Discovery of new diketopiperazines inhibiting Burkholderia cenocepacia quorum sensing in vitro and in vivo

Burkholderia cenocepacia, an opportunistic respiratory pathogen particularly relevant for cystic fibrosis patients, is difficult to eradicate due to its high level of resistance to most clinically relevant antimicrobials. Consequently, the discovery of new antimicrobials as well as molecules capable of inhibiting its virulence is mandatory. In this regard quorum sensing (QS) represents a good target for anti-virulence therapies, as it has been linked to biofilm formation and is important for the production of several virulence factors, including proteases and siderophores. Here, we report the discovery of new diketopiperazine inhibitors of the B. cenocepacia acyl homoserine lactone synthase CepI, and report their anti-virulence properties. Out of ten different compounds assayed against recombinant CepI, four were effective inhibitors, with IC50 values in the micromolar range. The best compounds interfered with protease and siderophore production, as well as with biofilm formation, and showed good in vivo activity in a Caenorhabditis elegans infection model. These molecules were also tested in human cells and showed very low toxicity. Therefore, they could be considered for in vivo combined treatments with established or novel antimicrobials, to improve the current therapeutic strategies against B. cenocepacia.


Supplementary
. Molecular docking analysis of compound 8b onto CepI structure. Cartoon representation of the CepI homology model (shown in grey, orientation as in Supplementary Figure S4) showing the results of molecular docking using compound 8b. Optimized candidate ligand binding sites identified by molecular docking are shown with colored dots. Candidate sites proximate to structural elements critical for substrate recognition are shown in magenta, others in cyan. Figure S6. Microphotographs of HeLa cells after 72h treatment with 8b and 6a. DMSO, cells treated with the highest amount of solvent. 10X and 20X indicate the magnification used.

Structural bioinformatics
Secondary structure prediction on CepI was performed through comparison of the results of multiple web servers including HHPRED S4 and PredictProtein S9 . The comparative evaluation of multiple structural prediction algorithms enabled unambiguous identification of the most prominent features of the proteins analyzed. The three-dimensional model of CepI was created using the homology modelling tools in HHPRED S4 and MODELLER S10 , and the alignment shown in Supplementary Fig. S1. The model was further optimized by geometry idealization using PHENIX S11 . Final model quality was assessed using PROCHECK and PDBSUM S12 and the Qmean server S13 . Structural figures were generated using ESPript S14 and PyMol S15 .

Molecular Docking
Molecular docking of compound 8b was performed using the CepI homology model obtained as described above and the three-dimensional structure of the inhibitor generated by the PRODRG server S16 . Unconstrained ligand exploration trials for identification of candidate ligand binding sites were performed by repeated searches using the SWISSDOCK S17 and PELE S18 docking servers. Out of more than 100 parallel independent attempts of unconstrained ligand binding search, the best candidate sites were selected based on the lowest stabilization energy and used as templates for ligand binding refinement using PELE S19 . Each candidate binding contact was subject to 24 parallel runs of refinement (using 12 hours of minimization for each run). Final selection used to generate Supplementary Figure S4  Step a). Suspension of 2',5'-Dihydroxyacetophenone (3.0 g, 19,7 mmol), benzyl chloride (4.5 ml, 39.4 mmol) and potassium carbonate (8.2 g, 59.1 mmol) in 75 ml of abs. ethanol was refluxed 4 hours, filtered off and mother liquid was evaporated under vacuum, residue treated by cold water and white solid was filtered off. Crude 2',5'-dibenzyloxyacetophenone was recrystallized from ethanol. The yield is 5.05 g (77%), mp. 77-78 o C (lit. 76-77 o C) S20 .
Step b). Suspension of 2',5'-dibenzyloxyacetophenone (3.0 g, 9.0 mmol) and selenium dioxide ( Step c). Suspension of [2,5-bis(benzyloxy)phenyl](oxo)acetic acid (2.5 g, 6.9 mmol) in 15 ml of carbon tetrachloride was treated by 2.5 ml of thionyl chloride and one drop of DMF. Reaction mixture was refluxed for 3 hours become clear solution and evaporated under vacuum. Oily residue was treated by toluene and evaporated under vacuum again. Colorless oil of [2,5bis(benzyloxy)phenyl](oxo)acetyl chloride (quantitative yield) was immediately used in next step without any additional purification.

MIC determination
The MICs (Minimal Inhibitory Concentrations) of the compounds reported in Supplementary Table   S1 were assessed against B. cenocepacia J2315 by using the 2-fold microdilution method in Ubottom 96-well microtiter plates.
About 10 5 CFU were used to inoculate each well of the microplate containing concentrations of compounds ranging from 1 to 128 µg/ml. Growth was determined by the resazurin method S22 after two days of incubation at 37 °C. 30 μl of a solution of resazurin sodium salt (Sigma Aldrich) at 0.01% in distilled water were added to each well, and the microtiters were incubated at 37 °C for about 4 h. The MIC was defined as the lowest concentration of the compound that prevented a change in color from blue to pink.