Homeostasis and immunological function of self-driven memory-phenotype CD4+ T lymphocytes

Abstract CD4+ T lymphocytes play an essential role in adaptive immune responses. In pathogen infection, naïve CD4+ T cells that strongly respond to foreign antigens robustly proliferate to differentiate into effector/memory cells, contributing to elimination of the pathogen concerned. In addition to this conventional T cell activation pathway, naïve T cells can also weakly respond to self antigens in the periphery to spontaneously acquire a memory phenotype through homeostatic proliferation in steady state. Such ‘memory-phenotype’ (MP) CD4+ T lymphocytes are distinguishable from foreign antigen-specific memory cells in terms of marker expression. Once generated, MP cells are maintained by rapid proliferation while differentiating into the T-bet+ ‘MP1’ subset, with the latter response promoted by IL-12 homeostatically produced by type 1 dendritic cells. Importantly, MP1 cells possess innate immune function; they can produce IFN-γ in response to IL-12 and IL-18 to contribute to host defense against pathogens. Similarly, the presence of RORγt+ ‘MP17’ and Gata3hi ‘MP2’ cells as well as their potential immune functions have been proposed. In this review, I will discuss our current understanding on the unique mechanisms of generation, maintenance, and differentiation of MP CD4+ T lymphocytes as well as their functional significance in various disease conditions.


Introduction
T lymphocytes are essential for adaptive immune responses. T cells are originally generated in the thymus where thymocytes are tested for their self reactivity through positive and negative selection [1]. Specifically, thymocytes expressing T cell receptor (TCR) that has high affinity to self antigen -MHC II complexes are deleted while those that fail to react to self undergo apoptosis by 'death by neglect.' As a consequence, only thymocytes that weakly react to self -MHC II molecules can survive and migrate into the periphery as CD4 þ or CD8 þ naïve T lymphocytes. In addition, thymocytes with relatively high TCR affinity to self antigens (but below the threshold for negative selection) can survive as regulatory T cells (Tregs).
In the periphery, T lymphocytes play a critical role in adaptive immunity. In pathogen infection, naïve CD4 þ T lymphocytes expressing TCRs that strongly recognize foreign antigens robustly proliferate to give rise to effector cells, thereby contributing to elimination of the pathogens concerned. After the pathogens are cleared most effector cells undergo apoptosis to avoid their detrimental effects on the host, leaving a small population of foreign antigen-specific memory cells that can more quickly and more strongly respond to secondary infection with the same pathogen. These T cell compartments are strictly regulated by a homeostatic mechanism in steady state [2].
In addition to the above 'conventional' activation pathway of T lymphocytes, accumulating evidence suggests that some naïve CD4 þ T cells can recognize self antigens weakly in the periphery and spontaneously acquire a memory phenotype at homeostasis [2,3]. Importantly, such 'memoryphenotype (MP) cells' have been found to possess innate effector function and can contribute to hose defense against pathogen infection [4]. In this review article, I will discuss our current understanding on the unique mechanisms of their development, maintenance, and differentiation in steady state as well as functional significance in disease conditions.
2. Identification of MP CD4 þ T lymphocytes as a population distinct from foreign antigen-specific memory cells Conventional CD4 þ T lymphocytes consist of naïve (CD44 lo CD62L hi ) and memory (CD44 hi CD62L lo ) compartments in steady state [2]. In secondary lymphoid tissues of unimmunized mice where overt immune responses are absent, the latter subpopulation occupies $10% of total CD4 þ T cells and increases in size with age. In the past, this cell population was tacitly assumed to all represent 'authentic' memory T lymphocytes that are specific for some foreign antigens derived of commensal microbiota and/or food. However, this simple assumption has been now called into question because of their unique mechanisms of generation, maintenance, differentiation, and effector function as well as distinctive phenotypic properties [3].
The first reason for the above notion is regarding the unique developmental pathway of CD44 hi CD62L lo Foxp3 À CD4 þ T lymphocytes in steady state. Thus, they are equally present in specific pathogen-free (SPF), germ-free (GF), and antigenfree (AF) mice [5,6]. Because AF animals lack both commensal and food antigens, it is likely that CD44 hi CD4 þ T lymphocytes in these animals develop in a manner dependent on self rather than foreign antigen recognition. Similarly, this cell population is present in CD4 þ T cells of foreign antigenspecific TCR-transgenic mice housed in SPF environment that lack their cognate foreign antigens [7]. Indeed, P25, OT-II, and Marilyn CD4 þ T lymphocytes that have TCRs specific for defined foreign antigens have been shown to have a small CD44 hi subset. Interestingly, the same study showed that strength of TCR affinity to self antigens that is reflected by CD5 expression levels on CD4 þ T cells [8] is positively correlated with the population size of CD44 hi cells, further supporting the notion that CD44 hi CD4 þ T lymphocytes arise dependently of self antigen recognition.
The second reason is concerning the mechanisms that govern maintenance of CD44 hi CD4 þ T cells in steady state. In the case of foreign antigen-specific memory cells, once generated from effector T lymphocytes they become largely independent of antigen recognition for their maintenance and stop cell division [9]. On the other hand, $30% of CD44 hi CD4 þ T lymphocytes are in cell cycle at any given time point at homeostasis [10], and more than 60% of the same cells have proliferated at least once over a four-week period [11]. This strongly suggests that rapidly cycling CD44 hi CD4 þ T lymphocytes are qualitatively distinct from typically quiescent memory cells.
Third, CD44 hi CD4 þ T cells have unique differentiation pathways at homeostasis. In conventional helper T lymphocytes, upon activation they differentiate into effector subsets including T-bet þ Th1, Gata3 hi Th2, and RORct þ Th17 [12]. This fate decision is mainly if not exclusively made by environmental cytokines. Specifically, naïve cells respond to IL-12 to differentiate into the Th1 subset while Th2 cells develop in the influence of IL-4. Th17 cells are directed by IL-6 and TGFb. Among these Th-differentiating cytokines, IL-12 is primarily produced by dendritic cells (DCs) that have been primed by microbial-derived agonists including LPS [13,14]. Similarly, CD44 hi CD4 þ T lymphocytes contain a T-bet þ Th1-like subset, differentiation of which is significantly promoted by IL-12 derived from type 1 DCs (DC1s) [7]. However, this IL-12 production by DC1s is homeostatically driven by self rather than foreign-derived agonists, and indeed can be equally detected in SPF, GF, and AF animals. These findings argue that T-bet þ CD44 hi CD4 þ T cell differentiation is self-sufficient, which is in sharp contrast to conventional Th1 differentiation where foreign-derived agonists play a dominant role.
Finally, my group has recently reported that steady-state CD44 hi CD4 þ T lymphocytes are phenotypically distinct from foreign antigen-specific memory cells [6]. Thus, the former cell population exhibits a Bcl2 lo CD127 lo$hi Sca1 lo$hi phenotype while the latter cells are Bcl2 hi CD127 hi Sca1 hi . Similarly in CD8 þ T lymphocytes, CD44 hi cells are known to be distinguishable from foreign antigenspecific memory using cell surface markers a4, a1, b1, and NKG2D [15,16], further supporting the notion that the two lymphocyte populations are different from each other in both CD4 þ and CD8 þ T cells.
Based on the above considerations, CD44 hi CD62L lo Foxp3 À CD4 þ T lymphocytes present at homeostasis are now thought to belong to a population that is distinct from foreign antigen-specific memory cells, and my group refers to the former cell population as 'memory-phenotype' (MP) cells. As described below, MP cells are a unique CD4 þ T lymphocyte population that develops from naïve precursors in a self antigen recognition-dependent manner in steady state and can exert innate effector function in host defense.

Homeostatic proliferation as a mechanism generating MP CD4 þ T lymphocytes
Because CD4 þ T lymphocytes newly generated from the thymus are essentially all naïve, MP cells are thought to arise from naïve precursors in the periphery. The mechanism of MP cell generation has been traditionally studied by transferring T lymphocytes into sublethally irradiated or Rag1/2-deficient mice. In such lymphopenic animals some donor cells spontaneously proliferate via 'homeostatic proliferation' to recover lymphosufficiency and acquire a memory phenotype [17,18]. Although in initial studies donor T cells contained both naïve and MP compartments, subsequent studies confirmed that sorted naïve T lymphocytes can generate the same proliferative response [19,20]. Thus, lymphopeniainduced homeostatic proliferation of naïve T lymphocytes is thought to reflect the mechanism of MP cell generation.
In lymphopenic environment, transferred naïve CD4 þ T lymphocytes exhibit two distinct types of proliferation: 'slow' and 'fast' [19,20]. The former is defined as a cell division every $3 days while the latter defined as a division more than once a day. Slow proliferation of naïve CD4 þ T lymphocytes contributes to turnover of their own although this might not be the case in CD8 þ T cells. Thus, donor CD4 þ cells largely maintain their CD44 lo CD62L hi phenotype, regenerating naïve cells. This proliferation requires TCR signaling presumably derived from self antigen recognition but not costimulatory molecules [17,19,[21][22][23][24][25]. In addition, the same response is known to be dependent on IL-7 provided by fibroblastic reticular cells in lymph nodes [2,[20][21][22]26].
On the other hand, a few clones of donor naïve CD4 þ T lymphocytes rapidly proliferate to acquire a CD44 hi CD62L lo 'MP-like' phenotype via fast homeostatic proliferation. Because the resultant cells are essentially CD25 À CD69 À , this phenotype is thought to reflect the one of memory rather than effector T lymphocytes. This form of proliferation is dependent on antigen recognition and costimulation including CD28 and OX40 signaling provided by DCs [17,21,24,[27][28][29][30], indicating that fast and slow cell divisions are differentially regulated. In addition, the same proliferative response is inhibited by TGFb as well as Tregs [23,[31][32][33][34]. Which antigens, foreign or self, drive the fast homeostatic proliferation of naïve CD4 þ T lymphocytes is an interesting question. A clue to this issue came from an early study where naïve CD4 þ T lymphocytes were transferred into lymphodeficient mice housed in SPF versus GF environments [19,22]. The fast proliferating population was significantly diminished in the latter hosts; however, the remaining cell population was reproductively detected. It is unlikely that the residual subpopulation is driven by food antigens because this proliferation is equally detected in AF conditions [23]. These findings argue that the fast homeostatic proliferation can be driven by both foreign and self antigens. In consistent, the fast population can be divided into commensal antigen-dependent a4b7 þ and -independent a4b7 À subpopulations [30,35], suggesting that the same integrin can be used as a marker to distinguish the two types of cells with different antigen specificity.
Does the fast homeostatic proliferation occur in physiological settings? It is well known that neonatal environment is lymphopenic in mice and in such conditions naïve CD4 þ T cells rapidly proliferate [36]. More recently, my group has reported that fast proliferation of naïve CD4 þ T cells can be driven in adult, lymphosufficient settings as well, although at a slower rate [37]. Thus, naïve CD4 þ T cells spontaneously proliferate in unchallenged, congenic animals and acquire a CD44 hi CD62L lo memory phenotype. Importantly, this proliferation is largely driven by self antigen recognition; CD5 hi naïve CD4 þ T cells that have higher TCR affinity to self Ags more rapidly proliferate than do their CD5 lo counterparts. Similarly, it has been shown that naïve CD8 þ T lymphocytes spontaneously proliferate to acquire a memory phenotype in a self antigen recognition-dependent manner [38]. Based on these considerations, I propose that homeostatic proliferation of naïve T lymphocytes is a physiologic process that generates MP cells in both neonatal and adult environments.

Heterogeneity of MP CD4 þ T lymphocytes
Once generated, MP cells tonically proliferate for their maintenance at homeostasis [10,37]. This is in contrast to foreign antigen-specific memory T lymphocytes that become largely independent of TCR signaling after generation and acquire a quiescent state [10]. Mechanisms of MP cell maintenance have been studied by again examining lymphopeniainduced homeostatic proliferation. Thus, when purified MP CD4 þ T lymphocytes are transferred into lymphopenic mice, donor MP cells exhibit slow and rapid proliferation in the recipient animals. Slow proliferation depends upon IL-7 whereas fast cell division requires TCR and OX40 signaling [21,28,39,40]. Hence, MP cell homeostasis is maintained by two different types of proliferative responses in lymphopenic environment.
Does slow and fast homeostatic proliferation of MP CD4 þ T cells occur in steady-state settings? We have recently reported that MP cells are heterogeneous in terms of CD127 and Sca1 expression (Figure 1(A)) and that $50% of CD127 lo Sca1 lo$hi MP cells are in cell cycle while those with a CD127 hi Sca1 hi phenotype are more quiescent [6]. This finding suggests that fast and slow homeostatic proliferation of MP cells occurs in lymphosufficient environment as well, as the forms of CD127 lo Sca1 lo$hi and CD127 hi Sca1 hi fractions, respectively. Moreover, we found that CD127 lo subpopulation represents newly generated MP cells and this fraction acquires a CD127 hi Sca1 hi phenotype with time, arguing that rapidly dividing cells represent a newly generated MP population while those with slow cell division are more differentiated and proliferationexperienced cells (Figure 1(B)).
In steady state, fast proliferation of MP lymphocytes reflected by Ki67 expression is regulated by costimulatory as well as coinhibitory signaling. Thus, the same expression is enhanced by CD80/86, OX40L, and GITRL while inhibited by PD-L1 [37,41]. Because these molecules are expressed on antigen-presenting cells including DCs and because DCs are essential for the induction of lymphopeniainduced fast homeostatic proliferation of MP CD4 þ T cells [28], it is likely that MP -DC interactions play an inevitable role in the tonic proliferation of MP cells and their later acquisition of a more quiescent state (Figure 1(B)).

Homeostatic differentiation of MP cells and its immunological function
Besides proliferative heterogeneity as described above, MP CD4 þ T lymphocytes are heterogeneous in terms of differentiation status as well (Figure 2). Thus, we previously reported that MP cells comprise a major T-bet þ 'MP1' and a minor RORct þ 'MP17' as well as T-bet À RORct À undifferentiated subpopulations [7]. Although the presence of Gata3 hi 'MP2' cells has not been observed in lymphoid organs, they might exist in some extra-lymphoid organs or in certain circumstances where MP1 or MP17 differentiation is hampered.
The MP1 subset develops in an IL-12p70-dependent manner in steady state [7]. This IL-12 is homeostatically produced by splenic DC1s that have been basally primed by some self-derived agonists via multiple TLRs -MyD88 pathways. Thus, AF mice that lack foreign-derived agonists from commensal microbiota and food have unaltered levels of IL-12 whereas TLR2, TLR7, and TLR9 knockout mice have significantly reduced levels of the same cytokine. Because high mobility group box 1 (HMGB1), b-defensin, surfactant protein A and D, and serum amyloid A can bind to TLR2 whereas single-strand RNA and IgG-chromatin complexes are known to activate TLR7 and TLR9, respectively [42][43][44][45][46][47][48][49][50], these self-derived molecules are possible candidates that tonically drive TLRs -MyD88 signaling in DC1s. Once primed, their homeostatic production of IL-12 is further stabilized by CD40L on CD4 þ T lymphocytes. In particular, MP as compared to naïve cells have upregulated levels of CD40L on their surface, which appears to tonically stimulate CD40 on DC1s. Hence, MP -DC1 interactions play an essential role in differentiation of MP1 in steady state. Importantly, DC1s are dispensable for MP cell development itself, arguing that DCs that generate MP cells and those that direct their type 1 differentiation are different. I propose this unique machinery of MP cell differentiation as 'homeostatic differentiation.' Differentiation pathway of the MP17 subset is still unclear. RORct þ CD44 hi CD62L lo Foxp3 À CD4 þ T lymphocytes are known to be present in secondary lymphoid tissues of SPF, GF, and AF animals [5,7], suggesting that foreign antigens or agonists are dispensable for their development. If acquisition of RORct by MP cells is self-sufficient, MP17 differentiation might be directed by IL-6 and/ or TGFb that are homeostatically produced by some other cells than DC1s in response to self-derived agonists. This hypothetical pathway needs to be validated by future studies.
Surprisingly, my group has demonstrated that MP cells and especially their MP1 subset can exert innate-like effector function in pathogen infection. Specifically, MP1 cells can produce IFN-c in response to IL-12, IL-18, and IL-2 in the absence of antigen recognition in vitro [6], which mediates host protection from Toxoplasma gondii and Mycobacterium tuberculosis infection in vivo [7,37]. This observation is in consistent with CD8 þ MP cells (referred to as virtual memory cells) that can exert innate effector function in an IL-12-, IL-18-, and IL-15-dependent manner [15,38]. Here it is noteworthy that CD4 þ MP cells are distinguishable from NKT cells using CD1d-tetramer [37]. My group and others thus proposed that together with NK, NKT, and innate lymphoid cells, CD4 þ and CD8 þ MP cells contribute to lymphocyte-mediated innate immunity that plays an important role in host defense until foreign antigen-driven adaptive immune responses have fully developed [4,51,52]. The above observation on CD4 þ MP cells suggests that MP1/2/17 CD4 þ T lymphocytes can be bystander activated by STAT activators and NF-jBactivating IL-1 family members as we previously proposed [37]. In addition to MP1 cells that can be activated by IL-12 (STAT4 activator) and IL-18 (IL-1 family member), a recent study showed that steady-state CD44 hi CD62L lo CD25 À CD4 þ T cells can produce IL-17 when stimulated with IL-23 (STAT3 activator) and IL-1 in the presence of IL-7 [53]. It remains to be determined whether this innate-like MP17 subset can contribute to host defense against pathogen infection. Similarly, it needs to be tested whether presumptive MP2 cells can be activated by TSLP (STAT5 activator), IL-33 (IL-1 family member), and/or IL-25 and exert innate effector function in vivo.

MP cells in humans
I have so far reviewed our current understanding on homeostatic proliferation and differentiation of MP cells as well as their host-protective function in mice. These observations pose the important question of whether the same T lymphocyte population exists in humans. In this regard, previous reports identified a rapidly proliferating CD4 þ T cell subpopulation with a memory phenotype in human cord blood and fetal spleen [54,55]. Furthermore, a recent paper showed that memory-like CD4 þ T lymphocytes are present in fetal intestine and can produce IFN-c in response to CD3/28 stimulation ex vivo [56]. Because foreign antigens are thought to be minimal in human cord blood and fetus, these findings support the notion that self-driven MP cells and especially their MP1 subset may exist in humans. Further investigation including phenotypical as well as functional analyses will be necessary to validate this hypothesis.
If the above functional MP cells are existent in humans, this cell population may be a potential therapeutic target in treatment of infectious diseases [57]. For example, my group previously showed that treatment with exogenous IL-12 enhances IFN-c production by MP cells, thereby contributing to host defense against T. gondii infection in mice [37]. This finding suggests that treatment with the same cytokine might be useful in activating MP1 cells in various infectious conditions. This 'immune-activating therapy' might be applied to presumptive MP2/ 17 cells as well.
On the other hand, because MP cells have selfreactivity and homeostatically express inflammationassociated transcription factors including T-bet and RORct, one can envision that the same cells can be the fundamental cause of autoimmune and/or inflammatory diseases. Indeed in mice, we have reported that MP cells can induce colitis in lymphopenic environment [6]. In addition, a recent study showed that MP CD4 þ T lymphocytes can produce IL-17A in response to IL-23 and IL-1 in the absence of antigen recognition in vitro, and suggested that such bystander activation of T lymphocytes contributes to exacerbation of experimental autoimmune encephalomyelitis driven by myelin-specific TCRtransgenic CD4 þ T cells [53]. Under healthy conditions, the inflammatogenic nature of MP cells may be inhibited by Tregs or other molecular signals; however, when such inhibitory mechanisms are disrupted, the former cells may exhibit excess proliferation and cytokine production to induce immunopathology. If this is the case in autoimmune and/or inflammatory diseases in humans, blockade of proliferation or activation of MP CD4 þ T lymphocytes may be beneficial as a novel approach for treatment of these diseases.

Acknowledgement
The author gratefully acknowledges A. Sher, W. E. Paul, and N. Ishii for their critical supports.

Disclosure statement
No potential conflict of interest was reported by the author(s).