The immunopathogenesis of SARS-CoV-2 infection: Overview of lessons learned in the first 5 years

Abstract

This review provides a broad overview of lessons learned in the five years since COVID-19 was identified. It is a bimodal disease, starting with an initially virus-driven phase, followed by resolution or ensuing inappropriate immune activation causing severe inflammation that is no longer strictly virus dependent. Humoral immunity is beneficial for preventing or attenuating the early stage, without benefit once the later stage begins. Neutralizing antibodies elicited by natural infection or vaccination are short-lived and highly vulnerable to viral sequence variation. By contrast, cellular immunity, particularly the CD8⁺ T cell arm, has a role in preventing or attenuating severe disease, is far less susceptible to viral variation, and is longer-lived than antibodies. Finally, an ill-defined phenomenon of prolonged symptoms after acute infection, termed “long COVID,” is poorly understood but may involve various immunologic defects that are hyperactivating or immunosuppressive. Remaining issues include needing to better understand the immune dysregulation of severe disease to allow more tailored therapeutic interventions, developing antibody strategies that cope with the viral spike sequence variability, prolonging vaccine efficacy, and unraveling the mechanisms of long COVID to design therapeutic approaches.

Introduction

From an initial trickle of cases in Wuhan, China, reported in late 2019, the novel virus SARS-CoV-2 spread rapidly and was declared a pandemic by the World Health Organization by February of 2020. As of November of 2024, the World Health Organization estimated a cumulative death toll from reported cases of COVID-19 to be just over 7 million persons,¹ although the true toll (as reflected by excess mortality) may be double to triple the reported cases.² However, in the few years since the debut of COVID-19, research has progressed at a remarkable pace thanks to reagents and tools developed in response to prior novel infectious outbreaks, including the original SARS pandemic³ and the last major pandemic due to HIV-1 that accelerated advances in virology (rapid isolation and sequencing of viruses), immunology, drug discovery (such as the advent of polymerase and protease inhibitors), and vaccine testing and development.^4,5

Demographic risk factors yield clues to pathogenesis

From the beginning, it was apparent that there is a wide spectrum of outcomes after SARS-CoV-2 infection ranging from absent/minimal symptoms to death from a systemic inflammatory response. As expected from properties of other respiratory viral infections, increasing age, various systemic illnesses/comorbidities (particularly pulmonary disease), and immunosuppressive conditions are predictive of greater morbidity and mortality,^6–9 reflecting the importance of host immunity against infection and end organ vulnerability. Less expected are observations that very young children typically have very mild infection,¹⁰ in stark contrast to influenza,¹¹ while hypertension^6,12 and obesity^12–14 are risk factors for severe illness, suggesting important factor(s) other than direct antiviral immunity in pathogenesis.

Clinical findings suggest 2 distinct disease stages

Early on, clinicians noted that COVID-19 is a biphasic syndrome.¹⁵ For those developing symptoms, a mild upper respiratory illness develops after a brief incubation period of about three to five days,^16,17 leading to a moderate flu-like illness (potentially requiring low-level oxygen supplementation) in many, although some persons recover without symptoms. Most persons recover spontaneously, but about 5 to 7 d after these initial symptoms, a subset progresses to severe disease (Figure 1) with complications such as respiratory failure, renal failure, and/or hemodynamic compromise.¹⁸ Key blood markers, including cytokines related to the inflammatory cascade, correlate to severe disease,¹⁹ indicating that it is mediated by intense inflammation. What determines who progresses is unclear, but aspects of immunity are less developed in young children, and hypertension²⁰ and obesity²¹ have clear links to proinflammatory mechanisms and states.

Responses to therapeutics further confirm distinct immune mechanisms in COVID-19 causing bimodal disease, initially driven by virus but then followed by relatively virus-independent immune dysregulation. The efficacy of antiviral therapeutics against SARS-CoV-2 have been noted to require early administration, during the first phase of illness.²² For example, nirmatrelvir is extremely effective preventing morbidity and mortality in high risk persons during the mild stage of disease²³ but has minimal benefit after the onset of moderate-to-severe illness.²⁴ This has been universally true for antiviral treatments, including neutralizing antibodies that have shown efficacy only in early infection.^25,26 By the time the inflammatory stage of disease is reached, virus in the lungs can be minimal,^27,28 indicating intrinsic immune dysregulation that is no longer predominately virus driven. Therapeutics shown to be effective during this later phase include immunomodulatory agents that reduce inflammation, including corticosteroids,²⁹ a Janus kinase inhibitor,^30,31 and interleukin (IL)-6–blocking antibodies.³² Moreover, early administration of immunosuppression may be associated with worse outcomes,²⁹ suggesting the importance of early immune containment of virus. Thus, it is clear that there are two distinct pathophysiologic phases of disease during which different interventions and immune responses have impact: the first due to direct effects of viral infection and the second due to immune dysregulation leading to inappropriately exuberant inflammation causing end organ damage (Figure 1).

The mechanism(s) leading to immune dysregulation in severe COVID-19 remain unclear, but ineffective early innate immunity is a likely contributor. While the downstream components of a dysregulated immune response have been described,¹⁹ accompanied by pathogenic activation of the clotting cascade,^18,33 the upstream mechanisms triggering this process are poorly understood. Ineffective viral clearance during the acute phase probably plays a key role. Type I interferons (IFNs) suppress SARS-CoV-2 replication,³⁴ but cellular production of type I IFNs appears to be delayed,^35,36 and IFN treatment of patients with early (but not late) infection accelerates viral clearance,³⁷ suggesting that inadequacy of this pathway may be responsible (although type I IFNs may then have a deleterious inflammatory effect in late infection). Other support for the importance of type I IFNs in early viral control includes observations of the development of type I IFN–neutralizing autoantibodies,^38,39 more severe disease when there are genetic defects in the TLR3−/IRF7− pathway,³⁹ and depletion of plasmacytoid dendritic cells that are a major source of type I IFNs.⁴⁰ The mechanism(s) whereby delayed viral clearance then triggers sustained dysregulated inflammation are still to be elucidated. An interesting hypothesis is that the viral spike protein may signal through toll-like receptor-4,⁴¹ leading to activation of inflammatory genes and ensuing production of proinflammatory cytokines.⁴² Another intriguing possibility relates to changes in the gut, including alterations of the microbiome^43,44 and loss of gut-blood integrity,^45,46 with a potential role for direct infection of enterocytes.⁴⁷ How other clinical risk factors contribute mechanistically remains somewhat obscure, such as protective effects of female sex⁴⁸ and possibly (controversially) ABO blood type.^49,50

The inflammatory phase of COVID-19 is marked by global immune dysregulation

Evaluations of this phase of disease have implicated the participation of numerous cell types of both innate and adaptive immunity leading to “cytokine storm” and end organ damage. Several viral components are likely activators of innate immune cells through pattern recognition receptors, such as spike, nucleoproteins, and single-stranded RNA,⁵¹ leading to activation of the nuclear factor κB and excessive release of key macrophage-associated cytokines including IL-6 and tumor necrosis factor.³⁶ IL-8 produced by macrophages or other innate cells is also systemically elevated in severe disease^52,53 and likely recruits damaging neutrophils to lungs.⁵⁴ Among many other reported proinflammatory abnormalities of innate immunity, dysregulated activation of the complement cascade⁵⁵ and reduction of antiviral NK cells⁵⁶ are also key features of severe infection. Abnormalities of adaptive immunity are also observed. In the B cell compartment, there are losses of transitional B cells (precursors of anti-inflammatory B regulatory cells) and reduction of IL-10–producing B cells.⁵⁷ Peripheral blood lymphopenia is consistently observed in severe disease, due to drops in both CD4⁺ and CD8⁺ T cell subsets,^53,58,59 and lack of severe lymphocytic infiltration in effector sites such as the lungs indicates that this may be true depletion rather than redistribution.^60,61 Data on T cells during severe infection reveal a picture of acute activation and exhaustion markers.^62,63

Cytokine production by CD4⁺ T cells likely participate in dysregulated inflammation, potentially including production of IFN-γ,^62,64 granulocyte-macrophage colony-stimulating factor,⁶⁵ IL-6,⁶⁵ and IL-17.⁶⁶ Dysregulation involving the CD4⁺ T cell subset is also suggested by observations that the frequency of CD4⁺ T regulatory cells is decreased in severe infection⁵³ and increased in mild infection.⁶⁷

Therapeutics for COVID-19 target the distinct phases of COVID-19

As mentioned previously, the efficacy of U.S. Food and Drug Administration–approved therapeutic approaches (reviewed in Sagar et al.)²² depends on the 2 distinct stages of disease (Figure 1). The virus-mediated early stage is responsive to antivirals, such as monoclonal neutralizing antibodies, remdesivir or molnupiravir (nucleoside analogues), or ritonavir-boosted nirmatrelvir (viral protease inhibitor). These agents are ineffective once the inflammatory phase of infection begins, when immunomodulatory therapies are indicated. These include the glucocorticoid dexamethasone, the kinase inhibitor baricitinib, and the IL-6–blocking antibody tocilizumab, all of which presumably act through dampening the inappropriate inflammation that causes end organ damage. Notably, such therapeutic interventions during the inflammatory phase of infection are all relatively blunt interventions, given the incomplete understanding of mechanisms behind dysregulated immunity. Future mechanistic investigations with animal models may provide the opportunity to dissect such mechanisms, such as a study showing in SARS-CoV-2–infected rhesus macaques that baricitinib reduces the production of proinflammatory cytokines by pulmonary macrophages.⁶⁸

SARS-CoV-2 infection elicits antiviral humoral immunity

Detecting antibody responses against the virus was an early priority for diagnostic and immune studies. Evaluations have focused on the spike protein, in particular the receptor binding domain (RBD) that mediates binding to the angiotensin-converting enzyme 2 receptor during target cell infection, making it the key target for neutralizing antibodies.^69,70 During acute infection, seroconversion occurs quickly after onset of symptoms, within about 10 to 12 d for IgM and 12 to 14 d for IgG.^71,72 The IgA response has been less studied but appears to have similar kinetics to IgM and IgG.⁷³ The measured quantity of serum anti-RBD antibodies generally correlates to serum neutralizing activity in infected persons, although there is evidence for ongoing somatic hypermutation and increasing potency for months after infection, particularly in combination with vaccination.^74,75 Early administration of monoclonal neutralizing antibodies has demonstrated efficacy in preventing symptomatic infection,⁷⁶ and correlative data have also suggested a contribution of endogenous neutralizing antibodies to resolving symptomatic infection.⁷⁷ As discussed above, passive immunization studies have shown a protective effect only in the early phase of infection. Antibody titers are markedly higher with in persons with severe infection,^72,73,75 indicating a virus-dependent expansion in later infection (rather than contribution to viral control), and even perhaps a deleterious contribution to inflammation.^78,79

SARS-CoV-2 messenger RNA vaccines also elicit antiviral humoral immunity

Two messenger RNA (mRNA) vaccine designs among those widely released early in the pandemic have had continued usage to date, including versions of BNT162b2⁸⁰ and mRNA-1273.⁸¹ These both encode the spike protein modified to stabilize its prefusion conformation to enhance neutralizing antibody responses.^82,83 Initial administration of these vaccines comprised a priming dose followed 21 or 28 d later by a boosting dose. In antigen-naïve persons, a low plateau of anti-RBD antibodies is achieved about 14 d after the first dose, and a higher plateau is reached about 6 d after the second dose; these levels roughly correspond quantitatively to peaks after mild or severe COVID-19.⁷⁵ Neutralizing antibodies, however, do not develop to levels comparable to natural infection until after the second dose,⁷⁵ indicating the need for qualitative maturation of the humoral response. In persons who had previously recovered from COVID-19, the first dose of an mRNA vaccine appears to drive antibody levels to a maximal concentration and high neutralizing activity,⁷⁵ reflecting the contribution of a potent recall response.

Spike sequence variation is a limitation faced by humoral immunity

Escape of SARS-CoV-2 from antibodies has been a key barrier for the efficacy of monoclonal antibodies and vaccines. The virus has undergone rapid and dramatic shifts in the sequence of spike into divergent lineages, with a succession of variants (denoted by the Greek letters) from Alpha to the most recent Omicron strains.⁸⁴ These strains clearly demonstrated the vulnerability of antibodies to sequence variation; for example, the Beta strain that emerged by mid-2020 (B.1.351) was insensitive to the first-generation monoclonal antibody bamlanivimab and neutralization by serum antibodies elicited by the first-generation vaccines.^85,86 Vaccine failures (“breakthrough infections”) were especially notable when the more deadly Delta variant became predominant by mid-2021.^84,87 Importantly, the spike mutations defining distinct strains cluster in the RBD that is the key target for neutralizing antibodies,^69,70 but the predominant selective pressure for these variants was most likely adaptations to become more transmissible after the leap from an animal host into humans (and binding the human angiotensin-converting enzyme 2 receptor), rather than immune selective pressure.^88,89 However, viral escape mutation has been observed within persons receiving monoclonal antibody therapy,⁹⁰ and immune-driven evolution within persons in some circumstances could play a role.⁹¹ Interestingly, the combination of natural infection and mRNA vaccination drives so-called hybrid immunity that results in antibodies with greater breadth that can often recognize new variants to which a person has not been exposed.^74,92,93 Regardless, the mRNA vaccines are now routinely updated to mirror the most common circulating strain(s) of SARS-CoV-2.⁹⁴

Humoral immunity from natural infection or vaccines has limited durability

The earliest evaluations of antibody persistence after infection were conflicting, with some observing rapid decay^95–97 and others suggesting stability.^98–100 Observations of rapid decay led some investigators to predict that immunity is short-lived, calling into question “immunity passports” or the possibility of herd immunity.^95,96 Such predictions were met with hostility and accusations calling these reports “questionable results” leading to “alarmist news media articles” and insistence on the “known” concept that most viral infections give prolonged immunity.¹⁰¹ Even the lay press weighed in on such arguments,¹⁰² citing long-lived spike-specific memory B cells,^74,103 without the context that the presence of memory itself is not necessarily a correlate of protection. However, even early in the pandemic it was already known from prospective human volunteer challenge/rechallenge experiments with the common cold coronavirus 229E elicits neutralizing antibodies that wane rapidly to allow reinfection before a year,¹⁰⁴ further confirmed in a large observational study showing that common cold coronavirus reinfections are common within a year.¹⁰⁵ It was also already known that antibodies against influenza vaccination have a half-life of about 26 wk, with near total decay within a year.¹⁰⁶ Anecdotal reports of reinfection within months after COVID-19 had already emerged very early in the pandemic.¹⁰⁷ Thus, even very early on, there were clear clues that immunity would be short-lived, and that predictions of short-lived immunity were neither unreasonable nor alarmist even without the benefit of hindsight. It was not at all surprising that early predictions of short-lived immunity were borne out, with vaccine-generated antibodies decaying similar to natural infection⁷⁵ and breakthrough infections happening with increasing frequency in the months after vaccination even in the absence of escape due to a strain shift.¹⁰⁸ It is now the norm that revaccination is recommended at least annually (every 6 mo for high-risk persons), with some experts advocating that all persons should be revaccinated every 6 mo.

Cellular immunity is elicited by SARS-CoV-2 infection

Given that antibodies are effective only in the early phase of infection, cellular immunity came to the forefront as a key protective factor against later severe illness. An early clue for the importance of T cell responses against the virus was the observation that an overall T helper 1 (Th1) profile correlates with better disease outcome,⁵³ followed by increasing observations that SARS-CoV-2–specific T cell responses correlate to resolution of infection and improved outcomes,^109–111 particularly the CD8⁺ subset.^59,110,111 Virus-specific T cells are detectable as soon as two to four days after symptom onset,^110,112 and early induction of T cells (but not antibody responses) is associated with attenuated infection.¹¹⁰ During active infection, the T cell response is skewed towards the CD4⁺ arm¹¹³ that is acutely Th1 biased^114–116 in keeping with an antiviral response, but it has been suggested that this proinflammatory phenotype becomes maladaptive in severe illness, when an anti-inflammatory profile of CD4⁺ T cells producing IL-10 has a beneficial regulatory role.¹¹⁷ The influx of CD8⁺ T cells in lungs is associated with moderation of lung damage,⁶⁰ indicating the importance of antiviral cellular immunity. After clearance of infection, the postacute memory T cell response appears to be a significant contributor to protection from reinfection.¹¹⁵ T cells can target most viral proteins¹¹⁸; among the structural proteins, the targeting of nucleocapsid is immunodominant.¹¹⁶ The implications of targeting in effective antiviral activity are unclear; one study showed that diverse targeting was superior for control of disease,¹¹⁹ although mRNA vaccines can elicit only spike targeting yet are protective. A role for targeting is suggested by the observation that certain HLA types may influence susceptibility to symptomatic infection and disease severity.^120,121 There is evidence that some advantageous T cell responses against SARS-CoV-2 may be primed by cross-reactive memory T cell responses against other previously circulating coronaviruses,^122,123 which may also be HLA dependent, as shown by the association of HLA B*1501 with presentation of cross-reactive spike protein epitopes from common cold coronaviruses OC43-CoV and HKU1-CoV and attenuated disease.¹²⁴

Cellular immunity is elicited by SARS-CoV-2 mRNA vaccines

Given that the mRNA vaccines lead to cellular translation of spike, they elicit both CD4⁺ and CD8⁺ T cell responses.¹²⁵ Evidence for the protective role of vaccine-elicited cellular immunity against severe disease is provided by the observation that mRNA vaccines afford partial protection from disease after a single dose in SARS-CoV-2–naïve persons, as T cell responses are generated before the generation of neutralizing antibodies,^126–128 reaching significant peak within the first 7 to 10 d after vaccination of prior unexposed persons.¹²⁸ Direct comparison of the immunodominance profiles has shown similar patterns of spike targeting between infection and mRNA vaccination.¹²⁸ Despite commonly occurring breakthrough infections due to failure of vaccine-induced antibodies from waning levels or viral spike variation, the vaccines have remained protective against progression to severe disease, underscoring the importance of vaccine-elicited cellular immunity independent of humoral immune protection. For example, antibodies from the original vaccines had very poor coverage of the Delta variant, yet mortality was still markedly diminished in those who had been vaccinated.¹²⁹

Cellular immunity against SARS-CoV-2 appears to be more durable than humoral immunity

Because cellular immunity is more technically challenging to quantify than antibodies, the durability of SARS-CoV-2–specific T cells is not well defined. Ultimately, the protection from severe infection after natural infection¹³⁰ or vaccination¹⁰⁸ does wane within several months in epidemiologic observational studies. It does appear that T cell responses elicited by natural infection are more stable in early convalescence^128,131,132 compared with antibodies,^95–97 but estimates of the half-life of antiviral T cell responses have varied dramatically. One interesting caveat of measuring cellular immune responses has been a unique property of responses to mRNA vaccines, which peak and disappear within days after each vaccination, when measured by some standard cytokine assays.^128,133 This is not seen after natural infection, or adenoviral vector–based SARS-CoV-2 vaccinations,¹²⁸ which yield responses that last months when detected via the same methodology. However, vigorous memory is apparent after brief ex vivo antigenic stimulation, illustrating the inadequacy of assays such as enzyme-linked immunosorbent spot and intracellular cytokine staining to predict memory T cell function and persistence in vivo.¹²⁸ Nonetheless, such memory assays do suggest that there is decay of cellular immunity after natural infection, and that T cells targeting nucleocapsid are immunodominant and longer-lived compared with those targeting spike.¹¹⁶ This raises the possibility that inclusion of nucleocapsid in mRNA vaccines could give more potent and long-lived cellular immunity.

SARS-CoV-2 sequence variation has less impact on cellular immunity

Unlike humoral immunity, the evolution of viral variants has had little impact on effectiveness of cellular immunity. With regard to spike-specific T cell responses, as mentioned previously the emergence of variants such as Delta evaded antibodies elicited by the first-generation vaccines, yet vaccinees were still protected from severe illness and death.¹²⁹ The main explanation is that the polymorphisms determining different variants tended to be clustered in and around the RBD that is the key target for neutralizing antibodies, which can also be readily affected by minor changes in tertiary structure. By contrast, most mRNA vaccines have responses against several epitopes distributed across all regions of spike,¹²⁸ and even the most divergent Omicron variant is still about 95% conserved (about 1,218 of 1,273 amino acids) compared with the original Wuhan sequence, again with changes mostly clustered in and around the RBD.¹³⁴ Consequently, most targeted T cell epitopes are conserved between spike variants. Relevant to T cell responses against nonspike viral proteins, other viral proteins such as nucleocapsid have demonstrated far more modest sequence evolution compared with spike, again suggesting that nucleocapsid might be a better vacine antigen for eliciting protective cellular immunity.¹³⁵

Other vaccine designs have been utilized during the pandemic

Beyond the mRNA vaccines discussed previously that were released early in the pandemic, several others saw usage across the world. These broadly included whole inactivated SARS-CoV-2 (e.g., Coronavac and Covaxin), spike-expressing recombinant adenoviruses (e.g., AD26-COV.S and AZD1222, utilizing human adenovirus serotype 26 and a chimpanzee adenovirus, respectively), and recombinant spike subunits (e.g., NVX-CoV2373). A detailed comparison of these vaccines is beyond the scope of this review, but these vaccines vary in key considerations of efficacy, safety, and logistics for production and distribution. Relative efficacy of different vaccine approaches is not well established, given the lack of controlled comparative trials. Given the previously mentioned importance of CD8⁺ T cell responses against SARS-CoV-2, mRNA and adenovirus vaccines have the theoretical advantage of directly accessing the HLA class I pathway to generate a CD8⁺ T cell response,¹²⁸ which may be reflected by the apparently lower protection offered by the inactivated virus vaccines.¹³⁶ The adjuvanted recombinant subunit NVX-CoV2373 vaccine, however, also generates cellular immunity¹³⁷ and seems to have comparable efficacy to the mRNA vaccines.¹³⁸ Another efficacy consideration is the need for repeated vaccinations, given the short duration of protective immunity. A theoretical consideration of repeated vaccination with adenovirus-vectored vaccines is that vector immunity from earlier exposure could dampen the desired response against the spike protein,¹³⁹ although that does not appear to have been borne out.¹⁴⁰ Each of the implemented vaccine strategies has appeared to be safe overall, with very rare incidences of serious adverse events, but these events have varied by vaccine type, such as cerebral venous sinus thrombosis being observed specifically in adenovirus-vectored vaccines.¹⁴¹ Finally, different vaccine designs vary in their logistical considerations. For example, classic whole virus inactivated vaccines or recombinant subunit vaccines are very inexpensive to manufacture, and can be stored at 4 °C versus the requirement for storage at −20 °C or −80 °C for the mRNA vaccines.

Variants of SARS-CoV-2 likely vary in their pathogenicity

As mentioned previously, the species leap of SARS-CoV-2 has resulted in shifts of viral sequence into defined genetic variants, predominately due to selection for greater transmissibility in humans as a new host. The pathogenicity of these variants has varied, with the clearest comparison showing the Delta variant to have significantly greater mortality risk than the preceding strains.^84,87 Since that observation, clearly delineating and comparing the pathogenicity of different variants has become increasingly difficult, as most persons are no longer naïve to infection and/or vaccines and have widely variable exposures and pre-existing immunity. Viral shifts may be decelerating as the virus nears optimization of human transmission, with only the Omicron variant and its subvariants dominating persistently after a quick succession of other variants the first 2 to 3 yr of the pandemic.⁸⁴ Animal data support the clinical impression that Omicron is less pathogenic than prior variants, perhaps due to less tropism for the lower respiratory tract.¹⁴² How differing pathogenicity between variants may be determined by other immune mechanisms, such as differing efficiencies triggering inflammatory pathways, is unclear given the limited understanding of the mechanisms determining these processes.

“Long COVID” has emerged as a concerning sequela of COVID-19

While this issue is complex and itself the topic of several lengthy reviews, some high points will be mentioned here. A significant percentage of persons clearing acute SARS-CoV-2 infection have lingering symptoms for several months or longer duration. The incidence appears to be related to the severity of acute illness, with estimates from roughly 20% of nonhospitalized cases¹⁴³ to 60% of hospitalized cases,¹⁴⁴ and a recent estimate of overall prevalence in the United States was perhaps 6.9% of all adults.^145,146 A major challenge to diagnosing and studying long COVID is the remarkable diversity of reported symptoms, spanning over 200 symptoms in practically every organ system, ranging from neuropsychiatric to cardiovascular to pulmonary to gastrointestinal involvement.¹⁴⁶ Some manifestations include objectively defined disease processes such as insulin-resistant diabetes mellitus¹⁴⁷ and postural orthostatic tachycardia syndrome,¹⁴⁸ and also more less definable symptoms or syndromes such as myalgic encephalomyelitis/chronic fatigue syndrome.¹⁴⁹ It thus seems likely that there is no single disease mechanism and that “long COVID” is a constellation of multiple disease processes with varying degrees of overlap, albeit all consequences of the initial infection. With regard to immunologic processes, a plethora of immunologic abnormalities have been reported, such as highly activated innate immune cells with persistently elevated type I/III IFNs¹⁵⁰ and elevated proinflammatory innate immune cytokines,¹⁵¹ reduced conventional dendritic cells,¹⁵² reduced and exhausted T cells,¹⁵² increased IL-6–secreting CD4⁺ T cells,¹⁵² expanded cytotoxic CD8⁺ T cells,¹⁵³ various autoantibodies,¹⁵⁴ and reactivation of herpesviruses.¹⁵⁵ Thus, there is a puzzling mix of possible immune hyperactivity and immunosuppression. What drives such immunologic processes is unclear, whether continuing abnormalities from acute infection despite viral clearance, or perhaps ongoing viral persistence in tissue reservoirs.^156,157 Given our lack of understanding, there are no proven treatments for long COVID, although prior vaccination¹⁵⁸ and antiviral treatment of acute infection¹⁵⁹ apparently reduce its incidence.

Conclusions

Infection with SARS-CoV-2 results in an initially asymptomatic to mild virus-driven illness, which in some persons progresses to a highly dysregulated inflammatory state that results in severe disease involving end organ damage. Early viral control, through immunologic (innate immune involving type I IFNs, antibodies, T cells) or therapeutic (antiviral medications) means is critical for preventing progression to severe disease. After the onset of moderate-to-severe disease, viral control has minimal impact, and some antiviral immune responses may even have deleterious effects.

Both natural infection and various vaccines elicit neutralizing antibody responses, but these have limited durability and are prone to escape due to variations in the spike protein. These two factors make reinfections and/or vaccine “breakthrough” infections common. However, vaccines still provide excellent protection from severe illness and death even when the virus is not controlled by humoral immunity, likely due to cellular immunity that limits virus, independently of antibody neutralization.

Cellular immune responses to natural infection and most vaccines are responsible for preventing severe illness or death in persons who develop symptomatic infection. These responses are weighted toward the CD4⁺ subset, although the CD8⁺ subset may have a more direct antiviral role. Virus-specific T cells also have limited persistence but probably better durability than antibodies, and are less prone to escape by spike protein sequence variation. While current vaccines encode only spike, targeting of other viral proteins such as nucleocapsid may provide more immunodominant and durable cellular immunity.

Outstanding questions still remain. A clearer understanding of the mechanisms of immune dysregulation in severe disease is needed to develop more tailored therapeutic interventions than the current broadly immunosuppressive treatments such as corticosteroids or IL-6 blockade. Establishing strategies for vaccine generation of broadly neutralizing antibodies covering spike variation is an important goal; the observation that “hybrid immunity” from combined infection and vaccination results in neutralizing antibodies against new variants suggests this is a possibility. Similarly, having monoclonal antibodies that span spike variability would be very useful for preventive and therapeutic use; monoclonal antibodies to date have been generally variant specific. Devising strategies to generate longer-lived protection from vaccines, through more persistent antibody and/or cellular immune responses, is an important aim. Greater immunodominance and persistence of cellular immune responses against nucleocapsid versus spike suggests that vaccine designs incorporating other proteins may be beneficial. However, the generally short-lived responses of natural immune responses against SARS-CoV-2 suggest that creating a vaccine to provide long-lived immunity is a tall order, given that the vaccine would have to improve upon nature. Finally, unraveling the processes leading to long COVID, including the contributions of immunologic mechanisms, will be key to developing interventions to reduce the ongoing and growing toll of the pandemic.

Acknowledgments

The author is grateful to the support of many colleagues for helpful discussions about the immunopathogenesis of COVID-19, including F. Javier Ibarrondo, Paul Krogstad, Nicole Tobin, and Grace Aldrovandi. Impetus to produce this review was also provided by an anonymous reviewer’s comment that this author “would do well to go back to basics of immunology and antiviral immunity. I suggest reading the following review article by Dr. Baumgarth and colleagues,”¹⁰¹ in response to a submitted manuscript regarding the decay of vaccine-elicited neutralizing antibodies against SARS-CoV-2.

Author contributions

Otto O. Yang (Conceptualization [lead])

Funding

None declared.

Conflicts of interest

O.O.Y. has served on the scientific board and as a consultant for CytoDyn Inc., for which he has received financial payments and stock options; and served on the board of directors for Applied Medical Corporation, for which he has received financial payments and stock.

References