# Permeability of Rheumatoid and Normal Human Synovium to Specific Plasma Proteins.

код для вставкиСкачать1550 PERMEABILITY OF RHEUMATOID AND NORMAL ‘HUMAN SYNOVIUM TO SPECIFIC PLASMA PROTEINS J. RODNEY LEVICK A method is described for the determination of the permeability of the blood-joint barrier to specific plasma proteins, using the ratio of protein concentration in synovial fluid to that in plasma. The inadequacy of the ratio per se as a direct index of permeability is discussed. Permeabilities are evaluated for the normal and rheumatoid human knee. Permeability increases in the rheumatoid knee by approximately 6 times for albumin and over 40 times for macroglobulins. The effect of protein molecular dimensions upon permeability is analyzed. Permeability shows less dependence upon solute dimensions in the rheumatoid knee than in the normal knee, i.e., molecular selectivity is reduced. From these data and synovial morphology, a twomembrane model of the blood-joint barrier is developed. The relative contribution of the component intima1 and endothelial layers to the total barrier is found to depend upon solute dimensions. The permeability of the blood-joint barrier to plasma proteins determines the oncotic pressure of synovial fluid. The difference in oncotic pressure across the barrier is one of several factors regulating flow into the joint space (1,2). Thus, changes in permeability to plasma proteins may contribute to the pathogenesis of rheumatoid and other inflammatory synovial effusions (2). Yet synovial permeability to plasma protein has been measured directly by only one Supported in part by the Arthritis and Rheumatism Council of Great Britain. Address reprint requests to J. Rodney Levick, Senior Lecturer in Physiology, St. George’s Hospital Medical School, London SW17 ORE, England. Submitted for publication November 20. 1980; accepted in revised form June 11. 1981. Arthritis and Rheumatism, Vol. 24, No. 12 (December 1981) group (3,4); synovial permeabilities to specific proteins of known physical dimensions have never been reported. Many investigators (5-7) have measured the concentration of specific plasma protein in synovial fluid (C,) and plasma (Cp), and have intuitively adopted the fluid/plasma concentration ratio C F / C ~as a direct measure of permeability. This paper demonstrates that CF/Cp is not a linear index of permeability. Also it depends on other variables in addition to permeability; consequently, the use of CF/Cp as a direct index of permeability is not justifiable. C&p may serve as a nonlinear index of permeability, but only if volume flow into the joint (equal to lymph outflow) is constant, and only after correction of CF for steric exclusion by hyaluronate. A method is described, however, by which permeability can be calculated from CF/CPand volume flow. The method is applied to data for rheumatoid and normal human knees; the relationship between permeability and solute dimensions is analyzed. METHODS Theory: Relationship between permeability and synovial fluidplasma concentration ratio. The permeability (P) of a membrane of area (A) is, by definition, the diffusional flux of solute (Js”) per unit concentration difference across the membrane. The membrane that separates synovial fluid from plasma consists of two layers in series, namely capillary endothelium and synovial intima (Figure 1). For this composite “blood-joint barrier,” permeability is given by: equation 1: P = JsD / A (CpCp‘) where C p is mean solute concentration in capillary plasma and CF’ is “effective” solute concentration in synovial fluid. PERMEABILITY OF HUMAN SYNOVIUM 1551 - BLOOD-JOINT BARRIER 1- [SYNOVIAL . . . . . . . . . . . . . .. .. . . . . . . . . . INTIMA~ I - [ PROTEI (Jd BLOOD C A P ILLAR Y . . . FLUID (Jv) LYMPH VESSEL Figure 1. Diagram of protein and fluid exchange between plasma, joint cavity, and lymph vessels in the steady state (highly schematic; thickness of endothelium greatly exaggerated). Protein passes from plasma (concentration C), at rate Js across two barriers in series. namely capillary endothelium and synovial intima; movement is mainly by diffusion. Fluid flows down pressure gradients from plasma into the joint cavity at rate J,. Protein concentration in synovial fluid (C,) is determined by the relative rates of protein influx and water influx. Because the system is in a steady state, fluid must also drain from the joint cavity at rate J,, probably into the lymphatic system. Local synthesis of immunoglobulins, which becomes significant in the rheumatoid joint (see text), is omitted for clarity. The effective solute concentration exceeds the measured concentration CEbecause solute is excluded from a fraction of the solvent volume by the polysaccharide fibers of synovial hyaluronate (8,9). If the fractional available volume is KAV,then equation 2: If a solvent flow occurs across the membrane, solute not only diffuses across the membrane but is also “dragged” along by the solvent (solute flux JsV). Kedem and Katchalsky (10) have shown by the methods of irreversible thermodynamics that net solute flux Js is then given by: equation 4: CF’ = C p / KAv The fractional available volume is given by the Ogston equation (8): equation 3: KAV= exp (-nL (rf + re,)2) where rf is hyaluronate fiber radius and L is fiber concentration in units of lengthlvolume. res is the Einstein-Stokes diffusion radius of the solute. where Jv is volume flow across the barrier, u is the solvent drag or osmotic reflection coefficient ( I I ) , and C is mean solute concentration within the barrier [approximately (Cp t CF’)/ 2 (12)]. In the steady state, solute influx (Js) must also equal the concentration of solute CF divided by the volume influx, since, by definition, concentration is the ratio of solute to volume (see Figure 1): LEVICK 1552 equation 5 : Cr = Js/Jv Volume influx JV is net filtration from plasma and is determined by such factors as capillary blood pressure and oncotic pressure (I). Substitution of equations 2 and 4 into equation 5 leads to an expression that describes the relationship between permeability and the observed ratio CF/Cp (represented by R): equation 6: biophysical details of the outflow system, since fluid is sampled before outflow.) Thus, lymph flow JV can influence synovial protein concentration CF, as is described by equation 7. Expressed nonmathematically , the lymph vessels affect synovial protein level by influencing the rate of washout of the joint cavity. Although lymph flow affects CF, it does not affect the permeability of the blood-joint barrier. Equation 7 is also applied below to data from rheumatoid joints. Because the assumption of negligible transport by bulk flow is, a priori, more dubious in such joints, the values of Jv/(R-l - KAV-') should be regarded only as approximations of permeability-surface area product PA; Jv/ (R- - KAV-I ) represents, strictly, total solute flux (by diffusion plus bulk-flow) per unit transsynovial concentration difference. It will be shown that these values provide important information regarding the physical processes by which protein is carried into rheumatoid joints. Data. C,JCp.The concentrations of specific plasma proteins or protein fractions in plasma and in the fluid from normal and rheumatoid human knees have been intensively investigated in many previous biochemical studies (5-7.1420). Data from 13 sources are summarized in Table 1. KAV. Fractional available volume in synovial fluid may be calculated from Ogston's equation (equation 3). Hyaluronate fiber radius rf is 3.5 x lo-' cm and L (cm/cm3) is given by: ' Equation 6 is not readily solved with the data available in the literature because synovial u for specific proteins has never been reported. However, a simpler equation, which can be solved, is obtained if solut? transport by solvent drag [the bulk flow term Jv(l - v)C] is neglected. Data exist that provide some justification for this simplification. Simkin and Pizzorno (3,4) found PA for plasma protein to be 0.008-0.011 ml/min for the normal human knee. The difference in total protein concentration across the normal bamer averages 49 mg/ml (Table 1); hence, the diffusional protein flux JsD is 39 x lo-* -54 x lo-' mglminute. The volume flow Jv into the normal knee is about lo-' ml/minute (2). If we assume u is 0.8 or more ( I ) , the protein flux due to solvent drag (Jsv) is 1 9 x lo-.' mglminute. Thus, diffusion accounts for at least 77-83% of plasma protein flux across the normal blood-joint barrier. If the relatively small nondiffusional component of protein influx is neglected, a simple relationship between steady-state concentration ratio R (= CF/Cp)and permeability is derived. Substitution of equations 1 and 2 into equation 5 gives: equation 7: [An analogous expression was first derived by Renkin (13) for lymph/plasma concentration ratios, where KAv is unity.] It can be seen that the steady-state fluid/plasma concentration ratio R is a hyperbolic rather than linear function of permeability. R depends on volume flow and fractional available volume as well as on permeability. Thus, the ratio R cannot serve as a direct index of permeability. Equation 7 does, however, enable permeability to be calculated from R for the normal joint. Although equation 7 is derived solely from consideration of influx into the joint cavity from plasma, it describes the steady-state situation, in which influx equals efflux. Therefore, Jv also represents fluid outflow, while protein efflux must equal JvCr;. Fluid and protein probably drain from the joint cavity via intercellular channels into lymph vessels, which lie deep to the intima and do not open directly into the joint cavity; thus, JV also equals lymph flow (Figure I). (It is stressed, however, that the above analysis is independent of any assumptions as to the anatomic or L = [HA] x NA X 10.3 lo00 x 379 X lo-' where [HA] is hyaluronate concentration in gdliter; 379 daltons is the molecular weight of the component disaccharide unit of length 10.3 x lo-' cm (9); and NA is Avogadro's number (6.02 x loz3).Hyaluronate concentrations in human synovial fluid were obtained from the literature (Table 1); they average 3 g d i t e r in the normal knee and 1 gdliter in the rheumatoid knee. J v . Net volume flow from plasma into the normal knee (Jv, equal to lymph outflow) is -lo-' ml/minute (2). Flow through the rheumatoid knee is estimated to be -2.10-2 ml/minute. This value was obtained by calculation of the flow which causes the calculated permeability to the smaller plasma proteins to lie in the range measured directly by Simkin and Pizzorno (4) for total synovial protein. The increase in fluid flow into (and therefore lymph flow out of) the rheumatoid joint tends to minimize the rise in protein concentration CF and renders CF in isolation a misleading guide (an underestimate) to the change of permeability in rheumatoid synovium. RESULTS Table 1 summarizes the fluid/plasma concentration ratios for various plasma proteins. Not all sources reported the CF and Cp values from which the published ratios were calculated; consequently, some of the values of R in the table differ slightly from the ratio of the quoted CF and Cp values. The fluid/plasma concentration ratio is always less than unity and PERMEABILITY OF HUMAN SYNOVIUM 1553 Table 1. Concentrations of plasma proteins and hyaluronate in synovial fluid from normal and rheumatoid human knees* .... -_ -. -__ ___.. Molecular Normal fluid, mg/mlS. Rheumatoid fluid, mg/ml -__. . weight Solute (ref) (KdV Fluid Plasma CF/Cp Fluid Plasma c JC p ~. _____ - __ __ -_ Total protein (6.14-16,19,20) I9.?)<. 2) 67.7(I .7) 0.28 50.0(4.8) 70.0(35) 0.71 a,-globulins (5.14-16.19) -44 1.3(0.1) 4.1(0.4) 0.32 ?.5(0.6) 4.0(0.3) 0.63 44. I 0.7 Orosomucoid (5,6,17,20) 0.8 1.3 0.65 0.31 Albumin (5.14-16,19,20) 69 12.0(1.4) 32.7(3.0) 0.37 i9.1(1.3) 29.2(0.2) 0.65 Transfenin (6,7) 74-90 0.24 1.4 2.3 0.62 Prothrombin-proconvertin (5) 68.5-74 0.10 0.36 P,-globulins (IS) 2.7 10.5 0.26 Haptoglobin (5,17,18,20) 100 0.09 I .8 0.05 0.9 2.4 0.38 IgG (6.7.20) -150 0.13 10.3 20.1 0.67 y-globulins (5.14-16.19) 160 2.4(0.4) I1.3( I . 5 ) 0.21 Il.O(O.6) 17.0(2.1) 0.65 Ceruloplasmin (6) 160 0.16 0.53 IgA (7,20) 160-400 1.6 3.6 0.47 a2-globulins (5,7,14-16.19) loo-820 I . l(O.2) 7.9(0.3) 0.14 3.2(1.2) 6.9(2.6) 0.46 a2-macroglobulin (6) 820 0.033 0.35 P2-globulins 340-1 ,OOO 0.5 4.9 0.09 (? inc. fibrinogen) (15) IgM (6.7) 900-1,OOO 0.045 0.5 1 1.44 0.43 Hyaluronate (21-23) -6,000 3.0 1 __ __ __ -_ -_ * Data are averages of concentrations presented in the references in parentheses. t Molecular weights are from these references and from F. W. Putnam’s The Plmrnci Proteins (33). $ Numbers in parentheses are standard errors of the mean; the presentation of the data do not enable SE to be calculated in all cases. Normal fluid was largely sampled postmortem. ~ ~ ~ - ~ ~ ~~ declines as protein dimensions increase. These low ratios might be due to steric exclusion of plasma protein by synovial hyaluronate (24), i.e., represent the fractional available volume in synovial fluid. This proposition is tested in Figure 2, which shows fractional available volume plotted against Einstein-Stokes diffusion radius of solute (re,), for fluids containing various concentrations of hyaluronate. On the same graph, the ratios CF/CP (normal knee) are plotted against res. The observed ratios are considerably smaller than can be accounted for by steric exclusion by hyaluronate at physiologic concentrations (2-4 gm/ liter); a hyaluronate concentration of at least 12 gm/ liter would be required to produce most of the observed ratios. Similarly, the ratios for rheumatoid knees cannot be explained by the rheumatoid concentration of hyaluronate (-1 gmfliter). The low ratios therefore do not represent an equilibrium state, but are caused by the low permeability of the blood-joint barrier to plasma protein. Permeability-area products (PA) for specific plasma proteins were calculated from the ratios in Table 1 via equation 7 and are tabulated in Table 2. In the normal knee, PA for plasma albumin (MW 69 Kdj is 7 X lop3cm3/minute. PA declines as protein molecular size increases, falling to 0.4 x lop3cm3/minute for u2-macroglobulin (MW 820 Kd). In the rheumatoid knee, the transsynovial flux of each protein per unit concentration difference (ap- ~ ~ proximate PA, see Theory) is increased. The relative increase is greater for large proteins than for small proteins. For example, PA for albumin increases by a factor of 6.36 in the rheumatoid knee, while PA for a?- es ( n m ) P Figure 2. Relationships between fractional available volume (KAV) and Einstein-Stokes diffusion radius of solute (rCJ in solutions containing various concentrations of hyaluronate. The number against each curve represents hyaluronate concentration in g d i t e r . Curves were calculated from equation 3. The solid triangles show the observed fluidplasma concentration ratios (CF/Cp) for the normal human knee, taken from Table 1 . The experimental data are not distributed around the curve for hyaluronate at physiologic concentration (-3 g d i t e r ) . LEVICK 1554 . .-C E \ E Q Q I .I 1 I 10 1 100 Figure 3. Relationships between permeability-surface area product (PA) and Einstein-Stokes diffusion radius (re,) for the blood-joint barrier of the human knee. Data are from Table 2 (solid symbols) and references 3 and 4 (open symbols); circles represent data from the normal joint and triangles data from the rheumatoid joint. The plotted values are total solute flux per unit transsynovial concentration difference, and, in the case of rheumatoid joints, should be regarded only as approximations of true PA (see Theory). Solid lines are lines of best fit for the protein permeabilities of Table 2 (linear regression analysis): the steeper of the two lines drawn through the rheumatoid data is the line of best fit if data for y-globulin, IgG, and IgM are excluded (because of local synthesis of these proteins by rheumatoid synovium). The parallel, dashed lines have a slope of - 1 and represent the decline in PA that would result purely from the decline of free diffusion coefficient as solute radius increases. macroglobulin increases by 41.3. The quantitative error inherent in the adoption of Cp/Cp as a direct measure of permeability is well illustrated here; whereas PA for albumin increases 6.36 times in the rheumatoid knee, the albumin concentration ratio CF/ Cp increases only I .76 times. The large increases in PA for plasma proteins in rheumatoid arthritis contrast with the slight reductions in PA for small solutes, such as urea, reported by Simkin and Pizzorno (4). The high concentrations of nonimmune proteins in rheumatoid fluid are caused by the raised permeabil- ity of the rheumatoid blood-joint barrier. An alternative potential cause of high protein levels, namely lymphatic obstruction (reduction of term Jv in equation 7), is discredited by the data, because l ) large proteins show a greater rise in concentration than do small proteins (molecular selectivity reduced), whereas lymphatic obstruction would elevate the concentrations of all proteins by an equal factor (equation 7); and 2) lymph flow appears to increase rather than decrease in the rheumatoid joint (see Methods). The relationship between permeability and solute radius for both normal and rheumatoid joints is illustrated in Figure 3. Also shown are values of PA for small solutes (THO, urea, urate, creatinine, glucose, sucrose) as reported by Simkin and Pizzorno (3,4). Data are plotted on logarithmic axes to facilitate comparison with the change in free-diffusion coefficient. The free-diffusion coefficient (D) of a solute is inversely proportional to the Einstein-Stokes radius ID cll/res or log D c1 - 1 x log res (27)l. Thus, on a log PA versus log res plot, a line of slope - 1 would indicate a decline in permeability which is wholly explicable by the decline in free-diffusion coefficient with solute size. For the normal blood-joint bamer, however, a line relating log PA and log re, for solutes of radius 0.5-3 nm is constrained to have a slope steeper than -1, i.e., permeability declines more steeply than can be explained by the change of free diffusion coefficient. This phenomenon has been termed “restricted” diffusion (28) and indicates that solutes of radius 0.5-3 nm do not diffuse freely within the normal blood-joint barrier. The decline in permeability as re, increases above 3 nm (protein data) is less dramatic but nevertheless appears to be slightly greater than that attributable to a decline in free-diffusion coefficient. The regression line relating log PA and log res for the protein data has a slope of -1.23 (or -1.61 if the prothrombin and haptoglobin data are omitted, see below). These slopes are not statistically significantly different from - 1, however (Student’s r-test). In regard to the rheumatoid knee, there is no evidence for restricted diffusion; the slope relating log PA to log res is no steeper than -1 throughout the range of molecular sizes. Indeed, the regression line for the protein data has a slope of only 0.09, which is significantly less steep than -1 or than the slope for the normal blood-joint barrier. Thus, although the permeability to each protein is increased in rheumatoid arthritis, the molecular selectivity of the bloodjoint barrier (i.e., decline in permeability per unit increase in solute radius) is considerably reduced. PERMEABILITY OF HUMAN SYNOVIUM 1555 Table 2. Permeability area products (PA) for the blood-joint barrier of normal and rheumatoid knees, calculated from CF/Cp Normal knee Solute a,globulins Prothrothrombinproconvertin Albumin Transferrin Orosomucoid Haptoglobin Ceruloplasmin y globulins IgG IgA macroglobulin pZ globulins IgM cr2 PA, pl/min P(X 10 cm/min)t Rheumatoid kneet PA. pl/min 3.0 5.17 I .87 38.2 -3.4 3.6 3.6-4.3 3.9 4.6 4.5-5.6 5.6 5.6 5.7 9. I 10.8 11.0 1.14 6.96 3.53 5.24 0.54 2.13 3.33 I .68 0.38 2.25 0.66 0.41 2.51 I .27 1.89 0.19 0.77 I .20 0.61 0.14 0.81 0.24 11.8 44.3 39.9 46.8 13.4 27.6 Solute radius, rcs. nm* 59.9$ 69.5$ 22.0 15.7 49.7$ * r,,, the Einstein-Stokes diffusion radii are from references 9, 25, 26. t Synovial area (A) in the normal adult human knee = 277 cmz (40). P for rheumatoid joints has not been calculated, because no value for A appears to be available. The PA values calculated for the rheumatoid joints prove to be only approximations of the true diffusional permeability of the rheumatoid bamer, because transport by bulk flow is not negligible (see Discussion); the listed values are strictly total transsynovial solute flux per unit concentration difference between plasma and synovial fluid. t. Probably an overestimate due to local synthesis of immunoglobulin by infiltrated cells in the rheumatoid joint (see text). A number of anomalous PA values are evident in Table 2. PA values for haptoglobin and prothrombin-proconvertin are unusually low compared with those for other proteins of similar molecular radius. The electrophoretic mobility of haptoglobin is relatively high for its molecular weight, indicating a relatively high negative charge at physiologic pH (25). The low permeability of the barrier to this densely charged macromolecule might therefore indicate the existence of a high negative-charge density in a component of the barrier. There is considerable evidence that molecular charge as well as size affects the permeability in other vascular barriers (29-32). Prothrombin, though of molecular weight 68.5-74, Kd (331, behaves as a larger molecule of equivalent molecular weight 90-100 Kd on Sephadex column chromatography (34) and appears to behave as an even larger solute in the blood-joint bamer. In the case of rheumatoid joints, the apparent PA values for y-globulin and IgG are unduly high compared with other proteins. This may be attributed to local immunoglobulin synthesis by infiltrated cells (6). Perhaps, the apparently high permeability to IgM might be similarly explained. DISCUSSION The blood-joint barrier has been regarded as an essentially passive obstacle composed of two rnembranes in series: the synovial intimal layer and the underlying capillary endothelium (4). The question arises whether the known properties of the components of these layers (Figure 4) are consistent with the relationships between permeability and solute dimensions described in the Results section. To answer this question, we must first evaluate the relative contributions of the two layers to the normal total resistance to solute exchange (resistance = l/PA). Such an evaluation is possible from morphometric data. The procedure may be justified by the work of Casley-Smith, O’Donoghue, and Crocker (35), who showed that morphometnc calculations yield parameters close to those determined by direct physiologic experiment for the blood-jejunal barrier. From Fick’s law of diffusion, the ratio of resistances of the layers to a very small solute, whose diffusion is negligibly restricted within either layer, is (Al/lI)/(AJL), where A is equivalent pore area and 1 is pathlength in membranes 1 and 2. For an extracellular 1556 LEVICK ICG ICC 1 f Synovial fluid 1 Intima j- Endothelium Plasma v ICJ F Figure 4. Schematic diagram of the blood-joint barrier, based on electron micrographs (not to scale). ICG = intercellular gel. ICC = interstitial channel of Casley-Smith and Vincent (48). V = vesicle in lumenal aspect of capillary endotheliurn. ICJ = intercellularjunction, between two endothelial cells. F = fenestral membrane. The intimal layer is represented for simplicity by a single layer of cells and interstitiurn, but is in reality often 2-3 cells deep over the subintimal capillary plexus. Endothelial basement membrane is not separately represented on this diagram, being treated as part of the extracellular matrix for the purposes of the present analysis. solute, AJA2 may be estimated from the ratio of the area of intimal intercellular matrix to area of endothelial “pores,” provided that the fractional content of impermeable matter is not grossly dissimilar in the two permeable regions. The area of endothelial pores is taken to be that of intercellular junctions [-0. 1% of endothelial area (36)] plus fenestrae, which are present in roughly half the capillaries and occupy 0.06% of the area of such vessels (2,37). Endothelial area in the subintimal capillary plexus at 10 pm depth may be calculated from the mean intercapillary distance [--120 prn (38,39)] and is -15.7% of synovial area, or 43.5 cm’ for the human knee [synovial area 277 cm’ (40)]. This calculated value is less than 1/100th of the estimate by Simkin and Pizzorno (3), which was based on capillary morphology in skeletal muscle; this brings into doubt their conclusion that aggregate endothelial resistance to small-solute exchange is negligible. (The issue highlights the urgent need for better, quantitative data regarding the synovial area and microcirculation.) The area of intimal matrix is 7% of synovial area [rabbit knee (2)l. If a similar percentage applies to human synovium, as published electron micrographs seem to show, then matrix area is 19.4 crn2 in the human knee, cf., the value of 0.57 cm2 inferred by Simkin and Pizzorno (3). The depth of the intimal barrier over the superficial capillary plexus averages -10 p.m. The thickness of the endothelial barrier is far less (-0.1 pm for the intercellular junction and 3-5 nm for the fenestral membrane). Substitution of these morphologic data leads to the conclusion that into the ratio A1/i2/A2/i1 the intimal layer accounts for 52-78% of the total resistance of the normal blood-joint barrier to the diffusion of a small extracellular solute. The capillary wall accounts for only 114-1/2 of the total resistance, in spite of its small pore area, because the pathlength across the fenestral membrane is only 1/3,000 of that across the intima. The theoretical effects of solute dimensions on the permeability of such a two-layer system may now be calculated and compared with the data (Figure 5 ) . First, the simplest system will be considered: many small, uniform endothelial pores and a uniform extracellular glycosaminoglycan matrix. Endothelial pore radius is estimated to be 5 5 nm, because such channels generate an osmotic reflection coefficient of 20.8 PERMEABILITY OF HUMAN SYNOVIUM h .-C E \ E u Q a Figure 5. Comparison of theoretical and experimentally determined permeabilities of the blood-joint barrier (PA) to solutes of increasing molecular rddius (rer). Data are represented by circles (normal knee) and triangles (rheumatoid knee) as described in the legend to Figure 2. Curves A to D are theoretical relationships between the permeability-surface area product and molecular radius for a barrier composed of two membranes (intirnal interstitiurn and capillary endothelium) in series. Curve A (slope -1) describes a b a m e r whose “pores” are sufficiently large not to restrict the diffusion of solute. Curve B describes a bamer containing restrictive 5 nm radius transendothelial channels (or a dense fiber matrix) in series with interstitial glycosaminoglycan gel equivalent to 5 gm hyaluronatefliter. Curve C is as for B plus a system of 30 nm radius vesicles transporting water and solutes across the capillary wall at 2.5 pYminute. Curve D is as for C plus a system of 71 nrn radius channels permeating the intimal gel phase. for plasma albumin (1,41). Solute diffusing through such a pore is “restricted” by steric exclusion effects at the pore entrance and by friction within the pore. The degree of restriction is proportional to solute radius, as described by Pappenheimer (28,36) whose equations (see Appendix) were used to construct curve B of Figure 5 . [A similar curve may be obtained if endothelial permeability is represented by a dense fiber matrix rather than cylindrical pores (42).] The extracapillary interstitial barrier is considered to re- 1557 strict diffusion to a degree not less than that of 5 gm hyaluronate/liter (43). The potential importance of molecular sieving by such a layer has been stressed by Nettelbladt and Sunblad (44). Restriction to diffusion through such a matrix was calculated from the equations of Ogston (8) and Ogston, Preston, and Wells ( 4 9 , as described by Curry and Michel (46) (see Appendix). Curve B of Figure 5 shows that the theoretical curve for such a composite barrier describes the small solute data of Simkin and Pizzorno (3) reasonably well but deviates increasingly from the experimental data as molecular dimensions increase. This is due mainly to the sharp decline in endothelial permeability as solute dimensions approach pore dimensions. The low but finite permeability to solutes such as the globulins, which are far larger than the “small” pores, demonstrates the existence of an additional pathway of wider dimensions across synovial endothelium. Ultrastructural studies of the transsynovial flux of ferritin indicate that protein can cross the capillary wall in cytoplasmic vesicles (47), of radius -30 nm. When the theoretical curve is recalculated to incorporate a vesicular turnover of 2.5 x 10 - 3 cm3/min and to allow for the vesicle excluded-volume effect as formulated by Garlick and Renkin (32), the new curve (curve C, Figure 5 ) now fits the experimental data more closely. Even so, the calculated restriction to the exchange of very large proteins is still too great. This is because the interstitial matrix begins to restrict the diffusion of very large solutes to a considerable degree. Casley-Smith and Vincent (48), however, have recently described aqueous channels of radius 71 nm that permeate the intimal matrix. When diffusion and solvent-drag through these large, relatively unrestrictive channels are included in the calculation of intimal solute flux, a further improvement between theory and data results (curve D of Figure 5). Therefore, recognized components of the normal blood-joint barrier (Figure 4) can together provide a reasonably satisfactory explanation of the permeability-solute dimension relationship presented in the Results section. By use of the above analysis, the relative contributions of the endothelial and intimal layers to the overall barrier can be calculated for solutes of various sizes (Figure 6). For small lipid-insoluble solutes such as electrolytes, urea, and glucose, the intimal layer contributes more to the net barrier to exchange. For larger solutes such as the principal oncotic agent, plasma albumin, the barrier resides mainly in the endothelial layer, because the small pores or dense LEVICK 1558 1 Figure 6. Relationship between the molecular radius of solute and the relative contributions of capillary endothelium and synovial intima to the total resistance (IIPA) of the normal blood-joint barrier to solute exchange. The relative contribution of each layer to the resistance to diffusion of a small, negligibly restricted, extracellular solute was calculated from morphometric data (see text). The changes in contribution with increasing solute size were calculated from the equations for restricted diffusion and steric exclusion (scc Appendix), applied to endothelium perforated by 5 nm radius channels and containing 30 nm radius vesicles and to intimal interstitiurn permeated by 71 nm radius channels. Note that for albumin, the chief oncotic constitutent of plasma, the major barrier is capillary endothelium; but, for macroglobulins, the intima begins to dominate the resistance to exchange. fiber matrix of this layer restrict the exchange of such solutes far more than does the intimal matrix. For very large solutes such as IgM, however, the intimal layer grows to dominate the net resistance to exchange, because the growth in interstitial resistance to diffusion exceeds the decline in vesicular transport across endothelium. None of the above theoretical curves (B, C, or D) adequately describe the permeability data for the rheumatoid joint. This might be expected, because the curves are derived for normal synovium. The rheumatoid data are, however, reasonably well fitted, at least for solutes up to the dimensions of small proteins [data of Simkin and Pizzorno (4)], by a line of slope - 1 (line A). This implies that the diffusion of such solutes across the combined endothelial-intimal barrier is not significantly restricted in the rheumatoid synovium. There are three possible explanations of the unrestrict- ed diffusion of small solutes and increased permeabilit y to large solutes: 1) the existence of wider channels than normal in the “tightest” component of the barrier, namely endothelium [for example, large interendothelial gaps such as those described by Schumacher in chemically inflamed synovium (4911; 2) a reduction in endothelial fiber matrix density; or 3) a greatly increased rate of vesicular transport. Although permeability to large solutes increases in rheumatoid synovium, there is a small reduction in the permeability of the combined bloodjoint barrier to small solutes (e.g., urea), as demonstrated by Simkin and Pizzorno (4). As these authors point out, this apparent anomaly is readily resolved by the different physical effects of channel width compared with channel length or total number of channels (i.e., membrane area). An increase in channel width increases permeability to large solutes more than to small solutes (see Appendix equations) and thereby reduces the molecular selectivity of the membrane. An increase in membrane thickness (channel length) or reduction in membrane area (number of channels) reduces permeability to large and small solutes to an equal degree and does not affect the molecular selectivity of the membrane. It has, therefore, been suggested (4) that the reduced permeability to small solutes may be due to an increase in pathlength (due to hypertrophy of the intimal layer) and/or a decrease in endothelial surface area (due to obliterative vasculitis). If the number of large endothelial gaps or the rate of vesicular transport increase in rheumatoid synovium, then the contribution of bulk flow of fluid to the net transport of protein may be expected to increase (because u declines), especially for the largest proteins. The data in Table 2 allow this supposition to be tested. If protein transport across rheumatoid synovium were purely by diffusion, then the tabulated values of Jv/(R-’ - KA” -’) (total flux per unit concentration difference) would equal PA. In this event, the slope relating log PA to log res would have a minimum slope of - 1 (i.e., - 1 or a more negative value). The slope of the log-log plot for the protein data, however, is 0.09 (or -0.67 if immunoglobulins are omitted from the regression analysis). Therefore, transsynovial flow contributes significantly to the transport of protein into the rheumatoid joint. The above considerations also confirm the suspicion that equation 7 at best yields only a rough approximation of the true diffusional permeability to protein in rheumatoid joints (see Theory ) . PERMEABILITY OF HUMAN SYNOVIUM APPENDIX 1559 (PA),, The equations from which the theoretical curves of Figures 2 and 5 were calculated are as follows. Net permeability (PA). For two membranes in series: where subscripts C and S refer to capillary and synovial intimal barriers, respectively. PA for a small, negligibly restricted solute (THO) is - I cm3/minute (3). This leads to (PA)c = 2.08 cm3/minute and (PA)s = 1.92 cm3/minute for THO (see morphologic data in Discussion). Capillary permeability (PA)c. This is the sum of permeability due to "small pores" (PA),, and that due to vesicular transport (PA),. (PA),, for a solute X is a fraction of that for water because of the change in free diffusion coefficient (a l/res) and restriction of that diffusion coefficient, i.e., = THO 4' 2.08 x r x rcrL X +THO + where is the function for restricted diffusion through cylindrical pores of radius rp as developed by Pappenheimer (28,36): + = (l-re5/r,)2[1-2. I(r.=\/r,) + 2 . ~ ( r ~ ~ / r ~ ) 3 - o . 9 5 ( r ~ , / r ~ ) ~ ] Permeability due to transport by vesicles of radius r, at V ml/minute is given by Garlick and Renkin (32) as: (PA),' = Synovial interstitial permeability (PA)s. This is the sum of the permeability of the gel phase (PA), and that due to aqueous interstitial channels (PA)i,. Representing interstitial gel by a hyaluronate fiber matrix: 2. 3. 5. 6. 7. 8. 10. where restriction to diffusion within the gel phase is given by the function of Ogston et al (45): ' The "permeability" due to aqueous interstitial channels that occupy -0.76% of matrix (48) is given by: x-- +' + Jv(1 - ~ rx)Cx (cp- c,..)' 1. Knight AD, Levick JR: The control of fluid movement 12. ( d J M + rcd'l' THO x REFERENCES 11. f(0PW) = e * (The term permeability is, strictly, inappropriate here, because part of the solute transport is by bulk flow.) wx, the solvent drag coefficient for solute X , was obtained from Curry (41), assuming Onsager reciprocity. Lateral diffusion down concentration gradients between the gel phase and the aqueous channels was not modeled and would further reduce the slope of curve D in Figure 5. 9. V(I-rc,'/rv)3 1.92 x 0.76 x 10 +'"O 4. (PA),,' == 13. 14. across rabbit synovium by intravascular pressures. Microvasc Res 20:256, 1980 Levick JR: Synovial fluid dynamics: the regulation of volume and pressure, Studies in Joint Disease. Vol 2. Edited by A Maroudas. Tunbridge Wells, Pitman Medical Ltd., in press Simkin PA, Pizzorno JE: Transynovial exchange of small molecules in normal human subjects. J Appl Physiol 36581-587, 1974 Simkin PA, Pizzorno JE: Synovial permeability in rheumatoid arthritis. Arthritis Rheum 22:689-696, 1979 Sunblad LE, Jonsson E, Nettelbladt E: Permeability of the synovial membrane to glycoproteins. Nature 192:1192, 1961 Kushner I , Somerville JA: Permeability of human synovial membrane to plasma protein. Arthritis Rheum 14:560-570, 1971 Veys EM: Comparative investigation of protein concentration in serum and synovial fluid. Scand J Rheumatol 3:l-12, 1974 Ogston AG: The spaces in a uniform random suspension of fibers. Trans Faraday SOC54: 1754-1757. 1958 Laurent TC: The interaction between polysaccharides and other macromolecules. 9. The exclusion of molecules from hyaluronic acid gels and solutions. Biochem J 93:106-112, 1964 Kedem 0. Katchalsky A: Thermodynamic analysis of the permeability of biological membranes to nonelectrolytes. Biochim Biophys Acta 27:229-246, 1958 Staverman AJ: The theory of measurement of osmotic pressure. Rec Trav Chim 70:344-352, 1951 Brace RA, Granger N, Taylor AE: Analysis of lymphatic protein flux data. 111. Use of the nonlinear flux equation to estimate u and PS. Microvasc Res 16:297303, 1978 Renkin EM: Transport of large molecules across capillary walls. The Physiologist 7: 13-28. 1964 Sandson J . Hamerman D: Paper electrophoresis of human synovial fluid. Proc SOCExp Biol Med 98:564-566, 1958 LEVICK 15. Schmid K, MacNair MB: Characterization of the proteins of certain postmortem human synovial fluids. J Clin Invest 37:708-718, 1958 16. Decker B, McKenzie BF, McGuckin WF, Slocumb CH: Comparative distribution of proteins and glycoproteins of serum and synovial fluid. Arthritis Rheum 2: 162-177, 1959 17. Nettelbladt E, Sunblad L: On the acid glycoproteins of serum and synovial fluid in rheumatoid arthritis. Arthritis Rheum 4: 161, 1961 18. Nettelbladt E, Sunblad L: Haptoglobin in serum and synovial fluid. Acta Rheum Scand 11:11-14, 1%5 19. Binette JP, Schmid K: The proteins of synovial fluid: a study of the al/a2-globulinratio. Arthritis Rheum 8: 1428, 1965 20. Willumsen L, Friis J: A comparative study of the protein pattern in serum and synovial fluid. Scand J Rheumatol 4:234-240, 1975 21. Sunblad L: Studies on hyaluronic acid in synovial fluids. Acta SOCMed Upsal 58:113-238, 1953 22. Balazs EA, Watson D, Duff IF, Roseman S: Hyaluronic acid in synovial fluid. I. Molecular parameters of hyaluronic acid in normal and arthritic human fluids. Arthritis Rheum 10:357-376, 1967 23. Hamerman D, Rosenberg LC, Schubert M: Diarthrodial joints revisited. J Bone Joint Surg 52A: 725-774, 1970 24. Schubert M, Hamerman D: The functioning of the diffuse macromolecules of joints. Bull Rheum Dis 14:345-348, 1964 25. Carter RD, Joyner WL, Kenkin EM: Effects of histamine and some other substances on molecular selectivity of the capillary wall to plasma proteins and dextrans. Microvasc Res 7:31-48, 1974 26. Felgenhauer K: Protein filtration and secretion at human body fluid barriers. Miigers Arch ges Physiol 384:9-17, I980 27. Einstein A: On the movement of small particles suspended in a stationary liquid demanded by the molecular kinetic theory of heat (1905), Investigations on the Theory of the Brownian Movement. Edited by R Fiirth. New York, Dover Publications, 1956, pp 1-18 28. Pappenheimer JR, Renkin EM, Borrero LM: Filtration, diffusion and molecular sieving through peripheral capillary membranes. Am J Physiol 167:13-46, 1951 29. Chang RLS, Deen WM, Robertson CR. Brenner BM: Permselectivity of the glomerular capillary wall. 111. Restricted transport of polyanions. Kidney Int 8:212218, 1975 30. Bohrer MP, Humes HD, Baylis C, Robertson CR, Brenner BM: Facilitated transglomerular passage of circulating polycations. Clin Res 25:505A, 1977 31. Venkatachalam MA, Rennke HG: The structural and molecular basis of glomerular filtration. Circ Kes 43~337-347, 1978 32. Garlick DG, Renkin EM: Transport of large molecules from plasma to interstitial fluid and lymph in dogs. Am J Physiol 219: 1595-1605, 1970 33. Putnam FW: The Plasma Proteins. New York, Academic Press, 1975 34. Kisiel W, Hanahan DJ: Purification and characterization of human factor 11. Biochim Biophys Acta 304: 103-1 13. 1973 35. Casley-Smith JR, O’Donoghue PJ, Crocker KWJ: The quantitative relationships between fenestrae in jejunal capillaries and connective tissue channels: proof of “tunnel-capillaries.” Microvasc Res 9:78-100, 1975 36. Landis EM, Pappenheimer JR: Exchange of substances through the capillary walls, Handbook of Physiology: Circulation. Vol 11. Washington, DC, American Physiology Society, 1963, pp 961-1034 37. Suter ER, Majno G: Ultrastructure of the joint capsule in the rat: presence of two kinds of capillaries. Nature 202:920, 1964 38. Davies DV, Edwards DAW: The blood supply of the synovial membrane and intra-articular structures. Ann R Coll Surg 2:142-146, 1948 39. Casley-Smith JR, Sims MA, Harris JL: Capillary lengths and areas and intercapillary distances in tissue near the human knee. Experientia 32:64-66, 1976 40. Davies DV: Synovial membrane and synvial fluid of joints. Lancet, Dec 7:815-822, 1946 41. Curry FE: A hydrodynamic description of the osmotic reflection coefficient with application to the pore theory of transcapillary exchange. Microvasc Res 8:236-252, 1974 42. Curry FE: Is the transport of hydrophilic substances across the capillary wall determined by a network of fibrous molecules? The Physiologist 239-93, 1980 43. Fox JR1 Wayland J: Interstitial diffusion of macromolecules in the rat mesentery. Microvasc Res 18:255-276, 1979 44. Nettelbladt E, Sunblad L: On the significance of hyaluronic acid changes in the pathogenesis of joint effusions. Opusc Med 12:224-232, 1967 45. Ogston AG, Preston BN, Wells JD: On the transport of compact particles through solutions of chain-polymers. Proc K SOC(Lond) A 333:297-316, 1973 46. Curry FE, Michel CC: A fibre matrix model of capillary permeability. Microvasc Res 20:96-99, 1980 47. Chamberlain MA, Petts V, Gollins E: lransport of intravenously-injected ferritin across the guinea-pig synovium. Ann Rheum Dis 31:493-499, 1972 48. Casley-Smith JR, Vincent AH: The quantitative morphology of interstitial tissue channels in some tissues of the rat and rabbit. Tissue Cell 10571-584, 1978 49. Schumacher HR: Fate of particulate material arriving at the synovium via the circulatici... an ultrastructural study. Ann Rheum Dis 32:212-218, 1973

1/--страниц