In the case of CD8⁺ T lymphocytes, the population of CD44^hi CD62L^lo but not CD44^hi CD62L^hi cells expands in response to immunization with foreign Ags (24). Indeed, the former cell compartment is large in feral mice while small in animals housed under SPF conditions (25). As a first step in comparing MP CD4⁺ T lymphocytes with foreign Ag-driven memory cells, we wished to determine if a similar foreign Ag-driven expansion of the CD44^hi CD62L^lo population occurs in CD4⁺ T cells as well. To do so, we infected SPF C57BL/6 mice with lymphocytic choriomeningitis virus (LCMV) Armstrong and waited for 6 weeks to generate foreign Ag-specific memory cells defined by tetramer staining. The CD44^hi CD62L^lo but not CD44^hi CD62L^hi CD4⁺ T-cell fraction was found to be significantly larger in infected versus uninfected control animals, while the naïve (CD44^lo CD62L^hi) cell compartment if anything was decreased in size ( Figure 1A ). As expected, viral Ag GP66-specific as well as NP309-specific memory CD4⁺ T cells were detected exclusively in the infected CD44^hi CD62L^lo cell population ( Figures 1B , Figure S1A ). These findings suggested that the CD44^hi CD62L^lo CD4⁺ T-cell population seen in infected animals represents a mixture of preexisting MP cells and LCMV-driven foreign Ag-specific memory cells.

To search for markers that are differently expressed in MP and foreign Ag-specific memory T lymphocytes, we compared the CD44^hi CD62L^lo cell population preexisting in the SPF environment with GP66-specific memory cells in infected mice by means of single-cell RNA sequencing (scRNAseq). Uniform Manifold Approximation and Projection (UMAP) analysis revealed that these two populations are transcriptomically distinct ( Figure 1C ), and subsequent scoring analysis indicated that expression of genes encoding positive regulators in the cell cycle (26) was higher in MP versus foreign Ag-specific memory cells, while the opposite was true for negative regulators ( Figure 1D ). Consistent with this finding, the uninfected CD44^hi CD62L^lo cell population displayed significantly higher Ki67 expression than did its counterpart in infected mice ( Figure 1E ), and closer analysis of the latter cell population revealed that GP66-specific memory cells are almost all Ki67^neg ( Figure 1F ). The above observations are in agreement with previous findings demonstrating that MP cells are rapidly proliferating while foreign Ag-specific memory cells are quiescent (22) and confirm the scRNAseq data shown in Figure 1C . Our data also support the hypothesis that the CD44^hi CD62L^lo CD4⁺ T-cell population is flexible in size, with CD44^hi CD62L^lo cells from infected mice, and especially their tetramer^neg subpopulation, representing a mixture of MP and foreign Ag-specific memory T lymphocytes since their Ki67⁺ fraction was intermediate in magnitude between those observed in uninfected CD44^hi CD62L^lo and infected foreign Ag-specific memory cells (as defined by GP66-tetramer staining) ( Figures 1E, F ).

Through further scoring analysis of the above scRNAseq data, we found that the expression of genes associated with memory T-cell formation (27) was significantly lower in the MP compared to the foreign Ag-specific memory compartment ( Figure 1D ). Because CD127 (IL-7 receptor α chain) and Sca1 (also referred to as Ly6a) are known to be expressed on foreign Ag-specific memory T lymphocytes (12, 28–30), we next compared their expression on CD4⁺ T cells in uninfected versus infected animals. MP (and minor population of CD44^hi CD62L^hi) cells in uninfected mice were found to consist of CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo subsets, while naïve cells were largely CD127^hi Sca1^lo ( Figures 1G , S2A ). In CD44^hi CD62L^lo cells, the CD127^hi Sca1^hi but no other subpopulations increased in size in infected CD4⁺ T lymphocytes ( Figure 1G ). These results suggested that in comparison with heterogeneous MP cells, foreign Ag-specific memory cells are CD127^hi Sca1^hi. Consistent with this hypothesis, GP66- as well as NP309-tetramer⁺ memory cells were essentially all CD127^hi Sca1^hi, while the tetramer^neg cells that presumably represent a mixture of MP and foreign Ag-specific memory populations as described above contained a frequency of CD127^hi Sca1^hi cells intermediate between that in uninfected MP and tetramer⁺ memory cells ( Figures 1H , S1B ). Furthermore, in CD44^hi CD62L^lo CD4⁺ T lymphocytes, the CD127^hi Sca1^hi subset had the lowest Ki67 expression in both infected and uninfected mice ( Figures 1I, J ). Together, these data demonstrate that MP cells preexisting in an SPF environment are subdivided into four different populations based on CD127 and Sca1 expression, while foreign Ag-specific memory T lymphocytes are all CD127^hi Sca1^hi, with both populations constituting the CD44^hi CD62L^lo CD4⁺ T-cell compartment at homeostasis.

Because foreign Ag-driven memory cells are CD127^hi Sca1^hi and this phenotype is partially shared by MP cells ( Figures 1G, H ), we sought to determine whether or not MP lymphocytes with the same phenotype represent a subpopulation of foreign Ag-specific memory cells. For this purpose, we re-analyzed the scRNAseq data obtained in Figure 1C and found that expression of anti-apoptotic genes (31) was significantly lower in MP versus foreign Ag-specific memory cells ( Figure 2A ). Given that Bcl2 plays a critical role as an anti-apoptotic factor in T lymphocytes (32), we measured its protein expression levels in these two populations. While CD127^hi Sca1^hi MP cells were largely Bcl2^lo, ~90% of CD127^hi Sca1^hi GP66- and NP309-tetramer⁺ memory T lymphocytes were Bcl2^hi ( Figures 2B , S1C ). Thus, CD127^hi Sca1^hi MP cells appear to represent a unique population that is distinct from foreign Ag-specific memory T lymphocytes in terms of Bcl2 expression. In addition, we observed a small fraction (~10%) of Bcl2^hi cells in CD127^hi Sca1^hi MP CD4⁺ T cells, and this fraction was more enriched in CD44^hi CD62L^hi cells ( Figures 2B , S2B ). The significance of this finding will be discussed later.

To further investigate whether the development of the CD127^hi Sca1^hi MP subset is foreign Ag-dependent or Ag-independent, we examined MP cells from SPF, GF, and antigen-free (AF) mice, the latter being deprived of both food and commensal Ags (33). MP as well as CD44^hi CD62L^hi and naïve CD4⁺ T cells were essentially intact in SPF, GF, and AF mice ( Figure 2C ). Surprisingly, CD127^hi Sca1^hi MP cells were largely unchanged in the three animal groups ( Figure 2D ). In addition, Ki67 expression was not significantly increased in GF or AF mice, and Bcl2 levels were not decreased and instead were elevated in the former animals ( Figures 2E, F ). This was also the case in the minor CD44^hi CD62L^hi CD4⁺ T-cell population ( Figure S2C, D ). Thus, CD127^hi Sca1^hi Bcl2^lo MP cells can be generated in the absence of foreign Ags, presumably in response to self Ags.

The above results identify CD127, Sca1, and Bcl2 as markers that are differently expressed in foreign Ag-specific memory versus MP cells and indicate that MP CD4⁺ T lymphocytes comprise 4 distinct subsets: CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo. Because CD127^hi Sca1^hi MP cells are present in almost equal numbers in SPF, GF, and AF mice, self Ags are thought to be the major stimuli for the generation of CD127^hi Sca1^hi as well as the other MP subsets in an SPF environment. To address how the four MP subpopulations defined by these phenotypic markers are generated and maintained, we measured CD127 and Sca1 levels in MP cells from mice of different ages. As expected, MP cells were rare in 1-week-old animals, and most of these cells were CD127^hi Sca1^lo ( Figures 3A, B ). Thereafter, the proportion of CD127^hi Sca1^hi lymphocytes increased progressively with age, in parallel with the total MP pool size.

In the case of CD8⁺ MP cells, a subpopulation referred to as innate memory cells are directly generated in the thymus (34). To ask whether CD127^hi Sca1^hi CD4⁺ MP cells arise in the thymus, we prevented lymphocyte egress from the thymus by injecting mice with FTY720 for 2 weeks. As expected, numbers of both CD4⁺ and CD8⁺ single-positive thymocytes accumulated in the thymus ( Figure S3A ), and in parallel, naïve CD4⁺ T cells decreased in number in secondary lymphoid tissues ( Figures S3B, C ). By contrast, peripheral MP CD4⁺ T lymphocytes as well as their four subfractions were largely unaffected by the treatment ( Figures S3B–D ), suggesting that MP cells are self-maintained in the periphery once generated. Consistent with this conclusion, the four MP subpopulations were not altered by adult thymectomy ( Figure S3E ). Based on these results, it is unlikely that peripheral CD127^hi Sca1^hi and other MP subpopulations are actively replaced by emigrants generated in the thymus.

To test the possibility that CD127^hi Sca1^hi MP cells are generated from peripheral naïve precursors, we transferred sorted naïve CD4⁺ T lymphocytes into wild-type (WT) recipients and analyzed the donor cells 1 to 4 weeks later. MP cells developed slowly with time ( Figure 3C ), with the CD127^hi Sca1^hi subset dominating by 4 weeks after transfer ( Figure 3D ). Thus, the four MP subsets including the CD127^hi Sca1^hi fraction appear to be generated from naïve T lymphocytes in the periphery.

To further examine the dynamics of MP cell maintenance following their development, we sorted for CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo MP subpopulations and transferred them individually to WT recipients. When the donor cell population was analyzed 2 weeks later, Sca1^lo cells were found to increase their Sca1 expression, while Sca1^hi cells remained Sca1^hi ( Figure 3E ). In addition, some CD127^hi cells converted to CD127^lo and vice versa ( Figure 3E ). Overall, these data argue that while naïve cells are CD127^hi Sca1^lo, as they differentiate into MP cells, they eventually acquire a CD127^hi Sca1^hi phenotype, either directly from CD127^hi Sca1^lo precursors or via CD127^lo intermediates. Consistent with this notion, when the levels of T-cell receptor (TCR) excision circles (TRECs) were measured in the four MP subpopulations sorted from intact mice, levels were the highest and lowest in CD127^hi Sca1^lo and CD127^hi Sca1^hi cell populations, respectively, with CD127^lo cells expressing TRECs at an intermediate level ( Figure 3F ). Together, these data identify CD127^hi Sca1^hi MP cells as the most mature of the four subsets, their generation reflecting extensive cell division of their precursors.

We previously reported that, as a whole population, MP cells become less dependent on TCR signaling once generated (4). This notion was based on the observation that treatment of SPF mice with either anti-I-Aβ monoclonal antibody (mAb) or cyclosporin A significantly inhibits the generation of MP cells from naïve precursors but has only a negligible effect on their steady-state proliferation. While these data established a lesser dependence of MP cell maintenance on Ag recognition, it was still unclear whether this function is completely independent of TCR signaling.

To address this issue, we analyzed the strength of TCR signaling that MP subpopulations receive at baseline. Initial experiments revealed that the four MP subsets express equivalent levels of TCRβ and CD3 as well as CD5 ( Figures 4A, B ), a marker that reflects TCR affinity to self Ags (35). Nevertheless, in Nur77-GFP reporter mice where the sum of TCR signal strength that T cells receive is reflected by reporter expression (36), CD127^lo Sca1^hi and CD127^lo Sca1^lo MP subpopulations showed significantly higher levels of GFP as well as CD69 than did their CD127^hi counterparts ( Figures 4C, D ). Given that CD127^hi and CD127^lo subsets are interchangeable in the steady state ( Figure 3E ), it is possible that the former MP subpopulation downregulates CD127 immediately after TCR ligation.

To test this hypothesis in vivo, we utilized CD4-CreERT2 TCRα^flox mice in which we previously established that ~50% of MP cells lose their TCR expression as a consequence of tamoxifen (TMX) treatment (5). Thus, by examining the four MP subfractions in TCRβ⁺ and TCRβ^neg MP subpopulations 10 days after TMX treatment, we could compare their steady-state proliferation in the presence or absence of normal levels of tonic TCR signaling in the same mouse. Using these animals, we found that Ki67 expression was significantly reduced by TCR ablation in all four MP subpopulations ( Figure 4E ). Moreover, the CD127^hi fraction increased while the CD127^lo decreased in size when TCR levels were reduced ( Figure 4F ). These results show that the four MP subpopulations proliferate in the presence of tonic TCR signaling, presumably delivered by self Ag recognition, and further suggest the existence of TCR-dependent interchangeability between CD127^hi and CD127^lo MP cells in a dynamic steady state.

In addition to TCR engagement, CD28 ligation plays an essential role in MP cell maintenance (4). To assess the function of the same signaling pathway in the four MP subpopulations, we treated mice with CTLA4-Ig for 10 days. In a similar manner to that induced by TCR signal blockade, Ki67 expression was reduced in the CTLA4-Ig-treated group, and CD127^hi cells increased while their CD127^lo counterparts decreased in the same animals ( Figures 4G, H ). Thus, optimal proliferation and dynamic equilibrium of MP cell subpopulations require tonic CD28 engagement in addition to TCR signaling.

The above observation that CD127, Sca1, and Bcl2 are differently expressed in MP and foreign Ag-driven memory CD4⁺ T cells revealed an unexpected phenotypic heterogeneity within the former lymphocyte population ( Figures 1 – 4 ). This finding prompted us to ask whether these four MP cell subpopulations possess different functions. The experiments performed above defined MP subpopulations based on markers (CD127 and Sca1) that are well known to characterize foreign Ag-driven memory cells, but left open the possibility that other molecules might better characterize functional MP subsets. To address this question, we analyzed the MP cell population in the scRNAseq dataset generated in Figure 1C by unsupervised clustering, to define its components based on the full transcriptome rather than on predefined markers. This analysis divided the MP cell population into four clusters (I–IV) ( Figures 5A, B ). Comparing the expression of Il7r and Ly6a among these four clusters showed a clear demarcation into Il7r ^hi Ly6a ^lo, Il7r ^hi Ly6a ^hi, Il7r ^lo Ly6a ^hi, and Il7r ^lo Ly6a ^lo gene expression patterns for clusters (I), (II), (III), and (IV), respectively; importantly, these correspond respectively to the CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo MP subpopulations defined in Figure 1G . The results of this analysis thus provide independent support to the conclusion that CD127 and Sca1 expression define four transcriptomically distinct MP subpopulations. Further clustering and scoring analyses revealed that the cluster (II), that is Il7r ^hi Ly6a ^hi and thus equivalent to the CD127^hi Sca1^hi MP subpopulation, has the highest Th1-associated gene signature (27) ( Figures 5B, C ), suggesting that this subset displays Th1-related function.

We have previously shown that MP CD4⁺ T lymphocytes can respond to IL-12 and produce IFN-γ in vivo (4). Consistent with this finding, when whole splenocytes were cultured in the presence of IL-12, IL-18, IL-2, or a combination of these cytokines in vitro, CD4⁺ MP cells produced IFN-γ in response to IL-12 ( Figure S4A ). The latter response increased with the addition of IL-18 and IL-2 and peaked at 24–48 h after cytokine stimulation ( Figures S4A, B ). With this in vitro system, we asked which of the MP subpopulation(s) can generate the above-described Th1-like response. To do so, we first checked the expression of T-bet, which we previously reported to play an essential role in the determination of Th1-like MP cell activity (4, 5). As shown in Figure 5D , CD127^hi Sca1^hi cells expressed the highest levels of T-bet, confirming the result of clustering analysis in Figure 5B . Consistent with this observation, when each MP subpopulation as well as naïve cells was sort-purified and cultured individually in the presence of IL-12, IL-18, and IL-2, CD127^hi Sca1^hi MP cells produced high amounts of IFN-γ ( Figure 5E ). Furthermore, the same MP subpopulation expressed high levels of Il12rb2, Il18r1, and Il18rap in the steady state ( Figure 5F ). In terms of IL-2 receptor expression, CD122 and CD132 were expressed on all four MP subsets ( Figure 5G ). However, CD25 (IL-2 receptor α chain) was only induced when CD127^hi Sca1^hi MP cells were stimulated by Th1 cytokines ( Figure 5H ), consistent with previous reports showing that IL-12 upregulates CD25 on Th1 cells (37, 38). Together, these results identify CD127^hi Sca1^hi MP T lymphocytes as a resting population with selective Th1 effector potential.

Finally, we sought to validate the above functional observations in vivo. Naïve CD4⁺ T lymphocytes are known to induce severe colitis when transferred to Rag2 knockout (KO) mice (39, 40). Using this approach, we tested the capacity of MP cell subpopulations to induce colitis by individually transferring CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo Foxp3⁻ MP CD4⁺ T cell subsets from Foxp3-reporter mice to Rag2 KO recipients. CD127^hi Sca1^hi cells were found to induce mild body weight loss and clinical as well as histological colitis ( Figures 6A–C ). Consistent with this observation, the total number of recovered donor cells was the highest in the CD127^hi Sca1^hi recipient group ( Figure 6D ). Thus, among the four MP subpopulations, the CD127^hi Sca1^hi subset induces the most severe colitis when transferred to Rag2 KO mice. We further asked whether colitis driven by the CD127^hi Sca1^hi MP subpopulation is dependent on IL-12 and/or IL-23. To do so, we blocked IL-12 p40 using mAb specific for the cytokine after donor cell transfer. As shown in Figures 6E–G , body weight loss, clinical symptoms, and histological colitis were almost completely abrogated as a result of mAb treatment. Consistent with this, the total number of donor cells and their IFN-γ⁺ fraction were dramatically reduced in the same mAb-treated animals ( Figures 6H, I ). These data suggest that in Rag2 KO mice, CD127^hi Sca1^hi MP cells can generate inflammatory responses, thus supporting their functional relevance.

C57BL/6 CD45.2⁺ WT mice were purchased from Taconic Biosciences (Rensselaer, NY, USA) or Japan SLC (Hamamatsu, Japan). Rag2 KO and CD45.1⁺ WT mice were obtained from the National Institute of Allergy and Infectious Diseases (NIAID) contract facility at Taconic Biosciences or breeding stock maintained at Tohoku University Graduate School of Medicine. Nur77-GFP and Foxp3-RFP reporter mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). CD4-CreERT2 TCRα^flox mice are previously described (5). All mice were maintained in SPF animal facilities in the NIAID, National Cancer Institute (NCI), National Institutes of Health (NIH), or Tohoku University Graduate School of Medicine except for GF and AF mice, which were bred and maintained in the animal facility of Pohang University of Science and Technology as previously described (33). All mice were used at the age of 8–16 weeks except in the case of Figures 3A, B , where mice of indicated ages were utilized. The care and handling of the animals used in our studies were in accordance with the animal study protocols approved by the NIAID or NCI Animal Care and Use Committee, by the Institutional Committee for the Use and Care of Laboratory Animals of Tohoku University, or by the Institutional Animal Care and Use Committees of the Pohang University of Science and Technology.

Thymectomy was performed on 8~12-week-old mice under aseptic conditions. Sham thymectomy was conducted with the same procedure except that thymic lobes were left intact. At the end of the experiments, the thymic deficiency was confirmed by careful macroscopic inspection.

LCMV Armstrong was propagated in baby hamster kidney-21 fibroblast cells [American Type Culture Collection (ATCC)]. Viral titers were determined by plaque assay using Vero African-green-monkey kidney cells (ATCC). Viral stocks were frozen at −80°C until use. For infection, mice were intraperitoneally injected with 2 × 10⁵ plaque-forming unit/mouse of the virus as previously described (22).

To activate CreERT2 recombinases, mice received an intraperitoneal injection of TMX (20 mg/mouse) dissolved in corn oil (both Sigma-Aldrich, St. Louis, MO, USA) as previously described (5). To block IL-12 p40 or CD80/86 signals, anti-IL-12B p40 (C17.8), CTLA4-Ig, or control IgG (300 μg/20 g body weight; all from Bio X Cell, West Lebanon, NH, USA) were administered every 3 days as reported (4, 5). To inhibit lymphocyte egress from the thymus and lymph nodes, FTY720 (20 μg/20 g body weight; Cayman Chemicals, Ann Arbor, MI, USA) dissolved in phosphate-buffered saline (PBS) or control PBS was given to mice every day (20).

Total CD4⁺ T lymphocytes were obtained from pooled splenocytes and lymph node cells of donor mice using a CD4⁺ T Cell Isolation Kit or CD4 Microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Naïve CD4⁺ T cells were then purified by sorting for CD4⁺ CD25^neg CD44^lo CD62L^hi cells using a fluorescence-activated cell sorting (FACS) Aria II (BD Biosciences, San Jose, CA, USA). To obtain MP cell subsets, CD127^hi Sca1^lo, CD127^hi Sca1^hi, CD127^lo Sca1^hi, and CD127^lo Sca1^lo subpopulations among CD4⁺ CD25^neg CD44^hi CD62L^lo cells from WT mice or CD4⁺ Foxp3^neg CD44^hi CD62L^lo cells from Foxp3-RFP reporter mice were sorted. Purity was >96%. For adoptive transfer experiments, depending on the experiments, 2 × 10⁵ to 1 × 10⁶ donor cells were intravenously injected into recipient animals.

After MP subpopulations were transferred to Rag2 KO mice, the animals were monitored for body weight. The disease activity index (DAI) score was assessed based on clinical symptoms as previously described (54). The histological score of the proximal portion of the colon was measured as previously reported (55).

Single-cell suspensions were prepared from spleens and red blood cells lysed in ACK buffer. In some experiments, splenic cells were further enriched for CD4⁺ T lymphocytes using a CD4⁺ T Cell Isolation Kit or CD4 Microbeads (Miltenyi Biotec). To obtain colonic cells, lamina propria mononuclear cells were isolated as previously described (54). Cells were suspended in staining buffer (PBS supplemented with 2% fetal bovine serum (FBS)) and incubated with CD16/32 mAb (2.4G2; Harlan Bioproducts, Indianapolis, IN, USA) for 10 min on ice. Cells were then incubated with the following mAbs or their combination for 20 min on ice: CD4 (RM4-5), CD8 (53-6.7), CD44 (IM7), CD62L (MEL-14), anti-NK1.1 (PK136), anti-TCRβ (H57-597) (Thermo Fisher Scientific, Waltham, MA, USA), CD3 (17A2), CD5 (53-7.3), CD25 (PC61), CD45.1 (A20), CD45.2 (104), CD69 (H1.2F3), CD122 (TM-β1), CD127 (A7R34), CD132 (TUGm2) (BioLegend, San Diego, CA, USA), and Sca1 (D7) (BD Biosciences). Tetramer staining used CD1d (PBS-57), I-Aβ-GP66, and I-Aβ-NP309 tetramers, obtained from the NIH Tetramer Core Facility (Emory University, Atlanta, GA, USA). To detect intracellular products, cells were fixed and permeabilized using Foxp3/Transcription Factor Staining Buffer Set for 30 min on ice after surface staining and then stained with mAbs against Bcl2 (10C4), Foxp3 (FJK-16s), Ki67 (SolA15) (Thermo Fisher Scientific), and/or IFN-γ (XMG1.2) (BioLegend) for 20 min on ice. For T-bet detection, fixed cells were stained with anti-T-bet (O4-46; BD Biosciences) mAb for 2 h at room temperature. Flow cytometry was performed using either Canto II, LSR II, Fortessa, or Symphony cytometers, and the data were analyzed with FlowJo software (BD Biosciences). Gating strategies are previously described (4, 5) and briefly summarized in Figure S6 .

MP and foreign Ag-specific memory CD4⁺ T lymphocytes were obtained by sorting for TCRβ⁺ CD4⁺ CD25^neg CD44^hi CD62L^lo CD1d-tetramer^neg cells from uninfected mice and TCRβ⁺ CD4⁺ CD25^neg CD44^hi CD62L^lo GP66-tetramer⁺ cells from LCMV-infected mice, respectively. Sorted cells were then loaded onto the 10× Chromium platform using the Chromium Single Cell 3′ Library & Gel Bead Kit V2 according to the manufacturer’s instructions (56). Libraries were sequenced on the Illumina NextSeq using paired-end 26 × 98 bp, and sequencing files were processed to extract count matrices using the Cell Ranger Single Cell Software Suite (v2.2.0). Further analyses were performed in R using the Seurat package (3.0) (57). Scoring analyses for memory and Th1 signatures were performed as previously described (27). For cell cycle and anti-apoptotic signatures, genes were determined based on previous studies (26, 31) ( Table S1 ). The obtained dataset is deposited on the Gene Expression Omnibus (GEO) (GEO number: GSE145999; token: kdspeciofrwhdod).

For detection of Il12rb1, Il12rb2, Il18r1, and Il18rap mRNA, total RNA was extracted from sorted cells using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and reverse-transcribed with SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). For detection of TRECs, total DNA was isolated from sorted cells using the DNeasy Blood & Tissue Kit (Qiagen). Real-time PCR was performed using the THUNDERBIRD SYBR qPCR Mix (Toyobo, Osaka, Japan). qPCR analysis was carried out using a 7500 real-time PCR system (Thermo Fisher Scientific). Relative gene expression was calculated by the ΔCt method and normalized to the amount of Gapdh. The following primer sets were used: Il12rb1, 5′-CCCCAGCGCTTTAGCTTT-3′ and 5′-GCCAATGTATCCGAGACTGC-3′ (58); Il12rb2, 5′-AATTCTTCTTCACTTCCGCATACG-3′ and 5′-GCTCCCAGAAGCATTTAGAAAGT-3′ (59); Il18r1, 5′-GCTCAGACCCTAATGTGCAAG-3′ and 5′-TGCAGTTTGCCTTCAGAAATC-3′; Il18rap, 5′-TGCAATGAAGCGGCATCTGT-3′ and 5′-CCGGTGATTCTGTTCAGGCT-3′ (60); TRECs, 5′-CATTGCCTTTGAACCAAGCTG-3′ and 5′-TTATGCACAGGGTGCAGGTG-3′ (61); and Gapdh, 5′-CCAGGTTGTCTCCTGCGACTT-3′ and 5′-CCTGTTGCTGTAGCCGTATTCA-3′ (4).

In Figures 5 , S4 , total splenocytes or FACS-sorted MP subsets were stimulated with IL-12p70 (10 ng/ml; PeproTech, Cranbury, NJ, USA), IL-18 (10 ng/ml; MBL, Nagoya, Japan), and/or IL-2 (10 ng/ml; Ajinomoto, Tokyo, Japan) in Roswell Park Memorial Institute (RPMI) complete media for the indicated period at 37°C. At the last 6 h of incubation, Brefeldin A (1 μg/ml; BioLegend) was added. In Figure 6I , total cell suspensions were incubated in RPMI complete media supplemented with phorbol myristate acetate (PMA) (20 ng/ml) and ionomycin (1 μg/ml; both from Sigma-Aldrich) for 5 h at 37°C in the presence of Brefeldin A. Cells were then harvested and subjected to intracellular staining as described above.

A Student’s t-test was employed to establish statistical significance. p-Values <0.05 were considered significant.