Targeted small molecule inhibitors blocking the cytolytic effects of pneumolysin and homologous toxins

Abstract

Pneumolysin (PLY) is a cholesterol-dependent cytolysin (CDC) from Streptococcus pneumoniae, the main cause for bacterial pneumonia. Liberation of PLY during infection leads to compromised immune system and cytolytic cell death. Here, we report discovery, development, and validation of targeted small molecule inhibitors of PLY (pore-blockers, PB). PB-1 is a virtual screening hit inhibiting PLY-mediated hemolysis. Structural optimization provides PB-2 with improved efficacy. Cryo-electron tomography reveals that PB-2 blocks PLY-binding to cholesterol-containing membranes and subsequent pore formation. Scaffold-hopping delivers PB-3 with superior chemical stability and solubility. PB-3, formed in a protein-templated reaction, binds to Cys428 adjacent to the cholesterol recognition domain of PLY with a K_D of 256 nM and a residence time of 2000 s. It acts as anti-virulence factor preventing human lung epithelial cells from PLY-mediated cytolysis and cell death during infection with Streptococcus pneumoniae and is active against the homologous Cys-containing CDC perfringolysin (PFO) as well.

Fig. 1. Structure-based virtual screening for the identification of PLY inhibitors.

a A structural model of a PLY dimer (blue and yellow are two neighboring monomers) was built by fitting crystallographic data (PDB entry: 4ZGH) into a cryo-EM map (PDB entry: 2BK2). b Identified potential binding sites shown as surface in red in the oligomerization and in the cholesterol binding domain (red circles). c Workflow of virtual screening resulting in 10 diverse hit molecules for testing in a hemolysis assay. d Molecular structure of PB-1.

Fig. 2. The evaluation of activity and binding kinetics of PB-3.

a IC₅₀ values of PB-1, PB-2, and PB-3 (distinct samples n = 3, 95% CI for IC₅₀ and error bars represent ± S.D.) corresponding to the ability of PB-inhibitors to block the PLY-induced hemoglobin release in sheep erythrocytes. b PLY was inhibited by PB-3, formed via protein-templated fragment ligation in the hemolysis assay, while PB-3.1 and PB-2.3 alone were no inhibitors (distinct samples n = 3, error bars represent ± S.D.). c Quantification of PB-3 formed by protein-templated ligation using HPLC-QTOF-MS. Concentrations were fitted to the one-phase saturation equation c(t) = c₀ + (c_max - c₀)*(1-exp(-K*t)) used to obtain c_max (795 nM) and t_1/2 (~20.5 min) (distinct samples n = 3, error bars represent ± S.D.). d, e PB-3 prevents human alveolar epithelial cells from PLY-associated impairment. Samples include controls (positive and negative), PLY alone, and PLY with PB-3 (1, 3 and 6 µM). Figures show LDH release (which is quantified as percent-cytotoxicity) after 4 h and 24 h. Mann Whitney test (two-tailed) is used for statistical evaluation (distinct samples n = 8, ***p = 0.0009 for d distinct samples n = 8 ***p = 0.0002 for (e), and error bars represent ± S.D.). f Binding kinetics of PB-3 and PLY complex measured in the BLI assay. PLY-C428A is used as reference protein and buffer is reserved as inhibitor reference. Binding is observed over 0.6, 1, and 6 µM of PB-3 (data shown are representing one of n = 3 distinct samples). The reference-subtracted data are presented in the graph, which display association and dissociation steps of wild-type PLY and PB-3. g The one-to-one kinetic fitting model was used to calculate the on- and off-rates (k_on, k_off) of binding and the binding affinity (K_D) of PB-3 to PLY.

Fig. 3. Investigation of PLY and PB-2 using Cryo-TEM.

a Cryo-electron tomography of liposomes treated with PLY: The right image shows a section of a central stack of 20 accumulated slices from the reconstructed 3D volume (left, as voltex presentation), one example of PLY-induced pre-pore and pore formation event is highlighted by a red circle. For a video representation see Supplementary Video 1. b Cryo-electron tomography of PLY-pores: Top: Reconstructed 3D volume of solubilized liposomes. Bottom: Central accumulated stack of 20 slices from the 3D volume showing protein pores formed by oligomerization of PLY monomers isolated by treatment of PLY-liposome mixture with Cymal-6 followed by Amphipol-35. For a video representation see Supplementary Video 2. c Cryo-electron tomography of liposomes in the presence of PLY and inhibitor (PB-2): A central stack of 20 accumulated slices (position corresponds to the blue frame in the 3D volume on the left) shows smooth vesicle membranes (top image and enlarged detail). A near-surface stack of 20 accumulated slices (position corresponds to the green frame in the same 3D on the left) shows suspended PLY particles not integrated in liposomal membranes (highlighted by red arrows in the bottom image and enlarged detail). For a video representation see Supplementary Video 3 (data shown in (a–c) are one of n = 3 and scale bars correspond to 100 nm).

Fig. 4. Investigation of the biological activity and binding interactions of PB-inhibitors against PLY-mutants and evaluation of PB-inhibitors against alternate CDC.

a IC₅₀ values of PB-3 against PLY-S61A, PLY-N85L, and PLY-T55A, which are mutants involving the potential binding site at the oligomerization interface (distinct samples n = 3, 95% CI for IC₅₀ and error bars represent ± S.D.). b Comparison of PB-3 activity against wild-type PLY and PLY-C428A (a UDP-cysteine mutant version of PLY). PB-3 blocks only wild-type PLY and is unable to neutralize the mutant PLY-C428A at all three concentrations (distinct samples n = 3, error bars correspond to ± S.D.). c Bio-layer interferometry assay: wild-type PLY, PLY-C428A-immobilized, and control NiNTA biosensors exhibit association with PB-3 at three concentrations (2, 4, and 8 µM). Only PLY (wild-type) shows significant association with PB-3 in contrast to PLY-C428A (data presented are one of n = 3 distinct samples). d IC₅₀ values of PB-1, 2, and 3 against perfringolysin (PFO, a toxin homologous to PLY) and PB-inhbitors show 10-fold increased potency toward PFO than PLY (distinct samples n = 3, 95% CI for IC₅₀ and error bars represent ± S.D.). e Intermedilysin (ILY, a cysteine-free CDC toxin) is not blocked by PB-inhibitors (distinct samples n = 3 for each experiment, 95% CI for IC₅₀ and error bars represent ± S.D.) f Covalent binding mode of PB-3 suggested docking, for details see text.

Fig. 5. Inhibition of PLY in native and S.pn.-infected human alveolar epithelial cells.

PLY neutralized by PB-3 in human alveolar epithelial cells (A549 cells). Cells were grown in Ham’s F12 culture medium with inclusion of a red mitochondrial dye (TMRE: tetramethyl-rhodamine, ethyl ester) to indicate healthy cells and a fluorogenic peptide substrate (DEVD) with green fluorescence to highlight activity of caspase 3 and 7 activity in apoptotic cells (data here represent one of n = 3, distinct samples). a (I) Healthy cells as a negative control, 24 h; (II) cells treated with PLY 15 nM, 24 h; (III) cells treated with PLY (15 nM) + PB-3 (6 µM), 24 h; (IV) cells with PB-3 as a control (24 h). b PB-3 blocked cellular deterioration in A549 cells infected with S.pn. D39 strain (data shown is one of n = 3, distinct samples). (I) healthy control cells, (II) cells infected with D39Δcps induced cellular injury (16 h), (III) cells infected with D39Δcps treated with PB-3 (6 µM) 4-times after regular intervals of 4 h (16 h), and (IV) D39ΔcpsΔply control, and it represents that only PLY provokes the cellular injury. All images were recorded by laser confocal microscopy after the mentioned time intervals. Scale bars correspond to 50 µm.