|Home | About | Journals | Submit | Contact Us | Français|
Eosinophilic esophagitis (EE) is an increasingly recognized disease that mimics gastro-esophageal reflux disease. Recently, EE has been associated with esophageal remodeling, but the mechanisms involved are poorly understood. We hypothesized that the development of EE in patients and in an experimental murine model would be associated with eosinophil-mediated tissue remodeling.
Histopathologic analysis of basal layer thickness and collagen accumulation was performed on the biopsy specimens of normal individuals, EE patients, and mouse esophageal tissue sections following experimental induction of EE in wild-type, eosinophil lineage-deficient, interleukin (IL)-5-deficient, and IL-5 transgenic mice, with the latter 2 mice groups having decreased and increased esophageal eosinophilia, respectively.
An impressive accumulation of collagen in the epithelial mucosa and lamina propria, as well as basal layer thickening, was observed in the esophagus of patients with EE as well as in mice with experimental EE compared with controls. Significantly reduced lamina propria collagen and basal layer thickness were observed in IL-5-deficient mice and eosinophil lineage-deficient mice compared with wild-type mice following the induction of experimental EE. Furthermore, the esophagus of CD2-IL-5 transgenic mice showed increased basal layer thickness and collagen accumulation compared with nontransgenic mice, yet IL-5 intestine transgenic mice did not have EE-like esophageal changes. Additional analysis revealed increased IL-5 levels in the esophagus of EE patients, allergen-challenged wild-type mice, and CD2-IL-5 transgenic mice but not in IL-5 intestine transgenic mice.
These findings provide evidence that local IL-5-mediated eosinophilia is essential in the induction of esophageal remodeling.
Tissue remodeling is the result of an imbalance in tissue regeneration and repair,1,2 and leukocyte infiltration is believed to contribute, at least in part.3 In the case of asthma, tissue remodeling involves epithelial cell hyperplasia and thickening of the epithelial mucosa with the development of subepithelial fibrosis from deposition of extracellular matrix proteins (collagen) beneath the basement membrane.4,5 Some of these characteristics have been reported recently in the esophageal biopsy specimens of patients with eosinophilic esophagitis (EE)6,7 and in a murine model of experimental EE.8,9 EE is a chronic disease characterized by eosinophil infiltration into the esophageal mucosa.10–14 Although EE clinically mimics gastroesophageal reflux disease, it is distinguished by the magnitude of eosinophils and basal layer hyperplasia and the lack of response to antireflux medication.15,16 Evidence from animal models suggests that eosinophils may have a role in tissue remodeling through the release of lipid mediators, cytokines, and cytotoxic proteins,17–19 yet their role in esophageal remodeling has been largely unexplored. Therefore, we focused the present study on understanding the role of eosinophils in the induction of esophageal tissue remodeling such as basal layer thickening and collagen accumulation.
To dissect the mechanism responsible for the induction of collagen deposition and basal layer thickening, we investigated mice genetically deficient in interleukin (IL)-5 and transgenic for IL-5 under the control of the T-cell CD2 promoter because we have previously shown that these genetic events induce and reduce EE,20,21 respectively. The levels of eosinophils, lamina propria collagen, and basal layer thickness in the esophagus following induction of experimental EE were assessed. Furthermore, we analyzed EE patient biopsy specimens for the degree of eosinophils, lamina propria collagen, and basal layer thickness. Our results demonstrate that esophageal biopsy specimens of EE patients and the esophagus of mice with experimental EE have multiple features of tissue remodeling. Notably, mice deficient in IL-5 were protected from the development of esophageal remodeling compared with wild-type (WT) mice. In contrast, transgenic mice that overexpress IL-5 (specifically in T cells) had increased esophageal collagen and basal layer thickness. Taken together, we provide evidence that eosinophils and local expression of IL-5 in the esophagus have a crucial role in esophageal remodeling.
Formalin-fixed, paraffin-embedded biopsy samples from the esophagus of normal or EE patients were obtained as per an institutional review board-approved protocol. A total of 32 esophageal biopsy specimens from patients was initially selected, and 8 EE and 6 normal biopsy specimens were subsequently selected on the basis of the presence of esophageal lamina propria to evaluate the presence of esophageal remodeling.7 EE patients were defined histologically as having >24 eosinophils/high-power field (hpf); normal individuals were defined as having 0 eosinophils/hpf and no esophageal abnormality.
The normal individuals and EE patients were selected without regard for age, atopic status, or gender. All EE patients’ biopsy samples selected for esophageal remodeling studies had endoscopic signs of EE. Diagnosis was established based on the maximum eosinophil counts per hpf (magnification, 400×) and basal layer expansion. Normal was defined as having 0 eosinophils per hpf and no basal layer expansion. The normal biopsy specimens were obtained from patients with symptoms typical of gastroesophageal reflux disease and EE but were found to have completely normal esophageal endoscopic and microscopic analyses. Patients with EE were defined by more than 24 esophageal eosinophils/hpf and extensive basal layer hyperplasia. Some of the patients also had allergic diseases, and 2 normal individuals had asthma treated with leukotriene receptor antagonists. The detailed summary of the normal individuals’ and EE patients’ characteristics including eosinophil levels in the esophageal epithelium, disease status, therapy, and trichrome staining scores are shown in Table 1.
BALB/c mice were obtained from Taconic Farms, Inc, New York, NY, and housed under specific pathogen-free conditions. The CD2-IL-5 transgenic,21 intestine fatty acid-binding protein promoter (FABPi)-IL-5 transgenic (iIL-5),22 and Δdbl-GATA and IL-5-deficient mice (all BALB/c) were used as described previously.21,23 All mice were maintained in a barrier facility, and the animals were handled under Institutional Animal Care and Use Committee approved protocols. There was no serologic evidence of pathogens in sentinel mice maintained with the colony.
A mouse model of allergic esophageal inflammation was established using methods described previously.20 In brief, mice were lightly anesthetized with isoflurane (Iso-Flo; Abbott Laboratories, North Chicago, IL), and 100 μg (50 μL) Aspergillus fumigatus (Bayer Pharmaceuticals, Spokane, WA) or 50 μL normal saline alone was applied to the nares using a micropipette with the mouse held in the supine position. After instillation, mice were held upright until alert. After 3 treatments per week for 3 weeks, mice were killed between 18 and 20 hours after the last intranasal challenge.
Esophageal tissue samples from mice and patients were fixed with 4% paraformaldehyde, embedded in paraffin, cut into 5-μm sections, and fixed to positively charged slides. Collagen staining was performed on the tissue sections using Masson’s trichrome (Poly Scientific R&D Corporation, Bay Shore, NY) for the detection of collagen fibers according to the manufacturer’s recommendations.
Multiple tissue pieces of the esophagus from 0.5 cm ascending to the stomach were taken from each mouse, and at least 4 or 5 random Masson’s trichrome-stained sections/mouse were analyzed. Using digital image capture, tissue lamina propria regions of the esophagus were quantified for the thickness of the total accumulated collagen with the assistance of digital morphometric analysis using the Metamorph Imaging System (Universal Imaging Corporation, West Chester, PA), as described previously.9,21 The morphometric analysis was performed on a total of 10 –12 mice/group in 3 experiments. Approximately 50 – 60 representative sections were analyzed, and results report the average of all experiments. The total area of lamina propria collagen was measured and normalized to the tissue length using the computerized software that provides both collagen area and length of the tissue measured and is expressed as collagen thickness in micrometers.
Multiple tissue pieces of the esophagus from 0.5 cm above the stomach were taken from each mouse, and at least 4 or 5 random H&E-stained sections/mouse were analyzed. Using digital image capture, the basal layer area of each esophageal section (1 to 5 layers thick) was measured by digital morphometric analysis using the Metamorph Imaging System computerized software (Universal Imaging Corporation, West Chester, PA), as described previously.9,21 The morphometric analysis was performed on a total 10 –12 mice/group in 3 experiments. Approximately 50 – 60 representative sections were analyzed, and results report the average of all experiments. The system provides total selected area and length of the tissue measured. The total area of the basal layer was normalized with the length measured and expressed as basal layer thickness in micrometers.
Masson’s trichrome-stained sections of normal individuals and EE patients were blindly examined under a light microscope for the semiquantitative collagen accumulation analysis. Absence of collagen in the tissue sections was graded as negative (−), and increased collagen staining was graded as positive (+) on a scale of 1–3.
The RNA samples (500 ng) were subjected to reverse-transcription analysis using Bioscript reverse transcriptase (Bio-Rad, Hercules, CA) according to the manufacturer’s instructions. IL-5, transforming growth factor (TGF)-β, and MUC5AC was quantified by real-time polymerase chain reaction (PCR) using the LightCycler instrument and LightCycler FastStart DNA master SYBR green I as a ready-to-use reaction mix (Roche, Indianapolis, IN). Results were then normalized to GAPDH amplified from the same complementary DNA (cDNA) mix and expressed as fold induction compared with the controls. cDNA was amplified using the mouse and human IL-5, TGF-β1, and MUC5AC primers obtained from SuperArray Bioscience Corporation (Frederick, MD) and GAPDH by using the following primers: tggaaatcccatcaccatct and gtcttctgggtggcagtgat.
Data are expressed as mean ± standard deviation (SD). Statistical significance comparing different sets of mice was determined by unpaired InStat GraphPad t test (San Diego, CA).
The esophageal basal layer thickness and collagen accumulation in the lamina popria were examined in biopsy samples from pediatric individuals with EE and in normal individuals who had no abnormal gastrointestinal pathology. EE patient biopsy specimens from the proximal and/or distal esophagus demonstrated >24 eosinophils/hpf. Histopathologic analysis revealed a prominent accumulation of collagen in the lamina propria and in the papillae of EE patient biopsy specimens, whereas low levels of collagen were detected in the esophageal biopsy specimens of normal individuals (Figure 1A and B). A large increase of the lamina propria collagen thickness was observed in the EE patients’ esophageal biopsy specimens compared with normal individuals’ biopsy specimens. Morphometric analysis showed a thickened basal layer in EE patients’ biopsy specimens compared with normal individuals’ (Figure 1C). Normal individuals and EE patients had a basal layer thickness of 3.0 ± 1.2 μm and 11.0 ± 2.6 μm, respectively (mean ± SD, n = 12–16 individuals, P < .001).
Lamina propria collagen accumulation and basal layer thickness in the esophageal tissue sections were examined in allergen- and saline-challenged mice. Histopathologic analysis of Masson’s trichrome-stained esophageal tissue sections revealed an impressive collagen accumulation in the lamina propria and a thickened basal layer in allergen-challenged mice compared with saline-challenged mice (Figure 2A–D). In addition, we also observed increase of muscularis mucosa thickness (Figure 2A–C) and impressive accumulation of trichrome-positive material in the elongated stromal papillae and into the muscularis mucosa (Figure 2C and D) of allergen-challenged mice. An increased thickening of the basal layer was observed in allergen-challenged mice compared with saline-challenged mice (Figure 2E and F). Morphometric analysis indicated ~2-fold increase in basal layer thickness (Figure 2G) and >2-fold increase in lamina propria collagen thickness in allergen-challenged mice compared with saline-challenged mice (Figure 2H). The basal layer and lamina propria collagen thickness in the esophagus of saline- and allergen-challenged mice was 3.8 ± 0.9 μm vs 9.2 ± 2.3 μm (mean ± SD, n = 12) and 7.4 ± 1.2 μm vs 19.9 ± 2.2 μm (mean ± SD, n = 12), respectively.
TGF-β and mucin MUC5AC genes have a significant role in the induction of tissue remodeling in the lung24,25; therefore, we examined transcript levels of TGF-β1 and MUC5AC in the esophageal biopsy specimens of normal individuals and EE patients as well as in the murine system following saline and allergen challenge. Real-time quantitative PCR analysis indicated that both TGF-β1 and MUC5AC messenger RNA (mRNA) levels were significantly induced in the esophagus of human EE patients (Figure 3A and B) and experimental EE (Figure 3C and D) compared with respective controls.
We next tested the hypothesis that allergen-induced esophageal tissue remodeling was dependent on IL-5-mediated tissue eosinophilia. To test this, experimental EE was induced in WT and IL-5 gene-deficient mice, and esophageal tissue remodeling was determined by analyzing the lamina propria fibrosis and basal layer thickness. The IL-5 gene-deficient mice demonstrated no induction of collagen in the lamina propria, stromal papillae, and muscularis mucosa compared with WT mice following allergen challenge (Figure 4A and data not shown). A significant reduction in basal layer thickness in allergen-challenged IL-5 gene-deficient mice compared with allergen-challenged WT mice was also observed (Figure 4B). The basal layer thickness of allergen-challenged WT and IL-5 gene-deficient mice was 10.4 ± 1.9 μm and 4.7 ± 1.0 μm (mean ± SD, n = 8, P < .01), respectively, and the lamina propria collagen thickness of allergen-challenged WT and IL-5 gene-deficient mice was 21.1 ± 2.2 μm and 5.2 ± 1.0 μm (mean ± SD, n = 8, P < .001), respectively.
We were next interested in establishing whether IL-5 overexpression in lymphocytes promoted esophageal tissue remodeling. Therefore, esophageal tissue sections from WT and CD2-IL-5 transgenic mice were analyzed for lamina propria collagen and basal layer thickness. The esophagus of CD2-IL-5 transgenic mice showed a thickened epithelium, expansion of the connective tissue, and collagen accumulation in the lamina propria and the extended stromal papillae (Figure 5A–C). Representative photomicrographs show increased basal layer and lamina propria collagen in CD2-IL-5 transgenic mice compared with WT mice (Figure 5A–C). The significant accumulation of collagen in the muscularis mucosa was also observed in IL-5 transgenic mice (Figure 5C). Morphometric analysis indicated ~2.5-fold increase in basal layer thickness (Figure 5D) and ~4-fold increase in lamina propria collagen thickness in IL-5 transgenic mice compared with WT mice (Figure 5E). Furthermore, we examined the contribution of eosinophils by examining IL-5 transgenic mice that were deficient in eotaxin-1. We have previously reported that these mice have decreased esophageal eosinophils compared with CD2-IL-5 transgenic mice.21 Morphometric quantitative analysis revealed a significant reduction in the basal layer and lamina propria collagen thickness in eotaxin-1-deficient CD2-IL-5 transgenic mice compared with CD2-IL-5 transgenic mice (Figure 5D and E). The basal layer and lamina propria collagen thickness in CD2-IL-5 transgenic mice was 11.0 ± 1.8 μm and 25.5 ± 4.7 μm (mean ± SD, n = 10), respectively, and in eotaxin-1-deficient CD2-IL-5 transgenic mice 6.7 ± 1.4 μm and 16.3 ± 3.3 μm (mean ± SD, n = 8), respectively, compared with 4.1 ± 0.9 μm and 6.2 ± 1.2 μm (mean ± SD, n = 10), respectively, in WT mice.
Next, we were interested in further establishing whether eosinophils had an essential role in promoting esophageal remodeling following allergen challenge using an independent approach. We addressed this by inducing experimental EE in eosinophil-deficient Δdbl-GATA and WT mice. Following allergen challenge, Δdbl-GATA mice had significantly reduced basal layer and lamina propria collagen thickness compared with WT mice. The saline-challenged Δdbl-GATA and WT mice had comparable lamina propria collagen and basal layer thickness (Figure 6A and B). The basal layer thickness of allergen-challenged WT and Δdbl-GATA mice was 11.4 ± 2.9 μm and 7.1 ± 2.2 μm (mean ± SD, n = 10 –12, P < .01), respectively, and the lamina propria collagen thickness of allergen-challenged WT and Δdbl-GATA mice was 22.8 ± 5.9 μm and 9.9 ± 2.7 μm (mean ± SD, n = 10 –12, P < .001), respectively. Notably, exposure of Δdbl-GATA mice to intranasal allergen did not induce any esophageal eosinophilia (data not shown).
We examined whether ectopic overexpression of IL-5 in the intestine induced esophageal tissue remodeling. To test this, we examined basal layer thickness and collagen accumulation in mice that overexpressed IL-5 under the control of the rat FABPi intestinal promoter. The transgene is specifically expressed in the small intestine.22 The basal layer and lamina propria collagen thickness in iIL-5 transgenic mice were comparable with that in WT mice (Figure 7A and B). Basal layer thickness in WT and iIL-5 transgenic mice was 4.26 ± 0.87 μm and 4.77 ± 0.9 μm (mean ± SD, n = 8), respectively, and lamina propria collagen thickness was 5.84 ± 1.17 μm and 6.59 ± 1.36 μm (mean ± SD, n = 8, P > .05), respectively.
Interestingly, CD2-IL-5 transgenic mice and iIL-5 transgenic mice both demonstrated increased levels of eosinophilia in the esophagus (Figure 7C), yet the CD2-IL-5 transgenic mice had substantially higher levels. The eosinophil levels in the esophagus of WT and CD2-IL-5 transgenic mice were 1.9 ± 0.9/mm2 and 121 ± 14/mm2 (mean ± SD, n = 10, P < .0001), respectively, whereas WT and iIL-5 transgenic had 1.26 ± 0.85/mm2 and 14.5 ± 3.2/mm2 (mean ± SD, n = 8, P < .001), respectively. These data suggest that systemic overexpression of IL-5 is not sufficient to induce esophageal remodeling, implying that local effects of IL-5 on esophageal eosinophils are required.
To test further the hypothesis that local expression of IL-5 is required for induction of esophageal remodeling, we performed quantitative real-time PCR analysis for IL-5 mRNA expression in the esophagus of WT, CD2-IL-5 transgenic, iIL-5 transgenic, and saline- or Aspergillus-challenged mice. CD2-IL-5 transgenic mice had ~20-fold relative increase in expression of IL-5 mRNA in the esophagus compared with iIL-5 transgenic and WT mice (Figure 8A). Both WT and iIL-5 transgenic mice expressed comparable levels of IL-5 mRNA in the esophagus. The allergen-challenged mice showed ~7-fold relative increase in the expression of esophageal IL-5 mRNA compared with the saline-challenged mice (Figure 8B). Furthermore, a similar observation of induced IL-5 mRNA was observed in EE patients’ esophageal biopsy specimens compared with the normal individuals’ (Figure 8C).
EE is a chronic inflammatory disease associated with marked eosinophil infiltration in the esophageal epithelial mucosa.12,26,27 Several clinical reports indicate that esophageal remodeling including basal layer thickening and collagen accumulation can develop in EE patients even in pediatric individuals.6,28–30 Previous reports indicate that macrophages, lymphocytes, and granulocytes possess the ability to stimulate fibrogenesis3,17,19,31,32; therefore, we investigated whether infiltrative cells, especially eosinophils, were necessary to induce esophageal tissue remodeling in EE. Previously, we showed that eosinophil infiltration induces esophageal epithelial hyperplasia in experimental EE,9,20 consistent with the basal layer thickening observed in human EE.26 Now, we present evidence that IL-5-mediated eosinophilia promotes esophageal tissue remodeling in mice. Translational studies first showed that, indeed, human esophageal biopsy specimens from EE patients had a thickened basal layer, elongated papillae filled with trichrome-positive materials, lamina propria fibrosis, and induced transcript levels of inflammatory mediators such as IL-5, TGF-β, and MUC5AC, consistent with previous studies.6,7 The identification of increased MUCA5AC distinguishes EE from Barrett’s esophagus33 and further supports an esophageal T helper cell type-2 environment in EE. Furthermore, we observed basal layer thickening and increased lamina propria collagen accumulation in association with a large influx of eosinophils in the esophagus and induced expression of inflammatory cytokines (IL-5 and TGF-β) and mucin MUC5AC genes in an experimental EE model of mice. Both TGF-β1 and MUC5AC have an important role in promoting tissue remodeling,7,25,34–36 and both of these genes are induced in EE patients as well as in experimental EE. We previously showed that IL-5 has a central role in regulating esophageal eosinophilia8,21; therefore, we further examined the role of IL-5 in allergen-induced esophageal tissue remodeling. Importantly, no induction of basal layer thickening and lamina propria collagen accumulation was observed in allergen-challenged, IL-5 gene-deficient mice. To support the theory that IL-5-mediated eosinophilia promotes esophageal tissue remodeling, we examined basal layer thickness and collagen accumulation in mice that overexpress IL-5. A thickened basal layer and densely positive trichrome-stained material were observed in the lamina propria of esophageal tissue sections of CD2-IL-5 transgenic mice. Furthermore, to evaluate the role of eosinophils in mediating esophageal remodeling, we examined CD2-IL-5 transgenic mice that were genetically deficient in eotaxin-1. We previously showed that eotaxin-1 had a significant role in IL-5-mediated esophageal eosinophilia.21 Our analysis indicated a significant reduction in both basal layer and lamina propria collagen thickness in the esophagus of eotaxin-1-deficient CD2-IL-5 transgenic mice compared with CD2-IL-5 transgenic mice. In addition, we also examined esophageal remodeling in allergen-challenged WT and Δdbl-GATA mice. Δdbl-GATA mice are totally deficient in eosinophils at baseline23,37,38 and do not induce eosinophilia following allergen challenge.23 The Δdbl-GATA mice following allergen challenge showed a significantly reduced induction of esophageal remodeling compared with WT mice. Collectively, these studies highlight the importance of eosinophils in the induction of esophageal remodeling. Furthermore, to determine whether esophageal remodeling was mediated by the local effect of IL-5, we also examined iIL-5 transgenic mice, in which IL-5 was overexpressed in the small intestine.22 Our data revealed that iIL-5 transgenic mice had unaltered basal layer thickening and lamina propria collagen accumulation in the esophagus. These data suggest that proper expression of IL-5 in a particular cell type is necessary for the induction of esophageal remodeling. Indeed, EE patients and CD2-IL-5 transgenic and allergen-challenged mice had local overexpression of IL-5 in the esophagus, consistent with prior studies demonstrating the presence of lymphocytes in the esophagus.26 Collectively, these results support a role for IL-5 and eosinophils in the development of collagen accumulation and basal layer thickening in the esophagus. It is noteworthy that T helper cell type-2 cytokines are strongly implicated in tissue remodeling and fibrosis in other tissues.39–41
In conclusion, our study demonstrates that eosinophil recruitment to the esophagus is associated with tissue remodeling. First, we show impressive collagen accumulation in the epithelial and subepithelial mucosa of esophageal biopsy specimens from human EE patients. Second, we report esophageal collagen accumulation and basal layer thickening in a murine model of experimental EE. Third, we demonstrate that IL-5 is required for allergen-induced esophageal collagen accumulation and basal layer thickening, at least in the setting of experimental EE. Finally, we demonstrate that IL-5 overexpression in lymphocytes is sufficient to induce eosinophil-associated esophageal remodeling. Taken together, these studies suggest a critical role for eosinophils and IL-5 in promoting esophageal tissue remodeling. These data support the potential clinical utility of anti-IL-5 in patients with EE, supporting our early clinical data with such an approach.42
Supported in part by grants NIH RO1 DK067255 (to A.M.) and NIH R01 AI45898 and NIH U19 AI070235 (to M.E.R.), the Digestive Disease Core Center (DDRDC) grant DK0789392, the Campaign Urging Research for Eosinophilic Disease (CURED), the Food Allergy Project, and the Buckeye Foundation.
The authors thank Andrea Lippelman for editorial assistance and Drs James and Nancy Lee (Mayo Clinic, Scottsdale, AZ) for the generous supply of anti-MBP.
Disclosure: M.E.R. is a paid consultant for Merck & Co. and Ception Therapeutics.