Background
Notch signaling underlies an evolutionarily conserved mechanism that regulates a wide range of cellular processes via direct cell-cell communication. In mammals, there are five Notch ligands (Jagged1, Jagged2, Delta-like 1/Dll1, Dll3, and Dll4) and four Notch receptors (Notch1-4). The interaction between the Notch ligand and its receptor triggers two proteolytic cleavages that release the Notch intracellular domain into the nucleus. There, the Notch intracellular domain forms complexes with members of the CSL transcription factor family (CBF1/RBP-Jκ/Suppressor of Hairless/LAG-1), leading to the expression of downstream transcription factors (reviewed in [
1]).
Among the four Notch receptors, Notch1 and Notch4 are expressed in the vascular endothelium, with Notch4 expression being more restricted to endothelial cells (ECs) [
2]. Mice homozygous for a null allele of
Notch1 develop vascular abnormalities and die
in utero shortly after primitive vascular plexus formation [
3,
4]. Homozygous
Notch4 null mice develop normally and are viable and fertile. However, the combination of homozygous loss of
Notch1 and
Notch4 results in a more pronounced vascular phenotype than the
Notch1 null homozygous alone [
3,
4]. Even though this finding suggests that lack of Notch4 can exacerbate the effect of Notch1 deficiency, the unique role for Notch4 in vascular development is unknown. In adult mice, genetic deletion of
Notch4 does not lead to any detectable abnormalities, other than a slightly elevated systolic blood pressure after experimental induction of hindlimb ischemia [
3,
5].
Angiogenesis, the formation of new vessels from pre-existing ones, plays a key role in cancer and other pathological conditions. Solid tumors depend on the development of blood vessels to provide oxygen and nutrients to support their growth [
6]. Interfering with Notch1 signaling dramatically affects angiogenesis in the developing and pathological vasculatures (reviewed in [
7,
8]). In solid tumors, blocking Dll4 leads to excessive sprouting and “non-productive” angiogenesis, which results in decreased vessel perfusion and inhibition of tumor growth [
9,
10]. Additionally, selective blocking of Notch1 with antibodies inhibits tumor growth by a mechanism consistent with that of Dll4 inhibition, namely by promoting abnormal growth of vessel sprouts that do not perfuse tissues [
11]. Thus while the importance of the Dll4-Notch1 signaling axis in tumor angiogenesis is well documented, whether Notch4, primarily expressed in the endothelium, is required for angiogenesis, vascular perfusion, and tumor progression is unknown.
To unravel the role of Notch4 in tumor angiogenesis, we chose breast cancer as a solid tumor model. Since Notch4 is predominantly expressed in the vascular endothelium in mice [
2], we therefore used orthotopic transplantation of mouse mammary carcinoma cells wild type for
Notch4 into syngeneic
Notch4-deficient hosts to test the importance of Notch4 in tumor angiogenesis.
Methods
Animals and human samples
C57BL/6J wild type mice were purchased from Jackson Laboratory. MMTV-PyMT transgenic mice in C57 background were provided by Dr. Zena Werb’s lab at the University of California, San Francisco (UCSF). All strains were bred and maintained in a C57BL/6J background. We crossed wild type mice to Notch4−/− mice to obtain Notch4+/− mice, which were then intercrossed to generate Notch4−/− and Notch4+/+ controls. Mice were kept in a pathogen-barrier facility from the time of breeding to euthanasia.
Human breast cancer samples (paraffin-embedded tissue blocks) were obtained from the UCSF Breast Cancer SPORE tissue core, which collected samples in compliance with a protocol approved by the UCSF Committee on Human Research. Patient consent was obtained using the standard surgical consent form, in combination with the brochure, Donating Tissue for Medical Research. The research involved collection of left-over specimens and, as such, involved no more than minimal risk to the subjects. Our use of these tissues was also approved by the UCSF Committee on Human Research.
Immunostaining
Paraffin tissue sections of human samples were processed using standard methods. Mouse tissue samples were fixed in 4% formaldehyde in phosphate-buffered saline (PBS), embedded in OCT (Tissue Tek), frozen, and cryosectioned. Immunostaining was performed using rabbit anti-Notch4 (Upstate, diluted 1/1000) and rat anti-mouse CD31 (MEC 13.3, Pharmingen, diluted 1/500) as previously described [
12,
13]. Secondary antibodies were Cy3-conjugated donkey anti-rabbit (diluted 1/1000) and alexa-488 donkey anti-rat (diluted 1/1000), both from Jackson ImmunoResearch. In experiments with mouse tissue tumors, sections were collected from three different regions of each tumor and quantification was performed from three different fields per section. When tumors were not detected by naked eye, the whole mammary fat pad was preserved for sectioning and microscopical analysis.
Animal procedures
Animal experiments were performed in compliance with the guidelines of UCSF Institutional Animal Care and Use Committee (IACUC), who approved this study under protocol #AN085404 and AN075264. We routinely monitored our mice targeted for tumorigenesis before and during tumor formation. Mice that displayed any of the following signs were euthanized: loss of 15% of their initial body weight, tumor size of 2 cm in the largest dimension, or decline in activity.
Isolation of primary tumor cells and orthotopic transplantation
Primary tumor epithelial cells were isolated from mammary carcinomas that developed in
MMTV-PyMT mice as previously described [
14]. Briefly, tumors were dissected from 4–5 month old females, minced with razor blades and collagenase-digested (Sigma Blend L 0.5 mg/ml in RPMI 1640, supplemented with 10 mM Hepes and 5% fetal bovine serum) for 1 hour at 37°C with continuous agitation. The resultant mammary epithelial “organoids” were washed in Hank’s medium with calcium and magnesium (with 5% fetal bovine serum), separated from single cells and blood through differential centrifugation, trypsinized for 20 minutes and treated with DNaseI to obtain single cell suspensions. These were immediately frozen at 5 × 10
6 cells/vial, and preserved in liquid nitrogen for later use. Each cell aliquot was then thawed and used for transplants in paired sets of
Notch4−/− and control animals. Cells were washed and resuspended in ice-cold PBS. Mice were anesthetized using isoflurane and a ventral incision was made to expose the fourth inguinal mammary glands. Approximately 10
6 cells (in a 10 μl volume) were injected into syngeneic C57BL/6J wild type or
Notch4−/− female mice at 3 weeks of age. Tumor development was monitored every other day by palpation.
Vessel perfusion studies
Mice were anesthetized with isoflurane and injected in the tail vein with a combination of 60 μg of biotinylated Lycopersicon esculentum lectin (Vector Laboratories, Burlingame, CA) and 60 μg of Cy3-streptavidin (Jackson ImmunoResearch). Mice were further anesthetized with ketamine/xylazine. After 5 minutes, the chest was opened and the vasculature was perfused with PBS through the left ventricle for 5 minutes, followed by 4% formaldehyde in PBS for 3 minutes at a pressure of 100 mm Hg.
Real-time PCR
Total RNA was isolated from normal mammary inguinal fat pads and from transplanted tumors using Trizol (Invitrogen), following the manufacturer’s instructions. RNA was further purified from contaminant genomic DNA with DNase I treatment and RNeasy columns (Qiagen), following the manufacturer’s instructions. RNA was retro-transcribed using SuperscriptIII (Invitrogen) and SYBR Green-based real-time-PCR was used to analyze gene expression. Normalized fold change in gene expression in tumor relative to normal gland was calculated according to Pffafl [
15], using
Tie2,
VE-cadherin, and C
D31 as reference genes. The following primer sequences were used:
Notch4: 5′-ctctgcagccctggctatac-3′, 5′-ggcatcgagcagtgtgtg-3′;
Tie2: 5′-atgcccttctccaccctctcc-3′, 5′-ccactacctactagtgaagaa-3′;
Cd31: 5′-ctcctcggcgatcttgctgaa-3′, 5′-gtcatggccatggtcgagta-3′;
VE-cadherin: 5′-gtaagtgaccaactgctcgtgaat-3′, 5′-tcctctgcatcctcactatcaca-3′
Immunoprecipitation and western blotting
Immunoprecipitation of Notch4 from lysates of normal mammary gland at pubertal stage followed by Western blotting analysis was performed as previously described [
16] using a polyclonal rabbit anti-Notch4 antibody (Upstate).
Statistical analysis
Vessel density and perfusion analysis was performed as previously described [
13]. Data are expressed as mean + standard error of the mean (s.e.m). P-values were calculated using a two-tailed t test except for tumor onset analysis where Fisher’s exact test was applied. Values of p≤0.05 were considered statistically significant.
Discussion
In this study we demonstrate that Notch4 expression is upregulated in the vasculature of mammary tumors. Furthermore, our results suggest that host Notch4 plays a role in the early emergence of MMTV-PyMT breast tumors following transplantation. We also found that host Notch4 is required for initial tumor vascular perfusion, but not vessel sprouting. To our knowledge, this is the first report demonstrating the functional relevance of host Notch4 upregulation in tumorigenesis and tumor vessel perfusion. Our findings shed light on the role of Notch4, independent of other Notch receptors, in tumor-host interactions.
Our results show increased levels of Notch4 in the blood vessels of mouse and human breast tumor tissues. Expression of Notch receptors and ligands has been described in a wide variety of different tumor types, such as colorectal, prostate, liver, pancreatic and breast cancer [
23,
24] and a role for Notch4 in regulating breast cancer stem cell activity has been proposed [
25]. Mittal
et al. reported that the levels of Notch receptors (1, 2, and 4) and ligands (Jagged 1, 2, Dll1, and 4) are increased in human breast cancer compared to normal breast tissue [
26]. Recently, a study by Speiser and collaborators showed increased Notch1 and Notch4 levels in tumor epithelial cells and vascular endothelial cells in triple-negative breast cancer samples [
27], therefore highlighting the relevance of Notch4 expression in the vasculature. However, in our studies, we did not observe Notch4 staining in either human (Figure
1c) or murine tumor cells (Figure
1a, ix, xii and
1b, iii), but rather observed Notch4 upregulation in the tumor vasculature. We used a well-characterized Notch4-specific antibody [
28] and verified its specificity using Notch4 knockout tissues (Additional file
1: Figure S1). Our study demonstrates Notch4 upregulation in the vasculature of both mouse models of mammary adenocarcinoma and human breast cancer. It is possible that different types and grades of breast carcinoma present different Notch4 expression levels and distribution. However, in our study, our mouse model provides evidence that host, likely vascular, Notch4 plays a role in breast cancer development. Consistent with our finding that Notch4 is expressed in the tumor vasculature, the Notch ligand Dll4 is detected in the vessels of infiltrating human breast adenocarcinoma samples [
29], making it a possible ligand for Notch4 in tumor vasculature.
Orthotopic transplantation of mammary tumor cells is a well-established model for
in vivo studies of breast tumorigenesis, and we chose this approach because it allowed us to study host Notch4-mediated effects on tumorigenesis independent of Notch4 activity in the tumor cell compartment. We demonstrate a contribution of the tumor microenvironment, namely the host Notch4, to tumorigenesis. We also pinpoint that the tumorigenic defect lies at the tumor onset following transplantation
. It is well documented that the growth of solid tumors depends on the development of new vasculature [
30]. Inhibition of members of the Notch signaling pathway leads to an increase in “non-productive” angiogenesis, characterized by a reduction in vessel perfusion despite an increase in vessel sprouting [
9‐
11,
31]. We thus examined vascular perfusion in tumors grown in
Notch4
−/−
hosts. Since both the mammary gland and its vasculature appear normal in
Notch4
−/−
mice, we reasoned that any vascular defects in tumors from
Notch4
−/−
hosts must be due to abnormalities that occur during tumorigenesis, and not as a result of preexisting vascular defects. We observed that vessel perfusion was reduced in tumors grown in
Notch4
−/−
vs. wild type hosts. This observation is consistent with many reports of Notch pathway inhibition leading to reduced perfusion [
9‐
11,
31].
Although poor tumor vessel perfusion correlated significantly with delayed tumor onset at early time points after transplantation, our results alone do not provide causal proof that reduced vessel perfusion leads to delayed tumor onset in
Notch4−/− mice. Given the complex dynamics in tumor-host interaction and tumor microenvironment, we cannot rule out the possibility that vascular Notch4 (and other Notch pathway proteins) may regulate tumor onset by mechanisms that are independent of vessel perfusion. It is also possible that differences in transplantation-associated immune responses contribute to the delayed tumor onset in
Notch4
−/−
hosts. Both tumor-associated fibroblasts and tumor-infiltrating leukocytes have been shown to play an important role in tumor onset and growth [
32]. Notch4 expression has been detected in immune cells of myeloid lineage [
21]. Although we detected Notch4 overexpression predominantly in the tumor vasculature, we cannot rule out the possibility that host myeloid cells may contribute to the difference in tumor onset between the two host genotypes.
Host Notch4 deficiency delays tumor onset and decreases initial perfusion, however, the growth of established tumors can ultimately progress in the absence of host Notch4. This result suggests that host Notch4 plays a unique role in the initiation of tumor onset after transplantation.
Surprisingly, our results indicate that Notch4 is dispensable for vessel sprouting in the tumor. Sprouting angiogenesis is a hallmark of tumor neovascularization, and Dll4/Notch1 signaling functions to inhibit vessel sprouting [
8]. Moreover, given that Notch1 appears to be the primary Notch receptor responsible for developmental angiogenesis [
3], together with the results obtained using specific anti-Notch1 antibodies in tumors [
11], Notch1 seems to be the predominant mediator of Notch signaling in tumor angiogenesis. It is therefore likely that the increased vascular network observed when inhibiting pan-Notch signaling or Notch ligands is mainly due to the inhibition of Notch1. Alternatively, it is possible that a subtle defect in vessel sprouting exists in the
Notch4−/− tumor vasculature, but current methodologies are not sensitive enough to detect such subtle phenotype.
Competing interests
The authors declare to have no competing interests.
Authors’ contributions
RAW, MJC, XW conceived and designed the experiments; MJC, XW, SKB and EC performed the experiments; MJC, XW, HC, RS, RAW, OM analyzed the data; TG contributed reagents/materials/analysis tools for this manuscript; HC, RAW, CAC, MJC, RS, OM wrote the manuscript. MJC and XW have contributed equally to this manuscript. All authors read and approved the final manuscript.