Tissue eosinophilia induced by recombinant human interleukin-5 in the hamster cheek pouch membrane

Interleukin-5 (IL-5) is a cytokine that preferentially effects the development and function of eosinophils, and is considered important in the pathophysiology of allergic inflammation. In this study, we evaluated the ability of recombinant human IL-5 (rHu IL-5) to promote tissue eosinophilia and the importance of this eosinophilia to pathological alterations in vascular function. Repetitive subcutaneous administration for 18 days of rHu IL-5 resulted in a 7-fold increase in the number of eosinophils found in the ipsilateral hamster cheek pouch membrane. The contralateral cheek pouch membrane and peritoneum of these animals showed lesser but significant elevations in the number of eosinophils. In contrast, denatured rHu IL-5 did not elevate eosinophils in these tissues. Through the use of intravital microscopy and fluorometric analysis, rHu IL-5 treated hamster cheek pouch membranes were evaluated for alterations in microvascular permeability, using plasma clearance of FITC-dextran 150 as an index. Despite promoting a prominent tissue eosinophilia, the repetitive subcutaneous injections of rHu IL-5 did not alter the clearance of FITC-dextran 150. Topical application of rHu IL-5 to the cheek pouch, also, had no effect on the clearance of FITC-dextran 150. Immunofluorescence observations using an antibody to the granule protein, eosinophil peroxidase, indicated that the recruited cells had not degranulated. Our results support the importance of IL-5 in the recruitment of tissue eosinophils, but further stimulation is probably required to cause degranulation of these cells and the ensuing tissue damage.


Introduction
Elevated numbers of eosinophils in the blood and tissues is a common feature associated with allergic reactions. [1][2][3] Through the generation of inflammatory lipid mediators, like platelet activating factor (PAF) or leukotrienes, and the release of biologically active granule proteins, the eosinophil leukocyte has been documented as a potent effector cell capable of contributing to the pathophysiology of allergic disorders.4-C-In support of this hypothesis, we and others have shown that eosinophil granule proteins can cause increased vascular macromolecular permeability, which may lead to oedema. 7'a Cytokines such as granulocyte-macrophage colony stimulating factor (GM-CSF) and the interleukins (IL) 3 and 5 have been identified as agents that may enhance eosinophil function ,in 91o allergic inflammatory reactions.' Recent experiments have documented that IL-5 is the pre-dominant cytokine that can cause the preferential proliferation, differentiation, and tissue recruitment of eosinophil leukocytes. 11 '12 Furthermore, IL-5 affects the activation, chemotatic responses, degranulation, and decreased apoptosis of per- [13][14][15] ipheral blood derived eosinophils.
Because IL-5 possesses biological activity across species, we used recombinant human IL-5 to induce eosinophilia in the Syrian hamster. '7 In order to investigate whether cytokine-induced tissue eosinophilia promotes oedema, we used the plasma clearance of fluoroscein isothiocyanate (FITC)-dextran 150 in the hamster cheek pouch preparation as a measure of changes in vascular permeability.

Materials and Methods
The male golden Syrian hamsters (80-100g) used for these experiments were cared for in accordance with the guidelines of the Animal (  . Animals phil influx into the cheek pouch membrane, were permitted unrestrained access to food and hamsters received repetitive injections of 0.01 mg water, of denatured or native rHu IL-5 in 0.1 ml of saline (Schering-Plough Research Institute, Kenilworth, Enumeration of eosinophils: In the studies NJ). The It-5 was denatured by boiling a lmg/ml designed to measure eosinophil infiltration into solution for 5 min, without loss of soluble the cheek pouch, hamsters were euthanized by protein. These injections were given on alternate COg asphyxiation. The oral surfaces of both days with a maximum of three injections per cheek pouches were rinsed with normal saline, week. This resulted in a total of seven injections the membranes were surgically removed, stret-of rHu IL-5 for the 18-day period or four injecched and stapled to a white index card. Mounted tions for the 9-day paradigm. Injections were membranes were fixed in phosphate buffered given subcutaneously to one cheek. Particular formalin (Fisher, Fairlawn, NJ). Using a number 6 care was taken to avoid puncturing the underear-punch, sections of cheek pouch membranes lying cheek pouch membrane. Treated animals were removed and stained for eosinophils using were then returned to their cages. the method of Duffy eta/. 18 Regions distant from blood vessels were chosen for examination using Macromolecular clearance and intravital micro-100 x magnification. From each tissue (n 3) scopy: To assess the integrity of the microcirculaper treatment group, at least five random fields tion following various treatments, separate sets of were chosen using 400 x magnification and the hamsters were anaesthetized with sodium pentonumber of eosinophils measured. The data per barbital (60mg/kg, i.p.) and tracheotomy Was treatment group were compiled and expressed performed. The fight jugular vein was cannulated as the mean number of eosinophils/high power for the administration of the fluorochrome tracer field (#/HPF; n 15). and supplemental doses of i.v. anaesthetic. The Eosinophil recruitment and degranulation in fight carotid artery was cannulated for collection the cheek pouch membrane was assessed using of arterial blood samples. The cheek pouch was an immunofluorescence technique. Membrane prepared according to previously published tissue samples were cut into 5 btm sections and methods. 8'21'22 Briefly, a two-piece Lucite subsequently stained with rabbit anti-human chamber with a 1-ml reservoir capacity was eosinophil peroxidase (anti-EPO; unpurified anti-attached to a single layer of the pouch, delineatserum diluted 1:20). This antiserum was develing a 2.3cm 2 area for intravital microscopy. The oped in our laboratory using purified human chamber reservoir was filled with bicarbonate eosinophil peroxidase as described previously. 19  peritoneal lavage. No reactivity of these tissues Throughout the surgery and experimental protowith normal rabbit serum was found (data not col, animals were kept on a heating pad to mainshown). Indirect fluorescence microscopy and tain body temperature at 37C. photography was performed using an affinity Macromolecular clearance from the cheek purified FITC-labelled goat anti-rabbit IgG pouch was measured using the plasma clearance (Southern Biotechnology Associates, Birmingham of FITC-dextran (MW 150000; Sigma Chemical AL). Samples were then stained in haematoxylin Co., St Louis; FITC-dx 150) as described preand eosin for leukocyte identification. 2 Preparaviously. '2 After surgical preparation, the hamster tion and analysis were performed without prior was positioned on a lucite board mounted on knowledge of the treatment applied, the stage of a Nikon Optiphot microscope. A 1 h To obtain peritoneal eosinophils, the perito-stabilization period followed the surgery, during neum was lavaged with 10ml of normal saline which the pouch was continuously suffused with and the accompanying cells were recovered using bicarbonate buffer (1 ml/min; 37C). a Jelco Teflon catheter (Critikon, Tampa, FL).
FITC-dx 150 was injected as a bolus (100mg/ Cells were centrifuged at 200 x g for 15 min at kg, i.v.) 30 min into the stabilization period. This 5C. Cell pellets were then resuspended in 2 ml was followed by the continuous infusion of of phosphate buffered saline (Gibco, Grand FITC-dx 150 (0.15mg/kg/min) for the duration Island, NY) with 0.1% foetal calf serum. Slides of the study. Blood samples were collected every from cytospin preparations were visualized with 30 min from the arterial cannula. In order to determine plasma clearance values, suffusate samples and plasma levels of FITC-dx 150 were determined according to previously published methods. 8 To evaluate the possible direct effects of IL-5 on the microcirculation, rHu IL-5 (0.1 mg/ml, I ml total volume) was topically applied to separate cheek pouch preparations as described previously. 8 Briefly, topical application of IL-5 was conducted by interruption of the suf-. : fusate, aspiration of the residual bicarbonate buffer, application of the IL-5 solution by a 1 ml syringe to the chamber reservoir, and by incubation for 5 min. All solutions were warmed to 35C for 5 min prior to application. After application the IL-5 solution was aspirated, the chamber bed washed three times (3 x I ml) with bicarbonate buffer (35C) and the suffusate re- treatments, within the given day, was determined using the The probe was connected to a 100W halogen Mann-Whitney non-parametric test, with p< 0.05. lamp. Epi-illumination was provided by a 50W mercury arc lamp, with appropriate filters for fluorescence. The recording system consisted of animals treated with denatured rHu IL-5 devela CCD camera (Dage, Michigan City, IN) con-oped a marginal eosinophilia in the ipsilateral nected to a Panasonic time generator, a Sony VO cheek pouch membrane only (Fig. 1). 5858 videotape recorder, and a RCA mono-Examination of eosinophil numbers in the chrome video monitor. In order to minimize contralateral cheek pouch membranes after rHu untoward effects from fluorescence epi-illumina-IL-5 treatment revealed a small but significant tion, viewing was kept to less than 45 s. 23 increase over 18 days of therapy (Fig. 2).
However, eosinophil numbers were considerably Statistical analysis: All values presented are mean less than those in the ipsilateral cheek pouch _+ S.E.M. Statistical significance (p < 0.05) was membrane. There was no significant increase in determined using the non-parametric, one-tailed, the recruitment of eosinophils to the con-Mann-Whitney test (InstatTM, GraphPad, San tralateral membrane in denatured rHu IL-5 Diego, CA). treated animals (Fig. 2).
Using a rabbit anti-Hu EPO antiserum, enumeration of immunopositive cells confirmed the Results increase in membrane eosinophils by rHu IL-5. Compared to the number of eosinophils that Eosinophil recruitment in cheek pouch mere-were found in untreated or denatured rHu IL-5 branes and peritoneum: Hamsters receiving membranes, substantially more cells were detecrepetitive subcutaneous injections of rHu IL-5 ted in the membrane exposed to native rHu IL-5 (0.01 mg) to the right cheek showed a substantial (Fig. 3a,b,c). Cells reactive with anti-EPO were increase in the number of eosinophils in the visible as discrete fluorescent ovals.
ipsilateral pouch membrane (Fig. 1). A statisti-Repetitive subcutaneous injections of rHu IL-5 cally significant rise in the number of membrane to the cheek pouch of hamsters resulted in a eosinophils occurred by the second day. Sub-marked increase in the percentage of recoverable sequent injections of rHu IL-5 further increased peritoneal eosinophils (Fig. 4). Rising from initial the number of eosinophils and, by day 18, a 7-values of 12 + 2.3%, the percentage of peritoneal fold increase in the number of eosinophils was eosinophils increased to 21 + 0.9% after 9 days seen in treated membranes (Fig. 1). In contrast, and 34 + 2.3% after 11 days. Following a similar The suffusate clearance values of FITC-dx 150 obtained from these animals did not deviate significantly from baseline during the 60 min timecourse (Fig. 5). Furthermore, the integrated clearance value of FITC-dx 150 obtained from the chronically rHu IL-5 treated animals was not significantly elevated compared to that from cheek pouch preparations exposed to topical rHu It-5 for 5 min (Fig. 6; p > 0.05).

Discussion
In allergic inflammatory diseases, such as atopic dermatitis and asthma, eosinophils have been hypothesized to contribute to the obseeeed tissue injury. There is increasing evidence that IL-Day 5 is an important mediator for promoting eosi-  24 26 nificance (*; p < 0.05) between treatments, for the given day, vability.
Additionally, IL-5 has been docuwas determined using the Mann-Whitney non-parametric test.
mented, in vitro, to increase the production of eosinophils from bone marrow cultures, umbilical cord cells, and a number of immortal tumour pattern of administration, denatured IL-5 did not cell lines. [27][28][29] Combined, these studies suggest increase the percentage of peritoneal eosinophils that IL-5 is a late-acting specific factor for eosino- (Fig. 4).
phil bone marrow production, release, directed migration and activation.

Macromolecular clearance of FITC-dx I50 and
However, few in vivo studies have been conintravital microscopy: To evaluate the acute ducted where rlL-5 has been directly adminiseffects of rHu IL-5, the microcirculation of the tered. More often, studies investigating the in hamster cheek pouch was surgically prepared vivo role of IL-5 on eosinophils have taken the and topically challenged for 5 min with either form of administering monoclonal antibodies bicarbonate buffer or rHu IL-5. Compared to (mAbs). Through the use of mAbs, researchers bicarbonate buffer, the application of rHu IL-5 have shown that IL-5, exclusively, is important for (0.1mg/ml, l ml) to the hamster cheek pouch the eosinophilia that is associated with allergic preparation did not result in a significant lung eosinophilia, parasitic infections and episoincrease in the clearance of FITC-dx 150. Furdic angio-oedema. 2' therrnore, the topical application of rHu IL-5 did While eosinophils normally populate the not alter vessel diameters, cause muscle fascicula-hamster cheek pouch membrane, the recruitment tions, or result in demonstrable leukocyte adheand activation of large numbers of eosinophils sion over the 60 min time-course of by the cytokine IL-5 might result in the release of experimentation, any number of inflammatory agents that induce Animals were then treated with repetitive suboedema. 4 This hypothesis is supported by the cutaneous injections of rHu IL-5 (0.01 mg/animal; clinical presentation of capillary leak syndrome in three times per week) to the cheek pouch for a cancer patients undergoing IL-2 treatment. Ele-9-day time-course. During the microsurgical prevated plasma levels of IL-5 and the activation of paration, no distinguishing alterations in the eosinophils have been associated with oedema membrane or vascular bed were noted. Furtherand morbidity in these patients. 5 more, throughout the stabilization period and the Our report conclusively demonstrates that injections of rHu IL-5 in the hamster is supported by our findings that: (a) histological examination of the underlying ipsilateral pouch membrane revealed a significant increase in the number of eosinophils; (b) less profound eosinophilia occurred in the contralateral pouch membrane; and (c) recoverable cells from peritoneal lavage were enriched in the percentage of eosinophils. However, although evoking a prominent eosinophilia, the repetitive administration of IL-5 did not alter vascular permeability, leading us to conclude that besides IL-5 other factors are required to cause the activation and degranulation of eosinophils under our experimental con-

ditions.
We have previously demonstrated that sub-nanomolar concentrations of the four prominent eosinophil granule proteins: major basic protein (MBP), eosinophil cationic protein (ECP), eosinophil derived neurotoxin (EDN) and eosinophil peroxidase (EPO) have the capacity to increase microvascular permeability in the hamster cheek pouch preparation. 8 Therefore, even partial degranulation of the recruited eosinophils should have resulted in a substantial increase in the clearance of FITC-dx 150. However, our immunofluorescenc studies indicated that the recruited eosinophils had not degranulated. Based on the discrete staining pattern with anti-EPO antiserum, this granule protein remained intracellular. Because EPO is found in the matrix of the specific granule, its presence within the cell indi-  cates that the three remaining proteins-MBP, ECP and EDN were still contained there as well.
It is plausible that the IL-5 induced eosinophilia present in the membranes represents a primed system and subsequent treatment with an appropriate agonist might provoke degranulation of the recruited eosinophils. As a preliminary test of this hypothesis, we treated one of the cheek The need for IL-5 bioactivity to cause the eosinophilia was confirmed by treating with denatured rHu IL-5 and failing to find a similar pattern of eosinophilia, both locally and systemically. Nevertheless, at the site of injection, within the underlying membrane, a small but detectable increase in the number of eosinophils was noted. Because no effect was detected in the contralateral membrane or in the peritoneal cell differential, we believe that this small local increase in eosinophils is probably due to the repetitive injections of a foreign denatured protein. The possibility that repetitive needle punctures (27 gauge) may have resulted in a fibrotic wound healing condition, causing the recruitment of eosinophils cannot be excluded. 39 However, we believe this to be an unlikely explanation for our observations. In the wound healing models of tissue eosinophilia, the cutaneous wounds are substantially larger than that provoked by our 39 40 needle punctures.
Furthermore, in our model we are studying the underlying oral membrane after the removal of the overlying, loose connective aerolar tissue.
The capacity of IL-5 to direct selective eosinophil chemotaxis has been reported from in vitro 41 42 studies.
Recent experiments also suggest that local production of IL-5 is responsible for eosinophil recruitment after antigen challenge in allergic individuals. 4 The cellular signaling for eosinophil chemotaxis is through the binding of IL-5 to specific surface receptors on the mature eosinophil leukocyte. 44 In our experimental model, the selected and directed recruitment of the eosinophils might be a result of the cells following a gradient in the administered IL-5, as the molecule was carried out into the systemic circulation. This hypothesis would also partially explain the lack of tissue accumulation of the eosinophils in the IL-5 transgenic animals. In this latter case, IL-5 is constitutively produced by all the T cells of the animal. Therefore, there is no directed tissue recruitment of eosinophils, as there is no specific gradient in IL-5.
In summary, our experiments document that repetitive exposure to rHu IL-5 can selectively recruit eosinophils to the hamster cheek pouch membrane. However, eosinophilia per se does not necessarily result in pathological alterations of vascular functions. This lack of pathologic sequelae in response to eosinophilia is consistent with findings in transgenic mice where eosinophilia is not, in general, associated with pathophysiological consequences. 45 However, it is in contrast with the observations of eosinophilia and IL-5 in the IL-2 induced capillary leak syndrome. 35 Because the hamster cheek pouch membrane microcirculation offers an accessible vascular network that has been intensely studied and responds to eosinophil granule proteins, it may be useful for establishing the parameters needed to elicit tissue damage by enhanced eosinophil degranulation.