PD-L2 is a second ligand for PD-1 and inhibits T cell activation.

Programmed death I (PD-I)-deficient mice develop a variety of autoimmune-like diseases, which suggests that this immunoinhibitory receptor plays an important role in tolerance. We identify here PD-1 ligand 2 (PD-L2) as a second ligand for PD-1 and compare the function and expression of PD-L1 and PD-L2. Engagement of PD-1 by PD-L2 dramatically inhibits T cell receptor (TCR)-mediated proliferation and cytokine production by CD4+ T cells. At low antigen concentrations, PD-L2-PD-1 interactions inhibit strong B7-CD28 signals. In contrast, at high antigen concentrations, PD-L2-PD-1 interactions reduce cytokine production but do not inhibit T cell proliferation. PD-L-PD-1 interactions lead to cell cycle arrest in G0/G1 but do not increase cell death. In addition, ligation of PD-1 + TCR leads to rapid phosphorylation of SHP-2, as compared to TCR ligation alone. PD-L expression was up-regulated on antigen-presenting cells by interferon gamma treatment and was also present on some normal tissues and tumor cell lines. Taken together, these studies show overlapping functions of PD-L1 and PD-L2 and indicate a key role for the PD-L-PD-1 pathway in regulatingT cell responses.

Costimulation light: activation of CD4+ T cells with CD80 or CD86 rather than anti-CD28 leads to a Th2 cytokine profile.

To examine the role of CD28 and CTLA-4 in Th cell differentiation, we used a novel microsphere-based system to compare the effects of CD28 ligation by Ab or CD80/CD86. One set of beads was prepared by coating with anti-CD3 and anti-CD28 Ab. Another set of beads was prepared by immobilizing anti-CD3 and murine CD80-Ig fusion protein or murine CD86-Ig fusion protein on the beads. The three sets of beads were compared in their effects on the ability to activate and differentiate splenic CD4 T cells. When purified naive CD4(+) cells were stimulated in vitro, robust proliferation of similar magnitude was induced by all three sets of beads. When cytokine secretion was examined, all bead preparations induced an equivalent accumulation of IL-2. In contrast, there was a marked difference in the cytokine secretion pattern of the Th2 cytokines IL-4, IL-10, and IL-13. The B7-Ig-stimulated cultures had high concentrations of Th2 cytokines, whereas there were low or undetectable concentrations in the anti-CD28-stimulated cultures. Addition of anti-CTLA-4 Fab augmented B7-mediated IL-4 secretion. These studies demonstrate that B7 is a critical and potent stimulator of Th2 differentiation, and that anti-CD28 prevents this effect.

CTLA-4 (CD152) can inhibit T cell activation by two different mechanisms depending on its level of cell surface expression.

CTLA-4 (CD152) engagement results in down-regulation of T cell activation. Two mechanisms have been postulated to explain CTLA-4 inhibition of T cell activation: negative signaling and competitive antagonism of CD28:B7-mediated costimulation. We assessed the contributions of these two mechanisms using a panel of T cell lines expressing human CTLA-4 with mutations in the cytoplasmic region. Under conditions of B7-independent costimulation, inhibition of IL-2 production following CTLA-4 engagement required the CTLA-4 cytoplasmic region. In contrast, under B7-dependent costimulation, inhibition of IL-2 production by CTLA-4 engagement was directly proportional to CTLA-4 cell surface levels and did not require its cytoplasmic region. Thus, CTLA-4 down-regulates T cell activation by two different mechanisms-delivery of a negative signal or B7 sequestration-that are operational depending on the levels of CTLA-4 surface expression. These two mechanisms may have distinct functional outcomes: rapid inhibition of T cell activation or induction of T cell anergy.

The inhibitory function of CTLA-4 does not require its tyrosine phosphorylation.

CTLA-4 is a negative regulator of T cell responses. Sequence analysis of this molecule reveals the presence of two cytoplasmic tyrosine residues at positions 165 and 182 that are potential Src homology (SH)-2 domain binding sites. The role of phosphorylation of these residues in CTLA-4-mediated signaling is unknown. Here, we show that sole TCR ligation induces zeta-associated protein (ZAP)-70-dependent tyrosine phosphorylation of CTLA-4 that is important for cell surface retention of this molecule. However, CTLA-4 tyrosine phosphorylation is not required for down-regulation of T cell activation following CD3-CTLA-4 coengagement. Specifically, inhibition of extracellular signal-regulated kinase (ERK) activation and of IL-2 production by CTLA-4-mediated signaling occurs in T cells expressing mutant CTLA-4 molecules lacking the cytoplasmic tyrosine residues, and in lck-deficient or ZAP-70-deficient T cells. Therefore, CTLA-4 function involves interplay between two different levels of regulation: phosphotyrosine-dependent cell surface retention and phosphotyrosine-independent association with signaling molecules.

CTLA-4 ligation delivers a unique signal to resting human CD4 T cells that inhibits interleukin-2 secretion but allows Bcl-X(L) induction.

We have assessed the functional effects of a panel of CTLA-4 mAbs on resting human CD4+ T cells. Our results demonstrate that some CTLA-4 mAbs can inhibit proliferative responses of resting CD4+ cells and cell cycle transition from G0 to G1. The inhibitory effects of CTLA-4 were evident within 4 h, at a time when cell surface CTLA-4 expression remained undetectable. Other CTLA-4 mAbs had no detectable inhibitory effects, indicating that binding of Ab to CTLA-4 alone is not sufficient to mediate down-regulation of T cell responses. Interestingly, while IL-2 production was shut off, inhibitory anti-CTLA-4 mAbs permitted induction and expression of the cell survival gene bcl-X(L). Consistent with this observation, cells remained viable and apoptosis was not detected after CTLA-4 ligation.

Beta 2-microglobulin with an endoplasmic reticulum retention signal increases the surface expression of folded class I major histocompatibility complex molecules.

With beta 2-microglobulin- (beta 2m-) cell lines such as R1E/Db, the surface expression of class I major histocompatibility complex molecules is greatly impaired, and class I molecules that are on the surface are generally misfolded. To determine whether beta 2m must be continually present with the class I heavy chain for the class I molecule to reach the surface in a folded conformation, a sequence encoding an endoplasmic reticulum (ER) retention signal (KDEL) was attached onto the 3' end of a beta 2m cDNA. After this chimeric cDNA was transfected into R1E/Db cells, beta 2m-KDEL protein was detectable by an anti-beta 2m serum within the cells but not at the cell surface. Interestingly, R1E/Db cells transfected with beta 2m-KDEL were found to express a high level of conformationally correct Db molecules at the cell surface. This observation implies that beta 2m has a critical and temporal role in the de novo folding of the class I heavy chain. We propose that the critical time for beta 2m association is when the class I molecule is docked with the transporter associated with antigen processing (TAP) and first interacts with peptide.

Characterization of class I MHC folding intermediates and their disparate interactions with peptide and beta 2-microglobulin.

Newly synthesized class I heavy chains achieve domain structure using disulfide bonds, assemble with beta-2 microglobulin (beta 2m), and bind peptide ligand to complete the trimeric complex. Although each of these initial events is thought to be critical for class I folding, their sequential order and effect on class I structure are unknown. Using mAb specific for distinct conformations of H-2Ld and Lq, we have defined folding intermediates of class I molecules. We show here that non-peptide-associated forms of Ld or Lq, detected by mAb 64-3-7 and designated L alt, lack numerous conformational epitopes surrounding their ligand binding sites. These results support the notion that L alt molecules have an open conformation. Interestingly, a significant proportion of L alt molecules were detected in association with beta 2m and these L alt/beta 2m heterodimers were preferentially folded by peptide in cell lysates. These findings indicate that class I heavy chain/beta 2m association can precede ligand binding and that peptide is probably the limiting factor for completion of the Ld/beta 2m/peptide trimeric complex in vivo. The characteristics of L alt molecules were investigated further by ascertaining the disulfide bond status of these molecules and their association with beta 2m and peptide. Treatment of cells with dithiothreitol (DTT), a membrane-permeable reducing agent, demonstrated that L alt molecules constitute a heterogeneous population including reduced, partially reduced and native class I molecules. Furthermore, partially reduced Ld alt molecules, in a cell line expressing a mutant Ld molecule lacking the alpha 2 domain disulfide bond, accumulated intracellularly, were not beta 2m-associated and displayed marginal peptide-induced folding in vitro. In accordance with this latter finding, peptide was found to preferentially convert fully disulfide-bonded forms of Ld alt to conformed Ld. Thus, we propose that intrachain disulfide bond formation precedes the association of class I heavy chain with beta 2m and peptide, and that disulfide bond formation is required for efficient assembly, ligand binding and folding of the class I heavy chain.

Peptide-induced rescue of serologic epitopes on class I MHC molecules.

To monitor conformational changes in MHC class I structure induced by interaction with peptide or beta 2-microglobulin (beta 2-m), we have taken a serologic approach. Previous studies by us and others have defined circumstances wherein specific peptides can decrease serologic recognition of class I molecules. However, such blocking of serologic epitopes has often been interpreted as steric hindrance by peptide side chains. In this paper, we describe peptide-induced gains in recognition by mAbs 30-5-7, 34-1-2, and B22/249. In experiments with mAb 30-5-7, impaired reactivity, which resulted from an Ld loop mutation, was specifically rescued by the binding of a beta-galactosidase-derived peptide to the Ld mutant. In studies with mAb 34-1-2, poor Ld detection was enhanced by mutations in Ld at beta 2-m interaction sites or by changes within the peptide-binding groove. To evaluate whether known peptides in the Ld groove could influence 34-1-2 recognition, we tested six peptide ligands, four of which increased the reactivity of 34-1-2 with the Ld-expressing cell to various degrees (up to 14-fold). It is of interest that Ld mutations at position 9 and 95/97 made significant differences in the ranking of the peptides in regard to their ability to increase recognition by 34-1-2 and B22/249. This finding suggests that mutations in the binding groove can alter peptide conformation and result in secondary changes in class I structure. On the basis of the cumulative serologic data, we propose that the class I molecule displays considerable fluidity, and is structurally influenced by both beta 2-m and peptide.

Exogenous peptide ligand influences the expression and half-life of free HLA class I heavy chains ubiquitously detected at the cell surface.

A pool of free HLA class I chains has been detected at the plasma membrane of all cells concomitantly expressing folded and assembled class I molecules. To determine the origin of these free HLA heavy chains, we have examined the biosynthesis of a single HLA class I molecule, HLA-B27, expressed by a murine cell line (L-B27). In L-B27 cells, as previously shown in Epstein-Barr virus-transformed lymphoblastoid cell lines, a precursor/product relationship exists, early in biosynthesis, between free (HC10-reactive) and beta-2-microglobulin (beta 2m)-associated (W6/32-reactive) class I heavy chains as demonstrated by pulse/chase experiments. At later stages in class I biosynthesis, both HC10- and W6/32-reactive heavy chains display complex oligosaccharides and accumulate at the cell surface. HC10- and W6/32-reactive molecules are both very stable at the cell surface, with half-lifes (t1/2) of > 7 h and approximately 4 h, respectively. Interestingly, cell surface expression and turnover of HC10- and W6/32-reactive molecules were affected by the addition of peptide ligands to the culture media. Culturing cells in the presence of HLA-B27 ligands resulted in the increased expression of W6/32-reactive molecules and the decreased expression of HC10-reactive molecules. Moreover, addition of exogenous peptide extended the t1/2 of W6/32-reactive molecules to > 7 h and reduced that of HC10-reactive molecules to 4 h. These results indicate that surface HC10-reactive molecules result largely from W6/32-reactive molecules following peptide and beta 2m dissociation. Therefore, HC10-reactive species are not only the precursors but also the end products in class I biosynthesis.

Binding of peptides lacking consensus anchor residue alters H-2Ld serologic recognition.

CTL recognize class I MHC/peptide complexes on the surface of target cells. Crystallographic and serologic data have indicated that peptide ligands can influence the conformation of class I molecules and hence T cell recognition. How the binding of peptides with disparate sequence motifs affects the conformation of distinct regions within a class I molecule remains unknown. A series of site-directed mutants of the murine class I molecule H-2Ld was studied to address this question. These mutants were generated by in vitro mutagenesis and used to map the serologic epitopes recognized by a panel of Ld-reactive mAb. The influence of six different ligands on serologic recognition by these mAb was then examined. Of 12 mAb tested, only one, B22/249, was found to be significantly influenced by the bound peptide. Peptide discrimination by B22/249 was observed at the cell surface and in immunoprecipitates of Ld after incubation with two of the six ligands. The two peptides that caused suboptimal B22/249 recognition of Ld/peptide lack a proline at position 2, which is present in the other four peptides and has previously been defined as an anchor residue for Ld ligands. The epitope on Ld detected by mAb B22/249 includes residues 63 to 70 on the alpha 1 domain helix. Two of these residues are in pocket B, which computer modeling predicts to be in contact with the second residue of Ld-binding peptides. Therefore, these data imply that a mAb to a class I molecule can distinguish peptides with different motifs, possibly reflecting peptide-dependent conformational changes in the class I molecule.

Residues in pockets B and F of HLA-B27 are critical in the presentation of an influenza A virus nucleoprotein peptide and influence the stability of peptide - MHC complexes.

Six pockets, designated A through F, which extend from the peptide binding site of class I HLA molecules, have been postulated to play an important role in determining peptide binding specificity. HLA-B27 mutant molecules with single amino acid substitutions at residues 9his-->phe, 24thr-->ser, 45glu-->thr, and 67cys-->ala in pocket B; 114his-->asn in pocket D; and 116asp-->phe in pocket F have been generated and characterized for their capacity to present an influenza A nucleoprotein peptide (NP 383-391) for cytotoxic T lymphocyte recognition. We report here that substitutions in residues 45, 67, and 116 affect presentation of NP 383-391 when peptide is processed and loaded during viral infection. Using 125I-labeled NP peptide, we demonstrate that substitutions in residues 67 and 116 alter the stability of NP-HLA-B27 complexes. A substitution at position 9 of the NP peptide complements the mutation introduced at residue 116, suggesting that the NP peptide binds with its carboxy terminal amino acid in pocket F. These findings indicate that polymorphic residues within pockets B and F of HLA-B27 play a crucial role in peptide binding and stability of peptide-MHC class I complexes. Furthermore, our results suggest that substitutions at allele-specific residues within pockets B and F alter the stability of NP-HLA-B27 complexes resulting in the diminution or abrogation of NP presentation during viral infection.

The peptide binding specificity of HLA class I molecules is largely allele-specific and non-overlapping.

To understand better the specificity of peptide binding by MHC class I molecules, we have evaluated the capacity of a panel of unrelated peptides to compete for the presentation of viral peptides presented by HLA-A3 and HLA-B27. The HIV-Nef7F peptide (74-82) was presented by HLA-A3 to Nef-specific HLA-A3-restricted CTL lines, and the influenza nucleoprotein peptide NP(380-393) was presented by HLA-B27 to NP(380-393)-specific HLA-B27-restricted CTL lines. In addition, we have extended studies from our group that have evaluated the capacity of a similar panel of peptides to inhibit presentation of an influenza nucleoprotein peptide NP (335-349) by HLA-B37 and a matrix peptide, M1 (57-68), by HLA-A2 to the appropriate peptide-specific CTL lines. Out of 41 peptides tested, only five bound to more than one of the MHC molecules analyzed. Pairwise comparisons of the peptide binding specificities among these four different class I molecules revealed no common competitor peptides in four of the six possible comparisons. Thus, each class I molecule appears to have a functionally distinct peptide binding site, as reflected by the ability to bind largely non-overlapping sets of peptides.

Peptide binding to HLA-A2 and HLA-B27 isolated from Escherichia coli. Reconstitution of HLA-A2 and HLA-B27 heavy chain/beta 2-microglobulin complexes requires specific peptides.

The specificity of peptide binding by human leukocyte antigen (HLA) class I molecules was investigated in a cell-free direct-binding assay. Peptides were assessed for binding to HLA-A2 and HLA-B27 by measuring the formation of heterotrimeric HLA complexes that consisted of iodinated beta 2-microglobulin, HLA heavy chain fragments isolated from the Escherichia coli cytoplasm, and peptide. In this system, no detectable HLA heavy chain-beta 2-microglobulin complexes were formed unless appropriate peptides were intentionally added to the reconstitution solution. Analysis with monoclonal antibodies demonstrated that these heterotrimeric complexes were correctly folded. Five nonhomologous peptides, known to form complexes with HLA-A2 or HLA-B27 from T-cell functional studies, were tested for their capacity to bind to HLA-A2 and HLA-B27 using the reconstitution assay. Four of the peptides bound to the appropriate class I molecule only. One peptide and some (but not all) substitution analogs of it bound to both HLA-A2 and HLA-B27. The effect of peptide length on binding to HLA-B27 was studied, and it was found that the optimal length was 9 or 10 amino acid residues; however, one peptide that bound to HLA-B27 was 15 amino acids long. All peptides that bound to HLA-B27 in the direct-binding assay also competed with antigenic peptides for binding to HLA-B27 on the surface of intact cells, as determined by a standard cytotoxic T-lymphocyte functional assay. Thus, we conclude that HLA-A2 and HLA-B27 bind distinct but partially overlapping sets of peptides and that, at least in vitro, the assembly of HLA heavy chain-beta 2-microglobulin complexes requires specific peptides.

Overlapping epitopes that are recognized by CD8+ HLA class I-restricted and CD4+ class II-restricted cytotoxic T lymphocytes are contained within an influenza nucleoprotein peptide.

Viral epitopes that are recognized by both HLA class I-restricted and class II-restricted T cells have been defined for a type A influenza virus nucleoprotein (NP) peptide. CD8+ and CD4+ CTL lines have been generated against a synthetic peptide encompassing residues 335 to 349 of NP that are restricted by HLA-B37 and HLA-DQw5, respectively. Both of these CTL populations were capable of specifically lysing influenza A virus-infected targets, indicating that a naturally processed NP peptide(s) was being mimicked by the NP (335-349) peptide. Amino acid residues that are critical for recognition of this NP determinant in the context of HLA-B37 and HLA-DQw5 were investigated by the use of panels of truncated and alanine-substituted NP peptides. The results demonstrate that: 1) truncations in the amino- or carboxy-terminal ends differentially affect CD8+ and CD4+ CTL recognition; 2) the NP (335-349) sequence contains two octapeptide epitopes that share a core of six amino acid residues (NP 338-343); and 3) alanine substitutions at five of these residues abrogated recognition by at least one of the CD8+ and CD4+ CTL lines. Thus, these class I- and class II-restricted CTL lines recognize similar but distinct epitopes, and different structural features of the NP peptide are required for presentation by HLA-B37 and HLA-DQw5. Comparison of the amino acid sequences of the NP peptide presented by HLA-B37 and HLA-DQw5 with other peptides known to be presented by both class I and class II molecules revealed a common motif among these peptides.

HLA-B37 and HLA-A2.1 molecules bind largely nonoverlapping sets of peptides.

T-cell recognition of peptides that are bound and presented by class I major histocompatibility complex molecules is highly specific. At present it is unclear what role class I peptide binding plays relative to T-cell receptor specificity in determination of immune recognition. A previous study from our group demonstrated that the HLA-A2.1 molecule could bind to 25% of the members of a panel of unrelated synthetic peptides as assessed by a functional peptide competition assay. To determine the peptide-binding specificity of another HLA class I molecule, we have examined the capacity of this panel of peptides to compete for the presentation of influenza virus nucleoprotein peptide NP-(335-350) by HLA-B37 to NP-peptide-specific HLA-B37-restricted cytotoxic T-lymphocyte lines. Forty-two percent of peptides tested were capable of inhibiting NP-(335-350) presentation by HLA-B37. Remarkably, none of these HLA-B37-binding peptides belong to the subset that was previously shown to bind to the HLA-A2.1 molecule. Only the NP-(335-350) peptide was capable of binding to both HLA-A2.1 and HLA-B37. These findings demonstrate that the peptide-binding specificities of HLA-B37 and HLA-A2.1 are largely nonoverlapping and suggest that, from the universe of peptides, individual HLA class I molecules can bind to clearly distinct subsets of these peptides.