Identification of small molecule inhibitors of Pseudomonas aeruginosa exoenzyme S using a yeast phenotypic screen

Abstract

Pseudomonas aeruginosa is an opportunistic human pathogen that is a key factor in the mortality of cystic fibrosis patients, and infection represents an increased threat for human health worldwide. Because resistance of Pseudomonas aeruginosa to antibiotics is increasing, new inhibitors of pharmacologically validated targets of this bacterium are needed. Here we demonstrate that a cell-based yeast phenotypic assay, combined with a large-scale inhibitor screen, identified small molecule inhibitors that can suppress the toxicity caused by heterologous expression of selected Pseudomonas aeruginosa ORFs. We identified the first small molecule inhibitor of Exoenzyme S (ExoS), a toxin involved in Type III secretion. We show that this inhibitor, exosin, modulates ExoS ADP-ribosyltransferase activity in vitro, suggesting the inhibition is direct. Moreover, exosin and two of its analogues display a significant protective effect against Pseudomonas infection in vivo. Furthermore, because the assay was performed in yeast, we were able to demonstrate that several yeast homologues of the known human ExoS targets are likely ADP-ribosylated by the toxin. For example, using an in vitro enzymatic assay, we demonstrate that yeast Ras2p is directly modified by ExoS. Lastly, by surveying a collection of yeast deletion mutants, we identified Bmh1p, a yeast homologue of the human FAS, as an ExoS cofactor, revealing that portions of the bacterial toxin mode of action are conserved from yeast to human. Taken together, our integrated cell-based, chemical-genetic approach demonstrates that such screens can augment traditional drug screening approaches and facilitate the discovery of new compounds against a broad range of human pathogens.

Figure 1. Overview of the yeast based approach to find inhibitors against the human pathogenic bacteria P. aeruginosa and phenotypes of the bacterial ORFs causing synthetic lethality in yeast.

(A) S. cerevisiae W303-1A was utilized to identify P. aeruginosa PAO-1 virulence factors or essential ORFs that inhibit yeast growth. P. aeruginosa ORFs that inhibited yeast growth when individually overexpressed are prioritized based on biological relevance. Genes of interest are subsequently screened for inhibitors by overexpressing the bacterial ORFs and assaying for restoration of yeast growth in the presence of small molecules. Finally, in vitro and in vivo experiments demonstrate that the inhibitor directly modulates the bacterial protein biological activity. (B) Yeasts harboring the plasmid with the P. aeruginosa ORFs were grown overnight in liquid media. The cultures were then robotically diluted 10, 100 and 1000 times before their transfer in duplicate on solid media containing either glucose (Control 100% growth - left panel) or galactose + raffinose (Induction of bacterial gene expression - right panel). Cell growth was compared to the yeast with the empty vector (negative control – red circle) and to the yeast harboring the vector pRS316 encoding the toxic gene TUB2 (positive control – green circle). Phenotype of yeast with the plasmid pYES-DEST52 coding for a Pseudomonas toxic gene is marked by a blue square. The plasmid pYES-DEST52 was selected based on its strong promoter GAL1 combined to its high copy number 2 µ origin of replication for a maximal protein expression. (C) Nine P. aeruginosa ORFs inhibiting yeast growth when overexpressed. Yeast were transformed with the pYES-DEST52 yeast expression vector encoding the nine bacterial ORFs, grown overnight and individually spotted in duplicate as a 10 fold serial dilution on plates containing either glucose (left panel) or galactose + raffinose (right panel).

Figure 2. Prevention of S. cerevisiae growth by ExoA-ADPRT, ExoY-adenylate cyclase and ExoS-ADPRT activities.

(A) Functional domains of the Pseudomonas ExoA, ExoY and ExoS and localizations of the point mutations abolishing the different enzymatic activities. (B) S. cerevisiae W303-1a was transformed with yeast expression vector alone (Empty vector), the yeast expression vector encoding ExoA wild type (ExoA wt) or ExoA E553A ADPRT mutant (ExoA-ADPRT mutant). Similarly, yeast was transformed with a vector containing ExoY wild-type (ExoY-wt), ExoY K81M AC mutant (ExoY-AC mutant). Finally, identical vector with ExoS wild type (ExoS-wt), ExoS R146W GAP mutant (ExoS-GAP mutant), ExoS E379A+E381A ADPRT mutant (ExoS-ADPRT mutant) or ExoS GAP and ADPRT double mutant (ExoS-GAP and ADPRT mutant) was incorporated in yeast. Toxicity of the different constructs in yeast were determined by spotting serial dilution of overnight cultures onto agar containing glucose (Control 100% growth - left panel) or galactose + raffinose (right panel).

Figure 3. Ras is a direct target of ExoS both in yeast and human.

(A) In mammalian cells, ExoS inactivates several targets. A similar mechanism could explain the observed growth defect in S. cerevisiae. In that case, overexpression of the yeast homologues of the ExoS human targets should restore yeast growth by titrating the toxin enzymatic activity. (B) In the presence of ExoS, yeast Ras2p overexpression reverts yeast growth to a level comparable to the one with the yeast harboring the empty vector. Overnight cultures of the yeast transformed with empty vector, the vector encoding ExoS alone or exoenzyme S with the yeast ORFs were adjusted to the same cell density (10⁶ cells/ml). The rescuing effect due to the yeast ORFs expression was estimated by spotting a 10 fold serial dilution of the yeast cultures on agar containing glucose - copper (upper panels) or galactose + raffinose + copper (lower panels). (C) Bmh1p acts as ExoS cofactor in yeast. The ExoS expression vector was transformed in wild-type, bmh1Δ and bmh2Δ yeast backgrounds. In absence of BMH1, but not BMH2, ExoS did not display any toxicity in yeast (left panel). Additionally, the vectors encoding ExoS and Bmh1p were cotransformed. Bmh1p overexpression restored ExoS toxicity in a yeast bmh1Δ background (right panel). (D) ExoS ADP-ribosylates Ras2p in vitro. Ras2p was incubated with ExoS and [³²P]NAD in presence or absence of the yeast activator protein Bmh1. The samples were separated by SDS-PAGE and incorporation of radioactive ADP-ribose analysed by phosphorimaging. The upper bands are caused by ExoS auto-ribosylation and served as a positive control. Due to the nature of the ExoS purification, the top band corresponds to the full-length ExoS and the lower band represents the auto-ribosylation of a truncated form of ExoS.

Figure 4. Exosin inhibits ExoS ADPRT activity.

(A) List of ExoS potential inhibitors isolated during the yeast chemical screen. In yeast, the growth inhibition caused by ExoS expression was used to screen for novel inhibitors of this bacterial protein by selecting those compounds that can restore growth to the yeast expressing the toxin. Structures of the six identified hits are displayed. Diosmin, Everninic acid, 4296-1011 and E216-5303 directly modulate ExoS ADPRT enzymatic activity. IC₅₀ for each molecule is defined by the compound concentration required to decrease ExoS ADPRT activity by 50%. ExoS, its cofactor FAS and human Ras were purified and used in the fluorescence-based ADPRT assay. The inhibitor IC₅₀ was determined by non-linear regression curve fitting. Flavokawain B and 0469-0706 possessed intrinsic fluorescence that interfered with the fluorescent ADPRT enzymatic assay, therefore the IC₅₀ for these compounds could not be determined. (B) Dose-response curve for E216-5303 on the ADPRT activity of ExoS. Various aliquots of a stock solution of E216-5303 prepared in DMSO were pre-incubated with 20 µM human Ras, 1 µM of FAS and 20 µM of ε-NAD⁺ in 100 mM NaCl, 2 mM MgCl₂, 200 mM sodium acetate, pH 6.0. The reaction was initiated with the addition of 50 nM ExoS and the transferase reaction was monitored by recording the time-dependent change in fluorescence intensity. The fluorescence excitation was at 305 nm with fluorescence emission at 405 nm. The inhibitor IC₅₀ was determined by non-linear regression curve fitting. (C) Lineweaver-Burk plot of the inhibition of ExoS ADPRT activity at 0(○), 20(▪), 30(▴) and 40(Π) µM of E216-5303. (D) Plot of the slope from (c) (K_M/V_max) against E216-5303 concentration (see Material and Methods for details). (E) Percentage of yeast growth induced by exosin analogues. Growth of yeast expressing ExoS was calculated for each of the 50 exosin analogues and compared to the cell expressing ExoS in the presence of exosin (100% growth control) and yeast with ExoS alone (no inhibitor – background growth control). (F) Structures, percentage of yeast growth recovery and IC₅₀ for the small molecule inhibitor exosin and its analogues. Yeast growth recovery was calculated as the difference of growth of yeast expressing ExoS in the presence of the inhibitor compare to yeast harboring the empty vector (100% growth control) and yeast with ExoS alone (no inhibitor – background growth control). Exosin and analogues directly modulate ExoS ADPRT enzymatic activity. The IC₅₀ values for exosin and analogues were determined as previously described. (G) Structure of the core molecule for exosin and analogues. An arrow indicates the para position of the benzyl ring, a position important for activity.

Figure 5. Inhibitors reduce P. aeruginosa cytotoxic effect during CHO cell infection.

CHO cells were seeded at a concentration of 1.25×10⁵ cells/well 12 hours prior P. aeruginosa PAK infection (MOI 10). (A) During the early stage of infection (2 hours), CHO cells were harvested by trypsination and resuspended in HBSS + 1% BSA. Induced cell death was measured by flow cytometry after 7-AAD (10 µg/ml) staining, using a Beckman-Coulter EPICS Elite flow cytometer. Dot plots shows the percentage of dead and living cells for different compound concentrations. The lower right quadrant shows living CHO cells that are positive in size (FSC-H) and negative by 7-AAD staining. In the upper left quadrant, the 7-AAD positive location indicates the number of dead or dying CHO cells after P.aeruginosa infection. These results are representative of 3 independent experiments. (B) The bar graph shows the percentage of dead/dying cells measured by flow cytometry. The error bars represent the SD with n = 3. The protective effect was compared to the number of dying/dead CHO cells in absence of both bacteria and compound as the 0% of cell lysis (background control) and to the number of dying/dead CHO cells in the presence of P. aeruginosa but in absence of any compound as the 100% lysis. (C) During the late stage of infection (4 hours), CHO cell supernatants were submitted to LDH release quantification using the Cytotoxicity Detection Kit (Roche). The error bars represent the SD with n = 3. The percentage of LDH release for each compound concentrations was compared to the LDH release of CHO cells in absence of both bacteria and compound (spontaneous lysis) and to the LDH release of CHO cells in the presence of P. aeruginosa PAK in absence of inhibitor (100% lysis).