Proximal immune-epithelial progenitor interactions drive chronic tissue sequelae post COVID-19

The long-term physiological consequences of SARS-CoV-2, termed Post-Acute Sequelae of COVID-19 (PASC), are rapidly evolving into a major public health concern. The underlying cellular and molecular etiology remain poorly defined but growing evidence links PASC to abnormal immune responses and/or poor organ recovery post-infection. Yet, the precise mechanisms driving non-resolving inflammation and impaired tissue repair in the context of PASC remain unclear. With insights from three independent clinical cohorts of PASC patients with abnormal lung function and/or viral infection-mediated pulmonary fibrosis, we established a clinically relevant mouse model of post-viral lung sequelae to investigate the pathophysiology of respiratory PASC. By employing a combination of spatial transcriptomics and imaging, we identified dysregulated proximal interactions between immune cells and epithelial progenitors unique to the fibroproliferation in respiratory PASC but not acute COVID-19 or idiopathic pulmonary fibrosis (IPF). Specifically, we found a central role for lung-resident CD8+ T cell-macrophage interactions in maintaining Krt8hi transitional and ectopic Krt5+ basal cell progenitors, thus impairing alveolar regeneration and driving fibrotic sequelae after acute viral pneumonia. Mechanistically, CD8+ T cell derived IFN-γ and TNF stimulated lung macrophages to chronically release IL-1β, resulting in the abnormal accumulation of dysplastic epithelial progenitors and fibrosis. Notably, therapeutic neutralization of IFN-γ and TNF, or IL-1β after the resolution of acute infection resulted in markedly improved alveolar regeneration and restoration of pulmonary function. Together, our findings implicate a dysregulated immune-epithelial progenitor niche in driving respiratory PASC. Moreover, in contrast to other approaches requiring early intervention, we highlight therapeutic strategies to rescue fibrotic disease in the aftermath of respiratory viral infections, addressing the current unmet need in the clinical management of PASC and post-viral disease.


INTRODUCTION
SARS-CoV-2 infection can lead to long-term pulmonary and extrapulmonary symptoms well beyond the resolution of acute disease, a condition collectively termed post-acute sequelae of SARS-CoV-2 (PASC) (1,2).With effective treatment strategies and vaccines to tackle acute COVID-19, the emerging challenge is to manage chronic sequelae in the 60+ million people currently experiencing PASC (3,4).Given the extensive damage to the respiratory tract during primary infection, the lungs are particularly susceptible to sustained impairments including dyspnea, compromised lung function, and radiological abnormalities which persist up to 2 years post infection in contrast to the majority of extrapulmonary sequelae (1,2,4).Some individuals also develop a non-resolving fibroproliferative response -PASC pulmonary fibrosis (PASC-PF) and typically require persistent oxygen supplementation and eventual lung transplantation (4)(5)(6)(7)(8).
Currently, mechanisms underlying the maintenance of these dysplastic progenitors and their contributions to post-viral respiratory sequelae remain largely elusive.By comparing the pathological, immunological, and molecular features of human respiratory PASC and mouse models of post-viral lung sequelae, here we have discovered that spatially defined interactions among CD8 + T cells, macrophages, and epithelial progenitors drive chronic tissue sequelae after acute viral injury.Furthermore, we identify nodes for therapeutic intervention, which may be adopted in the clinic to mitigate chronic pulmonary sequelae after COVID-19.

Ethics and biosafety
All aspects of this study were approved by the Institutional Review Board Committee at Cedars-Sinai Medical Center (IRB# Pro00035409) and the University of Virginia (IRB# 13166).Work related to SARS-CoV-2 was performed in animal biosafety level 3 (ABSL-3) facilities at the University of Virginia and influenza-related experiments were performed in animal biosafety level 2 (ABSL-2) facilities at the Mayo Clinic and the University of Virginia.

Cells, viruses, and mice
African green monkey kidney cell line Vero E6 (ATCC CRL-1587) were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), along with 1% of penicillin-streptomycin (P/S) and L-glutamine at 37°C in 5% CO 2 .The SARS-CoV-2 mouse-adapted strain (MA-10) was kindly provided by Dr. Barbara J Mann (University of Virginia School of Medicine).The virus was passaged in Vero E6 cells, and the titer was determined by plaque assay using Vero E6 cells.
Aged mice were received at 20 to 21 months of age from the National Institutes of Aging and all mice were maintained in the facility for at least 1 month before infection.All mice were housed in a specific pathogen-free environment and used under conditions fully reviewed and approved by the institutional animal care and use committee guidelines at the Mayo Clinic (Rochester, MN) and University of Virginia (Charlottesville, VA).
For primary influenza virus infection, influenza A/PR8/34 strain [75 plaque-forming units (PFU) per mouse] was diluted in fetal bovine serum (FBS)-free Dulbecco's modified Eagle's medium (DMEM) (Corning) on ice and inoculated in anesthetized mice through intranasal route as described previously (34).For SARS-CoV-2 infections, mice were infected with 5 x 10 4 PFU for C57/BL6 and 1000PFU for BALB/c virus of mouse-adapted (MA-10) intranasally under anesthesia as described previously.Infected mice were monitored daily for weight loss and clinical signs of disease for 2 weeks, following once a week for the duration of the experiments.
The mortality rate of mice calculated as "dead" were either found dead in cage or were euthanized as mice reached 70% of their starting body weight which is the defined humane endpoint in accordance with the respective institutional animal protocols.At the designated endpoint, mice were humanely euthanized by ketamine/xylazine overdose and subsequent cervical dislocation.

Evaluation of respiratory mechanics and lung function
Lung function measurements using FOT and the resulting parameters have been previously described (35).In brief, animals were anesthetized with an overdose of ketamine/xylazine (100 and 10mg/kg intraperitoneally) and tracheostomized with a blunt 18-gauge canula (typical resistance of 0.18 cmH 2 O s/mL), which was secured in place with a nylon suture.Animals were connected to the computer-controlled piston (SCIREQ flexiVent), and forced oscillation mechanics were performed under tidal breathing conditions described in (35) with a positive-end expiratory pressure of 3 cm H 2 O.The measurements were repeated following thorough recruitment of closed airways (two maneuvers rapidly delivering TLC of air and sustaining the required pressure for several seconds, mimicking holding of a deep breath).Each animal's basal conditions were normalized to their own maximal capacity.Measurement of these parameters before and after lung inflation allows for determination of large and small airway dysfunction under tidal (baseline) breathing conditions.Only measurements that satisfied the constant-phase model fits were used (>90% threshold determined by software).After this procedure, mice had a heart rate of ~60 beats per minute, indicating that measurements were done on live individuals.

Human lung tissue specimens
Human lung samples were obtained from patients enrolled in the IRB-approved Lung Institute BioBank (LIBB) study at Cedars-Sinai Medical Center, Los Angeles, CA.All participants or their legal representatives provided informed written consent.Lung tissues were processed within 24 hours after surgical removal.Specifically, the lung tissues were cut and immediately fixed in 10% normal-buffered formalin for 24 hours before tissue processing using the HistoCore PEARL -Tissue Processor, Leica Biosystem, Deer Park, IL, and embedded in paraffin for histological studies.The formalin-fixed paraffin-embedded cassettes were properly stored at room temperature until further sectioning.

Mouse tissue processing and flow cytometric analysis:
Animals were injected intravenously with 4 μg of CD45 or 2 μg of CD90.2 Ab labeled with various fluorochromes.Two minutes after injection, animals were euthanized with an overdose of ketamine/xylazine and processed 3 min later.After euthanasia, the right ventricle of the heart was gently perfused with chilled 1X PBS (10 mL).Right lobes of the lungs were collected in 5 mL of digestion buffer (90% DMEM and 10% PBS and calcium with type 2 collagenase (180 U/ml) (Worthington) and DNase (15 μg/ml) (Sigma-Aldrich) additives).Tissues were digested at 37°C for 1 hour followed by disruption using gentleMACS tissue dissociator (Miltenyi).Singlecell suspension were obtained by hypotonic lysis of red blood cells in ammonium-chloridepotassium buffer and filtration through a 70m mesh.Cells were washed with FACS buffer (2% of FBS and 0.1% of NaN 3 in PBS) and FC- receptors were blocked with anti-CD16/32 (2.4G2).
Surface staining was performed by antibody (details provided in Supplementary Table 2) incubation for 30 min at 4°C in the dark.After PBS wash, cells were resuspended with Zombiedye (Biolegend) and incubated at RT for 15 min.For IL-1β staining, cells were incubated with monesine (Biolegend) for 5 hours at 37°C and then stained with surface markers antibodies.
After washing with FACS buffer, cells were fixed with fixation buffer (Biolegend) and permeabilized with intracellular staining permeabilization wash buffer (Biolegend).The cells were then stained with anti-IL-1β at RT for 1hr and samples were acquired on an Attune NxT (Life Technologies).The data were analyzed with FlowJo software (Tree Star).

Caspase 1 FLICA analysis
Caspase 1 was detected by FAM-FLICA Caspase-1 Assay Kit (ImmunoChemistry Technologies) according to the manufacturer's instructions.Briefly, lung single cells or macrophages from in vitro culture were stained with fluorochrome-conjugated Ab cocktail for cell surface markers.After staining, cells were incubated with FLICA for 30min at 37°C, washed and detected by flow cytometry.

Cell isolation and ex vivo co-culture
To isolate myeloid and CD8 + T cells, single cell suspension of the lung was generated from influenza infected mice as described above and labeled and enriched with CD11c and CD11b or CD8 microbeads (Miltenyi Biotec, Auburn, CA) according to the manufacturer's instructions.
Purified myeloid cells were seeded in 96-well (200,000 per well) plates and incubated for 2 hours at 37°C 5% CO 2 to facilitate attachment.Wells were washed with 1X PBS to select for macrophages, following which selected wells were seeded with naïve or memory CD8 + T cells (40,000 per well) and further incubated for 16-18 hours.Supernatant and cell pellets were collected for cytokine measurement and gene expression analyses, respectively.

AT2 cell isolation and culture
AT2 cells were isolated from naïve mice as previously described (26,36,37).Briefly, mouse lungs were perfused with chilled PBS and intratracheally instilled with 1mL of dispase II (15U/mL, Roche), tying off the trachea and cutting away the lobes from the mainstem bronchi.
Lungs were incubated in 4mL of 15U/mL dispase II for 45min while shaking at room temperature, followed by mechanical dissociation with an 18G needle.Following passage through a 100μm filter, lungs underwent 10min of DNase I digestion (50μg/mL) and filtered through 70μm filter prior to RBC lysis.Single-cell suspensions were subject to CD45 depletion using microbeads (Miltenyi), incubated with anti-FcgRIII/II (Fc block) and stained with CD45, EpCAM, MHC-II, and viability die (see antibody details in Supplementary Table 2).
Fluorescence assisted cell sorting was performed on the BD Influx cell sorter to isolate AT2 cells as described previously (36) and collected in 500μL DMEM + 40% FBS + 2% P/S.Sorted AT2 cells (2x10 5 /well) were plated in a 96-well plate in DMEM/F12 + 10% FBS and cultured at 37°C, 5% CO 2 for 3 days prior to harvest.

RNA isolation and real time-quantitative polymerase chain reaction (RT-qpCR)
Cells were lysed in Buffer RLT and RNA was purified using the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer's instructions.Random primers (Invitrogen) and MMLV reverse transcriptase (Invitrogen) were used to synthesize first-strand complementary DNAs (cDNAs) from equivalent amounts of RNA from each sample.cDNA was used for real-time PCR with Fast SYBR Green PCR Master Mix (Applied Biosystems).Real-time PCR was conducted on QuantStudio 3 (Applied Bioscience).Data were generated with the comparative threshold cycle (ΔCt) method by normalizing to hypoxanthine-guanine phosphoribosyltransferase (HPRT) transcripts in each sample as reported previously (38).

COVID-19 convalescents cohort
Peripheral blood specimens were obtained from patients presenting to the University of Virginia Post-COVID-19 clinic.All participants provided written informed consent.Pulmonary function testing (PFT: spirometry, lung volume testing, diffusing capacity of the lungs for carbon monoxide [DLCO]) at the time of blood draws was used to define normal and abnormal lung function.

IL-1β cytokine evaluation
Plasma from COVID-19 convalescents, supernatant from ex vivo culture and murine BAL fluid following flushing of airway 3X with 600uL of sterile PBS were used to quantify IL-1β levels by ELISA as per manufacturer's instructions (R&D systems).Samples were first concentrated 5X (BAL) and 2X (cell culture supernatant) using Microcon-10kDa centrifugal filters (Millipore Sigma).

Immunofluorescence
Mouse lung tissues were routinely perfused with ice cold 1X PBS, fixed with 10% formalin and embedded in paraffin.Lung tissue sections (5μm) were deparaffinized in xylene and rehydrated.Heat-induced antigen retrieval was performed using 1X Agilent Dako target retrieval solution (pH 9) in a steamer for 20min (mouse lungs) or 45min (human lungs), followed by blocking and surface staining.For intracellular targets, tissues were permeabilized with 0.5% Triton-X 0.05% Tween20 for 1 hour at room temperature.Sections were stained with primary antibodies as listed in Supplementary Table 2 overnight at 4°C.Subsequently, samples were washed and incubated with fluorescent secondary antibodies as listed in Supplementary Table 2 for 2 hours at room temperature.Sections were counterstained with DAPI (1:1000, ThermoFisher Scientific) for 3 minutes and mounted using ProLong Diamond Antifade mountant (ThermoFisher Scientific).After 24 hours of curing at room temperature, images were acquired using the Olympus BX63 fluorescent microscope and pseudocolours were assigned for visualization.For each lung section, images were taken in at least 10-12 random areas in the distal lung.All images were further processed using ImageJ Fiji, OlyVIA, and/or QuPath software.

Spatial Transcriptomics analyses
Spatial transcriptomics (ST) data (generated from 10X Visium platform) was pre-processed using the spaceranger package (v2.0, genome version mm10).The Rpackage Seurat (v4.3.0) was used for quality control (QC) and preliminary analysis.Only high-quality spots with sufficient gene coverage (>=2000) were retained for downstream analysis.Like scRNA-seq data, the spatial expression data was first normalized to a log scale using the SCTransform method (v0.3.5).Top 2000 highly variable genes were then identified based on the variance of expression across spots for the principal component analysis (PCA) input.UMAP embeddings were generated for visualization at the reduced dimensionality (top 30 PCs).Spots were clustered with a shared nearest neighbor (SNN) modularity optimization-based clustering algorithm (built in Seurat).Samples were integrated and batch-effect-removed using the Harmony package (v0.1).In addition to the spatial spot view of the expression pattern, we used UMAP to visualize the expression pattern of spots associated with the identified cell types in a lower dimension using the visualization pipeline in the Seurat package.We determined each spot's potential cell type composition based on the following protocol: 1.We defined AE spots based on the k-means clustering analysis (using knowledge based key genes from different cell types).2. For non-AE spots, we used the average expression of Krt5, Krt8, Krt17, Cldn4, and Trp53 to estimate a Krt score, and similarly, average expression of Cd8a, Cd8b1, Itgae, Trbc1, Trbc2, Sell, Ccl5, Cd69, and Cd3d for Cd8 score.We set -0.4 as cutoff for both the Cd8 and Krt score: spots with Krt/Cd8 score >-0.4 are assigned as Krt/Cd8 associated spot, and assigned blue/red color, respectively in the spatial map.In this way, a spot can be Krt spot and Cd8 spot at the same time, and we use purple color to represent this category.For a given gene list, captured the expression of relevant genes in Krt spots and AE spots and compared the distribution of the expression index.The significance of the difference was estimated by Wilcoxon rank sum test.The p-values were then log 10 transformed and displayed as bar plot.

Statistical analyses
Quantitative data are presented as means ± SEM.Unpaired two-tailed Student's t test (twotailed, unequal variance) was used to determine statistical significance with Prism software (GraphPad) for two-group comparison.For multiple groups, analysis of variance (ANOVA) corrected for multiple comparisons was used when appropriate (GraphPad).Log-rank (Mantel-Cox) test was used for survival curve comparison and multiple t test was used to analyze differences in weight loss.We considered P < 0.05 as significant in all statistical tests and denoted within figures as a * for each order of magnitude P value.

Spatial association between dysplastic epithelial progenitors and CD8 + T cells is a hallmark of human PASC-PF.
We examined diseased lung sections from a cohort of PASC-PF patients that underwent lung transplantation at Cedars-Sinai Medical Center.Patients had a mean age of 52.5 years and exhibited persistent pulmonary impairment and hypoxemia, requiring oxygen supplementation (Supplementary Table 1).Lung histology was notable for extensive immune cell infiltration and collagen deposition in the alveolar epithelium (Fig. 1a).Consistent with the observed fibrotic sequelae, we found reduced levels of AT1 and AT2 cells in PASC-PF lungs compared to controls, suggesting a persistent defect in alveolar regeneration (Fig. 1b,c) (39).We also observed chronic persistence of ectopic Krt5 + basal cells and Krt8 hi transitional cells in PASC-PF lungs compared to controls, which is concordant with recent reports (5, 8, 30, 40) (Fig. 1d,e).
PASC-PF lungs also harbored widespread expression of alpha smooth muscle actin (αSMA), indicative of myofibroblast activity in addition to pockets of Krt5 -Krt17 + aberrant basaloid cells previously found in IPF lungs (Extended data Fig. 1a) (41,42).Therefore, PASC-PF is characterized by the sustained loss of functional alveolar epithelial cells, and the persistence of dysplastic Krt5 + pods and Krt8 hi transitional cells, which is histologically akin to other fibrotic lung diseases such as idiopathic pulmonary fibrosis (IPF) (28,29).
Previously, we reported that increased CD8 + T cell levels in the bronchoalveolar lavage (BAL) fluid were associated with impaired lung function of COVID-19 convalescents (9,10).In accordance, CD8 + T cell numbers were elevated in PASC-PF patient lungs (Fig. 1f).Moreover, CD8 + T cell abundance was significantly higher in acute COVID-19 but not in IPF lungs when compared to controls (Fig. 1g,h).Interestingly, a striking spatial association was observed between CD8 + T cells, and Krt8 hi and Krt5 + areas representing dysplastic repair upon analysis of their distribution in PASC-PF lungs (Fig. 1i,j Extended data Fig. 1b,c,h).Notably, this correlation between CD8 + T cells, and Krt8 hi and Krt5 + dysplastic areas was unique to PASC-PF lungs but not seen in lungs from control, acute COVID-19 or IPF conditions (Fig. 1k-m, Extended data Fig. 1d-g,i,j).Collectively, these data indicate that the spatiotemporal colocalization of CD8 + T cells and areas of dysplastic repair is a unique feature of post-viral pulmonary fibrosis and supports immune-epithelial progenitor interactions potentially contributing to the observed defects in alveolar regeneration and chronic pulmonary sequelae.

A mouse model of post-viral lung sequelae recapitulating features of human PASC-PF.
To investigate the role of immune-epithelial progenitor interactions, we aimed to develop a mouse model to capture the cellular and pathological features observed in PASC-PF lungs.We used a mouse-adapted (MA-10) strain of the SARS-CoV-2 virus to productively infect WT mice.
Notably, SARS-CoV-2 MA-10 infection is known to induce acute lung disease and pneumonia in mice, characterized by substantial damage to the airway epithelium, fibrin deposition, and pulmonary edema (43).Since aging is associated with an increased propensity to develop lung fibrosis post viral injury as well as severe disease after SARS-CoV-2 infection in mice, we included both young and aged mice in our study (34,44,45).As expected, aged C57BL/6 mice infected with SARS-CoV-2 MA-10 had increased morbidity and mortality compared to young mice (Fig. 2a, Extended data Fig. 2a).Indeed, we observed marked inflammation and tissue damage acutely (at 10 days post infection (dpi)) (Extended data Fig. 2b) (43,46).However, irrespective of age, the majority of the lungs recovered from the acute damage and only moderate pathology primarily restricted to the subpleural regions was observed at the chronic phase (35dpi) of infection (Fig. 2b).
Next, we tested SARS-CoV-2 MA-10 infection in aged BALB/c mice, which was previously reported to induce more robust inflammation and tissue sequelae compared to C57BL/6 mice (44) (Fig. 2c, Extended data Fig. 2e).Consistent with the report, we observed persistent immune cell infiltration, which was largely restricted to the peri-bronchiolar regions at 35dpi (Fig. 2d).However, similar to the aged C57BL/6 mice, minimal signs of collagen deposition were observed at later time points in the alveolar epithelium (Fig. 2d).Consistent with what was previously reported (44) , BALB/c mice failed to maintain pulmonary CD8 + T cells (Extended data Fig. 2c,f), a prominent feature of human respiratory sequelae (Fig. 1d).Moreover, no significant difference was observed in the development and persistence of Krt8 hi transitional cells between aged naïve and infected (35dpi) mice in both genetic backgrounds in spite of substantial alveolar damage during acute disease (Fig. 2e,f, Extended data Fig. 2g-i).Thus, SARS-CoV-2 infection in these two mouse models failed to recapitulate key features of tissue pathology and dysplastic lung repair observed in human PASC-PF.
Previously, we and others reported persistent lung inflammation and tissue pathology after influenza viral infection, particularly in aged C57BL/6 mice (9,34,47).Therefore, we infected young and aged WT C57BL/6 mice with influenza H1N1 A/PR8/34 strain, which causes substantial viral pneumonia and alveolar damage during acute disease (34,48).Similar to SARS-CoV-2 infection, aged mice exhibited increased morbidity and mortality post influenza infection (Fig. 2g, Extended data Fig. 2j).In contrast to SARS-CoV-2 infection however, we observed persistent immune cell infiltration and collagen deposition in the alveolar epithelium, particularly in aged mice that persisted to 60 dpi (Fig. 2h).Moreover, lungs from aged mice harbored significantly larger Krt5 + and Krt8 hi areas of dysplastic repair as well as higher levels of CD8 + T cells post influenza viral pneumonia, similar to human PASC-PF lungs (Fig. 2i, 3b,c).Of note, we observed a sustained defect in pulmonary function in aged mice up to 60dpi following influenza viral pneumonia, mimicking the impaired lung function observed in patients with severe pulmonary sequelae after acute COVID-19 (Fig. 2j,k

Exuberant tissue CD8 + T cell responses impair alveolar regeneration and promote dysplastic lung repair following viral pneumonia.
To further investigate the dynamics of the induction of lung fibrosis, we infected young and aged mice with influenza virus and characterized the immune and epithelial progenitor compartments over time (Fig 3a, Extended data Fig. 3a).The induction of epithelial progenitor activity was comparable in young and aged mice during the acute phase of infection (up to 14dpi) but diverged at later timepoints with increased Krt5 + and Krt8 hi progenitors in aged mice (Fig. 3a-c).
These trends correlated with a persistent age-associated defect in alveolar regeneration, exemplified by a sustained reduction in AT2 cell numbers (Extended data Fig. 3b,c).
Consistent with our previous study, lungs from aged mice harbored significantly higher levels of CD8 + T cells post influenza infection compared those of young mice (Fig. 3d) (9,34).Similar to human PASC-PF lungs, we also observed a spatial association between CD8 + T cells and Krt8 hi areas of dysplastic repair, reinforcing the relevance of this model to study chronic pulmonary sequelae (Fig. 3e, 1i).Furthermore, the association between CD8 + T cells and Krt8 hi areas was seen only at post-acute timepoints and strengthened over time, recapitulating features of human lungs after severe SARS-CoV-2 infection and suggesting these immune-epithelial progenitor interactions are primarily a feature of chronic sequelae of viral infections (Fig. 3f).
To understand the role of the persistent CD8 + T cells at this stage, we treated aged influenzainfected mice with CD8 + T cell-depleting Ab (αCD8) or isotype control Ab starting from 21dpi (Fig. 3g).This post-acute timepoint was chosen to ensure no interference with the essential antiviral activities of CD8 + T cells during acute infection as the virus is completely cleared by 15 dpi in this model (34).CD8 + T cell depletion improved histological evidence of disease in aged but not young mice (Fig. 3h, Extended data Fig. 3d).Importantly, Krt5 + and Krt8 hi areas were significantly reduced by depletion of CD8 + T cells (Fig. 3i,j), suggesting that CD8 + T cells are essential for the maintenance of dysplastic repair areas after recovery from acute disease.
Interestingly, AT1 and AT2 cells were markedly increased, and the alveolar architecture was restored after CD8 + T-cell depletion (Fig. 3k,l).We also observed a concomitant improvement in lung function after αCD8 treatment, suggesting that exuberant CD8 + T cell responses in the aftermath of acute disease affected the restoration of the alveolar spaces resulting in chronic impairment of alveolar gas-exchange function (Fig. 3m,n).Whether the observed decrease in Krt5 + and Krt8 hi cells post CD8 + T cell depletion is a result of complete differentiation into mature alveolar cell types or apoptosis of the progenitors remains to be elucidated using lineage-tracing studies, which are logistically challenging to perform in aged mice.Moreover, since the expansion of ectopic Krt5 + basal cells and Krt8 hi transitional cells was observed in the distal lung as early as 9dpi, it is unlikely that the depletion of CD8 + T cells 21dpi onwards affected the induction of the dysplastic repair program.Notably, depletion of CD8 + T cells resolved Krt5 + but not Krt8 hi areas in young mice and did not dramatically affect lung pathology or alveolar regeneration (Extended data Fig. 3d-i), suggesting that CD8 + T cells may specifically influence age-associated dysplastic lung repair.
To dissect the roles of circulating CD8 + T cells and lung resident memory CD8 + T cells in impairing alveolar regeneration, we used low and high dose αCD8 treatment to deplete circulating and pulmonary CD8 + T cells respectively, in aged influenza-infected mice (34,49).
We found that the resolution of areas of dysplastic repair only occurs upon depletion of pulmonary CD8 + T cells but not circulating CD8 + T cells (Extended data Fig. 3j,k) (34), suggesting that lung-resident CD8 + T cells are required for the maintenance of dysplastic lung repair.Taken together, our results strongly implicate the persistent tissue CD8 + T cell responses in the development of chronic pulmonary sequelae post viral pneumonia in aged animals.

Spatial transcriptomics reveal proximal interactions between CD8 + T cells, macrophages,
and Krt5 + and Krt8 hi -rich areas of dysplastic repair.
Given the heterogenous distribution of tissue pathology and the importance of capturing the spatially confined interactions between immune and epithelial progenitor cells within areas of dysplastic repair, we performed spatial transcriptomics on aged influenza-infected mouse lungs (60dpi) treated with control Ab or αCD8 (Fig. 4a).Following UMAP visualization and clustering of the capture spots (Extended data Fig. 4a,b), we observed a strong association between gene expression signatures of CD8 + T cells, and Krt5 + and Krt8 hi areas of dysplastic repair, similar to immunostaining results (Fig 4b , 3e).Consistent with previous data, gene expression signatures of Krt5 + and Krt8 hi progenitors were dramatically reduced in αCD8-treated lungs, with a concomitant increase in healthy alveolar epithelial cells (Fig 4b , 3i-k, Extended data Fig. 4d,e).This also corresponded with the expression pattern of fibrosis, with the strongest enrichment within Krt5 + and Krt8 hi areas and a similar reduction in αCD8-treated lungs (Extended data Fig. 4c).We further performed an agnostic evaluation of signaling pathways differentially regulated within the healthy alveolar epithelium compared to Krt5 + and Krt8 hi -rich areas of dysplastic repair using Gene Set Enrichment Analysis (GSEA) (Fig. 4c).Several pathways associated with inflammatory responses were highly active within areas of dysplastic repair, whereas growth factor responses and restoration of the vasculature were prominently observed in the healthy alveolar epithelium.
Upon investigation of specific gene expression-signatures, we observed an enrichment of Krt8 hi transitional cells (also known as alveolar differentiation intermediates (22), damage-associated transitional progenitors (21), and pre-alveolar type-1 transitional cell state ( 20)) (Fig. 4d) and aberrant basaloid cells (42) (Fig. 4e), particularly within areas of dysplastic repair.Evaluation of various immune and epithelial cell marker signatures revealed prominent enrichment of monocyte-derived macrophages in Krt5 + and Krt8 hi -rich areas in addition to other immune cells including CD4 + T cells, interstitial macrophages, B-cells, and natural killer cells (Fig. 4f, Extended data Fig. 4f,g).In contrast, gene expression signatures associated with pro-repair tissue-resident alveolar macrophages, AT1, and AT2 cells were primarily observed within the healthy alveolar epithelium (Fig 4f, Extended data Fig. 4e,h).To further characterize the immune-epithelial progenitor niche within areas of dysplastic repair, we performed a correlation analysis and found that CD8 + T cells and monocyte-derived macrophages were physically clustered around Krt5 + and Krt8 hi areas of dysplastic repair, and excluded from areas enriched with alveolar macrophages and mature alveolar epithelial cells (Fig. 4g).
To validate our findings from the spatial transcriptomics data, we immunostained aged influenza-infected mouse lungs and identified a similar enrichment of CX3CR1 + monocytederived macrophages within Krt8 hi areas of dysplastic repair (Fig. 4h,i).Interestingly, CD8 + T cell depletion resulted in a decrease in CX3CR1 + macrophage numbers, suggesting a potential role for their recruitment and maintenance in lungs (Fig. 4j,k, Extended data Fig. 4g).Similar to our mouse model of post-viral sequelae, human PASC-PF lungs exhibited an increase in macrophage populations compared to controls (Fig. 4l,m), which were also strongly enriched within areas of dysplastic repair (Fig. 4n,o).Together, our data revealed a conserved finding in both mouse and human post-viral lungs, where CD8 + T cells are present in fibrotic regions and in proximity to fibroproliferative mediators such as monocyte-derived macrophages, and Krt5 + and Krt8 hi dysplastic epithelial progenitors, representing a pathological niche after acute respiratory viral infections.

A CD8 + T cell-macrophage axis induces IL-1β release to arrest AT2 trans-differentiation in the transitional cell state.
We postulated the interactions between CD8 + T cells, macrophages, and epithelial progenitors within this pathological niche generated molecular cues to create a profibrotic microenvironment.We observed an enrichment of IL-1R signaling as well as inflammasome components in Krt5 + and Krt8 hi -rich areas of dysplastic repair compared to the healthy alveolar epithelium in aged influenza-infected mouse lungs at 60dpi (Fig. 5a, Extended data Fig. 5a).
IL-1β has been shown to promote the expansion of transitional Krt8 + cells upon bleomycin injury (21,50).Therefore, we investigated if IL-1β mediated the development of post-viral pulmonary fibrosis.First, we found that CD64 + macrophages were major producers of pro-IL-1β compared to CD64 -cells in influenza-infected aged lungs (Fig. 5b and Extended data Fig. 5b).Since mature IL-1β release from cells requires caspase-1 mediated pro-IL-1β cleavage (51), we examined caspase-1 activity in lung macrophages after CD8 + T cell depletion.Strikingly, CD8 + T cell depletion significantly reduced caspase-1 activity in lung macrophages (Fig. 5c).Moreover, inflammasome gene signatures were attenuated with αCD8 treatment (Extended data Fig. 5c), supporting the notion that CD8 + T cells persisting in the lungs post infection promote chronic inflammasome activation and IL-1β release by macrophages.CD8 + T effector and memory T cells are known express high levels of IFN-γ and TNF (52).Indeed, infected aged lungs harbor high levels of TNF (Fig. 5d) and IFN-γ (Fig. 5e) expressing CD8 + T cells.To determine whether IFN-γ and TNF mediate the observed inflammasome activation, we treated aged-influenza infected mice with neutralizing antibodies towards both IFN-γ and TNF starting 21dpi and observed a decrease in caspase-1 activity (Fig. 5f).As this effect was similar to αCD8-treatment (Fig. 5c, Extended data Fig. 5c), we directly tested if CD8 + T cell-derived IFN-γ and TNF regulated macrophage inflammation activation.We used an in vitro coculture of macrophages and CD8 + T cells isolated from mice previously infected with influenza (42 dpi) to assess IL-1β release (Fig. 5g).Indeed, we observed CD8 + T cells augmented macrophage Il1b mRNA expression (Extended data Fig. 5d), caspase-1 activity (Fig. 5h, Extended data Fig. 5e), and IL-1β release (Fig. 5i).The synergistic activity between macrophages and CD8 + T cells to produce IL-1β in vitro was not observed when isolated from naïve mice, suggesting that prior infection is necessary to prime the cells (Extended data Fig. 5f).Moreover, IL-1β released into the supernatant was significantly reduced upon treatment with IFN-γ and TNF neutralizing Ab in the coculture system, confirming the role of IFN-γ and TNF in promoting IL-1β release by macrophages (Fig. 5j).Consistently, we observed a similar decrease in BAL fluid IL-1β levels following treatment with either αCD8, or αIFN-γ + αTNF treatment (Fig. 5k,l).
Since CD8 + T cells, monocyte-derived macrophages, and Krt8 hi progenitors accumulate within dysplastic areas after influenza infection, we tested whether macrophage-derived IL-1β is a negative regulator of AT2 to AT1 trans-differentiation.Using a 2D primary AT2 cell culture model known to induce spontaneous differentiation into AT1 cells through the transitional cell stage, we examined if conditioned media from cocultured CD8 + T cells and macrophages influenced AT2 trans-differentiation (Fig. 5m) (18,37).AT1 cell marker expression was reduced upon exposure to conditioned media from CD8 + T cell-macrophage coculture compared to macrophages alone, which was rescued upon treatment with αIL-1β (Fig. 5n).In contrast, transitional cell marker expression exhibited the opposite pattern, with increased levels within the coculture group and a dramatic reduction following αIL-1β treatment (Fig. 5o).Conversely, rIL-1β treatment inhibited AT1 marker expression and promoted the expression transitional cell markers akin to previous reports (21).Collectively, our results suggest that exuberant CD8 + T cell-macrophage interactions promote chronic IL-1β release to inhibit AT2 cell trans-differentiation by arresting the cells in the transitional state.

Therapeutic neutralization of IFN-γ and TNF, or IL-1β activity enhances alveolar regeneration and restores lung function.
Our data thus far indicate a pathological role for CD8 + T cells persisting in human PASC-PF or virus-induced lung sequelae in an animal model in the development of chronic pulmonary sequelae.Since depletion of CD8 + T cells is not a clinically feasible treatment strategy, we explored the therapeutic efficacy of neutralizing the cytokines that are effectors of the profibrotic CD8 + T-cell.As expected, blocking IFN-γ and TNF activity in aged influenza-infected mice ameliorated fibrotic sequelae when compared to isotype controls (Fig. 6a,b).Moreover, IFN-γ and TNF neutralization attenuated Krt5 + and Krt8 hi areas of dysplastic repair (Fig. 6c,d), enhanced alveolar regeneration as evidenced by the increased numbers of AT1 and AT2 cells (Fig. 6c,e).The observed cellular changes also reflected in physiological benefit, with improved lung function following treatment (Fig. 6f).
Next, we tested the efficacy of neutralizing IL-1β in influenza-infected aged mice and observed dramatic attenuation of lung fibrosis, which phenocopies the results of CD8 + T cell depletion, and neutralization of IFN-γ and TNF (Fig. 6g,h, Extended data Fig. 6a,b).Improved alveolar regeneration was also observed, as evidenced by reduced Krt5 + and Krt8 hi areas and increased AT1 and AT2 cells (Fig. 6i-l).Further confirming the therapeutic efficacy of IL-1β blockade post infection, we found improved lung function in Ab-treated mice (Fig. 6m).Notably, improved outcomes following IL-1β neutralization were only seen in aged mice but not young mice, which is likely due to the elevation of IL-1β observed exclusively in aged mice (Extended data Fig. 6c,d).Thus, these data suggest that neutralization of IFN-γ and TNF, or IL-1β in the post-acute stage of viral infection may serve as viable therapeutic options to augment alveolar regeneration and dampen fibrotic sequelae observed following respiratory viral infections (Extended data Fig. 7).Consistent with the observed improvement in outcomes and a recent study identifying increased BAL IL-1β levels in respiratory PASC (7), we found that circulating IL-1β levels were elevated in individuals exhibiting persistent abnormal pulmonary function compared to those that had fully recovered (Fig. 6n), suggesting that chronic IL-1β activity may impede the restoration of normal lung function after acute SARS-CoV-2 infection.

DISCUSSION
Recent efforts have revealed several histopathological features conserved across various cohorts of respiratory PASC and PASC-PF patients including the prolonged reduction in alveolar epithelial cells (39), maintenance of Krt8 hi and Krt5 + dysplastic progenitors (5,8), and the persistence of various immune cell populations in the lungs (9,10,53).Although immunederived cues have recently been shown to influence lung repair (11,21,48,50,54), their exact interactions, if any, with the alveolar epithelium and role in post-viral fibrosis remain unexplored.
Here, we link these independent observations in PASC-PF lungs and an animal model of postviral fibrosis to describe spatially defined microenvironments composed of a dysregulated immune-epithelial progenitor niche that underlie dysplastic lung repair and tissue fibrosis after acute COVID-19.Intriguingly, these niches were observed specifically in PASC-PF lungs but not in acute COVD-19 or IPF lungs, indicating that these interactions are a unique feature of postviral fibrosis.
Understanding the mechanistic basis of respiratory PASC requires suitable animal models, however, SARS-CoV-2 MA-10 infection of neither C57BL/6 nor BALB/c mice resulted in severe alveolar pathology and fibrosis beyond acute disease.Although previous studies indicate that aged BALB/c mice develop significant pulmonary inflammation and prolonged pathology after SARS-CoV-2 infection (55), our data suggest these murine SARS-CoV-2 infection models may not fully recapitulate the pathophysiology leading to PASC-PF in humans.A major deficiency of the SARS-CoV-2 mouse infection model is the absence of persistent Krt8 hi and Krt5 + areas -a hallmark of human PASC-PF.Moreover, CD8 + T cells, which are enriched in human PASC-PF lungs to potentially promote the maintenance of Krt8 hi and Krt5 + dysplastic areas, are not appreciably increased post SARS-CoV-2 infection in BALB/c mice (9), which may be a reflection of a genetically programmed bias towards T H2 responses (56).Thus, it is imperative to develop clinically relevant animal models of SARS-CoV-2 post-viral fibrosis, validated by comparative analyses with human PASC, to uncover the underlying mechanisms and identify therapeutic targets.Nevertheless, in this study we show that influenza infection in aged C57BL/6 mice induced chronic pulmonary sequelae that faithfully recapitulate the immunopathological features of human PASC-PF lungs.
Excessive infiltration and accumulation of profibrotic monocyte derived macrophages has been reported in the context of severe acute COVID-19, IPF, and PASC (12,53,57,58).Here, we elucidate a previously unknown role for pulmonary CD8 + T cells in impaired recovery and fibrotic remodeling in PASC-PF but not acute COVID-19 or IPF lungs.This distinction is likely the product of a dysregulated and protracted antiviral response, originally aimed towards the clearance of the virus and virus-infected cells.Following viral infection-mediated alveolar injury, lung resident CD8 + T cells are recruited and maintained at sites of severe damage in order to protect these vulnerable sites in case of reinfection -previously termed as repair associated memory depots (59).These pulmonary CD8 + T cells typically gradually contract with successful alveolar regeneration in individuals that successfully recover from acute COVID-19 (9).However, long-term persistence of CD8 + T cells in human PASC, and aged influenza-infected mice impairs lung recovery post infection and drives the development of fibrotic disease.Given the overlap in fibrogenic pathways, it has been proposed that PASC-PF may represent an intermediate state prior to potential progression towards IPF (8,60).Whether CD8 + T cells primarily dictate the balance between functional recovery and PASC-PF post infection, or also play a pivotal role in the development of IPF is an open question (61).It is also currently unclear if the prolonged maintenance and activity of CD8 + T cells in lungs is a result of excessive TGF-β signaling reported to occur in PASC (8), chronic persistence of viral remnants (62,63), or other independent mechanisms.By combining imaging and spatial transcriptomics modalities, we show that respiratory sequelae post viral infections is, at least in part, a result of chronic IL-1β signaling downstream of aberrant interactions between CD8 + T cells and monocyte-derived macrophages, mediated by IFN-γ and TNF.Although chronic IL-1β was found to impair AT2 trans-differentiation in vitro, the use of aged mice in our experiments posed a logistical challenge to directly test the effect of IL-1β on Krt8 hi and Krt5 + progenitors in vivo using transgenic mice.Thus, there exists a distinct possibility for IFN-γ and TNF as well as IL-1β and relevant downstream mediators to influence epithelial progenitor cell fate through their actions on other lung immune and non-immune cells to ultimately result in fibrotic remodeling.Nevertheless, we show that neutralization of IFN-γ and TNF, or IL-1β activity in the post-acute phase of infection can augment alveolar regeneration and dampen fibrotic sequelae.Given the observed benefits in adults and pediatric patients hospitalized with acute COVID-19, United States Food and Drug Administration has already granted emergency use authorization to the IL-1 receptor antagonist, Anakinra and the JAKinhibitor, Baricitinib, in acute COVID-19.Our data strongly suggest that these drugs may also serve as promising candidates to treat ongoing respiratory PASC in the clinic.Representative immunofluorescence images staining CD8 + T cells (CD8α) and epithelial ) (9, 40).The exact mechanisms underlying the divergent trajectories in recovery following acute alveolar injury due to SARS-CoV-2 MA-10 and influenza viral infections remain unclear.Nevertheless, extensive evaluation of various mouse viral pneumonia models indicates influenza infection in aged mice has closer histopathological alignment with features of chronic pulmonary sequelae observed in PASC-PF lungs and can serve as a clinically relevant model to study the mechanisms of viral infectionmediated lung fibrosis. Fig.1