Therapy-Induced Acute Recruitment of Circulating Endothelial Progenitor Cells to Tumors

See allHide authors and affiliations

Science  22 Sep 2006:
Vol. 313, Issue 5794, pp. 1785-1787
DOI: 10.1126/science.1127592


The contribution of bone marrow–derived circulating endothelial progenitor cells (CEPs) to tumor angiogenesis has been controversial, primarily because of their low numbers in blood vessels of untreated tumors. We show that treatment of tumor-bearing mice with vascular disrupting agents (VDAs) leads to an acute mobilization of CEPs, which home to the viable tumor rim that characteristically remains after such therapy. Disruption of this CEP spike by antiangiogenic drugs or by genetic manipulation resulted in marked reductions in tumor rim size and blood flow as well as enhanced VDA antitumor activity. These findings also provide a mechanistic rationale for the enhanced efficacy of VDAs when combined with antiangiogenic drugs.

Angiogenesis, the growth of new blood vessels from an existing vasculature, requires expansion of vascular endothelial cells. This can occur locally by proliferation of differentiated endothelial cells, or systemically by mobilization of bone marrow–derived endothelial progenitor cells, which enter the peripheral blood circulation, migrate to sites of angiogenesis, and incorporate into growing vessels (1, 2). The role of such CEPs in tumor angiogenesis is controversial. Estimates of their contribution to the tumor endothelium in untreated tumors range from as much as 10 to 50% (1, 2) to 5% or less (3, 4), with the majority of studies showing figures in the lower range.

To investigate whether the levels of CEP recruitment to the tumor vasculature change during or after certain anticancer therapies, we studied mice treated with VDAs. In contrast to antiangiogenic drugs, which inhibit the formation of new blood vessels, VDAs cause acute occlusion of existing blood vessels, leading to a rapid and massive intratumoral necrosis (5, 6). However, VDA-treated tumors rapidly regrow from a characteristic rim of residual viable cells at the tumor periphery (5, 6) (fig. S1). We hypothesized that a rapid, reactive mobilization and subsequent tumor homing of CEPs might contribute to this regrowth. To test this hypothesis, we first treated non–tumor-bearing BALB/cJ mice with a single dose of a VDA, either combretastin-A4 phosphate (CA4-P) (5, 6) or OXi-4503, a second-generation derivative of CA4-P. Within 4 hours, CEP levels rapidly increased by a factor of 3, returning to basal levels after 24 hours (Fig. 1A and fig. S2A). An OXi-4503 dose range of 25 to 100 mg per kg body weight led to similar increases in viable CEPs (fig. S2B). Total and differential white blood cell (WBC) counts revealed an increase in neutrophils after 4 hours (fig. S2C). Similar changes in CEP levels occurred in other mouse strains (fig. S2D).

Fig. 1.

Elimination of the VDA-induced spike in CEPs, reduction in viable tumor rim blood vessel perfusion as well as blood flow, and increased hypoxia induced by prior treatment with VEGFR-2 monoclonal antibody DC101. (A) Eight-week-old non–tumor-bearing BALB/cJ mice (n = 5 mice per group) were bled from the retro-orbital sinus 4 and 24 hours after they were treated with OXi-4503 (100 mg/kg), DC101 (800 μg per mouse), or a combination of the two drugs, as indicated. CEP levels were determined using four-color flow cytometry as in (17) for each treatment group. Error bars ± SD; **0.05 > P > 0.01, ***P < 0.01. (B to D) The same drug schedule was used on orthotopically implanted MeWo human melanoma tumor cells. Six- to eight-week-old nude mice were subdermally transplanted with MeWo cells (2 × 106) that were allowed to reach ∼500 mm3, at which point they were treated with DC101 (800 μg), OXi-4503 (100 mg/kg), or DC101 24 hours before 100 mg/kg (at the same dosages). Three days after OXi-4503 treatment, tumors were harvested and evaluated for (B) necrosis (scale bar, 100 μm) and (C) hypoxia (green) and perfusion (blue) (scale bar, 50 μm). In a parallel experiment, mice were evaluated for functional blood flow (D) using high-frequency microultrasound (scale bar, 1 mm). See fig. S4 for summary of quantitative data.

Because prior studies have shown that antiangiogenic drugs suppress the mobilization and levels of CEPs (7), we reasoned that the VDA-induced spike might be prevented by prior treatment of mice with an antiangiogenic drug such as DC101, a monoclonal antibody to mouse vascular endothelial growth factor receptor–2 (VEGFR-2) (7, 8). Indeed, the CEP spike was not detectable when DC101 was injected into non–tumor-bearing mice 24 hours before OXi-4503 (Fig. 1A). Comparable experiments with MeWo human melanoma–bearing mice yielded similar results (fig. S3A). Three days after treatment with OXi-4503 alone, the tumors in these mice showed a prominent central necrotic area surrounded by a rim of viable tissue 3.3 ± 0.51 mm across. When DC101 was administered 24 hours before OXi-4503, the size of the viable tumor rim was reduced to 1.12 ± 0.33 mm (fig. S3B). Prevention of VDA-induced CEP mobilization may partially explain previous preclinical data showing an increased efficacy of therapies that combine a VDA with an antiangiogenic drug targeting VEGFR-2 (9).

We next evaluated the effect of the combination therapy (DC101 plus OXi-4503) on tumor vessel blood perfusion and flow on established (500 mm3) subdermally implanted MeWo tumors. On day 4, tumors were removed and analyzed for necrosis, hypoxia, perfusion, and functional blood flow [using speckle-variance analysis by high-frequency microultrasound imaging (10, 11)]. As seen in Fig. 1, B and C, and fig. S4, OXi-4503–treated tumors showed a viable and perfused rim encapsulating a central hypoxic and necrotic areas. A single dose of DC101 produced an increase in the hypoxic but not necrotic areas. However, the combination of DC101 and OXi-4503 produced a profound decrease in perfusion along with marked increases in hypoxia and necrosis. There was a surge in functional blood flow in peripheral functional vessels 3 days (but not 4 hours) after OXi-4503 treatment (Fig. 1D and fig. S4D), but this was not observed when DC101 was administered 24 hours before OXi-4503. As expected, the combination of DC101 and OXi-4503 markedly suppressed tumor growth (fig. S5), consistent with a previous study involving different drugs (9).

To determine whether the VDA-induced spike in CEPs and WBCs is followed by a preferential homing of the cells to the viable tumor rim, we studied lethally irradiated C57Bl/6J mice transplanted with green fluorescent protein–positive (GFP+) bone marrow cells. Human tumor cells cannot be grown in such immunocompetent mice, and thus these mice were used as recipients of a subcutaneous injection of syngeneic Lewis Lung carcinoma (LLC) cells (0.5 × 106). The mice were treated with the same drug combination (DC101 and OXi-4053) and schedule when tumors reached 500 mm3. Untreated and DC101-treated mice showed only a minor incorporation of GFP+ bone marrow–derived cells into the tumor periphery, consistent with previous reports showing low-level incorporation of bone marrow–derived circulating cells, including CEPs in tumor vessels (3, 4, 7), whereas OXi-4503 treated mice showed a substantial number of GFP+ cells at the tumor site, some of which were incorporated into the tumor vasculature, as assessed by staining with CD31 or VEGFR-2 specific antibodies (Fig. 2A and fig. S6, respectively). Prior treatment of the mice with DC101 markedly reduced the number of GFP+ cells in the tumor periphery (fig. S7). These results suggest that bone marrow–derived cells incorporate into or around the tumor vasculature associated with the viable tumor rim and support the hypothesis that the acute CEP spike observed shortly after OXi-4503 treatment contributes to tumor regrowth. It is possible that other types of proangiogenic bone marrow–derived cells are also recruited to the tumor rim, such as CD45+ (hematopoietic) myeloid/monocytic cell populations (12, 13) (fig. S8).

Fig. 2.

Homing and incorporation of GFP+ bone marrow–derived cells in mouse tumors after treatment of the tumor-bearing mice with DC101, OXi-4503, or the combination of the two drugs, and effect of OXi-4503 on CEPs and tumors in Idmut mice. (A) GFP+ bone marrow–transplanted C57Bl/6J mice (n = 5 mice per group) were used as recipients of LLC cells (0.5 × 106) injected subcutaneously and treated with DC101, OXi-4503, and the combination of the two drugs at the same doses and schedule described for Fig. 1, B and C; treatment was initiated when tumor volumes reached 500 mm3. Three days later, tumors were removed and stained for CD31 (red) to mark endothelial cells. GFP+ bone marrow–derived cells are green; scale bars, 50 μm for left images, 20 μm for right images. (B) Eight- to 10-week-old non–tumor-bearing Idmut mice and their age-matched wild-type (wt) controls (n = 4 or 5 mice per group) were treated with OXi-4503 (100 mg/kg). Blood drawn from the retro-orbital sinus at baseline, 4 and 24 hours, was processed for viable CEPs as in (17). Error bars ± SD; **0.05 > P > 0.01. (C) Eight- to 10-week-old Idmut mice, their age-matched wild-type controls, lethally irradiated Idmut mice transplanted with 106 bone marrow cells obtained from wild-type control mice, and lethally irradiated wild-type mice transplanted with 106 bone marrow cells derived from Idmut mice were implanted with LLC cells (0.5 × 107). When tumors reached 500 mm3, treatment with a single dose of OXi-4503 (100 mg/kg) was initiated. Three days later, mice were killed and tumors were harvested and evaluated for necrosis. Scale bar, 100 μm; BM, bone marrow. See fig. S9 for summary of quantitative data.

To confirm that the systemic CEP spike contributes significantly to the growth of the viable tumor rim after VDA treatment, we administered OXi-4503 to Id-1+/–Id-3–/– mutant (Idmut) mice, which are incapable of mobilizing CEPs (1416). As expected, treatment of non–tumor-bearing Idmut mice with a single dose of OXi-4503 did not produce the CEP spike that is seen in wild-type mice (Fig. 2B). To evaluate the antitumor activity of VDAs in such mice, we implanted LLC cells subcutaneously into Idmut mice and their wild-type controls. In addition, to confirm that the VDA effect is due to a systemic deficiency in CEP mobilization and not primarily to a local angiogenic defect, we “rescued” lethally irradiated Idmut mice by transplantation of bone marrow cells derived from wild-type mice as in (14); LLC cells were implanted 4 weeks later. When tumor volumes reached 500 mm3, mice were treated with a single dose of OXi-4503, and 3 days later the tumors were evaluated for levels of necrosis. The tumors in the Idmut or wild-type mice transplanted with Idmut bone marrow had a reduced viable tumor rim, with most of the tumor consisting of necrotic tissue, whereas the tumors in the wild-type or Idmut “rescued” mice showed a more central necrotic area surrounded by a conspicuous thick layer of viable tissue (Fig. 2C and fig. S9). The tumors in Idmut mice also showed extensive areas of hypoxia and minimally perfused rims (fig. S10). In the absence of treatment, little tumor necrosis was detected in wild-type or Idmut mice, the latter due to the treatment being initiated at early stages of tumor growth. Together, these results reinforce our hypothesis that CEPs mobilized from the bone marrow are a major contributor to the growth of the viable tumor rim after treatment with VDAs.

Our results illustrate that although CEP levels in vessels of untreated tumors are typically low, these levels can suddenly rise in response to acute stress, such as that caused by treatment with a VDA and possibly with other treatments such as maximum-tolerated-dose cytotoxic chemotherapy (17). This situation may be analogous to the rapid reactive mobilization and homing of CEPs to damaged vessels or arteries that occurs after pathological cardiovascular events such as myocardial infarcts (18). Our results also provide an additional mechanistic rationale for the enhanced efficacy of VDAs when combined with an antiangiogenic drug (9). Finally, they suggest that when a VDA is to be combined with chemotherapy, consideration should be given to the counterintuitive idea of administering chemotherapy shortly after VDA treatment, rather than the opposite sequence, because of the ability of chemotherapy to target CEPs (17, 19).

Supporting Online Material

Materials and Methods

Figs. S1 to S10


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article