Mapping Immune Responses to mRBP-3 1-16 Peptide with Altered Peptide Ligands

Abstract

purpose. Experimental autoimmune uveoretinitis (EAU) can be induced in C57BL/6 mice (I-A^b) using human retinoid-binding protein-3 (hRBP-3, previously IRBP) residues 1-20. This study of a truncated murine peptide (mRBP-3 1-16) was conducted to determine its pathogenic potential and to characterize partially its interaction with specific T cells.

methods. After immunization with mRBP-3 1-16 or hRBP-3 1-20, EAU was assessed by immunohistochemistry. The immune response was assessed by tritiated thymidine incorporation and cytokine production analyzed by enzyme-linked immunosorbent assay (ELISA). T-cell receptor (TCR)- and major histocompatibility complex (MHC)-binding of mRBP-3 1-16 was studied by modeling and by using altered peptide ligands (APLs) and T-cell clones.

results. mRBP-3 1-16 induced EAU in C57BL/6 mice, with severity and kinetics comparable to that after immunization with hRBP-3 1-20. T cells taken from mice immunized with mRBP-3 1-16 had a Th1 phenotype and proliferated in response to reactivation with mRBP-3 1-16, hRBP-3 1-20, or mRBP-3 1-16 APLs. mRBP-3 1-16 APLs elicited at least five distinct patterns of reactivity when tested with the mRBP-3 1-16-reactive T-cell clones.

conclusions. mRBP-3 1-16 immunizes and causes EAU in C57BL/6 mice. The studies using T-cell clones and APLs demonstrate that the immune response to mRBP-3 1-16 is drawn from a diverse population of antigen-specific T cells with a Th1 phenotype. Modeling and analysis of clones indicate that nonpathogenic T cells of an mRBP-3 1-16-reactive T-cell line recognize the peptide in a single register.

Experimental autoimmune uveoretinitis (EAU) serves as a model for noninfective, intraocular inflammation in humans, which is believed in many cases to have an autoimmune etiology. The histopathology of EAU is characterized by inflammatory infiltration in the posterior segment of the eye, ¹ which varies, depending on species, strain, dose, and route of immunization. In mice, immunization leads to a CD4⁺ Th1 T-cell–mediated disease, ² and the most frequently used antigen is retinoid-binding protein (RBP)-3 (previously known as interphotoreceptor retinoid-binding protein [IRBP]). This is a 140-kDa extracellular matrix protein that transports vitamin A derivatives between the photoreceptor cells and the retinal pigment epithelium (RPE). ³ Immunization with human (h)RBP-3 peptide 1-20 is known to induce EAU in C57BL/6 (I-A^b) mice, and removal of residues 1-5 generates a peptide that is poorly immunogenic and unable to induce EAU. ⁴ Consistent with other models of autoimmune disease, it has been shown that the level of RBP-3 expression in the thymus affects the susceptibility of mice to the induction of EAU with this protein, ⁵ and that deleting RBP-3 reveals T-cell reactivity that is not detected in wild-type mice. ⁶ More detailed characterization of the immune response to RBP-3 is therefore relevant to studies of EAU. 

Although CD4⁺ T cells are critical to the induction of EAU, the infiltrating myeloid cells (macrophages) also play a key role in both tissue destruction and tissue repair during EAU. ⁷ Furthermore, studies of uveoretinitis and other organ-specific autoimmune diseases show that susceptibility is governed by many genes. Thus, these are complex conditions influenced by both genes and environment. 

To penetrate this complexity, it is important to have a complete understanding of the autoantigen-specific responses involved. We have therefore extended the analysis of the T-cell response to RBP-3 in C57BL/6 mice. We noted that the sequence of hRBP-3 1-20 differs at position 17 (valine) from that of murine (m)RBP-3 (isoleucine at position 17) and that the four carboxyl-terminal amino acids are predominantly hydrophobic. We therefore synthesized mRBP-3 1-16 and studied its pathogenicity. In this report, we show that mRBP-3 1-16 induced EAU equivalent to hRBP-3 1-20 and that having an amino or carboxyl terminus did not alter the ability of mRBP-3 1-16 to induce disease. Analysis of the T-cell response shows that it is diverse and includes nonpathogenic T cells whose receptors we have mapped by using altered peptide ligands (APLs). Using a combination of activation of mRBP-3 1-16-reactive T-cell clones with APLs, a binding matrix to analyze interactions with major histocompatibility complex (MHC) class II, and classic studies of MHC-peptide binding, we have identified residues within mRBP-3 1-16 that are important T-cell receptor (TCR) contacts. 

Materials and Methods

Mice

C57BL/6 mice were originally obtained from Harlan UK Limited (Oxford, UK) and were housed in specific pathogen-free conditions with continuously available water and food. Mice immunized for disease induction were aged between 6 and 8 weeks. Treatment of the animals conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Reagents

hRBP-3 peptide 1-20 (GPTHLFQPSLVLDMAKVLLD), the truncated mRBP-3 peptide 1-16 (GPTHLFQPSLVLDMAK) (with either an amino or carboxyl terminus), and the nine altered peptide ligands (APLs) of mRPB-3 1-16, F6Y (GPTHLYQPSLVLDMAK), Q7R (GPTHLFRPSLVLDMAK), Q7S (GPTHLFSPSLVLDMAK), P8A (GPTHLFQASLVLDMAK), S9Q (GPTHLFQPQLVLDMAK), L10Q (GPTHLFQPSQVLDMAK), L10W (GPTHLFQPSWVLDMAK), L12N (GPTHLFQPSLVNDMAK), and D13I (GPTHLFQPSLVLIMAK) were obtained from Sigma-Genosys Ltd. (Poole, UK). (Bold letters indicate the single amino acid changes in the sequence.) The control peptide, murine hepatitis virus (KVIAKWLAVNVL) was synthesized by Quality Controlled Biochemicals, Inc. (Hopkinton, MA). The peptide purity was determined by HPLC. Peptide preparations were aliquotted and stored at −80°C. Culture medium, fetal calf serum (FCS), and supplements were supplied by Invitrogen (Paisley, UK), unless otherwise stated. 

Lymph Node Cell Analysis

Mice were immunized subcutaneously in both flanks and the scruff of the neck with 100 μg/mouse of peptide diluted in PBS (100 μL/site) in emulsion with CFA (1 mg/mL, 1:1 vol/vol; Invitrogen). Draining lymph nodes were removed 10 days after immunization, and single cell suspensions were prepared. Primed lymph node cells (LNCs) were seeded at 5 × 10⁵ per 200 μL of DMEM supplemented with 10% FCS, 100 U/mL penicillin-streptomycin, 100 μg/mL gentamicin, 2 mM l-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 1× MEM vitamin mixture, 0.1 mM asparagine (Sigma-Aldrich, Dorset, UK), and 5 × 10⁻⁵M 2-mercaptoethanol (culture medium). Each condition was plated in 200 μL per well in 96-well, round-bottomed plates (Corning-Costar, Corning, NY). 

Lymphocyte Proliferation

Primed LNCs were stimulated with peptide and incubated for 2 days at 37°C in 10% CO₂ in a humidified atmosphere. Supernatants were aspirated at 48 hours for cytokine assays and frozen at −80°C, and the plates were pulsed with 18.5 kBq tritiated thymidine (GE Healthcare, Bucks, UK) per well for the last 18 hours of incubation. Cells were harvested with a 96-well harvester (Tomtec, Hamden, CT), and thymidine uptake (measured in counts per minute [cpm]) was determined by liquid scintillation with a microbeta liquid scintillation counter (Wallac 1450; PerkinElmer Life Sciences, Cambridge, UK). 

Cytokine Assays

Cytokine production (TNF-α, IFN-γ, IL-2, IL-4, and IL-10) in culture supernatants was assayed by capture enzyme-linked immunosorbent assay (ELISA). Briefly, on day 0, capture antibody (in carbonate buffer, pH 9.6) was applied to flat-bottomed, 96-well plates (Nunc Immuno Plate; Fisher Scientific, Leicestershire, UK) and left overnight at 4°C. On day 1, nonspecific binding sites were blocked using 1% bovine serum albumin (BSA; Sigma-Aldrich) in PBS (1% PBSA) for 1 hour at 37°C. The supernatants were then added for 1 hour at 37°C, and the plates were washed with 0.5% Tween 20 (Sigma-Aldrich) in PBS. The appropriate detection antibody was added for 1 hour at room temperature, and plates were washed as just described. Extra-avidin peroxidase (Sigma-Aldrich) was applied to the plates for 30 minutes at room temperature, plates were washed, and the chromogen substrate (3,3′, 5,5′-tetramethylbenzidine (TMB) and hydrogen peroxide [H₂O₂]; BD Biosciences, Oxford, UK) were added. The reaction was stopped with 2 N sulfuric acid (H₂SO₄) (Sigma-Aldrich). The monoclonal antibody pairs for capture and detection were obtained from PharMingen, along with the TNF-α ELISA kit. Recombinant cytokines (PharMingen) were used as standards, with curves generated from doubling dilutions used to calculate concentrations of cytokine in the test sample. All data shown were obtained from supernatants diluted 1:2 with culture medium, and values are therefore 0.5 of the final cytokine concentration. 

T-Cell Lines and Clones

Mice were immunized as described earlier, and lymph nodes were removed 10 days later. Single-cell suspensions were prepared, and 5 × 10⁶ cells/mL were seeded at 1 mL/well in 24-well plates (Corning-Costar). Cells were stimulated with 20 μg/mL mRBP-3 1-16 peptide on day 0 and then with 20 μg/mL mRBP-3 1-16 and 1 × 10⁵ irradiated, syngeneic spleen cells/well on days 7, 14, and 21, and once per month thereafter to generate the mRBP-3 1-16-reactive T-cell line. The T-cell line (derived from the lymph nodes of four C57BL/6 mice) was in culture for at least 2 months before the experiments. Single-cell T-cell clones were obtained from the T-cell line by plating at limiting dilutions, followed by expansion as described earlier. 

EAU Induction and Scoring

Mice were immunized subcutaneously in one flank with 230 nanomoles peptide in PBS, in emulsion with CFA (1 mg/mL; 1:1 vol/vol) supplemented with 1.5 mg/mL Mycobacterium tuberculosis complete H37 Ra (BD Biosciences), and also 1.5 μg Bordetella pertussis toxin (Sigma-Aldrich) intraperitoneally (IP). At various time points after immunization, eyes were collected and carefully snap frozen, oriented in optimal cutting temperature (OCT) compound (R. Lamb Ltd., East Sussex, UK). After they were made and stored at −80°C, serial 8-μm sections were thawed at room temperature and fixed in acetone for 10 minutes. They were stained with rat anti-mouse monoclonal anti-CD45 antibody or anti-F4/80, or rat anti-human CD3ε (Serotec, Oxford, UK) and counterstained with hematoxylin (ThermoShandon, Pittsburgh, PA). Sections were scored for inflammatory infiltrate (presence of CD45-positive cells) and structural disease (disruption of morphology), as described previously. ⁸  

Analysis with the I-A^b Scoring Matrix

Using their scoring matrix, we analyzed the set of unique I-A^b binding peptide sequences reported by Zhu et al. ⁹ and compared them with mRBP-3 1-16. Briefly, each peptide was scored in every possible register throughout its length, using a program written in FORTRAN. For each peptide tested, the highest value was taken to indicate the most likely binding register. The mean highest value was 86.6, and maximum scores for all the known I-A^b binding peptides ranged from 43 to 122. 

I-A^b MHC-Binding Assays

The mouse B-cell lymphoma LB27.4 was used as the source of murine I-A^b molecules. LB27.4 cells were maintained, and I-A^b molecules purified by affinity chromatography using the anti-I-A^b,s,u monoclonal antibody Y3JP, ¹⁰ as previously described. ¹¹ Quantitative peptide-I-A^b binding assays were based on the inhibition of binding of radiolabeled ROIV peptide (sequence YAHAAHAAHAAHAAHAA) ^{12 13} to purified I-A^b molecules. Assays were performed at pH 7.0 in PBS containing 0.7% digitonin, and in the presence of a protease inhibitor cocktail. ¹¹ MHC binding of the radiolabeled peptide was determined by capturing MHC/peptide complexes on Y3JP antibody–coated plates (Lumitrac 600; Greiner Bio-one, Frickenhausen, Germany) and measuring bound counts per minute (TopCount; PerkinElmer) microscintillation counter. The average IC₅₀ of ROIV was 28 nM. Any change less than threefold was regarded as insignificant, and peptides with affinities <1000 were not considered as binders. 

Results

Effect of mRBP-3 1-16 Immunization

EAU can be induced in C57BL/6 mice with the human RBP-3 1-20 (hRBP-3 1-20) peptide. Amino terminal truncation of this peptide (residues 1-5) produces an antigen that does not immunize or produce disease. ⁴ To characterize the uveitogenic epitope further, we synthesized a peptide truncated at the carboxyl terminus, to produce mRPB-3 1-16 (mRBP-3 1-16; Fig. 1A ) and tested whether this peptide was immunogenic. Mice were immunized with mRBP-3 1-16 or hRBP-3 1-20 peptide, and draining lymph node cells (LNCs) were prepared 10 days later. LNCs from mice immunized with each antigen were tested with both peptides (Fig. 1B) . Cells from animals immunized with either peptide respond to both peptides, with similar dose responses, showing that they recruit and expand broadly overlapping populations of antigen-specific T cells. The pattern of cytokine production by ex vivo LNCs from mice immunized with either peptide was similar, in that it showed a Th1 type response. We detected IFN-γ and IL-2 but not IL-4 or IL-10 in the supernatants taken 48 hours after activating the T cells (data not shown). 

To characterize the immune response to mRBP-3 1-16 further, we established a T-cell line from mice immunized with this peptide, which was then expanded in tissue culture with the same peptide. When the line was established and growing stably (2 months in culture) we assessed proliferation and cytokine production. This demonstrated that the cells were peptide specific and had the same Th1 phenotype as the primary LNCs (no detectable IL-4 or IL-10, data not shown; Fig. 1C ), both features of T cells with the potential to induce organ-specific autoimmune disease, raising the possibility that mRBP-3 1-16 may induce uveoretinitis. Despite this phenotype, we found it difficult to induce EAU by transferring short-term T-cell lines from mice immunized with mRBP-3 1-16 and therefore wondered if these cells were truly pathogenic. To investigate this further, we attempted to induce EAU with mRBP-3 1-16. 

Induction of EAU in C57BL/6 Mice

C57BL/6 mice were immunized subcutaneously with 230 nanomoles each of mRBP-3 1-16 (with a C-terminal amide), hRBP-3 1-20, or mRBP-3 1-16 COOH (mRBP-3 1-16 with a carboxyl terminus) emulsified in complete Freund’s adjuvant (CFA), with Pertussis toxin intraperitoneally (as described in the Materials and Methods section). At a range of time points after immunization, eyes were snap frozen and stained with anti-CD45 antibody. Representative histology from a normal eye and eyes at day 21 after immunization with mRBP-3 1-16, hRBP-3 1-20, or mRBP-3 1-16 COOH is shown (Fig. 2A) . Three sections per condition were scored for both inflammatory infiltrate and changes in structural morphology. ⁸ CD45-positive cellular infiltrate and structural disruption were noted throughout the time course, demonstrating that mRBP-3 1-16 can induce EAU (Fig. 2B) . The peak of disease occurred at day 21 after immunization and the incidence of disease after mRBP-3 1-16 immunization was 100% (data not shown). Disease after mRBP-3 1-16 immunization mimicked that which occurred after hRBP-3 1-20 immunization, in that photoreceptor destruction occurred, and infiltrate was observed around interretinal vessels and in neuronal layers. Kinetics of disease were also similar to that observed previously with hRBP-3 1-20. ^{14 15}  

Sections were scored for EAU at the peak of disease (day 21 after immunization) after immunization with each of the three RBP-3 peptides (Fig. 2C) . Severity with regard to CD45-positive infiltrate and inflammation was comparable after immunization with each of the peptides. Further histology studies demonstrated that the colocalization of macrophages and T cells, determined by staining with anti-F4/80 or anti-CD3ε on day 21 after mRBP-3 1-16 (Fig. 2Di)or hRBP-3 1-20 (Fig. 2Dii)immunization, were similar. Thus, by these criteria, the disease induced with mRBP-3 1-16 and hRBP-3 1-20 is indistinguishable. We conclude that after immunization, mRBP-3 1-16 is an effective uveitogenic antigen in C57BL/6 mice. 

Given the Th1 phenotype of the mRBP-3 1-16-reactive T-cell line, we also tested whether adoptive transfer of these T cells into normal C57BL/6 recipients would lead to the development of uveoretinitis. We conducted four separate experiments—three times with 1 × 10⁷ and once with 1.5 × 10⁷ mRBP-3 1-16-reactive T cells/mouse intravenously injected, activated 3 days previously with 20 μg/mL mRBP-3 1-16 in medium with 1% normal mouse serum taken from C57BL/6 mice. Histologic examination of eyes on days 6, 11, and 15 after transfer showed no disease (data not shown). 

Having established that, after immunization, mRBP-3 1-16 induced EAU in C57BL/6 mice, but that a T-cell line did not, we were interested to determine which residues within the mRBP-3 1-16 peptide play a role in T-cell activation, so that mechanisms of disease induction (MHC-binding and TCR-ligation) could be further characterized. To investigate which residues were involved in MHC binding, we analyzed the mRBP-3 1-16 epitope with a scoring matrix based on earlier studies of the structure of I-A^b. 

Analysis of mRBP-3 1-16 Epitope

To understand the T-cell response to mRBP-3 1-16 we considered how the peptide might interact with MHC class II. We used a study of peptide binding to I-A^b in which the authors generated a matrix that allocates a score to any amino acid at every position. ⁹ This matrix produces a total score for any 9-mer peptide fitted to I-A^b. The main interactions with the MHC class II molecule are with six pockets (P1, P3, P4, P6, P7, and P9). When peptides are tested in several different registers, the highest score predicts the most favorable alignment. When a set of known I-A^b binding peptides is used, this algorithm successfully predicts binding alignment in 93% of cases. ⁹  

To elucidate how mRBP-3 1-16 might bind to I-A^b we reanalyzed the unique core sequences reported by Zhu et al. ⁹ and compared these with scores generated by analysis of the mRBP-3 1-16 peptide binding in all possible registers. Maximum binding scores for the set of known I-A^b binding peptides ranged from 43 to 122. When we analyzed mRBP-3 1-16 we found three binding registers with scores greater than 43. The highest score was generated by the mRBP-3 1-16 peptide binding with threonine (T3) in the P1 pocket, but this would place a phenylalanine (F6) in the P4 pocket, which is a forbidden interaction. On the basis of the motif scores, we therefore concluded that the native mRBP-3 1-16 peptide could bind with leucine (L5, score 64: register A) and/or phenylalanine (F6, score 63: register B) in P1 (Fig. 3) . 

These results suggested that the core of the peptide recognized by T cells lies between L5 and M14, but it also raised the possibility that the mRBP-3 1-16 could bind in two different registers. To clarify this further, we generated altered peptide ligands (APLs) with single amino acid substitutions in the central binding region, measured the binding of these peptides to I-A^b in competition assays, and examined the immune response they elicited. 

mRBP-3 1-16 APL Immunization of C57BL/6 Mice

Nine APLs of mRBP-3 1-16, each with a single amino acid substitution, were synthesized (F6Y, Q7R, Q7S, P8A, S9Q, L10Q, L10W, L12N, and D13I). These APLs encompass the putative core recognition motif of the peptide except position 11, which is an MHC-binding residue in either register. Mice were immunized with 100 μg/mouse mRBP-3 1-16 or each of the APLs. Ten days after immunization, recall responses were tested by stimulation of LNCs with 20 μg/mL control peptide or immunizing peptide. Figure 4Ashows the proliferation of the LNCs stimulated with the same peptide that was used to immunize the mice. Cells from mice immunized and reactivated with mRBP-3 1-16 or L10W showed the highest levels of proliferation. This finding indicated that a change from L to W at position 10 did not reduce recall responses, and we infer that this residue is not critical in MHC binding (binding relative to mRBP-3 1-16, 0.38). F6Y, P8A, and S9Q all immunized and induced a recall response, although the level of proliferation was lower than that with mRBP-3 1-16 or L10W. It is interesting that P8A and S9Q are respectively the worst (relative binding, 0.005) and the best (relative binding, 19.4) MHC-binding peptides. The ability of poorly binding peptides to elicit immune responses is well recognized for autoantigens. ^{16 17} There was minimal proliferation after recall with Q7R, indicating that this substitution significantly alters MHC binding, and this is confirmed by binding studies (Fig 4B) . However, a second substitution at this position, Q7S, bound and immunized better. L10Q bound better than L10W, but induced a lower recall response relative to mRBP-3 1-16. L12N bound better than mRBP-3 1-16, although this difference was not significant, whereas the binding of D13I was significantly reduced. We then compared the recall responses of cells taken from animals immunized with the different APLs to restimulation with the immunizing ligand or mRBP-3 1-16. For clarity Figure 4Bshows the IFN-γ production by these LNCs, but the proliferation data showed the same pattern (data not shown). We also show the relative binding of I-A^b MHC when compared with mRBP-3 1-16. It is striking that immunization with peptides with substitutions in the amino-terminal half of mRBP-3 1-16 induced a population of cells that were much more cross-reactive with mRBP-3 1-16 than did immunization with peptides with substitutions in the carboxyl-terminal half of mRBP-3 1-16. 

Next, to address which residues in mRBP-3 1-16 are important in TCR recognition, a panel of 15 mRBP-3 1-16-reactive T-cell clones were generated and activated with mRBP-3 1-16 or one of the nine APLs. 

Profiles of mRBP-3 1-16-reactive T-Cell Clone Activation with mRBP-3 1-16 APLs

Fifteen T-cell clones, derived from LNCs taken from mice immunized with mRBP-3 1-16, were activated with mRBP-3 1-16 and mRBP-3 1-16 COOH or one of each of the nine mRBP-3 1-16 APLs. Flow cytometric analysis of 14 of the clones with a panel of eight anti-Vβ antibodies (Vβ2, -3, -5, -6, -7, -8, -11, and -14) and anti-CD4 was performed. All clones tested stained positive for CD4. Clones stained positive for Vβ3, -6, -7, or -11 (data not shown), indicating that the T cells were drawn from a polyclonal population and not restricted to a limited TCR repertoire. We did not find a significant association between Vβ expression and patterns of reactivity. We then considered two possibilities to explain the reactive patterns of these clones: they might recognize the mRBP-3 1-16 peptide bound in a single register, in which case the response patterns of APLs would be predicted to show some uniformity, or the clones may be divided into some that recognize peptide bound in one register and others that are stimulated by peptide bound in a second register, in which case we would predict no uniformity in APL activation. We found that there were at least five distinct patterns (shown by two or more clones) of proliferation and IFN-γ and IL-2 production after stimulation of the T-cell clones. Four of these are illustrated (Fig. 5) , the fifth group of clones recognized mRBP-3 1-16 and F6Y only. Clone 1B10, and one other, proliferates above background and produces IFN-γ and IL-2 after stimulation with mRBP-3 1-16, mRBP-3 1-16 COOH, F6Y, and L12N (Figs. 5A and 5E) . The second pattern, shown by 1F7 and four other clones, was a low-level response to mRBP-3 1-16 COOH activation, activation with F6Y that was greater than with mRBP-3 1-16 (an average of more than 50 × 10³ cpm higher than activation with mRBP-3 1-16) and a response to Q7S. 1F7 alone also responds to D13I, although this APL induced much less IFN-γ than the other ligands (Figs. 5B 5F) . IFN-γ production after F6Y stimulation was also higher than that after stimulation with mRBP-3 1-16 (Fig. 5Bii) . Clone 1F3 and one other clone were the most broadly cross-reactive and showed equivalent levels of proliferation after activation with mRBP-3 1-16, mRBP-3 1-16 COOH, F6Y, Q7S, P8A, and L10Q and more variable responses to L12N (Figs. 5C 5G) . Although L10Q reproducibly induced proliferation, this ligand did not elicit equivalent levels of IFN-γ and IL-2 compared with the other peptides. The final pattern, represented by T-cell clone 1E11 and shared with one other clone, was the response to F6Y and a low but clearly reproducible level of proliferation in response to stimulation with S9Q (Figs. 5D 5H) . There was no detectable IL-2 production from this T-cell clone, but IFN-γ production mirrored the level of proliferation, with a low level observed as a result of S9Q activation (Fig. 5Dii) . 

In these experiments, all 15 mRBP-3 1-16-reactive T-cell clones proliferated in response to stimulation with mRBP-3 1-16, mRBP-3 1-16 COOH, and F6Y to a level higher than that seen with control peptide. This uniform response to F6Y suggests that all the T-cell clones we tested recognize mRBP-3 1-16 bound in a single register. 

Some of the T-cell clones also proliferated in response to P8A, S9Q, Q7S, L10Q, L12N, or D13I, but none of these clones proliferated or produced cytokines in response to Q7R or L10W, despite the fact that immunization with L10W induced good T-cell responses (Fig. 4) . We conclude that in the T-cell line there are clones that respond to the APLs P8A, S9Q, Q7S, L10Q, L12N, or D13I, but not to Q7R or L10W. Although the results with Q7R might be explained by its being a relatively poor MHC-binding peptide (although much better than P8A), L10W does not bind significantly differently to the cognate ligand. All these T-cell clones appear to recognize the peptide bound in a single register. 

Discussion

In this report, truncation of the hRBP-3 1-20, to produce mRBP-3 1-16, generated a peptide that was immunogenic and uveitogenic in C57BL/6 (I-A^b) mice. We found that the population of T cells that responded to mRBP-3 1-16 was diverse, as judged by the Vβ expression and response of a panel of 15 T-cell clones but that all the clones tested recognized the peptide bound in a single register. Disease after mRBP-3 1-16 immunization has histology, severity, and kinetics comparable to that after immunization with, hRBP-3 1-20, ^{14 15} but the shorter peptide is more soluble in aqueous solutions and is therefore easier to handle. 

Immunization with mRBP-3 1-16 or hRBP-3 1-20 leads to the production of a population of cells that proliferate in response to either peptide in vitro. A mature T-cell line reactive with mRBP-3 1-16 has a Th1 phenotype, but adoptive transfer of these cells does not induce EAU in C57BL/6 mice. Isolating mRBP-3 1-16-reactive cells from mRBP-3 1-16-immunized mice and culturing these cells in vitro, may have led to the selection of a T-cell line that comprises mainly cells that express TCRs that can no longer recognize the mRBP-3 epitope processed from whole RBP-3 protein, possibly because this epitope binds in a different register than that of exogenous peptide. Populations of T cells that can only recognize exogenous peptides have been termed “type B” T cells. ¹⁸ It is likely that type B T cells have a growth advantage in cell culture because their TCRs have a higher avidity than TCRs reactive with the processed native peptide, which mediates negative selection in the thymus. ¹⁹ However, it is clear that pathogenic cells must be generated by immunization with mRBP-3 1-16 and therefore this peptide must have the capacity to imitate the binding of processed antigen, which may allow for successful modulation of disease-relevant T cells with this peptide. ²⁰  

We have used binding studies, mRBP-3 1-16, APLs, and mRBP-3 1-16–reactive T-cell clones to dissect the TCR-peptide-MHC interaction and predict in which register the native peptide may bind to the MHC. Analysis of potential binding motifs showed that the wild-type peptide can bind equally well in either of two registers, with a shift of a single amino acid (Fig 3) . The binding studies (Fig. 4B)revealed that substitution of proline at position 8 and aspartate at position 13 significantly reduced binding compared with mRBP-3 1-16, which implies that the second substitution interfered with binding in the P9 pocket (register A). However, when we tested a panel of T-cell clones with different APLs, we found that all clones also responded to the APL F6Y, which is most consistent with their recognizing the peptide bound in register B. The binding motif analysis also indicates that the F6Y peptide is more likely than mRBP-3 1-16 to bind in register B (data not shown) and the uniform response F6Y elicits leads to the conclusion that all the clones also recognize mRBP-3 1-16 in register B. Therefore, for the clones that we have analyzed, we propose that L10 and D13 are the most important TCR contacts. Q7 is a secondary TCR contact, because the Q7S APL activates about half of the T-cell clones. Taking all the data together we hypothesize that naturally processed mRBP-3 1-16 binding is predominantly in register A, and therefore high-affinity TCRs that recognize the peptide in this register are deleted in the thymus. TCRs with the potential to recognize peptide in register B escape deletion, but immunization with free peptide allows the expansion of these cells in the periphery, and because these TCRs are higher affinity, they have a growth advantage in tissue culture. A similar mechanism has been identified in EAE induced with myelin basic protein (MBP). ²¹ An additional possible interpretation of the data obtained from immunization with L10W and with testing the panel of T-cell clones, is that L10W binds in a different register to all the other peptides. The previously reported observation that truncation of the first five amino acids of the human RBP peptide may be explained by a requirement for the presence of P-3 to P-1 for stable peptide binding. ⁹  

These studies are important because if we are interested in using APLs to induce a protective response in disease, we must have an understanding of the recognition of the native peptide before we can make educated alterations to its sequence. It has been shown in experimental autoimmune encephalomyelitis (EAE) models, that peptide analogues of autoantigens can inhibit disease, ^{22 23 24} but use of synthetic APLs as immunotherapy in EAU has been less well explored. A study investigating an MHC anchor-substituted variant of proteolipid protein (PLP, an autoantigen used to study EAE) peptide 139-151, termed 145D, rendered PLP 139-151–specific T-cell lines anergic in vitro. Administration of 145D in vivo before PLP 139-151 challenge led to a significant reduction in disease severity and incidence, and, importantly, 145D had the ability to reduce established EAE. ²⁵ This approach to using APLs (or more specifically MHC anchor-substituted variants) has several advantages over using APLs in which a TCR contact has been substituted. It is suggested that MHC anchor-substituted variants can tolerize a polyclonal population of autoreactive T cells and do not lead to the generation of T cells specific for the variant and that there was no observation of adverse reactions associated with the variant. ²⁵ Alterations in peptide sequence that lead to the production of peptides that are no longer bound by the I-A^b MHC, resulting in no binding affinity, would not be expected to be of any use in tolerance induction or bystander suppression by T regulatory cells, as antigen-presenting cells (APCs) cannot present the peptide to T cells in C57BL/6 mice. In a previous study, researchers have investigated altered recognition of human RBP-3 epitopes in spleen cells taken from C57BL/6 mice immunized with recombinant hRBP-3 repeat 1, then stimulated in culture with a panel of overlapping synthetic peptides. ⁶ That study found that mRBP-3 1-20 induced poor recall responses, which may be partly due to the complex pattern of binding, a recognition we describe with this peptide. 

In summary, in our study mRBP-3 1-16 immunization caused EAU in C57BL/6 mice. We found that the dominant component of a T-cell line recognizes the peptide in a single register. However, these T cells do not appear to be pathogenic, possibly because they recognize peptide bound in an alternative register. There are several possible explanations for this, however. One is that cells have lost the ability to traffic to the target organ ²⁶ and another is that we selected type B cells that do not respond to naturally processed peptide ¹⁸ binding in the same register. A further consideration is that it has been shown that binding in alternate registers can play a role in determining the balance between tolerance and susceptibility to autoimmunity. ²¹ It will therefore be interesting to determine how mRBP-3 1-16 is recognized by an as yet elusive pathogenic T-cell clone. 

 Supported in part by a grant from the National Eye Research Centre. CJG is funded by a University of Bristol studentship.

 Submitted for publication July 27, 2005; revised January 11, and February 1, 2006; accepted March 22, 2006.

 Disclosure: C.J. Guyver, None; D.A. Copland, None; C.J. Calder, None; A. Sette, None; J. Sidney, None; A.D. Dick, None; L.B. Nicholson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Peptide sequences, mRBP-3 1-16 immunization, and mRBP-3 1-16-reactive T-cell line phenotype. (A) Single amino acid sequences of hRBP-3 1-20, mRBP-3 1-20, and mRBP-3 1-16. (B) Proliferation of draining LNCs taken from C57BL/6 mice immunized 10 days previously with mRBP-3 1-16 or hRBP-3 1-20 (100 μg/mouse; two animals/peptide) in response to stimulation in vitro with mRBP-3 1-16 (▪) or hRBP-3 1-20 (▴) or control peptide (▾) at the concentrations indicated. Data are representative of two separate experiments, and each condition was plated in triplicate. (C) Proliferation and cytokine production of mRBP-3 1-16-reactive T-cell line in response to stimulation with mRBP-3 1-16 (▪) or control peptide (▾) at the concentrations indicated. Data are representative of two separate experiments. For proliferation, replicates of six were plated for each condition. Data are the mean counts per minute ± SE. For cytokine production (nanograms per milliliter) each condition was plated in duplicate and data represents the mean cytokine concentration measured by ELISA.

Figure 2.

 

Histology after disease induction with 230 nanomoles mRBP-3 1-16, hRBP-3 1-20, or mRBP-3 1-16 COOH (counterstained with hematoxylin) and disease scores. (A) Representative 8-μm sections from two separate experiments 21 days after immunization. Photographs show sections stained with anti-CD45 antibody from an untreated mouse (Normal) and mice immunized with 230 nanomoles mRBP-3 1-16, hRBP-3 1-20, or mRBP-3 1-16 COOH. (B) Mice were immunized with mRBP-3 1-16, eyes were harvested at the days after immunization indicated, and average disease score ± SE of inflammatory infiltrate and structural disease is shown. EAU developed in all animals in each group. n ≥ 6/time-point, data representative of two separate experiments. (C) Mice were immunized and eyes were taken at 21 days after immunization. Average disease score ± SE of inflammatory infiltrate and structural disease is shown. EAU developed in all animals in each group. n ≥ 6/peptide, data representative of two separate experiments. (D) Representative anti-F4/80 (F4/80) and anti-CD3ε (CD3) antibody staining (along with isotype control; No stain) of eyes at day 21 after immunization with mRBP-3 1-16 (Di) or hRBP-3 1-20 (Dii). Bar: (A) 200 μm; (D) 50 μm.

Figure 3.

 

Permitted binding motifs. Predicted binding of mRBP-3 1-16 to I-A^b reveals two potential registers (register A and B) in which the peptide binds. The squares indicate the location of MHC pockets (P1, P3, P4, P6, P7, and P9) in both registers.

Figure 4.

 

Immunization and recall with APLs of mRBP-3 1-16 and relative MHC-binding. (A) Two mice per group were immunized subcutaneously with 100 μg/mouse mRBP-3 1-16, F6Y, Q7R, P8A, S9Q, or L10W, and in a separate experiment mRBP-3 1-16, Q7S L10Q, L12N, or D13I. Draining lymph nodes were removed 10 days after immunization and primed LNCs were activated in vitro with 20 μg/mL of control peptide or immunizing peptide. The graph shows Δ mean counts per minute (cpm after recall with the immunizing peptide minus cpm after activation with the control peptide). Data shown is representative of two separate experiments of each set of results and each condition was plated in duplicate. (B) Supernatants were taken from the LNCs in the proliferation assay and IFN-γ production was determined after immunization with the indicated peptides. Recall with mRBP-3 1-16 (mRBP-3 1-16) or the immunizing peptide (Cognate) was then measured by ELISA. The graphs show Δ mean cytokine concentration (cytokine production after recall with the immunizing peptide minus cytokine production after activation with the control peptide) of duplicates. Data are representative of two separate experiments. Also shown in the I-A^b MHC binding by these peptides relative to mRBP-3 1-16.

Figure 5.

 

Stimulation of mRBP-3 1-16-reactive T-cell clones with mRBP-3 1-16 APLs. (Ai–Hi) mRBP-3 1-16-reactive T-cell clones 1B10, 1F7, 1F3, and 1E11 were stimulated with 20 μg/mL control peptide (Control), mRBP-3 1-16 (1-16 NH or 1-16), mRBP-3 1-16 COOH (1-16 COOH), or one of the five APLs. Graphs show mean cpm ± SE; each condition was plated in triplicate and data are representative of results in two separate experiments. (A–Hii) Supernatants were taken from the T-cell clones in the proliferation assays and cytokine production after ELISA was determined. Graphs show mean cytokine concentration. Each condition was plated in duplicate, and data are representative of results in two separate experiments. NT, not tested.

The authors thank David Wraith for a critical reading of the manuscript.