Identification of P. aeruginosa Drug Targets Modulating S. cerevisiae Growth
We developed a yeast-based strategy where S. cerevisiae
was initially used to identify P. aeruginosa
PAO1 virulence factors or essential ORFs that inhibit yeast growth (). These particular genes were selected because they provide an attractive starting point to develop antibacterial drugs. Accordingly, we developed a list of 505 potential drug targets of P. aeruginosa
. These bacterial ORFs were individually transferred into the yeast expression vector, pYES-DEST52 where the GAL1
promoter controlled their expression. Transformed yeast growing on 2% glucose served as control (i.e., wild type growth) because in these conditions, the expression of the exogenous bacterial genes is repressed. Expression of these genes was induced by growing the yeast on selective solid medium containing 2% galactose + 2% raffinose (). The experiment was repeated (four times), and involved inoculating yeast cultures at different dilutions and spotting variable volumes of culture on agar plates in an attempt to increase the consistency of this test.
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.
Of the 505 P. aeruginosa ORFs screened, nine strongly or partially impaired the yeast growth when overexpressed (). Five of these are essential genes, including; 1) the ribonuclease III – PA0770, 2) two probable transcription regulators - PA0906 and PA1520, 3) the transcription termination factor Rho – PA5239 and 4) a hypothetical protein – PA2702. In addition, four virulence genes were also detrimental to yeast growth, ExoA – PA1148, ExoS – PA3841, ExoT – PA0044, ExoY – PA2191. Interestingly, each of these four toxins are secreted or translocated by the type II (ExoA) or type III secretion system (ExoS, ExoT, ExoY) and each act within the infected host cell. By comparing the phenotype of yeast harboring the empty vector, we could assess the relative strength of the Pseudomonas gene overexpression effect and classify them into three groups ( – right panel). Firstly, ExoA, ExoY, PA1520 and PA2702 strongly inhibited S. cerevisiae growth. Secondly, ExoS, ExoT, PA0906 and transcription termination factor Rho showed an intermediate growth impairment whereas expression of ribonuclease III weakly affected yeast fitness.
Yeast Growth Inhibition Is Mediated by Exoa and Exos ADPRT, and Exoy Adenylate Cyclase Activities
To assess the influence of ExoA, ExoY and ExoS enzymatic activities on yeast growth, catalytic mutants were assayed. Residues important for the enzymatic activity of ExoA (E553A), ExoY (K81M) and ExoS (R146W and E379A+E381A) were previously reported 
and served to guide our mutant construction ().
Prevention of S. cerevisiae growth by ExoA-ADPRT, ExoY-adenylate cyclase and ExoS-ADPRT activities.
Compared to the empty vector control, overexpression of active ExoA-wt and ExoY-wt induced a severe growth defect ( – top and middle panels) whereas expression of the enzymatically inactive ExoA-ADPRT mutant and ExoY-AC mutant did not. This observation suggests that ExoA and ExoY toxicity is conferred by their ADPRT and AC activities, respectively. Moreover, whereas ExoS-wt expression reduced yeast growth, this dominant negative effect was totally abolished when expression of the ExoS-GAP and ExoS-ADPRT mutants were simultaneously induced, indicating one or both ExoS enzymatic activities are causative for the yeast growth defect ( – bottom panel). Because normal growth was observed only for the ExoS ADPRT domain mutant and not for the GAP mutant, this suggests that ExoS ADPRT enzymatic activity is responsible for the yeast toxicity consistent with previous observations 
. Taken together, these observations attribute the yeast growth inhibition to the ExoA-ADPRT, ExoY-AC and ExoS-ADPRT activities and validate the three toxins as appropriate drug target candidates for further study. Due to its critical role in the initial steps of chronic infections of immuno-compromised patients and in the pathogenesis of acute P. aeruginosa
infections, ExoS was selected for interrogation using our yeast-cell based inhibitor screen.
Exoenzyme S ADP-Ribosylates Identical Targets in Both Human and Yeast
To demonstrate that yeast can serve as a model system to mimic human cells during infection, we asked if these bacterial toxins modulate the biological activity of conserved eukaryotic targets. Following binding of P. aeruginosa
to human cells, the bacteria inject ExoS directly into the cytoplasm where it inhibits the activity of several targets by ADP-ribosylation. Therefore, overexpressing yeast homologues of ExoS human targets should restore yeast growth by titrating the toxin's enzymatic activity (). To test our hypothesis, forty-six yeast members of the Ras superfamily and cyclophilins were individually overexpressed in yeast in the presence of ExoS (Table S2
Ras is a direct target of ExoS both in yeast and human.
We first verified that individually, the overexpressed yeast proteins did not impair yeast growth. To accomplish this, cells were cultivated on galactose + raffinose in absence of copper such that only the yeast over-expressed candidates, but not ExoS, were expressed ( – top right panel). Yeast genes whose overexpression was toxic were eliminated from our analysis. In parallel, cells were grown on galactose + raffinose in presence of copper to assess the rescuing effect of yeast gene overexpression in the presence of the toxic ExoS ( – bottom right panel). Comparing yeast growth to the cell harboring the empty vector and yeast expressing ExoS alone ( – bottom left panel), ten yeast genes were found to rescue ExoS toxicity (Table S3
). Subsequently, only yeast genes demonstrating a strong growth rescue phenotype (such as RAS2
) were analyzed further whereas genes showing weak rescue (such as YPT1
) were not studied further (). S. cerevisiae
possesses two homologues of the human Ras protein (Ras1p and Ras2p). Interestingly, Ras2p was found among these ten ORFs, i.e. overexpression of Ras2p but not Ras1p rescued ExoS-induced toxicity ().
As previously described, ExoS requires Factor Activating Exoenzyme S (FAS) for its ADPRT activity 
. FAS is a member of the 14-3-3 protein family which has two yeast homologues, the Brain Modulosignalin Homolog (Bmh) 1 and 2. Accordingly, ExoS toxicity was assessed in the absence of Bmh1p or Bmh2p. As detected by the increase in yeast growth, ExoS-induced toxicity was diminished in cells lacking Bmh1p but not in those lacking Bmh2p ( – left panel). In a bmh1
Δ yeast background, the toxic effect of ExoS was again restored when introducing BMH1
in the presence of the toxin ( – right panel). Together, these data imply that Bmh1p acts as ExoS cofactor in yeast.
To better understand the mechanism of this toxicity, we demonstrated that yeast Ras2p was a direct target of ExoS and that Bmh1p was the ExoS cofactor in yeast, using a biochemical assay. To that end, an ADP-ribosyltransferase enzymatic assay was performed using the radioactive substrate [32P]-NAD+, purified P. aeruginosa ExoS, yeast Ras2p and Bmh1p. Autoradiographic analysis showed that radioactive ADP-ribose was incorporated by Ras2p (). Moreover, in absence of Bmh1p, no ADP-ribosylation was observed. These data reveal that in vitro, Ras2p is directly ADP-ribosylated by ExoS with Bmh1p as a cofactor.
Taken together, these results allow us to conclude the following; (i) in yeast, the growth inhibitory effect observed in the presence of the P. aeruginosa ExoS is mediated by its ADPRT activity, (ii) this growth inhibition is due, at least in part, to the inactivation of the yeast protein Ras2p by ADP-ribosylation, (iii) ExoS ADPRT activity is activated by the yeast cofactor Bmh1p. Most significantly, conservation of several toxin targets from yeast to human, such as Ras2p, Rsr1p, Ypt52p and Cpr6p, suggests that P. aeruginosa ExoS acts in a related manner in both organisms.
E216-5303 Modulates Exoenzyme S ADPRT Activity through Competitive Inhibition
The sensitivity and specificity of our yeast-based assay allowed us to use S. cerevisiae to detect potential inhibitors of the three selected P. aeruginosa drug targets. Because ExoS-wt inhibited yeast growth when overexpressed, we reasoned that any molecule that inhibits this enzyme should restore yeast growth (). Because we were unable to find any inhibitors when the bacterial toxin was expressed using the strong promoter GAL1, we exchanged the GAL1 promoter with the copper inducible promoter CUP1 which allows a titrable expression of the toxins. Expression from this promoter decreases the toxin level in yeast and renders the conditions of the yeast screen less stringent. Over 56,000 compounds, primarily synthetic small molecules, were tested against ExoS. Effect of the compounds was compared to the yeast growth in absence of compound (as control for inhibition) and to the cells dividing in absence of toxin (as a control for growth). With this strategy, we uncovered six potential inhibitors, Diosmin, Everninic acid, Flavokawain B, 0469-0796, 4296-1011 and E216-5303 based on their ability to restore yeast growth ().
Exosin inhibits ExoS ADPRT activity.
To determine if the observed yeast growth recovery was due to a direct modulation of the compounds on ExoS ADPRT activity, an in vitro fluorescent ADPRT enzymatic assay was performed. Diosmin, 4296-1011, Everninic acid and E216-5303 modulated ExoS ADPRT activity and their IC50 values were determined as 3, 6, 21 and 23 µM respectively (). Due to their intrinsic fluorescence, Flavokawain B and 0469-0706 effects could not be tested in our enzymatic assay. Because, only exosin protected CHO cells from lysis during P. aeruginosa infection in cell culture (data not shown) it was therefore selected for additional studies. Exosin acts as competitive inhibitor against the NAD+ substrate of ExoS as the Vmax values were largely unaffected, whereas the KM values increased from 9 to 30 µM (). The Ki value (dissociation constant for a competitive inhibitor) was 33.0 ± 3.0 µM for exosin (), which agrees favourably with the IC50 value for this compound (). Thus, the drug-like compound exosin directly modulates ExoS ADPRT activity in vitro via competitive inhibition. Therefore, exosin seems to restore ExoS dependant yeast growth defect by directly inhibiting ExoS ADPRT activity.
Exosin Protects Exoenzyme S Induced Cytotoxicity in CHO Cells
To determine if the small molecule inhibitor, exosin, could modulate the viability of CHO cells during P. aeruginosa infection, apoptotic CHO cells and living CHO cells were distinguished using the exclusion dye 7-AAD. Here, CHO cells were exposed to P. aeruginosa with or without the small molecule inhibitor for 2 hours, and the fraction of apoptotic cells was measured by 7-AAD staining and flow cytometry.
The mean fluorescent intensities of 7–AAD were plotted as a histogram (Figure S1
). When compared to the mean fluorescent intensity from the red peak (background fluorescent intensity - 7.42; n
3) and from the blue peak (control for P. aeruginosa
infection - 18.2; n
3), the green (20 µM), orange (40 µM), and light blue (80 µM) peaks gave mean fluorescent intensities of 13.4, 11.3, and 10.3, respectively, indicating that exosin exerted its effect in a dose-dependent manner. Therefore, a higher inhibitor dose reduced the number of cells undergoing apoptosis, reflecting a better protective effect. Similar observations were made in dot plots (). In the presence of 80 µM exosin, a significant increase in the percentage of living cells (79.35%; n
3) was observed with the serious reduction of dead cells (20.31%; n
3), compared to the infected CHO cells without inhibitor, 0 µM (49.72% and 50.28%, respectively). However, the protective effect of exosin at a concentration of 80 µM was not observed when CHO cells were infected by the P. aeruginosa
PA14 strain, a strain expressing ExoT, ExoY and the phospholipase exoenzyme U (ExoU) but not ExoS, indicating the specificity of the compound exosin against ExoS only ().
Inhibitors reduce P. aeruginosa cytotoxic effect during CHO cell infection.
In the CHO cell infection assay, the protective effect of exosin was monitored during an early stage of infection by detecting the number of dying and dead CHO cells using flow cytometry. Moreover, the effect of the inhibitor at the late stage of infection was assessed by the quantification of lactate dehydrogenase (LDH) released from the population of lysed CHO cells. Four hours after P. aeruginosa PAK infection () revealed a 6.93% decrease in lysis upon addition of 20 µM exosin, a 13.92% lysis reduction in the presence of 40 µM final of inhibitor and a 12.90% reduction at 80 µM. However, the protective effect of exosin at a concentration of 80 µM was not observed when CHO cells were infected by the P. aeruginosa PA14 strain that translocates ExoU instead of ExoS (). Together, these data strongly support the conclusion that the inhibitor exosin is specific for ExoS and is able to reduce ExoS cytotoxicity against mammalian cells.
PAK viability was tested by measuring optical density of cultures in the presence of 20, 40 and 80 µM of inhibitor over a period of 10 hours. Addition of exosin did not affect Pseudomonas
growth, further confirming the specificity of exosin for ExoS in CHO cells (Figure S2
Pre-Selection of Exosin Analogues in S. cerevisiae
Given the specificity of exosin, we screened 50 structural analogues of this compound in yeast to find molecules with increased potency against ExoS ADPRT activity. Seven analogues with an improved effect were found ( – exosin-5138, exosin-5316 and the compounds marked by an asterisk). According to the flow cytometry results, all of these compounds protected CHO cells when infected with P. aeruginosa in cell culture (data not shown). However, only exosin-5138 and exosin-5316 showed a protective effect when monitored with the LDH assay. Therefore, only these two compounds were used for further investigation. Exosin-5340 had no protective effect in yeast or in the CHO cell infection assay and served as a negative control. Importantly, results obtained from the yeast studies revealed the importance of the para position of the nitrobenzyl ring for the inhibitory activity of the compounds (). The three different analogues, exosin-5138, exosin-5316 and exosin-5340, were then selected for quantification of the yeast growth recovery and for IC50 determination. Exosin-5138 showed 36.8% recovery, almost double the protective effect of exosin whereas exosin-5340 conferred no protection (). The last analogue exosin-5316 (with 27.4% recovery) demonstrated a protective effect almost equal to the original compound (20.5% recovery). The fluorescent ADPRT enzymatic assay revealed that the three analogues directly modulate ExoS ADPRT activity in vitro (). The IC50 for each compound was calculated and these values paralleled the effect of the small molecules in yeast, most strongly for exosin-5138 and exosin-5340, and to a lesser extent for exosin-5316.
We extended our studies of these analogs in mammalian cells. For this purpose, protection provided by exosin-5138 and exosin-5316 was assessed in the CHO cell toxicity assay. Using the flow cytometry as described earlier, a strong protective effect of exosin-5138 was observed (). Dot plots of exosin-5138 showed a large reduction in dead cells at a compound concentration of 80 µM (). Exosin-5138 showed decreases of 25.20, 44.05 and 60.83% in the number of dead/dying CHO cells in the presence of 20, 40 and 80 µM of inhibitor, respectively (). In the LDH assay, exosin-5138 reduced cell lysis by 9.34, 18.61 and 24.64% in presence of the compound at 20, 40 and 80 µM inhibitor, respectively () demonstrating an improved efficacy of exosin-5138 against ExoS cytotoxicity versus the original hit.
In addition, as shown by flow cytometry, exosin-5316 exerted a protective effect (). The number of dead/dying CHO cells was detectably lower upon addition of the analogue exosin-5316; however, this reduction was not statistically significant compared to the original hit (p>0.05). In contrast, the LDH assay revealed a 7.94, 12.84 and 15.81% reduction in cell lysis at 20, 40 and 80 µM final concentration respectively (p<0.05). The analogue exosin-5316 showed similar protective effect compared the original hit (p<0.05) ().
The data show a correlation between the protective effect of exosin and its analogues in yeast and for the results obtained in the CHO cell infection assay. Moreover, these observations establish yeast as a powerful assay system to estimate the effect of analogues of an original hit and to prioritize lead compounds before tedious subsequent experiments in a more complicated model of infection are undertaken.