Negligible role for deletion mediated by cDC1 in CD8⁺ T cell tolerance

Abstract

Deletion of CD8⁺ T cells by dendritic cells (DCs) is recognized as a critical mechanism of immune tolerance to self-antigens. Although DC-mediated peripheral deletion of autoreactive CD8⁺ T cells has been demonstrated using T cells reactive to model antigens, its role in shaping the naturally-occurring polyclonal CD8⁺ T cell repertoire has not been defined. Using Batf3^−/− mice lacking cross-presenting CD8α⁺ and CD103⁺ DCs (cDC1), we demonstrate that peripheral deletion of CD8⁺ T cells reactive to a model tissue antigen is dependent on cDC1. However, endogenous CD8⁺ T cells from the periphery of Batf3^−/− mice do not exhibit heightened self-reactivity, and deep TCR sequencing of CD8⁺ T cells from Batf3^−/− and Batf3^+/+ mice reveals that cDC1 have a minimal impact on shaping the peripheral CD8⁺ T cell repertoire. Thus, while evident in reductionist systems, deletion of polyclonal self-specific CD8⁺ T cells by cDC1 plays a negligible role in enforcing tolerance to natural self-ligands.

Introduction

Stochastic genetic recombination of the T cell receptor (TCR) loci during thymic development generates a highly diverse T cell repertoire capable of responding to a broad range of foreign antigens^1,2. However, this process also gives rise to autoreactive T cells capable of recognizing self-ligands. To prevent autoimmunity, developing T cells expressing strongly self-reactive TCRs are eliminated in the thymus through negative selection (clonal deletion), or are diverted to alternative lineages, as is the case for thymic-derived regulatory T (T_reg) cells^3–6. Unique expression of transcription factors confers upon medullary thymic epithelial cells (mTECs) the ability to express and present tissue-specific antigens to developing thymocytes^7–9. Encounter with agonist self-ligands presented by mTECs in the context of class I major histocompatibility complex (MHC-I) molecules drives the negative selection of T cells destined for the CD8⁺ lineage^10,11.

Although deletional tolerance in the thymus is efficient, it is incomplete^12–14, particularly for thymocytes bearing TCRs with lower avidity for cognate self-antigens¹¹. Thus, a fraction of T cells expressing self-reactive TCRs escape negative selection, necessitating additional layers of peripheral immune regulation to prevent the development of autoimmunity. As professional antigen presenting cells (APCs), dendritic cells (DCs) have been implicated as critical mediators of T cell tolerance in peripheral tissues¹⁵. Through their constitutive acquisition of exogenous cellular proteins during normal turnover, DCs display derivative peptides in the context of MHC-I molecules to CD8⁺ T cells through a process called antigen cross-presentation¹⁶. Steady-state antigen cross-presentation by DCs in secondary lymphoid organs (SLO) has been shown to induce the abortive proliferation and deletion of CD8⁺ T cells bearing cognate TCRs in experimental systems using engineered model tissue antigens^11,17,18, supporting the notion that DC-mediated peripheral deletion is a key mechanism in the maintenance of self-tolerance. Basic leucine zipper transcription factor ATF-like3 (Batf3)-dependent CD8α⁺ and CD103⁺ DCs (also known as cDC1) are particularly adept at antigen cross-presentation to CD8⁺ T cells, and are required for the activation of CD8⁺ T cell responses against foreign and tumor-derived antigens^19–24. Because of their unique capacity for antigen cross-presentation, cDC1 have also been implicated in mediating the deletion of self-reactive CD8⁺ T cells^25,26.

In contrast, several lines of evidence have failed to demonstrate a major role for DCs in regulating the tolerance of peripheral CD8⁺ T cells. For example, CD8⁺ T cells from DC-deficient mice failed to expand following adoptive transfer into wild-type recipients, arguing against overt auto-reactivity of CD8⁺ T cells that have developed in the absence of DCs¹⁰. Further, the discovery of other peripheral tolerance mechanisms—including the immunologically silent clearance of apoptotic cells and antigen sequestration by macrophages^27,28, as well as T_reg-mediated suppression^29,30—coupled with the identification of self-reactive T cells in the periphery under homeostatic conditions^12,13,31,32, calls into question the efficacy and necessity of DC-mediated peripheral deletion in the maintenance of CD8⁺ T cell tolerance.

To assess the importance of peripheral deletion of autoreactive CD8⁺ T cells by cDC1 in the maintenance of self-tolerance, a number of experimental approaches were undertaken, including monoclonal and polyclonal CD8⁺ T cell transfer experiments and TCR repertoire analysis of CD8⁺ T cells from secondary lymphoid organs of Batf3^+/+ and Batf3^−/− mice. While cDC1 were indeed found to be required for the peripheral deletion of model tissue antigen-specific CD8⁺ T cells, their absence had no discernable effect on the polyclonal CD8⁺ T cell repertoire or its degree of reactivity toward naturally-occurring self-antigens. Together, our results suggest that DC-mediated peripheral deletion of self-reactive CD8⁺ T cells plays a minimal role in maintaining tolerance to self.

Materials and Methods

Mice

6–12 week-old C57BL/6 (H-2^b) and B6.SJL (B6.SJL-Ptprc^a/BoyAiTac) mice were purchased from Taconic laboratories (Germantown, NY). Batf3^−/− (129S-Batf3^tm1Kmm/J), RIP-mOVA (C57BL/6-Tg(Ins2-TFRC/OVA)296Wehi/WehiJ), and OT-I mice (C57BL/6-Tg(TcraTcrb)1100Mjb/J) were purchased from Jackson Laboratories. Batf3^−/− (Batf3^tm1Kmm/J) mice were backcrossed onto a B6 background for at least 10 generations. Vβ3-tg mice were generated by Dr. Peter Savage. TCRα^−/− (Tcra^tm1Mom/J) and CD4-Cre (Tg(Cd4-cre)1Cwi/BfluJ) mice on a B6 background were a gift from Dr. Albert Bendelac. Littermate controls were utilized where indicated. All strains were bred and maintained in our facility in a specific pathogen-free environment and used according to protocols approved by the Institutional Animal Care and Use Committee at the University of Chicago according to NIH guidelines.

Flow cytometry

Samples were stained with the following directly-conjugated antibodies (BD Bioscience, eBioscience or Biolegend): TCRβ (H57–597), CD4 (GK1.5), CD8α (53–6.7), CD3ε (145–2C11), CD44 (IM7), Thy1.1 (OX-7), CD45.1 (A20), CD45.2 (104), Vβ3 (KJ25), Vα2 (B20.1), PD-1 (RMP1–30), CD5 (53–7.3). Dead cells were excluded using fixable viability dyes (Invitrogen). Samples were run on LSRII or LSRFortessa (BD Bioscience) cytometers, and analysis was performed using FlowJo (Treestar).

Adoptive T cell transfer

CD8⁺ T cells were isolated from OT-I TCR transgenic, Batf3^−/−, or Batf3^+/+ mice via magnetic separation (Miltenyi), were CTV-labeled (for short-term 4–6 day transfers only), and 1 × 10⁶ OT-I or 5×10⁶ polyclonal CD8⁺ T cells were inoculated IV into recipient mice through the lateral tail vein. At the time points indicated, secondary lymphoid organs were harvested and stained with antibodies against TCRβ, CD3ε, CD8α, antibodies recognizing the congenic labels (Thy1.1 for OT-I, CD45.1 and CD45.2 for polyclonal transfer experiments), and Vα2 (OT-I only).

CFA/OVA vaccination

Six weeks following OT-I transfer, mice were vaccinated with SIINFEKL peptide (25 μg) emulsified in CFA. Six days later, cells from the spleen, vaccine-draining, pancreatic, and renal lymph nodes were stained with the same panel of antibodies used in the adoptive T cell transfer experiments, and the presence of OT-I T cells was assessed by flow cytometry. Diabetes was monitored every day using Bayer Diastix® measuring urine glucose.

TCR sequencing and data analysis

Spleen, inguinal, axillary, brachial, peri-aortic, renal, pancreatic, mesenteric, and cervical lymph nodes were pooled for individual Batf3^+/+ Vβ3-tg and Batf3^−/− Vβ3-tg littermates (n = 5 mice per group). TCRβ⁺ CD3ε⁺ CD8α⁺ CD4⁻ cells were sorted by FACS. RNA was isolated, and cDNA was generated. Samples were sent to iRepertoire. Data analysis was performed using R. CDR3 data was filtered based on recurrence, such that only clones appearing in all five mice of either group were included, resulting in a catalogue of >10,000 unique CDR3 amino acid sequences. Data were analyzed using row-wise Benjamini-Hochberg analysis to obtain a false discovery rate (FDR). Clones were deemed to be significantly differentially expressed if they had an FDR < 0.05. None of the TCRs in the catalogue reached significance. Morisita-Horn similarity index was calculated using Microsoft Excel® following the formula:

$$C_H=\frac{2{\sum }_{i=1}^S{x}_i{y}_i}{\left(\frac{{\sum }_{i=1}^S{x}_i^2}{X^2}+\frac{{\sum }_{i=1}^S{y}_i^2}{Y^2}\right)XY}$$

Where x_i is the total reads of a clone in one mouse, y_i is the total reads for the same clone in the other mouse, and X and Y are the total read counts in the two respective mice. Data was excluded from V-J usage analysis if either the V or J gene could not be determined. Heat maps of V-J usage for the top 1000 clones from each individual mouse were generated. Representative heat maps are depicted in Supplemental Figure S2. R code for V-J usage and identification of differentially expressed clones is available from the authors by request.

Retrogenic T cell studies

Recipient mice were sub-lethally irradiated with 500 rads 24 hours before injection. On the day of injection, donor double-negative thymocytes were isolated from thymi of TCRα^−/− Vβ3-tg CD4-Cre CD45.2 or Batf3^+/+ CD45.1 mice via magnetic bead depletion for CD11c, CD11b, B220, CD19, CD4 and CD8α. Retroviral transduction of thymocytes was performed by suspending the cells in 3 mL of supernatant from platinum-E cells which were transfected with pMIG retrovirus bearing the TCRα of interest (for TCRα^−/− Vβ3-tg CD4-Cre CD45.2) or empty pMIG (for Batf3^+/+ CD45.1), plus 12 μL of polybrene beads, and spinning at 37C, 1000g for 90 minutes. Cells were washed with RPMI containing 10% FBS before suspending in PBS at 5 × 10⁷ cells/mL. Retrogenic and empty-vector cells were mixed 1:1, such that the final concentration of each was 2.5 × 10⁷ cells/mL. Recipient mice were inoculated intra-thymically with 20 uL of the cell mixture (5 × 10⁵ of each, RG and EV). Each cell mixture was used for an equal number of Batf3^+/+ and Batf3^−/− recipients to control for batch effects related to transduction efficiency and mixture ratio.

Statistical analysis

Pair-wise comparisons were performed using 2-way ANOVA with Bonferroni post-tests. Data in Figure 1 with more than two groups was analyzed using 1-way ANOVA with Bonferroni post-test. All bar graphs are presented as mean + SEM. A p-value of <0.05 was considered to be statistically significant. Differences were not statistically significant unless indicated otherwise.

Figure 1. Peripheral deletion of CD8⁺ T cells reactive to a model self-antigen requires cDC1.

A) Experimental design. 10⁶ CTV-labeled Thy1.1⁺ OT-I T cells were adoptively transferred into Batf3^+/+ (n = 6), Batf3^−/− (n = 5), Batf3^+/+ RIP-mOVA⁺ (n = 6), and Batf3^−/− RIP-mOVA⁺ (n = 6 mice for each time-point analyzed, combined from two independent experiments). OT-I T cell proliferation was assessed by CTV dilution at day 4. Long-term OT-I T cell persistence was quantified at day 42. Mice in the vaccine cohort received SIINFEKL + CFA at day 42 and were analyzed at day 63. B-C) Analysis of OT-I T cell proliferation in pancreatic (pLN) and renal (rLN) lymph nodes 4 days following adoptive transfer, with representative CTV dilution histograms shown in (B) and quantified in (C). D-E) OT-I T cell frequencies among CD8⁺ T cells in spleens, pLNs, and rLNs at day 42; representative flow plots shown in (D) and data are quantified in (E). F-G) OT-I T cell frequencies among CD8⁺ T cells in the spleen and inguinal (vaccine-draining) LN at day 63 (21 days post-vaccination). Statistical significance was measured by two-way ANOVA with Bonferroni post-test. * p < 0.05, ** p < 0.01, *** p < 0.001

Results

Peripheral deletion of CD8⁺ T cells reactive to a model self-antigen requires cDC1

Batf3-dependent cDC1 have been shown to play a major role in activating CD8⁺ T cell responses to foreign and tumor antigens^19–24, and have also been implicated in the regulation of peripheral CD8⁺ T cell cross-tolerance^25,26. Thus, we hypothesized that this DC subset would also be required to mediate the deletion of autoreactive CD8⁺ T cells in peripheral tissues through the cross-presentation of tissue-specific antigens. To initially test this hypothesis in a well-characterized system in which peripheral CD8⁺ T cell deletion has been previously demonstrated¹⁸, Batf3^−/− mice were crossed with RIP-mOVA mice, which express chicken ovalbumin (OVA) protein under the control of the rat insulin promoter centrally in the thymus and peripherally in the pancreas and kidneys³³. CellTrace Violet (CTV)-labeled OVA-specific CD8⁺ T cells (OT-I cells) were adoptively transferred into Batf3^+/+ and Batf3^−/− littermates that were either negative or positive for the RIP-mOVA transgene (Figure 1A). Robust OT-I T cell proliferation occurred in the pancreatic lymph nodes (pLN) and renal lymph nodes (rLN) of Batf3^+/+ RIP-mOVA⁺ hosts. In contrast, OT-I T cell proliferation in Batf3^−/− RIP-mOVA⁺ hosts was largely abolished, comparable to that observed in RIP-mOVA^- animals (Figure 1B and C). To assess the subsequent fate of OT-I T cells, their frequencies and numbers were assessed six weeks after initial adoptive transfer. While OT-I T cells persisted in RIP-mOVA^- mice (Batf3^+/+ and Batf3^−/−), and in Batf3^−/− RIP-mOVA⁺ mice, very few OT-I T cells were recovered from Batf3^+/+ RIP-mOVA⁺ hosts (Figure 1D and E), suggesting that they had been deleted. Finally, a subset of each group of recipient mice was vaccinated with SIINFEKL peptide (the OVA epitope recognized by the OT-I TCR, which can be presented in a cDC1-independent manner) emulsified in complete Freund’s adjuvant (CFA) six weeks post-OT-I T cell transfer. Following vaccination, OT-I T cells expanded in Batf3^−/− RIP-mOVA⁺ hosts (Figure 1F and G), some of which had elevated urine glucose levels, indicative of diabetes (data not shown). In contrast, minimal OT-I T cell expansion was observed in vaccinated Batf3^+/+ RIP-mOVA⁺ recipients, and none developed diabetes. Thus, cross-presentation of a model tissue-restricted self-antigen by cDC1 is required for the deletion of adoptively-transferred antigen-specific CD8⁺ T cells.

Central tolerance mechanisms are intact in Batf3^−/− mice

Given these results, we reasoned that—due to a deficiency in antigen cross-presenting CD8α⁺ and CD103⁺ DCs—Batf3^−/− mice would harbor a pool of endogenous CD8⁺ T cells reactive toward tissue-specific antigens, which would normally be culled in the periphery of wild-type animals. Importantly, negative selection of CD8⁺ T cells has been shown to occur independently of thymic DCs^11,34. Furthermore, we observed similar frequencies of double-positive low (DP^lo) thymocytes in Batf3^+/+ and Batf3^−/− littermates^34,35, arguing against a gross defect in negative selection in the thymi of Batf3^−/− animals (Figure 2A). In addition, similar CD5 expression levels were found on peripheral CD8⁺ T cells from Batf3^+/+ and Batf3^−/− mice (Figure 2B), suggesting that they had experienced equivalent levels of TCR stimulation during thymic development³. To determine whether the absence of cDC1 altered the thymic CD8⁺ T cell repertoire, Batf3^−/− mice were crossed with mice bearing a fixed transgenic TCRβ (TCRβtg) chain⁵. TCRα deep sequencing was performed on CD8 single-positive T cells from the thymi of individual Batf3^+/+ and Batf3^−/− TCRβtg littermates. Comparative analysis identified no recurrent TCRs that were differentially expressed in the thymi of Batf3^+/+ mice versus Batf3^−/− mice (Figure 2C), illustrating that Batf3 deficiency does not impact the thymic CD8⁺ T cell repertoire. Collectively, these results suggest that defects in CD8⁺ T cell tolerance in Batf3^−/− mice, should they exist, would likely result from defective peripheral mechanisms.

Figure 2. Intact thymic selection in Batf3^−/− mice.

A) Representative flow cytometry and quantification of DP^lo PD-1^hi CD5^hi thymocytes undergoing negative selection in Batf3^+/+ and Batf3^−/− mice (n = 4 per group from one experiment, representative of two separate experiments). B) Representative histogram and quantification of CD5 expression by splenic CD8⁺ T cells in Batf3^+/+ and Batf3^−/− mice (n = 4 per group from one experiment, representative of two independent experiments). C) Volcano plot of false discovery rate (FDR) by log₂ fold-change for recurrent TCRα sequences used by CD8 single-positive thymocytes in Batf3^+/+ TCRβtg and Batf3^−/− TCRβtg littermates (n = 5 mice per group) following TCRα repertoire sequencing. Statistical significance was determined by one-way ANOVA with Bonferroni post-test (A and B) or Benjamini-Hochberg analysis (C).

Peripheral CD8⁺ T cells from Batf3^−/− mice do not exhibit enhanced self-reactivity

To elucidate the degree to which peripheral CD8⁺ T cells from Batf3^−/− mice exhibited enhanced self-reactivity, 5×10⁶ polyclonal Batf3^+/+ or Batf3^−/− CD8⁺ T cells were CTV-labeled and adoptively transferred into congenically-marked wild-type recipients. CD8⁺ T cells from Batf3^+/+ and Batf3^−/− mice engrafted with similar frequencies, and no differences in proliferation were observed in secondary recipients 6 days following adoptive transfer (Figure 3A and B). Five weeks following their adoptive transfer, similar frequencies of donor CD8⁺ T cells were again identified in SLO, and no differences in CD44 expression on donor CD8⁺ T cells were observed (Figure 3C and D). Examination of non-lymphoid organs revealed identical levels of minor tissue infiltration by donor Batf3^+/+ versus Batf3^−/− CD8⁺ T cells (Figure 3E and F). These results suggest that peripheral tissue antigen encounter by transferred CD8⁺ T cell populations was similar, regardless of the Batf3 genotype of the donor cells.

Figure 3. Polyclonal CD8⁺ T cells from Batf3^−/− mice do not exhibit hallmarks of enhanced auto-reactivity.

5 × 10⁶ polyclonal CD8⁺ T cells (CD45.2) were adoptively transferred from Batf3^+/+ or Batf3^−/− donors into Batf3^+/+ hosts (CD45.1). A-B) Representative flow cytometry plots and histogram showing engraftment and proliferation of donor CD8⁺ T cells at day 6 (A), quantified in (B), n = 8 mice per group, combined from two independent experiments. C-D) Representative flow cytometry plots showing long-term engraftment and CD44 expression on donor CD8⁺ T cells at day 35 (C), quantified in (D), n = 5 mice per group. E-F) Representative flow cytometry plots showing CD8⁺ T cell infiltration into eyes and salivary glands at day 35 (E), quantified in (F), n = 5 mice per group from one experiment, representative of two independent experiments. Statistical significance was determined using two-way ANOVA with Bonferroni post-test. aLN: axillary LN, bLN: brachial LN, iLN: inguinal LN, rLN: renal LN, Spl: spleen.

To next determine whether CD8⁺ T cells from Batf3^−/− mice preferentially expanded in response to tissue-specific antigens in the setting of an inflammatory stimulus, CD8⁺ T cells from Batf3^+/+ and Batf3^−/− donors were CTV-labeled and co-transferred at a 1:1 ratio into wild-type recipients, which were then treated with an α-CD40 agonist antibody to trigger systemic DC activation¹⁷. Again, no difference in proliferation or engraftment was observed, nor was there a difference in organ infiltration (Figure 4), further demonstrating that CD8⁺ T cells from Batf3^−/− mice as a collective are no more self-reactive than wild-type CD8⁺ T cells. These data are not confounded by differences in TCR signaling strength in the primary mice, as similar expression levels of CD69 and Egr2 were observed among peripheral CD8⁺ T cells in naïve Batf3^+/+ and Batf3^−/− mice, with only slightly higher levels observed in the mLN of Batf3^+/+ mice, where cDC1 could present antigens derived from gut pathobionts (Supplementary Figure S1).

Figure 4. Polyclonal CD8⁺ T cells from Batf3^−/− mice do not preferentially expand following systemic APC activation.

A-B) 2.5 × 10⁶ polyclonal CD8⁺ T cells from Batf3^+/+ (CD45.1) and Batf3^−/− (CD45.2) were CTV-labeled and co-transferred into Batf3^+/+ (CD45.1/.2) hosts (n = 8, combined from two independent experiments). Hosts were treated with α-CD40 agonist antibody immediately following transfer, and at days 2 and 4. Representative flow cytometry plots for input and day 6 output (iLN) and representative histograms for donor CTV dilution are shown in (A), and data are quantified in (B). C-D) 5 × 10⁶ polyclonal CD8⁺ T cells (CD45.2) from Batf3^+/+ or Batf3^−/− were transferred into Batf3^+/+ (CD45.1) hosts (n = 5 per group, from one experiment, representative of two independent experiments). Hosts were treated with α-CD40 agonist antibody as in (A). CD8⁺ T cell infiltration of the eyes and salivary gland was measured by flow cytometry at day 6. Representative flow cytometry plots are shown in (C) and data are quantified in (D). Statistical significance was determined by two-way ANOVA with Bonferroni post-test.

The peripheral CD8⁺ T cell repertoires of Batf3^−/− and Batf3^+/+ mice are highly similar

Although the above results revealed that CD8⁺ T cells educated in a cDC1-deficient environment were not overtly prone toward self-reactivity, it remained possible that rare tissue antigen-specific clones might be more prevalent in SLO of Batf3^−/− mice. To more sensitively assess the influence of cDC1 on shaping the peripheral CD8⁺ T cell repertoire, TCRα deep sequencing was performed on pooled SLO from individual TCRβtg Batf3^+/+ and TCRβtg Batf3^−/− littermates. This method has previously demonstrated differences in TCR repertoires of T_reg and T_conv cells within prostate tumors and in Aire^−/− mice^5,35. Peripheral CD8⁺ T cell repertoires were strikingly similar between the two groups, with overlapping dominant clones and similar distributions of frequencies for clones at each level of abundance (Supplementary Table S1 and Figure 5A), and the CD8⁺ T cell repertoires exhibited comparable diversity, regardless of genotype (Figure 5B). Further analysis of population similarity using the Morisita-Horn index revealed that the CD8⁺ T cell repertoires of Batf3^−/− and Batf3^+/+ mice were nearly identical Figure 5C). In addition, analysis of V-J pairing among the top 1,000 clones in Batf3^+/+ and Batf3^−/− mice revealed very minimal differences between the two groups of animals (Supplementary Figure S2). Further, no TCRs were identified as being significantly differentially-expressed between groups of Batf3^−/− and Batf3^+/+ mice (FDR < 0.05) (Figure 5D and E), although there were a few TCRs that approached statistical significance. One candidate TCR was identified as the most likely to be subjected to peripheral deletion by cDC1 (indicated in Figure 5E and Table 1) because it was associated with an FDR that approached significance, was ~25-fold more frequent in Batf3^−/− mice than in Batf3^+/+ mice, and was exceedingly rare in the SLO of Batf3^+/+ mice (frequency < 10⁻⁵), despite being found at similar frequencies in the thymic CD8 single-positive repertoire of Batf3^−/− and Batf3^+/+ littermates.

Figure 5. Batf3 deficiency has a negligible impact on the peripheral repertoire of CD8⁺ T cells.

TCRα repertoire sequencing was performed on CD8⁺ T cells from pooled secondary lymphoid organs of Batf3^+/+ TCRβtg and Batf3^−/− TCRβtg littermates (n = 5 mice per group). A) Pie chart showing the contribution to the overall repertoire by each binned rank for each group. B) D50, Shannon entropy, and reciprocal Simpson’s index of TCRα repertories from each mouse. C) Tile representation of Morisita Horn index showing the similarity of TCRα repertoires between individual mice. Index value is depicted in greyscale. Range: 0.911–0.990. D) Dot plot depicting mean frequencies of individual recurrent TCRα sequences in Batf3^+/+ TCRβtg and Batf3^−/− TCRβtg mice. Each dot represents a single TCRα sequence. E) Volcano plot of false discovery rate (FDR) by log₂ fold-change for recurrent TCRα sequences. FDR < 0.05 was considered significant. Significance was determined by one-way ANOVA with Bonferroni post-test (B) or Benjamini-Hochberg analysis (E).

Table I.

Statistical information for the identified TCR clone

V gene	TRAV13D-1
J gene	TRAJ26
CDR3	ALGNNYAQGLT
		Batf3+/+	Batf3−/−	Fold Change (Log2)	FDR
Mean Frequency (Thymus)		2.8 E −4	2.3 E −4	0.27	0.868
Mean Frequency (SLO)		8.46 E −6	2.21 E −4	−4.706	0.067

To determine whether T cells expressing the TCR of interest (Table 1) were regulated by cDC1 in the periphery, retrogenic T cells expressing this TCR were generated (Figure 6). Double-negative TCRβtg TCRα^−/− CD4-Cre thymocytes were transduced with the retroviral vector shown in Figure 6A containing cDNA encoding for the candidate TCRα (Rg), and control wild-type thymocytes were transduced with an empty vector (EV). To elucidate the extent to which this T cell clone was subjected to deletion by cDC1, Rg and EV thymocytes were mixed 1:1 and intra-thymically injected into Batf3^+/+ and Batf3^−/− mice³⁶ (Figure 6B). The ratio of Rg:EV cells in the periphery was used to measure the relative abundance of Rg T cells. If cDC1 were driving the deletion of this CD8⁺ T cell clone, then the ratio of Rg:EV CD8⁺ T cells would be higher in Batf3^−/− mice compared to Batf3^+/+ mice. Transduction of this TCRα led to the positive selection of this T cell clone in the thymus and subsequent development of mature CD8⁺ T cells. Two weeks after injection, there was no difference between Batf3^+/+ and Batf3^−/− hosts in the recovery of Rg T cells at peripheral sites (Figure 6C and D), indicating that CD8⁺ T cells bearing this TCR were not subjected to peripheral deletion by cDC1.

Figure 6. Expression of candidate TCR does not lead to clonal deletion.

A) Retroviral vector used to express TCRα of interest. B) Double-negative thymocytes from TCRβtg CD4-Cre TCRα^−/− CD45.2 mice were transduced with the retroviral vector depicted in (A) bearing the TCR indicated in Figure 5 and Table 1 (Rg), and were mixed 1:1 with empty vector-treated wild-type double-negative thymocytes (EV, CD45.1). 5 × 10⁵ cells were intra-thymically injected into sub-lethally irradiated Batf3^+/+ or Batf3^−/− recipients (CD45.1/.2; n = 6 mice per group, combined from two independent experiments). C) Representative flow cytometry plots showing CD8⁺ T cells from the inguinal and mesenteric lymph nodes (ILN and MLN, respectively) of recipient mice two weeks after intrathymic injections. D) Development and potential deletion were determined by the ratio of Rg to EV cells in SLO two weeks after transfer and quantified. Statistical was determined by two-way ANOVA with Bonferroni post-test.

Discussion

In isolation, the finding that peripheral deletion of a high-affinity CD8⁺ T cell clone specific for a model self-antigen is mediated by cDC1 might lead to the erroneous conclusion that DC-mediated T cell deletion is an important peripheral tolerance mechanism. However, under steady-state conditions, and in mice with polyclonal CD8⁺ T cell repertoires, we found no evidence of increased self-reactivity among CD8⁺ T cells educated in a Batf3-deficient environment, largely devoid of antigen cross-presenting CD8α⁺ and CD103⁺ DCs (Figures 3 and 4). This result is consistent with previous work, which revealed unperturbed CD8⁺ T cell development and homeostasis in mice constitutively lacking DCs¹⁰. While DC-mediated peripheral deletion has been observed time and again in systems that have incorporated the use of TCR transgenic T cells and model tissue antigens,^11,17,18 the role of DCs in regulating peripheral CD8⁺ T cell deletion in mice with polyclonal T cell repertoires and against natural self-ligands has not been conclusively demonstrated. Previous studies using mice with a DC-specific ablation of phagocytosis and cell migration showed that CD8⁺ T cells from these mice were more proliferative upon transfer into Rag1^−/− hosts, but only when co-transferred with T_reg-depleted CD4⁺ T cells¹⁸. Conversely, our data suggest that the maintenance of CD8⁺ T cell tolerance to peripheral antigens in the steady state does not require deletion of autoreactive T cell clones by cDC1.

Moreover, TCRα deep sequencing of fixed TCRβ Batf3^+/+ and Batf3^−/− mice revealed strikingly similar repertoires among CD8⁺ T cells resident in SLO. If a critical function of cDC1 under homeostatic conditions is to prune the autoreactive CD8⁺ T cell pool through cross-presentation of acquired tissue-specific antigens, then we would have expected to find a large number of unique TCRα sequences among CD8⁺ T cells from Batf3^−/− mice. Clearly, this was not the case (Figure 5). Our data do not rule out the possibility that cDC1 control the fate of very rare self-reactive CD8⁺ T cell clones, although our best candidate CD8⁺ T cell clone was not deleted in a cDC1-dependent manner (Figure 6). However, even if cDC1 do control fates of very rare autoreactive CD8⁺ T cell clones via deletion, this does not appear to be required for the maintenance of CD8⁺ T cell tolerance, as we failed to observe enhanced auto-reactivity among polyclonal CD8⁺ T cells that developed in a cDC1-deficient environment upon adoptive transfer into secondary hosts, even in the context of an inflammatory stimulus, such as in vivo administration of an agonist α-CD40 antibody (Figure 4). Taken together, our data demonstrate that the contribution of DCs to the maintenance of peripheral tolerance through deletion of self-reactive CD8⁺ T cells appears to have been over-estimated based on results obtained from reductionist models.

However, we do not discount other potential roles that DCs may play in maintaining tolerance. It is possible that constitutive cross-presentation of self-antigens by cDC1 leads to modulation of TCR signal strength in self-reactive clones, a tolerance mechanism that would not have been uncovered in our transfer experiments, TCR repertoire sequencing, or analysis of TCR signaling by the polyclonal CD8⁺ T cell population as a whole. A comprehensive study of all potential tolerance mechanisms enforced by cDC1 was beyond the scope of this work, which focused specifically on peripheral CD8⁺ T cell deletion. Perhaps a future study involving the unbiased retrogenic expression of many putative self-reactive TCRs could elucidate potential tolerance mechanisms conferred by cDC1 on a clonal level³⁵.

Although our results do not exclude the possibility that cDC2 are important for peripheral deletion of tissue antigen-specific CD8⁺ T cells, this seems highly unlikely based on the well-established functions of cDC1 in cross-presenting antigens to CD8⁺ T cells^{19–26,38–40}. In addition, cDC2 are highly specialized in presenting antigens on MHC-II^26,38, and producing cytokines leading to productive Th2 ⁴¹ and Th17 ⁴² responses. These canonical cDC2 functions are the result of a distinct genetic program^43,44 and inherent differences in cell biology affecting antigen presentation pathways in cDC2 versus cDC1⁴⁵. A potential caveat to our findings is the fact that small numbers of residual cDC1 in Batf3^−/− mice could confound the data^46,47. Future studies investigating DC-mediated peripheral deletion should involve the use of the recently characterized Wdfy4^−/− mice to specifically ablate cross-presentation of cell-associated antigens, without affecting cDC1 development as is the case in Batf3^−/− mice⁴⁸. However, combined with the many studies demonstrating that CD8⁺ T cell cross-priming is abolished in Batf3^−/− mice^19–24, our finding that peripheral deletion of OT-I T cells does not occur in Batf3^−/− RIP-mOVA⁺ mice (Figure 1) argues against the existence of a compensatory APC population that cross-presents tissue-specific self-antigens in Batf3^−/− mice. The high affinity of the OT-I TCR for SIINFEKL:K^b is not necessarily representative of endogenous autoreactive CD8⁺ T cells that escape thymic negative selection, which generally have lower affinity for cognate self-ligands^11,32, rendering them less susceptible to peripheral deletion than OT-I T cells in RIP-mOVA⁺ mice, in which OVA is expressed at very high levels³³.

Further, while we contend that peripheral deletion of CD8⁺ T cells does not substantially contribute to self-tolerance, we do not discount its potential role in other biologically meaningful contexts, such as the systemic introduction of a foreign antigen into the blood, or a similar encounter with antigens derived from circulating leukemia cells²⁴. Rather, we hypothesize that dominant peripheral tolerance mediated by T_reg cells, as well as antigenic ignorance and modulation of TCR signaling by antagonists such as CD5 are sufficient to prevent activation and tissue damage induced by self-reactive CD8⁺ T cells that escape negative selection in the thymus. Our work highlights the importance of re-examining basic immunological principles with modern technologies and in biologically relevant contexts.

Key Points.

Batf3-lineage cDC1 are required for peripheral CD8⁺ T cell deletion.

The absence of cDC1 does not increase CD8⁺ T cell autoreactivity or alter the TCR repertoire.

Peripheral deletion by cDC1 does not significantly contribute to CD8⁺ T cell tolerance to self.