Human Eotaxin Induces α₄ and β₂ Integrin-Dependent Eosinophil Accumulation in Rat Skin In Vivo: Delayed Generation of Eotaxin in Response to IL-4¹

Selective tissue infiltration of eosinophils at sites of inflammation is one of the striking features of allergic disease states such as asthma, rhinitis, or atopic dermatitis, but the mechanisms involved in specific eosinophil recruitment are not yet fully understood. In this context, leukocyte extravasation appears to follow a series of sequential leukocyte-endothelial cell interactions involving the selectins, the integrins, and the chemokines as the major molecular regulators (1, 2).

Eotaxin, a CC chemokine originally purified from the bronchoalveolar lavage fluid of allergen-challenged guinea pigs, is a potent and selective eosinophil chemoattractant that is active both on guinea pig and human cells (3, 4). Guinea pig (5, 6), mouse (7, 8), human (9, 10), and rat (11) eotaxin have now been cloned. Human eotaxin exhibits a similar high potency and selectivity for human eosinophils, inducing the elevation of intracellular free calcium concentrations, chemotaxis, actin polymerization, production of reactive oxygen species, and α_Mβ₂ (CD11b/CD18) up-regulation in vitro (9, 10, 12, 13). Although eotaxin is constitutively expressed at low levels in most human organs, marked up-regulation in protein or message has been detected in the epithelium and submucosa of nasal polyp tissues (9) and in the intestine of patients with inflammatory bowel diseases (10), where there is a clear eosinophil infiltration. The major cellular sources of eotaxin are thought to be the epithelium, the endothelium, and activated infiltrating leukocytes such as monocytes and eosinophils (9, 10, 14). Human eotaxin has been demonstrated by histologic examination to induce eosinophil accumulation in monkey skin (9). We found that human eotaxin was inactive on guinea pig eosinophils, both in vitro and in vivo (data not shown) despite the observation that guinea pig eotaxin is highly active on human eosinophils (4). We then found that ¹²⁵I-human eotaxin bound to rat eosinophils (data not shown). Therefore, we have used a previously described rat skin model (15, 16) to examine inflammatory responses induced by human eotaxin in vivo.

The initial interaction between leukocytes and the vascular endothelium, namely rolling, is a transient and reversible adhesion mainly mediated by the selectins and their carbohydrate counterligands (17, 18), although recently the α₄ integrins, α₄β₁ (very late activation Ag-4) and α₄β₇, have also been implicated in this process (19, 20). Rolling is thought to result in the exposure of leukocytes to chemoattractants, such as chemokines, presented on the surface of endothelial cells (21). The chemokines then trigger a firm adhesion of leukocytes that is thought to be mediated through the interaction of integrins with their endothelial counterreceptors: vascular cell adhesion molecule (VCAM⁴)-1 for α₄ integrins (for review, see 22 and intercellular adhesion molecule (ICAM)-1 and ICAM-2 for β₂ integrins (23, 24). These adhesive interactions serve to facilitate the subsequent transendothelial migration of leukocytes along chemoattractant gradients (25, 26).

Interactions of α₄ integrins with VCAM-1 and fibronectin can mediate eosinophil accumulation at sites of inflammation because eosinophils, but not normally neutrophils, express α₄ integrins on their surface (22), and mAbs against the α₄ integrin subunit or VCAM-1 inhibit eosinophil accumulation in several in vivo models of allergic and nonallergic inflammation (16, 27, 28, 29). Further, the recent cloning of the human eotaxin receptor, CCR3, and its high expression on eosinophils (30, 31) provide a molecular basis for the eosinophil selectivity of eotaxin. Hence, all these findings have led us to hypothesize that the combination of the chemoattractant activity of eotaxin together with its possible role in integrin activation might determine the selective eosinophil traffic observed at sites of eosinophilic inflammation. Our findings show that human eotaxin is a potent inducer of eosinophil accumulation in vivo, and that this response is dependent on α₄ and β₂ integrins and their endothelial cell counterligands. Further, we provide evidence that endogenously generated eotaxin is involved in eosinophil accumulation induced by IL-4.

Male Sprague Dawley cell donor rats (400–500 g) and male Sprague Dawley in vivo assay rats (200–300 g) were purchased from Harlan-Olac, Bicester, Oxfordshire, U.K.

Pentobarbitone sodium (Sagatal, 60 mg/ml) was purchased from May and Baker, Dagenham, Essex, U.K. Hypnorm (0.315 mg/ml fentanyl citrate and 10 mg/ml fluanisone) was from Janssen Pharmaceutical, Grove, Oxford, U.K. Hypnovel (5 mg/ml midazolam hydrochloride) was from Roche Products, Welwyn Garden City, U.K. ¹¹¹InCl₃ (10 mCi/ml in pyrogen-free 0.04 N hydrochloric acid), ¹²⁵I-human serum albumin (HSA; 20 mg of albumin per ml of sterile isotonic saline, 50 μCi/ml), [α-³²P]dCTP, Multiprime DNA labeling system, and Hybond N hybridization transfer membranes were from Amersham International, Amersham, U.K. BSA (<0.1 ng of endotoxin per mg), 2-mercaptopyridine N-oxide, control mAb MOPC-21 (mouse myeloma IgG1), platelet-activating factor (PAF), glycogen, and Freund’s complete and Freund’s incomplete adjuvants were from Sigma Chemical, Poole, Dorset, U.K. Sterile HBSS, HEPES, Tyrode’s salt solution, and TRIzol were from Life Technologies, Paisley, U.K. Percoll and protein A-Sepharose were from Pharmacia Fine Chemicals, Uppsala, Sweden. Pyrogen- and preservative-free heparin sodium (5000 U/ml) was from Pabyrn Laboratories, Greenford, U.K. Restriction enzymes and RNA m.w. markers were from Promega, Southampton, U.K. Fura-2 acetoxymethyl ester was from Cambridge Bioscience, Cambridge, U.K. Leukotriene B₄ (LTB₄) was from Cascade Biochemicals, Reading, U.K. Recombinant human TNF-α was from Biogen, Cambridge, MA. Recombinant rat IL-4 was from Serotec, Oxford, U.K. Synthetic human eotaxin (9) was from Leukosite, Cambridge, MA.

Anti-rat β₂ mAb WT.3 (mouse IgG1) was from AMS Biotechnology, Oxford, U.K. and was generated as described elsewhere (32). Anti-rat ICAM-1 mAb 1A29 (mouse IgG1) was generated by immunization of mice with endothelial cells obtained from high venules of rat lymph nodes as described in detail in a previous study (33). The anti-human α₄ integrin mAb HP2/1 (mouse IgG1), recognizing rat α₄ (34), and the anti-rat VCAM-1 mAb 5F10 (mouse IgG2a) (16) were from Biogen.

Rat leukocytes were elicited and purified as previously described (15, 16). Eosinophils were used only when their purity, as determined by Kimura staining, was >94%. The predominant contaminating cell type was mononuclear and a major exclusion criterion was the presence of neutrophils. When these cells were compared by FACS analysis to eosinophils immunostained in whole blood, there was no significant difference in the expression of α₄ integrins and L-selectin (16). Mononuclear cells (>97% pure at the upper interface) were prepared from the same donors and gradients as neutrophils (>98% pure at the lower interface).

Rat eosinophils, neutrophils, and mononuclear leukocytes (10⁷ cells/ml in Ca²⁺/Mg²⁺-free PBS containing 0.1% BSA) were loaded with fura-2 acetoxymethyl ester (2.5 μM, 30 min at 37°C). After two washes, cells were resuspended at 10⁶ cells/ml in Ca²⁺/Mg²⁺-free PBS containing 10 mM HEPES/0.25% BSA/10 mM glucose, pH 7.4. Aliquots were dispensed into quartz cuvettes and the external Ca²⁺ concentration adjusted to 1 mM with CaCl₂. Changes in fluorescence were monitored at 37°C using an LS50 fluorescence spectrophotometer (Perkin-Elmer, Beaconsfield, U.K.) at excitation wavelengths 340 nm and 380 nm, and emission wavelength 510 nm. [Ca²⁺]_i levels were calculated using the ratio of the readings at the two excitation wavelengths and a K_d for Ca²⁺ binding at 37°C of 224 nM (35). Responses were monitored for 3 min, and data are expressed as maximal increase in [Ca²⁺]_i over the basal levels.

Rat eosinophils or neutrophils were radiolabeled with ¹¹¹In as previously described (15, 16). Briefly, the cells (1–2 × 10⁷) were incubated with ¹¹¹InCl₃ (approximately 100 μCi chelated with 40 μg 2-mercaptopyridine-N-oxide in 0.1 ml of 50 mM sodium phosphate, pH 7.4) for 15 min at 20°C. The labeled leukocytes were washed three times and resuspended (1 × 10⁷ cells/ml) in HBSS, pH 7.4, containing cell-free citrated rat plasma to a final concentration of 10%. The final cell suspension normally carried approximately 60% of the total radioactivity used, and 5 × 10^{6 111}In-cells were injected into each recipient rat.

Leukocyte infiltration and edema formation were simultaneously measured using the local accumulation of i.v.-injected ¹¹¹In-labeled cells and ¹²⁵I-HSA, as previously described (15). Briefly, rats were anesthetized with a mixture of Hypnorm (0.1 ml/rat) and Hypnovel (0.1 ml/rat), injected i.p., and their dorsal skin shaved. ¹¹¹In-eosinophils or ¹¹¹In-neutrophils (5 × 10⁶ cells mixed with 2.5 μCi ¹²⁵I-HSA) were then injected i.v. via a tail vein. Five minutes later, the agents under investigation were injected (100 μl/site, in Tyrode’s salt solution containing 0.1% BSA) into the back skin. At the end of a 4-h test period, or 1 h for the time course experiment, the animals were reanesthetized and a cardiac blood sample was collected for determination of cell and plasma radioactivity. The animals were then killed by an overdose of sodium pentobarbitone, the back skin removed, and the injection sites punched out with a 17-mm-diameter punch. Skin, blood cell pellet, and plasma samples were counted in a Cobra Auto Gamma counter (Canberra Packard, Pangbourne, U.K.) and counts were cross-channel corrected for the two isotopes. The ¹¹¹In count per cell was determined and used to express eosinophil or neutrophil accumulation in each skin site in terms of the number of labeled leukocytes. Edema formation at each site was expressed as microliters of plasma by dividing skin sample ¹²⁵I counts by ¹²⁵I counts in 1 μl of plasma.

After their last wash, ¹¹¹In-eosinophils were divided into four groups for pretreatment with mAbs: 1) MOPC-21 (10 μg/5 × 10⁶ cells/0.5 ml), 2) HP2/1 (5 μg/5 × 10⁶ cells), 3) WT.3 (5 μg/5 × 10⁶ cells), and 4) WT.3 + HP2/1 (each at 5 μg/5 × 10⁶ cells). These mAb concentrations were determined by indirect immunostaining and FACS analysis to be saturating. After incubation for 20 min at 20°C, the cells were injected i.v. into recipient animals.

Rabbits were immunized with human synthetic eotaxin (0.2 mg/animal) emulsified in Freund’s complete adjuvant. Booster injections in incomplete adjuvant were given at 3 and 6 wk and the animals bled out 12 days after the last boost. The IgG fraction of the antiserum was purified by binding to protein A-Sepharose; elution with 0.1 M glycine-HCl, pH 3.0; collection into sodium phosphate, pH 7.4, to limit exposure to acid; and dialysis against saline. The IgG was adjusted to the original volume of serum, filtered, and stored at −20°C in aliquots.

A full-length rat eotaxin cDNA of 980 bp, isolated from an inflamed lung cDNA library (11), was excised from the vector pBKCMV (Stratagene, Cambridge, U.K.) by restriction digestion with BamHI and XhoI. To control for RNA loading, a full-length rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA of 1272 bp was used.

IL-4 (5000 U/100 μl) and Tyrode’s/0.1% BSA vehicle were injected into the skin of three rats 24, 18, 8, and 4 h before killing with an anesthetic overdose. The skin sites were quickly punched out and frozen at −80°C. Total RNA was extracted using TRIzol according to the manufacturer’s instructions. RNA samples (10 μg) were separated on 1% agarose gels, containing 2.2% formaldehyde, transblotted onto nylon membranes, and fixed by UV irradiation. Membranes were hybridized with the rat eotaxin cDNA probe (labeled using [α-³²P]dCTP and Multiprime kits). Membranes were washed at a final stringency of 0.1× SSC, 0.1% SDS, 55°C (2 × 20 min) and autoradiographed at −70°C. The membranes were stripped, rehybridized using the ³²P-labeled rat GAPDH cDNA probe, and washed at the same high stringency. Autoradiographs were assessed by laser densitometry. Data are expressed as the ratio of eotaxin to GAPDH mRNA: the sample in each rat with the highest ratio being given an arbitrary value of one and other samples expressed relative to this.

Results are expressed as the mean ± SEM for n animals or cell preparations. For the in vivo experiments, each datum unit is the average of responses in duplicate skin sites. Data were analyzed by two-way ANOVA of log-transformed data and statistical significance determined with the Newman-Keuls procedure for repeated comparisons. A p value <0.05 was considered statistically significant.

Human eotaxin induced a rapid and dose-dependent increase in [Ca²⁺]_i in rat eosinophils with a significant effect at 1, 10, and 30 nM, the maximal effect being at 10 nM (Fig. 1). Responses peaked within 30 s and returned to basal levels (76.1 ± 2.2 nM Ca²⁺_i) by approximately 2 min. The nonselective chemoattractant, PAF, used as a positive control, caused a similar increase in [Ca²⁺]_i when assayed at 10 nM (Fig. 1).

In contrast to the results with eosinophils, human eotaxin had no significant effect on the [Ca²⁺]_i of rat neutrophils and mononuclear cells. The positive control, 10 nM PAF, induced large and significant responses in both cell types. The [Ca²⁺]_i for neutrophils were: basal (88.4 ± 4.3 nM, n = 4 cell preparations), 10 nM eotaxin (6.9 ± 4.1 nM increase, n = 4), and 10 nM PAF (425.2 ± 65.8 nM increase, n = 4, p < 0.01). The results for mononuclear cells were: basal (203.1 ± 20.9 nM, n = 4), 10 nM eotaxin (5.3 ± 1.4 nM increase, n = 4), and 10 nM PAF (500.7 ± 168.6 nM increase, n = 4, p < 0.01).

Using a 4-h in vivo time period, intradermal (i.d.) administration of human eotaxin caused a dose-dependent accumulation of ¹¹¹In-labeled eosinophils in rat skin with significant responses being achieved at 10 and 100 pmol/skin site (Fig. 2,A). Intradermal PAF at 100 pmol/skin site also induced a significant eosinophil accumulation (Fig. 2,A). In contrast, human eotaxin did not induce any edema formation (Fig. 2 B) or ¹¹¹In-neutrophil accumulation (data not shown) while PAF (100 pmol/site, used as a positive control) induced significant responses in each case. Histologic examination of the skin injected with human eotaxin (100 pmol/site) demonstrated the presence of an eosinophil infiltrate, with no neutrophil infiltrate, after a 4-h in vivo test period in animals not injected i.v. with labeled cells (data not shown).

Human eotaxin at a dose of 10 pmol/skin site or Tyrode’s/BSA vehicle was injected i.d. into different skin sites 240, 120, 60, 30, and 0 min before the i.v. injection of ¹¹¹In-eosinophils. After a 1-h in vivo test period, the animals were killed and the responses quantified. Significant cell accumulation in response to the chemokine was detected within the first 2 h (i.e., at 0–60, 30–90, and 60–120 min) but not at later time points (Fig. 3).

To study the adhesion pathways involved in eotaxin-induced eosinophil accumulation in vivo, the effects of mAbs directed against different adhesion molecules were determined. Abs were i.v. administered, 15 min before injecting the labeled cells, at 3.5 mg/kg for anti-α₄ integrin mAb (HP2/1) and 5 mg/kg for anti-ICAM-1 mAb (1A29), anti-VCAM-1 mAb (5F10), and the control mAb (MOPC-21). In a previous study, these i.v. doses of mAbs were found to be sufficient to give the maximum attainable inhibitory effects on TNF-α-induced eosinophil accumulation (16). Once the cells were injected, human eotaxin at 10 and 100 pmol/skin site was i.d. administered and responses allowed to develop for 4 h. Anti-ICAM-1 mAb significantly reduced the eosinophil accumulation elicited by eotaxin (Fig. 4,A), the response to the top dose being reduced by 73%. Similarly, the anti-VCAM-1 and anti-α₄ integrin mAbs also significantly inhibited the eosinophil accumulation induced by human eotaxin; at 100 pmol eotaxin/skin site, responses were reduced by 43 and 67%, respectively (Fig. 4, B and C). These inhibitory effects could not be explained by a reduction in the circulating ¹¹¹In-eosinophils numbers, as the anti-α₄ integrin mAb had no significant effect on the circulating numbers, while the anti-ICAM-1 and the anti-VCAM-1 mAbs increased circulating ¹¹¹In-eosinophils (by 46 and 61%, respectively).

Anti-α₄ integrin mAb HP2/1 (16) and anti-β₂ integrin mAb WT.3 (data not shown) bind to rat eosinophils. The effect of in vitro pretreatment of ¹¹¹In-eosinophils with these mAbs, before i.v. injection of the cells, on their accumulation in eotaxin-injected skin sites is shown in Figure 5. Anti-α₄ integrin and anti-β₂ integrin mAbs significantly inhibited ¹¹¹In-eosinophil accumulation in response to eotaxin by 52 and 49%, respectively. Although the combination of both Abs appeared to cause a larger reduction in the chemokine-elicited responses (68%), this effect was not significantly different from the effects of either of the Abs alone.

Northern blot analysis showed a time-dependent eotaxin mRNA expression at sites of IL-4 injection (5000 U/skin site). Eotaxin message was virtually undetectable in vehicle-injected sites but was increased, beginning at 4 h and reaching a peak at approximately 18 h, after IL-4 injection (Fig. 6).

To determine whether eosinophil accumulation induced by IL-4 in rat skin is mediated by the generation of endogenous eotaxin, we used a polyclonal antiserum raised against human eotaxin. The IgG fraction of this rabbit antiserum, coinjected with i.d. agonist, inhibited eosinophil accumulation induced by human eotaxin but not that induced by LTB₄ or TNF-α (Fig. 7). IL-4 (5000 U/skin site) was coinjected with control rabbit IgG or anti-human eotaxin IgG at two different time points: −24 h and 0 h, each in the same animal so that each rat acted as its own internal control. Labeled cells were then injected i.v. and responses allowed to develop for 4 h. Anti-human eotaxin IgG reduced the ¹¹¹In-eosinophil accumulation induced by IL-4 injected at −24 h (i.e., the 24- to 28-h response) in all rats tested (n = 8, p < 0.05) but had no effect on the 0- to 4-h eosinophil accumulation response to IL-4 (Fig. 8).

Chemokines comprise a family of small m.w proteins (8–10 kDa) that have been implicated in the selective recruitment of leukocyte subsets observed in different inflammatory disease states. Among all the CC chemokines described to date, eotaxin is the only member that acts on a single receptor, CCR3. This receptor is present in high numbers on eosinophils, and the high affinity interaction can explain the high specificity of eotaxin for eosinophils (3, 4, 8, 9, 10, 36). Although human eotaxin has been tested extensively in vitro, little is known of its activity in vivo.

To characterize the eosinophil accumulation elicited by human eotaxin, we have used an in vivo assay system to measure ¹¹¹In-eosinophil accumulation and edema formation in rat skin (15, 16). Preliminary experiments showed that ¹²⁵I-human eotaxin bound to rat eosinophils in vitro (data not shown). This binding caused a rapid and dose-dependent increase in intracellular calcium levels (Fig. 1). In contrast, human eotaxin failed to elicit changes in intracellular calcium concentrations in either rat neutrophils or rat mononuclear cells. This demonstrates the eosinophil selectivity of eotaxin in the rat, in agreement with previous in vitro studies in other species (8, 9, 10, 36). Next, we demonstrated that human eotaxin elicits a potent and selective (Fig. 2) eosinophil accumulation in rat skin; no neutrophil accumulation or edema formation was seen. The eosinophil accumulation was a rapid event, significant responses occurring within the first hour (Fig. 3). Taken together with the in vitro data on calcium mobilization in isolated cells, this rapid eosinophil accumulation suggests a direct effect of eotaxin on eosinophils in vivo. Chemokines such as eotaxin are thought to be presented on the microvascular endothelium and stimulate adhesion (21).

The adhesive mechanisms that regulate chemokine-induced accumulation of leukocytes in vivo remain largely uncharacterized. As part of the present study, we have investigated the roles of the β₂ integrins/ICAM-1 and α₄ integrins/VCAM-1 adhesion pathways in the process of eosinophil accumulation elicited by eotaxin (Figs. 4 and 5). Pretreatment of ¹¹¹In-eosinophils in vitro with a saturating concentration of a neutralizing anti-β₂ integrin mAb, before their administration into the recipient animals, significantly suppressed the eotaxin-induced ¹¹¹In-eosinophil accumulation. Further, the i.v. administration of a neutralizing anti-ICAM-1 mAb also caused an inhibition of this response. These results suggest that eotaxin-induced leukocyte accumulation is at least partly mediated via the direct effects of the chemokine on the leukocytes causing up-regulation or activation of β₂ integrins in vivo. β₂ integrins can then mediate the firm attachment of the eosinophils to venular endothelial cells via basally expressed ICAM-1. Previous studies have demonstrated the importance of the β₂ integrins or ICAM-1 in eosinophil accumulation in other models of inflammation (37, 38, 39, 40). We have found that eotaxin up-regulates the expression of β₂ integrins on rat peripheral blood eosinophils in vitro (K. Nagai et al., unpublished observations). If this up-regulation also occurs in vivo, the newly expressed β₂ integrins will not be blocked by the in vitro pretreatment of the eosinophils with mAb. This would explain why we found that blocking the existing β₂ integrins in vitro (Fig. 5) was less effective than blocking the ICAM-1 counterligand in vivo (Fig. 4).

Neutralizing mAbs directed against α₄ integrins or VCAM-1 also suppressed the eosinophil accumulation induced by i.d. eotaxin. The anti-α₄ integrin mAb was effective at suppressing this response whether it was used to pretreat the ¹¹¹In-eosinophils in vitro or given i.v. These results suggest that, in addition to activating β₂ integrins, the eotaxin-induced eosinophil accumulation may be mediated via α₄ integrins although the mechanism by which eotaxin may activate α₄ integrins is still unknown. VCAM-1, one of the principal endothelial cell ligands for α₄ integrins, is generally considered to be expressed only on cytokine-activated endothelial cells (22). However, as an anti-VCAM-1 mAb was effective at attenuating the rapid eosinophil accumulation induced by eotaxin, our results suggest the presence of low levels (that may be undetectable by immunostaining) of basally expressed VCAM-1 on venular endothelial cells that are sufficient to mediate the accumulation of eosinophils. These results are in agreement with our previous findings in which we reported that an anti-α₄ integrin mAb could inhibit the eosinophil accumulation induced by rapidly acting chemoattractants such as LTB₄ and C5a^desArg (27).

The pretreatment of ¹¹¹In-eosinophils with a combination of anti-α₄ integrin and anti-β₂ integrin mAbs did not cause a significantly larger level of inhibition than that seen with either mAb alone. This differs from the additive effect seen with two similar Abs, which completely inhibited the eosinophil accumulation induced in guinea pig airways in response to i.v. Sephadex particles (29). It is possible that α₄ and β₂ integrins may be involved in different but sequential stages of the eotaxin-induced eosinophil migration, namely rolling and firm adhesion, respectively. In this respect, there is in vivo evidence for the involvement of α₄ integrins in eosinophil rolling (19), as well as in the subsequent phase of firm adhesion, whereas β₂ integrins are not thought to contribute to eosinophil rolling.

Recent in vitro studies have addressed the mechanisms involved in chemokine-mediated leukocyte adhesion and transendothelial cell migration. Eosinophil transmigration across unstimulated endothelial cell monolayers induced by RANTES can be virtually abolished by mAbs directed against β₂ integrins (41). Although this effect was reported to occur without an apparent increase in β₂ integrin expression on the leukocyte surface, recent reports have demonstrated that RANTES, monocyte chemoattractant protein (MCP)-3, and eotaxin can stimulate the up-regulation of α_Mβ₂ integrin (CD11b/CD18) on human eosinophils (13, 42). In addition, RANTES and MCP-3 can trigger a strong and prolonged adhesion of human eosinophils to ICAM-1, this effect being primarily mediated by direct activation of β₂ integrins (42). Eosinophils can also adhere to either VCAM-1 or fibronectin upon stimulation with chemokines (25, 26, 42, 43). This event was shown to be rapid and transient, and occurred without apparent changes in α₄β₁ integrin expression on the leukocyte surface (26, 42), although it was correlated with f-actin polymerization (26, 42), uropod-like formation, and asymmetrical distribution of very late activation Ag-4 (26). Hence, the above in vitro studies largely support the present in vivo findings and suggest that the eotaxin-induced eosinophil migration may be mediated by changes in the expression, affinity states, or signaling pathways of α₄ and β₂ integrins.

IL-4 is a Th2 cytokine that selectively induces VCAM-1 expression on endothelial cells with little or no effect on ICAM-1 or E-selectin expression (44, 45, 46). This feature of IL-4 has been associated with the selective adhesion and transmigration of eosinophils, as opposed to neutrophils (47, 48, 49, 50), across the vascular endothelium mediated by α₄ integrins on the eosinophil surface (22). In vivo, i.p. or i.d. injection of IL-4 induces selective eosinophil accumulation in nude mice and rats (49, 51). In agreement with in vitro studies, the IL-4-induced eosinophil accumulation in vivo appears to be α₄ integrin/VCAM-1 dependent (51). Besides inducing VCAM-1 expression, IL-4 can exert eosinophil accumulation through the endogenous generation of secondary inflammatory mediators. In other systems, IL-4 has been implicated in the release of stem cell factor from alveolar macrophages (52), in the increased expression of C10 in murine resident peritoneal macrophages (53), and in the synthesis and secretion of MCP-1 by cultured HUVECs (54).

Eotaxin message has been detected after transplantation of IL-4-secreting tumor cells in mice, an effect that occurs within 6 h and persists for 7 days (7). We conducted a series of experiments to detect the possible involvement of eotaxin generation in eosinophil accumulation induced by i.d.-injected IL-4 in the rat. Firstly, Northern blot analysis showed that eotaxin message began to increase at 4 h in IL-4-injected sites and was still present at 24 h (Fig. 6). Secondly, we used a neutralizing anti-human eotaxin Ab to block part of the response to IL-4. ¹¹¹In-eosinophil accumulation observed 24 to 28 h after, but not 0 to 4 h after, IL-4 injection was significantly suppressed when the cytokine was coinjected with the Ab (Fig. 8). These effects of anti-eotaxin are consistent with the mRNA data, which suggest the likely presence of eotaxin protein during the 24- to 28-h ¹¹¹In-eosinophil accumulation period but little or no eotaxin release during the 0- to 4-h period. Hence, we conclude that endogenous eotaxin generation partly mediates the later phases of eosinophil accumulation induced by IL-4. This response to eotaxin may, in turn, be dependent on the up-regulation of VCAM-1 on the endothelial cell surface in response to IL-4, as discussed above.

Th2 cytokines appear to play a key role in regulating eotaxin-induced eosinophil accumulation, since IL-5 has been shown to cooperate with eotaxin to induce eosinophil infiltration in the skin and in the lungs by mobilizing eosinophils from the bone marrow (10, 14, 55), and eosinophil infiltration induced by IL-4 can be partly mediated through eotaxin generation, as discussed above. Indeed, in two mouse models of Ag-induced pulmonary eosinophilia, it has recently been shown that elimination of T cell accumulation results in a reduction of eosinophil recruitment associated with eotaxin down-regulation (56, 57).

In conclusion, we have shown that human eotaxin is a potent and selective inducer of eosinophil accumulation in rat skin and provide the first in vivo evidence that eosinophil accumulation induced by chemokines can be abrogated by mAbs directed against both leukocyte and endothelial cell surface adhesion molecules. We have also demonstrated that the secondary generation of endogenous eotaxin is involved in the later phases of eosinophil accumulation in response to IL-4.