Cell-Intrinsic CD38 Expression Sustains Exhausted CD8⁺ T Cells by Regulating Their Survival and Metabolism during Chronic Viral Infection

ABSTRACT

Acute and chronic viral infections result in the differentiation of effector and exhausted T cells with functional and phenotypic differences that dictate whether the infection is cleared or progresses to chronicity. High CD38 expression has been observed on CD8⁺ T cells across various viral infections and tumors in patients, suggesting an important regulatory function for CD38 on responding T cells. Here, we show that CD38 expression was increased and sustained on exhausted CD8⁺ T cells following chronic lymphocytic choriomeningitis virus (LCMV) infection, with lower levels observed on T cells from acute LCMV infection. We uncovered a cell-intrinsic role for CD38 expression in regulating the survival of effector and exhausted CD8⁺ T cells. We observed increased proliferation and function of Cd38^−/− CD8⁺ progenitor exhausted T cells compared to those of wild-type (WT) cells. Furthermore, decreased oxidative phosphorylation and glycolytic potential were observed in Cd38^−/− CD8⁺ T cells during chronic but not acute LCMV infection. Our studies reveal that CD38 has a dual cell-intrinsic function in CD8⁺ T cells, where it decreases proliferation and function yet supports their survival and metabolism. These findings show that CD38 is not only a marker of T cell activation but also has regulatory functions on effector and exhausted CD8⁺ T cells.

IMPORTANCE Our study shows how CD38 expression is regulated on CD8⁺ T cells responding during acute and chronic viral infection. We observed higher CD38 levels on CD8⁺ T cells during chronic viral infection compared to levels during acute viral infection. Deleting CD38 had an important cell-intrinsic function in ensuring the survival of virus-specific CD8⁺ T cells throughout the course of viral infection. We found defective metabolism in Cd38^−/− CD8⁺ T cells arising during chronic infection and changes in their progenitor T cell phenotype. Our studies revealed a dual cell-intrinsic role for CD38 in limiting proliferation and granzyme B production in virus-specific exhausted T cells while also promoting their survival. These data highlight new avenues for research into the mechanisms through which CD38 regulates the survival and metabolism of CD8⁺ T cell responses to viral infections.

INTRODUCTION

CD38 is an ectoenzyme expressed on the surface of most innate and adaptive immune cells and is upregulated during T cell activation (1). Upon contact between antigen-presenting cells (APCs) and CD8⁺ T cells in vitro, CD38 localizes to the immunological synapse where its enzymatic functions increase intracellular Ca²⁺ signaling (2, 3). As an ectoenzyme, CD38 converts NAD⁺ to ADP-ribose (ADPR), cyclic-ADPR (cADPR), and NAADP⁺ (1, 4–6). CD38 enzymatic activity has numerous physiological impacts, as ADPR, cADPR, and NAADP⁺ all regulate cytoplasmic Ca²⁺ levels while NAD⁺, which is consumed by CD38, is a modulator of cellular metabolism, stress response, and circadian rhythms (1, 4). Additionally, CD38 acts as a receptor on the surface of T cells, the ligation of which can further increase T cell activation through Lck-mediated activation of MAP kinase and CD3ζ signaling pathways (7). CD38 continues to be of interest, as it is highly expressed on T cells during numerous viral infections and cancers (3). CD38 is a key indicator of T cell activation during viral infection, and CD38^hi CD8⁺ T cells have been detected in patients infected with hepatitis C virus (HCV), HIV, Dengue, H1N1 influenza A virus (IAV), H7N9, Ebola, and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (8–16). However, the functional role of CD38 in the antiviral T cell response remains to be fully investigated.

While CD38 is upregulated with T cell activation during chronic infection and cancer, it is known to promote immunosuppression. CD38 negatively regulates inflammatory responses through the CD38/CD203/CD73 axis by converting NAD⁺ to adenosine (17, 18). Adenosine signaling through the adenosine A2A receptor (A2AR) on T cells diminishes effector T cell functions, resulting in a loss of tumor control (19). Further, upregulation of CD38 on tumor cells increases adenosine production and limits T cell proliferation and responses to PD-1/PD-L1 checkpoint blockade (20). However, adenosine-driven resistance to immunotherapy can be overcome by the dual blockade of A2AR and PD-1, which enhances tumor-infiltrating CD8⁺ T cell gamma interferon (IFN-γ) and granzyme B production (21).

Recently, there has been increased interest regarding the role of CD38 in T cell exhaustion, as CD38 expression was found to mark terminally exhausted T cells in tumors (22). Prolonged T cell receptor (TCR) stimulation, as in the case of chronic viral infections and cancer, leads T cells toward a dysfunctional state known as T cell exhaustion (23, 24). Exhausted T cells can differentiate into the following two distinct populations: progenitor exhausted T cells (Tpex) and terminally exhausted T cells (Tex) (25–27). The Tpex population is capable of self-renewal and can respond to immune checkpoint blockade (ICB), while the Tex population retains some effector functions but is short-lived and unresponsive to ICB. Delineating these populations is of clinical interest, as patient response to ICB can be driven by the presence or absence of Tpex and Tex cells. CD38, along with CD101, has emerged as a marker for the CD8⁺ Tex population, which has a fixed epigenetic chromatin state that prevents these cells from regaining effector functions (22). Further, resistance to PD-1 ICB is driven by a population of suboptimally primed, dysfunctional PD-1⁺ CD38^hi CD8⁺ T cells (28).

While CD38 is an established marker of terminal T cell exhaustion, whether CD38 has a role in promoting exhaustion is still being investigated. In a murine melanoma model, both CD38 deletion and overexpression in antigen-specific CD8⁺ T cells were not sufficient to alter the exhaustion phenotype of tumor-infiltrating lymphocytes (29). In the case of chronic viral infections, such as HIV, CD38 expression is correlated with disease severity (30). As patients develop AIDS, the frequency of CD38-expressing CD8⁺ T cells increases and then declines with antiretroviral treatment (31–33). However, the impact of CD38 expression on the phenotype and function of exhausted T cells in chronic viral infection is currently unknown (30). In this study, we used acute Armstrong (Arm) and chronic Clone 13 (Cl13) lymphocytic choriomeningitis virus (LCMV) to investigate the cell-intrinsic role of CD38 on virus-specific T cells during acute and chronic infection. In a cotransfer model of virus-specific wild-type (WT) and Cd38^−/− P14⁺ CD8⁺ T cells, we show that Cd38^−/− P14⁺ CD8⁺ T cells had decreased survival in both acute and chronic infection. We show that CD38 expression maintained transferred P14⁺ Tpex and Tex populations in Cl13-infected mice but also restrained proliferation and granzyme B production in Tpex cells. We found that CD38 did not alter oxidative phosphorylation and glycolysis of adoptively transferred effector T cells generated during Arm infection, but exhausted Cd38^−/− P14⁺ T cells showed reduced metabolic function in both pathways. Taken together, our study shows that CD38 is an important regulator of virus-specific CD8⁺ T cell survival in both acute and chronic infection. Our work presents an interesting paradigm for CD38 on CD8⁺ Tpex cells, in which CD38 hinders their proliferation and granzyme B production but also ensures their survival.

RESULTS

CD38 expression is upregulated on exhausted CD8⁺ T cells during chronic viral infection.

To assess how CD38 is regulated during acute and chronic viral infections, we infected WT mice with either LCMV Arm or Cl13 and measured PD-1 and CD38 expression on virus-specific T cells over the course of infection. Consistent with previous studies, PD-1 was highly expressed on LCMV major histocompatibility complex (MHC) class I tetramer⁺ CD8⁺ T cells early in chronic Cl13 infection and remained higher than acute infection, albeit to a lower level, over the course of infection (Fig. 1A). While PD-1 was also expressed during acute Arm infection, it remained significantly lower on virus-specific T cells when compared to Cl13 (Fig. 1A). As with PD-1, CD38 was also highly upregulated on LCMV MHC class I tetramer⁺ CD8⁺ T cells from Cl13-infected mice, peaking at 17 days postinfection (dpi) and remaining elevated throughout infection (Fig. 1B). CD38 was also upregulated on tetramer⁺ CD8⁺ T cells during Arm infection; however, levels were significantly lower compared to Cl13 infection. We next determined whether cells upregulate PD-1 and CD38 together during infection (Fig. 1C). In Cl13-infected mice, the majority of LCMV MHC class I tetramer⁺ CD8⁺ T cells coexpressed PD-1 and CD38 (Fig. 1C). In contrast, acutely infected mice had low frequencies of PD-1⁺ CD38⁺ tetramer⁺ CD8⁺ T cells (Fig. 1C). Next, we examined the regulation of CD38 during chronic viral infection in patients, using a previously published data set (GEO accession number GSE150345) (34). We analyzed HCV-specific CD8⁺ T cell CD38 expression in patients both during chronic HCV (cHCV) infection and after these patients received direct-acting antiviral (DAA) treatment that cured their infection. In data collected from three separate patients, CD38 expression was high during cHCV infection, and levels decreased when these patients were cured from persistent infection (Fig. 1D). These findings showed that CD38 is expressed on virus-specific CD8⁺ T cells during acute and chronic viral infection, but high expression is sustained on exhausted CD8⁺ T cells.

FIG 1.

CD38 expression and coregulation with PD-1 during acute and chronic infection. WT mice were infected with 2 × 10⁵ PFU LCMV Armstrong i.p. or 2 × 10⁶ PFU LCMV Cl13 i.v. (A) Geometric mean fluorescence intensity (MFI) of PD-1 on GP_33–41, GP_276–286, and NP_396–404 tetramer-specific CD8⁺ T cells in blood of Arm- or Cl13-infected mice. (B) Geometric mean fluorescence intensity (MFI) of CD38 on GP_33–41, GP_276–286, and NP_396–404 tetramer-specific CD8⁺ T cells in blood of Arm- or Cl13-infected mice. (C) Frequency of PD-1⁺ CD38⁺ GP_33–41, GP_276–286, and NP_396–404 tetramer-specific CD8⁺ T cells in blood of Arm- or Cl13-infected mice. Virus-specific CD8⁺ T cells were isolated from patients with hepatitis C infection, and low-input RNA-sequencing was performed by Hensel et al., 2021 (34) (GEO accession number GSE150345). (D) This data was used to graph the average normalized CD38 expression in human HCV-specific CD8⁺ T cells both during cHCV infection and after direct-acting antiviral (DAA) cure. Significance was assessed by differential analysis with multifactor design of paired samples using DESeq2. Data are representative of 3 independent experiments with ≥5 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by multiple unpaired t test.

Cell-intrinsic CD38 expression supports the survival of virus-specific CD8⁺ T cells during acute and chronic viral infection.

Since we saw CD38 expression on T cells during LCMV infection, we evaluated the cell-intrinsic role of CD38 in CD8⁺ T cells during acute and chronic viral infection. We coinjected small numbers (1 × 10³ cells/each) of WT and Cd38^−/− P14⁺ T cell receptor (TCR) transgenic CD8⁺ T cells specific for LCMV peptide GP_33–41 into WT mice (Fig. 2A). WT host mice (CD45.2⁺) received WT P14⁺ (CD45.1⁺) and Cd38^−/− P14⁺ (CD45.1⁺ CD45.2⁺) T cells at a 1:1 ratio, and 1 day later were infected with LCMV Arm or Cl13. At 8 days post Arm infection, the frequencies and numbers of Cd38^−/− P14⁺ T cells in the spleen were significantly lower than WT P14⁺ T cells (Fig. 2B to D). Over the course of infection, this phenotype persisted, with Cd38^−/− P14⁺ T cells present at lower frequencies and numbers out to 30 dpi (Fig. 2B, E, and F). Cd38^−/− P14⁺ T cells were also present at lower levels than WT P14⁺ T cells in the blood at 8 and 30 dpi (see Fig. S1A and B in the supplemental material) and trended toward lower levels in the lymph nodes (Fig. S1B). Since lower frequencies of Cd38^−/− P14⁺ T cells were observed, we next investigated the survival of the cotransferred cells. At day 6 post Arm infection, Cd38^−/− P14⁺ T cells had significantly decreased frequencies of cleaved caspase-3⁺ apoptotic cells (Fig. 2G). However, by 8 dpi and continuing to 30 dpi, this trend had reversed and Cd38^−/− P14⁺ T cells showed significantly increased frequencies of caspase-3⁺ apoptotic cells (Fig. 2G).

FIG 2.

Cell-intrinsic kinetics and survival of WT and Cd38^−/− P14⁺ T cells during acute and chronic LCMV infection. (A) WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Arm or Cl13 infection a day later. (B) Ratio of cotransferred WT and Cd38^−/− P14⁺ T cells in the spleen at 8 and 30 dpi in Arm-infected mice. (C) Individual ratios (left) and averaged ratio (right) of cotransferred WT or Cd38^−/− P14⁺ T cells from six experiments at 8 dpi in the spleen of Arm-infected mice. Representative FACS plots of cotransferred WT or Cd38^−/− P14⁺ T cells at 8 dpi in the spleen of Arm-infected mice. (D) Individual numbers (left) and averaged numbers (right) of cotransferred WT or Cd38^−/− P14⁺ T cells at 8 dpi in the spleen of Arm-infected mice. (E) Individual ratios (left) and averaged ratio (right) of cotransferred WT or Cd38^−/− P14⁺ T cells from six experiments at 30 dpi in the spleen of Arm-infected mice. Representative FACS plots (E), individual numbers (F, left) and averaged numbers (F, right) of cotransferred WT or Cd38^−/− P14⁺ T cells at 30 dpi in the spleen of Arm-infected mice. (G) Frequencies of caspase-3⁺ apoptotic WT or Cd38^−/− P14⁺ T cells at 6, 8, and 30 dpi in the spleen of Arm-infected mice and representative FACS plots. (H) Ratio of cotransferred WT and Cd38^−/− P14⁺ T cells in the spleen at 8 and 30 dpi in Cl13-infected mice. (I) Individual ratios (left) and averaged ratio (right) of cotransferred WT or Cd38^−/− P14⁺ T cells from six experiments at 8 dpi in the spleen of Cl13-infected mice. Representative FACS plots of cotransferred WT or Cd38^−/− P14⁺ T cells at 8 dpi in the spleen of Cl13-infected mice. (J) Individual numbers (left) and averaged numbers (right) of cotransferred WT or Cd38^−/− P14⁺ T cells at 8 dpi in the spleen of Cl13-infected mice. (K) Individual ratios (left) and averaged ratio (right) of cotransferred WT or Cd38^−/− P14⁺ T cells from six experiments at 30 dpi in the spleen of Cl13-infected mice. Representative FACS plots (K), individual numbers (L, left), and averaged numbers (L, right) of cotransferred WT or Cd38^−/− P14⁺ T cells at 30 dpi in the spleen of Cl13-infected mice. (M) Frequency of caspase-3⁺ apoptotic WT or Cd38^−/− P14⁺ T cells at 6, 8, and 30 dpi in the spleen of Cl13-infected mice and representative FACS plots. Data are representative of 6 independent experiments with ≥5 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

Next, we investigated the role of CD38 in virus-specific T cells during chronic infection, using the same cotransfer setup described above. Adoptively transferred Cd38^−/− P14⁺ T cells were present at similar frequencies to WT P14⁺ T cells at 8 dpi in the spleen and then dropped to significantly lower frequencies over the course of Cl13 infection (Fig. 2H, I, and K). This was true of the number of Cd38^−/− P14⁺ T cells as well, which were present at similar numbers to WT P14⁺ T cells at 8 dpi (Fig. 2J) and then significantly lower numbers at 30 dpi (Fig. 2L). Cd38^−/− P14⁺ T cells were present at lower levels than WT P14⁺ T cells in the blood at 8 dpi (Fig. S1C) and similar levels to WT P14⁺ T cells in the blood and lymph nodes at 30 dpi (Fig. S1D). When apoptosis of adoptively transferred subsets was analyzed, Cd38^−/− P14⁺ T cells initially had reduced caspase-3⁺ frequencies at 6 and 8 dpi, and then significantly higher frequencies of caspase-3⁺ apoptotic cells than WT P14⁺ T cells at 30 dpi (Fig. 2M). Because we saw a reduction of Cd38^−/− P14⁺ T cells at multiple time points during acute and chronic infection, we wondered how these cells faired in competition with WT P14⁺ T cells in uninfected mice. When WT and Cd38^−/− P14⁺ populations were examined in lymph nodes and spleens 24 h after adoptive transfer, there was a significant decrease in the frequencies of Cd38^−/− P14⁺ T cells (Fig. S1E and F). These findings showed a cell-intrinsic role for CD38 both in promoting initial residence in uninfected mice as well as survival of CD8⁺ T cells during acute and chronic viral infection.

Cd38^−/− CD8⁺ T cells are activated after acute and chronic LCMV infection.

Because we saw that CD38 deletion impacted the survival phenotype of CD8⁺ T cells during LCMV infection, we wanted to examine whether loss of CD38 also impacted the effector phenotype of CD8⁺ T cells at 8 days post Arm and Cl13 infection. WT P14⁺ (CD45.1) and Cd38^−/− P14⁺ (CD45.1⁺/CD45.2⁺) cells were cotransferred in small numbers (2 × 10³) into mice at a 1:1 ratio, and 1 day later, mice were infected with LCMV Arm or Cl13. The frequency of cotransferred granzyme B⁺ cells was increased in Cd38^−/− P14⁺ T cells (Fig. 3A), while proliferating Ki67⁺ P14⁺ cell frequencies were similar between WT and Cd38^−/− P14⁺ T cells at 8 days post Arm infection (Fig. 3B). We next evaluated inhibitory receptor expression in WT and Cd38^−/− P14⁺ T cells and observed low frequencies of PD-1⁺ TIM-3⁺ at 8 days post Arm infection, with a slight increase in Cd38^−/− cells (Fig. 3C). CD39, a marker of antigen-specific T cell engagement and an ectoenzyme capable of converting ATP to adenosine, was also analyzed on transferred P14⁺ T cells and observed to be present at significantly higher frequencies in Cd38^−/− than in WT P14⁺ T cells (Fig. 3D) (35). When cytokine production was investigated, we observed similar frequencies of IFN-γ⁺ and a small but significant increase of IFN-γ⁺ tumor necrosis factor alpha-positive (TNF-α⁺) Cd38^−/− P14⁺ T cells compared to that of WT at day 8 post Arm infection (see Fig. S2A in the supplemental material). At day 8 post Cl13 infection, Cd38^−/− P14⁺ T cells had a small but significant increase in granzyme B production (Fig. 3E) and similar frequencies of Ki67⁺ cells (Fig. 3F) when compared to those of WT P14⁺ T cells. At 8 dpi, most transferred cells were PD-1⁺, with significantly more PD-1⁺ TIM-3⁺ Cd38^−/− P14⁺ T cells than WT (Fig. 3G). The frequency of CD39⁺ cells was also higher in transferred Cd38^−/− P14⁺ T cells than in WT P14⁺ T cells (Fig. 3H). Compared to Arm infection, overall cytokine production was lower at 8 days post Cl13 infection, although Cd38^−/− P14⁺ T cells had increased frequencies of IFN-γ⁺ and IFN-γ⁺ TNF-α⁺ cells compared to those of the WT (Fig. S2B). Together, these findings demonstrate that while CD38 deletion did not alter proliferation in cotransferred P14⁺ CD8⁺ T cells at 8 dpi, loss of CD38 did promote an increase in granzyme B⁺, PD-1⁺ TIM-3⁺, and CD39⁺ cells in both Arm and Cl13 infection.

FIG 3.

Effector phenotype and function of WT and Cd38^−/− P14⁺ T cells at day 8 of acute or chronic LCMV infection. WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Arm or Cl13 infection a day later. (A) Frequency and representative FACS plots of granzyme B⁺ cotransferred WT and Cd38^−/− P14⁺ T cells at 8 dpi from spleens of Arm-infected mice. Frequency and representative FACS plots of Ki67⁺ (B), PD-1⁺ TIM-3^low (C, left), PD-1⁺ TIM-3⁺ (C, right), and CD39⁺ (D) cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Arm-infected mice at 8 dpi. (E) Frequency and representative FACS plots of granzyme B⁺ cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Cl13-infected mice at 8 dpi. Frequency and representative FACS plots of Ki67⁺ (F), PD-1⁺ TIM-3^low (G, left), PD-1⁺TIM-3⁺ (G, right), and CD39⁺ (H) cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Cl13-infected mice at 8 dpi. Data are representative of 3 independent experiments with ≥5 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

Cd38^−/− CD8⁺ T cell proliferation is increased at late stages of Cl13 infection.

While we observed similar frequencies of Ki67⁺ WT and Cd38^−/− P14⁺ T cells at 8 days post Arm and Cl13 infection, we wanted to examine whether any differences occurred at 30 dpi. WT P14⁺ (CD45.1) and Cd38^−/− P14⁺ (CD45.1⁺/CD45.2⁺) cells were cotransferred in small numbers (2 × 10³) into mice at a 1:1 ratio, and 1 day later, mice were infected with LCMV Arm or Cl13 and spleens were analyzed at 30 dpi. The frequency of granzyme B⁺ and Ki67⁺ cells were ~10% in both WT and Cd38^−/− P14⁺ T cells at 30 days post Arm infection (Fig. 4A and B). At 30 dpi, most WT and Cd38^−/− P14⁺ T cells were PD-1⁻ TIM3⁻ (Fig. 4C). The majority of WT and Cd38^−/− P14⁺ T cells at 30 dpi were IFN-γ⁺ TNF-α⁺, with no differences in cytokine production observed (Fig. S2C). With Cl13 infection, we observed similar frequencies of granzyme B⁺ cotransferred cells at 30 dpi (Fig. 4D). However, Cd38^−/− P14⁺ T cells had a significant increase in their proliferation as shown by the increase in Ki67⁺ cells at 30 dpi (Fig. 4E). Most transferred cells were PD-1⁺ TIM-3^low at 30 dpi, and no differences were apparent in PD-1⁺ TIM-3⁺ populations between WT and Cd38^−/− P14⁺ T cells (Fig. 4F). Compared to Arm infection, there were decreased frequencies of IFN-γ⁺ TNF-α⁺ transferred cells at day 30 post Cl13 infection; however, no differences were observed between WT and Cd38^−/− P14⁺ T cells (Fig. S2D). These findings showed that CD38 expression limited the proliferation of exhausted CD8⁺ T cells while having minimal impact on cytokine production and inhibitory receptor expression.

FIG 4.

Effector phenotype and function of WT and Cd38^−/− P14⁺ T cells at day 30 of acute or chronic LCMV. WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Arm or Cl13 infection a day later. (A) Frequency and representative FACS plots of granzyme B⁺ cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Arm-infected mice at 30 dpi. Frequency and representative FACS plots of Ki67⁺ (B) and PD-1⁺TIM-3^low (C) cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Arm-infected mice at 30 dpi. (D) Frequency and representative FACS plots of granzyme B⁺ cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Cl13-infected mice at 30 dpi. Frequency and representative FACS plots of Ki67⁺ (E), PD-1⁺ TIM-3^low (F, left), and PD-1⁺ TIM-3⁺ (F, right) cotransferred WT and Cd38^−/− P14⁺ T cells from spleens of Cl13-infected mice at 30 dpi. Data are representative of 6 independent experiments with ≥4 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

CD38 expression promotes the maintenance of Tpex and Tex subsets during chronic infection.

During chronic viral infection and cancer, exhausted T cells form two distinct populations, progenitor (Tpex) and terminal (Tex) exhausted T cells (27). Tpex cells are characterized by expression of the transcription factor TCF-1 and are capable of self-renewal and seeding of the Tex population, which retain better effector function but are short-lived (26, 27). To assess how CD38 may impact Tpex and Tex populations, we cotransferred small numbers (2 × 10³) of WT P14⁺ and Cd38^−/− P14⁺ T cells into mice at a 1:1 ratio, infected mice with Cl13 1 day later, and then analyzed Tpex and Tex populations in the CD45.1⁺ transferred cells in the spleen at 8 and 30 dpi. Although both WT and Cd38^−/− P14⁺ T cells differentiated into Tpex and Tex subsets at 8 dpi, the frequencies and numbers of Cd38^−/− P14⁺ Tpex were significantly decreased compared to WT (Fig. 5A). In contrast, the Tex population was composed of similar frequencies and numbers of WT and Cd38^−/− P14⁺ T cells at 8 dpi (Fig. 5B). On day 30 post Cl13 infection, the Cd38^−/− P14⁺ Tpex and Tex frequencies and numbers were significantly lower than WT P14⁺ (Fig. 5C and D). When the frequencies of Tpex and Tex were tracked over the course of chronic infection, we found that Cd38^−/− P14⁺ T cells were more terminal than WT at 8 dpi (Fig. 5E). We next evaluated apoptosis in the transferred Tpex and Tex populations during infection and found that Cd38^−/− P14⁺ Tpex had more caspase-3⁺ cells at 6 and 8 dpi and trended toward more at 30 dpi than WT P14⁺ Tpex (Fig. 5F). Cd38^−/− P14⁺ Tex showed reduced caspase-3⁺ cells at 6 and 8 dpi and increased caspase-3⁺ cells at 30 dpi compared to that of WT P14⁺ Tex (Fig. 5F). Further analysis of exhausted Cd38^−/− P14⁺ T cells showed increased frequencies of effector markers such CD69 and KLRG1 when compared to those of the WT (Fig. 5G and H) (36). Expression of the terminal marker 2B4⁺ was similar in Cd38^−/− and WT P14⁺ T cells at 8 and 30 dpi (Fig. 5G and H). The frequencies of TCF-1⁺ cells were reduced at 8 and 30 dpi with CD38 deletion (Fig. 5G and H). These findings showed that CD38 expression is important for the survival and maintenance of Tpex and Tex cells, as well as TCF-1 expression, over the course of chronic infection. Additionally, cells lacking CD38 show increased expression of markers associated with effector-like exhausted populations.

FIG 5.

Analysis of exhausted populations in WT and Cd38^−/− P14⁺ T cells during Cl13 infection. WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Cl13 infection a day later. (A) Frequencies of WT or Cd38^−/− P14⁺ T cells of total cotransferred Tpex cells, representative FACS plots, and numbers of total Tpex at 8 dpi in spleen. (B) Frequencies of WT or Cd38^−/− P14⁺ T cells of total cotransferred Tex cells, representative FACS plots, and numbers of total Tex at 8 dpi in spleen. (C) Frequencies of WT or Cd38^−/− P14⁺ T cells of total cotransferred Tpex cells, representative FACS plots, and total numbers of Tpex at 30 dpi in spleen. (D) Frequencies of WT or Cd38^−/− P14⁺ CD8⁺ T of total cotransferred Tex cells, representative FACS plots, and total numbers of Tex at 30 dpi in spleen. (E) Frequencies of cotransferred Tpex and Tex cells in the spleen over the course of Cl13 infection. (F) Frequencies of apoptotic caspase-3⁺ cotransferred Tpex and Tex in the spleen over the course of Cl13 infection. Frequencies of 2B4⁺, CD69⁺, KLRG1⁺ and TCF1⁺ cotransfer cells at 8 dpi (G) and 30 dpi (H) in the spleen. Data are representative of 3 (8 dpi spleen), 6 (30 dpi spleen), or 2 (2B4 and CD69 expression) independent experiments with ≥4 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

CD38 negatively regulates the proliferation and function of progenitor exhausted CD8⁺ T cells.

Because we saw an increase in surface markers indicative of effector-like exhausted cells, we next examined the impact of CD38 deletion on the effector functions of Tpex and Tex transferred cells. On day 8 post Cl13 infection, the frequency of granzyme B⁺ Tpex cells was significantly increased in Cd38^−/− P14⁺ T cells compared to that of the WT (Fig. 6A), whereas Ki67⁺ cells were similar (Fig. 6B). Cd38^−/− P14⁺ Tex cells had a small but significant increase in granzyme B⁺ cells and similar Ki67⁺ cells compared to those of WT P14⁺ Tex cells (Fig. 6C). At 30 dpi, Cd38^−/− P14⁺ Tpex had significantly more Granzyme B production and Ki67 than WT (Fig. 6D and E). Cd38^−/− P14⁺ Tex had similar Granzyme B and increased Ki67 production compared to WT P14⁺ Tex (Fig. 6F). These findings showed that CD38 expression restrained granzyme B production and proliferation of Tpex cells during chronic infection.

FIG 6.

Cd38^−/− P14⁺ Tpex and Tex cell phenotypes in Cl13 infection. WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Cl13 infection a day later. (A) Frequencies of granzyme B⁺ WT or Cd38^−/− Tpex cells at 8 dpi in spleen and representative FACS plots. (B) Frequencies of Ki67⁺ WT or Cd38^−/− Tpex cells at 8 dpi in spleen and representative FACS plots. (C) Frequencies of granzyme B⁺ and Ki67⁺ cotransferred WT or Cd38^−/− Tex P14⁺ T cells in the spleen. (D) The 8 dpi frequencies of granzyme B⁺ WT or Cd38^−/− Tpex cells at 30 dpi in spleen and representative FACS plots. (E) Frequencies of Ki67⁺ WT or Cd38^−/− Tpex cells at 30 dpi in spleen and representative FACS plots. (F) The 30 dpi frequencies of granzyme B⁺ and Ki67⁺ cotransferred WT or Cd38^−/− Tex P14⁺ T cells in the spleen. Data are representative of 3 (8 dpi spleen) or 6 (30 dpi spleen) independent experiments with ≥4 mice per group. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

CD38 increased the metabolic fitness of CD8⁺ T cells during chronic infection.

Since CD38 has enzymatic functions, we next examined whether metabolic changes occurred in Cd38^−/− P14⁺ T cells. To examine the impact of CD38 expression on mitochondrial respiration and glycolysis during LCMV infection, we carried out the Seahorse XF Cell Mito stress test on live-sorted WT and Cd38^−/− P14⁺ T cells from Arm or Cl13-infected mice at 8 dpi. CD38 deletion during Arm infection did not impact the P14⁺ cell oxidative phosphorylation (Fig. 7A and B). Basal and maximal oxygen consumption rate (OCR) (a proxy for oxidative phosphorylation) and extracellular acidification rate (ECAR) (a proxy for glycolysis) were indistinguishable between WT and Cd38^−/− P14⁺ T cells from Arm-infected mice (Fig. 7C and D). The ratio of OCR/ECAR was also quantified to gain a relative measure of cellular preference for oxidative phosphorylation or glycolysis (Fig. 7E). At 8 days post Arm infection, CD38 deletion did not significantly impact the OCR/ECAR ratio. In contrast to Arm infection, CD38 deletion in P14⁺ T cells from mice infected with Cl13 resulted in decreased oxidative phosphorylation and glycolysis (Fig. 7F and G). The basal and maximal OCR of Cd38^−/− P14⁺ T cells were decreased compared to those of WT P14⁺ T cells from Cl13-infected mice (Fig. 7H). Basal and maximal ECAR levels were also significantly decreased with CD38 deletion (Fig. 7I). Basal and maximal OCR/ECAR ratios were similar between WT and Cd38^−/− P14⁺ T cells from Cl13-infected mice (Fig. 7J). Relative to Arm infection, the larger OCR/ECAR ratios in Cl13 samples demonstrated a preference for oxidative phosphorylation (Fig. 7J). These findings showed that cell intrinsic CD38 expression sustained the basal and maximal metabolic function of exhausted CD8⁺ T cells during Cl13 infection.

FIG 7.

Metabolism of Cd38^−/− virus-specific CD8⁺ T cells. WT and Cd38^−/− P14⁺ CD8⁺ T cells were transferred at a 1:1 ratio (1 × 10³ each) into WT naive mice, followed by LCMV Arm or Cl13 infection a day later. Live WT and Cd38^−/− P14⁺ CD8⁺ T cells were isolated from host mice at 8 dpi and a Seahorse XF Cell Mito stress test was performed. (A) Oxygen consumption rate (OCR) of live WT and Cd38^−/− P14⁺ T cells from Arm-infected mice at baseline and after addition of oligomycin, FCCP, and rotenone & antimycin A. (B) Extracellular acidification rate (ECAR) of live WT and Cd38^−/− P14⁺ T cells from Arm-infected mice at baseline and after addition of Mito stress test compounds. (C–E) Basal and maximal OCR and ECAR rates and quantified OCR/ECAR ratio for isolated WT and Cd38^−/− P14⁺ T cells from Arm-infected mice. (F, G) Seahorse XF Cell Mito stress test OCR and ECAR values from live WT and Cd38^−/− P14⁺ T cells isolated from Cl13-infected mice at 8 dpi. (H–J) Basal and maximal OCR and ECAR rates and quantified OCR/ECAR ratio for isolated WT and Cd38^−/− P14⁺ T cells from Cl13-infected mice. Seahorse data presented are median values of a representative biological replicate. Data are representative of 3 independent experiments with ≥5 mice per group for Arm and ≥15 mice for Cl13 infection. Data show mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001 by paired t test.

DISCUSSION

CD38 upregulation on activated T cells has been associated with response to infections (30). More recently, CD38 was found to mark terminal T cell dysfunction and has gained interest as a possible pharmacological target for the enhancement of T cell responses (22, 37). However, the functional role of CD38 on T cells in response to acute and chronic viral infection remains largely undefined. In this study, we used an adoptive cotransfer of Cd38^−/− P14⁺ and WT P14⁺ CD8⁺ T cells to investigate the cell-intrinsic role of CD38 on virus-specific T cells. We found that CD38 promotes the survival of CD8⁺ T cells during LCMV infection. In an acute infection, proliferation and granzyme B production were not limited by CD38 expression. While CD38 expression seems to play a relatively minor role in shaping the effector response of CD8⁺ T cells to acute infection, we found that CD38 is an important regulator of cell proliferation and granzyme B production, Tpex and Tex phenotype and maintenance, and metabolism during chronic infection.

During HIV infection, CD38 is upregulated on T cells and can serve as a marker of disease progression (32, 38, 39). In cases of H7N9 avian influenza, a larger population of CD38⁺ HLA-DR⁺ CD8⁺ T cells are found in patients who succumb to flu (13). In both Dengue- and Ebola-infected patients, CD8⁺ CD38⁺ T cells are upregulated in febrile patients (11, 13, 40). In concurrence with these data, we found that CD38 levels were elevated on T cells during Arm and Cl13 infection, although to a much higher extent in Cl13. Our study and others have shown that CD38 is coexpressed with PD-1 and correlates with activation as well as exhaustion (13). As with PD-1, CD38 is expressed during activation in response to acute infection, and CD38 levels are elevated and sustained on exhausted T cells during chronic infection. This differential expression is a result of divergent chromatin accessibility in T cells during acute and chronic infection (41). Regions of the Pdcd1 locus become uniquely accessible at day 5 and onward during tumorigenesis and at day 8 postinfection and onward during chronic infection (22, 41, 42). Similarly, the Cd38 locus is uniquely open in both virus and tumor-specific exhausted T cells (22, 43). The expression patterns and chromatin accessibility of CD38 make it a particularly good marker of chronic T cell stimulation and raises the question of how CD38 signaling may shape the phenotype of activated T cells. Interestingly, our analysis of the Hensel et al. (34) study showed that CD38 gene expression levels were high on CD8⁺ T cells from patients with chronic HCV, but these were decreased when the same patients were cured from their infection, suggesting that CD38 expression may be linked to persistent TCR stimulation.

As an ectoenzyme, the immunosuppressive function of CD38 has been well established through its contribution in regulating extracellular adenosine levels. CD38, in conjunction with CD203a and CD73 enzymatic activities, converts extracellular NAD⁺ to adenosine (17). In the tumor microenvironment, CD38-expressing tumor cells and myeloid-derived suppressor cells produce adenosine to limit cytotoxic T cell function through the adenosine receptor (20, 44, 45). Blocking the immunosuppressive adenosine pathway through targeting of the adenosine (A2A) receptor or CD73 has proven beneficial to antitumor T cell responses (21, 45, 46). While less is known about the cell-intrinsic role of CD38 on virus- or tumor-specific T cells, evidence suggests CD38 as a possible inhibitor of T cell function (11, 22, 28). We found that CD38 deletion did not impact cytokine production of CD8⁺ T cells during Arm and Cl13 infection. While CD38 deletion did promote increased PD-1 and TIM-3 expression at 8 dpi, by 30 dpi the impact of deletion on inhibitory receptor expression was minimal. These findings are consistent with a recent study showing that overexpression or deletion of CD38 on tumor-specific T cells did not alter their exhaustion phenotype (29). CD73 is downstream of CD39, another ectoenzyme capable of initiating conversion of ATP to adenosine (47, 48). The limited impact of CD38 deletion on virus-specific T cells at 8 days post LCMV infection may be partially due to the redundancy of enzymatic functions of CD39, CD73, and CD203 in adenosine formation (17). This is relevant, as we saw an increase in CD39⁺ cells in Cd38^−/− P14⁺ T cells. This indicates that CD38 deletion alone may not be enough to alter immunosuppressive adenosine signaling.

While effector functions were largely similar in cotransferred WT and Cd38^−/− P14⁺ CD8⁺ T cells at 30 dpi, we discovered an important cell-intrinsic role for CD38 expression in the survival of virus-specific CD8⁺ T cells. We found that adoptively transferred Cd38^−/− P14⁺ T cells were significantly decreased by day 30 post Arm and Cl13 infection due in part to increased apoptosis. Interestingly, when we performed cotransfers of naive Cd38^−/− and WT P14⁺ T cells into uninfected mice, cells lacking CD38 had diminished frequencies by 24 h. This indicated that CD38 may also play a cell-intrinsic role of homing of naive T cells. Our data indicate that Cd38^−/− P14⁺ T cells can expand in response to viral infection (as indicated by the similar frequencies in spleens of Cl13-infected mice at 8 dpi); however, these cells are unable to persist over the course of infection. A similar loss of adoptively transferred Cd38^−/− tumor-specific T cells compared to WT was seen in tumors 14 days after transfer (29). Interestingly, the decreased frequency of Cd38^−/− T cells was not attributed to apoptosis in the tumor model, whereas in the LCMV model, Cd38^−/− P14⁺ T cells had increased caspase-3 expression. Further research is needed to define the cell-intrinsic mechanism that drives the loss of Cd38^−/− T cells in uninfected mice and in tumor models. TCR signaling has been shown to regulate CD38, and a corresponding role for CD38 in promoting survival could explain its persistent expression on memory and exhausted T cells in mice and in patients with viral infections and autoimmune disorders (39, 49–52). Given its upregulation on activated and memory T cells, and considering we show that transferred Cd38^−/− P14⁺ T cells undergo significantly more apoptosis than WT P14⁺ T cells, it appears that CD38 expression promotes maintenance and survival of activated T cells. In vitro, CD38 limits proliferation of activated T cells, and small interfering RNA targeting of CD38 can increase T cell proliferation (28, 49). In accordance with this, we found that proliferation of adoptively transferred WT P14⁺ T cells was significantly constrained during Cl13 infection when compared to Cd38^−/− P14⁺ T cells at 30 dpi.

While we found that CD38 expression did not alter the development of T cell exhaustion, both Tpex and Tex frequencies and numbers were reduced during Cl13 infection with CD38 deletion. Phenotypically, we observed an increased effector response in Cd38^−/− P14⁺ Tpex cells, as shown by more proliferation and granzyme B production. The Tpex population is defined by high expression of the TCF-1 transcription factor and is vital for antiviral and antitumor T cell responses, as Tpex cells are stem-like, able to expand into the Tex population, and can respond to ICB (25, 27). Compared to Tpex cells, Tex cells are more effector-like, having higher cytokine and granzyme B production, as well as increased proliferation (26). In this context, it appears that CD38 may have a cell-intrinsic role in limiting an effector-like Tpex phenotype. However, while CD38 deletion promoted proliferation and KLRG1 expression, the overall frequencies of Cd38^−/− Tpex and Tex cells were decreased in cotransfer experiments, and Tpex cells lacking CD38 had higher apoptotic frequencies than their WT counterparts. Further, we found that cells expressing CD38 maintained higher TCF-1 frequencies over time, an important factor in Tpex formation. Our data present an interesting paradigm, in which CD38 hinders proliferation and granzyme B production in CD8⁺ T cells, particularly in the Tpex subset, but also ensures their survival.

Since we saw that CD38 deletion impacted survival of the transferred P14⁺ T cells during viral infection, we wanted to investigate whether cellular metabolism was also impacted by loss of CD38. The enzymatic activities of CD38 have been reviewed at length (3, 30, 53–55), establishing CD38 as a prominent regulator of cellular NAD⁺, NAADP, ADPR, and cADPR levels. The consumption of NAD⁺ by CD38 has the potential to limit T cell metabolism and effector functions, as NAD⁺ is a key substrate of the immunomodulatory proteins ART, PARP, and SIRT, which are involved in T cell fate, survival, and metabolism (3, 55). In vitro, CD38 deletion supports increased oxidative phosphorylation in CD4⁺ T cells and supports NAD⁺ levels in cultures of CD8⁺ T cells (29, 56). Interestingly, our analysis of virus-specific CD8⁺ T cell metabolism ex vivo revealed similar metabolic profiles in WT and Cd38^−/− P14⁺ T cells from Arm-infected mice. Given the established NADase activity of CD38, along with previous in vitro findings, we anticipated that CD38 deletion would increase the metabolic capabilities of transferred P14⁺ T cells. To our surprise, we discovered that Cd38^−/− P14⁺ T cells from Cl13 infection had reduced oxidative phosphorylation and glycolysis both at basal and maximal conditions. This is opposite of the predicted phenotype arising from NAD⁺ depletion. However, given that WT and Cd38^−/− P14⁺ T cells were transferred into C57BL/6 hosts with functional CD38, along with the recent finding that NAD⁺ levels are similar in B16-F10 tumors of WT and Cd38^−/− mice, the impaired metabolism seen in Cd38^−/− P14⁺ T cells may be a result of cell-intrinsic CD38 signaling rather than enzymatic activity (29). Additionally, the impact of CD38 on metabolism appears to be correlated with expression levels, as WT P14⁺ T cells from Arm infection expressed much lower levels of CD38 than those from Cl13 and were spared metabolic impacts upon CD38 deletion. This represents an exciting potential role for CD38 signaling in the mitochondrial fitness of P14⁺ T cells during Cl13 infection.

Our study shows the regulation of CD38 expression in effector and exhausted CD8⁺ T cells during viral infection. We found that CD38 is upregulated and expressed at higher levels on CD8⁺ T cells in Cl13 than in Arm infection. We showed that CD38 deletion has a cell-intrinsic role in regulating the survival of virus-specific CD8⁺ T cells throughout the course of viral infection. We found no differences in the metabolism of WT and Cd38^−/− P14⁺ T cells from Arm-infected mice, but Cd38^−/− cells arising during Cl13 infection had decreased oxidative phosphorylation and glycolysis. Cd38^−/− P14⁺ Tpex and Tex cells were decreased, and Cd38^−/− Tpex cells were more proliferative and functional than WT cells, indicating an inhibitory function of CD38 in this cell type. Further, exhausted Cd38^−/− T cells were decreased in TCF-1⁺ populations and increased in apoptotic Tpex cells during chronic infection. Our studies show a dual cell-intrinsic role for CD38 in limiting proliferation of virus-specific T cells while also promoting their survival. These data uncover important roles of CD38 in virus-specific T cell responses to infection and highlight new avenues for research into the mechanisms through which CD38 regulates the survival, effector function, and metabolism of exhausted CD8⁺ T cells.

MATERIALS AND METHODS

Mice.

Experimental male C57BL/6J (no. 000664) mice were obtained from The Jackson Laboratory and were used at 6 to 9 weeks of age. B6.129P2-Cd38^tm1Lnd/J (Cd38^−/−, no. 003727) mice were obtained from The Jackson Laboratory and bred for experiments. P14⁺ mice were obtained from The Scripps Research Institute (originally from Charles D. Surh) and were crossed with Cd38^−/− mice to obtain P14⁺ Cd38^−/− mice. Mice were used at ≥6 weeks of age. Animal care was in accordance with the UC Irvine Institutional Animal Care and Use Committees.

Virus infection.

Lymphocytic choriomeningitis (LCMV) Armstrong and Cl13 strains were propagated in baby-hamster kidney cells and titrated on Vero African green monkey kidney cells. Frozen stocks were diluted in Vero cell media, and 2 × 10⁵ PFU of LCMV Armstrong was injected intraperitoneally (i.p.) or 2 × 10⁶ PFU of LCMV Cl13 was injected intravenously (i.v.). For seahorse metabolism studies, mice were infected with Arm (i.p.) or Cl13 retro-orbitally (r.o.).

T cell adoptive transfer.

CD8⁺ T cells were enriched from spleens and lymph nodes (LNs) of WT or Cd38^−/− P14⁺ transgenic mice by column-free magnetic negative selection. Single cell suspensions from pooled spleen and LNs were incubated with biotinylated antibodies purchased from BioLegend against CD4 (GK1.5), B220 (RA3-6B2), CD19 (6D5), CD24 (M1/69), CD11b (M1/70), and CD11c (N418). Labeled cells were removed by mixing cell suspension with Streptavidin RapidSpheres (Stemcell Technologies) at room temperature (RT) for 5 min, followed by two 5-min incubations in an EasyEights EasySep Magnet (Stemcell Technologies). Enriched CD8⁺ T cells were washed in sterile phosphate-buffered saline (PBS) (1×) with fetal bovine serum (FBS) (2%) and purity determined by flow cytometry. For uninfected cotransfer experiments, WT P14⁺ (CD45.1⁺) and Cd38^−/− P14⁺ (CD45.1⁺ CD45.2⁺) T cells normalized (CD8a⁺ Va2⁺) and mixed at a 1:1 ratio (1 × 10⁴ cells/genotype) and injected retro-orbitally into WT (CD45.2⁺) recipient mice. Blood, spleen, and LNs were analyzed for adoptively transferred populations 24 h after injection. For infected cotransfer experiments, live WT P14⁺ (CD45.1⁺) or Cd38^−/− P14⁺ (CD45.1⁺ CD45.2⁺) T cells were normalized (CD8a⁺ Va2⁺) and mixed at a 1:1 ratio (1 × 10³ cells/genotype) and injected into WT (CD45.2⁺) recipient mice i.v. These mice were infected with LCMV 1 day later. Mice were bled r.o., and spleens and lymph nodes were isolated at the indicated time points. Schematic of the cotransfer was created with BioRender (https://www.biorender.com/).

Flow cytometry.

For cell surface staining, 2 × 10⁶ cells were incubated with antibodies in staining buffer (PBS, 2% FBS, and 0.01% NaN₃) and fixed in PBS with 1.85% formaldehyde for 20 min on ice. For LCMV tetramer staining, cells were incubated with H-2D^b-GP_33–41, H-2D^b-GP_276-286, H-2D^b-NP_396–404, or IA^b-_66–77 tetramers (NIH Core Facility) for 1 h 15 min at room temperature in staining buffer and then fixed with PBS with 1.85% formaldehyde for 20 min on ice. For intracellular cytokine stimulation and staining, cells were resuspended in complete RPMI 1640 (containing 10 mM HEPES, 1% nonessential amino acids and l-glutamine, 1 mM sodium pyruvate, 10% heat-inactivated fetal bovine serum [FBS], and antibiotics) supplemented with 50 U/mL IL-2 (NCI) and 1 mg/mL brefeldin A (BFA) (Sigma) and then incubated with 2 mg/mL LCMV GP_33-41 peptide (AnaSpec) at 37°C for 4 h. Cells were then fixed and permeabilized using a Cytofix/Cytoperm kit (BD Biosciences) before staining. For intranuclear transcription factor and Ki67 staining, cells were fixed and permeabilized using a Foxp3/transcription factor fixation/permeabilization kit (Fisher) and then stained with anti-human granzyme B (GB12) from Thermo Fisher and Ki67 (B56) from BD Bioscience. Fluorochrome-conjugated antibodies CD44 (IM7), CD45.1 (A20), CD45.2 (104), CD8α (53-6.7), CD4 (RM4-5), CD38 (90), CD160 (7H1), CD223 (C9B7W), CD279 (RMP1-30), CD366 (RMT3-23), CD244.2 (m2B4 [B6] 458.1), KLRG1 (2FI/KLRG1), and CD69 (H1.2F3) were purchased from BioLegend. Fluorochrome-conjugated Ly-108 (13G3) was purchased from BD Biosciences. Fluorochrome-conjugated TCF-1 (C63D9) was purchased from Cell Signaling Technology. Surface stains were performed at a 1:200 dilution, while intracellular and intranuclear stains were performed at a 1:100 dilution. Caspase-3 staining was done using PI (Sigma-Aldrich) and CaspGLOW fluorescein active caspase-3 staining kit (Thermo Fisher) and following manufacturer’s instructions. All data were collected on a NovoCyte 3000 (Agilent) and analyzed using FlowJo Software (Tree Star).

Seahorse assay.

Cotransferred WT or Cd38^−/− P14⁺ T cells were isolated from spleens of host mice at 8 days post Armstrong or Cl13 infection. CD8⁺ T cells were purified by negative selection (see methods in “T cell adoptive transfer”); stained for 10 min with propidium iodide (PI) in complete media RPMI 1640 (containing 10 mM HEPES, 1% nonessential amino acids and l-glutamine, 1 mM sodium pyruvate, 10% heat inactivated FBS, and 1% antibiotics [penicillin, streptomycin, l-glutamine] from Corning); surface stained with CD8 BV785 (53-6.7), CD45.1 PB (A20), and CD45.2 fluorescein isothiocyanate (FITC) (104) from BioLegend; and sorted on a BD FACSAria Fusion at >95% purity. Fluorescence-activated cell sorting (FACS) was used to isolate WT P14⁺ (PI⁻ CD8⁺ CD45.1⁺ CD45.2⁻) and Cd38^−/− P14⁺ (PI⁻ CD8⁺ CD45.1⁺ CD45.2⁺) T cells. Live cells were counted immediately after sorting and adhered onto the wells of a poly-d-lysine coated XF96 plate at between 500,000 and 550,000 cells/well (exact counts were noted and used for normalization). The Seahorse XF Mito stress test was performed to measure OCR and ECAR of cells plated in XF media (nonbuffered Dulbecco modified Eagle medium [DMEM] containing 10 mM glucose, 4 mM l-glutamine, and 2 mM sodium pyruvate) under basal conditions and in response to 1 μM oligomycin (Calbiochem), 1.5 μM fluoro-carbonyl cyanide phenylhydrazone (FCCP) (Enzo), and 1 μM rotenone + 1 μM antimycin A (Enzo) with the XF96 extracellular flux analyzer (Seahorse Bioscience). Data were normalized using Wave software (Agilent). Basal respiration values were determined as the mean OCR of the last 3 baseline data points minus the median of the 3 OCR data points after antimycin A and rotenone addition (57). Maximal OCR values were determined as the maximum of the three OCR values after FCCP addition minus the median of the last 3 values after rotenone and antimycin A addition. Basal ECAR values were determined as the mean of the last 3 baseline points. Maximal ECAR values were determined as the highest ECAR point after addition of FCCP. Basal OCR/ECAR was determined as the mean OCR of the last 3 baseline data points divided by the mean of the ECAR of the last 3 baseline data points. Maximal OCR/ECAR was determined as the highest of the 3 OCR values after the addition of FCCP divided by the corresponding ECAR value.

Quantification and statistical analysis.

Flow cytometry data were analyzed with FlowJo software (TreeStar). Graphs of flow cytometry data and Seahorse data were prepared with GraphPad Prism software. GraphPad Prism was used for statistical analysis to compare outcomes using a two-tailed unpaired Student's t test, Mann-Whitney, or a two-tailed paired t test where indicated; significance was set to P ≤ 0.05 and represented as follows: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; and ****, P ≤ 0.0001. Error bars show standard error of the mean (SEM).