HLA-B27: portraying immunodominant viral epitopes

Although the crystal structure of HLA-B27 has been known for a long time, only recently have X-ray diffraction studies of this molecule in complex with individual peptides become available. The report of three such structures involving viral epitopes that are immunodominant in HLA-B27-restricted T cell responses against influenza, Epstein-Barr and HIV viruses significantly improves our perception of critical aspects of the immunological and pathogenetic roles of HLA-B27, including (1) the molecular basis of its peptide-binding specificity and how this is modulated by subtype polymorphism, (2) the relationship between the structural and the antigenic features of immunodominant viral epitopes, (3) the basis for long term non-progression to AIDS of HIV-infected HLA-B27+ individuals, and (4) the structural features of microbial peptides influencing NK receptor engagement. Here, I discuss the implications of this and related studies for the relevance of HLA-B27 in host defense and as a pathogenetic molecule in spondyloarthritis.     

B*2707 differs in peptide specificity from B*2705 and B*2704 as much as from HLA-B27 subtypes not associated to spondyloarthritis

HLA-B*2707 is associated with ankylosing spondylitis in most populations. Like the non-associated allotypes B*2706 and B*2709, it lacks Asp116 and shows preference for peptides with nonpolar C-terminal residues. The relationships between the peptide specificity of B*2707 and those of the disease-associated B*2705 and the non-associated subtypes were analyzed by determining the overlap between the corresponding peptide repertoires, the sequence of shared and differential ligands, and by comparing allospecific T cell epitopes with peptide sharing. The B*2707-bound repertoire was as different from that of B*2705 as from those of B*2706, B*2709, or the two latter subtypes from each other. Differences between B*2707 and B*2705 were based on their C-terminal residue specificity and a subtle modulation at other positions. Differential usage of secondary anchor residues explained the disparity between the B*2707-, B*2706-, and B*2709-bound repertoires. Similar differences in residue usage were found between B*2707 and both B*2704 and B*2706, as expected from the high peptide overlap between the two latter subtypes. T cell cross-reaction paralleled peptide sharing, suggesting that many shared ligands conserve their alloantigenic features on distinct subtypes. Our results indicate that association of HLA-B27 subtypes with ankylosing spondylitis does not correlate with higher peptide sharing among disease-associated subtypes or with obvious peptide motifs.     

T-cell receptor usage in alloreactivity against HLA-B*2703 reveals significant conservation of the antigenic structure of B*2705

B*2703 is an exceptional HLA-B27 molecule in that it differs from the most common B*2705 subtype by a unique amino acid change (His59) altering N-terminal peptide anchorage. To assess how this unusual feature affects the antigenic structure of HLA-B27, TCR usage by alloreactive CTL raised against B*2703 from two individuals was analyzed. Only few CTL recognized B*2703 but not or at a lower level B*2705. Limited heterogeneity of these CTL was revealed by: 1) identity of TCR in two pairs of such CTL clones, 2) identity of β chains, paired to distinct α chains, in two clonotypes, and 3) almost identical fine specificity of these two clonotypes with site-specific HLA-B27 mutants. These results indicate that B*2703 "private" epitopes are rare. TCR usage among anti-B*2703 CTL was analogous as in anti-B*2705 responses in the predominant and donor-independent usage of Vβ segments from homology subgroup 4, more moderate and donor-dependent Vα skewing, N+Dβ diversity limited by motifs shared among clonotypes, and restricted Jα heterogeneity. Homology of N+Dβ motifs and Jα segments of anti-B*2703 with anti-B*2705 TCR suggested significant sharing of peptide-associated epitopes between both subtypes. The results indicate that allospecific TCR are recruited by B*2703 following similar rules as in the anti-B*2705 response, and suggest that the B*2703 change keeps unaltered much of the antigenic structure of the molecule relative to B*2705. Therefore, most of the peptides bound to B*2703 should be the same and keep a similar conformation as in B*2705.     

KIR2DL3 and KIR2DL1 show similar impact on licensing of human NK cells

Killer cell immunoglobulin‐like receptor/HLA class I (KIR/HLA‐I) combinations are associated with disease risk, implicating functional roles for NK cells (NKCs) or KIR⁺ T cells. KIR/HLA‐I interactions can act through inhibition of NKC activation by target cells and NKC licensing for greater intrinsic responsiveness. We compared licensing conferred by the weaker, HLA‐C group 1/KIR2DL3, and the stronger, HLA‐C group 2/KIR2DL1, inhibitory combinations. The “rheostat model” predicts weaker licensing by HLA‐C1/KIR2DL3 interactions than HLA‐C2/KIR2DL1. We analyzed degranulation in NKC subsets expressing single and multiple receptors for HLA‐I. NKG2A had the strongest licensing impact, while KIR2DL3, KIR2DL1, and KIR3DL1 were weaker, and not significantly different to each other. Presence of one or two matched HLA‐C allotypes did not alter licensing of KIR2DL3⁺ and KIR2DL1⁺ NKC. Coexpression of activating KIR2DS1 disarmed KIR2DL3⁺ and KIR2DL1⁺ NKC to a similar extent. KIR3DL1 and NKG2A combined for more enhanced licensing of double‐positive NKC than the combination of KIR2DL3 and KIR2DL1. Thus, KIR2DL3 and KIR2DL1 have similar capacity to license NKC, suggesting that inhibitory signal strength and amount of available HLA‐C ligands do not correlate with NKC licensing. Altogether, our results show that the basis for disease associations of HLA‐C and KIR2DL likely encompasses factors other than licensing.     

The differential impact of natural killer (NK) cell education via KIR2DL3 and KIR3DL1 on CCL4 secretion in the context of in‐vitro HIV infection

Carriage of certain inhibitory natural killer (NK) cell receptor (iNKR)/HLA ligand pairs is associated with protection from infection and slow time to AIDS implicating NK cells in HIV control. NK cells acquire functional potential through education, which requires the engagement of iNKRs by their human leucocyte antigen (HLA) ligands. HIV infection down‐regulates cell surface HLA‐A/B, but not HLA‐C/E. We investigated how NK cell populations expressing combinations of the iNKRs NKG2A, KIR2DL3 (2DL3) and KIR3DL1 (3DL1) responded to autologous HIV infected CD4 (iCD4) cells. Purified NK cells from HIV‐uninfected individuals were stimulated with autologous HIV iCD4 or uninfected CD4 T cells. Using flow cytometry we gated on each of the 8 NKG2A^+/–2DL3^+/–3DL1^+/‐ populations and analysed all possible combinations of interferon (IFN)‐γ, CCL4 and CD107a functional subsets responding to iCD4 cells. Infected CD4 cells induced differential frequencies of NKG2A^+/–2DL3^+/–3DL1^+/– populations with total IFN‐γ⁺, CCL4⁺ and CD107a⁺ functional profiles. 2DL3⁺NKG2A⁺ NK cells had a higher frequency of responses to iCD4 than other populations studied. A higher frequency of 2DL3⁺ NK cells responded to iCD4 from individuals that were not HLA‐C1 homozygotes. These results show that 2DL3⁺ NK cells are mediators of HIV‐specific responses. Furthermore, responses of NK cell populations to iCD4 are influenced not only by NK cell education through specific KIR/HLA pairs, but also by differential HIV‐mediated changes in HLA expression.     

The effect of LILRB1 but not LILRA3 gene polymorphism in immunopathology of ankylosing spondylitis—A parallel to KIR genes

LILR and KIR receptors recognize HLA‐B27 and may influence immune response in ankylosing spondylitis (AS) development. Purpose of the study was to analyse LILRB1/LILRA3 polymorphisms in AS. We observed a possible protective effect of the T allele of LILRB1 rs1061680:T>C and no association with insertion/deletion polymorphisms of LILRA3 with AS.     

High responsiveness of HLA-B57-restricted Gag-specific CD8+ T cells in vitro may contribute to the protective effect of HLA-B57 in HIV-infection

HLA-B57 has been shown to be associated with long-term asymptomatic HIV-1 infection. To investigate the biological mechanism by which the HLA-B57 allele could protect from HIV-1 disease, we studied both the number of CD8(+) T cells as well as CD8(+) T cell responsiveness directed to different HIV-1 Gag peptides presented by HLA-A2, -B8 or -B57. T cells specific for the HLA-B57 peptide KAFSPEVIPMF responded more readily and to a higher extend to antigenic stimulation in vitro than T cells specific for the HLA-A2 peptide SLYNTVATL or the HLA-B8 peptide EIYKRWII. This phenomenon was reproducible with T cells from individuals expressing HLA-B57 in combination with one or both of the other alleles and was persistent during long-term follow-up. Lower reactivity of A2- and B8-restricted T cells was not explained by mutations in the B8- or A2-restricted Gag-peptides. Moreover, no correlation between peptide mutation frequency and IFN-gamma production by the corresponding Gag-specific T cells was observed. In conclusion, functional differences were observed between T cells specific for HIV epitopes derived from the same protein presented by different HLA molecules. B57-restricted KAFSPEVIPMF-specific CD8(+) T cells have relatively high responsiveness, which could contribute to the protective effect of HLA-B57 in HIV infection.     

Lymphoblastoid cells express HLA-B27 homodimers both intracellularly and at the cell surface following endosomal recycling

The MHC class I allele HLA-B27 is very strongly associated with development of autoimmune spondyloarthritis, although the disease mechanism remains unknown. Class I molecules classically associate in the endoplasmic reticulum (ER) with beta2-microglobulin (beta(2)m) and antigenic peptides for cell surface expression and presentation to T cells. We have previously shown that HLA-B27 is capable of forming beta(2)m-free disulfide-bonded homodimers in vitro. Here we show that HLA-B27 forms disulfide-bonded homodimers in vivo by two distinct pathways. HLA-B27 homodimers form in the ER but appear unable to egress to the cell surface in human cells. Cell surface HLA-B27 homodimers are abundantly expressed in a variety of lymphoid cell lines. Experiments with inhibitors indicate that HLA-B27 homodimers can arise from cell-surface heterodimers via an endosome-dependent recycling pathway. HLA-B27 homodimer expression on the cell surface of 721.220 is dependent on the unpaired cysteine(67) and is inhibited by restoration of tapasin function or by incubation with peptides that bind strongly to HLA-B27 heterodimers. Cell surface expressed HLA-B27 homodimers are likely to be immunologically reactive ligands for NK family immunoreceptors and, hence, could play a pathogenic role in spondyloarthritis.     

Contribution of functional KIR3DL1 to ankylosing spondylitis

Increasing evidence points to a role for killer immunoglobulin-like receptors (KIRs) in the development of autoimmune diseases. In particular, a positive association of KIR3DS1 (activating receptor) and a negative association of KIR3DL1 (inhibitory receptor) alleles with ankylosing spondylitis (AS) have been reported by several groups. However, none of the studies analyzed these associations in the context of functionality of polymorphic KIR3DL1. To better understand how the KIR3DL1/3DS1 genes determine susceptibility to AS, we analyzed the frequencies of alleles and genotypes encoding functional (KIR3DL1*F) and non-functional (KIR3DL1*004) receptors. We genotyped 83 AS patients and 107 human leukocyte antigen (HLA)-B27-positive healthy controls from the Russian Caucasian population using a two-stage sequence-specific primer PCR, which distinguishes KIR3DS1, KIR3DL1*F and KIR3DL1*004 alleles. For the patients carrying two functional KIR3DL1 alleles, those alleles were additionally genotyped to identify KIR3DL1*005 and KIR3DL1*007 alleles, which are functional but are expressed at low levels. KIR3DL1 was negatively associated with AS at the expense of KIR3DL1*F but not of KIR3DL1*004. This finding indicates that the inhibitory KIR3DL1 receptor protects against the development of AS and is not simply a passive counterpart of the segregating KIR3DS1 allele encoding the activating receptor. However, analysis of genotype frequencies indicates that the presence of KIR3DS1 is a more important factor for AS susceptibility than the absence of KIR3DL1*F. The activation of either natural killer (NK) or T cells via the KIR3DS1 receptor can be one of the critical events in AS development, while the presence of the functional KIR3DL1 receptor has a protective effect. Nevertheless, even individuals with a genotype that carried two inhibitory KIR3DL1 alleles expressed at high levels could develop AS.     

Differences in the recognition by CTL of peptides presented by the HLA-B*4402 and the HLA-B*4403 molecules which differ by a single amino acid

The HLA-B*4402 and B*4403 molecules differ only at residue 156, which borders the peptide binding site. Strong in vivo allogeneic reactions mediated by cytolytic T lymphocytes (CTLs) were reported in patients who received a bone marrow graft mismatched for these B44 subtypes, indicating that HLA-B*4402 and B*4403 molecules present distinct antigens. This could be due either to the presentation of different sets of antigenic peptides or to the recognition by CTLs of conformational epitopes formed by the MHC molecules alone or in association with antigenic peptides. To address this question, we compared the two B44 subtypes in their presentation to tumor-specific CTLs of three peptides, encoded by genes MAGE-3, MUM-1 and Tyrosinase. The peptides bound with similar affinities to B*4402 or B*4403 molecules, as assessed by lytic competition assays. One HLA-B*4402-restricted and one HLA-B*4403-restricted CTL clone were derived against each peptide. When tested for lysis of B*4402 and B*4403 cells incubated with the antigenic peptides, most CTLs showed a marked preference for one of the two B44 subtypes. Using variant peptides incorporating single alanine substitutions, we compared a given CTLs' recognition of its antigenic peptide presented by both B44 subtypes. Some substitutions, which had no effect on the binding of the peptide, affected its recognition by the same CTL differently on B*4402 and B*4403 molecules. These results imply that the conformations adopted by the same peptide on the two HLA-B44 subtypes are different. We conclude that the B44 subtype specificity of T cells results mostly from distinct conformations adopted by the same peptides in the two B44 molecules. This does not exclude the possibility that in some cases the B44 subtype specificity results from the selective binding of a peptide to one subtype. We found several peptides, different from the three mentioned above, that contain the canonical HLA-B44 binding motif and bind to B*4403 but not to B*4402 molecules.     

Luminal bacterial antigen-specific CD4+ T-cell responses in HLA-B27 transgenic rats with chronic colitis are mediated by both major histocompatibility class II and HLA-B27 molecules

Rats transgenic (TG) for the human major histocompatibility complex (MHC) class I HLA-B27 and beta2-microglobulin genes develop chronic colitis under specific pathogen-free (SPF) but not sterile (germ-free, GF) conditions. We investigated the role of antigen-presenting molecules involved in generating immune responses by CD4+ mesenteric lymph node (MLN) cells from colitic HLA-B27 TG rats to commensal enteric micro-organisms. All TG MLN cells expressed HLA-B27. A higher level of MHC class II was expressed on cells from TG rats, both SPF and GF, compared to non-TG littermates. In contrast, rat MHC class I expression was lower on TG than non-TG cells. Both TG and non-TG antigen presenting cells (APC) pulsed with caecal bacterial antigens induced a marked interferon-gamma (IFN-gamma) response in TG CD4+ T lymphocytes but failed to stimulate non-TG cells. Blocking MHC class II on both TG and non-TG APC dramatically inhibited their ability to induce TG CD4+ T cells to produce IFN-gamma. Blocking HLA-B27 on TG APC similarly inhibited IFN-gamma responses. When the antibodies against MHC class II and HLA-B27 were combined, no APC-dependent IFN-gamma response was detected. These data implicate both native rat MHC class II and TG HLA-B27 in CD4+ MLN T-cell IFN-gamma responses to commensal enteric microflora in this colitis model.     

Reduced ratio of protective versus proinflammatory cytokine responses to commensal bacteria in HLA-B27 transgenic rats

Germ-free HLA-B27 transgenic (TG) rats do not develop colitis, but colonization with specific pathogen-free (SPF) bacteria induces colitis accompanied by immune activation. To study host-dependent immune responses to commensal caecal bacteria we investigated cytokine profiles in mesenteric lymph node (MLN) cells from HLA-B27 TG versus nontransgenic (non-TG) littermates after in vitro stimulation with caecal bacterial lysates (CBL). Supernatants from CBL-stimulated unseparated T- or B- cell-depleted MLN cells from HLA-B27 TG and non-TG littermates were analysed for IFN-gamma, IL-12, TNF, IL-10 and TGF-beta production. Our results show that unfractionated TG MLN cells stimulated with CBL produced more IFN-gamma, IL-12 and TNF than did non-TG MLN cells. In contrast, CBL-stimulated non-TG MLN cells produced more IL-10 and TGF-beta. T cell depletion abolished IFN-gamma and decreased IL-12 production, but did not affect IL-10 and TGF-beta production. Conversely, neither IL-10 nor TGF-beta was produced in cultures of B cell-depleted MLN. In addition, CD4(+) T cells enriched from MLN of HLA-B27 TG but not from non-TG rats produced IFN-gamma when cocultured with CBL-pulsed antigen presenting cells from non-TG rats. Interestingly, IL-10 and TGF-beta, but not IFN-gamma, IL-12 and TNF were produced by MLN cells from germ-free TG rats. These results indicate that the colitis that develops in SPF HLA-B27 TG rats is accompanied by activation of IFN-gamma-producing CD4(+) T cells that respond to commensal bacteria. However, B cell cytokine production in response to components of commensal intestinal microorganisms occurs in the absence of intestinal inflammation.     

The silent KIR3DP1 gene (CD158c) is transcribed and might encode a secreted receptor in a minority of humans, in whom the KIR3DP1, KIR2DL4 and KIR3DL1/KIR3DS1 genes are duplicated

Killer-cell Ig-like receptors (KIR) are structurally and functionally diverse, and enable human NK cells to survey the expression of individual HLA class I molecules, often altered in infections and tumors. Multiple events of non-reciprocal recombination have contributed to the rapid diversification of KIR. We show that approximately 4.5% of the individuals of a Caucasoid population bear a recombinant allele of KIR3DP1, officially designed KIR3DP1*004, that associates tightly with gene duplications of KIR3DP1, KIR2DL4 and KIR3DL1/KIR3DS1. The KIR3DP1 gene is normally silent, but the recombinant allele carries a novel promoter sequence and, as a consequence, is transcribed in all tested individuals. Messenger RNA of KIR3DP1*004 is made up of six exons; of these, exons 1-5 are similar to, and spliced like, those encoding the leader peptide and Ig-domains of KIR3D. By contrast, exon 6 is homologous to no other human KIR sequence, but only to possible homologs in chimpanzees and rhesus macaques, and encodes a short hydrophilic tail. The putative KIR3DP1*004 product, like those of the related genes LAIR-2 and LILRA3/ILT6/LIR4, is predicted to be secreted to the extracellular medium rather than anchored to the cell membrane.     

Interaction between KIR3DL1 and HLA-B*57 supertype alleles influences the progression of HIV-1 infection in a Zambian population

KIR and HLA loci are both highly polymorphic, and some HLA class 1 products bind and trigger cell-surface receptors specified by KIR genes. We examined whether KIR genes act in concert with HLA-B locus to control HIV-1 infection in a sample of Zambian patients. DNA samples from 88 Zambian patients with HIV-1 were examined. Patients were classified as either slow progressors (SP; n = 54) or rapid progressors (RP; n = 34) to AIDS. All were typed for HLA-B and KIR genes. Our results reveal an association between B*57 supertype (B*57s, which includes B*57 and B*58 alleles) and delayed progression to AIDS (p = 0.0007 by pc = 0.015; OR = 5.25). We also observed an increase incidence of Bw4-I80 in patients with slow progression (p = 0.001 by pc = 0.003, OR = 5). This increase was found to be secondary to B*57s. The presence of both KIR3DL1 and B*57S has a significant effect on progression to AIDS (p = 0.0008; OR = 5.61). B*57s genotypes with another HLA-B allele different from those in the trans position, which also had a specificity different to Bw4-I80 (Bw4-T80 or Bw6), was also greater in the SP than in the RP group (p = 0.00003; OR = 10.11). The presence of the inhibitory allele KIR3DL1 in combination with the HLA-B*57s alleles that contain the Bw4-I80 epitope, has a highly protective effect against progression to AIDS in Zambian patients.     

Increased frequency and function of KIR2DL1–3+ NK cells in primary HIV‐1 infection are determined by HLA‐C group haplotypes

The acquisition and maintenance of NK‐cell function is mediated by inhibitory killer‐cell immunoglobulin‐like receptors (KIRs) through their interaction with HLA class I molecules. Recently, HLA‐C expression levels were shown to be correlated with protection against multiple outcomes of HIV‐1 infection; however, the underlying mechanisms are poorly understood. As HLA‐C is the natural ligand for the inhibitory receptors KIR2DL1 and KIR2DL2/3, we sought to determine whether HLA‐C group haplotypes affect NK‐cell responses during primary HIV‐1 infection. The phenotypes and functional capacity of NK cells derived from HIV‐1‐positive and HIV‐1‐negative individuals were assessed (N = 42 and N = 40, respectively). HIV‐1 infection was associated with an increased frequency of KIR2DL1–3⁺ NK cells. Further analysis showed that KIR2DL1⁺ NK cells were selectively increased in individuals homozygous for HLA‐C2, while HLA‐C1‐homozygous individuals displayed increased proportions of KIR2DL2/3⁺ NK cells. KIR2DL1–3⁺ NK cells were furthermore more polyfunctional during primary HIV‐1 infection in individuals also encoding for their cognate HLA‐C group haplotypes, as measured by degranulation and IFN‐γ and TNF‐α production. These results identify a novel relationship between HLA‐C and KIR2DL⁺ NK‐cell subsets and demonstrate that HLA‐C‐mediated licensing modulates NK‐cell responses to primary HIV‐1 infection.     

Co‐expression of HLA‐B7 and HLA‐B27 alleles is associated with B7‐restricted immunodominant responses following influenza infection

It is recognized that host response following viral infection is characterized by immunodominance, but deciphering the different factors contributing to immunodominance has proved a challenge due to concurrent expression of multiple MHC class I alleles. To address this, we generated H2‐K^?/?/D^?/? double‐knockout transgenic mice expressing either one or two human MHC‐I alleles. We hypothesized that co‐expression of different allele combinations figures critically in immunodominance and examined this in influenza‐infected, double Tg MHC‐I mice. In A2/B7 or A2/B27 mice, using ELISpot assays with the A2‐restricted matrix I.58–66, the B7‐restricted NP418–426 or the B27‐restricted NP383–391 influenza A (flu) epitopes, we observed the expected recognition of both peptides for both alleles. In contrast, in flu‐infected B7/B27 mice, a significantly reduced level of B27/NP383‐restricted CTL response was detected while there was no change in the B7/NP418‐restricted CTL response. Flu‐specific tetramer studies revealed a partial deletion of Vβ8.1⁺ NP383/B27‐restricted CD8⁺ T cells, and a diminished Vβ12⁺ CD8⁺ T‐cell expansion in B7/B27 Tg mice. Using HLA Tg chimeric mice, we confirmed these findings. These findings shed light on the immune consequences of co‐dominant expression of MHC‐I alleles for host immune response to pathogens.     

Tripeptidyl peptidase II is dispensable for the generation of both proteasome-dependent and proteasome-independent ligands of HLA-B27 and other class I molecules

A significant fraction of the HLA-B27-bound peptide repertoire is resistant to proteasome inhibitors. The possible implication of tripeptidyl peptidase II (TPPII) in generating this subset was analyzed by quantifying the surface re-expression of HLA-B*2705 after acid stripping in the presence of two TPPII inhibitors, butabindide and Ala-Ala-Phe-chloromethylketone. Neither decreased HLA-B27 re-expression under conditions in which TPPII activity was largely inhibited. This was in contrast to a significant effect of the proteasome inhibitor epoxomicin. The failure of TPPII inhibition to decrease surface re-expression was not limited to HLA-B27, since it was also observed in several HLA-B27-negative cell lines, including Mel JuSo. Actually, HLA class I re-expression in Mel JuSo cells increased as a function of butabindide concentration, which is consistent with an involvement of TPPII in destroying HLA class I ligands. Inhibition of TPPII with small interfering RNA also failed to decrease the surface expression of HLA class I molecules on 143B cells. Our results indicate that TPPII is dispensable for the generation of proteasome-dependent HLA class I ligands and, without excluding its role in producing some individual epitopes, this enzyme is not involved to any quantitatively significant extent, in generating the proteasome-independent HLA-B27-bound peptide repertoire.