Leukemia (2019) 33:1557–1569 https://doi.org/10.1038/s41375-018-0353-0 ARTICLE Acute lymphoblastic leukemia NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia Belen Lopez-Millan1 Diego Sanchéz-Martínez1 Heleia Roca-Ho1 Francisco Gutiérrez-Agüera1 Oscar Molina1 ● ● ● ● ● Rafael Diaz de la Guardia1 Raúl Torres-Ruiz1,2 Jose Luís Fuster3 Paola Ballerini4 Ute Suessbier5 ● ● ● ● ● Cesar Nombela-Arrieta 5 Clara Bueno1,6 Pablo Menéndez1,6,7 ● ● Received: 20 September 2018 / Revised: 15 November 2018 / Accepted: 28 November 2018 / Published online: 11 January 2019 © The Author(s) 2019. This article is published with open access Abstract B cell acute lymphoblastic leukemia (B-ALL) is the most common childhood cancer, with cure rates of ∼80%. MLL- rearranged (MLLr) B-ALL (MLLr-B-ALL) has, however, an unfavorable prognosis with common therapy refractoriness and early relapse, and therefore new therapeutic targets are needed for relapsed/refractory MLLr-B-ALL. MLLr leukemias are characterized by the specific expression of chondroitin sulfate proteoglycan-4, also known as neuron-glial antigen-2 (NG2). 1234567890();,: 1234567890();,: NG2 was recently shown involved in leukemia invasiveness and central nervous system infiltration in MLLr-B-ALL, and correlated with lower event-free survival (EFS). We here hypothesized that blocking NG2 may synergize with established induction therapy for B-ALL based on vincristine, glucocorticoids, and L-asparaginase (VxL). Using robust patient-derived xenograft (PDX) models, we found that NG2 is crucial for MLLr-B-ALL engraftment upon intravenous (i.v.) transplantation. In vivo blockade of NG2 using either chondroitinase-ABC or an anti-NG2-specific monoclonal antibody (MoAb) resulted in a significant mobilization of MLLr-B-ALL blasts from bone marrow (BM) to peripheral blood (PB) as demonstrated by cytometric and 3D confocal imaging analysis. When combined with either NG2 antagonist, VxL treatment achieved higher rates of complete remission, and consequently higher EFS and delayed time to relapse. Mechanistically, anti-NG2 MoAb induces neither antibody-dependent cell-mediated not complement-dependent cytotoxicity. NG2 blockade rather overrides BM stroma-mediated chemoprotection through PB mobilization of MLLr-B- ALL blasts, thus becoming more accessible to chemotherapy. We provide a proof of concept for NG2 as a therapeutic target for MLLr-B-ALL. Introduction B cell acute lymphoblastic leukemia (B-ALL) is the most common childhood cancer [1]. Over the past 20 years, Supplementary information The online version of this article (https:// doi.org/10.1038/s41375-018-0353-0) contains supplementary progress in molecular diagnosis, patient stratification, material, which is available to authorized users. disease biology, therapy, and supportive care has * Belen Lopez-Millan Madrid, Spain blopez@carrerasresearch.org 3 Pediatric Hematology and Oncology Section, Hospital Clínico * Clara Bueno Virgen de la Arrixaca, Murcia, Spain cbueno@carrerasresearch.org 4 Pediatric Hematology, Armand Trousseau Hospital, Paris, France * Pablo Menéndez 5 pmenendez@carrerasresearch.org Hematology Department, University Hospital-University of Zurich, Zurich, Switzerland 6 1 Department of Biomedicine, School of Medicine, Josep Carreras Centro de Investigacion Biomedica en Red-Oncología Leukemia Research Institute, University of Barcelona, (CIBERONC), Zurich, Switzerland Barcelona, Spain 7 Instituciò Catalana de Recerca i Estudis Avançats (ICREA), 2 Molecular Cytogenetics Group, Human Cancer Genetics Program, Barcelona, Spain Centro Nacional de Investigaciones Oncológicas (CNIO), 1558 B. Lopez-Millan et al. considerably improved disease management and clinical Methods outcome. The backbone of current induction or re- induction post-relapse treatment protocols is vincristine, Patient samples and immunophenotyping glucocorticoids, and L-asparaginase (VxL), with or with- out an anthracycline [2], resulting in ∼90% of children Leukemic samples at presentation were used from n = 5 entering complete remission (CR) [3] and eventual cure independent MLLr-B-ALL patients with complete immu- rates of ∼80% [4, 5]. nophenotypic and molecular/cytogenetic diagnosis. Four Despite the overall progress in treatment, relapsed B- patients were t(4;11)/MLL-AF4+, and one patient was t ALL is the fifth most prevalent pediatric cancer, and B-ALL (1;11)/MLL-EPS15+. Patients’ mononuclear cells with remains the most common cause of death from malignancy >85% CD45lowCD19+CD34+CD10−NG2+ MLLr blasts in children [6, 7]. Moreover, several B-ALL cytogenetic– were isolated by density gradient centrifugation using molecular subgroups remain high risk, with 5-year event- Ficoll-Hypaque (Amersham Biosciences, Uppsala, Swe- free survival (EFS) and overall survival (OS) rates <30% den). MLL/(11q23) status was confirmed by fluorescence [8–10]. B-ALL carrying MLL rearrangements (MLLr-B- in situ hybridization [22]. Blasts were immunophenotyped ALL), particularly the t(4;11)+/MLL-AF4+ B-ALL, is of using the monoclonal antibodies (MoAbs) CD45-FITC, special interest because of its dismal prognosis, common CD19-APC, CD10-PerCP-Cy5.5, CD34-PE-Cy7 (BD therapy refractoriness, and central nervous system (CNS) Biosciences, San Jose, CA), and NG2−PE (Beckman, infiltration [11]. Patients experiencing early relapse or Barcelona, Spain), and the NG2+ and NG2− blast popula- therapy refractoriness have a poor long-term survival and, tions were isolated by fluorescence-activated cell sorting in these cases, the best therapeutic option is hematopoietic (FACS) using a FACSAria cell sorter (BD Biosciences). stem cell transplantation (HSCT) following induction into The Institutional Review Board of the Hospital Clinic of second remission. However, these patients commonly fail to Barcelona approved the study, and all patients’ parents gave achieve a second remission; moreover, current chemother- written informed consent. apy is associated with morbidity and serious side effects such as infertility, impaired development, and greater risk of Drugs and antibodies secondary neoplasms [12, 13]. Thus, innovative directed/ targeted therapeutic approaches are in high demand for Vincristine (Selleckchem, Houston, TX, USA) and dex- high-risk B-ALL and relapsed/refractory (R/R) B-ALL, and amethasone (Sigma-Aldrich, St Louis, MO, USA) were targeting leukemia-initiating cells and CNS-infiltrating reconstituted in dimethyl sulfoxide. L-asparaginase (Kidro- leukemia cells is key to overcome therapy resistance, lase®; EUSA Pharma, Oxford, UK) and chondroitinase- relapse, and CNS disease [14, 15]. ABC (Ch’ase, Sigma-Aldrich) were reconstituted in Chondroitin sulfate proteoglycan-4, also known as phosphate-buffered saline (PBS) according to the supplier’s neuron-glial antigen-2 (NG2), is a transmembrane pro- guidelines. Drugs were stored in aliquots at −20 °C. The teoglycan barely expressed in normal hematopoietic cells clone 9.2.27 MoAb was provided by Abcam (Cambridge, [16, 17]. Conversely, ~90% of 11q23/MLLr leukemias UK). The clone 7.1 MoAb-producing hybridoma was specifically express NG2, which has been incorporated into kindly provided by Professor Irwin Bernstein (Fred diagnostic workflows for leukemia immunophenotyping Hutchinson Cancer Center, Seattle, WA, USA). The anti- because of its predictive value for MLLr [16, 18–21]. NG2 7.1 MoAb was produced and purified using standard Against this background, we recently reported that NG2 is methods as previously detailed [18]. All drugs were admi- involved in leukemia invasiveness and CNS infiltration in nistered by intraperitoneal (i.p.) injection. MLLr-B-ALL, and high levels of NG2 expression in B- ALL blasts correlates with lower EFS, higher numbers of PDX models, in vivo treatment, and analysis of circulating blasts and more frequent CNS disease/relapse. In leukemia engraftment the present study, we investigated whether NG2 might represent a therapeutic target for MLLr-B-ALL. Using Eight- to-14-week-old NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ robust preclinical patient-derived xenograft (PDX) models, mice (NSG; n = 116) housed under pathogen-free condi- we show that NG2 antagonists synergize with VxL-based tions were used. The Animal Care Committee of The Bar- induction therapy, leading to an extensive mobilization of celona Biomedical Research Park approved all experimental MLLr-B-ALL blasts from bone marrow (BM) into periph- procedures with mice (HRH-17-0045-P2). A total of 1–2 × eral blood (PB) where they become more accessible/sensi- 105 iMLLr-B-ALL cells were transplanted intravenously tive to VxL-based chemotherapy, resulting in higher CR (i.v) into sublethally (2.25 Gy) irradiated mice as described rates (CRRs), and consequently, higher EFS and delayed [23]. Leukemia engraftment was monitored through weekly time to relapse. PB analysis, and human grafts were immunophenotyped by NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1559 flow cytometry using HLA-ABC-FITC combined with the cells (SEM (NG2+CD20−) and Daudi (NG2-CD20+)) were 7.1 MoAb indicated above. When human engraftment in PB pre-incubated with anti-NG2 (7.1 MoAb) or anti-CD20 was >0.5% (between weeks 6 and 8, depending on the (Rituximab) MoAb at 10 μg/mL for 30 min at 37 °C. Then, engraftment kinetics of each patient), mice were homo- they were incubated with either peripheral blood mono- geneously divided into the following treatment groups: (i) nuclear cells (PBMCs) at 10:1 effector:target ratio or 10% control, (ii) VxL alone, (iii) VxL plus Ch’ase, and (iv) VxL human AB serum to induce ADCC and CDC, respectively. plus 7.1 MoAb. Treatment schedules were as follows: Cytotoxicity was determined 4 and 24 h later by analyzing vincristine (V, 0.15 mg/kg) once weekly for 2 weeks, dex- the proportion of AnnexinV+/7-ADD+ by FACS. PBMCs amethasone (x, 5 mg/kg), and L-asparaginase (L, 1000 U/ were labeled with 3 µM eFluor to distinguish them from kg) daily during 5 days for 2 weeks. This standard induction target cells. Specific cell death was determined relative to treatment is known as VxL treatment. Ch’ase (0.06 U/ basal cell death in the absence of MoAb (consistently <5%). mouse) and 7.1 MoAb (10 mg/kg) were administered daily for 7 days (when given alone) or for 14 days (in combi- 3D confocal microscopy of BM nation with VxL). BM aspirates were always performed at the beginning and end of each treatment. Minimal residual 3D imaging of mouse femoral BM cavities was performed disease (MRD) was assessed in the BM of each mouse at as previously described [31]. Briefly, bones were collected, the completion of the 15-day treatment schedule. CR was fixed for 6 h in PBS/2% paraformaldehyde (PFA) at 4 °C, defined as the presence of ≤1% leukemic cells in BM (0.1% and dehydrated by incubation in PBS/30% sucrose for 72 h. in PB). Treatment was then stopped and mice were left Bones were then embedded in OCT, and snap frozen in untreated for 30 further days to follow-up potential relapse liquid nitrogen. Thick slices (400–700 µm) of BM were by weekly analysis of engraftment in PB. Analysis of EFS, obtained by iteratively sectioning both sides of the femur comparing mice treated with VxL vs. VxL plus NG2 using a cryostat until the BM cavity was bilaterally exposed. blockers, was performed with Kaplan–Meier curves from The remaining OCT embedding medium was removed by the end of the induction treatment (day 15) up to day 45. A washing BM slices three times in PBS for 5 min, followed leukemic engraftment in PB >0.5% was established to by an additional fixation step in 2% PFA for 2 h at 4 °C. define a relapse event. Mice were sacrificed at the end of the Slices were blocked overnight in blocking buffer (0.2% experiment. Triton X-100, 1% BSA, 10% Donkey serum in PBS) and incubated with primary antibodies against endomucin In vitro chemoresistance assay (CD105, Santa Cruz Biotechnology, Dallas, TX, USA), huCD45, and collagen (Abcam) in blocking solution for 72 Primary cells were maintained in a humidified atmosphere h at 4 °C. After repeated washing in PBS (3 × 2 h) slices with 5% CO2 at 37 °C. MLLr-B-ALL primary samples cells were incubated with a secondary antibody cocktail solution were cultured in StemSpam medium (Stem Cell Technol- (Jackson ImmunoResearch (Cambridge, UK and Invitro- ogies, Vancouver, BC, Canada) supplemented with stem gen) including 2 µm 4′,6-diamidino-2-phenylindole (DAPI) cell factor, FLT3 ligand, interleukin-3 (IL-3) and IL-7 (all for an additional 72 h at 4 °C. After successive washing from PeproTech, Rocky Hill, NJ, USA) as described [24]. steps, immunostained slices were optically cleared through Bone marrow-derived mesenchymal stromal cells (BM- overnight incubation in RapiClear 1.52 (Sunjinlab, Hsinchu MSCs) were obtained, grown, and characterized as exten- City, Taiwan). The slices were then mounted on glass slides sively described by our group [25–27]. A total of 1 × 105 for imaging using a SP8 Leica confocal microscope MLLr-B-ALL blasts were co-cultured in a 96-well plate equipped with hybrid detectors using ×10 (HCX-PL- with/without 2 × 104 irradiated BM-MSCs for either 30 min FLUOTAR) and ×20 (HC-PL-APO-CS2) objectives. Ima- or 24 h. Cells were then exposed to 0.5 or 50 µM VxL for ging data were analyzed and videos assembled using the 40 h. Viability (apoptosis) of CD19+ B-ALL blasts was Imaris v8.2 software (Bitplane AG). Supplemental videos measured using 7-AAD on a FACSCanto-II cytometer may be found in Supplementary data available with the running FACSDiva software (BD Biosciences) as pre- online version of this article. viously described [28]. Statistical analysis ADCC and CDC assays Data are expressed as mean ± SEM of independent experi- Antibody-dependent cell-mediated (ADCC) and ments unless otherwise specified. Statistical comparisons complement-dependent cytotoxicity (CDC) assays were were performed using either paired or unpaired Student’s t performed to directly address the anti-NG2-mediated cyto- test, as appropriate. For MRD studies, data are expressed as toxicity, as previously described [29, 30]. Briefly, target median (range) and significant differences were analyzed by 1560 B. Lopez-Millan et al. the median test. CRRs were statistically compared using Prism v6.0 (GraphPad Software Inc., La Jolla, CA, USA) or Pearson’s χ2 test and EFS curves were compared using the SPSS software (SPSS version 22, Chicago, IL, USA). Sta- log-rank test. Analyses were performed using GraphPad tistical significance was defined as a p value < 0.05. NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1561 Fig. 1 In vivo blockade of NG2 results in the robust mobilization of engraftment dynamics of MLLr-B-ALL blasts in PDX MLLr-B-ALL blasts into PB. a Left, experimental design of the models (Fig. 1b). A total of 105 NG2+ MLLr-B-ALL cells in vitro treatment with NG2 antagonists. Middle, engraftment capacity (n = 4 patients) were i.v. transplanted and, once PB of NG2+ or NG2− MLLr-B-ALL blasts 8 weeks after i.v. transplan- tation. Right, overnight in vitro exposure of NG2+ blasts to the NG2 engraftment reached >0.5%, mice were daily i.p. treated antagonists Ch’ase, 7.1 MoAb, or 9.2.27 MoAb abolishes their with the NG2 antagonist Ch’ase (0.06 U/mouse) or anti- engraftment potential [18]. b Experimental design of the in vivo NG2 7.1 MoAb (10 mg/kg/mouse) for 7 days (Fig. 1b). In treatment with NG2 antagonists. c Monitoring of the levels of leu- line with the in vitro data, NG2 blockage in vivo failed to kemic grafts in PB and BM before (day 0) and after (day 7) the indicated treatments. Each line represents the same mouse before and reduce tumor burden (Fig. 1c). However, when we com- after treatment. d Levels of leukemic grafts in BM (top panel) and PB pared vehicle-treated primografts with primografts treated (bottom panel) after the indicated in vivo treatments. Results are with either NG2 antagonist after completion of the treat- shown as mean ± SEM, relative to day 0 (before treatment). e Repre- ment, we found that treatment with NG2 antagonists sig- sentative FACS plots showing the identical leukemia NG2+ phenotype in both diagnostic samples and primografts. The right panel shows the nificantly reduced leukemic burden in BM that was in vivo effectiveness of the 7.1 MoAb, which abolishes NG2 accompanied by a massive leukemia infiltration in PB (p < expression in blasts recovered from primografts. n, 8–12 mice/group 0.01; Fig. 1c, d). Importantly, PDX models reproduced the from three different patients. *p < 0.05; **p < 0.01; ****p < 0.0001 immunophenotype of the de novo primary leukemia, and blasts recovered from primografts treated with NG2 Results blockers were mainly NG2− (Fig. 1e). These results show that blocking NG2 in vivo mobilizes MLLr-B-ALL blasts to NG2 blockage does not elicit cytotoxicity of MLLr-B- PB. This coupled to our previous data [18], reporting ALL blasts but results in in vivo robust blast similar engraftment levels of NG2+ MLLr-B-ALL blasts mobilization into PB when directly delivered intra-BM, suggest that NG2 reg- ulates mobilization to and engraftment in the BM. We first interrogated the ability of FACS-sorted NG2+ and In order to confirm FACS-based quantifications, we NG2− diagnostic leukemic blasts to reproduce the B-ALL employed advanced protocols for in situ 3D BM imaging, phenotype by engrafting in NSG mice. When equal num- which allows to visualize the presence and distribution of bers of NG2+ and NG2− B-ALL primary blasts (105 cells/ the human graft within the native BM microenvironment of patient, n = 4) were intravenously (i.v.) infused into NSG control and anti-NG2-treated mice at the end of treatment. mice, engraftment levels after 8 weeks were 3.5-fold higher Engrafted human B-ALL cells were labeled by immunos- in the PB of mice transplanted with NG2+ than in equiva- taining against huCD45 (Fig. 3a). Consistent with our lent mice transplanted with NG2− blasts (11.2 ± 1.8% vs. FACS data, we repeatedly observed the presence of large 3.2 ± 1.5%, p = 0.008) (Fig. 1a, left and middle panels). tissue volumes devoid of human cells, but filled with Overnight in vitro treatment of NG2+ B-ALL blasts with murine hematopoietic cells in the BM of anti-NG2-treated either Ch’ase (0.1U/ml) or anti-NG2 MoAb (either 7.1 (0.7 mice. These patches devoid of huCD45+ cells were sub- µg/ml) or 9.2.27 (100 µg/ml) clone) abolished their stantially smaller and almost absent in control BM (Fig. 3a, engraftment after i.v. transplantation (12 ± 1.7% vs. <2%, p Suppl Videos 1 and 2), thereby confirming that NG2 < 0.001) (Fig. 1a, right panel). Importantly, clinical data blockage efficiently induces egress and mobilization of BM from MLLr-BCP-ALL infants (n = 55) uniformly enrolled MLLr-B-ALL leukemic cells. in the Interfant treatment protocol reveals that patients with refractory or relapsed disease initially had at diagnosis 28% NG2 antagonists synergize with VxL therapy and more NG2-expressing blasts than those patients not render higher CR rates and EFS in preclinical PDX experiencing relapse (53% vs. 41%, p > 0.05). This suggests models of MLLr-B-ALL that NG2 plays a role in vivo in the propagation of MLLr-B- ALL cells. BM stroma confers chemoresistance to leukemic cells in a We then assessed in vitro and in vivo whether anti-NG2 variety of hematological malignancies [32, 33]. We there- (7.1 MoAb) effectively eliminates NG2+ MLLr-B-ALL fore first tested in vitro whether BM-MSCs also protect blasts. In vitro, anti-NG2 treatment was unable to activate NG2+ MLLr-B-ALL primary blasts from VxL. NG2+ effector mechanisms including CDC (Fig. 2a, b) and ADCC blasts pre-exposed to BM-MSC for 30 min or 24 h dis- (Fig. 2c), resulting in no cytotoxicity of NG2-expressing played ∼20% or ∼40% increased resistance to VxL, MLLr-B-ALL cells in 4- and 24-h assays. As a control, respectively, indicating a BM stroma-mediated chemopro- Rituximab (anti-CD20 MoAb) effectively activated both tection of MLLr-B-ALL blasts to induction therapy CDC and ADCC cytotoxicity mechanisms to eliminate (Fig. 3b). Mobilization of leukemic cells from BM to PB is Daudi cells (Fig. 2). We further addressed whether NG2 clinically desirable because circulating blasts become more blockage elicits cytotoxicity in vivo, thus impacting the accessible (and sensitive) to cytotoxic treatments due, in 1562 B. Lopez-Millan et al. Fig. 2 Anti-NG2 MoAb lacks both cellular-mediated (ADCC) and (right panels) with Daudi and SEM-NG2 cells using both anti-NG2 complement-mediated (CDC) cytotoxicity. a CD19, CD20, and (7.1 MoAb) and anti-CD20 (Rituximab) (n = 2). CT: no MoAb con- NG2 surface immunophenotype in Daudi (control) and SEM-NG2 trol. Human AB serum was used for CDC and human PBMCs were cells. b, c CDC (b) and ADCC (c) assays at 4 h (left panels) and 24 h used for ADCC. *p < 0.05; **p < 0.01; ***p < 0.001 part, to their detachment from the chemoprotective BM do this, engrafted PDXs were treated with vehicle, VxL niche [34–36]. Therefore, we next tested whether mobili- alone, VxL plus Ch’ase, or VxL plus 7.1 MoAb, and CR zation of MLLr-B-ALL blasts to PB with the NG2 (MDR <1% leukemic cells in BM (approx. <0.1% in PB) antagonist Ch’ase (Fig. 4) or 7.1 MoAb (Fig. 5) synergizes was assessed at completion of the treatment (day 15)). Mice with VxL treatment in PDX models of MLLr-B-ALL. To were then left untreated for 30 days and relapse and EFS NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1563 Fig. 3 a In situ 3D microscopy imaging of BM from primografts blockage, right panels). Large DAPI+ tissue regions depleted of treated with NG2 antagonists. Representative images of large tissue huCD45+ cells are only observed in the BMs from mice treated with volumes from the femoral diaphysis of transplanted mice treated with 7.1 MoAb. See also Supplementary Videos 1 and 2. b BM-MSCs the vehicle (CTRL, left panels), or with 7.1 MoAb for 7 days (NG2 protect MLLr-B-ALL cells against VxL chemotherapy (n = 2) were determined (Fig. 4a). In contrast to vehicle-treated (Figs. 4b, 5a). Moreover, despite comparable engraftment mice, VxL treatment was highly efficient and engrafted levels (∼24%) in BM at the time of treatment initiation (day PDXs treated with two cycles of VxL (14 days) showed a 0) (Figs. 4c, 5b), the levels of B-ALL engraftment in BM at massive decrease in leukemia burden in both BM and PB the end of treatment (day 15) was ∼3-fold lower when 1564 B. Lopez-Millan et al. either Ch’ase (0.8% vs. 2.2%, Fig. 4b, d) or 7.1 MoAb increased 2-fold when NG2 antagonists were co- (1.3% vs. 2.95%, Fig. 5a, c) were co-administered with administered with VxL: 63% vs. 33% for Ch’ase (Fig. 4d) VxL. As a consequence, the rate of animals achieving CR and 46% vs. 20% for 7.1 MoAb (Fig. 5c). To further assess NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1565 Fig. 4 In vivo treatment with Ch’ase sensitizes blasts to VxL, ren- to achieve a second remission; or (ii) anti-CD19-targeted dering higher CR rates and higher EFS in preclinical PDX models of immunotherapies, which have shown very promising out- MLLr-B-ALL. a Complete experimental design of the preclinical PDX comes although disease-free remissions are usually main- models detailing in vivo treatments with VxL and NG2 antagonists. After 2 complete cycles of VxL chemotherapy, MRD/CR was eval- tained for limited time [37–39]. Furthermore, current uated and relapse was followed-up for up to 30 days. b Monitoring of chemotherapy is associated with morbidity and serious side the levels of leukemic engraftment in PB and BM for the indicated effects such as infertility, impaired development, and treatments. PB engraftment was analyzed weekly. BM leukemic greater risk of secondary neoplasms [12, 13]. Thus, new engraftment was analyzed using BM aspirates at the end of VxL ± Ch’ase treatment (day 15) and at the end of the follow-up period (day targeted therapeutic approaches are urgently needed for de 50). Each line represents the same mouse before and after treatment. c novo high-risk B-ALL and R/R B-ALL. Levels of leukemia engraftment in BM at treatment initiation (day 0) Here we analyzed whether NG2, a surface antigen for VxL or VxL + Ch’ase mice cohorts. d BM levels of MRD at the exclusively expressed in 11q23/MLLr leukemias, represents end of two cycles of VxL ± Ch’ase (day 15). Each dot represents a single mouse. A mouse is considered in CR when the % of blasts in a target for MLLr-B-ALL. The therapeutic interest in tar- BM <1% (blue horizontal dotted line). The light blue bars represent the geting NG2 in MLLr R/R B-ALL is based on recent find- proportion of mice in CR (right Y-axis) for VxL and VxL + Ch’ase ings involving NG2 in leukemia invasiveness, CNS (n = 18 mice/group). e Kaplan–Meier survival curves for 45 days EFS infiltration, and clinical outcome. Indeed, high expression of (n = 16 mice). f Leukemic burden in BM at sacrifice. *p < 0.05; n.s.: no significant differences NG2 in B-ALL blasts has been associated with lower EFS, hyperleukocytosis, and more common relapse [18]. Our robust preclinical PDX models coupled to in situ 3D con- the potential clinical impact of the lower levels of MRD and focal imaging of native BM revealed that NG2 is crucial for higher CR rates, treatment was removed and PDXs were MLLr-B-ALL engraftment upon i.v. transplantation, and followed-up for a 30-day period. Importantly, 45-day EFS that in vivo blockade of NG2 using either a specific inhi- was higher in the group treated with either Ch’ase (65% vs. bitor Ch’ase or an anti-NG2-specific antibody results in an 31%, Fig. 4e) or 7.1 MoAb (55% vs. 40%, Fig. 5d) than extensive mobilization of MLLr-B-ALL blasts from BM to with VxL alone. Accordingly, at the time of sacrifice, mice PB. These findings are in line with previous studies impli- in the VxL-alone group had 50% higher leukemic burden cating NG2 antigen with MLLr leukemia invasiveness and than those receiving VxL combined with NG2 antagonists migration [18], and are consistent with the known biology (Fig. 4f). Collectively, these results indicate that NG2 of NG2. Importantly, anti-NG2 MoAb does not activate blockade overrides BM stroma-mediated chemoprotection either ADCC or CDC cytotoxicity pathways, thus resulting through PB mobilization of MLLr-B-ALL blasts, which in a lack of cytotoxicity in vitro and in vivo, which further consequently are more accessible to conventional che- supports the role of NG2 in regulating MLLr-B-ALL blast motherapy (Fig. 6). mobilization to and engraftment in the BM. NG2 is a transmembrane proteoglycan barely expressed in normal hematopoietic cells [16, 17], but specifically Discussion expressed in ~90% of MLLr acute leukemias [16, 17]. Accordingly, it has been incorporated into diagnostic There has been considerable progress in the past two dec- workflows for leukemia immunophenotyping due to its ades in understanding the biology of B-ALL, which toge- predictive value for MLLr [16, 18–21]. NG2 is a single-pass ther with better molecular diagnosis, patient stratification, type I transmembrane protein that has a very large extra- and therapy, has substantially improved clinical outcomes. cellular domain and short transmembrane and intracyto- VxL, with or without an anthracycline [2], is currently the plasmic domains [40, 41]. The large extracellular domain backbone of induction (or re-induction post-relapse) treat- consists of three subdomains known as D1–3 that signal ment protocols with ∼90% of children entering CR [3] and upon interaction with multiple extracellular matrix compo- an eventual 5-year cure rate of ∼80% [4, 5]. Nevertheless, nents, including laminins (binding through D1) collagen relapsed or refractory B-ALL (R/R B-ALL) is still common types II, V, VI, and fibronectin (binding through D2), and and remains non-curable [6, 7]. Moreover, age <1 year and with tenascin, galectins, and lectins such as P-selectin and MLLr are two major adverse prognostic factors in B-ALL. α3β1 integrin (binding through D1) [41–43]. It is therefore Accordingly, MLLr-B-ALL patients remain high-risk with likely that the NG2 molecule has an important role in the 5-year EFS and OS rates <30% [8–10]. The t(4;11)+/MLL- anchoring of blasts to the BM niche, such that the loss of AF4+ B-ALL is of special interest owing to its extremely such a cellular interaction results in detachment of MLLr short latency and dismal prognosis, common therapy blasts from the BM and their subsequent mobilization to refractoriness, and CNS involvement [11]. In R/R B-ALL PB. This scenario is similar to that of molecules commonly patients, the best therapeutic options are currently: (i) HSCT employed in clinical practice for the mobilization of either following induction into second CR, but they frequently fail hematopoietic progenitors or acute myeloid leukemia 1566 B. Lopez-Millan et al. Fig. 5 In vivo treatment with 7.1 MoAb sensitizes blasts to VxL, treatment initiation (day 0) for VxL or VxL ± 7.1 MoAb mice cohorts. rendering higher CR rates and higher EFS in preclinical PDX models c BM levels of MRD at the end of two cycles of VxL ± 7.1 MoAb (day of MLLr-B-ALL. a Monitoring levels of leukemic grafts in PB and 15). Each dot represents a single mouse. A mouse is considered in CR BM for the indicated treatments. PB engraftment was analyzed when the % of blasts in BM <1% (blue horizontal dotted line). The weekly. BM leukemic engraftment was analyzed using BM aspirates at light blue bars represent the proportion of mice in CR (right Y-axis) for the end of VxL ± 7.1 MoAb treatment (day 15) and at the end of VxL and VxL ± 7.1 MoAb. d Kaplan–Meier survival curves for follow-up period (day 50). Each line represents the same mouse before 45 days EFS (n = 10–11 mice in each group). *p < 0.05; n.s.: no and after treatment. b Levels of leukemia engraftment in BM at significant differences (AML) blasts into PB, such as granulocyte-colony- become sensitized to conventional chemotherapy (VxL) as stimulating factor or plerixafor [44–46]. demonstrated by lower levels of MRD and therefore higher The interaction of leukemic cells with the BM micro- rates of CR at treatment completion, resulting in higher EFS environment in functional niches is a major mechanism and delayed time to relapse. Previous preclinical studies underlying leukemia maintenance by favoring leukemic cell have successfully used an anti-NG2 MoAb to significantly growth and clonal evolution of malignant cells, ultimately reduce both adhesion-induced cell spreading and resulting in therapy resistance and relapse [47–52]. Thus, anchorage-independent growth of melanoma cells [55]. PB-mobilized MLLr-B-ALL blasts not only become more Conversely, ectopic expression of NG2 in NG2-null human accessible but might also be sensitized to conventional melanoma cells led to enhanced tumor spreading, unequi- chemotherapy by overriding BM stroma-mediated chemo- vocally implicating NG2 as a major regulator of migration protection, a major mechanism underlying refractoriness to and invasion in melanoma [56]. Of note, both NG2+ and chemotherapy [53, 54]. Our data further show that MLLr-B- NG2− cells are found in engrafted mice, suggesting a NG2 ALL blasts mobilized to PB with NG2 antagonists do in fact regulation in response to a homeostatic adaptation of MLLr NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1567 Fig. 6 A proposed mechanism of action for NG2 antagonists. NG2 is B-ALL blasts and BM stroma, promoting the mobilization of MLLr-B- well established to bind to common components of the extracellular ALL blasts into PB, thus overriding BM stroma-mediated chemo- matrix including collagen [40, 41, 57]. This supports a model in which protection and making leukemic cells more accessible (sensitive) to in vivo blockade of NG2 may disrupt the interaction between MLLr- conventional chemotherapy leukemic cells. This further supports NG2-targeted treat- a strategy based on a MoAb is expected to display less on- ment (along with current chemotherapy) as a promising target off-tumor toxicity. adjuvant therapy to disturb such homeostatic adaptation of MLLr blasts. Acknowledgements This work has been supported by the European Research Council (CoG-2014-646903 and PoC-2018-811220), the This is the first study targeting NG2 in human leukemias Spanish Ministry of Economy-Competitiveness (SAF2016), and the and provides a proof-of-concept preclinical model for NG2 Catalunya Government (SGR330 and PERIS 2017) to PM, and as a therapeutic target for MLLr-B-ALL (Fig. 6). Because the Asociación Española Contra el Cáncer (AECC), Beca FERO, and the NG2 is a malleable marker with enhanced expression in ISCIII/FEDER (PI17/01028) to CB RT-R, and OM are supported by postdoctoral fellowships from the AECC scientific foundation and the blasts infiltrating extramedullar hematopoietic sites [18], we Catalunya Government (Beatriu de Pinos, BP00048), respectively. PM propose that a MoAb-based or chimeric antigen receptor also acknowledges the financial support from the Obra Social La Caixa- (CAR) T-cell-based immunotherapy approaches will benefit Fundaciò Josep Carreras and “Premio Miguelín”. PM is investigator of current treatment of de novo or R/R MLLr-B-ALL patients. the Spanish Cell Therapy cooperative network (TERCEL). In addition, over 80% of leukemic blasts in MLLr AML are Author contributions BL-M conceived the study, designed, and per- NG2+, suggesting that MLLr AML may also be a candidate formed experiments, analyzed data. DS-M, HR-H, OM, RDdlG,FG-A, high-risk leukemia for NG2 targeting. As a note of caution, RT-R, US, and CN-A performed experiments and analyzed data. JLF we must bear in mind that any new immunotherapy despite and PB provided leukemic samples. CB conceived the study, designed being directed against a specific antigen not expressed in and performed experiments, analyzed dat,a and financially supported the study. PM conceived the study, designed experiments, wrote the healthy tissues/cells could be in the long-term accompanied manuscript, and financially supported the study. by potential secondary events, not anticipated a priori. CAR T cells or bispecific monoclonal antibodies-based clinical Compliance with ethical standards trials have revealed mechanisms of immune escape and even “on-target off-tumor toxicities” toxicities [57]. Conflict of interest The authors declare that they have no conflict of Therefore, future preclinical studies and phase I clinical interest. trials will definitively determine the implementation of a NG2-based immunotherapy for MLLr acute leukemias. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Although Ch’ase and 7.1 MoAb render a very similar effect, 1568 B. Lopez-Millan et al. Open Access This article is licensed under a Creative Commons blasts at different stages of immunophenotypic maturation have Attribution 4.0 International License, which permits use, sharing, stem cell properties. Cancer Cell. 2008;14:47–58. adaptation, distribution and reproduction in any medium or format, as 16. Bueno C, Montes R, Martin L, Prat I, Hernandez MC, Orfao A, long as you give appropriate credit to the original author(s) and the et al. NG2 antigen is expressed in CD34+ HPCs and plasmacy- source, provide a link to the Creative Commons license, and indicate if toid dendritic cell precursors: is NG2 expression in leukemia changes were made. The images or other third party material in this dependent on the target cell where leukemogenesis is triggered? article are included in the article’s Creative Commons license, unless Leukemia. 2008;22:1475–8. indicated otherwise in a credit line to the material. If material is not 17. Menendez P, Bueno C. Expression of NG2 antigen in MLL- included in the article’s Creative Commons license and your intended rearranged acute leukemias: how complex does it get? Leuk Res. use is not permitted by statutory regulation or exceeds the permitted 2011;35:989–90. use, you will need to obtain permission directly from the copyright 18. Prieto C, Lopez-Millan B, Roca-Ho H, Stam RW, Romero-Moya holder. To view a copy of this license, visit http://creativecommons. D, Rodriguez-Baena FJ, et al. NG2 antigen is involved in leuke- org/licenses/by/4.0/. mia invasiveness and central nervous system infiltration in MLL- rearranged infant B-ALL. Leukemia. 2018;32:633–44. 19. Smith FO, Rauch C, Williams DE, March CJ, Arthur D, Hilden J, References et al. The human homologue of rat NG2, a chondroitin sulfate proteoglycan, is not expressed on the cell surface of normal 1. Pui CH, Evans WE. A 50-year journey to cure childhood acute hematopoietic cells but is expressed by acute myeloid leukemia lymphoblastic leukemia. Semin Hematol. 2013;50:185–96. blasts from poor-prognosis patients with abnormalities of chro- 2. Szymanska B, Wilczynska-Kalak U, Kang MH, Liem NL, Carol mosome band 11q23. Blood. 1996;87:1123–33. H, Boehm I, et al. Pharmacokinetic modeling of an induction 20. Behm FG, Smith FO, Raimondi SC, Pui CH, Bernstein ID. regimen for in vivo combined testing of novel drugs against Human homologue of the rat chondroitin sulfate proteoglycan, pediatric acute lymphoblastic leukemia xenografts. PLoS ONE NG2, detected by monoclonal antibody 7.1, identifies childhood 2012;7:e33894. acute lymphoblastic leukemias with t(4;11)(q21; q23) or t(11;19) 3. Pui CH. Recent research advances in childhood acute lympho- (q23;p13) and MLL gene rearrangements. Blood. 1996;87:1134– blastic leukemia. J Formos Med Association. 2010;109:777–87. 9. 4. Pui CH, Mullighan CG, Evans WE, Relling MV. Pediatric acute 21. Wuchter C, Harbott J, Schoch C, Schnittger S, Borkhardt A, lymphoblastic leukemia: where are we going and how do we get Karawajew L, et al. Detection of acute leukemia cells with mixed there? Blood. 2012;120:1165–74. lineage leukemia (MLL) gene rearrangements by flow cytometry 5. Yeoh AE, Tan D, Li CK, Hori H, Tse E, Pui CH, et al. Man- using monoclonal antibody 7.1. Leukemia. 2000;14:1232–8. agement of adult and paediatric acute lymphoblastic leukaemia in 22. Catalina P, Montes R, Ligero G, Sanchez L, de la Cueva T, Bueno Asia: resource-stratified guidelines from the Asian Oncology C, et al. Human ESCs predisposition to karyotypic instability: is a Summit 2013. Lancet Oncol. 2013;14:e508–523. matter of culture adaptation or differential vulnerability among 6. Gaynon PS, Sun W. Oligoclonality and new agent evaluation in hESC lines due to inherent properties? Mol Cancer. 2008;7:76. acute lymphoblastic leukaemia. Br J Haematol. 2016;173:950–7. 23. Sanjuan-Pla A, Romero-Moya D, Prieto C, Bueno C, Bigas A, 7. Chessells JM, Veys P, Kempski H, Henley P, Leiper A, Webb D, Menendez P. Intra-bone marrow transplantation confers superior et al. Long-term follow-up of relapsed childhood acute lympho- multilineage engraftment of murine aorta-gonad mesonephros blastic leukaemia. Br J Haematol. 2003;123:396–405. cells over intravenous transplantation. Stem Cells Dev. 8. Sanjuan-Pla A, Bueno C, Prieto C, Acha P, Stam RW, Marschalek 2016;25:259–65. R, et al. Revisiting the biology of infant t(4;11)/MLL-AF4+ B- 24. Prieto C, Marschalek R, Kuhn A, Bursen A, Bueno C, Menendez cell acute lymphoblastic leukemia. Blood. 2015;126:2676–85. P. The AF4-MLL fusion transiently augments multilineage 9. Thomas M, Gessner A, Vornlocher HP, Hadwiger P, Greil J, hematopoietic engraftment but is not sufficient to initiate leukemia Heidenreich O. Targeting MLL-AF4 with short interfering RNAs in cord blood CD34(+) cells. Oncotarget. 2017;8:81936–41. inhibits clonogenicity and engraftment of t(4;11)-positive human 25. Menendez P, Catalina P, Rodriguez R, Melen GJ, Bueno C, leukemic cells. Blood. 2005;106:3559–66. Arriero M, et al. Bone marrow mesenchymal stem cells from 10. Bueno C, Montes R, Catalina P, Rodriguez R, Menendez P. infants with MLL-AF4+ acute leukemia harbor and express the Insights into the cellular origin and etiology of the infant pro-B MLL-AF4 fusion gene. J Exp Med. 2009;206:3131–41. acute lymphoblastic leukemia with MLL-AF4 rearrangement. 26. Rodriguez R, Rosu-Myles M, Arauzo-Bravo M, Horrillo A, Pan Leukemia. 2011;25:400–10. Q, Gonzalez-Rey E, et al. Human bone marrow stromal cells lose 11. Pui CH. Acute lymphoblastic leukemia in children. Curr Opin immunosuppressive and anti-inflammatory properties upon Oncol. 2000;12:3–12. oncogenic transformation. Stem Cell Rep. 2014;3:606–19. 12. Warris LT, van den Akker EL, Bierings MB, van den Bos C, 27. Diaz de la Guardia R, Lopez-Millan B, Lavoie JR, Bueno C, Zwaan CM, Sassen SD, et al. Acute activation of metabolic Castano J, Gomez-Casares M, et al. Detailed characterization of syndrome components in pediatric acute lymphoblastic leukemia mesenchymal stem/stromal cells from a large cohort of AML patients treated with dexamethasone. PLoS ONE 2016;11: patients demonstrates a definitive link to treatment outcomes. e0158225. Stem Cell Rep. 2017;8:1573–86. 13. Chow EJ, Pihoker C, Friedman DL, Lee SJ, McCune JS, Wharton 28. Romero-Moya D, Bueno C, Montes R, Navarro-Montero O, C, et al. Glucocorticoids and insulin resistance in children with Iborra FJ, Lopez LC, et al. Cord blood-derived CD34+ hemato- acute lymphoblastic leukemia. Pediatr Blood Cancer. poietic cells with low mitochondrial mass are enriched in hema- 2013;60:621–6. topoietic repopulating stem cell function. Haematologica. 14. Rehe K, Wilson K, Bomken S, Williamson D, Irving J, den Boer 2013;98:1022–9. ML, et al. Acute B lymphoblastic leukaemia-propagating cells are 29. Matas-Cespedes A, Vidal-Crespo A, Rodriguez V, Villamor N, present at high frequency in diverse lymphoblast populations. Delgado J, Gine E, et al. The human CD38 monoclonal antibody EMBO Mol Med. 2013;5:38–51. daratumumab shows antitumor activity and hampers leukemia– 15. le Viseur C, Hotfilder M, Bomken S, Wilson K, Rottgers S, microenvironment interactions in chronic lymphocytic leukemia. Schrauder A, et al. In childhood acute lymphoblastic leukemia, Clin Cancer Res. 2017;23:1493–505. NG2 antigen is a therapeutic target for MLL-rearranged B-cell acute lymphoblastic leukemia 1569 30. Montraveta A, Xargay-Torrent S, Lopez-Guerra M, Rosich L, cell subsets in peripheral blood and leukapheresis products from Perez-Galan P, Salaverria I, et al. Synergistic anti-tumor activity breast cancer patients mobilized with SCF plus G-CSF and of acadesine (AICAR) in combination with the anti-CD20 cyclophosphamide. Leukemia. 2001;15:430–9. monoclonal antibody rituximab in in vivo and in vitro models 45. Menendez P, Caballero MD, Prosper F, Del Canizo MC, Perez- of mantle cell lymphoma. Oncotarget. 2014;5:726–39. Simon JA, Mateos MV, et al. The composition of leukapheresis 31. Alvaro Gomariz PMH, Isringhausen Stephan, Suessbier Ute, products impacts on the hematopoietic recovery after autologous Becker Anton, Boss Andreas, Nagasawa Takashi, et al. Quanti- transplantation independently of the mobilization regimen. tative spatial analysis of hematopoiesis-regulating stromal cells in Transfusion. 2002;42:1159–72. the bone marrow microenvironment by 3D microscopy. Nat 46. Goker H, Etgul S, Buyukasik Y. Optimizing mobilization strate- Commun. 2018;9:2532–43. gies in difficult-to-mobilize patients: the role of plerixafor. 32. Korn C, Mendez-Ferrer S. Myeloid malignancies and the micro- Transfus Apher Sci. 2015;53:23–29. environment. Blood. 2017;129:811–22. 47. Tabe Y, Konopleva M. Advances in understanding the leukaemia 33. Karjalainen R, Pemovska T, Popa M, Liu M, Javarappa KK, microenvironment. Br J Haematol. 2014;164:767–78. Majumder MM, et al. JAK1/2 and BCL2 inhibitors synergize to 48. Sison EA, Brown P. The bone marrow microenvironment and counteract bone marrow stromal cell-induced protection of AML. leukemia: biology and therapeutic targeting. Expert Rev Hematol. Blood. 2017;130:789–802. 2011;4:271–83. 34. Nervi B, Ramirez P, Rettig MP, Uy GL, Holt MS, Ritchey JK, 49. Medyouf H, Mossner M, Jann JC, Nolte F, Raffel S, Herrmann C, et al. Chemosensitization of acute myeloid leukemia (AML) fol- et al. Myelodysplastic cells in patients reprogram mesenchymal lowing mobilization by the CXCR4 antagonist AMD3100. Blood. stromal cells to establish a transplantable stem cell niche disease 2009;113:6206–14. unit. Cell Stem Cell. 2014;14:824–37. 35. Bendall L. Extracellular molecules in hematopoietic stem cell 50. Schepers K, Campbell TB, Passegue E. Normal and leukemic mobilisation. Int J Hematol. 2017;105:118–28. stem cell niches: insights and therapeutic opportunities. Cell Stem 36. Aiuti A, Webb IJ, Bleul C, Springer T, Gutierrez-Ramos JC. The Cell. 2015;16:254–67. chemokine SDF-1 is a chemoattractant for human CD34+ 51. Iwamoto S, Mihara K, Downing JR, Pui CH, Campana D. hematopoietic progenitor cells and provides a new mechanism to Mesenchymal cells regulate the response of acute lymphoblastic explain the mobilization of CD34+ progenitors to peripheral leukemia cells to asparaginase. J Clin Invest. 2007;117:1049–57. blood. J Exp Med. 1997;185:111–20. 52. Konopleva M, Konoplev S, Hu W, Zaritskey AY, Afanasiev BV, 37. Terwilliger T, Abdul-Hay M. Acute lymphoblastic leukemia: a com- Andreeff M. Stromal cells prevent apoptosis of AML cells by up- prehensive review and 2017 update. Blood Cancer J. 2017;7:e577. regulation of anti-apoptotic proteins. Leukemia. 2002;16:1713– 38. Jabbour E, O’Brien S, Ravandi F, Kantarjian H. Monoclonal 24. antibodies in acute lymphoblastic leukemia. Blood. 53. Konopleva M, Tabe Y, Zeng Z, Andreeff M. Therapeutic target- 2015;125:4010–6. ing of microenvironmental interactions in leukemia: mechanisms 39. Maino E, Bonifacio M, Scattolin AM, Bassan R. Immunotherapy and approaches. Drug Resist Update. 2009;12:103–13. approaches to treat adult acute lymphoblastic leukemia. Expert 54. Lopez-Millan B, Díaz de la Guardia R, Roca-Ho H, Anguita E, Rev Hematol. 2016;9:563–77. Islam Abul BMMK, Romero-Moya D, et al. IMiDs mobilize acute 40. Campoli M, Ferrone S, Wang X. Functional and clinical relevance myeloid leukemia blasts to peripheral blood through down- of chondroitin sulfate proteoglycan 4. Adv Cancer Res. regulation of CXCR4 but fail to potentiate AraC/idarubicin 2010;109:73–121. activity in preclinical models of non del5q/5q- AML. Oncoim- 41. Stallcup WB. The NG2 proteoglycan: past insights and future munology. 2018;7:e1477460. prospects. J Neurocytol. 2002;31:423–35. 55. Harper JR, Reisfeld RA. Inhibition of anchorage-independent 42. Cooney CA, Jousheghany F, Yao-Borengasser A, Phanavanh B, growth of human melanoma cells by a monoclonal antibody to a Gomes T, Kieber-Emmons AM, et al. Chondroitin sulfates play a chondroitin sulfate proteoglycan. J Natl Cancer Inst. major role in breast cancer metastasis: a role for CSPG4 and 1983;71:259–63. CHST11 gene expression in forming surface P-selectin ligands in 56. Yang J, Price MA, Neudauer CL, Wilson C, Ferrone S, Xia H, aggressive breast cancer cells. Breast Cancer Res. 2011;13:R58. et al. Melanoma chondroitin sulfate proteoglycan enhances FAK 43. Fukushi J, Makagiansar IT, Stallcup WB. NG2 proteoglycan and ERK activation by distinct mechanisms. J Cell Biol. promotes endothelial cell motility and angiogenesis via engage- 2004;165:881–91. ment of galectin-3 and alpha3beta1 integrin. Mol Biol Cell. 57. Aldoss I, Bargou RC, Nagorsen D, Friberg GR, Baeuerle PA, 2004;15:3580–90. Forman SJ. Redirecting T cells to eradicate B-cell acute lym- 44. Menendez P, Prosper F, Bueno C, Arbona C, San Miguel JF, phoblastic leukemia: bispecific T-cell engagers and chimeric Garcia-Conde J, et al. Sequential analysis of CD34+ and CD34− antigen receptors. Leukemia. 2017;31:777–87.