Gene Therapy www.nature.com/gt

ARTICLE -based CAR-T cells demonstrate in vitro and in vivo cytotoxicity to MPL positive acute myelogenous and hematopoietic stem cells

1,2 1 1,3 1,2 1,3 1 Jaquelyn T. Zoine , Chengyu Prince , Jamie Y. Story , Gianna M. Branella , Allison M. Lytle , Andrew Fedanov , ✉ Jordan S. Alexander1, Christopher C. Porter1,4, Christopher B. Doering1, H. Trent Spencer1 and Shanmuganathan Chandrakasan 1,4

© The Author(s), under exclusive licence to Springer Nature Limited 2021

While targeting CD19+ hematologic malignancies with CAR T cell therapy using single chain variable fragments (scFv) has been highly successful, novel strategies for applying CAR T cell therapy with other tumor types are necessary. In the current study, CAR T cells were designed using a ligand binding domain instead of an scFv to target stem-like leukemia cells. Thrombopoietin (TPO), the natural ligand to the myeloproliferative leukemia (MPL) , was used as the antigen binding domain to engage MPL expressed on hematopoietic stem cells (HSC) and erythropoietic and megakaryocytic acute myeloid (AML). TPO-CAR T cells were tested in vitro against AML cell lines with varied MPL expression to test specificity. TPO-CAR T cells were specifically activating and cytotoxic against MPL+ leukemia cell lines. Though the TPO-CAR T cells did not extend survival in vivo, it successfully cleared the MPL+ fraction of leukemia cells. As expected, we also show the TPO-CAR is cytotoxic against MPL expressing bone marrow compartment in AML xenograft models. The data collected demonstrate preclinical potential of TPO-CAR T cells for stem- like leukemia through assessment of targeted killing of MPL+ cells and may facilitate subsequent HSC transplant under reduced intensity conditioning regimens.

Gene Therapy; https://doi.org/10.1038/s41434-021-00283-5

INTRODUCTION advantageous because of the immediate availability of known The use of T-cell based therapies for the treatment of cancer has binding domains, enhanced CAR specificity, predictable on–target advanced to engineering T cells to generate profound tumor- toxicities, and lack of immunogenicity [9]. specific immune responses leading to cancer remission in Thrombopoietin (TPO) is a hematopoietic growth factor and patients. Chimeric antigen receptors (CARs) are recombinant natural ligand to c-MPL (myeloproliferative leukemia) receptor [10]. receptors designed to bind tumor antigens, consequently activat- TPO drives and their progenitors proliferation and ing CAR T cells while bypassing major histocompatibility complex differentiation, and facilitates hematopoietic stem cell (HSC) self- recognition and priming [1]. The majority of CARs are designed to renewal and maintenance [11–13]. The liver is the main source of recognize a specific antigen through a single chain variable TPO; however, lower amounts are secreted by the kidney and bone fragment (scFv)- a variable heavy and chain of a monoclonal marrow niche [14]. Activation of c-MPL activates downstream antibody fragment joined by a linker sequence [2]. Four pathways including JAK2/STAT5, PI3K/Akt, and Raf1/MAP kinase, generations of CARs have been designed and tested, of which which are commonly implicated in hematological malignancies the second generation CARs—which contain one costimulatory [13]. MPL is abundantly expressed on minimal residual disease domain and a CD3ζ TCR signaling domain—have proven leukemias, erythropoietic, and megakaryocytic leukemias, which particularly successful in eliminating CD19+ hematological malig- typically have fewer treatment options and afflict a unique nancies [3–7]. population of patients with Down’s syndrome [15–21]. The success of CARs has been restricted due to the limited Acute myelogenous leukemia (AML) is a cancer composed of number of known tumor antigens and available antibodies or abnormal myeloblasts. AML accounts for 20% of pediatric malig- scFvs targeting them [8]. There are known obstacles in the CAR nancies and is the most common leukemia in adults. In AML, MPL field related to CAR design which may cause scFv failure, including expression has been characterized as an adverse prognostic factor length of the hinge and linker region, aggregation, and [22]. Overexpression of MPL correlates with shorter, complete immunogenicity against the scFv [4]. An alternative approach to remission in AML patients, suggesting a persistent and chemoresistant engage a CAR is using a ligand-based binding domain. Using a population of leukemia cells results in relapse [15]. Similar to ligand binding domain to engage tumor cells by T cells can be hematopoiesis, leukemia stem cells (LSC) are capable of self-renewal

1Aflac Cancer and Blood Disorders Center, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA. 2Cancer Biology Program, Graduate Division of Biological and Biomedical Sciences, Emory University School of Medicine, Atlanta, GA, USA. 3Molecular and Systems Pharmacology Program, Graduate Division of Biological and Biomedical ✉ Sciences, Emory University School of Medicine, Atlanta, GA, USA. 4Children’s Healthcare of Atlanta, Atlanta, GA, USA. email: [email protected] Received: 27 July 2020 Revised: 10 July 2021 Accepted: 21 July 2021 J.T. Zoine et al. 2 and propagation of the leukemia [23]. LSCs are characterized by Western blotting CD34+ CD38- expression and have been reported as difficult to target Cells were lysed using Cell Signaling lysis buffer (Danvers, MA). Cell lysates because they are resistant to chemotherapy [24–26]. Also, HSCs and were clarified by centrifugation and protein was quantified using a LSCs depend on the c-MPL/TPO pathway for survival [27, 28]. In Bradford Assay (Bio-Rad Laboratories, Hercules, CA). Equal quantities of addition, AML and other leukemiassuchasmegakaryocyticand protein were loaded, and cell lysates were separated by SDS-PAGE under reducing conditions and transferred to a nitrocellulose membrane (Bio-Rad erythropoietic leukemias, have been reported to have high MPL Laboratories, Hercules, CA). Next, the membrane was incubated with an expression that correlates with CD34 expression [29, 30]. Furthermore, – ζ + anti CD3 mAb (1:500) followed by an HRP-labeled goat anti-mouse IgG patients with MPL LSCs have a worse prognosis for patient outcomes secondary Ab (1:1000). and are resistant to conventional chemotherapies [28]. Taken together, there is an unmet need to target this subset of leukemias as well as Flow cytometry cancer stem cells. For these reasons, we designed a CAR targeting the Cells were washed with phosphate buffered saline (PBS) and centrifuged at MPL receptor. 100 × g. Supernatant was decanted and replaced with the appropriate antibody cocktail in PBS. The antibodies used from BD Biosciences (Franklin Lakes, NJ) include: BUV737 Mouse Anti-Human CD3 (Clone SP34-2), BUV496 MATERIALS AND METHODS Mouse Anti-Human CD38 (Clone HIT2), APC-Cy7 Mouse Anti-Human CD69 Cell lines and cell culture (Clone FN50), PE Mouse Anti-Human CD45, V450 Mouse Anti-Human CD3 HEL (DSMZ, Braunschweig, Germany), K562 (ATCC, Manassas, VA), and 697 (UCHT1), BV605 Rat Anti-Mouse CD16/32 (Clone 2.4G2), BV421 Rat Anti- (ATCC, Old Town Manassass, VA) cells were cultured in RPMI-1640 with Mouse CD150 (Clone Q38-480), PE-Cy7 Hamster Anti-Mouse CD48 (Clone L-glutamine (Corning CellGro, Manassass, VA) and 10% FBS and 1% HM48-1). Antibodies used from BioLegend (San Diego, CA) include: APC Penicillin/Streptomycin. CMK cells (gift from the Petrich laboratory, Emory Annexin V, PE Anti-Human CD110 (S16017E), Propidium Iodide Solution, University) were also cultured under previous conditions except with 20% APC Anti-Human CD38 (HIT2), FITC Anti-Mouse CD3/Gr-1/CD11b/CD45R FBS. Mo7e (DSMZ, Braunschweig, Germany) cells were cultured in (B220)/Ter-119 (“Lineage”), PE Anti-Mouse Ly-6A/E (Sca-1) (Clone D7), APC IMDM (1×) with L-glutamine and 25 mM HEPES, supplemented with 20% Anti-Mouse CD117 (c-) (Clone ACK2), PerCP/Cy5.5 Anti-Mouse CD34 FBS, 1% Penicillin/Streptomycin, and 10 ng/mL of TPO (BioLegend, San (Clone MEC14.7). Cells were analyzed by flow cytometry using an LSRII (BD Diego, CA). Biosciences, Franklin Lakes, NJ).

Primary cells pSTAT5 activation assay Whole blood leukoreduction filters were procured from the American Red To measure the response of leukemia cell lines to recombinant TPO

1234567890();,: Cross. Healthy donor T cells were isolated by negative selection from donor (BioLegend, San Diego, CA), cell lines were stimulated for 45 min with PBMC isolated from leukoreduction filters, as previously described [31, 32]. recombinant or TPO-CAR culture supernatant. After incubation with TPO, PBMCs were isolated after cells were isolated with Ficoll-Paque Premium cells were fixed and permeabilized for flow cytometry. Cells were stained sterile solution (GE Healthcare, Uppsala, Sweden). Leukocytes were washed with anti-hu phospho-STAT5 (Tyr694) clone SRBCZX (Invitrogen, Carlsbad, with PBS and T cells were isolated using EasySep Human T cell Isolation Kit CA). To account for shedding of TPO from the CAR, TPO-CAR or unmodified (Stem Cell Technologies, Cambridge, MA). Immediately after isolation, T cells were seeded at 1.5 × 106 cells/mL and 72 h later, cells were T cells were activated with CD3/CD28 DynaBeads (ThermoFisher Scientific, centrifuged, and the supernatant was collected. Leukemia cells were Waltham, MA) for 24 h. resuspended in 250 μL of conditioned media or 400 ng/mL of recombinant human TPO. Cells were processed to measure pSTAT5 by flow cytometry. To model circulating TPOs ability to block TPO-CAR engagement with Cloning of CAR constructs targets, cancer cells were cultured with 10 μg of human TPO antibody (R&D CAR sequences were cloned into a vector containing the necessary Systems, Minneapolis, MN) to facilitate blocking pSTAT5. Leukemia cells components for lentiviral production. The binding domain of TPO [33] was were additionally incubated with TPO-CAR T cells or unmodified T cells in used as the binding portion for the CAR. The codon-optimized construct addition to an external TPO source and pSTAT5 activation was measured. was designed to contain a CH3 hinge, CD28 transmembrane/costimulatory domain, and CD3 ζ as the signaling domain. The vector was optimized for human cell expression and cloned into an FUGW plasmid. All genes were Cytotoxicity assays obtained by gene synthesis from Genewiz (South Plainfield, NJ). Specific cytotoxicity of TPO-CAR T cells was assessed in co-culture experiments. Target cells (CMK, Mo7e, HEL) were labeled with Violet Proliferation Dye 450 (BD Biosciences, Franklin Lakes, NJ) and cytotoxicity Lentiviral production was assessed in flow cytometry-based cell death readouts. Target cells High-titer, recombinant, self-inactivating HIV lentiviral vector was produced were incubated with T cells at the varied effector to target (E:T) ratios: 0:1, using viral accessory plasmids including packaging plasmids encoding 1:2, 1:1, 2:1, 5:1 for 12 h at 37 oC. Target cell death was analyzed via flow Gag-Pol, Rev, and the VSVG envelope protein and CAR expression cytometry using dead cell stains Annexin V and PI, and effectors were plasmids. The expression plasmid encoding the TPO-CAR and viral analyzed for activation markers CD69 and CD38. Remaining targets accessory plasmids were transiently transfected in 293T-17(ATCC, Mana- were additionally analyzed for MPL surface expression. Antibodies were ssas, VA) cells using a calcium phosphate transfection (Sigma Aldrich, St. incubated for 60 min with shaking at room temperature and data were Louis, MO) method to generate lentiviral vectors. Cells were cultured in acquired after one volume PBS wash. To address the toxicity of TPO-CAR fi DMEM (Thermo Fisher Scienti c, Waltham, MA) supplemented with 10% T cells to non-MPL positive populations of cells target cells, K562 and 697 FBS. Conditioned media was collected for 3 days beginning at 48 h post- cell lines, were labeled with CFSE (ThermoFisher Scientific, Waltham, MA) fi transfection and passed through a 0.45-µm lter. Viral vectors were pooled and MPL+ cells were labeled with VPD450. Briefly, 1 μL of a 1 mM stock and concentrated by overnight centrifugation at 10,000 × g at 4 °C, solution was added to 10 × 106 cells in 1 mL of PBS and incubated at 37 oC fi fi fi − followed by ltration using a 0.22-µm lter lter and stored at 80 °C. for 15 min. Cells were washed with 10x the volume with complete media Viral concentrate titers were determined using quantitative real-time PCR and used in assay. MPL- and MPL+ populations were mixed at a 1:1 ratio analysis. Titers of the concentrated recombinant viral vectors were ~1 × 7 and T cells were added at a 1:2 effector to target ratio cytotoxicity was 10 TU/mL. measured in both populations by flow cytometry. To assess if the presence of circulating TPO inferences with the cytotoxicity of TPO-CAR T cells, we Lentiviral transduction established a competitive cytotoxicity assays, 0.1–400 ng/mL of recombi- Transduction of recombinant HIV lentiviral particles was carried out by nant human TPO (BioLegend, San Diego, CA) was added at the start of the incubating cells with viral vectors in complete medium supplemented with cytotoxicity assay at the same time as TPO-CAR T cells. Cytotoxicity was 8 μg/mL polybrene (EMD Millipore, Billerica, MA). Media was replaced 18 h measured by flow cytometry, as mentioned above. post-transduction. The transduced cells were cultured for at least 5 days before being used in experiments. Real time quantitative PCR Genomic DNA was extracted using the Qiagen DNeasy Blood & Tissue Kit using manufacturer’s recommended protocol (Qiagen, Germantown, MD).

Gene Therapy J.T. Zoine et al. 3 Oligonucleotide primers were designed for a 150 bp amplicon of the Rev- observed in control cell lines K562 (chronic myelogenous response element. Real-time PCR was performed in an Applied Biosys- leukemia) and 697 (B cell leukemia). tems® StepOne™ System (ThermoFisher Scientific, Waltham, MA). To assess MPL expression in normal bone marrow compartments, we evaluated MPL and pSTAT5 signaling following TPO stimulation In vivo mouse experiments in both mouse and human bone marrow samples. Our data show NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ(NSG)micewerepurchasedfrom the more stem-like cells/long-term hematopoietic stem cells (Lin−,c- Jackson Laboratory (Bar Harbor, ME) and were maintained in a kit+,sca-1+,CD48−, CD150+) have greater surface expression of pathogen-freeenvironment.Micewerecaredforaccordingtothe MPL and higher MFIs compared to more-differentiated short-term established principles of the Institutional Animal Care and Use hematopoietic stem cells (ST-HSCs; Lin−,c-kit+,sca-1+,CD48−, Committee (IACUC), and all animal protocols were approved by the CD150−), multipotent progenitors (MPP; Lin−,c-kit+,sca-1+, IACUC. Five-week-old mice were injected tail vein with 5 × 106 CMK CD48+, CD150−), and most differentiated progenitors. In human luciferase cells. Tumor growth and mice health were monitored three + − + − times per week by weighing, IVIS (In vivo Imaging System, PerkinElmer, bone marrow, HSCs (CD34 ,CD38,CD90, CD45RA ) have Waltham, MA) imaging, and bi-monthly complete blood counts. Luciferin greater MPL surface expression as compared to multipotent was made fresh immediately prior to imaging. Luciferin was injected at progenitors (MPPs, CD34+,CD38−,CD90−,CD45RA−), and 150 mg/kg intraperitoneal. Mice were imaged 10 min after injection and hematopoietic progenitors (HPCs, CD34+,CD38+, Supplementary bioluminescence was quantified. Fig. 1D, E). There was a significant difference in MFI by one-way ANOVA in the progenitor (165.2 ± 26.1, P < 0.0001), MPP (570.6 ± Statistical analysis 122.5, P < 0.0001), ST-HSC (1373 ± 234.3, P < 0.0001), and LT-HSC All statistical analysis and graphing were performed using Sigma Plot (2682 ± 253.2) (Fig. 1F, Supplementary Fig. 1F). Both human and version 13 (Systat Software Inc,) and GraphPad Prism. Exact methods are mouse recombinant TPO induced pSTAT5 expression in mouse bone described for each experiment as used. marrow, specifically in the LSK (lineage-, c-kit+,sca-1+) cells (Supplementary Fig. 1G).

RESULTS Development of ligand-based CAR targeting MPL Detection of MPL on LT-HSC and leukemia cells TPO is a 353 amino acid protein. To generate the TPO ligand for Data from the St. Jude PeCan Data Portal [34–40] database the antigen recognition domain, TPO was truncated at the 176 suggest many pediatric malignancies have MPL expression; amino acid position to include the biologically active portion and however, AML stand out as highly expressive (Fig. 1A). Acute the resulting cDNA was cloned into a lentiviral CAR construct megakaryoblastic (M7 or AMKL) and core binding factor (CBF) (Fig. 2A). Amino acid 176 is a glycosylation site and remaining AMLs tend to express higher levels of MPL than other subtypes protein consists of carbohydrate domains for stabilization of TPO (Fig. 1B). Interestingly, adult AMLs do not express the same level of [42]. Further, there is a thrombin cleavage site between amino MPL according to TCGA data (Fig. 1C). It is worthy to mention that acids 191 and 192 which could be detrimental to the shedding of a proportion of many tumors including B cell acute lymphoblastic TPO from the CAR structure [42]. Another source suggests suggest leukemia, T cell acute lymphoblastic leukemia, and neuroblastoma amino acids 1–163 are essential for binding [33]. Therefore, even may be treatable with TPO-CAR, as well, depending on the though the extra 13 amino acids may be irrelevant to binding they percentage of MPL+ surface expression. Current literature high- are essential to extend the antigen binding domain off the cell lights there is surface expression of MPL on patient leukemia membrane. We used a CH3 domain of IgG1 linker sequence, samples [15–21]. To confirm MPL expression on the surface of CD28TM/costimulatory domain, and codon optimized our CAR primary AML, we received two primary samples of PBMCs isolated construct using a custom codon usage bias table. The in silico from AML patients at time of diagnosis from our collaborators. We optimization using a commercial algorithm was made from a confirmed via flow cytometry a significant increase in MPL mean custom table rather than a species-specific or genome-based fluorescence intensity (MFI) compared to isotype control (Supple- tables [43]. Optimization parameters included removing the cis- mental Fig. 1A). Furthermore, MPL has limited expression in other acting motifs, destabilizing RNA structures, and minimizing GC tissues outside of HSCs; unlike other potential target such as c-KIT content. Human T cells were isolated from PBMCs and activated receptor which is more ubiquitously expressed and targeted as a for 24 h. T cells were then transduced and incubated for 18 h. Five mechanism of bone marrow/stem cell depletion (Table 1)[41]. days post-transduction, data demonstrated vector copy numbers RNA expression levels available from the Cancer Cell Line ranging from 0.20 to 0.96 copies per 50 ng of DNA suggesting no Encyclopedia (Table 2) demonstrate high MPL expression on significant difference in the VCN of the TPO-CAR construct HEL (erythroblastic leukemia) and CMK cells (AMKL), and low/ compared to the CD19 CAR control (Fig. 2B). Our efforts for direct undetectable expression on Mo7e, another AML line. We validated detection of the antibody-binding domain were limited by expression data by measuring the presence of surface MPL on absence of sufficient human TPO antibody for flow cytometry. various AML cell lines using flow cytometry (Fig. 1D, Supplemen- We tested recombinant TPO receptor conjugated to fluorophore tary Fig. 1B). Though Mo7e cells have low MPL surface expression, and due to technical difficulties, we could not fully validate the they are dependent on TPO for growth. Competitive inhibition of assay. In addition, secondary detection methods were also MPL antibody binding or downregulation of MPL by TPO could be unsuccessful. Due to these stipulations, we ran VCN and western potential reasons for low MPL surface expression. In addition, it is blot analysis on each T cell product generated. CAR activation also possible the MPL on Mo7e cells have a different splice variant induced by HEL, Mo7e, and CMK cells was measured after 12 h of that commercially available MPL antibody clones are not able to co-incubation by flow cytometry for CD69 and CD38 surface detect it. To verify these cell lines were responsive to TPO, we expression [44–46]. Cells transduced with TPO-CAR showed measured downstream upregulation of pSTAT5. We stimulated significantly higher activation compared to the non-transduced HEL, CMK, Mo7e cells as well as control cell lines K562 and 697 for T cells following incubation with either of the three cell lines (N ≥ 3 45 min with mouse or human TPO (Fig. 1E, Supplementary Fig. 1C). experimental, N ≥ 3 biological, P ≤ 0.001, Fig. 2C–E). TPO-CAR Upon stimulation with TPO, a significant increase in pSTAT5 was lysates were harvested and compared to control CD19 CAR T cells observed in HEL, CMK, and Mo7e cells and each cell line showed a and non-transduced T cells between days 5-7 post-transduction significant increase in MFI (Fig. 1E, P < 0.001, 2-way ANOVA) and CD3ζ expression was detected via western blot with compared to their unstimulated counterparts. Despite minimal endogenous CD3ζ used as a loading reference (Fig. 2F). In surface expression of MPL, Mo7e cells respond to TPO with multiple donors, the TPO-CAR had comparable protein expression downstream STAT5 phosphorylation. Notably, this shift was not to the CD19 CAR.

Gene Therapy J.T. Zoine et al. 4

Fig. 1 Establishing MPL as a target in cancer and stem cell biology. A Data were acquired from the St. Jude PeCan database. MPL RNA sequencing data were taken and formatted to show the expression measured by fragments per kilobase million (FPKM) across multiple pediatric subpopulations including adenocortical carcinoma (ACT), acute myeloid leukemia (AML), B cell acute lymphoblastic leukemia (BALL), choroid plexus carcinoma (CPC), ependymoma (EPD), high grade glioma (HGG), low grade glioma (LGG), medulloblastoma (MB), melanoma (MEL), mixed lineage leukemia (MLL), neuroblastoma (NBL), osteosarcoma (OS), retinoblastoma (RB), rhabdosarcoma (RHB), T cell acute lymphocytic leukemia (TALL), and Wilm’s tumor (WLM). Box and violin plots with minimum and maximum all data points shown are shown with median expression demonstrated by the dotted line for all tumors. B Data from pediatric AML patients demonstrating the acute megakaryoblastic leukemic (AMLM7, N = 102) and core binding factor (CBF, N = 44) leukemias have a higher for MPL compared to the uncharacterized AML population (N = 160). C St. Jude PeCan data portal expression and The Cancer Genome Atlas (TCGA) gene expression AML data sets for MPL expression. Pediatric N = 306, Adult N = 173. D Mean fluorescence intensity (MFI) by flow cytometry of MPL surface expression analysis showed significantly higher expression in the HEL (1008 ± 378.4) and CMK (1330 ± 160.5) cell lines compared to the Mo7e (316.7 ± 6.66) and Loucy (233 ± 8.66) lines. E Cells were stimulated for 45 min with recombinant human thrombopoietin, fixed, permeabilized and evaluated for pSTAT5 expression. Flow cytometry analysis of leukemia cell lines HEL (N = 3), CMK (N = 3), Mo7e (N = 3), and Loucy (N = 3) surface MPL expression show MFI of accompanied pSTAT5 stimulation with thrombopoietin. F Flow cytometry analysis for MPL surface expression was completed on whole mouse bone marrow (N = 13) and separated into progenitor and stem like compartments, with (long-term hematopoietic stem cells (LT-HSC) having the highest MPL surface expression compared to short-term hematopoietic stem cells (ST-HSC), multipotent progenitor (MPP), and progenitors). MFI of the MPL expression in each bone marrow compartment was evaluated.

Specific cytotoxicity targeting of MPL+ cells phenotype. Therefore, there is a threshold of killing that can be In vitro cytotoxicity of the TPO-CAR was evaluated against three reached with these target cell lines. MPL expressing leukemia cell lines as well as MPL negative T-cell Comparison of the 1:1 ratio from multiple donors showed acute lymphoblastic leukemia (ALL) cell line, Loucy, and B-cell consistent killing of the CAR-modified cells compared to controls leukemia cell line, 697. The TPO-CAR T cells significantly killed HEL, (Fig. 3D, Supplementary Fig. 2A, B). Our starting viability of target CMK, and Mo7e cells at effector to target ratios of 1:2, 1:1, 2:1, cells was between 80 and 90% which accounted for ~20% cell and 5:1 compared to non-transduced T cells and CD19 CAR death and there was a corresponding 10–15% increase in cell death T cells without significant cytotoxicity against non-MPL expressing with NT or CD19 CAR T cells. This could be due to background cell leukemia cell lines (N ≥ 3 experimental, N ≥ 3 biological Fig. 3A–C). death elicited by activated T cells or alloreactivity. However, the The effector to target ratios were optimized for CMK cells so it is TPO-CAR induced greater specific cell lysis of CMK compared to the possible if we tested lower ratios there would be a dose response non-transduced T cells and CD19 CAR T cells in all donors. in HEL and Mo7e cell lines. Our surface expression data for each of As a test of the specificity and efficacy,12 h after co-culture these cell lines demonstrates there is a mixed MPL+/− population with TPO-CAR T cells the remaining live target cell fraction of CMK of cells. We had previously done experiments to sort the MPL+ (N ≥ 3 experimental, N ≥ 3 biological, Fig. 3E, F) and HEL (N ≥ 3 population however, the cells revert to a mixed MPL expression experimental, N ≥ 3 biological, Supplementary Fig. 2E, F) were

Gene Therapy J.T. Zoine et al. 5 leukemia cells or cause graft versus host disease. This would Table 1. Comparison of KIT and MPL expression in healthy tissue. present an obvious problem when testing the effectiveness of our Tissue KIT MPL CAR in vivo. Therefore, to test these hypotheses, we first harvested fi Breast 25.5 0.6 media from TPO-CAR transduced T cells and unmodi ed T cells and performed a pSTAT5 activation assay to demonstrate if shed Ovary 15.3 0.3 TPO would cause activation of our leukemia cells (N = 3 Thyroid 12.7 0.3 experimental, Fig. 4C–E, Supplementary Fig. 4). The data show Brain 12.6 0.2 there was a significant activation of pSTAT5 by the T cell media of Salivary Gland 11.4 0.1 the TPO-CAR T cells in the CMK cell line, suggesting that shedding may be a significant problem when utilizing ligand-based CARs. Skin 11 0.1 Thus, shedding of the TPO from the CAR could lead to a Stomach 0.7 0.1 theoretical survival advantage for leukemia cells. Interestingly, we Lung 9.9 0.1 were able to block this induction of pSTAT5 by using an antibody Urinary Bladder 9.5 0.1 to TPO in the CMK cell line (Fig. 4E). The addition of recombinant TPO caused a significant reduction in cytotoxicity at doses of 400 High Low ng/mL in CMK cells when coincubated with the TPO-CAR (P < 0.05) but at physiological doses of TPO (~120 pg/mL), the TPO-CAR significantly kills compared to non-transduced T cells. Table 2. MPL RNA expression from CCLE. Cell Line MPL expression (TPM) Tissue Description Utilizing the TPO-CAR for targeting MPL in vivo We hypothesized that the TPO-CAR may not function well in vivo HEL 45.665 M6 AML JAK2p.V617F due to the variable expression of MPL on our available cell lines. CMK 22.130 M7 AML T21 Further, we demonstrated shedding of the ligand and naturally Mo7e 1.095 M7 AML infant occurring TPO could potentially activate our leukemia cells. Loucy 1.262 T-ALL SET-NUP214 However, we tested the TPO CAR in vivo against the CMK cell line (Supplementary Fig. 5A). NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice were irradiated with 100 Rads and injected screened for cell surface MPL expression. As expected, there was a intravenously with a luciferase positive CMK cell line. Ten days clear decrease in live MPL-positive cell percentages and MFI after after leukemia cells were injected, 5 × 106 non-transduced T cells coincubation with the TPO-CAR compared to targets incubated (N = 4) or TPO-CAR T cells (N = 4) were intravenously injected. The with non-transduced T cells, suggesting TPO-CAR T cells prefer- mice were imaged and weighed regularly to evaluate leukemic entially killed the MPL positive fraction of the leukemia cell lines. progression and overall health. Mice tolerated infusion of the Further, TPO-CAR T cells were mixed with CMK target cells and the modified T cells and showed no signs of weight loss or change in non-MPL expressing cells, K562 and 697 cells, at a 1:1:1 ratio. The activity. Interestingly, mice treated with TPO-CAR T cells and the TPO-CAR T cells demonstrated minimal toxicity against K562 and non-transduced T cells succumbed to sickness between days 34 697, while achieving cytotoxicity against the CMK cell line (N = 3 and 37 due to leukemia burden (Supplementary Fig. 5B). As experimental, N = 2 biological Fig. 3G). This was repeated with the expected, our in vivo model was not sufficient for tumor clearance. HEL target cell line and similar results were achieved (N = 3 To assess the impact the TPO-CAR had on MPL positive fraction experimental, N = 2 biological, Supplementary Fig. 2G). Again, our of leukemia cell lines, we repeated the experiment with a new data demonstrated significant killing of MPL+ cells compared to T cells donor and sacrificed the mice at day 30 to evaluate either MPL-negative cell line. leukemia burden and the bone marrow compartment. Mice were To evaluate for potential toxicity of TPO-CAR to MPL positive euthanized on day 30, after blood was drawn to evaluate overall bone marrow HSPC compartment, we cocultured murine bone health in complete blood counts (Supplementary Fig. 6A–M). marrow HSPC(LSK) with either TPO-CAR T cells or untransduced Spleens were dissociated for flow cytometry analysis. We T cells for 4 h and performed flow-based viability and CFU assay. analyzed the spleens from the leukemia only mice and the mice Our data suggest toxicity of the TPO-CAR to murine HSPC that received the TPO-CAR T cells for the amount of MPL (Supplementary Fig. 2C, D, Supplementary Table 1). expression on the remaining leukemia cells (mice treated with the non-transduced T cells were excluded due to a graft versus host Thrombopoietin impact on TPO-CAR function or graft versus leukemia effect). Leukemia cells were defined by To test the impact of natural TPO could have in our in vivo human CD3- and CD33+. Animals receiving the TPO-CAR modeling, effector cells and target cell lines, HEL, CMK, and Mo7e compared to leukemia only showed a significant reduction in were coincubated both with and without, a supraphysiological MPL surface expression (Fig. 5A–C). These results confirmed level of recombinant human TPO. The addition of TPO significantly our in vitro data demonstrating the TPO-CAR is specificfor impacted cytotoxicity with the TPO-CAR T cells, suggesting MPL+ cells. competition for the engagement of the MPL receptor (N = 3 In addition, we tested the bone marrow for depletion by experimental, N = 1 biological Fig. 4A, Supplementary Fig. 3A, B). measuring the total remaining bone marrow and the LK and LSK The addition of TPO caused a significant reduction in cytotoxicity. compartments in two femurs. There were fewer cells in the The CMK cells cultured with the TPO-CAR with TPO showed remaining bone marrow in two femurs for mice treated with TPO- significantly greater cell death when compared to un-transduced T CAR compared to that with non-transduced T cells (Fig. 5E, P ≤ cell killing. However, even in the presence of TPO, the cytotoxicity 0.001), however, while there is a trend that there were fewer in the with the CAR T cells against the Mo7e and CMK cell lines was still TPO-CAR treated mice compared to leukemia only, it was not significantly greater than cytotoxicity from untransduced T cells (2- significant. These data further suggest all mice receiving leukemia way ANOVA, P ≤ 0.001) and further analysis using a dose response cells had modest reduction in the amount of LK compartment starting at physiological levels demonstrate the TPO-CAR still (Fig. 5E, P ≤ 0.001). However, mice receiving TPO-CAR T cells had a retains significant target cells cytotoxicity even in the presence of greater reduction in the LK and the LSK compartment compared TPO (Fig. 4B, Supplementary Fig. 3C, D). to mice in non-transduced T cells arm (Fig. 5F P ≤ 0.05). As In addition, we hypothesized shedding of TPO from the expected, these results suggest the TPO-CAR results in on-target membrane-bound CAR could act as a survival advantage for collateral clearance of MPL expressing HSPC.

Gene Therapy J.T. Zoine et al. 6

A. TPO-CAR CH3 Ψ RRE cPPT cPPT 5’ LTR hUBC TPO-CAR CD28 CD3ζ 3’ LTR

IL-2 SS

B. C. ns 2.0

1.5

1.0

0.5

0.0 Genomic Copies / 50ng DNA TPO-CAR CD19 CAR

D. E. CAR Txd F. - eGFP P2A P2A eGFP CD19 CAR Non- HD9 TPO Non Codon Optimized CAR

100

75 ** ** 55.97 kDa CD19 CAR VCN: 0.89 * * 55.64 kDa TPO CAR VCN: 0.21 50 43.35 kDa NCO CAR VCN: 0.24 37

20 CD3 dimerization bands Endogenous CD3 Loading Control 15 Exposure: 315s Fig. 2 Generating a thrombopoietin ligand-based CAR to target MPL. A Schematic of the TPO-CAR. The construct is entirely codon optimized from the IL2 signal sequence to the end of the CD3ζ sequence and contains a CH3 hinge domain. It includes a 5′ long terminal repeat (LTR), human ubiquitin C promoter (hUBC), an -2 signal sequence (IL-2 ss), the TPO-CAR, the CD28 region, the CD3ζ intracellular domain and a 3′ LTR. B Genomic DNA was isolated for RT-PCR and vector copy analysis was performed. C-E Target cells were labeled with VPD450 proliferation dye or CFSE dye. Co-cultures were established at a 1:4 effector to target and incubated for 12 h and subsequently stained for CD3, CD69, CD38, MPL, Annexin V, and PI for flow cytometry. Activation after a 12-hour co-culture experiment was measured when cells were co incubated with HEL (C), CMK (D), or Mo7e (E) cells. Activation was measured by CD69, CD38, CD69+ CD38+. Data are demonstrating one T cell donor with experimental triplicates; however, data are representative across donors (N = 3 biological donors). F Protein cell lysates were collected and 40 μg were loaded into an SDS PAGE gel for western blot analysis using CD3ζ and HRP for detection. After 315 s of exposure, CD3ζ bands were detected in non-transduced T cells (HD9), a control ligand-based CAR (CAR band: 43.35 kDa), TPO-CAR (CAR band: 55.64 kDa), and the control CD19 CAR (CAR band: 55.97 kDa). Vector copy number of the cells are reported to the right to demonstrate similar copy numbers. Data show more CAR in the TPO-CAR compared to the CD19 CAR.

DISCUSSION benefit[9]. Ligand CAR designs can be advantageous because of Despite the promise of early CAR T therapy successes, there their ability to engage and disengage with receptors. However, remains a dearth of validated ideal targets and target-binding disadvantages of ligand antigen binding domains include weaker scFvs. Therefore, we explored alternative mechanisms to engage avidity, increased probability of CAR shedding, and enhanced cell surface receptors. We hypothesized that the repertoire of competition with the natural expressed ligand. To date, there are cancer targets could be expanded by utilizing a ligand-based CAR few ligand-based CARs in preclinical and clinical development— approach. In many cases, the understanding of ligand–receptor those that exist include the interleukin 13R2, interleukin 11, interactions is superior to antibody–antigen interactions, and thus adnectin, follicle stimulating , Fms-like tyrosine one can better predict the on-target, off-tumor side effects, which kinase 3, and granulocyte-macrophage colony stimulating factor helps to anticipate, and possibly exploit, side effects for clinical [47–51].

Gene Therapy J.T. Zoine et al. 7

Fig. 3 Cytotoxicity and specificity of the TPO CAR. The TPO CAR cytotoxic potential was measured against HEL (A), MO7e (B), CMK (C) cells in 12-h coculture assay. Increasing effector to target ratios (E:T) were tested including 0:1 (stained target cells alone), 1:1, 2:1, and 5:1 (y-axis) with non-transduced T cells, and the TPO-CAR. The CD19 CAR was only tested at the 1:1 E:T ratio, ***p < 0.001 D Cytotoxicity assays from multiple donors were pooled and the effector to target ratio 1:1 was compared within the CMK cell line. **P < 0.01 ***P < 0.001. E, F After the coculture cytotoxicity experiment, the remaining living CMK cells were evaluated for remaining MPL expression and MFI. G MPL- cells were stained with CFSE (K562 or 697) and MPL+ (CMK) cells were stained with VPD450. One hundred thousand cells from each cell line were mixed together and incubated with 100,000 TPO-CAR transduced T cells. Cytotoxicity was measured within the MPL- cells, K562’s cytotoxicity due the TPO- CAR was 11.7% ± 0.6 and 697’s 25.0% ± 5.74 versus CMK cell line 83.6 ± 8.6, P ≤ 0.0001 and 76.1 ± 1.0, P ≤ 0.0001.

Gene Therapy J.T. Zoine et al. 8

Fig. 4 Blocking TPO-CAR activity with thrombopoietin. A CMK cells were cocultured ±400 ng/mL of TPO with non-transduced T cells or TPO- CAR T cells at a 1 to 1 ratio for 12 h. Ordinary one-way ANOVA, ****P < 0.0001, ***P < 0.001 B CMK cells were using in the same assay mention but cocultured with doses of 0, 0.1 ng/mL, 1 ng/mL, 10 ng/mL, 100 ng/mL, and 400 ng/mL of recombinant TPO. The CMK cell line was stimulated with recombinant TPO, or media from non-transduced T cells, or TPO-CAR T cells +/− a polyclonal TPO antibody for 45 min. Representative histograms (C), percentage of pSTAT5 (D) and the MFI (E) demonstrate pSTAT5 staining ****P < 0.0001, ***P < 0.001.

We designed a ligand-based CAR targeting the MPL receptor the performance of the TPO-CAR in competition with supraphy- for leukemic cancer stem cell clearance [12, 52]. Our reason for siological dose of TPO. While we did observe decreases in pursuing a cancer stem cell population is due to their cytotoxicity, we found the TPO-CAR was still able to induce chemoresistance and ability to self-renew, making MPL an ideal substantial specific cell lysis compared to controls. This could candidate to treat LSC and prevent relapse [53]. Furthermore, potentially be resolved by depleting all natural TPO in vivo prior MPL has limited expression in healthy tissues, making the on- to infusion of CAR T cells. Another safety concern of this target, off-tumor side effects more predictable and manageable particular approach is the shedding of TPO from the CAR T cells (The GTex Portal). We have successfully targeted AML using the which induces activation of leukemia cells. It is plausible that the TPO-CAR targeting MPL. In vitro studies verified the functionality shed TPO could result in a growth advantage for MPL positive of the CAR, including activation by multiple indicators and cells. This effect could be one of the explanations as to why there cytotoxicity experiments with multiple cell lines. This is the first were no survival benefit in the in vivo experiments upon CAR report of a CAR designed to target MPL using a novel ligand- T cells transfer. However, despite potential TPO shedding based approach. concerns leading to survival of MPL positive cells, in vitro We hypothesized that naturally circulating TPO could suppress cytotoxicity assays showed significant killing of MPL+ cells. In the TPO-CAR’s cytotoxicity, so we designed experiments to test addition, most ligand-based CAR T cells do not include studies

Gene Therapy J.T. Zoine et al. 9

Fig. 5 In vivo specificity of TPO-CAR in leukemia xenografts against MPL. A–C Splenocytes from leukemia mice and TPO-CAR mice were evaluated for remaining MPL expression. Surface MPL expression was measured by flow cytometry (F, 30.3% ± 10.7 vs 77.1% ± 4.3, p = 0.002) and MPL MFI (G, 6421 ± 151 vs 3601 ± 535, p = 0.0009) on remaining leukemia compared to control mice. The bone marrow was analyzed for total cell count (D)LK(E) and LSK (F) bone marrow compartments. on ligand shedding and this could potentially be a problem hematopoietic human cells at higher effector to target ratios to across the platform. Finally, we would like to highlight that the evaluate safety. Our data support the conclusion we were patient’s endogenous TPO could cause significant growth specifically targeting the MPL+ population. Given that “catch all” advantage than the shed TPO from CAR-T cells. Our future plans therapies have not been successful in targeting heterogeneous include reengineering the CAR to reduce the amount of cancer populations and maintaining complete responses, the shedding and the suggested assay will allow us to move forward development of CARs to unique populations with minimal off- in that direction. We plan to test the TPO-CAR against more target effects are beneficial to the advancement of leukemia antigen negative cell lines and healthy hematopoietic and non- treatments. Our in vitro data strongly support the advancement

Gene Therapy J.T. Zoine et al. 10 of the TPO-CAR as a targeted agent for MPL+ stem like synNOTCH system for dual-antigen recognition to allow for leukemias. delayed transcription of the CAR, or an alternative short-lived Despite selecting for a monoclonal MPL+ population, we found immune cell source, such as γδ T cells or NK cells, could be utilized the CMK cell line would revert back to variable expression. [58–62]. In addition, lymphodepleting serotherapy or chemother- Therefore, future testing to the in vivo preclinical utility of the apy are usually part of a standard allogeneic HSCT conditioning TPO-CAR will require a 100% MPL positive cell line. Given, the regimen which could also decrease the chance of CAR-T cell success of our in vitro data in selecting for MPL+ leukemia cells persistence. we were aware of the benefit that the TPO-CAR could have in Since CAR therapy is advancing to target multiple antigens in combination with other relevant therapies. Nonetheless, we used one cellular product, one could adopt this strategy to target MPL the TPO-CAR as a single agent for in vivo testing using an immune and another antigen highly expressed on the surface. Therefore, compromised megakaryocytic leukemia model. We found the we anticipate targeting MPL using a bispecificCARTcell TPO-CAR treated animals succumbed shortly after the untreated approach could be a promising avenue for targeting AML. animals that received leukemia only. Our hypotheses were this Current, bispecific targeting approaches for AML include was either due to (i) on-target, off-tumor side effects or (ii) the bispecific antibodies, bispecific engager T cells, and ongoing TPO-CAR’s specific cytotoxicity against MPL+ leukemia cells testing of bispecific CAR T cells. This work impacts the study of allowed for outgrowth of MPL- populations. Due to the presence ligand- based CAR technology but highlights important con- of MPL on the stem cell compartment, we hypothesized the TPO- siderationssuchastheneedforhighaffinity ligands or using a CAR could rapidly clear bone marrow, leading to suppressed dual-ligand CAR to overcome some of the problems met in hematopoiesis. We viewed CAR effects in the stem and progenitor these studies. Taking these considerations into account, the cell compartment as a potential benefit by acting as a bridge to overall goal for the TPO-CAR would be to extend the use of this allogeneic HSC transplantation, possibly with decreased need for CAR to all relapsed hematopoietic cancers that have MPL genotoxic conditioning, while also acknowledging its potential as expression on the surface. a side effect. There is a pronounced need to target the TPO/MPL axis due to supporting data suggesting there is a biological sequestration of TPO by high MPL expressing leukemia [19]. REFERENCES Further, many patients presenting with this level of disease in the MPL+ leukemias typically have a poor prognosis and will 1. Sadelain M, Brentjens R, Rivière I. The basic principles of chimeric antigen receptor design. Cancer Discov. 2013;3:388–98. eventually need a bone marrow transplant following initial 2. Huston MM-H JS, Tai MS, McCartney J, Warren F, Haber E, Oppermann. H. Protein treatment [54]. engineering of single-chain Fv analogs and fusion . Methods Enzymol. Mice that received leukemia cells alone or leukemia cells with 1991;203:46–52. TPO-CAR T cells were evaluated for in vivo targeting of MPL+ 3. Brentjens RJ, Davila ML, Riviere I, Park J, Wang X, Cowell LG, et al. CD19-targeted leukemia cells. Our observations suggested that the TPO-CAR was T cells rapidly induce molecular remissions in adults with chemotherapy- targeting MPL+ cells, which have been documented as harder to refractory acute lymphoblastic leukemia. Sci Transl Med. 2013;5:177ra38. clear with conventional therapies, thereby verifying our in vitro 4. Chang ZL, Chen YY. CARs: synthetic immunoreceptors for cancer therapy and – results while leaving the MPL- portion to grow out leading to no beyond. Trends Mol Med. 2017;23:430 50. survival benefit. Due to previous findings suggesting MPL+ LSCs 5. Grupp SA, Kalos M, Barrett D, Aplenc R, Porter DL, Rheingold SR, et al. Chimeric antigen receptor-modified T cells for acute lymphoid leukemia. N Engl J Med. are resistant to chemotherapy, it is entirely possible that this CAR 2013;368:1509–18. could be paired with chemotherapy, which would target MPL- 6. Kochenderfer JN, Dudley ME, Carpenter RO, Kassim SH, Rose JJ, Telford WG, et al. populations of leukemia cells. Donor-derived CD19-targeted T cells cause regression of malignancy persisting In addition to targeting MPL+ leukemias, our data suggest after allogeneic hematopoietic stem cell transplantation. Blood. 2013;122: there is suppression in the bone marrow compartment of mice 4129–39. administered TPO-CAR. This can be partially due to the CAR 7. Chmielewski M, Abken H. TRUCKs: the fourth generation of CARs. Expert Opin Biol clearing the bone marrow HSPCs but could be an effect of the Ther. 2015;15:1145–54. ’ inflammatory bone marrow milieu from human T cells activation 8. Townsend MH, Shrestha G, Robison RA, O Neill KL. The expansion of targetable and proliferation in the mouse bone marrow leading to bone biomarkers for CAR T cell therapy. J Exp Clin Cancer Res. 2018;37:163. 9. Murad JM, Graber DJ, Sentman CL. Advances in the use of natural receptor- or marrow suppression. While we have performed limited studies to ligand-based chimeric antigen receptors (CARs) in haematologic malignancies. evaluate HSPC depletion of TPO-CAR T cells in the absence of Best Pract Res Clin Haematol. 2018;31:176–83. tumor, the current standard of care for AML refractory to 10. Lok S, Kaushansky K, Holly RD, Kuijper JL, Lofton-Day CE, Oort PJ, et al. Cloning chemotherapy is to proceed to allogeneic stem cell transplant and expression of murine thrombopoietin cDNA and stimulation of platelet [55–57]. If our approach is translated to the clinic, the TPO CAR will production in vivo. Nature. 1994;369:565–8. not only help to decrease the disease burden, but due to its added 11. Fox N, Priestley G, Papayannopoulou T, Kaushansky K. Thrombopoietin expands HSPC clearance it could potentially decrease the amount of hematopoietic stem cells after transplantation. J Clin Investig. 2002;110:389–94. conditioning chemotherapy or irradiation required for allogeneic 12. Kaushansky K, Lok S, Holly RD, Broudy VC, Lin N, Bailey MC, et al. Promotion of hematopoietic stem cell transplantation (HSCT) thereby could progenitor expansion and differentiation by the c-Mpl ligand thrombopoietin. Nature. 1994;369:568–71. decrease short term and long-term toxicity and morbidity. In 13. Yoshihara H, Arai F, Hosokawa K, Hagiwara T, Takubo K, Nakamura Y, et al. additional studies evaluating the persistence and homing of the Thrombopoietin/MPL signaling regulates hematopoietic stem cell quiescence TPO-CAR in non-tumor bearing animals to test toxicity and and interaction with the osteoblastic niche. Cell Stem Cell. 2007;1:685–97. evaluate homing so we can better address these safety concerns 14. Lok S, Foster DC. The structure, biology and potential therapeutic applications of are needed. Also, tumor lysis syndrome could be potential recombinant thrombopoietin. STEM CELLS. 1994;12:586–98. concerns from both depletion of tumor and parallel clearance of 15. Albitar M, Manshouri T, Kantarjian H, Keating M, Estrov Z, Faber J, et al. Corre- HSC. However, MPL is known to have a very limited expression lation between lower c-mpl protein expression and favorable cytogenetic groups – outside of the hematopoietic compartment and HSC contribute to in acute myeloid leukemia. Leukemia Res. 1999;23:63 69. <1% of bone marrow compartment. Hence the contribution of 16. Lange B. The management of neoplastic disorders of haematopoiesis in children with Down’s syndrome. Br J Haematol. 2000;110:512–24. HSC clearance to tumor lysis syndrome is likely to be limited. 17. Li H, Rao Q, Yu P, Chen SY, Li Z, Xing HY, et al. Expression of MPL in leukemia stem Another important concern is the persistence of CAR-T cell post- cells and its role in stemness maintainance. Blood. 2016;128:1723. HSCT, which could potentially impair donor engraftment. To 18. Quentmeier H, Zaborski M, Graf G, Ludwig WD, Drexler HG. Expression of the manage persistence of CAR-T cells and on-target/off-tumor receptor MPL and proliferative effects of its ligand thrombopoietin on human toxicity, a suicide switch could be introduced into the CAR, a leukemia cells. Leukemia. 1996;10:297–310.

Gene Therapy J.T. Zoine et al. 11 19. Rauch PJ, Ellegast JM, Widmer CC, Fritsch K, Goede JS, Valk PJ, et al. MPL 44. Alessio M, Roggero S, Funaro A, De Monte LB, Peruzzi L, Geuna M, et al. CD38 expression on AML blasts predicts peripheral blood neutropenia and thrombo- molecule: structural and biochemical analysis on human T lymphocytes, thy- cytopenia. Blood. 2016;128:2253–7. mocytes, and plasma cells. J Immunol. 1990;145:878–84. 20. Vigon I, Dreyfus F, Melle J, Viguie F, Ribrag V, Cocault L, et al. Expression of the 45. Chaturvedi V, Marsh RA, Lorenz AZ, Owsley E, Chaturvedi V, Nguyen T, et al. T cell c-mpl proto- in human hematologic malignancies. Blood. activation profiles distinguish hemophagocytic lymphohistiocytosis and early 1993;82:877–83. sepsis. Blood 2021;137:2337–46. 21. Yu P, Qiu SW, Rao Q, Lin D, Xing HY, Tang KJ. et al. Expression of c-MPL in 46. Shubinsky G, Schlesinger M. The CD38 lymphocyte differentiation marker: new leukemic stem cells from acute myeloid leukemia patients. Zhongguo Shi Yan insight into its ectoenzymatic activity and its role as a signal transducer. Immu- Xue Ye Xue Za Zhi.2012;20:1052–5. nity. 1997;7:315–24. 22. Wetzler M, Baer MR, Bernstein SH, Blumenson L, Stewart C, Barcos M, et al. 47. Huang G, Yu L, Cooper LJ, Hollomon M, Huls H, Kleinerman ES. Genetically Expression of c-mpl mRNA, the receptor for thrombopoietin, in acute myeloid modified T cells targeting interleukin-11 receptor alpha-chain kill human osteo- leukemia blasts identifies a group of patients with poor response to intensive sarcoma cells and induce the regression of established osteosarcoma lung chemotherapy. J Clin Oncol. 1997;15:2262–8. metastases. Cancer Res. 2012;72:271–81. 23. Guan Y, Gerhard B, Hogge DE. Detection, isolation, and stimulation of quiescent 48. Kahlon KS, Brown C, Cooper LJ, Raubitschek A, Forman SJ, Jensen MC. Specific primitive leukemic progenitor cells from patients with acute myeloid leukemia recognition and killing of glioblastoma multiforme by interleukin 13-zetakine (AML). Blood. 2003;101:3142–9. redirected cytolytic T cells. Cancer Res. 2004;64:9160–6. 24. Boyd AL, Aslostovar L, Reid J, Ye W, Tanasijevic B, Porras DP, et al. Identification of 49. Han X, Cinay GE, Zhao Y, Guo Y, Zhang X, Wang P. Adnectin-based design of chemotherapy-induced leukemic-regenerating cells reveals a transient vulner- chimeric antigen receptor for T cell engineering. Mol Ther. 2017;25:2466–76. ability of human AML recurrence. Cancer Cell. 2018;34:483–498 e5. 50. Perales-Puchalt A, Svoronos N, Rutkowski MR, Allegrezza MJ, Tesone AJ, Payne 25. Costello RT, Mallet F, Gaugler B, Sainty D, Arnoulet C, Gastaut JA, et al. Human KK, et al. Follicle-stimulating hormone receptor is expressed by most ovarian acute myeloid leukemia CD34+/CD38- progenitor cells have decreased sensi- cancer subtypes and is a safe and effective immunotherapeutic target. Clin tivity to chemotherapy and Fas-induced apoptosis, reduced immunogenicity, Cancer Res. 2017;23:441–53. and impaired dendritic cell transformation capacities. Cancer Res. 51. Nakazawa Y, Matsuda K, Kurata T, Sueki A, Tanaka M, Sakashita K, et al. Anti- 2000;60:4403–11. proliferative effects of T cells expressing a ligand-based chimeric antigen 26. Dong-Feng Z, Ting L, Yong Z, Cheng C, Xi Z, Pei-Yan K. The TPO/c-MPL pathway in receptor against CD116 on CD34(+) cells of juvenile myelomonocytic leukemia. J the bone marrow may protect leukemia cells from chemotherapy in AML Hematol Oncol. 2016;9:27. patients. Pathol Oncol Res. 2014;20:309–17. 52. Qian H, Buza-Vidas N, Hyland CD, Jensen CT, Antonchuk J, Mansson R, et al. 27. Ninos JM, Jefferies LC, Cogle CR, Kerr WG. The , c-Mpl, is Critical role of thrombopoietin in maintaining adult quiescent hematopoietic a selective surface marker for human hematopoietic stem cells. J Transl Med. stem cells. Cell Stem Cell. 2007;1:671–84. 2006;4:9. 53. Abdullah LN, Chow EK. Mechanisms of chemoresistance in cancer stem cells. Clin 28. Li H, Zhao N, Li Y, Xing H, Chen S, Xu Y, et al. c-MPL is a candidate surface marker Transl Med. 2013;2:3. and confers self-renewal, quiescence, chemotherapy resistance, and leukemia 54. Yogarajah M, Tefferi A. Leukemic transformation in myeloproliferative neoplasms: initiation potential in leukemia. Stem Cells. Stem Cells. 2018;36:1685–96. a literature review on risk, characteristics, and outcome. Mayo Clin Proc. 29. Sakashita K, Kato I, Daifu T, Saida S, Hiramatsu H, Nishinaka Y, et al. In vitro 2017;92:1118–28. expansion of CD34(+)CD38(-) cells under stimulation with hematopoietic growth 55. Breems DA, Van Putten WL, Huijgens PC, Ossenkoppele GJ, Verhoef GE, Verdonck factors on AGM-S3 cells in juvenile myelomonocytic leukemia. Leukemia. LF, et al. Prognostic index for adult patients with acute myeloid leukemia in first 2015;29:606–14. relapse. J Clin Oncol. 2005;23:1969–78. 30. Bouscary D, Prudhomme C, Quesnel B, Melle J, Picard F, Dreyfus F. c-mpl 56. Burnett AK, Goldstone A, Hills RK, Milligan D, Prentice A, Yin J, et al. Curability of expression in hematologic disorders. Leuk Lymphoma. 1995;17:19–26. patients with acute myeloid leukemia who did not undergo transplantation in 31. Wegehaupt AK, Roufs EK, Hewitt CR, Killian ML, Gorbatenko O, Anderson CM, first remission. J Clin Oncol. 2013;31:1293–301. et al. Recovery and assessment of leukocytes from LR Express filters. Biologicals. 57. Ganzel C, Sun Z, Cripe LD, Fernandez HF, Douer D, Rowe JM, et al. Very poor long- 2017;49:15–22. term survival in past and more recent studies for relapsed AML patients: the 32. Neron S, Dussault N, Racine C. Whole-blood leukoreduction filters are a source for ECOG-ACRIN experience. Am J Hematol. 2018. https://doi.org/10.1002/ajh.25162. cryopreserved cells for phenotypic and functional investigations on peripheral 58. Zoine JT, Knight KA, Fleischer LC, Sutton KS, Goldsmith KC, Doering CB, et al. Ex blood lymphocytes. Transfusion. 2006;46:537–44. vivo expanded patient-derived gammadelta T-cell immunotherapy enhances 33. Feese MD, Tamada T, Kato Y, Maeda Y, Hirose M, Matsukura Y, et al. Structure of neuroblastoma tumor regression in a murine model. Oncoimmunology. the receptor-binding domain of human thrombopoietin determined by com- 2019;8:1593804. plexation with a neutralizing antibody fragment. Proc Natl Acad Sci USA. 59. Rosenberg SA, Restifo NP, Yang JC, Morgan RA, Dudley ME. Adoptive cell transfer: a 2004;101:1816–21. clinical path to effective cancer immunotherapy. Nat Rev Cancer. 2008;8:299–308. 34. Downing JR, Wilson RK, Zhang J, Mardis ER, Pui CH, Ding L, et al. The pediatric 60. Patel S, Burga RA, Powell AB, Chorvinsky EA, Hoq N, McCormack SE, et al. Beyond cancer genome project. Nat Genet. 2012;44:619–22. CAR T cells: other cell-based immunotherapeutic strategies against cancer. Front 35. Zhang J, Ding L, Holmfeldt L, Wu G, Heatley SL, Payne-Turner D, et al. The genetic Oncol. 2019;9:196. basis of early T-cell precursor acute lymphoblastic leukaemia. Nature. 61. Martinez M, Moon EK. CAR T cells for solid tumors: new strategies for finding, 2012;481:157–63. infiltrating, and surviving in the tumor microenvironment. Front Immunol. 36. Zhang J, Wu G, Miller CP, Tatevossian RG, Dalton JD, Tang B, et al. Whole-genome 2019;10:128. sequencing identifies genetic alterations in pediatric low-grade gliomas. Nat 62. Morsut L, Roybal Kole T, Xiong X, Gordley Russell M, Coyle Scott M, Thomson M, Genet. 2013;45:602–12. et al. Engineering customized cell sensing and response behaviors using syn- 37. Zhang J, Benavente CA, McEvoy J, Flores-Otero J, Ding L, Chen X, et al. A novel thetic notch receptors. Cell. 2016;164:780–91. retinoblastoma therapy from genomic and epigenetic analyses. Nature. 2012;481:329–34. 38. Robinson G, Parker M, Kranenburg TA, Lu C, Chen X, Ding L, et al. Novel muta- ACKNOWLEDGEMENTS tions target distinct subgroups of medulloblastoma. Nature. 2012;488:43–8. NHLBI 1K08HL141635-01A1 and Atlanta Pediatric Scholars Program K12 Scholar 39. Chen X, Stewart E, Shelat AA, Qu C, Bahrami A, Hatley M, et al. Targeting oxidative supported by grant K12HD072245. stress in embryonal rhabdomyosarcoma. Cancer Cell. 2013;24:710–24. 40. Holmfeldt L, Wei L, Diaz-Flores E, Walsh M, Zhang J, Ding L, et al. The genomic landscape of hypodiploid acute lymphoblastic leukemia. Nat Genet. 2013;45:242–52. AUTHOR CONTRIBUTIONS 41. Arai Y, Choi U, Corsino CI, Koontz SM, Tajima M, Sweeney CL, et al. Myeloid JTZ designed research, performed research, collected data, analyzed/interpreted conditioning with c-kit-Targeted CAR-T cells enables donor stem cell engraft- data, and wrote the manuscript; CZP, JYS, AML, AF, GMB, JYS, and CCP designed ment. Mol Ther. 2018;26:1181–97. research, performed research, collected data, and analyzed/interpreted data. CBD, 42. Kuter DJ. The biology of thrombopoietin and thrombopoietin receptor agonists. HTS, and SC designed research, analyzed/interpreted data and edited the manuscript. Int J Hematol. 2013;98:10–23. 43. Brown HC, Zakas PM, George SN, Parker ET, Spencer HT, Doering CB. Target-cell- directed bioengineering approaches for gene therapy of hemophilia A. Mol Ther COMPETING INTERESTS Methods Clin Dev. 2018;9:57–69. The authors declare no competing interests.

Gene Therapy J.T. Zoine et al. 12 ADDITIONAL INFORMATION Reprints and permission information is available at http://www.nature.com/ Supplementary information The online version contains supplementary material reprints available at https://doi.org/10.1038/s41434-021-00283-5. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims Correspondence and requests for materials should be addressed to S.C. in published maps and institutional affiliations.

Gene Therapy