Interactions between various biological macromolecules control protein folding and assembly, DNA packing, protein-DNA interactions, tissue formation and stability, and many other processes . The Unit on Molecular Forces and Structure seeks to advance our understanding of these most basic molecular recognition reactions, with particular emphasis on the pathology of collagen fiber formation in brittle bone disease and other bone and connective tissue disorders. In the past year, we discovered that collagen monomers are thermally unstable at body temperature. For instance, at 37°C, human lung collagen completely denatures within two to three days. Initial melting of most thermally labile domains of collagen secreted by cells appears to trigger formation of fibers in which molecules are protected from further melting. In addition to their other effects, collagen mutations affect the melting temperature and thus disrupt proper timing of fiber formation, resulting in pathology. In particular, we described fibrillogenesis of several mutant collagens with different changes in melting temperature. Second, in collaboration with researchers at the Hereditary Disorders Branch of NICHD, we continued to characterize pathology caused by several mutations observed in human patients and by a mutation that was introduced in a murine brittle-bone-disease model. These studies are beginning to reveal some of the molecular origins of the symptoms associated with the bone disorder. Third, in collaboration with scientists at the Research Center Jülich (Jülich, Germany), we developed a theory relating the microscopic physics of helix-helix interaction to macroscopic properties of cholesteric assemblies observed in aggregates of DNA and collagen. Finally, we continued our study of the structuring of water at surfaces of biological macromolecules and the role of structured water in DNA-DNA and collagen-collagen interactions. The goal of the Unit on Molecular Forces and Structure is to undertake studies of the physical principles of interactions involving collagen and DNA that will lead to a better understanding of corresponding processes responsible for normal function and pathology in living organisms and to the development of new treatments for associated diseases.

Thermal Stability, Metabolism, and Function of Type I Collagen
Mertts, Kuznetsova, Mertz, Leikin in collaboration with E. Leikina (LCMB, NICHD)
Type I collagen is the most abundant animal protein; it forms the matrix of bone, skin, and other tissues. It was taken for granted that type I collagen triple helix melts several degrees above body temperature in all species from Arctic fish to humans. However, the collagen helix is less stable. We demonstrated that monomeric collagen already melts, albeit slowly, several degrees below body temperature. Its equilibrium state at body temperature is a random coil rather than a triple helix. We now argue that cells have to use molecular chaperones to fold this otherwise unstable protein. Once secreted from cells and processed by enzymes, collagen helices begin to unfold. Our preliminary data indicate that micro-unfolding of least stable collagen domains triggers self-assembly of fibers in which helices are protected from complete unfolding.

The melting time of monomeric type I collagen at body temperature varies between several hours (rats) to several days (humans); it is comparable to the time from collagen folding to its incorporation into fibers and may be one of the reasons why a large fraction of newly synthesized collagen is degraded and never achieves the state of a mature fiber. This is clearly an important factor in the control of synthesis, secretion, fibrillogenesis, and degradation of mature collagen. This property may also allow molecules confined in fibers to melt and refold locally, giving fibers their great combination of strength and elasticity. Apparently, nature adjusts collagen hydroxyproline content to ensure that the melting temperature of monomers is several degrees below rather than above body temperature.

We are now beginning to look more closely at possible molecular mechanisms relating changes in collagen thermal stability to connective tissue pathology caused by mutations. For instance, a decrease in the melting temperature by about 1°C reduces the unfolding time at body temperature from several days to several hours. As a result, the fine balance between the kinetics of fiber formation and the kinetics of collagen melting may be disrupted such that fibers may improperly form. Understanding the regulation of these processes may help us find treatments for a variety of connective tissue disorders.

Physical Biochemistry of G349C Collagen
Kuznetsova, Leikin in collaboration with A. Forlino and J.C. Marini (HDB, NICHD)
A substitution of almost any glycine in the (Gly-X-Y)n sequence of the collagen triple helix by any other amino acid produces brittle bones (osteogenesis imperfecta or OI). Mutations at some glycine positions are lethal while mutations at other positions can be relatively mild. The phenotype can vary from moderate to severe or lethal even between people with the same mutation and the same expression of the mutant protein. The recent development of the first true murine OI model that exhibits typical human-like symptoms and phenotype variability was based on a "knock-in" Gly349®Cys mutation. We are now investigating how the mutation affects the structure, stability, and interaction between collagen helices in different tissues.

Our measurements revealed that the mutation on its own affects neither the thermal stability of the triple helix nor collagen-collagen forces. Similarly, a popular model relating disease to formation of a kink in the collagen triple helix does not work for the G349Cmutation. We observed subtle changes in the protein stability and in vitro fiber formation that appear to be related to slight posttranslational overmodification of collagen caused by the mutation. Preliminary data indicate that the overmodification is more pronounced in bones than in skin and tendons. It may affect the phenotype but it does not seem to be the main factor controlling severity of the disease. We are therefore gradually shifting our focus to investigating the interaction of collagen with other matrix components, e.g., matrix proteoglycans.

Chain Register Disruption in Osteogenesis Imperfecta
Mertts, Leikin in collaboration with W. Cabral and J.C. Marini (HDB, NICHD)
We investigated the effect of shifting the register of the collagen helix by a single Gly-X-Y triplet on collagen assembly, stability, and incorporation into fibrils. The mutation causes lethal type II OI. While the normal allele encodes two identical Gly-Ala-Hyp triplets at amino acids 868-874 in a1 chain of type I collagen, the mutant allele encodes three. The shift in register delays helix formation and causes full-length overmodification. Differential scanning calorimetry and in vitro fibrillogenesis revealed a severe effect of the mutation on the stability and interaction between protein molecules. Molecules containing two mutant a1 chains melt at about 5°C below the denaturation point of the wild-type protein. Instead of being incorporated into fibers during in vitro fibrillogenesis, the molecules rapidly denature. Molecules containing only one mutant a1 chain are not as badly damaged. Their melting temperature is about 2°C lower than that of the wild-type protein, and they are incorporated into fibers. Nevertheless, their presence leads to significant inhibition of fiber formation. The rate and extent of fibrillogenesis of the wild-type protein are achieved only at four-fold protein concentrations. Gel electrophoresis of whole chains and their CNBr peptides clearly indicates posttranslational overmodification of the protein, but we do not know yet what (if any) role overmodification plays in the observed changes in the stability and function of the protein. The mutation results in slower cleavage of N-propeptides from pro-collagen even though the cleavage occurs 900 amino acids away from the mutation site. Preliminary data indicate that the change in the cleavage rate is not related to posttranslational overmodification of the protein. If true, the results suggest that the register shift persists, rather than relaxes, through the entire length of the collagen helix, e.g., by formation of a short loop. The mutation is under further investigation to learn more about the folding and function of normal collagen.

Electrostatic Chiral Interactions and Cholesteric Liquid Crystals of Biological Helices
Leikin in collaboration with S.V. Malinin and A.A. Kornyshev (Research Center Jülich, Jülich, Germany) and L.I. Krishtallik (Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Moscow, Russia)
Many biological macromolecules, including collagen and DNA, form chiral, liquid-crystalline aggregates. The aggregates consist of layers of parallel molecules. The orientation of molecules in each next layer is twisted by a constant small angle with respect to the previous layer. Such "cholesteric phases" are observed both in vitro and in vivo. They have unusual optical and mechanical properties that may be important for the function of various structures in living organisms. Cholesteric liquid crystals were discovered about 100 years ago, but a molecular theory relating their properties to intermolecular interactions still does not exist.

As a first step toward the formulation of such a theory, we previously derived expressions for electrostatic and hydration forces between idealized, infinitely long helices at all interaxial angles, and taking realistic, helical patterns of charges on molecular surfaces into account. The results suggested possible a interpretation of the surprisingly small value of the cholesteric twist angle but also revealed that the twist angle and other observable features of cholesterics may be directly related to molecular length. In the past year, we extended the theory and derived general expressions for the interaction between helices of finite length. We determined the balance of forces within a simplified cholesteric "unit cell" (a triad of two parallel molecules and a third molecule twisted with respect to them). From the corresponding local free energy density we evaluated the equilibrium cholesteric twist and the twist elastic constant. Good agreement with experimental data for DNA cholesterics suggests that our estimates may be a reasonable starting point for a more detailed statistical theory of such liquid crystals.

Water Structuring at Biomolecular Surfaces
Mertz, Leikin in collaboration with L.I. Krishtallik (Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Moscow, Russia)
Virtually all biomolecular recognition reactions, including assembly of collagen fibers, occur in water. It has been long recognized that water may be an active participant in these reactions rather than just a passive medium. Although a few possible roles of water in several specific reactions have been recently clarified, many phenomena remain unexplained. One of the poorly understood issues is the dielectric response of water at biomolecular surfaces to the electric field of charged surface groups. We previously reported a dielectric anomaly of such surface water. Our recent, more detailed data suggest that the anomaly may be a general feature of all hydrophobic interfaces. It is apparently related to the formation of highly labile, structured water clusters at such interfaces. The clusters may exhibit a resonant dielectric response to an electric field rapidly varying in space and resulting in dielectric overscreening.

Structured water clusters at protein surfaces may not only affect dielectric response, but they may also directly participate in protein-protein interactions. For instance, our preliminary dichroic Fourier Transformed Infra-Red (FTIR) data for fully hydrated rat tail tendon indicate that the dipoles of most water molecules are perpendicular to the collagen fibril axis and that the spectral properties of the water molecules are substantially different from those of water molecules with dipoles parallel to the axis. Our previous Raman and force measurement studies suggested that the energetic cost of restructuring and removing such structured water clusters upon dehydration of collagen fibers is the main energetic factor determining both specific and nonspecific collagen-collagen interactions. Substantial changes in local hydration of collagen triple helices can be induced by disease-causing mutations, such as substitutions of obligate glycine residues, as suggested by other authors. We thus anticipate that better understanding of the specific role played by water in collagen-collagen and other recognition reactions may enable us to devise new approaches to the study and treatment of molecular pathology.