Research Activities with Biomolecular Systems

In order to interpret experimental IRMPD spectra, two possibilities are offered. The first one consists in a systematic search for the different potential energy minima. IR spectra are subsequently calculated for each low-energy equilibrium conformations. This is usually done within the harmonic approximation with some appropriate scaling factor. Comparison between the experimentally observed spectrum and the predicted spectra is then used to identify populated isomers and/or conformers of the studied system.

An approach closer to experimental conditions consists in simulating the dynamical behaviour of peptides through DFT-based Car-Parrinello molecular dynamics (CPMD) conducted at the experimental temperature. This task is performed in the group of M.P. Gaigeot LAMBE Evry. MD is essential for including conformational dynamics (entropic effects being taken into account without any approximation), and is thus essential for the calculation of IR spectra of complex molecular systems that display numerous energetically equivalent conformations. The main point of IR spectra calculations based on the dipole time correlation function is that all anharmonic effects are naturally taken into account, and no approximations are made. This is to be opposed to the two successive harmonic approximations usually adopted for the determination of IR spectra from static ab initio calculations (harmonic approximation of the potential energy surface at the optimized geometries and mechanical harmonic approximation for the transition dipole moments, as used in normal modes analysis). In particular, the finite temperature dynamics takes place on all accessible parts of the potential energy surface, be they harmonic or anharmonic.

Mass-selected ions absorb photons that bring them from a v_i=0 to a v_f=1 vibrational state. Between two photon absorptions, internal vibrational redistribution (IVR) takes place and brings back ions to their initial v_i=0 vibrational state, while the acquired internal energy is redistributed over other modes. Following absorption of many photons, dissociation takes place. Infrared absorption is then monitored through detection of the ionic fragmentation yield.

IRMPD offers a direct possibility for determination of protonation sites by comparison between experimental IR absorption spectrum and the predicted spectra of the different tautomers. In the case of protonated dialanine, static calculations of the lowest energy configurations show that the proton can be located at the N-terninal (labelled A1 and A2) or C=O (labelled O1) amide groups at 300 K. These conformations are separated by energy barriers with transition states accessible at room temperature. Conformations A1 and A2 give the best agreement with the experimental spectrum while conformation O1 contributes to a much less extent to the experimental spectrum.

Car-Parrinello molecular dynamics (CPMD) simulations on Ala-AlaH⁺ peptide at 300K have shown a continual and recurrent isomerization dynamics between the two A1 and A2 conformers, i.e. the internal energy of the peptide at 300K is sufficient to overcome the energy barrier separating the A1 and A2 basins on the potential energy surface. Besides, one important result is the observation of spontaneous and reversible proton transfer events between NH3⁺ (A conformer) and the adjacent C=O amide (O conformer). This indicates that O1 is a metastable state which is thermally accessible at 300K. The IR spectrum extracted from the CPMD simulation is calculated from the Fourier Transform of the dipole time correlation function. All pertinent band shaping effects, such as anharmonicity and temperature are incorporated ab inito via the CPMD methodology.

An important recognized motif involved in cell adhesion and inflammatory processes is the arginine-glycine-aspartic acid RGD sequence, found in large glycoproteins such as fibronectin, participating to the composition of the intercellular medium. It is specially recognized by transmembrane proteins called integrins and initiates cell-signalling processes.

We have shown that the protonated RGD sequence adopts in the gas phase at least two conformations at 300K, one being similar to the structure of RGD as embedded in a protein (dendroaspin). In this neurotoxin homologue, the flexible RGD loop structure corresponds to residues 43-45 located on the surface of the protein. Our gas-phase study has shown that this loop structure is an intrinsic structural property of RGD, even if the acid aspartic carboxyl group is neutral while it is deprotonated for peptides in solution.

Structural information on acetylcholine and its two agonists, nicotine and muscarine has been obtained from the interpretation of IR spectra recorded in the gas-phase (IRMPD) or in low pH aqueous solutions (FT-IR). In the gas-phase, the IRMPD spectrum of acetylcholine ions exhibits three distinct transitions of the carbonyl stretch around 1750 cm^-1, which implies that at least three different conformations are populated at 300K. Among these conformers, only one corresponds to a N⁺-O distance (5.05 Å) that fits within the bioactive range, while the four others low energy conformers exhibit much shorter N⁺-O distance. However, in aqueous solution, this distance in the four lowest energy conformations is comprised in between 4.6 and 5.2 Å, all compatible with the bioactive range. The conformational space of these very flexible molecules is thus strongly modified by the presence of the solvent.

In the case of nicotine, the solvatation induces a complete change of the structure and the protonation site. Nicotine possesses two protonation sites on the nitrogen of the pyridine (labelled N₁(sp²)) and pyrrolidine (labelled N₁₂(sp³)) cycles. In solution, a strong transition at 1450 cm^-1 is observed and corresponds to the bending mode of N₁₂-H⁺. In the gas phase, the IRMPD spectrum of protonated nicotine exhibits an intense transition centred at 1540 cm^-1 that involves the bending motion of N₁-H⁺ group. This result has provided for the first time a clear experimental evidence for protonation of nicotine on the pyridine site, which is not the bioactive site of nicotine in condensed phase.

Nucleic acid secondary structures are determined by hydrogen bonding interactions between nucleic bases. Besides the well-known Watson-Crick base pairing motif, there is a variety of other possible hydrogen bonding configurations. For instance, guanine-rich sequences can fold into G-quadruplex structures owing to Hoogsteen hydrogen bonding between four guanines, forming G-quartets. As evidence is accumulating on in vivo G-quadruplex formation in telomeric DNA and in the promoters of some oncogenes, these structures have become key targets for anticancer strategies. G-quadruplexes are present in telomeres (ends of human chromosomes) that consist of long DNA single strands containing repeats of TTAGGG sequence.

We have investigated the IR signature of two quadruplex-forming sequences: dTG₄T, forming a tetrameric quadruplex [(dTG₄T)₄(NH₄⁺)₃] that is highly stable in solution and in the gas phase, and the human telomeric sequence dTTAGGGTTAGGGTTAGGGTTAGGG (noted T₄), forming an intramolecular antiparallel quadruplex in ammonium acetate solution. For both species, a significant red shift of the C=O stretch mode of the guanine base is observed for the non-covalent complexes with ammonium cations sitting in between two successive G-quartets. This first experimental study of large biomolecules has proven the ability of IRMPD spectroscopy for gas-phase conformational analysis of rather complex systems.

The binding of vancomycin to the tripeptide cell-wall precursor analogue Ac₂^LK^DA^DA involves several intermolecular hydrogen bonds, three with the carboxylate group of the tripeptide and two between amide groups of both species. The binding pocket thus contains the pseudopeptidic chain of vancomycin while the benzoic cycles and the sugar moiety do not seem to be directly involved in the formation of the complex.

The mid-IR spectra (1100-1800 cm^-1) of these mass-selected anionic species have been recorded by means of resonant infrared multiphoton dissociation (IRMPD) spectroscopy performed with the free-electron laser CLIO. Structural assignment has been achieved through comparisons with the low-energy conformers obtained from replica-exchange molecular dynamics simulations, for which IR spectra were calculated using a hybrid quantum mechanics/semi-empirical (QM/SE) method at the DFT/B3LYP/6-31+G^*/AM1 level. Comparison between deprotonated vancomycin and its non-covalently bound V+Ac₂^LK^DA^DA complex shows significant spectral shifts of the carboxylate stretches and the Amide I and Amide II modes that are satisfactorily reproduced by the structures known from the condensed phase.

The situation is rather different in the protonated form. Neutralization of the carboxylate group impairs the specific interaction of the deprotonated group of the receptor with vancomycin. Earlier gas-phase studies have suggested that the native structure of the complex might not be preserved in the protonated form, although no direct spectroscopic study has been yet undertaken.

The structure of doubly protonated vancomycin antibiotics with its cell-wall precursor analogue Ac₂^LK^DA^DA has been investigated in the gas phase through a combined laser spectroscopy, ion mobility (in collaboration with the group of Ph. Dugourd, LASIM) and theoretical modeling approach. Replica-exchange molecular dynamics simulations using the Amber99 force field were performed to explore the potential energy landscape of isolated vancomycin ions, as well as the different binding sites with the receptor. Among the low-energy conformers found, those with a calculated diffusion cross-section consistent with ion mobility experiments were selected for further optimization, and their IR spectra were simulated using a hybrid quantum mechanics/semi empirical (QM/SE) method at the DFT/B3LYP/6-31g(d):AM1 level. Both theoretical and experimental findings provide strong evidence that the native structure of the complex is not preserved in vacuo for the doubly protonated species.

This project aims to study the specific recognition involved in complexation processes of non-covalently bound systems. One of the unique advantages of this gas phase experiment is to study the local interactions between molecules in a free environment or within a controlled number of water molecules in systems in which the stoichiometry of the drug-receptor complexes is perfectly known.

The liquid droplets (50 µm diam) are generated on demand by a commercial droplet generator (Microdrop) and injected from 300 mbar into a high vacuum chamber (10^-5-10^-6 mbar) through two differential pumping stages. Upon irradiation by home-made mid-IR broad band laser pulses (centered on an absorption band of the solvent), analyte ions are ejected directly into high vacuum. At this stage, the desorption occurs in between the first stage of a two-field ion optics and the emitted ions are then accelerated and detected by micro channel plates (MCP) located 80 cm downstream in a second vacuum chamber.

We are currently building a reflectron-type TOF mass spectrometer that will be installed in the second vacuum chamber. In that configuration, the ions will pass through decelerating ion optics and will be spatially focussed in the extraction region of the orthogonal TOF mass spectrometer. The reflector assembly should ensure a good mass resolution. The mass-spectrometer vacuum chamber will be equipped with several laser feedthroughs before and into the reflector zone to perform laser photodissociation spectroscopy.

At the ARIBE Facility (Caen, France), with the groups of B.A. Huber (Caen) and T. Schalhölter (KVI, Groningen, Netherlands), we studied the interaction of 300 keV Xe²⁰⁺ ions with two amino acids: [D2]glycine (NH₂CD₂COOH) and valine [NH₂(CH)₂(CH3)₂COOH]. For the isolated molecule, the use of an incident projectile with a higher charge state has the advantage of being a soft ionisation tool due to gentle interactions at large distances. Thus, a smaller amount of energy is transferred, leaving the system in a less-excited state. An estimate of the internal energy of the fragmenting systems, obtained by applying the simplified model of an evaporative ensemble to the glycine molecule, yields values between 5 and 6.5 eV depending on the model parameters. To identify the effect of the environment, we extend the study to clusters of glycine and valine. The fragmentation patterns of the molecules isolated and embedded in clusters can be compared and the fragmentation dynamics discussed.

In the case of isolated amino acids in the gas phase, ion irradiation provokes a unimolecular dissociation process, in particular, the fragmentation is driven by the cleavage of the C-C_α bond, which appears to be the weakest bond of the system. For the larger valine molecule, the loss of the side chain appears as an alternative dissociation process. The valine cation is found to be unstable on the µs timescale.

In the case of amino acid clusters, the first striking result is the observation of the amino acid cation, which is totally absent for the isolated valine molecule and strongly enhanced in the case of glycine. The kinetic energy of these cations indicates that they result from the fragmentation of larger clusters cooling the system. This buffer effect of the cluster is also evidenced by the fact that the fragmentation yield is strongly reduced. The signal intensities associated with the main fragmentation channels for the C-C_α bond cleavage are reduced, which confirms that the amino acid clusters dissipate their excess energy by the evaporation of monomers or by cluster fission. The fragmentation pattern, that is, the possible pathways are also modified. In particular, the decay leading to the formation of the COOH⁺ ions is quenched in the case of both glycine and valine clusters. The disappearance of this fragmentation channel can be explained by a rapid charge redistribution within the cluster.