The adaptive immune system plays an essential role in protecting vertebrates against a broad range of pathogens and cancer. The MHC class I-dependent pathway of antigen presentation represents a sophisticated cellular machinery to recognize and eliminate infected or malignantly transformed cells, taking advantage of the proteasomal turnover of the cell's proteome. TAP (transporter associated with antigen processing) 1/2 (ABCB2/3, where ABC is ATP-binding cassette) is the principal component in the recognition, translocation, chaperoning, editing and final loading of antigenic peptides on to MHC I complexes in the ER (endoplasmic reticulum) lumen. These different tasks are co-ordinated within a dynamic macromolecular peptide-loading complex consisting of TAP1/2 and various auxiliary factors, such as the adapter protein tapasin, the oxidoreductase ERp57, the lectin chaperone calreticulin, and the final peptide acceptor the MHC I heavy chain associated with β2-microglobulin. In this chapter, we summarize the structural organization and molecular mechanism of the antigen-translocation machinery as well as various modes of regulation by viral factors and in genetic diseases and tumour development.
TAP (transporter associated with antigen processing) is a key player in the adaptive immune response. This heterodimeric transport complex, consisting of the ABC (ATP-binding cassette) half-transporter TAP1 and TAP2, translocates antigenic peptides from the cytosol into the ER (endoplasmic reticulum) lumen where they are loaded on to MHC class I molecules. MHC I molecules loaded with peptides are targeted to the cell surface for presentation of their antigenic cargo. Cytotoxic CD8+ T-lymphocytes scan these complexes by their T-cell receptor and eventually eliminate virally infected or malignantly transformed cells after specific recognition of MHC I molecules loaded with peptides derived from viral or oncogenic proteins (for reviews, see [1,2]).
TAP has evolved together with other factors of the adaptive immune system and is found in all higher (jawed) vertebrates. The presentation of endogenous peptides by MHC I molecules occurs in almost every nucleated cell (Figure 1). In professional antigen-presenting cells, such as dendritic cells, macrophages and B-lymphocytes, TAP is also critically involved in the cross-presentation of exogenously derived antigens, which is essential for the priming of CD8+ T-lymphocytes as well as the negative selection of developing T-cells in the thymus (for reviews, see [3–5]). In this process, antigens are taken up by pinocytosis or receptor-mediated endocytosis. Subsequently, the antigens are processed in the endosomes and loaded on to recycling MHC I molecules. Alternatively, antigens are translocated into the cytosol by a mechanism which still needs to be elucidated. After proteasomal processing, the peptides are translocated by TAP into the ER or post-ER compartment for loading to MHC I molecules.
Topology of TAP
Human TAP forms a heterodimer composed of TAP1 (ABCB2, 748 amino acids) and TAP2 (ABCB3, 686 amino acids), which both belong to subfamily B of ABC transporters. Both subunits are essential and sufficient for peptide transport into the ER lumen . Each subunit contains a TMD (transmembrane domain) followed by a cytosolic NBD (nucleotide-binding domain). On the basis of accessibility studies using a single cysteine-scanning approach, TAP1 and TAP2 comprise ten TMHs (transmembrane helices) each . TAP can be divided into a core complex, composed of the six C-terminal TMHs and the NBDs of TAP1 and TAP2, and the two unique N-terminal four-TMH-comprising domains (TMD0; Figure 2). Core TAP lacking both TMD0s is fully functional in respect of peptide binding and translocation . However, the core TAP complex does not interact with tapasin and is therefore impaired in forming a functional peptide-loading complex (see below), leading to a reduced MHC I cell-surface expression [8,9].
Peptide-binding site of TAP
The peptide-binding site of TAP has been mapped to the last cytosolic loop between TMHs 4 and 5 of the core complex and a stretch of 15 amino acids following TMH6 of both subunits . Additionally, a small region within CL1 (cytosolic loop 1) of TAP1, connecting TMH2 and TMH3 of the core complex, interacts with bound substrate  (Figure 2). TAP binds and transports most efficiently peptides with a length of 8–16 and 8–12 amino acids respectively [12,13]. The binding pocket of TAP features a high structural flexibility since 40-mer peptides and peptides with bulky side chains such as fluorophores, spin probes, chemical proteases or poly-lysine are also recognized by TAP [11,14–17]. In spite of the flexible nature of the binding site, only one peptide is bound with high affinity to TAP at a time, as demonstrated by fluorescence cross-correlation spectroscopy . The peptide specificity is restricted to the three N-terminal residues and the C-terminal residue . The specificity of the C-terminal residue matches between the proteasome, MHC I and TAP, suggesting a co-evolution of these structurally and functionally unrelated protein complexes. However, the length specificity and the preference for the N-terminal residues of these protein complexes are divergent. Peptides of variable length are delivered by TAP into the ER lumen and trimmed to 8–11-mers by amino exopeptidases ERAP (endoplasmic reticulum aminopeptidase) 1/2 to fit into the binding pocket of MHC I molecules . The sequence in between the ends of the peptide can be promiscuous in sequence and length, ensuring that a single transporter can translocate an almost unlimited pool of peptides.
The asymmetry of the motor domains
The NBDs of TAP are the motor domains, which convert the chemical energy of ATP into conformational changes, thereby driving peptide translocation across the membrane. For ATP hydrolysis, the NBDs have to dimerize in a head-to-tail fashion where two ATP molecules are enclosed at the dimer interface. As also seen in the case of other ABC transporters, the NBDs of TAP1 and TAP2 are functionally non-equivalent. The ATP-binding site I formed by the Walker A/B and H-loop from TAP1 and the C-loop (ABC signature motif) and D-loop of TAP2 displays the degenerate (non-consensus) site, since the conserved glutamate residue next to the Walker B motif and the invariant histidine residue of the H-loop are replaced by an aspartate and glutamine residue respectively. In addition, the canonical LSGGQ motif of the C-loop is modified to LAAGQ in TAP2. In contrast, ATP-binding site II forms the canonical site with all conserved consensus sequences. First evidence for the functional asymmetry came from mutational studies in which a chimaeric transporter containing two NBDs of TAP1 or TAP2 showed different transport activities . Furthermore, point mutations in motifs aligning site I interfered less with the transport activity of TAP than the equivalent mutations in site II [21–24]. Additionally, mutating the non-consensus C-loop of TAP2 to the canonical sequence created a TAP transporter with even higher transport activity than wild-type TAP . Despite this functional difference, both NBDs bind and hydrolyse ATP [25,26]. The difference in the nucleotide-binding behaviour was mapped to the non-homologous C-terminal ends of both subunits [27,28]. The nucleotide-binding state of TAP1 regulates the dissociation of MHC I molecules from the peptide-loading complex, a result, however, which has only been observed in the rat system . It is assumed that the consensus ATP-binding site II takes over the leadership in the catalytic cycle and ATP-binding site I has more a regulatory role in peptide transport.
The interaction of both NBDs during the transport cycle is underlined by the importance of an electrostatic interaction between an acidic residue in the Q-loop of one NBD and a basic residue in the C-loop of the opposite NBD. This intermolecular electrostatic network between both NBDs seems to be asymmetric and could be the basis for the cross-talk between both ATPase sites by closing one site, whereas the second site is open . At the same time, the importance of the aspartate residue next to the Walker B and the glutamine substitution in the H-loop was demonstrated using soluble NBD of human or rat TAP1 [31,32]. Exchanging both residues to the canonical one leads to a strong increase of ATPase activity. On the basis of further structural and biochemical studies, the function of these two residues in TAP was interpreted differently. The histidine residue of the H-loop may be directly involved in hydrolysis of the phosphoanhydride bond by stabilizing the transition state, whereas the glutamate residue orients the histidine residue for optimal catalysis . In the latter case, the hydrolytic water molecule is activated by the γ-phosphate (substrate-assisted catalysis) . Alternatively, the glutamate residue may be the catalytic base activating the hydrolytic water (general base catalysis), whereas the histidine residue stabilizes the transition state .
On the basis of the fact that ATP hydrolysis is strictly linked to peptide binding and transport [6,34], a tight coupling between the transmembrane domain and the NBD has to be postulated. Peptide binding seems to be sensed by a short stretch of amino acids in CL1 of TAP1. This peptide sensor (residues 282–288) changes its conformation during the transport cycle as shown by cycle-dependent peptide cross-linking to this site . This conformational change may represent the slow second step of the two-step peptide-binding reaction in which a quarter of all TAP residues are involved [35,36]. Remarkably, this peptide sensor corresponds to the EAA (Glu-Ala-Ala) motif found in bacterial ABC importers, which is involved in communication between the TMD and the NBD . Cysteine mutations in this sequence did not influence peptide binding, but disrupted peptide transport in TAP. This sensor was identified so far only in TAP1, pointing to the asymmetry of the subunits not only in the NBD but also in the TMD. On the basis of a homology model between core TAP and the putative multidrug ABC exporter Sav1866 from Staphylococcus aureus [37a], the transmission interface between the TMD and the NBD was assigned to CH (coupling helix) 1 and CH2, comprising residues 272–281 and 373–381 in TAP1, as well as residues 238–247 and 338–346 in TAP2 respectively . All of these helices display the membrane-distal part of the TMD and are in close contact with residues in the NBD. By oxidative cysteine cross-linking, the direct contact between CH1 and CH2 of TAP1 and the X-loop in the NBD of TAP2, which is in close proximity to the C-loop, was demonstrated . Furthermore, cysteine mutations in both loops affected TAP function differentially. Mutations in CH1 did not interfere with peptide binding, but with transport, whereas mutations in CH2 were mostly tolerated. The functional difference of both loops is also reflected in the observation that cross-linking of CH1 with the X-loop does not interfere with peptide binding, but inhibits peptide transport, whereas a covalent cross-link between CH2 and the X-loop still abrogates peptide binding . Since the peptide sensor and CH1 cover neighbouring sequences, it can be speculated that peptide binding induces a conformational change in the peptide sensor, which is predominantly transduced via CH1 to the X-loop and the Q-loop of the NBD. As a consequence, the ATP-loaded NBDs will dimerize, which could be the additional structural isomerization detected by a decreased lateral mobility of TAP in the ER membrane in the presence of peptide and ATP .
ATP-driven peptide transport
From biochemical and structural data of different ABC transporters and isolated NBDs, it is assumed that ATP functions as glue for the engagement of both NBDs (for reviews, see [40–42]). Subsequently, this dimerization turns the TMDs from an inward-facing to an outward-facing conformation in which the substrate can be released in the case of an ABC exporter. The present model of the transport cycle is illustrated in Figure 3. Some transporters, such as Pgp (P-glycoprotein) (ABCB1), seem to constantly undergo this non-productive cycle, since they show a high basal ATPase activity in the absence of substrates. In contrast, ATP hydrolysis of TAP is strictly coupled to peptide binding and no background ATPase activity is observed in the absence of peptides . Moreover, a tight quality control ensures that only peptides that are transported trigger ATP hydrolysis . Nevertheless, the longstanding question about the stoichiometry between solute transport and ATP hydrolysis has not yet been solved. An attractive possibility would be that the stoichiometry varies with the size or other properties of the substrate. This idea originates from bacterial ABC exporters of the PRT-HLY subfamily with haemolysin B as the most studied member, which translocate proteins up to 800 kDa in size in an unfolded manner. Since these substrates are larger than their transporters, it is hard to imagine that these polypeptides are transported in one step from an inward-facing to an outward-facing conformation. If there are transport intermediates, then the two-conformation model with an inward- and outward-facing conformation, probably with an additional occluded state in which the solute is buried in the transporter and not accessible from any site, is not valid for these transporters.
The macromolecular peptide-loading complex
As mentioned above, the heterodimeric TAP complex is involved in a dynamic protein network in which it resembles the central element in the peptide-loading complex responsible for the loading of MHC I molecules with high-affinity peptides. In the absence of a single factor of this loading complex, MHC I molecules are associated with low-affinity peptides displaying a decreased thermal stability and reduced MHC surface expression. Structural, as well as functional, studies have shed some light on the architecture of the peptide-loading complex. This macromolecular complex is composed of TAP1 and TAP2, tapasin, the oxidoreductase ERp57, the lectin chaperone calreticulin and MHC I molecules comprising the heavy chain and β2-microglobulin (for a review, see ). The stoichiometry of this complex is still under debate, as tapasin/TAP stoichiometries of 1:1, 1:2 and 1:4 have been found [44–46]. Both TAP subunits seem to bind the type I transmembrane glycoprotein tapasin independently from one another, since a complex composed of a wild-type and a core subunit still interacts with tapasin [8,47]. In tapasin, the interaction site is restricted to one flank of the transmembrane helix and part of the C-terminus placed in the cytosol . Beside its bridging function, tapasin enhances the thermostability of TAP  and increases the TAP expression level 3-fold in human and 100-fold in murine tapasin-deficient cells [50,51]. Complementary to its effect on TAP, tapasin stabilizes the peptide-loading complex by bridging all components and functions as peptide editor ensuring the loading of MHC class I molecules with high-affinity peptides (Figure 1).
TAP deficiency in genetic diseases
Patients with BLS I (bare lymphocyte syndrome I), in which one subunit of the TAP complex has a point mutation leading to a premature stop codon and thus to a truncated and non-functional protein, have a 30–100-fold reduced MHC I cell-surface expression compared with unaffected individuals . Strikingly, patients with this rare disease do not encounter severe viral infections and reach adult age. However, after a latency period of 4–7 years, patients suffer from recurrent bacterial infections of the upper respiratory tract. The lower respiratory tract as well is affected by infections in the progression of the disease. Routinely, Haemophilus influenzae, Streptococcus aureus, Klebsiella spp., Escherichia coli and Pseudomonas aeruginosa are the major causative pathogens. Additionally, half of the patients show necrotizing granulomatous lesions manifested in later childhood or at adult stage (for a review, see ). The proportion of γδ+ CD8+ T-cells in comparison with αβ+ CD8+ T-cells is much higher in TAP-deficient patients. The low abundance of αβ+ CD8+ T-cells, displaying the classical cytotoxic CD8+ T-cells during an adaptive immune response with high T-cell receptor diversity, results from the strongly decreased positive selection of this subclass of T-cells during T-cell development in the thymus caused by low cell-surface expression of MHC I because of TAP deficiency. In contrast, the selection of γδ+ CD8+ T-cells is not influenced by TAP deficiency, since this subset of T-cells recognizes predominantly non-peptidic antigens.
Remarkably, even lower levels of MHC I seem to be sufficient to select enough αβ+ CD8+ T-cells for a successful antiviral response. In addition, in some patients, αβ+ CD8+ T-cell clones could be identified, which recognize TAP-independent viral antigens . Antibody titres of viruses such as measles, mumps, influenza are normal. Interestingly, resting peripheral blood TAP−/− NK (natural killer) cells, which normally eliminate cells with low MHC I levels, show a hyporesponsiveness against host cells of the patients. However, after stimulation, these NK cells kill not only virally infected cells, but also normal autologous cells. Since in TAP-deficient patients, virus clearance is insufficient and delayed, neutrophils and T-cells show an increased stimulation. The high amount of proteolytic enzymes released by neutrophils together with the viral infection will damage especially the airways. Additionally, the cytokines and chemokines produced will attract and activate NK and γδ+ CD8+ T-cells for immune defence. However, these cells will also destroy bystander cells because of their low MHC I levels. This process will assist chronic alteration of the airway epithelium, promoting further bacterial infection leading to chronic inflammation.
Tumour development and TAP deficiency
During the progression of primary tumours, tumour cells develop sophisticated immune evasion strategies to escape the T-cell surveillance. In most cases, the tumour cells show a drastic decrease or even a loss of MHC I cell-surface expression, thereby evading the presentation of tumour-specific antigenic peptides (for reviews, see [54,55]). Besides the loss of MHC I caused at the transcriptional level or by gene mutations, TAP activity is also a target to escape immune surveillance [54,56]. Down-regulation of TAP expression is associated with disease progression and correlates with malignant melanoma with reduced survival . TAP expression can be restored in tumour cell lines by transfection of tap genes or by incubation with interferon-γ or the histone acetylation inhibitor trichostatin A [58,59]. In contrast, TAP expression is suppressed by IL-10 (interleukin 10) , which is of clinical relevance, since many tumours secrete IL-10. As TAP function can be restored by cytokine treatment, defects seem most likely to result from regulation failures. Apart from transcriptional regulation, the stability of the mRNA of TAP is also decreased in some tumour cells . In addition, mutations in the tap gene have been reported, which lead to a truncated and inactive TAP complex [62,63]. Different strategies are explored to prime an immune response against TAP-deficient tumours. For example, mice injected with TAP-transformed tumour cells showed a much higher survival rate than control animals .
Another concept is based on the activation of cytotoxic T-cells recognizing peptides bound to MHC I of TAP-deficient cells. These TAP-independent epitopes are derived from widely expressed self-proteins and are only presented by MHC I on the surface of cells with impaired antigen-processing function . For activation of T-cells specific for these epitopes, B6 wild-type mice were vaccinated with bone-marrow-derived dendritic cells of TAP−/− or wild-type mice and subsequently challenged with the TAP2-deficient RMA-S lymphoma or TAP1-deficient MCA fibrosarcoma cell line. In both cases, the TAP-deficient vaccination induced the activation of cytotoxic T-cells and drastically increased the survival rate of the mice in comparison with the animals inoculated with wild-type dendritic cells. In a further approach, the same group could show that T-lymphocytes specific for TAP-independent epitopes could also be activated by inoculation of mice with wild-type dendritic cells retrovirally transduced with the TAP inhibitor UL49.5. The advantage of this strategy seems to lie in the residual level of MHC I molecules on the cell surface, which elongates the lifetime by inhibiting the elimination by NK cells .
Viral escape mechanisms to shut down TAP function
To cope with the host's immune system, viruses have developed sophisticated strategies to escape its surveillance. Almost every step in the MHC I antigen-presentation pathway is targeted by viral factors. Meanwhile, different viral TAP inhibitors, almost all encoded by large double-stranded DNA viruses, are discovered that inhibit TAP function by entirely different strategies. By specific binding to TAP, mk3 of murine γ-herpesvirus-68 induces the degradation of TAP, tapasin and MHC I, whereas the E3/19k protein disrupts the MHC I/tapasin interaction with TAP, thus preventing efficient loading of antigenic peptides on to MHC I molecules. So, these two factors interfere with the assembly of the peptide-loading complex rather than directly inhibiting peptide translocation into the ER lumen. The latter strategy is discussed below (Figure 4).
The immediately-early gene product ICP47 of HSV (herpes simplex virus) type-1 and -2 inhibits peptide binding to human TAP. ICP47 of HSV-1 is an 88-amino-acid cytosolic protein, which binds with nanomolar affinity to human TAP [67,68]. The interaction of ICP47 induces a TAP conformation different from the peptide-bound state . The recognition mechanism by TAP must be different from unfolded peptides since the N- and C-termini of ICP47 can be modified, in contrast with peptides . Upon membrane association and before interaction with TAP, ICP47 undergoes a conformational change from an unstructured to a helix–loop–helix conformation [71,72]. The affinity of ICP47 to detergent-solubilized TAP is drastically decreased in the absence of the membrane environment. Therefore the membrane might have a dual function in inducing the active conformation of ICP47 as well as concentrating and localizing the inhibitor for efficient inhibition of TAP.
The type I transmembrane glycoprotein US6 of HCMV (human cytomegalovirus) impairs the peptide supply into the ER lumen by inhibiting ATP binding to TAP [73–75]. Since the ER-luminal domain of US6 is sufficient for the inhibition of ATP binding to TAP from the cytosol, an allosteric communication across the ER membrane and arrest of the antigen translocation complex is suggested [76,77]. US6 binds to several ER-luminal loops of TAP1 and TAP2 assembled in a pre-formed functional heterodimeric complex; whereas the interaction with both subunits is necessary to efficiently inhibit TAP activity .
The viral factor BLNF2a from the EBV (Epstein–Barr virus) blocks peptide and ATP binding to TAP. Because BLNF2a (60 amino acids) has a stretch of 20 hydrophobic amino acids at the C-terminus and because it does not contain a signal sequence for ER import, we propose that it belongs to the class of TA (tail-anchored) proteins. Insertion of BLNF2a into the ER membrane is TAP-independent. However, the TAP complex has a strong stabilizing activity , probably by protecting the viral factor from proteasomal degradation.
A very heterogeneous inhibition pattern is found in the viral factor UL49.5, which can be found in all herpesviruses sequenced today. Only UL49.5 from BHV (bovine herpesvirus), PRV (pseudorabiesvirus) and EHV (equine herpesvirus)-1 and -4 inhibit TAP function. All of these factors arrest TAP in a transport-incompetent conformation, but only the BHV inhibitor drives TAP into proteasomal degradation. UL49.5 of EHV-1 and EHV-4 inhibits ATP binding of TAP, whereas PRV does not impair ATP or peptide binding .
• TAP is a peptide transporter essential in the adaptive immune system.
• TAP is part of the peptide-loading complex which is crucial for efficient transport and loading of peptides on to MHC I molecules.
• The core complex is essential and sufficient for peptide translocation, whereas the unique TMD0 mediates the interaction with tapasin.
• Cytosolic loops of the TMD of TAP1 and TAP2 are involved in peptide binding, peptide sensing and signal transduction to the NBDs.
• The NBDs are asymmetric in sequence and function.
• Genetic diseases, tumour escape strategies as well as viral factors target TAP function.
- © The Authors Journal compilation © 2011 Biochemical Society