IE

AF

Baultn(ciasowIsaean1dEmiCIrr

Twdtopaatttsm

GASTROENTEROLOGY 2003;125:1419–1427

ntratracheal IL-13 Induces Eosinophilic Esophagitis by an IL-5,otaxin-1, and STAT6-Dependent Mechanism

NIL MISHRA and MARC E. ROTHENBERGrom the Division of Allergy and Immunology, Department of Pediatrics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio

ogabomveaiacmgpaRetlt

tt-flaothtpah

Bb

ackground & Aims: Eosinophil infiltration into the esoph-gus occurs in a wide range of diseases; however, thenderlying pathophysiologic mechanisms involved areargely unknown. We have previously reported that simul-aneous delivery of allergen to the lung and gastrointesti-al tract induces experimental eosinophilic esophagitisEE). We aimed to determine whether delivery of a Th2ytokine (interleukin [IL]-13) to the lung was sufficient fornduction of EE. Methods: IL-13 was delivered intratrache-lly to wild-type, signal transducer and activator of tran-cription (STAT) 6, eotaxin-1, or IL-5–deficient mice. Eosin-phil levels and 5�-bromodeoxyuridine (BrdU) incorporationere examined by immunohistochemical staining.Results:ntratracheal delivery of IL-13 induced dose-dependent eo-inophil accumulation in the esophagus (but not the stom-ch). In addition, intratracheal IL-13 induced esophagealpithelial hyperplasia. The ability of IL-13 to induce EE wasbolished in STAT6-deficient mice. IL-13-induced EE wasearly completely ablated in IL-5–deficient mice (37.3 �1.6 vs. 3.3� 3.2 eosinophils/mm2 in wild-type and IL-5–eficient mice, respectively). Additionally, IL-13–inducedE was significantly diminished in eotaxin-1–deficientice (48.7 � 10.3 vs. 14.1 � 12.5 eosinophils/mm2

n wild-type and eotaxin-1–deficient mice, respectively).onclusions: IL-13 delivery to the lung induces EE by anL-5, eotaxin-1, and STAT6-dependent mechanism. Theseesults further establish an intimate connection betweenespiratory and esophageal inflammation.

he accumulation of eosinophils in the esophagus is acommonly observed medical problem in patients

ith diverse diseases, including gastroesophageal refluxisease, eosinophilic esophagitis (EE), eosinophilic gas-roenteritis, and parasitic infections.1–9 Esophageal eosin-philia is associated with allergic responses; for example,atients with EE have a high rate of atopy (and asthma),nd their clinical symptoms and eosinophilic infiltrationsre ameliorated by an elemental diet or antiinflammatoryherapy (cromoglycate or glucocorticoids).10–12 Althoughhe role of allergens in the induction of eosinophilia inhe esophagus has been debated,13 recent experimentaltudies have established a link. Exposure of anesthetizedice to repeated challenges of aeroallergen (e.g., extracts

f Aspergillus fumigatus) using a protocol to induce aller-ic airway inflammation promotes marked EE.14 Thellergen-induced esophageal eosinophilia is accompaniedy intraepithelial eosinophils, extracellular granule dep-sition, and epithelial cell hyperplasia, features thatimic the pathophysiologic changes observed in indi-

iduals with various forms of EE.13–15 Importantly, theosinophilic inflammation occurs in the lungs and esoph-gus, but not the stomach or intestine, demonstrating anntimate immunologic connection between Th2-associ-ted allergic responses in the lung and esophagus. Of theytokines produced by Th2 cells, interleukin (IL)-5 is theost specific for eosinophils. IL-5 induces eosinophil

rowth, differentiation, activation, and survival andrimes eosinophils to respond to chemoattractants suchs eotaxin, an eosinophil selective CC chemokine.16–18

ecent experiments have demonstrated that IL-5 is over-xpressed in the esophagus of patients with EE.19 Addi-ionally, systemic overexpression of IL-5 (via pharmaco-ogic or transgenic approaches) promotes eosinophilrafficking to the esophagus in mice.20

In addition to producing IL-5, Th2 cells are thoughto induce eosinophil-associated inflammation throughhe secretion of an array of cytokines (IL-4, -5, -6, -9 -10,13, -25) that activate inflammatory and residential ef-ector pathways both directly and indirectly. In particu-ar, IL-4 and IL-13 are produced at elevated levels in thesthmatic lung and are thought to be central regulatorsf many of the hallmark features of disease. IL-13 appearso be particularly important because it is produced inigh quantities by Th2-cells and regulates multiple fea-ures of allergic disease (IgE production, mucus overroduction, eosinophil recruitment and survival, andirway hyperreactivity and the expression of CD23, ad-esion molecules, and chemokines [e.g., eotaxin]).21,22

Abbreviations used in this paper: BALF, bronchoalveolar lavage fluid;rdU, bromodeoxyuridine; EE, eosinophilic esophagitis; MBP, majorasic protein; STAT, signal transducer and activator of transcription.

© 2003 by the American Gastroenterological Association0016-5085/03/$30.00

doi:10.1053/S0016-5085(03)01352-0

Dysregulation of IL-13 has been reported in a variety ofallergic conditions including asthma,23 atopic dermati-tis,24–26 and allergic rhinitis.27–29 Overexpression of IL-13, by pharmacologic administration or transgenic ap-proaches, induces multiple features of asthma, includingeosinophilia, mucus overproduction, and airway hyper-responsiveness.30,31 Importantly, neutralization of IL-13activity by treatment with a soluble form of the IL-13receptor � 2 chain (IL-13R�2) reverses many of thecharacteristic features of experimental asthma.32,33

Based on the importance of IL-13 in asthma and thehigh concordance between asthma and EE in humans andin murine models of asthma,5,7,34 we hypothesized thatoverexpression of IL-13 in the lung may be associatedwith the development of EE. To test this hypothesis, wedelivered IL-13 to the lung by intratracheal administra-tion. Our results establish that pulmonary inflammation,triggered by IL-13, is associated with the development ofEE. Furthermore, we demonstrate that IL-13–inducedEE is primarily dependent on IL-5 and signal transducerand activator of transcription (STAT) 6 and, to a lesserextent, eotaxin-1.

Materials and MethodsMice

Specific pathogen-free Balb/c mice (8–12 weeks old)and STAT6-deficient mice were obtained from the JacksonLaboratory (Bar Harbor, ME). Eotaxin-1–deficient inbred miceof the Balb/c background were maintained with age- andsex-matched controls as described.35 Mice deficient in IL-5 (ona Balb/c background) were originally obtained from KlausMatthaei (Australian National University, Canberra, Australia)and have been previously described.36

Intratracheal Cytokine Delivery to the Mice

Mice (22–25 gm) were anesthetized by IP injection of500 �g of Ketaject (Ketamine HCL; Phoenix Pharmaceutical,Inc., St. Joseph, MO). Anesthetized mice were hung upright ata 60-degree angle on a vertical platform. Using flat forceps, thetongue was gently extended, and a long-loading pipette tipwas directly inserted into the trachea of anesthetized mice,followed by the delivery of 20 �L recombinant murine IL-13(generated in a HEK cell line [a generous gift of Dr. DebraDonaldson, Wyeth Research, Cambridge, MA]), IL-4, IL-9,IL-10 (R&D Systems, Minneapolis, MN), or saline. IL-4 wasadministered complexed to anti–IL-4 (these reagents werekindly provided by Fred Finkelman, University of Cincinnati),using previously reported methods because this increases the invivo half-life of IL-4.37 In some experiments, the PBS diluentwas spiked with methylene blue, and the amount of the dye inthe lung tissue homogenate was determined by optical densityat 550 nm.

Eosinophil Analysis in the Esophagus

The entire esophagus from the proximal to the distalend of adult mice was fixed in 4% paraformaldehyde in phos-phate buffer, pH 7.4, embedded in paraffin, cut into 5-�msections, fixed to positive charged slides, and immunostainedwith antiserum against mouse eosinophil major basic protein(anti-MBP), a kind gift of Drs. James and Nancy Lee (MayoClinic, Scottsdale, AZ), as described.36,38 In brief, endogenousperoxidase in the tissues was quenched with 0.3% hydrogenperoxide in methanol followed by nonspecific protein blockingwith normal goat serum. Tissue sections were then incubatedwith rabbit anti-MBP (1:16,000) overnight at 4°C, followedby 1:200 dilution of biotinylated goat anti-rabbit IgG second-ary antibody and avidin-peroxidase complex (Vector Laborato-ries, Burlingame, CA) for 30 minutes each. These slides werefurther developed with nickel diaminobenzidine-cobalt chlo-ride solution to form a black precipitate and counterstainedwith nuclear fast red. Negative controls include replacing theprimary antibody with normal rabbit serum to check endog-enous biotin and peroxidase activity. Quantification of theimmunoreactive cells was carried out by using a video-assistantintegrated computer software program (Image Pro Plus 4.1;Media Cytometrics, Silver Spring, MD). The eosinophil levelsare expressed as cells/mm2, following quantifications of at least4–6 fields per mouse.

Bronchoalveolar Lavage Fluid Collectionand Analysis

Mice were killed by CO2 inhalation. Immediatelythereafter, a midline neck incision was made, and the tracheawas cannulated. The lungs were lavaged 3 times with 1.0 mLPBS containing 1% FCS and 0.5 mmol/L EDTA. The recov-ered bronchoalveolar lavage fluid (BALF) was centrifuged at400g for 5 minutes at 4°C and resuspended in 200 �L PBScontaining 1% FCS and 0.5 mmol/L EDTA. Lysis of red bloodcells was carried out utilizing RBC lysis buffer (Sigma, St.Louis, MO) according to the manufacturer’s recommendations.Total cell numbers were counted with a hemacytometer. Cy-tospin preparations of 5 � 104 cells were stained with Giemsa-Diff-Quick (Dade Diagnostics of P.R., Inc., Aguada, PR), anddifferential cell counts were determined.

Analysis of Epithelial Cell Proliferation

To determine the degree of epithelial cell proliferation,5�-bromodeoxyuridine (BrdU) (Zymed Laboratories, San Fran-cisco, CA) incorporation analysis was performed according topreviously reported methods.14 In brief, saline or IL-13–treatedmice were injected intraperitoneally with 0.25 mL of 5�-BrdU(0.75 �g) 3 hours before death. The esophagus was fixed with10% neutral buffered formalin (Sigma) for 24 hours. Afterfixation, the tissue was embedded in paraffin, and 5 micronsections were processed using standard histologic approaches.Tissues were digested with trypsin (0.125%) for 3 minutes at37°C, followed by incubation for 30 minutes at room temper-ature. Sections were washed with PBS 3 times for 2 minutes

1420 MISHRA AND ROTHENBERG GASTROENTEROLOGY Vol. 125, No. 5

and further incubated with monoclonal biotinylated anti-BrdU antibody for 60 minutes at room temperature. Negativecontrols included replacing the primary antibody with PBS,and positive controls were provided by the manufacturer.BrdU nuclear incorporated positive cells were detected withstreptavidin-peroxidase and DAB substrate (Zymed Laborato-ries, San Francisco, CA) followed by counter staining withhematoxylin. The BrdU� cell quantitation was carried outwith the assistance of digital morphometry as described abovefor eosinophils.

Statistical Analysis

Statistical significance comparing different sets of micewas determined by Student t test. P values �0.05 were con-sidered statistically significant.

ResultsIntratracheal IL-13 Induces Dose- andConcentration-Dependent EsophagealEosinophilia

We were first interested in determining whetherintratracheal delivery of IL-13 to the lung was sufficientto induce EE. To test this, IL-13 was delivered to thelung of wild-type mice using doses and techniques thatwere previously known to induce experimentalasthma.32,33,39 Mice received 1–5 doses of recombinantIL-13 (or saline control) separated by 24 hours, and theeosinophil level in the esophagus was determined 24hours after each dose of IL-13. Additionally, differentdoses of IL-13 (0.5, 1.0, and 10 �g) were tested. Asshown in Figure 1, mice treated with 0.5 �g of intra-tracheal IL-13 developed detectable esophageal eosino-philia after 3 doses, and a plateau was seen between thefourth and fifth dose. In contrast, mice that received 1.0and 10 �g of IL-13 had detectable esophageal eosino-philia after 1 dose of IL-13. At these 2 higher doses ofIL-13, eosinophil levels continued to increase with sub-sequent doses and reached a near plateau between thefourth and fifth dose. Eosinophil levels increased witheach concentration of IL-13 (Figure 1), including a4.0-�g dose (data not shown). As a control, mice treatedwith intratracheal saline did not have significant levels ofesophageal eosinophils (Figure 1). As expected, IL-13also induced pulmonary eosinophilia in the same mice.For example, the levels of eosinophils in BALF were0.03 � 0.03 � 104, 1.5 � 0.4 � 104, 3.57 � 0.2 �104, and 29.2 � 3.8 � 104/lung following 5 doses of 0,0.5, 1.0, and 10 �g, respectively, of IL-13 (mean � SD,n � 4–6). In contrast to the effect of intratracheal IL-13on esophageal eosinophilia, intratracheal IL-13 had noeffect on eosinophil levels in the stomach (65.2 � 9.6 vs.56.1 � 23.6 eosinophils/mm2 in saline and IL-13 treated

mice, respectively [mean � SD, n � 4–5; P � 0.8]).We confirmed that our cytokine administration was in-deed resulting in intratracheal delivery by spiking theIL-13 diluent with methylene blue; notably, 97.9% �0.67% (mean � SD, n � 3 mice) of the dye was detectedin the lung tissue (5 minutes after delivery). To deter-mine whether the ability of IL-13 to induce EE was alsoa property of other Th2 cytokines, we intratracheallydelivered IL-4, IL-9, or IL-10. Five doses of 1.0 �g ofrecombinant IL-4, IL-9, or IL-10 were delivered to miceevery 48 hours. As a control, we compared levels ofesophageal eosinophils in mice treated with the sameamount of IL-13. Notably, each of these cytokines (ex-cept for IL-13) failed to induce EE. For example, the levelof esophageal eosinophils was 0.54 � 1.45, 0.84 � 1.4,1.11 � 2.7, 0.77 � 2.0, and 23.8 � 4.3 (mean � SD,n � 4 mice) for IL-4, IL-9, IL-10, and IL-13, respec-tively. Although we have not ruled out the possibilitythat these Th2 cytokines may induce EE under differentconditions (e.g., higher doses), these results suggest aspecial role for IL-13. Collectively, these results establishthat intratracheal delivery of IL-13 promotes dose- andconcentration-dependent EE.

IL-13–Induced Eosinophil-MediatedEpithelial Cell Hyperplasia

IL-13–induced esophageal eosinophilia occurredprimarily in the lamina propria and submucosa but alsoinvolved the epithelial layer (Figure 2). In the epitheliallayer, there was an apparent disruption of the organized

Figure 1. Kinetic and dose-response analysis of IL-13–induced EE.Wild-type mice were exposed to 5 doses (0.5, 1.0, and 10 �g) ofrecombinant murine IL-13 delivered by intratracheal administration onconsecutive days. The level of eosinophils in the esophagus 24 hoursafter each dose of IL-13 was quantitated by anti-MBP immunohisto-chemistry. The data are expressed as mean � SD (n � 5 or 6 mice ateach dose and time point) and are representative of 3 experiments.

November 2003 IL-13 INDUCES EOSINOPHILIC ESOPHAGITIS 1421

polarity of the basal layer (data not shown), prompting usto hypothesize that IL-13 might be inducing epithelialcell hyperplasia, a finding frequently seen in humanEE.5,7–9 To test this, we measured the incorporation ofBrdU after its administration to mice following the lastdose of IL-13. BrdU� cells were predominantly detectedin the basal layer of the epithelial layer (Figure 3A).Notably, IL-13–treated mice had a 2.5-fold increase inBrdU incorporation compared with mice that weretreated with saline (Figure 3B). The BrdU� cells insaline- and IL-13– challenged mice were 88 � 47 and217 � 86/mm2 (mean � SD, n � 5 or 6; P � 0.005).These results establish that intratracheal IL-13 inducesan experimental model of esophagitis characterized byeosinophil accumulation and epithelial hyperplasia.

IL-13–Mediated EE Is STAT6-Dependent

We were next interested in determining the re-lationship between IL-13 and STAT6 in regulatingesophageal eosinophilia. Extensive studies have shownthat STAT6 is important in regulating allergen andIL-13–induced eosinophilia in the lung. More recently,IL-13 has been shown to be important in regulating Th2

responses in the intestine.40–42 To test the role of STAT6in the development of IL-13–induced EE, STAT6 gene-deficient mice and strain-matched wild-type controlmice were treated with 5 doses of intratracheal IL-13 (4.0�g) (Figure 4A). As a negative control, mice were givenintratracheal saline. The eosinophil levels in the esoph-agus of wild-type and STAT6-deficient mice were37.6 � 13.4 and 1.12 � 1.15/mm2, respectively(mean � SD, n � 5–7; P � 0.0001), whereas, insaline-treated mice, they were 1.1 � 1.5 and 0.72 �1.6/mm2, respectively (mean � SD, n � 5–7). As acontrol, the eosinophil levels in the BALF were alsoshown to be reduced in the same mice (Figure 4B).

Figure 3. Epithelial proliferation following IL-13 delivery. Mice(BALB/c) were treated with 5 doses of intratracheal saline or IL-13(4.0 �g), and the incorporation of BrdU in the epithelial layer wasmeasured 3 hours after the last challenge. (A) A representativephotomicrograph showing BrdU� epithelial cells following IL-13 treat-ment is shown. (B) Quantification of BrdU incorporation in saline orIL-13-treated mice is shown. The results are mean � SD (n � 5 or 6mice) and are representative of 2 experiments. Arrowheads indicaterepresentative BrdU� cells. Original magnification �125.

Figure 2. Immunohistochemical analysis of IL-13–induced EE. Wild-type mice were exposed to 5 doses of saline (A) or recombinantmurine IL-13 (4.0 �g) (B–D) delivered by intratracheal administrationon consecutive days. Esophageal tissue was fixed in paraformalde-hyde, embedded in paraffin, cut into 5-�m sections, and immuno-stained with anti-MBP. Representative eosinophils are identified byblack staining. Eosinophils are identified only in the IL-13–treatedmice and are primarily localized to the lamina propria (LP), but also tothe muscularis mucosa (MM), and epithelial (EP) layers. Representa-tive eosinophils are identified with solid arrows, solid arrowheads,and open arrowheads in the LP, EP, and MM regions, respectively. Thephotomicrographs are representative of 3 experiments. Original mag-nification 125� for A and B and 400� for C and D.

1422 MISHRA AND ROTHENBERG GASTROENTEROLOGY Vol. 125, No. 5

Collectively, these results establish that IL-13–inducedEE occurs by a STAT6-dependent mechanism.

IL-5 Is Required for IL-13–InducedExperimental EE

We were next interested in establishing whetherIL-5 had an essential role in promoting esophagealeosinophilia following IL-13 delivery. We addressed thisby inducing experimental EE by IL-13 treatment ofwild-type and IL-5 gene-targeted mice (Figure 5A). Ex-posure of IL-5 gene-targeted mice to intratracheal IL-13did not induce EE, and wild-type control mice developed

marked EE (Figure 5A). For example, the eosinophillevels in the esophagus of wild-type and IL-5 gene-deficient mice following IL-13 treatment were 37.3�11.6 and 3.3 � 3/mm2, respectively (mean � SD, n �4 or 5; P � 0.0001), whereas, saline-treated mice had1.5 � 1.8 and 1.2 � 2/mm2, respectively. As a control,eosinophil levels in the BALF of IL-5 gene-targeted micefollowing intratracheal IL-13 were also shown to bereduced (Figure 5B). Collectively, these results establishan essential role for IL-5 in IL-13-induced EE.

Figure 5. IL-13–induced EE in IL-5 gene-targeted mice. The levels ofeosinophils in the esophagus of IL-5 gene-targeted and wild-type micewere analyzed following intratracheal IL-13 treatment. Wild-type(�/�) or IL-5–deficient (/) mice were challenged with 5 doses ofcontrol saline () or IL-13 (4.0 �g) (�), and the number of eosinophilsin the esophagus (A) was determined by anti-MBP staining and in theBALF (B) by differential counting. The results are expressed asmean � SD (n � 5 or 6 mice) and are representative of 2 experi-ments.

Figure 4. IL-13–induced EE in STAT6 gene-targeted mice. The levelsof eosinophils in the esophagus of STAT6 gene-targeted and wild-typemice were analyzed following intratracheal IL-13 treatment. Wild-type(�/�) or STAT6-deficient (/) mice were treated with 5 doses ofintratracheal control saline () or IL-13 (4.0 �g) (�), and the numberof eosinophils in the esophagus (A) was determined by anti-MBPstaining and in the BALF (B) by differential counting. The results areexpressed as mean � SD (n � 5–7 mice) and are representative of3 experiments.

November 2003 IL-13 INDUCES EOSINOPHILIC ESOPHAGITIS 1423

Eotaxin-1 Is Partially Required for IL-13–Induced Experimental EE

We were next interested in establishing whethereotaxin-1 had an essential role in promoting esophagealeosinophils following IL-13 challenge. We addressed thisby inducing experimental EE by IL-13 treatment ofeotaxin-1 gene-targeted mice. The level of eosinophils inthe esophagus was markedly increased in IL-13–treatedwild-type mice compared with eotaxin-1-deficient mice(Figure 6). In the absence of the eotaxin-1, there was a3.5-fold reduction in the number of eosinophils in theesophagus compared with IL-13–treated wild-type mice.However, in the absence of eotaxin-1, IL-13 treatmentstill induced 14-fold more eosinophils compared with

placebo treatment. For example, the eosinophil levels inthe esophagus of wild-type and eotaxin-1 gene-deficientmice following IL-13 treatment were 48.7 � 10.3 and14.1 � 12.5/mm2, respectively, whereas, in the saline-treated mice, they were 1.5 � 2 and 0.96 � 1.4/mm2,respectively (mean � SD, n � 4 or 5; P � 0.0001). Thepartial role for eotaxin-1 in regulating esophageal eosin-ophilia is similar to its partial role in the lung as assessedby IL-13–induced eosinophil accumulation in the BALF(Figure 6B).

DiscussionEosinophil infiltration into the esophagus is a

commonly observed medical problem in patients withdiverse diseases, including gastroesophageal reflux dis-ease, drug reactions, allergic EE, eosinophilic gastroen-teritis, and primary EE.3–5,8,43 Importantly, recentclinical studies have suggested that these disorders (es-pecially primary EE) are occurring with increasing fre-quency.5 A recent study has also shown that the levelof eosinophils in the esophagus negatively correlateswith response to conventional gastroesophageal refluxtherapy.44 Furthermore, a causal role for esophageal eo-sinophils in the development of epithelial hyperplasia, acardinal feature of primary EE, has been demonstratedin an experimental model of intranasal-aeroallergen-induced EE,14 consistent with the presence of activatedeosinophils in human EE.15 Collectively, these studieshighlight the importance of further dissecting the patho-genesis of EE. Accordingly, the present study was de-signed to examine molecular and cellular processes in-volved in the development of experimental EE. Based onthe high reported concordance between asthma and EEreported in clinical studies, we hypothesized that allergiclung responses may trigger EE. To address this, wefocused on IL-13, a cytokine that is well examined in thecontext of allergic airway inflammation but not in EE. Inparticular, we induced experimental asthma by intratra-cheal delivery of IL-13 and then examined the conse-quences on the esophagus.

We show that IL-13 delivery to the lung is sufficientfor the elicitation of eosinophil trafficking to the esoph-agus; in addition, we uncover several principles concern-ing the mechanism of IL-13–mediated EE. First, wedemonstrate a dose- and concentration-dependent abilityof IL-13 to induce EE following intratracheal cytokineadministration. Our prior studies with aeroallergen-induced EE were provoked by intranasal delivery ofallergen under conditions that simultaneously deliverantigen to the respiratory and gastrointestinal mucosa.14

Our current finding that intratracheal IL-13 induces EE

Figure 6. IL-13–induced EE in eotaxin-1 gene-targeted mice. Thelevels of eosinophils in the esophagus of eotaxin-1 gene-targeted andwild-type mice were analyzed following intratracheal IL-13 treatment.Wild-type (�/�) or eotaxin-1–deficient (/) mice were treated with5 doses of control saline () or IL-13 (4.0 �g) (�), and the number ofeosinophils in the esophagus (A) was determined by anti-MBP stain-ing and in the BALF (B) by differential counting. The results areexpressed as mean � SD (n � 4 or 5 mice) and are representative of3 experiments.

1424 MISHRA AND ROTHENBERG GASTROENTEROLOGY Vol. 125, No. 5

(under conditions that primarily deliver cytokine to thelung), strongly implicates an important role for respira-tory immune responses in the etiology of experimentalEE. The finding that intratracheal IL-13 does not elicitperipheral blood eosinophilia suggests that the effects ofIL-13 are manifested locally (e.g., by up-regulation ofspecific chemokines and/or adhesion molecules) ratherthan by systemic perturbations in eosinophil trafficking.In support of a local mechanism and a specific interactionbetween the lung and esophagus, it is notable that theeosinophils in the stomach were unaffected by intratra-cheal IL-13. Most studies in patients with EE haveprimarily focused on eosinophil accumulation in thesuperficial layers of the esophagus (because this is ame-nable to endoscopic biopsy), but recent ultrasound stud-ies have suggested that EE is also likely to involve thedeeper levels of the esophagus.45 Consistent with thesefindings, IL-13 induced eosinophil accumulation in alllayers of the esophagus. Second, we demonstrate thatIL-13–induced EE requires STAT6; this is similar to therequirement of STAT6 for IL-13–induced experimentalasthma.46 A clinical study has demonstrated elevatedlevels of IL-4 secreting T cells in the esophageal lesionsof patients with secondary EE.47 It is notable that IL-4and IL-13 are related cytokines that share a commonsignal transduction mechanism involving the IL-4 recep-tor � chain and STAT6.21 Third, we report an essentialrole for IL-5 in mediating the effects of IL-13 in theesophagus. Prior studies have shown that IL-5 is requiredfor the induction of experimental EE by aeroallergensand that IL-5 overexpression (by transgenesis or phar-macologic delivery) is sufficient to promote eosinophiltrafficking to the esophagus.14,20 In addition, esophagealeosinophilia induced by exposure of OVA-sensitizedmice to enteric-coated OVA beads is IL-5 dependent.20

This latter finding is particularly significant becausethese oral antigen-challenged IL-5–deficient mice stillmount intestinal eosinophilia,48 highlighting that thereis a differential requirement of IL-5 in regulating esoph-ageal and intestinal eosinophilia. Indeed, a clinical studyhas recently reported overexpression of IL-5 in theesophageal lesions of patients with primary EE.19 IL-5 isknown to prime eosinophils to respond to chemoattrac-tants and to induce eosinophil adhesion molecule expres-sion and activation.49 Thus, IL-5 may induce eosinophiltrafficking to the esophagus by enhancing eosinophilresponsiveness to endogenous chemokines expressed bythe esophagus, such as the eotaxins, or by up-regulatinghoming receptors specifically involved in eosinophil traf-ficking to the esophagus.50 Of note, eotaxin-1 is a con-stitutively expressed chemokine in the esophagus,36 sug-

gesting that IL-13 may further induce eotaxin-1expression or promote IL-5–dependent eosinophil adhe-sion to blood vessels in the esophagus. To evaluate therole of eotaxin-1 in mediating IL-13–induced eosinophiltrafficking to the esophagus, we examined mice that weregenetically deficient in eotaxin-1. These studies revealedthat eotaxin-1 had a large role in IL-13–mediated esoph-ageal eosinophilia. Although IL-13–induced EE waslargely dependent on eotaxin-1, the requirement ofeotaxin-1 was only partial because IL-13–treated eotaxin-1-deficient mice still had higher esophageal eosinophilsthan saline-treated wild-type mice. Consistent with this,eosinophils respond to a variety of chemokines, includingother CCR3 ligands (MCP-2, 3, RANTES, and eotaxin-2and 3)49,51,52; it remains to be determined which che-moattractants are responsible for mediating eotaxin-independent eosinophil accumulation in the esophagus.

Our finding that the cytokine IL-13 can induce ex-perimental EE further implicates Th2 cells in the im-munopathogenesis of EE. This suggests that drugs usedto treat allergy may also be useful for eosinophilic esoph-ageal disorders. Interestingly, recent clinical studies haveshown that topical delivery of glucocorticoids to theesophagus is effective therapy in some patients withEE.8,53,54 Additionally, humanized anti–IL-5, a thera-peutic reagent currently being studied for patients withasthma,55 as well as anti–IL-13 therapeutics may also beuseful for the treatment of EE.56 Although our studysuggests that IL-13–induced EE occurs by a similarmechanism as IL-13–induced asthma, a direct compari-son may not be valid because our studies in the lung areprimarily based on BALF analysis.

In summary, these investigations dissect the cellularand molecular mechanisms involved in eosinophil hom-ing to the esophagus. These data demonstrate that IL-13delivery to the lung simultaneously induces respiratoryand esophageal eosinophilia. In addition, we show acritical role for STAT6, IL-5, and (to a lesser extent)eotaxin-1 in IL-13–induced EE, strongly implicatingTh2-associated immune responses in the pathogenesis ofEE. Taken together with our prior studies demonstratingthat aeroallergen exposure in the respiratory tract inducesEE, these findings support an intimate connection be-tween pulmonary and esophageal eosinophil-associatedhypersensitivity responses and draw attention to the roleof IL-13.

References1. Winter HS, Madara JL, Stafford RJ, Grand RJ, Quinlan JE, Gold-

man H. Intraepithelial eosinophils: a new diagnostic criterion forreflux esophagitis. Gastroenterology 1982;83:818–823.

2. Brown LF, Goldman H, Antonioli DA. Intraepithelial eosinophils in

November 2003 IL-13 INDUCES EOSINOPHILIC ESOPHAGITIS 1425

endoscopic biopsies of adults with reflux esophagitis. Am J SurgPathol 1984;8:899–905.

3. Furuta GT, Ackerman SJ, Wershil BK. The role of the eosinophil ingastrointestinal diseases. Curr Opin Gastroenterol 1995;11:541–547.

4. Sampson HA. Food allergy. Part 1: Immunopathogenesis andclinical disorders. J Allergy Clin Immunol 1999;103:717–728.

5. Fox VL, Nurko S, Furuta GT. Eosinophilic esophagitis: it’s not justkid’s stuff. Gastrointest Endosc 2002;56:260–270.

6. Vasilopoulos S, Murphy P, Auerbach A, Massey BT, Shaker R,Stewart E, Komorowski RA, Hogan WJ. The small-caliber esoph-agus: an unappreciated cause of dysphagia for solids in patientswith eosinophilic esophagitis. Gastrointest Endosc 2002;55:99–106.

7. Orenstein SR, Shalaby TM, Di Lorenzo C, Putnam PE, SigurdssonL, Mousa H, Kocoshis SA. The spectrum of pediatric eosinophilicesophagitis beyond infancy: a clinical series of 30 children. Am JGastroenterol 2000;95:1422–1430.

8. Teitelbaum JE, Fox VL, Twarog FJ, Nurko S, Antonioli D, Gleich G,Badizadegan K, Furuta GT. Eosinophilic esophagitis in children:immunopathological analysis and response to fluticasone propi-onate. Gastroenterology 2002;122:1216–1225.

9. Walsh SV, Antonioli DA, Goldman H, Fox VL, Bousvaros A, Leicht-ner AM, Furuta GT. Allergic esophagitis in children: a clinicopath-ological entity. Am J Surg Pathol 1999;23:390–396.

10. Kelly KJ, Lazenby AJ, Rowe PC, Yardley JH, Perman JA, SampsonHA. Eosinophilic esophagitis attributed to gastroesophageal re-flux: improvement with an amino acid-based formula. Gastroen-terology 1995;109:1503–1512.

11. Liacouras CA, Wenner WJ, Brown K, Ruchelli E. Primary eosino-philic esophagitis in children: successful treatment with oralcorticosteroids. J Pediatr Gastroenterol Nutr 1998;26:380–385.

12. Spergel JM, Beausoleil JL, Mascarenbas M, Liacouras CA. Theuse of skin prick test and patch tests to identify causative foodsin eosinophilic esophagitis. J Allergy Clin Immunol 2002;109:368–368.

13. Furuta GT. Eosinophils in the esophagus: acid is not the onlycause. J Pediatr Gastroenterol Nutr 1998;26:468–471.

14. Mishra A, Hogan SP, Brandt EB, Rothenberg ME. An etiologicalrole for aeroallergens and eosinophils in experimental esophagi-tis. J Clin Invest 2001;107:83–90.

15. Justinich CJ, Ricci A, Kalafus DA, Treem WR, Hyams JS, KreutzerDL. Activated eosinophils in esophagitis in children: a transmis-sion electron microscopic study. J Pediatr Gastroenterol Nutr1997;25:194–198.

16. Sanderson CJ. Interleukin-5, eosinophils, and disease. Blood1992;79:3101–3109.

17. Collins PD, Marleau S, Griffiths-Johnson DA, Jose PJ, Williams TJ.Cooperation between interleukin-5 and the chemokine eotaxin toinduce eosinophil accumulation in vivo. J Exp Med 1995;182:1169–1174.

18. Zimmermann N, Hershey GK, Foster PS, Rothenberg ME. Chemo-kines in asthma: cooperative interaction between chemokinesand IL-13. J Allergy Clin Immunol 2003;111:227–242.

19. Straumann A, Bauer M, Fischer B, Blaser K, Simon HU. Idiopathiceosinophilic esophagitis is associated with a T(H)2-type allergicinflammatory response. J Allergy Clin Immunol 2001;108:954–961.

20. Mishra A, Hogan SP, Brandt EB, Rothenberg ME. IL-5 promoteseosinophil trafficking to the esophagus. J Immunol 2002;168:2464–2469.

21. Wills-Karp M. IL-12/IL-13 axis in allergic asthma. J Allergy ClinImmunol 2001;107:9–18.

22. Piliponsky A, Gleich G, Bar I, Levi-Schaffer F. Effects of eosino-phils on mast cells: a new pathway for the perpetuation of allergicinflammation. Mol Immunol 2002;38:1369.

23. Humbert M, Durham SR, Kimmitt P, Powell N, Assoufi B, Pfister R,

Menz G, Kay AB, Corrigan CJ. Elevated expression of messengerribonucleic acid encoding IL-13 in the bronchial mucosa of atopicand nonatopic subjects with asthma. J Allergy Clin Immunol1997;99:657–665.

24. Akdis M, Akdis CA, Weigl L, Disch R, Blaser K. Skin-homing,CLA� memory T cells are activated in atopic dermatitis andregulate IgE by an IL-13-dominated cytokine pattern: IgG4counter-regulation by CLA- memory T cells. J Immunol 1997;159:4611–4619.

25. Katagiri K, Itami S, Hatano Y, Takayasu S. Increased levels ofIL-13 mRNA, but not IL-4 mRNA, are found in vivo in peripheralblood mononuclear cells (PBMC) of patients with atopic derma-titis (AD). Clin Exp Immunol 1997;108:289–294.

26. Ying S, Meng Q, Barata LT, Robinson DS, Durham SR, Kay AB.Associations between IL-13 and IL-4 (mRNA and protein), vascu-lar cell adhesion molecule-1 expression, and the infiltration ofeosinophils, macrophages, and T cells in allergen-induced late-phase cutaneous reactions in atopic subjects. J Immunol 1997;158:5050–5057.

27. Pawankar RU, Okuda M, Hasegawa S, Suzuki K, Yssel H, OkuboK, Okumura K, Ra C. Interleukin-13 expression in the nasalmucosa of perennial allergic rhinitis. Am J Respir Crit Care Med1995;152:2059–2067.

28. Pawankar R, Okuda M, Yssel H, Okumura K, Ra C. Nasal mastcells in perennial allergic rhinitis exhibit increased expression ofthe Fc epsilonRI, CD40L, IL-4, and IL-13, and can induce IgEsynthesis in B cells. J Clin Invest 1997;99:1492–1499.

29. Al Ghamdi K, Ghaffar O, Small P, Frenkiel S, Hamid Q. IL-4 andIL-13 expression in chronic sinusitis: relationship with cellularinfiltrate and effect of topical corticosteroid treatment. J Otolar-yngol 1997;26:160–166.

30. Zhu Z, Homer RJ, Wang Z, Chen Q, Geba GP, Wang J, Zhang Y,Elias JA. Pulmonary expression of interleukin-13 causes inflam-mation, mucus hypersecretion, subepithelial fibrosis, physiologicabnormalities, and eotaxin production. J Clin Invest 1999;103:779–788.

31. Li L, Xia Y, Nguyen A, Lai YH, Feng L, Mosmann TR, Lo D. Effectsof Th2 cytokines on chemokine expression in the lung: IL-13potently induces eotaxin expression by airway epithelial cells.J Immunol 1999;162:2477–2487.

32. Wills-Karp M, Luyimbazi J, Xu X, Schofield B, Neben TY, Karp CL,Donaldson DD. Interleukin-13: central mediator of allergicasthma. Science 1998;282:2258–2261.

33. Grunig G, Warnock M, Wakil AE, Venkayya R, Brombacher F,Rennick DM, Sheppard D, Mohrs M, Donaldson DD, Locksley RM,Corry DB. Requirement for IL-13 independently of IL-4 in experi-mental asthma. Science 1998;282:2261–2263.

34. Guajardo JR, Plotnick LM, Fende JM, Collins MH, Putnam PE,Rothenberg ME. Eosinophil-associated gastrointestinal disor-ders: a world-wide-web based registry. J Pediatr 2002;141:576–581.

35. Rothenberg ME, MacLean JA, Pearlman E, Luster AD, Leder P.Targeted disruption of the chemokine eotaxin partially reducesantigen-induced tissue eosinophilia. J Exp Med 1997;185:785–790.

36. Mishra A, Hogan SP, Lee JJ, Foster PS, Rothenberg ME. Funda-mental signals regulate eosinophil homing to the gastrointestinaltract. J Clin Invest 1999;103:1719–1727.

37. Finkelman FD, Madden KB, Morris SC, Holmes JM, Boiani N,Katona IM, Maliszewski CR. Anti-cytokine antibodies as carrierproteins. Prolongation of in vivo effects of exogenous cytokinesby injection of cytokine-anti-cytokine antibody complexes. J Im-munol 1993;151:1235–1244.

38. Matthews AN, Friend DS, Zimmermann N, Sarafi MN, Luster AD,Pearlman E, Wert SE, Rothenberg ME. Eotaxin is required for thebaseline level of tissue eosinophils. Proc Natl Acad Sci U S A1998;95:6273–6278.

1426 MISHRA AND ROTHENBERG GASTROENTEROLOGY Vol. 125, No. 5

39. Pope SM, Brandt EB, Mishra A, Hogan SP, Zimmermann N,Matthaei KI, Foster PS, Rothenberg ME. IL-13 induces eosinophilrecruitment into the lung by an IL-5- and eotaxin-dependent mech-anism. J Allergy Clin Immunol 2001;108:594–601.

40. Urban JF Jr, Noben-Trauth N, Donaldson DD, Madden KB, MorrisSC, Collins M, Finkelman FD. IL-13, IL-4R�, and Stat6 are re-quired for the expulsion of the gastrointestinal nematode para-site Nippostrongylus brasiliensis. Immunity 1998;8:255–264.

41. Kweon MN, Yamamoto M, Kajiki M, Takahashi I, Kiyono H. Sys-temically derived large intestinal CD4(�) Th2 cells play a centralrole in STAT6-mediated allergic diarrhea. J Clin Invest 2000;106:199–206.

42. Heller F, Fuss IJ, Nieuwenhuis EE, Blumberg RS, Strober W.Oxazolone colitis, a Th2 colitis model resembling ulcerative coli-tis, is mediated by IL-13-producing NK-T cells. Immunity 2002;17:629–638.

43. Rothenberg ME, Mishra A, Collins MH, Putnam PE. Pathogenesisand clinical features of eosinophilic esophagitis. J Allergy ClinImmunol 2001;108:891–894.

44. Ruchelli E, Wenner W, Voytek T, Brown K, Liacouras C. Severity ofesophageal eosinophilia predicts response to conventional gas-troesophageal reflux therapy. Pediatr Dev Pathol 1999;2:15–18.

45. Fox VL, Nurko S, Teitelbaum JE, Badizadegan K, Furuta GT.High-resolution EUS in children with eosinophilic “allergic” esoph-agitis. Gastrointest Endosc 2003;57:30–36.

46. Akimoto T, Numata F, Tamura M, Takata Y, Higashida N, TakashiT, Takeda K, Akira S. Abrogation of bronchial eosinophilic inflam-mation and airway hyperreactivity in signal transducers and acti-vators of transcription (STAT)6-deficient mice. J Exp Med 1998;187:1537–1542.

47. Nicholson AG, Li D, Pastorino U, Goldstraw P, Jeffery PK. Fullthickness eosinophilia in oesophageal leiomyomatosis and idio-pathic eosinophilic oesophagitis. A common allergic inflamma-tory profile? J Pathol 1997;183:233–236.

48. Hogan SP, Mishra A, Brandt EB, Foster PS, Rothenberg ME. Acritical role for eotaxin in experimental oral antigen-inducedeosinophilic gastrointestinal allergy. Proc Natl Acad Sci U S A2000;97:6681–6686.

49. Rothenberg ME. Eosinophilia. N Engl J Med 1998;338:1592–1600.

50. Bochner BS, Schleimer RP. Mast cells, basophils, and eosino-phils: distinct but overlapping pathways for recruitment. ImmunolRev 2001;179:5–15.

51. Nickel R, Beck LA, Stellato C, Schleimer RP. Chemokines andallergic disease. J Allergy Clin Immunol 1999;104:723–742.

52. Lukacs NW, Oliveira SH, Hogaboam CM. Chemokines andasthma: redundancy of function or a coordinated effort? J ClinInvest 1999;104:995–999.

53. Langdon DE. Fluticasone in eosinophilic corrugated ringed esoph-agus. Am J Gastroenterol 2001;96:926–927.

54. Faubion WA Jr, Perrault J, Burgart LJ, Zein NN, Clawson M, FreeseDK. Treatment of eosinophilic esophagitis with inhaled cortico-steroids. J Pediatr Gastroenterol Nutr 1998;27:90–93.

55. Leckie MJ, ten Brinke A, Khan J, Diamant Z, O’Connor BJ, WallsCM, Mathur AK, Cowley HC, Chung KF, Djukanovic R, Hansel TT,Holgate ST, Sterk PJ, Barnes PJ. Effects of an interleukin-5blocking monoclonal antibody on eosinophils, airway hyper-responsiveness, and the late asthmatic response. Lancet 2000;356:2144–2148.

56. Barnes PJ. Therapeutic strategies for allergic diseases. Nature1999;402:B31–B38.

Received April 10, 2003. Accepted July 31, 2003.Address requests for reprints to: Marc E. Rothenberg, M.D., Ph.D.,

Division of Allergy and Immunology, Department of Pediatrics, Cincin-nati Children’s Hospital Medical Center, Cincinnati, Ohio 45229.e-mail: Rothenberg@cchmc.org; fax: (513) 636-3310.Supported in part by NIH grants R01 AI42242-05 and R01

AI45898-03, the Human Frontier Science Program, International LifeSciences Institute, and the Burroughs Wellcome Fund.The authors thank Andrea Lippelman for editorial assistance; Drs.

Fred Finkelman, Eric Brandt, Simon Hogan, and Paul Foster for helpfuldiscussions and/or reagents; Drs. James and Nancy Lee (Mayo Clinic,Scottsdale, AZ) for the generous supply of anti-MBP; Dr. Debra Donald-son (Wyeth Research, Cambridge, MA) for the generous supply ofrecombinant IL-13; and Klaus Matthaei (Australian National University,Canberra, Australia) for the original supply of the IL-5 (Balb/c) genetargeted mice.

November 2003 IL-13 INDUCES EOSINOPHILIC ESOPHAGITIS 1427