INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE IMPORTANCE OF TRANSCYTOSIS as a bulk transporter for macromolecules in the microvascular barrier has been vigorously debated over the past four decades. Ever since the discovery of plasmalemmal vesicles (19), morphologists have proposed that transcytosis is a major mode of protein transport across vascular endothelia (19). However, many studies (26-30, 46) have shown that transvascular large solute transport has characteristics that do not comply with "active" transcytosis. From a physiological perspective, transcapillary bulk transport of macromolecules appears to occur through aqueous channels, i.e., through pores or other size-selective structures (slits with a matrix of fibers) in the blood-tissue barrier. Such conclusions are mainly on the basis of a low cooling sensitivity of transvascular protein (albumin) transport (30, 33) and the marked coupling of protein transfer to transcapillary fluid filtration (29, 40). There have also been studies (2, 3, 7) that used ultrathin (~140 Å) serial sectioning electron microscopy (EM) of endothelial cells from various capillaries that show very few free plasmalemmal vesicles. Virtually all of the free vesicles observed by use of routine EM appear to be stagnant parts of racemous invaginations from the cell surface. This has been demonstrated by both conventional chemical fixation as well as after "ultrarapid" fixation by using slam-freezing (8).

During the past decade, novel approaches have been applied to discriminate active "vesicular transport" from passive "porous transport." N-ethylmaleimide (NEM) is an alkylating agent that appears to inhibit vesicular transport by interacting with the docking and fusion of vesicles with their target membranes (32, 34). Filipin is a cholesterol-binding agent that is assumed to inhibit transcytosis by binding to the caveolas and removing cholesterol from the plasmalemma, thereby causing disassembly of the vesicles (36). Recently, NEM has been claimed to inhibit albumin and small protein transfer in the murine heart (24, 25) as well as albumin transport across the rat lung microvasculature (34). Furthermore, the transport of radiolabeled albumin across the microvasculature of rat lungs appeared to be partly inhibitable by filipin (36). These studies were, however, performed in perfusion experiments in situ without continuous monitoring of pre- and postcapillary resistances, capillary filtration coefficients (L_pS), tissue weight, and without previous heparinization of the animals. Furthermore, the degree of vascular wash out achieved, whether complete or not, cannot be determined from these studies.

On the basis of this background, we decided to assess microvascular albumin transport at lowered temperature and after either NEM or filipin under well-controlled hemodynamic conditions in an isolated perfused rat lung. In this experimental model, vascular pressures and pre- and postcapillary resistances could be controlled and measured, respectively, whereas capillary hydraulic conductance could also be assessed. Transvascular tracer albumin transport was measured with the use of an isogravimetric albumin clearance [Cl_iso or "apparent permeability suface (PS) area product"], when net transcapillary fluid filtration flux (J_v) was zero. Also, an apparent reflection coefficient for albumin (_alb) was measured at elevated J_v values, by using a dual tracer albumin tissue uptake technique. With the use of these measurements there was no evidence that the "transcytosis inhibitors" employed reduced Cl_iso or increased _alb. On the contrary, whereas cooling from body temperature to room temperature slightly reduced the Cl_iso (largely in proportion to the concomitant increases in perfusate viscosity), both NEM and filipin produced increases in Cl_iso and reductions of _alb, which is compatible with an increased permeability of the lung endothelial barrier.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Isolated Perfused Rat Lung

1

1

Pulmonary arterial (P_A), pulmonary venous (P_V), and airway pressures (P_airw) were measured with the use of pressure transducers (Cobe, Lakewood, CO), and the lung weight was continuously recorded by using the force displacement transducer. P_A, P_V, P_airw, and tissue weight were continuously monitored on a polygraph recorder (model 7B, Grass Instruments, Quincy, MA). Dual perfusion systems, connected to each other via a stopcock just before the PA cannula (and before P_A line) and just after the LA cannula (and after the P_V line), and having a bypass circuit, allowed rapid shifts between perfusate I and perfusate II, leaving P_A, P_V, and tissue weight undisturbed. The second perfusion line allowed the nonperfusing solution, usually perfusate II, to bypass the lung preparation while recirculating. Lungs were always perfused in zone III conditions, unless otherwise specified.

Measurements of Pulmonary Capillary Pressure

Measurements of L_pS

(1)

1

1

(2)

Vascular Resistances

1

1

(3)

(4)

Assessment of Cl_iso and _alb

Tracer I was used to measure the initial Cl_iso during 7-8 min, whereas tracer II was used to measure the convective albumin clearance during transvascular filtration over a 7- to 8-min period after a defined increase in microvascular pressure produced by raising P_A and P_V by the same amount. Tracer I (4-5 µCi) was contained in a separate circuit (perfusate II) containing 70 ml of perfusate. It was possible to instantly switch to this perfusate containing tracer by using stopcocks. The perfusate reached the lung within 10 s, and the switch did not affect P_A, P_V, P_C, weight, or P_airw levels. A sampling from the tracer I reservoir (~0.3 ml at a time) was made at 2-min intervals. At 6 min, we suddenly added 4-5 µCi in 0.5 ml of tracer II to the tracer I (perfusate II) reservoir and stirred vigorously. An air bubble in the arterial line, which was later trapped within a bubble trap, delineated when the new tracer replaced the first. After the air bubble entered the bubble trap, P_A and P_V were rapidly and simultaneously raised to induce vascular filtration. Sampling of perfusate was then again performed at 2-min intervals. After 6-8 min of tracer I + tracer II perfusion, a switch was made to the unlabeled "cold" perfusate (perfusate I) to completely wash the lung's vasculature free of tracer. Washout fluid (~40-50 ml) reduced tracer concentration to usually <0.2% of the initial perfusate value. In a few preliminary experiments, only tracer I was used during strictly isogravimetric conditions. After a tracer washout, the lungs were removed, the right and left lungs were blotted and weighed in tared test tubes separately, and then radioactivity was assessed for ¹²⁵I and ¹³¹I. Appropriate corrections for spillover and radioactive decay were also performed. A minimum of 10,000 counts/min (CPM) per sample were also counted on a Wallac scintillator (model 1282, CompuGamma, Turku, Finland). Lung samples were dried at 60°C until a constant weight was obtained (~72 h).

Calculations

(5a)

(5b)

(6)

Note that Cl_iso contains both a diffusive term (PS) and a term due to "volume recirculation" of fluid between large and small pores during isogravimetric conditions (28-30). It could, however, be calculated that this term was rather small (~20% of Cl_iso) by using current data for small and large pore radius (60 and 250 Å, respectively) and for the fraction of L_pS accounted for by the large pores (0.01-0.15) (29, 40).

Clearance during transvascular filtration (Cl_F) was calculated as

(7)

Furthermore

(8)

J_V is the average filtration rate occurring during Time_II, defined as total weight gain minus the first minute weight gain divided by the elapsed time during pressure elevation. The use of Eq. 8 deviates from the linear equation (17) but is justified according to the nonlinear flux equation (22), because J_V was usually >2 ml · min¹ · 100 g¹, PS ~0.2 ml · min¹ · 100 g¹, and _alb ~0.65, yielding a Peclet number [J_V · (1)/PS], which is >3. Under this condition the Patlak equation (22) reduces to Eq. 8, and hence, the apparent PS term approaches zero (1, 28, 40).

The degree of edema (E) was calculated from dry-to-wet weight ratio of experimental lungs (DW/WW)_E and those of controls (DW/WW)_O, i.e., 0.18, from the formula

(9)

The edema factor, f_E = 1 + E/100, was used to always standardize the extravascular albumin space to tissue dry weight (actually to the wet tissue weight before any edema formation had occurred), by multiplication of all albumin spaces by f_E.

Experimental Protocol

Cooling experiments (n = 7). After preparation and a period of stabilization (10-15 min) during which the lung was allowed to become completely isogravimetric, a control L_pS (10-12 min) was assessed, after which the lung again reached isogravimetric conditions and P_C was determined by using the double-occlusion technique. By suddenly reducing the heater temperature and that of the heated (arterial) reservoirs to ~8°C (from 38°C), the perfusate temperature was reduced to ~22°C within 3-5 min. Perfusate temperature in the reservoir was measured by using simple mercury lab thermometers. After approximately another 10 min, a double-occlusion P_C was again determined, after which the tracer infusion started and Cl_iso and _alb were determined during 7 + 7 min, according to the procedures previously described.

NEM (n = 5) and filipin (n = 6) groups. NEM and filipin were both obtained from Sigma Chemical. When the preparation became isogravimetric after an initial L_pS determination, NEM was infused for 4 min directly proximal to the P_A cannula to produce a final concentration of 0.13 mM (24, 34). Cl_iso and _alb were then determined according to the description presented in Assessment of Cl_iso and _alb and Calculations. In the filipin group, filipin concentrations of 0.22, 0.44, 0.9, and 0.9 µg/ml (one concentration per animal, n = 4) were tested without yielding any significant effects on either Cl_iso, L_pS, or _alb (36). Therefore, in two experiments, a filipin concentration of 1.8 µg/ml was infused.

Control experiments (n = 8 + 5). In eight control experiments, a sham infusion of NaCl (0.3 ml/min) was given before the determinations of Cl_iso and _alb. In five preliminary control experiments, which are included in the main control group, only one albumin tracer (¹²⁵I-BSA) was employed, and therefore some  determinations were not performed in the control group.

Statistics

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of Cooling (n = 7)

1

1

1

1

1

1

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Effects of temperature reduction (solid bars) on isogravimetric albumin clearance (Cliso, ml · min1 · 100 g1), the capillary ultrafiltration coefficient (LpS, ml · min1 · cmH2O1 · 100 g1), the albumin reflection coefficient, alb (dimensionless), and total vascular resistance (Rtot, cmH2O · ml1 · min · g) compared with control (open bars). Moderate cooling produced only slight changes (~30%) in both Rtot (**P < 0.01) and LpS (*P < 0.05). Cliso tended toward reduction, but the reduction compared with control was not statistically significant according to the corrected ANOVA.

Effects of NEM

1

1

1

1

1

1

1

1

1

1

1

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Effects of 4 min of N-ethylmaleimide (NEM, 0.13 mM) exposure (solid bars) on Cliso (ml · min1 · 100 g1), LpS (ml · min1 · cmH2O1 · 100 g1),  (dimensionless), and Rtot (ml · min1 · cmH2O · 100 g) compared with control (open bars). Statistically significant (**P < 0.05) differences occurred in both the Cliso and LpS, which both increased. Furthermore, there was a reduction in alb (*P < 0.05).

Effects of Filipin (n = 6)

1

1

View larger version (13K): [in this window] [in a new window]   Fig. 3.   A nonlinear regression analysis of Cliso as a function of filipin dose delivered is Cliso = 0.154 · exp (0.841x); r = 0.98, where x represents the filipin dose.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Lumped results for all filipin studies (n = 6) for Cliso, LpS, alb, and Rtot (black bars) compared with control lungs (open bars). Dimensions are the same as in Figs. 1 and 2. The change in Cliso after filipin was statistically significant (*P < 0.05). However, the apparent reduction in alb was not statistically significant.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present results support the contention of passive, pore-type transfer of tracer albumin across the microvessels in ventilated rat lungs perfused under strictly controlled hemodynamic conditions by using a buffered PSS-albumin serum solution. Thus reducing perfusate temperature from 35° to 22°C only moderately reduced transvascular clearance of albumin during isogravimetric conditions (in due proportion to the increased viscosity) and not to the extent expected for an active transport process. Furthermore, when P_C was increased, there was a clear-cut increase in albumin transport, the coupling coefficient between albumin clearance and J_V being ~0.35, compatible with an average _alb of ~0.65 both at 35° and 22°C. Finally, the use of transcytosis inhibitors in concentrations expected to affect vesicular transport did not reduce transvascular albumin clearance from control in our experiments. In contrast, even low doses of NEM (0.13 mM) and high doses of filipin (1.8 µg/ml) actually increased Cl_iso and reduced _alb, at least after NEM, concomitant with increases in capillary hydraulic conductance, indicating that NEM and filipin both altered microvascular permeability.

The microvascular changes after NEM in the present experiments mimick those recently obtained in isolated perfused rat hindquarters (Carlsson O, Rosengren B-I, and Rippe B, unpublished observations) or in singly perfused frog mesenteric capillaries (Michel and Neal, personal communications), although the present doses of NEM were somewhat lower and did not markedly affect vascular resistance. However, they are in stark contrast to the apparent transport inhibition by NEM obtained by Schnitzer et al. (34-36) in in situ perfused rat lungs. The apparent PS (Cl_iso) during control in studies by Schnitzer et al. was about three- to fivefold higher than that in our experiments. The reason for this discrepancy is not known. A major difference between the experiments of Schnitzer et al. and the present study is that our perfusate always contained 10 vol% serum and that the albumin concentration was higher. Furthermore, lung cannulation, isolation, and perfusion were always preceded by heparinization of the animals in our experiments. In a preliminary study, in which only Cl_iso was assessed, we found that the apparent PS increased approximately threefold when serum was absent from the perfusate. These clearance data agree with those of Sunnergren et al. (39), who studied albumin clearances in serum free but PSS-albumin-perfused rat lungs in vitro. The serum effect is well documented for rat muscle capillaries (11, 12), rat glomerular vessels (9), frog mesenteric capillaries (4, 14), and partly for rat lung microvessels (35). The way that serum, or actually orosomucoid, exerts its effects is not known, but previous studies have indicated that orosomucoid affects the charge-selective properties of the capillary walls, making them physiologically negatively charged (4, 12). The serum effect may be one of the reasons for the observed discrepancy in control albumin clearance of the present experiments and those by Schnitzer et al. (34-36).

Another major discrepancy between the present experiments and the in situ perfusions by Schnitzer et al. is the relatively longer tracer-loading period in the present experiments than in those by Schnitzer et al., i.e., ~15 min for tracer I and 7-8 min for tracer II, versus 3 min in the studies of Schnitzer et al., and the seemingly better-controlled washout in the present study. Whereas the tracer loading and the tracer washout periods were about equal in length (3 min) in Schnitzer's (34-36) experiments, the washout periods in the present experiments accounted for only 20-50% of the tracer loading periods. In the original lung tissue uptake studies by Kern et al. (18), there were no significant reductions in Cl_iso as a function of tracer perfusion time during tracer perfusion periods up to 30 min, the half-time of the (exponential) lung albumin tissue (saturation) uptake curve being 77 min. The calculated theoretical reduction in clearance by using 15 min instead of 3 min of tracer loading is actually negligible (~5.5%). Therefore, the longer tracer loading periods used here are preferred because of the larger tracer albumin space monitored (more CPM/g to be counted), especially in relation to any residual tracer remaining in the intravascular space. This may have reduced errors in determinations of clearance and _alb in our experiments. Furthermore, and more importantly, strict precautions were taken in the present experiments to produce an efficient washout. In a majority of experiments the final rinse fluid contained only ~0.1% of the initial tracer concentration.

The average Cl_iso in our experiments is similar to that determined in rats in vivo from albumin kinetic studies (38) but slightly higher than values obtained by using dual isotope techniques for separation of extravascular from intravascular albumin spaces in vivo. With the use of such techniques, Haraldsson et al. (10) reported lung albumin clearance in rats in vivo to be 0.116 ml · min¹ · 100 g¹ from the 30-min extravascular tracer albumin space, similar to that of Dewey (6) (0.100 ml · min¹ · 100 g¹). The latter values are remarkably close to those reported by Ishibashi et al. (16) combining tissue uptake and lymphatic flux analyses in intact dog lungs (0.103 ml · min¹ · 100 g¹). Our Cl_iso data are one-third as high as those obtained in in vitro-perfused lungs without serum in the perfusate (39), which is most likely due to the presence of serum in our study. In one previous in vivo study rat lungs were cannulated in situ to wash out intravascular tracer albumin previously administered to the whole animal, the washout lasting for 3 min (5). In that study, however, albumin clearance was markedly higher than in the other in vivo studies mentioned above, and it is speculated that in situ vascular washout procedures, especially if the animal is not heparinized, may not have been adequate. If vascular washout had indeed been incomplete in the aforementioned in situ studies, including those of Schnitzer et al. (34-36), the effects of NEM or filipin may be ascribed to vasoconstriction and reductions in vascular capacitance. This would affect apparent distribution volume, hence, apparent clearance, of both small and large proteins, which show a large intravascular-to-extravascular space ratio in experiments of short duration. However, for small solutes, intravascular distribution volume is rather insignificant compared with the total extravascular distribution volume in experiments lasting a few minutes, and any changes in the former would not affect solute distribution volumes or estimates of blood-to-tissue clearances significantly.

It is not completely understood how NEM and filipin affect vascular permeability. However, NEM is an alkylating agent, which acts like a fixative, causing marked crosslinking of proteins of, i.e., the cytoskeleton as well as of those of the plasmalemma. Membrane-damaging effects may also be produced by high doses of filipin, because filipin is a sterol-binding agent, which removes cholesterol from the plasmalemma, thereby potentially disrupting the cell membrane. The present study and a parallel one by Carlsson O, Rosengren BI, and Rippe B (unpublished data) clearly point out the unsuitability of using NEM and filipin as transcytosis inhibitors. As it seems, the only physiological transcytosis inhibitor available at present is cooling. In the present experiments, however, cooling caused only moderate reductions in albumin clearance, which was reduced largely in proportion to the cooling-induced changes in viscosity produced.

During the past two decades, much has been learned about the general principles of vesicle shuttling between donor and target membranes in cells. The formation of a vesicle (vesicle budding) includes the formation of a protein shell from cytosolic proteins, named a "coat," which sculpts the vesicle out of the donor membrane. This vesicle budding is triggered by a GTP-dependent protein. In the next step the vesicle is detached and "decoated," allowing passage of the vesicle across the cell, usually by passive mechanisms. The vesicle and its target membrane have specific identifiers (so-called "SNAREs" or "SNAP receptors"), which can bind to each other, thereby eventually docking the vesicle to the target membrane, usually after minutes of transcellular passage. NEM-sensitive fusion protein, N-ethylmaleimide-sensitive factor (NSF), and soluble NSF attachment protein (SNAP) bind to the docking receptor complex, disrupting the complex and initiating fusion, when NSF hydrolyzes ATP. Thus at least two of the processes in vesicular shuttling, budding from the donor membrane and fusion with the target membrane, are dependent on the cellular metabolism. As such they should have a temperature coefficient of the order of 2.5-3, which would imply a reduction in shuttling velocity by ~70% for a temperature reduction from 35° to 22°C. However, if transport occurred by pure convection through large pores, then the reduction in transport would be exactly inversely proportional to the increase in viscosity. In that case, transport would be reduced by 25%. A diffusive process would, according to the Stokes-Einstein equation, be reduced by 28% for the same temperature reduction. These figures are more or less exactly consistent with the cooling-induced reduction found in albumin transfer in the present study, although statistically, the reduction in Cl_iso at 22°C was of borderline significance (P = 0.07). Our data thus strongly favor passive transport mechanisms as driving the transendothelial protein transport. Also the fact that the coupling between albumin transfer and fluid transfer, via (1_alb), was completely unchanged during cooling, supports this contention. In addition, although NEM even in low doses altered microvascular permeability, low doses of filipin left vascular permeability relatively unperturbed. Thus low doses of filipin did not reduce transendothelial albumin transport.

The present study does not refute the presence of endocytotic and exocytotic pathways or transcytosis of proteins across the endothelium in general. However, the mere existence of a pathway does not prove that it is quantitatively important. Hastings et al. (13) proved the existence of a monensin and nocodazole-inhibitable endocytotic pathway in cultured alveolar type II cells. They furthermore showed, by using immunohistochemical techniques, that transcytosis inhibition disturbed the albumin uptake from the alveoli in vivo. However, when the quantitative alveolar clearance of ¹²⁵I-labeled immunoglobulin G or ¹³¹I-labeled albumin was investigated in anesthetized rabbits, there was no significant inhibition of this bulk protein transport from the alveoli to the plasma. The authors concluded that endocytic pathways are insufficient to account for the clearance of large quantities of serum proteins from normal alveoli. It is then likely, that endocytotic and exocytotic mechanisms, in e.g., endothelia, have other primary functions than being a bulk transfer mechanism for proteins. This does not deny that transcytosis may be of importance in highly specific transendothelial protein transport processes of, e.g., signal peptides, such as procolipase (31), hormones, such as insulin (14), or even cytokines and chemokines. It is also possible that plasmalemmal vesicles may fuse to form large patent transendothelial channels (37). However, so far we have found no evidence supporting the hypothesis that active transcytosis produces bulk transfer of plasma protein from blood to tissues.

In summary, the present study supports the contention of passive transfer of albumin across the lung's microvascular endothelium. Tissue cooling failed to produce the marked reduction in albumin clearance, which would be expected for an active transport process. Furthermore, the transcytosis inhibitors NEM and filipin did not reduce albumin clearance across lung microvessels. In fact, both NEM and filipin, used as transcytosis inhibitors in many previous studies, produced changes in microvascular exchange parameters that are compatible with increases in microvascular permeability. The data suggest that both NEM and filipin produce some form of toxic endothelial damage of the microvascular barrier in isolated perfused rat lungs.