The concept of space and competition in immune regulation

Regulation of cell numbers and organ size in multicellular organisms is an important principle in biology. Experimental data in developmental biology indicate that there are mechanisms by which organs sense their total mass, linked to the regulation of cell size and proliferation, but not solely determined by either.^1,2 Active monitoring of organ size is suggested by regeneration experiments following removal of part of an organ; this has been best illustrated by the regeneration of mammalian liver after partial hepatectomy.³ Another example is the demonstration that, following transplantation of multiple spleen fragments, the total spleen graft mass tends to reach a plateau which is in the range of variation of normal spleen weights.⁴ There are a multitude of mechanisms that regulate the number of cells (reviewed in ref. 1). For instance, the number of cell divisions may be predetermined, as in the nematode Caenorhabditis elegans,⁵ or divisions may stop after a given time interval (a mechanism for cell number control used in cardiac myocytes). Furthermore, hormones and components of signal-transduction pathways regulate growth and cell division, e.g. growth hormone⁶ and components of the insulin-signalling pathway.⁷ Another widely used principle in biological systems is competition for limiting resources, such as secretory molecules or cell-contact mechanisms. This is evident, for instance, in the control of oligodendrocyte precursors, which are regulated via the concentration of platelet-derived growth factor, whereas the number of mature oligodendrocytes is related to axon-dependent survival signals that are limited in amount and assure that the final number of oligodendrocytes is matched to the number of axons.^8–10

In the mammalian immune system, size control checks maintain the number of peripheral T cells at more or less constant levels. This is not the result of a finite capacity for cell division, as it has been shown that T cells, following serial transfer into irradiated hosts, are able to divide up to 56 times in vivo, an expansion potential similar to that of colony-forming units.¹¹ Thus, alternative mechanisms are involved in controlling the numbers of T cells in the face of ongoing new production from the thymus (at least for a certain period in development) and continuous expansion in response to antigenic stimuli. The American physiologist, Walter Cannon, introduced the term ‘homeostasis’ to describe the tendency of an organism to restore its original status in the face of unexpected disturbances.¹² Permutations of this term are now widely used by immunologists referring to various response modes of T cells when the equilibrium is disturbed. The intrinsic dynamics of the immune system pose constant challenges threatening the equilibrium, and it is therefore understandable that the immune system has developed several layers of homeostatic control mechanisms.

How do t cells notice the presence or absence of other t cells?

T cells have a sensor for space that will drive them to react to changes in the number of other T cells. This can be visualized in an extreme scenario by labelling T cells with the cytoplasmic dye CFSE and transferring them into either a wild-type host or a T-cell-deficient host. In the T-cell-replete host, the adoptively transferred T cells retain their carboxy fluorescein diacetate succinimidyl ester (CFSE) label, whereas, in a T-cell-deficient environment, they will rapidly divide, even in the absence of cognate antigen,^13,14 resulting in gradual loss of CFSE fluorescence (see Fig. 1).

Figure 1.

Carboxy fluorescein diacetate succinimidyl ester (CFSE) labelled T cells transferred into a wildtype host will not divide and therefore retain their label (left panel), whereas they will divide and dilute the CFSE label after transfer into a lymphopenic host (right panel).

Under steady-state conditions, peripheral T cells are organized in two distinct pools: the naïve pool and the memory T-cell pool. Numbers in these two pools are kept stable under conditions where naïve cells are exported from the thymus and memory T cells are generated in the course of immune responses. Importantly, however, peripheral naïve versus memory phenotype T cells are maintained independently, at least under steady-state conditions.^15–17 Such independent homeostatic control of naïve and memory cells would guarantee preservation of a diverse repertoire of T-cell-receptor specificities to maintain readiness of the immune system to cope with novel pathogens and guard against gradual erosion of memory against pathogens from previous encounters. Independent homeostatic control of the two peripheral T-cell pools implies that the sensors for space must differ between naïve and memory T cells, and they are often referred to as occupying distinct ‘niches’. Lymphocytes are highly mobile: naïve T cells recirculate between lymphoid tissues, whereas memory cells are even more versatile and can be found in virtually all peripheral sites. It is therefore difficult to imagine them confined to a defined space or niche in a restricted topological sense, such as hepatocytes which are confined to the liver. T cells and dendritic cells colonize specific areas of lymph nodes and spleen known as T zones¹⁸ and it is thought that the proliferative reaction (homeostatic proliferation) which occurs when T cells find themselves in a lymphopenic environment is initiated in lymphoid organs.^19,20 The number of naïve CD4 and CD8 T cells is co-regulated, i.e. in the absence of either subset the other will compensate to achieve the same total number of naïve T cells;^11,21–23 thus, one would assume these two subpopulations to occupy the same ‘niche’. On the other hand, it was shown that secondary lymphoid organs may be essential only for the maintenance of naïve CD4, but not CD8, T cells.²⁴ This further emphasizes that control of cell numbers is not dictated by physical space in a defined location.

What defines ‘virtual space’ for t cells

Given that there is no inherent limitation of physical space that could be envisaged to restrain the accumulation of T cells, there must be a perception of ‘virtual’ space which limits the expansion and survival of T cells.

For naïve T cells, at least two components can be considered as limiting resources for their maintenance and survival: self-peptide major histocompatibility complex (MHC)-derived signals from antigen-presenting cells (APCs); and the cytokine interleukin (IL)-7, which is produced by stromal cells.^25–27 As these signals are also crucial for the generation of T cells in the thymus, it is difficult to disentangle their impact in the generation from that in maintenance and homeostasis. Overproduction of IL-7 in transgenic mice results in an increase in peripheral T-cell numbers,^28,29 largely owing to an increase in memory phenotype cells that have arisen as a consequence of homeostatic divisions. It is widely accepted that naïve T cells require MHC-derived signals in order to survive^30–32 and respond to changes in the homeostatic equilibrium.^{13,14,33–35} However, alternative interpretations have also been put forward.^36,37 One of the problems encountered in many experimental systems is the fact that naïve T cells lose their status of naivety in lymphopenia where they do not face competition for MHC and cytokine signals. While naïve T cells do not undergo cell division under steady-state conditions,^38–40 many, but not all, divide rapidly and change their phenotype in a lymphopenic environment.^{13,14,33,41–46} Once naïve T cells have undergone phenotypic conversion, they become independent of MHC-derived signals for their survival and therefore can no longer be counted as naïve T cells.

Interestingly, phenotypic conversion to a memory-like phenotype occurs naturally with thymic emigrants in the early neonatal period before peripheral lymphoid compartments have reached full cellularity.^47,48 As normal T-cell numbers are reached between 7 and 15 days after birth, these neonatal T cells are diluted by new thymic emigrants; in monoclonal T-cell receptor (TCR) transgenic mice, such phenotypically converted T cells, while prominent in the early neonatal period, are barely detectable in adults. The competitive nature of this process is illustrated by the fact that proliferation and activation of neonatal T cells is influenced by the number of T cells present in the periphery and strongly regulated by recent thymic emigrants.⁴⁷ The loss of naivety in peripheral T cells that encounter lymphopenic environments suggests that the naïve T-cell compartment should not really be considered a stable compartment, because its maintenance is strictly dependent on continuous thymic output and cannot be sustained by homeostatic mechanisms in the periphery.^49,50

Following encounter with foreign antigen, rather than the tonic interaction with self-peptide/MHC-presenting APC, antigen-specific T cells divide extensively and can transiently generate large numbers of effector T cells. CD8 T-cell numbers increase about fivefold during viral infections, and the majority of CD8 T cells can be antigen specific.^51–53 Competition for limiting resources includes access to antigen-bearing APC (reviewed in ref. 54), costimulatory molecules and growth factors, such as IL-2. Furthermore, T cells may compete by interfering with the expansion of neighbouring T cells, as evident in the suppression of T-cell responses to subdominant antigens.⁵⁵ Moreover, activated T cells may directly or indirectly inhibit the expansion of rival T cells by secretion of inhibitory cytokines, such as IL-10 or transforming growth factor-β (TGF-β), both of which have been implicated in T-cell regulation.^56,57 It is crucially important for the equilibrium of the immune system and the avoidance of immune pathology that the exuberant expansion of effector T cells is kept in check. The action of inhibitory receptors, such as cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4), on T cells, and programmed death-1 gene (PD-1)^58,59 activation of the Fas death pathway,^60,61 as well as passive death owing to loss of contact with protective cytokines,⁶¹ are all involved in the removal of effector T cells.

Activated T cells that escape apoptosis and differentiate to memory cells maintain their numbers by purely homeostatic mechanisms and are not perturbed by changes in thymic T-cell production. Memory T cells do not compete with each other on the basis of avidity for self-peptide/MHC complexes, as naïve T cells or effector T cells do, but their repertoire will also be perturbed by lymphopenic incidents.⁶² The sensor for space in memory T cells is thought to be based on cytokines. CD8 memory T-cell homeostasis is positively regulated by IL-15 and negatively affected by IL-2.^63,64 It is unclear what cytokine(s) control the long-term survival of CD4 memory T cells. While some results have shown that CD4 memory T cells may be maintained in the absence of the common cytokine gamma chain (gc), which is an essential component of the receptors for IL-2, -4, -7, -9 and -15,⁶⁵ a recent study described IL-7 as a survival factor for memory CD4 T cells,⁶⁶ although its joint usage by naïve and memory T cells does not appear to be compatible with independent regulation of the two pools. In contrast to naïve CD4 and CD8 T cells, which seem to occupy the same ‘niche’, the survival of these subpopulations as memory cells is not co-regulated, as evident from their distinct cytokine requirements.

Immune regulation through competition

Regulation of peripheral T-cell responses is an essential feature in the immune system, preventing not only the development of autoimmune disease but also excessive responses to exogenous antigens that could lead to immune pathology. It stands to reason that there is more than one mechanism operative to achieve this important goal, which includes the peripheral deletion of T cells,^67,68 differential action of cytokines,^56,57,69 induction of anergy after suboptimal activation,^69,70 restriction of clonal expansion by CTLA-4,⁷¹ and Fas/Fas ligand interactions.⁷² In the last decade, the notion gained prominence that there must be a dedicated lineage of cells responsible for immune regulation. Although the number of publications regarding regulatory CD4 CD25⁺ T cells has increased exponentially, there is still considerable uncertainty with respect to their developmental origin, characteristic markers and mode of action in vivo or in vitro.^73,74

The existence of a population of unique regulatory T cells was rekindled following a series of experiments that involved the neonatal thymectomy of mice. Thymectomy on day 3 after birth led to the development of a wide spectrum of organ-specific autoimmune diseases, whereas delaying thymectomy to day 7 after birth abrogated this problem. On the basis of these findings it was assumed that a population of regulatory T cells develops later in life than autoimmune effector cells, and that thymectomy on day 3 prevented selectively the emigration of such regulatory T cells.⁷⁵ Given what is known now about establishment of the peripheral T-cell pool and homeostatic mechanisms, there are alternative interpretations for the observed phenomena and it is doubtful that they indeed prove the existence of a subpopulation of T cells which is generated in the thymus and dedicated to the job of immune regulation. For instance, it is clear that the peripheral immune system of neonatal mice contains very few T lymphocytes and can therefore be considered lymphopenic (Fig. 2). Adult numbers of T cells are not reached until day 7 in the lymph nodes and day 15 in the spleen.⁷⁶ As a consequence, naïve T cells proliferate and acquire the phenotype and function of activated/memory T cells,^47,48 an effect that requires self-peptide/MHC interactions and is regulated by the size of the peripheral T-cell pool. In the absence of continuous thymic output, such T cells can expand to the same extent as T cells adoptively transferred into T-cell-deficient hosts. This has serious implications for the functionality of the immune system. Loss of the naïve T-cell pool as a result of homeostatic imbalances in lymphopenia causes selective outgrowth of oligoclonal populations which differ between individual mice and often cause immune pathology, or even autoimmunity.^77,78 There are various other experimental manipulations of mice where lymphopenia results in the development of organ-specific autoimmunity, such as neonatal administration of cyclosporin, high-dose fractionated total lymphoid irradiation and single-chain TCR transgenic mice (reviewed in ref. 79). T-cell depletion, leading to partial lymphopenia, can also result from a number of clinical conditions, such as chemotherapy, radiation, or treatment with cyclosporin A, and many viral or bacterial infections cause substantial lymphopenia.^80–84

Figure 2.

Thymectomy at day 3 after birth results in an oligoclonal peripheral T-cell repertoire (diversity illustrated by different colours) and expansion of a few clones (left panel). Delaying thymectomy to day 7 after birth restores polyclonality in the periphery and avoids autoimmunity.

It is therefore worth considering that dysregulated immune responses may not primarily be caused by the absence of an essential T-cell subpopulation whose function it is to prevent this from happening, but rather by the disturbances in the T-cell repertoire following re-establishment of homeostasis. On this basis we have argued previously that maintaining a balanced immune system may not be a unique job for a particular T-cell population, but rather a side-effect of normal T-cell competition for limited resources,⁸⁵ which prevents excessive outgrowth of potentially pathogenic T-cell clones. To prove this point we made use of a widely employed experimental system for the study of immune regulation in which naïve T cells are transferred into syngeneic lymphopenic Rag^–/– hosts, resulting in a gradual wasting disease about 4–6 weeks later. Disease was manifest in 100% of recipients of small numbers (2–5 × 10⁵) of naïve T cells, whereas we showed that increasing the inoculum of naïve T cells (thoroughly depleted of CD25⁺ and other activated CD4 T cells) transferred into lymphopenic hosts, to 6 × 10⁶ cells, delayed immune pathology and left 50% of the mice disease free. Increasing the inoculum further, to 12 × 10⁶ cells, completely prevented immune pathology. Furthermore, while 92% of the transferred T cells in recipients of small numbers of T cells showed an activated phenotype, only 58% of T cells in mice that had received high doses of T cells were activated.⁸⁶ Even transfer of a monoclonal naïve transgenic T-cell population was able to prevent pathology and this was linked with its ability to expand in response to ‘empty-space’ signals more efficiently than the naïve polyclonal population. These findings are compatible with the hypothesis that competition with neighbouring T cells for limited resources may restrain excessive activation and outgrowth of transferred T cells.

How do CD25⁺ CD4 t cells fit in the picture?

Irrespective of the argument that T-cell regulation may not require a particular dedicated regulatory subset, there is no doubt that CD25⁺ CD4 T cells are very effective at preventing autoimmunity and immune pathology in various experimental settings. While arguments over their phenotype and mechanism of action continue, e.g. whether or not CD25 expression is obligatory and whether they suppress via secretion of IL-10 or TGF-β, or both, it is clear that these cells are good at regulation in relatively small numbers. Nevertheless, one needs to consider that most, if not all, experimental systems for studying the function of regulatory T cells involve at least partial lymphopenia, such as transfer into T-cell-deficient, irradiated, or cyclophosphamide- or antibody-treated hosts. Even transfer into non-obese diabetic (NOD) mice may fall into this category because, despite apparently normal thymic export and peripheral T-cell numbers,⁸⁷ it remains to be determined whether the composition of the naïve T-cell pool is normal in these mice, given that thymectomy as late as 3 weeks of age accelerates disease onset.⁸⁸ This is in marked contrast to the finding that development of autoimmune features following neonatal thymectomy of (C57Bl/6 × A/J) F1 mice depends on strict timing of thymectomy. Removal of the thymus around day 3 after birth induces orchitis, but this is completely prevented if thymectomy is delayed to day 7.⁸⁹ While it is not proof that the underlying reason for prevention of autoimmune orchitis under these circumstances is the cessation of lymphopenia in the peripheral T-cell pool, it seems more than a coincidence that by day 7 after birth adult numbers of T cells are present in the lymph nodes,⁷⁶ so that one might expect a termination of lymphopenia-induced T-cell expansion.

A notable exception to the dependency on a lymphopenic environment for the asssessment of T-cell regulation is an experimental system in which regulatory CD4 T cells were generated in a TCR transgenic model. TCR transgenic mice that co-express their cognate ligand on thymic epithelium⁹⁰ differentiated to antigen-specific regulatory T cells whose suppressive function could be demonstrated in a T-cell-replete host.⁹¹ However, it remains to be determined if their mode of generation and function is related to that of polyclonal CD25⁺ T cells.

The most direct approach to assess the function of CD25⁺ CD4 T cells is their removal from intact animals via antibody treatment. It is clear that this regime does not result in onset of autoimmunity or immune pathology unless it involves subsequent transfer into T-cell-deficient hosts.⁷⁵ On the other hand, it has a very striking effect on immune responses to ‘foreign’ antigens. Depletion of CD25⁺ T cells allows induction of otherwise inefficient responses to tumours,^92–94 and markedly enhances antibacterial^95,96 and antiviral T-cell responses.⁹⁷ Interestingly, infections appear to result in enhanced expansion or recruitment of CD25⁺ CD4 T cells after the peak of effector activity.⁹⁵ It is therefore tempting to speculate that the main function of these cells may be damage limitation caused by exuberant effector cells that could cause immune pathology unless restrained. If viewed in the context of space and competition, one could argue that removing a fraction of T cells with an activated phenotype (all CD25⁺ T cells have the phenotype of an antigen-experienced T cell) creates favourable conditions for expansion of other T-cell clones that are currently exposed to antigen.

There may be several reasons for the fact that potentially autoreactive clones, which are known to exist in the peripheral repertoire, do not seem encouraged to expand in this scenario. Either the ‘hole’ created does not affect quiescent autospecific clones because it is in the activated/memory pool and not in the naïve pool, or alternatively, or in addition, the degree of T-cell depletion may be insufficient to allow significant expansion of autoreactive clones to cause pathology, as it is possible that only clones of relatively low avidity escape from thymic-negative selection. Furthermore, the state of activation of antigen-presenting dendritic cells will influence the expansion capacity of T cells responding to bona fide foreign antigens compared with self-antigens.⁹⁸

It is not yet clear what capacity underlies the competitive advantage of CD25⁺ T cells in regulating expansion of other T-cell clones. Their constitutive expression of a high-affinity IL-2 receptor would equip them well to compete for IL-2, thus depriving effector T cells of this cytokine, presumably compromising their expansion and contributing to their passive cell death by cytokine withdrawal. Furthermore, our recent studies indicate that IL-2 sequestered by CD25⁺ CD4 T cells from responder T cells drives their proliferation and differentiation towards production of the inhibitory cytokine, IL-10 (T. Barthlott, submitted). The constitutive expression of CTLA-4 on CD25⁺ CD4 T cells also appears to be involved in their regulatory function^99,100 and may contribute to competition for B7 molecules on APCs. Engagement of CTLA-4 was, furthermore, found to be linked to the production of TGF-β.¹⁰¹ Thus, in addition to competition for resources commonly required for the expansion of activated T cells, CD25⁺ CD4 T cells may also employ active interference strategies via the secretion of inhibitory cytokines. It is noteworthy that T cells engineered to secrete IL-10 by various means in vitro were shown to suppress immune responses, even in T-cell-replete hosts.^102,103 Similarly, the importance of TGF-β as a crucial factor in the suppression of islet-specific CD8 T cells by CD25⁺ CD4 T cells was shown in adoptive transfers into T-cell-replete hosts; expression of a dominant-negative TGF-β1 receptor type II transgene by islet-specific CD8 T cells abrogated their control by CD25⁺ CD4 T cells.¹⁰⁴

Many questions about this cell population remain. Are they really generated as a separate T-cell lineage in the thymus, or do they arise during antigen encounters in the periphery? If so, what stimulus induces their phenotype and defines their mode of action? The expression of a characteristic transcription factor, Foxp3,^105–107 could indicate differentiation analogous to that of T helper 1 (Th1) and T helper 2 (Th2) subsets, but it is still unclear as to when and how expression of Foxp3 is induced. Phenotypic heterogeneity of T cells, defined as regulatory T cells in operational terms, furthermore emphasizes that there are many ways to achieve regulatory function, and the diversity of experimental systems used to define regulation might have a strong influence on the interpretation of regulatory mechanisms.