**2. Experimental**

264 The Complex World of Polysaccharides

specific.

bases of NA.

biopolymers.

greater portion of N-acetylated and N-sulfated disaccharides (Lindahl U., et al. 1987). In fact, this finding demonstrates that synthesis of the glycoside moiety of proteoglycans is genetically determined *in vivo*, because their primary structure is tissue- and species-

A basic unit monomers of polysaccharides is glucose, which occurs in solution both in the α and in the β form owing to the mutarotation reaction. In addition, glucose can be epimerized to other hexoses and be oxidized to yield glucuronic and other hexuronic acids. Owing to such lability of glucose, biosynthetic systems always contain sufficient amounts of the α and β forms of hexoses and hexuronic acids. Modified with UDP at C1, these monosaccharides provide the main components for synthesis of polysaccharide chains.

As Fig. 1 and Table 3 demonstrate, the majority of known structural and storage polysaccharides each consist of two monomers, a hexose and hexuronic acid, which occur in the α or β form, are linked through (1-3), (1-4) O-glycoside bonds, and are modified to a various extent at various carbon atoms as a result of acetylation, amination, sulfation, etc. All these biopolymers can be combined in one group with a universal structure of multiply

We think that the periodicity of the primary structure is similar between polysaccharides and DNA tandem repeats. In view of this structural similarity, it was justified and important to study a possible complementarity between monosaccharides of glycans and

To check the hypothesis of complementarity of NA bases to hexoses and hexuronic acids, quantum chemical methods were used in our lab for a particular case of glucose and glucuronic acid contained in the heteropolysaccharide HA. It is clear that such an approach is computational and that the relevant conclusions need experimental verification. In view of this, it was necessary to obtain empirical information supporting or contradicting the results of quantum chemical computations. For this purpose, we employed UV spectrophotometry and dot hybridization, which allow detection of specific complexes of

Since the mechanism initiating GAG biosynthesis is unclear, it was expedient to study glycan synthesis in the rat liver upon administration of elevated doses of glucose. The use of 35SO42- as a radioactive label is inadequate for studying GAG synthesis, and we decided to label a precursor of the glycan polysaccharide chain. We used glucose as such a precursor: glucose is transformed into UDP-glucose and then into UDP-glucuronic acid, which is utilized in GAG synthesis. It should be noted that ribose 5-phosphate, which is formed from glucose 6-phosphate, is incorporated in NA. Glucose is converted into ribose in the pentose phosphate cycle by eliminating C1, which is released as carbon dioxide. Hence, we used [14С]glucose labeled at C1 to prevent generation of radiolabeled NA in our experiments. We studied the composition of rat liver polysaccharides in cell nuclei, microsomes, and in a total liver homogenate. In addition, the nuclear and microsomal fractions were used to monitor

the accumulation of radiolabeled saccharides, which are polysaccharide precursors.

repeated (A-B), (A-A), or (B-B) units, where A is a hexose and B is a hexuronic acid.

Quantum chemical modeling of biological structures, in particular, the geometric and electron structures of NA and polysaccharides, requires that the methods used report adequately the effects of weak intermolecular interactions, such as hydrogen bonds. The MP2 *(ab initio*) procedure, which utilizes bases with diffuse and polarization functions, meets this requirement quite well (Cybulski S.M., et al. 1989, Latajka Z. et al. 1990) but is hardly suitable for our objects because of their size: their analysis would be extremely timeconsuming and requiring excessive computational resources. A possible alternative in this situation is provided by the corresponding semiempirical methods. Early semiempirical methods (MINDO/3 and MNDO) considerably underestimated the energy of hydrogen bonds and, consequently, were unsuitable for studying biopolymers (Williams I.H. 1987). To eliminate such drawbacks, the AM1 (Dewar M.J.S., et al. 1985) and PM3 (Stewart J.J.P. 1989) methods were developed on the basis of neglect of diatomic differential overlap (NDDO). These methods were expected to adequately describe systems with hydrogen bonds. This was not the case with AM1: while computed energies of hydrogen bonds agreed well with experimental estimates, geometric parameters failed to represent the facts (Jurema J.M.W., et al. 1993). Parametrization was performed with a far greater body of experimental data in PM3 than in other semiempirical methods, which allowed PM3 to describe well the geometric structure of molecules and the heat of their generation. PM3 is indeed the first method yielding semiempirical estimates that agree with the results of experiments and *ab initio* calculations for hydrogen-bonded systems (Kallies B., et al.1995).

## **2.1. Method of calculation** *Ab initio*

*Ab initio* calculations are the main computational procedure in quantum chemistry and consists in solving Hartree-Fock one-electron equations. As initial data, the method utilizes the charges of nuclei, their positions in the molecule, and Slater- or Gaussian-type basis function sets. The method involves none of the observed physico-chemical properties of substances and, accordingly, is known as unempirical calculations. *Ab initio* calculations most commonly employ the MO LCAO (molecular orbitals as a linear combination of atomic orbitals) approximation, which takes account of all electrons of the system in question. This method is most accurate in quantum chemical computations, especially with intermediate (6-31G\*) and large (6-31\*\*) bases, and allows correct estimation of the electron and geometric structures of hydrogen bonds, which play an important role in biological objects.

Both unempirical *ab initio* calculations with basis 6-31G\* and the PM3 semiempirical method were employed in theoretical computations in this work. Geometric parameters and energy characteristics were computed by the unempirical and semiempirical methods for interacting nucleotides and saccharides and by the semiempirical methods for structures containing several pairs of nucleotides and saccharides. All computations were performed with complete optimization of geometrical parameters, using the GAMESS program (Schmidt M.W., et al. 1993). A global minimum of the total electron energy was sought by

the Newton-Raphson method with an energy gradient of 0.010 kcal/molÅ, starting from various initial approximations of the complex structure.
