Patients With Natural Killer (NK) Cell Chronic Active Epstein-Barr Virus Have Immature NK Cells and Hyperactivation of PI3K/Akt/mTOR and STAT1 Pathways

Abstract

Background

Chronic active Epstein-Barr virus (CAEBV) presents with high levels of viral genomes in blood and tissue infiltration with Epstein-Barr virus (EBV)–positive lymphocytes. The pathogenesis of CAEBV is poorly understood.

Methods

We evaluated 2 patients with natural killer (NK) cell CAEBV and studied their NK cell phenotype and signaling pathways in cells.

Results

Both patients had increased numbers of NK cells, EBV predominantly in NK cells, and immature NK cells in the blood. Both patients had increased phosphorylation of Akt, S6, and STAT1 in NK cells, and increased total STAT1. Treatment of 1 patient with sirolimus reduced phosphorylation of S6 in T and B cells, but not in NK cells and did not reduce levels of NK cells or EBV DNA in the blood. Treatment of both patients’ cells with JAK inhibitors in vitro reduced phosphorylated STAT1 to normal. Patients with T- or B-cell CAEBV had increased phosphorylation of Akt and S6 in NK cells, but no increase in total STAT1.

Conclusions

The increase in phosphorylated Akt, S6, and STAT1, as well as immature NK cells describe a new phenotype for NK cell CAEBV. The reduction of STAT1 phosphorylation in their NK cells with JAK inhibitors suggests a novel approach to therapy.

Patients with natural killer (NK) cell chronic active Epstein-Barr virus disease were found to have increased activation of Akt, S6, and STAT1 in their NK cells as well as large increases in immature NK cells in the blood.

Chronic active Epstein-Barr virus (CAEBV) presents with fever, hepatosplenomegaly, lymphadenopathy, hepatitis, and EBV-positive T, NK, or B cells infiltrating the tissues [1]. These patients have a disease lasting at least 6 months, no known immunodeficiency, and markedly increased levels of EBV in T, NK, or B cells in the blood. CAEBV occurs more frequently in patients of Asian and South or Central American descent [2, 3]. If untreated, these patients develop liver failure; T-cell, NK cell, or B-cell lymphomas; hemophagocytic lymphohistiocytosis; or opportunistic infections. Patients with NK cell CAEBV share many of the features of those with T-cell CAEBV, although they may also have mosquito bite hypersensitivity and elevated levels of immunoglobulin E (IgE) [4]. The only curative treatment is hematopoietic stem cell transplantation (HSCT) [5].

Here, we report 2 patients with NK cell CAEBV whose NK cells showed increased activation of the PI3K/Akt pathway and increased levels STAT1; however, unlike other immunodeficient patients with constitutive activation of these pathways our patients had no mutations identified in these proteins. These patients had an NK receptor repertoire indicative of a block in NK cell development.

METHODS

Patients

All patients were enrolled in a National Institute of Allergy and Infectious Diseases (NIAID) clinical protocol approved by the NIAID institutional review board (ClinicalTrials.gov identifier NCT00032513), and all gave signed consent. Human experimentation guidelines of the US Department of Health and Human Services were followed in the conduct of clinical research. CAEBV was defined as a progressive illness lasting at least 6 months with markedly elevated EBV DNA in the blood, infiltration of tissue with EBV-positive lymphocytes, and no other immunosuppressive condition.

S6 and Akt Signaling

Peripheral blood mononuclear cells (PBMCs) were incubated with eFluor 450 anti-CD3, PerCP anti-CD8, APC eFluor 780 anti-CD20, and BV711 anti-CD56 (BD) antibodies and then stimulated with either Fab immunoglobulin M (IgM), Dynabeads anti-CD3/CD28, or interleukin (IL) 15. The cells were fixed, washed, permeabilized, and washed, and intracellular staining was performed with Alexa Fluor 647 anti-pAkt S473, PE anti-pAkt T308, and Alexa Fluor 488 anti-pS6 S240. After overnight incubation, the cells were washed twice, resuspended in 4% paraformaldehyde, and analyzed by flow cytometry. For analysis, cells that were CD3⁺, CD20⁺, or CD56⁺ were identified and the median fluorescence intensity (MFI) for pAkt S473, pAkt T308, and pS6 S240 was determined. The fold change, comparing patients to healthy controls, for either unstimulated or stimulated cells was calculated based on the MFI of the cells.

NK Cell Receptor Panels

PBMCs were incubated with NK cell antibodies (see below), washed, and analyzed on a BD LSRII flow cytometer. NK cells were identified as CD3^–CD56⁺, and NK cell receptor positive cells were quantified. Antibodies used for the phenotype panel were PE-Cy5.5 anti-CD56, ECD anti-CD2, PE anti-CD11b, FITC anti-α _vβ3 integrin, APC anti-CD11a, BV785 anti-CD3, BV711 anti-CD8, BV605 anti-CD4, Pacific Blue anti-CCR7, and antimouse CD57; secondary: APC-Cy7 antimouse IgM, biotin anti-NKG2C, and BV650 streptavidin. Antibodies used to identify NK cell KIRs (killer-cell immunoglobulin-like receptors) were PE-Cy7 anti-NKG2A, PerCP-Cy5.5 anti-KIR2DL2/L3, ECD anti-CD56, FITC anti-KIR2DL3, Alexa Fluor 700 anti-KIR2DL1/S5, Alexa Fluor 647 anti-CD3, and BV421 anti-KIR3DL1.

STAT Assay

PBMCs were pretreated with JAK inhibitors for 30 minutes and then stimulated with cytokines for 20 minutes and incubated with anti-CD4 (PE-Cy7, APC, or PerCP OKT4), APC anti-CD56 (Biolegend), and Far Red live/dead stain. Cells were washed, fixed, washed, permeabilized, washed, and incubated with FITC anti-pSTAT1 Y701, Pacific Blue anti-pSTAT3 Y705, Pacific Blue anti-pSTAT5 Y694, or FITC anti-pSTAT6 Y641 antibodies. Cells were then washed and analyzed by flow cytometry. Live cells were gated and CD4⁺ and CD56⁺ populations identified. Histograms for each STAT were generated for CD4⁺ and CD56⁺ populations. Tofacitinib was used at 1 μM. Simulations were performed with IL-2 (10 ng/mL), IL-4 (100 ng/mL), IL-6 (10 ng/mL), IL-7 (10 ng/mL), IL-15 (100 ng/mL), or IL-21 (50 ng/mL). Interferons (IFN-α [10 ng/mL] and IFN-γ [10 ng/mL]) were used for stimulations. Additional methods are described in the Supplementary Data.

RESULTS

Clinical History

Patient NK1 is a 17-year-old female Alaska native (Yup’ik) who first presented at age 6 with abdominal pain, elevated liver aminotransferases, and elevated EBV DNA in the blood. Nine years later she presented with abdominal pain, and a liver biopsy showed EBV hepatitis with EBV in NK cells. Five months later, she had abdominal pain and a bone marrow biopsy showed an EBV-positive lymphoproliferative disorder with increased NK cells. She was treated with bortezomib and ganciclovir, but her viral load did not decline. She was evaluated at the National Institutes of Health (NIH) where laboratory data showed alanine aminotransferase (ALT) 66 U/L, aspartate aminotransferase (AST) 51 U/L, alkaline phosphatase 165 U/L, absolute eosinophil count 0.38 K/μL (normal range, 0.04–0.36), IgE 5087 IU/mL (normal range, 0–90), and normal levels of immunoglobulin G, IgM, and immunoglobulin A; EBV DNA was 1 349 000 copies/mL. While she had normal numbers of CD4 and CD8 cells, CD20 cells and NK cells were increased (Table 1). EBV was predominantly in NK cells (2 965 000 copies/10⁶ cells), compared to T cells (155 619 copies/10⁶ cells) or B cells (29 972 copies/10⁶ cells) and 90.2% of the NK cells were EBV DNA positive (Supplementary Figure 1). Due to her elevated liver aminotransferases, NK cell numbers, and EBV DNA level, she was treated with 3 cycles of etoposide-vincristine-doxorubicin-cyclophosphamide-prednisone, which lowered her EBV DNA to 39 000 copies/mL and NK cells to 336 cells/μL (from a high of 6000 copies/μL). She received conditioning therapy with busulfan, thiotepa, fludarabine, antithymocyte globulin, and rituximab and haploidentical CD34⁺ peripheral blood cells from her father. Her liver aminotransferases normalized and EBV DNA became undetectable. There was no family history of CAEBV; however, a maternal great-uncle and great-great grandmother both were diagnosed with autoimmune hepatitis (no liver biopsies were obtained). Whole exome sequencing did not detect pathogenic mutations in genes associated with severe EBV disease [5].

Table 1.

Epstein-Barr Viral Load in Lymphocyte Subsets in the Patients

Cell Typea	Patient 1		Patient 2
	EBV Copies/ Million Cells	No. of Cells/μL (%)	EBV Copies/ Million Cells	No. of Cells/μL (%)
NK	2 965  000	1145 (39.2)	14 400 000	2843 (77.9)
T	155 619	1180 (40.4)	324 000	708 (19.4)
B	29 972	566 (19.4)	938 000	106 (2.9)

Abbreviations: EBV, Epstein-Barr virus; NK, natural killer.

^aNormal values: NK, 126–729/μL (6.2%–34.6%); T cells, 714–2266/μL (60.0%–83.7%); B cells, 61–321/μL (3.3%–19.3%).

Patient NK2 was a 28-year-old man from Ecuador who first presented at age 24 with low-grade fever, night sweats, and weight loss which resolved without treatment. One year later his liver aminotransferases were elevated, a liver biopsy showed EBV lymphoproliferative disease, and his EBV DNA polymerase chain reaction (PCR) was 58 700 copies/mL. A PET/CT showed only hepatosplenomegaly. The following year he received 3 doses of third-party EBV latent membrane protein (LMP)1/2-specific cytotoxic T cells, since LMP1 and LMP2 are typically expressed in CAEBV disease [1]. After each dose of cells, the level of EBV DNA decreased, but increased 3–4 weeks later. Later that year he was evaluated at the NIH, where laboratory data showed ALT 122 U/L, AST 119 U/L, alkaline phosphatase 138 U/L, absolute eosinophil count 0.15 K/μL (normal range, 0.04–0.36), and IgE 93 IU/mL (normal range, 0–90). His EBV DNA PCR was 4 571 000 copies/mL. He had decreased numbers of CD4 cells (345 cells/μL; normal range, 359–1565 cells/μL) and NK T cells (7 cells/μL; normal range, 29–299 cell/μL), and increased numbers of NK cells (1682 cells/μL; normal range, 126–729 cells/μL). EBV DNA was predominant in NK cells (14 400 000 copies/10⁶ cells) compared to T cells (324 000 copies/10⁶ cells) or B cells (938 000 copies/10⁶ cells) (Table 1) and 95.5% of the NK cells were EBV DNA positive. Like patient 1, this patient also had high levels of virus in T cells, which has been reported previously in CAEBV [6]. There was no history of mosquito bite hypersensitivity. While the decision was made to initiate HSCT, he had progressive elevation of liver aminotransferases, bilirubin, and NK cell numbers; despite treatment with corticosteroids, combined ganciclovir and bortezomib, and cyclophosphamide, he died of liver failure. There was no family history of CAEBV or hepatitis. Whole exome sequencing did not detect pathogenic mutations in genes associated with severe EBV disease [5].

Increased Basal and Stimulated Phosphorylation of AKT and S6 in NK Cells From Patients With NK Cell CAEBV

Mutations in genes involved in the Akt signaling pathway have been reported in patients with EBV disease [4, 7–9]. While we did not detect mutations in genes important for Akt signaling in our patients with NK cell CAEBV, we evaluated the PI3K/Akt/mTOR signaling pathway in their cells and examined phosphorylation of a downstream target of mTOR, ribosomal protein S6, as well as 2 phosphorylated residues of Akt when activated [10]. Increased phosphorylation of S6 at amino acid S240 was observed in NK cells stimulated with IL-15 in patients NK1 and NK2 compared with healthy control cells (Figure 1 and Supplementary Figure 2). Increased phosphorylation of Akt at residue S473 was also noted in unstimulated and in stimulated NK cells from patients NK1 and NK2 compared with healthy controls. We also observed an increase in levels of pAkt S473 in unstimulated and stimulated T cells from patient NK1, and in pS6 S240 unstimulated and stimulated B cells from patient NK1 compared with control cells (Supplementary Figure 3). Phosphorylation of Akt and S6 in NK and T cells from the father, mother, and a brother of patient NK1 was similar to that of healthy controls (Supplementary Figure 4). These results indicate that the PI3K/Akt/mTOR pathway was hyperactivated in NK cells from patients NK1 and NK2.

Figure 1.

Increased basal and stimulated phosphorylation of S6 and Akt in natural killer (NK) cells from patients NK1 and NK2. Levels of phosphorylated S6 (S240) and Akt (S473 and T308) were measured in the patients’ NK cells and compared to healthy controls. Patient and healthy control peripheral blood mononuclear cells were stimulated for 20 minutes with interleukin 15 (which is important for NK cell development and differentiation) and analyzed by flow cytometry. NK (CD3^–, CD56⁺) cells were identified and histograms for pS6 and pAkt were then generated. Median fluorescence intensity (MFI) of stimulated minus unstimulated cells are shown with standard errors. The experiments for patient NK1 were performed twice and a representative result is shown.

Patients with gain-of-function mutations in the catalytic subunit p110δ of PI3K have increased activation in the PI3K/Akt/mTOR pathway and have been successfully treated with sirolimus, an mTOR inhibitor [8]. Therefore, we treated patient NK1 with oral sirolimus. After treatment for 20 weeks, reduced phosphorylation of S6 was observed in the patient’s T and B cells; however, phosphorylation of S6 remained elevated in her NK cells (Figure 2 and Supplementary Figure 5). During a 5-month course of treatment, the patient’s NK cell numbers and percentages, liver function tests, and level of EBV in the blood remained elevated; therefore, sirolimus was stopped.

Figure 2.

Sirolimus treatment of patient NK1 results in decreased phosphorylation of S6 in T and B cells, while S6 phosphorylation remains elevated in natural killer (NK) cells. Peripheral blood mononuclear cells from patient NK1 and a healthy control were analyzed by flow cytometry and gated to identify T (CD3⁺, CD20^–), B (CD3^–, CD20⁺), and NK (CD3^–, CD56⁺) cell subsets. Antibody to phosphorylated S6 (S240) was used for staining cells. The difference in median fluorescence intensity (MFI) between stimulated and unstimulated cells was calculated for pS6 and standard errors are shown. The experiment was performed 3 times and the means and standard error are shown.

Increased Phosphorylated and Total STAT1 in Cells From Patients With NK Cell CAEBV

The JAK/STAT pathway regulates many aspects of NK cell activity, from proliferation and maturation to cytotoxic activity [11]. Therefore, we evaluated STAT activity in our patients. Increased phosphorylation of STAT1 in NK cells following stimulation with IFN-α in patient NK1 and increased phosphorylation of STAT1 after stimulation with IL-21 was observed in NK cells from patients NK1 and NK2 (Figure 3A). STAT1 phosphorylation was increased in CD4 cells following stimulation with IFN-α or IL-21 in patients NK1 and NK2, and in monocytes following stimulation with IFN-α or IFN-γ. Increased phosphorylation of STAT1 has previously been described in patients with mutations in STAT1 leading to a gain-of-function phenotype [12, 13]. In addition to the increased phosphorylation of STAT1 in these gain-of-function patients, there is also an increase in levels of total STAT1 [14]. We found a marked increase in the level of STAT1 protein in patients NK1 and NK2, and a moderate increase in STAT1 in the mother of patient NK1, compared to healthy controls (Figure 3B). Furthermore, STAT3 expression was not elevated in the patients compared to healthy controls (Figure 3C), whereas STAT4 expression was markedly increased in the patients, and to a lesser extent in the mother and father of patient NK1 (Figure 3B). STAT 5 expression was increased in patient NK2 (Figure 3C). Surprisingly, despite the increased phosphorylation and the increased expression of STAT1 in patients NK1 and NK2, we did not identify any mutations in STAT1 or JAK/STAT regulatory proteins (SLIM, SOCS1/3, PIAS1/4, Ube1L, Ubp43, TCPTP, PTPN2/11, BRCA1, MCM5, and PRMT) that would account for the phenotype. This indicates that another mechanism is responsible for the increased phosphorylation and protein levels of STAT1 in our patients compared with prior patients reported with gain-of-function mutations.

Figure 3.

Increased STAT1 phosphorylation and total STAT1 in patients NK1 and NK2. A, Peripheral blood mononuclear cells (PBMCs) from patients or healthy controls were stimulated with interferon alpha (IFN-α; 10 ng/mL), interleukin 21 (IL-21; 50 ng/mL), or interferon gamma (IFN-γ; 10 ng/mL) for 20 minutes and analyzed by flow cytometry to assess STAT1 phosphorylation. The CD56⁺, CD4⁺, and monocyte populations were identified and median fluorescence intensity (MFI) for p-STAT1 Y701 was determined. The experiment was performed twice and a representative result is shown. B, Total levels of STAT1 and STAT4 were determined in PBMCs for patients NK1 and NK2, the mother and father of patient NK1, and 2 healthy controls by Western blot. C, Total levels of STAT3 and STAT5 were measured in PBMCs for patients NK1 and NK2, the mother of patient NK1, and 3 healthy controls by Western blot. Actin served as the loading control.

A JAK Inhibitor Reduces STAT1 Phosphorylation in the NK Cell CAEBV Patients’ Cells

Three JAK inhibitors, tofacitinib, ruxolitinib, and baricitinib have been approved by the US Food and Drug Administration (FDA) [15, 16]. Therefore, we determined if the elevated levels of pSTAT1 would be reduced by a JAK inhibitor and thus these if inhibitors might be a potential treatment for our patients. We pretreated PBMCs with tofacitinib (1 μM), prior to stimulation with cytokines. Tofacitinib reduced STAT1 phosphorylation in CD4 cells, monocytes, and NK cells from patients NK1 and NK2 following stimulation with cytokines to a level similar to that observed unstimulated cells (Figure 4). Importantly, tofacitinib reduced STAT1 phosphorylation in NK cells (Figure 4), whereas sirolimus did not reduce phosphorylation of S6 in NK cells from patient NK1 (Figure 2). JAK inhibitors did not reduce the level of total STAT1 protein in patient NK1 and only slightly reduced the level of STAT1 in patient NK2 PBMCs (Supplementary Figure 6).

Figure 4.

Tofacitinib reduces STAT1 phosphorylation in patients NK1 and NK2. Peripheral blood mononuclear cells from patient NK1, NK2, or a healthy control were pretreated with tofacitinib (1 μM) followed by stimulation with the indicated cytokine for 20 minutes and analyzed by flow cytometry to assess STAT1 Y701 phosphorylation. The natural killer, CD4, and monocyte cell populations were identified and antibody to phosphorylated S6 (Y701) was used for staining cells. The experiments were performed 3 times and representative results are shown. Abbreviations: IFN, interferon; IL, interleukin; MFI, median fluorescence intensity; NK, natural killer.

S6 and Akt Are Activated in NK Cells From Patients With T- and B-Cell CAEBV, but Levels of Total STAT1 Are Not Increased in PBMCs

Since our NK cell CAEBV patients presented with increased Akt and STAT1 signaling, we determined if increased activation of these signaling pathways was also observed in cells from patients with T- and B-cell CAEBV. Phosphorylation of S6, Akt at serine 473, and Akt at threonine 308 was increased in NK cells from 3 patients with T-cell CAEBV and 3 with B-cell CAEBV following stimulation with IL-15 (Figure 5A, top panels). Phosphorylation of pS6 in T cells was similar in B-cell and T-cell CAEBV patients to normal controls, while phosphorylation of Akt at serine 473 and Akt at threonine 308 was increased in T cells from patients with B-cell and T-cell CAEBV (Figure 5A, middle panels). Phosphorylation of S6 and Akt at serine 473 was increased in B cells from patients with T-cell CAEBV, whereas no differences between patients and controls were noted in phosphorylation of Akt at threonine 308 in B cells from patients with T-cell CAEBV or in S6 or Akt in B cells from patients with B-cell CAEBV (Figure 5A, lower panel). Taken together, this indicates that an increase in phosphorylation of S6 and Akt at serine 473 is commonly observed in NK cells from patients with NK, T-, and B-cell CAEBV. While an increase in phosphorylation of Akt was noted in T cells from patients with T-cell and B-cell CAEBV, this was not consistently seen in T cells from patients with NK cell CAEBV.

Figure 5.

Phosphorylation of S6 and Akt are increased in natural killer (NK) cells from patients with B- or T-cell chronic active Epstein-Barr virus (CAEBV), while levels of total STAT1 are not increased in peripheral blood mononuclear cells (PBMCs). A, PBMCs from 3 patients with T-cell and 3 with B-cell CAEBV and a healthy control were analyzed by flow cytometry and gated to identify NK (CD3^–, CD56⁺) (top panels), T cells (CD3⁺, CD20^–) (middle panels), and B cells (CD3^–, CD20⁺) (lower panels), and stimulated minus unstimulated median fluorescence intensity (MFI) values for pS6 and pAkt are shown. Antibodies to phosphorylated S6 (S240) and Akt (S473 and T308) were used to stain cells. The results show the mean with standard error from 3 T-cell and 3 B-cell CAEBV patients. B, Levels of STAT1 were determined in PBMCs from patients with T-cell (T2 and T4) or B-cell (B2 and B4) CAEBV and healthy controls by Western blot. Actin served as a loading control.

To determine if STAT1 was increased in patients with T- and B-cell CAEBV, we measured the level of total STAT1 in these patients. The level of total STAT1 protein in PBMCs from T-cell CAEBV and B-cell CAEBV patients 2 and 4 was similar to controls (Figure 5B). These data indicate that patients NK1 and NK2 have a distinctive phenotype, with increased phosphorylation of S6 and Akt in NK cells as well as increased phosphorylated and total STAT1. In contrast, the B-cell and T-cell CAEBV patients have increased phosphorylation of S6 and Akt in NK cells, but no increase in total STAT1. To determine if other patients with increased phosphorylation of STAT1 also have increased activation of the PI3K/Akt/mTOR pathway, we determined the level of S6 and Akt phosphorylation in patients with STAT1 mutations. These patients have increased phosphorylation of STAT1, due to their mutations leading to a STAT1 gain of function [12, 13]. Basal and stimulated levels of phosphorylated Akt were similar in patients with STAT1 gain of function mutations and the control (Supplementary Figure 7); levels of pS6 were increased in NK cells from the patients compared to the control. This further indicates that the phenotype of increased phosphorylation of STAT1 and activation of the Akt was unique to the patients with NK cell CAEBV.

The NK Cell CAEBV Patients’ NK Cells Have a Receptor Repertoire Consistent With an Immature Phenotype

Patients NK1 and NK2 had markedly increased numbers of NK cells; therefore, we further characterized these cells. We measured the level of several NK cell receptors on the surface of their cells at different stages of differentiation [17–19]. NK cells from patients NK1 and NK2 were identified as CD56^dim and were predominantly CD2⁺, CCR7^–, CD11a⁺, CD11b^–, NKG2A⁺, NKG2C^–, NKG2D^–, and CD57^–, consistent with an immature phenotype (Figure 6A). In contrast, NK cells from most of the normal controls and the family members of patient NK1 had lower levels of CD2 and NKG2A, and higher levels of CD11b and CD57, consistent with a mature phenotype. To assess the later stages of NK cell maturation, the level of KIRs on the surface of the cells was determined in the patients and controls [17–19]. NK cells from patients NK1 and NK2 were negative for all of the KIRs tested—KIR2DL2/L3, KIR2DL3, KIR2DL1/S5, and KIR3DL1—whereas the controls and parents of patient NK1 showed higher levels of these receptors (Figure 6B and Supplementary Figure 8). The brother of patient NK1 had levels of KIR2DL2/L3 and KIR3DL1 within the range seen in controls, but levels of KIR2DL3 and KIR2DL1/S5 were low. The presence of high levels of NKG2A, low levels of CD57, and low levels of KIRs in NK cells from patients NK1 and NK2 indicates that the cells have a block in maturation, which could lead to accumulation of these cells and account for their increased numbers [18].

Figure 6.

Natural killer (NK) cells from patients NK1 and NK2 express receptors that correspond with an immature phenotype. Peripheral blood mononuclear cells from patients NK1 and NK2; the mother, father, and brother of patient NK1; and healthy controls were stained with the indicated antibodies for NK cell markers (A) and NK killer cell immunoglobulin-like receptors (KIRs) (B) and analyzed by flow cytometry to determine the percentage of NK cells expressing each receptor. The experiment was performed twice for each of the patient and control cells, with the exception that cells from the brother of patient NK1 were only tested once.

DISCUSSION

We have identified a novel phenotype in patients with NK cell CAEBV. The patients have increased numbers and percentages of NK cells; EBV is predominantly in their NK cells; they have increased phosphorylation of Akt and S6 in NK cells and increased levels of phosphorylated and total STAT1; and, in the 2 patients tested, their NK cells had an immature phenotype.

We found that the NK cell CAEBV patients had increased activation of Akt and S6. Analysis of T-cell and B-cell CAEBV patients showed increased activation of Akt in NK cells from these patients. Activation of the PI3K/Akt/mTOR pathway has been reported in other viral infections, which could indicate that this is a general response to viral infection [20]. Expression of specific herpesvirus proteins has been shown to activate the PI3K/Akt/mTOR pathway, which aids in virus replication, latency, and reactivation [21]. EBV latent membrane protein 2A can activate Akt [22]. Hyperactivation of PI3K, due to gain-of-function mutations in PIK3CD, has been associated with high levels of EBV DNA in the blood and EBV lymphomas; however, these patients do not have NK cell proliferation [8]. Nonetheless, activation of the PI3K/Akt/mTOR pathway in CAEBV patients could contribute to EBV-driven lymphoproliferation. Inhibitors of the Akt pathway, such as sirolimus, have reduced lymphocyte proliferation in patients with mutations in PIK3CD; however, treatment of patient NK1 with sirolimus did not reduce the number of NK cells. Further investigation into the activation of the PI3K/Akt/mTOR pathway is warranted in patients with B- or T-cell CAEBV.

Patients NK1 and NK2 had activation of STAT1. Mutations in STAT1 have been associated with severe EBV disease, as well as viral, bacterial, and fungal diseases [12, 23]. Interestingly, 1 patient also had liver disease, similar to our patients, but unlike our patients, this patient had reduced numbers of T and NK cells [23]. STAT1 is induced in EBV-transformed B cells, which sustains viral latency and aids in evading immune surveillance [24]. Furthermore, EBV nuclear antigen 1 and BS-MLF1 both induce STAT1 expression, while EBV latent membrane protein 1 induces STAT1 phosphorylation [25–27]. STAT1 is also activated by heat shock proteins in EBV-transformed cell lines [23]. STAT3 is constitutively activated in cell lines and peripheral blood cells from patients with CAEBV; no mutations were found in the STAT3 gene [28]. JAK inhibitors have been used to treat patients with STAT1 gain-of-function mutations. Two patients with STAT1 gain-of-function mutations and increased phosphorylation of STAT1 were successfully treated with the FDA-approved JAK inhibitor, ruxolitinib [29, 30]. Surprisingly, none of our patients had mutations in STAT1. In our patients the JAK inhibitor, tofacitinib, reduced STAT1 phosphorylation in NK cells in vitro at a concentration (1 μM) that is achieved in the serum of volunteers treated with the drug [31]. This suggests that tofacitinib might serve as a potential treatment for these patients.

Patients NK1 and NK2 had NK cells that were CD56^dim, expressed low levels of KIRs, and high levels of NKG2A, which are also seen in patients with EBV infectious mononucleosis and persist for up to 6 months after the onset of disease [17]. In patients with EBV infectious mononucleosis these cells are actively proliferating (Ki67⁺) during the first month of disease, although they do not reach the levels observed in patients NK1 and NK2. Furthermore, the proliferation of CD56^dim NKG2A⁺ KIR^– NK cells positively correlated with high levels of EBV DNA in circulating cells during infectious mononucleosis [17], and EBV lytic replication of B cells was shown to trigger proliferation of NKG2A⁺, KIR^– NK cells [18]. A paucity of CD56^bright cells has been seen in GATA2 deficiency [32] and patients with this genetic disorder can develop severe EBV disease [33]. More recently, KIRs have been shown to enhance control of persistent virus infections in humans by strengthening CD8 T-cell responses [34].

In summary, we show that NK cell CAEBV patients have a unique phenotype with increased activation of the PI3K/Akt and STAT1 signaling pathways with immature NK cells that express low levels of KIRs.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Elaine Jaffe and Stefania Pittaluga for reviewing pathology and John O’Shea for tofacitinib.

Financial support. This work was supported by the Intramural Research Programs of the National Institute of Allergy and Infectious Diseases and the NIH Clinical Center.

Potential conflicts of interest. All authors: No reported conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.