|Home | About | Journals | Submit | Contact Us | Français|
In April 2010, a NIH workshop was convened to discuss the current state of understanding of lung cell plasticity, including the responses of epithelial cells to injury, with the objectives of summarizing what is known, what the field needs to know, and how to get there. The proximal stimulus for this workshop is the body of recent evidence suggesting that plasticity is a prominent but incompletely characterized property of lung epithelial cells, and that a focus on understanding this aspect of epithelial cell biology in particular, may be an important window into disease pathobiology and pathogenesis. In addition to their many vital functions in maintaining tissue homeostasis, epithelial cells have emerged as both a central target of disease initiation and an active contributor to disease progression, making a workshop to investigate the role of cell plasticity in lung injury and repair timely. The workshop was organized around four major themes: lung epithelial cell plasticity, signaling control of plasticity, fibroblast plasticity and crosstalk, and translation to human disease. Although this breakdown was recognized to be somewhat artificial, it was felt that this approach would promote cross-fertilization among groups that ordinarily do not communicate and lend itself to the generation of new approaches. The summary reports of individual group discussions below are followed by consensus priorities and recommendations of the workshop participants.
Respiration is dependent on a lung structure that is maintained by interactions among multiple cell types. This diversity of cell lineages adjusts to the mechanical and environmental challenges required for pulmonary function throughout life, providing a physical barrier to microbial pathogens, particles, and toxicants and enhancing host defense processes via mucociliary clearance and production of molecules mediating innate and acquired immunity. Lung structure is further maintained by the ability of pulmonary cells to proliferate, migrate, and differentiate to regenerate normal lung structures after injury. In spite of the elegance of this multitiered protective system, compromises in lung structure, whether initiated by inherited disorders or by interactions with the environment, are a major cause of morbidity and mortality associated with chronic lung disease. Common pulmonary disorders (e.g., pulmonary fibrosis, chronic obstructive pulmonary disease, asthma, cystic fibrosis, emphysema, bronchopulmonary dysplasia, and bronchiectasis) affect millions of people worldwide. In these disorders, marked changes in lung structure are accompanied by alterations in cell differentiation and function (e.g., terminal metaplasia, dysplasia, hyperplasia and perhaps epithelial–mesenchymal transition of the epithelium, and, in the case of fibroblasts, transition to a myofibroblast phenotype). These alterations are significantly influenced by cell–cell and cell–matrix interactions. Knowledge regarding the cellular and molecular processes maintaining normal lung cell homeostasis and mediating abnormal lung repair associated with chronic lung diseases will provide the framework for the development of strategies to prevent, diagnose, and treat pulmonary disorders that remain a significant health burden throughout the world. It was a testimony to the vitality and fluid nature of this field that there was no easy consensus among the participants on the definition of cell plasticity itself. Yet it was largely agreed that cell plasticity could be defined, at least for the purposes of the workshop, as “the ability of some cells to switch from one specific program of gene expression (phenotype) to another in response to specific signals from the environment in a regulated fashion. The changes may be reversible or irreversible and likely but not necessarily involve changes in cell shape, adhesion, and proliferation.” It was recognized that regulated phenotypic transitions may occur in a number of cell types besides epithelial cells, including endothelial cells and mesenchymal cells. However, so as not to dilute the focus of the workshop, discussion was largely limited to phenotypic transitions of epithelial cells and fibroblasts and cross-talk between these two cell types. In addition, one of the major limitations to advancing the field has been a lack of good models for pulmonary injury and fibrosis that involve epithelial injury independent of inflammation and that would more closely reflect human disease. Thus, discussion of alternate approaches to develop improved models of fibrosis was also a focus. Although it was recognized that many of the processes being discussed were relevant to other diseases, such as chronic obstructive pulmonary disease, to maintain the focus of the workshop, discussions of injury models were largely directed toward fibrosis.
The many specialized epithelial cells of the lung have long been defined by two inextricably linked properties: structure and function. This has led to definitions of lung epithelial cells that are based on their location and physical appearance, such as ciliated, goblet, serous, and basal cells of the conducting airways and type 1 and 2 epithelial cells of the alveoli. Classic microscopy studies using thymidine labeling to detect cell proliferation examined lung sections in the steady state or after injury (1–9). These studies defined subsets of airway and alveolar epithelial cells that exhibited the capacity to proliferate after injury. In addition, some labeled cells changed their shape, implying an alteration to the differentiated phenotype of the epithelial cells (10, 11). Developmental biologists established that multiple types of lung epithelial cells arose from the definitive endoderm, with lung epithelial progenitors being first detected in the primitive anterior-ventral foregut of the embryo and identified by expression of the transcription factor Nkx2.1 (Titf1) (12). The concept that a limited number of lung epithelial cell types were specified during embryonic development and more or less retained their committed differentiated state throughout life became conventional wisdom.
A number of advances in molecular and stem cell biology have begun to challenge this rigid paradigm, suggesting that much more diversity and fluidity may be present in various lung epithelial compartments (13–16). In instances where lung epithelia have appeared to derive from or turn into cellular phenotypes that were viewed as being a departure from the normal expected biology of that cell, the somewhat vague term “plasticity” was used to imply that a cell's program was pushed or pulled away from its natural homeostatic or developmental state by signaling mechanisms that are poorly understood. Because the diseased or injured lung frequently displays histology that is apparently disordered, metaplastic, or filled with mobilized cells, many investigators speculate that the genetic programs and lineage relationships present in the epithelial cells in these tissues may similarly deviate from normal. Whether these programs indicate cellular plasticity or the recapitulation of normal developmental programs is an active area of study that can best be advanced by the generation of new molecular tools, markers, lineage tags, and bioinformatics approaches.
Evolving technologies have also recently challenged rigid definitions of sets of epithelial cells. For example, the ability to track cells in vivo and over time after injury, with lineage tags or multiple marker genes, has suggested that no single marker can adequately define one type of lung epithelial cell. It is likely that significant heterogeneity and plasticity exist in each type of lung epithelial subset (17, 18). Furthermore, the advent of modern flow cytometery has allowed increasing fractionation of cells previously thought to contain only a single cell type, revealing heterogeneity in these cell subsets, based on gene expression, proliferative capacity, cell surface markers, or in vitro differentiation repertoire (19, 20).
Regarding lung epithelial cell biology, the recent discoveries in reprogramming have challenged investigators to acknowledge that extreme alterations in lineage, epigenetic memory, and differentiation repertoire are at least feasible in the laboratory dish in any cell type. The untested question becomes whether these alterations, such as reprogramming or cellular plasticity, ever happen naturally in vivo or whether some favorable resetting of cellular states can be engineered to treat lung epithelia damaged by injury or disease.
Plasticity rests upon the assumption that the cell has a definable “genetic program” (such as a transcriptome) in one state and a memory of that program. This memory of lineage and cell state is generally considered to be the epigenetic state of the cell, which is heritable, but it also can shift as the cell makes decisions regarding fate under the influence of a variety of stimuli.
Sophisticated imaging and the capacity to generate an increasingly diverse tool box of inducible genetic lineage tagging approaches now provides the seminal opportunity to again use the microscope to define the lineage relationships of parent and progeny during normal development as well as during disease and after injury. Induced genetic marking of certain lung cells or their niches will also allow study of micro- or macroenvironments and supporting niche cells that are increasingly described in other organ systems but virtually unstudied in lung biology.
Effective repair and regeneration of the lung after injury to the conducting airways or gas exchange apparatus requires precise control of progenitor cell fate. Thus, it is essential to identify the progenitor population(s) generating the cells and matrix comprising the effectively repaired lung and to understand how their proliferation and differentiation are regulated. Based on data imported from development, regeneration, and cancer, we review here knowledge about progenitor cell identity and control in the lung. We discuss the idea that mechanisms regulating a prototypical change in differentiated state—the epithelial–mesenchymal transition (EMT)—may reveal the multilevel linkage between signaling, gene expression, and phenotype.
Wnt signaling plays a critical role in specification and differentiation of early lung endoderm progenitors. The pathway appears to have a more profound effect on distal than on proximal lung development (21–23). These findings, along with data showing that Wnt signaling is important for tissue regeneration, suggest that this pathway may regulate lung epithelial regeneration and repair, especially in the distal or alveolar epithelium.
Wnt signaling may affect multiple cellular processes that are potentially antagonistic during tissue repair (e.g., progenitor expansion and fibrosis). Studies of this pathway need to include a more thorough understanding of how signaling networks integrate and control the response of adult lung epithelium to external injury and disease, especially with respect to epithelial–mesenchymal cross-talk. This should include studies exploring the integration of multiple signaling pathways, including FGF, BMP, Notch, and Shh, during injury and regeneration coupled to an expanded repertoire of injury and regeneration models.
Small airways of the human lung are sites of significant pathology. They are approximately 1.0 to 1.5 mm in internal diameter, the same diameter as mouse trachea and primary bronchi that are known to use p63+ basal cells as stem cells. Mouse airway basal cells that can be identified as p63+, Ngfr+, Krt5/14+, P-cadherin+, T1α+, and Egfr+ are multipotent stem cells that can self-renew and give rise to secretory and ciliated cells. Studies show that human basal cells are also multipotent stem cells that can generate surface epithelium and submucosal glands. However, we lack a rigorous comparison of human and mouse basal cells.
Although there is a paucity of studies on basal cells and their distribution in human lung disease, their capability as progenitors provides motivation to understand their potential role in hyperplasia, squamous metaplasia, and mucus metaplasia. We do not know whether aberrant basal cell function contributes to defective airway remodeling in human respiratory diseases. Defects, if present, may be intrinsic to basal cells or result from abnormalities in signaling from the niche, including stroma, endothelial cells, neighboring epithelial cells, and immune cells. Advances in this area will require injury–repair models in the mouse designed to study the role of basal cells in airway pathology together with ready access to well curated pathological samples of human airways disease.
Weinberg and Polyak have championed the concept that epithelial tumors are populated by a small subset of tumor cells with stem cell features (26). In murine breast models, CD44+/CD24−/lin− marks tumor-initiating cells that have mesenchymal features, the capacity to form mammospheres in vitro, and stem-like gene expression profiles enriched in Wnt-, Notch-, and EMT-related genes (24, 25). Furthermore, microRNAs involved in EMT regulation (miR-200, miR-335, miR-10b, and miR-9) are differentially expressed in breast tumors and correlate with poor prognosis (26, 27).
Although far less developed than the published work on breast cancer, studies on lung cancer report evidence of tumor cell populations with stem-like properties, including asymmetric cell division, polarized sphere formation in 3-D cultures, slow cell division and prolonged quiescence, prominent expression of drug efflux pumps, resistance to cytotoxic agents, and stem-like gene expression profiles (28–31). The cell surface markers that distinguish these cells from the larger tumor cell population have not been fully studied, but by far the most commonly cited marker of these cells is CD133 (28).
Lessons learned from lung tumor cells should be applied to normal lung progenitor cells with caution because, unlike normal lung progenitors, tumor cells have aberrant chromosomes, somatic mutations, abnormal cellular architecture, and altered expression of surface receptors. However, there are important parallels between lung cancer cells and nonmalignant progenitors residing in a diseased lung. Both share an “activated” microenvironment that is often hypoxic, fibrotic, inflammatory, angiogenic, and enriched in cytokines and growth factors.
Evidence of epithelial plasticity appears after many airway and parenchymal lung injuries, ranging from metaplasia and transdifferentiation in the major airways with viral infections to hyperplasia and potentially EMT in the lung parenchyma during fibrogenesis (13, 32, 33). It is not known whether only a small subset of preexisting, relatively undifferentiated cells accounts for regeneration and EMT after injury or whether most or all differentiated alveolar epithelial cells dedifferentiate, proliferate, or are reprogrammed toward different cell lineages. These are critical gaps in our understanding of lung repair.
The epithelial potential for proliferation and drastic phenotypic change after injury has elicited considerable interest in the underlying signaling mechanisms that drive epithelial responses to injury. Lung injury superimposes on regulated, epithelial cell TGF-β1, signaling additional insults that may alter the outcome. Persistent inflammation results not only in persistence of fibrin- and fibronectin-rich matrices in the lung that accentuate repair but also potentially the persistence of proinflammatory proteases, chemokines, and cytokines (e.g., IL-6 and IL-13) that foster ongoing TGF-β1 activation and fibrosis (34, 35). A second complicating insult is local hypoxia, in part from alveolar edema or collapse (36). Hypoxia results in accumulation of hypoxia inducible factor (HIF)-1α and epithelial reprogramming that resembles that of TGF-β1 signaling (37). Collectively, these insults challenge the self-regulating repair mechanisms of the distal lung and perpetuate a profibrotic state. The specific roles of alveolar collapse and edema and the resulting hypoxia and HIF signaling in lung fibrogenesis are not known.
Bidirectional interactions between epithelial cells and fibroblasts and interactions with matrix may strongly influence outcomes after lung injury. In addition, fibroblasts may exhibit phenotypic plasticity and exhibit heterogeneous phenotypes and fates after injury. The goals of this session of the workshop were to examine what is known regarding the effects of cell–cell and cell–matrix interactions on cell fate and to explore opportunities for influencing outcome after lung injury by modulating these interactions. Additional goals were to evaluate what is known regarding fibroblast fate and to explore opportunities for modulating cell fate to promote repair after injury.
Ultrastructural studies of the adult mammalian lung indicate that interstitial mesenchymal cells (which we refer to as “fibroblasts”) form direct contacts with alveolar epithelial cell type 2 (AEC2) and capillary endothelial cells on both sides of the air–blood barrier (38). Numerous small apertures in the basement laminae that support AEC1 and AEC2 allow for cytoplasmic extensions and interdigitations of fibroblasts with adjacent epithelial cells (39, 40). Interstitial fibroblasts appear to form an interconnected reticular network that extends from the distal alveoli to the more proximal conducting airways (41).
Fibroblasts are structurally well positioned to “sense” and respond to structural and functional changes in airway and alveolar epithelial cells and in vascular endothelial cells. Additionally, fibroblasts participate directly in immune cell trafficking to the airways; neutrophils and eosinophils migrate in contact with the fibroblasts that “guide” them through apertures in epithelial and endothelial basal laminae (39, 42, 43). In chronic obstructive pulmonary disease (COPD), a significant decline in the numbers of epithelial cell–fibroblast contacts with reduced numbers of basal lamina apertures are observed in emphysematous regions of lung parenchyma (40) and in conducting airways (44); this is associated with loss of airway caliber and thickening of the more proximal conducting airways.
In addition to direct cell–cell contact (“juxtacrine” signaling), communication between lung cells may occur through paracrine and autocrine signaling. The homeostatic mechanisms that maintain the differentiated state of resident lung cells are not well understood and are likely to require regulatory input from neighboring cells (45). In smokers and in patients with severe, chronic asthma, epithelial changes, such as squamous metaplasia and goblet cell hyperplasia, accompany the fibrotic remodeling response (46, 47). Such phenotypic changes may be perpetuated by bidirectional paracrine signaling between epithelium and mesenchyme (48–50). Fibroblasts may also regulate the phenotype and fate of cells by elaboration and remodeling of the extracellular matrix (ECM). The ECM provides mechanical support to the lung and critical contextual information for maintenance of epithelial cell phenotype in healthy lungs. Modulation of ECM facilitates tissue regeneration in response to injury, but sustained ECM alterations may lead to aberrant epithelial cell phenotype and function. Changes in ECM composition and structure affect cellular phenotype as a direct consequence of changes in signaling pathways controlled by integrins and other cell–ECM signaling molecules but also by virtue of the ability of the ECM to act as a reservoir and modulator of soluble signaling molecules.
There is a high degree of tissue-specific variability in epithelial–fibroblast interactions because epithelial differentiation is affected differently by mesenchyme derived from the trachea versus mesenchyme at distal structures, and mesenchyme derived from these different sites has differential sensitivity to soluble molecules produced by epithelial cells (51, 52). However, although much has been learned about the different types and functions of epithelial cells in the lung, the corresponding studies defining functional differences among lung fibroblasts are much less complete, limiting our ability to define functional interactions between epithelial cells and fibroblasts. Microarray studies have identified distinct gene expression profiles for lung fibroblasts as compared with other fibroblasts (53, 54) but not for distinct regions within the lung; much less is known about local changes in ECM composition and structure in the initiation and progression of lung diseases.
Fibroblasts are inherently versatile cells with remarkable plasticity. Although fibroblasts are of mesenchymal origin, no specific markers have been identified, and significant heterogeneity exists between different subsets within a population. Although reports of their ability to differentiate into epithelial and endothelial cells in vivo are limited, fibroblast differentiation into mesenchymal lineages (e.g., adipocytes, chondrocytes, and osteocytes) is well recognized. Fibroblast plasticity is well suited for their central role in tissue responses to injury; myofibroblast differentiation is a hallmark of wound healing responses in diverse organ systems, including the lung (55, 56). The origins of myofibroblasts and their relative contributions to lung repair and fibrosis remain unclear, although they may derive from local tissue mesenchymal progenitor cells (57), circulating fibrocyte-like cells (58–60), and epithelial–mesenchymal transition (16, 61, 62).
Pulmonary fibrosis complicates a heterogeneous group of disorders that are driven by chronic inflammation or repeated epithelial injury and involve regions of the lung varying from mainly airways to almost exclusively the lung parenchyma. More broadly, fibrosis is a final common pathway in many forms of chronic disease, affecting a range of tissues and leading to organ scarring and failure. Although fibrosis in various forms affects millions of patients in the United States, there is a paucity of adequate treatment options. Thus, there is a need to translate an understanding of basic disease mechanisms to the more pragmatic challenges associated with the clinical development of novel therapeutics. The most therapeutically challenging of these diseases in the lung is idiopathic pulmonary fibrosis (IPF). Although there have been many advances in clarifying the diagnosis and prognosis of IPF over the last decades, no medical therapy has significantly improved survival or halted the relentless progression of fibrogenesis (63, 64). The failure of antiinflammatory agents to favorably affect the course of IPF, coupled with the elucidation of defects that primarily involve AECs being associated with familial IPF, has turned attention to the pathobiology of epithelial–mesenchymal interactions in this disease. This has led to a number of clinical trials involving agents that, on the basis of preclinical studies in mouse models, were thought to focus on signaling pathways directly or indirectly involved in fibrogenesis. These include pirfenidone, etanercept, interferons, and, most recently, imatinib mesylate and bosentan. Unfortunately, none of these agents has demonstrated an unequivocal clinical benefit.
What lessons can one derive from the recent therapeutic failures? And what is needed in the field to inform and promote new clinical trials that better address critical elements of the pathobiology? Members of the workshop discussed key obstacles in translating basic studies to the clinical arena and attempted to identify several short- and more long-term advances that would increase the likelihood of developing promising therapies. Although much of the discussion was directed toward IPF, many of the principles elucidated are relevant to other chronic lung diseases. The group identified four major areas for high-impact discovery.
One lesson from recent clinical trials is that a molecular understanding of what drives the relentless fibrosis in IPF is unclear. A more complete understanding of the signaling pathways underlying fibrogenesis should result in new therapeutic targets more specifically directed toward fibrosis. In addition, because completely inhibiting a specific pathway may have adverse unintended consequences in the same or another cell type, more detailed knowledge of specific downstream targets and molecular interactions will be important.
A careful consideration of the model(s) in which preclinical data are obtained is important. Despite its limitations, the bleomycin-induced lung injury model remains the most widely used and best characterized model for studying potential antifibrotic agents. An important limitation of the bleomycin model is that fibrosis is highly dependent on inflammation induced by epithelial injury, whereas IPF has not proven responsive to antiinflammatory therapeutics. Inhibition of fibrosis in the bleomycin model by drugs given early after injury does not predict clinical responses. Although the bleomycin lung injury model has severe limitations, it has provided valuable insights into potential pathways involved in the pathogenesis of fibrosis, including molecular markers that reflect disease mechanisms impairing epithelial cell health and promoting fibrogenesis and that can reliably serve as interim indicators of a therapeutic response.
A major barrier to the development of therapeutics for fibrosis is the lack of interim indicators that reliably reflect drug effects on mechanisms of disease. Generating basic research tools and model systems to evaluate the relevance of target pathways to human disease and developing biomarkers to monitor the effectiveness of therapeutic intervention in clinical trials will greatly facilitate the development of novel therapies. Toward this end, understanding how molecular mechanisms regulating fibrotic disease in animal models overlap with human disease can lead to the identification of new biomarkers.
Personalized medicine refers to therapy based on the elucidation of family history, environmental exposures, distinct clinical features and outcomes, biochemical phenotypes, and pathological characteristics that together identify dominant signaling pathways affecting the progression and outcome of disease in an individual patient. Chronic illness represents over 75% of medical care in the United States. These diseases are complex disorders that result from a combination of genetic factors and environmental exposures. We tend to label them as distinct diseases; however, the diseases are more appropriately labeled syndromes to indicate involvement of a variety of molecular pathways and clinical subphenotypes within each category. Moreover, the diseases themselves often have overlapping genetic and environmental components with common pathologic manifestations. The goal of personalized medicine is to match specific aberrant molecular pathways with clinical subphenotypes and develop specific therapy for each individual patient.
The overlap of COPD and IPF represents the blurring of disease categories based upon common environmental exposures and pathologic phenotypes. At first glance, COPD and IPF appear to be diametrically opposed. Emphysema is a result of inflammation induced by cigarette smoke and proteolytic destruction of elastin and other ECM components, whereas IPF is not necessarily inflammation dependent and is characterized by relentless extracellular matrix deposition. However, around two thirds of patients with IPF are cigarette smokers. Moreover, epithelial cell apoptosis, matrix metalloproteinase activation, and net collagen accumulation are characteristic of COPD and IPF.
Even though the prognosis of IPF is generally poor, there is abundant evidence that there is substantial heterogeneity in the outcomes even within a biopsy-proven patient population. Cell plasticity may play a role with phenotypic switching, which can account for clinically distinct manifestations. Certain putative biomarkers of disease activity are elevated in some but not other patients with IPF. To effectively bring new therapeutic targets to the clinical arena, it will be crucial to identify at the outset the set of patients most likely to respond to a particular treatment. It may also be the case that the activity of one or only a few signaling pathways dominates disease progression in all patients, but there is little evidence to support this possibility.
This workshop was supported by the NHLBI and the NIH.
Author Disclosure: Z.B. received compensation for travel to the workshop. J.A.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. P.B.B. was a consultant for Sanofi-Aventis and owns a patent through the University of Minnesota. V.J.T., D.N.K., and S.D.R. do not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.A.K. is on the Board and owns stocks or options of Fate Therapeutics. D.W.B., E.E.M., B.L.H., J.M.K., D.C.W., D.C.R., and S.L.N. do not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.M.V. is employed by and owns stocks or options of Stromedix. P.W.N. was a consultant for Intermune, Boehringer-Ingelheim, Actelion, Novartis, Johnson & Johnson, Stromedix, Gilead Sciences, and Human Genome Sciences. S.D.S. was on the Advisory Board for Boehringer-Ingelheim and GlaxoSmithKline. C.J.B. is an NIH government employee. H.A.C. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.