The interaction of human plasma glycoaminoglycans with plasma lipoproteins.

Abstract Modifications of existing methods have allowed for the isolation and purification of various species of plasma glycosaminoglycans on the basis of their sulfate content and molecular size. All of the preparations precipitated human plasma low density lipoproteins (LDL); maximal precipitation occurred with amounts of glycans corresponding to 50 mug of hexuronate and 12 mg of LDL. The interaction of glycans with pyrene-labeled lipoproteins was also studied, measuring variations of the fluorescence emitted by the monomer (M) and excimer (E) species of the bound pyrene. The ratio IE/IM is proportional to c/eta, where c is the microscopic concentration of the pyrene confined to the hydrocarbon region of the lipoprotein and eta is the microviscosity of that region. To 0.12 mg of pyrene-labeled LDL, very low density lipoproteins (VLDL) or high density lipoproteins (HDL) were added increasing amounts of the various glycan preparations. The sulfate-rich species decreased the IE/IM ratio of LDL and HDL but not that of VLDL. This finding suggests that the glycan caused a change in lipoprotein conformation associated with either an increased volume or increased microscopic viscosity of the hydrocarbon region. The modification of LDL conformation could be prevented by proteolytic treatment of the sulfate-rich species or by addition to the system of suitable amounts of sulfate-poor species or of chrondroitin-4-sulfate, but could not be prevented by increased ionic concentration. These results suggest that the two main species of plasma glycans are important in maintaining adequate rheological properties of plasma lipoproteins.

of existing methods have allowed for the isolation and purification of various species of plasma glycosaminoglycans on the basis of their sulfate content and molecular size. All of the preparations precipitated human plasma low density lipoproteins (LDL); maximal precipitation occurred with amounts of glycans corresponding to 50 pg of hexuronate and 12 mg of LDL. The interaction of glycans with pyrene-labeled lipoproteins was also studied, measuring variations of the fluorescence emitted by the monomer (I@) and excimer (E) species of the bound pyrene. The ratio IEIIM is proportional to c/v, where c is the microscopic concentration of the pyrene confined to the hydrocarbon region of the lipoprotein and 11 is the microviscosity of that region. To 0.12 mg of pyrene-labeled LDL, very low density lipoproteins (VLDL) or high density lipoproteins (HDL) were added increasing amounts of the various glycan preparations.
The sulfate-rich species decreased the ZE/Znr ratio of LDL and HDL but not that of VLDL. This finding suggests that the glycan caused a change in lipoprotein conformation associated with either an increased volume or increased microscopic viscosity of the hydrocarbon region. The modification of LDL conformation could be prevented by proteolytic treatment of the sulfate-rich species or by addition to the system of suitable amounts of sulfate-poor species or of chondroitin-4-sulfate, but could not be prevented by increased ionic concentration.
These results suggest that the two main species of plasma glycans are important in maintaining adequate rheological properties of plasma lipoproteins.
The interaction between plasma lipoproteins and macromolecular polyanions has been studied by many authors with a variety of ingenious techniques (l-lo).
The presence of glycosaminoglycans in various cellular elements and extracellular spaces of the arterial wall, their postulated role in the structure and function of the vascular walls, their physicochemical properties, and their presence in blood (11-16) have prompted detailed studies of * This investigation was supported by Health, Education and Welfare Grants HL-14194.04 and HL-05435.14, and by a grant from The Robert A. Welch Foundation.
$ Established Investigator of the American IIeart Association.
their interaction with lipoproteins. III view of the determinant role ascribed to lipoproteins, and in particular, to the low densit) lipoproteins, in the pathogen&s of early atherosclerotic lesions (17-19) the relevance of their interactions with glgcosaminoglycans has been the object of protracted speculations (9,10,20,21).
In the present report, we have studied the interaction of plasma glycosaminoglycans with very low density lipoproteins, low density lipoproteins, and high density lipoproteins. The low levels of circulating glycosaminoglycans dictated that meaningful experiments be carried out with minimal amounts of them and with physiologically corresponding quantities of lipoproteins. Our previous studies (22) have demonstrated that plasma glycosaminoglycans consist essentially of chondroitin-4-sulfate, which may be separated into two different types on the basis of ester sulfate content: those capable of binding as such to ECTEOLA1 anion exchange-modified cellulose (operationally defined as "free" glycans, and having an average molar ratio 0.88, sulfate to hexosamine), and those able to do so only after treatment with proteolytic enzymes (defined as "bound" glycans, and having an average molar ratio of 0.48, sulfate to hexosamine).
For this study, it was necessary: (a) to harvest sufficient amounts of both types; (b) to isolate bound glycans without the use of proteolytic enzymes; and (c) to use methods capable of detecting their interaction with lipoproteins at levels lower than those forming precipitates.

Labeling of Lipoproteins with Pyrene
Immediately before use, lipoproteins were labeled with pyrene at room temperature with one of the following methods.
In the first (35), which is usefld for labeling LDL but not VLDL or HDL, 5 g of Bio-Glass 500 beads were stirred into 50ml of chloroform containing 4 mg of pyrene. The solvent was removed by flash evaporation at room temperature, and 0.5 g of the dry beads was used to prepare a column to which was applied 5 ml of LDL solution  (19).

Isolation and Purijication of Plasma
Glycosaminoglycans-Free and bound plasma glycans were isolated and purified as outlined in Scheme 1. The free glycans were eluted from Bio-Gel P-6 ( Fig. 1) in two broad peaks: Fraction I, emerging at the void volume (high molecular weight free), and Fraction II, less sharply defined, being retained (low molecular weight free). Rcchromatography of the latter on the same column (Fig. 1, inset) yielded an asymmetrical peak, suggesting various molecular species. Both peaks were utilized in our experiments. The bound glycans were eluted from Bio-Gel P-6 ( Fig. 2) in two peaks, a very small one, which eluted at the void volume and was discarded, and a major one, which was retained in the gel, and was subsequently utilized.  Fig. 1 for details. ble II shows that preparations of plasma glycans gave optimal precipitation of LDL at a concentration of 50 pg (as hexuronate) ; however, none of them was as effective as sodium heparinate.
Using pyrene-labeled lipoproteins, two fluorescence maxima were observed: the monomer at 390 nm, and the excimer at 470 nm. Fig. 3A shows the fluorescence of increasing amounts of 4% pyrene-labeled LDL; Fig. 3, B and C shows the spectra of 0.12 mg of VLDL and HDL labeled with 2% and 4y0 pyrene, respectively.
The ratio of IE/IM fluorescence of 4% pyrene-labeled LDL incubated with standard preparations of glycosaminoglycans with PPL and with free and bound glycans is shown in Fig. 4. While the standard preparations, including heparin and PPL, did not significantly change the IE/IM ratio of LDL, a substantial decrease was observed (Fig. 4B) when LDL was incubated with high or low molecular weight free glycans (maximum change, 61 y0 of the original value). Under the same conditions, corresponding amounts of bound glycans were without effect.
Besides LDL, also HDL and VLDL were incubated with increasing amounts of high or low molecular weight free and bound glycans (Fig. 5). While LDL and HDL show obvious spectral changes when incubated with either high or low molecular weight free glycans, they remain unchanged upon incubation with bound glycans. Pyrene-labeled VLDL failed to demonstrate +ny appreciable spectral change when incubated with either free or bound glycans. Fig. 6 shows that the effect of high molecular   weight free glycans on LDL is reversible only in part at high ionic concentration, while Fig. 7 shows that treatment of free glycans with papain completely eliminates their effects on LDL. Proteolytic treatment did not influence the already minimal effects of the bound glycans on LDL.
The interference of bound glycans and chondroitin-4-sulfate on the interaction of either high or low molecular weight free glycans with LDL is shown in Fig. 8. Chondroitin-4-sulfate (at levels 10 and 20 times higher than those of the free glycans) and bound glycans (at levels 10 times higher than those of the free glycans) reduce the decrease of the ZE/ZM ratio brought about by the free glycans. On the other hand, PPL and chondroitin6sulfate do not show similar effects. able to those of the previous method (22). Gel filtration and ultracentrifugal analyses of these preparations have demonstrated considerable heterogeneity. The high molecular weight free glycans probably consist of two chains attached to a peptide backbone, since treatment with papain reduces their average molecular weight from 37,100 to 18,000.
While the amount of high molecular weight bound glycans was insufficient for study, the low molecular weight fractions of both species were utilized in the present report. Their molecular parameters suggest that they represent oligosaccharides still attached to protein.
The availability of sufficient amounts of plasma glycans has permitted a study of their interaction with the various lipoproteins and particularly with LDL, the species which has been demonstrated to interact preferentially with various polyanions (1) and with the arterial wall (8).
The experiments have been performed with two levels of reactants. In the first series (which is comparable to experiments performed in the past in various laboratories), 12 mg of LDL were combined with 30 to 100 c(g of glycans, as hesuronate, in order to verify the occurrence of interaction between the two species of macromolecules.
The results presented in Table II indicated that all of the species tested are capable of precipitating LDL, 50 pg of each type being the most effective amount.
In the second series of experiments, the use of pyrene-labeled lipoproteins allowed us to use minimal amounts of reactants and to approach their physiological proportions. The parameter measured, the ratio of intensities ZE/Z,, is directly proportional to "c," the pyrene concentration within a lipoprotein, 6. Effects of increasing ionic strength on the interaction between 0.12 mg of 4y0 pyrene-labeled LI)L and free glycans. A, value of ZE/Iu ratio for LDL control; B, its decrease upon inubation with 1.5 rg of high molecular weight free glycans under standard ionic conditions (0.14 31 NnCl); C, its decrease n-hen, after such incubation, the NaCl concentration was raised to 0.9 M and incubation was resumed for 3 hours; D, its decrease when incubation was carried out directly in 0.9 n% NaCl. Control samples were identical in composition to experimental ones, glgcans being omitted.
proportional to "q," the viscosity of the hydrocarbon region of the lipoprotein (39,40). Thus, a decrease in the cscimer to monomer ratio observed in pyrette-labeled lipoproteins must, result from a change in the microscopic structure of the lipoprotein, although it is not possible at this titne to distinguish the relative importance of changes in microviscosity or hydrocarbon volume in producing this effect. If one visualizes the pyrette in the hydrocarbon region of the phospholipids and cholesteryl ester layers of lipoproteins, an increased viscosity or swelling of this region (which would decrease the microscopic concentration of pyrene, c) could reduce the rate of excimer formation by decreasing the monomers' collision rates (39,40).
The results shown in Figs. 4 and 5 indicate that the ratio IE/I~ of solutions containing either LDL or HDL decreases considerably upon interaction with high molecular weight and low molecular weight free glycans at a physiological pH and ionic concentration and with quantities of reactants approaching their physiological levels (250 to 500 pg of glycans as hexuronate, and 200 mg of LDL/lOO ml of plasma). Similar changes in the IE/IM ratio have been obtained with LDL labeled with 2% pyrene. The fact that only a very limited decrease of the IE/Ix ratio is observed as a result of interaction of LDL with large amounts of heparin (which is recognized to be the most efficient precipitant for LDL (8)) suggests that the formation of soluble and insoluble complexes between lipoproteins and glycans may not involve the same mechanisms. In fact, contrary to the formation of insoluble complexes (9, lo), the interaction between free glycans and LDL is not eliminated by a 6-fold increase in ionic concentration (Fig. 6), suggesting that it is not simply an electrostatic interaction.
The drastic effects elicited with proteolytic treatment of both high molecular weight and low molecular weight free glycans (Fig. 7) are consistent with this assumption, and suggest that, under the conditions of our experiments, protein-protein interactions might have occurred. Because of the low protein content of VLDL, its lack of interaction with plasma glycans would thus be understandable.
The free glycans, which are rich in sulfate, are the only ones producing detectable physical modification of LDL and HDL, regardless of their molecular size. They represent approximately 15y0 of the total plasma glycans, and from the results illustrated in Fig. 8, it would appear that their effects on LDL are prevented by the less sulfated species, the bound glycans. The latter ones, although capable of precipitating LDL at high concentrations (see Table II), do not cause significant changes of their structure (Figs. 4 and 5) under the conditions of our second set of experiments. Thus, it is possible that under physiological conditions, the unfavorable interaction between circulating LDL and free glycans might be prevented by the more abundant bound glycans. This interaction could take place, however, whenever the normal balance between free and bound glycans would be altered, as in the case of mucopolysaccharide storage diseases, in which the level of the free is absolutely and relatively increased over that of the bound glycans (22).
As previously discussed by Iverius (lo), when LDL is forced through the arterial wall by hydrostatic pressure, its concentration may increase because of a reduced rate of transport caused by the molecular sieving and steric exclusion effects of the proteoglycan network.
These phenomena have been considered sufficient for the entanglement and precipitation of LDL within the wall. However, in view of the results obtained in our experiments, additional relevant factors leading to LDL deposition and accumulation could be the increased viscosity or swelling of the hydrocarbon region resulting from their interaction with some of the plasma glycans. Moreover, it has been suggested that the latter two modifications could decrease the rate of catabolism of the LDL particles and contribute to their permanence within the arterial wall (40).
Finally, the favorable results claimed by Morrison (41) and by Nakazawa et al. (42) with the oral and parenteral administration of chondroitin-4-sulfate to experimental animals and 5391 patients with atherosclerosis might find a rational explanation in the protective effects that this glycan and its products of extensive degradation, such as the bound glycans, show in preventing the molecular modifications of LDL caused by plasma-free glycans.