Induced pluripotent stem cells for treating cystic fibrosis: State of the science

Induced pluripotent stem cells (iPSCs) are a recently developed technology in which fully differentiated cells such as fibroblasts from individual CF patients can be repaired with [wildtype] CFTR, and reprogrammed to differentiate into fully differentiated cells characteristic of the proximal and distal airways. Here, we review properties of different epithelial cells in the airway, and the in vitro genetic roadmap which iPSCs follow as they are step‐wise differentiated into either basal stem cells, for the proximal airway, or into Type II Alveolar cells for the distal airways. The central theme is that iPSC‐derived basal stem cells, are penultimately dependent on NOTCH signaling for differentiation into club cells, goblet cells, ciliated cells, and neuroendocrine cells. Furthermore, given the proper matrix, these cellular progenies are also able to self‐assemble into a fully functional pseudostratified squamous proximal airway epithelium. By contrast, club cells are reserve stem cells which are able to either differentiate into goblet or ciliated cells, but also to de‐differentiate into basal stem cells. Variant club cells, located at the transition between airway and alveoli, may also be responsible for differentiation into Type II Alveolar cells, which then differentiate into Type I Alveolar cells for gas exchange in the distal airway. Using gene editing, the mutant CFTR gene in iPSCs from CF patients can be repaired, and fully functional epithelial cells can thus be generated through directed differentiation. However, there is a limitation in that the lung has other CFTR‐dependent cells besides epithelial cells. Another limitation is that there are CFTR‐dependent cells in other organs which would continue to contribute to CF disease. Furthermore, there are also bystander or modifier genes which affect disease outcome, not only in the lung, but specifically in other CF‐affected organs. Finally, we discuss future personalized applications of the iPSC technology, many of which have already survived the “proof‐of‐principle” test. These include (i) patient‐derived iPSCs used as a “lung‐on‐a‐chip” tool for personalized drug discovery; (ii) replacement of mutant lung cells by wildtype lung cells in the living lung; and (iii) development of bio‐artificial lungs. It is hoped that this review will give the reader a roadmap through the most complicated of the obstacles, and foster a guardedly optimistic view of how some of the remaining obstacles might one day be overcome.

Subsequently iPSCs can be coaxed to differentiate into pulmonary, cardiac, neural, and other lineages, and finally into fully differentiated cells. 1 Recent studies show that ESCs and iPSCs are genetically identical when developed from the same genetically matched donor cells. 2 Genomic instability in human iPSCs can occur due to replication stress, 3 which can be controlled, for example, by nucleoside supplementation and increases in checkpoint kinase 1 (CHK1). 4 Nonetheless, among the most attractive properties of iPSCs is that they can proliferate rapidly, and can apparently propagate indefinitely. Thus IPSCs can provide large amounts of cells for personalized regeneration of a diseased tissue. This technology thus seems tailor-made for repairing genetic diseases such as cystic fibrosis (CF) in which mutations in the CFTR gene are responsible for a lifelimiting clinical phenotype in the CF lung. [5][6][7] Indeed, the timely matching of iPSC and genome editing technologies has recently been described as "forecasting experiments for the next decade." 8 For example, if the [508del]CFTR mutation from a CF patient's iPSCs could be replaced with a [wildtype]CFTR sequence, perhaps using CRISPR/ Cas9 or other editing procedures, 9,10 then the "corrected" iPSCs might be reprogrammed to replace epithelial cells in the diseased lung. This is the ultimate therapeutic promise for iPSC technology.
In vitro, the positive news is that gene editing has been successfully deployed to correct the [508del]CFTR to [wildtype]CFTR in iPSCs derived from a CF patient's fibroblasts. 11 Furthermore, it has been determined that [wildtype]CFTR function and protein can be detected in these now normal cells when they have been differentiated at the air-liquid interface to form proximal airway epithelial cells. 11 As proof of normal function, Firth et al were able to detect cAMPactivated chloride conductance by patch clamp, and mature CFTR protein "C" band by Western blot. CFTR is a cAMP-activated chloride channel, which, when properly glycosylated, trafficks to the apical plasma membrane of epithelial cells as a higher molecular weight "C" band. Importantly, equivalent experiments have been done by other investigators as well, not only for lung, 12,13 but also for CF pancreatic duct 14 and CF cholangiocytes. 15,16 Thus for treating CF lung disease with iPSCs, the vision for this workflow would be as follows: (i) to obtain fibroblasts from a CF patient; (ii) to induce iPSCs from the fibroblasts; (iii) to correct the CFTR mutation, either with CRISPR/Cas9 or zinc-finger nucleases (ZFNs); (iv) to grow up unlimited numbers of corrected iPSCs; and (v) to induce iPSCs to differentiate; and (vi) to deploy them back into the CF patient who donated the original fibroblast as personalized medicine. However, as we shall see below, the deployment part of this algorithm remains a significant challenge.
By focusing attention on the CF airway, a second profound challenge is understanding exactly what histological structures are being reconstituted, by which cells, and in which order. Which are the cells to which the iPSCs should be induced to develop, and how pure must the iPSCs and their induced progeny be? To develop iPSCs into therapies, should we be trying to use fetal development of the lung as a roadmap for repair and reconstitution, or is there another way? A final question is how to "administer" the iPSC-based therapy. Suggestions have ranged from whether to introduce cells as individuals or as pregrown sheets of mature endothelium through the airway; or to re-seed bronchi and bronchioles that have been stripped of mutant epithelium; or to inject the spheroids of mature lung cells of cadaveric origin intravenously, and let them spontaneously find their way to the lung. 17 However, there are clinical, structural, and physiologic barriers which must be overcome to move from these in vitro proof-ofprinciple experiments to deploying the technology therapeutically in CF patients. Many outstanding articles and comprehensive reviews have been published in the last few years that specifically address the lung. [18][19][20][21][22] Here, we will assess important challenges that the lung, and the CF disease itself, erects against different therapeutic strategies. In addition, there is much more to cystic fibrosis than the effects on the lung. We will discuss this aspect of CF below. Yet it is in the lung where the threat-to-life is most manifest, and where clinical work-arounds have been most challenging. We, therefore, focus primary attention in this review on replacing CF epithelial cells lining the CF airway, and on the challenges of the present state of the science. On a positive note, it has been calculated that by delivering just 25% of wildtype CFTR to the CF epithelial surface one might observe normal rates of mucus transport on the CF epithelium. 23 This calculation would thus seem to set a lower limit to the kind of success that added iPSCs might bring to a The physiology of the CF lung seems to be focused on failure of the cAMP-activated CFTR chloride channel to secrete chloride into the airway, with consequent dehydration of the airway. 24 If the channel had been active, first sodium and then water would have followed the chloride, and hydration of the airway would have been assured. The consequences of failure are catastrophic, especially when concatenated with the concomitant failure of the airway epithelial cells to discriminate between septic bacterial and viral infection, and aseptic damage such as dehydration, free neutrophil elastase, or an acidic environment. Thus in response to a damage stimulus, the CF goblet cells secrete mucus into a dehydrated airway. However, secreting mucus into a dehydrated environment limits mucin distribution, and results in thick, sticky mucus. Physiologically, this makes sense to the body because the function of the mucus is to protect airway cells by binding to bacteria and viruses. Therefore, on the explicit assumption that microbial invaders have arrived, and that there is a need to deflect them from their intended epithelial cell targets, mucus will be available to bind them. The ciliary cells will then ferry the mucin-bound microbes up and out of the lung. But because the CF POLLARD AND POLLARD | S13 airways are poorly hydrated, the mucus, along with the invaders stuck to it, becomes immobilized within the airway. 25 This leads to failure of mucociliary clearance, especially in the small airways that are denied air due to being blocked off by mucin plugs. 26 These are now sites of potential infection. Impaired PGE2-stimulated chloride and bicarbonate secretion in submucosal glands also contribute to cystic fibrosis airway disease. 27 The importance of bicarbonate is further manifest by a recent report of data from the CF rat, in which correction of low pH and bicarbonate transport is a specific target for restoring mucus clearance. 28 Further regarding oxygen detection in mucin-blocked airways, neurosecretory cells at bronchial bifurcations signal to the brain through their vagal nerve innervation that CO 2 levels are high and O 2 levels are low. Vagal innervation which contributes to the cough reflex in humans, 29 may be very important to the CF patient for helping clear blocked airways. 30 In the hypoxic mutant CFTR condition, it has been shown that hypoxia-induced 5HT (serotonin) release from neuroendocrine cells is likely suppressed, as well as the secretory response to high potassium. 31 In conclusion, even if immune cells do respond by entering into mucin-blocked, low oxygen regions, they do not function well in an environment that is additionally low in chloride because of mutant CFTR, and low in pH because of low bicarbonate.
Bacteria and viruses can then grow in these mucin plug blocked places where infection is further enhanced by denial of access to blood-born antibiotics. An important target for iPSC-based CF therapies would therefore be either suppression of mucin production or increased hydration, or both, in the CF submucosal glands and airways.

| Inflammation
CF airway epithelial cells chronically secrete high levels of chemokines such as IL-8 and cytokines such as IL-6, IL-1β, and TNFα. [32][33][34] The tendency to secrete high levels of IL-8 occurs under both baseline and activated conditions. The causative mechanism in CF is based on an intrinsically constitutive activation of NFκB signaling. [32][33][34][35][36] Subsequent experiments with the potent anti-inflammatory drug digitoxin have shown that digitoxin blocks NFκB activation in CF lung epithelial cells by inhibiting the interaction between TNFR1 and its first intracellular adaptor TRADD (TNFα-associated Death Domain). 35,36 TRADD is an obligatory mediator of TNFα/NFκB signaling. However, only recently has the biological mechanism of NFκB-driven chronic activation in CF lung been identified. 37  The result is that TRADD is now free to drive constitutive activation of NFκB, and also drive expression of NFκB-driven downstream proinflammatory cytokines and chemokines such as IL-8, IL-6, IL-1β, and many others. Consistently, a recent placebo-controlled, dose escalation trial of digitoxin in CF patients found that proiflammatory signaling was suppressed in biopsied nasal epithelia. In addition, neutrophil elastase in the sputum was significantly suppressed. This study had been initially planned by one of this review's authors, and executed at Johns Hopkins University (NCT00782288, clinicaltrials. gov). 38 Concurrently, new potentiator drugs such as VX-770 and correctors such as VX-809 and VX-661 were being developed for administration individually or in combinations. It has been reported that they can suppress proinflammatory signaling in some CF patients. [39][40][41][42] Not unexpectedly, rescue of CF cells with [wildtype] CFTR can suppress these constitutive chemokine and cytokine elevations. 33 Thus, suppression of the hyper-proinflammatory CF phenotype in the lung will be a necessary target for lung-centric iPSC-based therapies.

| Immune cell dysfunction
A third lung-centric CF problem is that CFTR mutations affect immune cell functioning. Examples include neutrophils, 43 macrophages, 44 and lymphocytes 45 in the immune system. Importantly, failure of neutrophil function is intrinsic to CF damage in the lung. For example, there is an abundance of IL-8 in the CF airway, which successfully attracts circulating CF neutrophils into the lung. The neutrophil's purpose there would be to consume bacteria and other microorganisms. However, the CF neutrophils are physiologically defective. Neutrophils normally use the chloride conducting properties of CFTR to bring ambient chloride into the phagolysosome where myeloperoxidase converts chloride to bacteriocidal HOCl ("chlorox"). However, ambient chloride concentration is low in the CF airway, and the chloride conducing properties of mutant CFTR are reduced. Therefore HOCl production is limited and bacterial killing becomes inefficient. 43,46 Another problem is that these crippled CF neutrophils chronically release ("dribble") neutrophil elastase (NE) into the airway, thus contributing to bronchial damage and bronchiectasis, reactive and sustained mucin secretion, and finally lung failure. However, it is possible that an iPSC-based therapy might reduce hypersecretion of IL-8 from the CF lung, thereby reducing the attractive signals for circulating CF neutrophils. It is already known that by reducing the absolute concentrations of circulating neutrophils by approximately 50% with high doseibuprofen, the chronic influx of CF neutrophils can be controlled, with positive therapeutic consequences for lung function in those CF patients able to sustain this therapy. 47

| Chronic infection
In the CF lung, there is an age-dependent progression of different types of chronic bacterial infections, starting with Staphylococcus aureus, and Haemophilus influenza, and progressing to Pseudomonas aeruginosa (controlled by intermittent tobramycin ® ) and finally Berkholderia cenocepacia complex. [48][49][50][51] The Cystic Fibrosis Foundation has also compiled detailed records on infection incidence in the entire US population. This problem has not necessarily gone away with the advent of correctors and potentiating drugs: while there is on average a prompt reduction in infection, in some patients there is a later rebound. 42 Thus an important target for iPSC therapy would be a profound change of the lung micro-environment such that it no longer preferentially accommodates the survival of these bacteria in the CF lung. Prospectively, perhaps iPSCs could also be used to make mesenchymal CD34+ stem cells, which would be used to permanently repair the mutant CFTR gene in the CF immune cell population.
However, one can envision the possibility that if the chemoattractive production of IL-8 could be suppressed by iPSC-based repair of mutant CFTR, massive invasion and damage to the CF lung by crippled neutrophils and macrophages might be minimized.

| The bystander/modifier gene problem
It has been increasingly realized that there is a great variation in disease severity for CF patients even with the same mutation. Environmental effects from smoking or second hand smoke became immediately apparent. 52 For the CF patient, the combustion products of smoke in the lung actually cause a significant reduction in CFTR gene expression. 53 Thus whatever residual function might still be available to the CF patient is diabolically suppressed by smoke exposure. More recently it has become apparent that a higher average temperature, found closer to the equator, enhances CF disease severity. 54 The mechanism is not known. However, this observation is reminiscent of the observation by Paul Di Sant'Agnese that in New York City during the hot and humid summers of 1948-1949, CF babies suffered from heat prostration, and sweated more chloride than normal babies. 55 The result was the classic sweat test for CF. On the other hand, increased chloride in inspired ocean air, as originally noted by a surfer with CF in Australia, reduces symptoms, leading to now-routine administration of inhaled sodium chloride. 56 However, there are also problems in CF which are not defined by the CFTR gene itself. Instead they are defined by non-CFTR bystander or modifier genes, that directly affect not only disease severity, 57 but in select cases also the capacity to respond to CF drugs. 41 We have estimated from a comprehensive survey of the literature that there may be as many as 56 such candidate genes. Variants associated with most of these genes significantly affect disease severity with respect to lung disease. A few examples of such lung-centric genes, many associated with inflammation, 58  and meconium ileus 69 ). Thus if iPSC-dependent therapies were directed exclusively to the lung, they might render effects of lungcentric bystander/modifier genes moot. But other non-lung CF organ pathologies would remain intact. Thus these non-lung CF organs would still be susceptible to enhanced CF disease severity by the non-lung-specific bystander/modifier genes.

| Other CF affected organs in the body
However, as noted above, CF is a type of chronic injury that affects functions not only in the lung, but also elsewhere in the body. Although the principal focus of this review is directed to iPSCs for repair of proximal airway epithelium in the CF lung, it should be appreciated that the disease associated with CFTR mutations affects more than epithelia in the lung.
For example, mutations in CFTR also affect smooth muscles in the lung which control the diameters of small airways. In the CF pig model, CFTR is associated with the sarcoplasmic reticulum (SR) and functions to increase the efficiency of calcium uptake. Consistently, in the cystic fibrosis SR, calcium uptake is reduced. 70 Consequently, elevated calcium in the cytosol leads to enhanced contractility of the smooth muscle. By contrast, Ivacaftor ® /VX-770 administration to CF patients with the [G551D] mutation enhances airway distensibility and reduced vascular tone both in small airways and in tissues outside the lung. 71 In the gastrointestinal tract, cells affected directly by CFTR mutations include cholangiocytes (the cells lining the biliary ducts), pancreatic ductal cells, and epithelia in the small and large intestine.
This recalls the historical situation that before pancreatic enzymes and bile salts became available for CF patients, CF was first and foremost a failure-to-thrive disease based on malnutrition. In fact, as iterated in the introduction, the use of iPSCs to reconstitute [wildtype]CFTR in CF cholangiocytes has been reported. 15,16 Similarly, iPSCs have also been developed to reconstitute [wildtype]CFTR expression in pancreatic ductal cells. 14 In addition, inflammation in the pancreas might still have consequences for lung disease, independent of mutant CFTR.
Although the mechanism is unknown, there is a well known clinical phenomenon connecting inflammation in the pancreas with pulmonary inflammation. 72 Thus pathology in these separate organs are linked in some manner, possibly by the shared vagus nerve.
Other systems sensitive to CFTR function include the male urogenital system, where congenital hypoplasia or aplasia of the vas deferens and seminal vesicles may occur either bilaterally or unilaterally. 73 Consequently, most CF males are phenotypically sterile.
Finally, in skin, the dysfunction of the sweat glands is life-threatening in hot, humid weather, and is patho-mnemonic for CF. 74,75 Mechanistically, mutant CFTR prevents the sweat glands from reabsorbing chloride. However, while these non-lung centric CF cell disorders remain unresolved, most would agree that the central threat to life for the CF patient lies squarely in the cells lining the proximal airway. these cells comes from the mouse, where possible most investigators have been careful to test for parallels between mouse and humans.
Remarkably and fortunately, the parallels have proved to be quite substantial.

| Proximal airway development, in vivo and in vitro
During fetal development in mouse and human, the foregut endoderm begins to form lung buds, which further undergo stereotypical branching morphogenesis. 19 The initial epithelium is composed of proximal progenitors, including basal cells, which closely adhere to the basal lamina. Basal cells have been presumed to be the stem cells for the rest of the pulmonary epithelium, based on studies dating back more than 25 years, [76][77][78] and more recently shown definitively. 79 The show that NOTCH signaling is required for differentiation, but not for self-renewal. 80

| Differentiation of neuroendocrine cells
During fetal development, the first epithelial cells to emerge from the progenitor basal stem cell compartment are the neuroendocrine cells. 81 These cells occur singly or in clusters, called neuroepithelial bodies, which appear to localize near bifurcations in the developing airway tubes. The function of neuroepithelial bodies is to detect local hypoxia/high CO 2 levels, and to signal this information through the vagus nerve to the brain. Development of neuroendocrine cells from basal stem cells is stimulated by NOTCH-hes1 signaling. The mature neuroendocrine cells store serotonin, calcitonin, calcitonin-gene related peptide (CGRP), and other bioactive amines in dense core vesicles. These cells also provide serotonin to both circulating platelets and to clusters of platelet-generating megakaryocytes where they occur in infected lung. Neuroendocrine cells, originating either from basal stem cells or club cells, can also self-replicate. Neuroendocrine cells are innervated from the basal side by the vagus nerve and can be uniquely destroyed by diphtheria toxin. 82 However, whether iPSCs can also reestablish connections between neuroepithelial cells and the vagus nerve is not yet known.

| Basal stem cell differentiation
Consistent with the data from Rock et al 79 and as summarized in Figure 1A, the basal stem cells are able to differentiate into club cells, . The exact source of Type II Alveolar Cells is not yet full understood. Basal stem cells (KRT14+), or club cells (SCGB1A1+) may be the origin of Type II Alveolar Cell, either directly, or by differentiation of club cells to basal stem cells. The Type II Alveolar Cell, marked by SFTPC+, is the source of surfactant (see the lamellar bodies) that allows the collapsed alveolar sac to re-expand with inhalation. Driven by WNT signaling, the Type II alveolar cell also differentiates to the Type I alveolar cell, across which O 2 and CO 2 exchange occurs ciliated cells, and goblet cells. Club cells themselves are able to replicate, as well as differentiate into ciliated cells with inhibition of NOTCH signaling. Club cells can also differentiate into goblet cells with increased NOTCH signaling. 80 In both mouse and human, club cell differentiation into goblet cells is also accompanied by SPDEF and FOXA3 expression following allergen exposure. 83 Finally, intermittent de-programming has been applied to club cells to generate a less differentiated state with the classical four component cocktail consisting of: Oct3/4, Sox2, Klf4, C-Myc. 84 The result is rapid expansion of a cell population that can be redifferentiated with 50% efficiency in vitro to a large club cell population. This redifferentiated club cell population can be used to repopulate denuded tracheal tissue in vivo, or in vitro at the air liquid interface, to yield a totally reconstituted epithelial surface, complete with apical expression of CFTR.

| Submucosal glands
Submucosal glands are found in the trachea in mouse, and in both the trachea and many generations of bronchiolar bifurcations in the human lung. 85 These glands are the result of substantial in-pocketing of the airway surface epithelium, which has the effect of vastly increasing surface area of the lung. It has been appreciated for many years that progenitor cells of the adult human airway contribute to submucosal gland development. 86 In the case of CF, where mucus is thickened, and clearance is profoundly compromised, the ducts connecting the submucosal glands to the airway surface epithelium are blocked by thickened mucus. Thus, it has been readily appreciated that cells in the submucosal glands may not have efficient access to medicinal agents administered to the airway, if the agent cannot easily navigate through mucin-blocked ducts. In fact, the basal stem cells in the submucosal glands are in a specialized niche, 87 where they are protected from toxic inhaled gases, particulates and microorganisms, and thus serve as a reserve to repopulate the airway epithelium. 88  | S17 response to regeneration following denuded tracheal xenografts.
Epithelial proliferation rates in the submucosal glands are apparently greater than in airway surface epithelium, when presented with equivalent regeneration challenges.

| Variant club cells as origin of type II and type I alveolar cells
Toward the end of gestation, the tips of distal branches further differentiate to form Type II Alveolar cells, which in turn give rise to Type I Alveolar cells, and thus the complex responsible for O 2 and CO 2 exchange. The transition between the proximal and distal airways may differ in some detail between mouse and human, so this is an area of lung biology that is still not fully understood. 87 Importantly there is little evidence that gas exchange, per se, is a critical problem for the CF patient. However, as further summarized in Figure 1B Dedifferentiation of club cells into functional basal stem cells has been reported following ablation of airway stem cells by doxycyclineinhalation-enabled diphtheria toxin expression. 91 The question mark in Figure 1B indicates the need for further work. Type II Alveolar epithelial cells can further differentiate into Type I Alveolar epithelial cells, thus providing the complete functional structure of the distal alveolar lung.
These remarkable examples of pluripotency for club cells has led some investigators to consider the club cell as yet another type of pluripotent stem cell, just like the basal stem cell. 20 The context seems important since under homeostatic conditions, turnover in the lung is slow.
However, in the event of major catastrophe, such as airway denudation or partial pneumonectomy, the remaining lung can rapidly enter a high production state to uniformly regenerate the lost functional capacity. 92

| iPSCs TO RESCUE CF LUNG DISEASE
As shown in Figure 2A-E, it has become possible to isolate iPSCs from a CF patient, convert the CFTR mutation to the wildtype 3 FIGURE 2 Differentiation of the iPSCs to cells in the pseudostratified squamous cell epithelial layer. A, Induced pluripotent stem cells (iPSCs). iPSCs can be prepared from CF patients who are homogeneous for the mutation [delF508]CFTR. The mutation can be edited to [wildtype]CFTR, and the modified iPSC can be expanded virtually without limit. B, Definitive Endoderm. IPSCs can be induced to differentiate into Definitive Endoderm by exposure to Activin A. Activation of Activin A occurs early in fetal development, in vivo, and was the source of motivation to start endoderm differentiation with this gene, in vitro. The code is red arrow, elevated and green arrow, reduced. C, Anterior Foregut Endoderm. Definitive endoderm can be differentiated into anterior foregut endoderm, marked by NKX.2.1, by reducing TGFβ with NOGGIN, and by reducing BMPs. FGF2 is elevated In order to delete thyroid precursors. Color code is as in part B. D, Lung endoderm. Lung endoderm can be directed to form from anterior foregut endoderm by exposure to WNT3a, BMP4, and FGF2. Color code is as in part B. E, Basal stem cells. These cells are precursors to all other cells making up the pseudostratified squamous cell epithelial layer lining the proximal airway. By changing to a recipe uniquely containing retinoic acid, basal stem cells be generated from lung endoderm. These cells are marked by TRP63 and at least 9 other markers that are shared by mouse and human. 18 Color code is as in part B. The REP symbol indicates that this cell can replicate itself. F, Activation of Basal Stem Cell differentiation. The problem of how basal stem cells are activated to differentiate into club cells, ciliated cells, goblet cells and neuroendocrine cells is not well understood. A compelling case has been made that it is driven by increases in cytokines and chemokines due to damage to the airway. 18 G, Type II and Type I Alveolar Cell. These cells represent the alveolar compartment and are responsible for gas exchange. The whorled objects in the Type II Alveolar cells represent surfactant-containing lamellar bodies. The surfactant is responsible for non-sticking expansion and contraction of alveoli. How Type II Alveolar cells are replicated remains poorly understood. They may be differentiated from basal stem cells, or from variant club cells, or from variant club cells that have dedifferentiated to basal stem cells. Type I Alveolar cells differentiate from Type II Alveolar cells in a WNT-dependent manner. H, Club cell. Club cells are another type of limited stem cell population that differentiates from Basal Stem Cells by an increase in NOTCH signaling. Club cells can de-differentiate to basal stem cells, for further differentiate to ciliated cells and goblet cells. It is possible that club cells differentiate into neuroendocrine cells. Club cells are uniquely sensitive to naphthalene ("moth balls"). When Club cells are killed by naphthalene, those surviving (variant) club cells near neuroepithelial bodies (NEBs) can restore the rest of the damaged epithelium. I, Ciliated cell. Ciliated cells are generated either from basal stem cells or from club cells when NOTCH signaling is reduced. Ciliated cells are responsible for mucociliary clearance of mucus and particulates in the airway. They are endstage cells that do not replicate or dedifferentiate. J, Goblet Cell. Goblet cells are the source of mucin secretion into the airway, and are generated either from basal stem cells or from club cells when NOTCH signaling is increased. Goblet cells respond to a variety of stimuli, including free neutrophil elastase, as well as airway dehydration, both characteristic of CF. There is evidence that goblet cells may be progenitors to ciliary cells. K. Neuroendocrine Cell. Neuroendocrine cells are the first of the epithelial cells to be generated during fetal development from airway precursor cells on the developing fetal airway. They contain calcitonin gene-related peptide (CGRP), substance P, somatostatin, serotonin, chromogranin A and synaptophysin. They occur either as single cells, or as clusters called Neuroepithelial Bodies (NEBs) at bronchial bifurcations. Their function is to detect hypoxia (and high CO 2 ) and to signal the details to the brain through contact with the vagus nerve. They also supply serotonin to platelets as they pass through the lung. Neuroendocine cells may be the source of human small cell lung cancer (SCLC) that often is associated with smoking condition, and then differentiate the corrected iPSCs to basal stem cells. The basal stem cell is a reasonable target, based on the information just summarized about this cell's apparent conditional capacity to form all known epithelial cell types in the lung. By way of a caveat for the foregoing, it is clear that the different cell populations are at best ca. 50% "pure," and sometimes substantially less. The actual pathway for differentiating iPSCs to definitive adult lung epithelial cells is full of many zigs and zags, well summarized in the review by Wong and Rossant. 19 Thus the in vitro proof-ofprincipal example given in the introduction represents more of a beginning than an end. The message seems to be that exactly what cells are being determined at any one time is stochastic or probabilistic, and is very dependent upon context. We have tended to agree that ". . .The lung can be considered a highly plastic and "democratic" tissue, in which a broad diversity of quiescent cell lineages can be induced to proliferate, dedifferentiate, or redifferentiate, and even change phenotype to repair an injured region." 20

| Definitive endoderm and anterior foregut endoderm
As shown in Figures 2A and 2B, it had been known from earlier embryological studies that high levels of Activin-A biomarked cells that had become definitive endoderm, and that NKX,2,1 biomarked cells that had become anterior foregut endoderm. Green et al 94  Consistently, pure Nkx2.1 cells, marked by Nkx2.1 mCherry from fetal mice, form clonal spheroids in semi-solid culture that can differentiate into polarized epithelium with multiple types of lung epithelial cells. 98

| Basal stem cells
After reaching the state of anterior foregut endoderm, this tissue can then be induced to differentiate into either Type II Alveolar Cells by a medium including retinoic acid (RA) (Figures 2C and 2G), or lung endoderm (Figures 2C and 2D) by alternatively adding WNT3a, BMP4, and FGF2. However, adding KGF, NOGGIN (to lower TGFβ) and retinoic acid to lung endoderm does finally yield a preparation enriched in basal stem cells (Figures 2D and 2E). Thus the Anterior Foregut Endoderm may stand at the bifurcation of development, leading either to proximal airway progenitor cells such as the Basal Stem Cell and its progeny, or to the distal airway, comprising the Type II Alveolar Cell and its Type I Alveolar Cell progeny.
Additional progress was also made in terms of increasing the efficiency of these development pathways by Mou et al, 99 who used a similar differentiation protocol, but differently timed, on both mouse and human iPSCs. Remarkably, these investigators showed that mouse and human progenitors were able to form respiratory epithelium in situ when injected subcutaneously in nude mouse models. Because NKX2.1 is also known as thyroid transcription factor-1, Longmire et al 100 were motivated to find a way to able to separately purify lung and thyroid NKX2.1 progenitors. They did so by inclusion of FGF2 in the differentiation medium (see Figures 2B and 2C). These pathways details are for the most part still definitive, 19 although significant progress in increasing efficiency has been accomplished. 95

| Human lung organoids (HLOs) in 3-D culture
A perceived limitation of the iPSC approach to therapy is that directed differentiation of iPSCs do not lead to a complete lung.

| Patient iPSCs as a tool for personalized drug discovery
As predicted by the inventors and developers of this technology, iPSCs could be deployed as a tool, either for analysis of a personalized CF disease phenotype, or as a personalized CF drug discovery platform. 102 For example, Wong et al 96 were the first to use iPSC-derived epithelium from a homozygous [delF508]CFTR CF patient to show that expression of mature [ΔF508]CFTR C-Band, defined by Western blot, could be induced by incubation with 10 μM C-18, an older VX-809-like compound. Cyclic AMP-activated iodide-efflux, a marker for CFTR chloride channel function, could also be detected in the presence of C-18. It is therefore possible that bystander/modifier genes than affect CFTR function in pulmonary epithelia may also be functionally detectable and, therefore, modifiable by gene editing or other drug regimens. In prospect, it would appear that the study of specific drug actions on these personalized iPSC-derived CF epithelial cells will be critical in both understanding patient-specific disease progression for different CFTR mutations, and in developing the best drug treatments for individual patients.
The concept of developing "lab-on-a-chip"(LOC) technology for CF has only recently begun to mature. 103-105 A LOC is a microfluidics device that integrates several laboratory functions on a single integrated circuit to achieve automation and high throughput. 106 For our present purposes, however, "lab-on-a-chip" has become "lungon-a-chip," in which cyclic mechanical stress can be applied to either proximal or distal lung epithelial cells. For example, it has been possible to mimic what might be happening in emphysema, where wall stresses might be great, 107 or ventilator-induced lung injury, where flow and stretch conditions can vary profoundly from normal states. 108 In pre-"lung on a chip" era studies on cyclic stretch effects on lung cells, it was found that stretch (15% strain for 4 h at 20 cycles/min), as induced by a ventilator-induced lung injury, could affect both glutathione biochemistry and secretion of IL-6 and IL-8 in the adenocarcinoma A 549 cell model of Type II Alveolar Cells. 109 Another early study in the same cell system showed that cyclic stretch could not only enhance proliferation, but also mitigate the detrimental effect of hypoxia on cell proliferation and viability. 110 However, Figure 3A-E demonstrates properties of a more recently developed sophisticated lung-on-a-chip," constructed by Guenat and coworkers, that is possibly more CF-relevant. 111 Figure 3A shows how a bioartificial alveolar membrane can be induced to stretch by a microdiaphragm, which is itself actuated by an electro-pneumatic driver. Figure 3B is a schematic for a triple "lung-on-a-chip," which consists of three microfluidic cell culture wells, each identical to the units described in Figure 3A. Figure 3C shows the actual physical assembly in which the lower chambers are filled with food dye for greater visibility. Figure 3D shows the xy-projection from a confocal microscope image of a confluent monolayer of normal human bronchial epithelial cells (16HBE14o-), growing as a monolayer on  Figure 3A. Here, the red fluorescence marks E-Cadherin (adherens junctions) and the blue fluorescence marks cell nuclei. Figure 3E shows the effect of static versus dynamic culture conditions, and of exposure time, on the secretion of IL-8 from 16HBE14o-cells. These experiments show that long periods of stretching, at a frequency approximating the normal breathing rate (10-12 breaths/min), and normal mechanical strain (5-12% linear elongation), significantly stratifies the level of IL-8 secreted in a dynamic culture over a 48 h period from a static culture over the same time period. In terms of deploying CF patient-derived iPSCs and their differentiated cellular products, this "lung-on-a-chip" platform could therefore provide a complete way to study drug effects and lung biology for individual CF patients, especially patients with the less common CFTR2 mutations. [112][113][114][115] Relevantly, tidal breathing has been shown to affect phasic motion-induced shear, with consequences for release of nucleotides to the extracellular space and regulation of pericellular liquid homeostasis in CF patients. 116 The lung-on-a-chip concept also permits a more global physiological view of CF, in which the disease is more than just a CFTR mutation. Rather it provides a more physiologically relevant context for analysis. 117

| Replacement of mutant lung cells by wildtype lung cells in the living lung
The so-far unresolved challenge has been how to use the iPSC technology for therapy: the replacement of mutant lung epithelial cells with their wildtype equivalents. We give several examples here to illustrate that this approach has been tested in different ways, and for different indications. One approach has been to attempt to perform the replacement in the otherwise diseased but still functional lung. One strategy has been to administer iPSCs directly by the intravenous route. [118][119][120] For example, acute lung injury (ALI) was induced in mice by administering endotoxin by the endotracheal route. 118 Mouse iPSC cells were then delivered into the mouse by tail vein injection. The consequences were that iPSC incorporation was enhanced; histopathologic changes, NFκB and neutrophil accumulations were reduced; and hypoxemia and pulmonary function were rescued. In these cases, iPSCs were apparently being used as a drug. In another strategy, immuno-competent mice, that had been treated with bleomycin to induce a model of interstitial fibrosis, were injected with cultured allogeneic and syngeneic adult lung spheroids. 17 Progression of fibrosis and inflammation were suppressed, without eliciting significant immune rejection. A human trial is said to be planned. In an alternative to the intravenous injection route, lungs of mice and pigs, whose lungs had been pre-injured/pre-conditioned with 2% polidocanol (PDOC), mature human airway epithelial cells were administered intra-tracheally. 121 Based on a fluorescent label, cell retention/ "engraftment" 2 days later was ca. 10% in mouse and ca. 22% in the pig. The purpose of the PDOC, an FDA-approved local anesthetic and anti-pruritic, was to temporarily remove the surface airway epithelial cells, thereby opening up a place for epithelial cell engraftment, and activating proliferation of normally quiescent epithelial cells by enhancing cytokine and chemokine expression in the airway.
The "pre-injured/pre-conditioned strategy" has also been attempted in studies with direct CF-centric relevance. In an early study with bone marrow (stem) cells (BMC), Wong et al 122  Recalling that transdifferentiation of BMCs was infrequent, nonetheless, apical [wildtype]CFTR protein was detected in the reconstituted airway; P aeruginosa infection was delayed; and survival was increased.
The benefit lasted for more than 6 months. Thus for those patients who are null for CFTR, and are therefore unlikely candidates for the newer corrector and potentiator drugs, this kind of approach, possibly with wildtype iPSCs instead of BMCs, may eventually be to their benefit.

| Decellularized lungs and artificial scaffolds
In response to the fact that there are so few lungs available for transplant, and so many patients in need, the idea of artificial lungs has developed significant appeal. 124,125 One more recent CF-relevant idea has been to use a patient's corrected iPSCs to repopulate a decellularized lung, and then to replace the diseased lung with the re-cellularized lung. The object of studies on lungs, ranging from mouse 126 to a non-human primate (Rhesus macaque), 127 has been to isolate an intact scaffold made up of a passive extracellular matrix, which can be repopulated with autologous stem or progenitor cells.
The first problem has been to remove as much cellular material as possible, without the entire scaffold/lung falling apart. The solution has been threefold: first, to wash the lung as completely as possible with divalent cation-free aqueous solutions; second, to extract as much as possibly with a succession of detergents such as sodium dodecyl sulfate (SDS), sodium deoxycholate (SDC), and CHAPS 128 ; and third, to then fix the entire organ in formaldehyde. Following many additional steps, a somewhat flexible scaffold is generated, upon which cellular re-population can hopefully be achieved. Presently, the experiments have been limited to 350 micron thick sections of the scaffold, which have been used to test the ability of stem cells to grow and assemble into a continuous proximal and distal lung. However, there is still much to learn even from this simple kind of experiment. progenitor cells bound to the scaffold surface and differentiated into mature airway epithelia, including ciliated cells (FOXJ1+, TUBB4A+), club cells (SCGB1A1+), and basal cells (TRP63+, KRT5+). CFTR protein was present, and was also functional, as defined by cAMP-dependent iodide efflux. Pit-like structures upon the surface could be discerned by optical microscopy. These were interpreted as nascent submucosal glands, and confirmed by scanning EM. Importantly, the epithelial differentiation process proved to be dependent upon the presence of heparinase I-sensitive heparin sulfate proteoglycans in the decellularized matrix. This paper also emphasized an important control: showing that decellurized kidney sections, a mesoderm-derived tissue, could not replace decellularized lung sections as a matrix for constitution of the airway epithelium. Thus this heavily washed and fixed scaffold is still biochemically "alive." Nonetheless, the lung physiologist reading this will think of many more organismic properties that this bioengineered lung must display in order to be durable and operational.

| CONCLUSION
Based on the state of the science, it appears that the best use of iPSCs for CF lies in using them from individual CF patients in their differentiated states for personalized medicine. For example, the impact of different bystander/modifier genes on drug responses can be studied today using a lung-on-a-chip approach. Thus the ability of specific variants in a patient's bystander/modifier genes can be assessed in terms of how a specific drug might affect CFTR trafficking and CFTR function, including cAMP-activated chloride transport or inflammation. However, in thinking of using iPSCs to create a durable physical therapy for CF lung disease, we quote the opinion of Shinya Yamanaka, the 2012 Nobel Laureate for the development of the iPSC technology, who concluded that "the potential is enormous, but many obstacles remain before this technology can become converted to therapy." 1 7 | GLOSSARY This glossary is provided as an aid to following the different workflows. Some of these entries are described in depth in the text. Unreferenced information has been abstracted from either Wikipedia or GeneCards. Gene: Gene can refer to the DNA, or it can refer to the cognate messengerRNA (mRNA). In both cases the abbreviation is in italics. When written in plain text the word gene refers to the protein that is made from the corresponding mRNA. Retinoic acid (RA): is an active metabolite of vitamin A. It binds to the retinoic acid receptor (RAR), which then binds to retinoic response elements (RAREs) on target DNA promoters. It is a potent morphogen in hind brain development and in the heart. SCG1A1: Secretoglobin family 1A member 1. This gene is also known as Uteroglobin. It is a biomarker for club cells. Variant club cells are also marked by SCGB1A1, and are located at branch points between a terminal bronchiole and the distal airways. It is hypothesized that they are the progenitors of Type II Alveolar Cells. Alternatively, they may dedifferentiate into basal stem cells, which are then the actual progenitors for Type II Alveolar Cells.  TRADD is the first intracellular adaptor to the TNF Receptor 1/TNFa complex. The function of this binding activity is to activate IKKαβγ, downstream NFκB signaling, and inflammation. In the presence of wildtype CFTR, TRADD is directed to the proteasome, NFκB is not activated, and no inflammation occurs. 37 Transcription Factor: Transcription factors bind to DNA sequences in a gene promoter, or to other transcription factors bound to promoter DNA, and either drive or inhibit gene expression. An example of a driver is NFκB. An example of an inhibitor is the glucocorticoid receptor.
TRP63: Transformation-related protein p63, also known as TP63. TRP63 is an evolutionary founding member of the p53 superfamily, and is regulated by SOX2 to induce the formation of basal stem cells.
TUBB4A: Tubulin beta-4A. TUBB4A is a form of tubulin (TUB), a major cytoskeletal protein, which together with TUBB4B is said to be preferentially and highly expressed in the central nervous system. Curiously, it is also a biomarker for ciliated cells.
UCHL1: ubiquitin carboxy-terminal hydrolase L1. UCHL1 is mostly found in the brain and testes/ovary. However, it is also found in lung and is a biomarker for neuroendocrine cells.
Vagus nerve: The vagus nerve is the tenth cranial nerve and originates in the area postrema. It provides both sympathetic and parasympathetic control of the heart, lungs and digestive tract. Conduction pathways in the vagus nerve are both afferent (going to the brain) and efferent (going away from the brain). In the lung, the neuroendocrine cells provide information on local hypoxia at airway bifurcations.
WNT: a combination name of the wingless gene and the Int1 gene, first discovered in Drosophila, but highly conserved. It forms three (3) signaling pathways, all activated by binding to one of the family of Frizzled receptors, which passes the information to an internal Disheveled protein. Pathways in embryology include cell fates, proliferation, and migration. An example is WNT3a.

ACKNOWLEDGMENTS
The authors wish to acknowledge the use of the cartoon in Figure 1A and parts of this cartoon in Figure 2