Research ArticleFibrosis

Eosinophil depletion suppresses radiation-induced small intestinal fibrosis

See allHide authors and affiliations

Science Translational Medicine  21 Feb 2018:
Vol. 10, Issue 429, eaan0333
DOI: 10.1126/scitranslmed.aan0333

Eosinophils foster fibrosis

Cancer patients receiving abdominal radiation can develop radiation-induced intestinal fibrosis (RIF). Takemura et al. observed eosinophils in the small intestine of RIF patients and irradiated mice. Lymphocyte-deficient mice still developed RIF, but eosinophils were indispensible. They probed the mechanism of interaction between stromal cells, and recruited eosinophils, and showed depletion of eosinophils with two types of antibodies that ameliorated RIF. As anti-eosinophil antibodies are moving forward clinically for other diseases, they may easily be repurposed to prevent RIF in cancer patients.

Abstract

Radiation-induced intestinal fibrosis (RIF) is a serious complication after abdominal radiotherapy for pelvic tumor or peritoneal metastasis. Herein, we show that RIF is mediated by eosinophil interactions with α-smooth muscle actin–positive (α-SMA+) stromal cells. Abdominal irradiation caused RIF especially in the submucosa (SM) of the small intestine, which was associated with the excessive accumulation of eosinophils in both human and mouse. Eosinophil-deficient mice showed markedly ameliorated RIF, suggesting the importance of eosinophils. After abdominal irradiation, chronic crypt cell death caused elevation of extracellular adenosine triphosphate, which in turn activated expression of C-C motif chemokine 11 (CCL11) by pericryptal α-SMA+ cells in the SM to attract eosinophils in mice. Inhibition of C-C chemokine receptor 3 (CCR3) by genetic deficiency or neutralizing antibody (Ab) treatment suppressed eosinophil accumulation in the SM after irradiation in mice, suggesting a critical role of the CCL11/CCR3 axis in the eosinophil recruitment. Activated α-SMA+ cells also expressed granulocyte-macrophage colony-stimulating factor (GM-CSF) to activate eosinophils. Transforming growth factor-β1 from GM-CSF–stimulated eosinophils promoted collagen expression by α-SMA+ cells. In translational studies, treatment with a newly developed interleukin-5 receptor α–targeting Ab, analogous to the human agent benralizumab, depleted intestinal eosinophils and suppressed RIF in mice. Collectively, we identified eosinophils as a crucial factor in the pathogenesis of RIF and showed potential therapeutic strategies for RIF by targeting eosinophils.

INTRODUCTION

Fibrosis is defined as an excessive deposition of extracellular matrix (ECM) based on dysregulated wound healing after repetitive tissue inflammatory injury (1, 2). Fibrosis occurs in various tissues after radiation injury (3). Among various organs, the small intestine is especially sensitive to radiation injury and therefore remains a major dose-limiting organ in abdominal radiation therapy for pelvic tumor or peritoneal metastasis (4, 5). Although many efforts have been made to minimize radiation toxicity of healthy tissues (for example, fractionated low-dose radiation) by careful exposure planning, more than half of the patients suffer from intestinal injury, including diarrhea, bleeding, and fistula due to acute death of crypt epithelium (46). The most troublesome adverse effect of abdominal radiation therapy is intestinal fibrosis [radiation-induced intestinal fibrosis (RIF)], which frequently occurs months or years after irradiation (4, 5). RIF causes intestinal narrowing and transmural fibrosis leading to obstruction. Thus, the development of a novel treatment for RIF is urgently required.

Immune responses are essential for inflammation and subsequent fibrosis of tissues (1, 2). Whereas innate immunity initiates inflammation in response to tissue damage, acquired immunity enhances and sustains these responses. In addition, the development and activation of α-smooth muscle actin–positive (α-SMA+) myofibroblasts (MFs) that produce large amounts of ECM are critical steps for fibrosis (7). Various immune cells infiltrate into damaged tissues and develop mature MFs from the progenitors including fibroblasts, pericytes/mural cells, and bone marrow (BM)–derived fibrocytes via the induction of cytokines and growth factors (1, 2, 7). Among immune cells, T cells, especially T helper type 2 cells (TH2), are important in the progression of many fibrotic diseases (8). However, whether T cells are required for the induction of RIF and the role of other immune cells remain understudied.

Here, we found that eosinophils specifically accumulated in fibrotic submucosa (SM) in the small intestines of human and mouse after abdominal irradiation. Eosinophils interacted with α-SMA+ stromal cells in SM to promote RIF. Depletion of intestinal eosinophils induced by knockout mouse models, eosinophil-targeting antibodies (Abs), and chemical inhibitors effectively prevented RIF. Thus, our findings provide an immunopathological mechanism of RIF and suggest eosinophil depletion as a potential therapeutic strategy.

RESULTS

RIF is induced even in the absence of T cells

The major pathological features of RIF in human patients include thickened SM caused by severe ECM deposition and inflammation characterized by infiltration of immune cells (fig. S1) (4). To investigate the pathological mechanisms of RIF, we established a mouse model of RIF by administering abdominal γ-irradiation with shielding of the head, thorax, and limbs using lead blocks (9). We preliminarily irradiated mice at various doses (6) and defined 12 Gy as the optimal dose for the analysis of RIF without inducing lethal acute gastrointestinal syndrome. At 12 weeks after irradiation, pathological lesions were analyzed macroscopically by magnetic resonance imaging (MRI), using T2-weighted image (T2WI) signals, which mainly indicate water content and are used to detect inflammatory lesions including edema (10). MRI analysis showed that both T2WI signals and thickness of the bowel wall were increased in irradiated intestine, indicating the existence of intestinal inflammation (Fig. 1A). Histological analysis showed typical morphological features, inducing SM thickening with edema and hyperplasia of muscular layer, both of which are well observed in human radiation-induced enteropathy (Fig. 1B) (4, 9). SM thickening was caused by significantly increased deposition of ECM, including collagen fibers composed of collagen type I and type III (P < 0.01; fig. S2). Therefore, this mouse model develops pathological features similar to human RIF.

Fig. 1 Abdominal irradiation induces fibrosis of small intestinal SM regardless of lymphocyte deficiency.

(A) Magnetic resonance imaging (MRI) of abdomen of wild-type (WT) mice at 12 weeks after abdominal irradiation (12 Gy). The lower panel shows bowel wall thickness (n = 3). Scale bars, 5 mm. (B) Azan staining of small intestines of WT mice at 12 weeks. Muc, mucosa; SM, submucosa; Mus, muscularis externa. Scale bars, 200 μm. The lower panel shows the thickness of the SM fibrous layer, as indicated by red arrows and insets (n = 6 to 7). (C) Immunohistochemistry of α-smooth muscle actin (α-SMA) and CD31 in normal small intestines of mice (n = 3). Scale bar, 100 μm. (D) Time course changes in the location of α-SMA+ cells and collagen deposition in the SM after irradiation. Each panel shows immunohistochemistry of α-SMA (left), zoomed image of inset in left panel (middle), and azan staining (right) at 0, 8, and 12 weeks. Yellow arrowheads denote α-SMA+ cells. Red arrowheads denote collagen deposited between crypts and α-SMA+ cells. Scale bars, 50 μm. (E) Azan staining of the small intestines from Rag2+/+ and Rag2−/− mice at 12 weeks. Scale bars, 200 μm. The lower panel shows the thickness of the SM fibrous layer (n = 4 to 9). Results are the means ± SEM of a representative of two (A) or three (B to E) independent experiments. *P < 0.05, **P < 0.01. N.S., not significant [unpaired t test for (A) and (B); Tukey-Kramer post hoc test for (E)].

ECM deposition is primarily mediated by MFs, which usually differentiate from various precursors in injured tissues (1). However, resident MFs are present in the subepithelial region of the small intestine like other specific tissues such as BM stroma and alveolar septa (11, 12). α-SMA+ subepithelial cells were located immediately subjacent to the basement membrane from villus to crypt in normal small intestine, juxtaposed to the basement of intestinal epithelial cells (Fig. 1C). Consistent with previous reports, subepithelial MFs exist as a syncytium, extending throughout the intestinal lamina propria (LP), which merges with α-SMA+ pericytes surrounding the blood vessels in some locations (CD31: marker of endothelial cells of blood vessels) (inset in Fig. 1C) (12). α-SMA+ cells in unirradiated mice were present just beneath the crypt (Fig. 1D). Although α-SMA+ cells were still located subjacent to epithelial cells in villi, pericryptal α-SMA+ cell numbers were diminished in the fibrotic small intestine at 12 weeks after irradiation. α-SMA+ cells in the SM kinetically shifted from the crypt to the muscular layer of mucosa after irradiation (yellow arrowheads in Fig. 1D and fig. S3). Collagen deposits were detected between crypts and α-SMA+ cells in the SM (red arrows in Fig. 1D). These data suggest that α-SMA+ cells are the primary collagen-producing cells in RIF and that excessive collagen deposition had broken down the syncytium structure of subepithelial α-SMA+ cells around crypts, allowing them to move toward the muscle layer after irradiation.

Under steady-state conditions, immune cells are located mainly in the villous core but not in the SM. However, infiltration of immune cells in the SM of irradiated mice was observed, consistent with RIF in humans. Previous studies reported that CD4+ T cells are crucial in the progression of fibrotic diseases (1, 2). Here, we detected T cell infiltration of the SM of irradiated intestines (fig. S4). To investigate T cell involvement in RIF, we used recombination activating gene 2 (RAG2)–deficient mice lacking mature lymphocytes. Rag2−/− mice developed a thickened SM fibrous layer similar to Rag2+/+ mice at 12 weeks after irradiation (Fig. 1E). Thus, RIF was induced even in the absence of T cells.

Intestinal eosinophils are indispensable for RIF

We next examined the contribution of other immune cells in RIF. In addition to T cells, we observed an excessive infiltration of eosinophils in SM in both mice and humans receiving abdominal irradiation, although eosinophil numbers in villi remained unchanged (Fig. 2, A and B). Transmission electron microscopy (TEM) analysis showed that eosinophils in the villi of irradiated mouse intestines had typical morphological features (for example, many eosinophilic granules) similar to those in the villi of healthy intestines. However, eosinophils infiltrating the SM after irradiation had cytoplasm with a loss of granular contents (Fig. 2C). Such degranulated eosinophils observed in inflammatory and allergic disorders were regarded as activated cells (13). Thus, these data suggest that eosinophils infiltrate the fibrotic SM and become activated after irradiation.

Fig. 2 RIF is associated with the accumulation of activated eosinophils in the SM.

(A) Anti–major basic protein (MBP) staining of small intestines of WT mice at 12 weeks after abdominal irradiation (12 Gy). Hematoxylin and eosin (H&E) staining shows a higher magnification of the SM. Red arrowheads denote eosinophils. Scale bars, 100 μm. The lower panels show the numbers of MBP-positive cells in the SM and villous core (n = 6 to 7). (B) H&E staining of the SM of healthy and fibrotic human small intestines. Fibrotic intestine indicated in the image was surgically resected from a patient who received fractionated radiation therapy for colorectal cancer (50 Gy in 25 fractions) at 215 days after the last irradiation. Small magnification images including the area of (B) is shown in the lowest panels of fig. S1. Red arrowheads denote eosinophils. Scale bars, 20 μm. The lower panel shows the numbers of eosinophils infiltrating the SM. Eosinophil numbers were measured at 10 fields per patient (n = 3). (C) Representative transmission electron microscopy image of eosinophils in the small intestinal SM and villi of irradiated WT mice at 12 weeks (n = 3). Scale bars, 10 μm. Results are the means ± SEM. **P < 0.01 (unpaired t test). A representative of three independent experiments is shown (A and C).

Next, we investigated the contribution of eosinophils to RIF using ΔdblGATA mice lacking eosinophils (14). We previously reported that LP innate immune cells are divided into four subsets based on their expressions of CD11b and CD11c: CD11blowCD11chi dendritic cells (DC), CD11bhiCD11chi DC, CD11bintCD11cint macrophages, and CD11bhiCD11cint eosinophils (15). CD11bhiCD11cint cells had bilobed nuclei, eosinophilic granules, and high expression levels of Siglec-F, a marker of eosinophils (Fig. 3A) (16, 17). ΔdblGATA mice lacked the CD11bhiCD11cint population in the small intestine (Fig. 3A). The MRI analysis showed that both T2WI signals and thickness of the intestinal wall were attenuated in ΔdblGATA mice at 12 weeks after 12 Gy of abdominal irradiation (Fig. 3B). Furthermore, the thickened SM caused by collagen deposition was markedly reduced in ΔdblGATA mice compared with wild-type (WT) mice (Fig. 3C). Thus, these results suggested the importance of eosinophils in the pathogenesis of RIF. In support of this, we observed the infiltration of eosinophils into the fibrotic SM in Rag2−/− mice to a similar extent as in Rag2+/+ mice (fig. S5). A recent study reported that ΔdblGATA mice also had decreased numbers of basophils (18). We therefore investigated RIF in ΔdblGATA mice reconstituted with eosinophils generated from BM progenitors (BMEos; fig. S6). The injection of ΔdblGATA mice with BMEos every week after irradiation substantially reconstituted the CD11bhiCD11cint cell population in the small intestine (WT mice, 23.3 ± 2.0% versus BMEos-reconstituted ΔdblGATA mice, 13.5 ± 3.5%; Fig. 3D). Reconstitution of intestinal eosinophils resulted in a marked increase in radiation-induced thickening of the SM fibrous layer of ΔdblGATA mice (Fig. 3E). In proportion to SM thickening, comparable numbers of eosinophils infiltrated into the fibrotic SM of BMEos-reconstituted ΔdblGATA mice (Fig. 3F). Together, these data suggest that eosinophils are essential for RIF.

Fig. 3 Intestinal eosinophils are critical for RIF.

(A) Flow cytometry of small intestinal lamina propria (LP) cells from WT and ΔdblGATA mice. Diff-Quik staining of CD11bhiCD11cintSiglec-F+ cells. Scale bar, 10 μm. (B) MRI images of the abdomens of WT and ΔdblGATA mice at 12 weeks after abdominal irradiation (12 Gy). Scale bars, 5 mm. The right-hand panel shows bowel wall thickness (n = 3). (C) Azan staining of the small intestines of WT and ΔdblGATA mice at 12 weeks. Scale bars, 200 μm. The right-hand panel shows the thickness of the SM fibrous layer (n = 4 to 6). (D to F) Flow cytometry of LP cells of the small intestines (D), azan staining of the small intestines (E), and the numbers of eosinophils in the SM of small intestines (F) from WT mice, noninjected ΔdblGATA mice, and BM-derived eosinophils (BMEos)–injected ΔdblGATA mice at 12 weeks. Data in (D) to (F) were obtained from the same mice. The right-hand panel in (E) shows the thickness of the SM fibrous layer (n = 5 to 6). Scale bars, 200 μm. Results are the means ± SEM of a representative of two independent experiments. *P < 0.05, **P < 0.01 (Tukey-Kramer post hoc test).

CCL11 from α-SMA+ cells induces eosinophil recruitment to the SM

We next investigated how eosinophils are recruited to the SM in RIF. Chemokine (C-C motif) ligand 11 (CCL11) is known to be induced in inflammatory sites to attract eosinophils in numerous diseases (17). Unlike other organs, CCL11 is constitutively expressed in the small intestine under baseline conditions and regulates the recruitment of eosinophils to the LP (19). To determine whether CCL11 was involved in the infiltration of eosinophils into the SM in RIF, we examined the expression of receptors for CCL11 by intestinal eosinophils. Although CCL11 is a ligand of C-C chemokine receptor 3 (CCR3) and CCR5 (16, 17, 20), we detected only CCR3 expression on intestinal eosinophils (Fig. 4A). Intestinal eosinophils migrated in response to CCL11 in vitro, and the CCR3 antagonist SB-328437 blocked the migration (Fig. 4B), suggesting that intestinal eosinophils can migrate via the CCL11/CCR3 axis. Furthermore, infiltration of eosinophils into the SM and thickening of the SM fibrous layer after irradiation were markedly suppressed in Ccr3−/− mice (Fig. 4, C and D). Consistent with this, eosinophil infiltration into the SM and the SM thickening after irradiation were also suppressed by repeated injection of anti-CCR3 neutralizing Ab, which induced a median decrease in small intestinal eosinophil counts of >50% from baseline by a single injection (Fig. 4E and fig. S7).

Fig. 4 α-SMA+ cells induce eosinophil recruitment via the CCL11/CCR3 axis in RIF.

(A) C-C chemokine receptor 3 (CCR3) and CCR5 expression by intestinal eosinophils of unirradiated mice (n = 3). (B) C-C motif chemokine 11 (CCL11)–mediated eosinophil migration in the presence of CCR3 antagonist SB-328437 (10 μM). Migration index is calculated as the ratio of the numbers of cells migrating in response to CCL11 relative to the numbers of cells migrating in response to media alone (n = 3). (C) Anti-MBP staining of the small intestines of Ccr3+/+ and Ccr3−/− mice at 12 weeks after abdominal irradiation (12 Gy). Scale bars, 100 μm. Red arrowheads denote eosinophils. The lower panel shows eosinophil numbers in the SM (n = 4 to 6). (D) Azan staining of the small intestines of Ccr3+/+ and Ccr3−/− mice at 12 weeks. Scale bars, 200 μm. The lower panel shows the thickness of the SM fibrous layer (n = 4 to 6). (E) Thickness of fibrous layer (left) and eosinophil numbers (right) in the small intestinal SM of 83103-treated WT mice were histologically measured at 12 weeks. Histological images are shown in fig. S7 (n = 5 to 6). (F) In situ hybridization (ISH) of Ccl11 mRNA with immunohistochemistry for α-SMA in the small intestines of WT mice at 4 weeks. Each panel shows ISH only (left), immunohistochemistry only (middle), and their combination (right). Arrows indicate Ccl11 mRNA-expressing α-SMA+ cells. Scale bars, 50 μm. Results are the means ± SEM of a representative of three (A and B) or two (C to F) independent experiments. *P < 0.05, **P < 0.01 [unpaired t test for (B) and (C); Tukey-Kramer post hoc test for (D) and (E)].

We next addressed what cell types were involved in CCL11 production for the recruitment of eosinophils to SM after irradiation. To determine the source of CCL11 in SM after irradiation tissues, we analyzed Ccl11 mRNA expression in the small intestine by in situ hybridization (ISH). Ccl11 mRNA was mainly expressed in LP cells at the base of villi where eosinophils normally reside in control mice (Fig. 2A and left panels in Fig. 4F). By contrast, we observed an up-regulation of Ccl11 mRNA in the SM of irradiated intestines. Double staining for ISH and immunohistochemistry with anti–α-SMA Ab indicated that Ccl11 mRNA expression was up-regulated in α-SMA+ cells in the SM of irradiated mice (Fig. 4F, right panels). These findings suggest that pericryptal α-SMA+ cells are activated after irradiation and induce CCL11 to attract eosinophils to the SM.

Chronic crypt necrosis triggers activation of pericryptal α-SMA+ cells

We then investigated how pericryptal α-SMA+ cells are activated for eosinophil recruitment after irradiation. Previous studies suggested that cytokines from TH2 induced CCL11 production (21). However, in this model, T cells had limited roles in the recruitment of eosinophils into the fibrotic SM after irradiation (fig. S5). As an initial event of radiation-induced gastrointestinal toxicity, irradiation causes acute and extensive cell death in crypts (6). Significantly higher levels of terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL)–positive dead cells were present in crypts but not in villi for up to 12 weeks after irradiation (at 6 hours; P < 0.01, at 4 and 12 weeks; P < 0.05; Fig. 5, A and B). Acute crypt cell death is induced by p53-mediated apoptosis (6). An increase of TUNEL and cleaved caspase-3 double-positive cells were observed in the crypts within 6 hours after irradiation (Fig. 5A). However, dead cells at 4 and 12 weeks after irradiation were negative for cleaved caspase-3, suggesting that crypt cell death in the chronic phase is caused by necrosis (22).

Fig. 5 Abdominal irradiation causes crypt necrosis and chronic ATP release.

(A) Terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) and anti-cleaved caspase-3 staining of small intestines after abdominal irradiation (12 Gy). Yellow arrowheads indicate TUNEL-positive crypt cells. Scale bars, 50 μm. (B) The numbers of TUNEL-positive crypt cells after abdominal irradiation (n = 4). HPF, high-power field. (C) Azan and anti-MBP staining of the small intestines of PPADS (pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid)–treated WT mice at 12 weeks. PPADS was intraperitoneally administered three times per week after irradiation. Scale bars, 100 μm. Red arrowheads denote eosinophils. The right-hand panels show thickness of fibrous layer and eosinophil numbers in the SM (n = 6). (D) Adenosine triphosphate (ATP) concentration in culture supernatant of ex vivo–cultured small intestines of WT mice collected after abdominal irradiation (n = 4). Results are the means ± SEM of a representative of three (A and B) or two (C and D) independent experiments. *P < 0.05, **P < 0.01 [Tukey-Kramer post hoc test for (B) and (C); unpaired t test for (D)].

In contrast to apoptosis, necrosis results in the extracellular release of high levels of intracellular components, which function as danger-associated molecular pattern (DAMP) that induces inflammation (23, 24). Recent studies have suggested that activation of P2X receptors by extracellular adenosine triphosphate (ATP) released upon cell damage can contribute to the development of fibrosis in various organs, including lung, kidney, liver, pancreas, and heart (25, 26). Deficiency of P2X7, one of the receptors for ATP, partially but significantly suppressed SM thickening and eosinophil infiltration into SM after irradiation (P < 0.05; fig. S8). Microarray analysis indicated that primary intestinal α-SMA+ cells also express other purinergic receptor such as P2X4 in addition to P2X7 (fig. S9), which might cause the redundant phenotype of P2x7r−/− mice. PPADS (pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid) is a pan-antagonist for P2X receptors including P2X4 and P2X7 (27, 28). At 12 weeks after irradiation, PPADS-treated mice showed significant attenuation of SM thickening and eosinophil infiltration into SM after irradiation when compared with nontreated mice (P < 0.01; Fig. 5C), and the suppressive effects were higher than P2X7 deficiency (fig. S10). Thus, P2X receptors are the crucial factors for the pathogenesis of RIF, indicating that extracellular ATPs are important DAMPs as a trigger of RIF. Then, we measured the release of extracellular ATP from the small intestine. ATP was not released in the immediate postirradiation period (Fig. 5D). However, ATP release was detected at later time points, suggesting that crypt necrosis triggers extracellular release of ATP in the small intestine after irradiation. We next tested which chemokines are induced in primary intestinal α-SMA+ cells by ATP using a chemokine array. Among the chemokines tested, only CCL11 expression was significantly up-regulated by ATP (P < 0.05; Fig. 6A and fig. S11). We also confirmed that ATP up-regulated the mRNA expression of CCL11 in intestinal α-SMA+ cells in a concentration-dependent manner (Fig. 6B). Collectively, these findings suggest that abdominal irradiation causes the chronic release of ATP at later time points after irradiation and that this ATP continuously activates pericryptal α-SMA+ cells to produce CCL11, thereby attracting eosinophils to the SM via CCR3.

Fig. 6 CCL11 and GM-CSF from α-SMA+ cells induce eosinophil recruitment and activation.

(A) Chemokine array analysis of cell lysates from intestinal α-SMA+ cells of WT mice treated with ATP (0 or 2.5 mM) for 2 days. The array images are shown in fig. S11. The signal intensity of the dot blot was measured in duplicate. (B) Quantitative real-time polymerase chain reaction (PCR) of Ccl11 mRNA expression in α-SMA+ cells stimulated with ATP for 1 day (n = 3). (C) Quantitative real-time PCR of mRNA expression of eosinophil activation-associated cytokines of intestinal α-SMA+ cells at 1 day after stimulation with ATP (0 or 2.5 mM; n = 3). (D) Flow cytometry of granulocyte-macrophage colony-stimulating factor receptor α (GM-CSFRα) expression by intestinal eosinophils of unirradiated WT mice (n = 3). (E) Immunohistochemistry for GM-CSF in the small intestine of WT mice at 4 weeks after irradiation. Red arrowheads denote pericryptal α-SMA+ cells. Scale bars, 50 μm. (F) Quantitative real-time PCR of Tgfb1 mRNA in intestinal eosinophils stimulated with GM-CSF for 4 hours (n = 4). (G) Col1a1 and Col1a2 mRNA in intestinal α-SMA+ cells at 1 day after incubation with supernatant from intestinal eosinophils, which were cultured with or without GM-CSF (10 ng/ml). For neutralizing the effects of transforming growth factor–β1 (TGF-β1), the supernatant was pretreated with anti-mouse TGF-β1 Ab (10 ng/ml) for 2 hours (n = 3). (H) Eosinophil peroxidase (EPX)–specific enzyme-linked immunosorbent assay analysis of the culture supernatants of intestinal eosinophils stimulated with GM-CSF (10 ng/ml) and CCL11 (100 ng/ml) for 2 hours (n = 3). Results are the means ± SEM of a representative of two independent experiments. **P < 0.01. [Tukey-Kramer post hoc test for (B), (F), (G), and (H); unpaired t test for (A) and (C)].

Eosinophils activated by GM-CSF from α-SMA+ cells promote fibrosis

We next investigated how intestinal eosinophils were activated in RIF. Although eosinophils are usually activated by TH2 during allergy and helminth infections, RIF and eosinophil activation were induced in the absence of T cells (Fig. 1E). Recent reports showed that interleukin-33 (IL-33) triggered eosinophil activation via innate lymphoid cell type 2 (ILC2) in various diseases (29). We therefore investigated whether IL-33 deficiency affected the development of RIF in vivo. However, there was no difference in collagen deposition and eosinophil infiltration in the SM between Il33−/− and WT mice (fig. S12). In addition, we used mice deficient in RAG2 and the common γ chain, which lacks ILC2 and lymphocytes (29, 30). Rag2−/−γc−/− mice showed a thickening of the SM fibrous layer similar to Rag2+/+γc+/+ mice after irradiation (fig. S13A). We also observed equivalent numbers of eosinophils infiltrating into the fibrotic SM of Rag2−/−γc−/− mice (fig. S13B). Thus, unexpectedly, the IL-33/ILC2 axis is not essential during the pathogenesis of RIF.

These results prompted us to examine whether activated α-SMA+ cells are also involved in eosinophil activation in addition to their recruitment after irradiation. We assessed whether ATP stimulation activates primary intestinal α-SMA+ cells to express cytokines, which are reported to induce eosinophil activation (19). ATP stimulation significantly increased the mRNA expression levels of IL-33 and granulocyte-macrophage colony-stimulating factor (GM-CSF; P < 0.01; Fig. 6C). Intestinal eosinophils expressed a receptor component for GM-CSF (Fig. 6D). In addition, pericryptal α-SMA+ cells expressed GM-CSF in the SM of irradiated intestines (Fig. 6E). We next investigated what responses GM-CSF induced in intestinal eosinophils. We comprehensively examined the profile of genes whose expression showed a >2-fold increase after GM-CSF stimulation in intestinal eosinophils (table S1). GM-CSF is known to enhance eosinophil survival (16). Accordingly, GM-CSF up-regulated prosurvival genes including the Bcl2 family and Pim-1. We found a significant up-regulation of transforming growth factor–β1 (TGF-β1; P < 0.01; Fig. 6F), a major profibrotic mediator (1, 2). We then examined whether TGF-β1 from intestinal eosinophils influenced collagen production by intestinal α-SMA+ cells. Culture supernatant of GM-CSF–stimulated, but not unstimulated, eosinophils significantly up-regulated the collagen type I mRNA expression by intestinal α-SMA+ cells in a TGF-β1–dependent manner (P < 0.01; Fig. 6G). In addition, as shown in Fig. 2C, we observed the degranulation of eosinophils in the SM of irradiated intestines. We therefore tested whether GM-CSF triggered the degranulation of intestinal eosinophils by detecting eosinophil peroxidase (EPX) release. Although GM-CSF alone had little effect on EPX release, costimulation with CCL11 enhanced EPX release (Fig. 6H). All these data suggest that GM-CSF derived from activated pericryptal α-SMA+ cells activates eosinophils after irradiation, leading to the induction of proinflammatory responses including production of the profibrotic cytokine TGF-β1 and eosinophil degranulation.

Anti–IL-5Rα Ab prevents RIF

Ongoing clinical trials suggest that Abs targeting IL-5, a critical cytokine for eosinophil maturation, proliferation, and survival, are safe, eosinophil lineage–specific, and an effective option for treating eosinophilic diseases (31). IL-5–neutralizing Abs, including mepolizumab and reslizumab, substantially reduce circulating eosinophils and ameliorate eosinophilic airway disorders such as asthma (31, 32) and are now approved for clinical usage (33). However, IL-5–deficient mice show reduced eosinophil levels in the blood but not in the small intestine (34), implying that IL-5 neutralization may be insufficient to induce potent and long-lasting reduction of small intestinal eosinophils for ameliorating RIF. Benralizumab is a humanized, afucosylated monoclonal Ab against human IL-5 receptor alpha (IL-5Rα) (31, 32). Benralizumab can induce almost complete depletion of circulating eosinophils not only by IL-5R antagonism but also by Ab-dependent cell-mediated cytotoxicity (ADCC) and is considered to have enhanced efficiency of eosinophil depletion when compared with IL-5–neutralizing Abs. We therefore assessed efficacy of anti–IL-5Rα Ab against RIF. We generated a chimeric monoclonal Ab cmIL5Ra1b12 with high binding affinity to mouse IL-5Rα (Kd = 3.17 × 10−11 M), composed of rat immunoglobulin G1 (IgG1) Fab fragment and afucosylated Fc portion of mouse IgG2a, which induces potent ADCC in mice (35). cmIL5Ra1b12 specifically recognized Ba/F3 cells ectopically expressing IL-5Rα (mIL-5Rα/Ba/F3 cells) and completely inhibited their proliferation in response to IL-5 (fig. S14, A and B). Notably, cmIL5Ra1b12 triggered ADCC against mIL-5Rα/Ba/F3 cells cultured with effector cells (fig. S14C). A single injection of cmIL5Ra1b12 into mice at doses above 1 mg/kg induced a median decrease in blood eosinophil counts of >90% from baseline (control, 1.62 ± 0.55 × 105 cells/ml versus 1 mg/kg, 0.12 ± 0.03 × 105 cells/ml) within 1 week (Fig. 7, A and B). This reduction was maintained for 2 weeks when injected with cmIL5Ra1b12 at 5 and 25 mg/kg and for 4 weeks at 25 mg/kg. Thus, consistent with human studies, anti–IL-5Rα Ab strongly depletes eosinophils in mice. cmIL5Ra1b12 reduced small intestinal eosinophils and achieved 70% reduction (control, 1.07 ± 0.12 × 106 cells per intestine versus 25 mg/kg, 0.23 ± 0.02 × 106 cells per intestine) through 4 weeks by a single injection at 25 mg/kg (Fig. 7, C and D). Therefore, we repeatedly injected cmIL5Ra1b12 (25 mg/kg) every 4 weeks from 4 weeks before irradiation. At 13 weeks after irradiation, cmIL5Ra1b12-treated mice showed 80 to 90% reductions in intestinal eosinophils (fig. S15), leading to marked reductions in eosinophil infiltration in SM and attenuation in SM fibrosis (Fig. 7E and fig. S16). As such, we propose that anti–IL-5Rα Ab treatment is a promising approach for suppressing RIF.

Fig. 7 cmIL5Ra1b12 effectively ameliorates RIF.

(A) Flow cytometry of blood leukocytes of WT mice at 1 week after injection of cmIL5Ra1b12 (25 mg/kg). CD11b+Gr-1int cells among SSChiCD11c leukocytes are eosinophils expressing Siglec-F. (B) Time course changes in blood eosinophil frequency in WT mice injected with cmIL5Ra1b12 (n = 3 to 4). (C) Flow cytometry of small intestinal LP cells of WT mice at 4 weeks after cmIL5Ra1b12 injection (25 mg/kg). Right panels show CD11bhiCD11cintSiglec-F+ eosinophils. (D) Small intestinal eosinophil levels in WT mice at 4 weeks after cmIL5Ra1b12 injection (n = 3). (E) Azan and anti-MBP staining of the small intestines of cmIL5Ra1b12-treated WT mice at 13 weeks after abdominal irradiation. Scale bars, 100 μm. Red arrowheads denote eosinophils. The right-hand panels show thickness of the SM fibrous layer and eosinophil numbers in the SM (n = 5 to 6). Results are the means ± SEM of a representative of two independent experiments. **P < 0.01 [Tukey-Kramer post hoc test for (E)].

DISCUSSION

Here, we performed a detailed analysis of the immunopathological mechanisms underlying RIF by using a mouse abdominal irradiation model, which mimics human chronic enteritis after radiation therapy and develops irreversible fibrosis in the small intestine. Consequently, we revealed that RIF is mediated by immune-nonimmune cell interactions between eosinophils and pericryptal α-SMA+ cells. Irradiation elevated extracellular ATP levels due to chronic crypt necrosis, which induced CCL11 and GM-CSF expression by pericryptal α-SMA+ cells that attracted and activated eosinophils, respectively. TGF-β1 from GM-CSF–stimulated eosinophils promoted collagen expression by α-SMA+ cells. Thus, radiation-induced injury mediated mutual activation of eosinophils and α-SMA+ cells that collectively contributed to SM fibrosis.

An interesting finding is that the activating mechanism of eosinophils to promote fibrotic reactions in RIF differs from those of other fibrotic diseases. Recent studies reported that eosinophils are important for tissue remodeling such as epithelial hyperplasia and collagen accumulation in eosinophilic esophagitis (EoE) (3638). Immune sensitization to various foods and subsequent TH2-polarized inflammation in the esophagus might be closely related with EoE development, indicating the TH2/eosinophil axis is important during EoE pathogenesis (39). Furthermore, the importance of CD4+ T cells was demonstrated in experimental fibrosis models of the liver, lung, and kidney, as well as those of inflammatory bowel diseases (IBD) (1, 2, 40). By contrast, RIF was induced in the absence of T cells. ILC2 also activates eosinophils via the secretion of type 2 cytokines including IL-5 and IL-13 (29). It was demonstrated that ILC2 is activated by IL-33 via the IL-33 receptor ST2 (29, 30). Recent reports showed that the IL-33/ILC2 axis was essential for fibrosis of the lung and liver (41, 42). The IL-33/ILC2 axis was also dispensable for RIF. Thus, the pathological mechanism of RIF is prominent compared with IBD, lung, and liver fibrosis models, in that eosinophil activation is not mediated by immune cells involved in type 2 immunity, but likely by a complete innate mechanism, as Rag2-deficient mice had unabated RIF. RIF is unique among eosinophil-mediated fibrotic diseases in that resident α-SMA+ cells are involved in both the recruitment and activation of eosinophils.

Our findings suggest that RIF may be suppressible by blockade of undesirable α-SMA+ cell-eosinophil interactions after irradiation, such as through targeting P2X receptor activation and the CCL11/CCR3 axis–dependent chemotaxis. On the basis of these findings, we aimed to test the potential of an eosinophil-directed therapeutic agent for RIF, particularly by using an anti–IL-5Rα Ab. This strategy was chosen as a reagent using the same mechanism of action that is now in advanced clinical stages. In particular, benralizumab, a monoclonal Ab against human IL-5Rα, induces near-complete depletion of blood eosinophils and shows strong antiasthma activity in humans (31). Depleting eosinophils is a safe strategy because eosinophil-deficient animals are generally healthy, and anti–IL-5 and anti–IL-5Rα are safe in humans (31). Therefore, we generated a new monoclonal Ab cmIL5Ra1b12 with high binding affinity to mouse IL-5Rα, which induces potent antagonism and ADCC, similar to benralizumab in humans. Anti–IL-5Rα treatment strongly suppressed eosinophil infiltration in the small intestinal SM and efficiently attenuated experimental RIF in mice. Thus, eosinophil-specific depletion by anti–IL-5Rα Ab might prevent the development or progression of RIF, allowing an increase in the maximum radiation tolerance dose of the intestine, which may help improve therapeutic outcomes in patients with pelvic tumors or peritoneal metastasis. As an alternative strategy to specifically reduce intestinal eosinophils, Ab therapy targeting CCR3 may also be promising.

Finally, there are some limitations of our study that will need to be investigated in the future. First, the cause of chronic crypt cell death after irradiation remains unclear. Although radiation-induced acute cell death is mediated by apoptosis, crypt cell death occurs by necrosis during the chronic phase, causing the release of extracellular ATP that functions as a DAMP. According to previous reports, vasculitis and subsequent vascular stenosis are hallmarks of chronic intestinal inflammation after irradiation (3, 4). Disturbance of the peripheral circulation around crypts may be one possible cause of crypt necrosis. Furthermore, epithelial cell–MF interactions are important for gut morphogenesis (12) because pericryptal MFs secrete various growth factors, cytokines, neurotransmitters, and hormones, which contribute to stability of the crypt epithelium and gut homeostasis (12). Subepithelial MFs are positioned parallel to intestinal blood vessels and exist as a syncytium with pericytes and fibroblasts (12). ECM deposition in SM may alter vascular arrangements and the location of α-SMA+ cells around crypts. Destruction of epithelial cell–MF interactions may also cause chronic crypt necrosis. Clarifying the mechanisms might provide clues for new strategies to prevent RIF. Second, this study tested the preventive effects of eosinophil depletion on RIF, but not its curative effects. It would be valuable to address whether eosinophil depletion induces regression of established RIF. Third, it is important to consider the influence of eosinophil depletion on cancer development in planning the clinical treatment for RIF. On the basis of this research, future studies will provide optimal therapeutic options for the management of RIF.

MATERIALS AND METHODS

Study design

The objective of this study was to establish viable therapeutic strategies for RIF by clarifying the pathological mechanisms using specimens of fibrotic lesions in small intestines from mice and humans that received abdominal irradiation. ΔdblGATA, Rag2−/−, Rag2−/−γc−/−, Il33−/−, and P2x7r−/− mice and their littermate controls on BALB/c background were exposed to 12 Gy of abdominal irradiation. Human normal and irradiated small intestinal samples were obtained from intact areas in patients with colorectal or cecal cancer and surgically resected specimens in patients with radiation enteritis during the chemoradiation therapy for colorectal or uterus cancer, respectively. In mice, manifestations of intestinal fibrosis at 12 weeks after irradiation were monitored by MRI and histologically investigated. α-SMA, eosinophil major basic protein (MBP), collagen type I, GM-CSF, and CD3 were detected by immunohistochemistry. Cell death was investigated by TUNEL- and anti-cleaved caspase-3 staining. mRNA expression of CCL11 was histologically detected by ISH. Expressions of surface molecules of intestinal eosinophils were determined by flow cytometry. Migration of intestinal eosinophils was investigated using an in vitro Transwell migration assay system. Gene expression of eosinophils was measured by quantitative real-time polymerase chain reaction (PCR) and next-generation sequencing. Eosinophil degranulation was observed by TEM analysis and enzyme-linked immunosorbent assay. Protein and gene expression of α-SMA+ stromal cells were measured by proteome profiler Ab array, microarray, and quantitative real-time PCR. Therapeutic effects of eosinophil depletion on RIF were examined by treating mice with mouse IL-5Rα–targeting Ab cmIL5Ra1b12 or anti-CCR3. No statistical method was used to preliminarily determine sample size. We did not use exclusion, randomization, or blinding procedures. Primary data are located in table S2.

Radiation treatment

The abdomens of mice anesthetized with an intraperitoneal injection of pentobarbital sodium (30 mg/kg body weight) were exposed to 12 Gy of γ-irradiation at a dose rate of 88 cGy/min using a 137Cs irradiator (Gammacell 40 Exactor; MDS Nordion). Other organs, above and below the abdomen, were shielded using lead blocks. PPADS (1 mg per mouse; TOCRIS Bioscience) was intraperitoneally administered three times per week for 12 weeks after irradiation. cmIL5Ra1b12 (25 mg/kg) was intraperitoneally administered at 4 weeks before and at 0, 4, 8, and 12 weeks after irradiation. CCR3 neutralizing Ab 83103 (5 mg/kg; R&D systems) was intraperitoneally injected every 2 weeks from 2 weeks before irradiation. MRI for in vivo imaging was conducted at 12 weeks (Supplementary Materials and Methods).

Histological analysis

Mouse small intestine was fixed in 10% formalin overnight and embedded in paraffin. Five-micrometer sections of mouse and human small intestine were stained with hematoxylin and eosin and observed under a light microscope (BZ-9000, Keyence). To visualize collagen fibers, we stained sections with azan staining. The thickness of the intestinal SM fibrous layer stained by azan was measured using the BZ-II Image Analysis Application (Keyence). α-SMA, CD3, CD31, cleaved caspase-3, collagen type I, GM-CSF, and MBP were detected by immunohistochemical analysis, and Ccl11 mRNA was detected by ISH (Supplementary Materials and Methods).

Culture of intestinal cells

CD11bhiCD11cintSiglec-F+ eosinophils were isolated from small intestinal LP as described previously (7). Intestinal α-SMA+ cells were isolated from the small intestines of 5-day-old WT BALB/c mice as described by Lahar et al. (43).

Statistical analysis

The statistical significance of differences was evaluated using unpaired two-tailed Student’s t tests. One-way analysis of variance (ANOVA) followed by the Tukey-Kramer post hoc test was performed for comparisons among multiple groups. P values less than 0.05 were considered statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/429/eaan0333/DC1

Materials and Methods

Fig. S1. Histology of human RIF.

Fig. S2. Detection of collagen fibers in mouse small intestine after abdominal irradiation.

Fig. S3. Migration of α-SMA+ cells from the pericryptal area after abdominal irradiation.

Fig. S4. Detection of T cells in mouse small intestine after abdominal irradiation.

Fig. S5. Eosinophil recruitment to the SM in Rag2−/− mice after abdominal irradiation.

Fig. S6. Mouse BM-derived eosinophils (BMEos).

Fig. S7. Anti-CCR3 Ab treatment ameliorates RIF.

Fig. S8. RIF in P2x7r−/− mice after abdominal irradiation.

Fig. S9. ATP receptor expression of intestinal α-SMA+ stromal cells.

Fig. S10. Suppression of RIF by blocking ATP-mediated activation of P2X receptors.

Fig. S11. Chemokine protein array analysis of intestinal α-SMA+ cells.

Fig. S12. RIF in Il33−/− mice after abdominal irradiation.

Fig. S13. RIF in Rag2−/−γc−/− mice after abdominal irradiation.

Fig. S14. In vitro analyses of anti–IL-5Rα activities of cmIL5Ra1b12.

Fig. S15. Eosinophil depletion by repeated treatment with cmIL5Ra1b12.

Fig. S16. Hydroxyproline levels of the small intestines of cmIL5Ra1b12-treated WT mice.

Table S1. Gene expression of intestinal eosinophils after GM-CSF stimulation.

Table S2. Primary data.

References (4450)

REFERENCES AND NOTES

Acknowledgments: We thank B. Baigalmaa, N. Kitagaki, K. Fujisawa, K. Abe, and T. Kikuchi for technical assistance and K. Ogawa and S. Watanabe for secretarial assistance. We are grateful to S. Koyasu, K. Moro, and K. Okoshi for the helpful discussions. Abs specific for mouse MBP and EPX were provided by the laboratory of J. J. Lee and N. A. Lee (Mayo Clinic, Scottsdale, AZ, USA). Funding: This study was supported by Grant-in-Aid for Scientific Research (B) (to S.U.); Grants-in-Aid for Young Scientists (B) (to N.T.); the Funding Program for World-leading Innovative R&D on Science and Technology from the Japanese Society for the Promotion of Science; Grant-in-Aid for Scientific Research on Innovative Areas (Homeostatic regulation by various types of cell death) (15H01367) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (to S.U.); the MEXT Translational Research Network Program of the MEXT (to S.U.); the Ministry of Health, Labour and Welfare in Japan (to S.U.); Practical Research Project for Allergic Diseases and Immunology from Japan Agency for Medical Research and Development (to S.U.); the Ichiro Kanehara Foundation for the Promotion of Medical Sciences and Medical Care (to S.U.); the Senri Life Science Foundation (to S.U.); the Astellas Foundation for Research on Metabolic Disorders (to S.U.); and the Takeda Science Foundation (to S.U.). Author contributions: N.T. conducted most of the experiments. N.T. and S.U. designed all of the experiments. Y. Kurashima measured ATP release from small intestines. Y.M. and Y.Y. performed MRI analyses. K.O. established cmIL5Ra1b12. Y. Kumagai and Y.S. helped with the next-generation sequence analysis. T.O., H.O., H.M., J.N., and M.M. helped with analysis of human samples. L.A., S.K., Y.T., K.M., and D.K. helped with the experiments. Y. Kurashima, E.J.P., S.S., Y.O., Y. Kumagai, Y.S., K.J.I., M.E.R., H.K., and S.A. provided advice on the experiments. N.T. and S.U. prepared the manuscript. S.U. directed the research. Competing interests: N.T., K.O., and S.U. are part of the inventors on patent application JP2017-212459 filed by The University of Tokyo and Kyowa Hakko Kirin Co. Ltd. that covers the treating method for radiation injury by eosinophil-depleting agent. The other authors declare that they have no competing interests. Data and materials availability: Next-generation sequencing and microarray data were deposited in the Gene Expression Omnibus (accession numbers GSE56786 and GSE57636). Requests for the Abs specific for mouse MBP and EPX, available through a material transfer agreement (MTA), should be sent to J. J. Lee and N. A. Lee at the Mayo Clinic.
View Abstract

Navigate This Article