The Epstein–Barr Virus Lytic Protein BZLF1 as a Candidate Target Antigen for Vaccine Development

The Epstein–Barr virus (EBV) is a B-cell–lymphotropic, γ-herpesvirus that infects greater than 90% of the world's population. EBV causes infectious mononucleosis and has a clear association with several lymphoid and epithelial-derived malignancies, including Burkitt lymphomas, Hodgkin disease, and gastric and nasopharyngeal carcinomas (1). Following primary infection, the virus establishes lifelong persistance within the host memory B-cell compartment utilizing restricted latent gene expression programs that can manifest as either resting state (Lat I) or cellular immortalization and transformation programs (Lat II, III; refs. 2–5). In immune-competent hosts, the outgrowth of EBV-immortalized B cells is prevented through a robust expansion of EBV-specific memory T cells directed toward both lytic and latent antigens (6). EBV-positive individuals that possess either congenital or acquired immune deficiencies are thus highly susceptible to viral reactivation and malignant transformation. Furthermore, EBV-seropositive patients treated with immune suppressive medications following organ/stem-cell transplantation are at significant risk for developing posttransplant lymphoproliferative disorder (PTLD), an often fatal B-cell lymphoproliferatve disease (LPD; ref. 7).

Current treatment options for these patients are limited, and although chemo/immunotherapeutic reagents can induce tumor remission, these therapies often lead to opportunistic infections and short disease-free survival due to ablation of EBV-specific immunity (8). Thus, new therapeutic and preventive approaches are needed that specifically target EBV-infected cells while promoting long-term, protective immunity. The development of adoptive cellular therapeutic strategies has proven to be effective (9–11); however, these options have limited availability, are not yet universally applicable, and may be impractical for preventing EBV-associated diseases in solid organ transplant patients. Alternatively, the development of vaccination strategies to enhance primary and recall memory immune responses is receiving increasing consideration as both therapeutic and preventive approaches to EBV-associated diseases, including infectious mononucleosis (12–15). However, progress in this area has been markedly slow, partly because biomarkers that correlate with host protection are not fully known.

The immediate-early protein BZLF1 is a transcriptional activator that initiates the disruption of latency in EBV-infected cells. Upon expression, BZLF1 binds to the promoters of early EBV lytic genes, triggering a signaling cascade that results in viral DNA replication and virion production (16–19). BZLF1 has also been shown to inhibit antiviral cytokine signaling (20, 21), disrupt CD4⁺ T-cell recognition of MHC-II molecules (22), and promote B-cell transformation and lymphomagenesis in a humanized NOD/LtSz-scid/IL2Rγ^null mouse model of EBV-related disease (23). Thus, BZLF1 expression is critical for viral activation, persistance, and disease pathogenesis, and, from an immunologic standpoint, may represent a prime target of adaptive immune responses. Indeed, BZLF1 is highly immunogenic and elicits robust CD4⁺ and CD8⁺ T-cell responses (6, 24, 25) that dominate the early immune responses in infectious mononucleosis patients and produce a healthy pool of EBV-specific memory T cells (26). Furthermore, we have shown that prevention of EBV-LPD in vivo is associated with the expansion of CD3⁺CD8⁺ T cells specific for BZLF1 (27). To the best of our knowledge, the potential therapeutic and preventative effects of BZLF1 vaccination have not been explored.

Here, we show that BZLF1 is expressed in EBV-LPD and is capable of eliciting a robust, memory T-cell response in healthy individuals. Furthermore, we demonstrate that presentation of full-length BZLF1 by human dendritic cells (DC) promotes the expansion of antigen-specific T-cell immunity and prolongs survival in a humanized, spontaneous murine model of EBV-LPD. These data suggest that BZLF1 is a critical target protein triggering immunosurveillance of EBV-infected cells and should be considered as a candidate for strategies to develop an EBV vaccine.

Deidentified, pathologist-verified EBV-LPD tumor samples were obtained from The Ohio State University Pathology Core Facility (Columbus, OH). Patient and hu-PBL-SCID EBV-LPD tumors were stained with anti-BZLF1 (1:300; Santa Cruz Biotechnology) and visualized by immunohistochemistry (IHC) according to Institutional Review Board (IRB)-approved protocol.

Secretion of human IFNγ was detected with the Human IFNγ enzyme-linked immunosorbent spot (ELISPOT) Kit (Mabtech) using MultiScreen-HA plates (Millipore) as per the manufacturer's instructions. Spots were visualized and counted using an Immunospot Imaging Analyzer (Cellular Technology Ltd.). Spot-forming units were calculated as the ratio between spots formed in response to antigen stimulation and spots formed in response to anti-CD3 mAb stimulation.

Six-week-old C.B.-17 SCID mice were purchased from Taconic Farms and housed in a specific pathogen-free environment. Animals showed no evidence of leaky phenotype (28) as determined by murine IgG ELISA (29). Animal work was approved by The Ohio State University Institutional Animal Care and Use Committee.

mAbs specific for the following antigens were used: anti-human CD3, CD8, and IFNγ (BD Bioscience). Human HLA-B8 tetramers complexed with allele immunodominant peptides for BZLF1 (RAKFKQLL; RAK) and EBNA-3A (FLRGRAYGL; FLR) and conjugated with allophycocyanin (APC) were provided by the NIAID Tetramer Facility and the NIH AIDS Research and Reference Reagent Program (27). Events were collected on an FC500 flow cytometer (Beckman-Coulter) and analyzed with FlowJo v7.6.1 (TreeStar) software.

EBV-transformed lymphoblastoid cell lines (LCL) were derived in vivo by engrafting SCID mice with human peripheral blood mononuclear cells (PBMC) from healthy EBV-positive donors as described (30). Cocultures for ELISPOT and IFNγ flow assays were created in 96-well U-bottom plates by combining irradiated LCLs (14,000 rad) with equal numbers of autologous PBMCs. Cultures were maintained in RPMI-1640 containing 10% FBS supplemented with antibiotics and 1% Glutamax (Life Technologies) in the presence of 10 U/mL IL2 (Prometheus).

The open reading frame sequence of EBV BZLF1 was amplified by RT-PCR using RNA isolated from an EBV-positive LCLs; forward primer: ATGATGGACCCAAACT, reverse primer: TTAGAAATTTAAGAG. The amplified product was cloned into Nde I/Not I sites of the pET26b(+) vector (Novagen), and subsequently introduced into Escherichia coli strain BL26 (Invitrogen) by standard transformation. Expression of BZLF1 protein was induced by 1.5 nmol/L IPTG treatment for 3 hours. In collaboration with ARVYS Proteins Inc., milligram quantities of low endotoxin-rBZLF1 protein (GMP-grade materials) were purified by a method comprising recovery and washing of inclusion bodies, two gel filtration chromatography steps, and filtration through anion-exchange filters. The identity of rBZLF1 protein was confirmed by mass spectrometry (Nano-LC/MS/MS system).

The rAd5F35/BZLF1 adenovirus was generated using the Adeno-X expression system 1 (Clontech Laboratories, Inc.). In brief, BZLF1 cDNA was cloned into Not I sites of the mammalian expression cassette pShuttle2. The I-CeuI/PI-SceI fragment containing the BZLF1 expression cassette was gel-purified and subcloned into I-CeuI/PI-SceI sites of the AdF35 genome (a gift from Dr. Malcolm Brenner, the Baylor College of Medicine, Houston, TX), in which the fiber gene of adenovirus serotype 35 is incorporated into an Ad5 capsid. Finally, Pac-I–digested pAd5F35/BZLF1 DNA was used to transfect HEK 293 cells. The individual viral plaque was isolated, and positive viral clones were used for large-scale production of rAd5F35/BZLF1 stock. The viral titer was determined using the Adeno-X rapid titer Kit (Clontech Laboratories, Inc.).

The rAAV2/BZLF1 virus consists of rAAV terminal repeats flanking a CMV-IE promoter BZLF1 gene expression cassette. To produce the rAAV2/BZLF1 vector, a Hela-derived cell line containing the above sequences, the AAV rep and cap genes, and a neomycin-resistant selectable marker was constructed as described (31). The rAAV2/GFP virus is identical to rAAV2/BZLF1 except the GFP gene is driven by the CMV-IE promoter. Large-scale rAAV2 viral preparation, purification, and quantification were performed as described (32–34).

The human amphotropic-packaging cell line HEK 293T (ATCC) was maintained in a 37°C humidified environment supplemented with 5% CO2 in DMEM media containing 5% FBS and grown to 80% confluence before infection. Following incubation, cells were cultured in the presence of rAAV2/BZLF1 [multiplicity of infection (MOI) = 10], rAAV2/GFP (MOI = 10), rAd5F35/BZLF1 (MOI = 10), or rAd5F35/null (MOI = 10) for 3 hours.

All immunoblotting was performed as described (35), using antibodies directed against BZLF1 (Santa Cruz Biotechnology) and actin (Santa Cruz Biotechnology). Staining of SDS-PAGE gels was performed using Coomassie Brilliant Blue Staining Solution (Bio-Rad).

PBMCs were obtained from healthy donors under an OSU IRB-approved protocol. Following Ficoll–Hypaque (GE healthcare Bioscience AB) separation and incubation of 5 × 10⁶ cells/mL in serum-free media for 2 hours at 37°C, the supernatant containing nonadherent cells was removed. Adherent monocytes were cultured for 3 days with human GM-CSF (50 ng/mL; Peprotech) and IFNα (1,000 U/mL; Schering Corporation) in the presence of rBZLF1 protein or an equivalent amount of BSA (control), or rAd5F35/BZLF1 (MOI = 10) or rAd5F35/null (MOI = 10; Vector Biolabs), or rAAV2/BZLF1 or rAAV2/GFP, respectively (36). DCs were then cocultured with autologous HLA-B8⁺ PBMCs at a ratio of 1:5 in the presence of 10 IU/mL IL2 for 10 days.

The engrafted humanized SCID (hu-PBL-SCID) mouse model of EBV-LPD has been described (37). Six-week-old SCID mice were injected i.p. with 5 × 10⁷ PBMCs that were obtained from a healthy HLA-B8⁺ EBV-seropositive donor. For depletion of endogenous murine natural killer (NK) cells, SCID mice were injected i.p. with anti–asialo-ganglioside 1 antiserum (ASGM-1; Wako Chemicals) 1 day before injection of human PBMCs and every 7 days thereafter for the duration of the study. Engraftment was confirmed by human IgG ELISA as described (37). Intraperitoneal vaccination with DCs previously cocultured with rBZLF1 or control BSA or virally transduced DCs (described above) began immediately following initial engraftment.

For the comparisons of IFNγ secretion in Figs. 2, 3E, and 5A, two sample t tests were used. Log-rank tests were used for the survival analyses in Fig. 5, and Kaplan–Meier survival curves were generated to display the results. All analyses were performed using SAS/STAT software version 9.2 (SAS Institute Inc.).

Recognition of EBV-encoded epitopes by effector lymphocytes is a major component of the adaptive immune system's ability to clear lytic and latently infected B cells. To determine whether BZLF1 is expressed in EBV⁺ malignancies, we performed IHC assays on primary EBV⁺ lymphoma tumor samples using an antibody specific for BZLF1. Of the tumors analyzed, 6 of 8 PTLD samples stained positive for BZLF1 (Fig. 1A and Table 1). Other EBV⁺ lymphomas staining for BZLF1 included lymphomatoid granulomatosis (1/3), diffuse large B-cell lymphoma (5/10), and double-hit lymphoma (1/1). Both EBV⁺ Hodgkin lymphoma (6 cases) and EBV⁺ unclassifiable lymphomas (8 cases) failed to demonstrate any BZLF1 signal. We next determined whether BZLF1 is expressed in the EBV-LPD tumors that spontaneously developed in hu-PBL-SCID mice engrafted from 3 separate donors of various HLA types (30). Consistent with our patient data, all hu-PBL-SCID tumors were positive for the BZLF1 protein (Fig. 1B). These results led us to hypothesize that enhanced recognition of BZLF1 by antigen-specific cellular immunity may provide a protective effect against the development of fatal EBV-LPD in hu-PBL-SCID mice.

We next evaluated the cell-mediated immune responses against BZLF1 in vitro. Antigen recognition by the adaptive immune system, specifically by T-cell subsets, is essential to providing long-term, protective immunity against EBV (6). PBMCs were isolated from 24 healthy, EBV-seropositive donors of various HLA types and infected with adenovirus encoding a recombinant BZLF1 (rAd5F35/BZLF1) or null control (rAd5F35/null) for 18 hours without cytokine treatment. To quantify the immune responses against BZLF1, an IFNγ ELISPOT assay was used to detect the number of cytokine-secreting T cells. The degree of responsiveness, as determined by level of IFNγ secretion, varied with each donor, highlighting the inherent diversity of antigen exposure and recall responses (Fig. 2).

We next examined whether DCs loaded with BZLF1 protein could mediate the expansion of BZLF1-specific memory CD8⁺ T cells in vitro. Milligram quantities of rBZLF1 protein were generated as described above and analyzed for purity by SDS-PAGE analysis (Fig. 3A). Purified rBZLF1 was then loaded onto monocyte-derived DCs at varying concentrations and cocultured with autologous PBMCs from an HLA-B8⁺ donor for 10 days (Fig. 3B). To analyze the ability of rBZLF1 protein to generate BZLF1-specific CTLs, a flow cytometric tetramer assay was used to detect the immunodominant HLA-B8–restricted RAK epitope of the BZLF1 protein (27). BZLF1-presenting DCs generated RAK-specific CD8⁺ T cells in a dose-dependent fashion (Fig. 3C). To assess the functional status of these T-cell subsets, we restimulated the DC-PBMC cocultures at day 13 with autologous LCLs. Restimulation induced a RAK-specific IFNγ response that was mediated by CD8⁺ subsets (Fig. 3D). Thus, BZLF1 antigen exposure by DCs confers specific CD8⁺ T-cell immunity in vitro for PBMCs from an HLA-B8⁺ individual.

To determine whether vaccination with rBZLF1 could prime BZLF1-specific T-cell immunity in vivo, hu-PBL-SCID mice were generated using HLA-B8⁺ PBMCs from a single EBV⁺ donor and injected once with either PBS, GM-CSF (5 μg/mouse), rBZLF1 (100 μg/mouse), or rBZLF1 with GM-CSF as an immune adjuvant (27). Four weeks after vaccination, ex vivo splenocytes were reexposed to rBZLF1-presenting DCs in vitro and assayed for IFNγ secretion by ELISPOT. In comparison with vehicle, vaccination with rBZLF1 alone significantly enhanced BZLF1-specific IFNγ responses (P = 0.0007, n = 3, Fig. 3E), indicating an efficient expansion of functional, BZLF1-specific T cells. Coadministration of rBZLF1 and GM-CSF did not significantly enhance the effect to BZLF1 re-exposure.

We next investigated if viral-mediated delivery of BZLF1 to DCs could serve as an effective vaccine strategy against EBV. We compared the ability of adenoviral and adeno-associated viral gene transfer to generate functional, BZLF1-presenting DCs in vitro. Monocyte-derived DCs were infected with either an adeno-associated virus (rAAV2/BZLF1) or adenovirus (rAdv5F35/BZLF1) encoding recombinant BZLF1 (or control); infection and overexpression of BZLF1 were confirmed in HEK 293T cells (Fig. 4A and C, respectively). Infected DCs were then cocultured with autologous HLA-B8⁺ PBMCs for 7 days and analyzed for RAK-specific CD8⁺ T-cell expansion by flow cytometry. DCs transduced with rAAV2/BZLF1 had a minimal effect on RAK-specific CTL expansion compared with DCs incubated with naked rBZLF1 protein (Fig. 4B); however, rAd5F35/BZLF1-transduced DCs induced a robust expansion of RAK-specific CD8⁺ T cells (Fig. 4D). Importantly, DCs transduced with rAd5F35/BZLF1 showed a greater expansion of antigen-specific T cells compared with that by DCs transduced with rAd5F35-null control or by rBZLF1-presenting DCs (compare Fig. 4B and D).

The hu-PBL-SCID model of spontaneous EBV-LPD has been used to study potential experimental therapeutic strategies for EBV-LPD (37, 38). We examined whether vaccination of hu-PBL-SCID mice with rAd5F35/BZLF1-transduced DCs (or control) would positively enhance cellular immunity and improve survival against fatal EBV-LPD. HLA-B8⁺ PBMCs from a healthy EBV-seropositive donor and 5 × 10⁶ autologous DCs transduced with either rAd5F35/BZLF1 or empty vector (rAd5F35/null) were injected i.p. into SCID mice (n = 10/group). Evidence of human cell engraftment was confirmed by assessing human IgG concentrations in serum. Two weeks after injection, mice in the “booster group” received a second i.p. injection of either rAd5F35/BZLF1-transduced or rAd5F35/null-transduced DCs as a booster. Five weeks after initial vaccination, 3 mice from the booster group were sacrificed and analyzed for immunosurveillance capacity against EBV-LPD. Splenocytes were cocultured ex vivo with autologous LCLs and assayed for IFNγ secretion by ELISPOT. Mice vaccinated with rAd5F35/BZLF1-transduced DCs showed significantly higher responsiveness to LCLs relative to that in vector-control–vaccinated mice (P < 0.0001, n = 3; Fig. 5A). Single vaccination with rAd5F35/BZLF1-transduced DCs showed a trend toward improved survival (P = 0.085, n = 10) at a later stage (Fig. 5B); however, vaccination followed by single booster delivery significantly delayed fatal EBV-LPD (P = 0.014, n = 10; Fig. 5C), with a median survival of 62 days for the rAd5F35/BZLF1 group and 48 days for the rAd5F35/null group.

EBV is the etiological agent of infectious mononucleosis and one of few known human oncoviruses. Although most carriers possess the immune components needed to limit the outgrowth of latently infected B cells, immunocompromised patients, such as those with congenital or acquired immunodeficiencies or those having undergone organ transplantation, are at increased risk for malignant transformation due to diminished EBV-specific immunity. EBV is also associated with a broad spectrum of benign (multiple sclerosis, lupus) and malignant diseases (nasopharyngeal, gastric carcinomas) in various immune-competent populations; thus, strategies to enhance immunity to EBV represent an unmet need. In the current study, we demonstrate that vaccination of hu-PBL-SCID mice against the EBV immediate-early protein BZLF1 can enhance specific cellular immunity and significantly delay the development of lethal EBV-LPD. These protective effects are associated with the expansion and activation of CD8⁺ T-cell subsets that recognize the immunodominant HLA-B8–restricted RAK peptide of BZLF1.

Traditionally, EBV-associated malignancies are thought to be driven largely by the cooperative functions of latent viral gene products. Our data support the emerging hypothesis that BZLF1 gene expression is a prominent factor in the development of EBV-associated malignancy (23). We found that both patient and hu-PBL-SCID EBV-LPD tumors show BZLF1-expressing tumor cells that, when targeted in hu-PBL-SCID mice, delay the development of fatal EBV-LPD. Consistent with this observation, Ma and colleagues (23) recently demonstrated that humanized NOD/LtSz-scid/IL2Rγ^null mice infected with a replication-defective BZLF1-deleted EBV mutant developed B-cell lymphomas less frequently than control virus–infected mice. Because BZLF1 expression has not currently been demonstrated to be sufficient to drive oncogenesis, its expression in malignant cells may provide a mechanism for evading tumor surveillance perhaps by downregulating proinflammatory cytokine signaling (20, 21) or by inhibiting the LMP1-induced upregulation of MHC class I molecules in infected cells (39). Horizontal EBV transmission to bystander EBV-negative B lymphocytes may also confer a distinct survival advantage to EBV⁺ lymphomas by replinishing latently infected B-cell populations (23).

Regardless, given the capacity for BZLF1 vaccination to promote EBV-specific immunity and, in turn, delay the progression of fatal EBV-LPD in hu-PBL-SCID mice, it is reasonable to hypothesize that immunosurveillance of BZLF1 by T-cell subsets is a critical factor in the pathogenesis of EBV-LPD and perhaps other associated diseases. Indeed, we have previously shown that prevention of EBV-LPD in hu-PBL-SCID mice treated with combination cytokine therapy is associated with a marked expansion of BZLF1-specific CD8⁺ lymphocytes (27). Furthermore, we have demonstrated that successful treatment of posttransplant LPD following renal allografting is associated with high frequencies of RAK-specific CD8⁺ T cells (40). Taken together, these findings point to the critical nature of BZLF1 immunosurveillance and the importance of restoring BZLF1-specific immunity in patients at risk or suffering from EBV-associated diseases.

Our vaccination experiment with BZLF1-transduced DCs clearly establishes that BZLF1 vaccination can induce specific cellular immunity capable of recognizing and eliminating EBV-transformed cells in hu-PBL-SCID mice. Indeed, the significant expansion of BZLF1-specific T cells (Figs. 3E and 5A), coupled with the increase in survival following repeated vaccination with BZLF1, suggests that BZLF1-specific T cells are expanding and clearing BZLF1-expressing tumor cells, and is doing so in a chimeric mouse model that is dominated by a dampened human immune response (41). Indeed, all vaccinated animals in our study ultimately developed fatal EBV-LPD. This effect is likely the result of hypofunctioning human T cells in the xenogeneic environment, the selection of viral antigen-loss mutants that have “hidden” BZLF1 from effector lymphocytes (42), and/or the presence of natural viral gene polymorphisms (43) that allow for immune escape. The latter possibility would support the rationale to develop a multivalent vaccine that includes additional target antigens covering a broad spectrum of EBV's genetic programs including proteins uniquely associated with malignancy. However, the single valent vaccine utilized in this study might be sufficient to prevent clinically fulminant infectious mononucleosis that can occur in genetically susceptible individuals (44). We are currently exploring multivalent vaccine approaches, including BZLF1, as one of the vaccine immunogens.

Although recombinant BZLF1 vaccination primed BZLF1-specific T-cell responses in vivo, it is particularly interesting that we did not observe an enhanced effect when GM-CSF was administered as an immune adjuvant (Fig. 3E), especially given our prior findings that combination GM-CSF and IL2 therapy protected hu-PBL-SCID mice from fatal EBV-LPD (27). Indeed, others have demonstrated that GM-CSF therapy enhances vaccine-mediated immune responses and confers protective immunity in vivo via the induction of professional APCs (45, 46). In contrast with these observations, however, Serafini and colleagues (47) have shown that high-dose GM-CSF therapy can impair antigen-specific T-cell expansion through the recruitment of myeloid-derived suppressor cells. Thus, in future vaccine studies with BZLF1, it may be useful to assess the delivery of low doses of immune adjuvants that promote specific cellular immunity while limiting the growth and recruitment of suppressor cell subsets.

Although our immunization strategy described here significantly prolonged survival from fatal EBV-LPD in hu-PBL-SCID mice engrafted with PBMCs from an HLA-B8⁺ individual, we cannot exclude the possibility that BZLF1 vaccination will not be as effective in individuals with other HLA types. Given that recognition of the immunodominant RAK peptide of BZLF1 is HLA-B8–restricted, it remains possible that BZLF1 exposure would produce a comparatively diminished immune response in individuals with other HLA types, especially if additional vaccine components compete with BZLF1 for T-cell receptor binding. Although this study does not directly compare cell-mediated immune responses to BZLF1 vaccination in HLA-B8⁻ versus HLA-B8⁺ individuals, we did observe IFNγ production in donors with a variety of HLA tissue types (Fig. 2). Of note, HLA-B8 individuals tested in this experiment produced modest to low levels of IFNγ, suggesting that non–HLA-related mechanisms are responsible for mediating the immune response to this target protein. Future work will need to directly address distinct differences in EBV antigen responsiveness between individuals with various HLA types.

In summary, we have demonstrated that vaccination with the EBV immediate-early protein BZLF1, a trascription factor that initiates the transition to lytic productive program, can induce a protective effect against the development of fatal EBV-LPD in SCID mice engrafted with PBMCs from an HLA-B8⁺ individual. As the hu-PBL-SCID model described here appears to recapitulate the pathogenesis of human EBV-LPD, our findings support consideration of this vaccination strategy in patients, especially HLA-B8⁺ patients, who are at risk for EBV-LPD (48) and other EBV-associated diseases. Our findings also support the consideration of BZLF1 as a component in a multivalent vaccine that consists of multiple EBV gene products. Efforts to identify other latent and lytic antigens that can be included in the development of a multivalent vaccine are currently under way.

No potential conflicts of interest were disclosed.

Conception and design: T. Liu, X. Zhang, M.E. Lustberg, R.A. Baiocchi

Development of methodology: T. Liu, J.T. Patton, S.L. Garman, G. Lozanski, R.A. Baiocchi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.S. Hartlage, J.T. Patton, S.L. Garman, H. Kurt, G. Lozanski, R.A. Baiocchi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.S. Hartlage, T. Liu, S.L. Garman, X. Zhang, M.E. Lustberg, R.A. Baiocchi

Writing, review, and/or revision of the manuscript: A.S. Hartlage, T. Liu, J.T. Patton, S.L. Garman, X. Zhang, G. Lozanski, M.A. Caligiuri, R.A. Baiocchi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.L. Garman, R.A. Baiocchi

Study supervision: S.L. Garman, R.A. Baiocchi

This work was supported by The Friends of Jason Gould Foundation, the Leukemia Lymphoma Society Translational Research Program (to R.A. Baiocchi and M.A. Caligiuri), the National Institutes of Health (to R.A. Baiocchi and M.A. Caligiuri, 1RC2CA148348-01), and NIH grant T32CA090223.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.