- Email: [email protected]
Hyaluronan–Phospholipid Interactions
JOURNAL OF STRUCTURAL BIOLOGY ARTICLE NO. SB973908
120, 1–10 (1997)
Hyaluronan–Phospholipid Interactions Ivonne Pasquali-Ronchetti, Daniela Quaglino, Giuseppe Mori, and Barbara Bacchelli Biomedical Sciences Department, University of Modena, via Campi, 287, 41100 Modena, Italy
and Peter Ghosh Raymond Purves Bone and Joint Research Laboratories (University of Sydney) at the Royal North Shore Hospital, St. Leonards, New South Wales, 2065, Australia Received February 29, 1997, and in revised form July 1, 1997
and tear are largely unknown. These cavities are filled by fluids containing amphophilic molecules, such as phospholipids, which might function as a ‘‘boundary’’ lubricant by binding to the surface membrane through their polar terminus and leaving the hydrophobic tail of the molecule exposed to equivalent hydrophobic chains extending outwardly from the opposite surface (Agostoni, 1972; Agostoni et al., 1969; Hills, 1989; Hills et al., 1982; Hills and Butler, 1985). As to the diarthrodial joint, its cavity is filled by the synovial fluid (SF) containing hyaluronan (HA), proteins, proteoglycans, and lipids. In normal joints, HA is the most represented constituent (2–4 mg/ml), has molecular mass in the order of 107 Da (Balazs and Denlinger, 1985), and is thought to be responsible for the retention of water within the joint cavity, especially under pressure, and to provide a hydroelastic cushion at the interface of the synovial membrane and between cartilage surfaces (Balazs, 1974; Laurent and Fraser, 1992; McDonald et al., 1994; Ogston and Steiner, 1953). The hydrodynamic and viscoelastic properties of concentrated solutions of HA are favored by the ability of HA to form an extended three-dimensional network by intra- and interchain bridges arising from hydrophobic interactions between regions along the HA chains (Gibbs et al., 1968; Kobayashi et al., 1994; Scott, 1992; Scott et al., 1991, 1990). Synovial fluid has been shown to contain also triglycerides, cholesterol, and phospholipids (PL), derived from the bloodstream or produced by synovial and cartilage cells (Bole, 1962; Rabinowitz et al., 1983; Wise et al., 1987). Early studies have suggested that lipids could exert a lubricant effect in joints by being adsorbed to the articular cartilage surface, as treatment of the articular cartilage with lipid solvents increased the joint friction more than
Hyaluronan–phospholipid interactions have been studied in vitro by negative staining and rotary shadowing electron microscopy. Hyaluronan (HA) molecules of different molecular weights (around 170 000; 740 000, and 1.9 3 106 Da) were added to phospholipid suspensions (DPPC or egg lecithin) that were in the form of either unilamellar particles or multilamellar vesicles. Suspensions were then gently stirred and incubated at different temperatures from 24 hr up to 7 days. After 24 hr, at temperatures just above the melting point of the phospholipid used, both unilamellar particles and multilamellar vesicles were already shown to change their organization in the presence of HA, giving rise to the formation of (1) huge perforated membranelike structures lying on the substrate; (2) 12-nmthick ‘‘cylinders’’ (rollers) with a tendency to aggregate and to form sheets. These structures were seen only in the presence of high-molecular-weight HA, whereas low-molecular-weight HA (170 kDa) induced fragmentation of liposomes and formation of a few short rollers. These data show that phospholipids and HA interact and suggest they may also do so in vivo within the joint cavity, where both chemical species are present, giving rise to complexes which might exhibit peculiar lubricating and protective properties. It is also proposed that such interactions may not be as efficient in arthritic joints, where HA is degraded to low-molecular-weight fragments. r 1997 Academic Press
INTRODUCTION
The mechanisms by which opposing biological surfaces lining closed cavities, such as pleural, pericardium, and synovial spaces, are capable of gliding over each other with low friction and minimal wear 1
1047-8477/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.
2
PASQUALI-RONCHETTI ET AL.
two times (Little et al., 1969) and rigorous rinsing of the articular cartilage with saline left articular surfaces very hydrophobic (Hills, 1990; Hills and Butler, 1985). Furthermore, electron microscopy showed the presence of lamellar bodies and of detachable lamellated PL structures on the surface of the articular cartilage (Hills, 1989, 1990). More recently, we could demonstrate that the surface of human and animal articular cartilages was covered by a discontinuous mono-multilayered membrane-like structure, adherent but not attached to the cartilage, which could be mechanically removed, was sensitive to lipid solvents, proteases, hyaluronidase, and chondroitin lyases, and, by immunocytochemistry, was positive for HA (Guerra et al., 1996). Therefore, it might be possible that HA and PL might interact at the cartilage surface. Formation of complexes between HA and PL has been suggested by 13C-NMR spectroscopy and laser–light scattering studies, as the chain flexibility of HA in solution was shown to increase when sonicated with dipalmitoylphosphatidylcholine (DPPC) (Ghosh et al., 1994). These findings were interpreted as arising from competition of DPPC for the hydrophobic centers along HA which are responsible for the interchain interactions and entanglements. In the present study, we describe the in vitro formation of HA–PL complexes using negative staining and rotary shadowing electron microscopic techniques. The results clearly indicate that the two molecular species have a strong tendency to interact and to form rather stable complexes, the nature of which is dependent on the molecular weight of the HA used. MATERIALS AND METHODS Dipalmitoyl-D,L-a-phosphatidylcholine (DPPC) and egg lecithin were from Sigma Chemical Co. (St Louis, MO). Highly purified 1.9 3 106 MW hyaluronan (HA) (lyophilized powder) was from Denki Kaguku Kogyo (Tokyo, Japan); 7.4 3 105 MW HA (10 mg/ml) and 1.7 3 105 MW HA (lyophilized powder) were kindly provided by Fidia SpA (Abano Terme, Italy). HA preparations were used without any further purification. The weight average molecular weights (MWW ) were provided by the nominated manufacturers and were supplied as the sodium salts. Lyophilized powders were reconstituted to 1% (10 mg/ml) sterile aqueous solutions prior to use. Uranyl acetate used as a negative stain was from Fluka AG, Switzerland. All other reagents used were of analytical grade. Double-distilled water was used to prepare solutions and buffers. Phospholipid Preparation (a) Aliquots of DPPC in pure ethanol (20 mg/ml) or of egg lecithin were added to Tyrode’s saline solution, pH 7.3, or to 0.05 M PBS containing 0.15 M Na Cl, pH 7.3, to the final concentration of 0.1, 0.5, and 1 mg/ml of phospholipid. The suspensions were mixed by vortex for 60 sec at room temperature and left at 50°C (DPPC) or at 35°C (egg lecithin) for 60 min, leaving the vials open. During that time, suspensions were mechanically stirred every 15 min for 60 sec. HA of the different molecular weights was added to
the phospholipid suspensions to the final concentrations of 0.1, 0.5, and 1 mg/ml, mixed by vortex for about 60 sec, and incubated under nitrogen at 50°C (DPPC) or at 35°C (egg lecithin) in the dark for times from 1 hr up to 4 days. (b) An aliquot of 20 mg/ml DPPC in pure ethanol was dried under nitrogen gas on the wall of a vial; Tyrode’s or PBS, pH 7.3, was added to reach the DPPC final concentration of 0.1, 0.5, and 1 mg/ml; phospholipid was suspended by sonication for 5 to 20 min in ice bath at power level 6 with a Branson sonifier, Model s-75. The dimensions of PL vesicles were verified by negative staining electron microscopy. HA was added to the phospholipid vesicles and the mixture was mechanically stirred and incubated at 50°C, under nitrogen atmosphere and in the dark, for times from 1 hr to 4 days. (c) DPPC (10 mg/ml) in Tyrode’s or in PBS, pH 7.3, was mechanically suspended by repeated passages at 50°C through a 6-in.-long 20-gauge needle and passed 7 to 10 times through an extruder apparatus (Thermobarrel Extruder, Lipex Biomembranes Inc., Vancouver, Canada), maintained at 50°C, and by the use of 0.1-µm mesh filters and a pressure of 20–40 bars. Vesicles were unilamellar and exhibited a rather homogeneous size (about 15–30 nm in diameter) as checked by negative staining electron microscopy. The phospholipid concentration of the final suspension was checked and vesicles were mechanically mixed with HA of the different molecular weights, at the same concentrations specified before, and were incubated at 50°C, under nitrogen atmosphere and in the dark, for times from 1 hr to 4 days. Negative Staining At different times of incubation, 10 µl of phospholipids or of HA alone or of the mixtures was placed on a holey carbon-coated copper grid and washed with 3–4 drops of 1% uranyl acetate in double-distilled water; after removal of the excess of stain the specimen was air dried and observed in the electron microscope. Rotary Shadowing At different times of incubation, about 100 µl of the various suspensions was gently spread on a 1-cm2 freshly cleaved mica sheet and the excess of material was removed by touching one edge of the mica with filter paper. Specimens were left to air dry and placed into a vacuum apparatus (Balzers, BAF 301 freeze etching). Platinum–carbon was evaporated on the specimen from an angle of 11° under rotation; then carbon was evaporated from 90°, forming a surface replica that was detached by floating on double-distilled water, collected on a copper grid, and observed in a Jeol JEM 1200 EXII electron microscope. RESULTS
As specified under Materials and Methods, hyaluronan was added to either monolamellar or multilamellar phospholipid suspensions. In all cases, the phospholipid organization was deeply affected by the presence of HA and the new structures resulting from the interactions between the two molecular species appeared to be rather independent from the starting phospholipid organization (monolamellar or multilamellar) and from the phospholipid nature (DPPC or egg lecithin), whereas they were markedly influenced by HA molecular weight. Moreover, although the techniques used in this study were not quantitative, it would appear that optimum complex formation occurred when both phospholipid and HA were at a final concentration of 1 mg/ml.
HYALURONAN–PHOSPHOLIPID INTERACTIONS
3
Negative Staining HA molecules could not be visualized by negative staining electron microscopy due to their small molecular dimensions. On the contrary, phospholipid vesicles, made either by extrusion or by sonication, as well as their structural modifications in the presence of HA, were readily seen by this method. DPPC particles prepared by extrusion and stained with 1% uranyl acetate appeared as an homogeneous population of 15- to 30-nm-wide unilamellar vesicles (Fig. 1a) which were stable for at least 4 days at 50°C. After incubation with 170 000 MW HA, for times from 24 to 48 hr, the number of particles on the grid appeared dramatically reduced, whereas large phospholipid aggregates, which could not be better defined by this technique, were observed on the same grids. These aggregates were never observed in the absence of HA. Hyaluronan with MW 740 000 induced the PL particles to form linear aggregates and to fuse. However, the most drastic modifications of PL particles were seen with 1.9 3 106 Da MW HA. After incubation at 50°C for times up to 2 hr, the 15to 30-nm-wide unilamellar vesicles obtained by extruder appeared to align into linear aggregates, forming interconnected short chains that, with time, appeared to fuse into strands, which were thinner than the original particles, exhibited a rather constant thickness of 12 nm, and were organized into a network (Fig. 1b). After 48 hr incubation, no more isolated vesicles could be identified and the material was completely organized into a meshwork of regular 12-nm-wide threads (Fig. 1c). Phospholipid suspensions obtained by sonication consisted of vesicles whose dimensions and structural organization (unilamellar or multilamellar) were dependent on time and frequency of sonication (see Materials and Methods) (Figs. 2a and 2b). When these phospholipid suspensions were incubated with HA, their appearance was modified again depending on the molecular weight of HA used. HA of MW 170 000 Da did not change the organization of DPPC vesicles substantially; however, a series of small globules could be seen on the grid (Fig. 2c). Incubation with the 740 000 MW hyaluronan induced the phospholipid vesicles to exhibit filopodia-like protrusions; that became longer for incubation times up to 24 hr and tended to stick into rather parallel 12-nmwide ‘‘tubules’’ (Fig. 2d). Incubation with the highmolecular-weight hyaluronan induced the phospholid vesicles to aggregate and fuse into ill-defined sheets, which could not be better defined by this technique. Phospholipid vesicles made by mechanical stirring of the suspension at room temperature consisted of huge concentric multilamellar structures, whose organization could not be easily studied by the nega-
FIG. 1. Negative staining of DPPC particles prepared by extrusion (a) and after addition of 1.9 3 106 Da HA (b,c). At different incubation times at 50°C (2 hr (b) and 48 hr (c), respectively) the high-molecular-weight HA induced a progressive fusion of phospholipid vesicles into a final network of 12-nm-wide strands. Bars, 0.05 µm.
tive staining technique, whereas it could be better defined by rotary shadowing. Rotary Shadowing Hyaluronan. Hyaluronan molecules observed by rotary shadowing electron microscopy appeared in form of rather rigid filaments, whose length and distribution on the substrate were strictly related to their molecular mass.
4
PASQUALI-RONCHETTI ET AL.
FIG. 2. Negative staining of sonicated phospholipid suspensions consisting of vesicles with different sizes depending on the sonication procedure used (a, b). In the presence of HA (c, d) liposomes changed their organization. 170 000 MW HA induced fragmentation of liposomes into small phospholipid vesicles and micelles (c); 740 000 MW-HA (d) induced the formation of tubular structures with a tendency to form 12-nm-wide stacks. Bars, 0.05 µm.
The 170 000 MW-HA consisted of isolated short filaments and of 15- to 20-nm-wide globules spread on the mica (Fig. 3a). The 740 000 MW HA appeared in form of globules and filaments which seldom formed discontinuous aggregates (Fig. 3b), while an approximately 2 million MW HA preparation appeared as very long filaments forming a threedimensional array, which flattened on the mica, giving rise to an irregular bidimensional network (Fig. 3c). In all cases the thickness of filaments was rather variable, as they appeared to be composed of a high number of molecules running parallel for long distances and splitting into progressively smaller entities, down to 2-nm-thick filaments, which is the resolution limit of this technique. Very often, highmolecular-weight HA was in the form of huge bundles of rather parallel elements (Fig. 3d). These features clearly indicate that pure HA molecules stick to-
gether through lateral adhesion forces and are consistent with observations of others (Ghosh et al., 1994; Scott, 1992; Scott et al., 1990, 1991). Phospholipid suspensions. By rotary shadowing, DPPC liposomes obtained by extruder or by sonication appeared as roundish globules tending to aggregate on the mica upon dehydration. On the contrary, when suspended in buffer by mechanical stirring, DPPC formed multilayered concentric large vesicles (Fig. 4a), whose size and number of lamellae in each vesicle were rather variable within the same preparation. These vesicles were rather stable with time up to at least 7 days at 50°C. Phospholipid/hyaluronan interactions. Given the high stability of DPPC multilamellar vesicles in buffer with time and their peculiar appearance when observed by rotary shadowing (Fig. 4a) as well as the
HYALURONAN–PHOSPHOLIPID INTERACTIONS
5
corresponding peculiar stable feature of hyaluronan when observed with this same technique, the rearrangement of the two molecular species when mixed together could be clearly and easily appreciated. Multilamellar phospholipid vesicles changed their organization in the presence of HA; these changes were dependent on the molecular weight of HA, were progressive with time, and were accelerated by temperatures higher than that of the melting point of the phospholipid. After 1 hr incubation at 50°C, DPPC membranes in the presence of 170 000 MW HA already showed large areas in the form of coalescent ill-defined globules, without any apparent ordered organization (Fig. 4b). Longer incubation times induced a fragmentation of the phospholipid membranes into small sheets and globules (Fig. 4c) and the appearance of short rigid rods forming small crystalline-like ordered structures (Fig. 4d, arrows). The 740 000 MW HA induced with time a more evident rearrangement of the phospholipid membranes with formation of interconnected tubules, of globules and membrane fragmentation (Fig. 4e). Moreover, these structures coexisted with a relevant number of crystalline-like structures (Fig. 4f ), which were identical to but much larger and longer than those observed with HA of low molecular weight (Fig. 4d). When incubated with the HA of MW ,2 million, the rearrangement of DPPC multilamellar vesicles was rather dramatic. For incubation times up to 1 hr at 50°C, the great majority of PL were still organized into multilamellar vesicles. However, 12-nm large threads of variable length were often observed at the periphery of or in close association with the phospholipid membranes, similar to those observed in the presence of 7.4 3 105 MW HA (Fig. 4e). For incubation times longer than 24 hr, the multilamellar DPPC vesicles almost disappeared and the specimen consisted of extremely large holey membranes (Fig. 4g), together with enormous numbers of crystallinelike structures (Fig. 4h). These were similar to, but much larger than, those already described for the low-molecular-weight HA–PL interactions (Figs. 4d and 4f ). The paracrystalline structures, in the form of flat or overlapping sheets on the mica, were large enough to be analyzed by optical diffraction, revealing a lateral periodicity of 12 nm. These crystallinelike structures may be interpreted as 12-nm-wide
FIG. 3. Rotary shadowing of HA molecules of different molecular weights. The 170 000 Da HA (a) consisted of short filaments and small globules; the 740 000 Da HA was composed of globules and rather long filaments (b); the 1.9 3 106 Da HA (c, d) formed long strands of adjacent filaments (arrows) that dissociated, giving rise to a spread network (c). Bars, 1 µm.
FIG. 4. Rotary shadowing of DPPC vesicles, prepared by vortexing, in the absence (a) and in the presence (b–h) of HA of different molecular weights. With time, the 170 000 Da HA (b–d) fragmented the phospholipid bilayers into small globules (b and c) and a few small crystalline-like structures (d, arrows). The effects of 740 000 MW HA were even more evident (e and f ). As already observed by negative staining, the most striking changes were observed with hyaluronan of 1.9 3 106 Da (g–h), in the presence of which phospholipids rearranged into huge perforated membranes (g) and sheets of 12-nm-wide cylinders (h). Bars, 1 µm.
HYALURONAN–PHOSPHOLIPID INTERACTIONS
‘‘cylinders,’’ which tended to aggregate by lateral apposition and to form extended roller sheets on the substrate. When egg lecithin multilayered vesicles obtained by vortexing (Fig. 5a) were incubated with the 2 million MW HA at 35°C, for times longer than 48 hr, phospholipid molecules appeared to be organized into huge thready networks (Fig. 5b), extended holey membranes (Fig. 5c), or crystalline-like structures (Fig. 5d). All these deep modifications of the egg lecithin membranes in the presence of HA were very similar to those already described for DPPC–HA interactions. DISCUSSION
Dipalmitoylphosphatidylcholine and egg lecithin molecules were shown to interact with hyaluronan, giving rise to complexes which might have some biological relevance. The interaction would seem to be independent of the PL used, since DPPC and egg lecithin afforted similar results and independent from ions of the medium (PBS or Tyrode’s). Moreover, the interactions would seem to be energetically favorable as they occurred irrespective of the previous organization of the phospholipid molecules (in the form either of large multilamellar structures or of unilamellar vesicles) and did not require sonication or other drastic procedures for formation. The data obtained can be interpreted as interactions between the two molecular species in which PL molecules aggregate around HA, the assembling being dependent on the length of HA molecules. Low-molecular-weight HA appeared to bind PL molecules favoring the fragmentation of large phospholipid membranes and the formation of globules with a tendency to aggregate. On the other hand, high-molecular-weight HA, which consisted of long filaments interconnected into a loose three-dimensional network, seemed to organize around them phospholipid molecules forming 12-nm-thick threads, which could either form a network or align parallel into large sheets. These features were observed by both negative staining (Figs. 1c, 1d, and 2d) and rotary shadowing (Figs. 4e, 4f, and 4h) electron microscopy. When observed by this latter technique, which provides a three-dimensional image of the objects, threads may be interpreted as ‘‘cylinders,’’ whose organization could correspond to that shown in Fig. 6a. The thickness and the appearance of a
FIG. 5. Rotary shadowing of lecithin vesicles, obtained by mechanical stirring (a), and incubated for 48 hr with hyaluronan of 1.9 3 106 Da (b–d). HA exherted structural modifications of the phospholipid bilayers similar to those previously observed with DPPC vesicles, such as a thready network (b), holey membranes (c), and crystalline-like structures (d). Bars, 1 µm.
7
8
PASQUALI-RONCHETTI ET AL.
FIG. 6. Proposed model for HA–lipid interactions. (a) HA molecules form a central filament surrounded by a bilayer of phospholipid molecules radially oriented along the central core of HA. The charged ends of the phospholipid molecules allow formation of a thin layer of water between the adjacent cylinders. (b) An inverted hexagonal lattice (HII phase) of dipalmitoylphosphatidylcholine induced by HA. This model is less probable, as discussed in the text.
single cylinder were consistent with complexes formed by one (or more) HA molecules acting as a central filament surrounded by a bilayer of phospholipid molecules radially oriented along the central HA core. As shown in Fig. 6a, the charged ends of the phospholipid molecules would favor a water layer between adjacent cylinders. Moreover, the dependence of the interactions on the molecular weight of HA is consistent with the model proposed in Fig. 6a. The longer the HA chain the more phospholipid molecules attached, which would result in more extensive and thus stronger intermolecular phospholipid interactions compatible with sheet formation. In any case, it cannot be excluded that the formation of rollers may be favored by phospholipid/mica interactions during drying. The idea that cylinders may represent inverted hexagonal phases (HII phase) (Seddon, 1990) (Fig. 6b) of phospholipids seems less probable for the following reasons. First of all, the phospholipid concentration was too low compared to that required for an inverted hexagonal lattice to occur, unless the cylinder is considered the result of local phospholipid concentration by progressing dehydration on the mica. In this case, however, cylinders would also have been observed in pure DPPC specimens. Second, 50°C for DPPC is not a tempera-
ture compatible with the inverted hexagonal phase (Luzzati, 1968) and because dehydration was performed at room temperature. Finally, the spacing for the inverted hexagonal lattice has been measured for dioleylphosphatidylethanolamine at 22°C from 4.0 to 6.2 nm, depending on the water content (Rand and Fuller, 1994). The cylinders illustrated in the present report are always 12 nm wide; that is, double that of a membrane bilayer. Another interesting feature obtained by incubation of phospholipid with medium and high-molecular-weight HA was the formation of huge perforated membrane-like sheets whose complexity could not be solved by the techniques used, but which might have important biological significance, as they could be the equivalent of the discontinuous membrane-like structures recently described as covering the surface of normal articular cartilage in diarthroic joints and which were shown by immunocytochemistry to be associated with HA (Guerra et al., 1996). The interaction of HA and the phospholipid DPPC has also been studied by gel-permeation chromatography, low-angle laser–light scattering, and 13CNMR techniques (Ghosh et al., 1994). These studies showed that the addition of DPPC to highly purified HA increased the flexibility of HA. It was suggested that this was due to competition of PL for intermolecular hydrophobic sites on the HA molecules involved in interchain associations. In the present study, the charged polar region of the phospholipid is considered to interact with HA, although the model shown in Fig. 6a does not exclude hydrophobic bonding by interactions between the hydrophobic patches on hyaluronan with the first angulated hydrocarbons on the phospholipid acyl chains (J. E. Scott, personal observation). The synovial fluid of normal joints contains several molecular species, excluding the proteins, the most abundant of which is HA. Hyaluronan is considered to play a critical role in the physiology of joint function, including lubrication of the synovial surfaces (Balazs et al., 1967; Laurent et al., 1995; Ogston and Steiner, 1953), although alternative glycoproteins have been proposed by others (Swann et al., 1981a,b). In addition, synovial fluid contains phospholipids (Bole, 1962; Wise et al., 1987) whose concentrations increase under pathological conditions (Wise et al., 1987); the phospholipids in synovial fluid have been proposed as the molecules responsible for articular cartilage ‘‘boundary’’ lubrication (Hills, 1989, 1990, 1995), and it has been suggested that the presence of HA may ‘‘fortify’’ lamellar lubrication (Hills, 1989, 1995). The model proposed in Fig. 6a is clearly compatible with the observations of Hills (1989, 1995) since the ability of HA to orientate phospholipid molecules into roller
9
HYALURONAN–PHOSPHOLIPID INTERACTIONS
structures, which can aggregate into sheets, might provide ideal conditions for lamellar lubrication. The aggregated rollers assembled as a layer would be analogous to graphite sheets which when subjected to shearing stresses would separate from each other with minimal friction. The presence of phospholipid–HA aggregates in the form either of rollers or of perforated membranelike sheets between the articulating surfaces of diarthroidal joints would be capable of providing excellent lubrication over a wide dynamic range and excellent protection of the cartilage surface from enzymes and free radicals in the synovial fluid. Rather interestingly, however, as shown in the present study, HA preparations with low molecular weights (around 170 000 or less) were less efficient in forming large membrane-like and roller sheets with phospholipids than the high-molecular-weight HA preparations. This finding may have important biological implications, as it is known that highmolecular-weight species of HA are degraded to low-molecular-weight species during joint inflammation and in osteoarthritis (Balazs et al., 1967; Dahl et al., 1985; Tulamo et al., 1994; Wise et al., 1987). Under pathological conditions, it would be expected that hyaluronan–phospholipid interactions would not produce the same sheets and rollers which are present under normal conditions and both lubrication and cartilage protection would be impaired. The injection of pure high-molecular-weight HA into osteoarthritic joints for symptomatic relief (Govoni et al., 1990; Peyron, 1993) would be expected to reestablish membrane-like and roller structures with endogenous phospholipids even though the majority of the free HA was observed to be cleared from the joint within a few days (Laurent et al., 1992). While the clearance rate of phospholipid–HA complexes from the synovial cavity is unknown, in this regard the observation by Laurent and co-workers (1992) that 2–5% HA injected into joints was retained and only slowly metabolized after intra-articular injection is noteworthy. Notwithstanding the ease with which these phospholipid–HA aggregates can be prepared in vitro, extrapolation to the in vivo situation must remain speculative until studies with normal and pathological synovial fluids have been completed. Studies in this direction are in progress and preliminary results confirm that phospholipids extracted and purified from normal bovine synovial fluid do exhibit a tendency to form aggregates with high-mass HA and that these aggregates have characteristics similar to those described in the present report. Funds for this study were from the Italian CNR and from the University of Modena (Progetto di Ricerca Applicata, 1995).
REFERENCES Agostoni, E. (1972) Mechanics of the pleural space, Physiol. Rev. 52, 57–128. Agostoni, E., Miserocchi, G., and Bonanni, M. V. (1969) Thickness and pressure of the pleural liquid in some mammals, Respir. Physiol. 6, 245–256. Balazs, E. A. (1974) The physical properties of synovial fluid and the special role of hyaluronic acid, in Helfet, A. (Ed.), Disorders of the Knee, 1st ed., pp 61–74, Lippincott, Philadelphia. Balazs, E. A., and Denlinger, J. L. (1984) The role of hyaluronic acid in arthritis and its therapeutic use, in Peyron, J. G. (ed.), Osteoarthritis-Current Clinical and Functional Problems, pp 165–174, Ciba Geigy, Paris. Balazs, E. A., Watson, D., Duff, I. F., and Roseman, S. (1967) Hyaluronic acid in synovial fluid. I. Molecular parameters of hyaluronic acid in normal and arthritic human fluids, Arthritis Rheum. 10, 357–376. Bole, G. G. (1962) Synovial fluid lipids in normal individuals and patients with rheumatoid arthritis, Arthritis. Rheum. 5, 589– 601. Dahl, L. B., Dahl, I. M. S. Engstrom-Laurent, A., and Granath, K. (1985) Concentration and molecular weight of sodium hyaluronate in synovial fluid from patients with rheumatoid arthritis and other arthropaties, Ann. Rheum. Dis. 44, 817–822. Ghosh, P., Hutadilok, N., Adam, N., and Lentini, A. (1994) Interactions of hyaluronan (Hyaluronic acid) with phospholipids as determined by gel permeation chromatography, multiangle laser-ligh-scattering photometry and 1H-NMR spectroscopy, Int. J. Biol. Macromol. 16, 237–244. Gibbs, D. A., Merrill, E. W., Smith, K. A., and Balazs, E. A. (1968) Rheology of hyaluronic acid, Biopolymers 6, 777–791. Govoni, E., Pasquali Ronchetti, I., Georgountzos, A., Perbellini, A., Cicchetti, F., Guerra, D., Vincenzi, D., Franchini, S., and Frizziero, L. (1990) Microarthroscopic and histopathological evaluation of synovial membrane in knee joint osteoarthritis after intraarticular treatment with hyaluronic acid, First Int. Symp. on Synovium, Philadelphia. Gross, J. (1948) Electron microscope studies of sodium haluronate. J. Biol. Chem. 172, 11–514. Guerra, D., Losi, M., Bacchelli, B., Mezzadri, G., and Pasquali Ronchetti, I. (1996) Ultrastructural identification of a membrane-like structure covering the surface of normal articular cartilage, J. Submicrosc. Cytol. Pathol. 28, 385–393. Hills, B. A. (1989) Oligolamellar lubrication of joints by surface active phospholipid, J. Rheum. 16, 82–90. Hills, B. A. (1990) Oligolamellar nature of the articular surface, J. Rheum. 17, 349–356. Hills, B. A. (1995) Remarkable anti-wear properties of joint surfactant, Ann. Biomed. Eng. 23, 112–115. Hills, B. A., and Butler, B. D. (1985) Phospholipids identified on the pericardiumm and their ability to impart boundary lubrication, Ann. Biomed. Eng. 13, 573–586. Hills, B. A., Butler, B. D., and Barrow, R. E. (1982) Boundary lubrication imparted by pleural surfactants and their identification, J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 53, 463–469. Kobayashi, Y., Okamoto, A., and Nishinari, K. (1994) Viscoelasticity of hyaluronic acid with different molecular weight, Biorheology 31, 235–244. Laurent, U. B. G., Fraser, J. R. E., Engstrom-Laurent, A., Reed, R. K., Dahl, L. B., and Laurent, T. C. (1992) Catabolism of hyaluronan in the knee of the rabbit, Matrix 12, 130–136.
10
PASQUALI-RONCHETTI ET AL.
Laurent, T. C., and Fraser, J. R. E. (1992) Hyaluronan, FASEB J. 6, 2397–2404. Laurent, T. C., Laurent, U. B. G., and Fraser, J. R. E. (1995) Functions of hyaluronan, Ann. Rheum. Dis. 54, 429–432. Little, T., Freeman, M. A. R., and Swanson, S. A. R. (1969) Experiments on friction in the human hip joint, in Wright V. (Ed.), Lubrication and Wear in Joints, pp. 110–122, Sector, London. Luzzati, V. (1968) Chapman D. (Ed.), Biological Membranes, Vol. I, Academic Press, London. McDonald, J. N., and Levick, J. R. (1994) Hyaluronan reduces fluid escape rate from rabbit knee joints disparately from its effect on fluidity, Exp. Physiol., 79, 103–106. Ogston, A. G., and Stanier, J. E. (1953) The physiological function of hyaluronic acid in synovial fluid: Viscous, elastic and lubricant properties, J. Physiol. 119, 244–252. Peyron, J. G. (1993) Intraarticular hyaluronan injections in the treatment of osteoarthritis: State of-the-art review, J. Rheumatol. 20, 10–15. Rabinowitz, J. L., Gregg, J. R., and Nixon, J. E. (1983) Lipid composition of the tissues of human knee joints. II. Synovial fluid in trauma, Clin. Ortho. 190, 292–298. Rand, R. P., and Fuller, N. L. (1994) Structural dimensions and their changes in a reetrant hexagonal-lamellar transition of phospholipids, Biophys. J. 66, 2127–2138. Scott, J. E. (1992) Supramolecular organization of extracellular
matrix glycosaminoglycans in vitro and in the tissues, FASEB J. 6, 2639–2645. Scott, J. E., Cummings, C., Brass, A., and Chen, Y. (1991) Secondary and tertiary structure of hyaluronan in aqueous solution, investigated by rotary shadowing-electron microscopy and computer simulation. Hyaluronan is a very efficient network-forming polymer, Biochem. J. 274, 699–705. Scott, J. E., Cummings, C., Geiling, H., Stuhlsatz, H. W., Gregory, J. D., and Damle, S. P. (1990) Examination of corneal proteoglycans and glycosaminoglycans by rotary shadowing electron microscopy, Int. J. Biol. Macromol. 12, 180–184. Seddon, J. M. (1990) Structure of the inverted hexagonal (HII) phase and non-lamellar phase transitions of lipids, Biochim. Biophys. Acta 1031, 1–69. Swann, D. A., Hendred, R. B., Radin, E. L., Sotman, S. L., and Duda, E. A. (1981a) The lubricating activity of synovial fluids, Arthritis Rheum., 24, 22–30. Swann, D. A., Slayter, S. H., and Silver, F. H. (1981b) The molecular structure of lubricating glycoprotein 1, the boundary lubricant for articular cartilage, J. Biol. Chem. 256, 5921–5925. Tulamo, R. M., Heiskanen, T., and Salonen, M. (1994) Concentration and molecular weight distribution of hyaluronate in synovial fluid from clinically normal horses and horses with diseased joints, Am. J. Vet. Res. 55, 710–715. Wise, C. M., White, R. E., and Agudelo, C. A. (1987) Synovial fluid lipid abnormalities in various disease states: Review and classification, Semin. Arthritis Rheum. 16, 222–230.
120, 1–10 (1997)
Hyaluronan–Phospholipid Interactions Ivonne Pasquali-Ronchetti, Daniela Quaglino, Giuseppe Mori, and Barbara Bacchelli Biomedical Sciences Department, University of Modena, via Campi, 287, 41100 Modena, Italy
and Peter Ghosh Raymond Purves Bone and Joint Research Laboratories (University of Sydney) at the Royal North Shore Hospital, St. Leonards, New South Wales, 2065, Australia Received February 29, 1997, and in revised form July 1, 1997
and tear are largely unknown. These cavities are filled by fluids containing amphophilic molecules, such as phospholipids, which might function as a ‘‘boundary’’ lubricant by binding to the surface membrane through their polar terminus and leaving the hydrophobic tail of the molecule exposed to equivalent hydrophobic chains extending outwardly from the opposite surface (Agostoni, 1972; Agostoni et al., 1969; Hills, 1989; Hills et al., 1982; Hills and Butler, 1985). As to the diarthrodial joint, its cavity is filled by the synovial fluid (SF) containing hyaluronan (HA), proteins, proteoglycans, and lipids. In normal joints, HA is the most represented constituent (2–4 mg/ml), has molecular mass in the order of 107 Da (Balazs and Denlinger, 1985), and is thought to be responsible for the retention of water within the joint cavity, especially under pressure, and to provide a hydroelastic cushion at the interface of the synovial membrane and between cartilage surfaces (Balazs, 1974; Laurent and Fraser, 1992; McDonald et al., 1994; Ogston and Steiner, 1953). The hydrodynamic and viscoelastic properties of concentrated solutions of HA are favored by the ability of HA to form an extended three-dimensional network by intra- and interchain bridges arising from hydrophobic interactions between regions along the HA chains (Gibbs et al., 1968; Kobayashi et al., 1994; Scott, 1992; Scott et al., 1991, 1990). Synovial fluid has been shown to contain also triglycerides, cholesterol, and phospholipids (PL), derived from the bloodstream or produced by synovial and cartilage cells (Bole, 1962; Rabinowitz et al., 1983; Wise et al., 1987). Early studies have suggested that lipids could exert a lubricant effect in joints by being adsorbed to the articular cartilage surface, as treatment of the articular cartilage with lipid solvents increased the joint friction more than
Hyaluronan–phospholipid interactions have been studied in vitro by negative staining and rotary shadowing electron microscopy. Hyaluronan (HA) molecules of different molecular weights (around 170 000; 740 000, and 1.9 3 106 Da) were added to phospholipid suspensions (DPPC or egg lecithin) that were in the form of either unilamellar particles or multilamellar vesicles. Suspensions were then gently stirred and incubated at different temperatures from 24 hr up to 7 days. After 24 hr, at temperatures just above the melting point of the phospholipid used, both unilamellar particles and multilamellar vesicles were already shown to change their organization in the presence of HA, giving rise to the formation of (1) huge perforated membranelike structures lying on the substrate; (2) 12-nmthick ‘‘cylinders’’ (rollers) with a tendency to aggregate and to form sheets. These structures were seen only in the presence of high-molecular-weight HA, whereas low-molecular-weight HA (170 kDa) induced fragmentation of liposomes and formation of a few short rollers. These data show that phospholipids and HA interact and suggest they may also do so in vivo within the joint cavity, where both chemical species are present, giving rise to complexes which might exhibit peculiar lubricating and protective properties. It is also proposed that such interactions may not be as efficient in arthritic joints, where HA is degraded to low-molecular-weight fragments. r 1997 Academic Press
INTRODUCTION
The mechanisms by which opposing biological surfaces lining closed cavities, such as pleural, pericardium, and synovial spaces, are capable of gliding over each other with low friction and minimal wear 1
1047-8477/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.
2
PASQUALI-RONCHETTI ET AL.
two times (Little et al., 1969) and rigorous rinsing of the articular cartilage with saline left articular surfaces very hydrophobic (Hills, 1990; Hills and Butler, 1985). Furthermore, electron microscopy showed the presence of lamellar bodies and of detachable lamellated PL structures on the surface of the articular cartilage (Hills, 1989, 1990). More recently, we could demonstrate that the surface of human and animal articular cartilages was covered by a discontinuous mono-multilayered membrane-like structure, adherent but not attached to the cartilage, which could be mechanically removed, was sensitive to lipid solvents, proteases, hyaluronidase, and chondroitin lyases, and, by immunocytochemistry, was positive for HA (Guerra et al., 1996). Therefore, it might be possible that HA and PL might interact at the cartilage surface. Formation of complexes between HA and PL has been suggested by 13C-NMR spectroscopy and laser–light scattering studies, as the chain flexibility of HA in solution was shown to increase when sonicated with dipalmitoylphosphatidylcholine (DPPC) (Ghosh et al., 1994). These findings were interpreted as arising from competition of DPPC for the hydrophobic centers along HA which are responsible for the interchain interactions and entanglements. In the present study, we describe the in vitro formation of HA–PL complexes using negative staining and rotary shadowing electron microscopic techniques. The results clearly indicate that the two molecular species have a strong tendency to interact and to form rather stable complexes, the nature of which is dependent on the molecular weight of the HA used. MATERIALS AND METHODS Dipalmitoyl-D,L-a-phosphatidylcholine (DPPC) and egg lecithin were from Sigma Chemical Co. (St Louis, MO). Highly purified 1.9 3 106 MW hyaluronan (HA) (lyophilized powder) was from Denki Kaguku Kogyo (Tokyo, Japan); 7.4 3 105 MW HA (10 mg/ml) and 1.7 3 105 MW HA (lyophilized powder) were kindly provided by Fidia SpA (Abano Terme, Italy). HA preparations were used without any further purification. The weight average molecular weights (MWW ) were provided by the nominated manufacturers and were supplied as the sodium salts. Lyophilized powders were reconstituted to 1% (10 mg/ml) sterile aqueous solutions prior to use. Uranyl acetate used as a negative stain was from Fluka AG, Switzerland. All other reagents used were of analytical grade. Double-distilled water was used to prepare solutions and buffers. Phospholipid Preparation (a) Aliquots of DPPC in pure ethanol (20 mg/ml) or of egg lecithin were added to Tyrode’s saline solution, pH 7.3, or to 0.05 M PBS containing 0.15 M Na Cl, pH 7.3, to the final concentration of 0.1, 0.5, and 1 mg/ml of phospholipid. The suspensions were mixed by vortex for 60 sec at room temperature and left at 50°C (DPPC) or at 35°C (egg lecithin) for 60 min, leaving the vials open. During that time, suspensions were mechanically stirred every 15 min for 60 sec. HA of the different molecular weights was added to
the phospholipid suspensions to the final concentrations of 0.1, 0.5, and 1 mg/ml, mixed by vortex for about 60 sec, and incubated under nitrogen at 50°C (DPPC) or at 35°C (egg lecithin) in the dark for times from 1 hr up to 4 days. (b) An aliquot of 20 mg/ml DPPC in pure ethanol was dried under nitrogen gas on the wall of a vial; Tyrode’s or PBS, pH 7.3, was added to reach the DPPC final concentration of 0.1, 0.5, and 1 mg/ml; phospholipid was suspended by sonication for 5 to 20 min in ice bath at power level 6 with a Branson sonifier, Model s-75. The dimensions of PL vesicles were verified by negative staining electron microscopy. HA was added to the phospholipid vesicles and the mixture was mechanically stirred and incubated at 50°C, under nitrogen atmosphere and in the dark, for times from 1 hr to 4 days. (c) DPPC (10 mg/ml) in Tyrode’s or in PBS, pH 7.3, was mechanically suspended by repeated passages at 50°C through a 6-in.-long 20-gauge needle and passed 7 to 10 times through an extruder apparatus (Thermobarrel Extruder, Lipex Biomembranes Inc., Vancouver, Canada), maintained at 50°C, and by the use of 0.1-µm mesh filters and a pressure of 20–40 bars. Vesicles were unilamellar and exhibited a rather homogeneous size (about 15–30 nm in diameter) as checked by negative staining electron microscopy. The phospholipid concentration of the final suspension was checked and vesicles were mechanically mixed with HA of the different molecular weights, at the same concentrations specified before, and were incubated at 50°C, under nitrogen atmosphere and in the dark, for times from 1 hr to 4 days. Negative Staining At different times of incubation, 10 µl of phospholipids or of HA alone or of the mixtures was placed on a holey carbon-coated copper grid and washed with 3–4 drops of 1% uranyl acetate in double-distilled water; after removal of the excess of stain the specimen was air dried and observed in the electron microscope. Rotary Shadowing At different times of incubation, about 100 µl of the various suspensions was gently spread on a 1-cm2 freshly cleaved mica sheet and the excess of material was removed by touching one edge of the mica with filter paper. Specimens were left to air dry and placed into a vacuum apparatus (Balzers, BAF 301 freeze etching). Platinum–carbon was evaporated on the specimen from an angle of 11° under rotation; then carbon was evaporated from 90°, forming a surface replica that was detached by floating on double-distilled water, collected on a copper grid, and observed in a Jeol JEM 1200 EXII electron microscope. RESULTS
As specified under Materials and Methods, hyaluronan was added to either monolamellar or multilamellar phospholipid suspensions. In all cases, the phospholipid organization was deeply affected by the presence of HA and the new structures resulting from the interactions between the two molecular species appeared to be rather independent from the starting phospholipid organization (monolamellar or multilamellar) and from the phospholipid nature (DPPC or egg lecithin), whereas they were markedly influenced by HA molecular weight. Moreover, although the techniques used in this study were not quantitative, it would appear that optimum complex formation occurred when both phospholipid and HA were at a final concentration of 1 mg/ml.
HYALURONAN–PHOSPHOLIPID INTERACTIONS
3
Negative Staining HA molecules could not be visualized by negative staining electron microscopy due to their small molecular dimensions. On the contrary, phospholipid vesicles, made either by extrusion or by sonication, as well as their structural modifications in the presence of HA, were readily seen by this method. DPPC particles prepared by extrusion and stained with 1% uranyl acetate appeared as an homogeneous population of 15- to 30-nm-wide unilamellar vesicles (Fig. 1a) which were stable for at least 4 days at 50°C. After incubation with 170 000 MW HA, for times from 24 to 48 hr, the number of particles on the grid appeared dramatically reduced, whereas large phospholipid aggregates, which could not be better defined by this technique, were observed on the same grids. These aggregates were never observed in the absence of HA. Hyaluronan with MW 740 000 induced the PL particles to form linear aggregates and to fuse. However, the most drastic modifications of PL particles were seen with 1.9 3 106 Da MW HA. After incubation at 50°C for times up to 2 hr, the 15to 30-nm-wide unilamellar vesicles obtained by extruder appeared to align into linear aggregates, forming interconnected short chains that, with time, appeared to fuse into strands, which were thinner than the original particles, exhibited a rather constant thickness of 12 nm, and were organized into a network (Fig. 1b). After 48 hr incubation, no more isolated vesicles could be identified and the material was completely organized into a meshwork of regular 12-nm-wide threads (Fig. 1c). Phospholipid suspensions obtained by sonication consisted of vesicles whose dimensions and structural organization (unilamellar or multilamellar) were dependent on time and frequency of sonication (see Materials and Methods) (Figs. 2a and 2b). When these phospholipid suspensions were incubated with HA, their appearance was modified again depending on the molecular weight of HA used. HA of MW 170 000 Da did not change the organization of DPPC vesicles substantially; however, a series of small globules could be seen on the grid (Fig. 2c). Incubation with the 740 000 MW hyaluronan induced the phospholipid vesicles to exhibit filopodia-like protrusions; that became longer for incubation times up to 24 hr and tended to stick into rather parallel 12-nmwide ‘‘tubules’’ (Fig. 2d). Incubation with the highmolecular-weight hyaluronan induced the phospholid vesicles to aggregate and fuse into ill-defined sheets, which could not be better defined by this technique. Phospholipid vesicles made by mechanical stirring of the suspension at room temperature consisted of huge concentric multilamellar structures, whose organization could not be easily studied by the nega-
FIG. 1. Negative staining of DPPC particles prepared by extrusion (a) and after addition of 1.9 3 106 Da HA (b,c). At different incubation times at 50°C (2 hr (b) and 48 hr (c), respectively) the high-molecular-weight HA induced a progressive fusion of phospholipid vesicles into a final network of 12-nm-wide strands. Bars, 0.05 µm.
tive staining technique, whereas it could be better defined by rotary shadowing. Rotary Shadowing Hyaluronan. Hyaluronan molecules observed by rotary shadowing electron microscopy appeared in form of rather rigid filaments, whose length and distribution on the substrate were strictly related to their molecular mass.
4
PASQUALI-RONCHETTI ET AL.
FIG. 2. Negative staining of sonicated phospholipid suspensions consisting of vesicles with different sizes depending on the sonication procedure used (a, b). In the presence of HA (c, d) liposomes changed their organization. 170 000 MW HA induced fragmentation of liposomes into small phospholipid vesicles and micelles (c); 740 000 MW-HA (d) induced the formation of tubular structures with a tendency to form 12-nm-wide stacks. Bars, 0.05 µm.
The 170 000 MW-HA consisted of isolated short filaments and of 15- to 20-nm-wide globules spread on the mica (Fig. 3a). The 740 000 MW HA appeared in form of globules and filaments which seldom formed discontinuous aggregates (Fig. 3b), while an approximately 2 million MW HA preparation appeared as very long filaments forming a threedimensional array, which flattened on the mica, giving rise to an irregular bidimensional network (Fig. 3c). In all cases the thickness of filaments was rather variable, as they appeared to be composed of a high number of molecules running parallel for long distances and splitting into progressively smaller entities, down to 2-nm-thick filaments, which is the resolution limit of this technique. Very often, highmolecular-weight HA was in the form of huge bundles of rather parallel elements (Fig. 3d). These features clearly indicate that pure HA molecules stick to-
gether through lateral adhesion forces and are consistent with observations of others (Ghosh et al., 1994; Scott, 1992; Scott et al., 1990, 1991). Phospholipid suspensions. By rotary shadowing, DPPC liposomes obtained by extruder or by sonication appeared as roundish globules tending to aggregate on the mica upon dehydration. On the contrary, when suspended in buffer by mechanical stirring, DPPC formed multilayered concentric large vesicles (Fig. 4a), whose size and number of lamellae in each vesicle were rather variable within the same preparation. These vesicles were rather stable with time up to at least 7 days at 50°C. Phospholipid/hyaluronan interactions. Given the high stability of DPPC multilamellar vesicles in buffer with time and their peculiar appearance when observed by rotary shadowing (Fig. 4a) as well as the
HYALURONAN–PHOSPHOLIPID INTERACTIONS
5
corresponding peculiar stable feature of hyaluronan when observed with this same technique, the rearrangement of the two molecular species when mixed together could be clearly and easily appreciated. Multilamellar phospholipid vesicles changed their organization in the presence of HA; these changes were dependent on the molecular weight of HA, were progressive with time, and were accelerated by temperatures higher than that of the melting point of the phospholipid. After 1 hr incubation at 50°C, DPPC membranes in the presence of 170 000 MW HA already showed large areas in the form of coalescent ill-defined globules, without any apparent ordered organization (Fig. 4b). Longer incubation times induced a fragmentation of the phospholipid membranes into small sheets and globules (Fig. 4c) and the appearance of short rigid rods forming small crystalline-like ordered structures (Fig. 4d, arrows). The 740 000 MW HA induced with time a more evident rearrangement of the phospholipid membranes with formation of interconnected tubules, of globules and membrane fragmentation (Fig. 4e). Moreover, these structures coexisted with a relevant number of crystalline-like structures (Fig. 4f ), which were identical to but much larger and longer than those observed with HA of low molecular weight (Fig. 4d). When incubated with the HA of MW ,2 million, the rearrangement of DPPC multilamellar vesicles was rather dramatic. For incubation times up to 1 hr at 50°C, the great majority of PL were still organized into multilamellar vesicles. However, 12-nm large threads of variable length were often observed at the periphery of or in close association with the phospholipid membranes, similar to those observed in the presence of 7.4 3 105 MW HA (Fig. 4e). For incubation times longer than 24 hr, the multilamellar DPPC vesicles almost disappeared and the specimen consisted of extremely large holey membranes (Fig. 4g), together with enormous numbers of crystallinelike structures (Fig. 4h). These were similar to, but much larger than, those already described for the low-molecular-weight HA–PL interactions (Figs. 4d and 4f ). The paracrystalline structures, in the form of flat or overlapping sheets on the mica, were large enough to be analyzed by optical diffraction, revealing a lateral periodicity of 12 nm. These crystallinelike structures may be interpreted as 12-nm-wide
FIG. 3. Rotary shadowing of HA molecules of different molecular weights. The 170 000 Da HA (a) consisted of short filaments and small globules; the 740 000 Da HA was composed of globules and rather long filaments (b); the 1.9 3 106 Da HA (c, d) formed long strands of adjacent filaments (arrows) that dissociated, giving rise to a spread network (c). Bars, 1 µm.
FIG. 4. Rotary shadowing of DPPC vesicles, prepared by vortexing, in the absence (a) and in the presence (b–h) of HA of different molecular weights. With time, the 170 000 Da HA (b–d) fragmented the phospholipid bilayers into small globules (b and c) and a few small crystalline-like structures (d, arrows). The effects of 740 000 MW HA were even more evident (e and f ). As already observed by negative staining, the most striking changes were observed with hyaluronan of 1.9 3 106 Da (g–h), in the presence of which phospholipids rearranged into huge perforated membranes (g) and sheets of 12-nm-wide cylinders (h). Bars, 1 µm.
HYALURONAN–PHOSPHOLIPID INTERACTIONS
‘‘cylinders,’’ which tended to aggregate by lateral apposition and to form extended roller sheets on the substrate. When egg lecithin multilayered vesicles obtained by vortexing (Fig. 5a) were incubated with the 2 million MW HA at 35°C, for times longer than 48 hr, phospholipid molecules appeared to be organized into huge thready networks (Fig. 5b), extended holey membranes (Fig. 5c), or crystalline-like structures (Fig. 5d). All these deep modifications of the egg lecithin membranes in the presence of HA were very similar to those already described for DPPC–HA interactions. DISCUSSION
Dipalmitoylphosphatidylcholine and egg lecithin molecules were shown to interact with hyaluronan, giving rise to complexes which might have some biological relevance. The interaction would seem to be independent of the PL used, since DPPC and egg lecithin afforted similar results and independent from ions of the medium (PBS or Tyrode’s). Moreover, the interactions would seem to be energetically favorable as they occurred irrespective of the previous organization of the phospholipid molecules (in the form either of large multilamellar structures or of unilamellar vesicles) and did not require sonication or other drastic procedures for formation. The data obtained can be interpreted as interactions between the two molecular species in which PL molecules aggregate around HA, the assembling being dependent on the length of HA molecules. Low-molecular-weight HA appeared to bind PL molecules favoring the fragmentation of large phospholipid membranes and the formation of globules with a tendency to aggregate. On the other hand, high-molecular-weight HA, which consisted of long filaments interconnected into a loose three-dimensional network, seemed to organize around them phospholipid molecules forming 12-nm-thick threads, which could either form a network or align parallel into large sheets. These features were observed by both negative staining (Figs. 1c, 1d, and 2d) and rotary shadowing (Figs. 4e, 4f, and 4h) electron microscopy. When observed by this latter technique, which provides a three-dimensional image of the objects, threads may be interpreted as ‘‘cylinders,’’ whose organization could correspond to that shown in Fig. 6a. The thickness and the appearance of a
FIG. 5. Rotary shadowing of lecithin vesicles, obtained by mechanical stirring (a), and incubated for 48 hr with hyaluronan of 1.9 3 106 Da (b–d). HA exherted structural modifications of the phospholipid bilayers similar to those previously observed with DPPC vesicles, such as a thready network (b), holey membranes (c), and crystalline-like structures (d). Bars, 1 µm.
7
8
PASQUALI-RONCHETTI ET AL.
FIG. 6. Proposed model for HA–lipid interactions. (a) HA molecules form a central filament surrounded by a bilayer of phospholipid molecules radially oriented along the central core of HA. The charged ends of the phospholipid molecules allow formation of a thin layer of water between the adjacent cylinders. (b) An inverted hexagonal lattice (HII phase) of dipalmitoylphosphatidylcholine induced by HA. This model is less probable, as discussed in the text.
single cylinder were consistent with complexes formed by one (or more) HA molecules acting as a central filament surrounded by a bilayer of phospholipid molecules radially oriented along the central HA core. As shown in Fig. 6a, the charged ends of the phospholipid molecules would favor a water layer between adjacent cylinders. Moreover, the dependence of the interactions on the molecular weight of HA is consistent with the model proposed in Fig. 6a. The longer the HA chain the more phospholipid molecules attached, which would result in more extensive and thus stronger intermolecular phospholipid interactions compatible with sheet formation. In any case, it cannot be excluded that the formation of rollers may be favored by phospholipid/mica interactions during drying. The idea that cylinders may represent inverted hexagonal phases (HII phase) (Seddon, 1990) (Fig. 6b) of phospholipids seems less probable for the following reasons. First of all, the phospholipid concentration was too low compared to that required for an inverted hexagonal lattice to occur, unless the cylinder is considered the result of local phospholipid concentration by progressing dehydration on the mica. In this case, however, cylinders would also have been observed in pure DPPC specimens. Second, 50°C for DPPC is not a tempera-
ture compatible with the inverted hexagonal phase (Luzzati, 1968) and because dehydration was performed at room temperature. Finally, the spacing for the inverted hexagonal lattice has been measured for dioleylphosphatidylethanolamine at 22°C from 4.0 to 6.2 nm, depending on the water content (Rand and Fuller, 1994). The cylinders illustrated in the present report are always 12 nm wide; that is, double that of a membrane bilayer. Another interesting feature obtained by incubation of phospholipid with medium and high-molecular-weight HA was the formation of huge perforated membrane-like sheets whose complexity could not be solved by the techniques used, but which might have important biological significance, as they could be the equivalent of the discontinuous membrane-like structures recently described as covering the surface of normal articular cartilage in diarthroic joints and which were shown by immunocytochemistry to be associated with HA (Guerra et al., 1996). The interaction of HA and the phospholipid DPPC has also been studied by gel-permeation chromatography, low-angle laser–light scattering, and 13CNMR techniques (Ghosh et al., 1994). These studies showed that the addition of DPPC to highly purified HA increased the flexibility of HA. It was suggested that this was due to competition of PL for intermolecular hydrophobic sites on the HA molecules involved in interchain associations. In the present study, the charged polar region of the phospholipid is considered to interact with HA, although the model shown in Fig. 6a does not exclude hydrophobic bonding by interactions between the hydrophobic patches on hyaluronan with the first angulated hydrocarbons on the phospholipid acyl chains (J. E. Scott, personal observation). The synovial fluid of normal joints contains several molecular species, excluding the proteins, the most abundant of which is HA. Hyaluronan is considered to play a critical role in the physiology of joint function, including lubrication of the synovial surfaces (Balazs et al., 1967; Laurent et al., 1995; Ogston and Steiner, 1953), although alternative glycoproteins have been proposed by others (Swann et al., 1981a,b). In addition, synovial fluid contains phospholipids (Bole, 1962; Wise et al., 1987) whose concentrations increase under pathological conditions (Wise et al., 1987); the phospholipids in synovial fluid have been proposed as the molecules responsible for articular cartilage ‘‘boundary’’ lubrication (Hills, 1989, 1990, 1995), and it has been suggested that the presence of HA may ‘‘fortify’’ lamellar lubrication (Hills, 1989, 1995). The model proposed in Fig. 6a is clearly compatible with the observations of Hills (1989, 1995) since the ability of HA to orientate phospholipid molecules into roller
9
HYALURONAN–PHOSPHOLIPID INTERACTIONS
structures, which can aggregate into sheets, might provide ideal conditions for lamellar lubrication. The aggregated rollers assembled as a layer would be analogous to graphite sheets which when subjected to shearing stresses would separate from each other with minimal friction. The presence of phospholipid–HA aggregates in the form either of rollers or of perforated membranelike sheets between the articulating surfaces of diarthroidal joints would be capable of providing excellent lubrication over a wide dynamic range and excellent protection of the cartilage surface from enzymes and free radicals in the synovial fluid. Rather interestingly, however, as shown in the present study, HA preparations with low molecular weights (around 170 000 or less) were less efficient in forming large membrane-like and roller sheets with phospholipids than the high-molecular-weight HA preparations. This finding may have important biological implications, as it is known that highmolecular-weight species of HA are degraded to low-molecular-weight species during joint inflammation and in osteoarthritis (Balazs et al., 1967; Dahl et al., 1985; Tulamo et al., 1994; Wise et al., 1987). Under pathological conditions, it would be expected that hyaluronan–phospholipid interactions would not produce the same sheets and rollers which are present under normal conditions and both lubrication and cartilage protection would be impaired. The injection of pure high-molecular-weight HA into osteoarthritic joints for symptomatic relief (Govoni et al., 1990; Peyron, 1993) would be expected to reestablish membrane-like and roller structures with endogenous phospholipids even though the majority of the free HA was observed to be cleared from the joint within a few days (Laurent et al., 1992). While the clearance rate of phospholipid–HA complexes from the synovial cavity is unknown, in this regard the observation by Laurent and co-workers (1992) that 2–5% HA injected into joints was retained and only slowly metabolized after intra-articular injection is noteworthy. Notwithstanding the ease with which these phospholipid–HA aggregates can be prepared in vitro, extrapolation to the in vivo situation must remain speculative until studies with normal and pathological synovial fluids have been completed. Studies in this direction are in progress and preliminary results confirm that phospholipids extracted and purified from normal bovine synovial fluid do exhibit a tendency to form aggregates with high-mass HA and that these aggregates have characteristics similar to those described in the present report. Funds for this study were from the Italian CNR and from the University of Modena (Progetto di Ricerca Applicata, 1995).
REFERENCES Agostoni, E. (1972) Mechanics of the pleural space, Physiol. Rev. 52, 57–128. Agostoni, E., Miserocchi, G., and Bonanni, M. V. (1969) Thickness and pressure of the pleural liquid in some mammals, Respir. Physiol. 6, 245–256. Balazs, E. A. (1974) The physical properties of synovial fluid and the special role of hyaluronic acid, in Helfet, A. (Ed.), Disorders of the Knee, 1st ed., pp 61–74, Lippincott, Philadelphia. Balazs, E. A., and Denlinger, J. L. (1984) The role of hyaluronic acid in arthritis and its therapeutic use, in Peyron, J. G. (ed.), Osteoarthritis-Current Clinical and Functional Problems, pp 165–174, Ciba Geigy, Paris. Balazs, E. A., Watson, D., Duff, I. F., and Roseman, S. (1967) Hyaluronic acid in synovial fluid. I. Molecular parameters of hyaluronic acid in normal and arthritic human fluids, Arthritis Rheum. 10, 357–376. Bole, G. G. (1962) Synovial fluid lipids in normal individuals and patients with rheumatoid arthritis, Arthritis. Rheum. 5, 589– 601. Dahl, L. B., Dahl, I. M. S. Engstrom-Laurent, A., and Granath, K. (1985) Concentration and molecular weight of sodium hyaluronate in synovial fluid from patients with rheumatoid arthritis and other arthropaties, Ann. Rheum. Dis. 44, 817–822. Ghosh, P., Hutadilok, N., Adam, N., and Lentini, A. (1994) Interactions of hyaluronan (Hyaluronic acid) with phospholipids as determined by gel permeation chromatography, multiangle laser-ligh-scattering photometry and 1H-NMR spectroscopy, Int. J. Biol. Macromol. 16, 237–244. Gibbs, D. A., Merrill, E. W., Smith, K. A., and Balazs, E. A. (1968) Rheology of hyaluronic acid, Biopolymers 6, 777–791. Govoni, E., Pasquali Ronchetti, I., Georgountzos, A., Perbellini, A., Cicchetti, F., Guerra, D., Vincenzi, D., Franchini, S., and Frizziero, L. (1990) Microarthroscopic and histopathological evaluation of synovial membrane in knee joint osteoarthritis after intraarticular treatment with hyaluronic acid, First Int. Symp. on Synovium, Philadelphia. Gross, J. (1948) Electron microscope studies of sodium haluronate. J. Biol. Chem. 172, 11–514. Guerra, D., Losi, M., Bacchelli, B., Mezzadri, G., and Pasquali Ronchetti, I. (1996) Ultrastructural identification of a membrane-like structure covering the surface of normal articular cartilage, J. Submicrosc. Cytol. Pathol. 28, 385–393. Hills, B. A. (1989) Oligolamellar lubrication of joints by surface active phospholipid, J. Rheum. 16, 82–90. Hills, B. A. (1990) Oligolamellar nature of the articular surface, J. Rheum. 17, 349–356. Hills, B. A. (1995) Remarkable anti-wear properties of joint surfactant, Ann. Biomed. Eng. 23, 112–115. Hills, B. A., and Butler, B. D. (1985) Phospholipids identified on the pericardiumm and their ability to impart boundary lubrication, Ann. Biomed. Eng. 13, 573–586. Hills, B. A., Butler, B. D., and Barrow, R. E. (1982) Boundary lubrication imparted by pleural surfactants and their identification, J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 53, 463–469. Kobayashi, Y., Okamoto, A., and Nishinari, K. (1994) Viscoelasticity of hyaluronic acid with different molecular weight, Biorheology 31, 235–244. Laurent, U. B. G., Fraser, J. R. E., Engstrom-Laurent, A., Reed, R. K., Dahl, L. B., and Laurent, T. C. (1992) Catabolism of hyaluronan in the knee of the rabbit, Matrix 12, 130–136.
10
PASQUALI-RONCHETTI ET AL.
Laurent, T. C., and Fraser, J. R. E. (1992) Hyaluronan, FASEB J. 6, 2397–2404. Laurent, T. C., Laurent, U. B. G., and Fraser, J. R. E. (1995) Functions of hyaluronan, Ann. Rheum. Dis. 54, 429–432. Little, T., Freeman, M. A. R., and Swanson, S. A. R. (1969) Experiments on friction in the human hip joint, in Wright V. (Ed.), Lubrication and Wear in Joints, pp. 110–122, Sector, London. Luzzati, V. (1968) Chapman D. (Ed.), Biological Membranes, Vol. I, Academic Press, London. McDonald, J. N., and Levick, J. R. (1994) Hyaluronan reduces fluid escape rate from rabbit knee joints disparately from its effect on fluidity, Exp. Physiol., 79, 103–106. Ogston, A. G., and Stanier, J. E. (1953) The physiological function of hyaluronic acid in synovial fluid: Viscous, elastic and lubricant properties, J. Physiol. 119, 244–252. Peyron, J. G. (1993) Intraarticular hyaluronan injections in the treatment of osteoarthritis: State of-the-art review, J. Rheumatol. 20, 10–15. Rabinowitz, J. L., Gregg, J. R., and Nixon, J. E. (1983) Lipid composition of the tissues of human knee joints. II. Synovial fluid in trauma, Clin. Ortho. 190, 292–298. Rand, R. P., and Fuller, N. L. (1994) Structural dimensions and their changes in a reetrant hexagonal-lamellar transition of phospholipids, Biophys. J. 66, 2127–2138. Scott, J. E. (1992) Supramolecular organization of extracellular
matrix glycosaminoglycans in vitro and in the tissues, FASEB J. 6, 2639–2645. Scott, J. E., Cummings, C., Brass, A., and Chen, Y. (1991) Secondary and tertiary structure of hyaluronan in aqueous solution, investigated by rotary shadowing-electron microscopy and computer simulation. Hyaluronan is a very efficient network-forming polymer, Biochem. J. 274, 699–705. Scott, J. E., Cummings, C., Geiling, H., Stuhlsatz, H. W., Gregory, J. D., and Damle, S. P. (1990) Examination of corneal proteoglycans and glycosaminoglycans by rotary shadowing electron microscopy, Int. J. Biol. Macromol. 12, 180–184. Seddon, J. M. (1990) Structure of the inverted hexagonal (HII) phase and non-lamellar phase transitions of lipids, Biochim. Biophys. Acta 1031, 1–69. Swann, D. A., Hendred, R. B., Radin, E. L., Sotman, S. L., and Duda, E. A. (1981a) The lubricating activity of synovial fluids, Arthritis Rheum., 24, 22–30. Swann, D. A., Slayter, S. H., and Silver, F. H. (1981b) The molecular structure of lubricating glycoprotein 1, the boundary lubricant for articular cartilage, J. Biol. Chem. 256, 5921–5925. Tulamo, R. M., Heiskanen, T., and Salonen, M. (1994) Concentration and molecular weight distribution of hyaluronate in synovial fluid from clinically normal horses and horses with diseased joints, Am. J. Vet. Res. 55, 710–715. Wise, C. M., White, R. E., and Agudelo, C. A. (1987) Synovial fluid lipid abnormalities in various disease states: Review and classification, Semin. Arthritis Rheum. 16, 222–230.