The Nonclassical Immune Surveillance for ERAAP Function

Abstract

The peptide repertoire presented by MHC class I molecules on the cell surface is essential for the immune surveillance of intracellular pathogens and transformed cells. The generation of this peptide repertoire is critically dependent on the endoplasmic reticulum aminopeptidase associated with antigen processing (ERAAP). Loss of ERAAP function leads to the generation of a profoundly disrupted peptide repertoire including many novel and immunogenic peptides. Strikingly, a large fraction of these novel peptides on ERAAP-KO cells are presented by the nonclassical MHC Ib molecule, Qa-1^b. One immunodominant Qa-1^b-restricted novel peptide is recognized by a unique CD8⁺ T cell population showing features of both conventional cytotoxic T cells and unconventional innate-like T cells. While much remains to be uncovered, here we summarize the latest discoveries of our lab on the important immune surveillance of ERAAP function mediated by nonclassical MHC Ib molecules and their unusual cognate T cells.

Introduction

A broad repertoire of peptides is constitutively presented by MHC I molecules on the cell surface. Through these peptide MHC I (pMHC I) ligands, the antigens within the cells, otherwise inaccessible, become ‘visible’ for screening by CD8⁺ T cells [1]. Abnormal cells, such as those infected by viruses or bearing mutations, display an altered peptide repertoire. With the foreign or mutated antigens presented on the cell surface, these potentially dangerous cells are efficiently detected and eliminated by the immune system.

The peptide repertoire of the cell is maintained through the antigen processing and presentation pathway. This mechanism allows for the generation of a diversity of the ‘ideal’ peptides for the MHC I molecule. The pathway starts in the cytoplasm where the protein precursors are fragmented by the proteasome. The peptide fragments are then transported into the endoplasmic reticulum (ER) via the peptide transporter associated with antigen processing (TAP). In the ER, these peptides are further customized to the optimal length and residue composition for the MHC I molecule. The final peptide products, being loaded onto the MHC I molecule with the help of several chaperone proteins, are then shuttled onto the cell surface [2, 3]. Each step of this pathway must be carefully monitored, since failure at any level can lead to significant immunological consequences [4]. For example, inhibition of the proteasome blocks the generation of peptide precursors and interferes with the assembly of the pMHC I complexes [5]. Disruption of the proteinase subunit LMP2 reduces the presentation of influenza virus antigen and impairs the viral specific T cell response in mice [6]. TAP deficiency severely impairs the generation of the pMHC I complexes on the cell surface, which further leads to the absence of CD8⁺ T cells in TAP-KO mice [7]. Similarly, loss of the ER chaperone Tapasin results in defects in the pMHC I surface expression and CD8⁺ T cell function in mice [8]. Many viruses have in turn evolved means to evade immune detection by inhibiting components of the antigen processing pathway. The Epstein-Barr virus (EBV) interferes with pMHC I complex generation by specifically inhibiting proteasomal processing [9]. Herpes simplex virus (HSV) ICP47 protein and human cytomegalovirus (HCMV) US6 protein can block peptide translocation and pMHC I generation by interacting with TAP [10, 11]. The HCMV US3 protein binds to Tapasin and interferes with the MHC I peptide loading process [12]. In addition to these well-known proteins, recent discoveries of novel key molecules further completed the map of antigen presentation pathway.

ERAAP (or ERAP1 in humans), the ER aminopeptidase associated with antigen processing, plays a crucial role in the final peptide customization step by trimming the N-terminal extensions from the antigenic peptide intermediates [13]. The function of ERAAP has been extensively studied since its identification by our lab. The critical role of ERAAP in the antigen processing pathway and the whole immune system was shown in our previous studies of cells and animals deficient in ERAAP. We found that loss of ERAAP function leads to a profoundly altered peptide repertoire, which generates a potent immune response in the wildtype animal [14]. The significance of appropriate ERAAP (ERAP1) function is also evident through an increasing number of associations between this molecule and disease states. For example, polymorphisms of ERAP1 and the homologous ERAP2 are associated with several autoimmune diseases, such as ankylosing spondylitis (AS), Birdshot chorioretinopathy, and psoriasis [15-17]. Altered ERAP1/2 protein levels are also associated with viral infection and poor tumor prognosis [18, 19]. The HCMV encoded microRNA miR-US4-1 downregulates ERAP1 expression during viral infection, which in turn leads to reduced presentation of HCMV derived peptides, and an impaired T cell response [19]. In our study of the immune response against the parasite Toxoplasma gondii, we found that loss of ERAAP function impairs expansion of the protective CD8⁺ T cell populations as a result of reduced presentation of an immunodominant peptide HPGSVNEFDF (HF10) by H-2L^d [20]. Clearly, disruption of ERAAP leads to profound immunological consequences.

Given the critical role of ERAAP in generating properly targeted immune responses, it’s perhaps not surprising that the immune system has also evolved mechanisms that monitor for appropriate ERAAP function. Here we review the latest research of our lab on the immune surveillance of ERAAP function. We will focus on the function of the nonclassical MHC Ib molecule Qa-1^b, analysis of the global peptide repertoire associated with ERAAP deficiency, the immunodominant ligand Qa-1^b-FL9 (QFL) presented on ERAAP-KO cells, and unique features of the CD8⁺ T cells that specifically recognize the QFL complex (Fig.1).

Fig 1.

The unconventional immune surveillance mechanism of ERAAP function.

ERAAP selectively affects the peptides being presented on the cell surface. The peptide repertoire of ERAAP deficient cells is composed of four groups of peptides: 1) the ERAAP-independent peptides (Ei-p) which are unaffected by ERAAP function; 2) the ERAAP-dependent peptides (Ed-p) that are absent on ERAAP-ko cells as compared with the wildtype peptide repertoire; 3) the ERAAP-sensitive peptides (Es-p) which are presented at significantly higher levels; 4) the ERAAP-unedited novel peptides (Eun-p) that are unique to the ERAAP-ko cells. The Eun-p contains longer peptides that are presented by both the classical MHC Ia molecule and the nonclassical MHC Ib molecule. The amount and diversity of peptides associated with the Qa-1^b molecule increases robustly on the ERAAP-ko cells. This altered peptide repertoire is highly immunogenic. The QFL-T cells, a unique CD8⁺ T cell population that specifically recognizes the immunodominant ligand Qa-1^b-FL9 presented on the ERAAP-ko cells, are relatively abundant in the naïve wildtype mice. These cells show some innate-like features including a CD44^hiCD122⁺ antigen experienced phenotype and the expression of a semi-invariant TCR α-chain(Vα3.2Jα21). QFL-T cells, together with the other as yet undefined T cells that are specific for the ligands presented on the ERAAP-ko cells, closely monitor for normal ERAAP function.

Qa-1^b, a versatile nonclassical MHC Ib molecule

In addition to the extensively-studied classical MHC Ia molecules, the important roles of the nonclassical MHC Ib molecules are now increasingly recognized. The murine MHC Ib molecule Qa-1^b (HLA-E in humans) is distinguished from the classical MHC Ia molecules in different aspects. Despite their structural similarity, the Qa-1^b molecule is monomorphic, tissue-restricted in expression, and functionally unique [21]. The Qa-1^b molecules are predominantly, but not exclusively, occupied by a nonamer peptide (AMAPRTLLL) derived from the leader sequence of the classical MHC Ia molecule, named Qa-1 determinant modifier (Qdm) [22]. The Qa-1^b-Qdm complex interacts with the CD94/NKG2 receptor family. Depending on which NKG2 protein is engaged, the outcome can be either inhibitory or stimulatory, although the NKG2A mediated inhibitory effect is believed to be predominant in most cases [23, 24]. The Qa-1^b molecule thus serves as a modulator for NK or CD8⁺ T cell function.

While the Qdm peptide is predominant, there are emerging reports of the Qa-1^b molecule presenting peptides derived from a variety of other sources under different conditions. In cells deficient for MHC Ia, the source of the Qdm peptide, Qa-1^b can present a heat shock protein 60 (HSP60)-derived peptide (GMKFDRGYI). The presentation of this peptide is enhanced on stressed macrophages which showed induced HSP60 expression [25]. Interestingly, on Salmonella infected cells, an analogous peptide (GMQFDRGYL) from the homologous Salmonella GroEL protein was presented by Qa-1^b. A Salmonella-specific cytotoxic T-lymphocyte (CTL) clone cross-reacts to both the HSP60 and the Salmonella GroEL derived peptide ligands [26]. In a study of type 1 diabetes, a pancreatic β-cell line showed upregulation of surface Qa-1^b expression in response to IFN-γ treatment. A preproinsulin leader sequence derived peptide (ALWMRFLPL) presented by Qa-1^b was identified as an immunogenic ligand recognized by a Qa-1^b-restricted CTL clone [27]. Assessment of the T cell response in pork insulin immunized nonresponder H-2b mice revealed that Qa-1^b can also present soluble insulin derived peptide in a TAP-independent manner [28]. In addition, a population of CD8αα-expressing TCRγδ intestinal intraepithelial lymphocytes (iIELs) was found to expand in a Qa-1^b dependent manner during the early response to Salmonella infection [29].

In contrast to its role in inducing immune responses, another interesting aspect of Qa-1^b function is its involvement in the prevention of experimental autoimmune encephalomyelitis (EAE) through activation of CD8⁺ T regulatory cells (Tregs). Application of the superantigen staphylococcus enterotoxin B (SEB) or vaccination of animals with the autologous myelin basic protein (MBP)-specific CD4⁺Vβ8⁺ T cell were both shown to protect mice from EAE. This protective effect was a result of the suppression of activated CD4⁺Vβ8⁺ T cells by CD8⁺ Tregs in a Qa-1 dependent manner. The proposed model is that activation of these pathogenic CD4⁺ T cells leads to the presentation of peptide(s) derived from the TCRβ chain by the Qa-1 molecule. The ligand further activates the Qa-1 restricted CD8⁺ Tregs, which function to maintain immune homeostasis [30-33]. Indeed, one study reported that a population of CD8 αα⁺ TCRαβ Tregs recognizes a peptide derived from the complementary determinant region 2 (CDR2) of the Vβ8.2 (GLRLIHYSY) presented by the Qa-1^a molecule on the CD4⁺ T cells [34]. These studies imply that through presentation of a diverse repertoire of peptides, the Qa-1^b molecule could be engaged with different T cells and potentially play versatile roles in defense against infectious diseases, stress conditions, and the development of certain autoimmune diseases. Collectively, it is evident that Qa-1^b has versatile functions extending beyond simple presentation of the Qdm peptide, and its role may be pro- or anti-inflammatory responses in different contexts. Based on our study of the peptide repertoire on the ERAAP deficient cells, we believe that the Qa-1^b molecule presents far more peptides than has previously been appreciated.

The peptide repertoire of classical and nonclassical MHC I is shaped by ERAAP

Early observations of the ERAAP knockout mice demonstrated a marked disruption of the natural peptide repertoire [14]. The altered peptide repertoire presented in the absence of ERAAP is composed of four groups of peptides: the peptides that are unaffected (ERAAP-independent, Ei-p), peptides that are missing from the repertoire (ERAAP-dependent ,Ed-p), significantly overrepresented peptides (ERAAP-sensitive, Es-p), and novel peptides unique to the ERAAP depletion (ERAAP-unedited novel, Eun-p) [35, 36]. While the variations among the four groups of peptides associated with ERAAP function are all interesting, the Eun-p was the most unanticipated discovery. High-throughput mass spectrometry and bioinformatics analysis of the peptides eluted from the surface of wildtype, ERAAP-KO, and ERAAP-Qa-1^b-KO bone marrow derived dendritic cells (BMDCs) further revealed numerus peptides associated with the classical MHC Ia and nonclassical MHC Ib (Qa-2^a, Qa-1^b) in the ERAAP deficient cells. Strikingly, the Qa-1^b molecule presents ten-fold more peptides on the ERAAP-KO cells compared to wildtype cells. Qa-1^b is the only MHC I molecule associated with a significantly increased number of the peptides unique to ERAAP-KO cells [37]. These observations reinforced the potential of the Qa-1^b molecule in presenting a broad range of peptides, and thus being versatile in function.

Peptide length and residue composition is crucial for proper MHC I binding and presentation. Thus, peptides unique to either the wildtype or the ERAAP-KO cells and those common to both were further compared for length and sequence. Assessment of their length showed distinctive differences between the ERAAP-KO and wildtype specific peptides. Compared to the canonical H2-K^b or H2-D^b presented peptides which contain eight or nine residues, on ERAAP-KO cells, both the classical MHC Ia and nonclassical MHC Ib molecules tend to present longer peptides. Comparison of the N-terminal pattern of these peptides revealed that the N-termini of the ERAAP-KO specific peptides are similar to the peptides common to both strains, whereas the wildtype specific peptides showed distinctive residue preference enriched for certain hydrophobic residues and with reduced representation of serine residues. This indicates that ERAAP edits a particular subset of the peptide precursors. Analysis of the source genes for the WT or ERAAP-KO associated peptides showed that when ERAAP mediated processing in the ER was compromised, proteins in the Golgi apparatus or those containing leucine-rich repeats (LRR) became the preferred source for peptides presented on the cell surface. These observations indicate that ERAAP deficiency leads to the presentation of novel peptides derived from different protein sources instead of the N-terminally extended versions of the wildtype specific peptides.

Among the ERAAP associated peptides, a novel immunogenic peptide DPSVHLLTA (DA9) was identified. This peptide is derived from an LRR-containing protein, FBXL19 [37]. Interestingly, in genome-wide association studies, the encoding gene of FBXL19 was shown to be associated with psoriasis, one of the autoimmune diseases related to ERAP1/2 polymorphism [16, 38]. It remains unknown whether this peptide is presented by a classical or nonclassical MHC I molecule. Research on this ligand and any corresponding ligand specific T cells might provide insight on the mechanism of psoriasis development.

Notably, deficiency of TAP, another critical molecule for antigen processing, could similarly generate a novel immunogenic peptide repertoire presented through Qa-1^b [39]. Interestingly, in contrast to the classical MHC Ia molecules which are poorly expressed on the cell surface in the absence of TAP, the expression of Qa-1^b/HLA-E molecules is less affected [44]. Similar to our observation of the ERAAP-KO associated peptide repertoire, the Qa-1^b dominant peptide Qdm was largely replaced by novel peptides upon TAP deficiency. Earlier studies showed that the novel peptides unique to TAP-deficient cells are mainly generated through cleavage of the N-terminal signal sequence or the C-terminus of proteins by signal peptide peptidase (SPP) in the ER [40-42]. These alternative antigen processing routes allow for the generation of sufficient pMHC I complexes for the immunosurveillance of compromised conventional antigen processing under conditions such as virus infection, cell transformation or stress in the microenvironment [43]. Although it remains to be assessed to what extent the Qa-1^b associated peptide repertoires on the ERAAP-KO and TAP-KO cells overlap, it is clear that Qa-1^b serves as a general indicator for antigen processing defects, at least at the level of the ER.

QFL, an immunodominant ligand on ERAAP-KO cells

Cross immunization experiments between wildtype and ERAAP-KO animals confirmed that the altered peptide repertoire on ERAAP-KO cells was immunogenic and elicited a robust CD8⁺ T cell response mediated by both classical MHC Ia and nonclassical Qa-1^b molecules [37, 45]. These observations indicate that loss of ERAAP function results in the presentation of highly immunogenic peptides through Qa-1^b. By screening a cDNA library prepared from ERAAP-KO cells using an ERAAP-KO specific CTL clone, BEko8Z, a nonameric peptide FL9 (FYAEATPML) derived from a highly conserved protein FAM49B was identified as an immunodominant peptide presented by the Qa-1^b molecule on ERAAP-KO cells [45] . FAM49A, the FAM49B homolog, is another candidate source for the peptide (unpublished data). So far, little is known about the function of the source protein FAM49B. It was recently reported that FAM49B is involved in the inhibition of T cell activation and protection against Salmonella infection through cytoskeletal remodeling [46, 47] . The protein is also associated with the suppression of tumor metastasis through regulating mitochondria function [48]. The mechanism of how FAM49B is processed to generate the FL9 peptide in the ERAAP-KO cells and its functional implications remains to be elucidated. Further studies on how and where Qa-1^b binds to the FL9 peptide, association between unique peptides presented by Qa-1^b, and defects of other components of the antigen processing pathway will reveal the fundamental molecular mechanism for Qa-1^b mediated immune surveillance.

The unique QFL-T cells

Unlike the Qa-1^b-Qdm ligand, which is recognized by the CD94/NKG2 molecules, the QFL ligand is recognized by the TCR of CD8⁺ T cells. These CD8⁺ T cells specific for the QFL ligand are thus named QFL-T cells [45]. Analysis of the QFL-T cells revealed that these cells showed unusual features of both conventional and unconventional innate-like T cells. The unconventional T cells, including invariant NKT (iNKT) cells, mucosal associated invariant T cells (MAIT cells) and γδ T cells, are groups of T cells that recognize unusual ligands, reside in non-lymphoid tissues, and show atypical MHC-restricted T cell phenotypes. iNKT cells are specific for the glycolipid a-galactosylceramide (a-GalCer) presented by CD1d [49], MAIT cells typically recognize the microbial-derived intermediates of vitamin B2 (riboflavin) presented by MR1 [50], whereas γδ T cells are mostly not MHC restricted. Unlike conventional T cells which mostly reside in the lymphoid tissues, iNKT cells are present at very high frequency in the mouse liver [49]; MAIT cells are mainly localized in the tissues colonized by microbiota, such as skin, lung, or small intestine [50, 51]; and γδ T cells are highly abundant in the skin and the large intestine. Both iNKT cells and MAIT cells express an invariant TCRα chain paired with a limited set of TCRβ chains [49, 50]. Functionally, these cells are believed to be a link between the innate and adaptive immune systems [52, 53].

We characterized the QFL-T cells based on the above features of the unconventional T cells. By labeling the splenocytes with QFL multimers, we found that similar to some unconventional T cells, QFL-T cells are present at a relatively high frequency (1 in 10,000 CD8⁺ T cells) even in naïve wildtype mice. A large fraction of these cells showed a CD44^hi CD122⁺ antigen experienced phenotype. Comparison of the QFL-T cell number and phenotype in Qa-1^b-KO and wildtype mice suggests that development of the QFL-T cells may not be exclusively dependent on the presence of Qa-1^b molecule, whereas the memory phenotype is likely to be imprinted by the exposure to Qa-1^b associated ligand [45]. Strikingly, assessment of TCR sequences showed the majority of QFL-T cells express an invariant α-chain Vα3.2Jα21 (TRAV9d-TRAJ21) [54]. Interestingly, a recent study of the Qa-1^b-restricted CTL clones which specifically recognize TAP-KO cells showed that similar to QFL-T cells, these T cell clones use an invariant TCR α-chain. This suggests that certain TCRs might predispose developing T cells to Qa-1^b restriction [55]. Analysis of the master transcriptional factors in the QFL-T cells showed these cells are T-bet⁺ PLZF⁻ RORγT. These results, together with the observation that activated QFL-T cells produce IFNγ and are cytotoxic to ERAAP-KO cells, suggest that functionally the QFL-T cells might resemble more conventional CD8⁺ CTLs [54].

The many potentials of QFL-T cells

In addition to the features of QFL-T cells we have so far uncovered, many outstanding and interesting questions remain to be answered and many potentials remain to be explored. First, how do the QFL-T cells acquire their unusual features? One possibility is that maturing T cells with the appropriate semi-invariant TCR receptors are imprinted with a memory phenotype through a unique developmental pathway in the thymus. It is well established that some other unconventional T cells, such as iNKT cells, are ‘born’ with memory through a thymic positive selection process [56]. Alternatively, expansion of naïve QFL-T precursors and subsequent memory phenotype acquisition could be a result of exposure to the QFL ligand in the periphery. It is possible that the QFL ligand is transiently expressed on the cell surface under conditions that interfere with ERAAP function, such as infection with pathogens seeking to evade classical immune surveillance by inhibiting ERAAP, stress conditions, or exposure to certain commensal bacteria.

A related question is the overall tissue distribution of QFL-T cells. It is clear they are present with a high frequency in the spleen of naïve wildtype mice. However, as is the case with other unconventional T cells, there may be an even higher enrichment of QFL-T cells in certain nonlymphoid tissues such as the liver, GI tract, or skin, particularly if peripheral antigen exposure is driving their localization, retention, and/or expansion.

The last and the most critical question is what are the functions of the QFL-T cells? They are undoubtably effector T cells that closely recognize alterations in ERAAP function. Given their cytotoxic abilities, QFL-T cells might be an important innate-like rapid responder seeking to maintain homeostasis by eliminating cells with defective ERAAP activity. In the context of an immune response, QFL-T cells could also play important regulatory roles analogous to the previously described Qa-1^b restricted CD8⁺ Tregs. In addition, the possibility of unique tissue-restricted functions of QFL-T cells remains to be assessed.

Conclusions and future perspectives

The trimming and customization of peptide precursors by the ER aminopeptidase ERAAP is a critical step in the MHC I mediated antigen presentation pathway. Loss of ERAAP function results in global quantitative and qualitative disruptions of the peptide repertoire presented by both classical and nonclassical MHC I molecules, including the presentation of numerous unedited novel peptides. Strikingly, these novel peptides, many of which are presented by nonclassical MHC Ib molecules, are immunogenic and can be recognized by CD8⁺ T cells. The function of ERAAP is closely monitored through recognition of the Qa-1^b-FL9 complex presented on ERAAP deficient cells by a unique population of CD8⁺ T cell population we have termed QFL-T cells. These unusual QFL-T cells share several features with unconventional innate-like T cells while maintaining the cytokine secretion and cytotoxic capabilities of conventional CD8⁺ CTL. The similarities between the QFL-T and these innate-like T cells, prompt the idea that QFL-T cells might also have an unusual tissue distribution pattern and carry out critical functions at the ‘border’ of innate and adaptive immunity through the immune surveillance of ERAAP. It is quite impressive that the nonclassical MHC Ib molecule plays an important role in monitoring defects in the antigen processing pathway in mice. In humans, it is not yet known if similar unconventional T cell populations surveil for defects in ERAP1/2 function or other defects in antigen processing via the Qa-1^b homolog HLA-E. As the two human ER aminopeptidases have distinct peptide sequence and length preferences, their roles in this immunosurveillance pathway might diverge. Analysis of HLA-E bound peptides in the absence of ERAP1 and/or ERAP2 might identify potential targets for unconventional T cell responses. Through our continued study of ERAAP and QFL-T cell function, we will gain more insights into pMHC Ib mediated immune surveillance of antigen processing and its importance in disease states.