Force-Regulated In Situ TCR–Peptide-Bound MHC Class II Kinetics Determine Functions of CD4⁺ T Cells

This article is featured in In This Issue, p.3503

The T cell receptor (TCR)–peptide-bound MHC (pMHC) interaction has intrigued many because of its low three-dimensional (3D) affinity (K_d ∼1–100 μM) but high specificity and sensitivity (1–3). One of the most widely used methods to characterize the TCR–pMHC interaction is surface plasmon resonance (SPR) analysis with soluble molecules under flow-governing conditions (2–6). These analyses show that the affinity and/or dissociation off-rate predict the T cell response, with the agonist pMHC having a higher affinity and/or slower off-rate than the antagonist (4). However, the cellular environment where the TCR resides differs from that where SPR experiments are performed. The native TCR–pMHC binding occurs across the junctional gap between two surfaces, that is, at a two-dimensional (2D) interface, whereas in SPR, soluble molecules capable of 3D movement bind surface-immobilized countermolecules. Although attempts have been made to account for this difference by physical considerations (7) and mathematical models (8, 9), a more critical problem may be the use of engineered TCR constructs in SPR experiments, which ignores potential effects of other subunits of the TCR complex and their interactions with other cellular components (e.g., coreceptors or other membrane and cytoplasmic proteins) or movements of cellular structures (e.g., cell migration, actin retrograde flow, and membrane undulation) that could actively restrict or promote TCR binding to pMHC in a force-dependent manner (10–12). Thus, analyzing 2D TCR–pMHC binding kinetics may yield important insights to T cell biology as direct measurement is made in situ in the physiological membrane environment (13–16).

Two types of techniques have recently been used to measure the 2D kinetics of TCR–pMHC interactions: one based on fluorescent imaging, for example, Förster resonance energy transfer (FRET) (14) or tracking molecular diffusion (17, 18), and the other based on mechanical assays, for example, adhesion frequency (12, 13, 15, 16, 19–23), thermal fluctuation (12, 13, 16, 24), force-clamp (12, 24), and flow chamber (25) assays. Huppa et al. (14) studied FRET between lipid bilayer-anchored pMHC class II (pMHC-II) and CD4⁺ T blasts. The FRET signal change between the fluorophore-labeled MCC:I-E^k and 5c.c7 TCR allowed the calculation of binding parameters. Compared with 3D, their results showed much faster 2D binding and unbinding kinetics with no change when the CD4 was blocked. The same system was studied by Axmann et al. (17) using diffusion analysis and O’Donoghue et al. (18) using fluorescence imaging combined with diffusion analysis, but reported drastically different results. The former article reported fast off-rates in agreement with the results of Huppa et al. (14), whereas the latter article reported much slower off-rates.

Using the adhesion frequency and thermal fluctuation assays, our group found that 2D kinetics of TCR binding to a panel of peptides bound to MHC class I (MHC-I) were faster, had a much larger dynamic range, and correlated better with T cell activation than their 3D counterparts in several CD8⁺ systems (13, 15, 16). Furthermore, the characteristics of mouse CD8 binding to H-2K^b and H-2D^b and human CD8 binding to HLA-A2 were measured to define the basal CD8 affinity for MHC-I (16, 19). Moreover, an intriguing cooperation between TCR and CD8 was observed for >1s contacts with pMHC-I, suggesting the formation of trimeric TCR–pMHC-I–CD8 bonds to abruptly increase the bond number (16, 20, 23). In addition, we recently showed that mechanical force impacts TCR–pMHC-I dissociation kinetics in a peptide-specific manner to further amplify T cell discrimination (24) and to trigger calcium signaling in CD8⁺ T cells (22, 24). However, it is not known how much of these observations are applicable to pMHC-II–restricted TCRs on CD4⁺ T cells.

In this study, we used mechanical-based assays to analyze in situ kinetics of 3.L2 TCR binding to a well-characterized panel of peptides presented by MHC-II at zero-force and in a range of constant forces. Consistent with the MHC-I–restricted TCR systems, the zero-force 2D kinetics of this MHC-II–restricted TCR correlated better to the T cell functions than the 3D kinetics. With applied force on the TCR–pMHC-II bonds, agonist and weak agonist behaved as catch-slip bonds where bond lifetime (t_b) first increased, reached a maximum, then decreased with increasing force (24, 26). In contrast, antagonists behaved as slip-only bonds where t_b decreased monotonically with increasing force (24, 26). This peptide-specific force-dependency of TCR–pMHC-II dissociation increased the dynamic range of the t_b differences. CD4 binding was barely detectible and did not display cooperation with TCR in pMHC-II binding. Our results show that in situ kinetics and force-regulated bond dissociation discriminate ligands and determine functions in CD4⁺ T cells.

Transgenic 3.L2 mice were housed at the Emory University Department of Animal Resources facility, and experiments followed a protocol approved by the Institutional Animal Care and Use Committee of Emory University. The mouse expressed I-E^k–restricted 3.L2 TCR specific to murine hemoglobulin (Hb) epitopes 64–76. Naive T cells were purified via magnetic negative selection from 6- to 8-wk-old mouse spleens using either CD4⁺ or CD8⁺ T cell isolation kit (Miltenyi Biotec) according to the manufacturer’s instructions. Cells were washed and stored at room temperature for up to 24 h in R10 media, which consists of RPMI 1640 (Cellgro) supplemented with 10% FBS (Cellgro), 2 mM l-glutamine (Cellgro), 0.01M HEPES buffer (Cellgro), 100 μg/ml gentamicin (Cellgro), and 2 × 10⁻⁵ M 2-ME (Sigma-Aldrich).

Recombinant pMHC monomers were from the National Institutes of Health Tetramer Core Facility at Emory University. These consisted of the wild-type peptide Hb, or an altered peptide ligand, covalently tethered to a C-terminally biotinylated I-E^k (27). The peptides used in this study, with sequence and relative activity given in parentheses, included Hb (GKKVITAFNEGLK, agonist), T72 (GKKVITAFTEGLK, weak agonist), I72 (GKKVITAFIEGLK, antagonist), and A72 (GKKVITAFAEGLK, weak antagonist). In brief, agonist provides the maximal level of functional response, weak agonist induces maximal levels with a >10-fold higher peptide concentration, and antagonists block agonist activity with low (antagonist) or high (weak antagonist) concentrations without inducing response by themselves. OVA peptides 257–264 (SIINFEKL) bound to H-2K^b and moth cytochrome c (MCC) peptides 88–103 (ANERADLIAYLKQATK) bound to I-E^k were prepared in the same way for irrelevant control. For 3D SPR measurement, single-chain TCR (only the V region connected by a flexible linker) was expressed in Escherichia coli and refolded from inclusion bodies as described previously (28).

Detailed procedures have been described previously (13, 19, 20). In brief, human RBCs were isolated from the whole blood of healthy volunteers according to a protocol approved by the Institutional Review Board of the Georgia Institute of Technology. RBCs were biotinylated, conjugated with streptavidin, and incubated with biotinylated pMHC for adhesion frequency assay with a micropipette. In the case of BFP experiments, silanized beads were covalently linked to streptavidin-maleimide (Sigma-Aldrich) and then conjugated with subsaturating biotinylated pMHC.

Site densities of the TCR, CD4, and pMHC were measured by flow cytometry (13, 19, 20) using three fluorescent Abs: a PE-conjugated anti-mouse Vβ8.3 TCR (1B3.3; BD Pharmingen), a PE-conjugated anti-mouse CD4 (GK1.5; eBioscience), and a PE-conjugated anti-mouse I-E^k (17-3-3 [Santa Cruz]; 14-4-4s [BD Pharmingen]). Abs were used at 10 μg/ml concentration in 100 μl FACS buffer (PBS without calcium and magnesium, 5 mM EDTA, 1% BSA, 25 mM HEPES, 0.02% sodium azide) at 4°C for 30 min, fluorescent intensities were measured by the BD LSR II flow cytometer (BD Biosciences), and site densities were calculated by comparing to the BD QuantiBRITE PE standard beads (BD Biosciences).

Detailed procedures have been described previously (13, 19, 20). In brief, two micropipettes holding a T cell on one side and a pMHC-coated RBC on the other (Fig. 1A) were controlled to make repeated contacts with a constant area (A_c). From the presence or absence of RBC membrane deflection upon retraction after a contact time (t_c), the adhesion events in 50 contact cycles were enumerated to calculate adhesion frequency (P_a). The P_a versus t_c data (Fig. 1B) were fitted by the least-mean-square method to the following probabilistic kinetic Eq. (29)

to determine the effective 2D affinity (A_cK_a) and off-rate (k_off) using the independently measured receptor (m_r) and ligand (m_l) densities.

Nonspecific adhesion to 3.L2 T cells was controlled using 5-s contacts to: 1) unmodified RBCs (P_a < 1%); 2) biotinylated RBCs (P_a < 1% for 800 μM biotinylated concentration); 3) biotinylated RBCs linked to streptavidin (P_a < 1% for 4335 molecules/μm²); 4) biotinylated RBCs linked to streptavidin and coated with irrelevant pMHC-I (P_a ∼1% for 1548 molecules/μm² OVA_257–264:H-2K^b); 5) biotinylated RBCs linked to streptavidin and coated with irrelevant pMHC-II (P_a < 5% for 4140 molecules/μm² MCC_88–103:I-E^k); and 6) biotinylated RBCs linked to streptavidin and coated with antigenic pMHC-II (P_a ∼ 75% for 144 molecules/μm² TCR and 36 molecules/μm² Hb_64–76:I-E^k) (Fig. 1C).

Detailed procedures have been described previously (30, 31). In brief, the bond formation and dissociation were detected by the reduction and resumption of the thermal fluctuations of the BFP bead. A t_b spans from the instant of bond association to the instant of bond dissociation. Modeling the kinetic process as a single-step first-order dissociation of a single monomeric TCR–pMHC-II bond, the probability P_b of a bond formed at time 0 to remain intact at time t_b is

Taking a natural log linearizes the exponential function on the right-hand side of this equation. Plotting the data as ln(number of events with a lifetime > t_b) versus t_b allows for estimation of k_off from the negative slope of the line (Fig. 1D).

Detailed procedures have been described previously (24). In brief, t_b were measured by the force-clamp assay, which loads a bond to a preset force and holds the force constant to measure how long the bond lasts under that force. More than 300 lifetime events were measured for each ligand in a force range of 5–40pN segregated into five to seven bins. To affirm >95% single-bond events, pMHC-II densities on the beads were adjusted to compensate their affinities to keep the adhesion frequency <20% (29, 32).

Detailed procedures have been described previously (28). In brief, concentration series of single-chain TCR (scTCR) covering at least two orders of magnitude were injected in a BIAcore 2000 instrument using a flow rate of 30 μl/min over CM5 sensor chips (GE Healthcare) that were coupled with pMHC-II monomer. All sensorgrams were corrected for nonspecific binding by subtracting the response from a surface coupled with CLIP/I-E^k and were corrected for bulk flow effects by subtracting the response obtained from plain buffer injection.

Sensorgrams were fitted to a 1:1 Langmuir binding model using BIAevaluation version 4.1 to derive the 3D k_off, k_on, and K_a (= 1/K_d = k_on/k_off). K_d and maximum response (R_max) values from equilibrium binding analysis were obtained by plotting the equilibrium response (R_eq) at each concentration and fitting these data to a one-step binding model using GraphPad Prism version 6.0a for Macintosh (GraphPad Software, San Diego, CA). Scatchard plots were generated to confirm 1:1 binding stoichiometry by graphing R_eq/[TCR] versus R_eq.

T cells were incubated with 20 μg/ml of a CD4 blocking Ab (GK1.5; eBioscience) (14, 33) or 50 μg/ml of an anti-3.L2 clonotypic Ab (CAb) (34) in 100 μl for 30 min in 4°C and then tested in the experiment chamber with the presence of the same concentrations of the respective Abs.

It has previously been shown by tetramer staining (35) and FRET experiments (14) that blocking CD4 does not affect CD4⁺ T cell binding to pMHC-II. To confirm this result, we used the adhesion frequency assay because it has been proved to possess higher sensitivity (21) (Fig. 1). Treatment with the anti-3.L2 CAb completely abolished the ∼70% steady-state adhesion frequency measured without treatment (Fig. 2A). In sharp contrast, treatment with the anti-CD4 Ab (GK1.5) resulted in a binding curve indistinguishable from the curve without treatment. Both curves show single-stage binding, with a transient increase followed by a plateau, which can be well-fitted by Eq. 1 (Fig. 2A) and return comparable effective 2D affinity and off-rate values (Table I). This is in sharp contrast with binding of pMHC-I to CD8⁺ T cells, which exhibits a two-stage curve consisting of an initial plateau followed by a second, higher plateau after an ∼1 s delay that is not predicted by Eq. 1 but characteristic of the cooperative TCR–pMHC-I–CD8 trimolecular binding, with the second stage blockable by anti-CD8 Ab (16, 20, 23). These results indicate that CD4 binding is much weaker than TCR, and the two molecules do not cooperate to bind pMHC-II to increase the bond number under our experimental conditions.

To directly measure CD4 binding, we coated a noncognate MCC_88–103:I-E^k to RBCs at a high density of >4000 molecules/μm² and tested binding of these RBCs to 3.L2 T cells at a long 5-s t_c. The resulting low-adhesion frequency translated to an effective 2D affinity of A_cK_a < 7.0 × 10⁻⁸ μm⁴, one to two orders of magnitude lower than that of CD8–H-2K^b (2.8–5.7 × 10⁻⁶ μm⁴) and CD8–H-2D^b (0.1–0.5 × 10⁻⁶ μm⁴) (19). This indicates that under zero-force, I-E^k binding to CD4⁺ T cells is largely bimolecular with the TCR, but not CD4.

The 3.L2 TCR was tested using a series of well-characterized ligands, including the wild-type peptide Hb and three altered peptide ligands (APLs) with a single amino acid alteration at residue 72 that possess a range of potency to stimulate T cell activation (6, 36). To obtain midrange adhesion frequencies amendable to fitting by Eq. 1, different densities for Hb (agonist), T72 (weak agonist), I72 (antagonist), and A72 (weak antagonist) were coated on the RBCs to compensate for their distinct affinities. Fig. 1B shows binding curves resulted from using these RBCs to test naive 3.L2 T cells by the adhesion frequency assay. Eq. 1 was fit to these data to evaluate the effective 2D binding affinities for all ligands (Table I), including the weak antagonist A72 with the lowest potency that was not previously detectable using SPR (6). The measured values span over a 6-fold range (A_cK_a ∼3.4–19.2 × 10⁻⁵ μm⁴), with the agonist Hb being the highest and the weakest antagonist A72 the lowest (Fig. 2B). This is a broader dynamic range than the 3D affinities for the same TCR–pMHC-II interactions measured by SPR (Fig. 2B), but the gain is much smaller than the previously studied MHC-I systems (13, 15, 16). The 2D and 3D affinities show positive correlation (p < 0.05; Fig. 3A), but the high R² value (0.9) may be because of their much higher values for the agonist Hb:I-E^k than the other three APLs, which have similar values in 3D but are better resolved by the 2D analysis.

The 2D off-rates can also be derived from fitting the adhesion frequency data with Eq. 1 (Table I). In addition, the thermal fluctuation assay was also used (Fig. 1D), because its increased temporal resolution allows us to obtain more accurate off-rate values from analysis of single t_b measurements (30, 31) (Table I). Our previous study on the CD8⁺ OT1 T cells finds drastically different 2D and 3D off-rate values for the same ligands that show opposite ranked orders between 2D and 3D for the same panel of ligands, with the ligand potency correlating positively with 2D off-rate but negatively with 3D off-rate. By comparison, the off-rates measured in this study on the CD4⁺ 3.L2 T cells have similar values in both 2D and 3D, show the same ranked orders, and correspond to ligand potency ranging from the Hb to A72 (Fig. 2C). As such, the 2D and 3D off-rates are well correlated (R² = 0.87, p < 0.1; Fig. 3B). The 2D effective on-rates were calculated from multiplying the effective 2D affinity by off-rate, A_ck_on = A_cK_a × k_off (Fig. 2D, Table I). The 3D on-rates measured by SPR are also plotted on Fig. 2D for comparison, which show poor correlation with the 2D on-rates (R² = 0.14, p > 0.5; Fig. 3C). The previously measured peptide potency, that is, reciprocal of the ligand concentration required to generate 40% of radiolabeled B cell apoptosis (1/EC₄₀) measured by liquid scintillation counting (36), was plotted versus the 2D binding parameters. These plots show positive correlation with the effective 2D affinity (Fig. 3D), negative correlation with 2D off-rate (Fig. 3E), and lack of correlation with the effective 2D on-rate (Fig. 3F).

We used a BFP force-clamp assay (24, 37) to measure TCR–pMHC-II t_b under mechanical force. To focus our studies on the TCR interaction with pMHC-II, we took advantage of a known property of the 3.L2 TCR transgenic mice: a propensity to generate a reasonable number of CD8⁺, CD4⁻ mature T cells (38) (Fig. 4A). Confirming the minimal contribution of CD4, the CD8⁺ 3.L2 T cells showed similar effective 2D affinity to CD4⁺ T cells for the same panel of pMHC-II (Fig. 4B, Table I). Under force, I72 and A72 (antagonists) t_b decreases monotonically with increasing force (Fig. 5A). In sharp contrast, the more potent ligands displayed t_b that first increased, reached a maximum around 10 pN, and decreased as force continued to increase (Fig. 5A). These characteristics define TCR dissociation from Hb and T72 (agonists) as catch-slip bonds and I72 and A72 (antagonist) as slip-only bonds (24). The t_b distributions were plotted for the four ligands (Fig. 5B–E). It follows from Eq. 2 that the off-rates are equal to the negative reciprocal slopes of the data lines. As such, the t_b distributions decay less rapidly for agonists that exhibited catch-slip bonds than the antagonists that showed slip-only bonds. The greatest resolution in t_b occurred when 5–15 pN force was applied where agonist ligand t_b were prolonged mostly by the optimal force (Fig. 5C, 5D).

To examine how force impacts the correlation between t_b and ligand potency, we replotted the EC₄₀ values from Fig. 3 against t_b at 0 and 10pN (Fig. 6A). Force tilted the EC₄₀ versus t_b line at zero-force, making it less steep (Fig. 6A, blue line). At 10pN (the force at which the longest t_b were observed for agonists), the slope of the EC₄₀ versus t_b line become the shallowest. This expanded the dynamic range of the t_b differences, allowing greater resolution of ligand potency than found at zero-force (Fig. 6A). To further reveal how force amplifies the power of ligand discrimination, we plotted the ratio of t_b of Hb to that of different peptide ligands versus force (Fig. 6B). The t_b ratios of Hb to I72 (green) and A72 (blue) increased with force to reach maximum at 10–15 pN before decreasing. This is different from what was found for T72 (brown), thus increasing the power of discrimination between agonists and antagonists under force.

In situ TCR–pMHC kinetic analyses using mechanical-based assays have mostly been on CD8⁺ T cell systems (12, 13, 15, 16, 19, 20, 22–25, 39). The micropipette adhesion frequency assay has been used to measure the 2D affinities of polyclonal CD4⁺ T cells reactive to self-MOG and pathogen GP61 peptides (21), of 2D2 TCR for MOG or NFM peptides, and of SMARTA TCR for GP61 peptide (40, 41), all presented by MHC-II. Whereas these studies showed that the micropipette assay has much higher sensitivity than 3D pMHC-II tetramer staining and the 2D affinity correlates with the T cell biology, this article represents a more systematic study using the mechanical assays to correlate the force-dependent in situ TCR–pMHC-II binding kinetics to corresponding SPR measurements and with CD4⁺ T cell function, allowing us to compare similarities and contrast differences between the CD4⁺ and CD8⁺ T cell systems. Although caution should be used in the data interpretation because only a modest number of altered peptide ligands were tested, our conclusions are strengthened by the combined use of 2D measurements both at zero-force and in a range of tensile forces, which is more comprehensive because these measurements account for the impact of cellular environment and the effects of mechanical forces on in situ TCR–pMHC interactions.

Comparisons can be made on the relative ranges between the binding parameters and the T cell responses, the parameter values, and their ranked orders. TCR–pMHC interactions are highly elegant in that they do not merely result in an all-or-none response; rather, the T cell “reads” the information presented by the pMHC, transduces it into intracellular signals, and responds to the complex input with a broad set of phenotypic outcomes. For a response range of 6–7 logs measured by functional avidity or peptide concentration required to achieve a certain level of function, the range of binding parameters measured in 3D typically spans ≤1 log, and thereby requires substantial signal amplification that is attributed to the work of intracellular signaling cascades. The in situ 3.L2 TCR affinity range is similar to that of the 42F3 TCR (15), which spans 1 log, but much smaller than that of OT1 TCR (13), which spans 3 logs. Compared with the OT1 system (13), the 3.L2 system shows effective 2D affinity and on-rate values within the respective ranges of those of A2 (agonist) and G4 (weak agonist), whereas the 2D off-rate for agonist Hb is below the level of E1 (weak agonist/antagonist). Similar to the CD8⁺ systems (13, 15, 16), the in situ TCR affinities of the present CD4⁺ system correlate with, and correspond to the T cell response better than, the 3D TCR affinities. In contrast with the OT1 TCR whose off-rates are much faster in 2D than 3D and have opposite ranked orders in 2D and 3D (12, 13), the 3.L2 TCR has similar 2D and 3D off-rates that are of the same ranked orders, which correlate well with the T cell response. Unlike the 2D OT1 TCR on-rates that span a broad 4-log range and correlate well with T cell responses (13), the 2D 3.L2 TCR on-rates span only 1 log and correlate poorly with the biology. These comparisons shed light on the TCR–pMHC interaction parameter ranges and values, and impose important constraints to models that relate TCR recognition to T cell function.

Unlike the CD8–MHC-I interaction that is readily measurable by the adhesion frequency and thermal fluctuation assays with kinetic parameters comparable with those of TCR interaction with antagonists (16, 19), CD4–MHC-II interaction was barely detectable by the same mechanical-based assays, allowing us to set only an upper bound for its effective 2D affinity with unmeasurable off-rate. Despite the much lower affinity of CD8 than that of TCR for agonist pMHC-I, CD8 and TCR cooperate to form a trimeric bond with pMHC-I to synergistically increase the bond number in a signaling-dependent manner (16, 20, 23). This phenomenon was not observed for the 3.L2 TCR in the present CD4⁺ T cell system. Future studies will determine what different properties of the CD4 and CD8 coreceptors may cause this distinction and whether other MHC-II–restricted TCRs or other conditions exist that may enable cooperation between the CD4 and TCR for synergistic MHC-II binding.

In three recent studies, TCR–pMHC-I t_b have been measured over a range of forces in several systems (24, 25, 39). The role of physical force in regulating TCR–pMHC-I dissociation kinetics and in triggering T cell activation has been highlighted in the findings that force elicits agonist-specific catch bonds (24, 39) that correlate to the triggering of calcium signals in T cells (24). Similar to the OT1 system, this study has demonstrated that the 3.L2 TCR also forms catch-slip bonds with agonist, and slip-only bonds with antagonist, peptides presented by MHC-II. Interestingly, the force ranges where catch bonds are observed are comparable for the OT1 (24), N15 (39), and 3.L2 TCRs, with the t_b reaching maxima around 10–15 pN. Similar to the OT1 system (24), force greatly amplifies the power of Ag discrimination by the 3.L2 TCR in a CD4⁺ T cell system. This is despite the fact that 10 pN force reverses the ranked orders of the off-rates of the OT1 TCR dissociating from a panel of pMHC-I measured at zero-force, whereas the ranked orders of the off-rates of the 3.L2 TCR dissociating from a panel of pMHC-II remain the same at 0 and 10 pN. These findings provide a new parameter dimension relevant to T cell mechanoimmunology, which is nonexistent for 3D kinetic analysis because 3D binding takes place under stress-free conditions. Future studies will likely determine how TCR engagement may induce intracellular forces that may exert on the TCR–pMHC bonds to regulate their dissociation and further trigger T cell activation.

We thank Laurel Lawrence for maintaining the mouse colony; Larissa Doudy for T cell purification; Drs. Lindsay Edwards, Joe Sabatino, Baoyu Liu, Wei Chen, and Zhenhai Li for support with experiments and helpful discussions; and the National Institutes of Health Tetramer Core Facility at Emory University for providing the MHC monomers.

The authors have no financial conflicts of interest.