IULTCS 2005 Heidemann Lecture

Transcription

1 IULTCS 2005 Heidemann Lecture A Molecular Dynamics Approach to the Supramolecular Structure of Collagen and Related Binding Properties Simona Bronco, Chiara Cappelli, Francesco Ciardelli, INFM, Polylab, c/o Dipartimento di Chimica e Chimica Industriale, Università di Pisa Susanna Monti, Istituto per i Processi Chimico-Fisici (IPCF-CNR) Pisa Abstract The change in the stability and conformational dynamics of a collagen-like microfibril segment (CMS) 23 amino acid residues long (around 5[3(GLY-X-Y)8]) as a result of the interaction with water, formaldehyde and gallic acid was studied by performing a series of nanosecond molecular dynamics simulations. The examination of average interaction energies per residue and their van der Waals and electrostatic components confirmed the results obtained with the structural analysis. Furthermore both van der Waals and electrostatic terms were important for the stabilization of the CMS: this was especially noted when CMS interacted with gallic acid. Also a molecular dynamics approach was carried out towards the structure, stability and conformational dynamics of an assembly of two pentameric bundles made of collagen-like triple helical segments. Stable supramolecular arrangements, where the two collagen unities were very close to each other at interacting distances, were identified via docking and energy minimization procedures. Analysis of the interaction with formaldehyde and gallic acid suggested that they perturbed the protein trace in a similar way depending on hydrogen bonding capability, hydrophobic association properties, size and concentration of the compound. The detailed information derived from the present theoretical approach is discussed to put in evidence the significant help to the experimental study and application of collagen related materials. Indeed it certainly provides a better understanding of the effect of chemicals on a very complex natural macromolecular system and can be of help in designing theoretically-driven experiments. Introduction Collagens are a large family of proteins sharing some common traits but also exhibiting wide differences and achieving different functional roles concerned principally with the maintenance of tissue and cellular shape, strength and structural integrity. Collagens are one of the most crucial components of the extracellular matrix and the fibril-forming biochemical types I, II, III, V, XI, are the most abundant and extensively studied. In addition to biological function they are the target of a variety of applications in industry (medical, pharmaceutical, food and leather sectors). Collagen is a protein with a complex supra-molecular organization; its molecules interact with each other at different hierarchical levels forming a characteristic high-order periodic structure with distinctive features and specific functions. The precise nature of collagen packing and self assembly in tissues has been widely investigated but the absolute three dimensional (3D) arrangement of collagen molecules, their lateral supramolecular organization, is not well understood. In 1968 Smith [1] proposed the so-called five-stranded pentagonal microfibril model which consisted of five macromolecules rolled up into a hollow cylinder, a description that was consistent with both the X- ray diffraction and the TEM data. Many other models attempting to combine X-ray diffraction data, electron microscopic results and biochemical data have been suggested over the years, each receiving confirmations and confutations so that no consensus structure has emerged so far. It is commonly accepted that a collagen molecule consists of three -chains arranged in a tight triple-helical entity. The helical conformation of each chain is dependent on the presence of Glycine (GLY) residues every two residues and on the high content of Proline (PRO) and Hydroxyproline

2 (HPR) residues. In all fibrillar collagens the -chains include an uninterrupted sequence of about 300 GLY-X-Y triplets flanked by much shorter terminal domains of different structure. In collagen type I, the most abundant, the triple helix organization leads to the formation of a long, rod-like structure, stiff but flexible, about 1.5 nm wide and over 300 nm long with globular domains at both ends. Collagen is also able to interact with various classes of chemicals [2]. Of particular interest is collagen binding with aldehydes and vegetal tannins: the ability of formaldehyde to interact with collagen has been known for a long time, and in fact it has been used as preserving agent for biological specimens or in embalming, as well as in the tanning process for the production of leather. Even more interesting is the binding between collagen and poly-phenolic compounds. In fact, not only these molecules can act as tanning agents, but their anti-oxidant role in carcinogenesis has also been evidenced. Despite such properties, only a little number of papers on the experimental study of interaction between collagen and modifying agents is present in the literature. This evidences the difficulties in treating this matter from the experimental point of view, which are mainly due to the intrinsic complexity of the collagen system, which indeed increases if the interaction with the "environment" is taken into account. For this reason, especially in this case the computational investigation can be of help in supplementing and rationalizing experimental findings. In addition, the computational approach permits to extend the study towards the understanding of the Collagen supra-molecular structure and how it is modified when Collagen is surrounded by chemicals. On the other hand, the treatment of a complex system needs the use of complex models and perhaps this is the reason why the literature on computational approaches to this matter is very limited. Methodology The collagen microfibril segment (CMS) used in these studies was generated by taking a a short segment 23 AA residues long (around 5[3(GLY-X-Y) 8 ]), made of five triple helical segments (THSs) and rich in hydrophilic residues (Figure 1). The CMS was extracted from a 3D computer model of the bovine type I collagen microfibril based on the Smith description made available to us by Professor Eleanor M. Brown [3]. The AA content of CMS is reported in Figure 2: notice that all the amino-acidic residues present in the large microfibril model can also be found in CMS and the ratio between the total amount of hydrophilic and hydrophobic residues is also preserved. The study of the supra-molecular structure was done by generating two-cms bundle models. In particular, 24 models of dimeric CMS helices were generated using medium-resolution docking approach followed by in vacuo energy minimizations: the most stable parallel and antiparallel conformation at neutral and at acidic ph were then chosen for molecular dynamics (MD) simulations. In Figure 3 a graphical representation of the two-cms bundles is reported. MD simulations were carried out using periodic boundary conditions in the NPT ensemble. Water was described using the TIP3P model [4]. A graphical representation of a portion of the two-cms bundles systems surrounded by a solution of gallic acid and water is depicted in Figure 4. All MD simulations were performed using the AMBER7 package [5] the Cornell et al. Force field [6] and the General AMBER Force Field (GAFF) running partly on computational resources at IPCF and partly on the IBM Linux Cluster CLX at CINECA supercomputer centre. Further details on the computational procedure can be found in the cited references [7, 8]

3 Figure 1. Definition of CMS Scheme I: structure of gallic acid

4 35 30 microfibril CMS % ALA ARG ASN ASP CYS GLN GLU GLY HIS HPR ILE LEU LYS MET PHE PRO SER THR TRP TYR VAL residue Figure 2. Top: AA sequence in the five triple helical bundles (THS) of CMS; the one-letter code is used to label the AA residues. Bottom: comparison of AA content in the original microfibril and in CMS.

5 Figure 3. Three dimensional structure of the two-cms bundles models. The thirty helices are drawn as cylinders. Three cylinders of the same colour form a THS and five THSs form a CMS.

6 Figure 4. A graphical representation of a portion of the two-cms bundles surrounded by a solution water/gallic acid. Results and Discussion The stability of CMS, its supramolecular structure and binding properties have been investigated in four different environments: a) 8.0%(v/v) formaldehyde/water solution; b) 1.4%(v/v) gallic acid (see scheme I)/water solution; c) pure water and d) pure formaldehyde. Bundle Stability and Structural Features [7, 8] To compare the relative stability of the models and the dynamic behavior of the helix bundles, various indexes, extracted from the MD trajectories were calculated. The detailed definition and

7 analysis of such indexes can be found in the cited reference. Here, only the findings which can be extracted from such an analysis will be reported. Major changes in the conformation of collagen occur when the CMS is in water. This is in agreement with data reported in the literature, showing that water molecules are one of the major contributors to the enthalpy of denaturation of collagen. Contrarily, minor changes in the collagen structure occur in the other cases. This is again in agreement with experiments on the direct determination of the influence of chemical modification on collagen molecular stability. In fact, it has been shown that even the alkylation of the collagen amino groups in the presence of NaBH 4 (at least performed with mild conditions) has little or no influence on collagen at the level of triple helical stability. Comparing the two simulations of collagen with formaldehyde, a larger deviation from the starting conformation is noted when CMS is surrounded by an aqueous solution of formaldehyde, thus confirming the denaturing role of water. Only a slight modification of the CMS structure is evidenced as a consequence of the interaction with gallic acid: again it has been experimentally shown that the interaction with polyphenols increases the hydrothermal stability of collagen. In all cases the greatest structure fluctuations are at the residues on the helical side exposed to the solvent. This can be explained by the fact that such residues have a large number of possible interactions with modifying agent and water molecules. However, the fluctuations are not large, thus indicating a general stability of CMS in all cases. Towards the supra-molecular structure: packing of the two CMS bundles [7, 8] The analysis of the two-cms models shows that, due to the dipolar interaction among the helical dipoles, which is more favorable when the CMSs helices are antiparallel arranged, the antiparallel arrangement of the two bundles leads to complexes more stable than the parallel ones. All the most stable arrangements, despite their structural differences and protonation states, share common features, being the angles between the two CMS bundles and the solvent-exposed surfaces very similar. In addition, in all cases the two CMS are very close to each other and the lateral chains are located at interacting distances. Similarly to what is found in the case of the single CMS, a general stability of all the models is noticed, being, as expected, the greatest fluctuations at the residues on the helical side exposed to the solvent. Also in this case the greatest fluctuations observed in pure water. The examination of the various crossing angles i.e. the angle between the long axes of the two CMSs, as a function of time shows that its range of variation is quite narrow and that the relative orientation of the helices at the inter-bundle interface has small fluctuations. All these findings suggest that there is no rearrangement of the helix packing within the various models during the course of the simulation, thus confirming a general stability of all five models. The analysis of the distances between the bundles shows that they are in contact and that the majority of contacts involve alanine residues. A small number of inter-helical hydrogen bonds at the inter-bundle interface are formed. These are often found between AA side chains, less frequently between the backbone atoms and are mostly localized at the middle toward the end of the helices. Structure and dynamics of solvent molecules [7, 8] In addition to the conformational dynamics of CMS and the CMSs bundles, it is possible with MD to analyze the interaction of the helices with their solvent environment made of cross-linking agent and water molecules. All solvent species, water, formaldehyde and gallic acid are found in the first solvation layer around both CMS and the two bundles, where they are in direct contact with the protein residues. In the case of formaldehyde, a high preference for ARG, LYS and GLN residues is found. Moving to gallic acid, the analysis suggests that both the phenolic OH groups and the carboxylic group of gallic acid molecules approach HPR, ARG, GLN, and LYS residues at hydrogen bonding distance and are part of a well-defined first solvation shell coordinated with these

8 amino acids side chain groups. These results are in agreement with experimental studies suggesting that HPR residues are among the preferred polyphenols binding sites and strong interactions also occur with ARG and GLU side chains. In Figure 5, a snapshot taken from the MD trajectory shows a gallic acid molecules bound to AA side chains of CMS. The examination of average interaction energies per residue and their van der Waals and electrostatic components confirm the results obtained with the structural analysis. Furthermore both van der Waals and electrostatic terms are important for the stabilization of the CMS: this is especially noted when CMS interacts with gallic acid. Figure 5. Snapshot taken from the simulation of CMS with gallic acid. Conclusions The results reported here, even if preliminary to a certain extent, on the either side provide a convincing evidence of the potentiality of the molecular dynamics simulation for the understanding of complex supramolecular systems. This in particular holds for the subject system based on collagen submitted to the treatment with chemicals which change its environment and can react with various functional groups present in the macromolecules. Such approach is of great help in investigating any system where collagen is involved, but in particular in the understanding of the tanning process from both scientific and applied view points. Indeed the tanning reaction is presently carried out by several different methodologies, all characterized by a substantially empirical approach where technologies and personal natural capacity are intimately mixed. No doubt that this approach gave excellent practical results and high quality leather was obtained for a variety of application with a great impact in our life and economy. Under the constant pressure deriving from the increasing environmental sensitivity of the Society and of the public administration a rationalization of the Leather Industry is absolutely necessary in order to maintain sustainability and competitive trade capacity. In this connection several advanced research groups in different countries performed extended experimental research activity also with

9 the support of the European Community [9]. Here it is not possible to discuss all these approaches and as a typical and high qualified example it is worth mentioning the contribution by Covington and co-workers [10], concerning detailed and very advanced fundamental studies on the mechanism of chrome tanning. Certainly the results obtained il all these studies provided a significant contribution to expand our knowledge on the topic, but at the same type evidenced once again how difficult and time consuming is to perform reliable scientific experiments with such complicated molecular mix as that present in the tanning reactor. [11]. In order to grant the technological evolution necessary to make the tanning process a sustainable and commercially competitive process, which can encounter the requirements of the modern industry, a deeper knowledge at molecular level is absolutely necessary This can be more conveniently approached if the expensive and complex experimental research is assisted by molecular modelling, which can also provide useful directions to design products with the desired properties. Acknowledgments This work has been partially funded by the European Project CRAFT Novel Approaches to Optimizing Penetration during Leather Manufacture (CHEMPEN). The INFM-Iniziativa Trasversale Calcolo Parallelo 2004 at Cineca supercomputer center is also acknowledged. References [1] Smith, J. W. Nature 1968, 219, [2] Covington, A. D. Chem. Soc. Rev. 1997, 26, [3] (a) King, G.; Brown, E. M.; Chen, J. M. Protein Eng. 1996, 9, (b) Brown, E. M.; Dudley, R. L.; Elsetinow, A. R. J. Am. Leather Chem. Assoc.1997, 62, (b) Qi, P. X.; Brown, E. M. J. Am. Leather Chem. Assoc. 2002, 97, [4] Jorgensen, W. L. J. Am. Chem. Soc. 1981, 103, [5] Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E., III; Wang, J.; Ross, W. S.; Simmerling, C. L.; Darden, T. A.; Merz, K. M.; Stanton, R. V.; Cheng, A. L.; Vincent, J. J.; Crouley, M.; Tsui, V.; Gohlke, H.; Radmer, R. J.; Duan, Y.; Pitera, J.; Massova, I.; Seibel, G. L.; Singh, U. C.; Weiner, P. K.; Kollman, P. A. AMBER 7; University of California: San Francisco, [6] (31) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M., Jr.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995, 117, [7] Bronco, S.; Cappelli, C.; Monti, S. Understanding the Structural and Binding Properties of Collagen: A Theoretical Perspective ; J. Phys. Chem. B 2004, 108, [8] Monti, S.; Bronco, S.; Cappelli, C. Towards the Supramolecular Structure of Collagen: A Molecular Dynamics Approach submitted. [9] GROWTH Project Radical Environmentally Sustainable Tannery Operation by Resource Management (RESTORM), GRD [10] Covington, A. D.; Menderes, O.; Brown, E. M.; Collins, M. J.; O Duwole, A. XXVI IULTCS, Cape Town, [11] Ciardelli, F.; Bronco, S.; Costantini, N.; Lazzaroni, R.; Taburoni, E.; Ghimenti, G.; Lombardi, G. Cuoio Pelli Materie Concianti 1999, LXXV n. 3, 109.