To bridge the gap between drug discovery and development, the characterization of drug targets has been a long-thought goal (Kramer and Cohen, 2004; Schenone
We have constructed a genome-wide gene deletion library with built-in bar codes in a gene-specific manner in fission yeast (Winzeler
Despite the recent advent of the cutting-edge CRISPR technology, the yeast screening system remains useful because of easier genetic handling and bar code detection. Since yeasts are classified as fungi, the yeast model system would be the best choice for screening of antifungals (Sipiczki, 2000). Terbinafine (TRB) has been developed as the first oral antimycotic in the allylamine class (Ostrosky-Zeichner
In this study, we identified target genes of TRB using the heterozygous gene deletion library and characterized the action mechanisms of the screened target genes as a proof-of-concept that supports harnessing this genome-wide screening system for the identification and characterization of target genes affected under any condition of interest.
All chemicals and reagents were obtained from Sigma-Aldrich (St Louis, MO, USA), unless stated otherwise. Yeast extract and agar were purchased from BD Difco (Sparks, MD, USA). TRB was provided by Korea United Pharm. Inc (Seoul, Korea). All oligonucleotides were purchased from Bioneer (Daejeon, Korea).
For the systematic screening of TRB target genes in fission yeast, we used the heterozygous gene deletion library (2N) constructed in a previous study (Kim
For the loss-of-function analyses of non-essential genes, cognate haploid gene deletion strains (1N) were derived from heterozygous gene deletion strains, if available. The ED668 strain (
The genome-wide screening of TRB target genes was performed via microarrays, as previously reported (Kim
Intracellular squalene levels were measured as previously described (Takami
To compare the relative drug sensitivity of target strains over control strains (SP286 for heterozygous and ED668 for haploid gene deletion strains), a sensitivity pattern map was employed by visualization of the relative sensitivity data from spotting assays, using a heat map. When the target strains showed no, mild (S,<5-fold), moderate (SS, 5-25-fold), or severe (SSS,>25-fold) susceptibility compared with the control strains, the relative drug sensitivity was represented in blue, light blue, light red, or red on the heat map, respectively.
To assess drug-drug target interactions by co-treatments of the heterozygous target strains with TRB and the other drug (ECZ or CHX), isobologram analysis was performed as previously described with slight modifications (Chou, 2006). In brief, IC50 values of a pair of drugs were determined individually in each target strain using 96 deep-well plates, and the IC50 values of drugs were plotted at
For the complementation test, the
As a first step of molecular modeling, protein databank (PDB) data were retrieved from RCSB PDB (https://www.rcsb.org/) and all amino acid sequence data were retrieved from the UniProt database (http://www.uniprot.org) as follows: human SQLE (ortholog of yeast Erg1, 574 amino acids), PDB code: 6C6P, UniProt ID: Q14534; human EIF3B (814 amino acids), PDB code: 5K1H, UniProt ID: P55884; budding yeast Erg1p (496 amino acids), UniProt ID: P32476; budding yeast Prt1p (ortholog of fission yeast Tif302, 763 amino acids), PDB code: 4U1F, UniProt ID: P06103; fission yeast Erg1 (457 amino acids), UniProt ID: Q9C1W3; fission yeast Tif302 (725 amino acids), UniProt ID: Q10425. After obtaining the amino acid sequences of interest, they were aligned to one another via BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi/). Using the available crystal structures, including the human SQLE, human EIF3B, and budding yeast Prt1p proteins as templates, unknown protein structures of interest, such as fission yeast Erg1 and Tif302 and budding yeast Erg1p were built via homology modeling using MODELER version 9.20 in Discovery Studio 2019 (BIOVIA, San Diego, CA, USA). Docked binding modes were obtained using the homology structure models. Among several homology structure models, the best one was selected based on the lowest probability density function (PDF) total energy and discrete optimized protein energy (DOPE) scores. The AutoDock 4.2.6 (The Scripps Research Institute, La Jolla, CA, USA) was then employed for molecular docking to assess the interactions between TRB and potential binding sites of homology structure models. The three dimensional (3D) structure of proteins and TRB was processed using AutoDock Tools 1.5.6. AutoGrid program (The Scripps Research Institute) was used for generating 3D affinity grid fields with a grid size and spacing of 40×40×40 Å3 and 0.375 Å, respectively. A total of 100 docking runs, 25×105 energy evaluations, and 27,000 iterations were carried out using the Lamarckian genetic algorithm method. For more details, see the supplementary material.
All experiments were analyzed using triplicate samples and repeated at least 3 times. Data are presented as the mean ± SD, unless indicated otherwise. Statistical comparisons between groups were performed using Student’s
Through the primary microarray screening and secondary spotting assay against TRB, we identified 24 heterozygous target strains corresponding to 7 essential and 17 non-essential genes (Supplementary Table 1). According to GO analysis of the relevant target genes in terms of biological process, 14 genes encoding mostly ribosomal proteins were related to ‘translation’ with a significant (
For further in-depth functional studies, we narrowed down the number of TRB target strains to 14 (10 non-essential and 4 essential genes) by using a cut-off of severe susceptibility (i.e., SSS), as shown in Table 1. According to GO analysis in terms of biological process, except for the well-known
Considering that the 14 heterozygous targets were identified by virtue of the TRB-induced haploinsufficiency, their cellular levels of
If all target strains had lower levels of
Taken together, the results indicate that the sensitivity of all heterozygous TRB targets strains screened under the principle of drug-induced haploinsufficiency is attributed to decreased levels of
When the relevant genes of heterozygous target strains were non-essential, their cognate haploid deletion strains were employed for elucidation of molecular mechanism through loss-of-function (knockout) analysis. Out of the 10 heterozygous target strains of non-essential genes, 9 cognate haploid strains corresponding to 7 RP and 2 SAGA-subunit genes, except for
To check whether the TRB-induced haploinsufficiency in the heterozygous target strains would reproduce in the cognate haploid deletion strains, we performed the sensitivity pattern map analysis with the 9 available haploid deletion strains against the same set of drugs used with the heterozygous target strains (Fig. 1A) by using the ED668 and 7 deletion haploids of random genes as controls (Fig. 2A). Surprisingly, compared with the controls, none of the 7 haploid deletion strains of RP genes showed susceptibility against the antifungals targeting ergosterol synthesis, whereas the 2 haploid deletion strains of SAGA-subunit genes (
To elucidate underlying mechanisms of the above phenomenon, first
The above results prompted us to elucidate the underlying action mechanism. The clue for the answer came from the experimental ratios of
As the first step to answer, we applied 3 speculations based on the experimental ratios of
Next, we took a couple of corrections into consideration to get the final numbers of repressor complexes: (i) the number of functional repressor complexes should be corrected based on a unit number of
We identified, for the first time, 3 novel essential genes (
First, SP286 was co-treated with a pair of ergosterol-targeting antifungals, TRB and ECZ, to expect a synergy via positive effects (Cavalheiro
However, it was hard to speculate how Tif302 could affect the cellular levels of
As all the heterozygous TRB target strains showed higher susceptibility to CHX compared with the SP286 control cells (Fig. 1C), CHX is likely to be related with modulation of ergosterol levels. In this regard, the strains were co-treated with TRB and CHX to gain more insight on the action mechanism of
To explain the serendipitous antagonistic effects, the action points of the 2 antifungals should be overlapping at least at a single point. As shown in the proposed model (Fig. 5B), Tif302 should be an overlapping point required for efficacy of both antifungals, as judged by the following findings. (i) All the heterozygous TRB target strains showed higher susceptibility to CHX compared with the SP286 control cells (Fig. 1C). The IC50 values of CHX decreased to 15-25 µM from 30 µM in SP286, as shown in the y-axis in the isobologram graphs. The lower levels of Erg1 made cells more vulnerable to stress (Bhattacharya
The above findings prompted us to assess whether TRB directly interacts with the Tif302 protein by the method of molecular modeling (Fig. 6, see Supplementary Fig. 1 and 2, and Supplementary Table 3 for more details). Regarding Erg1, TRB was found to bind to the 2 yeast Erg1 proteins with a binding energy of <–7 kcal/mol, which was lower than the –3.73 kcal/mol of the human SQLE protein (human ortholog of yeast Erg1). Based on the recent finding that human SQLE is a partial TRB target with a higher binding energy of –3.73 kcal/mol that the <–7.5 kcal/mol of yeast Erg1 (i.e.>1,000-fold higher IC50 than yeast Erg1) (Nowosielski
If TRB binds to Tif302, TRB treatment would affect the translation yield in a dose-dependent manner. Thus, we measured the translation yields in response to TRB treatment using GST as a reporter in wheat germ extracts (Supplementary Fig. 3). Expectedly, TRB treatment inhibited the
Overall, TRB has a secondary off-target of fission yeast Tif302, besides the conventional well-known Erg1 target. From the point of view of the off-target of Tif302, Tif302 can serve as a TRB target through 2 different action mechanisms, such as the decrease in
As shown in Fig. 7, we compared the target genes of non-essential genes identified from the heterozygous deletion library in the study with those previously reported from the haploid deletion library (Fang
With technological advances in the field of genomics, yeast gene deletion libraries (Winzeler
Compared with the previous study with the haploid gene deletion library (Fang
These novel findings shed light on some aspects of the antifungal roles of Tif302. First, Tif302 would be considered as a potential antifungal target, as its inhibition by TRB affected the translation yield (Supplementary Fig. 3). Consistently, there is an accumulating body of evidence that EIF3B could be a potential anti-cancer target in human cancers (Wang
In addition, we observed an unexpected phenomenon that non-essential RP genes are TRB targets only in heterozygous, but not in haploid, deletion strains, as shown in Fig. 3. Based on the comparison of the experimental ratio of
It is of note that all human orthologs of the heterozygous target genes have been reported to be associated with diverse human diseases, as shown in Table 1. Surprisingly, the human orthologs of the 5 RP-encoding genes screened in the present study are associated with the congenital erythroid aplasia DBA. Thus far, 21, including 19 RP genes, have been reported to be associated with human DBA via haploinsufficiency of RP genes caused by point mutations (Ulirsch
In this study, we have identified both essential and non-essential TRB target genes by the detection of gene-specific bar codes through microarrays. The detection method of bar codes is on upgrade from microarrays to next-generation sequencing, which will confer more accuracy to the screening system without bar code errors (Han
This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Ministry of Science & ICT (MSIT; NRF-2017M3A9B5060881, NRF-2017M3C9A5028693, and NRF-2017M3A9B5060880). In addition, human resources were supported by Chungnam National University and KRIBB. We also thank Bioneer for providing us with the fission yeast gene deletion library.
There are no conflict of interest.