Mg2+ Regulates Cytotoxic Functions of NK and CD8 T Cells in Chronic EBV Infection Through NKG2D

See allHide authors and affiliations

Science  12 Jul 2013:
Vol. 341, Issue 6142, pp. 186-191
DOI: 10.1126/science.1240094

Magnesium to the Rescue

Individuals with X-linked immunodeficiency with Mg2+ defect, Epstein-Barr virus (EBV) infection, and neoplasia (XMEN) disease are genetically deficient for expression of MAGT1, a magnesium transporter. Chaigne-Delalande et al. (p. 186) sought to better understand why these individuals are chronically infected with EBV at high viral loads and are susceptible to the development of lymphomas. CD8+ T cells and natural killer cells, which help to keep EBV infection in check, exhibited reduced cytotoxicity owing to their lower expression of the cell surface receptor NKG2D, which triggers cytolysis upon ligation. Magnesium supplementation in vitro and also in two XMEN patients restored levels of free Mg2+, increased NKG2D expression, and resulted in reduced amounts of EBV+ cells, suggesting that this may be an effective therapeutic approach for XMEN patients.


The magnesium transporter 1 (MAGT1) is a critical regulator of basal intracellular free magnesium (Mg2+) concentrations. Individuals with genetic deficiencies in MAGT1 have high levels of Epstein-Barr virus (EBV) and a predisposition to lymphoma. We show that decreased intracellular free Mg2+ causes defective expression of the natural killer activating receptor NKG2D in natural killer (NK) and CD8+ T cells and impairs cytolytic responses against EBV. Notably, magnesium supplementation in MAGT1-deficient patients restores intracellular free Mg2+ and NKG2D while concurrently reducing EBV-infected cells in vivo, demonstrating a link between NKG2D cytolytic activity and EBV antiviral immunity in humans. Moreover, these findings reveal a specific molecular function of free basal intracellular Mg2+ in eukaryotic cells.

Divalent metal cations, including Mg2+, play important roles in cellular processes. In eukaryotic cells, 95% of intracellular Mg2+ ([Mg2+]i) is bound (1). The remaining unbound, free [Mg2+]i (~0.5 to 1 mM) is tightly regulated, yet its specific molecular functions are unknown (13). Magnesium transporter 1 (MAGT1) selectively conducts Mg2+ across the plasma membrane (4, 5). We previously characterized a primary immunodeficiency, named X-linked immunodeficiency with Mg2+ defect, Epstein-Barr virus (EBV) infection, and neoplasia (XMEN) disease, caused by null mutations in the MAGT1 gene. We found that MAGT1 was required for a T cell receptor (TCR)–stimulated Mg2+ influx that transiently raises free [Mg2+]i to temporally coordinate T cell activation (2, 3).

XMEN disease is characterized by uncontrolled chronic EBV infection (2, 3). EBV is a B cell–tropic gammaherpesvirus that infects and establishes latency in 90% of humans worldwide (6). Uncontrolled EBV infection can lead to malignancy, especially in B cells (79). The cytotoxic functions of natural killer (NK) cells and cytotoxic CD8+ T lymphocytes (CTLs) are essential for control of viral infections and tumor immunosurveillance (10). XMEN patient cells contain normal bound [Mg2+]i but decreased free [Mg2+]i (2, 3), and whether the latter contributes to XMEN disease pathogenesis was unknown.

To understand defective EBV control in XMEN disease, we examined seven patients harboring mutations in MAGT1 (table S1). Similar to the first three patients characterized (A.1, A.2, and B.1) (2, 3), four additional XMEN patients with repeated minor viral infections and elevated EBV DNA in the blood were identified (table S1 and Fig. 1A). Furthermore, starting in childhood, some patients developed various EBV+ Hodgkin’s, Burkitt’s, and non-Hodgkin’s B cell lymphomas or related lymphoproliferative disorders (table S1 and Fig. 1B). Nevertheless, XMEN patients had normal adaptive immune responses, generating EBV-specific memory CD8+ T cells (fig. S1). However, unlike normal controls, EBV-specific CTLs from XMEN patients expanded by serial stimulations with irradiated autologous lymphoblastoid cell lines (LCLs) showed defective cytotoxicity against autologous LCLs (Fig. 1C) despite comparable expansions of EBV-tetramer–positive cells (fig. S2). Moreover, NK cells that were expanded in vitro with interleukin-2 (IL-2) exhibited defective cytotoxicity against K562 target cells (Fig. 1D). The NK cytotoxicity defect indicated that the previously described TCR defect could not fully account for the defective EBV immunity in XMEN patients. Decreased cytotoxicity also did not stem from deficient granule effectors or degranulation (fig. S3).

Fig. 1 XMEN patients exhibit high EBV levels, lymphoma development, and defective cytotoxicity.

(A) Levels of EBV DNA in XMEN patient blood (n = 7) were expressed as copies/ml blood (†), copies/106 cells (*), or copies/μg DNA (§) depending on the specific clinical laboratory. Each point represents an independent blood draw measurement at different times. Chronic active EBV (CAEBV) patient values are shown for comparison. Shaded area indicates normal range; EBV DNA is typically undetectable in EBV+ subjects, with less than 10% of normal individuals usually showing no more than 20 to 80 copies/ml in the blood (37). (B) Histopathology of the various EBV+ lymphoproliferative disorders (LPD) in XMEN patients. The upper panel shows the hematoxylin and eosin (H&E) staining, and the lower panel shows the EBV-encoded small RNA (EBER) staining by in situ hybridization. Yellow arrowheads show Reed-Sternberg cells. Magnification: ×400 for patients B.1 and E.1 (bottom panel) and ×1000 for patient F.1 and E.1 (top panel). (C) Cytotoxicity of EBV-specific CTLs from father (squares), mother (diamonds), and XMEN patients A.1 (filled circles) and A.2 (open circles) on autologous LCLs. (D) Cytotoxicity of IL-2–expanded NK cells from normal control (filled squares) or XMEN patients A.1 (filled circles) and A.2 (open circles) against K562 cells. Results shown in (C) and (D) are mean ± SEM and are representative of all XMEN patients tested (n = 4) in at least three independent experiments [two-way analysis of variance (ANOVA), ***P < 0.001; ****P < 0.0001].

MAGT1 not only mediates TCR-induced Mg2+ flux but also regulates the basal free [Mg2+]i (24, 11, 12). We found that all XMEN patients, but not chronic active EBV (CAEBV) patients lacking MAGT1 mutations or X-linked lymphoproliferative disease (XLP) patients chronically infected with EBV, had decreased basal free [Mg2+]i (Fig. 2, A and B). Given that the bound [Mg2+]i (95 to 99% of total) is not reduced in XMEN patients (3, 5), we hypothesized that Mg2+ supplementation might restore the free [Mg2+]i in XMEN patient cells. We therefore cultured CTLs and NK cells from XMEN patients in Mg2+-supplemented media and observed a dose-dependent increase in free [Mg2+]i with nearly full restoration after 5 days in 10 mM magnesium sulfate (MgSO4) (Fig. 2, C and D). Magnesium supplementation did not rescue the TCR activation defect (Fig. 2E) or TCR-induced Mg2+ and Ca2+ fluxes in XMEN T cells (fig. S4). However, Mg2+ supplementation did recover the cytotoxicity defect partially in CTLs and nearly completely in NK cells (Fig. 2, F and G), suggesting that the chronically low free [Mg2+]i caused impaired cytotoxicity.

Fig. 2 Mg2+ supplementation restores intracellular Mg2+ and enhances cytotoxicity but does not restore TCR activation.

(A) Flow cytometry profile of the ratio of the mean fluorescent intensity (MFI) of Mg2+-specific fluorescent probe MagFluo4 to the MFI of the Ca2+-sensitive probe Fura Red (MF4/FR) using PBMCs from normal control (red line) and XMEN patient A.1 (blue line). The ratio was arbitrarily set at 1 for normal control (dotted line). (B) Dot plot of the MF4/FR ratios from PBMCs of normal controls (squares, n = 5) and XMEN patients (circles, n = 6), non-XMEN CAEBV patients (triangles, n = 5), or XLP patients (diamonds, n = 4). (C) Flow cytometry profiles of MF4/FR ratio on normal control and XMEN patient CTLs cultured in media supplemented with the indicated amounts of MgSO4 for 5 days. (D) MF4/FR ratio of CTLs from normal control (red bars, n = 4) or XMEN patients (blue bars, n = 4). (E) Percentage of CD69+ in CD8+ cells after stimulation with anti-CD3 (1 μg/ml) for 24 hours in the presence of the indicated amount of MgSO4 or unstimulated. (F) Cytotoxicity of normal or XMEN patient EBV-specific CTLs, supplemented with or without 2 mM MgSO4 for 5 days, on autologous LCLs. (G) Cytotoxicity of normal or XMEN patient NK cells, supplemented with or without 2 mM MgSO4 for 5 days, on K562 targets. Results shown are mean ± SEM and are representative of all XMEN patients tested (n = 4) in at least three independent experiments (two-way ANOVA, *P = 0.01, **P = 0.0038, ****P < 0.0001).

NK cells use a combination of inhibitory and activating receptors to specifically recognize target cells altered by transformation or virus infection (13, 14). Certain activating NK receptors are also expressed on CTLs as costimulatory signals for cytotoxic function (15). We surveyed the major NK receptors and found a selective loss of NKG2D expression on XMEN CTLs and NK cells (Fig. 3, A and B). This was confirmed by surface staining, intracellular staining, and immunoblotting (Fig. 3, A to C, and fig. S5A). The NKG2D loss was not secondary to EBV infection, as cells from CAEBV or XLP patients expressed normal NKG2D levels (Fig. 3B). NKG2D is associated with the adapter DAP10 on NK cells, CTLs, and γδ T cells and mediates antiviral and antitumor cytotoxic responses (1620). We also observed decreased DAP10 in XMEN patients, which could be a cause or consequence of the NKG2D defect (20) (Fig. 3C). Their defective expression was posttranscriptional, as their steady-state mRNA levels were not reduced in XMEN samples (fig. S5B). Ubiquitin-conjugated peptides of NKG2D and DAP10 were detected by mass spectrometry, implying accelerated protein turnover.

Fig. 3 Effect of Mg2+ on NKG2D expression and function in vitro.

(A) Flow cytometry profiles of the surface expression of NK receptors on gated NK cells (CD3CD56+) from normal controls (red and green lines) and XMEN patient A.1 (blue line). Isotype antibody staining (gray shading) is shown as a reference for background fluorescence. Results are representative of all XMEN patients tested (n = 5). (B) Normalized MFI of NKG2D surface staining on gated CTLs from whole blood of normal controls (squares, n = 5) and XMEN patients (circles, n = 6), non-XMEN CAEBV patients (triangles, n = 5), or XLP patients (diamonds, n = 4). (C) Immunoblots of CTLs and NK cells from normal control and XMEN patient with the indicated antibodies. (D) NKG2D-specific redirected lysis of normal or XMEN patient IL-2–expanded NK cells on P815 cells expressing the NKG2D ligand, ULBP1. (E) NKG2D-specific redirected lysis of normal or XMEN patient CTLs on P815 expressing the NKG2D ligand ULBP1 or not in the presence of anti-CD3. (F) Flow cytometry profiles of MF4/FR ratio in CTL (left) and NKG2D expression on CTLs (middle) and NK cells (right) from normal control or XMEN patient supplemented with the indicated concentrations of MgSO4 for 5 days. (G) Immunoblot of cells lysates from CTLs from a normal control or an XMEN patient supplemented with or without 5 mM of MgSO4 for 5 days. (H) Cytotoxicity of normal control or XMEN patient EBV-specific CTLs supplemented with or without 2 mM MgSO4 for 5 days, on autologous EBV-transformed LCLs pretreated with or without soluble NKG2D-Fc. In (C) and (G), arrowheads show the various forms of the indicated protein. “nNKG2D” indicates the normal form of NKG2D and “pNKG2D” the patient’s forms; an asterisk (*) indicates a nonspecific band detected by anti-NKG2D. Results shown are mean ± SEM and are representative of all XMEN patients tested (n = 4) in at least three independent experiments {Student’s t test (B), two-way ANOVA [(D), (E), and (H)], ****P < 0.0001}.

In humans, NKG2D ligands include the major histocompatibility complex (MHC) class I–like proteins MICA and MICB, as well as the UL16-binding protein family of glycoproteins (ULBP1-6) that are induced by infection, cellular transformation, and cell stress (19, 21, 22). Using a ULBP1-expressing P815 target cell, we observed that NKG2D-mediated cytotoxicity by XMEN patient NK cells was abolished (Fig. 3D). Similarly, although defective cytotoxicity against P815 cells coated with an antibody to CD3 (anti-CD3) could involve the TCR signaling defect previously documented, further enhancement of the anti-CD3–induced cytotoxicity by ULBP1 was defective (Fig. 3E), suggesting that NKG2D loss contributed to the CTL killing defect in XMEN patients (2, 3).

We next examined Mg2+ and Ca2+ fluxes after stimulating various NK receptors and found that only CD16 stimulation induced a Mg2+ flux (fig. S6A). This flux was MAGT1-dependent and absent in XMEN NK cells but did not affect the Ca2+ flux in XMEN samples (fig. S6A). Most NK receptors work in synergy with each other (23), and we consistently observed a defect when triggering flux through combinations that included NKG2D in XMEN patient cells (fig. S6, B and C).

Because NKG2D was constitutively reduced in XMEN cells, we hypothesized that this resulted from decreased free [Mg2+]i. Correspondingly, we found that cultivation of normal CTLs in Mg2+-depleted medium caused a reduction of free [Mg2+]i and NKG2D expression (fig. S7). We therefore assessed NKG2D in XMEN NK cells and CTLs after 5 days of cultivation in medium with added MgSO4. We observed a rise in free [Mg2+]i that correlated with increased NKG2D expression in both cell types (Fig. 3, F to G). The supplementation did not change the expression or function of other NK receptors or Mg2+ fluxes (figs. S4 and S8, A and B). Consistent with the lack of flux restoration, Mg2+ supplementation did not restore the CD16-redirected lysis by XMEN NK cells (fig. S8C). EBV+ LCLs expressed substantial levels of NKG2D ligands (fig. S9). The enhancement of CTL killing of autologous LCLs by Mg2+ supplementation in XMEN CTLs (Fig. 3H) appears to be NKG2D-specific because it was blocked by NKG2D-Fc (Fig. 3H). Thus, NKG2D expression is sensitive to free basal [Mg2+]i and can be restored in XMEN patient cells by exogenous Mg2+ supplementation.

The defective killing by XMEN NK cells and CTLs associated with decreased free [Mg2+]i might account for the failed immune control of EBV. Therefore, we attempted to test this hypothesis in vivo by providing magnesium supplementation to patient A.1. First, we administered oral magnesium gluconate for 2 weeks. We observed a small but significant increase in basal free [Mg2+]i and a modest restoration of NKG2D expression on both CTLs and NK cells (fig. S10). The fraction of B cells harboring intranuclear EBV–encoded small RNA (EBER) declined during treatment (fig. S10C). Twenty-one days after treatment cessation (day 36), we observed that basal free [Mg2+]i dropped to the pretreatment level and the fraction of EBV-infected peripheral blood mononuclear cells (PBMCs) rebounded. These data indicate that continuous supplementation is required to maintain free [Mg2+]i and a beneficial effect on EBV control. We also tested the infusion of magnesium sulfate for 3 days and observed stronger and faster responses (fig. S11). We then administered oral supplementation to patients A.1 and A.2 for 175 days with magnesium threonate (MgT), which has better bioavailability and tolerability. Free [Mg2+]i and NKG2D expression increased in both patients within 2 days of treatment (Fig. 4, A and B). The free [Mg2+]i reached a plateau after 7 days of treatment and remained stable thereafter, whereas NKG2D expression increased progressively for the first 60 days (Fig. 4, A to C). The expression of NKG2D ligands observed on B cells was reduced in both patients (Fig. 4D). Moreover, the cells expressing NKG2D ligands were mostly EBV+ (Fig. 4, E and F). We noted a reversal of the fraction of EBV+ B cells in patient A.1 and to a lesser extent in patient A.2 at day 60 of treatment, which was later traced to a gap in Mg2+ administration. This indicates that the effects of Mg2+ supplementation were reversible and that control of EBV levels in vivo was linked to the level of NKG2D expression, which was continuously regulated by the free [Mg2+]i. The decrease in EBV-infected cells may be important because high levels of EBV are correlated with early lymphoma development and mortality in XMEN patients (table S1).

Fig. 4 In vivo Mg2+ supplementation of XMEN patient restored NKG2D expression and decreased EBV+ PBMCs.

Flow cytometry profiles of MF4/FR ratio in PBMC (A) and NKG2D expression on gated CTLs (CD3+CD8+) (B) from a normal control (Mother, untreated), patient A.1 (treated), and patient A.2 (treated) at the indicated times of magnesium supplementation in days (D). Mother was not available for sample on D14. (C) Quantification of the MF4/FR ratio (top), NKG2D MFI (middle) and percentage of EBV+ B cells measured by EBV-encoded small RNA (EBER) fluorescence in situ hybridization (FISH) (bottom) during the magnesium supplementation. The arrowhead and dotted line indicate the starting point of the magnesium supplementation. NKG2D-Fc staining on gated B cells (CD19+) from freshly isolated PBMCs of a normal control, XMEN patients (before D-63 and after D175 Mg2+ supplementation) alone by flow cytometry (D) or costained with EBER FISH in confocal microscopy (E). (F) Quantification of the EBER+/NKG2D-Fc+ cells shown as a percentage of EBV+ cells (±SEM) that are NKG2D-Fc+ or NKG2D-Fc in the patient’s PBMCs during in vivo Mg2+ supplementation as measured by EBV-specific FISH. Scale bars, 5 μm.

Our findings demonstrate that basal free [Mg2+]i is regulated at least in part by extracellular Mg2+ and has concentration-dependent biological functions in eukaryotic cells besides the cofactor function of bound Mg2+ (1, 24, 25). The dynamic control of NKG2D expression by free [Mg2+]i reveals the importance of [Mg2+]i homeostasis for antiviral and antitumor immunity (Fig. 4C). There are more than 20 mammalian magnesium transporters, and it will be interesting to determine which control free [Mg2+]i in XMEN patients during supplementation (5). XMEN patients manifest dysgammaglobulinemia, EBV infections, and susceptibility for lymphoma development similar to that exhibited by XLP patients (9, 26). In XLP1, 2B4- and NTB-A mediated cytotoxicity against EBV-infected B cells is defective (2729). Given the loss of synergy between NKG2D and 2B4 in XMEN patients (figs. S6C and S13), 2B4-mediated recognition of EBV-infected cells may be impaired in XMEN patients as well (23). Unlike cytomegalovirus, EBV does not have mechanisms to evade NKG2D recognition (3032). NKG2D ligands are up-regulated by lytic induction of EBV-infected cells and in tumors from EBV+ posttransplant lymphoproliferative disorders (22, 33). Moreover, recent studies show that NKG2D plays a critical role in the success of antitumor and hematopoietic stem cell transplants (HSCTs) (3436). Given that the only two known HSCT attempts in XMEN disease have been unsuccessful, Mg2+ supplementation may provide an adjunctive treatment in XMEN disease. Although the clinical utility of Mg2+ supplementation remains to be validated, our data illustrate in vivo that NKG2D is regulated by [Mg2+]i and apparently plays a key role in EBV control.

Supplementary Materials

Materials and Methods

Author contributions

Figs. S1 to S13

Tables S1 and S2

References (3844)

References and Notes

  1. Acknowledgments: This research was supported by the Division of Intramural Research, National Institute of Allergy and Infectious Diseases and the National Cancer Institute, NIH. We thank H. Malech, A. Snow, J. Niemela, G. Fahle, and T. Dimaggio for assistance in various aspects of the research and patient evaluation and S. Holland for clinical insights and support. We also thank R. Germain, P. Schwartzberg, R. Siegel, T. Waldmann, K. Schafer-Weaver, C. Kanellopoulou, S. Muljo, D. McVicar, J. Milner, R. Baker, and S. Holland for critically reading the manuscript. We thank E. Long and S. Rajagopalan for sharing antibodies and cell lines and helpful assistance with NK experiments; we also thank K. Dowdell, R. Orentas, and B. Lafont for sharing reagents and D. Douek, D. Price, and M. Quigley for assistance with tetramer staining. We are grateful to Merck Research Laboratories for generous support in kind of reagents and materials, and to P. R. Vagelos and R. Vessey for fostering the collaboration. F.-Y.L. was supported by the Medical Scientist Training Program at the University of California–San Francisco and the NIH M.D./Ph.D. Partnership program, and thanks K. Shannon and J. Toutolmin for support and encouragement. We also acknowledge support by the Pharmacology Research Associate Training (PRAT) program (to C.L.L.), National Institute of General Medical Sciences, NIH. The data presented in this paper are tabulated in the main paper and in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article