Background
Clinical strategies that use cytotoxic drugs to target tumor cells for destruction are frequently confounded by the responses of tumor-associated stromal cells, which can lead to stimulation of tumor growth, angiogenesis, invasion and metastasis and an immune suppressive environment [
1]. Cancer chemotherapy delivered on a metronomic schedule offers a novel approach to this problem by combining direct tumor cell cytotoxicity with repeated disruption of the tumor microenvironment. Clinical metronomic protocols primarily utilize daily low dose schedules and most commonly employ the classic cytotoxic drug cyclophosphamide (CPA) [
2,
3]. Metronomic administration of CPA and other cancer chemotherapeutic agents is thought to improve anti-tumor activity by combining direct tumor cell drug toxicity with tumor endothelial cell-directed anti-angiogenesis while minimizing toxicity to the patient [
4‐
7]. Recent findings implicate additional mechanisms in the action of metronomic chemotherapy, most notably immune-based mechanisms [
2,
8], including the activation of innate immunity by intermittent metronomic drug scheduling [
9,
10]. Thus, CPA administered on an intermittent, every 6-day metronomic schedule stimulates tumor recruitment of macrophages, natural killer (NK) cells, and dendritic cells with regression of large established tumors, as seen in several implanted glioma models [
10]. This potent innate anti-tumor immune response is not achieved using a traditional maximum tolerated dose schedule [
10], nor is it seen using an AUC-equivalent daily low dose metronomic CPA schedule [
9] that models daily metronomic schedules commonly used in the clinic [
2,
3].
Several cancer chemotherapeutic drugs have the potential to stimulate immunogenic cell death, independent of metronomic scheduling [
11,
12]. CPA promotes bone marrow generation of dendritic cell precursors capable of antigen presentation and differentiation of T-helper 17 cells [
13,
14], while doxorubicin can induce CD8 T-cell activation and interferon-γ production [
15]. Moreover, CPA, paclitaxel, temozolomide and vinorelbine can deplete regulatory T-suppressor cells [
16‐
19] and metronomic schedules of docetaxel and gemcitabine suppress myeloid-derived suppressor cells (MDSCs) [
20,
21]. MDSCs inhibit NK cell activity [
22] and thus have the potential to counter the NK cell-dependent regression of tumors treated with CPA on an intermittent metronomic schedule [
10]. Here, we investigate the effects of intermittent metronomic CPA treatment on tumor-associated MDSCs and on MDSC reservoirs in bone marrow and spleen.
Prior studies found that anti-angiogenesis drugs with VEGF receptor tyrosine kinase inhibitory activity block metronomic CPA-activated anti-tumor innate immunity and the associated tumor regression response [
10]. This inhibition of immune cell recruitment could result from the loss of tumor blood vessels trafficking immune cells into the tumor compartment. Alternatively, it could be a more direct consequence of the inhibition of VEGF signaling, which is common to both endothelial and immune cell lineages [
23] and is important for dendritic cell-endothelial cell cross-talk, trans-differentiation [
24] and tumor-associated macrophage infiltration [
25]. Endothelial cell VEGF signaling is also important for chemokine expression and secretion in pro-inflammatory responses [
26], which may be important for metronomic CPA-stimulated anti-tumor immunity [
9,
10]. A third possibility, suggested by the off-target effects of many receptor tyrosine kinase inhibitors (RTKIs), is that receptors other than VEGF receptor (VEGFR), such as C-FMS/CSF1-R on macrophages [
27] and FLT3 on dendritic cells and NK cells [
28], are involved in the observed innate immune cell inhibition. We presently investigate these issues using the VEGFR2-specific inhibitory monoclonal antibody DC101 [
29], which blocks VEGFR2-dependent angiogenesis without off-target effects.
The inhibition of intermittent metronomic CPA-activated anti-tumor innate immunity by VEGF receptor-targeted anti-angiogenic drugs [
10] indicates a need for therapies that circumvent this inhibition. One approach is to employ anti-angiogenesis drugs that act through mechanisms independent of VEGF receptor. Presently, we consider sorafenib [
30], a multi-RTKI with an IC
50 for VEGFR2 > 100-fold higher than the IC
50 values of the VEGFR-selective RTKIs axitinib (AG-013763), cediranib (AZD2171), and AG-028682 [
31‐
34], all of which strongly inhibit metronomic CPA-induced anti-tumor immunity and tumor regression [
10]. Our findings show that sorafenib is highly anti-angiogenic, yet it does not interfere with tumor recruitment of innate immune cells or metronomic CPA-induced tumor regression, supporting the conclusion that inhibition of innate immune cell recruitment is not an intrinsic feature of tumor anti-angiogenesis.
Discussion
Metronomic CPA activates a strong anti-tumor innate immune response that leads to major tumor regression when given on an intermittent, 6-day repeating schedule, but not when using a dose-equivalent daily schedule [
9], as seen in implanted glioma models [
10]. The strong innate immune modulatory effects of this intermittent metronomic CPA regimen are not accompanied by anti-angiogenesis, are seen in both xenograft (rat 9L gliosarcoma and human U251 glioblastoma grown in scid mice) and syngeneic tumor models (mouse GL261 gliomas grown in fully immune competent C57BL/6 mice), and are inhibited by anti-angiogenic drugs that target VEGF receptor tyrosine kinases [
10]. Here, we selected one of these tumor models, 9L, for further elaboration on the role of VEGFR2 signaling for anti-tumor innate immune function, based on its slower innate immune cell recruitment kinetics [
10], which allowed for a longer time period to elucidate early events integral to this response.
While the anti-angiogenic small molecule RTKIs used in our earlier work [
10] show selectivity for VEGF receptors, all RTKIs can be expected to have some off-target activity, suggesting that other receptors could be responsible for the observed inhibition of immune cell recruitment. Here, we report that the VEGFR2-specific inhibitory monoclonal antibody DC101 inhibits metronomic CPA-induced anti-tumor innate immunity and blocks tumor regression, implicating VEGFR2 in the anti-tumor innate immune cell response stimulated by intermittent metronomic CPA treatment. We also show that the interference of VEGF receptor inhibitors with anti-tumor innate immunity can be circumvented by using sorafenib, a multi-RTKI with greatly reduced VEGFR2 inhibitory activity compared to VEGF receptor-targeted anti-angiogenic drugs, which largely spares VEGFR from inhibition (Table
1) [
32,
34]. Thus, the undesirable antagonism previously seen between anti-angiogenic drugs and anti-tumor innate immunity activated by intermittent metronomic chemotherapy [
10] can be avoided by using angiogenesis inhibitors that act
via pathways independent of VEGFR2, which was found here to be essential for metronomic CPA-activated anti-tumor innate immunity.
Metronomic CPA increased tumor levels of CD11b, which marks bone marrow-derived innate immune cells, including monocytes/macrophages, dendritic cells and NK cells [
47], but when co-expressed with a second marker, Gr1, identifies bone marrow MDSCs [
35], which populate distant sites ahead of colonizing metastatic cells, creating conditions favorable for metastatic growth [
48]. MDSCs have also been linked to tumor immune evasion [
35], which may facilitate tumor metastasis. We found that metronomic CPA suppressed CD11b
+/Gr1
+ MDSC reservoirs in bone marrow and spleen, consistent with reports for metronomic schedules of other cancer chemotherapeutic drugs [
20,
21], but did not significantly increase tumor-associated MDSCs, which have the ability to counter innate immunity by suppressing NK cell activity [
22].
Metronomic CPA treatment induced large increases in tumor-associated lysozymes 1 and 2, effectors of macrophage cytotoxicity, suggesting an additional mechanism of metronomic CPA-induced tumor cell lysis that is distinct from the NK cell perforin-mediated granzyme lysis mechanism described earlier [
10]. Macrophages release many cytolytic factors, including lysozymes, following treatment with cancer chemotherapeutic drugs, leading to increased tumoricidal activity through filopodia extension and contact formation with target tumor cells [
49]. We also observed metronomic CPA-stimulated increases in expression of B220, a marker for interferon-producing killer dendritic cells and NK-dendritic cells [
46], suggesting these hybrid innate-adaptive immune cells contribute to innate immune cell-based tumor regression. Interferon-producing killer dendritic cells can kill cancer cells in their own right, but like dendritic cells, they also have the ability to present antigens, bridging innate immune killing and activating adaptive T- and B-cell responses [
46].
The present studies were designed to address several questions raised by our prior finding that VEGF-directed RTKIs can block metro-CPA-induced innate immune responses [
10], namely, whether the observed immune inhibition was: (a) due to the loss of tumor vascularity and hence a route to traffic immune cells into the tumor; (b) a result of off-target effects of the RTKIs (i.e., inhibition of kinases other than VEGFRs); or (c) a consequence of VEGFR inhibition unrelated to anti-angiogenesis. We have now addressed these issues by showing: (a) sorafenib induces a major loss of tumor vascularity without inhibiting innate immune cell recruitment, demonstrating that the loss of vascularity
per se is not the cause of the innate immune cell inhibition; and (b) monoclonal antibody DC101 – which is highly specific for VEGFR2 – recapitulates the immune response inhibitory effects of the small molecule RTKIs that we reported earlier, indicating that innate immune cell inhibition is not the consequence of RTKI off-target effects. Further, (c) our studies with DC101 provide strong evidence that VEGFR2 inhibition, unrelated to the loss of vasculature, is the cause of the observed immune inhibition, establishing VEGFR2 as an essential factor in the innate immune cell response.
Although sorafenib is reported to have VEGFR inhibitory activity based on
in vitro analysis and studies with cultured cell lines [
30,
33], that inhibition occurs at IC
50 values >100-fold higher than inhibition by the three VEGF receptor-selective inhibitors examined previously [
10] (Table
1). Indeed, we found that sorafenib did not inhibit VEGFR tyrosine phosphorylation in tumor blood vessels
in vivo at the dose used in the present study (Figure
3C). In fact, very few studies have reported sorafenib inhibition of VEGFR phosphorylation in either implanted tumor models or in cancer patients
in vivo, and even in those cases where inhibition was reported, the observed decreases in phospho-VEGF were partial, were not localized to tumor blood vessels, and mirrored the general decreases in tumor blood vessel density that result from sorafenib treatment [
50‐
52]. Nevertheless, sorafenib was highly anti-angiogenic in our experiments, presumably due to inhibition of its other tyrosine kinase targets important for tumor angiogenesis, including Raf, c-Kit, PDGFR-β and Flt-3 [
33]. Thus, sorafenib may be used under conditions where VEGFR2 is spared to avoid inhibition of the VEGFR2-dependent anti-tumor innate immune responses stimulated by intermittent metronomic CPA treatment.
Other anti-angiogenesis drugs that operate by a VEGFR2-independent mechanism may similarly be combined with intermittent metronomic therapies. These include Oxi4053, an endothelial cell tubulin-targeting cytotoxic agent, and TPN-470, which induces endothelial cell cell-cycle arrest; both drugs have been successfully combined with metronomic chemotherapy [
4,
53]. In addition, PPARγ agonists can boost the anti-angiogenic activity of metronomic chemotherapy by increasing endothelial cell expression of CD36, which binds TSP1 and initiates the extrinsic pathway of apoptosis [
54]. PPARγ agonists could thus be useful for potentiation of VEGFR2-independent anti-angiogenesis in combination with metronomic chemotherapy. Low-dose metronomic doxorubicin also activates VEGF receptor-independent, endothelial cell extrinsic apoptosis by increasing Fas expression and synergizing with an anti-angiogenic peptide fragment of TSP1 that up regulates FasL [
55]. Thus, a variety of non-VEGF receptor-targeting agents offer viable therapeutic options for combination with metronomic chemotherapy-induced anti-tumor innate immunity.
In a prior study, sorafenib increased tumor metastasis when administered at a dose 6 times higher than that used in the present study (150 mg/kg/day vs. 25 mg/kg/day in this study) [
56]. At that high dose sorafenib is likely to inhibit VEGFR2 signaling and thereby suppress global immune surveillance, leading to the observed increase in metastasis. Further investigation is required to determine whether a low-dose sorafenib regimen, such as that used here, might effect sufficient anti-angiogenesis while avoiding the increase in metastasis seen with several VEGF pathway inhibitory anti-angiogenic drugs [
56,
57], and whether it might be effective in combating metastases when combined with intermittent metronomic CPA. It will also be important to determine whether intermittent metronomic chemotherapy can activate anti-tumor innate immunity at metastatic nodules.
While our results establish that the VEGFR2-specific inhibitor DC101 suppresses tumor regression and the recruitment of innate immune cells (Figure
2), our findings do not require that VEGFR2 necessarily be expressed on the tumor-infiltrating innate immune cells themselves, or if it is, that DC101 inhibition of innate immune cell VEGFR2 signaling be the underlying mechanism for the block in immune cell recruitment. Other mechanisms that should be considered include a requirement for VEGFR2 signaling carried out by immune progenitors, or by one or more tumor-associated cells, including stromal and endothelial cells, or perhaps the tumor cells themselves, for attraction of the innate immune cells to the drug-treated tumors. DC101 can suppress endothelial cell progenitor mobilization from the bone marrow in response to chemotherapy [
41], suggesting that DC101, and the other VEGF receptor-selective drugs [
10], may block tumor recruitment of innate immune cells by inhibiting VEGFR2 signaling required for mobilization of immune cell hematopoietic progenitors [
23]. DC101 can also increase local tumor invasiveness and distant liver and lymph node metastasis [
57]. Our finding of increased SDF1α expression (Additional file
5), which also occurs during chemotherapy-induced circulating endothelial progenitor mobilization [
41], might be indicative of such immune mobilization.
Side effects of VEGF/VEGFR2 antagonists, such as internal bleeding or problems with post-operative wound healing reported for the anti-VEGF monoclonal antibody bevacizumab (Avastin) [
58], could result from decreased leukocyte levels, in particular platelets. Bevacizumab binds and inhibits human VEGF, but not mouse VEGF [
59], and consequently, systemic side effects such as immune suppression, while observable in human patients, would not be manifested with bevacizumab in preclinical mouse models. Several successful clinical protocols combine VEGF pathway-targeted anti-angiogenic dugs with metronomic CPA (e.g., [
60]), however, those protocols commonly use daily low dose CPA delivery, which based on our recent findings would not be expected to activate an innate immune cell response [
9].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JCD contributed to concept design, carried out animal studies and molecular analyses, and drafted the manuscript jointly with DJW. CS carried out additional animal studies and molecular characterization. DJW helped conceive the study, coordinated the overall project and revised and edited the manuscript. All authors read and approved the final manuscript.