Benchmarked against pandemic influenza in golden hamsters, SARS-CoV-2 uniquely sustains interferon signaling, chemokine production and myeloid activation in olfactory bulb and epithelium a month after clearance, alongside altered behavior and matching signatures in recovered human olfactory tissue.
Frere JJ, Serafini RA, Pryce KD, Zazhytska M, Oishi K, Golynker I, Panis M, Zimering J, Horiuchi S, Hoagland DA, Møller R, Ruiz A, Kodra A, Overdevest JB, Canoll PD, Borczuk AC, Chandar V, Bram Y, Schwartz R, Lomvardas S, Zachariou V, tenOever BR. SARS-CoV-2 infection in hamsters and humans results in lasting and unique systemic perturbations after recovery. Science Translational Medicine. 2022. 14(664), eabq3059.
DOI 10.1126/scitranslmed.abq3059. PMID 35857629. PMCID PMC9210449.
Correspondence is addressed to Venetia Zachariou and Benjamin R. tenOever.
Benchmarked against pandemic influenza in golden hamsters, SARS-CoV-2 uniquely sustains interferon signaling, chemokine production and myeloid activation in olfactory bulb and epithelium a month after clearance, alongside altered behavior and matching signatures in recovered human olfactory tissue.
Long COVID is defined clinically by symptoms persisting beyond four weeks, and its biological basis was unresolved. This study asks what is specific to SARS-CoV-2 by running a matched comparison against a pandemic influenza A virus in the same animal model at doses chosen to give comparable peak titers. Hamsters were profiled at peak infection, at one week after clearance, and at 31 days, across lung, heart, kidney, and six regions of the nervous system, with histology read by a board-certified pathologist and transcriptomes compared against human cadaver tissue. Acutely, the two viruses looked similar, with interferon and TNF signatures across tissues and comparable pulmonary infiltrates. After clearance the pictures diverged. Both viruses left peribronchiolar metaplasia and renal tubular atrophy, but the lesions were more extensive after SARS-CoV-2. In the nervous system, SARS-CoV-2 uniquely maintained interferon signatures in the olfactory bulb at 31 days, with elevated ISG15, MX2 and IRF7, chemokines including CXCL10 and CCL5, and enrichment for microglial and myeloid gene sets, in the absence of detectable viral RNA. The olfactory epithelium showed the same interferon signature plus T cell recruitment and activation. Hamsters showed transient anosmia at 3 days and reduced marble burying at 26 days. Post-mortem olfactory bulb and epithelium from donors who had recovered from documented COVID-19 more than a month before death showed correlated inflammatory programs.
By 2022 the acute biology of SARS-CoV-2 was reasonably well described. Infected cells fail to mount an efficient type I interferon response, much of the interferon that appears derives from uninfected phagocytic cells, and NF-kB-driven cytokine and chemokine production proceeds unchecked, drawing neutrophils and monocytes into an airway where antiviral defense is poorly engaged. Productive infection is largely confined to the respiratory tract in vivo, which the authors attribute not to tropism but to the systemic interferon response rendering distal tissues refractory, an interpretation supported by the observation that organoids and organotypic brain cultures are readily infected ex vivo while the same tissues are rarely infected in animals unless interferon biology is disrupted. What was missing was an account of the persistent phase. Post-acute sequelae of COVID-19 were well documented clinically, including breathlessness, fatigue, depression, anxiety, and impaired memory and concentration, but their cause was unknown, and there was no small animal system in which to study them. The golden hamster was already established as phenocopying acute COVID-19 without viral adaptation.
What distinguishes SARS-CoV-2 from another pandemic respiratory virus in the changes it leaves behind after infection is cleared, and can any of those lasting changes, in particular in the nervous system, be linked to functional and behavioral consequences and corroborated in humans who have recovered?
The central design choice is the comparator. By infecting parallel cohorts with the 2009 pandemic H1N1 influenza virus at a dose chosen to match SARS-CoV-2 peak titers, the study makes it possible to say which findings belong to SARS-CoV-2 rather than to severe respiratory viral infection generally, and the authors argue in the discussion that this benchmarking is a methodological requirement rather than an optional control. Three time points structure the work, 3 days for peak replication in both models, 14 days for the week after clearance, and 31 days, which corresponds to the clinical threshold for long COVID in a human patient. Tissue sampling is deliberately broad, taking peripheral organs where COVID-19 complications are reported alongside six nervous system regions chosen either because they were documented as virus positive in patients or because they govern sensory, motor, cognitive or affective functions altered in long COVID. Transcriptional profiling is paired throughout with histology, immunohistochemistry, in situ hybridization, and targeted quantitative PCR so that a signature can be checked against tissue. Because whole regions were sequenced, cell type inference is approached by deconvolution and by gene sets for neural and glial populations rather than by single-cell methods. Finally, behavior is assessed with a buried food test for olfaction and a marble burying assay for anxiety-like and repetitive behavior, and the animal findings are set against post-mortem olfactory tissue from donors with documented recovered COVID-19.
The study does not establish a mechanism for the persistent olfactory inflammation and it states this limitation directly, including that it does not demonstrate causality between brain inflammation and the behavioral changes measured. What the data constrain is the following. Sustained interferon and chemokine signaling in the olfactory bulb and epithelium at 31 days occurs when no infectious virus and no detectable viral RNA remain in those tissues, when animals have long since mounted an antibody response, and without a measurable increase in local apoptosis. The inflammatory cells involved are predominantly microglia and infiltrating macrophages in the bulb, with T cell involvement confined to the epithelium. Two explanations are offered by the authors as hypotheses and neither is tested here. One is that persistent defective viral genomes or residual viral debris in a tissue that was positive early continue to drive a host response. The other is that infection damages the olfactory epithelial barrier, allowing commensal microorganisms access to the bulb, an idea the authors support by pointing to reported thinning and sloughing of the epithelium. They also note that low-level persistence of viral material has been reported by others under some circumstances and should not be ruled out. The link from olfactory inflammation to behavior is an inference from the anatomical connection between the bulb and the limbic system together with prior literature associating bulb damage with depressive phenotypes, and the greater severity of SARS-CoV-2 damage relative to influenza is attributed speculatively to the larger amount of double-stranded RNA generated during subgenomic RNA production.
The work provides a tractable small animal system in which the persistent phase of SARS-CoV-2 infection can be studied, with molecular, histological and behavioral readouts in the same animals, and it establishes a methodological standard for that work by requiring a benchmark virus. Several findings would have read as COVID-specific without that comparator and turn out not to be, including the acute interferon response across tissues, peribronchiolar metaplasia, renal tubular atrophy, and loss of olfactory receptor transcripts, all shared with influenza and differing in degree rather than in kind. What survives the comparison is narrower and therefore more informative, namely the sustained olfactory bulb and epithelium inflammation with myeloid activation, present in both sexes, occurring without detectable virus, and mirrored in human tissue collected long after recovery. That combination supplies a candidate biological correlate for long COVID symptomatology that can now be tested causally.
The interpretive frame for the acute phase, in which SARS-CoV-2 produces a strong NF-kB-driven inflammatory response alongside a deficient type I interferon response and interferon arises largely from uninfected bystander cells, comes from earlier work by this laboratory and is cited throughout. The hamster model, the influenza benchmark, and the earlier characterization of SARS-CoV-2 tropism and ciliary loss in that model are likewise laboratory antecedents. What is new to the collaboration is the neuroscience arm, contributed with the Zachariou laboratory, and the olfactory system expertise contributed with the Lomvardas and Overdevest groups, whose work on the molecular basis of anosmia is cited for the loss of sustentacular cells and the negative enrichment of neuronal populations. Reading this paper together with other corpus entries on tissue-specific SARS-CoV-2 responses, including the ocular work, would be category 3 synthesis, and the common thread of a persistent inflammatory program decoupled from detectable virus should be assembled centrally rather than asserted here.
The authors are explicit about several boundaries. The different replication trajectories of the two viruses restricted the comparison to peak infection and to time points after clearance, since intermediate time points differ in replication and clearance rate and are hard to interpret. Causality between persistent brain inflammation and altered behavior is not demonstrated. The behavioral assays used were developed and characterized in mice and rats rather than hamsters, and the group sizes are smaller than standard for behavioral work because the experiments had to be performed inside high containment. The human comparison rests on very few donors, two recovered and one control for olfactory bulb and two recovered and three controls for epithelium, and it compares a moderate hamster infection with human cases that ended in death, a mismatch the authors call unavoidable because patients with moderate disease do not generally come to autopsy. Peripheral organ findings likely reflect a milder infection state than the human post-mortem literature on kidney involvement. Sequencing was performed on whole brain regions, so differentially expressed genes are summed across all cell populations present and cell type attribution rests on deconvolution rather than on single-cell measurement. The absence of detectable virus at 31 days is a limit of the assays used and the authors note that persistence under some circumstances has been reported by others. Finally, the study is confined to one SARS-CoV-2 isolate, one influenza comparator, one inoculation route, and animals that were not previously immune or vaccinated.
A month after clearing SARS-CoV-2, hamsters still show inflammation and myeloid activation in olfactory tissue and altered behavior, changes absent after influenza and mirrored in recovered people.
To find what is particular to SARS-CoV-2, hamsters were infected with either that virus or pandemic influenza and followed for a month past recovery. Both left scarring in lung and kidney, worse after SARS-CoV-2. Only SARS-CoV-2 left the olfactory bulb and the olfactory lining inflamed, with activated immune cells and interferon signaling still running despite no detectable virus. Infected animals behaved differently, and olfactory tissue from people who had recovered showed comparable inflammatory programs.
This study addresses the biology of post-acute sequelae of COVID-19 by comparing SARS-CoV-2 with 2009 pandemic influenza in the golden hamster at matched peak viral loads, profiling lung, heart, kidney and six nervous system regions at peak infection, one week after clearance, and 31 days. Acute responses were largely shared. At 31 days both viruses had left peribronchiolar metaplasia and renal tubular atrophy, more extensive after SARS-CoV-2, while lung transcriptomes had shifted to repair programs. Uniquely after SARS-CoV-2, the olfactory bulb retained interferon signatures, chemokine expression and microglial and macrophage activation with no detectable viral RNA and no increase in apoptosis, and the olfactory epithelium added T cell recruitment and activation. The phenotype occurred in both sexes. Hamsters showed transient anosmia acutely and reduced marble burying at 26 days. Post-mortem olfactory bulb and epithelium from donors recovered from documented COVID-19 showed correlated inflammatory enrichment, though from very few individuals.
co-led. Frere 2022 and Serafini 2023 are co-corresponding with the Zachariou laboratory, Zazhytska 2022 was led with the Lomvardas and Overdevest groups, and Heaton 2014 is co-led with Peter Palese and tenOever as joint senior and corresponding authors