Loss of CclA, required for histone 3 lysine 4 methylation, decreases growth but increases secondary metabolite production in Aspergillus fumigatus

Strains, growth conditions, and DNA manipulations

All A. fumigatus strains used in this study are derivatives of the clinical isolate AF293 (Xue et al., 2004) and are listed in Table 1. Strains were maintained on glucose minimal media (GMM) (Shimizu & Keller, 2001) and when appropriate the media was supplemented with 5 mM uracil and 5 mM uridine for pyrG1 strains, 5 mM arginine for argB1 strains, and hygromycin was used at a final concentration of 100 ug/ml. Radial growth experiments and 6-azauracil sensitivity assays were carried out essentially as in Palmer et al. (2008). All DNA manipulations were conducted according to standard protocols (Sambrook & Russell, 2001) and fungal transformation was done as previously described (Miller, Miller & Timberlake, 1985) with a minor modification of protoplasts being plated in 0.75% molten top agar. Primers used in this study are listed in Table 2.

Construction of mutant A. fumigatus strains

An A. fumigatus cclA disruption cassette was constructed using double-joint PCR consisting of the following: 1-kb DNA fragment upstream of the cclA start codon (primers Fbre5F and Fbre5R-EcoRI), a 3-kb EcoRI-HindIII DNA fragment of A. parasisitus pyrG via pJW24 (Calvo et al., 2004), and a 1-kb DNA fragment downstream of the cclA stop codon (primers Fbre3F and Fbre3R-HindIII). The fusion PCR product was then used to transform AF293.1 to prototrophy creating TJW84.1. Homologous single-gene replacement of cclA was confirmed by Southern analysis (Figs. 1A and 1B). In order to complement the ΔcclA mutant, a complementation plasmid (pJW109.5) was constructed. The plasmid contained a 3.8-kb wild-type cclA gene including a 1-kb native promoter and a 1-kb native termination cassette, which was amplified by primers FbreCOMF-NotI and FbreCOMR-SpeI. The PCR product was cloned into a NotI-SpeI site of pUCH2-8 (Alexander, Hohn & Mccormick, 1998), which contains the selectable marker hygromycin B phosphotransferase. TJW84.1 was transformed with pJW109.5 to hygromycin resistance and Southern blot confirmed single copy integration for TJW88.21 (Fig. 1B).

An hdmA disruption cassette was constructed by PCR amplifying a 1.1-kb DNA fragment upstream of the hdmA ORF using the primer pair JP1 and JP2 containing EcoRI restriction sites. The subsequent 1.1-kb EcoRI-EcoRI fragment was cloned into pJW24 to create pJMP1. A downstream 1.0-kb fragment was amplified using the primer pair JP3 and JP4 containing XbaI and NotI restriction sites respectively. The 1.0-kb XbaI–NotI fragment was cloned into pJMP1 to create the hdmA disruption cassette named pJMP2. The hdmA ORF was disrupted by transformation of linearized pJMP2 into AF293.6 creating the ΔhdmA argB1 auxotrophic strain (TJMP2.36), which was confirmed by Southern blot (Figs. 1C and 1D). In order to generate a prototrophic deletion strain, an argB complementation plasmid was constructed by amplifying a 2.6-kb fragment corresponding to A. fumigatus argB using the primer pair ‘JP Afumi argB For’ and ‘JP Afumi argB Rev’, which was subsequently cloned into pCR2.1 TOPO (Invitrogen) to create pJMP4. This plasmid was transformed into TJMP2.36 to generate a prototrophic ΔhdmA strain (TJMP3.52). For complementation of ΔhdmA, a 4.6-kb fragment corresponding to ∼0.9-kb up stream and ∼0.5-kb downstream of the hdmA ORF was amplified with the primer pair ‘JP lsdA comp For’ and ‘JP lsdA comp Rev’ each containing a BamHI site. The subsequent BamHI-BamHI fragment was cloned into pJMP4 to create pJMP13. TJMP2.36 was transformed with pJMP13 to create a complemented control strain TJMP29.1 (ΔhdmA  + hdmA) and single integration was confirmed by Southern blot (Fig. 1E).

Physiology experiments

Radial growth measurements and sensitivity to 6-azauracil (6AU) was conducted as described previously (Palmer et al., 2008). Four replicates were used for each assay and statistical significance was calculated with ANOVA analysis using Prism 5 software (Graph Pad).

Alterations in H3K4 methylation

Nuclear extracts were isolated as previously described (Palmer et al., 2008). Approximately 50 µg of nuclear protein extract was electrophoresed on a 10% Tricine-SDS-PAGE gel (Schägger, 2006) and subsequently electroblotted to nitrocellulose membranes. Detection of H3K4 modifications was conducted using the following primary antibodies and dilutions: 1:1,000 anti- histone 3 (Upstate, #07-690), 1:1,000 anti-H3K4 mono-methylation (Upstate, #07-436), 1:2,000 anti-H3K4 di-methylation (Upstate, #07-030), 1:2,000 anti-H3K4 tri-methylation (Upstate, #07-473). Chemiluminescent detection was employed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) and a secondary goat anti-rabbit-horseradish peroxidase conjugate antibody (Pierce, #31460) diluted 1:15,0000 (Fig. 2).

Northern analysis and secondary metabolite extraction

Fifty milliliter cultures of liquid GMM + 0.1% yeast extract were inoculated with 1 × 10⁷ spores per ml and incubated at 250 rpm 29 °C for 24 h, followed by a reduction in temperature to 25 °C for an additional 48 h. Mycelia were harvested, lyophilized overnight, and total RNA was extracted using Isol-RNA Lysis Reagent (5 Prime) according to manufacturers recommendations. Subsequent northern analysis was done using radiolabeled probes for the corresponding transcript (Fig. 4C) (primers are listed in Table 2). Secondary metabolites were extracted from 20 ml of culture filtrate with an equal volume of chloroform. The air-dried chloroform layer was resuspended in 75 µl chloroform and 5 µl was separated on a thin layer chromatography plate (Whatman, #4410 222) using chloroform: acetone (7:3) as a solvent. The TLC plate was then dried and imaged under 254 nm and 366 nm ultraviolet light (Fig. 4A). Organic extracts from solid minimal medium cultures was accomplished by taking equal sized cores from the center of point-inoculated cultures, homogenized in 3 ml of water, and extracted with an equal volume of chloroform. Quantification of TLC spots was achieved through densitometry analysis using Image J and normalized to wild type levels.

Pathogenicity assays

The neutropenic murine model of aspergillosis was conducted as described previously (Bok et al., 2005) with approval of the William S. Middleton VA and the University of Wisconsin Animal Care Committees in accordance with the MV2344 animal use protocol. Toll-deficient flies were generated by crossing flies carrying a thermosensitive allele of Toll (Tl r632) with flies carrying a null allele of Toll (Tl I-RXA) (Lionakis et al., 2005). Two- to four day old adult female Toll-deficient flies were used in all of the experiments. Twenty flies were infected with each A. fumigatus strain used in this study. All of the experiments were performed in triplicate. To minimize circadian rhythm variability, all experiments were performed at 9 AM. A. fumigatus isolates were grown on yeast extract agar glucose (YAG) at 37 °C. Conidia were collected in sterile 0.9% saline from 2 days old cultures. The conidial concentration suspension was determined by using a hemacytometer and adjusted to 1 × 10⁷ per ml. The dorsal side of the thorax of twenty CO₂ anesthetized flies was punctured with a thin (10 µm) sterile needle that had been dipped in a concentrated solution of A. fumigatus conidia (10⁷ cells/ml). In our previous work, this method was shown to deliver a reproducible inoculum of 700 to 800 conidia (Lionakis et al., 2005). As a negative control group, Toll-deficient flies were punctured with a 10 µm sterile needle and monitored daily for survival. Flies that died within 3 h of the injection were considered to have died as a result of the puncture procedure and were not included in the survival rate analysis. The flies were housed in a 29 °C incubator to maximize expression of the Tl r632 phenotype (Lemaitre et al., 1996). The Toll-deficient flies were transferred into fresh vials every 3 days. Fly survival was assessed daily over 8 days.

Deletion of cclA but not hdmA affects global histone 3 lysine 4 methylation patterns

Aspergillus fumigatus cclA was identified through BLASTp analysis of the A. fumigatus genome with the A. nidulans CclA amino acid sequence (Bok et al., 2009), which yielded one homolog Afu3g04120. The human LSD1 protein sequence (Genbank, #O60341) was used to BLASTp the A. fumigatus genome sequence, which provided one potential ortholog, Afu4g13000. Afu4g13000 is 29% identical to human LSD1 and contains the conserved SWIRM domain, a flavin amine oxidase domain, as well as an HMG Box domain (Shi et al., 2004) and thus was named hdmA for histone demethylase A.

The cclA and hdmA alleles were independently disrupted in the A. fumigatus AF293 genetic background. Southern analysis of transformants for each gene confirmed simple gene replacements (Fig. 1). One representative deletant, TJW84.1 for ΔcclA and TJMP3.52 for ΔhdmA, was chosen for each gene replacement for the studies presented below. Both deletants were complemented with their respective genes to generate a single copy cclA  + ΔcclA strain and a single copy hdmA  + ΔhdmA strain as shown in Table 1.

Our first experiment was to assess if loss of either gene affected histone 3 lysine 4 methylation (H3K4). The ΔcclA strain showed a clear loss of whole genome H3K4 tri-methylation and a considerable decrease in di-methylation (Fig. 2); this tallied with earlier studies in our lab assessing H3K4 methylation in an A. nidulans ΔcclA mutant (Bok et al., 2009). These data are also consistent with BRE2 mutants in Saccharomyces cerevisiae, where loss of BRE2 results in strains with diminished di-methylation and no tri-methylation of H3K4 (Schneider et al., 2005). The complemented strain was restored in H3K4 methylation to wild type levels. However, loss of hdmA did not exhibit a detectable impact on whole genome H3K4 methylation (Fig. 2).

CclA mutants are crippled in growth and sensitive to 6-Azauracil

As reported for the A. nidulans ΔcclA mutant (Giles et al., 2011), the A. fumigatus ΔcclA strain exhibited poor growth as exhibited by decreased radial growth on several different media, including rich media (YPD and Champs) as well as our standard minimal medium (GMM) (Fig. 3A). Additionally, the ΔcclA mutant had a measurable growth defect at all temperatures tested (Fig. 3B) and decreased mass when grown in liquid shaking culture (data not shown). This defect was rescued in the complemented strains. Neither the hdmA deletant nor its complement exhibited any growth phenotype.

The phenotype of the ΔcclA strain shared some similarities to that of a previously described A. fumigatus chromatin mutant, ΔclrD (Palmer et al., 2008), impaired in methylation of H3K9. As histone methyltransferase mutants are often found to be sensitive to chemicals that target various cellular processes, the ΔclrD mutant had been assessed for sensitivity to various chemical agents and found to be sensitive to 6-azauracil (6AU); an inhibitor of guanine nucleotide synthesis and indicator of transcriptional defects (Riles et al., 2004; Zhang et al., 2005; Palmer et al., 2008). Following that protocol, the ΔcclA and ΔhdmA strains were similarly screened and the former but not the latter found to be also sensitive to 6AU (Fig. 3C). Resistance to 6AU was rescued in the complement strain. Moreover, we did not find differences in sensitivity of ΔhdmA or ΔcclA strains compared to wild type when exposed to several chemical agents that included hydroxyurea (DNA synthesis inhibitor), thiabendazole (microtubule destabilizer), 1.2 M Sorbitol (osmotic stress), 0.6 M KCl (osmotic stress), Congo red (cell wall inhibitor), or Calcofluor white (cell wall inhibitor) (data not shown).

CclA loss results in increased secondary metabolism

Mutations in chromatin remodeling genes have frequently been associated with alterations in secondary metabolite production, both in A. nidulans (Shwab et al., 2007; Bok et al., 2009) and A. fumigatus (Lee et al., 2009). In particular the A. nidulans ΔcclA strain was found to produce novel metabolites with antimicrobial properties (Bok et al., 2009; Giles et al., 2011). An examination of extracts from three different conditions of the two mutants showed that the A. fumigatus ΔcclA strain but not the ΔhdmA strain was increased in the production of several metabolites (Fig. 4). In particular, this strain produced over 4 times as much gliotoxin compared to wild type when grown in liquid shaking culture (Figs. 4A and 4B). The high level of gliotoxin was associated with a large increase in transcription of gliZ, the C6 transcription factor required for gli expression (Fig. 4C). Under these growth conditions, loss of cclA also resulted in several fold increases of numerous other metabolites (Figs. 4A and 4B). The ΔcclA mutant was also analyzed under alternative growth conditions and produced similar amounts of metabolites on solid medium at 29 °C compared to wild type (Fig. 4D), however ΔcclA mutants produce more of a few unidentified metabolites when grown 37 °C on solid medium (Fig. 4D).

Virulence of ΔcclA is unaltered in a mammalian and Drosophila model

Although poorly growing A. fumigatus mutants have often been found to exhibit a decrease in virulence (D’enfert et al., 1996; Brown et al., 2000), the increased toxin production in the ΔcclA strain coupled with the poor growth presented a unique phenotype that disallowed an obvious prediction on its pathogenicity attributes. Therefore, to assess cclA or hdmA loss on virulence, we examined virulence in two animal models. Assessment of the mutants in a neutropenic murine model showed no significant difference between the wild type and mutant strains (Fig. 5A). This was also true for the Toll-deficient Drosophila model (where only ΔcclA was assessed, Fig. 5B).