To investigate the development of QFL specific T cells, we used tetramer enrichment of lymphocytes using Qa1^b tetramers loaded with the FL9 peptide (31) (hereafter called QFL tetramers). To increase the specificity of detection, we co-stained using both the QFL tetramer and an antibody specific for Vα3.2, which recognizes the invariant TCRα chain used by the majority of QFL T cells ( Supplementary Figures 1A, B ; Figure 1A ) (41). In this study, we focused on QFL tetramer⁺ and Vα3.2⁺ cells, hereafter called QFL T cells. The majority of QFL T cells in the thymus, spleen and small intestine (SI) intraepithelial lymphocyte (IEL) compartment of wild type mice were CD8α⁺CD4- ( Figure 1A ). Interestingly, while mature QFL T cells in thymus and spleen predominantly expressed the CD8αβ heterodimer, QFL T cells in the IEL compartment were a mixture of cells expressing CD8αα or CD8αβ ( Supplementary Figures 1C–E ). As previously reported (31), QFL T cells were relatively abundant in the spleen of wild type mice (~1300/spleen/mouse or 1/6000 CD8 T cells, Figures 1B, C ). For comparison, a study of conventional CD8 T cell frequencies reported a range of 1/30,000 to 1/160,000 (42, 44). Additionally, a substantial number of QFL T cells were identified in the thymus and IEL compartment of the small intestine, with an average of 240 QFL T cells and 1,200 QFL T cells respectively ( Figure 1B ). The frequency of QFL T cells out of mature CD8 T cells was higher in the SI IEL compared to the thymus and spleen ( Figure 1C ), suggesting that they undergo selective recruitment and/or expansion in this compartment.

As a complimentary method to characterize QFL T cells, we developed a TCR transgenic mouse that expresses the semi-invariant QFL TCR α-and β-chain (Vα3.2Jα21, Vβ1Dβ1Jβ2-7) used by a predominant clone (41), henceforth referred to as QFLTg. As expected, mature (CD24-) thymocytes expressing the QFL TCR were predominantly CD8 single positive (SP) ( Supplementary Figure 1A ). DP thymocytes from transgenic mice expressed relatively low levels of the QFL TCR, while CD4SP thymocytes expressed moderately high levels of the TCR but remained mostly immature (CD24+) ( Supplementary Figure 2A ). To confirm that selection of the QFL CD8SP T cells in these mice was driven by the transgene encoded TCR, we examined expression of Vα2, which is expressed on ~10% of T cells in wild type mice and serves as a representative of endogenous TCRα expression. As expected, mature CD8SP in thymus and spleen of QFLTg mice had reduced expression of Vα2, whereas mature CD4SP has levels of Vα2 similar to wild type, confirming that CD8SP are mostly selected using the QFL TCR, whereas CD4SP are selected using endogenous TCRs ( Supplementary Figures 2B, C ). Importantly, the population defined as QFL CD8SP in this study (gated: QFL tetramer+Vα3.2+CD8+CD4-) was almost completely devoid of endogenous TCRα expression. These data validate the efficacy of our TCR transgenic system and confirms that QFL CD8SP in transgenic mice are selected by the transgene-encoded TCR.

We also used thymocytes and splenocytes from QFLTg mice to confirm the specificity of the QFL TCR for the FL9-Qa1^b complex. As expected, the majority of QFL tetramer+ thymocytes and splenocytes did not stain with tetramers in which Qa1^b was loaded with the predominant QDM peptide ( Supplementary Figure 2D ). The small population of QFL tetramer+ cells that did stain with the QDM tetramer also expressed the QDM/Qa1^b receptor NKG2A, an NK receptor that is also expressed by activated T cells. The co-staining of QDM tetramer and NKG2A suggests TCR independent binding of QDM tetramer on QFL T cells and serves as a positive control for QDM tetramer staining. These data confirm the specificity of the QFL TCR for the FL9/Qa1^b complex.

Previous studies showed that QFL T cells respond to a self-peptide presented by Qa1^b (31). In addition, splenic QFL T cells from wild type mice display an antigen experienced phenotype, suggesting that they may receive strong TCR signals during their development in the thymus. To test this notion, we examined expression of CD5, a marker which positively correlates with self-reactivity (45–48). As predicted, CD5 is elevated in QFL CD8SP thymocytes from QFLTg and non-transgenic mice compared to conventional CD8SP T cells ( Figure 1D ). In addition, QFL CD8SP thymocytes from non-transgenic mice showed slight but detectable downregulation of CD8β compared to conventional CD8SP T cells, whereas QFL CD8SP thymocytes from QFLTg mice showed more pronounced CD8β downregulation ( Figure 1E ). This modulation of CD8β expression has been associated with thymocyte self-reactivity and agonist selection (49, 50). QFL CD8SP thymocytes from QFLTg mice also showed elevated levels of several markers associated with agonist selection, such as the transcription factors PLZF (51, 52) and Tbet ( Supplementary Figure 3 ). On the other hand, splenic, but not thymic, QFL CD8SP T cells express elevated levels of CD44, a marker associated with antigen experience ( Figure 1F ; Supplementary Figures 3A–C ). In addition, QFL CD8SP thymocytes from QFLTg mice did not show detectable upregulation of PD1 or α4β7, markers that are expressed by a subset of thymic IEL precursors ( Supplementary Figures 3A–C ) (22). Taken together these data suggest that QFL T cells experience relatively strong TCR stimulation and undergo agonist selection during their development in the thymus.

Some agonist selected T cell populations, such as regulatory T cells require a separate positive selection interaction prior to undergoing agonist selection (6). Previous reports that QFL T cells are detectable in the spleen of mice lacking Qa1^b, but undetectable in mice lacking β2m (31), a subunit of MHC I which is required for proper folding and surface expression of both classical MHC Ia and Qa1 (53), suggested the possibility that QFL T cells require positive selection on classical MHC Ia. To test this hypothesis, we generated K^bD^bKO mice and compared the number of QFL T cells in the thymus and spleen to that of WT and Qa1^bKO mice ( Figure 2A ; Supplemental Figures 4A, B ). The QFL CD8SP thymocytes were slightly reduced in K^bD^bKO and Qa1^bKO relative to wild type mice, but were undetectable in β2mKO mice ( Figure 2A ). Similar results were obtained with QFLTg mice, with substantial numbers QFL thymocytes and splenocytes found in the absence of Qa1^b or K^bD^b, but not in the absence of β2m ( Figure 2B ). Importantly, the QFL CD8SP thymocytes and splenocytes from QFLTg, QFLTgQa1^bKO mice have negligible expression of endogenous Vα2 ( Supplemental Figures 2B, C ), confirming that these cells were positively selected using the QFL TCR. Classical MHC Ia D^b is the source for the QDM peptide that is bound to Qa1^b in ERAAP sufficient cells, raising the possibility that loss of D^b could indirectly impact QFL T cell development by altering the peptides displayed on Qa1^b. However, QFL CD8SP thymocytes and splenocytes were found in similar numbers in K^bKO, D^bKO, and K^bD^bKO mice, arguing against this possibility ( Supplementary Figure 4C ). Altogether, these data suggest that neither classical MHC Ia, nor Qa1^b, are required for QFL T cell positive selection, although both may contribute to the efficiency of the process. Interestingly, CD8SP T cells in spleens of K^bD^bKO mice exhibit a higher frequency of Vα3.2⁺ cells compared to WT or Qa1^bKO mice ( Supplementary Figure 4D ), indicating that this V segment is preferentially used by T cells reactive to MHC Ib molecules.

Because Qa1^b presents agonist FL9 peptide to QFL T cells, we hypothesized that expression of Qa1^b might lead to agonist and negative selection of QFL thymocytes. In support of this, DP thymocytes in QFLTg (Qa1^b sufficient) mice exhibit reduced cellularity and a “DP^lo” phenotype associated with strong TCR signals (19, 54–56) whereas QFLTg Qa1^bKO mice express normal levels of CD4 and CD8α ( Figure 2C ). In addition, QFL CD8SP thymocytes from Qa1^b sufficient, but not Qa1^bKO mice, displayed CD8β downregulation, PLZF expression, and elevated CD5 expression compared to conventional mature CD8SP thymocytes ( Figures 2D-F ). In contrast, in the absence of classical MHC I (K^bD^bKO mice) QFL CD8SP T cells showed strong downregulation of CD8β expression, maintained PLZF expression and showed a slight reduction in CD5 expression compared to WT mice ( Figures 2D-F ). In the periphery, QFL T cells lost their antigen experienced phenotype in absence of Qa1^b, but not in absence of classical MHC I ( Supplemental Figure 4E ). These data suggest that Qa1^b is required for agonist selection of QFL T cells, although positive selection of QFL T cells can be driven by an alternative MHC 1 molecule.

To further explore the ligand-specificity of the QFL TCR we took advantage of the observation that MHC-naïve DP thymocytes are highly sensitive to in vitro TCR stimulation (57, 58). We examined expression of activation markers on pre-selection QFLTg (preQFLTg) thymocytes from a β2mKO background after co-culture with bone marrow derived dendritic cells (BMDC) isolated from mice lacking either Qa1^b, K^bD^b or β2m. PreQFLTg thymocytes showed stronger upregulation of the activation markers CD69 and CD5 upon 24-hour co-culture with WT and K^bD^bKO, compared to Qa1^bKO, BMDC ( Figure 2G ; Supplementary Figure 5A ). This is consistent with the hypothesis that Qa1^b is the primary MHC ligand for the QFL TCR. Interestingly, preQFLTg thymocytes showed a modest activation when co-cultured with Qa1^bKO BMDCs; this activation was significantly more compared to co-culture with β2mKO BMDCs ( Figure 2G ). A similar pattern of reactivity was observed when preQFLTg were cultured in thymic slices derived from WT, Qa1^bKO and β2mKO mice ( Supplementary Figures 5B, C ). This is consistent with the development of QFL T cells in Qa1^bKO mice, and suggests that the QFL TCR is cross-reactive with an alternative β2m-utilizing molecule.

To further explore the specificity of the QFL TCR for MHC 1 molecules, we used the DC-like cell line DC2.4 as a stimulator cell for preQFLTg thymocytes ( Figure 2H ; Supplementary Figure 5D ). The response of preQFLTg thymocytes to DC2.4 cells was abolished by CRISPR/Cas9 mediated gene knock out of β2m and partially reduced by loss of Qa1^b, paralleling the results from stimulation with BMDC, and pointing to the recognition of an alternative MHC-1 ligand in this system ( Figures 2G, H ). Interestingly, triple KO of Qa1^b, K^b, and D^b in DC2.4 cells reduced activation close to background levels, whereas double knock out of Qa1^b with either K^b, or D^b led to stimulation that was intermediate between the triple KO and Qa1^b KO cell lines. Together these data suggest that classical MHC-1 molecules may contribute to the positive selection of QFL T cells in Qa1^bKO mice.

While conventional αβT cells undergo positive selection by recognition of MHC molecules on thymic epithelial cells, MAIT cells and iNKT cells undergo selection via interactions with hematopoietic cells (4, 5). To investigate the cell type requirements for selection of QFL T cells, we generated reciprocal bone marrow chimeric mice in which either the donor cells or the host cells are β2mKO, and therefore lack surface expression of Qa1, as well as the classical MHC I molecules H2-D, H2-K ( Supplementary Figure 6 ). Interestingly, comparable numbers of QFL T cells were found in the thymus and spleen of the β2mKO>WT and WT>β2mKO chimeric mice ( Supplementary Figures 6B, C ), implying that QFL T cell development could occur efficiently on either non-hematopoietic or hematopoietic cells. To confirm these results, we generated reciprocal β2mKO chimeras using donor cells that expressed the QFL TCR transgene ( Figure 3A ). While there was some reduction in QFL T cell number in the thymus of β2mKO>WT compared to WT>β2mKO and wild type control chimeras ( Figure 3B ), similar numbers of QFL T cells were found in the spleen ( Figure 3C ). Thus, QFL T cell development is not strictly dependent on either non-hematopoietic or hematopoietic expression of MHC I.

We also examined whether QFL T cells that are selected exclusively by hematopoietic or non-hematopoietic cells retained their agonist selected phenotype. QFL thymocytes exhibited comparable CD8β downregulation but decreased PLZF expression when MHC I was restricted to non-hematopoietic cells ( Figures 3D, E ; Supplementary Figure 6D ). Similarly, expression of CD5 was decreased when MHC I was restricted to non-hematopoietic cells ( Figure 3F ; Supplementary Figure 6E ). In the periphery, QFL T cells in chimeric mice that lacked MHC I on hematopoietic cells did not display elevated expression of CD44 ( Figure 3G ; Supplementary Figure 6F ). Overall, the thymic phenotype of QFL T cells in β2mKO>WT chimeras is similar, but not identical, to that observed in Qa1^bKO mice ( Figures 2D–F ). These data suggest that Qa1^b on both hematopoietic and non-hematopoietic cells contribute to agonist selection, with hematopoietic cells playing the predominant role.

To test how agonist selection impacts the functionality of QFL T cells, we compared QFL T cells that arose in the presence or absence of Qa1^b for their ability to respond in-vitro to ERAAPKO splenocytes. QFL T cells from QFLTg showed extensive upregulation of activation markers and increased proliferation in response to stimulation compared to QFL T cells from QFLTg Qa1^bKO mice ( Figures 4A, B ). This implies that agonist selection on Qa1^b led to greater functional responsiveness, which could reflect greater functionality on a per cell basis, or an increased frequency of functional cells within the population. Since agonist selection partially correlates with selection on hematopoietic cells ( Figure 3 ; Supplementary Figure 6 ), we also examined QFL splenocytes from reciprocal β2mKO and wild type bone marrow chimeric mice as a further test of the impact of agonist selection on function. QFL T cells from QFLTg>β2mKO mice (Hematopoietic cell (HC) selected) responded more robustly to ERAAPKO APCs compared to cells from QFLTg β2mKO>WT mice (non- HC selected) ( Figures 4A, B ). Thus, QFL T cells that develop in the absence of Qa1^b, or in the absence of hematopoietically expressed MHC I, exhibit reduced functionality.

QFL thymocytes display an agonist selected phenotype that is enhanced by hematopoietic expression of MHC I. Given that agonist selection in the thymus can give rise to natural intraepithelial lymphocytes, we considered that QFL thymocytes might represent a population of IEL precursors. To test this idea, we injected Rag2KO neonatal mice with CD4-depleted QFL thymocytes from HC-only selected (QFLTg>β2mKO) or both HC and non-HC selected (QFLTg>WT) chimeric mice ( Figure 4C ). For comparison, we also injected Rag2KO neonatal mice with CD4-depleted thymocytes or splenocytes from intact QFLTg mice. QFL T cells were found in similar numbers in the spleen and IEL compartment of the SI for all 4 donor populations ( Figure 4D ). Interestingly, IEL T cells derived from HC-only selected QFL thymocytes showed more pronounced downregulation of CD8β (~50%) when compared to IEL T cells derived from HC+ non-HC selected QFL T cells (~10%) ( Figure 4E ). Altogether, these data indicate that both QFL thymocytes and splenocytes contain IEL precursors, and that selection by hematopoietic cells favors CD8β downregulation in QFL IELs.

B6 (C57BL/6), B6 Ly5.1 (B6.SJL-Ptprca Pepcb/BoyJ) and Rag2-/- (B6(Cg)-Rag2tm1.1Cgn/J) mice were from Jackson Labs. B2M-/- (B6.129-B2mtm1Jae N12) mice were from Taconic. Qa1^bKO mice (70) were obtained from the Shastri lab. TCR transgenic mice specific for FL9-Qa1^b (QFL) and H-2K1/H-2D1-/- (K^bD^bKO) mice were generated in our lab (described below). All mice were bred in the UC Berkeley animal facility and all procedures were approved by the Animal Care and Use Committee (ACUC) of the University of California.

The TCR alpha and beta chain sequences from the QFL specific BEKo8z hybridoma (31, 41) were cloned and amplified from the genomic DNA of the BeKoz Hybridoma. The TRAV9N/D-3 TCR alpha chain was cloned with the forward primer (5’ AAAACCCGGGCCAAGGCTCAGCCATGCTCCTGG) with an added XmaI cutting site at 5’ end of the DNA sequence and a reverse primer for TRAJ21 (5’ AAAAGCGGCCGCATACAACATTGGACAAGGATCCAAGCTAAAGAGAACTC) with an added Not1 cutting site at the 5’ end of the DNA sequence. The TCR beta chain was cloned with the forward primer (5’ AAAACTCGAGCCCGTCTGGAGCCTGATTCCA) with and added Xho1 cutting site at the 5’ end of the DNA and a reverse primer for TRBJ2-7 (5’ AAAACCGCGGGGGACCCAGGAATTTGGGTGGA) with a SacII cutting site flanking the 5’ end of the DNA sequence. The cloned TCR alpha chain was cloned into pTα cassette vector by inserting it between the Xmal and Not1 sites, while the TCR beta chains were cloned into pTβ cassette vector in between the Xhol and SacII sites (71). The ampicillin resistance gene was removed from pTα and pTβ cassette by EarI enzyme digest. The QFL transgenic mice were generated on the B6 background in the Cancer Research Laboratory Gene Targeting Facility at UC Berkeley under standard procedures. The QFL mice were maintained on the B6 background and bred once with B6 5.1 mice to generate (QFLTgxB65.1/2) background mice for use in experiments. Founder mice were identified by flow cytometry and PCR genotyping of tail genomic DNA using primers mentioned above.

K^bD^bKO were generated by the Gene Targeting Facility at UC Berkeley using Cas9/CRISPR-mediated gene targeting. The H-2K1 gene was targeted using an sgRNA (5’ GTACATGGAAGTCGGCTACG 3’) that aligned with the sense strand and the H-2D1 gene was targeted using an sgRNA (5’ AGATGTACCGGGGCTCCTCG 3’) that aligned with the antisense strand. Wild-type C57BL/6J mice were originally obtained from the Jackson Laboratories. Zygotes were obtained from super ovulated C57BL/6J females for CRISPR/Cas9 targeting knockout experiment. In brief, CRISPR mix (i.e., Cas9 protein and sgRNAs) was introduced to zygotes by electroporation as previously described (72).The embryos were then transferred to 0.5dpc pseudo pregnant females (CD-1, Charles River Laboratories) with oviduct transfer. When the pups were born, tails samples were collected for DNA extraction and genotyping. The resulting founder mice were identified by flow cytometry. The H-2K1 gene had a 2bp deletion (5’ TGCCTGGGCTTTCTGTGTCTCCCGCTCCCAATACTCGGGCCCCTCCTGCTCCATCCACCGC GCCCGCGGCTCATATCTCGGATTCTCCGCGTCGCTGTCGAAGCGCACGAACTCCGTGTCGTCCACGT—CCGACTTCCATGTACCGGGGCTCCCCGAGGCCGGGCCGGGACACGGCGGTGACGAAATACCTCAA 3’) where the sgRNA targeted. The H-2D1 gene had a 15bp deletion where the sgRNA targeted (5’ CCGTNGGGTCGTTCTGTTCCAAACCTCGGACTTGGGACCCGGGACGTCAGCGTCCCTGTGTCGGGAAGTGGAGGGGCCTGACCTCCCACGCGGGGTCACTCACCGCCCGCGCTCTGGTTGTAGTAACCNAGCAGGTTCCTCAGGCTCACTCGGAACCACTGCTCTTGGGCCTTGGNTTTCTGTGTTTCCCGCTCCCAATACTCCGGCCCCTCCTGCTCCATCCACGGCGCCCGCGGCTCATATCTCGGATTCTCCGCGTCGCTTTCGAAACGCACGAACTCCTTGTTGTCCACATAGCCAACAGAGATGTACCGGGGC—————CGGGACACGGCGGTCTCGAAATACCGCATCGAGTGTGGGCCTGGGGACGGCGCGCGGTGAGACCCCGACCTCCTCACCAAACCCCGGGCGGCTGCGCACGCCGGGAGGGGATCTGGGCGCGGGGCTCAGGTGGAGAAGGGGCGGAGGGTCCGNGGGGGCGACGA 3’).

Thymi, and spleens were mechanically dissociated in FACS buffer (0.5% BSA in PBS) or complete RPMI (10% FBS) to generate single-cell suspensions that were then passed through a 70μm filter. Intraepithelial lymphocytes (IELs) were isolated from the small intestine as previously described (73). Briefly, small intestine was cut to 1cm pieces and washed with cold CMF. Tissue pieces were allowed to settle and CMF was poured off. The tissue was then digested with DTE solution for 30 min at 37C in a 50mL conical tube. Tissue pieces were centrifuged at 1,500rpm for 5 min at 4C. Supernatant was collected and centrifuged at 1,500rpm for 5min at 4C. Lymphocytes were isolated by percoll separation utilizing 40% and 80% percoll (22). The percoll solution was centrifuged at 2000rpm with no brake for 20 min at room temperature. Lymphocyte layer was then washed with PBS and collected. Splenocytes were then RBC lysed using ACK lysis buffer (0.15M NH4Cl, 1mM KHCO3, 0.1mM Na2EDTA) for 5 minutes at room temperature.

Thymi, spleens, and IELs were stained in 2.4G2 supernatant for 30 minutes at 4°C with the following antibodies: (BD Biosciences) CD4 (RM4-4), CD8α (53-6.7), CD5 (53-7.3), PLZF (R17-809), (Biolegend) TCRβ (H57-597), CD8β (YTS156.7.7), B220 (RA3-6B2), Vα3.2 (RR3-16), CD45.2 (104), T-bet (4B10),(Invitrogen) CD8β (H35-17.2), CD24 (M1/69), integrin α4β7 (DATK32), CD69 (H1.2F3) (Tonbo) CD44 (IM7), and CD45.1 (A20). Cells were then washed in PBS and stained in Ghost Dye Violet 510 as described above. For intracellular staining, cells were fixed and permeabilized using the eBioscience FoxP3/Transcription Factor Staining Buffer Set (ThermoFisher) according to manufacturer’s instructions. Biotinylated peptide-MHC monomers were obtained from the NIH Tetramer Facility (Atlanta, GA). Tetramers were assembled by conjugating the biotin-labeled monomers with PE-labeled streptavidin (Agilent, #PJRS27-1) according to NIH Tetramer Facility protocols. Cell numbers were calculated using AccuCheck Counting Beads for count and pipetting accuracy (Life Technologies #PCB100) according to manufacturer’s instructions. All antibodies were from BD Biosciences, Biolegend, Invitrogen, or Tonbo Biosciences. Samples were processed using a Fortessa X20 (BD Biosciences) and analyzed using FlowJo software. For defining CD8α+CD8βlow thymocyte population, we adjusted the gate based on wild type controls samples analyzed in parallel to experimental samples, such that 5-6% of the wild type CD8SP cells were CD8βlow. This gate was then applied to our experimental samples (as shown in Figure 1E ). Similarly, the CD8α+CD8βlow gate in the IEL population was placed based on wild type samples run in parallel and then applied to experimental samples ( Supplementary Figure 1C ).

Single-cell suspensions of thymi and spleens were generated as described above. Cells were incubated in 50nM Datsatinib (Sigma Aldrich, CDS023389-25MG) for 30 minutes at 37°C and then stained with tetramer in 2.4G2 supernatant for 1 hour at room temperature. After staining, cells were washed and incubated with Anti-PE MicroBeads (Miltenyi Biotec, #130-048-801) in MACS buffer (0.5% BSA) for 30 minutes at 4°C. Cells were then positively enriched for tetramer+ T cells using a magnetic column (Miltenyi Biotec.) according to manufacturer’s instructions and washed before extracellular staining.

Bone marrow cells were harvested as described above and RBC lysed using ACK lysis buffer (0.15M NH4Cl, 1mM KHCO3, 0.1mM Na2EDTA) for 5 minutes at room temperature. Bone marrow cells were resuspended in cRPMI and seeded at 5 x 10⁶ cells per milliliter in 24 well plates. Cells were supplemented with GM-CSF (Peprotech, #315-03-20UG) until day 4 and adhering cells were harvested on day 6 using EDTA. 6 x 10⁵ CD11c+MHC-II+ bone marrow cells per milliliter were seeded in 24 well plates. Preselection QFL thymocytes were generated by crossing QFL TCR transgenic mice onto a non-selecting, MHC-I deficient background (β2M-/-). Thymic single-cell suspensions were generated as described above. Thymocytes were resuspended in cRPMI and seeded at 4 x 10⁶ cells per milliliter.

DC 2.4 cells were obtained from UC Berkeley’s Cell Culture Facility. DC 2.4 cells were transduced using the lentiCas9-Blast vector (Addgene plasmid # 52962; http://n2t.net/addgene:52962; RRID: Addgene_52962) and selected with 10 ug/mL of Blasticidin (AG Scientific #3513-03-9). Guide-RNA sequences targeting mouse β2m, Qa-1^b, H2-K^b, and H2-D^b were selected using CRISPick (portals.broadinstitute.org/gppx/crispick/public), and sequences are as follows: β2m: TCACGCCACCCACCGGAGAA, Qa-1^b: TACTACAATCAGAGTAACGA, H2-K^b: GTACATGGAAGTCGGCTACG, H2-D^b: AGATGTACCGGGGCTCCTCG. Guide-RNAs were stably transduced in SpCas9-expressing DC 2.4 cells using the LentiGuide-Neo vector and selected with 500 ug/mL of G418 (InvivoGen #108321-42-2). LentiGuide-Neo was a gift from Caroline Goujon (Addgene plasmid # 139449; http://n2t.net/addgene:139449; RRID: Addgene_139449). DC 2.4 cells that were negative for surface β2m, Qa-1^b, Qa-1^b x H2-D^b, Qa-1^b x H2-K^b, and Qa-1^b x H2-D^b x H2-K^b were independently sorted out on BD FACSAria Fusion (BD Biosciences).

WT, Qa1^bKO, Qa1^bK^bD^bKO, Qa1^bK^bKO, Qa1^bD^bKO, and β2mKO DC2.4 cells were plated in 48 Well Cell Culture Plates (Corning, Cat. No:3548) at 3x10⁵ cells per well. Cells were left to settle for 1 hour at 37°C 5% CO₂. DC2.4 cells were then treated with 5ng/ml Recombinant Mouse IFN-γ (Biolegend, Cat. No:575302) for 24 hours, then Recombinant Mouse IFN-γ was washed off. Single-cell suspensions of QFLTgβ2mKO thymi were prepared as described above. Thymocytes were overlaid at 1x10⁵ cells per well and cultured at 37°C 5% CO₂ for 24 hours, then harvested for flow cytometric analyses.

Thymic lobes from wild type, Qa1^bKO and β2mKO mice were gently isolated and any connective tissue was removed. Lobes were embedded into 4% agarose with a low melting point (GTG-NuSieve Agarose, Lonza) and sectioned into 400-500mm slices using a vibratome (VT1000S, Leica). Thymic slices were overlaid onto 0.4mm transwell inserts (Corning, Cat. No.: 353090) in 6 well tissue culture plates with enough cRPMI under the insert to reach the slices. Pre-selection QFL (2.5x10⁵) thymocytes were overlaid onto each slice and allowed to migrate for 3 hours, after which excess thymocytes were removed by gently washing with PBS. Slices were cultured at 37°C 5% CO2 until harvested for analysis. For flow cytometry, thymic slices were mechanically dissociated into single-cell suspensions prior to staining.

Host mice were depleted of NK cells by I.P. injecting anti-NK1.1 (PK136, Leinco Technologies, #N123) at 100ug/100uL every 24Hrs for two days, for a total of 200ug of depleting antibody. Mice were irradiated in two doses of 600 rads (total of 1,200rads), with a resting period of 16hrs between doses. Mice were maintained on antibiotic water (Trimethoprim/Sulfamethoxazole) 4 weeks following irradiation. Bone marrow was harvested from the femur of donor mice using standard techniques. Red blood cells were lysed using ACK lysis buffer (0.15M NH4Cl, 1mM KHCO3, 0.1mM Na2EDTA) for 5 minutes at room temperature. Cells were depleted of CD4+ T cells by staining with CD4 PE-conjugated antibody (RM4-4) for 20 minutes at 4°C and then with Anti-PE MicroBeads (Miltenyi Biotec.) as described above. The labeled cells were washed, resuspended in MACS buffer, and then passed through a magnetic column (Miltenyi Biotec.). Flow-through (CD4-depleted bone marrow cells) was washed, resuspended at (4x10⁶ cells) in 100uL of PBS and i.v. injected into recipient mice. Bone marrow chimeras were analyzed 8-11 weeks following reconstitution.

Cells were resuspended in 5uM CFSE proliferation dye (ThermoFisher #C34554) and incubated at 37°C for 9 minutes. Cells were then washed by addition of pre-warmed cRPMI while vortexing. Cells were then washed again with pre-warmed PBS and resuspended at the desired concentration.

Single cell-suspensions of splenocytes were generated and RBC lysed as described above. Antigen presenting cells (APCs) were prepared by depleting splenocytes of CD4, CD8, and NK1.1 expressing cells using a magnetic column as described above. APCs were seeded at 4x10⁵ cells per well in a 48 well plates. Responding cells were splenocytes from QFLTg mice that were depleted of CD4, NK1.1, B220, and CD19 expressing cells using a magnetic column as described above. Cells were then CFSE (ThermoFisher #C34554) labeled as described above and seeded at 1x10⁵ cells per well.

Donor thymocytes and splenocytes were isolated as described above. Single cell suspensions were depleted of CD4+ T cells, B cells and NK cells by staining with CD4 (RM4-4), CD19 (1D3), and NK1.1(PK136) PE-conjugated antibody for 20 minutes at 4°C and then with Anti-PE MicroBeads (Miltenyi Biotec) as described above. The labeled cells were washed, resuspended in cRPMI buffer, and then passed through a magnetic column (Miltenyi Biotec). Flow-through (CD4, CD19, NK1.1 depleted cells) was washed, resuspended at (1x10⁶ cells) in 100uL of PBS and intrahepatically injected into 4–6-day old Rag2KO recipients. Recipient mice were analyzed 14-15 weeks post-injection.