The barrier hypothesis and Oncostatin M: Restoration of epithelial barrier function as a novel therapeutic strategy for the treatment of type 2 inflammatory disease

ABSTRACT

Mucosal epithelium maintains tissue homeostasis through many processes, including epithelial barrier function, which separates the environment from the tissue. The barrier hypothesis of type 2 inflammatory disease postulates that epithelial and epidermal barrier dysfunction, which cause inappropriate exposure to the environment, can result in allergic sensitization and development of type 2 inflammatory disease. The restoration of barrier dysfunction once it's lost, or the prevention of barrier dysfunction, have the potential to be exciting new therapeutic strategies for the treatment of type 2 inflammatory disease. Neutrophil-derived Oncostatin M has been shown to be a potent disrupter of epithelial barrier function through the induction of epithelial-mesenchymal transition (EMT). This review will discuss these events and outline several points along this axis at which therapeutic intervention could be beneficial for the treatment of type 2 inflammatory diseases.

KEYWORDS:

• epithelial barrier function
• epithelial-mesenchymal transition
• neutrophil
• Oncostatin M
• type 2 inflammation

The barrier hypothesis

The role of mucosal epithelium is to maintain tissue homeostasis and defense. It achieves this through mucociliary clearance, anti-microbial peptide production, maintenance of air-surface liquid, regulation of ion permeability, and physical barrier function.¹ Epithelial barrier function is necessary for tissue homeostasis because it acts as a first line of defense against pathogens, allergens and environmental factors through the creation of a physical separation between the environment and the tissue. Epithelial barrier function is largely determined by the structure of intercellular epithelial junctions, including tight junctions, adherens junctions and desmosomes.² When the epithelial barrier is breached, pathogens, allergens and environmental factors can synergize with alarmins and pathogen- or danger-associated molecular patterns (PAMPs and DAMPs) that transit across the barrier and/or are released in response to cellular damage, resulting in the activation of innate and adaptive immune responses.³ The immune system's ability to neutralize the pathogenic stimulus and repair the resulting tissue damage is necessary for the restoration of the tissue to a normal homeostatic state. However, if barrier function is not restored, it may result in the development of an unnecessary chronic inflammatory state within the tissue and potential allergic sensitization via continued exposure facilitated passage of allergens across the defective barrier. The barrier hypothesis postulates that epithelial or epidermal barrier dysfunction thereby precedes the development of allergic sensitization and type 2 inflammatory disease.

Mucosal epithelium refers to epithelium from the airways, gastrointestinal tract, eyes and genitourinary tract, among others. Differentiated epithelium has a polarized structure where undifferentiated basal cells lie along the basement membrane and differentiated cells line the apical surface. This review will focus on airways and esophageal epithelium. Airways epithelium has a pseudostratified columnar structure, consisting of at least 3 common cell types, basal cells, ciliated cells, and goblet cells. Basal cells are undifferentiated progenitor cells that are attached to the basement membrane and express keratin 5, which can be used as a marker of undifferentiated epithelium.⁴ These cells will proliferate and migrate to cover a wound in the event of injury or infection.⁵ Ciliated and goblet cells are the differentiated cells within airway epithelium which lie along the apical surface of the epithelial cell layer. These cells express keratin 8, which can be used as a marker of differentiated airway epithelial cells.⁶ Goblet cells make mucins, which are released into the airway lumen to trap pathogens, allergens and other environmental factors that may otherwise result in injury or immune response. Ciliated cells contain cilia, comprised of acetylated α-tubulin, which beat in a coordinated fashion to clear the airways of the mucus and its trapped contents.⁷ Together these differentiated airway epithelial cells are responsible for ion transport, maintenance of air-surface liquid, mucus production, mucociliary clearance and barrier function.^5,8 Esophageal epithelium has a stratified squamous structure that consists of several layers, in which cells become more differentiated closer to the apical layers.⁹ The stratum basale comprises basal cells and lies along the basement membrane; these cells are undifferentiated progenitor cells.¹⁰ The most apical layer is the stratum corneum, which comprises dead, flattened cells that have lost their nuclei and organelles. While the stratum corneum of epidermis is keratinized, mucosal stratified squamous epithelium, as in the esophagus, has a non-keratinized stratum corneum.¹¹

The barrier function of airways and esophageal epithelium is largely determined through the apical junction complex, which includes tight junctions and adherens junctions.^5,12 Tight junctions localize to the apical edge of the epithelial layer, just adjacent to the lumen,¹³ and are comprised of intercellular proteins that self-adhere in the intercellular space, effectively sealing that area and regulating paracellular flux. These proteins include occludin, junctional adhesion molecule (JAM) family proteins, and claudin family proteins.¹⁴ These protein complexes are tethered to the actin cytoskeleton through adaptor proteins including the zonula occludens (ZO) protein family. The tight junction and actin cytoskeleton form a tight band along the apical edge of the cell layer that is necessary for proper barrier function.^15,16 Adherens junctions localize to the basal border of the tight junction complex and are also tethered to the actin cytoskeleton.⁵ These junctions consist of the intercellular self-adherent protein, E-cadherin, that maintains calcium dependent adhesion between epithelial cells. Adherens junctions also are comprised of accessory proteins, α-catenin, cateninδ1, and β-catenin, which collectively can act as a transcription factor upon disassembly of the junction.^17-21

This junctional structure is necessary for proper epithelial barrier function and junctional failure can lead to disease. For example, asthma is classically thought to involve a predisposition or skewing to type 2 inflammation, versus type 1 inflammation, that promotes persistent eosinophilic inflammation of the airways and subsequent tissue remodeling. However, this type 2 skewing model fails to fully explain many clinical features of asthma, including the heterogeneity of asthma phenotypes and the lack of association between airway eosinophilia and airway hyper-responsiveness.^22-26 The barrier hypothesis could help to explain some disease features that the type 2 skewing model does not because epithelial barrier dysfunction is proposed to be upstream of the generation of an immune response and could potentially result in type 1 or type 2 inflammation. Epithelial dysfunction has been demonstrated in asthma patients compared with control patients. Asthma patients have decreased expression of tight and adherens junction proteins and their epithelia have decreased barrier function as assessed by transepithelial electrical resistance (TEER) and dextran permeability across monolayers.^27-29 Additionally, reduced caveolin-1 expression in airway epithelium from asthma patients was shown to associate with decreased adherens junction expression, and promoted epithelial expression of the type 2-skewing cytokines TSLP and IL-33, suggesting that barrier dysfunction may initiate type 2 inflammation.²⁷ Thus, therapeutic intervention focused on either preventing barrier dysfunction, or restoring barrier once it is lost, may be effective in the treatment of type 2 inflammatory diseases, including asthma, chronic rhinosinusitis (CRS), eosinophilic esophagitis (EoE), and potentially allergic rhinitis and atopic dermatitis.

Type 2 inflammation has been shown to be important for immunity to helminths and toxins.^30,31 More recent studies have also shown that Type 2 immune responses are also important for the establishment of general tissue homeostasis.^32-34 Common features of type 2 inflammation include the expression of classical type 2 cytokines, IL-4, IL-5, and IL-13, tissue eosinophilia, mast cell activation and fibrosis. Prolonged type 2 inflammation can further promote or perpetuate allergic sensitization to otherwise innocuous food or aeroantigens that would normally be tolerated. Dysfunction of epithelial maintenance and repair of barrier has been implicated in the pathogenesis of many type 2 inflammatory diseases, including asthma, atopic dermatitis, eosinophilic esophagitis (EoE), and chronic rhinosinusitis (CRS).^35-38 Atopic dermatitis patients have been shown to express decreased levels of the tight junction protein, claudin 1 which negatively correlated with serum IgE and total eosinophil counts.³⁹ Recent genome-wide association studies (GWAS) have identified loss-of-function single nucleotide polymorphisms (SNPs) in the filaggrin gene that correlate with increased risk for the development of atopic dermatitis, allergic asthma, food allergy, and allergic rhinitis.^38,40-42 This observation is perhaps the best evidence so far in support of the barrier hypothesis. Filaggrin is a structural protein that is important for proper differentiation of the epidermal stratum corneum, which is the keratinized outer layer of the epidermis that is crucial for establishment of barrier function.^43,44 These loss-of-function SNPs result in impaired barrier function in the skin, leading to chronic inflammation and the potential generation of allergic sensitization.⁴⁵ Interestingly, airway epithelia do not express filaggrin, suggesting that the development of allergic asthma may be preceded by atopic dermatitis and subsequent cutaneous sensitization to aeroallergens or food allergens that drive disease at mucosal surfaces.^40,42 These observations and others provide evidence for the atopic march, which describes a characteristic progression of allergic disease in children that usually begins with the development of atopic dermatitis and allergic sensitization in infancy followed by food allergy in early childhood and finally allergic rhinitis and asthma later in childhood.⁴⁶

Healthy bronchial, nasal and esophageal epithelia are highly organized and polarized structures; healthy nasal epithelium is shown in Fig. 1A. However, epithelia from asthma, CRS and EoE patients have some common changes in structural elements including an expansion of basal cells, or acanthosis, associated with a loss of differentiation throughout the cell layer; acanthosis in nasal epithelium from a CRS patient is also shown in Fig. 1B. It is particularly striking that the morphology of the epithelia in these 3 different diseases has some similarities, potentially suggesting common mechanisms of epithelial dysfunction.^47,48 Many of these structural abnormalities are believed to lead to barrier dysfunction, and subsequent development of allergic sensitization, type 2 inflammatory disease or the exacerbation of established type 2 inflammation. A central theme of this review is that the barrier-disrupting cytokine, Oncostatin M (OSM), may play a role in mucosal allergic diseases affecting several different organs. Therapeutic interventions aimed at preventing barrier loss or restoring barrier function, either through targeting OSM or other mediators, could offer exciting alternative strategies for treating patients with type 2 inflammatory disease.

Figure 1. Epithelial morphology is abnormal in nasal polyps from CRS patients. Healthy nasal epithelium is a highly organized structure with undifferentiated progenitor cells along the basement membrane, and the differentiated cells, ciliated and goblet cells along the apical edge (A). However, in CRS the epithelium often has a large expansion of basal cells and is highly disorganized, which could lead to barrier dysfunction (B).

Oncostatin M and epithelial barrier function

OSM is a member of the IL-6 family of cytokines, which includes IL-11, IL-31 and leukemia inhibitory factor (LIF).⁴⁹ OSM has been shown to be expressed by many cell types of the haematopoietic lineage, including T cells, neutrophils, mast cells, macrophages and eosinophils.^50-53 OSM is not known to be expressed by epithelium, fibroblasts, or smooth muscle, which all express the 2 forms of the OSM receptor.⁵⁴ Human OSM signals through 2 heterodimeric receptors, both of which utilize glycoprotein 130 (gp130) for signaling. The type 1 OSM receptor is composed of the LIF receptor (LIFR) and gp130, and the type 2 OSM receptor is composed of the OSM receptor β-chain (OSMR) and gp130.^55,56 Of note, murine OSM does not bind the type 1 receptor, which is something to be considered regarding the implications of research done on OSM in mouse models.⁵⁷ The transcription factor STAT3 has been shown to be very important for epithelial homeostasis and repair.⁵⁸ Oncostatin M (OSM) is a potent activator of STAT3 and also activates other transcription factors, including STAT1, STAT5, PI3K, and the MAPK pathway through multiple tyrosine kinases including, janus kinase 1 (JAK1), janus kinase 2 (JAK2), or tyrosine kinase 2 (Tyk2).^49,59-64 OSM has been shown to have a wide range of physiologic effects, including in vitro inhibition of tumor growth, promotion of in vivo tumor invasiveness and metastasis, maintenance of bone metabolism, generation of joint inflammation and regulation of the hepatic acute phase response.^65-70 However, this review will focus on the effects of OSM that are related to epithelial repair and homeostasis, and inflammation.

OSM has been shown to be important in the repair of many tissue types. Biegel et al. have shown that OSM was important for regrowth after injury and that it decreased apoptosis in a STAT3 dependent manner in a colon carcinoma cell line. They also demonstrate elevated OSMR in inflamed colon biopsies from Crohn's disease and ulcerative colitis patients, which led them to suspect that OSM was important for epithelial repair and that it may be a potential drug target.⁷¹ In a chronic diabetes cutaneous wound model, the treatment of wounds with recombinant OSM decreased expression of proinflammatory cytokines, IL-1β and TNF, and improved wound closure during the early inflammatory phase of repair.⁷² In a model of cardiac repair, OSM mediated the dedifferentiation of cardiomyocytes, which is necessary for proper repair of cardiac injury.⁷³ Additionally, in a mouse demyelination spinal cord injury model, OSM induced many processes that are involved in repair of demyelination injuries including the mobilization of oligodendrocyte precursors, differentiation of oligodendrocytes, and the production of myelin.⁷⁴ These data suggest that OSM can play an important role in normal homeostatic tissue repair.

Epithelial damage repair occurs in 2 basic steps. First, basal cells proliferate and migrate into the wound to cover the damaged area, and second, the progenitor cells become contact inhibited, stop proliferating and differentiate back into ciliated or goblet cells. Intercellular junctions are then formed to reestablish barrier function in the repaired tissue.⁷⁵ The mechanism of OSM mediated barrier dysfunction in mucosal epithelium in vitro is currently unknown. OSM has been shown to be important for epithelial repair, as discussed above and OSM may be one signal that initiates the first step of repair following cellular damage through the induction of basal cell proliferation and migration to cover the wound. However, in vivo in a diseased state, either high levels of OSM, or the absence of a signal to initiate the second step of the repair process, could potentially prevent the transition to the second step of repair, leading to chronic proliferation of basal cells to create a state where differentiation and the generation of barrier function does not occur.

While OSM has been shown to be important for tissue repair, it has also been shown to have pathogenic effects in both type 1 and type 2 inflammation. In the context of type 1 inflammation, OSM treatment of dendritic cells (DC) induced polarization of Th1 cells through the induction of IL-12,⁷⁶ and Th1 cells can produce OSM.⁷⁷ OSM has been shown to play an important role in the autoimmune disease rheumatoid arthritis (RA). OSM is a biomarker of active RA compared with both quiescent RA and healthy control patients and it mediates its effects in RA through the modulation of matrix metalloproteinases.^78,79 OSM has also been shown to increase the expression of OSMR and IL-1R1 in synovial fibroblasts, thereby augmenting the pathogenic effects of OSM and IL-1 in RA.⁸⁰ In addition, OSM was sufficient to induce CCL13, a chemokine that is elevated in synovial fibroblasts of RA patients.⁶⁸

In the context of type 2 inflammation, previous studies have reported elevated levels of OSM in sinus tissue from allergic rhinitis patients, psoriatic skin and sputum of asthmatic patients with irreversible airflow obstruction.^81-83 Our group has shown elevated OSM in nasal polyps from CRS patients, in tissue biopsies and in induced sputum from asthma patients, and in biopsies from EoE patients compared with controls.^84,85 In a mouse model, intratracheal administration of adenoviral expressed OSM was shown to be sufficient to induce robust type 2 inflammation in the lung, without any specific antigen challenge.^86,87 Studies using reconstituted epidermis have shown that OSM treatment decreased expression of filaggrin, a molecule that is important for barrier function in keratinocytes, suggesting that it could also contribute to barrier dysfunction in the skin.⁸⁸ Our group has also shown that OSM was sufficient to induce epithelial barrier dysfunction and loss of tight junction structure in air-liquid interface cultures of differentiated nasal and bronchial epithelium. Additionally, we showed that levels of OSM correlated with levels of markers of epithelial leak in localized nasal secretions of CRS patients and in bronchoalveolar lavage fluids following allergen challenge in allergic asthmatic patients, suggesting that OSM may contribute to epithelial barrier dysfunction in vivo.⁸⁴

One potential mechanism for elevated OSM in type 2 inflammatory disease that has not been investigated is whether SNPs in the OSM or OSMR genes associate with the development of type 2 inflammatory disease. Gain of function SNPs in the OSM or OSMR gene could potentially result in elevated expression or increased receptor binding affinity or intrinsic activity that could amplify the effects of OSM and exacerbate type 2 inflammation, providing another potential mechanism of OSM mediated epithelial dysfunction. Alternatively, loss of function SNPs could be protective and confer reduced susceptibility to development of type 2 inflammation. SNPs in the OSMR gene have been found to associate with the severity of RA and systemic lupus erythematosus (SLE).⁸⁹ A SNP in the OSMR gene was shown to positively associate with tumor size in patients with papillary thyroid cancer.⁹⁰ Since SNPs in the OSMR gene have been shown to associate with disease, it would be worthwhile to seek SNPs in the OSM or OSMR gene that associate with type 2 inflammatory disease. If present, such SNPs could help to determine which patients are at risk and possibly provide insight about treatment modalities that patients might respond to.

It is paradoxical that OSM can play both pathogenic and protective roles as well as be involved with both type 1 and type 2 inflammation. Although speculative, one factor that may determine whether OSM induces pathogenic or protective responses may simply be the concentration of OSM. Perhaps small quantities of OSM are important for the protective, pro-repair effects of OSM and higher levels of OSM are pathogenic. Another potential explanation for the paradoxical effects of OSM may be tissue specificity. This idea has been outlined in the danger hypothesis, and postulates that the “choice” of whether the immune system should mount a type 1 or type 2 response is driven by the tissue where this immune response will be deployed.⁹¹ Type 1 immune responses have been shown to be very destructive in highly specialized tissues, such as the eye and brain.^92,93 These tissues may therefore skew the immune response toward a type 2 response in the event of an infection that would normally elicit a robust type 1 response to eradicate the pathogen and preserve tissue function. In support of this idea, fluid from the anterior chamber of the eye has been shown to be sufficient to both promote type 2 responses, and dampen type 1 responses.^94,95 Potentially, OSM may drive either a type 1 or type 2 response depending on which tissue is the site of inflammation. Another potential explanation for the paradoxical effects of OSM could relate to the signaling of OSM occurring through 2 distinct receptors, and activation by OSM of at least 5 different transcription factors. Activation of STAT3 may mediate the pro-repair effects of OSM, and OSM activation of other transcription factors may be responsible for OSM pathogenic effects. OSM roles in type 1 and type 2 inflammation may also be mediated through differing signaling mechanisms. Because OSM has so many varied effects throughout the body, therapeutically directly inhibiting OSM has the potential for having considerable side effects. Identifying the signaling mechanisms that mediate OSM induced barrier dysfunction could allow for a more targeted therapeutic approach while leaving the remaining OSM effects intact, thereby reducing the risk of side effects (Fig. 2).

Figure 2. Hypothetical mechanism of repair and potential therapeutic strategies for epithelial barrier restoration in the treatment of type 2 iflammatory disease. GM-CSF induces neutrophil derived OSM, which can mediate epithelial barrier dysfunction through the induction of EMT. Several potential mechanisms aimed at restoring epithelial barrier function for the treatment of type 2 inflammatory diseases are identified.

Neutrophil derived Oncostatin M in type 2 inflammation

Neutrophils are generally considered to be type 1 immune response effector cells and are not classically associated with type 2 immune responses. However, both CRS and asthma are known to have disease endotypes characterized by neutrophilia. While nasal polyps in Western countries tend to be highly eosinophilic, a significant proportion of nasal polyps from patients in Eastern Asian countries are neutrophilic.^96-98 Increased neutrophilia within nasal polyps has been linked to decreased responsiveness to corticosteroid treatment.⁹⁹ One endotype of severe asthma is often characterized by neutrophilia.^100,101 It is thought that neutrophilia in asthma associates with more severe disease, in part because neutrophils are not particularly responsive to corticosteroid treatment.¹⁰² However, little is known about the role of neutrophils in type 2 inflammatory disease, outside of the existence of neutrophilic subtypes of CRS and asthma.

Neutrophil-derived OSM has been implicated in the pathogenesis of many conditions including acute lung injury, asthma, breast cancer, and rheumatoid arthritis.^81,103-105 Osteopontin, prostaglandin E2, follistatin-like 1 (FSTL1), complement factor 5a, and thrombin have all been shown to induce OSM^106-110 but, only GM-CSF has been shown to induce OSM in neutrophils.^105,111 Elbjeirami et al. showed that OSM was induced during neutrophil transendothelial migration in response to endothelial production of GM-CSF.¹¹¹ Queen et al. showed that co-culture of neutrophils with breast cancer cells induced OSM production, a response that was lost when GM-CSF was blocked.¹⁰⁵ Additionally, Miller et al. have shown that FSTL1 was necessary for OSM induction in a mouse model of chronic asthma.¹⁰⁹ Both GM-CSF and FSTL1 have been shown to be elevated in allergic airways disease.^112-114 Our studies showed that GM-CSF induced OSM in neutrophils isolated from peripheral blood, while FSTL1 did not. Interestingly, we also showed that the OSM-producing neutrophils in nasal polyps also expressed GM-CSF, which may indicate that neutrophils induce their own production of OSM through autocrine production of GM-CSF.⁸⁵ There is precedent for such a concept, as Wardlaw and colleagues showed that eosinophils prolong their own survival through the autocrine production of GM-CSF.¹¹⁵ In addition, Durand et al, have shown that autoantibody ligation of FcγRIIIb (CD16) receptors on the surface of neutrophils protects against apoptosis through the induction of autocrine GM-CSF.¹¹⁶ Our group has previously shown elevated levels of autoantibodies in nasal polyps, suggesting that elevated autoantibodies in nasal polyps may trigger GM-CSF expression in the infiltrating neutrophils by a similar mechanism.¹¹⁷ Whether production of OSM by neutrophils via an autocrine GM-CSF dependent pathway occurs in type 2 inflammatory diseases is worthy of further investigation, and could potentially identify additional therapeutic targets (Fig. 2).

Neutrophils are classically thought to be a first line of defense in response to tissue damage,¹¹⁸ and neutrophil-derived OSM has been shown to be important in the first stages of epithelial repair during which basal cells proliferate to cover the wound.¹¹⁹ Neutrophils are known to be increased in CRS and severe asthma. Neutrophil-derived OSM could participate in both ongoing epithelial repair as well as promote the epithelial barrier dysfunction that is observed in these diseases. Several groups have shown that activated bronchial epithelium secretes GM-CSF into cell culture supernatants upon activation, e.g. by exposure to heat killed Staphylococcus aureus, which is a common pathogen that can cause epithelial damage.⁸⁵ One possible scenario is that epithelial damage induces GM-CSF, which in turn induces transient OSM production by the infiltrating neutrophils to promote the first stage of repair. If early OSM expression is transient, it would allow epithelial cells to redifferentiate back into functional epithelium during the later stages of repair. However, under pathological conditions, when neutrophils are more chronically recruited to the injury site, they may assume a phenotype that constitutively makes both OSM and GM-CSF.⁸⁵ The GM-CSF alone could be sufficient to induce chronic neutrophil-derived OSM in sufficient amounts to prevent late stage repair, causing a long-term state of barrier dysfunction. In addition, GM-CSF would also promote long-term survival of OSM-producing neutrophils. It will be important to understand the mechanism by which neutrophils “switch” from having a beneficial homeostatic role in epithelial repair to a pathogenic role in epithelial barrier dysfunction. Determining the molecular mechanism of this “switch” could uncover an important therapeutic target for the treatment of type 2 inflammatory disease.

Recent studies have described subtypes of polarized neutrophils, termed N1 and N2.¹²⁰ These N1 and N2 neutrophils have been shown to be very similar in function to their M1 and M2 macrophage counterparts.¹²⁰ N1 neutrophils are classical, proinflammatory effector cells, while N2 neutrophils specifically have been shown to promote tumor growth and metastasis, as well as play an important role in repair processes and the resolution of inflammation.^121-123 N1 and N2 neutrophil polarization has been best studied in mice, and N2 neutrophils have been shown to express the macrophage mannose receptor (MRC1), arginase 1 (ARG1), chitinase-like 3 (Ym1), IL-10 and TGFβ, which are also characteristic markers of M2 macrophages. Both M2 macrophages and N2 neutrophils have been associated with tissue repair mechanisms.¹²² Compared to their N2 counterparts, N1 neutrophils have been shown to express more proinflammatory cytokines and chemokines such as TNF, IL-1β, CCL3, CCL5, IL-6 and IL-12.¹²² We have shown that the majority of OSM⁺ neutrophils from nasal polyps expressed the N2 marker ARG1, indicating a phenotype resembling N2 neutrophils more than N1 neutrophils. We also showed that GM-CSF induced in vitro-stimulated neutrophils to express elevated MRC1 compared with unstimulated neutrophils and N1 polarized neutrophils, also indicating that OSM producing neutrophils could be skewed toward the N2 phenotype. However, since we did not see any induction of ARG1 in response to GM-CSF, another factor may be necessary for full induction of the N2 phenotype.⁸⁵ M2 macrophages have been shown to promote epithelial mesenchymal transition (EMT) in the context of cancer.¹²⁴ Given that M2 macrophages and N2 neutrophils have been shown to have similar functions, it is reasonable to speculate that N2 neutrophils could also induce EMT, potentially through the production of OSM, although this has not yet been studied. OSM has been shown to induce EMT,¹²⁵ which antagonizes epithelial differentiation, and expression of EMT biomarkers is well established in both CRS and asthma.¹²⁶

OSM has been shown to both promote and suppress cancer growth, and these effects should be considered in therapeutic targeting of OSM. OSM was originally discovered based on its ability to inhibit the proliferation of melanoma cell lines in vitro; however much of the data on OSM and cancer is conflicting and shows that OSM can both promote and prevent cancer cell growth. Some of this dichotomy is thought to reflect differential effects of OSM on tumor cells, where it may prevent growth, and on the immune cells within the tumor, where it may indirectly promote tumor growth. Whether OSM promotes or prevents cancer growth may depend on the relative abundance of immune cells within the tumor microenvironment. OSM has been shown to induce signatures of EMT in breast cancer cell lines, and the presence of OSM in breast carcinomas has been shown to be associated with increased evidence of EMT, decreased estrogen receptor expression, elevated risk of metastasis and poorer overall prognosis.^127-129 OSM has also been shown to associate with increased metastasis to the lung in a mouse model of cervical cancer.¹³⁰ Furthermore, OSM has been shown to promote the induction of M2 macrophages, and potentially could also promote N2 neutrophils through a similar mechanism.¹³¹ M2 macrophages and N2 neutrophils have been referred to as tumor-associated macrophages (TAMs) and tumor-associated neutrophils (TANs). Elevated TAMs have been linked with increased tumor growth, metastasis and poorer prognosis in many cancer types including breast, colon, gastric, lung, cervical and skin cancer.¹³² Elevated TANs have been linked with poor survival in head and neck, liver, kidney and pancreatic cancers.¹³³ These observations suggest that OSM could potentially promote cancer progression either directly or indirectly through the induction of M2 macrophages and possibly N2 neutrophils. Additionally, other aspects of type 2 inflammation have been shown to promote cancer progression, although no role for OSM in these responses was evaluated.^134-136

Given the association of cancer risk with type 2 inflammation, it seems plausible that type 2 inflammatory diseases may confer an elevated risk of cancer development, particularly in patients with severe or unmanaged disease. A few studies support a potential link between type 2 inflammation and cancer risk. A Swedish study found that asthmatic patients had a generally increased risk of cancer compared with the general population, and asthmatic patients diagnosed with cancer had a worse prognosis compared with non-asthmatic cancer patients.¹³⁷ A meta-analysis of American populations showed an increased risk of lung cancer in asthmatic patients that were never smokers.¹³⁸ Another study found a higher risk of breast cancer and lymphoma in atopic patients that had a positive skin prick test.¹³⁹ Taken together these data seem to suggest the possibility that cytokines associated with type 2 inflammation may promote cancer development and growth. Because of this potential association, proper management and/or prevention of type 2 inflammatory disease may be important not only to treat the patient's disease, but also to potentially decrease their lifetime risk for cancer development.

The role of Oncostatin M mediated epithelial- mesenchymal transition on epithelial barrier function in type 2 inflammatory disease

Epithelial-mesenchymal transition (EMT) describes the process whereby polarized epithelial cells transition to become non-polarized mesenchymal cells that are capable of proliferation and migration, and are less susceptible to apoptosis.^140,141 EMT has been shown to be involved in many cellular processes including embryogenesis, tissue repair, and cancer metastasis.^142,143,144 Growth factors TGFβ, VEGF, FGF, and members of the EGF family are all able to induce EMT.¹⁴⁵ The induction of EMT can be mediated by many signaling cascades, including STAT3, the Notch pathway, the wnt pathway, SMAD pathways, gp130-YAP-Src, and various receptor tyrosine kinase pathways.^146,147

The transition from polarized epithelial cells to mesenchymal cells requires the downregulation of genes that are important for maintenance of the epithelial phenotype, namely junctional proteins, and the upregulation of genes important for maintenance of the mesenchymal phenotype. Downregulation of the adherens junction protein, E-cadherin destabilizes the intercellular junctions, and is a hallmark of EMT.¹⁴⁸ During EMT, decreased E-cadherin is balanced by elevated levels of the mesenchymal cadherin, N-cadherin, which further destabilizes the intercellular junctions.^149,150 Additionally, genes that are required for maintenance of the mesenchymal phenotype are upregulated during EMT. Fibroblast specific protein (FSP1), also known as S100A4, is an important mediator of the mesenchymal phenotype through the alteration of cell motility, survival and cytoskeletal organization.^151,152 Actin polymerization is important for motility of mesenchymal cells, and α-smooth muscle actin (ACTA) is elevated in mesenchymal cells.^153,154 Elevated levels of the intermediate filament protein vimentin (VIM) are a canonical indicator of EMT because mesenchyme-associated vimentin intermediate filaments replace the keratin intermediate filaments that are present in epithelium.^155,156 Heat shock protein 47 (SERPINH1), is also elevated during EMT, and plays a role in collagen synthesis.¹⁵⁷ Discoidin domain receptor 2 (DDR2) is a collagen receptor that is capable of inducing EMT when activated by type I collagens.¹⁵⁸ Integrin α5 (ITGA5) and fibronectin have been shown to be important markers of the mesenchymal phenotype.¹⁴² Additionally, the extra domain A (EDA) splice variant has shown to be expressed in mesenchymal cells.¹⁵⁹ Expression of the genes necessary for the mesenchymal phenotype is mediated through transcription factors that drive EMT, including the Snail (SNAI) family proteins, twist-related protein (TWIST) family proteins, and zinc finger E-box binding homeobox (ZEB) family proteins.¹⁶⁰ As already mentioned, Oncostatin M (OSM) induces EMT, and we have implicated OSM in the barrier dysfunction that occurs in type 2 inflammatory disease.⁸⁴ In a cohort of CRS and control patients, we observed positive correlations of OSM with the mesenchymal markers, VIM, SERPINH1, ITGA5, and DDR2; with the EMT inducer TGFβ; and with the EMT transcription factors, TWIST1, TWIST2, ZEB1, and ZEB2 (unpublished). Additionally, when we treated bronchial epithelial ALI cultures with OSM we saw increased cell counts and elevated staining of the proliferation marker ki67 and the mesenchymal markers S100A4 and ACTA (unpublished). These data suggest that OSM is sufficient to induce EMT in airway epithelium, and that OSM induces epithelial barrier dysfunction through the induction of EMT.

The restoration of epithelial barrier function as a therapeutic strategy for the treatment of type 2 inflammatory disease

Our work has identified OSM as a potentially important inducer of barrier dysfunction in type 2 inflammatory diseases. Notably, OSM induced barrier disruption in nasal epithelial cells from CRS patients grown at ALI was reversible,⁸⁴ suggesting that therapeutic intervention targeting OSM has the potential to be beneficial in the treatment of CRS through the restoration of epithelial barrier function.

OSM has been therapeutically targeted in rheumatoid arthritis (RA) through the neutralization of OSM, and the inhibition of the JAK/STAT pathway.^161,162 JAK inhibitors CP-690, 550 and INCB028050, which inhibit JAK-1, JAK-2 and JAK-3, were shown to effectively block the activation of STAT1, STAT3 and STAT5 in OSM-treated fibroblasts in vitro.¹⁶² In contrast, a phase II trial in rheumatoid arthritis (RA) using an anti-OSM monoclonal antibody, GSK315234, did not show any benefit, potentially due to low affinity OSM binding. However, the drug was well tolerated in patients and the study investigators concluded that a high-affinity anti-OSM antibody might be beneficial in the treatment of RA.¹⁶¹ Another potential strategy for the neutralization of OSM is a receptor fusion protein (RFP) that consists of OSM ligand binding subunits of OSMR and gp130 covalently linked by a flexible domain, and such a construct has been shown to be a highly potent and specific inhibitor of OSM.¹⁶³

While the neutralization of OSM may result in either the prevention of barrier loss or the restoration of barrier function, it may not be optimal to target OSM to achieve these effects. As discussed earlier, OSM plays many diverse roles, including maintenance of liver homeostasis, bone metabolism, joint homeostasis and epithelial homeostasis. Broad systemic inhibition of OSM may thus interfere broadly with OSM functions and result in side effects associated with dysregulated OSM mediated homeostasis. OSM has also been shown to both inhibit growth of tumor cell lines in vitro, and promote tumor invasiveness and metastasis in vivo, as discussed earlier. Because OSM may play a role in the suppression of some types of cancer, further study is warranted to determine whether systemic inhibition of OSM could have the potential to elevate the risk of cancer development or progression. Given the considerable potential for unwanted side effects as a result broad systemic inhibition of OSM, such drugs should be approached carefully or alternatives should be explored. One potential alternative is local inhibition of OSM for the treatment of type 2 inflammatory disease, which could be achieved through medicated nasal lavages, nasal sprays, swallowed slurries or inhalers. Inhibition of OSM restricted locally in these diseases could potentially restore barrier function and allow chronic type 2 inflammation to resolve with a diminished risk of systemic side effects. As discussed earlier, the identification of the signaling pathway responsible for OSM induced barrier dysfunction could uncover alternative therapeutic targets that could help restore barrier function while leaving the remaining essential functions of OSM unchanged.

Another potential alternative to systemic inhibition of OSM is through inhibition of GM-CSF. GM-CSF is elevated in nasal polyps, expressed by neutrophils and sufficient to induce neutrophil derived OSM in vitro.⁸⁵ Given that GM-CSF may promote prolonged survival in neutrophils as well as induce OSM, therapeutic targeting of GM-CSF could potentially be beneficial in the treatment of type 2 inflammatory diseases through the restricted indirect inhibition of OSM. Inhibition of GM-CSF could block neutrophil-derived OSM production, allowing epithelium to enter the late phase of repair and differentiation, potentially resolving chronic inflammation resulting from barrier dysfunction. Two strategies have been used to inhibit GM-CSF in human trials, monoclonal antibodies against GM-CSF, and a monoclonal antibody against GM-CSFRα. Clinical trials have been conducted with monoclonal antibodies against GM-CSF (MOR103, KB003, and namilumab) in rheumatoid arthritis, plaque psoriasis, and severe asthma. A phase II trial using MOR103 in RA patients showed that patients given higher doses of the antibody had improved symptom scores compared with placebo control and low dose MOR103.¹⁶⁴ A phase II trial was conducted using KB003 in asthmatic patients that were poorly controlled with long-acting bronchodilators and either inhaled or oral corticosteroids. The investigators did not observe improved forced expiratory volume in 1 second (FEV₁) in the entire patient cohort. However, when patients were separated based on markers of poorly controlled asthma, peripheral blood eosinophilia, low baseline FEV₁, and high bronchodilator reversibility, improvement in FEV₁ was detected in response to treatment with KB003.¹⁶⁵ Clinical trials have also been conducted to test a monoclonal antibody against GM-CSFRα, mavrilimumab, for the treatment of RA, and showed reduced symptom scores in RA patients. The investigators concluded that further study of mavrilimumab for the treatment of RA was warranted.¹⁶⁶ Additionally, mavrilimumab, a monoclonal antibody against GM-CSFRα, has been shown to be safe, and a phase 2 trial of this drug showed no difference in the frequency of adverse effects between mavrilimumab and placebo treatment.¹⁶⁷ Identification of the factor or factors that are necessary to induce neutrophil derived GM-CSF could also provide another exciting therapeutic target, although it is important to note that any effect of GM-CSF inhibition could be independent of OSM (Fig. 2). The success of this approach would require not only that OSM is the major inducer of barrier dysfunction but also that GM-CSF is the major inducer of OSM; neither of these conclusions is firmly established. Another potential alternative to broad systemic inhibition of OSM could be through blocking the downstream effects of OSM by blocking the induction of EMT, which could include the inhibition of transcription factors like Snail, Zeb or Twist that drive EMT, or the inhibition of the EMT inducers, TGFβ, EGF, VEGF and FGF (Fig. 2).

Conclusion

Epithelial barrier dysfunction associated with EMT is an important feature of the pathogenesis of many type 2 inflammatory diseases, including CRS, asthma and EoE. We have identified a novel pathogenic role for the neutrophil derived cytokine, OSM, in the disruption of epithelial barrier function through the induction of EMT. Therapeutic targeting of OSM, its signaling, mediators of EMT, neutrophils, or GM-CSF may be beneficial in the treatment of these diseases through the restoration of epithelial barrier function, and further work is warranted to determine the validity of these potential therapeutic strategies.

Disclosure of potential conflicts of interest

Disclosures for RPS are consultancies with: Intersect ENT, GlaxoSmithKline, Allakos, Aurasense, Merck, BioMarck, Sanofi, Astra/Zeneca/Medimmune, Genentech, Exicure Inc, Otsuka Inc.

Funding

The work was supported in part by Grants R37HL068546 and U19AI106683 (Chronic Rhinosinusitis Integrative Studies Program (CRISP)) from the NIH, and by the Ernest S. Bazley Charitable Fund.