When one considers immune responses to viruses in general, there is broad acceptance, largely based on studies on rodents and some studies on infections in experimentally studied primates, that whereas protection from infection requires neutralizing antibodies, viral clearance is primarily dependent on the elimination of infected cells by cytotoxic CD8⁺ T cells. Indeed, studies on T-cell transfer and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) challenge in rhesus macaques have shown that, as for other viruses, CD8⁺ T cells are key for protection.¹

Naïve CD8⁺ T cells are activated to yield cytotoxic effector T cells and CD8⁺ memory T cells. There are a few distinct CD8⁺ memory T-cell subsets currently recognized; these include central memory, effector memory, tissue-resident memory (T_RM), and stem-like memory T cells (Figure 9.1). Central memory T cells home to secondary lymphoid organs (lymph nodes, the spleen, and Peyer patches)
and recirculate following the same path as naïve T cells. Effector memory T cells home to the tissue sites of infection but can still return to secondary lymphoid organs. These two memory cell subsets are sustained by self-renewing stem-like memory T cells. Some memory T cells that arrive at the tissue site of infection differentiate into cells that acquire the ability to be retained at that tissue site. These cells are called T_RM cells and are likely able to self-renew.

Activated CD8⁺ T cells are part of broader Type I immune responses that include interferon (IFN)-γ-producing Th1 CD4⁺ T cells, natural killer (NK) cells, and ILC1 cells. For some viruses, CD4⁺ cytotoxic T lymphocytes (CD4⁺ CTLs) that emerge in the context of Type I immunity are relevant as well and these cells will be discussed later in the SARS-CoV-2 context. B-cell lymphoma 6 (Bcl-6)-expressing T follicular helper (T_FH) cells represent another key CD4⁺ T-cell subset of relevance in viral infections as well, because these cells, found only in lymphoid organs, are required for germinal center formation and for the generation of high-affinity antibody responses and robust B-cell memory.

The unprecedented numbers of patients with coronavirus disease 2019 (COVID-19) and the worldwide interest this disease generated among immunologists, virologists, and other scientists resulted in an enormous depth of investigation over a relatively short period of time. It is probably fair to say that human immune responses have been studied at a depth in this disease not seen prior for any other human viral infection, not even human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS).

Did the study of COVID-19 help provide new insights about T-cell responses to viruses? The reality is that although many of the core principles of antiviral T-cell immunity have been obtained from decades of studies in rodents, human studies of T-cell immunity in viral infections, especially severe viral infections, had hitherto been quite limited. The very large numbers of patients seen in many centers across the globe and the enormous degree of scientific interest in the study of COVID-19 have undoubtedly solidified knowledge as well as provided some new insights about T-cell responses to viruses in humans. Although the case has frequently been made for the protective relevance of T cells in COVID-19² and this is not disputed, all rigorous correlates of protection so far in human studies of COVID-19 and vaccination relate to antibodies, especially neutralizing antibodies.^3,4 It is important to emphasize, relatively early in this chapter, that although it is safe to assume that T cells do contribute to immune protection against SARS-CoV-2 (and this will be elaborated upon later), there are no rigorous correlates of protection mediated by T cells against this virus that have been established in the context of either infection or vaccination.⁵

In this chapter, we will initially review knowledge about preexisting cross-reactive T-cell immunity and other factors that may explain why some people develop severe, life-threatening COVID-19. We will then discuss T-cell responses in patients with mild-to-moderate COVID-19—essentially patients who typically did not require hospitalization or admission to intensive care units and did not develop a cytokine storm/septic shock–like phenotype. We will then discuss T-cell responses in severe COVID-19 and separately consider memory and T-cell immunity linked to vaccination and hybrid immunity. We will conclude by exploring whether there have been new insights gained about T cells in viral infections that can inform both prophylactic and therapeutic strategies in future pandemics.

The most severe clinical presentations of COVID-19 and the highest mortalities were observed at a time when prior exposure to this virus had not occurred and populations had not been vaccinated. Although the importance of specific adaptive immunity in reducing disease severity is well established, the possibility was considered, even early in the pandemic, that severity might have been influenced by previous adaptive immune exposure. An early hypothesis that attempted to explain susceptibility to severe COVID-19 was that previous exposure to common cold coronaviruses may have resulted in cross-reactive T cells that contributed to immune protection and thus resulted in protection from infection or resulted in mild or moderate COVID-19. This hypothesis has gained support from studies showing that asymptomatic Individuals preferentially inherit HLA-B*15:01. This allele presents a peptide from Spike that is conserved across all variants of SARS-CoV-2 and differs by only one amino acid from a very similar peptide in two common cold coronaviruses HKU.1-CoV and OC43-CoV. Cross-reactive high affinity T cells have been identified and are the likely source of protection from symptoms.⁶

In initial studies examining T-cell responses in both unexposed individuals and in COVID-19 patients, the approach used was an activation-induced marker (AIM)-based assay gating on T cells that expressed activation-induced proteins such as CD69, OX40, and CD137, after peripheral blood mononuclear cells were exposed to pools of antigenic peptides from SARS-CoV-2.^{7, 8 and 9} These and related studies on prepandemic samples¹⁰ suggested that between 20% and 50% of unexposed individuals have T cells, especially CD4⁺ memory T cells, directed against the common cold coronaviruses HCoV-43, HCoV-229E, HCoV-NL63, and HCoV-HKU1. However, although early AIM assay–based studies suggested that preexisting cross-reactive T cells were mainly of the memory CD4⁺ phenotype, subsequent studies on a range of prepandemic samples using more specific functional cytokine release–based approaches and the use of peptide-major histocompatibility complex (MHC) tetramers have suggested a lower degree of cross-reactivity and also indicated that most cross-reactive T cells are actually of a naïve phenotype and also include CD8⁺ T cells.^{11, 12 and 13}

It is, however, possible that many earlier studies may have failed to detect relevant preexisting cross-reactive memory CD8⁺ T cells because they explored limited portions of the SARS-CoV-2 peptidome. One particularly interesting study that supports the relevance of preexisting T-cell memory to protection from COVID-19 was performed on health care workers (HCW) in the UK.¹⁴ In this study, overlapping peptides of both structural and nonstructural SARS-CoV-2 proteins were used in conjunction with enzyme-linked immunosorbent spot (ELISpot) assays to detect memory T cells. Subjects with overt COVID-19 were compared with seronegative HCWs (SN-HCW). Subjects in this latter group were repeatedly negative when tested by polymerase chain reaction (PCR) and serology, but many in this group had preexisting memory T cells against the SARS-CoV-2 replication transcription complex, especially against the viral RNA polymerase itself. These proteins are highly conserved, at the primary sequence level, across coronaviruses. In this SN-HCW group, many subjects with strong reactivation of T cells against RNA polymerase exhibited concomitant transcriptional induction of IFI27, a very robust innate marker suggestive of an abortive infection.¹⁵ These data strongly suggest that individuals with robust preexisting T-cell memory expand this memory pool whenever they are infected with SARS-CoV-2; in many, this pool of memory T cells is protective and results in abortive subclinical and “sub-serologic” infections. However, although preexisting memory T cells may attenuate infection, no correlation between the absence of preexisting memory T cells and susceptibility to severe disease has been established.

It was noted early in the pandemic that very severe disease (characterized by hospitalization, intensive care unit [ICU] admittance, septic shock, organ failure, or respiratory failure) and mortality were linked to older age, male gender, and underlying conditions that included obesity, Type II diabetes, cardiovascular disease, and chronic lung disease. Although it has been argued that the relative absence of naïve T cells in the elderly might have made them more susceptible to COVID-19, there is at present no firm epidemiologic or immunologic data to argue that there are differences in preexisting or initially generated virus-specific T-cell immunity in broad swaths of susceptible and nonsusceptible individuals (obese versus nonobese, diabetic versus nondiabetic, old versus young, men versus women) that support the notion that a relative lack of T-cell immunity drives the propensity for the development of severe COVID-19. It is also worth noting that although polymorphic differences in human leukocyte antigen (HLA) genes have been linked to many diseases, susceptibility to severe COVID-19 or to any aspect of this disease, in spite of the very large numbers studied, had until recently, not shown any HLA linkage. The data described above with the preferential inheritance of HLA-B*15:01 in asymptomatic individuals has established that CD8+ T cells are likely important in protecting from symptom acquisition and disease progression. Indeed, a completely different point of view has developed and will be elaborated on in this chapter, suggesting that it is susceptibility to COVID-19 that results in severe disease and consequently a temporal window within which T-cell immunity is defective.

We live with viruses and other microbes in a state of equilibrium. However, the commensals that we harbor constantly elicit effective innate immune responses as well as adaptive immune responses, so our immune systems are never really “at rest.” Most commensal viruses that do enter and replicate in our mucosal epithelial cells are initially prevented from expanding because of the antiviral state induced by Type I IFNs (and Type III IFNs as well, not discussed here). Type I IFNs are synthesized by virally infected cells themselves, largely triggered by cytosolic sensors, but these cytokines are also made by plasmacytoid dendritic cells (pDCs). The latter cells are capable of producing and secreting large amounts of Type I IFNs but do so after viral capture into endosomes where viral nucleic acids trigger endosomal toll-like
receptors (TLRs). These Type I IFNs bathe neighboring cells and induce the expression of a very large number of IFN-stimulated genes. The numerous products of these genes include PKR, OAS2, MX1, IFITM1/2 and 3, tetherin, and many, many other proteins that contribute to the antiviral state.^16,17

Apart from generating the antiviral state in neighboring cells, Type I IFNs also activate DCs, inducing higher levels of the CD80 and CD86 costimulatory ligands, enhance antigen presentation, and contribute to the strength of activation of naïve T cells. Type I IFN signaling can also enhance cross-presentation on conventional DCs (cDCs) and pDCs, inhibiting lysosomal acidification and diverting antigens to the MHC class I pathway.

Pathogens, by definition, have evolved mechanisms to evade innate immunity. Most viruses that are linked to severe disease have the ability to partly or almost completely compromise the induction of the antiviral state by the host innate immune system. About a third of the 29 proteins encoded by the SARS-CoV-2 genome compromise Type I IFN production by infected cells and the subsequent generation of the antiviral state. In addition, because viruses are typically cleared by cytotoxic CD8⁺ T cells, many pathogenic viruses have also evolved the ability to compromise the presentation of viral peptides on host MHC class I molecules. SARS-CoV-2 is no exception.¹⁸

Susceptibility to severe COVID-19 is best understood in the context of genetic defects in the Type I IFN pathway or the presence of autoantibodies to Type I IFNs.^{19, 20 and 21} The attenuation of the production of or responses to Type I IFNs allows SARS-CoV-2 to replicate more efficiently and establish itself in the lungs. These data, taken together, represent an important advance in the study of antiviral susceptibility, and this knowledge has come from attempts to better understand why some individuals are more susceptible to severe COVID-19 than others.

The failure of the host to harness an effective Type I IFN response results, in susceptible individuals, in both the buildup of large amounts of viral antigens and also a more severe inflammatory response that creates an altered milieu for T-cell activation. The accumulation of virus and the resultant cell damage and/or cell death result in the overexuberant activation of neutrophils, monocytes, and macrophages by pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), leading to the development of acute respiratory distress syndrome, cytokine storm, and progressive lung damage. It is this antigenic abundance combined with an altered inflammatory milieu for the initial activation of naïve T cells and the subsequent reactivation of effector cells in infected tissues that likely causes many of the changes in T-cell immunity that have been observed in severe COVID-19 as we will discuss later.

The studies on Type I IFNs in the context of susceptibility to severe COVID-19 have raised many interesting issues. Does the endogenous Type I IFN “tone” differ between men and women? TLR7 has been shown to escape X inactivation in immune cells,²² and the possibility has been considered that transcripts from human endogenous retroviruses might activate TLR7 and establish higher basal levels of Type I IFN in women, offering them some baseline protection from severe viral infection (but possibly making them more susceptible to autoimmunity). Although genetic variants in genes linked to the Type I IFN pathway may reduce this basal Type I IFN “tone,” presumably this “tone” is also reduced with age, though this is not an issue that has been adequately studied. A very small fraction of the individuals who develop autoantibodies to Type I IFNs have a known AIRE mutation; the AIRE protein is required in the thymic medulla to induce tolerance to proteins in the periphery and these include Type I IFNs. It is, however, not well understood why, in the absence of any known germline mutations, aging also results in a break in self-tolerance and thus somehow contributes to an increase in autoantibodies against Type I IFNs, more so in men than in women.²¹

The major questions of interest in the context of “moderate” COVID-19 infections in nonimmunocompromised individuals, typically diagnosed in an outpatient setting, are the following: How do T-cell responses in moderate COVID-19 differ from those seen in severe COVID-19? How long does T-cell memory in COVID-19 last? Are T-cell responses in COVID-19 protective?

One of the main messages from the pandemic that can be taken by immunologists who look at the human immune system in the context of viral infections is that the milieu generated by the nature of the initial innate immune response can in somewhat unpredicted ways determine the
nature of the T-cell response. Another way in which relatively new knowledge accrued during the era of COVID-19 was a consequence of the more recent emergence of tools like high-dimensional flow cytometry, mass cytometry, and single-cell RNA sequencing in human studies that allowed a deeper analysis of the antiviral T-cell response.

Overall, the “normal” expected antiviral immune response was best studied in the initial (prevaccination) year of the COVID-19 pandemic when (apart from studying severe disease) investigators also interrogated mild-to-moderately ill patients, essentially outpatients who presented with smaller elevations of acute-phase reactants like C-reactive protein and fibrinogen. In these patients, many studies described the fairly robust activation of CD4⁺ T cells and CD8⁺ T cells (as mentioned earlier using either AIM assays or peptide-MHC tetramers). Viral peptides were mainly derived from the S, M, and N proteins or from a number of selected open reading frames (ORFs).

The vast majority of human studies on cellular immunity in COVID-19 have been performed on circulating T cells. Mild COVID-19 more readily induced antigen-specific T-cell responses than severe COVID-19²³ (Figure 9.2). Indeed, activated antigen-specific T cells such as circulating PD-1⁺ ICOS⁺ CD4⁺ T cells (that correspond best to precursors of T_FH cells in lymph nodes) and activated CD38⁺ HLADR⁺ CD4⁺ and CD8⁺ T cells were seen transiently in the blood of acutely ill patients with moderate disease, prior to their recovery.^24,25 When examining dominant T-cell responses to a specific peptide form of the N protein in an HLA B*07:02 context, functionally active antigen-specific cytokine-secreting CD8⁺ T cells were robustly generated in mild disease as compared with poor responses in severe disease.¹¹ These data, taken together, raised the possibility that T-cell responses were protective in mild/moderate COVID-19. The protective relevance of T-cell memory has been, however, best established by the study of the immunodeficient, particularly cancer, patients who have been studied in the largest numbers. The T-cell responses in most solid cancer patients who have had COVID-19 were similar to the responses seen in patients without cancer, and survival correlated well with the expansion of activated CD8⁺ T cells that synthesize and secrete IFN-γ. The most telling results have come from hematologic cancer patients who have undergone B-cell depletion therapy.²⁶ The majority of these patients recovered after generating T-cell responses to the virus, supporting the notion that T-cell responses are protective. However, death rates linked to SARS-CoV-2 are higher in patients lacking B cells—so clearly B cells also protect, and T cells alone provide incomplete protection.²⁷