Two Strikes and You’re Out? The Pathogenic Interplay of Coinhibitor Deficiency and Lymphopenia-Induced Proliferation

A system of inter- and intraclonal competition for available resources for T cell survival serves to govern the number of T cells in a host as well as maintain TCR diversity (1, 2). Perturbations that induce a state of lymphopenia, classically defined as an abnormally low number of blood lymphocytes, result in a glut of available resources for T cells to exploit. Such resources include peptide:MHC (pMHC) complexes (3–9), including self-pMHC, which can provide a tonic signal through the TCR, as well as homeostatic cytokines, particularly IL-7 (1, 10–21). Thus, T cells that either survive an initial lymphopenia-generating stimulus, are transferred to a lymphopenic host, or are newly generated posthematopoietic stem cell transplants will undergo a process of lymphopenia-induced proliferation (LIP) to fill the available niche as defined by these resources. LIP is associated with acquisition of a memory phenotype and effector function (22). Importantly, LIP is linked to inflammatory disease, such as immune reconstitution inflammatory syndrome after rebound of the CD4 T cell compartment in antiretroviral drug–treated HIV patients (23, 24) or colitis associated with LIP of regulatory T cell (Treg)-depleted T cell populations (25). Reconstitution of the T cell compartment of an otherwise healthy lymphopenic host does not normally result in an overt autoimmune phenotype, thus the occurrence of immune pathology in this context has been considered as a two-hit phenomenon requiring a cofactor (26). In the case of HIV immune reconstitution inflammatory syndrome, this cofactor could be Ags associated with comorbid infections for which a large proportion of the rebounding CD4 T cells have high affinity, whereas in the aforementioned colitis model it would be a deficiency in Tregs.

Our group (27, 28) and others (29, 30) have reported that defects in T cell–expressed molecules with coinhibitory function or other inhibitory roles in TCR signaling represent additional cofactors that can drive autoimmunity in league with LIP. However, at least in the case of PD-1 deficiency we have found that a third predisposing factor, newly generated T cells, must be present for overt autoimmunity (27). Furthermore, the age of the lymphopenic host and availability of lymphoid stroma influence autoimmune potential, with neonatal or lymph node–deficient hosts being less susceptible. Herein we review factors that have been reported to influence the extent of LIP at various stages of T cell ontogeny and the occurrence of autoimmunity.

Naive CD4 and CD8 T cells expand in response to TCR:pMHC interaction and/or IL-7 and fail to expand significantly in MHC-deficient lymphopenic hosts (5, 8, 31–35), whereas central memory T cells require only cytokines (IL-7 or IL-7+IL-15) (36) [although pMHC can still influence their LIP (37)]. Efficient LIP requires lymphocytes to gain entrance to secondary lymphoid organs (SLOs), particularly lymph nodes (38–40), where specialized IL-7 producing cells known as fibroblastic reticular cells are situated, along with pMHC-bearing dendritic cells (DCs). The concept of competition between T cells for space in the SLOs was demonstrated (39) by using pertussis toxin to block entry of competitor T cells into the T cell zones of SLOs. Thus, lymph nodes represent an important and finite resource reservoir, particularly for CD4 T cell survival and LIP. The role of lymph nodes can confound interpretation of LIP studies that employ common γc–deficient recipients, as their lymph node deficiency is rarely recognized.

Variances in ability to compete for resources between T cell clones would be expected to influence their rate of LIP. Accordingly, the intensity of the signal resulting from TCR:pMHC interaction is considered to be key (22, 41). Various TCR transgenic (Tg) T cells have been shown to undergo differential LIP (9, 37, 42, 43). The level of T cell CD5 expression has been reported to correlate with the intensity of the positively selecting TCR:self-pMHC interaction in the thymus (44), as well as the ability of some (43, 45–47) but not all (37) TCR Tg T cells to undergo LIP in a host lacking the cognate Ag. LIP in polyclonal T cells showed that some cells would undergo very rapid proliferation upon transfer into a lymphopenic host (spontaneous proliferation), in contrast to a proportion of cells that remained CFSE positive during the experimental timeframe (<∼7 divisions). This gave rise to the idea that two forms of LIP exist: spontaneous or endogenous proliferation, and slow or homeostatic proliferation. The fast, spontaneous form of LIP in polyclonal T cells was shown to depend largely on commensal flora (48, 49). However, the effect was not entirely dependent on a specific T cell response to microbial Ags; TCR Tg Rag^−/− OT-II CD4 T cells also underwent spontaneous LIP in the absence of OVA that was abrogated in germ-free hosts (50) and rescued by provision of cecal bacterial lysate-pulsed DCs, which depended on MyD88 and IL-6 expression in the DCs (50). Thus innate immune stimuli promote LIP of T cells, possibly through modulation of costimulation (discussed below). However, the division of LIP into two qualitatively different categories based on CFSE dilution is artificial as it depends on the length of transferred cell residency in the recipient (differing vastly between studies). In addition, there is no compelling evidence that cells that have undergone seven versus greater than eight divisions, for example, after an arbitrary period of time, have chosen opposite fates in a binary decision process leading to two qualitatively distinct types of LIP. Indeed, transfer of a small number of CD8 OT-I Rag^−/− T cells (1 × 10⁶) into Rag^−/− hosts gave rise to what would be considered both spontaneous and homeostatic proliferation via this scheme, whereas transfer of 1 × 10⁷ of the same cells gave rise to only the slow homeostatic version (51). The same phenomenon occurs in CD4 T cells upon transfer of low versus high numbers of hemagglutinin (HA) Rag^−/− (52) and DO11.10 Rag^−/− TCR Tg cells (53). Therefore, even within monoclonal T cells, some cells will undergo significantly more divisions than others and competition plays a large role in shaping this outcome. Thus, any qualitative dichotomy between spontaneous and homeostatic proliferation is likely resolved simply by recognizing that T cell interactions with self-pMHC and costimulatory ligands would generally be expected to be weaker in germ-free mice. In addition, TCR:foreign-pMHC interactions could more strongly promote LIP in the subset of T cells specific for microbial Ags. Although the fast spontaneous and slow homeostatic proliferation can have differential requirements for IL-7 (12, 54), the inability of anti–IL-7Rα blockade to inhibit spontaneous proliferation may result from strong TCR signals overcoming the need for IL-7R–mediated signals. TCR ligation blocks IL-7/IL-7R–mediated survival signals in an affinity-dependent manner in both CD4 and CD8 T cells (55), such that high-avidity TCR:pMHC interactions might block responsiveness to IL-7 entirely. Conversely, coinhibitory dampening of TCR signals leads to dependence on homeostatic cytokines (56).

Costimulation is important for productive T cell activation in lymphoreplete hosts. However, the role of cosignaling molecules during LIP has received far less study. Polyclonal CD28^−/− T cells underwent relatively normal LIP upon transfer into irradiated hosts, including those additionally lacking CD40 or 4-1BBL (57), suggesting that costimulation is not a requirement for LIP. However, mixed wild-type (WT) and CD28^−/− polyclonal naive T cells transferred to Rag^−/− recipients demonstrated that costimulation played a minimal role in LIP of CD8 T cells, but the CD4 compartment became dominated by WT cells (52). Importantly, treatment with CTLA-4-Ig or transfer into B7-deficient hosts could completely block this competitive advantage. Similarly, CTLA-4-Ig treatment has been demonstrated to inhibit LIP of DO11.10 (52, 53) and HA TCR Tg CD4 cells (52). CD24, a GPI-linked glycoprotein that costimulates T cells, is expressed on newly generated T cells (58) as well as post-TCR triggering and promotes both CD4 and CD8 LIP (59). Interestingly, CD24 has an additional inhibitory role in DCs via Siglec G (60) and its deficiency on non–T cells in lymphopenic hosts strongly promotes LIP and LIP-driven autoimmunity (61).

Major functional mechanisms of Tregs include regulating the access of conventional T cells (Tcons) to pMHC by potentially acting as a super competitor (62), or regulating the level of costimulation provided by APCs (63–67). Tregs can control LIP (51, 52, 67–73), although there are mixed results (25) likely owing to the variety of experimental approaches (particularly the use of polyclonal versus monoclonal Tregs and responder populations).

Using a polyclonal system, equal coinjection of CFSE-labeled CD45RB^high CD4 T cells (naive responders) and CD45RB^low CD25⁺ Tregs into Rag2^−/− mice resulted in decreased LIP (69) and a 10-fold decrease in T cell recovery. Later studies also demonstrated that polyclonal CD4⁺CD25⁺ cells could inhibit the accumulation of responding CD4⁺CD25⁻ T cells more efficiently than an equal number of CD4⁺CD25⁻ cells (52, 68), suggesting that polyclonal Tregs are more efficient competitors. It should be noted that CD4 Tregs also directly suppress LIP of CD8 T cells (51), in which case they are unlikely to be directly competing for pMHC class I. CD4 conventional T cells can also help LIP of CD8 T cells (51) and therefore the extent that Treg suppression affects LIP of individual populations can be challenging to decipher.

The ability of Tregs to suppress LIP in vivo has been reported to depend in part on the affinity of the responding cells’ TCR, as gauged by CD5 expression. Expansion of CD5^low cells was more efficiently suppressed (70), suggesting that T cells on the higher end of self-pMHC affinity might have an even greater advantage during LIP in the presence of Tregs. However, Vβ/CDR3 spectratyping has suggested that Tregs preserve the overall diversity of the repertoire in LIP (73).

Mechanistically, Tregs from IL-10^−/− mice were unable to reduce the accumulation of responding naive T cells, suggesting that IL-10 was critical for Treg control of LIP (69), although other studies have instead found that IL-10 production by Tregs was only important for the control of colitis (51). Knockout of the CTLA-4 molecule specifically in Tregs results in a scurfy-like autoimmune disease (74), and CTLA-4 was shown to be critical for the in vivo control of LIP (67, 72). Inhibition of LIP by Tregs can be due to CTLA-4–dependent downmodulation of B7-1/B7-2 expression on DCs, and the ratio of Treg:DC in lymph nodes critically determines the extent to which this occurred (67). This supports the concept that in LIP-driven autoimmunity, the ratio of Treg:DC may be more important for preventing disease than the ratio of Treg:Tcon.

Finally, cotransfer of either WT polyclonal CD4⁺CD25⁺ Tregs or central memory CD4 T cells at a 2:1 ratio with DO11.10 TCR Tg Rag2^−/− CD4 cells could equally suppress LIP of the responders (71). Under these conditions suppression of LIP appeared dependent on competition for IL-7, as IL-7R^−/− polyclonal central memory cells or Tregs were ineffective. Given that these DO11.10 cells and polyclonal Tregs likely had minimal overlap in their pMHC specificity, this suggests that the ability of Tregs to suppress LIP in a non-pMHC–specific way is primarily through competition for IL-7 and their ability to do so is approximately equal to central memory T cells. Moving forward, it will be important to fully clarify the role of Treg TCR specificity in controlling LIP of responding T cells.

Several receptors with coinhibitory function have also been demonstrated to control LIP. TGF-β, an immunoregulatory cytokine important for the generation of Tregs, was recently shown to act directly on Tcons to restrain LIP (29, 75). With equally mixed polyclonal WT and TGF-βRII–deficient cells in Rag1^−/− recipients, TGF-βRII^−/− T cells predominated by a factor of five by 7 d posttransfer. Using the spontaneous versus homeostatic LIP paradigm discussed previously, only a subset of the WT control cells underwent spontaneous LIP after transfer, compared to all of the TGF-βRII^−/− cells. Moreover, antibiotic treatment of the recipients to deplete flora could block spontaneous LIP of the WT cells to a large extent, but not that of the TGF-βRII^−/− cells (29). TGF-βRII–mediated signals also restrained LIP of OT-I TCR Tg T cells. It remains unclear whether the rapid LIP of these TGF-βRII^−/− cells in vivo was due to an exaggerated response to self-pMHC or other resources present during lymphopenia.

Similar results have been found in studies of T cells deficient in expression of the coinhibitor BTLA (76). Whether BTLA suppresses LIP via control of pMHC-generated signals or other resources was not determined. Additional coinhibitory molecules that have been shown to inhibit LIP include LAG-3 (77) and the transmembrane adapter protein SIT (78). The coinhibitor PD-1 is recruited to TCR microclusters during activation and recruits the phosphatase SHP-2 to modulate TCR signaling (79–81). We found that polyclonal T cells deficient in PD-1 have a proliferative or survival advantage during LIP (82), and have recently extended this observation to a TCR Tg system where LIP could only occur in response to nonagonist pMHC (K.K. Ellestad, J. Lin, L. Boon, and C.C. Anderson, manuscript in preparation).

Other molecules not considered as coinhibitors but with the ability to attenuate TCR signaling have also been shown to modulate LIP in a TCR-dependent manner. PTPN2, a protein tyrosine phosphatase, is upregulated in thymocytes and newly generated naive T cells as well as in CD8 T cells undergoing rapid LIP, potentially as a mechanism of T cell tuning. T cells with conditional deletion of PTPN2 underwent more rapid LIP (30). Similarly, loss of the protein tyrosine phosphatase PTPN22 also leads to enhanced T cell LIP (83) and cytokine production in response to weak pMHC stimulation.

To date, deficiencies in PD-1 (27), TGF-βRII (29), and PTPN2 (30) have been linked to the development of autoimmunity in conjunction with T cell LIP. Here we will focus on findings from the PD-1^−/− model and potential differences compared with TGF-βRII or PTPN2 deficiency that may shed light on critical requirements for LIP-driven autoimmunity.

The C57BL/6 PD-1^−/− mouse spontaneously develops a late-onset, mild, lupus-like disease (84). In stark contrast to this mild phenotype, reconstitution of the lymphoid compartment of Rag^−/− mice by transfer of PD-1^−/− HSC gave rise to a rapid, severe, systemic autoimmune disease shortly after the first newly generated T cells emerged from the thymus (27), whereas WT HSC recipients were unaffected. Treatment of recipients with broad-spectrum antibiotics was unable to prevent disease, suggesting that gut flora did not drive this autoimmunity (82). Disease was also not attributable to defective Treg generation/conversion due to PD-1 deficiency (82). Furthermore, Rag^−/− recipients of PD-1^−/− HSC or thymocytes, but not splenocytes, developed autoimmunity, suggesting that recent thymic emigrants (RTEs) were critical for generation of disease (27). This also indicated that PD-1’s role is in the establishment rather than the maintenance of self-tolerance.

The inability of established PD-1^−/− peripheral T cells to cause disease in lymphopenia suggests that rather than being a two-hit model, this LIP-driven autoimmune disease relies additionally on a third factor, newly generated T cells (Fig. 1). One might consider that this population has not yet gone through peripheral tolerance mechanisms in a lymphoreplete setting, and thus might contain a higher proportion of self-reactive T cells or have a general shift to higher avidity to self as compared with established peripheral T cells. In addition, newly generated T cells would lack a peripherally generated Treg (pTreg) population and thus upon transfer to a lymphopenic host, the absence of pTregs may allow expansion of Tcons (due, for example, to an insufficient ratio of pTreg:APC presenting pMHC). Indeed, early transfer of purified Tregs to adult Rag1^−/− PD-1^−/− HSC recipients can block disease (82). If the third hit, newly generated T cells contribute to LIP-driven autoimmunity because this population lacks pTregs, it could explain why TGF-βRII^−/− T cells did not need to be a newly generated population to cause autoimmunity (29); pTreg numbers are controlled by TGF-β (85) such that the established T cell pool in TGF-βRII^−/− mice is likely to be pTreg-deficient similar to newly generated T cells of WT mice (see Fig. 1). In addition, the ability of a population composed mostly of established T cells from TGF-βRII^−/− mice to cause LIP-driven autoimmunity may require some RTEs. The autoimmune potency of established versus newly generated T cells has yet to be examined in the setting of TGF-βRII deficiency.

The potency of newly generated T cells in driving autoimmunity during LIP initially seems to counter the well-established immaturity and limited effector function of RTEs (58). RTEs are predisposed to tolerance within a lymphoreplete environment (86). However, RTEs have heightened responses to IL-7 (87–89) and low-level (90) or low-affinity pMHC signals (91) that become available in lymphopenia and they express CD24 that promotes LIP (59). Thus, the competitive disadvantage of RTEs is reversed in lymphopenia (92).

Consistent with a critical role for lymph node–dependent LIP in driving disease following PD-1^−/− HSC transfer, lymph node–deficient Rag^−/−γc^−/− and lethally irradiated LTα^−/− mice were found not to be susceptible (27). Furthermore, although all adult Rag^−/− recipients of PD-1^−/− HSC developed disease, almost all neonatal Rag^−/− recipients remained healthy. This protection of neonates correlated with a drastic reduction in effector memory (CD44^hiCD62L^lo) phenotype acquisition (31%) compared with adult recipients (90%). Compared with the adult, the available lymphoid stroma and space in the neonate would be smaller, thus resources may be exhausted more rapidly, preventing autoimmunity. The same explanation for lack of disease may apply to the PD-1^−/− mouse in which the first newly generated T cells seed the periphery during the neonatal period. Beyond anatomic size considerations, it was recently suggested that a subset of innate lymphoid cells (ILC3) expressing CD4 and the IL-7R are present in neonates and can act as inhibitors of CD8 T cell LIP via an unknown mechanism dependent on IL-7R signaling (93). These innate lymphoid cells may act as sinks for IL-7, reducing LIP potential in the neonatal period and the ability of high IL-7 concentrations to promote hyperstimulation of T cells (94). At first glance, the role of the neonatal period in controlling LIP of PD-1–deficient T cells appears counter to the findings with TGF-βRII–deficient T cells where neonatal lymphopenia supports LIP-mediated autoimmunity (29, 95, 96). However, if one examines lymphopenia quantitatively these findings are not incongruous. In terms of quantity of lymphopenia in the recipients, the order would be adult Rag^−/− > neonatal Rag^−/− > neonatal WT > adult WT. Although the mild lymphopenia of the neonatal period is not sufficient to drive PD-1–deficient T cells to initiate more than a mild and slowly developing lupus-like autoimmunity, a more pronounced enhancement of resource-mediated signaling in TGF-βRII–deficient T cells could allow even the less severe lymphopenia in neonates to support LIP-mediated autoimmunity.

The traditional criteria for lymphopenia is an abnormally low number of lymphocytes (Criteria 1). One might also consider a host lymphopenic if it supports LIP of T cells transferred to it (Criteria 2). These criteria, however, are not always simultaneously true and do not adequately encompass all of the factors that influence the extent to which T cells will undergo LIP or acquire spurious effector function and autoimmunity.

Consider the following: compared with a normal adult mouse, which contains ∼46 × 10⁶ peripheral naive T cells, the DO11.10 TCR Tg mouse contains ∼25 × 10⁶ naive T cells (97). Therefore by Criterion 1 above, the DO11.10 TCR Tg is lymphopenic. However, transfer of DO11.10 T cells to a DO11.10 TCR Tg host yields no LIP (97), therefore by Criterion 2 the DO11.10 TCR Tg mice are not lymphopenic. The input DO11.10 T cells compete for the same self-pMHC resources as the endogenous DO11.10 T cells which have already achieved homeostatic balance with the available resources, thus HA Rag2^−/− TCR Tg cells transferred to a DO11.10 TCR Tg host undergo efficient LIP (97) as they compete for a discrete set of resources (different pMHC). Similarly, the description of various hosts as lymphopenic does not convey differences in the degree of lymphopenia in that host; LIP will depend in part on the number of pre-existing T cells in the various lymphopenic hosts. As a final example, consider two adult WT mice, one of which has just received an injection of IL-7. Provision of more of the IL-7 resource will effectively enlarge the niche allowing some T cells to undergo expansion (18–21). Was the mouse that received IL-7 then made lymphopenic?

In sum, we suggest that a more informative way to view the current homeostatic state of the T cell compartment in a host is in terms of a quantitative LIP potential, which takes into account the totality of the available resources for a given set of T cells in the host versus the number of competitors for those resources. This concept is useful because it provides a clear and inclusive quantitative framework to make predictions for how the overall system of T cell and resource interactions will respond to perturbations. For example, such a framework could inform strategies to ameliorate autoimmune diseases that target the size of the resource or competitor pool specifically for disease-causing T cells in order to have their effect, avoiding the use of broad immunosuppression. The concept of LIP potential may also be applied to smaller compartments within a host, for example the environment within a tumor or a tissue. Although blockade of PD-1 may be effective in tumors due to enhanced collateral damage (98), it may also be especially effective in the tumor microenvironment because blockade of PD-1 increases intratumoral LIP potential, facilitating expansion and/or activation of pre-existing tumor-infiltrating T cells.

In order to derive an equation describing the proportional relationship between resources, competition, and LIP potential, the following aspects must be considered: 1) resources are finite, and consist of homeostatic cytokines and, for a given T cell, the set of pMHC present in the host with which its TCR can interact (tonic self-pMHC or agonist pMHC if present). The size of the host or compartment in question influences the total available resources; 2) T cells compete for access to homeostatic cytokines and, both intra- and interclonally, for pMHC; 3) Tregs can inhibit LIP and may be viewed as professional competitors for pMHC and/or other factors; 4) TCR avidity influences the ability to compete for pMHC and may secondarily influence ability to compete for homeostatic cytokines (by modulation of cytokine receptor expression etc.); 5) costimulation and coinhibition can influence LIP positively and negatively, respectively. Tregs can modulate costimulation; 6) naive T cells require tonic TCR:pMHC stimulation and homeostatic cytokine signals (chiefly IL-7) for survival. Central memory T cells rely on homeostatic cytokines but not TCR:pMHC stimulation for survival.

In general, for a given T cell clone (A):

Variable definitions:

• [HC $(a)$]: concentration of homeostatic cytokines useable by clone $A$

• [pMHC $(a)$]: the available concentration of the set of pMHC complexes that can mediate tonic or greater interactions with the TCR of clone $A$

• affinity $(a)$⁠: the affinity of the interaction between the TCR of $A$ and pMHC $(a)$

• [Tcon $(a)$]: the concentration of competitor Tcons that can recognize pMHC $(a)$

• [Treg $(a)$]: the concentration of Tregs that can recognize pMHC $(a)$

• $costimcoinhib(a)$⁠: the basal level of costimulation divided by coinhibition provided by APCs presenting pMHC $(a)$

• size $(a)$⁠: represents the size of the SLOs of the host or the area (e.g., a single lymph node, intratumoral space) accessible to clone $A$ for which LIP potential is being measured.

Then for naive T cells (which require pMHC for survival/LIP):

For central memory T cells (which use homeostatic cytokines for survival but do not require pMHC, although pMHC interactions may influence the efficiency of their LIP):

These equations predict methods to control LIP potential. It is also important to realize that the set of pMHC recognizable by clone $A$ could overlap with the set of pMHC recognizable by clone $B$ (i.e., $pMHC(b)$⁠), clone $C$⁠, and so on, thereby illustrating how interclonal competition could influence LIP potential for all three T cell clones.

By what mechanisms does a global (across all T cell clones) increase in T cell LIP potential lead to promotion of autoimmunity? We found that neither the Fas nor perforin-mediated effector pathways are required for autoimmunity post–PD-1^−/− HSC or thymocyte transfer (K.K. Ellestad, G. Thangavelu, L. Boon, and C.C. Anderson manuscript in preparation). However, the abundance of numerous inflammatory cytokines in the serum (27) might suggest that the disease results primarily from a cytokine storm. T cells within the polyclonal repertoire with TCRs naturally on the higher end of the self-pMHC affinity spectrum may, in the context of LIP and PD-1 deficiency (which globally affects the strength of the TCR signal), be pushed to spuriously secrete proinflammatory cytokines. A sufficient number of such activated cells could raise serum cytokine concentrations to a threshold level for causing the overt features of autoimmunity in this model and potentially others (29, 30, 75).

A complex network of intra- and interclonal competition between T cells for available self-pMHC helps to manage the diversity of the T cell repertoire (1). Indeed, contraction of clonally expanded Ag-specific CD4 T cells following an immune response is regulated by their interclonal competition with neighboring T cells that share self- but not foreign-pMHC specificity (99). During LIP, T cells with higher affinity for pMHC have an advantage as they can compete more strongly for pMHC resources and proliferate faster, and in WT mice some evidence exists that LIP leads to altered T cell repertoires (100). In the absence of a coinhibitor, an upward shift in the average TCR signal may exacerbate the imbalance between lower- and higher-affinity T cells that recognize the same self-pMHC, and this may be exponentially magnified in a host with high LIP potential. One could envision a situation where critical low-affinity neighboring T cells may be outcompeted during LIP to the extent that they are no longer numerous enough to adequately act as relevant competitors of clones with higher self-pMHC affinity, paving the way for autoimmunity. Whether coinhibitor deficiency exacerbates disparities in T cell diversity normally associated with LIP is unclear. However, the idea of neighboring T cells acting as modulators of LIP potential for T cells with overlapping pMHC specificity itself has applications in terms of treating autoimmunity or cancer.

The microenvironment within a solid tumor might be viewed as having low LIP potential due to the local presence of Tregs, limited relevant neoantigen-derived pMHC (neo-self-pMHC), and/or paucity of homeostatic cytokines. The concept that blocking coinhibition would generate immunotherapy for tumors was proposed 45 y ago (101) and in recent years, anti–PD-1 or anti–PD-L1 blocking Abs have proven highly effective (102). By blocking PD-1 signals, every neo-self-pMHC is made more valuable as a resource because it can effectively supply a larger TCR signal. Thus with anti–PD-1 treatment, the available neo-self-pMHC within a tumor may support the survival of increased numbers of neoantigen-specific T cells and/or promote their activation due to the stronger TCR signals. Indeed TCR signals from both tumor neoantigens and self-pMHC would be enhanced and the latter may contribute to the response to agonist (103, 104). Specifically increasing LIP potential further within this microenvironment might be particularly beneficial in the situation of decreasing Ag load due to successful tumor clearance, where sufficient neo-self-pMHC might not persist to sustain high T cell numbers. For example, boosting homeostatic cytokines in the intratumoral environment might be achievable by using bispecific therapeutic approaches such as IL-7 or IL-15 tethered to a tumor Ag-directed mAb. Such an approach may help to maintain sufficient survival signals to sustain an expanded population and effectively keep the pressure on the tumor.

In summary, TCR affinity for pMHC, the level of available pMHC and homeostatic cytokines, costimulation, coinhibition, Tregs, and competition from Tcon all regulate LIP to varying extents. Importantly, not all autoimmunity-predisposing cofactors have an equal magnitude effect on LIP/T cell activation, thus autoimmunity during LIP is not always a two-hit phenomenon (e.g., the PD-1^−/− autoimmune model described herein requires LIP + newly generated T cells + PD-1 deficiency). Traditional definitions of lymphopenia lack the ability to describe the potential for a given clone of T cells to undergo LIP, or LIP potential. LIP potential is governed by the resource availability experienced by a particular T cell clone in a host and this concept can be expanded to subcompartments within a host (e.g., a tumor). Taken together, we hypothesize that the risk of LIP-driven autoimmunity upon reconstitution of a polyclonal T cell repertoire is proportional to the global LIP potential of the host.

The authors have no financial conflicts of interest.