A Dapl1⁺ subpopulation of naïve CD8 T cells is enriched for memory-lineage precursors

Abstract

Memory CD8 T cells provide long-lasting immunity, but their developmental origins remain incompletely defined. Growing evidence suggests that functional heterogeneity exists within the naïve T cell pool, shaping lineage potential before antigen stimulation. Here, we identify a subpopulation of naïve CD8 T cells expressing death-associated protein-like 1 (Dapl1) that contains preprogrammed precursors biased toward memory differentiation. The differentiation of these precursors is independent of Dapl1 but relies on the transcription factor B-cell lymphoma/leukaemia 11b (Bcl11b), resulting in the generation of Dapl1⁺ central memory–like CD8 T cells after infection and stem-like memory cells in cancer. Dapl1⁺ naïve T cells originate among mature thymocytes and gradually appear in the periphery postnatally. Peripheral Dapl1⁺ and Dapl1⁻ populations show limited plasticity, supporting a thymic-imprinting model. These findings reveal a developmentally imprinted subset of naïve CD8 T cells committed to memory fate, uncovering an alternative pathway for memory T cell generation offering new avenues for therapeutic application.

Thymic-derived naïve Dapl1⁺ CD8 T cells contain a subset preprogrammed to form memory-like cells upon activation.

INTRODUCTION

CD8 T lymphocytes are indispensable for immune defense, targeting and eliminating host cells infected with intracellular pathogens or transformed by cancer. After antigen clearance, most effector cytotoxic T lymphocytes (CTLs) undergo apoptosis, with only a small fraction persisting as memory CD8 T cells. These memory cells play critical roles in long-term immune protection, rapidly responding to antigen reexposure, and are central to tumor immunotherapies, including adoptive T cell transfer, chimeric antigen receptor T cell (CAR-T) therapy, and checkpoint blockade (1–16). Furthermore, memory CD8 T cells are essential for the efficacy of vaccines, making their study important for advancing immunotherapies and vaccine development.

Although much progress has been made, the molecular mechanisms underlying memory CD8 T cell differentiation remain incompletely understood. Two competing models have sought to explain memory T cell ontogeny. The “linear” model suggests that memory cells arise from effector cells, while the “bifurcative” model suggests that asymmetric division of activated CD8 T cells generates distinct effector and memory lineages (17–27). Despite their differences, both models share the assumption that memory fate is determined during antigen-driven differentiation in the immune response. Recent findings, however, challenge this paradigm, suggesting that heterogeneity within the naïve T cell population may predetermine memory or effector potential before antigen stimulation (28–39). Progress in this area remains hindered by the absence of reliable markers to identify this heterogeneity, highlighting the need for innovative approaches to unraveling the origins of memory CD8 T cells.

Here, we report a previously unidentified marker death-associated protein-like 1 (Dapl1), the expression of which reveals the heterogeneity of naïve T cells. The Dapl1 gene encodes a 108–amino acid intracellular protein with predicted localization in the cytoplasm. Although very little is known about Dapl1 protein function, it has been reported that Dapl1 may interact with many major pathways including cascades of MYC, mitogen-activated protein kinase, nuclear factor κB, and nuclear factor of activated T cells (40–45). Furthermore, in other models, Dapl1 is portrayed as a potential “gate keeper” of stem-like phenotypes, suppressing generation of reactive oxygen species and protein translation (46, 47). It has been demonstrated that Dapl1 is highly expressed in the skin, retina, and testis. Furthermore, the Dapl1 gene appeared among the top differentially expressed genes (DEGs) in stem-like memory CD4 and CD8 T cells in many recent studies (15, 48–63). The generation of anti-Dapl1 monoclonal antibodies and a Dapl1-reporter mouse model enabled us to detect and study Dapl1-expressing T cells. With these tools, we identified a distinct subpopulation of Dapl1⁺ cells within naïve T lymphocytes, revealing an additional layer of heterogeneity. Furthermore, we found that within Dapl1⁺ CD8 T cells, a subset of preprogrammed lymphocytes preferentially differentiates into memory-like cells during immune responses to pathogens and cancer. Thus, our findings uncover a complementary mechanism contributing to memory formation in CD8 T cells.

RESULTS

Dapl1 expression reveals heterogeneity of naïve T cells

Previously, we demonstrated that the expression of lethal-7 (let-7) microRNAs (miRNAs) markedly promotes generation of memory CD8 T cells (15). We found that the evolutionarily conserved Dapl1 gene is among the top DEGs between let-7–transgenic (Let-7Tg) and let-7–deficient (Let-7KD) CTLs. Let-7Tg CTLs with memory potential expressed very high levels of Dapl1 mRNA, while terminally differentiated Let-7KD CTLs had almost none (Fig. 1A).

Fig. 1. Dapl1 expression is restricted to specific T cell populations.

(A) Normalized Dapl1 reads [fragments per kilobase of transcript per million mapped reads (FPKM)] from RNA-seq of WT (gray), Let-7KD (red), and Let-7Tg (blue) CTLs (GSE232541). (B) Normalized Dapl1 expression (DESeq2) in αβ T cell receptor (TCRαβ) T cell subsets, mined from the ImmGen genome browser. DP, double positive; CM, central memory.(C) Dapl1 protein expression [Western blot (WB)] in total thymocytes or CD4 and CD8 T cells from peripheral lymph nodes. (D) Confocal microscopy of CD8 T cells stained for Dapl1 (red), CD8α (green), and 4′,6-diamidino-2-phenylindole (DAPI) (blue). (E) Representative flow cytometry of lymphocytes taken from peripheral lymph nodes of B6 mice and intracellularly stained for Dapl1. PB, Pacific Blue. (F) Dapl1 expression by proportion (top) or mean fluorescence intensity (MFI) (bottom) in CD4 T cells (light blue), CD8 T cells (blue), or CD4⁻CD8⁻ [double negative (DN); gray] cells from the blood (n = 6), spleen (n = 3), or lymph nodes (n = 3) of B6 mice intracellularly stained for Dapl1. (G) UMAP of multicolor flow cytometry data from splenic T cells of B6 mice (n = 3 per group). (H) Flow cytometry analysis of CD122 and CD44 expression on Dapl1⁺ or Dapl1⁻ splenocytes (left), with summary statistics (right). (I) Surface protein expression on naïve (CD44⁻CD62L⁺) T cells calculated either as the proportion of Ly6C⁺ cells or MFI relative to the Dapl1⁻ population (CD5, CD44, CD62L, CD127, and CD122). Statistical analyses were calculated using either an ordinary one-way analysis of variance (ANOVA) using Tukey’s correction [(A) and (F), top], two-tailed unpaired t tests [(F), bottom], or two-tailed paired t tests [(G) and (I)]. All data are representative of three biological replicates from one RNA-seq analysis (A) and two or more independent experiments [(B) to (G)] or are pooled from three biological replicates from two independent experiments [(E) to (I)].

On the basis of Immunological Genome Project (ImmGen) RNA sequencing (RNA-seq) datasets (64), we found that Dapl1 mRNA expression is restricted to peripheral naïve T cells and also persists at lower levels in memory precursor effector cells (MPECs) and memory CD8 T cells but is not present in short-lived effector cells (SLECs) (Fig. 1B). Furthermore, recent publications have shown the association of Dapl1 expression with stem-like memory phenotypes for both CD4 and CD8 T cell lineages (15, 48–63). We hypothesized that the Dapl1 expression pattern may reflect the gradual loss of multipotent traits of naïve T cells during differentiation into effector cells. Therefore, Dapl1 is a candidate marker for tracing memory lineage cells, where these traits are still preserved.

To detect Dapl1-expressing cells, we generated our own monoclonal antibody (mAb) against the mouse Dapl1 protein (clone 6H9). The specificity of this anti-Dapl1 mAb was validated by enzyme-linked immunosorbent assay (ELISA) using recombinant protein and by Western blot (WB) with Dapl1-transduced NIH-3T3 cells (fig. S1, A and B). The 6H9 mAb identified a 12-kDa band corresponding to Dapl1 protein in both CD4 and CD8 T cells from wild-type (WT) mice by WB, but no Dapl1 expression was detected in thymocytes (Fig. 1C), recapitulating the mRNA expression patterns (Fig. 1B). Using the 6H9 mAb for confocal microscopy, we confirmed the predicted cytoplasmic localization of Dapl1 within the cells (Fig. 1D and fig. S1C). Unexpectedly, we observed that not all CD8 T cells were Dapl1⁺. Intracellular staining for flow cytometry revealed that only 30 to 50% of naïve peripheral T cells expressed Dapl1 (Fig. 1, E and F). Notably, the intensity of Dapl1 expression and the frequency of Dapl1⁺ cells were higher in CD8 T cells than in CD4 T cells (Fig. 1F). We also observed that the proportion of Dapl1⁺ lymphocytes and their expression levels of Dapl1 were slightly elevated in blood compared to the spleen and lymph nodes. We next compared the phenotype of Dapl1⁺ and Dapl1⁻ T cells based on the expression of selected surface markers for naïve and activated T cells (fig. S1D). Uniform manifold approximation and projection (UMAP) analysis of multicolor flow cytometry data revealed that Dapl1⁺ T cells clustered separately from Dapl1⁻ T lymphocytes for both CD4 and CD8 populations (Fig. 1G). Furthermore, most of the Dapl1⁺ CD4 and CD8 T cells exhibited quiescent CD44^lowCD122^low phenotypes and had fewer activated CXCR3⁺CD103⁻ cells compared to their Dapl1⁻ counterparts (Fig. 1H). These differences were more pronounced with increasing Dapl1 protein expression (fig. S1E). Notably, among the few Dapl1⁺ CD44^hiCD122^hi CD8 T cells, most displayed a virtual memory phenotype (CD49d⁻), whereas antigen-experienced CD49d ⁺ cells were relatively rare (fig. S1F). In addition, gated naïve (CD44^lowCD62L^hi) Dapl1⁺ CD8 T cells contained a high frequency of Ly6C⁺ cells and expressed elevated levels of CD5, CD44, CD62L, CD127, and CD122 compared to naïve Dapl1⁻ cells (Fig. 1I and fig. S1, G and H). In contrast, naïve Dapl1⁺ CD4 T cells exhibited a milder phenotype, characterized by increased expression of CD5 and CD127. Thus, Dapl1 expression uncovers previously unrecognized heterogeneity within the naïve CD4 and CD8 T lymphocyte populations.

Dapl1-reporter model uncovers heterogeneity among many T cell subsets

To investigate the properties of Dapl1⁺ T cells, we generated a Dapl1-reporter mouse model, Dapl1^ZsG, in which the fusion construct of Dapl1, self-cleaving 2A peptide (P2A), and fluorescent protein ZsGreen (ZsG) is expressed under the control of the Dapl1 gene’s promoter and all its regulatory elements (Fig. 2A). Overall, the mice appeared phenotypically identical to littermate controls. At steady state, 20 to 30% of CD4 and 40 to 50% of CD8 peripheral T cells in Dapl1^ZsG/WT mice were ZsG⁺ (Fig. 2B). Moreover, the intensity of ZsG expression and the frequency of ZsG⁺ lymphocytes were higher in CD8 T cells than in CD4 T cells. This expression pattern of ZsG correlates well with Dapl1 mRNA (Fig. 1B) and protein (Fig. 1E) expression in WT T lymphocytes. Unexpectedly, we found that the Dapl1^ZsG allele does not result in production of Dapl1 protein, despite mRNA expression (Fig. 2C and fig. S2A). It is possible that extra amino acids from the P2A peptide modify the C terminus of Dapl1 protein leading to its instability. We confirmed this by overexpressing N- and C-terminal hemagglutinin (HA)–tagged Dapl1 constructs (fig. S2B), where only the N-terminal HA tag was well tolerated. Thus, serendipitously, we created a Dapl1^ZsG knockout–reporter allele. By leveraging this model, we used heterozygous Dapl1^ZsG/WT mice as reporters, with ZsG marking Dapl1-expressing cells. In contrast, homozygous Dapl1^ZsG/ZsG mice, which lack Dapl1 expression, allowed us to track “Dapl1-wannabe” cells through ZsG expression.

Fig. 2. Characterization of Dapl1 reporter.

(A) CRISPR-Cas9–mediated Dapl1^ZsG reporter gene construct strategy. 5′UTR, 5′ untranslated region. (B) Dapl1 expression in peripheral lymph nodes of Dapl1^ZsG/WT mice (n = 3; left), with proportions of Dapl1⁺ cells and Dapl1⁺ MFIs of CD4 T cells (light green), CD8 T cells (green), or CD4⁻CD8⁻ cells (DN, gray) (right). (C) WB of Dapl1 and Zap70 expression in cell lysates from CD8 T cells from lymph nodes of WT, Dapl1^ZsG/WT, or Dapl1^ZsG/ZsG mice. (D) Representative UMAP projection (left) of multicolor flow cytometry data from spleens of Dapl1^ZsG/WT mice with ZsG fluorescence. DC, dendritic cell; pDCs, plasmacytoid dendritic cells. (E) ZsG and RFP expression on splenocytes from a Dapl1^ZsG/WTFoxp3^RFP/RFP mouse. (F to H) Proportion of ZsG expressing cells within (F) spleen, (G) liver, and (H) gut subsets: small intestine intraepithelial lymphocyte (IEL), small intestine lamina propria lymphocyte (LPL), colon IEL, or colon LPL. Statistical analyses were calculated using ordinary one-way ANOVAs with Tukey’s correction for multiple comparisons. Data are representative of more than three independent experiments [(B), (E), and (F)], one experiment [(C) and (D)], three biological replicates from one experiment (G), or at least two biological replicates pooled from three independent experiments (H).

In Dapl1^ZsG/WT mice, ZsG expression was restricted to T cells (Fig. 2D and fig. S2, C to E). While αβ T cell receptor (TCRαβ) T cells in secondary lymphoid organs contained a large proportion of ZsG⁺ cells, few TCRγδ T cells and virtually no natural killer T (NKT) lymphocytes expressed ZsG. Using double reporter Dapl1^ZsG/WTFoxp3^RFP/RFP mice, we found that CD4 Foxp3⁺ T regulatory cells (T_reg cells) contained only 2 to 5% of ZsG⁺ cells, while a rare population of CD8 Foxp3⁺ T_reg cells had ~20% ZsG⁺ cells (Fig. 2, E and F). Notably, the presence of ZsG⁺ T cells was comparable between splenic and liver T cell subsets but sharply reduced in the small intestine for both innate-like and conventional populations (Fig. 2, G and H; and figs. S3, A and B, and S4, A to D). In the colon, innate-like T lymphocytes were predominantly ZsG⁻, whereas conventional T cells contained 20 to 30% of ZsG⁺ cells, highlighting the tissue-specific distribution of Dapl1-expressing cells. The notable difference in ZsG⁺ T cell frequencies between the small intestine and colon may also reflect differences in the composition of distinct T cell populations and warrants further investigation.

Next, we investigated the phenotype of conventional and regulatory ZsG⁺ T cells from Dapl1^ZsG/WTFoxp3^RFP/RFP mice. As expected, the conventional (Foxp3⁻) ZsG⁺ CD4 and ZsG⁺ CD8 T cells closely phenocopied WT Dapl1⁺ T cells stained with anti-Dapl1 mAb (Figs. 1F and 3, A to C, and fig. S5, A to D). Most of the ZsG⁺ T cells were quiescent and, when gated on the naïve (CD44^lowCD62L^hi) population, exhibited significantly elevated levels of CD5, CD44, CD62L, CD127, and CD122. Similar to Dapl1⁺ cells from WT mice, ZsG⁺ CD8 T cells were enriched with Ly6C⁺ cells. ZsG⁺Foxp3⁺ CD4 T_reg cells displayed reduced surface expression of CD5 and CD44, an increased proportion of CD62L⁺ and Ly6C⁺ cells, and a lower frequency of PD-1⁺ cells (fig. S5, E and F). This phenotype may indicate that Dapl1⁺ T_reg cells have a less activated state, which has been previously linked to reduced suppressive function (65–67). Together, our results confirmed that ZsG reliably reports Dapl1⁺ subpopulations of both CD4 and CD8 T cells in Dapl1^ZsG/WT mice.

Fig. 3. Dapl1 expression marks a distinct subpopulation of naïve CD8 T cells.

(A) Representative UMAP projection of multicolor flow cytometry data from splenic T cells of Dapl1^ZsG/WTFoxp3^RFP/RFP mice. (B) CD122 and CD44 expression on ZsG⁻ or ZsG⁺ splenocytes (left), with summary statistics (right). (C) Surface marker expression on naïve (CD44^lowCD62L^high) T cells, calculated either as the proportion of Ly6C⁺ cells or MFI relative to the ZsG⁻ population (CD5, CD44, CD62L, and CD127). (D) ZsG expression among circulating WT Dapl1^ZsG/WT, Dapl1^ZsG/WTP14⁺Rag2^−/−, or Dapl1^ZsG/WTOT-I⁺Rag2^−/− CD8 T cells. (E) ZsG expression in CD45.2⁺ CD8 T cells 30 days post–adoptive transfer of sorted ZsG⁻ or ZsG⁺ CD45.2⁺ CD8 T cells into CD45.1 hosts. (F) scRNA-seq analysis of naïve CD8 T cells from P14⁺Rag2^−/−, OT-I⁺Rag2^−/−, or WT datasets (GSE131847, GSE199563, GSE213470, GSE221969, GSE181784, and GSE186839), consisting of (top) violin plots representing Dapl1 heterogeneity, (middle) descriptions of the datasets, and (bottom) a heatmap of DEGs consistent among datasets. FC, fold change. (G) Pathway enrichment analysis denoting significantly different pathways between Dapl1⁺ and Dapl1⁻ naïve CD8 T cells. JAK, Janus kinase; STAT, signal transducers and activators of transcription. Statistical analysis for (B) and (C) was calculated via two-tailed paired t tests for difference [(B) and (C)] or two-way ANOVA (E). All data are pooled from three biological replicates from at least two independent experiments [(A) to (C)], or representative of at least three independent experiments [(D) and (G)].

To test whether the heterogeneity of Dapl1 expression is due to differences in TCR specificity, we generated two different TCR-transgenic mice (P14 and OT-I) on a Dapl1^ZsG/WTRag2^−/− background. On the basis of ZsG expression, we found that similar to polyclonal CD8 T cells, ~40% of both naïve P14 and OT-I CD8 T cells expressed Dapl1 (Fig. 3D), revealing intraclonal heterogeneity of naïve CD8 T cells. Next, we investigated the stability of Dapl1 expression in naïve CD8 T cells. To this end, we adoptively transferred sorted ZsG⁺ and ZsG⁻ P14 CD8 T cells from lymph nodes of P14 Dapl1^ZsG/WTRag2^−/− mice into congenically marked recipient mice (Fig. 3E and fig. S6). One month later, most donor cells retained their original phenotype, with only a small fraction of initially ZsG⁻ cells up-regulating ZsG expression overtime. These results indicate that Dapl1 expression is largely stable in the naïve state, with limited plasticity between ZsG⁻ and ZsG⁺ naïve T cell subsets. Furthermore, reanalysis of existing single-cell RNA-seq (scRNA-seq) datasets for naïve P14, OT-I, and polyclonal cells confirmed the presence of both Dapl1⁺ and Dapl1⁻ CD8 T cells at the mRNA level (Fig. 3F). These cells exhibited markedly different transcriptomes, indicating the presence of two distinct populations (Fig. 3F). Notably, naïve Dapl1⁺ CD8 T cells up-regulated a unique set of genes associated with cytokine signaling, interferon (IFN) signaling, and immunoregulation, while down-regulating genes involved in differentiation and apoptosis (Fig. 3G) (68–73). Together, these findings highlight heterogeneity of different T cell subsets even within monoclonal naïve populations.

Dapl1⁺ CD8 T cells have biased differentiation toward memory lineage

Naïve Dapl1⁺ and Dapl1⁻ CD8 T cells exhibited significant differences in their phenotype and transcriptional profiles. Therefore, we investigated whether Dapl1⁺ and Dapl1⁻ naïve CD8 T cells are two separate lineages with distinctive differentiation potentials. First, we tested the persistence of Dapl1 expression in CTLs generated from bulk naïve P14 CD8 T cells. On the basis of both endogenous expression of Dapl1 protein in WT mice and ZsG in our reporter, we observed that only a small fraction of differentiated cells were Dapl1⁺ (Fig. 4, A and B, and fig. S7, A and B). Next, we examined the phenotype of Dapl1⁺ and Dapl1⁻ CTLs, using the expression of CD62L as a proxy of memory lineage and Tim-3 for terminally differentiated effector cells (15). Unexpectedly, most Dapl1⁺ CTLs displayed a memory-like (CD62L⁺Tim-3⁻) phenotype and expressed low levels of effector markers such as CD44, CD122, CD160, PD-1, and 2B4. In contrast, Dapl1⁻ CTLs showed significantly higher expression of these molecules and also included a population of terminally differentiated (CD62L⁻Tim-3⁺) cells. Furthermore, we found that the frequency of ZsG⁺(Dapl1⁺) CTLs increased after treatment with rapamycin, a potent inhibitor of TORC1 known to promote the formation of memory cells (Fig. 4C) (74). These results indicate that Dapl1 expression in differentiating cells may be associated with memory formation.

Fig. 4. Biased differentiation of Dapl1⁺ CD8 T cells.

(A) Schematic of P14⁺Rag2^−/− CD8⁺ T cell differentiation into CTLs, with Dapl1 expression, CD62L⁺Tim-3⁻ and CD62L⁻Tim-3⁺ proportions, and relative surface marker expression (MFI fold change) in Dapl1⁺ versus Dapl1⁻ subsets (n = 3). Irr.Spln, irradiated splenocytes. (B) ZsG expression in CTLs from P14⁺Dapl1^ZsG/WTRag2^−/− mice, with CD62L⁺Tim-3⁻ and CD62L⁻Tim-3⁺ subset proportions and surface marker expression in ZsG⁻ versus ZsG⁺ cells (n = 3). (C) Proportions of ZsG⁺ CTLs after dimethyl sulfoxide or rapamycin supplementation of medium for 48 hours after activation (n = 3). (D) Schematic of LM^GP33 infection and adoptive transfer, with ZsG expression in CD8 T cells from the blood (days 7 and 14) or spleen (≥day 30) (n = 8). (E) KLRG1 versus CD127 and CD62L versus CD44 expression in CD8 T cells over time, with summary (n = 8). INT, intermediate. (F) Public scRNA-seq datasets from LCMV or influenza A virus (IAV)-induced memory CD8 T cells, with Dapl1 heterogeneity (violin plots), dataset details (table), and shared DEGs (heatmap). d, day. (G) Fold changes in Dapl1⁺ and subset frequencies after retroviral-induced transcription factor (TF) overexpression (n = 3). Statistical analysis was calculated using two-tailed paired t tests [(A) and (B)], a two-tailed unpaired t test (C), ordinary two-way ANOVA with false discovery rate (FDR) correction (D), or ordinary one-way ANOVAs using multiple comparison corrections [(E) and (G)]. Data are representative of two or more independent experiments [(A) to (D)] or are pooled from three independent experiments (G).

The small number of Dapl1⁺ CTLs may reflect either a proliferative or survival disadvantage of Dapl1⁺ cells or the down-regulation of Dapl1 expression during the course of differentiation. To investigate these possibilities, we first assessed the proliferation of naïve CD8 T cells from P14⁺Dapl1^ZsG/WTRag2^−/− mice at 24, 48, and 72 hours poststimulation in vitro. On the basis of CellTrace Violet dye dilution, both Dapl1⁺ and Dapl1⁻ lymphocytes underwent comparable rounds of cell division, although Dapl1⁺ cells showed a slight delay (fig. S7C). Activation markers CD44 and CD69 were similarly expressed in both populations, indicating equivalent activation states (fig. S7D). Unexpectedly, proliferating Dapl1⁺ cells expressed low levels of surface CD25, with this difference becoming more pronounced over time. Notably, low CD25 expression during CD8 T cell differentiation has been associated with increased memory potential (75), suggesting a link between Dapl1⁺ cells and memory fate commitment. To test whether some Dapl1⁺ cells down-regulate Dapl1 expression during differentiation, we generated CTLs in vitro from sorted ZsG⁺(Dapl1⁺) and ZsG⁻(Dapl1⁻) naïve CD8 T cells from P14⁺Dapl1^ZsG/WTRag2^−/− mice. Both populations expanded at similar rates, ruling out a survival defect (fig. S8A). However, a substantial proportion of initially Dapl1⁺ cells lost the expression of Dapl1 (Ex-Dapl1) during CTL differentiation and were phenotypically similar to CTLs derived from naïve ZsG⁻(Dapl1⁻) cells. In contrast, the small fraction of CTLs that retained ZsG expression had exclusively memory-like phenotypes (CD62L⁺Tim-3⁻) with reduced levels of effector markers (fig. S8B). Notably, the vast majority of effector descendants of naïve ZsG⁻(Dapl1⁻) T cells remained Dapl1⁻, suggesting limited plasticity from Dapl1⁻ to Dapl1⁺ subpopulations during differentiation.

To directly investigate the fate of differentiating Dapl1⁺ and Dapl1⁻ CD8 T cells, we used an in vivo infection mouse model, where we used Listeria monocytogenes (LM) that expressed the GP33 peptide (LM-GP33), a cognate antigen for P14 TCR-transgenic CD8 T cells. Sorted Dapl1⁺ and Dapl1⁻ naïve P14 T cells were separately injected into LM-infected host mice, and their phenotypes were assayed over time (Fig. 4, D and E). During the immune response, many initially Dapl1⁺ cells lost Dapl1 expression, becoming Ex-Dapl1 cells, which confirmed our in vitro observations. The frequency of Dapl1⁺ cells gradually recovered, with the majority of long-lived memory cells expressing Dapl1. On days 7 and 14 postinfection, CD8 T cells that retained Dapl1 expression (ZsG⁺) had acquired an MPEC phenotype (KLRG1⁻CD127⁺), whereas Ex-Dapl1 cells differentiated into both MPECs and SLECs (KLRG1⁺CD127⁻). Notably, Ex-Dapl1 cells were enriched with double-positive (KLRG1⁺CD127⁺) effectors, which have the potential to generate diverse memory subsets, including effector memory T cells (T_EM cells; CD44^hiCD62L⁻) (76). After 30 days postinfection, most of the Dapl1⁺ memory cells exhibited the phenotype of central memory CD8 T cells (T_CM cells; CD44^hiCD62L⁺). In contrast, Ex-Dapl1 memory cells, in addition to a T_CM population, contained a significantly higher proportion of T_EM cells. Overall, the number of generated memory cells was comparable between descendants of Dapl1⁺ and Dapl1⁻ naïve T cells (fig. S8C). Notably, the differentiated progeny of Dapl1⁻ naïve CD8 T cells remained preferentially ZsG⁻(Dapl1⁻) at all times and gave rise to both SLECs and MPECs as expected. In this case, the appearance of a very small proportion of Dapl1⁺ memory cells (2 to 4%) can be explained by initial sort impurity or a very limited generation of Dapl1⁺ lymphocytes de novo. We also noticed that in both in vitro and in vivo experiments, cells that retained expression of Dapl1 expressed higher surface levels of CD62L than their Dapl1⁻ counterparts (fig. S8, B and D), suggesting that Dapl1⁺ and Dapl1⁻ memory lineage cells may be different. To this end, we reanalyzed publicly available scRNA-seq datasets for P14 T_CM, T_EM, and resident memory T (T_RM) cells generated in response to acute lymphocytic choriomeningitis virus Armstrong (LCMV-Arm) infection (62, 68, 70, 71, 77, 78). We confirmed that a proportion of Dapl1⁺ cells is present in T_CM but not T_EM or T_RM populations after acute LCMV-Arm infection (68) (fig. S8E). Notably, the absence of Dapl1⁺ cells in the T_RM dataset from the small intestine is consistent with our earlier findings using the Dapl1-reporter model, which showed a marked reduction of Dapl1⁺ cells in this tissue at steady state (Fig. 2H). Moreover, DEGs up-regulated in Dapl1⁺ T_CM cells included genes associated with stem-like memory cell differentiation (Lef1, Id3, Sell, Ccr7, etc.) and those involved in the regulation of translation and survival (Fig. 4F and fig. S8F), while the genes linked to leukocyte activation were down-regulated. These data demonstrate that differentiated Dapl1⁺ T cells are transcriptionally different from Dapl1⁻ counterparts and express higher levels of key memory-related genes. Overall, our results suggest that the pool of Dapl1⁺ naïve CD8 T cells harbors a distinct subpopulation precommitted to memory lineage, with their progeny marked by persistent expression of Dapl1 and unique transcriptional profile. This finding indicates that, for a subset of naïve CD8 T cells, memory fate may be predetermined before antigen encounter (fig. S9).

To define the mechanisms regulating Dapl1⁺ CD8 T cell differentiation, we focused on the transcription factors (TFs) that were up-regulated in Let-7Tg CTLs where Dapl1 was also strongly expressed (15). Specifically, we tested the ability of selected TFs (Bcl11b, Klf2, Tcf1, Lef1, Sox4, Bach2, Foxo1, and Id3) from memory signature genes of Let-7Tg CTLs to support the differentiation of Dapl1⁺ CTLs in vitro. Compared to the empty–retrovirus (RV) control, WT CTLs transduced with RVs encoding Bcl11b, Klf2, and Tcf7 showed robust generation of Dapl1⁺ CTLs (Fig. 4G and fig. S10A). In addition to increased proportion of Dapl1⁺ cells, these transduced CTLs also exhibited higher frequency of CD62L⁺ cells and reduced frequency of Tim-3⁺ cells (Fig. 4G and fig. S10A). In contrast, CTLs transduced with Lef1, Sox4, Bach2, Foxo1, Id3, or Id2 (a TF involved in effector cell differentiation) did not significantly alter the generation of Dapl1⁺ cells. Given the fact that Bcl11b was the most potent inducer of Dapl1⁺ CTLs in vitro and has been previously implicated in the formation of MPECs and CD8 T memory cells (79, 80), we next assessed the expression of Dapl1 in WT and Bcl11b^−/− MPECs generated in response to LM–ovalbumin (OVA) infection using a publicly available RNA-seq dataset (80). On day 9 postinfection, in addition to compromised cell numbers (80), Bcl11b^−/− MPECs had significantly reduced expression of Dapl1 mRNA compared to WT cells (fig. S10B), while, as expected, all SLECs had low Dapl1 expression. Together, these results suggest that the TF Bcl11b promotes the differentiation of Dapl1⁺ memory lineage CD8 T cells.

During chronic infection or cancer, effector cells become exhausted T cells (T_EX cells), and the formation of memory cells is altered leading to the generation of a distinct memory lineage known as progenitors of exhausted T cells (T_PEX cells) with stem-like potential (3, 6–8, 16). These stem-like cells are the primary responders to various tumor treatments including immunotherapies such as T cell adoptive transfer, CAR-T, and checkpoint blockade. Phenotypically, T_PEX cells express CD62L, Ly108, and low levels of inhibitory receptors such as Tim-3 on their surface (3). We therefore investigated the presence and phenotype of differentiated Dapl1⁺ T cells within the tumor microenvironment using B16 melanoma mouse model. Approximately 10 to 20% of CD8 tumor-infiltrating lymphocytes (TILs) isolated from Dapl1^ZsG/WT or WT B16-bearing mice were Dapl1⁺. These cells were antigen experienced (CD44^hi) and displayed the T_PEX cell phenotype (CD62L⁺Tim-3⁻ and Ly108^hi) (fig. S11, A to C). The reanalysis of an available TIL scRNA-seq dataset further confirmed our findings that T_PEX cells but not terminally differentiated T_EX cells contain a Dapl1⁺ population (50) (fig. S11D). Furthermore, the up-regulated DEGs in Dapl1⁺ CD8 TILs included genes that are linked to a stem-like memory cell profile (Lef1, Klf2, Myb, Sell, Slamf6, etc.), cell differentiation, translation, and survival (6, 11, 50, 81–83) (fig. S11, E and F). These data suggest that during antitumor immune responses, a precommitted subpopulation of naïve Dapl1⁺ cells also exhibits differentiation bias and contributes to the formation of stem-like memory CD8 T cells in TILs.

Dapl1 expression is not required for the differentiation of Dapl1⁺ memory CD8 T cells but may contribute to their maintenance and cytokine production

Whether Dapl1 plays a role in the differentiation of memory CD8 T cells is unknown. To explore this question, we took advantage of our mouse model, where homozygosity of the Dapl1-reporter allele results in Dapl1 deficiency (Fig. 2C). In Dapl1^ZsG/ZsG mice, ZsG expression serves as a marker for Dapl1-deficient T cells, allowing us to track and characterize these “wannabe” cells. After adoptive transfer of naïve Dapl1^WT/WT or Dapl1^ZsG/ZsG P14 T cells followed by LM-GP33 infection, we consistently observed a mild but significant decrease in the proportion of Dapl1-wannabe cells at every stage of the immune response including formation of memory cells (Fig. 5A), while the phenotype of Dapl1-wannabes resembled the memory lineage phenotype of Dapl1⁺ cells (Fig. 5B and fig. S12A). We obtained similar results in differentiation experiments in vitro (fig. S12B). Notably, loss of Dapl1 expression had no impact on the generation of Dapl1-wannabe TILs or their stem-like memory phenotype in Dapl1^ZsG/ZsG mice (fig. S12C). Furthermore, in a separate in vivo experiment, the phenotype of differentiating CD8 T cells transduced with Dapl1-RV was indistinguishable from that of cells transduced with an empty-RV control, indicating that Dapl1 overexpression does not affect the lineage fate of CD8 T cells (fig. S12D). Thus, our results suggest that Dapl1 may be involved in maintenance of differentiating cells rather than the control of their fate.

Fig. 5. Role of Dapl1 expression in differentiation of CD8 T cells.

(A) Schematic of LM^GP33 infection and adoptive transfer, with ZsG or Dapl1 expression in CD8 T cells from blood (days 7 and 14) or spleen (≥day 30). (B) KLRG1 versus CD127 and CD62L versus CD44 expression in CD8 T cells over time, with summary. (C) TNFα and IFN-γ expression in day 30+ memory CD8 T cells 4 hours poststimulation with ionomycin and phorbol 12-myristate 13-acetate (PMA) (left), with quantification (right). Summary plots are pooled from 12 (WT) and 13 (Dapl1^ZsG/ZsG) biological replicates at days 7 and 14 and 8 biological replicates at day 30+ for both groups from three independent experiments [(A) and (B)] or 8 (WT) or 6 (Dapl1^ZsG/ZsG) biological replicates pooled from two independent experiments (C). Statistical analyses were conducted via ordinary two-way ANOVA with FDR correction (A), Brown-Forsythe and Welsh ANOVA tests (B), or a mixed-effects analysis with Šidák’s correction (C).

We next sought to test functional differences between Dapl1⁺ and Dapl1-wannabe memory CD8 T cells. Restimulation of memory cells collected after 30 days postinfection demonstrated that true Dapl1⁺ memory CD8 T cells contained a higher proportion of polyfunctional [Ifn-γ⁺ tumor necrosis factor–α (Tnfα)⁺] cells compared to Dapl1-wannabe or Dapl1⁻ memory T cells (Fig. 5C). However, adoptively transferred sorted P14 Dapl1^WT/WT and Dapl1^ZsG/ZsG memory T cells control B16-GP33 tumor growth with similar efficiency in tumor-bearing mice, suggesting that the absence of Dapl1 does not affect the overall performance of memory T cells (fig. S12E). In addition, although a previous report suggests that mice with a global CRISPR-mediated knockout of Dapl1 exhibit enhanced effector function of CD8 T cells in antitumor immune responses (45), we did not observe differences in MC38 tumor growth control between Dapl1^ZsG/ZsG and Dapl1^WT/WT mice, pointing to potential discrepancies between these two models of Dapl1 deficiency (fig. S12F). Collectively, our findings suggest that while Dapl1 is dispensable for the generation and function of memory CD8 T cells, it contributes to the maintenance of memory lineage cells and production of effector cytokines.

The development of naïve Dapl1⁺ T cells

Previous studies have shown that neonatal CD8 T cells exhibit reduced memory potential and tend to differentiate into SLECs rather than MPECs or memory cells (34, 84, 85). On this basis, we hypothesized that the development of Dapl1⁺ T cells may be limited in neonatal mice. To test this, we assessed the frequency of naïve Dapl1⁺ T cells in the spleens of Dapl1^ZsG/WT mice at 1 day, 1 week, and 1 month after birth. In 1-day-old neonatal mice, no Dapl1⁺ T cells were detected in the periphery. However, these cells gradually appeared with age (Fig. 6A), suggesting a developmental delay in generating Dapl1⁺ T cells during the neonatal period.

Fig. 6. Dapl1⁺ naïve T cells originate from mature thymocytes.

(A) Foxp3 and ZsG expression on splenocytes from 1-day-old (n = 4), 1-week-old (n = 4), or 1-month-old (n = 3) Dapl1^ZsG/WTFoxp3^RFP/RFP mice (left), with quantification of ZsG expression among CD4 (light green), CD8 (dark green), and T_reg cell (red) subsets (right). (B) Expression of ZsG in Dapl1^ZsG/WT mice on mature thymic CD4 and CD8 populations, as marked by TCRβ (left), with summary (right). ISP, immature single positive. (C) Dapl1 expression by RNA-seq of thymic subsets sorted by maturity (top; GSE148973), with heatmaps of defining characteristics by subset (bottom). n.s., not significant. (D) UMAP plots regenerated from a single-cell multimodal thymocyte dataset (GSE186078) denoting pseudotime, Cd4, Cd8, and Dapl1 expression. (E) Surface CD24 expression in CD8 T cells from the thymus (Thy) or lymph nodes (LN) of Dapl1^ZsG/WT mice stratified by ZsG expression (left), with summary (right). (F) ZsG MFI in CD8 T cells from the thymus or lymph nodes of Dapl1^ZsG/WT mice (n = 3). (G) ZsG expression in thymocytes from Dapl1^ZsG/WT mice treated with daily intraperitoneal injections of either ethanol (n = 2) or FTY720 (n = 3), with summary of replicates (right). All data are representative of two to four biological replicates from one experiment [(A) and (G)] or are representative of at least two independent experiments [(B), (E), and (F)]. Statistical analysis was conducted using either an ordinary two-way ANOVAs using Tukey’s correction [(A) and (G)], matched one-way ANOVA with Geisser-Greenhouse correction (B), DESeq2 (C), one-way Welsh’s ANOVA (E), or two-tailed unpaired t test (F).

We then examined whether Dapl1⁺ T lymphocytes originate within the thymus before populating secondary lymphoid organs or whether they arise directly in peripheral tissues. In the adult thymus, we detected a small population of Dapl1⁺ cells only among mature CD4 and CD8 single-positive (SP) thymocytes (Fig. 6B and fig. S13A). To confirm these findings, we analyzed Dapl1 mRNA levels in a bulk RNA-seq dataset of sorted thymocyte populations at different developmental stages (86). The results corroborated that only postselected thymocytes expressed Dapl1 (Fig. 6C). In addition, projecting Dapl1 expression on pseudotime analysis of a thymocyte scRNA-seq dataset indicated that Dapl1 mRNA expression is indeed up-regulated in a subset of the most mature thymocytes (Fig. 6D) (87).

To investigate whether the mature Dapl1⁺ T cells found in the thymus could originate from recirculation from the periphery, we used CD24 as a marker of thymocyte maturation. The surface expression of CD24 is highest on immature thymocytes and declines with maturation, while peripheral T cells are generally CD24⁻. We observed that ZsG⁺(Dapl1⁺) TCR ⁺ CD8 thymocytes exhibited intermediate levels of CD24—lower than immature double-positive thymocytes but significantly higher than peripheral T cells from lymph nodes (Fig. 6E), suggesting their thymic origin. Moreover, the lower CD24 expression on ZsG⁺(Dapl1⁺) TCR ⁺ CD8 thymocytes compared to mature ZsG⁻(Dapl1⁻) TCR ⁺ CD8 thymocytes indicated that Dapl1⁺ cells represent a more advanced maturation stage. Last, ZsG intensity was lower in Dapl1⁺ thymocytes than peripheral T cells (Fig. 6F), together ruling out recirculation from the periphery as a likely source.

The presence of Dapl1⁺ T cells among the most mature thymocytes implies that SP thymocytes up-regulate Dapl1 immediately before exiting the thymus, potentially explaining the low frequency of Dapl1⁺ cells in the thymus relative to peripheral lymphoid organs. To test this, we blocked T cell egress from the thymus using fingolimod (FTY720), a potent agonist of S1PR1 receptor that is involved in T cell migration (88). After 1 week of FTY720 treatment, we observed a significant increase in the proportion of ZsG⁺(Dapl1⁺) T cells among mature SP4 and SP8 populations in Dapl1^ZsG/WT mice (Fig. 6G), while peripheral ZsG⁺(Dapl1⁺) T cell frequency remained unchanged (fig. S13B). Together, these findings demonstrate that Dapl1⁺ T cells first emerged among mature thymocytes shortly before their exit into peripheral lymphoid organs, highlighting the thymus’s potential role in preprogramming naïve T cell populations.

DISCUSSION

Understanding the origins of memory T cells has been a key focus in immunology. It is generally accepted that the steps leading to memory lineage commitment occur after antigen recognition. Consequently, naïve T cells have been regarded as precursors capable of differentiation into memory and effector cells during immune responses. This bipotential ability was elegantly demonstrated in adoptive transfer experiments of single naïve CD8 T cells (19, 21, 26). These and many other findings support the popular model of memory formation proposed by Reiner’s group, where naïve T cells undergo asymmetric division after antigen encounter, producing one daughter cell committed for the effector lineage and another for the memory lineage (20, 22, 27). Here, we report a discovery of a novel subpopulation of naïve CD8 T cells whose differentiation does not align with the conventional model of memory formation. These naïve cells are already biased to differentiate into memory-like lineage cells even before encountering an antigen. They can be identified by persistent expression of the Dapl1 gene during differentiation, both in vitro and in vivo.

Using in-house mAb (6H9) against the Dapl1 protein and a Dapl1-reporter mouse model, we uncovered substantial heterogeneity among naïve T cells based on Dapl1 expression. At steady state, most of Dapl1⁺ T cells had a true naïve phenotype and were largely excluded from activated or differentiated subsets, with this phenotype positively correlating with the levels of Dapl1 expression. However, naïve Dapl1⁺ cells also exhibited distinct features, including slightly elevated levels of CD5, CD44, CD62L, CD122, and CD127 (in CD8 T cells), and a higher frequency of Ly6C⁺ cells with up-regulated Ly6C expression. This heterogeneity was also observed in naïve CD8 T cells from major histocompatibility complex (MHC) class I–restricted TCR-transgenic mice, indicating previously unrecognized intraclonal diversity. Furthermore, scRNA-seq analysis of multiple datasets of naïve CD8 T cells revealed a unique transcriptional signature of Dapl1⁺ lymphocytes, further supporting their distinct identity within the naïve T cell pool.

During the immune response, a small subset of naïve Dapl1⁺ CD8 T cells retained Dapl1 expression and differentiated into T_CM cells. Notably, naïve Dapl1⁻ cells neither up-regulated Dapl1 nor exhibited any bias in their differentiation. These findings strongly suggest that within the Dapl1⁺ naïve CD8 T cell population, a small population (10 to 20%) is precommitted to the memory lineage even before antigen stimulation. This previously unrecognized pathway of memory formation adds more complexity to the current model of CD8 T cell differentiation, which is based on the fact that memory lineage commitment occurs only after antigen exposure (17). Although the preprogramming of naïve CD8 T cells into memory has not been previously described, the dominance of the bifurcative model has been challenged by a few other findings. For instance, nuclear envelope invaginations have been shown to distinguish naïve CD8 T cells predisposed to differentiate into effector-like cells (30). Similarly, type I IFN exposure in naïve CD4 T cells caused a bias toward central memory cell differentiation by influencing their transcriptional and functional programming (28). In addition, biased differentiation has been observed between naïve CD5^high and CD5^low CD4 T cells (33).

We found that Dapl1 expression in differentiated cells is associated with the CD8 T cell subsets having memory potential, including virtual memory cells, MPECs, central memory cells, and stem-like cells. This observation aligns with prior studies in which Dapl1 was identified among top DEGs in memory-like lymphocytes (15, 48–63). Furthermore, the transcriptional difference between Dapl1⁺ and Dapl1⁻ memory cells revealed a new layer of heterogeneity. This raises the possibility that differentiated Dapl1⁺ lymphocytes represent a previously unknown lineage of memory CD8 T cells. It is important to mention that even Ex-Dapl1 cells, which lost the expression of Dapl1 during differentiation, were enriched with KLRG1⁺CD127⁺ memory precursors. Further research is needed to determine whether these phenotypic differences translate into functional variations, to identify additional markers that can reliably pinpoint memory precursors within the Dapl1⁺ naïve T cell pool, and to assess whether Dapl1^hi and Dapl1^low naïve cells follow divergent fates during differentiation.

The general biological function of the Dapl1 protein remains largely unexplored. In lower vertebrates such as zebrafish and Xenopus, the homologous Dap1b protein functions as a translational inhibitor by directly interacting with ribosomes, rendering them dormant and contributing to ribosomal storage and translational repression (46). In mammals, Dapl1 has been primarily studied outside the immune system (40–44, 47), although one study reported its expression in both murine and human T cells and implicated it in the negative regulation of antitumor activity of CD8 T cells (45). Using our homozygous reporter mouse model, we investigated Dapl1-deficient lymphocytes but did not observe enhanced antitumor activity. Instead, we found that Dapl1 expression influences the frequency of Dapl1⁺ memory CD8 T cells and their production of effector cytokines such as IFN-γ and TNFα. Further research is required to elucidate the molecular mechanisms through which Dapl1 drives these observed phenotypes.

Among the TFs known to promote memory formation, Bcl11b emerged as the most potent in supporting the differentiation of Dapl1⁺ memory lineage cells in vitro and in vivo. This aligns with previous studies showing that, beyond its essential role in T cell lineage specification, Bcl11b is required for memory CD8 T cell differentiation and counteracts the innate-effector program (79, 80, 89). However, the precise role of Bcl11b in this context remains to be elucidated. It is currently unclear whether Bcl11b directly regulates Dapl1 expression or whether it indirectly promotes the generation of Dapl1⁺ memory-like T cells by influencing the stability of Dapl1⁺ cells or rescuing the conversion of naïve Dapl1⁺ cells to Ex-Dapl1 cells upon activation. As a future direction, it will be critical to define the transcriptional hierarchy and network dynamics that govern the balance between proeffector and promemory TFs in these cells during activation and differentiation. Given the critical role of Bcl11b in T cell development, future investigation should determine whether its expression is necessary for the formation of naïve Dapl1⁺ T cells and whether it contributes to preprogramming of memory precursors within this population.

The origin of Dapl1 cells represents a novel and important area of investigation. Although most Dapl1⁺ naïve T cells are located in the periphery, we detected the earliest Dapl1⁺ naïve T cells among the most mature thymocytes, which rapidly accumulated following blockade of thymic egress, supporting a thymic developmental origin. Moreover, peripheral Dapl1⁺ and Dapl1⁻ populations exhibit limited plasticity, making it unlikely that Dapl1⁺ cells arise de novo in the periphery. However, the signals that drive the development of Dapl1⁺ cells in the thymus remain unidentified. It is also unclear whether the memory commitment observed in a subset of Dapl1⁺ naïve T cells occurs in the thymus or in the periphery. Addressing these questions is of great interest, as it could inform strategies to enhance vaccine efficiency and improve T cell–based therapies.

Furthermore, we found that the development of Dapl1⁺ T cells is age dependent. Peripheral Dapl1⁺ T cells emerge gradually within the first week after birth, coinciding with the transition of the neonatal immune system to an adult-like state. This transition is thought to be governed by a postnatal switch in the source of hematopoietic stem cells (HSCs): from embryonic HSCs derived from the liver to adult HSCs originated in the bone marrow (90). Notably, neonatal CD8 T cells are reported to have diminished memory potential, favoring differentiation into SLECs (34, 84, 85). This process is largely mediated by the transient expression of Lin28b, an RNA binding protein that inhibits the biogenesis of evolutionary conserved let-7 miRNAs (91, 92). Although Lin28b-transgenic CTLs (Let-7KD) lack detectable Dapl1 mRNA expression (Fig. 1A), it remains to be determined whether Dapl1 expression is regulated by let-7.

In summary, our study predicts the existence of a novel subpopulation within a pool of Dapl1⁺ naïve CD8 T cells that appears biased to differentiate into memory-like cells even before antigen exposure. Moreover, we demonstrated that Dapl1 expression serves as a valuable marker for tracking the progeny of this population among long-lived cells generated during immune responses to cancer or infection. Future investigations are needed to elucidate the precise roles these cells play in immunity and to uncover the molecular mechanisms underpinning their commitment to memory lineage.

MATERIALS AND METHODS

Ethics statement

This study was performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH). All animals were handled according to approved Institutional Animal Care and Use Committee protocol #5781 of the University of Massachusetts, Amherst.

Animals

B6 (C57BL/6J, stock no. 000664), Rag2^−/− (B6.Cg-Rag2^tm1.1Cgn/J, stock no. 008449), OT-I [C57BL/6-Tg(TcraTcrb)1100Mjb/J, stock no. 003831], and Foxp3^RFP (C57BL/6-Foxp3^tm1Flv/J, stock no. 008374) mice were obtained from the Jackson Laboratory. P14 [B6;D2-Tg(TcrLCMV)327Sdz/JDvs/J] mice were a gift from A. Singer (National Cancer Institute, NIH). Dapl1^ZsG mice were generated in the University of Massachusetts Animal Models core facility (see below). P14⁺ Lin28-transgenic and P14⁺Let-7Tg mice on a Rag2^−/− background were generated as previously described (93). Animals were housed in a specific pathogen–free facility under a 12-hour dark/12-hour light cycle at 20° to 22°C and a humidity range of 30 to 70%. All breedings were maintained at the University of Massachusetts, Amherst.

Generation of Dapl1^ZsG mice

Dapl1^ZsG knock-in mice were generated in the University of Massachusetts Animal Models core facility using CRISPR-Cas9–mediated genome editing. The knock-in template construct was designed to replace the Stop codon of the Dapl1 gene and contained a GSG-P2A-ZsG fragment flanked by 120–base pair (bp) 5′ and 3′ homology arms. Two single guide RNAs (sgRNAs) were used to target protospacer adjacent motif (PAM) (AGG) 10 bp upstream (antisense sgRNA1, GTCCTGGTCTAACATTTTCG) and PAM (AGG) 3 bp downstream (sense sgRNA2, AACCTCGAAAATGTTAGACC) of Stop codon of the Dapl1 gene, such that the original target sequence is disrupted after insertion of the construct and therefore is no longer subject to Cas9 recutting. Single-strand DNA template (GenScript Inc.), Cas9 protein, and sgRNAs (Integrated DNA Technologies) were microinjected into zygotes isolated from superovulated B6D2F1 (C57BL/6J × DBA2J) female mice (8 to 10 weeks old) mated with B6D2F1 males, followed by euthanasia 20 hours post–human chorionic gonadotropin injection for zygote collection from the oviducts. Microinjected zygotes were cultured in potassium simplex optimized medium (EMD Millipore) at 37°C in a humidified atmosphere of 5% CO₂/5% O₂ balanced with N₂. After 3 days of in vitro culture, early blastocysts [embryonic day 3.5 (E3.5)] were transferred into the recipients (E2.5) using nonsurgical embryo transfer (#60010, ParaTechs Corporation) to produce founder mice. Correctly targeted mice [identified by polymerase chain reaction (PCR) using the primers for 5′ end Dapl1 forward GTAGAGACTGCGACTTGACC/ZsG reverse CTTCATGGTCATCTCCTTGG and for 3′ end ZsG forward TCCTGCGAGAAGATCATCC/Dapl1 reverse GAAAAGGGAAGATACAAAAGTTGG and confirmed by sequencing) were backcrossed to C57BL/6J mice for 10 generations.

Generation of anti-Dapl1 antibodies

Anti-Dapl1 antibodies were generated as described previously (94). Briefly, recombinant Dapl1 protein was cloned and expressed using the Novagen pET Expression System (EMD Millipore) according to the manufacturer’s instructions. Eight-week-old BalbC/J mice were immunized intraperitoneally with 100 μg of the purified Dapl1 protein mixed in with Freund’s adjuvant. Spleens were harvested after two boosts, and splenocytes were fused with SP20 myeloma cells [American Type Culture Collection (ATCC)] using electrofusion (BTX, Harvard Apparatus). Hyrbidomas were cultured in RPMI 1640 supplemented with 20% fetal bovine serum (FBS), penicillin (100 U/ml), streptomycin (0.1 mg/ml), 2 mM l-glutamine, 1 mM sodium pyruvate, and HAT (0.1 mM sodium hypoxanthine, 0.4 μM aminopterin, and 16 μM thymidine). Antibody clones were tested by ELISA on Dapl1 recombinant protein and by immunofluorescence on NIH-3T3 cells transduced with a Dapl1 overexpression RV.

Flow cytometry analysis

Flow cytometry data were acquired on BD LSRFortessa or Cytek Aurora and analyzed in FlowJo v10.9.0 software. The following monoclonal antibodies from BioLegend were used: CD8α (53-6.7), CD8β (YTS156.7.7), CD4 (RM4-5), CD44 (IM7), CD62L (MEL-14), KLRG1 (2F1/KLRG1), CD127 (A7R34), PD-1 (29F.1A12), Tim-3 (RMT3-23), 2B4 (m2B4(B6)458.1), CD45.2 (104), CD122 (TM-β1), CD5 (53-7.3), TCRβ (H57-597), TCRγ/δ (GL3), CD11b (M1/70), Ly6C (HK1.4), Ly6G (1A8), CD3 (145-2C11), NK1.1 (PK136), B220 (RA3-6B2), F480 (BM8), SiglecH (551), CD11c (N418), CD24 (M1/69), CD172 (P84), CXCR3 (CXCR3-173), CD103 (2E7), Ly108 (330-AJ), TNFα (MP6-XT22), streptavidin-AF647, phycoerythrin (PE), and BV711. CD160 (CNX46-3), MHCII (M5/114.15.2), and IFN-γ (XMG 1.2) were from eBioscience. CD45 (30-F11) was from Invitrogen. PE-CF594 rat anti-mouse immunoglobulin G (IgG) was from BD Biosciences. PE-conjugated CD1d tetramers were obtained from the Tetramer Core Facility of the US NIH. Live cells were incubated with anti-CD16/32 Fc block (2.4G2, BD Pharmingen) before staining with antibodies against surface markers. Staining for surface proteins was performed at 4°C for 40 min, and fluorescence-activated cell sorting (FACS) buffer [phosphate-buffered saline (PBS) + 0.5% bovine serum albumin (BSA) + 0.01% sodium azide] was used for washes. Dead cells were stained with 4′,6-diamidino-2-phenylindole (DAPI).

For intracellular cytokine staining, cell suspensions were restimulated in vitro for a total of 4 hours with phorbol 12-myristate 13-acetate (PMA; 50 ng/ml; Sigma-Aldrich) and 1 μM ionomycin (Sigma-Aldrich), in the presence of 2 μM monensin (eBioscience). After surface antibody staining, cells were stained with the LIVE/DEAD fixable Blue Dead Cell Stain Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Cells were then fixed and permeabilized using the Cytofix/Cytoperm solution kit (BD Biosciences) according to the manufacturer’s instructions, followed by staining with antibodies against intracellular molecules, including Dapl1.

Data from Cytek Aurora were gated on live CD45⁺ TCRβ⁺ cells (fig. 1G, 3A) or total live CD45⁺ cells (fig. 2E) and exported to csv files as channel values with all compensated parameters. The exported data were analyzed by the Seurat package (95) using R software version 4.2.1. Data were loaded as count matrices and converted to a Seurat object using the CreateSeuratObject() function. Fluorescence intensity representing Dapl1 or ZsG expression was added to the metadata and included (Figs. 1G and 3A) or excluded (Fig. 2E) as a variable for clustering. Data were normalized and scaled, and all features were used for principal components analysis and UMAP analyses. Clusters were annotated on the basis of the fluorescence level of cell type markers. Cells representing CD4⁺ or CD8⁺ cells were subset and reclustered (Figs. 1G and 3A).

Cell isolation

Lymph nodes, spleens, and thymuses were gently tweezed to isolate lymphocytes. ACK lysing buffer (KD Medical) was used to remove red blood cells from spleen suspensions. For purification of CD8 or CD4 T cells, lymphocyte suspensions were incubated with anti-mouse CD4 (GK1.5) or anti-mouse CD8 (2.43) antibodies, respectively, followed by incubation with anti-rat IgG magnetic beads (BioMag, QIAGEN).

For the isolation of intraepithelial lymphocytes (IELs) and lamina propria lymphocytes (LPLs), the small intestine and colon were first removed from the rest of the gastrointestinal tract, onto collection medium (RPMI 1640 supplemented with 25 mM Hepes, 1% l-glutamine, 1% penicillin/streptomycin, 50 μM β-mercaptoethanol, and 3% FBS). Peyer’s patches lining the small intestine were removed. Tissues were cleaned by flushing out feces with collection medium and rinsing in PBS. Tissue fragments were agitated at 37°C for 20 min in collection medium containing 5 mM EDTA and 1 mM dithiothreitol (DTT) and then further shaken in serum-free collection medium containing 2 mM EDTA. The suspension was washed several times in collection medium and passed through a 70-μm filter. For LPLs, tissue fragments were minced and agitated at 37°C for 25 min in serum-free collection medium containing collagenase D (1 mg/ml) and deoxyribonuclease (0.5 mg/ml). The suspension was passed through a 70-μm filter, washed in collection medium, and then purified by a 30% Percoll gradient. Last, IELs and LPLs were resuspended in collection medium containing 10% FBS.

For TIL isolation, mice were injected subcutaneously in the right flank with 2.5 × 10⁵ tumor cells. On day 8, mice were intravenously injected with CD45.2-BV785 antibodies (3 mg per mouse). Tumors were collected 3 min postinjection. They were then minced and passed through a 40-μm filter, spun at 1250 rpm for 5 min, and resuspended in FACS buffer.

For isolation of liver lymphocytes, livers of mice were perfused with 10 ml of PBS, followed by collection and manual disassociation of tissues via filtration through a 40-μm filter. Cells were then spun at 1250 rpm for 5 min, followed by resuspension in 5 ml 37% Percoll at room temperature. Resuspended cells were separated by a 20-min room temperature spin at 800g. The uppermost layers of fluid were carefully aspirated to preserve the lymphocyte pellet, which was then stained for flow cytometry.

In vitro culture

All cultures were kept in 37°C and 7% CO₂ atmosphere in a humidified incubator. For CTL generation, CD8 T cells were stimulated with irradiated splenocytes and soluble anti-CD3 mAbs (1 μg/ml). Cells were cultured in RPMI 1640 supplemented with 10% FBS, 1% Hepes, 1% sodium pyruvate, 1% penicillin/streptomycin, 1% l-glutamine, 1% nonessential amino acids, and 0.3% β-mercaptoethanol. Forty-eight hours after activation, interleukin-2 (IL-2) was added to culture medium (100 U/ml), and cells were differentiated for an additional 3 days. For CTL differentiation with inhibitors, rapamycin (Sigma-Aldrich) was used at the final concentration of 0.5 μM for the first 48 hours of culture. For 24-hour activations, T cells were stimulated with plate-bound anti-CD3 mAbs (1 μg/ml) and anti-CD28 mAbs (5 μg/ml).

NIH-3T3, MC38 and Jurkat cells were obtained from ATCC. B16-GP33 (15), and NIH-3T3 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS. Jurkat and MC38 cells were cultured in RPMI 1640 supplemented with 10% FBS.

In vivo tumor experiments

After receiving a sublethal dose of irradiation (5 Gy), mice were injected subcutaneously in the right flank with 2.5 × 10⁵ MC38 cells. For studies involving adoptive T cell transfer, irradiated mice were injected intravenously with 5 × 10⁴ P14⁺ memory cells (96), followed by subcutaneous injection of 2.5 × 10⁵ B16-GP33 cells the next day. Tumors were measured every 2 to 3 days with a caliper, and tumor volume was determined using the following formula: ½ × D × d², where D is the length and d is the width. Mice were euthanized when tumors reached 2 cm³, when tumors became ulcerated or in instances when tumors interfered with normal behavior.

CellTrace Violet proliferation assay

Naïve CD8 T cells were labeled with CellTrace Violet (Invitrogen) for 15 min at 37°C. Cells were stimulated using plate-bound anti-CD3 mAbs (1 μg/ml) and anti-CD28 mAbs (5 μg/ml), followed by a 24- to 72-hour culture period and flow cytometry analysis.

In vivo plasticity of Dapl1 expression

ZsG⁺ and ZsG⁻ CD8 T cells were sorted from the lymph nodes of a Dapl1^ZsG/WT p14⁺Rag2^−/− mouse on a CD45.2 background, and 250,000 cells were adoptively transferred to CD45.1 hosts via intravenous injection. One month later, splenocytes were isolated, and ZsG expression in CD45.2⁺ CD8 T cells was assessed via flow cytometry.

Confocal microscopy

For microscopy, iBiTreat dishes (μ-slide 8 well, 80826-90) were coated with fibronectin (5 μg/ml) overnight and then washed before seeding 1 × 10⁴ NIH-3T3 cells per cell. NIH-3T3 cells were imaged 1 day after seeding the iBiTreat dishes. Cultures were rinsed with PBS and then fixed with 3.7% formaldehyde at room temperature for 25 min. After rinsing three times with PBS, cells were permeabilized with 0.5% Triton X-100 at room temperature for 5 min. Cells were blocked with PBS containing 1% BSA for 1 hour at room temperature. Cells were stained as described in the “Flow cytometry analysis” section, using PBS + 1% BSA for washes. For naïve P14 T cells, cells in suspension were stained, fixed, and permeabilized as described above. Stained cells were mixed with ProLong Gold (P36934, Thermo Fisher Scientific) and sealed onto a glass slide with a coverslip. Images were taken with a Nikon Eclipse Ti Series inverted microscope with a C2 confocal attachment, using the Immersion oil type F (Nikon) objective for the 100× lens. Analyses were performed using the Nikon Elements Analysis 3.1 software.

Experiments with LM

Recombinant LM expressing glycoprotein 33-41 epitope was a gift from J.T. Harty (University of Iowa). A frozen aliquot of LM (stored at −80°C) was thawed, cultured in Tryptic Soy Broth medium (KD Medical) with streptomycin (50 μg/ml), and optical density at 600-nm wavelength (OD₆₀₀) was measured hourly. OD₆₀₀ of 1 corresponds to 1 × 10⁹ colony-forming units (CFU) of LM. Once OD₆₀₀ reached 0.08 to 0.09, cells were spun down at 6000 rpm for 15 min at 4°C and resuspended in 0.9% NaCl at the concentration of 6 × 10⁷ CFU/ml. Each recipient mouse was infected intravenously with 6 × 10⁶ CFU of LM. Before infection but during the same day, mice were injected intravenously with 2 × 10⁴ naïve P14⁺ CD8 T cells.

WB analysis

Cells were collected and lysed in M2 lysis buffer [20 mM tris (pH 7.0), 0.5% NP-40, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 2 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 20 mM β-glycerol phosphate, 1 mM sodium vanadate, and leupeptin (1 μg/ml)] and then resolved by SDS–polyacrylamide gel electrophoresis. Blots were probed with anti-Dapl1, anti-Zap70 (BD Biosciences), anti-HA (Cell Signaling Technology), and anti–α-tubulin (Sigma-Aldrich) and visualized using enhanced chemiluminescence (Thermo Fisher Scientific) with horseradish peroxidase–conjugated anti-mouse IgG (Calbiochem) or anti-rabbit IgG (Jackson ImmunoResearch).

Cloning, retroviral vectors, and transduction

Open-reading frames (ORFs) of TFs (ID2, ID3, and Klf2) and Dapl1 were PCR amplified using Phusion High-Fidelity DNA Polymerase [New England Biolabs (NEB)] and ORF clones as templates (Horizon Discoveries). The rest of TFs (Bcl11b, Foxo1, Bach2, Tcf7, Lef1, and Klf2) and HA-tagged Dapl1 constructs were synthesized (GenScript). Inserts were then cloned into pMRX–internal ribosomal entry site (IRES)–green fluorescent protein (GFP) vector using NEBuilder HiFi DNA Assembly kit (NEB). RVs were produced from the transfection of Platinum-E cells with empty and gene of interest–expressing pMRX-IRES-GFP vectors using the Transporter 5 transfection reagent (Polysciences) according to the manufacturer’s instructions. Viral supernatants were collected 24 hours after transfection, concentrated with PEG-it (System Biosciences), and flash frozen in aliquots in liquid nitrogen. For retroviral transduction of T cells, naïve lymphocytes were stimulated with irradiated splenocytes in the presence of anti-CD3 mAbs (2 μg/ml), irradiated splenocytes for 14 hours, and then spinfected (2000 rpm for 90 min at 37°C) with virus and polybrene (4 μg/ml). Medium was changed 4 hours after transduction, after which IL-2 was added at 100 U/ml daily until day 5 postplating to generate CTLs. Jurkat cells were spinfected with virus and polybrene at 2000 rpm for 90 min at 37°C; NIH-3T3 cells were spinfected at 900 rpm for 90 min at 37°C.

Thymic egress inhibition experiments

Age- and sex-matched female Dapl1^ZsG/WT mice were injected with the thymic egress inhibitor FTY720 (1 mg/kg; MilliporeSigma) or with a 10% ethanol in PBS vehicle control intraperitoneally every 48 hours for 8 days. Mice were euthanized at day 9, and lymph nodes and spleens were collected.

Isolation of RNA and quantitative PCR

RNA was isolated using The Monarch Total RNA Miniprep Kit (NEB). cDNA was synthesized using the SensiFast cDNA Synthesis Kit (Thomas Scientific). SYBR Green quantitative PCR was performed using the SensiFAST SYBR Lo-Rox Kit (Thomas Scientific). SYBR Green primers (Integrated DNA Technologies) used are as follows: Dapl1, ATGGCAAACGAAGTACAAGTTCT (forward) and TCTTTCCAAAACGCCCATCTC (reverse); Rpl13A, CGAGGCATGCTGCCCCACAA (forward) and AGCAGGGACCACCATCCGCT (reverse).

Analysis of public RNA-seq datasets

For reanalysis of bulk RNA-seq data comparing Dapl1 expression in Bcl11b knockout MPECs, normalized counts from GSE186283 were compared via an ordinary one-way analysis of variance (ANOVA). Graphs were generated via GraphPad Prism 10. For reanalysis of sorted thymic precursor data (Fig. 6C), raw counts were obtained from GSE148973 (86). Using the DESeq2 (97) package in R, normalized counts and differential expression were calculated.

Analysis of public scRNA-seq datasets

scRNA-seq datasets were first screened for those that included clusters with a sufficient amount of cells expressing Dapl1 for calculating statistics. First, read count matrices were downloaded from National Center for Biotechnology Information Gene Expression Omnibus and then analyzed using the Seurat package in R with standard procedures or with version 5.0 integration methods when appropriate. Next, cell clusters were annotated on the basis of the expression level of marker genes suggested by the original publications. Last, within respective clusters, differential gene expression was calculated between cells expressing Dapl1 at a level greater than 0.1 and cells expressing Dapl1 at a level less than 0.1, using the FindMarkers() function and the test = “wilcox” option. For the meta-analysis of each cell type of interest (naïve cells, memory cells from acute infection, and stem-like cells from TILs), the list of DEGs were compared across different datasets and filtered for those that were commonly up-regulated or commonly down-regulated in at least two datasets. The intensity of differential expression of these genes were represented by plotting the average log₂ fold change values with the pheatmap package in R. Prediction of differentially enriched pathways based on these genes were performed with Metascape (98), and the results were plotted using the ggplot2 package in R.

Datasets used for naïve cells are GSE131847 (71), GSE199563 (69), GSE213470 (72), GSE221969 (73), GSE181784 (68), and GSE186839 (70). Datasets used for memory cells from acute infection models are GSE152841 (78), GSE131847 (71), GSE182275 (77), GSE181784 (68), GSE119940 (62), and GSE186839 (70). Datasets used for stem-like cells from tumors are E-MTAB-13073 (50), GSE180094 (6), GSE122712 (11), GSE182509 (81), GSE221118 (83), and GSE217038 (82).

For the multimodal thymic CITEseq dataset GSE186078 (87), the h5ad files were downloaded and converted to a Seurat object in R. The FeaturePlot() function was used to visualize the expression of select markers and differentiation pseudotime.

Statistical analysis

Data statistical analysis was performed with Prism 10 (GraphPad Software). P values for comparisons of only two groups were determined using paired or unpaired two-tailed Student’s t tests, with Mann-Whitney corrections if normal distributions could not be assumed. Ordinary Brown-Forsythe and Welsh one-way ANOVAs comparisons were used to compare multiple conditions, and two-way ANOVAs were used to compare tumor growth curves, cell population proportions over time, and thymic egress inhibition experiments. Mixed-effects analyses were used to compare multiple groups of independently normalized data. Survival comparisons were conducted using Kaplan-Meier tests. Sample size was not predetermined by statistical methods. Two mice were excluded from analysis of day 30+ data from Fig. 5B due to signs of secondary infection or malocclusion in adoptive transfer recipients; exclusion of these data points does not affect the interpretation of results.

Supplementary Materials

The PDF file includes:

Figs. S1 to S13

Legend for data S1

Other Supplementary Material for this manuscript includes the following:

Data S1