The wild-type strain P. roqueforti CECT 2905 (ATCC 10110) was used in this work. The strain was routinely kept on Power medium agar (Fierro et al., 1996). Other media used in this work were Czapek minimal medium, YES agar (Rojas-Aedo et al., 2017) and CM liquid medium (Gil-Durán et al., 2015).

To generate the sfk1 knockdown construct, plasmid pJL43-RNAi (Ullán et al., 2008) was used. This plasmid has two promoters in opposite directions, and an NcoI site between the promoters (Ullán et al., 2008).

A 450-bp fragment of the sfk1 gene was amplified with primers RiSfkF (5′- AGACTCCCATGGCCACTGGACTATGATCG -3′) and RiSfkR (5′- AGACTCCCATGGAACGACAATGAACGAGGA -3′). Both primers contain an NcoI site (underlined). The amplified product was digested with NcoI and ligated into pJL43-RNAi (also previously digested with NcoI and dephosphorylated), giving rise to plasmid pSfk1-RNAi. The pSfk1-RNAi plasmid was confirmed by sequencing (data not shown) and used to transform P. roqueforti CECT 2905 according to the protocol described by Gil-Durán et al. (2015). In addition, a control strain that contains the empty plasmid pJL43-RNAi was obtained.

DNA and total RNA from P. roqueforti were extracted as described by Gil-Durán et al. (2015). RT-PCR experiments were performed as described by García-Rico et al. (2017). Briefly, total RNA extracted was treated with RNase-free DNase I (Roche, Germany) and then quantified using a μDrop plate in a MultiSkan GO quantification system (Thermo Scientific, Germany). To synthesize sfk1 cDNA, RevertAid Reverse Transcriptase (Thermo Scientific, Germany) was used according the manufacturer’ instructions. Reactions contained 200 ng of treated RNA and 10 pmol of primers RiSfkF and RiSfkR. Amplification of β-tubulin cDNA (internal control) was performed as described by García-Rico et al. (2017). The products of the reactions were resolved by electrophoresis and quantified by densitometry analysis using the “myImageAnalysis” software (Thermo Scientific, Germany). Values were expressed as percentage of the ratio between sfk1/β-tubulin.

The deduced sequence of Sfk1 protein from P. roqueforti CECT 2905 was used to find similar sequences using BlastP. Sequences obtained were used to perform phylogenetic analysis. Briefly, sequences were aligned by Clustal Omega and the phylogenetic trees were reconstructed using the Neighbor-joining method with the MEGA version 7 program (Kumar et al., 2016). Evolutionary distances data were calculated using the Poisson correction model. The support of internal nodes was performed by bootstrap analyses with 1,000 replications.

The apical extension rates, the production of conidia, and the measurements of conidial germination were done exactly as described by Gil-Durán et al. (2015). The thermal and hypertonic stress analyses were performed as described by García-Rico et al. (2017).

The cell wall integrity assays were performed using calcofluor white as described by Mendoza et al. (2016). Briefly, conidia from P. roqueforti at a final concentration of 1 × 10⁴ conidia/mL were inoculated in 24-well plates (lined with 12-mm glass coverslips) containing 1 mL of CM medium. Cultures were incubated at 28°C for 15 h until the germination of conidia. After this time, the medium was removed and glass coverslips with P. roqueforti hyphae adhered were washed three times with 0.9% NaCl, and stained with calcofluor white during 10 min. Then the glass coverslips containing hyphae were washed again and mounted on slides. The fluorescence of P. roquerfoti hyphae stained with calcofluor white was observed under a fluorescence microscope with Blue/Violet (110033V2) filter. The fluorescence intensity was quantified using the program “myImageAnalysis” v2.0 (Thermo Scientific, Germany). For this purpose, the fluorescence intensity of hyphae from three randomly-chosen fields of each sample was recorded. The average of the fluorescence intensities recorded from the wild-type strain was set as 100% of fluorescence. The rest of the samples were normalized with respect to the wild-type strain. As control of cell wall damage, germinated conidia from the wild-type strain were treated with Lysing Enzymes from Trichoderma harzianum (20 mg/mL) during 2 h at 28°C, and stained with calcofluor white as above.

The production, extraction and HPLC analyses of mycophenolic acid and andrastin A from P. roqueforti CECT 2905 have been described elsewhere (Del-Cid et al., 2016; Rojas-Aedo et al., 2017). In the case of roquefortine C, the strains were grown on YES agar for 15 days at 28°C. The mycelia were collected, suspended in a chloroform-methanol (2:1) mixture, and then submitted to sonication during 30 min. The mixture was filtered through nylon membranes (mesh 30 μm). The extracted mycelia were kept for further dry-weight determination (see below), whereas the filtrated was evaporated to dryness in a rotary evaporator. Previous to HPLC analysis, the dried filtrated was suspended in 500 μl of methanol (HPLC grade) and briefly centrifuged. Twenty μl of this suspension were submitted to HPLC analysis. HPLC analysis was performed using a Waters 1525 HPLC equipment (Waters, Ireland) with a 4.6 mm × 250 mm (5 μm) SunFire C18 reverse phase column. The column was held at 35°C. The mobile phase was water with 0.02% TFA (solvent A) and acetonitrile with 0.02% TFA (solvent B). The gradient program was as follows: 15% B to 47% B linear over 15 min, 47% B to 100% B over 1 min, and isocratic 5 min. The flow rate used was 1.2 mL/min. For the detection and quantification of the compound, pure roquefortine C (Santa Cruz Biotechnology) was used as standard. The detection of roquefortine C was performed at 254 nm. Under the conditions described, roquefortine C showed a retention time of 10.9 min. Roquefortine C production was normalized by dry weight of fungal mycelia. For this purpose, the mycelium from each sample was dried as described by García-Rico et al. (2009).

The nucleotide sequence of the sfk1 gene described in this work has been deposited in the GenBank database under accession number MF614690.

In a previous work, we performed SSH experiments to looking for genes from P. roqueforti CECT 2905 that are regulated by Gαi (Gil-Durán et al., 2015). Among the sequences obtained from these experiments, a full cDNA matching with hypothetical sfk1 genes from ascomycetous fungi (mainly Penicillium species) was obtained. This cDNA sequence is 914 nt- long, excluding the poly-A tail. By using genomic DNA from P. roqueforti CECT 2905 and suitable primers (designed from the cDNA sequence), we amplified the sfk1 gene by PCR. The gene was sequenced and compared with the cDNA sequence, revealing the presence of four introns (data not shown). The sfk1 gene from strain CECT 2905 was also submitted to BlastN analysis against the genome of P. roqueforti FM164 (Cheeseman et al., 2014). We found that sfk1 gene from P. roqueforti CECT 2905 exactly matches (100% identity, 100% coverage, and no gaps) with the sfk1 gene from strain FM164, located at contig ProqFM164S01-10 in the genome sequence of this strain (data not shown).

The sfk1 gene from P. roqueforti CECT 2905 encodes a protein of 235 amino acids. BlastP analysis of the deduced protein revealed the presence of orthologs in a wide range of species of filamentous fungi from the phylum Ascomycota, mainly from the genera Penicillium, Aspergillus, and Talaromyces. Using several of these sequences, a phylogenetic analysis was performed (Figure 1A). This analysis revealed that Sfk1 sequences are clustered in agreement with the expected evolutionary relationship among these genera. On the other hand, using all Sfk1 sequences from Penicillium species retrieved by BlastP (22 sequences including Sfk1 from P. roqueforti), a second phylogenetic analysis was performed (Figure 1B). Eighteen sequences (including Sfk1 from P. roqueforti) belong to species from the subgenus Penicillium, whereas five sequences belong to species from the subgenus Aspergilloides (Figure 1B). In this case, the Sfk1 sequences also are clustered in agreement with the expected evolutionary relationship among these Penicillium species (Figure 1B).

Finally, and according to bioinformatics predictions (TMHMM Server v. 2.0 and TMpred at Expasy), the Sfk1 protein from P. roqueforti is a transmembrane protein. It contains six putative transmembrane helix domains, with its amino and carboxyl ends inside oriented (data not shown).

RNA-mediated gene silencing was used to perform the functional characterization of sfk1 in P. roqueforti CECT 2905. The fungus was transformed with pSfk1-RNAi, and 57 phleomycin-resistant transformants were obtained. Fifteen transformants were randomly selected and subjected to RT-PCR (see section “Materials and Methods”). Two of these transformants (hereafter SFK19 and SFK38) showed a reduction of sfk1 transcripts of approximately 55% compared to the wild-type strain (Figure 2A). In addition, the presence of silencing cassette in these transformants was also confirmed by PCR (Figure 2B). Taken together, all these data indicate the successful silencing of sfk1 in transformants SFK19 and SFK38.

Since the deletion of sfk1 produces changes in the growth pattern of S. cerevisiae (Jin et al., 2008), we analyzed the effect of the RNA-mediated silencing of sfk1 on growth in P. roqueforti. Figure 3A shows that the colonies of the transformants SFK19 and SFK38 grown on Czapek minimal medium or Power rich medium were smaller than the wild-type strain, suggesting that the knockdown of sfk1 altered growth of P. roqueforti. Figure 3B shows the apical extension rates of transformants SFK19 and SFK38 on the same media. As observed, and independently from the medium used, the attenuation of sfk1 altered normal growth of the fungus. On Czapek medium, the apical growth rate of strains SFK19 and SFK38 ranged between 0.170 and 0.185 mm/h, versus 0.240 mm/h in the wild-type strain, whereas on Power medium, the apical growth rate of strains SFK19 and SFK38 ranged between 0.20 and 0.22 mm/h, versus 0.31 mm/h in the wild-type strain (Figure 3B). These results indicate that on Czapek medium, the transformants grew at a rate between 72 and 77% of wild-type strain levels. Similarly, on Power medium the transformants grew at a rate between 64 and 75% of wild-type strain levels. Therefore, we conclude that the knockdown of sfk1 in P. roqueforti caused a reduction in the apical extension rate.

We also tested conidia production in all the strains. Interestingly, we found that the RNA-mediated silencing of sfk1 does not affect conidiation (Figure 4A). However, the knockdown of sfk1 affected conidial germination. Figure 4B shows that all the strains follow a sigmoidal pattern of conidial germination, but in transformants SFK19 and SFK38, this process was earlier (around 1 h) compared with the wild-type strain. For example, at 9 h, around 20% of conidia from transformants have already germinated, whereas the wild-type strain reaches the same percentage of germination at 10 h. Similar differences are observed along all the exponential phase of germination (Figure 4B).

In S. cerevisiae, Sfk1 is important for response to heat shock (Audhya and Emr, 2002). For this reason, we tested thermal stress resistance of P. roqueforti strains. Remarkably, we did not found any difference in the ability to resist heat shock between the wild-type strain and transformants SFK19 and SFK38 (Figure 5A). On the other hand, we also tested hypertonic stress resistance, and we found interesting differences. Thus, the RNA-mediated silencing of sfk1 decreased osmotic stress resistance to NaCl or KCl (Figure 5B). As average, the relative apical extension rate of transformants SFK19 and SFK38 was approximately 66% of wild-type levels in the presence of KCl or NaCl.

It has been suggested that Sfk1 could be important for cell wall integrity (see section “Discussion”), so we analyzed the effect of the RNA-mediated silencing of sfk1 on the cell wall integrity in P. roqueforti by using calcofluor white (Figure 6). This compound is a fluorochrome molecule which binds to the chitin in the cell walls of the fungi. Therefore, a decrease in cell wall fluorescence intensity produced by calcofluor white is indicative of cell wall damage (Mendoza et al., 2016). Our results indicate that compared to the wild-type strain, transformants SFK19 and SFK38 have lower fluorescence intensity (Figure 6A). In the case of strain SFK19, it shows approximately 74% of the fluorescence of wild-type strain levels, whereas in SFK38 the effect is more dramatic, because this strain shows approximately 45% of the fluorescence of wild-type strain levels (Figure 6B). These data suggest that the knockdown of sfk1 affected the cell wall integrity in transformants SFK19 and SFK38.

Roquefortine C, mycophenolic acid, and andrastin A are some of the main secondary metabolites produced by P. roqueforti (García-Estrada and Martín, 2016), so we studied if RNA-mediated silencing of sfk1 affected their production. As shown in Figure 7 the production of these secondary metabolites is depleted in transformants SFK19 and SFK38 compared to the wild-type strain. In the case of andrastin A, the wild-type strain produces approximately 900 μg/mg of the compound, whereas in the same conditions, the transformants SFK19 and SFK38 produce approximately 250 μg/mg (Figure 7B). More evident is the effect on roquefotine C. In this case, the wild-type strain produces approximately 560 μg/mg, whereas the transformants produce barely 15 μg/mg (Figure 7B). Finally, the most drastic effect was observed for mycophenolic acid. The wild-type strain produces approximately 300 μg/mg of this compound, but is almost undetectable in transformants SFK19 and SFK38 (Figure 7B). These results strongly suggest that the knockdown of sfk1 decreases the production of the main secondary metabolites from P. roqueforti.