Update July 2008
In the Chapter on Lithium and non-psychiatric diseases, there was presented an hypothesis, proposed in 1968, suggesting an active immune tolerance mechanism. As envisaged, it would occur within the thymic cortex (in particular) and possibly in the secondary lymphoid tissue, to a lesser degree. For want of any other term, “antigen factors” was suggested for those factors presumed to be multiplied in the proliferating cortical thymocytes. They would be released upon the latter’s dissolution and to be taken up by the thymocyte progenitors for amplification, consequent upon cell proliferation. This simplistic approach was the best that could be proposed in 1968. More recent studies of immune processes characterize the extensive thymocyte dissolution as apoptosis from “neglect” and negative selection. Perhaps these processes may not be as simple as now proposed, and the nuclear or other products may still be put to more use than as recycled small molecule waste. Here, there will be a more detailed examination of the old hypothesis within the strictures of current knowledge and, in particular, those proposed vague and mysterious “antigenic factors.”
For various reasons, interest in the topic has waned for the moment, so that what follows are notes relevant to the proposed topic. Whilst no clear evidence for such a proposed amplification process has been identified (although such a mechanism may still be possible) there may be an amplification process using the dispersal of high affinity TCRs amongst early thymocytes destined to be Tregs, so that the latter end up with about half the numbers compared to the thymocytes destined for an active immune role. The overall conclusion reached was that further progress in the topic must await more experimental data, particularly with respect to the Thymic Nurse Cells which, from the present study, seem likely to play a far more important role than may be generally regarded. In the current format, perhaps these notes may be of interest for students of immunology.
Predictions from 1968-1970
In the published version of the hypothesis in 1970, there were two predictions :
· “. . .that competition for tolerance by antigens may be found.” This prediction applied particularly to the thymus, but with lesser involvement peripherally. Whilst not involving the thymus, the work comparing the immunogenic responses to guinea pig-derived MBP against those by rat-derived MBP (at least) would seem to confirm this prediction in this context.
· “. . . that thymic grafts, enclosed in cell-impermeable chambers, and introduced into patients with neoplastic disease immediately after anti-cancer treatment, may aid tumour rejection.” Whilst the proposed grafts may not have been performed as described, there is ample evidence in the literature that elimination or inhibition of thymus-derived regulatory T cells (Treg, Tr; at least) can combat tumours (see Chapter on Lithium and non-psychiatric diseases).
Whilst tolerance may involve the effector cells and their origins, such as B cells and NK cells, this discourse will be concerned with thymic function and how it may initiate tolerance and may be identified by responses involving the periphery.
Current Tolerance Concepts
Recent historical documentation of the features of tolerance to “self” includes, in general, the failure of lymphocytes to respond to an antigen, and this may be effected by deletion (clonal abortion; negative selection; apoptosis), clonal anergy, editing or immunological ignorance,,. (There may be some confusion as to what the term “tolerance” means. From a clinical aspect, it would seem to represent a failure of an organism to respond, at a grossly observable and disease-related level, to an antigen, however the antigen may be presented. In some contexts, the ability of a cell to recognize an antigen and yet not actively respond to it is not considered “tolerance,” which confuses the issue. In this discussion, “tolerance” will mean the clinical use.) Whilst there has been much work and discussion regarding tolerance without deletion (ie without negative selection) and the factors that influence it, the biochemical “switches” responsible have been elusive, apparently not involving the TCR repertoires, which are uninfluenced by the gene for Foxp3.
“Inefficient antigen recognition” is recognized to be the general common influence leading to peripheral immune tolerance, as when there is MHC-peptide:TCR signalling without the costimulatory factor(s). This requires non-involvement by CD4 and other co-stimulatory molecules, such as B7 (B7.1>B7.2). This is associated with prolonged expression of APC-T-cell union or activity. Signalling occurs, but with differential information transduction; low B7.1 expression being associated with tolerogenic potential. The model predicts that the thymus sets the responsiveness of the T cells released into the periphery, and that this setting is maintained in the periphery.
There are now recognized a number of ways in which MHC class I & II molecules and fragments of membrane, including receptors, may be transferred between cells,. One mechanism for tolerance conversion involves the transfer, by trogocytosis, of HLA-G1, an immunosuppressive surface marker. This does require cell-to-cell contact, but not the traditional receptors; only becoming effective on activated (eg CD4+, CD8+ & NK) cells, so that its function in adult vivo is probably to switch off excessive immune activation, cytolytic activity and cell proliferation rapidly, but only briefly (< ~24 h). Since the acquired expression can be used by the recipient for transfer to other cells, the tolerance attribute can be transmitted, but with dilution and without amplification, and relies on its surface location. As such, it is not immediately applicable to the thymus. However, it remains of interest, because it must utilize a signalling pathway with the surface HLA-G1(?ILT2) most upstream, even in cells that do not otherwise express HLA-G1, leaving the possibility that other signals may tap into the pathway further downstream. The speed and ephemeral nature of the suppression effects argue against inhibition at the transcription level. Immune cells are also susceptible to immune-suppression induced by peptide fragments derived from DNA viruses, such as those related to LMP1 produced by EBV,. These 6-17 aa fragments are plasma membrane-associated, where they are independent of MHC molecules and more active when acetylated and hydrophobic. They are secreted on exosomes and probably associate with clusters of G-proteins, PKC and adenylate cyclase, up-regulate IL-10 and down-regulate IL-2, and may be involved in the signalling with NF-κB.
The current understanding of the early basic thymic process involves inhibited apoptosis, cell fate determination and tolerance through Notch signalling,,-associated proliferation of waves of progenitor T cells proceeding via prothymocytes to double positive thymocytes within the machinery provided by the thymic epithelial cells. They then reach a stage when T cell receptor (TCR) molecules may be expressed with various affinities,,,, to peptides attached to APCs in association with MHC molecules, including some non-specificity by “degeneration,” , with various pathways :
No TCR Apoptosis (dissolution) by “neglect” in the Cortex
TCRhigh Negative selection (ie elimination of potentially self-harmful cells by apoptosis [dissolution], generally in the medulla/cortico-medullary region), an ill-defined process that seems to involve promiscuous gene expression of most antigens of the body (chiefly) by the medullary thymic epithelium, presented either by them or more effectively, by marrow-derived APC’s (dendritic cells).
TCRweak/low Positive selection (signal to proceed) proliferation (amplification) and release. The specific antigen does not interfere with the responses of mature lymphocytes to other agonists. This process may involve the Calcineurin/NFAT pathway signalling, providing an unique hypersensitivity to ERK in the early part of the thymocyte maturation pathway. The weakest signals favour CD8, the stronger ones include CD4, probably associated with reduced Notch-dependent calcium responses, with the co-receptors CD4 and CD8 sequestering Lck to bring about MHC restriction selectivity.
The cells that have been positively selected, yet not deleted by negative selection, are released to the periphery, where they respond to affinities at levels above the level required for positive selection, provided the density is adequate (affinity & density); the positive selection being based upon (affinity & density) rather than the MHC/peptide:TCR exposure, or contact time, and the negative selection, likewise, upon tissue-specific affinities, rather than site, together with co-factors, accessibility within the thymus, thymocyte maturation and the number of TCR types per cell. There is the possibility that CD30 and its ligand may play a part in thymic negative selection. However, this does little to explain how tolerance may be preset in those cells that are released from the thymus. The enigma of tolerance maintenance in the periphery may involve immature dendritic cells, possibly having special nodal forms to process the migratory DCs carrying the products of apoptotic cells. Again, this does not address how the DCs made, or acquire, tolerance (other than by deletion).
The overall thymic demarcations can be shown :
THYMUS STRUCTURAL DIFFERENTIATION
Double positive (DP; CD4+8+) mostly
95% eliminated ultimately
? multiple costimulatory factors for negative selection
Single positive (SP; CD4+ & CD8+) mostly
Most apoptosis, involving BM-derived APCs
HSA+ (~70%) = sensitive to negative selection
But there are anomalies in that :
i. Self-MHC recognition augments T cell responses, with the TCRs of Tregs having a greater affinity and,
ii. Thymocytes with “self-recognizing” TCRs with strong avidity escape negative selection, to provide ~50% levels of those with that specificity as Treg in lymph nodes and,
iii. Low numbers of T cells, when presented with “self” peptide, proliferate to clonal positive selection, instead of negative selection and,
iv. Thymocytes proliferating in response to positive selection of a peptide, subsequently become “desensitized” to that peptide and,
v. Thymocytes with weak self-antigen recognition are released to the periphery in a tolerant state
vi. Tregs from the thymus have TCRs with similar repertoires to Tregs in the periphery, with the gene for Foxp3 non-contributory to the TCR repertoire, and
vii. Bone marrow-derived APCs have only a finite life span, meaning that tolerance attributes, if involving APCs, need to “load” the bone marrow-derived cells, either in the bone marrow, or in the thymus and,
viii. Whilst most (~95%) cortical DP thymocytes are destined to “apoptosis,” the latter occurs mainly in the medulla or cortico-medullary regions.
ix. The tolerance to C-reactive protein (which was not expressed at detectable levels by cortical epithelial cells) seemed determined by intra-thymic antigen epitope hierarchy rather than circulating levels, and there seemed some negative selection at the DN stage, being more pronounced at the SP stage. The DN deletions would seem likely to be in the cortex, making explanation difficult. (See later, with respect to CEA & C5 [Żal].)
Whilst deletions occur before, during and after the DP stage, when concentrating on the latter, the proposal put forward by Punt et al. is that DP thymocytes in vitro may be subject to two main types of deletion :
a. A CD28-dependent pathway, where thymocytes with both TCR and concomitant CD28 costimulatory signals are eliminated in the medulla, in association with APC &/or epithelial cells, which have B7.1. This pathway is independent of PI3-kinase and cycloheximide inhibition but involves PKCγ and caspases.
b. A CD28-independent pathway for those thymocytes with an high affinity TCR-self Ag in the cortex would provide the thymocytes with a “death tag,” which marks them out for deletion when they come upon APCs &/or macrophages in the cortico-medullary region (Notch and AIRE are not considered).
Studies from human thymic tissue (without culture) showed APCs (dendritic cells or macrophages) in the medulla producing one or more self antigens and often having a rosette of TUNEL positive (apoptotic) thymocytes attached. This type of self-antigen expression was observed at a lower level in spleen and lymph nodes.
Thymic Nurse Cells; Nursing thymic epithelial cells
A feature of the thymic cortex is the presence of thymic nurse cells,(“TNC”). These have received limited attention, with researchers falling into two camps, a small number who regard TNCs as very important, believing them to have something to do with positive and negative selection, although the processes are unknown, and those who seem to consider them some quaint aberration that does not fit in with accepted ideas on selection, and are therefore best disregarded. There have been doubts about their homogeneity, inter-study differences as well as the relationship of the in vitro studies with the in vivo existence and function. However, the general features seem to be as follows. They are found principally in the subcapsular and perivascular regions of the thymic outer cortex, as confirmed by a rat, anti-mouse antibody, ph91, although there is the view that they are heterogeneous, with smaller forms subcapsular and medullary, with a tendency to appear during isolation. They are apparently playing a role in thymocyte binding and maturation (including mitosis) after the first proliferation phase,, before positive selection from αbTCRlo CD4lo &/or CD8lo onwards, and maintaining the health of the cortical epithelial cells. They contain many vacuoles with osmiophilic material within (possibly membrane remnants). A rat strain of cells seem to be able express MHC class I and, when stimulated with IFN-γ, MHC II. With these and ICAM-1 molecules some thymocytes can attach to their surface, probably involving LFA-1, CD18, CD2, Thy-1, CD4 and CD8, the blocking of these inhibited binding ~50%, although CD4, CD8, CD45, CD11a/CD18, CD11b/CD18, CD25 and CD5 did not appear involved using a mouse SV40 transformed cell line. CD4 and CD8 seem weak factors in binding (if at all). The method of cell-cell binding is unclear, but may possibly involve ph91 and also Mls antigens, which are carried by the nursing TEC and can engage when in mixed lymphocyte reactions involving MHC Class II molecules that are essential for T cell development, leading to apoptosis. Foetal mouse thymocytes, induced into apoptosis by 6-8 h of Dexamethasone (artificial) became attached to human TNCs, with hSR-B1 seemingly playing an important (but not the only) role. A form of immune synapse seems involved. The TNC then can “engulf” the thymocytes, the latter showing directional motility into channels, then vacuoles (~60% inhibited by anti-CD18 antibody), holding 15 – 50 thymocytes. “Engulfment” would seem an active process for both thymocytes and the TNC. Macrophages seem able to insinuate themselves under the TNC surface ruffles by using a different mechanism, and meander amongst the thymocytes caressing them with cytoplasmic ruffles, causing less actin/tubulin organization in the TNC. The TNC-thymocyte contacts appear to involve an immune synapse-like organization of both actin and microtubules. (In the chick, at least, the TNCs seem to have the endosomal-lysosomal machinery to process peptides for MHC II presentation. There may also be heterogeneity of responding TNC thymocytes in GVH reactions on chorioallantoic membranes,, with the intra-TNC lymphocytes showing a higher level of maturation compared to those in the surrounds, considered to be positive selection.) Views differ, but some of the thymocytes “engulfed” are protected from induced apoptosis (including cortisone-induced), some 80% undergo physiological apoptosis and many (in the rat) maturate from immature CD4- CD8+ to CD4+ CD8+, then (in mice) from αbTCRloèintèhi, with CD69- to CD69+ (if IL-1b is present) and Bcl-2 up-regulated. These changes can occur on the attached surface thymocytes. Overall, only ~8% of DP thymocytes that encounter TNCs remain viable and proliferation of thymocytes within the TNCs occurs, but is not appreciable, despite tritiated thymidine uptake in many. When anti-CD3 was administered to mice to induce rapid apoptosis of thymocytes, some ~95% of ingested thymocytes at ~16 h were apoptotic, and the apparent number of TNCs had risen. These observations suggest that many TNCs are “inactive” and in reserve, becoming active and apparent when there is a glut of apoptotic thymocytes; a finding in accord with the seemingly mixed antibody binding results with CDR1. Also, the TNC number seemed roughly related to the thymocyte number or mass of the thymus. Trypanosoma cruzi is believed to produce a trans-sialidase which mobilizes the sialyl groups from the plasma membrane. This enzyme results in thymocyte depletion from the cortex, affecting DP cells onwards by inducing apoptosis, which is identified within the TNC complexes; an enforced and inappropriate negative selection. Sadly, mechanisms were not examined in greater detail, however, from all this we may conclude that the TNCs comprise a very influential processing site in the cortex which, when perturbed, can result in appreciable cell depletion; the inference being, that most cortical thymocytes probably enter and pass through these units in normal physiology. Using the artificial conditions of dexamethasone-induced thymocyte apoptosis, attachment of these cells to TNCs involved both phosphatidylserine receptors (PSR) and scavenger receptor B1 (SR-B1), with TNF-α up-regulating the expression of PSR, and TGF-b SR-B1. This study presumes that apoptosis starts before the thymocytes reach the TNC, and does not consider that viable cells may be able to enter the sinus/vacuolar system before the stimulus’ effects for apoptosis becomes obvious. (Since Lithium is believed to cause production of TNF-α, the involvement of TNCs in the Lithium-induced thymocyte deletion may be partly explained; the TNC involvement with DP thymocytes being after and before Lithium has direct or co-stimulatory actions on DN and SP thymocytes respectively.) Even with both TNF-α and TGF-b, there was only ~50% of thymocyte binding to TNCs, indicating that there may be a reasonable percentage of thymocytes not destined for deletion by the TNCs. The involvement of TGF-b, generally an immune suppressant, is noteworthy. Work with a cell line, considered representative of TNCc, indicated that the TNC can attach and internalize apoptotic thymocytes (induced by dexamethasone) preferentially (even dead cells), yet can stimulate others that have been internalized to proliferate. Whether the thymocytes were internalized in true vacuoles or channels was not evaluated.
A likely function for the TNC’s is as a site for selection and maturation of those DP thymocytes that respond to the “self” antigens found promiscuously expressed in the thymic cortex. They can be protected and held until they mature past the DP stage, when they can proceed to the thymic medulla, immune to the deletion process of negative selection because they can resist the apoptosis processes. They then mature to thymic Tregs, with the expression of the relatively high affinity TCR approximately 50% of the non-Treg lines. This raises the possibility that the cells with high affinity TCRs lose and/or donate the TCR to those with lesser amounts – a form of amplification of TCR-equipped thymocytes. Those that are not processed by the TNCs, perhaps because all the potential TNC niches are filled, are deleted by negative selection, or if the TCR affinity is less, released as CD4+ CD25-. This may explain the positive selection claimed for the TNCs, but it does little to explain the claimed negative selection, unless this is for the TCR transfers.
Since apoptotic cells tend to round-up and lose adhesion, once the thymocytes have entered the intimately secluded location within the TNCs, the latter may provide an environment where “antigenic factors” (such as plasma membrane fragments with specific TCRs and MHC class I &/or II and antigenic peptides) may be released from apoptotic thymocytes in a very controlled environment that does not allow the cells or fragments to float away into the mouths of the macrophages, until appropriate. This could allow a transfer process to be completed, moving “antigenic factors” from the apoptotic cells to those more immature that are “protected” from apoptosis, and may enjoy limited proliferation, producing an amplification stage.
The major difference then, between central cell processing in the thymus and the peripheral processing in lymph nodes and spleen, is that, in the thymus, there is a very large proportion of cells produced with no or minimal TCR avidity subjected to deletion by a process that is not yet clearly characterized (whether typical apoptosis or some variant, that is not clearly self-antigen dependent).
So, the major turnover in the thymus seems to involve the deletion of the DP thymocytes of the cortex, specified by no or minimal TCR affinity, thereby subject to “neglect” and deletion. Since any TCR affinity for this group is below a survival threshold, the sheer volume of the thymocyte turnover could have this whole group of cells contributing some products towards an active thymus process.
The events can be illustrated in flow diagrams :
If the “antigen factor(s)” (F) released from the strong epitope-expressing cells highlight the developing specific epitopes (E) during the double positive stage, such that the F:E ratio is above some critical level, the cells with the strong F-tagged specific epitopes of a moderate concentration may be selectively labelled for destruction, releasing more of the specific F. The resulting specific tagging of lesser specific E levels will then lower the threshold level for the critical destruction-inducing TCR concentration, thereby specifically accentuating the removal of “self”-sensitive cells by negative selection. If the F:E ratio is high and the combined component level falls below a critical level, the cells so labelled can escape negative selection and be specifically marked by F for tolerance (Treg), eg :
Epitopehigh + F = negative selection & production of much F
Epitopemod – = positive selection only
Epitopemod + F = negative selection with smaller F production
Epitopelow – = positive selection only
Epitopelow + F = positive selection + tolerance è Treg
The appreciable cellular turnover was known in 1968. Today, the trigger for most of this cell destruction is known. However, there are still cells with weak-moderate TCR molecules which, for the most part, do not initiate auto-immune diseases, and this needs explaining. The issues are, then, whether the substantial proliferation and destruction of thymocytes through negative selection represents a dead-end branch, producing non-specific molecular waste which re-enters the metabolic pathways, or whether some “antigen factors,” as representing the destroyed TCRhigh cells is, in some way, processed back in to the early level of a proliferation stage, in order to amplify, convey and/or convert the thymocytes that are proceeding towards release, to be (more) tolerant with regard to those “antigenic factors.” Possible contenders for such a feedback could be:
· DNA fragments, as perhaps with CpG motif-rich oligonucleotides, which may, instead of stimulating immunity, block immunity, particularly at the promoter level,
· Histones, that may bind DNA and modify gene expression,
· RNA fragments. These could be incorporated into ribosomes to produce peptides, or they could act as interfering RNA, blocking specific RNA translation,
· Ribosomal protein, that can bind and modify RNA transcription,
and possibly others, as from carbohydrates and lipids.
· Reasonably intact proteins, such as antigen/antibody complexes, gene promoters etc.
For tissue remodelling, the removal, by apoptosis, of unwanted cells without initiating any inflammatory responses, is a necessity. Such staged clearance is usually associated with anti-inflammatory signals,, such as TGF-b, IL-10, PGE2, MIP-2, NO, keratinocyte-derived chemokine and, on occasions, the reinforcement of tolerance. Necrosis, on the other hand, provokes inflammation signals and responses,. Cells undergoing apoptosis tend to inhibit the maturation of dendritic cells, unless co-stimulated by necrotic cells or their products, as may occur if there is delayed clearance of an excess of apoptotic cells. However, these observations shed little light upon the basic mechanisms involved in the apoptosis immune suppression. Mature peripheral T cells stimulated by the Superantigen Staphylococcus enterotoxin A normally undergo deletion (apoptosis, producing tolerance), but a small proportion of the group of CD4 cells affected may be protected by a different antigen that binds to the same variable b epitope, indicating a competition for this type of tolerance induction (see prediction from 1968-70, above), with the nature of the differing receptor signals not being elucidated. This tolerance by deletion may be relevant to thymic negative selection. An infectious-like tolerance involving induced inhibition can be transferred with antigen specificity when the antigen is presented by an APC to CD4 cells with Serrate1 the ligand and Notch1 the receptor on the CD4 cell. This induces a 70-90% reduction in immune response. The antigen/Serrate1 combination also suppresses responses to nearby “linked,” epitopes on the antigen, and can be transferred by CD4 T cells to naïve recipients. It is reasonably long lasting. Whilst the underlying mechanism of this process is unknown at present, it may be relevant to an active tolerance induction involving “antigen factors” in the thymus, where Notch signalling is known. (ie an [activated TCR + “antigen factor”] complex, derived from one or both deletion processes, could bind to developing thymocytes such that, with a Serrate1 signal, the complex may induce tolerance in a “linked” fashion in the thymocyte, which would then negotiate [possibly] the positive selection and the negative selection process, before release to the periphery.)
Whilst caspase-activated DNAse (CAD) is the enzyme involved in the classical apoptosis DNA fragmentation, other mechanisms can be involved. This has been shown in mice with CAD/ICAD mutations, with varying degrees of DNA degeneration. Phagocytes degrade ingested DNA in lysosomes with DNAse II, so there appears to be a two-step process. In life, the latter step seems the major clearance mechanism, the former being dispensable, with the macrophages being able to sense early signs of cell death and dispatch the cell.
Caspases Break down many cell structures and molecules and trigger CAD
CAD Degrades DNA to ~50 kDa fragments by attacking nuclear scaffold regions. These fragments are then broken down to nucleosomal units. It and ICAD are found abundantly in the lymphocytes of the spleen and thymus
DNAse II Lysosomal enzyme in macrophages
To simply describe physiological cell death by the all-embracing term “apoptosis” ignores the variations that are possible in particular situations. Of particular note is the absence of MFG-E8 from thymic macrophages, which would point to an inability of the macrophages to engulf dying cells and, thereby presumably, insulate the cellular components from the destructive macrophage enzyme DNAse II. The Mer receptor may be a replacement in the thymus; deficiency resulting in a failure of macrophages to engulf cells made apoptotic from Dexamethasone application (artificial). Negative selection (dissolution; normally sparse) in the thymus seems to occur in the medulla, involving Mac-3+ macrophages, (Mac-3 probably decorating both surface and endosomic elements). These macrophages in lung and spleen (at least) express IFNAR1-dependent iNOS, and contrast with thymocyte dissolution in the cortex, which is prominent and seems to be related to the lack of MHC/TCR molecules and “neglect,” resulting in phagocytosis by F4/80+ macrophages (cells that may present “self” epitopes). In the human thymic cortex, phagocytes resembling macrophages and immature dendritic cells are associated with, and take up apoptotic (physiological) thymocytes; phagocytes that may present antigens and may be human equivalents of the murine F4/80 macrophages. These forms of apoptosis seem different.
When Caenorhabdilis elegans is allowed to permit its cells to follow endogenous apoptosis (without physical or chemical inducements) the apoptosis process is augmented by influence from the engulfing cells; when the latter are defective, the cell destined for apoptosis can waver in its commitment. The mechanism seemed obscure. If this is extrapolated to mammals, there would be expected interplay between the cells destined for apoptosis and the macrophages or dendritic cells, and not simply the mopping-up of cellular debris.
One of the early indicators of apoptosis is the formation of blebs on the plasma membrane, probably involving breakdown of sub-membrane fodrin filaments. The exterior then exposes phosphatidylserine, which establishes a procoagulant cell surface. The blebs contain DNA as chromatin and nucleosomal fragments. Autoantibodies to dsDNA add a significant specificity to the more general antibody binding to chromatin. For specific B-cells, the antibody-binding to blebs can supply the stimulus for clonal expansion, which may become more specific against the dsDNA epitopes.
Rapid destruction of DNA and RNA causes the release of the purine derivative uric acid (urate). This has been shown to be an immune adjuvant, particularly if crystalline, which may be more likely in the cytosol where the concentration is much higher. For this reason, rapid apoptosis may have appreciably different effects than slow apoptosis (allowing control of the cytosolic uric acid concentration); the latter may have less adjuvant effect.
Cell breakdown products
These can be expected to be taken up by phagocytes and/or dendritic cells by phagocytosis (for the particulate debris) with pinocytosis and lysosomal receptor-mediated uptake for soluble material, from which antigens may be detected,. Macrophages show the greatest phagocytic activity, being able to deprive nearby immature dendritic cells of dissolution products, and downgrade any inflammatory responses involving dendritic cells, seemingly by production of IL-10 and TGFb (and possibly others). Immature dendritic cells, however, are the only phagocytic cells capable of cellular engulfment and subsequent maturation for cross presentation to the CTLs, using MHC class I molecules. Of particular interest is the finding that some EGF:ferritin complexes ended up in the rough endoplasmic reticulum (which may be able to incorporate the pinocytosed or endocytosed RNA) and that phagocytosed material (as from cells undergoing apoptosis) and soluble antigens linked to endosome directed receptors are processed through a weakly acidic phase, with the phagosome contents being subjected to the more strongly acidic environment for dissolution subsequently. The former, weak acid environment, would allow sensitive molecules to be separated and moved into the cytoplasm (perhaps similar to the results of the gene gun for naked DNA, and PLGA:DNA complex microspheres to insert genes for peptide production, the microspheres not reaching the late endosomes or lysosomes). Bacterial DNA inserted into the cytoplasm induces a set of genes less inflammatory than those induced via phagosome vesicles, by an IRF3-dependent pathway that can involve NOD2 for amplification. Whilst interest has been with the MHC class I pathway, the initial phase may be a common stage for the process proposed by the “antigen factors” influence for tolerance. Chromatin, probably as nucleosomes released by intact, but ~naturally degenerating lymphocytes in ~physiological culture, may selectively bind to receptors on activated monocytes and act as antigens (at least). Thus, there are mechanisms described which may result in the introduction of both particular and/or soluble breakdown products (“factors”) into the phagocytes, monocytes or dendritic cells, which could provide for the transfer of genetic material which, in turn could result in the production of transgenic peptides that may influence the responses and products of the recipient cells.
The engulfment of thymocytes induced into apoptosis using corticosteroid treatment (artificial) by macrophages reveals complexities involving the non-active state of the macrophages or the active state. Non-active macrophages recognize exposed phosphatidylserine, but are inhibited by vitronectin receptors. The activated macrophages (peritoneal, following thioglycollate) seem able to detect molecules presented earlier, indicating a progression of events for cell surface ligands. The activity status of normally functioning macrophages in the thymus is not clear, but is probably “not active.” The thymus may be able to clear supra-physiological thymocyte apoptosis using the macrophage Mertk (Mer) receptor, but this pathway does not seem to be required in normal physiology.
The fate of DNA and RNA is important. Generally, the non-methylated oligonucleotides (especially with many CpG motifs, as found in mammalian promoter regions) are immunogenic for DCs when coupled with an antibody, being processed via the Toll receptor TLR-9 in association with the endosomes,,. This indicates an antigenic-specific selection. However, those oligonucleotides that are methylated or otherwise modified (eg [CpG] S-ODN 2088), may become tolerogenic, indicating a potential role for specifically-linked chromatin fragments to induce an amplification process for tolerance generation.
There are a number of ways that cells can take up debris, particulate antigenic material and soluble antigens. Phagocytic, engulfment processes will be dealt-with elsewhere. Soluble antigens typically attach either to cell receptors or to MHC I molecules attached to cells, such as dendritic cells and lymphocytes, or attached to exosomes produced by macrophages and dendritic cells. Once attached to the receptor concerned, there are a number of pathways that generally lead to the endosome/lysosome vesicles and network, with considerable debate over the factors which influence the pathway choice. The binding and internalization of oligodeoxynucleotides by Mac-1+ phagocytes is stimulated by TNFα, arachidonic acid and IL-8. The process seems to involve vesicles. Those materials that reach the lysosomes are destined for proteolytic destruction. Those that associate with the MVBs, may become attached to the walls of the vesicles (exosomes-to-be) for exteriorization via a number of possible routes.
The APC-Effector Dynamics
Peripheral immature dendritic cells are able to take up and process debris for antigenic peptide presentation by MHC class II molecules, being the breakdown products from both apoptotic and necrotic cells. These can then mobilize and move to lymph nodes when they, in turn, undergo apoptosis and transfer the original peptides to nodal dendritic cells. This appears applicable for both tolerance and immune activation, the tolerance status perhaps decided by the nodal environment, with DCs + IL-2 better able to amplify specific Tregs than macrophages, and the maintenance of Foxp3+ status needing only DCs. Human blood CD4+ CD25- T cells can be stimulated by TCR/CD28 to express Foxp3 and become Tregs. The exposure of CD4+ CD25- cells in peripheral lymphoid tissue, together with an antigen, CD28 stimulation and TGF-b results in the induction of Foxp3, the down-regulation of Smac7 with the adoption of Treg behaviour and the production of more TGF-b, to reinforce tolerance.
The basic features of MHC class I APC-CD8+ CTL signalling are presented by Huang et al. 1999; describing and illustrating the transfer of the MHC + antigenic peptide into the endosomic/lysosomic compartments of the sensitized effector CTL; the TCRs of the CTL disappearing within minutes of contact. Dendritic cell processing of immune complexes involving the MHC class I & II-restricted pathways were described by Regnault et al., 1999, and an endosomic/lysosomic tubular secretion process to the plasma membrane in mature dendritic cells has been demonstrated,. Examination of the APC-effector cells relationship showed that, provided there was a bond between the cells, as by an antigen-independent B7-CD28+ union, various molecules associated with the MHC region can be transferred to the effector cell and internalized. Such internalized proteins and peptides can be subjected to proteasome digestion or recycled. The ease of this was related to the expression of CD28+, with resting CD4+ > CD8+, and (activated cells) >> (resting forms) and provides a potential pathway for peptides to participate in an antigen diluting cell-cell transmission sequence. Similarly, for MHC class II T:APCs, activation of the recipient effector through the its TCR results in transfer and incorporation of MHC/peptide and bystander molecules, probably by means of vesicles, the MHC/peptide molecules then to be found on/in the recipients’ plasma membranes; a process that may be involved in tolerance transfer, as by an inefficient signal recognition. Thus, there needs to be an amplification of the antigenic factor in order to provide an infectious tolerance.
The plasma membrane
Trogocytosis can transfer part of the plasma membrane, chiefly of haematopoietic cells, with varying ease (carcinoma cells do so only poorly), the process being antigen independent, and retaining signalling capabilities,:
The phospholipids of the plasma membrane, such as phosphatidylserine are important in revealing early apoptosis to activated macrophages, the latter recognizing the oxidized phospholipids that are believed to be produced by membrane peroxidation, as well as other factors. (The study used thymocytes induced to apoptosis by dexamethasone, with peritoneal macrophages induced by thioglycollate; both artificial influences.) The similarity with the in vivo thymic cortex, with its DP thymocytes and macrophages is unclear, although at some point in the cell destruction, oxidized phospholipids exposed at the plasma membrane seems likely.
Factors such as IL-1b and HSP70 may be secreted by a process involving the endoplasmic reticulum and the endolysosomal organelles, with release to the exterior. Whilst extracellular HSP70, a chaperone protein that can bind and carry proteins and peptides, is generally pro-inflammatory,, requiring CD40 and CD4+ for an adjuvant-like role, it can be immuno-suppressive in chronic conditions and suppress apoptosis. The immune-promoting effect of necrosis over apoptosis is attributed (in part) to the release of HSP molecules.
Splenic DCs (at least) can be divided into two main classes, those that are CD8α+ and those that are CD-. The former class present FasL to the CD4 lymphocytes on immune contact and stimulation, resulting in an appreciable loss through apoptosis. Elaboration of this form of peripheral negative selection showed that macrophages could also destroy specific T cells in using FasL-Fas interaction both in vitro and in vivo, but the factors influencing the APC’s selection pathway between immune enhancement, anergy or clonal deletion are not identified.
The thymic medulla has Hassall’s corpuscles, the epithelial cells that express TSLP. This can stimulate the surrounding dendritic cells to maturity (LAMP+) which, in turn are able to transform the CD4+ CD8- CD25- thymocytes to CD25+ Treg status (Foxp3+), and stimulate them to proliferate. This requires medium-high affinity, antigen-specific CD4+ SP thymocytes to have survived the “neglect” in the cortex, and also the negative selection in the medulla, considered to occur in the cortico-medullary region by LAMP- immature dendritic cells. The maturation of the CD4+ CD25hi in mice indicates that these cells are positively selected, based upon their specific TCR affinities. They then migrate to the thymic medulla, in response to chemokines influenced by Aire (its expression stimulated by thymocytes, which may occur in TNCs also), with the epithelial component being maintained by lymphotoxinbR, at least . Those precursors that go on to become Treg show some selection differences, attributed to the TCR affinities being just below the threshold for negative selection. Those destined to Treg status may then be stimulated by IL-2 (or IL15) plus an unknown factor, via STAT5, perhaps by paracrine release of IL-2 upon the apoptosis of others in the cohort, being those with lesser affinities. This results in the up-regulation of Foxp3 and the Treg phenotype. Those thymocytes destined to become Treg have to “run the gauntlet,” and require a special pass ticket directing them to the central location of the Hassall’s corpuscles, and allowing safe passage through the negative selection traps.
Peripheral tolerance, involving CD4 T cells of an anergic/suppressive type (Treg-P) seems to involve cell surface and active TGF-b. When TGF-b is present with a specific TCR (or agonist anti-CD3 antibody) and either an agonist anti-CD28 or APCs, applied to naïve CD4 CD25- T cells, the latter become CD25+, express Foxp3, are anergic in vitro, but undergo proliferation and are suppressive in vivo for T cells sensitized with the same antigen peptide (no specificity for cell origin) and, in turn, express TGF-b. Accordingly, they resemble the “professional” T regulatory cells of thymic origin (Treg-T).
TGF-b has become a significant factor in the immune tolerance associated with tumours. In rat and mice models, tumours can produce a (currently unknown) factor that “licenses” immature DCs (CD11c+ CD11b+ co-factorslow) to produce TGF-b. This, together with MHC Class II + peptide specifically activates the proliferation of Treg. These suppress the immune assault on the tumour. In addition, the presence of TGF-b on the membrane of Tregs may be important in suppressing the anti-tumour actions of NK cells (including NK cell proliferation). Tregs can induce suppression of the NK cell expression of NKG2D, a costimulatory factor for NK anti-tumour attack. (Bolus Lithium in a human patient with lung cancer resulted in the CD56+ subtype count [total lymphocyte count] in her peripheral blood to rise from 0.21[1.8] x109/L to 0.25 [1.66] x109/L on day 6 of treatment, later falling to 0.23 [1.4, CD8 depleted > CD4] x109/L on day 38. These changes may indicate that Lithium, in the protocol used, may suppress the Tregs associated with tumours, as is also claimed for low dose “metronomic” Cyclophosphamide of 50 mg twice a day for a week repeated after another week. The authors noted no significant changes in the total lymphocyte counts. The latter observations were on terminal patients with very basic assessments of the tumour responses. Whilst the cyclophosphamide protocol may not be optimal, on current assessment, bolus Lithium may be a safer form of treatment to improve anticancer effects.)
AIRE seems to participate in tolerance induction, particularly involving maintenance of the thymic medullary epithelium in adults by RANKL and CD40L from CD4+ αbTCRhi thymocytes. How it operates is not clear, but it is expressed by medullary epithelial cells, DCs and also DP thymocytes within the thymus.
Dendritic cell exocytosis and vesicles
Dendritic cells discharge vesicles (exocytosis) apparently collected in multi-vesicular bodies, in order to direct recipient effector lymphocytes. Their size is variable – diameters ~40-90 nm. Most are found in immature dendritic cells (~5-8/cell), the number falling with maturation (~1-2/cell). Maturation induction reduces exosome production by ~70%, possibly related to endocytosis changes. These endosomes may also be produced by “pinching off” the plasma membrane (which has multiple villous-like “processes” with slightly bulbous ends), and signalling may occur weakly without cell-to-cell contact. When dendritic cells are disrupted by sonic treatment, there are produced large quantities of vesicle-like particles. With peptide, adhesion molecules and co-stimulatory molecules, those from mature dendritic cells are highly immunogenic and may not require viable APCs (contrast with CD4+ signalling) although the effects are more transitory than the viable APC/endogenous peptide induction, possibly due to stability (or other factors, perhaps some within the vesicles). Those from immature dendritic cells are much less immunogenic (or more tolerogenic ?). Thus, if associated with peptide (ie peptide outside the vesicles), effector cells (CD8+) can be activated, provided that some co-stimulatory and adhesion factors are present (CD28/B7 & LFA-1/CD54); a MHC:TCR association is not necessary,. The stimulation was peptide specific. With peptide and without “vesicles” (in vitro) the cells died within 2-3 days.
The composition of DC-derived exosomes using GM-CSF stimulated cells and the apoptosis blebs form the DC cells induced to apoptosis by ultraviolet light, showed appreciable differences. (However, considerable caution should be shown in considering these results, because the DCs were stimulated and presumably mature, and ultraviolet light is not a normal, physiological apoptosis trigger. As has been elaborated elsewhere, “physiological” dissolution [or apoptosis] would be expected to be slower and probably more organized and, in some contexts, have active participation involving surrounding cells.) In particular, chromatin, with DNA fragments and histones were prominent in the apoptosis blebs. Exosome components vary by cell-types and, in the case of DCs between immaturity and maturity. (There are some differences in the components found compared to the earlier assessments; in particular, the earlier report lists ribosomal proteins and does not list enolase, etc. Another study found enolase and GAPDH in addition to PrP, apparently as part of the exosomal membrane wall.)
Enzymes represented in 2006 (MD-DC) include :
Pyruvate kinase M1
Thioredoxin peroxidase 2
Peptidyl prolyl cis-transisomerase A (Cyclophilin A)
Non-enzymic major proteins within exosomes include :
HSC73 A chaperone protein that may carry antigenic proteins/peptides into cells via surface receptor mechanisms, to enter the proteasomal or endosome pathways
Gi2α A G-protein chain, strongly represented, 32% of total protein
A later review provides an accumulated and imposing list of protein compounds only, found in mouse-derived exosomes, using mass spectrometry. The relative importance of each is not given, and enolase is not listed, but lists Histones 1-3 and Ribosomal proteins 16 & s18, proteins that may bind DNA or RNA fragments respectively.
Protein electrophoresis showed bands of proteins with sizes below ~20 kDa. These were not characterized. Another study examined the ubiquinated proteins associated with exosomes (whether from immature of mature DCs is not clear). There were many intra-exosomal polyubiquinated proteins with the possibility of mono-ubiquinated proteins as well. These were released upon opening the exosomes (unlike MHC class II). The authors speculated that these proteins would act in some regulatory way, as determining the fates of the MVB; whether to fuse with lysosomes or secrete the exosomes. Exosomes have enriched levels of the tetraspanin proteins. These may be involved in membrane fusion and/or fission, and can influence motility.
Exosomes (apparently) within multi-vesicular bodies and below the plasma membrane were prominent in immature dendritic cells. There appear appreciable differences in the production and release of exosomes between immature and mature dendritic cells, with the mature cells and their exosomes acquiring MHC class II and the adhesion molecule ICAM-1, in particular, and losing MFG-E8. Those produced by sonic disruption may be quite artificial.
Note that Enolase and GAPDH are the only members of the Embden-Meyerhof-Parnas glycolytic pathway that are represented. Presumably they are not included “for fun.” Their relationship in other compartments raises the possibility that they may interact, as at the plus ends of microtubules beneath the plasma membrane and may be involved in an energy process of the Needham pathway type (see elsewhere on the website).
Exosomes are internalized by splenic CD8- dendritic cells, to become part of the late endocytic pathway and probably subjected to the processing involving the ESCRTs. These process and prepare peptides and MHC molecules for release with the new exosomes in the late endosome/multivesicular stores to the exterior,.
Particularly relevant work with delayed-type hypersensitivity reaction and collagen-induced arthritis in mice showed that exosomes (or the DC’s from which they are gathered) can, if expressing FasL induced by viral transfection, reduce inflammation, not only on the side of the administration, but the contralateral side as well. (The exosomes were prepared from bone marrow dendritic cells [CD11c+] matured with GM-CSF & IL-4, meaning that they had “inflammatory” characteristics.) This seemed to require “live” exosomes and was MHC class II dependent and showed antigenic specificity. Despite the expectation that FasL-FasR induced apoptosis might be important, there was no evidence that apoptosis was involved. Explanation was limited, but the conclusion was, that what was being demonstrated, was a particular immune-suppression mechanism utilizing exosomes, that could be induced by FasL and IL-10, and which can have “infectious” features. Exosomes from tumours may also modulate the expression of NHG2D expression on immune cells and thereby affect their function. The reason for this, being, as yet, unclear.
Whether thymocytes undergoing proliferation and maturation can produce exosomes is unclear. However, during the negative selection dissolution (apoptosis ?) apoptotic blebs can be expected to be produced, liberated and dispersed, to be picked up by macrophages and possibly dendritic cells or other cells. Consideration needs to be given to the possible factors that may be transferred that way, and how they may affect those cells that may participate in their uptake.
Tumours in the subcutaneous compartment are relatively refractory to bone marrow-derived dendritic cells unless the latter are administered subcutaneously, following which they can relocate in the T-call regions of the draining lymph nodes, where they seem most effective in local tumour rejection. Dendritic “vesicles” are not so restricted, and can be administered effectively intravenously (Kovar et al 2006). Memory CD8 + CTL are specific for the lymphoid location. (Those with innate responsive feature are probably largely of thymus origin, having processing involving haematopoietic cells.) When they are activated by subcutaneously administered dendritic cells (which usually reach and remain within the local nodes) the CTLs disperse generally to other lymphoid deposits and, presumably, the tumour deposit. Intravenously administered dendritic cells appeared in the spleen, and this acted like a lymph node with respect to tumour metastasis to the lungs.
In general, many workers studying dendritic cells assess their ability to be immunogenic (positive). Those studying the tolerogenic role (negative) of dendritic cells have been concerned with antigen collection, recognition, dendritic cell responses and cytokine production, with these factors downgrading any immune activation. Often these studies present an artificially large cohort of apoptotic cells. The maintenance of tolerance may be by the continuous phagocytosis of apoptotic cells under the cover of IL-10, assisted by the opsonin MFG-E8, without which, autoimmune phenomena develop.
Local tolerance to self antigens expressed by tumours (melanoma) may be broken by using bone marrow-derived dendritic cells loaded with a peptide self-antigen, when injected subcutaneously near the tumour site (also subcutaneous). An explanation for this breakdown of tolerance was by hypothesis, with the proposal that the dendritic cells, by providing a more professional APC function, caused the “switch on” of the CD8+ CTLs, yet with little downstream effect upon the normal melanocytes. This option raises the possibility that there were other antigenic factors involved. Another possibility, now possibly understood better and which may be involved to a varying degree, is that the peptide-loaded dendritic cells were able to inhibit or “switch off” specifically those Treg cells in the regional lymph nodes, thereby allowing the inherently sensitized CTL activity to inhibit the local tumour, but having minimal effect upon the melanocytes of the skin, perhaps because their antigen production was so low.
Tumours seem able to act as APCs and present certain antigens such as LAGE1 and ARTC1 to CD4 T cells, converting them to, and maintaining them as Treg, thereby suppressing the anti-tumour responses.
CEA seems unable to produce an immune rejection response in humans because it is expressed by the thymic medullary epithelial cells, in particular (slight CEA seemed expressed by the separated cortical epithelial cells and macrophages, both groups not expressing GAP75, the medullary marker). (cf C5; Żal 1994)
The earlier study of promiscuous antigenic gene expression relating to C-reactive protein (CRP; Klein et al. 1998, see earlier) used the rat monoclonal antibody CDR1 for cell separation prior to gene expression studies. This antibody bound to mouse mTECs, including the cell surface of TNCs, although, judging from the photomicrograph shown, binding could be variable and may indicate heterogeneity. CDR1 binding appears at post natal day 9 and does not bind to rat or human mTECs (Rouse et al. 1988). It seems to differ from a previously described antibody ERTR4. Bos et al. 2005, referred to thymic cell separation. For TNCs, Li et al. (2005) used the separation method described by Nihei et al. (2005), which relied upon foetal calf serum gradient sedimentation, with TNCs “heavier.”
The general impression that the promiscuous gene expression occurs within the thymus is a medullary function is dispelled by the study by Gotter et al. (2004) that shows that most genes are expressed in the epithelial cells of both medulla and cortex, and that the dendritic cells are also expressors, though less so. Examination of their Fig. 3 graphs indicates that the regressions lines for all cell types are similar, slopes of ~1. However, there are genes less expressed in the cortex, also less in the dendritic cells, particularly in the weaker range for those expressed in the medulla. There would be expected to be a considerable exposure of early thymocytes in the cortex to most of the gene products to be found in the medulla. Exposure to major epitopes may be expected in the cortex, with a range of weaker, lesser epitopes in the medulla.
This means that tumour hosts have tolerance largely induced by deletion. Some weak epitopes may escape this process, but subsequent immune stimulation may be suppressed by other means, such as Treg.
An endogenously-derived mouse tumour that induced hypoglycaemia did not produce tolerance. Rather, there was heightened immunity with increased tumour load. This finding goes contrary to the general experience. Perhaps the difference is in the hypoglycaemia.
Using lymph node cells responding in adjuvant-induced arthritis, plasmid-introduced DNA for Hsp60, Hsp70 and Hsp90 can ameliorate the arthritis, associated with T cell proliferation, reduction of TNFα production and up-regulation of IL-10 and TGFb (Th3 & Th3). These results point to the possible importance of the Hsp family and their peptide epitopes in peripheral tolerance induction and the potential for transfer of DNA fragments to induce these effects.
The peripheral blood of humans carries unresponsive CD4+ CD25+ T cells which are able to suppress the function of CD25— cells by cell-to-cell contact, without suppressing APC function. This was thought to involve the Notch pathway in the CD25+ T cells, which may involve γ-Secretase.
Autoantigen. For an “antigen factor” to become “infectious” it must multiply or amplify and not be diluted with cell proliferation. There are well known mechanisms for this that depend upon some form of template, from which copies can be generated, for example, DNA, RNA or certain proteins, such as prions. If the template does not reproduce, then the copies must be able to reform a template in the progeny. Auto-antigens are, by definition, derived ultimately from the host’s own DNA transcription (but perhaps with post-translational change), so that at cellular dissolution (apoptosis-like breakdown) certain DNA or derived RNA oligonucleotides, together with their binding proteins, may be selected, as by (perhaps antibody-) association with the antigenic peptides, RNA may be produced (± chaperones), and be transferred, in a transgenic-like process (probably using exosomes containing ribosomal proteins and chaperones; see Segura et al., 2005) to thymic progenitors, which can then proliferate, and thereby amplify the factors. These are subsequently released to the thymic dendritic cells at the dissolution of the clone’s multiplied cohort. Activation and MHC involvement are not required initially: this would be a role for immature dendritic cells.
Copyright © MA Traill 5/7/2008
 Traill MA. Winthrop Impulse 1970; 8(11):1
 Walker MR White GA et al. J. Leukoc. Biol. 1999; 66:120-126
 Pakravan N Hassan A et al. Cell. Mol. Immunol. 2007; 4(3):197-201
 Nossal GV. Eur. J. Biochem. 1991; 202:729-737
 Nossal GV. Cell 1994; 76:229-239
 Strasser A Puthalakath H et al. Immunol. Cell Biol. 2008; 86(1):57-66
 Hsieh C-S Liang Y et al. Immunity 2004; 21:267-277
 Hsieh C-S Zheng Y et al. Nat. Immunol. 2006; 7(4):401-410
 Matzinger P. Annu. Rev. Immuno. 1994; 12:991-1045
 Germain RN. Immunology 2008; 123:20-27
 Davis DM. Nat. Rev. 2007; 7:238-243
 Joly E & Hudrisier D. Nat. Immunol. 2003; 4(9):815
 LeMaoult J Caumartin J et al. Blood 2007; 109(5):2040-2046
 Caumartin J Favier B et al. EMBO J 2007; 26(5):1423-1433
 Dukers DP Meij PF et al. J. Immunol. 2000; 165:663-670
 Flanagan J Middeldorp J et al. J. Gern. Virol. 2003; 84:1871-1879
 Hoyne GF Dallman MJ et al. Immunol. 2000; 100:281-288
 Jenkinson EJ Jenkinson WE et al. Nat. Rev. 2006; 6:551-555
 Hoyne GF Dallman HJ et al. Immunol. Rev. 2001; 182:215-227
 Anderson G Lane PJ et al. Nat. Rev. Immunol. 2007; 7(12):954-963
 Robey E & Fowlkes BJ. Annu. Rev. Immunol. 1994; 12:675-705
 Von Boehmer H. Cell 1994; 76(2):219-228
 Punt JA Havran W et al. J. Exp. Med. 1997; 196(11):1911-1922
 Sprent J & Kishimoto H. Phil. Trans. R. Soc. Lond. B 2001; 356:609-616
 Pugliese A. Immunology 2004; 111:138-146
 McBeth C Seamons A et al. J. mol. Bio. 2008; 375:1306-1319
 Sprent J & Webb SR. Curr. Opin Immunol. 1995; 7:196-205
 Strasser A Puthalakath H et al. Immunol. Cell Biol. 2008; 86:57-66
 Gotter J Brors B et al. J. Exp. Med. 2004; 199(2):155-166
 Gallegos AM & Bevan MJ. Immunol. Rev. 2006; 209:290-296
 Ignatowicz L Kappler J et al. J. Immunol. 1996; 157:1827-31
 Gallo EM Winslow MM et al. Nature 2007; 450(7170):731-735
 Laky K & Fowlkes B. J. Exp. Med. 2007; 204(9):2115-2129
 Kappes DJ. Immunity 2007; 27:691-693
 Kronenberg M & Rudensky A. Nature 2005; 435:598-604
 Liu C-P Crawford F et al. Proc. Natl. Acad. Sci. USA 1998; 95:4522-4526
 Goldman KP Park CS et al. 2005; 35(3):709-717
 Żal T Volkman A et al. J. Exp. Med. 1994; 180:2089-2099
 Lambroploulou M Tamiolakis D et al. Med. Sci. Monit. 2007; 13(12):BR280-285
 Steinman RM Turley S et al. J. Exp. Med. 2000; 191(3):411-416
 Klein L Klein T et al. J. Exp. Med. 1998; 188(1):5-16
 Pugliese A Brown D et al. J. Clin. Invest. 2001; 107:555-564
 Pezzano M Samms M et al. Microbiol. Molec. Biol. Rev. 2001; 65(3):390-403
 Webb O Kelly F et al. Cell. Immunol. 2004; 228(2):119-129
 Pezzano M King KD. Cell. Immunol. 1998; 185(2):123-133
 Toussaint-Demylle D Scheiff J-M et al. Cell Tissue Res. 1990; 261:115-123
 Brelińska R. Cell Tissue Res. 1989; 258:637-643
 Alvarez-Vallina L González A et al. J. Immunol 1993; 150(1):8-16
 Čolić M Vučević D et al. Immunology 1994; 83:449-456
 Li Y Pezzano M et al. Cell Immunol. 1992; 140(2):495-506
 Hiramine C Nakagawa T et al. Cell Immunol. 1995; 160(1):157-162
 Anderson G Moore NC et al. Annu. Rev. Immunol. 1996; 14:73-99
 Imachi H Murao K et al. Lab. Invest. 2000; 80(2):263-270
 Philp D Pezzano M et al. Cell Immunol. 1993; 148(2):301-315
 Penninger J Rieker T et al. Eur. J. Immunol. 1994; 80(2):263-270
 Rieker T Penninger J et al. Eur. J. Immunol. 1993; 23(4):904-910
 Rieker T Penninger J et al. Dev. Comp. Immunol. 1995; 19(4):281-289
 Samms M Philp D et al. Cell Immunol. 1999; 197(2):108-115
 Li A Lie X et al. Cell. Mol. Immun. 2005; 2(4):301-305
 Pezzano M Philp D et al. Cell Immunol. 1996; 169(2):174-184
 Aguilar LK Aguilar-Cordova E et al. J. Immunol. 1994; 152(6):2645-2651
 Rouse RV Bolin LM et al. J. Histochem. Cytochem. 1988; 36(12):1511-1517
 Mucci J Hidalgo A et al. Proc. Natl. Acad. Sci. USA 2002; 99(6):3896-3901
 Cao WM Murao K et al. J. Mol. Endocrin. 2004; 32:497-505.
 Péret-Cruet J & Dancey JT. Experimentia 1977; 33(5):646-649
 Muños-Fernández MA Pimentel-Muiños FX et al. 1992; 76:439-445
 Vučević D Čolić M et al. Dev. Immunol. 2002; 9(2):63-72
 Jordan MS Boestreanu A et al. Nat. Immunol. 2001; 2(4):301-306
 Brelińska R Seidel H-J et al. Cell Tissue Res. 1991;264:175-183
 Savill J & Fadok V. Nature 2000; 407(6805):784-788
 Fadok V Bratton DL et al. J. Clin. Invest. 2001; 108:957-962
 Shibata T Nagata K et al. J. Immunol. 2007; 179(6):3407-3411
 Gregory CD. Curr. Opin. Immunol. 2000; 12(1):27-34
 Sauter B Albert ML et al. J. Exp. Med. 2000; 191(3):423-433
 Rovere P Sabbadini MG et al. J. Leukoc. Biol. 1999; 66:345-349
 McCormack JE Kappler J et al. Proc. Natl. Acad. Sci. USA 1994; 91:2086-2090
 Hoyne GF Le Roux I et al. Int. Immun. 2000; 12(2):177-185
 Nagata S Nagase H et al. Cell Death Differ. 2003; 10:108-116
 Nagase H Fukuyama H et al. Cell Death Differ. 2003; 10:142-143
 Hanayama R Tanaka M et al. Science 2004; 304:1147-1150
 Scott RS McMahon et al. Nature; 411(6834):207-211
 Flotte TJ Springer TA et al. Am. J. Pathol. 1983; 111:112-124
 Ho M-K & Springer TA. J. Biol. Chem. 1983; 258(1):636-642
 Painz R Walter I et al. Immunobiology 2008; 212:863-875
 Surh CD & Sprent J. Nature 1994; 372(6501):100-103
 Throsby M Homo-Delarche F et al. Endocrinology 1998; 138:2399-2406
 Paessens LC Fluitsma DM et al. J. Pathol. 2007; 214(1):96-103
 Reddien PW Cameron S et al. Nature 2001; 412(6843):198-202
 Green DR & Beere HM. Nature 2001; 412(6843):133-135
 Cascuila-Rosen L Rosen A et al. Proc. Natl. Acad. Sci. USA 1996; 93:1624-1629
 Neeli I Richardson MM et al. Mol. Immunol. 2007; 44(8):1914-1921
 Shi Y Evans JE et al. Nature 2003; 425(6957):516-521
 Burgdorf S & Kurts K. Curr. Opin. Immunol. 2008; 20:89-95
 Haigler HT & McKanna JA et al. J. Cell Biol. 1979; 81:382-395
 Ronchetti A Revera P et al. J. Immunol. 1999; 163:130-136
 Albert ML Pearce SF et al. J. Exp. Med. 1998; 188(7):1359-1368
 Larregina AT Morelli AE et al. Blood 2004; 103:811-819
 Trombone AP Silva PL et al. Genetic Vaccines Ther. 2007; 5:9
 Leber JH Crimmins GT et al. PLoS Pathog. 4(1): e6
 Emlen W Niebur J et al. J. Immunol. 1994; 152:3685-3692
 Emlen W Holers VM et al. J. Immunol. 1992; 148(10):3042-3048
 Pradhan D Krahling S et al. Mol. Biol. Cell 1997; 8:767-778
 Seitz HM Camenisch TD J. Immunol. 2007; 178(9):5635-5642
 Häcker H Mischak H et al. EMBO J. 1998; 17(21):6230-6240
 Leadbetter EA Rifkin IR et al. Nature 2002; 416(6881):573-576
 Rifkin IR Leadbetter EA et al. Immunol. Rev. 2005; 204:27-42
 Hislop JN Marley A et al. J. Biol. Chem. 2004; 279(21):22522-22531
 Benimetskaya L Loike JD et al. Nat. Med. 1997; 3(4):414-420
 Inaba K Turley S et al. J. Exp. Med. 1998; 188(11):2163-2173
 Yamazaki S Patel M et al. Proc. Natl Acad. Sci. USA 2006; 103(8):2758-2763
 Walker MR Kasprowicz DJ et al. J. Clin. Invest. 2003; 112:1437-1443
 Fantini MC Becker C et al. J. Immunol.2004; 172:5149-5153
 Huang J-F Yang Y et al. Science 1999; 286:952-954
 Regnault A Lankar D et al. J. Exp. Med. 1999; 189(2):371-380
 Kleijmeer M Ramm G et al. J. Cell Biol. 2001; 155(1):53-63
 Chow A Toomre D et al. Nature 2002; 418(6901):988-994
 Hwang I Huang J-F et al. J. Exp. Med. 2000; 191(7):1137-1148
 Rudensky AY Preston-Hurlburt P et al. Nature 1991; 353(6345):622-627
 Patel DM Arnold PY et al. J. Immunol. 1999; 63:5201-5210
 Poupot M Pont F et al. J. Immunol. 2005; 174:1414-1722
 Rechavi O Goldstein I et al. PLoS ONE. 2007; 2(11):e1204
 Chang M-K Bergmark C et al. Proc. Natl. Acad. Sci. USA 1999; 96:6353-6358
 Mambula SS Stevenson MA et al. Methods 2007; 43(3):168-75
 Dai S Wan T et al. Clin. Cancer Res. 2005; 11(20):7554-7563
 Millar DG Garza KM et al. Nat. Med. 2003; 9(12):1469-1476
 Beere HM & Green DR. Trends Cell Biol. 2001; 11(1):6-10
 Süss G & Shortman K. J Exp. Med. 1996; 183:1789-1796
 Zhang H Su X et al. J. Immunol. 1999;162:1423-1430
 Watanabe N Wang YH et al. Nature 2005; 436(7054):1181-1185
 Chen J Yang W et al. Immunol. Invest. 2008; 37(3):203-214
 Zhu M Chin RK et al. J. Immunology 2007; 179:8069-8075
 Lio CW & Hsieh CS. Immunity 2008; 28(1):100-111
 Chen W Jin W et al. J. Exp. Med. 2003; 198(12):1875-1886
 Ghiringhelli F Puig PE et al. J. Exp. Med. 2005; 202(7):919-929
 Ghiringhelli F Ménard C et al. J. Exp. Med. 2005; 202(8):1075-1085
 Ghiringhelli F Menard C et al. Cancer Immunol. Immunother. 2007; 56:641-648
 Anderson MS Venanzi ES et al. Science 2002; 298:1395-1401
 White AJ Withers et al. Eur. J. Immunol. 2008; 38(4):942-947
 Suzuki E Kobayashi Y et al. Autoimmunity 2008; 41(2):133-139
 Théry C Regnault A et al. J. Cell Biol. 1999; 147:599-610
 Sprent DM. Blood Cells Mol. Dis. 2005; 35(1):17-20
 Hwang I Shen X et al. Proc. Natl. Acad. Sci. USA. 2003; 100(11):6670-6675
 Kovar M Boyman O et al. Proc. Natl. Acad. Sci. USA. 2006; 103:11671-11676
 Théry C Boussac M et al. J. Immunol. 2001; 166:7309-7318
 Chaput N Flament C et al. J. Leuk. Biol. 2006; 80:471-478
 Segura E Nicco C et al. Blood 2005; 106:216-223
 Fevrier B Vilette D et al. Proc. Natl. Acad. Sci. USA, 2004;101(26):9683-9688
 Castellino F Boucher PE et al. J. Exp. Med. 2000; 191(11):1957-1964
 Mignot G Roux S et al. J. Cell. Mol. Med. 2006; 10(2):376-388
 Buschow DI Liefhebber JM et al. Blood Cells Mol. Dis. 2005; 35(3):398-403
 Hemler ME. Annu. Rev. Cell Dev. Biol. 2003; 19:397-422
 Segura E Amigorena S et al. Blood Cells Mol. Dis. 2005; 35:89-93
 Slagsvold T Pattni K et al. Trends Cell Biol. 2006; 16(6):317-326
 Morelli AE Larregina AT et al. Blood 2004; 104:3257-3266
 Van der Goot FG & Gruenberg J. Trends Cell Biol. 2006; 16(10):514-521
 Kim SH Bianco N et al. Mol. Ther. 2006; 13(2):289-300
 Xu Y Zhan Y et al. J. Immunol. 2007; 179:7577-7584
 Clayton A & Tabi Z. Blood Cells Mol. Dis. 2005; 34:206-213
 Eggert AA Schreurs MW et al. Cancer Res. 1999; 59:3340-3345
 Horai R Mueller KL et al. Immunity 2007; 27:775-785
 Mullins DW Sheasley SL et al. J. Exp. Med. 2003; 198(7):1023-1034
 Morelli AE Larregina AT et al. Blood 2003; 101:611-620
 Asano K Miwa M et al. J. Exp. Med. 2004; 200(4):459-467
 Schreurs MW Eggert AA et al. Cancer Res. 2000; 60:6995-7001
 Wang HY Peng G et al. 2005; 174:2661-2670
 Bos R van Duikeren S et al. Cancer Res. 2005; 65(14):6443-6449
 Nguyen LT Elford AR et al. J. Exp. Med. 2002; 195(4):423-435
 Quintana FJ Carmi P et al. Arthritis Rheum. 2004; 50(11):3712-3720
 Ng WF Duggan PJ et al. Blood 2001; 98(9):2736-2744
 Boulton ME Cai J et al. J. Cell. Mol. Med. 2008; “Postprint”;10.1111/j.1582-4934.2008.00274.x
 Larregina AT Morelli AE et al. Blood 2004; 103:811-819