|Home | About | Journals | Submit | Contact Us | Français|
Direct identification and isolation of antigen-specific T cells became possible with the development of “MHC tetramers”, based on fluorescent avidins displaying biotinylated peptide-MHC (pMHC) complexes. This approach, extensively used for MHC class I–restricted T cells, has met very limited success with MHC class II tetramers (pMHCT-2) for the detection of CD4+ specific T cells. In addition, a very large number of these reagents while capable of specifically activating T cells after being coated on solid support, are still unable to stain. In order to try to understand this puzzle and design usable tetramers, we examined each parameter critical for the production of pMHCT-2 using the I-Ad-OVA system as a model. Through this process the geometry of pMHC display by avidin tetramers was examined, as well as the stability of recombinant MHC molecules. However, we discovered that the most important factor limiting the reactivity of pMHCT-2 was the display of peptides. Indeed, long peptides, as presented by MHC class II molecules, can be bound to I-A/HLA-DQ molecules in more than one register as suggested by structural studies. This mode of anchorless peptide binding allows the selection of a broader repertoire on single peptides and should favor anti-infectious immune responses. Thus, beyond the technical improvements that we propose, the redesign of pMHCT-2 will give us the tools to evaluate the real size of the CD4 repertoire and help us in the production and testing of new vaccines.
The adaptive immune response is characterized by the selective activation and expansion of a very limited number of antigen-specific precursor cells. Studying the repertoire and the dynamics of T cell responses as well as their functional and phenotypic characteristics is determinant for the development of new therapy against cancer, infectious or autoimmune diseases. Accessing individual antigen-specific T cells ex and in vivo has become possible with the development of peptide-MHC complexes tetramers (pMHCT). Indeed, multimerization allows to circumvent the low affinity of pMHC/T cell receptor (TCR) interactions by taking advantage of the avidity effect, without losing the exquisite antigen specificity of T cell recognition (1, 2). Therefore, pMHCT appear to be the ideal tool to study both CD4+ and CD8+ T cell responses. As such, class I pMHCT (pMHCT-1) have been used successfully for the identification, enumeration and phenotypic characterization of CD8+ T cell responses in a large number of antigenic systems (3, 4). In contrast, class II pMHCT (pMHCT-2) have had a very limited number of applications (4-7). This discrepancy has puzzled immunologists for the past 10 years and has been attributed to a wide variety of reasons: i) difficulty in the preparation of soluble pMHC class II complexes, ii) low frequency of antigen-specific CD4+ T cells, iii) low affinity of MHC class II binding peptides, iv) low affinities of TCR/pMHC class II binding, v) activation status of CD4+ T cells, vi) stringency in the geometry of display. Some of these issues have been successfully addressed; the intrinsic instability of MHC CLASS II molecules was overcome by the addition of leucine zippers at the C-terminus of both α– and β-chains to optimize pairing (8); the low affinity of MHC class II/peptide binding was answered by covalently linking the peptide to the β-chain to facilitate association (9); the low frequency of CD4+ specific T cells was overcome by enrichment techniques using magnetic beads (10). Artificial APCs built with liposomes decorated with pMHC class II molecules have also been used to increase avidity. However, they remain difficult to use and are confined to laboratories familiar with the technology (11). In the end, the technology of pMHCT-2 remains unreliable and the results of pMHCT-2 staining, positive or negative, remain difficult to interpret.
To address some of the fundamental issues that may explain this situation and separate de facto the worlds of pMHCT-1 and 2, we have taken advantage of the anti-ovalbumin response in H-2d mice. Indeed, I-Ad-OVA323-339 pMHCT-2 reagents do not “work” in this antigenic system (e.g. no immunostaining) despite being capable of activating I-Ad-OVA323-339 specific T cells such as DO11.10. In the present study, each of the 3 main components of a pMHCT-2 molecule was critically re-evaluated: geometry, stability and peptide display. While re-engineering of the MHC αβ interface dramatically increased stability and yield of recombinant I-Ad molecules, peptide display appeared as the single most important characteristic to make a pMHCT-2 operational. Re-engineering tethered peptides to limit their association to MHC to a single register allowed the production of pMHCT-2 capable of staining both activated and naïve OVA323-339 -specific T cells. Thus, beyond the technical improvements that we propose, the re-design of pMHCT-2 will give us the tools to evaluate the real size of the CD4 repertoire and help us in the production and testing of new vaccines.
DO11.10 hybridoma T cell, specific for the peptide 323–339 of the Chicken Ovalbumin protein in the context of I-Ad, was a gift from Dr. S. Webb (TSRI). These cells were maintained in RPMI 1640 (Invitrogen, Carlsbad, CA) supplemented with 10% FCS, 2 mM L-glutamine, 1 mM HEPES buffer and 1× MEM-NEAA.
DO11.10 TCR- (12), and OT-2 TCR-transgenic mice (13) were kindly provided by Dr. L. Sherman (TSRI) and Dr. C. Surh (TSRI), respectively. BALB/c mice were bred at The Scripps Research Institute Animal Facility. BALB/c mice were immunized in the footpad and tail base with 10 μg peptide or 200 μg of ovalbumin protein in CFA unless indicated otherwise. Draining lymph nodes (LNs) were harvested 8 days after immunization and single-cell suspensions were prepared for flow cytometry analysis (FACS).
Soluble MHC class II molecules were prepared as previously reported (14). Biotinylatable molecules were produced by adding a biotinylation sequence to the C-terminus of the α-chain following the acidic zipper. The I-Ad molecules were produced with two different peptides: chicken OVA323–339 (15) or influenza HA126-138 (16). After purification, MHC class II molecules were biotinylated with the BirA enzyme according to the manufacturer's instructions (Avidity, Denver, CO). Biotinylation was measured by immunodepletion using SA-agarose beads and SDS-PAGE. Biotinylated molecules were kept at 4°C and tetramerized with PE-labeled SA (BioSource International) just before flow cytometry.
The DO11.10 TCR Fv construct was engineered as already published (17) and transfected into the Escherichia coli strain BL21 Codon+ (Stratagene). Protein was extracted from bacteria with 50 mM Tris–HCl (pH 8.0), 0.37 M sucrose, 1 mM EDTA, dialysed against 10 mM Tris–HCl (pH 8.0), 0.4 M NaCl, for 48h before capture on Ni-agarose beads and elution of the Histidine-tagged bound protein with PBS-200 mM imidazole. Soluble protein was finally purified by size-exclusion chromatography on a Superdex 75 column (Amersham Biosciences).
Organs from several mice were harvested and pooled in complete RPMI (10% FCS, 10 mM Hepes, 2 mM L- Glutamine). Splenocytes were isolated on cell strainer in complete RPMI. Red blood cells were removed by 3 min incubations at room temperature in NH4Cl 0.165 M followed by two washes in complete RPMI and one in FACS buffer (PBS, 2% FCS, 25 mM Hepes, 2 mM EDTA). Popliteal and inguinal LNs were simply mashed on cell strainer and washed once in FACS buffer.
Single-cell suspensions were first incubated for 15 min at 4°C in 5 μg/mL anti-CD16/CD32 (FcBlock) in FACS buffer. Tetramer staining was performed at room temperature for 1h with 75μg/mL tetramers in 100 μl FACS buffer (PBS, 2% FCS, 25 mM Hepes, 2 mM EDTA), followed by a 15-min staining at 4°C with APC-labeled anti-CD8α (53-6.7) -B220 (RA3-6B2), -CD11c (M1/70), -CD49b (DX5) (APC channel: exclusion), FITC-labeled anti-CD3ε (145-2C11) and APC*Cy7-labeled anti-CD4 (GK1.5) antibodies. Propidium iodide (1μg/ml) was used for exclusion of dead cells. For magnetic sorting, the tetramer incubation was followed by an incubation with anti-PE microbeads (Miltenyi Biotech). Automatic positive separation of tetramer-PE positive cells was carried out with an AutoMACS cell separator (Miltenyi Biotech). Depleted fractions were processed following the same protocol. Events were acquired on a LSRII flow cytometer (BD Biosciences, Mountain View, CA) and analyzed using the FlowJo software (Tree Star, Ashland, OR). Single-cell suspensions prepared from DO11.10-transgenic mice or immunized mice were stained using the same protocol. All antibodies were purchased from Biolegend.
Hybridoma T cells (4×105 cells/well) were stimulated at 37°C for 5 or 24h with various concentrations of plate-bound MHC class II molecules in PBS. Supernatants were harvested and tested for the presence of IL-2 by assessing the proliferation of an IL-2-dependent cell line. [3H]Thymidine (0.5 μCi) was added to each well after a 24h incubation. Radioactivity in the cellular fraction was then measured 16h later. Results are expressed as mean cpm of triplicates (+/-SD).
Cells sorted with I-Ad-OVA or I-Ad-HA tetramer-PE were spun on glass slides using a cytospin centrifuge (600 rpm, 4 min) and fixed for 10 min in 4% paraformaldehyde before successive staining with WGA-AF588 and DAPI (Invitrogen, Carlsbad, CA). Slides were washed in FACS buffer and mounted using a Prolong Gold antifade reagent (Invitrogen, Carlsbad, CA). Images were acquired using a Bio-Rad MRC1024 laser-scanning confocal microscope fitted with a krypton/argon mixed gas laser (Richmond, CA).
Affinity binding experiments of His-tagged I-Ad complexes for DO11.10 Fv TCR were done by using a BIAcore 2000 (GE-Biacore, NJ). Two to four hundred resonance units (RUs) of DO11.10 Fv TCR or pMHC were covalently bound to a Biacore CM5 sensor chip using standard amine coupling. Analyte was injected onto the surface at a flow rate of 20μl/min in successive twofold dilutions from 10 or 20 μM. No regeneration procedure was necessary because of the low affinity. Experimental curves obtained at each concentration were normalized by subtraction from a reference surface. Sensorgrams were analyzed using BIAevaluation 3.0 software and fit globally to determine kinetic constants. The suitability of the fit was evaluated based on ×2 values and the appearance of residuals.
Display on an avidin platform is rigid and exposes the four MHC molecules in the four cardinal directions (11, 18, 19). Such a module limits the accessibility of multiple pMHCs from a single tetramer when binding to a cell surface. We thought of interrogating this issue by modifying the site of addition of the biotinylation sequence on the pMHC. The ingenuity of this idea seemed to be validated by a serendipitous observation in which a working H-2Kb tetramer was rendered inefficient by moving the site of biotinylation from the “bottom” (C-terminus) to the side (between α2 and α3 domains) of the molecule (EL and LT, unpublished observation). In order to display the pMHC off of the avidin molecule in as many positions as possible, we introduced the BirA biotinylation site in each of the six external loops of the α2 domain (Fig. 1). The biotinylation sequence was introduced in between two GlySerGlySer linkers at position 65,127,152,158,185 and 198 (Fig. 1). The modified α chains were transfected with cDNAs coding for I-Ad β chains tethered to the OVA323-339 or influenza hemagglutinin HA126-138 peptides into S2 fly cells. All the variants were expressed (Fig. 2A), and could be efficiently biotinylated (not shown). With the exception of the biot65 mutant, all MHC class II monomers were functional and comparable to the original construct with a C-terminal biotinylation site, as tested in a plate-bound assay using DO11.10 hybridoma cells as reporter (Fig. 2B). The anion exchange profile of the biot65 monomers indicated that it was probably misfolded. When the five functional variants were tested as tetramers for FACS staining of splenocytes from DO11.10 TCR transgenic mice, no staining was obtained (Fig. 2C). Similar negative results were obtained after footpad immunization with OVA323-339 peptide (not shown). These results indicated that the geometry of the tetramers, and the orientation of the monomers on the streptavidin, were not likely to be critical for the ability of pMHCT-2 to stain T cells.
A review of the literature on pMHCT-2 highlights that most of the successful reagents are among the most stable MHC CLASS II molecules, I-E in the mouse and HLA-DR in human, namely (20-22). We have already reported that improving αβ pairing by adding zippers was determinant for the expression of I-A molecules (8). However, this mode of stabilization is outside the αβ interface and we thought that modifying directly the αβ interfacial contact residues could further increase stability. In order to do so, a series of pairs of residues was chosen to introduce artificial disulfide bridges by mutation to cysteines (Fig. 1). Five residues were chosen for mutagenesis (2 on the α-, 3 on the β-chain) based on the orientation of their side chains and the distance between pairs of facing residues. All 3 mutants, with one artificial interfacial disulfide bridge each, were expressed with the OVA323-339 and HA126-138 peptides (Fig. 3A). The disulfide-linked dimers migrated on SDS-PAGE gel as a single 75 kDa band that could be reduced in two 40 and 35 kDa bands corresponding to the α- and β-chains, respectively (Fig. 3A). Yields for the disulfide complexes were significantly higher (2 to 3 folds) than for the original molecules indicating that a higher stability translated once again into higher expression and secretion. All three of the disulfide complexes bearing the OVA peptide could activate DO11.10 hybridoma T cells (Fig. 3B). However, the α96β149 I-Ad-OVA molecule did not stimulate DO11.10 cells as well as the other disulfide complexes. When tested for the staining of transgenic DO11.10 T cells, none of the disulfide-stabilized constructs produced staining different from the pMHCT-2 HA negative control (Fig. 3C). In conclusion, further stabilization of the MHC class II molecule increases recombinant production but does not result in staining pMHCT-2 reagents.
Tetramer staining has been linked to T cell activation in both MHC class I and class II antigenic systems (23, 24). To examine the possibility that our pMHCT-2 could stain only activated T cells, DO11.10 transgenic splenocytes were purified and activated in vitro by addition of OVA323-339 peptide. The tetramer staining was assessed at different times after peptide activation, together with TCR expression. Three days after activation, the staining with the pMHCT-2 OVA was higher than the staining with pMHCT-2 HA control (Fig. 4A). This weak specific staining disappeared quickly within 24h. Thus, it appeared that activation could be linked to late events that favored pMHCT-2 binding but staining remained very weak compared to TCR expression, and did not appear to be suitable to studies of small CD4+ T cell populations as we sought. In addition, in vivo activation by injection of the OVA323-339 peptide in DO11.10 transgenic mice was not followed by pMHCT-2 specific staining in the draining lymph nodes (Fig. 4B).
Very recently, the usage of similar reagents as ours was reported for the OVA antigenic system in H-2b animals after local immunization and tetramer enrichment using magnetic beads (25). To confirm these results in the H-2d background, BALB/c mice were immunized with either OVA323-339 or HA126-138 peptide, and magnetic enrichment was performed on draining lymph nodes with I-Ad pMHCT-2 OVA or HA reagents for FACS analysis. As reported in H-2b mice, each tetramer could stain specifically a small population of CD4+ T cells in each immunization (Fig. 4C). The percentage of detected cells could be titrated with a range of pMHC-T concentrations (not shown). Calculated back to the number of total CD4+ T cells (pre-sort), 0.07% OVA-specific and 0.11% HA specific CD4+ T cells could be identified in this particular experiment. Since the efficiency of one of the rare working pMHCT-2 reagent has been shown to depend on active cellular processes such as clustering and/or endocytosis (24) and because the usage of antibody-coated beads could induce the aggregation of multiple tetramer/TCRs conjugates and consequent internalization, we analyzed the cellular pattern of tetramer staining after sorting by confocal microscopy. Compared to the specific membrane staining with wheat germ agglutinin (WGA) and DAPI nuclear staining, it appeared that the pMHCT-2 were not internalized but clustered at the cell membrane (Fig. 4D). The beads used for enrichment could still simply increase the avidity and avoid the dissociation of pMHCT-2 reagents, mimicking the effect of liposomes. However, this possibility was not confirmed using DO11.10 transgenic splenocytes neither by incubating the cells with the beads after staining with I-Ad-OVA323-339 tetramers nor by performing a specific sorting (not shown and Fig. 4B).
Therefore, we had in our hands pMHCT-2 reagents that could detect a very small number of antigen-specific T cells in a polyclonal population after immunization but that could not stain an homogenous monoclonal population of naïve or activated T cells from transgenic mice. This paradox led us to believe that what we observed ex vivo after immunization was only a subset of the peptide reactive cells, and that there was a fundamental flaw in our pMHCT-2 design.
In a crystallographic analysis of two distinct I-Ad-peptide complexes we have previously highlighted the unique anchorless peptide-binding mode of I-Ad molecules and suggested the possible binding of I-Ad-bound peptides in alternative registers (26). This occurrence was confirmed for the OVA323-339 peptide, which was shown to bind I-Ad in at least 3 functional registers (27, 28). However, the possible impact of these observations on the design of pMHCT-2 was never evaluated. Indeed, the first correlate to multiple registers is that at any given time, a population of I-Ad molecules binding to a peptide with multiple registers is a mixture of molecules bound to individual registers at ratios obeying the laws of thermodynamics. In other words, unless one of the registers had a much higher affinity for I-Ad than the others, the population is a mixture. It also means that when tetramers are made with this population of I-Ad molecules, the probability of having the same peptide displayed by all 4 individual molecules of a single tetramer is very low. To address directly this concept in the context of the ovalbumin system, the OVA323-339 peptide was truncated into 9mers to favor the display of 4 individual registers (Fig. 5A). All four proteins were efficiently expressed at very similar level indicating that all four peptides were binding to I-Ad with very comparable affinities. Indeed, we have learned from expressing more than 60 different I-A-peptide complexes that expression paralleled peptide affinity for any given I-A haplotype (LT unpublished observation). In plate-bound activation assay, DO11.10 hybridoma T cells were exquisitely sensitive to the display of register 2 whereas they were not responsive to registers 1 and 3 and showed a 2-3 log lower recognition of the full length peptide and register 4 peptide (Fig. 5B). This result confirmed that register 2, previously described as the core epitope (OVA329-337) for DO11.10 T cells (28), was indeed the peptide specifically recognized by this particular TCR. The weaker stimulation seen for the full-length and register 4 peptide is most likely due to plastic immobilization and assay conditions, two very important issues that we will develop in the discussion. When the same reagents were tested for their capacity at staining naive DO11.10 transgenic T cells only pMHCT-2 presenting register 2 worked (Fig. 5C). Tetramer staining with register 2 matched percentage and intensity of staining obtained with the KJ1-26 anti-idiotype antibody (Fig. 5C). Since the OVA323-339 peptide also binds the closely related allele I-Ab, and OT-2 T cells recognize the same epitope as their I-Ad-restricted counterparts DO11.10 (28), we thus generated the four I-Ab-OVAregister pMHC complexes. Interestingly, OT-2 cells could only be stained with I-Ab-OVAreg2 pMHCT-2 (Fig. S1). The specific binding of DO11.10 TCR (and OT-2 cells) to complexes displaying register 2 could be titrated and was also confirmed by surface plasmon resonance measurements with recombinant molecules (Fig. 5D). The affinity KD= 4.95+/-2.0 μM, calculated from the kinetic rates constants from 3 independent experiments (Kass= 5.90+/-1.5×103 M.s-1 and Kdiss= 0.028 s-1), was comparable to other pMHC CLASS II/TCR pairs (29, 30).
Interestingly, when we compared our version of the I-Ab-OVA323-339 pMHCT-2 with the one produced at the NIH tetramer core facility on naïve OT-2 transgenic T cells (using I-Ab-CLIP pMHCT-2) a sharp discrepancy emerged: whereas our home product could stain OT-2 T cells at a level comparable to I-Ab-OVAreg2 pMHCT-2, the core reagent gave only a weak staining of less than 50% of the cells when used at the highest concentration (Fig. 6A and Fig. S1). Since the only difference between the 2 reagents was the length (7 and 18 for “in house” and NIH, respectively) and nature (S-GS3 and GS6+thrombin cleavage site for “in house” and NIH, respectively) of the linker, it was easy to draw 2 conclusions: 1) the length of the linker was very important for proper display, 2) with a short linker OVA323-339 was preferentially displayed in register 2 by I-Ab molecules. To expand these results, we tested within I-Ad-OVAregister 2 complexes Gly-Ser tethers of 6, 7and 8 residues in length (Fig. 6B, C). Six-residues linker constructs were expressed but failed to stain or activate DO11.10 T cells. Staining and activation were optimal and identical for 7- and 8-residue linkers (Fig 6C).
In addition to adjusting linker length and nature, to obtain the correct display of short register peptides, we also tried to anchor the side chain of the first residue in the P1 pocket since this pocket is the largest one in I-Ad and the most amenable to manipulations. Based on the observation that P1 can accommodate residues with large side chains (26), we chose Met and Glu (31) as anchoring P1 residues. However, as shown in figure 6D, both Met and Glu had negative impact on register 2 display suggesting that subtle changes as induced by anchoring peptides were poorly tolerated by T cell recognition.
To test the possibility that long peptides generated in vivo were indeed displayed in multiple registers and capable of selecting multiple T cell populations corresponding to each register BALB/c mice were immunized subcutaneously with ovalbumin protein and adjuvant. At day 8, draining lymph nodes were examined for the presence of register-specific T cells by FACS after magnetic enrichment. As seen in figure 7, cells specific for each of the 4 OVA323-339 registers were detected. With respect to quantity, register 2-specific cells were the more numerous, followed by register 4, 1 and 3-specific cells for a total of 2342 cells whereas the long peptide tetramer sorted 342 cells only (14.6%). To identify which register-specific population that long peptide population corresponded to, a series of register-specific depletions was performed followed by a new staining with the long peptide tetramer. As suspected based on numbers (342 versus 368), the long peptide population was identical to the register 1-specific population (Fig. 7C, D).
In conclusion, these results indicated that indeed long peptides such as the OVA323-339 peptide, are displayed in vivo in multiple registers, each selecting a distinct OVA-specific T cell population.
The development of the pMHC tetramer technology has been the most important advance in the characterization of antigen-specific T cells in recent years. The discrepancy between pMHCT-1 and pMHCT-2 performances has remained as one of the most puzzling and annoying issue with tetramers. Indeed, as much as CD8 T cell responses have been studied in details, CD4 T cells enumeration and dynamics are still poorly characterized. The intrinsic instability of MHC class II molecules (8), the low frequency of CD4+ T cells (10), and the low affinity of MHC class II binding peptides (9) were thought for a long time to explain the poor performance of MHC class II tetramers. However, after each of these issues had been addressed successfully, pMHCT-2s were still performing poorly. The ovalbumin system, one of the most studied antigens in immunology, and that we have used in this study, has lacked pMHCT-2 that could stain naïve transgenic T cells in either I-Ab or I-Ad genetic background. The current study was aimed at evaluating some of the most critical features of class II tetramers. It revealed that the geometry of display of pMHC on the avidin platform was not a limiting factor and that the further stabilization of the MHC dimer with artificial disulfide bridges could only increase protein production yields without altering the staining behavior of the reagent. These failures brought us to re-examine one fundamental difference between MHC class I and class II molecules and its consequences for the design of tetramers: peptide display. Indeed, we have learnt from early biochemical studies and later from structures that the 2 classes of MHC molecules differed greatly with respect to peptide binding. Foremost, a majority of MHC class I bound peptides are short (8-10mers) to fit almost perfectly the length of a groove that is closed at both ends, and bind through the side chain of anchor residues into MHC pockets. This “anchored” mode of binding results in peptides with high affinity for MHC and pMHC with long half-lives. To the opposite, MHC class II bound peptides are long (12-25 mers) and protrude out of both open ends of the groove. Moreover, whereas HLA-DR/I-E molecules use mostly an anchored mode of binding to associate with peptide, HLA-DQ/I-A dimers use preferentially an anchorless mode of display. This alternate mode of association results in more promiscuous molecules (26, 32, 33) and a fairly low affinity of peptide binding (34). The potential evolutionary advantage of such a mode of peptide binding has remained unclear. Theoretically one could argue that higher promiscuity of peptide binding could translate into higher diversity of CD4 T cell response but no experimental data has supported this idea, yet.
As a first step in this direction we have now shown that long peptides such as produced in MHC class II loading compartments, could be displayed in multiple registers and select T cells specific for each of these registers. This observation becomes critical for the design of proper pMHC tetramers and more importantly for the way we should dissect HLA-DQ/I-A-restricted T cell responses. It is clear now that I-A pMHCT-2s built with long peptides (tethered or non-tethered) are mere MHC tetramers but mixes of pMHC complexes. The mix will obey the simple laws of thermodynamics with more representation of the peptide register binding with the highest affinity, e.g. register 1 and register 2 for ovalbumin in I-Ad and I-Ab, respectively. However, our immunization studies show that the immune response does not simply follow the hierarchy of peptide binding. For instance, in I-Ad OVA register 1 response is subdominant whilst register 2 and 4 responses are dominant even though register 1 has the highest affinity. The higher affinity of register 1 is not only based on direct binding studies (35) but more convincingly on the fact that when the long ovalbumin peptide is tethered to I-Ad, this particular register is the one that is displayed preferentially as indicated by structural studies (26) and the results obtained with the long peptide tetramer for staining. This observation will be critical for the design of subunit vaccines which, so far, has been guided by the identification of “motif” of binding in polypeptides and peptide binding studies. The current study shows that the choice of peptides should rely on much subtler parameters. In addition to the register issue, our failure to produce functional P1 mutants of register 2 demonstrates that trifling alterations of peptide display such as minimal torsion or height differences applied by a single mutation can modify T cell recognition altogether. This idea is supported by Unanue and colleagues studies showing that two distinct conformers of the same HEL48-61 peptide bound to I-Ak could lead to differential recognition by specific CD4+ T cell clones (36, 37). This result is not that surprising when we consider that T cell agonism, super-agonism and antagonism are supported by very minor structural differences (38, 39) but it indicates very stringent limitations to our design of register-specific pMHCT-2. Indeed, we are pretty much left with the single option of limiting the length of linker to produce our reagents. The usage of non-tethered peptides does not appear as a viable option as the affinity of 9mer peptides is very low as exemplified by the 4 registers of OVA323-339 for which we could not measure an affinity by using classical peptide binding methods (EL, LT, unpublished results). In any event, as we already discussed, peptide binding studies appear of limited interest to design appropriate pMHCT-2.
On the technical side, the current study also indicates the limitation of the T cell stimulation assay on plastic-immobilized pMHC molecules. Indeed, we thought for years that the pMHCT-2 made with the OVA323-339 peptide was functional because capable of stimulating DO11.10 cells in vitro. However, as shown in Figure 5 the proper register pMHC was 100 fold more efficient in the same assay. Since, register 4 also supported some activation of DO11.10 in the same assay, it is likely that plastic immobilization resulted in peptide exchange between neighbor molecules and some degree of register shift sufficient to activate T cells. Therefore, activation assay on immobilized pMHC is not sufficient to determine the characteristics of pMHCT-2.
On the more important functional side, it is quite stunning to see that the usage of the original OVA323-339 pMHCT-2 would limit our detection of the anti-OVA323-339 response to register 1 and only 15% of the overall response. This result will incite us to re-evaluate the real diversity and amplitude of CD4+ T cell responses in systems where register shifts are likely such as many I-A-restricted responses
In conclusion, our findings by demonstrating that long peptides are displayed in vivo in multiple productive registers should have important implications for studying CD4+ T cell responses and designing appropriate subunit vaccines.
We would like to acknowledge contributions in the form of technology development and support from the staff of the N.I.H. Tetramer Facility. We would also like to thanks Randy Stefanko for his technical assistance and Kenji Yoshida for technical help and valuable discussion.
This work was supported by National Institute of Health Grants RO1-DK055037 and U19-AI050864 (to L.T), RO1-AI48540 (to K.C.G), and RO1-CA58896 (to I.A.W) and by the Howard Hughes Medical Institute (to K.C.G.). JS and JDA received support from the NIAID contract for the NIH Tetramer Facility (N01-AI-25456).
Publisher's Disclaimer: This is an author-produced version of a manuscript accepted for publication in The Journal of Immunology (The JI). The American Association of Immunologists, Inc. (AAI), publisher of The JI, holds the copyright to this manuscript. This version of the manuscript has not yet been copyedited or subjected to editorial proofreading by The JI; hence, it may differ from the final version published in The JI (online and in print). AAI (The JI) is not liable for errors or omissions in this author-produced version of the manuscript or in any version derived from it by the U.S. National Institutes of Health or any other third party. The final, citable version of record can be found at www.jimmunol.org