INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

SEVERAL INTRAVITAL MICROSCOPY studies in recent years have demonstrated the importance of leukocyte-endothelial cell adhesion in the development of increased venular permeability during acute inflammation (12-14). Much has been learned from these studies about the inflammatory response in postcapillary venules; however, a large part of the tissue fluid accumulation that accompanies inflammation is likely to derive from the upstream capillaries, which have a greater surface area available for fluid filtration (8). Only recently has intravital microscopy been applied to determine the role of leukocytes during changes in capillary permeability to fluid (4-7).

Recent studies of mesenteric exposure to either platelet-activating factor (PAF) or N ^G-nitro-L-arginine methyl ester (L-NAME) indicate that the mechanism whereby capillary fluid filtration rate (J_V/S, where J_V is the volume filtered flux and S is the surface area of the capillary segment) is increased during these models of acute inflammation involves leukocyte adhesion to endothelial cells (4, 5, 7). The PAF-induced increase in J_V/S is attenuated by about 75-80% with either antineutrophil serum or an antibody against the adhesion molecule P-selectin (5), suggesting a prominent role for neutrophil adhesion. The remaining 20-25% of the PAF-induced increase is likely due to increased hydrostatic pressure (7).

A role for leukocyte-endothelial cell adhesion in capillary permeability is difficult to explain, inasmuch as leukocyte adhesion is a phenomenon rarely observed in capillaries (or arterioles) in mesenteric preparations. (Leukocyte adhesion is essentially confined to the postcapillary segment of the microcirculation.) Proposed explanations (4) of PAF-induced increases in permeability include 1) a direct action of leukocytes on endothelial cells during transit as the leukocytes squeeze through the capillaries (in the absence of apparent leukocyte adhesion); 2) an upstream propagated response from sites of venular leukocyte adhesion, using gap junction communication; or 3) venule-to-arteriole communication, in which one or more mediators produced at the site of venular leukocyte adhesion reaches a nearby paired arteriole (usually arranged in a countercurrent direction), which delivers the mediator downstream to branching capillaries.

Therefore, the aim of this study was to determine which, if any, of these three potential mechanisms is responsible for leukocyte-mediated increases in capillary J_V/S in a model of PAF exposure of the rat mesentery. The first possibility (direct action during transit) was investigated by determining whether adhesion molecules are expressed in rat mesenteric capillaries and by determining whether there was a correlation between the number of leukocytes passing through a selected capillary and the increase in capillary J_V/S. The second possibility (upstream gap junction communication) was investigated by determining whether there was a correlation between the amount of postcapillary leukocyte adherence and the increase in capillary J_V/S and by determining whether inhibition of venule-to-capillary gap junction communication could attenuate the increase in capillary J_V/S. The third possibility (venule-to-arteriole communication) was investigated by determining whether there was a correlation between the extent of venule-to-arteriole pairing and the increase in capillary J_V/S. Pairing was characterized as 1) the distance x (see Fig. 1) along an arteriole from the branch point of a capillary to the point at which a postcapillary venule came within 15 µm of the arteriole, and 2) the percentage of the arteriole length located within 15 µm of a postcapillary venule (100y/(x + y) in Fig. 1).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Depiction of countercurrent arteriovenular pairing, where x is distance from branching capillary to point at which pairing comes within 15 µm, and y is length of arteriole in close pairing with a venule.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Animal preparation. Forty-seven male Sprague-Dawley rats (2-3 mo old) were fasted overnight and then anesthetized with an intraperitoneal injection of 135 mg/kg thiobutabarbital (Inactin; Research Biochemicals, Natick, MA). A thoracotomy was performed to facilitate breathing, and the right carotid artery was cannulated to monitor systemic blood pressure and to administer an overdose of pentobarbital sodium (160 mg/kg) at the conclusion of each experiment. The small intestine was exteriorized through a midline abdominal incision, and the rat was placed on its side on a Plexiglas board so that a selected section of mesentery could be draped over a glass coverslip glued on a hole centered in the board. The exposed intestine, except for the selected mesenteric section under study, was covered with gauze soaked in bicarbonate-buffered saline (BBS) consisting of (in mM) 132 NaCl, 4.7 KCl, 1.2 MgSO₄, 20 NaHCO₃, and 2.0 CaCl₂. After the board was mounted onto the stage of an inverted microscope (Nikon Diaphot, Tokyo, Japan), the mesentery and intestine were kept moist with a 2 ml/min superfusion of BBS bubbled with a 95% N₂-5% CO₂ gas mixture and warmed to 37°C. Rectal temperature was monitored and maintained near 37°C by positioning an infrared heat lamp over the rat.

Video microscopy. The mesentery was observed through a ×40 objective (Nikon Fluor 40, 0.85 numerical aperture, Tokyo, Japan) using a 100-W halogen light source, and bright-field images were captured with a color camera (Sony DXC-107A, Tokyo, Japan). The images were then directed through a time-date generator (Panasonic WJ-810, Tokyo, Japan) into a videocassette recorder (JVC BR-S601MU, Yokohama, Japan) with the live image displayed on a monitor and the taped image used for playback analysis using an image grabber (Imaging Technology Visionplus-AT, Bedford, MA) and image processor (Bioscan Optimas, Edmonds, WA).

Measurement of capillary filtration. Capillary fluid filtration rate was measured using a modified Landis technique (15). As described previously (4-7), a selected capillary was occluded and the filtration rate determined by measuring the decreasing distance between two red blood cells (16). All selected capillaries had diameters <10 µm and were "true" capillaries, i.e., divergent at their upstream end and convergent at their downstream end. Each selected capillary was briefly occluded near its venous end with a glass micropipette drawn to a tip diameter of ~5 µm, or with a metal rod with a rounded tip of similar dimensions (Institute for Micromanufacturing, Louisiana Tech, Ruston, LA). The glass micropipette tip was rounded with a microforge (Stoelting 51512, Wood Dale, IL) to minimize damage to the capillary during occlusion. Positioning of the occluding tip was accomplished with the use of a micromanipulator (Narishige MO-302, Japan) mounted on the microscope. During capillary occlusion, individual red blood cells within the vessel gradually move closer together and toward the occlusion site as the intravascular fluid separating the cells filters across the endothelial barrier into the surrounding tissue. To measure J_V/S, two red blood cells ~80-100 µm apart were selected, and the distance between those cells was monitored for a period of 16 s.

PAF protocol. After a baseline period of ~30 min, a 100 nM PAF solution was substituted for the control superfusing buffer solution. The mesentery was exposed to PAF for 10-15 min before 1) measurements of J_V/S or 2) excision of the tissue for mounting and staining for adhesion molecule expression.

Capillary leukocyte flux. In experiments where capillary flux of leukocytes was measured, the basic PAF protocol (specified above) was performed with the following modification. After the final PAF occlusions were performed, rats were injected intravascularly with 0.2 ml of a 0.1% solution (in saline) of rhodamine 6G (Molecular Probes, Eugene, OR). Rhodamine 6G fluorescently stains the mitochondria of leukocytes, which we observed in the green spectrum (450-490 nm excitation; 510 nm dichroic/520 nm barrier filter). Each capillary was videotaped for a period of only 30 s to minimize exposure of fluorescent light and to have nearly simultaneous measures in each selected capillary. The number of leukocytes passing through the capillary during the 30-s period was counted during playback of the videotape and multiplied by a factor of 2 to determine leukocyte flux per minute. Additionally, leukocyte velocity through the capillaries was determined with an image processor during playback of the videotape.

Adhesion molecule expression. In experiments to determine adhesion molecule expression, the mesentery was superfused for 10-15 min with either BBS or 100 nM PAF in BBS. The rat was then exsanguinated by extracting blood from the carotid artery while injecting an equal volume of BBS through the cannulated jugular vein. (The red blood cells were removed because of their absorbance of fluorescent light.) Three sections of mesentery were excised from each rat for determining 1) P-selectin expression, 2) intercellular adhesion molecule 1 (ICAM-1) expression, and 3) control fluorescence in the absence of a binding antibody.

Venular leukocyte adherence. An adherent leukocyte was defined as one that was instantaneously stationary on the vessel wall during a period of scanning the venule downstream of each capillary in which J_V/S had been measured. The videotaped length of vessel began where the capillary converged with the venule and continued through 500 µm downstream. Venular leukocyte adherence was measured after capillary occlusions during the baseline and PAF superfusion periods of the basic PAF protocol described above.

Localized halothane exposure. To inhibit gap junction communication between a selected capillary and the venule with which it converged, we injected halothane (Ayerst Laboratories, Philadelphia, PA) locally through a micropipette, in a method similar to that described by Frame and Sarelius (3). The solution was prepared by mixing 1 ml liquid halothane into 10 ml BBS; the solution was allowed to equilibrate (to 0.345% halothane) in the BBS solution overnight, stored in the dark at room temperature. Immediately before use, 2 ml of the equilibrated solution were mixed with 8 ml BBS and ~5 µg fluorescein isothiocyanate to form a fluorescent 0.069% solution of halothane. The halothane solution was delivered onto a junction of a capillary and postcapillary venule through a micropipette (with a tip diameter of 8-12 µm), which was attached by tubing to a height of 30 cm above the pipette tip to allow consistent flow from the pipette (3). Existence of flow from the tip onto the capillary-venule junction was verified by observing the fluorescent solution leaving the pipette. The halothane solution was introduced to the junction after baseline capillary occlusions and continued to be injected throughout PAF superfusion. Because halothane is light sensitive, these experiments were performed in the dark.

Venule-to-arteriole pairing. Arterioles and venules are often paired in a parallel countercurrent arrangement. In this study, three characteristics of this pairing were quantified. The first characteristic was the distance from the selected capillary (in which J_V/S was measured) along the feeding arteriole to the point at which the arteriole came within 15 µm of a postcapillary venule (see Fig. 1). The distance of 15 µm was obtained from the results of Zamboni et al. (24), who found that ischemia-reperfusion caused arteriolar constriction but only when the arterioles were paired with a postcapillary venule less than 15 µm away, suggesting that pertinent mediators were capable of diffusing through this distance from venules to arterioles. The second characteristic was the percentage of the arteriole (upstream from the branching capillary) that remained within 15 µm of a paired venule. The length of the arteriole was measured as far as possible until it reached the fat cells surrounding the major arcade vessels. The third characteristic was a combination of the first two: the percentage of venule-to-arteriole pairing length was divided by the distance from the selected capillary to the initial point at which a venule-to-arteriole pairing occurred. This will be referred to as the pairing quotient.

Statistics. Two sets of data were compared using standard t-tests, and comparisons between more than two sets were made with Bonferroni's post hoc test. Linear regression was performed to determine whether relationships existed between various parameters. Each test was performed with Instat software (GraphPad Software, San Diego, CA) using a 95% confidence level to determine significant differences. Data are presented as means ± SE. When more than one capillary was studied per rat, values were averaged before means were compared. Throughout, n represents the number of animals.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Baseline capillary J_V/S (n = 22 rats; including 36 capillaries) averaged 0.027 ± 0.002 µm/s in all PAF groups excluding halothane exposure and adhesion molecule expression. Capillaries with baseline values >0.040 µm/s were excluded from this average and from further study with PAF, because of the possibility that they were already undergoing an inflammatory response. In the same rats, J_V/S increased to 0.052 ± 0.008 µm/s following 10-15 min of PAF superfusion (P = 0.003 compared with baseline). In time-matched control experiments, J_V/S remained essentially constant (0.018 ± 0.003 to 0.019 ± 0.006 µm/s; P = 0.76; n = 7 rats; including 8 capillaries). As reported previously (5, 7), the PAF-induced increase cannot be explained by changes in hydrostatic or osmotic pressure and is therefore considered to be a result of increased vascular permeability. PAF induces an ~7-mmHg increase in arteriolar pressure, which in itself may increase J_V/S by ~25% (7). Plasma protein concentrations drop by ~0.2 g/dl during the protocol (5), which would only add ~1 mmHg to the pressure gradient promoting filtration.

To investigate the possibility that leukocyte-endothelial cell adhesion within the capillaries is the cause of PAF-induced increases in capillary filtration, we excised sections of mesentery from 12 rats and stained the sections for expression of the adhesion molecules P-selectin and ICAM-1. Analysis of confocal images revealed that antibodies against both ICAM-1 and P-selectin were localized to the venular and capillary endothelium (data not shown). The staining present in the capillaries did not extend into the arteriolar endothelium and was significantly higher than in control samples using a nonbinding antibody (see Fig. 2). The tendency for adhesion molecule expression to be enhanced with PAF compared with control BBS superfusion was not statistically significant.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Expression of P-selectin and intercellular adhesion molecule-1 (ICAM-1) in capillaries of rat mesentery. Values are given as gray value fluorescence in arbitrary units. Two bars at far left give results of including fluorescent dye without antibody to either adhesion molecule. Data are presented as means ± SE (n = 6 in each group). PAF, platelet-activating factor. *P < 0.05, **P < 0.01, ***P < 0.001 compared with no antibody.

Existence of adhesion molecules on capillary endothelium is consistent with a hypothesis that leukocyte-endothelial cell adhesion within the capillaries could be involved in increased capillary filtration. To further test this hypothesis, leukocyte flux through individual capillaries was quantified to see whether there was a correlation between flux and the increase in capillary J_V/S. If this hypothesis is correct, then it would be expected that capillaries with few leukocytes passing through would be minimally affected, whereas capillaries with a high leukocyte flux would be affected most. After baseline and PAF capillary occlusions (13 capillaries in 5 rats), circulating leukocytes were fluorescently labeled with rhodamine 6G, and the flux of labeled cells was counted for each capillary. Leukocyte flux varied from 2 to 36 cells/min; however, as shown in Fig. 3, no correlation (P = 0.35) existed between flux and the relative increase in capillary J_V/S (PAF J_V/S divided by baseline J_V/S). We also tested the hypothesis that the time spent by the leukocytes within these capillary segments (assessed by their velocity) could determine the PAF-induced increase in J_V/S. For example, a slowly moving leukocyte could potentially influence the capillary endothelium to a greater extent than a faster one. However, there was also no correlation (P = 0.69) between average leukocyte velocity through the capillaries vs. increased J_V/S (data not shown). We also performed two additional experiments (observing 3 capillaries) to see whether leukocyte flux at 15 min of PAF superfusion (when flux was measured in the preceding experiments) represents the levels present immediately after PAF exposure. The capillary with the highest flux of the three after 15 min of superfusion (22 leukocytes/min) also had the highest average flux during the 0- to 5-min period (18.8 leukocytes/min). The capillary with the lowest flux after 15 min of superfusion (2 leukocytes/min) also had the lowest average flux during the 0- to 5-min period (0.4 leukocytes/min). Therefore, our values of flux (obtained after the PAF occlusions) appear to represent typical values throughout 15 min of PAF superfusion.

View larger version (11K): [in this window] [in a new window]   Fig. 3.   PAF-induced increase in capillary fluid filtration rate (JV/S) (calculated as PAF JV/S divided by baseline JV/S) as a function of leukocyte flux through each of 13 capillaries. JV, volume filtered flux; S, surface area of capillary segment.

If PAF-induced increases in capillary J_V/S are not due to leukocyte-endothelial cell adhesion within the capillaries, then it is likely that leukocyte adhesion downstream in the postcapillary venules is involved in the capillary response. One hypothesis explaining this mechanism is that a cellular response could be initiated from sites of venular leukocyte adhesion, with a permeability-increasing signal propagated upstream to capillary endothelial cells using gap junction communication. If this hypothesis is true, then it would be expected that if a given capillary converged into a postcapillary venule that contained a large amount of leukocyte adhesion, the venule would send a strong signal to the capillary. Alternatively, if very few leukocytes are sticking to the venule immediately downstream, a very weak signal would be expected to reach the capillary. The results shown in Fig. 4A argue against such a scenario, with no correlation between the amount of PAF-induced postcapillary leukocyte adherence (through a distance of 500 µm from the capillary-venule junction) and the relative increase in capillary J_V/S. However, it must be considered that the values of adherence include all subsets of leukocytes; the PAF-induced increase in J_V/S may be specifically dependent on neutrophils (5). Leukocyte adherence increased from an average of 14.7 ± 5.3 to 37.9 ± 11.4 cells per 500 µm (P = 0.008; n = 9). The sections with the most adherence after PAF also had the most during the baseline period (r² = 0.90; P < 0.0001).

View larger version (12K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Fig. 4.   A: PAF-induced increase in JV/S as a function of leukocyte adherence through a distance of 500 µm downstream of the selected capillaries. , Experiments with PAF plus exposure to halothane at the capillary-venule junction; , experiments with PAF alone. B: PAF-induced increase in JV/S in experiments with PAF alone (n = 6) or with added exposure of halothane (n = 4) at capillary-venule junction. Data are means ± SE. *P < 0.05 compared with baseline.

The possibility of gap junction communication being responsible for increased capillary J_V/S was also tested by using a localized injection of 0.069% halothane at the capillary-venule junction. Halothane inhibits gap junction communication (3), therefore, if a signal is initiated by venular leukocyte adhesion, blocking the signal to prevent its arrival at the capillary would be expected to attenuate the PAF-induced increase in capillary J_V/S. However, no such attenuation (compared with PAF superfusion alone) was noted in these experiments, as demonstrated in Fig. 4B.

Another way in which postcapillary leukocyte adherence could increase capillary J_V/S is through venule-to-arteriole communication, in which a mediator produced at the site of venular leukocyte adhesion reaches a nearby paired arteriole (usually arranged in a countercurrent direction), which delivers the mediator downstream to branching capillaries. If the mediator has a short half-life, or is metabolized rapidly, or diffuses away once it enters the arteriolar circulation, then the distance from the mediator's entry into the arteriole to its delivery to the capillary could determine whether it has an effect on capillary permeability. Figure 5A demonstrates a tendency toward a correlation between 1) the distance from the capillary-arteriole junction to a venule-arteriole pairing (defined as <15 µm separation between the arteriole and venule); and 2) the relative increase in capillary J_V/S caused by PAF superfusion. Additionally, if venule-to-arteriole communication is the mechanism of increased capillary J_V/S, then it would be expected that arterioles with a high percentage of their length paired within a close distance (15 µm) of a postcapillary venule would deliver a greater concentration of the permeability-increasing mediator to the capillaries than those arterioles with a small percentage pairing. In support of this hypothesis, Fig. 5B demonstrates a significant correlation (P < 0.05) between the percent paired length of arteriole and the relative increase in capillary J_V/S. One possible reason for the amount of scatter present in Fig. 5, A and B, is that some arterioles with a high percentage pairing (which would promote an increase in J_V/S) may also have a long distance to initial pairing (which would not promote an increase in J_V/S), and vice versa. If both the distance to the initial venule pairing and the percentage pairing length determine the capillary response, then a combination of the two factors should be the most accurate predictor of which capillaries will respond the most to PAF. For example, a capillary whose feeding arteriole is a short distance from a venule pairing and has a high percentage of its length paired with a venule will have a high pairing quotient [(%pairing)/(distance to initial pair)] and would be expected to respond vigorously to PAF. Alternatively, if either the percent pairing is small or the distance to the initial pair is large, then the pairing quotient will be smaller and the capillary would not be expected to respond as much. This relationship is demonstrated in Fig. 5C, in which a correlation of the pairing quotient with the relative increase in capillary J_V/S is better than with either individual pairing parameter alone (P < 0.0001). Each of these three correlations are dependent on two outlying data points (the 2 with the highest increase in J_V/S). Although the correlation in Fig. 5B is no longer statistically significant in their absence, the correlation in Fig. 5A actually improves (from P = 0.06 to 0.02) without the variation introduced by the two points. However, even if neither of the first two correlations were statistically significant, the more important measure of pairing takes both factors into consideration, as presented in Fig. 5C. This correlation gives the best relationship between pairing and the increase in J_V/S and remains statistically significant (P < 0.05) even without the two outlying points. The two parameters of the pairing quotient were not found to be dependent on each other (r² = 0.11;P = 0.20); if this were not the case, the correlation of Fig. 5C would be overestimated.

View larger version (17K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Fig. 5.   PAF-induced increase in JV/S as a function of distance from arteriole-capillary junction to arteriole-venule (a-v) pairing (A); percentage of arteriole length paired with a postcapillary venule (B); and pairing quotient (C), which divides the abscissa of B by abscissa of A.

The analysis described in the preceding paragraph included arteriolar pairing with venules of any size. During an acute inflammatory response, less leukocyte adhesion is observed to occur in mesenteric venules <15 µm in diameter, possibly because of higher shear rates and/or lower expression of endothelial adhesion molecules. As the diameter of a venule exceeds 15 µm, leukocyte rolling increases, and as the diameter exceeds 20 µm, leukocyte rolling is frequently accompanied by firm adherence and subsequent emigration of the leukocytes into the surrounding tissue (see Fig. 6). Figure 7 demonstrates that the correlation between the pairing quotient and increased J_V/S depends on the size of venules included in the analysis. With each size of venule included, the correlation coefficient r² is 0.73, with a corresponding P value of <0.0001. The correlation drops from statistical significance (P = 0.054) when venules <25 µm in diameter are excluded from the analysis, indicating that venules <25 µm in diameter play a critical role in arteriolar communication. However, the fact that the correlation does not continue to improve by including venules with diameters <20, 15, and 10 µm indicates that the smallest venules may not release as many permeability-increasing mediators as those with diameters >20 µm. Of the 60 venules included in this analysis, 54 were >10 µm in diameter; 50 were >15 µm; 39 were >20 µm; 31 were >25 µm; and 26 were >30 µm in diameter.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Depiction of arteriole-venule pairing in mesenteric microcirculation. Mediator transfer from venule to arteriole may be highest when initiated from the 25- to 35-µm-diameter venules, which have a higher density of adherent leukocytes [white blood cells (WBC)].

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Square of correlation coefficient (r2) derived from relationship between pairing quotient and relative increase in PAF-mediated JV/S. Correlation is statistically significant (P < 0.0001) when including venules in diameter ranges indicated with open bars but not significant (0.05 < P < 0.10) when excluding venules with diameters <25 µm (hatched bars).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This study is the first (of which the authors are aware) to show that venule-to-arteriole communication may be responsible for increased capillary permeability. The rationale for the study originates from observations that edema is frequently a result of increased capillary fluid filtration and that the capillary response is dependent on leukocytes (4-7). In several organs, such as the lung, heart, and liver, leukocytes are occasionally observed to plug the capillaries for a prolonged period. In these organs, it is tempting to propose that leukocytes damage the capillary endothelium during their entrapment. However, in many other tissues, leukocyte adhesion within the capillaries is rarely observed; leukocyte rolling and firm adherence is essentially restricted to the postcapillary venules. In these tissues, it is more difficult to visualize how the leukocytes could inflict damage during transit, as prolonged contact within the capillaries is prevented by high shear rates and lower expression of endothelial adhesion molecules (10, 18). The present study provides some support for a mechanism whereby leukocyte adhesion modulates capillary permeability through venule-to-arteriole communication.

The phenomenon of venule-to-arteriole communication has also been found to be responsible for delivering vasoactive signals from venules to arterioles. For example, studies by Hester (9), Tigno et al. (23), and Falcone and Bohlen (2) have suggested that vasoactive metabolites can diffuse from postcapillary venules into paired arterioles to alter their resting diameters. Additionally, a recent study by Zamboni et al. (24) has shown the importance of such communication in a model of acute inflammation. In their study, ischemia-reperfusion in skeletal muscle caused arteriolar constriction but only when the arterioles were paired with a postcapillary venule <15 µm away (the separation criterion used in our study).

A more direct hypothesis (4) concerning how leukocytes increase capillary permeability postulates that it occurs by the release of mediators during transit through the capillaries. Ley et al. (17) have shown that there are preferred pathways for leukocytes through the microcirculation, which result in more leukocytes traversing distal rather than proximal capillaries. We used this phenomenon to see whether the number of leukocytes passing through a selected capillary (leukocyte flux) correlated with the increase in fluid filtration induced by PAF. Although adhesion molecule expression is typically observed to be sparse or even nonexistent in capillaries, we did show significant expression in capillaries of the rat mesentery. Therefore, if leukocytes interact, however transiently, with adhesion molecules as they squeeze through the capillaries' small diameters, this potentially provides a connection between leukocyte adhesion and increased capillary permeability. However, our results demonstrated an absence of a correlation between leukocyte flux and increased J_V/S. It is likely that the interaction between leukocytes and endothelial cells within the capillaries is too brief to initiate leukocyte release of permeability-increasing agents.

The second hypothesis tested in this study was that a signal could be propagated through gap junction communication from sites of postcapillary leukocyte adhesion upstream to the capillaries. Although no evidence was obtained to support this hypothesis in our study, a few items need to be addressed concerning our experiments. First, in our attempts to correlate downstream leukocyte adherence to the increase in J_V/S, it was not always possible to follow the venule for more than 500 µm, even though gap junction communication can be propagated for at least 2,000 µm (19, 20). Second, we quantified only the firm adherence of leukocytes because quantification of leukocyte rolling at any one cross section, much less through a distance of hundreds of micrometers, is highly time consuming. Therefore, we are not able to make any conclusive statements concerning a possible correlation between postcapillary leukocyte rolling and the increase in capillary J_V/S. Third, in our halothane experiments, we did not investigate whether gap junction communication may coexist with venule-to-arteriole communication, propagating a signal from the site of venule-to-arteriole pairing downstream along the arteriolar endothelium, reaching the capillary. This possibility could be tested by inhibiting gap junction communication at the junction of arterioles and capillaries. However, our negative results with halothane and the lack of a relationship between venular leukocyte adherence and increased capillary J_V/S greatly reduce the possibility that direct venule-to-capillary gap junction communication plays a significant role in this model.

Analysis of the size of venules paired with the arterioles indicates that although venules of all diameters may contribute to arteriolar communication, those <20 µm in diameter may not contribute as much as the larger venules. This is probably due to the fact that vessels <20 µm in diameter have less frequent leukocyte adhesion because of higher shear rates and lower expression of adhesion molecules. Kimura et al. (11) have presented evidence that PAF-induced leukocyte adherence is greatest in venules ranging from 25 to 35 µm in diameter. Therefore, these vessels might be expected to release the highest concentration of mediators that communicate with the paired arterioles. Indeed, our analysis demonstrated the importance of including venules 20 µm in diameter.

Results from the venule-to-arteriole pairing analysis raise numerous interesting questions. What mediators are produced at the site of leukocyte-endothelial cell adhesion in venules that reach arterioles? Does the arteriolar endothelium in turn release a secondary mediator, which then is delivered to the capillaries? Is there a possibility that interstitial mast cells play an intermediate role in the communication between the venules and arterioles? Are the mediators released from inside the venule, or from the venular endothelium, or is the mediator released from leukocytes that have emigrated out of the venules? If the permeability-increasing agents can be delivered by arterioles as far as the capillaries, are they also delivered further downstream to increase postcapillary permeability?