CD4 Inhibits Helper T Cell Activation at Lower Affinity Threshold for Full-Length T Cell Receptors Than Single Chain Signaling Constructs

CD4+ T cells are crucial for effective repression and elimination of cancer cells. Despite a paucity of CD4+ T cell receptor (TCR) clinical studies, CD4+ T cells are primed to become important therapeutics as they help circumvent tumor antigen escape and guide multifactorial immune responses. However, because CD8+ T cells directly kill tumor cells, most research has focused on the attributes of CD8+ TCRs. Less is known about how TCR affinity and CD4 expression affect CD4+ T cell activation in full length TCR (flTCR) and TCR single chain signaling (TCR-SCS) formats. Here, we generated an affinity panel of TCRs from CD4+ T cells and expressed them in flTCR and three TCR-SCS formats modeled after chimeric antigen receptors (CARs) to understand the contributions of TCR-pMHCII affinity, TCR format, and coreceptor CD4 interactions on CD4+ T cell activation. Strikingly, the coreceptor CD4 inhibited intermediate and high affinity TCR-construct activation by Lck-dependent and -independent mechanisms. These inhibition mechanisms had unique affinity thresholds dependent on the TCR format. Intracellular construct formats affected the tetramer staining for each TCR as well as IL-2 production. IL-2 production was promoted by increased TCR-pMHCII affinity and the flTCR format. Thus, CD4+ T cell therapy development should consider TCR affinity, CD4 expression, and construct format.


INTRODUCTION
CD4 + T cells are critical for tumor elimination through both indirect and direct mechanisms. Indirectly, CD4 + T cells target tumor cells by activating tumor-killing cells such as CD8 + T cells, macrophages, B cells, and natural killer cells (1)(2)(3)(4). CD4 + T cells have direct cytotoxic effects against tumor cells that express major histocompatibility complex II (MHCII) (1)(2)(3)(4) and direct CD4 + T cell responses are less toxic to the patient than a CD8 + T cell response, especially when responding to overexpressed tumor associated antigens (TAA) (5). The presence of tumor-specific CD4 + T cells is correlated with improved patient survival following vaccination with cancer-associated peptides whether or not they are directly involved in tumor suppression (6)(7)(8). Furthermore, CD4 + T cells can sustain an immune response when CD8 + -specific antigens are lost which otherwise might result in tumor escape (9). Despite these clear benefits, only one published clinical study (10) focuses on the immunotherapeutic benefits of CD4 + T cell receptors (TCRs) (10,11).
CD4 + T cells are activated by interactions between the TCR and its cognate peptide presented on MHCII (pMHCII) (12). TCRs can detect a single amino acid change and distinguish between self-proteins and mutated neoantigens (11), uniquely suiting TCR-based therapies for specific tumor targeting. Furthermore, unlike antibody-based chimeric antigen receptors (CARs), which are limited to extracellular targets, TCRs can target intracellular antigens presented by MHC molecules (11). To rationally design optimal targeting strategies, it is essential to understand how the TCR:pMHC interaction impacts T cell responses. The relationship between TCR affinity and T cell activation is complex, but in general, T cell functional activity correlates with TCR binding affinity for pMHC (13)(14)(15)(16)(17)(18)(19). However, there are important nuances to this general theme. For example, tumor-associated antigens may be skewed towards lower-affinity clones due to thymic negative selection (20,21), even the lowest-affinity TCRs can induce T cell proliferation, cytokine production and memory formation (19,22). On the other end of the spectrum, high affinity TCRs have been shown to enhance immune responses in some cases (23) and attenuate responses in others (24)(25)(26)(27)(28)(29), with some reports showing evidence of an affinity threshold beyond which increased affinity does not impact the magnitude of the response (17,18). An additional consideration is that even when high-affinity TCRs are capable of heightened cytotoxicity and tumor control, these TCRs may be predisposed to autoimmunity (30). Thus, the optimal affinities for TCRs engineered against tumor-specific peptides may lie within a low or intermediate affinity (14,(24)(25)(26)(30)(31)(32)(33)(34)(35)(36). As most affinity studies to date have focused on CD8 + TCRs, CD4 + T cell affinity thresholds are less well characterized.
The role of the CD4 coreceptor is an important consideration when associating TCR-pMHCII affinity to CD4 + T cell activation. CD4 binds to MHCII as part of the TCR complex and contributes to proximal TCR signaling, proving especially critical for T cell function when cognate pMHC ligands are limiting (<30 complexes) (37). TCR signaling dependence on CD4 is affected by the quality of TCR:pMHCII interaction and is unnecessary upon stimulation with optimal ligands (38). Thus, CD4 may be restricted to improving the TCR dwell time on pMHCII for lower affinity interactions (39). As CD4 + TCRs can function in natural killer cell lines without CD4 (40), CD4 may not have as great of an effect on T cell activation as CD8, particularly with high affinity TCRs.
To determine how TCR-pMHCII affinity and CD4 coreceptor interactions affect CD4 + T cell activation, we examined activation of the CD4 transgenic murine T cells LLO118 and LLO56 that are stimulated by the same Listeria monocytogenes epitope. These TCRs differ by 15 amino acids and recognize the LLO 190-205 peptide presented by the MHCII molecule I-A b with similar affinity (41,42). LLO118 has a more robust primary response and LLO56 has a more robust secondary response, indicating that TCR affinity is not the only parameter affecting activation in these cells. To examine the role of affinity in the activation responses of LLO56 and LLO118, we engineered an affinity panel of CD4 + TCRs (ranging from 4 µM to 200 nM) using yeast display (43,44). After characterizing their affinity and avidity, the activation characteristics of two low affinity clones, two intermediate affinity clones, and one high affinity clone were examined in the full length TCR (flTCR) format or in three TCR-SCS CAR formats (CD28-and 4-1BB-based second generation CARs, and CD28/4-1BB third generation CAR). T cell receptor single-chain signaling chimeric antigen receptors (TCR-SCS CARs) are an exciting potential therapeutic option and as CD4 + T cells are potent responders to cancer, we sought to understand how CD4 + TCRs respond to a variety of affinities. TCR-SCSs constructs avoid mispairing with endogenous TCR chains, which is an inherent risk for engineered flTCRs (45). CARs also produce more cytokines and are activated by higher antigen densities than flTCRs and may be more likely to ignore healthy cells with low amounts of TAAs, which may improve clinical outcomes (46)(47)(48).
We found that increased TCR affinity promotes production of IL-2 regardless of flTCR or TCR-SCS format. The flTCRs are more responsive to lower amounts of peptide stimulation, and contrary to CD8 + TCR findings (49), produce more cytokine than TCR-SCSs. While there are some observable trends dependent on second and third generation TCR-SCS CAR format, IL-2 production varies depending on whether the TCRs were engineered from the LLO56 or LLO118 TCRs. CD4 promotes the activation of low affinity flTCRs and TCR-SCSs, but CD4 is inhibitory in intermediate affinity flTCR and high affinity TCR-SCS CARs. The flTCR reaches CD4 inhibition at a lower affinity than TCR-SCSs, suggesting that flTCRs perceive a stronger initial activation signal. These findings suggest that therapeutic CD4 TCR development should consider construct features, TCR affinity, and coreceptor activation contributions when choosing or engineering therapeutic TCRs and cell lines.

Stability Clone Selection
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 Va2 or antimouse TCR Vb2 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 Va or Vb 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.

Affinity Clone Selection
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).

Tetramer Dissociation
Each affinity and stability clone K D was determined through tetramer dissociation (55). Aliquots of 1x10 6 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).

Tetramer Decay
Half-life (t 1/2 ) was determined by staining 3x10 6 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.

scTCR Expression, Refolding, and Purification
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 b-Dthiogalactoside). 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 N 2 . 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).

Bio-Layer Interferometry (BLI)
BLI experiments were performed with an Octet RED96. Streptavidin (SA) biosensors (ForteBio) were hydrated and equilibrated in 1x HEPES buffered saline (HBS, 50 mM HEPES, 150 mM NalCl, pH 7.2), 2mM CaCl 2 , 1 mM MgCl 2 , 1 mg/ml milk, 0.1% Tween, and 0.02% NaN 3 . 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 .
T Cell Hybridoma Peptide-Specific Activation and IL-2 Measurement 2.8x10 4 T cell hybridoma clones were incubated with 2.8x10 5 splenocytes (1:10) isolated from BL6.C57 mice with varying amounts (10 -8 M to 10 -3 M) of peptide  , 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
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). EC 50 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.

Yeast Displayed TCR Panel Has Varied Affinities
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) CDR3b (amino acids 96-108, 111, 116, 118), CDR2b (aa52), and CDR3a (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). Vb2 stability mutations were conserved between constructs while Va2 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 CDR3b 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 TCRa and TCRb 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 CDR3b 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 2.5log better than stability mutant LLO118 low ( Figure 1C). LLO118 int had three CDR3b mutations and LLO118A high had five additional CDR3b mutations ( Figure 1E).    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, r 2 = 0.76) was 165-times longer than LLO118 int and LLO56 int (t ½ =~1 min each, r 2 = 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, r 2 = 0.97) and LLO56 int (t ½ = 3.5 mins, r 2 = 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).

Construct Format Impacts Surface Expression and pMHCII-Affinity Independently
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-Vb2 antibodies via magnetic column selection for >85% GFP + and TCR expression ( Figure 3C).
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 Vb2 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 affinitydependent 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 (CD4p = 0.0122 and CD4 + p= 0.0086), as did LLO118 high CD28 compared to LLO118 low (CD4p = 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.

CD4 Inhibits High Affinity TCR IL-2 Production
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 CD4clones 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 CD4clones may also respond uniquely across antigenic concentrations, however this is likely an artifact due to variability or the limits of detection.

Lck Sequestration by CD4 Inhibits Some TCR IL-2 Production
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 CD4TDbind does not bind MHCII, it still contains the cytoplasmic tail necessary for binding Lck (72,73).Finally, 58 -/-CD4T + Dbind (CD4T + Dbind) 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 + Dbind 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. CD4hybridomas 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 CD4had 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).
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 + Dbind compared to CD4T + significantly reduced intermediate TCR-SCS construct IL-2 production to CD4levels 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 formatdependent manner. IL-2 production data interpretation broken down into base TCR (LLO56 or LLO118), construct, and TCR-pMHCII affinity. "-" indicates that the condition inhibits or does not promote IL-2 production, "NA" indicates that effects on IL-2 production were "not appreciable", and "+" indicates that the condition promotes or least does not inhibit IL-2 production. Bolded interior boxes highlight the phenotype shared by intermediate affinity flTCR and high affinity TCR-SCS clones.

DISCUSSION
Here, we engineered and characterized a panel of MHCIIspecific TCRs with increasing pMHC affinity in order to interrogate the relationships between TCR format, TCR-pMHCII affinity, and the coreceptor CD4 on CD4 + T cell activation. In addition to the generation of a high affinity MHCII-dependent TCR model, we identify a CD4-dependent phenotype potentially relevant for cancer-immunotherapeutic development and show that high affinity flTCRs outperform TCR-SCS formats, that TCR-SCS format effects on T cell activation are more dependent on the TCR than the TCR-SCS format, and that CD4 can inhibit both flTCR and TCR-SCS activation in an Lck dependent and independent fashion. This study utilized 58 -/-T cell hybridomas as a proxy for T cell activation activity. While not as physiologically relevant as using primary T cells, this system has the advantage of enabling the survey of T cell activation characteristics for multiple constructs as we have done here and has been frequently used as a springboard for further exploration of high affinity TCRs in primary T cells (47,65,74,75). The contributions of these factors were assessed using IL-2 production, which is a proxy, but not a complete indication of T cell activation. flTCRs produced more IL-2 than all TCR-SCS constructs at each affinity level and IL-2 production generally increased with rising TCR affinity for all constructs. In low affinity TCRs, CD4 enhanced IL-2 production for both flTCR and TCR-SCS formats. For intermediate or high affinity TCR clones, IL-2 production was abrogated by CD4-Lck sequestration and an unknown CD4-dependent mechanism. These effects, activation promotion by increased affinity and CD4-MHCII, or activation suppression by Lck-sequestration and CD4 itself, had unique affinity thresholds that are dependent on construct type (flTCR or TCR-SCS). Lck sequestration affected activation for all intermediate and high affinity constructs, while CD4-MHCII ceased to promote activation and CD4-dependent inhibition repressed IL-2 production at unique affinity thresholds for flTCR constructs (intermediate affinity) and TCR-SCS constructs (high affinity). It is possible that Lck sequestration or the constructs themselves, have unique and important effects on other activation markers (such as the early activation markers CD25, pLCk, pCD3z, pERK, CD69), inhibitory markers (such as PD1 and LAG3), tonic signaling, T cell proliferation and effector function, which should be investigated in primary T cells in the future. The balance of free unbound Lck and coreceptor-bound Lck affects T cell developmental fate, and T cell responsiveness in the periphery. Following colocalization to the TCR, CD4 signals via Lck bound to its cytoplasmic tail (68,69). Lck phosphorylates immune-receptor tyrosine-based activating motifs (ITAMs) of the CD3 subunits of the TCR complex, which then initiates other early signaling machinery of the T cell (69,76,77). Bound and unbound Lck signal independently and can alter T cell development and function (78,79). During thymic selection, the intracellular coreceptor-bound or unbound state of Lck determines whether ab TCRs are MHC-restricted or independent (78). Lck association with coreceptor proteins determines MHC restriction (78), and coreceptor-Lck binding stoichiometry is the limiting factor for signaling during selection (80). In particular, CD8, which binds Lck more preferentially than CD4, has a greater effect on TCR selection and increases CD8 + T cell reactivity to low affinity and self-reactive antigen compared to CD4 + T cells (80). In both mature CD8 + T cells and T cell hybridomas, free Lck has higher mobility, more activating Y394 phosphorylation, higher kinase activity, and mediated higher T cell activation compared to coreceptor-bound Lck (81). Additionally, during activation, TCR-CD3 is first phosphorylated by unbound-Lck followed by MHC-dependent CD3-CD8 interaction and the less activated coreceptor-bound Lck (82)(83)(84). CD4-bound Lck activation may be reliant on a mechanism distinct from CD4-free Lck activation, which is likely mediated by tyrosine-protein kinase Fyn and may obscure mechanism comparison (37,85,86). Additionally, it is also possible that CD4 may function differently in T cells expressing native TCRs or CAR cytoplasmic domains. However, despite these potential complications, CD4-Lck-dependent inhibition could occur in two fashions. First, optimal TCR affinity-mediated signaling is dependent on fine-tuning the intensity and duration of the Lck phosphorylation cascade and high affinity TCRs may have early intense Lck phosphorylation resulting in acute transient activation (87). Conversely, if CD4 is not recruited to the TCR, it could sequester Lck away from the activation complex, which prevents the activation phosphorylation cascade thereby attenuating T cell activation (49). The first option suggests that all high-affinity TCR signaling would be attenuated regardless of whether Lck was interacting with CD4; however, IL-2 output reduction in the presence of CD4-Lck sequestration is clearly demonstrated by our intermediate and high affinity CD4+ T cell hybridoma clones. It is also possible that with an increase in affinity and the subsequent decrease in off-rate or increase in half-life, CD4-Lck fails to cycle through the TCR-pMHC synapse, thereby decreasing CD3 phosphorylation and thus downstream activation. Signaling activation is affected by both TCR-pMHCII dwell time and CD4-Lck interactions (70,88,89). CD4 increases TCR signaling on low-affinity pMHCII by increasing TCR-CD3 dwell time (39). CD4 dwell time on pMHCII is proportional, yet faster, to TCR dwell time, suggesting that TCR:pMHCII interaction kinetics would directly affect the duration that CD4 molecules cycle through the immunological synapse in a processive-like manner (88). Additionally, compared to coreceptor-bound Lck, CD4-free Lck is phosphorylated more at its Y394 activation site, with higher kinase activity and mobility (81); thus, it may be that if CD4-free Lck is prevented from interacting with the immunological synapse, activation may be reduced. TCR-pMHCII interactions are highly ordered and uniform, increasing the likelihood that the spatial relationship between Lck and the ITAMs of the TCR-SCS or flTCRs are consistent. Thus, kinetic factors, such as TCR-pMHCII affinity would greatly influence the stability of the macrocomplex and consequently the duration of Lck interactions with the ITAMs (39,90). These kinetics alone could explain the drop in activation observed for our high-affinity, slow off-rate TCR clones. To support this idea, CD8 also acts as a dominant negative inhibitor for ligands that do not recruit fresh CD8 to the TCR-CD3 complex (49).
Previous research suggests that CD4 can send an inhibitory signal independent of Lck via post activation antibody-mediated ligation, which attenuates IL-2 production and ongoing activated T cell response (91). This response was also observed in a clonal variant expressing a form of CD4 unable to associate with Lck, suggesting that CD4 has independent inhibitory or regulatory function (91). Furthermore, CD4-mediated inhibition has also been observed during CD4-MHCII interactions leading to a decrease in IL-2 mRNA (91). While we did not seek the source for our Lck-independent CD4 inhibition nor acquire IL-2 mRNA levels, we noted that there was an affinity threshold for this behavior that was independent of MHC interaction, and therefore may be a unique mechanism to that reported in Chervin et al. (49). The affinity threshold for this Lckindependent CD4 inhibition was lower for flTCR (intermediate affinity) than TCR-SCS (high affinity). This may be due to the signaling power of each construct: flTCR-CD3 complexes have 10 ITAMs with 20 tyrosine residues available for phosphorylation, whereas TCR-SCS domains have only 3 ITAMs and 6 tyrosine residues (47,92,93). The increased availability of ITAMs per activated Lck may also explain why LLO56 int flTCR experienced less IL-2 production inhibition in the presence of CD4-more signal per Lck molecule despite CD4-Lck movement restriction. Whether SCS-TCRs function as dimers is unclear and remains a topic of study (67,94,95). It is also curious that the CD4-MHCII interaction supports activation in intermediate affinity TCR-SCS clones, suggesting that while CD4 may not contribute to the overall affinity of TCR-SCS constructs, it may stabilize the interaction between TCR-pMHCII or provide an additional Lck-independent activation signal. The increased interaction stability is more likely as high affinity TCR-SCS IL-2 production is not significantly improved when CD4 interacts with MHCII, suggesting high affinity constructs likely have stable interactions independent of CD4 contributions. Taken together this data suggests an affinity threshold where, up to a point, increased time for CD4-MHCII interactions improves TCR-dependent signaling when it is not Lck-limited, but after a certain affinity point, increased dwell time slows TCR-dependent signaling and positive benefits of CD4-MHCII interactions become redundant.
In addition to the stability challenges presented by scTCR format, the TCR-SCS intracellular format also affected the stability of each TCR. TCR-SCS CD28 format was more stably expressed than other TCR-SCS or flTCR formats, and as noted in other studies, the enhanced surface expression of TCR-SCS CD28 formats via increased stability may explain their improved avidity and T cell activation (96)(97)(98). However, it is difficult to ascertain whether the increased IL-2 production of TCR-SCS CD28 is due to enhanced stable surface expression or the innate characteristics of CD28-intracellular signaling domains. As observed in numerous antibody-based CAR studies comparing CD28 domains to 4-1BB domains, intracellular signaling domains differentially impact multifactorial T cell response characteristics, including cytokine production (99). For example, CD28-CAR constructs, which can directly bind Lck, are well known for their Lck-binding-dependent enhanced IL-2 production, increased tonic signaling, and subsequent T cell exhaustion compared to 4-1BB CARs (100)(101)(102)(103). Thus, the observed increase in IL-2 production for TCR-SCS CD28 constructs may be attributable to the innate characteristics of CD28-intracellular signaling domains rather than increased stable surface expression. As CD28-CARs phosphorylate CD3 more quickly yet do not exceed the levels of CD3 phosphorylation exhibited by 4-1BB CARs, this may be due to signaling intensity (101). Additionally, because CD28 recruits Lck to lipid rafts where it associates with CD4, CD28 may be better able to recruit Lck (104,105). While TCR-SCS 3 rd generation constructs had mixed activation success and overall reduced cytokine production compared to TCR-SCS CD28 constructs, this may be attributable to 3 rd generation CAR T cells improved expansion and persistence and may mimic some characteristics of 4-1BB CAR T cells, like reduced cytokine production (106,107). Our TCR-SCS CD28 constructs demonstrated similarities to antibody-based CD28 CARs, including enhanced tonic signaling in some clones, suggesting that TCR SCS CD28 may also have increased T cell exhaustion. However, unlike CD28 CARs, CD4 expression ameliorated tonic signaling in our TCR-SCS constructs. It will be important to examine the role of CD4 in primary T cells to determine if CD4 prevents exhaustion in clones expressing TCR-SCS CD28 constructs.
CD4 + T cells are promising newcomers to immunotherapy. CD4 + TCRs convey exquisite target specificity and direct robust immune responses through indirect mechanisms that avoid tumor antigen escape. While much development and thought has been devoted to the activation benefits and off-target effects of increased TCR-pMHC affinity, especially for CD8 + TCRs, further TCR-therapeutic development should give consideration to the unique affinity thresholds of TCR-SCS and flTCR formats and the potential inhibitory effects of CD4.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by Brigham Young University's Institutional Animal Care and Use Committee (IACUC protocol #18-0708).

AUTHOR CONTRIBUTIONS
DJ conceived the experiments. DJ, WM, SM, JF, JH, and TO conducted the experiments. KW and KC provided experimental advice. SP cloned the initial LLO118 and LLO56 constructs in the yeast display constructs and permitted use of previously published data. DJ analyzed data and wrote the manuscript. All authors reviewed the manuscript. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank National Institutes of Health Tetramer Core Facility at Emory University for providing the MHC tetramers. We thank Dr. Claudia Tellez Freitas for her help with initial cell culture training and advice. We thank Dr. Dan Harris and Dr. David Kranz (University of Illinois Urbana-Champaign) for Platinum Ecotrophic, 58 -/and 58 -/-CD4 + cell hybridoma lines, TCR-SCS constructs and flTCR expression advice. We also thank Dr. Michael Kunz (University of Arizona) for CD4T + and CD4T + D bind constructs and for his excellent critical review of the manuscript.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2020.561889/ full#supplementary-material SUPPLEMENTARY FIGURE 1 | LLO118 and LLO56 single-chain TCRs stabilizing mutations. Wild type templates (LLO118 WT and LLO56 WT ) compared to stabilized single-chain TCR (scTCR) templates (LLO118 low and LLO56 low ). The original LLO56 WT scTCR template included mutations of the amino acid K42bG, H36aY and S74aT (highlighted gray) known to enhance surface display levels in related TCRs (108). Stability mutations selected by random mutagenesis and directed evolution are marked in red. Boxed amino acids show joint LLO118 low and LLO56 low selection (K42bG and T93bA), and mutations in another known stability hotspot (L45aI and I49aM) are unmarked. LLO118 low independently selected I115aK, and LLO56 low selected T93bA and S9aT.
SUPPLEMENTARY FIGURE 2 | Protein sequence map for 4-1BB SCS-TCR formats for (A) LLO56 low and LLO56 int , and (B) LLO118 low , LLO118 int , and LLO118 high . Protein fragments are delineated by color bar above the amino acid sequence. Stability mutations (highlighted grey or red font) are marked as indicated in Supplemental Figure 1. CDR3b mutations for additional clones are listed below the sequence map.
SUPPLEMENTARY FIGURE 3 | Protein sequence map for CD28 SCS-TCR formats for (A) LLO56 low and LLO56 int , and (B) LLO118 low , LLO118 int , and LLO118 high . Protein fragments are delineated by color bar above the amino acid sequence. Stability mutations (highlighted grey or red font) are marked as indicated in Supplemental Figure 1. CDR3b mutations for additional clones are listed below the sequence map.
SUPPLEMENTARY FIGURE 4 | Protein sequence map for 3 rd gen SCS-TCR formats for (A) LLO56 low and LLO56 int , and (B) LLO118 low , LLO118 int , and LLO118 high . Protein fragments are delineated by color bar above the amino acid sequence. Stability mutations (highlighted grey or red font) are marked as indicated in Supplemental Figure 1. CDR3b mutations for additional clones are listed below the sequence map.
SUPPLEMENTARY FIGURE 5 | Protein sequence map for flTCR formats for LLO56 low and LLO56 int . Protein fragments are delineated by color bar above the amino acid sequence. Stability mutations were removed for flTCR constructs. CDR3b mutations for additional clones are listed below the sequence map.