Search

The precise molecular mechanisms that determine the three-dimensional architectures of tissues remain largely unknown. Within tissues rich in extracellular matrix, collagen fibrils are frequently arranged in a tissue-specific manner, as in certain liquid crystals. For example, the continuous twist between fibrils in compact bone osteons resembles a cholesteric mesophase, while in tendon, the regular, planar undulation, or “crimp”, is akin to a precholesteric mesophase. Such analogies suggest that liquid crystalline organisation plays a role in the determination of tissue form, but it is hard to see how insoluble fibrils could spontaneously and specifically rearrange in this way. Collagen molecules, in dilute acid solution, are known to form nematic, precholesteric and cholesteric phases, but the relevance to physiological assembly mechanisms is unclear. In vivo , fibrillar collagens are synthesised in soluble precursor form, procollagens, with terminal propeptide extensions. Here, we show, by polarized light microscopy of highly concentrated (5–30 mg/ml) viscous drops, that procollagen molecules in physiological buffer conditions can also develop long-range nematic and precholesteric liquid crystalline ordering extending over 100 μm 2 domains, while remaining in true solution. These observations suggest the novel concept that supra-fibrillar tissue architecture is determined by the ability of soluble precursor molecules to form liquid crystalline arrays, prior to fibril assembly.

International audience; The solubility limit of purified chick type I procollagen, incubated at 37 degrees C in phosphate-buffered saline, was found to be in the range 1 to 1.5 mg/ml. At higher concentrations large aggregates formed. These comprised: (1) D-periodic assemblies; (2) narrow filaments with no apparent periodicity; and (3) segment-long-spacing-like aggregates. The D-periodic assemblies, which predominated at high concentrations, were separated from the other types of aggregate and found to be ribbon-like. Ribbons were uniform in thickness (approximately 8 nm) and up to 1 micron wide. Staining patterns showed features similar to those in native-type collagen fibrils. Immunolabelling indicated that the carboxyl-terminal propeptide domains were close to the carboxyl-terminal gap-overlap junction, and that the amino-terminal propeptide domains were folded over into the amino-terminal side of the overlap zone. Both propeptide domains appeared to be located on the surface of the assemblies. These observations show that intact propeptide domains hinder, but do not prevent, the formation of D-periodic assemblies. The presence of the propeptide domains on the surface of a growing assembly could restrict its lateral growth and limit its final thickness.The solubility limit of purified chick type I procollagen, incubated at 37 degrees C in phosphate-buffered saline, was found to be in the range 1 to 1.5 mg/ml. At higher concentrations large aggregates formed. These comprised: (1) D-periodic assemblies; (2) narrow filaments with no apparent periodicity; and (3) segment-long-spacing-like aggregates. The D-periodic assemblies, which predominated at high concentrations, were separated from the other types of aggregate and found to be ribbon-like. Ribbons were uniform in thickness (approximately 8 nm) and up to 1 micron wide. Staining patterns showed features similar to those in native-type collagen fibrils. Immunolabelling indicated that the carboxyl-terminal propeptide domains were close to the carboxyl-terminal gap-overlap junction, and that the amino-terminal propeptide domains were folded over into the amino-terminal side of the overlap zone. Both propeptide domains appeared to be located on the surface of the assemblies. These observations show that intact propeptide domains hinder, but do not prevent, the formation of D-periodic assemblies. The presence of the propeptide domains on the surface of a growing assembly could restrict its lateral growth and limit its final thickness.

We have examined the state of aggregation of type I procollagen in the concentration range 5 to 800 micrograms/ml. Electron microscopy typically indicates a high proportion of aggregated material (greater than 50%), when a range of preparative techniques are used. Aggregates of in-register molecules (segment-long-spacing-like aggregates) are frequently observed, often with units of in-register molecules connected via the C-terminal propeptides. In contrast, studies using gel-filtration chromatography and density-gradient ultracentrifugation demonstrate only limited aggregation in solution (less than 5%) even at 800 micrograms/ml. The aggregated material is mainly dimeric and probably not segment-long-spacing-like. We conclude that aggregation of procollagen is strongly favoured by adsorption to a surface when samples are prepared for electron microscopy. The possible relevance of these observations to the fate of procollagen secreted by cells in vivo is discussed.

A brief review is given of attempts to interpret the low-angle meridional X-ray diffraction patterns of tendon to yield the axially projected electron density of collagen fibrils. Both X-ray/electron microscope correlations and more conventional crystallographic methods have been used. For wet tendon, there have been two fundamental assumptions; either that those parts of the fibril which are positively stained in electron micrographs are intrinsically electron dense, or that the electron density is centrosymmetric. The validity of these assumptions is tested by synthesising models for the electron density based on the known amino acid sequence of collagen and the Hodge-Petruska scheme. Allowance is made for the effect of the water environment. These models are then Fourier transformed to give a set of calculated intensities for comparison with the observed low-angle meridional X-ray data. The calculated intensities are very sensitive to the conformation of the non-triple-helical terminal peptides (telopeptides) and to the D period. Difference Fourier syntheses are used to refine a model which gives the best agreement with the observed data whilst being both physically and chemically acceptable. The calculated phases of this model (model E) are then combined with the observed amplitudes to give a Fourier synthesis of the axially projected electron density in wet collagen fibrils. The implications of this result for the study of various metabolic and pathological situations in connective tissue are discussed.

The amino acid sequence in the triplet region of the α1 chain of collagen was analyzed for complementary relationships that would explain the stagger of multiples of 670 A between the rod-like molecules in the fibril. The analysis was done by moving the sequence of 1011 amino acids past itself and scoring for complementarity between opposing amino acids allowing a range of ±2 to 3 residues. It was found that interactions between amino acids of opposite charge and between large hydrophobic amino acids in the overlapping region between two chains are maximal when the chains are staggered by 0D, 1D, 2D, 3D and 4D, where D = 234 ± 1 residues. The residue repeat derived from this value is 2.86 ± 0.02 A. The existence of a D separation between interacting residues was shown to be reflected in the actual distribution of large hydrophobic amino acids. Surprisingly, the distribution approximates the pattern ( 2D 11 )5( D 11 ) repeated over 4.4D intervals. The regularity may arise from structural constraints imposed by super-coiling. The distribution of charged residues is less regular and does not show a well-defined periodicity. However, positively-charged residues tend to be near negatively-charged residues, allowing intramolecular charge neutralization as well as strong intermolecular charge interactions at 0D.