Abstract
Malassezia encompasses a monophyletic group of basidiomycetous yeasts naturally found on the skin of humans and other animals. Malassezia species have lost genes for lipid biosynthesis, and are therefore lipid-dependent and difficult to manipulate under laboratory conditions. In this study we applied a recently-developed Agrobacterium tumefaciens-mediated transformation protocol to perform T-DNA random insertional mutagenesis in Malassezia furfur. A total of 767 transformants were screened after exposure to 10 different stresses, and the 19 mutants that exhibited a phenotype different from the wild type were further characterized. The majority of these strains had single T-DNA insertions, which were identified within the open reading frames of genes, within untranslated regions, and in intergenic regions. Some T-DNA insertions generated chromosomal rearrangements, and others could not be characterized. To validate the findings of the forward genetic screen, a novel CRISPR/Cas9 system was developed to generate targeted deletion mutants for 2 genes identified in the screen: CDC55 and PDR10. This system is based on co-transformation of M. furfur mediated by A. tumefaciens to deliver both a CAS9-gRNA construct that induces double-strand DNA breaks, and a gene replacement allele that serves as a homology directed repair template. Targeted deletion mutants for both CDC55 and PDR10 were readily generated with this method. This study demonstrates the feasibility and reliability of A. tumefaciens-mediated transformation to aid in the identification of gene functions in M. furfur through both insertional mutagenesis and CRISPR/Cas9-mediated targeted gene deletion.
Introduction
The genus Malassezia is a lipophilic, monophyletic group of basidiomycetous yeasts that colonize sebaceous skin sites and represents more than 90% of the skin mycobiome (Findley et al. 2013; Wu et al. 2015; Byrd et al. 2018). In addition to a ubiquitous presence on the skin of human and animals, recent data support the hypothesis that Malassezia fungi are much more widespread than previously thought. Metagenomics studies revealed the presence of Malassezia DNA in a number of unexpected areas such as in association with corals and sea sponges in the ocean, although Malassezia marine species have yet to be isolated in axenic culture (Amend et al. 2019). There are currently 18 species within the Malassezia genus. One defining characteristic of the Malassezia genus is the lack of a fatty acid synthase, Δ9 desaturase, and Δ2,3 enoyl CoA isomerase, making them lipid-dependent and difficult to study and manipulate under laboratory conditions. Malassezia are highly divergent from other fungi that are commonly found on the skin, such as Candida species and the dermatophytes. Furthermore, Malassezia species belong to the Ustilaginomycotina subphylum, which includes the plant pathogens Ustilago, Sporisorium, and Tilletia, and are highly divergent from other basidiomycetous fungi that infect humans, such as Cryptococcus neoformans. Recent classifications revealed that Malassezia represents a sister group to the blast yeast-like fungi Moniliella (Wang et al. 2014; Wang et al. 2015), which includes species reported to be pathogenic on human and animal skin (McKenzie et al. 1984; Pawar et al. 2002) as well as others that are of interest in sugar alcohol production in industrial settings (Kobayashi et al. 2015).
In the last decade, there has been increasing scientific interest in Malassezia, with several sequencing projects aimed at defining genomic features and gene content for 15 broadly recognized Malassezia species (Xu et al. 2007; Gioti et al. 2013; Triana et al. 2015; Wu et al. 2015; Park et al. 2017; Zhu et al. 2017; Kim et al. 2018; Lorch et al. 2018; Cho et al. 2019; Morand et al. 2019). All haploid Malassezia species have small genomes compared to other phylogenetically related fungi (7 to 9 Mb compared to ~20 Mb), and have lost genes involved in carbohydrate metabolic processes and hydrolysis activity. Genome analyses have revealed intriguing features, such as i) loss of the RNA interference pathway components; ii) evidence of horizontal gene transfer events from bacteria; iii) the presence of genes unique to Malassezia; and iv) the expansion of secreted protein, lipase, and protease gene families that encode products predicted to breakdown lipids and proteins important for growth and host and microbial interactions.
The typical Malassezia genome is between ~7 and 9 Mb, which is about half the size of other basidiomycetous fungi, with the exception of M. furfur hybrid species whose genomes are twice the size of other Malassezia species. It is likely that the genomes of Malassezia species have reduced over time concomitantly with their evolution as a commensal organism and adaptation to the skin (Wu et al. 2015). There are other cases in which fungal genome reduction correlates with niche specialization, with the most remarkable examples being the obligate Pneumocystis species with genomes of ~7-8 Mb (Ma et al. 2016), and Microsporidia species with genomes as small as 2.9 Mb (Cuomo et al. 2012).
Aside from their commensal lifestyle, Malassezia fungi have been associated with several skin disorders, including pityriasis versicolor, dandruff, severe atopic dermatitis in humans, and otitis in dogs (Gaitanis et al. 2012; Wu et al. 2015). However, the exact role of Malassezia in these clinical conditions has been controversial, with recent studies even hypothesizing a protective role of M. globosa against Staphylococcus aureus, a bacterium that is associated with severe atopic dermatitis (Li et al. 2017; Ianiri et al. 2018). The lack of knowledge regarding Malassezia function within the skin mycobiome is due, in part, to the dearth of experimental systems for studying Malassezia-host interactions; current knowledge is based solely on in vitro experiments with isolated host cells (Watanabe et al. 2001; Ishibashi et al. 2006; Donnarumma et al. 2014; Glatz et al. 2015; Sparber and Leibundgut-Landmann 2017). Recently, 2 groundbreaking studies reported novel experimental murine models for studying Malassezia interactions with the skin and intestinal mucosa (Limon et al. 2019; Sparber et al. 2019). Sparber and colleagues demonstrated that the application of M. sympodialis, M. pachydermatis, and M. furfur on the dorsal ear skin of mice resulted in robust colonization of the epidermis and a rapid cytokine response dominated by IL-17 and related factors. This response was found to be critical for preventing fungal overgrowth on Malassezia-exposed skin and exacerbates inflammation under atopy-like conditions (Sparber et al. 2019). Another study by Limon and colleagues demonstrated the involvement of Malassezia in inflammatory bowel disease. The authors characterized the mycobiome associated with the intestinal mucosa of healthy individuals and patients with Crohn’s disease, and found that M. restricta, one of the most common inhabitants of human skin, was especially abundant in Crohn’s disease patients. Moreover, the presence of M. restricta was linked with a polymorphism in the gene for CARD9, a signaling adaptor critical for defense against fungi (Limon et al. 2019). The importance of these studies has been highlighted in 2 commentaries (Dawson 2019; Wrighton 2019).
Although these models represent an important advance in understanding the mechanisms of host responses to Malassezia, a lack of technologies for functional genetic studies has hampered the identification and characterization of the fungal components that promote inflammation and induce host responses. We were the first group to develop a transformation system based on transconjugation-mediated by Agrobacterium tumefaciens (AtMT, A. tumefaciens-mediated transformation) that is effective for both insertional and targeted mutagenesis and enabled the first genetic manipulation of M. furfur and M. sympodialis (Ianiri et al. 2016; Ianiri et al. 2017). Subsequently, M. pachydermatis has also been transformed (Celis et al. 2017).
CRISPR (clustered regularly interspaced short palindromic repeats)/Cas9 was originally discovered as a mechanism of adaptive bacterial immunity for defense against invading DNA elements (Jinek et al. 2012). The CRISPR/Cas9 system has been modified for use in other organisms, and at present, represents a revolutionary technology that has allowed gene editing in a number of cell types, including fungi (Shi et al. 2017; Adli 2018). The system consists of 2 elements: a specific endonuclease (Cas9) and a guide RNA (gRNA) that form a complex that catalyzes double-strand breaks (DSBs) at a specific DNA site flanking a protospacer adjacent motif (PAM) sequence of the host genome. After the DSB is generated, the DNA can be repaired either through non-homologous end joining (NHEJ) or through homology directed repair (HDR) when donor DNA is provided (Shi et al. 2017).
The present study is divided into 2 sections. In the first, we build upon the previously developed AtMT technology to perform the first T-DNA-mediated genetic screen in M. furfur. The aim was to generate a library of random insertional mutants, select for mutants with a phenotype of interest, and characterize insertion sites within the M. furfur genome to infer the function of genes involved in processes of physiological and clinical interest. In the second part of this study, we developed the first efficient, transient CRISPR/Cas9 mutagenesis system for Malassezia, and successfully generated targeted deletion mutants of two genes identified in the forward screen: CDC55, which encodes a subunit of protein phosphatase 2A (PP2A), and PDR10, which encodes an ABC transporter predicted to be involved in pleiotropic drug resistance. When validating the effectiveness of the T-DNA insertional mutagenesis for gene function studies, this novel CRISPR/Cas9 technology overcomes issues related to the reduced rate of homologous recombination observed in M. furfur, and we expect that it will facilitate molecular research on Malassezia fungi.
Materials and Methods
Strains and culture conditions
The haploid M. furfur strain CBS14141 (previously known as JPLK23) was used as the wild type (WT) strain for transformation experiments. This strain was maintained on modified Dixon’s media (mDixon) [mycological peptone (10 g/L), malt extract (36 g/L), glycerol (2 ml/L), tween 60 (10 ml/L), desiccated ox-bile (10 g/L) and agar (20g/L) for solid media]. Transformants were maintained on mDixon supplemented with the antifungal agents nourseothricin (NAT) or G418 (NEO).
Forward genetics screen in M. furfur
Insertional mutagenesis was performed through AtMT using Agrobacterium tumefaciens strain EHA105 engineered with the binary vectors pAIM2 or pAIM6, which contain NAT and NEO resistance markers under the control of M. sympodialis ACT1 promoter and terminator, respectively (Ianiri et al. 2016). Initially, transformations were performed using previously developed methods (Ianiri et al. 2016; Celis et al. 2017). Selected transformants were colony-purified on selective media and arrayed in 96 well plates containing 100 µL of mDixon + NAT or mDixon + NEO for in vitro assays and long-term storage.
For the primary screen, a 1.5 µL aliquot of cellular suspension of transformants was spotted on mDixon agar containing the following chemicals: Congo red (0.5%), sodium chloride (NaCl, 1M), sodium dodecyl sulfate (SDS, 0.3%), or fluconazole (FLC, 150 µg/ml) for cell wall and plasma membrane stress; NaNO2 (100 mM) for nitrosative stress; or CdSO4 (30 µM) for protein-folding defects and heavy metal stress. Transformants were also exposed to UV light (250 to 450 µJ × 100), elevated temperature (37°C), pH (pH 7.5), and nutrient-limiting conditions [yeast nitrogen base media (YNB)]; when used, arginine and tyrosine were added at 30 mg/L. Transformants selected in the primary screen as having a phenotype different than the WT were confirmed through a standard 1:10 serial dilution method by spotting 1.5 µL of cellular suspension on mDixon agar in the conditions that allowed their selection.
Molecular characterization of the T-DNA insertional mutants of M. furfur
Insertional mutants with a phenotype of interest were single-colony purified and grown overnight in 25 mL of liquid mDixon for genomic DNA extraction using a CTAB extraction buffer (Pitkin et al. 1996). To identify the insertion sites of the T-DNA in the M. furfur genome, inverse PCR (iPCR) was performed according to previously published methods (Idnurm et al. 2004; Ianiri and Idnurm 2015). Briefly, approximately 2 µg of DNA were digested with the restriction enzymes PvuII, XhoI, SacII, ApaI, EcoRI (6-bp recognition site) or TaqI (4-bp recognition site), column purified, and eluted in 30 µL of elution buffer. Then, 8.5 µL of digested DNA were self-ligated with T4 DNA ligase (New England Biolabs) overnight at 4°C, and 1 µL was used as template for iPCR using primers ai76-ai77 for DNA digested with restriction enzymes that cut outside the T-DNA region, or ai076-M13F and ai077-M13R where restriction enzymes that cut inside the T-DNA were used (Idnurm et al. 2004). iPCR conditions were: initial denaturation at 94°C for 2 min, denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec, and extension at 72°C for 2.5 min. PCR reactions were performed using ExTaq polymerase (Taqara Bio, Japan) according to manufacturer’s instructions. When ExTaq PCRs were unsuccessful, LaTaq polymerase (Taqara Bio, Japan) suitable for high G+C rich regions was used with an annealing temperature of 55°C and 60°C. Amplicons were either PCR- or gel-purified and subjected to Sanger sequencing. Sequences were subjected to BLASTn analysis against the M. furfur CBS 14141 genome assembly available on NCBI (reported as JPLK23) (Wu et al. 2015) and against an unpublished PacBio assembly. Gene boundaries and regulatory regions were determined using unpublished RNAseq data, which allowed us to define the accurate locations of T-DNA insertions. Retrieved M. furfur sequences were subjected first to BLASTx analysis against the latest genome assembly of M. sympodialis (Zhu et al. 2017) and subsequently on SGD (Saccharomyces Genome Database) to identify orthologs and infer gene function. Genes were named based on orthologous genes in Saccharomyces cerevisiae. Gene annotation was carried out manually based on BLAST searches and with the automated software Augustus (http://bioinf.uni-greifswald.de/augustus/submission.php) using RNAseq for untranslated regions (UTRs) and introns.
For Southern blot analysis, ~2 µg of genomic DNA were digested with SacII (no cut sites are within the NAT or NEO cassette, thus allowing us to determine the number of T-DNA insertions), resolved on a 0.8% agarose gel in 1x Tris-acetate EDTA (TAE) buffer, transferred to a Zeta-Probe membrane, and probed with NAT or NEO cassettes labeled with [32P]dCTP. NAT and NEO cassettes were amplified from plasmids pAIM2 and pAIM6, respectively, with universal M13F and M13R primers.
RNA extraction was performed using the standard TRIzol method (Rio et al. 2010). RNA was treated with the TURBO DNAse enzyme (Thermo Fisher Scientific) according to the manufacturer’s instructions, and quality was assessed using a NanoDrop spectrophotometer. Then, 3 µg of purified RNA were converted into cDNA via the Affinity Script QPCR cDNA synthesis kit (Agilent Technologies) according to manufacturer’s instructions. For each sample, cDNA synthesized without the RT/RNAse block enzyme mixture was used as a control for genomic DNA contamination. Approximately 500 pg of cDNA were used to measure the relative expression level of target genes through quantitative real-time PCR (RT-qPCR) using the Brilliant III ultra-fast SYBR green QPCR mix (Agilent Technologies) in an Applied Biosystems 7500 Real-Time PCR System. For each target, a “no-template control” was performed to analyze melting curves and to exclude primer artifacts. Technical triplicates and biological triplicates were performed for each sample. Gene expression levels were normalized using the endogenous reference gene TUB2 and determined using the comparative ΔΔCt method.
Generation of plasmid for CRIPSR/Cas9 targeted mutagenesis in M. furfur
Plasmids for targeted mutagenesis of M. furfur CDC55 and PDR10 through A. tumefaciens-mediated transformation were assembled in S. cerevisiae using the binary vector pGI3 as previously reported (Ianiri et al. 2016; Ianiri et al. 2017). The NAT cassette was amplified from plasmid pAIM1 using primers JOHE43277 and JOHE43278. The 5′ and 3′ flanking regions for homologous recombination were amplified from the genomic DNA of M. furfur CBS14141 using primer pairs JOHE45209-JOHE45210 and JOHE45211-JOHE45212 for CDC55 and JOHE45201-JOHE45212 and JOHE45203-JOHE45204 for PDR10, respectively. The PCR products and the double-digested (KpnI and BamHI) binary vector pGI3 were transformed into S. cerevisiae using lithium acetate and PEG 3750 as previously reported (Ianiri et al. 2016). To assess correct recombination of the newly generated plasmids, single colonies of S. cerevisiae transformants were screened by PCR using primers specific for the NAT marker (JOHE43281–JOHE43282) in combination with primers homologous to outside of the region of the plasmid pGI3 involved in the recombination event (JOHE43279-JOHE43280). Positive clones of S. cerevisiae were grown ON in YPD and subjected to phenol-chloroform-isoamyl alcohol (25:24:1) plasmid extraction using a previously reported protocol (Hoffman 2001). The plasmid DNA obtained was then introduced into the A. tumefaciens EHA105 strain by electroporation, and the transformants were selected on LB + 50 µg/mL kanamycin. PCRs were performed using ExTaq and/or LATaq polymerase as described previously, with the only difference being an extension time of 1.5 min.
To generate the components of the CRISPR/Cas9 system in Malassezia, the histone H3 was identified in the M. sympodialis ATCC42132 genome assembly (Zhu et al. 2017) through BLASTp analysis using S. cerevisiae H3 as query. The 813-bp upstream and 257-bp downstream regions, including the M. sympodialis H3 promoter and terminator (indicated as pH3, and tH3), respectively, were amplified by PCR using JOHE46457-JOHE46458, and JOHE46461-JOHE46462, respectively. High Fidelity (HF) Phusion Taq polymerase (New England Biolabs) was used according to manufacturer’s instructions, with an annealing temperature of 55°C for 30 sec, and 1 min extension at 72°C. Primer JOHE46457 includes a chimeric region for recombination in pPZP-201BK and a multicloning site, and primers JOHE46458 and JOHE46461 include chimeric regions for recombination with primers JOHE46459 and JOHE46460, which were used to amplify CAS9 open reading frame (ORF) from plasmid pXL1-Cas9 (Fan and Lin 2018). Primer JOHE46462 has SacII and SpeI restriction sites, and a region for recombination with the promoter of the 5SrRNA of M. sympodialis used to drive expression of the single guide RNA (gRNA). CAS9 amplification did not work well with HF Phusion Taq, so we used ExTaq polymerase as described above, but with fewer cycles (20 cycles) and a 4-min extension. The M. sympodialis ATCC 42132 ribosomal cluster was identified in the latest genome assembly and annotation (Zhu et al. 2017) through BLASTn analysis using ITS sequences from M. sympodialis CBS 7222 available on GenBank (accession number NR_103583). A 674-bp region from the end of the rRNA-eukaryotic large subunit ribosomal RNA (position 612351 on chromosome 5), including the rRNA-5S ribosomal RNA gene (position 613025 on chromosome 5), was amplified by PCR using primers JOHE46463-JOHE46464. Primer JOHE46463 has a chimeric region complementary to primer JOHE46462. This PCR was performed using the touchdown protocol, with an initial denaturation of 94°C per 5 min, followed by 24 cycles of denaturation at 94°C for 30 sec, annealing at a gradient temperature of 62°C for 30 sec minus 1°C per cycle, and extension at 72°C for 1 min. This was followed by 16 cycles of denaturation at 94°C for 30 sec, annealing at 50°C for 30 sec, extension at 72°C for 1 min, with a final extension of 72°C for 5 min. The gRNA scaffold was amplified from plasmid pSDMA64 (Arras et al. 2016) using primers JOHE46465-JOHE46466. JOHE46466 includes SpeI and SacII restriction sites, 7 thymine residues (6T terminator), and a chimeric region for recombination in pPZP-201BK.
The specific target sequence for CDC55 was identified using the program EuPaGDT (http://grna.ctegd.uga.edu/) available on FungiDB (https://fungidb.org/fungidb/). Specific target sequence for the gene CDC55 was added by PCR with primers JOHE46468-JOHE46466 using the gRNA scaffold as template. Primer JOHE46468 has a chimeric region for recombination with both the 5SrRNA sequence and the gRNA scaffold, with an intervening target sequence specific for CDC55. These PCRs were performed using HF Phusion Taq as reported above. All components were gel purified, and equimolar amounts of the purified amplicons were used for overlap PCR to generate the Cas9 expression cassette (pH3-CAS9, tH3) and the complete gRNA (5S rRNA promoter fused with the gene-specific gRNA scaffold). PCRs were carried out using HF Phusion taq and the touch down protocol as above, with the only difference being extension times of 5 min and 1 min, respectively. The 2 resulting amplicons were cloned within the T-DNA of pPZP201BK digested with KpnI and BamHI through HiFi (New England Biolabs) assembly according to manufacturer’s instructions and recovered in Escherichia coli DH5α. E. coli clones were screened for recombinant plasmids by PCR using primers specific for the plasmid backbone (JOHE43279 and JOHE43280) in combination with JOHE46458 and JOHE46463, respectively. The plasmid sequence for CRISPR/Cas9 deletion of CDC55 (named pGI40) was confirmed by Sanger sequencing.
To generate the CRISPR/Cas9 plasmid for targeted mutagenesis of PDR10, the binary vector pGI40 was digested with SpeI to remove the CDC55-specific gRNA, recovered from the gel and purified. A PDR10-specific target sequence designed using EuPaGDT was added to the gRNA scaffold by PCR using primers JOHE46466-JOHE46467, which have chimeric regions for recombination with both the 5SrRNA sequence and the gRNA scaffold, with the specific target sequence for PDR10 in between.
This amplicon and the 5SrRNA previously generated were recombined through HiFi assembly within the T-DNA of the SpeI-digested pGI40, and the novel recombinant plasmids with the PDR10-specific gRNA (named pGI48) were identified by Sanger sequencing. This procedure is reported in Figure 4B. Recombinant plasmids were introduced in A. tumefaciens through electroporation.
AtMT was performed with modifications that increase transformation efficiency compared to our previous protocol used to generate insertional mutants. Briefly, M. furfur was grown for 2 days at 30°C and the culture was diluted to OD600 ~1. The engineered A. tumefaciens strains with the gene deletion cassettes and the CRISPR/Cas9 expression system were grown overnight, diluted to an OD600 ~0.1, and incubated for 4 to 6 h in shaking cultures (30°C) in liquid induction medium (IM) until OD600 reached a value of 0.6 to 0.8. These bacterial cellular suspensions were mixed in 1:1, 1:2, and 2:1 ratio, respectively, and they were added to M. furfur cellular suspension at 1:2 and 1:5 ratios, respectively. These cellular suspensions were centrifuged at 5200 g for 15 min, the supernatants were discarded, and ~500 µl to 1 mL of these fungal and bacterial mixes were spotted directly onto nylon membranes placed on mIM agar containing 200 µM acetosyringone. These were coincubated for 5 days at room temperature (plates maintained without Parafilm) prior to transferring the dual cultures to mDixon supplemented with NAT (100 µg/mL) to select for fungal transformants and cefotaxime (350 µg/mL) to inhibit Agrobacterium growth.
M. furfur transformants resistant to NAT were colony purified and subjected to phenotypic and molecular characterization. Putative mutants for the CDC55 gene were exposed to UV light (250 to 300 µJ × 100) to identify those with impaired growth according to the results of the forward genetic screen. For molecular analysis, 23 representative NAT resistant transformants sensitive to UV light were subjected to phenol-chloroform-isoamyl alcohol (25:24:1) DNA extraction, and the correct replacement of the target loci was assessed by PCR. Diagnostic PCRs to identify homologous recombination events for the CDC55 gene were carried out with primers JOHE45213 or JOHE45874 in combination with specific primers for the NAT gene (JOHE43281 and JOHE43282, respectively), and with primers JOHE45215-JOHE45216 specific for the internal region of CDC55. To evaluate the overall rate of homologous recombination (HR) of the CRISPR/Cas9 system, a larger number of cdc55Δ candidate mutants were tested for sensitivity to hydroxyurea, which was found to be the stressor with the strongest phenotype. Similarly, putative pdr10Δ mutants were exposed to FLC (150 µg/mL) for phenotypic characterization, and transformants displaying impaired growth were subjected to DNA extraction for molecular characterization. Diagnostic PCRs were carried out using primers JOHE45205 and JOHE45206 alone and in combination with specific primers for the NAT gene (JOHE43281 and JOHE43282, respectively), and with primers JOHE45207-JOHE45208 specific for the internal region of PDR10. PCR analyses consisted of 34 cycles of denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec, extension at 72°C of 1 min/kb, with an initial denaturation at 94°C for 2 min and a final extension at 72°C for 5 min. PCR analyses were performed using ExTaq (Takara) according the manufacturer’s instructions. To detect homologous recombination events at the 3′ region of CDC55 and to amplify the full length PDR10 gene, LATaq polymerase (Takara) with Buffer I was used. PCR for CAS9 was carried out with ExTaq and the touchdown protocol with primers JOHE46459-JOHE46461. PCR for the gRNA was carried out with JOHE46465-JOHE46466 using ExTaq as reported above. All the primers used are listed in Table S1.
Phenotypic analysis of the target mutants was performed on mDixon agar by spotting 1.5 µL of 1:10 dilutions of each cellular suspension in the following conditions: UV (250 to 450 µJ × 100), hydroxyurea (50 mM), benomyl (50 µM), FLC (150 µg/mL), amphotericin B (AmB, 50 µg/mL), 5-flucytosine (5FC, 1 mg/mL), caspofungin (Caspo, 100 µg/mL), cyclosporin A (CsA) at 100 µg/mL both alone and in combination with 10 mM of lithium chloride, tacrolimus (FK506) at 100 µg/mL both alone and in combination with 10 mM of lithium chloride, and dimethyl sulfoxide (DMSO, 40 µL) which was used to dissolve benomyl (10 mM).
Genomic comparison and phylogeny of the ABC transporter of M. furfur and M. sympodialis
The predicted amino acid sequences of S. cerevisiae Pdr10, Pdr5, Pdr15, Pdr12, Snq2, Pdr18, Aus1, and Pdr11 were used as queries for tBLASTn and BLASTp searches against the genomes of M. furfur CBS14141 and M. sympodialis ATCC 42132 available on GenBank (Wu et al. 2015; Zhu et al. 2017). The Malassezia best hits were retrieved, and the encoded proteins for M. furfur were predicted using Augustus (http://bioinf.uni-greifswald.de/augustus/submission.php) based on RNAseq evidence for UTR regions and introns.
The web portal of ACT Artemis (https://www.webact.org/WebACT/home) was used for synteny analysis of a ~15000 bp region of M. sympodialis and M. furfur containing orthologues of the Pdr10 encoding genes. tBLASTx analysis with a E-value of 0.100000 was performed.
For phylogeny, the aforementioned predicted ABC transporter proteins were aligned using MUSCLE and the phylogenetic tree was generated with MEGA 7 (http://www.megasoftware.net/) (Kumar et al. 2016) using the maximum likelihood method (LG model, 5 discrete gamma categories) and 100 bootstrap replications.
Results
Molecular and phenotypic characterization of M. furfur insertional mutants
Insertional mutants of M. furfur were generated through AtMT using both NAT and NEO dominant drug resistance markers. A total of 767 insertional mutants were isolated and their growth was tested under several different stress conditions. A total of 19 mutants (~2.5%) with a phenotype different than the WT were selected for further characterization.
Inverse PCR (iPCR) was utilized to identify the genes inactivated by the T-DNA insertions. The sequenced amplicons were compared to the unannotated M. furfur CBS14141 genome assemblies (one reported in NCBI as M. furfur JPLK23, and another unpublished based on PacBio sequencing) coupled with RNAseq data, which facilitated the identification of the coding and regulatory sequences. This allowed an accurate determination of T-DNA insertion sites. Gene names were assigned according to the Saccharomyces Genome Database. Southern blot analysis was performed to determine the number of T-DNA insertions, revealing that 16 transformants harbored single T-DNA insertions, and 3 transformants had 2 T-DNA insertions (strains 4A10, 4B1, and 6B2) (Fig. 1). In parallel, iPCR allowed the characterization of 15 T-DNA insertions (Fig. 2). Six strains had T-DNA insertions within a predicted open reading frame (ORF), 2 strains had insertions within UTRs, and 3 strains had T-DNA insertions in intergenic regions with no RNAseq read coverage. Of the remaining 8 strains, 4 were suspected to have chromosomal rearrangements because the T-DNA borders were found in different locations in the M. furfur CBS 14141 genome, and the junctions between the T-DNA and the M. furfur genome could not be identified in another 4 strains. Table 1 summarizes the results of the forward genetics approach performed in this study, and properties of the T-DNA insertions are reported in Figure 2.
Two mutants that displayed reduced growth on YNB were identified, and analysis of the genome sequence flanking the T-DNA revealed insertions in genes involved in amino acid biosynthesis (Fig. 3A). Strain 6C8 had a non-standard T-DNA insertion that generated a deletion of ~800 bp in the genome of M. furfur. Moreover, we were not able to identify the sequence of the left border (LB) from iPCR, and the first nucleotides obtained mapped within the ORF of TYR1, which encodes prephenate dehydrogenase, an enzyme involved in tyrosine biosynthesis (Mannhaupt et al. 1989). Conversely, the right border (RB) was found within the ORF of the adjacent gene encoding an uncharacterized protein with no conserved domains that shares similarity with several other Malassezia species and basidiomycetes. Addition of tyrosine did not restore the growth of strain 6C8 to the WT level (Fig. 3A), which instead was achieved in SD media supplemented with all amino acids, suggesting the hypothesis that Tyr1 is also involved in the biosynthesis of other amino acids. In strain 2A8, the T-DNA inserted within the ORF of ARG1, which encodes the enzyme arginosuccinate synthetase that catalyzes the formation of L-argininosuccinate from citrulline and L-aspartate in the arginine biosynthesis pathway (Jauniaux et al. 1978). Addition of L-arginine to YNB was sufficient to restore a WT phenotype, confirming that M. furfur ARG1 is involved in arginine biosynthesis (Fig. 3 A).
Four insertional mutants that showed decreased growth at elevated temperature (37°C) were identified. Of these, only strain 7H6 had a standard T-DNA insertion. In strain 7H6, the T-DNA integrated between 2 genes: downstream of an RNA-binding domain-containing protein and upstream of JEN1 (Fig 3B). While the RNA-binding domain-containing protein is uncharacterized in S. cerevisiae, Jen1 is a plasma membrane monocarboxylate/proton symporter that transports pyruvate, acetate, lactate, and other substrates (Casal et al. 1999). To assess which gene was affected by the T-DNA insertion and therefore responsible for the phenotype of interest, an RT-qPCR analysis was performed. Expression levels were normalized to the TUB2 gene of M. furfur WT grown at 30°C. Expression of the uncharacterized gene encoding the RNA-binding domain-containing protein in strain 7H6 was ~60% lower compared to the WT, whereas expression of JEN1 was undetectable, indicating that either or both genes could be responsible for the temperature sensitive phenotype of strain 7H6 (Fig 3 B). The other transformants that displayed a temperature-sensitive phenotype included strain 5F1 (which showed a chromosomal rearrangement involving the 5′ regions of the gene INO80 located on chromosome 1 and of the GDP1 gene located on chromosome 3), strain 6B2 (which had 2 T-DNA insertions, one of which could be identified and was found within the ORF of a uncharacterized RhoGTPase), and strain 7D5 (whose T-DNA insertion could not be characterized by iPCR) (Table 1).
Strain 1 A7 showed increased sensitivity to UV light (250 and 350 µJ × 100) compared to WT M. furfur CBS 14141 (Fig. 3C). In strain 1A7, the T-DNA inserted into the third exon of the CDC55 gene. In S. cerevisiae, CDC55 encodes a regulatory subunit of protein phosphatase 2A. CDC55 is involved in cell cycle control, and it is required for successful chromosome segregation and nuclear division (Healy et al. 1991; Bizzari and Marston 2011).
Six strains showed increased sensitivity to the antifungal fluconazole (FLC, 150 µg/ml) compared to the WT strain (Fig. 3D). Strain 6A10 had a T-DNA insertion in the predicted stop codon of the S. cerevisiae ortholog SIP5. The function of this protein is unknown, and it has no known domains. However, it has been reported to interact with both the Reg1/Glc7 phosphatase and the Snf1 kinase in response to glucose starvation (Sanz et al. 2000). In strain 7D9, the T-DNA was found in the intergenic region between the 5′ end of an ATP-binding cassette (ABC) multidrug transporter gene and the 3′ end of the UBC6 gene. As shown in Figure 6E, the closest S. cerevisiae homolog is the ABC transporter PDR10, which is the designation that we adopted. ABC multidrug transporters are involved in pleiotropic drug responses that mediate resistance to xenobiotic compounds including mutagens, fungicides, steroids, and anticancer drugs (Sipos and Kuchler 2006). UBC6 encodes a ubiquitin-conjugating enzyme involved in ER-associated protein degradation (Walter et al. 2001). As confirmed by targeted mutagenesis (discusses below), the FLC-sensitive phenotype of strain 7D9 is due to T-DNA insertion in the promoter region of PDR10.
In strain 7F8 the T-DNA inserted within the 3’UTR of ADY2 (Fig. 3D). ADY2 encodes an ammonium and acetate transmembrane transporter involved in nitrogen utilization (Rabitsch et al. 2001; Paiva et al. 2004). In strain 2H11 T-DNA integration generated a rearrangement involving the ERG5 gene and the region close to the 5′ end of the PDA1 gene. In addition to increased sensitivity to FLC, this strain showed increased sensitivity to sodium chloride compared to WT. Erg5 is a cytochrome P450 enzyme that is a C-22 sterol desaturase involved in ergosterol biosynthesis (Lees et al. 1995). Although it is known that FLC targets membrane ergosterol, and ERG5 deletion in S. cerevisiae leads to increased FLC sensitivity (Kapitzky et al. 2010), it cannot be excluded that the FLC and NaCl sensitivity of strain 2H11 is due both to ERG5 mutation and the intrachromosomal rearrangement itself. In strains 2G9 and 5D11, the T-DNA likely integrated in tandem repeats because iPCR amplicons consisted of both the left and right borders of the T-DNA fused together, and this prevented retrieval of the junctions between the T-DNA and the genome.
Three strains showed sensitivity to cadmium sulfate (CdSO4, 30 µM), and only one (strain 2F4) showed a standard T-DNA insertion in the 5′ regions of both SEC13 and PRP43 (Fig. 3E). SEC13 in S. cerevisiae encodes an essential protein that is a structural component of the COPII (coat protein complex II), of the nuclear pore outer ring, and of the Seh-1 associated complex. It is involved in COPII-coated vesicle budding from the ER to the Golgi, nuclear pore distribution, and the ubiquitin-dependent ERAD (ER-associated ubiquitin-dependent protein breakdown) pathway, which is involved in protein degradation by cytoplasmic proteasomes (Menon et al. 2005; Dokudovskaya et al. 2011; ČopiČ et al. 2012). S. cerevisiae PRP43 encodes an RNA helicase protein that is also essential for viability and contributes to the biogenesis of ribosomal RNA, and it is also involved in spliceosomal complex disassembly (Arenas and Abelson 1997; Giaever et al. 2002). qPCR did not show clear downregulation of either gene (data not shown), and whether either or both genes are responsible for the cadmium sulfate-sensitive growth defect remains to be established. Because the T-DNA inserted in the 5′ region of SEC13 and PRP43, whose orthologs are essential in S. cerevisiae, we speculate that the functions of both genes are affected or that the phenotype observed is unlinked to the T-DNA insertion. In strain 3A1, a rearrangement involving the JLP2 and TCP1 genes was found, and for strain 4B1, Southern blot indicated 2 T-DNA insertions, one of which was identified and found within the 3’UTR of the MAE1 gene, which encodes a mitochondrial malic enzyme that is important for sugar metabolism and acts as a precursor for many amino acids (Boles et al. 1998). For strain 1F12, iPCR using different restriction enzymes was unsuccessful. We also identified a strain (4A10) that was sensitive to sodium nitrite (NaNO2) and SDS. According to Southern blot analysis, strain 4A10 has 2 T-DNA insertions, one of which could be identified and was found in an uncharacterized enoyl-CoA hydratase gene. Another strain (5F10) was sensitive to NaCl and iPCR revealed the presence of a chromosomal rearrangement involving the 3′ region of the DUG1 gene and the 5′ UTR of the RPC10 gene.
Development of a CRISPR/Cas9 gene deletion system to generate cdc55Δ M. furfur mutants
To validate the results of the insertional mutagenesis screen, the insertional mutants 1A7 and 7D9 and their mutated genes were chosen for further analysis as a proof of principle. First, we focused on the UV-sensitive strain 1A7 with a T-DNA insertion in the CDC55 gene. We were intrigued by this strain because CDC55 mutation is not known to be responsible for UV sensitivity in other fungi. The aim was to generate an M. furfur cdc55Δ targeted mutant, determine if the UV phenotype of the original insertional mutant is attributable to CDC55 mutation, and investigate any further functions of the gene in M. furfur.
For targeted mutagenesis of CDC55, molecular biology techniques were performed following our previously published methods (Ianiri et al. 2016). Regions of 1500 and 1000 bp flanking the 5′ and 3′ ends of the CDC55 target gene, respectively, were amplified from M. furfur genomic DNA and fused with the NAT marker within the T-DNA borders of plasmid pGI3. The recombinant plasmid (pGI41) bearing the cdc55Δ::NAT allele was identified in S. cerevisiae by colony PCR and A. tumefaciens EHA105 transformed by electroporation. Several rounds of Agrobacterium-transconjugation were performed, and NAT-resistant transformants of M. furfur were single colony-purified and subjected to diagnostic PCR to confirm CDC55 targeted mutagenesis. None of the transformants tested (0 out of more than 100) showed full replacement of the gene CDC55.
Next we developed a CRISPR/Cas9 system for M. furfur to increase homologous recombination efficiency. Because the plasmid for targeted gene replacement of CDC55 was already available, we generated an additional plasmid to make a DNA DBS in CDC55, and then used the available cdc55Δ::NAT allele as HDR template to repair the break. For expression of Cas9, the ORF of the CAS9 endonuclease was cloned under the control of the strong promoter and terminator of the histone H3 gene of M. sympodialis ATCC42132. To drive expression of gRNA specific for the target CDC55 gene, the promoter of the 5S rRNA was chosen. Because the ribosomal cluster is well annotated in the newly released genome of M. sympodialis (Zhu et al. 2017) and we have evidence that M. sympodialis promoters and terminators are functional in M. furfur (Ianiri et al. 2016), a 689 bp region including the 5S rRNA and its upstream region was amplified from M. sympodialis ATCC42132. The forward primer for the p5S rRNA contained restriction sites for SacII and SpeI to facilitate genetic manipulations. The scaffold gDNA also was obtained by PCR, and 6 thymine residues (6-T) were included as terminator. A 20-nt oligonucleotide target of gRNA was designed to match a region of the CDC55 gene adjacent to a PAM site, and it included the 5′ and 3′ regions that overlapped with the 5S rRNA promoter and the gRNA scaffold, respectively. This target oligonucleotide was added to the gRNA scaffold through PCR as reported in Figure 4B. The 5 PCR fragments (pH3; CAS9; tH3; p5S rRNA; gRNA) were used as template for overlap PCRs, and two final amplicons (pH3-CAS9-tH3 and p5SrRNA-gRNA) were cloned in pPZP201BK to generate plasmid pGI40 (Table 2) (Figures 4A and 4B).
AtMT of M. furfur CBS14141 was conducted to test the developed CRISPR/Cas9 system to generate targeted gene replacement of the CDC55 gene. Since our previous reports of AtMT of Malassezia (Ianiri et al. 2016; Ianiri et al. 2017), we have optimized the protocol to achieve a higher transformation efficiency. The main change included the use of a 2:1 to 5:1 Malassezia:A. tumefaciens mixture that was concentrated through centrifugation before the coincubation step on modified induction media [mIM, (Ianiri et al. 2016)]. The detailed procedure is reported in the Materials and Methods. For co-transformation of M. furfur using A. tumefaciens strains bearing the binary vectors pGI40 (CRISPR/Cas9 expression system) and pGI41 (HDR cdc55Δ::NAT template), induced bacterial strains were mixed in ratios of 1:1, 1:2, and 2:1, then added to ratios of 1:2 and 1:5 with M. furfur cells (Fig. 5A). The co-cultures were centrifuged to eliminate the supernatants, and the pellet containing the mix of the 3 components was spotted on nylon membranes placed on mIM agar. The plates were incubated at room temperature for 5 days without parafilm. The coincubation cultures were recovered and plated on mDixon containing NAT and CEF. A representative subset of 23 M. furfur transformants was single colony-purified and subjected to molecular characterization through PCR. Genotyping was performed using 1) primers designed beyond the regions of DNA used in the generation of the deletion allele in combination with specific NAT primers; 2) primers internal to the gene CDC55; 3) primers specific for the CAS9 genes, and primers specific for the gRNA (Figure 5B). Specific amplicons of ~1.6 kb and ~1.4 kb for the 5′ (left) and 3′ (right) T-DNA-genomic DNA junctions, respectively, were obtained for all of the 23-randomly selected transformants. Accordingly, the internal region of CDC55 was amplified only from the WT strain. No amplicons for CAS9 or the gRNA were obtained. These results indicate that all transformants tested had full replacement of the CDC55 gene and absence of CAS9 and gRNA integration in the genome (Fig. 5C). Furthermore, 64 additional random cdc55Δ mutant candidates were tested for sensitivity to hydroxyurea, which we found to be the most effective stressor for the cdc55Δ phenotype, and found that 62 displayed impaired growth compared to WT (Fig. S1). Therefore, molecular and phenotypic analyses revealed that out of 87 transformants analyzed, 85 were cdc55Δ mutants, resulting in a rate of homologous recombination of 97.7%.
In S. cerevisiae, CDC55 positively regulates mitotic entry at the G2/M phase transition and negatively regulates mitotic exit, and it regulates the mitotic spindle assembly and the morphogenesis checkpoint (Wang and Burke 1997; Bizzari and Marston 2011). Null cdc55Δ mutants display abnormally elongated buds; decreased growth rate; and increased sensitivity to gamma rays and hydroxyurea (DNA-damaging agents that interfere with DNA replication), to benomyl and nocodazole (which interfere with microtubule polymerization), and cold-induced stress. Phenotypic characterization of 2 representatives independent cdc55Δ M. furfur mutants confirmed that they were sensitive to UV light (Fig. 5D), corroborating the phenotype of the insertional mutant 1A7. Moreover, the M. furfur cdc55Δ mutant had a slower growth rate compared to the WT strain, and increased sensitivity to hydroxyurea and benomyl (Fig. 5D). Due to the inability of M. furfur WT to grow at low temperature, cold sensitivity could not be determined for M. furfur cdc55Δ. M. furfur cdc55Δ mutants were subjected to microscopy analysis both under normal and stress conditions, and when exposed to hydroxyurea they displayed cells with abnormal morphology and elongated buds, similar to S. cerevisiae cdc55Δ mutants (Fig. 5E).
Generation of a pdr10Δ M. furfur mutant with CRISPR/Cas9
The other insertional mutant of interest was strain 7D9, which has a T-DNA insertion between the PDR10 and UBC6 genes and exhibited FLC sensitivity. It was hypothesized that the phenotype of strain 7D9 was due to the T-DNA interfering with the function of PDR10, which is well known to mediate antifungal drug response and therefore was chosen for targeted mutagenesis. In M. furfur, PDR10 is a large, 4470-bp gene. The pdr10Δ::NAT gene disruption cassette was generated as previously described for CDC55, and the vector was named pGI42. For the gRNA, a primer with a specific PDR10 target between regions that overlap with the 5S rRNA promoter and the gRNA scaffold was added to the gRNA scaffold by PCR. The resulting amplicon was then cloned together with the p5S rRNA in the T-DNA of pGI40 digested with SpeI as reported in Figure 4B; this vector was named pGI48.
Co-transformation of M. furfur CBS14141 was performed using A. tumefaciens strains bearing plasmids pGI42 and pGI48 as reported in Figure 5A. 60 M. furfur NAT-R transformants were single colony purified, and streaked onto mDixon + FLC. Five (8.3%) transformants that displayed FLC sensitivity plus a randomly-selected FLC-resistant control strain were subjected to molecular characterization (Fig. 6B). PCR analysis using external screening primers designed beyond the region of DNA utilized to generate the pdr10Δ::NAT deletion allele produced 2 amplicons: a 6183-bp amplicon for the WT and the FLC resistant strain, and a 3933-bp amplicon for the 5 transformants that displayed FLC sensitivity. For these 5 transformants, PCR carried out using the external primers with specific NAT primers generated amplicons of ~1.1 kb and ~1.3 kb on the 5′ and 3′ regions, respectively. PCR using primers internal to the PDR10 gene generated an amplicon of 486 bp only in the WT and the randomly selected NAT-R strain. These PCR results indicate full replacement of the PDR10 gene in the 5 transformants that displayed FLC sensitivity.
Mutants pdr10Δ showed hypersensitivity to FLC, indicating that the phenotype of strain 7D9 was due to the T-DNA interfering with the function of PDR10. Moreover, because ABC transporters are known to be involved in pleiotropic drug resistance and cellular detoxification, the phenotypic response of pdr10Δ mutants was tested against other antifungal drugs of clinical relevance. Surprisingly, M. furfur pdr10Δ mutants showed only sensitivity to FLC and grew at the WT level on amphotericin B, 5-flucytosine, caspofungin, tacrolimus (FK506), and cyclosporine A (CsA) both alone and in combination with the plasma membrane stressor lithium chloride (Fig. 6C and data not shown), which we previously showed enhances antifungal activity of tacrolimus against M. furfur (Ianiri et al. 2017). Moreover, M. furfur pdr10Δ mutants did not display sensitivity to the DNA-damaging agents UV or hydroxyurea and only displayed sensitivity to the fungicide benomyl (Fig. 6C).
During BLAST searches, we noted that M. furfur has 2 adjacent ABC transporter-encoding genes that are orthologs of 3 adjacent ABC transporter-encoding genes in M. sympodialis (Fig. 6D), a Malassezia species that we use as a model for genomics comparison within the genus because of the high quality of its genome assembly (Zhu et al. 2017). Interestingly, BLASTp of these ABC transporters against S. cerevisiae revealed high similarity (ie E-value 0.0) with several ABC transporters, such as Pdr18, Pdr12, Pdr5, Pdr10, Pdr15, Aus1, and Pdr11. Reciprocal BLAST (BLASTp and tBLASTn) of these proteins against M. furfur and M. sympodialis finds only the aforementioned adjacent ABC transporters, which we named Mf (M. furfur) and Ms (M. sympodialis) PDR10_1, PDR10_2, and PDR10_3. The mutated gene in M. furfur corresponds to PDR10_1. Phylogenetic analysis revealed that ABC transporters of M. furfur and M. sympodialis cluster together in a maximum likelihood tree and are related to the S. cerevisiae Pdr10 ABC transporter, which is the gene designation that we selected. This analysis suggests a common duplication event of the Malassezia PDR10 (green dot on Fig. 6E), followed by another more recent duplication in M. sympodialis (blue dot on Fig. 6E).
Discussion
A. tumefaciens-mediated transformation is considered a “silver bullet” in functional genomics of fungi, and its main applications as well as the major discoveries that it has allowed have been recently reviewed (Idnurm et al. 2017a). Because A. tumefaciens can grow under a variety of conditions, the transformation method is versatile and has been successfully applied in a number of fungi, including those with particular nutrient requirements and that are recalcitrant to other transformation approaches, such as Malassezia (Ianiri et al. 2016; Celis et al. 2017).
In this report, we present the first application of forward genetics in M. furfur, a representative species of the fungemia-causing Malassezia group. The goal was to generate random insertional mutants, expose them to stress conditions to isolate those displaying sensitivity compared to the WT, and identify the corresponding T-DNA insertion sites to determine the function of the genes causing the phenotypes. Given the lack of knowledge on gene function in Malassezia, insertional mutants were assayed on a variety of conditions that are known to interfere with i) important cellular processes, such as those involved in plasma membrane and cell wall maintenance, growth under nutrient limiting conditions, and protein folding; ii) response to environmental stresses, such as osmotic and nitrosative stresses, UV light, elevated temperature and pH, and heavy metals; and iii) response to the antifungal FLC, which is of clinical relevance.
This loss-of-function screen allowed the characterization of 8 M. furfur insertional mutants (1A7, 2A8, 2F4, 6A10, 6C8, 7D9, 7F8, 7H6) that had 1 T-DNA insertion as determined by Southern blot analysis (Fig. 1) and that displayed sensitivity to one or more stress conditions (Fig. 3; Table 1). In 4 strains, the T-DNA inserted within the ORF of genes, and in another it was found to lie within a 3′ UTR (Table 1, Fig. 2 - 3), thus allowing us to define with high probability a direct link between genotype and phenotype. Clear examples of this were M. furfur transformants 2A8 and 1A7. Strain 2A8 was selected because of its reduced growth on minimal medium (YNB), and it was found to have a T-DNA insertion in the ARG1 gene. Strain 1A7 was selected for its increased sensitivity to UV light, and found to have a T-DNA insertion in the CDC55 gene. Two different approaches were employed to validate the findings of the insertional mutagenesis screen. For strain 2A8 the addition of arginine was sufficient to rescue growth to a WT level (Fig. 3A), while for strain 1A7, targeted M. furfur cdc55Δ mutants (Fig. 5) confirmed UV sensitivity.
In 3 other mutants of interest, the T-DNA inserted between 2 adjacent genes, and further experiments were conducted to identify the gene(s) responsible for the observed phenotype. A successful approach for strain 7H6 was gene expression analysis through RT-qPCR, which revealed downregulation of both genes flanking the T-DNA insertion (Fig. 3B). Strain 7D9 was sensitive to FLC and had a T-DNA insertion between the 3′ region of UBC6 and 5′ region of PDR10, and targeted mutagenesis confirmed that the observed phenotype was due to T-DNA insertion in the promoter region of PDR10 (Fig. 6). Lastly, the RT-qPCR approach did not allow us to define which genes was responsible for the cadmium sulfate sensitive phenotype of strain 2F4 (data not shown).
Despite the benefits of an AtMT random mutagenesis approach, analysis of the T-DNA insertion events also revealed limitations. Eleven of 19 M. furfur insertional mutants selected (~58%) were not useful for gene function analysis. Three transformants had 2 T-DNA insertions as determined by Southern blot analysis (Fig. 1), and although we determined at least one insertion site (Table 1), it was not possible to determine which gene was responsible for the mutant phenotype. Moreover, the insertion sites could not be identified through iPCR for 2 transformants (1F12 and 7D5). Strains 2G9 and 5D11 contained tandem T-DNA insertions, and in 4 strains (2H11, 3A1, 5F1, 5F10), chromosomal rearrangements following the integration of the T-DNA in the M. furfur genome were observed. Although AtMT represents a powerful method for random mutagenesis, we and other authors have commonly found non-standard T-DNA insertion events in the genome of both ascomycetous and basidiomycetous fungi [for more details see the following reviews and references within them (Michielse et al. 2005; Bourras et al. 2015; Idnurm et al. 2017a; Hooykaas et al. 2018)]. For example, in a recent study on systematic T-DNA insertion events in the red yeast Rhodosporidium toruloides, Coradetti and colleagues found that only 13% of mutants had regular T-DNA insertions and a total of 21% of insertions were useful to identify the genes mutated by the T-DNA (Coradetti et al. 2018). Moreover, in classical forward genetic screens in which loss-of-function events are selected, it is common to isolate strains with chromosomal rearrangements that originated following the insertion of the T-DNA, because these strains are generally less fit and display increased sensitivity to stress (Ianiri and Idnurm 2015). These undesirable events have been also described following AtMT of plants (Clark and Krysan 2010).
Typically in forward genetics the linkage between the T-DNA insertion and the phenotype is confirmed through: 1) sexual crosses and analysis of the phenotype in the recombinant progeny, 2) functional complementation, or 3) generation an independent targeted mutation for the gene identified (Idnurm et al. 2017a). Because of the lack of a known sexual cycle in Malassezia, and the difficulty of genetic manipulations for complementation studies, in the present study we aimed to generate M. furfur mutants for the genes CDC55 and PDR10 to validate their involvement in UV and FLC resistance, respectively. Following our previously reported protocol for targeted mutagenesis in M. furfur (Ianiri et al. 2016; Ianiri et al. 2017), several transformation rounds were performed, but we did not obtain any CDC55 or PDR10 mutants. Therefore, a system based on CRISPR/Cas9 was developed to increase homologous recombination and facilitate the generation of targeted mutants in M. furfur.
Because AtMT is the only effective transformation technique for Malassezia, a functional CAS9 cassette and a gRNA needed to be cloned within the T-DNA of a binary vector, together with a marker for selection. For homologous recombination-mediated targeted mutagenesis, a specific gene replacement construct to serve as template to repair the BSB was also necessary. Cloning of all the required components within the T-DNA of one binary vector is technically challenging and time consuming. In one study, Kujoth and colleagues generated a large T-DNA that included one or more gRNA, a Cas9 expression cassette, and a gene marker, and successfully applied this system in gene editing strategies through NHEJ in Blastomyces dermatitidis (Kujoth et al. 2018). For CRISPR/Cas9 in Leptosphaeria maculans, Idnurm and colleagues reported a system based on 2 binary vectors, one with CAS9 and a marker, and the other with gRNA and another marker, that could be successfully delivered at the same time through co-transformation employing A. tumefaciens to perform efficient gene editing through NHEJ (Idnurm et al. 2017b). For CRISPR/Cas9 of Malassezia, we opted for a system that would be suitable for targeted gene replacement through homologous recombination based on co-transformation of M. furfur with 2 A. tumefaciens strains, one bearing the binary vector with the HDR gene deletion allele, and another with a binary vector engineered for the CRISPR/Cas9 system without a gene marker. The rationale for generating a marker-free binary vector was to: 1) have a CRISPR/Cas9 transient expression system with a reduced rate of CAS9 and/or gRNA ectopic integration in the host genome, similar to a system developed for C. albicans and C. neoformans (Min et al. 2016; Fan and Lin 2018); 2) allow further genetic manipulation of the NAT-generated mutant using the other Malassezia-specific gene marker available, which encodes for resistance to neomycin; and 3) reduce recombination within the actin promoter and terminator regions of the NAT and NEO gene markers.
Considering gRNA expression, the choice of an appropriate promoter has represented a major challenge for the application of CRISPR/Cas9 technology in fungi. Currently, common approaches include the use of a strong promoter recognized by RNA polymerase II, such as that of the ACT1 or GDP1 genes, coupled with a hammerhead ribozyme and/or hepatitis delta virus ribozyme for gRNA excision (Idnurm et al. 2017b; Kujoth et al. 2018); the use of RNA polymerase III promoters, such as the U6 promoters of small nuclear RNA used for C. neoformans (Wang et al. 2016; Fan and Lin 2018); or the promoters of the tRNA or rRNA with or without ribozymes (Shi et al. 2017). In this study, we first tested a strategy based on the use of the 5S rRNA promoter of M. sympodialis (Fig. 4B). While we were working on developing this system, Zheng and colleagues reported a similar approach in A. niger, demonstrating high efficiency of gene editing using both the 5S rRNA promoter alone or combined with the HDV ribozyme (Zheng et al. 2018). Cas9 expression is usually achieved using a strong promoter and terminator; in this study the regulatory regions of the histone H3 gene of M. sympodialis served this purpose (Fig. 4A). During the first CRISPR/Cas9 attempt, we were able to generate M. furfur cdc55Δ mutants. Surprisingly, both molecular and phenotypic analysis revealed a homologous recombination rate of 98% (Fig 4C, and Fig. S1). This high rate of homologous recombination is similar to that in the study of Zheng and colleagues (Zheng et al. 2018) and other CRISPR/Cas9-mediated gene deletion approaches (Fan and Lin 2018). Given these positive results, the use of ribozymes flanking the 5S rRNA promoter was not tested. Phenotypic analysis confirmed the involvement of CDC55 in UV resistance, and further assays revealed sensitivity of the M. furfur cdc55Δ mutant to benomyl and hydroxyurea (Fig. 5D), which also induced an abnormal bud morphology (Fig. 5E). This indicates a conserved function of the cell division cycle protein Cdc55 in M. furfur and S. cerevisiae.
This CRISPR/Cas9 technology was then tested for targeted mutagenesis of another gene of interest, PDR10. Corroborating results obtained for CDC55, we were able to promptly obtain pdr10Δ mutants, although the rate of homologous recombination was lower for this gene. This could be due to several factors, such as shorter flanking regions of ~800 bp used in the HDR pdr10Δ::NAT template, the length of the PDR10 gene (more than 4 Kb), the genomic location, or to lower activity of the PDR10-specific gRNA. Analysis of M. furfur pdr10Δ mutants revealed an unexpected specificity of M. furfur PDR10 for resistance to the clinical-relevant drug FLC, and to the antifungal agent benomyl (Fig. 6C). While there are multiple studies on the pleiotropic drug resistance function of ABC transporters in non-pathogenic (S. cerevisiae) and pathogenic (C. albicans) yeasts (Sipos and Kuchler 2006; Coste et al. 2008; Paul and Moye-Rowley 2014), in these cases specific analysis of Pdr10 in response to several drugs has yet to be performed, and therefore it is not possible to provide a detailed comparison analysis that supports conserved or divergent functions of PDR10 in M. furfur. A recent study reported the involvement of S. cerevisiae Pdr10 in double-strand break repair via sister chromatid exchange (MuÑoz-GalvÁN et al. 2013), but we could not confirm this function in M. furfur because of the lack of sensitivity of pdr10Δ mutants to DNA-damaging agents (Fig. 5C). Further bioinformatics analyses revealed that Pdr10 is the only ABC transporter present in the genome of two Malassezia species, and that it underwent ancestral and more recent gene duplication events (Fig. 6D-E). This suggests profound differences with other fungi, and further studies are needed to elucidate the evolution and specific roles of these Malassezia ABC transporters in resistance to chemicals and their network of interactions.
Our understanding of Malassezia genetics is still limited, and the T-DNA-mediated random insertional mutagenesis applied in this study coupled with a novel and efficient CRISPR/Cas9 system represent straightforward approaches to advance molecular genetics in this understudied organism. Indeed, while T-DNA-mediated random insertional mutagenesis is of particular relevance to discover novel gene functions, such as the UV sensitive phenotype of CDC55, or the FLC sensitivity due to mutations in the genes SIP5 and ADY2, the efficiency of CRISPR/Cas9 is a critical requirement to perform large-scale analyses while also validating the results of the genetics screen.
Historically Malassezia research has been hampered by the fastidious nature and particular growth requirements of species within this genus and by their difficult identification and classification. Nevertheless, in addition to the available genome sequence and annotation of most Malassezia species, the recent introduction of animal models to study Malassezia interactions with the skin and the gastrointestinal tract (Limon et al. 2019; Sparber et al. 2019), and the development of this novel CRISPR/Cas9 system and other existing molecular technologies (Ianiri et al. 2016; Celis et al. 2017; Ianiri et al. 2017) represent key scientific advances to study the biology and pathophysiology of Malassezia, the main fungal inhabitants of mammalian skin.
FUNDING INFORMATION
This work was supported by NIH/NIAID R01 grant AI50113-15 and by NIH/NIAID R37 MERIT award AI39115-21 (to J.H.).
ACKNOWLEDGMENTS
We thank Sheng Sun for assistance with Southern blot analysis, and Tom Dawson for sharing unpublished genome and RNAseq information prior to publication. We thank Ci Fu, Shelby Priest and Cecelia Wall for critical comments on the manuscript.