Leukocyte Ig-Like Receptors – A Model for MHC Class I Disease Associations

MHC class I (MHC-I) polymorphisms are associated with the outcome of some viral infections and autoimmune diseases. MHC-I proteins present antigenic peptides and are recognized by receptors on natural killer cells and cytotoxic T lymphocytes, thus enabling the immune system to detect self-antigens and eliminate targets lacking self or expressing foreign antigens. Recognition of MHC-I, however, extends beyond receptors on cytotoxic leukocytes. Members of the leukocyte Ig-like receptor (LILR) family are expressed on monocytic cells and can recognize both classical and non-classical MHC-I alleles. Despite their relatively broad specificity when compared to the T cell receptor or killer Ig-like receptors, variations in the strength of LILR binding between different MHC-I alleles have recently been shown to correlate with control of HIV infection. We suggest that LILR recognition may mediate MHC-I disease association in a manner that does not depend on a binary discrimination of self/non-self by cytotoxic cells. Instead, the effects of LILR activity following engagement by MHC-I may represent a “degrees of self” model, whereby strength of binding to different alleles determines the degree of influence exerted by these receptors on immune cell functions. LILRs are expressed by myelomonocytic cells and lymphocytes, extending their influence across antigen-presenting cell subsets including dendritic cells, macrophages, and B cells. They have been identified as important players in the response to infection, inflammatory diseases, and cancer, with recent literature to indicate that MHC-I recognition by these receptors and consequent allelic effects could extend an influence beyond the immune system.

Thus, many studies have examined the nature of the peptides presented by disease-associated alleles and of T cell responses restricted by these alleles (7,8). For example, a number of studies have examined the peptide specificities of HLA-B27 subtypes (9). In the context of HIV infection, a dominant HLA-B27 restricted viral peptide is thought to play a key role in the association of this allele with control of infection. Immune escape from the response against the dominant peptide results in a decrease in HIV-1 replication (10).
In humans, classical MHC-I are also recognized by members of the killer Ig-like receptor (KIR) family, which are encoded in the leukocyte receptor complex (LRC) on chromosome 19. KIR demonstrate allele (and in some cases, peptide) specificity (11), albeit at a lower level of precision for individual peptide/MHC complexes than that shown by classical TCR. KIR are expressed on natural killer (NK) cells and T cells where they inhibit the ability of these cytotoxic cells to lyse target cells that express self MHC-I alleles. As knowledge regarding their biology and MHC specificities has grown, KIR has been studied alongside MHC-I in conditions such as spondyloarthropathy, HIV, and HCV infections (5,12,13). There is considerable variation in KIR haplotypes such that any individual may not carry the relevant MHC ligand for every KIR receptor that they express and vice versa. A number of studies suggest that particular combinations of KIR and HLA alleles, believed to result in functional receptor/ligand interactions, are associated with protection from progression to AIDS following HIV infection (14).
A lesser-studied family of proteins encoded within the LRC is also capable of recognizing MHC class I. These leukocyte Ig-like receptors (LILR) do not appear to be involved in the cytolytic removal of targets bearing non-self MHC-I protein complexes (15). Instead, they are predominantly expressed on cells of the myelomonocytic lineage, and some of them show a broad specificity encompassing both classical and non-classical MHC-I (16). The observation that LILR vary in the strength of their binding to individual MHC-I alleles, however, raised the possibility that these innate immune receptors may contribute in some manner toward MHC-I disease associations (17). In support of this theory, a recent study of a large cohort of HIV-1 infected patients demonstrated that the overall binding strength of LILRB2 for the MHC-I haplotypes expressed by these individuals was positively associated with the level of viremia (18).

LeUKOCYTe ig-LiKe ReCePTORS
The various members of the LILR family are broadly categorized as inhibitory (LILRB) or activating (LILRA), according to the presence or absence of tyrosine-based signaling motifs in their cytoplasmic tail. In some cases, putative activating receptors have been shown to elicit inhibitory effects and vice versa for inhibitory receptors (19). Receptor engagement results in intracellular phosphorylation of the tyrosine-based motifs within the receptors themselves (LILRB) or on associated adaptor molecules (LILRA) (19). Downstream signaling events can be mediated by phosphatases such as SHP-1, SHP-2, and SHIP (20,21) and vary according to the receptor and/or cellular context. For example, SHP-2 may mediate production of IL-6 via the NF-kB pathway following LILRB2 engagement on dendritic cells (22) or inhibition of the mTOR pathway following LILRB1 engagement on T lymphocytes (23).
There are multiple similarities between KIR and LILR in terms of Ig domain-based structure, gene location within the LRC, and ability to recognize MHC-I (15). Unlike their NK receptor counterparts, however, LILR orthologs (known as PIR) are found in rodents, where they demonstrate similar ligand binding, expression, and functional profiles (24,25). This may indicate a higher degree of evolutionary conservation for LILR than for KIR, with bovine orthologs also identified (26) and similar proteins documented in chickens and fish (27,28). Within the murine system, there is a single inhibitory receptor, PIR-B, and multiple activating receptors (PIR-A). PIRs are involved in the regulation of lymphocyte, antigen-presenting cell, and granulocyte functions (29), and their study has enabled the identification of functions for both these receptors and their human counterparts, such as the regulation of synaptic plasticity (30) and platelet activation by PIR-B and LILRB2 (31). Figure 1 shows the known expression profiles of LILR on leukocyte subsets according to current literature. The known expression profiles for LILR are not exhaustive; expression of individual members of the family has been documented for macrophages, B-cells, NK cells, and other non-immune cells (32)(33)(34)(35)(36)(37)(38)(39)(40). These receptors are, therefore, likely to have far-reaching effects on a range of immunological functions. Immune cells, which have yet to be characterized in full for LILR expression, include invariant NK (iNKT), gamma delta (γδ), regulatory (Treg) and T helper 17 (Th17) T-cells, B-cell subsets, as well as the various APC subsets and granulocytes.
Leukocyte Ig-like receptor activity can result in the upregulation or downregulation of both innate and adaptive functions with a range of effects on different cell types. For example, LILR and PIR have been shown to inhibit TLR-mediated functions of antigen-presenting cells such as inflammatory cytokine secretion (38,(41)(42)(43). Inhibitory LILR have been shown to inhibit the upregulation of co-stimulatory proteins on antigen-presenting cells (36,(44)(45)(46), thus favoring regulatory T cell responses (47)(48)(49)(50). On lymphocytes, inhibitory LILR have been shown to inhibit T and B cell receptor signaling and downregulate antibody and cytokine production (51-53). Activating LILR have been shown to mediate monocyte activation and secretion of inflammatory cytokines (54) and on basophils to trigger release of histamine (55).

MHC ReCOGniTiOn BY LiLR
Following the initial identification of LILRB1 as a receptor for self and viral MHC-I (56), structural studies predicted that several other members of the family would also recognize MHC-I (57). Members of the family were allocated into two groups on this basis, with Group 1 containing receptors predicted to bind MHC-I and Group 2 containing receptors that were not predicted to bind MHC-I (57). It was confirmed subsequently that the Group 1 members LILRA1, LILRA2, LILRA3, LILRB1, and LILRB2 can engage MHC-I (17,58 the broadest specificity, with the ability to recognize all classical and non-classical self MHC-I alleles and forms tested to date. Although LILRB2 binds to both the α3 and β2m regions of the MHC-I antigen-presenting structure, the major portion of its binding site lies within the highly conserved α3 domain (59). The degree of interaction between this receptor and the α3 domain is sufficient to allow LILRB2 to bind open conformers of MHC-I, which lack β2m. In contrast, the major LILRB1 binding site lies within β2m, thus this receptor can only associate with β2massociated MHC-I. Recognition of open MHC-I conformers has also been observed for LILRA1 and LILRA3, which were shown in one study to have stronger binding to open confomers than to β2m-associated MHC-I (17). These findings indicate that alternatively folded forms of MHC-I may play a functional role in the immune response. It is also important to note that the members of the LILR family may interact in cis with MHC-I on the cell surface, as been demonstrated for PIR-B and LILRB1 (60,61). Despite their broad specificity, LILRB1 and LILRB2 show variation in their strength of binding to different MHC-I alleles (17). Binding occurs predominantly through the D1-D2 domains of the receptor (57), but it has been suggested that secondary binding sites in the D3 and D4 domains may contribute to allelic variations in the strength of LILR binding (62). The potential importance of such variations was first highlighted by the observation that MHC-I complexes differing by only one amino acid in the bound peptide showed different affinities for LILRB2, which corresponded with the extent of LILRB2-mediated modulation of antigen-presenting cell phenotype (63). A subsequent comparison of binding strength for different MHC-I alleles to LILRB1 and LILRB2 identified distinct preferences (17). LILRB1 has a lower affinity for some HLA-A alleles; those with Ala 193 and Val 194 have shown lower binding ability. Ser 207 and Gln 253 alleles also show weaker binding to LILRB1 and are in linkage disequilibrium with Ala 193 and Val 194 . LILRB2 has been shown to bind most strongly to HLA-A and weakly to HLA-B alleles but with greater variability for these alleles than LILRB1. Its binding is weakest to a subset of alleles including HLA-B27 and HLA-B*5701. Some of these outliers were MHC-I alleles with known disease associations, leading to the suggestion that LILR recognition of MHC-I might influence susceptibility to, and outcome of, some viral infections or autoimmune diseases.

LiLR, MHC, AnD inFeCTiOn
Viral infection may be regarded as the primary pathology in which MHC-I recognition is essential to achieve a successful immune response. MHC-I proteins present fragments of intracellular proteins to T cells in order to enable the lysis of infected cells, and the peptide binding specificity of particular MHC-I alleles may thus influence the course of disease. There is evidence to suggest that LILR expression is induced in response to infection (64) and can be regarded as an indicator of an effective adaptive immune response (65). Studies are now beginning to highlight the relevance of LILR in particular infections and the influence of MHC-I recognition in the process.
Distinct LILR expression profiles were found to be associated with dendritic cell dysfunction during acute HIV-1 infection (66) and with "elite" control of infection (39). As there are wellcharacterized associations for different MHC-I alleles with either HIV viral control or progression to AIDS (67) and given that LILR have been implicated in its disease pathology, this viral infection represented a suitable model for testing the hypothesis that LILR may mediate MHC-I disease associations. Support for this theory was provided by studies, which demonstrated that MHC-I alleles and complexes associated with disease progression were preferential ligands for the inhibitory receptor LILRB2, whereas those associated with delayed onset of AIDS showed weaker binding to the receptor (17,63,68,69). It could therefore be hypothesized that weaker affinity for LILRB2 would result in a lack of inhibition of dendritic cell functions, resulting in a more effective anti-HIV immune response. One study sought to examine the MHC-I haplotype of HIV-1 patient cohorts in combination with the strength of their LILR binding in order to assess whether LILR recognition might influence the course of disease. An association with LILRB2, but not LILRB1, binding strength was observed, indicating that the strength of MHC-I recognition correlates with control of viral load (18). This study provided the first strong evidence that, despite the broad specificity of LILR, the strength of their binding preference for different MHC-I alleles could represent a novel mechanism for an MHC-I association during infection.
Binding of MHC-I by "Activating" members of the LILR family may also be relevant in HIV-1 infection. LILRA1 and LILRA3 preferentially bind HLA-C open confomers (17), and HLA-C variants have been associated with different outcomes of HIV infection. One particular polymorphism, −35C/T, lies 35 kb upstream of the HLA-C locus. The −35C allele corresponds with increased HLA-C expression, which in turn is associated with delayed onset of AIDS (70). HLA-C proteins are more stable in open conformer form than their HLA-A and -B counterparts and are upregulated following immune cell activation. It is, therefore, possible that LILRA1 or LILRA3 recognition of HLA-C might provide a further mechanism for MHC-I disease associations during HIV infection.
Leukocyte Ig-like receptor binding preferences for MHC-I alleles may influence the outcome of other viral infections. Expression of HLA-B27 is associated with spontaneous clearance of hepatitis C virus infection (71), and by analogy with HIV-1, it could be hypothesized that the low binding preference of LILRB2 for this allele might influence disease outcome. Another viral infection where LILR may be responsible for MHC-I-associated protective effects is dengue. Large case-control studies have identified MHC-I alleles with protective effects in dengue infection (72). Antibody opsonized dengue has recently been shown to co-ligate the inhibitory receptor LILRB1 when engaged by FcγR, leading to inhibition of FcγR signaling (73) and indicating that LILRB1 may play a role in antibody-dependent dengue. Infection with DENV is highly inflammatory and results in a large influx of activated B-cells.

AUTOiMMUniTY
Individual LILR have been implicated in autoimmunity, and their preferences for MHC-I alleles may be relevant in these conditions. Of the receptors known to recognize MHC-I, LILRA3 has been found to be associated with a number of inflammatory conditions. Expressed only in a soluble form, LILRA3 possesses no known signaling capacity of its own but can bind ligands of cellassociated LILR. Some individuals do not express LILRA3 due to a large 6.7 kbp sequence deletion. The prevalence of this deletion polymorphism is population-dependent and ranges from 6 to 84% (74,75), with a particularly high relevance in the Japanese population, where a number of non-functional spliced isoforms have also been identified (76). The deletion has been associated with increased susceptibility and early onset of multiple sclerosis (MS) symptoms in a number of studies (77,78), although conflicting data have been observed in other populations (74).
LILRA3 deficiency may also be a risk factor for Sjögrens syndrome (SS), with increased prevalence of null allele homozygous individuals (79) in certain populations, while the functional allele is a suggested risk factor in others (75). More recent studies have linked LILRA3 to rheumatoid arthritis (RA). In contrast to MS, increased serum level of functional LILRA3 is a proposed genetic risk factor for RA, with serum levels correlating directly with disease severity (80). Of further note is the prominent expression of LILRA2, A5, B2, and B3 in synovial tissues of RA patients (81), and the reduction of LILRA2, LILRB2, and LILRB3 in patients responsive to disease-modifying antirheumatic drugs (DMARDs) (82). Functional LILRA3 has also been suggested as a risk factor for systemic lupus erythematosus (SLE) following a genotyping study in Han Chinese populations, which also found higher levels of LILRA3 mRNA in SLE patients (75).

OTHeR LiGAnDS AnD FUnCTiOnS OF LiLR
Direct recognition of dengue virus by LILRB1 highlights the relevance of future studies to characterize the full range of ligands for these receptors and compare their relative binding strengths. As described above, LILRB2 is known to be the most promiscuous receptor in the family in terms of its broad specificity for classical and non-classical MHC-I in folded and unfolded forms. LILRB2 has also been shown to bind a range of non-MHC ligands including angiopoietin-like proteins (32) and NOGO, a myelin component (30). More recently, LILRA3 has also been shown to bind NOGO (83). These findings extend the relevance of LILR beyond immune responses to situations such as neurodegeneration, neural plasticity, angiogenesis, and other, as yet, unidentified scenarios where MHC-I may compete with other ligands for receptor binding (84). In the future, comparative binding assays may indicate how MHC-I allelic preferences might influence the ability of LILR to bind alternative ligands. Such investigations could cast light on previous observations regarding the relevance of MHC-I in neural plasticity and regeneration (85,86) and associations with non-immune conditions such as Alzheimer's disease.

FUTURe DiReCTiOnS
Studies on HIV-1 have provided proof of concept that LILR binding preferences for MHC-I alleles could represent a novel mechanism to explain some of the associations of MHC-I alleles with autoimmune diseases and the outcome of certain viral infections. According to this model, the influence of LILR can vary according to the strength of their binding to MHC-I alleles, representing a "degrees of self " model. MHC polymorphisms could, therefore, determine the degree of LILR signaling and consequent regulation of functions for a range of immune cell subsets as indicated in through which LILR might alter disease outcomes will require an enhanced understanding of LILR biology. It will be necessary to obtain a full characterization of the LILR expression repertoire on immune cell subsets and identify the functional effects of LILR on each cell type. For example, in the context of dengue infection, LILR expression on B cell subsets may also be relevant in viral uptake and/or generation of non-neutralizing antibodies. It will also be necessary to characterize LILR expression and function on non-immune cells. Comparative binding assays between MHC-I alleles and alternative ligands should then help to explain the wide-ranging influence of these proteins.

AUTHOR COnTRiBUTiOnS
RA and LH were responsible for the drafting and editing of the manuscript.