Previously we reported that the C-terminally truncated mouse CD4 T1 mutant (Table 1 and Figure 1 arrow) reduces but does not prevent IL-2 production relative to WT CD4 despite lacking most of the intracellular domain, including the CQC clasp and IKRLL motifs shown to mediate CD4-LCK association (Lee et al., 2022). Because T1 maintains both the transmembrane GGXXG and juxtamembrane (C/F)CV+C motifs, we reasoned that it had utility for studying the contributions these motifs make to pMHCII-specific signaling in the absence of CD4-LCK interactions.

Accordingly, we generated 5c.c7⁺ 58α^-β^- cells expressing WT CD4, T1, or T1 combined with our previously reported Δbind mutant that disrupts binding to pMHCII in the CD4 D1 domain. This T1^Δbind mutant allowed us to eliminate the contributions of CD4-pMHCII interactions as well as CD4-LCK interactions (Table 1 and Figure 2—figure supplement 1; Glassman et al., 2016; Glassman et al., 2018; Parrish et al., 2015). The goals were to: (1) determine if T1 significantly impairs pMHCII-specific signaling events, relative to WT, as expected because it lacks the CQC and IKRLL motifs that mediate CD4-LCK interactions; (2) use T1^Δbind to establish the baseline above which pMHCII-dependent assembly of CD4 with TCR-CD3 contributes to signaling. The T1 mutant produced measurable IL-2 (response magnitude ~61% of WT, Table 2) that was consistently reduced across a range of MCC peptide concentrations compared to WT, whereas the T1^Δbind mutant did not produce IL-2. We also observed reduced pMHCII-specific endocytosis for T1 and T1^Δbind compared to WT (Figure 2A and Figure 2—figure supplement 2). These data indicate that, in our system, extracellular CD4-pMHCII engagement is essential for IL-2 production while the CD4 intracellular domain C-terminal of R422 contributes to IL-2 production, but is not absolutely necessary in the absence of other elements of the intracellular domain.

Next, we evaluated the impact of these mutations on proximal signaling events measured by flow cytometry in response to antigen presenting cells (APCs) expressing agonist MCC:I-E^k after background subtraction of signal levels in response to APCs expressing null Hb:I-E^k (Figure 2—figure supplement 3). Our analysis of five independently generated pairs of WT and T1 cell lines showed variable difference in pCD3ζ levels. For three pairs of lines, the T1 response was higher than the WT and for two it was lower (Figure 2B). The averages of the T1 response normalized to the paired WT controls showed no difference between the population. There were also no differences for the average percent of T1 cells with response levels to MCC that were above background responses to the null Hb peptide compared to their paired WT controls. Similar results were observed for pZAP70 and pPLCγ1 levels (Figure 2—figure supplement 4). These data, summarized in Table 2, provide evidence that CD4-LCK interactions via the CQC clasp and IKRLL motifs are not key determinants of pMHCII-specific early signal events in this experimental system.

In comparison, the T1^Δbind mutant significantly reduced pCD3ζ, pZAP70, or pPLCγ1 levels compared to the T1 control but, interestingly, did not impact the percent of cells responding to agonist MCC:I-E^k (Figure 2C and Figure 2—figure supplement 5). Because the antibodies used to detect pCD3ζ, pZAP70, and pPLCγ1 recognize single phosphorylated tyrosines, these data indicate that CD4 enhances the number of TCR-CD3 complexes per cell with phosphorylated CD3ζ molecules at Y142 (same in human and mouse), as well as the number of phosphorylated ZAP70 and PLCγ1 molecules per cell, in a manner that is dependent on TCR-CD3 and CD4 co-engagement of pMHCII but independent of the intracellular domain C-terminal of R422. A summary of these results can be found in Table 3. A key takeaway is that the average pCD3ζ levels for three independently generated T1^Δbind mutants was 54.86% of the T1 control (p<0.001). CD3ζ phosphorylation therefore occurs upon TCR engagement of cognate pMHCII in the absence of CD4-pMHCII interactions, and CD4 engagement of pMHCII increases the number of phosphorylated TCR-CD3 complexes. Importantly, the pCD3ζ levels observed in the T1^Δbind mutants is subthreshold for IL-2 production at the bulk population level. These results allow us to conclude that pMHCII-specific baseline pCD3ζ levels achieved in our system in the absence of CD4-pMHCII interactions are approximately half of that achieved when CD4 can interact with pMHCII on the outside of the cell. Furthermore, consistent with the kinetic discrimination model of T cell activation, CD4-pMHCII interactions on the outside of the cell appear to be essential to push pCD3ζ levels above the threshold level needed for IL-2 production (Rabinowitz et al., 1996).

Having established that T1 contributes to early pMHCII-specific signaling (e.g. pCD3ζ) as well as signaling output (i.e. IL-2), we generated 5c.c7⁺ 58α^-β^- cells expressing WT, T1, or T1 bearing mutants of the GGXXG and/or (C/F)CV+C motifs (Table 1 and Figure 3—figure supplement 1). To study the function of the GGXXG motif alone, we used our previously described TMD mutant (GGXXG to GVXXL) wherein bulky side chains replace the glycines that compose a flat surface on the transmembrane helix that could mediate protein:protein or protein:cholesterol interactions (Fessler, 2016; Parrish et al., 2015; Song et al., 2014; Teese and Langosch, 2015; Wacker et al., 2013). We called these mutants T1-TMD. To test the function of the palmitoylation motif, which contains the core CVRC (418–421) sequence in mouse and humans, we made note of a broader (C/F)CVRC motif in eutherians wherein the majority of ortholog sequences had CCVRC, as found in mouse, and a minority had FCVRC as found in humans. Importantly, position 417 is under purifying selection for all extant CD4 orthologs we analyzed by the fixed effects likelihood (FEL) method (Figure 1), indicating that this residue evolved under purifying selection due to functional importance (Lee et al., 2022). We therefore made T1-Palm (3C) mutants wherein we mutated all three cysteines to serines as well as T1-Palm (2C) mutants where only the core cysteines common to mouse and humans were mutated to serines. Finally, we made mutants containing both the TMD and Palm mutants that we called T1-TP(3C) or T1-TP(2C).

To confirm that truncating CD4 relieved CD4-LCK association, and evaluate if the mutations impacted any residual CD4-LCK interactions that may occur due to colocalization in particular membrane domains, we performed sucrose gradient analysis of detergent lysates of WT, T1, T1-TMD, T1-Palm(2C), T1-Palm(3C), T1-TP(2C), and T1-TP(3C). We observed significantly reduced CD4-LCK association in detergent resistant membranes (DRMs, aka membrane rafts) and DSMs for T1 compared to WT (Figure 3—figure supplement 2; Pike, 2006). The mutants did not impact residual CD4-LCK association in the DRMs or DSMs. Furthermore, the total LCK signal associated with CD4 was significantly lower for T1 compared with the WT, and the residual association was not impacted by the mutants (Table 3 and Figure 3—figure supplement 3). We therefore conclude that, in our experimental system, any functional differences between the T1-TMD, T1-Palm(2C/3C), or T1-TP(2C/3C) mutants and the T1 control is a function of the motif we are studying in the absence of direct CD4-LCK interactions.

Next, we measured IL-2 production by WT, T1, and CD4 mutant cells to evaluate the impact of the mutations on signaling output in response to MCC. Here again, T1 was slightly lower than WT, while the T1-TMD reduced the IL-2 response magnitude and sensitivity relative to the T1 control (Figure 3A). This was expected given our prior work showing that the TMD mutant reduced IL-2 production driven by a shorter CD4 truncation mutant (CD4T, mutated at R420) (Parrish et al., 2015). Mutating the GGXXG or (C/F)CV+C motifs individually or together did not impact pMHCII-induced TCR endocytosis relative to the T1 control, and activation-induced CD4 endocytosis was inhibited for all truncated CD4 molecules (Figure 3—figure supplement 4A and B). The T1-Palm(2C), T1-TP(2C), T1-Palm(3C), and T1-TP(3C) all reduced IL-2 production relative to T1, indicating that they also influence the LCK-independent contribution of CD4 to pMHCII-specific signaling (Figure 3B). Of note, the T1-TP(2C) IL-2 response magnitude was lower than that of T1-Palm(2C) alone, consistent with what we previously observed with the full-length versions of these CD4 mutants (Figure 3B; Lee et al., 2022). Furthermore, the T1-Palm(3C) response magnitude was lower than that of T1-Palm(2C), suggesting that the additional cysteine contributes to signaling. Together, these data indicate that the GGXXG and (C/F)CV+C motifs contribute to the signaling cascade that leads to pMHCII-specific IL-2 production when CD4 does not directly interact with LCK via the intracellular CQC clasp and IKRLL motifs. These data do not tell us where in the signaling pathway the contributions are made.

To evaluate if the GGXXG and (C/F)CV+C motifs influence early signaling events, we measured pCD3ζ, pZAP70, and pPLCγ1 levels for four independently generated T1-TMD lines, three independently generated T1-Palm(2C) lines, and three independently generated T1-TP(2C) lines relative to their matched T1 controls in response to APCs expressing MCC:I-E^k (Figure 4 and Figure 4—figure supplements 1 and 2, and Table 3). For the T1-TMD, T1-Palm(2C), and T1-TP(2C) mutant lines we observed lower average levels of pCD3ζ, pZAP70, and pPLCγ1 relative to the T1 controls. These data indicate that the GGXXG and (C/F)CV+C motifs contribute to CD4’s ability to enhance proximal signaling in the absence of CD4-LCK interactions.

Of note, when we considered the average fold change of the means of the independently generated mutant cell lines relative to their paired T1 controls, we included the mean of the fold change of the T1^Δbind mutant cells from Figure 2C relative to their paired T1 controls for comparison (dotted line, Figure 4B and Figure 4—figure supplements 1B and 2B). Doing so provides visual boundaries within which to consider the contribution that CD4 T1, which can bind pMHCII, makes to proximal signaling in relationship to T1^Δbind (dotted line) that cannot bind pMHCII. These boundaries then provide an additional guide for evaluating the impact of the mutations on proximal signaling. For example, the T1-TMD and T1-Palm(2C) showed reduced pCD3ζ levels midway between T1 and T1^Δbind, which equated to a reduced contribution to pCD3ζ levels made by these mutants to ~48% of the contribution to pCD3ζ levels made by T1, while the T1-TP(2C) contributed ~16% of the pCD3ζ levels contributed by T1. The double mutant therefore nearly eliminated the contribution made by T1 to pCD3ζ levels. Although these are relative comparisons, the data further suggest a role for the GGXXG and (C/F)CV+C motifs in contributing to CD4-mediated pMHCII-specific signaling in the absence of CD4-LCK interactions.

Because we had found that T1^Δbind cells failed to produce IL-2 in response to high peptide concentrations in our dose-response experiments (Figure 1A), whereas the T1-TP(2C) mutant cells made IL-2 in response to high peptide concentrations (Figure 3B), we found it interesting that the early signaling responses we measured to APCs expressing high ligand densities of tethered pMHCII were similar between the T1^Δbind and T1-TP(2C) cells. We therefore evaluated IL-2 production of the T1, T1^Δbind, and T1-TP(2C) in response to the APCs used in these signaling studies to ask if they followed the same pattern of responses to high peptide doses in our prior experiments. We found that the T1-TP(2C) cells made approximately half as much IL-2 as the T1 controls, whereas IL-2 production by the T1^Δbind cells was negligible (Figure 4—figure supplement 3). These data indicate that the ability of CD4 to interact with pMHCII can partially overcome the loss of function of the GGXXG and (C/F)CV+C motifs at high ligand densities with respect to a downstream signaling output, such as IL-2 production, even if the early signaling events measured at 2 min after pMHCII engagement were similar.

Finally, we considered the percent of responders (i.e. those cells that signaled above background) for the mutant cell lines. We observed small decreases in responding T1-TP(2C) cells for pCD3ζ and pZAP70 relative to the control, while pPLCγ1 trended lower for this mutant. These data indicate that mutating the GGXXG and (C/F)CV+C motifs together in the absence of the intracellular domain can reduce the number of cells that initiate pMHCII-specific signaling (Figure 4C and Figure 4—figure supplements 1C and 2C). We did not observe this phenotype in our prior study when the intracellular domain was present, suggesting this is unique to the absence of the intracellular domain (Lee et al., 2022).

We also measured early signaling for three independently generated T1-Palm(3C) and T1-TP(3C) mutant cell lines compared to their paired T1 controls. For all lines, pCD3ζ, pZAP70, and pPLCγ1 levels were significantly lower than the controls, which was reflected in the normalized fold change (Figure 5A and B and Figure 5—figure supplement 1A and B, 2A and B). The fold change for the T1-TP(3C) levels relative to the T1 was similar to T1-TP(2C) (Figure 5B and Figure 5—figure supplements 1B and 2B and Table 3). These data provide further evidence that the GGXXG and (C/F)CV+C motifs together are key determinants of the LCK-independent contributions that CD4 makes to amplifying pMHCII-specific proximal signaling.

Finally, the percent responders were lower for pCD3ζ and pZAP70 for T1-TP(3C) compared to WT (Figure 5 and Figure 5—figure supplements 1C and 2C). These data are consistent with the idea that, in the absence of the intracellular domain, mutating these motifs influences initiation of key signaling events. See Table 3 for a summary of the data.

Available CD4 orthologs were identified through reciprocal blast-based searches and downloaded from GenBank. BLAST may not only identify orthologs, so additional criteria were used to include putative orthologous CD4 sequences in our analyses: presence of a domain structure consisting of four extracellular Ig domains followed by a transmembrane domain and a C-terminal intracellular domain, including the presence of the LCK binding clasp (CxC). Sequences that were shorter, contained frameshift mutations, or displayed high sequence variability were excluded from the analysis. For the current study, teleost fish were considered to be the oldest living species that contain a CD4 molecule given that a CD4 ortholog was not reported in the elephant shark (Callorhinchus milii), although future analyses of other cartilaginous fishes might yield more distant orthologs (Venkatesh et al., 2014). The final dataset contained 99 unique CD4 orthologs, ranging from teleost fish to human. These (putative) coding sequences were translated to amino acids and aligned using MAFFT (Katoh et al., 2002). For codon-based analyses, the aligned amino acid sequences were back-translated to nucleotides to maintain codons. The multiple sequence alignments were further processed to remove all insertions (indels) relative to the mouse CD4 sequence (NM_013488.3) to maintain consistent numbering of sites. The 5’ and 3’ regions of the CD4 molecules were not consistently aligned due to different start codon usage or extensions of the ICD, respectively. The alignment was edited to start at the codon (AAG) coding for K48 within mouse CD4. The alignment that includes all 99 CD4 sequences ends at the last cysteine residue that makes up the CQC clasp. For the mammalian only dataset, the alignment terminates at the mouse CD4 stop codon.

FastTree was used to estimate maximum likelihood trees (Price et al., 2010). For amino acid-based trees, the Jones-Taylor-Thornton (JTT) model of evolution was selected, while the general time-reversible model was used for nucleotide trees (Jones et al., 1992; Waddell and Steel, 1997). In all conditions, a discrete gamma model with 20 rate categories was used. A reduced representation tree (based on the amino acid alignment) is shown in Figure 1A. We used the same JTT model to estimate the marginal reconstructions of nodes indicated in Figure 1A. Phylogenetic trees and logo plots were visualized in Geneious and further edited using Adobe Illustrator. Ancestral sequences were estimated using GRASP (Foley et al., 2020).

Codon-based analysis of selection was performed using the hypothesis testing FEL model as implemented within the phylogenies (HyPhy) package (version 2.5.14) (MP) (Kosakovsky Pond et al., 2020; Pond et al., 2005; Kosakovsky Pond and Frost, 2005; Weaver et al., 2018). The back-translated codon-based alignments described above were used for these analyses. FEL uses likelihood ratio tests to assess a better fit of codons that allowed selection (p<0.1). When calculating values for all CD4 orthologs included in the initial phylogenetic analysis, we analyzed and identified the sequences within the mammalian clade as the foreground branches on which to test for evolutionary selection in order to maximize statistical power.

Covariation between protein residues was calculated using the MISTIC2 server. We calculated four different covariation methods (MIp, mFDCA, plmDCA, and gaussianDCA) (Colell et al., 2018). Protein conservation scores were calculated based on the protein alignment using the ConSurf Server (Ashkenazy et al., 2016; Ashkenazy et al., 2010). ConSurf conservation scores are normalized, so that the average score for all residues is 0, with a standard deviation of 1. The lower the score, the more conserved the protein position. For the purpose of this study, residues were considered to covary if the MI was larger than 4 and both residues had a ConSurf conservation score lower than –0.5. Also, pairs with an MI larger than 8 were considered to covary if the conservation score was below –0.3. Using these criteria, we selected 0.5% of all possible pairs as recommended (Buslje et al., 2009; Colell et al., 2018).

Statistical analyses and associated raw data for the figures are available on Dryad (doi:10.5061/dryad.p8cz8w9vw).

58α^-β^- T cell hybridoma lines were generated from Kuhns Lab stocks of parental 58α^-β^- T cell hybridoma cells (obtained from YH Chien at Stanford University) by retroviral transduction and maintained in culture by standard techniques as previously described (Glassman et al., 2018; Letourneur and Malissen, 1989). 58α^-β^- T cell hybridomas lack expression of endogenous TCRα and TCRβ chains, are CD4 negative, make IL-2 in response to TCR signaling, and are variant of the DO-11.10.7 mouse T cell hybridoma (Balb/c T cell fused to BW5147 thymoma) (Letourneur and Malissen, 1989). We validate the cells lines by these characteristics as well as expression of H2-D^d to validate Balb/c origin. In brief, 1 day after transduction the cells were cultured in 5 μg/mL puromycin (Invivogen) and 5 μg/mL zeocin (Thermo Fisher Scientific) in RPMI 1640 (Gibco) supplemented with 5% FBS (Atlanta Biologicals or Omega Scientific), penicillin-streptomycin-glutamine (Cytiva), 10 µg/mL ciprofloxacin (Sigma), and 50 µM beta-2-mercaptoethanol (Thermo Fisher Scientific). The next day drug concentrations were increased to 10 μg/mL puromycin (Invivogen) and 100 μg/mL zeocin (Thermo Fisher Scientific) in 10 mL in a T25 flask. Aliquot of 1×10⁷ cells were frozen at days 5, 7, and 9. Cells were thawed from the day 5 freeze and cultured for 3 days in 10 μg/mL puromycin and 100 μg/mL zeocin, and maintained below 1×10⁶ cells/mL to use in the functional assays. Cells used in the functional assays were grown to 0.8×10⁶ cells/mL density and replicates of three functional assays were performed every other day. If cells exceeded 1×10⁶ cells/mL at any point in the process they were discarded as they lose reactivity at high cell densities and a new set of vials was thawed. Typically, two independent WT and mutant pairs were generated for any given mutant and tested for IL-2 to gain further confidence in a response phenotype. When cells lines are presented together in a graph, that indicates that the cell lines (WT and mutants sets) were generated at the same time from the same parental cell stock.

Given the number of mutant CD4 cell lines generated and handled in this study, the identity of the transduced CD4 gene was verified by PCR sequencing at the conclusion of three independent functional assays. 58α^-β^- T cell hybridomas were lysed using DirectPCR Tail Lysis Buffer (Viagen Biotech) with proteinase K (SIGMA) for 2 hr at 65°C. Cells were then heated at 95°C for 10 min. Cell debris was pelleted, and the supernatants were saved. CD4 was amplified by PCR using Q5 DNA Hot Start Polymerase (New England BioLabs) in 0.2 μM dNTP, 0.2 μM primer concentration, Q5 Reaction buffer, and water. CD4 was amplified using the following primers:

5’ primer: acggaattccgctcgagcgccaccatggtgcgagccatctctctcttagg.

3’ primer: ctagcaagcttgtcgactcaagatcttcattagatgagattatggctcttctgc.

Product were purified using SpinSmart Nucleic Acid Purification Columns (Thomas Scientific) and sent to Eton Bioscience for sequencing with the following 5’ CD4 primer: gtctctgaggagcagaag.

The I-E^k+ M12 cells used as APCs were previously reported (Glassman et al., 2018). M12 cells are a murine B cell lymphoma from Balb/c mice (H2-D^d validated) (Kim et al., 1979). Cells were cultured in RPMI 1640 (Gibco) supplemented with 5% FBS (Atlanta Biologicals or Omega Scientific), penicillin-streptomycin-glutamine (Cytiva), 10 µg/mL ciprofloxacin (Sigma), 50 μM beta-2-mercaptoethanol (Thermo Fisher Scientific), and 5 μg/mL puromycin (Invivogen), and 50 μg/mL zeocin (Thermo Fisher). The parental cells are maintained in Kuhns Lab stocks and were originally obtained from MM Davis stocks (Stanford University).

Parental 58α^-β^- T cell hybridoma and M12 cells are periodically treated with Plasmocin and tested for mycoplasma contamination by PCR using primer sequences available from ATCC (5' primer sequence: TGCACCATCTGTCACTCTGTTAACCTC; 3' primer sequence: GGGAGCAAACAGGATTAGATACCCT). All transduced cell lines used here were grown in ciprofloxacin and generated from parental cells that had tested negative for mycoplasma.

For retroviral production we used Phoenix-eco cells from the Nolan Lab (ATCC CRL-3214).

Cell surface expression of CD4 and TCR-CD3 complexes were measured by flow cytometry. In brief, cells were stained for 30 min at 4°C in FACS buffer (PBS, 2% FBS, and 0.02% sodium azide) using anti-CD4 (clone GK1.5, eFluor 450 conjugate, Thermo Fisher Scientific), anti-TCRα (anti-Vα11, clone RR8-1, APC conjugate, Thermo Fisher Scientific), anti-CD3ε (145-2C11, Thermo Fisher Scientific) and GFP was detected as a measure of the TCRβ-GFP subunit. Analysis was performed on a Canto II or LSRII (BD Biosciences) at the Flow Cytometry Shared Resource at the University of Arizona. Flow cytometry data were analyzed with FlowJo Version 9 software (Becton, Dickinson & Company).

IL-2 production was measured to quantify pMHCII responses. 5×10⁴ transduced 58α^-β^- T cell hybridomas were cocultured with 1×10⁵ transduced I-E^k+ M12 cells in triplicate in a 96-well round-bottom plate in RPMI with 5% FBS (Omega Scientific), Pen-Strep+L-Glutamine (Cytiva), 10 ng/mL ciprofloxacin (Sigma), and 50 μM beta-2-mercaptoethanol (Fisher) in the presence of titrating amounts of MCC 88-103 peptide (purchased from 21st Century Biochemicals at >95% purity) starting at 30 μM MCC and a 1:3 titration (Glassman et al., 2018). For experiments with APCs expressing tethered pMHCII, 5×10⁴ 58α^-β^- T cell hybridomas were cultured with 1×10⁵ MCC:I-E^k+ or T102S:I-E^k+ M12 cells in triplicate in a 96-well round-bottom plate using the same culture conditions as above. The supernatants were collected and assayed for IL-2 concentration by ELISA after 16 hr of coculture at 37°C. Anti-mouse IL-2 (clone JES6-1A12, BioLegend) antibody was used to capture IL-2 from the supernatants, and biotin anti-mouse IL-2 (clone JES6-5H4, BioLegend) antibody was used as the secondary antibody. Streptavidin-HRP (BioLegend) and TMB substrate (BioLegend) were also used.

To assess engagement-induced endocytosis, CD4 surface levels were measured by flow cytometry 16 hr after coculture with APCs and peptide as described above for IL-2 quantification. 96-well plates containing cells were washed with ice-cold FACs buffer (PBS, 2% FBS, 0.02% sodium azide), transferred to ice, and Fc receptors were blocked with Fc block mAb clone 2.4G2 for 15 min at 4°C prior to surface staining for 30 min at 4°C with anti-CD4 (clone GK1.5 EF450, Invitrogen) and anti-Vβ3 TCR clone (clone KJ25, BD Pharmingen) antibodies. Cells were washed with FACS buffer prior to analysis on a LSRII (BD Biosciences) at the Flow Cytometry Shared Resource at the University of Arizona. Flow cytometry data were analyzed with FlowJo Version 10 software (Becton, Dickinson & Company). The average of the geometric mean of the TCR or CD4 signal was taken for the triplicate of the post 58α^-β^- cells cocultured with M12 I-E^k+ cells at 0 μM MCC concentration. Each value of the raw gMFI of TCR or CD4 for cells cultured at 10 μM MCC was subtracted from the average gMFI at 0 μM. The values show the change of gMFI from 0 μM to 10 μM.

Membrane fractionation by sucrose gradient was performed similarly to previously described methods (Hur et al., 2003; Parrish et al., 2016). For cell lysis, 6×10⁷ 58α^-β^- T cell hybridomas were harvested and washed 2× using TNE buffer (25 mM Tris, 150 mM NaCl, 5 mM EDTA). Cells were lysed on ice in 1% Triton X detergent in TNE in a total volume of 1 mL for 10 min and then dounce homogenized 10×. The homogenized lysate was transferred to 14×95 mm² Ultraclear Ultra Centrifuge tubes (Beckman). The dounce homogenizer was rinsed with 1.6 mL of the 1% lysis buffer, which was then added to the ultracentrifuge tube. 2.5 mL of 80% sucrose was added to the centrifuge tube with lysate and mixed well. Gently, 5 mL of 30% sucrose was added to the centrifuge tubes, creating a 30% sucrose layer above the ~40% sucrose/lysate mixture. Then, 3 mL of 5% sucrose was added gently to the centrifuge tube, creating another layer. The centrifuge tubes were spun 18 hr at 4°C in a SW40Ti rotor at 36,000 rpm.

Analysis of membrane fractions was performed via flow-based fluorophore-linked immunosorbent assay (FFLISA) as previously described (Parrish et al., 2016). In brief, 88 μL of Streptavidin Microspheres 6.0 μm (Polysciences) were coated overnight at 4°C with 8 μg of biotinylated anti-CD4 antibody (clone RM4-4, BioLegend). Prior to immunoprecipitation (IP), beads were washed with 10 mL of FACS buffer (1× PBS, 2% FBS, 0.02% sodium azide) and resuspended in 3.5 mL of 0.1% Triton X-100 lysis wash buffer in TNE. For each cell line lysed, 10 FACS tubes were prepared with 50 μL of the washed RM4-4-coated beads. Upon completion of the spin, 500 μL was carefully taken off the top of the centrifuge tubes and discarded. Following this, 1 mL was extracted from the top of the tube, carefully as to not disrupt the gradient, and added to a FACS tube with coated beads and capped. This was repeated for 10 individual fractions in separate FACS tubes. Following the extraction, lysates were incubated with the beads for 90 min, inverting the tubes to mix every 15 min.

Following the IP, FACS tubes were washed 3× using 0.1% Triton X lysis wash buffer in TNE. Tubes were then stained using 1 μL anti-CD4 (APC conjugate; clone GK1.5, BioLegend), 1.5 μL anti-LCK (PE conjugate, clone 3A5, Santa Cruz Biotechnology), and 1 μL CTxB (AF 488 conjugate, Thermo Fisher Scientific, resuspended as per the manufacturer’s instructions) for 45 min at 4°C. Following the stain, tubes were washed using 0.1% Triton X lysis wash buffer in TNE. Analysis of beads was performed on a LSRII (BD Biosciences) at the Flow Cytometry Shared Resource at the University of Arizona. 10⁴ events were collected per sample. Flow cytometry data were analyzed with FlowJo Version 10 software (Becton, Dickinson & Company).

For FFLISA analysis, raw gMFI values for fraction 1 were subtracted from the rest of the fractions to account for background, such that the gMFI of fraction 1 is 0. To normalize the data, the percentage of CD4 within any given fraction (fx) relative to the total CD4 gMFI (CD4 signal % of total) was calculated by dividing the gMFI signal in a given fraction (fx) by the sum of the total CD4 gMFI signal [sum(f1:f10)CD4 gMFI] and multiplying by 100 [e.g. fx % of total = fx CD4 gMFI/Sum(f1:f10)CD4 gMFI × 100]. To normalize the CTxB and LCK signal in any given fraction relative to the CD4 signal in that same fraction (CTxB or LCK normalized to CD4) the gMFI of CTxB or LCK in fx was divided by the CD4 gMFI of that fx and then multiplied by the percentage of CD4 within fx (e.g. Normalized fx LCK = fx LCK gMFI/fx CD4 gMFI × fx CD4% of total CD4 gMFI). Area under the curve (AUC) analysis was performed with GraphPad Prism 9 for fractions 1–6 to determine the AUC for the DRM domains due to their floating phenotypes, and for fractions 6–10 to determine the AUC for the DSM domains.

M12 cells expressing Hb:I-E^k (null) or MCC:I-E^k (cognate) tethered pMHCII complexes were labeled with Tag-it Violet according to the manufacturer’s instructions (BioLegend). M12 cells and 58α^-β^- cells were then chilled on ice for 30 min, 5×10⁵ of each cell type were mixed together in 1.5 mL snap cap tubes, and the cells were pelleted at 2000 rpm for 30 s at 4°C to force interactions. The supernatant was removed and the tubes were transferred to a 37°C water bath for 2 min to enable signaling. Fixation Buffer (BioLegend Inc) was then added for 15 min at 37°C. Cells were washed twice with FACs buffer, pelleted at 350×g for 5 min at room temperature, resuspended in 1 mL True-Phos Perm Buffer (BioLegend Inc), and incubated at –20°C for 16 hr.

Cells were blocked with anti-mouse FcRII mAb clone 2.4G2 hybridoma supernatants (ATCC) for 30 min, pelleted, and stained on ice for 60 min with anti-pCD3ζ (clone K25-407.69, Alexa Fluor 647 conjugate, BD Biosciences) in one sample tube, or with anti-pZAP70 (clone n3kobu5, APC conjugate, Invitrogen) and anti-pPLCγ1 (clone A17025A, PE conjugate, BioLegend) in a separate sample tube at the vendor-recommended concentrations. Finally, cells were washed 2× with FACs buffer at 1000×g for 5 min at room temperature and analyzed on a Canto II (BD Biosciences) at the Flow Cytometry Shared Resource at the University of Arizona or on a BD Fortessa. 1×10⁴ 58α^-β^- cell:M12 cell couples were collected per sample.

Flow cytometry data were analyzed with FlowJo Verison 10 software (Becton, Dickinson, & Company) by gating on 58α^-β^- and M12 cell couples, as described previously (Glassman et al., 2018). Histograms of the pCD3ζ, pZAP70, or pPLCγ1 intensity for the gated population were then generated and data expressing the gated populations as numbers of cells within intensity bins was exported from FlowJo into Microsoft Excel where the number of cells for each bin intensity value for MCC:I-E^k stimulated cells was subtracted from Hb:I-E^k stimulated cells on a bin-by-bin basis. This allowed us to enumerate the intensity differences per bin upon stimulation with the agonist pMHCII over background. Mean intensity and standard error of the mean (SEM) were calculated based on the background-subtracted (MCC:I-E^k-Hb:I-E^k) data. The data was then transferred to Prism 9 where we performed smoothing analysis with 500 nearest neighbors to smooth the line profile for graphing purposes. Those intensity bins with positive values were considered to contain cells that had responded to the MCC:I-E^k stimuli above background.

Statistical analyses of sucrose gradient and functional assays were performed with GraphPad Prism 9 software as indicated in the figure legends. For each functional assay (IL-2 production and CD4 endocytosis), each individual experiment (biological replicate) was performed with triplicate analysis (technical replicates) and each experiment was repeated at least three times (three biological replicates) or as indicated in figure legends. For sucrose gradient analysis, 10⁴ beads were collected by flow cytometry in each experiment (technical replicates) and each experiment was performed three times (biological replicates). Three biological replicates were chosen for each analysis as per convention, and no power calculations were determined. One-way ANOVA was performed with a Dunnett’s post-test when all mutants tested in an experiment were compared to a control sample (e.g. WT). Sidak’s post-test were applied when comparing between two specific samples. These post-tests were chosen based on Prism recommendations. Student’s unpaired t-tests (two-tailed) were performed when comparing WT and LL mutant samples only for phosphorylation analysis.

The sequences of 5cc7α, 5cc7βG, CD4 WT, and CD4 T1 constructs were previously described (Lee et al., 2022). Full-length CD3δ, -ε, -γ, and -ζ were encoded on a poly-cistronic construct as previously described (Holst et al., 2008; Kuhns and Davis, 2007).

The following CD4-T1 constructs: T1-TMD, T1-Palm (2C), T1-Palm (3C), T1-TP (2C), and T1-TP(3C) were cloned by conventional molecular biology techniques, including PCR-based mutagenesis where needed, into pUC18 via 5’ EcoRI and 3’ HindIII. After sequence verification they were subcloned into pP2 puromycin resistance MSCV vector (MCS-IRES-puro) via 5’ XhoI and 3’ BglII, midi-prepped (QIAGEN), and sequence-verified. Mutated codons are shown in bold uppercase letters, while the motif under interrogation is shown in bold and underlined (any non-mutated codons within the motif are bold lowercase letters and underlined). The sequences for these constructs are as follows:

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcaaaggtgtattaattagaggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtGTGtccttcCTCtttctgggtttccttgggctctgcatcctctgctgtgtcaggtgccggtgataaagatcttgagtcgacaagcttgctag.

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcaaaggtgtattaattagaggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtggctccttcggctttctgggtttccttgggctctgcatcctctgcTCTgtcaggTCCcggtgataaagatcttgagtcgacaagcttgctag.

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcaaaggtgtattaattagaggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtggctccttcggctttctgggtttccttgggctctgcatcctcTCCTCTgtcaggTCCcggtgataaagatcttgagtcgacaagcttgctag.

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcaaaggtgtattaattagaggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtGTGtccttcCTCtttctgggtttccttgggctctgcatcctctgcTCTgtcaggTCCcggtgataaagatcttgagtcgacaagcttgctag.

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcaaaggtgtattaattagaggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtGTGtccttcCTCtttctgggtttccttgggctctgcatcctcTCCTCTgtcaggTCCcggtgataaagatcttgagtcgacaagcttgctag.

The gene of T1^Δbind was made by cloning the in-frame extracellular domain of CD4 that encodes the Δbind extracellular domain (Parrish et al., 2015) flanked by XhoI and BamHI into the T1 gene. The gene was sequence-verified in pUC 18, subcloned into pP2 puromycin resistance MSCV vector (MCS-IRES-puro) via 5’ XhoI and 3’ BglII, midi-prepped, and sequence-verified. The sequence for the construct is as follows with mutated codons shown in bold uppercase letters, while the motif under interrogation is shown in bold and underlined (any non-mutated codons within the motif are bold lowercase letters and underlined):

acggaattccgctcgagcgccaccatggtgcgagccatctctcttaggcgcttgctgctgctgctgctgcagctgtcacaactcctagctgtcactcaagggaagacgctggtgctggggaaggaaggggaatcagcagaactgccctgcgagagttcccagaagaagatcacagtcttcacctggaagttctctgaccagaggaagattctggggcagcatggcGATggtGATTCAGATAGCggaggttcgccttcgcagtttgatcgttttgattccaaaaaaggggcatgggagaaaggatcgtttcctctcatcatcaataaacttaagatggaagactctcagacttatatctgtgagctggagaacaggaaagaggaggtggagttgtgggtgttcaaagtgaccttcagtccgggtaccagcctgttgcaagggcagagcctgaccctgaccttggatagcaactctaaggtctctaaccccttgacagagtgcaaacacaaaaagggtaaagttgtcagtggttccaaagttctctccatgtccaacctaagggttcaggacagcgacttctggaactgcaccgtgaccctggaccagaaaaagaactggttcggcatgacactctcagtgctgggttttcagagcacagctatcacggcctataagagtgagggagagtcagcggagttctccttcccactcaactttgcagaggaaaacgggtggggagagctgatgtggaaggcagagaaggattctttcttccagccctggatctccttctccataaagaacaaagaggtgtccgtacaaaagtccaccaaagacctcaagctccagctgaaggaaacgctcccactcaccctcaagataccccaggtctcgcttcagtttgctggttctggcaacctgactctgactctggacaaagggacactgcatcaggaagtgaacctggtggtgatgaaagtggctcagctcaacaatactttgacctgtgaggtgatgggacctacctctcccaagatgagactgaccctgaagcaggagaaccaggaggccagggtctctgaggagcagaaagtagttcaagtggtggcccctgagacagggctgtggcagtgtctactgagtgaaggtgataaggtcaagatggactccaggatccaggttttatccagaggggtgaaccagacagtgttcctggcttgcgtgctgggtggctccttcggctttctgggtttccttgggctctgcatcctctgctgtgtcaggtgccggtgataaagatcttgagtcgacaagcttgctag.

All data generated and analyzed for inclusion in this study are included in the manuscript and supporting files. These include alignments and phylogenetic trees associated with Figure 1, as well as the plotted values and associated statistical values shown in figures (see Dryad: http://doi.org/10.5061/dryad.p8cz8w9vw). Cell lines and constructs are available upon request. All sequences are provided in the Materials and methods.