Lung Remodeling Regions in Long-Term Coronavirus Disease 2019 Feature Basal Epithelial Cell Reprogramming

Abstract

Respiratory viruses, including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), can trigger chronic lung disease that persists and even progresses after expected clearance of infectious virus. To gain an understanding of this process, the current study examined a series of consecutive fatal cases of coronavirus disease 2019 (COVID-19) that came to autopsy at 27 to 51 days after hospital admission. In each patient, a stereotyped bronchiolar-alveolar pattern of lung remodeling was identified with basal epithelial cell hyperplasia, immune activation, and mucinous differentiation. Remodeling regions featured macrophage infiltration and apoptosis and a marked depletion of alveolar type 1 and 2 epithelial cells. This pattern closely resembled findings from an experimental model of post-viral lung disease that requires basal-epithelial stem cell growth, immune activation, and differentiation. Together, these results provide evidence of basal epithelial cell reprogramming in long-term COVID-19 and thereby yield a pathway for explaining and correcting lung dysfunction in this type of disease.

Graphical abstract

The coronavirus disease 2019 (COVID-19) pandemic spotlights the need to better understand both acute and chronic disease triggered by severe respiratory viral infection. The acute phase of disease with severe pneumonia and lung injury dominated the early management and study of the crisis.¹ However, even then, it appeared likely that progressive and often long-term disease was also a significant cause of morbidity and mortality. Indeed, a high percentage of patients with COVID-19 survived the acute infectious illness only to experience the major degree of organ dysfunction over a more prolonged time course during and after the initial hospitalization.^2,3 This outcome was consistent with previous observations that other types of respiratory viruses can also trigger a pathway to long-term immune-mediated disease.⁴ Thus, viral initiation, exacerbation, and progression of chronic lung disease can be found in clinical observations and corresponding experimental models of asthma, chronic obstructive pulmonary disease, and related inflammatory disease phenotypes.⁴ Together, these observations increase the likelihood that the host response to the virus can be reprogrammed from protection to an abnormal remodeling response in susceptible individuals. Whether and how this alternative pathway might also be linked to progressive and/or long-term COVID-19 still need to be defined.

To address this issue, lung tissue samples from autopsies of patients with COVID-19 were studied to obtain cell and molecular insights in the context of previous observations.⁴ The analysis focused on a consecutive series of patients with COVID-19 who died long after onset of illness (27 to 51 days after hospital admission) to provide a snapshot of long-term post-viral lung disease. In fact, this opportunity for analysis of lung tissue was largely unprecedented in human studies of this disease process. In that context, the study found consistent and striking basal-epithelial cell hyperplasia that extended beyond the usual airway location and instead moved into the distal airspaces of the lung. As introduced above, this remodeling process was captured long after the start of the initial illness but in close association with morbidity and mortality from the disease. Moreover, this same pattern of progression was predominant in experimental models of viral infection using natural pathogens for mice, such as Sendai virus,5, 6, 7 and for humans, such as influenza A virus and respiratory enterovirus-D68.^8,9 Those studies ultimately identified a subset of basal-epithelial stem cells (basal-ESCs) that are critical for homeostasis but also reprogrammable to become an essential driver of lung remodeling signatures.⁷ Herein, the study identified similar circuitry for lung remodeling in post-viral patients with COVID-19 with the possibility that the same components can be targeted to address long-term COVID-19 in the lung and perhaps other tissue sites.

Materials and Methods

Human Sample Procurement and Processing

Human lung tissue was obtained from a series of consecutive autopsies performed from April to August 2020 at Barnes-Jewish Hospital (St. Louis, MO). As nondisease controls, human lung samples were obtained from an Advanced Lung Disease Tissue Registry that contains whole lung explants harvested but not used for lung transplantation, as described previously,^10,11 and from a tissue procurement service (IIAM, Edison, NJ). A summary of clinical characteristics is provided in Supplemental Table S1. All human studies were conducted with protocols approved by the Washington University (St. Louis, MO) Institutional Review Board.

Tissue Staining, Immunohistochemistry, and Analysis

Tissues were fixed with 10% formalin, embedded in paraffin, cut into sections (5 μm thick), and adhered to charged slides. Sections were deparaffinized in Fisherbrand CitroSolv (Fisher Scientific, Hampton, NH; number 04-355-121), hydrated, and treated with heat-activated antigen unmasking solution (Vector Laboratories, Inc., Newark, CA; number H-3301-250 or H-3300-250). Tissue processing and then hematoxylin-eosin, periodic acid–Schiff–hematoxylin, and Gomori trichrome staining were performed, as described previously.^6,8 Immunostaining was performed using the following primary antibodies: rabbit anti–angiotensin-converting enzyme 2 (ACE-2) polyclonal antibody (pAb) (ab65863; Abcam, Cambridge, UK), mouse anti–ACE-2 monoclonal antibody (mAb) (clone 171606; R&D Systems, Minneapolis, MN), rabbit anti–keratin 5 (anti-KRT5) pAb (ab53121; Abcam) and mAb (clone EP1601Y; ab52635; Abcam), rabbit anti–aquaporin 3 (anti-AQP3) pAb (ab125219; Abcam), mouse anti-CD68 mAb (clone Kp-1; Sigma-Aldrich, St. Louis, MO), rabbit anti-CD163 mAb (clone D6U1J; Cell Signaling, Danvers, MA), rabbit anti-CD31 mAb (clone EPR17259; Abcam), rabbit anti–collagen IV mAb (clone EPR20966; Abcam), mouse anti–secretoglobin 1A1 (anti-SCGB1A1) mAb (clone E-11; Santa Cruz Biotechnology, Dallas, TX), mouse anti–acetylated tubulin (clone 6-11B-1; Sigma-Aldrich), mouse anti–mucin 5AC (anti-MUC5AC) mAb (clone 45M1; ThermoFisher Scientific, Waltham, MA; and Santa Cruz Biotechnology), mouse anti–HT2-280 pAb (TB-27AHT2-280; Terrace Biotech, San Francisco, CA), rabbit anti–surfactant protein C pAb (ab90716; Abcam), rabbit anti-podoplanin mAb (clone EPR7072; Abcam), rabbit anti–mucin 5B (anti-MUC5B) pAb (ab87276; Abcam), rabbit anti–Ki-67 mAb (clone D2H10; Cell Signaling), rabbit active (cleaved) caspase-3 mAb (clone 5 A1E; Cell Signaling), and mouse anti-CXCL17 mAb (clone 422204; R&D Systems). Antibody binding was detected with Alexa Fluor 488 or 594 conjugated secondary antibodies (ThermoFisher Scientific; numbers A-21206, A-21207, A-21202, and A-21203) matched to primary antibodies. All sections were counterstained with DAPI and were imaged using a DM5000 B microscope (Leica, Wetzlar, Germany) for conventional imaging. Staining was quantified in whole lung sections using a NanoZoomer S60 slide scanner (Hamamatsu, Shizuoka, Japan) and ImageJ software version 1.53v (NIH, Bethesda, MD; https://imagej.net/ij/index.html, last accessed January 7, 2023), as described previously.⁷ For the present experiments, an entire lung section was scanned and analyzed for each patient or subject (n = 5 per group) to assess lung remodeling region phenotype.

Statistical Analysis

All data are presented as means ± SEM and are representative of five patients with COVID-19 or five subjects without the disease. Unpaired t-test as well as mixed-model repeated measures analysis of variance with Tukey correction for multiple comparisons were used to assess statistical significance between means. In all cases, significance threshold was set at P < 0.05. The number of patients and subjects for each experimental condition is defined in the figure legends.

Results

To understand clinical observations of long-term disease after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, a protocol was developed to obtain and analyze the lung tissue samples from a consecutive set of hospitalized patients who underwent autopsy for COVID-19 as a primary cause of death. This group of five patients was highly relevant to long-term post-viral disease because they died at 27 to 51 days after initial presentation for care and positive testing for SARS-CoV-2 infection (Supplemental Table S1). Examination of hematoxylin-eosin staining of lung biopsy sections from all patients showed histopathologic changes of organizing diffuse alveolar damage (Figure 1A). In addition, this staining revealed a stereotyped pattern of airway epithelial cell hyperplasia that extended into distal airspaces that were filled with cellular and mucinous material (Figure 1A). Periodic acid–Schiff–hematoxylin staining confirmed these findings, including accumulation of mucus in these same locations (Figure 1A). Gomori trichrome staining indicated that these same bronchiolar-alveolar lung remodeling regions were also sites for collagen accumulation consistent with a fibrotic reaction (Figure 1A).

Figure 1.

Basal epithelial cell hyperplasia-metaplasia and mucinous differentiation accompanied by fibrosis and macrophage infiltration in lung remodeling regions in patients with COVID-19. A: Hematoxylin-eosin, periodic acid–Schiff (PAS)–hematoxylin, and Gomori trichrome staining of lung biopsy and biopsy sections from patients with COVID-19 and nondisease controls. For trichrome staining: collagen/mucus blue, cytoplasm/keratin/muscle red, and nuclei black. B: immunostaining for keratin 5 (KRT5) and aquaporin 3 (AQP3) with DAPI counterstaining in lung sections from patients with COVID-19 and nondisease control subjects. C: Representative white colorization of KRT5⁺ and whole tissue signals derived from image analysis using ImageJ software version 1.53v. D: Quantitation of staining derived from image analysis in B for conditions in C. E: Immunostaining for CD68 and CD163 with DAPI counterstaining in lung sections for conditions in B. F: Quantitation of staining derived from image analysis method in B for conditions in C. G: Immunostaining for CD68 plus collagen IV (CollIV) or CD31 with DAPI counterstaining in lung sections for conditions in B. Data are representative of five patients (at 27 to 51 days after hospital admission for COVID-19) and five control subjects per staining condition. Values represent means ± SEM (D and F). n = 5 patients or subjects per group (D and F). ∗P < 0.05. Scale bars: 1200 μm (A, left panels); 200 μm (A, middle and right panels, and G); 100 μm (B and E); 5 mm (C).

Immunostaining with basal epithelial cell markers KRT5 and AQP3⁷ more specifically identified the presence of basal-epithelial cell clusters extending into former alveolar spaces (Figure 1B). Quantitative levels of KRT5⁺ and AQP3⁺ staining were each significantly increased in patients with COVID-19 compared with subjects without disease (Figure 1, C and D). These remodeling sites (matched for size and morphology) also showed increased staining for monocyte-macrophage surface receptors CD68 and CD163 compared with nondisease controls (Figure 1, E and F). Accumulation of CD68⁺ macrophages was most prominent in remodeled airspaces defined by collagen IV⁺ basement membrane and CD31⁺ vascular borders (Figure 1G). These results served as an index of basal-epithelial cell hyperplasia-metaplasia and monocyte-macrophage infiltration to mark the extent of lung remodeling regions in patients with COVID-19.

The spatial distribution and morphology of alveolar type 2 (AT2) and type 1 (AT1) cells during late-stage disease were examined to define the observed damage to the alveolar epithelium. As expected, immunostaining showed that KRT5⁺ basal cells were restricted to airway mucosal sites, and HT2-280⁺ surfactant protein C⁺ AT2 cells and podoplanin⁺ AT1 cells were confined to alveolar sites in nondisease control subjects (Figure 2, A and B). In contrast, KRT5⁺ cells were abundant, whereas HT2-280⁺, surfactant protein C⁺, and podoplanin⁺ cells were nearly absent, in the lung remodeling regions of patients with COVID-19 (Figure 2, A and B). Expression of the SARS-CoV-2 receptor ACE-2 was localized to KRT5⁺ basal epithelial cells in airway mucosa and remodeling regions, with rare expression on α-tubulin⁺ ciliated cells and no significant expression on HT2-280⁺ AT2 cells, SCGB1A1⁺ club cells, or CD68⁺ macrophages (data not shown) (Figure 2C). These findings suggested viral capacity for primary infection of basal epithelial cells in proximal airway and distal lung remodeling sites.

Figure 2.

Alveolar epithelial cell loss and viral receptor expression in lung remodeling regions in COVID-19. A: Immunostaining for keratin 5 (KRT5) plus HT2-280, surfactant protein C (SFTPC), and podoplanin (PDPN) with DAPI counterstaining in lung sections from patients with COVID-19 and nondisease controls. B: Quantitation of staining for conditions in A. C: Immunostaining for angiotensin-converting enzyme 2 (ACE-2) alone and with KRT5, HT2-280, secretoglobin 1A1 (SCGB1A1), or α-tubulin with DAPI counterstaining in lung sections for conditions in C; immunostaining with rabbit anti–ACE-2 polycloncal antibody for serial sections and mouse anti–ACE-2 monoclonal antibody for costaining sections. White dashed-line boxed area indicates ACE-2⁺ α-tubulin⁺ cells. Data are representative of five patients and five control subjects per staining condition. Values represent means ± SEM (B). n = 5 patients or subjects per group (B). ∗P < 0.05. Scale bars: 100 μm (A); 50 μm (C).

As a further readout of post-viral lung remodeling, mucinous differentiation, as defined by expression of the predominant lung gel-forming mucins MUC5AC and MUC5B, was characterized. Nondisease control subjects had the usual pattern of mixed MUC5AC⁺ and MUC5B⁺ staining in airway mucosal epithelium, predominant MUC5B⁺ staining in submucosal glands, and no detectable MUC5AC⁺ or MUC5B⁺ staining in alveolar epithelium (Figure 3A). As noted previously,⁶ nondisease control subjects (who were on mechanical ventilation, similar to patients) exhibited variable amounts of airway mucus staining, but none of the controls showed lung remodeling regions as found in patients with COVID-19. In contrast, mixed MUC5AC⁺ and MUC5B⁺ staining that was generally colocalized to mucous cells in lung remodeling regions was observed for each of the patients with COVID-19 (Figure 3B). Quantitation of mucin staining showed a significant increase in MUC5AC⁺ and MUC5B⁺ staining at these sites compared with comparable bronchiolar-alveolar sites in nondisease controls (Figure 3C).

Figure 3.

Mucin (MUC)5AC with MUC5B expression in lung remodeling regions in COVID-19. A: Immunostaining for MUC5AC plus MUC5B with DAPI counterstaining in lung sections from nondisease control subjects. B: Immunostaining for MUC5AC plus MUC5B with DAPI counterstaining in lung sections from patients with COVID-19. C: Quantitation of staining for conditions in B. Data are representative of five patients per staining condition. Values represent means ± SEM (C). n = 5 patients or subjects per group (C). ∗P < 0.05. Scale bars = 1 mm (A and B). Original magnification: ×5 (A, insets); ×3 (B, insets).

The study also aimed to better define the net increases in monocytes-macrophages as an index of epithelial and immune cell activation. A slight but significant increase in Ki-67⁺ cells in patients with COVID-19 was observed as a sign of cell proliferation, but this growth signal was not seen in the HT2-280⁺ AT2 cell or CD68⁺ macrophage population in patients with COVID-19 or nondisease controls (data not shown) (Figure 4, A and B). In contrast, a marked increase in active caspase-3⁺ cells as a marker of programmed cell death (apoptosis) was observed in patients with COVID-19 compared with that in controls (Figure 4, C and D). Moreover, the active caspase-3 signal was not localized to the decreased population of HT2-280⁺ AT2 cells but instead was colocalized primarily to the increased population of CD68⁺ lung macrophages (Figure 4, E and F). This combination of results suggested that macrophage accumulation resulted from chemokine-dependent migration into the remodeling sites. In that regard, a marked increase was observed in chemokine CXCL17⁺ cells that colocalized at least in part to KRT5⁺ basal epithelial cells in remodeling regions in patients with COVID-19 (Figure 4, G and H). Cytoplasmic staining for CXCL17 was consistent with localization to the secretory pathway. Together, these findings were consistent with monocyte-macrophage accumulation based on tissue infiltration despite an increased level of programmed cell death even at this late stage of post-viral lung disease. This recruitment was associated with immune activation of basal epithelial cells for production of macrophage chemokine CXCL17.

Figure 4.

Cell proliferation and apoptosis in lung remodeling regions in COVID-19. A: Immunostaining for Ki-67 plus CD68 with DAPI counterstaining in lung sections from patients with COVID-19 and nondisease controls (NDs). B: Quantitation of staining for conditions in A. C: Immunostaining for active caspase-3 with DAPI counterstaining in lung sections for conditions in A. D: Quantitation of staining for conditions in C. E: Immunostaining for active caspase-3 plus HT2-280 or CD68 with DAPI counterstaining for conditions in A. F: Quantitation of staining for conditions in E. G: Immunostaining for keratin 5 (KRT5) plus CXCL17 with DAPI counterstaining for conditions in A. H: Quantitation of staining for conditions in G. Data are representative of five patients and five control subjects per staining condition. Values represent means ± SEM (B, D, F, and H). n = 5 patients or subjects per group (B, D, F, and H). ∗P < 0.05. Scale bars: 100 μm (A and C); 50 μm (E and G). Original magnification: ×4 (A, insets); ×3 (E and G, insets).

Discussion

This study analyzed a series of autopsies from patients with COVID-19 with late-stage lung disease at least 27 to 51 days after initial viral infection. The results reveal a bronchiolar-alveolar lung remodeling process that is characterized by the following: i) basal epithelial cell hyperplasia and metaplasia with extension into former alveolar spaces; ii) colocalized depletion of alveolar types of epithelial cells normally found in these spaces; iii) epithelial stem–progenitor cell differentiation to mucous cells with mucins characteristic of mucosal and submucosal locations; and iv) basal epithelial cell production of chemokine CXCL17 associated with lung monocyte-macrophage infiltration. This study presents these findings in the context of human clinical studies and comparable animal models.

Data on post-viral lung remodeling in humans are limited. The COVID-19 pandemic generated an unprecedented opportunity to define this process in vivo. Initial data from patients with COVID-19 documented the early phase of diffuse alveolar damage typical of lung injury and acute respiratory distress syndrome.12, 13, 14 Thereafter, basal-epithelial cell repair,^15,16 AT2 cell loss,¹⁷ macrophage accumulation,^18,19 predominant MUC5AC⁺ mucus production,²⁰ and selective MUC5B⁺ microcysts²¹ were reported at subsequent stages of disease in lungs of patients with COVID-19. The present series of entirely long-term cases was further characterized by the consistent development of lung remodeling regions with KRT5⁺-AQP3⁺ basal-epithelial cell hyperplasia-metaplasia, combined AT1 and AT2 cell dropout, mixed MUC5AC⁺ and MUC5B⁺ mucus production, and CD68⁺CD163⁺ monocyte-macrophage accumulation, despite accelerated caspase-3⁺ apoptotic turnover. Together, the data provide histologic, cellular, and molecular definition of the switch from acute lung injury to chronic bronchiolization that was suggested in descriptive reports of autopsies after presumed viral infection²² and basal cell growth derived from single-cell RNA-sequencing analysis of COVID-19 samples.^16,23, 24, 25, 26, 27 The present finding of viral receptor ACE-2 on basal epithelial cells suggests that this cell population might be a direct target for reprogramming toward long-term growth, immune activation, and mucinous differentiation.

These new findings also enable comparison of the present data to animal models to better define the pathogenesis of post-viral lung disease. At present, experimental models of SARS-CoV-2 infection do not yet provide a model of long-term lung disease with high fidelity to the human phenotype noted above. This discrepancy is likely a reflection of the inability to duplicate the usual pattern of severe respiratory infection. In that regard, the usual inhaled mechanism of spread (extending from upper airway to alveolar sites) and the consequent severity of infection can be achieved with the natural Sendai virus pathogen in mice. This type of infection thereby yields a remarkably similar pattern of acute infectious illness and, in turn, manifests the key signatures of post-viral lung remodeling disease found in COVID-19.^7,9,28 This experimental model also reveals a basal-ESC subset that jumps the usual bronchiolar-alveolar boundary and grows into distal airspaces as a new site for differentiation and immune activation that is required for long-term dysfunction.⁷ Single-cell analysis marks this cell subset and its descendant lineage with macrophage chemokine Cxcl17 expression in mice⁷ and humans (https://www.proteinatlas.org, last accessed September 6, 2022). Furthermore, CXCL17 mRNA is increased markedly in bronchoalveolar lavage samples from patients with COVID-19.²⁹ Herein, CXCL17 was induced in basal epithelial cells in long-term disease, implicating immune activation of basal-ESCs and a possible effector signal for monocyte-derived macrophage and dendritic cell participation. This participation is essential for long-term disease in the Sendai virus mouse model^5,28,30,31 and is consistent with monocyte-macrophage infiltration found in remodeling regions of patients with long-term COVID-19 here and in previous reports using single-cell RNA sequencing.^26,27 A similar post-viral immune process can be detected after severe infection with influenza A virus in mice,⁸ which can also manifest basal cell hyperplasia-metaplasia.32, 33, 34, 35 Furthermore, post-viral lung disease is found even in milder infection with enterovirus (respiratory enterovirus-D68) in mice engineered for deficiency in airway epithelial interferon signaling.⁹ A similar type 1 interferon deficiency is associated with increased severity in COVID-19,³⁶ consistent with observations that severity of illness is linked to long-term disease in mouse models and humans.^3,8,9

The present findings also provide additional insights in comparison to previous studies of AT2 cells. For example, AT2–basal cell trans-differentiation was proposed as a mechanism for basal cell hyperplasia/metaplasia based on human organoid co-culture and mouse xenotransplantation models of pulmonary fibrosis.³⁴ However, lineage tracing shows that basal cell hyperplasia/metaplasia derives from Krt5-expressing basal cells in the Sendai virus mouse model of post-viral lung disease that resembles long-term COVID-19.⁷ Moreover, AT2-basal intermediates (eg, Krt8-Sftpc–expressing cells) are not detected in the post-viral mouse model, and AT2 cell markers are not found in lung remodeling regions in humans with COVID-19. In a related point, AT2 cells were not a primary site for viral receptor ACE-2 expression using mAb and pAb detection to account for variation in epitope recognition. This result is similar to expression on a relatively rare subset of AT2 cells in other studies of human lung tissue.^17,37, 38, 39 These data also highlight the distinctive nature of the basal-epithelial versus AT2 cell lineage in the setting of respiratory viral infection.

Together, the present data for patients with COVID-19 fit a consistent paradigm for long-term post-viral lung disease found in natural and experimental settings. Our perspective is that epithelial barriers maintain a basal-ESC subset that is tasked for repair and regeneration after injury, including the damage from severe viral infection. However, this same system can orchestrate a tissue remodeling process that is pathogenic because basal epithelial cell depletion can attenuate post-viral disease in the lung and disrupt homeostasis at other barrier sites.⁷ The presently identified markers for the relevant basal-ESC subset thereby suggest an approach to disease modification via correction of basal-ESC reprogramming. Suitable targets already exist in the form of IL-33–nucleokine and mitogen-activated protein kinase 13–stress kinase activation that allow for basal-ESC growth and mucinous differentiation in post-viral lung remodeling disease^7,40 and perhaps other virus-triggered lung diseases.⁴ The present study provides a basis to define these and related strategies in lung remodeling disease due to COVID-19.

Author Contributions

K.W. designed and performed experiments; H.Y.-D. performed experiments; Y.Z. analyzed data; S.R.A. and D.E.B. obtained and analyzed clinical data; E.C.C. identified autopsy cases and assisted with histopathologic analysis; and M.J.H. directed the project and wrote the manuscript. All authors acquired and analyzed data.