Effect of depletion of glycosaminoglycans and non-collagenous proteins on interstitial hydraulic permeability in rabbit synovium

Abstract

1. The hydraulic resistance of synovial interstitium helps to retain a lubricating fluid within the joint cavity. The contributions of sulphated glycosaminoglycans to resistance were assessed by selective depletion by chondroitinase ABC, keratanase and heparinases I, II and III in vivo. Also, since glycosaminoglycans do not account fully for the resistance, the contribution of non-collagenous, structural proteins in interstitium was assessed by treatment with chymopapain, a collagen-sparing protease.

2. Ringer solution containing enzyme was injected into the synovial cavity of the knee in anaesthetized rabbits. After ≥ 30 min the intra-articular pressure was raised and the relation between pressure (P_j) and trans-synovial outflow (Q̇_s) determined. The slope dQ̇_s/dP_j at low pressures, i.e. below yield pressure, represents the hydraulic conductance of the lining, i.e. 1/resistance. The contralateral joint received Ringer solution without enzyme as a control. Action of enzymes on the tissue was confirmed by histochemical and immunohistochemical studies.

3. Treatment with chondroitinase ABC (5 joints) increased the hydraulic conductance of the lining by 2.3 times (control, 1.34 ± 0.22 μl min⁻¹ cmH₂O⁻¹; post-enzyme, 3.11 ± 0.45 μl min⁻¹ cmH₂O⁻¹). This was significantly less than the effects of leech, Streptomyces and testicular hyaluronidases, which caused an average 4.7 times increase (P < 0.001, ANOVA). Analogous findings were made above yield pressure.

4. Treatment with keratanase (3 joints) or heparinases I, II and III (3 joints) caused no significant increase in trans-synovial flows or conductance, even though the concentration of heparan sulphate in synovium is higher than that of chondroitin sulphates or hyaluronan.

5. Treatment with chymopapain (7 joints) caused the greatest increases in trans-synovial flow, which exceeded control flow by an order of magnitude in one case. After 0.1 U chymopapain the average conductance was 6.6 times the control conductance below yield pressure. Immunohistochemical studies confirmed that chymopapain treatment removed the synovial proteoglycans.

6. It is concluded that, despite their similar resistivities in vitro, the different glycosaminoglycans do not contribute equally, weight for weight, to interstitial resistance in vivo. Hyaluronan is the dominant glycosaminoglycan governing synovial interstitial resistance. In addition, non-collagenous structural proteins contribute significantly to interstitial resistance.

The hydraulic resistance of interstitium influences many aspects of body fluid physiology, including fluid transport between microcirculation and lymph, fluid drainage from specialized cavities like the anterior eye chamber or the synovial joint cavity, and the preservation of the normal, non-pitting biomechanical state of soft tissues. This last function is particularly highly developed in articular cartilage, where an unusually high concentration of sulphated glycosaminoglycans (chondroitin and keratan sulphates) generates a very high interstitial hydraulic resistance (Maroudas, 1980; Buchmann et al. 1995). In certain tissues, however, the non-sulphated glycosaminoglycan hyaluronan is more important than chondroitin and keratan sulphates in determining the hydraulic resistance, e.g. in the anterior angle of the eye (Knepper et al. 1984). Although there is a popular acceptance of the dominant role of sulphated/non-sulphated glycosaminoglycans in determining tissue resistance, protease digestion studies show that non-collagenous interstitial proteins, presumably proteoglycan core proteins and glycoproteins, also contribute significantly to interstitial resistance (Day, 1952; Hedbys, 1963). Similarly, biochemical and biophysical data show that, in order to account quantitatively for interstitial resistance in most tissues, the interstitial non-collagenous proteins, as well as glycosaminoglycans, have to be taken into account (Levick, 1987).

In joints, the interstitial resistance of synovium is important because the synovial lining must encapsulate and retain the lubricating synovial fluid. The cells of the lining are separated by gaps several micrometres wide that form 20–33 % of the free surface. Thus the hydraulic resistance of the matrix plugging the gaps is the key factor retarding fluid escape from the joint cavity. Analysis of rabbit synovium shows that the matrix contains approximately 0.8 mg hyaluronan, 1.2 mg chondroitin sulphates and 1.9 mg heparan sulphate per millilitre of extrafibrillar space, making a net glycosaminoglycan concentration of approximately 4 mg ml⁻¹ (Price et al. 1996a). The hydraulic importance of some of these glycosaminoglycans is confirmed by a 5-fold increase in the hydraulic conductance of the lining after treatment with testicular hyaluronidase. Testicular hyaluronidase removes both chondroitin sulphates and hyaluronan from the matrix (Scott et al. 1997). Treatment with Streptomyces and leech hyaluronidases causes equally large increases in conductance, even though these enzymes degrade only hyaluronan (Coleman et al. 1998b). It was inferred, therefore, that chondroitin sulphates may contribute less than hyaluronan to interstitial hydraulic resistance in synovium. The first objective of the present study was to evaluate further the hydraulic significance of the individual sulphated glycosaminoglycans, by studying the effects of chondroitinase ABC, keratanase and heparinase on synovial lining conductance. The presence of keratan sulphate in rabbit synovium has only recently been confirmed (Coleman et al. 1998a).

The 4 mg ml⁻¹ glycosaminoglycan found in synovial interstitium is not enough to account fully for the tissue resistance indicated by pressure-flow studies. A mathematical model of synovial resistance requires a total biopolymer concentration of ∼14 mg ml⁻¹, if the biopolymer is uniformly distributed (Levick, 1994). It seems likely, therefore, that additional biopolymers, particularly interstitial non-collagenous proteins such as proteoglycan core proteins and glycoproteins, contribute substantially to hydraulic resistance in synovium, as they do in many other tissues (Levick, 1987). If this proposal is correct, proteolysis of non-collagenous proteins should cause a large rise in interstitial hydraulic conductance, i.e. 1/hydraulic resistance. Indeed, from the above estimates of concentration, the effect of proteolysis could exceed that of glycosaminoglycan depletion. A second objective of the present study was to test this by measuring the conductance of the joint lining after digestion with chymopapain, a collagen-sparing protease from the papaya tree, Carica papaya (Stern, 1969). The effect of protease on synovial lining permeability is also of clinical interest, because the activity of metalloproteinases is greatly raised in arthritic joint effusions (Lohmander et al. 1993).

METHODS

The animal preparation and general procedures were described in detail by Coleman et al. (1998b). Experiments complied with UK animal legislation and animals were killed at the end of the pressure-flow studies by an overdose of i.v. pentobarbitone before excision of samples for immunohistochemistry.

Materials

The control intra-articular infusate and enzyme vehicle was sterile, non-pyrogenic Baxter Ringer solution, which contains 147 mM sodium, 4 mM potassium, 2 mM calcium and 156 mM chloride, pH 7.2 (Baxter Healthcare Ltd, Thetford, Norfolk, UK). Chondroitinase ABC (EC 4.2.2.4) was obtained from ICN Flow (High Wycombe, UK). All other enzymes, labelled antibodies and other reagents were purchased from Sigma. The monoclonal antibodies and biotinylated hyaluronan binding region (HABR, the G1 domain of aggrecan) were gifts; see Acknowledgements.

Animal preparation and determination of synovial pressure-flow relation

New Zealand white rabbits (2–4 kg, ∼2.5-8 months old) were anaesthetized by i.v. pentobarbitone (30 mg kg⁻¹) and urethane (500 mg kg⁻¹), tracheostomized and maintained by smaller half-hourly doses. Two cannulae were inserted into the suprapatellar bursa. One was connected to a water-calibrated Validyne CD 15 differential pressure transducer (Linton Instruments, Diss, Norfolk) to measure intra-articular fluid pressure (P_j, ± 0.1 cmH₂O). The other was connected to a saline-filled 20 ml infusion reservoir, the vertical height of which controlled intra-articular pressure. Flow of Ringer solution from the reservoir into the joint cavity, Q̇_in, was recorded by an interposed photoelectric drop counter with a drop size of 13.3 μl. Pressures and flows were recorded on a chart recorder.

The pressure-flow relation was determined over imposed values of P_j ranging from approximately 2 cmH₂O above atmospheric pressure, which is reached in a normal flexed joint and is the lowest pressure at which flow is measurable, to 22–25 cmH₂O, which is reached in some joint effusions. Pressure was raised in steps of 2–4 cmH₂O by raising the infusion reservoir vertically by 2–4 cm at 15 min intervals. After the initial, fast, filling transient, inflow stabilized. Trans-synovial flow in the steady state at 15 min, Q̇_s, was calculated from the sustained rate at which the joint took up fluid from the reservoir (Q̇_in). A small fraction of Q̇_in at 15 min is caused by viscous creep of the stressed cavity walls, and Q̇_s was obtained by subtracting the volume creep rate (Q̇_creep), i.e. Q̇_s=Q̇_in - Q̇_creep. Creep rate (μl min⁻¹) is given by the relation Q̇_creep= 0.23P_j+ 0.4 (Levick, 1979). The creep correction was only a small percentage of the measured flows.

Enzymatic digestion protocols

In each animal one knee received an intra-articular injection of 500 μl of control solution (vehicle, Ringer solution) and the other received 500 μl of enzyme in vehicle. The order of the treatment (control or enzyme) was varied. After 30 min or more (see below) the pressure-flow relation was determined. Previous work with the enzymes testicular and Streptomyces hyaluronidases showed that their effect on synovial conductance was complete within 3 min of intra-articular injection, with no further effect over 30 min (Coleman et al. 1998b). Chondroitinase ABC was studied in five rabbits, keratanase in three rabbits, heparinases in three rabbits (see below), and chymopapain in seven joints of six rabbits.

Chondroitinase ABC (EC 4.2.2.4) breaks the β(1→4) glycosidic bond between N-acetylgalactosamine and glucuronic acid by an eliminase reaction, which causes degradation to unsaturated disaccharides. The enzyme degrades chondroitin 4-sulphate, chondroitin 6-sulphate, dermatan sulphate and, more slowly, hyaluronan (Yamagata et al. 1968; Meyer, 1971; Derby & Pintar, 1978; Knepper et al. 1984). The protease activity was assessed by azocasein proteolysis and was negligible (Coleman et al. 1998b). The eliminase activity was confirmed by the permeability enhancement and by immunohistochemistry. Treatment consisted of exposing the joint lining in vivo to 10 U of enzyme for 30 min. Chondroitin sulphate content is 0.55 mg (g synovium)⁻¹ (Price et al. 1996a), from which the content of the whole lining is calculated to be 0.025-0.043 μmol disaccharide. Since 10 U of enzyme by definition release 10 μmol disaccharide min⁻¹ at 37°C and pH 8, the treatment was more than adequate to degrade the chondroitin sulphate in synovium.

Keratanase (EC 3.2.1.103; 19 U joint⁻¹) is an endo-β-galactosidase, 1 U of which releases 1 μmol galactose h⁻¹ at 37°C and pH 7.4. The activity of the commercial preparation was confirmed by keratan sulphate depletion in immunohistochemical preparations of treated synovium (see below).

Heparinase activity was confirmed in vitro by reduction of the relative viscosity of a heparan sulphate solution to ∼1 by the enzyme preparation in < 45 min. To cover the spectrum of heparan sulphates, which are a diverse group of glycosaminoglycans, 50 U heparinase I (EC 4.2.2.7) was given intra-articularly for 40 min and was followed by 10 U heparinase II for an additional 40 min in the first animal, prior to the pressure-flow measurements. Since this treatment had no detectable effect on permeability, the treatment was expanded to include three heparinases. Intra-articular injections of 50 U heparinase I, 10 U heparinase II and 10 U heparinase III (EC 4.2.2.8) were given sequentially in two more animals. A period of 40 min was allowed between each injection. The pressure-flow relation was determined after a total digestion period of 2 h. Each heparinase cleaves a different linkage within heparin and heparan sulphates. The heparinases yield disaccharides when acting in concert (Lohse & Linhardt, 1992). One unit of heparinase releases, by definition, 0.1 μmol uronic acid h⁻¹ at 25°C and pH 7. The heparan sulphate content of the synovial lining, given a concentration of 0.92 mg (g synovium)⁻¹ (Price et al. 1996a), is calculated to be 0.042-0.072 μmol disaccharide.

Chymopapain activity was confirmed by the release of 41 % of the azo dye from azocasein in a standard test in vitro (Coleman et al. 1998b). Chymopapain (EC 3.4.22.6; molecular size, 27–45 kDa; active pH range, 6.5-8.5) has a narrower specificity than papain and has been used clinically to degrade proteoglycans in intervertebral discs (‘chemonucleolysis’; Watts et al. 1975; Oegema et al. 1988). It rapidly degrades proteoglycans, and possibly also matrix glycoproteins, but does not attack collagen to any significant degree, as estimated by hydroxyproline release (Bradford et al. 1984). Its activity depends on -SH groups. The enzyme is activated by reducing agents such as cysteine, and it requires a chelating agent such as sodium edetate for maximal activity. For the chymopapain experiments, therefore, the vehicle was changed to Ringer solution containing 10 mM L-cysteine hydrochloride and 10 mM EDTA.

To test whether the activating agents themselves altered permeability, e.g. by activating endogenous degradative enzymes, an experiment was carried out in which the control joint received the cysteine-EDTA vehicle, while the opposite joint received the cysteine-EDTA vehicle containing 0.1 U chymopapain. Since the vehicle had no discernible effect on permeability (see Results), control joints were injected with ordinary Ringer solution in all other animals. Similarly, chymopapain-activating solution has no significant effect in the nucleus pulposus of human intervertebral discs (Bradford et al. 1984).

Joints were treated with chymopapain in doses of 10 U (3 animals), 5 U (1 animal), 0.5 U (1 animal) and 0.1 U (2 animals, with a final injection to 0.2 U in 1 case). Decreasing concentrations were explored because 10 U, the dose first studied, caused marked intra-articular bleeding within a minute or so. This caused a slow rise in intra-articular pressure with time, following the initial rise in pressure due to the volume of injectate. Since it was not possible to measure trans-synovial outflow when pressure was changing, it was necessary to wait 1–1.5 h, during which time the intra-articular bleeding slowed or ceased, as shown by a falling pressure. (After an initial step in pressure due to the injected volume, pressure normally falls slowly, owing to net fluid absorption.) The pressure-flow relation was then determined. Intra-articular haemorrhage was greatly reduced after 0.5 U chymopapain and was negligible after 0.1 U (see Results).

The non-collagenous protein concentration is 41.3 mg (g synovium)⁻¹ (Price et al. 1996a). The total mass is 0.95-1.61 mg joint⁻¹, of which < 0.18 mg (11 %) may be interstitial protein and the rest intracellular. By definition, 1 U chymopapain degrades 0.34 mg of test substrate, benzoyl-arginine ethyl ester, per minute at 25°C and pH 6.2. On this basis, and in view of the huge increases in permeability, even the lowest concentration of enzyme was sufficient to digest the interstitial non-collagenous proteins over 1.5-2 h.

Preparation of tissue for microscopy

After measurement of the pressure-flow relation on each side, samples of synovium and underlying tissue were excised from the lateral and medial aspects of the suprapatellar bursa. Samples were fixed in 4 % paraformaldehyde in 0.05 M Tris-HCl buffer, pH 7.2, overnight at room temperature. The tissue was washed in buffer, and dehydrated in graded ethanols, cleared in methyl salicylate, embedded in paraffin wax and sectioned at 7 μm. Full details of the histochemical and immunohistological procedures were described by Coleman et al. (1998a, b).

The hyaluronan content of the tissue was determined histochemically using HABR as a probe (see Materials). This binds to hyaluronan. Bound probe was detected by incubation with streptavidin peroxidase and a standard peroxidase substrate.

The presence of chondroitin sulphate proteoglycan was assessed immunohistochemically using mouse monoclonal antibodies to the chondroitin 4-sulphate attachment region (antibody 2-B-6) and the chondroitin 6-sulphate attachment region (antibody 3-B-3). The epitopes of these antibodies are oligosaccharides of chondroitin sulphate that remain attached to the core protein after digestion of the section by chondroitinase ABC. The antibodies do not detect the native chondroitin sulphate chains (Caterson et al. 1985). Therefore, to reveal the epitopes, the tissue sections were digested with chondroitinase ABC.

When synovium had been exposed to intra-articular chondroitinase ABC in vivo, treatment of tissue sections by chondroitinase ABC was omitted in some cases. Binding of the antibody in these sections indicated that the enzyme digestion in vivo had removed the main glycosaminoglycan chains.

The presence of keratan sulphate was assessed with monoclonal antibody 5-D-4. The epitope of 5-D-4 is the main glycosaminoglycan chain. Unlike antibodies 2-B-6 and 3-B-3 no digestion of sections is required to reveal the epitope in vitro (Mehmet et al. 1986). The second antibody for both the chondroitin sulphate and keratan sulphate primary antibodies was goat anti-mouse IgG conjugated with alkaline phosphatase. No anti-heparan sulphate antibodies were available to us.

Statistical analysis

In control joints the P_j-Q̇_s relation commonly shows a marked steepening at about 7–14 cmH₂O (‘yield pressure’) with little or inconsistent curvature above or below this. Consequently the relation is usually represented, for purposes of numerical comparison, by two linear regressions, one fitted to results above yield pressure (high pressure range) and the other below it (low pressure range) (Edlund, 1949; Levick, 1979). Yield pressure was determined by inspection. Regression slopes were compared by Student's paired t test or repeated measures one-way analysis of variance (ANOVA) with Bonferonni's post hoc test as implemented in Graphpad Prism (Graphpad Software Inc., San Diego, USA), with P < 0.05 accepted as a significant difference. The infusion-driven intra-articular pressures varied a little between experiments, so to compare flows at identical pressures between experiments the flows were interpolated to standard values, namely 2.5, 5.0, 7.5 cmH₂O etc., by linear interpolation between the two bounding measurements. Means are followed by standard error of mean (s.e.m.) throughout. Ratios were compared using non-parametric tests.

RESULTS

Normal structure of synovium

The normal joint lining consists of a sheet of cellular tissue that is two or three cells thick and has an abundant interstitial matrix (see control sections in Figs 1A, 1C, 6A and 6E). The histochemical and immunohistochemical results showed that the normal matrix contained hyaluronan (Fig. 1A), keratan sulphate (Fig. 1C), chondroitin 4-sulphate (not shown; see Coleman et al. 1998b) and chondroitin 6-sulphate (Fig. 6E). Within the synovium, and close to the free surface, there were abundant capillaries and postcapillary venules (Levick, 1995). Beneath the synovium there was a less dense, wider, less vascular layer of loose connective tissue, the subsynovium.

Figure 1. Photomicrographs of synovium (S), subsynovium (SS) and underlying muscle (M) from the suprapatellar bursa of the knee joint cavity (JC), showing effects of treatment by glycosaminoglycanidases.

A, control synovium showing normal distribution of hyaluronan (HABR preparation). B, synovium treated with chondroitinase ABC in vivo. Only small amounts of hyaluronan remain (HABR preparation). The hydraulic conductance of this tissue, measured prior to sampling, was raised. C, control synovium showing normal distribution of keratan sulphate (5-D-4 antibody). D, synovium treated with keratanase in vivo, showing absence of keratan sulphate (5-D-4 antibody). Synovial hydraulic conductance was not significantly changed. All photomicrographs are at the same magnification; bar in A, 10 μm.

Figure 6. Effect of chymopapain on synovial structure and immunohistochemistry.

A, the normal, control synovium (S) is a thin layer of cells and interstitium with small blood vessels (arrows). The underlying subsynovium (SS) is a loose connective tissue with few cells. It lies on muscle (M). (Haematoxylin and Eosin staining.) B, synovium and subsynovium after treatment with 0.1 U chymopapain in vivo are oedematous and thickened. The cells are more widely spaced and the blood vessels (arrows) are enlarged. The hydraulic conductance of this tissue, measured prior to excision, was greatly increased. (Haematoxylin and Eosin staining.) C, synovium after treatment with 10 U chymopapain in vivo is disrupted. The cells that remain on the surface appear to be embedded in a hyaline layer (arrows). The subsynovium is filled with red cells. The blood vessels (BV) are very swollen. (Haematoxylin and Eosin staining.) D, after treatment with 0.1 U chymopapain in vivo much of the hyaluronan is removed from the synovium (HABR preparation). Compare with Fig. 1A. E, control synovium, showing the distribution of chondroitin 6-sulphate oligosaccharides (3-B-3 antibody). F, after treatment with 0.1 U chymopapain in vivo, little if any chondroitin 6-sulphate remains (3-B-3 antibody). All photomicrographs are at the same magnification; bar in A, 10 μm.

Effects of chondroitinase ABC treatment

Physiological results

Treatment with chondroitinase ABC increased the trans-synovial flow markedly at any given pressure (Fig. 2A). The mean flow in five pairs of joints, and the statistical significance of the flow increases, are shown in Fig. 2B. When the results were expressed as a ratio of post-enzymatic flow to the control flow in the same animal at the same pressure, trans-synovial flow was increased, on average, by 3.45 ± 0.44 times over the whole pressure range (n = 41).

Figure 2. Effect of chondroitinase ABC on the rate of absorption of Ringer solution from the cavity of the rabbit knee.

A, results from contralateral knees of the same rabbit. One joint was pre-treated for 30 min with 10 U chondroitinase ABC in Ringer solution (•) and the other with plain Ringer solution (○), prior to determining the pressure-flow relation. Dashed lines show regressions through the results in the low and high pressure ranges (below and above yield pressure). B, comparison of interpolated flows at the same intra-articular pressure, with () or without chondroitinase ABC pre-treatment (□) in pairs of knees from five rabbits (means ±s.e.m.). Statistical significance is indicated (*P≤ 0.05, **P≤ 0.01, ***P < = 0.001; Student's paired t test).

The slope of the pressure-flow relation increased markedly at pressures around 9 cmH₂O (yield pressure). The slope dQ̇_s/dP_j below yield pressure defines the hydraulic conductance of normal synovial lining, whereas the slope above yield pressure depends on the rate of increase of conductance with pressure (Knight & Levick, 1985). Conductance increases because of distension of the lining and increased hydration with pressure (Price et al. 1996b). The slope increase occurred at similar pressures in the control and chondroitinase ABC-treated joints, namely 7–14 cmH₂O (control) and 6–14 cmH₂O (enzyme treated; differences not significant).

Chondroitinase ABC increased dQ̇_s/dP_j below yield pressure by a factor of 2.3, that is, from the control value of 1.34 ± 0.22 to 3.11 ± 0.45 μl min⁻¹ cmH₂O⁻¹ after treatment (n = 5). Above yield pressure, chondroitinase ABC increased the slope 3.8 times, from 1.90 ± 0.24 (control) to 7.17 ± 0.33 μl min⁻¹ cmH₂O⁻¹ (n = 5). The differences between slopes below and above yield pressure, and between control and enzyme-treated joints, were statistically significant (P < 0.001; ANOVA with Bonferroni's test post hoc).

Since the substrate specificities of chondroitinase ABC, leech, Streptomyces and testicular hyaluronidases overlap (see Introduction), statistical comparisons with the hyaluronidase results of Coleman et al. (1998b) were carried out. Treatment with chondroitinase ABC, which raised flows 3.45 ± 0.44 times overall (n = 41), had less effect than treatment with Streptomyces hyaluronidase, which raised flows 6.22 ± 0.56 times (n = 45; P < 0.001, Mann-Whitney U test), or treatment with testicular hyaluronidase, which raised flows 4.93 ± 0.49 times (n = 42; P < 0.001). Similarly, the effects of chondroitinase ABC treatment on dQ̇_s/dP_j were significantly less than the effects of leech, Streptomyces and testicular hyaluronidases (Table 1; P < 0.001, ANOVA). Treatment with chondroitinase ABC increased the slopes below, and above, yield pressure by 2.3 and 3.8 times, respectively, whereas the hyaluronidases increased the slopes by 4.7 and 5.4 times, respectively (pooled hyaluronidase data).

Table 1.

Effect of selective enzymatic degradation of extracellular matrix components on slope dQ̇_s/dP_j (hydraulic permeability) of relation between trans-synovial flow (Q̇_s) and intra-articular pressure (P_j), ordered by size of effect

	dQ̇s/dPj
Treatment	Pressure range < Py	Pressure range > Py
Keratanase	0.69 ± 0.25 (3)	1.66 ± 0.43 (3)
Control	0.43 ± 0.13 (3)	1.78 ± 0.17 (3)
Heparinases	1.01 ± 0.45 (3)	1.82 ± 0.82 (3)
Control	1.48 ± 0.47 (3)	3.23 ± 1.96 (3)
Chondroitinase ABC	3.11 ± 0.45 (5)	7.17 ± 0.33 (5)
Control	1.34 ± 0.22 (5)	1.90 ± 0.24 (5)
Hyaluronidases*	4.38 ± 0.55 (15)	12.79 ± 1.06 (16)
Control*	0.94 ± 0.12 (10)	2.35 ± 0.13 (11)
Chymopapain	6.41 ± 2.06 (7)	29.51 ± 5.84 (8)
Control	1.64 ± 0.45 (5)	2.46 ± 0.55 (5)

Yield pressure, P_y, is the pressure at which the slope of the pressure-flow relation increases markedly.

^*Calculated by pooling the slopes observed after treatment with testicular hyaluronidase, Streptomyces hyaluronidase and leech hyaluronidase; results of Coleman et al. (1998b). Results are means ±s.e.m. (in μl min⁻¹ cmH₂O⁻¹) with number of experiments in parentheses.

Histochemical and immunohistochemical results

Although the HABR probe for hyaluronan was still bound to synovium that had been treated in vivo with chondroitinase ABC, the binding was greatly reduced in comparison with control tissue (Fig. 1A and B). The binding of HABR probe was totally abolished by treatment with Streptomyces hyaluronidase (Table 2).

Table 2.

Effect of enzymatic treatments on synovial histochemistry and immunohistochemistry

Enzymatic treatment in vivo	Hyaluronan (HABR)	Chondroitin 4-sulphate attachment region (2-B-6)	Chondroitin 6-sulphate attachment region (3-B-3)	Keratan sulphate (5-D-4)
Controla	++	++	+	++
Chondroitinase ABC (in vivo)	+	++c	+c	+
Keratanase	++	++	+	0
Heparinases	++	++	+	++
Hyaluronidasesb	0	++	+	++
Chymopapain, 0.1 U	+	+	0	0

Key: ++, much antibody or probe bound; +, antibody or probe bound; 0, antibody or probe not bound.

^aResults fromColeman et al. (1998a).

^bResults from Coleman et al. (1998b).

^cBinding was similar with or without digestion of the sections with chondroitinase ABC. Biotinylated hyaluronan binding region probe (HABR) and antibodies used are shown in parentheses.

Antibodies 2-B-6 and 3-B-3, for the chondroitin 4-sulphate and chondroitin 6-sulphate attachment regions, respectively, were bound to tissue treated with chondroitinase ABC in vivo, both with and without digestion of the sections with chondroitinase ABC (Table 2). Tissues that had not been treated with chondroitinase ABC, in vivo or as sections, did not bind the antibodies.

Antibody to keratan sulphate (5-D-4) was bound after chondroitinase treatment in vivo (Table 2).

Effect of keratanase treatment

Physiological results

Treatment with keratanase was limited to three animals, because it became clear that, compared with the effects of chondroitinase ABC and the hyaluronidases, its effect was negligible (see Fig. 3). Trans-synovial flows were not increased significantly by treatment with keratanase compared with vehicle, being slightly lower than control in seventeen out of twenty-seven comparisons at matched pressures. The ratio of flow after keratanase treatment to control (median, 0.85) was not significantly different from equality (P = 0.6, Student's t test on Gaussian-distributed logarithmic ratios). The slope dQ̇_s/dP_j, either below yield pressure or above it, was not increased significantly by keratanase treatment (see Table 1). Differences in slope between control and keratanase-treated joints were not statistically significant, either by Student's paired t test (P = 0.8; 6 pairs of regression slopes) or by comparison of regression slopes fitted to pooled data below or above yield pressure (P > 0.5).

Figure 3. Effect of keratanase treatment on rate the of absorption of Ringer solution from the cavity of the rabbit knee.

A, results from contralateral knees of the same rabbit. One was pre-treated for 30 min with keratanase in Ringer solution (•) and the other with plain Ringer solution (○) prior to determination of the pressure-flow relation. Note expanded scale of the y-axis compared with Fig. 2. Dashed lines are regressions through the results below and above yield pressure. B, comparison of interpolated flows at the same intra-articular pressure with () or without keratanase pre-treatment (□) in pairs of knees from three rabbits (means ±s.e.m.), plotted on the same y-axis scale as Fig. 2 for comparison. None of the differences between control and keratanase-treated joints are statistically significant.

Histochemical and immunohistochemical results

After keratanase treatment in vivo, the synovium no longer bound the 5-D-4 antibody for keratan sulphate (compare Fig. 1C and D). The HABR probe for hyaluronan was still bound. Likewise, the 2-B-6 and 3-B-3 antibodies were still bound to the chondroitinase ABC-treated sections.

Effect of heparinase treatment

Physiological results

Treatment with heparinases did not, on average, increase the trans-synovial flow or synovial hydraulic conductance, either in the joint treated with heparinases I and II or in those treated with heparinases I, II and III. Trans-synovial flows were actually slightly lower after heparinase injection than after control injection in nineteen out of twenty-six comparisons at matched pressures, and the flow ratios (median, 0.80) just attained significant difference from equality after logarithmic normalization (P = 0.05, Student's t test). The slope dQ̇_s/dP_j, either below yield pressure or above it, was not increased by heparinase treatment (see Table 1). Differences in slope between control and heparinase-treated joints were not statistically significant, either by comparison by Student's paired t test (P = 0.2; 6 pairs of regression slopes) or by comparison of regression slopes fitted to pooled data below or above yield pressure (P > 0.5).

Histochemical and immunohistochemical results

Treatment of synovium with heparinase in vivo did not affect the synovial binding of the HABR probe, 2-B-6 and 3-B-3 antibodies in sections digested with chondroitinase ABC, and 5-D-4 antibody (Table 2).

Effects of chymopapain treatment

Physiological results

The effect of chymopapain was the most marked of all the enzymes studied, with trans-synovial flows at 20 cmH₂O generally exceeding 300 μl min⁻¹ (cf. Fig. 2), and reaching 700 μl min⁻¹ in one joint treated with 0.1 U chymopapain (Fig. 4A). Such high flows were never observed in joints treated with chondroitinase ABC or hyaluronidases.

Figure 4. Effect of chymopapain treatment on the rate of absorption of Ringer solution from the synovial cavity of rabbit knees.

The scale on the y-axis is contracted relative to Figs 2 and 3 in order to accommodate the very large flows. A, synovium injected with 0.1 U chymopapain, 90 min before measuring the pressure-flow relation. Dashed lines are regressions through the results below and above yield pressure. B, comparison of interpolated flows at same intra-articular pressures (means ± s.e.m) after chymopapain pre-treatment () and in control joints (□; Student's t test, *P≤ 0.05, **P < = 0.01).

The magnitude of the increase in trans-synovial flow increased with intra-articular pressure, from 1.5 times control flow at 5 cmH₂O to 7.8 times control flow at 22.5 cmH₂O; see Fig. 4B. The limited increase at low pressures may have been due in part to slight, residual bleeding into the cavity (see Methods), which would compete with the inflow from the infusion reservoir. Indeed, the trans-synovial flows at 2.5 cmH₂O in joints treated with 10 U chymopapain were, in some cases, smaller than the control flows, even though the hydraulic conductance of the lining, dQ̇_s/dP_j, had increased 3.9 times. Inspection of aspirated intra-articular fluid and of the joint lining at the end of the experiment showed severe bleeding in joints treated with 5–10 U chymopapain, slight bleeding after 0.5 U and virtually no bleeding after 0.1 U. Perhaps because of the absence of bleeding, a joint treated with only 0.1 U chymopapain generated the biggest net trans-synovial absorption rates of the study (Fig. 4A).

There was a very clear increase in the slope of the pressure-flow relation in the chymopapain-treated joints at a certain pressure, i.e. a clear yield phenomenon. The pressures at which the increase in slope occurred ranged from 7.8 to 13.0 cmH₂O (mean, 11.1 ± 0.7 cmH₂O; n = 7), which is within the usual range for yield pressure.

The effect of chymopapain on conductance, dQ̇_s/dP_j, bore no consistent relation to the dose injected; the regression slope of a plot of dose versus dQ̇_s/dP_j was not significantly different from zero. This was as expected, because an excess of enzyme activity was present in all cases (see Methods). The conductance results were, therefore, analysed as a single group. Treatment increased dQ̇_s/dP_j below yield pressure almost 4-fold, from a control value of 1.64 ± 0.45 to 6.41 ± 2.06 μl min⁻¹ cmH₂O⁻¹ after treatment. Above yield pressure the slope was increased 12-fold, from a control value of 2.46 ± 0.55 to 29.51 ± 5.84 μl min⁻¹ cmH₂O⁻¹ after chymopapain treatment. The differences between slopes below and above yield pressure, and between control and enzyme-treated joints, were statistically significant (P < 0.001; ANOVA with Bonferroni's test post hoc).

The results of various enzyme treatments are summarized in Fig. 5. The effect of chymopapain treatment on conductance exceeded even the effect of the hyaluronidases (Table 1; P < 0.001, one-way ANOVA), although only the difference above yield pressure attained statistical significance (P < 0.001, Bonferonni's test).

Figure 5. Comparison of the effects of various glycosaminoglycanidases and a protease, chymopapain, on the pressure-flow relation of the synovial lining.

The keratanase results (Fig. 3) are not shown because they almost exactly overlie the heparinase results. The hyaluronidase results are the means of the pooled results for testicular, Streptomyces and leech hyaluronidases from Coleman et al. (1998b; n = 14). The control values are the means of the pooled control flows from the chondroitinase ABC, keratanase, heparinase and chymopapain studies (n = 16). Results shown are means ±s.e.m.

As noted under Methods, one control joint was injected with the activating vehicle, viz. 10 mM cysteine and 10 mM EDTA in Ringer solution, rather than with plain Ringer solution. The resulting trans-synovial flows of 8 μl min⁻¹ at 5 cmH₂O, rising to 54 μl min⁻¹ at 20 cmH₂O, were normal for Ringer solution, namely mean values of 8.9 ± 2.5 μl min⁻¹ at 5 cmH₂O rising to 43 ± 8 μl min⁻¹ at 20 cmH₂O in the present study. The flows after injection of activating vehicle contrast with the flows after injection of chymopapain in activating vehicle, namely 13.4 ± 6.0 μl min⁻¹ at 5 cmH₂O, rising to 307 ± 73 μl min⁻¹ at 20 cmH₂O. Thus the vehicle itself was not responsible for the increased trans-synovial flows. The vehicle caused no intra-articular bleeding.

Histological, histochemical and immunohistochemical results

After treatment with 0.1 U chymopapain (Fig. 6B), the synovial lining was several cells thick and possessed abundant capillaries and venules. There was no evidence of haemostasis. The lining appeared thicker than in control tissue (Fig. 6A), with a swollen, oedematous appearance; but it must be noted that even normal synovium varies considerably in thickness (Knight & Levick, 1983). The subsynovium consisted of areolar connective tissue with no haematomata and few extravascular leukocytes.

After treatment with 0.5 U chymopapain (not shown), the synovium was reduced to a monolayer of cells. Haematomata were absent, but extravasated subsynovial polymorphonuclear leukocytes and haemoconcentration in subsynovial venules indicated inflammation.

After treatment with 5–10 U chymopapain the tissue showed evidence of massive haemorrhage, namely large accumulations of red cells in the subsynovium and joint cavity (Fig. 6C). Watts et al. (1975) attribute this to disruption of endothelial cell junctions. The synovial lining cells were few in number and formed a discontinuous monolayer. The synovial interstitium appeared as a thin, glassy, refractile layer. This hyaline layer had sufficient mechanical integrity to retain large haematomata in the subsynovium, because in some regions red cells were piled up against it. Strands of the hyaline material extended into the subsynovium.

The binding of the HABR probe, and antibodies 2-B-6, 3-B-3 and 5-D-4, was greatly reduced or abolished after treatment with 0.1 and 0.5 U chymopapain in vivo, when compared with control tissue (compare Fig. 6D with Fig. 1A, and Fig. 6F with Fig. 6E). The tissue treated with higher concentrations of chymopapain was not examined immunohistochemically.

The results of the histochemical and immunohistochemical studies are summarized in Table 2.

DISCUSSION

Effects of individual glycosaminoglycanidases

The first objective was to assess the contribution of individual sulphated glycosaminoglycans to synovial hydraulic resistance. Synovium contains chondroitin sulphates at a concentration of 1.2 mg (ml extrafibrillar space)⁻¹, heparan sulphate at 1.9 mg (ml extrafibrillar space)⁻¹, keratan sulphate at an unknown concentration, and hyaluronan at 0.8 mg (ml extrafibrillar space)⁻¹ (Price et al. 1996a). Treatment of synovium with leech, Streptomyces and testicular hyaluronidases has already demonstrated the great hydraulic importance of hyaluronan (Coleman et al. 1998b).

Chondroitinase ABC

Chondroitinase ABC treatment removes the main chondroitin sulphate chains, leaving behind residual chondroitin 4-sulphate and 6-sulphate oligosaccharides on the proteoglycan core protein, which serve as epitopes for 2-B-6 and 3-B-3 antibodies, respectively (see Methods). The binding of antibody to synovium that had been treated with intra-articular chondroitinase ABC in vivo, without further application of the enzyme to the sections, confirms that the enzyme removed the main chondroitin sulphate chains in vivo. The moderate hyaluronidase activity of chondroitinase ABC (see Methods) accounts for the partial depletion of synovial hyaluronan (Fig. 1B and Table 2). No available chondroitinase is entirely specific for chondroitin sulphates.

The modest effect of chondroitinase ABC treatment on conductance indicated that, in synovium, chondroitin sulphate may be less important hydraulically than hyaluronan. Although chondroitinase ABC treatment increased synovial hydraulic conductance, the increase was less than that caused by hyaluronidase treatment (Table 1 and Fig. 6). Similarly, the effect of chondroitinase ABC on the outflow conductance of the anterior chamber of the rabbit eye is less than that of testicular hyaluronidase (Knepper et al. 1984). Moreover, at least part of the limited effect of chondroitinase ABC was due to the partial depletion of synovial hyaluronan. Coleman et al. (1998b) likewise concluded that chondroitin sulphate contributes less than hyaluronan to synovial resistance. They compared synovial conductance after treatment with Streptomyces hyaluronidase with the conductance in the contralateral joint treated with Streptomyces hyaluronidase plus chondroitinase ABC. The increases in conductance on the two sides were similar, which indicates that chondroitin sulphate depletion does not significantly enhance the effect of hyaluronan depletion. The mechanisms by which hyaluronan depletion outstrips the effect of chondroitin sulphate depletion, despite the greater concentration of the latter in normal synovium (Table 3), and greatly exceeds the effect predicted by a model of synovial interstitium (Fig. 7), were discussed by Coleman et al. (1998b).

Table 3.

Summary of extrafibrillar matrix composition and effect of enzymatic digestion of components on synovial hydraulic conductance below yield pressure

Matrix component	Concentration (mg (ml extrafibrillar space)−1)a	Hydraulic conductance below yield pressure after enzyme (% control)
		Experimental result (from Table 1)	Prediction for complete removal of hydraulic drag of uniformly distributed chainsb
Chondroitin sulphates	1.2	223***d	123 (143)g
Keratan sulphate	unknown	161 (n.s.)	—
Heparan sulphates	1.9	68 (n.s.)	140
Hyaluronan	0.8	502**e	115
Non-collagenous interstitial proteins	10c	606***f	15700h

^aSynovium comprises 66% interstitium by volume; the interstitium comprises 23% collagen fibrils by volume (Levick et al. 1996; Price et al. 1996a).

^bDistributed model of Levick (1994) modified by Coleman et al. (1998b), based on 13.9 mg (ml extrafibrillar space)⁻¹ of total biopolymer distributed uniformly in the extrafibrillar space below yield pressure, the extrafibrillar space being distributed non-uniformly across the tissue (Levick et al. 1996). The model takes no account of matrix disorganization after component removal.

^cEstimate based on glycosaminoglycan total of 3.9 mg ml⁻¹ and uniform model requirement of 13.9 mg ml⁻¹; compatible with total protein content of tissue (Price et al. 1996a).

^dSome of this increase may be due to concomitant removal of hyaluronan by chondroitinase ABC.

^eResults for Streptomyces hyaluronidase (Coleman et al. 1998b).

^fMean result after 0.1 U chymopapain, which preserves lining cellularity.

^gValue in parentheses refers to removal of both chondroitin sulphate and hyaluronan.

^hIn reality, total interstitial protein depletion would involve loss of sulphated glycosaminoglycans too.

^**P≤ 0.01

^***P≤ 0.001.

Figure 7. Comparison of observed effects of glycosaminoglycanidases and protease on synovial hydraulic conductance (data points, means ±s.e.m.) with predictions of a mathematical model (see text).

The vertical axis is hydraulic conductance below yield pressure after enzyme treatment, i.e. after partial depletion of biopolymer, relative to control conductance. The dot-and-dashed horizontal line marks the average, 6.6-fold increase in conductance caused by treatment with 0.1 U chymopapain, to 10.8 μl min⁻¹ cmH₂O⁻¹ (below yield pressure). This corresponds to a residual extrafibrillar biopolymer concentration of 6 mg ml⁻¹ (vertical line).

Keratanase

The presence of keratan sulphate in rabbit synovium is inferred from immunohistochemical rather than biochemical evidence (Coleman et al. 1998a), but is strongly supported, in the present study, by the depletion of the 5-D-4 epitope by keratanase treatment in vivo (Fig. 1D and Table 2). Without information on the keratan sulphate concentration, it is not possible to assess the lack of effect of keratan sulphate removal on hydraulic permeability. The keratanase did not remove hyaluronan, chondroitin 4-sulphate attachment region or chondroitin 6-sulphate attachment region from synovium, which indicates an absence of significant protease contamination.

Heparinases

Although heparan sulphates are present in synovium in a greater concentration than either chondroitin sulphates or hyaluronan, heparinase treatment did not increase synovial conductance. The lack of effect of heparinases on synovial lining conductance, despite activity in vitro (see Methods), might arise in principle from lack of action in vivo (since no antibody was available to check this) or from the non-uniformity of the extracellular compartment. Heparan sulphate proteoglycans are, in general, associated with cell surfaces and basement membranes, and in synovium heparan sulphates occur around B-type synoviocytes, in association with laminin (Revell et al. 1995). The sub-regions of the extracellular space that contain most of the heparan sulphate, namely basement membranes and pericellular matrix, may carry little of the trans-synovial flow, or might normally carry some flow but collapse after heparinase treatment (explaining the slight, just significant reduction in flow). In other words, extracellular matrix organization, as well as the averaged concentration of a component, may be important in determining tissue resistance. Bert & Reed (1998), studying flow through rat dermis, likewise emphasize the importance of the organization of matrix for hydraulic resistance, although in dermis there are grounds for believing that collagen rather than glycosaminoglycans may dominate the resistance (Levick, 1987).

Effect of glycosaminoglycanidases compared with model predictions

Interstitial flow depends on local pressure gradients and on the hydraulic drag created by biopolymer chains, namely glycosaminoglycans, proteoglycans and glycoproteins (Levick, 1987). A distributed, mathematical model of this process in synovium enables quantitative predictions to be made of the effects of enzymatic treatments. The model incorporates tissue surface area and thickness, interstitial volume fraction, compositional gradients normal to the surface, pathway tortuosity, collagen fibril volume fraction, capillary density and transport distances (Levick, 1991, 1994). Flows are calculated from the boundary pressures by a finite difference method, using Darcy's law for flow between interstitial microdomains, and the Starling principle for transcapillary fluid exchange. The model was upgraded by Coleman et al. (1998b) to take account of the organization of synovial collagen fibrils into bundles (fibres), and the hydraulic drag of the microfibrillar network and cell surfaces. Figure 7 shows a semilogarithmic plot of results from the revised model for synovial-lining conductance below yield pressure, for the case of uniformly distributed molecular chains (cf. heparan sulphates).

Starting with a control concentration in non-collagenous interstitial space of 13.9 mg biopolymer ml⁻¹, which is the minimum concentration required by a uniform distribution model to account for the control hydraulic resistance, depletion of chondroitin sulphates alone (1.2 mg ml⁻¹) or chondroitin sulphates plus hyaluronan (2.0 mg ml⁻¹) should raise conductance by 23 and 43 %, respectively; concentration affects conductance non-linearly. The actual effects of chondroitinase ABC and testicular hyaluronidase, namely increases by 174 ± 23 and 436 ± 86 %, respectively, greatly exceed the model predictions (Fig. 7). Hyaluronan depletion is the dominant factor (see above), and mechanisms have been proposed for its ‘excessive’ effect, based on its pivotal role in matrix organization and/or non-uniformity of distribution (Coleman et al. 1998b).

For depletion of randomly distributed heparan sulphate chains (1.9 mg ml⁻¹), conductance is predicted to increase by 40 % (Table 3). The observed fall by 32 % was not statistically significant. An increase of only 40 % could only be detected by a larger study, because of the high variance of the control conductances. A power calculation indicates that an n value of twenty-three would be required to detect a 40 % increase with a power of 0.9, using Student's two-tailed t test at a significance level of 0.05. As noted above, however, the model calculation applies to an evenly dispersed glycosaminoglycan, and heparan sulphate is not evenly distributed.

The amount of heparan sulphate in control synovium is virtually identical to the amount of chondroitin sulphate plus hyaluronan (Table 3). The specific resistivity of each glycosaminoglycan species in vitro is similar (Comper & Zamparo, 1990). Consequently, the model predicts almost equal contributions to synovial resistance by heparan sulphate and by chondroitin sulphate plus hyaluronan (Table 3). The observed effects of their depletions, however, were vastly different. The results thus indicate that the different categories of glycosaminoglycans do not contribute equally, weight for weight, to hydraulic resistance in vivo. Hyaluronan is far more important to the hydraulic resistance of synovium than are the sulphated glycosaminoglycans.

Effect of chymopapain

The extrafibrillar interstitial space contains ∼4 mg ml⁻¹ glycosaminoglycan, while the estimated tissue resistance requires a total concentration, for all forms of biopolymer, of ∼14 mg ml⁻¹, depending on distribution pattern (Levick, 1994). It has been suggested, therefore, that the hydraulic drag of non-collagenous protein biopolymers, particularly proteoglycan core proteins and glycoproteins, contributes substantially to resistance (Price et al. 1996a). Similar apparent discrepancies between extrafibrillar glycosaminoglycan concentration and hydraulic resistance occur in many other tissues, and can be resolved, quantitatively, by taking into account the proteoglycan core protein and glycoprotein content of the tissue (Levick, 1987; Price et al. 1996a). The importance of non-collagenous structural proteins was supported here by the effect of interstitial proteolysis. The increases in fluid escape rate and hydraulic conductance after interstitial proteolysis exceeded those caused by depletion of hyaluronan or sulphated glycosaminoglycans, and were amongst the highest that we have ever observed. Large increases in tissue conductance after proteolysis were likewise observed in mouse flank fascia by Day (1952) and in corneal stroma by Hedbys (1963).

Yield pressure was unaffected by chymopapain. It is speculated that the yield occurs when tension exceeds a criticial value in synovium. This tension may develop when the ‘slack’ or excess area present as concertina-like folds of synovium at low intra-articular volume and pressure (seen in scanning electron micrographs; McDonald & Levick, 1988) becomes fully extended. If this is correct, the yield pressure should depend primarily on gross anatomy, and be unaffected by enzymes acting on interstitial permeability.

Although loss of lining cells contributed to enhanced conductance after treatment with 5–10 U chymopapain, a multicellular lining remained after treatment with 0.1 U (Fig. 6B). The immunohistochemical results showed that 0.1 U chymopapain removes much of the proteoglycan, as well as hyaluronan, from the interstitium (Fig. 6D-F). The contribution of non-collagenous proteins to hydraulic resistance may, therefore, be multiple, as follows.

(i) Protein chains exert hydraulic drag, as discussed by Price et al. (1996a).

(ii) Proteoglycan core protein anchors the sulphated glycosaminoglycan chains. In Wharton's jelly proteolysis releases all the sulphated glycosaminoglycans plus 20 % of the hyaluronan (Meyer et al. 1983). Similarly, in synovium there was almost total depletion of sulphated glycosaminoglycans and partial loss of hyaluronan after treatment with 0.1-0.5 U chymopapain. The increased conductance after proteolysis is caused by the removal not only of interstitial structural proteins, but also of the associated glycosaminoglycans.

(iii) Glycoproteins may generate a restraining force that opposes the glycosaminoglycan osmotic or ‘swelling’ pressure. Without a mechanical restraint, interstitium will expand by osmosis, and the increased hydration will raise the hydraulic conductivity. For example, trypsin digestion of the glycoprotein microfibrils in Wharton's jelly increases the interstitial volume 3-fold, because microfibrillar tension normally opposes the glycosaminoglycan swelling pressure (Meyer et al. 1983). Interstitial expansion of this magnitude would increase hydraulic conductivity ∼12 times in synovium, as read from Fig. 7. The thickened, oedematous appearance of synovium treated with 0.1 U chymopapain (Fig. 6B), though only a qualitative observation, is compatible with interstitial expansion.

(iv) Proteases might affect interstitial architecture by degrading anchorage sites between cell and matrix, such as the β₁-integrins. The β₁-integrin link allows fibroblasts to influence interstitial fibre tension and hence hydration actively (Reed et al. 1998). In two pilot experiments where RGD sequence peptide (3 mM GRGDTP peptide, Novabiochem, Nottingham, UK) was injected intra-articularly to block β₁-integrin sites on synoviocytes, there were modest increases in trans-synovial outflows and conductance at low pressures, from a control level of 1.23 ± 0.17 to 2.07 ± 0.16 μl min⁻¹ cmH₂O⁻¹ in RGD-treated joints (D. Scott, P. J. Coleman, R. M. Mason & J. R. Levick, unpublished results). If substantiated by further work, this would indicate a possible link between β₁-integrins and synovial permeability.

Effect of protease compared with model predictions

Although many non-collagenous, structural proteins occur in synovial interstitium (for review see Levick et al. 1996), their concentrations are unknown. If the simplest scenario is adopted, namely that the estimated 14 mg biopolymer (ml interstitium)⁻¹ comprises the known 4 mg ml⁻¹ glycosaminoglycan plus 10 mg ml⁻¹ non-collagenous, chymopapain-digestible protein chains, then total protein depletion would leave, at most, 4 mg ml⁻¹ residual biopolymer (Table 3). This would cause a 15.7 times increase in synovial conductance below yield pressure (point on theoretical curve in Fig. 7 corresponding to residual 4 mg ml⁻¹ biopolymer), or > 15.7 times increase when concomitant glycosaminoglycan loss is taken into account. The largest chymopapain-induced rise in conductance below yield pressure, seen experimentally, was to 11 times the control value.

The model can be used, conversely, to estimate the reduction in mean biopolymer concentration that would explain observed increases in conductance. For example, the average increase in conductance below yield pressure was 6.6-fold after 0.1 U chymopapain, a treatment that left an intact cell layer. The curve in Fig. 7 shows that this can be modelled by a reduction in extrafibrillar biopolymer concentration (protein plus glycosaminoglycan) to 6 mg ml⁻¹, less than half the control level, if the remaining biopolymer is uniformly distributed and the synovial geometry unchanged (thickness, pathway tortuosity etc.). The larger increases in dQ̇_s/dP_j above yield pressures could be caused partly by mechanical weakening of the lining by chymopapain, affecting both membrane geometry at a given pressure (area, thickness) and biopolymer content, and partly by the non-linear relation between biopolymer concentration and conductance. In view, however, of recent evidence that matrix organization as well as content governs hydraulic resistance (Coleman et al. 1998b; Bert & Reed, 1998), quantitative model predictions of enzyme effects may be seriously in error without information about changes in pathway geometry and organization.

Conclusions

The different classes of glycosaminoglycan do not contribute equally, weight for weight, to synovial hydraulic resistance. Hyaluronan, and possibly chondroitin sulphates, contribute substantially to resistance, but heparan and keratan sulphates contribute little. The greatest fall in resistance occurs when non-collagenous structural proteins are removed from the interstitium. This may be of pathophysiological significance, because the joint effusions of traumatic, osteoarthritic and inflammatory arthritis contain activated metalloproteinases, such as stromelysin, which have a broad proteolytic spectrum against matrix proteins. The metalloproteinases have been implicated in cartilage breakdown (Lohmander et al. 1993). On the basis of the effect of chymopapain, it seems probable that the permeability of the joint lining, as well as cartilage, will be affected by free metalloproteinases in arthritides.