The role of senescent T cells in immunopathology

Abstract The development of senescence in tissues of different organs and in the immune system are usually investigated independently of each other although during ageing, senescence in both cellular systems develop concurrently. Senescent T cells are highly inflammatory and secrete cytotoxic mediators and express natural killer cells receptors (NKR) that bypass their antigen specificity. Instead they recognize stress ligands that are induced by inflammation or infection of different cell types in tissues. In this article we discuss data on T cell senescence, how it is regulated and evidence for novel functional attributes of senescent T cells. We discuss an interactive loop between senescent T cells and senescent non‐lymphoid cells and conclude that in situations of intense inflammation, senescent cells may damage healthy tissue. While the example for immunopathology induced by senescent cells that we highlight is cutaneous leishmaniasis, this situation of organ damage may apply to other infections, including COVID‐19 and also rheumatoid arthritis, where ageing, inflammation and senescent cells are all part of the same equation.


| INTRODUC TI ON
As we age, we accumulate cells in many organs that exhibit signs of DNA damage, have poor proliferative capacity and are highly secretory. These cells are senescent, defined as being in a state of cell cycle arrest associated with phenotypic and functional changes (Campisi, 2013;Campisi & Fagagna, 2007). This process is primarily thought to prevent cancerous transformation of dividing cells (Braig et al., 2005;Chen et al., 2005), but senescence plays a vital role in developmental biology as well as in wound healing/tissue repair (Demaria et al., 2014;Krizhanovsky et al., 2008;Kurundkar et al., 2016;Pitiyage et al., 2011;van Deursen, 2014). While transient senescence is a beneficial mechanism earlier in life, the accumulation of senescent cells with increasing age leads to organ dysfunction, driving inflammation and may underlie many age-related diseases such as atherosclerosis , osteoarthritis (Jeon et al., 2017), neurodegenerative diseases (Chinta et al., 2018;Kritsilis et al., 2018) and cirrhosis (Wiemann et al., 2002).
Senescence is triggered by activation and subsequent maintenance of DNA damage repair (DDR) signalling pathways. This can by induced by oncogene activation, replicative stress related to telomere erosion and oxidative stress (Allsopp, 2008;Blackburn et al., 2015) that kickstart a DDR. This involves proteins such as γ-H2AX, ATM, ATR, p16 and p53 (Nassour et al., 2016) that leads to cycle arrest which becomes permanent if repair fails. Importantly, while senescent cells do not proliferate, they are still metabolically active and secrete a wide array of cytokines, chemokines and matrix metalloproteinases that vary according to the cell type involved (Coppé et al., 2010). It is this senescence-associated secretory phenotype (SASP) that is thought to play a crucial role in age-related pathology such as the chronic low grade inflammation observed with advanced age, called inflammageing (Franceschi et al., 2000). Therefore, two recognized characteristics of senescent cells are the lack of proliferative activity and the presence of a senescence-associated secretory phenotype (SASP, Campisi, 2013).
While senescence was first discovered in fibroblasts and extensively worked on in other non-leukocytic cells, it has become increasingly clear that immune cells undergo senescence as well.
Within the immune system, the existence of non-proliferative leukocyte populations that have high capacity for biologically active mediator secretion has been recognized for many decades, albeit under a different name. These are the effector T lymphocytes that secrete pro-inflammatory cytokines and cytotoxic granules but do not proliferate after activation . Recent studies show that these cells also harbour DNA damage, short telomeres, low telomerase activity and engage signalling pathways associated with cellular senescence (Callender et al., 2018;Henson et al., 2014;Lanna et al., 2014). Therefore, the terms effector T cells and senescent T cells may be synonymous and refer to the same T cell populations. Effector T cells that are expanded in number during the acute phase of a human anti-viral response to Epstein-Barr Virus (EBV) are highly expanded in number and lose the expression of the anti-apoptotic molecule Bcl-2 that makes them short-lived (Akbar et al., 1993).
These cells also have relatively long telomeres due to upregulation of the enzyme telomerase and are in cell cycle (Maini et al., 1999).
The difference betwen such short-lived effector cells and senescent T cells in the steady state is that the latter are non-proliferative, do not express telomerse after activation, exhibit short telomeres and express senescence related signalling molecules as described above.
Nevertheless senescent cells in the steady state in both CD4 and CD8 compartments still express lower levels of Bcl-2 that the other subsets. Collectively this indicates that the extent of T cell proliferation the acute phase of a viral infection drives T cells to senescence.
These cells are still susceptible to apotosis but can persist if sufficient antiapototic cytokines are tissue niches. It can be argued that senescent T cells derive from a subpopulation of effector T cells that do not undergo apoptosis, instead becoming senescent and lingering long term. In this review, we discuss recent observations suggesting that certain infectious agents can drive the accumulation of CD8 + T cells that exhibit all the hallmarks of senescence as described above (henceforth referred to as T sen ). We will focus on patients infected with Leishmania braziliensis where the increased T sen numbers in the skin may induce the characteristic skin lesions associated with this disease. We propose a hypothesis, bases on existing data, that this pathology occurs because senescent non-lymphoid cells in the skin are be killed by infiltrating T sen, a novel interaction between senescent lymphoid and non-lymphoid cells that may also have implications for ageing.
2 | FUN C TI ONAL PROPERTIE S OF CD8 + T s e n CD8 + T cells can be subdivided based on expression of the costimulatory molecules CD27 and CD28, where naïve CD8 + T cells co-express both markers and as they differentiate to an effector phenotype lose CD28 expression and subsequently CD27 (Henson et al., 2012). Sometimes these will be subdivided further into T EMRA cells, so named because they re-express CD45RA (Henson et al., 2012). The CD27 − CD28 − compartment harbours T sen cells which can be identified further by expression of KLRG1 and CD57 (Brenchley et al., 2003;Henson et al., 2009). Moreover, many studies have shown a plethora of markers that are upregulated on T sen cells, such as downregulated telomerase, shortened telomeres, DNA damage responses and constitutive MAPK activity, features specific to T sen over effector T cells. T sen also exhibit a SASP consisting of proteases and pro-inflammatory mediators such as TNF and IL-1β (Callender et al., 2018). These functional changes are summarized in Table 1 ( Akbar et al., 1993;Brenchley et al., 2003;Callender et al., 2018Callender et al., , 2019Geginat et al., 2003;Gumá et al., 2004;Henson et al., 2009Henson et al., , 2012Henson et al., , 2014Henson et al., , 2015Libri et al., 2011;Maini et al., 1999;Ouyang et al., 2003;Pereira et al., 2020;Plunkett et al., 2001;Tarazona et al., 2001Tarazona et al., , 2002Voehringer et al., 2002;Weng et al., 2009). Single cell RNAseq analysis of peripheral blood derived CD8 + T cells show that CD8 + T cells identified by the above functional criteria also express multiple features associated with cellular senescence (Table 1, Pereira et al., 2020). This indicates that the terms senescent and effector CD8 + T cells are synonymous and identify identical populations.
CD8 + T sen have limited proliferative capacity after activation via the T cell receptor (TCR) complex (Akbar et al., 2016) that is due in part to downregulation of key TCR signalling molecules such as LCK, LAT and PLC-γ (Pereira et al., 2020). CD4 + T sen also exhibit identical markers of senescence to their CD8 + counterparts  but are present at much lower frequency that CD8 + T sen in the peripheral blood of healthy donors. The development of senescence characteristics in T cells was initially considered to indicate the dysfunction of these cells during ageing (Akbar et al., 2004). However, CD8 + T sen populations express surface receptors that are associated with NK cells such as NKG2D, NKG2C, NKG2A and Killer immunoglobulin-like receptor (KIR) families compared to undifferentiated and non-senescent CD28 + CD27 + CD8 + T cells (Abedin et al., 2005;Michel et al., 2016;Pereira et al., 2020;Vallejo et al., 2011). This suggests rather than being dysfunctional, these cells acquire an alternative functional profile as they differentiate towards senescence. This is supported by observations that these T sen express DAP12, an NK cytotoxicity adaptor molecule, and are capable of killing tumour cell lines in an MHCI-independent manner (Pereira et al., 2020). Mechanistically, the switch from TCR to NKR expression in CD8 + T sen is regulated by stress proteins known as the sestrins in both humans and mice (Pereira et al., 2020). Of note, CD4 + T sen also express NKRs, suggesting that they may also mediate effector functions through these receptors (Pereira et al., 2020).

| CD8 + T s e n ACCUMUL ATE DURING AG EING AND PER S IS TENT VIR AL INFEC TI ON
Previous reports have shown that CD8 + T sen accumulate in older humans and that this is largely driven by ongoing sub-clinical responses to persistent viruses especially cytomegalovirus (CMV) (Jackson et al., 2017;Nikolich-Žugich et al., 2020;Savva et al., 2013).
Moreover, this population has increased capacity to rapidly respond to signals mediated by inflammatory cytokines (Freeman et al., 2012). Furthermore, proinflammatory cytokine production associated with persistent CMV infection may induce bystander senescence in non-CMV specific T cells . Therefore the cytotoxic CD8 + T sen populations that accumulate during ageing and during persistent viral infections are now re-focussed to preferentially recognize NK ligands that may be expressed by infected, malignant or "stressed" tissues (Pereira et al., 2019;Sagiv et al., 2016).
Therefore, T sen can mediate NKR dependent cytotoxicity, independently of their antigen specificity.

| THE K ILLING OF S ENE SCENT NON -LYMPHOID TISSUE CELL S BY NK AND CD8 + T CELL S
Senescent non-lymphoid cells accumulate in many tissues during ageing (Campisi, 2013) and are associated with tissue dysfunction that lead to frailty. Transgenic mouse models that enable the specific removal of senescent cells from different tissues in vivo show that the elimination of these cells increases lifespan, improves overall fitness and reduces age-associated characteristics of the animals (Baar et al., 2017;Baker et al., 2011Baker et al., , 2016). An exciting observation was that cells of the immune system, including NK, CD4 + and CD8 + T Concurrently, these cytokines activate the endothelium which upregulates adhesion markers like E-selection. This binds to CLA expressing senescent T cells enabling their infiltration into the skin.

cells, can also recognize and eliminate senescent cells in vitro and
Here, these senescent T cells, expressing a host of NKRs including NKG2D interact with resident skin cells. The interaction between NKG2D + T sen and MICA/B + fibroblasts results in killing of the latter and contributing to off-target tissue pathology.
parasite clearance (Rossi & Fasel, 2017). There is a strong correlation between the severity of the disease and the number of CD8 + T cells present in the lesion (Faria et al., 2009;Santos et al., 2013), but this is independent of the presence of parasites in the lesions (Carvalho et al., 2007;Murray et al., 2005;Pearson et al., 1996).
Despite this, if untreated, the lesions increase in size progressively causing the characteristic pathology of CL. This raises the question as to the cause of the lesions in the skin suggesting the possibility that chronic inflammation and non-specific CD8 + T or NK cell cytotoxic responses may lead to non-specific tissue destruction.

| HYP OTHE S IS: THE NON -S PECIFIC NATURE OF CD8 + T s e n INDUCE S IMMUNOPATHOLOGY DURING INFEC TIONS SUCH A S CUTANEOUS LEIS HMANIA S IS
We hypothesize that dermal macrophages infected by Leishmania initiate an inflammatory response that preferentially attracts L. braziliensis specific CD8 + T sen from the blood which clear the infection. However, the inflammation in the skin, induced by the interaction between infected macrophages and the infiltrating T cells, induces the expression of stress ligands including those that bind to NK receptors by the surrounding stroma (Gasser et al., 2005;Groh et al., 2001;Molinero et al., 2004). The interaction between these ligands and NK receptors on T sen leads to cytotoxic killing and tissue damage. In turn, this damage leads to exacerbated inflamma-

| FUTURE PER S PEC TIVE S
The investigation of the cellular infiltrate in patients with cutaneous leishmaniasis by gene expression (Amorim et al., 2019;Christensen et al., 2016) and histological analyses is essential to determine the presence of alternative ligands for CD8 + T sen within the tissue. It is also crucial to identify the source of inflammation at these sites, are the myeloid cells, the stromal cells, the leukocytes or all of them involved? The hypothetical positive feedback loop described here that involves T sen and excessive inflammation and non-specific tissue damage may also be involved in other non-resolving inflammatory diseases where accumulation of T sen cells have been demonstrated including Chagas disease (Molina & Kierszenbaum, 1989) and malaria (Cockburn et al., 2014). The accumulation of senescent T cells with increased pro-inflammatory potential has been implicated during age-related diseases such as rheumatoid arthritis (Goronzy et al., 2013;Weyand et al., 2017), Alzheimer's (Gate et al., 2020), and cardiovascular diseases (Youn et al., 2019;Yu et al., 2015), where the non-specific tissue damage driven by T sen cells needs to be investigated and might offer new opportunities for prevention and treatment (Andersson et al., 2011). Furthermore, it would important to know if inflammation that is induced by infections such as SARS-CoV-2 that have a disproportionate pathological impact in older individuals (Merad & Martin, 2020) results from non-specific T sen recruitment and induction of non-antigen specific damage in the lung and other tissues. Furthermore, it is currently unclear how senolytic drugs that are being used to clear senescent cells will affect senescent leukocyte populations and if this may have beneficial or detrimental consequences. Collectively we suggest that T cells that have been driven to extreme differentiation and senescence by infectious agents acquire potent MHC-I-unrestricted cytotoxic and capacity for inflammatory cytokine secretion together with the ability to cause non-specific NK-related damage. Given the increased burden of tumours and infections with age, the expansion of NK celllike functions in CD8 + T cells could be an advantageous adaptation that would enable the recognition and elimination of infected and transformed cells. However, the accumulation of senescent T cells during ageing is a double-edged sword that may induce pathology, especially in inflammatory environments.

ACK N OWLED G EM ENTS
Authors would like to thank the Fundação de Amparo a Pesquisa do

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S ' CO NTR I B UTI O N S
All four authors conceived and wrote the manuscript.