key: cord-0695892-4oyhemxa
authors: Evangelou, Konstantinos; Veroutis, Dimitris; Paschalaki, Koralia; Foukas, Periklis G.; Lagopati, Nefeli; Dimitriou, Marios; Papaspyropoulos, Angelos; Konda, Bindu; Hazapis, Orsalia; Polyzou, Aikaterini; Havaki, Sophia; Kotsinas, Athanassios; Kittas, Christos; Tzioufas, Athanasios G.; de Leval, Laurence; Vassilakos, Demetris; Tsiodras, Sotirios; Stripp, Barry R.; Papantonis, Argyris; Blandino, Giovanni; Karakasiliotis, Ioannis; Barnes, Peter J; Gorgoulis, Vassilis G.
title: Pulmonary infection by SARS-CoV-2 induces senescence accompanied by an inflammatory phenotype in severe COVID-19: possible implications for viral mutagenesis
date: 2022-02-03
journal: Eur Respir J
DOI: 10.1183/13993003.02951-2021
sha: 06c01a337bd772e0837a885af2af68b8518d082d
doc_id: 695892
cord_uid: 4oyhemxa

BACKGROUND: SARS-CoV-2 infection of the respiratory system can progress to a multi-systemic disease with aberrant inflammatory response. Cellular senescence promotes chronic inflammation, named as senescence-associated secretory phenotype (SASP). We investigated whether COVID-19 disease is associated with cellular senescence and SASP. METHODS: Autopsy lung tissue samples from 11 COVID-19 patients and 43 age-matched non-COVID controls with similar comorbidities were analysed by immunohistochemistry for SARS-CoV-2, markers of senescence and key SASP cytokines. Virally-induced senescence was functionally recapitulated in vitro, by infecting epithelial Vero-E6 cells and a three-dimensional alveosphere system of alveolar type 2 (AT2) cells with SARS-CoV-2 strains isolated from COVID-19 patients. RESULTS: SARS-CoV-2 was detected by immunocytochemistry and electron microscopy predominantly in AT2 cells. Infected AT2 cells expressed the angiotensin-converting-enzyme 2 (ACE2) and exhibited increased senescence (p16(INK4A) and SenTraGor(TM) positivity) and IL-1β and IL-6 expression. In vitro, infection of Vero-E6 cells with SARS-CoV-2 induced senescence (SenTraGor(TM)), DNA damage (γ-H2AX) and increased cytokine (IL-1β, IL-6, CXCL8) and Apolipoprotein B mRNA-editing (APOBEC) enzyme expression. Next-generation-sequencing analysis of progenies obtained from infected/senescent Vero-E6 cells demonstrated APOBEC-mediated SARS-CoV-2 mutations. Dissemination of the SARS-CoV-2-infection and senescence was confirmed in extra-pulmonary sites (kidney and liver) of a COVID-19 patient. CONCLUSIONS: We demonstrate that in severe COVID-19, AT2 cells infected by SARS-CoV-2 exhibit senescence and a proinflammatory phenotype. In vitro, SARS-CoV-2 infection induces senescence and inflammation. Importantly, infected senescent cells may act as a source of SARS-CoV-2 mutagenesis mediated by APOBEC enzymes. Therefore, SARS-CoV-2-induced senescence may be an important molecular mechanism of severe COVID-19, disease persistence and mutagenesis.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes coronavirus disease 2019 (COVID-19) that primarily affects the respiratory system. The clinical course of the patients ranges from asymptomatic to a life-threatening respiratory failure accompanied by a multi-systemic inflammatory disease [1, 2] . Systemic disease may occur through a virally-mediated inflammatory response that consists of a variety of proinflammatory cytokines and chemokines, including interleukin (IL)-1β, IL-6, IL-12, IFN-γ, TNF-α, CXCL-8, CXCL-10 and CCL-2 [3, 4] . The link between viral infection of cells and development of severe lung disease and systemic manifestations is still poorly understood. Viral infection results in the activation of complex innate and adaptive immune responses that are orchestrated sequentially, involving several cell types and inflammatory mediators [5, 6] . At the cellular level, intrinsic defence mechanisms are activated and outcomes range from complete recovery to cell death [7] [8] [9] [10] [11] . An "intermediate" and essential cellular state that is overlooked, due to lack of efficient methodological tools, is cellular senescence [12, 13] .

Cellular senescence is a stress response mechanism that preserves homeostasis. Senescent cells are characterized by prolonged and generally irreversible cell-cycle arrest and resistance to apoptosis [12, 14] . Additionally, they also exhibit secretory features, collectively described as the senescence-associated secretory phenotype (SASP) [12] . SASP includes a variety of cytokines, chemokines, growth factors, proteases and other molecules, depending on the type of senescence and the cells involved [12, 15] . These inflammatory proteins are released in the extracellular space as soluble factors, transmembrane proteins following ectodomain shedding, or as molecules engulfed within extracellular vesicles [16] [17] [18] . Cellular senescence plays a key role in several lung diseases, involving the senescence of several cell types in the lung [19, 20] . Under physiological conditions, senescence is transiently activated and SASP mediates the recruitment of immune cells for senescent cell clearance. In addition, other SASP factors promote tissue regeneration and repair, overall ensuring cellular/tissue homeostasis. On the contrary, persistence of senescent cells exerts harmful properties promoting tissue dysfunction and the maintenance of chronic inflammation, via paracrine and systemic SASP [12, 15] .

There is little published evidence linking viral infection to cellular senescence [21] [22] [23] [24] [25] [26] .

Given the implication of an inflammatory response in the progression of COVID-19 and the SASP secretion by senescent cells, we investigated whether cellular senescence occurs in COVID-19. Since individual markers are not adequate to unequivocally detect senescence, especially in vivo, as they may also be present in non senescent conditions, we followed in a clinical setting a detailed multi-marker algorithmic approach that we and others recently published and was approved by the senescence community [12, 27] .

We were the first to provide evidence supporting the proof for senescence in infected lung tissue by applying this algorithm in a previous, preprint (bioRvix), version of the current manuscript [28] , which subsequently has been confirmed by others [29] . This is the largest clinical study verifying viral-induced senescence by SARS-CoV-2 infection linked to a proinflammatory phenotype, which may contribute to acute and chronic clinical manifestations in severe COVID-19. Importantly, we also provide novel evidence for the generation of viral mutations promoted by SARS-CoV-2 persistence in senescent cells.

Formalin-fixed and paraffin-embedded autopsy lung tissue samples were obtained from 11 patients that died from severe COVID-19 (confirmed by RT-qPCR) ( Table 1 and Table S1A ).

Lung parenchyma displaying analogous lesions (atelectasia, fibrosis and infiltration by immune cells) from 43 age-matched patients with similar comorbidities from previously published cohort [30] and new cases, resected prior to the COVID-19 outbreak, were used for comparison with COVID-19 samples for all experiments detailed below (negative controls) ( Table 1 and Table S1B ). A non-COVID cohort of 60 age-matched lung samples was patients. The virus was recovered from a nasopharyngeal swab, rinsed in 1 ml saline and filtered twice through a 0.22 nm filter. Virus stock was prepared by infecting fully confluent Vero-E6 cells in DMEM, 10% fetal bovine serum (FBS) with antibiotics, at 37 o C and 5% CO 2 .

Virus stock was collected four days after inoculation, sequenced by NGS (online supplement) and the supernatant was frozen (−80°C) until use. Infections were carried out in 24-well plates, using SARS-CoV-2 at a 0.01 MOI. Cells were either fixed with 4% paraformaldehyde or lysed with NucleoZOL (Macherey-Nagel) 17 days post infection. Administration of the specific ataxia-telangiectasia-mutated (ATM) protein kinase inhibitor KU-55933 was carried out as described elsewhere [31] . Three independent experiments were performed. All manipulations were carried out in a Biosafety level 3 facility.

Formalin fixed and paraffin embedded sections from a three-dimensional alveosphere system infected with SARS-CoV-2 and corresponding controls (non-infected) were analysed.

The alveospheres consist of self-renewing AT2 cells that express the well-established AT2 cell markers, HTII-280 and surfactant protein C as well as of some cells expressing the AT1 cell marker HTI 56 [32] . Alveospheres were gently opened using pipetting to allow infection and to avoid cellular stress [32] .

Mouse immunization and antibody collection, selection and specificity determination are described in the online supplement. Transcriptome analysis of hybridomas and amino acid determination of selected clones are also provided. Four clones, (479-S1, 480-S2, 481-S3 and 482-S4) are under patent application (Gorgoulis VG, Vassilakos D, Kastrinakis N. GR-patent application no: 22-0003846810).

RNA extraction and Reverse-Transcription real-time PCR (RT-qPCR) detection were performed as previously described (online supplement) [33] .

NGS was performed as previously described (online supplement) [34] .

Protein extraction and immunoblot analysis was performed as described elsewhere [31] . For details see online supplement. Horse Radish Peroxidase conjugated anti-goat, anti-mouse and anti-rabbit secondary antibodies (1:1000 dilution) (Cell Signaling) were used. Primary antibodies were: anti-APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like) 3G/A3G (Abcam ab109727 and ab172694), anti-APOBEC3H (LSBio LS-C151868) and anti-GAPDH (Cell Signaling).

Immunocytochemistry and immunohistochemistry were performed according to published protocols [34] . The following primary antibodies were applied overnight at 4 o C: anti-SARS-CoV-2 (G2) monoclonal antibody (at a dilution 1:300), anti-SARS-CoV-2 monoclonal antibody (1A9 clone, Genetek), anti-ACE-2 (Abcam), anti-thyroid transcription factor (TTF)-1 (Dako),

anti-IL-1β (Abcam), anti-IL-6 (R&D systems), anti-phospho-histone (Ser 139 ) H2AX (γΗ2ΑΧ) (Cell Signaling), anti-Ki67 (Abcam) and anti-p21 Waf1/Cip1 (Cell Signaling).

SenTraGor TM (trademark of GL13 compound) staining and double staining experiments were performed and evaluated as previously described [27, 35] . The mean percentage of SenTraGor positive alveolar cells in at least 10 high power fields (x400) per patient was quantified.

Representative area from hematoxylin and eosin-stained paraffin sections of the lung autopsy of COVID-19 patients and non COVID-19 controls were chosen under the light microscope and marked. Paraffin-embedded tissue was deparaffinized, rehydrated and fixed in 2.5% glutaraldehyde in PBS for 24h and post-fixed in 1% aqueous osmium.

As described in detail in online supplement, candidacy for APOBEC deamination of C→U mutational substitutions in the SARS-CoV-2 genome of strains available in GISAID database [36, 37] and in those obtained after cell culture infection, was examined and verified against experimentally validated APOBEC motifs ( Figure S7 ).

Data are expressed as mean ± SD. Comparisons were performed with unpaired nonparametric Mann-Whitney U test (comparison between two groups) or Kruskal Wallis test followed by Dunn's post-hoc analysis (comparison between three groups). The Wilcoxon paired non-parametric test was used to compare values of infected cells in vitro.

In order to detect SARS-CoV-2 in lung tissue we developed monoclonal antibodies which react against receptor-binding domain (RBD) of the spike protein of SARS-CoV-2 and identified a high affinity antibody (G2), the validity of which was recently verified [38] ( Figure   S1 , Table S1 ). SARS-CoV-2 was detected (using both our 'in house' G2 clone and a commercially available antibody from GeneTek) predominantly in alveolar type 2 (AT2) cells, which were identified by TTF-1 and Surfactant Protein B (SP-B) positivity, and in sparse inflammatory cells (alveolar and tissue macrophages) in all COVID-19 patients, ranging from <5 cells/4mm 2 tissue to >50 cells/4mm 2 tissue ( Figure 1A , Table S1A ). Surface epithelial cells in small peripheral airways stained also positive in certain cases ( Figure S2 ). SARS-CoV-2 infected AT2 cells were occasionally large and appeared isolated (denuded or syncytial) or clustered (hyperplasia), exhibiting a variety of topological distribution ( Figure 1A) . These cells co-expressed the angiotensin-converting enzyme 2 (ACE2) receptor ( Figure 1B) , supporting SARS-CoV-2 infection being mediated by the ACE2 receptor [39] . In addition, electron microscopy analysis in representative COVID-19 cases confirmed the presence of virus within AT2 cells ( Figure 1C ) and high magnification revealed virions in the proximity of the endoplasmic reticulum, indicating their likely assembly and budding, as well as virions residing in cytoplasmic vesicles, implying their transfer and release into the extracellular space.

A proportion of SARS-CoV-2 infected AT2 cells (range 8-21%) displayed a senescent phenotype, with positive staining for SenTraGor TM and p16 INK4A (Figure 2A -C, Figure S3 ) [12, 29, 30] . SenTraGor TM is an established marker of senescence and has the unique property of detecting senescent cells in any setting, including archival material [formalin-fixed paraffin embedded (FFPE) tissue)] [12, 27, 35] . By contrast, lung tissues exhibiting fibrosis, atelectasia and inflammatory infiltrates from age-matched non-COVID-19 cases with similar comorbidities ( Table 1 and Table S1 ), including a separate control cohort of patients with acute pneumonia (aspiration pneumonia), showed significantly lower senescence (range 1-2%, p<0.0001) (Figure 2A-C, Figure S4 ), suggesting that SARS-CoV-2 infection may induce senescence.

In order to demonstrate that SARS-CoV-2 can induce cellular senescence, we infected Vero cells with a viral strain isolated from a COVID-19 patient. Vero-E6 cells is an established cellular system for viral propagation and studies, as apart from their high infectivity to SARS-CoV-2 they are among the few cell lines demonstrating SARS-CoV-2-mediated cytopathic effects, an essential aspect in diagnostics [40, 41] . Infection was carried out at a low multiplicity of infection (MOI) to mimic natural coronavirus infection [42] . In line with our hypothesis, the infected cells following an initial surge of cell death reached an equilibrium demonstrating clear evidence of senescence (increased SenTraGor positivity, increased p21 WAF1/CIP1 and reduced Ki-67), as compared to the non-infected control cells at 17 days post infection (Figure 3 and S5A) . As Vero cells lack p16 INK4A [43] , the most likely trigger of senescence is DNA damage, as previously reported [12, 13] . DNA damage measured by γ-H2AX immunostaining was evident in SARS-CoV-2 infected cells ( Figure 3A4 ). Treatment of infected cells with a selective ATM inhibitor (KU-55933) dramatically decreased senescence assessed by SenTraGor staining (Figure S6 ). ATM has been previously reported as a key driver of NF-κB-dependent DNA-damage-induced senescence [44] . It appears that genotoxic stress results from a vicious cycle imposed by the virus in host cells as it hijacks most intracellular protein machineries [11, 45] . Likewise, cellular senescence was identified in infected alveolar cells/alveospheres of an established primary lung alveolar threedimensional model, confirming the in vivo observations and the findings in Vero cells ( Figure   3B , Figure S5B ).

We found very high expression of both IL-1β and IL-6 by senescent AT2 cells in the lungs of COVID-19 patients, while in the non COVID-19 control cases, including samples from patients with acute pneumonia, expression was very low in the few senescent AT2 cells detected in lung tissue with fibrosis, atelectasia and inflammatory infiltrates (p<0.0001) ( Figure 4A -C, Table S1 ). As both cytokines are key components of the inflammatory response to SARS-CoV-2, our findings may imply this pattern of inflammation may be due to the SASP as a result of cellular senescence in COVID-19 patients. Likewise, SARS-CoV-2 senescent Vero cells displayed expression of SASP-related cytokines, as assessed by our algorithmic assessment of senescence, supporting our in vivo findings ( Figure 3C) .

SARS-CoV-2 immunoreactivity was additionally detected by applying both our 'in house' G2 clone and a commercial one from GeneTek in serial sections of available kidney and liver tissue from one COVID-19 patient ( Figure 5A) . A clear cytoplasmic signal was evident in a number of renal tubules as well as in areas of the liver parenchyma. In contrast, in tissues from non-COVID-19 cases the signal was absent, as expected ( Figure 5A) . Interestingly, by serial section analysis, areas harboring the virus also exhibited senescence (SenTraGor positivity), implying occurrence of SARS-CoV-2 induced senescence in extra pulmonary sites ( Figure 5B ).

We recently hypothesized that senescent cells could represent a source of SARS-CoV-2 quasispecies generation [46] . This assumption was based on the fact that the apoptotic tolerant nature of senescent cells allows the virus to be hosted for longer periods compared to other cells with higher cell turnover [47, 48] , thus rendering the SARS-CoV-2 virome more susceptible to host mediated editing. APOBEC enzymes are well known to participate in viral genome editing [12, 33, 49] . Notably, we recently showed these enzymes to be highly expressed in senescent cells [49] .

Updating our initial bioinformatic investigation [46] , we screened ~ 4,500,000 SARS-CoV-2 strains from the GISAID database [36, 37] . This analysis mostly indicated that the predominant signature of SARS-CoV-2 variants is APOBEC-mediated (Figure 6 ), in line with recent literature [50] . To validate our hypothesis, we proceeded to cell culture infection with two strains isolated from patients and obtained progenies from senescent cells ( Figure 6B ). Next generation sequencing (NGS) analysis demonstrated that the collected viral genomes acquired mutations. Importantly, the predominant pattern was APOBECdriven, in agreement with the bioinformatic analysis of the GISAID database. Moreover, and in line with our presumption and our previous study [49] , the SARS-CoV-2 induced senescent cells demonstrated significantly higher mRNA and protein levels of the APOBEC enzymes, particularly G and H (RNA editing cytoplasmic variants), which are reported to play a pivotal role in viral RNA editing [12, 33, 49] . Other mutation signatures were also found both in the cell culture isolated virome genomes and in the strains from GISAID, and maybe the outcome of the oxidative stress present in the senescent cells [51] .

We have demonstrated the presence of SARS-CoV-2 in AT2 cells of patients who died from COVID-19 using our novel and a commercial anti-SARS-CoV-2 antibody against its spike protein and by electron microscopy. We have shown for the first time following the senescence detecting algorithm in vivo that a proportion of SARS-CoV-2-infected AT2 cells exhibit features of cellular senescence (as demonstrated by significantly increased staining with the novel senescence marker SenTraGor TM and also of p16 INK4A ) [12, 27, 35] . In this respect, our investigation encompassing serial section analysis and co-staining is the first to demonstrate in vivo which cells are truly senescent and infected. The finding that in agematched non-COVID-19 controls, including lung tissue samples from a cohort of patients with acute pneumonia, the percentage of senescent cells was much lower (1-2%) than that of the COVID-19 patients (8-21%), strongly indicates that SARS-CoV-2 triggers senescence. To confirm that SARS-CoV-2 induces cellular senescence per se, we demonstrated that infection of epithelial cells with SARS-CoV-2 virus (B.1.222 strain) in vitro increased SenTraGor TM staining and induced DNA damage, measured by increased γ-H2AX expression. Importantly, we found that inhibition of ATM (ataxia-telangiectasia-mutated protein kinase), an apical orchestrator of the DNA damage response pathway, dramatically reduced senescence in infected samples, suggesting SARS-CoV-2 induced senescence being mediated by DNA damage and activation of the DNA damage response pathway. In line with our in vivo and in vitro findings, SARS-CoV-2 induced senescence was identified in a previously established primary lung alveolar three-dimensional model [32] . The fact that alveospheres underwent gentle mechanical manipulations is unlikely to have contributed to the occurrence of senescence, given that such interventions are frequently applied on 3D organoid systems and no senescence phenotypes have been reported nor are evident from the post-infection proliferation rates and culture [52] .

We also demonstrated that cells infected with SARS-CoV-2 exhibit high expression of IL-1β and IL-6, both components of the SASP and implicated in systemic features of severe COVID-19 [3, 4] . Therefore, our in vitro and in vivo findings suggest that SARS-CoV-2 attaches to AT2 cells via ACE2 to infect these cells and through activation of DNA damage response signalling roots, induces cellular senescence and associated proinflammatory phenotype (SASP). In line with our observations, Lee et al recently demonstrated that SARS-CoV-2-induced senescence exhibited enhanced γ-H2AX DNA damage foci that were abrogated by the reactive oxygen species (ROS) scavenger N-acetylcysteine (NAC) [29] , suggesting implication of ROS and oxidative stress ( Figure S8) . Moreover, infected cells displayed SASP-positive senescence mediated by p53 and activation of cGAS/STING signaling pathway [29] . They also demonstrated that senolytics reduced COVID-19 lung disease and inflammation in infected hamsters and mice [29] , further suggesting SASP as an outcome of viral induced senescence.

Camell et al also reported that treatment with senolytics of old mice infected with a SARS-CoV-2-related mouse β-coronavirus reduced inflammation [53] . Based on our findings and recent evidence a potential mechanistic scenario could be the following ( Figure S8 ): upon entrance, the virus highjacks certain energy consuming functions related to RNA processing, translation and the ER [11] . Thus, an energy shift occurs resulting in increased ROS production, oxidative stress and DNA damage/DNA Damage Response (DDR) pathway activation. In addition, DDR and the subsequent cell cycle arrest may be driven by an interaction of Coronavirus nsp13 protein and DNA polymerase δ [54] and SASP via the cGAS/STING and other DNA damage dependent pathways ( Figure S8 ). Therefore, our previous work showing that DNA damage/DDR is a potent inducer of senescence, signifies its importance in the establishment of the viral induced senescence phenotype [31] .

Senescent cells are in a state of cell cycle arrest but remain metabolically active and secrete a typical profile of inflammatory proteins known as the SASP. SASP components include the proinflammatory cytokines IL-1β and IL-6, which are elevated in plasma of COVID-19 patients that have acute respiratory distress syndrome (ARDS) or systemic inflammatory features. We demonstrate that in COVID-19, senescence alters the properties and function of AT2 cells, producing cytokines that could be released into the systemic circulation, amplifying and perpetuating chronic inflammation. Of course, the availability of tissue samples and the absence of more refined COVID-19 in vitro and in vivo models make it difficult to draw robust conclusions. It is likely that SARS-CoV-2 spreads from epithelial cells in the lower airways to infect AT2 cells in the alveolar walls, which express ACE2, and causes local senescence and inflammation via SASP in the lung. In line with this notion, it has been previously reported in vitro that cellular senescence has also a role as a natural antiviral defence mechanism and that SASP acts as the major contributor of this response, activating and recruiting the immune system to clear out the infection [21, 22] . Subsequently, other cellular compartments (stem/progenitor, endothelial and inflammatory cells) may be affected via different cell and non-cell autonomous mechanisms, leading to aberrant immune responses, chronic inflammation, tissue dysfunction and/or fibrosis and eventually to lung damage and failure [26, 55] . In particular, senescence in stem/progenitor cells impairs lung regenerative capacity. Immunosenescence results in elevated neutrophils-tolymphocytes ratio (NLR) and high IL-6 production. In addition, IL-1, IL-6, CXCL8 and TNF-α might be crucial for maintaining chronic inflammation and TGF-β, PAI-1, and MMPs might favour fibrosis via the expression of fibrotic genes (ACTA2, COL1A1, COL1A2, and FN1) in the surrounding microenvironment. The virus may then enter the circulation and senescence may subsequently spread systemically to affect other organs (Figure 5) , leading to multiorgan failure/multi-morbidity and death in the acute phase [1, 25] , or leading to post-acute sequelae of COVID-19 (PASC or long-COVID), an evolving syndrome with long-term complications [26, 56] . In this manner, senescence could also predispose to disease severity upon re-infection by SARS-CoV-2 or infection by other viruses and even negatively impact vaccine efficacy, similarly to what has been proposed for aged individuals [26] .

An additional implication relates to the prolonged survival of senescent cells that are infected with the virus, as senescent cells are resistant to apoptosis and clearance by efferocytosis [12, 14] . Such a context can provide an extended time "window" for virus replication, therefore exposing its genome to host-mediated editing. Along this vein we showed that the senescent cell compartment acts as a "fertile" environment for mutational evolution of the virus, as it is more susceptible to APOBEC-3G and 3H RNA editing ( Figure 6 ) [33, 49] . Of course, other mechanisms extending survival and replication of SARS-CoV-2 may take place. Coronaviruses code for an important multifunctional enzyme termed papain-like protease (PLP) that exerts intrinsic deubiquitinating and deISGylating activities. The latter is related to interferon-stimulated genes (ISG) up-regulation and can serve to antagonize the host's immune response that would otherwise hinder infection [57] . In this manner increased and prolonged viral replication and exposure of its genome to host mediated editing and quasispecies generation could also occur. Taking into account that interferons are known senescence inducers, the picture gets even more complicated, rendering difficult to discriminate IGS-mediated mechanisms from that involving virally-induced senescence [58] . Of course, additional studies are needed and the guideline algorithmic approach for in vivo senescence assessment, followed in the current work, is anticipated to clarify in the future a lot of unanswered questions in clinical settings [12, 27] . Notably, in the recently identified Omicron (B.1.1.529) SARS-CoV-2 variant/strain the APOBEC signature was identified as the predominant mutational profile (Figure S9 ). To better understand the role of APOBEC in the generation of viral mutations, inhibition of APOBEC proteins by RNA interference would be valuable. However, this is very challenging, since APOBEC suppression induces DNA damage and sensitises cells to stress induced death [59] [60] [61] . Interestingly and in line with our concept linking cellular senescence with viral mutagenesis, by conducting a bioinformatic analysis, we found that viruses bypassing cellular senescence (oncogenic) exhibit a significant lower mutation burden compared to viruses inducing senescence ( Figure   S10 ). Given that mutations can affect the effectiveness of vaccines, our findings could imply a potential implication of viral induced senescence in vaccination strategies. Further studies are required towards this root.

Despite the fact that up to the present this is the largest clinical study demonstrating virally-induced senescence by SARS-CoV-2, a limitation of our study is the relatively small sample size of the COVID-19 lung autopsies, due to difficulty of accessing this rare material, given that the autopsies are limited and were performed only in the initial phase of the COVID-19 outbreak in order to identify the pathological basis of this new entity. Another limitation due to the nature of the disease is that pathological features, such as cellular senescence in lungs, can only be investigated in cadaverous material, which represents the most severe outcome of the spectrum of COVID-19 clinical manifestations. Therefore, evaluation of senescence in less severe conditions is not currently feasible. Moreover, in our study we examined 18 cases of aspiration pneumonia, which displayed similar degrees of senescence to other non-infected control cases. It would be interesting to extend these studies in other types of pneumonia. However, material from patients with an acute but not COVID-19 related inflammatory response is rare, given that these patients commonly recover following treatment and in the infrequent cases of fatal outcome autopsy is rarely All non-COVID samples were obtained from distal lung tissue to resected lung carcinoma (lobectomy or pneumonectomy). All non-COVID acute pneumonia samples were from patients with aspiration pneumonia.

10μm 10μm 10μm 10μm

Anti-SARS-CoV-2 ab TTF-1 

Anti-SARS-CoV-2 ab(+)/p16 INK4A (+) 1. Bioinformatic screening and identification of accumulating mutations in the genome of Sars-CoV-2 isolates from patients STEP 2. STEP 1.

Spectrum of mutations observed STEP 3.

Define APOBEC sites that overlap C > U mutations APOBEC non-APOBEC

A.

i.

ii.

C → U C → U in strains from senescent cells Screen using APOBEC consensus (see Fig S7 for 

RNA was collected from biological duplicates of generated hybridomas as described elsewhere [2] . RNA samples were processed according to manufacturer's instructions, Fastq files were demultiplexed with Flexbar [3] . Quality control of the Fastq files was assessed with FastQC tools [4] . Adapter sequences were removed with Cutadapt program [5] with the following parameters: quality trimming was set to 20 and the minimum allowed nucleotide length after trimming was 20 nucleotides using --pair-filter=any to apply the filters to both paired reads. A two way alignment mode was followed to identify the antibody clone. More precisely alignments were performed with Bowtie2 [6] with parameters set as following: -D 20 -R 3 -N 1 -L 20 -i S,1,0.50 -no-mixed --nodiscordant against an index made from IMGT database http://www.imgt.org/ having downloaded all mouse and human IG genes. Also this mode of alignments was executed for quality control and visualization of the aligned reads spanning the IG gene segments on the genome browser. The second mode refers to the determination and reconstruction of the clones. This was performed with MiXCR suite [7] . At first, alignments against the IG repertoire were performed with kaligner and visualization of alignments was assessed. It was observed that the use of kaligner gave better results with higher clone hits regarding the VH and VL segments. Full assembly of the clones was performed. A full report of the number of reads and assembly of CDR and FR clones is provided in clones479_S1kalign.txt. The clones with the highest number of reads and coverage across the V, D, J segments were considered. The reported matched sequences were also checked with IgBlast tool https://www.ncbi.nlm.nih.gov/igblast/. In addition, after the assembly of the amino acid reconstruction of the FR and CDR regions of the full variable fragment for both the Heavy and Light antibody chains, a 3D visualization was also determined via folding the V protein fragment with iTassser suite [8] . The above analysis has been extensively described in Gorgoulis VG, Vassilakos D and Kastrinakis N.

(2020) GR patent application no: 22-0003846810.

Method: ICC and IHC were performed according to previous published protocols [9] . In brief, 3 μm thick sections from formalin-fixed paraffin embedded (FFPE) lung tissues were Evaluation of G2 staining: Cells were considered positive irrespective of the staining intensity. Two different semi-quantitative IHC evaluation approaches, previously described were adopted [11, 12] According to the first, the number of G2 positive cells for positive staining in >50 cells per 4 mm 2 [10] . Regarding the second one, the number of G2 positive cells per whole slide was estimated and subsequent scores were assessed:

(+) between one and five positive cells per whole slide (scattered cells), (++) more than five cells per whole slide but no foci (isolated cells) and (+++) more than 10 cells in one × 20 field (with foci) [11] . For p16 INK4A, the mean percentage of positive alveolar cells in at least 10 high power fields (x400) per patient was measured. This information has been included in the methods section (Main and Online Suppl Data). For IL-6 and IL-1b, the percentage of immunopositive cells was encountered [13] . Evaluations were performed blindly by four experienced pathologists (KE, PF, CK and VG) and intra-observer variability was minimal (p≤0.05).

To investigate the mutational patterns of the SARS-CoV-2 genome we downloaded from GISAID database (https://www.gisaid.org/) 4,672,296 available strains that were distributed globally (Step 1 in Figure S7 ). These strains were aligned with Bowtie aligner [6] , having as reference the Wuhan first assembly NC_045512, obtained from NCBI (https://www.ncbi.nlm.nih.gov/sars-cov-2/) (Step 2 in Figure S7 ). More information regarding the commands used for the alignments is provided in Part 1 of the supplementary code file (bioinformatic.analysis.sh). The identification of mutations is performed with an "in-house" script (Part 2 of supplementary code file) using calmd function from SAMtools [14] , which is based on proteome occupancy profile studies (Step 3 in Figure S7 ) [15] . Candidate APOBEC sites need to have a C→U frequency of mutation at the same nucleotide position of more than 5 reads. Furthermore, in order to investigate at RNA secondary structure level the filtered deaminated sites (with more than 5 C→U mutation counts), we extracted windows of ±60 nucleotides around the C→U most frequent sites and interrogated the folding of the RNA sequences using the Vienna RNA fold algorithm (Step 4 in Figure S7 ) [16] . To guide the RNA folding we incorporated SHAPE reactivities from SHAPE-seq data for SARS-CoV-2 [17] . To decipher the candidate motifs we counted the frequency of letters starting from k-mers of ±5 nucleotides from the most frequent deaminated position up to 30-mers, which is usually taken as an upper limit in most RBP pull-downs and usually corresponds to the protein bound protected fragment [18, 19] (Step 4 in Figure S7 ). The frequency for each letter was determined via a perl script, which extracts all possible k-mers and their frequencies.

The highest motif consensus was around ±5 nucleotides from the C→U deaminated nucleotide, as the motif becomes more degenerate when extending above ±7

nucleotides. Next, the frequency of each k-mer was plotted with Web-logo motifs [20] . In Figure S7 ). The dominant RNA structure motif was identified using the BEAM software [21] . Overall the motif analysis regarding the sequence composition around the most frequently deaminated sites, along with the RNA structure, revealed an UACCA enrichment around regions of open hairpin structures, in agreement with the results from the literature [22] [23] [24] [25] [26] [27] .

Filtering and alignments of Fastq files for the SARS-CoV2 strains from the infected Vero E6 cells and from actual patients were demultiplexed with Flexbar [3] . Quality control of the Fastq files was assessed with FastQC tools [4] . Adapter sequences were removed with Cutadapt program [5] with the following parameters: quality trimming was set to 28 and the minimum allowed nucleotide length after trimming was 21 nucleotides. Potential, very high over-represented k-mers, at the beginning of the reads, were removed.

Alignments were performed with Bowtie2 [6] with the parameter -very-sensitive using as a reference genome the B. Black arrows depict senescent cells in comparison to non senescent ones (yellow arrow).

Absence of senescence is clearly evident in non infected alveolar cells. Step 1

Call mutations

Step 3a

Step 3b

Extract k-mer around C→U:

[ (-N k ...N 1 ) C→U(+N 1 ...N k ) ] Step 4 

Characteristics of SARS-CoV-2 and COVID-19

Distinct phenotypes require distinct respiratory management strategies in severe COVID-19

An inflammatory cytokine signature predicts COVID-19 severity and survival

Cytokine Storm in COVID-19: The Current Evidence and Treatment Strategies

SARS-CoV-2 immunity: review and applications to phase 3 vaccine candidates

Adaptive immune responses to SARS-CoV-2 infection in severe versus mild individuals

Immunity and immunopathology to viruses: what decides the outcome?

Viral infection and the evolution of caspase 8-regulated apoptotic and necrotic death pathways

Regulating the adaptive immune response to respiratory virus infection

Breaking Bad: How Viruses Subvert the Cell Cycle

Actionable Cytopathogenic Host Responses of Human Alveolar Type 2 Cells to SARS-CoV-2

Cellular Senescence: Defining a Path Forward

The spatio-temporal landscape of lung pathology in SARS-CoV-2 infection. medRxiv : the preprint server for health sciences

Senescence and apoptosis: dueling or complementary cell fates?

Senescence and senotherapeutics: a new field in cancer therapy

Unbiased analysis of senescence associated secretory phenotype (SASP) to identify common components following different genotoxic stresses

A proteomic atlas of senescence-associated secretomes for aging biomarker development

Unmasking senescence: context-dependent effects of SASP in cancer

Cellular Senescence as a Mechanism and Target in Chronic Lung Diseases. American journal of respiratory and critical care medicine

The aging lung: Physiology, disease, and immunity

Cell senescence is an antiviral defense mechanism

The Interaction of Viruses with the Cellular Senescence Response

Cell fusion induced by ERVWE1 or measles virus causes cellular senescence

Cellular senescence: friend or foe to respiratory viral infections? The European respiratory journal

The struggle of a good friend getting old: cellular senescence in viral responses and therapy

Cellular senescence as a potential mediator of COVID-19 severity in the elderly

Algorithmic assessment of cellular senescence in experimental and clinical specimens

SARS-CoV-2 infects lung epithelial cells and induces senescence and an inflammatory response in patients with severe COVID-19

Virus-induced senescence is a driver and therapeutic target in COVID-19. Nature 2021. 30. Gorgoulis VG

Oncogene-induced senescence is part of the tumorigenesis barrier imposed by DNA damage checkpoints

SARS-CoV-2 infection of primary human lung epithelium for COVID-19 modeling and drug discovery

A prototypical nonmalignant epithelial model to study genome dynamics and concurrently monitor micro-RNAs and proteins in situ during oncogene-induced senescence

Sample pooling strategies for SARS-CoV-2 detection

Robust, universal biomarker assay to detect senescent cells in biological specimens

disease and diplomacy: GISAID's innovative contribution to global health

Global initiative on sharing all influenza data -from vision to reality

SARS-CoV-2 Antigenemia as a Confounding Factor in Immunodiagnostic Assays: A Case Study

Structural and Functional Basis of SARS-CoV-2 Entry by Using Human ACE2

Culture of SARS-CoV-2 in a panel of laboratory cell lines, permissivity, and differences in growth profile

Comparative tropism, replication kinetics, and cell damage profiling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study

Protease-mediated enhancement of severe acute respiratory syndrome coronavirus infection

The genome landscape of the african green monkey kidney-derived vero cell line. DNA research : an international journal for rapid publication of reports on genes and genomes

ATM is a key driver of NF-κB-dependent DNA-damage-induced senescence, stem cell dysfunction and aging

Cellular senescence as a source of SARS-CoV-2 quasispecies

Viral Infections: Negative Regulators of Apoptosis and Oncogenic Factors

SARS-CoV-2 triggers inflammatory responses and cell death through caspase-8 activation

Induction of APOBEC3 Exacerbates DNA Replication Stress and Chromosomal Instability in Early Breast and Lung Cancer Evolution

Potential APOBEC-mediated RNA editing of the genomes of SARS-CoV-2 and other coronaviruses and its impact on their longer term evolution

Recent Advances: Molecular Mechanism of RNA Oxidation and Its Role in Various Diseases

Modelling Cryptosporidium infection in human small intestinal and lung organoids

Senolytics reduce coronavirus-related mortality in old mice

SARS-CoV-2 triggers DNA damage response in Vero E6 cells

Spike Protein Induces Paracrine Senescence and Leukocyte Adhesion in Endothelial Cells

Deubiquitinating Enzymes in Coronaviruses and Possible Therapeutic Opportunities for COVID-19

Type I interferons and related pathways in cell senescence

APOBEC3G acts as a therapeutic target in mesenchymal gliomas by sensitizing cells to radiationinduced cell death

APOBEC3G rescues cells from the deleterious effects of DNA damage

Inhibition of APOBEC3G activity impedes double-stranded DNA repair

Potential Clinical Benefits of Quercetin in the Early Stage of COVID-19: Results of a Second, Pilot, Randomized, Controlled and Open-Label Clinical Trial

Possible Therapeutic Effects of Adjuvant Quercetin Supplementation Against Early-Stage COVID-19 Infection: A Prospective, Randomized, Controlled, and Open-Label Study

Porcine epidemic diarrhea virus nucleoprotein contributes to HMGB1 transcription and release by interacting with C/EBP-β

A prototypical nonmalignant epithelial model to study genome dynamics and concurrently monitor micro-RNAs and proteins in situ during oncogene-induced senescence

FLEXBAR-Flexible Barcode and Adapter Processing for Next-Generation Sequencing Platforms

FastQC: a quality control tool for high throughput sequence data

Cutadapt removes adapter sequences from high-throughput sequencing reads

Fast gapped-read alignment with Bowtie 2

Antigen receptor repertoire profiling from RNA-seq data

The I-TASSER Suite: protein structure and function prediction

Robust, universal biomarker assay to detect senescent cells in biological specimens

Activation of the DNA damage checkpoint and genomic instability in human precancerous lesions

In situ detection of SARS-CoV-2 in lungs and airways of patients with COVID-19. Modern pathology : an official journal of the United States and Canadian Academy of Pathology

Unspecific post-mortem findings despite multiorgan viral spread in COVID-19 patients

Physiological hypoxia restrains the senescence-associated secretory phenotype via AMPK-mediated mTOR suppression

The Sequence Alignment/Map format and SAMtools

Differential protein occupancy profiling of the mRNA transcriptome

The Vienna RNA websuite

Genome-wide mapping of SARS-CoV-2 RNA structures identifies therapeutically-relevant elements

CLIP and complementary methods

A large-scale binding and functional map of human RNA-binding proteins

WebLogo: a sequence logo generator

A novel method for the identification of conserved structural patterns in RNA: From small scale to high-throughput applications

Crystal structure of the catalytic domain of HIV-1 restriction factor APOBEC3G in complex with ssDNA

The double-domain cytidine deaminase APOBEC3G is a cellular site-specific RNA editing enzyme

Stem-loop structure preference for site-specific RNA editing by APOBEC3A and APOBEC3G

Evidence for hostdependent RNA editing in the transcriptome of SARS-CoV-2

Nanoscale Characterization of Interaction of APOBEC3G with RNA

APOBEC Enzymes as Targets for Virus and Cancer Therapy

SARS-CoV-2/ SENESCENCE

SenTraGor(+)

1,553 0,94 9,999 9,999 9,999 9,999 3,253 2,818 0,716 0,569 0,539 0,81 D 9,999 9,999 9,999 9,999 9,999 9,999 1,366 1,093 1,332 1,728 9,999 9,999 E 9,999 9,999 0,362 0,363 9,999 9,999 9,999 9,999 9,999 9,999 9,999 9,999 F 9,999 9,999 9,999 9,999 0,384 0,348 2,713 2,172 9,999 9,999 9,999 9,999 G 9,999 9,999 2,517 2,657 9,999 9,999 9,999 9,999 2,265 9,999 1,71 2,454 H 9,999 9,999 9,999 9,999 2,324 2,419 9,999 9,999 1,026 0,377 0,342 0,398