Chromatin mapping identifies BasR, a key regulator of bacteria-triggered production of fungal secondary metabolites

The eukaryotic epigenetic machinery can be modified by bacteria to reprogram the response of eukaryotes during their interaction with microorganisms. We discovered that the bacterium Streptomyces rapamycinicus triggered increased chromatin acetylation and thus activation of the silent secondary metabolism ors gene cluster in the fungus Aspergillus nidulans. Using this model, we aim understanding mechanisms of microbial communication based on bacteria-triggered chromatin modification. Using genome-wide ChIP-seq analysis of acetylated histone H3, we uncovered the unique chromatin landscape in A. nidulans upon co-cultivation with S. rapamycinicus and relate changes in the acetylation to that in the fungal transcriptome. Differentially acetylated histones were detected in genes involved in secondary metabolism, in amino acid and nitrogen metabolism, in signaling, and encoding transcription factors. Further molecular analyses identified the Myb-like transcription factor BasR as the regulatory node for transduction of the bacterial signal in the fungus and show its function is conserved in other Aspergillus species.


Introduction
The eukaryotic epigenetic machinery can be influenced by bacteria. For example, bacteria can secrete chromatin modifiers or proteins such as methyltransferases that cause chromatin silencing in eukaryotic cells (Yoshida et al., 1990;Rolando et al., 2013). As an early example, we discovered that the silent secondary metabolite (SM) gene cluster for orsellinic acid (ors) in the filamentous fungus Aspergillus nidulans is activated upon physical interaction with the bacterium Streptomyces rapamycinicus. The interaction of the fungus with this distinct bacterium led to increased acetylation of histone H3 lysines 9 and 14 at the ors gene cluster and thus to its activation (Schroeckh et al., 2009;Nützmann et al., 2011;Nützmann et al., 2013). The lysine acetyltransferase (KAT) responsible for the acetylation and activation of the ors gene cluster was shown to be GcnE (Nützmann et al., 2011).
Using this model, we aim to gain an understanding of the molecular mechanisms of microbial communication based on bacteria-triggered chromatin modification. In order to obtain a holistic view on the fungal-bacterial interaction that, in the future, might allow predicting interaction partners and discovering the molecular elements involved, we developed a genome-wide chromatin immunoprecipitation (ChIP)-seq analysis specifically during co-cultivation. This led to the discovery of major alterations of epigenetic marks in the fungus triggered by the bacterium and to the identification of BasR as key regulatory node required for linking bacterial signals with the regulation of SM gene clusters.

Results
Genome-wide profiles of H3K9 and H3K14 acetylation in A. nidulans change upon co-cultivation with S. rapamycinicus A. nidulans with and without S. rapamycinicus was analyzed by genome-wide ChIP-seq for enrichment of acetylated (ac) histone H3 at lysines K9 and K14 (Figure 1; Appendix 1 -Details of ChIP analysis). To account for reads originating from S. rapamycinicus we fused the genomes of A. nidulans (eight chromosomes) and S. rapamycinicus. The resulting fused genome also served as reference for mapping of chromatin marks (see Appendix 1 -Details of the ChIP analysis).
H3K14ac and H3K9ac showed a higher degree of variability across the genome than on H3, implying that the regulatory dynamics of histone acetylation are more specific than those that would be achieved by H3 localization alone. Some areas, such as a region in the first half of chromosome four, were particularly enriched in these acetylation marks, potentially indicating distinctive acetylation islands, which are short loci with continuous enrichment of histone modifications. Such islands have been identified previously in the intergenic and transcribed regions of the human genome, and some of these have been shown to colocalize with known regulatory elements (Roh et al., 2005). An island that is particularly enriched for H3K9ac was found around the ors gene cluster (Figure 1c and 2), thus supporting our previous data (Nützmann et al., 2011). The coverage profiles of H3, H3K14ac and H3K9ac consistently change in co-culture compared to monoculture, as seen in Figure 1. In particular, the promoter region of the genes orsD and orsA showed reduced nucleosome occupancy (see Figure 1c). This could be due to a redistribution of nucleosomes that ultimately changes the distribution of histone marks. Such nucleosome rearrangements might represent the prevailing driver of H3K14ac change which is associated with a reduction in overall acetylation level. This might explain the local H3K14ac decrease at the translation start sites (TSSs) shown in Figures 1  and 2. In comparison to the changes to H3K14ac, the changes to H3K9 acetylation levels are stronger, leading to an increase in H3K9 acetylation despite nucleosome rearrangements. This finding is supported by the observation that unmodified H3 was depleted throughout the ors cluster, especially at the orsA and orsD TSSs.
Co-cultivation of A. nidulans with S. rapamycinicus had a major impact on SM gene clusters, nitrogen assimilation, signaling and mitochondrial activity We employed two strategies to measure changes in histone modification levels. The first analysis was based on the finding that histone acetylation can mostly be found on histones within a gene, in particular on nucleosomes +1 and +2 (Jiang and Pugh, 2009) (Appendix 1-figure 1). Therefore, for each library, we counted mapped reads that overlapped genes. This formed the basis for a quantitative comparison between monocultures and co-cultures using standard read-counting methods for sequencing data (see 'Materials and methods'). Throughout this study, we refer to this method as differential chromatin state (DCS) analysis. The second analysis was based on a first round of peak-calling and subsequent quantification of the peaks. Comparison of the generated data sets showed 84 ± 1.7% similarity. The data obtained from the gene-based DCS method (Supplementary file 1) were used for both further analyses and comparisons of the culture conditions using a false discovery rate (FDR) cut-off of 0.01. This does not include further filtering on the log-fold changes (LFCs) to capture the possible biological relevance of the detected changes. windows (bins) that segment the fused genome at a resolution of 50 kb for each library separately (see 'Materials and methods' for details). The read count values are plotted at the midpoint of each bin, which are connected by lines. Gene density is reported likewise by counting the number of genes for each bin instead of reads. Background values derive from S. rapamycinicus (brown) and A. nidulans (green) grown in monoculture. The red arrow indicates the location of the ors gene cluster. (b) Zoom into chromosome II. The red lines mark the ors gene cluster. Data from three replicates are shown, which share the same tendency. Overall intensities for background, H3K9ac, H3K14ac and H3(Cterm) are compared between A. nidulans monoculture (blue) and co-culture (green). The average genome density (black) is also shown. (c) Example of an Integrative Genomics Viewer (IGV) screenshot showing the region of the ors gene cluster at the bottom of the figure labeled with black arrows. Other differentially acetylated gene bodies are listed in Supplementary file 1. White gene arrows indicate genes that do not belong to the ors gene cluster. Data obtained from monocultures of the fungus are depicted in blue and from co-cultivation in green, whereas background data are shown in gray. DOI: https://doi.org/10.7554/eLife.40969.002

A. nidulans
A. nidulans + S. rapamycinicus  Quality and the absence of possible biases introduced by the co-culture or other sources were further investigated by MA plots. They showed a symmetrical and even distribution around LFC = 0, meeting the requirements for the statistical tests described in the 'Materials and methods' (Appendix 1-figure 2). DCS analysis of H3, as a proxy for nucleosome occupancy, was found to be lower (FDR < 0.01) in 37 genes and higher in two genes during co-cultivation. Using the same cut-off, H3K14ac levels during bacterial-fungal co-cultivation were found to be lower for 154 genes and higher for 104 genes. Differential acetylation of chromatin was found for H3K9ac, with 297 genes with significantly lower and 593 with significantly higher acetylation (Supplementary file 1). The analysis of microarray data obtained under identical conditions showed a positive correlation of higher gene expression with H3K9 acetylation (r = 0.2 for all genes and r = 0.5 for a subset of genes showing differential acetylation; Appendix 1-figures 3 and 4). Data for selected genes are summarized in Supplementary file 2, which shows the LFCs of H3K9ac ChIP-seq data with their corresponding microarray data. In total, higher acetylation during co-cultivation was seen in histones belonging to six SM gene clusters, the ors, aspercryptin (atn), cichorine (cic), sterigmatocystin (stc), anthrone (mdp) and 2,4-dihydroxy-3-methyl-6-(2-oxopropyl)benzaldehyde (dba) gene clusters, with the emericellamide (eas) and microperfuranone clusters being the only ones with reduced acetylation and expression (Supplementary file 2, section V). With a few exceptions, the genes covered by histone H3 that had increased acetylation are involved in calcium signaling and asexual development Figure 3. Influence of S. rapamycinicus on fungal nitrogen metabolism and mitochondrial functions. (a) Normalized ChIP-seq read counts were used to quantify the chromatin state (H3, H3K14ac, H3K9ac) of nitrogen metabolism genes. Counts were obtained by counting reads mapping to the promoter area of each gene, which is defined as the sequence 500 bp down-and 1000 bp upstream from the TSSs. Depicted bars are calculated from three data points. (b) Transcription analysis of randomly selected genes of primary and secondary nitrogen metabolism by qRT-PCR during co-cultivation. Relative mRNA levels were measured after 3 hr and normalized to the b-actin gene expression. The transcription of orsA was used as a positive control. (c) Respiratory activity comparing A. nidulans grown in co-culture with S. rapamycinicus and A. nidulans in monoculture. Respiratory activity was determined using a resazurin assay. Data were normalized to medium. The black line shows the time points that are significantly different between A. nidulans and A. nidulans grown in co-culture with S. rapamycinicus. ***p<0.001. DOI: https://doi.org/10.7554/eLife.40969.006 The following figure supplement is available for figure 3: Genes assigned to mitochondrial function showed decreased acetylation of H3K9, which implied reduced mitochondrial function. This assumption was confirmed by measuring the respiratory activity of fungal cells. In monoculture, the fungus showed a high metabolic activity, which was significantly reduced during co-cultivation ( Figure 3c).

Bacteria induce elements of the fungal cross-pathway control
To identify transcription factors that are involved in transducing the bacterial signal to the fungal expression machinery, and because a transcription factor gene is missing in the ors gene cluster, we searched the 890 differentially H3K9 acetylated genes for those annotated as putatively involved in transcriptional regulation. In total, 22 putative transcription factor-encoding genes fulfilled this requirement (Supplementary file 2, section VII). Most of them (18 genes) showed significantly higher acetylation in co-culture, whereas only four genes had lower acetylation. Among the genes with increased acetylation in co-culture were cpcA, coding for the central transcriptional activator of the cross-pathway control CpcA, as well as the bZIP transcription factor gene jlbA (jun-like bZIP). Both of these genes have been shown to be highly expressed during amino-acid starvation in A. nidulans (Hoffmann et al., 2001;Strittmatter et al., 2001). In addition, a putative ortholog (AN7174) of the S. cerevisiae bas1 gene showed an increase in acetylation. In yeast, Bas1p (together with the homeodomain protein Bas2p) is involved in the regulation of amino-acid biosynthesis (Springer et al., 1996;Valerius et al., 2003). Consistently, a number of genes related to amino-acid metabolism showed increased acetylation of H3K9 during the co-cultivation of A. nidulans with S. rapamycinicus (Supplementary file 2, section II). qRT-PCR analysis was carried out to correlate the ChIP-seq data with the expression levels of cpcA, jlbA and AN7174 , and this demonstrated upregulation of cpcA and AN7174 during co-cultivation ( Figure 4a). In S. cerevisiae, it was shown that Gcn4 (CpcA in A. nidulans) and Bas1p share a similar DNA-binding motif and that both activate the transcription of the histidine biosynthesis gene HIS7 independently of each other (Springer et al., 1996). Consistent with a possible involvement of these transcription factors in cross-pathway control (CPC) is the observation that the addition of the histidine analogue 3-aminotriazole (3-AT), which is known to induce the CPC via amino-acid starvation, led to the production of orsellinic acid in the fungal monoculture ( Figure 4b) and to an increased expression of orsA, cpcA and AN7174 ( Figure 4c).
To analyze a possible involvement of these genes in the bacteria-induced activation of the ors gene cluster, the genes cpcA (data not shown) and AN7174 (Figure 5-figure supplement 1a) were deleted. Deletion of cpcA in A. nidulans showed no effect on the induction of the ors gene cluster in response to S. rapamycinicus (data not shown), whereas deletion of AN7174 resulted in a significantly reduced expression of orsA and orsD, and in complete loss of orsellinic acid production (Figure 5). Therefore, AN7174 was named basR and analyzed in detail.

The transcription factor BasR is a central regulatory node in bacteriatriggered regulation of the SM gene cluster
Further analysis of the A. nidulans genome revealed a second gene (AN8377) encoding a putative ortholog of the S. cerevisiae bas1 gene ( Figure 6 and Figure 6-figure supplement 1). Both genes (basR and AN8377) code for Myb-like transcription factors whose function in filamentous fungi is completely unknown. We compared the H3K9 acetylation and gene expression of both genes upon co-cultivation. The basR gene showed increased H3K9 acetylation (LFC = 0.6) and drastically increased transcription (LFC = 5.85) during co-cultivation compared to AN8377 (H3K9ac In S. cerevisiae, Bas1p needs the interaction with Bas2p for the transcriptional activation of several genes that are required for histidine and purine biosynthesis (Springer et al., 1996). The C-terminal activation and regulatory (BIRD) domain of Bas1, which was described as mediating this Bas1p-Bas2p interaction (Pinson et al., 2000), is missing in BasR. It is thus not surprising that we did not find an ortholog for the S. cerevisiae bas2 gene in the A. nidulans genome. Although the addition of 3-AT to monocultures of A. nidulans led to the production of orsellinic acid and derivatives thereof, the effect of 3-AT was abolished in the basR deletion mutant strain ( Figure 4b).
The transcriptional activation of HIS7 by Bas1/Bas2 upon adenine limitation in yeast requires a functional Gcn5 (GcnE in A. nidulans) (Valerius et al., 2003), so we raised the question of whether GcnE is needed for full basR expression. Addition of S. rapamycinicus or 3-AT to the gcnE deletion mutant led to decreased basR gene expression compared to levels of gene expression seen in the wild type in co-culture or in a monoculture with 3-AT ( Figure 4c). These data indicate that GcnE is required for basR expression. Inspection of the basR mutant strain on agar plates did not reveal further obvious phenotypes (data not shown).
To further substantiate the influence of basR on the ors gene cluster, we generated a basR overexpression strain ( Figure 5-figure supplement 1b) by employing the inducible tet On -system (Helmschrott et al., 2013). Addition of doxycyline to the media induced basR expression as well as the expression of the ors gene cluster (Figure 5a). However, basR gene expression was detectable even without doxycycline addition, indicating 'leakiness' of the tet On -system. Nevertheless, production of orsellinic and lecanoric acid was only detected upon doxycycline addition (Figure 5b), supporting the important role of BasR for their biosynthesis. To address the question of whether other SM biosynthesis gene clusters are regulated by BasR, we performed RNA-sequencing (RNA-seq) analysis. We examined the transcription profiles of A. nidulans with and without S. rapamycinicus, and compared it to those of the basR overexpression strain with and without the addition of doxycycline. Obvious candidate gene clusters, regulated by BasR, were the eight differentially acetylated gene clusters found in the ChIP-seq data (Supplementary file 2). Five of the eight differentially acetylated SM gene clusters, namely the dba, cic, eas and microperfuranone gene clusters, were also differentially transcribed in response to the streptomycete as well as in the basR overexpression strain (Figure 7a; Supplementary file 3), emphasizing the importance of BasR in bacteria-induced secondary metabolite regulation. In addition, this finding was perfectly mirrored when we applied matrix assisted laser desorption/ionization (MALDI)-mass spectrometry (MS) imaging, which showed reduced levels of emericellamides both in basR-overproducing colonies of A. nidulans and in cogrown colonies, but not in colonies without the streptomycete or doxycycline addition (Figure 7b).
Interestingly the microperfuranone gene cluster, which is acetylated at lower levels and transcribed in response to the bacterium, is transcriptionally upregulated in the basR-overexpression strain, suggesting a transcriptional regulation that is independent of the signal(s) induced by S. rapamycinicus.
The presence of BasR in fungal species makes it possible to forcast the inducibility of ors-like gene clusters by S. rapamycinicus To address the question of whether basR homologs exist in other fungi and whether such potential homologs have similar functions, we analyzed fungal genomes using BlastP. Surprisingly, obvious basR homologs are only present in a few other Aspergillus spp. including Aspergillus sydowii and Aspergillus versicolor, and are apparently lacking in many others ( Figure 6 and Figure 6-figure supplement 1). Interestingly, in addition to three additional genes in both fungi, a gene cluster similar to the ors gene cluster of A. nidulans was also identified (Figure 6b). We overexpressed basR in A. sydowii using the tet On -system to analyze its function ( Figure 6-figure supplement 3). LC-MS analyses revealed the appearance of novel masses that were assigned to orsellinic acid derivatives ( Figure 6c).
Finally, we addressed the question of whether the presence of the basR gene and the ors gene cluster allows the forecasting of their inducibility by S. rapamycinicus. As shown in Figure 6, also cocultivation of A. sydowii with S. rapamcinicus led to the activation of the fungal ors gene cluster, again linking BasR with the bacteria-triggered induction of the production of orsellinic acid derivatives.

S. rapamycinicus induces a unique chromatin landscape in A. nidulans
We were able to use genome-wide ChIP-seq analysis of acetylated histone H3 (H3K9ac, H3K14ac) and the quantification of H3 to uncover the chromatin landscape in the fungus A. nidulans upon cocultivation with S. rapamycinicus. In an attempt to characterize the general distribution of nucleosomes and acetylation marks over the genome, we compared the intensity of chromatin states with gene density. A lower gene density was typically found in heterochromatic regions such as the centromeres and telomeres, creating a repressing environment (Allshire and Ekwall, 2015). We found reduced H3 occupancy in heterochromatic regions, indicating either replacement of H3 by the centromere-specific H3 CENP-A or reduced nucleosome occupancy (Smith et al., 2011;Allshire and Ekwall, 2015).
We observed distinct peaks for H3K9ac in A. nidulans grown in co-culture with S. rapamycinicus. One of the areas with the greatest increase in H3K9ac was the ors gene cluster, nicely confirming our previous findings (Nützmann et al., 2011) ( Figure 8). Furthermore, previous ChIP qRT-PCR experiments indicated a distinct increase of H3K9ac inside the cluster borders, which did not expand to neighboring genes (Nützmann et al., 2011). By contrast, the H3K14ac modification seemed to be of a more global nature and not exclusively confined to specific regions such as the ors gene cluster. These conclusions were extended here by the pattern detected in the genome-wide ChIP-seq data, which showed no spreading of H3K9ac to genes adjacent to the ors gene cluster, thereby demonstrating the quality of the genome-wide ChIP data generated here. Furthermore, these results are also consistent with our previous finding of reduced expression of SM cluster genes as a consequence of the lack of H3K14 acetylation (Nützmann et al., 2013). In contrast to H3K14ac, H3K9ac is less uniformly distributed over the genome. It only showed strong enrichment in the promoters of certain genes. Especially high levels of acetylation were found at orsA and in the bidirectional promoters of orsD and orsE. This observation was recently confirmed by the finding that H3ac and H3K4me3 were increased at the orsD gene only when the ors cluster was transcriptionally active (Gacek-Matthews et al., 2016).
We also assessed the distribution of H3K9ac and H3K14ac, as well as that of the C-terminus of H3 (H3Cterm), at the TSSs and translation termination sites (TTSs) (Appendix 1 -Chromatin profiles at translation start sites and translation termination sites). For H3K9, an enrichment of acetylation~500 bp downstream of the TSSs as well as immediately upstream of the TSSs was observed. This was expected as similar results were obtained with an antibody targeting the acetylated N-terminus of histone H3 in A. nidulans (Gacek-Matthews et al., 2016) and in other fungi such as S. cerevisiae and Cryptococcus neoformans (Haynes et al., 2011;Mews et al., 2014). Increased acetylation coincides with reduced levels of H3 around the TSSs, which are most probably due to a depletion of nucleosomes at the promoter. The profile plots for H3K14 acetylation are similar, although not as highly enriched around the TSSs as those for H3K9 (Appendix 1-figure 5 a). As expected, a comparison of LFCs for both modifications showed high similarity, suggesting that the modifcations are established interdependently (Gacek and Strauss, 2012;Waters et al., 2015). At the 3' end of the ORF, H3 density drastically increased accompanied by reduced levels of H3K9ac and H3K14ac (Appendix 1- figure 5). Likewise, reduced acetylation at the TTSs was observed in A. nidulans (Gacek-Matthews et al., 2016) and S. cerevisiae (Mews et al., 2014). It is interesting to notice that the increase in nucleosome density directly correlated with a decrease in the gene expression rate  1-figure 6e). Previous studies suggested a direct correlation between the presence of nucleosomes and the stalling of RNA polymerase II (Grosso et al., 2012).

Increased gene expression directly correlates with histone H3K9 acetylation
Acetylation is generally regarded as an activating chromatin mark that promotes the transcription of eukaryotic genes (Bannister and Kouzarides, 2011). Our study suggests a more differentiated picture. When we compared data from this study with microarray data (Nützmann et al., 2011) (Appendix 1-figures 3 and 4), the acetylation of H3K9 directly correlated with gene expression levels. A similar finding was reported for other fungi (Wiemann et al., 2013). By contrast, this was not observed for the acetylation of H3K14. This could partly result from the low number of targets for this modification. By contrast, gene promoters showed a distinct increase of H3K14ac at the TSSs in dependence on the average transcription level (Appendix 1-figure 6c). The low correlation between active gene transcription and acetylation at H3K14 confirmed earlier results (Reyes-Dominguez et al., 2008;Nützmann et al., 2011). Previously, we showed that a mimicry of a hypoacetylated lysine 14 on histone H3 drastically altered the phenotype and the expression of SM gene clusters (Nützmann et al., 2013). This effect was overcome, however, when later time points of cultivation were considered. Taken together, the primary location at the TSSs and the major defect in SM production at earlier stages indicate a role for H3K14ac in transcriptional initiation. Hyper-acetylation at H3K14 could be also relevant for marking active genes and providing a docking site for regulatory proteins.

S. rapamycinicus silences fungal nitrogen metabolism
A substantial number of genes that are involved in primary and secondary nitrogen metabolism were strongly depleted for H3K9ac upon co-cultivation with S. rapamycinicus. This correlated with reduced expression of the respective genes. Thus, upon contact with the bacterium, A. nidulans showed reduced nitrogen uptake and reduced degradation of various nitrogen sources, leading to nitrogen starvation.
Under nitrogen starvation or low availability of primary nitrogen sources, such as glutamine and ammonium, the intracellular level of glutamine drops (Tudzynski, 2014). This was in fact observed for the intracellular concentration of amino acids in A. nidulans when the fungus was co-cultured with the bacterium (Appendix- figure 7). Thus, in presence of S. rapamycinicus but not of noninducing streptomycetes such as S. lividans the fungus is in a physiological state of nitrogen starvation ( Figure 8). Nitrogen limitation has been shown before to represent a trigger for the activation of a number of SM gene clusters including the ors gene cluster Studt et al., 2012). Nitrogen starvation also activates the expression of the anthrone (mdp) gene cluster , which we also observed in our data. However, induction of orsellinic acid production by nitrogen starvation took about 60 hr, whereas co-cultivation with S. rapamycinicus had already triggered expression of the cluster genes after 3 hr. Therefore, it is unlikely that the bacteria-triggered activation of the cluster is exclusively achieved by restricting nitrogen availability for the fungus. Furthermore, shortage of nitrogen leads to de-repression of genes that are involved in the usage of secondary nitrogen sources, which was not supported by our data. In S. cerevisiae, it has been reported that a shift from growth under nutrient sufficiency to nitrogen starvation induced the degradation of mitochondria (Eiyama et al., 2013). Similarly, decreased acetylation and transcription of genes with mitochondrial function were also detected upon contact of A. nidulans with the bacterium. This was further supported by a lower mitochondrial respiratory activity in the fungal cells during co-cultivation (Figure 3c).

BasR is a central regulatory node for integrating bacterial signals leading to regulation of SM gene clusters
Another consequence of nitrogen starvation is the reduced availability of amino acids in the cell. Consequently, as shown here, the amino-acid biosynthetic pathways represented a major group of de-regulated genes at both the acetylation and expression levels. Amino-acid biosyntheses in fungi are regulated by the CPC system upon starvation for distinct amino acids (Tudzynski, 2014;Krappmann and Braus, 2005). Since deletion of cpcA in A. nidulans did not affect the induction of the ors gene cluster, whereas the artificial inducer of the CPC system 3-AT does (Sachs, 1996), it is conceivable that CPC somehow plays a role. 3-AT is a structural analogue of histidine that triggers histidine starvation in the fungal cell and thereby the CPC (Sachs, 1996). In S. cerevisiae, other regulators such as the heterodimeric transcription factor complex Bas1p/Bas2p, which is even bound by Gcn5p, have also been shown to induce the CPC (Valerius et al., 2003;Daignan-Fornier and Fink, 1992). We identified two putative orthologous genes in the genome of A. nidulans, but further analysis revealed that only basR (AN7174) was involved in ors gene cluster activation during the fungalbacterial co-cultivation (Figure 8). Despite the fact that AN8377 seems to resemble S. cerevisiae bas1 more closely (Figure 7 and Figure 6-figure supplement 1), it is not needed for the ors gene cluster activation.
On the basis of bioinformatic analysis, BasR of A. nidulans consists of 305 amino acids and thus is rather different from its closest homolog which is Bas1p of S. cerevisiae with 811 amino acids (Zhang et al., 1997). The BIRD region of Bas1p that mediates the Bas1p-Bas2p interaction (Pinson et al., 2000) is missing in BasR. The basR gene was highly upregulated in the microarray data, and the upregulation of this gene coincided with the increased H3K9 acetylation of its promoter. basR deletion and overexpression clearly demonstrated the function of this transcription factor gene in activating the ors gene cluster in response to S. rapamycinicus. A functional GcnE seems to be required for efficient basR expression, indicating a dependency similar to that observed for bas1 in yeast (Valerius et al., 2003).
Interestingly, the basR gene could not be found in all of the fungal genomes analyzed here but it was found, for example, in A. sydowii and A. versicolor, which were also found to encode ors gene clusters. As in A. nidulans, overexpression of the A. sydowii basR gene led to the activation of its silent ors gene cluster. On the basis of this finding, we predicted that S. rapamycinicus also induces the ors gene cluster in A. sydowii an this was indeed the case. We did not find a basR homolog in A. fumigatus, although the formation of fumicyclines is induced by S. rapamycinicus (Kö nig et al., 2013). This might be due to the fact that the available genome data lack the basR gene due to missing annotation or, alternatively, because a different regulatory response mechanism to S. rapamycinicus is present in A. fumigatus.
Genome-wide ChIP-seq analysis also indicated that the interaction of S. rapamycinicus with A. nidulans influenced other SM gene clusters and leads to a downregulation of the fungal nitrogen metabolism (Figure 3 and Supplementary file 2), which might be regulated via BasR. Further analyses revealed that BasR is also required for the transcriptional regulation of the dba, cic, microperfuranone and eas gene clusters (Figure 7), as well as being important for the downregulation of genes belonging to the nitrate-assimilation gene cluster (Supplementary file 3). These data indicate that overexpression of basR phenocopies the regulation by S. rapamycinicus and highlights the importance of BasR for the regulation of SM gene clusters and its role in transducing the bacterial signal(s) in the fungus. As implied by the finding that the presence of basR and the ors cluster in several fungi coincided with their inducibility by S. rapamycinicus, in future it might be possible to predict which microorganisms communicate with each other based on their genetic inventory.

Microorganisms, media and cultivation
A. nidulans strains were cultivated in Aspergillus minimal medium (AMM) at 37˚C, 200 rpm (Brakhage and Van den Brulle, 1995). When required, supplements were added as follows: arginine (871 mg/ml), p-aminobenzoic acid (3 mg/ml) and pyridoxine HCl (5 mg/ml). Pre-cultures were inoculated with 4 Â 10 8 spores per ml. 10 mg/ml doxycycline was used to induce the tet On -inducible system. A. nidulans gcnE-3xflag strain was used for ChIP-seq analysis. For the measurement of orsellinic acid, mycelia of overnight cultures (~16 hr) in AMM were transferred to fresh medium and inoculated with S. rapamycinicus, as previously described (Schroeckh et al., 2009). RNA extraction for expression analysis during co-cultivation was performed after 3 hr of cultivation; for analysis of the basRoverexpression mutant after 6 hr of monoculture, samples for HPLC analysis were taken after 24 hr. A. sydowii was cultivated at 28˚C, 200 rpm in malt medium (Scherlach et al., 2010). For the induction of the ors cluster in A. sydowii, 48-hr-old precultures were transferred to fresh AMM and inoculated with S. rapamycinicus or doxycycline. 10 mg/ml doxycycline was added twice over the course of 48 hr. Samples were taken for LC-MS analysis after 96 hr for A. sydowii co-cultivation and after 48 hr for the A. sydowii basR-overexpression mutant. For MALDI-MS Imaging analysis, conductive ITO slides (Bruker Daltonics, Bremen, Germany) were coated with 3 ml 0.5% (w/v) AMM agar and incubated at room temperature for 30 min (Aiyar et al., 2017;Araújo et al., 2017). Identical conditions were ensured by supplementation of all slides with arginine regardless of the fungal genotype. S. rapamycinicus was applied by filling 5 ml of a preculture in a tube and by point inoculation of 15 ml of the settled mycelium on the agar. For A. nidulans, 500 conidia of wild type and mutants were point inoculated onto the agar. For co-cultivation experiments, both microorganisms were inoculated 1 cm apart from each other. The slides were incubated at 37˚C in a Petri dish for 4 days. The slides were dried by incubation in a hybridization oven at 37˚C for 48 hr.

Quantitative RT-PCR (qRT-PCR)
Total RNA was purified with the Universal RNA Purification Kit (Roboklon, Berlin, Germany). Reverse transcription of 5 mg RNA was performed with RevertAid Reverse Transcriptase (Thermo Fisher Scientific, Darmstadt, Germany) for 3 hr at 46˚C. qRT-PCR was performed as described before (Schroeckh et al., 2009). The A. nidulans b-actin gene (AN6542) served as an internal standard for calculation of expression levels as previously described (Schroeckh et al., 2009). Primers for the amplification of probes are listed in Supplementary file 4.

ChIP coupled to quantitative RT-PCR (qRT-PCR)
Cultures were grown as described in the cultivation section. After 3 hr, the isolated DNA was crosslinked to proteins as described before (Boedi et al., 2012). Powdered mycelium was dissolved in 1 ml of sonication buffer (Boedi et al., 2012) and 330 mL aliquots were then subjected to sonication for 30 min with cycles of 2 min maximum intensity followed by a 1 min pause. Sheared chromatin was separated from cell wall debris and incubated with 40 mL of a protein A slurry for 30 min at 4˚C on a rotary shaker. A purified 1:10 dilution of the supernatant was then incubated overnight at 4˚C with 3 mL of antibody directed against the desired target. Antibodies were precipitated with 40 mL of Dynabeads (Invitrogen, Carlsbad, USA) and were immediately incubated with the sample for 40 min at 4˚C on a rotary shaker. Samples were washed three times with low salt buffer followed by washing once with high-salt buffer (Boedi et al., 2012). Washed beads were dissolved in 125 ml TES buffer and reverse cross-linked with 2 mL of 0.5 M EDTA, 4 mL of 1 M Tris-HCl pH 6.5 and 2 mL of 1 mg/ml proteinase K for 1 hr at 45˚C. Subsequent DNA purification was conducted with a PCR purification kit and samples were eluted in 100 mL of 1:10 diluted elution buffer. The DNA concentration of genes of interest was quantified using qRT-PCR as described above. The antibodies used are the following: mouse monoclonal ANTIFLAG M2 (Sigma-Aldrich, F3165-5MG, Taufkirchen, Germany), rabbit polyclonal anti-histone H3 (Abcam ab1791, Cambridge, UK), rabbit polyclonal histone H3K9ac (39137, Active Motif, La Hulpe, Belgium)) and rabbit polyclonal anti-acetyl-histone H3 (Lys14) (07-353, Merck Millipore, Darmstadt, Germany).
The samples of A. sydowii were loaded onto an ultrahigh-performance liquid chromatography (LC)-MS system consisting of an UltiMate 3000 binary rapid-separation liquid chromatograph with photodiode array detector (Thermo Fisher Scientific, Dreieich, Germany) and an LTQ XL linear ion trap mass spectrometer (Thermo Fisher Scientific, Dreieich, Germany) equipped with an electrospray ion source. The extracts (injection volume, 10 mL) were analyzed on a 150 x 4.6 mm Accucore reversed-phase (RP)-MS column with a particle size of 2.6 mm (Thermo Fisher Scientific, Dreieich, Germany) at a flow rate of 1 ml/min, with the following gradient over 21 min: initial 0.1% (v/v) HCOOH-MeCN/0.1% (v/v) HCOOH-H 2 O 0/100, which was increased to 80/20 in 15 min and then to 100/0 in 2 min, held at 100/0 for 2 min, and reversed to 0/100 in 2 min.
Identification of metabolites was achieved by comparison with an authentic reference. Samples were quantified via integration of the peak area using Shimadzu Class-VP software (version 6.14 SP1).

MALDI-MS imaging analysis and data processing
Sample preparation and matrix coating were performed as previously described (Aiyar et al., 2017). Samples were analyzed (Aiyar et al., 2017) in an UltrafleXtreme MALDI TOF/TOF (Bruker Daltonics, Bremen, Germany), in reflector positive mode with the following modifications: 100 -3000 Da range, 30% laser intensity (laser type 4) and raster width 200 mm. The experiments were repeated three times (2 nd and 3 rd replicates with 250 mm raster width). Calibration of the acquisition method, spectra procession, visualization, analysis and illustration were performed as described before (Aiyar et al., 2017). Chemical images were obtained using Median normalization and weak denoising.

Resazurin assay
Respiratory activity was measured by reduction of resazurin to the fluorescent dye resorufin. 10 4 conidia of A. nidulans in 100 mL AMM were pipetted into each well of a black 96-well plate. The plate was incubated for 16 hr at 37˚C. The pre-grown fungal mycelium was further cultivated in monoculture or with 10 mL of an S. rapamycinicus culture. Cultures were further supplemented with 100 mL of AMM containing resazurin in a final concentration of 0.02 mg/ml. Fluorescence was measured (absorption wavelength 560 nm, emission wavelength 590 nm) every 30 min for 24 hr at 37˚C in a Tecan fluorometer (Infinite M200 PRO, Mä nnedorf, Switzerland). For all conditions, measurements were carried out in triplicates for each of the two biological replicates. The significance of values was calculated using a two-way ANOVA Test with GraphPad Prism 5 (GraphPad Software Inc., La Jolla, USA).

ChIP-seq pre-processing
The A. nidulans FGSC A4 genome and annotation (version s10-m03-r28) were obtained from the Aspergillus Genome Database (AspGD) (Cerqueira et al., 2014). The S. rapamycinicus NRRL 5491 genome was obtained from NCBI (GI 521353217). Both genomes were concatenated to a fused genome which served as the reference genome for subsequent mapping. Raw ChIP-seq reads were obtained using FastQC v0.11.4. Trimming and filtering were achieved by applying Trim Galore utilizing Illumina universal adapter and phred +33 encoding. Reads were not de-duplicated because the duplication rate was <15% for most libraries. Bowtie2 (version 2.2.4) using default parameters was employed to map reads to the fused genome. Quantification of reads was carried out using the Bioconductor 'GenomicAlignments' package forming the basis for three subsequent approaches. First, a genome-wide equi-spaced binning across the genome with different resolutions (50 k and 2 k bp bins) counting reads overlapping each bin was applied. Library normalization on bin counts was performed by only considering reads mapping to the A. nidulans genome. Second, reads overlapping genes were counted, using the AspGD (Cerqueira et al., 2014) annotation. They formed the basis for the subsequent DCS analysis (see below). Third, average profile plots to assess relative histone distributions around TSS and TTS were generated using the bioconductor package regioneR (Gel et al., 2016).

DCS analysis
To identify genes exhibiting differences in their chromatin state, we employed the bioconductor package edgeR (Robinson and Oshlack, 2010) originally developed for RNA-seq differential expression analysis. The ChIP-seq data follow the same pattern, that is, negative binomial distribution of reads. Library normalization was achieved with the trimmed mean of M values method (Robinson and Oshlack, 2010) based only on A. nidulans gene counts for calculating the effective library sizes, not taking into account reads mapping to S. rapamycinicus which would otherwise artificially influence the effective library size. Comparisons were made between libraries for all ChIP targets separately obtained from monocultures of A. nidulans and co-cultures with S. rapamycinicus. These targets were H3, H3K9ac and H3K14ac. Results including normalized read counts (RPKM), statistics and LFCs are reported in Supplementary file 1. Normalized counts and LFCs were also further used for comparisons with the corresponding microarray-based gene expression and the calculated LFCs.

MACS analysis
Candidate peaks were identified using two methods: a differential binding analysis (EdgeR) and a peak-calling approach (MACS, version 2.0.1) (Zhang et al., 2008). The peak caller performed several pairwise comparisons between samples with the same antibody and between different conditions in order to retrieve the peaks with significant change of ChIP signal indicating differential binding for that particular comparison. The program kept the track of different replicates, the signal was reported per million reads and produced a BED format track of the enriched regions, other parameters were used with default values. The BED files were subsequently converted to Big Wig format for visualization through the tool Integrative Genomics Viewer (Thorvaldsdó ttir et al., 2013).

Generation of A. nidulans deletion strains
The transformation cassettes for the basR and AN8377 deletion strains were constructed as previously described (Szewczyk et al., 2007). Approximately~1000 bp sequences homologous to the regions upstream and downstream of basR and AN8377 were amplified and fused to the argB deletion cassette (Schroeckh et al., 2009). Transformation of A. nidulans was carried out as described before (Ballance and Turner, 1985).

Generation of inducible A. nidulans and A. sydowii basR-overexpressing strains
For overexpression of basR, the tetracycline-controlled transcriptional activation system (tet On ) was used (Helmschrott et al., 2013). The basR gene sequences together with their~1000 bp flanking regions were amplified from A. nidulans and A. sydowii genomic DNA. The tet On -system was amplified from plasmid pSK562. All DNA fragments were assembled using NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs, Frankfurt, Germany). The A. nidulans pabaA1 gene was used as a selectable marker to complement the p-aminobenzoic acid auxotrophy of the A. nidulans DbasR mutant. For A. sydowii, the Aspergillus oryzae hph cassette was used as the selectable marker. 200 mg/ml hygromycin (Invivogen, Toulouse, France) was used for selection of transformant strains.

Phylogenetic analysis
The amino-acid sequences for the two Myb-like transcription factors from A. nidulans (AN7174 (basR) and AN8377) and Bas1 from S. cerevisiae were used for a Blast search in the UniProtKB database. For each sequence, the first 50 hits were retrieved. All hits were grouped together, and redundant and partial sequences removed. The obtained 54 hits were first aligned using MUSCLE (Edgar, 2004). The phylogenetic tree was obtained using the Maximum Likelihood method contained in the MEGA6 software facilities (Tamura et al., 2013).

Measurement of amino acids
Amino acids were extracted from 10 mg samples with 1 ml of methanol and the resulting extract was diluted in a ratio of 1:10 (v:v) in water containing the 13 C, 15 N labeled amino-acid mix (Isotec, Miamisburg, Ohio, USA). Amino acids in the diluted extracts were directly analyzed by LC-MS/MS as described, but with the modification that an API5000 mass spectrometer (Applied Biosystems, Foster City, California, USA) was used (Docimo et al., 2012).

cDNA library construction and sequencing
Total RNA was isolated as described for qRT-PCR analysis, from three replicates of A. nidulans cultivated with and without S. rapamycinicus and from the inducible A. nidulans basR-overexpression strain with and without the addition of doxycycline. Samples were taken after six hours of co-cultivation or addition of doxycycline to induce the basR overexpression. Total RNA from the three replicates was pooled and 9 mg of RNA were processed for the library preparation. Library construction, Illumina next-generation sequencing, and the mapping and normalizing of the transcript reads were performed by StarSEQ GmbH (Mainz, Germany). Transcript levels were normalized by counting the number of transcripts per million (TPM) (Wagner et al., 2012).

Availability of data and materials
ChIP-seq data were deposited in the ArrayExpress database at EMBL-EBI (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-5819. The code for data processing and analysis can be obtained from https://github.com/seb-mueller/ChIP-Seq_Anidulans (Müller, 2018; copy archived at https://github.com/elifesciences-publications/ChIP-Seq_Anidulans). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.  1-figure 4). Since the used bin size is large, this could point at global H3 occupancy to be higher for high gene density regions such as euchromatin and low H3 occupancy for heterochromatin. Noteworthy, the correlation between read and gene density was found to be lower for H3K14ac and H3K9ac (r=0.14 and 0.15 respectively), which might indicate a more subtle regulatory mechanism for those marks targeting individual genes as opposed to larger domains. Notably, the highest correlation was found between the two acetylation marks (r=0.53) hinting at some potential cross-talk or common regulation between them (Appendix 1-figure 4).

Chromatin profiles at translation start sites and translation termination sites
We assessed the location of H3K9ac, H3K14ac and histone H3 relative to promoters and gene bodies by plotting the average read count frequency for all genes to either the TSSs or TTSs (Appendix 1-figure 1) (Yu et al., 2015). Due to missing information about the 5' and 3' transcriptional start and stop sites, we used translational start and stop sites for transcription start and stop sites, respectively, as surrogates. The results obtained apply to both A. nidulans in mono-and co-culture with the bacterium. According to Kaplan et al. (2010), the peaks correspond to highly positioned nucleosomes relative to the TSSs with a nucleosome-free region (NFR) directly upstream of the TSS. H3K9ac and H3K14ac showed highest enrichment for the first and second nucleosomes up-and downstream of the TSS and drastic reduction downstream of the third nucleosome after the TSS (Appendix 1-figure 1). Reduced occupancy of unmodified histone H3 was observed at the TSS (Kaplan et al., 2010). Towards the 3' end of genes, histone H3 occupancy gradually increased, which was accompanied by a decrease in acetylation. Plotting of differentially acetylated H3K9ac against H3K14ac showed a strong correlation between the localization of the two modifications ( Figure 2-figure  supplement 2a). Acetylation is generally described as a transcription activating mark (Gacek and Strauss, 2012). To test for this general assumption we correlated our acetylation data to microarray data generated under the same condition (Nützmann et al., 2011) ( Figure 2-figure supplement 2b). This allowed us to compare the log-fold changes (LFCs) of the differential chromatin states with the LFCs calculated from the RNA expression data. H3K9ac correlates with the differentially expressed genes with a coefficient of 0.5 in contrast to H3K14ac (-0.05) and histone H3 (-0.01) for which no detectable dependency was observed (Appendix 1- figure 3). Similarly, Appendix 1- figure 3 shows the same trend, i.e., a correlation of 0.2 for gene expression changes versus H3K9ac changes. This finding was expected since the calculation included all genes most of which did not change. To determine the correlation of acetylation at the TSS and the TTS according to the grade of expression of the genes, we separated the differentially expressed genes into four quartiles (q1 lower 25 %, q2 the medium lower 25-50 %, q3 are the medium higher 50-75 %, q4 higher 25 %). The increase of acetylation at the TSS correlated with the expression level of genes (Appendix 1figure 6a and c). Decreased expression coincided with an increase of histone H3 at the TSS (Appendix 1-figure 6e).

Reduced intracellular amino acid concentration in response to the bacterium
The increased expression of cpcA and other genes involved in amino acid biosynthesis implied a reduced availability of amino acids in the cell upon bacterial-fungal co-cultivation. Therefore, we measured the internal amino acid pool in A. nidulans both grown in monoculture and with S. rapamycinicus (Appendix 1- figure 7). As an additional control and to further confirm the specificity of the interaction, the fungus was co-cultivated with Streptomyces lividans, which does not induce the ors gene cluster. As shown in Appendix 1-figure 7, significantly reduced levels of glutamine, histidine, phenylalanine, asparagine, threonine and reduced metabolism of arginine, which was supplemented to the medium, were observed. The monoculture of A. nidulans, co-cultivation of A. nidulans with S. lividans as well as the addition of S. rapamycinicus after 24 hours of fungal cultivation served as controls. All of the controls showed comparable amino acid levels.