In detail

With the aim to provide a more comprehensive and biologically relevant picture of the drug-membrane interactions and how the effect of these interactions can modify the biophysical properties of the membranes in relation with pharmacological activities, most of the studies are performed by using cells (bacteria or mammalian cells) and membrane models (supported bilayers, liposomes [SUVs, LUVs; GUVs]) mimicking (i) bacterial and (ii) eukaryotic membranes. In close collaboration, we used a range of complementary methods including AFM, 31P NMR, dynamic light scattering, fluorescence spectroscopy (Laurdan, DPH, TMA-DPH, DHE, calcein, octadecylrhodamine B...) and confocal and electronic microscopy.

With the aim to provide a more comprehensive and biologically relevant picture of the drug-membrane interactions and how the effect of these interactions can modify the biophysical properties of the membranes in relation with pharmacological activities, most of the studies are performed by using cells (bacteria or mammalian cells) and membrane models (supported bilayers, liposomes [SUVs, LUVs; GUVs]) mimicking (i) bacterial and (ii) eukaryotic membranes. In close collaboration, we used a range of complementary methods including AFM, 31P NMR, dynamic light scattering, fluorescence spectroscopy (Laurdan, DPH, TMA-DPH, DHE, calcein, octadecylrhodamine B…) and confocal and electronic microscopy.

Cardiolipin domain as target for amphiphilc aminoglycoside derivatives?

Combination of existing lipid diversity and functions with biophysics of bacterial membranes is a unique opportunity to discover new antibiotics. Bacteria (as mammalian cells) have capacity to maintain specialized zones in their membranes for fruitfully fill in their biological functions.

In the frame of our work, we focus on areas characterized by high curvature and enriched in cardiolipin, as encountered at poles and division septa of Gram-negative bacteria, with the aim to understand if and how membrane-acting antibiotics (amphiphilic neamine derivatives) might modify bacterial physiological processes.

Intensive medicinal chemistry development was performed in collaborationwith Prof. JL Decout and coll. (Grenoble, F) from a group of oldantibiotic drugs called aminoglycosides, which target ribosomal RNA. Molecular foundations and structure–activity relationships made on the central backbone (neamine versus neosamine), the nature of the hydrophobic tail (naphtyl, alkyl) as well as the position and the number of substitution on the central backbone to define optimal amphiphilicity, led to the emergence of amphiphilic antibacterial aminoglycosides. More than 80 derivatives were synthesized with very promising compounds active against Gram-positive and Gram-negative sensitive and resistant bacteria.In addition, we did not observe any emergence of resistance in P. aeruginosatreated for 35 days with amphiphilic aminoglycoside derivatives at subinhibitory concentrations.

To decipher the molecular mechanism involved in their activity, we used both living bacteria (P. aeruginosa) as well as membrane model systems including LUVs (Large Unilamellar Vesicles) for membrane permeability and depolarization, GUVs (Giant Unilamellar vesicles) for confocal microscopy and lipid monolayers, for Langmuir isotherm compression. Wedemonstrated the interaction of the amphiphilic neamine derivativeswith outer membrane’s lipopolysaccharides and inner membrane’s anionic phospholipids mostly cardiolipin leading to membrane permeabilization (NPN and PI assays) and depolarization (DiSC3(5) fluorescence). Targeting cardiolipin bacterial micro-domains mainly located at the cell poles, led to relocation of cardiolipin domains associated with bacterial morphological changes including a severe length decrease.

These results suggest an effect of amphiphilic aminoglycoside antibiotics on cardiolipin domains with in turn changes in the activity of proteins dependent upon cardiolipin and involved in bacterial division (FtsZ)and/orbacterial shape (MreB).

3’,6-dinonyl Neamine targets P. aeruginosa microdomains of cardiolipin leading to their redistribution (top) and changes in width and length (bottom). Top: non treated bacteria (left) and treated (5 μM for 10 min at 37°C) bacteria (right); arrows indicate cardiolipin domains. (El Khoury et al, 2017, Sci. Reports)

At a glance, our results bring into light fundamental concepts which could be important in membrane-lipid therapy in which the molecular targets are the lipids and the structure they form. The role of lipids can be (i) to facilitate membrane bending and the formation of highly curved intermediates, reducing the energy barriers of fission and fusion and (ii) to recruit specialized proteins. Influencing curvaturedirectly as well as indirectly by targeting negative intrinsic curvature of lipids or in impairing the soft mechanical behavior could be a new approach for antibiotic design.

Lipid-proteins interactions could be also critical in cell physio- and pathology of erythrocytes

In this context, we investigated (i) whether enriched lipid domains in cholesterol and sphingomyelin could contribute to function-associated cell (re)shaping, (ii) whether the seminal concept of highly ordered rafts could be refined with the presence of lipid domains exhibiting different enrichment in cholesterol and sphingomyelin and association with erythrocyte curvature areas and (iii) how differences in lipid order between domains and surrounding membrane are regulated and whether changes in order differences could participate to erythrocyte deformation and vesiculation

For studying the first question, we probed by vital imaging the lateral distribution of cholesterol and sphingomyelin (using either specific Toxin fragments or trace insertion of BODIPY-SM) in relation with: (i) membrane biconcavity of resting red blood cells; (ii) membrane curvature changes and Ca2+exchanges upon mechanical stretching of healthy red blood cells or in elliptocytes, a red blood cells model of impaired shape; and (iii) membrane vesiculation upon red blood cells aging.

In this aspect, we are exploring the role of P. aeruginosa outer membrane asymmetry and of the Outer Membranes Vesicles (OMVs) formation in bacterial physiology as well as the potential impact for the activity of amphiphilic aminoglycoside antibiotics.

We revealed the specific association of cholesterol-and sphingomyelin-enriched domains with distinct curvature areas of the erythrocyte biconcave membrane. Upon erythrocyte deformation, cholesterol-enriched domains gathered in high curvature areas. In contrast, sphingomyelin-enriched domains increased in abundance upon calcium efflux during shape restoration. Upon erythrocyte storage at 4 °C (to mimick aging), lipid domains appeared as specific vesiculation sites

Recruitment of cholesterol-enriched domains in increased curvature areas of the red blood cells rim upon stretching. Green arrowheads indicate high curvature areas. Scale bars 2μm. Leonard et al, 2016, Sci. Reports.

The second and third questions benefit from the use of a fluorescent hydration-and membrane packing-sensitive probe, Laurdan, to determine the Generalized Polarization (GP) values of lipid domains vs the surrounding membrane. Sphingomyelin-and cholesterol-enriched domains were less ordered than the surrounding lipids in erythrocytes at resting state. Upon erythrocyte deformation (elliptocytes and stimulation of calcium exchanges) or membrane vesiculation (storage at 4°C), lipid domains became more ordered than the bulk.Upon aging and in membrane fragility diseases (spherocytosis), an increase in the difference of lipid order between domains and the surrounding lipids contributed to the initiation of domain vesiculation. Altogether, results demonstrated the critical role of domain-bulk differential lipid order modulation for erythrocyte reshaping probably related with the pressure exerted by the cytoskeleton on the membrane

The existence of clusters of proteins and lipids and especially, the transient nanometric cholesterol-and sphingolipid-enriched domains, called rafts,are described as signaling platforms for a wide range of cellular responses to stimuli including reactive oxygen species (ROS) generation, inflammatory cytokines expression and cell death. we explored the role of cholesterol and cholesterol-enriched domains for cellular toxicity ofthe potential anticancer drug, the ginsenoside Rh2 and the anti-inflammatory complex budesonide: HPβCD.

Taking benefit from our previous studies investigating the mechanisms involved in nephrotoxicity induced by aminoglycoside antibiotics, we explored thecapacity of new antibiotics to accumulate within the cells and to induce accumulation of undigested lipids within the lysosomes. More recently, we started to explore the mitochondrial alterations induced by oxazolidinone antibiotics.

I. Cholesterol-enriched domains and cellular toxicity of anticancer drug (Ginsenoside Rh2)

Pursuing our studies on the molecular mechanism involved in necrosis and apoptosis in leukemic monocytes induced by saponins (α-hederin, a monodesmosidic triterpenoid) and especially the critical role of cholesterol and cholesterol-enriched domains, we extend to ginsenoside Rh2, a steroid saponin (protopanaxatriol) known as one of the active principles of Panax ginsengroot. This work is performed in close collaboration with J. Leclercq’s team.

We demonstrated that membrane cholesterol could delay the activity of ginsenoside Rh2, renewing the idea that saponin cytotoxicity is ascribed to an interaction with membrane cholesterol.

The cytotoxic activity of Rh2 is accelerated in human leukemic U937 cell lines upon cholesterol depletion via the pretreatment with methyl-β-cyclodextrin, a cholesterol-sequestering agent. Mechanistically, Rh2 alters plasma membrane fluidity by compacting the hydrophobic core of lipid bilayer (DPH anisotropy) and relaxing the interfacial packaging of the polar head of phospholipids (TMA-DPH anisotropy). The treatment with Rh2 consequently conducts to the dephosphorylation of Akt and the activation of the intrinsic pathway of apoptosis (loss of mitochondrial membrane potential, caspase-9 and -3 activation).

Rh2 decreases the phosphorylation of Akt faster upon cholesterol depletion. Cells depleted in cholesterol with 5 mM MβCD for 2 h, or not, were incubated for the indicated times with 60 μM Rh2 or with 0.1% DMSO (vehicle). Equal amounts of cell extracts were subjected to western-blot analysis for pAkt, Akt and β-actin protein (Verstraeten et al, 2018, Toxicol. Appl. Pharmacol.)

All these features are induced faster in cholesterol-depleted cells, which could be explained by faster cell accumulation of Rh2 in these conditions.

II. Cholesterol-enriched domains and cellular toxicity of antiinflammatory drug (budesonide) complexed with HPβCD

Budesonide (BUD), a poorly soluble anti-inflammatory drug, is used to treat patients suffering from asthma and COPD (Chronic Obstructive Pulmonary Disease). Hydroxypropyl-β-cyclodextrin (HPβCD), a biocompatible cyclodextrin known to interact with cholesterol, is used as a drug-solubilizing agent in pharmaceutical formulations. Budesonide administered as an inclusion complex within HPβCD (BUD:HPβCD) required a quarter of the nominal dose of the suspension formulation and significantly reduced neutrophil-induced inflammation in a COPD mouse model exceeding the effect of each molecule administered individually. This suggests the role of lipid domains enriched in cholesterol for inflammatory signaling activation.

We first showed that BUD:HPβCD induced an increase in membrane fluidity and permeability induced by BUD:HPβCD in vesicles containing cholesterol. We also demonstrated on giant unilamellar vesicles (GUVs) and lipid monolayers, the disruption of cholesterol-enriched raft-like liquid ordered domains aswell as changes in lipid packing and lipid desorption from the cholesterol monolayers, respectively. Except for membrane fluidity, all these effects were enhanced when HPβCD was complexed with budesonide as compared with HPβCD.

Confocal fluorescence microscopy imaging of membrane phase separation in GUVs upon incubation with BUD:HPβCD, and HPβCD. Imaging of membrane domains in GUVs composed of (left) DOPC:pSM (1:1) and (right) DOPC:pSM:Chol (1:1:3) before (top, control) and after (descending) 5, min with the BUD:HPβCD complex or HPβCD. DOPC:pSM vesicles were labeled with Rho-DOPE (red channel) to visualize the liquid disordered (ld)/solid ordered (so) phase separation in red/dark. DOPC:pSM:Chol were labelled with Rho-DOPE (red channel) and NBD-PE (greenchannel) to visualize the liquid disordered (ld)/liquid ordered (lo) phase separation in red/green channels, respectively. Dos Santos et al, 2017, Biochim. Biophys. Acta. Biomembranes

Since changes in biophysical membrane properties have been linked to membrane signaling including pathways involved in inflammation processes, we moved on cellular models (A549) and demonstrated that BUD:HPβCD could limit (i) hydrogen peroxide-and lipopolysaccharide-induced ROS generation, (ii) alveolar cell death mainly due to HPβCD, and (iii) CXCL8/interleukine-8 expression mainly due to BUD. Our results suggest that BUD:HPβCD would potentially be more beneficial than BUD to deal with COPD-related inflammation.

III. Mitochondrial toxicity of oxozolidinones

Oxazolidinones extert their antibacterial effect by inhibitingprotein synthesis in bacteria. We evidenced a specific inhibition of the synthesis of protein encoded by the mitochondrial genome accompanied by an inhibition of the respiratory function and morphological alterations (swelling of mitochondria and disappearance of cristae). We are now exploring whether these changes may contribute to explain the thrombocytopenia and anemia reported in patients treated by these drugs.Our current data suggest that oxazolidinones prevent the maturation of platelet precursors.

Left: Influence of increasing concentrations of linezolid (LZD) and tedizolid (TZD) on CYTox I expression in HL-60 promyelocytes incubated for 120 h in the presence of increasing concentrations of these drugs.Western blots of CYTox I (protein encoded by the mitochondrial genome) and of Tom 20 (encoded by the nuclear genome) of mitochondrial protein fractions. Right: electron microscopy images of mitochondria from HL-60 cells exposed to 15 mg/L linezolid or 3 mg/L of tedizolid.Milosevic et al,2018, Antimicrobial Agents and Chemotherapy

Bacterial persistent or recurrent infections are associated with two specific lifestyles, namely intracellular survival and biofilms. We are studying antibiotic activity against these two forms of infections in relationship with antibiotic pharmacokinetics (factors determining antibiotic access to the target). 

Cellular pharmacokinetics
We study the cellular accumulation (including the mechanisms of entry) and the subcellular localization of novel molecules in preclinical and clinical development, as a basis for further studies examining their intracellular activities in specific compartments. We try to decipher the mechanisms for their penetration and distribution within the cells. Over the last years, we have focused our interest on new antibiotic classes, like lipoglycopeptides, ketolides, new fluoroquinolonesand new oxazolidinones now present on the market. We are now examining innovative antibiotic classes acting on still unexploited targets in order to define their capacity to accumulate within the cells and then to define their interest for the treatment of intracellular infections.

Cellular pharmacodynamics
In parallel, we study the activity of antibiotics against intracellular bacteria sojourning in different subcellular compartments, mainly Listeria mono-cytogenes(cytosol), Staphylococcusaureus(phagolysosomes), andPseudomonasaeruginosa.We have also extended this model to other bacterial species of medical interest. We developed an in vitro pharmacodynamic approach to comparethe efficacy and the potency of the drugs. In brief, we showed thatantibiotics are in general less effective but equipotent against intracellular than against extracellular bacteria, irrespective of their accumulation level. The data generated with these models have been incorporated inthe dossier having led to the registration of the last antibiotics brought on the market.            We are now trying to elucidate the mechanisms by which intracellular bacteria become tolerant to antibiotics. We specially focus on trying to identify genes involved in intracellular persistence. To this effect, we ran a transcriptomic analysis of intracellular S. aureus surviving antibiotic exposure within permissive eukaryotic cells. We found that these survivors were persisters, i.e. phenotypic variants exhibiting transient non-growing state and antibiotic tolerance. This phenotype was stable but reversible upon antibiotic removal, unveiling a reservoir for relapsing infection.These persisters harboreda major transcriptomic reprogramming but remain metabolically active. Regulation mechanisms were not solely dependent on stringentresponse but includeda network of responses displaying multiples entries, comprising the activation of cell wall stress stimulon, SOS and heat shock responses. These changes led to multidrug tolerance after exposure to a single antibiotic.We have also characterized the capacity to survive inside these cells of clinical isolates collected from persistent infections.

Overview of intracellular persistence regulation of S. aureus. In vacuolar nutrient-rich compartments, persisters are metabolically active cells shielding cell wall, DNA and translation products. Under environmental factors from the host cell, including a carbon source shift and antibiotic pressure, persisters promote a network of stress or adaptive responses displaying multiple entries. Stringent response does not show signs of activity for prolonged periods but rather contributes partly to initiate the switch to a persister phenotype through (i) post-translational modifications, contributing to an almost immediate blockade of bacterial division, and (ii) transcriptional regulation, silencing energy-consuming processes. Regulation circuits also include the cell wall stress stimulon, the SOS response, and the heat shock response. These active responses, together with a decrease in oxidative phosphorylation and in translation levels, lead to multidrug tolerance upon exposure to a single antibiotic. This stable phenotype allows bacteria to maximize the chances of long-term survival. Finally, depending on the level of stress, this state could either revert to replicative forms, or promote the evolution to resistant forms, through increased probability of mutations and horizontal gene transfer. Peyrusson et al. 2020, Nature Communication.

We developed in vitro pharmacodynamic models to evaluate the activity of antibiotics against biofilms made of S. aureus, S. pneumoniaeor P. aeruginosa. We showed that antibiotic efficacy and relative potency are considerably reduced in biofilms as compared to planktonic cultures. With S.aureus, we found that biofilms made of clinical strains isolated from patients suffering from persistent infections are still more refractory to antibiotics. We could demonstrate that this was mainly due to a default of penetration of the antibiotics within these biofilms, which could attribute to the matrix composition (polysaccharide content). On these bases, we are exploring innovative strategies in order to disrupt this matrix and increase antibiotic activity. 

In parallel, we have also started to develop more pertinent models of biofilms, like biofilms growing in artificial sputum medium mimicking the viscoelastic properties of the mucus found in the respiratory tract of patients suffering from cystic fibrosis, or multispecies biofilms developing on orthopedic implants.

Fluorescence microscopy imaging of a multispecies biofilm made of Candida albicans(glycans stained in blue), Staphylococcus aureus and Escherichia coli (stained respectively in green and red by FISH probes)

We have provioulsy demonstrated the role of active efflux as a mechanism responsible for the intrinsic resistance of P. aeruginosato specific antibiotics, like temocillin, or macrolides.

We have now started to evaluate the ipact of this mechanism of resistance in Achromobacter xylosoxidans, a bacterial species which follows P. aeruginosain the colonization of the lung of patients with cystic fibrosis (CF).

NPN (fluorescent substrate of efflux pumps) accumulation in reference strains and clinical CF isolates of Achromobacteras a function of their AZI/AMK MIC

We could demonstrate that effluxindeed plays a major role in the poor susceptibility of this species to commonly used antibiotics. Importantly, also we could evidence some mutations in these proteins that affect their substrate specificity.

In a world of increasing resistance, discovery of antibiotics acting on new, unexploited targets is an important medical need. In coworking with groups active in pharmaceutical chemistry or in pharmacognosy (within the institute or outside), we evaluate the activity of new compounds and try to decipher their mode of action. In this context, wehave discovered with the CMFA group new inhibitors of peptidoglycan synthesis and with the GNOS group, agents reversing resistance to lactams in Staphylococcus aureus or active against Leishmania Mexicana Mexicana and Trypanosoma brucei brucei.

In collaboration with the team of JM Bolla at the Université Aix-Marseille (France), we are also evaluating the activity of original compounds originally designed as inhibitors of efflux but showing much broader synergistic effects with antibiotics, in our models of infections, including intracellular infections and activity against strains that show resistance to other antibiotic classes or mutations in their efflux systems.

Our clinical researchaims at optimizing the scheme of administration of antibiotics in terms of ease of administration, safety, and efficacy, taking into account their pharmacodynamic properties.

At the present time, we are evaluating administration by continuous infusion or prolonged infusion of beta-lactams.More specifically, we perform pharmacokinetic studies in specific patients populations (like haemodialysis patients, critically-ill patients or children) in order to propose optimize therapeutic doses. We investigate the parameters that can affect protein binding of drugs, as only the unbound fraction is thought to be important for activity.