Differentiation of naive cord-blood T cells into CD19-specific cytolytic effectors for posttransplantation adoptive immunotherapy

Banked unrelated umbilical cord blood (UCB) is source of hematopoietic stem cells for patients with B-lineage acute lymphoblastic leukemia (B-ALL).¹  However, despite maximally intensive preparative regimens, disease-relapse remains a significant cause of mortality after umbilical cord-blood transplantation (UCBT).

Adoptive therapy after allogeneic hematopoietic stem cell transplantation (HSCT) with ex vivo–expanded donor-derived tumor-specific T cells might be used to augment the graft-versus-leukemia (GVL)–effect, thereby reducing the incidence of leukemic relapse, without exacerbating graft-versus-host disease (GVHD).^2,3  While the feasibility of isolating, expanding, and infusing antigen-specific αβ T-cell receptor (TCR)⁺ T cells from peripheral blood (PB) has been validated in animals and humans,^4-10  the naive function of neonatal T cells precludes a priori identification of resident tumor-specific T cells. Redirecting the specificity of T cells through enforced expression of antigen-specific immunoreceptors and differentiating UCB-derived T cells into cytotoxic T lymphocytes (CTLs) is one approach to overcoming this lack of endogenous tumor-specific T cells specific for desired targets.^11,12  We and others are developing adoptive immunotherapy platforms using single-chain chimeric immunoreceptors to redirect the specificity of primary human T cells and NK cells for cell-surface proteins expressed on tumor targets, such as the B-lineage–specific antigen CD19, a molecule expressed on normal and neoplastic B cells.^13-22  These chimeric immunoreceptors typically fuse extracellular single-chain antibodies to the intracellular domain of the CD3-ζ chain or FcR-γ chain and are distinguished by an ability both to bind antigen and to transduce T-cell activation signals.²³  These immunoreceptors are “universal,” as they bind antigen in an HLA-independent fashion; thus, one receptor construct can be used to treat a population of patients with antigen-positive tumors.

To determine whether genetic introduction of a single-chain chimeric immunoreceptor successfully overcomes the naivety and immaturity of UCB-derived T cells, we generated T cells from UCB that expresses a CD19-specific chimeric immunoreceptor designated CD19R.

The plasmid CD19R/HyTK-pMG has been described previously.¹⁷  The plasmid HyTK-pMG,²⁴  which expresses the bifunctional fusion gene of hygromycin phosphotransferase and HSV-1 thymidine kinase (HyTk),²⁵  was modified to express a trifunctional fusion gene composed of North American firefly (Photinus pyralis) luciferase (ffLuc) fused to HyTK (ffLucHyTK). The ffLucHyTK transgene was cloned into HyTK-pMG to create the plasmid ffLucHyTK-pMG. The chimeric receptor gene CD19R was inserted to create the plasmid CD19R/ffLucHyTK-pMG (Figure 1).

Nucleated cells (5 × 10⁷) were obtained from anticoagulated umbilical cord blood by density-gradient centrifugation over Ficoll-Paque-Plus (Pharmacia Biotech, Piscataway, NJ) and stimulated on day 0 with 30 ng/mL OKT3 (Ortho Biotech, Raritan, NJ). rhIL-2 at 25 U/mL was added on the following day. On day 3 the cells were genetically modified by electroporation with Pac I–linearized CD19R/HyTK-pMG or CD19R/ffLucHyTK-pMG using previously described methods.^17,21  Beginning 2 days after electroporation, hygromycin B (Stratagene, La Jolla, CA) was added to cultures at a cytocidal concentration of 0.2 mg/mL active drug. Daudi²⁶  lymphoma cells were genetically modified to express the ffLuc gene by electroporating with the plasmid ffLucZeo-pcDNA, as previously described, and propagated in a cytocidal concentration (0.2 mg/mL) of zeocin (InvivoGen, San Diego, CA).²¹  U251T were genetically modified to express the truncated CD19 (tCD19)²⁷  gene by lipofectamine transfection with the DNA plasmid pCI-rCD19-neo (kindly provided by Dr Michio Kawano, Yamaguchi University School of Medicine, Japan), propagated in a cytocidal concentration (0.25 mg/mL) of G418 (Calbiochem, San Diego, CA) and sorted for high expression of CD19.²¹  8 × 10⁶ MHC class I^–/II^– CD19^–K562^28,29  erythroleukemia cells were genetically modified by electroporation in hypo-osmolar buffer with plasmid tCD19/ffLucHyTK-pMG to express tCD19. On day 3 after transfection, the cells were selected in cytocidal concentration (0.4 mg/mL, active drug) of hygromycin B, sorted for high expression of tCD19, and cloned.

Lymphoblastoid cells (LCLs), CD19⁺ Daudi, and CD19^–K562 cells were maintained in RPMI 1640 (Irvine Scientific, Santa Ana, CA) supplemented with 2 mM L-glutamine (Irvine Scientific), 25 mM HEPES (Irvine Scientific), 100 U/mL penicillin, 0.1 mg/mL streptomycin (Irvine Scientific), and 10% heat-inactivated defined fetal calf serum (FCS) (Hyclone, Logan, UT), referred to as culture media (CM). U251T human glioblastoma cells were cultured in Dulbecco modified Eagle medium (Irvine Scientific) supplemented with 10% heat-inactivated FCS and antibiotics as described for CM. UCB was obtained after informed consent (in accordance with the Helsinki protocol) after severing of the umbilical cord or from the discarded tubing set after completion of UCBT. A modification of the Riddell et al method³⁰  was used to propagate the genetically modified UCB-derived T cells. Two-week stimulation/expansion cultures were established using 10⁶ T-cells, 30 ng/mL OKT3, 5 × 10⁷ (fresh) or 10⁸ (frozen) γ-irradiated (3500 cGy) peripheral-blood mononuclear cells (PBMCs), obtained from healthy volunteer donors under informed consent, and γ-irradiated (8000 cGy) LCLs at a final 5:1 PBMC/LCL ratio in CM. rhIL-2 was added to culture at 25 U/mL every 48 hours, beginning on day 1 of each 14-day expansion cycle. Hygromycin B at 0.2 mg/mL was added on day 5 of subsequent stimulation cycles. For large-scale expansions, cells from multiple flasks were pooled and transferred to a LifeCell bag (Miltenyi, Auburn, CA). Beginning 14 days after electroporation, some T-cell bulk cultures were cloned by limiting dilution. Genetic modification and expansion of CD19-specific T cells from adult PB was described previously.¹⁷ 

On day 0, 0.5 × 10⁶ UCB-derived T cells were stimulated with γ-irradiated 50 × 10⁶ PBMCs and 10⁷ LCLs in 25 mL CM supplemented with 30 ng/mL OKT3. Two millileters of cell suspension was plated per well (4 × 10⁴ T cells/well) in 5 μM GCV (Hoffmann-La Roche, Nutley, NJ). Then, 25 U/mL of rhIL-2 was added beginning on day 1 and repeated every 48 hours. Total and viable cells (trypan blue exclusion) were counted after 14 days.

Expression of the chimeric 66-kDa CD3-ζ was accomplished using a mouse anti–human CD3-ζ mAb (BD Biosciences, San Jose, CA) under reducing and nonreducing conditions, based on methods previously described.¹⁷ 

The following fluorescein isothiocyanate (FITC), phycoerythrin (PE), and PE-Cy5 (CyChrome)–conjugated reagents were obtained from BD Biosciences: anti-TCRαβ, anti-CD2, anti-CD3ϵ, anti-CD4, anti-CD8α, anti-CD11a, anti-CD18, anti-CD19, anti-CD25, anti-CD26, anti-CD27, anti-CD28, anti-CD45RA, anti-CD45RO, anti-CD49a, anti-CD50, anti-CD54, anti-CD58, anti-CD69, anti-CD70, anti-CD80, anti-CD122 (IL2-Rβ), anti-CD132 (IL-2Rγ), anti-CD95L (FasL), anti-CD95 (Fas), anti–HLA DR, antiperforin, and anti–granzyme A. PE-conjugated anti-NKG2D (clone 149810), anti-CCR7, and anti–IL-7Rα were from R&D Systems (Minneapolis, MN). F(ab')₂ fragment of FITC-conjugated goat anti–human Fcγ (catalog number 109-096-008, Jackson Immunoresearch, West Grove, PA) was used at 1/20 dilution to detect cell-surface expression of CD19R. Purified goat antibody (Jackson Immunoresearch) was used as a nonspecific control. In some experiments, CyChrome-conjugated mAbs were replaced with 1 μg/mL propidium iodide (PI) to exclude nonviable cells from analyses. Intracellular staining was performed on fixed and permeabilized cells per manufacturer's instructions (Cytofix/Cytoperm and Perm/Wash solutions, BD Biosciences). Data acquisition was on a FACS Calibur (BD Biosciences) analyzed using CellQuest version 3.3 (BD Biosciences). Fluorescent activated cell sorting was performed using a MoFlo MLS (Dako-Cytomation, Fort Collins, CO).

CD8⁺ UCB-derived responder T cells (10⁶) were cocultured at a 1:1 ratio in 12-well tissue culture plates with γ-irradiated (8000 cGy) CD19^– (parental) U251T or tCD19⁺ U251T stimulator cells in CM. After 12 hours at 37°C, the conditioned media was assayed by cytometric bead array (CBA) per manufacturer's instructions (BD Biosciences) on a FACS Calibur.

The cytolytic activity of T cells was determined by 4-hour chromium release assay (CRA) as previously reported using effector to target (E/T) cell ratios of 50:1, 25:1, 5:1, and 1:1.^17,21  Data are reported as mean ± the standard deviation (SD).

Imaging was undertaken at 37°C on 4 identically equipped Eclipse TS100 microscopes with Plan Fluar 4 ×/0.13 numeric aperture PhL DL phase contrast objective lenses (Nikon, Melville, NY) coupled to Sanyo video CCD cameras (Sanyo, Chatsworth, CA), digitized at 640 × 480 pixels with a Matrox frame grabber board (Matrox, Quebec, Canada). 0.2 × 10⁶ T cells in 200 μL Hanks balanced salt solution, supplemented with 0.1% human serum albumin, were added to T-25 flasks containing adherent parental U251T or tCD19⁺ transfected U251T cell lines, plated 4 hours prior to assay at 0.5 × 10⁶ cells/flask. U251T cells were chosen as targets based on an ability to identify living, dividing, and dying/dead cells by phase contrast dynamic morphology. Adherent U251T cells are spread, except during division when they round up, become figure eights, and divide. Dying cells round up and implode. Acquisition rate was at 4 frames per minute. MetaMorph version 6.31 (Molecular Devices/Universal Imaging, Downing-town, PA) was used to mark events as T-cell engaging or disengaging from a tumor cell, tumor-cell death, and tumor-cell division. We quantified on/kill/off/divide events from the first approximately 900 frames after inspecting all 4799 frames.

ffLuc transgene bioluminescence activity was measured from 10⁵ cells with 0.14 mg/mL D-luciferin (Xenogen, Alameda, CA) using a Victor2 Luminometer (PerkinElmer, Boston, MA).

Six days prior to experimentation, 10- to 14-week-old female NOD/Scid (nonobese diabetic, severe combined immunodeficient, NOD/LtSz-Prkdc^Scid/J) mice (Jackson Laboratory, Bar Harbor, ME) were γ-irradiated to 2.5 Gy using a ¹³⁷Cs-source (JL Shepherd Mark I Irradiator, San Fernando, CA) and maintained under pathogen-free conditions at the City of Hope Animal Resources Center. On day –5, the NOD/Scid mice were injected in the peritoneum with 10⁶ ffLuc⁺ CD19⁺ Daudi cells. Beginning on day –4, tumor engraftment was evaluated by bioluminescent imaging (BLI). Mice with progressively growing tumors were segregated into 2 treatment groups. One group received 3 doses, in 3-day intervals, of a 50 × 10⁶ UCB-derived CD8⁺ CD19-specific T-cell clone by intraperitoneal injection. The other group received no T cells. Both groups received rhIL-2 (25 000 U/mouse) every other day by intraperitoneal injection, beginning the day of T-cell injection. Animal experiments were in compliance with protocols approved by the institutional animal care and use committee.

On day –7, NOD/Scid mice were sublethally γ-irradiated (2.5 Gy). On day 0, 10⁷ γ-irradiated (80 Gy) Daudi cells were injected into the peritoneal cavity. Three hours later, the mice received by intraperitoneal injection 10⁷ cells of an UCB-derived CD8⁺CD19R⁺ffLuc⁺HyTK⁺ T-cell bulk line. That same day, the T cells were imaged, and the mice were segregated into 2 treatment groups so that the average T-cell–derived ffLuc activity was approximately the same between the groups. Beginning on the day of adoptive transfer of T cells, one group of mice received 0.125 mg in 200 μL by intraperitoneal injection of GCV (diluted in PBS) twice a day for 8 days. The other group received intraperitoneal injections of 200 μL PBS in a parallel schedule. All mice received rhIL-2 (25 000 U/mouse) on days 0 (after T-cell injection), 2, 5, 7, and 9 by intraperitoneal injection. T-cell persistence was evaluated by serial biophotonic imaging (see next paragraph). Animal experiments were in compliance with protocols approved by the institutional animal care and use committee.

The in vivo luciferase activity was noninvasively detected in NOD/Scid mice using an intensified CCD camera (biophotonics), based on previously published methods.²¹  In brief, up to 6 anesthetized mice were simultaneously imaged. One unmanipulated mouse in each imaging group was injected D-luciferin to determine the background bioluminescence. The tumor-derived and T-cell–derived ffLuc-activity was measured using an IVIS 100 series system (Xenogen), beginning approximately 15 minutes after intraperitoneal injection of 150 μL (4.29 mg/mouse) D-luciferin potassium salt (Xenogen). Photons were quantified using Living Image version 2.20 (Xenogen), and the bioluminescence signal was measured as total photon flux normalized for exposure time and surface area and expressed in units of photons/second/cm²/steradian (p/s/cm²/sr). For anatomic localization, a pseudocolor image representing light intensity (blue, least intense; red, most intense) was superimposed over a digital grayscale body surface reference image. WinMorph (www.DebugMode.com) version 3.01 morphing software and Adobe (San Jose, CA) Premiere Pro version 7.0 was used to convert 2 sets of 4 images into a cinematic sequence.

UCB-derived T cells were genetically modified with a nonviral gene transfer system to express the CD19-specific chimeric immunoreceptor (Figure 1A,B) and propagated using a reiterative 14-day numeric expansion process, which can be readily adapted to manufacture T cells in compliance with cGMP for phase 1/2 adoptive immunotherapy trials. The genetically modified UCB-derived T cells expanded on cytocidal concentrations of hygromycin B express CD19R transgene by Western blot analysis, as revealed by an approximately 66-kDa chimeric CD3-ζ protein under reducing conditions (Figure 2A) and by flow cytometry analysis using antibody specific for human Fc (Figure 2B). Western blot analysis under nonreducing conditions reveals an approximately 190-kDa chimeric CD3-ζ protein, which is consistent with oligomerization of CD19R (based on the predicted molecular weight of 69 kDa), favored by the CPSC → CPPC amino acid substitution within the IgG4 hinge-region of the CD19R transgene (Figure 1A).

The redirected functional activity of the ex vivo–expanded genetically modified UCB-derived T cells was determined by examining whether CD19R⁺ T cells could produce a T_c1-type cytokine in response to CD19 stimulation. The CD19-specific T cells were incubated with CD19⁺ tumor cells and, after 12 hours, the conditioned media was analyzed for cytokine content. It was determined that UCB-derived CD19R⁺ T cells specifically released approximately 700-fold more IFN-γ in response to tumor cells bearing CD19 antigen, compared with these same T cells incubated with tumor cells that do not express CD19 (Figure 2C). No significant difference in the pattern of cytokine expression was detected when T cells and stimulator cells were incubated for up to 48 hours (data not shown). We did not detect secretion of IL-2 in response to CD19⁺ stimulator cells (data not shown), which is consistent with a differentiated state of the ex vivo–propagated CD8⁺ CD28^neg T cells stimulated with OKT3 and rhIL-2 (Figure 3) and delivery of an activation signal solely through chimeric CD3-ζ.

Freshly isolated CD3⁺ UCB-derived T cells expressed a naive immunophenotype^31-33  characterized by expression of CD45RA (96%), CD27 (100%), and CCR7 (95%) (Figure 3A-C), whereas T cells directly isolated from adult PB expressed a heterogeneous pattern of staining of these markers. This finding reflects the fact that PB contains a mixture of naive, memory, and effector T-cell subpopulations and is consistent with the observation that the proportion of CD45RO⁺ memory cells in PB increases with age.³⁴  Our ex vivo genetic modification and propagation of both UCB- and PB-derived T cells resulted in differentiation toward a CD8⁺ effector-memory phenotype characterized by up-regulation of CD45RO expression and concurrent down-regulation of expression of CD45RA, CD27, CCR7, CD28, and CD127 (Figure 3A-F) and induction of activation markers CD69, HLA DR, and CD26 (Figure 3G-I). In addition, we observed an increase in the expression of T-cell adhesion/immunomodulatory molecules CD49a³⁵  (VLA-1), CD58, and CD70 (Figure 3J-L). The ex vivo activation and propagation of UCB- and PB-derived genetically modified T cells is dependent in part on the addition of exogenous rhIL-2 and, as expected, the cell-surface expression of the 3 component chains of the T-cell high-affinity IL-2 receptor (IL-2Rα, IL2-Rβ, IL-2Rγ) increased during the period of ex vivo cell culture for both UCB- and PB-derived T cells. Thus, the overall pattern of these flow cytometry determinants is similar on genetically modified ex vivo–propagated T cells derived from both UCB and adult PB.

The decreased incidence of severe GVHD after pediatric UCBT may be due to a reduction in the magnitude of the cytotoxic alloresponse in neonatal T cells,^36-39  and this reduction may compromise the ability of genetically modified UCB-derived T cells to kill tumor targets. Therefore, we determined whether UCB-derived T cells expressing the CD19R transgene could specifically lyse CD19⁺ tumor targets. Four-hour CRAs demonstrated that both bulk and cloned populations of the genetically modified UCB-derived T cells specifically lysed tumor targets expressing CD19 with approximately 70% of CD19⁺ tumor cells lysed at an E/T cell ratio of 50:1 (Figure 4). Specificity for CD19 is indicated by the lysis of CD19⁺ U251T cells but not parental CD19^neg U251T cells (Figure 4A). To determine whether target-cell lysis was due to recognition of allelic HLA variants by endogenous αβ TCR, we expressed tCD19 in K562 cells, which do not express classic HLA class I or II molecules. The genetically modified UCB-derived T-cell line and clone exhibited 5-fold (at E/T of 1:1) greater lysis of CD19⁺ K562 compared with CD19^neg K562 cells, which is consistent with activation of the T cells for cytolysis through introduced chimeric immunoreceptor and not via the endogenous αβ TCR (Figure 4B). ⁵¹Cr-released from CD19^–K562 is most likely due to NK T-cell activity inherent in the genetically modified T cells, since K562 cells are susceptible to NK-mediated lysis. We evaluated whether CD19-specific UCB-derived T cells could lyse primary B-ALL cells (Figure 4C). Four primary B-ALL samples were loaded with ⁵¹Cr and between 40% and 60% of the blasts were lysed at E/T cell ratio of 50:1. The killing of B-ALL blasts was more efficient using a CD8⁺ T-cell clone compared with the T-cell line, reflecting a higher percentage of cells with cell-surface expression of the CD19-specific chimeric immunoreceptor on the clone relative to the bulk population (data not shown).

The T-cell effector-mechanism for the CD19-specific lysis of target cells occurs through granule exocytosis and death-ligand–dependent mechanisms. Flow cytometry was used to determine whether critical proteins of both cytolytic pathways were expressed in genetically modified UCB-derived T cells (Figure 4D). Lymphocytes (containing a mixture of NK and T cells) from unmanipulated UCB and UCB-derived genetically modified T cells could both be demonstrated to be perforin⁺ and granzyme A⁺, but the propagated cells had a relatively decreased expression of perforin. UCB-derived genetically modified T cells expressed Fas, but only very low amounts of membrane-bound Fas ligand (FasL) and tumor necrosis factor–related apoptosis-inducing ligand (data not shown). While genetically modified T cells derived from PB are also Fas⁺ and mostly FasL^–, the level of expression of Fas in UCB- and PB-derived genetically modified T cells was similar, whereas lymphocytes from unmanipulated PB, as previously reported,⁴⁰  had higher expression of Fas compared with unmanipulated lymphocytes from UCB (data not shown). Based on these flow cytometry data, we surmise that the CD19-specific cytolytic activity of UCB-derived T cells appears to be mediated predominately through the perforin/granzyme lytic pathway.

Video time-lapse microscopy (VTLM) allows direct inspection of CD8⁺CD19R⁺ UCB-derived T cells killing CD19⁺ tumor cells and determination of whether T cells can recycle their killing ability. The video microscopy reveals the engagement and disengagement of T cells (small, white, retractile bodies) to adherent spindle-shaped parental U251T tumor cells and U251T genetically modified to express CD19 (Figure 5A). The viability of the tumor cells during the assay can be evaluated by observing their ability to divide, compared to tumor cells undergoing T-cell–mediated destruction that leads to their rounding up and death. We observed that 20% of the CD19⁺ tumor cells are killed by CD19R⁺ UCB-derived T cells, compared with no killing of CD19^neg tumor cells or CD19⁺ tumor cells cocultured with CD19R^neg T cells (data not shown). When killing data are plotted over time, we demonstrate a net loss of CD19⁺ tumor cells cocultured with CD19R⁺ T cells, compared with a net gain of tumor cells in controls, reflecting the ongoing cell division of tumor cells in the control wells (Figure 5B).

T cells transiently scan the surface of dendritic cells for cognate antigen presented by HLA-restricted molecules using an antigen-independent adhesion mechanism with maximal cell-cell conjugation occurring approximately 20 minutes after the initial incubation of T cells with dendritic cells (DCs).^41,42  Based on 3 coculture videos, we determined that the mean residence time a UCB-derived CD19-specific T cell spent on a CD19⁺ tumor cell before that tumor cell was killed was approximately 22 ± 26 (mean ± SD) minutes, with a median of 15 minutes (Figure 5C). This was similar to the residence time the CD19R^– T cells spent on CD19⁺ tumor targets (data not shown). A small number (approximately 3%) of the CD19-dependent killing events occurred soon after (< 2 minutes) a T cell contacted a CD19⁺ tumor cell, 90% of the tumor killing took place in less than 36 minutes, and 5% of the killing occurred after more than 60 minutes of sustained contact (Figure 5C).

Antibody molecules have a higher functional affinity for antigen compared with the binding of αβ T-cell receptor (TCR) to peptide in the context of MHC. Thus, the use of an scFv to direct the specificity of UCB-derived T cells for CD19 raises the possibility that CD19-specific T cells may not be able to recycle their cytolytic activity. However, the pattern of T-cell lytic activity demonstrated by VTLM reveals that between 45% to 67% of the CD19⁺ tumor cells were killed by CD19-specific T cells that had previously killed other CD19⁺ targets (Table 1). T cells that were capable of repetitive killing represented 20% to 30% of the observed T cells. VTLM data that capture these events are shown in Figures 5D and S1, and in Video S1 (see the Supplemental Materials link at the top of the online article, at the Blood website).

We evaluated the ability of UCB-derived CD19-specific T cells to control the growth of established CD19⁺ tumor in NOD/Scid mice (Figure 6). To noninvasively quantify tumor burden before and after adoptive immunotherapy, CD19⁺ Daudi cells were genetically modified to constitutively express the ffLuc transgene. A CD19-specific UCB-derived CD8⁺ T-cell clone was adoptively transferred into mice with growing tumors, and the longitudinal tumor-derived ffLuc-enzyme activity was evaluated in vivo using BLI. The mice in the group that received adoptive immunotherapy experienced a reduction in tumor bulk, with 60% of the mice obtaining a complete remission. All the mice in the control group, which received no cell-based therapy, experienced continued tumor growth. By calculating the cumulative area under the curve (AUC) for each mouse and then applying a permutation test for AUC means across all groups, we demonstrate that the antitumor effect achieved a statistical significance of P = .008.

Adoptive immunotherapy with donor-derived CD19-specific T cells after UCBT is anticipated to improve the GVL effect. However, adoptive transfer of these UCB-derived T cells may cause or exacerbate GVHD. Therefore, to improve patient safety, the CD19-specific chimeric immunoreceptor was coexpressed with the HyTK bifunctional selection/suicide gene in the plasmid CD19R/HyTK-pMG.¹⁷  An in vitro assay was developed to validate that the CD19R⁺HyTK⁺ T cells are sensitive to selective eradication by the prodrug. Using this assay, after 14 days of culture with 5 μM GCV, approximately 95% of UCB-derived T cells (clone and line) expressing the HSV-1–derived TK gene are killed, whereas UCB-derived genetically modified hygromycin-resistant T cells that do not express TK are apparently not eliminated by GCV (Figure 7A). The average number of viable TK⁺ T cells after 14 days of exposure to GCV was 0.43% to 0.93% of the number of cells grown in the absence of GCV. We also examined whether the HyTK⁺ T cells could be eliminated in vivo. To visualize noninvasively the survival of adoptively transferred T cells in mice, a new trifunctional suicide protein was developed fusing ffLuc with HyTK cDNAs to create the fusion gene ffLucHyTK. This multifunction gene was coexpressed along with the CD19R gene in the DNA expression plasmid designated CD19R/ffLucHyTK-pMG (Figure 1C,D). We adoptively transferred the UCB-derived T cells with the phenotype CD19R⁺ffLuc⁺HyTK⁺ into 2 groups of mice (Figure 7B). One group received GCV, and one group received PBS to control for the inherent loss of T cells in this model. The T-cell death was significantly faster in the GCV-treated group, compared with the group that received PBS (P < .005). The in vivo half-life of CD19-specific T cells in mice receiving PBS was approximately 2.1 days, and the half-life of CD19-specific T cells in mice receiving GCV was approximately 0.56 days. Even after stopping GCV the ffLuc-activity from T cells in the treatment group could not be detected above background flux for a further 12 days of imaging (data not shown).

To evaluate the feasibility of generating UCB-derived CD19-specific T cells for clinical infusion after UCBT, we determined whether genetically modified T cells could be generated from small numbers of cord-blood cells remaining in the tubing-set after infusion of the UCB-derived hematopoietic cells. These cells were stimulated with OKT3 and rhIL-2 and, after nonviral gene transfer with the linearized plasmid CD19R/HyTK-pMG and selection in a cytocidal concentration of hygromycin B, the genetically modified T cells could be numerically expanded on average 39 ± 11 (mean ± 1 SD) -fold over a 14-day growth cycle. These propagated T cells successfully expressed the CD19-specific chimeric immunoreceptor and demonstrated redirected specificity for CD19 (Table 2).

The inability of maximally intensive preparative regimens combined with immunologic graft-versus-tumor reactivity to eradicate minimal residual disease (MRD) is a mechanism of treatment failure after UCBT. Targeting CD19⁺ MRD early after HSCT by adoptive transfer of donor-derived tumor-specific T cells is one strategy to consolidate the tumor cytoreduction achieved with preparative regimens and establish tumor-specific immunosurveillance.

While UCB-derived T cells can be propagated ex vivo,^43-49  the functional immaturity of T cells in circulating umbilical cord blood typically prevents ex vivo identification of endogenous leukemia/lymphoma-specific T cells. Therefore, we developed a genetic engineering strategy to introduce a chimeric immunoreceptor in order to generate UCB-derived T cells with desired specificity. We demonstrate that UCB-derived T cells can be rendered specific for CD19, expanded ex vivo, and observed to target CD19 specifically on malignant B cells in vitro and in vivo. Our approach uses methods that are currently in practice at our institution for phase 1/2 T-cell trials and that can be accomplished using small numbers of residual mononuclear cells remaining in the tubing set after infusion of UCB-derived hematopoietic stem cells.

Age-related differences in T-cell differentiation, activation, adhesion, and costimulatory molecules between T cells isolated from neonates and adults influence their relative ability to achieving fully competent activation.^50-53  Although the starting population of T cells derived from UCB exhibited the markers for a naive population, they differentiated to an effector-cell phenotype after genetic manipulation, stimulation, and culture, and exhibited a similar pattern of cell-surface markers as CD19-specific effector T cells derived from adult PB, which are currently being evaluated in clinical trials.

VTLM was used to observe these killing events directly and to establish that UCB-derived T cells can disengage from a target after specific killing and recycle cytolysis. This recycling occurred despite the use of an scFv with a high affinity for CD19 (Ka ∼2 × 10^–9M^–1) compared with the low-affinity (micromolar range) binding of T-cell adhesion/costimulatory molecules.^54,55  The time to achieve CD19R-mediated killing of CD19⁺ tumor cells can be compared to the lytic potential of T cells recognizing antigen through an endogenous TCR. Antigen-specific lysis of target cells, mediated by recognition of αβTCR of HLA and cognate antigen, by CD8⁺ T cells preloaded with cytotoxic granules can occur within 5 minutes, although the killing process can stretch out over hours.⁵⁶  This compares with a mean of approximately 20 minutes for UCB-derived T cells to kill CD19⁺ tumor cells as revealed by morphologic changes. Future experiments using bispecific²¹  T cells coexpressing a TCR with a known specificity as well as CD19R will help determine if this difference in time to kill is due to the type of T cell used or is a function of relative antigen density and/or T-cell physiology.

The redirected cytolytic activity of UCB-derived T cells also was evaluated in vivo. The adoptive transfer of CD19-specific UCB-derived T cells was demonstrated to control tumor growth. Furthermore, the development of multifunction reporter genes,^57,58  such as the trifunctional ffLucHyTK gene, allows BLI to demonstrate that the infused TK⁺ UCB-derived T cells can be eliminated in vivo by GCV-mediated ablation. This elimination may be critical in a clinical setting should adoptive immunotherapy lead to adverse events such as GVHD.

The CD19 molecule is expressed on most ALLs, chronic lymphocytic leukemias, and lymphomas of the B lineage.^59,60  In vitro assays have indicated that progenitor cells of B-ALL express CD19.^61,62  However, recent data have challenged this assertion by demonstrating that (1) sorted CD34⁺CD19^– leukemic progenitors can give rise to B-ALL in NOD/Scid mice and, (2) that approximately 50% of immature CD34⁺CD19^– bone marrow cells from children with high-risk B-cell precursor ALL are marked by chromosomal translocations present in peripheral blasts.^63,64  Refuting these data are possible technical issues associated with down-modulation of CD19 due to in vitro culture of progenitor cells, and when using CD19-specific antibodies to stain cells,⁶⁵  microarray analyses demonstrating that CD34⁺ leukemic B-cell progenitors expressed CD19,⁶⁶  and that for some B-ALLs, leukemia-specific chromosomal translocations cannot be detected in the CD34⁺CD19^– primitive lymphoid-restricted progenitor/stem-cell population.⁶⁷  Additional investigation is needed to clarify this issue.

One of the criticisms of UCBT, compared with other unrelated sources of hematopoietic stem cells, has been that donor anonymity precludes augmenting the GVL effect by infusions of donor-derived lymphocytes.^68-70  The ability to generate B-ALL–specific T cells from UCB by enforcing expression of a chimeric immunoreceptor provides a potential solution to this limitation. Furthermore, the setting of UCBT may be particularly well suited to adoptive immunotherapy, since the slow pace of hematopoietic recovery provides a period of relative lymphopenia in which adoptively transferred T cells might expand using homeostatic mechanisms inherent to recovering the lymphocyte pool, and the relatively low incidence of severe GVHD means that recipients will be less likely to be taking immunosuppressive medications that might impair the function and persistence of the adoptively transferred T cells. In aggregate, the data presented support a clinical trial to evaluate safety and feasibility of adoptive immunotherapy with CD19-specific UCB-derived T cells.

We thank Drs Merrill Lewen and Bryan Jick at Huntington Memorial Hospital (Pasadena, CA) for assistance with procurement of umbilical cord blood, Dr Mark Sherman (COH's Division of Information Sciences) for assistance with molecular modeling, COH's flow cytometry facility (Lucy Brown, Manager), COH's VTLF (James Bolen, Manager), COH's Animal Resource Center under direction of Dr Richard Ermel, COH's Stem Cell Processing Laboratory (Candace Kay, Supervisor) in Department of Transfusion Medicine, Dr Stephen Hunger (University of Florida, College of Medicine, Gainesville, FL) and the Children's Oncology Group ALL cell bank for providing primary ALL samples, Dr Waldemar Debinski (Wake Forest University, NC) for providing U251T cells, and Dr Julie Mayo for editorial assistance.