HSV-sr39TK Positron Emission Tomography and Suicide Gene Elimination of Human Hematopoietic Stem Cells and Their Progeny in Humanized Mice

The genetic modification of hematopoietic stem cells (HSC) is an attractive approach for the treatment of disease, first demonstrated in primary immune deficiencies (1–3). Transplantation of gene-modified HSCs into patients resulted in long-term correction of disease in the majority of subjects, and paved the way for future applications using viral vectors to modify hematopoietic cells (4). Gene therapy has also proved to be a promising modality for engineered immunity. Preclinical studies and clinical trials that engineered peripheral T cells with cancer-antigen reactive T-cell receptors (TCR) and chimeric antigen receptors (CAR) have achieved tumor regression in patients (5–8). Unfortunately, not all patients developed a lasting and complete response with most demonstrating transient antitumor reactivity. The observation that many patients initially responded with a reduction in tumor burden, yet ultimately relapsed, is hypothesized to be due to the nature of the ex vivo T-cell expansion protocol, which pushes T cells to a differentiation state characterized by robust cytotoxic effector function at the cost of regenerative capacity (9–11). The ability to generate an antigen specific T-cell infusion product with long-lasting in vivo persistence, such as central memory T cells, is an area of active preclinical and clinical investigation (12–16).

HSCs represent the most primitive hematopoietic cells with the greatest regenerative potential, and recent preclinical studies have examined the modification of HSCs for cancer immunotherapy. The introduction of a prearranged TCR to HSCs was first demonstrated in mice (17), and later in humanized mouse models (18–20). These studies demonstrated that engineered HSCs give rise to progeny T cells expressing the introduced transgenic TCR, and are reactive against cells expressing the target antigen. CARs have also been shown useful in the modification of HSCs for therapeutic immunotherapy, specifically against CD19 for B-cell malignancies (21). The duration of de novo T-cell production in this chimeric setting is currently unknown, though clinical evidence supports the notion that HSCs support long-lasting thymopoiesis (22, 23).

The use of strong enhancer/promoter sequences within the vector necessary to achieve therapeutic levels of the introduced transgene can result in activation of proto-oncogenes in proximity of the integration site, and clonal expansion culminating in leukemic transformation of modified hematopoietic cells (24). These events, although rare, mandate the incorporation of safety elements in vector design including insulators (25) or internal promoters with self-inactivating long terminal repeats (LTR) lacking strong enhancers (26–28). An additional concern particular to T-cell immunotherapy is that the introduction of a self-antigen–specific TCR or CAR has the potential to induce an autoimmune reaction. There have been several reports of cytokine storm syndrome after the transplant of CAR-transduced T cells (29, 30), which may benefit from an approach to decrease the number of transgenic cells through the use of a suicide gene. Immunotherapy is designed to focus primarily on tumor-specific antigens, though low level of these antigens may be expressed by normal tissue, leading to unintended off-target reactivity. In clinical trials targeting melanoma by transfer of T cells engineered to express a human TCR against the _27–35MART-1 peptide, acute skin rash and auto-immune vitiligo are often observed because of reaction against normal melanocytes that also express the MART-1 antigen (31). More concerning is the recent report of the death of 2 patients in a clinical trial using autologous T cells modified with an affinity-enhanced TCR against the MAGE3 antigen due to unpredicted reactivity to cardiac Titin (32). The possibility of occult cytotoxicity of the TCR or CAR further supports the inclusion of a method to eliminate gene-modified cells in vivo.

Suicide genes can be incorporated as a safety switch to selectively ablate gene-modified cells during an adverse event. These have been demonstrated in the setting of clonal outgrowth from activation of a proto-oncogene (33) and graft versus host disease (GvHD) and on-target/off-organ cytotoxicity (34). Selective uptake of DNA replication chain terminator drugs by engineered nucleoside kinases, such as native or modified herpes-simplex-virus-thymidine-kinase (sr39TK; ref. 35), initiation of apoptosis mediated by inducible caspase systems by chemical dimerizers (36, 37), or surface proteins designed as antibody targets (38), have all been used to eliminate gene-modified cells. sr39TK (39) is advantageous over other modalities in that it additionally serves as a positron emission tomography (PET) reporter gene, allowing in vivo imaging to noninvasively track gene-modified cells using radiolabeled substrates such as 9-(4-[¹⁸F]-fluoro-3-[hydroxymethyl]butyl)guanine ([¹⁸F]-FHBG; ref. 40). Despite clear potential benefit, the characterization of the utility of sr39TK as both a PET reporter and suicide gene in human HSCs and their progeny has yet to be demonstrated.

Here, we report the use of a lentiviral vector encoding sr39TK to gene-modify human HSCs, demonstrate a lack of developmental skewing due to the transgene; visualization of gene-modified HSCs and their progeny at high resolution serial scans in vivo; and the ablation of gene-modified cells in hematopoietic tissues after a single course of the prodrug ganciclovir as evaluated by biochemical, cell-biologic, and molecular biologic techniques. These results lend support for the inclusion of sr39TK in clinical trials for the modification of HSCs with a cancer antigen–reactive TCR or CAR to both monitor successful transplant and provide a safety-feature allowing the ablation of cells during a serious adverse event.

Cells were thawed in a 37°C water bath and transferred to 50-mL tubes. X-VIVO-15 was added drop-wise with agitation to dilute thawed cell product at 1:10. Cells were spun at 500 × g for 5 minutes and supernatant was aspirated. Cells were resuspended in 50-mL X-VIVO-15 and counted using a ViCELL Cell Viability Analyzer (Beckman Coulter). Cells were spun down at 500 × g for 5 minutes, and supernatant was aspirated. Cells were resuspended in X-VIVO-15 (Lonza) + (50 ng/mL) stem cell factor, (50 ng/mL) Fms-related tyrosine kinase 3 ligand, (50 ng/mL) thrombopoietin, and (20 ng/mL) interleukin-3 (Peprotech) at a density of 4 × 10⁶ cells/mL. Twenty-four–well non-tissue culture–treated plates coated with RetroNectin (TaKaRa) were seeded with 0.25 mL (1.0 × 10⁶ cells) of cell suspension and incubated overnight. The following day, concentrated NY-ESO-1-TCR/sr39TK lentiviral prep was added for a final vector concentration of 1.0 × 10⁸ TU/mL in a final volume of 500 μL X-VIVO-15 + cytokines as described above. Cells were incubated overnight. The following day, cells were collected from wells and rinsed three times in X-VIVO-15 without cytokines. Cells were counted and resuspended at a density of 2.0 × 10⁷ cells/mL in X-VIVO-15 + cytokines as described above.

Humanized mice were generated by the intrahepatic transfer of 1.0 × 10⁶ NY-ESO-1-TCR/sr39TK- or mock-transduced CD34⁺ peripheral blood stem cells (PBSC) to neonatal NSG-HLA-A2.1 mice on days 3 to 5 after birth using a 28G tuberculin syringe (18). Neonates were preconditioned immediately before injection with 100 cGy irradiation from a ¹³⁷Cs source (J.L. Shepherd & Associates, San Fernando, CA). For tissue harvest, animals were euthanized by 5% CO₂ asphyxiation immediately before dissection. Single-cell suspensions of thymus and spleen were prepared by dissociating organs with a 3-mL syringe plunger over 70-μm mesh in FACS buffer (DPBS, 2% FBS, and 2 mmol/L EDTA). Individual bones (femurs, humeri, and sternum) were kept separate to investigate potential differences in marrow spaces by flow cytometry and ddPCR. Marrow spaces were flushed with a 23 G needle through 70-μm mesh. Cells were enumerated and 1 × 10⁶ splenocytes, 1 × 10⁶ cells from the marrow, and 1 × 10⁵ thymocytes were stained with antibodies as described below. Immunologic cytotoxicity assays were performed as previously described (41).

Blood was drawn from the retro-orbital sinus using heparin-coated capillary tubes (Thermo Fisher). The following antibodies [Becton Dickinson (BD)] were used to assess human engraftment: murine CD45-V500 clone 30-F11, human CD45-V450 clone HI30, human CD19-PE-Cy7 clone SJ25C1, human CD3-PerCP clone SK7, human CD4-APC clone RPA-T4, and human CD8-FITC clone HIT8a. Expression of the NY-ESO-1-TCR was determined by binding to a phycoerythrin (PE)-labeled HLA-A2.1 MHC-tetramer loaded with the _157–165NY-ESO-1 SLLMWITQC (Beckman Coulter). Antibodies were added to 80-μL whole blood, incubated in the dark for 30 minutes, RBC lysed with 1-mL FACS Lyse (BD), washed with 3-mL FACS buffer, spun at 500 × g for 5 minutes, and resuspended in 250-μL FACS buffer. Data were acquired on a FACS Fortessa (BD). Analysis was performed on an average of 2,000 to 10,000 hCD45⁺ cells per 80-μL peripheral blood drawn per mouse.

[¹⁸F]-FHBG was synthesized as described previously (42). Mice were injected i.v. with 250 μCi [¹⁸F]-FHBG in 50 to 100 μL, and allowed a 3 hour conscious uptake. Mice were anesthetized with 2% isoflurane for sequential imaging in the Siemens Preclinical Solutions MicroPET Focus 220 and MicroCAT IICT (Siemens). PET data were acquired for 10 minutes and reconstructed with a filtered background projection probability algorithm. PET/CT images were coregistered. Quantification of PET signal was performed by drawing 3D regions of interest (ROI) using AMIDE software (http://amide.sourceforge.net/). Maximum Activity Projections (MAP) were generated for display in figures. The max intensity of the muscle ROI, based on the percentage of injected dose per gram (%ID/g), was subtracted from each hematopoietic niche ROI to normalize for background. Images are presented in false-color volumetric renderings generated in AMIDE.

Descriptive statistics for quantitative variables such as the mean and standard error by experimental groups were summarized and presented. Differences between experimental groups were assessed by unpaired t test (Fig. 2A–G) or pairwise comparison (Fig. 6A–D) within the framework of one-way analysis of variance (ANOVA). To account for variation from individual animals, linear mixed effect models with random intercept (43) were used to evaluate the between-experimental group difference (Figs. 5A–C and 6E and F) as well as pre- and posttreatment difference (Fig. 5A–C). For all statistical investigations, tests for significance were two-tailed unless otherwise specified. A P value less than the 0.05 significance level was considered to be statistically significant. All statistical analyses were performed using SAS version 9.3 (SAS).

To test the function of sr39TK, we generated a lentiviral vector composed of a codon-optimized NY-ESO-1-TCR linked by a 2A cleavage-peptide to sr39TK (ESO/TK) driven by the strong retroviral LTR promoter MSCV (Fig. 1A). Humanized mice were generated by transplanting neonatal NSG-A2.1 mice with ESO/TK transduced CD34-enriched PBSCs via intrahepatic injection (Fig. 1B). At 2 months after transplant, mice were screened by peripheral blood immunophenotyping. Human cell chimerism in the mice was determined by evaluating lymphocytes for human CD45% divided by total (human + murine) CD45%. Human cells were gated into hCD19⁺ B cells and hCD3⁺ T cells, and the CD3⁺ population was subfractioned to CD4 helper and CD8 cytotoxic subsets with the NY-ESO-1 tetramer-binding activity of each assayed (Fig. 1C). The transplant of PBSCs to neonatal NSG-A2.1 mice resulted in human chimerism in peripheral blood beginning at 2 months after transplant. The transduction of PBSCs with an ESO/TK lentiviral vector neither result in a significant change in total human cell chimerism nor alter the composition of human lymphoid cells (Fig. 2A–E and Supplementary Table S1). NY-ESO-1-TCR⁺ cells, identified by costaining with the _157–165NY-ESO-1 HLA-A2.1 tetramer, were only observed in the animals transplanted with gene-modified cells. CD4⁺ T cells bearing NY-ESO-1-TCR were not observed (Fig. 2F). CD8⁺ NY-ESO-1-TCR–bearing cells developed solely in the ESO/TK–transduced group, and 8 out of 15 mice had readily detectable TCR-positive CD8 T cells in the periphery as early as 2 months after transplant (Fig. 2G).

To validate the effector function of NY-ESO-1-TCR–bearing T cells developed in vivo from transduced HSCs, experimental mice were harvested, splenocytes dissociated, and expanded by coculture with artificial antigen-presenting cells loaded with the _157–165NY-ESO-1 peptide. Controls were generated from healthy adult donor peripheral blood T cells activated by CD3/CD28 beads and transduced with the ESO/TK vector or mock transduced. Ex vivo–expanded splenocytes from humanized mice or control human T cells were cocultured with non-HLA-A2.1 (M257) or HLA-A2.1 (M257/A2.1 and M407) patient-derived melanoma cell lines expressing the NY-ESO-1 antigen. ⁵¹Chromium release assays to assess cytotoxicity revealed that humanized mouse–derived T cells killed target cells in an HLA-restricted fashion (Fig. 3A and B), comparable with control normal donor T cells transduced with the NY-ESO-1-TCR (Fig. 3C). Minimal background cytotoxicity in nontransduced donor T cells was observed (Fig. 3D). ELISA assays revealed similar results, with both humanized mouse derived- and healthy donor–transduced NY-ESO-1 antigen-specific T cells secreting the effector cytokine interferon-gamma when cultured in the presence of target cells (Fig. 3E).

A subset of mice were selected for PET imaging studies (nontransduced humanized, N = 3; ESO/TK-transduced humanized, N = 10) based on equivalent human chimerism and lymphocyte composition, with an additional (N = 3) nontransplanted age-matched NSG-A2.1 control animals to examine background biodistribution.

The PET reporter/suicide gene sr39TK is an engineered herpes-simplex-virus-thymidine-kinase with approximately 300× greater affinity for ganciclovir than wild-type HSV-TK (44). The ESO/TK vector was first tested in Jurkat cells in vitro. Cells transduced at a multiplicity of infection (MOI) of 10 and 100 expressed the NY-ESO-1-TCR (Supplementary Fig. S1A), showed selective uptake of [18F]-FHBG (Supplementary Fig. S1B), and were selectively killed by ganciclovir (Supplementary Fig. S1C), confirming the functional activity of the ESO/TK vector.

To test the function of sr39TK as a PET reporter/suicide gene in vivo, we designed an experiment to serially scan humanized mice with the PET reporter [¹⁸F]-FHBG before and after treatment with the prodrug Ganciclovir followed by investigation of cell composition by cell- and molecular-biologic methods (Fig. 4A). Nontransplanted NSG-A2.1 mice and transplant recipients of mock transduced or ESO/TK gene-modified human PBSCs were injected with 250 μCi [¹⁸F]-FHBG and imaged on a Siemens MicroPET scanner followed by CT scan for overlay. Nontransplanted NSG-A2.1 mice were imaged to determine background biodistribution of [¹⁸F]-FHBG, which is known to have a high background in the abdominal area due to the probe elimination through the biliary tree and the gastrointestinal tract in mice (45). As expected, non-humanized NSG-A2.1 mice exhibited predominantly gastrointestinal tract, gall bladder, and bladder signal, with no signal in presumptive hematopoietic niches, or areas of high metabolic activity such as the brain or heart (Fig. 4B). Evaluation of uptake in the spleen was occluded by gastrointestinal signal. Nontransduced humanized mice showed similar background biodistribution of [¹⁸F]-FHBG probe, and lack of hematopoietic niche signal (Fig. 4C). In contrast, mice humanized with ESO/TK–transduced PBSCs exhibited strong signal in hematopoietic compartments (i.e., long bones, skull, vertebrae, and thymus) in addition to background gastrointestinal biodistribution (Fig. 4D). Signal quantitation was performed in Amide software by drawing three-dimensional ROI on individual femurs, humeri, the thymus, and arm muscle (Supplementary Fig. S2A). The maximum %ID/g was determined for each ROI, and muscle was subtracted from hematopoietic niche ROIs to normalize background tissue uptake. Significant accumulation of probe in ROIs was observed in hematopoietic compartments in the ESO/TK-transduced cohort versus the nontransduced humanized group (Supplementary Fig. S2B).

To test the suicide gene function of sr39TK in transduced human cells in vivo, previously scanned nontransduced humanized mice and ESO/TK-transduced humanized mouse cohorts were treated intraperitoneally for 5 days with vehicle or (50 mg/kg) of the nucleoside prodrug ganciclovir, which is converted to a cytotoxic nucleotide when phosphorylated by sr39TK. PET/CT imaging was performed 1 week after the final drug injection to allow ablation of gene-modified cells and clearance of residual ganciclovir. Vehicle-treated ESO/TK mice demonstrated specific uptake in hematopoietic niches in pre- and posttreatment scans (Fig. 5A); however, ganciclovir completely ablated PET signal in posttreatment scans in all hematopoietic niches previously observed to harbor probe accumulation in ESO/TK-transduced humanized mice (Fig. 5B). No difference in signal accumulation was detected in pre- and posttreatment scans in the nontransduced humanized cohort (Supplementary Fig. S2C). Vehicle-treated ESO/TK–transduced recipient mice showed no significant difference in signal accumulation in hematopoietic compartments as determined by pre- and posttreatment scans (Fig. 5C). Ganciclovir-treated ESO/TK–transduced recipient mice showed significant ablation of [¹⁸F]-FHBG PET signal in hematopoietic compartments in posttreatment scans (Fig. 5D). The posttreatment signal of ganciclovir-treated ESO/TK mice was not significantly different than background uptake in nontransduced humanized mice.

Animals were euthanized 1 day after the final scan, and tissues were collected and dissociated. Cell suspensions were enumerated, and allocated for subsequent analyses. Flow cytometry of splenocytes to measure chimerism revealed human cells present in all cohorts; nontransduced humanized, vehicle-treated, and ganciclovir-treated ESO/TK–transduced humanized mice. There was not a significant reduction of human chimerism in ganciclovir-treated ESO/TK mice (Fig. 6A). CD19 B cells and CD3 T cells were detected in all cohorts at endpoint analysis with no significant difference between vehicle- and ganciclovir-treated ESO/TK mice (Figs. 6B and 6C). In contrast, NY-ESO-1-TCR–bearing CD3⁺CD8⁺ T cells were reduced to background levels in the ganciclovir-treated ESO/TK–transduced humanized mice (Fig. 6D).

Quantitation of PET signal and flow cytometric analyses demonstrated ablation of gene-modified cells while sparing non–modified cells. However, cells with low metabolic activity may neither be sensitive to drug selection nor show specific uptake of [¹⁸F]-FHBG. In addition, as surface TCR expression requires coexpression of CD3, flow cytometry is unable to measure the presence of this transgene in non–T cells. To investigate persistence of other gene-modified cells, qPCR was performed to measure the amount of lentiviral vector psi element per human genome in each organ compartment. No vector genomes were detected in nontransduced humanized mice (Fig. 6E). The amount of vector present in the ESO/TK-transduced mice treated with vehicle varied among different animals (mean, 0.918 ± 0.131; range, 0.552–1.72), but was relatively consistent among the different tissues tested for each recipient (Fig. 6F). In the cohort treated with a course of ganciclovir, there was a significant reduction of integrated vector (mean, 0.123 ± 0.131) compared with the vehicle-treated cohort (Fig. 6G; P < 0.001).

Gene therapy using HSCs has proved to be an efficacious treatment for monogenetic diseases, and is currently of interest for immunotherapy applications. Preclinical studies have provided evidence that HSCs transduced to express a transgenic TCR are capable of producing antigen-specific effector T cells in vivo paving the way for a first-in-man study nearing phase I clinical trial (CIRM Disease Team Grant DR2A-05309). However, several questions remain. Enthusiasm for engineered immunity is tempered by the possibility of on-target/off-organ reactivity of the modified cells, and the cautionary tales of clonal outgrowth in HSC gene therapy patients merit the inclusion of safety measures in vector design. The inclusion of a suicide gene could provide a safety switch capable of ablating gene-modified cells in the event of undesirable off-target reactivity or clonal transformation. The ability to noninvasively track gene-modified cells in vivo would allow early detection of successful engraftment, active thymopoiesis, and homing to tumor tissue.

The humanized mouse allows the study of HSCs and development of their progeny in vivo. We used this model system to investigate the potential application of the PET reporter/suicide gene sr39TK in the setting of HSC-based engineered immunotherapy to noninvasively locate and ablate gene-modified cells. We observed no detrimental effect of lentiviral transduction with the ESO/TK vector on the engraftment of PBSCs as evidenced by equivalent human chimerism and lymphoid composition between transduced and mock-transduced cohorts. Detection of gene-modified cells by PET was ubiquitous in ESO/TK–transduced humanized mice (N = 15), though only 8 of 15 (53.33%) had detectable NY-ESO-1-TCR⁺ cells in peripheral blood at 2 months after transplant. Therefore, PET imaging allowed early assessment of engraftment of gene-modified cells before NY-ESO-1-TCR⁺ cells have developed and migrated to the periphery in sufficient numbers for flow cytometric analysis.

A previous report used bioluminescent imaging and the luciferase reporter to visualize gene-modified human HSCs and their progeny residing in hematopoietic niches in a humanized mouse model (46). Our work expands on this pioneering study by using PET imaging, a higher-resolution, directly clinically translatable approach to locate human HSCs in vivo. HSCs modified to express sr39TK were observed in hematopoietic niches, such as the long bones of the arms and legs and the thymus after dosing with [¹⁸F]-FHBG. Strong sternal signal in mice led us to include this hematopoietic niche in our harvests, a practice not routinely performed in humanized mouse studies yet an abundant source of hematopoietic cells. Punctate murine vertebral marking with engraftment of vector-bearing cells (Supplementary Fig. S3) directly demonstrates the high-resolution possible with this imaging technology. The limit of detection using [¹⁸F]-FHBG as a probe with the HSV-sr39TK PET reporter gene was previously determined to be 1 × 10⁶ cells/mm³ (47). The thymus of a well-engrafted humanized mouse is populated by approximately 2.5 × 10⁶ human thymocytes, the majority of which are TCR-positive in transduced cohorts, and is approximately 1 mm³ in volume (E.H. Gschweng; Unpublished Data). In the clinical setting, the number of transduced cells along with the richer soil of a human host for transduced/transplanted human HSCs is likely to result in robust PET imaging in excess of seen in our humanized mouse study.

Although the immunogenicity of sr39TK has been reported in human studies of gene-modified T cells (48, 49), in the setting of gene-modified HSCs, de novo–generated DCs may home to the thymus and induce tolerance to the introduced gene product (50). Currently, only in silico predictive models of human immunogenicity exist, and the only true test is to evaluate the development of an immune reaction to a transgene in clinical trials. Still, there are alternative approaches that do not rely on viral-derived or otherwise xenogeneic reporter genes (37, 51).

Although PET signal was completely ablated after ganciclovir treatment, we detected a small amount of vector-containing cells in harvested hematopoietic compartments by qPCR. This may indicate that some transduced HSCs were ganciclovir resistant and generated new cells after ganciclovir treatment. Longitudinal studies to examine these possibilities in small animals are technically difficult owing to the paucity of human cells generated, though a recent study examining sr39TK mediated ablation of Rhesus macaque HSCs provides evidence that a single round of ganciclovir is sufficient to ablate stem cells (52). The elimination of the majority of modified cells should be sufficient to control major toxicities.

sr39TK allows evaluation of successful engraftment of gene-modified HSCs in vivo with high resolution, and the detection of thymic engraftment indicative of developing anticancer TCR-expressing T cells. It may further be used to examine the homing of gene-modified T cells to intended tumor targets and eradication of disease. In the event of off-target cytotoxicity by engineered T cells, GvHD, or insertional oncogenesis, the suicide gene function of sr39TK could be harnessed to eliminate modified cells while importantly sparing the remaining unmodified graft. Our study supports the hypothesis that a clinical approach to engineered HSC immunotherapy would benefit from the inclusion of an imaging/suicide gene.

No potential conflicts of interest were disclosed.

This article is dedicated to the memory of CW.

Conception and design: E.H. Gschweng, M.N. McCracken, A. Ribas, O.N. Witte, D.B. Kohn

Development of methodology: E.H. Gschweng, M.N. McCracken, M.L. Kaufman, R.C. Koya, A. Ribas, O.N. Witte, D.B. Kohn

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.H. Gschweng, M.N. McCracken, M.L. Kaufman, N. Saini, R.C. Koya, T. Chodon, A. Ribas, O.N. Witte

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.H. Gschweng, M.N. McCracken, X. Wang, A. Ribas, D.B. Kohn

Writing, review, and/or revision of the manuscript: E.H. Gschweng, M.N. McCracken, M.L. Kaufman, R.P. Hollis, X. Wang, N. Saini, R.C. Koya, T. Chodon, A. Ribas, O.N. Witte, D.B. Kohn

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.H. Gschweng, M.N. McCracken, M. Ho, R.P. Hollis, D.B. Kohn

Study supervision: E.H. Gschweng, D.B. Kohn

The authors thank the Broad Stem Cell Research Center (BSCRC) for administrative and infrastructure support, Jessica Scholes and Felicia Codrea of the BSCRC Flow Cytometry Core for technical assistance, Dr. Sam Sadeghi and Jeffery Collins for radiosynthesis of [¹⁸F]-FHBG, and Dr. Waldemer Lando, Dr. David Stout, and Darrin Williams in the Crump Institute (Los Angeles, CA) for Molecular Imaging facility for their technical help with PET/CT scans. The authors also thank the Cell Processing and Manipulation Core in the Translational Cores, Physicians and Nurses at Cincinnati Children's Hospital Medical Center (CCHMC; Cincinnati, OH), and the CCHMC Translational Research Trials Office.

These studies were supported by awards from the National Cancer Institute, NIH (PO1 CA132681). E.H. Gschweng and M.N. McCracken are supported by the BSCRC—California Institute of Regenerative Medicine (CIRM) TG2–01169. A. Ribas is supported by the CIRM New Faculty Award RN2–00902–1. O.N. Witte is an investigator of the Howard Hughes Medical Institute.

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.