Main

MHC class II-restricted antigen presentation is essential for CD4+ T cell-dependent immune responses. Different professional antigen-presenting cells (APCs) participate in a wide range of processes that are necessary for the generation of an effective and specific immune response. Dendritic cells (DCs) sample their environment and capture foreign antigens, such as those derived from bacteria or viruses, and initiate adaptive immune responses against these antigens by activating CD4+ and CD8+ T cells1. DCs and epithelial cells in the thymus express (and also capture) numerous self proteins and contribute to central tolerance and peripheral tolerance. By contrast, each B cell generally captures a single antigen after binding to its antigen-specific surface B cell receptor (BCR)2 and then presents peptides from this antigen to specific T cells. Regardless of APC type, all APCs specifically interact with distinct subsets of T cells that express antigen-specific T cell receptors (TCRs) on their surface. The specificity of this interaction depends on the ability of APCs to display antigenic peptides, immobilized by MHC class I and class II molecules, on their surface. MHC class II binds antigenic peptides that are generated by proteolysis of self and non-self proteins in endosomes and lysosomes, and 'presents' them to antigen-specific CD4+ T cells3. Recognition of peptide–MHC class II by CD4+ T cells stimulates their activation and differentiation into T helper cell subsets and also mediates interactions between antigen-specific B cells and T helper cells. Recognition of peptide–MHC class I complexes on DCs by CD8+ T cells stimulates the differentiation of CD8+ T cells into cytotoxic T cells. Understanding the mechanisms that are involved in the effective generation (and maintenance) of peptide–MHC class II complexes in APCs provides much needed insight into T cell function in normal and pathological states. In this Review, we discuss the existing literature regarding MHC class II biosynthesis, maintenance and degradation, with a particular emphasis on the processes that are important for APC function.

Basics of peptide loading of MHC class II

The basic steps that are required for MHC class II molecules to bind antigenic peptides are similar in all APCs and have been recently reviewed4, and for this reason we briefly summarize these steps before discussing cell type-specific differences in antigen processing and presentation in APCs. The antigenic peptides presented by MHC class II molecules on splenic DCs and B cells are derived from almost all intracellular locations5 and use a variety of mechanisms to access endosomal and/or lysosomal antigen-processing compartments (described below). MHC class II molecules must also access these compartments for efficient peptide binding (Fig. 1). Newly synthesized MHC class II molecules bind to a non-polymorphic protein termed invariant chain (Ii), which contains targeting motifs that direct the Ii–MHC class II complex to endosomal–lysosomal antigen-processing compartments6. After trafficking from the trans-Golgi network to the plasma membrane, recognition of the targeting motifs in Ii enables cell surface Ii–MHC class II complexes to undergo clathrin-mediated endocytosis7,8. Following endocytosis9, Ii–MHC class II complexes traffic through the endocytic pathway from early endosomes to late endosomal–lysosomal antigen-processing compartments, which contain antigenic proteins and peptides.

Figure 1: Generation of peptide–MHC class II complexes in antigen-presenting cells.
figure 1

MHC class II αβ dimers associate with the invariant chain (Ii) in the endoplasmic reticulum (ER), traffic through the Golgi apparatus and are delivered to the plasma membrane. Ii–MHC class II complexes are internalized by clathrin-mediated endocytosis and traffic to multivesicular antigen-processing compartments. Some of these complexes are sorted into the intraluminal vesicles (ILVs) of multivesicular bodies (MVBs), in which sequential Ii proteolysis leads to the generation of a fragment of Ii, termed class II-associated invariant chain peptide (CLIP), which remains in the peptide-binding groove of MHC class II. CLIP is removed from CLIP–MHC class II complexes by the enzyme HLA-DM, which is present in the MVB internal and limiting membranes, thereby facilitating peptide binding onto nascent MHC class II. The activity of HLA-DM is regulated by HLA-DO, but the mechanism of regulation remains unknown. ILV-associated peptide–MHC class II somehow associates with the MVB limiting membrane, and endosomal–lysosomal tubules directed towards the plasma membrane either directly fuse or give rise to transport vesicles that fuse with the plasma membrane. MVB membranes are rich in the lipids that constitute lipid raft membrane microdomains, and fusion of peptide–MHC class II from MVB-derived membranes with the plasma membrane leads to the deposition of lipid raft-associated peptide–MHC class II directly into the plasma membrane. In circumstances in which an entire MVB fuses with the plasma membrane, the ILVs of MVBs are released from the APC in the form of exosomes. Surface expressed peptide–MHC class II can be internalized through a clathrin-independent endocytosis pathway and can be targeted for lysosomal degradation or can recycle back to the plasma membrane.

PowerPoint slide

MHC class II is not able to bind antigenic peptides until Ii is proteolytically degraded and dissociates from the Ii–MHC class II complex10. Although peptide–MHC class II complexes can be generated throughout the endocytic pathway11, the typical antigen-processing compartment is a late endosomal or lysosomal structure that is enriched in proteolytic enzymes and disulphide reductases and has a sufficiently low pH to activate proteases for antigen proteolysis3,4,12. In B cells, DCs and macrophages, these compartments exist in various forms, have complex internal membrane structures and are all forms of multivesicular bodies (MVBs). Ii is degraded by stepwise proteolysis in these compartments and a small fragment of Ii, termed the class II-associated invariant chain peptide (CLIP), remains bound to the MHC class II peptide-binding groove. This fragment of Ii must be removed from MHC class II to allow binding of antigenic peptides. The release of CLIP is facilitated by the enzyme HLA-DM (known as H2-M in mice), which is an MHC class II-like protein encoded in the MHC locus13. HLA-DO (known as H2-O in mice) is another MHC class II-like protein that suppresses HLA-DM activity in mildly acidic antigen-processing compartments14. After CLIP removal, the nascent MHC class II molecule then binds antigenic peptides, leaves the MVB in tubulovesicular endosomes and is inserted into the plasma membrane for recognition by CD4+ T cells.

Mechanisms of antigen acquisition and processing

Macropinocytosis. APCs use various different cellular processes to capture and to deliver diverse antigens to antigen-processing compartments (Fig. 2). DCs are very efficient APCs as a result of their ability to internalize extracellular material through a nonspecific process termed macropinocytosis15. Immature DCs have a very robust process of constitutive macropinocytosis and this is thought to be a major mechanism of antigen acquisition by DCs and the main mode of entry of vaccinia virus, adenovirus type 3 and herpes simplex virus into cells16. By contrast, B cells have little macropinocytic activity; these cells rely almost exclusively on receptor-mediated endocytosis for antigen acquisition (described below). It has also been suggested that the self proteins and peptides present in lymph fluid are important for peripheral tolerance17, although the extent to which macropinocytosis mediates the uptake of these antigens by APCs remains to be determined.

Figure 2: Pathways of antigen endocytosis in antigen-presenting cells.
figure 2

Antigens can enter the endocytic pathway of antigen-presenting cells (APCs) by several distinct mechanisms. Receptor-mediated endocytosis via clathrin-coated vesicles requires antigen binding to one of several receptors on APCs — such as lectin receptors — which results in their internalization into early endosomes. Macropinocytosis, which is a form of cell 'drinking', is an actin-dependent process that leads to the uptake of soluble material into the cell in a macropinosome. In each of these processes, the internalized early endosomes eventually fuse with multivesicular late endosomal–lysosomal antigen-processing compartments. It is in these compartments that internalized antigen proteolysis and peptide–MHC class II complex formation takes place. Phagocytosis is an endocytic process in which opsonized particles bind to various receptors and, following actin-dependent membrane reorganization around the particle, enter cells in membrane-derived phagosomes. Phagosomes are not particularly rich in proteases or MHC class II and, after Toll-like receptor (TLR)-induced fusion with lysosomes or potentially with MHC class II-containing late endosomal–lysosomal compartments, the resulting phagolysosome generates peptide–MHC class II complexes. Autophagy is a process by which membranes (often derived from the endoplasmic reticulum (ER)) envelop cytosolic antigens to form an autophagosome. Upon autophagosome fusion with lysosomal compartments, the resulting autophagolysosome generates peptide–MHC class II complexes. It is important to remember that this is a schematic and that each APC does not contain only one early endosome and one antigen-processing compartment; there is a gradient of endosomes containing varying amounts of the essential components that are necessary to generate peptide–MHC class II complexes in every cell and it is for this reason that the nature of the antigen-processing compartment used by different antigens will differ.

PowerPoint slide

Activation of DCs in vitro leads to a transient burst in macropinocytosis that is eventually (within hours) terminated18,19. Macropinocytosis involves the formation of plasma membrane ruffles and is regulated by the RHO GTPase cell division cycle 42 (CDC42)20 and by RAC-dependent remodelling of the actin cytoskeleton19. Studies using bone marrow-derived DCs revealed that DC maturation inhibits macropinocytosis by reducing the amount of GTP-bound CDC42 and that macropinocytosis can be restored in mature DCs by overexpressing a constitutively active form of CDC42 (Ref. 20). Similarly, DC activation is associated with the suppression of uptake of larger particles (such as 1 μm latex beads) by nonspecific phagocytosis21,22. Given this suppression of nonspecific antigen uptake in mature DCs, the general tenet was that immature DCs are the main samplers of antigen in peripheral tissues and that maturation converts DCs from antigen-sampling cells to antigen-presenting cells. Recent studies have challenged this dogma. Activation of DCs in vivo, by injection of the Toll-like receptor (TLR) ligands CpG-containing oligodeoxynucleotides, lipopolysaccharide (LPS) or Escherichia coli, did not markedly suppress soluble antigen uptake23,24. Accordingly, in vivo-matured DCs could effectively process and present soluble (pinocytosed) antigens to antigen-specific CD4+ T cells ex vivo22,24. Moreover, in vivo maturation of DCs followed by immunization with soluble antigen in the presence of adjuvant has been reported to promote antigen-specific CD4+ T cell proliferation in mice24, but this has not been observed in other studies22,23.

Receptor-mediated endocytosis. In addition to nonspecific antigen uptake, APCs have several cell surface receptors that facilitate antigen entry into antigen-processing compartments through clathrin-mediated endocytosis. B cells with antigen-specific BCRs are 1,000-times more efficient in processing and presenting antigens to cognate T cells than B cells with BCRs specific for an irrelevant antigen25. B cells also have cell surface complement receptors and Fc receptors for IgG (FcγRs) that are involved in both clathrin-dependent and clathrin-independent antigen internalization, processing and presentation26,27. Unlike macropinocytosis, clathrin-coated pits and vesicles persist in activated DCs20. In fact, FcγRs that are present on both immature and mature DCs target immune complexes to antigen-processing compartments for efficient processing and presentation to antigen-specific CD4+ T cells22. DCs and B cells also express lectin receptors, such as the mannose receptor and DEC205 (also known as LY75), which recognize carbohydrate residues on self proteins and on some pathogens and target them for internalization either by clathrin-dependent endocytosis or by phagocytosis. Conjugation of antigens to ligands for these endocytic receptors has been used to deliver them to endosomes and/or lysosomes for efficient antigen processing and presentation22. Targeting of antigens to endocytic receptors on DCs such as DEC205 can enhance the efficiency of priming of naive CD4+ T cells by 1,000-fold compared with antigen internalized by macropinocytosis28. Similarly, it has recently been shown that binding of DEC205-targeted antigen by B cells amplifies antigen-specific CD4+ T cell responses in vivo29.

Phagocytosis. Phagocytosis is probably the most important mechanism of foreign antigen uptake by DCs and macrophages in vivo30. Not all DC subtypes efficiently phagocytose foreign antigens; plasmacytoid DCs have severely compromised phagocytosis compared with CD11chi conventional DCs31. Phagocytosis of apoptotic cells by APCs generates a wide range of self peptides that, when presented on the surface of a peripheral DC, can induce peripheral tolerance against self antigens32. Similarly to macropinocytosis, nonspecific phagocytosis is also suppressed during DC activation20,21,22, a finding that is attributed to decreased expression of active CDC42 by mature DCs20. However, phagocytosis can also be mediated by receptors such as C-type lectins, Fc receptors, complement receptors, scavenger receptor A and CD36, which recognize ligands on pathogens and apoptotic cells. Although many of these receptors also internalize ligands by clathrin-mediated endocytosis, their ability to facilitate the internalization of large (>3 μm) particles (such as opsonized pathogens) is clathrin independent and does not rely on other components of the clathrin-mediated endocytosis machinery33. Activation of DCs in vivo does not markedly alter their ability to capture antigens by receptor-dependent phagocytosis and even after in vivo activation, DCs are capable of internalizing and processing phagocytosed cargo, and presenting it to CD4+ T cells22.

Phagocytosis requires large amounts of membrane for the capture of cargo, and proteomic analysis of purified phagosomes has implicated both early endosomes and the endoplasmic reticulum (ER) as sources of phagosome membrane. The recruitment of ER membrane to phagosomes is especially interesting given that DCs can activate MHC class I-restricted CD8+ T cells through cross-presentation of phagocytosed antigen34,35,36. Individual phagosomes are fairly non-proteolytic compartments; however, phagosomes undergo a series of membrane fusion events with lysosomes to generate hybrid organelles termed phagolysosomes30. The phagolysosome is a highly destructive organelle that contains MHC class II molecules and the complete MHC class II peptide-loading machinery. It has been reported that TLR signalling enhances phagosome 'maturation' in macrophages and DCs37,38; however, another study showed that TLR signalling suppresses phagosome maturation in macrophages39. Recruitment of TLRs to the phagosome is mainly mediated by the delivery of the endosome-associated adaptor protein AP3 to the phagosome membrane40, probably during the process of phagosome–endosome fusion.

Most pathogens that are internalized by phagocytosis will ultimately be destroyed in phagolysosomes. However, some pathogens, such as Mycobacterium tuberculosis, have evolved mechanisms to avoid destruction in a phagosome (Box 1). Proteolysis of pathogen-derived proteins that are present in phagolysosomes provides a plentiful source of antigenic peptides for loading onto MHC class II molecules, although it should be noted that the proteolytic activity of phagolysosomes in DCs is regulated to prevent the complete destruction of immunogenic peptides (as described below). Curiously, different phagosomes within a single DC can behave autonomously, such that one phagosome can mediate the presentation of antigens containing a TLR ligand, whereas a different phagosome is unable to present antigen (that does not contain a TLR ligand)37,41. A recent study found that phagosome content could mix between individual phagosomes as long as each phagosome contained a TLR-activating signal42. One advantage of such a mechanism would be to allow APCs to distinguish between phagosomes with self (non-stimulating) antigens and non-self (stimulating) antigens; however, definitive testing of this model has not been carried out.

Autophagy. Approximately 20–30% of the peptides that are eluted from MHC class II in B cells and human DCs are derived from cytosolic or nuclear proteins43, which suggests a role for macroautophagy in transporting cytosolic antigens to antigen-processing compartments44. Autophagosomes engulf cytosolic macromolecules and organelles and can fuse with endosomal–lysosomal antigen-processing compartments to form autophagolysosomes45. The presentation of autophagosome-encapsulated antigens requires intact lysosomal function, which highlights the importance of lysosomal proteolysis in T cell epitope formation46. Autophagy also leads to the generation of citrullinated proteins and peptides by peptidylarginine deiminase47, and antibody responses to citrullinated proteins have been linked to autoimmune rheumatoid arthritis48. Mutation of autophagy gene 5 (ATG5) significantly disrupts both positive and negative selection of CD4+ T cells by thymic epithelial cells49,50 and leads to a quantitative alteration of the MHC class II peptidome in cortical thymic epithelial cells49. ATG5-deficient DCs are also unable to activate herpes simplex virus-specific CD4+ T cells51, which suggests an important role for macroautophagy in peripheral DC function.

Regulation of endosomal and lysosomal proteolysis in APCs. Regulation of proteolysis in endosomes and lysosomes is essential for many aspects of MHC-restricted antigen processing and presentation. 'Lysosomal' proteases, such as asparaginyl endopeptidase52 and cathepsin S53,54, are essential not only for Ii degradation and dissociation from MHC class II but also for the generation of the antigenic peptides that ultimately bind to MHC class II. Cathepsin S, B, H and L redistribute from true lysosomes to MHC class II-containing late endosomal antigen-processing compartments upon DC activation55 — a process that promotes peptide–MHC class II complex assembly. Activation of DCs also increases the activity of the ATP-dependent vacuolar proton pump and acidifies antigen-processing compartments56, thereby acutely enhancing antigen proteolysis and peptide–MHC class II complex formation in maturing DCs56,57,58.

Although enhanced protease activity leads to increased antigen proteolysis, this does not necessarily lead to enhanced epitope generation and antigen presentation by DCs. Indeed, there are many T cell epitopes that are generated in early endosomes that are actually destroyed by lysosomal proteolysis59,60. A recent study has shown that the delivery of a model antigen to early endosomes by conjugation to a CD40-specific monoclonal antibody resulted in efficient antigen presentation to CD4+ T cells, whereas delivery of antigen to late endosomes or early lysosomes by conjugation to a DEC205-specific antibody resulted in fairly poor presentation61. Therefore, professional APCs have developed mechanisms to limit antigen proteolysis, thereby preserving antigenic peptide integrity. Although DCs are better than macrophages at stimulating naive T cells, lysosomal extracts of DCs are 50-times less active than those of macrophages60. Unlike macrophages (which rapidly and completely degrade internalized proteins), DCs can retain intact internalized antigens for many hours in vivo57,58,60.

Phagosomal compartments in DCs are less acidic than their counterparts in macrophages and this difference is mediated by the RAC2-dependent association of NADPH oxidase 2 (NOX2; also known as cytochrome b245 heavy chain) to the membrane of the developing phagosome62. It has recently been shown that during phagocytosis (but not macropinocytosis or receptor-mediated endocytosis), TLR2 signalling recruits the classic autophagy markers ATG8 and LC3 to the developing phagosome63. ATG8- and LC3-tagged phagosomes also recruit NOX2, which results in increased phagosome pH, prolonged antigen persistence in developing phagosomes and prolonged antigen presentation by MHC class II to CD4+ T cells63. These studies highlight the importance of maintaining a balance between antigenic protein proteolysis and complete proteolytic destruction of antigenic MHC class II-binding peptides in antigen-processing compartments in APCs.

Regulation of MHC class II biosynthesis

The expression of MHC class II is tightly regulated in all APCs both transcriptionally and post-transcriptionally. MHC class II transactivator (CIITA) is the master regulator of MHC class II gene expression64. CIITA is constitutively expressed only by professional APCs, such as B cells, DCs, macrophages and thymic epithelial cells. However, CIITA activity and ultimately MHC class II expression can be regulated in both professional and non-professional APCs65; for example, interferon-γ (IFNγ) stimulates CIITA expression and converts monocytes from MHC class II-negative cells to MHC class II-expressing functional APCs. Expression of mRNA encoding MHC class II also depends on the maturation status of APCs. Resting B cells express MHC class II mRNA and protein, and activation of B cells in vitro and in vivo upregulates this expression. Activation of CD11c+ conventional DCs in the spleen also leads to an increase in MHC class II protein synthesis; however, within 24 hours of stimulation, MHC class II synthesis in DCs is abrogated66. This termination of MHC class II biosynthesis does not occur following activation of plasmacytoid DCs67 owing to their sustained expression of CIITA68, and this suggests that there are fundamental differences in the role of newly synthesized MHC class II molecules in conventional DCs and plasmacytoid DCs69. Enhanced MHC class II synthesis following TLR signalling is thought to enable maturing CD11c+ DCs to respond to pathogens through the formation of new pathogen-specific peptide–MHC class II complexes in antigen-processing compartments, whereas the subsequent termination of MHC class II biosynthesis in mature DCs is thought to limit the generation of peptide–MHC class II complexes that are irrelevant to the invading pathogen70. As discussed below, the ability of DCs to effectively generate antigen-specific peptide–MHC class II complexes even when protein synthesis is terminated22 suggests that previously synthesized MHC class II molecules can enter antigen-processing compartments. Thus, the termination of MHC class II synthesis upon DC activation might limit, but not completely prevent, the formation of new antigenic peptide–MHC class II complexes.

MHC class II trafficking

Transport to antigen-processing compartments. In immature DCs, Langerhans cells and macrophages, large amounts of MHC class II are present in endosomal–lysosomal antigen-processing compartments before cell activation71,72,73. Activation of DCs by various TLR ligands leads to a marked change in MHC class II expression and cellular localization; in immature DCs, considerable amounts of MHC class II reside in both intracellular endosomes and lysosomes and on the plasma membrane, whereas in mature DCs, the vast majority of MHC class II resides on the plasma membrane (Fig. 3a). The increased expression of MHC class II on the plasma membrane following DC activation can probably be accounted for by considering several factors: the retention of MHC class II in antigen-processing compartments and the lysosomal degradation of intracellular MHC class II in immature DCs66,74,75; the increased biosynthesis of MHC class II upon DC activation67,76; the movement of intracellular MHC class II to the plasma membrane upon DC activation74,77,78,79; and the increased stability of MHC class II complexes in mature DCs74,76.

Figure 3: Activation of antigen-presenting cells redistributes intracellular MHC class II molecules.
figure 3

a | In immature dendritic cells (DCs), up to 75% of all MHC class II molecules reside in lysosome-like antigen-processing compartments. It is in these compartments that peptide–MHC class II complexes are generated (albeit slowly in immature DCs). The predominant intracellular localization of MHC class II in immature DCs is due to robust synthesis of invariant chain (Ii)–MHC class II complexes and the movement of these complexes to antigen-processing compartments. Although some peptide–MHC class II complexes formed in these compartments do traffic to the plasma membrane, much of the MHC class II that is synthesized in immature DCs is degraded in lysosomes. Maturation of DCs by pathogen uptake and/or Toll-like receptor (TLR) stimulation leads a transient burst in Ii–MHC class II biosynthesis, efficient generation of peptide–MHC class II complexes that traffic to the plasma membrane in tubular endosomes and degradation of residual MHC class II present in lysosomes. Fully mature conventional DCs do not synthesize Ii–MHC class II and therefore express very little MHC class II in antigen-processing compartments. Arrow thickness indicates prominence of a particular pathway and dashed arrows indicate minor pathways. In mature DCs, any internalized MHC class II is simply recycled to the plasma membrane where peptide–MHC class II complexes can interact with naive antigen-specific CD4+ T cells. Note that after DC activation, structures such as antigen-processing compartments still exist in the mature DC, but peptide–MHC class II complexes do not accumulate in these compartments. b | In resting B cells, large amounts of MHC class II are present on the plasma membrane and in peripheral early endosomal vesicles. Binding of antigen to the B cell receptor (BCR) triggers BCR endocytosis and recruits the peripheral MHC class II-containing vesicles towards the antigen–BCR-containing lysosome-like antigen-processing compartments. Internalized antigen is processed into peptides in these compartments and the resulting peptide–MHC class II complexes traffic to the plasma membrane for recognition by primed antigen-specific CD4+ T cells. TCR, T cell receptor.

PowerPoint slide

In contrast to DCs and macrophages, very little MHC class II resides in late endosomal compartments in resting B cells80 (Fig. 3b). MHC class II trafficking to the conventional antigen-processing compartment is induced by BCR signalling, which targets both MHC class II and BCR-associated antigen to the same late endosomal HLA-DM+ antigen-processing compartment for efficient peptide–MHC class II complex formation80,81. MHC class II redistribution is regulated by the actin-based motor protein myosin II and the spleen tyrosine kinase (SYK)-dependent reorganization of the actin cytoskeleton82,83. BCR-stimulated recruitment of MHC class II to antigen-processing compartments is not only an interesting cell biological process but also has important implications for immune function. Delivery of the model antigen hen egg lysozyme (HEL) to B cells by the HEL-specific BCR or by the macropinocytosis pathway generates HEL46–61 peptide–MHC class II complexes with different conformations that can be recognized by different HEL-specific CD4+ T cell clones84. This is presumably due to the initial formation of HEL46–61 peptide–MHC class II complexes in either MHC class II+HLA-DM+ late endosomes (following BCR-mediated HEL processing) or endosomal compartments that contain MHC class II but are devoid of HLA-DM (following soluble HEL processing). Macropinocytosis of soluble HEL together with nonspecific BCR crosslinking leads to generation of the macropinocytosis-associated peptide–MHC class II conformation84. This suggests that, similarly to phagosome autonomy in DCs38, dichotomy of antigen loading onto MHC class II in distinct intracellular compartments occurs in B cells. Similarly, autoreactive pathogenic CD4+ T cells in non-obese diabetic mice recognize an insulin peptide that binds to MHC class II in a different orientation to that of insulin-peptide–MHC class II complexes that are generated in the thymus and are therefore not negatively selected85. These peptide–MHC class II complexes form when preprocessed peptide (but not intact insulin protein) is incubated with APCs ex vivo85. Although the mechanism responsible for the generation of these distinct insulin-peptide–MHC class II complexes is not completely understood, it is likely that they form in distinct intracellular compartments that may have differences in pH, protease activity and levels of HLA-DM.

Transport to the plasma membrane. Following peptide binding, the peptide–MHC class II complexes leave the antigen-processing compartments and traffic to the plasma membrane. Activation of DCs results in the formation of elongated tubules that emanate from antigen-processing compartments77,79 and these tubules (or vesicles derived from these tubules) constitutively deliver peptide–MHC class II complexes to the DC plasma membrane. Interaction of antigen-loaded DCs with antigen-specific CD4+ T cells results in the transport of these peptide–MHC class II-containing tubules directly towards the immunological synapse that is formed between a DC and a T cell78. During DC activation, the late endosomal pool of MHC class II redistributes to peripheral vesicles that no longer contain late endosomal markers and finally the MHC class II is inserted into the plasma membrane74. It has recently been shown that complexes of MHC class II with an Eα fusion protein that is internalized by phagocytosis accumulate in punctate cytoplasmic structures that are devoid of markers of either endosomes or lysosomes40. It remains to be determined whether these structures are equivalent to the class II-containing transport vesicles (CIIVs) observed in B cells86; however, it is likely that these structures are MVB-derived transport vesicles that are destined for the plasma membrane.

Although B cells can clearly bind soluble antigens in vitro, most B cells in vivo acquire antigen that is immobilized on the surface of other cells2. The process by which this occurs has recently been identified. Similar to the polarized secretion of lysosome-derived peptide–MHC class II complexes at the immunological synapse between DCs and T cells78, B cells were shown to polarize their lysosomes towards the synapse formed at the site of contact with antigen immobilized on latex beads87. This process depends on reorganization of the B cell microtubule-organizing centre and leads to the secretion of lysosomal enzymes into the synapse, the proteolysis-dependent extraction of antigen from the surface of latex beads and the generation of specific peptide–MHC class II complexes for subsequent presentation to antigen-specific CD4+ T cells.

The precise molecular mechanisms that regulate the movement and the fusion of peptide–MHC class II-containing vesicles and tubules with the plasma membrane are poorly understood. Peptide–MHC class II-containing vesicles move in a 'stop-and-go' manner along microtubules towards the plasma membrane and various molecular motors, actin-binding proteins and GTPases control this process88. Disruption of the actin cytoskeleton inhibits LPS-induced peptide–MHC class II movement to the DC plasma membrane58. It is likely that peptide–MHC class II complex redistribution from intracellular antigen-processing compartments to the plasma membrane in activated DCs is regulated by the TRIF (also known as TICAM1)–RHOB pathway89,90. Genome-wide short interfering RNA screens for genes that regulate the movement of peptide–MHC class II molecules revealed that the GTPase ADP ribosylation factor-like protein 14 (ARL14; also known as ARF7) and myosin 1E are important for the actin-based movement of antigen-processing compartments to the DC surface91. However, exactly how these proteins regulate vesicle movement remains to be elucidated.

MHC class II secretion from APCs. When an MVB fuses directly with the plasma membrane, the entire contents of the MVB (including the intraluminal vesicles (ILVs)) are released into the extracellular space and the secreted ILVs are termed exosomes. Exosomes are readily released from all APC subtypes examined; these small (<100 nm) vesicles contain various T cell ligands (such as peptide–MHC class II, co-stimulatory molecules and adhesion molecules) that can both directly and indirectly stimulate antigen-specific CD4+ T cells92,93. Exosome release is constitutive in APCs, but it can also be stimulated following encounter of antigen-loaded APCs with antigen-specific CD4+ T cells94,95, which can then trap the antigen-bearing exosomes on their surface95. In addition to functioning as 'mini-APCs' to stimulate primed CD4+ T cells, exosomes can be internalized by APCs and can function as a source of antigen. Plasmacytoid DCs are fairly poor stimulators of naive CD4+ T cells, partly as a result of their poor phagocytic capacity, but these cells can effectively internalize exosomes and apoptotic bodies, which results in efficient T cell proliferation96. Peptide–MHC class II-bearing exosomes can also become attached to the surface of cells — such as follicular DCs — that do not themselves synthesize MHC class II proteins, enabling them to present antigen97. MHC class II molecules are also known to be present on mouse CD4+ T cells and this acquisition is likely to be a consequence of exosome binding to the T cell surface.

Given their ability to stimulate immune responses in vivo, DC-derived exosomes are promising candidates for use in cancer immunotherapy98. Tumour peptide-pulsed exosomes bearing MHC class II, adhesion and co-stimulatory molecules have been used as cell-free vaccines to eliminate tumours99, and tumour-derived exosomes have also been used as a source of antigen in DC-based therapy for malignant melanoma100. The many clinical protocols using a range of cell-derived exosomes highlight the promise of using these MVB-derived vesicles in a clinical setting.

Clustering in membrane microdomains. A considerable proportion of cell surface peptide–MHC class II is localized in discrete membrane microdomains known as lipid rafts. These cholesterol-rich microdomains are found in all mammalian cells and the association of MHC class II with lipid rafts has been observed in human and mouse B cells, DCs, monocytes, macrophages and thymic epithelial cells (reviewed in Ref. 101). MHC class II has also been reported to associate with tetraspanin proteins in structures termed tetraspan webs102. However, given the close relationship between membrane cholesterol, lipid rafts and tetraspan webs, it is unclear how these microdomains differ in vivo101. The presence of peptide–MHC class II in membrane microdomains is particularly important for T cell activation by APCs that have low levels of specific peptide–MHC class II on their surface103,104,105, a finding that has been attributed to the local concentration of small amounts of relevant peptide–MHC class II in discrete domains. Pulse–chase biosynthetic radiolabelling studies and the use of protein transport inhibitors have shown that MHC class II localizes to lipid rafts in antigen-processing compartments before peptide binding106,107. MVBs are enriched in the lipids that are constituents of lipid rafts, and we have proposed that coordinated peptide loading onto MHC class II in these compartments and transport of the complexes to the plasma membrane mediates the delivery of a portion of relevant peptide–MHC class II complexes that is present in a pre-formed lipid raft108. We have recently confirmed that peptide–MHC class II complexes generated in antigen-processing compartments first appear at the plasma membrane of DCs in small clusters and that the integrity of these clusters is altered by the disruption of lipid rafts105. These studies show that DCs have a mechanism that allows temporal coordination of intracellular peptide–MHC class II formation and the subsequent insertion of these same peptide–MHC class II complexes into the plasma membrane in membrane microdomains. Whether this is true in APCs other than DCs remains to be determined.

The fate of surface MHC class II in APCs

Recycling from the plasma membrane. Once expressed on the APC surface, peptide–MHC class II complexes do not simply persist there permanently. The plasma membrane of eukaryotic cells is constantly being internalized; macrophages internalize the equivalent of their entire plasma membrane in only 30 minutes109. Although many internalized plasma membrane proteins are transported to lysosomes for eventual degradation, many proteins efficiently recycle from the plasma membrane to early endosomes and back to the plasma membrane. However, little is known about the molecular machinery that specifically regulates MHC class II recycling in APCs.

Although some antigenic epitopes are only generated in the acidic antigen-processing compartments, others are only generated in early endosomes by limited proteolysis110. For example, influenza virus has an antigenic epitope that simply requires unfolding and mild proteolysis in the early endosomes of DCs111. Presentation of these 'early endosome' epitopes to T cells does not require Ii-dependent transport of MHC class II or HLA-DM-mediated peptide editing of CLIP–MHC class II, or even neosynthesis of MHC class II molecules112,113. Together, these findings suggest that such epitopes are presented on pre-existing MHC class II molecules that have been recycled from the plasma membrane into the early endocytic pathway. This alternative pathway of processing and presentation is likely to be important not only for epitopes that are easily destroyed in very acidic antigen-processing compartments but also for conditions in which MHC biosynthesis is terminated (such as following DC activation in vivo).

However, not all antigens that bind to recycling MHC class II molecules do so in early endosomes. Mature DCs that internalize antigenic Eα protein via receptor-mediated endocytosis generate Eα peptide–MHC class II complexes in conventional late endosomal–lysosomal antigen-processing compartments that are independent of new MHC class II biosynthesis22. Taken together, these data are consistent with a model in which 'old' (perhaps recycling) MHC class II complexes in mature DCs can generate 'new' peptide–MHC class II complexes following antigen internalization; this model challenges the long-held belief that mature DCs are unable to acquire, process and present antigens to T cells.

Regulation by ubiquitylation. Perhaps the greatest effect of DC activation on MHC class II cell biology is the regulation of peptide–MHC class II stability. Immature DCs synthesize large amounts of MHC class II, express peptide–MHC class II complexes on their surface and, under steady-state conditions, have a high rate of constitutive MHC class II degradation76. Activation of DCs dramatically reduces the kinetics of peptide–MHC class II turnover74,76 and in fully activated DCs, the half-life of surface peptide–MHC class II complexes is more than 100 hours76. The rapid turnover of peptide–MHC class II in immature DCs is primarily due to the ubiquitylation of a conserved lysine residue that is present in the MHC class II β-chain by the E3 ubiquitin ligase MARCH1 (Refs 114,115,116,117). Analysis of APCs isolated from MARCH1-deficient mice, or from mice in which this conserved MHC class II lysine residue has been mutated, reveals a direct role for ubiquitylation by MARCH1 in regulating intracellular accumulation and stability of peptide–MHC class II complexes117,118,119.

Ubiquitylation is not the only mechanism that regulates the turnover of peptide–MHC class II from the surface of DCs. The encounter of antigen-loaded APCs with antigen-specific CD4+ T cells results in the release of exosomes loaded with peptide–MHC class II complexes from APCs94,95, thereby reducing the cell-associated level of peptide–MHC class II. Whereas the targeting of MHC class II molecules for lysosomal degradation is ubiquitin dependent115,116,117,119, the sorting of MHC class II into the exosome pathway is ubiquitin independent95,120, which shows the complexity of protein sorting in the late endosomal and MVB system in DCs. We have recently shown that encounter of antigen-specific primed CD4+ T cells with antigen-pulsed DCs or crosslinking of surface peptide–MHC class II with an MHC class II-specific monoclonal antibody leads to rapid peptide–MHC class II endocytosis and lysosomal degradation121. This phenomenon is also independent of MHC class II ubiquitylation, as it is observed in DCs isolated from MARCH1-deficient mice121. Surprisingly, engagement of antigen-pulsed DCs with antigen-specific T cells also results in the degradation of immunologically irrelevant peptide–MHC class II complexes, and this is due to the presence of both relevant and irrelevant peptide–MHC class II complexes in lipid raft microdomains121. So, although ubiquitylation seems to regulate the steady-state turnover of MHC class II in DCs, it does not seem to control ligation-induced changes in peptide–MHC class II expression on the surface of DCs.

MARCH1 is expressed by resting B cells, immature DCs, monocytes and macrophages122, but not by epithelial cells in the thymus123. The mechanisms that control this selective expression of MARCH1 remain to be determined. MARCH1 targets not only MHC class II but also CD86 for degradation124,125, and therefore regulates the basal expression of at least two important T cell ligands in APCs. The anti-inflammatory cytokine interleukin-10 (IL-10) stimulates the synthesis of mRNA encoding MARCH1 in monocytes126, which explains the observation that IL-10 downregulates MHC class II expression and promotes lysosomal MHC class II accumulation in human monocytes127.

Expression of MARCH1 mRNA is rapidly downregulated following activation of conventional DCs118,128, thereby limiting the expression of MARCH1 to immature DCs. MARCH1 is also expressed by plasmacytoid DCs, but its levels are not altered by activation of these cells67, which might reflect important differences in the role of MHC class II turnover in conventional DCs and plasmacytoid DCs. Although MARCH1 is also expressed by resting B cells and macrophages, it remains to be determined whether activation of these cells also suppresses its expression. MARCH1 has a very short half-life (less than 30 minutes) and its expression is partly regulated by auto-ubiquitylation129,130. Therefore, DC activation rapidly leads to the termination of MARCH1 mRNA synthesis118,128, the downregulation of MARCH1 protein118 and the termination of MARCH1-mediated MHC class II ubiquitylation115,116. Interestingly, MARCH1 targets the internalizing peptide–MHC class II complexes in early endosomes of immature DCs131 in a process that could promote the degradation of 'older' surface-expressed peptide–MHC class II complexes and spare the 'new' peptide–MHC class II complexes that are still present in the antigen-processing compartment (Fig. 4).

Figure 4: Degradation of peptide–MHC class II complexes in antigen-presenting cells.
figure 4

After arrival at the cell surface, peptide–MHC class II complexes are internalized into early endosomes by clathrin-independent endocytosis in both immature and mature dendritic cells (DCs) at the same rate (kint); however, whether this is true in all antigen-presenting cell (APC) subsets remains to be determined. The E3 ubiquitin ligase MARCH1 is constitutively expressed by resting B cells and DCs, and its expression can be enhanced by exposure of monocytes or macrophages to interleukin-10 (IL-10; not shown). In immature DCs, ubiquitylation of internalized peptide–MHC class II complexes by MARCH1 at the plasma membrane and in early endosomes targets these complexes for lysosomal degradation. By contrast, inefficient ubiquitylation of peptide–MHC class II complexes by MARCH1 in B cells prevents their delivery to lysosomes and instead redirects them back to the plasma membrane (indicated by the dashed arrow). Activation of immature DCs rapidly terminates MARCH1 expression and peptide–MHC class II ubiquitylation. As internalized peptide–MHC class II complexes are not ubiquitylated in mature DCs, these complexes are spared from lysosomal degradation and enter a re-iterative recycling pathway back to the plasma membrane (indicated by the thicker arrow).

PowerPoint slide

In addition to transcriptional and translational control of MARCH1 expression, MARCH1 activity and MHC class II ubiquitylation are regulated by protein–protein interactions. The DC activation marker CD83 enhances MHC class II expression by suppressing the interaction of MHC class II with MARCH1 (Ref. 132). Both CD83 and MARCH1 can be found in early endosomes and recycling endosomes118,133, which are the intracellular compartments that are thought to be the sites of MHC class II ubiquitylation131. Recent data showing that Toll-interacting protein (TOLLIP) reduces the expression of MARCH1 and enhances the expression of MHC class II134 confirm the hypothesis that modulation of MHC class II ubiquitylation is a major factor in the regulation of MHC class II expression in APCs.

Mechanism of MARCH1-mediated MHC class II downregulation. In B cells, MHC class II is also ubiquitylated, but it does not seem to accumulate in lysosomes. Recent data suggest that differences in ubiquitin chain length in B cells and DCs may affect MHC class II localization. In DCs, MHC class II molecules contain longer ubiquitin chains that preferentially target MHC class II molecules to late endosomal–lysosomal compartments, whereas in B cells MHC class II molecules contain short ubiquitin chains that target MHC class II molecules to earlier endosomes135. The relationship between ubiquitin chain length and the steady-state intracellular distribution of MHC class II was confirmed by expressing genetically engineered MHC class II containing different ubiquitin conjugates in each cell type135. Similar to many other ubiquitylated integral membrane proteins, it is probable that ubiquitylated MHC class II molecules interact with the multiprotein endosomal sorting complex required for transport (ESCRT) that directs ubiquitylated proteins to MVBs for eventual lysosomal degradation136. It is probable that MHC class II molecules with longer ubiquitin chains bind with higher affinity to ESCRT components and traffic to MVBs more effectively than those with short ubiquitin chains (or none at all). We have recently found that internalized peptide–MHC class II recycles very poorly in immature DCs and very efficiently in mature DCs (P.A.R. and K.-J. Cho, unpublished observations). The failure of peptide–MHC class II complexes to recycle in immature DCs is MARCH1 dependent, which is consistent with the idea that ubiquitylation targets internalizing peptide–MHC class II for ESCRT-dependent lysosomal degradation. Such a process could be important for generating a rapidly changing peptide–MHC class II repertoire in immature DCs that can be 'fixed' upon DC activation.

Concluding remarks

To maximize the chance of antigen presentation to antigen-specific T cells, professional APCs have adapted various processes to deliver both exogenous and endogenous antigens, whether self or foreign, into antigen-processing compartments. Depending on its form and on the role of the APC that is recognizing the antigen, different endocytic processes are involved in antigen processing; for example, phagocytosis of pathogens for stimulation of naive CD4+ T cells by DCs, or BCR-mediated endocytosis of pathogens or pathogen proteins by B cells for stimulation of primed CD4+ T cells. In each of these diverse situations, peptide–MHC class II complexes form in intracellular compartments and traffic to the APC plasma membrane, where they are either very stable (as in mature DCs) or undergo rapid turnover (as in immature DCs). This Review highlights the steps involved in peptide–MHC class II biosynthesis, transport and turnover in different types of APC and provides an insight into the complexity of these processes. Understanding the molecular mechanisms underlying these processes can potentially initiate new approaches to manipulate distinct steps in the antigen processing and presentation pathway in APCs in an attempt to combat human disease.