Electron Microscopic Studies of Proteoglycan Aggregates from Bovine Articular Cartilage*

Proteoglycan aggregates from bovine articular cartilage have been visualized by electron microscopy of mixed proteoglycan-cytochrome c monolayers. The proteoglycan aggregates consist of proteoglycan subunits arising laterally at fairly regular intervals (20 to 30 nm) from the opposite sides of an elongated filamentous structure. The filamentous backbone in individual aggregates varies in length from 400 to 4000 nm. The individual proteoglycan subunits in the aggregate vary in length from 100 to 400 nm. However, there is no difference in the average size of the proteoglycan subunits associated with the largest or smallest aggregates. The sizes of the individual aggregates are determined mainly by the lengths of their filamentous backbones. The stoichiometry of binding of subunits to filament, calculated from the data reported here, is close to that for the binding of subunits to hyaluronic acid reported by others.


SUMMARY
Proteoglycan aggregates from bovine articular cartilage have been visualized by electron microscopy of mixed proteoglycan-cytochrome c monolayers. The proteoglycan aggregates consist of proteoglycan subunits arising laterally at fairly regular intervals (20 to 30 nm) from the opposite sides of an elongated filamentous structure.
The filamentous backbone in individual aggregates varies in length from 400 to 4OGO nm. The individual proteoglycan subunits in the aggregate vary in length from 100 to 400 nm. However, there is no difference in the average size of the proteoglycan subunits associated with the largest or smallest aggregates. The sizes of the individual aggregates are determined mainly by the lengths of their filamentous backbones.
The stoichiometry of binding of subunits to filament, calculated from the data reported here, is close to that for the binding of subunits to hyaluronic acid reported by others. The  Proteoglycan complex containing roughly equal amounts of 16 S protcoglycan subunit and 70 S proteoglycan aggregate was isolated by equilibrium density gradient ccntrifugation under associative conditions, following dissociative extraction of bovine proximal humeral articular cartilage with 3 111 MgCl,, as previously described (5). The appearance of proteoglycan aggregates invariably found in preparations of proteoglycan complex is illustrated in Figs. 1 to 3. The aggregates consist of proteoglycan subunits of varying length that arise laterally at fairly regular intervals from the opposite sides of an elongated, filamcntous, thread-like structure.
From a large number of electron micrographs of many preparations, clcctron micrographs of seven protcoglycan aggregates were selected in which the molcculcs were spread on the monolayer without kinking, entanglement, or overlapping of the filamcntous backbone or proteoglycan subunits.
In these molecules, measurements were made of the contour lengths of the filamcntous backbone, and of the long axis (core protein) of the protcoglycan subunits, and the number of subunits per aggrcgatc were counted (Table I). Fig. 1 shows the structure of one of the largest protcoglycan aggregates encountered.
It contains 140 protcoglycan subunits of varying length, which arise in roughly perpendicular fashion from the opposite sides of a filamentous backbone approximately 4200 nm in length. Fig. 2, at higher magnification, shows an aggregate of intermcdiatc size, consisting of 77 subunits arising from a filament approsimatcly 1700 nm in length.
In this cast, both subunits and filament were unusually well extended on the monolayer and clearly defined. Fig. 3 shows two small aggregates surrounded by free proteoglycan subunit,s, which are also in variably present in proteoglycan complex preparations. The conditions of spreading and staining employed (described under "Experimental Procedures") rcprcscnt the best set of conditions SO far identified for demonstrating the molecular architecture of the proteoglycan aggregate, specifically the filamentous backbone of the aggregate and the numbers of subunits per aggregate. Under these conditions, the chondroitin sulfate and kcratan sulfate chains of proteoglycan subunits arc usually not extended but appear to be clumped alongside the protein core. A proteoglycan subunit is occasionally found in which the mucopolysaccharidc side chains arc extended (Fig. 3, inset). Under other COllditiOllS (spreading solution 0.05 M ammonium acetate, hypophase 0.01 M ammonium acetate), the mucopolysaccharide chains of the protcoglycan subunits are more uniformly extended.
In this case, proteoglycan subunits and the filamentous backbone of proteoglycan aggregates are entangled and overlapping, and the molecular architecture of the aggregat.c is obscured.
Measurements of the length of the filamentous backbone of each aggregate, of the average length of the long axis (protein core) of proteoglycan subunits, and of the number of proteoglycan subunits per aggregate are presented in Table I. These measurements indicate that the size of the proteoglycan aggregate is determined mainly by the length of its filamentous backbone. Consider first the size of the proteoglycan subunits associated with individual aggregates (Table I, Column 1). Measurements of the average lengths of the long axis (protein core) of the subunits indicate that there is no difference in the average size of the subunit's associated with the largest or smallest aggregates.
The long axis of the subunits associated with each aggregate are all very close to an over-all average length of 226 nm. Consider next the relationship between the number of subunits per aggregate and t,he length of the filamcntous backbone of each aggregate (Table I, Columns 2 and 3). The number of subunits per aggregate increases roughly in proportion to the length of the filamentous backbone.
The calculated average interval between subunits is approximately 26 nm. Direct measurements of the intervals between sites of attachment of the subunits to the filament gave an average value of 29 rim.
The latter studies, viewed in context with the work reported here, strongly suggest that the filamentous backbone of the proteoglycan aggregate demonstrated by electron microscopy is hyaluronic acid. Hardingham and Muir (6) demonstrated that the addition of small amounts of hyaluronic acid (0.7 %) to proteoglycan subunit from pig laryngeal cartilage resulted in a large increase in the hydrodynamic size of the subunit, demonstrated by gel chromatography or viscosity studies. They postulated that the interaction was a cooperative binding of many proteoglycan subunits with a single hyaluronic acid chain. The complex formed between subunits and hyaluronic acid was dissociated under the same conditions as proteoglycan aggregate. Hyaluronic acid was directly isolated from pig laryngeal cartilage and found to bind proteoglycan subunits isolated from the same cartilage (7).
Hascall and Heincg%rd (10-12) have confirmed and greatly extended the observations of Hardingham and Muir. Working with proteoglycans from bovine nasal cartilage, Hascall and Heineggrd found that 0.4 to 0.8% hyaluronic acid (W/W) was present in proteoglycan aggregate (10). Studies of the binding of proteoglycan subunit to hyaluronic acid preparations of different molecular weights suggested that the sizes of proteoglycan aggregates are determined mainly by the sizes (lengths) of the hyaluronic acid molecules, in accord with the results reported here. Hascall and Heinegbrd also studied the specificity of the interaction of proteoglycan subunit, and of proteoglycan subunit core preparations in which chondroitin sulfate was removed with chondroitinase, with hyaluronic acid. It was found that hyaluronic acid decasaccharides or nonasaccharides interacted strongly with proteoglycan subunit, whereas smaller hyaluronic acid oli-   5 Hardingham and Muir (8) 5 Hascall and Heineg%rd (11) S-10 Hascall and Heineg%rd (11) 24-48 Hascall and Heineg%rd (11) 24 Hardingham and Muir (8) 20-30 This paper References a Calculated assuming that the filamentous backbone of the proteoglycan aggregate is hyaluronic acid, using a molecular weight of 416 and a length of 1 nm for the hyaluronic acid disaccharide repeating unit. gomers interact only weakly (11). Subsequently, Heineg%rd and Hascall were able to isolate that region of the proteoglycan subunit core protein which binds to hyaluronic acid.
In Table II, the stoichiometry of binding of proteoglycan subunits to hyaluronic acid derived from chemical studies is compared with that calculated from the measurements of the electron micrographs of the proteoglycan aggregates.
Both the studies of Hardingham and Muir, and those of Hascall and Heineggrd indicate that the shortest chain length of hyaluronic acid to which a single proteoglycan subunit can bind strongly is approximately 5 nm in length (Table II, Lines 1 and 2). However, when many proteoglycan subunits bind to a single hyaluronic acid chain of high molecular weight (Table II, Lines 4 and 5), they bind at intervals far greater than 5 nm. The spacing between neighboring subunits bound to the same hyaluronic acid chain appears to be related to the lengths of the mucopolysaccharide chains arising from the proteoglycan subunit core protein, and to the resulting steric effects between adjacent subunits (8,11). Thus, core molecules from which chondroitin sulfate chains have been removed, consisting of kcratan sulfate chains -6 nm in length arising from core protein, bind to high molecular weight hyaluronic acid chains at 8. to IO-rim intervals (Table II, Line  3). Native protcoglycan subunits with chondroitin sulfate chains, 40 to 50 nm in length, arc spaced at a minimum of 24 nm. Consider finally the actual intervals at which protcoglycan subunits arise from the filamentous backbone of the protcoglycan aggregates, based or1 direct mcasuremcnts of the clcctron micrographs (Table II, Line 6). The ratio of the length of the filamentous backbone of each aggregate to the number of subunits per aggregate indicates that the avcragc spacing between subunits is from 20 to 30 nm. This correspondence between the spacing of subunits calculated from binding studies with that demonstrated by electron microscopy strongly suggests that the filamentous backbone of the protcoglycan aggregate is hyaluronic acid.  Table I were also measured.
The length distribution of the subunits is pfescnted in Fig. 4. The long axes of the subunits vary greatly in length from 100 to 400 nm. Thcrc is no indication of the cxistcncc of a population of monomers and dimers.
These results 'are in accord with, and support the suggestion of HcinegBrd and Hascall (12), that the core protein of protcoglycan subunit contains a polysaccharide attachment region of variable length.