Contribution of Endothelial Selectins and α₄ Integrins to Eosinophil Trafficking in Allergic and Nonallergic Inflammatory Reactions in Skin¹

An important early event in the process of leukocyte recruitment from the microcirculation to the interstitium is interaction with vascular endothelial cells. Initially, the circulating leukocytes are captured and roll on the endothelial cells of postcapillary venules; this process is mediated by selectins present on leukocytes (L-selectin) and endothelial cells (P-selectin and E-selectin) as well as their carbohydrate-expressing ligands (e.g., P-selectin glycoprotein ligand-1) (1, 2, 3, 4). The exact interplay between these molecules is not fully understood, although recent observations suggest that rolling can be mediated by P-selectin and L-selectin sequentially (reviewed in Refs. 4 and 5), with E-selectin being responsible for slow rolling (6). Moreover, it is clear from studies assessing neutrophil migration in vivo that the function of selectins is partially redundant (7, 8, 9). For example, thioglycolate-induced neutrophil recruitment to the peritoneal cavity of mice at 6 h was blocked by an anti-P-selectin mAb in E-selectin-deficient mice only, but not in wild-type mice (9). Moreover, granulocyte migration to the inflamed pleural cavity was abrogated only when P- and E-selectin were blocked simultaneously (8). Most studies to date have assessed the role of cell adhesion molecules in mediating neutrophil and lymphocyte migration into tissues (1, 2), and much less is known about the role of these molecules and in particular the role of selectins in mediating eosinophil recruitment in vivo.

It is believed that eosinophils play an important role in the pathophysiology of allergic diseases such as asthma and atopic dermatitis (10, 11). In these diseases, eosinophils accumulate in tissues in which they can secrete cationic proteins (e.g., major basic protein), oxygen radicals (e.g., superoxide anions), lipids (e.g., platelet-activating factor and leukotriene (LT)D₄),⁴ cytokines (e.g., IL-5 and granulocyte-macrophage CSF), and chemokines (e.g., RANTES and macrophage inflammatory protein-1α (MIP-1α)) that have the potential to damage tissue cells directly or contribute to eosinophil survival and activation (12). In the lung, the blocking of eosinophil recruitment is associated with a significant amelioration of lung function (13); therefore, agents that modulate eosinophil recruitment in vivo may be beneficial in the treatment of allergic diseases and other conditions in which eosinophil infiltration is a feature (14).

We have recently shown a role for selectins in mediating the eosinophil recruitment induced by LPS injection in the pleural cavity of mice (8). Blocking Ab against L-selectin or a combination of anti-P- and anti-E-selectin mAbs virtually abolished LPS-induced eosinophil recruitment (8). In these experiments, the anti-selectin mAbs were administered before the injection of LPS, and eosinophil accumulation in the pleural cavity was measured after 24 h (8). Since LPS requires both lymphocytes and resident macrophages to induce eosinophil recruitment (15), it is possible that the anti-selectin mAbs inhibited eosinophil recruitment indirectly by modulating the recruitment and/or activation of lymphocytes and macrophages. In this respect, an anti-α₄ integrin mAb reportedly attenuates neutrophil influx in mouse skin in a delayed-type hypersensitivity (DTH) reaction, presumably by an indirect effect as a result of inhibiting mononuclear cell recruitment and/or activation (16).

The aim of the present study was to investigate the role of endothelial selectins in directly mediating the recruitment of ¹¹¹In-labeled eosinophils in response to chemoattractants (eotaxin, MIP-1α, and LTB₄) and in response to allergen in sensitized animals. These experiments were conducted as described in a recently developed mouse model in which chemokine responsiveness was investigated (17). To dissociate the contribution of selectins to the development of an allergic reaction from their role in the actual process of eosinophil recruitment, anti-selectin mAbs were administered before a pulse of ¹¹¹In-labeled eosinophils but after the start of the reaction. In addition, we evaluated whether the expression of P- and E-selectin was up-regulated in sites of delayed-onset allergic reactions using a localization of radiolabeled mAbs. Because we found evidence of a selectin-independent component to eosinophil accumulation in a DTH reaction, we also examined the role of α₄ integrins using a blocking mAb.

Female CBA/Ca mice (18–20 g) were purchased from Harlan (Bicester, U.K.). CBA/Ca mice overexpressing the murine IL-5 gene (Tg1 mice) (18) were obtained from Glaxo Wellcome (Greenford, U.K.) and were bred in-house.

The following were obtained from Sigma (Poole, U.K.): OVA, methylated BSA (MBSA), 2-mercaptopyridine-N-oxine, and Percoll. PBS (calcium- and magnesium-free, pH 7.4) and HBSS were obtained from Life Technologies (Paisley, U.K.). LTB₄ was purchased from Cascade (Reading, U.K.). Dextran T500 and Percoll were supplied by Pharmacia (Milton Keynes, U.K.). Na¹²⁵I, Na¹³¹I, and ¹¹¹InCl₃ were purchased from Amersham (Little Chalfont, U.K.). Murine eotaxin and MIP-1α were supplied by PeproTech (London, U.K.). Anti-mouse B220 (rat IgG2b) and anti-mouse CD2 (rat IgG2b) were obtained from PharMingen (Cambridge Bioscience, Cambridge, U.K.) and goat anti-rat IgG microbeads were supplied by Miltenyi Biotec (Camberley, U.K.).

Anti-L-selectin (MEL-14) and anti-α₄ integrin (PS/2) mAb-producing cell lines were purchased from American Type Culture Collection (Manassas, VA) and grown in a hollow fiber bioreactor; mAbs (both rat IgG2b) were purified by ammonium sulfate precipitation followed by extensive dialysis against PBS. Rat IgG1 to murine P-selectin (both the blocker 5H1 and the nonblocker 10A10) were prepared as described previously (8). Rat IgG2b to murine E-selectin (10E6, a blocker) and a nonblocking anti-murine E-selectin mAb (14E4, rat IgG2b) were as described previously (8). Anti-E- and P-selectin mAbs and the isotype controls 8B9 (IgG1) and 2–4A1 (IgG2b) were gifts from Dr. B. Wolitzky (Hoffman-La Roche, Nutley, NJ).

Eosinophils were purified from the blood of CBA/Ca mice overexpressing the IL-5 gene. In our transgenic mouse colony, eosinophils account for ∼60% of all circulating blood leukocytes (data not shown). Animals were anesthetized with pentobarbitone (2 mg per mouse injected i.p. with 50 IU of heparin). Blood was obtained by cardiac puncture (three to four donor mice per experiment) after 5 min, and RBCs were sedimented using dextran T500 (one part of blood to four parts of 1.25% dextran). The leukocyte-rich supernatant was removed, centrifuged at 300 × g for 7 min, and layered onto a discontinuous 4-layer Percoll gradient (densities: 1.070, 1.075, 1.080, and 1.085 g/ml). The gradients were centrifuged at 1500 × g for 25 min at 20°C, and eosinophils and lymphocytes were collected from the 1.080/1.085 interface. Lymphocytes were removed using negative immunoselection with rat anti-mouse CD2 and B220 mAbs on a magnetic cell separation system (MACS) BS column according to the manufacturer’s guidelines (Miltenyi Biotec). Briefly, the eosinophil and lymphocyte pellet was resuspended in PBS/BSA (1 × 10⁷ cells/500 μl) and incubated with 10 μg/ml of anti-CD2 and 7.5 μg/ml of anti-B220 for 20 min on ice. The cells were washed and resuspended in PBS/BSA (80 μl of PBS/BSA per 1 × 10⁷ cells). A total of 20 μl of goat anti-rat IgG microbeads per 1 × 10⁷ cells were added, and the cells were incubated for 20 min at 6 to 8°C. The cell suspension was put through the immunomagnetic selection column, and the eosinophils were collected with the column effluent. The purified eosinophils obtained in this manner were >96% pure and >98% viable; contaminating cells were mononuclear.

For the in vivo experiments, eosinophils were radiolabeled as described previously for guinea pig cells (19). Briefly, purified mouse eosinophils were incubated with ¹¹¹In (∼100 μCi in 10 μl) that had been chelated to 2-mercaptopyridine-N-oxine (40 μg in 0.1 ml of 50 mM PBS, pH 7.4) for 15 min at room temperature. Subsequently, cells were then washed twice in PBS/BSA and then resuspended at a final concentration of 1 × 10⁷¹¹¹In eosinophils/ml.

In this study, two immunization procedures were used that, after Ag challenge, result in inflammatory reactions that are of relevance to allergic skin diseases in humans (20). These procedures were designed to induce a classic DTH reaction and a late-phase reaction to allergen. While DTH reactions are not normally characterized by the accumulation of eosinophils, recruitment of these cells has been observed in a DTH reaction in both human skin (20) and mouse skin when circulating eosinophils were elevated (21). These observations may be relevant for several skin diseases, including atopic dermatitis, in subjects whose blood eosinophil numbers are increased.

A DTH reaction was induced by immunizing nontransgenic CBA/Ca mice with MBSA in CFA (22). The main advantage of this sensitization procedure is that it uses a soluble Ag to induce a DTH reaction, which enables discrete sites on the back skin of animals to be injected with Ag, rather than using a topical application of an insoluble or poorly soluble Ag (22). Briefly, mice received two intradermal (i.d.) injections in the abdominal skin of 50 μl of MBSA (5 mg/ml) that had been emulsified in CFA. After 7 to 8 days, the animals were anesthetized and shaved; MBSA (1 and 10 μg/site) was subsequently injected i.d. in the back skin.

In the second procedure, animals were immunized with OVA, which had been adsorbed to aluminum hydroxide, as described previously (23). Briefly, CBA/Ca mice were injected s.c. on days 1 and 8 with 0.2 ml of a solution containing 100 μg of OVA and 70 μg of aluminum hydroxide. The animals were anesthetized and shaved at 7 to 8 days after the last immunization, and OVA (0.1 to 1 μg.site) was injected i.d. The latter reaction, which is associated with an early increase in plasma leakage and mast cell degranulation (data not shown), will be referred to as an active cutaneous anaphylactic (ACA) reaction.

Using the protocol shown in Figure 1, time course experiments were conducted initially to determine the ideal measurement periods for ¹¹¹In eosinophil accumulation in the ACA and DTH reactions. Sensitized animals were challenged i.d. with Ag at 20 h, 4 h, and immediately before the i.v. injection of radiolabeled cells. The recruitment of ¹¹¹In eosinophils in mouse skin was then assessed over a period of 4 h, resulting in the following measurement periods: 0 to 4 h, 4 to 8 h, and 20 to 24 h after Ag challenge (see Fig. 1).

For the experiments assessing ¹¹¹In eosinophil recruitment in response to direct-acting chemoattractants, animals received i.d. injections of eotaxin (10 pmol/site), MIP-1α (10 pmol/site), LTB₄ (150 pmol/site), and PBS (50 μl/site) at 10 min after the injection of ¹¹¹In eosinophils. The dose of each chemoattractant was chosen to induce a comparable degree of eosinophil recruitment, which was virtually complete within 4 h (data not shown).

At the end of the 4-h measurement period, blood was obtained by cardiac puncture, and the animals were killed using an overdose of pentobarbitone. The back skin was removed, and skin sites were punched out using a 10-mm diameter wad punch and counted in a gamma-counter together with aliquots of the infused ¹¹¹In eosinophils and the blood samples. By comparing the counts in a skin site with the number of cpm associated with one eosinophil, the number of ¹¹¹In eosinophils at each site was calculated. The percentage of ¹¹¹In eosinophils circulating at the end of the 4-h measurement period was calculated by comparing the cpm of the blood sample with the number of cpm injected and assuming a blood volume of 70 ml/kg. In all experiments, each animal received up to six i.d. injections of Ag (sensitized animals) or chemoattractant.

We used binding, nonblocking mAbs as controls for the endothelial selectins, although these mAbs gave results that were no different from those obtained with isotype-matched IgG, as we have shown previously (8). Blocking and nonblocking (control) mAbs were administered i.v. at 10 min before the injection of ¹¹¹In eosinophils. Thus, in the ACA (4–8-h measurement period) and DTH (20–24-h measurement period) reactions, mAbs were administered after the start of the reaction but before the i.v. injection of ¹¹¹In eosinophils. The dose of each mAb was as follows: 5H1 (anti-P blocking, 30 μg/mouse), 10A10 (anti-P nonblocking, 30 μg/mouse), 10E6 (anti-E blocking, 30 μg/mouse), 14E4 (anti-E nonblocking, 30 μg/mouse), MEL-14 (anti-L-selectin, 200 μg/mouse), PS/2 (anti-α₄ integrin, 40 μg/mouse), and 2–4A1 (IgG2b, 40 or 200 μg/mouse). None of the mAbs mentioned above had any significant effect on the levels of circulating ¹¹¹In eosinophils as measured at 2 or 4 h after mAb injection i.v. (data not shown).

Purified eosinophils (5 × 10⁵ cells) were incubated at 4°C for 30 min with a saturating concentration of MEL-14 in PBS/BSA. After two washes, FITC-conjugated goat anti-rat IgG was added and incubated for 30 min at 4°C. The cells were washed twice, and FITC fluorescence was determined on a FACScan flow cytometer (Becton Dickinson, Oxford, U.K.) and analyzed using CellQuest software.

The expression of E-selectin and P-selectin at skin sites was assessed using a method that quantifies the accumulation of radiolabeled anti-selectin mAbs and compares this result with the accumulation of radiolabeled, isotype-matched, nonbinding mAbs (24, 25). 5H1 and 10E6 were radiolabeled with ¹²⁵I to a specific activity of ∼2.5 μCi/μg using the Iodogen method. Isotype-matched mAbs (8B9 and 2–4A1) were radiolabeled with ¹³¹I to a similar specific activity. Free radiolabel was separated from radiolabeled mAbs by gel filtration (Sephadex PD10, Pharmacia) followed by overnight dialysis. The free iodine at the time of experiments was <3%.

The expression of E- and P-selectin was assessed in sites of ACA and DTH reactions. At 4 h and 20 h after Ag challenge of ACA and DTH sites, respectively, ¹²⁵I/5H1 or ¹²⁵I/10E6 together with ¹³¹I/8B9 or ¹³¹I/2–4A1 were injected i.v. into mice (∼1 μg of each mAb) and allowed to accumulate at sites of inflammation for 5 min. Animals then received an overdose of pentobarbitone, blood samples were collected into heparin, and plasma was prepared. The back skin was removed, and skin sites were punched out and counted together with plasma samples in the gamma-counter. The activity of each mAb at each skin site was compared with the plasma activity. The specific accumulation of anti-E or anti-P-selectin mAbs was calculated by subtracting the accumulation value for control ¹³¹I-labeled mAbs and was expressed as microliter plasma equivalents.

All results are presented as the mean ± SEM. Normalized data were analyzed by one-way ANOVA, and differences between groups were assessed using the Student-Newman-Keuls posttest. A p value of <0.05 was considered significant. The percent inhibition was calculated by subtracting the background values obtained in response to an i.d. injection of PBS.

The i.d. injection of LTB₄, eotaxin, and MIP-1α in mouse skin induces a recruitment of ¹¹¹In eosinophils that is fast in onset and mostly complete within 3 to 4 h (21). Figure 2 shows the effects of treatment with blocking and nonblocking anti-P- and anti-E-selectin mAbs on the ¹¹¹In eosinophil recruitment induced by comparably effective doses of LTB₄ (150 pmol/site), eotaxin (10 pmol/site), and MIP-1α (10 pmol/site). Treatment with the blocking anti-E-selectin mAb 10E6 had no significant effect on the ¹¹¹In eosinophil recruitment induced by the chemoattractants tested (Fig. 2,a). In contrast, the anti-P-selectin mAb 5H1 significantly inhibited LTB₄-, eotaxin-, and MIP-1α-induced ¹¹¹In eosinophil recruitment by 49%, 74%, and 91%, respectively (Fig. 2,b). However, combined treatment with both anti-P- and anti-E-selectin mAbs resulted in a further inhibition of ¹¹¹In eosinophil recruitment, such that responses were abolished (i.e., reduced to the PBS background level) or reduced by >95% (Fig. 2 c). There was no effect on the percentage of ¹¹¹In-labeled eosinophils circulating in the blood after the 4-h measurement period (data not shown).

Two immunization procedures were used to induce late-onset eosinophil recruitment in mouse skin: OVA that had been adsorbed to aluminum hydroxide (ACA reaction) or MBSA in CFA (DTH reaction). Figure 3 shows the time-course of ¹¹¹In eosinophil recruitment after Ag challenge of sensitized animals. In these experiments, ¹¹¹In eosinophils were injected i.v. at various intervals after Ag challenge, and recruitment in skin sites was assessed over a 4-h measurement period. In the ACA reaction, there was significant (p < 0.05) ¹¹¹In eosinophil recruitment in all measurement periods following challenge with OVA. The number of ¹¹¹In eosinophils at skin sites peaked at 4 to 8 h, but there was ∼40% as much eosinophil-recruiting activity at 20 to 24 h (p < 0.05 compared with 4–8 h) following Ag challenge. In the DTH reaction, significant (p < 0.05) ¹¹¹In eosinophil recruitment was delayed until 4 to 8 h following challenge with MBSA, although the recruitment was substantially greater at 20 to 24 h. Because ¹¹¹In eosinophil recruitment peaked at 4 to 8 h and 20 to 24 in the ACA and DTH reactions, respectively, these measurement periods were used in the majority of subsequent experiments. Eosinophil recruitment into 8-h ACA and 24-h DTH sites was confirmed by histology (data not shown).

i.d. injections of increasing doses of OVA (0.1–1 μg) in the skin sites of animals that had been sensitized previously to OVA induced a dose-dependent recruitment of ¹¹¹In eosinophils as measured at 4 to 8 h after Ag challenge (Fig. 4). Pretreating animals with blocking anti-P- or anti-E-selectin mAbs alone just before the i.v. injection of ¹¹¹In eosinophils but at 4 h after the injection of Ag had no effect on the recruitment of these cells to skin sites (Fig. 4, a and b). In contrast, a combined treatment with blocking anti-P- plus anti-E-selectin mAbs abolished the recruitment of ¹¹¹In eosinophils at all doses of Ag (Fig. 4 c). The percentage of ¹¹¹In eosinophils circulating after 4 h was not different between the groups.

The i.d. injection of 1 and 10 μg of MBSA in the skin sites of animals that had been sensitized previously to MBSA induced a dose-dependent recruitment of ¹¹¹In eosinophils as measured at 20 to 24 h after Ag challenge (Fig. 5). Treating animals with blocking anti-E-selectin mAb immediately before the pulse of ¹¹¹In eosinophils partially inhibited ¹¹¹In eosinophil recruitment (Fig. 5,a); in contrast, the blocking anti-P-selectin mAb 5H1 failed to alter significantly the number of eosinophils that were accumulating (Fig. 5,b). A combined treatment with both anti-P- and anti-E-selectin mAbs significantly inhibited ¹¹¹In eosinophil recruitment in the DTH reaction, but the combination was only more effective than the anti-E-selectin mAb alone at the dose of 1 μg/site of MBSA (Fig. 5 c). Thus, in contrast to the ACA reaction, in which there was a total suppression of ¹¹¹In eosinophil accumulation when P- and E-selectin were blocked, the trafficking of the same cells clearly shows some independence of endothelial selectins in the DTH reaction.

In these experiments, mAbs were administered just before the i.v. injection of ¹¹¹In eosinophils but at 20 h after the injection of Ag, when endothelial selectin expression could have been very high. Thus, the failure to inhibit DTH completely could have been due to insufficient an plasma concentration of blocking mAbs to saturate all endothelial selectins. Therefore, we used a six times higher dose of each blocking mAb (200 μg/mouse), which achieves a plasma concentration of >90 μg/ml for at least 24 h (8). However, even at this dose, anti-P- plus anti-E-selectin mAbs failed to cause any further reduction in the recruitment of ¹¹¹In eosinophils (data not shown).

Thus, it was clear from the previous experiments that P- and E-selectin-independent ¹¹¹In eosinophil recruitment occurred in the DTH reaction, specially at the Ag dose of 10 μg/site. Additional experiments were performed to test whether L-selectin played any role in mediating this endothelial selectin-independent component. Mouse eosinophils that had been purified from blood using the procedure described in this study expressed L-selectin, as assessed by the binding of the anti-L-selectin mAb MEL-14 (data not shown). In addition, we have reported previously that this level is similar to that found on eosinophils in whole blood (17). Pretreating animals with MEL-14 alone failed to alter the recruitment of ¹¹¹In eosinophils in the DTH reaction (data not shown). Moreover, a combined treatment with anti-E-selectin, anti-P-selectin, and anti-L-selectin had no further inhibitory effect than anti-E-selectin alone in the DTH reaction (Fig. 5 d). In contrast to the lack of effect of the anti-L-selectin at inhibiting ¹¹¹In eosinophil recruitment in the DTH reaction in mouse skin, the same batch of MEL-14 significantly (>80%) blocked KC- and LPS-induced neutrophil recruitment into the pleural cavity without an effect of circulating leukocytes (data not shown and 8 .

Because a combined treatment with anti-selectin mAbs failed to suppress completely the recruitment of ¹¹¹In eosinophils in the 20- to 24-h-old DTH reaction but abolished the recruitment of these cells in a 4- to 8-h-old ACA reaction, we tested whether selectin-independent eosinophil migration would also occur in a 20- to 24-h-old ACA reaction. Figure 6 shows that, in contrast to their effects in the DTH reaction, the combined treatment with anti-P- and anti-E-selectin mAbs virtually abolished ¹¹¹In eosinophil recruitment in the later ACA reaction (Fig. 6). When either mAb was used alone, no inhibitory effect was observed (data not shown).

To investigate the expression of P- and E-selectin in sites of ACA and DTH reactions in mouse skin, immunolabeling studies using ¹²⁵I-anti-P- and ¹²⁵I-anti-E-selectin mAbs were conducted. Figure 7 shows the specific accumulation of anti-E-selectin (10E6)- and anti-P-selectin (5H1) mAbs in sites of ACA and DTH reactions in mouse skin. In accordance with the in vivo data described above, there was a marked (∼10-fold) expression of E-selectin in sites of 20- to 24-h-old DTH reactions and a significant but less marked expression of E-selectin in sites of 4- to 8-h-old ACA reactions (Fig. 7,a). Similarly, the expression of P-selectin was significantly increased in sites of 4- to 8-h-old ACA and 20- to 24-h-old DTH reactions (Fig. 7 b).

Recently, it has become clear that the α₄ integrin very late Ag-4 can mediate the rolling of α₄ integrin-positive cells both in vitro and in vivo in addition to mediating firm adhesion (26, 27). Although α₄ integrin-mediated rolling is thought to be more efficient at lower shear rates (5), an anti-α₄ integrin mAb significantly inhibited the rolling of human eosinophils in rabbit mesentery (27) as well as the rolling in sensitized mouse cremasteric postcapillary venules at 8 h after Ag challenge (28). Pretreating mice with a blocking anti-α₄ integrin mAb (PS/2, administered before the i.v. injection of ¹¹¹In eosinophils but at 20 h after Ag challenge) inhibited ¹¹¹In eosinophil recruitment in a 20- to 24-h-old DTH reaction by 92 to 100% (Fig. 8,b). In a 4- to 8-h-old ACA reaction, the ¹¹¹In eosinophil recruitment induced by 0.1 and 1 μg of OVA was inhibited by 76 to 85%, respectively (Fig. 8,a). Similarly, ¹¹¹In eosinophil recruitment in response to 1 μg of OVA in a 20- to 24-h-old ACA reaction was inhibited by 65% (Fig. 8 a). A dose of 0.1 μg of OVA failed to induce any significant recruitment of ¹¹¹In eosinophils when measured from 20 to 24 h (data not shown).

In the present study, we have evaluated the role of endothelial selectins (P- and E-selectin) in mediating the recruitment of eosinophils into sites of acute and late-onset inflammation in mouse skin. mAbs that block selectin function were administered at different times after Ag challenge but immediately before a pulse of radiolabeled eosinophils. This strategy was adopted to dissociate the contribution of selectins to the development of a delayed-onset reaction from their role in the actual process of eosinophil recruitment. The data presented in the present study suggest a direct effect of endothelial selectins on the mediation of eosinophil trafficking.

Eosinophil accumulation in nonallergic inflammation was induced by the lipid mediator LTB₄ and two CC chemokines, MIP-1α and eotaxin. Allergic inflammation was induced by the i.d. injection of OVA into sensitized mice to cause a rapid increase in plasma leakage (data not shown), and a peak of radiolabeled eosinophil trafficking was measured 4 to 8 h later. The signals present at challenged skin sites at this time, which induce eosinophil accumulation, are not fully characterized, but eotaxin and its receptor, CC chemokine receptor 3 play a major role (17). Allergic inflammation was also induced by Ag in MBSA-sensitized mice to elicit a DTH reaction. While DTH reactions are not normally characterized by the accumulation of eosinophils, recruitment of these cells has been observed in a DTH reaction in human skin (20) and mouse skin when circulating eosinophils were elevated (21). These observations may be relevant for several skin diseases, including atopic dermatitis, in subjects whose blood eosinophil numbers are increased. In our experiments, the trafficking of ¹¹¹In eosinophils was delayed compared with the ACA reaction but was of similar magnitude, suggesting that eosinophil chemoattractant molecules were present at skin sites. Using a polyclonal Ab directed against murine eotaxin, we have found that this CC chemokine plays an important role in eosinophil accumulation in the DTH reaction (our manuscript in preparation).

The present study shows that P-selectin plays a major role in mediating eosinophil recruitment in the acute inflammatory reactions induced by direct-acting chemoattractants; however, neither P- or E-selectin alone is sufficient to mediate eosinophil recruitment in an ACA reaction, despite the fact that the expression of both molecules is increased. In addition, in a DTH reaction, in which there is also an increased expression of both E- and P-selectin, eosinophil recruitment is largely selectin-independent at a high dose of the Ag. The finding that P-selectin plays a major role in mediating eosinophil recruitment in acute reactions is in agreement with data showing an important role for P-selectin in mediating early leukocyte rolling in vivo (29, 30, 31). Similarly, our observations concur with data obtained in P-selectin knockout mice and anti-P-selectin mAb-treated mice that show the importance of this adhesion receptor in mediating leukocyte influx in acute inflammatory reactions (7, 29, 32, 33). Interestingly, there was a significantly greater inhibition of chemoattractant-induced eosinophil recruitment when an anti-E-selectin mAb was also added to the anti-P-selectin mAb (see Fig. 2). These results suggest that there is a low basal expression of E-selectin in mouse skin that may be important to support leukocyte rolling under conditions in which P-selectin is blocked. Although we failed to observe any increase in E-selectin expression following the injection of chemoattractants, there was a consistent, specific ¹²⁵I-anti-E-selectin mAb accumulation in naive skin sites (see Fig. 7) suggesting basal expression of this molecule. Moreover, several studies have reported a low but measurable amount of E-selectin expression in the skin of naive subjects (4).

In contrast to the effects on the chemoattractants, neither an anti-P-selectin nor an anti-E-selectin mAb alone had any significant inhibitory effect on a 4- to 8-h-old ACA reaction in the mouse, while combined treatment abolished eosinophil recruitment. Moreover, studies with radiolabeled mAbs showed that there was a significant increase in the expression of both P- and E-selectin at the skin sites of ACA reactions. These results are remarkably similar to studies evaluating the role of P- and E-selectins in mediating thioglycolate-induced late neutrophil recruitment (4–8 h) in the mouse (9, 32). A blockade or lack of expression of both P- and E-selectin was essential if a complete inhibition of neutrophil recruitment was to be observed (9, 32, 34). Taken together, these findings highlight the redundant function of P- and E-selectin, when both molecules are expressed simultaneously, at mediating both neutrophil and eosinophil recruitment in models of chronic inflammation (4). Moreover, they suggest that a blockade of both endothelial selectins is necessary if an effective inhibition of leukocyte recruitment at sites of chronic inflammation is to be achieved pharmacologically.

Although platelet P-selectin could contribute to the accumulation of radiolabeled mAb in this reaction, extensive work in the mouse by Eppihimer et al. (25) showed that blocking platelet function in vivo did not alter the accumulation of a ¹²⁵I-labeled P-selectin mAb (RB40.34) in inflamed murine tissues. Along with the observation of negligible radiolabeled mAb binding to platelets in whole blood (25), the data suggest that of ¹²⁵I-P-selectin mAb bind to endothelial cells.

While there was a marked increase in the expression of both P- and E-selectin in a 20- to 24-h-old DTH reaction, the eosinophil recruitment was only partially inhibited when a combination of anti-P- and anti-E-selectin mAbs was used. The relative lack of effect of a combined treatment with anti-P- and anti-E-selectin mAbs was most marked at the highest dose of Ag tested (10 μg/site). This was not due to a lack of blocking mAbs, since no further inhibition occurred when a 6 times higher dose was used. The addition of an anti-L-selectin mAb failed to inhibit eosinophil recruitment any further, demonstrating that eosinophil recruitment was mostly selectin-independent in the DTH reaction induced by a high dose of Ag. In contrast, eosinophil recruitment in a 20- to 24-h ACA reaction was still selectin-dependent even at this late measurement period. Our findings are in contrast with previously published data demonstrating a significant inhibitory effect of a single or combined blockade or a lack of expression of L-, P-, or E-selectin on the recruitment of neutrophils and/or mononuclear cells in DTH reactions in mouse skin (9, 35, 36, 37, 38). In the latter studies, neutrophil recruitment was inhibited by 50 to 60%; however, an effect on the recruitment and activation of intermediate cell types (e.g., lymphocytes) may account for this finding. For example, in L-selectin knockout mice, the lack of migration of lymphocytes to the draining lymph nodes after sensitization and/or challenge was deemed responsible for a reduced DTH reaction in knockout animals (36). In mice that were deficient in both P- and E-selectin, the inhibition of leukocyte recruitment into DTH sites could be at least partially explained by the lack of migration of memory T cells into skin sites (38). In support of this possibility, recent studies have demonstrated a role for endothelial selectins and P-selectin glycoprotein ligand-1 in mediating the trafficking of Th1 cells but not Th2 cells to DTH sites in mouse skin (39, 40). The inability of anti-selectin mAbs to block eosinophil recruitment in the DTH reaction in the present study suggests that there are additional selectin-independent adhesion pathways that may mediate eosinophil rolling in these late-phase skin reactions.

Recently, Binns and colleagues (41) showed that an anti-E-selectin mAb effectively blocked both lymphocyte and neutrophil recruitment into cutaneous DTH reactions in sensitized pigs. Similar to our studies, the mAb was administered just before the i.v. radiolabeled cells and at different times after the i.d. challenge with Ag (41). However, we failed to show a marked inhibitory effect of anti-E-selectin on eosinophil recruitment into sites of DTH reactions in mouse skin. Although E-selectin, by inference, does support eosinophil rolling in mouse skin microvessels (as discussed above), other work has shown that this molecule plays a greater role in mediating the rolling of neutrophils than eosinophils in vivo (42). It is also possible that the requirements for selectins vary in different species.

It is now known that the α₄ integrin very late Ag-4 can mediate the rolling of α₄ integrin-positive cells both in vitro and in vivo in addition to mediating firm arrest (26, 27, 28, 43). The presence of α₄ integrin on the tips of microvilli is consistent with the capacity of this integrin to mediate rolling (43). Although α₄ integrin-mediated rolling is thought to be more efficient at lower shear rates (5), an anti-α₄ integrin mAb significantly inhibited the rolling of eosinophils in rabbit mesentery (27) and the rolling of leukocytes after Ag challenge in mouse cremaster (28). These results suggest that α₄ integrin may play a significant role in mediating eosinophil rolling at shear rates that occur in vivo; however, the most recent evidence suggests that this is dependent upon prior tethering by endothelial selectins (28). In our studies, an anti-α₄ integrin mAb virtually abolished eosinophil recruitment in the 20- to 24-h-old DTH reaction in marked contrast to the partial inhibitory effects of anti-selectin mAbs. However, the same anti-α₄ integrin mAb also inhibited eosinophil recruitment in the ACA reaction by ≤85%. It is not clear from our results whether the greater efficacy of the anti-α₄ integrin mAb in the DTH reaction reflects an action of the mAb at inhibiting leukocyte rolling, firm adhesion, or both. Similarly, the ligand for α₄ integrin that putatively mediates rolling (or firm adhesion) has not been identified.

In conclusion, the present study highlights the importance of endothelial selectins for eosinophil migration in vivo and suggests that these molecules may be good therapeutic targets for the development of drugs for the treatment of certain allergic skin diseases. In this respect, we have recently shown that the polysaccharide fucoidin effectively inhibits eosinophil infiltration in acute inflammatory reactions in guinea pig skin (44). However, selectin-based therapy may not be effective in all situations, because selectin-independent eosinophil recruitment can occur in late-phase inflammation (for example, under conditions that mimic the DTH reaction). Clearly, additional studies are needed to define the cell adhesion pathway(s) that are responsible for selectin-independent eosinophil recruitment, but the α₄ integrins are attractive candidates. Whether α₄ integrin-based strategies will be more effective than selectin-based strategies at inhibiting eosinophil recruitment in human disease remains to be determined.