Yeast display allows for the external presentation and screening of large libraries of genetically altered proteins (43, 44). Single chain T cell receptor (scTCR) contain linked variable TCR α and β domains but lack constant domains and any signaling domains. The scTCR constructs for LLO56 (residues 1–116) and LLO118 (residues 1–120) (Invitrogen) consist of the mature Vβ domain, a 13-aa linker (DAKKDAAKKDDAS) (50), followed by the mature Vα domain (LLO118 residues 1–112 or LLO56 residues 1–113), and an N-terminal HA tag (PYDVPDYA). To display scTCRs on yeast, the constructs were placed in pCT302 (NheI and BglII) (Addgene plasmid # 41845; http://n2t.net/addgene:41845; RRID : Addgene_41845) (51). Stability clones were selected from scTCR transcripts replicated by error-prone PCR (Standard Taq, New England BioLabs, B9014S) (44). Affinity libraries were generated using site directed mutagenesis of 5 amino acids in the CDR3β region and splicing by overlap extension (SOE) PCR (44, 52) using LLO118 and LLO56 specific primers (Q5 High Fidelity DNA Polymerase, New England BioLabs M0491) (Supplementary Materials).

To generate yeast libraries, 150 ml cultures of growth phase EBY100 yeast were collected and washed twice with 50 ml ice-cold water and once with ice-cold electroporation buffer (1 M Sorbitol/1 mM CaCl2) then resuspended in 0.1 M LiAc/10 mM DTT and incubated at 30°C and 225 rpm for 30 min (53). Cells were washed with 50 ml electroporation buffer, resuspended in 200 µl electroporation buffer and aliquoted with digested pCT302 backbone (NheI and BglII, 1,250 ng) and inserted (6,250 ng) into 0.2 mm gap cuvettes then electroporated (2.5 kV and 25 µF). Cells were allowed to recover for 1 h in 4 ml 1 M sorbitol:YPD media (1:1) and were resuspended in SD-CAA media and incubated for 2–3 days at 30°C before quantification. Stability and affinity library sizes ranged from 1.1x10⁷ to 1.9x10⁹.

Libraries calculated to have at least 10 copies of each clone were placed in 5 ml SG-CAA media for 36–48 h to induce scTCR expression (54). To select stability clones, yeast libraries were incubated with either 2 µg/ml anti-mouse TCR Vα2 or anti-mouse TCR Vβ2 phycoerythrin-conjugated antibodies (BioLegend, clone B20.1 and B20.6, respectively) in 5 ml PBS 1% BSA for 2 h at 4°C, washed with 15 ml PBS 1% BSA and stained with 50 µl anti-PE MicroBeads in 2 ml PBS 1% BSA (Millitenyi 130-048-801) for 20 min at 4°C. Labeled clones expressing properly folded Vα or Vβ were positively selected in magnetic LS columns (Millitenyi 130-042-401). Selected cells were grown in 3 ml SD-CAA media (48 h) before induction in SG-CAA (36–48 h). Each library was subjected to three rounds of growth and sorting, and the most stable clone identified via flow cytometry (BD Accuri C6). Stability clones were used as templates for subsequent stability or affinity libraries.

To select affinity clones, induced yeast libraries were incubated with tetramer (LLO_190-201/I-A^b) (I-A(b)CC (NEKYAQAYPNVS), NIH 22201), and sorted like stability clones. To isolate high affinity clones, libraries were exposed to an increasingly strict temperature and incubation regimen. Initially, libraries were subjected to high concentrations of tetramer (13.0 µg/ml), high temperatures (37°C), and long incubation times (3 h), and in later rounds, combinations of lower tetramer concentrations (3.25 µg/ml), lower temperatures (RT or 4°C), and shorter incubation times (1 h) were used to isolate the clones with highest affinity. Each library was column sorted three times. Isolated clones with increased tetramer binding were identified via flow cytometry (BD Accuri C6).

Each affinity and stability clone K_D was determined through tetramer dissociation (55). Aliquots of 1x10⁶ induced cells were stained with 100 µl of various concentrations of LLO_190-201/I-A^b tetramer (0.152 nM to 12.16 nM) for 1.5 h at room temperature and quantified via flow cytometry. Tetramer binding was assessed as MFI of positive population and normalized to the highest recorded MFI using FlowJo. K_D was defined as 50% maximum binding concentration (55).

Half-life (t_1/2) was determined by staining 3x10⁶ cells of each affinity clone with 6.5 µg/ml of tetramer for 1.5 h at room temperature (56). Samples were washed three times in PBS 1% BSA to remove excess tetramer. Following an initial timepoint measurement, 90 µl of 0.1 µg/ml or 1 µg/ml anti-mouse MHC class II (I-A/I-E) (clone: M5/114.15.2, eBioscience) was added and the decrease of tetramer binding was quantified at various time points (2, 5, 10, 15, 20, 30, 45, and 60 min) by placing 10 µl of cells into 90 µl of buffer and running immediately on the flow cytometer.

The following protocol was modified from Garcia et al[61]. Briefly, scTCR constructs were cloned into pET28a (Novagen) using NcoI and SacI restriction sites. Constructs were expressed in BL21 T7 Express E. coli (New England Biolabs) and protein expression was induced for 4 h (0.4 mM isopropyl β-D-thiogalactoside). Cells were lysed with 1 mg/ml lysozyme (ThermoFisher Scientific), 5 mM MgCl2, 1 µl/ml DNase I (Promega), 1% Triton-X 100, and 10 mM dithiothreitol followed by two rounds of sonification (Branson Digital Sonifer) for 1 min at 0.5 s alternations at 40% power. Fifty to 200 mg of inclusion body slurry was dissolved in 1 ml of 7M GnHCl and 10 mM beta-mercaptoethanol. Four hundred ml of 2 M GnHCl, 50 mM Tris-HCl, 2mM GSH, 0.2 mM GSSG, and 0.1% NaAz were dripped into dissolved inclusion bodies for 2–4 h at 4°C. Then 2–2.5 L of 200 mM NaCl, 50 mM Tris-HCl, and 0.1% NaAz were dripped for 24 h (1.5 ml/min speed) at 4°C. Following an additional 24 h spinning at 4°C, the refolded TCR solution was vacuum filtered with 0.22 µm PES membranes (Olympus Plastics), and then concentrated in an Amicon 8400 unit (Ultracel 10 kdal Ultrafiltration Discs) under 55psi N₂. Once the volume was reduced to 50–100 ml of refolded scTCRs, the samples were again filtered with 0.45 µm CA-membrane and GF prefilter syringe filter and purified by FPLC (AKTAstart) on a HisTrap column (GE Life Sciences). Purified scTCRs were concentrated using Amicon centrifugal filters (Ultra 4 10k) and quantified by Pierce BCA Protein Assay kit (Thermo Scientific).

BLI experiments were performed with an Octet RED96. Streptavidin (SA) biosensors (FortéBio) were hydrated and equilibrated in 1x HEPES buffered saline (HBS, 50 mM HEPES, 150 mM NalCl, pH 7.2), 2mM CaCl₂, 1 mM MgCl₂, 1 mg/ml milk, 0.1% Tween, and 0.02% NaN₃. SA sensors were loaded with 2.0 µg/ml biotinylated LLO_190-201/I-A^b monomer or DQB_187-101/I-A^b monomer to 1.0–2.0 nm. Loaded biosensors were equilibrated in assay buffer until baseline was achieved. scTCR association was probed in wells with assay buffer (stability clones 2, 1, 0.5, 0.25, 0.125, 0.061 µM; affinity clones 800, 400, 200, 100, 50, and 25 nM, or 20, 10, 5, 2.5, 1.25, and 0.625 nM) with a blank reference-subtraction well for 400–600 s. Ideal concentration range spanned one log above and below the K_D where possible; however, this range had to be optimized depending on the sensitivity of the assay, and on the amount of protein available. Matching of sample and baseline imidazole and milk concentrations (through serial dilution of sample buffer into baseline wells) was critical for detection of scTCR binding. Blocking with bovine serum albumin increased non-specific binding while milk efficiently blocked NSB. Dissociation was observed in baseline assay buffer (600–1,200 s). Assays were run at 30°C with a plate shake speed of 1,000 rpm.

Data was collected at 5 Hz, using 20-point signal averaging and analyzed using custom kinetic analysis. Due to non-specific binding at the later stages of the association and dissociation steps, K_D was calculated by extracting and selecting the data points from the initial association to determine k_obs (2–100 s depending on the affinity of the constructs), plotting concentration vs rate, and then plotting those slopes against scTCR concentration and estimating k_assoc from the slope. k_dissoc is the slope of concentration vs rate of the dissociation step data (2–100 s depending on the affinity of the constructs). K_D was determined by dividing k_diss/k_assoc and t_1/2 = ln2/k_D.

All 58^-/- T cell hybridoma cell lines were cultured in RPMI 1640, 10% FBS, 2g/L NaHCO₃ (23.8 mM), HEPES (4.2mM), L-glutamine (3.24 mM), 1% Penn-strep and split 1:5 or 1:10 every 2–3 days. Platinum Ecotrophic cells (Plat E) were cultured in DMEM, 10% FBS, 1% pen-strep, 1 µg/ml puromycin, and 10 µg/ml blasticidin and split 1:4 every other day.

Affinity mutations were cloned into four possible constructs: full length TCRs (flTCRs), and three TCR-single chain signaling formats based on chimeric antigen receptor (CAR) formats (second generation 4-1BB and CD28 CARs, and third generation 4-1BB/CD28 CAR). Inserts were cloned into pMSCV-IRES-GFP II (pMIGII) (Addgene plasmid # 52107; http://n2t.net/addgene:52107; RRID : Addgene_52107) using MfeI and XhoI (GenScript) (57). All constructs were led by a Kozak sequence and either Vα2 signal peptide (MDKILTASFLLLGLHLAGVSGQ) and an additional Vβ2 signal peptide (MWQFCILCLCVLMASVATD) for flTCRs or high affinity M33 3^rd gen CAR signal peptide (MLLALLPVLGIHFVLRDAQA) for all TCR-SCS CAR constructs (58). flTCR constructs have a P2A cleavage domain (GSGATNFSLLKQAGDVEENPG) (59) between Cα2 and Vβ2 domains. All TCR-SCS and flTCR were linked to GFP by an IRES domain and GFP expression mirrors TCR construct expression in cell lines.

Vectors were transfected into Plat E packaging cells grown overnight in 6-well plates with TransIT-VirusGEN (Mirus, MIR 6703). Forty-eight hours later, 1 ml of viral supernatant was mixed with 1 ml of 1x10⁶ 58^-/- CD4^- or 58^-/- CD4⁺ cells in a 6-well plate and spinfected for 2 h at 30°C at 1,000 G (acceleration 6, brake 2). After 48 h recovery, clones Vβ2, Vα2, and GFP⁺ expression was checked by flow cytometry. Clones with under 85–90% GFP expression were sorted 1–3 times with magnetic LS columns (Miltenyi Biotec, 130-042-401) using 10 µl Vβ2-PE antibodies and 10 µl anti-PE MicroBeads (Miltenyi Biotec, 130-048-801) per manufacturer specifications. Clones were checked for TCR expression after each sort round. CD4T⁺ and CD4T⁺ Δbind (60) were cloned into pMIGII (MfeI/XhoI) and retrovirally transfected into existing 58^-/- CD4^- flTCR and TCR-SCS clones and sorted for >95% CD4 expression (CD4 PE-Cy7, GK1.5, Biolegend) by flow sorting (BD FACSAria II). Twenty-five thousand cells were stained with respective antibodies or tetramer for all affinity and stability measurements and measured with flow cytometry (BD Accuri).

2.8x10⁴ T cell hybridoma clones were incubated with 2.8x10⁵ splenocytes (1:10) isolated from BL6.C57 mice with varying amounts (10^-8 M to 10^-3 M) of peptide (LLO_190-205, GenScript) in 75 µl 58^-/- media in 96 well plate for 24 h. IL-2 production was measured using an IL-2 ELISA kit (KIT) and measured on a microplate reader. This study was approved and carried out in accordance with principles of the Basel Declaration and recommendations of Brigham Young University’s Institutional Animal Care and Use Committee (IACUC protocol #18-0708).

Statistical analysis was performed via one-way ANOVA with Tukey’s multiple comparison test (p < 0.05 was significant, no alpha adjustments required). Half-life (t_1/2) was determined by linear regression between time point 0 and the time point where no tetramer binding was detected (56). To determine the K_D, we fit the data with a non-linear curve, based on one site-specific binding kinetics (55). EC₅₀ was determined with Sigmoidal, 4PL, X is log(concentration) least squares fit. Standard deviation is reported for each value. All analyses were conducted in GraphPad Prism.

Murine transgenic helper T cells LLO56 and LLO118 bear TCRs, which recognize the same naturally occurring Listeria monocytogenes peptide (LLO_190-205) presented on MHCII (I-A^b). The LLO56 and LLO118 TCR bind cognate pMHC with similar affinity (27.4 µM and 28.3 µM, respectively), yet have unique primary and secondary responses to TCR stimulus (summarized in Supplemental Table S1) (41, 42). LLO56 and LLO118 differ from each other by 15 amino acids located in the complementarity determining regions (CDR) CDR3β (amino acids 96–108, 111, 116, 118), CDR2β (aa52), and CDR3α (aa93) regions (Supplemental Figure S1). To further elucidate the effects of TCR-pMHCII affinity on CD4⁺ T cell activation, the variable regions of LLO56 and LLO118 (Figure 1A) were used as templates for generating a panel of single-chain TCRs (scTCRs) with low (wild type), intermediate, and high affinities. scTCR libraries generated by random mutagenesis and expressed via yeast surface display (Figure 1B) were selected for protein folding stability through magnetic column sorting (Figure 1C). scTCR expression levels vary according to yeast cell cycle stage and can result in multiple peaks. The left peak is the non-displaying fraction and there can be intermediate and high displaying yeast (61). Vβ2 stability mutations were conserved between constructs while Vα2 stability mutations clustered in known stability hotspots (Supplemental Figure S1). To generate affinity mutants, five amino acids in the stability mutants LLO56_low and LLO118_low CDR3β region were mutated by site directed mutagenesis and selected for improved binding affinity for LLO_190-201/I-A^b tetramers by magnetic column sorting. Increases in scTCR affinity cannot be explained by increases in scTCR expression, as HA and TCRα and TCRβ antibody binding remained the same across each experiment (Figure 1B). Additionally, none of the isolated stability or affinity mutants bound significantly to a non-target peptide tetramer (DQB1_87-101/I-A^b), indicating that the increase in tetramer binding is due to peptide-specific binding and not increased affinity for I-A^b alone (Figure 1D). Affinity mutant LLO56_int with four CDR3β mutations (Figure 1E) bound LLO_190-201/I-A^b 1.5 log better than stability mutant LLO56_low (Figure 1C). Affinity mutant LLO118_high bound to the LLO_190-201/I-A^b tetramer 1.0-log better than affinity mutant LLO118_int and 2.5-log better than stability mutant LLO118_low (Figure 1C). LLO118_int had three CDR3β mutations and LLO118A_high had five additional CDR3β mutations (Figure 1E). While LLO56_int and LLO118_high mutants were used as templates for mutant libraries of the complementary determining region 3 of the α chain (CDR3α), no further clones with increased-affinity for LLO_190-201/I-A^b tetramer were isolated, suggesting that CDR3β is primarily responsible for LLO_190-201 peptide interactions for these specific TCRs. It is important to note that the stability clones initially isolated from yeast libraries relied on a frameshift mutation at the stop codon that added 19-amino acids from the yeast expression vector to the carboxy end of Vα2 (RSDNNSVDVTKSTLFPPYF). While LLO56_low and LLO56_int successfully retained stability and affinity gains without the 19 amino acids, several attempts to create new LLO118 stability and affinity clones without the additional amino acids were unsuccessful. Therefore, the stabilizing 19 amino acids were maintained for LLO118 clones. Other studies have utilized TCR formats that express constant domains in order to maintain scTCR folding while adding additional affinity mutations to the TCRs, therefore this observation was not unexpected (62–64).

The multivalent binding avidity of each clone was determined by LLO_190-201/I-A^b tetramer titration (150 pM–15.00 nM) of scTCR expressed on yeast (Figure 2A). Avidities ranged from the highest clone LLO118_high (7.30 nM), to intermediate avidity clones LLO56_int (39.20 nM) and LLO118_int (44.80 nM) (Figures 2A, B and Table 1). Stability clones LLO56_low and LLO118_low were excluded from these analyses because binding was undetectable even at the highest concentrations of LLO_190-201/I-A^b tetramer (Figure 2B). Tetramer decay analysis of clones displayed on yeast determined that the multivalent half-life for LLO118_high (t_½ = 165 min, r² = 0.76) was 165-times longer than LLO118_int and LLO56_int (t_½ = ~1 min each, r² = 0.97 each) suggesting that the increased avidity of LLO118_high is predominantly due to a lengthened off-rate (Figure 2C). A second round of tetramer decay with lower levels of MHC inhibiting-antibody better resolved the half-lives of LLO118_int (t_½ = 6.7 mins, r² = 0.97) and LLO56_int (t_½ = 3.5 mins, r² = 0.98) (Figure 2D), indicating that LLO118_int has a longer dissociation rate than LLO56_int. The resulting panel of TCRs provides a range of tetramer avidities ranging from high to low (Figure 2E).

While tetramer avidity measurements may be more physiologically relevant as multiple TCR-pMHCs interact simultaneously during T cell activation, TCR-pMHC affinity measurements provide a standard measurement to compare between TCR systems. Therefore, TCR:pMHC affinity was measured by quantifying the interaction of monomeric refolded scTCR with monomeric LLO_190-201/I-A^b via bio-layer interferometry. Due to non-specific binding at the later stages of the association and dissociation steps, the K_D was calculated manually by extracting the data from the early measurements; k_obs slopes ((Figure 2F) were plotted against scTCR concentration (Figure 2G) and k_assoc estimated from the slope. k_diss is the slope of dissociation graphs (Figure 2H). K_D was determined by dividing k_diss/k_assoc. LLO118_high (20.0 ± 13.9 nM) K_D was 215-fold higher than LLO118_low (4.3 ± 0.7 µM) (Table 1). Intriguingly, while LLO118_int and LLO56_int avidity measurements were similar, their affinity measurements were markedly different (20-fold). LLO118_int (1.3 ± 0.3 µM) was only 3-fold higher affinity than LLO118_low and LLO56_int (66.2 ± 39.8 nM) was 43-fold higher than LLO56_low (3.8 ± 1.3 µM) (Table 1).

To quantitatively assess the effects of TCR-pMHC affinity, CD4, and construct format on helper T cell activation, TCR constructs were retrovirally transduced into murine T cell hybridomas, 58^-/- CD4^- (CD4^-) and 58^-/- CD4⁺ (CD4⁺), which do not express endogenous TCRs. 58^-/- T cell hybridoma cell lines have been a useful cell line for examining TCR kinetics and IL-2 production prior to primary cell line observations (47, 49, 65, 66). LLO56_low and LLO56_int were placed in the three TCR-SCSs formats, and LLO56_WT and LLO56_int were placed in flTCR constructs without stability mutations (Figures 3A, B, Supplemental Figures S2–S5). Because of the necessity of the additional 19 amino acids, LO118_low, LLO118_int, and LLO118_high affinity changes were not transferred to flTCR constructs. LLO118_low, LLO118_int, and LLO118_high were placed in three TCR-SCSs formats based on commonly used second and third generation chimeric antigen receptor (CAR) formats (Figures 3A, B, Supplemental Figures S2–S4). The transduced cell lines were sorted with anti-Vβ2 antibodies via magnetic column selection for >85% GFP⁺ and TCR expression (Figure 3C).

TCR stable surface expression varied by individual constructs. As assessed by Vβ2 expression, flTCR constructs were less stably expressed than all TCR-SCS constructs perhaps due to CD3 subunit availability, while TCR-SCS CD28 constructs were the most stably expressed format for both LLO56 and LLO118 constructs (Figure 3D). While most construct expression was equitable between CD4^- and CD4⁺ cell lines, TCR-SCS 4-1BB constructs had the most expression variability between constructs as CD4 expression destabilized LLO118_low and LLO118_int TCR-SCS 4-1BB expression (p = 0.0018 and p = 0.0429, respectively) (Figure 3D). CD4^- LLO118_high 4-1BB constructs were less stable than CD4^- LLO118_low 4-1BB and LLO118_int 4-1BB (p = 0.0030 and p = 0.0002, respectively) (Figure 3D). CD4⁺ LLO118_high 4-1BB was also less stable than either LLO118_low or LLO118_int (p = 0.0285 and p = 0.0493) (Figure 3D). Overall CD4 expression did not significantly destabilize 3^rd gen constructs except LLO118_low (p = 0.0401) and LLO56_int where CD4 stabilized VB2 expression (p = 0.0109) (Figure 3D). GFP expression does not correlate with expression differences of the constructs for each format between CD4^- and CD4⁺ expression nor expression differences between TCR (Supplemental Figures S6).

Tetramer titrations were used to approximate the avidity of each flTCR or TCR-SCS construct. Intriguingly, the intracellular format strongly influenced the avidity of each intermediate and high affinity TCR construct (Figure 4A). There is no clear link across all clones between stable Vβ2 expression and construct avidity, although the most stable constructs—TCR-SCS CD28—did have the highest apparent avidity (LLO56_int and LLO118_high) (Figure 4A). Overall, CD4 expression (dotted lines) did not affect the avidity of the constructs, excepting LLO118_int and LLO118_high 3^rd gen constructs where CD4 lessened and heightened avidity, respectively (Figure 4A). The MFI measured for each clone at 10^-8 M (a non-saturated concentration) were used to compare avidity differences between affinity clones. LLO56 4-1BB, 3^rd gen and flTCR constructs had no significant differences between LLO56_low and LLO56_int (Figure 4B). This may be due to the small affinity differences between LLO56_low and LLO56_int as measured in tetramer and bio-layer interferometry assays. However, LLO118 3^rd gen constructs also did not show affinity-dependent avidity changes, thus intracellular signaling domains may also affect the avidity of extracellular scTCRs. There were significant avidity differences for LLO118 4-1BB clones; CD4^- LLO118_high 4-1BB had significantly better avidity than its cognate CD4⁺ pairing (p = 0.0004), and was also significantly higher than CD4^- LLO118_low and LLO118_int 4-1BB (p = 0.0003 and p = 0.0092, respectively) (Figure 4B). Additionally, TCR-SCS CD28 constructs for both LLO56 and LLO118, which are the most stably expressed constructs (Figure 3D), showed increased MFI by increasing TCR affinity (Figure 4B). LLO56_int CD28 had significantly greater avidity than LLO56_low CD28 (CD4^- p = 0.0122 and CD4⁺ p= 0.0086), as did LLO118_high CD28 compared to LLO118_low (CD4^- p = 0.0129 and CD4⁺ p= 0.0113) (Figure 4B). IL-2 production is not correlated with GFP intensity (Supplemental Figure S6). Taken together, while there is no systematic correlation, this data suggests that construct stability may influence avidity measurements, as CD28 clones had the highest stability and avidity, and confirms that generally, CD4 does not affect perceived avidity.

To assess the effects of TCR-pMHCII affinity, CD4 expression, and format on T cell activation we measured IL-2 expression in response to increasing agonist peptide concentrations. As anticipated, LLO56_low flTCR IL-2 production improved with CD4 expression, but CD4 expression unexpectedly reduced IL-2 production for LLO56_int flTCR (Figure 5A). Despite the inconsistent role of CD4, flTCRs produced significantly more IL-2 at all affinity levels (Figure 5A) and were at least 1-log fold more sensitive to peptide than all TCR-SCSs (Figures 5B–D). IL-2 production for CD4^- clones rose with increased TCR affinity for most constructs except 3^rd gen constructs; LLO56_int 3^rd gen failed to produce more cytokines than LLO56_low 3^rd gen (Figure 5B) and LLO118_high 3rd gen that produced less IL-2 than LLO118_int 3^rd gen (Figure 5C). This pattern of uneven gains across affinity and base TCR was also observed for 4-1BB constructs (Figures 5D, E); while LLO56 4-1BB did see gains across affinity (Figure 5D), LLO118 4-1BB constructs had limited affinity gains across the affinity gradient (Figure 5E). CD4^- LLO56_int CD28 and CD4^- LL0118_high CD28 produced more IL-2 than other TCR-SCS constructs which suggested that their heightened stable expression may promote IL-2 production (Figures 5F, G). As noted in low affinity scTCR clones (LLO56_low 3^rd gen and CD28, and LLO118_low 3^rd gen) (Figures 5B, C, F), CD4⁺ and CD4^- clones may also respond uniquely across antigenic concentrations, however this is likely an artifact due to variability or the limits of detection.

While CD4 promoted the activation of all low affinity clones, it unexpectedly suppressed IL-2 production for all intermediate and high affinity constructs (Figure 5). The magnitude of IL-2 suppression is greatly dependent on whether the construct was a flTCR or TCR-SCS construct. For example, while LLO56_low flTCR IL-2 production was assisted by CD4 expression, LLO56_int flTCR IL-2 production was reduced 2.2-fold (p = 0.0707) at 10^-3 M peptide stimulation (Supplemental Figure S7). In contrast, only one TCR-SCS had such a mild IL-2 reduction. CD4⁺ LL056_int 4-1BB IL-2 production was reduced by 2.5-fold (p = 0.1551) (Supplemental Figure S7). The IL-2 production for the other intermediate and high affinity TCR-SCS constructs was intermediately reduced for LLO56_int CD28 (6.4-fold, p = 0.0104), and severely reduced for LLO118_int 3^rd gen (28.1-fold, p = 0.0004), LLO118_int CD28 (21.2-fold, p = 0.0400), LLO118_high 3^rd gen (16.5-fold, p = 0.0051), and LLO118_high CD28 (25.9-fold, p < 0.0001) (Supplemental Figure S7). Peptide sensitivity, defined as the lowest concentration where IL-2 response exceeds baseline IL-2 production, was equitable between CD4^- and CD4⁺ for constructs LLO56_int flTCR, LLO56 3^rd gen, and LLO56_int 4-1BB (Figures 5A, D, Table 2), but delayed 1-log fold for LL056_int CD28 and at least 2-log fold for all LLO118_int and LLO118_high constructs (Figures 5C, E–G, Table 2). This suggests that CD4 reduced peptide sensitivity for most TCR-SCS constructs, possibly in a TCR-dependent manner.

Lck is an early proximal signaling kinase that colocalizes to the cytoplasmic domain of CD4 (68, 69). If Lck is poorly recruited to the TCR-pMHCII synapse, then T cell activation may be diminished (49). We hypothesized that our high affinity clones may poorly recruit CD4-Lck to the immunological synapse, decreasing activation, and therefore reducing IL-2 production as observed in CD4⁺ intermediate and high affinity clones. To parse out the potential contributions of CD4-Lck sequestration, CD4-MHCII interactions, and any CD4-dependent inhibition, we expressed a selection of our flTCR and TCR-SCS clones in four 58^-/- T cell hybridoma lines (49, 60). LLO56 TCR-SCS 3^rd gen and LLO118 TCR SCS 4-1BB clones were dropped due to their poor performance in the first IL-2 tests. The 58^-/- CD4^- T cell hybridoma cell line (CD4^-) lack CD4, which allows Lck to interact freely with the TCR-pMHCII complex and nullifies CD4-MHCII interactions (Figure 6A). The 58^-/- CD4⁺ T cell hybridoma cell line (CD4⁺) has wild type CD4 which sequesters Lck to its cytoplasmic tail and binds to MHCII (Figure 6B). 58^-/- CD4T⁺ T cell hybridoma line (CD4T⁺) is truncated C-terminally (maintains amino acids 1-421) which allow Lck to colocalize but not bind to CD4 while CD4 still binds to MHCII (Figure 6C) (60, 70). CD4T expressed in both T cells and hybridomas has been documented in many sources to not bind Lck (60, 70, 71) and is suggested to produce IL-2 in a Lck-independent manner (60, 70, 71). Previous work has also demonstrated in cells that, while CD4TΔbind does not bind MHCII, it still contains the cytoplasmic tail necessary for binding Lck (72, 73).Finally, 58^-/- CD4T⁺ Δbind (CD4T⁺ Δbind) frees Lck and is mutated to prevent CD4 binding to MHCII by altering residues 68–73 from KGVLIR to DGDSDS (Figure 6D) (60).

CD4T⁺ and CD4T⁺ Δbind constructs were retrovirally transduced into existing CD4^- T cell hybridomas containing TCR-SCS or flTCR constructs and the clones were sorted for GFP, TCR, and CD4 expression by flow sorting. CD4^- hybridomas did not express CD4, and there was consistent CD4 expression between the various CD4⁺ clones (Supplemental Figure S8) while GFP levels varied (Supplemental Figure S9A). TCR surface expression was consistent across cell lines for most TCR constructs, except LLO56_low and LLO56_int CD28 clones which were most stably expressed in CD4⁺ cells (p < 0.0001 and p < 0.0001, respectively) (Supplemental Figure S9B). Similarly, avidity measured by tetramer was mainly consistent between clones except for LLO56_low 4-1BB where CD4^- had higher avidity than all CD4⁺ clones (p = 0.0014), and LLO118_int CD28 where CD4⁺ clone had the highest avidity (p = 0.0063) (Supplemental Figure S9C).

As expected, LLO56_low flTCR IL-2 production was promoted by the presence of CD4-MHCII interactions (CD4⁺ p = 0.0105 and CD4T⁺ p = 0.0014) and the absence of CD4-MHCII interaction in the CD4^- and CD4T⁺ Δbind abrogated LLO56_low flTCR IL-2 production (Figure 6E). CD4 Lck-sequestration did not affect LLO56_low flTCR IL-2 production since the CD4⁺ and CD4T⁺ clones responded the similarly to antigen (Figure 6E). In contrast, intermediate affinity LLO56_int flTCR IL-2 production is inhibited by CD4 Lck-sequestration, as CD4⁺ produced significantly less IL-2 than CD4T⁺ (p = 0.0162) (Figure 6F). However, the more striking phenotype is LLO56_int flTCR CD4-dependent inhibition, as IL-2 production is significantly reduced by the presence of CD4 in any form (Figure 6F). CD4^- LLO56_int flTCR cells produced significantly more IL-2 than any clone expressing CD4 (CD4⁺ p < 0.0001, CD4T⁺ p = 0.0019, and CD4T⁺ Δbind p = 0.0002) (Figure 6F). Furthermore, CD4-MHCII interaction was not a significant contributor to intermediate affinity flTCR IL-2 production as there was no significant change in IL-2 production between the CD4T⁺ and CD4T⁺ Δbind clones (Figure 6F).

Low affinity TCR-SCS clones LLO56_low CD28 (Figure 6G) and LLO56_low 4-1BB (Figure 6H) were low IL-2 producers and the role of CD4 was conflicting as all CD4 iterations inhibited IL-2 production for 4-1BB but promoted IL-2 production for CD28 whether or not CD4 binds to MHCII. It was also difficult to draw conclusions about Lck-sequestration for low affinity TCR-SCS constructs due to low levels of IL2 production. Intermediate TCR-SCS clones LLO56_int CD28 (Figure 6I), LLO56_int 4-1BB (Figure 6J), LLO118_int CD28 (Figure 6K), and LLO118_int 3^rd gen (Figure 6L) had a unique ubiquitous phenotype comparable to the phenotype described for intermediate affinity flTCR clones. IL-2 production was most reduced when CD4 sequestered Lck in the CD4⁺ clones (Figures 6I–L). However, intermediate TCR-SCS CD4T⁺ constructs produced the most IL-2, indicating unrestricted Lck promotes the greatest T cell activation (Figures 6I–L). CD4T⁺ Δbind compared to CD4T⁺ significantly reduced intermediate TCR-SCS construct IL-2 production to CD4^- levels suggesting that CD4-MHCII binding supports IL-2 production for intermediate TCR-SCS affinity (Figures 6I–L). Noticeably, high affinity LLO118 TCR-SCSs followed the same inhibition patterns seen for LLO56_int flTCR where inhibition by Lck sequestration and CD4 presence was not significantly affected by MHCII-CD4 binding (Figures 6M, N). Taken together, these data indicate that flTCRs and TCR-SCS have independent affinity thresholds for the inhibitory effects of Lck-sequestration and CD4-dependent inhibition, and the activation promoting effects of CD4-MHCII interactions (summarized in Table 3). Thus, IL-2 inhibition is affected by CD4-Lck sequestration, CD4-pMHCII interaction, and by a CD4-dependent mechanism in an affinity- and format-dependent manner.