Inhibiting fungal multidrug resistance by disrupting an activator-Mediator interaction.

Eukaryotic transcription activators stimulate the expression of specific sets of target genes through recruitment of co-activators such as the RNA polymerase II-interacting Mediator complex. Aberrant function of transcription activators has been implicated in several diseases. However, therapeutic targeting efforts have been hampered by a lack of detailed molecular knowledge of the mechanisms of gene activation by disease-associated transcription activators. We previously identified an activator-targeted three-helix bundle KIX domain in the human MED15 Mediator subunit that is structurally conserved in Gal11/Med15 Mediator subunits in fungi. The Gal11/Med15 KIX domain engages pleiotropic drug resistance transcription factor (Pdr1) orthologues, which are key regulators of the multidrug resistance pathway in Saccharomyces cerevisiae and in the clinically important human pathogen Candida glabrata. The prevalence of C. glabrata is rising, partly owing to its low intrinsic susceptibility to azoles, the most widely used antifungal agent. Drug-resistant clinical isolates of C. glabrata most commonly contain point mutations in Pdr1 that render it constitutively active, suggesting that this transcriptional activation pathway represents a linchpin in C. glabrata multidrug resistance. Here we perform sequential biochemical and in vivo high-throughput screens to identify small-molecule inhibitors of the interaction of the C. glabrata Pdr1 activation domain with the C. glabrata Gal11A KIX domain. The lead compound (iKIX1) inhibits Pdr1-dependent gene activation and re-sensitizes drug-resistant C. glabrata to azole antifungals in vitro and in animal models for disseminated and urinary tract C. glabrata infection. Determining the NMR structure of the C. glabrata Gal11A KIX domain provides a detailed understanding of the molecular mechanism of Pdr1 gene activation and multidrug resistance inhibition by iKIX1. We have demonstrated the feasibility of small-molecule targeting of a transcription factor-binding site in Mediator as a novel therapeutic strategy in fungal infectious disease.

Eukaryotic transcription activators stimulate the expression of specific sets of target genes through recruitment of co-activators such as the RNA polymerase II-interacting Mediator complex[1,2]. Aberrant function of transcription activators has been implicated in a number of diseases. However, therapeutic targeting efforts have been hampered by a lack of detailed molecular knowledge of the mechanisms of gene activation by disease-associated transcription activators. We previously identified an activator-targeted three-helix bundle KIX domain in the human MED15 Mediator subunit that is structurally conserved in Gal11/Med15 Mediator subunits in fungi[3,4]. The Gal11/Med15 KIX domain engages pleiotropic drug resistance transcription factor (Pdr1) orthologues, which are key regulators of the multidrug resistance (MDR) pathway in S. cerevisiae and in the clinically important human pathogen Candida glabrata[5,6]. The prevalence of C. glabrata is rising, partly due to their low intrinsic susceptibility to azoles, the most widely used antifungal[7,8]. Drug-resistant clinical isolates of C. glabrata most commonly harbour point mutations in Pdr1 that render it constitutively active[9-14] suggesting that this transcriptional activation pathway represents a linchpin in C. glabrata MDR. We have carried out sequential biochemical and in vivo high-throughput screens to identify small molecule inhibitors of the interaction of the C. glabrata Pdr1 activation domain with the C. glabrata Gal11A KIX domain. The lead compound (iKIX1) inhibits Pdr1-dependent gene activation and re-sensitizes drug-resistant C. glabrata to azole antifungals in vitro and in animal models for disseminated and urinary tract C. glabrata infection. Determining the NMR structure of the C. glabrata Gal11A KIX domain provided a detailed understanding of the molecular mechanism of Pdr1 gene activation and MDR inhibition by iKIX1. We have demonstrated the feasibility of small molecule targeting of a transcription factor-binding site in Mediator as a novel therapeutic strategy in fungal infectious disease. Based on our previous findings that deletion of the KIX domain of Saccharomyces cerevisiae GAL11 or Candida glabrata GAL11A abrogates Pdr1-dependent transcriptional responses and xenobiotic tolerance we hypothesized that the CgPdr1-CgGal11A interaction interface might serve as a promising target for novel anti-MDR compounds[3]. A fluorescently tagged CgPdr1 activation domain (AD) was used in an in vitro fluorescence polarization (FP) screen[15] of ~140,000 chemically diverse compounds to identify small molecules that block the interaction between the CgGal11A KIX domain and the CgPdr1 AD (Extended Data Fig. 1). Based on the high degree of conservation between S. cerevisiae and C. glabrata, we followed up the top hits from the FP screen with an azole growth inhibition screen in S. cerevisiae to identify hits with in vivo efficacy (Fig. 1). We identified 5 compounds that reproducibly inhibited growth in a concentration-dependent manner only in the presence of ketoconazole (Extended Data Fig. 1). The most potent compound is referred to as iKIX1 (Fig. 1). In vitro binding studies revealed that the Kd of the CgPdr1 AD for the CgGal11A KIX domain is 0.32 µM and the apparent Ki for iKIX1 is 18 µM (Fig. 1).

Extended Data Figure 1

Left: Table of compound libraries that were screened using a fluorescence polarization assay at the Institute of Chemistry & Cell Biology (ICCB) facility at Harvard Medical School.

Right: An S. cerevisiae viability screen identifies small molecules that preferentially inhibit growth of S. cerevisiae in a concentration-dependent manner in the presence of 5 µM ketoconazole (KET). Top hits from the screen are shown; OD600 values are the average of values from duplicate plates.

Figure 1

Discovery of inhibitors of the CgGal11A KIX-CgPdr1AD interaction interface

(a) Schematic of the screening process.

(b) iKIX1 inhibits cell growth in a concentration-dependent manner in the presence of 5 µM ketoconazole (KET); error bars represent means +/− s.d. from duplicate plates.

(c) iKIX1 structure.

(d) FP titration curve showing the interaction of CgGal11A KIX domain with CgPdr1 AD30 fitted to a Kd of 319.7 nM ± 9.5 nM (left). iKIX1 competes out CgPdr1 AD30 with an IC50 of 190.2 µM ± 4.1 µM (right). The measured Kd and IC50 values were used to calculate an apparent Ki of 18.1 µM for iKIX1. Data represent mean of two replicates and standard error from the fit is shown.

To facilitate the elucidation of the mechanism of action of iKIX1, we determined the high-resolution solution structure of the CgGal11A KIX domain with a backbone RMSD of 0.7 Å (Fig. 2 and Extended Data Table 1; PDB # 4D7X). The CgGal11A KIX domain has 51% sequence identity and 61% similarity with the S. cerevisiae Gal11/Med15 KIX domain[3] with an overall RMS deviation of 2.0 Å (Fig. 2 & Extended Data Fig. 2). The CgGal11A KIX domain forms a three-helix bundle harbouring an extensively hydrophobic core and a short helix at the N-terminus (Fig. 2). We determined interaction interfaces of the CgGal11A KIX domain with the CgPdr1 AD and iKIX1 (Fig. 2) by chemical shift perturbation (CSP) analysis. The CgPdr1 AD and iKIX1 target the same large hydrophobic groove harboured by the three helices. Residues from all three helices constitute the interaction interface, and titration of an ILV-methyl labeled CgGal11A KIX domain reveals large CSPs on the three leucines (L19, L23 and L51) upon addition of CgPdr1 AD and iKIX1 (Extended Data Fig. 2). The basic interaction interface on the KIX domain complements the acidic residues of CgPdr1 AD (Fig. 2). Residues of the CgGal11A KIX domain that interact with CgPdr1 AD and iKIX1 overlap strongly, suggesting direct competitive binding as the mechanism of inhibition. Docking of iKIX1 to the CgGal11A KIX domain suggests extensive hydrogen bonding and hydrophobic interactions between iKIX1 and KIX domain residues (Fig. 2 and Extended Data Fig. 2), matching the interaction interface mapped by CSP analysis.

Figure 2

Elucidation of the CgGal11A KIX domain structure; CgPdr1 AD and iKIX1 bind to a similar interface on the CgGal11A KIX domain

(a) Backbone representation of the 10 lowest energy NMR structures of CgGal11A KIX domain; backbone RMSD ~ 0.7 Å, left. Overlay of CgGal11A (purple) and S. cerevisiae Gal11/Med15 (blue) KIX domains, with an overall RMSD of 2.0 Å[25], right.

(b) Chemical shift perturbations (CSPs) on the CgGal11A KIX domain in the presence of CgPdr1 AD (red) or iKIX1 (blue). Residues coloured in red or blue indicate a chemical shift perturbation greater than 2 s.d. Residues highlighted in green (L19, L23 and L51) represent significant CSPs in the side-chain methyl groups of an ILV labeled sample.

(c) The iKIX1 and CgPdr1 AD target the hydrophobic groove on the CgGal11A KIX domain, which is surrounded by a basic patch. Residues H43, K54, K68, K78, R79, F47 and M72 present a positive electrostatic surface enclosing the binding interface.

(d) iKIX1 docked to the CgGal11A KIX domain. iKIX1 is depicted as red sticks and spheres. Residues that experience significant methyl CSP upon addition of iKIX1 are depicted as blue sticks.

Extended Data Table 1

NMR and refinement statistics for CgGal11A KIX domain

Summary of quality statistics for the ensemble of 10 structures calculated with AMBER explicit water refinement and list of experimental restraints.

	Protein
NMR distance and dihedral constraints
Distance constraints
Total NOE	1718
Intra-residue	602
Inter-residue	1116
Sequential (|i−j| = 1)	517
Medium-range (|i−j| ≤ 4)	488
Long-range (|i−j| ≥ 5)	111
Hydrogen bonds	0
Total dihedral angel restraints	158
phi	79
psi	79
Structure statistics
Violations (mean and s.d.)
Distance constraints (Å)	0.083 ± 0.039
Dihedral angel constraints (°)	2.407 ± 0.357
Max. dihedral angel violation (°)	10.267
Max. distance constraint violation (Å)	0.248
Deviations from idealized geometry
Bond lengths (Å)	0.013
Bond angles (°)	1.9
Average pairwise r.m.s.d.** (Å)
Heavy	1.1
Backbone	0.7

Pairwise r.m.s.d. was calculated among ordered residues (3–83) of 10 refined structures.

Extended Data Figure 2

(a) 2-dimensional representation of the H-bonding network between the CgGal11A KIX domain and iKIX1 based on docking studies.

(b) Chemical shift perturbations (CSPs) of ILV methyl resonances. Left: 1H-13C HSQC showing ILV methyl resonances of CgGal11A KIX domain in presence (brown) and absence (teal) of CgPdr1 AD (2-fold excess). Right: 1H-13C HSQC showing ILV methyl resonances of CgGal11A KIX domain in presence (purple) and in absence (teal) of iKIX1 (4-fold excess). Three leucines (L19, L23, L51) show significant CSPs in both spectra.

(c) Sequence alignment of the C. glabrata Gal11A and S. cerevisiae Gal11/Med15 KIX domains[26].

To assess the in vivo effects of iKIX1 on Pdr1-dependent transcription, we initially utilized a strain in which the two S. cerevisiae PDR1 orthologues (ScPDR1 and ScPDR3) are deleted and which carries a plasmid expressing CgPDR1[3], and a heterologous luciferase gene driven by 3 pleiotropic drug response elements (PDREs). Luciferase activity was strongly induced by ketoconazole treatment; iKIX1 co-treatment was able to block this induction in a concentration-dependent manner (Fig. 3).

Figure 3

iKIX1 blocks Gal11/Med15 recruitment and upregulation of Pdr1 target genes

(a) iKIX1 inhibits ketoconazole (KET)-induced upregulation of luciferase activity in a dose-responsive manner in a Sc pdr1Δpdr3Δ strain containing plasmid-borne CgPDR1 and 3XPDRE-luciferase. (UT): untreated control; ** P<0.001.

(b,c) iKIX1 prevents the ketoconazole (KET)-induced recruitment of Gal11/Med15/Mediator to the upstream activating sequences (UAS) of the PDRE-regulated promoter ScPDR5 (B), and ScPDR5 induction (C). Representative experiment from two biological replicates (ChIP DNA and RNA from same experiment) is shown. Error bars indicate s.d. of technical replicates; *** P<0.00001, ** P<0.0005, and * P<0.01 by two-tailed Student’s t-test.

(d) iKIX1 inhibits ketoconazole (KET)-induced transcriptional upregulation of CgCDR1 and CgCDR2 in a CgPDR1 wild-type strain (SFY114). ** P<0.005.

(e) RNA-Seq analysis of a C. glabrata SFY114 (PDR1 wild-type) strain pre-treated with iKIX1 or vehicle alone then induced with ketoconazole (iKIX1 + KET and KET, respectively).

(f) iKIX1 inhibits xenobiotic-induced CgPdr1 transcription in CgPdr1 gain-of-function mutants (amino acid changes indicated). Samples shown were induced with ketoconazole. * P<0.05 and ** P<0.01 as compared to DMSO + ketoconazole control.

(g) iKIX1 inhibits rhodamine 6G efflux in C. glabrata as compared to vehicle control. * P<0.05, ** P<0.005 as compared to DMSO + ketoconazole control.

(a,d,e,f,g) Data represent the means of three biological replicates. Two-tailed student’s t-test used to determine P values; error bars represent means +/− standard deviation.

A chromatin immunoprecipitation (ChIP) assay was used to examine Gal11/Med15 recruitment to Pdr1-regulated target genes in S. cerevisiae after iKIX1 treatment. Gal11/Med15 was rapidly recruited to the promoters of the Pdr1 target genes PDR5 and SNQ2 after ketoconazole addition; in contrast, ketoconazole-induced recruitment of Gal11/Med15 was abrogated when the cells were pre-treated with iKIX1 (Fig. 3, Extended Data Fig. 3). iKIX1 did not impede the constitutive occupancy of Pdr1 at the same Pdr1-regulated target genes (Extended Data Fig. 3). Consistent with the ChIP data, iKIX1 strongly inhibited azole-induced transcription of ScPdr1 target genes (Fig. 3, Extended Data Fig. 3).

Extended Data Figure 3

(a) iKIX1 prevents the ketoconazole (KET)-induced recruitment of ScGal11/Med15/Mediator to the upstream activating sequences (UAS) of the PDRE-regulated promoter ScSNQ2 and transcriptional upregulation of ScSNQ2.

(b) HA-Pdr1 occupies PDRE-regulated promoters of ScPDR5 and ScSNQ2 in the presence of 20 µM iKIX1 or vehicle (DMSO) control prior to and following ketoconazole (KET) addition.

(c) 20 µM iKIX1 inhibits ketoconazole-induced upregulation of ScPdr1 target genes ScPDR5 and ScSNQ2 in the HA-Pdr1 strain. RNA was harvested concurrently with representative chromatin immunoprecipitation experiment shown in panel (b) at t=0 min. (DMSO, 20 µM iKIX1) and t=15 min. after ketoconazole induction (DMSO + KET, 20 µM iKIX1 + KET). Transcripts are normalized to ScSCR1 and un-induced DMSO control.

(a–c) Representative experiment from two biological replicates is shown. Error bars represent mean +/− s.d. of technical replicates; *P<0.05, **P<0.01 and ***P<0.001 as calculated by two-tailed Student’s t-test.

(d) RNA-Seq analysis of a wild-type S. cerevisiae strain (BY4741) pre-treated with iKIX1 or vehicle alone then induced with ketoconazole (iKIX1 + KET and KET, respectively) demonstrate a blunted induction of Pdr1 target genes following iKIX1 pre-treatment. Data represents means of three biological replicates.

(e) iKIX1 pre-treatment does not significantly alter the transcript levels of PDR1 or GAL11/GAL11A in S. cerevisiae or C. glabrata after azole induction. Cells were pre-incubated with vehicle (DMSO) or iKIX1 and then induced with 40 µM ketoconazole (+KET) for 15 minutes before harvest. Average value of three biological replicates is shown and error bars represent mean +/− standard deviation; * P<0.05, ** P<0.001 as compared to DMSO or DMSO+KET control, calculated by two-tailed Student’s t-test.

Next, we determined the effect of iKIX1 on the transcription of C. glabrata Pdr1-regulated genes involved in drug efflux and MDR (CgCDR1, CgCDR2 and CgYOR1). CgPdr1 targets were strongly up-regulated after ketoconazole treatment[12,16]. However, pre-treatment with iKIX1 reduced target gene induction in a durable and concentration-dependent manner (Fig. 3 and Extended Data Fig. 4). Treatment with iKIX1 alone did not significantly affect Pdr1-target gene induction (Extended Data Fig. 4).

Extended Data Figure 4

With iKIX1 pre-treatment, CgPdr1-dependent transcription of (a) CgCDR1 and (b) CgYOR1 remains repressed 120 minutes after ketoconazole induction. SFY114 (PDR1 wild-type) cells were pre-incubated with vehicle (DMSO) or iKIX1 and then induced with 40 µM ketoconazole (+KET). Transcript levels were assessed by quantitative RT-PCR prior to and for 120 minutes following ketoconazole induction. Transcript levels are normalized to CgRDN25-1 and un-induced vehicle control (DMSO) at t=0.

(c) iKIX1 treatment alone does not have significant effects on CgPdr1 target gene induction either in the presence of wild-type (SFY114) or gain-of-function mutant CgPDR1 (amino acid alterations indicated).

(d) Table of average CgCDR1 delta Cp values (Cp – Cp) and corresponding standard deviation for quantitative real-time PCR experiments shown in Figure 3f and Extended Data Figure 4c.

(a–d) For all panels of Extended Data Figure 4, average value of three biological replicates is shown and error bars represent +/− standard deviation; * P<0.05, ** p <0.01, and *** p <0.005 as compared to vehicle + ketoconazole control. P values calculated using two-tailed Student’s t-test.

Next generation RNA sequencing (RNA-Seq) was employed to query the genome-wide effects of iKIX1 and azole treatments alone and in combination on the transcriptome in both S. cerevisiae and in C. glabrata. In accord with previous reports[16,17], azole treatment up-regulates Pdr1-dependent genes in both yeasts, such as the drug efflux pumps ScPDR5 and CgCDR1 (Supplementary Tables 1 and 2). Combined azole and iKIX1 treatment strongly blunted expression of many azole-activated and Pdr1-dependent genes in both S. cerevisiae and C. glabrata (Fig. 3, Extended Data Fig. 3 and Supplementary Tables 1 and 2), consistent with prior data and the proposed mechanism of action of iKIX1. iKIX1 alone affected very different sets of genes in S. cerevisiae and C. glabrata (Supplementary Tables 1–3). Treatment of S. cerevisiae and C. glabrata cells with iKIX1 did not significantly alter the expression of PDR1 or GAL11/MED15 after azole treatment (Extended Data Fig. 3). Together, these findings suggest that the primary mechanism of synergistic antifungal effects of iKIX1 with azoles is through blocking the azole-stimulated and Pdr1-dependent drug efflux pathway.

To ascertain iKIX1 efficacy in azole-resistant C. glabrata strains, we examined the effects of iKIX1 on CgPdr1 target gene expression in a set of isogenic strains with gain-of-function CgPDR1 mutations originally identified in azole-resistant C. glabrata clinical isolates[9]. iKIX1 reduced azole-induced transcription of CgPdr1 target genes (e.g., CgCDR1) in a concentration-dependent manner in all strains tested (Fig. 3 and Extended Data Fig. 4).

To investigate whether these transcriptional effects translated to functional effects on drug efflux rates, we utilized the fluorescent compound rhodamine 6G, a substrate of the CgCdr1 efflux pump[18,19]. Maximum efflux rates were significantly decreased in PDR1 wild-type or gain-of-function strains pre-treated with iKIX1, as compared to vehicle control (Fig. 3 and Extended Data Fig. 5).

Extended Data Figure 5

iKIX1 inhibits efflux of rhodamine 6G in PDR1 wild-type, PDR1 and PDR1 strains. Data points indicate mean of three biological replicates and error bars represent mean +/− s.d.

Due to its ability to reduce efflux pump gene expression and pump activity, we predicted that iKIX1 could restore azole-sensitivity to CgPDR1 gain-of-function mutant strains. Isogenic C. glabrata strains with wild-type or single gain-of-function alterations across CgPdr1 (Fig. 4) were tested for their sensitivity to fluconazole or ketoconazole on gradient plates with increasing concentrations of iKIX1 or vehicle. As expected, a CgPDR1 wild-type strain was sensitive to both fluconazole and ketoconazole, whereas CgPDR1 gain-of-function mutant strains grew robustly in the presence of azoles. iKIX1 restored azole-sensitivity to PDR1 gain-of-function mutant strains in a concentration-dependent manner (Fig. 4). CgPDR1 wild-type strains also exhibited increased growth inhibition in the presence of both iKIX1 and azole versus single agents alone (Extended Data Fig. 6).

Figure 4

iKIX1 as a co-therapeutic in models of C. glabrata disseminated disease

(a) Schematic showing CgPdr1 gain-of-function alterations in relation to putative functional domains. DBD: DNA-binding domain, ID: inhibitory domain, MHR: middle homology region, AD: activation domain.

(b) iKIX1 restores the efficacy of azoles towards CgPDR1 gain-of-function mutants. Plates contained increasing concentrations of vehicle control (DMSO) or iKIX1 to 150 µM in the absence or presence of fluconazole (FLU) or ketoconazole (KET).

(c) iKIX1 in combination with fluconazole but not fluconazole alone significantly extended survival of G. mellonella larvae injected with CgPDR1 (SFY115, n=9). For SFY114, n=10. * P<0.05, *** P<0.001 as compared to PBS vehicle control. Statistical differences measured using a log-rank (Mantel-Cox) test.

(d) iKIX1 combination treatment with 25 mg/kg fluconazole (low FLU) reduces fungal tissue burden in the kidney or spleen of mice injected with CgPDR1 wild-type (SFY114); iKIX1 in combination with 100 mg/kg fluconazole (high FLU) reduces fungal tissue burden in the kidney or spleen of mice injected with CgPDR1 (SFY115). N=5 mice for each treatment condition; * P<0.05, ** P< 0.005 and *** P< 0.0001 as compared to no treatment. Statistical differences measured using a Wilcoxon rank-sum test; error bars represent means +/− standard deviation.

Extended Data Figure 6

(a,b) iKIX1 increases the sensitivity of Cg strains bearing wild-type CgPDR1 to azole treatment. Two strains bearing wild-type CgPDR1 alleles (SFY114, DSY759) were plated at concentrations differing by ten-fold (10×, 1×) on plates containing increasing concentrations of (a) iKIX1 to 300 µM in the presence or absence of 1 µM ketoconazole (KETO) or (b) iKIX1 to 250 µM in the presence or absence of 50 µM fluconazole (FLU).

(c) iKIX1 and ketoconazole (KET) have additive effects on the growth of a CgPDR1 wild-type strain.

(d) iKIX1 and ketoconazole (KET) synergistically inhibit the growth of the CgPDR1 mutant.

(c,d) The EUCAST broth microdilution method[27] was used to assess the effects of iKIX1 and ketoconazole combination treatment. Growth, as assessed by OD540, was normalized to no drug control. All combination indices (CI) for the CgPDR1 mutant were less than 1, indicating synergy. A representative of three biological replicates is shown and the red line indicates a combination index of 1.

Based on the strong combination effect of azoles and iKIX1 in the CgPDR1 mutant we focused follow-up studies on this mutant strain. To investigate whether azoles and iKIX1 act in a synergistic or additive manner in CgPDR1 wild-type and CgPDR1 mutant strains, we assessed growth in checkerboard assays with ketoconazole and iKIX1. In the wild-type CgPDR1 strain, the combination of ketoconazole and iKIX1 was additive (Extended Data Fig. 6). However, the CgPDR1 mutant exhibited synergistic growth inhibition with iKIX1 and ketoconazole combination treatment, with combination indices <1 (Extended Data Fig. 6), in concordance with the spot-plating assay.

We carried out a limited analysis exploring the chemical space around the iKIX1 scaffold using commercial and custom synthesized iKIX1 analogs, identifying several compounds that lost activity in all assays; one analog (A2) is shown in Extended Data Figure 7. This example, together with data from iKIX1 analogs and the docked structure of iKIX1 to the CgGal11A KIX domain, supports a model where iKIX1 engages the core of the KIX domain using an array of hydrophobic and hydrogen bond contacts.

Extended Data Figure 7

Electron-withdrawing groups in the aromatic ring of iKIX1 complement the basic binding interface of the CgGal11A KIX domain and thus play a key role in iKIX1 function.

A structurally similar iKIX1 analog (A2) lacking electron-withdrawing groups increases the IC50 in the FP assay (a) and abolishes activity in the S. cerevisiae luciferase reporter assay (b), repression of CgCDR1 expression (c), and synergistic C. glabrata cell growth inhibition with azoles (d).

Error bars in (b,c) indicate mean +/− s.d. of technical replicates (reads/real-time PCR reactions, respectively). ** P<0.005; statistical differences calculated using two-tailed Student’s t-test.

(e) iKIX1 inhibits viability of HepG2 cells at concentrations >50 µM. The mean of 3 biological replicates is shown; error bars represent means +/− s.d.

(f) iKIX1 exhibits no effect on transcription of SREBP-target genes in HepG2 cells at concentrations up to 100 µM. Biological duplicates were assessed; representative experiment is shown and error bars represent means +/− s.d. of technical (real-time PCR) replicates.

(g) Mouse plasma stability of iKIX1 and mouse and human microsomal stability of iKIX1, n=1

(h) In vivo pharmacokinetic parameters of iKIX1, n=3 mice per time point.

We utilized two metazoan model systems to evaluate the potential utility of iKIX1 as a co-therapeutic with fluconazole to treat disseminated C. glabrata infection. The larvae of the moth Galleria mellonella has been used as a model to test the pathogenicity of a wide variety of human pathogens[20]. We utilized a G. mellonella survival assay to determine the virulence of C. glabrata PDR1 wild-type or PDR1 strains in the presence of fluconazole, iKIX1, or a combination of the two (Fig. 4). Larvae were injected with C. glabrata and a single injection of fluconazole (50 mg/kg), iKIX1 (25 mg/kg), a combination of the two, or vehicle; survival was monitored every 24 hours. G. mellonella injected with wild-type CgPDR1 was sensitive to fluconazole alone, and exhibited no significant alterations in survival with a fluconazole-iKIX1 combination. However, in G. mellonella larvae injected with a CgPDR1 strain, whereas the single agents fluconazole or iKIX1 did not significantly increase survival compared to vehicle, co-treatment with iKIX1 and fluconazole significantly increased survival (P<0.001).

Prior to mammalian studies, we sought to evaluate the potential toxicity of iKIX1 in mammalian cells (Extended Data Fig. 7). Human HepG2 cells treated with iKIX1 revealed toxicity only at high concentrations of iKIX1 (IC50 ~100 µM). iKIX1 had no effect on the transcription of SREBP-target genes at concentrations up to 100 µM, indicating its specificity for the fungal Gal11/Med15 KIX domain[4]. We also assessed the in vitro stability and in vivo mouse pharmacokinetics of iKIX1 and found that iKIX1 exhibited favorable drug-like properties and in vivo exposure in these studies (Extended Data Fig. 8).

Extended Data Figure 8

(a) Clinical isolates DSY562/DSY565 (azole sensitive and PDR1 azole-resistant strains, respectively) behave similarly to SFY114/SFY115 (isogenic PDR1+ and PDR1 strains, shown in Figure 4d) in the mouse infection model. n=10 mice for each treatment condition; * P<0.01, ** P<0.005 and *** P<0.0001.

(b) iKIX1 combination treatment with fluconazole reduces fungal tissue burdens in the spleen or kidney of mice injected with C. glabrata PDR1 (SFY116). n=5 mice for each treatment condition; ** P < 0.01 and * P < 0.05.

(c) 100 mg/kg/day iKIX1 (high iKIX1) treatment of mice infected with SFY93 (pdr1Δ) significantly reduces fungal burden in a mouse infection model (CFU/g kidney) alone as compared to SFY114 (PDR1) or SFY115 (PDR1). n=10 mice for each treatment condition; * P<0.01, ** P<0.005, *** P<0.0001.

(d) Mice infected with SFY114 (PDR1), SFY115 (PDR1) or SFY93 (pdr1Δ) were treated with low (10 mg/kg/day) iKIX1, low fluconazole (low FLU; 25 mg/kg/day), fluconazole at 100 mg/kg/day (FLU) or combination with the two. iKIX1 did not confer additional reductions in CFU/g kidney with SFY93 infection. n=10 mice for each treatment condition. *** P<0.0005.

(e) iKIX1 and ketoconazole (KETO) reduce adherence of CgPDR1 (SFY116) to CHO-Lec2 cells. Adherence is normalized to SFY114 DMSO control; each column represents the average of 4 biological replicates. * P<0.05 as compared to SFY114 DMSO control.

(f) iKIX1 (100 mg/kg/day) or fluconazole (FLU) significantly reduces fungal burden in the bladder and kidney in a urinary tract infection model in mice. n=15 mice were infected in each group and points at 0 log10 CFU/g organ fell below the detection limit of the method (50 CFU/g organ). * P< 0.05, ** P<0.005

(a–f) Statistical differences were measured using a Mann-Whitney/Wilcoxon rank-sum test as compared to no treatment control; error bars represent means +/− standard deviation.

To evaluate the therapeutic potential of iKIX1 and azole antifungal co-therapy in a mammalian model, we initially turned to an established mouse model of disseminated fungal disease[11]. Mice were inoculated with C. glabrata by tail-vein injection and were dosed peritoneally once-daily with 100 mg/kg fluconazole (high FLU), 100 mg/kg iKIX1, a combination of the two, or vehicle alone. After 7 days, mice injected with a CgPDR1 wild-type strain exhibited significantly reduced tissue fungal burden in the kidney and spleen following fluconazole treatment alone; iKIX1 co-treatment did not result in further reductions (Fig. 4). In contrast, in mice injected with the azole-resistant CgPDR1 strain, only co-treatment with iKIX1 and fluconazole resulted in significant (~10-fold) reductions in fungal burdens in the kidney and spleen (P<0.0001) (Fig. 4). Similar results were observed with the clinically isolated CgPDR1 and CgPDR1 strains DSY562 and DSY565 (Extended Data Fig. 8). Consistent with previous studies[9], the fungal burden in mice infected with the CgPDR1 strain was higher than those infected with wild-type CgPDR1 strains, suggesting that PDR1 mutant strains may be more virulent in vivo. Similar but less pronounced results were found in mice injected with a CgPDR1 strain (Extended Data Fig. 8). When mice were injected with a CgPDR1 wild-type strain and dosed with 25 mg/kg fluconazole (low FLU) alone or in combination with iKIX1, fluconazole alone poorly reduced tissue burden, whereas combination treatment resulted in significant (~10-fold) reductions in fungal burdens in both organs (P<0.0001) (Fig. 4). These results suggest that iKIX1 combination treatment with azole may be therapeutically desirable even in the absence of CgPDR1 gain-of-function mutations. Mice infected with a Cgpdr1 null strain were more sensitive to iKIX1 alone; unlike mice infected with CgPDR1 or CgPDR1 strains, low doses of iKIX1 did not further reduce fungal burden in Cgpdr1 null infections (Extended Data Fig. 8c,).

CgPDR1 gain-of-function mutations are also known to control adherence to host cells. As previously observed[21], a PDR1 mutant increased relative adherence to epithelial cells as compared to a PDR1 wild-type strain. Strikingly, iKIX1 treatment alone reduced adherence to levels similar to a PDR1 wild-type strain (Extended Data Fig. 8). Ketoconazole alone or co-treatment with iKIX1 also reduced relative adherence to levels comparable to a PDR1 wild-type strain. To assess the role of iKIX1 in modulating adhesion in an infection model, we turned to a mouse model of urinary tract infection[22]. In both the bladder and kidney, iKIX1 alone was sufficient to decrease fungal load after infection with either a PDR1 wild-type strain or a PDR1 strain (Extended Data Fig. 8), suggesting that iKIX1 may indeed modulate adhesion.

The proportion of azole-resistant C. glabrata (up to 20% in the US) and the emergence of multidrug resistance (approximately 40% of echninocandin-resistant isolates are azole-resistant) argues for the need for novel treatments that can target these resistant populations[23,24]. Our results demonstrate that small molecule disruption of the interaction between the CgGal11A KIX domain and the CgPdr1 activation domain is a therapeutically tractable method for resensitizing azole-resistant C. glabrata to standard azole antifungal treatment (Extended Data Fig. 9).

Extended Data Figure 9

Model of iKIX1 function as a co-therapeutic in combination with an azole, blocking the azole-induced recruitment of Gal11/Med15-Mediator to Pdr1 target genes upon azole-treatment and preventing the upregulation of Pdr1 target genes, including those which encode drug efflux pumps.