Accepted Manuscript Intracellular siderophore but not extracellular siderophore is required for full virulence in Metarhizium robertsii Bruno Giuliano Garisto Donzelli, Donna M. Gibson, Stuart B. Krasnoff PII: DOI: Reference:

S1087-1845(15)30001-3 http://dx.doi.org/10.1016/j.fgb.2015.06.008 YFGBI 2863

To appear in:

Fungal Genetics and Biology

Received Date: Revised Date: Accepted Date:

18 May 2015 26 June 2015 27 June 2015

Please cite this article as: Donzelli, B.G.G., Gibson, D.M., Krasnoff, S.B., Intracellular siderophore but not extracellular siderophore is required for full virulence in Metarhizium robertsii, Fungal Genetics and Biology (2015), doi: http://dx.doi.org/10.1016/j.fgb.2015.06.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Intracellular siderophore but not extracellular siderophore is required for full virulence in Metarhizium robertsii Bruno Giuliano Garisto Donzelli1*, Donna M. Gibson2, Stuart B. Krasnoff2

Affiliations 1

School of Integrative Plant Science - Plant Pathology and Plant-Microbe Biology Section,

Cornell University, Ithaca, NY 14853 2

USDA ARS, Robert W. Holley Center for Agriculture and Health, 538 Tower Road, Ithaca, NY

14853

* Corresponding author E-mail: [email protected]

1

Abstract Efficient iron acquisition mechanisms are fundamental for microbial survival in the environment and for pathogen virulence within their hosts. M. robertsii produces two known iron-binding natural products: metachelins, which are used to scavenge extracellular iron, and ferricrocin, which is strictly intracellular. To study the contribution of siderophore-mediated iron uptake and storage to M. robertsii fitness, we generated null mutants for each siderophore synthase genes (mrsidD and mrsidC, respectively), as well as uptake-deregulated mutants by inactivating the transcriptional repressor mrsreA. All of these mutants showed impaired germination speed, differential sensitivity to hydrogen peroxide, and differential ability to overcome iron chelation on growth-limiting iron concentrations. RT-qPCR data supported regulation of mrsreA, mrsidC, and mrsidD by supplied iron in vitro and during growth within the insect host, Spodoptera exigua. We also observed strong upregulation of the insect iron-binding proteins, transferrins, during infection. Insect bioassays revealed that ferricrocin is required for full virulence against S. exigua; neither the loss of metachelin production nor the deletion of the transcription factor mrsreA significantly affected M. robertsii virulence.

Keywords Siderophore; iron; metachelins, ferricrocin, virulence, GATA transcriptional repressor

Abbreviations BAW, beet armyworm (Spodoptera exigua); TFs, transferrins; AsMM, Aspergillus Minimal Medium 2

1. Introduction Iron is an abundant (Yanagi, 2011) but poorly bioavailable nutrient in both soils (Schwertmann, 1991) and aquatic environments (Boyd and Ellwood, 2010). In the vast majority of living organisms, decreased iron acquisition due to low iron concentration or impaired uptake limits growth and fitness. In contrast, an excess of iron can have detrimental effects associated with the production of toxic free radicals through the Fenton-Haber Weiss reaction (Fenton, 1896; Haber and Weiss, 1932). In plants and animals, excessive free iron also increases vulnerability to pathogens (Bullen et al., 1991; Jones and Wildermuth, 2011), and for this reason, iron sequestration is considered a universal innate antimicrobial defense (Dellagi et al., 2005; Hood and Skaar, 2012; Kieu et al., 2012). Thus, within or outside their hosts, microbial pathogens face iron scarcity which has driven the evolution of high affinity uptake systems. Fungi possess two main high affinity iron uptake pathways: one depends on the reduction of highly insoluble Fe3+ to the more soluble Fe2+ by ferric reductases (reductive iron assimilation, RIA) while the second employs nonribosomally synthesized Fe3+ chelators (siderophore-mediated iron acquisition, SIA). Fungal pathogens vary in their reliance on these acquisition systems during the interaction with their hosts. In the plant pathogens Cochliobolus heterostrophus, Gibberella zeae and Alternaria alternata, SIA and not RIA is essential for full virulence (Chen et al., 2013; Condon et al., 2014; Oide et al., 2006; Park et al., 2006; Schrettl et al., 2004). In Colletotrichum graminicola, SIA and RIA have complementary roles: RIA is deployed during the biotrophic phase, when extracellular siderophore production is downregulated to avoid host detection; SIA is then used to support 3

necrotrophic growth (Albarouki and Deising, 2013; Albarouki et al., 2014). In Ustilago maydis, SIA is dispensable for virulence whereas RIA impairment strongly affects growth and sporulation in planta. (Eichhorn et al., 2006; Mei et al., 1993). A similar variability is observed in animal pathogens some of which can acquire iron directly from host macromolecules. For instance, Candida albicans and Cryptococcus neoformans do not produce siderophores, and are reduced in virulence when their RIA system is disrupted (Ramanan and Wang, 2000; Saikia et al., 2014). Both fungi are also able to exploit iron reservoirs within their hosts using dedicated pathways (Cadieux et al., 2013; Hu et al., 2013; Weissman and Kornitzer, 2004). Null mutants of A. fumigatus that are incapable of extracellular siderophore biosynthesis showed reduced virulence in murine models of invasive pulmonary aspergillosis, with RIA providing a minor contribution to iron uptake in the host environment (Schrettl et al., 2004). A role for SIA during host interaction is also likely in the human pathogens Histoplasma capsulatum and Paracoccidioides spp. where the siderophore system has been only partially characterized (Hilty et al., 2011; Silva-Bailao et al., 2014). In the vast majority of Ascomycota, siderophores also play a role in intracellular iron homeostasis, assisting in iron transport within the colony and reducing formation of reactive oxygen species (Eisendle et al., 2006; Oide et al., 2007). Loss of intracellular siderophores is often associated with reduced conidiation, impaired sexual development, increased sensitivity to oxidative damage and delayed conidial germination, and in some pathogenic fungi, reduction of virulence (Hof et al., 2007; Matzanke et al., 1988; Schrettl et al., 2007; Wallner et al., 2009). In Ascomycetes, iron acquisition and iron consuming pathways are regulated by orthologs of the A. fumigatus transcription factors hapX and sreA (Baek et al., 2008; Haas et al., 1999; Hortschansky et al., 2007; Lopez-Berges et al., 2012; 4

Pelletier et al., 2003; Schrettl et al., 2010). sreA orthologs from different fungal species were shown to bind similar consensus sequences (HGATAR) found in promoter regions of ironregulated genes (Chao et al., 2008; Chi et al., 2013; Gauthier et al., 2010; Pelletier et al., 2002; Schrettl et al., 2008; Voisard et al., 1993; Zhou and Marzluf, 1999). The role of siderophore-mediated iron acquisition in entomopathogenic fungi has yet to be explored. M. robertsii produces intracellular ferricrocin (vide infra), and, when deprived of iron, secretes a mixture of iron binding coprogen-type siderophores that include unusual Oglycosylated, N-oxidized compounds (metachelins) and as well N(α)-dimethyl coprogen (NADC) and dimerumic acid (DA) which are known from other fungi (Fig. 1) (Krasnoff et al., 2014). In this work, we analyzed the role of metachelins and ferricrocin in M. robertsii fitness and virulence using deletion mutants for each siderophore synthetase as well as strains in which the general iron uptake transcriptional repressor mrsreA was inactivated.

2. Materials and methods 2.1. Strains, culture conditions and fungal transformation Conidia from M. robertsii ARSEF 2575 and its derivatives were obtained on ¼ strength Sabouraud dextrose agar with yeast extract (SDAY/4) (Moon et al., 2008). Phenotypic characterizations were carried out using strains grown on AsMM amended with FeCl3.7H2O when required. Solid AsMM was produced with 1% agarose, since standard agar was found to contain significant amounts of iron. Iron content of AsMM broth, AsMM plus 1% agarose and AsMM plus 1.5% standard agar, were 1 ppb (0.018 μM), 72 ppb (1.3 μM, agarose iron content 5

was 7.25 μg/g) and 3,046 ppb (55 μM), respectively. Iron concentrations were determined by analysis on an inductively coupled argon plasma emission spectrometer (ICP-ES, model 61E Trace Analyzer, Thermo Jarrell Ash Corp., Franklin, MA). Media and methods used for Agrobacterium-mediated transformation are described elsewhere (Moon et al., 2008). Chrome azurol S (CAS) plates were produced as described (Schwyn and Neilands, 1987) with the exclusion of casaminoacids, since other components provided sufficient N for fungal growth.

2.2. Bioinformatic analyses Orthologous genes were identified by reciprocal best blast between predicted proteins identified in genomes of M. robertsii ARSEF2575, Gibberella zeae PH-1, Cochliobolus heterostrophus C5 and Aspergillus fumigatus Af293. Protein domains were identified using the on-line services Interproscan (ver 5) and NCBI CDD search (Marchler-Bauer et al., 2011). Alignment of genomic regions was performed with Multigeneblast (Medema et al., 2013) using a database containing 46 fungal genomes covering all Ascomycota subdivisions except Saccharomycotina. Primer design for the detection of BAW TFs was based on two fragments deposited to GeneBank as accessions AEW24428 (transferrin-1, TF1) and EW24429 (transferrin2, TF2) which show high similarity to Bombix mori transferrin-like proteins XP_004926508 and NP_001037014. Primers for TF3 and TF4 were designed using sequences from the Spodobase (http://bioweb.ensam.inra.fr/Spodobase/), clusters Se1E19898-5-1 and Sf9LR491641-5-1-C1, respectively. These 2 fragments displayed the highest similarity to B. mori melanotransferrinlike genes XP_004927448 and XP_004927626, respectively. Conserved motifs were identified 6

with the oligo-analysis tool in RSAT and a custom background model (Thomas-Chollier et al., 2011) using promoter regions extracted from genes likely to be involved in iron-metabolism in M. robertsii. These included mrsreA (EXV00864), mrsidD (EXV04699), mrsidC (EXV05490), mrsidA (EXV05491), mrsidF (EXV04697), mrsidI (EXV04698), mrsidJ (EXV05489.1), mrsit1 (EXU96430), mrftr1 (EXU98046), mrfetC (EXU98047), mrhapX (EXU97861), mrabcB (EXV04695), mrfre3 (EXV00893), and mrctr3 (EXV00894). Phylogenetic trees based on either mrsreA ortholgs or concatenated RNA polymerase II largest and second largest subunits (rpb1 and rpb2), and the elongation factor 1 alpha (tef1) from 41 Ascomycota were reconstructed with PhyML 3.1 (JTT matrix and SPRs topology search) after MAFFT 7.1 alignment (G-INS-I method) and manual editing (Guindon et al., 2010; Katoh et al., 2002; Schoch et al., 2009). Trees were compared using the Nye topological score (Nye et al., 2006).

2.3. Generation of mrsidC, mrsidD and mrsreA null mutants Null mutants for mrsidC, mrsidD and mrsreA were produced using double crossover gene replacement constructs. For each gene, two homing DNA fragments and the intercalating bar selection marker were obtained by PCR with primers listed in Table 1 followed by directional assembly into pBDU as described elsewhere (Giuliano Garisto Donzelli et al., 2012; Pall and Brunelli, 1993). Amplification templates were M. robertsii ARSEF2575 and pUCAP Bar NOSII (Donzelli, unpublished), respectively. pBDU derivatives carrying gene replacement constructs were mobilized into Agrobacterium tumefaciens EHA105 and then delivered to M. robertsii ARSEF2575 as described (Moon et al., 2008). Screening for putative knockouts in early 7

transformants was carried out by standard PCR (see Table 1 for primers). Gene deletion and bar cassette insertion number were subsequently assessed in single conidial progenies by Southern analyses using described methods (Moon et al., 2008) and PCR-synthesized DIG-labeled probes (Table 1).

2.4. Metachelin identification and quantification One hundred ml batches of AsMM with no iron added were inoculated with 106 conidia, and then shaken at 150 RPM at 25 °C for 12 days. Culture broths were filtered through Whatman #1, extracted with half-volumes of EtOAc and then chromatographed on Amberlite XAD-16 which had been batch extracted in MeOH to remove impurities, equilibrated by several exchanges of H2O and then slurry-packed into 1.4 cm ID x 9.0 cm polypropylene columns. After loading of 100 ml of EtOAc extracted broth, columns were eluted with 5 bed volumes of H2O. The siderophore fraction was then eluted with 10 bed volumes of MeOH, dried in vacuo, and stored at 4 °C. Yields of crude extract ranged from 15 - 45 mg/100 ml broth. After weighing, samples were redissolved in 2 ml MeOH and diluted 20-fold in MeCN-H2O (15:85) with 0.1% formic acid (0.4-1 mg/ml) and 5 µL aliquots were analyzed by HPLC-MS. HPLC-MS employed a Waters Acquity UPLC system with an Acquity UPLC® BEH C18 column (2.1 x 50 mm, 1.7 µm) eluted at 0.3 or 0.5 ml/min using a mobile phase consisting of A, 0.1% formic acid in MeCN and B, 0.1% formic acid (Gradient: Held at 10% A for 5 min, ramped to 100% A in 2 min and held for 2 min, returned to 10% A in 2 min). UV detection was accomplished using the Acquity PDA detector scanning the range from 190-500 nm. Mass detection was accomplished with a Waters XEVO G2 QTOF spectrometer scanning the mass range from m/z 150-1300 in 0.5 sec. 8

Positive ion mode spectra were acquired using sampling cone and capillary voltages of 50 V and 2.8 kV, respectively. Chromatograms were extracted from the total ion chromatogram (TIC) for the calculated exact mass of the singly charged molecular ion of metachelin A (m/z 1095.5197). Accurate masses were determined by combining scans through the extracted ion chromatogram peaks and then using lock mass correction with leucine enkephalin (calculated at m/z 556.2771 for [M+H]+) as the external lockmass standard). Scans for the metachelin A peak were combined and centered using the automatic peak detection function supplied by MassLynx 4.1 software. Metachelin A in the iron-free form was identified by cochromatography with an authentic sample purified as described previously (Krasnoff et al., 2014). The limit of detection (LOD) for metachelin A (defined as signal/noise ≥ 3) using this method was 500 pg on column which established the LOD in the biological sample at 20 µg/L of fermentation filtrate. The identity of metachelin A was confirmed by averaging accurate mass measurements for scans through the putative metachelin A peak in the extracted mass chromatogram; (observed: m/z 1095.5217; calculated for [M+H]+ 1095.5197, Δ = 1.8 ppm) and co-chromatography with authentic metachelin A.

2.5. Ferricrocin identification and quantification Ferricrocin for use as a chromatographic standard was extracted from conidia using a modification of a scheme described previously (Krasnoff et al., 2007; Oide et al., 2007). Briefly, ca. 25 g batches of conidia were transferred to 250 ml polypropylene centrifuge bottle and sequentially extracted in hexane, MeOH, and MeOH/H2O/0.5% HCOOH with vortex mixing (30 sec) followed by sonication for 30 minutes (5 min for the hexane fraction. After direct-infusion 9

ESI-MS analysis, the MeOH/H2O/0.5% HCOOH fraction was dissolved in MeOH, applied to a column of Sephadex® LH-20 and eluted with MeOH. Fractions were monitored by ESI-MS and those bearing the mass signature of ferricrocin were recombined, concentrated, dissolved in mobile phase and (vide infra) fractionated by repetitive semi-preparative reversed-phase HPLC on a Phenomenex Prodigy® ODS3 column (10 x 250 mm; 5 μm; 100 Å) eluted at 4 ml/min with a mixture of MeCN : H2O : formic acid (150 : 850 : 2, v/v). Detection by UV absorbance at 435 nm, which affords a good S/N ratio for ferrated siderophores, revealed a single major peak that was collected and concentrated to yield purified ferricrocin in its iron bound form. The identification of ferricrocin was established by verifying its molecular formula by high resolution ESI-MS and by co-chromatography with an authentic standard (kindly provided by Hubertus Haas, Innsbruck Medical University, Austria). Quantitative comparisons of WT and KO strains were made using crude extracts of mycelium. 100 ml batches of AsMM + 10 µM FeCl3 were inoculated with 106 conidia and shaken at 150 RPM at 25 °C for 19 days. Cultures were then filtered through 2 layers of Whatman #1 and the mycelium was lyophilized. The dried mycelium was weighed (0.31 - 0.49 g/100 ml), ground with mortar and pestle, and then extracted as described for extraction of conidia. After weighing, the MeOH/H2O/0.5% HCOOH fractions were redissolved in 2 ml MeOH and diluted 10-fold in 13% MeCN in 0.1% FA (0.4-1 mg/ml) and 5 µL aliquots were analyzed by HPLC-MS. The identity of ferricrocin in crude extracts was confirmed by averaging accurate mass measurements for 14 scans through the putative ferricrocin peak in the extracted mass chromatogram (observed: m/z 771.2486, 793.2313, and 809.2034; calculated for ferricrocin

10

m/z 771.2487, Δ = 0.1 ppm; 793.2306, Δ = -0.9 ppm; and 809.2046, Δ = -1.5 ppm, for [M+H]+, [M+Na]+ and [M+K]+, respectively, and co-chromatography with authentic ferricrocin.

2.6. Phenotypic analyses Sensitivity to the membrane impermeable Fe2+ chelator bathophenanthroline disulfonic acid (BPS), the membrane permeable iron chelator 2,2'- dipyridyl (DP), sodium dodecyl sulfate (SDS), calcofluor white, congo red, and hydrogen peroxide were tested in AsMM plus 1% agarose as the solidifying agent. Assays were conducted in 12-well (3.5 ml medium/well) or 24-well (1.5 ml medium/well) microtiter plates inoculated with 0.5 μl of a 107 conidia/ml suspension in 0.01% Silwet L77 placed at the center of the well followed by incubation in the dark at 25°C for 5 to 14 days, depending on the assay. Colony growth was measured as the average of 2 orthogonal radii. Sensitivity to iron depletion was tested in 24-well plates containing AsMM broth (18 nM) where iron was added up to 500 nM. After 6 days at 25°C, with shaking at 90 RPM, mycelium growth was measured by absorbance at 900 nm (OD900) using a plate reader (Bio-Tek Synergy HT), with 4 replicates per treatment. ED50 were estimated using a logistic 3-parameter sigmoidal model in JMP PRO 11.0.0 (SAS Institute Inc., Cary, NC). Hydrogen peroxide sensitivity of conidia was tested by exposing 2x106 spores/ml suspensions to hydrogen peroxide for 30 min at RT followed by two consecutive washes in ~10 and ~100 volumes of 0.05% Silwet-L77 and recovery by centrifugation for 2 min at 6,000 x g. Conidia were resuspended in 200 μl AsMM broth plus 10 μM FeCl3 and 100 μL were transferred to a 96-well plate containing the same medium. After incubation for 36h at 25°C in the dark Growth was assessed at OD900 after resuspending the biomass by pipetting, and using 4 replicates per 11

treatment. Data was fit using a Gompertz 3-parameter model in JMP PRO 11.0.0 (SAS Institute Inc., Cary, NC). Germination tests were conducted in 24-well microtiter plates using ~5x104 conidia in 250 μl/well of AsMM broth amended with 0, 0.1, 1, 10, and 100 μM FeCl. Cultures were incubated for 16 h at 20 or 23 °C with no agitation followed by trypan blue staining, one rinse with 50 mM NaOH. Digital imaging at 400x magnification was accomplished on an inverted microscope equipped with a MagnaFire digital camera. Pictures were subsequently processed with CellProfiler 2.1 automated pipelines developed to identify and measure size and shape of both non-germinated and germinated conidia (Carpenter et al., 2006; Wahlby et al., 2012). A conidium was considered germinated if the length of its major axes exceeded the sum of the length and width of the average WT conidium. Statistical analyses were conducted using JMP PRO 11.0.0 (SAS Institute Inc., Cary, NC). Dose-response curve models were fit using sigmoidal models available in JMP PRO and selected by comparing the respective Akaike Information Criterion (AIC). Parameters and comparison tests were considered significant at p < 0.05.

2.7. Insect virulence assays Virulence assays on 3rd instar BAW larvae (Benzon Research, Carlisle, PA) were conducted as described in (Giuliano Garisto Donzelli et al., 2012) using inocula containing 104 to 107 conidia/ml in 0.01% Silwet L-77 (Loveland Industries Inc., Greeley, CO) or carrier only. Each inoculum concentration was applied to batches of 24 larvae followed by incubation at 25°C and 15 h:9 h light:dark photoperiod. Each treatment had at least 3 independent replicates. LD50 for each strain and replicate was estimated by probit analysis. The effect of strain was tested by 12

comparison of the estimated LD50 at day 7 using ANOVA followed by the Dunnett post hoc test where WT was used as the control. Statistical analyses were conducted with JMP PRO 11.0.0 (SAS Institute Inc., Cary, NC) and p < 0.05.

2.8. Gene expression by Reverse Transcription Quantitative-Polymerase Chain Reaction (RTqPCR) To compare gene expression in the iron deplete and replete conditions, WT, ΔmrsidC, ΔmrsidD and ΔmrsreA strains were inoculated in 6-well microtiter plates carrying 5 ml/well AsMM medium in the presence of 100 nM iron, at 1x104 conidia/well. Cultures were shaken at 90 rpm at 25°C. At this iron concentration, all strains grew very slowly and especially ΔmrsidC, requiring 6-7 days of growth to obtain workable biomass. Fungi were then transferred to 6-well microtiter plates carrying 5 ml/well fresh AsMM containing either 18 nM or 100 μM iron and incubated for 6 h at 90 rpm at 25°C before RNA extraction. In vivo expression of mrsidC, mrsidD and mrsreA and insect TFs was assessed in BAW dusted with M. robertsii conidia to assure a high fungal burden. Treated larvae and non-inoculated controls were sampled at 24, 48, 72, 96 and 120h. The 120h time point had 2 sub-samples: one included larvae that were still alive (120h sample) and the other contained larvae that appeared to be dead (120D). Before RNA extraction, all but the 24h samples (when fungal penetration is still occurring) were surface sterilized with 20% commercial bleach plus 0.05% Silwet L77 for 30 min at RT and quickly rinsed twice with sterile distilled water. RNA extraction and cDNA synthesis was performed as described (Giuliano Garisto Donzelli et al., 2012). Primers (Table 1) were designed using PrimerSelect (Lasergene), tested for amplification efficiency and specificity using genomic DNA 13

and cDNA from both M. robertsii and BAW as the templates. Reactions were performed with SYBR Universal SYBR Green Supermix (Bio-Rad Cat# 172-5121) and iQ5 thermocycler (BioRad) using a two-step amplification protocol and 58°C as the annealing temperature. Ct values obtained for either M. robertsii or S. exigua transcripts were normalized using Ct values for L32 ribosomal proteins from the respective organisms. Initial tests using 3 housekeeping normalizing genes were not significantly different from single gene normalization. Each treatment had 2 technical replicates and 3 biological replicates. Relative expression calculation and statistical analyses were conducted using REST 2009 (Pfaffl et al., 2002).

14

3. RESULTS 3.1. mrsidC, mrsidD and mrsreA are part of conserved gene clusters sidD/nps6, sidC/nps2 and sreA/sre1 orthologs in M. robertsii genome were identified by reciprocal best blast and correspond to M. robertsii genome accessions EXV04699, EXV05490 and EXV00864, respectively (Oberegger et al., 2001; Oide et al., 2007; Oide et al., 2006; Schrettl et al., 2007; Wall et al., 2003). Domain analysis based on InterproScan and Genbank Conserved Domain Database signatures indicated that mrsidD encodes a predicted 1,800 aa NRPS with a domain structure identical to that found in other coprogen-assembling NRPSs (ATC TC, where A = adenylation domain; T = thiolation domain and C = condensation domain). MRSIDC is a 4,845 aa protein with type IV ferrichrome synthetase architecture (ATC ATC TC ATC TC TC) (Bushley et al., 2008). Predicted MRSREA carries 2 zinc finger domains sharing the C-x2-C-x17-C-x2-C motif with its orthologs (Harrison, 1991; Teakle and Gilmartin, 1998). mrsidC is located between a putative glycine-betaine biosynthesis cluster (badA and codA, (Lambou et al., 2013)), one of 2 candidate orthologs of the triacetylfusarinine esterase sidJ on one side and sidA (L-ornithine N5-monooxygenase) ortholog on the other (Supplementary Fig. S1A) (Grundlinger et al., 2013). Gene organization around mrsidD is conserved in related fungi irrespective of their lifestyle, but is not shared by more distant species such as A. fumigatus and C. heterostrophus (Supplementary Fig. S1B). The association of mrsidD with the sidF (triacetylfusarinine transacylase) and sidI (triacetylfusarinine enoyl-CoA hydratase) orthologs mrsidF and mrsidI is also conserved in most of the examined Hypocrealeans with the exception of G. zeae, C. purpurea, F. oxysporum and E. festucae where either the sidD/nps6 ortholog is located in clusters more similar to that found in A. fumigatus (where sidF, sidH, sidD, sit1, mirD are co15

localized) or, in the case of E. festucae, where sidD/nps6 is absent as the ferrichrome-type siderophore, epichloënin acts as the extracellular siderophore (Johnson et al., 2013). The ABC transporter EXV04695.1 is the ortholog of A. fumigatus Afu3g03670, which is upregulated under iron deficiency; it shares a divergently transcribed promoter with EXV04696.1 which in turn is the M. robertsii ortholog of Afu1g17180, an sreA-regulated oxidoreductase adjacent to sidI and sidC in the A. fumigatus genome (Schrettl et al., 2008). No high confidence orthologs for either sidH (enoyl-CoA hydratase, Afu3g03410) or sidG (transacetylase, Afu3g03650) were found in M. robertsii. mrsreA is located in a third distinct region of the M. robertsii genome. This region is also conserved among Hypocreales and contains several genes putatively associated with development and epigenetic regulation (Supplementary Fig. S1C). We used the upstream regions of 12 M. robertsii orthologs of iron-regulated genes characterized in either Aspergillus spp. and/or Saccharomyces cerevisiae to identify putative regulatory motifs and their recurrence. RSAT analysis found 2 significant motifs (Supplementary Fig. S2A). The first motif (GATCWGATAA) is virtually identical to that identified in A. fumigatus and was found in all the test sequences with the exception of the fre3/ctr3 promoter region (Schrettl et al., 2008). The second motif (ATCAATC) is found in all of the test sequences (Supplementary Fig. S2B) which is nearly identical to the binding site for the iron activator hapX (Hortschansky et al., 2015). Phylogenetic analysis of mrsreA orthologs from 43 Ascomycetes displayed a high degree of congruence to that inferred using concatenated rpb1, rpb2, and tef1 (Nye topological score = 90%, for identical trees the score is 100%) (Supplementary Fig. S3).

16

3.2. Identification of deletion mutants PCR screening identified several KO candidates for each of the targeted genes. After single conidial isolation, selected isolates were confirmed to carry a single insertion of the bar gene at the targeted locus (Supplementary Fig. S4). Exceptions were the mrsreA deletion mutant S18 and the ectopic integrant S26 in which at least two copies of the transgene were detected. Isolates C7, D3, and S35, which are mutants for mrsidC, mrsidD and mrsreA genes, respectively, were used for further analyses.

3.3. Effect of mrsidC, mrsidD and mrsreA deletion on siderophore production in M. robertsii Growth of WT in AsMM broth containing 50 nM iron for 12 days resulted in the accumulation of an estimated 10.0 mg/L broth metachelin A along with related glycosylated and unglycosylated coprogen-type siderophores reported previously (Fig. 1) (Krasnoff et al., 2014). No metachelin A or related compounds were detected in the extracts from the ΔmrsidD mutant D3 (Fig. 2A) or other independently obtained mrsidD knockouts (data not shown). WT conidia and mycelium also produced ferricrocin. Ferricrocin content in extracts obtained from mycelium recovered per liter of AsMM amended with 10 µM iron was estimated at ca. 12.3 µg, whereas no ferricrocin was detected in mycelial extracts from ΔmrsidC strains (Fig. 2B). Deletion of mrsreA had no significant effect on siderophore production in iron deplete AsMM (data not shown). However, when tested in iron replete media, ΔmrsreA strains displayed an increased capacity to solubilize environmental iron compared to the WT (Fig. 2C).

3.4. mrsidC, mrsidD and mrsreA mutants display differential sensitivity to iron availability. 17

Deletion of mrsidC, mrsidD had no impact on radial growth on AsMM containing either 10 or 1 µM iron. ANOVA indicated that on the same medium ΔmrsreA strains showed ~10% (p60% compared to the untreated control. CW provided complete inhibition of the ΔmrsidD strain regardless of the iron content of the medium, while CR inhibition of the same strain depended on iron content of the medium (p = 0.05) (Supplementary Fig. S5). The increased sensitivity to both CW and CR by the ΔmrsidC mutant was not statistically significant, while loss of mrsreA produced a small but significant decrease in iron-associated sensitivity to CR (p=0.0014) but not to CW. In +Fe medium all tested strains displayed similar sensitivity to SDS. However, deletion of mrsidD and, to a lesser extent, deletion of mrsidC, significantly increased sensitivity to SDS when iron availability was reduced (Supplementary Fig. S5).

3.7. ΔmrsidC, ΔmrsidD and ΔmrsreA mutations affect germination. After 15h incubation in AsMM an average of 88% WT conidia had germinated as compared to 79.7% for ΔmrsreA, 81.6% for ΔmrsidD and 60.9 % for ΔmrsidC (Fig. 5A), and all differences were statistically significant (p 3`(1)

Remarks

MRSidC-AF2 MrsidC-AR3 MrsidC-BF MRSidC-BR2 MrSidD-AF MrSidD-AR MrSidD-BR MrSidD-BF SreAKO-AF SreaKO-AR SreaKO-BF SreaKO-BR BarExprS-F BarExprSR-NOS MrSidC 5062F

GGGAAAGdUTCTGCTCATCACGCTG ATCATCCdUTACTTTGCAAGACGGCTCT ACTTGTGGdUTCTTGTTGAAGACGAGGTGGC GGAGACAdUATTTCGTCTTGAGAGTGGCA GGGAAAGdUAATCAAATGGGGAAATCTCTG ATCATCCdUGCATGCATCTTTGTATCAGTG GGAGACAdUGACGTGCCAAATAATGCCAA ACTTGTGGdUCTCCAGCAGATAGCCGTAAAG GGGAAAGdUTTTTTGCGCAAGTGGTG ATCATCCdUATGTGCAAATGCGCTGGTAG ACTTGTGGdUGATCTAGCGGTCAGGT GGAGACAdUCAGAATCCATAAAGAAATAC AGGATGAdUAGAAGATGATATTGAAGGA ACCACAAGdUCATGTTTGACAGCTTATCAT GCGGGCAATGCTGGCTCAAA

MrSidC 5760R

CCCGAAACAACCGAACCGAGTAAG

MrSidD 3159F

GGAAGCCGACCGAATGAACG

MrSidD 3707R

GCTGACCCCTGATTTTGACTTGTT

MrSreA 1093F

CAAACAAACGACCACTGACAACTG

MrSreA 2733R

CGGAGAAAGAAAAGAGGCAGGAGA

BarF BarR MrSreaScreenF

GTCTGCACCATCGTCAAC CGTCATGCCAGTTCCCGT GCGTCCCCACAAACCTGAGAACTT

MrSreaScreenR

CGCGACGACTTCCCACCATC

MrSidC-ScreenF

ATGGCATAGTACTCCGTCCCTCTG

MrSidC-ScreenR

CATCATGCCTGGCTAGAAAGTGG

MrSidD754F

ACCAGTCAGTCAATCGAGTCAATA

mrsidC KO left flank mrsidC KO left flank mrsidC KO right flank mrsidC KO right flank mrsidD KO left flank mrsidD KO left flank mrsidD KO right flank mrsidD KO right flank mrsreA KO left flank mrsreA KO left flank mrsreA KO right flank mrsreA KO right flank bar cassette (all KO vectors) bar cassette (all KO vectors) DIG-labeled probe for mrsidC DIG-labeled probe for mrsidC DIG-labeled probe for mrsidD DIG-labeled probe for mrsidD DIG-labeled probe for mrsreA DIG-labeled probe for mrsreA DIG-labeled probe for bar DIG-labeled probe for bar KO screening mrsreA. Anneals outside the targeted region (2) KO screening mrsreA. Anneals within fragment replaced by the bar gene (2) KO screening mrsidC. Anneals outside the targeted region (2) KO screening mrsidC. Anneals inside the targeted region (2) KO screening mrsidD. Anneals outside the targeted region (2)

49

MrSidD7845R

ACGAGAAAGGTGCTACGACAA

bar848F

ACTGGCATGACGTGGGTTTCTGG.

PTRPC80R

CCGCCTGGACGACTAAACC

Mr-L32-268F

AACCGAACCTACGCCGCTGAGATT

Mr-L32-351R

GCCTTGGCATTTGTGACCTTGACT

MrSidC5231R MrSidC5062F MrSreA-1804F MrSreA-2008R MrSidD-3827F MrSidD-3938R BAW-L32-324F

TGGCTCCGAAAATAACACAGACAC GCGGGCAATGCTGGCTCAAA CGCCGAAGCCGACACCCTCTGGA GTTGAGCTGGGCAGTTTTTGACA CGCCGACGCTGGTTGCCTTTATT CCTTTGCGCCCATGTCCTTGTCTA ACAATGTCCGTGAGCTGGAGAT

BAW-L32-473R

GCAGACGGGCCGCAGAGTT

BAW-TF1-F BAW-TF1-R BAW-TF2-F BAW-TF2-R BAW-TF3-F BAW-TF3-R BAW-TF4-F BAW-TF4-R

CGTTTTATCGACTCATCCCTACAACA AGAGTTACCAGAAACACGAAGAAAGA CGCAGCTGTCGCCTTTGAGAG TCGGCTTTATAGGGATGTTGTCG ACGATGCTGGCTGGTGGTG CGAGCGGATTGGCGATGTGTA ACATAGCTTGCATAATTTGTCAGAGTTA CATGTTTTAATGCAAGAAGGAGGTT

(1) dU = deoxyuridine

50

KO screening mrsidD. Anneals outside the targeted region (2) Anneals towards the 3` end of the bar gene (2) Anneals towards the 5` end of the trpc promoter in bar expression cassette (2) M. robertsii ribosomal protein L32, qPCR M. robertsii ribosomal protein L32, qPCR mrsidC, qPCR mrsidC, qPCR mrsreA, qPCR mrsreA, qPCR mrsidD, qPCR mrsidD, qPCR BAW ribosomal protein L32, qPCR BAW ribosomal protein L32, qPCR BAW transferrin 1, qPCR BAW transferrin 1, qPCR BAW transferrin 2, qPCR BAW transferrin 2, qPCR BAW transferrin 3, qPCR BAW transferrin 3, qPCR BAW transferrin 4, qPCR BAW transferrin 4, qPCR

Highlights 

M. robertsii core biosynthetic genes for metachelin A and ferricrocin were deleted.



Null mutants for the transcriptional repressor mrsreA were also produced.



Metachelins were found to be dispensable for virulence while ferricrocin was not.



M. robertsii mrsreA mutants were unaltered in virulence.

51