502 immunoprecipitated DNA by the input DNA. Each PCR reaction was replicated. To normalize the 503 amount of DNA precipitated with histone H3-acetyl K9 and H3-trimethyl K4, the quantities from 504 precipitation with these antibodies was divided by the previously calculated ratio of the anti-C-505 terminus histone H3 precipitation to input DNA.    Table S4  567   Table S1. Differential gene expression of all strains. 568 Table S2. All Gene Ontology enrichment results. 569 Table S3. All secondary metabolic gene clusters in Aspergillus fumigatus. 570 Table S4. T-tests for significant differences between supernatant and mycelial secondary 571 metabolites shown in Figure S1. 572