Research ArticleImmunodeficiency

Inhibition of diacylglycerol kinase α restores restimulation-induced cell death and reduces immunopathology in XLP-1

See allHide authors and affiliations

Science Translational Medicine  13 Jan 2016:
Vol. 8, Issue 321, pp. 321ra7
DOI: 10.1126/scitranslmed.aad1565

SAPping immunopathology

Individuals with deficient immune systems may also paradoxically experience hyperimmune side effects. X-linked lymphoproliferative disease (XLP-1)—an immunodeficiency caused by defects in the T cell receptor adaptor protein SAP [signaling lymphocytic activation molecule (SLAM)–associated protein]—is associated with expansion of activated T cell after viral infection. Now, Ruffo et al. report that down-regulating diacylglycerol kinase α (DGKα) in SAP-deficient T cells restores restimulation-induced cell death, preventing this excess expansion. If these data hold true in humans, targeting DGKα may prevent viral-induced immunopathology in XLP-1 patients.

Abstract

X-linked lymphoproliferative disease (XLP-1) is an often-fatal primary immunodeficiency associated with the exuberant expansion of activated CD8+ T cells after Epstein-Barr virus (EBV) infection. XLP-1 is caused by defects in signaling lymphocytic activation molecule (SLAM)–associated protein (SAP), an adaptor protein that modulates T cell receptor (TCR)–induced signaling. SAP-deficient T cells exhibit impaired TCR restimulation-induced cell death (RICD) and diminished TCR-induced inhibition of diacylglycerol kinase α (DGKα), leading to increased diacylglycerol metabolism and decreased signaling through Ras and PKCθ (protein kinase Cθ). We show that down-regulation of DGKα activity in SAP-deficient T cells restores diacylglycerol signaling at the immune synapse and rescues RICD via induction of the proapoptotic proteins NUR77 and NOR1. Pharmacological inhibition of DGKα prevents the excessive CD8+ T cell expansion and interferon-γ production that occur in SAP-deficient mice after lymphocytic choriomeningitis virus infection without impairing lytic activity. Collectively, these data highlight DGKα as a viable therapeutic target to reverse the life-threatening EBV-associated immunopathology that occurs in XLP-1 patients.

INTRODUCTION

X-linked lymphoproliferative disease (XLP-1) is a heritable immune disorder caused by germline mutations in the SH2D1A gene, which encodes the signaling lymphocytic activation molecule (SLAM)–associated protein (SAP) (1). SAP is a small SH2 domain–containing adaptor primarily expressed in T, natural killer (NK), and invariant NKT (iNKT) cells (1). XLP-1 is best recognized for the increased susceptibility of affected males to develop overwhelming lymphoproliferation after primary Epstein-Barr virus (EBV) infection (2). Also known as fulminant infectious mononucleosis (FIM), this lymphoproliferative process is characterized by the massive accumulation of activated CD8+ T cells, which infiltrate multiple organs and inflict severe tissue damage. FIM is the most common and clinically challenging manifestation of XLP-1, with up to 65% of patients dying despite the use of chemoimmunotherapy (3). Accordingly, alternative and more effective treatment strategies are sorely needed for XLP-1 patients who develop FIM.

T lymphocytes derived from XLP-1 patients exhibit multiple functional defects, including reduced cytotoxic activity (4) and impaired restimulation-induced cell death (RICD) (5). RICD is a self-regulatory apoptosis program triggered by repeated T cell receptor (TCR) stimulation that maintains peripheral immune homeostasis by constraining the accumulation of activated T cells (6). A similar death defect is present in the activated T cells of Sh2d1a−⁄− mice (7). It is proposed that defective RICD, combined with impaired clearance of EBV-infected B cells, sustains and amplifies the expansion of activated T cells that typifies FIM (5, 6).

SAP binds to immunotyrosine-based switch motifs within the cytoplasmic domains of the SLAM family receptors (SLAM-Rs) (8), thus competing with the binding of SH2 domain–containing inhibitory lipid and tyrosine phosphatases such as SH2-containing inositol polyphosphate 5-phosphatase (SHIP) and SH2-containing protein tyrosine phosphatase 1 and 2 (SHP-1)/SHP-2 (9). In addition, SAP facilitates recruitment of kinases such as FynT and Lck to SLAM-Rs to promote optimal signaling within T, NK, and NKT cells (10, 11). Indeed, RICD resistance in XLP patient T cells results in part from weak TCR signaling associated with excess SHP-1 activity and defective recruitment of Lck to the NTB-A receptor, which colocalizes with the TCR (5, 11). Although SAP links SLAM-R signaling to several downstream functions via activation of Src-family kinases [for example, interleukin-4 (IL-4) secretion (12), iNKT cell development (13)], this signaling axis is not the only pathway in which SAP is involved for signal regulation. For example, the requirement for SAP in the provision of CD4+ T cell–mediated “help” for B cell differentiation is Fyn-independent (14). To fully understand XLP-1 pathogenesis and develop more effective therapeutic interventions, the mechanistic characterization of signaling molecules involved in these “alternative” SAP-dependent signaling pathways is imperative.

We recently observed that after TCR stimulation, SAP selectively inhibits diacylglycerol kinase α (DGKα) without requiring FynT or Lck (15). DGKα and DGKζ phosphorylate diacylglycerol (DAG) to generate phosphatidic acid, thereby modulating TCR signal strength by regulating DAG levels and downstream biochemical events (16, 17). In activated T cells, silencing SAP expression results in persistently active DGKα and thus impaired DAG signaling, leading to reduced protein kinase Cθ (PKCθ) membrane recruitment, NFAT (nuclear factor of activated T cells) and ERK1/2 (extracellular signal–regulated kinase 1/2) activation, and IL-2 production (15). These data collectively suggest that upon antigen stimulation, SAP inhibits DGKα activity to facilitate optimal DAG accumulation and full TCR signal strength, ultimately leading to cell activation.

Because TCR signal strength directly correlates with RICD sensitivity (18), we hypothesized that the reduced RICD of XLP-1 T cells might be linked to deregulation of DGKα in the absence of functional SAP. Consistent with this notion, we show herein that the loss of SAP in T cells results in reduced DAG polarization to the immune synapse (IS) and impaired TCR-induced DAG-dependent TCR signaling. Both of these events are due to persistent DGKα activity and contribute to RICD resistance. Consequently, the inhibition of DGKα in XLP-1 T cells restored DAG signaling and RICD by rescuing IS architecture and triggering a specific DAG-dependent apoptotic process mediated by the orphan nuclear receptors NR4A1 (NUR77) and NR4A3 (NOR1). Strikingly, in vivo inhibition of DGKα activity reduced the excessive CD8+ T cell accumulation and interferon-γ (IFNγ) production that occur in Sh2d1a−⁄− mice infected with lymphocytic choriomeningitis virus (LCMV), a murine model of FIM. Our findings illuminate the SAP/DGKα signaling axis as a key regulator of TCR-induced apoptosis. These results highlight DGKα as a novel, druggable target for treating FIM by promoting RICD, reducing the accumulation of pathogenic, activated CD8+ T cells, and thus mitigating the life-threatening immunopathology that often occurs in EBV-infected XLP-1 patients.

RESULTS

DGKα inhibition rescues RICD in SAP-deficient T cells

To investigate whether reduced DAG signaling contributes to the T cell–driven pathologic manifestations of XLP-1, we examined whether silencing or inhibition of DGKα could restore the sensitivity of XLP-1 T cells to RICD. SAP-deficient XLP-1 T cells exhibit reduced RICD relative to control T cells after stimulation with increasing concentrations of the agonistic anti-CD3 antibody (Ab) OKT3 (5) (Fig. 1, A and B). Remarkably, this defect in RICD was substantially rescued by the small interfering RNA (siRNA)–mediated silencing of DGKα (Fig. 1, A to C) or by pretreatment with the DGKα inhibitor R59949 (Fig. 1, D and E) or R59022 (Fig. 1F) (19). The rescue in RICD obtained upon DGKα inhibition was likely due to the induction of apoptosis, as indicated by an increased percentage of AnnexinV+ cells (Fig. 1G). Conversely, the inhibition or silencing of DGKα had little effect on RICD in activated T cells from healthy subjects (Figs. 1 and 2).

Fig. 1. DGKα silencing or inhibition restores RICD in XLP-1 patient T cells.

(A and B) Activated T cells from normal donors (Ctrl) or indicated XLP-1 patients were transfected with control (Cntrl) or DGKα-specific siRNA, and then restimulated 4 days later with OKT3 Ab. After 24 hours, % cell loss was evaluated by propidium iodide (PI) staining. Data are means ± SD of two experiments (A) or one experiment (B) performed in triplicate, representative of two independent experiments using different control donors. (C) DGKα relative expression (rel exp) in siRNA-transfected cells from (A) measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR) (upper panel, mean ± SEM, n = 4) or by Western blotting, with tubulin as loading control (lower panel). (D to F) Ctrl or XLP patient T cells were restimulated with OKT3 Ab after pretreatment with DGK inhibitor R59949 or R59022 (5 to 10 μM) or dimethyl sulfoxide (DMSO). After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SD of three experiments (E) or one experiment (D and F) performed in triplicate representative of two independent experiments using different control donors. (G) Cells used in (D) were pretreated with R59949 (10 μM) or DMSO and restimulated with OKT3 (100 ng/ml) for 0, 6, and 12 hours. The % of apoptotic cells was measured by AnnexinV staining. Representative histograms are shown; marker numbers denote % AnnexinV+ cells. The net increase in AnnexinV+ cells at 12 hours is shown at the right. Data are means ± SD of six independent experiments using four separate controls and two XLP patients. Asterisks denote statistical significance by two-way analysis of variance (ANOVA) with Sidak correction (A, B, D, and F) or paired t test (C and G).

Fig. 2. DGKα silencing or inhibition restores RICD in SAP-silenced T cells.

(A) Activated normal donor T cells were transfected with control or SAP siRNA and restimulated 4 days later with OKT3 Ab. After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SEM of three experiments performed in triplicate. (B) SAP expression in siRNA-transfected T cells from (A) was measured by qRT-PCR (upper panel, mean ± SEM of four experiments) or by Western blotting, with actin as a loading control (lower panel). (C and D) siRNA-transfected cells (A) were restimulated with OKT3 Ab after pretreatment with DMSO and DGK inhibitor R59949 or R59022 (5 to 10 μM). After 24 hours, the % cell loss was evaluated by PI staining. Data are means ± SEM of five experiments (C) or five (control) and eight (SAP siRNA) independent experiments (D) performed in triplicate. (E) siRNA-transfected cells as in (A) were pretreated with DMSO or R59022 (10 μM) and restimulated with OKT3 (10 ng/ml). After 12 hours, the % apoptotic cells was evaluated by AnnexinV staining. Representative histograms are shown; marker numbers denote % AnnexinV+ cells. The net increase in AnnexinV+ cells at 12 hours is shown at the right. Data are means ± SD of four experiments. (F) siRNA-transfected cells (A) were treated with C8-DAG (50 μM) and restimulated with OKT3 Ab. After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SEM of five experiments performed in triplicate. Asterisks denote statistical significance by two-way ANOVA with Sidak correction (A and C to F) or paired t test (B and E). (G) Schematic cartoon: Proapoptotic TCR signaling is governed by DGKα inhibition in activated T cells.

Because patient-derived cells were limited, we repeated these assays using siRNA to knock down SAP expression in activated T cells from healthy donors (Fig. 2) (5). In agreement with our previous findings, SAP-silenced cells exhibited defective RICD that was rescued by concomitant silencing of DGKα (Fig. 2, A and B) or by treatment with the DGKα inhibitor R59949 (Fig. 2C) or R59022 (Fig. 2D). This restoration of RICD in SAP-silenced T cells was associated with enhanced apoptosis, as indicated by increased AnnexinV staining (Fig. 2E). For other isoforms expressed in T cells, silencing of DGKζ, but not DGKδ, also partially rescued RICD in SAP-silenced cells (fig. S1, A to D). Conversely, overexpression of DGKα or DGKζ conferred partial resistance to RICD in normal T cells (fig. S1, E and F). These findings suggest a link between the RICD resistance of SAP-deficient lymphocytes and unrestrained DAG depletion caused by enhanced DGK activity. To explore this further, we supplemented cultures with the DAG analog 1,2-dioctanoyl-sn-glycerol (C8-DAG), which is rapidly incorporated into the cell membrane and triggers DAG-dependent signaling (20). Indeed, C8-DAG treatment markedly enhanced RICD in SAP-silenced but not control T cells (Fig. 2F). Collectively, these data demonstrate that excessive DGKα activity contributes to RICD resistance in SAP-deficient T cells and that this process can be reversed by inhibition of DGKα. These data suggest that SAP promotes TCR signal strength and RICD sensitivity by attenuating DAG metabolism carried out by DGKα in activated T cells (Fig. 2G).

Inhibition of DGKα rescues defective DAG polarization and signaling at the IS in SAP-deficient cells

DAG generation and polarization at the IS are required for TCR-induced cellular responses (21). To investigate whether the deregulated DGKα activity caused by SAP deficiency affects DAG polarization toward the IS, we imaged DAG localization using a PKCθ–cysteine-rich domain (CRD)–based biosensor (22). After activation by superantigen-loaded Raji B cells, we observed that PKCθ-CRD polarization to the IS was strongly reduced in SAP-silenced versus control Jurkat cells (Fig. 3, A to D). In contrast, co-silencing of DGKα (Fig. 3, A and B) or pretreatment with the DGK inhibitor R59949 (Fig. 3, C and D) restored PKCθ-CRD polarization in SAP-silenced T cells. Consistent with the finding that DGKα shapes the DAG gradient at the IS (23), DAG polarization was also reduced in SAP-expressing, DGKα-silenced Jurkat cells (Fig. 3, A and B).

Fig. 3. DGKα silencing or inhibition restores synapse formation in SAP-deficient T cells.

ShCNTRL or shSAP Jurkat T cells were transiently transfected as indicated. (A to J) After 48 hours (C, D, and G to J) or 96 hours (A, B, E, and F), T cells were challenged with SEE-loaded Raji B cells, and confocal live-cell images were captured during T cell–antigen presenting cell (APC) conjugation. In (C), (D), (G), (H), (I), and (J), T cells were pretreated for 30 min with 10 μM R59949 or DMSO. (A and C) Top row: Enhanced green fluorescent protein (EGFP)–tagged PKCθ-CRD (pseudocolor) together with the perimeter of the APC (dotted line). Bottom row: Phase-contrast images with APC denoted by *. Scale bar, 10 μm. (B and D) Quantification of EGFP-PKCθ-CRD accumulation at the IS. Mean ± SEM of >20 conjugates per condition from three experiments. (E and G) Top row: LifeAct–GFP (green fluorescent protein) (green). Bottom row also shows Raji B cells stained with CellTracker Red CMTPX (red). Scale bar, 10 μm. (F and H) Quantification of LifeAct-GFP accumulation at the IS. Mean ± SEM of >20 conjugates per condition from three experiments. (I) Top row: GFP-tubulin (green). Bottom row also shows Raji B cells stained with CellTracker Red CMTPX (red). Scale bar, 10 μm. (J) Quantification of MTOC polarization index. Mean ± SEM of >35 conjugates per condition from two experiments. Asterisks in all panels denote statistical significance by one-way ANOVA with Sidak correction.

Polarized DAG signaling triggers F-actin polymerization and microtubule organizing center (MTOC) orientation (24, 25). Consistent with reduced DAG polarization, SAP-silenced T cells exhibited a strong defect in F-actin accumulation at, and MTOC orientation toward, the IS upon contact with superantigen-loaded Raji cells (Fig. 3, E to J). Again, silencing or inhibition of DGKα partially restored these processes (Fig. 3, E to J). These findings indicate that SAP regulates the architecture of the IS by inhibiting DGKα, thereby limiting DAG metabolism locally.

We next investigated whether inhibition of DGKα restores DAG-mediated signaling downstream of the TCR in SAP-silenced primary human T cells. PKCθ and RasGRP1 are recruited to the IS in a DAG-dependent manner (26, 27) and are required for induction of RICD (28, 29). Consistent with our hypothesis, SAP-silenced primary T cells exhibited defective recruitment of PKCθ and RasGRP1 to the IS, which was fully restored upon DGKα silencing (Fig. 4, A, B, E, and F) or pharmacological inhibition (Fig. 4, C, D, G, and H). Considering inhibition of DGKα also rescues defective ERK1/2 activation in SAP-deficient T cells (15), these data underscore the importance of the SAP/DGKα axis in regulating DAG-dependent signaling.

Fig. 4. DGKα silencing or inhibition restores PKCθ and RasGRP1 recruitment to the IS in SAP-deficient cells.

(A, C, E, and G) Activated T cells were transfected with the indicated siRNA and after 72 hours were incubated with SEE-loaded Raji B cells (denoted with *) for 15 min and fixed and stained for PKCθ (A and C) or RasGRP1 (E and G). Top rows: Target protein (green); bottom rows also show phase contrast. Scale bar, 5 μm. (B) Percentage of cells displaying PKCθ at the IS. Data are means ± SEM of six replicates from two independent experiments. (D) Percentage of cells displaying PKCθ at the IS. Data are means ± SEM of three experiments. (F) Percentage of cells with RasGRP1 at the IS. Data are means ± SD of one representative experiment performed in quadruplicate. (H) Percentage of cells displaying RasGRP1 at the IS. Data are means ± SEM of three experiments. Asterisks in all panels denote statistical significance by two-way ANOVA + Sidak correction.

To test if inhibition of DGKα rescues RICD in SAP-deficient T cells by restoring specific DAG-mediated signaling pathways, we examined whether the rescue of RICD requires the activity of PKCθ or RasGRP1. Silencing of PKCθ (Fig. 5A) or RasGRP1 (Fig. 5B) reduced RICD in control siRNA–transfected T cells and completely abrogated the rescue of RICD in SAP and DGKα siRNA–transfected cells. Moreover, pharmacological inhibition of PKC or MEK (mitogen-activated protein kinase)/ERK enzymatic activity also prevented the restoration of RICD noted after DGKα silencing in SAP-deficient T cells (fig. S2).

Fig. 5. DGKα silencing restores TCR-induced PKCθ and Ras-mediated signaling pathways to drive RICD in SAP-deficient cells.

(A and B) Activated normal donor T cells were transfected with the indicated siRNA and restimulated 4 days later with OKT3 Ab (10 ng/ml). After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SEM of seven (A) or six (B) experiments performed in triplicate. Right panels: Expression of PKCθ (A) or RasGRP1 (B) was measured by Western blotting, with actin as a loading control. (C and D) qRT-PCR for IL2 mRNA in T cells pretreated with R59949 (10 μM) (C) or transfected with DGKα siRNA (D) after restimulation with OKT3 (10 μg/ml, 4 hours). GUSB served as the reference gene. Graphs represent mean ± SEM of six (C) or seven (D) experiments. (E) Left: Representative flow cytometric histograms showing CD25 surface expression on siRNA-transfected T cells from (A) ±OKT3 restimulation (24 hours). Right: Graph depicts mean fluorescence intensity (MFI) of CD25 expression. Data are means ± SEM of four experiments. Asterisks in all panels denote statistical significance by two-way ANOVA with Sidak correction. IgG, immunoglobulin G. (F) Schematic cartoon: SAP-mediated inhibition of DGKα activity ensures a sufficient pool of DAG required for proper IS organization and recruitment of PKCθ and RasGRP, which mediates downstream signaling for RICD.

TCR activation stimulates DAG-dependent induction of IL-2 and the high-affinity IL-2 receptor CD25 (30, 31), which are both required for RICD (32). Indeed, inhibition or silencing of DGKα restored induction of CD25 in SAP-silenced T cells after TCR restimulation, and a similar trend was observed with IL-2 expression (Fig. 5, C to E). These findings further establish the SAP/DGKα axis as a critical regulator of DAG signaling potency. Collectively, these findings underscore the vital role of SAP-dependent inhibition of DGKα in sustaining DAG signaling, leading to the activation of PKCθ and Ras-ERK and RICD (Fig. 5F).

NUR77 and NOR1 mediate the rescue of RICD that is induced by DGKα inhibition in SAP-deficient T cells

We next investigated the mechanism by which the enhancement of DAG signaling obtained after inhibition of DGKα restores RICD sensitivity in SAP-deficient T cells. We previously showed that in SAP-deficient T cells, TCR-induced expression of key proapoptotic genes such as FASLG and BCL2L11 is impaired (5). We observed that silencing or inhibition of DGKα failed to rescue FASLG or BCL2L11 expression after TCR restimulation of SAP-silenced T cells (fig. S3, A and B). Similarly, DGKα blockade failed to restore the induction of all three major isoforms of BIM protein (extra long, EL; long, L; and short, S), as well as full-length and soluble FASL protein in SAP-silenced and XLP-1 patient T cells after restimulation (fig. S3, C to E). These observations imply that DGKα inhibition does not restore all SAP-dependent, proapoptotic effector functions that contribute to RICD sensitivity.

Instead, we found that SAP-deficient T cells exhibit a previously unrecognized defect in TCR restimulation–induced up-regulation of NR4A1 (NUR77) and NR4A3 (NOR1), two nuclear receptors involved in negative selection of thymocytes and RICD of mature T cells (33). DGKα silencing or inhibition selectively restored TCR-dependent induction of both NR4A1 and NR4A3 in SAP-silenced activated T cells (Fig. 6, A to D). DGKα inhibition also partially rescued NUR77 and NOR1 protein induction in XLP-1 T cells after TCR restimulation (Fig. 6E). Upon TCR engagement, NUR77 and NOR1 proteins are phosphorylated by the ERK1/2-regulated 90-kD ribosomal S6 kinase (RSK), triggering the intrinsic apoptosis pathway (34). Indeed, the RSK-specific inhibitor SL0101 (35) significantly reduced RICD in control T cells, confirming that phosphorylation of NUR77 and NOR1 is an important component of RICD execution (Fig. 6, F to H). SL0101 significantly blunted the RICD rescue triggered by DGKα inhibition in XLP-1 T cells, as well as in SAP/DGKα-silenced T cells (Fig. 6, F to H). These data indicate that the rescue of RICD afforded by DGKα blockade in SAP-deficient T cells is dependent on RSK activity. Moreover, concomitant knockdown of NUR77 and NOR1 reduced the rescue of RICD induced by DGKα inhibition in XLP-1 T cells (Fig. 6, I to K). Together, these observations indicate that inhibition of DGKα boosts RICD in SAP-deficient T cells in part by selectively restoring TCR-induced up-regulation and RSK-dependent phosphorylation of NUR77 and NOR1 (Fig. 6L).

Fig. 6. Silencing or inhibition of DGKα restores RICD sensitivity in SAP-deficient T cells via induction of proapoptotic molecules NUR77 and NOR1.

(A to D) qRT-PCR for NR4A1 (A and B) or NR4A3 (C and D) from activated normal donor T cells transfected with control or SAP-specific siRNA ± DGKα-specific siRNA (A and C) or 5 μM R59949 (B and D) and restimulated with OKT3 (10 μg/ml) for 4 hours. GUSB served as the reference gene. Data are means ± SEM of eight (A), five (B), seven (C), or six (D) experiments. (E) Activated T cells from normal donor (Ctrl) or XLP-1 patient 7 were pretreated for 30 min with R59022 or R59949 (10 μM), then restimulated with OKT3 (100 ng/ml). Cell lysates were analyzed by Western blotting for NUR77, NOR1, and β-actin content. Data are representative of two independent experiments using different donors. (F and G) Activated T cells from normal donors (Ctrl) or XLP-1 patients were pretreated for 30 min with DMSO, SL0101 (90 μM), R59949 (10 μM), or both, followed by restimulation with OKT3 (100 ng/ml). After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SD of one experiment each performed in triplicate using different donors. (H) Activated donor T cells were transfected with the indicated siRNA and treated 4 days later with SL0101 (50 μM) for 30 min, followed by OKT3 (10 ng/ml). After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SEM of five experiments performed in triplicate. (I and J) Activated T cells from normal donors (Ctrl) or XLP-1 patients were transfected control or NUR77 + NOR1 siRNA and treated 4 days later with DMSO or R59022 (10 μM) for 30 min, followed by OKT3 (100 ng/ml). After 24 hours, % cell loss was evaluated by PI staining. Data are means ± SD of one experiment each, performed in triplicate using different donors. Asterisks in all panels denote statistical significance by two-way ANOVA with Sidak correction. (K) Western blot for NUR77 and NOR1 expression in OKT3-restimulated, siRNA-transfected T cells from (I). Actin served as a loading control. (L) Schematic cartoon: Mechanism of proapoptotic TCR signaling governed by SAP-dependent DGKα inhibition in activated T cells. n.s., not significant.

DGKα inhibition reduces CD8+ T cell accumulation and activation in LCMV-infected Sh2d1a−⁄− mice

Defective RICD is thought to contribute to the aberrant T cell activation and accumulation that occur in EBV-infected XLP-1 patients (36). The demonstration that DGKα silencing or inhibition sensitizes SAP-deficient lymphocytes to RICD in vitro prompted us to assess whether DGKα inhibition might influence T cell–mediated immunopathology in vivo. Toward this end, we used a murine model in which Sh2d1a−⁄− mice are infected with LCMV. In this model, Sh2d1a−⁄− mice develop many of the cardinal manifestations of FIM, including CD8+ T cell expansion, proinflammatory cytokine production, and tissue infiltration (37, 38). For these experiments, wild-type (Sh2d1a+/+) or Sh2d1a−⁄− mice were infected with LCMV Armstrong and 4 days later were treated with vehicle or R59022 (39). On day 8, the peak of the antiviral T cell response, mice were euthanized and evaluated for hyperinflammation.

After LCMV infection, both wild-type and Sh2d1a−⁄− mice developed marked and comparable splenomegaly (Fig. 7, A and B) that was associated with an increase in the absolute number of total splenocytes (Fig. 7C). Examination of splenocyte immunophenotype revealed a significant increase in the percentage and absolute number of total as well as LCMV-specific (gp33+) CD8+ T cells, most of which exhibited an activated CD44+ phenotype (Fig. 7, D to I). Treatment of LCMV-infected wild-type mice with R59022 did not significantly affect any of these parameters (Fig. 7). Conversely, R59022 treatment of LCMV-infected Sh2d1a−⁄− mice appeared to lessen organomegaly (Fig. 7, A and B) and decrease the total splenic lymphocyte count (Fig. 7C). Although R59022 treatment did not affect the percentage of activated splenic CD8+ T cells in either mouse strain, it did induce a significant decrease in the number of total as well as LCMV-specific CD8+ T cells selectively in the Sh2d1a−⁄− animals (Fig. 7, D to I).

Fig. 7. In vivo DGKα inhibition reduces the number of activated virus-specific CD8+ T cells in LCMV-infected Sh2d1a−⁄− mice.

(A) Images of spleens from uninfected [phosphate-buffered saline (PBS), “P”] and LCMV-infected mice without (LCMV, “L”) or with R59022 treatment (LCMV + R59022, “L + R”). Representative spleens from each cohort from B6 wild-type (WT) (top panel) and Sh2d1a−⁄− mice (lower panel) are shown. (B and C) Ratio of spleen over body weight (B) and total splenocyte count (C) for animals in each group are presented. B6 WT mice, red symbols; Sh2d1a−⁄− mice,  blue symbols. (D to I) Representative flow cytometric (density) plots showing the percentages of CD8+ CD44+ (top) and LCMV-specific CD8+ gp33+ (bottom) in the spleens (D) and livers (G) of B6 WT and Sh2d1a−⁄− mice. Percentages (E and H) and absolute numbers (F and I) of CD8+ CD44+ and CD8+ CD44+ gp33+ cells in the spleens (E and F) and livers (H and I) of B6 WT (red symbols) and Sh2d1a−⁄− (blue symbols) mice. Data are from one of two experiments in which a total of 6 to 10 mice in each cohort were examined. Error bars represent SD. Asterisks denote statistical significance that was determined by two-way ANOVA with Sidak correction.

Compared to wild-type animals, LCMV-infected Sh2d1a−⁄− mice also exhibited a trend toward higher serum IFNγ levels (Fig. 8A) and greater degrees of tissue inflammation (Fig. 8, B to D). To evaluate whether DGKα inhibition affected CD8+ T cell functions such as cytokine production or degranulation, splenocytes from LCMV-infected or uninfected, R59022 treated or untreated mice were cultured directly ex vivo with the major histocompatibility complex (MHC) class I–restricted LCMV peptide gp33 and examined for expression of intracellular TNFα (tumor necrosis factor–α), IFNγ, and surface CD107a. R59022 treatment did not affect the percentage of CD8+ T cells that secreted cytokines (TNFα or IFNγ) or degranulated (CD107α exposure) in LCMV-infected wild-type or Sh2d1a−⁄− mice (Fig. 8, E and F). R59022 treatment actually enhanced the cytolytic activity of Sh2d1a−⁄− CD8+ T cells against autologous B cell targets in vitro (fig. S5). However, such treatment did reduce the absolute number of cytokine-producing and degranulating cells only in the Sh2d1a−⁄− animals (Fig. 8G). Consistent with these findings, only R59022-treated Sh2d1a−⁄− mice exhibited a significant reduction in the serum IFNγ levels (Fig. 8A). Finally, R59022 treatment significantly reduced the number and size of hepatic inflammatory infiltrates in LCMV-infected Sh2d1a−⁄− but not wild-type mice (Fig. 8, B to D). These findings collectively indicate that inhibition of DGKα selectively decreases the magnitude of the CD8+ T cell effector pool in LCMV-infected Sh2d1a−⁄− mice. DGKα inhibition had no adverse effects on viral clearance, as LCMV was efficiently cleared from wild-type and Sh2d1a−⁄− mice by day 8 with or without R59022 treatment (fig. S6). These preclinical data suggest that pharmacologic inhibition of DGKα might reduce the accumulation of aberrantly activated CD8+ T cells and subsequent hypercytokinemia and tissue inflammation that occur in EBV-infected XLP-1 patients, without impairing CD8+ T cell activity or viral clearance.

Fig. 8. In vivo DGKα inhibition reduces the number but not incidence of virus-specific CD8+ cytokine-producing cells of LCMV infected Sh2d1a−⁄− mice.

B6 WT (red symbols) and Sh2d1a−⁄− mice (blue symbols) were either uninfected (PBS, “P”) or infected with LCMV without (LCMV, “L”) or with R59022 treatment (LCMV + R59022, “L + R”). (A) Serum IFNγ levels were assayed on day 8 after infection by enzyme-linked immunosorbent assay (ELISA). Data are compiled from two experiments in which a total of 6 to 10 mice in each cohort were examined. Error bars represent SEM. (B to D) Hematoxylin and eosin–stained liver sections from mice in each group were analyzed for the number of inflammatory foci (B) and area of the inflammatory infiltrate (C). For each sample, five random fields were captured at ×20 magnification and scored. Histology of the livers from representative mice in each group under ×20 magnification (top row) is shown (D). Arrows point to the inflammatory foci. Micrographs in the bottom row are the respective computer analyzed images shown in the top row. (E to G) Splenocytes (2 × 106) from PBS (P), LCMV-infected (L), and LCMV-infected mice with R59022 treatment (L + R) groups were left unstimulated or stimulated with gp33 peptide (0.4 ng/ml) in the presence of monensin (1000 μg/ml) for 5 hours. Cells were then analyzed for intracellular cytokine production and degranulation. Representative flow cytometric (density) plots gated on CD8+ CD44+ splenocytes showing the percentages of CD8+ IFNγ+ (top), IFNγ+ TNFα+ (middle), and IFNγ+ CD107a+ (bottom) cells from B6 WT and Sh2d1a−⁄− mice (E). Percentages (F) and absolute numbers (G) of CD8+ IFNγ+, IFNγ+ TNFα+, and IFNγ+ CD107a+ cells. Absolute numbers were calculated by multiplying the percentages with the respective absolute numbers of CD8+ gp33+ cells. Error bars in (B), (C), (F), and (G) represent SD. Asterisks denote statistical significance that was determined by two-way ANOVA with Sidak correction.

One major limitation of this study is its focus on the role of the SAP/DGKα axis strictly in T cells. It does not address the putative role of DGKα in other SAP-deficient immune cells such as NK or NKT cells, which also likely contribute to the development of various XLP-1 manifestations. In addition, although LCMV infection of Sh2d1a−⁄− mice is a widely used murine model of FIM, it does not fully recapitulate the pathogenesis of EBV infection in humans. Finally, the translation of these findings to the clinic will require the development and characterization of novel, clinical-grade DGKα-specific inhibitors. Nonetheless, our data clearly provide proof of concept that DGKα may be a novel drug target for treating XLP-1–associated FIM.

DISCUSSION

Our results demonstrate that inhibition of DGKα restores sensitivity to RICD in SAP-deficient T cells and reduces hyperinflammation in LCMV-infected Sh2d1a−⁄− mice. These data support the hypothesis that in SAP-deficient T cells, persistent DGKα activity increases DAG metabolism at the IS, thus reducing DAG signaling and RICD sensitivity, and underscore the role of SAP in modulating DGKα activity (15). In SAP-expressing T cells, further inhibition of DGKα only marginally influenced sensitivity to RICD in vitro and did not dampen the LCMV-induced CD8+ T cell response of wild-type mice in vivo.

Our finding that inhibition of DGKα restores proper IS organization in SAP-deficient T cells indicates that SAP, through regulation of DGKα, controls DAG polarization at the IS, thus promoting F-actin polymerization and MTOC orientation. Indeed, DGKα is recruited to the peripheral SMAC (pSMAC), where it shapes the DAG gradient responsible for the recruitment of novel PKC isoforms (ε, η, and θ), which control MTOC polarization and actin polymerization (Fig. 3) (23, 25). DGKα inhibition also partially rescued impaired cytolytic activity in SAP-deficient cytotoxic T lymphocytes (CTLs) (fig. S5), further highlighting the link between SAP-dependent inhibition of DGKα and IS function (4). Together, these data indicate that SAP-mediated negative regulation of DGKα controls IS structural organization and function by regulating the TCR-induced gradient of DAG.

Both DGKα and DGKζ regulate TCR-induced DAG signaling (16, 17). Consistently, inhibition or overexpression of either isoform rendered T cells more or less sensitive to RICD, respectively, confirming that both isoforms regulate DAG signalling in this context (fig. S1). Accordingly, administration of exogenous DAG partially rescued the defective RICD in SAP-deficient T cells (Fig. 2F). However, only DGKα is regulated by SAP and shapes the DAG gradient at the IS (15, 23). We speculate that DGKα, which colocalizes with F-actin at the pSMAC, regulates the DAG gradient and F-actin polymerization at the IS, whereas DGKζ, which is more evenly distributed in the IS, metabolizes most of the DAG generated there, consistent with proposed models (23, 40). Notably, silencing of DGKδ, which is highly expressed in T cells, did not affect RICD, underscoring the specific roles of DGKα and DGKζ in regulating the DAG pool relevant for signaling and RICD onset.

Our finding that inhibition of DGKα restored RasGRP1 and PKCθ recruitment to the IS in SAP-deficient T cells, and subsequent DAG-dependent induction of IL-2 and CD25, illuminates a biochemical link between the SAP/DGKα axis, IS restoration, and downstream signaling events required for RICD (2831). Indeed, activation of RasGRP1 and PKCθ was required to rescue RICD in SAP-deficient T cells upon DGKα blockade (Fig. 5, A and B, and fig. S2). DGKα inhibition cannot recapitulate all SAP-dependent signaling functions, such as TCR-induced expression of FASLG or BCL2L11, genes previously implicated in SAP-associated induction of RICD (5). This observation suggests that rescue of DAG-mediated signaling activates other proapoptotic pathways that partially compensate to boost RICD sensitivity. Here, we show that the induction of NUR77 and NOR1 is defective in TCR-restimulated SAP-deficient T cells, and that the expression of these genes is restored by inhibition of DGKα. NUR77 and NOR1 are orphan nuclear receptors known to trigger thymocyte apoptosis during negative selection and mediate RICD (33, 41). The proapoptotic activity of NUR77 and NOR1 depends on their phosphorylation by RSK, an ERK1/2-dependent kinase, and subsequent translocation to the mitochondria to promote mitochondrial depolarization and apoptosis (33, 34). We observed that ERK and RSK activity, as well as NUR77 and NOR1, were required for RICD rescue triggered by DGKα inhibition in XLP-1 T cells. These observations provide a mechanistic connection between the rescue of DAG signaling and the execution of RICD.

EBV-induced FIM is proposed to result from defective RICD of CD8+ T cells and impaired cytotoxic elimination of EBV-infected B cells by CD8+ T cells and NK cells. These events contribute to the excessive accumulation of activated effector CD8+ T cells and life-threatening damage to the liver, the bone marrow, and other organs (2, 5). Using a murine model of FIM (37, 38), we showed that DGKα inhibition had no significant effect on reducing the activation status or function (for example, cytokine secretion or degranulation) of effector CD8+ T cells in either wild-type or Sh2d1a−⁄− mice. These data suggest that treatment with R59022 after LCMV infection does not impair initial lymphocyte activation. However, this treatment did significantly decrease the absolute number of activated CD8+ T cells in Sh2d1a−⁄− mice, leading to fewer and smaller lymphocytic infiltrates within the liver and marked reductions in the level of IFNγ in the serum. These results suggest that RICD resistance is connected to aberrant DGKα activity and serves as a key driver of virus-induced immunopathology in Sh2d1a−⁄− mice. Remarkably, the apoptosis resistance of activated T cells in LCMV-infected Sh2d1a−⁄− mice can be overcome via DGKα inhibition, even when such inhibition is initiated well after infection is established. These data are relevant to the clinical setting, where patients often present with FIM days to weeks after primary EBV infection.

In conclusion, our findings underscore the importance of SAP-mediated DGKα inhibition in maintaining lymphocyte homeostasis by ensuring sufficient TCR-induced DAG signaling strength for apoptosis. These data provide proof of principle that treatment with a DGKα inhibitor could serve as a novel, reasonable strategy to counteract pathological EBV-driven lymphohistiocytosis that occurs in EBV-infected XLP-1 patients by restoring the RICD sensitivity of activated CD8+ T cells.

MATERIALS AND METHODS

Study design

This was a preclinical study to (i) determine if DGKα inhibition could rescue RICD in SAP-deficient T cells and (ii) assess the efficacy of a DGKα inhibitor in attenuating CD8+ T cell lymphocytosis and immunopathology in LCMV-infected SAP-deficient mice, a model of FIM. Although in vitro experiments utilizing XLP-1 patient T cells were often constrained by limited sample availability, each RICD experiment was performed with at least two separate XLP patients and different control donors (for example, Fig. 1, A and B). We also generated robust corroborating data using siRNA-mediated SAP knockdown in T cells from multiple human donors (n ≥ 3 experiments each). Once we established that DGKα blockade restored RICD sensitivity in SAP-deficient T cells, we focused on delineating the biochemical mechanism that explains this phenomenon. For all in vitro data, the number of experiments (including technical replicates) is defined in each figure legend. For in vivo experiments, numbers of mice are outlined in each figure legend. All statistical analyses described below were verified by consultation with an experienced biostatistician [C. Olsen, Uniformed Services University of the Health Sciences (USUHS)].

Cell culture

Peripheral blood mononuclear cells (PBMCs) were isolated from normal controls or XLP-1 patients by Ficoll-Paque PLUS (GE Healthcare) density gradient centrifugation, washed, and resuspended at 2 × 106 cell/ml in complete media (cRPMI): RPMI-GlutaMAX (Life Technologies) containing 10% heat-inactivated fetal calf serum (FCS) (Lonza), 2 mM glutamine, and penicillin and streptomycin (100 U/ml)(Life Technologies). T cells were activated with anti-CD3 (1 μg/ml) (clone UCHT1) and anti-CD28 (clone CD28.2) antibodies. After 3 days, activated T cells were washed and cultured in cRPMI plus recombinant human IL-2 (rhIL-2) (100 IU/ml) (PeproTech) at 1.2 × 106 cells/ml for ≥7 days before apoptosis assays were conducted (media changed every 2 to 3 days).

Jurkat A3 cells were from American Type Culture Collection, and 293FT from Life Technologies. Cells were cultured in RPMI or Dulbecco’s modified Eagle’s medium (Life Technologies) with 10% FCS and antibiotics/antimycotics (Sigma-Aldrich).

DGK inhibitors R59949 and R59022 (Sigma-Aldrich) were dissolved in DMSO. All reagents and antibodies used are listed in table S1.

siRNA transfections

PBMCs were transfected with 200 pmol of Stealth Select siRNA or Stealth RNAi Negative Control Duplexes (Life Technologies). siRNA sequences are listed in table S2. Transient transfections were performed using Amaxa Nucleofector Kits for human T cells (Lonza) and the Amaxa Nucleofector II or 4D Systems (program T-20 or EI-115). Cells were cultured in IL-2 (100 IU/ml) for 4 days to allow target gene knockdown. Knockdown efficiency was periodically evaluated by RT-PCR and Western blotting.

Conjugation and live-cell imaging of Jurkat T cells

Jurkat T cells and Raji B cells were labeled, imaged, and analyzed as described in the Supplementary Methods.

Immunofluorescence experiments with primary T cells

Human T cells were stimulated with soluble anti-CD3 and anti-CD28 (1 μg/ml) for at least 7 days and transfected with Amaxa Nucleofector Kit for human T cells (Lonza) with control, SAP-specific, and/or DGKα-specific siRNA. After 72 hours, T cells were incubated with Raji B cells loaded with mixed SEE (staphylococcal enterotoxin E) and SEB (staphylococcal enterotoxin B) superantigens (1 μg/ml) for 15 min, fixed, and stained for either PKCθ or RasGRP1. For some experiments, transfected T cells were pretreated with R59949 (10 μM, 30 min, 37°C) or DMSO before conjugation.

Cytofluorimetry

To examine RICD, activated T cells (105 cells per well) were plated in triplicate in 96-well round-bottom plates and treated with anti-CD3ε mAb (monoclonal antibody) OKT3 (1 to 100 ng/ml) in cRPMI + rhIL-2 (100 IU/ml) for 24 hours. R59949 (5 to 10 μM), R59022 (5 to 10 μM), DAG (50 μM), U0126 (5 μM), FR180204 (10 μM), or Rottlerin (6 μM) inhibitors were added 30 min before restimulation. At 24 hours after restimulation, cells were stained with PI (1 μg/ml) and collected for 30 s per sample on FACScan or Accuri C6 flow cytometers (BD). Cell death was analyzed with CellQuest/CFlow software (BD) or Flowing software (Turku Bioimaging) as percentage of cell loss = (1 − [number of viable cells (treated)/number of viable cells (untreated)]) × 100 (5).

For AnnexinV assays, ~1 × 106 cells were treated with OKT3 (10 ng/ml) as above. Cells were stained 6 to 12 hours later with AnnexinV–phycoerythrin (BioLegend) and analyzed on Accuri C6.

To evaluate CD25 expression, 1 × 106 cells were stimulated with OKT3 (100 ng/ml) for 24 hours and fixed and stained with anti-CD25 plus anti-mouse Alexa Flour 488. Stained cells were collected on a FACSCalibur.

Western blotting

Lymphocytes (1 × 106 to 10 × 106 cells) were stimulated, lysed, and subjected to SDS–polyacrylamide gel electrophoresis and immunoblotting as described (5, 15). Immunoblot images were acquired and quantified using a Versadoc Model 4000 Imaging System (Bio-Rad) or ImageJ software (for film). Spot densitometry analyses are summarized in table S4.

Quantitative RT-PCR

Activated lymphocytes (30 × 106 cell/ml) were stimulated in cRPMI with OKT3 (10 μg/ml) for 4 hours. R59949 (5 μM) was added 30 min before restimulation. Cells were washed with cold PBS, and mRNA was extracted using a ChargeSwitch Total RNA Cell Kit (Life Technologies). RNA was reverse-transcribed using High-Capacity cDNA Reverse Transcription Kits (Life Technologies), and cDNA targets were quantified by RT-PCR (C1000 Thermal Cycler CFX96, Bio-Rad) using TaqMan gene expression assays (see table S3), with GUSB as the housekeeping control (Life Technologies).

Mice and in vivo experiments

Sh2d1a−/− mice were as described (38). C57BL/6 (B6) mice were purchased from Jackson Laboratories. To establish LCMV infection, mice received 2 × 105 plaque-forming units of LCMV Armstrong by intraperitoneal injection on day 0, and experiments were carried out until day +8 after infection. Beginning at day 4, mice were given twice daily intraperitoneal injections of R59022 at a dose of 2 mg/kg body weight, dissolved in DMSO. Mice in all groups were sex- and age-matched. Stimulation of mouse splenocytes in vitro, assessment of liver histology, and quantification of viral titers were performed as described in the Supplementary Methods.

Statistical analysis

Evaluation of in vitro assays across multiple treatments (RICD, RT-PCR, and ELISA), and in vivo experiments was analyzed using two-way ANOVA (α = 0.05) with Sidak’s multiple comparisons correction using GraphPad PRISM software. When comparing two groups (RT-PCR, AnnexinV+ cells), a two-tailed paired Student’s t test was performed in Microsoft Excel. Error bars are described in the figure legends as ± SEM or ± SD where appropriate. Asterisks denote significance in all experiments; P values are included in table S5.

Study approval

Blood samples were obtained with informed consent under protocols approved by the respective Institutional Review Boards (Cincinnati Children’s Hospital Medical Center, National Institute of Allergy and Infectious Diseases, and University of Piemonte Orientale). Experimental procedures on animals were approved by the Institutional Animal Care and Use Committee at The Children’s Hospital of Philadelphia and St. Jude Children’s Research Hospital.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/321/321ra7/DC1

Methods

Fig. S1. Activity of DGKα and DGKζ contributes to RICD resistance in T cells.

Fig. S2. DGKα silencing restores RICD in SAP-deficient cells through PKCθ- and Ras-mediated signaling pathways.

Fig. S3. DGKα blockade fails to rescue TCR-induced up-regulation of proapoptotic mediators FASL and BIM in SAP-deficient cells.

Fig. S4. Major steps of the automated segmentation and fluorescence quantification algorithm.

Fig. S5. DGKα inhibition enhances SAP-deficient CD8+ T cell cytotoxicity against autologous B cell targets.

Fig. S6. R59022 DGKα inhibitor does not impair viral clearance in the livers and spleens of LCMV infected Sh2d1a−⁄− mice.

Fig. S7. Gating strategies used in Figs. 7 and 8.

Table S1. Reagents.

Table S2. siRNA sequences.

Table S3. TaqMan gene expression arrays.

Table S4. Spot densitometry analysis for Western blotting.

Table S5. Statistical analyses.

Unmodified Western blot images.

Source data (excel).

REFERENCES AND NOTES

  1. Acknowledgments: We thank the patients and their families for participating in our study. We also thank biostatistician C. Olsen (USUHS) for consultation on statistical analysis, J. Wherry for providing LCMV and assistance with performing LCMV infections, D. Cantrell for supplying the PKCθ-CRD construct, R. Wedlich-Soldner for supplying the LifeAct-GFP construct, and J. Downward for supplying GFP-tubulin. Funding: K.N. was supported by the XLP Research Trust and the Sean Fischel Fund for HLH (hemophagocytic lymphohistiocytosis) research. A.L.S. and S.E.L. were supported by grants from the NIH (1R01GM105821), XLP Research Trust, and USUHS. A.G., E.R., and V.M. were supported by grants from Telethon (GGP10034 and GGP13254) and AIRC (Associazione Italiana per la Ricerca sul Cancro) (IG13524 and IG5392). G.B. is supported by the University of Piemonte Orientale (Young Investigators). V.M. was supported by a grant from the Compagnia di San Paolo. R.D. was supported by an NIH K22 grant (1 K22 CA188149-01). Author contributions: E.R., V.M., S.E.L., A.L.S., and G.B. designed and carried out in vitro experiments; R.D., K.V., P.T., and K.E.N. designed and carried out in vivo experiments; S.E.L. and A.L.S. designed and carried out experiments on XLP-1 patient lymphocytes; V.M., I.R., and C.B. designed and performed imaging in Jurkat cells; L.P. and C.T.B. performed imaging on primary lymphocytes; C.W. performed imaging on mouse lymphocytes; S.M.K. and P.L.S. performed cytotoxicity assays; G.B., A.G., K.E.N., and A.L.S. designed the project and some of the experiments, supervised research, and wrote the paper. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article