Eric Hajjar, Amit Kumar, Enrico Spiga, Francesca Collu, Atilio Vargiu, Paolo Ruggerone and Matteo Ceccarelli

Transcription

1 University of Cagliari, Dept of Physics CNR-SLACS: Sardinian LAboratory for Computational Materials Science Eric Hajjar, Amit Kumar, Enrico Spiga, Francesca Collu, Atilio Vargiu, Paolo Ruggerone and Matteo Ceccarelli

2 University of Cagliari, Dept of Physics Eric Hajjar Wednesday talk CNR-SLACS: Sardinian LAboratory for Computational Materials Science How molecular simulations can help understand permeation properties of antibiotics through porins Towards Antibiotic in-silico drug design methodology and proof of concept. Amit Kumar, Thursday talk Successful stories using this methodology; methods, presentation of previous and novel results. 5 POSTERS -Molecular simulations of cephalosporins diffusion through OmpF -Searching for similar patterns on the translocation of fluoroquinolones through OmpF -Methodologies and perspectives in the simulation of antibiotics translocation -Molecular Simulation of Penicillin's Diffusion through Bacterial Poring OmpF -Kinetic monte carlo study of translocation

3 Outline 1. Introduction / Biological problems 2. Methods / Theoretical solutions From classic Molecular Modeling to advanced Metadynamic algortithm Our strategy 3. Antibiotic in silico drug design / Proof of concept Study of a common antibiotic in wild type and a natural strain of OmpF mutation identification of the determinants for translocation. 4. Extending methodology to cephalosporins 5. Conclusions, Perspectives, Work in progress

4 Biological Problem resistance BACTERIA ANTIBIOTICS inhibit growth of bacterias cell wall Focus on Gram-negative bacteria: - Pathogenic for humans, increased antibiotic resistance - Outer Membrane rich with lipopolysaccharides (LPS) and porins.

5 Problem of Bacterial resistance to Antibiotics American Medical Association (AMA): "The global increase in resistance to antimicrobial drugs has created a public health problem of potentially crisis proportions." Many ways for Bacteria to resist to Antibiotics -Production of B-lactamases -Under-expressing porins -Over-expressing efflux pumps -Mutating porins to affect antibiotic uptake Need of a new way to design antibiotics / a bottom-up approach: focus on the molecular basis of antibotic transport and bacterial resistance.

6 Antibiotics are transported in bacteria via Outer Membrane Proteins Antibiotics have to diffuse passively through some general porins at the outer membrane OmpF and OmpC are the most abundant in Gram-negative bacteria Have been thoroughly investigated by many techniques General diffusion proteins (poor substrate selectivity, often open)

7 OMPF described in literature as general diffusion protein OmpF: X-ray structure in View on OmpF monomer: - Beta Barrel, 8 loops which are extracellular...except L3: folds back inside Constriction region with a few described important residues.

8 Is OmpF just a tube channel? a) There is a constriction region: with the L3 that folds back inside (radius ~ 7Å) b) There is a particular electrostatic field, above / at / bellow the constriction region! Z-slices of the electrostatic potential

9 Is OmpF just a tube channel? a) There is a constriction region: with the L3 that folds back inside (radius ~ 7Å) b) There is a particular electrostatic field, above / at / bellow the constriction region! Maybe not Justify the approach: focus on the molecular basis of antibotic transport, to design antibiotics with improved permeation properties In this rational drug-design scenario molecular simulations can have an important role.

10 Molecular modeling Broad range of computational methods with associated analysis tools A molecule is: composed of atoms and springs between them Energy: any arrangement of atoms and molecules in the system E ~ f (atomic positions) The potential energy function (E) is a sum of terms r réq φ θ éq θ

11 Molecular Dynamic (MD) Calculates the time dependent behavior of a molecular system - Strategy: de F = dr Calculate forces on each atom dv d2x F =ma=m. =m. 2 dt dt Iterative integration of Newton s laws (over dt). - Result: trajectory, specifies how the positions of the atoms vary with time

12 Energy landscapes The case of butane As the size of the molecule increases: bigger steric effects; more complex energy landscape

13 Advanced (realistic) energy landscape Simple (trivial) energy landscape and in search for the minum in energy The most stable conformation of a molecule is the one with the lowest energy

14 In our case we want to use computational simulations to study ANTIBIOTICS translocation through PORINS Our Systems of study The Porin systems: Antibiotics: different famillies with different properties Wild Type OmpF (and OmpC in preparation) OmpF-variants: R42A, R82A, R132A, D113A, D113N, E117A. according to litterature and discussions with partners Thursday s talk

15 Antibiotics of different charge, size, hydrophobicity...interest! penicillins fluoroquinolones cephalosporins Cefpirome, Ampicillin Ciprofloxacine Norfloxacine Cephepime Penicillin-G Ceftizoxime Moxifloxacine Cefetamet Carbenicillin Enrofloxacine Thursday s talk Cefpodoxime

16 The case of Antibiotics passage through porins Experimentally, from electrophysiology experiments on BLM, the time of this process is ~100 µs (Winterhalter & Bezrukov). This time exceeds typical simulations time ~ the process is classified as rare event Using MD, free energy barriers are difficult to cross LIMITATIONS of MD for studying antibiotic translocation Reaction Coordinates Our strategy to overcome this problem: Accelerated Molecular Dynamic (MD) simulations

17 Accelerated Molecular Dynamic METADYNAMICS (Laio and Parrinello, PNAS 2002) Construct a bias potential that discourages the system from revisiting configurations that have already been explored. Free Energy landscape is being filled up by metadynamic Very efficient method to sample free energy landscapes / accelerate evolution of the system Thursday s talk

18 Simulations of antibiotic translocations OMP Building Complexes + Monomer / Trimer Wild Type / Mutants antibiotic FF field parametrization detergent molecule / lipid bilayer Solvation cubic pre-eq. water box, counter ions) 4 CPU with proper reaction coordinates Metadynamic run ~ 40 days (samples millions of conformations) Quantitative: free energy landscape of translocation process Analysis Qualitative: Inventory of the interactions, area, atomic fluctuations.

19 III) Proof of concept Towards in-silico design of antibiotics Start with a common antibiotic study its tranlocation through OmpF WT and Mutants Propose a better antibiotic in accordance with the findings Verify / Extend the hypothesis

20 Ampicillin translocation through OmpF

21 Contour plot of the free energy surface (FES) for the Amp-OmpF WT simulation Constriction zone Highly populated (deep) minimum in energy at Z {-1:1) and Θ {120:160}. The energy (ΔG) to overcome this barrier is ~8kcal. Thursday s talk

22 Visual Analysis? complicated and untractable to analyse trajectories with the eyes We used various computational methods

23 Some ways of quantifying the translocation process Inventory of interactions Cross sectional area calculation Simulation time Atoms of Antibiotic Thursday s talk

24 Identify the conformations corresponding to the free energy minima extract minima I II III VI - Quantify the barrier between each minima, - Launch standard equilibrium MD for each of them IV V

25 Conformations along MD simulations of Minima s Simulation along Mini-I&II above constriction region along Mini-III & IV At the constriction region Amp- OmpF WT along Mini-V below constriction region F D At constriction region D113 induce repulsion Strong binding site D113 Hbonds to antibiotic / slows down escape

26 III) Proof of concept Towards in-silico design of antibiotics 1 Start with a common antibiotic study its tranlocation through OmpF WT and Mutants 2 Propose a better antibiotic in accordance with the findings 3 Verify / Extend the hypothesis The N+ group of Amp slows down its diffusion due to interactions with D113

27 Simulations of Ampicillin with OMPF-D113A WT D113A Amp successfully translocate through OmpF-D113A A single mutation changes drastically strength and localization of minima on the FES

28 Conformations along MD simulations of Minima s above constriction region At the constriction region below constriction region Quickly and passively translocate Amp find the proper configuration to translocate faster as there is no repulsion with A113. stronger hydrophobic interactions with the porin.

29 III) Proof of concept Towards in-silico design of antibiotics 1 Start with a common antibiotic study its tranlocation through OmpF WT and Mutants 2 3 Propose a better antibiotic in accordance with the findings Verify / Extend the hypothesis The N+ group of Amp slows down its diffusion due to interactions with D113 The mutation D113A strengthen the hydrophobic character which helps diffusion PenG, lacks the Nterm + group

30 Translocation of PenG is optimal! Does not require interactions on Nterm side, but is directed by its COO- Above the constriction region; PenG avoids the repulsion (of its COO-) with D113 as it adopts a particular folded structure. Importance of the flexibility of the antibiotic along standard MD of Mini-II

31 Translocation of PenG is optimal! Below the constriction region; PenG phenyl group is attracted by hydrophobic pocket while COO- interact with basic cluster. PenG undergo a 180degree rotation, then quickly translocate in this orientation with this phenyl down. Importance of the hydrophobicity of the antibiotic along standard MD of Mini-IV

32 Outcome map of the antibiotics interactions Penicillin-G Ampicillin (N1) E117, E113, F118, G119 (N3) E113 (O5) R42, R82 Importance of (O & OXT) R42, R82,distribution charge R132, K16, Y40 (N3) E113 (O2) R168, K80, R82, R132, R168, R167, Y102 (O & OXT) S125, R42, R82, R132, K16 (O5)K80, R82, R132

33 III) Proof of concept Towards in-silico design of antibiotics 1 Start with a common antibiotic study its tranlocation through OmpF WT and Mutants 2 Propose a better antibiotic in accordance with the findings The N+ group of Amp slows down its diffusion due to interactions with D113 PenG, lacks the Nterm + group Cephalosporins, (CFR) 3 Verify / Extend the hypothesis

34 Simulations of Cefpirome (CFR) translocation Binding site at constriction region Presence of binding sites Below the constriction region WT I IV II V VI D113A VII III

35 CFR-WT The antibiotic take advantage of a great flexibility to quickly slide down at the constriction region: Mini-V D113 plays moderate role in the binding as CFR diffuse down Below the constriction region Mini-VI Mini-VII CFR is now favorably interacting with two walls of hydrophobic pockets

36 CFR- OmpF(D113) Mutation enhance (hydrophobicity) the binding site above the constriction region just above constriction region Strong central binding site Mini-VI From there: CFR released far from basic cluster wall, escape much faster than WT... Mini-VI and downward

37 Conclusions from our simulations...(1/2) Importance of Flexibility: - Being able to adopt a wide range of conformations is an advantage to translocate. Importance of Hydrophobicity: - Several hydrophobic pockets found. -Mutation D113A enhance hydrophobicity ~ opens a Hpocket at the constriction region, CFR enters there Importance of Polarity: - Several partners for making hydrogen bonds. - Particular position of basic residues: a staircase for translocation

38 Conclusions from our simulations...(2/2) Our results refute presence of a single barrier and/or single binding site. There are several of them along the translocation path. Importance of the topology of the free energy surface: Characterize the nature, strength and localisation of each minima Wolfgang R. Bauer, Bezrukov, our kinetic monte-carlo schemes... This work provide a rational basis for -designing antibiotics with improved properties -predicting antibiotic susceptibility -interpreting experiments -improving the model of antibiotic diffusion

39 Work in Progress / Perspectives Improve our quantitative and qualitative description of translocation: - Add new reaction coordinates in the metadynamic algorithm - Estimate and minimize the error in free energy. - Effect of desolvatation - Kinetic monte Carlo to estimate the barriers between minima s Improve the comparison with experiments / synchronisation with partners Thursday s talk Electrophysiology-disagreement but Liposome swelling-agreement One particular strength of the project: many partners towards same goal (electrophysiology, swelling assays, fluorescence, drug design, antibiotic susceptibility)

40 Acknowledgments Project No.: RTN Translocation TO ALL OF YOU