INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MORPHOLOGICAL EVIDENCE of free vesicles in the cytoplasm of mesothelial cells of peritoneum, pericardium, and pleura has been provided for a long time (17, 22, 25, 29, 32, 40, 51, 52). Moreover, morphological evidence for vesicular transport of macromolecules from the luminal to the interstitial side of mesothelium has been provided in rat parietal pericardium (28) and mice parietal peritoneum (17, 18, 23). This information, however, is insufficient compared with the wealth of information on vesicular transport available for endothelium (12, 19, 33, 34, 45-47, 50), which, however, has been challenged (42) even recently (43). By analogy to what is known about the endothelium, and on the basis of some hints on the pleural mesothelium, morphologists (23, 51, 52) suggested that vesicles of the pleural mesothelium might provide transcytosis. On the other hand, the prevailing view among physiologists about protein exit from the pleural space is that most of it occurs through the lymphatic stomata of the parietal pleura (35, 37-39, 48). This view is based on the finding that, after ligation of the right lymphatic duct and thoracic duct, only a small fraction of the labeled albumin injected into the pleural space reaches the blood (13, 35). This finding, however, does not prove that most labeled albumin leaves the pleural space through the lymphatic stomata, because albumin leaving the pleural space outside the stomata is eventually drained by the lymphatics of the interstitial space (3). Therefore, following the suggestions of morphologists, it has been proposed that proteins could leave the pleural space also by transcytosis, in addition to lymphatic drainage through the stomata of the parietal mesothelium and solvent drag through the visceral mesothelium (3). Recently, we determined the permeability (P) of the mesothelium to small, medium, and large molecules and determined the equivalent radius of the "small pores" of its intercellular junctions (2, 8, 55). P to albumin was markedly greater than predicted by the relationship between P and Stokes-Einstein radius (a) of the solutes. This suggested that albumin transfer across the mesothelium should mainly occur through "large pores" and/or transcytosis. Indirect evidence for transcytosis was provided by analyzing, in a way similar to that used by Renkin (41) for capillary endothelium, the P-a relationship of macromolecules through the mesothelium (8).

The purpose of this research is, therefore, to provide evidence for transcytosis in the mesothelium from lumen to interstitium (L-I). To this end, we used specimens of parietal pericardium because specimens of this serosa may be obtained with less tissue damage than specimens of pleura (55). First, we determined albumin transfer at 37°C and albumin concentration (C_alb) 0.5% in the direction interstitium to lumen (I-L; i.e., opposite to that previously measured; Ref. 8) to ascertain the occurrence of a net albumin flux in the direction L-I. Second, we determined albumin transfer in both directions at 12°C and C_alb 0.5%, because it has been shown in other tissues that at about this temperature, transcytosis ceases (24). Third, we determined albumin transfer at 37 and 12°C with C_alb lower or higher than that which was previously used (0.5%) in an attempt to ascertain whether C_alb affects transcytosis and whether the vesicular transport of albumin occurs, besides in fluid phase, also through binding with the vesicular surface as in capillary endothelium (45). Fourth, we determined the transfer of dextran 70 in the direction I-L (i.e., opposite to that which was previously measured; Ref. 8) at 37°C and C_alb 0.5%, L-I at 12°C and C_alb 0.5%, and at low C_alb and 37°C, to extend the study to a different molecule, which should not be catabolized during the transfer and should be transferred only in fluid phase. This allows the determination of its vesicular concentration and, hence, the computation of the vesicular liquid flow. Fifth, we determined the transfer of dextran 70 under the conditions used in the previous measurements (8) after the addition of an inhibitor of transcytosis (nocodazole; Ref. 16).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Specimen collection and preparation. Specimens of the retrosternal parietal pericardium were obtained from 117 rabbits (body wt 4-6 kg, age 5-10 mo). Rabbits were purchased from bmg farm (Cividate al Piano, Bergamo). Animal experimentation was authorized by the Ministry of Health by decree N. 36/94A according to decree law 116/92, in compliance with guidelines 86/609/CE. The animals were anesthetized with a solution containing pentobarbital sodium (Sigma, 10 mg/ml) and urethane (Sigma, 250 mg/ml), 2 ml/kg iv, or with a solution containing ketamine hydrochloride (Sigma, 10 mg/ml) and xylazine hydrochloride (Sigma, 3 mg/ml), 2 ml/kg iv. They were placed supine on a tilting board 20° head up. The trachea was cannulated to ensure adequate ventilation during the preliminary surgical procedure, and air flow and tidal volume were recorded on a 7418 Hewlett-Packard thermopaper oscillograph. Collection and preparation of the specimens were performed with the procedure previously described, which minimizes manipulation and air exposure of the mesothelium (55). Briefly, after killing the rabbit by an overdose of anesthetic, we removed a segment of sternum, leaving undamaged the underlying parietal pericardium, which, in this region, should be free of lymphatic stomata (25). While albumin-Ringer solution was being poured on the pericardium to prevent air exposure of the mesothelium, a roughly rectangular specimen of pericardium (~3 × 2 cm) was hooked and excised. The specimen was never stretched during removal, and the whole procedure was completed within 4 min after the death of the animal. The specimen, covered by albumin-Ringer solution, was pinned with its interstitial side facing upward, at its in situ length and width, to a layer of Sylgard (Dow Corning) adhering to the bottom of a petri dish. The solution was bubbled continuously with a 95% O₂-5% CO₂ gas mixture. Small vessels, fat patches, and, when present, blood clots were removed from the interstitial side of the specimen until a transparent area of ~1 × 1.5 cm was obtained; the mesothelium of the central part of the specimen was never touched. The cleaning procedure took 20-25 min.

Solutions and labeled molecules. The composition of the Ringer solutions used during specimen collection and preparation, as well as during the measurements (see below), was (in mM): 139 Na⁺, 5 K⁺, 1.25 Ca²⁺, 0.75 Mg²⁺, 119 Cl, 29 HCO, and 5.6 D-glucose. Bovine serum albumin (Sigma) was added to the solution to have a 0.1% concentration in the preparation periods, a 0.5% concentration in the first incubation period (see below), and the following concentrations (C_alb) in the various series of experiments: 0.005% (a value a little above the minimum required to keep normal permeability; Ref. 15), 0.05%, 0.5% (that which was previously used; Ref. 8), and 2.5%. The labeled molecules used were bovine ¹²⁵I-albumin (specific activity ~1 mCi/mg, ICN Biomedicals) and 70 kDa FITC-dextran (0.013 mol FITC/mol glucose, Sigma). The mean activity of ¹²⁵I-albumin in the solution placed in the donor chamber of the Ussing apparatus was ~1 µCi/ml (ranging from 0.7 to 1.8 µCi/ml). Because of the fast decay of ¹²⁵I-albumin (half-life 59.7 days), increasing amounts of the ICN albumin (which is only partly labeled) had to be added as time elapsed: the amount added to 1 ml of solution ranged from 1 to 3 µg (i.e., 1.4 × 10⁵ to 4.3 × 10⁵ µmol). This amount is negligible relative to C_alb, except when the latter is 0.005%, because in this case the ICN albumin added is 2-6% of C_alb. Unbound ¹²⁵I is present in the labeled albumin solution, and a correction for the inherent radioactivity was made (see below). To check whether a relatively high concentration of unbound ¹²⁵I affected the results despite the correction, labeled albumin solution was filtered before use in part of the experiments (see below and RESULTS). Filtration was made by centrifuging at 5,000 g for 30 min at 4°C through low protein binding membrane (Millipore) with 30-kDa nominal molecular weight cut off (8). FITC-dextran was used at a concentration of 0.78 × 10² µmol/ml. Unlabeled dextran was added at the same concentration in the recipient chamber. To minimize the concentration of free FITC, solutions containing FITC-dextran were dialyzed for ~16 h at ambient temperature before the experiments.

Measurements of unidirectional flux and permeability. Specimens were mounted as planar sheets between the frames of a Ussing apparatus (rectangular window: 0.5 cm², chambers vol 4 ml). Both chambers were immediately and simultaneously filled with albumin (0.5%)-Ringer solution. A first incubation period of 30 min was allowed for tissue recovery. Solutions contained in the chambers were oxygenated and stirred throughout the experiment by bubbling 95% O₂-5% CO₂ through ports opening near the bottom of the frame in each chamber; the apparatus was water-jacketed to maintain the temperature appropriate to the kind of experiment (37 or 12°C, see below), and, accordingly, solutions were heated at 37°C or cooled at 12°C before adding them to the chambers. At the end of the first incubation period, both chambers of the Ussing apparatus were simultaneously emptied and refilled with labeled solution in the donor chamber and unlabeled solution in the recipient chamber. At this stage, the solution with the appropriate C_alb for a given experimental series was used in both chambers. A second incubation period of 40 min was allowed to attain steady state of the tracer in the specimen. At the end of this period, a 50-µl sample was withdrawn from the donor chamber while the recipient chamber was emptied and immediately refilled with 3.95 ml of fresh unlabeled solution. At the end of this procedure, which required 3-4 s, the first measurement period started. The duration of the measurement period was 30 min. At the end of this period, a second measurement period of equal duration was performed after the above procedure was repeated. The values of the two experimental periods were averaged except when the second one exceeded the first one by >20%. In these cases (<12%), only the first one was taken. No short-circuit current was applied because no electrical potential difference was found across the in vitro specimens of parietal pericardium (9).

Series of experiments. Labeled albumin flux, P_alb, and albumin flux were determined in the following series of experiments. Series 1 was in the direction I-L under the conditions (37°C, C_alb 0.5%) previously used for the measurement in the direction L-I (8). This series was performed to ascertain, by comparison with L-I, the occurrence of a net albumin flux. Moreover, four additional experiments L-I, 37°C, C_alb 0.5% were done in which labeled albumin solution was filtered before use (see above). Series 2 was in the direction L-I at 12°C with C_alb 0.5%. Series 3 was in the direction I-L at 12°C with C_alb 0.5%. Series 2 and 3 were performed to ascertain whether the net flux disappears, because it has been shown in other tissues that at ~12°C, transcytosis ceases (24). The values obtained in series 2 and 3 (and in the following series at 12°C, see below) were corrected to bring water viscosity and kinetic energy of solutes to 37°C. Other series were L-I at 37°C with C_alb 0.005%, 0.05%, and 2.5% and L-I at 12°C with C_alb 0.005%, 0.05%, and 2.5%. These series were performed for the following reasons: 1) to assess whether C_alb affects vesicular formation and, hence, vesicular transport and 2) to assess whether the vesicular transport of albumin, besides in fluid phase, also occurs through binding to the vesicular membrane, like in capillary endothelium (45). In this case, a competition for binding should occur between albumin molecules, and, therefore, the vesicular flux of labeled albumin should decrease with increasing C_alb. In three experiments of the series I-L, 12°C, C_alb 0.5%, and in all of those of the series L-I, 12°C, C_alb 0.05%, labeled albumin solution was filtered before use (see above).

2

2

Statistics. Data are expressed as means ± SE. Statistical significance of differences between groups was determined by unpaired t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Albumin. The value of P_alb of the parietal pericardium in the direction I-L at 37°C was 0.092 × 10⁵ cm/s, i.e., 53% (P < 0.02) of that in the direction L-I, 0.172 × 10⁵ cm/s (Table 1). Because the catabolism of albumin in the luminal chamber was essentially nil (see METHODS), the greater albumin flux in the direction L-I than in I-L cannot be due to a diffusion of peptides in the direction L-I. Therefore, most of the net albumin flux should represent an active transport in the L-I direction. The value of P_alb L-I at 12°C (corrected to bring water viscosity and kinetic energy of solutes to 37°C) was 0.089 × 10⁵ cm/s, i.e., 52% (P < 0.02) of that L-I at 37°C (see Table 1). The value of P_alb I-L at 12°C (corrected as before) was 0.084 × 10⁵ cm/s, i.e., similar to that L-I at the same temperature (Table 1). The finding that at 12°C the net albumin flux disappears shows that it is due to an active transport, likely transcytosis, because at this temperature the vesicles do not occur (24) and because of the results obtained with dextran 70 (see Dextran).

2

5

5

5

5

4

4

4

1

2

4

1

2

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Total (circles) and diffusive (broken line) albumin flux through specimens of parietal pericardium vs. albumin concentration in a log-log plot. , lumen to interstitium at 37°C. +, lumen to interstitium at 12°C (corrected to bring water viscosity and kinetic energy of solutes at 37°C). , interstitium to lumen at 37°C. , interstitium to lumen at 12°C (corrected as before). Vesicular albumin flux is given by vertical distance between circles and dotted line.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Vesicular albumin flux lumen to interstitium through specimens of parietal pericardium vs. albumin concentration.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Total (continuous line) and diffusive (broken line) flux of labeled albumin through specimens of parietal pericardium vs. albumin concentration. Vesicular flux of labeled albumin is given by the vertical distance between continuous line and broken line. Symbols same as in Fig. 1. cpm, Counts per minute.

Dextran. The value of P_dx of parietal pericardium I-L at 37°C and C_alb 0.5% was 0.026 × 10⁵ cm/s, i.e., lower (P < 0.01) than that L-I at the same temperature and C_alb, 0.095 × 10⁵ cm/s (Table 4). The value of P_dx L-I at 12°C and C_alb 0.5% (corrected to bring water viscosity and kinetic energy of solutes to 37°C) was 0.038 × 10⁵ cm/s, i.e., lower (P < 0.05) than that at 37°C (Table 4). Moreover, the value of P_dx with 40 µM nocodazole L-I, 37°C, Calb 0.5% (n = 10) was 0.032 (± 0.012) × 10⁵ cm/s, i.e., lower (P < 0.02) than that without nocodazole L-I, 37°C, C_alb 0.5% and similar to those I-L, 37°C, C_alb 0.5% and L-I, 12°C, C_alb 0.5%. These findings show the occurrence of transcytosis in the mesothelium from L to I. The value of P_dx L-I at 37°C and C_alb 0.005% was 0.030 × 10⁵ cm/s, i.e., lower (P < 0.02) than that with C_alb 0.5% (Table 4). This finding shows that transcytosis in mesothelium ceases when C_alb is 0.005%.

5

1

2

5

1

2

5

1

2

5

1

2

2

1

2

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results show the occurrence of an active transport of albumin and dextran from the luminal to the interstitial side of the mesothelium of the parietal pericardium. The findings that this active transport involves dextran and ceases at 12°C or with nocodazole show that this transport is due to transcytosis, in line with our previous indirect evidence (8). Moreover, the results show that transcytosis decreases progressively at low C_alb and eventually vanishes when C_alb 0.005%. This suggests that this vesicular transport is not constitutive, but appears to be activated by albumin. With C_alb similar to that which occurs under physiological conditions in the pleural or pericardial liquid, ~1% or 0.14 µmol/ml (20, 36, 44), J_alb,ves through the mesothelium should be ~5 × 10⁴ µmol · h¹ · cm². This is likely an underestimation because no correction was made for TCA-soluble peptides formed during transcytosis (see METHODS and RESULTS). Vesicular liquid flow (J_liq,ves), computed from the vesicular flux of dextran 70 (which is only fluid phase) with C_alb 0.5%, should be ~3.5 µl · h¹ · cm².

In a study on the absorption of colloidal thorium dioxide from the pericardial cavity of rats, morphological evidence suggested that it is carried by vesicles through the mesothelium of the parietal pericardium and, to a much smaller extent, of the visceral pericardium (28). Thorotrast was then shown to be drained by lymphatics (30). Moreover, radioiodinated serum albumin placed in the pericardial cavity of rabbits was found to leave through the parietal pericardium and to be drained by lymphatics (21). Finally, the lymphatic drainage from the pericardial cavity of conscious sheep, estimated from the egress of labeled serum albumin, was found to be 0.04 ml/(h kg) (10).

An active transport of bovine serum albumin through bullfrog alveolar epithelium was found by Kim et al. (27). Part of the albumin transferred was intact. Net P in the alveolar to pleural direction was 0.018 × 10⁵ cm/s, i.e., one-fifth of that presently found in rabbit mesothelium (0.084 × 10⁵ cm/s; Table 1, from the value of the first line minus the average value from the second, third, and fourth lines). In this connection, one has to consider that their measurements were performed at room temperature. Moreover, an active transport of dog albumin through canine bronchial epithelium was shown by Johnson et al. (26). Most of the albumin transferred was catabolized. Net P in the L-I direction was 0.049 × 10⁵ cm/s, i.e., approximately one-half that presently found in rabbit mesothelium. No marked fall in albumin transport appeared at low C_alb.

The time taken by vesicles to form and move to the other side of the cell (turnover time) has been determined or indirectly computed in different preparations of capillary endothelium and of pericardial mesothelium. Renkin (41), after having indirectly determined J_liq,ves through the capillary endothelium of dog leg, computed the volume of vesicles from estimates of the volume of the capillary endothelium in his preparation and of the percentage of this volume occupied by the vesicles. Dividing vesicle volume by J_liq,ves, he got a turnover time of ~5 min. This time is similar to that of the vesicular transport of thorium dioxide through the mesothelium of parietal pericardium of rats studied by electron microscopy (28). On the other hand, Milici et al. (34), with immunocytochemical procedures, showed that albumin carried by plasmalemmal vesicles through the capillary endothelium of mouse myocardium appears in the pericapillary space <15 s after the beginning of perfusion, with no gradient from the intercellular junctions. From our data, the turnover time can be obtained following Renkin's approach with only one estimate, because the surface area of the membrane is known in our experiments. The average thickness of the mesothelium in rabbit is ~2 µm (51). Hence, the volume of 1 cm² of mesothelium is 2 × 10⁴ ml. From electron micrographs (28, 29), one can estimate that the vesicles occupy ~15% of this volume. Hence, vesicle volume in 1 cm² of mesothelium is 3 × 10⁵ ml. Because J_liq,ves through 1 cm² of mesothelium is 3.5 µl/h or 9.7 × 10⁴ µl/s, the turnover time should be ~30 s.

To the extent that the data obtained on the specimens of parietal pericardium can be applied to the pleura under physiological conditions (for morphological and cytochemical similarity of the mesothelium of these serosas, see Ref. 49), transcytosis appears to be a relevant mechanism that removes liquid and protein from the pleural space. Therefore, we now consider the other mechanisms that are known to remove liquid and protein from the pleural space to attempt a rough comparison. 1) The volume of liquid absorbed by the Starling forces of the capillaries of the visceral pleura was found in dogs to be 0.5 µl · h¹ · cm² per mmHg of driving pressure (4), but under those experimental conditions the mesothelium was damaged because of air exposure and manipulation (6, 55). This was not a problem for the aim of that research, but it is if one wants to know the liquid flow through the visceral mesothelium. Moreover, the driving pressure across this mesothelium is uncertain and controversial (6, 37). As a rough estimate, the volume of liquid absorbed by the Starling forces through the visceral mesothelium in rabbit could be ~1.5 µl · h¹ · cm². The amount of albumin following this flow by solvent drag should be small, but probably not negligible (3). 2) The lymphatic drainage from the pleural space under physiological conditions has been estimated in dogs from egress kinetics of labeled albumin injected into the space and has been found to be 0.02 ml/(h kg) (35). Taking into account that the surface area of the parietal pleura in dogs is 110 cm² · kg^2/3 (Ref. 36, plus an estimate of the surface area of the costo-phrenic sinus) and that dog weight in the above research was 17.5 kg, the lymphatic drainage from the pleural space should be 0.5 µl · h¹ · cm². Considering that the above measurements should be a little underestimated, as pointed out by the authors (35) and that the turnover of liquid per unit surface area in rabbits appears to be greater than in dogs (37), the lymphatic drainage from the pleural space in rabbits could be ~2 µl · h¹ · cm². Because albumin concentration in the pleural liquid is ~1% or 0.14 µmol/ml (36, 44), this lymphatic drainage should imply an albumin removal from the pleural space of 2.8 × 10⁴ µmol · h¹ · cm². This value, however, should be greater than that corresponding to the lymphatic drainage through the stomata of the parietal pleura because albumin egress from the pleural space (as it has been measured; Ref. 35) includes most of the albumin that has left the pleural space outside the stomata and is then removed from the interstitium by the inherent lymphatics (see Introduction). Therefore, if transcytosis occurs in the mesothelium of the parietal pleura, to get the albumin flux occurring through the lymphatic stomata of the parietal pleura, most of the vesicular albumin flux should be subtracted from albumin egress from the pleural space as it has been measured. Hence, the lymphatic drainage of albumin through the stomata of the parietal pleura (no stomata occur in the visceral pleura; Ref. 51) should be only a fraction of albumin egress from the pleural space, as it has been measured (35). The same consideration applies to liquid flow. Finally, the finding that vesicular albumin flux in the specimens of parietal pericardium, ~5 × 10⁴ µmol · h¹ · cm², is greater than albumin egress from the pleural space, ~2.8 × 10⁴ µmol · h¹ · cm², indicates that the vesicular albumin flux in the parietal pleura should be smaller than that found in our specimens of parietal pericardium. This fits with the morphological evidence of a smaller concentration of vesicles in the pleural (51, 52) than in the pericardial mesothelium (28, 29).

Even if vesicular liquid flow in the pleura under physiological conditions were one-half that found in our specimens of parietal pericardium, i.e., ~1.7 instead ~3.5 µl · h¹ · cm² (see above), the contribution of vesicular transport to the removal of liquid and protein from the pleural space would be substantial, appearing to be greater than that of the other known mechanisms. In particular, the present findings provide experimental support to one argument raised (3, 6) against the view that under physiological conditions, the lymphatic drainage through the stomata of the parietal pleura accounts for 75-80% of liquid removal from the pleural space and that, therefore, it is the main mechanism setting pleural liquid pressure under physiological conditions (35, 37-39). Indeed, if transcytosis occurred in the pleural mesothelium, the lymphatic drainage through the stomata should not be considered the main mechanism setting pleural liquid pressure under physiological conditions. Moreover, even if the vesicular albumin flux through the mesothelium were one-half that found in our specimens of parietal pericardium, i.e., ~2.5 instead of ~5 × 10⁴ µmol · h¹ · cm² (see above), the role of lymphatic drainage through the stomata in preventing protein accumulation into the pleural space (1, 37, 48) should be reevaluated. It might be that under physiological conditions, the peculiar function of the lymphatic stomata of the parietal pleura is that of allowing removal of cells, debris of cells, and bacteria from the pleural space. This function, which implies a simultaneous removal of liquid and protein, appears more consistent with the size of the stomata, which is 2-6 µm or more in diameter (51). The situation changes when the volume and the pressure of the pleural liquid increase, because then the lymphatic drainage from the pleural space may increase >10 times (11).

We have previously provided indirect evidence of a Na⁺-coupled liquid absorption from the pleural space, which appears to be ~0.7 µl · h¹ · cm² (5, 6, 53, 54). This absorption of essentially protein-free liquid increases when C_alb in the pleural liquid decreases below its normal value (5, 6). Conversely, the present findings show that transcytosis decreases when C_alb is relatively low (Figs. 1, 3). Because the former mechanism increases C_alb by removing protein-free liquid from the pleural space, whereas the latter mechanism would decrease C_alb if albumin binding occurred in the vesicles, these two mechanisms could contribute automatically to the constancy of C_alb in the pleural liquid. Indeed, a decrease in C_alb stimulates the former and inhibits the latter and vice versa.