Knufia petricola A95 (CBS 123872) was used as recipient (WT) for genetic manipulation. K. petricola genes analyzed in this study have been described previously (Voigt et al., 2020) or were identified in the genome sequence of A95 (~28 Mb, ~340×, 12 contigs) (Heeger et al., unpublished) using the tblastN tool of Geneious Prime 2021 (Biomatters Ltd., Auckland, NZL). For the sequences see GenBank accessions OM802156-OM802160 or Supplementary Sequences 1–5. A95 and derivatives (Supplementary Table 1) were cultivated in malt extract broth/agar (MEB/MEA) at 25 °C in darkness (Nai et al., 2013). Basal synthetic medium was SDNG (synthetic-defined–nitrate–glucose) [0.17% Difco™ Yeast Nitrogen Base without Amino Acids and Ammonium Sulfate (BD Biosciences, Franklin Lakes, NJ, USA), 0.3% NaNO₃, 2% glucose, 2% agar]. SDYG contained 0.1% yeast extract instead of NaNO₃. For inoculation of growth assays, cells were taken from surface-grown colonies, resuspended in phosphate-buffered saline (PBS), and dispersed using glass beads (3–5 mm) and a Ribolyser (Hybaid) (20 s at 40 Hz). Cell titers of 1:10 dilutions were determined using a Thoma cell counting chamber and adjusted with PBS to 1 × 10⁶ cells ml⁻¹ for droplet assays and 2.5 × 10⁶ cells ml⁻¹ for growth assays. Dilution series down to 1 × 10³ cells ml⁻¹ were prepared and used for inoculation of solidified media with 10-μl droplets. For growth assays, 200 μl of the cell suspensions (2.5 × 10⁶ cells ml⁻¹) were evenly distributed with 10 glass beads (3–5 mm) on agar. Stock solutions of selective agents were hygromycin B (HYG; AppliChem, Darmstadt, GER) (41 mg ml⁻¹ H₂O), nourseothricin (NTC; Werner BioAgents GmbH, Jena, GER) (100 mg ml⁻¹ H₂O), geneticin (G418; Sigma-Aldrich, St. Louis, MO, USA) (50 mg ml⁻¹ H₂O), glufosinate ammonium (GFS; ChemPur, Karlsruhe, GER) (100 mg ml⁻¹ H₂O), and chlorimuron ethyl (CME; Alfa Aesar, Heysham, UK) (100 mg ml⁻¹ DMF).

Genomic DNA from K. petricola was prepared as described previously (Voigt et al., 2020). DNA was mixed with Midori Green Direct (Biozym Scientific GmbH, Oldendorf, GER) and separated in 1% agarose gels using the MassRuler™ DNA Ladder Mix (Thermo Scientific, Waltham, MS, USA) as size standard. Total RNA from K. petricola was extracted using the TRI Reagent RNA Isolation Reagent (Sigma-Aldrich, St. Louis, MO, USA) and purified using the Monarch RNA Cleanup Kit (New England Biolabs, NEB, Ipswich, MA, USA). 1 μg of total RNA was submitted to reverse-transcription (RT) using the iScript gDNA Clear cDNA Synthesis Kit (Bio-Rad Laboratories Inc., Hercules, CA, USA). Standard PCR reactions were performed using desalted primers from Eurofins Genomics Germany GmbH (Ebersberg, GER) listed in Supplementary Table 2, the Q5® High-Fidelity DNA Polymerase (NEB) for cloning and sequencing purposes and the Taq DNA Polymerase (NEB) for diagnostic applications. PCR products were purified with the Monarch™ PCR & DNA Cleanup Kit (NEB) and sequenced. Plasmids listed in Supplementary Table 3 were assembled by homologous recombination in Saccharomyces cerevisiae FY843 (Oldenburg et al., 1997; Schumacher, 2012) or using the NEBuilder® HiFi DNA Assembly Cloning Kit (NEB). Plasmid DNA from Escherichia coli and S. cerevisiae was extracted with the Monarch® Plasmid Miniprep Kit (NEB). For extraction of larger amounts of plasmid DNA from E. coli the NucleoBond® Xtra Midi Kit (Macherey-Nagel, Düren, GER) was used. Sequencing of PCR fragments and plasmids was accomplished with the Mix2Seq Kit at Eurofins Genomics. Cloning procedures were supported by using SnapGene® 4.0.8 (GSL Biotech LLC, Chicago, IL, USA).

Short-homology (SH) fragments for replacement of gene of interests (goi) by resistance cassettes were amplified by PCR using pNDH-OGG (hygR), pNDN-OGG (natR) (Schumacher, 2012), pNDG-OGG (genR), pNDP-OGG (baR), or pNDS-OGG (suR) (this study, Figure 2) as template and primers goi-SH5F (binding in TniaD) and goi-SH3R (binding in PoliC) containing ~75-bp-long 5′ overhangs homologous to target genomic regions. For replacement of genes by a resistance cassette fused to an expression cassette, the constructs were amplified by PCR with primers Txxx-goi-SH5F (binding in TniaD/TniiA of the resistance cassette) and Txxx-goi-SH3R (binding in TtrpC/Tgluc/Ttub1 of the expression cassette) containing ~75-bp-long homologous sequences as 5′ overhangs. Per transformation approach 10–15 μl of the PCR were used. Constructs with long-homology (LH) fragments were isolated from cloned plasmids via digestion (2 μg DNA per transformation approach). Protospacers (PSs) for CRISPR/Cas9 were identified with the CRISPR site finder of Geneious Prime® 2021 (Biomatters Ltd.). AMA1-bearing vectors for mediating the extrachromosomal expression of Cas9 and target-specific sgRNAs (liberation from transcripts by attached ribozymes or endogenous RNases) were cloned as previously described (Nødvig et al., 2015, 2018; Wenderoth et al., 2017) (Supplementary Figure 2) and transformed in the circular form (2 μg of each plasmid per transformation approach). Transformations carried out with different combinations of CRISPR/Cas9 plasmids and donor DNA are summarized in Supplementary Table 4. PEG-mediated transformation of protoplasts was performed as described previously (Noack-Schönmann et al., 2014a; Voigt et al., 2020), but an osmotically stabilized synthetic-defined medium lacking amino acids [KTM, Knufia transformation medium: SDNG supplemented with 0.49 M (16.8%) sucrose] was used for regeneration of protoplasts. Distributed protoplasts (5 × 10⁵ protoplasts on 20 ml of KTM per Petri dish) were overlaid after 24 h with 5 ml of warm KTM supplemented with 0.4% (w/v) agar and 250 μg ml⁻¹ HYG (final conc. 50 μg ml⁻¹), 25 μg ml⁻¹ NTC (final conc. 5 μg ml⁻¹), 500 μg ml⁻¹ G418 (final conc. 100 μg ml⁻¹), 200 μg ml⁻¹ GFS (final conc. 40 μg ml⁻¹) or 375 μg ml⁻¹ CME (final conc. 75 μg ml⁻¹). Putative transformants were transferred after 2–3 weeks to MEA containing 25 μg ml⁻¹ of HYG and/or 5 μg ml⁻¹ of NTC, or to SDNG containing 100 μg ml⁻¹ of G418, 40 μg ml⁻¹ of GFS or 75 μg ml⁻¹ of CME. Targeted integration of replacement and expression constructs was detected by diagnostic PCRs with primers binding up- and downstream of the integration sites and within the transformed constructs. Gene editing events were detected by amplicon sequencing.

For measuring GFP fluorescence with a spectral photometer, the Poi::gfp strains were cultivated for 7 days in 20 ml of MEB at 25°C and 100 rpm. Cultures were dispersed in grinding jars with eight steel beads (5 mm) by using a TissueLyser™ (Retsch GmbH, Haan, GER; 10 min at 30 Hz). Cells were washed twice with PBS and finally suspended in 5 ml of PBS. Cell titers were determined using a Thoma cell counting chamber. Ninety-six-well plates were prepared with 1 × 10⁹ cells in 200 μl PBS per well. GFP fluorescence (excitation 485/20, emission 528/20) was quantified with a Synergy HT microplate spectrophotometer and the Gen5™ software (BioTek; Agilent Technologies, Inc., Santa Clara, CA, USA). Fluorescence and light microscopy of cells suspended in PBS were performed with a Zeiss AxioImager M2m microscope. Cells were resuspended in PBS-based solution in a final concentration of 2 μg ml⁻¹ of 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) (1 mg ml⁻¹ H₂O) for staining the nuclei prior to microscopy. GFP fluorescence was examined with filter set 38 HE [excitation BP 470/40 (HE), beam splitter FT 495 (HE), emission BP 525/50 (HE)], mCherry fluorescence using filter set 14 (excitation BP 510-560, beam splitter FT 580, emission LP 590) and DAPI fluorescence using filter set 49 (excitation G 365, beam splitter FT 395, emission BP 445/50). Images were captured with a Zeiss AxioCam 503 mono camera and analyzed using the ZEN 2 (blue edition) v3.2.0.0000 software package (Carl Zeiss AG, Jena, GER).

In genome editing, the term “multiplexing” refers to the ability to edit more than one genomic site in an organism (Schuster and Kahmann, 2019). Multiplex CRISPR/Cas9 may include using a single sgRNA targeting the conserved region of related genes or using several sgRNAs targeting unrelated genes. In K. petricola, CRISPR/Cas9-mediated simultaneous replacement of pks1 and phs1 resulting in white mutants was achieved using two Cas9/sgRNA-delivering plasmids (pAMA/ribo−pks1^PS2, pAMA/ribo−phs1^PS1) or two RNPs (RNP-pks1^PS2, RNP-phs1^PS1) which were co-transformed with resistance cassette-containing replacement fragments (Voigt et al., 2020). For increasing the efficiency of simultaneous mutations with the perspective for generation of triple and quadruple mutants, an alternative for the expression of sgRNAs was quested. Considering that AMA1-bearing plasmids work in K. petricola when a sgRNA is released by ribozymes from a mRNA transcript (pAMA/ribozyme), the functionality of plasmid-based expression of single/multiple sgRNAs from tRNA splicing cassettes according to Nødvig et al. was assessed. In this system (pAMA/tRNA), a tRNA-sgRNA-tRNA cassette is flanked by the U3 regulatory sequences of Aspergillus fumigatus for mediating the transcription by the RNA polymerase III and subsequent release of sgRNA(s) from the RNA transcript by the endogenous tRNA maturation machinery (Nødvig et al., 2018) (Supplementary Figure 1).

To evaluate the efficiency in K. petricola, pks1, and phs1 were targeted by using the same protospacers (pks1^PS2, phs1^PS1) and repair templates as in the previous study. For testing the expression of one and two sgRNAs from a tRNA splicing cassette, the plasmids pAMA/tRNA-pks1^PS2, pAMA/tRNA-phs1^PS1 and pAMA/tRNA-pks1^PS2-phs1^PS1 were cloned (Supplementary Table 3). For gene replacement, wild-type or pks1- protoplasts were co-transformed with the different Cas9/sgRNA-delivering plasmids (pAMA/tRNA vs. pAMA/ribozyme) and appropriate replacement fragments (Δpks1 [N], Δphs1 [H]) conferring resistance to nourseothricin (NTC) and hygromycin (HYG), respectively (Supplementary Table 4; Figure 2A). Single sgRNA expressed from tRNA and ribozyme cassettes resulted in similar numbers of differentially pigmented resistant colonies, i.e., pink (transformation of WT with sgRNA^pks1 + Δpks1 [N], upper row) and white (transformation of pks1- with sgRNA^phs1 + Δphs1 [H], middle row). However, the use of a single plasmid for the expression of two sgRNAs (pAMA/tRNA-pks1^PS2-phs1^PS1) in combination with both replacement fragments resulted in a greater number of white natR hygR colonies compared to the control approach with the same sgRNAs expressed from two different pAMA/ribozyme plasmids (Supplementary Figure 3).

In a second experiment, WT and pks1- protoplasts were transformed with the three pAMA/tRNA plasmids alone for random gene editing or together with the corresponding 80-bp-long single stranded DNA oligonucleotides pks1-oligo2.7 or phs1-oligo1 as repair templates for targeted gene editing (Supplementary Table 4; Figure 2B). Protoplasts were overlaid with medium containing HYG for transient selection of pAMA-containing cells potentially becoming edited strains. From the transformation plates shown, it is evident that the addition of oligonucleotides overlapping the Cas9 cutting sites considerably increased the number of differentially pigmented colonies which is in accordance with previous experiments with ribozyme-released sgRNAs. Furthermore, the rate of double gene editing (white mutants) by co-transformation of pAMA/tRNA-pks1^PS2-phs1^PS1 and the two oligonucleotides was increased [53 vs. ~40% for expression of the two sgRNA from two plasmids (Voigt et al., 2020)]. Double gene editing events in twelve white mutants (W1-12) were analyzed by amplicon sequencing of the DSB/PS-spanning regions of pks1 and phs1 (Supplementary Figures 1, 4). Accordingly, two different mutations causing in-frame deletions of 3 bp in pks1 were identified resulting in PKS1^ΔL118 (in seven mutants; mutation was present in used pks1-oligo2.7) or PKS1^ΔL119 (five mutants). All twelve mutants contained the same 2-bp-mutation (present in used phs1-oligo1) in phs1 resulting in a premature stop codon and a truncated protein (PHS1^1−63aa).

Interestingly, the transformation of melanin-deficient pks1- protoplasts with pAMA/tRNA-phs1^PS1 resulted in several colonies with an orange pigmentation instead of the expected white phenotype (Figures 2A,B). Three of them (O1-3) were studied by amplicon sequencing revealing that they achieved different deletions in the 5′ region of phs1: O1 contains an in-frame deletion of 9 bp resulting in a 3-aa-long deletion plus one aa exchange (PHS1^{Δ63−65, V66L}), O2 an in-frame deletion of 12 bp resulting in a 4-aa-long deletion (PHS1^Δ63−66), and O3 a deletion of 23 bp resulting in a 8-aa-long deletion plus one aa exchange (PHS1^{Δ60−67, G68S}) (Figure 2C; Supplementary Figures 1, 4). Considering the color of the mutants indicating that carotenoids are still produced, and that the identified deletions retain the reading frame, it appears likely that the three PHS1 variants maintained the function to catalyze the formation of phytoene (first step in carotenoid synthesis) via the C-terminal domain but are affected in the cyclization of carotenoids via the N-terminal lycopene cyclase domain.

Taken together, expression of sgRNA(s) from pAMA/tRNA plasmids works well for gene editing and gene replacement approaches, indicating that the U3 regulatory sequences from A. fumigatus mediate moderate transcription of the tRNA-sgRNA-tRNA cassettes by the RNA polymerase III in K. petricola. Double mutants are obtained more easily, and the generation of triple and quadruple mutants becomes a realistic objective. Because of the eligibility and the flexibility of this system—one primer pair per sgRNA allows for assembling of plasmids with various combination of sgRNAs—it will replace the pAMA/ribozyme strategy in the long term.

The CRISPR/Cas9 technique can be applied for mutating multiple genes without introducing the equal number of selection markers; however, replacing genes by resistance cassettes facilitates the screening for all desired mutations by selection and diagnostic PCR. Moreover, replacement approaches eliminate the entire gene and yield independent mutants with identical genotypes. Formerly, two selection systems, i.e., hygR and natR, were available in K. petricola. First triple mutants were generated by replacing genes in the selection marker-free pigment mutant pks1-/phs1- that derived from a gene editing approach (not shown). Nevertheless, additional selection systems are beneficial for various approaches. Further frequently used selection marker systems in fungi are bleR [phleomycin (PHLEO)/ble], genR [geneticin (G418)/nptII], baR [glufosinate (GFS)/bar], suR [chlorimuron ethyl (CME)/sur] and fenR [fenhexamid (FEN)/erg27^*] (Sweigard et al., 1997; Cohrs et al., 2017; Lichius et al., 2020) (Table 1). The usage of bleR was excluded as phleomycin introduces DSBs in DNA and thus may lead to additional undesired mutations. A first rapid assay showed that growth of K. petricola is prevented in amino acid-lacking SDNG medium supplemented with geneticin (G418), glufosinate ammonium (GFS) and chlorimuron ethyl (CME), but is not affected by fenhexamid (data not shown). As a result, genR, baR, and suR were evaluated as putative new selection systems.

In a previous study, primers (goi-SH5F, goi-SH3R) were designed allowing for the amplification of the hygR [H] and natR [N] cassettes from pNDH-OGG and pNDN-OGG, respectively (Voigt et al., 2020). These plasmids derive from the pNXR-XXX series which have all the same modular structure (Schumacher, 2012) (Supplementary Figures 5, 6). The 3′ ends of the primers are standardized and bind to the identical regulatory sequences of the resistance cassettes, while the 5′ overhangs are flexible and represent the ~75-bp-long 5′- and 3′-noncoding sequences of the goi (SH sequences). Other compatible templates are pNDB-OGG (bleR) and pNDF-OCT (fenR) (Schumacher, 2012; Cohrs et al., 2017) (Supplementary Figure 5). Accordingly, plasmids with compatible genR, baR and suR cassettes were cloned. The resistance genes nptII, bar, and sur were amplified from pKS-Gen, pCB1524 or pCB1532 (Sweigard et al., 1997; Bluhm et al., 2008; McCluskey et al., 2010) with primers generating overlaps with PtrpC or TniaD and obtained amplicons were assembled with the pND–OGG backbone. Resulting pNDG-OGG (genR), pNDP-OGG (baR) and pNDS-OGG (suR) (Supplementary Table 3) were used as templates for the PCR-based generation of Δpks1 fragments with genR [G], baR [P], or suR [S] cassettes (Table 2). Because GFS and CME depend on an amino acid-free medium for preventing growth, the complex osmotically stabilized transformation medium (MEAS) was replaced by synthetic-defined osmotically stabilized SDNG (KTM). Protoplasts of WT:A95 were co-transformed with pAMA/ribo-pks1^PS2 and Δpks1 [G], Δpks1 [P], or Δpks1 [S] fragments (Supplementary Table 4), and overlaid with KTM-TOP containing G418, GFS and CME in different concentrations. Pink colonies appeared after 1–2 weeks on the top and could be clearly separated from the black cells growing in the lower layer (background) (Figure 3). By adjusting the concentrations in following experiments, background growth was reduced by using G418 and GFS in higher concentrations. However, increased concentrations of CME did not have the same effect likely because of the relatively short half-life of the compound (Choudhury and Dureja, 1996). Thus, putative transformants must be isolated from the transformation plates as soon as possible. Notably, pink genR colonies were found only when KTM (and not MEAS) was used. As usage of KTM also results in faster protoplast recovery which shortens the period until top-grown transformants can be transferred, KTM is used in the meantime for all transformations independent of the selective agent. Twelve pink colonies per selection marker were transferred to SDNG supplemented with the appropriate selective agent. All mutants displayed moderate growth, thus six mutants per selection marker system were examined by diagnostic PCR (Supplementary Figure 7). The detection of the expected amplicons for HR events at 5′ and 3′ of pks1 for most of the mutants confirmed the stable integration of the different resistance cassettes into the genome of K. petricola for conferring resistance.

In a second experiment, expression constructs comprising the three different resistance cassettes linked to a gfp cassette should be integrated in a melanin-containing wild-type-like background. Therefore, the cloned pNDX-OGG plasmids were used as templates for amplifying the expression constructs with primers binding in TniaD (resistance cassette) and Tgluc (expression cassette) and providing 5′ overhangs homologous to intergenic region 2 (igr2, see Section Identification of Two Neutral Genome Loci for Targeted Insertion of Expression Constructs). For generating resistant black mutants, WT:A95 protoplasts were co-transformed with the SH fragments and pAMA/ribo-igr2^PS1 (Supplementary Table 4). Many black genR, baR and suR transformants were obtained. Six transformants per selection system were studied by diagnostic PCR detecting the correct insertion in most of the transformants (Supplementary Figure 8A). Finally, the growth phenotypes of resistant strains and WT:A95 were examined on solidified SDNG with different concentrations of G418, GFS, or CME, using two different modes of inoculation (Figure 3; Supplementary Figure 8B). In the first method, 5 × 10⁵ cells per Petri dish were distributed with glass beads, which corresponds to the number of protoplasts per Petri dish used for transformation. The second inoculation method simulates the transfer of transformants to fresh selection medium for secondary selection by inoculating an undefined number of cells from 1-week-old surface-grown colonies using an inoculation loop. Generally, higher concentrations of the selective agents are needed to prevent growth of wild type A95, when undefined numbers of cells are streaked out on solidified medium. Based on these results (inhibitory concentrations) as well as observations from transformation approaches, the most suitable concentrations for primary and secondary transformant selection were defined (see Section Materials and Methods).

The performed experiments demonstrated the applicability of the three selection marker systems for transforming protoplasts of K. petricola. The set of compatible plasmids was expanded allowing for the PCR-based generation of gene replacement fragments with seven different resistance cassettes (five usable in K. petricola) by using a single pair of ~100-bp-long primers, and for allowing the PCR-based generation of expression constructs for different applications including knock-in in native gene loci (Supplementary Figure 6) or other genomic loci as described in the following.

Promoters are the basis for gene expression, though the nature of the gene and the terminator may significantly influence the stability and half-life of the transcript and the protein. Appropriate promoter activities are essential for most molecular tools, e.g., for driving expression of endogenous or foreign genes including reporter genes. Few promoters are known to mediate high and constitutive gene expression in several but not all fungi. Known functional promoters in K. petricola derive from A. nidulans and are: PtrpC used for expression of resistance cassettes, Ptef1 used for expression of cas9 from AMA1-bearing plasmids, and PgpdA and PoliC used for expression of fluorescent reporter gene constructs (Voigt et al., 2020) (Supplementary Table 3). However, in the same study inconclusive results were obtained for Pact1 from Botrytis cinerea that allows for moderate expression of gfp in different fungi. To learn more about the performance of already used promoters and to identify those enabling controllable gene expression, experiments were set-up to compare promoter (of interest, Poi) activities using gfp as reporter gene for quantifying GFP fluorescence and expressing these Poi::gfp constructs in identical genetic backgrounds. For taking advantage of the pigmentation phenotypes that are already visible on the transformation plates due to the haploid and uninucleate nature of K. petricola, the pks1 locus was chosen for targeted integration (knock-in) of the expression constructs. In addition, melanin-free strains appeared more favorable for quantifying GFP fluorescence. Six different promoters were included in the study (Table 3); the previously mentioned PoliC and PgpdA from A. nidulans, Pact1 from B. cinerea and Pgln1 from Fusarium fujikuroi as constitutive promoters, as well as PniaD from A. nidulans and Pgal1 from K. petricola (Supplementary Sequence 1) as candidates for controllable promoters. At least in some fungi, promoters of nitrogen assimilation and galactose metabolism genes underlie positive and negative regulation by substrate induction and catabolite repression, respectively (Krappmann and Braus, 2005; Christensen et al., 2011).

Expression constructs for the first four promoters were available (Schumacher, 2012), compatible constructs for PniaD and Pgal1 were cloned (Supplementary Table 3). As the application of CRISPR/Cas9 mediates efficient HR with short homologous (SH) sequences and the used plasmids/constructs comprise identical modular structures, the six different expression constructs could be amplified from the six plasmids with (four) primers binding in TniaD/TniiA of the resistance cassette and in Tgluc/TtrpC of the expression cassette, and containing ~75-bp-long 5′ overhangs homologous to noncoding regions of pks1 (Figure 4A; Supplementary Figure 5). The co-transformation of WT:A95 protoplasts with pAMA/ribo-pks1^PS2 and the six Poi::gfp expression constructs (Supplementary Table 4) resulted in plenty of pink hygR transformants (data not shown). The correct integration event at pks1 that left the flanking sequences intact, was exemplarily examined for 16 strains of PniaD::gfp and Pgal1::gfp by diagnostic PCR as described in Supplementary Figure 9 (data not shown). In fact, the expected amplicons for HR at 5′ and 3′ of pks1/the Cas9 cutting site were obtained in all transformants demonstrating the suitability of the applied black-pink screening for the fast identification of expression construct-containing transformants. For further analyses, three strains per Poi::gfp construct were chosen. Cell suspensions from seven-day-old liquid cultures were prepared, and different numbers of cells in 200 μl PBS were transferred to microtiter plates for detecting GFP fluorescence in a microplate spectrophotometer. This pilot experiment revealed that 10⁹ melanin-free GFP-expressing cells per well resulted in the most reliable data while no fluorescence could be detected for black GFP-expressing (PoliC::gfp) cells by this set-up. The experiment, i.e., cultivation of the pink Poi::gfp strains followed by the detection of fluorescence was repeated twice yielding the same results (Figure 4B). Thus, PoliC (100%) mediated the highest GFP fluorescence followed by PgpdA (54%) and Pgln1 (46%). Very low levels of fluorescence were detected for PniaD (19%), Pgal1 (6%), and Pact1 (2%). Overall, these data are in accordance with the observations made by fluorescence microscopy when images with identical exposure times were captured.

Inspired by the first promoter studies using gfp as reporter and pks1 as integration site, another approach was taken using the endogenous pks1 (pigmentation) as a reporter. The same plasmids were used as templates but only the resistance cassette with the Poi was amplified by PCR with primers attaching 75-bp-long homologous sequences to the 5′-noncoding region (as before) or the 5′ end of pks1 to the expression constructs (Figure 4A; Supplementary Figure 10). Thus, by co-transformation of these amplicons with pAMA/ribo-pks1^PS4 for inserting a DSB in the 5′-noncoding region of pks1 (Supplementary Table 4), the endogenous promoter was replaced by the Poi resulting in Poi::pks1. As “background” control, Ppks1 (1.0 kb) was replaced by the hygR cassette only (ΔPpks1). To obtain clearer nuances of melanin, the constructs were introduced into the carotenoid-free phs1- mutant. Obtained hygR transformants were checked by diagnostic PCR, and for two transformants per promoter, the 3′ region comprising the Poi-pks1 transitions were amplified and sequenced to confirm the absence of mutations in pks1. Cell suspensions of Poi::pks1 and control strains were dropped onto different solidified media including SDYG (control; Figure 4C) and media containing nitrate or galactose as single nitrogen or carbon source to induce PniaD and Pgal1, respectively. However, no obvious differences to the control plates were found (Figure 4C). The experiment revealed that deletion of 1 kb of the 5′-noncoding region (considered promoter) of pks1 did not eliminate pks1 expression by leading to gray colonies. In this expression system, Pgal1 showed the lowest activity by resulting in colonies pigmented similarly to the background control. All other Poi::pks1 fusions led to fully melanized colonies not allowing for detection of gradual promoter activities. The mutations Pgal1::pks1 and ΔPpks1::pks1 were introduced into the wild-type background as well for yielding mutants with reduced melanin contents for further phenotypic characterization (Supplementary Tables 1, 2; Figure 4D).

In sum, the promoter studies clarified that PoliC or PgpdA (and not Pact1) must be used for appropriate expression of fluorescent reporter genes in K. petricola. No controllable promoters could be identified in this pilot study; however, the established protocols for generating (melanin-free) strains with identical genetic backgrounds and studying promoter activities by quantification of GFP fluorescence will allow screening further candidate promoters in the future. Finally, the knock-in of expression constructs by replacing pks1 enables the rapid identification of correct (pink) transformants (black-pink screening).

To elucidate biological processes in detail, protein localization and protein-protein interaction studies in vivo might be crucial and are more favorable than performing interaction studies in heterologous hosts. Both co-localization and bimolecular fluorescence complementation (BiFC) studies require the integration of two expression constructs into the genome, and the two-step generation of such strains can be laborious and time-consuming.

The BiFC technology is based on the association of non-fluorescent protein fragments, e.g., split GFP or YFP that are fused to proteins of interest and expressed in living cells. A physical interaction of the proteins of interest will bring the reporter fragments in proximity, allowing the formation of the native reporter protein and the emission of its fluorescent signal. Therefore, BiFC studies identify both the localization and interaction partners of the protein of interest in different organisms including fungi (Kerppola, 2008). Co-localization studies refer to a spatial overlap between two (or more) different fluorescently labeled proteins with varying emission wavelengths. A suitable pair of fluorescent proteins represents GFP and DsRED/mCherry. In contrast to BiFC, these approaches are not suitable to detect protein-protein interactions, but they can be extremely helpful to locate the protein of interest to a specific cellular compartment by avoiding the application of fluorescent dyes.

With the aim of demonstrating the functionality of BiFC studies in K. petricola using available plasmids with split-GFP constructs of B. cinerea codon-optimized gfp (Leroch et al., 2011; Schumacher, 2012), a method was quested allowing for the easy and fast generation of strains containing two expression constructs. Considering the results for double editing by plasmid-based multiplex CRISPR/Cas9 and the well-working black-pink screening, the color-based screening method was expanded by using the phs1 locus for the knock-in of the second expression construct. Successful replacement of pks1 and phs1 by two expression constructs comprising hygR or natR cassettes, mediated by the two RNPs expressed from pAMA/tRNA-pks1^PS2-phs1^PS1, is expected to yield white hygR and natR transformants which can be immediately identified on the transformation plates and should contain both expression constructs (black-white screening) (Figure 5A).

As proof-of-principle, the interaction of the conserved White Collar GATA-type transcription factors (TFs) in K. petricola was studied. White Collar 1 is a blue light receptor that forms a complex (called WCC) with White Collar 2 in the nucleus to induce the transcription of light-responsive genes (Yu and Fischer, 2019). The K. petricola orthologs WCL1/2 (white collar-like1/2) contain the characteristic protein domains for sensing blue light (LOV—light-oxygen-voltage), for protein-protein interactions (PAS—Per-Arnt-Sim) and DNA binding (ZF—GATA-type zinc finger) (Supplementary Sequences 2, 3) and share 28/27% aa identity with A. nidulans LreA/B, 34/41% aa identity with Neurospora crassa WC-1/WC-2, 42/43% aa identity with B. cinerea WCL1/2, and 54/56% aa identity with the Exophiala dermatitidis orthologs. The coding regions of the K. petricola genes were amplified and fused to full-length gfp, mch and split gfp (gfpN, gfpC). As (the non-functional) Pact1 is contained in some basal plasmids (Pact1::gfpN), the promoter was replaced by PgpdA in the same cloning step, yielding fusion constructs wcl1-gfp, wcl1-mch, gfp-wcl2, wcl1-gfpC, and gfpN-wcl2 under control of PgpdA or PoliC (Supplementary Table 3). The expression constructs were amplified using plasmids as template and primers containing the sequences homologous to the 5′ and 3′-noncoding regions of pks1 or phs1 as 5′ overhangs. WT:A95 protoplasts were co-transformed with different combinations of expression constructs and pAMA/ribo-pks1^PS2 (single integration of wcl1/2 fused to full-length gfp/mch via selection with HYG) or pAMA/ribo-pks1^PS2-phs1^PS1 (double integration of wcl1/2 constructs at pks1 and phs1 via selection with HYG and NTC) as schematically depicted in Figure 5B and specified in Supplementary Table 4. As expected, pink hygR transformants were obtained for wcl1-gfp and gfp-wcl2 fusion constructs targeted to the pks1 locus and white hygR/natR transformants for pairs of wcl1/2 fusion constructs targeted to pks1 and phs1 (data not shown). Cell suspensions of three transformants per approach were examined by fluorescence microscopy revealing identical results. Thus, the single expression of WCL1-GFP or GFP-WCL2, and the co-expression of WCL1-mCH and GFP-WCL2 (co-localization approach) as well as of WCL1-GFP^C and GFP^N-WCL2 (BiFC approach) led to bright fluorescence in the nuclei indicating that the method works and that WCL1 and WCL2 do interact in the nuclei of K. petricola. For none of the included negative controls, i.e., strains co-expressing GFP^C + GFP^N, WCL1-GFP^C + GFP^N, or GFP^C + GFP^N-WCL2, fluorescence was detectable (Figure 5B, data not shown). As a pleasant side effect, it was found that white strains are more convenient for microscopy as no mechanical separation of the cells is necessary (compared to black cells) and cells do not form clumps (compared to black and pink cells).

In conclusion, the performed experiments demonstrated the practical application of BiFC coupled to a color-based screening approach for protein-protein interaction studies in K. petricola. Furthermore, the co-localization and interaction of the two K. petricola White Collar orthologs in the nucleus were proven. A set of four cloning vectors—pNAH-OGCGc, pNAH-OGCGn (Schumacher, 2012), pNDN-GGNTn, pNDN-GGNTc (this study)—is now available that allows to fuse the proteins of interest via variable linker sequences to the N- and C-terminal halves of GFP to test eight distinct combinations of fusion proteins for bimolecular fluorescence complex formation in K. petricola and beyond.

Ectopic integrations of expression constructs may disrupt other genes, or the integration in a silent genomic region (heterochromatin) may prevent the expression of the inserted constructs. By targeted integration of constructs in defined genomic regions (knock-in) such negative effects are bypassed, and strains with predictable genetic backgrounds are generated. The established color-based screening procedures are suitable for promoter, localization and BiFC studies for which the pigmentation phenotype can be ignored as far as genetic backgrounds are identical. However, the phenotypic characterization of complemented deletion mutants or over-expression strains requires neutral genetic backgrounds, i.e., that inserted genes are expressed from noncoding genomic regions leaving all genes intact.

Aiming the identification of such neutral insertion sites in the genome of K. petricola, two intergenic regions in transcriptionally active areas of contig 01 (data not shown), hereafter named igr1 and igr2, were chosen as candidates for the experimental validation. Igr1 and igr2 represent shared terminator sequences of two adjacent genes, have sizes of ~2.3 kb, exhibit regular GC contents of 50%, lack restriction sites for frequently used enzymes and contain appropriate Cas9 sites (PS+PAM) in the middle of the sequences (Figure 6A; Supplementary Sequences 4, 5). The PSs were fused with the sgRNA scaffold for yielding pAMA/ribo-igr1^PS1 and pAMA/ribo-igr2^PS1. Cloning vectors were assembled comprising ~1.5-kb-long sequences that flank the considered Cas9 cutting sites/insertion sites, adapter sequences and multiple cloning sites for subsequent cloning, different resistance cassettes for selection (pIGRXR) and additonal expression cassettes containing gfp or mcherry (pIGRXH-AGT, pIGRXN-OCT) (Supplementary Table 3; Supplementary Figure 11). For the generation of first igr insertion strains, WT:A95 protoplasts were co-transformed with a CRISPR plasmid (pAMA/ribo-igr1^PS1 or pAMA/ribo-igr2^PS1) and the corresponding insertion construct isolated by digestion from pIGR1H-AGT, pIGR1N-OCT, pIGR2H-AGT, or pIGR2N-OCT (Supplementary Table 4). Four resistant transformants per construct were screened by diagnostic PCR for the correct insertion in igr1 and igr2, respectively (Supplementary Figures 12, 13). Finally, the growth phenotypes of two strains per igr (from each approach) with verified insertion events and the wild-type strain A95 were assessed under 10 different stress conditions (Figure 6B). For this, cell suspensions of the strains were dropped onto solidified SDYG-based medium, that was adjusted to extreme pH values or supplemented with stress-inducing agents including hydrogen peroxide, calcofluor white and SDS. Heat and UV stress were applied to dropped cells on control medium (SDYG, pH 5) (see Supplementary Figure 14 for details). The tested igr1/2 insertion strains showed wild-type-like growth phenotypes under all conditions, suggesting that the insertion of the ~4-kb-long sequences into igr1 and igr2 does not result in obvious phenotypes.

To analyze whether genes inserted into igr1 or igr2 are efficiently expressed, further gfp expression cassettes were inserted into igr1 and igr2 as well as into ura3 (gene locus, considered as control; Supplementary Figure 15). For this, the expression cassettes were amplified by PCR using pNAH-GGG, pNDH-OGG, or pNDH-AGT as template and suitable primers attaching 75-bp-long homologous sequences to the expression constructs as described previously. These SH constructs were co-transformed with appropriate CRISPR plasmids (Supplementary Table 4), and six transformants per approach were screened by diagnostic PCR (Supplementary Figures 12, 13, data not shown). As a consequence, strains were generated that contain the same expression constructs (PoliC::gfp, PgpdA::gfp, or Pact1::gfp) at different genomic loci, i.e., at igr1 or igr2 via insertion in noncoding regions or at ura3 or pks1 via replacement of coding sequences resulting in wild-type-like and melanin-free pigmentation. Finally, cell suspensions of the strains were examined by fluorescence microscopy. Representative results for one strain each are shown (Figure 6C). Moderate levels of GFP fluorescence were detected for PoliC::gfp and PgpdA::gfp constructs integrated at the four different loci, indicating that fluorescence microscopy works for melanin-containing (black) and melanin-free (pink) cells and that inserted constructs at igr1 and igr2 are well expressed (Figure 6D). As observed before (Figure 4), PoliC mediated best GFP fluorescence, followed by PgpdA. Fluorescence levels of Pact1::gfp were comparable to those of the negative controls (gfp-lacking cells) confirming the unsuitablity of Pact1 for driving gene expression in K. petricola.

As a result, two intergenic regions in the genome of K. petricola were identified in which expression cassettes can be inserted by application of plasmid-based CRISPR/Cas9. Given the experimentally proven neutrality of insertions at igr1 and igr2, phenotypes of insertional strains are due to the expression of the introduced gene(s). Therefore, constructs can be inserted in both igr1 and irg2 in a single step using pAMA/tRNA−igr1^PS1−igr2^PS1 (Supplementary Table 3) e.g., for co-overexpression of two genes in a wild-type-like background. While expression constructs from any plasmid can be amplifed using primers with short homologous sequences to igr1 and igr2, the generated plasmids with long homologous sequences and five different resistance cassettes (pIGRXR) are particulary suitable for cloning of complementation constructs as depicted in Supplementary Figure 11. Constructs for transformation of K. petricola can be isolated by digestion due to attached multiple cloning sites (might be crucial for large constructs) or amplified by PCR using primers binding in the igr1/2^PS1-flanking sequences. In fact, first deletion mutants were successfully complemented by targeting expression cassettes to igr2 (R. Gerrits, pers. comm.).