In vivo maturation and migration of dendritic cells.

Abstract

The hallmark of immunity is antigen (Ag) recognition. From early life until death we face countless Ags; many are deleterious and come from the outside world. However, lymphocytes, the mediators of specific acquired immunity, are usually not exposed where foreign antigens are encountered. Thus a system must ferry these Ags from the periphery into inner tissues where lymphocytes comfortably reside or traffic. Clearly, translocating Ags from the periphery to the lymphoid niches is a pivotal function whereby the dendritic cell (DC) system initiates immune responses. DCs, however, are not passive Ag couriers; a dynamic process not fully understood is triggered by Ag deposition. Strategically poised at the boundaries between the inner and the outside world, in a way bridging innate and acquired immunity, DCs sample, trap, engulf, and digest Ags of very diverse origin, in what is called Ag processing. Antigenic fragments are then regurgitated, exposed at the cell surface through one of three molecular pathways [Major histocompatibility complex (MHC) CI and CII, and CD1)] responsible for an efficacious Ag presentation. Concomitantly, DC bear an arsenal of powerful costimulatory molecules [CD40; CD80; CD86; DC-specific/intercellular adhesion molecule type 3-grabbing, non-integrin (DC-SIGN)] and the potential to produce critical cytokines [chemokines, interleukin-12 (IL-12), etc.], thus ensuring the initiation and the fate of acquired immunity. The DC system is particularly efficient, not just translocating, but also transforming antigens sampled at the earlier local milieu into an immunologically accesible language, becoming an excellent reporter of the previous antigenic past, translating the conundrum posed by viruses, bacteria, proteins, etc., into T-cell receptor (TCR) ligands; a short-peptide vocabulary that T cells especially now understand. A highly dynamic process is thus triggered by encountering Ags; while these are carried and processed to be appropriately presented, DCs in turn experience a variety of changes: migratory, phenotypical, functional; all encompassed in the term maturation. Essentially, DC maturation means to change from an antigen-capturing to an antigen-presenting, T-cell-priming mode, a process whereby DCs convert antigens into efficacious immunogens, express the necessary cytokines and costimulatory molecules, thus appropriately initiating the specific, acquired clonal immunity. Though the term DC lineage has been loosely used, that seems changing to a multilineage concept, as, for instance, different DC types can be generated from a variety of different progenitors and precursors.1 With the exception of FDC that are located exclusively within follicles and do not seem to traffic,2 all DC types known are BM derived. However, the potential mechanisms controlling the production of DC or their progenitors within BM, are not known. Nor it is known whether these putative controls are autonomous, exclusive of BM, or dependent on exogenous-paracrine influences, or even affected by external environmental stimuli like microbial products. For instance, the daily BM production of DC, of DC progenitors or precursors, as well as their subsequent release into the blood stream, is virtually unknown. Certainly, however, DC circulating in blood are scarce (less than 1%), but given the ceaseless steady-state DC traffic between blood, lymph and tissues,3 a constant BM supply of DC or DC precursors is assumed. Research into when during ontogeny, or under which mechanisms of control, DC or DC precursors colonize peripheral tissues like intestine, skin, respiratory airways or the genital mucosa, is scarce. Nor it is known whether developmental synchrony exists both for colonizing the periphery and for phenotype maturation. In mouse skin for instance, epidermal DC are present in newborns at day 0, but with the exception of adenosine triphosphatase (ATPase) and CD45, they lack most adult markers (Heras M, manuscript in preparation), which appear later. Also, it is not known whether these DC in newborns will perform any of the various functions that their adult counterparts do (Fig. 1). Dendritic cell (DC) system without antigens (Ags). 1 Haemopoietic pluripotential cells (HPC) under as yet unknown influences constantly generate dendritic cell (DC) progenitors, which, likewise, 2 generate DC precursors. 3 These enter the blood stream and, together with circulating monocytes (Mn), Mn from the blood marginating pool (BMP), and plasmacytoid precursors, constitute the potential largest reserve of blood DC precursors. 4 By undefined mechanisms, precursors and DC leave the blood to colonize peripheral tissues (skin, intestinal, genital and airways epithelia) as sentinel immature DC. 5 In absence of exogenous Ag there is a constitutive ‘steady-state’ flux toward lymphatics, some DC carry apoptotic remains from fraternal cells, and some Mn might be stationed at the abluminal side and reverse transmigrate into lymphatics. 6 Lymph-derived DC congregate in T-cell zones or paracortex, presumably exposing self-Ags without costimulators, and peripherally tolerizing self-reactive lymphocytes (Lc). Plasmacytoid precursors leave blood through HEV and, if encountering appropriate signals (CD40L, IL-3) become DC localizing in T-cell zones and probably inside germinal centres (GCDC). Meanwhile, T cells specific for exogenous Ag remain naive (TN) and continue trafficking unaltered. 7 Presumably, a big proportion of apoptotic DC death occurs here by as yet unknown mechanisms, but likely the outcome of at least Fas–Trance–CD40L interactions. Whether senescent DC ‘retire’ and remain here, and for how long, is also unknown. 8 Some DC might exit lymph nodes (LN) and start a still undefined pathway to recirculate. Concomitantly, unperturbed T cells continue recirculating, still in a naive stage. (MHC CII, dark-green colour.) Classical experiments suggest that in skin, for instance, the turnover of epidermal DC is of several weeks in the steady state. When labelled thymidine is administered to mice during five days, no epidermal DC are marked,4 whereas most DC in mesenteric lymph are labelled, implying a more rapid turnover of the latter cells. Furthermore, prolonged survival of DC has been documented in skin transplants, for as long as 6 months.4 Of note, resident DC in the respiratory tract decline substantially within 3 days of irradiation, and BM reconstitution indicates a half life of about 2 days, comparable to that proposed for gut wall DC.5 Holt thus suggests that DC in mucosal sites have shorter replenishment times than in tissues such as the skin. Recently, Matsuno and colleagues6 described hepatic DC in an immature, postmitotic end-stage, which, according to BrdU feeding have been recently generated. These cells ingest particulate material, translocate from blood to lymph localizing in the T-cell zone of celiac nodes, and their monocytic origin is implied.6 Nevertheless, it would appear that once in the periphery, DC become terminal cells unable to divide, as DC in draining lymph nodes (DLN) and most tissues seem radioresistant,7 and no mitotic figures of tissue DC have ever been described.7 Though in vitro several cell types can generate DC, perhaps the more abundant potential source of BM-derived DC precursors in vivo are blood monocytes, as their in vitro treatment with granulocyte macrophage–colony-stimulating factor (GM–CSF) and IL-4 rapidly generates functional DC.8 Further in vitro experiments showed that in apparent absence of these cytokines, phagocytosing monocytes differentiate into DC while reverse-transmigrating endothelium toward the lumen,9 reminiscent of DC moving into lymphatics. Initial experiments in monocyte-deficient op/op mice implied that the relative contribution of this potential pathway might not be affecting the overall DC status in vivo, as these animals have readily detectable DC in skin, spleen, lymph nodes and Peyer's Patches, despite the severe monocyte deficiency.10 However, a re-examination of these op/op mice studies revealed that the acquisition and presentation of particulate antigens in vivo, such as subcutaneously injected microspheres, might define a population of monocyte-derived DC responsible for these discrete functions.11 Another potential source of circulating DC precursors is the human plasmacytoid cell type,12 apparently identical to the natural interferon-α/β (IFN-α/β) -producing cells (IPC).13,14 Under viral stimulation, IPC become DC inducing IFN-γ and IL-10 production by T cells.14 However, when cultured with CD40-ligand (CD40L) and IL-3, these cells become functional ‘type 2’ DC, the putative human lymphoid-derived DC, which seems to skew T-cell responses toward T helper type 2 (Th2) polarization.15 Circulating plasmacytoid monocytes also produce large amounts of type I IFN but, unlike DC, enter inflammed lymph nodes from blood through high endothelial venules (HEV).16 The propounded counterpart DC ‘type 1’, deemed the monocyte-derived DC, shifts T-cell responses toward Th1.15,17 It has been recently argued that the same DC population has the plasticity to lead to both T helper polarizations, depending on the DC stage.1,18 In any case, an important number of monocytes are in the blood marginating pool, constituted by microvasculatures of liver and spleen where blood flow is substantially slow. They represent good candidates for blood DC precursors in vivo, especially as particulate materials can be ingested at these locations.7 A fundamental achievement in DC research was their in vitro production, either from pluripotential progenitors19,20 or monocytes8 treated with GM–CSF, generating a prevailing belief that this cytokine was indispensable to produce DC in vivo. However, genetically manipulated animals lacking GM–CSF,21 or its receptor,22 soon proved otherwise as these mice exhibited other anomalies, but were able to develop apparently normal DC. Other genetically altered animals such as RelB–/–23 and transforming growth factor-B (TGFB–/– mice24 did reveal alterations in the DC system, as the former has Langerhans cells (LC) but not fully mature splenic DC, whereas the latter lacks epidermal LC,24 implying at least differential pathways during in vivo DC ontogeny. Studies in OX40Ligand–/– mice have recently uncovered that peripheral DC might be defective only at a defined stage of their functioning. These mice exhibited an impaired contact hypersensitivy (CHS) response, apparently due to a defective DC performance during the induction of T-cell priming, but not during the challenge phase.25 In contrast, cutaneous and lymph node DC from Rag 2–/– mice seem functionally defective both in the induction and the elicitation phase, and their functional maturation seemed dependent on the presence of normal T cells.26 Likewise, Ikaros mutant mice27 as well as athymic animals have also revealed alterations within the DC system,26 though less well characterized. Once an Ag overcomes barriers such as the skin stratum corneum, mucus, genital secretions, etc., it will then encounter other ones such as epithelial cells, monocytes, and probably DCs. Given their anatomical location, many cell types like epithelial cells of the genital tract or keratinocytes in skin might encounter incoming Ags; however, only some are capable of performing the next steps toward immunity: Ag uptake and degradation. While some cells might have these latter abilities, only few will be able to continue up to Ag processing. Even further on, very few cell types will have the ability to do what is required next: directed migration toward lymphoid tissues. The culmination of this process initiated by peripheral Ag deposition will be: (a) the non-random accumulation of these Ag-ferrying and processing cells in the paracortex and T-cell areas of peripheral lymphoid tissues, (b) the phenotypic and functional switch in DC from an Ag-sampling to an Ag-presenting T-cell priming stage, and (c) the intimate clustering of DCs with a matched, low-frequency Ag-specific clone, a phenomenon deemed the immunological synapse (Fig. 2). Dendritic cell (DC) system with antigens (Ags). 1 Haemopoietic pluripotential cells (HPC) presumably increase production of DC progenitors and, 2 DC precursors, 3 augmenting influx to blood of DC precursors, of Mn DC precursors from the BMP and, depending on the Ag (viral), of plasmacytoid DC precursors also. Activated endothelium is now able to interact with scarce circulating DC, perhaps facilitating their egress. 4 Encountering Ags (red colour) in peripheral tissues are likely to be accompanied by microbial products (lipopolysaccharide, LPS) and early proinflammatory molecules (interleukin-1, IL-1; tumour necrosis factor, TNF; CD40-ligand, CD40L). 5 All these events concur to differentiate DC precursors and activate peripheral immature DC. These downregulate molecules holding them locally (E-cadh) and crawl toward lymphatics using MDR1, integrins and chemokine receptors to move and orientate themselves. Some abluminal Mn become DC while phagocytosing and reverse migrating to the lumen, while lymphatics upregulate molecules (adhesion molecules, SLC) helping DC to navigate to lymphoid tissues. 6 Whether the bifurcation for Ag-bearing DC to enter follicles or to congregate outside in the T-cell zones is simply stochastic, is unknown. Mobilized maturing DC downregulate phagocytosis, increase costimulators (CD40, CD80, CD86) as well as the proportion and time of Ag (red)–MHC (dark green) complex exposure. They also produce chemokines attracting naive (DC-CK1) and activated (MDC) T cells, facilitating (DC-SIGN) intimate encounters with low-frequency specific naive T lymphocytes, which scan their membranes during the immunological synapse. If Ag is of viral origin, presumably more plasmacytoid-derived DC are produced, some localizing within GC. In the medullary cords, other DC associate with plasmablast rescuing them from apoptosis. 7 DC fate afterwards is unclear. Presumably, most DC subsets die here by apoptosis, perhaps the outcome of tight clustering with strongly activated (FasL-expressing) T cells and the interplay between Fas, CD40L and TRANCE. Whether putative senescent DC survive, and if they do, for how long, and how they die, remains unkown. 8 T cells emerge activated (T act) to recirculate through blood and tissues becoming effector cells, and DC-rescued plasmablasts become Ab-producing cells. 9 Certain activated T cells will be homing in to the sites of their initial Ag encounter, mounting a response there. It is unclear whether matured Ag-bearing or senescent DC recirculate and how. (Exogenous Ag, red colour; MHC CII, dark green colour.) Though substantial work has been done on DCs recently, the very early steps ensuing Ag encountering in vivo are mostly unkown, especially regarding the initial DC mobilization. Nevertheless, most of the knowledge derives from skin DCs, either murine or from human skin explants. Early after applying contact sensitizers, several proinflammatory cytokines are induced locally, especially IL-1 and tumour necrosis factor-α (TNF-α).28 Interfering with TNF-α has revealed that this is a key cytokine for the initial Ag-triggered DC mobilization.29 For instance, animals lacking TNF receptor have defective DC migration;30 anti-TNF Abs inhibit both delayed-type hypersensitivity (DTH) reactions31 and mobilization of DC from the periphery into lymph32 and DLNs.29 Conversely, cutaneous TNF application induces migration of peripheral DC into DLNs.33,34 As at least in skin this occurs within minutes of Ag contact, a cellular source with preformed ready-to-use TNF is inferred. Mast cells are good candidates, as, in addition to their location at mucosae and skin and their early involvement in inflammation, they bear important molecules that can affect DC mobilization, maturation and survival, such as TNF, IL-4 and CD40-ligand.35 Recently, the p-glycoprotein, MDR1, was implicated in the early mobilization of skin LC, as these remained in the epidermis when using an MDR1 antagonist in skin explant cultures.36 It has also been shown that E-cadherin on DC, by allowing binding to keratinocytes, contributes locally to control early DC mobilization from skin.37 Indeed, E-cadherin downregulation on LC seems to be one of the first signals in this process, allowing DC to detach from their cutaneous neighbours to start crawling to lymphatics. Likewise, most epidermal LC, but not lymph node DC, express α6-integrin, whose ligand Laminin is found in the extracellular matrix (ECM) of epidermis, lymphatics and lymph node (LN) paracortex. Both the spontaneous and the induced LC migration from epidermis, as well as the arrival of DC into DLN, are blocked with anti-α6 antibodies. However, while epidermal DC retracted from keratinocytes and adopted a round morphology as if preparing to emigrate, they did not leave the epidermis.38 Further, after skin culture or in vivo stimulation, non-random ‘cords’ of DC appear within putative lymphatics, both in human and mouse dermis.39,40 Additional work in CCR2 null mice has also shown that this chemokine receptor seems required for the DC to migrate from dermis into the DLN.41 Interestingly, apparently the first alteration of in vivo DC migration related to a clinical entity, was recently described.34 Indeed, CD40L–/– mice, the equivalent of human hyper-immunoglobulin M (hyper-IgM) syndrome, exhibited defective Ag-induced mobilization of epidermal DC into the DLN, with a consequently faulty CHS response. This is perhaps the earliest DC migratory defect identified in vivo, especially in the context of Ag exposure. The IL-10–knockout (KO) mice uncovered an apparently inhibitory effect of IL-10 upon epidermal DC migration during Ag deposition, as these animals display increased DC traffic into DLN and a consequently enhanced CHS response.42 Nevertheless, the earlier migratory steps from epidermis into dermis are not fully understood yet. It is also unknown whether DC mobilization into lymphatics, being steady state or Ag-triggered, is mainly a passive phenomenon due solely to the ‘absorbing’ effect of the negative pressure of these vessels. Nor it is known whether the DC trafficking under steady-state conditions is the same population mobilized upon Ag deposition, or a differently specialized DC subset performs this function. Conjoint action of chemokines and adhesion molecules provide a form of traffic signals that guide DC to navigate through different tissues. In vitro studies have shown that DC equivalent to immature peripheral DC populations bear themselves functional chemokine receptors of an inflammatory pattern, whereas mature DC switch these receptors to what is called a primary response pattern.43,44 Blood DC bearing the P-selectin-ligand tether and roll on a permissive activated endothelium expressing P- and E-selectin,45 resembling DC leaving blood toward inflammed tissues. In the Peyer's Patches, CCR6/MIP3α recruit myeloid DC toward mucosal surfaces whereas CCR7/MIP3β attracts lymphoid DC into the T-cell zones.46 Also, TNF-α-activated but not resting LC upregulate CCR7 expression, whose ligand secondary lymphoid tissue chemokine (SLC), is found both in lymphatic endothelium and in secondary lymphoid tissues,47 thus providing the initial guidance to DC entering lymphatics on their way to the DLN paracortex (Fig. 2). Studies in knockout animals have revealed the involvement of cytokines such as lymphotoxin in the lymphoid tissue organogenesis.48 However, it is conceivable that DC residing within lymphoid tissues, by virtue of the great variety of chemokines they can produce, might have discrete roles upon the fine segregation of T and possible B cells in lymphoid tissues. In fact, DC have been shown to be involved in the de novo formation of organized lymphoid tissue during the induction and maintenance of autoimmune responses.49 Situations such as these are likely, especially when highly intimate Ag-driven clustering between DC and lymphocytes are occuring. Perhaps this will clarify a phenomenon variously described as ‘lymphocyte trapping’ or ‘lymphocyte shutdown’, refering to the dramatic fall in lymphocyte output from lymph nodes upon Ag challenge. Indeed DC have been shown to produce a large array of chemokines in vitro including IL-8, RANTES (regulated on activation, normal, T-cell expressed, and secreted), thymus and activation regulated chemokine (TARC), etc.,50 to congregate T cells and plasmablasts in vivo51,52 and B cells in vivo and in vitro53,54 during Ag-driven reactions. The chemokine macrophage-derived chemokine (MDC) for instance, is upregulated during in vivo maturation and migration of epidermal LC into the DLN.55 Interestingly, MDC from DLN–DC was very good at attracting activated but not naive T cells, whereas the chemokine DC–CK1, expressed in situ by DC in the paracortex and germinal centre (GC), is chemotactic for naive T cells.56 Moreover, the rediscovered molecule DC-SIGN, functions on DC engaging intercellular adhesion molecule-3 (ICAM-3) on naive T cells, partially explaining the exclusive striking DC potency to activate naive T lymphocytes.57,58 Thus, differential expression of defined molecules and chemokines by DC subsets within lymphoid tissues, by coordinately attracting naive or activated T cells, will ensure contact with Ag-specific T-cell clones, whether for a primary or a secondary immune response. While most studies emphasize T-cell activation by DC, little is known regarding DC effects upon B cells. This is plausible because, as DC migrate and mature downregulating Ag degradation, remains of native Ag might be kept, suitable for B-cell recognition.59,60 Evidence in vitro and in vivo indicate that DC might directly affect naive B cells to become antibody-producing cells, thus influencing the priming of antibody responses also.53,54,59 For instance, an in vivo protocol of low-dose DC-targeted immunization proved highly efficient to rapidly induce (5 days) elevated antibody titres in a single step.61 Though human naive B cells have been demonstrated in close association to interdigitating DC (IDC) in situ,53 the extrafollicular B-cell growth, for instance, is spatially different.52 Regarding extrafollicular naive B-cell priming, Gordon suggests60 that given the few probabilities for this triad (DC, B and T cells) to coincide temporally and spatially, activated DC might be ‘licensed’ to provide themselves all signals required, including CD40-ligands. This seems supported by earlier findings because, when omitting DC from limiting dilution cultures, PP B cells stop producing IgA.62 Furthermore, the tissue origin of DC, but not of T cells, determined not only the isotype secreted but also the extent of pre-B-cell differentiation in vitro.63 On the other hand, recent in vivo work from MacLenann's group52 indicates that an apparently different DC subset located at the medullary cords and not associated with T cells might be responsible for plasmablast differentiation and survival, thus regulating antibody production extrafollicularly. Also, in vivo work from Kosco's group has identified another DC type that early after cutaneous immunization (first 24 hr) localizes not in the paracortex but within primary follicles, and, upon in vivo transfer into naive mice, induces immunoglobulin production.64 Thus, it is not clear yet if DCs delivering Ag into the paracortex are exactly the same population bringing this same Ag into primary or secondary follicles, nor it is known whether routing Ag extra-or intrafollicularly is simply stochastic. Thus, the role of DC in intiating or maintaining both extra-follicular as well as primary or secondary follicular B-cell reactions, is a topic yet to be investigated (Fig. 2). Apart from endogenous cytokines, exogenous bacterial products such as lipopolysaccharide (LPS) can mobilize DC from the intestine to lymph when administered systemically,32,39 and to increase interstitial DC mobilization from tissues such as heart and kidneys. It is not known whether equivalent products from gram-positive microbes, from viruses or from mycobacteria, might also affect in vivo DC migration, maturation, or survival, an issue currently under investigation. Teleologically, it would make sense that endogenous and exogenous products such as, for instance TNF-α and LPS behave as strong inducers of both migration and maturation of DC, as they are likely to be present at the early microenvironment during Ag exposure. DC somehow appear to ‘sense’ microenvironmental factors to finely regulate intracellular Ag processing. For instance, inflammatory stimuli are associated with the key shift in DC from an immature Ag-capturing to a mature Ag-presenting stage. This shift seems dependent on substantially increasing the expression, as well as the half-life (over 100 hr), of membrane MHC CII molecules. Both these latter phenomenon apparently result from decreasing degradation and endocytosis of MHC CII molecules.65,66 However, more recent information revealed that the relative low efficiency of immature DC to present Ag apparently results from a blockade in the intracellular peptide loading of MHC CII, a blockade released in a timely fashion upon DC exposure to inflammatory stimuli like TNF-α, LPS and CD40L.67 As DC mature in lymphoid tissues, MHC CII-Ag complexes – the TCR-ligands – will accumulate for much longer periods at the membrane, thus providing passenger low-frequency-specific T-cell clones more opportunities to scrutinize DC surfaces in search of Ags of interest. At the same time, DC accessory molecules such as CD40, CD81 and CD86 will be upregulated, both in vitro and in vivo.2,26,53 The counterpart is that in situ, human and murine peripheral immature DC lack costimulatory molecules such as CD80, CD86, and CD40,26,53 and MHC CII is mostly intracellular, somehow ‘sequestered’ avoiding premature futile Ag exposure, apparently reserved for better times within lymphoid tissues. Arguably, the increasing time of surface Ag exposure on maturing DC has been described as antigenic memory, as the previous antigenic experiences of DC now remain longer.65 It is likely that most of these Ags will be derived from the original site of Ag encounter, probably associated to inflammation (TNF, IL-1) and exposure to microbial products (LPS). While in most cases examined DC appear to run away from inflammatory stimuli, in the respiratory tract airways DCs are apparently the earliest cells recruited into the tracheal epithelium upon local Ag exposure, mirrored only by the PMN infiltration. Concomitant to this DC influx into the airway epithelium, there is an increased DC traffic into the draining lymph nodes. These experiments also showed nicely that simultaneous to these mobilizations, DC undergo morphological maturation.68 Interestingly, immature DC from liver also show an intriguing peculiarity; migration in opposite direction to blood flow, from the sinusoids to portal areas.6 It is known from early work that there is a ceaseless migration (‘steady-state flux’) of DCs through afferent lymph at a rate of about 104–5/hr, independent of exogenous Ag deposition,3 which is increased several times upon Ag exposure.2 However, as neither the amount of skin DC nor their efflux explain the total DC circulating in afferent lymph, other sources yet unknown are implied. Very recent work has elegantly shown that these constantly migrating immature DCs might be ferrying autologous apoptotic cells from the periphery into the draining lymph nodes, likely resulting from intestinal epithelium turnover in these experiments.69 This phenomenon was previously observed in vitro70 and for apoptotic epithelial vaginal cells in situ.71 A conceivable consequence is that self-Ags from apoptotic cells will be ferried and presented to lymphocytes by immature DCs, thus inducing tolerance instead of immunity against fraternal cells.72 A parallel thinking is that the latter mechanism will spare the reactivity to exogenous non-self, leaving it intact in a stand-by situation, ready for when this Ag encounter eventually occurs. There is scarce research about the intriguing fate of DC once they have inititated specific, acquired immune responses within lymphoid tissues. For instance, Ingulli has elegantly shown in vivo that, unlike Ag-unpulsed DC, Ag-loaded DC clustered many Ag-specific naïve T cells around but disappeared from DLN in about 48 hr.73 Calculating the average number of DC yield per DLN with respect to the number of DC arriving through afferent lymph per hour (around 104), it is assumed that most DC must die inside lymph nodes, unless they escape into the blood flow or become converted into another kind of cell. But there are no experimental proofs for either of these two last alternatives. Therefore, LN have been proposed as the final destination for DC. Intriguingly enough, however, there is no experimental evidence for this latter alternative either, besides the calculations between DC entering and exiting LNs. Certainly, revealing apoptotic cell death in vivo and in situ has not proven easy, even in locations where is deemed to occur extensively. It is still possible then that the inferred massive DC death occuring inside LN is not readily observed because of factors such as a narrow apoptotic window during Ag-induced reactions, or perhaps because another cell type efficiently performs the ‘corpse clearance’ function,74 as tingible body macrophages seem to do with apoptotic B cells within GC.75 Even in vitro there is little work concerning DC apoptosis, either on the potential signals to induce it (Fas, UV irradiation, Dexamethasone) or to counteract it [CD40-ligand, TNF-related activation-induced chemokine (TRANCE), FLICE-like inhibitory protien (cFLIPL), GM–CSF]. In vitro experiments suggest that whereas putative immature DC that have not interacted yet with T cells are more prone to apoptotis via Fas, DC that have seen the CD40-ligand of T cells become resistant.76 While this might prevent the killing of DC during the intimate synapse with strongly activated T cells, thus ensuring early T cell–DC interactions, it poses the question as to how T-cell activation subsides and when the DCs die. It has even been suggested that the inferred disposal of DC within lymphoid tissues after an immune response could be associated with the downregulation of T-cell responses,26,73,76 as the mechanisms to explain the latter (for instance ‘clonal downsizing’) are not well understood either.26 Despite the extensive research on DCs recently, there remain significant gaps to be addressed, especially within the in vivo context. For instance, the mechanisms whereby BM ensures a constant supply of DC progenitors, whether these mechanisms are exclusive or influenced by factors as diverse as, for instance, hormones or microbial products. There is little work on the ontogeny of DC in vivo, especially regarding the phenotype and the colonization of peripheral tissues, which may be relevant to a better planning of paediatric vaccination. Likewise, more understanding of DC mobilization from skin, intestine, airways and genital mucosa may help in the formulation of new and better prophylactic and therapeutic strategies. The involvement of DC both in extra- and intrafollicular reactions, as well as the putative DC death within lymphoid tissues, also merit careful attention. Another subject to be explored is the DC–nervous system interactions, whether for unique DC subsets or for potential controls upon DCs, as shown for epidermal LC.77 A crucial topic nowadays is certainly the relationship between the DC subsets described in vitro (two myeloid and one lymphoid) and their correlates in vivo and in situ, because, for instance, populations like epidermal LC seem elusive to categorize. Examining DC subsets in vivo along their maturation pathways may be an aid in identifying their precise molecular characterization and therefore be a device to enable better potential therapeutic strategies. Some previously unsuccesful prophylactic approaches perhaps should be re-examined in light of the emerging knowledge of the DC system. Finally, DC involvement in diseases, whether from developed (cancer, atherosclerosis, etc.) or developing (old, new, and re-emerging infections) countries, is a completely open field, where, as recently discussed,78,79 immunology must lend itself to understand immunity. I would like to acknowledge numerous colleagues who contributed to this work in different ways, especially J. Calderón, M. Heras, A. Escobar, V. Shreedhar, M. Kripke, J. Banchereau and R. Steinman. This is to honour the memory of Professor C. Flores-Edeza, who dedicated his life to advance education. Supported by Conacyt 30757M.