Design of Soluble Recombinant T Cell Receptors for Antigen Targeting and T Cell Inhibition

Functional optimisation of soluble TCRs

1

Design of soluble recombinant T cell receptors for

antigen targeting and T cell inhibition

Bruno Laugel*, Jonathan M. Boulter†, Nikolai Lissin‡, Annelise Vuidepot‡, Yi Li‡,

Emma Gostick*, Laura E. Crotty§, Daniel C. Douek§, Joris Hemelaar¶, David A.

Price§#, Bent K. Jakobsen‡ and Andrew K. Sewell*||**

*The T-cell Modulation Group, The Peter Medawar Building for Pathogen Research,

University of Oxford, South Parks Road, Oxford OX1 3SY, UK †Department of Medical Biochemistry and Immunology, University of Wales College of

Medicine, Heath Park, Cardiff CF14 4XN, UK ‡Avidex Ltd., 57 Milton Park, Abingdon, Oxon OX14 4RX, UK

§Human Immunology Section, Vaccine Research Center, National Institute of Allergy and

Infectious Diseases, 40 Convent Drive, National Institutes of Health, Bethesda, MD

20892, USA ¶Harvard Medical School, Department of Pathology, 200 Longwood Avenue, Boston, MA

02115, USA

||Correspondence to: Andrew Sewell

Tel: (+)44-1865-281539

Fax: (+)44-1865-281530

Email: [email protected]

Running title: Functional optimisation of soluble TCRs

JBC Papers in Press. Published on November 4, 2004 as Manuscript M409427200

Copyright 2004 by The American Society for Biochemistry and Molecular Biology, Inc.


2

Summary The use of recombinant T cell receptors (TCRs) to target therapeutic interventions has been hindered by the

naturally low affinity of TCR interactions with peptide-major histocompatibility complex (pMHC) ligands.

Here, we use multimeric forms of soluble heterodimeric αβ TCRs for specific detection of target cells

pulsed with cognate peptide, discrimination of quantitative changes in antigen display at the cell surface,

identification of virus-infected cells, inhibition of antigen-specific cytotoxic T lymphocyte (CTL) activation

and identification of cross-reactive peptides. Notably, the A6 TCR specific for the immunodominant HLA

A2-restricted HTLV-1 Tax11-19 epitope bound HLA A2-HuD87-95 (KD = 120 µM by surface plasmon

resonance), an epitope implicated as a causal antigen in the paraneoplastic neurological degenerative

disorder anti-Hu syndrome. A mutant A6 TCR that exhibited dramatically increased affinity for cognate

antigen (KD = 2.5 nM) without enhanced cross-reactivity was generated; this TCR demonstrated potent

biological activity even as a monomeric molecule. These data provide insights into TCR repertoire selection

and delineate a framework for the selective modification of TCRs in vitro that could enable specific

therapeutic intervention in vivo.


3

Introduction Peptide-major histocompatibility complex (pMHC) antigens displayed on the surface of target cells are

recognized by T cells via their specific T cell receptor (TCR) (1). The TCR coreceptors CD8 or CD4 bind to

invariant domains of pMHC class I (pMHCI) or pMHC class II (pMHCII) respectively, and are known to

facilitate the process of antigen recognition by T cells (2). Recent advances have enabled the generation of

high quality soluble TCR, pMHCI, pMHCII, CD8 and CD4 proteins; these, in turn, have allowed the

biophysical characterization of the interactions between these molecules. Accordingly, the TCR and CD4/8

co-receptors have been shown to have very low affinities (KD ~ 10-3-10-6 M) for cognate pMHC. Despite

these low affinity interactions, however, the process of antigen engagement can initiate T cell recognition of

antigen-presenting cells (APCs) bearing fewer than 10 copies of a specific pMHC complex (3,4). The

mechanisms by which these weak recognition events result in such exquisite sensitivity are not fully

understood.

The production of soluble recombinant αβ T cell receptors has proved challenging. The main technical

pitfall is heterodimeric instability in the absence of anchoring transmembrane domains and α/β pairing

through an interchain disulphide bond. One of the commonest protein engineering strategies used in TCR

studies to date has been the construction of single-chain TCRs (scTCRs). This technique, which takes

advantage of the structural similarities between antibodies and TCRs, is based upon the single-chain Fv

technology used to generate antibody fragments (5). In short, for the TCR, it involves the cloning and

expression of a unique chimerical open-reading-frame where the Vα and Vβ domains are paired with a

protein linker (6,7). These reagents have been successfully used in structural and biophysical studies (7,8)

but widespread application of the method has proven more difficult. Alternative approaches have included

shuffling the variable (V) and constant (C) domains of the TCR to the C-region of an immunoglobulin κ

light chain to generate a soluble heterodimeric protein (9); the resultant TCR was shown to react with

several anti-TCR antibodies, but was not shown to bind specific antigen.

In this study, we generate soluble versions of the human HLA A2-restricted JM22 and A6 TCRs specific for

dominant viral epitopes derived from the influenza matrix and human T cell leukaemia virus type 1 (HTLV-


4

1) Tax proteins respectively. Unlike most of the recombinant TCRs described to date, these proteins

comprise an α- and a β-chain that are expressed separately and paired by a non-native disulfide bond (10).

We overcome limitations imposed by the intrinsically low binding affinity and correspondingly short half-

life of the monomeric TCR/pMHCI interaction by building multimeric forms of these proteins. The exquisite

binding sensitivity and specificity exhibited by these multimeric TCRs allowed us to monitor quantitative

modifications of antigen display on APCs and, in the case of A6, to investigate the binding parameters of the

TCR with several syngeneic cross-reactive ligands. These latter data, which represent the first biophysical

demonstration of the self-reactivity of a human TCR, have important implications for our understanding of T

cell repertoire selection by thymic editing. Further, we characterize these soluble multimeric TCRs

functionally and show that such reagents can be used to inhibit antigen-specific CD8+ T cell activation.

Finally, we employ a novel method that allows the generation of TCRs with cognate ligand affinities in the

antibody range; these latter reagents hold the potential to revolutionize immunotherapeutics by enabling

targeted drug delivery and antigen-selective immunosuppression.


5

Experimental Procedures

B cell lines

LBL 721.174 cells, T1 cells and EBV-immortalized HLA A2+ B cells (PK) were maintained at 37°C in

RPMI medium supplemented with 10% fetal calf serum (FCS), L-glutamine (2 mM), penicillin (100 U/ml)

and streptomycin (100 µg/ml) (R10 medium).

T cell lines

The E6 CTL line was derived from peripheral blood mononuclear cells (PBMCs) obtained from an HTLV-1

infected donor. Cells specific for the HLA A2-restricted HTLV-1 Tax11-19 epitope were initially expanded by

exposure to peptide-pulsed autologous PBMC in R10, and then further stimulated with peptide-pulsed PK B

cells and mixed irradiated allogeneic feeder PBMC from 3 unrelated donors in Isocove’s modified

Dulbecco’s Minimum Essential Medium (D-MEM) (Sigma) supplemented with FCS and antibiotics as

above, together with PHA (4 µg/ml) and T-STIM (10%; Becton Dickinson). The specificity of the line was

confirmed by pMHCI tetramer staining. The 003 CTL clone specific for the HLA A2-restricted HIV-1

epitope SLYNTVATL (p17 Gag; residues 77-85) was isolated and maintained as previously described (11).

Manufacture of soluble heterodimeric TCRs

The generation of soluble TCR heterodimers was based on the procedure described by Boulter et al. (10).

Each TCR chain was individually cloned in the bacterial expression vector pGMT7 and expressed in E.coli

BL21-DE3(pLysS). Residues threonine 48 and serine 57 of, respectively, the α− and β-chain TCR constant

region domains were both mutated to cysteine. A biotinylation target motif (12) was also fused to the C-

terminus of the TCR beta chain to allow tetramerization with extravidin. Expression, refolding, purification

and biotinylation of soluble TCR heterodimers have been described previously (13). Development and

production of the high affinity A6c134 TCR are described elsewhere1.


6

Multimerization of T cell receptors

Tetramerization of TCR heterodimers was performed by the addition of extravidin or R-phycoerythrin-

labelled extravidin (Sigma) in aliquots, in order to saturate its binding sites, to a total TCR:extravidin molar

ratio of 4:1.

Flow cytometry

106 B cells per staining were pelleted and pulsed with the appropriate concentration of HLA A2-restricted

peptides diluted in RPMI for 90 min at 37 °C. Cells were then washed once in 5 ml FACS buffer (PBS with

2mM EDTA and 2% FCS) and resuspended in 100 µl TCR tetramer solution at 100 µg/ml (with respect to

TCR). Incubation was carried out at 37 °C for 30 min except when specified. Cells were washed twice in 5

ml FACS buffer prior to analysis. All samples were collected on a FACSCalibur flow cytometer and data

were analyzed with CellQuest (BD BioSciences) software; a minimum of 5000 live cells was analyzed per

sample.

Peptides

All HLA A2-restricted peptides (>95% purity; Invitrogen, Paisley, UK) were dissolved in DMSO and

diluted in RPMI medium to the desired concentrations. Binding affinity of peptides to HLA A2 was

estimated using the algorithm developed by Rammensee et al (14) available online at the following url:

http://syfpeithi.bmi-heidelberg.com/Scripts/MHCServer.dll/EpitopePrediction.htm.

CTL inhibition assay

HLA A2+ B cells (PK) were pulsed with the indicated peptide concentrations as described above or infected

with vaccinia virus. 5x103 target cells were then incubated with TCR tetramers or monomers, or bovine

serum albumin (BSA), at a final concentration of 10 µg/ml for 4 hr at 37 °C in 50 µl R5 medium. CTLs

maintained overnight in R5 were then added to the targets at a 5:1 (Tax-specific CTL) or 3:1 (HIV-1 Gag-

specific) effector:target ratio in 100 µl final volume and incubation was carried out for a further 4 hr at 37°C.

Sample supernatant was then harvested and MIP1β concentration quantified in duplicate assays by ELISA

according to the manufacturer’s instructions (R&D technologies, Abingdon, UK). Absorbance was


7

measured at 450 nm with a Bio-Rad 550 microplate reader; standard curves were constructed with each

assay.

Vaccinia virus

The construction of the recombinant vaccinia virus expressing the pX region of HTLV-1 used in this study

was reported in Parker et al. (15). PK B cells were infected with viral particles at a multiplicity of infection

(MOI)=5 in D-MEM with 0.1% BSA for 90 min at 37 °C. Cells were then washed once, resuspended in D-

MEM supplemented with 10% FCS and antibiotics as above and then incubated for 12 hr to allow for

protein expression.

Surface plasmon resonance

A Biacore 3000TM machine and CM-5 sensor chips were used. Approximately 5000 RU (response units) of

streptavidin was covalently linked to the chip surface in all four flow-cells using the amino-coupling kit

according to manufacturer’s instructions. Biotinylated pMHCI proteins and biotinylated control protein

(OX68) were bound to the sensor surfaces by flowing dilute solutions (50 µg/ml) of protein over the relevant

streptavidin-coated flow cell. 500 RU (for kinetics measurements) or 1000RU (equilibrium affinity

measurements) of protein ligand were bound to each flow cell. If biotinylated TCR monomers were to be

used, surfaces were blocked with 1 mM biotin in HEPES buffered saline (HBS). Soluble A6, JM22 and 1G4

TCRs were then flowed over the relevant flow-cells at a rate of 5 µl/minute (equilibrium) or of 50 µl/min

(kinetics) at the concentrations indicated. The 1G4 HLA A2-restricted, NY-ESO specific TCR was used as a

negative-binding control. All measurements were performed at 25°C using HBS buffer. Responses were

recorded in real time and analyzed using BIAevaluation software (BIAcore, Uppsala, Sweden). Equilibrium

dissociation constants (KDs) were determined assuming a 1:1 interaction (A + B <-> AB) by plotting specific

equilibrium binding responses against protein concentrations followed by non-linear least squares fitting of

the Langmuir binding equation: AB = B*ABmax/(KD + B), and were confirmed by linear Scatchard plot

analysis using Origin 6.0 software (Microcal; Northampton, MA). Kinetic binding parameters (kon and koff)

were determined using BIAevaluation software. koff values for TCR tetramers were estimated using

dissociation phase data at least 10 minutes after reagent binding to prevent interference from the small


8

fraction of weakly bound TCR tetramers on the BIAcore™ chip surface.


9

Results

Comparative kinetic parameters of A6 and JM22 TCRs in monomeric and multimeric form

The TCRs derived from the CTL clones A6 and JM22 were selected for this study. Each of these TCRs is

specific for a peptide presented in the context of HLA A*0201 (HLA A2 from hereon). The A6 TCR

recognizes an antigenic peptide that comprises residues 11-19 of the lymphotrophic retrovirus HTLV-1

transcription factor Tax (16). The co-crystal structure of the A6 TCR and its cognate ligand has been

determined (17). This TCR/pMHCI interaction has the highest affinity for a human TCR and a syngeneic

ligand measured to date (18). The JM22 TCR is specific for residues 58-66 of influenza A matrix protein

(19). The affinity and kinetic parameters of this interaction have been characterized (20) and the crystal

structure of the TCR complexed to its cognate ligand has recently been determined (21). Whereas

dissociation rates are similar for both TCRs and fall within the spectrum of values observed for all

TCR/pMHC interactions studied to date (22), the on-rate measured for the A6 TCR is unusually fast and

accounts for most of the difference in affinity between these two TCR/pMHCI interactions. Equilibrium

binding analyses of biotinylated A6 and JM22 TCRs in monomeric and multimeric forms are shown in

Figure 1. The multimers took longer than the monomers to reach equilibrium binding. This difference is

believed to reflect the steric rearrangements required for binding of the tetrameric molecules. Importantly,

binding specificity of the multimers was not affected and no response was monitored with irrelevant pMHCI

complexes. The most striking feature of this data is the substantial reduction in dissociation rate exhibited by

the multimerized TCRs compared to monomers. TCR tetramers exhibited complex dissociation kinetics,

with a fraction of TCR tetramers (14% for A6 and 43% for JM22) exhibiting a more rapid off-rate (koff); this

likely represents those reagents binding via only two pMHCI molecules or those binding in a sterically

compromised manner. The vast majority of the TCR tetramers, presumably those bound to three pMHCI

molecules, have a considerably slower (true) koff. The greatly reduced koff of tetramerized TCR results in

reagents that bind with a half-life of 154 min and 43 min for the A6 and JM22 TCR tetramers respectively

(Figure 1). Thus the A6 and JM22 TCR tetramers exhibit a 1339- and 444-fold slower dissociation rate than

the respective monomeric interactions. The dissociation of A6 TCR tetramers did not exhibit concentration

dependence (Supplementary Material Figure 2). The greatly increased half-life of tetramerized TCRs


10

provides a basis for the use of such reagents in cellular binding assays where the half-life of interaction is

critical for stable adhesion to the cell surface.

Soluble A6 and JM22 T cell receptors bind antigen-pulsed B cells specifically

Experimental parameters with multimerized TCRs were tested systematically to define the optimal staining

conditions described in the Experimental Procedures. Figure 2 shows a typical staining of peptide-pulsed

HLA A2+ B cells with the A6 and JM22 TCR multimers under optimal conditions. Background staining of

unpulsed cells was negligible as the mean fluorescence intensity (MFI) was similar to the value obtained for

unstained cells (data not shown). Antigen-specific staining was observed with peptide concentrations of

10-5 M and more in all cases. The A6 TCR multimer stained cognate peptide-loaded APC more efficiently

than the JM22 TCR multimer (compare Figures 2A and 2B). This difference in staining at high

concentrations of exogenous peptide probably reflects ligand affinity and off-rate differences (Figure 1).

TCR multimers do not exhibit cross-reactivity with unrelated self and non-self peptides

Multimerized A6 and JM22 TCRs do not stain HLA A2+ B cells without the addition of exogenous antigen

implying that they do not cross-react with the repertoire of self-peptides displayed by HLA A2 molecules on

the cell surface at physiological densities. In addition, experiments in which B cells pulsed with either the

HTLV-1 Tax11-19 peptide or the influenza matrix58-66 peptide were stained with the irrelevant TCR tetramer

failed to show detectable cross-staining with non-cognate ligand (Figure 3). These results indicate that the

recombinant proteins bind to the surface of APC in a peptide-dependent manner and that non-specific

binding does not occur. Thus, in vitro protein engineering and refolding do not seem to affect the binding

specificity of these TCRs. However, the possibility remained that these reagents could bind high amounts of

other unrelated, non-self peptides displayed by HLA A2 molecules on the cell surface. To rule out this

possibility, we stained HLA A2+ B cells pulsed with high concentrations (200 µM) of several HLA A2-

restricted peptides derived from different pathogens, some of them exhibiting a degree of sequence

similarity with HTLV-1 Tax11-19 and influenza matrix58-66 at non-anchor residues. No non-specific binding of

TCR multimers to cells pulsed with any of these peptides was detected (Figure 3). Even though this does not


11

entirely rule out the possibility of cross-reactivity, these results confirm that the A6 and JM22 TCRs retain

cognate antigen specificity in soluble multimeric form.

TCR multimers can sense quantitative modifications of antigen display at the cell surface and stain APCs

processing endogenous antigen

Monoclonal antibodies (mAbs) have been used previously to examine the presentation of specific pMHC

(23-33). Generation of such reagents is laborious and involves extensive screening of mAbs, although some

recent technical advances may facilitate this process (34,35). The use of soluble TCRs may prove to be

advantageous in this respect. O’Herrin et al. used soluble TCR dimers to monitor quantitatively and

qualitatively the up-regulation of pMHC molecules by IFNγ (36). We repeated this result in our systems;

HLA A2+ B cells cultured in medium supplemented with 50 U/ml IFNγ for 48 hrs, pulsed with peptide and

then stained with soluble TCR multimers exhibited higher levels of specific staining than cells grown in

IFNγ-free medium (data not shown). As the staining of TCR multimers seemed to reflect the up-regulation

of MHC molecules induced by IFNγ at the cell surface, we examined whether we could monitor similar

effects induced by other biomolecules. β2-microglobulin (β2m) is known to enhance the activation of CTLs

when added exogenously in bioassays (37,38). Although the precise molecular events responsible for this

effect are not completely understood, it is clearly associated with an increase in the number of antigenic

pMHCI complexes on the cell surface. The effect of β2m on TCR multimer staining was monitored using

cells pulsed with varying concentrations of peptide in the presence or absence of high concentrations of

exogenous β2m (Figure 4). In the absence of β2m, the minimum peptide concentration for which specific

staining occured above background (detection threshold) was 10 µM (Figures 2, 4A&4B). At 10 µM peptide

concentration, the MFI values were substantially increased in the presence of exogenous β2m (from 10.43 to

36.28 for the A6 TCR multimer, and from 10.76 to 21.00 for the JM22 TCR multimer in the experiment

shown) (Figure 4A&B). Further, the addition of exogenous β2m lowered the detection threshold, for both

the A6 and JM22 TCR multimers, to allow specific staining of cells pulsed with 1 µM cognate peptide

(Figure 4A&B). These data are consistent with previous observations from studies using conformation-

specific antibodies and highlight the dramatic effect of exogenous β2m on the density of cell surface pMHCI

complexes. The effect of β2m on TCR binding was dose-dependent with a dynamic range from 1 – 100


12

µg/ml (data not shown). The stainings shown in this study were carried out with higher β2m concentrations

to ensure optimal effect. In Figure 4C, similar concentrations of exogenous β2m were used to facilitate

staining of HLA A2+ cells infected with recombinant vaccinia virus grown in medium supplemented with

IFNγ. Cells were stained 12 hours post-infection by vaccinia expressing the HTLV-1 Tax protein or an

irrelevant vaccinia. The A6 TCR multimer stained only HLA A2+ B cells infected with vaccinia expressing

the Tax protein (Figure 4C). That the detection of naturally processed ligand required the presence of

exogenous β2m indicates that β2m-mediated enhancement of pMHCI antigen density on the cell surface

does not operate through peptide exhange catalysis, but likely reflects an effect on conformational

stabilization.

Increasing the order of multimerization does not enhance the sensitivity of antigen detection

The ability of TCR multimers to detect APCs bearing cognate peptide on their cell surface is a function of

the reduced dissociation rate compared to the monomeric molecule. We therefore reasoned that increasing

the order of multimerization might lead to further incremental increases in detection threshold. However,

microbeads coated with fluorochrome-conjugated TCRs were not able to distinguish lower levels of cell

surface antigen than TCR tetramers; addition of a cross-linker induced further increases in background

staining (Supplementary Material Figure 3 and data not shown).

Antigen-specific inhibition of cytotoxic T lymphocyte (CTL) activation by soluble TCRs

The detection threshold of specific antigens on the cell surface using TCR multimers in combination with

flow cytometry, even under optimal conditions, is limited to a relatively high peptide concentration

(i.e. 10-6 M); this probably corresponds to a cell surface antigen density in excess of that produced by natural

intracellular processing (34). To test whether binding could occur at lower antigen densities, a more

sensitive experimental system was required. Previously, we have shown that phycoerythrin (PE)-conjugated

pMHCI multimers can induce functional effects in CD8+ T cells at concentrations more than 2 orders of

magnitude lower than can be detected by flow cytometry (Sewell, unpublished observations). We reasoned

that TCR multimers could compete for antigens displayed on the cell surface of APCs with TCRs expressed

by T cells specific for the same epitope, and therefore inhibit activation of CTL. Figure 5A shows the effect


13

of pre-incubating target cells with multimerized TCRs on the activation of an HLA A2-restricted HTLV-1

Tax11-19 specific CTL line. The A6 TCRs inhibited macrophage inflammatory protein (MIP)1β release by

over two-fold for peptide concentrations ranging from 10-5 M to 10-8 M as well as for peptide produced

endogenously by recombinant vaccinia virus. CTL activation did not seem to be completely abrogated by

pre-incubation with the TCRs, since for higher peptide concentrations (>10-7 M), MIP1β release remained

above background. Pre-incubation with the JM22 TCR tetramer failed to decrease MIP1β production,

indicating that the inhibition was antigen-specific. TCR binding thus occurs at low antigen density, more

related to physiological levels, and with peptides produced in an endogenous manner, conditions in which

flow cytometry fails to detect any binding of these reagents. This result highlights the limitations of the latter

technique in terms of sensitivity when the amount of ligand on the cell surface is restricted. The limits of

detection of PE on a FACSCalibur are >150 molecules/cell (BD Biosciences Scientific Support, UK).

pMHCI tetramers are thought to cross-link at least three different TCRs on the T cell surface (39). Assuming

the same kind of binding for TCR tetramers, >450 antigenic pMHCI molecules/cell will be required to

observe positive staining by flow cytometry in our experiments. Interestingly, TCR monomers failed to

mediate inhibition of CTL activation (Figure 5C). This probably reflects a requirement for multimerization

in order to achieve stable binding of soluble TCRs, but could also indicate that inhibition of CTL activation

is mediated through steric hindrance due to the large PE molecule in the complex rather than specific

masking of the antigen. Such steric effects might be expected to prevent access of CTL to cognate and non-

cognate pMHCI molecules on the APC surface. Figure 5B shows the effect of A6 TCR multimers on the

activation of a CTL clone specific for an HLA A2-restricted HIV-1 p17 Gag-derived peptide

(SLYNTVATL; residues 77-85). In this experiment, HLA A2+ B cells pulsed with both the Tax11-19 peptide

and the HIV-1 p17 Gag77-85 peptide were stained with A6 TCR tetramers and assayed for their ability to

elicit activation of either HIV-1 Gag- or HTLV-1 Tax-specific CTLs. Incubation of the targets with A6 TCR

multimers inhibited activation of the HTLV-1 CTL but did not inhibit activation of the HIV-1 specific CTLs

by these same targets. We observed similar TCR-specific inhibition of CTL activation using tetramers made

with unconjugated streptavidin, thus excluding steric hindrance due to the large PE moiety in the observed

CTL inhibition (data not shown).


14

The A6 recombinant TCR multimers show specificity similar to the native cellular protein

Given that functional effects can occur below the level of detection by flow cytometry, cross-reactivity

becomes an even more significant concern when considering the application of these reagents for cellular

targeting and therapeutic intervention in vivo. We therefore undertook experiments to characterize this

potential confounding factor further. In an effort to understand the importance of individual amino-acid side

chains of a peptide in the TCR/pMHC interaction for T cell activation, Hausmann et al. extensively tested

the effect of single amino-acid substitutions on the effector functions of CTL clones specific for the HLA

A2-restricted HTLV-1 Tax11-19 antigen (40). From these substitutions they inferred different recognition

motifs, which were used to search databases for protein sequences matching these motifs. Several peptides,

all of them partial agonists eliciting significant levels of activation in A6 CTL, were identified. These

“mimotopes” represented a good system to test whether the recombinant A6 TCR retained the specificity of

the native molecule expressed on the T cell surface. Staining was carried out with seven of these peptides,

chosen according to their activation potential. One was derived from the Tel1 protein of Saccharomyces

cerevisiae (Tel1p549-557 MLWGYLQYV) and the other six from human proteins (HuD87-95 LGYGFVNYI,

BENE54-62 LLQGWVMYV, Phosphofructokinase572-580 TMGGYCGYL, Protein tyrosine phosphatase1073-1081

DLKGFLSYL, Protein tyrosine kinase32-40 SLHGYKKYL and HuR61-69 LGYGFVNYV). Flow cytometry

was used to determine that tetramerized A6 TCR bound to the surface of HLA A2+ APC pulsed with just

two of these mimitopes (Figure 6). These were HuD87-95, a nonameric peptidic fragment derived from a

protein expressed in neurons, and Tel1p. Given the characteristically weak nature of TCR/pMHC

interactions, the affinity of a particular TCR for cross-reactive, non-optimal ligands was expected to be very

low. Accordingly, A6 TCR multimers only stained APC pulsed with very high concentrations of cross-

reactive peptides. The detection threshold was 0.5 and 1 mM for Tel1p and HuD87-95 respectively; staining of

HLA A2+ B-cells pulsed with 0.1 mM of the Tax11-19 peptide is shown for comparison (Figure 6). Although

weak, binding was peptide-specific and concentration dependent. These results highlight the fact that the

recombinant molecules retain the specificity of the native molecule and confirm the cellular data published


15

by Hausmann et al. (40). The Tel1p and HuD87-95 peptides selected using the A6 TCR tetramers were taken

forward for further analysis by SPR.

SPR analysis of the affinity between A6 TCR and cross-reactive ligands

A6 TCR tetramer staining of APCs pulsed with HuD87-95 and Tel1p was inefficient compared to staining of

APCs pulsed with Tax11-19, likely reflecting differences in TCR/pMHCI binding affinities. However, other

parameters not assessed in this study, such as the efficiency of peptide binding to HLA A2, could influence

staining of APCs with TCR multimers. Estimations of relative binding affinities of both cross-reactive

peptides and Tax11-19 peptide for HLA A2 were calculated. Predictive scores are shown in Table I and

indicate that both peptides are likely to bind HLA A2, although more weakly in the case of HuD87-95. In

order to measure the actual affinities of the interactions between the antigen and the A6 TCR, soluble HLA

A2 molecules folded around the HuD87-95 and Tel1p peptides were synthesized and purified. Figures 7A&C

show the binding response data of soluble A6 TCR at a range of concentrations (224 µM and two-fold

dilutions thereof) injected over, respectively, HLA A2-Tel1p and HLA A2-HuD87-95 complexes immobilized

on the BIAcore flow cell. The equilibrium binding response of both interactions is plotted for each TCR

concentration in a non-linear fit of the Langmuir isotherm (Figures 7 B&D). Fits of three independent

measurements gave a mean KD of 38.6 µM for HLA A2-Tel1p and 123.3 µM for HLA A2-HuD87-95. These

values and the corresponding standard deviations are shown in Table I. A6 TCR/ HLA A2-HuD87-95 binding

did not reach saturation; this is likely to result in a lower accuracy of the KD estimation. The 1G4 TCR

specific for the tumor epitope HLA A2-NY-ESO (10) failed to bind to any of these ligands (data not shown).

Positive control data are shown for comparison (Figures 7E&F). Altogether, these results show that the

cross-reactive ligands exhibit significant binding affinities with the A6 TCR.

A high affinity variant of the A6 TCR inhibits CTL activation as a monomeric molecule

Although multimeric TCRs could inhibit the activation of CTL with identical specificities, the detection of

cross-reactivity with self-ligands is concerning in terms of translation of this technology into a therapeutic

setting. Further, the administration of multimeric TCRs may be problematic in vivo, and monomeric A6

TCR failed to inhibit the activation of HTLV-1-specific CTLs. The absence of a detectable biological effect


16

with monomeric wildtype TCR is presumably a consequence of the short half-life of the TCR/pMHCI

interaction. We therefore developed a phage display system to select a mutant of the A6 TCR that binds to

the cognate HLA A2-LLFGYPVYV antigen with extremely high affinity (KD = 2.5 nM) without apparent

loss of specificity; the affinity increment is due to a substantial increase in off-rate (1 and Supplementary

Material Figure 4). Previous studies have shown that the removal of unfavourable structural features can

repair T cell recognition of antagonist peptides (18,41). Similarly, structural analysis demonstrated that the

enhanced affinity of our mutant TCR (A6c134) is due to contact optimization with bound peptide1. The

A6c134 TCR binds to HLA A2-LLFGYPVYV antigen with a half-life of 3900s, and can stain cells pulsed

with as little as 10-8M antigen1. This enhanced affinity translated into a greatly increased biological potency

compared to wildtype recombinant TCR protein; indeed, the A6c134 TCR protein inhibited the activation of

HTLV-1 Tax11-19 specific CTL even as a monomeric molecule (Figure 5C). The inhibition of CTL activation

by monomeric high affinity A6c134 TCR was comparable to that seen with multimeric wildtype A6 TCR

(Figure 5C) for an identical peptide concentration. A6c134 TCR was also able to inhibit the release of lytic

granules by an HTLV-1 Tax11-19 specific CTL line (Supplementary Material Figure 5).

The A6 TCR variant exhibits increased affinity selectively for the HLA A2-restricted Tax11-19 parent epitope

The A6c134 TCR exhibited minimal cross-reactivity with self peptides at physiological concentrations as

demonstrated by the absence of background staining in flow cytometric assays; further, there was no

detectable staining of APCs pulsed with several unrelated peptides1. To demonstrate formally that the

increased affinity was specific for the wildtype Tax11-19 epitope, we studied the interactions of A6c134 with

the Tel1p and HuD87-95 epitopes to which cross-reactivity had been detected with the parent A6 TCR. SPR

showed that A6c134 and A6 TCRs bound to the Tel1p epitope with very similar affinity (KD = 46.5 and 38.6

µM respectively; Figure 7B & 8D). A6c134 bound the HuD87-95 epitope with an approximately five-fold

increased affinity compared to the A6 parent TCR (KD= 21 µM compared to 123 µM; Figures 7D & 8E)

however this difference is minimal when compared to the 400-fold increased binding observed with HLA

A2-Tax11-19 (KD = 2.5 nM compared to > 0.9 µM1). Figure 8A shows the comparative binding responses of

A6c134 with HLA A2-Tax11-19, the crossreactive ligand HLA A2-HuD and a negative control. The

association rate appears significantly slower in the case of HLA A2-Tax11-19. Yet, consistent with the


17

findings of Li et al.1, the most striking feature is the dramatically slower dissociation rate exhibited by the

A6c134 monomers. This is illustrated in the first two injection cycles, during which dissociation of the TCR

is barely detectable, and by the saturation of immobilized complexes as concentration of the analyte

increases. Altogether, these results indicate that TCR affinity can be selectively increased for a specific

peptide without concomitant major increases in the binding of related ligands.


18

Discussion

In this study, we generated soluble human TCRs that, like membrane-bound TCRs, comprised two

polypeptide chains; these were expressed separately from two distinct bacterial vectors and paired by a non-

native disulfide bond. This strategy of protein engineering produces a protein similar to the native globular

TCR that is extremely stable. The A6 and JM22 TCRs studied here appear structurally authentic, maintained

their ligand specificity in this form and were used to optimize the targeting of cells expressing cognate

antigen. To overcome the intrinsically low affinity of TCR/pMHC interactions, we adopted two approaches.

First, we increased overall avidity by multimerization to reduce the composite dissociation rate; this

approach has been exploited successfully to allow identification of antigen-specific T cells with soluble

pMHC molecules (39,42). Second, we adopted mutational strategies to engineer a TCR from its A6 parent

that had dramatically increased affinity for cognate ligand due to a reduction in off-rate. The TCR yielded by

this latter approach was more sensitive than multimeric forms of the parent TCR in terms of its ability to

detect cognate antigen on the target cell surface, and exhibited potent biological activity with minimal cross-

reactivity. The soluble heterodimeric TCRs manufactured in this study were utilized in a variety of

functional assays including specific detection of target cells pulsed with cognate peptide, discrimination of

quantitative changes in antigen display at the cell surface, identification of virus-infected cells, inhibition of

antigen-specific CTL activation and identification of cross-reactive peptides.

Part of this work has developed systems that utilize TCRs for the specific identification of cell surface

pMHCI complexes. In this regard, several noteworthy features emerge from the data. First, PE-conjugated

TCR multimers failed to emit a signal detectable by standard flow cytometric techniques when target cells

were pulsed with low antigen concentrations (Figure 2). However, functional assays clearly demonstrated

specific and biologically relevant binding at antigen densities well below the fluorescence detection limit

(Figure 5). The requirement for high antigen density in order to visualize TCR multimer binding is therefore

a methodological issue. In terms of detection sensitivity, the apparent discrepancy between TCR and pMHCI

multimers, which are widely used to identify and characterize antigen-specific T cell populations directly ex

vivo by flow cytometry, can be explained by respective ligand density differences on the target cell surface.

A T cell is thought to bear approximately 30,000-100,000 TCRs (43), all of identical idiotypic specificity; at


19

physiological antigen densities, however, only a small fraction of the 50,000 or so MHCI molecules on the

surface of an APC (44) will present the epitope cognate for the TCR clonotype. Thus, it appears that flow

cytometric detection of specific pMHCI at low antigen densities using soluble recombinant TCR-based

reagents is limited by poor signal intensity. Second, despite these limitations, TCR multimers proved to be

useful tools that enabled the discrimination of quantitative changes in antigen presentation on the target cell

surface. Indeed, the effects of both β2m and IFNγ on the loading of exogenous peptide could be monitored

by flow cytometry using these reagents. It is well documented that the addition of exogenous β2m can

enhance the apparent sensitivity of CTL to antigen (37). Other studies have shown that the addition of

exogenous β2m concomitant with peptide caused a vast increase in the number of conformationally correct

pMHCI molecules on the cell surface as monitored with conformation-specific antibodies (38,45). In the

present study, the addition of exogenous β2m at a given concentration of peptide increased the number of

TCR-cognate ligands by at least two-fold and sufficiently increased the sensitivity of flow cytometric

detection to allow the visualization of endogenously processed pMHCI antigen (Figure 4). These

observations could be extended to enable tracking of specific viral antigen expression in many settings. The

synergistic effects of IFNγ, which enhances processing and presentation of endogenously generated antigen

(46), might further enhance the general applicability of either TCR multimers or high affinity variant TCRs

for this purpose. Third, TCR multimers were used to screen for and identify syngeneic cross-reactive

pMHCI ligands (Figure 6); this approach eliminates the effects of adhesion/costimulatory molecular

interactions that potentially confound cellular screening assays (47,48) and has recently been validated in

class II-restricted systems (49). Surface plasmon resonance studies confirmed binding, with measured

affinities for both HLA A2-Tel1p and HLA A2- HuD87-95 lying at the lower end of the spectrum of

previously defined TCR/pMHC interactions (22) (Figure 7). In order to ensure an efficient adaptative

immune response, it is advantageous that a single T cell clonotype can potentially engage in a functionally

productive manner with several foreign peptides (50). On the molecular level, the corollary of this

hypothesis is that a TCR clonotype will exhibit sufficient intrinsic conformational diversity to cross-react

with several molecular mimics and less structurally related pMHCI complexes (51-53). Although

alloreactive and/or xenoreactive ligands have been identified for other TCRs (51,53-56), few cross-reactive

syngeneic interactions between naturally occurring pMHCI and cognate TCR have been identified (57-60).


20

Indeed, to the best of our knowledge, this study is the first to characterize the interaction between a

pathogen-reactive human TCR and a cross-reactive self-peptide in biophysical terms.

The cross-reactivity identified in this study has several important implications. First, it instructs on the limits

of thymic selection. Of the cross-reactive human peptides capable of eliciting functional responses through

the A6 TCR in the study by Hausmann et al., only HuD87-95 bound the corresponding TCR tetramer in our

hands by flow cytometry. Thus, this peptide likely ranks at the upper end of the A6 TCR affinity spectrum in

terms of interactions with human peptides; indeed, this assumption was confirmed by SPR analysis for those

peptides tested (data not shown). In a broader context, we have examined the binding of TCRs from several

human immunodominant anti-viral CTL to their cognate ligand in addition to the A6 and JM22 TCRs

studied here. All of these TCRs have a relatively high affinity for their cognate ligand (KD 0.9-10 µM). The

TCR/pMHCI interaction of anti-tumour CTL appears to be weaker than this (KD = 20-40 µM) (Boulter et

al., manuscript in preparation). Thus, the affinity of the interaction between HLA A2-HuD87-95 and the A6

TCR (KD =120 µM) is only 4 times weaker than that of TCRs capable of mediating functional T cell

responses in the periphery. These data suggest that thymic selection operates within narrow limits, and are

consistent with studies in murine systems (61). Second, the cross-reactivity of A6 with HLA A2-HuD87-95

might be relevant in autoimmunity. Paraneoplastic neurological degenerations (PNDs) are disorders that

develop in patients with coexistent malignancies. PNDs are believed to be triggered by an anti-tumour

immune response directed against neuronal antigens expressed inside tumour cells (62). Curiously, the HuD

protein has been implicated as a source of such antigen in one form of PND called anti-Hu syndrome (62-

64). It is believed that the expression of HuD by small cell lung cancer triggers an immune response against

this protein, which is exclusively neuron-expressed in healthy individuals (63,64). A recent study examined

the recognition and processing of 14 computer-predicted HLA A*0201-restricted, HuD-derived epitopes

(64). Only one of these peptides, comprising HuD residues 86-95, was generated through the MHCI antigen-

processing pathway; HuD86-95 was found to be immunogenic in experimental systems (64). HuD86-95

contains the shorter 9mer HuD87-95 peptide (LGYGFVNYI) that cross-reacts with the A6 TCR in our study;

HLA A2-HuD87-95 binds the A6 TCR tetramer to a slightly greater extent than the longer HuD86-95 peptide

(data not shown). Cross-recognition of a neuron-derived epitope by a Tax11-19-specific TCR might be


21

relevant to HTLV-1-associated pathology. 1-2% of HTLV-1 infected individuals, particularly those with a

high viral load, develop HTLV-1-associated myelopathy/tropical spastic paraparsis (HAM/TSP). The HLA

A2-restricted Tax11-19 epitope is highly immunodominant in HLA A2+ individuals infected with HTLV-1,

and appears to elicit especially large CTL responses in the setting of HAM/TSP (65,66). Further, HTLV-1-

specific lymphocytes are enriched in the cerebrospinal fluid of individuals with HAM/TSP (65). These

findings suggest that such CTL might cause HTLV-1-associated myeloneuropathy (65-67); this could

potentially result either from direct recognition of viral antigen in central nervous system (CNS) tissue or

from cross-recognition of CNS-derived self antigens such as HuD87-95. The latter hypothesis is supported by

elegant studies showing that Tax11-19-stimulated T cell lines from two HLA A2+ individuals with HTLV-1-

associated myelopathy killed targets pulsed with HuD86-95 peptide (40).

The ability of TCR multimers to target cells expressing specific antigens provides a vehicle for the delivery

of therapeutic interventions selectively to sites of disease. In comparison to systemic delivery systems,

targeted therapeutics might both enhance efficacy and minimize side-effects. We show here, for example,

that both wildtype multimeric TCRs and engineered high affinity monomeric TCRs can be used to conceal

specific antigens from cognate CTL (Figure 5). In future, it might be possible to use similar technology in

vivo to down-modulate cellular autoimmune reactions. Alternatively, soluble TCRs could be used to deliver

cytotoxic agents or to redirect pathogen-specific immune responses; this latter principle has recently been

demonstrated using pMHCI-conjugated antibodies that allowed antiviral CTL to induce regression of human

tumour xenografts in a murine model (68). However, the observation of TCR cross-reactivity is a potentially

confounding factor in the context of targeted therapy. In this respect, the development of engineered TCRs

with artificially enhanced binding affinities for cognate ligand represents a dramatic advance ((69,70) and 1).

Importantly, this increased affinity is not accompanied by concomitant increases in cross-reactivity (Figure

8)1. Thus, a strategy for the optimization of targeted therapeutics is defined that could be developed for

multiple applications, most notably perhaps within the fields of tumour immunology and autoimmunity.


22

REFERENCES

1. Davis, M. M., Boniface, J. J., Reich, Z., Lyons, D., Hampl, J., Arden, B., and Chien, Y. (1998) Annu Rev Immunol 16, 523-544

2. Konig, R. (2002) Curr Opin Immunol 14, 75-83. 3. Irvine, D. J., Purbhoo, M. A., Krogsgaard, M., and Davis, M. M. (2002) Nature 419, 845-849 4. Purbhoo, M. A., Irvine, D. J., Huppa, J. B., and Davis, M. M. (2004) Nat Immunol 5, 524-530 5. Huston, J. S., Levinson, D., Mudgett-Hunter, M., Tai, M. S., Novotny, J., Margolies, M. N., Ridge,

R. J., Bruccoleri, R. E., Haber, E., Crea, R., and et al. (1988) Proc Natl Acad Sci U S A 85, 5879-5883

6. Hoo, W. F., Lacy, M. J., Denzin, L. K., Voss, E. W., Jr., Hardman, K. D., and Kranz, D. M. (1992) Proc Natl Acad Sci U S A 89, 4759-4763

7. Plaksin, D., Polakova, K., McPhie, P., and Margulies, D. H. (1997) J Immunol 158, 2218-2227 8. Lake, D. F., Salgaller, M. L., van der Bruggen, P., Bernstein, R. M., and Marchalonis, J. J. (1999)

Int Immunol 11, 745-751 9. Gregoire, C., Rebai, N., Schweisguth, F., Necker, A., Mazza, G., Auphan, N., Millward, A.,

Schmitt-Verhulst, A. M., and Malissen, B. (1991) Proc Natl Acad Sci U S A 88, 8077-8081 10. Boulter, J. M., Glick, M., Todorov, P. T., Baston, E., Sami, M., Rizkallah, P., and Jakobsen, B. K.

(2003) Protein Eng 16, 707-711 11. Sewell, A. K., Harcourt, G. C., Goulder, P. J., Price, D. A., and Phillips, R. E. (1997) Eur J

Immunol 27, 2323-2329 12. Cull, M. G., and Schatz, P. J. (2000) Methods Enzymol 326, 430-440 13. Willcox, B. E., Gao, G. F., Wyer, J. R., O'Callaghan, C. A., Boulter, J. M., Jones, E. Y., van der

Merwe, P. A., Bell, J. I., and Jakobsen, B. K. (1999) Protein Sci 8, 2418-2423 14. Rammensee, H., Bachmann, J., Emmerich, N. P., Bachor, O. A., and Stevanovic, S. (1999)

Immunogenetics 50, 213-219 15. Parker, C. E., Daenke, S., Nightingale, S., and Bangham, C. R. (1992) Virology 188, 628-636 16. Utz, U., Koenig, S., Coligan, J. E., and Biddison, W. E. (1992) J Immunol 149, 214-221 17. Garboczi, D. N., Ghosh, P., Utz, U., Fan, Q. R., Biddison, W. E., and Wiley, D. C. (1996) Nature

384, 134-141 18. Ding, Y. H., Baker, B. M., Garboczi, D. N., Biddison, W. E., and Wiley, D. C. (1999) Immunity 11,

45-56 19. Lehner, P. J., Wang, E. C., Moss, P. A., Williams, S., Platt, K., Friedman, S. M., Bell, J. I., and

Borysiewicz, L. K. (1995) J Exp Med 181, 79-91 20. Willcox, B. E., Gao, G. F., Wyer, J. R., Ladbury, J. E., Bell, J. I., Jakobsen, B. K., and van der

Merwe, P. A. (1999) Immunity 10, 357-365 21. Stewart-Jones, G. B., McMichael, A. J., Bell, J. I., Stuart, D. I., and Jones, E. Y. (2003) Nat

Immunol 4, 657-663 22. Van Der Merwe, P. A., and Davis, S. J. (2003) Annu Rev Immunol 21, 659-684 23. Aharoni, R., Teitelbaum, D., Arnon, R., and Puri, J. (1991) Nature 351, 147-150. 24. Chung, D. H., Belyakov, I. M., Derby, M. A., Wang, J., Boyd, L. F., Berzofsky, J. A., and

Margulies, D. H. (2001) J Immunol 167, 699-707 25. Dadaglio, G., Nelson, C. A., Deck, M. B., Petzold, S. J., and Unanue, E. R. (1997) Immunity 6,

727-738 26. Eastman, S., Deftos, M., DeRoos, P. C., Hsu, D. H., Teyton, L., Braunstein, N. S., Hackett, C. J.,

and Rudensky, A. (1996) Eur J Immunol 26, 385-393. 27. Messaoudi, I., LeMaoult, J., and Nikolic-Zugic, J. (1999) J Immunol 163, 3286-3294 28. Morkowski, S., Raposo, G., Kleijimeer, M., Geuze, H. J., and Rudensky, A. Y. (1997) Eur J

Immunol 27, 609-617. 29. Murphy, D. B., Rath, S., Pizzo, E., Rudensky, A. Y., George, A., Larson, J. K., and Janeway, C. A.,

Jr. (1992) J Immunol 148, 3483-3491. 30. Polakova, K., Plaksin, D., Chung, D. H., Belyakov, I. M., Berzofsky, J. A., and Margulies, D. H.

(2000) J Immunol 165, 5703-5712 31. Puri, J., Arnon, R., Gurevich, E., and Teitelbaum, D. (1997) J Immunol 158, 2471-2476.


23

32. Reay, P. A., Matsui, K., Haase, K., Wulfing, C., Chien, Y. H., and Davis, M. M. (2000) J Immunol 164, 5626-5634.

33. Zhong, G., Reis e Sousa, C., and Germain, R. N. (1997) Proc Natl Acad Sci U S A 94, 13856-13861.

34. Cohen, C. J., Sarig, O., Yamano, Y., Tomaru, U., Jacobson, S., and Reiter, Y. (2003) J Immunol 170, 4349-4361

35. Biddison, W. E., Turner, R. V., Gagnon, S. J., Lev, A., Cohen, C. J., and Reiter, Y. (2003) J Immunol 171, 3064-3074

36. O'Herrin, S. M., Lebowitz, M. S., Bieler, J. G., al-Ramadi, B. K., Utz, U., Bothwell, A. L., and Schneck, J. P. (1997) J Exp Med 186, 1333-1345

37. Shields, M. J., Kubota, R., Hodgson, W., Jacobson, S., Biddison, W. E., and Ribaudo, R. K. (1998) J Biol Chem 273, 28010-28018

38. Rock, K. L., Rothstein, L. E., Gamble, S. R., and Benacerraf, B. (1990) Proc Natl Acad Sci U S A 87, 7517-7521

39. McMichael, A. J., and O'Callaghan, C. A. (1998) J Exp Med 187, 1367-1371 40. Hausmann, S., Biddison, W. E., Smith, K. J., Ding, Y. H., Garboczi, D. N., Utz, U., Wiley, D. C.,

and Wucherpfennig, K. W. (1999) J Immunol 162, 5389-5397 41. Baker, B. M., Gagnon, S. J., Biddison, W. E., and Wiley, D. C. (2000) Immunity 13, 475-484 42. Altman, J. D., Moss, P. A., Goulder, P. J., Barouch, D. H., McHeyzer-Williams, M. G., Bell, J. I.,

McMichael, A. J., and Davis, M. M. (1996) Science 274, 94-96 43. Schodin, B. A., Tsomides, T. J., and Kranz, D. M. (1996) Immunity 5, 137-146 44. Yewdell, J. W. (2001) Trends Cell Biol 11, 294-297 45. Otten, G. R., Bikoff, E., Ribaudo, R. K., Kozlowski, S., Margulies, D. H., and Germain, R. N.

(1992) J Immunol 148, 3723-3732 46. Hengel, H., Lucin, P., Jonjic, S., Ruppert, T., and Koszinowski, U. H. (1994) J Virol 68, 289-297 47. Hemmer, B., Vergelli, M., Pinilla, C., Houghten, R., and Martin, R. (1998) Immunol Today 19,

163-168 48. Hiemstra, H. S., van Veelen, P. A., Willemen, S. J., Benckhuijsen, W. E., Geluk, A., de Vries, R.

R., Roep, B. O., and Drijfhout, J. W. (1999) Eur J Immunol 29, 2385-2391 49. Crawford, F., Huseby, E., White, J., Marrack, P., and Kappler, J. W. (2004) PLoS Biol 2, E90 50. Mason, D. (1998) Immunol Today 19, 395-404 51. Garcia, K. C., Tallquist, M. D., Pease, L. R., Brunmark, A., Scott, C. A., Degano, M., Stura, E. A.,

Peterson, P. A., Wilson, I. A., and Teyton, L. (1997) Proc Natl Acad Sci U S A 94, 13838-13843 52. Holler, P. D., and Kranz, D. M. (2004) Mol Immunol 40, 1027-1031 53. Reiser, J. B., Darnault, C., Gregoire, C., Mosser, T., Mazza, G., Kearney, A., van der Merwe, P. A.,

Fontecilla-Camps, J. C., Housset, D., and Malissen, B. (2003) Nat Immunol 4, 241-247 54. Reiser, J. B., Darnault, C., Guimezanes, A., Gregoire, C., Mosser, T., Schmitt-Verhulst, A. M.,

Fontecilla-Camps, J. C., Malissen, B., Housset, D., and Mazza, G. (2000) Nat Immunol 1, 291-297 55. Corr, M., Slanetz, A. E., Boyd, L. F., Jelonek, M. T., Khilko, S., al-Ramadi, B. K., Kim, Y. S.,

Maher, S. E., Bothwell, A. L., and Margulies, D. H. (1994) Science 265, 946-949 56. Lang, H. L., Jacobsen, H., Ikemizu, S., Andersson, C., Harlos, K., Madsen, L., Hjorth, P.,

Sondergaard, L., Svejgaard, A., Wucherpfennig, K., Stuart, D. I., Bell, J. I., Jones, E. Y., and Fugger, L. (2002) Nat Immunol 3, 940-943

57. Hiemstra, H. S., Schloot, N. C., van Veelen, P. A., Willemen, S. J., Franken, K. L., van Rood, J. J., de Vries, R. R., Chaudhuri, A., Behan, P. O., Drijfhout, J. W., and Roep, B. O. (2001) Proc Natl Acad Sci U S A 98, 3988-3991

58. Hagerty, D. T., and Allen, P. M. (1995) J Immunol 155, 2993-3001 59. Quaratino, S., Thorpe, C. J., Travers, P. J., and Londei, M. (1995) Proc Natl Acad Sci U S A 92,

10398-10402 60. Wucherpfennig, K. W., and Strominger, J. L. (1995) Cell 80, 695-705 61. Alam, S. M., Travers, P. J., Wung, J. L., Nasholds, W., Redpath, S., Jameson, S. C., and Gascoigne,

N. R. (1996) Nature 381, 616-620 62. Albert, M. L., and Darnell, R. B. (2004) Nat Rev Cancer 4, 36-44 63. Szabo, A., Dalmau, J., Manley, G., Rosenfeld, M., Wong, E., Henson, J., Posner, J. B., and

Furneaux, H. M. (1991) Cell 67, 325-333


24

64. Plonquet, A., Garcia-Pons, F., Fernandez, E., Philippe, C., Marquet, J., Rouard, H., Delfau-Larue, M. H., Kosmatopoulos, K., Lemonnier, F., Farcet, J. P., Gherardi, R. K., and Langlade-Demoyen, P. (2003) J Neuroimmunol 142, 93-100

65. Elovaara, I., Koenig, S., Brewah, A. Y., Woods, R. M., Lehky, T., and Jacobson, S. (1993) J Exp Med 177, 1567-1573

66. Utz, U., Banks, D., Jacobson, S., and Biddison, W. E. (1996) J Virol 70, 843-851 67. Jacobson, S., Shida, H., McFarlin, D. E., Fauci, A. S., and Koenig, S. (1990) Nature 348, 245-248 68. Lev, A., Noy, R., Oved, K., Novak, H., Segal, D., Walden, P., Zehn, D., and Reiter, Y. (2004) Proc

Natl Acad Sci U S A 101, 9051-9056 69. Shusta, E. V., Holler, P. D., Kieke, M. C., Kranz, D. M., and Wittrup, K. D. (2000) Nat Biotechnol

18, 754-759 70. Holler, P. D., Holman, P. O., Shusta, E. V., O'Herrin, S., Wittrup, K. D., and Kranz, D. M. (2000)

Proc Natl Acad Sci U S A 97, 5387-5392


25

Footnotes

**This work was funded by the Wellcome Trust. AKS is a Wellcome Trust Senior Fellow.

#DAP is a Medical Research Council Clinician Scientist.

1Li et al submitted

We thank Anton van der Merwe for helpful suggestions and critical reading of the manuscript


26

Figure legends Figure 1. Binding of A6 and JM22 TCR monomers and tetramers to HLA A2-Tax and HLA A2-Flu

complexes. Streptavidin was linked to a BIAcore™ CM-5 chip by amine coupling, biotin-tagged pMHCI

was loaded onto each flow cell, and data were collected at 25°C with a flow rate of 5 µl/min. 5 µl of each

biotinylated TCR monomer at 1 mg/ml was flowed over all flow cells, as was 25 µl of each TCR tetramer at

50 µg/ml. Negligible response was observed to non-cognate pMHCI for both TCR monomers and tetramers.

To facilitate visual comparison of monomer and tetramer binding events, the much larger monomer response

values were normalized to the peak values for the tetramers. Kinetic binding parameters for the tetramers

were estimated using BIAEvaluation™ software (see table). Monomer KDs were taken from experiments

with unbiotinylated A6 TCR (Supplementary Material Figure 1) or JM22 TCR (20). There are two apparent

off-rates for the TCR tetramers: (i) a minority fast off-rate thought to correspond to those tetramers binding

less than three antigens; and, (ii) a slow (true) off-rate for those tetramers likely binding three pMHCI

molecules. The latter off-rate is shown in the table. Some irreversible binding of biotinylated TCR

monomers is observed owing to the incomplete blocking of the streptavidin-coated chip surface with soluble

biotin.

Figure 2. TCR multimer staining of APCs. HLA A2+ B-cells were pulsed with exogenous peptides at the

indicated concentrations and stained with A6 or JM22 TCR multimers as described in the Experimental

Procedures.

Figure 3. TCR multimers exhibit exquisite staining specificity. HLA A2+ B cells were pulsed with a

variety of A2-restricted peptides at 200 µM or incubated with medium only (unpulsed cells) and stained with

the indicated TCR tetramers as described in the Experimental Procedures. Bars show the standard deviation

from the mean fluorescence intensity (MFI) values from 2 independent stainings.

Figure 4. Characterization of TCR binding by flow cytometry. LBL721.174 cells were incubated for 90

min with the HLA A2-HTLV-1 Tax (A) or HLA A2-Flu matrix (B) peptides at the indicated concentrations


27

or with medium only in presence of either 200 µg/ml BSA (empty bars) or 200 µg/ml β2m (filled bars) and

stained with the relevant TCR tetramer conjugated with extravidin-PE as described in the Experimental

Procedures. (C) PK B cells (HLA A2+) infected with recombinant vaccinia virus (see Experimental

Procedures) were stained 12 hr after infection as above.

Figure 5. TCR-mediated specific inhibition of CTL activation. (A) HLA A2+ B cells (PK) were pulsed

with the indicated concentrations of HLA A2-HTLV-1 Tax11-19 peptide or infected with vaccinia virus

expressing the full-length Tax protein and used as targets in a CTL activation assay (see Experimental

Procedures). Bars show the standard deviation from the mean of two replicate assays. Data shown are

representative of three experiments. (B) HLA A2+ B cells (PK) were pulsed with 1 µM HLA A2 HTLV-1

Tax peptide and either 0.1 µM (grey filling) or 10 µM (black filling) HLA A2-HIV Gag peptide, or pulsed

with 0.1 µM HTLV-1 Tax peptide (inset). Cells were then pre-incubated with 100 µg/ml A6 Tax TCR

tetramer or BSA and used in CTL activation assays with 003 HIV Gag CTLs or with E6 Tax CTLs, as

indicated. Bars show the standard deviation from the mean of three replicate assays. Background MIP1β

release by E6 CTL was 60 pg/ml. (C) HLA A2+ B cells (PK) were pulsed with 1 µM HLA A2-HTLV-1 Tax

peptide and incubated with 10 µg/ml A6 wt monomers/tetramers, A6c134 monomer or 1G4 TCR tetramer

prior to CTL activation assay as described in the Experimental Procedures. Data are expressed as percentage

of MIP1β release compared to an assay performed in absence of TCR (100%). Bars show the standard

deviation from the mean of three replicate assays and are representative of 2 independent experiments.

Background MIP1β release without addition of antigenic peptide was 91.8 pg/ml (or 14.3% of that with

added antigen). The A6c134 monomer was able to inhibit CTL activation even at 0.1 µg/ml (data not

shown).

Figure 6. A6 TCR tetramer staining of HLA A2+ B cells pulsed with cross-reactive peptides. PK B cells

were pulsed with indicated concentrations of HuD87-95 (A) and Tel1p (B) peptides in the presence of 200

µg/ml β2m for 90 min and stained as described in the Experimental Procedures. Staining of cells pulsed with

with 0.1 mM Tax11-19 peptide is shown for comparison.


28

Figure 7. Affinity of the A6 TCR for HLA A2-Tel1p and HLA A2-HuD87-95. Kinetic binding

experiments for A6 TCR flowed over immobilized A2-Tel1p (A), A2-HuD87-95 (C) and A2-Tax11-19 (E).

Equilibrium binding specific response is plotted against A6 TCR concentration and a non-linear fit of the

Langmuir binding isotherm is shown for A2-Tel1p (B), A2-HuD87-95 (D) and A2-Tax11-19 (F). A6 TCR was

flowed over the BIAcore chip at concentrations of 224 µM and two-fold dilutions thereof down to 1.75 µM.

Saturation binding with A2-Tax11-19 was reached at low concentration. Only data with lower concentrations

of TCR is shown in F.

Figure 8. Affinity of the A6c134 TCR for HLA A2-Tel1p and HLA A2-HuD87-95. Kinetic binding

experiments with A6c134 TCR flowed over immobilized ligands are shown. Overlay of the binding

responses for A6c134 flowed over A2-Tax11-19 (solid line), A2-HuD87-95 (dotted line) and control A2-

hTERT540 complexes (dashed line) (A). Kinetic binding experiments for A6c134 flowed over A2-Tel1p (B)

and A2-HuD87-95 (C). Equilibrium binding specific response is plotted against A6c134 TCR concentration

and a non-linear fit of the Langmuir binding isotherm is shown for A2-Tel1p (D) and A2-HuD87-95 (E).

A6c134 TCR was flowed over the BIAcore chip at concentrations of 48 µM and two-fold dilutions thereof,

down to 0.0937 µM.

Table I. Summary of binding data of A6 TCR with various ligands measured by SPR.

LIGAND

HLA A2-Tax

HLA A2-Tel1p

HLA A2-HuD

11-19

86-95

Bindingprediction*

28

28

19

SEQUENCE

LLFGYPVYV

MLWGYLQYV

LGYGFVNYI

Calculated KD in independentBIAcore experiments

Average KD ± SD

(µµµµM)

0.95 0.82 1.4

45.7 46.5 23.6

115.5 151.3 103.2 123.3±35.4

38.6±20.9

*Binding prediction of peptide for HLA A2 (see reference 14)

1.06±0.3

A B

Laugel et al. Figure 1

JM22 monomer JM22 tetramer A6 monomer A6 tetramer

k on (M-1 s-1) 3 × 104 ° - 7 × 104 $ -

k off (s-1) 0.12 ° 2.7 × 10–4 0.1 $ 7.5 × 10–5

Approx KD (M) 6.6 × 10–6 ° - 1.4 × 10–6 $ -

t1/2 (s) 5.7 ° 2580 6.9 $ 9240

° (20) $ (data from experiment shown in Supplementary Material Figure 1)

100

0100 101 102 103

no peptide no peptide10 µM 10 µM

100 µM 100 µM

Cel

l cou

nt

FL-2 fluorescence intensity

JM22 Flu A6 TaxA

Laugel et al Figure 2

B100

0100 101 102 103


unpulsed cells

HTLV-1 Tax LLFGYPVYV

EBV GLCTLVAML

HCV-22 GLQDCTMLV

HCV-33 ALYDVVTKL

HIV Gag SLYNTVATL

Flu matrix GILGFVFTC

Mean Fluorescence Intensity

A6 Tax TCR Flu TCR

0 20 40 60 80 20 40 60 800 100


5

10

15

20

25

0

10

20

30

40

200 µg/ml BSA200 µg/ml β2m

200 µg/ml BSA200 µg/ml β2m

Tax11-19 Flu matrix58-66

Mea

n Fl

uore

scen

ce I

nten

sity

irrelevantvaccinia

A B C

Tax vaccinia

Cou

nts

Fluorescence intensity

1 µMnone 10 µM 1 µMnone 10 µM 101100

[Peptide]


250

500

003 E6

Non

e

1G4

1mer

A6

1mer

s

A6

4mer

s

A6c

134

1mer

Rel

ativ

e M

IP1β

rele

ase

(%)

0

20

40

60

80

100

[MIP1β] (pg/ml)

10-5

10-6

10-7

10-8

tax-vaccinia

unpulsed cells

0 200 400 600

JM22 tetramer

A6 tetramer

BSA

[Pep

tid

e]

(M)

[MIP

1β]

(p

g/m

l)

A6 Tax TCR

BSA

BA C


No peptide

2.5mM

1mM

Tax 0.1mMTax 0.1mM

No peptide

1mM0.5mM

2.5mM

HuD87-95 Tel1p

A B

Cel

l cou

nt

FL-2 fluorescence intensity

B

0

1 0 0

2 0 0

3 0 0

KD= 38.6 µµµµM

0 1 0 0 2 0 0 3 0 0

0 1 0 2 0 3 0 4 0 5 0 6 0 7 00

5 0

1 0 0

1 5 0

KD= 0.95 µµµµM

F

A2-Tel1p

A

48028150

28200

28250

28300

28350

28400

28450

28500

280 300 320 340 360 380 400 420 440 460

Res

po

nse

Un

its C

28100

28200

28250

275 295 315 335 355 375 395 415 435 455 475

28150

28050

28000

27950

D

1 0 0 2 0 0 3 0 00

5 0

1 0 0

1 5 0

KD= 123.3 µµµµM

0

28400

28500

28600

28700

28800

28900

29000

29100

29200

320 340 360 380 400 420 440 460

E

Sp

ecif

ic R

esp

on

se

A2-HuD

A2-Tax

Time (s) TCR concentration (µµµµM)



Res

po

nse

Un

its

Time (s)

28300

28350

28400

28450

28500

28550

28600

560 580 600 620 640 660 680 700

A2-Tel1p

28050

28100

28150

28200

28250

28300

560 580 600 620 640 660 680 700

A2-HuD

0 10 20 30 40 50

0

10

20

30

40

50

60

0 10 20 30 40 50

0

50

100

150

200

250

Sp

ecif

ic R

esp

on

se

TCR concentration (µµµµM)

B C

D E

KD=46.5 µµµµM KD=21 µµµµM

0

500

1000

1500

2000

2500

0 200 400 600 800 1000 1200 1400

A

-20

0

20

40

60

80

100

120

150 200 250 300 350 400 450 500 550 600

Laugel et al. Supplementary Material Figure 1

Bin

din

g R

esp

on

se

Time (s)

Kinetic parameters of the A6/HLA-A2 interaction. Streptavidin was covalently linked to a BIAcore CM-5 chip by aminecoupling at 5,000 RU and biotinylated pMHC protein or biotinylated control protein were bound to sensor surface atapproximately 500 RU in each flow-cell. A6 TCR solutions at a maximum concentration of 1 µM and two-fold dilutionsthereof, down to 0.0625 µM, were flowed over the cells at a fast flow-rate (50 µl/min.). Binding response curves (plainlines) and corresponding fits (dashed lines) are shown as overlays. The calculated association (kon) and dissociation rateconstants (koff) corresponding to this figure are shown in the table of Figure 1.


-20

0

20

40

60

80

100

120

140

160

180

-500 -100 300 700 1100 1500 1900 2300 2700 3100 3500Time (s)

Bin

din

g R

esp

on

se

1 µM0.25 µM0.125 µM

Binding of A6 TCR tetramers to HLA-A2 Tax. Streptavidin was covalently linked to a BIAcore CM-5 chip byamine coupling at 5,000 RU and biotinylated pMHC protein or biotinylated control protein were bound to sensorsurface at approximately 500 RU in each flow-cell. Each A6 TCR tetramer solution at the indicated concentrationwas flowed over a separate flow-cell of the chip at the injection rate of 50 µl/min. The half-life of tetramerinteraction calculated for the linear part of the dissociation phase of each tetramer concentration differs by <3.5%.

No antigen

10 -6 M Tax peptide

10 -5 M Tax peptide

Cel

l cou

nt

FL-1 Mean Fluorescence Intensity


Staining of B-cells with microbeads-linked TCRs. Purified TCR at a concentration of 1mg/ml was labeled with Alexa Fluor 488by standard N-hydroxy-succinidimide linkage to -NH2 groups (Molecular Probes, Oregon) and purified according to the manufacturer’s directions. Labeled TCR was then mixed with microMACS streptavidin-conjugated magnetic microbeads(Miltenyi Biotech, Germany) at a ratio of 20 TCR molecules per streptavidin molecule for 15 min at room temperature. TCR-streptavidinbead multimers were then purified on microMACS columns and eluted in 150 µl PBS. For staining, T1 cells were suspended at 106 cells/ml in R10 containing varying concentration of Tax11-19 peptide, as indicated, for 90 minutes at 370C. Cells were washed in PBS, resuspendedin 50 µl aliquots at 3x106 cells/ml, mixed with 16 µl of TCR multimers, and incubated at room temperature for 2 hrs. Cells were then brought up to a total volume of 200 µl with PBS without washing and immediately analyzed by flow cytometry.

Time (s)

Res

pons

e (R

U)

-50

0

50

100

150

200

250

2000 3000 4000 5000 6000 7000 8000


Binding of A6c134 monomers to HLA A2-Tax. Streptavidin was linked to a Biacore™ CM-5 chip by amine coupling, biotin-tagged pMHCI was loaded onto each flow cell at a low density (500 RU), and data were collected at 25°C with a flow rate of 50 µl/min. 120 µl of TCR monomer (0.5 µM) was injected during the association phase.

100101102103104

100 101 102 103 104100101102103104

100 101 102 103

100101102103104

100 101 102 103

100101102103104

100 101 102 103


0.12%

0.27%

Anti-Lamp FITC

Ant

i-C

D8

APC

2.07%

HLA-A2 Tax11-19 Tetramer-PE

Ant

i-C

D8

APC

A

B C

ED

2.33%

TCR mediated inhibition of lysosomal membrane associated proteins (LAMPs) up-regulation. LAMPs are proteins present in the membrane of cytotoxic granules and are displayed on the cell-surface following CTL activation, when the granules fuse with the cytoplasmic membrane (Betts et al. 2003. Sensitive and viable identification of antigen-specific CD8+ T cells by a flow cytometric assay for degranulation. J Immunol Methods 281:65-78). Soluble A6c134 TCR inhibited antigen-induced LAMP upregulation by HLA-A2 Tax11-19 specific CTLs (E). Approximately 25,000 cells of the D1 line containing about 2% of HLA-A2 Tax11-19 specific cells (A) were incubated either without

peptide (B) or in presence of 10-6 M (C, D and E) of Tax11-19; and with BSA (B and C), with the irrelevant 1G4 TCR (D) or with A6c134 (E) at a final concentration of 10 µg/ml. The assay was carried out over 1 hour in presence of the protein transport inhibitor monensin (Golgistop, BD Pharmingen) and anti-LAMPs FITC (BD Pharmingen) antibody. Cells were then washed with FACS buffer and stained with anti-CD8 APC antibody prior to the FACS analysis. Percentage of total cells in gate is indicated in B-E.

2.13%

100101102103104

100 101 102 103

104

104

104

Design of Soluble Recombinant T Cell Receptors for Antigen Targeting and T Cell Inhibition

Documents

Transcript of Design of Soluble Recombinant T Cell Receptors for Antigen Targeting and T Cell Inhibition