VEGFC Antibody Therapy Drives Differentiation of AML Free

VEGFC is one of the VEGF family members with a unique role in lymphangiogenesis as well as in angiogenesis in normal homeostasis and cancer (1–4). VEGFC can bind to kinase insert domain receptor (KDR, i.e., VEGFR-2) and fms-related tyrosine kinase-4 (FLT-4, i.e., VEGFR-3) receptors expressed by vascular endothelial cells, lymphatic endothelial cells, and leukemic blasts (1–3, 5, 6). KDR is expressed extracellular on the acute myeloid leukemia (AML) cell membrane, intracellular in the cytoplasm and on the nuclear membrane of AML cells, whereas FLT-4 mainly stains positive within the cytoplasm of AML blasts (5–7). This phenomenon implicates that the extrinsic VEGFC/KDR axis is more likely to support AML cells due to the limited availability of FLT-4 in these AML cells.

High VEGFC levels were identified as an independent prognostic factor in AML and associated with decreased complete remission rates and a reduced survival (8). Exogenous VEGFC can protect AML cells from chemotherapy-induced apoptosis (5). We previously showed that endogenous VEGFC expression is associated with decreased drug responsiveness in childhood AML (9). We therefore hypothesized that VEGFC is an important autocrine growth factor involved in CD34⁺ AML blast maintenance.

Current literature supports an important function for VEGFC in AML progression and therapy resistance (5, 8, 9). Nevertheless, the downstream mechanism of VEGFC signaling in AML blasts is still unknown, and its potential as therapeutic target in AML is an unexplored field of research. Therefore, we set out to investigate the contribution of VEGFC on AML cell functions and the associated downstream signal transduction regulation.

Medical Ethical Committee approved METC 2010.036 and 2013.281, University Medical Center Groningen, the Netherlands. After obtaining written-informed consent (according the declaration of Helsinki), patient samples were handled as was previously described (7). Supplementary Table S1 includes patient with AML characteristics as French-American-British (FAB) classification, karyotype, blast (%), VEGFC levels (pg/mL), KDR (%), FLT-4 (%), CD34 (%), and FLT3 mutational status.

THP-1 and OCI AML3 AML cells were obtained from the ATCC (DSMZ), cultured in RPMI-1640 medium (Thermo Fisher), and supplemented with 1% penicillin/streptomycin (Thermo Fisher) and 10% FCS (Bodinco). MS5 bone marrow stromal feeder layer (a kind gift from J.J. Schuringa from the Department of Experimental Hematology, University Medical Center Groningen). Cell lines were all tested mycoplasma free and approximately 25 times passaged. Cell line karyotypes were regularly tested and were maintained among passages.

shRNA sequences targeting VEGFC and VEGFR-2/KDR (Supplementary Table S2) were genetically modified into a pLKO1-mCherry vector, and PCR-amplified FOXO3A was cloned into the pRRL-GFP vector. Lentiviral particles were generated by 293T cells using psPAX2, pMD2.G (VSV-G), and FuGENE (Roche).

Cells were serum blocked and stained with primary antibodies and secondary antibodies (Supplementary Table S2). Intracellular stainings were performed according to the manufacturer's protocol (Fix & Perm; Life Technologies). Annexin V-FITC/propidium iodide (PI) staining for apoptotic cells following manufacturer's protocol (Annexin-V-FLUOS Staining Kit, Roche). Samples were analyzed using LSRII (BD FACS DIVA software; BD Biosciences) and FlowJo software (Tree Star Inc.).

The VEGFC protein expression in patient samples was measured in duplicates using a VEGFC ELISA (R&D Systems) following the manufacturer's protocol.

VGX-100 is a human monoclonal VEGFC antibody (a kind gift from Vegenics Pty Ltd). VGX-100 binds to and precipitates all forms of VEGF-C in both the human and mouse.

CD34⁺ cells were isolated using a MoFlo-XDP sorter (Beckman Coulter). After CD34 sorting, the CD34 percentages exceeded 95% in all samples.

A total of 1,000 CD34⁺ sorted mononuclear cells were cultured for 2 weeks in 1 mL methylcellulose (MethoCult H4435 Enriched, Stem Cell Technologies) according to the manufacturer's protocol, and experiment is performed in duplicate per patient sample.

CD34⁺ sorted blasts are cultured on MS5 mouse bone marrow stromal cells, in Gartner's media, αMEM (Thermo Fisher) containing 12.5% FCS, 12.5% horse serum (Thermo Fisher), 1% penicillin/streptomycin, 57.2 μmol/L β-mercaptoethanol (Sigma-Aldrich), and 1 μmol/L hydrocortisone (Sigma-Aldrich), supplemented with 20 ng/mL trombopoietin (a kind gift from Kirin Brewery), IL3 (Gibco), and G-CSF (Invitrogen), and experiment is performed in duplicate per patient sample.

Long-term culture-initiating cell (LTC-IC) assays plated at a density range (1/5/10/50/100/250/40,000/120,000 cells/well) were subjected to colony-forming cell (CFC) methylcellulose on top of the stroma (MS5) at week 5 of coculturing. Experiment is performed in 10-plo per density per patient sample.

Cytospins were stained with May–Grunwald–Giemsa. Images were taken with a Leica DM 3000 or Leica DM IL microscope with a Leica DFC420C camera (Leica Geosystems B.V.). Histologic analysis of the sternum (bone marrow) and spleens of AML-xenografted mice was outsourced to the Histology Core Facility of VIB, Leuven.

Cells were lysed in Laemmli sample buffer (Bio-Rad). Proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes, incubated overnight with primary antibodies (Supplementary Table S2), washed, and incubated with horseradish peroxidase–conjugated secondary antibodies. Protein bands were visualized by chemiluminescence. Phosphoproteome array (R&D Systems) analysis was performed according to the manufacturer's protocol, and data analysis and normalization were performed as previously described (7).

FLT3 fragment length analysis by fluorescent labeling technology to establish the ratio of the mutant (ITD) FLT3 allele to the wild-type (WT) FLT3 allele. Ratios up to 0.5 are indicative for a heterozygous FLT3-ITD–mutant allele present in 100% of the AML cells [1 ITD peak/(1 ITD peak + 1 WT peak) = 0.5]. Newly diagnosed patients with AML harbor a heterozygous FLT3-ITD mutation in >92% of the cases (10).

Proteomic profiling was performed using newly diagnosed pediatric AML samples (n = 31) and CD34⁺ normal bone marrow (NBM) samples (n = 10) using reversed phase protein array (RPPA), as described previously (11).

FOXO3A, CD11b, and p21 mRNA expressions together with HPRT as a reference gene were analyzed in triplicates using SYBR Green qRT-PCR (Bio-Rad Laboratories). Relative mRNA expression from triplicates was determined using the ΔΔC_t method (primer sequences in Supplementary Table S2).

NOD-SCID/IL2γ^−/− (NSG) mice were purchased from Charles River. This animal study was approved by the ethical animal committee at KU Leuven (P262/2015). Animals received anti-VEGFC 40 mg/kg treatment twice a week via i.p. injections. Animals were injected with 10⁶ primary AML cells i.v. white blood cell (WBC) counts were measured using a micro-semi C-reactive protein hematology analyzer (Axonlab).

Statistical package for the social science (SPSS 17) software was used for graphing box plots. The Mann–Whitney U test was used to determine differences between AML and NBM or two experimental groups of mice, and two-tailed Student t tests or a paired sample t test was used for analysis comparing untreated and treated AML cells based upon Levene test for equality of variance, and the Kruskal–Wallis test was used to define significant differences between more than two groups.

VEGFC is an important prognostic factor in AML supporting AML blast growth and apoptosis-evading signals (Supplementary Fig. S1A–S1C). The CD34⁺ and CD34⁻ cell populations within primary AML patient samples expressed significantly higher levels of VEGFC as compared with NBM controls (Fig. 1A; Kruskal–Wallis test, P = 0.013). Associated to higher VEGFC expression, primary AML patient samples present elevated KDR membrane protein expression levels (Fig. 1A; Mann–Whitney U test, P = 0.001), whereas FLT-4 membrane protein expression was absent (Fig. 1A; Mann–Whitney U test, P = 0.381; ref. 7). In support of these data, VEGFC and KDR knockdown effects on the proliferation of AML cell lines were comparable (Supplementary Fig. S2A). Cell-cycle inhibitor p21 mRNA expression was significantly induced in anti–VEGFC-treated as well as VEGFC knockdown AML cells (Supplementary Fig. S2B). In addition, VEGFC-supporting effects on AML cell growth were suppressed in KDR-knockdown cells (Supplementary Fig. S2C). These findings challenged us to explore VEGFC (30 μg/mL) monoclonal antibody treatment effects on AML cell functions (Fig. 1B).

In KDR-expressing AML cell line THP-1, VEGFC antibody therapy significantly induced myelocytic differentiation, supported by an increased population of cells that express differentiation markers CD11b and CD14 (Fig. 1C and D and Supplementary Fig. S3A; Student t test, both P < 0.05. In addition, VEGFC antibody therapy induced apoptosis in a dose-dependent matter (Fig. 1E). VEGFC antibody therapy reduced KDR membrane expression, which implicates an eradication of the VEGFC/KDR axis in these leukemic cells (Fig. 1F).

Next, VEGFC antibody therapy effect on CD34⁺ primary AML samples (Supplementary Fig. S3B and S3C) was examined in a variety of AML stem/progenitor cell assays. CFC assays (3D semisolid media) highlight a 25% reduction in four AML patient samples, and one AML patient sample showed decreased colony formation solely after serial replate (Fig. 2A, combining all performed CD34⁺ AML patient CFC assays, Mann–Whitney U test, P = 0.0192). VEGFC antibody therapy–suppressive colony formation was supported by 35% lower total CFC cell counts in all AML samples (Fig. 2A; Mann–Whitney U test, P = 0.0028). In LTC-IC assays (3D coculture assay), VEGFC antibody therapy decreased the outgrowth of CD34⁺ AML cells in 6 of the 7 AML patient samples, overall reducing the LTC-IC outgrowth by 28% (Fig. 2B; Mann–Whitney U test, P = 0.003). Although LTC-IC assay outgrowth was reduced by VEGFC antibody therapy, LTC-IC cultures were retained in the presence of VEGFC antibody therapy. Limiting dilution LTC-IC assays revealed a further decrease up to 49% in the CD34⁺-initiating leukemic cell outgrowth potential in the presence of anti-VEGFC (Fig. 2C and D; Mann–Whitney U test, P = 0.003). Morphologic analysis revealed VEGFC antibody therapy–induced myelocytic differentiation that appeared already after 1 week of treatment (Fig. 2E and F).

Besides an overall 28% reduction in the outgrowth of VEGFC antibody–treated CD34⁺ AML blasts, a significant 3.3-fold induction of differentiation along the myelocytic lineage could be appreciated in LTC-IC assays (Fig. 2G; Mann–Whitney U test, P = 0.001). Differentiation marker analysis confirmed that increasing percentages of cells stained positive for CD38, CD11b, and CD14 or CD15 cells in liquid cultures, CFC assays, and LTC-IC assays as compared with untreated controls (Fig. 2H and Supplementary Fig. S3D and S3E), whereas CD34 percentages were decreased. In addition, VEGFC antibody–treated cultures presented increased percentages of apoptotic cells (Fig. 2I and Supplementary Fig. S3D).

FLT3-ITD fragment analysis showed identical ratios of heterozygous FLT3-ITD–mutant cells in untreated and VEGFC antibody–treated cultures in CFC and LTC-IC assays (Supplementary Table S3), supporting that anti-VEGFC treatment affects the leukemic clone. Although the FLT3-ITD leukemic cells were affected by VEGFC antibody–enforced myelomonocytic differentiation, the cellular responses for FLT3-WT AML samples were superior to the FLT3-ITD AML samples (Supplementary Fig. S4A; Mann–Whitney U test, P = 0.042). VEGFC antibody therapy of control CD34⁺ NBM cultures presented an approximate 3- to 4-week latency in myelocytic lineage skewing as compared with AML CD34⁺ cells (Supplementary Fig. S4B and S4C). Overall, these findings highlight that VEGFC antibody treatment is a novel new differentiation therapy, which targets the leukemic clonogenic capacity of CD34⁺ AML blasts.

As a first approach to define VEGFC downstream targets, we analyzed phosphoproteome arrays of three independent untreated and VEGFC antibody–treated AML samples. Phosphorylation of MEK1/2 (S218/S222, S222/S226), AMPKα2 (T172), HSP27 (S78/S82), paxillin (Y118), STAT2 (Y689), and STAT5_b (Y699) was significantly reduced in anti–VEGFC-treated AML samples (Fig. 3A, paired sample t test, mean ± SEM; *, P < 0.05). Decreased phosphorylation of specifically ERK1/2 was confirmed by immunoblot analysis, and STAT5a/b was reduced in some cases in anti–VEGFC-treated CD34⁺ AML samples, and not in CD34⁺ NBM (Fig. 3B; Mann–Whitney U test, P = 0.008 for Erk, and P = 0.151 for STAT5). In the previously performed LTC-IC assays in limiting dilutions, we observed a loss of erythropoiesis in some but not all AML patient samples, which is supported by reduced STAT5 phosphorylation. Although reduced levels of Erk1/2 phosphorylation can explain a potential drop in expansion potential of the AML blasts by VEGFC antibody therapy, these findings cannot explain the induction of myelomonocytic differentiation (7).

Next, we combined flow cytometry VEGFC and KDR protein expression analysis together with RPPA analysis of the same set of pediatric AML samples. The VEGFC/KDR protein association network showed strong significant overlapping correlations for CCND3, LGALS3, FOXO3 (S318/321), PRKCD (S645), KIT, LSD1, NPM1, EIF2AK2, PTPN11, SSBP2, and STAT5A/B (Fig. 3C, Pearson correlations, all P < 0.05).

FOXO3 suppression is described to mediate differentiation of the AML blasts (12). In line with previous reports in adult AML samples, the basal and phosphorylated FOXO3A protein expression levels were significantly increased in pediatric AML as compared with CD34⁺ NBM samples (Supplementary Fig. S4D; Student t test, both P < 0.001; ref. 13). Immunoblot and flow cytometry analysis revealed that FOXO3A protein expression was decreased upon VEGFC antibody treatment in THP-1 cells and primary AML samples (Fig. 3D). To define whether VEGFC antibody therapy–induced myelomonocytic differentiation was facilitated via its suppression of FOXO3A, we generated THP-1 FOXO3A overexpression cells. Constitutive FOXO3A overexpression was shown to rescue the anti–VEGFC-induced expression of CD11b in a dose-dependent manner (Fig. 3E and Supplementary Fig. S4E and S4F, Student t test, 30 μg/mL and P = 0.027; 60 μg/mL and P = 0.019). These findings implicate that anti–VEGFC-induced differentiation of leukemic cells was enforced via the suppression of FOXO3A.

To investigate the in vivo VEGFC-targeting efficacy, we injected a primary AML patient sample (EVI1 ASXL1) into NSG mice that progressively developed leukemias. These patient-derived AML xenografted mice were treated with DMSO or VEGFC antibody therapy. Leukemia was presented by increased WBC counts in the peripheral blood of these animals. The WBC counts were significantly reduced by VEGFC antibody therapy (Fig. 4A; Mann–Whitney U test, P = 0.014). Upon disease progression, histologic bone marrow examination showed that the human AML blast population was only slightly reduced in VEGFC antibody–treated animals (Fig. 4B; Supplementary Fig. S5; Mann–Whitney U test, P = 0.127). In the bone marrow, we observed a significant induction of eosinophilic compartment in VEGFC antibody–treated mice, supported by elevated levels of human CD11b expression (Fig. 4B and C; Mann–Whitney U test, histology P = 0.046, and flow cytometry P = 0.007). The spleens of VEGFC antibody–treated human PDX AML mice showed a minor decrease in size (Fig. 4D). When focusing on spleen infiltration of human xenografted AML cells, histologic examinations revealed a significant reduction in the amount of AML blasts that localized to the spleens in VEGFC antibody–treated mice (Fig. 4D; Supplementary Fig. S5; Mann–Whitney U test, P = 0.011). The overall in vivo efficacy of VEGFC antibody treatment was characterized by a modest reduction in human PDX AML blast homing to the bone marrow of NSG mice, leading to a stronger decrease in human PDX AML engraftment to secondary AML sites as the spleen, where we observed an approximately 50% reduction of human AML blasts. Taken together, this in vivo study shows that VEGFC antibody therapy suppresses the AML progression in vivo via the induction of differentiation.

VEGFC has been shown to be an independent prognostic factor and showed to interfere with AML survival in vivo and ex vivo (5, 8, 9). Our study highlights VEGFC-targeted treatment as potential new differentiation therapy in AML. This study is one of the few that showed potent differentiation of the CD34⁺ leukemic clone by VEGFC-targeting antibody therapy in vitro and in vivo. The modest VEGFC antibody therapy–mediated reduction of AML blasts in the bone marrow of NSG mice in vivo underscores the supportive therapeutic potential of VEGFC-targeting differentiation therapy in addition to conventional treatment regimens for patients with AML. ATRA is a differentiation therapy available as conventional therapy in the clinic, applied to all patients with acute promyelocytic leukemia (APL) that harbor the PML-RARA fusion protein. This differentiation therapy significantly improved the outcome of APL (14, 15). More recently, the IDH2 inhibitor Enasidenib was approved in the clinics as new differentiation therapy of IDH2-mutant AMLs (16).

VEGFC targeting antibody therapy is currently under investigation in a phase I clinical trial in combination with Bevacizumab (VEGFA-targeting antibody) for advanced solid tumors using a maximum dosage of 20 mg/kg, which showed to be well tolerated (NCT01514123), and final results should be available soon. VEGFC-targeting antibody therapy in mice was previously shown using daily dosage of 20 mg/kg, which was well tolerated (17, 18). In our study, VEGFC-targeting antibody therapy at 40 mg/kg twice weekly in NSG mice did not affect the animal body weight nor showed aberrant histology of organs, and was therefore well tolerated.

Notably, we found that VEGFC antibody treatment blocked the erythroid outgrowth of 2 of 5 patient samples in long-term CD34⁺ AML cultures, which can be caused by the inhibition of STAT5 phosphorylation that is known to guide erythropoiesis (19, 20). Evidence of similar effects should be paid attention to in the ongoing clinical trial of this compound. The VEGFC-targeting therapeutic approach might be useful for other cancer subtypes as well, as for example high VEGFC expression levels have been shown to modulate the breast cancer–metastasizing capacity (4).

Our study describes that VEGFC-mediated FOXO3A levels are important for the preservation of immature AML blasts. Interactions between AMPKα2 and FOXO3A have previously been described (21). We speculate that the decreased levels of AMPKα2 protein phosphorylation by VEGFC antibody treatment may be the key substrate for FOXO3A to control AML cell fate. So far, our findings indicate an important regulatory function for VEGFC in CD34⁺ AML cell fate decisions. Anti-VEGFC therapy enforced CD34⁺ AML blast myelocytic differentiation by FOXO3A suppression, creating new opportunities for differentiation therapy besides high-dose chemotherapy in AML.

No potential conflicts of interest were disclosed.

Conception and design: K.R. Kampen, E.S.J.M. De Bont

Development of methodology: K.R. Kampen, H. Mahmud, E.S.J.M. De Bont

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.R. Kampen, H. Mahmud, A.B. Mulder, H.J.M.P. Verhagen, L. Smit, S.M. Kornblau, E.S.J.M. De Bont

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.R. Kampen, F.J.G. Scherpen, A.B. Mulder, V. Guryev, S.M. Kornblau, E.S.J.M. De Bont

Writing, review, and/or revision of the manuscript: K.R. Kampen, H. Mahmud, K.D. Keersmaecker, S.M. Kornblau, E.S.J.M. De Bont

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F.J.G. Scherpen, A.B. Mulder, V. Guryev, L. Smit

Study supervision: A. ter Elst, K.D. Keersmaecker, E.S.J.M. De Bont

Other (performed experiments): H. Mahmud

We thank Megan E. Baldwin and Robert Klupacs from Vegenics for their generous supply of anti-VEGFC treatment reagent (VGX-100). We thank Kirin Brewery for providing TPO used in LTC-IC assays. Henk Moes, Roelof Jan van der Lei, and Geert Mesander assisted in cell sorting. J.J. Schuringa provided the MS5 feeder cell line. We thank Bart-Jan Wierenga for the lentiviral shRNA plasmid. The authors would like to thank the patients who donated leukemia specimens and nurse practitioners and clinicians who acquired specimens. The authors thank Hein Schepers for sharing the FOXO3A overexpression plasmid.

K.R. Kampen was supported by a PhD grant from the Foundation for Pediatric Oncology Groningen, the Netherlands (SKOG, 12001); animal experiments were granted by Foundation Beatrix Children's Hospital (2014 to K.R. Kampen); a subsequent grant was received from the Jan Kornelis de Cock Stichting (project code 2015-43 to K.R. Kampen); and a postdoctoral fellowship was received from Lady Tata Memorial Trust International Award for Research in Leukaemia (2016–2017). A. ter Elst and H. Mahmud shared the KiKa grant for the kinomics/proteomics AML project (2010-57 to E.S.J.M. de Bont and S.M. Kornblau).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.