Background
Breast cancer is the most common non-cutaneous cancer form in the Western world with more than 250,000 annual new cases in the USA alone. Around one in eight women develop invasive breast cancer over the course of life.
Increased proliferative rate of breast cancer cells necessitates high metabolic activity, which again requires a constant supply of metabolic substrates and O
2. High O
2 utilization rates and insufficient blood supply lead to inadequate tissue oxygenation at least in micro-areas of human breast cancer tissue [
1]. As evidenced by association between tumor perfusion and breast cancer O
2 consumption [
1], O
2 availability limits oxidative metabolism in cancer tissue, which supports the anti-cancer potential of interventions that manipulate O
2 and nutrient delivery to tumors.
The clinical impact of anti-angiogenic drugs used to target generation of breast cancer arteries has been disappointing [
2]. Modifying tumor perfusion by targeting functions of already formed tumor arteries provides an attractive alternative: selective vasoconstriction of cancer arteries will lower tumor perfusion and starve cancer cells; and if sufficiently pronounced, blood flow arrest can cause tumor infarction as observed in response to thrombus induction in tumor blood vessels [
3]. Selective vasodilation of cancer arteries during treatment periods also has therapeutic potential by facilitating delivery of chemotherapeutic agents and enhancing radiotherapeutic responses through elevated local O
2 pressures [
4]. Delivery of drugs to cancer tissue is challenged by irregular flow patterns [
5]: large variations in blood flow between tumor regions result in inhomogeneous drug delivery; and even in individual vessel segments, blood flow is temporally variable with blood flow periodically stopping or reversing. Controlling precapillary resistance is also important for fluid exchange, which again plays a role in extravasation and shunting of drugs.
The heterogeneous and typically insufficient blood supply to cancer tissue creates local hypoxia and accumulation of acidic waste products. The extracellular microenvironment of breast cancer tissue is fundamentally different from normal tissue and contains, for instance, high concentrations of H
+, ATP, and paracrine signaling factors (e.g., endothelin and vascular endothelial growth factor) [
6]. These local biochemical disturbances are hostile for most normal cells including those of the immune system [
7,
8]. When the metabolic stress of the tumor microenvironment remains within the limits of cancer cell survival, it provides a survival advantage for cancer cells compared to normal cells and constitutes a selection pressure that favors cancer cells with more malignant phenotypes [
9]. Targeted vasodilation of cancer arteries may relieve intermittent blood flow patterns and provide more uniform blood distributions that can improve the typical mismatch between blood flow and metabolism, minimize disturbances in local environment, inhibit upregulation of hypoxia-inducible genes, and boost the ability of the immune system to recognize and fight cancer cells.
The architecture of the cancer vasculature is distorted with leaky and disorganized blood vessels sharing characteristics of arterioles, capillaries, and venules [
10]. Variable density of contractile elements challenges interventions to substantially modify tumor perfusion. Rather than studying dysmorphic intratumoral blood vessels, we here focused on feed arteries that supply cancer tissue with blood, yet in their overall wall composition resemble normal arteries and thus are expected to maintain ability for considerable tension development.
In the current study, we explored the structure and function of breast cancer feed arteries from mice with ErbB2-induced breast cancer. ErbB2 is an orphan growth factor receptor that facilitates breast carcinogenesis: ~ 20% of human breast cancers show overexpression or gene amplification of ErbB2, and targeted treatment—e.g., with the functional monoclonal antibody trastuzumab—improves the prognosis [
11]. We tested the hypothesis that breast cancer feed arteries are functionally and structurally distinct from equivalent control arteries and find that breast cancer arteries (a) are thin-walled with overall lower capacity for vasocontraction and (b) display deregulation of adrenergic and thromboxane-mediated signaling leading to reduced perivascular, nerve-mediated, contractile responses. We propose that attenuated constriction of breast cancer feed arteries enhances tumor perfusion and that preferential changes in tumor vascular resistance relative to other vascular beds are therapeutically attainable.
Methods
We isolated breast cancer feed arteries and matched control arteries from female virgin FVB mice (FVB/N-Tg(MMTVneu)202Mul/J; Jackson Laboratories, ME, USA) with breast epithelial overexpression of ErbB2—also known as neu and HER2—under transcriptional control of the mouse mammary tumor virus promoter [
12]. Mice were housed in the animal facility at the Department of Biomedicine, Aarhus University, under a 12-h light/12-h dark cycle with free access to food and water. ErbB2-overexpressing mice develop macroscopically identifiable breast cancer with median latency of 205 days [
12], and they were palpated weekly for tumor detection. Cancer feed arteries were adherent to breast carcinomas that developed along the mammary lines extending from the axilla to the inguen; control arteries were similarly located anatomically but distanced at least 3 cm from macroscopically identifiable tumor tissue. All animal handling was approved by the Danish Animal Experiments Inspectorate (2012-15-2935-00002) and performed according to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.
Small artery myography
Mice were euthanized by cervical dislocation or CO
2 inhalation, and breast cancer feed arteries and matched control arteries isolated under a stereomicroscope. Arteries were mounted on 40-μm stainless steel wires for isometric myography in multi-channel myograph chambers (610 M; DMT, Denmark) filled with Ca
2+-free physiological saline solution (PSS, see composition below) in order to induce full relaxation and avoid stretch-induced damage. Myograph chambers were then washed with standard PSS and heated to 37 °C. Breast cancer feed arteries and matched control arteries had very similar internal diameters (171 ± 5 μm
vs. 178 ± 6 μm;
n = 56;
P = 0.32; paired two-tailed Student’s
t test) when normalized to 90% of the internal diameter corresponding to a transmural pressure of 100 mmHg [
13]. Cancer and control arteries from a given mouse were both excluded from further analyses if either artery developed less than 1 mN force in response to application of 63.9 mmol/L extracellular (o) K
+. Data were collected in LabChart Pro (ADInstruments, New Zealand).
As per previous experience with murine arteries [
14], vasocontractile responses to endothelin-1, norepinephrine, thromboxane analog U46619, and depolarization with elevated [K
+]
o were tested through cumulative applications. Endothelium-dependent vasorelaxation to acetylcholine was evaluated as single doses in order to avoid spontaneous relaxations or tachyphylaxis in arteries pre-contracted with norepinephrine. Experiments involving elevated [K
+]
o were performed in the presence of 1 μmol/L of the non-selective α-adrenoceptor antagonist phentolamine in order to avoid the effects of norepinephrine released from perivascular nerve endings in response to depolarization.
The PSS contained (in mmol/L) 140 Na+, 4 K+, 1.6 Ca2+, 1.2 Mg2+, 124 Cl−, 22 HCO3−, 1.2 SO42−, 1.2 H2PO4−, 10 HEPES, 5.5 glucose, and 0.03 ethylenediaminetetraacetic acid (EDTA); pH was adjusted to 7.4 when gassed with 5% CO2/balance air at 37 °C. In experiments mimicking metabolic acidosis, pH of the bath solution was reduced to 6.8 by lowering HCO3− to 5.5 mmol/L through substitution with Cl−. In Ca2+-free PSS, the 1.6 mmol/L Ca2+ and associated 3.2 mmol/L Cl− were omitted.
Measurements of intracellular [Ca2+]
We evaluated intracellular [Ca
2+] in vascular smooth muscle cells (VSMCs) using fluorescence microscopy based on the Ca
2+-sensitive fluorophore Fura-2 (Life Technologies, CA, USA) [
15]. Arteries mounted in single-channel wire myographs (320A; DMT) were loaded with 5 μmol/L of the acetoxymethyl ester form of Fura-2 for 2 × 30 min at 37 °C and investigated using a Leica DM IRB inverted microscope (Germany) with a ×20 objective (numerical aperture 0.5) connected to a PTI DeltaScan fluorescence system (NJ, USA). Arteries were excited alternately at 340 and 380 nm and emission light collected around 530 nm. Background fluorescence prior to loading was subtracted from recorded emissions before ratio (340/380) calculation. Relative changes in Fura-2 fluorescence were calculated for each artery and normalized to the average response in control arteries.
Membrane potential measurements
Arteries were mounted in a customized, single-channel, water-heated wire myograph (DMT). Membrane potential (V
m) was measured using aluminum silicate microelectrodes (World Precision Instruments, UK; resistance ~40–120 MΩ when backfilled with 3 mol/L KCl) coupled to an Intra-767 amplifier (World Precision Instruments), visualized on an oscilloscope (Gould-Nicolet Technologies, UK), and continuously stored with a PowerLab system (ADInstruments) [
16]. Electrode entries into cells resulted in abrupt drops in voltage followed by sharp returns to baseline upon retraction [
17].
Artery structure and stereological analyses
Arterial wall structure was assessed based on transmission light microscopy of arteries mounted in dual-channel wire myographs (420A; DMT) with glass windows in the chamber base [
18]. Media thickness was measured under gentle stretch using a filar micrometer at six individual points for each artery [
19]. To determine the relationship between arterial wall structure and tumor size, we calculated the volumes of excised tumors based on the formula:
$$ \mathrm{Volume}=\mathrm{Length}\times \mathrm{Width}\times \mathrm{Depth}\times \uppi /6 $$
After normalization, arteries were fixed for 20 min in 4% paraformaldehyde and stored in phosphate-buffered saline at 5 °C. Paraffin-embedded arteries cut longitudinally—i.e., perpendicular to the long axis of the VSMCs—into 3-μm sections were stained with Giemsa or hematoxylin and eosin. Through identification of nuclear ends—i.e., nuclei present in one section but absent in the neighboring section—the numerical cellular density and VSMC dimensions were calculated for the media volume demarcated by two consecutive sections based on the unbiased stereological disector method [
18,
20,
21]. We calculated media stress—as wall tension divided by media thickness—in order to evaluate if differences in contractile function are explained by structural variation between the arteries.
Electrical field stimulation
Perivascular nerve endings of arteries mounted in dual-channel wire myographs (420A; DMT) were stimulated using bipolar electrical fields created between 200-pm thick platinum electrodes (DMT). Single trains of 10-s duration consisting of 0.1-ms square-wave pulses of 35 mA with frequencies from 0.5 to 64 Hz were delivered by a CS200 current stimulator (DMT). Dependency on nerve conduction was confirmed with 1 μmol/L of the specific voltage-gated Na
+-channel inhibitor tetrodotoxin (TTX) [
22]. Prazosin was applied in order to test involvement of α
1-adrenoceptors [
23,
24].
Labelling of α1-adrenergic receptors
We evaluated expression of α
1-adrenergic receptors using the boron dipyrromethene (BODIPY™)-labelled α
1-adrenoceptor antagonist prazosin [
25]. Freshly dissected arteries incubated with 1 μmol/L BODIPY™ FL prazosin (B7433; ThermoFisher Scientific, MA, USA) for 90 min at room temperature in the dark were investigated by confocal microscopy: arteries were excited at 488 nm and emission light collected at wavelengths between 505 and 530 nm. Z-stack image series of 1-μm optical steps were constructed and the second image within the media used for quantification. Strong autofluorescence of the internal elastic lamina defines the transition between tunica media and tunica intima and ensures that the confocal planes of quantified images are at comparable levels in the media. Pixels with fluorescence intensities above a pre-selected threshold (2000 arbitrary units) were considered positive. Labeled areas were expressed relative to the total area of the optical sections covered by tunica media. Essentially no fluorescence was observed under the experimental conditions when no BODIPY™-labelled prazosin was added (data not shown).
Reverse transcription and quantitative PCR analyses
Expression of messenger RNA for α
1-adrenergic receptors was evaluated by reverse transcription and quantitative PCR analyses. Commercially available primers and probes (TaqMan® Gene Expression Assays; Applied Biosystems, CA, USA) against α
1A- (Mm00442668_m1), α
1B- (Mm00431685_m1), and α
1D-adrenoceptors (Mm01328600_m1) were used with the transferrin receptor (Mm01344478_m1) serving as reference. Isolated arteries were disrupted in lysis buffer using Qiagen TissueLyser (Denmark). Total RNA was isolated with the RNeasy Micro QiaCube kit and—after DNase treatment—reverse transcribed using Reverse Transcriptase III (Invitrogen, CA, USA), RNase inhibitor Superase (Invitrogen), and random decamer primers (Eurofins Genomics, Germany). To control for genomic amplification, RT– experiments were performed without reverse transcriptase added. PCRs performed with a Stratagene MX3000P qPCR system (AH Diagnostics, Denmark) consisted of 1 cycle at 95 °C for 4 min followed by 50 cycles at 92 °C for 10 s, 55 °C for 20 s, and 72 °C for 30 s. Expression levels for α
1-adrenoceptors in cancer arteries relative to control arteries were calculated based on the 2
–ΔΔCt method [
26].
Statistics
Data are expressed as mean ± SEM; n indicates the number of mice. P < 0.05 was considered statistically significant. We used the paired two-tailed Student’s t test to compare one variable in cancer and matched control arteries. Arteries from different mice were compared using the unpaired two-tailed Student’s t test. The one-sample t test was used to compare single distributions to hypothetical means. We evaluated the effects of two variables on the measured variable, measured multiple times for each mouse, using repeated measures two-way analysis of variance (ANOVA) followed by the Sidak or Bonferroni post-hoc test. Dependency of media structure on tumor volume was tested by least-squares linear regression analyses. Concentration-response relationships were fitted to sigmoidal functions and compared with extra-sum-of-squares F tests. Right-skewed data were log-transformed to achieve normal distribution. Statistical analyses were performed using GraphPad Prism 7.03 software.
Discussion
We show here that murine breast cancer feed arteries are functionally and structurally abnormal: compared to similarly-sized control arteries from equivalent anatomical locations along the mammary lines, vasocontractile responses of breast cancer feed arteries are reduced (Fig.
1 and Additional file
1: Figure S1) because of (a) structural thinning of tunica media due to fewer VSMC layers (Fig.
2) and (b) reduced activation during norepinephrine stimulation (Fig.
3) explained by diminished α
1A-adrenoceptor expression (Fig.
5), leading to smaller VSMC depolarizations (Fig.
4c) and decreased VSMC intracellular Ca
2+ responses (Fig.
4a). In particular, our data document striking attenuation of sympathetic vasocontraction in response to endogenous norepinephrine released from perivascular nerve endings (Fig.
6).
We observed overall reduction in α
1-adrenoceptor expression in VSMCs of breast cancer feed arteries—evidenced by attenuated binding of the α
1-adrenoceptor antagonist prazosin in the tunica media (Fig.
5b-
d)—and confirmed lower levels of messenger RNA transcripts coding for α
1A-adrenoceptors (Fig.
5e). Noteworthy, media stress in response to depolarization induced with elevated [K
+]
o was unaffected in breast cancer feed arteries compared to control arteries (Fig.
3b). Endothelin-1-induced VSMC depolarization, VSMC intracellular Ca
2+ dynamics, and media stress also did not differ between breast cancer feed arteries and control arteries (Fig.
3a and
4b,
c). These findings support the concept that apart from differences in receptor expression, individual VSMCs of breast cancer feed arteries are functionally unaffected. Our structural evaluation of the arterial media assumes that it consists of uniformly sized VSMCs without infiltration from non-contractile—e.g., fibroblastic—cells, and this assumption is supported by the equivalent media stress in cancer feed arteries and control arteries depolarized with elevated extracellular [K
+] (Fig.
3b).
Reduced norepinephrine-induced vasoconstriction of breast cancer feed arteries—which have similar relaxed diameters as control arteries—leads to lower vascular resistance and predictably increases perfusion of tumor tissue. Because of lower norepinephrine responses, we expect that tension development in breast cancer feed arteries is more disconnected from systemic cardiovascular regulation than in control arteries, and we predict a strong influence from local metabolic and paracrine factors. Indeed, our findings support that vasomotor activity in response to locally accumulated vasodilator metabolites (e.g., H
+ (Additional file
1: Figure S1)) and paracrine signaling substances (e.g., endothelin-1 (Fig.
1c and Additional file
1: Figure S1B)) is preserved in breast cancer feed arteries. In accordance, studies show that endothelin-1 is important for myogenic tone development in tumor arterioles [
28].
Thinner tunica media of breast cancer feed arteries compared to control arteries is explained by fewer VSMCs per unit artery length resulting in fewer VSMC layers (Fig.
2j,
k). Sizes of individual VSMCs and the volume fraction of VSMCs in tunica media did not differ between breast cancer feed arteries and control arteries (Fig.
2f-
i). Possible explanations for altered arterial structure and function include increased mechanical forces due to reduced downstream resistance. Disturbed flow patterns regulate arterial structure with high shear stress generally leading to outward remodeling and low shear stress causing inward remodeling [
29,
30]. In particular, the blood flow patterns modulate vessel growth during vascular endothelial growth factor-induced angiogenesis [
31]. Altered biochemical composition of the extracellular tumor microenvironment may also influence the structure and function of breast cancer feed arteries. Disturbed local acid-base conditions of cancer tissue have potential to modify arterial remodeling [
32]: recent evidence supports that intracellular pH influences VSMC proliferation whereas local pH gradients along VSMC protrusions modify VSMC migration [
29]. Altered pH
o conditions have also been shown to regulate, for instance, cell-cell and cell-matrix adhesion [
33,
34]. Therefore, acid-base disturbances are powerful signals predicted to impact arterial structure development. The altered wall structure was present even in arteries from the smallest investigated tumors (Fig.
2d), suggesting that the change in arterial development or adaptation of arterial structure occurs early in cancer development. The specialization of the cancer feed arteries towards reduced resistance likely facilitates oxygen and substrate delivery during carcinogenesis and early tumor expansion, which emphasizes its potential as a therapeutic target.
Systemic application of vasoactive substances leads to heterogeneous changes in vascular resistance between different organ systems and often influences perfusion pressure. In addition to the effects of cardiac function, perfusion pressure is a function of total peripheral resistance. As a consequence, tumor perfusion depends on changes in the resistance of the tumor vasculature relative to other vascular beds. Considering the diminished vasocontractile response of breast cancer feed arteries, increased sympathetic tone is predicted to increase perfusion pressure, have marginal effects on tumor vascular resistance, and ultimately increase tumor perfusion. Reduced precapillary resistance due to attenuated sympathicus-mediated vasoconstriction will also increase capillary pressure and fluid filtration and likely contribute to edema formation in tumor tissue. Thus, our results indicate that variation in surface receptor expression between cancer and normal arteries can be utilized to change resistance of the tumor vasculature relative to the peripheral vasculature as a whole and thereby modify oxygen and nutrient delivery to cancer tissue. As suggested for bradykinin signaling in tumor arteries, a relative decrease in tumor vascular resistance can facilitate chemotherapy delivery [
35,
36].