Introduction
In addition to factors intrinsic to tumor cells, elements of the tumor microenvironment also have profound influences on mammary tumor progression and metastasis [
1‐
3]. Even when mammary epithelial cells are transformed by oncogenes, such as mouse mammary tumor virus-driven polyoma middle T (MMTV-PyMT) [
4], stromal/epigenetic factors are required for progression from hyperplasias to malignancies [
5,
6]. Key mammary tumor stromal elements include the vasculature, adipocytes, fibroblasts, myeloid cells, and the extracellular matrix, as well as matrix-associated proteases and growth factors. Of these stromal components, the vasculature is the most universal and widely recognized [
7]. The transition of hyperplastic foci to neoplasms requires the recruitment of a vascular supply, an event known as the angiogenic switch [
7,
8]. Effective tumor vascularization is critical not only for growth of the primary neoplasm, but also for tumor metastasis to distant sites.
We have shown that the NG2 proteoglycan (also known as CSPG4, AN2, HMPG, and HMW-MAA) is a prominent cell surface component of microvascular pericytes and can serve as a reliable marker for detection of these cells during the early stages of microvessel development [
9‐
12]. In addition, because NG2 is important for cell proliferation, motility, and cell-cell interaction [
13], genetic ablation of NG2 [
14] results in deficient vascularization in several models of postnatal angiogenesis. In both ischemic retinal vascularization and corneal angiogenesis, the NG2 null mouse exhibits reduced microvessel formation due to retarded pericyte function. In these models, deficits in both pericyte recruitment and proliferation lead to reduced pericyte investment of endothelial cells [
15]. In an allograft model of brain tumor progression, ablation of NG2 causes a several-fold reduction in tumor progression due to deficits in pericyte/endothelial cell interactions that lead to poor vascular function. In particular, the reduced ensheathment of endothelial cells by NG2 null pericytes causes deficiencies in basal lamina assembly, vessel patency, and vessel integrity that compromise vessel performance [
16]. These results emphasize the functional importance of NG2 in stimulating pericyte proliferation and motility, possibly via NG2-mediated enhancement of pericyte responses to growth factors such as FGF2 [
10,
12,
13], as well as the role of NG2 in mediating β1 integrin activation that promotes pericyte/endothelial cell interaction during early phases of neovascularization [
17].
The MMTV-PyMT transgenic mouse provides a means of studying the stromal role of NG2 in a model of spontaneous breast cancer initiation and progression. Although in some cases of human basal-like breast cancer NG2 is reportedly expressed by tumor cells that are triple-negative for estrogen receptor, progesterone receptor, and HER2 [
18,
19], we have found that NG2 is not expressed by mammary tumor cells in the MMTV-PyMT mouse [
11,
20]. Thus, in the transgenic mouse model, any effects of NG2 ablation on mammary tumor progression must be due to alterations in stromal influences. The MMTV-PyMT mouse is ideal for this work for several reasons. First, mammary tumors occur in 100% of female MMTV-PyMT mice [
21]. Second, tumor onset is rapid compared to most other mammary tumor models [
22]. Third, the model exhibits many features of human breast cancer [
4]. Fourth, the fact that NG2 is not expressed by the mammary tumor cells in this model allows us to focus on the stromal, as opposed to tumor cell-autonomous, roles of the proteoglycan. This pattern of expression is relevant to human breast cancer, where NG2 is also strongly expressed by tumor stromal elements. Stromal roles of NG2 are equally relevant to NG2-positive and NG2-negative mammary tumor types. Here we demonstrate that ablation of NG2 in the MMTV-PyMT mouse causes vascular deficits during the early stages of tumor development and that these deficits correlate with a reduction in early mammary tumor growth.
Materials and methods
Animals
Wild type (NG2+/+) and NG2 null (NG2-/-) females [
14] were crossed with males carrying the mouse mammary tumor virus promoter-driven polyoma middle T transgene (Tg) (MMTV-PyMT) [
21] to generate C57Bl/6 wild type and NG2 null versions of this breast cancer model [
23]. β-actin/EGFP transgenic mice (C57Bl/6; Jackson Laboratories, Bar Harbor, ME, USA) were maintained on both wild type and NG2 null backgrounds. Mice were housed in the Sanford-Burnham Vivarium (fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care). All animal procedures were performed in accordance with Office of Laboratory Animal Welfare regulations and were approved by Sanford-Burnham Institutional Animal Care and Use Committee review prior to execution.
Human tissues
Samples of human ductal adenocarcinoma were obtained from the Sanford-Burnham tumor bank. Surgical samples were obtained with the consent of the patients, and were taken prior to initiation of any treatment. Specimens were fixed in formalin at the time of surgical resection.
Tumor cell lines
The Py230 and Py8119 mammary tumor cells lines were established from spontaneous mammary tumors arising in C57Bl/6 MMTV-PyMT females. Tumors were minced and rotated at 250 rpm at 37°C for three hours in 3 ml of Ham's F12K medium containing 1 mg/ml collagenase (Worthington Biochemicals, Lakewood, NJ, USA), 2 mg/ml soybean trypsin inhibitor (Sigma-Aldrich, St. Louis, MO, USA), and 2% BSA. After addition of FCS-containing medium, the suspension was passed through a 70 μm nylon filter (Fisher Scientific, Pittsburg, PA, USA). Single cells were pelleted by centrifugation and cultured in Ham's F12K medium containing 5% FCS, 2.5 μg/ml amphotericin B, 50 μg/ml gentamycin, and MITO+. Cell lines were cloned by limiting dilution in this same medium.
Spontaneous tumor progression
For monitoring the progression of spontaneous MMTV-PyMT neoplasms, #4 mammary glands were spread as whole mounts on glass slides and stained with carmine-alum [
6,
24] to visualize the mammary epithelium, lymph nodes, mammary intraepithelial neoplasias (MINs), and incipient tumors. Early neoplastic progression (weeks 6 to 12) was quantified by using image analysis to determine the total area occupied by MINs. This analysis of MIN area did not include sub-areolar tumors, which were the earliest neoplasms to develop in both wild type and NG2 null MMTV-PyMT females. During weeks 14 to 20, the total mammary tumor burden was determined by weighing tumors dissected from all mammary glands.
Tumor transplantation/engraftment
For mammary tumor transplantation, tumor fragments (1 mm3) were prepared from a single tumor (1 cm in diameter) taken from a wild type MMTV-PyMT donor, and transplanted bilaterally into the #4 fat pads of wild type and NG2 null recipient females. Tumor sites were palpated three times per week to determine the initial time at which tumors could be detected.
Engraftment of mammary tumor cell lines was accomplished by injection of Py230 and Py8119 cells (106 cells in 50 μl of PBS) into the #2 and #4 mammary fat pads of wild type and NG2 null mice. Tumor sites were palpated three times weekly to determine the initial time at which tumors could be detected.
Immunostaining and microscopy
Dissected tumors or tumor-containing tissues were fixed in 1% paraformaldehyde and cryoprotected by overnight immersion in 20% sucrose. Tissue blocks were frozen in Optimal Cutting Temperature embedding medium (OCT, Sakura Finetek, Torrance, CA, USA) and used to prepare 20 μm sections on a Reichert-Jung 1800 cryostat microtome. Immunofluorescence was performed essentially as described [
9] using the following primary antibodies: rabbit and guinea pig anti-NG2 [
9,
25], rat anti-mouse CD31, rat anti-mouse F4/80, and rat anti-mouse CD11b (BD Biosciences, La Jolla, CA, USA), mouse anti-α-smooth muscle actin (αSMA, Sigma-Aldrich), rabbit anti-desmin (Millipore, Billerica, MA, USA), rabbit anti-collagen IV (Millipore), and goat antibodies against VEGF and VEGFR-3 (R&D Systems, Minneapolis, MN, USA). Images were acquired using the Fluoview 1000 Laser Point Scanning Confocal Microscope (Olympus, Tokyo, Japan) and the Radiance 2100 Multiphoton Confocal Microscope (BioRad, Hercules, CA, USA). Additional fluorescence imaging was performed with the TE300 Nikon Fluorescence Microscope (Nikon, Tokyo, Japan). Image analysis was performed using Image-Pro Plus 4.5 software (Media Cybernetics Inc, Bethesda, MD, USA).
Bone marrow transplantation
Bone marrow transplantation from female β-actin/EGFP donors to female MMTV-PyMT recipients was carried out as previously described [
12]. Bone marrow was collected from tibiae and femurs of both wild type and NG2 null donors. Recipients at eight weeks of age (both wild type and NG2 null MMTV-PyMT females) were gamma irradiated with two doses of 5 Gy each, administered three hours apart, and were immediately reconstituted by retro-orbital injection of 5 × 10
5 bone marrow cells in 100 μl of Ringer's solution. Wild type bone marrow was transplanted to both wild type and NG2 null recipients, and a parallel procedure was performed with NG2 null bone marrow. Engrafted mice were maintained on antibiotic water (1.1 g/L neomycin sulfate and 455 mg/l Polymyxin B) for six weeks. Peripheral blood samples were collected from 14-week old mice for flow cytometric analysis of the extent of EGFP engraftment. Animals with at least 75% engraftment were utilized for analysis of macrophage populations as described below. In C57Bl/6 MMTV-PyMT mice, 14 weeks is roughly the earliest time at which mammary tumors can be detected by palpation [
23]. Tumors were collected for flow cytometric analysis at ages ranging from 16 to 18 weeks.
Flow cytometry
Spontaneous MMTV-PyMT mammary tumors (between 1 to 2 cm3 in volume) were used for flow cytometric analysis of myeloid cell populations derived from β-actin/EGFP bone marrow donors. Tumors were taken from four sets of mice: wild type recipients engrafted with wild type bone marrow, wild type recipients engrafted with NG2 null bone marrow, NG2 null recipients engrafted with wild type bone marrow, and NG2 null recipients engrafted with NG2 null bone marrow. Tumor-bearing mice under Avertin anesthesia were perfused transcardially with 10 ml of cold saline. Tumors were dissected free of normal tissue, finely minced with a razor blade in (D)MEM/F12 medium, and digested with 2 mg/ml collagenase for one hour at 37°C. Digests were filtered through 70 μm nylon mesh to produce single cell suspensions, which were treated on ice for 30 minutes with Live/Dead Aqua (Invitrogen, Carlsbad, CA, USA) to allow exclusion of dead cells from flow cytometric analysis. Cells were then washed in PBS, pelleted at 1000 × g, and resuspended at a density of 105 cells in 100 μl PBS. These cell suspensions were treated on ice for 30 minutes with the following fluorochrome-labeled antibodies: CD11b APC, 1/100 dilution (eBioscience; San Diego, CA, USA); F4/80 PE-Cy5, 1/20 dilution (Biolegend; San Diego, CA, USA); CD45 PE-Cy7, 1/50 dilution (eBioscience); Gr1 APC-Cy7, 1/100 (Biolegend); Tie2 PE, 1/50 dilution (eBioscience); CD206 Alexa Fluor 647, 1/20 dilution (Biolegend); and CD11c Cy7, 1/50 dilution (eBioscience).
Labeled cells were pelleted at 1000 × g, resuspended for five minutes at room temperature in 100 μl of 1% paraformaldehyde, and then brought to 400 μl with PBS. Fluorescence activated cell sorting (FACS) was performed using the LSR Fortessa instrument (BD Biosciences; La Jolla, CA, USA). FloJo software (Tree Star, Inc; Ashland OR, USA) was used for quantitative analysis of the flow cytometric data.
Vessel leakiness and tumor hypoxia
Leakiness of mammary tumor vessels and tumor hypoxia were determined after intravenous injection of fluorescein isothiocyanate (FITC)-dextran (250 kDa; Sigma-Aldrich) or pimonidazole hydrochloride (HPI Inc, Burlington, MA, USA), respectively [
16].
Statistical analysis
Quantitative results are expressed as means ± SE. Most statistical analyses were performed using two-tailed t-tests. Wilcoxon signed rank tests were used to evaluate statistical significance in the case of tumor onset studies. P values less than 0.05 were considered statistically significant.
Discussion
Previous work has shown that the NG2 proteoglycan promotes cell proliferation and motility, along with cell-cell and cell-matrix interactions [
13]. Mechanisms underlying these effects involve the ability of NG2 to potentiate cellular responses to growth factors [
13,
31] and to activate signaling by β1 integrins [
32,
33]. These functions of NG2 make it an important player in pericyte biology, as evidenced by our findings that genetic ablation of the proteoglycan leads to vascularization deficits in both tumor [
16,
34,
35] and non-tumor [
15] models. However, the detailed effects of NG2 ablation on vascular structure and function have been examined only in the case of intracranial melanoma allografts [
16]. Thus, the importance of vascular NG2 in other types of tumors outside the brain, especially spontaneous tumors, remains unexplored. In this report we examine effects of NG2 ablation on the cellular and functional properties of tumor vessels in mammary tumors.
In three different experimental paradigms (orthotopic tumor cell line allografts, tumor transplantation, and
de novo tumor development), the appearance of detectable MMTV-PyMT mammary tumors is significantly delayed in the NG2 null mouse. The normal pattern of mammary gland development seen in the NG2 null mouse suggests that impaired mammary tumorigenesis is not the result of NG2-dependent deficits in normal mammary morphogenesis. Instead, the role of NG2 manifests itself during the process of tumorigenesis. In the case of melanomas and gliomas, NG2 is highly expressed by components of the tumor stroma [
13,
36,
37] but can also contribute to tumor progression as a component of the tumor cells themselves [
32,
38]. In human breast cancer, NG2 is also reported to promote tumor progression via its expression on so-called triple negative tumor cells [
18,
19]. However, in the case of MMTV-PyMT mammary tumors, our immunocytochemical studies establish that NG2 is not expressed by either the mammary epithelium or by neoplastic mammary tumor cells derived from this normal tissue. Instead, the proteoglycan is found on adipocytes in the mammary fat pad, myeloid cells that invade tumors from the circulation, and on perivascular cells associated with tumor microvessels. We see this same pattern of NG2 expression in samples of non-triple negative human ductal adenocarcinoma. The tumor-promoting properties of stromal NG2 further attest to the powerful influence of stromal elements on mammary tumor progression. These stromal effects of NG2 will be important for both mammary and other types of tumors, regardless of whether the tumor cells themselves are NG2-positive or NG2-negative.
Significantly, increased tumor latency is the most apparent effect of NG2 ablation in all three paradigms that we tested (orthotopic allografts, transplants, and
de novo tumors). While tumor onset is delayed in the NG2 null mouse in each of these models, once tumor growth begins, it occurs at roughly the same rate observed in wild type mice. These observations seem compatible with the evidence we present concerning the importance of NG2 for effective early tumor vascularization, and also with the idea that early establishment of a functional vascular supply is a critical event in the successful progression of a tumor. Since robust tumor vascularization is required for the angiogenic switch from hyperplasia to neoplasia [
7,
8,
39,
40], for continued tumor growth, and for eventual dispersal of tumor cells to distant sites, factors that reduce vascularization frequently are inhibitory to tumor progression. The retarded early progression of mammary tumor growth we have observed in the NG2 null mouse is consistent with the impaired progression in this mouse of other vascularization-based pathologies [
15], including brain tumor progression [
16].
The fact that NG2, an early marker of activated pericytes, plays an important functional role in vascularization is a testament to the precocious role of pericytes in the neovascularization process. As opposed to the traditional view of pericytes as late participants in the vascularization process, more recent studies have demonstrated the early presence of these cells in nascent microvessels [
9,
10,
12,
25,
41‐
43], especially in the context of tumor vascularization [
11,
16]. Several pieces of our current evidence point to differences in early vascularization of mammary tumors in wild type and NG2 null mice. As observed in pathological ocular neovascularization [
15] and brain tumor vascularization [
16], NG2 null pericytes exhibit reduced ensheathment of endothelial cells in mammary tumors. Although a reduction in pericyte number is not apparent in tumors growing in NG2 null mice [
16], reduced pericyte interaction with endothelial cells nevertheless compromises pericyte contribution to vessel development. This suggests that a key role of pericyte NG2 is the mediation of pericyte/endothelial cell communication via stimulation of β1 integrin signaling in endothelial cells [
17]. This communication deficit is also reflected by reduced assembly of the vascular basement membrane in mammary tumors grown in the NG2 null mouse. Deposition of the vascular basal lamina is a key result of pericyte/endothelial cell collaboration, since this structure is critical for vessel maturation, maintenance, and function [
44,
45].
The reductions in both pericyte coverage of endothelial cells and basal lamina deposition in mammary tumor vessels in the NG2 null mouse are accompanied by several other deficits in vessel structure and function. Vessels in NG2 null tumors are smaller in diameter than those in wild type tumors. The 20% reduction in vessel diameter seen in both spontaneous and engrafted tumors might restrict blood flow to tumor tissue in the NG2 null mouse. Pericyte maturation is retarded in tumor vessels in the NG2 null mouse and endothelial cell investment by mature pericytes is impaired to a greater extent than investment by immature pericytes. Endothelial cell sprouting is also reduced in tumor vessels in the NG2 null mouse, consistent with numerous studies demonstrating the importance of extracellular matrix attachment for activation of key signaling pathways in endothelial cells [
46‐
48]. As a result of these multiple defects, early tumor vessels in the NG2 null mouse exhibit greater than three-fold increased leakiness compared to early vessels in wild type tumors. The overall consequence of these deficits is an almost three-fold increase in mammary tumor hypoxia in the NG2 null mouse. All of these vascular deficiencies are observed during the same early time period in which tumor establishment and growth are negatively impacted in the NG2 null mouse.
Elevated hypoxia in early-stage NG2 null tumors is accompanied by expression of increased levels of VEGF beyond what is seen in wild type tumors. Levels of vessel-associated VEGF are similar in tumors in both genotypes, so that increased amounts of non-vascular VEGF account for the difference in VEGF levels seen in wild type and NG2 null tumors. This VEGF localization pattern may account for our observation that increased VEGF levels in NG2 null tumors are not accompanied by changes in vascular density or morphology. Our previous study with tumors in collagen VI null mice suggested that tumor vessel remodeling is more readily induced by vessel-associated VEGF [
49]. Extracellular matrix-bound VEGF species have been shown to have angiogenic/morphogenetic properties that differ from those of non-vascular VEGF species [
47]. Increased levels of non-vascular VEGF are apparently not sufficient to induce vascular growth and remodeling, at least at this early stage of tumor development. It remains to be determined whether increased VEGF levels in NG2 null tumors can have effects on vascular density and/or morphology at later points in tumor development. It will also be important to investigate the long-term effects of NG2 ablation on tumor vascularization and hypoxia, with specific attention to tumor growth, invasion, and metastasis. An important feature of the MMTV-PyMT mouse mammary tumor model, metastasis is a critical factor in determining the survival of human breast cancer patients.
Reduced mammary tumor progression in the MMTV-PyMT mouse due to pericyte-dependent deficits in vascularization has a parallel in endothelial cell-dependent deficits in the tumor vasculature. In the context of the MMTV-PyMT model, ablation of T-cadherin, which is normally expressed by vascular endothelial cells, causes deficiencies in mammary tumor vascularization that lead to diminished tumor progression [
50]. In addition to highlighting the importance of pericyte/endothelial cell crosstalk, these combined findings once again emphasize the critical dependence of mammary tumor progression on vascularization. The interplay between pericytes and endothelial cells validates recent attempts at dual targeting of these two cell types as a means of improving the efficacy of anti-angiogenic therapy [
51,
52]. A dual targeting strategy might offer the means of improving therapy in cases of breast cancer that are resistant to other types of treatment.
In spite of our evidence that ablation of pericyte NG2 is an important factor in the reduced mammary tumor progression seen in NG2 null mice, we must still confront the possibility that ablation of NG2 in myeloid cells and adipocytes may also contribute to the observed effects. The frequent perivascular localization of myeloid cells [
12,
27], along with the emerging importance of these cells in several aspects of tumor progression, including inflammation, vascularization [
26,
28,
40,
53], and metastasis [
54‐
56], demands that we consider the effects of NG2 ablation on myeloid cell function. Our own work has demonstrated the participation of NG2-positive myeloid cells in the earliest stages of fibroblast growth factor 2 (FGF2)-induced neovascularization [
12]. Our initial flow cytometry evidence indicates that NG2 ablation reduces the number of both TAMs and TEMs present in mammary tumors. Both of these macrophage sub-populations are thought to have tumor promoting properties [
26‐
28,
53‐
56], consistent with our observation of delayed mammary tumor growth in NG2 null MMTV-PyMT mice. Along these same lines, we have previously seen that ablation of NG2 in a model of spinal cord demyelination diminishes macrophage recruitment to demyelinated lesions and shifts macrophages from a pro-inflammatory to anti-inflammatory phenotype [
57]. These apparent effects of NG2 on macrophage recruitment and/or maturation emphasize the need for additional work to determine the role of the proteoglycan in macrophage contributions to tumor vascularization, growth and metastasis. We cannot conclude from our current results whether changes in macrophage populations are merely correlated with changes in tumor growth or whether they are causally involved in altering tumor growth.
Adipocytes also have the potential to play a key stromal role in mammary tumorigenesis. New discoveries are defining the role of adipocytes in controlling metabolism, as well as in producing adipokines that promote mammary tumor progression [
58,
59]. It is of considerable interest that ablation of the NG2 ligand collagen VI, which is a product of adipocytes, leads to impaired mammary tumor progression in the MMTV-PyMT mouse [
60]. Since NG2 and collagen VI are both produced by adipocytes, and since NG2 serves as a cell surface receptor for collagen VI [
61,
62], NG2 on the adipocyte surface might be important for collagen VI anchorage and localization, and perhaps for its effects on the behavior of mammary tumor cells.
In order to resolve these questions regarding the various stromal roles of NG2 in breast cancer, we are developing Cre-lox capabilities for cell type-specific ablation of the proteoglycan in the context of the MMTV-PyMT model. This will allow us to define the importance of NG2 in mediating the respective effects of pericytes, myeloid cells, and adipocytes on mammary tumor progression. Understanding the respective roles of NG2 in these different stromal populations will be important for any attempts to use the proteoglycan as a target for therapeutic purposes. For cases of human breast cancer in which the tumor cells also express NG2, targeting of the proteoglycan may be an even more powerful approach due to inhibitory effects on both tumor and stromal compartments.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KG and WKY contributed equally to this work, participating in study design, data acquisition and analysis, and preparation of the manuscript. KK, HH, LJTY, and YC performed data acquisition and analysis. LGE contributed key resources and participated in manuscript preparation. RDC and WBS contributed to experimental design and to preparation of the manuscript. All authors read and approved the final manuscript.