CD34-based enrichment of genetically engineered human T cells for clinical use results in dramatically enhanced tumor targeting

Abstract

Objective clinical responses can be achieved in melanoma patients by infusion of T cell receptor (TCR) gene transduced T cells. Although promising, the therapy is still largely ineffective, as most patients did not benefit from treatment. That only a minority of the infused T cells were genetically modified and that these were extensively expanded ex vivo may have prevented their efficacy. We developed novel and generally applicable retroviral vectors that allow rapid and efficient selection of T cells transduced with human TCRs. These vectors encode two TCR chains and a truncated CD34 molecule (CD34t) in a single mRNA transcript. Transduced T cells were characterized and the effects of CD34-based enrichment of redirected T cells were evaluated. Both CD8(+) and CD4(+) T cells could be transduced and efficiently co-expressed all introduced transgenes on their surface. Importantly, more than fivefold enrichment of both the frequency of transduced cells and the specific anti-tumor reactivity of the effector population could be achieved by magnetic beads-based enrichment procedures readily available for clinical grade hematopoietic stem cell isolation. This CD34-based enrichment technology will improve the feasibility of adoptive transfer of clinically relevant effectors. In addition to their enhanced tumor recognition, the enriched redirected T cells may also show superior reactivity and persistence in vivo due to the high purity of transduced cells and the shortened ex vivo culture.

Fig. 1

Retroviral constructs. a The α- and β-chains of each TCR were linked by the P2A sequence using PCR. Cloning behind the hybrid 5′ LTR, containing the human cytomegalovirus enhancer and promoter fused to MMLV 5′ LTR, created SAMEN-1383IαP2Aβ and SAMEN-HCVcl3αP2Aβ. b IRES/neo^r cassettes were replaced with a truncated version of the CD34 molecule (CD34t) linked to the β-chain of the respective TCR via the T2A sequence. The resulting vectors were designated TIL 1383I TCR/CD34t and HCV cl 3 TCR/CD34t, respectively. LTR long terminal repeat, ψ packaging signal, SD splice donor, SA splice acceptor, α alpha chain, β beta chain

Fig. 2

TIL 1383I TCR/CD34t retrovirus mediates efficient co-expression of CD34t and TIL 1383I TCR β-chain on the cell surface of transduced primary human T cells. Activated bulk PBL were transduced twice with TIL 1383I TCR/CD34t retrovirus or sham transduced and expanded. After 1 week, cells were harvested, beads removed by a magnet and cells cultured overnight. The following day, cells were stained with mAbs specific for CD34, Vβ12 and CD4 or CD8 or the respective isotype controls before exposed to 7AAD and analyzed by flow cytometry. All plots show the relative logarithmic immunofluorescence intensities of FL2 versus FL1 and depict cells gated as viable lymphocytes, CD4⁺ cells or CD8⁺ cells, respectively. The data shown is from one representative experiment of five

Fig. 3

Enrichment of redirected T cells using CD34-based stem cell isolation procedures results in dramatically enhanced antigen specific tumor targeting. T cells transduced with TIL 1383I TCR/CD34t retrovirus were separated in two fractions using Dynal CD34 progenitor cell selection system. After addition of anti-CD34 mAb-coated magnetic beads, unbound cells were collected as the “34t-Negative” fraction. The remaining cells were then dissociated from the beads and collected as the “34t-Positive” fraction. The “Unsorted” control fraction consisted of sham-sorted cells, exposed to the same buffers and procedures but not the anti-CD34 mAb-coated beads. a Cells from each of the three fractions were stained with mAbs specific for CD8, Vβ12 and CD34 or the respective isotype controls before exposed to 7AAD and analyzed by flow cytometry. All plots show the relative logarithmic immunofluorescence intensities of FL2 versus FL4 for the viable lymphocytes. b In parallel, 5 × 10⁴ cells from each fraction were co-cultured overnight with 5 × 10⁴ peptide-pulsed T2 cells, irrelevant gp100_{209–217,210M} (209M) or cognate Tyrosinase_368–376 (YMD), or HLA-A2 positive melanoma 624 MEL cells or matched HLA-A2 negative 624-28 MEL cells. A mixture of PMA and ionomycin was used to induce polyclonal responses, serving as an internal positive control for each of the three T cell populations. The amount of human IFN-γ released into the co-culture supernatants was measured by sandwich ELISA and the respective concentrations of cytokine present in wells containing various combinations of effector and target cells were estimated. Bars represent the mean concentration and error bars the standard deviation between duplicate cultures. The data shown is from one representative experiment of three

Fig. 4

Fluorescence-activated cell sorting of transduced T cells into fractions expressing distinct cell surface levels of CD34t. Activated PBL were transduced once with TIL 1383I TCR/CD34t retrovirus and expanded for 1 week. The beads were then removed and the T cells frozen. Upon thawing, cells were stained with anti-CD34 mAb and 7AAD and viable cells sorted by FACS into four different fractions. The CD34t⁺ T cells were divided down the middle in two equal portions, “34t-High” (cells sorted as the half having highest CD34t expression) and “34t-Low” (the remaining CD34t⁺ cells). In addition, CD34t⁻ cells were isolated as the “34t-Negative” fraction. “Unsorted” cells were 7AAD⁻ lymphocytes FACS isolated irrespective of their CD34t expression. a After overnight culture, cells from each of the FACS fractions were stained with mAbs specific for CD34, Vβ12 and CD8 before exposed to 7AAD and analyzed by flow cytometry. All plots show the relative logarithmic immunofluorescence intensities of FL2 versus FL1 for the viable lymphocytes. b In parallel, 5 × 10⁴ cells from each fraction were co-cultured overnight with 2.5 × 10⁴ peptide-pulsed, irrelevant HCV NS3_1406–1415 or cognate tyrosinase_368–376, T2 cells or 5 × 10⁴ HLA-A2 positive melanoma 624 MEL cells or HLA-A2 negative 624-28 MEL cells. The supernatants were analyzed by sandwich ELISA and the concentrations of IFN-γ in wells containing the respective combination of effector and target cells were estimated. Bars represent the mean concentration and error bars the standard deviation between duplicate cultures. The data shown is from one representative experiment of two

Fig. 5

CD34-based enrichment of the function of redirected T cells transduced with a non-traceable transgene. Activated PBL were transduced once with HCV cl 3 TCR/CD34t retrovirus, expanded for 1 week and separated in two fractions using the magnetic bead-based enrichment system. Cells bound to anti-CD34 mAb-coated magnetic beads were collected as the “34t-Positive” fraction, while the remaining cells constituted the “34t-Negative” fraction. The “Unsorted” control fraction consisted of sham-sorted cells. a Cells from each of the three fractions were stained with mAbs specific for CD34 (or its isotype control), CD8 and 7AAD. All plots show the relative logarithmic immunofluorescence intensities of FL4 versus FL1 for the viable T lymphocytes. b 1 × 10⁵ cells from each fraction were co-cultured overnight with 1 × 10⁵ target cells. Effectors were stimulated with T2 cells pulsed with irrelevant CMV_pp65 peptide or cognate HCV NS3_1406–1415 peptide, or the HLA-A2 positive hepatocellular carcinoma cell line HepG2 with or without a HCV NS3_1406–1415 minigene. A mixture of PMA and ionomycin was used to induce polyclonal responses, which served as internal positive control for each of the sorted T cell populations. The concentrations of IFN-γ in the co-culture supernatants were estimated by sandwich ELISA. Bars represent the mean concentration and error bars the standard deviation between duplicate cultures. The data shown is from one representative experiment of five