Molecular modelling of leather related collagen structures

Tanning is the stabilization of leather against water, swelling and rot1. There are several ways to tan hides but of greatest significance today by far is chrome tanning which uses salts of Chromium III and was discovered in the mid 19th century. Alternatively there is vegetable tanning which is composed of large polyphenolic molecules found in the bark and roots of  certain plants and has been known for thousands of years. Research over the past 100 years has brought great strides forward in the understanding of all aspects of leather production. However, there is still a lack of detailed understanding at the molecular level of tanning and the interaction of leather processing chemicals with the leather, ie with collagen molecules.  Such an understanding could greatly contribute to a more rational design and development of chemicals for all aspects of leather chemistry.  One goal could be the replacement of chromium salts with environmentally friendlier molecules while retaining their performance advantages.  Leather is an incredible material: it is tough, abrasion resistant, waterproof yet breathes, and more.  It owes these attributes to its hierarchical structure (Figure 1) in that substructure motifs are combined to form larger structures with different physical and mechanical properties.  For example: tropocollagen strands coalesce to micro fibrils and these to collagen fibres and these in turn to leather. The bulk structure of leather and the collagen fibres has been well studied by optical and electron microscopy down to the nanometer scale. Also the interaction of many leather chemicals and finishing products can be studied and optimized at this level of detail, eg surface finishes. In order to gain an understanding of the structure of collagen at the molecular level several techniques have played crucial roles: high resolution microscopy (TEM, SEM, AFM), x-ray crystallography, protein sequencing and theoretical methods. Type-1 collagen consists of  linear triple helix strands in a ‘coiled-coil' geometry analogous to a three stranded rope with the three helically coiled amino acid chains (Figure 1)2.  Type-1 collagen is circa 1,000 amino acids long and consists of GLY-X-Y triplets, ie every third amino acid is glycine. This is important so that the repeating helix structure can exist. The other two amino acids (X and Y) in the triplets consist of circa 10% proline, circa 10% hydroxyproline, circa 10% alanine and the rest a distribution of the other naturally occurring amino acids. The end regions of the collagen molecule consist of short sections of less ordered telopeptides. The role of the hydroxyproline (formed from proline and the co-factor ascorbic acid) is to stabilise the fibrils via interstrand hydrogen bonding interactions rather than intrafibril stabilisation. The proline provides the chain with stiffness but not interstrand stability. The tropocollagen molecules are offset to their neighbours along the axis of the fibril. This is seen as regular bands in the electron micrographs.  In 2006 Orgel et al3,4  published a crystal structure of the complete collagen type-1 fibril, including the telopeptide region. The primary amino acid sequences of the different types of collagen are known and deposited in sequence databases such as SwissProt or PIR5. Individual tropocollagen molecules are too long to currently simulate in total. However, short segments have been simulated beginning in the 1980s6-19. The emphasis was on the structure of the model tropocollagen strands and their interaction with one another via side chain functionality. The possible interactions of acidic side chains with chromium clusters were also briefly examined, based on geometric arguments and forcefield methods.  These studies provided the ground work for understanding the intra and interstrand interactions such as hydrogen bonding and the role of tightly bound water. The role of hydroxyproline has been shown to be interstrand stabilisation, ie fibril stabilisation rather than tropocollagen stabilisation.  Leather chemists are interested in the interaction of process chemicals, tanning agents, pickling salts, dyes etc, with collagen. Therefore, in order for theoretical models to have relevance for lab chemists the models must be extended.  It is evident that with increasing computing power and more efficient algorithms that the size and complexity of the systems that can be studied with molecular dynamics and quantum chemical approaches is rapidly expanding. BASF became interested in leather simulations in 2003.  At that time they set out to develop the most realistic and flexible model available at the time. The models developed since then are periodic in all three dimensions, take water into account as well as salts and additives (the model and eight images are shown in Figure 2). In 2004 they initially reported on molecular dynamics simulations on the initial models at pH 1.7 and 14 with the appropriate protonation and counterions to ensure charge neutrality21. To validate their model they ran molecular dynamics simulations at constant pressure and temperature (300K, 1 bar) to check the equilibrium density of the neutral molecule.  The simulation box had a density of 1.44 g/cm3 which is in close agreement with a -variable- experimental value of ca. 1.35 g cm3. A further validation of the model came from simulations of swelling at the different pHs and pickling with various salts (NaCl, CaCl2, Na2SO4, as well as urea (1% and 12% w/w)22. Results show that addition of salt prevents swelling at very low and high pH to varying degrees, just as one would expect in the lab.  The strongest interaction was with sulfate rather than chloride. This is in accordance with the Hofmeister series for the interaction of salts with proteins (F-, SO42- > HPO42- > acetate > Cl- > NO3- > Br- > ClO3-)23. Urea was found to concentrate at the collagen surface during the course of the simulation. However, the simulation time of 2ns was not sufficient to see the beginnings of denaturation with the 12% solution that would be expected and much longer simulations would be required. This question of simulation time of large systems with many degrees of freedom is of general interest. Conformational changes and large translocations etc, generally take place at time scales that are not accessible to molecular dynamics, with its one femto-second step size.  Fortunately, methods are being developed to overcome some of the shortcomings.  In the group of Parrinello the so called ‘metadynamics' method has been implemented to force molecules to cover the desired phase space and one can extract the relative free energies for the process under investigation24-25. In the case of collagen simulations it is possible with the use of metadynamics to simulate polymeric retanning agents.   The authors examined the effect of adding polyacrylic acid (PAA) and poly methyl acrylic acid (PMAA) as 200mer polymers to the 34 TH fibril model at pH7. PMAA is a retanning agent and PAA does not work in this application. PMAA is more hydrophobic and stiffer with a larger radius of gyration than PAA. Otherwise they are structurally quite similar and the question was whether or not a difference could be detected and explain the difference in tanning properties.  Their simulations indicate that PMAA forms a better buffer between the collagen microfibrils, due to stiffness and less hydrogen bonding, whereas PAA forms many, relatively strong hydrogen bonds with the fibrils which could allow the fibrils to come into close proximity forming a parchment-like structure. More simulations are needed to confirm these results. Even with the use of metadynamics there is still a lot of phase space that would need to be covered before these questions can be definitively answered. The good news is that computing power is continuously going up and the price coming down. Also the software is being optimised for highly efficient parallel  processing.  These facts combine to make clusters of several hundred linux based PCs running efficient academic code such as NAMD financially accessible to even small organizations interested in extending their knowledge of these highly complex systems. Authors Lorenz Siggel1, Rosa E Bulo2, Ferenc Molnar1, Horst Weiss1 and Tilman Taeger1 1) BASF Aktiengesellschaft, Ludwigshafen, Germany 2) Department of Chemistry, Free University of Amsterdam, The Netherlands