A Deficiency in the Cytokine TNFSF14/LIGHT Limits Inflammation and Remodeling in Murine Eosinophilic Esophagitis

Abstract

Eosinophilic esophagitis (EoE) is a chronic type 2 allergic disease, with esophageal tissue remodeling as the mechanism behind clinical dysphagia and strictures. IL-13 is thought to be a central driver of disease, but other inflammatory factors such as interferons and TNF superfamily members have been hypothesized to play a role in disease pathogenesis. We recently found that the cytokine TNFSF14/LIGHT is upregulated in the esophagus of EoE patients and that LIGHT promotes inflammatory activity in esophageal fibroblasts. However, the global effects of LIGHT on EoE pathogenesis in vivo remain unknown. We investigated the impact of a LIGHT deficiency in a murine model of EoE driven by house dust mite (HDM) allergen. Chronic intranasal challenge with HDM promoted esophageal eosinophilia, increased CD4+ T cell numbers, and IL-13 and CCL11 production in WT mice. Esophageal remodeling was reflected by submucosal collagen accumulation, increased muscle density, and greater numbers of fibroblasts. LIGHT−/− mice displayed normal esophageal eosinophilia, but exhibited reduced frequencies of CD4 T cells, IL-13 expression, submucosal collagen, and muscle density, and a decrease in esophageal accumulation of fibroblasts. In vitro, LIGHT increased division of human esophageal fibroblasts and selectively enhanced IL-13-mediated expression of a subset of inflammatory and fibrotic genes. These results show that LIGHT contributes to various features of murine EoE, impacting the accumulation of CD4 T cells, IL-13 production, fibroblast proliferation, and esophagus remodeling. These findings suggest that LIGHT may be a novel therapeutic target for the treatment of EoE.

INTRODUCTION

Eosinophilic esophagitis (EoE) is a chronic type 2 allergic disorder of the esophagus characterized by the accumulation of eosinophils in the esophageal epithelium (1, 2). EoE is often diagnosed during childhood or early adolescence. Elimination diets that restrict multiple foods are used to avoid allergen exposure and control disease progression (3, 4). A number of patients also undergo anti-inflammatory therapy, where topical corticosteroids are currently used (5, 6). However, many patients suffer from persistent inflammation that ultimately leads to abnormal esophageal tissue remodeling, characterized by an excessive deposition of extracellular matrix (ECM) components combined with muscle hypertrophy (7–9). Ultimately, this contributes to esophageal stiffness and lumen narrowing, causing a loss of compliance and food impactions. Patients suffering from fibrotic complications may require invasive dilation therapy to restore cross sectional area, which highlights an urgent need for therapies to control the progression toward fibrotic remodeling.

A number of studies in mice have highlighted the ability of airway or food allergens to cause esophageal eosinophilic infiltration and have provided models to understand the development of EoE features (10, 11). Studies using allergens such as ovalbumin or Aspergillus fumigatus have aided the discovery of the key roles of eosinophil receptors such as siglec-F, and cytokines such as IL-5, IL-13 and TGF-β1, in the development of esophageal inflammation (10, 12–15). IL-5 is associated with eosinophilia, whereas TGF-β1 and IL-13 are linked to fibroblast activation that increases ECM deposition, as well as other aspects of inflammation that contribute to fibrosis and remodeling (12, 14–17). However, it is likely that additional inflammatory factors may be important and act in complementary manners to these cytokines. In this regard, we have recently shown that a member of the TNF superfamily of proteins termed TNFSF14/LIGHT is transcribed by several different subsets of T cells in the esophagus of patients with EoE, including the pathogenic Th2 effector subset that produces IL-13 (18). Our studies also showed that LIGHT is a strong activator of a pathogenic pro-inflammatory response in human esophageal fibroblasts in vitro (18, 19).

LIGHT has two receptors, the herpes virus entry mediator (HVEM) and the lymphotoxin beta receptor (LTβR), molecules expressed on many hematopoietic and non-hematopoietic cells including fibroblasts, epithelial cells, T cells and eosinophils (19–23). From studies in other in vivo or in vitro systems, LIGHT has been found to regulate a range of inflammatory activities in these cell types, including promoting survival and cytokine production in memory Th2 cells (24). However, the overall importance of LIGHT to the development of EoE and to different features of EoE disease is not yet known. In the present manuscript, we investigated LIGHT’s contribution to esophageal inflammation and remodeling in a murine model of EoE. We show that a deficiency of LIGHT protects from features of murine EoE including T cell-mediated inflammation, IL-13 accumulation and esophageal remodeling. In vitro, we show that LIGHT drives fibroblast proliferation and synergizes with IL-13 to enhance the expression of select inflammatory and fibrosis mediators, corresponding to weak accumulation and/or proliferation of fibroblasts in vivo in the absence of LIGHT.

MATERIALS AND METHODS

Animals and EoE model

C57BL/6 wild type mice used in this study were purchased from Jackson laboratories (Bar Harbor, ME) and housed in the animal facility at La Jolla Institute for Immunology. All experiments were carried out in compliance with ethical requisitions under approved protocols. LIGHT−/− mice were bred in house by crossing of homozygous LIGHT−/− male and female mice. Female mice in an age range of 6-10 weeks were used for experiments. For induction of disease, mice were sensitized by intraperitoneal administration of 4 μg house dust mite (HDM, dermatophagoides pteronyssinus) (Greer laboratories, Lenoir, NC) and 50 μl of Alum adjuvant (Sigma, St Louise, MO) diluted in PBS to a final volume of 250μl. This was repeated twice in one week on a 4-day interval, and mice were then rested for 4 days (Figure 1A). Mice were then challenged by intranasal administration of 50 μg HDM diluted in PBS with 4 challenges on 4 consecutive days (acute period) followed by 6 additional challenges administered on a twice a week regime for 3 weeks (chronic phase, Figure 1A). Mice were sacrificed and esophagus collected for analysis on day 31.

Figure 1. LIGHT−/− mice exposed to HDM show defective T cell infiltration in the esophagus but normal eosinophilia.

A, Schematic of immunization protocol with HDM to drive esophageal inflammation. B, Images of esophagi stained with hematoxylin & eosin in WT mice exposed to PBS or HDM (arrows mark eosinophils, data representative of 2 independent experiments with a total of 3 PBS and 5 HDM mice). C, representative flow cytometry depicting CD45⁺ cells (left) and CD3⁺ or CD3⁻ cells within the CD45+ population, comparing PBS or HDM immunized wild type mice. D, representative flow of CD11b and CD11c within CD45⁺CD3⁻ cells (left) and Siglec-F within the CD11b⁺CD11c− fraction (right) in mice treated as in C. E, representative flow depicting CD4 within the CD45⁺CD3⁺ fraction. F-H, numbers of CD11b⁺CD11c−, Siglec-F⁺, and CD3⁺CD4⁺ cells in the esophagus of wild type mice exposed to PBS, or wild type and LIGHT−/− mice exposed to HDM. Data from 4 independent experiments is presented, each dot corresponds to a single mouse (*p<0.05, **p<0.01 or ***p<0.001).

Flow cytometry analyses

All information regarding antibodies used in this study can be found in supplementary Table 1. For flow cytometry of murine immune cell populations, esophagi were collected on PBS supplemented with 3% fetal bovine serum, 2mM EDTA and 0.01% Sodium azide. The tissues were opened longitudinally, chopped into small fragments and sequentially incubated with 0.25% trypsin-EDTA on ice with agitation for 2 hours, followed by a 1:1 mix of trypsin-EDTA and dispase on ice with agitation for 1 hour, and then incubated with 2 mg/ml collagenase IV at 37⁰C for 30 minutes. For analyses of structural cell populations, the cell purification was performed using a Lamina Propria Dissociation Kit (Miltenyi Biotec, San Jose, CA). The cells obtained were counted and the total number recorded. The cells were pre-incubated with Fc-blocking antibody for 10 minutes, followed by staining with the appropriate antibody panels. Where staining for intracellular markers was required, cells were first incubated with antibodies to surface markers for 20 minutes, washed with PBS, and then fixation/permeabilization performed using the eBiosciences FoxP3 fixation/permeabilization kit (Thermo, Waltham, MA) following the manufacturers indications. Cells were analyzed using a BD LSR or BD Fortessa. Cells were first gated for high CD45 expression and next segregated into CD3⁻ and CD3⁺ populations. For the identification of eosinophils, CD45⁺ CD3⁻ cells were gated as CD11b^highCD11c^lowSiglec-F⁺. For identification of T cells, CD45⁺CD3⁺ were first gated for CD4⁺, then CD3⁺CD4⁻ cells were gated for CD8⁺. In experiments aimed at analyzing fibroblast abundance, cells were negatively selected as CD45⁻CD31⁻EpCAM⁻. These triple negative populations were further analyzed for expression of PDGFRα and vimentin. Vimentin^highPDGFRα^high were further gated for CD90. For IL-13 analysis, CD45⁺ cells were first gated for IL-13. CD45⁺IL-13⁺ cells were next analyzed for Siglec-F (eosinophil) and CD4 (CD4 T cell) lineages. Additionally, CD45⁺IL-13⁺Siglec-F^low cells were gated for CD11b expression. For immune populations the number of cells per 10⁵ single events is presented. For the analysis of fibroblasts, the number of cells gated was extrapolated to the total number of cells counted per esophagus.

Histology

For histologic analyses, esophagi were processed under standardized conditions in the histology CORE in La Jolla Institute for Immunology. Whole length esophagi were cut in 4-6 equal segments to obtain comparable histologic cuts for each esophagus. Tissues were formalin-fixed, paraffin-embedded and stained with H&E to assess inflammation and the presence of eosinophils, or with Masson’s Trichrome to assess collagen deposition. Quantification of collagen abundance and muscle density was performed using ImageJ (Bethesda, ME) and an in-house developed methodology. In brief, images were converted to RGB, colors separated and the appropriate channel isolated. Monochromatic images were then adjusted for signal threshold and converted to black over white background. For muscle quantification, a drawing tool was used to delineate the muscularis and separate the signal from other tissue areas. For collagen quantification, the signal from both the muscularis and submucosa was combined. The percentage of trichrome stained area was then calculated and compared.

Protein and gene expression analyses

For protein extraction from mouse esophagi, snap frozen specimens were thawed, chopped in ice cold radio-immunoprecipitation assay buffer (RIPA) and homogenized using a tissue-lyser (Qiagen, Hilden, Germany). Protein was normalized using a Bradford-based methodology and equal amounts of protein used for target analysis. ELISA for IL-13, CCL11, IL-5 (R&D systems, Minneapolis, MN) and TGF-β1 (Sigma, St Louise, MO) were performed according to the manufacturers indications. For gene expression analysis, cells were lysed in trizol and RNA extracted using a phenol-chloroform-based method. RNA was reverse transcribed to obtain cDNA using a QuantiTect reverse transcription kit (Qiagen, Hilden, Germany) and gene expression analyzed by real time-PCR. Primers were: RPL13A (FWD, CATAGGAAGCTGGGAGCAAG; REV, GCCCTCCAATCAGTCTTCTG); ICAM1 (FWD, TTGTTGGGCATAGAGACCCC; REV, GGTTTTAGCTGTTGACTGCCC); IL32 (FWD, AGAGCTGGAGGACGACTTCA; REV, CTCGGCACCGTAATCCATCT); CXCL5 (FWD, CACGCAAGGAGTTCATCCCA, REV, TCCTTCCCGTTCTTCAGGGA); IL6 (FWD, AGATGTAGCCGCCCCACACA; REV, TTTTCACCAGGCAAGTCTCCTCA); ACTA2 (FWD, CCGACCGAATGCAGAAGGA; REV, ACAGAGTATTTGCGCTCCGAA); SERPINE1 (FWD, TGGTTCTGCCCAAGTTCTCCCTG; REV, TGCCACTCTCGTTCACCTCG); IL4RA (FWD, TTCTGCTCTCCGAAGCCCAC; REV, CTGGGTTTCACATGCTCGCT); GATA6 (FWD, ACCACCTTATGGCGCAGAAA; REV, ATAGCAAGTGGTCTGGGCAC); CCL26 (FWD, CCCTCCTGAGTCTCCACCTT; REV, GAAGGGGCTTGTGGCTGTAT); LOXL3 (FWD, CACGGATGTGAAGCCAGGAA; REV, CGTTCAAACCTCCTGTTGGC); LOXL4 (FWD, TTGCGCTTCTCCACACAGAT; REV, CAGCTATCGCGTCCAGTCTT); POSTN (FWD, GTCTTTGAGACGCTGGAAGG; REV, AGATCCGTGAAGGTGGTTTG); CCL11 (FWD, AACTCCGAAACAATTGTACTCAGCTG; REV, GTAACTCTGGGAGGAAACACCCTCTCC)

Human Cells and RNA sequencing

Human primary esophageal fibroblasts were obtained from esophageal mucosa of healthy donors (Arkansas Regional Organ Recovery Agency). For experiments analyzing cell proliferation, cells were untreated (phosphate buffered saline, PBS) or treated with 50 ng/ml of LIGHT and incubated with 10 μM BrdU for 24 hours prior to collection and staining (see flow cytometry section). The cells were permeabilized, fixed and stained using a FITC-BrdU flow cytometry kit according to manufacturer’s recommendations (BD Biosicences, Franklin Lakes, NJ). For analysis of IL-13 and LIGHT-mediated gene expression, cells were untreated (PBS) or stimulated with 50 ng/ml of either cytokine or a combination of both for 24 hours. After treatment, the cells were lysed in trizol and RNA extracted using a phenol-chloroform-based method. RNA sequencing was performed according to our previously described methods (18, 19). Data from LIGHT-stimulated cells was previously submitted (GSE143482). Data from IL-13-stimulated fibroblasts can be found in GSE212731 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE212731). RNA was sequenced at the Genomics CORE in La Jolla Institute for Immunology.

Statistical analysis

Multiple comparisons were preformed using one-way ANOVA analysis of variance with a Newman-Keuls or Turkey’s multiple comparison post-test. Statistical significance was considered when the p-value was below 0.05. In mouse experiments, each dot represents an independent mouse. In cell experiments, each dot represents an independent donor unless otherwise indicated.

RESULTS

LIGHT-deficient mice display normal esophageal eosinophilia but reduced accumulation of CD4 T cells in a murine model of eosinophilic esophagitis.

To address the involvement of LIGHT in inflammation in the esophagus we established a murine model of EoE using the allergen house dust mite (HDM). Previous studies have shown that intranasal challenge with several antigens can induce esophageal eosinophilia and inflammation (10, 11, 25). We used a chronic protocol in which mice were sensitized with two rounds of intra-peritoneal HDM and alum adjuvant in one week, followed by an acute elicitation period of 4 intranasal HDM challenges in one week and then a chronic elicitation period of 2 challenges per week for three weeks (Figure 1A). Hematoxylin and eosin staining showed evidence of submucosal esophageal inflammation and the presence of eosinophils that accumulated near the basal epithelium (Figure 1B). Flow cytometry analysis of CD45⁺ cells revealed both CD3⁺ and CD3⁻ populations (Figure 1C). The CD45⁺CD3⁻ population contained CD11b^high cells, of which most were CD11c^low (Figure 1D and Supplementary Figure 1A). The CD11b^highCD11c^low cells contained a Siglec-F⁺ eosinophil population and a Siglec-F⁻ population (Figure 1D). These were seen in control mice esophagi, and WT HDM-exposed mice had a significant increase in both CD11b^highCD11c^low cells and Siglec-F⁺ eosinophils, and a similar trend for total CD11b^high cells, although the latter was not statistically significant (Figure 1D, F, G, and Supplementary Figure 1A–B). The CD45⁺CD3⁺ population was a mixture of CD4⁺ T cells and CD8⁺ T cells (Figure 1E, and Supplementary Figure 1C). WT HDM-exposed mice had increased numbers of CD4⁺ T cells, with a trend to more CD8⁺ T cells albeit not significant (Figure 1E, H, and Supplementary Figure 1C–D). There was also a relatively high number of MHCII-expressing cells, and to a lesser extent CD117-expressing cells, in the esophagus, but the number of cells with these markers did not change with HDM exposure (Supplementary Figure 1E–F).

To address whether LIGHT played a role in esophageal inflammation, LIGHT−/− mice were then exposed to HDM and compared to WT mice. Interestingly, LIGHT−/− had similar enhanced numbers of total CD11b⁺ cells, CD11b⁺CD11c⁻ cells, and Siglec-F⁺ eosinophils that accumulated in the esophagus compared to those found in WT mice (Figure 1F–G and Supplementary Figure 1B). In contrast, an analysis of the number of T cells in HDM-exposed mice showed a statistically decreased number of CD4⁺ T cells in esophagi of LIGHT−/− mice, with a trend toward fewer CD8⁺ T cells although with the caveat that CD8 T cell accumulation was quite variable in WT mice (Figure 1H and Supplementary Figure 1D). Thus, our model recapitulates some key features of the inflammatory process seen in human EoE, including infiltration of eosinophils and T cells to the esophagus, and a loss of LIGHT impaired T cell but not eosinophil accumulation.

IL-13 levels are reduced in the esophagus of LIGHT−/− mice challenged with HDM.

Several inflammatory factors are thought important in human EoE, including eotaxins, IL-5 and IL-13 (12, 13, 26). After 4 weeks of HDM challenges in WT mice, we found a significant increase in levels of IL-13 and CCL11 (eotaxin-1), but not TGF-β1, while IL-5 was undetectable, in esophageal protein extracts (Figure 2A–C, and not shown). In LIGHT-deficient mice, a pronounced reduction in the abundance of IL-13 was observed, significantly lower than that found in WT mice after exposure to HDM (Figure 2A). In contrast, CCL11, one of the chemokines involved in eosinophil migration, was not different between WT and LIGHT−/− mice exposed to HDM, in line with no difference being observed in eosinophil accumulation (Figure 2B).

Figure 2. LIGHT−/− mice display reduced HDM-driven IL-13 in the esophagus.

A-C, IL-13, CCL11 and TGF-β1 protein levels in esophageal protein extracts from wild type mice exposed to PBS, or wild type and LIGHT−/− mice exposed to HDM (data from 2 independent experiments is presented, each dot corresponds to a single mouse). D-E, representative flow and quantification of CD45⁺IL-13⁺ cells; F-G, representative flow and quantification of CD45⁺IL-13⁺CD4⁺ cells; H-I, representative flow and quantification of CD45⁺IL-13⁺Siglec-F⁺ cells; and J-K, representative flow and quantification of CD45⁺IL-13⁺CD11b⁺ cells, in the esophagus of mice treated as in A-C. Each dot corresponds to a single mouse, *p<0.05, **p<0.01 or ***p<0.001.

We next used flow cytometry to understand whether the IL-13 reduction was associated to a specific immune population. A distinct IL-13⁺ population was observed within CD45⁺ cells in the esophagus of WT mice challenged with HDM (Figure 2D). Comparison of CD45⁻ and CD45⁺ populations in WT mice revealed only CD45⁺IL-13⁺ cells induced by HDM (Supplementary Figure 2A). Consistent with our protein analysis, frequencies of CD45⁺IL-13⁺ cells were increased in WT mice after exposure to HDM, but significantly this was not observed to the same extent in LIGHT−/− mice (Figure 2D and E). Further analysis within this CD45⁺IL-13⁺ population showed that approximately half of the cells were CD4⁺ T cells. These were increased in WT mice, although variability from mouse to mouse meant this difference did not reach statistically significance (Figure 2F–G). Fewer CD4⁺IL-13⁺ cells were seen in LIGHT−/− mice (Figure 2G). The mean fluorescence intensity of IL-13 in CD4⁺ cells was however not different between WT and LIGHT−/− mice challenged with HDM (Figure 2F and not shown). These data collectively imply that the deficiency in LIGHT did not impair differentiation into a Th2 cell but, rather, altered the accumulation and/or survival of Th2 cells in the esophagus.

Although non-CD4 cells accounted for approximately 50% of the IL-13-staining CD45⁺ cells, the intensity of fluorescence for IL-13 was reduced in these cells, implying they were likely to contribute less IL-13 than the CD4⁺ T cells (Figure 2F, H, J). Few CD8⁺ T cells were contained within this CD45⁺IL-13⁺CD4⁻ population (Supplementary Figure 2B), and staining for Siglec-F also revealed only a low number of eosinophils expressing IL-13, with no statistical increase after HDM challenge (Figure 2H–I). In contrast, the majority of other CD45⁺CD4⁻ cells that stained for IL-13 were CD11b⁺ (Figure 2J). These were increased in WT mice challenged with HDM, and their numbers were lower in LIGHT−/− mice (Figure 2K). The nature of these cells is not known, but they could be macrophages that have been reported in some situations to make IL-13. These data then suggest that in the mouse esophagus, IL-13 may derive from both CD4⁺ and CD4⁻ cells that are induced in response to allergen. Accumulation of both IL-13-producing populations were controlled by LIGHT, although it is likely that the primary influence of LIGHT on IL-13 production was explained by its effect on CD4⁺ T cells.

Chronic HDM exposure induces esophageal remodeling that is attenuated in LIGHT-deficient mice.

Increased extracellular matrix deposition and muscle hypertrophy are features of remodeling in EoE patients (2, 27, 28). In our EoE model, WT mice exposed to HDM had increased submucosal collagen deposition, as well as evidence of more collagen in the muscularis externa, as shown by Masson’s Trichrome staining (Figure 3A). To quantify the abundance of collagen and compare it between WT and LIGHT−/− mice, we developed an ImageJ-based method applying filters to isolate the collagen signal and selecting specific areas of interest for quantification (Figure 3B). This analysis revealed an increase in collagen in the submucosa and muscularis regions combined in WT mice exposed to HDM, whereas significantly less collagen was apparent in LIGHT−/− mice exposed to HDM (Figure 3B–C). We also quantified the density of the muscularis externa layers (Figure 3D–E). The muscularis of WT mice exposed to PBS had numerous empty spaces in between the muscle fibers (white spaces among black signal), but HDM-exposed WT mice exhibited denser muscle layers with fewer spaces (Figure 3D–E). Interestingly, LIGHT−/− mice exposed to HDM had less compacted muscle layers, with numerous empty spaces in the muscularis indicating reduced muscle density (Figure 3D–E). These data suggest that prolonged exposure to HDM causes structural alterations of the esophagus that are attenuated in LIGHT−/− mice.

Figure 3. LIGHT−/− mice have attenuated esophagus submucosal collagen and muscle density after allergen exposure.

A, representative 10X (top) and 20X (bottom) images of mouse esophagi stained with Masson’s trichrome blue from WT mice exposed to PBS, or WT and LIGHT−/− mice exposed to HDM. B, representative images of mouse esophagi stained with Masson’s trichrome blue after image processing to highlight stained area, and C, quantification of collagen staining. (D) representative images of mouse esophagi after image processing to highlight muscle layers, and E, quantification of muscle density. Data from 3 independent experiments, each dot a single mouse (*p<0.05, **p<0.01 or ***p<0.001).

A LIGHT-deficiency reduces fibroblast abundance caused by HDM exposure in the esophagus.

Fibroblasts are one of the primary sources of collagens and other matrix proteins. To address the abundance of fibroblasts in our murine EoE model, we used flow cytometry. We have previously shown that CD90 and vimentin (VIM) are expressed in human esophageal fibroblasts, while high expression of platelet-derived growth factor receptor (PDGFR)-α has been found to be a marker of fibroblasts in tissues such as the colon (18, 29). Stromal lineage cells were analyzed within a triple negative gate of CD45⁻/CD31⁻/EpCAM⁻ cells (Figure 4A), showing an abundant and distinct population of cells expressing high levels of VIM and PDGFRα corresponding to a fibroblast phenotype (Figure 4B). VIM^high PDGFRα^high cells were further found to segregate into two subpopulations based on expression of CD90 (high or low, Figure 4C). Exposure to HDM increased the total number of VIM^high PDGFRα^high fibroblasts per esophagus in WT mice, but this increase in number was lower in LIGHT−/− mice when compared to PBS controls (Figure 4D and E). Both CD90^high and CD90^low cells were elevated within the VIM^highPDGFRα^high population in wild type mice, but we found fewer CD90^low cells accumulated in LIGHT−/− mice (Figure 4E and not shown). This was not a developmental defect, in that in the absence of HDM exposure, the basal number of VIM^highPDGFRα^high and VIM^highPDGFRα^highCD90^low cells was comparable between WT and LIGHT−/− mice (Supplementary Figure 3).

Figure 4. LIGHT regulates esophagus fibroblast accumulation.

A-C, representative flow plots of gating to identify populations of fibroblasts, comparing the expression of VIM, PDGFRα and CD90 within CD45-CD31-EpCAM− cells. D-E, quantification of the number of VIM+PDGFRα+ cells and VIM+PDGFRα+CD90^low cells within the CD45-CD31-EpCAM− population in the esophagus of WT mice exposed to PBS, or WT and LIGHT−/− mice exposed to HDM. Data from 3 independent experiments, each dot a single mouse (*p<0.05, **p<0.01 or ***p<0.001). F-G, analysis and quantification of BrdU uptake and PDGFRα expression in human esophageal fibroblasts cultured with PBS or LIGHT from three independent donors (p<0.05).

CD90^high fibroblasts, that produce epithelial-sustaining mediators such as WNT2B, are reduced in number in ulcerative colitis, suggestive of a protective role for CD90^high cells, and a potential pathogenic role for CD90^low cells (30, 31). Thus, the changes in fibroblast populations in LIGHT−/− mice might be indicative of alterations in stromal cell activity associated with reduced disease. Fibroblasts from multiple organs in the mouse and human express the receptors for LIGHT, namely HVEM and LTβR. Thus, the reduction in CD90^low cells might have suggested that LIGHT directly regulated the survival and/or division of these fibroblasts that would determine their accumulation in the esophagus. To address this, we used fibroblasts from the esophagus of three independent human healthy donors. Fibroblasts were cultured with PBS or recombinant LIGHT and incorporation of BrdU was assessed over 24 hours. Cells exposed to LIGHT had a two times higher rate of BrdU uptake, with the majority of the BrdU⁺ cells detected within the PDGFRα^high gate (Figure 4F–G). Thus, the accumulation of subpopulations of fibroblasts in the esophagus after HDM exposure is at least in part dependent on expression of LIGHT, and promoting proliferation is a potential mechanism whereby LIGHT might contribute to expansion of a specific PDGFRα^high fibroblast population.

LIGHT and IL-13 cooperate to promote select inflammatory and fibrotic responses in human esophageal fibroblasts

We showed previously that LIGHT induces transcripts and protein expression of a number of inflammatory molecules in human esophageal fibroblasts, including ICAM-1, IL-32, CXCL5, and IL-6, but had no obvious effect on ECM proteins (18, 19), implying that the changes we observed in collagen in the LIGHT−/− mouse esophagus might be due to fewer pathogenic fibroblasts accumulating and/or an indirect effect related to the reduction in IL-13. A comparison by RNA-seq of the transcriptional responses induced by LIGHT and IL-13 in human esophageal fibroblasts showed little overlap in the genes targeted by each cytokine alone, with only 12 common transcripts upregulated (Figure 5A). This included GATA6, a transcription factor previously linked to induction of myofibroblast differentiation (32), and IL4RA, implying the potential to increase responsiveness of fibroblasts to IL-13. As reported in other publications (26), transcripts upregulated by IL-13 included key inflammatory chemokines involved in EoE (CCL11 and CCL26), collagens (COL5A2, COL6A1, COL6A2 and COL7A1), matrix metalloproteases (MMP15 and MMP17) and other fibrotic mediators such as periostin (POSTN) or lysyl oxidase-like proteins (LOXL3) (Figure 5B). These were not significantly upregulated by LIGHT alone.

Figure 5. LIGHT enhances the effects of IL-13 in promoting transcription of a select number of genes in esophageal fibroblasts.

A-B, venn diagram comparing the number of upregulated transcripts induced by LIGHT and IL-13 with respect to PBS, in human esophageal fibroblasts (>1.5 fold, padjust<0.05), and showing the common gene targets (A), and heat map comparing the effects of PBS, LIGHT and IL-13 on selected genes (B). C-G, RT-PCR analysis of gene expression induced by LIGHT, or IL-13, or a combination of LIGHT and IL-13, after 24 hours of treatment, in human esophageal fibroblasts. Cells from at least 3 independent donors were used, each dot represents data from an independent experiment (*p<0.05, **p<0.01 or ***p<0.001).

To understand if LIGHT and IL-13 have co-operative activities, we treated human esophageal fibroblasts with a combination of both cytokines and assessed induction of a select group of the above molecules by PCR. Firstly, we assessed the expression of two of the common gene targets, GATA6 and IL4RA. Correlating with the RNA-seq data, transcription of GATA6 was significantly increased by LIGHT or IL-13 and this was further enhanced when both cytokines were combined (Figure 5C). By PCR we did not observe a statistically significant increase in IL4RA. LIGHT in isolation had no effect on the IL-13-regulated chemokines CCL11 and CCL26, but interestingly, it selectively and significantly enhanced the effects of IL-13 on CCL26 expression but not on CCL11 (Figure 5D). To determine if the cooperative effect seen with GATA6 and CCL26 was present for other genes, we assessed the expression of several LIGHT target genes. IL-13 alone, unlike LIGHT alone, did not induce transcripts for IL32, ICAM1, or CXCL5, and did not enhance the actions of LIGHT in this regard (Figure 5E). IL-13 alone did induce IL6 transcripts similar to LIGHT, and there was an additive effect of combined cytokine treatment on IL6 expression (Figure 5E). We further analyzed the expression of several fibrosis-related molecules. Similar to CCL26, POSTN was uniquely regulated by IL-13 in isolation, but its expression was enhanced by the combination of LIGHT and IL-13 (Figure 5F). LOXL4 mRNA was also greatest with combined LIGHT and IL-13 treatment, although this gene was also induced by either cytokine in isolation, whereas LOXL3 and COL6A2 were uniquely up-regulated by IL-13 and not affected by LIGHT (Figure 5F). Interestingly ACTA2, that encodes the myofibroblast marker alpha-smooth muscle actin (α-SMA), was not significantly increased by either LIGHT or IL-13 alone, but was induced when these cytokines were combined, whereas SERPINE1 (which encodes the plasminogen activator inhibitor-1) was only regulated by LIGHT in isolation and IL-13 had a moderate effect in enhancing this expression (Figure 5G). Collectively these results suggest a potential disease-driving cooperative effect between LIGHT and IL-13 in select fibroblast responses in EoE.

DISCUSSION

In recent years, many advances have been made in our understanding of various molecules upregulated in EoE patient samples and transcriptional signatures that associate with disease. However, studies of patient tissue and in vitro cell cultures are still limited in being able to more definitively predict the likely crucial drivers of the inflammatory and fibrotic responses in the esophagus. In this manuscript, we add to our recent publications describing the potential significance of LIGHT to human EoE by showing in a model of allergen-mediated eosinophilic esophageal inflammation that LIGHT-deficient mice have an attenuated disease phenotype. This was characterized by decreased esophageal T cell infiltration, lower levels of IL-13, improved structural remodeling, and an attenuated fibroblast response.

Intranasal administration of airway allergens was first shown to promote eosinophilic inflammation in the mouse esophagus by Mishra and colleagues, in studies comparing administration of A. Fumigatus, dust mites, and ovalbumin by either oral instillation, intragastric release, or intranasal inhalation (10). These experiments demonstrated that intranasal administration induces the greatest esophageal eosinophilia, and other authors have also shown eosinophilic infiltration in the esophagus after exposure to similar intranasal antigen preparations (11, 25, 33). These models have been instrumental for the study of the contribution of the type 2 cytokines, IL-5, IL-13, and IL-33, to the recruitment and activation of eosinophils, and more recently, to the study of the impact of esophageal epithelial permeability in EoE. Although less data is available on other disease features such as esophageal fibrosis, a recent study showed increased deposition of fibronectin in the submucosa of mice chronically exposed to peanut extract (11). The data presented here similarly shows that repetitive inhalation of HDM also induces recruitment of eosinophils, accompanied by increases in eotaxin (CCL11) in the esophagus. In addition, our model was also characterized by an increased infiltration of T cells into the mouse esophagus and increased collagen deposition and muscle density, and thus provides a reasonable representation of some aspects of the disease seen in humans. Accumulation of CD4⁺ and CD8⁺ T cells were observed in the esophagus, and CD4⁺ T cells were potentially the greatest source of IL-13 in the model, which is in line with previous studies analyzing T cell diversity in the esophagus of patients with active EoE (34). Our histologic analyses found that eosinophils were mostly located to the submucosa, which is in contrast to the largely epithelial eosinophilia detected in human EoE, but in line with what has been observed by various authors in other murine EoE models.

We first connected the cytokine LIGHT with the process of fibrosis and tissue remodeling from studies of intranasal administration of a chronic regime of HDM that induced asthmatic lung inflammation, and demonstrated that a loss of LIGHT expression in gene-deficient animals, or blocking LIGHT therapeutically, reduced several of these features in the lungs (35). Similarly, intranasal administration of bleomycin elicited a lung fibrotic response that was strongly dependent on LIGHT (36), and other studies with skin inflammation models have supported a central role for LIGHT in tissue remodeling (23, 37). Given the broad expression of LIGHT’s receptors, HVEM and LTβR, on hematopoietic and non-hematopoietic cells, its targets of action to drive these responses could be multiple cell types, including T cells, eosinophils and fibroblasts. Interestingly, our previous asthma studies showed that the number of airway eosinophils in the lungs of HDM-exposed mice was not affected in LIGHT−/− mice, but that lung eosinophils were defective in their expression of IL-13, reflective of a direct activity of the LIGHT receptor HVEM that is expressed on these cells (35). In the current study, we again found no difference in eosinophilia in the absence of LIGHT, although notably few eosinophils recruited to the esophagus expressed IL-13, suggesting that these eosinophils were in a relatively less differentiated state. Whether additional intranasal challenges with HDM for an extended period of time will result in significant numbers of IL-13⁺ eosinophils in the esophagus is not clear, but this would be interesting to study in the future. Regardless, we did find that IL-13 was strongly upregulated in the esophagus in WT mice, but its primary source appeared to be CD4⁺ T cells, based on the fluorescence intensity of staining of this cytokine in these cells together with the increased numbers of CD4⁺ T cells seen in the esophagus. CD11b⁺ cells were also found to make IL-13, and their numbers were increased with allergen challenge, but the potential contribution to total IL-13 might have been less as suggested by the flow cytometry data revealing a lower intensity of staining for IL-13. Most interestingly, LIGHT−/− mice had reduced levels of total esophageal IL-13, corresponding to lower numbers of CD4⁺ T cells, including those that were Th2-like and made IL-13, as well as reduced numbers of CD11b⁺ cells making IL-13. The likely reason for reduced CD4 T cell infiltrates is that HVEM is constitutively present on all T cells, regardless of differentiation status, and that its signals from LIGHT can control either T cell expansion or T cell survival, a phenomenon previously shown for effector memory Th2 cells responding to allergen (24) and in various model systems for other CD4 or CD8 T cells (20, 38, 39). We cannot however rule out from the current experiments indirect effects of LIGHT that might have modulated the T cell infiltrate in the esophagus. For example, it is possible that LIGHT may regulate the production of T cell recruiting chemokines such as CCL17 and CCL22, a possibility that has not been addressed here. The identity of the CD11b⁺ cells producing IL-13 was also not fully elucidated in the present manuscript, but they might be macrophages which are known to produce IL-13 in some situations. Macrophages also express both receptors for LIGHT (35), and again a direct effect on these cells is possible.

The source of LIGHT in our model system is not known. LIGHT was first found expressed in activated human T cells (21), both CD4+ and CD8⁺, and these cells are the likely major source in many situations. Our previous published (24) and unpublished murine studies with adoptive transfer of LIGHT-deficient T cells have supported this notion, although other cell types such as NK cells, dendritic cells, and neutrophils are also capable of making this cytokine. LIGHT could function in this scenario in an autocrine manner, or paracrine between one T cell and another, as we showed in lung studies focused on effector memory Th2 cells (24). Recent single cell analysis of T cells from esophagus biopsies of EoE patients described the existence of several T cell subclusters, including a population of CD4⁺ T cells that produced IL-5 and IL-13 (34), and in line with the aforementioned, we found that LIGHT was transcribed at levels similar to these other two cytokines in that CD4 T cell population, as well as being expressed in esophagus infiltrating CD8 T cells (18).

IL-13 is thought to be central to EoE, and the relevance of IL-13 to the pathogenesis of allergic diseases is well illustrated by promising clinical results attained with anti-IL-4Rα and anti-IL-13 antibodies in EoE, asthma, and atopic dermatitis (40–42). Additionally, several reports suggest that IL-13 can contribute to fibrosis via indirect as well as direct pro-fibrotic effects on fibroblasts or by enhancing the effect of other cytokines such as TGF-β1 (14, 43, 44). We have previously shown abundant expression of LIGHT’s receptors HVEM and LTβR in human esophageal fibroblasts and demonstrated that LIGHT also has profound inflammatory effects on these cells. Our studies now additionally show that IL-13-driven expression of select inflammatory and fibrotic mediators in esophageal fibroblasts is enhanced by combined stimulation with LIGHT, as well as each molecule driving distinct responses in these cells, reminiscent of what we previously reported for lung fibroblasts (22). With LIGHT possessing the ability to promote IL-13 production from CD4 T cells, eosinophils, and mast cells (24, 35, 45), either by a direct activity or through regulating the numbers of these cells, we then provide further support of an important role for a LIGHT-IL-13 axis in the promotion of inflammation and fibrosis in many allergic diseases, including EoE. The action of LIGHT enhancing the expression of the IL-13-driven transcripts CCL26 and POSTN (periostin) is interesting. CCL26 is strongly involved in eosinophil recruitment and activation, and although we found no evidence that LIGHT can directly affect eosinophilia in the current experiments, it may potentiate IL-13-driven eosinophilia in other situations or when responses become more chronic. Similarly, the ability of LIGHT to enhance periostin expression may contribute to the fibrotic response found in the esophagus, although we did not investigate in situ periostin expression in the current studies. However, this is a possibility in that our previous results in a murine model of atopic dermatitis demonstrated that LIGHT was crucial for maximal expression of periostin in the skin (23). Collectively, this adds to our previous studies showing that aspects of the LIGHT-induced transcriptome in esophageal fibroblasts are conserved in active EoE patients and patient-derived fibroblasts, and that a network between LIGHT and TGF-β1 also promotes an alternate pathogenic esophageal fibroblast differentiation state (18, 19). Moreover, we found that LIGHT promotes esophageal fibroblast proliferation in vitro, and thus factors such as LIGHT that might influence proliferation as well as differentiation of fibroblasts are likely to be important for the extracellular matrix remodeling and rigidity seen in the esophagus of EoE patients with more chronic disease.

Lastly, we found that LIGHT-deficient mice displayed evidence of reduced muscle density, shown by numerous empty spaces in the muscularis layers. An increase in the muscle thickness of the esophagus has been seen in human EoE patients, and is thought to additionally contribute to difficulties in swallowing (7, 28, 46). An increase in collagen deposition may have contributed to increased muscle thickness in WT mice, and explained the defect in LIGHT−/− mice, but direct effects of LIGHT on smooth muscle or skeletal muscle cannot be ruled out at present. Whether one or both receptors for LIGHT are expressed on esophagus muscle in either the mouse or in humans is not known, but this might be interesting to examine in future studies.

In summary, we show that LIGHT is a contributor to esophageal inflammation and fibrosis in a model of murine EoE. We show that the esophagi of LIGHT-deficient mice exposed to inhaled allergen have attenuated T cell infiltration, reduced IL-13 expression, and decreased fibrosis associated with altered fibroblast responses. Together with our previous studies in human fibroblasts and patient biopsies, the evidence presented in this manuscript suggests that LIGHT will be a key mediator of EoE and a potential target for therapy.

KEY POINTS.

• LIGHT promotes esophageal T cell infiltration and IL-13 levels in a model of EoE

• LIGHT regulates esophageal tissue remodeling in a model of EoE

• LIGHT drives esophagus fibroblast proliferation and is synergistic with IL-13