Redundant and specialized roles for diacylglycerol kinases α and ζ in the control of T cell functions

See allHide authors and affiliations

Science Signaling  28 Apr 2015:
Vol. 8, Issue 374, pp. re6
DOI: 10.1126/scisignal.aaa0974


The diacylglycerol kinases (DGKs) attenuate diacylglycerol (DAG)–mediated signals by catalyzing the conversion of DAG to phosphatidic acid. In T lymphocytes, the antigen-stimulated generation of DAG links signal strength to the intensity and duration of signaling by the Ras–extracellular signal–regulated kinase (ERK) and protein kinase C (PKC)–dependent pathways. The generation of DAG at the plasma membrane of T cells lies at the core of the mechanisms that delimit T cell functions. DGKα and DGKζ are the two main isoforms that are found in T cells, and several approaches define their precise contribution to T cell responses. Each of these isoforms has specialized and redundant functions that limit the intensity of DAG-regulated signals downstream of antigenic stimulation. This ability, which in normal T cells contributes to maintaining homeostasis and function, is exploited by tumors to evade immune surveillance. Modification of DGK activity offers new perspectives for the therapeutic manipulation of T cell functions for treatment of autoimmune pathologies, or for overcoming tumor-induced T cell tolerance. Precise knowledge of the mechanisms that sustain DGK isoform–specific regulation in T lymphocytes is indispensable for the development of new tools for pharmacological intervention.

Diacylglycerol Signaling Coordinates Transcriptional Control in Response to Antigen Recognition

T lymphocytes are highly tuned to recognize and eliminate foreign invaders (1). They do so through T cell antigen receptors (TCRs), which recognize peptide antigens bound to major histocompatibility complex (MHC) proteins on the surface of an antigen-presenting cell (APC). Discrimination between self and nonself antigens lies at the core of TCR signaling functions that determine T cell differentiation in the thymus and their activation in the periphery. Stimulation of the TCR organizes a structured signaling platform that is accompanied by extensive membrane remodeling and organelle assembly termed the immunological synapse (2). After antigen recognition, the TCR activates Src family kinases, which promote the recruitment of adaptor molecules, such as SLP-76 (SH2 domain–containing leukocyte protein of 76 kD) and LAT (linker of activated T cells). Phosphorylation of these molecules generates docking sites that facilitate the activation of additional signaling molecules, including phospholipase C-γ1 (PLC-γ1) (3).

After activation, PLC-γ1 hydrolyzes phosphatidylinositol 4,5-bisphosphate (PI45P2) in the plasma membrane to generate inositol trisphosphate (IP3) and diacylglycerol (DAG). Soluble IP3 stimulates the depletion of intracellular Ca2+ stores and the subsequent activation of store-operated Ca2+ entry (4). DAG generated at the cell membrane serves as a platform to stabilize and activate targets that share a cysteine-rich domain known as the protein kinase C type 1 (C1)–binding domain. The numbers and functions of DAG effectors are extensive and include not only members of the classical and novel protein kinase C (PKC) and PKD families but also regulators of the small guanosine triphosphatases (GTPases) of the Ras and Rac families (5). In T lymphocytes, the DAG generated by stimulation of the TCR recruits and activates the Ras guanine nucleotide exchange factor (GEF) RasGRP1 (Ras guanyl nucleotide–releasing protein 1) (6). Activation of the Ras–extracellular signal–regulated kinase (ERK) pathway in T lymphocytes is thus largely linked to DAG generation at the plasma membrane (7). Antigen-stimulated generation of Ca2+ and DAG drives a transcriptional program that elicits the expression of activation-induced genes (8). Increases in the concentration of intracellular Ca2+ activate the phosphatase calcineurin, which dephosphorylates and activates the transcription factor nuclear factor of activated T cells (NFAT) (9). The Ras-ERK pathway controls the nuclear localization of the transcription factor AP-1 (activator protein 1), which, once in the nucleus, forms dimers with NFAT (9). Cooperation between NFAT and AP-1 thus integrates Ca2+- and DAG-regulated signals, leading to the expression of genes encoding activation-induced T cell markers (Fig. 1A).

Fig. 1 Ca2+- and DAG-regulated transcription programs mediate T cell activation.

(A) Antigen-mediated stimulation of the TCR initiates the cascade of signals responsible for the PLC-γ–mediated generation of IP3 and DAG. These two mediators stimulate the Ca2+-NFAT and Ras–ERK–AP-1 pathways, which lead to target gene transcription. (B) Costimulatory receptors, such as CD28, promote activation of the PKCθ–NF-κB pathway, which results in expression of the genes encoding IL-2 and IL-2R. (C) Unbalanced Ca2+- and DAG-regulated signals lead to NFAT-mediated transcription of target genes, which results in anergic responses. MEK, mitogen-activated or extracellular signal–regulated protein kinase kinase; IKK, IκB kinase.

Although it is critical, gene expression by NFAT and AP-1 alone is not sufficient to stimulate the full T cell activation program. Costimulatory molecules such as CD28 act in concert with the TCR after their recognition of B7 ligands presented by the APC. Costimulation does not affect the Ca2+-induced transcription of target genes by NFAT, but it enhances the phosphorylation of inhibitor of κB α (IκBα) and activation of the nuclear factor κB (NF-κB) family of transcription factors. Ligand binding to CD28 activates PKCθ, which is recruited to the plasma membrane by direct interaction with the cytoplasmic tail of CD28 and requires DAG generation for its full activation (10). In addition, CD28-mediated recruitment of additional signaling molecules, such as phosphoinositide 3-kinase (PI3K), growth factor receptor–bound protein 2 (Grb2), or filamin A, contributes to enhanced T cell activation (11). Engagement of costimulatory receptors facilitates secretion of the cytokine interleukin-2 (IL-2) and the cell surface expression of the IL-2 receptor (IL-2R) α chain (CD25), which, together with the IL-2Rβ chain (CD122) and the common γ chain, constitute the high-affinity IL-2R (Fig. 1B) (12). As a result, quiescent resting T cells enter the cell cycle, facilitating the proliferation of effector T cell populations.

The coordinated integration of Ca2+- and DAG-regulated pathways that follows the simultaneous recognition of foreign antigens by the TCR and ligands by costimulatory receptors permits effective defense against pathogens. In the absence of costimulatory signals or in response to weak self antigens, engagement of the TCR causes anergy, a hyporesponsive state that renders T cells refractory to subsequent stimulation (13). The most consistent feature of anergic stimuli is that they stimulate the Ca2+-mediated activation of NFAT, but they cannot fully activate the DAG-dependent PKC–NF-κB or ERK–AP-1 pathways (Fig. 1C). Most of the genes whose expression is increased in tolerized or anergic T cells are dependent on NFAT for their expression (14) or are dependent on NFAT-regulated transcription factors, such as early growth response 2 (Egr2) (15). Anergy-induced genes encode proteins responsible for dampening signals downstream of the TCR and inhibiting the production of IL-2 (16).

DAG Kinases Act to Restrain DAG Signals

DAG kinases (DGKs) are a family of lipid kinases that terminate DAG signaling by transforming it into phosphatidic acid (PA) (17). DGKs have a conserved catalytic region and at least two atypical C1 domains that, except in the case of DGKβ and DGKγ, do not recognize DAG (18). Mammals express 10 different isoforms of DGKs, which are classified into five subgroups on the basis of distinct regulatory motifs in their primary sequences (17). Type I DGKs (α, β, and γ) have EF-hand structures and recoverin homology domains; type II DGKs (δ, η, and κ) have a pleckstrin homology (PH) domain and a sterile α motif (SAM); the lone type III DGK (ε) is the simplest and has no regulatory motif; whereas the two type IV DGKs (ζ and ι) have myristoylated, alanine-rich PKC substrate (MARCKS), ankyrin, and PDZ-binding domains. DGKθ, which has three C1 domains and a PH-like domain, is classified as type V DGK.

DGKα and DGKζ are the two main DGK isoforms found in T cells (Fig. 2). DGKα, a type I DGK, has a unique expression pattern, which is characterized by an increased abundance in the thymus and peripheral T cells that decreases after T cell activation (19, 20). Early experiments in T cell lines demonstrated that ectopic expression of DGKα represses the TCR-dependent activation of Ras-ERK signaling and AP-1–mediated transcription of the gene encoding the activation marker CD69 (21). In addition, expression of a kinase-inactive DGKα mutant promotes the plasma membrane localization of RasGRP1 and the activation of Ras-ERK signaling, suggesting a role for this isoform as an inhibitor of the RasGRP1-Ras-ERK pathway (22). The identification of the DGKα gene (DGKA) as one of the genes that is increased in expression after the induction of clonal anergy (14) reinforced the idea that the DGKα-mediated consumption of DAG leads to anergy. Further studies in mouse models confirmed a direct correlation between DGKα and T cell anergy; ectopic expression of DGKα in murine primary T cells induces anergy (23), and DGKα-deficient mice have an anergy-resistant phenotype (24). Another study identified the Egr2 transcription factor as necessary for expression of the gene encoding DGKα in models of superantigen- and tumor-induced anergy (Fig. 2) (25).

Fig. 2 Summary of the mechanisms proposed to regulate DGKα and DGKζ in T cells.

The regulatory domains of the DGKα (left) and DGKζ (right) isoforms are depicted. Whereas both isoforms have the characteristic C1 and catalytic domains, specific features define their specialized function. DGKα has two EF-hand motifs that bind Ca2+, thus providing a mechanism for conformational change and localization at the plasma membrane. In addition, the C1 regions of DGKα bind to the PI3K product PIP3, and Lck phosphorylates the enzyme at Tyr335 (pY). DGKζ contains a MARCKS-like motif that is phosphorylated by PKC, as well as a C-terminal ETAV sequence, which interacts with proteins containing PDZ domains. In lymphoid cells, binding partners of DGKα include RasGRP1 and Lck, whereas DGKζ interacts with RasGRP1 and SNX27. The PI3K- and AKT-mediated inactivation of FoxO represses expression of the gene encoding DGKα, whereas anergy leads to the increased expression of genes encoding both DGK isoforms through transcriptional regulation by Egr2. Another mechanism of genetic control is mediated by miRNAs. The expression of DGKζ is repressed by miR-34a during T cell activation, whereas DGKα is targeted by miR-297 in glioblastoma.

T lymphocytes also have DGKζ, a ubiquitously expressed type IV DGK (Fig. 2). Similar to DGKα, the overexpression of DGKζ substantially impairs Ras activation and CD69 transcription in T cell lines (26). Germline DGKζ knockout mice also show enhanced activation of Ras-ERK signaling, as well as increased CD69 abundance after TCR stimulation (27). Consistent with this function, loss of DGKζ renders T cells hyperproliferative in response to antigenic stimulation and confers resistance to anergy. Similar to the case for DGKα, deletion of Egr2 prevents the increased abundance of DGKζ after anergy is induced (25). Initial studies with human T cell lines and mouse models thus suggested redundant functions for DGKα and DGKζ in the DAG-mediated control of the Ras-ERK pathway. However, more detailed analysis has begun to define important isoform-related contributions to specific T cell functions. DGKα and DGKζ have a common function as DAG consumers and PA generators; however, they are not fully redundant, which is probably a result of their distinct activation mechanisms and/or expression patterns. As members of different DGK subtypes, each isoform has a distinct modular organization, which suggests the existence of specific activation mechanisms (Fig. 2). In addition, other studies suggest that distinct mechanisms are responsible for the selective regulation of isoform abundance.

DGKα Is Activated by Ca2+ and DGKA Is Transcriptionally Repressed by the PI3K-AKT-FoxO Axis

DGKα has two EF-hand domains, which are characteristic of Ca2+-regulated enzymes; in the absence of Ca2+, these domains prevent the localization of the cytosolic DGKα to the plasma membrane (21). Experiments with green fluorescent protein (GFP)–tagged DGKα in Jurkat cells (a human CD4+ T cell line) showed its transient translocation to the plasma membrane in response to Ca2+ and the activation of tyrosine kinases. Pharmacological inhibition or point mutation of the adenosine triphosphate (ATP)–binding site of GFP-DGKα resulted in sustained plasma membrane localization of the construct (21), suggesting that its enzyme activity reduces its localization to the plasma membrane. The Src family kinase Lck phosphorylates DGKα at the hinge between the second C1 region and the catalytic site (Y335 in the human sequence). Analysis of endogenous DGKα showed that tyrosine-phosphorylated DGKα is associated with the plasma membrane and probably corresponds to the active pool of the enzyme. The time course of this phosphorylation mirrors that of PLCγ-1 (Fig. 2) (28), suggesting that the generation and consumption of DAG are concerted. In addition to Ca2+- and Lck-mediated phosphorylation, the membrane localization of DGKα requires the C1 domains, which recognize the lipid products of PI3K (29). The Lck-dependent phosphorylation of DGKα is enhanced by the Ca2+ ionophore ionomycin, which suggests a model in which Ca2+ induces a transition from the closed to the open conformation of DGKα, which enables its localization to the plasma membrane. The description of the Ca2+-induced dissociation of the intramolecular interaction between the EF-hand motifs and the C1 domains of DGKα is consistent with this model (30).

Experiments with a specific anti-pDGKαY335 antibody confirmed the phosphorylation of endogenous DGKα in Jurkat cells and peripheral blood lymphocytes (PBLs) treated with anti-CD3 antibodies, which stimulate the TCR (28). Costimulation with an anti-CD28 antibody results in more transient phosphorylation, suggesting that costimulation accelerates the termination of DGKα-mediated signals. These results correlate with data from experiments with Jurkat cells and PBLs, which showed that costimulatory signals delivered by CD28 or by signaling lymphocytic activation molecules (SLAMs) inhibit DGKα activity (31). The SLAM-associated protein (SAP) is thought to mediate the inactivation of DGKα; however, the exact mechanism remains unclear. Ectopic expression of a kinase-defective DGKα mutant enhances the activation of Ras-ERK signaling in T cell lines (22), suggesting that the extent of inhibition of DGKα is an effective mechanism by which costimulatory signals increase DAG generation and promote DAG-dependent signals (Fig. 3).

Fig. 3 Differential contributions of DGKα and DGKζ to the regulation of DAG in response to stimulation of the TCR and CD28.

(A) In the absence of costimulatory (CD28-dependent) signals, both DGKα and DGKζ act as negative regulators of the DAG that is generated in response to the TCR-stimulated activation of PLC-γ. DAG binds to RasGRP1, which activates the Ras-ERK pathway and cooperates with Ca2+ to activate PKCα, which also activates Ras-ERK signaling and the shedding of CD62L from the cell surface by the metalloprotease ADAM17 (a disintegrin and metalloprotease 17). Lck-mediated phosphorylation of DGKα and the Ca2+ influx downstream of the TCR promote the activation of DGKα, which limits DAG signaling. DGKζ limits the extent of PKCα activation, which in turn modulates the membrane localization and activation of DGKζ, providing another layer of control of the Ras-ERK signaling pathway. (B) CD28 signaling recruits PKCθ close to the plasma membrane and inactivates DGKα, which results in increased DAG abundance. The increased amounts of DAG further activate RasGRP1 and supply the necessary lipid for the activation of PKCθ, a mechanism that ensures the appropriate function of DGKζ as the main DAG consumer in response to strong TCR activation and costimulatory signals.

In addition to posttranslational regulation downstream of the TCR and CD28, the gene encoding DGKα is transcriptionally repressed after T cell activation (20). The mechanism that governs this inhibition of expression was partially resolved by the identification of three binding sites for the transcription factor FoxO in Dgka and that are conserved in mammalian DGKα-encoding genes (32). FoxO-mediated transcription is inactivated by the PI3K-dependent activation of AKT, which in turn controls the phosphorylation state and nuclear localization of FoxO. AKT-mediated phosphorylation of FoxO promotes its nuclear exclusion and terminates transcription (Fig. 2). FoxO-mediated transcription couples the temporal pattern of DGKα abundance to costimulatory signals; TCR stimulation elicits PI3K activity through the p110δ catalytic isoform, which leads to weak, transitory phosphorylation of FoxO (33). On the other hand, costimulation results in the recruitment and activation of additional PI3K isoforms, which enhances the phosphorylation of FoxO (34). The binding of IL-2 to the IL-2R also stimulates the PI3K-AKT pathway and represses FoxO-mediated transcription. This PI3K-AKT-FoxO pathway regulates critical factors that sustain cell proliferation (35) and promote the differentiation of T cells into effector cytotoxic T lymphocytes (CTLs) (36). The gene encoding DGKα is transcriptionally repressed by IL-2 (32), which helps to explain the ability of IL-2 to reverse T cell anergy (37). Transcriptional regulation of the gene encoding DGKα downstream of IL-2 in CTLs suggests the presence of additional layers of control that couple the intensity of PI3K-AKT-FoxO signals to DAG-regulated signals. CTLs derived from DGKα-deficient mice thus show increased IL-2–dependent proliferation and Ras-ERK activation, as well as an enhanced IL-2–dependent cytotoxic program (32).

In contrast to the Egr2/3 transcription factors, which regulate DGKα and DGKζ downstream of the TCR (25), the transcriptional control of DGK activity by the PI3K-AKT-FoxO pathway appears specific to the DGKα isoform. The abundance of DGKZ mRNA (which encodes DGKζ) also decreases during TCR activation, but these changes are similar in the presence or absence of costimulatory signals, and the inhibition of AKT does not alter expression of DGKZ (32). Repression of DGKζ expression by the microRNA (miRNA) miR-34a is proposed as another layer of genetic control that helps to limit the isoform-specific functions of DGKs during effective T cell activation (38). The miRNA-dependent regulation of DGKA expression has not been described in T cells; however, this isoform is reported to be a target of miR-297 in glioblastoma (Fig. 2) (39).

DGKζ Is Functionally Coupled to PKC and RasGRP1 and Limits the DAG-Regulated Threshold for T Cell Activation

In contrast to the functions of DGKα, those of DGKζ in T lymphocytes are regulated by PKC-dependent phosphorylation in its MARCKS region (Fig. 2) (40). A mutant DGKζ bearing serine-to-alanine mutations in this region did not relocate to the plasma membrane in Jurkat cells, did not consume DAG, and had no effect on expression of the gene encoding IL-2 (41). Phosphorylation of the DGKζ MARCKS domain is thought to release the negative restriction imposed by the C-terminal domain, which has four ankyrin-like repeats and a PDZ-binding domain that interacts with PDZ-containing proteins (PDZbm) (42, 43). Proteomic studies identified sorting nexin 27 (SNX27) as a DGKζ partner in T lymphocytes and showed that the interaction between both proteins facilitates the localization of DGKζ in endosomal and recycling compartments (42). Knockdown of SNX27 in Jurkat cells elicits activation of the ERK pathway similarly to that observed in DGKζ-deficient cells, which suggests that both proteins act in the same pathway (44).

Whereas initial analyses reported a similar hyperactive phenotype in T cells from DGKα- and DGKζ-deficient mice, a study suggests a dominant role for DGKζ in suppressing the Ras signaling pathway (45). The dominant role of DGKζ does not extend to other pathways; for example, DGKα- and DGKζ-deficient T cells show a similar increase in the PKCθ-dependent phosphorylation of IκBα (Fig. 3). The predominant contribution of DGKζ in suppressing the Ras-ERK pathway in primary T lymphocytes confirms findings from a study of the genetic silencing of these two isoforms in Jurkat cells (41). Although endogenous DGKα and DGKζ both associate with the immunoprecipitated TCR-CD28 complex, knockdown of only the DGKζ isoform impairs complex-associated DGK activity. The lesser contribution of DGKα to the enzymatic activity associated with the TCR-CD28 complex is consistent with data showing the inhibition of DGKα in response to CD28-dependent costimulatory signals (31). The inhibition of DGKα by CD28 would also explain the predominant contribution of DGKζ to the TCR-mediated control of DAG metabolism (Fig. 3).

The underlying cause of the predominant role for the DGKζ isoform in regulating Ras-ERK signaling, even in the absence of costimulatory signals, is not clear. Joshi et al. (45) suggested that although DGKζ is less abundant than DGKα in T cells, it has higher affinity for RasGRP1. This difference contrasts with a mathematical model that characterizes the RasGRP1 and SOS pathways as analog and digital determinants, respectively, in the regulation of Ras activation in T cells (46). The discrepancies between the original mathematical predictions and experimental results imply that the DGKζ deficiency has effects other than merely promoting the interaction between DAG and RasGRP1 and are suggestive of the participation of additional DAG effectors (47). A study of Jurkat cells demonstrated that the enhanced activation of Ras-ERK signaling in DGKζ knockdown T cells requires PKCα (48) because PKC inhibitors or small interfering RNA (siRNA)–mediated knockdown of PKCα reversed the increased Ras-GTP loading and enhanced ERK activation induced by the loss of DGKζ. DGKζ limits the transient translocation of PKCα to the immunological synapse, a function linked to the TCR-stimulated shedding of CD62L from primary mouse T lymphocytes (Fig. 3) (48).

These findings indicate that DGKζ restricts the activation of Ras-ERK signaling not only by limiting the DAG-mediated recruitment of RasGRP1 to the plasma membrane but also by inhibiting PKCα (48). The role of DGKζ as a regulator of PKCα upstream of Ras-ERK signaling concurs with a study that described the PKC-mediated regulation of RasGRP1 (49) and also explains the discrepancies between the original mathematical predictions and experimental results (47). The contribution of the DGKζ-PKCα axis to regulation of the antigen-dependent shedding of CD62L, which is itself associated with virus clearance and antitumor immunity, is also consistent with the strong antiviral and antitumor activity exhibited by DGKζ-deficient mice (47, 50). The precise role of PKCα in modulating Ras activity through RasGRP1, DGKζ phosphorylation, or both and its effect on the interaction between DGKζ and RasGRP1 remain to be clarified.

DGKα and DGKζ Regulatory Mechanisms Delimit Their Biological Function as DAG Suppressors

The data cited thus far suggest that the attenuation of DAG-regulated functions in T lymphocytes requires distinct mechanisms to ensure precise spatiotemporal activation of DGKs. The activation of Ras by a plasma membrane lipid, either directly by RasGRP1 or indirectly through PKC-mediated modulation, is part of the sophisticated machinery that translates receptor affinity into signal strength. The specific contribution of DGKζ as an effector or modulator of PKC correlates well with its role in coupling the activation of the RasGRP1-Ras signaling pathway to the magnitude of the signals (Fig. 3). The interaction between DGKζ and PDZ-containing proteins, such as SNX27, suggests a regulatory mechanism that targets DGKζ to specific membrane compartments.

Whereas DGKζ is an inhibitor of Ras signaling in response to strong TCR signals, most studies suggest that the predominant contribution of DGKα is to sustain weak, incomplete signals. Structure-function studies predict that stimuli that cause tolerance or anergy will cause disproportionate activation of DGKα through Ca2+-induced conformational changes and Lck-dependent phosphorylation. In such cases, the DGKα response would block the Ras-ERK pathway, contributing to incomplete, nonproductive T cell activation (Fig. 3). In addition, weak TCR signals limit the PI3K-dependent activation of AKT, which facilitates FoxO-mediated transcription of the gene that encodes DGKα.

DGK-Mediated Generation of PA and Its Contribution to T Cell Functions

DGKs terminate DAG-regulated signaling by catalyzing the conversion of DAG into PA, another lipid with pleiotropic functions. When inserted into lipid bilayers, PA provides flexibility and curvature, modulating membrane shape. In addition, the negative charge of PA facilitates the plasma membrane recruitment and activation of signaling proteins that have planar positive charge domains (51). Although several studies have shown functions for the PA generated by DGKα and DGKζ, most studies were not performed with T cells. The mammalian target of rapamycin (mTOR), a master regulator of cell growth and proliferation that integrates nutrient and growth factor signals, is a key target of PA. Although the PA generated by phospholipase D has been widely implicated in the activation of mTOR complex 1 (mTORC1), PA generated by DGKζ also activates mTORC1 (52, 53). In Drosophila and Caenorhabditis elegans, the knockdown of DGKζ orthologs limits the activation of the mTORC1 effector ribosomal S6 kinase (S6K) and prolongs life span (54). These experiments suggest that the DGKζ-mediated activation of mTORC1 is a conserved mechanism that may contribute to control aging and stress tolerance.

The role of DGKζ as a stimulator of mTORC1 contrasts with the enhanced TCR-dependent mTOR signaling reported in thymocytes doubly deficient in DGKα and DGKζ (55). This apparent incongruity is probably a result of the distinct role of DAG in T lymphocytes as a direct regulator of Ras activation downstream of the TCR. The contribution of DAG as a stimulator of mTOR through activation of RasGRP1-Ras signaling correlates with the impaired activation of mTOR exhibited by RasGRP1-deficient thymocytes (55). The correlation between activating mutations of RasGRP1 and basal mTOR activation (56) confirms the relevance of DAG as a positive regulator of a RasGRP1-mTOR axis in T cells.

Another important target of the PA generated by DGK is phosphatidylinositol 4 phosphate 5 kinase (PI4P5K), an enzyme that catalyzes the conversion of PI4P into PI45P2. In addition to acting as a substrate for PLC-γ and PI3K, PI45P2 regulates adhesion of the cytoskeleton to the plasma membrane, actin dynamics, and the assembly of endocytic vesicles (57). PI45P2 is proposed to modulate T cell rigidity at the immunological synapse (58). A study in human embryonic kidney (HEK) 293 cells showed that DGKζ interacts with PI4P5K type Iα in lamellipodial protrusions, where it enhances actin polymerization (59). A similar functional interaction of DGKζ with PI4P5Kζ in T lymphocytes (although it has not yet been described) could contribute to the regulation of cytoskeletal reorganization and endocytosis by PI45P2. Indeed, CD28-mediated regulation of PI45P2 turnover through the recruitment and activation of PI4P5Kα is proposed to promote CD28 costimulatory signals (60).

Studies of neuronal and skeletal muscle cells showed that DGKα- and DGKζ-derived PA participates in Rac-mediated cytoskeletal remodeling. DGKζ-mediated generation of PA promotes neurite outgrowth by activating p21-activated kinase 1 (PAK1), which phosphorylates Rho-family guanine nucleotide dissociation inhibitors (RhoGDIs) and causes the dissociation of this inhibitor from Rac1 (43, 61). A similar mechanism is proposed for DGKα in epithelial cells, in which the PA-dependent recruitment of the atypical PKCζ mediates the release of Rac from the inhibitory RhoGDI complex (62). At the immunological synapse, PKCζ promotes polarization of the microtubule-organizing center (MTOC) in both CD4+ and CD8+ T cells (63). It is unclear whether a DGKα-PA-PKCζ axis operates during formation of the immunological synapse, but studies report defects in MTOC polarization as a result of DGKα deficiency (64).

Finally, several reports linked the DGK-mediated generation of PA with the control of membrane and protein trafficking. An interaction between DGKα and the protein MICAL-L1 (microtubule-associated monooxygenase, calponin, and LIM domain containing–like 1) is proposed to facilitate the biogenesis of tubular recycling endosomes and the recycling of MHC I in HeLa cells (65). This function of DGKα in endosome recycling correlates with its regulation of integrin recycling and tumor invasiveness through the PA-dependent tethering of the Rab11 effector RCP (Rab-coupling protein) to invasive pseudopods (66). In T lymphocytes, DGKα is critical for the polarization of multivesicular bodies and the release of exosomes toward the immunological synapse (67), whereas DGKζ interacts with SNX27 in the recycling compartment (44). Future studies should explore the function of DGKα- and DGKζ-generated PA in membrane traffic during T cell activation.

Contributions of DGKα and DGKζ to T Cell Development

The previous sections summarize the differences and similarities between the functions of DGKα and DGKζ as consumers of DAG and generators of PA. Their distinct regulatory domains, mechanisms of transcriptional regulation, and the nature of their substrates suggest their participation in various aspects of T cell function. Analysis of DGKα- and DGKζ-deficient mice provides important clues to both the redundant and the distinct contributions of these two isoforms during T cell development.

T cells acquire self-tolerance properties during their development in the thymus, in a process that impedes the generation of immunodeficiency and autoimmune pathologies. Thymus-generated T cell subsets include CD4+ T cells, CD8+ T cells, CD4+ regulatory T (Treg) cells, invariant natural killer T (iNKT) cells, and γδ T cells. Thymocytes progress from the double negative (DN) to the double positive (DP) stage, and the intensity of TCR-stimulated signals determines the fate of distinct T cells. As discussed in the next sections, the DGK-mediated control of DAG is a key factor in the regulation of the signals that ultimately drive the development of these T cell populations.

DGKζ Specifically Limits the Development of Treg Cells

Two studies identified DGKζ as being critical for the development of natural Treg (nTreg) cells (45, 68). nTreg cells contribute to peripheral T cell tolerance by limiting the responses of effector T cells, at least in part through the secretion of immunosuppressive cytokines, such as transforming growth factor–β (TGF-β) and IL-10 (69). Treg cells are characterized by the presence of the transcription factor Foxp3, which can form heterodimers with NFAT and bind to the promoters of the genes encoding IL-2, CD25, and CTL4. Foxp3 competes with AP-1 for binding to the same binding site in the IL2 promoter and represses the expression of IL-2 (70). The cell surface expression of the high-affinity IL-2R on nTreg cells leads them to compete with effector T cells for the available IL-2, which contributes to their suppressive phenotype (71).

Compared to wild-type mice, DGKζ-deficient mice have increased numbers of CD25+Foxp3CD4+ nTreg precursors and of CD25+Foxp3+CD4+ nTreg cells (68). This increase in the number of nTreg precursors might be explained by the enhanced activation of NF-κB and AP-1, transcription factors that stimulate CD25 production and are sensitive to DAG signaling. DGKζ deficiency also facilitates the transition of precursor cells to mature CD25+Foxp3+CD4+ nTreg cells, which requires CD28-stimulated signals (72). This gain of function in DGKζ-deficient mice is attributed to enhanced nuclear translocation of c-Rel, the NF-κB family member that opens the Foxp3 locus to other transcription factors. Reducing either c-Rel abundance or ERK phosphorylation reverses the DGKζ-deficient phenotype (73). These data suggest the selective consumption of DAG by DGKζ downstream of the TCR and CD28 and confirm the contribution of strong TCR-mediated signals in the development of Treg cell populations.

DGKα and DGKζ Have Redundant Functions in the Development of Effector T Cells

Strong TCR signals stimulate the development of nTreg cells (74) and also control the selection of effector T cells to prevent immunodeficiencies and autoimmune diseases (75). Thymocytes with high affinity for self antigens are deleted during negative selection, and those selected cells progress through developmental stages based on their TCR signaling strength. Genetic deletion of RasGRP1 impairs the positive selection of thymocytes in mice, which confirms the DAG-mediated activation of Ras-ERK as a critical decoder of signal intensity downstream of the TCR (76). Deficiency in DGKα or DGKζ alone does not alter the normal development of effector T cells. Altered T cell development is only observed in DGKα and DGKζ double knockout mice, which suggests that the two isoforms are functionally redundant. As is the case for RasGRP1−/− mice, DGKα and DGKζ doubly deficient mice show a developmental impairment at the DP stage (77). Analysis of DGKα- and DGKζ-deficient thymocytes demonstrated the enhanced phosphorylation of ERK1/2, S6K1, 4E-BP1, and AKT after TCR engagement, which is suggestive of enhanced activation of the Ras and mTOR signaling pathways (55). The severe developmental blockade at the DP stage as a result of the combined deficiency in DGKα and DGKζ implies that a tight control of DAG signaling drives the progression of thymocytes through this stage. The partial thymocyte development observed after the addition of PA suggests that there is synergy between DGKα and DGKζ not only through DAG metabolism but also through the generation of PA (77). The exact mechanism by which PA promotes the maturation of DP thymocytes remains to be clarified.

Studies of transgenic mice that express a constitutively active mutant DGKα (caDGK) in the thymus provided a different model to confirm the importance of adequate DAG metabolism in T cell development. These mice, whose phenotype resembles that of RasGRP1−/− mice, have defects in the positive selection of thymocytes, as well as altered CD4 and CD8 cell lineage commitment (78). Impaired production of DAG also affects early T cell development checkpoints, as shown by the partially impaired DN-to-DP progression and the accumulation of immature single positive populations of cells in the caDGK mice. These developmental defects translate to peripheral organs, as exhibited by lymphopenia and a hyperactive mature T cell phenotype that again closely resembles that of RasGRP1−/− mice.

DGKα and DGKζ Have Redundant Roles in the Development and Functions of iNKT Cells

iNKT lymphocytes are innate-like, αβ T cells with a semi-invariant TCR that recognizes lipid antigens presented by CD1d. iNKT cells are derived from DP thymocytes; they are selected by interacting with cortical thymocytes through signals from their invariant TCR and costimulatory signals delivered by SLAM family members (79). After positive selection, iNKT cells mature functionally in three differentiation steps that are characterized by differential expression of the activation markers CD44 and NK1.1. Immature stage 1 iNKT cells (CD44NK1.1) undergo several rounds of cell division before differentiating into stage 2 cells (CD44+NK1.1) and then increase the abundance of NK1.1 for further maturation to stage 3 iNKT cells (CD44+NK1.1+). DAG-mediated Ras activation is important for iNKT cell development, as demonstrated by the substantial decrease in numbers of these cells in RasGRP1-deficient mice (80) and in transgenic mice that express a caDGK (78). Loss of either DGKα or DGKζ alone does not markedly affect the development of iNKT cells; however, DGKα and DGKζ doubly deficient mice show an acute reduction in the number of iNKT cells, with increased cell death and a blockade in the transition from stage 2 (CD44+NK1.1) to stage 3 (CD44+NK1.1+) (81). Enhanced activation of the RasGRP1-Ras-ERK and PKCθ–IKK–NF-κB pathways as a result of a double deficiency in DGKα and DGKζ correlates with the defects in the transition from stage 2 to stage 3, as well as in the increased cell death observed in mice bearing constitutively active K-Ras or IKKβ, respectively (81). The essential roles for mTORC1 and mTORC2 in the development of the iNKT cell lineage and in the effector functions of these cells (82, 83) strengthen the idea of synergy between DGKα and DGKζ in the activation of mTOR. The contributions of both isoforms correlate with results showing that a deficiency in Egr2 impairs the generation of iNKT cells (84), and suggest that the need for positive and negative signals downstream of the TCR during the selection of conventional αβ T cells is conserved in iNKT cells.

Isoform-Specific DGK Function in the Spatial Control of DAG During Formation of the Immunological Synapse

Formation of the immunological synapse defines a situation in which the Golgi apparatus, the MTOC, and mitochondria all polarize toward the APC (85). The DAG generated as a result of antigen recognition is maintained and restricted to the contact area between the T cell and the APC, as shown by different approaches that stimulate T cells (5, 86). The localized generation of DAG at the immunological synapse acts as a signaling platform to recruit and activate DAG effectors (87). Given that RasGRP1 is the main Ras activator, and considering the need for DAG binding to activate RasGRP1, the accumulation of DAG at the immunological synapse links it directly to Ras activation. T cells use the Ras-ERK pathway to translate antigen affinity into biological output, which includes positive selection in the thymus, cytolytic functions, and peripheral tolerance. Studies of thymocytes, for example, showed that the compartmentalization of the RasGRP1-Ras signal is linked to the establishment of T cell activation thresholds (88).

In effector T cells, the restricted accumulation of DAG mediates reorientation of the MTOC by the sequential recruitment to the immunological synapse and the activation of three novel PKC isoforms (PKCε, PKCη, and PKCθ) (89). The importance of the localized accumulation of DAG at the immunological synapse for MTOC polarization was demonstrated in experiments with phorbol esters and pharmacological inhibitors of DGK activity. These treatments impaired MTOC polarization and the CTL-mediated killing of target cells without inhibiting degranulation (86). Studies in Jurkat cells and primary T lymphocytes addressed the relevance of each DGK isoform to the spatial regulation of DAG abundance during formation of the immunological synapse. GFP-fused DGKα and DGKζ constructs and a fluorescent DAG sensor (the C1 tandem domain of PKCθ) were used to assess translocation dynamics during formation of the immunological synapse in both models. In Jurkat cells in contact with SEE (staphylococcal enterotoxin E superantigen)–loaded Raji B cells, ectopic GFP-DGKζ shows rapid, sustained accumulation at the plasma membrane, including the immunological synapse, which correlates with the impaired accumulation of a Cherry-DAG biosensor at this site (Fig. 4). The limited accumulation of DAG at the T cell–APC contact area requires DGKζ activity and correlates with impaired expression of Il2 (41). In this model, consistent with previous findings (21), sustained membrane localization of GFP-DGKα was only observed for a kinase-deficient mutant.

Fig. 4 Spatial organization of DGKα and DGKζ during formation of the immunological synapse.

(A) Formation of the immunological synapse leads to polarized movement of the MTOC and the Golgi and endocytic recycling compartments toward the cell-cell contact area. The accumulation of DAG mediates polarization of the MTOC; the scheme shows the proposed redistribution of DGKα and DGKζ during formation of the immunological synapse. (B and C) The relative distributions of DAG and PIP3 across the cell-cell contact area during formation of the immunological synapse. (B) The generation of PIP3 at the peripheral synapse area facilitates the recruitment of DGKα, which consumes DAG; this limits the recruitment and activation of nPKC, which drives MTOC polarization. (C) The interaction between SNX27 and DGKζ limits the consumption of DAG in the recycling compartments, which promotes the localization of RasGRP1 and PKCα to the immunological synapse and their activation.

In contrast, similar overexpression experiments in mouse primary T cells showed comparable recruitment of GFP-DGKα and GFP-DGKζ to the immunological synapse (45). Another study revealed a substantial defect in MTOC polarization in DGKα−/− CD4+ T cells during immunological synapse formation, a defect that is not reproduced by loss of DGKζ (64). Analysis of the distribution of DAG with lipid bilayers and a DAG sensor suggests that DAG is distributed more broadly in DGKα-deficient cells than in wild-type cells, and total internal reflection fluorescence microscopy approaches demonstrated the localization of GFP-DGKα at the peripheral ring of F-actin (Fig. 4). The translocation of DGKα requires PI3K activation, consistent with early studies that showed the PI3K-dependent control of DGKα translocation (29). Pharmacological inhibition of PI3K mimics the DGKα−/− phenotype, with enhanced diffusion of DAG and impaired MTOC polarization, and also leads to the homogeneous distribution of ectopically expressed GFP-DGKα at the immunological synapse (64).

These two cell models thus identified DGKζ as the isoform that mainly controls overall DAG abundance, whereas analysis of DGKα-deficient mice suggests that this isoform metabolizes DAG at the peripheral supramolecular adhesion complex (pSMAC) (Fig. 4). This specific contribution of DGKα to the consumption of DAG at the pSMAC correlates with the impaired recruitment of PKCθ to the immunological synapse in SAP-deficient T cells and with the rescue of this defect by siRNA-mediated silencing of DGKα (31). The defect in MTOC polarization is reversed by reconstitution of the cells with catalytically active DGKα, which suggests that, in addition to DAG consumption, PA-regulated functions might be necessary at the pSMAC. Additional studies should test whether PA generated by DGKα at the pSMAC regulates Rac functions through a PKCζ-dependent axis, similar to that described in epithelial cell models (62). This function is consistent with studies showing that Rac is necessary for MTOC polarization and F-actin ring formation (90), and that the localization of PKCζ to the immunological synapse in CTLs regulates MTOC polarization (63).

Despite the reported defects, a deficiency in DGKα does not alter cytolytic functions (64). This could be attributed to the low abundance of DGKα in CTLs (32) and would be consistent with data showing that PKCζ activity is necessary for MTOC polarization but not for CTL-mediated killing of targets (63). Whereas a deficiency in DGKα has no apparent effect on CTL function, its overexpression, activation, or both as a result of impaired costimulatory signals might contribute to defective CTL responses. This would coincide with the proposed effect of DGKα on impaired T cell responses in patients with X-linked lymphoproliferative syndrome (XLP), a severe, rare primary immunodeficiency linked to a mutation in SAP that leads to defective cytotoxic responses and fatal mononucleosis after infection with Epstein-Barr virus (91). This finding would also be consistent with the enhanced responses of CTLs to tumors after the loss of DGKα and DGKζ (92). The inhibition of DGKα could be a mechanism to reduce defects in CTL responses in immunodeficiencies and in cancer. The induction of T cell tolerance by ectopic expression of DGKs could also be of clinical interest to reduce responses to allergens, as was shown in experimental asthma models (93).

Cancer, Immunotherapy, and DGKs: Blocking the Blocker

DGK research has received much attention in recent years because evidence suggests that these kinases contribute to tumor immunosuppression. Immune evasion by cancer cells was long ago recognized as a hallmark of cancer, and fostering the intrinsic ability of T cells to recognize and kill tumors was immediately perceived as an attractive alternative for cancer treatment. Many years of trial and error were nonetheless needed before scientists appreciated the complexity of tumor strategies to escape immune surveillance. The initial concept of tumors as unrecognized foreign invaders drove approaches aimed to “train” T lymphocytes to recognize tumor-specific antigens, which led to the development of chimeric antigen receptor (CAR)–expressing T cells. These patient-derived cells are genetically modified to express tumor antigen–specific antibodies linked to intracellular domains that mediate T cell activation (94). Reinfusion of CAR T cells that target the CD19 B cell antigen yielded positive results in patients with hematological malignancies and is a target of ongoing clinical trials (94). Tumors nonetheless not only camouflage themselves as self but specifically devise immunosuppressive approaches that impair T cell functions.

Tumor-infiltrating lymphocytes (TILs), even those that are genetically modified, acquire a hypofunctional phenotype when they reach the tumor milieu (Fig. 5). The immunosuppressive capacity of tumor cells derives from cytokine secretion and from the recruitment of cells with immunosuppressive functions, such as Treg cells (95). The metabolic features of the tumor microenvironment, which is acidic and hypoxic, are also highly immunosuppressive. Tumors induce hypofunctional T cells by increasing the cell surface expression of inhibitory co-receptors that impair the capacity of T cells to proliferate, secrete cytokines, and mediate cytolysis (96). These so-called immune checkpoint receptors include CTLA-4, which competes with CD28 for binding to its ligands on the target cell (97), PD-1 (programmed cell death receptor 1), which interacts with its ligand to cause the apoptotic death of activated T cells (98) and T cell exhaustion (99), as well as BTLA (B and T lymphocyte attenuator) (100) and LAG-3 (lymphocyte activation gene 3) (101). Antibodies to some of these molecules have yielded positive results in patients with advanced cancers, mainly melanoma. Ipilimumab, which blocks CTLA-4, was the first immune checkpoint antibody approved for clinical use in metastatic melanoma (102).

Fig. 5 DGKs are increased in abundance in hypofunctional, tumor-infiltrating lymphocytes.

Enhanced tumor metabolism generates a hypoxic, acidic microenvironment that promotes immunosuppression through cytokine secretion and the recruitment of cells with immunosuppressive functions, including Treg cells and macrophages. This intratumor milieu facilitates the transition of TILs from an active functional state to a hypofunctional state, which is characterized by the increased cell surface abundance of inhibitory receptors (including CTLA-4, PD-1, and LAG-3) and increased amounts of DGKα and DGKζ. This phenotype limits the responses of T cells and enables tumor cells to evade the immune system.

The induction of hyporesponsive TILs by tumors is also mediated by the increased abundances of DGKα and DGKζ (Fig. 5). Studies of human renal carcinoma showed the increased abundance of DGKα in TILs, which correlates with their inability to stimulate ERK phosphorylation or mount an effective cytotoxic response (103). In these studies, the treatment of TILs with DGK inhibitors promoted ERK activation and the delivery of cytotoxic granules. DGKα acts similarly in the hypofunctional phenotype of tumor-infiltrating NK cells (104). The importance of DGK function to the hypofunctional state of TILs was confirmed by studies of human CAR T cells. Although these cells can slow tumor growth, they fail to cause total regression. As was observed previously for TILs, the authors of this study also found loss of functional activity, a phenotype associated with the increased abundance of inhibitory receptors, the tyrosine phosphatase SHIP-1, and of DGKα and DGKζ (105). These studies highlight a correlation between DGK function and decreased T cell function, and they suggest that by consuming DAG, DGK drives the equilibrium of TCR-derived signals toward Ca2+-dependent pathways.

Loss of TIL function is not a terminal, nonresponsive stage because inhibitory receptors and DGK abundance both rapidly decrease when T cells are cultured in complete growth medium (105). Identifying the factors in the tumor microenvironment that increase negative regulation and determining how they do so will provide valuable information for the design of new therapeutic approaches. The reversibility of tumor-mediated immunosuppression and the increased abundance of DGKs in this process is a starting point for these treatments.


T cell homeostasis requires continuous control of the environment, with quiescent T cells trafficking between hematopoietic organs, ready to respond to foreign antigens. After antigenic challenge, proper cellular activation guarantees adequate responses, but inhibitory mechanisms are also needed to control the intensity and duration of signals generated after engagement of the TCR. The equilibrium between effector and regulatory T cell populations, the elimination of terminally differentiated effector cells, and the generation of memory cells are distinct outcomes that rely on the correct decoding of signal intensity. The consumption of DAG by DGKα and DGKζ is a powerful inhibitory mechanism that couples the intensity of TCR-stimulated signals with T cell function. Modulating either the abundance or localization of DGKs is a potent means by which T cells direct their responses. Detailed phenotypic analysis of DGKα- and DGKζ-deficient mice demonstrates the specific contributions of each isoform to T cell functions. Tumors that use immunosuppressive approaches to survive can evade immune attack by increasing the abundances of these two isoforms in T cells that infiltrate tumors. The enhanced abundance, activation, or both of DGKα in the T and NK cells of XLP patients might help limit their capacity to eliminate infected B cells. The development of siRNA-based approaches, inhibitory constructs, and pharmacological inhibitors of DGK function could be of therapeutic use in promoting the thymic generation of Treg cells as well as promoting the cytolytic killing of tumors and infected cells.


Acknowledgments: We thank the members of the Mérida group, as well as Y. Carrasco and E. Díez for helpful comments and critical discussion, and C. Mark for excellent editorial assistance. Funding: E.A. holds an FPU (Formación de Profesorado Universitario) fellowship from the Spanish Ministry of Education. A.A.-F. was supported by the Spanish Anti-Cancer Association and the Madrid regional government. S.I.G. is supported by Madrid regional government grant S2010/BMD-2305. The work in the Mérida laboratory is supported in part by grants from the Spanish Ministry of Economy and Competitivity (BFU2013-47640-P), the Spanish Ministry of Health (Instituto de Salud Carlos III; Cancer Program Grant RD12/0036/0059), and the Madrid regional government (IMMUNOTHERCAM S2010/BMD-2326) to I.M.
View Abstract

Stay Connected to Science Signaling

Navigate This Article