Since the immune system is constituted of multiple cell types, it is expected that different cells cross-talk to each other to perform specific functions. As discussed in the previous section, CTLs can execute their killing functions locally through direct contact with target cells in an MHC-I dependent manner using a wide-spectrum of effector molecules ( Figure 2A ). However, they should be present in sufficient numbers at peripheral tissues to control the pathogen, which is not the case prior to infections. To circumvent such dilemma, following infection, T cells migrate to non-lymphoid tissues and differentiate into tissue-resident non-circulating memory T cells (T_RMs). Both antigen presentation and cytokines are required for the differentiation of T_RMs. For instance, in mice, naïve cells require cross-talking to DNGR-1⁺ dendritic cells (cDC1, CD103⁺ CD8a⁺ DCs) for the generation of T_RMs in response to Flu and Vaccinia viruses (48, 49). This type of communication also involves IL-12, IL-15, and CD24 co-stimulation signals as well (50–54). In humans, the cross-talk of CD1c⁺ DCs with naïve CD8 T cells plays a pivotal role in generation of T_RMs in a TGF-β-dependent manner (55). Further, both effector T cells (T_EFF) and central memory T cells (T_CM) have the capacity to differentiate into T_RMs (56). However, the question remains: “how such small numbers of T_RMs still can control pathogen dissemination”.

One way to overcome this challenge is to start communicating with other immune cell types. As shown in Figure 2B , upon antigen and cytokine stimulation, memory T cells rapidly express IFNγ and chemokines promoting recruitment and activation of innate myeloid and lymphoid cells including monocytes and NK cells. These cells further amplify the recruitment of memory cells via expression of CXCL9/10 chemokines resulting in the generation of systemic and/or tissue broad anti-microbial state ( Figure 2B ). The Masopust lab and others spearheaded elegant studies to examine this model (57–61). For example, Schenkel et al. showed that T_RMs were able to recruit bystander circulating memory CD8 T cells to peripheral tissues through VCAM-1/IFNγ axis by using an OT-I or P14 chimeric mouse model. In this model, naïve OT-I or P14 CD8 T cells were adoptively transferred to B6 mice followed by VV-OVA or LCMV infection respectively. To activate T_RMs at the female reproductive tract (FRT), OVA or LCMV-specific peptides were injected transcervically (t.c). Concurrently, there was an upregulation of VCAM-1 (a4β1-CD49d) on vascular endothelium and IFNγ by reactivated T_RMs. These events were associated with recruitment of OT-I specific CD8 T cells (bystander) in response to LCMV infection, which was blocked by neutralization of VCAM-1 or IFNγ. Further, the recruitment of additional immune cell types, including B cells, to FRT as well as the activation of innate cells such as DCs and NK cells were observed (57, 58). Along the same lines, it has been shown by Soudja et al. that memory CD4 and CD8 T cells plays an essential role in orchestrating activation of splenic innate immune cells following secondary infections in an IFNγ dependent manner (60). In conclusion, T_RMs can deploy their cytotoxicity indirectly through recruitment of a wide-range of immune cell types acting as guardians of peripheral tissues in case of reinfection.

Indirect cytotoxicity has been further supported by other studies using tumor vaccine mouse models. For example, Kalinski’s lab demonstrated that CD8 T cells can act as de fecto helper cells supporting an effective anti-tumoral DC immunological response (62). In these studies, they showed that the adoptive transfer of autologous DCs loaded with poorly immunogenic MC38 tumor lysate to the animals bearing the same tumor was only marginally effective. However, the inclusion of OVA257–264 epitope into the vaccine supported the generation of MC38-specific CTL responses in wild-type B6 animals and tumor clearance. These data suggest that non-specific CD8 T cells could play an indirect cytotoxic role through harnessing the killing capacity of antigen-specific T cells via unknown mechanism(s). Similar data were also obtained in a model of wild-type mice harboring memory responses against LCMVgp33–41, a dominant epitope of a natural mouse pathogen, where the inclusion of LCMVgp33–41 peptide strongly enhanced the induction of CTLs against MC38 tumors. These vaccines not only elevated CTLs’ function against the poorly immunogenic MC38 adenocarcinoma but also against the highly immunogenic OVA-expressing EG7 lymphoma (62). These studies provide a mechanistic insight into the role of memory CD8 T cells in enhancing the anti-tumoral effect, which suggest indirect cytotoxic function. However, additional studies are required to determine which T cells are responsible for the killing of the tumor, is it the tumor-specific, non-specific (bystander), or both?

The helper function assumed by T cells has been extended to encompass the role of CD8 T cells in inducing antibody production by B cells and their involvement in killing tumors, pathogen clearance and autoimmunity. More than 30 years ago, the Le Gros lab showed that polyclonal stimulation (PMA/Ionomycin + IL-2 + IL-4) of total CD8 T cells isolated from murine lymph nodes (LNs) resulted in (1) down regulation of CD8α, (2) decrease in cytotoxicity, (3) downregulation of IFNγ and perforin, (4) upregulation of T_H2 cytokines IL-4, IL-5, and IL-10, and (5) help for B cells to produce IgG antibodies (63). The authors took their analyses one step further and examined whether this phenomenon was MHC-I restricted. Indeed, stimulating CD8 T cells with MHC allo-antigens in the presence of IL-4 resulted in non-cytolytic phenotype (63). Co-culturing these activated cells with autologous B cells resulted in the secretion of IgG antibodies in the culture supernatant. This early study put CD8 T cells at a crossroad with antibody producing B cells, which underlined a possible indirect cytotoxic role of CD8 cells in the pathogenesis of autoimmunity.

Later on, extensive body of literature discussed the existence of T follicular helper CD4 cells (CD4 Tfh cells) and their role in providing help to B cells for antibody production (64–70). Similar to CXCR5⁺ PD1⁺ CD4 Tfh cells, CXCR5⁺ CD8 T cells exhibit a B cell helper function, where they support antibody production either (1) through a direct interaction with B cells (71–74) or (2) via enhancement of CD4 T cell-B cell interaction (75). Indeed, upon TCR stimulation, these cells upregulate CD70, OX40 and ICOS molecules, which are required for T cell dependent humoral responses.

The indirect cytotoxic function of CXCR5⁺ CD8 T cells in antibody production and B cell support had been demonstrated in various disease states. For example, in gastric cancer, the accumulation of CXCR5⁺ CD8 T cells in the tumor is associated with better patient overall survival (OS) (76). Similarly, IL-21 producing CXCR5⁺ CD8 T cells accumulate in the hepatocellular carcinoma tumor tissues in close proximity to CD19⁺ B cells, which predicts better disease prognosis (71). These studies raise the question: what type of cross-talk is taking place within the tumor microenvironment. One can predict interaction between B cells and CXCR5⁺ CD8 T cells. Indeed, co-culturing these cells with B cells resulted in enhanced in vitro differentiation of B cells as well as an increase in IgG and reduction in IgM production. Hence, the indirect cytotoxic role of CD8 T cells against tumors could be explained by helping B cells to produce antibodies, which in turn bind tumor cells and recruit NK cells to initiate antibody-dependent cell cytotoxicity (ADCC). In another study, CXCR5⁺ ICOS⁺ CD8 T cells had been shown to infiltrate tumoral lymph nodes (LNs) in Hodgkin lymphoma (HL) (77). This subpopulation was shown to upregulate the expression of IL-2, IL-4 and IL-21, key cytokines for antibody production and B cell support. However, they showed weak expression of effector molecules including GzmB, perforin, and IFNγ. Similarly, CXCR5⁺ PD-1⁺ ICOS⁺ CD8 T cells isolated from nasal polyp tissue promote antibody production when co-cultured with B cells (78). Along the same lines, the Youngblood lab showed elegantly that HIV-specific CD8 T cells isolated from Elite controllers (ECs) expressed high levels of CXCR5 transcript compared to ART-suppressed non-controllers (79), which suggested the protective role of CXCR5⁺ CD8 T cell in EC patients. Further, the in vitro stimulation of CD8 T cells isolated from ECs with HIV-specific peptides (gag) upregulates CXCR5 (80).

The indirect cytotoxic role of these cells had been further described in autoimmune diseases. For instance, in an autoimmune hemolytic anemia murine model, Valentine et al. demonstrated a significant increase of CXCR5⁺ PD1⁺ CD8 and CD4 T cells in secondary lymphoid tissue early during pathogenesis (75). The two subpopulations upregulated ICOS, IL-21 and Bcl-6. However, treating the mice with CD8 and CD4 depleting antibodies resulted in increased survival, improved anemia, reduced B cell survival and decreased anti-erythrocyte IgG autoantibodies, suggesting the pathogenic role of these cells. Thus far, these data support the potential protective indirect cytotoxic function of T cells in the context of tumor development and viral control, while pathogenic in case of autoimmune diseases. Hence, it is important whether to harness or inhibit these indirect cytotoxic functions in the context of cancer or autoimmunity, respectively.

In line with previous studies highlighting the capability of the transcription factor Stat5 in negatively controlling CD4 Tfh cells and maintaining B cell tolerance (81, 82), Chen et al. demonstrated that the deficiency of Stat5 in CD8 T cells led to an increased autoantibody production in Ig ^HEL sHEL transgenic mice. This deficiency resulted in an increase in germinal center B cells and expansion of CXCR5⁺ PD-1⁺ CD8 T cell population after an acute viral infection. These data suggest that Stat5 negatively control CXCR5 PD1 CD8 T cell population as well (72). In conclusion, CD8 T cells can provide help to B cell resulting in enhancement of antibody production.

The specific cell surface molecules and cytokines expressed by CD8 T cells involved in B cell support and antibody production had been explored by Shen et al. (74). The authors demonstrated that CD8 T cells that localize to B cell follicles in tonsils and LNs express CXCR5 (74). Further, CXCR5⁺ CD8 T cells upregulate CD40L and ICOS, while polyclonal stimulation of these cells resulted in increased expression of IFNγ, IL-4 and IL-21 (74). Additionally, co-culturing TCR stimulated CXCR5⁺ CD8 T cells with autologous B cells resulted in increased production of antibodies, where this phenomenon was completely abolished by blocking either CD40L or IL-21. Finally, Loyal et al. defined a CD40L⁺ helper CD8 memory subpopulation that expresses IL-6 receptor and lacks the cytotoxicity surface marker SLAMF7 (83). Ironically, this indirect cytotoxic mechanism seems to be a double-edged sword in a way where antibodies can protect against pathogens, or kill tumor cells but also they can precipitate autoimmunity and induce a self-damage.

Alloreactive memory CD8 T cells are considered as the main drivers of allograft rejection through their cytotoxic machinery (106, 107). However, Krausnick et al. demonstrated a non-cytotoxic role of CD8 T cells in regulating the alloimmune response during lung transplantation (108). In this study, the authors showed that B6 CD8 depleted mice or B6 CD8^-/- mice acutely reject their pulmonary allografts from Balb/c mice with a significant inflammatory infiltration in the grafted lungs. Further, the adoptive transfer of wild-type B6 CD8 T cells into immunosuppressed B6 CD8^-/- recipients restored tolerance to BALB/c lung allografts. The authors took their analyses one step further and performed a mixed lymphocyte reaction (MLR) to further understand the role of CD8 T cells in this model. In these analyses, they observed that CD8 T lymphocytes isolated from tolerated BALB/c→B6 lung allografts, but not spleens could inhibit proliferation of B6 congenic CD4⁺ and CD8⁺ T lymphocytes (responders) in the presence of BALB/c splenocytes (stimulators). These findings suggest that CD8 T cells with regulatory capacity accumulate in lung allografts enhancing graft tolerance, where a large proportion of CD8 T cells infiltrating tolerated lung grafts acquire an IFNγ⁺ central memory phenotype.

To further understand this phenomenon, the authors pretreated recipient mice with IFNγ–neutralizing antibody or Ifnγ^–/– animals were used as hosts. Surprisingly, they observed a break in tolerance and graft rejection. Injection of Ifnγ^–/– CD8 T cells into CD8^–/– mice failed to rescue BALB/c lung allografts from rejection, despite costimulatory blockade (108). Additionally, the authors observed that trafficking of these central memory CD8 T cells was chemokine dependent. Indeed, injection of Pertusis toxin treated CD8 central memory cells (which irreversibly inactivate Gαi-coupled chemokine receptors) into immunosuppressed B6 recipient of Balb/c lung impaired migration of central memory cells to the lung. To this end, the obvious question is: How does IFNγ act to prevent rejection? Is it a signal related to the lung microenvironment or intrinsic to the T cells? The authors took their study one step further and showed that IFNγ exerted its regulatory effect via a Nitric oxide (NO)-pathway. In fact, inhibition of iNOS abrogated the suppressive capacity of T cells. Hence, they showed that NO was essential in allowing graft acceptance and maintaining tolerance locally. Taken together, these data demonstrated the non-cytotoxic role of host CD8 T cells in lung allograft tolerance in an IFNγ dependent-manner. This work provided a deep insight into the role of CD8 T cells in regulating the alloimmune response in a non-cytotoxic manner, albeit such role seems to be milieu dependent as similar functions were not shown in other transplanted tissues.

Early after an acute myocardial ischemic attack, lymphocytes and macrophages migrate to the necrotic myocardial area (111). Infiltrating immune cells phagocytose the necrotic debris and initiate the scar tissue formation (111). Along the same lines, Curato et al. demonstrated that a subset of CTLs play a key role in this process (110). In this study, the authors showed that CTLs expressing the Angiotensin II receptor (AT2R) are protective against myocardial ischemia through upregulation of the immunomodulator cytokine IL-10 and downregulation of the proinflammatory cytokines such as IFNγ. Further, co-culturing post-ischemic AT2R⁺ CD8 T cells with adult cardiac myocytes resulted in significantly lower apoptotic rate when compared to AT2R^- CD8 T cells. Adding neutralizing IL-10 antibodies to the co-culture led to an increase in the cardiac myocyte apoptotic rate in both AT2R⁺ and AT2R^- CD8 T cells, suggesting that IL-10 is critical for the non-cytotoxic cardioprotective effect of AT2R⁺ CD8 T cells. These findings highlighted the role of AT2R⁺ CD8 T cells in locally regulating cytokine expression, skewing it towards a reparative profile. Furthermore, the protective effect of this population was emphasized by the adoptive transfer of AT2R⁺ CD8 T cells in cardiac tissue, which reduced myocardial ischemia. Thus far, this process reduces the bystander inflammatory injury to the healthy myocardial tissue, maintains cardiac myocyte viability, and prevents one of the most drastic post infarctions sequalae, which is autoimmunity against cardiac proteins and possibly Dressler syndrome. Although the authors demonstrated the pivotal non-cytotoxic role CD8 T cells in this disease model, the means by which they are recruited to the necrotic area is still to be determined. A possible mechanism could be via the release of damage associated molecular patterns (DAMPs) by necrotic myocytes that might activate infiltrating macrophages, which in turn create a chemokine rich niche that helps recruitment of CD8 T cells to the site of injury.

Later on, interest has increased to further understand the involvement of CD8 T cells in various tissue repair mechanisms. Indeed CD8 T cells have been shown to play an important role in post-traumatic skeletal muscle regeneration. Although the muscle repair process depends namely on progenitor satellite cells and anti-inflammatory macrophages, the recruitment of CD8 T cells to the inflammatory microenvironment suggests a crucial role for these cells in the regenerative process (112). Using cardiotoxin induced mouse skeletal muscle injury model, Zhang et al. demonstrated that depletion of CD8 T cells impaired skeletal muscle regeneration and increased scar formation by excessive matrix deposition (109). Consistently, adoptive transfer of CD8 T cells into CD8 knockout mice improved myofibroblast size and inhibited matrix deposition post cardiotoxin injury. CD8 knockout mice have limited recruitment of the Gr1^high anti-inflammatory macrophages, which are essential for skeletal muscle repair (113), into the inflammatory environment leading to a reduction in the number of satellite cells (109). CD8 T cells were further shown to play a key role in the recruitment of Gr1 ^high macrophages through the secretion of MCP-1 ( Figure 4A ). The mechanism through which CD8 T cells are recruited to the injured skeletal muscle tissue is still to be identified to provide basis for a therapeutic regenerative model.

CD8 T cells can interact with various cell types of their innate counterparts, orchestrating and fine tuning the immune response. For instance, one study demonstrated that blood circulating memory CD8 T cells, as opposed to the tissue effector CD8 T cells, have a reduced cytotoxic ability towards DCs. These cells were shown to characteristically express GzmB and perforin at a lower level. They were shown to confer a helper signal to DCs mediated by IFNγ, supporting the production of IL-12p70, a key cytokine for a Th1 immune response. Memory CD8 T cells help protecting antigen presenting DCs from the cytotoxic killing by effector T cells through the upregulation of the endogenous anti-granzyme protease inhibitor-9 (PI-9) in a TNFα dependent-manner (114) ( Figure 4B ). This provides a feedback mechanism that optimizes an effective antigen presentation and allows for a stronger immune response where potentiation of antigen presentation has a multitude of clinical implications in the area of anti-microbial and cancer vaccines.

Consciousness, the state of internal and external awareness of a living-being, remains a controversial topic among scientists and philosophers. Although it is not completely understood how conscious the immune system is, one way to explain it is through the cross-talk between wide-spectrum of immune and non-immune cells. For instance, soluble mediators such as chemokine gradients guide different immune cell types from one organ to another. Additionally, cells can crosstalk to each other through receptor-ligand interactions. Whether there is an additional means by the cells to remain conscious is yet to be discovered. Along the same lines, once memory T cells see a foreign antigen, they undergo a rapid transition to a highly activated proliferative state. Consequently, they become “conscious” of the chemokine gradient and hence, they migrate to the site of infection to clear the pathogen. For instance, TCR activation of naïve T cells results in down regulation of CCR7 and CXCR4 chemokine receptors and upregulation of inflammatory chemokine receptors including CCR3, CCR5, and CXCR3. Consequently, they acquired the capacity to migrate to inflamed tissues (115, 116). During the migration of activated memory T cells, a portion of them surprisingly migrate to antigen-free lymph nodes (117, 118) yet, the biological significance behind this route of trafficking is still enigmatic. Along the same lines, a recent study discussed the migration of unconventional T cell cells (UTCs) from peripheral tissues to draining LNs (119).

This scenario raises the question: “what is the function of memory T cells following pathogen clearance? Do they have any role(s) during homeostasis?”. One possibility could be a cross-talk between activated memory CD8 T cells and naïve CD8 T cells that continuously patrol between the periphery and LNs (120–122). Indeed, recently our lab showed that activated memory CD8 T cells acquire a novel non-cytotoxic function by which they can interact and influence the phenotype and transcriptome of naïve CD8 T cells. In this scenario, they acquire two states (1) an activated/memory T cell state and (2) a unique hybrid state between naïve and effector/activated memory cells (123) ( Figure 4C ). Since both cell populations were sorted from the same healthy subject, we speculate that activated memory T cells are presenting self-antigen to naïve T cells generating auto-reactive T cells. These findings may explain the non-cytotoxic functions of activated memory T cells and their contribution for the rise of autoimmunity following vaccination or transplantation.