Wang et al.

Journal of Experimental & Clinical Cancer Research (2019) 38:235

https://doi.org/10.1186/s13046-019-1211-2

RESEARCH Open Access

Metformin inhibits metastatic breast cancer

progression and improves chemosensitivity
by inducing vessel normalization via PDGF-B
downregulation
Ji-Chang Wang1,2†, Guang-Yue Li3†, Bo Wang2, Su-Xia Han4, Xin Sun5, Yi-Na Jiang6, Yan-Wei Shen8, Can Zhou7,
Jun Feng1, Shao-Ying Lu1, Jian-Lin Liu1, Mao-De Wang9* and Pei-Jun Liu2,10*

Abstract
Background: Vascular maturity and functionality are closely associated with tumor progression and
chemosensitivity. The antidiabetic agent metformin has shown its ability to inhibit tumor angiogenesis in metastatic
breast cancer models. However, it remains unclear if or how metformin remodels the abnormal vasculature of
metastatic breast cancer, while inhibiting angiogenesis.
Methods: Metastatic breast cancer models were constructed to compare microvessel density (MVD), vascular
maturity and function, lung metastasis and chemosensitivity in metformin-treated or untreated mice. Protein array
assay and transcriptome sequencing were performed for genetic screening. Lentiviral shRNA-PDGF-B transfection
was used for observing the contribution of PDGF-B knockdown to metformin’s vascular effects.
Results: Metastatic breast cancers were characterized by an excessively angiogenic, immature and morphologically
abnormal vasculature. Compared to control, metformin significantly reduced MVD, leakage and hypoxia, and
increased vascular mural cells coverage and perfusion, namely, “vessel normalization”. Metformin at human blood
concentrations had no direct effect on the migration and proliferation of cancer cells. Based on that, reduced lung
metastasis of the primary tumor and improved chemosensitization by metformin were assumed to be mediated via
metformin’s vascular effects. Further results of genetic screening and in vivo experiments showed that the
downregulation of platelet-derived growth factor B (PDGF-B) greatly contributed to the metformin-induced vessel
normalization.
Conclusions: These findings provide pre-clinical evidences for the vascular mechanism of metformin-induced
metastasis inhibition and the chemosensitization of metastatic breast cancers.
Keywords: Metformin, Metastatic breast cancer, Vessel normalization, Chemosensitization, PDGF-B

* Correspondence: [email protected]; [email protected]

†
Ji-Chang Wang and Guang-Yue Li are co-first author.
9
Department of Neurosurgery, First Affiliated Hospital of Xi’an Jiaotong
University, No. 277 of the Western Yanta Road, Xi’an 710061, Shaanxi
Province, China
2
Center for Translational Medicine, First Affiliated Hospital of Xi’an Jiaotong
University, No. 277 of the Western Yanta Road, Xi’an 710061, Shaanxi
Province, China
Full list of author information is available at the end of the article

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 2 of 17

Background biguanides also have the potential to enhance the in vivo

Angiogenesis that mediates the formation of new blood toxicity of chemo-drug for cancer treatment [21, 22], but it
vessels serves as a hallmark for cancers [1], of which the was still unclear whether this chemosensitization involves a
critical role in cancer progression has now been widely vascular mechanism.
accepted [2]. Hence, anti-angiogenic drugs (AADs) have The aim of the present study was to investigate the ef-
been extensively developed whose usage constitutes a fects of metformin on vascular maturity and functional-
major modality of anti-tumor therapy [3]. However, ity and angiogenesis. Further results of genetic screening
mechanisms underlying AADs-induced antitumor effects imply the deep involvement of platelet-derived growth
remained unclear. Currently, there are two major hypoth- factor B (PDGF-B) in metformin-induced vessel
eses highly relevant to AADs-related antitumor activities. normalization.
One offers a possible mechanism that cancer cells are
killed through the blocking of blood supply by AADs via Methods
the inhibition of tumor angiogenesis [2]. Up until now, Cell culture, proliferation, colony formation and migration
this hypothesized tumor-starving mechanism has not been assays
clinically verified. Another hypothesis involves the remod- HUVECs and murine 4T1 and human MDA-MB-231
eling of the remaining abnormal vessels [4, 5], also known metastatic breast cancer cell lines were obtained from the
as “vessel normalization”. In the latter hypothesis, the American Type Culture Collection (ATCC, Manassas,
drugs not only suppress both the growth and metastasis VA) and cultured in Dulbecco’s Modified Eagle’s medium
of the tumor but also enhance the chemosensitization of (DMEM) (Invitrogen) supplemented with 10% fetal bovine
cancer cells by improving the vascular maturity and func- serum (FBS). All cell lines used in the study were not
tionality, and ameliorating tumor hypoxia [6]. listed in the database of commonly misidentified cell lines
Conventional therapies targeting tumor angiogenesis is maintained by International Cell Line Authentication
efficacious (in terms of survival benefit) only for some Committee (ICLAC). Cell line Cross-Contamination was
cancers, such as colorectal cancer and renal cell carcin- tested using the Short Tandem Repeat (STR) genotyping
oma, etc., but not for others (e.g. breast cancer, melan- analysis method. Mycoplasma contamination was tested
oma) [6, 7]. Anti-angiogenic benefit in term of survival using Myco-Test Kit of MP Biomedicals (No.093030000)
cannot be seen in all patients with cancers [8, 9], which every three months, or when the growth rate and morph-
have been clinically demonstrated to be responsive to ology of the cell lines were found to be abnormal. All cell
anti-angiogenic therapies. For instance, bevacizumab lines were cultured and maintained in an atmosphere con-
added to chemo-drug did not significantly improve the sisting of CO2 (5%) and the room air (95%) at 37 °C. In
overall survival of the patients with metastatic breast vitro migration was analyzed by inoculating cancer cells in
cancers [9]. This is partially due to the lack of vascular the upper chamber of a transwell (Millipore; 8 μm pore in-
parameters available for predicting the treatment efficacy sert). To assess in vitro colonization ability, cancer cells
[10]. Moreover, intrinsic and acquired resistance have were cultured in a 6-well culture plate, which were finally
been shown to even impair the survival benefit already fixed and stained with crystal violet solution (C0775,
achieved clinically in some cancer patients [3, 11]. Thus, Sigma). Cellular proliferation rate was measured by count-
there is a pressing need for researchers to develop a ing the number of cells in the culture dish (6 cm diameter)
more effective treatment regimen. each day.
Population- and clinic-based studies have demonstrated
the potential anti-proliferative and anti-metastatic activities Chemicals and reagents
of the antidiabetic agent metformin, a member of bigua- Metformin (No.13118), cyclophosphamide (CTX,
nides, when used in cases with malignant diseases [12–14]. No.13849), cisplatin (CPT, No.13119) and imatinib
Data from preclinical studies have revealed the pleiotropic (No.13139) were purchased from Cayman Chemical.
effects of metformin [15, 16]. However, the mechanisms of The active 4-hydroxy-CTX (4-OH-CTX) was obtained
metformin’s effects in carcinogenesis were not fully under- by the ozonization of cyclophosphamide using the
stood, and more details concerning metformin’s effects method described previously [23, 24].
should be further studied. The anti-angiogenesis potential
of metformin has recently been reported by several labora- Protein array assay
tories [17–19]. However, little is known to date about if or To analyze the proteome profile of angiogenesis-related
how metformin remodels the abnormal tumor vasculature, factors, mouse angiogenesis array (ARY015, R&D Sys-
while inhibiting angiogenesis. Since vascular maturity and tem) was incubated with fresh cellular lysates containing
functionality are closely associated with hypoxia and metas- 1 mg protein according to the manufacturer’s instruc-
tasis [20], further researches with a focus on the vascular tions. The RapidStep™ ECL reagent (No.345818, Milli-
mechanism would be hugely meaningful. Additionally, pore) was used, and spot intensities were measured
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 3 of 17

using the NIH ImageJ software (Bethesda, MD) for data quickly and then put on the ice, and each sample was
capture. immediately cryopreserved with liquid nitrogen. Messen-
ger RNA (mRNA) extraction, cDNA synthesis, PCR en-
Mouse models richment, library construction, quality control and
All animals were obtained from Animal Experiment sequencing were performed by Beijing Biomarker Cor-
Center of Xi’an Jiaotong University. To assess the in vivo poration (China, Beijing). 3 independent samples in each
tumor growth, 5Х105 4T1 and 2Х106 MDA-MB-231 group were used for gene expression analysis. Heatmaps
cells were suspended in 100 μL precoated PBS, and then were presented to show the change of gene expression
injected into the fat pad of the 4th left mammary gland levels using Prism 7.0 (GraphPad, USA).
of mice (4T1: female BALB/c; MDA-MB-231: female
nude mice). After 6–8 weeks, the mice without obvious
abnormality in appearance were randomly divided into Immunofluorescence, Histomorphometry and H&E
different groups (8 mice per group) when the mean Staining
tumor volume reached about 50–100 mm2. For the ob- Mouse tissues were fixed in 4% PFA for 12 h at 4 °C, and
servation of chemosensitization, tumor-bearing mice sequentially dehydrated in the 20 and 30% sucrose solu-
were pretreated with metformin (orally) for 5 days before tions, respectively. For the 2D and 3D confocal imaging,
receiving the intraperitoneal injection of CTX. Before tissue samples were cut into 6 μm-thick and 40 μm-thick
the experiment started formally, the proper sample size sections, respectively. The prepared sections were stored
that ensures adequate power for a statistical difference in a − 80 °C Laboratory Freezer (DW-86L728J, Haier).
was estimated using the following formula: N = Single or double immunostaining was performed with
2·[(u0.05) + u(0.10))·S/X]2, “S” indicates standard deviation the following antibodies. Primary antibodies: CD31
of the overall sample, and “X” indicates the difference of (anti-rabbit, ab28364, Abcam; anti-Rat, ab7388, Abcam),
the mean tumor weight between two groups. Tumor PDGF-B (BA0519–2, Boster), α-SMA (BM0002, Boster),
volume was measured with a caliper every two or three VE-cadherin (No.138101, BioLegend), cl-PARP (#9542,
days and calculated using the formula V = 0.523•[a2•A] Cell Signaling), PCNA (BM3888, Boster), cl-Caspase-9
(“a” indicates the minor tumor axis; “A” indicates the (#9542, Cell Signaling), cl-Caspase-3 (#9661, Cell Signal-
major tumor axis). To observe the change of lung me- ing); NG-2 (R&D, MAB6689). Secondary antibodies:
tastasis (4 T1) from primary tumors, tumor-bearing mice Alexa fluor 488-conjugated Goat anti-Rabbit antibody
were fed for at least 28 days after inoculation. (A-11008, Invitrogen), Alexa fluor 488-conjugated Goat
anti-Rat antibody (A-11006, Invitrogen), DyLight
Flow cytometry analysis 550-conjugated Donkey anti-Rat antibody (SA5–10027,
Necrotic and late apoptotic cells were labeled with pro- Invitrogen), Alexa fluor 546-conjugated Goat anti-Rabbit
pidium iodide (PI) at a concentration of 5 μg/mL, and antibody (A-11010, Invitrogen), Alexa fluor
PI+ dead cells were identified by Flow Cytometry (BD 546-conjugated Donkey anti-Mouse antibody (A-10036,
Bioscience, USA). Invitrogen), Alexa fluor 647-conjugated Donkey
anti-Mouse antibody (A-31571, Invitrogen), Alexa fluor
PDGF-B knockdown by shRNA 546-conjugated Goat anti-Rat antibody (A-11081, Invi-
Lentivirus-mediated PDGF-B silencing was performed trogen). Sections were washed with 0.1% PBST, blocked
by transfecting 4T1 cells with control shRNA (against with 5% BSA in PBST at 37 °C for 1 h, and permeabilized
scrambled sequence) or mouse PDGF-B shRNA. The with the 0.2% triton X-100 solution for 15–30 min.
transfection procedure was carried out according to the For fluorescent 3D-reconstruction, 40 μm-thick sec-
manufacture’s protocol. Positively transfected cells were tions were treated with the 0.1% trypsin retrieval solu-
selected using puromycin, and the silencing efficiency tion at 37 °C for 15–20 min to get enhanced signal for
was investigated with the quantitative real-time polymer- signal detection. Sections were then incubated with pri-
ase chain reaction (PCR). The primers for detecting mary antibodies diluted in 5% BSA (0.1% PBST) at 4 °C
mRNA level of mouse PDGF-B (NM_011057.3 → for no less than 24 h, followed by staining with the ap-
NP_035187.2, CCDS: CCDS27656.1) were as follows: the propriate, fluorescently conjugated secondary antibodies.
forward primer: 5`-TCTCTGCTGCTACCTGCGTC Nuclei were counter-stained by DAPI (2-5 μg/mL) at the
T-3`, the reverse primer: 5`- CAGCCCCATCTTCATC- room temperature for 15 min before the fluorescent im-
TACGG -3`. aging. The fluorescent single- or multi-layer images
(2.5–3.5 μm per layer) were obtained using confocal
Transcriptome sequencing assay laser scanning microscopy (Leica, German). Software of
100 mg tissue samples from the 4T1 tumors in LAS AF Lite (Leica, German) was used to perform the
metformin-treated or untreated mice were extracted 3D-reconstrution of CD31 or CD31/α-SMA fluorescent
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 4 of 17

signaling. Furthermore, 5–10 fields (20 x magnification) Results

per tumor were randomly selected and analyzed [25]. Metastatic breast cancers were characterized by an
To observe the perfusion status, perfused vessels were excessively angiogenic and immature vasculature
labeled by the intravenous injection of 20 mg/kg Both MDA-MB-231 and 4T1 breast cancer cell lines
Rhodamine-labeled lectin (RL-1102, Vector Labs) 15 min were characterized by distant lung metastasis when
before the intracardiac perfusion of 40 ml 4% parafor- orthotopically implanted, thus being selected. To ob-
maldehyde with a flux of 10 ml/min. Tumors were then serve the vascular morphology of both tumors, the
extracted, fixed in 4% PFA for 1 h at 4 °C, sequentially 3D reconstruction of those CD31-stained vessels was
dehydrated, embedded in OCT (Tissue-Tek #4583, performed. Microvessels in both cancers manifested
Sakura Finetek, USA), and cut into 6 μm-thick sections. the strong abilities of sprouting and branching
All perfused or un-perfused vessels were immunostained (Fig. 1a), and were distinctly larger. This kind of vas-
with anti-CD31 antibodies (ab28364, Abcam) followed cular phenotype was similar with what was found in
by staining with Alexa fluor 488-conjugated Goat angiogenic cancers. It was further confirmed that a
anti-Rabbit secondary antibodies (A-11008, Invitrogen). large number of vessels were found to be PCNA+ in
For the observation of the vascular leakage of tumors, 4T1 cancers (53.4%), indicating that endothelial cells
100 mg/kg Fluorescein Isothiocyanate (FITC)-conjugated (ECs) were proliferative (Fig. 1b). These data sug-
Dextran (70kD, No.53471, Sigma) was intravenously gested the involvement of excessive angiogenesis in
injected into tumor-bearing mice 10 min before tumors the progression of metastatic breast cancers.
were harvested. To detect tumor hypoxia, 60 mg/kg Next, we focused on the vascular maturity and func-
PIMO (Hypoxyprobe Inc.) was intravenously injected tion. Vascular lumens in the normal breast tissue were
into the tail vein of tumor-bearing mice 90 min before extensively covered by vascular smooth muscle cells
tumors were wholly extracted. PIMO+ hypoxic cells in (VSMCs, marker: α-SMA), a common vascular mural
tumor sections were then immunostained with cell, presenting an inerratic lumen morphology (Fig. 1c).
anti-PIMO antibody (Hypoxyprobe Inc.) according to However, vessels in the 4T1 cancer were poorly covered
the manufacturer’s instructions. by VSMCs. Considering that α-SMA was reported to be
The H&E-stained paraffin sections (4 μm) of those fixed abundantly expressed in cancer-associated fibroblasts,
tumors or lung tissues were assessed for tumor necrosis, the pericyte coverage was further examined using a
hemorrhage and metastatic nodules. To observe the che- NG-2 antibody. Consistently, only few NG-2+ pericytes
mosensitization, tumor sections from Cisplatin-treated (PCs) covered the vessel lumen (indicated by white tri-
mice were stained with anti-Cisplatin-modified DNA anti- angles in Fig. 1d). These data suggested that vessels in
body (GTX17412, Genetex). Paraffin-embedded sections metastatic breast cancers were structurally immature. It
were sequentially incubated with secondary antibodies was worth noting that most of VSMCs and PCs were
(SV0002, Boster), 3, 3′-diaminobenzidine (DAB, ZLI-9017, disassociated or detached from the vascular lumen (Fig.
ZSGB-Bio) and the hematoxylin (H9627, Sigma) solution. 1c and d).

Statistical analysis Metastatic breast cancers were hypoxic with leaky vessels
Quantitative analysis was performed using the Prism 5.0 Hypoxia and the vascular leakage of tumors were fur-
or 7.0 software (GraphPad, San Diego, CA). All quantita- ther investigated, since the poor vascular maturity
tive data were represented by mean ± SEM. Kolmogorov- contributes to both [26, 27]. Vessel leakage was de-
Smirnov normality test was performed to analyze the nor- tected by the intravenous injection of the
mal distribution, and coefficient of variation (CV) was FITC-conjugated Dextran (70kD) and the counter-
used to estimate the variation of data within each group. staining with CD31. In 4T1 cancer, a large amount of
When CV was greater than 15%, the data was considered FITC-conjugated dextran was found to be located
to be abnormal. Bartlett’s test was performed for investi- outside the CD31+ vessel (Fig. 1e, white arrows), indi-
gating the homogeneity of variances between the groups. cating that it was extravasated from the blood vessel
For any set of data which was not normally distributed, to extravascular regions. To observe the hypoxic sta-
nonparametric Wilcoxon or Kruskal-Wallis test was per- tus, PIMO, a reagent for detecting hypoxia, was intra-
formed to investigate the statistical difference between venously injected to tumor-bearing mice after 2 weeks
two or multiple independent samples. The statistical sig- of administration. Further IHC staining showed that
nificance between two groups and multiple groups was 28.75 ± 5.58% areas were stained by the hypoxic
defined as P < 0.05 by two-tailed student’s t test or marker PIMO in 4 T1 tumors (Fig. 1f ), while the nor-
one-way ANOVA t-test. Two-way ANOVA analysis was mal breast tissue was almost devoid of any hypoxic
performed when an additional factor or variant was in- region. As a key regulator of angiogenesis in cancer,
volved in the experiment. hypoxia was then investigated to show if there was a
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 5 of 17

Fig. 1 Vascular morphology, maturity and functional status of the metastatic breast cancers. 4 T1 and MDA-MB-231 metastatic breast cancer cells
were orthotopically injected into the fat pad of mice. (a) 3D-reconstruction of CD31+ vessels in 4 T1 and MDA-MB-231 (MM-231) tumors from
untreated mice. Tumor sections were stained with an anti-CD31 antibody. Bars: 100 μm. (b) Immunofluorescent double-staining for PCNA (Green)
and CD31 (Red) in sections of untreated CT-26 tumors. White arrows indicate PCNA nucleus positive endothelial cells that are proliferating. Nuclei
were counterstained with DAPI. Bars: 100 μm. (c) Double immunostaining (Left) for CD31 (Green) and α-SMA (Red) in frozen sections of normal
breast and 4 T1 tumors, and quantification (right) of percentage of α-SMA+/CD31+ vessels (of CD31+ vessels; n = 8). White and yellow triangles
indicate disassociated vascular smooth muscle cells (VSMCs) and associated VSMCs, respectively. Nuclei were counterstained with DAPI. Bars:
100 μm. (d) Double immunostaining for CD31 (Green) and NG-2 (Red) in frozen sections of untreated 4 T1 tumors. White triangle indicates the
pericyte associated with the vessel wall. Bars: 100 μm. (e) Representative images showing FITC-conjugated dextran perfused CD31+ vessels (Red)
in 4 T1 tumors from untreated mice. FITC-conjugated dextran was injected through the tail vein in advance. White triangles indicate CD31+
vascular lumen containing FITC-Dextran; white arrows indicate dextran leaking outside the vessel wall. Bars: 100 μm. (f) Representative image for
showing Pimonidazole (PIMO; brown) staining, and quantification of PIMO+ hypoxic areas in 4 T1 tumors and normal breast of BALB/c mice
(Lower Right Corner, n = 8). Bars: 100 μm. Quantitative data are indicated as mean ± SEM. ***p < 0.001

discrepancy in hypoxic areas (%) between hypo- and that of the hypo-vascular region. These evidences sug-
hyper-vascular regions. As shown in Additional file 1: gest metastatic breast cancer showed signs of leaky
Figure S1A and S1B, the hypoxic fraction of the vessel and hypoxia, which were closely associated
hyper-vascular region was significantly higher than with the immature vasculature.
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 6 of 17

The cell non-autonomous mechanism of metformin in Furthermore, the clinically relevant dose of metformin
inhibiting the growth and metastasis of tumors also greatly decreased the lung metastasis index by
To bring out the clinical antitumor effect of metformin, 53.9% (Fig. 2c and d). As the metastasis index indicates
its routine clinical dose (about 30 mg/kg•day) widely pre- the number of metastatic nodules per gram of the pri-
scribed in antidiabetic use was increased to 225 mg/kg•day mary tumor, the inhibition of tumor metastasis may be
for anti-tumor use in the mice by the body surface area independent of the direct suppression of the primary
normalization [28]. To exclude any low dose-associated tumor growth. Intriguingly, however, metformin did not
hermetic response to metformin, low dose groups (0.1, 1 alter the in vitro colonization of 4T1 cells (Fig. 2e), while
and 25 mg/kg•day) were concurrently set up. The dose of inducing great suppression of the metastatic burden of
225 mg/kg•day was referred to as the clinically relevant the lungs (Fig. 2f ), but not of the liver or the spleen.
dose of metformin. As shown in Fig. 2a, 4T1 cancer did Then, it was explored whether metformin directly af-
not respond to low doses of metformin (0.1, 1 and 25 mg/ fected the proliferation and the metastasis of breast can-
kg•day), indicating the absence of any hermetic dose re- cer cells. Since the plasma concentration of metformin
sponse for taking metformin. Compared to the control, was estimated to be approximately 60 μM in patients,
metformin at a dose of 225 mg/kg•day significantly re- both 50 μM and 100 μM concentrations were used for
duced the growth of 4T1 cancers by 63.7% (Fig. 2a and b). further in vitro analyses. At both doses, metformin did

Fig. 2 Effects of metformin on the growth and metastasis of the metastatic breast cancers. (a and b) Growth curve (cm3) and measurement of
weight (g) of 4 T1 tumors from BALB/c mice untreated or treated with metformin at different concentrations (0.1, 1, 25 and 225 mg/kg•day; n = 8).
(c and d) H&E staining for 4 T1 primary tumor lung metastasis, and quantification of 4 T1 metastatic index (metastatic nodules per gram of the
primary tumor; n = 8). Blue arrows indicate metastatic nodules in lung. (e) Representative images of in vitro colonization of 4 T1 cancer cells,
which were untreated or persistently treated with 50 μM or 100 μM metformin (repeated for 5 times). Cancer cells were inoculated into each well
of a 6-well culture plate. (f) Bar graphs showing reduced lung weight, and comparable weights of liver and spleen of mice untreated or treated
with metformin at concentration of 225 mg/kg•day or 25 mg/kg•day (n = 6). Organ weight was normalized for body weight. (g) In vitro
proliferation curve of 4 T1 cancer cells untreated or treated with metformin at concentrations of 50 or 100 μM (repeated for 5 times). (h)
Representative images showing 4 T1 cancer cell migration in vitro, and quantification of migrated cancer cells per 400 X field (n = 5). (I) H&E
staining of paraffin-processed tumor tissues from metformin-treated or untreated mice, showing the comparably direct invasion of cancer cells to
the peritumoral muscle. Black arrows indicate muscle fibers penetrating into the tumor. Bars: 200 μm. Quantitative data are indicated as mean ±
SEM. **p < 0.01; ***p < 0.001
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 7 of 17

not alter the proliferation and the migration of breast weaker than in the control group (Fig. 4a), while the
cancer cells in vitro (Fig. 2g and h). In addition, CD31 signal intensity did not differ between groups (Fig.
metformin-treated 4 T1 cancers did not have the same 4d). In addition, metformin induced a shift of the size of
sharp, demarcated borders as the control (Fig. 2i), indi- the tumor vessels towards a smaller one (Fig. 4e).
cating that metformin had no effect on the direct inva- The inhibition of angiogenesis can often be accompan-
sion of cancer cells to the peritumoral muscle. Overall, ied by a change of vascular maturity [31, 32]. VSMCs
these data suggested that the antimetastatic and antipro- and pericytes coverage, the status of the vascular base-
liferative effects of metformin probably were not medi- ment membrane and the signatures of vascular maturity
ated by a cancerous cell-autonomous mechanism. [4] were observed by the staining of α-SMA, NG-2 and
VE-Cadherin. Metformin-treated 4T1 tumors exhibited
Metformin enhanced the susceptibility of breast cancer a higher percentage of vessels with α-SMA+ VSMCs
cells to chemotherapy coverage (White arrows, Fig. 4f and g). It is worth noting
Vascular maturity and functionality have been reported to that metformin-treated 4T1 tumors had fewer VSMCs
be associated with chemosensitization. Therefore, the ef- disassociated with vessels than those of the control (Yel-
fects of the metformin pretreatment on in vivo sensitivity low arrows, Fig. 4f ). In addition, less VSMCs disasso-
of cancer cells to low-dose cyclophosphamide (CTX, 20 ciated with vessels were found in 4T1 tumors of the
mg/kg•day) was investigated. Tumor-bearing mice were metformin-treated mice. Metformin administration
orally pretreated with metformin starting from the sev- shifted the discontinuous VE-Cadherin+ vascular base-
enth day after inoculation. Compared to the single CTX ment membrane towards a continuous phenotype while
treatment, the pretreatment combining of CTX and met- increasing its abundance (Fig. 4h). Consistently, metfor-
formin resulted in a more significant anticancer effect and min treatment resulted in a significant increase of the
greatly prolonged the survival time of the tumor-bearing percentage of vessels with the coverage of NG-2+ peri-
mice (Fig. 3a and b). This metformin-mediated chemosen- cytes in orthotopic MDA-MB-231 tumors (Fig. 4i and j).
sitization was accompanied by the aggravation of tumor
necrosis and hemorrhage (Fig. 3c-e). Metformin improved the vascular functionality of
Double staining for CD31 and cl-PARP further verified metastatic breast cancers
the phenomenon described above. Compared to single Structural maturity is closely associated with the vascu-
CTX group, the combined metformin pretreatment re- lar functional status [33]. We then focused on the effect
sulted in a significant increase of the number of of the administration of metformin on the perfusion sta-
cl-PARP+ 4T1 cells (Fig. 3f-h). Interestingly, most of tus. A TRITC-conjugated Lectin, which can bind to the
these apoptotic cancer cells were located close to CD31+ surface of ECs lining along the blood flow, was intraven-
vessels, suggesting the involvement of the vascular ously injected. Critically, both metformin-treated 4T1
mechanism in the enhancement of the toxicity of and MDA-MB-231 tumors exhibited significantly higher
chemo-drug. It is worth noting that the combined met- percentage of Lectin+ vessels than those of the control
formin pretreatment did not increase the proportion of (White arrows, Fig. 5a-d). This functional improvement
PARP+ vessels in 4T1 cancer (Fig. 3h). Furthermore, was further validated by the fact that metformin greatly
metformin pretreatment significantly increased the num- decreased tumor hypoxia (Fig. 5e and f ), indicated by re-
ber of cisplatin-DNA adduct-positive 4T1 tumor cells duced positive areas of pimonidazole (PIMO). Further-
(Red arrows; Fig. 3i and j), suggesting that more cyto- more, vessels in metformin-treated tumors were less
toxic drugs were delivered to the tumor. More import- leaky than in the control group (Fig. 5g). Collectively,
antly, metformin did not have the potential to enhance these findings suggested that metformin improved the
CTX-induced cell necrosis (indicated by propidium iod- vessel functionality by increasing the vascular perfusion
ide (PI)+ cells, Fig. 3k) in vitro. and decreasing the vascular leakiness, thus serving as a
mediator for the normalization of breast cancer vessels.
Metformin inhibited angiogenesis and improved vascular
maturity Metformin downregulated tumoral PDGF-B and reduced
Angiogenesis inhibition is one of the mechanisms sup- the vascular compression
pressing the tumor growth [29, 30]. Therefore, it was Efforts were further devoted to the gaining of mechanistic
then investigated whether metformin affects breast can- insights into the metformin-induced vessel normalization.
cer angiogenesis. CD31 staining for microvessels showed Excessive pro-angiogenic factors contributed to the abnor-
that normal dose of metformin (225 mg/kg) significantly mal angiogenesis [34, 35], resulting in an immature vascu-
reduced the micro-vessel density (MVD) and vascular lature [36], which thus motivated us to investigate the
branching points (Fig. 4a-c). The sprouting ability of the underlying mechanism with a special focus on angiogenic
micro-vessels in metformin-treated 4T1 tumors were factors. Results of the angiogenesis protein array showed
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 8 of 17

Fig. 3 The in vivo effects of metformin pretreatment on cancer responses to cyclophosphamide (CTX). (a and b) Growth curve and survival curve
of 4 T1 cancers from mice untreated or treated with metformin (225 mg/kg•day), low dose CTX (Cyclophosphamide, 20 mg/kg•day) or the
combined treatment (n = 8). Mice were pretreated with metformin for 5 days before receiving CTX. (C-E) (c) H&E staining of sections of 4 T1
cancers, and quantification of (d) necrotic and (e) hemorrhagic areas, revealing enhanced cancer responses to CTX by metformin pretreatment
(n = 8). “N” indicates “Necrosis”; black arrows indicate hemorrhage; “C & M” indicates the combination of CTX and metformin. Bars: 200 μm. (f)
Double staining for CD31 and PARP of 4 T1 tumor sections, and quantification of (g) PARP+ CD31+ cancer cells and (h) PARP+ CD31+ vessels (n =
8). Tumor-bearing mice were treated with metformin (pretreatment for 5 days), CTX or combined treatment. White arrows indicate PARP+
endothelial cells (ECs); yellow triangles indicate PARP+ cancer cells. (i) Representative images showed increased cisplatin-DNA adducts in 4 T1
tumors from metformin pre-treated mice (225 mg/kg•day, lasting for 5 days). After that, mice were treated with intraperitoneal injections of
cisplatin (CPT, 5 mg/kg, every 2 days). The right column refers to the reverse-color image of the left column. Red arrows indicate the cells with
positive cisplatin-DNA adducts. Magnification:200Х. (j) Quantification of cells with nuclear cisplatin adduct in 4 T1 cancers (n = 8). (k) Quantification
of PI+ necrotic ECs and 4 T1 cells in vitro upon combined 4-OH (activated CTX) treatment (n = 5). Quantitative data are indicated as mean ± SEM.
*p < 0.05; **p < 0.01; ***p < 0.001

that metformin treatment resulted in a great reduction of than 50 FPKM (Fig. 6c), but not greatly affected. Among
PDGF-B protein levels of 4T1 cancer cells in vitro (Fig. 6a PDGFs, the expression of PDGF-B was the most signifi-
and b), while the protein levels of some other factors were cantly decreased (Fig. 6d), indicating the initial high
slightly affected. To validate this initial screening result, PDGF-B expression in 4T1 cells. Consistently, metformin
RNA sequencing was performed to detect the change of treatment resulted in a great reduction in the PDGF-B sig-
those affected factors in vivo. Expression levels of endo- nal intensity of 4T1 tumors (Fig. 6e). These results were
glin, endostatin, MMP-9 and osteopontin were higher consistent with the finding of our previously published
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 9 of 17

Fig. 4 Effects of the metformin administration on angiogenesis and the vascular maturity. (a) CD31 immunostaining in sections of 4T1 tumors
from mice untreated or treated with metformin at concentrations of 25 mg/kg•day or 225 mg/kg•day. White arrows indicate vascular sprouts of
microvessels. Bars: 100 μm. (B and C) quantifications of vascular (b) sprouts and (c) branches (per mm2; n = 8). (d) Quantification of mean
fluorescent intensity of CD31 signaling of vessels in 4 T1 tumors (n = 8), revealing the unaffected CD31 expression level of vessels. (e) Analyzes of
diameter distribution of vessels in 4T1 tumors from untreated or metformin-treated mice (n = 8). (f) Double immunostaining for CD31 (Green) and
α-SMA (Red) in frozen sections of 4T1 tumors from untreated or metformin-treated mice, and (g) quantification of percentage of α-SMA+ CD31+
vessels (of CD31+ vessels; n = 8). White arrows indicate vessels with VSMCs coverage; yellows arrows indicate VSMCs disassociated with vessels.
Bars: 100 μm. (h) Double immunostaining for CD31 and VE-cadherin of 4T1 tumor sections, revealing continuous and more abundant in
metformin-treated than untreated tumors. Bars: 50 μm. (i) Double staining for CD31 (green) and NG-2 (red), revealing more vessels with pericyte
coverage in metformin-treated than untreated MDA-MB-231 (MM-231) tumors. White and yellow arrows indicate pericytes disassociated and
associated with vessels, respectively; white triangles indicate pericytes that are partially associated with vessels. Bars: 100 μm. (j) Quantification of
NG-2+ CD31+ vessels (of CD31+ vessels) in MDA-MB-231 tumors (n = 8). Quantitative data are indicated as mean ± SEM. *p < 0.05;
**p < 0.01; ***p < 0.001

article that metformin reduced the PDGF-B signal inten- T1 tumors (Additional file 1: Figure S1C). These evidences
sity in the peri-necrotic regions of the 4 T1 tumor [37]. suggested that the down-regulation of PDGF-B by metfor-
Overall, these data suggested PDGF-B downregulation min might be contributed to reduce the compression of
was deeply involved in metformin’s vascular remodeling cancer cells to vessels.
effects.
PDGF-B has been demonstrated to increase the intersti- High PDGF-B expression was associated with poor
tial fluid pressure (IFP), which is assumed to compress the prognosis and high CD31 expression
intratumoral vessels [38], thus affecting the vascular func- To determine if PDGF-B expression is associated with
tionality. In 4T1 tumors of the control group, those major prognosis in female patients with breast cancers, the
arteries with multiple layers and VSMCs coverage were published TCGA dataset containing both gene expres-
found to be severely compressed (Fig. 6f), with only very sion and survival data was employed. 1064 patients were
few arteries were patent with open lumen. Compared to divided into high (≥10.14 FPKM) and low (< 10.14
the control group, the percentage of patent vessels was FPKM) PDGF-B expression groups. High PDGF-B
significantly elevated in 4T1 tumors of the metformin mRNA expression levels were associated with a signifi-
group (Fig. 6f). PDGF-B signaling has been reported to cantly decreased 5-year survival rate (Fig. 6g). Further
regulate the interstitial fluid pressure accompanied with analysis for the relationship between mRNA levels of
the hyper-activated Hippo signaling. As was expected, the PDGF-B, PCNA, Ki-67 and CD31 was performed to in-
expression level of YAP was decreased by metformin in 4 vestigate what contributed to the poor prognosis. Both
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 10 of 17

Fig. 5 Effects of metformin administration on hypoxia and the vascular functionality. (a-d) Representative images showing more lectin-perfused
CD31+ vessels in metformin-treated than untreated (A) 4T1 and (c) MDA-MB-231 cancers (n = 8). TRITC-conjugated lectin (Red) was intravenously
injected into tail veins 15 min before the sacrifice. Tumor sections were further immunostained with anti-CD31 antibody (Green). White arrows
indicate CD31+ vessels with lectin perfusion. Higher (B) percentage (of CD31+ vessels) and (d) density (per mm2) of Lectin+ CD31+ vessels in
metformin-treated than untreated tumors. Bars: 100 μm. (e) Immunohistochemical staining for pimonidazole (PIMO; brown) and (f) quantification
of PIMO+ hypoxic area in 4 T1 cancers from untreated or metformin-treated mice, revealing ameliorated tumor hypoxia by metformin (n = 8).
Bars: 500 μm. (g) Representative images showing decreased vascular leakage in metformin-treated than untreated 4 T1 tumors. FITC-dextran
(green) was injected through tail vein in advance. Tumor sections was counterstained with anti-CD31 antibody (Red). White triangles indicate the
dextran leaking outside the vessel wall. Bars: 100 μm. Quantitative data are indicated as mean ± SEM. *p < 0.05

PCNA and Ki-67 mRNA levels were not correlated with PDGF-Rβ blockade abrogated metformin’s vascular
PDGF-B mRNA levels (Fig. 6h). Interestingly, increased remodeling effects
CD31 mRNA levels were associated with significantly in- PDGFR-β is the receptor of PDGF-B and has been ex-
creased PDGF-B mRNA levels (Fig. 6h). These evidences tensively studied for its critical role in vascular remodel-
suggested the contribution of PDGF-B to poor prognosis ing. Therefore, we then focused on exploring whether
was associated with the vascular mechanism, but not the PDGF-Rβ was involved in metformin’s vascular remodel-
direct proliferative promotion of breast cancer cells. ing effects. Imatinib, a drug that can specifically block
PDGFR-β signaling, was used in this study. Similar to
sh-PDGF-B and metformin, PDGFR-β blockade by ima-
PDGF-B knockdown in breast cancers with high PDGF-B tinib led to decreased CD31+ areas and an increased vas-
expression improved vascular maturity and function cular coverage of VSMCs (Fig. 7a and b). Surprisingly,
To further validate if PDGF-B downregulation mediates imatinib apparently abrogated the metformin-induced
the vessel normalization in cancers with PDGF-B ex- increase of the VSMCs coverage on vessels (Fig. 7a and
pression, 4T1 cells were transfected by lentiviral d), indicating the involvement of PDGFR-β signaling in
shRNA-PDGF-B. As shown in Fig. 6e, PDGF-B knock- metformin’s vascular remodeling effects. Notably, the
down greatly decreased the mean fluorescent PDGF-B combination of imatinib and metformin decreased the
intensity from 20.7 to 6.2 pixels in the sections of im- percentage of VSMCs associated with vessels to a com-
planted tumors (Fig. 6e), indicating a successful knock- parable level of the shRNA control group (Figure7B).
down of PDGF-B. Furthermore, PDGF-B knockdown These evidences suggested that PDGFR-β is indispens-
markedly reduced the MVD and CD31+ areas of 4T1 able for the subsequent vascular remodeling when
cancers compared with the control group (Fig. 7a and PDGF-B was downregulated.
b), suggesting the angiogenic effect of tumoral PDGF-B
level. Similar with metformin, PDGF-B knockdown sig- PDGF-B knockdown improved the chemosensitization of
nificantly increased percentages of α-SMA+ vessels and CTX and reduced lung metastasis
VSMCs, demonstrating the improved vascular maturity As PDGF-B knockdown induces the vessel normalization,
(Fig. 7a and b). In addition, vessels in shRNA-PDGF-B 4 we then investigated whether PDGF-B knockdown im-
T1 tumors were less leaky than those in shRNA-Con tu- proves the chemosensitization and limits the primary lung
mors (Fig. 7c), and there were more Lectin-perfused ves- metastasis of breast tumors. As shown in Fig. 7e, PDGF-B
sels in shRNA-PDGF-B 4 T1 tumors (Fig. 7d). knockdown induced a slight reduction in 4T1 cancer
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 11 of 17

Fig. 6 Downregulation of PDGF-B by the metformin treatment. (a) Immunoblotting for cancer cell-derived angiogenic factors using mouse
angiogenesis arrays, and (b) quantification of grey intensity of each factor (independent of 3 experiments). Cancer cells were treated with
metformin (100 μM) for 48 h. (c) Column chart for showing the transcriptional levels of angiogenesis-related factors. FPKM indicates Fragments
Per Kilobase of transcript per Million fragments mapped. Met indicates metformin; Met1, Met2 and Met3 indicate different reports of data of
independent tumors in metformin group. (d) Heatmap analysis of the effect of metformin on transcriptional levels of top 9 angiogenesis-related
genes of 4T1 tumors (n = 3). The fold change ranges from − 0.5 to 0.5. “Sig.” indicates “Significance”. (e) Double immunostaining for CD31 and
PDGF-B in 4T1 orthotopically implanted tumors (Left). shRNA-PDGF-B- and shRNA-Control-transfected 4T1 tumor-bearing mice received control or
metformin treatment (225 mg/kg•day). Quantification of PDGF-B fluorescent intensity and microvessel density (Right; n = 8). (f) Immunostaining
for CD31 and α-SMA in 4T1 cancers, revealing increased percentage of patent microarteries (n = 8). White arrows indicate the compressed lumen
of arteries distributed in 4T1 cancer stroma. (g-h) Survival curve for analyzing the association of PDGF-B mRNA expression in primary breast tumor
with survival probability of patients with breast cancer (g). PDGF-B mRNA levels was positively correlated with mRNA level of CD31, but not PCNA
and Ki-67. Data was obtained from TCGA dataset. 1064 samples were included. FPKM: Fragments Per Kilobase of transcript per Million fragments
mapped. Quantitative data are indicated as mean ± SEM. “ns” indicates not statistically significant; *p < 0.05; **p < 0.01; ***p < 0.001

necrosis without affecting hemorrhage. When treated with compared with those in the shRNA control group (Fig.
CTX, shRNA-PDGF-B 4T1 cancers exhibited a significant 7e). Since increased blood perfusion can bring more drugs
increase in both tumor necrosis and hemorrhage into the tumor tissue, these findings thus suggest that the
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 12 of 17

Fig. 7 (See legend on next page.)

Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 13 of 17

(See figure on previous page.)

Fig. 7 Effects of PDGF-B knockdown on metastasis, chemosensitization and vascular maturity and functionality. (a) Double staining for CD31
(green) and α-SMA (red) of 4T1 tumor sections. shRNA-PDGF-B- and shRNA-Control-transfected 4 T1 tumor-bearing mice received control,
metformin (225 mg/kg), imatinib (60 mg/kg) for the combined treatment. Fluorescent signaling of series thin-layer scanning were reconstructed
for 3D observation of VSMCs on vessels. VSMCs No. is at the bottom of the panel and indicated as mean ± SEM (n = 8). (b) Quantification of ratio
of α-SMA+ area/CD31+ area (upper), CD31+ area per μm2 (middle) and percentage of α-SMA+ VSMCs associated with vessels (bottom) in 4 T1
cancers (n = 8). (c) Representative images showing decreased vascular leakage in shRNA-PDGF-B- than shRNA-Con-transfected 4T1 tumors. Fitc-
dextran (green) was injected through tail vein 10 mins before sacrifice. Tumor sections was counterstained with anti-CD31 antibody (Red). White
arrows indicate the dextran leaking outside the vessel wall. Bars: 100 μm. (d) Representative images showing more lectin-perfused CD31+ vessels
in shRNA-PDGF-B- than shRNA-Con-transfected 4T1 tumors (n = 8). Red: perfused lectin; Green: CD31+ vessels. White arrows indicate CD31+
vessels with lectin perfusion. Percentage of Lectin+/CD31+ vessels (of CD31+ vessels) is indicated as mean ± SEM (at the bottom of the panel). (e)
H&E staining of sections of 4T1 cancers (Left) and quantification of necrotic and hemorrhagic areas (Right; n = 8). shRNA-PDGF-B or shRNA-Con-
transfected 4T1 tumor-bearing mice received low dose CTX (20 mg/kg•day). “N” indicates “necrosis”; black arrows indicate tumor hemorrhage. (f)
Decreased primary tumor lung metastasis in mice bearing shRNA-PDGF-B 4T1 cancer cells than that in mice bearing shRNA-Con 4T1 cells. (Upper)
H&E staining for 4T1 tumor sections; (Lower) quantification of 4T1 metastatic index (metastatic nodules per gram of primary tumor; n = 8). Red
asterisk indicates lung metastasis nodule. Magnification: 200Х. (g) Schematic diagram of metformin-induced vascular remodeling in metastatic
breast cancers. Metastatic breast cancers are angiogenic with hypoperfusion and vascular immaturity, which contribute to vascular leakage,
chemoresistance, hypoxia and distant metastasis. By decreasing PDGF-B of those metastatic breast cancers, metformin inhibits angiogenesis and
improved the vascular maturity and functionality, therefore improving the chemosensitivity and reducing the distant metastasis

inhibition of tumoral PDGF-B enhances chemosensitiza- Therefore, metformin is assumed to prune the breast
tion by increasing the drug delivery owing to the im- cancer vasculature more excessively. However, to date, it
proved vascular function. In further characterization of has not been reported that metformin aggravated tumor
the primary lung metastasis (Fig. 7f), the metastatic index hypoxia [44]. Consistently, metformin ameliorated the
of the shRNA PDGF-B group was significantly reduced, hypoxia of two metastatic breast cancers, while MVD
compared to the shRNA control group. was greatly reduced. These evidences suggest there ex-
ists a mechanism independent of affecting the vascula-
Discussion ture, which is accountable for the ameliorated hypoxia.
Bevacizumab, an AAD approved at an early time, did Metformin is an AMPK activator that induces energetic
not significantly improve the overall survival of patients stress [45]. In this condition, tumor cells were metabol-
with metastatic breast cancers [9, 39]. Bevacizumab was ically reprogrammed to consume less oxygen, thus coun-
initially designed to neutralize the VEGF, thus inhibiting teracting or reversing the vascular pruning-induced
VEGFR-2-mediated angiogenesis and tumor growth. hypoxia [44].
However, as previously reported, metastatic breast can- Currently, the issue of tumoral PDGF-B’s effect on the
cer cells had high expressions of several other vascular maturity has become controversial [46].
pro-angiogenic factors in addition to VEGF [40], such as Platelet-derived growth factor B (PDGF-B), a member of
FGF-2, Ang-2 and PDGF-B [37]. Thus, in theory, AADs the PDGFs family [46], binds to its receptors, such as
designed for targeting a single factor may be not enough PDGF receptor β (PDGFR-β), to induce the cell survival,
[41], which offers an explanation for AAD’s ineffective- proliferation and migration [47]. It is now widely ac-
ness in treating cancers derived from some organ sys- cepted that the endothelial PDGF-B regulates the re-
tems. Our laboratory previously found that metformin cruitment of PCs [47]. Thus, the downregulation of
inhibited the expression of VEGF, Ang-2 and FGF-2 in a PDGF-B should reduce the vascular maturity. However,
metastatic breast cancer model [37]. Besides, PDGF-B PDGF-B was highly expressed by some tumors [48, 49],
was screened out by transcriptome sequencing, and which were inversely characterized by an excessively an-
angiogenesis was greatly inhibited by PDGF-B knock- giogenic and immature vasculature. Further blockade of
down. This result was validated by the clinical data that PDGF-B/PDGFR-β significantly increased the vascular
the PDGF-B expression level was positively associated maturity of tumors with high PDGF-B expression [49],
with the CD31 expression level. Due to these findings, but reduced that of tumors with low PDGF-B expres-
metformin should be considered as a reagent with a sion. These evidences indicated that high and low ex-
broad range of targeted factors possibly more available pression of PDGF-B in tumors might have opposite
than the conventional AADs. effects on vascular maturity. To date, this mechanism
Another disadvantage of traditional AADs is the exces- has not been reported in studies on metastatic breast
sive pruning of the tumor vasculature [20], thus leading cancers. Herein, it was found that high PDGF-B expres-
to the hypoxia-mediated tumor cell dissemination [42, sion was detected in the metastatic breast cancer model.
43]. Compared to AADs, metformin has more angio- By reducing tumoral PDGF-B, the metformin treatment
genic targets, such as VEGF, PDGF-B and FGF-2. resulted in the suppression of angiogenesis and a more
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 14 of 17

mature vasculature of metastatic breast cancers, thus non-malignant disorders, which has attracted a great
limiting the distant metastasis and improving chemosen- deal of attention as the safety and frequency of side ef-
sitization. These evidences are further supported by the fects of these medicines have been already proven. For a
poor prognosis of patients with breast cancers of high typical instance, ticlopidine (purinergic receptor P2Y12
PDGF-B expression. These data indicate that the down- inhibitor), which is an anti-coagulant drug to prevent
regulation of PDGF-B in tumors with high expression of the transient ischemic attack (TIA) and stroke, and has
PDGF-B inhibits angiogenesis and improves the vascular been shown to be effective for low-grade glioma and
functionality and maturity. high-grade astrocytoma. This P2Y12 inhibitor increases
As early as in 1970’s, biguanides were reported to po- the intracellular cAMP level and promotes the autoph-
tentiate the antitumor effects of CTX and other agy flux [58]. Notably, tricyclic antidepressants such as
chemo-drugs in vivo [21]. As shown in a recent clinical imipramine promote autophagy in glioma cells synergis-
trial [50], diabetic patients receiving metformin had a tically with this drug by further elevating the intracellu-
greater response rate to chemo-drug than non-diabetic lar cAMP concentration [59].
patients. Despite the increasing efforts that have been Metformin activates AMPK signal pathway, which not
made [22], it was still unclear whether or how metfor- only decreases insulin resistance in type 2 diabetes melli-
min sensitizes cancer cells to chemo-drugs. Recently, tus but also blocks AMPK-mediated mTOR activation
metformin has been demonstrated to directly enhance even in cancer stem cells (CSCs) [60]. mTOR signal is
the toxicity of chemo-drugs [51–53]. However, concen- regulated by amino-acid transporters [61], characterized
trations used in those studies were higher than the by the L-type amino acid transporter 1 (LAT1; SLC7A5)
blood concentration of patients. In the current paper, and the glutamine/amino acid transporter (ASCT2;
metformin was not found to significantly inhibit the SLC1A5), which explains why the AMPK-mTOR axis
proliferation and migration of 4 T1 cancer cells at the functions as a sensor of the dynamic change in the nutri-
blood concentrations in vitro, indicating an indirect ent/growth factor microenvironment. Particularly, the leu-
chemosensitization mechanism. Furthermore, our re- cine uptake via LAT1 activates the mTOR signal pathway
sults suggest metformin-mediated chemosensitization leading to poor prognosis. Because EpCAM is a functional
resulted from the enhancement of drug delivery rather CSC marker that forms a complex with amino-acid trans-
than the direct enhancement of the toxicity of porters such as LAT1 [62], it is reasonable that the LAT1
chemo-drug. expression level would be positively correlated with poor
Interestingly, metformin pretreatment results in a re- prognosis. Therefore, the LKB1-AMPK-mTOR axis is or-
sponse of metastatic breast cancer cells to CTX at a chestrated by the amino-acid concentration in the tumor
lower dose, which was further supported by the in- microenvironment, and this axis promotes the metabolic
creased CPT nucleus adduct+ cells. Critically, those reprogramming of cancer cells in response to the micro-
CTX-induced apoptotic cells were located to the regions environment [58, 63].
adjacent to vessels. Thus, metformin’s chemosensitiza- More clinical and pre-clinical evidences should be
tion effect may be due to the increased delivery of che- provided to validate the vascular mechanism, and if
mo-drugs to the deep tumor. Vessel normalization activity metformin targets a broad range of angiogenesis-
increases the tumor oxygen and reduces the interstitial related factors. As metformin is a drug widely pre-
fluid pressure [36, 54], thus enhancing the generation of scribed for metabolic disorder, further efforts should
oxygen radicals and promoting egress of cytotoxic agents also be devoted to investigating the involvement of the
to the perivascular region in tumors. Consistent with the metabolic mechanism and its contribution to amelio-
results from other laboratories [44], metformin treatment rated hypoxia.
increased tumor oxygenation, reduced tumor hypoxia
and improved radiotherapy response. Furthermore, Conclusions
metformin-induced chemosensitization may be contrib- Our current work provided the pre-clinical evidences for
uted by the reduction in tumoral PDGF-B, which mediates metformin’s effect in remodeling the abnormal breast
resistance by PDGFR-β signaling and increasing IFP [55]. cancer vasculature. Herein, the antidiabetic agent met-
It has also been reported that metformin’s chemosensitiza- formin inhibited the progression of metastatic breast
tion effect was contributed by increased uptake of chemo- cancers, and induced chemosensitization by a vascular
therapeutic by tumor cells [56]. mechanism. By decreasing tumoral PDGF-B, metformin
Given that metformin has long been used for the treat- inhibited the excessive angiogenesis and improved the
ment of type 2 diabetes mellitus, the results of this re- vascular maturity and functionality. The normalized vas-
search is remarkable in terms of drug re-positioning culature was with more mural cells coverage and better
(DR) [57]. DR is a screening for anti-cancer therapeutic basement membrane [30], thus limiting distant metasta-
effects of conventionally administered medications for sis. As the structure determines the function, improved
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 15 of 17

vascular maturity led to an increase in the blood perfu- Consent for publication
sion of tumors, thus allowing more chemo-drugs or Not applicable.

therapeutic particles to be delivered into the tumors

Competing interests
[64]. On the basis of that, the elucidated vascular mech- No potential conflicts of interest were disclosed by all the contributing
anism of metformin is of great significance and value for authors.
the clinical treatment of metastatic breast cancers.
Publisher’s Note
Additional file Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Additional file 1: Figure S1. Effects of metformin on expression level of
Author details
YAP and discrepancy in hypoxia between hypo-vascular and hyper- 1
Department of Vascular Surgery, First Affiliated Hospital of Xi’an Jiaotong
vascular regions. (DOCX 1571 kb)
University, Xi’an 710061, Shaanxi Province, China. 2Center for Translational
Medicine, First Affiliated Hospital of Xi’an Jiaotong University, No. 277 of the
Abbreviations Western Yanta Road, Xi’an 710061, Shaanxi Province, China. 3Department of
3D: 3-dimension; 4-OH-CTX: 4-hydroxy-cyclophosphamide; AADs: Anti- Science and Technology, First Affiliated Hospital of Xi’an Jiaotong University,
angiogenic drugs; ATCC: American Type Culture Collection; CD31: Alternative Xi’an 710061, Shaanxi Province, China. 4Department of Oncological
name of PECAM-1, platelet endothelial adhesion molecule; CPT: Cisplatin; Radiotherapy, First Affiliated Hospital of Xi’an Jiaotong University, Xi’an
CPT-1: Carnitine acyltransferase 1; CTX: Cyclophosphamide; DAB: 3,3′- 710061, Shaanxi Province, China. 5Department of Thoracic Surgery, First
diaminobenzidine; DAPI: 4′,6-diamidino-2-phenylindole; DMEM: Dulbecco Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710061, Shaanxi
modified eagle medium; DR: Drug re-positioning; ECL: Enhanced Province, China. 6Department of Pathology, First Affiliated Hospital of Xi’an
chemiluminescence; ECs: Endothelial cells; EpCAM: Epithelial cell adhesion Jiaotong University, Xi’an 710061, Shaanxi Province, China. 7Department of
molecule; FBS: Fetal bovine serum; FGF-2: Fibroblast growth factor-2; Medical Oncology, First Affiliated Hospital of Xi’an Jiaotong University, Xi’an
GFP: Green fluorescent protein; H&E staining: Hematoxylin and eosin 710061, Shaanxi Province, China. 8Department of Breast Surgery, First
staining; HRP: Horseradish peroxidase; ICLAC: International Cell Line Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710061, Shaanxi
Authentication Committee; IF: Immunofluorescence; IFP: Interstitial fluid Province, China. 9Department of Neurosurgery, First Affiliated Hospital of
pressure; MM-231: MDA-MB-231; MMP: Metalloproteinase; MVD: Microvessel Xi’an Jiaotong University, No. 277 of the Western Yanta Road, Xi’an 710061,
density; OCT: Optimal cutting temperature compound; PARP: Poly-ADP- Shaanxi Province, China. 10Key Laboratory for Tumor Precision Medicine of
ribose polymerase; PBS: Phosphate-buffered solution; PCNA: Proliferating cell Shaanxi Province, First Affiliated Hospital of Xi’an Jiaotong University, No. 277
nuclear antigen; PCR: Polymerase chain reaction; PDGF: Platelet-derived of the Western Yanta Road, Xi’an 710061, Shaanxi Province, China.
growth factor; PDGFR-β: Platelet-derived growth factor receptor-β;
PFK: Phosphofructokinase; PI: Propidium iodide; PIMO: Pimonidazole; Received: 22 January 2019 Accepted: 3 May 2019
shRNA: Small hairpin RNA; STR: Short Tandem Repeat; TCGA: The cancer
genome atlas; TRITC: Tetramethyl rhodamin isothiocyanate; VE-
cadherin: Vascular endothelial-cadherin; VEGF: Vascular endothelial growth References
factor; VSMCs: Vascular smooth muscle cells; α-SMA: Smooth muscle actin α 1. Kerbel RS. Tumor angiogenesis. N Engl J Med. 2008;358:2039–49.
2. Folkman J. Role of angiogenesis in tumor growth and metastasis. In:
Acknowledgements Seminars in oncology: Elsevier; 2002. p. 15–8.
We thank Ruiqi Wang and He Chen for their help with section preparation; 3. Ebos JML, Kerbel RS. Antiangiogenic therapy: impact on invasion, disease
Pan Li for help with immunohistochemical and fluorescent staining; Nan progression, and metastasis. Nat Rev Clin Oncol. 2011;8:210.
Dong for help with mouse breeding; Yan Zhang and Mei Xue for their help 4. Goel S, Duda DG, Xu L, Munn LL, Boucher Y, Fukumura D, Jain RK.
with the technical support in confocal fluorescent imaging and 3D- Normalization of the vasculature for treatment of cancer and other diseases.
reconstruction; Jun Lv and Jie Zheng for their assistance in statistical analysis. Physiol Rev. 2011;91:1071–121.
5. Ouarne M, Bouvard C, Boneva G, Mallet C, Ribeiro J, Desroches-Castan A,
Funding Soleilhac E, Tillet E, Peyruchaud O, Bailly S. BMP9, but not BMP10, acts as a
This work was financially supported by Grants from the National Natural quiescence factor on tumor growth, vessel normalization and metastasis in
Science Foundation of China (No.81672876, No.81602638, No.81602502 and a mouse model of breast cancer. J Exp Clin Cancer Res. 2018;37:209.
No.81502413), Keypoint Research and Invention Program of Shaanxi Province 6. Jain RK. Antiangiogenesis strategies revisited: from starving tumors to
(No.2017SF017 and No.2016SF196), Nature Science Foundation of Shaanxi alleviating hypoxia. Cancer Cell. 2014;26:605–22.
Province (No.2016JM8125), and Fundamental Research Funds for the Central 7. Willett CG, Boucher Y, Di Tomaso E, Duda DG, Munn LL, Tong RT, Chung
Universities (No. xjj2016104). DC, Sahani DV, Kalva SP, Kozin SV. Direct evidence that the VEGF-specific
antibody bevacizumab has antivascular effects in human rectal cancer. Nat
Availability of data and materials Med. 2004;10:145–7.
The TCGA dataset obtained and/or analyzed in the current paper are 8. Bear HD, Tang G, Rastogi P, Geyer CE Jr, Robidoux A, Atkins JN, Baez-Diaz L,
available from TCGA official website and the corresponding author Brufsky AM, Mehta RS, Fehrenbacher L. Bevacizumab added to neoadjuvant
reasonably requested. chemotherapy for breast cancer. N Engl J Med. 2012;366:310–20.
9. Miller K, Wang M, Gralow J, Dickler M, Cobleigh M, Perez EA, Shenkier T,
Authors’ contributions Cella D, Davidson NE. Paclitaxel plus bevacizumab versus paclitaxel alone
J.C. Wang and G.Y. Li performed experiments and summarized the results. B. for metastatic breast cancer. N Engl J Med. 2007;357:2666–76.
Wang and S.X. Han established the gene knockdown cancer models. X. Sun, 10. Tolaney SM, Boucher Y, Duda DG, Martin JD, Seano G, Ancukiewicz M, Barry
Y.W. Shen and C. Zhou prepared cancer tissues and sections for staining. J. WT, Goel S, Lahdenrata J, Isakoff SJ. Role of vascular density and
Feng, S.Y. Lu and J.L. Liu were in charge of dataset analysis. P.J. Liu and M.D. normalization in response to neoadjuvant bevacizumab and chemotherapy
Wang designed the study, analyzed the results and wrote the manuscript. All in breast cancer patients. Proc Natl Acad Sci. 2015;112:14325–30.
authors read and approved the final manuscript. 11. Rapisarda A, Melillo G. Overcoming disappointing results with
antiangiogenic therapy by targeting hypoxia. Nat Rev Clin Oncol. 2012;9:
Ethics approval and consent to participate 378.
All animal procedures were performed in accordance with animal care 12. Kisfalvi K, Eibl G, Sinnett-Smith J, Rozengurt E. Metformin disrupts crosstalk
committee of Xi’an Jiaotong University. Informed consent for participation between G protein-coupled receptor and insulin receptor signaling systems
was not available. and inhibits pancreatic cancer growth. Cancer Res. 2009;69:6539–45.
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 16 of 17

13. Gonzalez-Angulo AM, Meric-Bernstam F. Metformin: a therapeutic induces a pressure gradient across the vasculature and improves drug
opportunity in breast cancer. Clin Cancer Res. 2010;16:1695–700. penetration in tumors. Cancer Res. 2004;64:3731–6.
14. Daugan M, Dufaÿ Wojcicki A, d’Hayer B, Boudy V. Metformin: An anti- 37. Wang JC, Li GY, Li PP, Sun X, Li WM, Li Y, Lu SY, Liu PJ. Suppression of
diabetic drug to fight cancer. Pharmacol Res. 2016;113(Part A):675–85. hypoxia-induced excessive angiogenesis by metformin via elevating tumor
15. Martin-Castillo B, Vazquez-Martin A, Oliveras-Ferraros C, Menendez JA. blood perfusion. Oncotarget. 2017;8:73892–904.
Metformin and cancer: doses, mechanisms and the dandelion and hormetic 38. Chauhan Vikash P, Boucher Y, Ferrone Cristina R, Roberge S, Martin John D,
phenomena. Cell Cycle. 2010;9:1057–64. Stylianopoulos T, Bardeesy N, DePinho Ronald A, Padera Timothy P, Munn
16. Pan Q, Yang GL, Yang JH, Lin SL, Liu N, Liu SS, Liu MY, Zhang LH, Huang YR, Lance L, Jain Rakesh K. Compression of pancreatic tumor blood vessels by
Shen RL, et al. Metformin can block precancerous progression to invasive Hyaluronan is caused by solid stress and not interstitial fluid pressure.
tumors of bladder through inhibiting STAT3-mediated signaling pathways. J Cancer Cell. 2014;26:14–5.
Exp Clin Cancer Res. 2015;34:77. 39. Sledge GW. Anti–vascular endothelial growth factor therapy in breast
17. Qian W, Li J, Chen K, Jiang Z, Cheng L, Zhou C, Yan B, Cao J, Ma Q, Duan W. Cancer: game over? J Clin Oncol. 2015;33:133–5.
Metformin suppresses tumor angiogenesis and enhances the 40. Wang J, Li G, Wang Y, Tang S, Sun X, Feng X, Li Y, Bao G, Li P, Mao X, et al.
chemosensitivity of gemcitabine in a genetically engineered mouse model Suppression of tumor angiogenesis by metformin treatment via a
of pancreatic cancer. Life Sci. 2018;208:253–61. mechanism linked to targeting of HER2/HIF-1alpha/VEGF secretion axis.
18. Garrido MP, Vera C, Vega M, Quest AFG, Romero C: Metformin prevents Oncotarget. 2015;6:44579–92.
nerve growth factor-dependent proliferative and proangiogenic effects in 41. Mitsuhashi A, Goto H, Saijo A, Trung VT, Aono Y, Ogino H, Kuramoto T, Tabata
epithelial ovarian cancer cells and endothelial cells. Therapeutic advances in S, Uehara H, Izumi K, et al. Fibrocyte-like cells mediate acquired resistance to
medical oncology 2018, 10:1758835918770984–1758835918770984. anti-angiogenic therapy with bevacizumab. Nat Commun. 2015;6:8792.
19. Han J, Li Y, Liu X, Zhou T, Sun H, Edwards P, Gao H, Yu FS, Qiao X. 42. Cassavaugh J, Lounsbury KM. Hypoxia-mediated biological control. Journal of
Metformin suppresses retinal angiogenesis and inflammation in vitro and in Cellular Biochemistry. 2011;112:735-44.
vivo. PLoS One. 2018;13:e0193031. 43. Conley SJ, Gheordunescu E, Kakarala P, Newman B, Korkaya H, Heath AN,
20. Carmeliet P, Jain RK. Principles and mechanisms of vessel normalization Clouthier SG, Wicha MS. Antiangiogenic agents increase breast cancer stem
for cancer and other angiogenic diseases. Nat Rev Drug Discov. 2011; cells via the generation of tumor hypoxia. Proc Natl Acad Sci. 2012;109:
10:417–27. 2784-9.
21. Dilman V, Anisimov V. Potentiation of antitumor effect of 44. Zannella VE, Dal Pra A, Muaddi H, McKee TD, Stapleton S, Sykes J, Glicksman
cyclophosphamide and hydrazine sulfate by treatment with the antidiabetic R, Chaib S, Zamiara P, Milosevic M. Reprogramming metabolism with
agent, 1-phenylethylbiguanide (phenformin). Cancer Lett. 1979;7:357–61. metformin improves tumor oxygenation and radiotherapy response. Clin
22. Lin YC, Wu MH, Wei TT, Lin YC, Huang WC, Huang LY, Lin YT, Chen CC. Cancer Res. 2013;19:6741–50.
Metformin sensitizes anticancer effect of dasatinib in head and neck 45. Liu X, Romero IL, Litchfield LM, Lengyel E, Locasale JW. Metformin targets
squamous cell carcinoma cells through AMPK-dependent ER stress. central carbon metabolism and reveals mitochondrial requirements in
Oncotarget. 2014;5:298–308. human cancers. Cell Metab. 2016;24:728–39.
23. Peter G, Wagner T, Hohorst HJ. Studies on 4-hydroperoxycyclophosphamide 46. Cao Y. Multifarious functions of PDGFs and PDGFRs in tumor growth and
(NSC-181815): a simple preparation method and its application for the metastasis. Trends Mol Med. 2013;19:460–73.
synthesis of a new class of "activated" sulfur-containing cyclophosphamide 47. Kazlauskas A. PDGFs and their receptors. Gene. 2017;614:1–7.
(NSC-26271) derivatives. Cancer Treat Rep. 1976;60:429–35. 48. Cao R, Björndahl MA, Religa P, Clasper S, Garvin S, Galter D, Meister B, Ikomi
24. Wagner T, Peter G, Voelcker G, Hohorst HJ. Characterization and quantitative F, Tritsaris K, Dissing S. PDGF-BB induces intratumoral lymphangiogenesis
estimation of activated cyclophosphamide in blood and urine. Cancer Res. and promotes lymphatic metastasis. Cancer Cell. 2004;6:333–45.
1977;37:2592–6. 49. Hosaka K, Yang Y, Seki T, Nakamura M, Andersson P, Rouhi P, Yang X,
25. Fischer C, Jonckx B, Mazzone M, Zacchigna S, Loges S, Pattarini L, Jensen L, Lim S, Feng N. Tumour PDGF-BB expression levels determine dual
Chorianopoulos E, Liesenborghs L, Koch M, De Mol M, et al. Anti-PlGF effects of anti-PDGF drugs on vascular remodelling and metastasis. Nat
inhibits growth of VEGF(R)-inhibitor-resistant tumors without affecting Commun. 2013;4.
healthy vessels. Cell. 2007;131:463–75. 50. Jiralerspong S, Palla SL, Giordano SH, Meric-Bernstam F, Liedtke C, Barnett
26. Jain RK, Booth MF. What brings pericytes to tumor vessels? J Clin Investig. CM, Hsu L, Hung M-C, Hortobagyi GN, Gonzalez-Angulo AM. Metformin and
2003;112:1134–6. pathologic complete responses to neoadjuvant chemotherapy in diabetic
27. Raza A, Franklin MJ, Dudek AZ. Pericytes and vessel maturation during patients with breast cancer. J Clin Oncol. 2009;27:3297–302.
tumor angiogenesis and metastasis. Am J Hematol. 2010;85:593–8. 51. Li L, Han R, Xiao H, Lin C, Wang Y, Liu H, Li K, Chen H, Sun F, Yang Z.
28. Reagan-Shaw S, Nihal M, Ahmad N. Dose translation from animal to human Metformin sensitizes EGFR-TKI–resistant human lung Cancer cells in vitro
studies revisited. FASEB J. 2008;22:659–61. and in vivo through inhibition of IL-6 signaling and EMT reversal. Clin
29. Hicklin DJ, Ellis LM. Role of the vascular endothelial growth factor pathway Cancer Res. 2014;20:2714–26.
in tumor growth and angiogenesis. J Clin Oncol. 2005;23:1011-27. 52. Choi YW, Lim IK. Sensitization of metformin-cytotoxicity by dichloroacetate
30. Maes H, Kuchnio A, Peric A, Moens S, Nys K, De Bock K, Quaegebeur A, via reprogramming glucose metabolism in cancer cells. Cancer Lett. 2014;
Schoors S, Georgiadou M, Wouters J. Tumor vessel normalization by 346:300–8.
chloroquine independent of autophagy. Cancer Cell. 2014;26:190–206. 53. Bai M, Yang L, Liao H, Liang X, Xie B, Xiong J, Tao X, Chen X, Cheng Y, Chen
31. Chauhan VP, Stylianopoulos T, Martin JD, Popovic Z, Chen O, Kamoun WS, X, et al. Metformin sensitizes endometrial cancer cells to chemotherapy
Bawendi MG, Fukumura D, Jain RK. Normalization of tumour blood vessels through IDH1-induced Nrf2 expression via an epigenetic mechanism.
improves the delivery of nanomedicines in a size-dependent manner. Nat Oncogene. 2018;37:5666–81.
Nanotechnol. 2012;7:383–8. 54. de Oliveira RL, Deschoemaeker S, Henze A-T, Debackere K, Finisguerra V,
32. Huang Y, Yuan J, Righi E, Kamoun WS, Ancukiewicz M, Nezivar J, Takeda Y, Roncal C, Dettori D, Tack E, Jönsson Y. Gene-targeting of Phd2
Santosuosso M, Martin JD, Martin MR, Vianello F. Vascular normalizing doses improves tumor response to chemotherapy and prevents side-toxicity.
of antiangiogenic treatment reprogram the immunosuppressive tumor Cancer Cell. 2012;22:263–77.
microenvironment and enhance immunotherapy. Proc Natl Acad Sci. 2012; 55. Pietras K, Rubin K, Sjöblom T, Buchdunger E, Sjöquist M, Heldin C-H, Östman
109:17561–6. A. Inhibition of PDGF receptor signaling in tumor stroma enhances
33. Carmeliet P, Jain RK. Molecular mechanisms and clinical applications of antitumor effect of chemotherapy. Cancer Res. 2002;62:5476–84.
angiogenesis. Nature. 2011;473:298–307. 56. Davies G, Lobanova L, Dawicki W, Groot G, Gordon JR, Bowen M, Harkness
34. Carmeliet P. Angiogenesis in life, disease and medicine. Nature. 2005;438: T, Arnason T. Metformin inhibits the development, and promotes the
932–6. resensitization, of treatment-resistant breast cancer. PLoS One. 2017;12:
35. Jayson GC, Kerbel R, Ellis LM, Harris AL. Antiangiogenic therapy in oncology: e0187191.
current status and future directions. Lancet. 2016. 57. Yoshida GJ. Therapeutic strategies of drug repositioning targeting
36. Tong RT, Boucher Y, Kozin SV, Winkler F, Hicklin DJ, Jain RK. Vascular autophagy to induce cancer cell death: from pathophysiology to treatment.
normalization by vascular endothelial growth factor receptor 2 blockade J Hematol Oncol. 2017;10:67.
Wang et al. Journal of Experimental & Clinical Cancer Research (2019) 38:235 Page 17 of 17

58. Yoshida GJ. Metabolic reprogramming: the emerging concept and

associated therapeutic strategies. J Exp Clin Cancer Res. 2015;34:111.
59. Shchors K, Massaras A, Hanahan D. Dual targeting of the Autophagic
regulatory circuitry in gliomas with repurposed drugs elicits cell-lethal
autophagy and therapeutic benefit. Cancer Cell. 2015;28:456–71.
60. Barbieri F, Thellung S, Ratto A, Carra E, Marini V, Fucile C, Bajetto A,
Pattarozzi A, Wurth R, Gatti M, et al. In vitro and in vivo antiproliferative
activity of metformin on stem-like cells isolated from spontaneous canine
mammary carcinomas: translational implications for human tumors. BMC
Cancer. 2015;15:228.
61. Wullschleger S, Loewith R, Hall MN. TOR signaling in growth and
metabolism. Cell. 2006;124:471–84.
62. Yoshida GJ, Saya H. EpCAM expression in the prostate cancer makes the
difference in the response to growth factors. Biochem Biophys Res
Commun. 2014;443:239–45.
63. Xu D, Hemler ME. Metabolic activation-related CD147-CD98 complex. Mol
Cell Proteomics. 2005;4:1061–71.
64. Jiang W, Huang Y, An Y, Kim BY. Remodeling tumor vasculature to enhance
delivery of intermediate-sized nanoparticles. ACS Nano. 2015;9:8689–96.