Regulation of Secondary Metabolism by the Velvet Complex Is Temperature-Responsive in Aspergillus

Sensing and responding to environmental cues is critical to the lifestyle of filamentous fungi. How environmental variation influences fungi to produce a wide diversity of ecologically important secondary metabolites (SMs) is not well understood. To address this question, we first examined changes in global gene expression of the opportunistic human pathogen, Aspergillus fumigatus, after exposure to different temperature conditions. We found that 11 of the 37 SM gene clusters in A. fumigatus were expressed at higher levels at 30° than at 37°. We next investigated the role of the light-responsive Velvet complex in environment-dependent gene expression by examining temperature-dependent transcription profiles in the absence of two key members of the Velvet protein complex, VeA and LaeA. We found that the 11 temperature-regulated SM gene clusters required VeA at 37° and LaeA at both 30 and 37° for wild-type levels of expression. Interestingly, four SM gene clusters were regulated by VeA at 37° but not at 30°, and two additional ones were regulated by VeA at both temperatures but were substantially less so at 30°, indicating that the role of VeA and, more generally of the Velvet complex, in the regulation of certain SM gene clusters is temperature-dependent. Our findings support the hypothesis that fungal secondary metabolism is regulated by an intertwined network of transcriptional regulators responsive to multiple environmental factors.

regulators of fungal development and secondary metabolism (Bayram and Braus 2012;Calvo et al. 2016). In the absence of light in A. nidulans, two Velvet complex members, VeA and VelB, enter the nucleus, where VeA interacts with the chromatin-modifying protein LaeA (Bayram et al. 2008). The resulting heterotrimeric protein complex modulates expression of SM gene clusters and developmental processes in many fungi (Wiemann et al. 2010;Hoff et al. 2010;Chettri et al. 2012;Lind et al. 2015), including the opportunistic human pathogen A. fumigatus (Perrin et al. 2007;Dhingra et al. 2012Dhingra et al. , 2013. While most master regulators of secondary metabolism are known in the context of the individual environmental cues that activate them, it is likely that these regulators combinatorically control SM production to fine-tune the metabolic profile of a fungus to changing environments. The possibility of combinatorial regulation is supported by recent studies showing that multiple environmental cues can regulate production of the SM terrain in A. terreus (Gressler et al. 2015), that both the lightresponsive regulator VeA and the nitrogen regulator AreA are required for wild-type (WT) levels of SM-producing gene transcription in Fusarium oxysporum (López-Berges et al. 2014), and that glucose concentration can impact SM production in A. nidulans through changes in the subcellular localization of VeA (Atoui et al. 2010).
The fungal genus Aspergillus is an excellent system to examine the influence of environmental variation in SM regulation, as the mechanisms for SM production have been widely studied in this group of organisms. The SM gene clusters (Inglis et al. 2013) and SM production profiles (Chiang et al. 2008;Frisvad et al. 2009) of several species are described in depth, and several master SM regulators are well characterized (Brakhage 2013). Furthermore, although variation of SM production in response to environmental cues, including temperature (O'Brian et al. 2007;Yu et al. 2011), pH (Tilburn et al. 1995;Bignell et al. 2005), light (Bayram et al. 2008), and hypoxia (Blatzer et al. 2011;Barker et al. 2012) has been observed, it has not been systematically characterized or mechanistically understood. For this study, we chose A. fumigatus, the most common cause of a suite of diseases known collectively as aspergillosis (Latge 1999). A. fumigatus produces a diverse array of SMs, including the immune-suppressing SM gliotoxin, which is thought to promote its virulence (Scharf et al. 2012). Additionally, A. fumigatus is highly thermotolerant; it can grow at 55°and can survive at temperatures up to 75° (Beffa et al. 1998;Ryckeboer et al. 2003;Abad et al. 2010). It is unknown whether changes in temperature affect global patterns of gene expression in the secondary metabolic pathways of this opportunistic pathogen.
To test whether variation in environmental cues other than the known light response can influence Velvet complex-based SM regulation in A. fumigatus, we examined global gene expression using RNA sequencing (RNA-seq) in response to different temperatures in WT, DveA, and DlaeA backgrounds. We found that change in temperature had a marked impact on the expression of SM genes, and that VeA regulates the genes required for producing at least four SMs at 37 but not at 30°, suggesting that the Velvet complex is involved in both temperature-and light-based regulation of secondary metabolism in Aspergillus.

RNA isolation
Mycelial mats were collected and immediately frozen in liquid nitrogen. Samples were then lyophilized and ground. Total RNA was extracted using Direct-zol RNA MiniPrep Kit from ZYMO, following the manufacturer's instructions. RNA was resuspended in autoclaved double-distilled H 2 O. Samples were stored at 280°. Expected veA and laeA expression patterns in the WT and corresponding deletion mutants were verified by quantitative RT-PCR (qRT-PCR; Supplemental Material, Figure S2).

RNA-seq
RNA-seq libraries were constructed and sequenced at the Vanderbilt Technologies for Advanced Genomics Core Facility at Vanderbilt University, using the Illumina Tru-seq RNA sample prep kit, as previously described Lind et al. 2015). Briefly, total RNA quality was assessed via Bioanalyzer (Agilent Technologies). Upon passing quality control, poly-A RNA was purified from total RNA and second-strand complementary DNA (cDNA) was synthesized from messenger RNA. cDNA ends were then blunt repaired and 39 ends were adenylated. Barcoded adapters were ligated to the adenylated ends and the libraries were PCR-enriched, quantified, pooled, and sequenced on an Illumina HiSequation 2500 sequencer. Two biological replicates were generated for each strain sequenced.

Gene expression analysis
Raw RNA-seq reads were trimmed of low-quality reads and adapter sequences using Trimmomatic with the suggested parameters for singleend read trimming (Bolger et al. 2014). After read trimming, all samples contained between 9.5 and 14.1 million reads, with the average sample containing 12 million reads. Trimmed reads were aligned to the A. fumigatus Af293 version s03_m04_r11 genome from the Aspergillus Genome Database (Arnaud et al. 2010(Arnaud et al. , 2012. Read alignment was performed with Tophat2, using the reference gene annotation to guide alignment and without attempting to detect novel transcripts (parameter: -no-novel-juncs) (Kim et al. 2013). Reads aligning to each gene were counted using HTSeq-count, with the union mode . Differential expression was determined using the DESeq2 R package (Love et al. 2014). Genes were considered differentially expressed if their Benjamini-Hochberg adjusted p-value was , 0.1 and their log 2 fold-change was . 1 or , 21.

Functional enrichment analysis
Functional category enrichment was determined for overexpressed and underexpressed genes in all conditions tested, using the Cytoscape plugin BiNGO (Shannon et al. 2003;Maere et al. 2005). To allow for a high-level view of the types of differentially expressed gene sets, the Aspergillus GOSlim v1.2 term subset was used (The Gene Ontology Consortium 2014). The Benjamini-Hochberg multiple testing correction was applied and functional categories were considered significantly enriched if the adjusted p-value was , 0.05.

Gene cluster expression
A. fumigatus secondary metabolic gene clusters were taken from a combination of computationally predicted and experimentally characterized gene clusters (Inglis et al. 2013;Lind et al. 2015). A list of all SM gene clusters used in this study is available in Table 1. SM gene clusters were designated as differentially expressed if half or more of the genes in the cluster were differentially expressed.Gene clusters where half or more genes were significantly differentially expressed (adjusted p-value , 0.1) but with a |log 2 fold change| less than 1 were considered weakly differentially expressed. Clusters containing a mix of overexpressed and underexpressed genes were considered to have mixed expression.

Temperature-shift experiments
Conidia from WT strains were inoculated in Czapek-Dox (10 7 spores/ml) and grown as liquid shake cultures in the dark at 30°. After 24 hr of growth, equal biomass (1 g) was transferred to new flasks, which were then cultured at 30 or 37°. Mycelia were harvested and RNA extracted as previously described, at 24 and 72 hr time points, with three biological replicates. This temperature-shift experiment was also performed with a starting culture temperature of 37°. For expression analysis, 5 mg of total RNA was treated with RQ1 RNase-Free DNase (Promega, Madison, WI). cDNA was synthesized with Moloney murine leukemia virus reverse transcriptase (Promega). qRT-PCR was performed with the Applied Biosystems 7000 Real-Time PCR System, using SYBR green dye for fluorescence detection. To determine expression values, cDNA was normalized to 18S ribosomal gene expression. Expression of two backbone biosynthetic genes, gliP and psoA of the gliotoxin and pseurotin gene clusters, was assayed using the primers in Table S1.

Data availability
All RNA-seq data files are available from the NCBI's Short Read Archive database (accession number: SRP080951).

RESULTS
Temperature shift changes the expression of 10% of all genes and of more than half of the genes in SM gene clusters To investigate the effect of temperature on gene expression, we compared the transcriptomes of A. fumigatus WT grown at 37°Compared with WT grown at 30°. This comparison identified 1101 differentially expressed genes (log 2 fold-change . 1, adjusted p-value , 0.1), which corresponds to . 10% of the A. fumigatus transcriptome. Of these genes, 402 were expressed at a higher degree (overexpressed) and 699 genes were expressed at a lower degree (underexpressed) at 37°t han at 30°(File S1). Genes overexpressed at 37°were enriched (adjusted p-value , 0.05) for the functional categories CARBOHYDRATE METABOLIC PROCESS and EXTRACELLULAR REGION; genes underexpressed at 37°were enriched for the categories CELL ADHESION, SECONDARY METABOLIC PROCESS, TOXIN METABOLIC PROCESS, and OXIDOREDUCTASE ACTIVITY ( Figure 1A).
As functional category enrichment analysis indicated that genes involved in secondary metabolism were expressed at lower levels in WT at 37°than at 30°, we next investigated the impact of temperature on expression of each of the 37 previously identified secondary metabolic gene clusters (Inglis et al. 2013;Lind et al. 2015). We found that half or more of the genes in 13 gene clusters were expressed at lower levels at 37°than at 30°, including the clusters encoding the conidial melanin pigment, fumigaclavine, endocrocin, trypacidin, fumipyrrole, gliotoxin, fumiquinazoline, fumitremorgin, fumagillin, pseurotin, and three gene clusters that do not encode known products (cluster 15, cluster 30, and cluster 35) (Figure 3 and File S1). As previous analysis has shown that endocrocin is not produced at temperatures above 35°, these results indicate that this is attributable to changes in gene expression (Berthier et al. 2013). Three other gene clusters that do not encode known products, namely cluster 21, cluster 25, and cluster 36, were overexpressed at 37° (Figure 3 and File S1). Additionally, half or more genes in six gene clusters (cluster 5, cluster 6, cluster 18, cluster 23, cluster 28, and cluster 31) were differentially expressed but contained a mixture n Not known Afu8g02350, Afu8g02360, Afu8g02380, Afu8g02390, Afu8g02400, Afu8g02410, Afu8g02420, Afu8g02430 Inglis et al. (2013) of both overexpressed and underexpressed genes; none of these gene clusters encode known products. The effect of temperature on SM production on two of these gene clusters, gliotoxin and pseurotin, were further tested using temperatureshift experiments. Cultures were grown at 30°and then shifted to either 37 or 30°and harvested after 24 and 72 hr of growth. The shift experiment was also performed by growing the starting culture at 37°and then shifting to either 30 or 37°. Recapitulating our RNA-seq based results, the backbone synthesis gene from the gliotoxin gene cluster, gliP, was more highly expressed at 30 than 37°at 24 and 72 hr in both temperature up-and down-shift experiments ( Figure S3A). The backbone synthesis gene from the pseurotin gene cluster, psoA, was more highly expressed at 37 than 30°at the 24 hr time point during the temperature up-shift experiment; however, at 72 hr, the gene was more highly expressed at 37°( Figure S3B). Further, psoA was more highly expressed at 30 than 37°at all time points for temperature down-shift experiments.
VeA regulates a much large number of genes at 37°t han at 30°T o investigate how temperature influences VeA's role in controlling gene expression, we compared the transcriptomes of a DveA strain with WT grown at either 37 or 30°. In agreement with previous studies Lind et al. 2015), we found a very large number (3404) of differentially expressed genes in DveA at 37°, with 1821 overexpressed genes and 1583 underexpressed genes (File S1). Far fewer genes were differentially expressed in the DveA strain at 30°. Specifically, 1986 genes were differentially expressed in DveA, with 1026 genes overexpressed and 960 genes underexpressed (File S1). A comparison of the 3404 differentially expressed genes at 37°with the 1986 differentially expressed genes at 30°revealed that a subset of 1468 genes were differentially expressed in DveA at both temperatures, suggesting that their regulation by VeA is temperature independent ( Figure 1B). However, while 518 genes were differentially expressed solely at 30°, almost four times as many genes (1935) were differentially expressed solely at 37°; these results indicate that the regulatory impact of VeA is much greater at 37°than at 30°.
To determine the functions of differentially expressed genes in DveA vs. WT at 30 and 37°, we performed functional category enrichment analyses. Overexpressed genes in DveA were enriched for functional categories relating to transcription and translation activity at both 30 and 37°, while the categories DNA-DEPENDENT TRANSCRIPTION, TRANS-FERASE ACTIVITY, NUCLEOTIDYLTRANSFERASE ACTIVITY, REGULATION OF BIOLOG-ICAL PROCESS, and DNA BINDING were only enriched at 37°( Figure 1A). Further, the number of overexpressed genes in each category was higher for all significantly enriched categories at 37°, with the Genes underexpressed in DveA were enriched for functional categories related to secondary metabolism, including SECONDARY METABOLIC PROCESS and TOXIN METABOLIC PROCESS. The only significantly enriched category for genes underexpressed in DveA at 37°that was not enriched for genes underexpressed at 30°was CARBOHYDRATE METABOLIC PROCESS ( Figure 1A). The number of underexpressed genes annotated to each functional category was higher at 37°, with the exception of RIBOSOME, CYTOSKELETAL ORGANIZATION, and CELL ADHESION, which remained unchanged (File S2). These enrichment analyses indicate that though many more genes are differentially expressed in DveA at 37°, VeA is regulating similar categories of genes at both temperatures.

LaeA regulates similar numbers and types of genes at 30°and 37°T
o investigate how temperature influences LaeA's role in gene regulation, we compared the transcriptomes of a DlaeA strain with WT grown at either 37 or 30°. While DveA strains showed temperaturedependent differences in the number of differentially expressed genes, DlaeA strains showed similar numbers of differentially expressed genes at both 30 and 37°. In total, 1971 genes were differentially expressed in DlaeA strains compared with WT at 37°(632 overexpressed and 1339 underexpressed), while 1809 genes were differentially expressed at 30°(452 overexpressed and 1357 underexpressed) (File S1). There was moderate overlap of the sets of differentially expressed genes at the two temperatures; 1109 genes were differentially expressed in DlaeA at both temperatures, while 770 and 862 genes were only differentially expressed at 30 and 37°, respectively ( Figure 1C).
To identify the functions of genes differentially expressed in the DlaeA strain compared with WT at 37 and 30°, we performed functional category enrichment analyses. Genes overexpressed at both 37 and 30°in DlaeA were enriched for the categories TRANSPORT, TRANSPORTER ACTIVITY, MEMBRANE, and PLASMA MEMBRANE. However, genes overexpressed at 37°were enriched for an additional 14 functional categories related to cell division, filamentous growth, and DNA metabolism that were not enriched in genes overexpressed at 30°( Figure 1A). Underexpressed genes at both 30 and 37°were enriched for categories relating to secondary metabolism, in agreement with LaeA's well-documented role as a master regulator of secondary metabolism (Bok et al. 2006;Bayram and Braus 2012). Two functional categories, CARBOHYDRATE METABOLISM and CELL WALL, were enriched for underexpressed genes at 30 but not 37°.
VeA and LaeA have greater regulatory overlap at 37°t han at 30°A s VeA and LaeA are both members of the Velvet complex and are known to interact, it is very likely that they exhibit substantial overlap in the genes they regulate (Calvo 2008). To examine the effect of temperature on this regulatory overlap, we determined the intersection of genes differentially expressed in DveA vs. WT and DlaeA vs. WT at 30 and 37°. In total, 741 genes were underexpressed in both DveA and DlaeA at 37°(this number corresponds to 41% of all underexpressed genes in DveA and 55% of all underexpressed genes in DlaeA) and 579 genes were underexpressed in both DveA and DlaeA at 30°(41% of all underexpressed genes in DveA and 55% of all underexpressed genes in DlaeA) (Figure 2, B (Figure 2A and File S3), no functional categories were significantly enriched at 30°. Because fewer genes were overexpressed than were underexpressed in both DveA and DlaeA at either temperature, and many more genes were overexpressed in VeA's absence than in LaeA's absence, these results suggest that LaeA may primarily function as a positive regulator of gene expression.
Many SM gene clusters are regulated by both VeA and LaeA at 37°, but only by LaeA at 30°W e expect that SM clusters regulated by the Velvet complex, comprised of the VelB, VeA, and LaeA proteins (Bayram et al. 2008), will require both VeA and LaeA for WT levels of expression. SM gene clusters not controlled by this protein complex, however, may not show differential gene expression in DveA or DlaeA strains, or may be differentially expressed in only one strain. At 37°, 12 SM gene clusters were underexpressed in both DveA and DlaeA, suggesting that they may be regulated by the Velvet protein complex; these clusters include 1,8dihydroxynaphthalene (DHN) melanin pigment, fumigaclavine, endocrocin, trypacidin, fumipyrrole, gliotoxin, fumiquinazoline, cluster 28, cluster 30, fumitremorgin, fumagillin, and pseurotin ( Figure 3). Interestingly, six of these SM gene clusters were either normally expressed in DveA at 30°or had much less of a change from WT expression, suggesting that VeA's regulatory role may be temperature-dependent. These clusters include DHN melanin pigment, fumiquinazoline, cluster 28, fumitremorgin, fumagillin, and pseurotin ( Figure 3, Figure 4, Figure  S1, and File S1). Furthermore, clusters that expressed more highly at Figure 3 Differential expression of SM gene clusters in all conditions. Dark red boxes indicate half or more genes are underexpressed, dark blue boxes indicate half or more genes are overexpressed, and dark purple boxes indicate that half or more genes are a combination of overexpressed and underexpressed genes. Light-colored boxes indicate that half or more genes in that gene cluster meet the statistical significance cutoff for differential expression but have less than a twofold change in expression.
37°than at 30°in WT A. fumigatus were also often underexpressed in DveA and DlaeA strains; of the 12 clusters underexpressed in both DveA and DlaeA strains at 37° (Figure 3), 11 were expressed at higher levels in WT at 30°than at 37°. The exception was cluster 28, which was underexpressed in both DveA and DlaeA strains at 37°but was expressed at similar levels in WT at 30 and 37°.
Several clusters were differentially expressed in either DlaeA or DveA, but not in both. The hexadehydroastechrome cluster, while underexpressed in DlaeA at both 30 and 37°, was overexpressed in DveA at 30°and showed mixed expression in DveA at 37° (Figure 3). Two gene clusters, cluster 14 and cluster 21, were underexpressed in DveA at both temperatures but showed mixed expression in DlaeA. Finally, the neosartoricin/fumicycline cluster, which was very lowly expressed in WT at both 30 and 37°, contained some up-regulated genes in DveA at both temperatures, but showed no change in expression in DlaeA strains. The expression patterns of these gene clusters indicate that, although VeA and LaeA play roles in their regulation, these proteins may in some cases be acting independently of each other.

DISCUSSION
Production of SMs in A. fumigatus and other filamentous fungi is triggered by diverse environmental cues, such as temperature, pH, and nutrient sources, and several master SM regulators that respond to these cues have been identified. However, the extent to which master SM regulators can respond to multiple environmental cues to regulate SM production is not known. Considered together, our findings that temperature regulates global SM production in A. fumigatus and that the light-responsive master SM regulator VeA is also responsive to changes in temperature, provide support for the hypothesis that regulation of SM production occurs in response to multiple environmental cues.
Growth at 37°Compared with 30°had a marked impact on gene expression in A. fumigatus WT, significantly changing the expression levels of 10% of its genes. Importantly, genes involved in secondary metabolism were disproportionately affected ( Figure 1A); 13 of the total 37 SM gene clusters were expressed at higher levels at 30°than at 37°, while three clusters were expressed at lower levels at 30°( Figure 3A). These results are in accordance with studies in A. flavus that find a global pattern of higher SM cluster expression at 30°than at 37°, the optimal temperature for growth in both fungi (Yu et al. 2011). Additional support for our findings that temperature plays a significant role in SM gene expression was provided by qRT-PCR assays of two SM genes, gliP and psoA, in temperature-shift experiments ( Figure S3). Specifically, a temperature shift from 37 to 30°increased the expression of both genes, supporting our conclusion that temperature modulates SM gene expression.
To elucidate the effects of temperature on SM regulation, we exposed deletion strains of genes encoding two key members of the Velvet protein complex, veA and laeA, to different temperature conditions. At 37°, the optimal temperature for A. fumigatus growth, we find that VeA and LaeA are both involved in regulating genes in many SM gene clusters. While the lists of which genes are parts of the known SM gene clusters are not identical to the lists used in previous analyses of LaeA's regulatory role of controlling secondary metabolism, our RNA-seq results generally agree with previously published microarray data (Perrin et al. 2007). One notable difference from previous reports is our finding that a putative terpene-producing cluster on chromosome 5 (Afu5g00100-00135) is under LaeA regulation. Further, our findings that VeA transcriptionally regulates many gene clusters agrees with chemical data that show that VeA is required for the synthesis of fumagillin, fumitremorgin, and fumigaclavine at 37° . The sets of genes that are increased in VeA and LaeA's absence do not show broad overlap in their functions ( Figure 1A), suggesting that VeA and LaeA's regulatory roles are distinct from each other. This inference is further supported by the observation of six SM gene clusters that are differentially regulated by VeA but not by LaeA at different temperatures. The Velvet protein complex formed by LaeA, VeA, and VelB has been implicated as a regulator of secondary metabolism in many fungi (Bayram et al. 2008;Calvo 2008;Wiemann et al. 2010;Bayram and Braus 2012;Chettri et al. 2012); these data provide additional evidence that the LaeA and VeA have functionally distinct roles in regulating SM clusters (Bayram and Braus 2012;Lin et al. 2013).
Our finding that VeA's regulation of SM gene clusters is temperature-dependent raises the hypothesis that, in addition to its critical role in controlling dark-responsive secondary metabolism by localizing in the nucleus under dark conditions and, to a lesser degree, under light conditions (Bayram et al. 2008), VeA may also be involved in controlling the response to temperature. Interestingly, previous work in A. nidulans has shown that glucose concentration influences both VeA's subcellular localization and sterigmatocystin production, altering the effect of light on the biosynthesis of this mycotoxin (Atoui et al. 2010); thus, light and temperature might just be two of the many environmental cues to which VeA responds.
How might VeA, a single protein, mediate such a diversity of regulatory controls on multiple SM gene clusters in response to several different environmental cues? One possibility is that VeA's regulatory diversity is mediated through the protein's multiple interaction partners. VeA forms a heterodimer with another Velvet family protein, VelB, and both proteins are necessary for sexual fruiting body formation in A. nidulans. Many of VeA's interacting partners impact its subcellular localization. For example, in A. nidulans VeA interacts with the methyltransferases LlmF and the VipC-VapB heterodimer, which respectively increase and repress VeA's nuclear import Sarikaya-Bayram et al. 2014). VeA is also known to interact directly with the red light sensing protein FphA and therefore indirectly with the blue light sensing White Collar homologs LreA and LreB, which may modulate VeA's light responsive capabilities and subcellular location, as well as potentially playing a role in glucose response (Purschwitz et al. 2008(Purschwitz et al. , 2009Atoui et al. 2010;Sarikaya-Bayram et al. 2015). Another possible mechanism explaining VeA's multifaceted role is offered by recent experiments in A. nidulans showing that phosphorylation of different combinations of residues of VeA generates distinct phenotypes, including changes in sterigmatocystin production (Rauscher et al. 2015).
Irrespective of what the precise molecular mechanism(s) contribute to VeA's diverse array of regulatory controls, the emerging picture from recent studies, including this one, is that VeA is responding to multiple environmental signals, including light (Bayram et al. 2008), glucose (Atoui et al. 2010), nitrogen source (López-Berges et al. 2014), and temperature (this study), allowing filamentous fungi to modulate cellular processes such as secondary metabolism in response to changing environments.

ACKNOWLEDGMENTS
We thank members of the Rokas laboratory for useful discussions. This work was conducted in part using the resources of the Advanced Computing Center for Research and Education at Vanderbilt University (http://www.accre.vanderbilt.edu/). This work was supported by the U.S. National Library of Medicine training grant 2T15LM007450 (to A.L.L.). The funders had no role in study design, data collection, data analysis, decision to publish, or preparation of the manuscript.