C57BL/6 (WT mice; CD45.1⁻CD45.2⁺CD90.1⁻CD90.2⁺), B6.SJL-PtprcaPepcb/BoyCrCrl (SJL mice; CD45.1⁺CD45.2⁻CD90.1⁻CD90.2⁺) mice, and B6.PL-Thy1a/CyJ (Thy1.1 mice; CD45.1⁻CD45.2⁺CD90.1⁺CD90.2⁻) were purchased from the Jackson laboratory or Charles River Laboratories. Generation of DGKζ KO mice were described previously (20, 21). Sevenmaker (ERK^SEM) (CD45.1⁻CD45.2⁺CD90.1⁻CD90.2⁺) mice were provided by L. Samuelson from the National Institutes of Health and were originally developed by S. Hedrick from the University of California, San Diego (22). Wild type P14 mice (CD45.1⁻ CD45.2⁺CD90.1⁻CD90.2⁺) were provided by E.J. Wherry (1). SJL mice were crossed to P14 mice to generate WT-P14 (CD45.1⁺CD45.2⁺CD90.1⁻CD90.2⁺ or CD45.1⁺ CD45.2⁻CD90.1⁻CD90.2⁺). DGKζ KO mice and ERK^SEM mice were crossed to P14 mice to generate DGKζ KO-P14 and ERK^SEM-P14 (CD45.1⁻CD45.2⁺CD90.1⁻CD90.2⁺). Unless otherwise specified, all mice were 7 to 12 weeks old at the time of use, were housed in pathogen-free conditions, and were treated in strict compliance with the Institutional Animal Care and Use Committee regulations at the University of Pennsylvania.

For flow cytometric analyses, cells were stained with antibodies against cell surface antigens at 4°C for 30 min in phosphate-buffered saline (PBS). LIVE/DEAD Fixable near-IR Dead Cell Stain Kit was used to exclude nonviable cells. Intracellular cytokine staining was performed with the BD Cytofix/Cytoperm Kit according to the manufacturer’s protocol. Flow cytometry was performed with an LSR II or LSR Fortessa or FACS Canto flow cytometer (BD Biosciences). Data were analyzed using FlowJo software (TreeStar). All fluorochrome-conjugated anti- bodies are listed in Table S1 . The gating strategy for tetramer⁺CD8⁺ T cells is shown in Supplementary Figures 1A, B .

For CD8⁺ T cell activation assays, freshly isolated splenocytes from LCMV CL13-infected mice were cultured together with anti-CD107a Ab and monensin for 5 h in tissue culture plates. During 5 h in culture, cells were restimulated with GP33 peptide (200 ng/ml; GP_33-41), peptide pool (mixed for 200 ug/ml each; GP_33-41, GPC_92-101, GPC_118-125, GPC_221-228, GPC_276-286, L_156-163, L3_38-346, L_349-357, L_455-463, L_775-782, L_1428-1435, L_2062-2069, NP_165-175, NP_205-212, NP_238-248, NP_396-404), or PMA (100 ng/ml; Sigma-Aldrich) and ionomycin (1 mg/ml; Sigma-Aldrich). Peptides were provided E.J. Wherry (23). After 5 h in culture, CD8⁺ T cells were analyzed for anti-CD107a Ab staining and intracellular IFNγ by flow cytometry.

TCR transgenic GP33-specific cells (P14) were isolated from the peripheral blood or splenocytes of donor mice using gradient centrifugation with Histopaque-1083 (Sigma-Aldrich) as previously described (1). For experiments using LCMV infection, CD45.1⁺ WT-P14 T cells were mixed 1:1 with CD45.1⁻ P14 T cells of the desired genotype (DGKζ KO -P14 and ERK^SEM-P14) and a total of 0.5-1.0 × 10³ T cells were adoptively transferred by intravenous injection into 7-10-week-old recipient (Thy1.1) mice 1 day before infection as previously described (1). This enabled us to distinguish both donor populations from the host T cells (CD90.2⁺ vs CD90.2⁻). In some experiments, Rapamycin (Sigma-Aldrich, R8781; pre-diluted in DMSO) was diluted further in PBS (75 μg/kg) and injected i.p. at a volume of 0.2 ml for 14 days. See Supplementary Figure 1C for gating strategy.

LCMV CL13 were propagated and titers were determined as previously described (5). One kidney per LCMV CL13 infected-mice was homogenized (Cole-Parmer) in 1 ml of MEM-alpha media (Gibco) supplemented with 10% FBS. The virus titer of homogenized kidney fluid or serum was determined by plaque assay (5) and represented as plaque-forming units per ml (PFU/ml). To test the anti-viral activity of ASP1570, the plaque assay was performed in the presence or absence of ASP1570 (1 μM) with a fixed concentration of virus (1.4 × 10⁷ PFU/ml) added to the cell cultures. Mice were infected intravenously with 4 × 10⁶ PFU of LCMV CL13.

For pharmacological inhibition of DGKζ activity in vivo, mice were administered p.o. once daily with 0.5% Methyl Cellulose (10 ml/kg; Wako Chemicals USA, Inc 13317815) or ASP1570 (3 mg/kg; Astellas Pharma, Inc; Patent: WO2021132422) for 3 Days before takedown. The drug was not given if the mouse was <80% of initial body weight on the day of administration.

For in vitro phosphorylation assays, splenocytes were pretreated for 30 min with fluorescently labeled anti-CD8 and anti-CD44 antibodies and LIVE/DEAD Fixable near-IR Dead Cell Stain in the presence or absence of ASP1570 (1 μM). Splenocytes were washed and stimulated with anti-CD3 antibody (BD Biosciences) for the indicated times in the presence or absence of ASP1570 (1 μM). After stimulation, splenocytes were fixed in 2% paraformaldehyde and Perm Buffer III (BD Biosciences) and stained for anti-pERK and anti-pS6 intracellularly. Plots were gated on CD44^loCD8⁺ T cells. All fluorochrome-conjugated antibodies are listed in Table S1 .

Statistical tests for flow-cytometry data were performed using GraphPad Prism software. A P value of <0.05 was considered significant in these analyses. A Student t-test (two-tailed) was used for comparisons between two independent conditions. A paired Student t-test was used when the samples being compared originated from the same mouse. Log-rank test was used to determine significance of P value of <0.05.

To test the role of DAG-mediated signaling in T cell activation during the acute phase of chronic viral infection, we infected WT and DGKζ KO mice with LCMV CL13 and examined the phenotype and function of LCMV-specific T cells against two immunodominant epitopes (GP33 and GP276 peptides) at Days 7 and 10 post infection. At Day 7 post infection, other than a slight increase in the fraction of GP33-specific CD8⁺ T cells (GP33-tetramer⁺) in the blood, the fraction of LCMV-specific T cells was unchanged in the spleen and blood of DGKζ KO compared to WT mice ( Figure 1A ). In contrast, at Day 10 post infection, the fraction of LCMV-specific CD8⁺ T cells was decreased in DGKζ KO compared to WT mice ( Figure 1A ). Moreover, owing to an overall decrease in splenocyte count in LMCV CL13-infected DGKζ KO mice ( Supplementary Figure 2 ), the absolute number of LCMV-specific CD8⁺ T cells was significantly decreased at both Days 7 and 10 post infection in DGKζ KO compared to WT mice ( Figure 1B ). Despite the overall decreased number, the fraction of LCMV-specific CD8⁺ T cells expressing the effector differentiation marker, KLRG1 was significantly increased at both Days 7 and 10 post infection in DGKζ KO compared to WT mice ( Figure 1C ). Functionally, an increased fraction of splenic CD8⁺ T cells from DGKζ KO compared to WT mice were capable of degranulating (CD107a⁺) and producing IFNγ upon GP33 peptide restimulation ex vivo ( Figure 1D ). However, there was no difference in the virus titer of DGKζ KO and WT mice at either Day 7 or 10 post infection ( Figure 1E , Supplementary Figure 3 ). These results suggested that DGKζ KO LCMV-specific CD8⁺ T cells showed increased effector differentiation and functional capacity but potentially impeded expansion and/or survival. Analysis of T cells beyond Day 10 was not performed due to significantly exacerbated weight loss and mortality of LCMV CL13-infected DGKζ KO mice ( Figures 1F, G ).

To test the impact of DGKζ deficiency on LCMV-specific CD8⁺ T cells beyond Day 10, we examined the activity of DGKζ KO T cells on a fixed LCMV GP33 peptide-specific TCR background (P14 TCR transgenic mice). This allowed us to directly compare WT and DGKζ KO T cells in the same environment and avoided confounding issues associated with altered TCR selection in the thymus of DGKζ KO mice. Naïve CD8⁺ T cells from congenically disparate (CD45.2⁺ vs CD45.1⁺) WT P14 and DGKζ KO P14 mice were mixed at a 1:1 ratio, adoptively transferred into WT mice (Thy1.1⁺), and subsequently infected with LCMV CL13 ( Figure 2A ). At Days 7 and 10 post infection, DGKζ KO P14 CD8⁺ T cells out-competed co-transferred WT P14 CD8⁺ T cells ( Figure 2B ), suggesting that DGKζ deficiency afforded an expansion advantage. Moreover, the fraction of inhibitory receptor-expressing cells and short-lived effector cells (SLEC; KLRG1⁺CD127⁻) were increased in DGKζ KO P14 CD8⁺ T cells compared to WT P14 CD8⁺ T cells ( Figure 2C , Supplementary Figures 4A, B ), suggesting increased effector cell differentiation. Almost all P14 CD8⁺ T cells of either WT or DGKζ KO origin degranulated and >70% produced IFNγ upon restimulation with GP33 peptide ( Supplementary Figure 5A ). A statistically significant increase in the fraction IFNγ-producing of DGKζ KO P14 CD8⁺ T cells compared to WT was seen on Days 7 and 10 post infection ( Supplementary Figure 5A ). In contrast to the increased SLEC phenotype, a smaller fraction of DGKζ KO P14 CD8⁺ T cells displayed a memory precursor effector cell (MPEC; KLRG1⁻CD127⁺) or central memory cell (T_CM; CD62L⁺CD127⁺) phenotype. Additionally, fewer DGKζ KO P14 CD8⁺ cells expressed the memory-associated transcription factor Eomesodermin⁺ (Eomes⁺) phenotype ( Figure 2D , Supplementary Figure 4B ). Interestingly, an abrupt disappearance of DGKζ KO P14 CD8⁺ T cells was seen on Day 14 ( Figure 2B , Supplementary Figure 3B ), suggesting that DGKζ KO P14 CD8⁺ T cells underwent cell death. Consistent with these data, we found that the sudden crash in DGKζ KO P14 CD8⁺ T cells was preceded by a larger fraction of cells expressing the pro-apoptotic molecule Bim and fewer expressing the anti-apoptotic molecule Bcl-2 at Day 10 post infection ( Figure 2E , Supplementary Figure 4B ). Together, these data suggest that enhanced DAG-mediated signaling in LCMV-specific CD8⁺ T cells leads to rapid early expansion and effector differentiation with decreased proportion of cells expressing memory markers, which subsequently leads to disappearance of these cells.

ERK activation is an important signaling event that is downstream of DAG. To test whether the effect of DAG-mediated signaling on T cells was mediated by ERK activation, we utilized mice with a gain-of-function of ERK (sevenmaker mutation; ERK^SEM) in T cells. This mutation makes ERK more resistant to dephosphorylation, leading to prolonged activation of ERK (22, 24). The T cell compartment of ERK-SEM mice is largely intact, having a slight reduction in total number of thymocytes but normal numbers of lymph node T cells (22). LCMV CL13 infection of ERK^SEM mice did not cause increased mortality as seen in DGKζ KO mice, although some exacerbation of weight loss was seen at Day 9 post infection ( Figure 3A ). The fraction and number of LCMV-specific T cells of WT and ERK^SEM mice were largely similar at Days 7 and 10 post infection, although slight increases were seen in the absolute number of LCMV-specific CD8⁺ T cells at Day 7 and the fraction of GP33-specific CD8⁺ T cells at Day 10 post infection in ERK^SEM mice ( Figure 3B ). Compared to DGKζ KO mice, there was only a modest increase in the fraction of KLRG1⁺ cells, whereby an increase was only seen in GP33-specific CD8⁺ T cells in ERK^SEM mice at Day 7 post infection ( Figure 3C ). Similarly, an increased fraction of degranulating and IFNγ-producing CD8⁺ T cells upon LCMV GP33 peptide restimulation was seen in ERK^SEM mice at Day 7 but not at Day 10 post infection ( Figure 3D ). Virus titers were significantly decreased in the kidneys of ERK^SEM mice at both Days 7 and 10 post infection ( Figure 3E ). The virus load in serum was ~2 log lower than kidney, but no difference was seen between WT and ERK^SEM mice ( Supplementary Figure 3B ). These results suggested that the anti-viral activity of ERK^SEM T cells was enhanced, but the activation phenotype may not have been pronounced due to decreased viral titers.

Since ERK^SEM mice had decreased viral titers relative to WT mice, next crossed ERK^SEM mice to a P14 background to enable a more rigorous comparison of T cell differentiation in a setting with normalized viral burden. To this end, naïve CD8⁺ T cells from CD45.2⁺ ERK^SEM P14 mice and CD45.1⁺ WT P14 mice were mixed in a 1:1 ratio and adoptively transferred into Thy1.1⁺ mice and infected with LCMV CL13 ( Figure 4A ). Similar to DGKζ KO P14 CD8⁺ T cells ( Figure 2B ), ERK^SEM P14 CD8⁺ T cells outcompeted WT P14 T cells at Days 7 and 10 ( Figure 4B ). Moreover, almost all P14 CD8⁺ T cells of either WT or ERK^SEM origin degranulated and >60% produced IFNγ upon restimulation with GP33 peptide ( Supplementary Figure 5B ). A statistically significant increase in the fraction IFNγ-producing of ERK^SEM P14 CD8⁺ T cells compared to WT was seen on Day 10 post infection ( Supplementary Figure 5B ). Interestingly, while DGKζ KO P14 CD8⁺ T cells disappeared at Day 14 ( Figure 2B ), ERK^SEM P14 CD8⁺ T cells continued to increase even after Day 14 post infection, representing >90% of P14 T cells by Day 28 ( Figure 4B ). Although the fraction of inhibitory receptor-expressing cells and SLEC phenotype was unchanged ( Figure 4C , Supplementary Figures 6A, B ), the proportion of ERK^SEM MPEC, T_CM, and Eomes⁺ P14 cells were increased on Day 10 post infection ( Figure 4D , Supplementary Figure 6B ). Consistent with their increased survival, the proportion of ERK^SEM P14 CD8⁺ T cells expressing Bim was significantly lower with a trend towards an increased proportion of Bcl-2-expressing cells ( Figure 4E , Supplementary Figure 5B ). These data suggest that the selective enhancement of ERK does not phenocopy DGKζ deficiency. Rather, isolated ERK augmentation leads to an increased proportion of CD8⁺ T cells with a memory phenotype and to increased proliferation and survival of CD8⁺ T cells without a loss in effector differentiation and anti-viral activity.

Since the constitutive lack of DGKζ in T cells led to the disappearance of LCMV-specific CD8⁺ T cells beyond Day 14 post LMCV CL13 infection, we could not test how DGKζ deficiency affected T cell exhaustion in the chronic phase of LCMV CL13 infection. Thus, we employed a recently developed highly potent DGKζ-selective inhibitor (ASP1570) to temporally inhibit DGKζ function. We first tested the effect of ASP1570 in the acute phase of LCMV CL13 infection. ASP1570 or vehicle was administrated for 3 Days before takedown at Day 7 post infection ( Figure 5A ). Short term ASP1570 treatment early in infection led to a higher proportion of LCMV-specific CD8⁺ T cells in blood but not in the spleen at Day 7 ( Figure 5B ). The proportion of GP276-specific but not GP33-specific CD8⁺ T cells expressing KLRG1 and displaying a SLEC phenotype was increased in ASP1570-treated mice ( Figure 5C ). However, no differences were seen in fraction of LCMV-specific CD8⁺ T cells expressing Eomes or displaying a T_CM or MPEC phenotype ( Figure 5D ). Moreover, the fraction of LCMV-specific CD8⁺ T cells expressing Bim or BCL-2 remained unchanged ( Figure 5E ). Ex vivo restimulation with GP33 peptide revealed an increased fraction of degranulating and IFNγ-producing CD8⁺ T cells ( Figure 5F ). Interestingly, there was a small but statistically significant decrease in viral titers in the kidney but not in the serum with ASP1570 treatment ( Figure 5G , Supplementary Figure 3C ), which could potentially explain some of the differences in T cell phenotype seen between the DGKζ KO mice and DGKζ inhibitor treatment. Of note, ASP1570 did not display any direct anti-viral effect against LCMV CL13 ( Supplementary Figure 7 ).

We next tested the effect of short term DGKζ inhibition in the chronic phase of LCMV CL13 infection. ASP1570 or vehicle was administrated for 3 Days before takedown at Day 35 post infection ( Figure 6A ). A significant increase in the proportion and a trend towards an increase in absolute number of LCMV-specific CD8⁺ T cells were observed after ASP1570 treatment ( Figure 6B ). The fraction of GP276-specific CD8⁺ T cells expressing KLRG1 was increased and the proportion of GP276-specific CD8⁺ T cells expressing some of the inhibitory receptors (PD-1, LAG3, TIM3, TIGIT) was decreased after ASP1570 treatment ( Figure 6C ). In the chronic phase of LCMV CL13 infection, exhausted T cells express low levels of KLRG1 (25) and the co-expression of Eomes and PD-1 marks terminally exhausted T cells (26). While there was a decrease in the proportion of exhausted (PD-1⁺KLRG1⁻) LCMV-specific CD8⁺ T cells, the fraction of the most terminally exhausted T cells (PD-1⁺Eomes⁺) was not different ( Figure 6D ). To test for T cell function, restimulation with GP33 peptide and a LCMV peptide pool was used (23), since T cells specific for other LCMV-derived peptides are increased in the later phases of LCMV CL13 infection. Although degranulation was unchanged, the fraction of IFNγ-producing CD8⁺ T cells was increased in ASP1570-treated mice at Day 35 post infection compared to vehicle-treated mice ( Figure 6E ). Importantly, short term ASP1570 treatment reduced virus titers at Day 35 post infection ( Figure 6F ). These results suggest that short-term DGKζ inhibition might provide some benefit in reinvigorating exhausted T cells in the setting of chronic viral antigen exposure.

We next sought to test the impact of the selective activation of ERK in the chronic phase of LCMV CL13 infection. LCMV CL13-infected ERK^SEM mice harbored a higher proportion and absolute number of LCMV-specific CD8⁺ T cells at Day 35 post infection compared to WT mice ( Figure 7A ). The fraction of LCMV-specific CD8⁺ T cells expressing KLRG1 was higher and those expressing inhibitory receptors ( Figure 7B ) were markedly lower in ERK^SEM compared to WT mice. Moreover, the proportion of exhausted (PD-1⁺KLRG1⁻) and the most terminally exhausted LCMV-specific CD8⁺ T cells (PD-1⁺ Eomes⁺) was significantly lower in ERK^SEM compared to WT mice ( Figure 7C ). Functionally, the fraction of CD8⁺ T cells degranulating and producing IFNγ was higher after LCMV peptide restimulation ( Figure 7D ). Moreover, ERK^SEM mice displayed reduced virus titers ( Figure 7E ). Thus, the selective activation of ERK pathway leads to favorable anti-viral T cell responses that extends into the chronic phase with less evidence of T cell exhaustion.

To examine the extent of ERK activation in DGKζ KO, ERK^SEM, and ASP1570-treated CD8⁺ T cells, we quantified phospho-ERK (pERK) after TCR stimulation. We found that both DGKζ deficiency, ERK^SEM expression, and ASP1570 treatment augmented the fraction of CD8⁺ T cells expressing pERK ( Figure 8A , Supplementary Figure 8A ). The fraction of pERK⁺ ERK^SEM CD8⁺ T cells was less than DGKζ deficiency. This could be because the ERK^SEM mutation is located only in ERK2 and thus. Moreover, the ERK^SEM mutation delays ERK dephosphorylation but may not necessarily increase the probability of ERK phosphorylation. Consistent with this notion, the amount (MFI) of pERK in pERK⁺ CD8⁺ T cells was increased ( Supplementary Figure 8B ). Since DAG also stimulates the AKT/mTOR pathway, we next examined the phosphorylation of S6 (a surrogate readout of the AKT/mTOR pathway) in DGKζ KO, ERK^SEM, and ASP1570-treated CD8⁺ T cells. We found that DGKζ deficiency and ASP1570 treatment augmented the fraction of CD8⁺ T cells expressing pERK, whereas this was not consistently seen in ERK^SEM CD8⁺ T cells ( Figure 8B , Supplementary Figure 8C ).

Blocking mTOR activity promotes the expression of Eomes, Bcl-2, and CD62L, which leads to increased memory generation and KLRG1^low cells (27). Thus, we wondered whether the enhanced mTOR signal was responsible for the difference in survival seen in DGKζ KO vs. ERK^SEM CD8⁺ T cells. To test this possibility, WT CD8⁺ P14 and DGKζ KO CD8⁺ P14 mice were mixed at a 1:1 ratio, adoptively transferred into WT mice (Thy1.1⁺), and subsequently infected with LCMV CL13 and treated with vehicle or rapamycin daily ( Figure 9A ). We found that treatment with rapamycin significantly augmented the survival advantage of DGKζ KO P14 CD8⁺ T cells at Day 10 post infection ( Figure 9B ). Strikingly, at Day 14 post infection when DGKζ KO P14 CD8⁺ T cells disappear in vehicle-treated mice, rapamycin-treated mice still show a significant survival advantage of DGKζ KO P14 CD8⁺ T cells ( Figure 9B ). The survival of DGKζ KO P14 CD8⁺ T cells at Day 14 was preceded by a decrease CD8⁺ T cells expressing Bim and an increase in cells expressing Eomes with rapamycin compared to vehicle treatment at Day 10 post infection ( Figure 9C ). However, DGKζ KO P14 CD8⁺ T cells expressing a SLEC, MPEC, or T_CM phenotype, or Bcl-2 was unaltered by rapamycin treatment at Day 10 post infection ( Figure 9C ). These results suggest that increased mTOR signaling in DGKζ KO might explain why the survival of CD8⁺ T cells is different from the selective enhancement of ERK by the ERK^SEM mutation.