Introduction
The 2026 Conference on Retroviruses and Opportunistic Infections (CROI) showcased several high-quality abstracts, covering exciting work in the areas of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-focused therapeutics, long COVID, SARS-CoV-2 immunity, and the impact of HIV on SARS-CoV-2 outcomes.
SARS-CoV-2 Therapeutics
Monoclonal Antibodies
In an oral abstract session focused on long-acting antivirals and novel delivery platforms, Tebas (Abstract 180) presented data from a first-in-human study of DNA-encoded monoclonal antibody (DMAb) technology.1 Although this particular study evaluated a SARS-CoV-2-specific monoclonal antibody (mAb) that is no longer effective against current circulating strains, the potential uses for DMAb technology are much broader. Just 1 or 2 injections of DNA in an individual could result in 2 or more years of steady delivery of mAbs to the circulation, targeting HIV, cancer antigens, autoimmune diseases, or neurologic conditions. DMAb technology could circumvent the current limitations of mAb infusions such as the facilities and time needed to manufacture mAbs, the logistics of transporting and storing mAbs at cold temperatures, the expense and time required to administer an intravenous medication, and the need for repeat infusions.
This was a phase I, dose-escalation study of the electroporation of synthetic plasmid DNA-encoding mAbs that are modified versions of tixagevimab and cilgavimab, which bind distinct sites on the SARS-CoV-2 spike protein. Tixagevimab/cilgavimab was granted emergency use authorization (EUA) by the US Food and Drug Administration (FDA) in December 2021 for COVID-19 preexposure prophylaxis for vulnerable individuals. Its EUA was revoked in January 2023, at which time more than 90% of circulating SARS-CoV-2 variants in the US were resistant to it. In this study, 2 separate plasmid DNA constructs (one of tixagevimab, one of cilgavimab) were codelivered intramuscularly via localized electroporation 1, 2, or 4 times over the course of 4 weeks at doses ranging from 0.5 mg to 2 mg in 44 healthy individuals. In the 39 individuals who completed the study, blood was sampled monthly for 8 months and then every 1 to 6 months for 96 weeks. More than 90% of participants reported local pain and scabbing and more than 40% reported local erythema, although the pain was reported to resolve within 5 to 10 minutes. Systemic adverse events (AEs) were rare, with myalgia reported in approximately 10% and headache reported in fewer than 10%.
Pharmacokinetic profiling was favorable. The 2 mAbs originating from the separate electroporations of the 2 DNA constructs were coexpressed in the circulation at similar levels. There was rapid production of mAbs by muscle cells, peaking at 6 to 8 weeks after administration. Levels of these antibodies remained stable at biologically relevant levels during the 2-year follow-up period presented here. No antidrug Abs were detected in any participant during the follow-up period. Binding assays demonstrated that mAbs elicited by this platform bound well to several strains of SARS-CoV-2, including ancestral SARS-CoV-2, Delta, and Omicron strain BA.2, although binding to BA.4/5 was decreased, as expected. Neutralization assays of resulting mAbs demonstrated half-maximal inhibitory concentration (IC50) levels similar to tixagevimab/cilgavimab. Although this particular SARS-CoV-2 mAb therapeutic no longer has activity against current strains and will not be further developed, this was a successful proof-of-concept study examining DMAb technology in humans and demonstrating the safety and durability of DMAb-elicited production of a functional mAb cocktail.
Although EUAs for almost all other SARS-CoV-2 mAbs have been revoked due to their loss of efficacy as a result of viral evolution and escape, pemivibart has retained its EUA for COVID-19 preexposure prophylaxis (PrEP) in individuals who are immunocompromised. Pemivibart is an mAb that targets the SARS-CoV-2 spike receptor binding domain and competes with binding of spike to its receptor, human angiotensin-converting enzyme 2 (ACE2). Narayan (Abstract 740) presented a posthoc analysis of pemivibart efficacy for COVID-19 PrEP in people with advanced HIV infection from the phase III CANOPY (A Study to Evaluate the Efficacy and Safety of VYD222 for Prevention of COVID-19) trial of pemivibart administered twice, 3 months apart.2 In the study, 27 individuals had advanced HIV, defined as baseline CD4+ T count below 350 cells/μL. Although this subgroup had a higher rate of serious AEs in the study than the study group as a whole, they had similar rates of treatment-associated AEs. Neutralizing Ab titers were elevated to levels associated with effective protection by pemivibart and no cases of reverse transcription polymerase chain reaction (RT-PCR)-confirmed symptomatic COVID-19 were reported through month 6. These data suggest that pemivibart is a viable PrEP agent against symptomatic COVID-19 in people with HIV (PWH) with CD4+ counts below 350 cells/μL.
Of note, the most recent Infectious Diseases Society of America (IDSA) Clinical Practice Guidelines recommend that pemivibart be considered for a more narrow population of PWH, ie, those with CD4+ counts below 200 cells/μL who are at risk of progression to severe COVID-19, and only when the predominant SARS-CoV-2 strains are susceptible to pemivibart.3 Although one can infer that PWH with CD4+ counts below 200 cells/μL had no RT-PCR-confirmed symptomatic COVID-19 since none from the broader HIV group did, it was difficult to determine from Narayan's presentation whether the neutralizing Ab titers in those with CD4+ counts below 200 cells/μL were elevated to levels associated with protection against COVID-19. It was also difficult to determine whether the rates of serious AEs or treatment-related AEs are different in this narrower group that would be more likely to receive pemivibart in clinical practice.
Small-Molecule Antivirals
Unlike most mAbs developed to date, SARS-CoV-2-targeted small-molecule antiviral drugs such as nirmatrelvir/ ritonavir, remdesivir, and molnupiravir have remained effective against all strains of SARS-CoV-2. However, each has their limitations. Nirmatrelvir/ritonavir is an effective oral medication, but it must be taken twice daily, has bothersome taste and gastrointestinal AEs, and has drug-drug interactions. Molnupiravir is also an oral medication, but studies suggest it may be less effective in preventing hospitalization and death from COVID-19,4–6 and given its mutagenic mechanism of action, some fear it may induce host mutations.7 In addition, in immunocompromised individuals or in other rarer cases, molnupiravir use can result in mutated SARS-CoV-2 that could be transmitted onward.8,9 Remdesivir use currently requires intravenous infusion, although oral prodrugs are currently in development. In this environment, advancements in the SARS-CoV-2 oral antiviral space are of great interest.
Okamoto (Abstract 738) presented preclinical work developing investigational quinazoline derivatives targeting SARS-CoV-2. They started with amodiaquine, an antimalarial drug with antiviral activity against ebolavirus and SARS-CoV-2 that may also inhibit interleukin-6 (IL-6) production from macrophages in an animal model.10 Its AEs, including hepatotoxicity and agranulocytosis, can be severe so the group replaced its quinoline core with a quinazoline backbone and synthesized and screened hundreds of derivatives to identify candidates with improved SARS-CoV-2 activity with reduced toxic effects. They identified YMSA-0958, a compound with a half-maximal effective concentration (EC50) in cell culture virus infection experiments similar to remdesivir and nirmatrelvir, although with a lower 50% cytotoxic concentration (CC50). The compound also performed as effectively as remdesivir in cell culture experiments when introduced 2 hours before or simultaneously with SARS-CoV-2 infection. YMSA-0958, like remdesivir, inhibited the SARS-CoV-2 RNA-dependent RNA polymerase. It was effective in decreasing viral RNA in the lung in a BALB/C mouse model of SARS-CoV-2 infection. Given YMSA-0958's potential for higher toxicity than remdesivir, a more detailed description of these preclinical mouse infection experiments and preclinical toxicity data will be needed to advance this compound. It will be interesting to see whether decreased viral replication in the lung is accompanied by decreased levels of IL-6 in the mouse model, given the effects of its precursor molecule and given that elevated IL-6 is an important biomarker of human COVID-19 severity.11
Nirmatrelvir/ritonavir is an inhibitor of the SARS-CoV-2 3C-like protease (3CLpro), also known as main protease (Mpro). 3CLpro is a cysteine protease and is required for the cleavage and maturation of proteins involved in viral replication. 3CLpro sequences from SARS-CoV, SARS-CoV-2, and MERS-CoV share a high degree of similarity, so it is an attractive target for antiviral development. Bardiot et al reported on the discovery and successful preclinical profiling, including animal model testing, of ALG-097558, an oral 3CLpro inhibitor with pan-coronavirus activity and no need for concomitant administration of ritonavir.12 Phase I studies indicated good safety and pharmacokinetics with 7 or fewer days of treatment. At CROI 2026, Fitzgerald (Abstract 737) presented safety and pharmacokinetic data from a phase I study of ALG-097558 administration to people with moderate hepatic impairment (Child-Pugh class B, n=8) or severe renal impairment (estimated glomerular filtration rate [eGFR] <30 mL/min, n=6), and healthy participants with normal renal and hepatic function. Safety data indicated that ALG-097558 was generally well tolerated, with only 1 treatment-emergent AE of epigastric pain in the hepatic impairment study. Pharmacokinetic data indicated that moderate hepatic impairment or severe renal impairment do not result in a more than 2-fold increase in mean plasma exposures of the study drug and its metabolite compared with healthy controls. The data support the dosing of ALG-097558 in individuals with mild-to-moderate hepatic impairment or mild-to-severe renal impairment and advancing this investigation to phase II trials.
Long COVID and Postacute Sequelae
Long COVID continues to affect millions of people worldwide. Purpura and colleagues (Abstract 752) surveyed people in the C-PIC (COVID-19 Persistence and Immunology Cohort) study of Columbia University. Of 159 individuals who received long COVID-specific health care who answered the survey, 108 (68%) reported having experienced stigma or having felt dismissed by clinicians regarding their long COVID symptoms. People with more severe long COVID were more likely to report stigma or dismissal. This should be a wake-up call to clinicians to work harder to help people with long COVID feel seen, validated, and supported.
Various population-based and electronic health record (EHR)-based studies have demonstrated that SARS-CoV-2 infection is associated with increased incidence of new medical diagnoses and worsening of prior conditions.13 Now that virtually every person has immunity to SARS-CoV-2, whether via vaccine or infection or both, it remains unknown to what extent COVID-19 in the current era is associated with postacute sequelae at a population scale. Sax and colleagues (Abstract 748) used the US-based Veradigm Network EHR to identify approximately 250,000 adults with a COVID-19 diagnosis or medical claim in 2023 with 12 months of data available before and after the index diagnosis. A total of 73% of these individuals had evidence of at least 1 COVID-19 vaccination. Documentation of a neurologic condition in the follow-up period that was not documented in the preinfection period was found in 13% of these individuals. Similarly, 9%, 8%, and 4% of the individuals had documentation of cardiovascular, respiratory, and chronic kidney disease (CKD) that was not in the records in the preinfection period. Increases of 3% to 17% in neurologic-, cardiovascular-, respiratory-, and CKD-specific health care utilization rates were found in this population. Even in the post-Omicron era of widespread SARS-CoV-2 immunity, this work provides evidence to suggest that COVID-19 is still associated with increased incidence of new diagnoses and worsening of prior diagnoses in the postacute period.
Byrne and colleagues (Abstract 749) utilized data from the DC Cohort (DC Cohort Longitudinal HIV Status Neutral Study) of PWH and determined rates of incident CKD and change in eGFR in the 2 years after COVID-19 diagnosis. The investigators propensity score-matched 463 PWH with COVID-19 2-to-1 with 855 control PWH. COVID-19 in PWH was associated with eGFR loss of 5.3 mL/min/1.73 m2 over 2 years of follow-up compared with PWH without COVID-19. A total of 7.3% of PWH with COVID-19 developed CKD during the follow-up period. An eGFR decline was more likely to occur in people with higher preCOVID-19 creatinine level, prior hepatitis B, and lower albumin levels. A decline in eGFR of more than 25% was associated with female sex, diabetes, and higher creatinine at COVID-19 diagnosis, whereas CD4+ count more than 500 cells/μL was protective. These data suggest that PWH who have COVID-19 might benefit from the monitoring of kidney function in the months after diagnosis.
Opsteen and colleagues (Abstract 753) examined immune signatures of long COVID in PWH and people without HIV (PWoH), ultimately showing that the immune dysregulation associated with HIV infection overshadows that associated with long COVID. In this study, 73 PWH and 197 PWoH were surveyed and sampled a median of 2.5 years after initial SARS-CoV-2 infection. Long COVID was defined as self-reporting 5 or more new symptoms post COVID or self-reporting 1 or more new symptom together with moderate-to-extreme symptoms or meeting the RECOVER (Researching COVID to Enhance Recovery) 2023/2024 Research Index. In this study, PWH were more likely to be male, be black, have hypertension, use tobacco, and to have had a COVID-19 vaccine prior to first SARS-CoV-2 infection, and were less likely to have been hospitalized for COVID-19, than PWoH. All PWH were virally suppressed on antiretroviral therapy (ART) for a median of 20 years and had a median CD4+ count of 740 cells/mL. The group first analyzed which symptoms associated with each other. Regardless of HIV status, obesity was associated with a greater number of correlated symptom pairs, largely defined by sleep problems, gastrointestinal issues, weakness, dyspnea, and cough. Flow cytometry was performed on blood samples, but significant differences in cell type frequency were associated with HIV status rather than long COVID status. An evaluation of the association of cell frequency with individual symptoms determined that elevated CD16+ nonclassical monocytes were associated with new-onset malaise, which concurs with a prior study demonstrating the significance of monocytes in the immune dysregulation associated with long COVID.14
The mechanisms underlying long COVID remain unknown, but one hypothesis is that persisting viral antigen or viral replication may cause long COVID. Mateu and colleagues (Abstract 747) presented data on levels of SARS-CoV-2 antigen detected in the blood in people with and without long COVID. In this study, 425 adults were followed for 2 years, including 167 people with long COVID, 148 people with full recovery from COVID-19, and 110 people without SARS-CoV-2 infection (uninfected). Blood samples were collected at 6 to 12 months post infection and 18 to 24 months post infection. The Simoa platform was used to quantify SARS-CoV-2 spike protein, spike S1 subunit, and nucleocapsid antigen from plasma.
At 6 to 12 months post infection, 31% of people with long COVID, 20% of those with full recovery, and 1 uninfected individual had detectable levels of any antigen. At 18 to 24 months post infection, only 3% of people with long COVID and no recovered or uninfected people had detectable SARS-CoV-2 antigen. Spike protein was detected in plasma in 20% of participants with long COVID, 16% of those with full recovery, and 4% of uninfected individuals. Spike S1 subunit was detected in 4%, 2%, and 2% and nucleocapsid was detected in 7%, 3%, and 2% of people in the long COVID, full recovery, and uninfected groups, respectively. Importantly, there was no significant association between long COVID and detectable antigen, whether examined separately or combined. The P-value of this test was not reported, so it is unclear whether there is a nonsignificant trend toward antigen persistence and long COVID. This adds to an uncertain evidence base in which few studies associate the presence of circulating SARS-CoV-2 antigen with long COVID and more studies find no association.
Long COVID and the Brain
One of the hallmarks of long COVID is brain fog and many research groups are working to uncover the mechanisms behind this neurocognitive dysfunction. Singh (Abstract 108) presented a talk during a session on neuropathogenesis about her group's preliminary work to uncover the link between long COVID, myeloid-compartment immune dysfunction, and cerebral vascular dysfunction. This work builds on earlier findings that severe COVID-19 is associated with epigenetic changes in hematopoietic stem and progenitor cells that affect progeny innate immune cells that persist for months to years.15 This earlier work determined that long COVID was associated with persistent epigenetic changes in monocytes and their precursors as well as increased expression of the calgranulins S100A8/A9 (which together form calprotectin) in monocytes. S100A8/ A9 are expressed in neutrophils and monocytes and are released during inflammation to stimulate leukocyte recruitment and cytokine secretion.
To test the hypothesis that monocyte activation and S100A8/A9 expression/secretion exacerbates neurologic dysfunction in people with long COVID, the group examined brain imaging and blood samples from 138 individuals without long COVID and 105 individuals who had experienced long COVID (90% of whom had current long COVID at the time of the study). Cerebral microvascular perfusion, glymphatic function, and blood-brain barrier function were assessed with different magnetic resonance imaging (MRI) protocols. Using subsets of participants, long COVID was associated with a higher frequency of circulating classical monocytes and an increased frequency of classical monocytes correlated negatively with MRI measures of microvascular perfusion. Long COVID was also associated with increased plasma levels of S100A8/ A9, and plasma S100A8/A9 levels are inversely associated with MRI measures of glymphatics. Monocytes from 3 to 5 people with long COVID exhibited increased excretion of IL-6 and CC motif chemokine ligand 2 from endothelial cells after monocytes were stimulated with inflammatory stimuli compared with 4 or 5 people with recovery. More participants are needed to see if these associations are significant, and the study is ongoing.
In a separate study of neurologic long COVID, Bernad and colleagues (Abstract 754) studied symptoms and blood samples in 131 individuals, 43 of whom were a median of 2 or more years post infection with neurologic long COVID, 29 of whom had long COVID, 28 of whom had full recovery after SARS-CoV-2, and 32 of whom were not infected. Assays quantifying herpesvirus viral load and serology were performed alongside flow cytometry for immunophenotyping and functional characterizations of SARS-CoV-2 and herpesvirus-specific T-cell responses. Levels of immunoglobulin G (IgG) against Epstein-Barr virus, cytomegalovirus, and human herpesvirus 6 did not differ across groups. Flow cytometry analyses demonstrated that CD8+ terminally differentiated effector memory T cells reexpressing CD45RA were elevated in long COVID (but not neurologic long COVID) compared with people with full recovery. People with neurologic long COVID had stronger CD4+ T-cell responses to SARS-CoV-2 antigens than to herpesviruses. The investigators concluded that at 2 or more years post infection, people with neurologic long COVID were no more likely to have evidence of herpesvirus reactivation or higher levels of herpesvirus T-cell responses than those who were recovered or never infected by SARS-CoV-2. The take-home message on long COVID from CROI 2026 is that its biomarkers remain elusive.
Immunity and Acute COVID-19
There were fewer studies on acute COVID-19 and SARS-CoV-2 immunity after acute infection at CROI 2026 than in prior years, reflecting waning interest in acute COVID-19 in immunocompetent individuals. Kapira and colleagues (Abstract 316) examined upper and lower respiratory tract immune responses in PWoH living in Malawi. Bronchoalveolar lavage (BAL), nasal swab, and peripheral blood sampling were performed on 43 individuals after COVID-19 and levels of SARS-CoV-2-specific immunoglobulin A (IgA) and IgG were quantified. Higher levels of spike-specific IgA than nucleocapsid-specific IgA were found in all 3 sampled sites, whereas this was not seen when examining spike and nucleocapsid IgG. Higher frequencies of spike-and nucleocapsid-specific CD4+ cells were observed in BAL than in blood. Circulating spike and nucleocapsid IgG correlated more strongly with BAL/lower respiratory tract Abs than nasal/upper respiratory tract Abs. This suggests that serum levels of SARS-CoV-2 Abs may be viewed as a rough indicator of the level of Abs in the lower respiratory tract and they may not be correlated with protection against infection in the upper respiratory tract.
Nanayakkara and colleagues (Abstract 744) examined whether prior infection or vaccination or both (hybrid immunity) played a role in preventing infection or time to symptom resolution in a prospective observational cohort study of adults with acute COVID-19 in North Carolina. A total of 2731 participants were enrolled beginning in October 2022 with follow-up through August 2025. Studies of this kind are not as common as they were at the beginning of the pandemic, and this work provides an idea of whether vaccination and hybrid immunity still provide benefit to people with mild-to-moderate COVID-19 in the post-Omicron era. Symptoms were assessed daily for 14 days via an online survey. Time of symptom resolution was defined as the first of 2 consecutive days reporting no symptoms. Hybrid immunity was associated with a 2.7-day faster time to symptom resolution than no prior vaccination/infection. Unvaccinated adults, with or without prior infection, had the slowest symptom resolution. People aged 65 years or older experienced faster resolution of symptoms, perhaps in part due to higher vaccine uptake in this group. These results provide evidence that COVID-19 vaccination is associated with a faster time to symptom resolution in the post-Omicron era.
Conclusion
CROI 2026 showcased a variety of interesting studies pertaining to acute COVID-19, long COVID and postacute sequelae, and SARS-CoV-2 therapeutic agents. These studies offer crucial insights into the therapeutic drugs, pathophysiology, epidemiology, and lived experiences associated with SARS-CoV-2 infection.
Notes
All abstracts cited in the text appear in the CROI 2026 Abstract eBook, available online at www.CROIconference.org
The IAS-USA has identified and resolved ahead of time any possible conflicts of interest that may influence CME activities with regard to exposition or conclusion. All financial relationships with ineligible companies for the author and planners/reviewers are below.
Financial affiliations in the past 24 months: Dr Antar reported no relevant financial relationships with ineligible companies. (Updated April 8, 2026)
Additional References Cited in Text
P Tebas, A Patel, JT Agnes, . Safety and pharmacokinetics of SARS-CoV-2 DNA-encoded monoclonal antibodies in healthy adults: a phase 1 trial. Nat Med. 2025;31(12):4150–4159. doi: 10.1038/s41591-025-03969-0
CR Wolfe, J Cohen, K Mahoney, . Safety and efficacy of pemivibart, a long-acting monoclonal antibody, for prevention of symptomatic COVID-19: interim results from a Phase 3 Randomized Clinical Trial (CANOPY). Clin Infect Dis. 2025;81(3):439–450. doi: 10.1093/cid/ciaf265
A Bhimraj, Y Falck-Ytter, LR Baden, . 2025 clinical practice guideline update by the Infectious Diseases Society of America on the treatment and management of COVID-19: pemivibart for pre-exposure prophylaxis, vilobelimab for critical illness, and abatacept or infliximab for severe or critical illness. Clin Infect Dis. 2026;81(Supplement_3):i1–i5. doi: 10.1093/cid/ciaf424
C Torti, PP Olimpieri, P Bonfanti, . Real-life comparison of mortality in patients with SARS-CoV-2 infection at risk for clinical progression treated with molnupiravir or nirmatrelvir plus ritonavir during the Omicron era in Italy: a nationwide, cohort study. Lancet Reg Health Eur. 2023;31:100684. doi: 10.1016/j.lanepe.2023.100684
YYH Cheung, EHY Lau, G Yin, . Joint analysis of vaccination effectiveness and antiviral drug effectiveness for COVID-19: a causal inference approach. Int J Infect Dis. 2024;143:107012. doi: 10.1016/j.ijid.2024.107012
WHK Schilling, P Jittamala, JA Watson, . Antiviral efficacy of molnupiravir versus ritonavir-boosted nirmatrelvir in patients with early symptomatic COVID-19 (PLATCOV): an open-label, phase 2, randomised, controlled, adaptive trial. Lancet Infect Dis. 2024;24(1):36–45. doi: 10.1016/S1473-3099(23)00493-0
CB Frederick, RF Schinazi, R Swanstrom. Reader comment on: “Comprehensive genotoxicity and carcinogenicity assessment of molnupiravir.” Toxicol Sci. 2025; 204(1):116–117. doi: 10.1093/toxsci/kfae156.
NM Fountain-Jones, R Vanhaeften, J Williamson, . Effect of molnupiravir on SARS-CoV-2 evolution in immunocompromised patients: a retrospective observational study. Lancet Microbe. 2024;5(5):e452–e458. doi: 10.1016/S2666-5247(23)00393-2
T Sanderson, R Hisner, I Donovan-Banfield, . A molnupiravir-associated mutational signature in global SARS-CoV-2 genomes. Nature. 2023;623(7987):594–600. doi: 10.1038/s41586-023-06649-6
S Aoki, T Tanaka, K Yokoi, . Design, synthesis, and anti-SARS-CoV-2 activity of amodiaquine analogs. Chem Pharm Bull (Tokyo). 2025;73(4):355–368. doi: 10.1248/cpb.c24-00647
D Cihakova, MB Streiff, SP Menez, . High-value laboratory testing for hospitalized COVID-19 patients: a review. Future Virol. 2021. doi: 10.2217/fvl-2020-0316
D Bardiot, DC McGowan, K Gupta, . Discovery and preclinical profile of ALG-097558, a pan-Coronavirus 3CLpro inhibitor. J Med Chem. 2025;68(13):13436–13454. doi: 10.1021/acs.jmedchem.5c00088
Z Al-Aly, Y Xie, B Bowe. High-dimensional characterization of post-acute sequelae of COVID-19. Nature. 2021; 594(7862):259–264. doi: 10.1038/s41586-021-03553-9
S Kumar, C Li, L Zhou, . A distinct monocyte transcriptional state links systemic immune dysregulation to pulmonary impairment in long COVID. Nat Immunol. 2026;27(2):200–212. doi: 10.1038/s41590-025-02387-1
JG Cheong, A Ravishankar, S Sharma, . Epigenetic memory of coronavirus infection in innate immune cells and their progenitors. Cell. 2023;186(18):3882–3902. doi: 10.1016/j.cell.2023.07.019