A Public T cell Receptor Recognized by a Monoclonal Antibody Specific for the D-J Junction of the β-chain

Authors

  • T. Frigstad,

    1. Centre for Immune Regulation and Institute of Immunology, University of Oslo and Oslo University Hospital, Oslo, Norway
    Search for more papers by this author
  • G. Å. Løset,

    1. Centre for Immune Regulation and Department of Biosciences, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • I. Sandlie,

    1. Centre for Immune Regulation and Department of Biosciences, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • B. Bogen

    Corresponding author
    1. K.G. Jebsen Centre for Research on Influenza Vaccines, University of Oslo and Oslo University Hospital, Oslo, Norway
    • Centre for Immune Regulation and Institute of Immunology, University of Oslo and Oslo University Hospital, Oslo, Norway
    Search for more papers by this author

Correspondence to: B. Bogen, Institute of Immunology, Rikshospitalet, 0027 Oslo, Norway, E-mail: bjarne.bogen@medisin.uio.no

Abstract

It is becoming increasingly clear that T cell responses against many antigens are dominated by public α/β T cell receptors (TCRs) with restricted heterogeneity. Because expression of public TCRs may be related to resistance, or predisposition to diseases, it is relevant to measure their frequencies. Although staining with tetrameric peptide/major histocompatibility complex (pMHC) molecules gives information about specificity, it does not give information about the TCR composition of the individual T cells that stain. Moreover, next-generation sequencing of TCR does not yield information on pairing of α- and β-chains in single T cells. In an effort to overcome these limitations, we have here investigated the possibility of raising a monoclonal antibody (moAb) that recognizes a public TCR. As a model system, we have used T cells responding to the 91–101 CDR3 peptide of an Ig L-chain (λ2315), presented by the MHC class II molecule I-Ed. The CD4+ T cell responses against this pMHC are dominated by a receptor composed of Vα3Jα1;Vβ6DβJβ1.1. Even the V(D)J junctions are to a large extent shared between T cell clones derived from different BALB/c mice. We here describe a murine moAb (AB10) of B10.D2 origin that recognizes this public TCR, while binding to peripheral T cells is negligible. Binding of the moAb is abrogated by introduction of two Gly residues in the D-J junction of the CDR3 of the β-chain. A model for the public TCR determinant is presented.

Introduction

The αβ TCR recognizes short antigenic peptides presented in the groove of MHC class I or class II molecules. The TCR is assembled by clonally unique recombinations of V(D)J gene segments in thymocytes. Stochastically generated combinations, with the addition of junctional diversity, generate a theoretical αβ TCR repertoire of about 1015 in the mouse, of which only a fraction is tapped at a given time point because a mouse only contains ~1–2 × 108 T cells [1]. To date, the structures of 84 TCR–pMHC trimeric complexes have been solved (http://www.imgt.org/3Dstructure-DB/). For some antigens, the TCR repertoire is partially (semi-public) or highly (public) conserved in the sense that certain TCR compositions dominate the response, even in different individuals [2-10]. The underlying mechanisms for these public TCR responses have been elusive, but may possibly be explained by convergent recombination in the development of the näive repertoires [11].

A striking and early example of a public TCR is the BALB/c CD4+ T cell response to an idiotypic (Id) CDR3 sequence of syngeneic Ig L-chain. T cell recognition of this Id-MHCII on B cells has been shown to have consequences for T-B collaboration, autoimmunity and development of B lymphomas [12]. More specifically, T cells in this model recognize residues 91–101 of the λ2315 L-chain (ALWFRNHFVFG) presented by the I-Ed MHC class II molecule [13, 14]. αβ TCR sequences were established from seven independent clones [3]. Independency was assured by isolation of T cell clones from different mice or by differences in nucleotide sequences. The TCRs used by these clones were conserved in that six of seven clones used Vβ6 and five of seven clones used Vα3. Moreover, of the five Vα3;Vβ6 clones, four used the same (D)J segments for both α- and β-chains (Jα1;DβJβ1.1). Of these four highly conserved clones, two clones had identical aa in the junctional regions even though they differed at the nucleotide level. A third clone had a GlyGly doublet at the D-J junction of the β-chain (one substitution and one insertion), while a fourth clone had an I→M substitution at the V-J α-junction. These four clones, having an extremely conserved TCR (Vα3Jα1;Vβ6DβJβ1.1), were marked by a peculiar specificity in that they cross-reacted between peptides having either Phe94 (λ2315 sequence) or Tyr94 (germline λ2) in the P3 position. This difference (Phe versus Tyr) arises as a consequence of a somatic mutation in codon 94 of the V-gene segment of the MOPC315 myeloma.

Polymeric pMHC stains antigen-specific T cells regardless of TCR composition. By contrast, moAb specific for TCR rather recognizes antigenic determinants on the TCR regardless of TCR specificity. Anti-TCR moAb may be specific for Vβ [15] or Vα [16] independently of which other V segment the αβ TCR is composed of. In addition, clonotype-specific anti-TCR moAb has been described that binds the relevant T cell clone but only negligibly to polyclonal T cells [17-20]. Clonotype-specific moAbs are highly likely to bind to the unique CDR3-related sequences, and indeed, there is strong evidence for this with the moAb 1B2 specific for the murine 2C TCR [21].

We have here generated a TCR-specific mouse moAb with pMHC-like properties in that it recognizes the CDR3 of the β-chain of the public TCR. Thus, this moAb is a unique supplement to multimeric pMHC reagents in that it is a high-affinity probe that can be used to track a dominant fraction of T cells in an immune response.

Materials and methods

Mice

B10.D2/n were from OLAC (Bicester, UK), while BALB/c were from Taconic (Ry, Denmark).

Cell lines

Independent BALB/c T cell clones 4B2A1, 4B5B1, 7A10B2, 8A5A5, 8A1A8, 11A11A3 and 11A12A5 specific for aa 91–101 of the λ2315 Ig L-chain of the BALB/c myeloma protein M315, presented on the MHC class II molecules I-Ed, have been described [13, 14, 22], including their TCR sequences [3].

Immunizations

Twelve B10.D2/n mice were immunized with 107 cloned 7A10B2 T cells in PBS, emulsified 1:1 in 100 μl complete Freund's adjuvant i.p. Mice were boosted with the same number of 7A10B2 cells in PBS i.p on days 20 and 40. Dilutions of day 50 sera were tested in flow cytometry against 7A10B2 and 4B2A1 cells. Five mice had antibodies that bound stronger to 7A10B2 than 4B2A1. A single mouse with the most serum Abs was boosted with 3 × 107 7A10B2 cells in PBS i.p.

Fusion to obtain B cell hybridomas, screening, cloning, purification and conjugations

Three days after the last boost, spleen cells were fused to Sp2/0 following a previously described protocol [20]. 14/1248 wells had Ab that bound 7A10B2 in flow cytometry; in two of these cases, no binding to 4B2A1 and 4B5B1 T cell clones was observed. One clone (AB10) was cloned twice by limiting dilution and was typed by flow cytometry to be an IgG2a/κ. AB10 was purified from culture SN on protein G and conjugated to biotin and FITC by standard procedures.

Flow cytometry

Cells were washed once with PBS and blocked with 60% heat-inactivated rat serum and 200 μg/ml rat anti-murine CD32 moAb (clone 2.4G2, ATCC® number HB-197, Manassas, VA, USA) in PBS. Cells were then stained with AB10 or GB113 or mouse IgG2a/κ isotype control (anti-SV40 moAb, Pab 108, BD Pharmingen, San Jose, CA, USA) or rat anti-Vβ6 moAb (RR4-7, BD Pharmingen) (all moAbs biotinylated, 5 μg/ml) before incubation with 2 μg/ml streptavidin–PE conjugate (BD Pharmingen). In some experiments (Fig. 4), cells were stained with 10 μg/ml anti-Vβ6 moAb RR4-7-FITC or FITC-conjugated rat IgG2b/κ isotype control (both BD Pharmingen), rather than biotinylated versions. Stained cells were fixed in 2% paraformaldehyde, and events were collected on a FACSCalibur flow cytometer (BD Biosciences). Data were analysed using CellQuest software (BD Biosciences, San Jose, CA, USA).

moAb-induced T cell proliferation

Wells (96-well plates, flat bottom) were coated with triplicates (200 μl) of AB10 or GB113 [20] overnight at 4 °C. Concentrations of AB10 and GB113 were 10 μg/ml, 2 μg/ml and 0,4 μg/ml. Wells were washed with RPMI1640 (10% FCS), and 7A10B2 or 4B2A1 T cells were added (2 × 104/well, 200 μl). After 24 h, wells were pulsed with 2 μCi [3H]dThd for 24 h, harvested and counted.

Surface biotinylation, immunoprecipitation and Western blotting

Surface biotinylation, immunoprecipitation and Western blotting were performed as described [23]. Monoclonal antibodies used for immunoprecipitation were AB10 or anti-Vβ6 moAb (44.22.1). Reduced [β-mercaptoethanol (Sigma)] and non-reduced samples were analysed by Western blotting, and precipitated proteins were detected by streptavidin–HRP (GE HealthCare, Oslo, Norway).

Homology modelling and molecular visualization

A homology model of the 7A10B2 TCR was made using Modeller v9.8 (http://salilab.org/modeller/) employing the 2C (PDB: 1tcr) and JM22 (PDB: 2vlm) as crystallographic TCR template structures (Figure S1). The 8A5A5 TCR model was made by in silico mutating the 7A10B2 CDR3α residue Ile94 to Met using the Swiss-PdbViewer v4.01 [24]. The 11A5A1 TCR model was made by CDR3β loop modelling of the 7A10B2 structure using ModLoop [25]. Molecular surfaces with electrostatic potential contouring were visualized using GRASP2 [26], whereas the Swiss-PdbViewer and Pov-Ray v3.62 (http://www.povray.org/) were used for loop visualization and image rendering, respectively.

Results

Immunization and screening strategy

The aim was to obtain a moAb against a public TCR (Vα3Jα1;Vβ6 DβJβ1.1) that is repetitiously used by BALB/c CD4+ T cells in response to a particular pMHC, aa 91–101 of the Ig L-chain λ2315 presented on the I-Ed molecule [3]. The choice of mouse strain for immunization was B10.D2 because this strain has the same H-2d haplotype as BALB/c but a different TCRα-chain haplotype (TCRαb versus TCRαa). This strategy should minimize induction of anti-H-2 antibodies, but increase the likelihood of obtaining TCR-specific moAbs. B10.D2 mice were repetitiously immunized with cloned 7A10B2 T cells that express the public TCR, first in CFA i.p. and then in saline. A mouse that had comparatively higher amounts of serum Ab directed against immunizing 7A10B2, compared with another λ2315-specific BALB/c T cell clone (4B2A1) with a non-recurrent TCR (Vα1Jα19, Vβ8.2, DβJβ1.2), was used as donor of spleen cells for fusion. A B cell hybridoma producing AB10 moAb (IgG2a,κ) was selected for further studies.

AB10 moAb binds TCR of 7A10B2 T cells

The AB10 moAb stained 7A10B2 T cells in flow cytometry but not 4B2A1 T cells (Fig. 1). Conversely, the previously published GB113 moAb clonotype specific for the non-recurrent λ2315-specific 4B2A1 TCR [20] bound 4B2A1 but not 7A10B2 (Fig. 1). Neither AB10 nor GB113 [20] bound BALB/c peripheral lymph node T cells to a significant extent above the isotype-matched control moAb (Fig. 1). This holds true for combined CD4+ and CD8+ populations analysed together (Fig. 1) as well as for each population analysed separately (not shown). As a positive control, a rat anti-Vβ6 moAb bound 5% of CD4+ and CD8+ cells (not shown).

Figure 1.

AB10 specifically binds 7A10B2 T cells. The indicated T cell clones were stained with biotinylated AB10 and previously described GB113 [clonotype specific for 4B2A1 (20) anti-TCR mAbs (both mouse IgG2a) followed by streptavidin–PE]. The rightmost panel shows staining of combined CD4+ and CD8+ BALB/c lymph node T cells with AB10 and GB113 mAbs compared with an isotype control mAb. A positive control, rat anti-Vβ6 mAb, bound about 5% of lymph node T cells (not shown).

The AB10 moAb precipitated a surface molecule from 7A10B2 T cells that had a MW of ≈ 60 kD under non-reducing conditions. Upon reduction, two bands at ≈ 33 kD were seen (Fig. 2, lane 3). These findings are consistent with an αβ TCR heterodimer [27]. Confirming the TCR identity of the precipitated molecule, an anti-Vβ6 moAb precipitated a molecule with identical appearance (Fig. 2) (7A10B2 is Vβ6+). These observations strongly indicate that the AB10 moAb binds the αβ TCR of 7A10B2. In addition to the αβ TCR bands, a weak band of unknown identity of ≈ 45 kD was seen after reduction with β-mercaptoethanol (Fig. 2, lanes 3 and 4).

Figure 2.

AB10 precipitates a TCR molecule. 7A10B2 T cells were surface biotinylated, and lysates were incubated with either AB10 mAb (lanes 1 and 3) or 44.22.1 mAb specific for Vβ6. Immunoprecipitates were run under reducing or non-reducing conditions on SDS-PAGE and transferred to membranes that were probed with streptavidin–HRP.

AB10 moAb activates 7A10B2 T cells

AB10 moAb immobilized on solid phase was able to induce proliferation of 7A10B2 T cells but not 4B2A1 cells. Conversely, the GB113 clonotype-specific moAb stimulated 4B2A1 but not 7A10B2 cells (Fig. 3). It may be concluded that immobilized AB10 stimulates 7A10B2 T cells independently of costimulatory signals delivered through CD28.

Figure 3.

Solid phase-bound AB10 induces proliferation of 7A10B2 T cells but not 4B2A1 T cells. Microtitre wells were coated with the indicated concentrations of AB10 or GB113 anti-TCR mAbs. 7A10B2 T cells or 4B2A1 T cells were added in triplicates, and proliferation was measured as incorporated 3H thymidine.

AB10 moAb recognizes a public TCR used in the response against the λ2315 idiotype

The seven independent clones of Fig. 4 were stained with the AB10 moAb. The following observations were made: (1) the moAb failed to stain TCRs of 4B2A1 and 8A1A8, which both are distinct from that of 7A10B2; (2) the moAb was not specific for either Vα3 or Vβ6 alone or a combination of Vα3Vβ6 independent of CDR3s, because it failed to stain 4B5B1 (which has Vα3Vβ6 in common with 7A10B2, but different Jα and DβJβ); (3) the moAb stained 11A12A5, as would be expected because that T cell clone has an identical αβ TCR to 7A10B2 at the amino acid level (albeit originating from a different mouse and having silent nucleotide substitutions in the junctional regions [3]); (4) the moAb stained 8A5A5, which differs from 7A10B2 by an I→M in the α-chain V-J junction and (5) the moAb failed to stain 11A5A1, the only difference from 7A10B2 being a T→GG (substitution + insertion) in the D-J junction of the β-chain. In conclusion, these data demonstrate that the AB10 moAb recognizes CDR3s of a public TCR. Moreover, recognition is abrogated by a Thr→GlyGly in the CDR3β.

Figure 4.

AB10 detects an epitope present on public TCRs expressed by independent T cell clones. T cells were stained with biotinylated AB10 followed by streptavidin–PE or RR4-7-FITC (specific for Vβ6). Data for seven independent clones are shown. Amino acid sequences and V(D)J gene segment usage are indicated. GG indicates a mutation and an insertion in Dβ-Jβ in the 11A5A1 clone that abolishes AB10 staining. M indicates a mutation in the N-region of the α-chain of 8A5A5, which has no influence on AB10 binding (see Snodgrass et al. [3] for amino acid and nucleotide sequences).

Modelling of the antigenic determinant recognized by AB10

The binding profile of AB10 strongly suggests that key epitope elements are situated in the somatically recombined CDR3 loop of the β-chain (Fig. 4). However, it is not possible to determine exactly where AB10 binds to the public TCR without crystallographic data. In lack of such information, we therefore generated structural models of the three TCRs of 7A10B2, 8A5A5 and 11A5A1 (Figs. 5 and S1).

Figure 5.

Structural models of the 8A5A1, 7A10B2 and 11A5A1 TCRs. (A) The models are depicted in top-down view onto the binding site of the TCRs. Models were generated based on homologous 2C and JM22 TCR structures, as detailed in supplementary Fig. 2. GRASP2 [26] generated the electrostatic potential mapped onto the molecular surface (red, negative; blue, positive). (B) The CDR3β loops of 7A10B2 (left) and 11A5A1 (right) are visualized in ball and sticks (C, grey; N, blue; O, red), with intraloop hydrogen bonds and their corresponding distances in green. (C) Superimposed overlay of CDR3β loops of 7A10B2 and 11A5A1 shown in ribbon, together with the invariant CDR3α.

As could be expected, the overall structures of the three TCRs are very similar both in shape and electrostatics (Fig. 5A). However, a loop rearrangement of CDR3β distinguishes 7A10B2 and 11A5A1 (Fig. 5B and C), which translates into a loop protrusion in 11A5A1, and local change in the surface electrostatics (Fig. 5A). In more detail, both Thr97 and Asn98 in 7A10B2 are solvent exposed at the distal loop tip (Fig. 5B); the residues are present in all AB10-reactive T cell clones, indicating crucial hydrogen bonding interactions with AB10. In the non-reactive 11A5A1 TCR, the combined effect of Thr97→Gly97 and the insertion of Gly98 leads to a loss of the Thr97 hydroxyl to Asn98 amide hydrogen bond, as well as a displacement of Asn98 away from the solvent-exposed surface, thus redirecting its hydrogen bond to the Thr100 backbone nitrogen. The result is a protruding loop in 11A5A1, with changed electrostatics; hence, AB10 binding is abrogated.

The difference between Ile94 (7A10B2) and Met94 (8A5A5) in CDR3α is not sensed by AB10 (Fig. 4), which appears reasonable in the light of the very low structural impact of this aa (Figs. 5A and S2). Thus, both Ile94 and Met94 are positioned away from the top of the surface and are partially hidden by the neighbouring Arg95 (Figure S2). Nevertheless, the CDR3α Ile94 to Met94 substitution confers a heteroclitic phenotype to the 8A5A5 T cell clone [14] manifested by a reduced responsiveness to Phe94 (mutated residue of the λ2315immunogen) but maintained cross-reactivity to Tyr94 (germline-encoded residue of λ2) [14]. It is tempting to speculate that the Ile94→Met94 substitution somehow negatively affects recognition of Phe94 (perhaps by reducing hydrophobic interactions). Collectively, the data therefore support the hypothesis that the peptide fine specificity of the λ2315-specific T cell clones primarily is restricted by CDR3α, whereas the AB10 epitope primarily locates to CDR3β.

Discussion

Recent empirical evidence [11] suggests that the TCR repertoire may be more restricted than what would be theoretically expected [1]. Moreover, antigen-specific T cell responses in inbred mice, as well as in humans, are often dominated by conserved and public TCRs [28]. Expression of public TCRs could be associated with either resistance or predisposition to infections and autoimmune diseases [29]. Furthermore, examination of antigen-specific TCR expression in EBV-infected patients over a period of several years has shown that the dominant pattern changes during a prolonged immune response [4]. It should therefore be of interest not only to detect pMHC specificity by use of tetramers and pentamers, but also to be able to determine TCR V-gene composition and expression levels of public TCRs in response to antigen. Herein, we describe a moAb, AB10, which by virtue of detecting a public TCR expressed by Id(λ2315)-specific CD4+ T cells [3], combines features of clonotype-specific moAbs and pMHCs. To our knowledge, this is the first moAb specific for a public TCR.

The anti-TCR moAb AB10 shares features with clonotype-specific moAbs [17-20] in that it binds the T cell clone used for immunization, but not polyclonal T cells in lymphoid organs of normal mice. However, in distinction from clonotype-specific moAbs, AB10 bound not only the immunizing clone but three of four independently derived T cell clones that express an extremely conserved public TCR (Vα3Jα1;Vβ6DβJβ1.1) repetitively used by different mice in response to immunization with the λ2315 L-chain. We were able to demonstrate that the binding specificity of AB10 was influenced by a change in sequence and structure of the CDR3 β-chain loop. More specifically, as defined by modelling, an epitope-destroying Gly doublet (one substitution and one insertion) generated an altered CDR3β loop topology and changes in surface electrostatics. The most straightforward interpretation is that the public TCR-specific AB10 moAb binds a conserved motif characteristic of the clonotypic CDR3β and that the structural changes introduced by the GlyGly doublet abrogate binding. However, it is not ruled out that the moAb binds another determinant that is conformationally changed upon the GlyGly introduction in the CDR3β loop. Structure determination is required to definitely resolve this issue. It may be concluded that specificity of AB10 is influenced my microheterogeneity within a panel of public TCRs. Such exquisite specificity may limit the use of public TCR-specific moAbs because conserved TCR with very minor aa differences in CDR3 loops may escape detection.

The finding that a public TCR-specific moAb is influenced by changes in the CDR3β loop is consistent with current models for how TCRs recognize pMHC. Thus, TCR CDR3α and CDR3β loops straddle the antigenic peptide bound to MHC molecules [30], and the loops are thought to greatly influence specificity for peptide. Consistent with this, the hypervariability between TCRs is mainly localized to CDR3 loops generated by V(D)J recombination and N-region diversity – because TCR is not subject to somatic hypermutation [1].

Specificity of the clonotype-specific moAb 1B2 and its cognate ligand, the 2C TCR, has been reported [21]. An alanine scan of the 2C TCR suggested that the 1B2 epitope is localized to the CDR3β loop. Such a specificity for CDR3 loops is likely to be true for clonotype-specific moAbs in general. Thus, clonotype- and public TCR-specific moAbs may be rather similar, the difference being that the latter are specific for a TCR that is recurrently used in response to an antigen. In fact, as more and more T cell responses are found to be dominated by conserved TCR, many clonotype-specific moAbs may on closer scrutiny turn out to be public TCR specific. Nevertheless, truly clonotype-specific moAbs most likely exist. An example is the GB113 moAb used herein that uniquely binds a TCR that lacks any resemblance to the public TCR in our panel of Id(λ2315)-specific T cell clones [20]. Even though we favour the idea that public TCR-specific and clonotype-specific moAbs both bind CDR3, it is not excluded that such moAbs could bind non-CDR3 regions if the latter contain epitopes that are conformationally dependent on particular CDR3 sequences. It has recently been suggested in the cytochrome C model that CD4+ clonotypic diversity is skewed towards dominance by a few high-affinity TCRs when the antigenic peptide has low affinity for MHCII [31]. The same mechanism may apply to the λ2315 system because the P9 residue in the nonameric 92–100 peptide is naturally Phe100, while a basic residue such as Lys is preferred in P9 of I-Ed [32]. Thus, the 91-101 CDR3 λ2315 peptide is likely to bind relatively weakly to I-Ed. This may explain why Id(λ2315)-specific T cell clones obtained after immunization and prolonged tissue culture express a public TCR, detected by the current AB10 moAb.

Monoclonal antibodies specific for public TCRs could have a number of advantages compared with pMHC tetramers. Thus, the former, but not always the latter, can be used with ease on tissue sections. It is also considerably easier to use moAbs in flow cytometry, and the signal is often brighter. Moreover, moAbs specific for public TCRs may indicate both specificity and TCR composition of T cells, while pMHC tetramers only indicate specificity. Public TCR-specific moAbs also have an advantage compared with next-generation sequencing because the latter technique does not allow allocation (‘pairing’) of TCR α- and β-chain sequences to single T cells, while staining with public TCR-specific moAbs does. A drawback with public TCR-specific moAbs is the difficulty in obtaining such reagents.

Acknowledgment

The work was funded by grants from the Norwegian Research Council [175358/S10 (BB) and 174796/130 (GÅL)]. Peter O. Hofgaard did some of the flow cytometry stainings of Fig. 1. The Norwegian Research Council had no involvement in study design, collection, analysis or interpretation of data or in the writing of the report.