RESEARCH ARTICLE

Palmitoylethanolamide Modulates
Inflammation-Associated Vascular
Endothelial Growth Factor (VEGF) Signaling
via the Akt/mTOR Pathway in a Selective
Peroxisome Proliferator-Activated Receptor
Alpha (PPAR-)-Dependent Manner
a11111

Giovanni Sarnelli1*, Alessandra DAlessandro1, Teresa Iuvone2, Elena Capoccia3,

Stefano Gigli3, Marcella Pesce1, Luisa Seguella3, Nicola Nobile3, Giovanni Aprea1,
Francesco Maione1, Giovanni Domenico de Palma1, Rosario Cuomo1, Luca Steardo3,
Giuseppe Esposito3*
1 Department of Clinical Medicine and Surgery, University of Naples Federico II, Naples, Italy, 2 Department
of Pharmacy, University of Naples Federico II, Naples, Italy, 3 Department of Physiology and Pharmacology
Vittorio Erspamer, La Sapienza University of Rome, Rome, Italy
* sarnelli@unina.it (GS); giuseppe.esposito@uniroma1.it (GE)

OPEN ACCESS
Citation: Sarnelli G, DAlessandro A, Iuvone T,
Capoccia E, Gigli S, Pesce M, et al. (2016)
Palmitoylethanolamide Modulates InflammationAssociated Vascular Endothelial Growth Factor
(VEGF) Signaling via the Akt/mTOR Pathway in a
Selective Peroxisome Proliferator-Activated Receptor
Alpha (PPAR-)-Dependent Manner. PLoS ONE 11
(5): e0156198. doi:10.1371/journal.pone.0156198
Editor: Sujit Basu, Ohio State University, UNITED
STATES
Received: February 23, 2016
Accepted: May 10, 2016
Published: May 24, 2016
Copyright: 2016 Sarnelli et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information files.
Funding: The authors have no support or funding to
report.

Abstract
Background and Aim
Angiogenesis is emerging as a pivotal process in chronic inflammatory pathologies, promoting immune infiltration and prompting carcinogenesis. Ulcerative Colitis (UC) and Crohns
Disease (CD) represent paradigmatic examples of intestinal chronic inflammatory conditions in which the process of neovascularization correlates with the severity and progression of the diseases. Molecules able to target the angiogenesis have thus the potential to
synergistically affect the disease course. Beyond its anti-inflammatory effect, palmitoylethanolamide (PEA) is able to reduce angiogenesis in several chronic inflammatory conditions,
but no data about its anti-angiogenic activity in colitis have been produced, yet.

Methods
The effects of PEA on inflammation-associated angiogenesis in mice with dextran sulphate
sodium (DSS)-induced colitis and in patients with UC were assessed. The release of Vascular Endothelial Growth Factor (VEGF), the hemoglobin tissue content, the expression of
CD31 and of phosphatidylinositol 3-kinase/Akt/mammalian-target-of-rapamycin (mTOR)
signaling axis were all evaluated in the presence of different concentrations of PEA and
concomitant administration of PPAR- and - antagonists.

Competing Interests: The authors have declared

that no competing interests exist.

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

1 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Results
Our results demonstrated that PEA, in a selective peroxisome proliferator activated receptor
(PPAR)- dependent mechanism, inhibits colitis-associated angiogenesis, decreasing
VEGF release and new vessels formation. Furthermore, we demonstrated that the mTOR/
Akt axis regulates, at least partly, the angiogenic process in IBD and that PEA directly
affects this pathway.

Conclusions
Our results suggest that PEA may improve inflammation-driven angiogenesis in colonic
mucosa, thus reducing the mucosal damage and potentially affecting disease progression
and the shift towards the carcinogenesis.

Introduction
Angiogenesis is the process of new vessels development from preexisting vasculature in adult
tissues and it is emerging as pivotal in the pathogenesis and progression of chronic inflammatory pathologies [14].
There is evidence that angiogenesis contributes to a significant dysfunction of vessel architectures, promotes the recruitment of pro-inflammatory cells, and results in a progressive loss
of the epithelial integrity [5,6]. Inflammatory bowel diseases (IBD), such as Crohns disease
(CD) and ulcerative colitis (UC) are paradigmatic examples of chronic inflammatory diseases
in which angiogenesis-related factors affect diseases progression and severity [58].
A variegate class of signaling molecules/cytokines, involved in inflammation and tissue
remodeling processes, co-promotes angiogenesis, such as nitric oxide (NO) or prostaglandins
(PGs), but a prominent role has been identified for Vascular Endothelial Growth Factor A
(VEGF). This mediator, through the activation of a complex signaling network, yields to neovascularization, worsening tissue damage and promoting the carcinogenic drift [910]. In
keeping with this, the inhibition of angiogenetic process may represent a potential therapeutic
target in IBDs, acting on both inflammation and carcinogenic risk [11,12].
Although the release of VEGF is regulated by different molecular pathways, the upstream
activation of phosphatidylinositol 3-kinase/Akt/mammalian target of rapamycin (mTOR) signaling axis (Akt/mTOR pathway) has been recognized as pivotal in VEGF-related neovascularization. Indeed, the activation of this pathway determines also the overexpression of the
Hypoxia-Inducible Factor (HIF)-1, a specific transcriptional factor, which, in turn, further
increases the release of VEGF [12,13]. This complex network, is physiologically induced by
hypoxia in order to guarantee the appropriate tissue oxygenation, stimulating vessels formation, however a pathological over-activation of this pathway has been also described in different inflammatory diseases and several tumors [14,15].
Palmitoylethanolamide (PEA) is an N-acylethanolamide (NAE), structurally and functionally related to anandamide (AEA), with anti-inflammatory and analgesic activities. The antiinflammatory effect of PEA depends on its ability to activate peroxisome proliferator activated
receptor (PPAR)-, a member of nuclear hormone receptor superfamily of ligand activated
transcription factors [16,17]. In both mice and human colitis, PEA has been reported to
decrease the release of several pro-inflammatory cytokines [1820], and there are data suggesting that PEA also exerts a significant anti-angiogenic activity in other chronic inflammatory

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

2 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

conditions [21, 22]. However, we recently demonstrated that PEA is able to directly reduce the
release of pro-angiogenic factors in an in vitro model of colon cancer cells [23]. The potential
anti-angiogenic activity of PEA during colitis has never been reported, yet. With the present
study we aimed to evaluate the ability of PEA to reduce the inflammation-related angiogenesis
in the colon of mice with dextran sulphate sodium (DSS)-induced colitis and in UC patients,
and to characterize its mechanisms of action.

Materials and Methods

Animals and experimental design
Six-weeks-old wild-type (WT) male CD-1 mice (Harlan Laboratories, Udine, Italy) were
used for experiments. All procedures on mice were approved by La Sapienza University's
Ethics Committee. Animal care was in compliance with the IASP and European Community
(EC L358/1 18/12/86) guidelines on the use and protection of animals in experimental
research. Animals were randomly divided into six groups (n = 10 per group): non-colitic
control group; colitic group; colitic group receiving PEA 2 and 10 mg/kg, [24, 25]; colitic
group receiving PEA (10 mg/kg) and selective PPAR- antagonist MK866 (10 mg/kg); colitic group receiving PEA and selective PPAR- antagonist GW9662 (1 mg/kg) [26]. Two
internal control groups (n = 5 per group) were also considered: colitic group receiving
PPAR- or PPAR- antagonist; non-colitic group receiving daily PEA 10 mg/kg. Immunohistochemistry was performed on five groups, excluding colitic group receiving the lowest
dosage of PEA.
Colitis was induced by administrating 4% DSS (MP Biomedicals, Solon, Ohio, USA) in
drinking water for six consecutive days. PEA alone, or combined with PPAR antagonists, was
given by intraperitoneal administration from day 2 to 6 and then animals were sacrificed at day
7 by carbon dioxide inhalation followed by cervical dislocation. Colons were isolated to perform histochemical and biochemical analyses as described below.
In order to further confirm the involvement of PPAR- in mediating the effects of PEA,
additional sets of experiment were conducted in six-weeks-old wild-type PPAR- null (KO)
mice (Taconic, Germantown, New York, USA), that were divided in the following groups: vehicle; colitic; colitic receiving daily PEA 10, 50 or 100 mg/kg, n = 5 for each group, respectively.

Cultured human intestinal biopsies

The experimental group comprised 10 patients with a new diagnose of UC (4 women; age
range 2158 years; endoscopic MAYO score>2) and 5 control subjects (3 men; age range 42
60 years; absence of gastrointestinal symptoms) undergoing colonoscopy for colon cancer
screening. Exclusion criteria were history of cancer, use of 5-aminosalicylic acid, immunosuppressant, anti-platelet and anti-coagulant drugs. Also patients suffering from cardiovascular,
renal or respiratory comorbidities were excluded. All subjects received and signed an informed
consent and the Federico II University Ethical committee approved the protocol.
Four mucosal biopsies from the sigmoid region of UC patients and two biopsies from the
same site of controls were collected and cultured in FBS-supplemented Dulbecco Modified
Eagles Medium (DMEM) at 37C in 5% CO2/95% air. All biopsies were cultured for 24 hours,
with or without PEA at the following concentrations 0,001, 0,01, 0,1 M [27]. Biopsies were
then homogenized and analyzed by western blot as described below. PFA-fixed samples were
used for immunohistochemistry, this analysis was only performed on biopsies stimulated with
the highest dosage of PEA (0,1 M).

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

3 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Protein extraction and western blot analysis

Mice and human specimens were homogenized in ice-cold hypotonic lysis buffer and protein
concentration was determined using Bio-Rad protein assay kit (Bio-Rad, Milan, Italy). Analysis
of total Akt, phosphor-Akt, total mTOR, phosphor-mTOR, total p70S6K, phosphor-p70S6K,
anti-VEGF-R, anti-EGF receptor, anti-HIF1 and -actin protein expression was performed
on total protein fractions of homogenates. Equivalent amounts of homogenates (50 g) underwent electrophoresis through a polyacrilamide minigel. Proteins were then transferred onto
nitrocellulose membranes that were saturated by incubation with 10% nonfat dry milk in
1 PBS overnight at 4C and then incubated, according to the experimental protocols with:
mouse anti-total Akt (1:1000 v/v, Cell signaling technology, Euroclone, Pero, MI, Italy); rabbit
monoclonal anti phosphor-Akt (Ser 473), (1:2000 v/v, Cell signaling technology, Euroclone,
Pero, MI, Italy); rabbit polyclonal anti total mTOR (1:1000, Abcam, Cambridge, UK); rabbit
polyclonal anti phosphor-mTOR (pSer2448) (1:1000 v/v, Cell signaling technology, Euroclone,
Pero, MI, Italy); rabbit polyclonal anti total p70S6K (1:1000 v/v Cell Signaling Technology,
Euroclone, Pero, MI, Italy); and rabbit polyclonal anti-phosphor-p70S6K (Thr421/Ser424,
Thr389); (1:1000 v/v, Cell Signaling Technology, Euroclone, Pero, MI, Italy); rabbit monoclonal anti-VEGF receptor (1:1000 v/v, Cell Signaling Technology, Euroclone, Pero, MI, Italy);
rabbit polyclonal anti-EGF receptor (1:1000 v/v, Abcam, Cambridge, UK); mouse monoclonal
anti-HIF1 (1:500 v/v, Sigma Aldrich, MI, Italy) and mouse anti--actin (1:2000 v/v, Santa
Cruz Biotechnology, Santa Cruz, California, USA). Membranes were then incubated with the
specific secondary antibodies conjugated to horseradish peroxidase (HRP) (Dako, Milan,
Italy). Immune complexes were revealed by enhanced chemiluminescence detection reagents
(Amersham Biosciences, Milan, Italy). Blots were analyzed by scanning densitometry (GS-700
imaging densitometer; Bio-Rad). Results were expressed as OD (arbitrary units; mm2) and normalized on the expression of the housekeeping protein -actin.

Enzyme-linked immunosorbent assay for VEGF and EGF

Enzyme-linked immunosorbent assay (ELISA) for VEGF (Abcam, Cambridge, UK) was carried out on mice and human specimens supernatants according to the manufacturer's protocol.
Absorbance was measured on a microtitre plate reader. In a subset of experiments an ELISA
for EGF (Abcam, Cambridge, UK) was also carried out on plasma of PPAR- null (KO) mice;
VEGF and EGF levels were thus determined using standard curves method.

Immunohistochemistry
Mice and human specimens were fixed in buffered formalin, embedded in paraffin and cut
into 5m-thick serial sections. According to manufacturers instructions, after heat-mediated
antigen retrieval, the tissue was formaldehyde fixed and blocked with serum. The tissue was
incubated with the primary antibody anti CD31 (1:50 v/v, Abcam, Cambridge, UK) for 20 minutes. After three 5-min washes, the secondary antibody was added and the samples were incubated at room temperature for 20 min. The streptavidin-HRP detection system (Chemicon
Int., Temecula, CA, USA) was added and samples were incubated at room temperature. After
three 5-min washes, 50 L of chromogen was added and the reaction terminated after 1 min in
water. Sections were then counterstained with haematoxylin eosin at room temperature. Negative controls were performed by omitting primary antibody. Slides were thus analyzed with a
microscope (Nikon Eclipse 80i by Nikon Instruments Europe), and images were captured at
20X magnification by a high-resolution digital camera (Nikon Digital Sight DS-U1). The
amount of vascularization in each colon section was quantified as a percentage of tissue area

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

4 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

immunopositive for CD31 at a 20X magnification in 6 regions, each amounting to a 0,20- mm2
area, and expressed as vessel density of (%).

Haemoglobin content measurement

As previously demonstrated, hemoglobin content measurement represents an appropriate
method for the detection and quantification of angiogenesis in tissues [28,29]. Mice and
human colonic specimens were collected and weighted, samples were then homogenized in
1 PBS. After centrifugation at 2500 g for 20 min at 4C, the supernatants were further centrifuged at 5000 g for 30 min, and hemoglobin concentration in the supernatant was determined spectrophotometrically at 450 nm by the hemoglobin assay kit (Sigma Aldrich, MI,
Italy). Values were expressed as mg haemoglobin/g of wet weight.

Statistical analysis
Results were expressed as meanSD of n experiments. Statistical analysis was performed using
parametric one-way analysis of variance (ANOVA) and multiple comparisons were performed
by Bonferronis posthoc test; p values <0.05 were considered significant.

Results
Mice DSS-induced colitis is associated with an increase of angiogenesis
that is inhibited by Palmitoylethanolamide
In mice with DSS-induced colitis, bloody diarrhea together with loss of body weight and
increase of spleen size were observed from day 4 until the sacrifice. As expected, the immune
infiltrate and the inflammatory mediators (NO, PGE2 and TNF) were also significantly
increased in treated mice (data not shown).
Hemoglobin tissue content and the expression of CD31, a blood vessel endothelial marker,
were evaluated to detect the effect of PEA on angiogenesis. The hemoglobin content was significantly increased in mice with DSS-induced colitis compared to controls (36,62 vs 11,61,6
mgHb/gr tissue; p<0,0001), but such increase was significantly reduced in mice receiving PEA
(2 and 10 mg/kg) in a dose-dependent fashion (-36% and -60%, respectively; all p<0,001, Fig
1A). Co-administration of the PPAR- antagonist, MK866, but not PPAR- antagonist,
GW966, significantly reverted the effects of PEA on hemoglobin content, likely indicating the
selective involvement of PPAR- (Fig 1A).
A significant higher density of CD31 positive cells was also observed in the inflamed mucosa
of DSS-treated mice compared to controls, and, in line with the above described results, this
was significantly inhibited by PEA in PPAR- dependent mechanism (Fig 1B).
In order to evaluate whether VEGF regulates the inflammatory-related angiogenesis in mice
colitis and if PEA may directly affect this specific pathway, we assessed the release of VEGF
and the expression of its receptor (VEGF-R). As expected, in DSS-treated mice the release of
VEGF was significantly higher than in controls, and this was associated with an increased
expression of VEGF-R (18,76,2 vs 137,514 pg/mL and 1,250,5 vs 5,270,7 OD mm2,
respectively; all p<0,0001; Fig 1C and 1D). Again, the administration of PEA (2 and 10 mg)
significantly inhibited the release of VEGF and the expression of its receptor in a dose-dependent manner (-32 and -72%, -33 and -66% vs DSS-treated mice, respectively; all p<0,001, Fig
1C and 1D), and this effect was significantly affected by MK866, but not by GW9662, coadministration (Fig 1C and 1D).
In Fig 2 it is summarized that PEA failed to induce any significant effect in DSS-treated
PPAR- null mice, further supporting the specific involvement of PPAR-.

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

5 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Fig 1. Palmitoylethanolamide (PEA) inhibits colitis-associated angiogenesis in mice. (A) DSS-induced

colitis caused a significant increase of Hb-content in colonic mucosa, PEA is able to reduce, in a dosedependent fashion, the Hb-content in colitis mice; this effect persisted in presence of PPAR antagonist
(GW9662) while it was nullified by PPAR antagonist (MK866). (B) Immunohistochemical images showing
the expression of CD31 on untreated mice colonic mucosa (panel 1), DSS-treated mice colonic mucosa
(panel 2), DSS-treated mice colonic mucosa in presence of PEA (10 mg/Kg) alone (panel 3), PEA (10 mg/
Kg) plus MK866 10 mg/Kg (panel 4), and PEA (10 mg/Kg) plus GW9662 1 mg/Kg (panel 5). Magnification
20X; scale bar: 100m. The graph summarizes the relative quantification of CD31 expression (%) on mice
colonic mucosa in the same experimental groups, showing the reduction of CD31 expression in colitic mice
after PEA administration, except for the group also treated with the antagonist of PPAR. (C) VEGF release
resulted increase in DSS-treated mice and it was significantly reduced by PEA treatment in a PPAR
dependent manner. (D) Western blot analysis and relative densitometric analysis (arbitrary units normalized
on the expression of housekeeping protein -actin) of VEGF-receptor (VEGF-R) expression, showing similar
results to VEGF release. Results are expressed as meanSD. *p<0.05, **p<0.01 and ***p<0.001 versus
DSS-treated mice.
doi:10.1371/journal.pone.0156198.g001

Palmitoylethanolamide inhibits the angiogenesis in the mucosa of

patients with UC
In order to verify the effect of PEA in the context of human colonic inflammation we performed the same experimental protocol on mucosal biopsies from UC patients. As previously
reported [6,28,29], in the mucosa of UC patients the concentration of hemoglobin was higher
than in controls (436 vs 11,83 mgHb/g tissue p<0,001, Fig 2A) and challenge with PEA
induced a significant reduction of at hemoglobin content, in a dose dependent manner (-33,
-55 and -67%, for 0,001, 0,01 and 0,1 M respectively; all p<0.01, Fig 2A). Similarly to what
observed in the inflamed colon of mice, 0,1 M of PEA were also able to significantly reduce
the number of CD31 positive cells in the mucosa of patients with UC (-50%; p<0,001, Fig 2B).
As compared to control biopsies, the release of VEGF and the expression of its receptor
were also significantly increased in the mucosa from UC patients (191,612 vs 41,68 pg/mL
and 100,8 vs 0,50,1 OD mm2, respectively; all p<0,001, Fig 2C and 2D). Again, the challenge with PEA was demonstrated to induce a significant and dose-dependent reduction of
both VEGF release and VEGFR expression (-37, -53 and -70%, and -30, -50 and -70%, for
0,001, 0,01 and 0,1 M respectively; all p<0,001, Fig 2C and 2D). Similarly to what observed in
the mice all the above described effects of PEA were dependent by PPAR- since co-

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

6 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Fig 2. PEA had no effects on colitis-associated angiogenesis in PPAR null (KO) mice. (A) DSSinduced colitis caused a significant increase of Hb-content in colonic mucosa compared to untreated group,
PEA at different concentrations (10100 mg/kg) did not show any effect on the Hb-content in colitis PPAR
(KO) mice. (B) Immunohistochemical images showing the expression of CD31 on untreated (panel 1), DSStreated (panel 2), and DSS-treated mice in presence of PEA (10 mg/Kg) (panel 3), PEA (50 mg/Kg) (panel 4)
and PEA (100 mg/Kg) (panel 5). Magnification 20X; scale bar: 100m. The graph summarizes the relative
quantification of CD31 expression on mice colonic mucosa in each respective group of mice, and shows the
lack of any significant effect of PEA in reducing CD31 expression in colitic PPAR null mice. In PPAR null
mice, DSS-treatment significantly increased the release of VEGF (C) and EGF (D) and these were
unchanged by PEA administration, regardless of the concentrations used. (E) Western blot analysis and
relative densitometric analysis (arbitrary units normalized on the expression of housekeeping protein -actin)
of VEGF-receptor (VEGF-R) and EGF-receptor (EGF-R) expression, showing similar results to VEGF and
EGF release. Results are expressed as meanSD. ***p<0.001 versus untreated mice.
doi:10.1371/journal.pone.0156198.g002

administration of MK866, but not of GW9662 significantly inhibited its effects even at the
highest dose (Fig 2A2D)

The anti-angiogenic effect of PEA depend upon Akt/mTOR/p70S6K and

HIF-1 pathways downregulation
In order to verify whether PEA has a direct anti-angiogenic effect or this effect was related to
its anti-inflammatory properties, we evaluated the Akt/mTOR/p70S6K pathway, that is directly
involved in the angiogenesis. In the inflamed colon of DSS-treated mice phosphor-Akt, phosphor-mTOR and p70S6K- phosphorylation, were significantly up-regulated, and all were dosedependently and significantly reduced by intraperitoneal administration of PEA (Fig 3A and
3B). The expression of the hypoxia-Inducible Factor (HIF)-1 was also significantly reduced
by -43,3 and -64% after treatment with PEA at 2 and 10 mg/kg, respectively (Fig 3A and 3B).
Similarly to what observed in the mouse colon, the expression of phosphor-Akt, -mTOR,
-p70S6K and HIF-1 were all significantly overexpressed in the mucosa of UC patients (14,7
1,5 vs 1,40,7, 131 vs 1,40,3, 120,8 vs 10,4 and 8,52,7 vs 0,40,1 OD = mm2, respectively vs. control; all p<0,001; Fig 3C and 3D), and they were significantly reduced by PEA in a
dose-dependent fashion (all p<0,001, Fig 3C and 3D).
In both mouse and human specimens the ability of PEA to reduce the overexpression of
Akt/mTOR/p70S6K and HIF-1 pathways was significantly inhibited by concomitant administration of MK866 but not of GW9662, further supporting the concept that its effect involves
the PPAR activation (Fig 4A4D). This finding was also supported by the observation that

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

7 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Fig 3. Effects of palmitoylethanolamide (PEA) on molecular markers of angiogenesis in human. (A)

Ulcerative colitis caused a significant increase of Hb-content in colonic mucosa, PEA is able to reduce, in a
dose-dependent fashion, the Hb-content in colitis patients; this effect persisted in presence of PPAR
antagonist (GW9662) while it was nullified by PPAR antagonist (MK866). (B) Immunohistochemical images
showing the expression of CD31 on: colonic mucosa of controls (panel 1), UC patients colonic mucosa
(panel 2), UC patients colonic mucosa in presence of PEA (0,1 M) alone (panel 3), PEA (0,1 M) plus
MK866 3 M (panel 4), and PEA (0,1 M) plus GW9662 (9nM) (panel 5). Magnification 20X; scale bar:
100m. The graph summarizes the relative quantification of CD31 expression (%) on human colonic mucosa
in the same experimental groups, and, as described in mice, PEA administration caused a significant
reduction of CD31 expression, except after co-administration of PPAR antagonist. (C) VEGF release
resulted increase in uncreative colitis patients and it was significantly reduced by PEA treatment in a PPAR
dependent manner D) Western blot analysis and relative densitometric analysis (arbitrary units normalized on
the expression of housekeeping protein -actin) of VEGF-receptor (VEGF-R) expression, showing a similar
behavior to VEGF release. Results are expressed as meanSD. *p<0.05, **p<0.01 and ***p<0.001 versus
untreated biopsies from ulcerative colitis patients.
doi:10.1371/journal.pone.0156198.g003

PEA failed to significantly affect the Akt/mTOR/p70S6K and HIF-1 pathways in PPAR null
mice (Fig 5).

Discussion
Current therapies for inflammatory bowel disease are still challenging because of the relatively
high rate of failure, the high costs and the risk of severe side effects due to immunosuppressive
agents [30,31]. New molecules able to target different steps and pathways of the inflammatory
response appear therefore a promising strategy for the treatment of these diseases.
Angiogenesis has been recently identified as a key event in the context of intestinal inflammation, whose extent significantly correlates with both the severity and the progression of the
diseases [58], and favors the drift toward colonic carcinogenesis [32]; inhibition of angiogenesis appears thus as a synergistic and promising therapeutic strategy in IBDs [33,34].
VEGF is implicated in the regulation of the angiogenic process in sustained inflammation,
and contributes to mucosal tissue remodeling, vascular permeability and leukocyte infiltration
of the inflamed mucosa [3537]. Here we demonstrated that, both in vivo and in ex vivo, PEA,
in a concentration dependent manner, significantly reduced VEGF release and the expression
of its receptor, in mice and human inflamed colon, respectively; this result was also associated
with a significant decrease of mucosal hemoglobin content and CD31 positive vessels density.

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

8 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

Fig 4. Effects of palmitoylethanolamide (PEA) on Akt/mTOR/p70S6K axis activation and HIF-1

expression in DSS-induced colitis and in ulcerative colitis. (A) Western Blot analysis and (B) relative
densitometric analysis showing the effects of PEA (2 mg/Kg and 10 mg/Kg), given alone or in the presence of
MK866 (10 mg/Kg) or GW 9662 (1 mg/Kg), on the expression of phosphor-Akt, phosphor-mTOR, phosphorp70S6K and HIF-1 in mice with DSS-induced colitis. Western blot analysis (C) and relative densitometric
analysis showing the effects of PEA at increasing concentration (0.001 M, 0.01 M and 0.1 M) given alone
and in the presence of MK866 (3 M) or GW9662 (9 nM), on the expression of phosphor-Akt, phosphormTOR, phosphor-p70S6K and HIF-1 in UC patients biopsies. In both human and mice, colitis induced the
activation of pro-angiogenic Akt/mTOR/p70S6K pathway, and PEA resulted able to reduce it, in a dosedependent and PPAR dependent fashion. Results are expressed as mean SD. *p<0.05, **p<0.01 and
***p<0.001 versus DSS-treated mice (A and B) or vs untreated colonic biopsies from UC patients (C and D).
doi:10.1371/journal.pone.0156198.g004

Fig 5. Palmitoylethanolamide (PEA) failed to significantly affect DSS-induced Akt/mTOR/p70S6K axis

activation and HIF-1 expression in PPAR null (KO) mice. (A) Western Blot analysis and (B) relative
densitometric analysis showing the effects of PEA (10100 mg/kg) on the expression of phosphor-Akt,
phosphor-mTOR, phosphor-p70S6K and HIF-1 in DSS-treated PPAR null (KO) mice. DSS exposure
significantly induced the activation of the pro-angiogenetic Akt/mTOR/p70S6K pathway as compared to
untreated mice, but treatment with PEA had no significant effect in treated mice, regardless of the
concentration used. Results are expressed as mean SD. ***p<0.001 versus untreated mice.
doi:10.1371/journal.pone.0156198.g005

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

9 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

It has been described that PEA is able to exert an antiangiogenic activity in other experimental models of chronic inflammation, likely supporting the concept that this ability is dependent
by its anti-inflammatory effects [21,22,38]. However, we recently demonstrated that PEA
directly reduces the release of pro-angiogenic factors in an in vitro model of colon cancer cells,
through an Akt-mTOR pathway-dependent VEGF inhibition; these results suggest that the
anti-angiogenic activity displayed by PEA in the inflamed colon is not solely related by its antiinflammatory effects [23].
In our setting, we therefore investigated whether PEA-dependent VEGF signaling inhibition
modulated by the AkT/mTOR axis. Different molecular pathways are involved in the angiogenesis, but the activation of AkT/mTOR axis has been specifically related to neo-vascularization in
the development of inflammation-sustained colon cancer [3941]. In particular, it induces the
over-expression of HIF-1, a transcriptional factor related to hypoxia, that cooperates with reactive oxygen species (ROS), stimulating the release of VEGF and eventually neo-angiogenesis
[4244]. Our results demonstrated that PEA, in a PPAR- selective and concentration dependent-manner, significantly reduced the phosphorylation of Akt, mTOR and p70S6 proteins in
mice colon and ex-vivo human mucosa, leading to downstream inhibition of HIF-1 with consequent inhibition of VEGF and EGF secretion and the respective receptors expression.
Remarkably, here we also showed that all the above-described pleiotropic effects of PEA are
specifically related to the activation of the PPAR- pathway. Although the role of PPAR- as
putative site of action of anti-inflammatory and anti-cancer drugs has been specifically
addressed [45,46], the importance of PPAR- pathway is recently emerging [16,23,27]. In
keeping with this, we have demonstrated that the inhibition of the mTOR/AkT axis depends
on PPAR- activation, supporting its contribution in IBD-related angiogenesis and suggesting
its protective role in inflammation-associated carcinogenesis. In addition, we provide data suggesting that PEA is able to act on the process of angiogenesis by directly modulating the endothelial cells functioning, as demonstrated by its effect to significantly inhibit inflammatoryassociated proliferation and migration of HUVEC cells (S1 Fig).
As stated, anti-angiogenetic drugs represent an intriguing approach to treat IBDs, due to the
effect on both inflammation and tumorigenesis. However, the efficacy of anti-angiogenic drugs
is limited by the complexity and redundancy of the molecular pathways converging in neovascularization. In this context, PEA appears as a very interesting compound, since together with
its activity on the AkT/mTOR pathway, it also significantly reduces the p38/MAPK/NF-kB
axis [4749]. It has been indeed previously demonstrated that PEA is able to inhibit the NF-kB
(nuclear factor kappa-light-chain-enhancer of activated B cells) pro-inflammatory pathway,
determining a strong downregulation of cyclooxygenase (COX)-2 and inducible nitric oxide
synthase (iNOS) expression, with a consequent reduction of prostaglandins and nitric oxide
release [48, 49]. While the role of these mediators in the inflammatory process is well established, there is evidence about their involvement in neo-vascularization and tumor growth,
supporting the inflammation-associated carcinogenesis assumption [50, 51]. Interestingly,
besides such anti-angiogenic and anti-inflammatory activity, PEA, as the others cannabinomimetic fatty acid derivatives, also exerts an antiproliferative effect on cancer cells, supporting its
protective effects in both inflammation and cancer prevention [23, 52, 53].
To date, mesalamine is the unique drug, widely used in IBD, with both anti-inflammatory
and potential anti-carcinogenic proprieties [54, 55]. However, even if rare, severe side effects to
this compound, such as pancreatitis and interstitial nephritis, have been described [56]. PEA is
a safe drug with a well known toxicological profile and it is already available as orally administered supplement in clinical practice [57, 58].
Although further studies are needed, palmitoylethanolamide, due to its anti-inflammatory
and anti-angiogenic effects, might represent a promising food therapy for the prevention of

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

10 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

inflammation-associated colon cancer, and, most importantly, be part of a combined and

multi-target therapy in the management of inflammation-associated angiogenesis and the
potential anti-carcinogenic activity.

Supporting Information
S1 Fig. Palmitoylethanolamide reduced migration and proliferation in DSS-treated
HUVEC cells.
(DOCX)

Author Contributions
Conceived and designed the experiments: GS GE. Performed the experiments: TI EC SG L.
Seguella NN. Analyzed the data: AD MP FM. Contributed reagents/materials/analysis tools:
RC GA GdP L. Steardo. Wrote the paper: AD MP.

References
1.

Paleolog EM. Angiogenesis in rheumatoid arthritis. Arthritis Res. 2002; 4:8190.

2.

Sluimer JC, Daemen MJ. Novel concepts in atherogenesis: angiogenesis and hypoxia in atherosclerosis. J Pathol. 2009; 218:729. doi: 10.1002/path.2518 PMID: 19309025

3.

Crawford TN, Alfaro DV 3rd, Kerrison JB, Jablon EP. Diabetic retinopathy and angiogenesis. Curr Diabetes Rev. United Arab Emirates; 2009 Feb; 5(1):813.

4.

Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature. 2000; 407:24957. PMID:
11001068

5.

Tas SW, Maracle CX, Balogh E, Szekanecz Z. Targeting of proangiogenic signalling pathways in
chronic inflammation. Nat Rev Rheumatol. 2016 Feb; 12(2):11122. doi: 10.1038/nrrheum.2015.164
PMID: 26633288

6.

Jackson JR, Seed MP, Kircher CH, Willoughby DA, Winkler JD. The codependence inflammation of
angiogenesis and chronic. FASEB J. 1997; 11:45765. PMID: 9194526

7.

Alkim C, Alkim H, Koksal AR, Boga S, Sen I. Angiogenesis in Inflammatory Bowel Disease. Int J Inflam.
2015; 2015:970890 doi: 10.1155/2015/970890 PMID: 26839731

8.

Danese S, Sans M, de la Motte C, Graziani C, West G, Phillips MH, et al. Angiogenesis as a novel component of inflammatory bowel disease pathogenesis. Gastroenterology. AGA Institute American
Gastroenterological Association; 2006 Jun; 130(7):206073.

9.

Olsson A-K, Dimberg A, Kreuger J, Claesson-Welsh L. VEGF receptor signallingin control of vascular
function. Nat Rev Mol Cell Biol. 2006 May; 7(5):35971. PMID: 16633338

10.

Ferrara N. Role of vascular endothelial growth factor in the regulation of angiogenesis. Kidney Int.
1999; 56:794814. PMID: 10469350

11.

Davaatseren M, Hwang J-T, Park JH, Kim M-S, Wang S, Sung MJ. Allyl isothiocyanate ameliorates
angiogenesis and inflammation in dextran sulfate sodium-induced acute colitis. PLoS One. 2014 Jan;
9(7):e102975. doi: 10.1371/journal.pone.0102975 PMID: 25051185

12.

Son MK, Jung KH, Hong S-W, Lee H-S, Zheng H-M, Choi M-J, et al. SB365, Pulsatilla saponin D suppresses the proliferation of human colon cancer cells and induces apoptosis by modulating the AKT/
mTOR signalling pathway. Food Chem. Elsevier Ltd; 2013 Jan; 136(1):2633.

13.

Karar J, Maity A. PI3K/AKT/mTOR Pathway in Angiogenesis. Front Mol Neurosci. 2011 Jan; 4(51):18.

14.

Clottes E. Hypoxia-inducible factor 1: regulation, involvement in carcinogenesis and target for anticancer therapy. Bull Cancer. 2005 Feb; 92(2):11927. PMID: 15749641

15.

Hadjipanayi E, Schilling AF. Hypoxia-based strategies for angiogenic induction: the dawn of a new era
for ischemia therapy and tissue regeneration. Organogenesis. 2013 Oct 1; 9(4):26172. doi: 10.4161/
org.25970 PMID: 23974216

16.

Lo Verme J, Fu J, Astarita G, La Rana G, Russo R, Calignano A, et al. The Nuclear Receptor Peroxisome Proliferator-Activated Receptor-alpha Mediates the Anti-Inflammatory Actions of Palmitoylethanolamide. Mol Pharmacol. 2005; 67(1):159. PMID: 15465922

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

11 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

17.

Conti S, Costa B, Colleoni M, Parolaro D, Giagnoni G. Antiinflammatory action of endocannabinoid palmitoylethanolamide and the synthetic cannabinoid nabilone in a model of acute inflammation in the rat.
Br J Pharmacol. 2002; 135:1817. PMID: 11786493

18.

Costa B, Conti S, Giagnoni G, Colleoni M. Therapeutic effect of the endogenous fatty acid amide, palmitoylethanolamide, in rat acute inflammation: inhibition of nitric oxide and cyclo-oxygenase systems. Br J
Pharmacol. 2002 Oct; 137(4):41320. PMID: 12359622

19.

Ross R a., Brockie HC, Pertwee RG. Inhibition of nitric oxide production in RAW264.7 macrophages by
cannabinoids and palmitoylethanolamide. Eur J Pharmacol 2000 Aug; 401(2):12130. PMID:
10924916

20.

Farquhar-smith WP, Jaggar SI, Rice ASC. Attenuation of nerve growth factor-induced visceral hyperalgesia via cannabinoid CB 1 and CB 2 -like receptors. Pain. 2002; 97:1121. PMID: 12031775

21.

De Filippis D, DAmico A, Cipriano M, Petrosino S, Orlando P, Di Marzo V, et al. Levels of endocannabinoids and palmitoylethanolamide and their pharmacological manipulation in chronic granulomatous
inflammation in rats. Pharmacol Res. Elsevier Ltd; 2010 Apr; 61(4):3218.

22.

Cipriano M, Esposito G, Negro L, Capoccia E, Sarnelli G, Scuderi C, et al. Palmitoylethanolamide Regulates Production of Pro-Angiogenic Mediators in a Model of beta Amyloid-Induced Astrogliosis In
Vitro. CNS Neurol Disord Drug Targets. United Arab Emirates; 2015; 14(7):82837.

23.

Sarnelli G, Gigli S, Capoccia E, Iuvone T, Cirillo C, Seguella L et al. Antiproliferative and antiangiogenic
effects of palmitoylethanolamide in Caco-2 human colon cancer cell involve a selective PPAR-alpha
dependent inhibition of Akt/mTOR pathway. Phytotherapy Research; 2016 2016 Mar 1. [Epub ahead of
print]

24.

Capasso R, Izzo AA, Fezza F, Pinto A, Capasso F, Mascolo N, et al. Inhibitory effect of palmitoylethanolamide on gastrointestinal motility in mice. Br J Pharmacol. 2001; 134:94550. PMID: 11682441

25.

Costa B, Comelli F, Bettoni I, Colleoni M, Giagnoni G. The endogenous fatty acid amide, palmitoylethanolamide, has anti-allodynic and anti-hyperalgesic effects in a murine model of neuropathic pain:
involvement of CB(1), TRPV1 and PPARgamma receptors and neurotrophic factors. Pain. International
Association for the Study of Pain; 2008 Oct 3; 139(3):54150.

26.

Brown WH, Gillum MP, Lee H, Paulo J, Camporez G, Zhang X. Fatty acid amide hydrolase ablation promotes ectopic lipid storage and insulin resistance due to centrally mediated hypothyroidism. Proc Natl
Acad Sci USA. 2012; 109:1496671. doi: 10.1073/pnas.1212887109 PMID: 22912404

27.

Esposito G, Capoccia E, Turco F, Palumbo I, Lu J, Steardo A, et al. Palmitoylethanolamide improves

colon inflammation through an enteric glia/toll like receptor 4-dependent PPAR- activation. Gut. 2014
Aug; 63(8):130012. doi: 10.1136/gutjnl-2013-305005 PMID: 24082036

28.

Tsujii M, Kawano S, Tsuji S, Kobayashi I, Takei Y, Nagano K, et al. Colonic mucosal hemodynamics
and tissue oxygenation in patients with ulcerative colitis: investigation by organ reflectance spectrophotometry. Gastroenterol. 1995 Apr; 30(2):1838.

29.

Bhaskar L, Ramakrishna BS, Balasubramanian KA. Colonic mucosal antioxidant enzymes and lipid
peroxide levels in normal subjects and patients with ulcerative colitis. J Gastroenterol Hepatol. 1995
Mar-Apr; 10(2):1403. PMID: 7787158

30.

Grevenitis P, Thomas A, Lodhia N. Medical Therapy for Inflammatory Bowel Disease. Surg Clin North
Am. 2015 Dec; 95(6):115982. doi: 10.1016/j.suc.2015.08.004 PMID: 26596920

31.

Martnez-Montiel MP, Casis-Herce B, Gmez-Gmez GJ, Masedo-Gonzlez A,Yela-San Bernardino

C, Piedracoba C, et al. Pharmacologic therapy for inflammatory bowel disease refractory to steroids.
Clin Exp Gastroenterol. 2015 Aug 17; 8:25769. doi: 10.2147/CEG.S58152 PMID: 26316792

32.

Setia S, Nehru B, Sanyal SN. The PI3K/Akt pathway in colitis associated colon cancer and its chemoprevention with celecoxib, a Cox-2 selective inhibitor. Biomed Pharmacother. Elsevier Masson SAS;
2014 Jul; 68(6):7217.

33.

Hatoum O a, Heidemann J, Binion DG. The intestinal microvasculature as a therapeutic target in inflammatory bowel disease. Ann N Y Acad Sci 2006 Aug; 1072:7897. PMID: 17057192

34.

Chidlow JH, Langston W, Greer JJM, Ostanin D, Abdelbaqi M, Houghton J, et al. Differential angiogenic
regulation of experimental colitis. Am J Pathol [Internet]. American Society for Investigative Pathology;
2006 Dec; 169(6):201430.

35.

Ferrante M, Pierik M, Henckaerts L, Joossens M, Claes K, Van Schuerbeek N, et al. The role of vascular endothelial growth factor (VEGF) in inflammatory bowel disease. Inflamm Bowel Dis. 2006 Sep; 12
(9):8708. PMID: 16954806

36.

Goebel S, Huang M, Davis WC, Jennings M, Siahaan TJ, Alexander JS, et al. VEGF-A stimulation of
leukocyte adhesion to colonic microvascular endothelium: implications for inflammatory bowel disease.
Am J Physiol Gastrointest Liver Physiol. 2006 Apr [cited 2015 Sep 25]; 290(4):G64854. PMID:
16293653

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

12 / 13

Antiangiogenic Effect of Palmitoylethanolamide in Colitis

37.

Scaldaferri F, Vetrano S, Sans M, Arena V, Straface G, Stigliano E, et al. VEGF-A links angiogenesis
and inflammation in inflammatory bowel disease pathogenesis. Gastroenterology. AGA Institute American Gastroenterological Association; 2009 Feb; 136(2):58595.e5.

38.

De Filippis D, Russo A, De Stefano D, Maiuri MC, Esposito G, Cinelli MP et al. (2007). Local administration of WIN 55,2122 reduces chronic granuloma-associated angiogenesis in rat by inhibiting NFkappa B activation. J Mol Med 85: 635645 PMID: 17447045

39.

Hay N. The Akt-mTOR tango and its relevance to cancer. Cancer Cell. 2005 Sep; 8(3):17983. PMID:
16169463

40.

Morgensztern D, McLeod HL. PI3K/Akt/mTOR pathway as a target for cancer therapy. Anticancer
Drugs. 2005 Sep; 16(8):797803. PMID: 16096426

41.

Shaw RJ, Cantley LC. Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature. 2006 May
25; 441(7092):42430. PMID: 16724053

42.

Yuan G, Nanduri J, Khan S, Semenza GL, Prabhakar NR. Induction of HIF-1alpha expression by intermittent hypoxia: involvement of NADPH oxidase, Ca2+ signaling, prolyl hydroxylases, and mTOR. J
Cell Physiol [Internet]. 2008 Dec; 217(3):67485.

43.

Laughner E, Taghavi P, Chiles K, Mahon PC, Semenza GL. HER2 (neu) signaling increases the rate of
hypoxia-inducible factor 1alpha (HIF-1alpha) synthesis: novel mechanism for HIF-1-mediated vascular
endothelial growth factor expression. Mol Cell Biol. 2001 Jun; 21(12):39954004. PMID: 11359907

44.

Glover LE, Colgan SP. Hypoxia and metabolic factors that influence IBD pathogenesis. Gastroenterology. 2011; 140(6):174855. doi: 10.1053/j.gastro.2011.01.056 PMID: 21530741

45.

Han S, Roman J. Peroxisome proliferator-activated receptor gamma: a novel target for cancer therapeutics? Anticancer Drugs. 2007 Mar; 18(3):23744. PMID: 17264754

46.

Dubuquoy L, Rousseaux C, Thuru X, Peyrin-Biroulet L, Romano O, Chavatte P, et al. PPARgamma as

a new therapeutic target in inflammatory bowel diseases. Gut. 2006 Sep; 55(9):13419. PMID:
16905700

47.

Dormond-Meuwly A, Roulin D, Dufour M, Benoit M, Demartines N, Dormond O. The inhibition of MAPK

potentiates the anti-angiogenic efficacy of mTOR inhibitors. Biochem Biophys Res Commun. Elsevier
Inc.; 2011 Apr 22; 407(4):7149.

48.

Scuderi C, Esposito G, Blasio A, Valenza M, Arietti P, Steardo L, et al. Palmitoylethanolamide counteracts reactive astrogliosis induced by -amyloid peptide. J Cell Mol Med [Internet]. 2011 Dec; 15
(12):266474.

49.

Gee E, Milkiewicz M, Haas TL. p38 MAPK activity is stimulated by vascular endothelial growth factor
receptor 2 activation and is essential for shear stress-induced angiogenesis. J Cell Physiol. 2010 Jan;
222(1):1206. doi: 10.1002/jcp.21924 PMID: 19774558

50.

Xu W, Liu LZ, Loizidou M, Ahmed M, Charles IG. The role of nitric oxide in cancer. Cell Res. 2002 Dec;
12(56):31120. PMID: 12528889

51.

Montrose DC, Nakanishi M, Murphy RC, Zarini S, McAleer JP, Vella AT, et al. The role of PGE2 in intestinal inflammation and tumorigenesis. Prostaglandins Other Lipid Mediat. 2015 Jan-Mar; 116117:26
36 doi: 10.1016/j.prostaglandins.2014.10.002 PMID: 25460828

52.

De Petrocellis L, Bisogno T, Ligresti A, Bifulco M, Melck D, Di Marzo V. Effect on cancer cell proliferation of palmitoylethanolamide, a fatty acid amide interacting with both the cannabinoid and vanilloid signalling systems. Fundam Clin Pharmacol. 2002 Aug; 16(4):297302. PMID: 12570018

53.

Di Marzo V, Melck D, Orlando P, Bisogno T, Zagoory O, Bifulco M et al. Palmitoylethanolamide inhibits

the expression of fatty acid amide hydrolase and enhances the anti-proliferative effect of anandamide
in human breast cancer cells. Biochem J. 2001 Aug 15; 358(Pt 1):24955. PMID: 11485574

54.

Tang J, Sharif O, Pai C. Mesalamine protects against colorectal cancer in inflammatory bowel disease.
Digestive and Liver disease 2010; 55(6):1696

55.

Rubin DT, LoSavio A, Yadron N, Huo D, Hanauer SB. Aminosalicylate therapy in the prevention of dysplasia and colorectal cancer in ulcerative colitis

56.

Cunliffe RN, Scott BB. Monitoring for drug side-effects in inflammatory bowel disease APT 2002 66
(4):647662

57.

Paladini A, Fusco M, Cenacchi T, Schievano C, Piroli A, Varrassi G. Palmitoylethanolamide, a Special

Food for Medical Purposes, in the Treatment of Chronic Pain: A Pooled Data Meta-analysis. Pain Physician 2016; 19:1124 PMID: 26815246

58.

Orefice NS, Alhouayek M, Carotenuto A, Montella S, Barbato F, Comelli A,et al. Oral Palmitoylethanolamide Treatment Is Associated with Reduced Cutaneous Adverse Effects of Interferon-1a and Circulating Proinflammatory Cytokines in Relapsing-Remitting Multiple Sclerosis. Neurotherapeutics. 2016
Apr; 13(2):42838. doi: 10.1007/s13311-016-0420-z PMID: 26857391

PLOS ONE | DOI:10.1371/journal.pone.0156198 May 24, 2016

13 / 13