INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The occurrence of episodes of systemic hypoxia is a common component of many pathological episodes in intensive care medicine. Often, systemic hypoxia can be associated with a local or generalized inflammatory process, but the relationships between these two pathological manifestations is still under discussion. Indeed, the links between systemic hypoxia and inflammation are important to study, because this could lead to a better evaluation of specific treatments of hypoxia (such as NO inhalation or postural maneuvers) in limiting the generalization of the inflammatory process.

Regarding the possible role of hypoxia in the genesis of tissue injury mediated by leukocyte-dependent inflammatory processes, several mechanisms have been shown in in vitro models.

In cultured endothelial cells, hypoxia (even in the absence of anoxia or reperfusion) can induce activation of xanthine oxidase (1), alteration of eicosanoid synthesis (2), increase in transcription of several genes including those coding for platelet-derived growth factor or endothelin (3). These stimuli could be responsible for an increase in leukocyte-endothelial cell interaction. In addition, direct triggering of endothelial cell adhesion mechanisms has also been shown. For instance, hypoxia has been generally shown to increase the avidity of the endothelial-leukocyte adhesion by increasing expression of several endothelial cell selectins such as endothelial leukocyte adhesion molecule-1 (ELAM-1) (4, 5) and intercellular adhesion molecule-1 (ICAM-1) (6) or lymphocyte function-associated antigen (LFA)-1 endothelial cell adhesion ligands (6). Note however that a decrease in the adhesiveness of human umbilical vein endothelial cells (HUVEC) for granulocyte, when submitted to hypoxia, has also been reported (7).

The possibility that hypoxia can directly affect leukocytes has been evoked. Hypoxia has been shown to alter the surface ultrastructure and morphological characteristics of leukocytes (8) but some controversy exists as to the effect of hypoxia on leukocyte adhesion molecules. Indeed, Scannell and coworkers (9) have found that hypoxia increased the expression of CD11b/CD18 in leukocytes exposed for 2 h to hypoxia, but others have reported the absence of effect after 1 h hypoxia exposure (10). It should be noted that in most in vitro studies the levels of O₂ tension considered are much lower than those usually observed in clinical situations of hypoxia.

With regard to in vivo experiments, it should be noted thatin contrast to the numerous in vivo studies showing the increase of leukocyte-endothelial cell adhesion during episodes of ischemia-reperfusion (11)very few data are available concerning the effects of systemic hypoxia per se in the absence of anoxia or stop-flow phenomenon.

In this context, since leukocyte-venule interaction is a major initial step in the inflammatory process, it seemed to us important to explore in vivo the possible consequences of systemic hypoxia on this parameter. Thus, the present study was carried out to address the following questions: (1) Can systemic hypoxia, at levels found in clinical situations, induce leukocyte activation in peripheral tissues? (2) What are the mechanisms involved in this process and in particular the role of local hypoxia and local hemodynamic changes in the targeted peripheral tissue?

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Male Sprague-Dawley rats weighing 223 ± 4 g (mean ± SE) were anesthetized by intraperitoneal injection of 50 mg/kg sodium penthiobarbital. A patent airway was maintained with a tracheotomy tube. The carotid artery was cannulated for measurement of systemic mean arterial blood pressure with a Uniflow 33600 (Baxter, Maurepas, France) transducer.

Preparation of the Cremaster

After anesthesia, the right cremaster muscle was surgically prepared for in vivo visualization, by a new technique described in detail elsewhere (12). Briefly, the muscle was detached from the scrotum but was not longitudinally incised as in the original procedure. A transverse buttonhole slit about 5 mm long was made in the proximal part of the cremaster pouch. The testicle, epididymis, and the cremaster itself were then drawn out through the buttonhole. This procedure led to the invagination of the cremaster, which acquired a finger shape, with the cremaster pouch now turned inside out. The small pedicle that attaches the cremaster to the testicle was tied up with two stitches and cut between them, to separate the cremaster completely from the testicle, which was reincorporated into the abdominal cavity. To prepare the cremaster muscle for transillumination microscopy, a flexible extendible ovoid ring was made with metal wire (diameter, 0.1 mm) covered with Silastic rubber. The ring was then secured in its axis with a clamp and introduced longitudinally into the cremaster pouch. When the clamp was removed the ring expanded gently, spreading out the cremaster that acquired a racket shape. The ring had been positioned so that the main cremaster artery was in the center of the racket's upper surface. Throughout these procedures, the muscle was continuously bathed with warm saline solution (Figure 1).

View larger version (21K): [in this window] [in a new window]   Figure 1.   Schematic representation of the surgical preparation of cremaster muscle (A) and experimental set up (B). The rat was placed in supine position on thermostated blanket and breathed either normal or low PO2 gas. After surgical preparation, the cremaster muscle was placed in the microcirculatory chamber and covered with a plexiglass plate. During the experiment, the muscle was continuously superfused with warm Krebs solution. This superfused solution bubbled with different gas mixtures in the different groups, thus allowing one to change the local oxygenation of the muscle (see Table 1).

This new preparation procedure involves minimal incision of the cremaster and so reduces considerably the risk of hemorrhage and of lesions to the muscle and its microcirculation. Because the size of the ring is adapted to the dimensions of the cremaster, the extension of the muscle is sufficient to allow good optical resolution, but does not affect the microcirculation.

The muscle was continuously superfused at 2 ml · min¹ with a modified Krebs-Henseleit solution containing, in millimolar per liter: 118 NaCl, 5.9 KCl, 2.5 CaCl₂ · 2 H₂O, 0.5 MgSO₄ · 7 H₂O, 28 NaHCO₃, and 10 glucose at a temperature of 34.5° C in the muscle chamber. In all groups studied except one, the superfusate solution bubbled with a 6% CO₂-94% N₂ gas mixture fixed the pH, PO₂, and PCO₂ of this solution in the muscle chamber at 7.40 ± 0.02, 22 ± 2, and 40 ± 0.5 mm Hg, respectively. The chamber was covered with a plexiglass plate to isolate it from the atmosphere.

To visualize the microcirculation, the chamber was placed on the movable stage of a modified Leitz microscope and the cremaster muscle was transilluminated using a 100-W tungsten-halogen lamp. The image, magnified by a ×20 objective and ×10 oculars, was projected into a charge-coupled device (CCD) camera (Sony) connected to a professional videotape recorder (Sony VO 9600 P).

Parameters Measured

Systemic parameters. Blood pressure was recorded continuously, and blood gases were measured at the end of the stabilization period and at the end of the observation period.

Microvascular parameters. Two randomly selected venules were examined in each experiment. Venular orders were identified by their relative locations in the network according to morphological characteristics. Venular diameters were measured by playback analysis of the video record and a distance measurement device (Model 303 dimension analyzer; IPM, San Diego, CA). Red cell velocity (V_RBC) was measured using a photodiode system (IPM, San Diego, CA) connected to a cross-correlator (Model 102B; IPM). Venular mean blood flow velocity (V_mean) was estimated as V_RBC/1.6 and venular wall shear rate () was estimated using Poiseuille's approximation  = 8 · V_mean/D where D was the venule diameter.

Leukocyte rolling velocity was determined from the time required for a leukocyte to traverse a given distance along the length of a venule (100 µm). The number of adherent leukocytes was determined off-line during playback analysis of the videotape by counting the number of firmly adherent leukocytes in 100 µm of venule length. As proposed by House and Lipowsky (13), we considered leucocytes adhering more than 10 s as firmly adherent.

In the following paragraphs, N refers to the number of rats used and n to the number of venules studied.

Experimental Protocols

After 30-min stabilization period, terminal venules (i.e., fourth or third order) were randomly selected. Basal venular diameter was studied, then rats were allocated to several groups (Table 1):

Group 1 (control n = 12). The rats breathed spontaneously a 21% O₂-5% CO₂ gas mixture and the rat cremaster muscle was exposed topically to Krebs at a low O₂ tension (ranging between 15 and 20 mm Hg).

In Group 2 (n = 6) systemic hypoxia was induced by making the rats breathe a gas mixture at 10% O₂-90% N₂. In this group too, the rat cremaster muscle was exposed topically to Krebs at a low O₂ tension (ranging between 15 and 20 mm Hg).

In Group 3 (n = 9) systemic hypoxia was induced similarly to Group 2, but the rat cremaster muscle was exposed topically to Krebs with a high PO₂ value ranging between 85 and 100 mm Hg. Consequently in this group, despite the low value of blood PO₂, tissue oxygenation was maintained at a high level by the superfusion bath (14).

In order to explore the involvement of the CD11/CD18 complex in the effect of systemic hypoxia, rats in Group 4 (n = 5), were pretreated by bolus intravenous administration of 1 mg/kg anti-rat LFA-1 monoclonal antibodies at the beginning of the stabilization period. Then systemic hypoxia was induced as in Group 2. In Group 5 (n = 6) rats were administered pentoxifylline (20 mg/kg by bolus intravenously + 20 mg/kg · h during the experiment [15]), a drug known to inhibit leukocyte activation in in vitro (16) and in vivo experimental models (17, 18). Then, they were submitted to systemic hypoxia + low superfusion bath PO₂ as in Group 2.

Because it is known that leukocyte adhesion or rolling can be affected by changes in shear rate and because in our experiments (as in those previously reported by others), hypoxia affected the systemic as well as microvascular hemodynamic, we carried out an additional series of experiments to test the hypothesis that the changes in hemodynamic associated with systemic hypoxia could be sufficient per se to induce the effects observed on leukocytes. Consequently, in Group 6 (n = 8), we studied rats under normoxic conditions but for which a partial occlusion of the spermatic artery had been made in order to decrease venular blood flow by the same amplitude as observed during systemic hypoxia and during the same period of time (1 h). The total duration of the experiment was similar in all groups studied.

Immunohistochemistry

In a complementary study, nine rats were used for immunohistochemistry experiment. In these experiments rats were submitted to the same protocols as in Groups G1, G2, and G3 (i.e., control, systemic hypoxia with low or high PO₂). At the end of the protocol, the cremaster muscle was fixed in buffered formalin solution (4% formaldehyde), embedded in paraffin and cut. The tissue sections were subsequently deparaffinized (toluene and 100% ethanol), washed using buffered solution (tris-buffered saline [TBS]; pH 7.6). The localization of immunoreactive P-selectin was achieved using the avidin-biotin- peroxidase technique. Briefly, endogenous peroxidase activity was eliminated by incubation with 3% H₂O₂ for 5 min. The tissue sections were then passed in a microwave oven in a citric-acid/sodium citrate solution for 3 × 5 min. The sections were washed in TBS solution and incubated for 30 min with diluted normal blocking serum. The slides were then incubated with the primary antibody (anti-P-selectin) for 1 h at room temperature at dilutions of 1:100, 1:200, and 1:400. Of these dilutions, 1:100 dilution gave the highest degree of immunolocalization with the least amount of nonspecific background staining. The slides were subsequently processed as described in the Vectastain ABC kit (VECTOR Laboratory, Burlingame, CA) and counterstained with Harris's hematoxylin.

Control preparations were created by omission of the primary antibody. The positive staining for P-selectin was determined by detection of the brown peroxidase reaction product. The number of positive immunostained vessels was counted for each slide. The measurements were performed only on the vessels that were cut in cross section. Thirty-five venules per tissue section were examined and the percentage of positive staining venules was determined for each slide. Each tissue section was analyzed blindly.

Drugs

Anti-rat LFA-1 monoclonal antibody from R&D Systems (Abingdon, UK) recognizes the 2 leukocyte integrin chain. Therefore the antibody recognizes LFA-1, Mac-1, and P150,95. Pentoxifylline was provided by Dr. M. Lenoble from Laboratory Heochst-France. For immunochemistry experiments, we used a purified rabbit anti-human CD62P polyclonal antibody which reacts with rat P-selectin (PharMingen, San Diego, CA).

Statistical Analysis

For all parameters, possible difference among groups at baseline was tested by one-factor analysis of variance (ANOVA). The changes between values at baseline and those measured after 1 h in the different experimental conditions were compared by two-way ANOVA. Because treatments were studied in parallel groups and because in each group measurements were repeated in the same rats (for systemic parameters) or in the same venules (for microvascular parameters) at baseline and after 1 h, the two-way ANOVA model was built with one between factor (i.e., factor "group") and one within factor (i.e., factor "time": repeated measurements at baseline and 1 h made in the same rat or in the same venule). When interaction between the two factors was significant, thus showing that the time-dependent changes were different between groups, we compared all groups at 1 h by one-way ANOVA. Only when this global analysis was significant, post hoc tests (Duncan's test) were made. This last analysis was also used to compare the mean percentages of positively stained venules in immunohistochemistry experiments. For all tests significance level was fixed at 5%. All results are reported as mean ± SEM.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Systemic Parameters

No significant difference was found between mean blood pressures in the different groups at baseline values. After 1 h mean blood pressure was found significantly lower (16% in average, p < 0.001) in groups submitted to hypoxia than in those not submitted to hypoxia. No significant difference was found among the different groups submitted to the same level of O₂ in the inspired gas.

At baseline, arterial PO₂ was not significantly different between the different groups (81 ± 2 mm Hg). After 1 h, arterial PO₂ was stable when rats were maintained in normoxia (86 ± 4 mm Hg) and significantly decreased in hypoxia groups (global mean: 51 ± 3 mm Hg). No significant difference was found among arterial PO₂ in the four groups submitted to hypoxia. In all groups PCO₂ was lower at 1 h than at baseline (43 ± 1 mm Hg versus 36 ± 1 mm Hg). No difference was found between groups with regard to baseline values or variations in PCO₂ (interaction between factor "time" and factor "group" not significant [NS]).

Microvascular Parameters

Venular diameter. The mean diameter of the venules studied was 24 ± 1 µm and no significant difference between groups was found for this parameter. No significant variation with time was found (mean value was 25 ± 1 µm at 1 h).

Shear rate. At baseline no significant difference was found between shear rates in the different groups. A significant decrease in shear rate was found (p < 0.05) (i.e., even 20% in average for control group). A trend to a larger decrease in shear rate was observed in the groups submitted to hypoxia (G2-G5) than in the control group (G1) but the difference did not reach the significance level (interaction between factor "time" and factor "group" NS). Note that in the group submitted to partial occlusion of spermatic artery (G6) the decrease in rate was similar to that in the groups submitted to hypoxia (Figure 2).

View larger version (39K): [in this window] [in a new window]   Figure 2.   Shear rate (mean ± SEM) in the different groups studied. At baseline, no significant difference was found; the white column corresponds to the pooled values of the six groups studied. After 1 h, a significant decrease in shear rate was found (p < 0.05 factor "time" in the two-way ANOVA), but this decrease was similar for all the conditions tested (interaction between factors of the two-way ANOVA; NS). Despite a trend to a larger decrease in shear rate in the groups submitted to systemic hypoxia, the difference did not reach the significance level. G1: normoxia; G2: systemic hypoxia + low O2 tension in the superfusion bath; G3: systemic hypoxia + high O2 tension in the superfusion bath; G4: systemic hypoxia + anti-CD18 antibody; G5: systemic hypoxia + pentoxifylline; G6: normoxia + arterial occlusion (see METHODS).

Leukocyte rolling velocity. At baseline, no significant difference was found between the various groups and mean leukocytes rolling velocity was 63 ± 1 µm/s. A significant interaction was found between factor "time" and factor "group" (p < 0.001, two-way ANOVA) thus showing that the evolution between baseline and 1 h was significantly different between groups.

Indeed at 1 h, as shown by Figure 3, leukocyte rolling velocity was found significantly different between groups (p < 0.001, one-way ANOVA). Post hoc comparisons showed that leukocyte rolling velocity was significantly lower (p < 0.05) in groups submitted to hypoxia with low PO₂ in the superfusion bath (whether or not associated with anti-CD11/CD18i.e., G2 and G4, respectively) than in groups without hypoxia (G1 and G6), or with hypoxia associated with high PO₂ in the superfusion bath or pentoxifylline (G3 and G5). No significant difference was found among these last four.

View larger version (29K): [in this window] [in a new window]   Figure 3.   Leukocyte rolling velocity (mean ± SEM) in the different groups studied. At baseline, no significant difference was found; the white column corresponds to the pooled values of the six groups studied. After 1 h, the changes in the different groups were significantly different (p < 0.001, interaction between factors of the two-way ANOVA). After 1 h, the leukocyte rolling velocity was found significantly lower in the groups submitted to systemic hypoxia with low O2 tension in the superfusion bath, whether or not associated with anti-CD18 antibody (p < 0.05 for Duncan test for post hoc comparisons) than for the four others which did not differ significantly. Groups defined as in Figure 2 (see METHODS).

Leukocyte adhesion. At baseline no significant difference was found between the different groups and mean number of long-lasting adherent leukocytes per mm² was 462 ± 100. A significant interaction was found between factor "time" and factor "group" thus showing that the evolution between baseline and 1 h was significantly different between groups.

Indeed at 1 h, as shown by Figure 4, leukocyte adhesion was found significantly different between groups (p < 0.001 one-way ANOVA). Post hoc comparisons showed that leukocyte adhesion was significantly higher (p < 0.05) in groups submitted to hypoxia with low PO₂ or high PO₂ in the superfusion bath (G2 and G3) than in groups without hypoxia (G1 and G6) or with hypoxia association with anti-CD11/CD18 or pentoxifylline (G4 and G5). No significant difference was found among these four latter.

View larger version (25K): [in this window] [in a new window]   Figure 4.   Number of adherent leukocytes (mean ± SEM) in the different groups studied. At baseline, no significant difference was found; the white column corresponds to the pooled values of the six groups studied. After 1 h, the changes in the different groups were significantly different (p < 0.001, interaction between factors of the two-way ANOVA). After 1 h the number of adherent leukocytes was significantly larger for the groups submitted to systemic hypoxia with low or high O2 tension in the superfusion bath (p < 0.05 for Duncan test for post hoc comparisons) than for the four others which did not differ significantly. Groups defined as in Figure 2 (see METHODS).

Immunohistochemistry Parameters

The mean percentage of positively stained venules was significantly different among the three groups (p < 0.001; Figure 5). After 1 h very few venules were stained positively for P-selectin in normoxic rats (G1) (3.9 ± 0.8%), whereas significant numbers of stained venules were observed in the hypoxic with low O₂ tension in the superfusion bath group (G2) (26.1 ± 8.3%; p < 0.005). In hypoxic rats with high O₂ tension in the superfusion bath group (G3) the percentage of positively staining venules was equal to 3.4 ± 1.9%, a value significantly lower than that found in G2 (p < 0.005) and not statistically different from that found in G1.

View larger version (11K): [in this window] [in a new window]   Figure 5.   Percentage of positively stained venules for P-selectin antibody during normoxia (G1) or systemic hypoxia with low O2 tension in the superfusion bath (G2) or high O2 tension in the superfusion bath (G3). Values are means ± SEM. Numbers in bars = no. of slides studied. After 1 h, the mean percentage was significantly different between the three groups (p < 0.005 for ANOVA). The mean percentage of positively stained venules was significantly larger for the group submitted to systemic hypoxia with low O2 tension in the superfusion bath (p < 0.05 for Duncan test for post hoc comparisons) than for the two others which did not differ significantly.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Leukocyte adhesion to microvascular endothelium is a major initial step in the pathogenesis of inflammatory disorders. Regarding this phenomenon, findings from local ischemia or local ischemia-reperfusion studies are commonly extrapolated to propose pathophysiological mechanisms related to systemic hypoxia which can be retrieved during intensive care in patients. However, these two situations clearly differ in several important aspects, such as for instance the local or systemic nature of the disturbance in oxygenation, the degree of blood or tissue hypoxia, and the presence or absence of a stop-flow phenomenon. This is why the present study was carried out to evaluate the specific effects on leukocyte-endothelium interactions in peripheral muscles of level of hypoxia corresponding to that found in clinical situations. We found that: (1) systemic moderate hypoxia resulting from purely respiratory disturbance even in the absence of local stop-flow phenomenon or circulatory shock can induce an increase in leukocyte adhesion and a decrease in leukocyte rolling velocity in the microcirculation of peripheral tissues; (2) to be present, this increase in leukocyte adhesion does not require tissue hypoxia of the peripheral tissue but the effect of systemic hypoxia on rolling velocity is prevented by tissue oxygenation; (3) this increase in leukocyte adhesion is mediated by CD11/CD18 integrins.

The baseline values for leukocyte rolling velocity and number of adherent leucocyte in the present experiments are very similar to those found by others in muscle (19) or in mesentery (20). With regards to the changes in these parameters observed after 1 h, several potential confounding factors should be taken into account for proper interpretation of the mechanisms involved in the present results. First, after surgical exposure of the muscle a spontaneous time-dependent activation of the preparation can be present. For instance after 1 h, a limited decrease in shear rate and a limited trend to decrease in leukocyte rolling velocity, but not in number of adherent leukocytes, can be found even under the control conditions. It is the reason why in the present study, conclusions were drawn only on the basis of significant differences between groups at the same time, and not on the basis of time-dependent changes within each group. In addition, since the present study was carried out in whole animals, interactions of the induced systemic hypoxia, with systemic or local hemodynamic and local oxygenation should be specifically considered.

Role of Changes in Venular Shear Rate

In our experiments hypoxia was associated with an average 16% decrease in mean arterial pressure. This observation was similar to the 20% decrease in arterial pressure reported by Marshall and Metcalfe (21) and related by these authors to arteriole dilation. It is likely that a decrease in pressure can induce decrease in the venular shear rate. When the venular shear rate reaches low values, it is known that an increase in leukocyte adhesion and a decrease in leukocyte rolling velocity are induced. For this reason we considered that, even if the difference between the shear rate in nonhypoxic and hypoxic groups did not reach the significance level in our experiments, it was important to address specifically the question of whether or not the hypoxia-induced effects might be only due to its effects on venular shear rate. Several elements allowed us to exclude this hypothesis. First, it appears from the study by Perry and Granger (22) that a large increase in leukocyte adhesion only appears for shear rates below 250 s¹. Indeed, in our experiments, hypoxia in the different groups studied was associated with a decrease in average shear rate from average values in the range 579 to 490 s¹ to values in the range 486 to 305 s¹ thus remaining higher than the threshold value which can be extrapolated from the cited work. In addition, in order to evaluate the role of the decrease in shear rate, we studied an experimental group in which the spermatic artery was partially occluded, thus inducingin the absence of hypoxemiaa decrease in the shear rate similar to that of the groups submitted to hypoxemia. In this group, adhesion and leukocyte rolling velocity were not significantly different from that observed in control animals.

Role of Integrins

In our experiments, administration of antibody to CD11/ CD18 prevented the effect of hypoxia on leukocyte adhesion. Regarding the role of leukocyte integrins, it should be noted that leukocyte adhesion induced by systemic hypoxia did not differ from that observed in several in vivo ischemia-reperfusion (IR) studies (23). However, present findings show that this mechanism can be triggered by a moderate level of hypoxia even in the absence of the large local release of proinflammatory mediator associated with ischemia-reoxygenation and stop-flow/reflow phenomenon.

With regard to the relatively rapid time course of the effects observed (i.e., within 1 h), it should be noted that in other situations of leukocyte activation, several investigators have found an increase in CD11/CD18 at the surface of the leukocytes after only a few minutes (24, 25).

With respect to hypoxia, several in vitro experiments have demonstrated that hypoxia increase in leukocyte adhesion to endothelial cells can be inhibited by anti-CD11/CD18 (26, 27). Further experiments will be necessary to define more precisely the respective role of increase in Mac-1, LFA-1 expression in leukocytes, and whether ICAM-1 or other endothelial surface proteins (27) were involved in this hypoxia-induced adhesion in vivo.

With regard to the effect of hypoxia on leukocyte rolling velocity, it was not inhibited by CD11/CD18 antibody. This observation is in line with the idea generally admitted that selectins rather than 2 integrins mediate leukocyte rolling (28).

Role of Muscle Hypoxia

Since the thickness of the cremaster muscle is less that 200 µm, the superfusion bath with high O₂ can induce a high level of oxygenation in the muscle even if arterial blood PO₂ is low. This possibility has been used in several experiments studying the effect of high O₂ tension on microvessels in vivo (29). In addition, Klitzman and coworkers (14) gave a mathematical demonstration of this possibility of oxygenating the muscle via the superfusion bath by calculating O₂ tension within the muscle in several conditions of superfusion bath PO₂. In addition the possibility that high O₂ concentrations in the bath actually reach the endothelium has been demonstrated by several researchers showing that high O₂ concentrations can modulate arteriole diameter via endothelium-dependent mechanisms. Consequently, another aspect of the present study was the possibility to study the effect of systemic hypoxia in the absence of hypoxia in the tissue under study. These experimental conditions helped us to determine whether or not local hypoxia was a necessary factor for the changes in leukocyte behavior associated with systemic hypoxia.

From the present results, it appears that even when the muscle itself was protected from hypoxia (high-PO₂ superfusion bath), the systemic hypoxia-induced increase in leukocyte adhesion was similar to that observed when the muscle was not protected from hypoxia (low-PO₂ superfusion bath). Such a phenomenon can be explained by several mechanisms: (1) activation of circulating leukocytes by systemic hypoxia, thus leading to adherence via constitutive endothelial adherence molecules in nonhypoxic areas; (2) release by leukocytes of proinflammatory mediators (such as cytokines) in the circulating blood or at the level of the nonhypoxic muscle leading to the activation of the endothelium at this level; (3) as shown in in vitro experiments, release of proinflammatory factors such as platelet-activating factor (PAF) (30, 31) and interleukin-1 (IL-1) (7, 32) by endothelial cells perfused by hypoxic blood and not oxygenated by high-PO₂ superfusion (i.e., the whole vasculature with the exception of the cremaster muscle studied). These factors will be responsible for the increase in adhesion in the nonhypoxic areas. Obviously all these mechanisms are not mutually exclusive and can be synergistic.

In contrast to that observed for leukocyte adhesion, the hypoxia-induced decrease in rolling velocity was prevented by oxygenation of the muscle through the high-PO₂ superfusion bath. This indicates that in our conditions of moderate systemic hypoxia, local tissue hypoxia is a necessary factor for the observed increase in rolling. Because rolling largely depends on P- or L-selectin, it may be hypothesized that hypoxia can interfere with selectin expression. Indeed, an increase in P-selectin expression reaching its maximum in 90 min has been reported in endothelial cells submitted to deep hypoxia in vitro (33). An even more rapid kinetics of appearance of P-selectin expression has been described during reperfusion after ischemia (34). In the present study, in a complementary protocol using immunohistochemistry, a significant increase in P-selectin was found in rats submitted to systemic hypoxia + low superfusion bath PO₂. The percentage of positively stained venules under these conditions was smaller than that obtained after ischemia- reperfusion (34). This was not surprising considering the possible differences in the level of tissue hypoxia in the case of moderate systemic hypoxia, as in the present study, or in the case of ischemia, as done by Weyrich and colleagues (34). As shown in Figure 5, the increase in the percentage of positively stained venules was prevented by tissue oxygenation through the high-PO₂ superfusion bath. These findings are qualitatively similar to our in vivo observations regarding Groups 1 to 3 described in Figure 3, thus making it likely that the decrease in leukocyte rolling velocity was related to the increase in P-selectin expression associated with tissue hypoxia.

Effect of Pentoxifylline

We found that pentoxifylline significantly inhibited hypoxia-induced increase in leukocyte adhesion and decrease in leukocyte rolling velocity. The effect of pentoxifylline on adhesion is similar to that described in several models of I/R and in line with the protective effect of pentoxifylline on tissue injury post I/R in several experimental models. This might be related to the proposed mechanisms of action of pentoxifylline via the modulation of cAMP level in the leukocyte (35). Even if in the present experiments, the inhibitory effect of pentoxifylline on leukocyte adhesion was very potent it should be borne in mind, before considering any extrapolation to clinical situations that very high doses of the drug were used in these experimental studies.

In conclusion, the present study shows that moderate levels of systemic hypoxia can be responsible for the dissemination in peripheral tissues of the leukocyte-endothelium adhesion process, which is the first step of the inflammation process.