Over but not gone: Lingering epigenetic effects of COVID-19

Abstract

Long COVID affects nearly one in five adults who have had COVID-19, yet the mechanisms underlying this disorder remain poorly understood. In a new study, Cheong et al. show that the epigenetic and transcriptional state of myeloid immune cells and their progenitors are durably altered in patients following severe COVID-19.

Long COVID, or post-acute COVID syndrome (PACS), is now estimated to occur in 19% of adults having presented with COVID-19 [1]. As with other post-infectious inflammatory conditions -- including multisystem inflammatory syndrome in children (MIS-C), acute disseminated encephalomyelitis (ADEM), and post-streptococcal inflammatory syndromes -- PACS is believed to be caused by immunologic changes that persist after acute infection [2]. Further, it is now increasingly recognized that these changes include not only those in the adaptive immune system, including production of autoantibodies or autoreactive T cells due to molecular mimicry, but also changes in the innate immune system. Indeed, epigenetic changes in innate immune cells, such as macrophages, and their long-lived progenitors are thought to enable innate immune responses that are faster and more powerful than prior to an acute infection.

A new study by Cheong et al in Cell reveals crucial differences in the transcriptional and epigenetic state of innate immune cells and their progenitors among a group of adults that were severely SARS-CoV-2-infected during the first COVID-19 wave in New York City in 2020 [3]. By comparing single-cell ‘omics analyses of healthy people with those during the early post-infection or late post-infection phases, as well as individuals hospitalized for non-COVID-19 illness, the study team demonstrated that myeloid cells and hematopoietic progenitor cells harbored transcriptional and chromatin accessibility changes that did not return to baseline even one year after acute SARS-CoV-2 infection. These data provide powerful evidence that persistent changes, i.e., trained immunity, may underlie PACS and other post-infectious inflammatory syndromes.

Emergency hematopoiesis normally arises upon acute infections or inflammation to elevate the innate immune cells’ production and facilitate pathogen clearance; Cheong and colleagues predictably found that myeloid differentiation and platelet activation were significantly increased in early post-COVID-19 patients. However, the frequencies of myeloid progenitors were also significantly elevated in the late post-infection group, which is interesting as it suggests that skewed differentiation is a durable process occurring in innate immune cells and their progenitors for as long as twelve months following severe COVID-19 –long after emergency hematopoiesis is resolved.

Notably, this work mirrors observations from murine models that showed similar long-term production of memory innate immune cells conferred by transcriptomic, epigenetic, and functional reprogramming of myeloid progenitors after lipopolysaccharide (LPS) exposure [4]. In a murine model of Mycobacterium avium infection, our group recently demonstrated persistent functional characteristics of bone marrow-derived macrophages, including increased killing capacity and altered metabolism, long after infectious exposure and transplant of hematopoietic stem and progenitor cells (HSPCs) [5]. These changes were partially recapitulated by administration of IFNγ alone in mice, suggesting that inflammatory cytokines are key mediators of trained immunity. Furthermore, these changes were sufficient to confer cross-protection against murine infection with influenza virus, an antigenically distinct pathogen [5]. Collectively, these studies highlight the concept that infection history can be epigenetically inscribed in myeloid progenitors and have long-term effects on downstream innate immune responses.

From a technological standpoint, Cheong and coworkers employed an innovative enrichment approach to circumvent the rarity of HSPCs for their snRNA/snATAC-sequencing (seq) studies: they combined both bone marrow mononuclear cells (BMMC) and peripheral mononuclear cells (PBMCs) for their analyses [3]. This pooled approach was justified by their finding that bone marrow HSPCs and peripheral circulating HSPCs co-clustered in RNA-seq and ATAC-seq analyses and shared similar cell-defining lineage markers. This unique method, coined “PBMC analysis with Progenitor Input Enrichment (PBMC-PIE)”, provided sufficient HSPC numbers while obviating the need for bone marrow sampling in most study subjects.

Mechanistically, the authors identified chromatin accessibility signatures and expression of transcription factors (TF) shared in HSPCs and monocytes. In both the early- and late-post-disease cohorts, AP-1(FOS/JUN) and IRF family motifs were enriched. HSPCs maintained increased C/EBP family and JUN activity in the last post-disease cohort, even when most other AP-1 activity resolved. These findings are supported by work describing the mechanism behind inflammatory memory in epidermal stem cells: AP-1 family members FOS and JUN were also found to play a crucial role in enhanced wound healing following an initial immunological insult in mice, suggesting that this pathway might be universal across diverse cell types [6]. Of note, AP-1 family members including BATF2 can amplify pro-inflammatory signaling pathways and promote myeloid differentiation following IFNγ- mediated inflammatory stress in murine hematopoietic stem cells [7]. Cheong et al showed that while AP-1 family member TF expression was enhanced, key erythroid TF GATA1 expression was diminished, with analogous changes in chromatin accessibility [3]. These findings were mirrored by enhanced myelopoiesis and reciprocally diminished erythropoiesis in early- and late- post-COVID-19 patients (Figure 1).

Figure 1. The balance between myelopoiesis and erythropoiesis is altered after acute SARS-CoV-2 infection.

The altered balance between myelopoiesis and erythropoiesis occurs with no return to baseline up to 12 months post-infection, indicative of durable changes in the innate immune system. These changes have been mitigated in COVID-19 patients who have received IL-6R blockade therapy [3]. Figure created with Biorender.com.

To validate these findings, the authors generated a mouse model recapitulating the effects of COVID-19 on innate immune cells. Upon mouse hepatitis virus 1 (MHV-1) infection, innate immune cells were more pro-inflammatory and mobilized to various tissues, including bronchoalveolar lavage fluid and the brain, compared to controls -- findings that are consistently seen in COVID-19. Of note, such central nervous system inflammation was associated with a significant and persistent decrease in myelin basic protein (MBP) in murine brains, indicative of widespread demyelination. These findings suggest that inflammation and monocyte infiltration can lead to long-standing tissue remodeling which might contribute to persistent symptoms such as brain fog in PACS. These observations are consistent with reports in other murine models; for example, mesenchymal stromal cells that are responsible for maintaining the bone marrow niche of HSPCs can be persistently damaged after systemic LPS exposure [8]. These microenvironmental changes may also contribute to changes in hematopoiesis such as myeloid bias, which might persist after an acute infection resolves.

Overall, murine studies have highlighted a role for inflammatory signaling in HSPC reprogramming and trained immunity [5, 7, 9]. Severe COVID-19 is associated with a hyperinflammatory state, and anti-inflammatory therapy can improve outcomes when applied early in the disease course [10]. Accordingly, Cheong and colleagues reported that IL-6R blockade (tocilizumab) therapy appeared to mitigate long-term reprogramming of HSPC and myeloid cells [3]. Specifically, human COVID-19 subjects who received tocilizumab showed an improved return to baseline transcriptional and epigenetic states compared to those who did not. Furthermore, IL-6R blockade mitigated post-infectious inflammatory states, including increased frequencies of myeloid progenitors post-infection, enhanced IRF and C/EBP TF activity, as well as a shifted balance between myelopoiesis and erythropoiesis in the murine MHV-1 model. These findings raise the possibility that anti-inflammatory therapy might not only improve outcomes from acute SARS-CoV-2 infection, but might perhaps also help to prevent PACS. Overall, the study by Steven Josefowicz’s group [3] adds to the growing literature on trained immunity with implications for treatment and prevention of post-infectious inflammatory disorders, vaccination strategies, and more.