Skip to main content
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 30.
Published in final edited form as: Nat Immunol. 2018 Apr 16;19(5):487–496. doi: 10.1038/s41590-018-0092-4

Monomeric TCRs Drive T-Cell Antigen Recognition

Mario Brameshuber 1,*, Florian Kellner 2,#, Benedikt K Rossboth 1,#, Haisen Ta 3, Kevin Alge 2, Eva Sevcsik 1, Janett Göhring 1,2, Markus Axmann 4, Florian Baumgart 1, Nicholas R J Gascoigne 5, Simon J Davis 6, Hannes Stockinger 2, Gerhard J Schütz 1, Johannes B Huppa 2,*
PMCID: PMC7612939  EMSID: EMS146254  PMID: 29662172

Abstract

T-cell antigen recognition requires T-cell antigen receptors (TCRs) binding to MHC-embedded antigenic peptides (pMHCs) within the contact region of a T-cell with its conjugated antigen-presenting cell. Despite micromolar TCR:pMHC affinities, T-cells respond to even a single antigenic pMHC, and higher order TCR-structures have been postulated to maintain high antigen sensitivity and trigger TCR-proximal signaling. We interrogated the stoichiometry of TCRs and their associated CD3 signaling chains on the surface of living primary T-cells with the use of (i) single molecule brightness and (ii) single molecule coincidence analysis, (iii) photon-antibunching based fluorescence correlation spectroscopy and (iv) Förster Resonance Energy Transfer measurements. While all four approaches unambiguously confirmed the accepted subunit ratio within TCR/CD3 complexes, we found exclusively monomeric TCR/CD3 complexes driving the recognition of antigenic pMHCs. Our findings underscore the exceptional capacity of single TCR/CD3 complexes to elicit robust intracellular signaling, which may prove critical for optimizing T-cell-based immunotherapies.

Keywords: TCR/CD3 complex, T-cell antigen recognition, immunological synapse, single molecule biophotonics, Förster Resonance Energy Transfer (FRET)

Introduction

T-cell antigen recognition is paramount to adaptive immunity and invariably tied to the formation of an immunological synapse, the transient area of contact between the T-cell and the antigen-presenting cell (APC) 1. It is driven by T-cell antigen receptors (TCRs) on the T-cell binding to antigenic peptide fragments presented by APCs in the context of MHC molecules. When measured in vitro TCR interactions with nominal peptide/MHC complexes (pMHCs) are typically of moderate affinities 2. Nonetheless T-cells manage to detect the presence of even a single antigenic pMHC among thousands of non-stimulatory but structurally similar pMHCs 3, 4. The molecular and cell biological mechanisms underlying this exceptional sensitivity are not well understood, and it is not clear how extracellular TCR-engagement triggers the first intracellular activation events. This persistent gap in understanding still exists because comprehensive approaches are complicated by the transient nature of the recognition events, which take place under the highly dynamic geometric constraints present within the nascent immunological synapse. Given these circumstances, biochemical methodologies lose much of their resolving power since they require the disruption of the membrane environment of at least one of the synaptic binding partners. Minimally invasive imaging approaches with molecular resolution are hence preferable especially when complemented with biochemical and structural insights.

A logical starting point to address the underpinnings of the extraordinary T-cell antigen sensitivity is to determine the composition dynamics of the TCR prior to and during ligand binding 5, 6. Based on genetic, biochemical and cell biological evidence it is widely accepted that the clonotypic and disulfide-linked αβTCR heterodimer, the true ligand binding unit, is non-covalently associated with the signaling chains of the CD3 complex, namely CD3γ, δ, ε and the disulfide-linked ζ2 homodimer. Co-and posttranslational ER quality control mechanisms ensure the exclusive egress of properly assembled TCR/CD3 (TCRαβ/CD3γδε2 ζ2) complexes from the ER to the plasma membrane, where they unfold their physiological function 7, 8, 9, 10.

The existence of multimerized TCR/CD3 complexes was first proposed based on density gradient centrifugation and immunoprecipitation experiments involving T-cell lysates 11. Later analyses based on blue native gel electrophoresis of T-cell lysates and electron microscopy of immunogold labeled T-cell surfaces invoked the presence of higher order TCR/CD3 structures as a key factor in promoting and maintaining T-cell antigen sensitivity 12, 13. However, the use of detergent for membrane solubilization and also cell fixation required for EM specimen preparation complicates data interpretation, and derived conclusions were in fact challenged by single molecule spectroscopy on live T-cell hybridoma 14. Furthermore, the clustered appearance of anti-TCR immunogold particles may reflect enrichment of mobile and monomeric TCRs in specialized plasma membrane compartments 15, 16 rather than receptor oligomerization.

Ligand-induced dimerization or oligomerization has long been considered the decisive trigger for TCR/CD3 complexes, in analogy to the initial steps driving receptor tyrosine kinase (RTK) - mediated signaling, even though TCR/CD3 complexes lack, unlike RTKs, kinase activity and require the presence of the TCR-proximal p56lck kinase for activation. This concept has been mostly fueled by the observation that anti-TCR or CD3 antibodies do not stimulate as monovalent Fabs or single-chain variable fragments (scFVs), but do so as bivalent intact antibodies, especially when further aggregated 17. Furthermore, monomeric soluble pMHCs do not activate T-cells while soluble pMHC-dimers do 18, 19. Also, the fact that many MHC class II complexes form dimers when crystalized for structure determination 20, 21, 22 has been regarded as an inherent propensity of MHC class II molecules to form higher order structures. Furthermore, elastic light scattering studies conducted on mixtures of soluble TCRs and pMHC suggested multimerization of TCRs in a ligand-dependent fashion 23. However, some of these findings could not be reproduced for all MHC class II proteins studied 24 and are hence the cause of much controversy. Of note, to this date, ligand-induced multimerization of TCRs has not been directly shown within the confines of the immunological synapse.

As a consequence, many questions regarding the existence of higher order TCR-structures, their exact nature, the possible regulation and mechanism of their formation as well as their impact on pMHC binding and downstream signaling have so far remained open. Here, we have analyzed in considerable detail the stoichiometry of TCRs within the immunological synapse of living T-cells with the use of (i) single molecule brightness and (ii) single molecule coincidence analysis, (iii) a novel method combining photon arrival time analysis with fluorescence correlation spectroscopy (PA/FCS) and (iv) Förster Resonance Energy Transfer (FRET)-based imaging. All four experimental approaches allowed us to further substantiate the widely-accepted subunit stoichiometry within a single TCR/CD3 complex, serving as benchmark for further analyses, yet revealed TCR monomers as the predominant receptor species driving ligand recognition. We hence conclude that TCRs act as single receptor entities in an autonomous fashion with important mechanistic implications for TCR-proximal signal amplification, the maintenance of T-cell antigen sensitivity and therapeutic T-cell engineering.

Results

Single molecule-based fluorescence brightness analysis of TCR/CD3 complexes

The TCR surface density of 75 ± 24 molecules per μm2 on the plasma membrane of living T-cells (figure 1b) is too high to allow for direct, diffraction-limited imaging of individual TCRs or potential TCR oligomers (figure 1c, left panel). To reveal the TCR stoichiometry on living cells we opted to employ a single molecule-based fluorescence brightness analysis approach termed TOCCSL (Thinning Out Clusters while Conserving Stoichiometry of Labeling), as this methodology enables assessing the degree of protein oligomerization below the diffraction limit 25, 26, 27. Figure 1c outlines the principle of a typical TOCCSL experiment. T-cell blasts from 5c.c7 TCR transgenic mice were decorated with site-specifically Alexa Fluor 647-conjugated scFVs derived from the monoclonal TCRβ-reactive antibody H57-597 (AF647-H57-scFV) 28. T-cells were allowed to adhere on a planar glass-supported lipid bilayer (SLB) functionalized for cell adhesion with laterally mobile ICAM-I before imaging with an epifluorescence microscope in total internal reflection (TIR) mode (figure 1a). At room temperature, the half-life of TCRβ:AF647-H57-scFV binding is 44 minutes (supplementary figure 1a), i.e. substantially longer than the time window of 5-15 minutes allowed for performing TOCCSL experiments. By applying a high-powered laser pulse, we quantitatively ablated the fluorescence of a spatially defined area and recorded a control image 1 millisecond after bleaching to verify complete photobleaching (figure 1c, center panel). After a recovery phase lasting 3-10 seconds we acquired the so-termed TOCCSL image to record isolated diffraction-limited spots originating from laterally mobile TCR entities, which had entered from the masked cell surface and had therefore not undergone any bleaching (figure 1c, right panel). The corresponding laser illumination sequence for a representative TOCCSL experiment is shown at the top of figure 1c. The brightness distribution of diffraction-limited spots detected in TOCCSL images from numerous TOCCSL experiments (n > 120) is shown as a probability density function (pdf) in figure 1d. In order to quantify the number of AF647-H57-scFV -labeled TCRβ subunits per diffraction limited spot, we compared the measured distribution of fluorescence brightness values ρ(B) with the brightness distribution of single AF647-H57-scFVs, ρ1. The latter was obtained from the same experiment after prolonged and repeated photo-bleaching, or alternatively from recording T-cells with 150-fold substoichiometric TCR-labeling (see Methods, Microscopy - TOCCSL). A linear combination of n-mer contributions, ρn, was derived from ρ1 and used to fit the distribution ρ(B) (see Methods, TOCCSL and mobility analysis, equation 1). The fit yielded predominantly TCRβ monomers. A small fraction of 3 ± 4% (figure 1d, left) potentially reflects dimeric TCRβ entities, however, this value is close to the detection limit of the TOCCSL method 27.

Figure 1. Single-color TOCCSL provides no evidence for the existence of laterally mobile higher order TCR-structures but confirms the presence of two CD3ε subunits per TCR/CD3 complex.

Figure 1

(a) Monovalent scFVs and Fabs derived from indicated TCRβ- and CD3ε-reactive mAbs serve as probes to score for epitope stoichiometry within laterally mobile TCR/CD3 entities as imaged in TIR illumination on the surface of living primary T-cells in contact with non-stimulatory SLBs.

(b) TCR surface densities were measured for T-cells (n = 30) in contact for 3-9 minutes with SLBs functionalized with ICAM-1.

(c) The laser intensity/pulse duration and image acquisition protocol applied in TOCCSL experiments is indicated. T-cells quantitatively decorated with monovalent antibody fragments were placed onto SLBs harboring ICAM-1. A defined synaptic T-cell region was illuminated in TIR mode as controlled by an aperture (dashed area) implemented within the excitation pathway (i). The average TCR surface density of 75 ± 24 μm-2 (figure 1b) was too high to allow for the resolution of single or potentially clustered TCR/CD3-associated fluorescent probes, yielding many TCR/CD3 entities per camera pixel (see scheme below). As demonstrated by the image acquired 1 ms after the 400-ms bleach pulse (ii) fluorescent probes were quantitatively ablated in the synaptic area of interest. Following a recovery time of 3-10 seconds, fluorescence-labeled TCR/CD3 entities having diffused from the masked periphery into the field of view became recordable in the TOCCSL image as diffraction limited spots (iii). For better visualization, the contrast in (ii) and (iii) was increased five-fold. The scheme illustrates how a single fluorescent entity composed of either a single or two TCR/CD3 complexes affects TOCCSL-based readout. Scale bar: 2 μm.

(d) The brightness distribution ρn of single fluorescent entities corresponding to TCRβ-reactive AF647-H57-scFVs and to CD3ε-reactive AF647-KT3-scFVs detected in the TOCCSL image (c, iii) is plotted as a probability density function. Recorded intensities (black line) were fitted as described in the Methods section (red line) assuming monomer (solid blue line) and dimer (dashed blue line) contributions. The fit yielded a dimer fraction of 3 ± 4% (n = 30 cells) for the TCRβ-reactive AF647-H57-scFV and 74 ± 4% (n = 42 cells) for the CD3ε-reactive AF647-KT3-scFV. (e) The fraction of mobile TCR molecules was determined by utilizing FRAP-experiments (n = 15 cells), and confirmed by step-length distribution analysis of single molecule tracking data (n = 21 cells). Error bars present standard error of the mean.

In contrast, decorating T-cells with site-specifically and stoichiometrically AF647-labeled scFVs derived from the CD3ε -reactive KT3 antibody (AF647-KT3-scFV) gave rise to a high proportion of double single molecule brightness events but no triple or quadruple single molecule brightness events (figure 1d, right). These observations are consistent with the existence of two CD3ε copies per TCR/CD3 complex.

To determine the fraction of laterally mobile TCR species, which are accessible to TOCCSL analysis, we performed ensemble Fluorescence Recovery after Photobleaching (FRAP) and single molecule tracking (SMT) experiments, which yielded 73 ± 6% (FRAP) and 64 ± 3% (SMT) mobile TCRs (figure 1e and supplementary figure 1b, c).

Taken together, laterally mobile TCR/CD3 entities present on non-activated T-cells predominantly accommodate one TCR-reactive H57-scFV and two CD3ε-reactive KT3-scFVs. These findings are in good agreement with a TCRαβ/CD3γδε2ζ2 stoichiometry previously described for monovalent TCR/CD3 complexes 7.

Co-localization-based single molecule analysis of TCR/CD3 stoichiometry

Since stochastic noise associated with single molecule photon emission and also the limited accuracy of the fitted single molecule brightness signal complicate TOCCSL-based detection of rare higher order TCR-structures, we extended the acquisition protocol by introducing a second spectrally resolvable fluorescence channel. This dual color TOCCSL methodology allows for unambiguous identification of interaction probabilities below 0.1% 29 through simple co-localization of differentially labeled single TCR/CD3 complexes. Experiments were carried out and analyzed as schematically depicted in figure 2. In brief, we decorated surface-exposed TCRs in equal proportions with AF488-H57-scFVs and AF647-H57-scFVs. Following rapid and quantitative fluorescence ablation of a synaptic region of interest and partial fluorescence recovery, TCR oligomerization was quantified by co-localizing single molecule fluorescence events originating from differentially labeled TCR species. For this the fluorescence emission was split into two channels for individual detection of single AF488 and AF647 fluorescence events. We either excited both fluorophores during the recovery phase simultaneously (figure 2a) or separated the recording of the AF488 and AF647 channels by one millisecond (supplementary figure 2d) to enable the detection of FRET events between differentially labeled H57-scFVs as an additional readout for TCR oligomerization. With unhindered fluorophore rotation, FRET yields are inversely proportional to the sixth power of the distance between the corresponding fluorescent dyes. Hence FRET can be employed to score for spatial proximity in the nanometer range. For the FRET pair AF488:AF647 the calculated Forster radius R0, i.e. the inter-dye distance giving rise to half-maximal energy transfer, amounts to 5.6 nanometers (supplementary table 1), roughly covering the length of the TCR.

Figure 2. Examining the oligomerization of laterally mobile TCRs via dual color co-localization-based TOCCSL.

Figure 2

(a) As schematically illustrated, dual color TOCCSL analysis relies on the degree of co-localization of differentially labeled monovalent antibody fragments targeting identical or distinct, non-overlapping epitopes on the TCR/CD3 complex. Sample illumination was performed following the indicated intensity/pulse duration and image acquisition protocol and with the use of a blue (488nm) and a red (647nm) laser giving rise to a pre-bleach (i), bleach control (ii) and TOCCSL (iii) image collected simultaneously or in short succession on the camera CCD and spectrally resolved by an emission beam splitter. Photo-bleaching was always performed simultaneously for both colors. TCR oligomerization events are determined by co-localizing the two differentially labeled probe-species without the need for brightness analysis. Dual color TOCCSL images represent T-cells labeled with AF488-H57-scFVs and AF647-H57-scFVs targeting TCRβ (upper panel) and CD3ε-reactive AF488-KT3-scFVs and AF647-KT3-scFVs (lower panel) recorded 10 s after the bleach pulse. Positions of diffraction-limited spots were determined for both color channels and are displayed as white circles in the red AF647 channel (left) and the green AF488 channel (right) representing H57-scFV or KT3-scFV respectively. Corrected positions of one probe projected into the other color channel are shown as green and red crosses, respectively. If two fluorescence events registered in different color channels were found within a radius of 1 pixel (corresponding to 200nm), they were counted as co-localizing entities (yellow arrow).

(b) The degree of co-localization was strictly dependent on fluorescent probes applied and is consistent with the subunit stoichiometry of single but not oligomerized TCR/CD3 complexes. Scale bars: 4 μm.

Positions of individual TCRs were determined for both spectral channels with an accuracy of about 20 nanometers, corrected for chromatic aberration and overlaid to probe for co-localization within a 200-nanometer radius (supplementary figure 2a-c). After correcting for false-positive events (see Methods for details), the fraction of co-localized events was determined (figure 2b and table 1). Where applicable, FRET yields were also quantified for all double positive single molecule events. As summarized in figure 2b and table 1, the degree of AF488/AF647-H57-scFv probe pairing at 24°C and 37°C, was found negligible.

Table 1. Results of two-color TOCCSL co-localization experiments.

Co-localization events Single molecule FRET events
Probe H57-scFv H57-scFv KJ25-Fab H57-
scFV/KJ25-
Fab
KT3-scFv H57-scFv KJ25-Fab H57-
scFV/KJ25-
Fab
KT3-scFv
Epitope β, H57 β, H57 β, KJ25 β, H57& β,
KJ25
ε, KT3 β, H57 β, KJ25 β, H57& β,
KJ25
ε, KT3
Labelling color H57 H57 KJ25 KJ25 KT3 H57 H57 KJ25 KT3
H57 H57 KJ25 H57 KT3 H57 H57 H57 KT3
Temperature 24 °C 37 °C 24 °C 24 °C 24°C 24 °C 24°C 24 °C 24°C
Number of cells 47 48 30 20 31 21 30 20 13
Number of AF488 signals 798 558 998 339 966 1882 998 384 97
Number of AF647 signals 1321 762 449 209 305 2055 449 381 122
Number of co-
localization/FRET events
10 9 20 49 38 9 1 36 3
Number of false positive events 6 6 19 1 0 6 1 3 0
Fraction of co-
localization/FRET events
0.4 % 0.5 % 0.1 % 21 % 10 % 0.15 % 0 % 8.7 % 2.5 %

To maximize the likelihood of revealing TCR oligomers, which might be cryptic to H57-scFV-based detection, we repeated dual color TOCCSL experiments using dye-conjugated KJ25-Fabs targeting the variable region of the TCRβ chain. The epitope of KJ25 overlaps with the TCRβ region involved in antigen binding and should hence be accessible on all TCRβ subunits capable of pMHC binding in putative oligomeric structures 30. Of note, we found TCR surface densities measured with KJ25-Fabs and H57-scFVs comparable (supplementary figure 2e), indicating similar if not identical accessibility of both antibody fragments to surface-resident TCRs. Again, as was observed with the use of the H57-scFV, we only detected negligible co-localization and single-molecule FRET events with KJ25-Fabs (figure 2b and table 1). However, when applied in combination as AF488-H57-scFV and AF647-KJ25-Fabs (or vice versa), antibody fragments gave rise to a high fraction of dual color signals, which likely represented one TCRβ subunit carrying two labels in close proximity as substantiated by their high FRET values (figure 2b, table 1 and supplementary figure 2f).

In line with the results of the one color TOCCSL experiments, the use of the CD3ε -reactive AF488-/AF647-KT3-scFVs yielded a large proportion of double positive events in dual color TOCCSL experiments (figure 2b and table 1) reflective of two CD3ε copies per TCR/CD3 complex.

In summary, dual color TOCCSL results from all label combinations applied were consistent with a subunit stoichiometry of monovalent but not of higher order TCR/CD3 entities. Moreover, the close label apposition within (i) AF488-KT3-scFVs/AF647-H57-scFv double positive TOCCSL entities as well as (ii) AF488-KJ25-Fab/AF647-H57-scFv double positive TOCCSL entities invariably gave rise to quantifiable single molecule FRET signals (supplementary figure 2f and table 1). Together, these findings underscore the robustness of the TOCCSL approach and imply that laterally mobile TCR/CD3 complexes do not form higher order structures with a lifetime larger than 3-10 seconds, i.e. the duration of fluorescence recovery in our TOCCSL experiments.

Combining photon-antibunching with fluorescence correlation spectroscopy (PA/FCS) to enhance temporal resolution

Both TOCCSL methods involve a step of fluorescence recovery lasting 3-10 seconds. If transient protein complexes dissociate during this time, they elude detection in the subsequent TOCCSL image. To investigate the possibility that TCR-CD3 complexes form transient higher order structures with a half-life below the duration of fluorescence recovery, we probed the number of dye molecules associated with a given TCR/CD3 complex via analysis of coinciding photon arrival times 31, 32, 33 and fluorescence correlation spectroscopy (FCS). This approach relies on the fact that a single fluorophore cannot emit more than one photon when excited with a light pulse significantly shorter than its characteristic fluorescence lifetime. As a consequence, coincident photons resulting from such excitation pulses must originate from different independent emitters, a property which can be exploited to quantify the number of fluorophores in a given observation area within the plasma membrane 33. Unrelated emitters can be discriminated from those attached to the same complex because they diffuse through the illuminated spot in an uncorrelated fashion. In contrast, fluorophores attached to the same entity move in a correlated manner (figure 3a, right panel). FCS experiments yielded a coefficient for lateral TCR diffusion D of 0.034 ± 0.014 μm2s-1 as determined from the normalized auto correlation function (ACF) (figure 3b, left panel), which is in good agreement with the value measured through single molecule tracking experiments (with D = 0.037 ± 0.002 μm2/s, supplementary figure 1b). Considering a confocal spot area of 0.08 μm2 the degree to which fluorophores undergo uncorrelated diffusion can be readily quantified within a single experiment for lag times equal to or larger than 80 milliseconds. This in turn allows determining the average number of transient higher order TCR/CD3 entities with a life-time between 100 and 300 milliseconds as quantified by the average FCS transition time 32.

Figure 3. PA/FCS measurements exclude the existence of short-lived higher order TCR structures.

Figure 3

(a) Scheme illustrating how T-cells quantitatively decorated with AS635P-H57-scFV and in contact with a lipid bilayer containing ICAM-1 were subjected to fluorescence correlation spectroscopy experiments combined with photon-antibunching measurements. A pulsed laser beam with a temporal width much shorter than the fluorescence lifetime was focused as a diffraction limited spot onto the T-cell plasma membrane touching the SLB as a means to probe the number of molecules moving in a correlated manner. Coincident photon pairs originating from TCR/CD3 complexes moving in an uncorrelated fashion are time lag time-independent, whereas coincident photon pairs resulting from fluorophores diffusing in a correlated manner and corresponding to higher order TCR-structures are lag-time dependent. The autocorrelation of coincident photon pairs for two different lag-times (here: from 0 ns to 2 s) hence reveals to the oligomeric state of the diffusing molecule.

(b, left panel) Shown are the measured normalized autocorrelation functions (ACFs) of diffusing, T-cell associated AS635P-H57-scFV (upper graph), AS635P-KJ25-Fab (center graph) and AS635P-KT3-scFV (bottom graph). The measured ACF (blue) was fitted with a 2D diffusion function with two components and one triplet state (red dashed line). The average diffusion yielded 0.034 ± 0.014μm2/s (AS635P-H57-scFV, n = 7 cells).

(b, right panel). Fitting respective ACFs at 0 s (blue), the 1st period (100 ns for H57 and 400 ns for KJ25 and KT3) (green) and 2 s (red) lag time yielded indicated values for the number of emitters per freely diffusing TCR/CD3 complex.

(c) Comparison of the average number of emitters for the three fluorescent probes diffusing either in solution (H57-scFV: 1.17 ± 0.01 with n = 11 measurements; KJ25-Fab: 1.26 ± 0.01 with n = 12 measurements; KT3-scFV: 1.14 ± 0.01 with n = 11 measurements) or when bound to plasma membrane-resident TCR/CD3 complexes (H57-scFV: 1.09 ± 0.08 with n = 7 cells; KJ25-Fab: 1.14 ± 0.08 with n = 7 cells; KT3-scFV: 1.86 ± 0.22 with n = 8 cells). Error bars represent the standard error of the mean.

For PA/FCS experiments (figure 3a-c) we decorated T-cells quantitatively with either Abberior Star 635P (AS635P) -conjugated H57-scFv (figure 3b, upper row), AS635P-KJ25-Fab (figure 3b, center row) to target TCRβ or with the CD3ε-reactive AS635P-KT3-scFV (figure 3b, bottom row) serving as positive control for correlated diffusion behavior of dye-pairs. Measurements were conducted with T-cells adhering to non-stimulatory SLBs featuring ICAM-1 to promote cell adhesion (figure 3a, left panel). For cells decorated with the TCRβ-reactive antibodies, coincident photon pairs with zero lag time gave rise to almost the same autocorrelation amplitude as photon pairs with two seconds lag time (blue curve and red curve, figure 3 upper and center row, right panels). We can therefore conclude that coinciding photon pairs were emitted from fluorophores attached to TCRs diffusing independently with respect to one another. Fitting the autocorrelation curves with a theoretical model (see Methods, Data Analysis PA/FCS experiments) confirmed an average number of emitters of either 1.09 ± 0.08 (AS635P-H57-scFv) or 1.14 ± 0.08 (AS635P-KJ25-Fab) (figure 3c) per freely diffusing TCR complex. As monomer reference, PA/FCS-experiments were carried out with soluble AS635P-H57-scFvs and AS635P-KJ25-Fabs (supplementary figures 3a and b) diffusing freely in imaging buffer. These studies yielded values of 1.17 ± 0.01 emitters per AS635P-H57-scFv and 1.26 ± 0.01 emitters per AS635P-KJ25-Fab (figure 3c), attributable to a small fraction of antibody fragments carrying more than one AS635P-dye.

In stark contrast, a large fraction of T-cell bound AS635P- KT3-scFV showed correlated diffusion behavior with 1.86 ± 0.22 emitters (figure 3b, bottom row, right panel), as would be expected given the presence of two CD3ε subunits per TCR/CD3 complex. Control experiments carried out with this probe in solution yielded an average number of emitters of 1.14 ± 0.01 (figure 3C and supplementary figure 3c), which is similar to the number of emitters detected in AS635P-H57-scFVs.

In line with the results of the single and dual color TOCCSL experiments, the PA/FCS approach did not reveal the existence of transient higher order TCR/CD3 structures, while it was sensitive enough to confirm the presence of two yet not more CD3ε subunits per TCR/CD3 entity. We therefore conclude that mobile TCR/CD3 complexes are exclusively monomeric in nature.

Probing TCR stoichiometry via Forster Resonance Energy Transfer (FRET)

Both TOCCSL methodologies and PA/FCS do not account for immobile entities as the rationale of all three methodologies rests on the lateral mobility of the structures of interest. To address TCR homo-association in a fashion that is independent of TCR mobility, we measured FRET between antibody fragment-labeled TCRs. To this end 5c.c7 TCR transgenic T-cell blasts were decorated with equimolar amounts of H57-scFVs site-specifically conjugated to either Alexa Fluor 647 or Alexa Fluor 555 and placed onto lipid bilayers harboring ICAM-1, B7-1 and optionally I-Ek/MCC, the nominal 5c.c7 TCR ligand for T-cell stimulation, (figure 4a). H57-scFV-binding to TCR does not interfere with antigen recognition 28, 34 rendering this antibody fragment suitable to probe both free and ligand-engaged TCRs. Positioning the fluorophores on the H57-scFV probes at a site close to the antibody´s paratope 28 (supplementary figure 6) ensured FRET yields reflecting most closely distances between TCRs. After cells had fully adhered, we determined the FRET efficiency by FRET donor recovery after FRET acceptor photobleaching in TIR mode.

Figure 4. Assessing the molecular proximity between individual TCRs by FRET donor recovery after FRET acceptor photo bleaching.

Figure 4

(a) Experimental system employed to assess synaptic epitope proximities and rationale underlying the FRET-based approach: 5c.c7 TCR-transgenic T-cells quantitatively decorated with a mixture of AF555-H57-scFVs and AF647-H57-scFVs are placed onto SLBs featuring ICAM-1, B7-1 and optionally for stimulation SLBs featuring I-Ek/MCC. FRET efficiencies between TCRs are determined in TIR mode within T-cell-SLB interfaces through FRET donor recovery after FRET acceptor photobleaching.

(b) Shown are FRET donor (green), as well as FRET donor intensities in false-color representation before and after FRET acceptor ablation, which were then used to assay inter-epitope distances between and within TCR/CD3 complexes as schematically depicted in (a). For FRET quantification (see Methods) brightness values were averaged for the entire T-cell contact area (white) or microclusters (green dashed line). Crosslinking of TCR-bound biotinylated H57-scFV through bivalent streptavidin (SA) or dimerization of H57-scFV decorated TCRs via intact KJ25 mAb led to a large increase in FRET and served as positive controls (c, d). Scale bars: 3 μm.

(c) Synaptic FRET efficiencies measured under non-activating (n = 20 cells) and activating (n = 21 cells) conditions. Control experiments (Ctrl) without FRET acceptor bleach pulses (control, n = 21 cells) accounted for photobleaching of the FRET donor. FRET yields of divalent streptavidin (SA)-crosslinked TCR-associated H57-scFV probes (see also supplementary figure 4d) served as positive control (1nM: n =10 cells; 10nM: n = 11 cells).

(d) FRET efficiencies measured before and after addition of the KJ25 antibody served as another positive control. T-cells labeled with AF568-H57-scFV and AF647-H57-scFV (ratio = 1:3) were either placed directly onto SLBs or first exposed to intact KJ25 mAb for TCR dimerization and then placed onto SLBs featuring ICAM-1 and B7-1. FRET was measured via FRET donor recovery after FRET acceptor photobleaching for entire synapses (control: n = 15 cells, 0 ng/μl: n = 12 cells, 10 ng/μl: n= 11 cells) or within antibody induced TCR microclusters (10 ng/μl: n= 29 microclusters).

(e) FRET efficiencies measured under stimulatory conditions in the presence of ICAM-1, B7-1 and I-Ek/MCC within cSMACs and TCR microclusters at a FRET donor to FRET acceptor ratio of 1:1 (n = 10 cells / 36 TCR microclusters) and 1:3 (n = 13 cells / 24 TCR microclusters).

(f) Degree of enrichment of TCRs within TCR microclusters compared to average synaptic TCR density (n = 39 cells).

(g) Targeting an alternative TCRβ epitope involved in pMHC binding with a mixture of AF488-KJ25-Fabs and AF647-KJ25- Fabs in different ratios yielded no detectable FRET (black circles, 1:1 n = 21 cells, 1:3 n = 19 cells). In contrast, simultaneous application of H57-scFVs and KJ25-Fabs, i.e. two probes targeting two separate epitopes on the same TCRβ protein, yielded a high FRET yield (red (n = 21 cells) and blue circles (n = 17 cells)).

(h) Substantial FRET yields were determined when labeling CD3ε with AF568-KT3-scFV and AF647-KT3-scFV at a ratio of 1:3 (n = 15 cells) or targeting TCRβ and CD3ε simultaneously with AF568-H57-scFV and AF647-KT3-scFV (n = 21), confirming close proximity of the respective subunits. Error bars represent standard error of the mean.

Placing T-cells onto non-stimulatory SLBs featuring ICAM-1 and B7-1 did not give rise to measurable FRET values (figure 4b, first row, and figure 4c) above those expected to result from random molecular encounters (simulated using a Monte Carlo approach, see Methods section and supplementary figure 4a). No FRET was detectable in the presence of nominal antigenic pMHCs (figure 4b, second row, and figure 4c) – neither within central supramolecular activation clusters (cSMACs) of mature synapses, nor within emerging TCR microclusters of nascent immunological synapses (figure 4e), which were ~2.5-fold enriched in TCRs (figure 4f). We then enhanced the sensitivity of the FRET-based assay by increasing the FRET donor to FRET acceptor ratio from 1 to 3, but did not observe FRET, both in the absence (supplementary figure 4b) and presence (figure 4e) of activating I-Ek/MCC.

We also switched the FRET donor from Alexa Fluor 555 to Alexa Fluor 568 as a means to increase the Forster radius of the FRET system from 5.1 to 8.3 nanometers, yet without detecting FRET above values expected from random molecular encounters (supplementary figure 4a, c). To ensure that we had not missed TCR multimerization events in the course of initial TCR-pMHC engagement, we focused our analysis on initial T-cell contacts made with SLBs, which had been functionalized with ICAM-1, B7-1 and high densities of agonist pMHCs (about 500 μm-2). While such contacts showed vigorous TCR-proximal signaling as verified via simultaneously recorded synaptic recruitment of retrovirally transduced ZAP70-GFP, they did not give rise to measurable FRET between TCR-bound H57-scFVs (supplementary figure 4e and f). Only after crosslinking site-specifically biotinylated TCR-associated H57-scFV-FRET probes via divalent streptavidin (figure 4b, third row, figure 4c and supplementary figure 4d) or after dimerizing H57-scFV-decorated TCRs via the Vβ3 (i.e. 5c.c7 TCR) -reactive mAb KJ25 (figure 4b, fourth row, figure 4d), we recorded FRET efficiencies of about 9±1% and 17±1%, respectively, testifying to the overall robustness of the FRET-based assay.

Of note, we measured substantial FRET yields when employing probe combinations, which had given rise to double positive dual color TOCCSL entities in previous experiments. These pairings included (i) AF555-H57-scFV/AF647-KJ25-Fab (FRET yield = 18±3%) reporting the close distance of two non-overlapping epitopes within TCRβ (figure 4b, fifth row and figure 4g) and (ii) AF568-KT3-scFV/ AF647-KT3-scFV (FRET yield = 18±0.5%) visualizing the two juxtaposing copies of CD3ε within individual TCR-CD3 complexes (figure 4b, sixth row and figure 4h). Synaptic FRET yields between TCR/CD3-bound AF568-H57-scFV and AF647-KT3-scFV amounted to even 25±2% (figure 4b, seventh row, and figure 4h), reflecting the presence of two FRET acceptor dyes at close distance to one FRET donor fluorophore within the complex. In contrast, we did not observe any significant FRET signals when decorating TCRs with the AF555-KJ25-Fab/AF647-KJ25-Fab combination targeting 5c.c7 TCRβ at a site involved in pMHC binding (figure 4g). This is consistent with both (i) the absence of double positive KJ25-Fabs signals in dual color TOCCSL experiments (figure 2b) and (ii) uncorrelated diffusion of KJ25-Fabs deduced by PA/FCS (figure 3c).

In summary, while the FRET-readout faithfully testified to the close proximity between CD3ε and TCRβ and also between the two CD3ε subunits within individual TCR/CD3 complexes, it did not reveal any FRET-sensitive proximity between individual TCRs. Combined our measurements imply that at least the large majority if not all of immobile or ligand-engaged TCRs are monomeric in nature.

FRET-based analysis of TCR-engaged pMHCs

Despite being significantly reduced in size compared to full antibodies, the employed scFVs and Fabs may produce steric constraints when binding their epitope. To circumvent potential complications associated with the use of TCR-attached probes we followed an alternative FRET-based strategy involving site-specific fluorescence-labeling of pMHCs, i.e. the TCR-ligands (figure 5a). To this end the MHC-embedded moth-cytochrome-c (MCC) was coupled at either the N- or the C-terminus to AF555 or AF647 to serve as FRET donor and FRET acceptor when presented via SLB-embedded I-Ek to 5c.c7 TCR-transgenic T-cell blasts. For FRET-measurements T-cells were confronted with SLBs harboring ICAM-1, B7-1 and AF555-/AF647-labeled I-Ek/MCC present at a density of 100 molecules μm-2 or higher, and FRET-donor increase was monitored after FRET-acceptor ablation.

Figure 5. FRET-based evaluation of synaptic interactions between TCR-engaged pMHCs.

Figure 5

(a) Rationale underlying the FRET-based assessment of interactions between TCR-engaged pMHCs. 5c.c7 TCR-transgenic T-cell blasts are confronted with SLBs featuring ICAM-1, B7-1 and a mixture of stimulatory I-Ek/MCC conjugated either at the N- or C-terminus of peptide cargo with the FRET donor dye AF555 or the FRET acceptor dye AF647 (left). Recruitment of pMHCs into discernible synaptic clusters are reflective of emerging TCR microclusters giving rise to a high degree of pMHC binding. The pMHC density within such microclusters was found to be up to four-fold increased when compared to the density outside the synapse (right, n = 52 cells).

(b) Representative FRET-based experiments to score for homotypic synaptic interactions between TCR-engaged pMHCs as illustrated in (a). Shown are FRET donor (green), as well as FRET donor intensities in false-color representation before and after FRET acceptor ablation. FRET efficiencies were determined on a pixel-per-pixel basis and are shown in the right column. Relative enrichment of I-Ek/MCC within TCR microclusters and cSMACs compared to the initial

I-Ek/MCC density within the SLB prior to addition of T-cells is indicated. Scale bar: 3 μm.

(c) FRET efficiencies measured on SLBs featuring I-Ek/MCC-AF555(N), I-Ek/MCC-AF555(C), I-Ek/MCC-AF647(N) and I-Ek/MCC-AF647(C) in combinations and densities as indicated (n ≥ 8 regions measured on SLB per density). FRET values expected from randomized molecular encounters via Monte Carlo simulations are shown at indicated pMHC densities (AF555:AF647 = 1:1).

(d, e) FRET efficiencies as determined within entire synapses (e, N/C n = 19 cells, C/C n = 18 cells, N/N n = 15 cells) and individual TCR microclusters (d, N/C = 17 cells / 53 microclusters, C/C = 18 cells / 31 microclusters, N/N = 15 cells / 40 microclusters). Areas outside the synapse refer to SLBs, which are not in contact with T-cells, as is indicated by the demarcation lines in (b).

Error bars represent standard error of the mean.

In the absence of T-cells, all three different pMHC combinations, i.e. MCC-AF555(N)/MCC-AF647(N), MCC-AF555(C)/MCC-AF647(C) and MCC-AF555(N)/MCC-AF647(C) gave rise to slight FRET signals due to random molecular encounters of freely diffusing pMHCs, which increased with increasing pMHC densities and matched values predicted by Monte Carlo simulations (figure 5c and Methods). Of note, synaptic FRET yields were equal or even lower than those measured outside the synapse, despite considerable pMHC enrichment due to TCR-binding (figure 5d, e). In fact, the fraction of TCR-bound pMHCs can be readily quantified within a region of interest (encircled region in figure 5b) by means of pMHC entrapment above initial pMHC densities prior to the addition of T-cells 28. We hence focused our analysis on synaptic areas enriched in pMHCs such as TCR microclusters as FRET values measured therein are reflective of the oligomerization state of TCR-engaged pMHCs. As shown in figure 5e we did not observe any FRET within synapses attributable to TCR binding. On the contrary, minimally elevated FRET yields detected underneath TCR microclusters (figure 5d) were significantly below those measured in cell-free SLBs featuring corresponding pMHC densities and those predicted by our Monte Carlo simulations (figure 5c). This finding suggests pMHC entrapment by activated TCRs are actively spaced from one another in microclusters at distances that are larger than those susceptible to FRET-based detection.

In conclusion, even the minimally invasive FRET approach employing site-specifically labeled agonist ligands as FRET probes yielded no indication for TCR-induced homotypic pMHC interactions. This observation combined with the lack of FRET between scFV- and Fab-probed TCRs (see above) renders scenarios, in which synaptic TCRs engage nominal pMHC-ligands as higher order structures, highly unlikely.

Discussion

The extent to which surface TCRs are pre-organized in or assemble into higher order structures when engaging pMHCs, continues to dominate the quest for the molecular mechanisms underlying the maintenance of T-cell antigen sensitivity and the initiation of TCR-proximal signaling 6. What has in fact been entirely missing in the ongoing debate is direct evidence that oligomers of TCR/CD3 complexes, constitutive or ligand-induced, exist within the plasma membrane of living T-cells. Adequate experimental systems need to both conserve the unique properties of the T-cell plasma membrane and provide at the same time molecular readout.

Advanced and minimally invasive live cell imaging and spectroscopy approaches are poised to meet such criteria especially when paired with fluorescent probes engineered to inform about nanometer length scales. In this sense, we took advantage of functionalized SLBs serving as surrogate APCs and the non-invasive TOCCSL and PA/FCS methodology to monitor the stoichiometry of laterally mobile surface TCRs, which comprise in our hands up to 70% of all surface-resident TCRs prior to ligand recognition. To score for interactions between immobile TCRs or TCRs with reduced mobility as well as pMHC-engaged TCRs, we have carried out quantitative FRET measurements, which are independent of lateral TCR-mobility. For precise quantitation, we employed (i) TCRβ-reactive H57-scFVs and CD3ε-reactive KT3-scFVs, which had been stoichiometrically dye-conjugated in a site-specific manner, (ii) TCRβ-reactive Fabs binding to a TCRβ-epitope involved in pMHC binding and (iii) site-specifically labeled pMHC molecules embedded within the SLB to be recognized and spatially reorganized via engaging TCRs by contacting T-cells. Anisotropy measurements of fluorophores associated with pMHCs and T-cell bound antibody fragments were consistent with rotational freedom for meaningful FRET measurements (supplementary figure 1f).

Despite affording single molecule sensitivity, none of the applied approaches provided any evidence for the existence of constitutive or ligand-induced higher order TCR-structures. Neither the more quantitative single color brightness nor the more sensitive single molecule coincidence TOCCSL analysis yielded mobile TCR structures other than monomers while both approaches unambiguously supported the widely-expected TCR/CD3 subunit composition. This was subsequently confirmed by PA/FCS, allowing us to rule out the existence of short-lived TCR higher order structures not accessible to TOCCSL.

Importantly, FRET-based measurements conducted to account for immobile, slowly diffusing or ligand-engaged TCR entities failed altogether to provide any data in favor of higher order structures while they proved sensitive enough to reveal TCR dimers induced via divalent streptavidin or full anti-Vβ3 antibody. For a number of reasons, we consider it highly unlikely that the absence of FRET signals resulted merely from limitations of the FRET systems applied: (i) Positioning FRET dye combinations with 5.1 nm to 8.2 nm Förster radii (R0) within the TCR/CD3 complex reveals inter-dye distances of up to 15 nm, i.e. spanning the dimensions of more than two TCR/CD3 complexes (figures 6a, b), even when such interactions are present in low abundance. (ii) The further use of site-specifically labeled stimulatory pMHCs as FRET probes provided us with additional flexibility in label positioning (figure 6c), yet did not give rise to any detectable ensemble FRET signals above background, even when focusing predominantly on pMHCs accumulating as a result of TCR-engagement in TCR microclusters. In fact, the opposite was true, as yields concerning inter-pMHC FRET underneath these TCR-enriched structures tended to be significantly lower than those measured in cell-free SLBs featuring freely diffusing pMHCs in comparable amounts. This suggests that signaling TCRs coalescing in macroscopically discernible membrane entities are actively kept apart by yet to be identified cell-intrinsic mechanisms. Once formed, such entities do not require F-actin or intact microtubules, as is indicated by pharmacological inhibition studies (supplementary figure 5d). Addressing the role of signalosome assembly will invariably require conditional interference with components of TCR-proximal signaling (e.g. p56lck, ZAP70, LAT, SLP76), as inhibition of the TCR proximal kinase p56lck via pp2 disturbed TCR microcluster formation only in part. FRET yields measured between TCR-bound pMHCs accumulated underneath such microclusters were still reduced (supplementary figure 5d).

Figure 6. FRET-based assessment of epitope proximities within putative dimeric TCR and pMHC configurations.

Figure 6

(a, b) Composite structure of the 5c.c7 αβTCR decorated with H57-scFV in two dimer configurations as previously proposed 39. The position of both-FRET dyes is indicated by a red and green star. Putative dimerization motives are shown in yellow and orange. FRET efficiencies as calculated for the AF555/AF647 (A) and AF568/AF647 (B) dye combinations with Forster radii of 5.1 nm and 8.2 nm respectively are indicated.

(c) If I-Ek-embedded peptides (yellow) served as FRET sensors for the detection of higher order TCR-structures, both I-Ek-dimer configurations would give rise to substantial FRET yields as indicated. Scale bar: 5nm.

It should be noted, that all FRET-based experiments involving scFVs or Fabs as FRET probes were carried out at 24°C to allow for a quantitative readout unaffected by probe dissociation, hence leaving open the possibility of TCR-oligomerization at 37°C. In such case, higher order TCR-structures would be nonetheless irrelevant for maintaining T-cell antigen sensitivity, as this is unaffected at 24°C 28. In addition, FRET-based experiments conducted with the use of pMHC-FRET probes at 37°C gave rise to identical results (supplementary figure 5e and f), rendering temperature-dependent TCR-oligomerization as such improbable.

A possibility we have not addressed directly is the existence of glycosylated and immobile higher order TCR entities assembled via galectins and other glycoproteins 35, which could in principle link TCRs at distances that are larger than those accessible by FRET. Such structures should be detectable by superresolution microcopy and dissolvable into mobile TCR monomers after exposure to lactose competing with galectin-binding. However, since lactose treatment did not cause an increase in mobile TCRs (supplementary figure 1g), we deem such scenario rather unlikely.

Since neither TOCCSL- nor PA/FCS- nor FRET-based experimentation showed any indication for homotypic TCR associations, we consider constitutive as well as ligand-induced TCR-oligomers for the maintenance of antigen sensitivity and signal generation irrelevant.

It should be noted though that we cannot entirely rule out that the use of fluorescent antibody fragments in our studies prevented TCR oligomerization and hence interfered with the detection of higher order structures. However, the existence of such hypothetical oligomers cannot explain the exquisite sensitivity of T-cell antigen recognition, as this is not affected by TCRβ-H57-scFV binding28, 34. Alternatively, already formed TCR oligomers might be accessible to one antibody probe only as the result of a unique quaternary structure, and could evade in this fashion detection via TOCCSL, PA/FCS and FRET. However, such a scenario is highly unlikely given that targeting TCRβ at two different non-overlapping epitopes gave rise to virtually identical results while probing CD3ε with KT3-scFVs in TOCCSL and PA/FCS experiments resulted predominantly in dimer signals reflective of single TCR/CD3 complexes but no tetramer-, hexamer or octamer signals corresponding to TCR dimers, trimers or tetramers. In this context, it is also noteworthy that decorating 5c.c7 TCR-transgenic T-cells with the KJ25-Fab causes a complete blockade of T-cell antigen recognition due to its epitope being involved in pMHC engagement. Hence, if there were any unaccounted TCRs as part of larger complexes due to antibody inaccessibility, such TCRs would not be able to engage pMHCs in a productive manner. In such case TCR/CD3 oligomers would not be able to bind more than one pMHC at any given time. Consistent with this, we did not witness any TCR-engaged pairs or oligomers of agonist ligands in close enough proximity for FRET-based detection regardless of how we labeled the SLB-embedded pMHC.

How can one then reconcile previous observations without any evidence for higher order TCR-structures? The clustered appearance of gold-conjugated anti-TCR-antibodies in electron microscopy-studies as reported by Schamel and colleagues 12 can be explained with the existence of nanoscopic membrane domains of 20 to 200 nm in diameter, in which mobile monovalent TCRs were reported to be highly enriched 16, 36. Incomplete disruption of such nanodomains during partial cell lysis 12 could very well cause the reduced migration of TCR/CD3 complexes in blue native PAGE as well as increased sedimentation in the course of ultracentrifugation.

A recent EM-based study on negatively stained TCR/CD3-antibody complexes extracted from cell lysates suggested pMHC-induced TCR-dimer formation as a trigger for T-cell activation 37. In view of the detergent solubilization step required for sample preparation it remains however unclear whether the observed TCR entities had already been present prior to, or formed during, cell lysis. In any case, if present on living cells such structures would have given rise to a readily detectable FRET signal with at least one of our applied FRET-based approaches.

The fact that we did not observe higher order TCR-structures under resting or activating conditions does not necessarily contradict any mechanisms by which T-cells can be activated via TCR-multimerization as enforced by soluble ligands. The underlying trigger could involve TCR-segregation from otherwise accessible phosphatases, yet is in view of the data presented here likely to deviate from the mechanisms responsible for physiological stimulation, especially when antigen is sparse. In line with this notion, TCR-proximal signaling induced by soluble antibodies is barely influenced by treatment with actin-depolymerizing drugs such as cytochalasin D or latrunculin A, while APC- or SLB-mediated stimulation is severely affected by it, even when drug interference occurs after initial contact formation to ensure TCR-pMHC engagement 38 (and own observations). Employing an ectopic erythropoietin receptor-based expression system, Kuhns and colleagues quantified cell proliferation driven by close subunit proximity as a readout for TCR-dimerization 39. This assay produced signals referred to as TCRα- and TCRβ-homo-association, which amounted however to less than 3% of the signal of the positive control involving the interaction between TCRα/β and CD3ε. Frequent but random collisions of mobile monomeric TCR/CD3 complexes enriched within specific membrane domains 16, 36 could very well explain the modest effect on cell proliferation in this system. Such collisions have indeed been observed on immortalized T-cell lines with low TCR expression levels via two color coincidence spectroscopy 14 and also for tagged TCR/CD3 complexes ectopically expressed in HEK cells through recording Bioluminescence Resonance Energy Transfer (BRET) 40.

In conclusion, employing four independent live cell imaging and spectroscopy methodologies we did not observe TCRs of higher order structure on the surface of living primary T-cells both under resting and ligand-engaged conditions, even though our methods proved sensitive enough to confirm the subunit stoichiometry of the TCR/CD3 complex. We therefore rule out that TCR oligomerization, dynamic or constitutive in nature, plays a significant role in antigen sensitization and the generation of intracellular signaling events following extracellular TCR-engagement within the immunological synapse. Hence factors other than ligand valence per TCR entity and receptor oligomerization may be instrumental in highly sensitized TCR-based signal initiation. These may include serial engagement of multiple TCRs by agonist pMHCs, kinetic segregation as a direct consequence of the molecular dimensions of receptor-ligand pairs involved and the synaptic space permitted 41, 42, as well conformational changes within the TCR 43 and the TCR/CD3 complex 44, 45, 46, 47 upon ligand engagement and even deformations within the receptor complex as brought about by cellular forces 48, 49. Given the extraordinary efficacy of T-cell antigen detection, the monomeric nature of ligand-engaged TCR/CD3 complexes highlights their function as catalytic centers for the recruitment, activation and subsequent rapid release of the TCR-proximal kinase ZAP70 for downstream signal amplification 50.

Obviously, more work will be necessary to clarify the significance of individual parameters influencing T-cell antigen sensitivity, the relevance of which can hardly be overestimated for adaptive immune functions and the design of T-cell-based immunotherapies. Reductionist approaches involving T-cells modified with much simplified TCR-surrogates, such as chimeric antigen receptors (CARs), may help in this quest 6 but may not always prove adequate, since CAR-T-cells require 100 to 500 times more antigen for stimulation. Given the transient nature of protein-protein and protein-lipid interactions and also the non-linear properties of the conjugated cells involved, the use of non-invasive molecular live cell approaches will be a sine qua non in this endeavor.

Methods

Tissue culture and retroviral transduction

Primary T cells were isolated from 5c.c7 αβTCR transgenic mice and cultured as described 1 in T-cell media, i.e. RPMI containing 10% FCS (Sigma-Aldrich), 2 mM L-glutamine (Irvine Scientific), 100 U/ml penicillin/streptomycin, 50 μM β-mercaptoethanol (Sigma-Aldrich), which had been supplemented with 1 μM C18-reverse phase HPLC-purified moth cytochrome c (MCC) 88-103 peptide (sequence: ANERADLIAYLKQATK, T-cell epitope underlined, Elim Biopharmaceuticals Inc, USA). T-cells were used for experiments 7-9 days after stimulation. Retroviral transfection of T-cells with ZAP70-GFP was performed as published 1, 2. Freshly isolated primary OT-1 TCR-transgenic T-cells were cultured in T-cell media containing 1 μM C18-reverse phase HPLC-purified ovalbumin peptide 257-264 for (sequence: SIINFEKL, T-cell epitope underlined, Elim Biopharmaceuticals Inc) as described 3 and used 8 days after stimulation.

Protein expression and functionalization

ICAM-1-His12 and B7-1-His12 were expressed using a baculovirus expression system and purified as described 4. I-Ekα-His6 and I-Ekβ-His6 was expressed as inclusion bodies in E. coli, refolded in vitro in the presence of the MCC88-103 or the MCC(ANP) space-holder peptide (sequence: ANERADLIAYL[ANP]QATK) (Elim Biopharmaceuticals Inc) as described 5, 6, 7 and purified via Ni2+-NTA resin affinity chromatography followed by size-exclusion S-200 chromatography (S-200) using an Äkta purifier system (GE Healthcare Life Sciences, UK). Peptides used for pMHC-based FRET measurements included MCC FRET (N) (CSDLIAYLKQATKGG, T-cell epitope underlined) and MCC FRET (C) (sequence, SSDLIAYLKQATKGGSC, T-cell epitope underlined) (both from Elim Biopharmaceuticals Inc). After purification via C18 reverse-phase high-performance liquid chromatography (HPLC), peptides were conjugated at the cysteine residue with Alexa Fluor 555 or Alexa Fluor 647 maleimide (Thermo Fisher Scientific, USA), purified again by HPLC and confirmed in identity by liquid chromatography-electrospray ionization mass spectroscopy. For production of AF555-/AF647-pMHC-(N) or -pMHC-(C) I-Ek-2xHis6 preloaded with MCC(ANP) were incubated in peptide exchange buffer (PBS plus 100mM citric acid, pH5.1) with the corresponding quantitatively fluorescence-labeled MCC FRET (N) or MCC FRET (C) peptides present in 20-fold excess for 48 hours at room temperature. Following peptide exchange, properly folded I-Ek molecules loaded with fluorescence-labeled MCC FRET (N/C) peptides were purified via S-200 size chromatography and subjected to spectrophotometry to verify quantitative peptide labeling and peptide exchange. For KT3 scFV generation, mRNA was prepared from KT3 hybridoma using TRIzol (Thermo Fisher Scientific) according to the manufacturer’s instructions (to serve as a template for 5′ rapid amplification of cDNA ends (RACE, Thermo Fisher Scientific). VH and VL antibody domains were fused as illustrated in supplementary figure 5 and mutagenized for site-specific label attachment (Quikchange; Agilent, USA).

H57-scFV and KT3-scFV were expressed in inclusion bodies, refolded, site-specifically labeled with maleimide-functionalized Alexa Fluor 488, Alexa Fluor 555, Alexa Fluor 568, Alexa Fluor 647 (all Thermo Fisher Scientific) or Abberior Star 635P (Abberior GmbH, Germany) and purified as described 4. Label positons within the H57- and KT3-scFVs are indicated in supplementary figures 6 and 7. The protein to dye ratio ranged between 0.95 and 1.0 for all Alexa dye conjugation products as determined by photo-spectrometry. To serve as positive FRET donor control, H57-scFV was equipped via a C-terminal cysteine (J4, supplementary figure 6) with maleimide-TCO (Jena Bioscience, Germany) and then conjugated via click-chemistry to a synthetic N-terminally biotinylated FRET DONOR peptide linker (biotin-GGGGSY(GGGGS)2KGGGGSC-maleimide-6-Me-Tetrazine, Elim Biopharmaceuticals, USA), which had been labeled with Alexa Fluor 555 N-succinimidyl ester at its lysine residue (underlined). The positive FRET acceptor control consisted of a J4 H57-scFV linked via its C-terminal cysteine residue to maleimide-DBCO (Jena Bioscience) and subsequently conjugated via click-chemistry to a synthetic N-terminally biotinylated FRET ACCEPTOR peptide linker (biotin-GGCGS(GPGGA)5GGKYGGSK-Azide, INTAVIS GmbH, Germany), which was labeled at its lysine residue (underlined) with Alexa Fluor 647 N-succinimidyl ester. Prior to click-conjugation both FRET DONOR and FRET ACCEPTOR peptide linkers had been purified by C18-reverse phase HPLC and confirmed in identity by mass-spectrometry. T-cells were decorated with both H57-scFV-biotin FRET dye linker probes in equimolar amounts, which were subsequently crosslinked with streptavidin engineered to harbor two biotin binding sites positioned in trans 8, 9.

H57-scFV used in photon arrival time experiments was conjugated with Abberior-Star 635P maleimide (AS635P-H57- scFV) yielding a protein to dye ratio of 0.9.

KJ25 mAbs were purified from hybridoma culture supernatant (serum-free medium, Gibco, 12045076) by affinity chromatography using GammaBind Plus Sepharose (GE Healthcare Life Sciences) according to manufacturer’s instructions and then subjected to size-exclusion S-200 chromatography for the removal of aggregates. For Fab production, purified KJ25 mAbs were processed using the Pierce™ Fab Preparation Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Fabs were subjected to S-75 size exclusion chromatography and concentrated to 1mg/ml by ultrafiltration (Merck, UFC901096). After addition of 10% 1M NaHCO3 the protein was allowed to react with a 2-fold molar surplus of fluorescent dyes functionalized with succinimidyl ester. The reaction mix was lastly subjected to S-75 size exclusion chromatography to remove unreacted dye and arrive at monomeric dye-conjugated KJ25-Fabs.

Decoration of T-cells with antibody fragments

5x105 T-cells were washed in ice- cold Hank’s buffered salt solution (HBSS, Lonza, Switzerland) supplemented with 2% FCS (imaging buffer,) and re-suspended in a volume of 30 μl imaging buffer. Respective fluorescence-labeled probes were added until saturation of binding had been achieved after incubation for 20 minutes at 4°C, as determined in preceding titration experiments (all H57-scFVs: 150 ng for 105 cells in 30 μl; KT3-scFVs: 3 μg for 105 cells in 30 μl; KJ25-Fabs: 300 ng for 105 cells in 30 μl, see also supplementary figure 1e). Quantitative cell labeling was verified by flow cytometric analysis applying different incubation times and probe concentrations. After incubation, cells were washed twice with imaging buffer and stored on ice prior to experiments.

Major deviations from stoichiometric labeling of T-cells with H57-scFVs and KJ25-Fabs could be excluded for a number of reasons:

  • (1)

    The site-directed nature of maleimide-/unpaired cysteine- mediated labeling procedure and measured protein:dye ratios of close to 1 (see above) ensured the presence of one fluorophore per probe.

  • (2)

    Single color TOCCSL experiments involving AF647-KT3-scFV gave rise to ~ 74% double positive events (figure 1d) indicating that dark states for AF647 amount at the most to (1-square root(0.74)) x 100%) 14%. This is an overestimate due to potential complications associated with the TOCCSL method itself (e.g. partial bleaching of CD3ε dimers located at the boundary of the bleach spot) and with labeling to saturation, which is more complex for KT3-scFVs than for H57-scFVs and KJ25Fabs, as indicated by the T-cell/label saturation curves (supplementary figure 1 e).

  • (3)

    Measuring TCR-surface densities employing H57-scFV conjugated to AF488, AF555, AF568 and AF647 gave rise to almost identical results, implying that dark AF488, AF555 and AF568 dyes are present at a fraction of 14% or less.

  • (4)

    Exposing T-cells to the KJ25-Fab leads to a complete blockade of T-cell antigen recognition of nominal pMHCs (I-Ek/MCC) indicating occupation of all TCRs involved in antigen binding by this antibody, as is expected given that the KJ25 epitope overlaps with the TCR paratope binding to pMHC. All TCRs, which are relevant for ligand recognition, are hence accessible to KJ25-Fab labeling. Substoichiometric labeling of TCRs by H57-scFVs is also highly improbable, since TCR-surface density measurements, which are based on KJ25-Fabs and H57-scFVs, give rise to identical results. This assessment is consistent with the lack of FRET between TCR-engaged pMHCs.

Drug perturbations and glycosylation interference

For pharmacological inhibition of actin/tubulin polymerization after the formation of microclusters, Latrunculin B (Sigma-Aldrich) and Nocodazole (Sigma-Aldrich) was added to the measurement chamber to a final concentration of 10 and 100 μM, respectively. For p56lck kinase inhibition, cells were pre-treated with 10 μM PP2 (Sigma-Aldrich, P0042) for 60 minutes before addition to the SLB. PP2 remained in solution during the measurement. For glycosylation interference, T-cells were incubated for 20 min on ice with 20 mM lactose or 20 mM sucrose diluted in imaging buffer. Control experiments were carried out in imaging buffer. During measurements, lactose or sucrose was present at 2 nM.

Preparation of planar glass-supported lipid bilayers (SLBs)

Vesicles containing 90% 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 10% 1,2-dioleoyl-sn-glycero-3-[N(5-amino-1-carboxypentyl)iminodiacetic acid] succinyl[nickelsalt] (Ni-DOGS NTA; Avanti Polar Lipids) were prepared as described 4 and stored in PBS buffer. Cover glass slides (#1.5, 24x60 mm, Menzel, Germany) were cleaned for 30 min in Piranha solution containing concentrated H2SO4 and 30% H2O2 (both from Sigma-Aldrich) at a ratio of 7:3. Slides were rinsed with purified water, dried under a nitrogen stream and attached with the use of dental imprint silicon putty (Picodent twinsil 22, Picodent, Germany) to LabTek 8-well chambers (Nunc, USA), from which the glass bottom had been removed as published 7. Glass slides were incubated with a tenfold diluted vesicle solution for 10 min, before they were extensively rinsed with phosphate buffered saline (PBS, Lonza). For functionalization, formed SLBs were incubated with His-tagged proteins for 75 minutes at room temperature and then rinsed with PBS. Prior to the addition of T-cells, PBS was exchanged with imaging buffer. pMHC surface densities were calculated by comparing the ensemble fluorescence signal with that of individual fluorophores.

Determination of TCR surface densities

Average surface densities were calculated from T-cells in contact with ICAM-1 functionalized SLBs and labeled to saturation with AF647-H57-scFV. The T-cell brightness per μm2 (measured in TIR configuration) was then divided by the fluorescence signal of a single AF647-H57-scFV. The latter was determined by incubating T-cells with a mixture of unlabeled/AF647-H57-scFVs at a ratio of 150:1 and measuring the brightness of individual diffraction-limited spots.

Microscopy - TOCCSL

Single molecule experiments were performed as described 10, 11. Briefly, an Axiovert 200 microscope equipped with a 100x NA=1.46 Plan-Apochromat objective (Zeiss, Germany) was used for illuminating samples in objective-based total internal reflection (TIR) configuration via the epiport by 488 nm (Sapphire, Coherent, USA) or 647 nm (Innova I-90, Coherent) laser light sources with a power density of 1-5 kW/cm2 on the sample. For exact timing, acousto-optic modulators (Isomet, USA) were used. A slit aperture (Thorlabs, USA) with a width of 5 to 10 μm in the sample plane was used as field stop. Timing protocols were generated and executed with the use of an in-house written program package implemented in LABVIEW (National Instruments, USA). After appropriate filtering (zt488/640 rpc, Chroma, USA; FF01-538/685-25, Semrock, USA), emitted signals were split into two color channels using a custom-made dichroic wedge (Chroma) and imaged on the same back-illuminated, liquid-nitrogen cooled CCD-camera (Micro-Max, Roper Scientific, USA). For precise temperature control an in-house built incubator equipped with a heating unit and an objective heater (PeCon, Germany) were used. Unless noted otherwise experiments were carried out at 24°C.

After recording a pre-bleach image with a power density of 1kW/cm2 and an illumination time of five milliseconds (figure 1c, pulse i), samples were bleached with a laser pulse applied for 400 milliseconds with a power density of 5kW/cm2. Following a recovery time of two to ten seconds, up to 18 images were recorded at 20 milliseconds intervals with the same settings as used for the pre-bleach image. Complete photoablation was tested by acquiring an image one millisecond after bleaching (figure 1c, pulse ii). The recovery time was adjusted to allow for reentry of unbleached signals and the first image (figure 1c, pulse iii) was used for brightness analysis (TOCCSL image). The last image served to determine the brightness of a single fluorophore. Due to the low photostability of Alexa Fluor 647, the probability of observing more than one fluorescent Alexa Fluor 647 dye per diffraction limited spot in the last image is negligible. Alternatively, T-cells were decorated with a 1:150 mixture of AF647-scFs and unlabeled scFVs, placed on SLBs and the single spot brightness was determined for individual AF647-scFVs. Both approaches gave rise to similar results. For dual color TOCCSL experiments (figure 2a), the pre-bleach images were recorded with a delay of 20 milliseconds, the bleach pulses were applied simultaneously and the TOCCSL images were either recorded simultaneously (figure 2a, pulse iii) or in series with 1 millisecond time gap between the separate acquisitions of the two channels (supplementary figure 2d).

To ensure TOCCSL analysis unbiased by the potentially slower diffusion of higher order TCR/CD3 entities we took the following steps: (1) After having tracked individual TCR/CD3 entities over time, a statistical analysis of observed step lengths (of individually tracked entities) yielded two distinct fractions: an immobile fraction (~30%) and a mobile fraction (~70%). A histogram of the diffusion constants of individually traced entities gave rise to a unimodal distribution (supplementary figure 1d). (2) Prolonging the recovery time (by up to 10 seconds) to get access to slowly diffusing entities did not influence the outcome. (3) Depletion of fast TCR/CD3 entities through repetitive TOCCSL procedures on the same cell to enrich for slowly diffusing entities did not alter the results.

Photon-Antibunching/FCS experiments

All experiments were conducted on a customized setup equipped with a 635 nm pulsed diode laser (LDH-D-C-635, PicoQuant, Germany) featuring a pulse width below 100 ps and operated at 9.5 MHz for fluorescence excitation. The excitation beam was fed into the objective (HC PL APO 100x/1.40 OIL CS2, Leica Microsystems, Germany) via an appropriate dichroic mirror (zt 625-745 rpc, Chroma) separating excitation from emission. The fluorescence was collected by the same objective, split into four identical channels by three 50:50 beam splitters, each filtered by a band pass filter (685/70 ET, Chroma), coupled into four multimode fibers and detected by four avalanche photodiodes (SPCMAQR-13-FC, Perkin Elmer Optoelectronics, USA). The fibers acted as confocal pinholes with a diameter of approximately 1 airy disc and were aligned to the focal excitation maximum. Data acquisition and analysis software were written in C++ and MATLAB (The Mathworks Inc.). A three-axis piezo stage (TRITOR 102 CAP, Piezosystem Jena, Germany) was used to raster scan the T-cell plasma membrane in contact with the SLB and to find regions of interest for FCS measurements, which were controlled by a multifunction analog output computer card NI PCI-6731 (National Instruments). Data acquisition relied on time-correlated single-photon counting in absolute timing mode with the use of a DPC-230 card (Becker & Hickl GmbH, Germany) and additional custom-made electronic cards allowing for the use of 4 or more independent detection channels. FCS-antibunching experiments were conducted for 180 seconds with 1μW illumination power at the back aperture of the objective. Control experiments involving soluble scFVs or Fabs in imaging buffer were performed for 30 minutes with an illumination power of 5 μW at the back aperture of the objective. To block non-specific scFV/Fab -binding to the glass surface, the coverslips were blocked with biotinylated bovine serum albumin (Sigma-Aldrich, 1mg/ml in HBSS) for ten minutes followed by a five-minute incubation step in imaging buffer.

FRET experiments

A microscope setup similar to that employed for TOCCSL-based experiments was used with the following deviations: Alexa Fluor 555 and Alexa Fluor 568 were excited by 532nm solid state laser (Millennia Pro, Spectra-Physics, USA). Fluorescence emission was separated from excitation by a dichroic mirror (R405/488/532/635, Semrock), split into two color channels via a Dual View system (Photometrix, USA) equipped with appropriate filter sets (640dcxr, HQ585/40m, HQ700/75m, Chroma) and detected with a back illuminated EMCCD-camera (iXon Ultra 897, Andor, UK). For measuring ensemble FRET yields through FRET donor recovery after FRET acceptor photo bleaching, a custom timing protocol was applied to record the following sequence: a five millisecond low intensity (100 W/cm2) Alexa Fluor 647 acquisition was followed by a five millisecond low-intensity Alexa Fluor 555/Alexa Fluor 568 acquisition (to monitor the Alexa Fluor 555/ Alexa Fluor 568 signal before acceptor photo bleaching), a 200 millisecond high-intensity 647 nanometer bleach pulse (5 kW/cm2, to ablate Alexa Fluor 647), a second acquisition of Alexa Fluor 555/Alexa Fluor 568 (to monitor the FRET donor signal increase after acceptor bleaching) and finally a low-intensity Alexa Fluor 647 acquisition (to verify Alexa Fluor 647 ablation). Excitation was performed in TIR mode. For FRET analysis, the background-corrected average Alexa Fluor 555/Alexa Fluor 568 brightness of the entire contact area between T-cell and SLB, of synaptic TCR microclusters (MC) or of the central supramolecular activation cluster (cSMAC) was calculated before (Ibb) and after acceptor ablation (Iab). The FRET efficiency was calculated by dividing the difference (Iab-Ibb) by Iab. To compensate for the possible movement of TCR MCs during one FRET experiment, FRET yields were calculated by averaging the FRET yield of MCs which were selected in the image before acceptor bleaching and MCs which were selected in the image after acceptor bleaching. As a negative control, the FRET experiment was performed without acceptor bleaching, yielding the amount of photo-bleaching after two subsequent donor excitation pulses. For every experiment, the respective FRET efficiencies were corrected for photo-bleaching.

As shown in supplementary figure 5g, inter-pMHC FRET yields within synaptic areas, which were not enriched for pMHCs (i.e. areas with little or no pMHC binding), and inter-pMHC FRET yields measured outside the synapse were indistinguishable. However, when determined on a pixel by pixel level, different degrees of heterogeneity in FRET yields were observed outside and inside the synapse. These heterogeneities resulted from (i) temporal fluctuations in pMHC distribution on the lipid bilayer and (ii) the nature of the dequenching-based FRET-quantitation requiring the acquisition of two FRET donor images separated by a bleach pulse with a duration of 200 ms. pMHCs inside the synapse move more slowly due to random interactions with the T-cell´s glycocalyx, thereby reducing pixel-by pixel heterogeneities in measured FRET yields. As a consequence, pixel-by-pixel heterogeneities in inter pMHC FRET were also found significantly reduced in synapses of mismatched OT-1 TCR-transgenic T-cells confronted with SLBs featuring I-Ek -FRET probes, unlabeled B7-1 and ICAM-1 (supplementary figure 5g).

Experiments involving the simultaneous visualization of the synapse-recruited ZAP70-GFP were performed on a slightly modified TIRF microscopy setup, which included in addition to the system outlined above a third laser for excitation of GFP (Sapphire) and an triple emission beam splitter for simultaneous recording of the GFP, the FRET donor and FRET acceptor channel (Optosplit III, CAIRN Research, UK, equipped with BS640 (Semrock), zt532RDC (Chroma), HC510/20 (Semrock), 570/60 (Chroma) and 675/50 (Chroma)). Settling of the T-cells on the SLB was monitored in DIC. As soon as synaptic ZAP70-70 recruitment became observable in TIRF via attenuated 488 nm excitation, FRET yields were determined with the dequenching-based approach employing 532 and 647 nm TIRF excitation as described above.

Anisotropy measurements

Anisotropy measurements were performed as described in 12. Briefly, polarized low intensity light (~150 W/cm2) from 532 nm or 647 nm laser sources was used to excite the probes for 5 ms in non-TIR, epi-illumination configuration. After appropriate filtering (see above, Microscopy-TOCCSL) the emission was split by a Wollaston prism into two polarization components, Ip and Is, and imagined on the same back-illuminated, liquid-nitrogen cooled CCD-camera (see above, Microscopy-TOCCSL). To compensate for different detection efficiencies in the p- and s-polarized channels, the correction factor, g, was determined with a 1 μM solution of Alexa Fluor 555 (gAF555 = 0.87), Alexa Fluor 568 (gAF568 = 0.62) and Alexa Fluor 647 (gAF647 = 0.77) dyes placed on an anti-absorptive cover glass. Anisotropy was calculated by r=IpgIsIp+2gIs according to Lakowicz 13.

TOCCSL and mobility analysis

TOCCSL experiments were analyzed as described in 10. Briefly, a MATLAB (The MathWorks, USA) - based maximum likelihood estimator was used to determine position, integrated brightness B, full width at half maximum (FWHM), and local background of individual signals in the TOCCSL images. A detailed description of the applied brightness analysis is provided in 14 and 15. The brightness values B of single AF647-scFV molecules pooled from the last images of all TOCCSL sequences were used to calculate the probability density function (pdf) of monomers, ρ1(B). Due to the independent photon emission process, the corresponding pdfs of N co-localized emitters can be calculated by a series of convolution integrals, ρN(B)=ρ1(B)ρN1(BB)dB. A weighted linear combination of these pdfs was used to calculate the brightness distribution of a mixed population of monomers and higher-order multimers:

ρ(B)=N=1NmaxαNρN(B) equation 1

The brightness values from all TOCCSL images of multiple cells for one experimental condition were used to calculate ρ1(B). A least-square fit with equation 1 was employed to determine the weights of the individual pdfs, αN, with N=1NmaxαN=1. For all fits, no higher contributions than dimers (α2) were observed. A minimum of 250-500 brightness values 16 was applied to calculate ρ1(B) and ρN(B). TCR surface densities from images obtained in standard fluorescence microscopy experiments were determined by dividing mean intensities per μm2 by the average single molecule brightness. To calculate the size of the mobile fraction size by FRAP, TOCCSL sequences were applied on single cells several times. Mobile fraction sizes could then be estimated by measuring the relative recovery of fluorescence intensity in the pre-bleach image of consecutive experiments with delay times of 20 seconds. Single molecule tracking was performed as described in 17. A bi-exponential fit to the probability distribution of square displacements yielded the fraction of the mobile component, as described in more detail in 18. Both methods yielded comparable values for the laterally mobile fraction (figure 1e). Mean square displacement analysis of individual trajectories yielded a unimodal distribution of diffusion constants (supplementary figure 1e).

A detailed description of the applied co-localization analysis performed for dual color TOCCSL experiments is provided in 11. For correcting chromatic aberrations, fluorescent multi-color beads (TetraSpeck, Life Sciences, USA) were immobilized on glass slides and imaged under identical conditions as for dual color TOCCSL experiments. Positions of beads in both color channels were determined (supplementary figure 2a) and used to calculate the relative shift and stretch of the two channels to each other. As a control, the algorithm was applied to determine the virtual distance of single beads in the corrected color channels (supplementary figure 2a, right image). The average distance yielded about 20 nanometers (supplementary figure 2b, right panel), which can be attributed to the correction error of the method, since localization errors were in this case negligible (about five nanometers, supplementary figure 2b, left and center panel). For dual color TOCCSL data, detected positions were corrected and the virtual distances between AF488-scFV and AF647-scFV signals were determined. Signals within a distance of 200 nanometers were counted as co-localized, visually inspected and the total number of co-localization events, Nco-loc, for one experimental condition was normalized by the total number of signals, NscFv-Alexa Fluor 488 and NscFv-Alexa Fluor 647, yielding the average co-localization fraction,

fcoloc=2NcolocNscFvAF488+NscFvAF647.

Data Analysis Photon-Antibunching FCS experiments

PA/ FCS data were analyzed as published 19. The correlation function was fitted by the model functions

A+Bj=0exp(|tjTrep|τ)Cexp(tτ)

for the lag time t close to zero and

A+Bj=0exp(|tjTrep|τ)

for the infinite lag time t (figure 3b and supplementary figure 3a-c). Trep is the time between subsequent laser pulses, whereas A, B, C, A′, B′ and τ constitute fitting parameters with τ being the fluorescence life time. If there is no blinking state, as is true for the triplet state, the average number of molecules present in one cluster of molecules diffusing together can be calculated as follows:

nnt=BBC.

In order to investigate the triplet population of the fluorescence-labeled scFVs associated with the T-cell plasma membrane or in solution (serving as monomer reference), we calculated the FCS curves respectively and fitted them for two-dimensional diffusion with one triplet time model

G2(t)=α+βexp(tγ)1+tδ+σ,

and for three-dimensional diffusion with one triplet time model

G2(t)=a+bexp(tc)(1+t/d)1+t/e+f,

with α, β, γ, δ, σ, a, b, c, d, e and f serving as fitting parameters (figure 3c and supplementary figure 3a).

The triplet-state-corrected average number of molecules in each cluster is calculated by

n=BBCα+β+σα+σ(2D-diffusion)orn=BBCa+b+fa+f(3D-diffusion).

FRET efficiency Monte Carlo simulation

Matlab was used to simulate random x- and y-positions of pMHC molecules (TCRs) for various surface densities between 50-4000 (10-200) molecules per square micrometer. For every acceptor molecule, the distance d to the nearest donor molecule was calculated and used to determine the FRET efficiency E:

E=11+(dR0)6

With R0 = 5.1 (8.2) nm, the Förster radius for the Alexa Fluor 555 (568)/ Alexa Fluor 647 dye pair (supplementary table 1). FRET efficiencies of all molecules were added up and normalized by the number of acceptor molecules. For every surface density, the simulation was repeated 20 times and mean and standard error of the mean were determined.

Supplementary Material

Supplementary information

Acknowledgements

This work was supported by Austrian Science Fund (FWF) projects I953-B20 (MB), P 25775-B2 (KA, JBH), P27941-B28 (FB), the Vienna Science and Technology Fund (WWTF) project LS13-030 (FK, GJS). and by the Singapore National Medical Research Council under NMRC/CBRG/0064/2014 (NRJG).

Footnotes

Competing Financial Interests

The authors declare no competing financial interests

Author Contributions

MB and JBH conceived the project and wrote the manuscript. MB performed or supervised most of the imaging experiments. JBH performed the initial FRET-based and contributed to early TOCCSL experiments. MA performed early TOCCSL experiments and provided ideas. BKR performed TOCCSL and FRET-based experiments. FK and KA produced important imaging probes. FK and JG performed FRET-based experiments. ES built molecular models, analyzed anisotropy data and provided ideas. GJS provided imaging infrastructure and important ideas. HT performed photon-antibunching experiments and provided imaging infrastructure, respectively. FB, HS, NRG and SJD contributed key reagents and important ideas.

Data availability

The data that support the findings of this study are available from the authors on reasonable request, see author contributions for specific data sets.

Code availability

Every custom code employed for the analysis of all single color TOCCSL, dual color TOCCSL, PA/FCS and FRET-based experiments will be made accessible at the reader's request.

References

  • 1.Bromley SK, et al. The immunological synapse. Annu Rev Immunol. 2001;19:375–396. doi: 10.1146/annurev.immunol.19.1.375. [DOI] [PubMed] [Google Scholar]
  • 2.Davis MM, et al. Ligand recognition by alpha beta T cell receptors. Annual review of immunology. 1998;16:523–544. doi: 10.1146/annurev.immunol.16.1.523. [DOI] [PubMed] [Google Scholar]
  • 3.Irvine DJ, Purbhoo MA, Krogsgaard M, Davis MM. Direct observation of ligand recognition by T cells. Nature. 2002;419:845–849. doi: 10.1038/nature01076. [DOI] [PubMed] [Google Scholar]
  • 4.Purbhoo MA, Irvine DJ, Huppa JB, Davis MM. T cell killing does not require the formation of a stable mature immunological synapse. Nat Immunol. 2004;5:524–530. doi: 10.1038/ni1058. [DOI] [PubMed] [Google Scholar]
  • 5.Huppa JB, Davis MM. The interdisciplinary science of T-cell recognition. Advances in immunology. 2013;119:1–50. doi: 10.1016/B978-0-12-407707-2.00001-1. [DOI] [PubMed] [Google Scholar]
  • 6.Taylor MJ, Husain K, Gartner ZJ, Mayor S, Vale RD. A DNA-Based T Cell Receptor Reveals a Role for Receptor Clustering in Ligand Discrimination. Cell. 2017;169:108–119.:e120. doi: 10.1016/j.cell.2017.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Call ME, Pyrdol J, Wiedmann M, Wucherpfennig KW. The organizing principle in the formation of the T cell receptor-CD3 complex. Cell. 2002;111:967–979. doi: 10.1016/s0092-8674(02)01194-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Huppa JB, Ploegh HL. In vitro translation and assembly of a complete T cell receptor-CD3 complex. J Exp Med. 1997;186:393–403. doi: 10.1084/jem.186.3.393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Huppa JB, Ploegh HL. The alpha chain of the T cell antigen receptor is degraded in the cytosol. Immunity. 1997;7:113–122. doi: 10.1016/s1074-7613(00)80514-2. [DOI] [PubMed] [Google Scholar]
  • 10.Weissman AM, Samelson LE, Klausner RD. A new subunit of the human T-cell antigen receptor complex. Nature. 1986;324:480–482. doi: 10.1038/324480a0. [DOI] [PubMed] [Google Scholar]
  • 11.Exley M, Wileman T, Mueller B, Terhorst C. Evidence for multivalent structure of T-cell antigen receptor complex. Molecular Immunology. 1995;32:829–839. doi: 10.1016/0161-5890(95)00046-h. [DOI] [PubMed] [Google Scholar]
  • 12.Schamel WW, et al. Coexistence of multivalent and monovalent TCRs explains high sensitivity and wide range of response. J Exp Med. 2005;202:493–503. doi: 10.1084/jem.20042155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kumar R, et al. Increased sensitivity of antigen-experienced T cells through the enrichment of oligomeric T cell receptor complexes. Immunity. 2011;35:375–387. doi: 10.1016/j.immuni.2011.08.010. [DOI] [PubMed] [Google Scholar]
  • 14.James JR, et al. Single-molecule level analysis of the subunit composition of the T cell receptor on live T cells. Proc Natl Acad Sci U S A. 2007;104:17662–17667. doi: 10.1073/pnas.0700411104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Wilson BS, Pfeiffer JR, Surviladze Z, Gaudet EA, Oliver JM. High resolution mapping of mast cell membranes reveals primary and secondary domains of Fc(epsilon)RI and LAT. The Journal of cell biology. 2001;154:645–658. doi: 10.1083/jcb.200104049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Lillemeier BF, et al. TCR and Lat are expressed on separate protein islands on T cell membranes and concatenate during activation. Nat Immunol. 2010;11:90–96. doi: 10.1038/ni.1832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Tang Q, et al. The Src family kinase Fyn mediates signals induced by TCR antagonists. J Immunol. 2002;168:4480–4487. doi: 10.4049/jimmunol.168.9.4480. [DOI] [PubMed] [Google Scholar]
  • 18.Cochran JR, Cameron TO, Stern LJ. The relationship of MHC-peptide binding and T cell activation probed using chemically defined MHC class II oligomers. Immunity. 2000;12:241–250. doi: 10.1016/s1074-7613(00)80177-6. [DOI] [PubMed] [Google Scholar]
  • 19.Krogsgaard M, et al. Agonist/endogenous peptide-MHC heterodimers drive T cell activation and sensitivity. Nature. 2005;434:238–243. doi: 10.1038/nature03391. [DOI] [PubMed] [Google Scholar]
  • 20.Brown JH, et al. Three-dimensional structure of the human class II histocompatibility antigen HLA-DR1. Nature. 1993;364:33–39. doi: 10.1038/364033a0. [DOI] [PubMed] [Google Scholar]
  • 21.Stern LJ, et al. Crystal structure of the human class II MHC protein HLA-DR1 complexed with an influenza virus peptide. Nature. 1994;368:215–221. doi: 10.1038/368215a0. [DOI] [PubMed] [Google Scholar]
  • 22.Fremont DH, Hendrickson WA, Marrack P, Kappler J. Structures of an MHC class II molecule with covalently bound single peptides. Science. 1996;272:1001–1004. doi: 10.1126/science.272.5264.1001. [DOI] [PubMed] [Google Scholar]
  • 23.Reich Z, et al. Ligand-specific oligomerization of T-cell receptor molecules. Nature. 1997;387:617–620. doi: 10.1038/42500. [DOI] [PubMed] [Google Scholar]
  • 24.Baker BM, Wiley DC. alpha beta T cell receptor ligand-specific oligomerization revisited. Immunity. 2001;14:681–692. doi: 10.1016/s1074-7613(01)00160-1. [DOI] [PubMed] [Google Scholar]
  • 25.Moertelmaier M, Brameshuber M, Linimeier M, Schutz GJ, Stockinger H. Thinning out clusters while conserving stoichiometry of labeling. Applied Physics Letters. 2005;87 [Google Scholar]
  • 26.Brameshuber M, et al. Imaging of mobile long-lived nanoplatforms in the live cell plasma membrane. J Biol Chem. 2010;285:41765–41771. doi: 10.1074/jbc.M110.182121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Brameshuber M, Schutz GJ. Detection and quantification of biomolecular association in living cells using single-molecule microscopy. Methods Enzymol. 2012;505:159–186. doi: 10.1016/B978-0-12-388448-0.00017-6. [DOI] [PubMed] [Google Scholar]
  • 28.Huppa JB, et al. TCR-peptide-MHC interactions in situ show accelerated kinetics and increased affinity. Nature. 2010;463:963–967. doi: 10.1038/nature08746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Ruprecht V, Brameshuber M, Schutz GJ. Two-color single molecule tracking combined with photobleaching for the detection of rare molecular interactions in fluid biomembranes. Soft Matter. 2010;6:568–581. [Google Scholar]
  • 30.Xie J, et al. Photocrosslinkable pMHC monomers stain T cells specifically and cause ligand-bound TCRs to be 'preferentially' transported to the cSMAC. Nature immunology. 2012;13:674–680. doi: 10.1038/ni.2344. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kimble HJ, Dagenais M, Mandel L. Photon Antibunching in Resonance Fluorescence. Physical Review Letters. 1977;39:691–695. [Google Scholar]
  • 32.Sýkora J, et al. Exploring Fluorescence Antibunching in Solution To Determine the Stoichiometry of Molecular Complexes. Analytical Chemistry. 2007;79:4040–4049. doi: 10.1021/ac062024f. [DOI] [PubMed] [Google Scholar]
  • 33.Ta H, Kiel A, Wahl M, Herten DP. Experimental approach to extend the range for counting fluorescent molecules based on photon-antibunching. Phys Chem Chem Phys. 2010;12:10295–10300. doi: 10.1039/c0cp00363h. [DOI] [PubMed] [Google Scholar]
  • 34.Varma R, Campi G, Yokosuka T, Saito T, Dustin ML. T cell receptor-proximal signals are sustained in peripheral microclusters and terminated in the central supramolecular activation cluster. Immunity. 2006;25:117–127. doi: 10.1016/j.immuni.2006.04.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Demetriou M, Granovsky M, Quaggin S, Dennis JW. Negative regulation of T-cell activation and autoimmunity by Mgat5 N-glycosylation. Nature. 2001;409:733–739. doi: 10.1038/35055582. [DOI] [PubMed] [Google Scholar]
  • 36.Sherman E, et al. Functional nanoscale organization of signaling molecules downstream of the T cell antigen receptor. Immunity. 2011;35:705–720. doi: 10.1016/j.immuni.2011.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Birnbaum ME, et al. Molecular architecture of the alphabeta T cell receptor-CD3 complex. Proceedings of the National Academy of Sciences of the United States of America. 2014;111:17576–17581. doi: 10.1073/pnas.1420936111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Valitutti S, Muller S, Cella M, Padovan E, Lanzavecchia A. Serial triggering of many T-cell receptors by a few peptide-MHC complexes. Nature. 1995;375:148–151. doi: 10.1038/375148a0. [DOI] [PubMed] [Google Scholar]
  • 39.Kuhns MS, et al. Evidence for a functional sidedness to the alphabetaTCR. Proc Natl AcadSci US A. 2010;107:5094–5099. doi: 10.1073/pnas.1000925107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.James JR, et al. The T cell receptor triggering apparatus is composed of monovalent or monomeric proteins. J Biol Chem. 2011;286:31993–32001. doi: 10.1074/jbc.M111.219212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Davis SJ, van der Merwe PA. The structure and ligand interactions of CD2: implications for T-cell function. Immunol Today. 1996;17:177–187. doi: 10.1016/0167-5699(96)80617-7. [DOI] [PubMed] [Google Scholar]
  • 42.Chang VT, et al. Initiation of T cell signaling by CD45 segregation at 'close contacts'. Nat Immunol. 2016;17:574–582. doi: 10.1038/ni.3392. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Beddoe T, et al. Antigen ligation triggers a conformational change within the constant domain of the alphabeta T cell receptor. Immunity. 2009;30:777–788. doi: 10.1016/j.immuni.2009.03.018. [DOI] [PubMed] [Google Scholar]
  • 44.Gil D, Schamel WW, Montoya M, Sanchez-Madrid F, Alarcon B. Recruitment of Nck by CD3 epsilon reveals a ligand-induced conformational change essential for T cell receptor signaling and synapse formation. Cell. 2002;109:901–912. doi: 10.1016/s0092-8674(02)00799-7. [DOI] [PubMed] [Google Scholar]
  • 45.Lee MS, et al. A Mechanical Switch Couples T Cell Receptor Triggering to the Cytoplasmic Juxtamembrane Regions of CD3zetazeta. Immunity. 2015;43:227–239. doi: 10.1016/j.immuni.2015.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Wang F, Beck-Garcia K, Zorzin C, Schamel WW, Davis MM. Inhibition of T cell receptor signaling by cholesterol sulfate, a naturally occurring derivative of membrane cholesterol. Nat Immunol. 2016;17:844–850. doi: 10.1038/ni.3462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Swamy M, et al. A Cholesterol-Based Allostery Model of T Cell Receptor Phosphorylation. Immunity. 2016;44:1091–1101. doi: 10.1016/j.immuni.2016.04.011. [DOI] [PubMed] [Google Scholar]
  • 48.Kim ST, et al. The alphabeta T cell receptor is an anisotropic mechanosensor. The Journal of biological chemistry. 2009;284:31028–31037. doi: 10.1074/jbc.M109.052712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Liu B, Chen W, Evavold BD, Zhu C. Accumulation of dynamic catch bonds between TCR and agonist peptide-MHC triggers T cell signaling. Cell. 2014;157:357–368. doi: 10.1016/j.cell.2014.02.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Katz ZB, Novotna L, Blount A, Lillemeier BF. A cycle of Zap70 kinase activation and release from the TCR amplifies and disperses antigenic stimuli. Nat Immunol. 2017;18:86–95. doi: 10.1038/ni.3631. [DOI] [PMC free article] [PubMed] [Google Scholar]

References

  • 1.Huppa JB, Gleimer M, Sumen C, Davis MM. Continuous T cell receptor signaling required for synapse maintenance and full effector potential. Nat Immunol. 2003;4:749–755. doi: 10.1038/ni951. [DOI] [PubMed] [Google Scholar]
  • 2.Huse M, et al. Spatial and temporal dynamics of T cell receptor signaling with a photoactivatable agonist. Immunity. 2007;27:76–88. doi: 10.1016/j.immuni.2007.05.017. [DOI] [PubMed] [Google Scholar]
  • 3.Cai E, et al. Visualizing dynamic microvillar search and stabilization during ligand detection by T cells. Science. 2017;356 doi: 10.1126/science.aal3118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Huppa JB, et al. TCR-peptide-MHC interactions in situ show accelerated kinetics and increased affinity. Nature. 2010;463:963–967. doi: 10.1038/nature08746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Altman JD, Reay PA, Davis MM. Formation of functional peptide complexes of class II major histocompatibility complex proteins from subunits produced in Escherichia coli. Proc Natl Acad Sci U S A. 1993;90:10330–10334. doi: 10.1073/pnas.90.21.10330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Xie J, et al. Photocrosslinkable pMHC monomers stain T cells specifically and cause ligand-bound TCRs to be 'preferentially' transported to the cSMAC. Nature immunology. 2012;13:674–680. doi: 10.1038/ni.2344. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Axmann M, Schutz GJ, Huppa JB. Single Molecule Fluorescence Microscopy on Planar Supported Bilayers. Journal of visualized experiments : JoVE. 2015:e53158. doi: 10.3791/53158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Howarth M, et al. A monovalent streptavidin with a single femtomolar biotin binding site. Nat Methods. 2006;3:267–273. doi: 10.1038/NMETHXXX. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Fairhead M, Krndija D, Lowe ED, Howarth M. Plug-and-play pairing via defined divalent streptavidins. J Mol Biol. 2014;426:199–214. doi: 10.1016/j.jmb.2013.09.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Brameshuber M, et al. Imaging of mobile long-lived nanoplatforms in the live cell plasma membrane. J Biol Chem. 2010;285:41765–41771. doi: 10.1074/jbc.M110.182121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ruprecht V, Brameshuber M, Schutz GJ. Two-color single molecule tracking combined with photobleaching for the detection of rare molecular interactions in fluid biomembranes. Soft Matter. 2010;6:568–581. [Google Scholar]
  • 12.Harms GS, Sonnleitner M, Schutz GJ, Gruber HJ, Schmidt T. Single-molecule anisotropy imaging. Biophys J. 1999;77:2864–2870. doi: 10.1016/S0006-3495(99)77118-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Lakowicz JR. Principles of Fluorescence Spectroscopy. Springer; Boston, MA: 2006. [Google Scholar]
  • 14.Moertelmaier M, Brameshuber M, Linimeier M, Schutz GJ, Stockinger H. Thinning out clusters while conserving stoichiometry of labeling. Applied Physics Letters. 2005;87 [Google Scholar]
  • 15.Schmidt T, Schütz GJ, Gruber HJ, Schindler H. Local Stoichiometries Determined by Counting Individual Molecules. Analytical Chemistry. 1996;68:4397–4401. [Google Scholar]
  • 16.Brameshuber M, Schutz GJ. Detection and quantification of biomolecular association in living cells using single-molecule microscopy. Methods Enzymol. 2012;505:159–186. doi: 10.1016/B978-0-12-388448-0.00017-6. [DOI] [PubMed] [Google Scholar]
  • 17.Wieser S, Schutz GJ. Tracking single molecules in the live cell plasma membrane-Do's and Don't's. Methods. 2008;46:131–140. doi: 10.1016/j.ymeth.2008.06.010. [DOI] [PubMed] [Google Scholar]
  • 18.Schutz GJ, Schindler H, Schmidt T. Single-molecule microscopy on model membranes reveals anomalous diffusion. Biophys J. 1997;73:1073–1080. doi: 10.1016/S0006-3495(97)78139-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Sýkora J, et al. Exploring Fluorescence Antibunching in Solution To Determine the Stoichiometry of Molecular Complexes. Analytical Chemistry. 2007;79:4040–4049. doi: 10.1021/ac062024f. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary information

Data Availability Statement

The data that support the findings of this study are available from the authors on reasonable request, see author contributions for specific data sets.

Every custom code employed for the analysis of all single color TOCCSL, dual color TOCCSL, PA/FCS and FRET-based experiments will be made accessible at the reader's request.

RESOURCES