IPF AT2 cells are stuck in transition and biophysically dysfunctional

Abstract

Idiopathic Pulmonary Fibrosis (IPF) is an incurable disease with extensive molecular, cellular, and organ level dysfunction. A major gap in IPF research is the lack of understanding of how short-term cellular behavior causes long-term tissue remodeling. By optimizing lung slices from explanted human lungs, we discovered foci of migratory non-canonical alveolar type 2 (AT2) cells in regions of established lung fibrosis and found that these cells are trapped in states of cellular transition that are driven by persistent developmental repair programs. Consistent with these biophysical behaviors, pharmacological activation of β-catenin reproduced persistent migration, whereas YAP activation restrained it. We conclude that imbalanced developmental programs drive AT2 cell motility and lesion heterogeneity, providing a mechanistic link between short-term cellular dynamics and slowly progressive fibrosis of IPF.

Idiopathic pulmonary fibrosis (IPF) is an incurable lung disease that begins with microscopic fibroproliferative lesions heterogeneously distributed throughout the lung. It is conventionally understood that these lesions expand to remodel the lung parenchyma on an organ-size scale over extended time periods of months to years. However, it remains unclear whether, and through what mechanisms, short-term cellular biophysical behaviors drive the progressive, chronic process of lung remodeling.

In murine and in vitro models, we previously showed that key injury-associated processes involved in the development of lung fibrosis result in IL-6- and YAP-dependent migration of distal airway epithelia, a behavior associated with fibrotic cyst formation^1,2. Consistent with these findings, others have reported alveolar type 2 (AT2) cells also migrate in an ArpC3-dependant manner in injured murine lungs³. Following injury to the terminal airway or alveolar epithelium, tissue-resident stem cells (e.g., basal, club, or AT2 cells) activate diverse transcriptional programs that initiate self-renewal and differentiation towards terminally committed cell types^4–6. During this process, cells transit through intermediate cell fates and reach distinct compartments through migration^7–12. Together, these findings suggest a biophysical framework for IPF progression. However, in the native fibrotic human lung, the dynamic migratory behaviors of epithelial cells and their relationship to specific intermediate cell fates remain completely unknown.

EpCAM^pos cells migrate in lung fibrosis

To investigate migratory phenotypes of epithelial cells in human lung tissue, we performed precision-cut lung slices (PCLS) experiments on explanted human lungs from unaffected control subjects and patients with IPF (Fig. 1a, Extended Data Table 1). To label epithelial cells, lung slices were incubated with fluorescently labeled anti-EpCAM antibody and imaged in a time-lapse manner. Individual labeled cells and their trajectories were identified, analyzed, and statistically compared. To determine if a cell is actively migrating, we analyzed cell speed as a metric for temporal motion along its path, and confinement ratio as a metric for overall motion throughout its path. In control lungs and histologically normal regions of IPF lungs, most epithelial cells remained at their initial positions, exhibiting confinement ratio below 0.5 and displacement of only a few dozen microns (Fig. 1b, Supplementary Video 1). In contrast, in regions of established fibrosis of IPF lungs, we found foci of intensely migrating cells, characterized by substantially higher speeds and confinement ratios. Within these areas of established fibrosis, migration was unique to each cell, persistent, and characterized by frequent changes in trajectory directions. However, this intense migration was not observed in every region of established lung fibrosis or uniformly across the entire field of view (FOV) of imaging. These findings are consistent with the known pathogenic heterogeneity of IPF and suggest that the movement of cells might be a consequence of the underlying lung fibrosis¹³. Alternative to migration that is unique to each cell, collective migration of airway epithelial cells has been reported in IPF-derived airway cultures and in injured murine airway^1,2. Despite sampling multiple fields of view across IPF lungs, we did not detect a collective migration phenotype in human lung tissue.

Fig. 1. EpCAM^pos migrating cells are found in regions of established lung fibrosis.

a, Overall design of experiment to study migration of epithelial cells in PCLS from healthy and IPF lungs. Two parameters were used to study cells migration: speed and confinement ratio. Confinement ratio is a ratio of net displacement between initial and last time points divided by the total path length (confinement ratio is a continuous measure with 0 indicating no displacement and 1 indicating a straight-line trajectory). Created with BioRender.com. b, Representative phenotype of EpCAM^pos cells migration of healthy and IPF lungs. From top to bottom: representative histological regions (H&E); trajectories of cells over 12 hours (colored) and background EpCAM signal (grey); cells distribution of mean speed per track; distribution of confinement ratio. Scale bar is 50 μm. c, Cumulative data of migration of control lung parenchyma (N=8 lungs, 24 regions, 31,213 cell tracks), IPF lung parenchyma (N=4 lungs, 10 regions, 17,536 cell tracks), and IPF fibrotic regions (N=7 lungs, 31 regions, 75,931 cell tracks). Left: mean speed probability distribution, right: confinement ratio probability distribution. d, Pairwise comparison of migration statistics of IPF fibrotic regions to regions of control lung parenchyma in panel C. Left panels: mean track speed comparisons. Right panels: confinement ratio comparisons. Upper panels: Kolmogorov-Smirnov test p-values. Bottom panels: Cohen’s D-values. White color represents no significance.

To quantitatively compare migratory phenotypes across fibrotic and non-fibrotic lung regions, we selected multiple FOVs from fibrotic regions of IPF lungs and compared these with control lungs and normal-appearing parenchyma of IPF lungs. Average cell speeds were significantly increased in IPF normal parenchyma relative to control lung parenchyma and were further elevated in regions of established fibrosis (Fig. 1c, Extended Data Table 2). Confinement ratio distributions were almost identical between unaffected control lungs and normal-appearing parenchyma from IPF lungs. In contrast, areas of established lung fibrosis demonstrated elevated confinement ratios compared with parenchyma from both control and IPF lungs. These differences in both speed and confinement ratio were further supported by pairwise statistical comparisons across all regions imaged (Fig. 1d). The broad distribution of D-values underscores the substantial variation in speed and confinement migratory phenotypes, reflecting the intrinsic spatial heterogeneity of lung fibrosis in IPF.

To visualize the pathological features of areas characterized by intense epithelial migration, we isolated regions of statistically significant migratory phenotypes using 2–4 mm punch biopsies from PCLS and performed histological analyses (Fig. 2). Histologic evaluation confirmed that migrating EpCAM^pos cells are localized to regions of fibrotic lung remodeling; although the degree of fibrosis varied from patchy involvement to nearly completely fibrotic (Extended Data Fig. 1). Notably, our findings indicate that regions of intense migration are histologically characterized as areas of transition between minimally remodeled lung and end-stage lung fibrosis. However, of all regions of intense migration, only 22.5% had fibroblastic foci and none had honeycomb cysts, suggesting that these two hallmark features of IPF are not associated with the observed migration.

Fig. 2. Captured regions of intense EpCAM^pos migration are in fibrotic regions of variable density.

Example of three regions of migration captured for histology. Trajectories of EpCAM^pos cells: trajectories of cells over 12 hours (colored) and background EpCAM signal (grey). The region of imaging was noted and punch-excised, then embedded to FFPE block. Left panel (trajectories) matches the right panel (H&E staining) with precision of ~200 μm. Due to tissue distortion often occurring during FFPE processing of thin PCLS tissue, we focused on characterizing the whole punch-excised regions (2–4 mm in size) that contained the FOVs of intense migration within. Further analysis covers such whole punch-excised regions. Scale bars from left to right: 100 μm, 500 μm, 500 μm, 500 μm, 100 μm.

Migratory cells express HT2–280 and KRT5

Next, we characterized cell types present in areas of migration. All (100%) analyzed migratory regions contained abundant HT2–280 positive cells, identifying AT2 cells as a consistent component of these regions (Fig. 3a). As expected, HT2–280-positive cells were also present in unaffected control lung parenchyma. In contrast, whereas normal lung parenchyma contained RAGE-positive alveolar type 1 (AT1) cells, these cells were absent from regions of intense migration in IPF lungs (Extended Data Fig. 2). To assess the presence of airway epithelial cells, we first examined airway-associated cell morphology and detected these cells in 19% of the regions of intense migration (Extended Data Fig. 3a). Consistent with this observation, immunostaining for the club cells marker SCGB1A1 revealed positive cells in only 12% of regions of intense migration (Extended Data Fig. 3b).

Fig. 3. Migrating EpCAM^pos cells are likely HT2–280 and Krt5 double positive.

a, Imaged regions of control and IPF lungs have multiple HT2–280 positive cells (AT2 cells), including regions of intense migration (27 FOVs of migration from 6 IPF analyzed). Scale bars are 100 μm. b, Lysotracker trajectories often repeat EpCAM trajectories in regions of migration of IPF lungs. Scale bars are 50 μm. c, CD206 and FolR2 trajectories (blue) do not repeat highly migratory EpCAM positive cells (yellow). Scale bars are 50 μm. d, Regions of intense EpCAM migration (N=23 from 6 IPF lungs analyzed) of IPF lungs contain HT2–280 and KRT5 double positive cells (magenta arrows), as well as single positive HT2–280 cells (orange arrows), and single positive Krt5 cells (white arrows). All HT2–280 positive cells in such regions are negative for KRT17. Scale bars are 20 μm. e, Few regions of intense EpCAM migration (N=23 from 6 IPF lungs analyzed) of IPF lungs contain KRT5 and KRT17 double positive cells (white arrows). Scale bar is 20 μm. f, Uncultured control lungs (N=7) do not have HT2–280 and KRT5 double positive cells. All tested uncultured IPF lungs (N=7) have HT2–280 and KRT5 double positive cells that are negative for KRT17. Scale bars are 100 μm for zoomed out and 20 μm for zoomed in images. g, Normal parenchyma of uncultured IPF lungs contain low number of HT2–280 and KRT5 double positive cells (as portion of all HT2–280 cells) but their abundance is significantly increased in areas of progressed fibrosis (Student’s t-test). h, Illustration describing observed regions of migration. In contrast with healthy lung parenchyma containing AT1 and AT2 cells, IPF regions of migration contain AT2 cells that are positive for KRT5, some KRT5 and KRT5/KRT17 cells. Most cells in the IPF lung are migrating independently while monolayers are also present.

To confirm that migrating EpCAM^pos cells were AT2 cells, we stained live PCLS tissue with both anti-EpCAM antibody and Lysotracker, a combination frequently used to identify AT2 cells by flow cytometry¹⁴. Many migratory trajectories observed in the EpCAM^pos cells were recapitulated in the LysoTracker channel, indicating that the same cells were identified by these markers (Fig. 3b, Supplementary Video 2). To further characterize migrating EpCAM^pos cells in areas of established lung fibrosis, we performed high-magnification confocal imaging of IPF lungs (Supplementary Video 3). Migrating EpCAM^pos cells in these FOVs were approximately 10–12 μm in diameter consistent with the reported size of AT2 cells^15,16. To distinguish our EpCAM^pos migratory population from possible immune cell migration, we stained multiple PCLS samples using anti-CD206 and anti-FolR2 antibodies – surface markers of alveolar and tissue-resident interstitial macrophages, respectively¹⁷. Migration of EpCAM^pos cells did not overlap with the CD206^pos or FolR2^pos cells (Fig. 3c, Supplementary Video 4 and 5). In aggregate, these findings indicate that AT2 cells constitute the predominant EpCAM^pos migratory population in fibrotic regions of IPF lungs.

Accumulation of alveolar-airway transitional cells is a hallmark of IPF and is often associated with incomplete differentiation of AT2 cell^{10,11,18–21}. To determine whether migratory AT2 cells in IPF possessed markers of transitional alveolar cells, we stained for KRT5, KRT8, and KRT17²². In 72% of migratory regions, KRT5^pos/HT2–280^pos, KRT5^pos/HT2–280^neg and KRT5^neg/HT2–280^pos cells co-existed, with KRT5^pos/HT2–280^pos cells markedly outnumbering the other cell types (Fig. 3d). These three cell populations often occurred as isolated cells but occasionally formed contiguous monolayers spanning ~100–200 μm or more. KRT17^pos cells were detected in 37% of the migratory regions (Fig. 3e). Nearly all KRT17^pos cells co-expressed KRT5 and none were positive for HT2–280, distinguishing them from the predominant KRT5^pos/HT2–280^pos migrating cells. KRT8 was present in AT2 cells from both control and IPF lungs (Extended Data Fig. 4). However, quantitative analysis of median KRT8 signal revealed no enrichment of KRT8^high signature in AT2 cells within the migratory regions^9,10.

To determine if KRT5/HT2–280 co-positivity arose as an artifact of PCLS culture^23,24, we examined uncultured control and IPF lungs. KRT5^pos/HT2–280^pos cells were absent from control lungs (Fig. 3f, Extended Data Fig. 5). In contrast, all IPF lungs contained regions of KRT5^pos/HT2–280^pos cells, organized as both single cells and monolayers, resembling the architecture observed in PCLS migratory regions (Fig. 3f, Extended Data Fig. 5). While KRT5^pos/HT2–280^pos cells were detectable at low abundance in some non-fibrotic regions, their frequency was significantly increased in regions of established lung fibrosis (Fig. 3g). Notably, KRT5^pos/HT2–280^pos cells expressed lower levels of KRT5 than airway basal cells, indicating a distinct epithelial population (Extended Data Fig. 5d).

Published single-cell RNA-sequencing (scRNA-seq) studies report accumulation of KRT5-positive AT2 cells in IPF lungs^25,26. These cells are noted to co-express few other airway basal cell markers. To further evaluate this cellular population, we interrogated additional IPF-related scRNA-seq datasets^19,20. Although robust co-expression of KRT5 with canonical AT2 markers was not consistently detected, KRT5 expression was evident within a SCGB3A2-positive distal club cells population (Extended Data Fig. 6)²⁷. These distal lung KRT5-positive cells shared a subset of selected basal markers reported with KRT5^pos AT2 cells. Pathway analysis of KRT5-associated transcriptional programs in distal airway epithelial cells revealed activation of RhoA-mediated migratory signaling, increased energy utilization, and activation of multiple pathways implicated in IPF pathogenesis. Together, these findings support our conclusion that a population of noncanonical, KRT5-expressing AT2 cells represent the primary migratory population within regions of established lung fibrosis (Fig. 3h).

Migration relies on altered WNT and YAP

To investigate the molecular mechanisms underlying epithelial cell migration in regions of established lung fibrosis, we performed a series of targeted perturbations of candidate signaling pathways. Activation of β-catenin, mimicking canonical WNT signaling, initiated migration of EpCAM^pos cells in normal unaffected lung parenchyma, phenocopying the migratory behavior observed in fibrotic regions of IPF lungs (Fig. 4a, Supplementary Video 6). In these β-catenin stimulated samples of control lungs, migration was unique to each cell, persistent, and characterized by frequent changes in trajectory direction. In regions of established lung fibrosis of IPF lungs, migration of EpCAM^pos cells was significantly attenuated by inhibition of pan-PI3K or ROCK1/2 (Fig. 4a). Immunostaining assays revealed no evidence of increased proliferation (Ki67), DNA damage (phospho-H2A.X), or epithelial-mesenchymal transition (ZEB1) of AT2 cells within migratory regions compared with uncultured control lung parenchyma or IPF lungs (Fig. 4b). Together, these findings indicate that AT2 cells in fibrotic regions of IPF lungs engage a distinct β-catenin-driven program of cell migration that is uncoupled from proliferation, DNA damage, or EMT.

Fig. 4. Migratory EpCAM^pos cells are stimulated by WNT and inhibited by YAP.

a, Migration of EpCAM cells in control lungs is stimulated by β-catenin transcriptional activator BML-284 (N=3 lungs, 8 regions, 5566 cell tracks) compared to not-treated samples (N=3 lungs, 9 regions, 4,383 cell tracks). Migration of EpCAM cells in IPF lungs is downregulated by PI3K inhibitor Pictilisib (N=4 lungs, 9 regions, 1753 cell tracks), ROCK inhibitor Y-27632 (N=4 lungs, 9 regions, 2605 cell tracks), and YAP activator XMU-MP-1 (N=4 lungs, 9 regions, 2302 cell tracks) compared to not-treated samples (N=4 lungs, 15 regions, 5510 cell tracks). 20% of the fastest cells were used in the analysis from each group. b, AT2 cells (HT2–280^pos) in regions of migration do not have increased markers of proliferation, DNA damage or EMT. Kruskal-Wallis test used. c, Compared to uncultured control lungs (N=19 samples, 7 control lungs) YAP nuclear localization is diminished in regions of migration of IPF lungs (N=26, 6 IPF lungs), as well as in patchy fibrotic regions of uncultured IPF lungs (N=18, 7 IPF lungs). Kruskal-Wallis test used. d, Illustration describing response of migrating cells to treatments. In control lungs, activation of β-catenin leads to intense migration. In IPF lungs, YAP activation and PI3K/ROCK inhibition leads to decreased migration.

YAP is a critical co-transcriptional activator required for successful AT2-to-AT1 differentiation, especially in transitional cells^10,28–30. To determine its role in AT2 cells migration, we induced YAP nuclear localization and co-transcriptional activity in regions of established fibrosis. Surprisingly, this treatment reduced migration of EpCAM^pos cells (Fig. 4a). In contrast, our previous observations in small airway epithelia demonstrated that YAP activation induced a migratory phenotype². These differences likely result from distinct roles of YAP in airway epithelia (proliferation and migration)³¹ versus AT2 cells (differentiation)^28–30.

To quantify endogenous YAP activity in regions of migration, we measured the nuclear-to-cytoplasmic ratio of YAP. HT2–280^pos cells within regions of migration of IPF lungs exhibited significantly reduced nuclear YAP compared to uncultured control lungs (Fig. 4c). Interestingly, fibrotic uncultured samples of IPF lungs demonstrated a similar decreased level of YAP nuclear localization in AT2 cells, indicating HIPPO activation in the majority of AT2 cells in these regions. Together, these data suggest that migrating AT2 cells initiate differentiation toward an AT1 identity but fail to complete this program due to insufficient YAP activation¹⁰. These findings lead us to conclude that KRT5^pos/HT2–280^pos migratory AT2 cells appear to be trapped in a transitional state driven by dysregulated developmental programs (Fig. 4d).

Discussion

Despite major advances in understanding IPF pathogenesis, the cellular dynamics and biophysical alterations underlying IPF pathogenesis remain completely unknown. For the first time, we directly examined the biophysical state of epithelial cells in human IPF lung tissue. We discovered that the slowly progressive fibrotic features of IPF are driven by fundamental defects in lung cell development that prevent normal repair of damaged alveoli by AT2 cells. These fundamental defects manifest as discrete foci of migratory AT2 cells within fibrotic regions of the lung. These AT2 cells appear to be trapped in a state of transition, ultimately preventing differentiation of AT2 to AT1 cells and restoration of the alveolar epithelium. These findings establish a clear relation between biophysical dysfunction of the alveolar epithelium with the slowly progressive microscopic fibrotic lesions that define IPF.

Reactivation of cell developmental programs is a central determinant of effective repair following injury. During embryogenesis and lung repair, coordinated signaling between epithelial, immune, mesenchymal, and endothelial compartments is mediated by major pathways such as WNT, SHH, HIPPO, NOTCH, PI3K, and TGF-β^32–37. Throughout lung development, stem cells induce proliferation, migration, and differentiation to establish specialized epithelial lineages, to drive branching morphogenesis, and to form alveoli. Similarly, during lung repair, stem cells are guided by these development programs to repopulate injured regions with appropriate epithelial lineages^38,39. However, accumulated epigenetic modifications and senescence-associated stress in de-differentiated or recruited stem cells can constrain spatial and temporal control of developmental programs, limiting tissue repair potential. While dysregulation of developmental pathways in IPF alveolar epithelia has been described^19,40, a corresponding biophysical defect had remained unknown. Our observation that persistently migratory non-canonical AT2 cells cluster within regions of established lung fibrosis indicates that these cells are ineffective in restoring the alveolar epithelium.

Our findings reveal that both canonical and non-canonical AT2 cells migrate within fibrotic regions of the lung and appear to be stuck in transition, preventing their normal differentiation to AT1 cells. This incomplete alveolar epithelial differentiation limits AT1 cell formation and impairs restoration of gas exchange. Onco-suppressor pathway P53, usually activated in repair to prevent uncontrolled proliferation, can also accelerate cell senescence in prolonged transitional state^40,41. Moreover, sustained migration of de-differentiated cells imposes a substantial energetic burden, increasing susceptibility to metabolic stress and oxidative damage^42,43. Collectively, our findings demonstrate that ineffective transition of AT2-to-AT1 cells is detrimental to tissue homeostasis and leads to persistent migration of non-canonical KRT5^pos/HT2–280^pos cells driven by developmental repair programs. Focusing on the inherent defects in AT2-to-AT1 cell transition that are associated with biophysical dysfunction represents a promising approach to identifying critical pathogenic mechanisms and key therapeutic targets for IPF, an incurable and progressive type of lung fibrosis.

Methods

Human PCLS culture

IPF lungs were obtained as lung transplantation explants at University of Colorado Anschutz Medical Campus Hospital. Informed consent was obtained from all participants under the COMIRB (Colorado Multiple Institutional Review Board) 11–1664 study entitled “Functional Profiling of Interstitial Lung Disease”. Control non-fibrotic non-smoker lungs were obtained as non-transplantable donor lungs from The International Institute for the Advancement of Medicine (IIAM), with informed consent collected directly by the organization. Blood was perfused out through vasculature with ice-cold sterile PBS (Gibco 10010049) containing 1x Pen-Strep (100 units/mL of penicillin and 100 μg/mL of streptomycin, Gibco 15140122). A lobe or a segment of a lobe was infused with Agarose solution: 2% ultra-low gelling temperature Agarose (Sigma-Aldrich A2576), 1x MEM (Gibco 11430030) pH adjusted to 7.5, 1x Pen-Strep at 40 ⁰C. Tissue was immersed in ice-cold PBS plus 1x Pen-Strep at 4 ⁰C for 1 hour. Distal lung tissue was excised with scalpel within 2 cm region from pleura or, in case of dense subpleural fibrosis, 2 cm region from dense fibrosis region. For IPF lungs, normal or transitional tissue was selected based on CT-guided lung dissection. Tissue was sliced at Compresstome (Precisionary Instruments VF-510–0Z) at 500 μm thickness. Tissue was cultured in 1:1 ratio of DMEM (Corning 10–013-CV) and BEBM (Corning CC-3171) supplemented with selected Bronchial Epithelial SingleQuots (Corning CC-4175) kit components BPE, Insulin, Hydrocortisone, GA-1000, Transferrin, Triiodothyronine, Epinephrine, additionally supplemented with Retinoic acid (Sigma-Aldrich R2625) at 50 nM, hEGF (Corning 354052) at 25 ng/mL, BSA (Fisherbrand BP9703100) at 1.5 μg/mL, 1x Pen-Strep, Amphotericin B (Sigma-Aldrich A2942–100ML) at 0.5 μg/mL. Media was changed daily. Viability of tissue was assessed retrospectively via H&E staining with criteria similar to reported²³. Tissue that did not pass viability criteria was discarded from the data collection.

Human PCLS cells labeling, imaging and treatments

Slices were incubated with conjugated anti-EpCAM antibody (Bioss bs-1513R-BF647) at 20 ug/mL for 3 hours, washed 4 times for 10 minutes with fresh culture media, then de-stained for 2 hours. LysoTracker Green DND-26 (Invitrogen L7526) labeling was done at 1 μM concentration for 1 hour, followed by 4 washes for 10 minutes with fresh media and immediate imaging after microscope adjustment time. Anti-CD206 (Proteintech 60143–1-Ig) and anti-FolR2 (Biolegend 391702) antibodies were conjugated with Alexa Fluor 555 using antibody labeling kit (Invitrogen A20187), incubated with slices at 50 ug/mL for 12 hours prior anti-EpCAM labeling, washed 4 times for 10 minutes and de-stained during anti-EpCAM labeling. Labeled slices were placed into individual wells on a 12-well plate with 1 mL media and weighed down with truncated 12-mm Transwells (Corning 3460). Keyence BZ-X810 with humidified stage incubator (37 °C, 5% CO2) was used for imaging at the following parameters: 10x magnification, 200 μm z-stack through depth of tissue with 7 μm steps, time lapsed imaging each 3 to 6 minutes for 12 hours. Corresponding fluorescent filters were used for each channel: ET-CY5 for EpCAM, ET-CY3/R for CD206 and FolR2, GFP for LysoTracker. For high magnification confocal imaging, Andor BC43 (Oxford Instruments) was used with the same labeling process. Imaging parameters were: 40x magnification, 100 μm z-stack through depth of tissue with 5 μm steps, time lapsed imaging each 3 to 6 minutes for 12 hours.

For treatments, BML-284 (Selleck Chemical S8178) at 10 μM, Pictilisib (Medchemexpress HY-50094) at 5 μM, Y-27632 (Selleck Chemical S1049) at 10 μM or XMU-MP-1 (Tocris Bioscience 6482) at 5 μM were added to slices starting 5 hours before the imaging (during anti-EpCAM staining) and then kept in the media throughout de-staining and imaging. 0.1% DMSO was used in control samples to account for the vehicle.

Image processing and migration quantification

Collected images were processed for z-stack maximum intensity projection, and stabilized with Stackreg, Fiji^44,45. Resulting time-lapsed 2D projections were processed for cells segmentation and migration tracking using TrackMate⁴⁶. Migration statistics were analyzed for cells speeds and confinement ratio distributions. Kolmogorov-Smirnov and Cohen’s D-value were used as statistical tests using Python libraries. For treatments, mean cells speeds were calculated for the 20% of the fastest cells per group and distributions were compared between groups.

Histological samples preparation

PCLS regions of interest were excised with 2–4 mm biopsy punches (Med-Vet International BP2MM, BP3MM, BP4MM), fixed in 4% formaldehyde (StatLab 28530–5) and 1x PBS (Fisher BioReagents, BP39920) for 24 hours, washed 3 times for 10 minutes with 1x PBS. Fixed PCLS tissue was stained with Hematoxylin Solution, Gill No. 2 (Leica 3801520), then positioned flat between two lens papers and embedded in paraffin. Uncultured tissue from control and IPF lungs were excised within 2 cm from the sliced regions and were fixed similarly (avoiding Hematoxylin staining). Uncultured IPF lungs normal-appearing parenchyma and patchy interstitial fibrosis were annotated by histopathologist using H&E staining. These regions were excised from FFPE blocks using 3 mm biopsy punches and organized in Tissue Microarrays (TMA), then used as uncultured controls for reported staining.

Immunofluorescent staining

FFPE blocks were cut into 5 μm sections, collected on positively charged glass slides. Sections were heated to 65 ⁰C for 10 minutes, deparaffinized in xylene and re-hydrated, then boiled in citrate buffer for antigen retrieval for 25 minutes, then blocked and permeabilized in 1x PBS with 5% bovine serum albumin (BSA, Fisher BioReagents BP9703100) and 0.1% Triton-X (Sigma-Aldrich X100–100ML) for 1 hour at room temperature. Samples then were washed 3 times for 5 minutes with 1x PBS plus 0.1% Tween-20 (Sigma-Aldrich P9416–100ML), incubated with primary antibodies in 1x PBS with 5% BSA at 4 ⁰C overnight, washed 4 times for 5 minutes in 1x PBS plus 0.1% Tween-20, incubated with secondary antibodies in PBS with 5% BSA at room temperature for 1 hour, washed 3 times in 1x PBS plus 0.1% Tween-20, incubated with DAPI for 30 minutes, washed 2 times in 1x PBS plus 0.1% Tween-20, washed 1 time in 1x PBS, coverslipped and imaged. Full list of antibodies and dilutions is provided in Extended Data Table 3. Secondary antibodies were always used at 1:500 dilution.

Immunofluorescent quantification

For quantification, slides were scanned on either Keyence BZ-X810 (20x magnification) or Olympus VS120 (40x magnification) scanner using 4 channels for DAPI, Alexa Flour 488, Alexa Flour 555 and Alexa Flour 647. QuPath 0.5.1 was used to segment the cells using DAPI nuclear signal⁴⁷. Resulted signal intensities were normalized to median tissue brightness for each sample. Proteins of interest were quantified as mean intensity per cell, per cytoplasm or per nucleus. Median of mean intensities within sample were quantified and tested between samples and groups using Kruskal-Wallis test. For YAP nuclear localization, ratio of mean nuclear intensity to mean cytoplasmic intensity was calculated for each cell, then the median of the ratio was calculated within samples and compared between groups using Kruskal-Wallis test.

KRT5 co-expression of scRNA-seq analysis

Single-cell RNA sequencing (scRNA-seq) datasets GSE135893 and GSE136831 were accessed through Gene Expression Omnibus. All analyses included IPF and control subjects only. Single cell counts were transformed using variance stabilizing transformation using sctransform R package⁴⁸. Genes were filtered out if expressed in 10% or less across all cells in the club/SCGB3A2+ cell population. We used a linear mixed model on KRT5 vst transformed counts adjusting for disease status and including a random effect for subject. To adjust for multiple comparisons across genes, we applied a Benjamini-Hochberg correction; genes with FDR < 0.05 were considered significant. Pathway analysis was performed using Ingenuity Pathway Analysis software on significant gene lists⁴⁹. Analysis was performed using Seurat version v4 and R version 4.3.0.

Extended Data

Extended Data Fig. 1. Regions of intense migration mapped to histological samples.

More examples of representative histology of regions of migration. Migratory regions are located in regions of progressed fibrosis of distal lung, but the degree of progression varies. Red or yellow rectangles represent approximate region of imaging. Red circles represent approximate regions of biopsy-punch excision. Scale bars from left to right: 100 μm, 500 μm, 500 μm, 100 μm.

Extended Data Fig. 2. AT1 cells are not detected in regions of intense EpCAM^pos migration.

Pro-SFTPC for AT2 cells (red), RAGE for AT1 cells (grey). Normal parenchyma of imaged control lung contains multiple AT1 cells. Region of migration of EpCAM^pos cells contain AT2 cells but lack AT1 cells.

Extended Data Fig. 3. Majority of IPF regions of migration lack canonical morphology of airway epithelial cells and SCGB1A1-positive cells.

a, H&E staining (zoomed out and zoomed in) demonstrates airway cells morphology in regions of migration (N=31 regions from 7 IPF lungs analyzed). Only 19% of the regions have canonical airway cells. Scale bars are 500 μm and 100 μm. b, SCGB1A1 protein is present in 12% regions of migration (N=25 regions from 6 IPF lungs analyzed). Scale bars are 250 μm and 20 μm.

Extended Data Fig. 4. KRT8^high cells are not present IPF regions of migration.

Quantification of KRT8 in AT2 cells. Regions of migration (N=26, 6 IPF lungs) do not have elevated amount of KRT8 in AT2 cells compared to uncultured control parenchyma (N=19, 7 lungs), IPF normal parenchyma (N=15, 7 lungs) and IPF patchy fibrotic regions (N=18, 7 lungs). Kruskal-Wallis test used. Scale bars are 20 μm.

Extended Data Fig. 5. KRT5^pos/HT2–280^pos cells are present in all IPF lungs and in none of control lungs.

HT2–280 (red), KRT5 (yellow) and KRT17 (green) staining of uncultured IPF lungs (a-d) and control unaffected lungs (e-h). All IPF lungs tested (N=7) contain KRT5^pos/HT2–280^pos cells. None of control lungs (N=7) have such cells. Panel D reflects airway-like cells that are KRT5-high and partially KRT17 positive (top) and KRT5pos/HT2–280pos cells. The latter have less KRT5 signal than airway cells. Zoomed out to zoomed in regions scale bars: 400 μm, 100 μm, 20 μm.

Extended Data Fig. 6. KRT5 co-expression in distal club cells reveals migration and IPF-related pathways.

a-b, Differentially expressed genes (DEGs) associated with KRT5 expression in SCGB3A2 positive cells from GSE135893 (114 DEGs) and club cells from GSE136831 (487 DEGs). KRT15, KRT17 and S100A2 are present in both lists matched with reported KRT5^pos AT2 (Supplementary Table 1 and 2)²⁶. c, SCGB3A2 expression in epithelial cell type highlights this gene expression in club cells in dataset GSE136831. d-f, Ingenuity Pathway Analysis (IPA) for DEGs presented in panel b (GSE136831). (d) Upstream regulators (for panel B DEGs) highlighting activation of β-catenin transcriptional activity, as well as multiple IPF-related effectors: TGFB1, TP53, TNF, IL1B, FOS. (e) Overview of the most significant biological processes (for panel B DEGs); multiple migration pathways upregulated (highlighted with red). (f) Key cell function pathways (for panel B DEGs) highlight high energy consumption, RHO GTPases regulation of actin remodeling, estrogen receptor signaling.

Extended Data Table 1.

List of lungs used for PCLS experiments with demographic information.

Disease status	Sex	Age	Ethnicity	Race	Smoking status
IPF	M	63	Non-Hispanic	Black	former smoker
IPF	M	65	Non-Hispanic	White	former smoker
IPF	F	69	Non-Hispanic	White	never smoker
IPF	M	70	Non-Hispanic	White	never smoker
IPF	M	66	Non-Hispanic	White	never smoker
IPF	M	65	Non-Hispanic	White	never smoker
IPF	M	68	Non-Hispanic	White	former smoker
Control	M	71	Non-Hispanic	White	never smoker
Control	M	46	Hispanic	White	never smoker
Control	F	59	Non-Hispanic	White	never smoker
Control	M	33	Non-Hispanic	Black	never smoker
Control	F	67	Non-Hispanic	White	never smoker
Control	M	62	Non-Hispanic	White	never smoker
Control	F	71	Hispanic	White	never smoker
Control	M	48	Non-Hispanic	White	never smoker

Extended Data Table 2.

Statistical tests of pulled migration data presented in Fig. 1c.

Comparison	Control parenchyma vs IPF parenchyma	Control parenchyma vs IPF fibrotic	IPF parenchyma vs IPF fibrotic
Speeds KS test p-value	< 1×10−10	< 1×10−10	< 1×10−10
Speeds D-value	0.13	0.200	0.102
Conf. ratio KS test p-value	< 1×10−9	< 1×10−10	< 1×10−10
Conf. ratio D-value	0.032	0.166	0.140

Extended Data Table 3.

Immunofluorescence antibodies list.

Protein	Manufacturer	Catalog number	Dilution	Secondary antibody manufacturer	Secondary antibody catalog
Gamma H2A.X (phosphor S139)	Abcam	ab2893	1:200	Invitrogen	A31573
HT2–280	Terrace Biotech	B-27AHT2–280	1:150	Invitrogen	A21042
Keratin 17	Invitrogen	MA1–06325	1:100	Invitrogen	A21141
Keratin 5	Biolegend	905504	1:100	Invitrogen	A31573
Keratin 8	Abcam	ab53280	1:1000	Invitrogen	A32731
Ki67	Abcam	ab15580	1:100	Invitrogen	A32731
Pro-SFTPC	Abcam	ab90716	1:500	Invitrogen	A31572
RAGE	R&D Systems	AF1145	1:100	Invitrogen	A21447
SCGB1A1	R&D Systems	MAB4218	1:250	Invitrogen	A21208
YAP	Sigma-Aldrich	MABS2029	1:100	Invitrogen	A48265
ZEB1	Invitrogen	14–9741-82	1:50	Invitrogen	A21241

Supplementary Material

Supplementary Table 1.

List of differentially expressed genes for KRT5 co-expression analysis in SCGB3A2 positive cells from GSE135893.

Supplementary Table 2.

List of differentially expressed genes for KRT5 co-expression analysis in club cells from GSE136831.

Supplementary Video 1.

Representative phenotype of EpCAM-positive cells migration in healthy and IPF lungs. 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary Video 2.

Same field of view of an I PF lung for comparison of EpCAM and Lysotracker migration. EpCAM-positive signal is in grey, Lysotracker signal is in green. 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary Video 3.

High magnification imaging of migrating cells in IPF lungs. EpCAM-positive signal is in magenta, Lysotracker signal is in green (with noticeable autofluorescence present). 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary Video 4.

Same field of view of an IPF lung for comparison of EpCAM and CD206 migration. EpCAM-positive signal is in grey, CD206 signal is in red. 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary Video 5.

Same field of view of an IPF lung for comparison of EpCAM and FolR2 migration. EpCAM-positive signal is in grey, FolR2 signal is in red. 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary Video 6.

Effect of β-Catenin on EpCAM^pos cells migration. Left: Control lung at baseline. Middle: same control lung treated with β-Catenin activator. Right: migratory region of an IPF lung. 1 second of video corresponds to 3 hours of time-lapsed imaging.

Supplementary information: Supplementary Tables 1 to 2; Supplementary videos 1 to 6. Supplementary information legends are provided at the end of this document.

Data availability statement

All data used in the analysis is supported by images, supplementary videos and tables.