Airway remodeling, in addition to inflammation, abnormal neurogenic and contractile response, has recently been appreciated as one of the contributors to the symptoms, abnormal physiology and natural history of asthma and therefore is likely to generate manifestation of the disorder as phenotype. This novel concept led to a series of studies delineating cellular and molecular mechanisms of airway remodeling and clinical trials aiming to identify subpopulations of asthmatics with similar disease mechanism (endotype) [64, 65]. Airway remodeling is a collective term for structural alterations of airways encompassing the subepithelial fibrosis, myofibroblast hyperplasia and smooth muscle hypertrophy. The role of EMT in the development of subepthelial fibrosis, has been considered [28, 66] upon observation of elevated TGFβ1 production by eosinophils and fibroblasts in patients with severe and moderate asthma [67, 68]. Studies of allergen-induced airway remodeling in transgenic mice suggested an important role of TGFβ1, VEGF, Th2 cytokines (IL-5, IL-4, IL-13), and epithelial derived NF-κB regulated chemokines in airway remodeling, while eosinophils have received attention as cells contributing to thickening of reticular basal membrane . Mechanistically, eosinophils are the source of potent cytotoxic mediators, such as leukotrienes, metalloproteinases, and growth factors, including TGFβ1 . In patients with severe asthma, eosinophils constitute the majority of TGFβ1 producing cells as bronchial biopsies showed that 65% of TGFβ1 mRNA-positive cells are eosinophils and 75% of lung eosinophils were positive for TGFβ1 mRNA . Animal studies of airway inflammation using interleukin-5 and eotaxin-transgenic models indicate that goblet cell hyperplasia, epithelial hypertrophy, and focal collagen deposition are eosinophil-dependent [71, 72]. Complementing these observations, ΔdblGATA eosinophil-deficient and IL-5-deficient mice showed significant reduction in subepithelial collagen deposition and smooth muscle proliferation upon single challenge or chronic exposure to allergen [56, 73]. Furthermore, allergen challenge experiments in IL-5-deficient mouse showed decreased numbers of TGFβ1 -positive cells in the peribronchial region and reduced expression of TGFβ1 in the whole lung . The role of eosinophils in mediating EMT was directly shown in experiments where intratracheal instillation of bone marrow-derived eosinophils into mouse airways resulted in a marked deposition of type I collagen and significant fibrosis . These changes were accompanied by decreased ECad and increased α-SMA expression 21 d after eosinophil instillation. These findings complemented an earlier study employing anti-IL-5 antibody in patients with severe eosinophilic asthma refractory to corticosteroids that demonstrated that a reduction in eosinophils was associated with decreased basement membrane deposition of tenascin, lumican and procollagen type III .
However, there are several observations suggesting that airway remodeling in asthma may occur without significant involvement of eosinophils  or even in the absence of inflammation. For example, repeated methacholine-induced bronchoconstriction in mild atopic asthmatics resulted in airway remodeling comparable to that of induced by allergen challenge . This remodeling was accompanied by increased expression of TGFβ1 in the bronchial epithelium without airway eosinophilia suggesting heterogeneous upstream mechanisms mediate airway remodeling. Besides eosinophils, macrophages of alternative phenotype (known as the “M2 type”) were shown to participate in lung fibrotic processes in experimental models of nematode infection in IL-4R deficient mice [78, 79]. Macrophage polarization toward the M2 phenotype is characterized by increased expression and secretion of TGFβ1 and can be achieved by stimulation of macrophages with IL-4, IL-13  and glucocorticoids . Although the ability of M2 macrophages to produce TGFβ1 upon stimulation of cytokines overexpressed in asthmatic airways implied their involvement in fibrotic and remodeling processes in asthma, recent observations have, however, shown overlapping of M1/M2 phenotypes and conversion of one phenotype to another depending upon the inflammatory response . Similarly to macrophages, bronchial smooth muscle cells and dendritic cells were considered to play role airway remodeling after observation of enhanced secretion of TGFβ1 in response to neutrophil-derived elastase [83, 84]. Since neutrophils infiltrate airways in severe and chronic asthma, their ability to mediate production of TGFβ1 may affect airway remodeling seen in this group of patients .
The correlation of airway remodeling with features of EMT, expression of TGFβ1, severity of disease, resistance to glucocorticoid therapy and airway eosinophilia or neutrophilia can be used to define distinct molecular phenotypes of asthma and to some degree asthma endotype. The asthma “phenotype” is the result of a complex constellation of pathophysiological processes, whereas an “endotype” represents a subtype of disease defined functionally and pathologically by a singular molecular mechanism translating into a treatment response [86, 87]. Although no single endotype of asthma has yet been fully characterized, several potential endotypes were proposed based on clinical characteristics, biomarkers, histology and treatment response. In this regard, the phenotype of steroid-refractory asthma may include the potential endotype of steroid-insensitive eosinophilic asthma, a relatively rare form of asthma with airway eosinophilia, histological features of airway remodeling, glucocorticoid resistance and sensitivity to anti-IL-5 treatment . This endotype mechanistically resembles IL-5 transgenic or IL-5 deficient mouse models .
Future research exploring the role of EMT in airway remodeling may delineate that EMT may play a role common to several endotypes. For example, EMT may contribute to steroid insensitivity as suggested by decreased induction of anti-inflammatory genes by glucocorticoids in TGFβ1-EMT transformed cells A549 cells . In neutrophilic asthma, another subtype of severe steroid refractory asthma characterized by Th17-mediated neutrophilic inflammation and the absence of airway eosinophilia, airway remodeling may result from EMT induced by TGFβ1 released from elastase-stimulated bronchial smooth muscle cells and/or dendritic cells . Furthermore, even the endotype of Th2-high, early-onset asthma with high eosinophilia and good response to steroid treatment exhibits some features of airway remodeling  as shown by thickening of reticular basement membrane by median age of 29 months in preschoolers with recurrent wheezing . Although not yet defined, this endotype, driven by IL-4 and IL-13, may feature a role of TGFβ1 released from M2 polarized macrophages or from macrophages activated during phagocytosis of apoptotic eosinophils. Similarly, the severe, late-onset eosinophilic asthma endotype, representing 20% of the severe asthma population , also shows features of TGFβ1 mediated EMT and airway remodeling in spite of good response to systemic corticosteroid therapy.
In conclusion, future studies aiming at defining asthma endotypes should consider identification of mechanisms and respective biomarkers of TGFβ1 mediated EMT. Search for specific endotype biomarkers must focus on homogenous group of patients according to underlying disease mechanisms described by histological features (e.g. eosinophilia, M2 polarized macrophages), cytokine signatures (e.g. TGFβ1, IL4, IL-5) and responsiveness to anti-cytokine and glucocorticoid therapy. Systemic studies need to be directed at the cellular, molecular, and genetic factors that are responsible for determining why only some asthmatics develop significant remodeling. The information gathered from these studies will impact strategies employed to combat this increasingly complex and heterogeneous disorder.