I obtained my Ph.D. in February 1993 in Organic synthesis (@ the "Laboratoire de Biochimie Structurale", Université d'Orsay-Paris XI, Orléans, France; Prof. J.-M. Beau and Dr. J. Prandi). During my thesis, I studied how asymetric synthesis methods are efficient to synthesis unusual thiosugars, components of enediyne antibiotics.
Then, I did two post-docs: During the first one (@ the "Biomembrane Institute", University of Washington, Seattle, WA, USA; Prof. S.-i. Hakomori and Dr T. Toyokuni), I studied the electrophilic azidation as a new source of 2-aminosugars. I also worked on the development of glycopeptide vaccines usable against breast cancer.
During the second one (@ the "Laboratoire de Synthèse de Biomolécules", Université d'Orsay-Paris XI, Orsay, France; Prof. J.-M. Beau), I was involved in the development of a new method relevant to reduce azido-oligosaccharides, based on Samarium diiodide.
In October 1996, I got a position of Assistant Professor or "Maître de Conférences" (@ the Faculté de Pharmacie, Université de Picardie Jules Verne, Amiens; Prof. J.-P. Monti and then Prof. J. Rochette), and I worked on the synthesis and structural studies of proteins and/or oligopeptides.
Then, from September 2003 up to September 2005, I worked as "Research Associate" (@ The Scripps Research Institute, Dpt. of Molecular Biology, La Jolla, CA, USA; Prof. D.-A. Case). I was involved in the study of electrostatic interactions in proteins and nucleic acids.
Since September 2006, I am Professor of Organic chemistry (@ the Faculté de Pharmacie, Université de Picardie Jules Verne), and I am interested in empirical force field development, atomic charge derivation and molecular dynamics studies in implicit and explicit solvent conditions of non-natural oligosaccharides, oligonucleotides and polypeptides.
Present research interests
I) Synthesis and Structural Study of Insulin isolated Chains
Insulin is a polypeptide hormone that is composed of two chains [the A-chain (21 residues) and the
B-chain (30 residues)]. These chains are linked together by two interchain disulfide bridges (residues A7-B7 and A20-B19), a third
disulfide bridge (intrachain) being present in the A-chain (residues A6-A11). In the crystal structures, two different conformations,
referred to as the T and R states, are observed. In the T state, the B-chain includes an extended arm (residues B1-B8), a central α-helix
(residues B9-B19), a β-turn (residues B20-B23) and a C-terminal β-strand (residues B24-B28).
The A-chain includes two α-helixes (residues A1-A8 and A12-A18). In the R state, an additional α-helix appears between the residues B1-B8 (1).
In solution, the insulin structure obtained from NMR studies is in agreement with the T state (2).
Although the insulin structure is well known (in solution and in crystal), insulin variants have been studied to determine which part
of insulin is important for its biological activity. So far, insulin-insulin receptor interactions are not well understood since
no crystal structure of this complex has been determined. The only part of the insulin receptor that has been crystallised is the
intracellular tyrosine kinase domain (3). Thus, major information on the functional surface of insulin has been deduced from
structure-function relationship, using insulin mutants (4-8 among many others !).
We have been implied in the study of the isolated A and B-chains of insulin:
Numerous works have been done on the insulin isolated chains. In particular, it has been shown that insulin can be re-synthesized from
its isolated chains (9). Thus, it has been postulated that the insulin isolated chains present enough structural information to re-form
insulin (10, 11). The structure of both chains has also been widely studied (12, 13).
- In collaboration with Prof Larreta-Garde (Université de Cergy-Pontoise, France) and Dr. J.-M. Wieruszeski (Institut Pasteur,
Lille, France), we studied and compared the conformation of an isolated B-chain (sulfonate derivative or oxidized B-chain) in water
and in solvents, mimicking biological media with restricted water activity (i. e. glycerol). Based on a CD, NMR and molecular modeling
study, we have demonstrated that the conformation of this B-chain is affected by the solvent. Our results have shown that in water the
tertiary structure of the oxidized B-chain loses its central helix and presents a fold between the residues B18-B24 while, in glycerol, the
peptide is more rigid and not folded (14).
- In collaboration with Prof Davoust's laboratory (Université de Rouen, France),
we studied a B-chain analog where the cysteines implied in the two interchain disulfide bridges with the A-chain were replaced by two
serines (residues B7 and B19). Using CD and NMR spectroscopies, and molecular modeling, we compared the conformation of this novel B-chain
and of the oxidized B-chain. We have clearly observed that the tertiary structure of the "bis-serine" derivative (PDB code: 1HO0) is
better resolved than the oxidized B-chain one, and is more in agreement with the B-chain structure found in the native insulin (13, 15).
This could be explained by the presence of the two sulfonate groups in the oxidized B-chain.
By their negative charge, such a chemical group could destabilize the B-chain tertiary structure and modify peptide folding (15).
- Finally, in collaboration with Prof Pujol (Université de Caen, France), we are studying a novel A-chain analog. In this peptide, the cysteines
implied in the interchain disulfide bridges with the B-chain have been replaced by serines (residues A7 and A20) and the intrachain disulfide
bridge has been reconstituted (residues A6-A11). This A-chain mutant has been tested in in vitro cell cuture assays.
The data clearly show that this A-chain mimics insulin effects on DNA synthesis, glucose uptake and glycogen synthesis (16).
To our knowledge, these are the first results showing that an insulin isolated chain displays functional properties similar to those of
insulin. The determination of this A-chain mutant structure is underway and should provide additional information on the mode of action of this "mini-insulin" (16, 17).
(1) Derewenda et al., Nature 1989, 338, 594-596. (2) Kline et al., Biochemistry 1990, 29, 2906-2913. (3) Hubbard, EMBO J. 1997, 16, 5573-5581. (4) Derewenda et al., J. Mol. Biol. 1991, 220, 425-433. (5) Mirmira et al., J. Biol. Chem. 1991, 266, 1428-1436. (6) Nakagawa et al., Biochemistry 1992, 31, 3204-3214. (7) Kaarsholm et al., Receptor 1994, 5, 1-8. (8) Hua et al., J. Mol. Biol. 1996, 264, 390-403, and references cited herein. (9) Dixon et al., Nature 1960, 20, 721-724. (10) Wang et al., TIBS 1991, 16, 279-281, and references cited herein. (11) Tsou, TIBS 1995, 20, 289-292. (12) Hawkins et al., Biochim. Biophys. Acta 1994, 1209, 177-182. (13) Hawkins et al., Int. J. Pept. Protein Res. 1995, 46, 424-433, and references cited herein. (14) Dupradeau et al., Biochim. Biophys. Acta 1999, 1429, 446-458. (15) Dupradeau et al., J. Pept. Res. 2002, 2002, 60, 56-64. (16) Le Flem et al., Biorg. Med. Chem. 2002, 10, 2111-2117. (17) Manuscript in preparation. These publications are simply a short list that has to be completed...
II) Development of new Force Field Parameters using Quantum and Molecular Mechanics
During molecular modeling of insulin isolated A and B-chains (using NMR restraints and force field based simulations), we met two problems:
- As the thiol chemical group of the cysteine residue, found in the insulin isolated chains, is unstable (1, 2), it has been replaced by different chemical groups such as sulfonate (RSO3-), S-sulfonate (RSSO3-) or S-thiomethyl (RSSMe) in several studies (3-5).
In particular, we have recently shown that the sulfonate group, by its charge, presents a de-structuring effect on the B-chain tertiary structure and favors conformational averaging (6).
To accuratly model insulin chains with these unusual amino-acids (i. e. sulfonate and S-sulfonate), new force field parameters for these residues have to be developed.
- As some of these peptides were not soluble in aqueous solution (allowing a NMR study), we decided to carry out some of our NMR study in DMSO. Although different DMSO models were available for molecular mechanics simulations (7, 8), we decided to develop our own model.
We selected the Amber force fields (9-14) to develop these new force field parameters since AMBER force fields are freely available in the public domain and the procedure to develop such parameters is well documented.
At the present time and in collaboration with Dr. G. Moyna (University of the Sciences in Philadelphia, PA) and Dr. P. Cieplak (formerly, University of Warsaw, Poland, then Accelrys Inc., San Diego, now Burnham Institute for Medical Research, San Diego), we are working on force field parameter development and more generally on methods simplifying this tedious and time consuming procedure. We wrote three programs, freely available for academic users:
- The first one, named AmberFFC, automatically converts all the AMBER and GLYCAM force fields to the Accelrys format (15).
It was written entirely with the perl programming language making AmberFFC highly flexible and portable.
AmberFFC is intended to any modeler who is interested in using the current AMBER and GLYCAM force fields (i.e. eleven force fields, so far) within the Accelrys CDiscover and FDiscover programs.
- The second one, named R.E.D. is again a Perl script (16).
It automatically derives 'RESP' and 'ESP' charges starting from an un-optimized PDB structure (17-19).
It sequentially executes (i) either GAMESS or Gaussian to minimize the structure studied and to compute the corresponding Molecular Electrostatic Potential,
and (ii) RESP to fit the atom-centered charges to the grid previously determined.
Format conversions needed during the procedure and "RESP", "GAMESS" and "Gaussian" inputs are automatically generated by R.E.D.
Moreover, a new RESP fitting procedure is proposed allowing to get highly reproducible 'RESP' and 'ESP' charges whatever the QM software is and whatever the starting Cartesian coordinates are. R.E.D. makes the development of the 'RESP' and 'ESP' charges a straightforward, simple and highly reliable procedure.
A graphical user-friendly interface X R.E.D. has been also developed to execute R.E.D. and modify R.E.D. variables.
- The last one, named FFParamDev (still unavailable) (20), is also a Perl script.
It automatically calculates bond, angle and torsion force field parameters. FFParmDev executes GAMESS
to generate quantum mechanics energies and CDiscover to get non-bonding molecular mechanics energies.
Then, it fits Emm (MM energy) to Eqm (QM energy) and determines force field parameters using Hopfinger's method (21).
(1) Andreu et al., Method. Mol. Biol. 1994, 35, 91-169. (2) Annis et al., Method. Enzymol. 1997, 289, 198-221. (3) Du et al., Int. J. Pept. Protein Res. 1982, 20, 47-55. (4) Wei et al., Biochim. Biophys. Acta 1992, 1120, 69-74. (5) Hawkins et al., Int. J. Pept. Protein Res. 1995, 46, 424-433, and references cited herein. (6) Dupradeau et al., J. Pept. Res. 2002, 2002, 60, 56-64. (7) Fox et al., J. Phys. Chem. B 1998, 102, 8070-8079. (8) Vishnyakov et al., J. Phys. Chem. A 2001, 105, 1702-1710. (9) Weiner et al., J. Am. Chem. Soc. 1984, 106, 765-784. (10) Weiner et al., J. Comput. Chem. 1986, 7, 230-252. (11) Cornell et al., J. Am. Chem. Soc. 1995, 117, 5179-5197. (12) Kollman et al. In Computer Simulation of Biomolecular Systems, Wilkinson, P. Weiner, W. Van Gunsteren, Ed., Elsevier 1997, 3A, 83-96. (13) Cheatham et al., J. Biomol. Struct. Dyn. 1999, 16, 845-861. (14) Wang et al., J. Comput. Chem. 2000, 21, 1049-1074. (15) Dejoux et al., J. Mol. Model. 2001, 7, 422-432. (16) Dupradeau et al. Manuscript in preparation (17) Singh et al., J. Comput. Chem. 1984, 5, 129-145. (18) Cornell et al., J. Am. Chem. Soc. 1993, 115, 9620-9631. (19) Bayly et al., J. Chem. Phys. 1993, 97, 10269-10280. (20) To be published (21) Hopfinger et al., J. Comput. Chem. 1984, 5, 486-499. These publications are simply a short list that has to be completed...
III) Modeling of 3-dimensional structures: the Hemochromatosis protein HFE
Hemochromatosis is a disease of the iron metabolism that affects 2-5 in 1000 individuals of European origin (1). A gene responsible for hemochromatosis has been identified close to the
major histocompatibility complex (2). The gene product is a glycoprotein, named HFE, that is homologous to class I MHC proteins and associated with the β2-microglobulin (β2m).
- In collaboration with B. Alterberg (EMBL, Heidelberg), and with the help of Dr. Vriend (CMBI, Nijmegen),
we built a 3-dimensional model of this protein (PDB code 1C42) (3) based on homology modeling, using the program WHATIF(4).
This model represents the first 3-dimensional structure of HFE available in the public domain (recently, this web page moved here).
Then, X-ray structures of HFE and of HFE complexed with the transferin receptor have been released in the protein data bank (PDB codes 1A6Z and 1DE4, respectively) (5, 6).
However, our goal was not to compete with the HFE X-ray structure to be first, but simply to emphasize that the HFE model was built without knowing the X-ray structure (it is important to underline that the
authors of the HFE X-ray structure have protected the Cartesian coordinates for one whole year).
Two common genetic mutations have been identified in the HFE gene (7, 8). The first one results in a cysteine for tyrosine substitution in position 282 (C282Y). In northern Europe, between 70-90 % of patients with hemochromatosis are homozygous for this mutation.
The second mutation corresponds to the replacement of a histidine in position 63 by an aspartic acid (H63D) but the involvement of this mutation in the disease remains unclear. Other rare pathological mutations as well as silent polymorphisms have also been described (9, 10).
The 3-dimensional model of HFE allowed us to present 3-dimensional characteristics of this "class I MHC like" protein, to predict immunogenic loops to develop an antibody (named HFE-RWα1) able to recognize HFE, and to discuss about the consequences of the C282Y and H63D mutations.
In particular, we proposed an hypothesis explaining why the H63D mutation could have no effect on the HFE activity: According to the HFE model, an explanation could be that the introduction of an aspartic acid in that position generates a new salt bridge either with
Arg-66 or Arg-67 in the same loop or with Lys-92 in a nearby helix. On the contrary, the HFE model displays a structural limitation: it does not take into account of the translation of the α1 domain helix towards the α2 domain, narrowing the class I MHC groove.
However, our work shows that a strategy based on homology modeling is sufficient to undertake biological investigations in the absence of the X-ray structure. Finally, evolutionary hypotheses have been considered based on multiple sequence alignments (3).
An interesting comparison of the HFE model and HFE X-ray structure is done as an exercice (exercice 8) by students during "BMB/Bi/Ch_170" courses.
- In collaboration with Prof Ferec's laboratory (Université de Bretagne occidentale, France), we studied the Q283P mutation and demonstrated its important impact on the wild type 3-dimensional structure by unrestrained molecular dynamics (11).
We are also studying other HFE mutations and the structure/mutations of other proteins involved in the iron metabolism...
(1) Merryweather-Clarke et al., J. Med. Gen. 1997, 34, 275-278. (2) Feder et al., Nat. Genet. 1996, 13, 399-408. (3) Dupradeau et al., Biochim. Biophys. Acta 2000, 1481, 213-221. (4) Vriend, J. Mol. Graph. 1990, 8, 52-56. (5) Lebron et al., Cell 1998, 93, 111-123. (6) Bennett et al., Nature 2000, 403, 46-53. (7) Feder et al., J. Biol. Chem. 1997, 272, 14025-14028. (8) Waheed et al., Proc. Natl. Acad. Sci. USA 1997, 94, 12384-12389. (9) Rochette et al., Am. J. Hum. Gen. 1999, 64, 1056-1062. (10) Wallace et al., Gastroenterology 1999, 116, 1409-1412. (11) Le Gac et al., Blood Cell Mol Dis 2003, 30, 231-237. These publications are simply a short list that has to be completed...