|Home | About | Journals | Submit | Contact Us | Français|
Central tolerance is critical in establishing a peripheral T-cell repertoire purged of functional autoreactive T cells. One of the major requirements for effective central tolerance is the presentation of self and other innocuous antigens (Ags), including food, gut flora or airway allergens, to developing T cells in the thymus. This seemingly challenging task can be mediated in some cases by ectopic expression of tissue-specific Ags by thymic epithelial cells, or by entry of systemic blood borne Ags into the thymus. More recently, thymic homing peripheral dendritic cells (DCs) have been proposed as cellular transporters of peripheral tissue-specific Ags or foreign innocuous Ags. The aim of this viewpoint is to discuss the three principal thymic DC populations and their trafficking properties in the context of central tolerance. We will first discuss the importance of peripheral DC trafficking to the thymus and then compare and contrast the three DC subsets. We will describe how they were characterized; describe their trafficking to and their microenvironmental positioning in the thymus; and discuss the functional consequence of thymic trafficking and localization on thymic selection events.
Central tolerance is the mechanism by which newly developing T cells as well as B cells are rendered non-reactive to self antigens (Ags). In order for central tolerance of T cells to occur, self Ags, including tissue-specific Ags (TSAs), must be available in sufficient concentrations in the thymus to be processed and presented on MHC molecules to developing thymocytes. Early studies have shown that blood-borne Ags can access the thymic medulla, and induce clonal deletion mediated by thymic dendritic cells (DCs) [1, 2, 3]. It was therefore proposed that developing thymocytes are censored mainly to systemic peripheral Ags that can access the thymus via the blood stream, but that many potentially autoreactive T-cell clones specific to compartmentalized TSAs would ignore their cognate Ag in the periphery via “clonal ignorance” [4, 5]. However the discovery of promiscuous gene expression by specialized thymic epithelial cells (TECs) in later studies and the identification of the nuclear transcription factor AIRE have broadened our understanding of how the thymus projects the entire “peripheral self” to developing thymocytes . In these exciting studies, many TSAs were found to be expressed by medullary thymic epithelial cells (mTECs) under the control of AIRE [7-9], although AIRE-independent mechanisms of promiscuous TSA expression by mTECs have been reported . In addition to mTECs, BM-derived thymic DCs were also shown to play a major role in presenting ectopically expressed TSA to developing thymocytes [11, 12]. Therefore the relationship between thymic DCs and mTECs in the induction of central tolerance has been addressed in several studies where mTECs can either (i) act as an Ag pool for cross presenting thymic DCs  and/or (ii) autonomously present Ags directly to developing T cells via macroautophagy [13, 14]. However the relative contributions of thymic DCs and mTECs to central tolerance induction are still not clear. In some studies, mTECs have been shown to be more efficient in selecting T reg cells, whereas thymic DCs supported deletion [15, 16]. In contrast, BM-derived APCs were also shown to generate Treg cells but under rather artificial experimental conditions [17-19]. More importantly though, ectopic gene expression in the thymus might not efficiently present temporally regulated gene products, or the plethora of Ags of the gut microbiome, food Ags or other foreign but innocuous Ags that access mucosal sites . Therefore alternative mechanisms must exist to broaden the spectrum of TSAs or innocuous Ags presented to developing T cells.
The role of extrathymic DCs in peripheral tolerance is well established: immature DCs or DCs matured in the absence of pathogen pattern recognition signals can efficiently induce adaptive regulatory T (Treg) cells in peripheral lymphoid tissues . However, studies in the past decade suggest that peripheral migratory DC populations can also transport peripheral Ags to the thymus for central tolerance . Seminal studies by Bonasio et al.  showed that exogenous bulk DCs from the spleen can access the thymus after adoptive transfer, and that endogenous peripheral DCs also home into the thymus in parabiosis experiments. These studies also established the trafficking programs utilized by bulk splenic DCs to access the thymus, which involve P-selectin and interactions of the DC integrin α4β1 with its endothelial ligand VCAM-1 . The multistep DC homing cascade from the blood into the thymus was found to be also dependent on a pertussistoxin sensitive G-protein coupled chemokine receptor , and subsequent studies from our group defined a key role for the chemokine receptor CCR9 in thymic recruitment of pDCs . Chemokines and their receptors also regulate the homing of DC subsets within the thymus, directing localization to the medulla, corticomedullary (CM) junction and/or perivascular spaces [23, 24].
The functional consequences of DC trafficking and microenvironmental homing in the thymus have begun to be elucidated (reviewed in [12, 25, 26]), with evidence that DCs can transport peripheral Ags to the thymus to mediate clonal deletion of developing Ag-specific thymocytes, and the DCs may also direct the selection of natural Treg (nTreg) cells depending on the nature of the DC population [26, 27]. Thymic DCs are heterogenous, and not all subsets are migratory [28, 29]. Up to 50% of thymic DCs appear to arrive from the peripheral circulation , potentially bearing peripheral tissue Ags for thymocyte selection. Three subtypes of thymic DCs have been characterized in the mouse: CD11clow B220+ plasmacytoid DCs (pDCs) and two phenotypically and functionally distinct subsets of CD11c+ B220- conventional DCs (cDCs) . The thymic cDC populations have been described as CD8αlow CD11b+ SIRPα+ cDCs and CD8αhigh CD11b- SIRPα- cDCs, which we will denote in this Viewpoint as SIRPα+ and SIRPα- cDCs respectively . Similar thymic DC populations have been described in the human . Table 1 summarizes the thymic DC populations and their role in central tolerance.
Thymic DCs were originally thought to arise intrathymically from a common DC/T-cell precursor . Experiments performed a decade later, using selective combinations of normal, parabiotic and radioablated mice revealed distinct origins of CD8α+ vs CD8α- cDC subsets in the adult mouse thymus (SIRPα expression was not monitored on DCs in these earlier studies) . Subsequent studies, using adoptive transfer experiments, parabiosis and fetal thymic transplantation, established that the SIRPα+ (CD8α-) cDC population migrates into the thymus from the blood [26, 31], whereas the SIRPα- (CD8α+) subset develops intrathymically from bone marrow (BM) derived precursors (see section on “Thymus resident (non-migratory) SIRPα- cDCs” below). In co-culture with thymocytes, SIRPα+ cDCs induced clonal deletion of Ag specific T cells, and also the development of thymic Treg cells . In fact, to date SIRPα+ cDCs are the only thymic DC population that has been shown to induce natural Treg cells in vitro from developing thymocytes. Even though studies on SIRPα+ cDCs have focused mainly on a model of CD4+ T-cell deletion , the poor-to-modest cross presentation ability of SIRPα+ cDCs  would suggest that they do not play a major role in the thymic deletion of CD8+ T cells.
Unlike the bulk of thymic DCs, which are found in the medulla and are sparsely detectable in the cortex [34-36], SIRPα+ cDCs are found primarily in the cortex and the perivascular regions of the thymus . They are highly endocytic, and have been shown to sample blood borne antigens . SIRPα+ cDCs, but not SIRPα-DCs or pDCs, were shown to be selectively decreased in the thymus of CCR2-deficient mice, suggesting a role for CCR2 and its ligands in their development, homing or survival . Moreover, consistent with a role for SIRPα+ cDCs in thymic selection, CCR2-deficient mice exhibited a significant if modest impairment in the ability of intravenously injected Ags to induce clonal deletion of Ag specific thymocytes . BM SIRPα+ cDCs have been shown to egress into peripheral blood in response to CCR2-mediated signals, which might explain the deficiency in the thymus; but monocyte chemotactic protein-2 (MCP-2 or CCL8), a potential ligand for CCR2 , was constitutively detected in the thymic perivascular region where the SIPRα+ cDCs are localized . Moreover, CCR2 was expressed by a portion of intrathymic SIRPα+ cDCs, but not SIRPα- cDCs . Thus, although direct experimental confirmation is needed, it is likely that CCR2-CCL8 contributes to intrathymic localization of SIRPα+ cDCs, particularly in the perivascular spaces. Taken together, the ability of SIRPα+ thymic cDCs to sample blood-borne Ags, their unique intrathymic localization, and their ability to induce clonal deletion and nTreg induction [23, 27] suggests a specialized role for SIRPα+ thymic cDCs in the development of central tolerance to blood-borne Ags. Since SIRPα+ cDCs migrate into the thymus from the blood, they may also contribute to central tolerance through the presentation of Ags that they acquire in the periphery prior to thymic localization, although this possibility remains to be assessed.
SIRPα- cDCs comprise the most abundant thymic DC population: they make up approximately two thirds of cDCs and one half of all DCs in the thymus [29-31]. SIRPα- cDCs express CD8α and, like splenic CD8α+ cDCs, they can cross-present exogenous Ags into the MHC class I pathway for presentation to CD8+ T cells. They are found predominantly in the medulla, and have been shown to cross-present TSAs expressed ectopically by mTECs to developing thymocytes for central tolerance [29, 38, 39]. Unlike migratory SIRPα+ cDCs, SIRPα- cDCs are resident cells: they arise from intrathymic precursors, and do not exchange efficiently with the circulating peripheral DC pool . Despite being derived from intrathymic precursors, there has been significant debate on whether thymic SIRPα- cDCs derive from a common T/DC progenitor or from separate T-cell and DC-progenitors that seed the thymus [[30, 31, 33] versus [40, 41] respectively]. Recent cell lineage tracing experiments and studies using fluorescent reporter mice argue that T cells and myeloid cells (including thymic DCs) arise from distinct precursors in the thymus [42, 43]. Nevertheless, the intrathymic precursor of thymic SIRPα- cDCs is derived from thymic homing BM progenitors. CCR7 and CCR9 have been implicated in thymic recruitment of BM-derived thymocyte precursors [44-46], and it is likely that one or both of these chemokine receptors are involved in controlling the representation of the SIRPα- cDC population in the thymus as well. In fact, in mixed BM competitive chimeras where WT- and CCR9-deficient BM-derived DCs develop side-by-side, the majority of SIRPα- thymic cDCs are derived from WT vs CCR9-deficient progenitors; whereas WT and CCR9-/- BM contribute equally well to the migratory SIRPα+ cDC compartment in the thymus . Moreover thymic SIRPα- cDCs have been found to be even more significantly reduced in numbers in CCR7/CCR9-double deficient mice (our unpublished findings). Regardless of the mechanisms of precursor recruitment, SIRPα- cDCs are believed to represent thymic resident cDCs that can tap into selfAg reservoirs that may be less accessible to the migratory DC subsets . It would be interesting to assess in future studies, whether incoming migratory SIRPα+ cDCs could also tap into the promiscuously expressed thymic Ag pool, or whether instead migratory cDCs with potentially poor cross-presenting capabilities import only extrathymic Ags into the thymus, particularly those from mucosal sites.
Mechanisms of DC recruitment to the thymic medulla, where most thymic DCs and, in particular, the resident thymic subset reside, have recently been elucidated. The chemokine XCL1 (also known as lymphotactin) is produced predominantly by the MHC class IIhigh subpopulation of mTECs and mediates medullary accumulation of thymic DCs that express the cognate XCR1 receptor . In contrast, CCR7 and CXCR4 play a minor role, if any, in the accumulation of DCs in the thymic medulla. Among the thymic DC subpopulations, XCR1 was most highly expressed on lymphoid CD11b- DCs, a population that largely overlaps with resident SIRPα- cDCs (these studies did not characterize thymic DCs based on SIRPα expression). Moreover, in Xcl1-deficient mice, thymic DCs failed to localize in the medulla and instead accumulated in the deep cortex and CM regions . XCL1-mediated accumulation of CD11b- cDCs to the medulla may contribute to efficient interactions between mTECs and CD11b- cDCs. Xcl1 deficiency also resulted in reduced numbers of thymic nTreg cells  suggesting, albeit indirectly, that SIRPα-, as well as SIRPα+ cDCs (as discussed in the “Migratory (thymus homing) SIRPα+ cDCs” section above), may be competent to induce nTreg cells, and that their induction of nTreg cells may depend upon their medullary localization.
In sum, thymic cDCs can be divided into two major subsets: a “resident” SIRPα- CD8α+ CD11b- population and a “migratory” SIRPα+ CD8α- CD11b+ cDC population. XCL1 expressed by thymic mTECs targets the “resident” SIRPα- cDCs to the medulla  where they cooperate with mTECs [11, 47]. The interaction between such SIRPα- cDCs and mTECs is thought to contribute to nTreg-cell induction. CCR2-CCL8 interactions on the other hand, may be important in positioning SIRPα+ migratory cDCs to the cortex and the perivascular regions of the thymus where they can sample soluble blood-borne Ags , and could also potentially present Ags that they transport from the periphery.
Plasmacytoid DCs are important innate immune cells that produce type I interferons in response to viral infections . Early studies showed that thymic pDCs are recruited from the blood, as opposed to originating from intrathymic precursors . In their unactivated immature state, pDCs are poor presenters of Ags, leading to suggestions that a primary role of pDCs in the thymus, as elsewhere, might simply be to protect the tissue from viral infections . However, in the absence of microbial stimulation, pDCs can induce immune tolerance in the periphery through the induction of Treg cells .
We described a tolerogenic population of pDCs in lymphoid tissues that expresses the chemokine receptor CCR9 , and is involved not only in homing of memory and effector lymphocyte populations to the small intestines [51, 52], but also in progenitor T-cell homing to the thymus . CCR9+ pDCs efficiently induce Treg cells from peripheral T cells, and inhibit immune responses in vitro and in vivo . The high expression of CCR9 by these immunosuppressive pDCs, the involvement of CCR9 in T-cell progenitor recruitment to the thymus and the known expression of its chemokine ligand CCL25 by thymic epithelial cells, led us to the finding that CCR9 mediates peripheral pDC trafficking to the thymus . Interestingly, CCR9 deficiency did not completely block thymic pDC recruitment in our studies, suggesting additional, overlapping homing mechanisms: CCR7 is a candidate, since CCR7 as well as CCR9 participates in T-cell progenitor homing into the thymus [44, 45], and a recent study has revealed a functional role (in lymph node homing) for low level CCR7 expression on circulating pDCs . In agreement with previous studies implicating P-selectin and α4 integrins in thymic DC homing [21, 54], we also found that pDCs expressed α4 and P-selectin ligands, and that endogenous thymic pDC migration was inhibited by α4 integrin blockade . Thus peripheral pDC homing to the thymus likely employs the same adhesion cascade used by T-cell progenitors and bulk splenic DCs, although pDC recruitment to the thymus appears to be more dependent on CCR9 [22, 44, 45].
Ag-loaded peripheral pDCs that access the thymus localize to the CM junction, and mediate efficient clonal deletion of developing Ag-specific thymocytes . In our hands, we did not observe thymic nTreg-cell induction after intravenous injection of Ag-loaded peripheral pDCs . This is consistent with earlier studies in which mouse thymic pDCs failed to efficiently induce Treg-cell production from thymocytes in vitro . In contrast, human thymic pDCs can induce FOXP3+ Treg cells in culture models, and TSLPR+ pDCs co-localize with FOXP3+ Treg cells in the human thymus . The failure of Treg-cell induction by thymic DCs in many mouse model studies ([21-23] and reviewed in [12, 26]) may be a function of the transgenic TCR-Ag model systems employed, reflecting unique characteristics of signaling through the transgenic TCR. Several studies have indeed correlated TCR affinity and signal strength with differential effects on clonal deletion versus Treg-cell induction, such that high affinity interactions mediate clonal deletion, whereas lower affinities, in comparison, rescue developing thymocytes from death and shunt their development into Treg cells [13, 56-59]. Therefore caution has to be exercised in interpreting the results: additional studies will be required to determine critically the importance of pDC versus “migratory” and “resident” cDC populations, and of TCR affinity, in Treg-cell induction versus clonal deletion modalities of central tolerance.
It is interesting to consider that, in addition to peripheral self Ags, thymic homing DCs may also transport innocuous foreign Ags to the thymus, such as those from allergens in the respiratory tract or food or innocuous flora in the digestive tract . Indeed, CCR9 can mediate pDC localization to the intraepithelial compartment of the gut wall, a site well positioned to sample Ags in the gut lumen [22, 60]. Whether peripheral DCs, in particular pDCs, can transport locally endocytosed mucosal Ags to the thymus, remains to be determined, however.
Importantly, transport of peripheral Ags into the thymus needs to be carefully controlled to prevent central tolerance to pathogen-associated molecules. Such control appears to be achieved at least in part by suppression of thymic homing of peripheral DC populations by pathogen-associated pattern recognition through toll like receptors [21, 22]. In the case of pDCs, for example, we showed that TLR9 ligands, which mimic microbial DNA, efficiently downregulate CCR9, preventing pDC-mediated transport of pathogen-associated Ags into the thymus .
In conclusion, CCR9+ pDCs and SIRPα+ cDCs comprise the rapidly exchanging migratory DCs in the thymus . CCR9 is critical in targeting pDCs to the CM region of the thymus where they participate in clonal deletion of developing Ag-specific thymocytes . Moreover pDCs may be specialized in transporting innocuous peripheral Ags into the thymus, potentially including food or intestinal microflora Ags. Migratory SIRPα+ cDCs on the other hand are targeted to the thymic cortex and localize to perivascular spaces in part by means of CCR2-CCL8 interactions , and contribute to clonal deletion and nTreg-cell induction  to systemic blood borne Ags . Whether they also transport peripherally endocytosed Ags to the thymus for central tolerance, supplementing the transport function of pDCs, remains to be determined. Finally the resident SIRPα- cDC population, believed by some to arise from a common T-cell/DC progenitor (reviewed in ), is targeted by XCR1-dependent migration to the thymic medulla , where these cells cross-present mTEC-expressed Ags to developing thymocytes (reviewed in ). In general all the thymic homing DCs, and the migratory precursors of thymic resident DCs, are assumed to home from the blood into the thymus by means of common mechanisms of P-selectin rolling and α4β1/VCAM-1-mediated arrest on vessels of the CM junction ; but the involvement of multiple chemoattractants allows differential recruitment and microenvironmental positioning of these specialized antigen presenting cells that mediate central tolerance to self and innocuous non-self Ags.
H.H. is a recipient of an Investigator Career Award from the Arthritis Foundation and was a fellow under NIH Training Grant AI07290. Supported in part by NIH grants AI093981, DK084647, and AI047822 to E.C.B., and by a CIRM grant to H.H. and E.C.B.
Conflict of interest: The authors declare no financial or commercial conflict of interest.