Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

When exposed to sufficient concentration of an inflammatory stimulus, neutrophils are primed to migrate out of the bloodstream at the site of inflammation. Intravital microscopy of neutrophils in postcapillary venules during a systemic inflammatory insult has demonstrated that the migratory process involves a sequence of adhesive steps (1, 2). In the first step of the adhesion cascade, neutrophils roll along the endothelial surface. Chemoattractants associated with the endothelial cell surface at a site of inflammation can trigger the rolling neutrophils and, if sufficiently activated, the neutrophils then firmly adhere to and traverse the endothelium to migrate into the extravascular space toward the inflammatory locus.

Firm adhesion in postcapillary venules has been shown to be mediated by members of the 2 (or CD18) integrin subfamily expressed by the neutrophil. Integrins are cell- surface proteins that integrate the activities of the extracellular matrix and the cytoskeleton. They are transmembrane heterodimers consisting of an  and a  chain and are grouped into subfamilies based on their  subunit. Neutrophils express two members of the CD18 family of integrins, CD11a/CD18 (or LFA-1) and CD11b/CD18 (or Mac-1), both of which are believed to play a role in neutrophil adhesion to endothelial cells (3). CD11b/CD18 is constitutively expressed on the surface of neutrophils in a low affinity state. Exposure of the neutrophil to locally derived endothelial chemoattractants in the postcapillary venules triggers a conformational change in the integrin heterodimer, resulting in greater affinity for endothelial ligands. Activation can also stimulate release of intracellular stores of CD11b/CD18 onto the neutrophil surface (4).

Although neutrophil adherence and subsequent migration out of postcapillary venules during systemic inflammation is generally dependent on CD18, the existence of CD18-independent neutrophil migration has been demonstrated. Doerschuk and coworkers (5) compared neutrophil migration in the systemic and pulmonary circulation of rabbits exposed to hydrochloric acid (HCl), Streptococcus pneumonia, Escherichia coli endotoxin, and phorbol 12-myristate 13-acetate (PMA) and found that, although neutrophil migration into the abdominal wall in response to all stimuli tested was CD18-dependent, this was not the case with respect to neutrophil migration to the lungs. The CD18-dependency of neutrophil migration in the pulmonary vasculature varied depending on the inflammatory stimulus, with PMA and E. coli endotoxin inducing CD18-dependent migration, whereas migration in response to HCl and S. pneumonia occurred by a CD18-independent mechanism. A more recent study by Kumasaka and coworkers (6) demonstrated that although neutrophil migration in response to acute Pseudomonas aeruginosa-induced pneumonia was CD18-dependent, when recurrent pneumonia occurred at a previously inflamed site, CD18 was not required. Whether neutrophil migration occurs via the CD18-dependent or CD18-independent pathway was suggested to depend not only on the inflammatory stimulus but also on the cytokines produced at the inflammatory locus (7). Initially, because CD18-independent migration had not been observed in previous studies of systemic neutrophil migration, this alternate migratory route was suggested to be lung-specific (5). However, more recent reports have demonstrated CD18-independent neutrophil migration in organs other than the lung. Anti-CD18 monoclonal antibodies (mAbs) have been found ineffective in protecting against nephritis in a number of studies of kidney inflammation (reviewed in Reference 8) or in blocking neutrophil migration in the liver during ischemia-reperfusion injury (9).

Although CD18-independent neutrophil transendothelial migration was first reported in 1990 in animal models of inflammation, the adhesive interactions mediating this migration have not yet been identified, possibly because of the difficulty in examining neutrophil-endothelial interactions in organs in vivo. To characterize the adhesive interactions involved in CD18-independent neutrophil trans-endothelial migration, an in vitro model where this alternative migratory route is predominant would be ideal. The degree of involvement of the CD18-independent migratory route during inflammation has been shown to be tissue-specific (10, 11), and selection of this migratory pathway has been demonstrated in vitro to be regulated by the endothelial cells (12). An in vitro study using human umbilical vein endo-thelial cells (HUVECs) demonstrated that in a number of stimulatory conditions the predominant migratory route across these endothelial cells is dependent on CD18 (13).

The objective of the present study was to examine neutrophil migration across endothelial cells from the lung an organ known to support CD18-independent migration. Using human pulmonary artery endothelial cells (HPAECs), CD18 involvement in neutrophil migration was examined in response to three different types of chemoattractants, the N-formyl bacterial product formylmethionyl leucyl-phenylalanine (FMLP), the leukotriene (LT) B₄, and the chemokine interleukin (IL)-8. In addition, involvement of the 1 integrins (CD29) and the endothelial selectins in neutrophil migration in this model was examined.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials

FMLP, LTB₄, human placental Type IV collagen, and 2,2'-azino- bis(3-ethylbenzthiazoline-6-sulfonic acid) were purchased from Sigma-Aldrich Chemical Company Ltd., Dorset, UK. Function-blocking mAbs specifically recognizing the 2 integrin subunit CD18, the 1 integrin subunit CD29, and the endothelial selectin E-selectin, as well as the corresponding isotype-matched immunoglobulin (Ig) G control mAbs, were purchased from Becton Dickinson, Oxford, UK. Phycoerythrin (PE)-conjugated anti-CD11b, anti-CD29, anti-L-selectin, and PE-labeled isotype-matched IgG control antibodies were also from Becton Dickinson. Function-blocking anti-P-selectin mAb and human recombinant IL-8 were from R&D Systems, Oxon, UK. The matrix metalloproteinase (MMP) inhibitors CGS 27.023A and BB-94, and the human neutrophil elastase inhibitor ONO 5046, were supplied by Bayer plc, Slough, UK.

Cell Culture

HPAECs (Clonetics Corp., San Diego, CA) were grown in 100% humidity and 5% CO₂ at 37°C in endothelial growth medium (EGM) supplemented with 5% fetal calf serum (FCS), epidermal growth factor (10 ng/ml), hydrocortisone (1 µg/ml), bovine brain extract containing heparin (10 µg/ml), gentamicin (50 ng/ml), and amphotericin-B (50 ng/ml) (Clonetics). When approximately 80% confluent, cells were harvested, resuspended in fresh EGM, and seeded as previously described (14) at a density of 1.5 × 10⁵ cells in 200 µl onto Transwell polycarbonate membrane filters (6.5 mm diameter, 3.0 µm pore size; Corning Costar Corporation, Cambridge, MA) that had been coated with human Type IV collagen. The Transwell filter inserts were suspended in 24-well culture plates so that the filter separated the upper and lower compartments. A total of 600 µl of culture medium was placed in the lower compartment and the cells were cultured for 4 d. Scanning and transmission electron microscopy confirmed monolayer confluence and integrity (14).

All experiments were carried out on HPAECs between passages 6 and 9 and, after completion of the experiments, the cells were confirmed to be free of mycoplasma infection. Chromosome analysis also confirmed that cells remained diploid.

Isolation of Neutrophils

Human venous blood was collected by venipuncture into sterile vacutainers containing 0.105 M sodium citrate as an anticoagulant and allowed to cool to room temperature for 10 to 15 min. Neutrophils were isolated using density gradient centrifugation on Polymorphprep (Nycomed Pharma As, Oslo, Norway) at 450 × g for 35 min at 20°C. Contaminating erythrocytes were removed by hypotonic lysis and the isolated neutrophils were resuspended in Hanks' balanced salt solution (HBSS) without Ca²⁺ or Mg²⁺ for cell number and viability to be assessed before being resuspended in HBSS containing Ca²⁺ and Mg²⁺ (cHBSS) at a concentration of 1 × 10⁷ neutrophils/ml. Neutrophils isolated in this way were  97% pure and > 95% viable.

Transmigration Assay

After 4 d in culture, HPAECs on the Transwell filter inserts were transferred to a fresh 24-well tissue culture plate and neutrophil transmigration was monitored as previously reported (14). In brief, culture medium was carefully removed from other filter inserts and, after gentle rinsing of the cells with cHBSS prewarmed to 37°C, medium was replaced with cHBSS in the upper compartment of the Transwell system and either chemoattractant in cHBSS or cHBSS alone was added to the lower compartments. After a 45-min preincubation period, the migration assay was initiated by the addition of 1 × 10⁶ neutrophils to all upper compartments. The plates were incubated at 37°C in 100% humidity and 5% CO₂ for 3 h. When the effects of protease inhibitors on transmigration were being analyzed, inhibitor (1 µM) was present in both the upper and lower compartments throughout the preincubation period and during the migration experiment. After the 3-h incubation time, the plate was placed on ice and nonadherent neutrophils (upper compartment) and migrated neutrophils (lower compartment) were collected by gentle washing with cHBSS followed by centrifugation at 300 × g for 10 min. The neutrophils were then lysed by suspension in cHBSS containing 0.25% (wt/vol) Brij-35. The HPAEC monolayer with associated adherent neutrophils was removed by carefully cutting the filter membrane out of the insert and lysed by addition of cHBSS containing 0.25% Brij-35. For each experiment, a range of neutrophil concentrations was prepared and incubated in either cHBSS, FMLP, IL-8, or LTB₄ for 3 h at 37°C, after which time the neutrophils were collected by centrifuging at 300 × g for 10 min before being resuspended in cHBSS containing 0.25% Brij-35 and lysed concurrent with the migration assay samples. All neutrophil lysates were assayed for myeloperoxidase (MPO) activity using an adaptation (14) of the method of Madara and coworkers (15). A standard curve of number of neutrophils versus MPO activity was constructed and the numbers of nonadhered, adherent, and migrated neutrophils were quantified by extrapolation of the MPO activity present in upper, monolayer, and lower compartments, respectively. A linear relationship between number of neutrophils and MPO activity was obtained in the range of 0.05 to 1.0 × 10⁶ neutrophils/ml.

Assessment of the Role of Surface Adhesion Molecules in Neutrophil Transmigration of HPAECs

The involvement of the -integrins in neutrophil migration across HPAECs was analyzed by preincubating neutrophils for 15 min at 37°C with either anti-CD18 (the common subunit of the 2 integrins), anti-CD29 (the common subunit of the 1 integrins), or a corresponding isotype-matched IgG control mAb (2 to 4 µg of purified Ig/10⁵ neutrophils) before adding them to the HPAEC monolayers. Migration of untreated neutrophils and of neutrophils treated with antibody was analyzed in response to FMLP (10 nM), IL-8 (10 nM), and to LTB₄ (0.1 µM) by the method detailed in the TRANSMIGRATION ASSAY section. The function-blocking antibodies were present for the duration of the migration assay.

The involvement of the endothelial selectins E- and P-selectin was examined by treating HPAEC with anti-E- or anti-P-selectin mAb, alone and in combination. HPAECs were incubated with antibody or an isotype-matched IgG control (final antibody concentration of 5 µg/ml) in the upper compartment of the Transwells for the duration of the 45-min preincubation period and during the 3-h transmigration assay.

Quantification of Surface Adhesion Molecules on Migrating Neutrophils

PE-conjugated anti-CD11b, anti-CD29, anti-L-selectin, and fluorescein isothiocyanate (FITC)-conjugated anti-CD11a mAbs were used to fluorescently label neutrophil surface-adhesion molecules. Isotype-matched Ig labeled with PE (or FITC for CD11a) was used to assess nonspecific binding. Migrated neutrophils were collected and fixed for 10 min on ice with an equal volume of 0.4% formaldehyde in phosphate-buffered saline (PBS). After washing and resuspending in ice-cold PBS, 10 µl of PE- or FITC-labeled antibody was added to 100 µl aliquots of cell suspension containing 1 × 10⁵ neutrophils and incubated in the dark at 4°C for 45 min. Cells were then washed twice in PBS containing 1% (vol/vol) FCS before being resuspended in 1% (wt/vol) paraformaldehyde in PBS. Cell-surface fluorescence was quantified using a FACScan flow cytometer (Becton Dickinson) and expressed as the relative fluorescence index (RFI), where RFI was calculated as the ratio of the mean fluorescence intensity of specific versus nonspecific fluorescence intensity for each sample. Isotype-matched PE- or FITC-labeled Ig was used to assess nonspecific binding.

Assessment of Permeability of HPAEC System

Permeability of the HPAEC monolayers was assessed by measuring the transendothelial transfer of trypan blue-labeled albumin by an adaptation of previously described methods (16, 17). HPAECs were grown on Transwell filters for 4 d, washed, and incubated at 37°C with either cHBSS, FMLP (10 nM), IL-8 (10 nM), or LTB₄ (0.1 µM), for 3 h 45 min (i.e., the total exposure time of HPAECs to chemoattractant during a transmigration assay). After this time, the permeability of the monolayers to 4% Trypan blue-labeled albumin was determined over 30 min at 37°C in a shaking waterbath. Trypan blue-labeled albumin transfer across HPAECs was quantified by measuring the absorbance at 570 nm of aliquots sampled from the lower Transwell compartment. To determine whether the presence of neutrophils in addition to chemoattractant affected HPAEC permeability, in a separate series of experiments, neutrophils were included in the incubation period as described in the Transmigration assay procedure and permeability was measured as described earlier.

Statistical Analysis

Results are summarized as means ± standard error of the mean (SEM). Treatment versus control effects were analyzed by Student's t test. Multiple comparisons were performed using analysis of variance followed by the Bonferonni post-test. Statistical significance was considered P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Neutrophils migrated across monolayers of HPAECs in response to IL-8 and LTB₄ (Figure 1). Previous observations with FMLP-stimulated neutrophil migration demonstrated that the maximum amount of neutrophil migration to this chemoattractant (74.4 ± 14.7% of added neutrophils) occurred at a concentration of 10 nM FMLP (14). With IL-8 as the chemoattractant, 74.5 ± 21.9% of neutrophils migrated to the highest concentration examined (50 nM) after 3 h, whereas maximum migration to LTB₄ occurred at a concentration of 100 nM, with 94.1 ± 3.3% of added neutrophils migrating (Figures 1a and 1b, respectively). The FMLP, IL-8, and LTB₄ concentrations that stimulated maximum migration did not cause neutrophils to release MPO in that the MPO content of neutrophils incubated with chemoattractant for 3 h was not different from that of neutrophils incubated in HBSS alone and MPO activity was not detected in the supernatants of the transmigration assays.

View larger version (45K): [in this window] [in a new window]   View larger version (78K): [in this window] [in a new window]   Figure 1.   Neutrophil migration across HPAECs in response to different concentrations of (a) IL-8 and (b) LTB4. Neutrophils were allowed to migrate for 3 h across HPAECs as described in MATERIALS AND METHODS. The percentage of migrated neutrophils was determined by comparison of MPO in neutrophils recovered from the lower compartment of the Transwell system with the total MPO content of neutrophils added to the system. Results are means ± SEM for three independent experiments.

The involvement of CD18 in the migratory process was initially tested using concentrations of the three chemoattractants that caused maximal neutrophil migration (Figure 2a). Treating neutrophils with anti-CD18 mAb decreased FMLP-stimulated migration by 71.7 ± 4.4% compared with the isotype-matched IgG control (P < 0.001). In contrast, migration in response to LTB₄ was decreased by only 20.5 ± 10.2% (P < 0.01) and no significant decrease was observed with migration to IL-8. To determine whether the CD18-dependency of migration to IL-8 and LTB₄ might be different at submaximal migration, lower concentrations of the two chemoattractants that were previously found to stimulate neutrophil migration of approximately 40% were used. Similar to what was observed when migration was maximal, the predominant migratory route to either 5 nM IL-8 or 1 nM LTB₄ was not dependent on CD18 (Figure 2b). Migration to 1 nM IL-8 was also found to be CD18-independent (6.2 ± 0.9% of control neutrophils migrated compared with 5.3 ± 0.8% of neutrophils treated with anti-CD18 mAb).

View larger version (54K): [in this window] [in a new window]   View larger version (33K): [in this window] [in a new window]   Figure 2.   Neutrophil migration across HPAECs in response to either (a) FMLP (10 nM), IL-8 (10 nM), or LTB4 (100 nM); or (b) lower concentrations of IL-8 (5 nM) and LTB4 (1 nM) in the presence of antibody to CD18 (lightly shaded bars) and isotype-matched IgG1 (darker shaded bars). Neutrophils were allowed to migrate for 3 h across HPAECs as decribed in MATERIALS AND METHODS. The percentage of migrated neutrophils was determined by comparison of MPO in neutrophils recovered from the lower compartment of the Transwell system with the total MPO content of neutrophils added to the system. Results are means ± SEM for eight and three independent experiments for a and b, respectively.

To determine whether the inability of anti-CD18 mAb to block IL-8-stimulated and LTB₄-stimulated migration could be attributed to these chemoattractants altering the integrity of the HPAEC monolayers, the effect of exposure to the three chemoattractants on endothelial permeability was determined. Without cells, the collagen-coated filters allowed 65.19 ± 0.14% of a Trypan blue/bovine serum albumin (BSA) solution to permeate through. The presence of HPAEC monolayers decreased this permeability to 9.1 ± 3.1%. Incubation with FMLP, IL-8, or LTB₄ at concentrations used in the migration assay did not effect HPAEC monolayer permeability (Table 1). To determine whether neutrophils activated by IL-8 or LTB₄ caused increased endothelial cell injury compared with FMLP-activated neutrophils, the permeability experiments were repeated in the presence of neutrophils. Neutrophils stimulated to migrate by IL-8 or LTB₄ did not increase HPAEC permeability more than FMLP-stimulated neutrophils (Table 1).

Levels of cell-surface CD11a and CD11b were measured on nonmigrated neutrophils and on neutrophils that had migrated in response to FMLP, IL-8, and LTB₄. Levels of CD11a were not increased on neutrophils that had migrated. Compared with the amount of CD11a on neutrophils that were not stimulated to migrate (RFI value of 3.7 ± 0.6), levels of this integrin on neutrophils after 3 h migration to FMLP, IL-8, or LTB₄ were not increased (RFI values of 3.1 ± 0.3, 3.4 ± 0.5, and 2.9 ± 0.1, respectively). To determine whether CD11a levels may have peaked before the 3-h time point, amounts were quantified on neutrophils after 60 min of migration. At this earlier time point, the amount of surface CD11a was similar to that observed after 3 h of migration, with no difference between the amount of CD11a on nonmigrated and migrated neutrophils.

In contrast to CD11a, levels of CD11b did change during migration (Figure 3a). After 3 h, neutrophils that had migrated to FMLP had 1.7-fold more surface CD11b expression compared with nonmigrated neutrophils (P < 0.05). By comparison, no significant increment in the amount of CD11b on neutrophils stimulated to migrate to IL-8 or LTB₄ was observed. To determine whether CD11b levels on neutrophils that had migrated to IL-8 and LTB₄ may have peaked at an earlier time point in the 3-h transmigration assay, migrated neutrophils were analyzed for surface CD11b after 60 min. At this stage, CD11b levels were increased on all migrated neutrophils compared with levels on nonmigrated neutrophils. However, similar to that after 3 h, the greatest increase in CD11b was on neutrophils that had migrated to FMLP. Compared with levels on nonmigrated neutrophils (RFI value of 44.5 ± 6.3), after 60 min neutrophils migrated to FMLP had 2-fold more CD11b (RFI value of 93.6 ± 6.9, P < 0.01) whereas neutrophils migrated to IL-8 and LTB₄ had 1.8- and 1.7-fold more CD11b, respectively (RFI values of 81.3 ± 8.9 and 76.2 ± 4.7, respectively; P < 0.05).

View larger version (63K): [in this window] [in a new window]   View larger version (42K): [in this window] [in a new window]   Figure 3.   Expression of CD11b and L-selectin on the surface of neutrophils that had migrated in response to FMLP, IL-8, and LTB4. Migrated neutrophils were incubated at 4°C for 30 min with either (a) 5 µg/ml of PE-conjugated mAb to CD11b or (b) 1 µg/ml of mAb to L-selectin. After washing twice in PBS containing 1% (wt/vol) BSA, the neutrophils were then resuspended in 1% (wt/vol) paraformaldehyde. Cell-surface fluorescence was quantified using a FACScan flow cytometer and expressed as RFI. Isotype-matched PE-labeled Ig was used to assess nonspecific binding. Results are means ± SEM for seven and three independent experiments for a and b, respectively.

After 3 h, migrated neutrophils had less surface L-selectin compared with nonmigrated neutrophils (Figure 3b). Neutrophils that had migrated to FMLP had 3.3-fold less L-selectin (P < 0.05), whereas neutrophils migrated to IL-8 and LTB₄ had levels of L-selectin that were 1.9- and 2.3-fold less, respectively, than the control counterparts (not significant).

To determine whether the CD18-independent migration observed in response to IL-8 and LTB₄ was mediated by interactions involving the 1-integrins, the effect of preincubating neutrophils with a function-blocking antibody to the common 1 integrin subunit CD29 was examined. Anti-CD29 mAb did not decrease transmigration in response to FMLP, IL-8, or LTB₄ (Table 2). When neutrophils were treated with a combination of anti-CD18 and anti-CD29 mAbs, the effect observed was similar to that observed with anti-CD18 mAb treatment alone. Anti-CD29 mAb did not decrease the CD18-independent fraction of migration observed. Although neutrophils did express CD29 on their surface, levels were low. Neutrophils isolated and incubated in HBSS for 3 h at 37°C had an RFI value of 2.87 ± 0.15 when labeled with PE-conjugated anti-CD29 mAb. Including FMLP, IL-8, or LTB₄ in the incubation did not cause a significant increase in this basal level of CD29. Incubation with HPAECs, however, did cause a slight increase in the surface expression of CD29 (from an RFI value of 2.87 ± 0.15 to 3.76 ± 0.22; P = 0.03). This level was not further increased on neutrophils that had migrated across the HPAEC monolayers. In fact, CD29 levels were slightly lower on neutrophils that had migrated in response to FMLP (RFI value of 2.67 ± 0.29; P = 0.04) compared with levels on nonmigrated neutrophils.

The involvement of endothelial E- and/or P-selectin in CD18-dependent and -independent neutrophil migration across HPAECs was also determined. Treatment of HPAEC monolayers with function-blocking anti-P- or anti-E-selectin mAb, either alone or in combination, did not decrease neutrophil migration to any of the three chemoattractants (Table 2), indicating that these endothelial selectins are not required for either CD18-dependent or -independent neutrophil migration.

Because we had previously demonstrated that CD18-dependent migration to FMLP in this in vitro system did not require matrix degradation by MMPs or serine proteases (14), the effects of MMP and serine protease inhibitors on CD18-independent migration to IL-8 and LTB₄ were also examined for the present study. Neither inhibitor type significantly decreased CD18-independent migration (Table 3), indicating that matrix degradation is not required in this migratory pathway either.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study is the first description of an in vitro model in which transendothelial migration of neutrophils can be stimulated to occur predominantly by the novel CD18-independent route. In this model, HPAECs were found to be highly responsive in supporting chemoattractant-induced transmigration, with approximately 74% of added neutrophils migrating in response to FMLP and IL-8, whereas LTB₄ stimulated a maximal migration of 94%. Similar to in vivo observations, CD18 involvement in the migratory process depended on the chemoattractant used. When stimulated to migrate by FMLP, 72% of neutrophils traversed HPAECs by a mechanism that required CD18. In contrast, neutrophil migration across HPAECs in response to IL-8 and LTB₄ was predominantly CD18-independent, with 91 and 79%, respectively, of total migration occurring by a route not involving CD18.

In some animal models of inflammation, CD18-independent migration has been reported to be accompanied by decreased endothelial integrity (18, 19). It was therefore important to determine whether the lack of blocking effect by anti-CD18 mAb on IL-8-stimulated and LTB₄-stimulated migration observed in this study could be accounted for by a decrease in the integrity of the HPAEC monolayers. Thus, the permeability of the HPAEC system was measured after incubation with the chemoattractants both with and without neutrophils. No increase in albumin transfer was observed with IL-8 and LTB₄ at concentrations used in the migration assay either in the absence or presence of neutrophils, indicating maintenance of monolayer integrity. Although Biffl and coworkers, in one study, reported that IL-8 could alter endothelial monolayer integrity in vitro (20), the observations in the present study are in accordance with results of Huang and coworkers (21), who reported that neutrophil migration across HUVEC monolayers in response to FMLP or LTB₄ was not accompanied by an increase in vascular permeability. The CD18-independent migration in response to IL-8 and LTB₄ observed in this study cannot therefore be explained by chemotaxis through areas of damaged endothelium or exposed filter.

FMLP-stimulated migration was accompanied by an increase in the amount of CD11b/CD18 on the neutrophil surface. Although IL-8 and LTB₄ could stimulate 74 and 94% (respectively) of added neutrophils to migrate, these neutrophils did not increase surface expression of CD11b or shed L-selectin to the same extent as neutrophils that had migrated to FMLP. This difference in CD11b/CD18 upregulation supports the observation that IL-8 and LTB₄ can stimulate neutrophils to migrate by an alternative migratory route that involves adhesive interactions other than those mediated by CD18.

Interestingly, in FMLP-stimulated assays where neutrophil migration was blocked by the anti-CD18 mAb, 57% of these anti-CD18 mAb-treated neutrophils were found associated with the monolayer. This suggests that the inability of these neutrophils to migrate when CD18 is blocked may not be entirely due to lack of ability to adhere to the endothelium and may suggest an involvement for CD18 at other points in the transmigratory process. Neutrophil adhesion (as distinct from migration) to HPAECs in this instance is mediated by interactions not involving CD18, an observation supported by the finding that neutrophils from patients with leukocyte adhesion deficiency syndrome who lack functional CD18 maintain the capacity to bind to lipopolysaccharide (LPS)-pretreated human endothelial cells by a CD18-independent mechanism in vitro (22).

Previous studies have shown that up to 70% of neutrophil migration across human dermal fibroblast monolayers occurs by a CD18-independent route (23) that can be blocked by an antibody recognizing CD29, the common subunit of the 1-integrin family. Although cell-surface expression of 1 integrins is limited on blood neutrophils, members of this integrin subfamily are induced on extravasated neutrophils (24, 25) and have been reported to account for some of the CD18-independent neutrophil extravasation observed in animal models. Studies examining neutrophil migration into the inflamed joints of rats with adjuvant arthritis demonstrated that, although migration was not completely inhibited by anti-CD18 antibody treatment, it could be blocked when combined with an antibody recognizing the 1 integrin very late antigen (VLA)-4 (CD49d/CD29). The same combination of antibodies also virtually abolished neutrophil accumulation in cutaneous inflammation (10). CD18-independent migration has also been demonstrated with monocytes, and members of the 1-integrin family of surface adhesion molecules, in particular VLA-4, were found to mediate some of this CD18-independent migration (23, 26).

In the present study, however, the CD18-independent neutrophil migration observed could not be accounted for by interactions involving 1 integrins, inasmuch as a function-blocking antibody recognizing the common 1 subunit CD29 did not decrease IL-8-stimulated or LTB₄-stimulated migration of neutrophils across HPAECs. The lack of 1-integrin involvement during neutrophil transmigration observed in this in vitro model reflects in vivo observations of acute lung inflammation, where blocking 1 integrins, either alone or in combination with blocking CD18 interactions, did not decrease neutrophil accumulation into inflamed rat lungs (26). Such CD18- and CD29-independent mechanisms also account for a significant component of neutrophil migration into glomeruli in a rat model of antiglomerular basement membrane nephritis (27). Although migrated neutrophils in this study were found to express 1 integrins on their surface, these levels were low and were not increased either by the migratory process or by exposure to chemoattractant, an observation that supports the lack of inhibitory effect of 1-integrin blocking antibody on neutrophil transmigration.

Although selection of the migratory route used to traverse endothelial cells may be a direct result of the chemoattractant interacting with its receptor on the neutrophil, there is evidence that the endothelial cells themselves can function to mediate the selection process (12). Endothelial surface adhesion molecules, in particular E-selectin, have been suggested to mediate CD18-independent adhesion to endo-thelial cells in vitro (21) and, indeed, CD18-independent adhesion to endothelial cells has been called endothelial-dependent adherence (28). The involvement of the endo-thelial surface adhesion molecules E- and P-selectin in inflammation has been examined in a number of animal models with conflicting results. Anti-P-selectin mAb significantly inhibited (50 to 75%) neutrophil migration to dermal inflammatory reactions induced by zymosan-activated serum containing the chemotactic factor C5ades Arg, LPS, interferon-, and tumor necrosis factor- in arthritic rats (29). However, after instillation of bacteria in the trachea, no reduction of neutrophil extravasation was seen in P-selectin knockout mice or in P- and E-selectin double-knockout mice (30). In the current study, treatment of HPAECs with anti-E- or P-selectin mAb, either alone or in combination, did not affect neutrophil transmigration in response to any of the three chemoattractants examined. Such lack of selectin involvement is in agreement with other in vitro observations. Neutrophil adhesion to HUVECs exposed to monosodium urate crystals was reported to promote neutrophil adhesion that occurs by a firm CD18-independent and selectin-independent adhesive mechanism (31), whereas mAb to E-selectin failed to inhibit neutrophil migration across IL-1-stimulated HUVEC-amnion and also when neutrophils were additionally stimulated by LTB₄ (32). Our results and those of Reinhardt and coworkers (31) indicate that the adhesive interactions mediating CD18-independent migration do not involve E- or P-selectin.

To conclude, we have described neutrophil migration across HPAECs in vitro in response to FMLP, IL-8, and LTB₄ and report that, although these chemoattractants elicit similar ultimate neutrophil migratory responses, the adhesion molecules that mediate this transendothelial migration differ. Migration stimulated by FMLP is predominantly dependent on neutrophil CD18, whereas IL-8 and LTB₄ elicit migration by a route independent of this integrin. This CD18-independent migration is not mediated by the 1 (or CD29) integrins or by the endothelial selectins. Although CD18- and CD29-independent neutrophil migration has previously been identified in animal models of inflammation, the adhesive interactions that mediate this process have not been elucidated. The model described here provides an ideal tool for examining CD18-independent migration at the level of adhesion molecule involvement and also at the level of endothelial cell mediation.