0
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      Cystic Fibrosis: “Ionocyte Modulators”?

      editorial
      1 , 2 , 3 , 1 , 2 , 3 , 4
      American Journal of Respiratory Cell and Molecular Biology
      American Thoracic Society

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          The most prominent manifestations of cystic fibrosis (CF) include thickened respiratory secretions and recurrent pneumonia, which, before the advent of cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapy, commonly led to respiratory failure and eventually to death. CF is caused by mutations in the CFTR gene that result in the production of a defective CFTR protein. CFTR modulators that directly target the dysfunctional CFTR protein have dramatically improved outcomes for patients with certain mutations. Normally, in the airway, CFTR is localized to the apical surface of airway epithelial cells and is responsible for the transport of chloride and bicarbonate ions, whose concentration determines the tonicity and pH of the airway surface fluid, both essential parameters for effective mucociliary transport and host defense. The recent discovery of the pulmonary ionocyte as the airway epithelial cell type that expresses the most CFTR per cell upended our understanding of the localization of CFTR within the airway epithelium. This created a mystery in which one could plausibly ask, is CF a “rare cell disease”? However, airway secretory cells, and surprisingly even airway basal stem cells, also express low levels of CFTR transcript. Recently, unique ionocyte-specific patterns of CFTR regulation have been documented, suggesting that CFTR may have cell type–specific functionality (1). So, how does each of these CFTR-expressing cells participate in CF, and what is their normal physiologic function? Indeed, the initial papers reporting the discovery of the pulmonary ionocyte implicated a role for ionocytes in regulating airway physiology, including a role in modulating airway viscosity and CFTR-mediated current in murine and human large airways (2, 3). However, a subsequent paper suggested that, in the human small airways, ionocytes were scarce and that the more abundant secretory cell population housed the majority of epithelial CFTR (4). In this issue of the Journal, Cai and colleagues (pp. 295–309) add to this nascent literature by presenting rigorous evidence that supports a central role for ionocytes in CF airways (5). Using human tissue sections, human air–liquid interface (ALI) cultures, and ferret genetic modulation, the investigators report that sonic hedgehog (SHH) signaling pathway activation is increased in CF. Furthermore, they found that the increased SHH signaling activity results in the presence of greater numbers of ionocytes. Perhaps most important, the increased number of ionocytes correlated with the increased CFTR current. To start, Cai and colleagues demonstrate that SHH signaling is increased in CF by immunohistochemical analysis of airway tissue sections. In CF samples, BSND (Barttin)-positive ionocytes approximately doubled in number. Of note, these samples were collected from patients before 2000, so presumably, these CF epithelial cells were exposed to a pathologic CF milieu because effective modulator therapy with elexacaftor-tezacaftor-ivacaftor had not yet been discovered. Remarkably, the elevated SHH signaling found in actual CF primary patient tissue persisted when patient-derived epithelium was cultured. This suggests that airway basal stem cells retain a memory of a “CF state,” at least transiently. It also suggests the hypothesis that CFTR dysfunction itself or the resulting pathologic CF milieu leads to increased SHH signaling, which in turn leads to a rise in ionocyte numbers. Indeed, when the authors inhibited CFTR function, SHH signaling pathway components increased. However, there is no mechanistic understanding linking CFTR physiology to SHH signaling. Does SHH signaling influence ionocyte numbers? To address this, Cai and colleagues manipulated the SHH pathway pharmacologically in human control (non-CF) ALI-cultured cells. They demonstrate that the SHH agonist, SAG, increases the mRNA expression of ionocyte genes FOXI1 and BSND by 1.5- to 2-fold. There was also an increase in CFTR protein by Western blot analysis. Looking at functional data, treatment with SHH agonist leads to increased forskolin/IBMX (3-isobutyl-l-methyl-xanthine)-stimulated short-circuit current, and GlyH101 inhibited CFTR currents. Conversely, inhibiting SHH signaling in both fully mature and maturing ALI reduced ionocyte numbers, ionocyte mRNA expression, CFTR protein by Western blot analysis, and CFTR short-circuit current. Thus, modulating ionocyte numbers leads to a corresponding change in CFTR-mediated chloride current. In contrast, SHH pathway inhibition leads to increased secretory cell numbers but less CFTR protein and current, demonstrating an anticorrelation of secretory cell numbers and CFTR expression and function. How does SHH signaling increase ionocyte numbers? Because ciliated cells are known to express SHH pathway components (6), the mechanistic link likely involves non–cell autonomous processes (7). To gain further mechanistic clarity, ionocyte progenitors need to be well defined to establish how SHH signaling eventually leads to increased ionocyte differentiation. Is the role of SHH in mediating ionocyte numbers conserved in animal models of CF? To obtain rigorous genetic evidence of the impact of SHH signaling on ionocytes and CFTR, the authors made a thoughtful choice by turning to the ferret model. This model not only allows rigorous and elegant genetic modulation but also has the virtue that CFTR function is preserved in the setting of serial passaging (5), unlike the case with human cells, where CFTR function deteriorates with propagation in culture (8). SHH pathway pharmacologic modulation and CRISPR/Cas9 targeting of the SHH pathway components in the ferret epithelia recapitulated the changes in ionocyte gene expression and chloride and bicarbonate currents that were seen in human epithelia. Are there different flavors of ionocytes? Indeed, the authors found heterogeneous populations of ionocytes that express differing sets of proteins, suggesting the hypothesis that there are functional subtypes of ionocytes, as seen in other tissues and organisms (9). Specifically, they identify ionocytes with various combinations of BSND, ATP6V1G3 (a component of the V-ATPase), and FOXI1 protein expression. Lending more intrigue to this finding, not all ionocytes expressed CFTR. Interestingly, CFTR current increased in proportion to the numbers of BSND+ CFTR+ ionocytes in human ALI. Furthermore, recent single-cell RNA-sequencing data from patients without and with CF showed a twofold statistically significant increase in CFTR expression specifically in ionocytes (10). In aggregate, these findings point to the need to interrogate not just the number of ionocytes but also their specific patterns of gene and protein expression as a prelude to defining the functional attributes of different subclasses of ionocytes. In summary, Cai and colleagues have demonstrated that SHH signaling is elevated in CF epithelia. Increased SHH signaling increases ionocytes while concomitantly decreasing secretory cell numbers. The changes in ionocyte numbers seem to correlate with levels of CFTR protein and ion current in human and ferret, implying that ionocytes may actually play a dominant role with respect to CF-related physiology in bronchial epithelia. This suggests that the role of the ionocyte in the small airway needs to be revisited in a variety of models, although it remains possible that the function of ionocytes and secretory cells differs depending on where they are found in the airway tree. The demonstration that ionocyte numbers are directly related to CFTR current suggests a new therapeutic approach to increasing CFTR activity. Modulating ionocyte numbers twofold leads to physiologically relevant increases in CFTR current. This, of course, is entirely plausible because rare cells in other systems subtend key physiologic roles. Thus, the rigorous and cutting-edge work by Cai and colleagues has a number of important implications, and, like all such work, it raises a myriad of fascinating questions. Pharmacologic manipulation designed to increase ionocyte numbers could serve as a therapeutic strategy for increasing CFTR-mediated anion secretion in CF. Even more speculatively, ionocytes might be reduced in diseases as diverse as chronic obstructive pulmonary disease, ciliary dyskinesia, or even asthma and pneumonia, in which case “ionocyte deficiency” might lead to poor mucociliary transport, thickened secretions, and abnormalities of host defense and therefore may also benefit from “ionocyte modulators.” However, quantifying ionocyte deficiency and establishing a correlation to the severity and symptoms of complex pulmonary disease is challenging because of the scarcity of ionocytes and the need for extensive sampling to determine their numbers. In the case of SHH modulation specifically, increasing SHH signaling has been shown to decrease the ciliary beat frequency and to acidify airway surface liquid pH because of dampening cAMP signaling (6). Therefore, increasing SHH signaling might impair mucociliary transport and host defense even if this strategy does increase ionocytes and CFTR expression. In addition, the modulation of ionocytes is not equivalent to the modulation of CFTR, so ionocyte modulators may have unintended and unpredictable beneficial or harmful effects, given our incomplete understanding of the physiologic role of ionocytes, including possible roles that are not directly linked to CFTR biology. Moreover, ionocyte numbers are likely to be influenced by multiple developmental signaling pathways that presumably act on a very rare and as yet poorly defined airway epithelial progenitor cell. Perhaps different mechanistic classes of ionocyte modulators targeting these putative pathways could be found. Perhaps existing drugs already have “off-target” effects that alter ionocyte numbers. Finally, ionocyte modulators may have a role in instances where CFTR modulators fail to fully rescue CFTR-mediated anion secretion or to complement gene therapy strategies that result in an incomplete restoration of normal CFTR expression and function.

          Related collections

          Most cited references10

          • Record: found
          • Abstract: found
          • Article: not found

          A revised airway epithelial hierarchy includes CFTR-expressing ionocytes

          We combine single-cell RNA-seq and in vivo lineage tracing to study the cellular composition and hierarchy of the murine tracheal epithelium. We identify a new rare cell type, the FoxI1-positive pulmonary ionocyte; functional variations in club cells based on their proximodistal location; a distinct cell type that resides in high turnover squamous epithelial structures that we named “hillocks”; and disease-relevant subsets of tuft and goblet cells. With a new method, Pulse-Seq, we show that tuft, neuroendocrine, and ionocyte cells are continually and directly replenished by basal progenitor cells. Remarkably, the cystic fibrosis gene, CFTR, is predominantly expressed in the pulmonary ionocytes of both mouse and human. Foxi1 loss in murine ionocytes causes a loss of Cftr expression and disrupts airway fluid and mucus physiology, which are also altered in cystic fibrosis. By associating cell type-specific expression programs with key disease genes, we establish a new cellular narrative for airways disease.
            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            A single cell atlas of the tracheal epithelium reveals the CFTR-rich pulmonary ionocyte

            The functions of epithelial tissues are dictated by the types, abundance, and distribution of the differentiated cells they contain. Attempts to restore tissue function after damage require knowledge of how physiological tasks are distributed among cell types, and how cell states vary between homeostasis, injury/repair, and disease. In the conducting airway, a heterogeneous basal cell population gives rise to specialized luminal cells that perform mucociliary clearance 1 . We performed single cell profiling of human bronchial epithelial cells and mouse tracheal epithelial cells to obtain a comprehensive picture of cell types in the conducting airway and their behavior in homeostasis and regeneration. Our analysis reveals cell states that represent known and novel cell populations, delineates their heterogeneity, and identifies distinct differentiation trajectories during homeostasis and tissue repair. Finally, we identified a novel, rare cell type, which we call the ‘pulmonary ionocyte’, that co-expresses FOXI1, multiple subunits of the V-ATPase, and CFTR, the gene mutated in cystic fibrosis (CF). Using immunofluorescence, signaling pathway modulation, and electrophysiology, we show that Notch signaling is necessary and FOXI1 expression sufficient to drive the production of the pulmonary ionocyte, and that the pulmonary ionocyte is a major source of CFTR activity in the conducting airway epithelium. The conducting airway is lined by a pseudostratified epithelium consisting of basal, secretory and ciliated cells, as well as rare pulmonary neuroendocrine cells (PNECs) and brush cells 2 . Studies of lineage tracing and regeneration post-injury show that basal cells are a heterogeneous population containing the epithelial stem cells 3,4 . Basal cells differ in their expression of cytokeratins 14 and 8 (Krt14 and Krt8) and luminal cell fate determinants that are upregulated upon injury 2,5 . To identify the full repertoire of basal cell molecular states, and to identify candidate gene expression programs that might bias basal cells to self-renew or to adopt differentiated fates, we performed single-cell RNA profiling on airway epithelial cells. We also sought to elucidate the molecular composition of rare PNECs and brush cells, which have fewer lineage markers and are harder to define functionally 6,7 . Because our approach is unbiased and comprehensive, it could also identify new cell types with a role in mucociliary clearance. We performed single-cell RNA-seq 8 (scRNA-seq) on 7,662 mouse tracheal epithelial cells and 2,970 primary human bronchial epithelial cells (HBECs) differentiated at an air-liquid-interface (ALI) 9 (Fig. 1a,b). As there are well-documented differences between mouse and human airways 10 , using these two systems allows comparative analyses and prioritization of common findings between mouse and human. This also provided in vivo validation of findings in the culture model, which lacks non-epithelial cells and uses defined culture conditions. A similar analysis of mouse tracheal epithelial cells in a co-submitted paper (Montoro et al., co-submitted) corroborates many of our findings. We visualized the single cell data using a graph-based algorithm (SPRING 11 ) that conserves neighboring relationships of gene expression, facilitating analysis of differentiation trajectories. The resulting graphs revealed a non-uniform continuum structure spanning basal-to-luminal differentiation, with rare gene expression states representing satellite clusters (see ‘Data availability’). Using spectral clustering, we partitioned cells into populations with specific, reproducible gene expression signatures (Fig. 1c,d). Based on enrichment of previously annotated markers (Supplementary tables 1, 2), we identified clusters in mouse (Fig. 1c) and human (Fig. 1d) which represented known cell types 2,7 : basal, secretory, ciliated, brush and PNECs. We performed pairwise correlation analysis as a measure of relatedness between clusters, and curated a list of transcription factors, surface molecules and kinases expressed in each cluster (Extended Data Fig. 1, Supplementary Tables 1-3). Our analysis confirmed previous findings 5,12–14 that basal and secretory populations are heterogeneous, and uncovered additional molecular heterogeneity (Extended Data Fig. 2, 3). Basal cells formed a continuum of states defined by gene modules associated with a basal-to-luminal gene expression axis (Krt5 vs Krt8) as well as by variable expression of genes associated with basement membrane deposition and remodeling. In both mouse and human, Col17a1/COL17A1 and Igfp family members (Igfp3/IGFBP6) correlated with the basal cell sub-population marker Krt14, while in mouse an independent Krt14 - module associated with basal cell adhesion molecule Bcam and with Decorin Dcn, which regulates collagen fibrilogenesis. Among secretory cells, many differences were associated with different levels of maturity, with the least mature cells expressing basal cell transcripts (e.g. Krt5/KRT5 and Trp63/TP63), and the most mature expressing Muc5b/MUC5B. Yet secretory cells also differed in other ways. In human, one cluster associated with antigen presentation (HLA gene family members). In mouse, the secretory cells appeared to associate with two distinct trajectories from the basal layer: those expressing Krt4, and those emerging from a Krt4-low state marked by Trp63, Bcam and Dcn. Heterogeneity of both basal and secretory cells was also associated with tens of other genes with diverse functions, including signaling molecules (e.g., Wnt10a) and early specifiers of mature lineages (e.g., FoxJ1) (Extended Data Fig. 2, 3). Our analysis also revealed gene signatures of epithelial cell states not previously described. First, the paired cytokeratins 4 and 13 (Krt4/13) defined a unique cluster in the mouse dataset located between basal and secretory, suggesting this may be a transitional cell state (Fig. 1c, Supplementary Table 1). Immunofluorescence for Krt4 in mouse tracheal epithelium demonstrated that it was co-enriched in subsets of Krt5+ basal cells, Krt8+ luminal cells and Scgb1a1+ club (secretory) cells, but not in Foxj1+ ciliated cells (Extended Data Fig. 4a). This pattern is reminiscent of the proposed model for basal luminal precursors (BLPs), a subset of non-transit amplifying basal cells with upregulated luminal markers 13 . In addition, KRT4/13 expression was closely correlated and defined a major axis of heterogeneity in basal and differentiating HBECs (Extended Data Fig. 3b). Second, in the human single cell map, we identified a FOXN4 + cluster that was highly enriched for the ciliated cell specification factor FOXJ1 but low for markers of maturation, including the ciliary component TUBB4B (Fig. 1d, Supplementary Table 2). Foxn4 is known to drive robust transcription of ciliated genes during multiciliated cell differentiation in Xenopus 15 , suggesting that this cluster represents a state of multiciliated cell differentiation. We confirmed the existence of this cluster by immunofluorescence, showing that FOXN4 was indeed enriched in a subset of FOXJ1hi cells but not in cells containing mature cilia (Extended Data Fig. 4b). Third, in the human data we identified a novel cluster enriched for SLC16A7 (Fig. 1d, Supplementary Table 2), which encodes the monocarboxylate transporter 2 (MCT2) previously shown to be involved in acidification of CF HBEC cultures 16 , as well as AIRE, the gene that drives negative selection of self-reactive T cells in thymic epithelium 17 . This cluster contained the largest number of highly specific genes in the dataset, with a greater percentage of mitochondrial genes. This cluster may reflect cellular stress 18 or may represent a unique antigen-presenting airway epithelial cell. Finally, a single cluster identified in both mouse and human was enriched for ion transporters and the transcription factors Foxi1, Ascl3 and Tfcp2l1 (Fig. 1c,d, Supplementary Tables 1,2). The novel cluster expressed subunits of the Vacuolar ATPase (V-ATPase) proton pump, which are also expressed in Foxi1-expressing ionocytes in the mucociliary epithelium of Xenopus larval skin, intercalated cells of the mammalian kidney and in forkhead-related (FORE) cells of the inner ear 19,20 . This cluster was highly enriched for Cftr, the gene that encodes a critical chloride channel that is mutated in CF, as well as for genes encoding multiple ClC chloride channels (e.g., NKCC1, ClC-Kb), the calcium-activated potassium channel KCNMA1, and members of the Slc9 family of Na+/H+ exchangers (Nhe4 in mouse and NHE7 in human). We named these cells pulmonary ionocytes. To further identify cell states in the conducting airway that may emerge or expand following injury, we performed scRNA-seq and immunofluorescence on regenerating tracheas at 1, 2, 3, and 7 days after polidocanol-induced injury (Fig. 2a). We again visualized cell transcriptomes using a SPRING graph, now expanded to 14,163 cells to reveal detailed changes in epithelial cell states during repair (Fig. 2b). This identified two states specific to injury response (Extended Data Fig. 5a, methods). The first state appeared at 1 dpi (Fig. 2b,c light gray), corresponding to Krt5 + basal cells in cycle and co-expressing additional cytokeratins including Krt14, Krt8, and Krt4/13, which were largely non-overlapping in homeostasis (Fig. 2d). The second injury-specific state, which appeared 2 and 3 dpi, included cells transiting directly from basal to ciliated (Fig. 2b,c dark gray) rather than differentiating through a secretory progenitor (Fig. 1a). We detected 1,237 genes varying in expression during multiciliated cell differentiation, including the specification factors Foxj1, Myb and Mcidas 21 (Extended Data Fig. 6, Supplementary Table 4). Early secretory cell states also reappeared at 2 and 3 dpi. By 7 dpi, the relative abundance of cell populations, including rare populations (PNECs, brush cells and pulmonary ionocytes) largely returned to that seen in uninjured tracheae (Fig. 2c, Extended Data Fig. 5b). Our data open up a range of possible avenues for future research, from the significance of the gene modules defining basal and secretory cell heterogeneity, to the catalog of potential regulators and components for rare PNEC and brush cells, premature ciliated cells, Krt4/Krt13 + cells, and pulmonary ionocytes. In this study, we focus on the localization, specification and function of the newly identified pulmonary ionocyte. We first validated the presence of the pulmonary ionocyte population by immunofluorescence. FOXI1 labeled 1–2% of HBECs and, as predicted, was distinct from basal cells (TP63), secretory cells (MUC5B), ciliated cells (FOXJ1) or neuroendocrine cells (ASCL1) (Fig. 3a). Immunostaining demonstrated apical enrichment of the V-ATPase in FOXI1+ HBECs, similar to what has been shown for other Foxi1+ epithelial lineages 19,20,22 , as well as Nerve Growth Factor Receptor (NGFR, Extended Data Fig. 4c), confirming the marker gene enrichment identified by scRNA-seq (Extended Data Fig. 1b). Also as predicted, CFTR mRNA was highly enriched in Foxi1-expressing cells in mouse trachea and primary human bronchial tissue (Fig. 3b), compared to low expression throughout the epithelium (Extended data Fig. 7c,d). Interestingly, FOXI1 + cells were more concentrated in bronchial gland ducts than in the surface epithelium (Extended data Fig. 7d), a pattern similar to previously described rare, CFTR-high cells 23 . In other proton-secreting cells Foxi1 specifies the lineage and regulates expression of V-ATPase subunits 19,20 ; therefore, we next asked whether FOXI1 was sufficient to specify pulmonary ionocytes in HBEC cultures. We performed scRNA-seq of cells transduced with GFP:FOXI1 (n=10,330) or GFP alone (n=9,436) and mapped the data onto the reference HBEC state map (Fig. 3e, Extended Data Fig. 8a-e). Cultures transduced with GFP:FOXI1 had significantly higher numbers of cells classified as ionocytes (23-fold increase, p 5% of total counts in at least one cell. Data visualization using SPRING and clustering. To visualize the high dimensional gene expression data, we applied SPRING 11 , a method for building a k nearest neighbors (kNN) graph of cells and representing it in 2D using a force directed layout. Clusters were identified by applying spectral clustering on the same adjacency matrix as used for SPRING (implementation in python, sklearn.cluster.SpectralClustering(affinity=‘precomputed’, assign_labels=‘discretize’)). Clusters were assigned labels (e.g., secretory, basal) based on marker gene expression. In the SPRING plot of human data (Fig. 1), clusters representing intermediate states with no unique gene expression are shown in gray. Cell population-specific gene identification (Fig. 1). To be considered as specific to population i, a gene had to satisfy the following criteria: Be statistically significantly higher in population i compared to all other cells as determined by a two-sided permutation test using the difference in sample means as the test statistic (FDR 50 TPM in population i. Average expression in population i at least 1.5-fold higher than in any other population (i.e., max-to-second-max ratio > 1.5). At pseudo value of 10 TPM was added before division. Be max in population i for 4/4 (mouse) or 2/3 (human) of the biological replicates. Fig. 1c-d shows the expression level of the top 50 such hits ordered by decreasing max-to-second-max ratio. For each gene, 100% was set at the maximum expression per cluster (average of all replicates). The color was saturated at 20% (low) and 100% (high). Detailed gene lists are provided as Supplementary Tables 1-3. For Extended Data Fig. 1, transcription factor (TF) lists were obtained from animaltfDB 35 , and GO terms GO:0016301 and GO:0009986, including any descendent terms, were used for kinases and surface molecules, respectively. Identification of correlated gene modules within basal and secretory cells To characterize the heterogeneity within basal and secretory cells, we identified modules of correlated genes. For mouse, we performed the following steps (see also Extended Data Figures 2 and 3): Select basal cells (same procedure for secretory). Identify 3000–5000 most variable genes. Calculate gene-gene rank correlation. Retain genes with r>0.2 with at least 4 other genes. In mouse, Krt14 did not meet this criterion therefore was included manually. Heat map rows and columns (Extended Data Figures 2 and 3) were hierarchically clustered (distance defined as 1-rSpearman, Ward linkage). For human data, we first considered the basal, secretory, and intermediate state cells collectively to identify two main modules of anti-correlated genes (Extended Data Figure 3). From there, we selected genes specific to basal and recalculated gene-gene correlation but within basal cells only. The same was performed with secretory. Smoothing (data imputation) Smoothing was carried in Fig. 2c,d, and Extended Data Figures 5a and 9. All data shown in other figures is not subject to smoothing/imputation. Data smoothing, or equivalently imputation, was carried out using a graph diffusion approach on the k-nearest neighbor graph G defined above by SPRING. G is an unweighted undirected graph. The smoothing operation replaces a scalar quantity x i on node i of the graph, e.g., raw expression level of a gene, with a smoothed value x(s) = Osx, where the smoothing operator is Os = eLβ and L is the random walk graph Laplacian of G. The smoothing operator accepts a single parameter, β,which determines the kernel size, i.e., the extent of smoothing. This parameter is equivalent in physical terms to diffusion time: longer times lead to broader diffusion. For all plots shown, we used β = 1. Analysis of cell density changes relative to uninjured To visualize which cell populations are enriched at a given time point relative to uninjured (Fig. 2) the following was performed for every time point t of mouse recovery data: Get every cell from t to vote for its 10 nearest neighbors among all mouse cells and count votes. Smooth vote counts on the graph (see previous section for smoothing). Smoothed vote counts are a proxy for the density of cells from time point t on the graph (see also the two left-most plots of Extended Data Fig. 5a). Normalize the total vote count to 1. Divide the density at time point t by the density of cells in uninjured. Identification of recovery-specific cell populations The procedure is summarized in Extended Data Fig. 5a. To identify recovery specific cell populations in the SPRING plot combining all mouse data (populations in gray in Fig. 2b), we first performed steps a) and b) described in the previous paragraph to determine the density of injured cells on the graph. Next, a threshold of 25 smoothed counts was selected by visual inspection of the distribution of votes, and cells receiving fewer than 25 votes were considered depleted in uninjured (i.e., recovery-specific). Recovery-specific cells were split into two clusters by spectral clustering, and labels were assigned based on characteristic gene expression. Cells from the mouse recovery time course experiment that were not recovery-specific inherited the label of their single nearest neighbor in uninjured mouse data (Euclidean distance in principle component (PC) space of most variable genes). Analysis of recovery-specific trajectory from basal to ciliated The procedure is also summarized in Extended Data Fig. 6. 609 recovery-specific cells from 24–72 hours post-injury, and forming a continuum between basal and ciliated cells, were manually selected on the SPRING plot and used for Population Balance Analysis (PBA), a method developed in our lab for describing differentiation trajectories 36 . For this analysis, ‘source’ and ‘sink’ cell populations were defined as the basal and multiciliated tips of the cell kNN graph respectively. Cells were then ordered on the graph by the diffusion ‘potential’ parameter of PBA (a measure pseudotime of progression from source to sink). To smooth the gene expression of individual cells, a moving average with window size of 100 cells was calculated. Identification of differentially expressed genes along the basal-to-ciliated trajectory Temporally varying genes were identified using a previous method 37 with minor changes. Prior to statistical testing, the following filters were applied on the full gene list considering only the 609 cells forming the basal-to-ciliated trajectory: Expression level: at least 3 normalized counts in at least 3 cells. Variable: Fano factor >1. Notably, none of these filters considers the cell ordering. For each gene i of the surviving 4651 genes, a statistic t was calculated: ti,observed = mi,max – mi,min, where mi is a vector with the expression level of gene i in the 510 average cells after application of a moving average over cells ordered using PBA. The procedure was repeated on shuffled cells for multiple permutations, each time resulting in a ti,random value. The one-sided p-value for gene i was defined as the fraction of times ti,observed ≥ ti,random. To account for multiple hypothesis testing, the false discovery rate was controlled at 5% using the Benjamini-Hochberg procedure. For each of the 4651 genes used in the permutation test, we also calculated the maximum fold-change defined as: FC max = m i , max + 100 TPM m i , min + 100 TPM . 1237 genes with FCmax≥2 and FDR≤5% were considered differentially expressed along the basal-to-ciliated trajectory. Polidocanol-induced injury. Polidocanol-induced injury was performed as previously described 38 . Briefly, mice were anesthetized and delivered one dose of 15 μL 2% Polidacanol or PBS vehicle control by oralpharyngeal aspiration to induce injury. Tracheas were harvested at 1 day (1d), 2d, 3d and 7d following injury for scRNA-seq or for fixation and immunofluorescence. Immunofluorescence, microscopy and cell counting. For RNAscope® and immunofluorescence of paraffin embedded sections, mouse tracheas were dissected under sterile conditions and HBEC transwell cultures were isolated using 8mm biopsy punch (Integra Miltex, 33–37). Primary human bronchial tissue was obtained through the International Institute for the Advancement of Medicine. All tissues were immediately fixed in 10% Normal buffered formalin for 18–24 hours at room temperature (RT) then transferred to PBS and kept at 4°C until paraffin embedding. For immunofluorescence of mouse tracheas, 5μm sections were baked and deparaffinized using standard procedures. After antigen retrieval using pH6 Citrate buffer (Abcam), sections were rinsed in PBS and blocked in 10% normal goat serum (NGS) or 10% normal donkey serum (NDS) for 30 min at RT. Primary antibody was added overnight at 4°C. Sections were washed 3X in PBS for 5 min each, and secondary antibody was added for 1h at RT and sections were again rinsed in PBS, followed by Hoechst (1:1000) for 30 sec. For RNAscope®, 5 μm sections were prepared according to RNAscope® procedures for multiplex fluorescent assay (Advanced Cell Diagnostics, 320850) or dual chromogenic assay (322430). RNAscope® probes used were FOXI1 (476351), CFTR (603291), and FOXJ1 (430921). Mounting medium and coverslip were applied and slides were stored at 4°C for immunofluorescence or RT for chromogenic ISH. For immunofluorescence of whole mount HBEC transwell cultures, cells were fixed in 4% paraformaldehyde for 30 min RT, washed 3X 10 min in IF buffer (130 mM NaCl, 7 mM Na2HPO4, 3.5 mM NaH2PO4, 7.7 mM NaN3, 0.1% bovine serum albumin, 0.2% Triton X-100, and 0.05% Tween- 20), blocked in 10% NGS IF buffer, stained in primary antibody diluted in 10% NGS IF buffer overnight at 4°C, washed 3X 20 min in IF buffer, counterstained in secondary antibody diluted in 10%NGS IF buffer plus 1:5000 Hoechst for 1h RT, washed 3X 20 min in IF buffer, and washed 2X in PBS before mounting. The following antibodies were used: rabbit anti-FOXI1 (1:200, Sigma-Aldrich HPA071469), mouse anti-FOXI1 (1:100, Origene TA800146), goat anti-FOXI1 (1:200 Abcam ab20454) rabbit anti-ATP6V1B1 (1:100, Sigma-Aldrich HPA031847), mouse anti-acetylated αtubulin (1:1000 Sigma-Aldrich T6793), rabbit anti-Scgb1a1 (1:200, Millipore 07–623), rabbit anti-MUC5B (Santa Cruz sc-20119), rabbit anti-FOXJ1 (1:200, Sigma-Aldrich HPA005714), mouse anti-FOXJ1 (EBioscience, 1:200, 14–9965-80), rabbit anti-FOXN4 (Sigma-Aldrich HPA050018), mouse anti-ASCL1 (1:00, Beckton-Dickinson 556604), mouse anti-Krt4 (1:100, abcam ab9004), rabbit anti-Krt4 (1:100 Proteintech 16572–1-AP), rabbit anti-Krt5 (1:250, abcam ab52635), chicken anti-Krt5 (1:1000, BioLegend 9059), mouse anti-NGFR (1:200, ThermoFisher, MA1–18418) and chicken anti-Krt8 (1:200, abcam ab107115). Secondary antibodies used were Alexa Fluor 488, 568, 647 (Life Technologies) at 1:500. Fluorescent images were collected on a confocal microscope (Axiovert 200; Carl Zeiss), with a 40X objective (Zeiss, Plan-Apochromat 40X/1.3 Ph3 M27), a Yokogawa CSU-X1 spinning disc head, and an electron-multiplying charge-coupled device camera (Evolve 512; Photometrics). Scale bars were added, and images were processed using Zen Blue software (Zeiss) and Photoshop (Adobe). FOXI1+ and FOXJ1+ cells were counted using ImageJ software. Chromogenic signals were acquired using a Nuance™ FX multispectral imaging system (PerkinElmer) with an Olympus BX61 microscope interfaced with a liquid crystal based camera and tunable filter from 420 nm to 720 nm at 20 nm intervals. Spectral components were unmixed and pseudo-colored for individual channels. Lentivirus production. For overexpression, FOXI1 (GeneID 2299) was cloned into the plenti6/V5-DEST.NGFP Gateway® vector, which was generated by transferring the N-EmGFP ORF from pcDNA6.2/N-EmGFP-DEST (ThermoFisher, Cat# V35620) into pLenti6/V5-DEST (ThermoFisher, V49610). Lentiviral packaging 4 × 106 293T cells were seeded in a 100 mm Poly-D-Lysine coated dish (Corning® BioCoat™, 356469) one day before transfection with 14 ml of cell growth medium (DMEM (ThermoFisher, Cat# 11965092), 10% FBS (Clontech 631106), 2mM L-Glutamine (Invitrogen 25030), 0.1mM MEME Non Essential Amino Acids (Invitrogen 11140), and 1mM Sodium Pyruvate MEM (Invitorgen 11360)). For transfection, 7 μg of packaging plasmid DNA (ViraPower lentiviral Packaging Mix, ThermoFisher K497500) was mixed with 5 μg of expression construct DNA and 36 μl Fugene6 (Promega, E2691). OptiMEM (ThermoFisher, 31985062) was then added the mixture to a total volume of 800 μl. 293T cells were included with the transfection reagent mixture for 24 hours before the growth medium was refreshed. At 72 hours after transfection, virus was harvested, and frozen for future experiments. Packaged virus was added to HBEC cultures 1 hour after cell seeding and then removed at feeding the following day. Flow cytometry and cell sorting. Cells were harvested using 0.05% Trypsin-EDTA (ThermoFisher, 25300054), pelleted at 300g for 5 min, suspended in 2% FBS DMEM with EDTA and filtered through a 40μm strainer before being analyzed by flow cytometry or cell sorting. RNA was extracted with Trizol (Invitrogen, 15596026). cDNA was synthesized from 1 ug of RNA with qScript XLT cDNA Super Mix kit (Quanta Biosciences, 95161–100). qPCR was carried out using FastStart Universal Probe Master kit (Roche, 04914058001) with 40 ng of cDNA per reaction. Taqman probes for qPCR (Applied Biosystems) are shown below: FOXI1, Hs00201827_m1 FOXJ1, Hs00230964_m1; P63, Hs00978340_m1; GAPDH, Hs99999905_m1; CFTR, Hs00357011_m1; ATP6V1B1, Hs00266092_m1; ITGA6, Hs01041011_m1; DNAI2, Hs01001544_m1; SCGB1A1, Hs00171092; MUC5B, Hs00861588_m1; NRARP, Hs01104102_s1; HES5, Hs01387464_g1; HES1, Hs00172878_m1; MUC5AC, Hs01365601_m1 Short-circuit current (Isc) measurements in Ussing Chambers. For Ussing studies, HBECs were cultured in 6-well Snapwell® plates (Corning, 3801) at a density of 83,000 cells/Snapwell®. Snapwell inserts containing differentiated HBECs were then mounted in chambers bathed in Buffer (Kreb’s Ringer Solution; 400 mL H2O, 25 mL 2.4M NaCl, 25 mL 0.5M NaHCO3, 25 mL 66.6M KH2PO4 + 16.6 mM K2HPO4, 25 mL 24 mM CaCl2 + 24 mM MgCl2, .9g Dextrose). Amiloride (Sigma, A9561) was added apically at 10μM to inhibit Na+ absorption, then Forskolin (Sigma, F6886) was added apically at 20μM to stimulate cAMP and finally, CFTR-172 (Sigma-Aldrich, C2992) inhibitor was added apically and basally at 30μM. Under these conditions, cAMP-stimulated Isc due to addition of Forskolin could be attributed to CFTR-mediated Cl- secretion from basolateral to apical solution. Statistical analysis. The standard error of the mean was calculated from the average of at least three independent HBEC cultures. The Student’s t-test (unpaired, two-tailed) was used to compare data between groups with a p-value of less than 0.05 considered significant. Pearson correlation and its associated p-value between ΔIsc and FOXI1+ or FOXJ1+ cell number/mm 2 was calculated using the MATLAB corr function. Multivariate regression was carried out using the MATLAB fitlm function. Sensitivity was defined as the fractional change in Isc induced by a fractional change in FOXI1 (x_1) or FOXJ1 (x_2) cell number/mm 2 , at the ΔIsc value across all samples, estimated from the slope and intercept of multivariate regression as S_i = (dIsc/dn)/(Isc/n) = [slope_i* ]/ with I=1,2, respectively for FOXI1, FOXJ1. Code availability. Python scripts implementing the methods as described can be obtained upon request. Data availability. All sequencing data are available in the Gene Ontology Omnibus repository under the accession number GSE102580, the NCBI Sequence Read Archive under the accession number SRR5881096, the Klein lab SPRING viewer, and the Single Cell Portal (https://portals.broadinstitute.org/single_cell). To explore the single cell data: Klein lab SPRING viewer (we recommend using Google Chrome): https://kleintools.hms.harvard.edu/tools/springViewer_1_6_dev.html?datasets/uninjured_MTECs/uninjured_MTECs https://kleintools.hms.harvard.edu/tools/springViewer_1_6_dev.html?datasets/all_MTECs/all_MTECs https://kleintools.hms.harvard.edu/tools/springViewer_1_6_dev.html?datasets/reference_HBECs/reference_HBECs https://kleintools.hms.harvard.edu/tools/springViewer_1_6_dev.html?datasets/GFP_GFPFOXI1_HBECs/GFP_GFPFOXI1_HBECs Single Cell Portal: Go to https://portals.broadinstitute.org/single_cell Log in with the following Google credentials: scp.wingert@gmail.com h7J-cD5-fpG-kLX View study: “A single cell atlas of the conducting airway reveals the CFTR-rich pulmonary ionocyte” Go to “explore” Explore. Under “Load a cluster” you can switch between the 3 SPRING plots used in the paper. NOTE: when you start typing a gene, it will autofill. Make sure to select mouse genes for mouse and human genes (all capitals) for human. Failing to do so will display a given gene as not expressed. Extended Data Extended Data Figure 1: Atlas of transcription factors, surface molecules and kinases enriched in proximal airway lineages of mouse and human. Transcription factors, kinases and surface molecules in mouse (a) and human (b) identified among the list of cell type-specific genes that met the following criteria: significantly enriched in lineage (false discovery rate (FDR) 0.2. Gene-gene correlation heat map shows 4 gene modules in mouse airway basal cells (a) and 6 gene modules in mouse airway secretory cells (b). SPRING plots show where gene modules are expressed in a given population. Multiple genes are combined in a single signature defined as the mean rank of expression (dense ranking). Extended Data Figure 3: Gene modules identified in human bronchial lineages. Two major modules of anti-correlated genes were identified by selecting variable genes within the basal to secretory continuum that were correlated with at least 4 other genes with rank correlation > 0.12. Genes within each module were then separately considered within basal and secretory cells, keeping genes with a correlation > 0.35 with at least 4 other genes. Gene-gene correlation heat map shows 3 gene modules in human airway basal cells (a) and 4 gene modules in human airway secretory cells (b). SPRING plots show where gene modules are active in a given population. Multiple genes are combined in a single signature defined as the mean rank of expression (dense ranking). Extended Data Figure 4: Validation of novel lineages in mouse and human by immunofluorescence. a, Immunofluorescence in mouse tracheal epithelium for Krt4 (green, arrowheads), Krt5 (basal), Krt8 (luminal), Scgb1a1 (club, secretory) and Foxj1 (ciliated) (n=3 animals). b, Immunofluorescence in differentiated HBEC cultures for FOXN4 (red, arrows), FOXJ1 (arrowheads mark FOXJ1low cells) and Acetylated αTubulin (cilia) (n=2 donors). c, Immunofluorescence in HBEC cultures for the ionocyte markers FOXI1, ATP6V1B1 and NGFR (n=3 donors). Arrowhead shows apical enrichment of ATP6V1B1. Arrows highlight lateral protrusions. Scale bar, 20 μm. Extended Data Figure 5: Identification of recovery-specific cell states and population dynamics during regeneration. a, Cells from uninjured mouse airway do not equally populate all regions of the SPRING plot of all mouse data combined. Each cell from the uninjured condition voted for its 10 nearest neighbors among all mouse cells profiled, and smoothed vote counts are used as a proxy for uninjured cell density on the map (two leftmost plots). By visual inspection of the smooth vote distribution a threshold of 25 votes was chosen to binarize regions of the SPRING plot into present vs depleted in uninjured. b, Barcharts representing abundance of rare populations as a fraction of all cells, over time post-injury. Error bars represent the 95% binomial proportion confidence interval (normal approximation). Total number cells = 7,898 from n=4 mice (uninjured), 898 from n=1 mouse (24h), 1,964 from n=1 mouse (48h), 1,082 from n=1 mouse (72h) and 2,321 from n=4 mice (1 week). c, Barcharts showing the fraction of all cells that are express Foxi1 in each population during recovery. Values shown correspond to fraction of all cells at each time point (cell and mouse numbers as in b above). Error bars defined as in b. Extended Data Figure 6: Analysis of basal to ciliated differentiation trajectory following injury. a, Population Balance Analysis (PBA, see Methods) was used to order 609 cells highlighted in black along the pseudotime of their basal-to-ciliated progression, followed by application of a moving average over a window of 100 cells. The resulting ordering of averaged cells is referred to as the basal-to-ciliated trajectory. PBA requires manually selecting source and sink cells for calculating the pseudotime. b, Heat map of the 1237 genes differentially expressed genes along the basal-to-ciliated trajectory (permutation test, FDR 50 TPM in at least on cluster were considered. The bottom of the heat map shows the top 20 enriched genes identified treating all four transduction-specific states as one population. e, Bar chart showing fold-changes in population size following GFP:FOXI1 vs GFP transduction (extension of Fig. 3d). f, Expression of transgene in identified cell populations. Extended Data Figure 9: Notch pathway component enrichment in airway lineages. SPRING plots show enrichment of Notch pathway components in mouse (a) and human (b) airway lineages. Normalized counts are shown for the Notch ligands JAG1, JAG2 and DLL1 and the Notch receptors NOTCH1, NOTCH2 and NOTCH3. The Notch target gene signature combines HES1, HES5 and NRARP into a single gene signature defined as the mean expression rank (dense ranking). All gene expression and signature values are smoothed (see Methods for smoothing). Extended Data Figure 10: Notch signaling inhibition decreases ionocyte markers in HBECs. a, Expression of Notch target genes and airway lineage markers in cultures treated with 3.3 μM DAPT compared to cultures treated with DMSO. Notch target genes (NRARP p=.03, HES5) and secretory cell markers (MUC5B p=.001, MUC5AC) are decreased while ciliated cell markers (FOXJ1, DNAI2 p=.01) and basal cell markers (ITGA6 p=.006 and TP63) are increased upon DAPT treatment. Note that ionocyte markers (FOXI1 p=.02, CFTR) are also decreased upon DAPT treatment. P-value determined by two-tailed t-test (n=8 experiments in 2 donors). b, FOXI1 cell counts in HBEC cultures treated with antibodies that neutralize individual NOTCH receptors (n=5 experiments in 2 donors). All data are mean ± SEM. Supplementary Material 1 2 3 4 5
              Bookmark
              • Record: found
              • Abstract: found
              • Article: not found

              Pharmacological Rescue of Conditionally Reprogrammed Cystic Fibrosis Bronchial Epithelial Cells.

              Well-differentiated primary human bronchial epithelial (HBE) cell cultures are vital for cystic fibrosis (CF) research, particularly for the development of cystic fibrosis transmembrane conductance regulator (CFTR) modulator drugs. Culturing of epithelial cells with irradiated 3T3 fibroblast feeder cells plus the RhoA kinase inhibitor Y-27632 (Y), termed conditionally reprogrammed cell (CRC) technology, enhances cell growth and lifespan while preserving cell-of-origin functionality. We initially determined the electrophysiological and morphological characteristics of conventional versus CRC-expanded non-CF HBE cells. On the basis of these findings, we then created six CF cell CRC populations, three from sequentially obtained CF lungs and three from F508 del homozygous donors previously obtained and cryopreserved using conventional culture methods. Growth curves were plotted, and cells were subcultured, without irradiated feeders plus Y, into air-liquid interface conditions in nonproprietary and proprietary Ultroser G-containing media and were allowed to differentiate. Ussing chamber studies were performed after treatment of F508 del homozygous CF cells with the CFTR modulator VX-809. Bronchial epithelial cells grew exponentially in feeders plus Y, dramatically surpassing the numbers of conventionally grown cells. Passage 5 and 10 CRC HBE cells formed confluent mucociliary air-liquid interface cultures. There were differences in cell morphology and current magnitude as a function of extended passage, but the effect of VX-809 in increasing CFTR function was significant in CRC-expanded F508 del HBE cells. Thus, CRC technology expands the supply of functional primary CF HBE cells for testing CFTR modulators in Ussing chambers.
                Bookmark

                Author and article information

                Journal
                Am J Respir Cell Mol Biol
                Am J Respir Cell Mol Biol
                ajrcmb
                American Journal of Respiratory Cell and Molecular Biology
                American Thoracic Society
                1044-1549
                1535-4989
                14 June 2023
                1 September 2023
                14 June 2023
                : 69
                : 3
                : 250-252
                Affiliations
                [ 1 ]Department of Internal Medicine

                Division of Pulmonary and Critical Care
                [ 2 ]Center for Regenerative Medicine

                Massachusetts General Hospital

                Boston, Massachusetts
                [ 3 ]Klarman Cell Observatory

                Broad Institute

                Cambridge, Massachusetts
                [ 4 ]Harvard Stem Cell Institute

                Cambridge, Massachusetts
                Author information
                https://orcid.org/0000-0002-3605-8066
                Article
                2023-0169ED
                10.1165/rcmb.2023-0169ED
                10503302
                37315655
                8ebbbb96-c634-445c-9b4f-fcaedd8933fd
                Copyright © 2023 by the American Thoracic Society

                This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0. For commercial usage and reprints, please e-mail dgern@ 123456thoracic.org .

                History
                Page count
                Figures: 0, Tables: 0, References: 10, Pages: 3
                Funding
                Funded by: Cystic Fibrosis Foundation, doi 10.13039/100000897;
                Award ID: CFF003338L121
                Funded by: National Institutes of Health, doi 10.13039/100000002;
                Award ID: R01HL118185-08
                Award ID: R01HL142559-04
                Award ID: R01HL157221-01
                Categories
                Editorials

                Comments

                Comment on this article