Abstract The frequency of antifungal resistance, particularly to the azole class of ergosterol biosynthetic inhibitors, is a growing global health problem. Survival rates for those infected with resistant isolates are exceptionally low. Beyond modification of the drug target, our understanding of the molecular basis of azole resistance in the fungal pathogen Aspergillus fumigatus is limited. We reasoned that clinically relevant antifungal resistance could derive from transcriptional rewiring, promoting drug resistance without concomitant reductions in pathogenicity. Here we report a genome-wide annotation of transcriptional regulators in A. fumigatus and construction of a library of 484 transcription factor null mutants. We identify 12 regulators that have a demonstrable role in itraconazole susceptibility and show that loss of the negative cofactor 2 complex leads to resistance, not only to the azoles but also the salvage therapeutics amphotericin B and terbinafine without significantly affecting pathogenicity. Subject terms: Antifungal agents, Antimicrobial resistance, Fungal genomics, Fungal pathogenesis __________________________________________________________________ Resistance to primary treatments of invasive aspergillosis is growing. Here, the authors generate a knockout library for 484 transcription factors in Aspergillus fumigatus, and show that loss of the NCT complex leads to cross-resistance to all primary and some salvage therapeutics without affecting pathogenicity. Introduction Aspergillus fumigatus is the most important airborne mould pathogen and allergen worldwide. Estimates suggest that over 3 million people have invasive or chronic infections that lead to in excess of 600,000 deaths every year^[68]1. Only three classes of drugs are currently recommended for the treatment of aspergillosis with the azole class being recommended for primary therapeutic purposes, and amphotericin B and the echinocandins (caspofungin and micafungin) for salvage therapy. With effective treatment, mortality rates for invasive disease remain ~50%^[69]2. It is of great concern, however, that drug resistance to the azoles is rapidly emerging. For individuals that are infected with a resistant isolate the mortality rate exceeds 80%^[70]3,[71]4. As disease progression is so rapid in invasive aspergillosis (IA)^[72]5, therapy failure is attributable to delays in administering alternative agents. Methods to rapidly detect resistance are critical to ensure effective transition to alternative appropriate therapies. Our understanding of the factors governing azole resistance in A. fumigatus is not fully defined. The azoles act by inhibiting the hemoprotein lanosterol demethylase (Cyp51A) resulting in a reduction of the key sterol, ergosterol, in the fungal membrane and an accumulation of toxic sterol intermediates^[73]6. Azole resistance is frequently associated with an allelic variant of cyp51A that incorporates a tandem repeat in the promoter, typically TR[34] or TR[46], with a secondary non-synonymous mutation within its coding sequence^[74]7. These modifications appear to have no substantial impact on pathogenicity in murine models of invasive aspergillosis^[75]8. The mechanism of resistance in a significant proportion of other isolates remains unclear^[76]7. This hinders the development of rapid molecular diagnostics to detect drug resistance in the clinic. Our limited understanding of the mechanisms of azole resistance also prevents the development of combination therapeutic strategies that specifically target drug-resistance mechanisms^[77]9. We and others have recently reported on the role played by various A. fumigatus transcriptional regulators in response to azole antifungal drugs^[78]10–[79]12. The sterol regulatory element-binding protein (SREBP), SrbA is a basic helix–loop–helix (bHLH) transcriptional activator which directly regulates at least seven genes in the ergosterol biosynthetic pathway, including cyp51A^[80]10. Loss of srbA through gene replacement results in a significant increase in susceptibility to azoles^[81]13 in part due to significant reductions in cyp51A mRNA levels^[82]14. The binding site for SrbA in the cyp51A promoter falls within the 34 and 46 mers duplicated in TR34/46 pandemic azole-resistant isolates^[83]12. The repeat duplicates the DNA-binding site leading to SrbA-mediated upregulation of cyp51A and a concomitant increase in azole resistance. AtrR, a Zn2-Cys6 transcription factor, also positively regulates sterol biosynthesis and directly binds the cyp51A promoter at the TR site, and additionally the promoter of an azole exporter, cdr1B^[84]11,[85]15. The CCAAT-binding domain complex CBC, a heterotrimer comprising HapB, HapC and HapE, is a negative regulator of sterol biosynthesis directly binding the promoters of 14 ergosterol biosynthetic genes, including cyp51A^[86]12. Loss of CBC function leads to pan-azole resistance. Notably, a clinical azole-resistant isolate with a defect in HapE that results in perturbed DNA binding at the cyp51A promoter has been described^[87]12,[88]16. Binding of the CBC at the cyp51A promoter is facilitated by another transcriptional regulator, HapX^[89]12. HapX is an iron-responsive basic region leucine zipper (bZIP) transcription factor that regulates the expression of genes linked to iron acquisition, storage and metabolism and facilitates binding of the CBC^[90]17. It is also notable that loss of SrbA, the CBC, HapX or AtrR is associated with significant reductions in virulence in murine models of invasive pulmonary aspergillosis^[91]11–[92]13,[93]17. The transcriptional network governing azole resistance is therefore highly complex and involves multiple regulators, some of which remain to be identified as our current models of this network are not able to explain all of the existing clinically significant mechanisms of azole resistance. We therefore postulated that other perturbations of the transcriptional network would lead to alterations in azole resistance without affecting pathogenicity. In this study, we have generated and screened a library of 484 A. fumigatus transcription factor null mutant strains and identified a cohort of 12 factors that govern azole resistance and sensitivity. Here, we describe in detail the role of two CBF/NF-Y family transcription regulators, AFUB_029870 (NctA) and AFUB_045980 (NctB) where loss of function leads to azole resistance. We show that the A. fumigatus NctA and NctB (Negative cofactor two A and B) are part of the same transcriptional regulatory complex and demonstrate that the NCT complex is a key regulator of ergosterol biosynthesis and the azole exporter CDR1B. We also report that loss of the NCT complex leads to a multi-drug-resistance phenotype, including the azoles (itraconazole, voriconazole and posaconazole) as well as the salvage therapeutic amphotericin B^[94]18 and terbinafine, an agent used in the treatment of chronic and allergic disease^[95]19. Furthermore, loss of this complex results in a notable increase in the immunogenic properties of A. fumigatus, but does not result in loss of virulence. The results of our study highlight that loss of function of a single gene can give rise to cross-resistance to many of the primary drugs used to treat aspergillosis and should prompt clinical centres to look for these mechanisms of resistance. As we also show that strains lacking a functional NCT complex are hypersensitive to the echinocandins, switching patients harbouring strains with these mutations to caspofungin could be a preferred course of action. Results Generation of a library of transcription factor null mutants A systematic review of the genes previously annotated as transcription factors (TFs) in the databases at ENSEMBL fungi, ASPGD and DBD^[96]20 resulted in the identification of 495 putative TFs (Supplementary Data [97]1). To further characterise this cohort, we classified each TF according to its Pfam domains, identified using Hidden Markov Model profiling (hmmscan). The majority of TFs were shown to have either one (n = 245) or two (n = 158) functional domains associated with transcriptional regulation (Supplementary Fig. [98]1). The most common domains identified were binuclear zinc cluster (n = 192 (PFAM00172) and fungal-specific transcription domains (n = 195 (PFAM04052 and 11951); Supplementary Fig. [99]1). This is consistent with previous reports that describe domains of transcription factors in ascomycetes^[100]21. Sixty-two proteins lacked any Pfam domains associated with transcription factor function. However, following cross-referencing of these to orthologues in the AF293 genome (whose genome annotation has been enhanced by comparison to RNA-seq data) to reconcile annotation differences, and additional evaluation to identify PROSITE, SUPERFAMILY, SMART and CDD domains, we were able to identify transcription factor-associated domains for all, but six proteins (Supplementary Data. [101]1). For two of these proteins, MedA and LaeA, a functional role in transcriptional regulation has been defined experimentally^[102]22–[103]24. We were unable to identify any functional domains in the orthologue of the TFIIIC subunit tfc6 or three proteins annotated as RfeD, F and G. Unlike in the model yeast, Saccharomyces cerevisiae, gene replacement strategies in A. fumigatus are complicated by relatively low levels of homologous recombination. This problem can be mitigated by the use of strains lacking components of the non-homologous end joining machinery^[104]25, such as Ku70^[105]26, Ku80^[106]27 and Lig4^[107]28, however, even in strains lacking these factors, gene replacement cassettes require around 1 kb of homologous sequence flanking each side of a target gene^[108]27. To facilitate the disruption of all of the TFs identified, we chose to employ a fusion PCR approach similar to that described by Szewczyk et al.^[109]29 (see schematic Supplementary Fig. [110]2) and used custom developed scripts to design the primers for amplification of the gene replacement cassettes (see the Methods section). Cassettes were successfully amplified for all 495 of the TFs, and were used to transform MFIG001, a Δku80, pyrG+ strain derived from FGSC strain A1160^[111]30. We isolated null homokaryons for 97.7% (484) transcription factor genes as defined by our ability to isolate strains, in which we could amplify from the hygromycin resistance cassette to a region beyond the gene replacement cassette in addition to a lack of a PCR product corresponding to the target gene (see schematic Supplementary Fig. [112]2). Precise replacement was further confirmed for a randomly selected subset of 12 mutants by Southern blot analysis (Supplementary Data [113]1). Despite several attempts (minimum n = 3), we were unable to isolate null mutants for 11 genes (Supplementary Data [114]1). The majority of these genes encode components of the RNA polymerase I/II/III transcription factor complexes, or are transcription factors, which locate at a higher-level of a regulatory hierarchy^[115]31 whose deletion would likely to cause a lethal phenotype. An azole susceptibility screen of the transcription factor mutant library The minimum inhibitory concentrations (MIC) of itraconazole for the isogenic strain (MFIG001) was determined to be 0.5 mg/L in the RPMI-1640 media, following procedures outlined by EUCAST^[116]32. To identify novel transcriptional regulators associated with azole sensitivity and resistance, we screened the transcription factor null library at itraconazole concentrations representing sub-minimum inhibitory concentrations (MIC) (0.06 and 0.12 mg/L) and at the MIC (0.5 mg/L). Six transcription factor null mutants exhibited clear and reproducible fitness defects in sub-MIC levels of itraconazole when compared with the cohort of mutants in the collection (Fig. [117]1a). The transcription factors knocked out in two of these mutants have previously been defined as activators of the ergosterol biosynthetic pathway, including cyp51A, namely SrbA and AtrR. The other null mutants identified in the screen lacked the carbon catabolite repressor CreA^[118]33, the calcium-responsive regulator ZipD^[119]34, the SAGA complex subunit AdaB^[120]35 and the orthologue of the S. cerevisiae stress-responsive regulator GIS2 herein described as GisB^[121]36. When the library was screened at itraconazole levels at the MIC, we identified six transcription factor mutants, two of which we have previously described (∆hapX and ∆hapB)^[122]12 and four of which we associate with azole resistance in A. fumigatus for the first time (∆nctA, ∆nctB, ∆areA and ∆rscE). Fig. 1. Identification of transcription factors associated with azole tolerance in A. fumigatus. [123]Fig. 1 [124]Open in a new tab a Relative growth fitness of the 484 constructed TFKO mutants in MIC (0.5 mg/L) and sub-MIC (0.06 and 0.12 mg/L) levels of itraconazole. Each dot represents the mean of three replicate experiments. The TFKOs which exhibited significant fitness defects (fitness less than −4 for sub-MIC, and more than 1.5 for MIC condition) are indicated by coloured circles. b, c Heatmaps showing drug susceptibility profiles of the screened TFKO mutants to different antifungal drugs. Susceptibility assays were performed in triplicate, where optical density readings of fungal growth were standardised to no-drug control wells and shown as a relative growth values. Source data are provided as a Source Data file. We extended our phenotypic profiling of these isolates to assess their general growth fitness (Supplementary Fig. [125]3) and sensitivity profiles to additional antifungal drugs (Fig. [126]1b, c). The identified transcription factor null mutants showed variable growth fitness profiles, depending on the culture conditions tested. However, we did not see a clear correlation between their growth fitness and azole-sensitivity phenotypes. In comparison with the isogenic wild-type control, ΔcreA and ΔzipD showed increased sensitivity to the ergosterol biosynthetic inhibitor terbinafine. Interestingly, the nctA and nctB null mutants phenocopied each other (Supplementary Fig. [127]3) and were resistant to the triazoles voriconazole (>32-fold increase in MIC) and posaconazole (>128-fold increase), terbinafine (twofold increase in MIC), miltefosine (eightfold increase in MIC) and amphotericin B (eightfold increase in MIC). Conversely, the nctA and nctB null mutants showed hypersensitivity to the cell wall perturbing agents Congo Red, Calcofluor White, caspofungin and micafungin. NctA and NctB are members of the CBF/NF-Y family of regulators Pfam domain searches indicate that NctA and NctB are members of the evolutionarily conserved CBF/NF-Y family of transcription factors, which include the CBC transcription regulator complex (Supplementary Fig. [128]4)^[129]37. NctA (AFUB_029870) encodes a 247 aa protein, and is a reciprocal BLAST match of the S. cerevisiae negative cofactor 2 (NC2) complex α-subunit Bur6. NctA and Bur6, however, show little sequence similarity with the exception of the CBF/NFYB domain where they share 49% identity over 75 contiguous amino acids. NctA has two paralogues in A. fumigatus, HapE and an as yet un-named regulator encoded by AFUB_058240. Consistent with the similar phenotypes of the nctA and nctB null mutants, nctB (AFUB_045980) encodes a 142 aa protein that is the reciprocal BLAST match of the S. cerevisiae NC2 complex β-subunit (Ncb2) sharing 49% sequence identity over 86% of the protein. NctB has one readily identifiable paralogue in A. fumigatus, HapC. Loss of NctA leads to a reduction in growth rate To determine if the effects of nctA loss were due solely to the gene replacement and not other mutagenic events from transformation, an nctA reconstituted strain was generated. The growth rate of the nctA null alongside the parent strain MFIG001 (WT) and the nctA reconstituted strain (nctA rec) was assessed on Aspergillus complete media (ACM), Aspergillus minimal media (AMM), RPMI-1640 and DMEM. In all conditions, the nctA null mutant had a significantly reduced growth rate (36–44% reduction) compared with MFIG001 (Fig. [130]2a, b). Importantly, the reconstituted strain was indistinguishable in these assays from MFIG001. This reduction in growth rate is associated with a decrease in the initial rate of germination (5–6 h) and hyphal extension (Fig. [131]2c, d). Fig. 2. Impact of nctA deletion upon A. fumigatus growth. [132]Fig. 2 [133]Open in a new tab a Colonial growth phenotypes of the wild-type MFIG001 (WT), the nctA null (∆nctA) and the nctA reconstituted strain (nctA rec) grown on a solid Aspergillus complete medium (ACM) for 72 h at 37 °C. b Radial growth phenotypes on solid ACM, Aspergillus minimal medium (AMM), RPMI-1640 and DMEM. The error bars represent standard error of the mean of three independent experiments. c Germination rates in liquid RPMI-1640 at 37 °C. P-value was calculated by repeated measures two-way ANOVA with Sidaks correction: ****P < 0.0001; NS, P > 0.05. The error bars signify the standard deviations. d Hyphal extension rates in RPMI-1640 at 37 °C. P-value was calculated by Kruskal–Wallis test with Dunn’s correction: ****P < 0.0001; NS, P > 0.05. Percentages of germination rates and hyphal extension rates were measured under the microscope. The error bars represent standard error of the mean. Source data are provided as a Source Data file. NctA and NctB co-operatively regulate the same network of genes To assess the role of NctA and NctB on global regulation of gene expression in A. fumigatus, we carried out transcriptomic analysis (RNA-seq) using RNA extracted from cultures grown in the RPMI-1640 media. In the absence of itraconazole, 1244 genes were upregulated (>2-fold; FDR < 0.05) and 1049 (>2-fold; FDR < 0.05) genes were downregulated in the nctA null when compared with the isogenic parent strain (Supplementary Data [134]2). Similarly, 1183 genes were upregulated and 1104 genes were downregulated in the nctB null mutant. In the presence of itraconazole, 735 genes were upregulated and 736 genes were downregulated in the nctA null mutant, and 865 genes were upregulated and 852 genes were downregulated in the nctB null mutant. Direct comparison of the two data sets shows that the regulons of both transcription factors have a very high degree of similarity in both conditions (Fig. [135]3a, b), suggesting that both transcription factors work co-operatively. Fig. 3. Effects of nctA and nctB deletion on the global and the ergosterol biosynthetic gene expression. [136]Fig. 3 [137]Open in a new tab a, b Comparison of differential gene expression in the nctA and the nctB null mutants. Scatter plot comparison of the log2 differential expression ratio of gene expression data in (a) no-drug, and (b) 0.5 mg/L itraconazole conditions. The log2 expression ratio between the nctA null mutant and the wild-type are plotted on the x-axis, and the nctB null mutant and the wild-type are plotted on the y-axis. The Pearson’s correlation (r) between the two gene expression data sets is shown. c, d Effects of nctA deletion on the expression of the genes involved in ergosterol biosynthesis. c The putative ergosterol biosynthetic pathway in A. fumigatus. The genes highlighted in yellow are those whose expression levels were upregulated more than 1.5-fold in the nctA null mutant compared with the wild-type. The genes highlighted in light blue are those whose expression levels were downregulated more than 1.5-fold in the nctA null mutant compared with the wild-type. b Heatmap showing the RNA-seq expression levels of the genes involved in ergosterol biosynthesis. Log2 differential expression values are scaled between −1.5 and 1.5, and displayed. Source data are provided as a Source Data file. To assess if NctA physically interacts with NctB, and to determine if any other proteins are found complexed with NctA, we generated a strain which encoded a functional C-terminally S-tagged derivative of NctA (as shown by the fact that the strain that was indistinguishable from the wild-type isolate) (NctA-S-tag, see Supplementary Fig. [138]5), and performed co-immunoprecipitation followed by LC/MS identification of interacting proteins. In addition to NctA, 20 unique proteins were identified (Supplementary Table [139]1; >2 matched peptides), including NctB, the TBP-associated transcriptional regulator Mot1, the transcriptional co-repressor Cyc8 and 9 ribosomal or ribosome-associated proteins. The interaction between NctA and NctB and Mot1 was supported with reciprocal co-immunoprecipitation using a C-terminally S-tagged version of NctB (NctB-S-tag, see Supplementary Fig. [140]5 and Supplementary Table [141]1). Taken together these data suggest that, consistent with the role of Bur6 and Ncb2 in yeast^[142]38,[143]39, NctA and NctB form a complex with the TBP-associated co-regulator Mot1 in A. fumigatus and are responsible for regulating the same cohort of genes. The NCT complex is a global regulator of sterol biosynthesis The negative cofactor complex has been defined as a general regulator of transcription, however, it is clear from the data presented in Fig. [144]3 that the NctA/B complex is responsible for regulating only a subset of genes at any one time. To assess if these subsets of genes are enriched for particular processes or biochemical pathways, GO term and Metabolic Pathway Enrichment analysis was performed on genes dysregulated more than twofold. Such analysis are somewhat limited when applied to A. fumigatus as GO terms and metabolic maps for filamentous fungi are poorly defined, however, we were able to identify enriched classes of genes in our upregulated cohort that are associated with secondary metabolism, transcriptional and translational processing, transport and notably steroid biosynthesis (Supplementary Fig. [145]6). As both nctA and nctB null mutants are pan-azole resistant and our metabolic enrichment analysis identified steroid biosynthesis as an enriched class in genes upregulated in the null mutants, we assessed the mRNA levels of the genes in the ergosterol biosynthetic pathway (Fig. [146]3c, d)^[147]40. Of the 29 genes annotated as being associated with the pathway, 7 (erg13A, hmg2, erg12, erg11A, erg24A, erg6 and erg2) have increased mRNA levels more than 1.5-fold in the nctA null mutant (Fig. [148]3; Supplementary Data [149]2), while nine have reduced mRNA levels (erg10A, erg13B, erg7C, erg24B, erg25A, erg26B, erg27, smt1 and erg5). Intriguingly for three of the downregulated genes, their paralogues are upregulated (erg13B, erg24B, smt1 [paralog to erg6]). Notably, erg7C has recently been identified as an oxidosqualene:protostadienol cyclase which diverts 2,3-epoxysqualene from ergosterol biosynthesis into the helvoic acid biosynthetic pathway^[150]41. Potentially, reduced levels of erg7C would result in increased flux through the ergosterol biosynthetic pathway. Similarly, we observed significant downregulation of several siderophore biosynthesis genes in the nctA and the nctB null mutants, especially sidI, which acts to divert the ergosterol precursor mevalonate into the siderophore biosynthesis pathway^[151]42 (Supplementary Data [152]2). Together with the upregulation of hmg2 and erg12 in the NCT mutants, this might suggest a potential role of the NCT complex in balancing ergosterol and siderophore biosynthesis. Our RNA-seq analysis indicates that the NCT complex is a key regulator of other transcription factors associated with azole resistance. Loss of nctA leads to upregulation of activators of the ergosterol biosynthetic pathway (2.2-fold for srbA and 1.6-fold for atrR in the no-drug conditions) and downregulation of the negative regulator-encoding gene hapC (approximately twofold, Supplementary Data [153]2). These results suggest that the NCT complex affects transcript levels of ergosterol biosynthesis genes both directly and indirectly by modulating the expression levels of these transcription factors. NCT binds the promoters of genes linked to azole susceptibility To identify the promoters of genes that are directly bound by the NCT complex, we performed chromatin-immunoprecipitation sequencing (ChIP-seq) using an anti-S-tag polyclonal antibody to isolate DNA bound to the NctA-S-tag fusion protein in vivo. We identified 4811 and 4290 NctA-binding peak regions in the absence and the presence (0.5 mg/L) of itraconazole, respectively (q-value < 0.01, fold enrichment >1.5; Supplementary Data [154]3). A comparison of the ChIP-seq data sets showed that >70% of the peak regions are common between the conditions (Fig. [155]4a). Analysis of genome-wide occupancy of NctA revealed that the large majority (~70%) of NctA-binding peaks are located within 1.5 kb upstream of the translational start site (TSS) of an annotated gene (Fig. [156]4b). In total, 33.1% (3316 genes in no-drug conditions) and 29% (2945 genes in 0.5 mg/L itraconazole conditions) of the total predicted ORFs were assigned to have at least one NctA-binding event within their upstream region. Frequency distribution of NctA-binding peak summits showed that NctA is predominantly positioned around 300 bp upstream from the TSS (Fig. [157]4b). Applying the Multiple Em for Motif Elicitation (MEME) de novo motif discovery program^[158]43, we identified several conserved nucleotide motifs within the ChIP peak regions (Fig. [159]4c). TATA-box like AT-rich motives were identified as the common nucleotide motifs with the highest e-value. This is consistent with the result from previous studies with NCT orthologues, which have been shown to form a stable complex with the TATA-box-binding protein (TBP)^[160]44–[161]47. The results of the motif discovery suggest that the NCT complex in A. fumigatus also interacts with TBP to regulate gene expression as also suggested by the physical interaction studies. Fig. 4. Genome-wide binding profile of NctA. [162]Fig. 4 [163]Open in a new tab a Venn diagram showing the overlap of NctA-binding peaks (q-value < 0.01, fold enrichment >1.5) obtained from no-drug and the 0.5 mg/L itraconazole-treated conditions. b Distribution of the ChIP-seq peak location with respect to annotated genes. The frequency distribution of the distance between the merged peak summit of the significant binding peak regions (q-value < 0.01, fold enrichment >1.5) and the translation start site (ATG) of the nearest annotated gene is plotted. c Conserved nucleotide motifs identified in the NctA ChIP-seq peak regions. A summary of the de novo motif discovery analysis is shown with the identified consensus motifs and its sequence logos, the calculated e-values, and the frequency of the appearance of the motifs within the ChIP-seq peak set are analysed. Source data are provided as a Source Data file. We investigated the correlation between NctA occupancy and mRNA levels by comparing the ChIP-seq and the RNA-seq data sets. We expected that regulation of genes in close proximity to the NctA-binding sites would be altered upon loss of nctA. However, for the total 3361 genes having at least one NctA-binding peak within 1.5 kbp of upstream region, only 755 genes (22%) showed differential mRNA levels, and the remaining 2606 genes exhibited no significant change (Supplemental Fig. [164]7a, b). Among the 755 differentially expressed genes, 352 genes were upregulated and 403 genes were downregulated in the nctA null mutant in no-drug conditions. Similar results were obtained for the itraconazole-treated (0.5 mg/L) sample (Supplemental Fig. [165]7c, d). These results suggest that binding of the NCT complex alone is not sufficient for it to elicit its regulatory function. In order to obtain further insight into the molecular mechanisms driving the itraconazole resistance in the nctA and the nctB null mutants, we analysed binding of NctA on the promoter region of the genes related to ergosterol biosynthesis and their known transcriptional regulators (Supplementary Data [166]2, Supplementary Fig. [167]8a). Of the 16 genes, which showed differential expression in the RNA-seq analysis in the absence of itraconazole, nine genes (erg10A, erg13B, erg11A, erg24A, erg25A, erg27, erg6, smt1 and erg5A) were found to have at least one NctA-binding peak within 1.5 kb of the upstream region. Similarly, among the 14 differentially expressed genes in the presence of itraconazole, 11 genes were shown to have the peak regions within the defined upstream region (erg10A, erg13A, erg13B, erg11A, erg11B, erg24A, erg24B, erg25B, erg6, smt1 and erg3). Interestingly, the genes encoding the regulatory proteins HapC, SrbA and AtrR, and Mot1 were also confirmed to have an NctA-binding peak in their upstream region (Supplementary Fig. 8b, Supplementary Data [168]2). Ergosterol levels are elevated in NCT complex mutants To analyse the effect of the loss of the NCT complex on ergosterol biosynthesis, sterol levels were quantified using GC–MS. Consistent with our hypothesis, we observed a 60% increase in ergosterol content (w/w) in both mutant strains when compared with the isogenic control (P < 0.04; Fig. [169]5). Ergosterol levels in the reconstituted strain were indistinguishable from the wild-type (P > 0.85), and lower than observed in the knockout (P = 0.0138). Taken together, our data suggest that NctA and NctB are negative regulators of ergosterol biosynthesis via their interaction at the promoters of cyp51A and other genes that encode components of the ergosterol biosynthetic pathway. Fig. 5. Loss of nctA and nctB increases the cellular ergosterol content. Fig. 5 [170]Open in a new tab Ergosterol levels of the wild-type (WT), the nctA null mutant (∆nctA), the nctB null mutant (∆nctB) and the nctA reconstituted isolate (nctA rec) in RPMI-1640 incubated for 24 h were determined by GC–MS. Ergosterol content of each mutant was normalised to that of the wild-type, and shown as a relative fold change. Samples were assessed in biological triplicates. P-values were calculated using one-way ANOVA: *P < 0.05; ns, P > 0.05. The error bars represent standard error of the mean. Source data are provided as a Source Data file. Levels of the azole transporter CDR1B are elevated in NCT mutants The relatively modest increase in expression of cyp51A (ca. 2.5-fold) and cellular ergosterol content (ca. 1.6-fold) in the nctA mutant did not appear consistent with the relatively large increase in itraconazole resistance (>64-fold), suggesting that factors independent of ergosterol biosynthesis may be influencing azole resistance in the NCT complex-deficient strains. Recently, an association between azole resistance and increase in mRNA levels of the ABC transporter CDR1B has been described^[171]30, and the transcriptional regulator AtrR has been shown to co-regulate both cdr1B and cyp51A expression^[172]11. We therefore examined our RNA-seq data to see if cdr1B was dysregulated in the nctA and nctB null mutants. mRNA levels of cdr1B were increased by 3.1 (FDR = 2.5 × 10^2) and 2.1 (FDR = 0.15)-fold, respectively, in the nctA and the nctB null (Fig. [173]6a). Furthermore, evaluation of our ChIP-seq data suggested that this regulation was related to a direct interaction between the NCT complex and the cdr1B promoter (Fig. [174]6b). To assess if this increase in transcript levels led to a concomitant increase in translated protein levels, we quantified levels of the transporter using an anti-CDR1B antibody^[175]48. In keeping with our transcriptomic results, CDR1B levels were increased in the nctA null mutant by >2.4-fold (Fig. [176]6c, d). Fig. 6. Defects in the NCT complex leads to transcriptional de-repression of cdr1B and over-production of CDR1B protein. [177]Fig. 6 [178]Open in a new tab a Expression levels of cdr1B transcripts in the nctA null and the nctB null mutant in RNA-seq analysis: *, FDR < 0.05. b In vivo binding of NctA on the 5′-upstream region of cdr1B. Tracks for the NctA ChIP-seq (ChIP) and their input DNA control (Input) are visualised in the UCSC genome browser together with annotated gene models and their transcript, which are expressed in the no-drug conditions. The direction of the target gene and the 5′-proximal gene are shown in red arrows and blue arrows, respectively. c Representative western blot (WB) showing the increase in CDR1B protein levels in the ∆nctA mutant. Cell-free extracts were resolved via SDS-PAGE, and probed for CDR1B using a CDR1B-specific antibody. Ponceau S staining was performed as an overall loading control. d Relative protein levels of CDR1B. The relative intensity of the CDR1B signal was quantified by densitometric scanning. The data represent the mean results of two biological replicates, and the error bars signify the standard deviations. Source data are provided as a Source Data file. The nctA null mutant shows enhanced immunogenic properties Given the requirement of the NCT complex in the resistance to two leading classes of therapeutic agents used to treat aspergillosis, it was important to assess if NCT complex-mediated regulation is important for pathogenicity of A. fumigatus. As our evidence to data suggests that the roles of NctA and NctB are non-redundant with respect to each other, we investigated pathogenic properties of the nctA mutants using both in vitro and in vivo infection models. To assess contact-mediated cytotoxicity, the nctA null mutant was co-cultured with human A549 alveolar epithelial cells and the murine macrophage RAW 264.7 cell line (Fig. [179]7a, b). Our results indicate that loss of NCT complex results in a significant reduction in cell damage when compared with either isogenic control or the reconstituted isolate. This result is consistent with the growth reduction seen for this strain in the media used for this experiment (Fig. [180]2b; Supplementary Fig. [181]5c). We next examined immunogenic properties of the nctA null mutants to mammalian immune cells. Murine macrophages challenged with live spores from the nctA mutant showed an increase in production of TNF-α, IL-6 and IL12p40 (Fig. [182]7c–e), whereas human epithelial cells challenged with the nctA mutant showed an increase in production of IL-8 (Fig. [183]7f). The nctA null mutant also resulted in more activation of dendritic cells compared with the wild-type (Fig. [184]7g, h). These findings suggest that the nctA null mutant could potentially cause an aberrant immunogenic response during infection. The increased immunogenic properties observed for the nctA null mutant could be associated with an alteration in the cell wall structure or spore surface, as the null mutant was hypersensitive to the cell wall synthesis inhibitory drugs and the perturbing agents Congo Red and Calcofluor White (Fig. [185]1c). However, examination by TEM revealed that the thickness of the cell wall was unchanged in the nctA null mutant (Supplementary Fig. [186]9). Fig. 7. Cytotoxicity and immunogenic properties of the nctA null mutant. [187]Fig. 7 [188]Open in a new tab a, b A549 epithelial cells or macrophages were infected with the wild-type (WT), the nctA null mutant (∆nctA) or the nctA reconstituted isolate (nctA rec) for 24 h. Cytotoxicity of each mutant was evaluated by measuring the release of lactate dehydrogenase (LDH) activity into the culture medium. The data represent five different infection challenges with triplicate LDH activity measurements. Data are shown in fold change of LDH activity relative to the wild-type-infected cells. The error bars mean the standard error of the mean (SEM), and P-values were calculated by Kruskal–Wallis test with Dunn’s correction: *P < 0.0180; **P < 0.0056 (for a, A549 cells). **P < 0.0032; ***P < 0.0007 (for b, macrophages). c–f Granulocyte macrophage colony-stimulating factor-induced bone marrow-derived dendritic cells (gm-csf BMDCs) were infected with the A. fumigatus strains with the multiplicity of infection (MOI) = 5:1. Proinflammatory cytokines were quantified by ELISA. Data represent three biological replicates (c–e) or five biological replicates (f) with ± SEM. P-values were calculated by ANOVA with Tukey’s correction: ***P < 0.002; ****P < 0.0001. g, h Activation of dendritic cells measured by CD40 and CD80 markers by flow cytometry. Data represent three biological replicates with ±SEM. P-values were calculated by ANOVA: ***P < 0.0002; ****P < 0.0001 (for g). ***P < 0.0004; ****P < 0.0001 (for h). Source data are provided as a Source Data file. An NCT mutant is virulent in a model of invasive aspergillosis To assess the impact of the loss of the NCT complex in pathogenicity, we compared the virulence of the nctA null mutant with that of the isogenic isolate MFIG001 in a leukopenic model of invasive aspergillosis. All mice challenged with the MFIG001 strain succumbed to infection (100% mortality) within 7 days post infection. Despite the significant growth defect (Fig. [189]2b) and the reduced cytotoxicity (Fig. [190]7a, b) observed in NCT mutants, the virulence of the nctA null mutant was statistically indistinguishable from the isogenic isolate (Fig. [191]8a). Fig. 8. Assessing the effect of loss of nctA on the virulence of A. fumigatus in murine infection models. [192]Fig. 8 [193]Open in a new tab Mice were infected with the wild-type (WT), the nctA null mutant (∆nctA) or the reconstituted isolate (nctA rec). a, b Kaplan–Meier curve for murine survival for mice treated via (a) intranasal infection with 5.0 × 10^5 conidia after mice were rendered neutropenic by treatment with cyclophosphamide, and (b) intranasal infection with 7.0 × 10^6 conidia after cortisone acetate treatment. A log rank analysis was used to compare the results between the strains. c Relative expression levels of proinflammatory cytokines in the infected lung tissues. Data represent the mean of biological triplicates and error bars illustrate the standard deviation. Statistical significance was calculated by Student’s t test. d Histopathology of representative sections of lungs after 36 h post infection with WT, ∆nctA, or nctA rec. Lung sections were stained with haematoxylin–eosin (H&E) for visualisation of the host cells, and Grocott’s Methenamine Silver (GMS) for visualisation of the fungal elements. Scale bar: 50 µm. e Quantification of the fungal burden in the infected lungs. The total genomic DNA was extracted from the same lung lobe samples used for the cytokine expression analysis, and fungal DNA concentration was determined by qPCR. Data represent the means of biological replicates, and the error bars mean standard deviation. The statistical significance of variances between fungal burdens was calculated by using a non-parametric Mann–Whitney t test. Source data are provided as a Source Data file. In the leukopenic aspergillosis model, the host cellular innate response is quantitatively attenuated particularly during the initiation of infection allowing the microbe to proliferate and cause host damage. Thus, we investigated the role of the NCT complex in virulence in a murine model that uses cortisone acetate as an immunosuppressive agent resulting in qualitative defects in innate immunity. This model has been shown to better reflect the host’s immune response in relation to detection of fungal-specific PAMPs. In this model, overall murine mortality with the isogenic strain was 45% at day 10 post infection. Similarly, the nctA null mutant also showed 50% mortality at the end of the infection time course. Although the nctA rec strain showed a slightly increased mortality compared with the other strains, no significant differences were observed in virulence between the all tested strains (Fig. [194]8b). One potential hypothesis that could explain the retention of virulence of the nctA null mutant is the aberrant immunogenic properties observed in the mutant in vitro. However, no significant differences were observed in the expression levels of cytokine-encoding genes (TNF-α, IL12 and IL-6) between the strains at 36 h post inoculation (p.i.) (Fig. [195]8c). Furthermore, no obvious difference in the inflammatory response was observed between the strains in histopathology (Fig. [196]8d). Contrary to our expectation from the in vitro studies, the nctA null mutant showed an indistinguishable level of fungal burden with the isogenic control at 36 h p.i. (Fig. [197]8e). These results suggest that pathogenicity is maintained in this strain as its growth is not significantly altered in vivo, and our results add to the growing body of evidence that poor in vitro growth is not an absolute indicator of virulence defects^[198]49,[199]50. Discussion In this study, we have constructed a 484 members transcription factor null mutant library with the aim of providing a systematic evaluation of regulators that contribute to azole resistance in A. fumigatus. This library now provides an opportunity for the fungal community to further explore regulatory mechanisms and factors in the pathobiology of this important human fungal pathogen. To date, much of our understanding of azole tolerance has been driven by hypotheses derived from model organisms that have significantly superior functional genomic resources. Functional screens of transcription factor null mutant libraries have been performed in the model yeasts S. cerevisiae^[200]51 and Schizosaccharomyces pombe^[201]52, the model filamentous fungus Neurospora crassa^[202]53 and the pathogenic yeasts Cryptococcus neoformans^[203]54 and Candida albicans^[204]55. Although these studies have proven to be effective in uncovering a number key regulators of drug resistance, the roles of transcription factors can vary significantly from species to species and large-scale transcriptional rewiring is frequently observed (for a review see ref. ^[205]56). For example, loss of the A. fumigatus pH-responsive transcriptional regulator PacC, results in a >20-fold increase in flucytosine sensitivity in A. fumigatus^[206]57, however, loss of the orthologue in yeast (RIM101) leads to an increase in flucytosine resistance^[207]54,[208]55. Our screen of azole resistance in A. fumigatus highlights significant differences in the transcriptional regulation of azole tolerance in filamentous fungi compared with yeasts. Of the 12 transcription factors identified in our screen as having altered susceptibility to itraconazole, only half HapB, SreA (SrbA)^[209]54, AdaB^[210]35,[211]58, GisB^[212]59 CreA and HapX^[213]60 may have been predicted from previous screening efforts. Of the remaining regulators, functional orthologues of AreA exist in yeast (eg. Gat1p and Gln3p in S. cerevisiae), however, they have not been associated with changes in azole resistance while orthologues of three (NctA, NctB and RscE) are essential for viability, and two (ZipD and AtrR) are absent in yeasts. We have explored, in detail, the role of the two CBF/NF-Y family transcription regulators, AFUB_029870 (NctA) and AFUB_045980 (NctB) in azole resistance. The orthologues of these regulators in S. cerevisiae, known, respectively, as Bur6 and Ncb2, are subunits of a heterotrimeric transcriptional regulator called Negative Cofactor 2 (NC2). The NC2 complex, originally identified as a TBP (TATA-box-binding protein)-associated factor, acts as a negative regulator of RNA polymerase II transcription by inhibiting formation of the pre-initiation complex (PIC)^[214]61,[215]62. Assembly of the heteromeric PIC is required for transcription from RNA pol II-dependent promoters, and its assembly is contingent upon recruitment of the TBP and the general transcription factors TFIIA and TFIIB. NC2 inhibits PIC formation by preventing the interaction of TBP with TFIIA and TFIIB [ref. ^[216]63 and references therein]. The interaction between NCT and