Inhibition of both paracrine and autocrine VEGF/ VEGFR-2 signaling pathways is essential to induce long-term remission of xenotransplanted human leukemias

HL-60 pro-myelomonocytic leukemia cells (obtained from American Type Tissue Culture Collection) were cultured in RPMI (Life Technologies, Grand Island, NY) medium, supplemented with 10% FBS, penicillin, and streptomycin. Primary leukemic samples were isolated from the peripheral blood of leukemic patients, and cultured overnight in serum-free RPMI. For each in vivo experiment, viable leukemic cells were counted and ressuspended in serum-free RPMI.

Human umbilical vein EC (HUVEC) and bone marrow-derived EC (BMEC) were cultured in complete endothelial medium as described (21). For leukemia proliferation experiments, EC were cultured in 6-well plates (Costar) in complete EC medium until the start of the experiment. Before starting the proliferation experiments, EC were placed in serum-free medium.

Leukemic cells were cultured alone, or in the presence of a confluent layer of HUVEC or BMEC in serum-free medium. To avoid direct cell/cell contact, 1 × 10⁵ leukemic cells were placed on transwell (Costar) inserts with a pore size of 0.4 μm. Viable cells were counted by trypan blue exclusion, using a hemacytometer.

Conditioned medium was collected from BMEC or HUVEC monolayers, cultured in serum-free conditions over a 48-h period. Granulocyte–macrophage colony-stimulating factor (GM-CSF) and IL-6 levels were measured by ELISA, following conventional protocols (Cytokine Core Laboratory, Baltimore).

Neutralizing mAbs against human (clone IMC-1C11) or mouse (clone DC101) VEGFR-2 (KDR/Flk-1) and against human (clone 6.12) or mouse (clone mF-1) VEGFR-1 were provided by ImClone Systems. These antibodies are specific for one or the other VEGFR and have strict species specificity. A human Ig preparation (Bayer, Elkhart, IN) was used at the same dose as the neutralizing antibodies, as a negative control.

Nonobese diabetic immunocompromised (NOD-SCID) mice, age-matched (6–8 weeks) and sex-matched, were used in all experiments. Briefly, mice were irradiated with 3.5 Gy from a ¹³⁷Cs g-ray source at a dose rate of ≈0.90 Gy/min and intravenously inoculated with 1 × 10⁷ HL-60 cells in 0.2 ml serum-free RPMI on day 0. Three days after inoculation, mice were divided into four groups of eight mice. One group was treated with 400 μg human IgG thrice weekly (control), one group received IMC-1C11 alone (400 μg/mouse thrice weekly), one was treated with DC101 alone (800 μg/mouse thrice weekly), and the last group was treated with IMC-1C11 and DC101 twice weekly, at the same dosage. Each experiment was done twice. Mice were evaluated for any signs of toxicity throughout the experiment, and survival was recorded. At days 7, 14, 21, 42, 60, and 120 after the start of the experiment, two mice from each group were killed, and their organs and peripheral blood were collected for histology and quantification of plasma VEGF levels and circulating leukemic cells, respectively.

The presence of human leukemia cells in the bone marrow of inoculated mice was quantified by flow cytometry. Briefly, surgically removed tibias day 21 after the start of the experiment were flushed from their bone marrow content with a 19-g syringe, and the single cell suspension was stained for the presence of leukemic cells. Antibodies used for the FACS analysis were: mAb against human CD15 [phycoerythrin (PE)-labeled, clone DU-HL60–3, Sigma]; mAb against human VEGFR-2 (clone 6.64, ImClone Systems), with a secondary goat anti-mouse IgG PE-labeled antibody (Kirkegaard & Perry Laboratories); mAb against murine VEGFR-2 (FITC-labeled DC101) and an isotypic control, MSIgG1 (FITC/PE-labeled clone 2T8–2F5, Coulter). The percentage of human CD15⁺ VEGFR-2⁺ leukemic cells and murine VEGFR-2⁺ cells in mouse bone marrow samples was determined by using a Coulter Elite flow cytometer. These results are representative of three mice/treatment group. Each experiment was done twice.

Human and murine-VEGF plasma levels were evaluated by using commercial ELISA kits (R&D Systems), according to the manufacturer's instructions. These assays had a sensitivity of 7.5 pg/ml. Each sample was analyzed in duplicate, and two samples (a total of six mice) were assayed per time point. The results shown are representative of three mice/time point and two separate experiments.

The FACS results are expressed as mean ± SEM. Statistical analyses were performed by using the unpaired two-tailed Student's t test. For survival analysis, the nonparametric one-tailed Mann–Whitney u test was used.

Freshly isolated leukemias, or their cell line counterpart, if cultured in serum-free condition survive and even show a modest increase in proliferation over a 48-h period (ref. 8 and Fig. 1). Earlier, we showed that leukemia survival in serum-free conditions is caused by the generation of an autocrine VEGF/VEGFR-2 loop (8). However, if cultured in the presence of bone marrow-derived or umbilical vein EC, leukemic cells proliferate significantly more (Fig. 1). This finding suggested that EC support leukemia growth and survival in vitro, perhaps by secreting specific growth factors into the cell culture supernatants. Therefore, we analyzed EC supernatants for the presence of growth factors that promote leukemic growth, such as GM-CSF, IL-6, or IL-8 (Table 1). In the present study, both endothelial sources released significant amounts of GM-CSF, IL-6, or IL-8 into the cell culture supernatants, as measured by ELISA (Table 1). This finding indicates that EC may support leukemia survival and proliferation in a paracrine fashion, by secreting specific growth factors that will subsequently act on the leukemic cells. Importantly, a constant source of growth factors appears to be critical for the survival and proliferation of the leukemic cells, because addition of exogenous GM-CSF or IL-6 to leukemic cultures was less effective at promoting leukemia proliferation (Fig. 1).

As they expand, the leukemic cells produce angiogenic growth factors such as VEGF and fibroblast growth factor, which stimulate both proliferation as well as growth factor production by the endothelial layer. Notably, activation of HUVEC or BMEC by angiogenic growth factors such as VEGF in vitro significantly increased the production of GM-CSF, IL-6, and IL-8, mimicking the coculture system described above (data not shown). Taken together, these results suggest that generation of such paracrine loops in vivo may be essential for leukemia engraftment and proliferation of a malignant clone.

Next, we examined the relative contribution of the paracrine (endothelial-derived) as well as autocrine (endothelial-independent) loops toward leukemic growth and engraftment in vivo.

Sublethally irradiated NOD/SCID mice inoculated with HL-60 leukemic cells, if left untreated, or treated with an irrelevant human IgG, die within 2–3 weeks (Fig. 2A, mean survival time = 18.2 ± 1.3 days. In these mice, clinical signs and symptoms of acute leukemia, such as severe anemia, lymphadenopathy, and cachexia became apparent within 2 weeks. As with leukemic patients (3), clinical progression in inoculated mice is characterized by increased circulating human VEGF levels. As shown in Fig. 2B, human, but not murine, VEGF plasma levels increased in inoculated mice with disease progression.

To investigate whether the growth of leukemia in vivo is angiogenesis-dependent, or is promoted solely via an autocrine VEGF/human VEGFR stimulation, we used neutralizing mAbs to murine (clone DC101) or human (clone IMC-1C11) VEGFR-2 and mAb to murine (clone mF1) or human (clone 6.12) VEGFR-1. These antibodies have strict species specificity and are capable of blocking VEGF-induced receptor phosphorylation and endothelial proliferation (9–11). In addition, we have previously shown that the HL-60 leukemic cells used in our studies express functional VEGFR-2, and therefore respond to treatment with IMC-1C11 (8). Moreover, DC101 (antimurine VEGFR-2) has been shown to delay tumor growth, through inhibition of angiogenesis, in several murine solid tumor models (10).

Treatment of leukemic mice with DC101, three times a week, at 800 μg/injection (a treatment schedule determined for other tumor models), prolonged survival by more than 2-fold (Fig. 2A, mean survival time = 36.8 ± 6.3 days, P < 0.05 compared with the control group), but the mice eventually died within 42 days. Treatment with IMC-1C11 (three times a week, at 400 μg/injection) prolonged survival by more than 3-fold (Fig. 2A, mean survival time = 70.3 ± 3.3 days, P < 0.01) compared with the human IgG-treated group but, again, all of the mice still died within 70–80 days. The results obtained by treating leukemic mice with DC101 show that blocking VEGF-induced angiogenesis in vivo, via interaction with murine VEGFR-2, delays leukemic growth, but is not sufficient to eradicate the disease. Targeting the autocrine VEGF/VEGFR-2 loop on leukemic cells with IMC-1C11 significantly blocked leukemic growth and invasiveness, but also failed to induce long-term remission of inoculated mice. Notably, administration of either mF1 or 6.12 to leukemia-inoculated mice, to target the paracrine or the autocrine VEGF/VEGFR-1 loops, respectively, had no effect on mouse survival. Mice treated with either mF1 or 6.12 died because of leukemia proliferation 23 and 25 days after the start of the experiment, similarly to those in the control group (Fig. 2A). These results demonstrate that the VEGF/VEGFR-2 loops are critical for leukemia proliferation in vivo, whereas VEGFR-1 appears to play a marginal role.

Coadministration of IMC-1C11 and DC101 had a synergistic effect on survival of leukemic mice. Eighty percent (10/12, and 11/12 in a total of two experiments, P < 0.001) of the mice treated with both antibodies survived beyond 100 days (Fig. 2A). Moreover, 40% of these mice achieved prolonged remission (up to 200 days, Fig. 2A) with no clinical symptoms or histological evidence of disease. Given the overall short life expectancy of NOD-SCID mice (≈10–12 months), this result is highly significant. As predicted, plasma levels of human VEGF remained lower in the DC101 + IMC-1C11-treated group, presumably reflecting slower tumor progression (Fig. 2B).

We determined whether treatment of leukemic mice with the anti-VEGFR-2 antibodies also blocked the dissemination of the disease. Gross inspection of control mice on days 7 and 14 revealed marked lymphadenopathy and infiltration of HL-60 cells into the bone marrow, spleen, liver, and lung. Infiltrating HL-60 cells also were found in the liver, spleen, bone marrow, and lungs of control mice on day 21 (Fig. 3A). In addition, bone marrow sections showed abundant vascularization, suggesting increased angiogenesis (Fig. 3A). Similarly, leukemia mice treated with mF1 or 6.12 also had evidence of leukemia proliferation in peripheral organs and bone marrow (data not shown), which correlated with the inability of these antibodies to delay leukemic growth. In contrast, all organs from IMC-1C11- or DC101-treated mice were free of HL-60 cells microscopically on days 7 and 14, and, even after 21 days, tumor infiltration was limited to the perivascular areas of the liver (Fig. 3B). DC101-treated mice did eventually develop lymphadenopathy and massive tumor infiltration of the spleen (Fig. 3B), at which point they succumbed to the proliferating leukemia. In mice treated with IMC-1C11, tumor infiltration was reduced and still limited to the perivascular area of the liver on day 42 (Fig. 3B). However, in mice treated with both antibodies, there was no evidence of disseminated leukemia until day 60 (Fig. 3B), at which time selected liver sections had evidence of micrometastases (Fig. 3B). All mice in this group survived beyond 100 days and, upon achieving remission, had no evidence of metastatic disease up to day 120 (last time point analyzed, data not shown).

To determine the effects of anti-VEGFR-2 antibody therapy on the capacity of human leukemic cells to remain engrafted, we examined the bone marrow cells of leukemic mice by flow cytometric analysis, using human-specific antibodies to leukemic (CD15) and endothelial (VEGFR-2) markers. Twenty one days after the start of the experiment, there were large numbers of human-CD15⁺, VEGFR-2⁺ cells in the control group treated with human-IgG (Fig. 3, human CD15⁺, 7% of total bone marrow cells; VEGFR-2⁺, 5% of bone marrow cells). On the other hand, in IMC-1C11, DC101- or IMC-1C11 + DC101-treated mice, at the same time point, the number of leukemic cells in the bone marrow was significantly reduced, with human CD15⁺ VEGFR-2⁺ cells comprising less than 1% of the total bone marrow cells. Similar results were obtained in peripheral blood samples from the same mice (data not shown). In addition, to confirm that leukemia engraftment increases bone marrow angiogenesis, we analyzed bone marrow samples for the presence of VEGFR-2⁺ (Flk-1⁺) murine EC. Twenty one days after the start of the experiment, the percentage of murine VEGFR-2⁺ bone marrow cells increased from 0.4% in noninoculated normal mice (baseline levels), to over 1.4% in human IgG-treated, leukemia-inoculated mice (Fig. 4). This result correlated with the bone marrow histology (Fig. 2A) and suggests that bone marrow angiogenesis increases after inoculation with leukemic cells. The percentage of murine VEGFR-2⁺ cells in bone marrow aspirates was reduced to baseline levels in IMC-1C11-, DC101-, and IMC-1C11 + DC101-treated mice (Fig. 4). Similar results were obtained in mice treated with IMC-1C11 alone (up to day 60) or the combination of IMC-1C11 + DC101 up to day 120 (data not shown).