The outbreak of coronavirus disease 2019 (COVID-19) is an unprecedented global health crisis. Tissue and peripheral blood analysis indicate profound, aberrant myeloid cell activation, cytokine storm, and lymphopenia, with unknown immunopathological mechanisms. Spatiotemporal control of the quality and quantity of the antiviral immune responses involves synchronized cellular and molecular cascades and cross-talk between innate and adaptive immunity. Dysregulated responses in immunity, such as at the stages of immune sensing, alarming, polarization, and resolution, may contribute to disease pathology. Herein, we approach SARS-CoV-2 through an immunomodulatory lens, discussing possible mechanisms of the asynchronized antiviral immune response and proposing potential therapeutic strategies to correct the dysregulation.
The recurring outbreaks of viruses within the Coronaviridae family, such as Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV), Middle East Respiratory Syndrome Coronavirus (MERS-CoV), and now SARS-CoV-2, demonstrate the grave threat that these viruses pose to humanity. SARS-CoV-2 is considered less lethal than SARS-CoV and MERS-CoV but more contagious (Perlman and Netland, 2009; Sanche et al., 2020). Despite some knowledge regarding the epidemiology and clinical presentation of the resulting disease, COVID-19, much less is understood regarding its pathobiology (Guan et al., 2020).
Angiotensin-converting enzyme 2 (ACE2) has been identified as the major SARS-CoV-2 receptor critical for the infection of lung epithelial cells and other tissues (Hoffmann et al., 2020; Zhou et al., 2020a). However, viral load from the nasopharynx does not appear to be a good biomarker correlating with disease severity and mortality; indeed, patients with severe disease can present with low viral titers, while high viral titers have been observed in asymptomatic individuals (Lescure et al., 2020; Zou et al., 2020). SARS-CoV-2 and its family members share some puzzling, unique pathological features, most notably acute respiratory distress syndrome (ARDS), cytokine release syndrome (CRS), and lymphopenia, despite excessive myeloid cell–dominant inflammation, which has been correlated with COVID-19 severity (Vabret et al., 2020). Moreover, single-cell RNA sequencing studies on the nasopharyngeal and bronchial samples from COVID-19 patients revealed the existence of the viral genome in immune cells (Bost et al., 2020; Chua et al., 2020). These findings highlight possible direct viral–immune engagement and an essential, pathogenic signature of immune dysregulation. Here, we present a model of intermolecular and intercellular immune asynchrony, based on dysregulation of timing, localization, quality, and quantity of the immune response, to explain the immunopathogenesis of COVID-19. We also propose potential strategies to correct the viral-induced immune dysregulation.
Antiviral immunity: A multifactorial but synchronized system
The activation, recruitment, and resolution of an antiviral immune response is composed of efficiently organized cellular and molecular cascades that tightly regulate the balance between viral clearance and immune toxicity (Fig. 1). Upon virus infection, multiple innate immune recognition mechanisms can detect the virus, recognizing pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs; Vabret et al., 2020), or viral proteins (Koch et al., 2013). Within hours, the sensing apparatus triggers a rapid antiviral response to inhibit viral replication, primarily through type I/III IFN production (Stetson and Medzhitov, 2006), together with other cytokines (e.g., IL-1, IL-18, and IL-6) and chemokines (e.g., CCL2 and CCL7). Over the course of several days, adaptive immunity is activated and subsequently down-regulates innate immunity to avoid nonspecific damage to the host (Kim et al., 2007). As the pathogen is eliminated, various innate (e.g., M2 macrophages and regulatory dendritic cells) and adaptive regulatory cell types (e.g., regulatory T and B cells) also help resolve inflammation. In addition to this temporal regulation, spatial considerations also affect the outcome of an immune response (Dorward et al., 2020). Because of their different homeostatic requirements, different organs are differentially immunologically poised; lungs, for example, are more prone to T helper cell 2 (Th2) immunity, mediated in part by unique tissue-resident immune cells and diverse tissue factors that also influence local priming (Ardain et al., 2020).
This synchronized antiviral, intercellular coordination is reinforced by intracellular regulation and enacted via intermolecular interactions. Brakes for the immune system present themselves at an intrinsic level, with some immune cells possessing a short lifespan (e.g., neutrophils), a poor capacity to proliferate (e.g., some tissue-resident macrophages), or a sensitivity to activation-induced cell death (e.g., lymphocytes). Immune responses are also controlled externally by a multitude of costimulatory (e.g., CD28, 4-1BB) and coinhibitory (e.g., PD-1, CTLA-4, LAG-3) receptor–ligand interactions and cytokine/chemokine networks (Chen and Flies, 2013). Many of these inhibitory pathways are induced in response to inflammation to tame the immune response: for example, the secretion of decoy receptors such as IL-1Ra or IL-18BP (Dinarello et al., 2013; Mantovani et al., 2019) in response to IFN-γ (Hurgin et al., 2002; Tilg et al., 1993). Thus, immune homeostasis is achieved via cross-talk between molecules that is integrated within and across innate and adaptive immune cells, ensuring appropriate timing, location, quality, and quantity of a successful antiviral immune response.
COVID-19 immunopathogenesis: An immune asynchrony model
Postmortem analysis of COVID-19 patients showed a heavy infiltration of mononuclear cells in affected areas, including lung, heart, and kidney, coupled with cytokine storm and lymphopenia (Merad and Martin, 2020; Tian et al., 2020). While the underlying cause of pathology remains unclear, below we summarize unique features of SARS-CoV-2 and potential immunological asynchronies that may drive aberrant hyperinflammation (Fig. 1).
Pathogenic human coronaviruses may initially overwhelm the body owing to their efficient entry into lung epithelial cells via ACE2 and rapid replication rates (Chu et al., 2020). ACE2 can also play an immunomodulatory role, serving as a protector against acute lung injury and inflammation in SARS-CoV (Kuba et al., 2005; Li, 2016). Moreover, its more established role as a negative regulator of angiotensin II in the renin-angiotensin system may also lead to pathology, as Ang-II infusion has been shown to increase plasma cytokine levels (e.g., IL-6, IL-1, and TNF-α; Zhang et al., 2009; Eguchi et al., 2018) and aggravate hypertension, the top comorbidity associated with COVID-19 (Richardson et al., 2020).
Under the pressures of the host–virus arms race, many viruses have adapted to use multiple receptors for entry, and recent studies suggest the existence of additional receptors involved in SARS-CoV-2 cellular entry. Neuropilin-1 was found to potentiate SARS-CoV-2 infectivity, though this binding was mainly ACE2 dependent (Cantuti-Castelvetri et al., 2020,Preprint; Daly et al., 2020,Preprint). Recent data revealed the existence of SARS-CoV-2 in some immune cell populations, such as myeloid and T cells, that express little ACE2 (Chua et al., 2020). While viruses may enter immune cells through direct phagocytosis or the antibody-dependent enhancement (ADE) effect, not all immune cells (e.g., T cells) are capable of phagocytosis, and the extent and importance of ADE to immune dysregulation remains unknown (Vabret et al., 2020). Moreover, a recent D614G mutation in SARS-CoV-2 that is uninvolved in ACE2 binding has been shown to increase infectivity (Zhang et al., 2020,Preprint). In light of all this evidence, it is possible that novel SARS-CoV-2 receptors on immune cells could exist and are responsible for virus-induced immune dysregulation. The contribution of glycosylation to virus binding also warrants further investigation, as the high glycosylation profile of SARS-CoV-2 has been well documented, involving more than a dozen of N- or O-glycosylation sites. It has been suggested that the virus spike glycosylation constitutes a “glycan mask” to reduce viral immunogenicity (Walls et al., 2016); however, their roles in viral evasion, receptor binding, and immune modulation remain an unclear but important factor in infection.
After receptor binding, the virus gains access to the host cytosol through intracellular protease–mediated proteolytic cleavage of virus spike protein. While proteases such as TMPRSS2 and cathepsin are used by both SARS-CoV-2 and SARS-CoV (Hoffmann et al., 2020), virus spike protein cleavage by furin is unique to SARS-CoV-2 owing to the presence of a polybasic site at the SARS-CoV-2 S1–S2 junction (Xia et al., 2020). With a broader tissue expression compared to TMPRSS2, which is more restricted to epithelial cells (Vaarala et al., 2001), the ability to use furin may result in increased depth and breadth of SARS-CoV-2 pathogenesis compared with SARS-CoV. Thus, SARS-CoV-2 infection is extended spatially not only by receptor expression but also by protease expression and function. It will be important to systematically explore whether other factors facilitate viral infection through ACE2 or other receptors.
Innate viral sensing and myeloid dysregulation
Coronaviruses have evolutionarily developed mechanisms to counteract innate sensing mechanisms (Merad and Martin, 2020; Perlman and Dandekar, 2005). Studies have shown that SARS-CoV–infected mice exhibit a delayed type I IFN response (Channappanavar et al., 2016), which may allow for initial, high rates of viral replication. Many SARS-CoV proteins (Vabret et al., 2020), some of which are conserved in SARS-CoV-2, act as antagonists for IFN and other innate sensing elements, such as inhibition of IRF-3 or RIG-I (Hu et al., 2017; Kopecky-Bromberg et al., 2007). These mechanisms may explain the low amounts of IFN observed in the peripheral blood and lungs of patients with severe COVID-19 (Blanco-Melo et al., 2020; Hadjadj et al., 2020), although this role is somewhat controversial, as elevated levels have been observed in other reports (Lee et al., 2020; Zhou et al., 2020b). Type I IFN initiated by PAMP signals can further activate inflammatory monocytes, macrophages (Channappanavar et al., 2016), and neutrophils (Middleton et al., 2020; Zuo et al., 2020). Moreover, sustained IFN signaling also up-regulates ACE2 expression (Ziegler et al., 2020), serving as a deadly feed-forward loop exploited by the virus for increased infectivity. Rapid viral replication may activate inflammasome signaling or increase cytopathy of target cells, leading to increased DAMP signaling induced by numerous danger signals (e.g., HMGB1 and heat shock protein), as well as innate immune cytokines (e.g., IL-6, IL-1, IL-18, and GM-CSF). Excessive production of IL-1 and IL-18 may also result from the inflammasome-pyroptosis pathway, as conserved envelope and ORF3a proteins are capable of directly activating the NLRP3 inflammasome (Cagliani et al., 2020; Shah, 2020).
The dysregulation of adaptive immunity
Lymphopenia has also been observed in severe COVID-19 patients (Fig. 2; Vabret et al., 2020). While the cause is unknown, proinflammatory cytokines, such as IL-1, IL-6, GM-CSF, and CSF1 (M-CSF), are capable of directly acting on hematopoietic progenitors, polarizing differentiation to the myeloid lineage in a process called “emergency myelopoiesis,” which may compete with lymphoid differentiation (Boettcher and Manz, 2016). In addition, impaired and delayed type I IFN signaling may hinder recruitment and activation of adaptive immunity (Channappanavar et al., 2014). Some of these cytokines, such as IL-6 and IL-8, have been shown to inhibit dendritic cell priming of naive T cells (Yoshikawa et al., 2009). Of the T cells that remain, a hyperactive and exhausted phenotype with coinhibitory molecule expression (e.g., PD-1, TIGIT, and CTLA-4) is observed (Zheng et al., 2020), with increased populations of potentially pathogenic Th17 cells (Xu et al., 2020). These attenuated yet sustained adaptive immune responses may be pathological, for example via potential activation of bystander T cells that cause organ damage and epitope spreading (Fujinami et al., 2006; Kim and Shin, 2019; Whiteside et al., 2018). Delayed IFN or Fas/FasL signaling may trigger apoptosis of virus-specific T cells in activation-induced cell death, as suggested in SARS-CoV-2–infected spleens and LNs (Chen et al., 2020,Preprint). Other potential mechanisms for lymphopenia include the direct viral infection of lymphocytes (Gu et al., 2005), lymphocyte exhaustion (Diao et al., 2020), relocation to other tissue sites, or suppression from other immune cells, perhaps through interactions with tolerogenic mixed M2 myeloid populations attempting to repair wounded tissue. However, many of these mechanisms are currently speculative, and more evidence is required for further validation. Overall, the suspended arrival of adaptive immunity has many consequences, including ineffective viral clearance along with a failure to temper innate immune responses, as typical innate negative feedback regulators, such as cytokine antagonists IL-1Ra or IL-18BP, are largely induced by IFN-γ released by activated lymphocytes.
Lymphopenia may also cause poor immunological memory, especially regarding memory B cell antibody responses, likely due to lack of T cell help and declining B cell levels in infected patients. In the case of SARS-CoV-2, infected patients have been shown to lose antibody titers within weeks or months after recovery (Liu et al., 2020,Preprint; Long et al., 2020), though this also may be attributed to a naturally contracting antibody response after infection. The full breadth of the antibody repertoire must be carefully evaluated, as not all antibodies generated are beneficial in virus neutralization (Iwasaki and Yang, 2020). Antibodies in SARS-CoV have also been shown to be pathological by skewing macrophage responses, leading to fatal acute lung injury through hypercytokinemia (Liu et al., 2019).
Intrinsic Th2-focused pulmonary immunity, and perhaps the dysregulation of M2-like macrophages and their cytokine profiles (e.g., IL-10 and TGF-β), may have the potential to increase scarring and fibrosis in COVID-19 patients compared with non–COVID-19-related ARDS. The wound healing process is also intimately linked to blood coagulation, as patients with severe COVID-19 often present with elevated levels of D-dimer, thrombosis, and fibrin/fibrinogen degradation products (Connors and Levy, 2020). Key regulators of the coagulation process have been associated with disease and affect the immune system (Gupta et al., 2020,Preprint); for example, the prothrombinase fibrinogen-like protein 2 is induced in murine coronavirus and causes fibrin deposition (Marsden et al., 2003) while also serving as an immunosuppressive factor for regulatory T cell functionality (Shalev et al., 2008). Vascular leakage from the dysfunction of endothelial cells may also contribute to coagulopathy and promote inflammation (Teuwen et al., 2020).
In earlier stages of disease, the focus should be on antiviral therapy, but treatment strategies in the late stage of disease should focus on immunosuppression to alleviate immunopathology, given that CRS and ARDS are the dominant causes of death in severe COVID-19 (Fig. 3). To avoid destroying antiviral immunity, key pathways must be identified to correct the dysregulated immune response, with a special emphasis on targeting hyperactivated innate dysregulation. With the recent modest effect of anti–IL-6 in clinical trials (Atal and Fatima, 2020), we here propose other potential pathways with a mechanistic basis that warrant attention.
Innate immune targets
Correction of the dysregulated innate immune sensing and alarming mechanisms may prove helpful to alleviate COVID-19 pathogenesis, such as inhibitors of the inflammasome and pyroptosis (Yap et al., 2020) or quelling DAMP signaling through agents such as anti-HMGB1 and CD24-Fc (Andersson et al., 2020; Tian et al., 2020). IL-1 family cytokines in particular are up-regulated and act as key mediators of early myeloid overactivation, lung inflammation, and fibrosis (Borthwick, 2016; Conti et al., 2020; George et al., 2020; Huang et al., 2005). Moreover, IL-18 treatment in mouse models causes severe pneumonitis, especially in the context of IL-2 (Okamoto et al., 2002), reminiscent of the features of COVID-19. COVID-19–induced lymphopenia and suppression of IFN-γ secretion may lead to delayed or impaired production of IL-1Ra and IL-18BP, the natural inhibitors of this pathway that are up-regulated in response to IFN-γ. While one study has suggested high levels of IL-1Ra in patients with severe infection (Yang et al., 2020 Preprint), pathology may arise from delayed secretion of receptor antagonists that is missed by collecting data at a single time point. Detailed, longitudinal measurement of free IL-1 and IL-18 and their natural antagonists may provide biomarker candidates for COVID-19 patients and present possible opportunities for therapeutic intervention.
Other pathways that affect myeloid cell activation and cytokine secretion are also worth consideration. Small, conditional RNA sequencing analysis of COVID-19 samples indicated that CSF-1 and CSF-2 may be associated with CRS (Wen et al., 2020). A key growth factor for macrophages, the CSF-1/CSF1-R pathway plays an important role in pulmonary fibrosis, with aberrant expression observed in a subset of COVID-19 patients (George et al., 2020). In addition, IL-8, a proinflammatory chemokine for neutrophils as well as macrophages and lymphocytes, correlated with disease severity in COVID-19 and SARS (Gong et al., 2020,Preprint; Wang et al., 2004; Wong et al., 2004), serving as the rationale for an ongoing clinical trial with anti–IL-8. Efforts are also underway to target other cytokines (e.g., anti-GM-CSF; Lang et al., 2020) and signaling pathways (e.g., JAK-STAT and IRAK4 inhibitors) activated in the innate immune population.
Adaptive immune targets
T cells are critical for viral clearance but also cause pathology. While stimulators of adaptive immunity may be beneficial in the earlier stages of disease to optimize antiviral defense, we advocate for the use of negative T cell regulators in severe COVID-19 patients to combat hyperinflammation in the late stage of disease. Immunosuppression cannot be too general at the risk of destroying antiviral immunity; thus, it is important to specifically dampen the activity of pathogenic T cells while sparing naive T cells. Drugs such as CTLA4-Ig (Abatacept), which blocks CD28 interactions and more affects naive T cells, may not be a good option for COVID-19 treatment. Other coinhibitory pathways, such as PD-1 and LAG-3, are up-regulated on antigen-experienced cells and may be more virus specific. Anti–PD-1 antagonist therapies are currently being tested in a phase II clinical trial, under the rationale that anti-PD blockade may compensate for T cell depletion; however, caution should be exercised regarding anti-PD in COVID-19, as anti–PD-1 treatment has induced lethal pneumonia in cancer patients (Nishino et al., 2015). Although agonists for the PD-L1/PD-1 pathway have not yet been successful, agonists for other coinhibitory pathways, such as LAG-3 and VISTA, exist in the form of agonistic antibody and/or natural ligand (Grebinoski and Vignali, 2020; Prodeus et al., 2017; Wang et al., 2019). Given the pathogenic Th17 signature, RORγt and RORα inhibitors, or anti–IL-17 and anti–IL-17R blockade, may also prove useful for treatment.
Declining B cell numbers and function in patients can potentially be corrected via boosting T cell help (e.g., targeting the CD40/CD40L interaction) or by directly supporting germinal center B cells and long-lived plasma or memory B cell responses (e.g., BAFF or APRIL). However, care must be taken to not increase inflammation while supporting B cell antibody responses, especially as the quality and quantity of the antibody repertoire can be difficult to control. The use of vaccines and neutralizing antibodies has been considered the most promising long-term strategy to prevent further outbreaks of SARS-CoV-2. Dozens of vaccine programs, including intact inactivated and attenuated virus vaccines and virus-based subunits of virion structural proteins or nucleic acids (DNA/RNA) vaccines, are currently underway, some of which have shown initial encouraging data in generating neutralizing antibody titers and/or T cell responses in humans (Jackson et al., 2020; Keech et al., 2020,Preprint; Mulligan et al., 2020; Zhu et al., 2020). However, the quality and quantity of the antiviral antibody response over time awaits further evaluation. The use of convalescent plasma is another approach, although there are significant challenges regarding its supply, quality, and dosing regimen (Abraham, 2020). Other than these approaches, therapeutic antibody programs mostly target the receptor binding domain to block the SARS-CoV-2 spike/ACE2 interaction; however, antibodies targeting spike epitopes outside of the receptor binding domain were found to be functional in decreasing viral infection (Chi et al., 2020). Given that the virus binds immune cells that lack expression of ACE2, there is also an urgent need to identify additional viral receptors, especially those that are functionally involved in COVID-19 immunopathogenesis, for the design of antibody cocktails or multispecific antibodies that block all viral–host interactions. Rational antibody engineering should be undertaken to prevent virus mutational escape (Baum et al., 2020), and antibody affinity, Fc-mediated effector function, and possible ADE should also be considered as important factors in therapeutic design.
Overall, a hyperactive innate immune system coupled with an asynchronized and inadequate adaptive immune system result in a cacophony of severe tissue damage without viral clearance and immunological resolution. From this overall picture of viral pathogenesis, a general strategy for COVID-19 treatment emerges, consisting of controlling the virus at the early stage of disease and taming the immune system at the late stage of disease. The necessity of large-scale and carefully controlled clinical trials to properly identify disease biomarkers and evaluate efficacy of potential therapeutics cannot be overstated. In addition, the discovery of novel receptors for viral entry and immune dysregulation will lead to the reevaluation of proposed neutralizing antibodies and highlights the urgency for further research on critical viral immune interactions to identify key regulators for clinical use.
We thank Drs. Xu Zhou and Siyuan Ding for critically editing the paper and all Jun Wang Lab members for helpful discussion.
Author contributions: J. Wang, T. Zhou, T.T. Su, and T. Mudianto conceived the study and wrote the manuscript.
Disclosures: The authors declare no competing interests exist.
T. Zhou and T.T. Su contributed equally to this paper.