Neuroendocrine cells orchestrate regeneration through Desert hedgehog signaling

SUMMARY

Understanding the mechanisms underlying mammalian regeneration may enable development of novel regenerative therapies. We present a mechanism wherein Desert hedgehog (Dhh), secreted from epithelial neuroendocrine cells, elicits a regenerative/protective response from mesenchymal cells. In mammalian airway, this mesenchymal response strikingly amplifies the initial signal from rare neuroendocrine cells to activate the entire tissue for survival and regeneration upon injury from SO₂ gas inhalation or following influenza or SARS-CoV-2 infection. Similar epithelial-mesenchymal feedback (EMF) signaling directed by Dhh from neuroendocrine β-cells likewise protects mouse pancreatic islets from streptozotocin injury. A role for EMF signaling in human pancreatic islets is suggested by higher incidence of diabetes in patients treated with Hedgehog pathway inhibitors. Remarkably, EMF augmentation by small molecule Hedgehog pathway agonism protects against streptozotocin injury of pancreatic β-cells and shields against airway injury from SO₂ and influenza infection, with potential protective/therapeutic utility in chemical or infectious airway injury and in diabetes.

In brief (eTOC)

Upon chemical or viral injury of airway or pancreatic islets, Desert hedgehog signaling from epithelial neuroendocrine cells elicits regenerative and protective feedback responses from neighboring mesenchymal cells thus having therapeutic implications for mammalian regeneration in airway, islets and other organs.

Graphical Abstract

INTRODUCTION

An understanding of the intrinsic mechanisms underlying adult mammalian tissue repair and maintenance may suggest novel therapeutic solutions to the challenges of regenerative medicine¹. Despite this potential therapeutic value, however, our understanding of these mechanisms does not yet encompass general molecular principles like those known to underlie embryonic patterning, such as morphogenic signaling gradients^2,3 and spatially patterned transcription factor cascades⁴. To help establish a more general understanding of adult tissue regeneration we explore here a surprising similarity in molecular and cellular mechanisms underlying injury-induced regenerative responses of the mammalian airway^5,6 and of the endocrine organs, or islets, of the pancreas⁷.

The mammalian respiratory system supplies oxygen for the entire organism by employing millions of gas exchange units, termed alveoli. The thin-walled structure of these delicate air sacs ensures a minimal barrier that facilitates passive gas exchange but also renders them vulnerable to injury from inhaled particulates, toxins, and pathogens. To minimize such exposure, the airway tree is tasked with the protective functions of trapping and clearing injurious agents before they reach more distal alveoli. Should such agents injure the airway itself, the airway epithelium is able to activate a vigorous regenerative program. Although recent advances have illuminated the roles of diverse cell types in mammalian airway function^8,9, the molecular mechanisms that preserve airway integrity remain largely unknown.

To investigate these mechanisms, we focus here on tracheal airway response to injury caused by inhalation of SO₂ gas, a major toxic component in gases originating from volcanic eruptions, fossil fuel combustion, and other natural and anthropogenic sources^10–12. We have discovered a critical regulatory role for signaling initiated by expression of the Desert hedgehog (DHH) member of the Hedgehog family of signaling proteins within solitary pulmonary neuroendocrine cells (PNECs) of the proximal airway, comprising fewer than 1% of total cells within the tracheal epithelium¹³. In a striking response to and amplification of this limited initial signal, upon injury the entire airway is activated by DHH-dependent expression of IL-6 from a subset of mesenchymal cells. This epithelial-mesenchymal feedback (EMF) signaling promotes airway integrity by protecting epithelial cells from apoptosis and promoting their proliferation. EMF signaling is also protective against influenza and SARS-CoV-2 infection. Dhh expression in the trachea is restricted to neuroendocrine cells, and this unique specificity is preserved in the pancreas, in which distinct types of neuroendocrine cells also comprise a small proportion (1–2%) of the total cell mass, but are easily identified within prominent clusters, the islets of Langerhans^14,15. We find that similar EMF signaling operates within pancreatic islets, with DHH expression from insulin-producing β-cells eliciting a protective/regenerative response from islet mesenchymal cells. This response shields β-cells from injurious effects of the cytotoxic agent, streptozotocin (STZ). Of potential therapeutic significance, we find that pharmacologic augmentation of EMF signaling with a small molecule Hedgehog pathway agonist is protective in both tissues, and the human relevance of these observations is highlighted by an increased incidence of diabetes in patients treated with Hedgehog pathway inhibitors.

RESULTS

Hedgehog signaling in the tracheal airway

The molecular mechanisms that regulate airway homeostasis or regeneration after injury are poorly understood. In other endodermal organs, such as the urinary bladder, epithelial cells produce Hedgehog protein signals which trigger a mesenchymal feedback response that supports the homeostatic and regenerative activity of the epithelium^16,17. Despite little current evidence of epithelial or mesenchymal cell expression of Hedgehog transcripts in trachea of the adult mouse^8,9,18–20, we detected low transcript levels for Desert hedgehog (Dhh) and key components of Hedgehog signaling in bulk RNA-seq analysis of the tracheal transcriptome, whereas RNAs encoding the IHH (Indian hedgehog) or SHH (Sonic hedgehog) members of the Hedgehog family were essentially absent (Figure 1A; S1A,B; Table S1). Tamoxifen induction of reporter mice (Gli1Cre^ERT2;R26^Ai6/+) revealed that ~20% of tracheal mesenchymal cells express Gli1 (Figure 1B; S1C), a zinc finger transcription factor that is both a target and a mediator of Hedgehog signaling^21,22. This mesenchymal expression of Gli1 was markedly reduced in Dhh^−/− homozygotes (fully viable²³; Figure 1C,D).

Figure 1. Desert hedgehog signaling via GLI1 in SO₂ inhalation injury.

(A) Expression of Desert hedgehog (Dhh), Sonic hedgehog (Shh), and Indian hedgehog (Ihh). Bulk RNAseq shows that Dhh is the predominant Hedgehog ligand expressed in trachea (FPKM, fragments per kilobase of exon per million mapped fragments). RNA samples pooled from two male and two female mice. (B) ZsGreen expression in tamoxifen-induced Gli1^CreERT2/+;R26^Ai6 reporter mice marks a subset of tracheal mesenchymal cells (Gli1, orange). Ciliated cells were visualized by immunostaining for AcTUB (acetylated Tubulin), and nuclei by DAPI staining (all figures). Scale bar=20μm. (C) β-galactosidase activity (X-gal staining) shows reduced Gli1 expression in mesenchyme underlying the epithelium of Dhh^−/− homozygotes (comparison of Dhh^−/−;Gli1^nLacZ/+ and Dhh^+/+;Gli1^nLacZ/+ mice). Scale bar=25μm. (D) Average number of cells in C stained by X-Gal within 30μm from the epithelium (n=4). (E) Experimental scheme (upper left). Representative tracheal sections from mice of the indicated Dhh genotypes, uninjured (Ctrl) or 24 hours post-injury (hpi). Secretory cells, and basal cells were visualized by immunostaining for Secretoglobin Family 1A Member 1 (SCGB1A1), and KRT5 (Keratin 5; all figures), respectively. Scale bars=20μm. (F) Average number of ciliated, secretory, and basal cells in E (n=6). (G) Representative tracheal sections from mice of indicated genotypes, uninjured (Ctrl) or 24 hpi. Scale bars=20μm. (H) Average number of ciliated, secretory, and basal cells in G (n=6). (I) Schematic summaries of injury effects in mice of genotypes indicated, Ctrl or 24 hpi. In all figures “n” is number of biological replicates; P values ≤0.05, ≤0.01, and ≤0.001are indicated by *, **, ***, and were calculated with a two-tailed t-test unless otherwise specified; “ns” is not significant or no indication; error bars indicate standard deviation; “Epi” denotes epithelium and “Mes” denotes mesenchyme. See also Figure S1 and Table S1.

Desert hedgehog signaling maintains epithelial integrity

We interrogated tracheal regeneration by subjecting mice to a standard three-hour period of SO₂ gas inhalation (500 ppm; schematic diagram, Figure 1E), which non-specifically damages epithelial cells of the proximal airway in a variety of experimental animals^24,25. This injury has been used in lineage tracing experiments to identify secretory and basal cells in mouse tracheal epithelium as stem/progenitor cells ^26,27. In wild-type mice ~85% of ciliated cells (AcTUB) and ~41% of secretory cells (SCGB1A1) were lost by 24 hours post-injury (hpi) (Figure 1E,F), with accompanying proliferation of basal cells (KRT5) (see below). By contrast, Dhh^−/− homozygous mutants lost ~96% of ciliated cells and ~88% of secretory cells 24 hpi and basal cells failed to proliferate (Figure 1E,F).

Although Gli1 loss only partially reduces Hedgehog pathway activation and is compatible with normal development and reproduction^28,29 Gli1^−/− homozygous mice essentially phenocopied the abnormal injury response of Dhh^−/− homozygous mutants, whereas Gli1^+/− heterozygous littermates were similar to Dhh^+/+ (wild-type) mice in severity of injury (Figure 1G,H; schematic summary in Figure 1I).

Expression of Desert hedgehog in pulmonary neuroendocrine cells

We introduced the CreERT2 allele as an in-frame translational fusion to the Dhh ATG start codon (Dhh^CreERT2; Figure 2A; S2A,B,C) for tamoxifen-inducible Cre recombinase to mark and identify cells expressing Dhh. In marked contrast to Shh^{CreERT2 30} or Ihh^{CreERT2 31} drivers, the Dhh^CreERT2 driver combined with the Rosa26^m™G bi-fluorescent protein reporter³² generated GFP-expressing cells (Figure S2D). With the Rosa26^Ai6 reporter, which readily undergoes Cre-mediated recombination to express ZsGreen³³, ~1% of cells in the tracheal epithelium were marked (Figure 2B,C,D), a low percentage as compared to basal cells expressing either KRT5 or KRT14 (Figure S2E) but consistent with our bulk RNA-Seq data (Figure S1A; and Table S1). Some of these cells displayed wedge- or flask-shaped morphologies, characteristically associated with solitary pulmonary neuroendocrine cells (PNECs)^13,34,35. These cells communicate with the basal surface of the tracheal epithelium, or with both basal and luminal surfaces (Figure 2C,D), corresponding to ‘open’ or ‘closed’ PNECs, respectively³⁶.

Figure 2. Dhh expression in neuroendocrine cells of the airway epithelium.

(A) A cassette for bi-cistronic expression of CreERT2 and eGFP inserted in-frame following the ATG translation start codon of the Dhh gene by homologous recombination. IRES, internal ribosome entry sequence; WPRE, woodchuck hepatitis virus posttranscriptional regulatory element. (B) Experimental scheme to visualize Dhh expression (C-D) by tamoxifen-induced ZsGreen expression in Dhh^CreERT2/+;R26^Ai6 reporter mice. (E) Dhh expressing cells co-localize with pulmonary neuroendocrine cell (PNEC) markers Calcitonin gene-related peptide (CGRP, red), and Synaptophysin (SYP, red) in both tracheal and intra-pulmonary airway epithelium. Outlined cells display the distinctive wedge- or flask-shaped appearance of solitary tracheal cells or of cells within intra-pulmonary neuroepithelial bodies (NEBs). (F) Dhh expression in intra-pulmonary airway. Scale bars in C=10μm, D=50μm, insets=20μm, E=10μm, F=50μm. (G) Percentage of Dhh-expressing cells that co-express CGRP or SYP within trachea, or within NEBs in lung (n=10). Most NEBs (>80%) contain Dhh-expressing cells; some intra-pulmonary Dhh-expressing cells are not within NEBs. See also Figure S2.

All or most cells expressing the neuroendocrine cell marker proteins Synaptophysin (SYP) and Calcitonin Gene-Related Polypeptide (CGRP) in trachea are marked by ZsGreen in tamoxifen-treated Dhh^CreERT2;Rosa26^Ai6 mice, although only 67% and 16% of ZsGreen-marked cells expressed SYP or CGRP, respectively (Figure 2E; tabulated in 2G). This pattern is consistent with universal Dhh expression in different classes of solitary tracheal PNECs of the airway epithelium (EPCAM⁺), but not in endothelial (CD31⁺) or hematopoietic (CD45⁺) cells (Figure S2F). Dhh-marked cells in intrapulmonary airway epithelium in contrast show the distinctive locations, morphologies, and protein expression of PNECs clustered in neuroepithelial bodies (NEBs), the characteristic organization of intrapulmonary neuroendocrine cells^13,37 (Figure 2E,F,G). Although over 80% of NEBs contained Dhh-marked cells, only a subset of these clustered neuroendocrine cells were marked, ranging from ~10–90% (Figure 2E,F,G).

Neither the level of Dhh gene expression, nor the numbers of Dhh-expressing or CGRP- or SYP-expressing cells was significantly altered by injury (Figure S2G,H,I,J); nor was expression of Shh or Ihh induced by injury (Figure S2D,K). Neither the proliferative status of mesenchymal Gli1-expressing cells (Figure S2G,H) nor the overall expression levels of Gli1, Gli2, and Gli3 (Figure S2K) were significantly altered upon injury.

Epithelial DHH activates GLI1-mediated transcriptional activation of IL-6

RNA-Seq analysis (Table S1) and quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) of mice exposed to SO₂ identified cytokine IL-6 (Interleukin-6) and no other cytokine or growth factor genes as significantly induced at 6, 9, 12, 18 and 24 hpi in a Gli1- and Dhh-dependent manner (Figure 3A,B; S3A,B; Table S1). IL-6^−/− homozygous mutant mice phenocopied the severe epithelial damage noted in Dhh^−/− and Gli1^−/− mice at 24 hpi (compare Figure 1E,F,G,H to Figure 3C,D) and, consistently, IL-6 expression upon injury was reduced in Dhh^−/− and Gli1^−/− mutant mice (Figure 3B).

Figure 3. Neuroendocrine epithelial-mesenchymal feedback (EMF) signaling triggers a broad tissue response.

(A) Experimental scheme, panels B-G. (B) Tracheal IL-6 RNA (RT-qPCR) normalized to Hprt, from indicated genotypes, uninjured or 24 hpi (n=6). (C) Representative tracheal sections from mice of indicated genotypes. Scale bar=20μm. (D) Average number of ciliated, secretory, and basal cells in C (n=3). (E) RNA in situ hybridization for IL-6 receptor alpha (IL-6ra, yellow puncta; dapB, bacterial probe) in tracheal sections from uninjured wild-type mice. Scale bar=25μm. (F) Immunostaining for phospho-STAT3 (pSTAT3, red) in tracheal sections from mice of indicated genotypes, uninjured or 6 hpi. Scale bar=30μm. (G) Average number of pSTAT3⁺ epithelial cells in E (n=3). (H) Experimental scheme for tamoxifen induction of GLI1-FLAG. (I) ChIP-PCR with anti-FLAG, from tracheal cells of mice in H, showing injury-dependent GLI1-FLAG binding to the IL-6 promoter (n=2). (J) Schematic of neuroendocrine epithelial-mesenchymal feedback (EMF) signaling in airway (see text). See also Figure S3.

Phosphorylation of the STAT3 transcription factor, a downstream event dependent on IL-6 signaling³⁸, was detected in the majority of epithelial and mesenchymal cells upon injury of Gli1^+/− but not in Gli1^−/− homozygotes, consistent with expression of the IL-6 receptor in most epithelial and mesenchymal cells (Figure 3E,F,G). Expression of IL-6 initially is restricted to Gli1-expressing cells 3h after initiation of injury, but then expands to include non-Gli1-expressing cells present in mesenchyme and epithelium (Figure S3C), suggesting injury-dependent recruitment of additional IL-6-expressing cells.

To address the question the question of how injury enables GLI1 augmentation of IL-6 expression without increasing Gli1 expression, we assessed GLI1 binding at the IL-6 promoter by ChIP-PCR³⁹ in Gli1^CreERT2/+;Rosa26^{Gli1-FLAG-YFP/+} mice, in which tamoxifen treatment activates expression of FLAG-tagged GLI1 in cells that express wild-type Gli1. These mice retained normal injury-dependent expression of IL-6 (Figure S3E), and use of the highly specific anti-FLAG antibody to precipitate GLI1-FLAG-bound chromatin⁴⁰ revealed that GLI1-FLAG is associated with IL-6 promoter DNA only upon injury (Figure 3H,I). This injury-induced access and binding of GLI1 to the IL-6 promoter may be enabled by chromatin remodeling proteins and other IL-6-regulating transcription factors that are induced by injury (Figure S3F,G).

Amplification of epithelial-mesenchymal feedback (EMF) signaling in airway injury response

We ablated Dhh-expressing cells in tamoxifen-treated adult mice of the genotype Dhh^CreERT2/+;Rosa26^DTR/DTR either through systemic administration of diphteria toxin (DT), or with local epithelial ablation by intranasal instillation of DT into the airway. Both methods produced an impaired injury response that phenocopies that of Dhh^−/− mutant mice (Figure S3H,I). Tamoxifen-treated mice of the genotype Gli1^CreERT2/+;Smo^flox/flox, in which the essential Hh response component Smoothened (Smo) is ablated in mesenchymal cells, also phenocopied the impaired injury response of Dhh^−/− mutant mice (Figure S3H,I), as did intranasal instillation of an IL-6 receptor-inactivating monoclonal antibody (Figure S3H,I).

In these experiments, cell ablation by topical administration of DT indicates that injury response requires neuroendocrine cell production of DHH in epithelium, and Gli1^CreERT2-driven ablation of Smo indicates that response to this epithelial signal occurs in cells of the mesenchyme. The topically-administered IL-6-receptor-blocking antibody in turn demonstrates a requirement for IL-6 response in the epithelium, and this reciprocal exchange between epithelial and mesenchymal compartments establishes the essential role of an epithelial-mesenchymal feedback (EMF) signaling cascade in airway response to inhalational injury (Figure 3J). These experimental manipulations also indicate that EMF signaling initiated by DHH (not SHH or IHH; Figure 1A; S2D,K) is required in adults, at the time of injury.

Our analysis reveals a rapid expansion of EMF signaling from <1% of epithelial cells for Dhh expression (Figure 2C,D,E; S2D,E,F), to induction of Gli1 expression in ~20% of mesenchymal cells (Figure 1B,C; S1C). GLI1-activated IL-6 expression (Figure 3A,B; Figure S3C,D) then expands to non-Gli1-expressing cells (Figure S3C), ultimately producing sufficient IL-6 to cause phosphorylation and activation of STAT3 throughout most cells of the tracheal mesenchyme and epithelium. This dramatic expansion occurs within six hours following injury (Figure 3F).

EMF signaling directs regeneration

Neither Ki-67 expression (cell proliferation) nor TUNEL staining (cell death) is significantly detected in uninjured trachea, but substantial expression of Ki-67 occurs in injured trachea of Gli1^+/− mice by 6 hpi, predominantly in KRT5⁺ basal cells (Figure 4B,D; S4A,B,C,D). TUNEL staining in Gli1^+/− trachea emerges by 9 hpi and peaks at 12 hpi, most prominently in ciliated and secretory cells, and declines thereafter (Figure 4C,D; S4E,F,G). Ki-67 expression in Gli1^−/− trachea in contrast is barely detectable, even by 24 hpi (Figure 4B,D; S4A,C,D). Extensive TUNEL staining, however, emerges and peaks at 6 hpi (Figure 4C,D; S4E,G), and Dhh^−/− mutant mice display similar effects (Figure S4H,I,J,K,L). Epithelial expression of BCL-2, protective against apoptosis, is much higher at 18 hpi in epithelium of Gli1^+/− as compared to Gli1^−/− mice (Figure 4E,F,G).

Figure 4. EMF signaling directs cell survival, proliferation, and regeneration of injured epithelium.

(A) Experimental scheme for panels B,C,D. (B) SO₂ -induced proliferation (Ki-67, red), and (C) apoptosis (TUNEL staining, red), in tracheal sections from mice of indicated genotypes, uninjured or hpi as indicated. Scale bars=30μm. (D) Proliferation index (%Ki-67⁺, B) and apoptosis (TUNEL⁺ cells/field, C) (n=3). (E) Experimental scheme to evaluate epithelial regeneration. (F,G) Epithelial BCL2⁺ cells (red) in tracheal sections from mice of indicated genotypes, either uninjured or 18 hpi (n=3). Scale bar=30μm. (H) Tracheal sections from wild-type, Gli1^−/− and Dhh^−/− mice at 14 days post-injury (dpi). Scale bar=20μm. (I,J) Number of epithelial cells (all types, I; ciliated cells, J) from H (see also S4M) in tracheal sections from the indicated genotypes, either uninjured or 3, 7, and 14 dpi (n=3). (K,L) Schematic summaries of normal and abnormal epithelial regeneration (K) and the role of EMF signaling in restoring epithelial integrity (L). See also Figure S4.

Augmentation of epithelial injury caused by loss of EMF signaling persists over time, with epithelial cell numbers in Dhh^−/− and Gli1^−/− homozygotes remaining deficient at 3 and 7 days p.i., and ciliated cell deficiency persisting to 14 days p.i. (Figure 4E,H,I,J,K; S4M). These results indicate that EMF signaling maintains epithelial integrity by preserving differentiated airway epithelial cells and through regenerative proliferation in basal and secretory cells (Figure 4L).

Pharmacologic Hedgehog pathway activation protects against airway injury

Confirming the requirement for EMF activity at the time of injury, administration of the SMO antagonist XL-139⁴¹ markedly increased SO₂ sensitivity (Figure 5A,B,C,D), with loss of ~96% of ciliated and ~91% of secretory cells at 24 hpi (Figure 5E,F), and also blocked injury-induced augmentation of IL-6 and nuclear pSTAT3 (Figure 5G,H,I,J). Remarkably, administration of SAG21k, a small molecule agonist of SMO^41–44, dramatically protected SO₂-injured trachea, with survival of ~66% vs. ~9.7% of ciliated and ~82% vs. ~43% of secretory cells in SAG21k-treated as compared to vehicle-treated mice at 24 hpi (Figure 5A,E,F). This enhanced protection was accompanied by augmented X-Gal staining in an increased proportion of tracheal mesenchymal cells of Gli1^nLacZ/+ reporter mice (Figure 5B,C) and significant induction of Gli1 expression (Figure 5D), without, however, any change in the number of epithelial PNEC cells (Figure S5A,B). SAG21k administration also dramatically augmented the injury-induced expression of IL-6, and of nuclear pSTAT3 (Figure 5G,H,I,J), both with and without SO₂ injury. These protective effects of SAG21k treatment were observed in Dhh^−/− homozygous mutants, but not in Gli1^−/− or IL-6^−/− homozygotes (Figure S5C,D), consistent with SAG21k action through SMO, downstream of DHH but upstream of GLI1 and IL-6 in the EMF signal cascade.

Figure 5. Pharmacologic modulation of EMF signaling governs severity of SO₂ inhalation injury.

(A) Pharmacologic modulation of EMF signaling using Hh antagonist XL-139 (Hh ant), Hh agonist SAG21k (Hh ag), or vehicle (veh). (B) Tracheal sections from Gli1^nLacZ/+ reporter mice with pharmacologic suppression (Hh ant) or augmentation of cell number and intensity of X-Gal staining in tracheal mesenchyme. Scale bar=50μm. (C) Average number of X-gal stained mesenchymal cells in B (n=3). (D) Gli1 expression (RT-qPCR) normalized to Hprt (n=6). (E,F) Tracheal sections showing ciliated, secretory, and basal cells from uninjured or injured mice treated as indicated (n=3). Scale bar=20μm. (G) Tracheal sections from uninjured or injured mice treated as indicated showing phospho-STAT3 (pSTAT3, red). Scale bar=40μm. (H) IL-6 expression (RT-qPCR) normalized to Hprt (n=6). (I) Average number of pSTAT3⁺ cells per field in G (n=3). (J) Schematic diagrams showing effects of Hh pathway modulation on STAT3 activation (red nuclei) and epithelial integrity. See also Figure S5.

Influenza virus infection severely injures Gli1 mutant airway

To test the role of EMF signaling in injury from respiratory viruses, we infected male mice with the H1N1 strain of influenza A virus (IAV). We noted no major deficit of tracheal epithelial cells 24 hours post-infection (Figure 6A,Bi), although KRT5⁺ basal cells displayed a significant increase in proliferation (Figure 6A,Bii). The tracheal epithelium of Gli1^−/− homozygous mutant males in contrast displayed a dramatic reduction of ciliated cells (Figure 6Biii,C), although the basal cell proliferative response remained intact. Consistent with a generally increased sensitivity to H1N1 in female mice^45,46, Gli1^−/− homozygous females displayed denudation of tracheal epithelium and died within 24 hours (Figure 6Biv,C).

Figure 6. EMF signaling protects against damage from influenza and SARS-CoV-2 virus infection.

(A) Experimental scheme for infection with the A/Puerto Rico/8/1934 strain of the Influenza A Virus (H1N1; hpi, hours post-infection). (B) Tracheal sections from mice of the indicated genotypes and genders, uninfected (i) or 24 hpi (ii,iii,iv); quantification in (C). Elevated basal epithelial expression of Ki-67 in males (ii,iii), with complete loss of ciliated cells in Gli1 homozygotes (iii); severe loss of all epithelial cells and death ensued 24 hpi in Gli1^−/− females (iv) Scale bar=20μm, (n=3). (D) Experimental scheme for panels E,F,G. (E) Tracheal sections from uninfected or infected mice at indicated days post-infection (dpi). (F) Quantification of surviving ciliated cells in (E). Scale bar=20μm, (n=3). (G) Schematic summaries of tracheae from uninfected (Ctrl) or infected mice at 7dpi, with indicated treatments. Ki-67+ basal nuclei indicated in red. (H) Experimental scheme for H1N1 infection in mice of indicated genotypes and pharmacologic treatments in panels (I,J,K). (I) Weight loss as percent decrease after H1N1 infection for male mice of the genotype and treatment indicated (n=7). (J) Survival curve showing that infected Gli1^−/− homozygotes are moribund by 4 dpi vs. 7 dpi for wild-type mice; most mice treated with Hh ag (SAG21k) survived to 8 dpi. Log-rank test compared vehicle-treated Gli1^+/+ to SAG21k-treated Gli1^+/+ mice (p=0.002), vehicle treated Gli1^+/+ to vehicle-treated Gli1^−/− mice (p=3.9 X 10⁻⁴), and SAG21k-treated Gli1^+/+ to Gli1^−/− mice (p=1.7 X 10⁻⁴) (n=7 for each group). (K) IL-6 RNA expression (RT-qPCR normalized to Hprt) from tracheae of mice uninfected or 2 dpi, with treatments as indicated (n=6). (L) Experimental scheme for infection of mice expressing human ACE2 under control of the keratin 18 promoter with the WA-1 isolate of the SARS-CoV-2 virus, with Hh antagonist (ant, XL-139), Hh agonist (Hh ag, SAG21k), or vehicle (veh) treatment. (M) Tracheal sections from uninfected mice or mice 48 hpi, with veh, Hh ant, or Hh ag treatment. Scale bar=20μm, quantification in Figure S6I,J. (N) Lung sections from mice either uninfected or 48 hpi. Scale bar=100μm, quantification in Figure S6K,L. Abbreviations: PDPN, podoplanin; αSMA, alpha smooth muscle actin; SARS-CoV-2, SARS-CoV-2 nucleocapsid protein. See also Figure S6.

Pharmacologic Hedgehog pathway activation protects against H1N1 injury

Extended H1N1 infection in wild-type males produced a dramatic loss of tracheal ciliated cells by 3 days post-infection, with continued absence at 7 days post-infection (Figure 6D,E,F). In striking contrast, SAG21k treatment preserved tracheal ciliated cells in infected mice, even at 7 days post-infection (Figure 6D,E,F; results summarized in Figure 6G).

In bronchial airways ciliated cells are lost 1–3 days post-infection in Gli1^−/− homozygous mutant males, but this loss is delayed to 3–7 days post-infection in wild-type males (Figure S6A), and ciliated cells are still present in the lung at 7 days post-infection in males treated with SAG21k (Figure S6A). Augmented EMF signaling thus protects against injury from H1N1 infection, whereas blockade aggravates injury.

EMF signaling also protected against infection-associated weight loss with extreme, intermediate, and least severe weight loss observed in Gli1^−/−, wild-type, and SAG21k-treated male mice, respectively (Figure 6H,I). Overall survival paralleled weight loss, with no Gli1^−/− mutant males surviving beyond 5 days post-infection, wild-type mice surviving 6–7 days post-infection, and most mice receiving SAG21k surviving beyond 8 days post-infection (Figure 6J).

The protective effect of EMF signaling against viral injury entails a mesenchymal response, with infection-induced IL-6 expression dramatically suppressed in Gli1^−/− mice, and augmented by treatment with SAG21k (Figure 6K). Expression of viral genes encoding the M1 and NS1 proteins indicates viral load and viral spread, respectively^47,48. Augmented and diminished expression of M1 and NS1 viral genes over time (2d, 5d, 7d post-infection) occurred in Gli1^−/− homozygous and SAG21k-treated males as compared to wild-type (Figure S6B,C,D), aligning with effects of EMF signaling activity on epithelial integrity, weight loss, and IL-6 induction.

Hedgehog pathway blockade exacerbates airway damage from SARS-CoV-2 infection

To determine whether EMF signaling affects SARS-CoV-2-associated injury we utilized mice expressing the human SARS-CoV-2 receptor ACE2 under control of the Keratin18 promoter (K18-hACE2) and infected with the ancestral isolate of the SARS-CoV-2 virus (WA-1; Figure 6L–N; Figure S6E–L). No observable changes in the tracheal epithelium of vehicle-treated or SAG21k-treated wild-type mice were noted 2d post-infection (Figure 6L,M; S6I,J). In stark contrast mice treated with the Hedgehog pathway antagonist XL-139 displayed extensive loss of tracheal epithelial cells, especially ciliated cells (Figure 6L,M; S6I,J). By RNA in situ hybridization, IL-6 tracheal expression increased ~6-fold 2d post-infection, and was augmented by SAG21k treatment in both uninfected and infected mice (Figure S6E,G). Treatment with XL-139 in contrast reduced baseline IL-6 expression ~2-fold in uninfected mice and fully suppressed the induction of IL-6 expression by viral infection (Figure S6E,G). The WA-1 isolate of SARS-CoV-2 targets the gas exchange region of the deeper lung (Figure 6N; S6F), and we also noted similar effects of SAG21k and XL-139 treatment on IL-6 expression by RT-qPCR (Figure S6H). Epithelial cells, and especially ciliated cells, were lost in the intrapulmonary airway of infected mice treated with XL-139 (Figure 6N; S6F,K,L), a loss not observed in vehicle- or SAG21k-treated mice infected by the WA-1 or Delta or Omicron variants of SARS-CoV-2, even at 7 days post-infection (Figure S6M,N). We also noted a strikingly disorganized appearance of alveolar structure in WA-1-infected mice treated with XL-139 (Figure 6N).

Expression of Desert hedgehog and Gli1 in pancreatic islets

We noted in pancreas of Dhh^CreERT2;R26^Ai6 reporter mice a pattern of Dhh expression strikingly suggestive of the morphology of pancreatic islets (Figure 7A). Endocrine cells within pancreatic islets share many characteristics of neuroendocrine cells in other epithelial organs^49–55, and we noted that >95% of marked cells displayed co-expression of insulin, indicating β-cell identity (Figure 7B). Dhh-expressing cells do not co-express glucagon or pancreatic polypeptide, markers of α and δ islet cells, respectively (Figure S7A). With tamoxifen treatment of the Gli1^CreERT2;R26^Ai6 reporter mouse, we identified Hedgehog-responsive Gli1-expressing cells that are elongated and predominantly localized to the islet perimeter or occasionally intermixed with insulin-expressing β cells (Figure 7C). In their morphology and location these cells closely resemble islet mesenchymal support cells^56,57 and do not co-express insulin, glucagon, or pancreatic polypeptide (Figure 7C; S7B).

Figure 7. EMF signaling protects mouse and human pancreatic islets.

(A,B) Dhh expression (green) in tamoxifen-induced Dhh^CreERT2/+;R26^Ai6 mice (A), co-localizes with insulin (red immunostain) expressed in β-cells (B). (C) Gli1 expression (green) in tamoxifen-induced Gli1^CreERT2/+;R26^Ai6 reporter mice does not co-localize with insulin. Dashed outlines indicate marked cells. Scale bars=100μm in (A) and 20μm in (B,C). (D) Experimental scheme for acute streptozotocin (STZ) injury (1 dose, 120 mg/kg) in panels E-H. (E) Relative expression of IL-6 (RT-qPCR) normalized to Hprt in pancreas of indicated genotype and treatment, uninjured or 12 hpi (n=3). (F) Pancreatic islets from mice of indicated genotypes, uninjured or 48 hpi, with immunostaining for glucagon and insulin to visualize α- and β-cells. Scale bar=20μm. (G) Percentage of islets within the indicated size range (maximum diameter) in the indicated genotypes and treatments. (H) Percentage of islets within the indicated range of β-cell composition (β-cells/(α-cells + β-cells) in the indicated genotypes. P values in G and H determined using the Kolmogorov-Smirnov test, where “s” indicates significance at a confidence level of 95%, and “ns” indicates no significant difference (n=6). (I) Schematic drawings of islets from mice of the genotypes and treatments indicated. (J) Patients aged 18–89 years from two distinct datasets (Stanford and National) that were diagnosed with basal cell cancer but not diabetes mellitus (DM) and treated with an FDA-approved Hedgehog pathway inhibitor (one of the SMO antagonists, vismodegib, sonidegib or glasdegib) were propensity score matched 1:10 with other adults as controls that were diagnosed with basal cell cancer but not DM and not treated with SMO antagonists. In the Stanford Dataset, 9 out of 90 (10.0%) participants taking SMO inhibitors developed DM compared to 40 out of 900 (4.4%) controls (OR 2.39 CI: 1.12–5.10). In the National Dataset, 16 out of 251 (6.4%) participants taking SMO inhibitors developed DM compared to 98 out of 2510 (3.9%) controls (OR 1.68 CI: 0.97–2.89). Meta-analysis of the two non-overlapping cohorts demonstrated increased risk of DM in patients taking SMO inhibitors (OR 1.89 CI: 1.22–2.94, p=0.005 by random-effect meta-analysis). See also Figure S7.

EMF-dependent, injury-induced expression of IL-6 in pancreatic islets

To investigate potential injury-induced activation of EMF signaling in pancreatic islets a single dose of streptozotocin (STZ), a natural product toxic to β cells⁵⁸, was injected intraperitoneally (120 mg/kg) in mice genetically or pharmacologically manipulated to block Hedgehog pathway activity. Dhh^−/− homozygous male mice succumbed rapidly (Figure S7C,D), whereas female mutants survived, likely due to the protective effect of estrogen against STZ injury in females^59,60. Pancreatic IL-6 expression in female mice was induced ~10 fold within 12 hours of STZ treatment, but not in Gli1^−/− or Dhh^−/− mutant females, nor in XL-139-treated wild-type females (Figure 7D,E).

EMF signaling protects β-cells in STZ injury

Intraperitoneal injection of glucose in uninjured mice causes a spike in blood glucose from a baseline of ~180 mg/dL to ~400 mg/dL within 30 minutes, with a return to baseline by ~90 minutes in female wild-type and Dhh^−/− homozygotes (Figure S7E, dashed lines). In STZ-injured wild-type mice blood glucose spiked from a baseline of ~300mg/dL to ~500mg/dL, then slowly and steadily decreased to ~400mg/dL by 120 minutes after challenge (Figure S7E, solid black line). Glucose levels of STZ-injured Dhh^−/− mice averaged ~475mg/dL and upon glucose challenge spiked to the maximum reading range of 600mg/dL, with no decrease throughout the subsequent 120 minutes (Figure S7E, solid red line).

Histologic analysis showed that although islet number decreased by ~40% in both wild-type and Dhh^−/− mice at 48 hours after STZ treatment (Figure S7F,G), the injury-induced decrement in islet size differed dramatically, from an initial range of ~100–400 μm in both to <50μm for ~50% of islets in Dhh^−/− homozygotes but only ~10% of islets in wild-type (Figure 7F,G; S7F).

The islets of uninjured wild-type and Dhh^−/− female mice were very similar in their composition, with β cells comprising between ~70–90% of total α + β cells (Figure 7F,H). More than 40% of islets in STZ-injured Dhh^−/− mice, however, dropped below 10% β cells, whereas those of injured Dhh^+/+ mice mostly contained 50% or greater β cells (Figure 7F,H). Neuroendocrine EMF signaling thus appears to operate in pancreatic islets and in Dhh^−/− homozygous mutant mice STZ injury leads to a markedly more severe loss of β cells.

We also analyzed more moderate injury like that used to model diabetes⁶¹ by dosing mice at 40 mg/kg STZ on several consecutive days (Figure S7H). We noted by in situ hybridization that IL-6 expression is induced specifically in Gli1-expressing mesenchymal cells (Figure S7I,J). As compared to wild-type mice, five days of such STZ treatment in Dhh^−/− homozygotes resulted in loss of injury-induced proliferation (Ki-67; Figure S7K,L), dramatically augmented apoptosis (TUNEL staining; Figure S7M,N), and augmented levels of fasting glucose (Figure S7O). Dhh expression in β-cells, Gli1 expression in mesenchymal cells, and Dhh- and Gli1-dependent STZ induction of IL-6 together indicate that injury-responsive EMF signaling occurring in the endocrine compartment of the pancreas is functionally nearly identical to that found in airway (Figure S7P).

Pharmacologic Hedgehog pathway activation protects pancreatic β-cells

In SAG21k-treated mice we noted that overall islet size and number as well as proportion of β-cells were not reduced by injury with the single high dose of STZ (120 mg/kg; Figure 7F,G,H). Of particular interest, this preservation of β-cells by SAG21k treatment, even in Dhh^−/− mutants, strikingly exceeded that of endogenous neuroendocrine EMF signaling in wild-type mice (Figure 7F,G,H; summarized schematically in 7I), suggesting that preservation of β-cells may be enhanced by pharmacologic enhancement of endogenous EMF signaling.

EMF signaling reduces the incidence of human diabetes

Type 1 and type 2 diabetes mellitus (T1DM and T2DM) share the common feature of β-cell loss, whether from immune attack (T1DM) or by association with metabolic syndrome, which is affected by many factors including diet, lifestyle, and genetics (T2DM)⁶². T2DM may also be associated with environmental chemicals including persistent organic pollutants such as dioxins, and metals such as arsenic^63,64. The association of T2DM with arsenic is particularly intriguing, as arsenic inhibits Hedgehog signal response^65–67. Whether this association of arsenic with T2DM is due to reduced EMF signaling and consequent β-cell loss is unclear, however, given the pleiotropic nature of arsenic effects.

We therefore examined the adult onset of diabetes in two human cohorts, selected because they were undergoing Hedgehog pathway inhibitor treatment (with SMO antagonists, vismodegib, sonidegib, or glasdegib⁶⁸) for basal cell cancer (BCC). Control cohorts also suffering from BCC but not undergoing Hedgehog inhibitor treatment were paired with treatment cohorts using 10:1 propensity score matching to mitigate potential confounding effects, as previously described^69,70. As seen in Figure 7J, patients in both the Stanford Dataset and the National Dataset undergoing Hedgehog inhibitor treatment displayed a higher probability of developing diabetes than patients not treated with Hedgehog pathway inhibitors, and a meta-analysis combining both groups showed a 1.89-fold increased incidence of diabetes as compared to the control cohort (p=0.005). This association of diabetes with highly specific Hedgehog inhibitors suggests that EMF signaling indeed may play a role in maintaining the integrity of human β-cells.

DISCUSSION

An essential neuroendocrine cell function in regenerative signaling

Our work finds that EMF initiated by DHH signal from sparse solitary pulmonary neuroendocrine cells (PNECs; 1% of tracheal epithelial cells) functions to protect and promote repair of the entire airway. Amplification of EMF signaling entails GLI1-mediated expression of IL-6 in ~20% of mesenchymal cells, with rapid expansion to non-Gli1-expressing cells that may include infiltrating cells that produce IL-6 and the soluble IL-6 receptor (IL-6r), enabling a generalized response via IL-6 trans-signaling, as previously reported in lung and other tissues^71–74. These amplification steps extend to and encompass essentially all epithelial and mesenchymal cells of the airway, in a manner sufficiently rapid as to limit the apoptosis caused by injury, followed by a wave of regenerative proliferation. Both protection from apoptosis and regenerative proliferation are dependent on EMF signaling initiated by DHH secretion from neuroendocrine cells and GLI1-mediated response in mesenchymal cells.

Previously characterized functions of PNECs, primarily in the context of neuroepithelial bodies (NEBs) within the lung, include oxygen- chemo- and mechano-sensing, hormonal control of pulmonary blood flow^75–78, immune modulatory activity with implications for the pathogenesis of emphysema and asthma^79,80, and some progenitor cell activity with specific types of injury^81,82. But the functions of solitary PNECs have remained mysterious, with the exception of a report that chronic hypoxia over a period of 2–4 weeks increases solitary PNECs which can eventually aid in providing protection from hypoxia⁸³. The acute response we describe here is quite distinct, occurring within hours instead of weeks and with no increase in PNEC numbers. This role for neuroendocrine cells in rapid response and protection from acute injury is mirrored in pancreatic islets, where DHH expression from β-cells provides protection against acute injury by STZ.

EMF signaling in other organs

An emerging theme in our understanding of postembryonic tissue repair mechanisms is that they often employ signaling pathways, such as the Hedgehog pathway, that guide the development of specialized structures and organs during embryogenesis. The neuroendocrine EMF signaling characterized here in airway and pancreatic islets may have counterparts in the bladder, intestine, prostate, and other endodermal epithelial organs that also employ Hedgehog-initiated reciprocal epithelial-mesenchymal signaling for control of regenerative activity^16,17,84. The universal features shared by EMF signaling in these endodermal derivatives include an epithelial Hedgehog signal and a GLI1-mediated mesenchymal response with feedback to the epithelium. In the bladder, for example, widespread expression of Shh occurs in basal epithelial cells, and transcriptional activation of proliferation and differentiation factors such as Wnt2, Wn4, Fgf16, Bmp4, and Bmp5 are induced via GLI1 in bladder mesenchyme^16,17,85. In the intestine, other Hedgehog signals, largely IHH, elicit a response via GLI1 in a subset of mesenchymal cells that includes Wnt signals critical for maintenance of the epithelium^86–89; mesenchymal GLI1-mediated response also appears to control and pattern the branched outgrowth of regenerating murine prostate in male castrates supplemented with testosterone^17,31.

Another common feature of EMF signaling is baseline expression of GLI1 in mesenchyme, but with activation of GLI1 transcriptional targets dependent upon injury. As demonstrated for airway and pancreatic islets in this work, injury-dependent activation of GLI1 targets in mesenchyme was also previously established in bladder. In the airway we show that the molecular basis for this injury-dependence is that GLI1 binding to the promoter of its target gene is triggered by injury. Further work will be required to determine whether injury triggers GLI1 binding to target gene promoter regions in other endodermal tissues.

In addition to distinctive mesenchymal targets that are activated by GLI1 in EMF signaling in different organs (see above), the actual Hedgehog family member and the cellular source of the epithelial Hedgehog signal differ. In airway and pancreatic islets the DHH signal originates from neuroendocrine cells, whereas SHH and IHH from other types of epithelial cells are the signals in bladder and intestine, respectively. These differences warrant a consideration of the evolution of EMF signaling. The three-membered Hedgehog gene family appears to have arisen in tandem with the two sequential genome duplications that are associated with emergence of the vertebrate lineage⁹⁰. The Shh and Ihh genes are more closely related and represent the re-duplicated members of one branch, whereas only Dhh survives from the other branch⁹¹. One aspect of duplication within the vertebrate Hedgehog family is the diversification of expression and consequently the distinct functions of each family member, as exemplified by the unique role of Dhh in development of the testis and in peripheral nerve sheaths^23,91–93. Another aspect of Hedgehog family evolution, however, may be retention of an ancestral role in regenerative EMF signaling as described here, which would have arisen after initial evolution of the Hedgehog gene (in the common ancestor with cnidarians) but predating the genome duplications that occurred in the founding of the vertebrate lineage. Of particular interest given this apparent deep evolutionary conservation and the association of Dhh with neuroendocrine cells in airway and islets is the widespread presence of neuroendocrine cell networks in many epithelial organs, suggesting that a broader exploration of the role of neuroendocrine EMF signaling circuitry in injury response of other organs may prove fruitful.

EMF protection against respiratory viral injury

Although previous work suggested that influenza A virus (IAV) infection may induce expression and activation of the GLI1 protein within the mouse lung^94,95, the role of Hedgehog signaling in IAV-associated injury remained unknown. We have found here that DHH signaling within the neuroendocrine EMF circuit is protective against viral injury, as accelerated loss of ciliated cells in tracheal and intrapulmonary airway is a particularly prominent feature of genetically or pharmacologically impaired EMF signaling during IAV infection in mice. EMF signaling thus is critical to the preservation of ciliated cells, whose mucociliary clearance activity may be important in limiting viral spread during active infection. The prolonged survival of H1N1 virus-infected mice treated with SAG21k agonist correlates with and may result from this preservation of ciliated cells and their mucociliary clearance activity, as this activity would be expected to slow or prevent viral spread to the gas exchange regions of the lung and slow or prevent disease progression in our H1N1 infection experiments.

A similar effect of pharmacologic EMF impairment on ciliated cells was noted in SARS-CoV-2 infection, again in airway of the trachea and deeper lung. The mesenchymal target of Hedgehog signaling in viral infection, as in acute inhalational injury, is IL-6. This protective role for IL-6 is intriguing, given the harmful role proposed for IL-6 in advanced, severe SARS-CoV-2 infection^96,97. In contrast, we show here a beneficial protective role of IL-6-inducing EMF signaling at early stages of SARS-CoV-2 infection. Consistent with this beneficial effect in early infection, some work suggests that blockade of IL-6 signaling using the receptor-blocking antibody tocilizumab in SARS-CoV-2 patient therapy does not actually enhance recovery or survival^98,99. A general beneficial effect of IL-6 signaling in viral response is also supported by the observation that IL-6 mutant mice fare more poorly in IAV infection. The cellular mechanism of this beneficial effect may involve immune cells and remains to be further elucidated¹⁰⁰.

Therapeutic activation EMF circuit activity

Our work highlights a neuroendocrine EMF signal cascade headed by DHH from epithelial neuroendocrine cells and requiring injury-activated mesenchymal response mediated by GLI1, resulting in expression of IL-6. Although prior studies have identified a role for IL-6/STAT3 activity in promoting airway cell survival, regeneration, and differentiation^101–104, the induction of IL-6 by EMF signaling sheds light on the physiologic response to airway injury in vivo and provides new points of access to therapeutic intervention. Full tissue effects thus can be accessed through upstream manipulations of EMF signaling that initially affect only a minority of cells. To illustrate, we note that a small molecule Hedgehog pathway antagonist in mice blocks the response to DHH of a small number of mesenchymal cells yet dramatically exacerbates injury throughout the entire airway epithelium. In contrast, and of potential therapeutic significance, treatment with a small molecule SMO agonist affects these same mesenchymal cells and generates a potent and widespread protective and regenerative effect in injury. Similar small molecule-based pharmacologic augmentation of airway EMF circuit activity may improve protection and/or recovery from acute airway injury and chronic disease such as that reported in factory workers¹⁰⁵, firefighters and other first-responders¹⁰⁶, and military service members¹⁰⁷ subjected to short-term inhalation of toxic chemicals and toxins.

The operation of neuroendocrine EMF signaling in pancreatic islets and its ability, when activated, to protect islet β-cells suggests an approach to preservation of β-cells in the setting of diabetes. Loss of β-cells occurs early in life in type 1 diabetes, whereas type 2 diabetes is initiated by insulin resistance secondary to metabolic stress, with an average of ~40% reduction in β-cell mass⁶². The feasibility of therapeutic intervention in diabetes is suggested by: (i) the protection of β-cells from STZ injury that we observe upon pharmacologic augmentation of EMF signaling; and (ii) the apparent protective effect of EMF signaling evident from increased diabetes incidence in BCC patients undergoing Hedgehog pathway inhibitor therapy. In addition to EMF augmentation in vivo, Hedgehog pathway activation may also provide an islet-enhancing effect with potential value in the setting of islet cell transplantation, which increasingly is viewed as a promising treatment for diabetes¹⁰⁸. An approach focused on protection of endogenous tissues or amplification of their native regenerative abilities, however, may offer certain advantages over transplantation by circumventing concerns relating to achievement of efficient engraftment and management of transplant immunogenicity.

Limitations of study

Our study considers neuroendocrine EMF signaling in airway and pancreatic islets as a specific subclass of EMF signaling within a larger group of endodermal organs that includes intestine, bladder, and prostate. Our study has the limitation, however, that it is not sufficiently comprehensive to establish the prevalence of EMF signaling in all endoderm-derived organs. We also demonstrate here that binding of GLI1 to its target gene, IL-6, is injury-dependent, suggesting a role for injury-induced chromatin remodeling factors or transcription factors that may facilitate GLI1 binding. We have not in this work established the identities of these ancillary factors in airway or pancreatic islets, nor have we established more generally the requirement for such ancillary factors in EMF-mediated injury response. We also have noted a greater sensitivity of females to IAV infection and greater sensitivity of males to STZ injury. The basis for these previously reported sex differences has not been established, and we were unable to assess how or whether they interact with EMF signaling.

RESOURCE AVAILABILITY

Lead Contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Philip A. Beachy (pbeachy@stanford.edu).

Materials availability

The Dhh^CreERT2 mouse generated in this study will be made available on request, but we may require a completed materials transfer agreement if there is potential for commercial application.

Data and code availability

Unprocessed RNAseq data, in FASTQ format, have been deposited to National Institutes of Health’s Sequence Read Archive (PRJNA1256524, SAMN48185833, SAMN48185834, SAMN48185835, SAMN48185836, SAMN48185837). Raw mapped reads of RNAseq data, and corresponding analysis, in Excel format, have been deposited to Mendeley Data (https://data.mendeley.com/datasets/gr87pgn56d/2). All data reported in this paper, Tuxedo package mapping conditions, and any additional information required to reanalyze the data reported in this paper is available from lead contact upon request. This paper does not report original code.

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

MICE

Strains

This study was conducted in accordance with National Institutes of Health (NIH), Institutional Animal Care and Use Committee (IACUC), and United States Army Medical Research and Development Command (USAMRDC) Animal Care and Use Review Office (ACURO) guidelines. All procedures were performed under protocols approved by the Administrative Panel on Laboratory Animal Care (APLAC) at Stanford University. Mice were housed under specific-pathogen-free conditions on a 12 h dark/light cycle at a constant temperature (20–25°C) and humidity (45–50%), with standard laboratory rodent chow and water available ad libitum. Targeted Gli1 (Jackson Laboratory, 008211), Dhh (Jackson Laboratory, 002784), and IL-6 (Jackson Laboratory, 002650) mutants were purchased and back-crossed to the FVB background strain for at least 9 generations. The Dhh^CreERT2 mouse was generated by the Gene Targeting and Transgenics facility at HHMI Janelia Research Campus. Briefly, a targeting vector with CreERT2, eGFP, and neomycin resistance gene was inserted into exon 1, immediately following the ATG methionine translation initiation codon of the endogenous mouse Dhh locus. The neomycin resistance cassette was excised through FLP recombinase¹¹⁰ for experiments. R26^{GLI1-FLAG-EYFP}, and R26^iDTR were form Jackson Laboratory (013123 and 007900, respectively). Gli1^CreER and Shh^CreER were from Jackson Laboratory (007913 and 008211, respectively), and Ihh^CreER was generated at the Stanford transgenic facility (Lim et al., 2014) Gli1^CreERT2, Dhh^CreERT2, Ihh^CreERT2, and Shh^CreERT2 were crossed to R26^m™G (Jackson Laboratory, 007576), R26^Ai6 (Jackson Laboratory, 007906), R26^Ai9 (Jackson Laboratory, 007909), or R26^Ai14 (Jackson Laboratory, 007908) to generate reporter lines.

Breeding and mouse details

Gender-matched, virgin mice between 8 and 12 weeks of age, unaffected by prior non-related experiments were used for all experiments. Only littermates were used to compare each experiment replicate. Heterozygous siblings were inbred to generate litters with wild-type, heterozygous, and homozygous mutants. Gender and age matched littermates are allocated to experimental groups (based on genotypes) by random assignment. Breeders were at least 10 weeks old, and replaced between 6–8 months of age. Pilot experiments were done to evaluate gender differences. Initial SO2 injury did not result significant observable differences in tracheal the epithelium. Each bulk tissue RNAseq experiment group were pooled RNA from tracheas of 2 male and 2 female mice. Subsequent SO₂ injury was performed on male mice for tracheal analysis and quantifications. Pilot H1N1 experiments resulted in more severe infections in females, and subsequent infections were performed on male mice. All SARS-CoV-2 infections on K18-hACE2 female mice were performed at an AAALAC-accredited ABSL3 facility at Gladstone Institutes by our collaborator Rahul Suryawanshi as previously described¹¹¹. Pilot STZ injury resulted in higher morbidity in males; and subsequent experiments were performed on female mice unless otherwise stated.

Genotyping

Tail snips were immersed in QuickExtract^™ DNA Extraction Solution (Thermo/Fisher NC9904870) and incubated at 65°C for 60 minutes, followed by 95°C heat inactivation for 5 minutes. Two μl of extract was added to 1X GoTaq^® G2 Hot Start Green Master Mix (Promega M7422 or M7423). Dhh^CreER primers: Dhh F – TCCAGGTACTCTGGGACCTT, Dhh R1 – TGCTTGTATAGCAGAGGCAC, and Dhh R2 – CCTGAACATGTCCATCAGGT. Primers for all other genetic models are available on the Jackson Laboratory website. PCR cycling conditions: 95°C for 10 minutes, and 35 cycles of 95°C for 45s, 55°C for 45s, 72°C for 45s, and final extension at 72°C for 10 minutes. All other genotyping was performed using the same conditions, but with primers found in protocols available from The Jackson Laboratories.

HUMANS

Analysis of patient cohorts

This study utilized de-identified clinical data provided through Atropos Health’s real-world evidence platform. All data were fully de-identified in accordance with HIPAA Safe Harbor standards, and no identifiable information was accessed or used. The incidence of new diabetes mellitus (DM) among adults diagnosed with basal cell cancer who were or were not treated with an FDA-approved Smoothened (SMO) inhibitor was evaluated in a dataset of patients seen at Stanford Hospital and Clinics, and in a nationally representative electronic health record dataset (1 January 2015 to 25 January 2023) from Eversana Life Sciences. Data elements extracted for analysis included demographics (age, sex, race/ethnicity); diagnostic and treatment codes; numbers of database encounters; and comorbidity profiles (Charlson Comorbidity Index [CCI] score¹¹²; Tables S2, S3). Basal cell cancer and DM diagnoses were identified using International Classification of Diseases, 10^th edition (ICD-10) codes, and SMO inhibitor therapies were identified with Anatomical Therapeutic Chemical codes (Table S4). Adults aged 18 to 89 years with (1) a diagnosis of basal cell cancer at baseline and (2) no diagnosis of DM during the baseline period were included in the cohort. For the analysis of incident DM, the cohort was further stratified into those who did receive the SMO inhibitor therapy (vismodegib, sonidegib or glasdegib) and those who did not receive SMO inhibitor during the study period (controls). Given the large sample size of the SMO-inhibitor free cohort, 25,000 randomly selected patients were sampled for analysis, and this control cohort was matched to the treatment cohort using propensity score matching to mitigate potential confounding effects as previously described⁷⁰.

Briefly, a propensity score using age, sex, CCI score¹¹², diagnostic codes, procedure codes, medication codes, and number of encounters in the EHR database was calculated. Subsequently, a 10:1 control cohort was created to match the treatment groups by matching patients according to their propensity scores (Tables S2, S3). Outcome comparisons between these balanced groups permitted attribution of any observed differences to treatment with SMO inhibitors.

The incidence of DM was compared between cohorts who were or were not treated with SMO inhibitors. Odds ratios with 95% confidence intervals (CIs) were computed and reported in the unadjusted and propensity-matched datasets. Random-effect meta-analysis was used to analyze the combined non-overlapping cohorts. All analyses were performed using R (v4.2.1, R Core Team 2022) using the Atropos Health real-world evidence platform.

METHODS DETAILS

Cre-mediated recombination

To induce Cre-mediated recombination in mice harboring CreER alleles, up to five consecutive daily doses of tamoxifen (Sigma-Aldrich, T5648) dissolved in corn oil (Sigma-Aldrich, C8267) at 20mg/ml, was administered by intraperitoneal injection at approximately 12 mg per 30g body weight. The fifth dose was administered if animals showed no signs of distress/morbidity. Animals were analyzed 3–21 days after first injection to characterize marking pattern, and allow time for efficient marking. In all variants of Dhh^CreERT2 fluorescent reporter mice generated and tested (mTmG, Ai6, Ai9, and Ai14), marked cells are present only after tamoxifen administration. Hh pathway agonist SAG21k or antagonist XL139 dissolved in buffered saline was administered by IP at 2mg/kg and 100mg/kg, respectively. Alternatively, XL139 was also administered by oral gavage as previously described⁴¹. All intranasal instillations were performed with 20ul volume, on unanesthetized mice. Recombinant mIL-6 (R&D, 406-ML) was instilled intranasally at 0.2 ug/kg, at 90, 45 minutes, immediately prior to, and after injury. Anti IL-6r (BioXCell, BE0047) at 5mg/kg, 24, 12 hours, immediately before, and after injury. Alternatively, anti-IL-6r was also administered by IP at 15mg/kg. Diphtheria toxin (Sigma, D0564) was administered 2 weeks after the tamoxifen wash out period, with 7 consecutive daily doses (10ng by intranasal instillation, or 100ng by IP) followed by a 7 day washout period prior to SO₂ injury.

SO₂ injury

Custom order 500ppm SO₂ balanced with air was purchased through ProSpec^™ by Praxair. The compressed gas cylinder is connected to a stainless steel regulator (PRS30222321–660) and flow meter (PRSFM4360–0EPDM) to direct gas into a 2 liter induction chamber measuring 3.75”W × 9.00”D × 3.75”H (VetEquip, 941444) for injuring 1 cage of mice. Flow rate was held around ~4.3 standard cubic foot per hour, for 3 hours. Uninjured control littermates were exposed to filtered air. Upon completion of exposure, mice were returned to a new cage. Exposure always began at 11:00 AM. For injury of up to 12 mice, a Mouse Pie Cage (Braintree Scientific, INC, MPC-3 AERO) was used instead. For injury of strains not backcrossed into the FVB background for at least 9 generations, the flow rate was empirically optimized to around ~6.5 cubic foot per hour, for 4 hours.

Influenza virus infection

For Influenza infection, mice were anesthetized by IP injection of Ketamine/Xylazine mixture and infected intranasally using 20ul of the diluted A/Puerto Rico/8/34 virus (Charles River, Cat# 10100374). Virus was diluted (1:1000) in sterile PBS from the stock (Stock-5×10⁶ PFU/ml) before infection. Viral titer of the stock was determined by plaque assay in low passage Madin-Darby canine kidney (MDCK) cells.

SARS-CoV-2 virus infection

SARS-CoV-2 infection of mice was performed by Rahul K. Suryawanshi as previously described¹¹³, under protocol approved by the Institutional Animal Care and Use committees at the University of California, San Francisco and Gladstone Institutes.

Streptozotocin injury and blood glucose measurements

For high dose streptozotocin (STZ) injury, (Sigma, S0130), mice were fasted at 5 PM, with access to water for 18 hours, and then injected with 120 mg/kg of STZ in citrate buffer by intraperitoneal injection at 11 AM the following day. Mice then had access to food and water unless stated otherwise. For low dose STZ injury, mice were fasted for 4 hours starting at 7 AM, and then injected with 40 mg/kg of STZ in citrate buffer by intraperitoneal injection at 11 AM on the same day. Mice then had access to food and water. This dosing was repeated for 5 consecutive days. For blood glucose measurements, approximately 3mm of tail were snipped from fasted or un-fasted mice, and the first 5 drops of blood were blotted onto a clean Kimwipes^®, followed by sampling using an Accu-Chek^® meter, repeated twice for every measurement. Glucose tolerance test by intraperitoneal injection was performed 48 hours after streptozotocin administration, when animals had been fasted for 18 hours, with access to water. Baseline blood glucose was measured, followed by IP injection with glucose in PBS at 1.5g/kg; blood glucose was then measured every 30 minutes over a period of 2 hours.

RNA Isolation for RNAseq and Taqman RT-qPCR

For RNAseq and Taqman RT-qPCR of tracheal RNA, a segment of trachea below the distal tip of the thyroid glands and above the carina cartilage was harvested after dissection away from esophagus and overlying adventitia. For RNAseq, each experimental condition was a pool of tracheas harvested from two male and two female mice, to minimize samples submitted to sequencing for batch effects considerations. For Taqman^® RT-qPCR, each experimental condition was a pool of tracheas, or entire pancreas harvested from one male and one female mouse. All RNA isolation was performed using TRIZOL (Thermo/Fisher 15596026) according to protocol, with modifications, as follows. PureLink^™ Carrier RNA (Thermo/Fisher 12183016) was added at 25ng per 350ul of TRIZOL. For RNAseq, samples were dissolved in 1.5ml TRIZOL. For RT-qPCR, samples were dissolved in 1ml TRIZOL. The aqueous phase was taken after mixing with 0.2mL of chloroform (Sigma, C2432) per 1mL of Trizol used. RNA was precipitated with 0.5mL of 2-propanol (Thermo/Fisher A416) per 1mL of Trizol used, at room temperature for least 3 hours, and subsequently at −20°C overnight. Pellet was washed twice with 80% Ethanol, Decon^™ Labs 200 proof Ethanol, Molecular Biology Grade (Thermo/Fisher 07678003), diluted with UltraPure^™ DNase/RNase-Free Distilled Water (Thermo/Fisher 10977023) prior to drying in room temperature for at least 2 hours prior to re-suspension with non-DEPC treated nuclease free water (Thermo/Fisher AM9937). RNAseq was performed with the Illumina Hi-Seq platform at Stanford Functional Genomics Facility. Data were processed using the Tuxedo protocol¹¹⁴. For qPCR, 20ng/ul of total RNA was used to generate cDNA using the High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Thermo/Fisher 4368814). Taqman^® probes were used with TaqMan^® Universal Master Mix II (MMII), no UNG (Thermo/Fisher 4440040), on Applied Biosystems HT7900 with standard cycling protocol. Stock cDNA was diluted 5 fold with non-DEPC treated nuclease free water and used at a final dilution of 1:5 in 1X MMII, with the following 20X primer Taqman^® probes: IL-6(Mm00446190_m1), Gli1(Mm00494654_m1), and Hprt(Mm00446968_m1).

RNAseq and Taqman RT-qPCR analysis

For RNAseq, the 5 datasets were Gli1^+/− uninjured, Gli1^+/− injured plus 12 and 24 hours, and Gli1^−/− injured plus 12 and 24 hours. FPKM generated from the Tuxedo protocol¹¹⁴ were used to determine the coefficient of variation for the two Gli1^+/− injured samples, and Gli1^−/− injured samples. Data was filtered if the sum of the CV for each genotype was higher than 0.75. Data was further filtered for any transcript where the sum of reads for all data sets was less than 3.0. Injury dependent up/down regulated gene lists were determined by the ratio of average FPKM of injured Gli1^+/− over the FPKM of uninjured Gli1^+/− FPKM for each gene, or vice versa. Injury and Gli1 dependent genes were then determined by the taking the ratio of average injured Gli1^+/− FPKM and average injured Gli1^−/− FPKM for each gene. For qPCR where ΔΔCt is shown, the ΔCt values were first determined by subtracting the Ct value for Hprt from Ct value for gene of interest. Then the ΔΔCt parameter is obtained from the difference between ΔCt values for control (i.e. vehicle treated, and uninjured animals), and individual experimental group. A positive value corresponds to an increase in expression, and negative values correspond to a decrease in expression. For example, a ΔΔCt value of 2 corresponds to an increase of ~4 fold, and a value of −2 corresponds to a decrease of ~4 fold in expression compared to control, assuming near-perfect efficiency of amplification. Relative fold change is determined from calculating (2^−ΔΔCt).

Tissue processing/microscopy

For microscopy, tracheae were separated from the esophagus, followed by a cut just below the level of the thyroid and another just above the level of the carina cartilage. For unfixed frozen tracheal sections, dissected trachea were then filled with Tissue-Tek^® O.C.T. Compound (Thermo/Fisher 50–363-579) from the thoracic opening with a 25 gauge needle (BD 305122) and BD Luer-Lok^™ 1-ml syringe (BD 309628), and then positioned in a Tissue-Tek^® Standard Cryomold measuring 25mm X 20mm X 5mm (Thermo/Fisher NC1533005). Cassettes were filled with O.C.T. compound for snap freezing. Blocks were kept at −80°C for short term storage or immediately sectioned with a Leica CM3050S Cryostat, at a thickness of ~14μm. Pancreas from PBS and PFA perfused mice were dissected and fixed overnight in 4% PFA, followed by 3 X 30 minutes washes in 1X PBS, then 30% sucrose soak for 48 hours, and equilibrated in O.C.T. Pancreas were then frozen and sectioned at 12μm. The first ~25% depth of tissue was trimmed and discarded, then consecutive sections were mounted onto Superfrost^™ Plus Gold slides (Thermo/Fisher Scientific 15–188-48) until reaching ~75% depth. Sections were air dried in a fume hood at room temperature for 5 minutes and flooded with 4% PFA, diluted from Pierce^™ 16% Formaldehyde (w/v) Methanol-free (Thermo/Fisher Scientific 28908) with 1X buffered saline, and incubated at room temperature for 20–60 minutes, depending on antibodies used. Slides were washed liberally and soaked in 1X buffered saline at least 3 times for 10 minutes each. Slides were incubated in blocking buffer (1X buffered saline with 5% Gibco^® goat serum (Thermo/Fisher Scientific 16210–064) and 0.5% Bovine Serum Albumin (Sigma A7906) with or without Triton^™ X-100 (Sigma, T8787), depending on antibodies used. Primary antibodies were diluted in 1:1 blocking buffer and 1X buffered saline, applied to sections, and incubated at room temperature overnight in a humidified chamber protected from light. Slides were briefly washed in 1X buffered saline, and incubated at room temperature with secondary antibody at 1:2000 dilution in 1:1 blocking buffer and 1X buffered saline for three hours in a humidified chamber protected from light. Slides were briefly washed in 1X buffered saline with or without Tween^® 20 (Sigma, TWEEN^® 20) depending on antibodies used. This was followed by a brief gentle rinse with distilled water. Slides were allowed to dry at room temperature and mounted with ProLong^® Gold Antifade Reagent with DAPI (Thermo/Fisher P-36931) using Fisherbrand 24X60–1.5 Cover Glass (Thermo/Fisher 22266882). For unfixed frozen lung sections, euthanized mice were perfused with 20ml buffered saline; dissected lung was inflated via the trachea with 50/50 Tissue-Tek^® O.C.T. compound/PBS. Subsequent steps were identical to tracheal processing. For fixed frozen sections, euthanized mice were first perfused with 20ml buffered saline, followed by 15ml 4% PFA diluted in buffered saline. Dissected tissues were fixed in 50 ml 4% PFA at room temperature for 3 hours, then kept overnight 4°C. Tissues were washed in at least three hourly changes of 50ml buffered saline the following day. Tissues were then soaked in 30% sucrose dissolved in buffered saline at 4°C until the tissue sank, but not for more than 60 hours. Subsequent steps were the same as for unfixed frozen tissue. For paraffin embedding and sectioning, mice were euthanized and perfused with both buffered saline and PFA as described above. Dissected tracheae and lungs were submerged in 4% PFA overnight, and transferred into 70% EtOH until paraffin embedding by a local histology service, Tissue-Tek. Paraffin-embedded specimens were sectioned at a thickness of 10μm. Gli1^nLacZ β-Galactosidase expression in fresh frozen sections was detected by X-Gal staining using β-galactosidase Staining Kit (Clontech, 631780), according to protocol. Apoptotic cells were detected by TUNEL staining using In Situ Cell Death Detection Kit, TMR red (Sigma, 12156792910), according to protocol. RNA in situ hybridization was carried out using RNAScope^™.

IL-6 promoter ChIP-PCR

Gli1^CreERT2/+;R26^{GLI1-FLAG-EYFP} mice were injected with five consecutive daily doses of tamoxifen, followed by a 2 week wash out period. Mice where then subjected to SO₂ injury, then euthanized at 12 hpi, perfused with PBS, and 1% PFA prior to tracheal dissection. Dissected, cleaned, and minced tracheas were then fixed for another 20 minutes at room temperature. The fixed tracheas were then enzymatically dissociated using 3mg/ml of Collagenase Type 1 (Thermo/Fisher, 17100017), in 1X Dispase (StemCell Technologies, 07913), with 1% FBS (Omega Scientific, FB-01), and 0.5% Penicillin-Streptomycin (Gibco, 15140122), in a volume of 1mL per trachea, at 37°C. Cell suspensions were then filtered through 40 μm cell strainers (Fisher, 087711), and topped with 10% FBS in FBS, followed by swing bucket centrifugation at 500g at 4°C for 5 minutes. Cell pellets were washed twice with ice cold PBS for chromatin immunoprecipitation as previously described^115,116. PCR to detect the region of IL-6 promoter with GLI1 DNA-binding motifs was performed as previously described³⁹.

Image acquisition

Fluorescent confocal images were acquired on a Zeiss LSM800 with Airyscan and Definite Focus using ZEN 2.1 (blue edition). Images were tiled, and Z-stacks orthogonally projected, followed by brightness, and contrast adjustments made within ZEN software. Images were exported as .tif or .png files. Brightfield images were acquired on a custom Leica DMI6000 B and DMC4500 digital camera, using Leica Application Suite X version 3.4.2.18368. Images were exported as .tif. Non microscope images were taken using a Canon G9X digital camera. Images were exported as .jpg. Images for comparison in figures were subjected to identical adjustments in color, brightness, and contrast in Microsoft Powerpoint.

QUANTIFICATION AND STATISTICAL ANALYSIS

Final experiment conditions, and sample sizes were determined based on pilot studies. Independence of observations was utilized to evaluate if the data met assumptions of the statistical approach. A biological replicate is a tissue/organ from one animal, with a minimum of three age and gender matched replicates per experimental condition, done on separate days, but at same time of day as specified in the injury sections. Randomization was achieved by using different litters for replicates. Blinding was done by identifying each test subject with a number that contains no information on experiment conditions, for quantification and analysis. Experiment conditions were then revealed for that subject number for graphing. For tracheal cell counts, cells per field of view were determined and reported from male biological replicates due to concerns with differences in physiological size of organs observed in male and females. To quantify tracheal epithelial cells, the central portion of the trachea was sampled from male mice at intervals of 60μm, as longitudinal sections. Three random regions from each section were sampled to obtain average numbers of cell types. The averages from all sections were then averaged to obtain the number for each experimental group. This count was repeated in biological replicates to determine the final average. To quantify PNECs, and NEBs, the central portion of lung lobes were sampled at intervals of 60μm. The numbers of Dhh^CreERT2-marked cells co-localizing with CGRP, or SYP are the sum of counts from all sections coming from multiple biological replicates. For islet quantification, numbers were determined and reported from female biological replicates due to STZ toxicity in Dhh^−/− males. To quantify islet numbers, the central portion of the pancreas was sampled at intervals of 60μm, and the number of islets from the entire section was counted. The counts from all sections were then averaged. This count was repeated in biological replicates to determine the final average. To quantify islet size, the central portion of the pancreas was sampled at intervals of 60 μm, and the longest diameter for each islet was measured. The islets were then binned into one of six categories, depending on diameter. This count was repeated in biological replicates. Each category was then summed and divided by the total number of islets measured for the experimental group to determine percentage. To quantify proportion of β-cells, the central portion of the pancreas was sampled at intervals of 60 μm, and the proportion of β-cells was calculated as a ratio of (β-cells/(β-cells + α-cells)) for each islet. The islets were then binned into one of ten categories, depending on ß-cell proportion. This calculation was repeated in biological replicates. Each category was then summed and divided by the total number of islets measured for the experimental group to determine percentage. For RNAseq and qPCR, additional sample replicate information is described in the RNA isolation section. Bar graphs chart the average, where error bars represents standard deviation. P values were calculated using a two-tailed t-test throughout unless otherwise stated. Pairwise comparisons to determine significance p, and replicate numbers (n) are provided in the figures and/or legends.

Supplementary Material

1

Figure S1. Hedgehog pathway components in trachea, related to Figure 1.

(A and B) Expression of genes, ascertained from bulk RNAseq of tracheal cells (FPKM). RNA sample pooled from two male and two female mice. (A) Dhh, Krt14, and Krt5, showing that Dhh expression is low compared to that of basal cell markers in trachea. (B) Genes encoding essential Hedgehog pathway components are expressed in trachea. (C) Gli1 expression (green) in tamoxifen-induced Gli1^CreERT2/+;R26^Ai6 reporter mice marks a subset of tracheal mesenchymal cells, identified by Vimentin staining (red, first row); Gli1-expressing cells do not co-express markers for epithelial cells (EpCAM - white), leukocytes (CD45 - red, second row), or endothelial cells (CD31 - red, third row). Scale bar=30μm.

2

Figure S2. Expression of Hedgehog family members in tracheal epithelium, related to Figure 2.

(A) Detailed schematic of a cassette for bi-cistronic expression of CreERT2 and eGFP inserted in-frame with the ATG translation start codon of the endogenous Dhh gene by homologous recombination. The FRT-flanked neomycin resistance gene was excised by flippase (FLP) recombinase to generate the Dhh^CreERT2 allele used in experiments. (B) Testes from male Dhh^CreERT2/+ and Dhh^{CreERT2/CreERT2} littermates reveal rudimentary testicular development in adult homozygotes, consistent with earlier observations for Dhh^−/− homozygous mutants²³. (C) Dhh expression in testis in a pattern consistent with Sertoli cell localization¹⁰⁹ here is visualized by tamoxifen-induced expression of ZsGreen in Dhh^CreERT2;R26^Ai6 reporter mice. Scale bar=200μm for whole testis section (left panel), and 25μm for the magnified image (right panel). (D) Representative images visualizing Dhh, Shh, and Ihh expression by tamoxifen-induction of GFP expression in mice before and 24 hours after SO₂ injury. The Dhh^CreERT2/+;R26^m™G reporter mice showed marking of rare epithelial cells, with no apparent increase in PNEC number at 24 hpi. The Shh^CreERT2/+;R26^m™G, and Ihh^CreERT2/+;R26^m™G reporter mice showed no marking of cells in trachea before or after injury. Scale bar=20μm. (E) Tracheal Dhh expression visualized by tamoxifen-induced ZsGreen in Dhh^CreERT2/+;R26^Ai6 mice. Marked cells are rare as compared to basal cells expressing KRT5 (yellow - second row), and/or KRT14 (red - third row), consistent with PNEC frequency in airway epithelium, and data from bulk tissue RNAseq showing Dhh expression is lower than Krt5 and Krt14 (Figure S1A). Scale bar=50μm. (F) Dhh expression in trachea and lung visualized by tamoxifen-treatment of Dhh^CreERT2/+;R26^Ai9 reporter mice marks a subset of epithelial cells (fuschia) that do not co-express markers for the basement membrane (first row: Laminin in trachea; smooth muscle actin in lung (αSMA)); for endothelial cells (second row, CD31); for leukocytes (third row, CD45). Marked cells co-express the epithelial marker EpCAM (white). Scale bars: trachea=20μm, lung=60μm and insets=20μm. (G,H) Representative images (G) and quantification (H) of Ki-67 staining 24 hpi showing that cells expressing Dhh, and Gli1 (visualized by indicated reporters) do not proliferate upon injury. Scale bar=10μm. P values derived from Ki-67 expression in basal cells as compared to Dhh- and Gli1-expressing cells (n=6). (I,J) Representative images and quantification showing that PNECs expressing CGRP or SYP did not increase at 24 hpi compared to controls. Scale bar=10μm. Quantification in panel J, with P value derived by comparison of CGRP or SYP expression in injured vs. controls (n=6). (K) RNAseq from two male and two female mice (pooled RNA) showing predominant Dhh ligand gene expression in trachea, and expression of Dhh, Shh, Ihh, Gli1, Gli2, and Gli3 largely unchanged by injury (average of 12 and 24 hpi).

3

Figure S3. Injury response: gene expression, role of DHH-expressing cells, and dependence on Smo-, Gli1- and IL-6-mediated signaling, related to Figure 3.

(A,B) RNAseq data showing fold induction of IL-6 (A), and other secreted factors induced 1.2-fold or more by injury (B) (see also Table 1). Fold-induction was calculated as the FPKM ratio of injured Gli1^+/− heterozygous or Gli1^−/− homozygous mice (average of 12 and 24 hpi ± s.d.) to uninjured trachea from Gli1^+/− heterozygous mice. Each RNA sample was pooled from two male and two female mice. Note that, unlike other secreted factors, full injury-induced expression of IL-6 depends on Gli1 function (see also Figure 3B, and panel D, below). (C) Representative images of tracheal sections from tamoxifen-treated Gli1^CreERT2/+;R26^Ai6 mice, before injury and at 3 and 7 hours after beginning of injury (0 and 4 hpi), showing red puncta from IL-6 RNA in situ hybridization initially in proximity to Gli1-expressing cells (green) and then becoming more widespread. Scale bar=10μm. (D) RT-qPCR of IL-6 in tracheal RNA from the genotypes indicated. P values were calculated by comparison of Ctrl to injured samples (n=3). (E) Experimental scheme for tamoxifen induction of GLI1-FLAG in Gli1-expressing cells of Gli1^CreERT2/+;R26^{GLI1-FLAG-EYFP} mice, with RT-qPCR to detect IL-6 expression. Note that tamoxifen activation of the Rosa26^{GLI1-FLAG-EYFP} does not activate IL-6 expression in the absence of injury. P values were calculated by comparison of all experiment groups against the uninjured vehicle group (n=4). (F,G) RNAseq data showing transcription factors (F), and chromatin remodelers (G) induced 1.2-fold or more by injury (see also Table 1). Fold induction was calculated as in panes A,B. (H) Representative images of sections from dissected trachea of mice of the genotypes and treatments indicated, with intranasal instillation of diphtheria toxin (DT) and anti-IL-6 receptor antibody (IL-6r Ab). Scale bar=20μm. (I) Quantification of panel H, showing average number of ciliated, secretory, and basal cells. P values were calculated by comparing Ctr to treatment groups, as indicated (n=6).

4

Figure S4. EMF-dependent proliferation, survival, and regeneration following SO₂ injury,, related to Figure 4.

(A) Proliferation of basal and non-basal epithelial cells visualized by Ki-67 (red), and Krt5 (white) in representative images of tracheal sections from mice of Gli1 genotypes indicated, either uninjured (Ctrl) and hpi (h), as specified. Scale bar=60μm. (B) Inset of image A4, with white arrowhead pointing to a proliferating basal cell; the white arrow indicates a non-basal epithelial cell expressing a high level of Ki-67. The yellow arrowhead indicates a cell with low but detectable Ki-67. Scale bar=20μm. (C) Quantification of panel A showing average basal and non-basal cell proliferation in the left and right panels, respectively, from mice of the Gli1 genotypes indicated, Control and hpi as specified. P values were calculated by comparing Gli1^+/− to Gli1−/− (n=3). (D) Proliferation index from panel A based on any detectable level of Ki-67 in epithelial cells (left) or only high levels of Ki-67 (right), from control mice or mice at the hpi as indicated. P values were calculated by comparing Gli1^+/− to Gli1^−/− at each time-point (n=3). (E) Apoptosis of basal and non-basal epithelial cells visualized by TUNEL assay (red) in representative images of tracheal sections from mice of Gli1 genotypes indicated, either uninjured (Ctrl) and hpi (h), as specified. Scale bar=60μm. (F) Inset of image Eii, with white arrow pointing to an apoptotic basal cell, white arrowhead pointing to a non-apoptotic basal cell, yellow arrow pointing to an apoptotic non-basal cell, and yellow arrowhead pointing to a non-apoptotic non-basal cell. Scale bar=20μm. (G) Quantification of panel E showing average basal and non-basal epithelial cell apoptosis in the left and right panels, respectively, from mice of the Gli1 genotypes indicated, Control and hpi as specified. P values were calculated by comparing Gli1^+/− to Gli1^−/− (n=3). (H) Proliferation of basal and non-basal epithelial cells visualized by Ki-67 (red), and Krt5 (white) in representative images of tracheal sections from mice of the Dhh genotypes indicated, either uninjured (Ctrl) and hpi (h), as specified. Scale bar=20μm. (I) Quantification of panel H showing average basal and non-basal cell proliferation in the left and right panels, respectively, from mice of the Dhh genotypes indicated, Control and hpi as specified. P values were calculated by comparing Dhh^+/+ to Dhh^−/− (n=3). (J) Proliferation index from panel A based on any detectable level of Ki-67 in epithelial cells (left) or only high levels of Ki-67 (right), from control mice or mice at the hpi as indicated. P values were calculated by comparing Dhh^+/+ to Dhh^−/− at each time-point (n=3). (K) Apoptosis of basal and non-basal epithelial cells visualized by TUNEL assay (red) in representative images of tracheal sections from mice of Dhh genotypes indicated, either uninjured (Ctrl) and hpi (h), as specified. Scale bar=60μm. (L) Quantification of panel K showing average basal and non-basal epithelial cell apoptosis in the left and right panels, respectively, from mice of the Dhh genotypes indicated, Control and hpi as specified. P values were calculated by comparing Dhh^+/+ to Dhh^−/− (n=3). (M) Representative images of tracheal sections from wild-type, Gli1^−/− and Dhh^−/− mice at 3, 7, and 14 days p.i. Note the impaired recovery of secretory and ciliated cells in Gli1^−/− and Dhh^−/− mice. Scale bar=30μm.

5

Figure S5. Local modulation of EMF signaling in adult airway, related to Figure 5.

(A) Representative images of sections from trachea of mice treated with vehicle or SAG21k (Hh Ag), showing ciliated cells (AcTUB, green), and no notable expansion of solitary PNECs (CGRP/SYP, red). Scale bar=20μm. (B) Experimental scheme and quantification of panel A, showing that the average number of PNECs does not increase with SAG21k treatment. P value was calculated by comparison of SAG21k to vehicle treatment (n=6). (C) Representative images of sections from dissected trachea of mice of the genotypes and treatments indicated, uninjured (Ctrl) or 24 hpi. Scale bar=20μm. IP, intraperitoneal; IN, intranasal instillation. (D) Quantification of C, showing average number of ciliated, secretory, and basal cells. P values were calculated by comparing vehicle (Veh) treatment (IP) to SAG21k or IL-6 treatment.

6

Figure S6. EMF signaling attenuates severity of airway damage from H1N1 and SARS-CoV-2 infection, related to Figure 6.

H1N1 infection. (A) Representative images of lung sections from WT, Gli1^−/−, or SAG21k-treated WT mice, either uninfected (Ctrl), or infected with H1N1 for 1, 3, or 7 days (Experimental scheme, Figure 6D), and showing progressive ciliated airway cell loss. Scale bar=50μm. (B, C, D) Expression quantification of viral genes M1 and NS1 by RT-qPCR normalized to Hprt in tracheal RNA from mice of the genotype and treatment specified, either uninfected (Ctrl) or infected with H1N1 for 2 (B), 5 (C), or 7 (D) days. P values in (B) are from comparison of uninfected (Ctrl) to infected mice (no significant differences were seen between WT, Gli1^−/−, and WT+Hhag groups at 2 dpi); P values in (C,D) are from comparison of Gli1^−/−, and WT+Hhag to WT. Note the progressive divergence in viral load and spread by 5 and 7 dpi.

SARS-CoV-2 infection. (E,F,G,H) Representative images showing tracheal sections (E, in situ hybridization to IL-6 mRNA - red puncta, scale bar=20μm), or lung sections (F, immunostaining as indicated, abbreviations as in Figure 6N), from vehicle-treated, Hh ant-treated, or Hh ag-treated mice (veh, XL-139, and SAG21k, respectively), either uninfected (Ctrl) or infected with SARS-CoV-2. (G) Quantification of IL-6 RNA in situ hybridization puncta in (E) (n=3). (H) Quantification of IL-6 RNA by RT-qPCR normalized to Hprt in lung 48 h post-infection (n=6). (I,J) Average number of total epithelial cells (I) and ciliated cells (J) in trachea from Figure 6M (n=6). (K,L) Average number of lung epithelial cells (K) and ciliated cells (L) from Figures 6N and S6F, (n=3). (M) Representative images of lung sections from K18-hACE2 mice infected with the WA1, Delta, or Omicron variants for 4 or 7 days, showing minimal loss of airway ciliated cells at all time points. Scale bar=250μm. (N) Quantification of panel M, showing average number of ciliated cells in intrapulmonary airways. P values were calculated by comparing uninfected to infected mice (n=3).

7

Figure S7. EMF signaling reduces diabetogenic effects and toxicity of streptozotocin, related to Figure 7.

(A) Dhh-expressing cells in pancreatic islets (green) are visualized by tamoxifen-induced ZsGreen expression in Dhh^CreERT2/+;R26^Ai6 reporter mice, and do not co-localize with α-cell marker glucagon (row 1, red), or δ-cell marker pancreatic polypeptide (row 2, red). (B) Gli1 expressing cells (green) are visualized by tamoxifen-induced ZsGreen expression in Gli1^CreERT2/+;R26^Ai6 reporter mice, and do not co-localize with α-, or δ-cell markers (rows 1, and 2, respectively). Dotted outlines show borders of marked cells. Scale bars in A and B=20μm.

(C) Experimental scheme for severe acute streptozotocin STZ injury (120 mg/kg, 1 dose) in panels D, E, F, and G. (D) Hh pathway activity strikingly affected survival of STZ-injured male mice, with highly significant survival differences (log-rank test) when comparing Dhh^−/− to Dhh^+/+ (p < 1.4X10⁻⁴) or Dhh^−/− to SAG21k-treated Dhh^−/− mice (p < 1.4X10⁻⁴), but not when comparing Dhh^+/+ to SAG21k-treated Dhh^−/− mice (p=0.317) (n=7 for each group). (E) Intraperitoneal glucose tolerance test (IPGTT) was performed in fasted mice by measurement of glucose prior to and 30, 60, 90, and 120 minutes after glucose challenge (1.5g/kg IP glucose) in mice either uninjured (Ctrl) or 48 hours after STZ injury. P values were calculated for STZ-treated Dhh^−/− and Dhh^+/+ groups by comparing glucose values, n=4. (F) Representative images of pancreatic sections from female mice of the Dhh genotypes indicated, either uninjured (Ctrl) or 48 hpi, with α- and β-cells visualized by immunostaining for Glucagon and Insulin, respectively. Scale bar=500μm. (G) Average number of islets per pancreatic section in mice of the indicated genotypes and treatment. P values were calculated by comparison of injured to uninjured (Ctrl) groups (n=4). (H) Experimental scheme for panels I-O, with two and five days of low dose STZ for panels I,J and K-O, respectively. (I) Representative images of pancreatic sections from tamoxifen-treated Gli1^CreERT2/+;R26^ZsG/+ mice showing IL-6 mRNA (red puncta) from in situ hybridization. (J) Quantification of the average number per field of IL-6 mRNA puncta closer than (+) or more distant than (−) 10μm from ZsGreen signal (Gli1-expressing cells, green) after 2 days of low dose (40 mg/kg) STZ injury (see experimental scheme in panel H). Scale bar=10μm. P value was calculated by comparing cells Gli1-expressing cells to cells not expressing Gli1 (n=3). (K) Representative images of pancreatic sections from Dhh^+/+ and Dhh^−/− mice either uninjured (Ctrl) or injured daily five times with low dose STZ (40 mg/kg) as indicated in the experimental scheme in panel H. Note the presence after injury of proliferative pancreatic β-cells (Ki-67-positive, red; Insulin-positive, green) largely in islets from Dhh^+/+ but not Dhh^−/− mice. Scale bar=50μm. (L) Quantification of images in panel K, showing the average of proliferative β-cells. P values were calculated by comparing injured (STZ) and uninjured (Ctrl) Dhh^+/+ to Dhh^−/− mice (n=6). (M) Representative images of pancreatic sections from Dhh^+/+ and Dhh^−/− mice either uninjured (Ctrl) or injured daily five times with low dose STZ (40 mg/kg) as indicated in the experimental scheme in panel H. Note the presence after injury of more apoptotic pancreatic β-cells (TUNEL-positive, red; Insulin-positive, white; Glucagon-positive, green) in islets from Dhh^−/− as compared to Dhh^+/+ mice. Scale bar=50μm. (N) Quantification of images in panel M, showing the average of apoptotic β-cells. P values were calculated by comparing injured (STZ) and uninjured (Ctrl) Dhh^+/+ to Dhh^−/− mice (n=6). (O) Fasting blood glucose measurements from female mice of Dhh genotypes indicated, injured daily five times with low dose STZ (40 mg/kg) as indicated in the experimental scheme in panel H. P value was calculated by comparison of Dhh^+/+ to Dhh^−/− mice (n=6). (P) Schematic of neuroendocrine EMF signaling in pancreatic islets, showing induction of GLI1 expression in mesenchymal cells (blue) by DHH expressed in β-cells, and GLI1- and injury-dependent induction of IL-6.

8

Supplemental Table 1, SO2 injury-regulated gene expression, Related to Figure 1.

9

Supplemental Table 2, Stanford cohort demographics, Related to STAR Methods “Analysis of patient cohorts.”

10

Supplemental Table 3, Nationwide cohort demographics, Related to STAR Methods “Analysis of patient cohorts.”

11

Supplemental Table 4, Medication and ICD 10 codes used for cohort analysis, Related to STAR Methods “Analysis of patient cohorts.”

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken anti-GFP	Abcam	(Abcam Cat# ab13970, RRID:AB_300798)
Mouse anti-Vimentin	Abcam	(Abcam Cat# ab8978, RRID:AB_306907)
Rabbit anti-alpha smooth muscle Actin antibody	Abcam	(Abcam Cat# ab5694, RRID:AB_2223021)
Rabbit anti-BCL2	Abcam	(Abcam Cat# ab59348, RRID:AB_2064155)
Rabbit anti-Keratin5	Abcam	(Abcam Cat# ab53121, RRID:AB_869889)
Rabbit anti-Ki-67	Abcam	(Abcam Cat# ab15580, RRID:AB_443209)
Rabbit anti-pSTAT3	Abcam	(Abcam Cat# ab76315, RRID:AB_1658549)
Rabbit anti-Sypnatophysin	Abcam	(Abcam Cat# ab32127, RRID:AB_2286949)
Chicken anti-Keratin5	Biolegend	(BioLegend Cat# 905901, RRID:AB_2565054)
Rabbit anti-Keratin14	Biolegend	(BioLegend Cat# 905301, RRID:AB_2565048)
Rat anti-mouse CD326 (Ep-CAM) Alexa Fluor® 647 conjugated	Biolegend	(BioLegend Cat# 118212, RRID:AB_1134101)
Rat anti-mouse CD326 (Ep-CAM) PE conjugated	Biolegend	(BioLegend Cat# 118205, RRID:AB_1134176)
Rat anti-mouse CD45 Alexa Fluor® 647 conjugated	Biolegend	(BioLegend Cat# 103124, RRID:AB_493533)
Rat anti-mouse CD45 PE conjugated	Biolegend	(BioLegend Cat# 103106, RRID:AB_312971)
Rat anti-mouse Podoplanin Alexa Fluor® 647	Biolegend	(BioLegend Cat# 156204, RRID:AB_2750408)
Hamster anti-Pecam-1	Millipore	(Millipore Cat# MAB1398Z, RRID:AB_94207)
Rabbit anti-CGRP	Millipore	(Millipore Cat# PC205L, RRID:AB_2068524)
Rabbit anti-Laminin Alexa Fluor® 647 conjugated	Novus	(Novus Cat# NB300-144AF647, RRID:AB_2891039)
Rabbit anti-Glucagon	Proteintech	(Proteintech Cat# 15954-1-AP, RRID:AB_2878200)
Rabbit anti-Insulin	Proteintech	(Proteintech Cat# 15848-1-AP, RRID:AB_10597100)
Rat anti-Insulin	R and D Systems	(R and D Systems Cat# MAB1417, RRID:AB_2126533)
Chicken anti-RFP	Rockland	(Rockland Cat# 600-901379, RRID:AB_10704808)
Mouse anti-RFP	Rockland	(Rockland Cat# 200-301379, RRID:AB_2611063)
Rabbit anti-RFP	Rockland	(Rockland Cat# 600-401-379, RRID:AB_2209751)
Rabbit anti-SCGB1A1	Santa Cruz Biotechnology	(Santa Cruz Biotechnology Cat# sc-25555, RRID:AB_2269914)
Mouse anti-acetylated tubulin	Sigma-Aldrich	(Sigma-Aldrich Cat# T6793, RRID:AB_477585)
Mouse anti-FLAG M2	Sigma-Aldrich	(Sigma-Aldrich Cat# A2220, RRID:AB_10063035)
Mouse IgG–Agarose	Sigma-Aldrich	(Sigma-Aldrich Cat# A0919, RRID:AB_1172448)
Mouse anti-SARS-CoV/SARS-CoV-2 Nucleocapsid Antibody	Sino Biological	(Sino Biological Cat# 40143-MM05, RRID:AB_2827977)
Mouse anti-ZsGreen	Takara Bio	(Takara Bio Cat# 632598, RRID:AB_2943078)
Rabbit anti-ZsGreen	Takara Bio	(Takara Bio Cat# 632474, RRID:AB_2491179)
Rabbit anti-GFP	Thermo Fisher Scientific	(Thermo Fisher Scientific Cat# A-11122, RRID:AB_221569)
Bacterial and virus strains
Influenza A/PR/8/34 (H1N1)	AVS Bio	10100374
SARS-CoV-2 WA1	Suryawanshi et al.117
SARS-CoV-2 Delta	Suryawanshi et al.117
SARS-CoV-2 Omicron	Suryawanshi et al.117
Chemicals, peptides, and recombinant proteins
QuickExtract™ DNA Extraction Solution	Biosearch Technologies	Cat#QE09050
Recombinant Mouse IL-6 Protein	Bio-Techne/R&D Systems	Cat#406-ML
InVivoMAb anti-mouse IL-6R	BioXCell	Cat#BE0047
InVivoMAb rat IgG2b isotype control, anti-keyhole limpet hemocyanin	BioXCell	Cat#BE0090
InVivoPure pH 7.0 Dilution Buffer	BioXCell	Cat#IP0070
Andwin Scientific Tissue-Tek™ CRYO-OCT Compound	Fisher Scientific	Cat#1437365
Decon™ Labs 200 proof Ethanol, Molecular Biology Grade	Fisher Scientific	Cat#07-678-003
Gibco™ Goat Serum, New Zealand origin	Fisher Scientific	Cat#16-210-064
Gibco™ PBS, pH 7.4	Fisher Scientific	Cat#10-010-049
Gibco™ Penicillin-Streptomycin (10,000 U/mL)	Fisher Scientific	Cat#15-140-122
Invitrogen™ Nuclease-Free Water (not DEPC-Treated)	Fisher Scientific	Cat#AM9937
Invitrogen™ UltraPure™ DNase/RNase-Free Distilled Water	Fisher Scientific	Cat#10977023
Thermo Scientific™ Pierce™ 16% Formaldehyde (w/v), Methanol-free	Fisher Scientific	Cat#PI28908
500ppm SO2 balanced with air	Linde PROSPEC	Custom order
Omega Scientific Inc Fetal Bovine Serum, USDA Approved, 500 Ml	Omega Scientific	Cat#FB-01
GoTaq® G2 Hot Start Master Mixes	Promega	Cat#M7423
Bovine Serum Albumin	Sigma-Aldrich	Cat#A7906
Chloroform	Sigma-Aldrich	Cat#C2432
Corn oil	Sigma-Aldrich	Cat#C8267
Diphtheria Toxin	Sigma-Aldrich	Cat#D0564-1MG
Sodium citrate tribasic dihydrate	Sigma-Aldrich	Cat#S4641
Sodium hydroxide	Sigma-Aldrich	Cat#S8045
Streptozocin	Sigma-Aldrich	Cat#S0130
Tamoxifen	Sigma-Aldrich	Cat#T5648
Triton™ X-100	Sigma-Aldrich	Cat#T8787
TWEEN® 20	Sigma-Aldrich	Cat#P1379
Dispase (5 U/mL)	STEMCELL Technologies	Cat#7913
2-Propanol, ACS Grade	Thermo Fisher Scientific	Cat#A416
Collagenase, Type I, powder	Thermo Fisher Scientific	Cat#17100017
ProLong Gold antifade reagent with DAPI	Thermo Fisher Scientific	Cat#P-36931
PureLink™ Carrier RNA	Thermo Fisher Scientific	Cat#12183016
TRIzol™ Reagent	Thermo Fisher Scientific	Cat#15596026
Critical commercial assays
RNAscope™ Multiplex Fluorescent Reagent Kit v2	Bio-Techne/ACDBio	Cat#323100
Applied Biosystems™ High-Capacity cDNA Reverse Transcription Kit	Fisher Scientific	Cat#4368814
Applied Biosystems™ TaqMan™ Universal Master Mix II, no UNG	Fisher Scientific	Cat#4440040
In Situ Cell Death Detection Kit, TMR red	Millipore	Cat#12156792910
Beta-Galactosidase Staining Kit	Takara Bio	Cat#631780
Deposited data
Illumina HiSeq™ Data (RNA-seq): SRR33346779, SRR33346776, SRR33346778, SRR33346780, SRR33346777	This paper	BioProject: PRJNA1256524
RAW mapped reads	This paper	https://data.mendeley.com/datasets/gr87pgn56d/2
RNA-Seq analysis	This paper	https://data.mendeley.com/datasets/gr87pgn56d/2
Experimental models: Cell lines
MDCK (NBL-2)	ATCC	Cat#CCL-34
Experimental models: Organisms/strains
Gli1CreERT2: Gli1tm3(cre/ERT2)Alj/J	The Jackson Laboratory	RID:IMSR_JAX:007913
ShhCreERT2: B6.129S6-Shhtm2(cre/ERT2)Cjt/J	The Jackson Laboratory	RRID:IMSR_JAX:005623
Ihh CreERT2	Lim et al.37	Available upon request.
Dhh CreERT2	This paper	Available upon request, and will be donated to The Jackson Laboratory upon publication.
Gli1nLacZ: Gli1−/−: Gli1tm2Alj/J	The Jackson Laboratory	RRID:IMSR_JAX:008211
Dhh−/− : B6;129-Dhhtm1Amc/J	The Jackson Laboratory	RRID:IMSR_JAX:002784
IL-6−/− : B6.129S2-Il6tm1Kopf/J	The Jackson Laboratory	RRID:IMSR_JAX:002650
R26mTmG: Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J	The Jackson Laboratory	RRID:IMSR_JAX:007576
R26ZsGreen: R26Ai6: B6.Cg-Gt(ROSA)26Sortm6(CAG-ZsQreen1)Hze/J	The Jackson Laboratory	RRID:IMSR_JAX:007906
R26tdTom: R26Ai9: B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J	The Jackson Laboratory	RRID:IMSR_JAX:007909
R26tdTom: R26Ai14: B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J	The Jackson Laboratory	RRID:IMSR_JAX:007908
R26GLI1-FLAG-EYFP: Gt(ROSA)26Sortm6(Gli1)Amc/J	The Jackson Laboratory	RRID:IMSR_JAX:013123
R26iDTR: C57BL/6-Gt(ROSA)26Sortm1(HBEGF)Awai/J	The Jackson Laboratory	RRID:IMSR_JAX:007900
FVB: FVB/NCrl	Charles River	RRID:IMSR_CRL:207
K18-hACE2: B6.Cg-Tg(K18-ACE2)2Prlmn/J	The Jackson Laboratory	RRID:IMSR_JAX:034860
Oligonucleotides
DhhCreERT2 Forward primer - TCCAGGTACTCTGGGACCTT	IDT DNA	Custom order
DhhCreERT2 Reverse primer 1 - TGCTTGTATAGCAGAGGCAC	IDT DNA	Custom order
DhhCreERT2 Reverse primer 2 (alternative) - CCTGAACATGTCCATCAGGT	IDT DNA	Custom order
RNAscope™ Negative Control Probe - DapB	Bio-Techne/ACDBio	Cat#310043
RNAscope™ Probe - Mm-Il6	Bio-Techne/ACDBio	Cat#315891
RNAscope™ Probe - Mm-Il6ra	Bio-Techne/ACDBio	Cat#435921
Mouse Gli1 TaqMan	Thermo Fisher Scientific	Cat#Mm00494654_m1
Mouse Hprt TaqMan	Thermo Fisher Scientific	Cat#Mm00446968_m1
Mouse IL-6 TaqMan	Thermo Fisher Scientific	Cat#Mm00446190_m1
Software and algorithms
SPSS	IBM	https://www.ibm.com/spss
ZEN Blue	Zeiss	https://www.zeiss.com/microscopy/us/products/software/zeiss-zen.html
Tuxedo Suite	Trapnell et al.120	https://cole-trapnell-lab.github.io/papers/trapnell-tuxedo-protocol/
Image Lab	Bio-Rad	https://www.bio-rad.com/en-us/product/image-lab-software?ID=KRE6P5E8Z
Tuxedo Suite mapping parameters	This paper	https://data.mendeley.com/datasets/gr87pgn56d/2
Other
Mouse Pie Cage for Aerosol Delivery	Braintree Scientific	Cat#MPC-3- AERO
Andwin Scientific Cryo Mold Intermediate	Fisher Scientific	Cat#NC9642669
BD Disposable Syringes with Luer-Lok™ Tips 1mL	Fisher Scientific	Cat#14-823-30
BD General Use and PrecisionGlide Hypodermic Needles 25G	Fisher Scientific	Cat#14-826AA
BD PrecisionGlide™ Single-use Needles 27G	Fisher Scientific	Cat#14-826-48
BD PrecisionGlide™ Single-use Needles 30G	Fisher Scientific	Cat#14-821-13A
Falcon™ Cell Strainers 40 μM	Fisher Scientific	Cat#08-771-1
Fisherbrand™ Rectangular Cover Glasses	Fisher Scientific	Cat#12-541-037
Fisherbrand™ Tissue Path Superfrost™ Plus Gold Slides	Fisher Scientific	Cat#15-188-48
Stainless Flow Meter with EPDM Seals and O-rings range 4.5–45 sccm	Praxair	PRSFM4360-0EPDM
Single Stage Stainless Regulator 0–50psi range	Praxair	PRS30222321-660
Induction Chamber, 2L	Vet Equip	Cat#941444

Highlights.

Desert hedgehog (Dhh) is expressed in airway and pancreatic islet neuroendocrine cells (86)

Dhh elicits protective/regenerative response via epithelial-mesenchymal feedback (EMF) (86)

Amplification from rare neuroendocrine cells expands the EMF effect to the entire trachea (89)

Drug-enhanced EMF activity protects against chemical and viral injury of airway and β-cells (87)