TCR Affinity Biases Th Cell Differentiation by Regulating CD25, Eef1e1, and Gbp2

This article is featured in In This Issue, p.2515

CD4⁺ T lymphocytes are critical for controlling infections through their ability to provide help to B cells, cytotoxic T cells, or myeloid cells (1). CD4⁺ T cells provide these diverse functions by differentiating into specialized subsets following TCR recognition of peptide:MHC class II (p:MHCII) complexes on the surface of dendritic cells (DCs) and in the context of cytokines from the innate immune system (1–3). Work by our group and others has shown that TCR dwell time on p:MHCII, which strongly correlates with TCR affinity, also influences Th cell differentiation (4–6). In our experiments, increases in TCR affinity corresponding to p:MHCII dwell times of 0.9–2.3 s correlated with increased differentiation of macrophage-helping Th1 cells and decreased formation of B cell–helping T follicular helper (Tfh) cells (4, 5, 7). Although Th1 cell differentiation fostered by high-affinity TCR interactions is related to strong induction of the IRF4 transcription factor (8), additional aspects of the mechanism by which TCR affinity affects T cell differentiation have yet to be determined.

IL-2R signaling promotes Th1 cells and suppresses Tfh cell differentiation by driving STAT5 activation and induction of Blimp1, a repressor for the Tfh cell–promoting transcription factor Bcl-6 (9–15). Potentially, TCR affinity regulates Th cell differentiation, in part, by influencing IL-2 signaling. We therefore tested two TCR affinity–regulated, IL-2 signaling–based mechanisms, one rooted in DC Ag presentation and another focused on IL-2R α-chain (CD25) expression. The two major classical DCs in the spleen differ in p:MHCII presentation and IL-2 consumption potential (16–18). XCR1⁺ DCs are potent producers of the Th1 cell–inducing cytokine IL-12 (18) but are relatively poor producers of p:MHCII complexes, whereas SIRP⍺⁺ DCs consume IL-2 and are weak IL-12 producers but are strong producers of p:MHCII complexes (17). Thus, it is possible that high TCR affinity could bias toward Th1 cell differentiation because only Th cells with high-affinity TCRs cells could access the small number of p:MHCII complexes displayed on XCR1⁺ DCs, whereas low-affinity cells could access the abundant p:MHCII complexes on SIRP⍺⁺ DCs favoring Tfh cell differentiation (19). Alternatively, because IL-2R expression is proportional to the strength of TCR signaling and drives Th1 cell–promoting STAT5 activation (9–15), Th cells with high-affinity TCRs may be intrinsically more likely to become Th1 cells than Th cells with low-affinity TCRs.

We tested these models by examining the influence of TCR affinity on differentiation and T cell/DC interactions using two TCR transgenic (Tg) strains that contain T cells with differing TCR affinities for the same p:MHCII ligand (20). We found that Th cells with low-, medium-, or high-affinity TCR T cells tended to adopt uncommitted, Tfh, or non-Tfh cell fates, respectively, after exposure to bacteria or viruses expressing the relevant peptide. In all cases, Th cells interacted more frequently with SIRP⍺⁺ DCs than with XCR1⁺ DCs, indicating that differential interactions with DCs did not account for TCR affinity–based differences in Th1/Tfh cell formation. Rather, TCR affinity influenced Th differentiation by controlling the expression of the IL-2R and eukaryotic translation elongation factor 1 ε 1 (Eef1e1), which promoted Th1 cells, and guanylate binding protein 2 (Gbp2), which promoted Tfh cells. Our results suggest that TCR affinity–based effects on Th differentiation are related to the capacity of the TCR to induce different genetic programs at different affinity levels.

Six- to eight-week-old C57BL/6 (B6) mice were purchased from The Jackson Laboratory or the National Cancer Institute Mouse Repository (Frederick, MD). Il2ra^−/− mice were purchased from The Jackson Laboratory. Rag1^−/− B3K506 TCR Tg, Rag1^−/− B3K508 TCR Tg (20), and Rag1^−/− Ubc^GFP TEa TCR Tg mice were bred and housed in specific pathogen-free conditions in accordance with guidelines of the University of Minnesota Institutional Animal Care and Use Committee and National Institutes of Health. Clec4a4^DTR mice (21) were backcrossed onto the B6 background and used as bone marrow donors. The Institutional Animal Care and Use Committee of the University of Minnesota approved all animal experiments.

For intranasal and i.v. infections, mice were injected with 1 × 10⁷ CFUs of ActA-deficient Listeria monocytogenes expressing the 3K peptide, L. monocytogenes/P5R, L. monocytogenes/P2A, or L. monocytogenes/Flic bacteria, as described previously (5, 22, 23). Mice injected i.v. with L. monocytogenes/P5R or L. monocytogenes/P2A bacteria contained similar numbers of organisms in their livers 24 h later (D.I. Kotov, unpublished observations). L. monocytogenes/3K, L. monocytogenes/P5R, and L. monocytogenes/P2A strains were generated by inserting sequences encoding each peptide in frame with the Listeriolysin O signal sequence and promoter that drives maximal production under in vivo infection conditions, as described previously (22). The P5R epitope was inserted into the PR8 influenza A virus genome as previously described (24, 25). Mice were infected intranasally with 40 PFU of PR8 influenza A virus–expressing P5R.

Lymph nodes were collected from Rag1^−/− B3K506 and Rag1^−/− B3K508 (20) TCR Tg mice and hand-mashed into a single-cell suspension. Aliquots were stained with allophycocyanin-labeled CD4 (RM4–5; Tonbo Biosciences) Ab and analyzed with fluorescent AccuCheck Counting Beads (Invitrogen) to assess CD4⁺ T cell numbers and purity using an LSR II (BD Biosciences) flow cytometer. For imaging experiments, 1 × 10⁶ TCR Tg cells were transferred into B6 mice 24 h before infection. A total of 1 × 10⁵ TCR Tg cells were transferred into B6 mice for flow cytometry experiments examining the initial 3 d following L. monocytogenes infection, whereas 3 × 10³ TCR Tg cells were transferred for experiments examining the response at 7 d postinfection with L. monocytogenes or influenza.

Single-cell suspensions were generated by dissociating spleens with the gentleMACS Dissociator (Miltenyi Biotec) or by hand in a petri dish. Single-cell suspensions were stained for 1 h at room temperature with 0.1 μg of FITC-labeled CD90.1 Ab (HIS51; Thermo Fisher Scientific) and 2 μg of BV650-labeled (L138D7; BioLegend) or BUV395-labeled (2G8; BD Biosciences) CXCR5 Ab. Samples were then enriched for CD90.1 Ab–bound cells using magnetic bead–based enrichment, as described previously (26), with the minor modification that EasySep Mouse APC or FITC Positive Selection Kits (Stemcell Technologies) and EasySep Magnets (Stemcell Technologies) were used.

For identification of surface markers, the sample was stained on ice with various combinations of the following Abs: allophycocyanin-ef780–labeled B220 (RA3–6B2; Thermo Fisher Scientific), allophycocyanin-ef780–labeled CD11b (M1-70; Thermo Fisher Scientific), allophycocyanin-ef780–labeled CD11c (N418; Thermo Fisher Scientific), BV786-labeled CD4 (GK1.5; BD Biosciences), AF700-labeled CD44 (IM7; Thermo Fisher Scientific), FITC-labeled CD45.1 (A20; BioLegend), BUV395-labeled CD45.2 (104; BD Biosciences), BUV395-labeled CD25 (PC61; BD Biosciences), PE-labeled CD69 (H1.2F3; Thermo Fisher Scientific), and FITC-labeled CD90.1 (HIS51; Thermo Fisher Scientific). All samples were also stained with a fixable viability dye (Ghost Dye Red 780; Tonbo Biosciences). For transcription factor staining, samples were fixed with the eBioscience Foxp3/Transcription Factor Staining Kit (Thermo Fisher Scientific) and then stained with BV421-labeled RORɣt (Q31-378; BD Biosciences), BV605-labeled or BV421-labeled T-bet (4B10; BioLegend), and AF488-labeled Bcl-6 (K112-91; BD Biosciences) Abs. To calculate cell numbers, fluorescent AccuCheck Counting Beads (Invitrogen) were added to each sample after the final wash step. Cells were then analyzed on an LSR II or Fortessa (BD Biosciences) flow cytometer. Data were analyzed with FlowJo (Tree Star).

Spleens were harvested from mice 48 h after i.v. infection with L. monocytogenes/P5R bacteria. The spleens were chopped into small pieces and digested with Collagenase P (MilliporeSigma) for 20 min at 37°C prior to hand-mashing in a petri dish to generate a single-cell suspension. The single-cell suspension was enriched for CD11c-expressing cells using a CD11c-positive selection kit (Miltenyi Biotec). The enriched samples were stained for 30 min at room temperature with BV421-labeled XCR1 (ZET; BioLegend), PE-labeled CD64 (X54-5/7.1; BioLegend), allophycocyanin-labeled SIRP⍺ (P84; BioLegend), allophycocyanin/Cy7-labeled Ly-6G (1A8; BioLegend), allophycocyanin/Cy7-labeled Siglec F (E50-2440; BD Biosciences), allophycocyanin-eF780–labeled NKp46 (29A1.4; Thermo Fisher Scientific), allophycocyanin-eF780–labeled CD90.2 (53-2.1; Thermo Fisher Scientific), PE/Cy7-labeled Ly-6C (HK1.4; Thermo Fisher Scientific), FITC-labeled CD11c (N418; Thermo Fisher Scientific) Abs, and a fixable viability dye (Ghost Dye Red 780; Tonbo Biosciences). The cells were sorted with a FACS Aria II (BD Biosciences) to isolate live XCR1⁺ (CD64⁻NKp46⁻Ly-6C⁻CD90.2⁻Ly-6G⁻SIRP⍺⁻CD11c⁺XCR1⁺) and SIRP⍺⁺ (CD64⁻NKp46⁻Ly-6C⁻CD90.2⁻Ly-6G⁻XCR1⁻CD11c⁺SIRP⍺⁺) DCs.

For coculture, naive B3K508 CD4⁺ T cells were isolated from Rag1^−/− B3K508 TCR Tg mice as described in the cell transfer section. B3K508 cells were cultured with the sort-purified XCR1⁺ or SIRP⍺⁺ DCs at 1:1, 1:3, and 1:10 ratios of DCs to T cells in complete IMDM for 24 h at 37°C. Some T cells were also cultured without DCs to serve as negative staining controls for markers of activation. After culture, T cells were stained with Ab and analyzed on a flow cytometry as described in the cell enrichment and flow cytometry section.

Confocal microscopy was performed using a Leica SP5 confocal microscope with two hybrid detectors and two photomultiplier detectors as well as 405, 458, 488, 514, 543, 594, and 633 lasers. Twenty-micrometer paraformaldehyde-fixed splenic sections from naive or L. monocytogenes/P5R-infected mice were imaged with a 63× oil immersion objective lens with a numerical aperture of 1.4 NA. The splenic sections were stained with F4/80 BV421-labeled (BM8; BioLegend), Pacific Blue–labeled B220 (RA3-6B2; BioLegend), CF405L-labeled CD8⍺ (53-6.7; BioLegend), AF488-labeled phospho-S6 kinase (p-S6) (2F9; Cell Signaling Technologies), CF555-labeled CD86 (GL-1; BioLegend), AF647-labeled CD45.2 (104; BioLegend), AF700-labeled MHCII (M5/114.15.2; BioLegend), CF514-labeled CD11c (N418; BioLegend), BV480-labeled CD3 (17A2; BD Biosciences), and AF594-labeled SIRP⍺ (P84; BioLegend) Abs. CF-labeled Abs were generated by conjugating purified Abs from BioLegend to CF405L, CF514, or CF555 with Biotium Mix-n-Stain labeling kits (Biotium). The mark and find feature in the Leica Application Suite was used to image 12 T cell zones in each spleen, with each image consisting of a 20-μm z-stack acquired at a 0.5-μm step size. Additionally, the Leica SP5 microscope was used to image single color–stained Ultracomp eBeads (Thermo Fisher Scientific) for generating a compensation matrix.

Image analysis was performed using Chrysalis software (27). In brief, a compensation matrix was created by automatic image-based spectral measurements on calibration samples in ImageJ by using Generate Compensation Matrix (27). The compensation matrix was used to perform linear unmixing on three-dimensional images and movies with Chrysalis. Chrysalis was also used for further automated image processing, including merging images from the same tissue into a single stack, generating preview time-lapse sequences to identify informative fields and generating new channels based on mathematical and morphological operations on existing channels. Imaris 8.3, 8.4, 9.0, and 9.1 (Bitplane) were used for surface creation to identify cells in images and to run the Sortomato, XTCreateSurfaces, and XTChrysalis Xtensions (27). These Xtensions were used for creating surfaces based on protein expression, quantifying cell/cell interactions, and exporting cell surface statistics. These exported statistics were imported into FlowJo for quantitative image analysis.

A total of 1 × 10⁶ B3K506 or B3K508 TCR Tg cells were transferred into B6 mice 24 h before i.v. infection with L. monocytogenes/P5R bacteria. The spleen and lymph nodes were harvested 48 h postinfection and stained with FITC-labeled CD90.1 (HIS51; Thermo Fisher Scientific), allophycocyanin-ef780–labeled B220 (RA3–6B2; Thermo Fisher Scientific), allophycocyanin-ef780–labeled CD11b (M1-70; Thermo Fisher Scientific), allophycocyanin-ef780–labeled CD11c (N418; Thermo Fisher Scientific), BV786-labeled CD4 (GK1.5; BD Biosciences), AF700-labeled CD44 (IM7; Thermo Fisher Scientific), and a fixable viability dye (Ghost Dye Red 780; Tonbo Biosciences). The TCR Tg cells were then enriched with the EasySep Mouse FITC Positive Selection Kit (Stemcell Technologies) and EasySep Magnets (Stemcell Technologies). Live TCR Tg CD4⁺ T cells were sorted from the enriched sample into TRIzol (Thermo Fisher Scientific) using a FACS Aria II (BD Biosciences). RNA was extracted using chloroform (Thermo Fisher Scientific) and purified with the RNeasy Micro Kit (Qiagen). RNA quantities were measured with a Qubit 2.0 (Thermo Fisher Scientific) and 41 ng RNA of each sample was converted to cDNA with SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific). Primers (Supplemental Table I) were designed with the National Center for Biotechnology Information Primer-Blast tool (28). Quantitative PCR (qPCR) was performed with FastStart SYBR Green (MilliporeSigma) in a thermocycler (Eppendorf). The resulting data were analyzed by normalizing to the housekeeping gene Gapdh and then calculating fold change in gene expression for B3K506 cells relative to B3K508 cells with the comparative C_T method (29).

A CRISPR/Cas9 system using two MSCV-based γ retroviral vectors was used to target genes of interest. One virus encoded Cas9 and the fluorescent protein mNeongreen (30) and the other encoded guide RNAs (gRNAs) and the fluorescent protein mAmetrine (31). These vectors were created by modifying the LMP-Amt vector (32), which was a gift from S. Crotty (La Jolla Institute). The vector encoding Cas9 and mNeongreen was generated by replacing the short hairpin RNA–encoding segment, PGK promoter, and mAmetrine gene with Cas9-P2A and mNeongreen (Allele Biotechnology) through In-Fusion Cloning (Takara Bio). The Cas9-P2A fragment was cloned from the lentiCRISPRv2 puro plasmid, which was a gift from B. Stringer (plasmid no. 98290; Addgene).

To generate the gRNA vector, the LMP-Amt vector (32) was modified by removing the SapI site and replacing the short hairpin RNA–encoding segment with the bacterial toxin gene CCDB flanked by AarI sites using In-Fusion Cloning (Takara Bio). This CCDB-containing fragment was cloned from the pMOD_B2303 plasmid, which was a gift from D. Voytas (University of Minnesota). This modified LMP-Amt plasmid served as the recipient for Golden Gate cloning reactions that replaced the CCDB gene with a transfer RNA (tRNA)/gRNA array (33–35) using AarI (Thermo Fisher Scientific), SapI (New England BioLabs), and T4 (New England BioLabs) enzymes. The gRNAs used in the tRNA/gRNA arrays were designed in Benchling to target functional motifs identified in InterPro (36) with two gRNA designed for each target gene. Primers encoding the gRNAs were then generated using the Voytas Lab Plant Genome Engineering Toolkit (37), and the cloning of the tRNA/gRNA array was performed as described previously (37).

Retrovirus was prepared as described previously (38), with minor alterations. In brief, Platinum-E cells (Cell Biolabs) were grown in complete DMEM (Life Technologies) prior to transfection with Polyethylenimine, Linear, MW 25,000 (Polysciences), pCL-Eco, and retroviral plasmids encoding Cas9 or tRNA/gRNA arrays. After transfection, the media was supplemented with ViralBoost (ALSTEM Cell Advancements) and 30 μM water soluble cholesterol (MilliporeSigma). The virus-containing supernatant was collected 24 and 48 h after transfection and filtered with a 0.45-μm Nylon, 25-mm Syringe Filter (Thermo Fisher Scientific). Aliquots were stored at −80°C for up to 3 mo.

Retroviral transduction was performed as described previously (38), with minor alterations. Specifically, naive B3K508 CD4⁺ T cells or naive Cas9⁺ B3K508 CD4⁺ T cells were isolated from Rag1^−/− B3K508 TCR Tg or Rag1^−/− Cas9⁺ B3K508 TCR Tg mice, respectively, as described in the cell transfer section. These B3K508 T cells were grown in a 96-well plate with complete IMDM (MilliporeSigma) containing IL-7 (Tonbo Biosciences). Cells were activated in plates coated with anti-CD3 (2C11; Bio X Cell) and anti-CD28 (37.51; Bio X Cell). The cells were transduced with retroviral supernatant and polybrene (MilliporeSigma) 24 and 40 h after activation. B3K508 T cells were transduced with retrovirus-encoding Cas9 and retrovirus-encoding tRNA/gRNA arrays, whereas the Cas9⁺ B3K508 T cells were only transduced with tRNA/gRNA array–encoding retrovirus. For transductions, plates were spun at 1500 rpm for 2 h at 37°C, and then the media was exchanged for complete IMDM-containing IL-2 (PeproTech). After 3 d of culture, the cells were moved to plates that were not coated with CD3 or CD28 Abs, and 2 d later, the media was switched to complete IMDM-containing IL-7 (Tonbo Biosciences). After 24 h, cells were either directly transferred into B6 mice or stained with BV786-labeled CD4 (RM4-5; BD Biosciences) Ab and a fixable viability dye (Ghost Dye Red 780; Tonbo Biosciences). The stained cells were sorted with a FACS Aria II (BD Biosciences) to isolate live Cas9⁺gRNA⁺CD4⁺ T cells, which were then transferred into B6 mice. Each B6 mouse received 2500 T cells. Four days after transfer, mice were infected with L. monocytogenes/3K bacteria, and the spleens were harvested 7 d postinfection for flow cytometric analysis.

Statistical significance was determined using Prism (GraphPad) software for unpaired two-tailed Student t test and one-way and two-way ANOVA tests. Prism (GraphPad) was also used to calculate linear correlations and R².

All of the custom code generated for image processing or analysis can be downloaded at https://histo-cytometry.github.io/Chrysalis/.

We used B3K506 and B3K508 TCR Tg strains (20) to examine the relationship between TCR affinity and T cell differentiation. B3K506 and B3K508 mice contain monoclonal populations of CD4⁺ T cells with TCRs that bind to a peptide called P5R complexed with the I-A^b MHCII molecule of B6 mice. The B3K506 TCR, however, binds more strongly to this ligand than B3K508 TCR (K_Ds of 11 and 93 μM), whereas the B3K508 TCR binds to a related ligand P2A:I-A^b more strongly than the B3K506 TCR (K_Ds of 175 and 278 μM). The K_Ds of 11 and 93 μM will be considered high and medium affinities, respectively, whereas K_Ds of 175 and 278 μM will be considered to be low affinities. The effect of these affinity differences on differentiation was examined by adoptive transfer of CD90.1⁺ B3K506 or B3K508 T cells into CD90.2⁺ mice (39) that were then infected i.v. with L. monocytogenes bacteria engineered to express P5R (L. monocytogenes/P5R) or P2A (L. monocytogenes/P2A) peptide.

As expected from the work of Huseby and colleagues (20), the expansion of B3K506 and B3K508 T cells in response to L. monocytogenes/P5R and L. monocytogenes/P2A infection correlated positively with TCR affinity (Fig. 1A). Previous work (9) indicated that L. monocytogenes infection drives the differentiation of Th1 cells, which express the T-bet transcription factor (40) and Tfh cells, which express the CXCR5 chemokine receptor (41–43). B3K506 and B3K508 cells also adopted these fates (Fig. 1B), although high TCR affinity (lower K_D) biased differentiation to the Th1 cell fate and medium affinity toward the Tfh cell fate as we reported previously (5), whereas lower-affinity T cells predominantly remained uncommitted, as defined by a lack of T-bet or CXCR5 (Fig. 1C). Despite having a lower percentage of Tfh cells than the medium TCR affinity population, the high TCR affinity T cell population had a greater absolute number of Tfh cells because of greater proliferation (Fig. 1C). The TCR affinity–mediated differentiation bias was recapitulated in mice infected with influenza expressing P5R (PR8/P5R). After PR8/P5R infection, T cells with a higher TCR affinity were again biased toward the Th1 cell fate, whereas lower TCR affinity biased cells toward the Tfh cell fate, demonstrating that the influence of TCR affinity on T cell differentiation is not restricted to L. monocytogenes infection (Fig. 1D).

The effect of TCR affinity on T cell differentiation was also tested after intranasal L. monocytogenes/P5R infection, which induces Th17 cells along with Th1 and Tfh cells (23). Th17 cell differentiation was identified by expression of the lineage-defining transcription factor RORɣt (44). B3K506 cells generated larger fractions of Th17 and Th1 cells, an equal fraction of Tfh cells, and a lower fraction of uncommitted cells than B3K508 cells (Fig. 1E), suggesting that high TCR affinity also promotes the Th17 cell fate. The large percentage of uncommitted Th cells observed after this infection was likely a consequence of low Ag presentation due to poor adaptation of this enteric bacterium to the intranasal route.

Before testing the differential Ag presentation hypothesis, we first examined the kinetics of Th cell differentiation to identify the time when critical T cell/DC interactions occur. B3K506 and B3K508 cell differentiation was examined over the first 3 d after L. monocytogenes infection because Th1/Tfh cell bifurcation occurred in this time frame in another system (11, 38). A larger fraction of the B3K508 effector T cell population consisted of CXCR5⁺ cells than the B3K506 population on days 1, 2, and 3 postinfection (Fig. 2A) and did not change after day 3 (Figs. 1C, 2A). These results suggest that the Th1/Tfh cell bifurcation occurs within the first 3 d following L. monocytogenes infection.

We also confirmed that SIRP⍺⁺ DCs produce more p:MHCII complexes than XCR1⁺ DCs after L. monocytogenes infection, as reported in another system (17). The amount of p:MHCII presentation was determined by measuring the capacity of purified splenic DCs from day 2 i.v. L. monocytogenes/P5R-infected mice to stimulate CD69 and CD25 expression by cognate Th cells in vitro (10, 17) (Fig. 2B). SIRP⍺⁺ DCs were more potent activators of B3K508 cells than XCR1⁺ DCs in this assay (Fig. 2B), indicating that SIRP⍺⁺ DCs display more P5R:I-A^b complexes than XCR1⁺ DCs, as expected.

We then tested how the different DC types affected Th cell differentiation using a cell ablation strategy. The majority of SIRP⍺⁺ DCs, which express Clec4a4 (DCIR2), were ablated by administering diphtheria toxin to radiation chimeras generated with Clec4a4^DTR bone marrow (21). The effect of XCR1⁺ DCs was studied with Batf3^−/− mice, which lack these cells (45). A 60% reduction in SIRP⍺⁺ DCs led to an increase in Th1 cells and a decrease in Tfh cell differentiation by B3K506 and B3K508 cells following L. monocytogenes/P5R infection, irrespective of TCR affinity (Fig. 2C). Ablation of XCR1⁺ DCs had the opposite effect: increased Tfh and decreased Th1 cell formation for high- and medium-affinity TCR T cells (Fig. 2D). These results indicate that XCR1⁺ DCs are critical for optimal Th1 cells and SIRP⍺⁺ DCs for optimal Tfh cell differentiation after L. monocytogenes infection for high- and low-affinity T cells. Thus, this approach indicated that differential Ag presentation by XCR1⁺ and SIRP⍺⁺ DCs cannot explain the tendency of high-affinity cells to produce Th1 cells and low-affinity cells to produce Tfh cells.

Because cell ablation can result in compensatory effects (46), we also tested the DC preference model by direct observation of T cell/DC interactions in mice with a normal complement of DCs. Histo-cytometry was used to assess B3K506 and B3K508 T cell interactions with XCR1⁺ and SIRP⍺⁺ DC because two-photon microscopy cannot simultaneously resolve all the T cell and DC populations of interest. The analysis focused on T cell zones of splenic sections from L. monocytogenes/P5R-infected and naive CD45.1⁺ mice that received CD45.2⁺ B3K506 or B3K508 cells because the T cells were primarily located in this location (39). TCR Tg cells were identified based on the congenic marker, CD45.2, which was used to generate digital surfaces corresponding to these cells (Fig. 3A, 3B). TCR Tg Th cells receiving acute TCR signals, and thus likely engaged in productive interactions with DCs, were identified by staining p-S6, a T cell activation–induced phosphorylation event associated with the mTORC1 pathway (47) (Fig. 3B, 3C).

Although B3K506 and B3K508 cells had equal p-S6 levels 1 d postinfection, B3K508 cells had significantly lower values 1 d later (Fig. 3D). This result suggests that higher TCR–affinity B3K506 cells had more prolonged TCR signaling than lower TCR–affinity B3K508 cells.

Th cell interactions with the DC subsets were then assessed. DCs were identified by generating surfaces on a DC channel that contained voxels with strong MHCII and CD11c, but weak B220, CD3, F4/80, and CD45.2 signal (Fig. 4A). XCR1⁺ DCs and SIRP⍺⁺ DCs were identified based on CD8⍺ and SIRP⍺ expression, respectively (16, 18, 48, 49) (Fig. 4B). Identification of XCR1⁺ and SIRP⍺⁺ DCs was validated by comparing the DC gating strategy in wild-type (WT) and Batf3^−/− mice, which lack XCR1⁺ DCs, 1 d after L. monocytogenes/P5R infection (Fig. 4B). DCs in the WT mice consisted of∼30% XCR1⁺ DCs and 70% SIRP⍺⁺ DCs, whereas over 90% of the DCs were XCR1⁺ DCs in Batf3^−/− mice (Fig. 4B). Productive T cell/DC interactions were identified as T cells undergoing T cell activation (p-S6⁺) and touching a DC(s) (Fig. 4C). B3K506 and B3K508 cells interacted equally with DCs 1 and 2 d after L. monocytogenes/P5R infection (Fig. 4D). The ratio of T cell contacts with SIRP⍺⁺ or XCR1⁺ DCs was analyzed to determine whether low- or high-affinity T cells interacted preferentially with one DC subset or the other. The analysis revealed that high-affinity B3K506 and medium-affinity B3K508 cells interacted 1.5 times more frequently with SIRP⍺⁺ than with XCR1⁺ DCs 1 and 2 d postinfection (Fig. 4E). Thus, TCR affinity–mediated bias in T cell differentiation was not associated with preferential access of low- or high-affinity T cells to a particular DC subset.

The lack of impact on T cell/DC interactions suggested that TCR affinity regulates differentiation in a T cell intrinsic manner. We, therefore, tested a model in which increasing TCR affinity for p:MHCII drives greater IL-2R expression, STAT5 signaling, and Th1 cell differentiation. As described previously (4), B3K506 and B3K508 cells expressed CD25 (IL-2Rα) in a TCR affinity–dependent manner 2 d postinfection with L. monocytogenes/P5R, with higher TCR affinity correlating with higher CD25 expression (Fig. 5A, 5B). This difference in CD25 expression inversely correlated with Tfh cell differentiation for high and medium TCR–affinity T cells, in line with IL-2R signal transduction suppressing Tfh cell differentiation (12) (Fig. 5A). T cells with a low-affinity TCR had the lowest CD25 expression, but less Tfh cell differentiation than T cells with a medium affinity TCR, which could be attributed to a larger proportion of uncommitted cells (Fig. 5A). Moreover, CD25⁺ Th1 cells expressed more CD25 than CD25⁺ Tfh cells (Fig. 5B), adding support to the hypothesis that CD25 signaling represses Tfh cell differentiation. These results suggest that high-affinity TCR/p:MHCII interactions promote Th1 cell differentiation by maintaining CD25, which represses the Tfh cell fate.

TCR signaling induces many proteins in addition to CD25, some of which could be mediators of TCR affinity–based effects on Th cell differentiation. We focused on nine candidates from a published list of genes induced in naive TCR Tg T cells in vitro to a greater extent by a large amount of agonist peptide than by a lower amount (10). The candidate genes were identified by excluding genes with <10 reads per kilobase of transcript, per million mapped for the no-peptide condition, less than a 4-fold increase in expression in the high- versus no-peptide conditions, and less than a 2-fold increase in expression in the high- versus low-peptide conditions. Of the 12 most abundantly expressed genes, Nr4a1 and Irf4 were excluded from further analysis because of their well-established roles in T cell differentiation (8, 50–55), whereas CD25 was used as a positive control.

Expression of the candidate genes was validated by qPCR analysis of B3K506 and B3K508 T cells 2 d after L. monocytogenes/P5R infection. Of the nine genes analyzed, seven were induced to a greater extent in B3K506 than B3K508 T cells, indicating a dependence on TCR signaling (Fig. 5C). The role of these seven genes in Th cell differentiation was determined by disrupting each gene in B3K508 cells using CRISPR/Cas9-mediated gene targeting. Cas9-expressing B3K508 cells were transduced with a retrovirus encoding the fluorescent protein mAmetrine and gRNAs targeting the candidate genes or the bacterial gene LacZ as a control. The transduced T cells were then transferred into B6 mice before i.v. infection of the mice with L. monocytogenes bacteria expressing 3K peptide (L. monocytogenes/3K). The transferred gRNA⁺ T cells were identified based on CD90.1 and mAmetrine expression and evaluated for adoption of the Th1, Tfh, and uncommitted cell lineages (Fig. 5D, 5E). The expansion of gRNA⁺ B3K508 cells in the recipient mice was unaffected in the majority of cases, with the exceptions of Eef1e1 and Srm, loss of which resulted in reduced or enhanced expansion, respectively (Fig. 5F). As expected, targeting the Il2ra gene–encoding CD25 reduced Th1 cell differentiation relative to the LacZ control population (Fig. 5G). Of the candidate genes, only Eef1e1 (encoding Eef1e1, also known as AIMP3 and p18) and Gbp2 (encoding Gbp2) had significant effects on T cell differentiation. Ablation of Eef1e1 reduced Th1 cell differentiation (Fig. 5G), whereas targeting Gbp2 led to enhanced Th1 cell differentiation. These results demonstrate that TCR-driven Eef1e1 promotes Th1 cell differentiation, whereas Gbp2 inhibits this process.

Our results indicate that uncommitted effector Th cells are induced by low-affinity TCR/p:MHCII interactions. Tfh cells then become the preferred effector cells as TCR affinity increases, until Th1 and Th17 cells become predominant at higher affinities. The fact that XCR1⁺ DCs are potent producers of IL-12 (18), but relatively poor producers of p:MHCII complexes, whereas IL-2–consuming SIRP⍺⁺ DCs have the opposite properties suggested an extrinsic explanation of these TCR affinity–based effects on Th cell differentiation. XCR1⁺ DCs could display fewer p:MHCII complexes than SIRP⍺⁺ DCs, creating a situation in which Th cells with low-affinity TCRs would be unable to interact with XCR1⁺ DCs thereby receive IL-12 and become Th1 cells, but could still interact with SIRP⍺⁺ DCs to become Tfh cells. There was potential for this model to be correct because we confirmed that XCR1⁺ DCs present fewer P5R:I-A^b complexes than SIRP⍺⁺ DCs 2 d after L. monocytogenes/P5R infection, the time frame when Tfh cell differentiation is initiated. However, we found medium- and high-affinity Th cells equally interacted with each DC subset, 20% with SIRP⍺⁺ DCs, and 15% with XCR1⁺ DCs. Thus, the frequency of interaction with XCR1⁺ DCs cannot explain the greater tendency of high-affinity Th cells to produce non-Tfh cells in this system.

We found better evidence in favor of a Th cell intrinsic mechanism. The CD25 component of the IL-2R was an attractive mediator of such an effect because its expression is proportional to TCR signaling (7, 10, 13, 14) and biases differentiation by inhibiting Tfh cell formation through Blimp1 induction (12). We found that high TCR affinity positively correlated with stronger TCR signaling, maintenance of CD25 expression, and Th1 cell differentiation. Longer expression of the IL-2R would provide more time for IL-2 binding and induction of the STAT5-regulated Tfh cell repressor Blimp1 (11, 12). DiToro et al. (56) recently showed that the capacity to produce IL-2 is also a key factor in the Tfh/Th1 cell bifurcation process. They found that Th1 cells are derived from precursors that express CD25 and experience strong STAT5 signaling, but do not produce their own IL-2, whereas the Tfh cells are derived from precursors that produce the IL-2 used by the CD25hi Th1 cell precursors, but do not respond to IL-2 because of low IL-2R expression. Our result that Th1 cells express more CD25 than Tfh cells is consistent with this model. TCR affinity likely influences the non-Tfh/Tfh cell balance by determining how many precursor cells respond to IL-2. This process may also operate during type 2 immune responses because the population of house dust mite allergen p:MHCII-specific Th cells induced by allergen inhalation is comprised of Th2 and Tfh cells (57), and the Th2 cells are CD25 dependent (58–60).

Our results indicate that the Eef1e1 is another TCR signal strength–responsive driver of the Th1 cell fate. Eef1e1 binds to Lamin A (61), and overexpression studies performed in cell lines demonstrated that Eef1e1 facilitates Lamin A ubiquitination and degradation via the ubiquitin ligase, Siah1 (62). Interestingly, Lamin A–deficient T cells have reduced in vivo Th1 cell differentiation (63), whereas Siah1 promotes Th17 cell differentiation in vitro (64). Thus, strong TCR signaling may favor the differentiation of non-Tfh cells by engaging a pathway involving Eef1e1, Lamin A, and Siah1. The Th1 cell–promoting effects of Eef1e1 may also be related to the role that we identified for this molecule in proliferation. Craft and colleagues (65) showed that IL-2–mediated activation pathways associated with cellular proliferation promote Th1 over Tfh cell differentiation. Eef1e1, which interacts with ATM (66), a protein critical for TCR signaling-induced proliferation (67), may contribute to this non-Tfh cell–promoting process.

Increased expression of Eef1e1, CD25, and IRF4 and induction of downstream pathways is a plausible explanation for the increase in Th1 cell formation that occurs as TCR affinity increases from moderate to high levels. This mechanism, however, cannot explain the increase in Tfh cell formation that occurs as TCR affinity increases from low to moderate levels. Our results indicate that Gbp2, a TCR-induced protein that promoted Tfh cell formation, could be involved in this transition. Gbp2 is associated with reduced mitochondrial fission (68), which is important for aerobic glycolysis (69–71), and operates at low levels in Tfh cells (65, 72). Therefore, Gbp2 may promote Tfh cell formation by keeping aerobic glycolysis in check. An alternative hypothesis is that Gbp2 represses Th1 cell differentiation, and high TCR affinity allows T cells to overcome this inhibition.

Together, our results suggest a general model in which low TCR affinity and signaling induces Gbp2, thereby promoting Tfh cell differentiation, whereas high TCR affinity and signaling drives Eef1e1 and CD25, which represses the Tfh cell fate and promotes non-Tfh cell formation. The type of non-Tfh cells would then depend on the type of cytokines produced by the innate immune system. For example, intranasal infection, which elicits IL-6 and TGF-β production by cervical lymph node DCs (73), would favor Th17 cell differentiation, whereas i.v. infection, which stimulates IL-12 production by splenic XCR1⁺ DCs (18), would bias toward Th1 cell differentiation. This mechanism would ensure that polyclonal populations of epitope-specific naive Th cells, which always contain a spectrum TCR affinities (26), generate effector cells capable of promoting humoral and cellular immunity suited to clearance of the microbe. The tendency of moderate-affinity T cells to become Tfh cells may foster affinity maturation by only allowing B cells with the highest-affinity Igs to garner T cell help, whereas diversion of high-affinity T cells to become myeloid cell helpers would ensure that the helped cells achieve the optimal state of microbicidal activity.

We thank J. Walter, C. Ellwood, the University of Minnesota Center for Immunology Imaging Core, the University of Minnesota Imaging Centers, and the University of Minnesota Flow Cytometry Resource for technical assistance; D. F. Voytas and J. J. Belanto for helping generate the CRISPR/Cas9 system; and M. Y. Gerner for suggestions on histo-cytometry.

The authors have no financial conflicts of interest.