First posted online on 2 February 2017 as 10.1242/dmm ......2017/02/01 · A phenotype-based...
Transcript of First posted online on 2 February 2017 as 10.1242/dmm ......2017/02/01 · A phenotype-based...
Miconazole Protects Blood Vessels from Matrix Metalloproteinase 9-Dependent Rupture
and Hemorrhage
Running Title: Miconazole Inhibits Hemorrhage
Ran Yang1,2, Yunpei Zhang3, Dandan Huang3, Xiao Luo2, Liangren Zhang2, Xiaojun Zhu1,2,
Xiaolin Zhang4, Zhenming Liu2,*, Jingyan Han3,*, and Jing-Wei Xiong1,2,*
1Beijing Key Laboratory of Cardiometabolic Molecular Medicine, Institute of Molecular
Medicine; 2State Key Laboratory of Natural and Biomimetic Drugs, School of Pharmaceutical
Sciences, and 3School of Basic Medical Sciences, Peking University, Beijing 100871, China; 4AstraZeneca Asia and Emerging Market Innovative Medicine and Early Development, Shanghai
201203, China
*To whom correspondence may be addressed:
Jing-Wei Xiong, Ph.D.
Institute of Molecular Medicine
Peking University, Beijing 100871, China
Tel.: 86-10-62766239
Email: [email protected]
or
Jingyan Han, Ph.D.
Email: [email protected]
or
Zhenming Liu, Ph.D.
Email: [email protected]
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
© 2017. Published by The Company of Biologists Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproductionin any medium provided that the original work is properly attributed.
http://dmm.biologists.org/lookup/doi/10.1242/dmm.027268Access the most recent version at DMM Advance Online Articles. Posted 2 February 2017 as doi: 10.1242/dmm.027268http://dmm.biologists.org/lookup/doi/10.1242/dmm.027268Access the most recent version at
First posted online on 2 February 2017 as 10.1242/dmm.027268
Summary Statement
A phenotype-based chemical screen in zebrafish identifies miconazole as a novel hemorrhagic
suppressor. Miconazole inhibits vessel rupture and hemorrhages by decreasing pErk and matrix
metalloproteinase 9 in zebrafish and rats.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Abstract
Hemorrhagic stroke accounts for 10-15% of all strokes and is strongly associated with mortality
and morbidity worldwide, but its prevention and therapeutic interventions remain a major
challenge. Here, we report the identification of miconazole as a hemorrhagic suppressor by a
small-molecule screen in zebrafish. We found that a hypomorphic mutant fn40a, one of known -
pix mutant alleles in zebrafish, had the major symptoms of brain hemorrhage, vessel rupture, and
inflammation as those in hemorrhagic stroke patients. A small-molecule screen with mutant
embryos identified anti-fungal drug miconazole as a potent hemorrhagic suppressor. Miconazole
inhibited both brain hemorrhages in zebrafish and mesenteric hemorrhages in rats by decreasing
matrix metalloproteinase 9 (mmp9)-dependent vessel rupture. Mechanistically, miconazole
down-regulated the levels of pErk and Mmp9 to protect vascular integrity in fn40a mutants.
Therefore, our findings have demonstrated that miconazole protects blood vessels from
hemorrhages by down-regulating the pERK-MMP9 axis from zebrafish to mammals and have
shed light on the potential of phenotype-based screens in zebrafish for the discovery of new drug
candidates and chemical probes for hemorrhagic stroke.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Introduction
Cardiovascular diseases are the leading causes of mortality worldwide, and stroke ranks
among the top three. About 6.7 million people died of stroke, accounting for ~17% of
total mortality from non-communicable diseases globally in 2012(WHO, 2015).
Hemorrhagic stroke accounts for ~10-15% of all strokes and is strongly associated with
mortality and morbidity. Although the morbidity and mortality resulting from
hemorrhagic stroke are high, we know less about its molecular pathology than we have
learned from ischemic stroke. Current studies suggest that the inflammatory response is a
major cellular event after hemorrhagic stroke; it accompanies with neuronal death,
leukocyte infiltration, and the activation of microglia/macrophages and astrocytes. Major
inflammatory mediators include matrix metalloproteinases (MMPs), nuclear factor
erythroid 2-related factor 2 (Nrf2), heme oxygenase (HO), and ferric irons(Wang, 2010).
Various anti-inflammatory strategies in hemorrhagic stroke are explored in both
preclinical and clinical trials, such as microglia/macrophage inhibitory factor, MMP
inhibitors, the Nrf2 inducer sulforaphane, HO inhibitors, and deferoxamine for iron-
mediated toxicity(Egashira et al., 2015; Wang, 2010). However, none of the above target-
based treatments has been translated into the clinic so far.
The current failure to develop medical treatments, which primarily rely on target-based
treatments, suggests that the essential therapeutic targets for this disease are not yet
discovered, or alternatively, multiple genetic factors in parallel are involved in this
complex disease. At the same time, phenotype-based drug discovery is gradually being
recognized and actively pursued by both academic and industrial scientists using model
organisms such as zebrafish(MacRae and Peterson, 2015). The zebrafish is a powerful
model organism for both genetic and chemical screens, so establishing a suitable
zebrafish model for hemorrhagic stroke can provide an alternative method in searching
for medical treatments on this devastating disease.
Here, we present an ethylnitrosourea (ENU)-induced mutant fn40a, one of known-pix
mutant alleles(Chen et al., 2001; Liu et al., 2007), as a hemorrhagic stroke model for
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
chemical suppressor screens in zebrafish. By performing a small-molecule screen, we
found that miconazole, a known anti-fungal drug, is a potent hemorrhagic suppressor.
Further molecular and cellular analyses suggested that miconazole inhibited both brain
hemorrhages in zebrafish and mesenteric hemorrhages in rats by decreasing MMP9-
dependent vessel rupture. Therefore, our findings demonstrate the great potential of
phenotype-based screens in zebrafish for the discovery of new drug candidates and
chemical probes for hemorrhagic stroke.
Results
The fn40a mutant is a suitable hemorrhagic stroke model for chemical suppressor screens
in zebrafish
fn40a was identified as a hemorrhagic mutant from an ENU-induced mutagenesis screen of the
zebrafish genome at Massachusetts General Hospital, Boston(Chen et al., 2001; Liu et al., 2007).
This mutant failed to complement with bubbleheadm292, a β-pix mutant, suggesting that fn40a is
allelic to m292. However, no mutations were found in the coding region and splicing sites of β-
pix, while β-pix mRNA decreased dramatically in homozygotic fn40a mutants(Liu et al., 2007),
suggesting that mutations might occur in the promoter or enhancer region. While m292 mutants
are embryonic-lethal, we found that fn40a mutants are able to recover and survive to adulthood,
consistent with that fn40a is a hypomorphic allele to -pix(Liu et al., 2007). This particular
feature of viable homozygous fn40a adults, from which 100% mutant embryos can be obtained,
was then exploited for a chemical suppressor screen as previously described for gridlock mutants
in zebrafish(Peterson et al., 2004). Briefly, we inbred and collected the fn40a mutant embryos,
and raised those with severe brain hemorrhage to the next generation. After two generations, the
hemorrhage phenotype was severe and stable and the hemorrhage rate reached almost 100%
(n>1000). Hemorrhagic site occurred most frequently in the hindbrain and occasionally in the
midbrain and forebrain of the mutants at 2 days post-fertilization (dpf) (Fig. 1C-D; n>100),
which is similar to another -pix allele m292(Liu et al., 2007; Liu et al., 2012) and redhead
mutant pak2a (mi149)(Buchner et al., 2007). Since the hemorrhages of homozygous mutants
dissolved gradually and disappeared around ~4 dpf, they could survive to adulthood with normal
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
body growth and fertility. By using flk1:eGFP and gata1:DsRed double transgenic zebrafish, we
found that gata1:DsRed-labeled erythrocytes flowed completely inside well-patterned brain
vessels labeled by flk1:eGFP in wild-type sibling embryos (Fig. 1E-F, I; Movie S1). In contrast,
erythrocytes accumulated in the intracerebral region of fn40a mutants (Fig. 1H, K), accompanied
by disruption of the central arteries (CtA) between the basilar artery (BA) and the primordial
hindbrain channel (PHBC) in the hindbrain (Fig. 1G, K; Movie S2). The recruitment of
coro1a:eGFP-labeled inflammatory cells to the hemorrhagic region in mutants (Fig. 1L)
compared with the evenly distributed inflammatory cells in wild-type siblings (Fig. 1J). In
addition, it has been reported that β-pix interacts with rap1b and ccm1, in which ccm1 is one of
the mutated genes causing cerebral cavernous malformations(Gore et al., 2008), thus establishing
a potential link between this mutant and intracranial hemorrhages in humans. These data suggest
that, like intracerebral hemorrhagic models in rodents(Wang, 2010), the fn40a mutant mimics the
phenotypes of hemorrhagic stroke such as brain-vessel rupture, intracerebral hemorrhage, and
inflammation; and the transient brain hemorrhages of the mutants can be exploited for a chemical
suppressor screen in zebrafish.
Using an in-house chemical library with diversified chemical structures of 923 drug-like
compounds, we performed a hemorrhagic suppressor screen. The outline of this screen includes
three major steps (Fig. 1M). During the first step (primary screen), 5 fn40a mutant embryos were
arrayed in each well of 96-well plates, and were then subjected to administration of 10 µmol/L of
small molecules from 6 to 24 hours post-fertilization (hpf). At 24 hpf, the compounds were
washed away, replaced, and incubated with fresh egg water until the desired stages. Each
compound repeated in three wells. The hemorrhagic suppression efficiency was scored based on
the hemorrhage rate for each compound. The second step was to validate the efficacy of
candidate compounds from the primary screen by using an average of 200 mutant embryos. In
the final step of validation, we tested the dose-dependent efficacy and toxicity of each candidate
compound with a series of concentrations and more than 70 mutant embryos per condition.
Through this screen, we isolated 7 candidate hemorrhagic suppressors, of which their chemical
structures were shown (Fig. S1) and their hemorrhagic suppression efficacy was shown by
hemorrhage rate (Table S1). Miconazole nitrate (the nitrate form of miconazole) stood out of the
seven candidates because of its great efficacy on hemorrhagic suppression and least toxicity to
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
embryos. Importantly, miconazole effectively suppressed brain hemorrhages in a dose-dependent
manner (Fig. S2N). By applying 5 μmol/L miconazole to fn40a mutant embryos from 6 to 24
hpf, we found that it suppressed as high as 70% of brain hemorrhage compared with that in
DMSO-treated embryos (Fig. S2N). In addition, application of miconazole from 6 to 24 hpf had
more suppression on brain hemorrhage in fn40a mutants, while its treatments, either from 24 to
36 hpf or 24 to 48 hpf, had little inhibition, and its treatments from 36 to 48 hpf had no inhibition
(Fig. S2O), suggesting that miconazole plays a protective role in generating mature cerebral
vessels before their ruptures and hemorrhages occur in fn40a mutants around 40 hpf, and/or
miconazole might take longer time to reach its target and have action in vivo. Taken together, we
have identified a suitable hemorrhagic stroke model in zebrafish that can be exploited for
identifying anti-fungal drug miconazole as a novel, potent hemorrhagic suppressor of fn40a
mutants.
Miconazole suppresses intracerebral hemorrhages in fn40a mutants
O-dianisidine staining of live embryos allowed us to compare the amounts of erythrocytes in
heterozygous and homozygous fn40a mutants in the presence of PTU (0.003%) (Fig. 2A, A’, C,
C’). While erythrocytes were found in the heart, common cardinal vein, dorsal aorta, and other
major vessels in heterozygous siblings treated with DMSO or miconazole at 2 dpf (Fig. 2A-B,
A’-B’), brain hemorrhages were evident in homozygous mutants (Fig. 2C’) and were suppressed
by miconazole treatment (Fig. 2D’). We found that hemorrhage rate in normal pigmented mutant
embryos was similar to that in PTU-treated mutant embryos, which were also suppressed by
miconazole (Fig. 2A-D, A’-D’; Fig. S2A-H). To quantitate the amount of hemorrhage, we
applied Tg(flk1:eGFP) to label vascular endothelial cells and Tg(gata1:DsRed) to label
erythrocytes. Consistent with the data by O-dianisidine staining, we found that brain hemorrhage
was evident in homozygous mutants compared with heterozygous siblings, which miconazole
inhibited hemorrhage (Fig. S2I-M). By testing five known analogues of miconazole, we found
that sulconazole or tioconazole efficiently inhibited brain hemorrhages while voriconazole,
fluconazole, or ketoconazole had little inhibition on hemorrhages (Fig. S3; n>50), suggesting
that the imidazole group is essential for the activity of hemorrhagic suppression of miconazole.
To address whether vessel patterning defects occur in homozygous mutants, we bred flk1:eGFP
transgenic zebrafish into the fn40a mutant background. The flk1:eGFP-labeled cerebral vessels
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
were well patterned and symmetrical in heterozygous siblings treated with DMSO or miconazole
at 2 dpf and 3 dpf (Fig. 2E-F, 2I-J). On the other hand, some of the central arteries (CtA) were
disrupted in homozygous fn40a mutants treated with DMSO at 2 dpf (Fig. 1G) and 3 dpf (Fig.
1K), consistent with the previous report on bubbleheadm292(Liu et al., 2007). Remarkably,
miconazole treatment protected the CtA from ruptures and hemorrhages (Fig. 2D’, H, L).
To gain cellular insights into the hemorrhage processes, we first set up time-lapse 3-dimensional
imaging to monitor the development of brain vessels in live embryos using a LSM700 confocal
microscopy. The fli1a:nuc-eGFP transgenic reporter was used to label the nuclei of both
endothelial cells and erythroid cells, allowing us to easily identify proliferation and migration of
these cells during development. In fn40a mutants, cerebral vessels were well-patterned and grew
normally up to 1.5 dpf, around which time point blood cells frequently leaked out from the
central arteries to form hematomas, resulting in further inflammation and disruption of the
surrounding vessels (Movie S4; n>5) compared with heterozygous siblings (Movie S3; n>5).
These data suggested cranial vascular patterning is relatively normal but those vessels are
immature and fragile in fn40a mutant.
To visualize the subcellular structures of cerebral vessels and surrounding cells, we used
transmission electron microscopy (TEM) and found that blood cells were normally located inside
blood vessels and did not infiltrate into surrounding neurons in heterozygous mutants (DMSO
treatment) (Fig. 2M, Q), which were not affected by miconazole treatment (Fig. 2N, R). In
contrast, blood cells invaded neuronal tissues (Fig. 2O) in mutants and blood vessels were
disrupted and a large amount of blood plasma, with high electron-density, diffused outward and
interrupted the interactions of surrounding cells in mutants (Fig. 2S). These abnormalities were
rescued by miconazole, which restored the restriction of blood cells and plasma within blood
vessels in mutants (Fig. 2P, T). Similarly, microvessels were also disrupted in mutants and these
defects were rescued by miconazole treatment compared with heterozygous siblings with or
without miconazole treatment (Fig. S4). Therefore, our data reveal the critical role of miconazole
in protecting cerebral vessels from ruptures and hemorrhages in fn40a mutants.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Miconazole improves vessel integrity by decreasing MMP9 in fn40a mutants
The above data suggested that the integrity of brain vessels was compromised in fn40a mutants.
We then tested whether vascular permeability was affected in this mutant by examining the
leakage of DAPI injected through the sinus venous of wild-type embryos (Fig. 3E-F, E’-F’, I-J,
I’-J’) and fn40a mutants (Fig. 3G-H, G’-H’, K-L, K’-L’) at 36 hpf (Fig. 3E-H, E’-H’; before
evident brain hemorrhages) and 48 hpf (Fig. 3I-L, I’-L’; after initiation of brain hemorrhages). As
expected, we found a little amount of DAPI leakage from brain vessels in wild-type siblings at
36 hpf (Fig. 3E, E’) and at 48 hpf (Fig. 3I, I’), and this was not affected by miconazole treatment
(Fig. 3F, F’, J, J’). In contrast, we found a large amount of DAPI leaked from blood vessels,
resulting in DAPI-stained neurons in fn40a mutants (DMSO control) before (36 hpf) or after (48
hpf) brain hemorrhage occurred respectively (Fig. 3G, G’, K, K’), and this was rescued by
miconazole (Fig. 3H, H’, L, L’). These data strongly support the notion that brain vessels are
compromised with higher permeability in fn40a mutants.
In mammals, the neurovascular unit (NVU) maintains cerebral homeostasis(Posada-Duque et al.,
2014). NVU includes the cellular components of vascular endothelia, neurons and non-neuronal
cells (microglia, astrocytes, and pericytes), as well as the molecular components of intercellular
junction proteins like tight junctions and extracellular proteins like matrix metalloproteinases
(MMPs)(Xing et al., 2012). Since we found minimal changes in the structure and number of
neurons and endothelial cells in fn40a mutant by fluorescent transgenic reporters, we wondered
whether the molecular components were affected. Fluorescence immunohistochemistry showed
comparable zonula occludens-1 (ZO1) expression in wild-type siblings and mutants, and this was
not affected by miconazole (Fig. 3A-D, A’-D’), suggesting that tight junctions were probably less
affected at this level. Using TEM analysis, we found that cerebral endothelial cells were
disrupted in mutants, and these defects were rescued by miconazole treatment (Fig. S4). On the
other hand, we did not find the blood-brain barrier structure outside of endothelium as that in
mammals by TEM, while blood cells and plasma flew exclusively in blood vessels with no or
little leakage of DAPI in 2.5 dpf embryos. These data further support that miconazole protects
these immature cerebral vessels at this early embryonic stage in zebrafish.
Previous research in rodents has demonstrated that MMPs are essential components for
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
regulating vascular integrity(Xing et al., 2012). The increasing level of MMPs enhances
proteolysis in the extracellular matrix (ECM) and facilitates cell proliferation and migration
during development and metastasis. More importantly, MMPs, well-known inflammatory
mediators, play an important role in brain injury (del Zoppo, 2010; Xing et al., 2012). In the
zebrafish genome, only six MMP members are well annotated: mmp2, mmp9, mmp13b, mmp14a,
mmp14b, and mmp23aa. We then tested whether these MMPs were affected in fn40a mutants,
and if so, whether they were responsive to miconazole treatment. Quantitative real-time PCR
revealed that mmp9, but not the other mmps, was significantly up-regulated in fn40a mutants
(DMSO control), and this was suppressed by miconazole, compared with wild-type siblings
(DMSO control) at 2 dpf (Fig. 3M). Additional analyses showed that both Mmp9 protein (Fig.
3N) and mmp9 mRNA (Fig. 3O) were up-regulated in fn40a mutants, and these were effectively
rescued by miconazole. Miconazole had no effects on mmp9 mRNA while decreased Mmp9
proteins in wild-type siblings (Fig. 3N), suggesting that it might also regulate Mmp9 protein
stability. In addition, whole-mount in situ hybridization revealed that mmp9 RNA was enriched
in the brain at 1.5 and 2 dpf, and it increased in homozygous mutants, which was suppressed by
miconazole (Fig. S5A-H). Importantly, elevated Mmp9 proteins were associated with brain
vessels of homozygous mutants compared with heterozygous mutants in the absence or presence
of miconazole; and like mmp9 mRNA, its proteins in brain vessels of this mutant were also
suppressed by miconazole (Fig. S5I-Q).
Furthermore, MMP9 is highly conserved among species from zebrafish, mouse, rat, and rabbit to
human (Fig. S6). Consistently, MMP9 inhibitor I (3 µmol/L) effectively decreased the brain
hemorrhage rate from 100% (in mutants with DMSO) to ~50% (in mutants with MMP9 inhibitor
I) (Fig. 3P-Q). Mutants treated by MMP9 inhibitor I had no or little hemorrhage compared with
those in mutants by DMSO (Fig. 3Q). In addition, two other MMP9 inhibitors (II and V) had
similar effect on inhibiting hemorrhage in a dose-dependent manner (Fig. S7E-F). MMP9
inhibitor II had very low solubility in E3 water accompanying with precipitation at 4 µmol/L,
which might cause less efficiency of its inhibition on hemorrhage (Fig. S7E). In addition, MMP9
zymography assay confirmed that either MMP9 inhibitor I, MMP9 inhibitor II, or miconazole
decreased Mmp9 activities of homozygous fn40a mutants compared with DMSO-treated mutants
(Fig. S7G). Taken together, our data suggest that the increased permeability of brain vessels and
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
the resultant hemorrhages are at least partly due to elevated MMP9 in fn40a mutants, and
miconazole suppresses the hemorrhages by decreasing MMP9 expression.
Miconazole has a conserved function in protecting mesenteric vessels against MMP9-
dependent hemorrhagic rupture in rats
Having found that miconazole is an effective hemorrhagic suppressor in zebrafish, we then tested
whether it also played a similar role in mammals. Extending this study to mammals was critical
for both therapeutic utility and mechanistic studies. To achieve this goal, we established a
mesenteric hemorrhagic model in rats by using the urokinase-type plasminogen activator (uPA)-
dependent activation of MMP9, degradation of the ECM, and resultant mesenteric hemorrhages.
Compared with control treatment with physiological saline (Movies S5 and S6), we found that
miconazole (5 or 10 mg/kg by intravenous injection) did not cause mesenteric hemorrhages (Fig.
4A-C), suggesting that miconazole had little toxicity at the concentrations we used. By applying
uPA (30,000 IU/kg) in rats, we started to detect mesenteric hemorrhages at ~60 min, and more
hemorrhages from 90 to 120 min after uPA induction (Fig. 4D; Movies S7 and S8). In contrast,
pre-treatment with miconazole (5 or 10 mg/kg) 30 min before uPA injection prevented
mesenteric hemorrhages (Fig. 4E-F) in terms of both the number of hemorrhagic spots and the
hemorrhagic area (Fig. 4G-H). Together with its role in the maturation of developing cerebral
vessels in zebrafish (Figures 2-3; Figure S2O), these pharmacological data suggest that
miconazole plays a conserved role in protecting vessels against MMP9-dependent rupture in rats.
Therefore, identification of chemical suppressors such as miconazole for developing cerebral
vessel ruptures has important implications in decreasing adult pathological phenotypes including
vessel ruptures and hemorrhages.
Previous studies have demonstrated that uPA cleaves plasminogen to generate the active
proteinase plasmin; and plasmin can reciprocally activate pro-uPA. Plasmin cleaves and activates
pro-matrix metalloproteases (pro-MMPs) including pro-MMP9 to activated MMPs; and both
plasmin and MMPs break down many components of the ECM(Smith and Marshall, 2010). In
the absence of uPA, we found a well organized ECM, labeled by either collagen type IV or
laminin (Lama1) in mesenteric microvessels treated with physiological saline (Fig. 5A, A’, G,
G’), 5 mg/kg miconazole (Fig. 5B, B’, H, H’), or 10 mg/kg miconazole (Fig. 5C, C’, I, I’). CD31
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
labeled the vascular endothelium and DAPI staining for nuclei. In the presence of uPA, the
fibrillary ECM of mesenteric vessels was partly broken down and became less organized on
transverse sections (Fig. 5D, D’, J, J’), and this was rescued by pre-treatment with miconazole at
either 5 or 10 mg/kg (Fig. 5E-F, E’-F’, K-L, K’-L’). In addition, the tight-junction proteins ZO1
and Claudin 5 decreased in this uPA-induced hemorrhagic model, and this was partially rescued
by pre-treatment with miconazole (data not shown). As expected, MMP9 was up-regulated in the
ECM close to the endothelium in this uPA-induced model, and this was effectively suppressed by
pre-treatment with miconazole (Fig. 6A-F, A’-F’). Western blot analysis further substantiated
this conclusion (Fig. 6G). Our data has confirmed previous studies that uPA-induced MMP9
disrupts the ECM and tight junctions, establishing a suitable model of mesenteric hemorrhage,
and, like in zebrafish, miconazole effectively suppresses MMP9-dependent vessel rupture and
hemorrhage, highlighting the idea that miconazole acts on highly-conserved protein components
or signaling pathways.
Miconazole suppresses brain hemorrhage by decreasing the Erk1/2-MMP9 signaling
Through a quantitative mass spectrum analysis of whole embryos, we found that β-Pix and five
Pak-family members were down-regulated in fn40a mutants but these changes were not rescued
by miconazole treatment, compared with sibling heterozygotes (data not shown), suggesting that
miconazole functioned down-stream to the Pix/Pak pathway. Besides, previous studies have
revealed a strong correlation between extracellular signal-regulated kinase 1 and 2 (ERK1/2) and
MMP9 at both the transcriptional and post-transcriptional levels(Garavello et al., 2010). MMP9
is inducible and its expression is regulated by many transcriptional factors including activating
protein 1 and 2, specificity protein-1, NF- B, and polyoma enhancer A binding protein 3/E
twenty-six(St-Pierre et al., 2004). Through the Ras/Raf/MEK/ERK pathway, these transcriptional
factors are activated and translocated into the nucleus to regulate MMP9 gene expression(Chang
et al., 2011; Garavello et al., 2010; Wu et al., 2013). The correlation between ERK1/2 and
MMP9 is quite controversial, given that their regulation could be opposite in different cell types.
However, it is noted that the up-regulation of MMP9 is associated with activation of MEK/ERK
in neurological diseases like ischemic stroke(Maddahi et al., 2009). Therefore, we examined
whether Erk1/2 also elevated Mmp9 in fn40a mutants. Western blot analysis showed that
phosphorylation of Erk1/2 increased in the hemorrhagic mutants, and this was protected by
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
miconazole treatment, compared with wild-type siblings treated with DMSO or miconazole at 24
hpf (Fig. 6H). Importantly, by comparing with heterozygous mutants in the absence or presence
of miconazole (Fig. S8A’-B’; F’-G’), pErk in the brain vessels of this mutant increased at 36 hpf
(Fig. S8C, C’) and at 48 hpf (Fig. S8H, H’), which decreased after miconazole treatment (Fig.
S8D, D’, I, I’), suggesting that miconazole is able to suppress elevated pErk in the brain vessels
of this mutant.
Furthermore, each of the three MEK inhibitors U0126, PD0325901, and SL-327 effectively
suppressed the brain hemorrhage rate of fn40a mutants in a dose-dependent manner (Fig. 6I; Fig.
S7B-D) (Shin et al., 2016). Although PD0325901 was somewhat toxic to embryos, its
suppression was highly efficient. Remarkably, when a MEK inhibitor (U0126 or PD0325901)
and miconazole were combined at lower doses, each of which were less effective in suppressing
hemorrhages, they synergistically inhibited brain hemorrhages in fn40a mutants (Fig. 6J-K).
Regarding hemorrhage rate of mutants treated by miconazole or MEK inhibitors, or combined
ones in Fig. 6I-K, we considered that mutants with little or no hemorrhage were “hemorrhage-
negative” compared with that hemorrhagic mutants by DMSO were “hemorrhage-positive”.
Western blot analysis further confirmed that miconazole and U0126 (both at a lower
concentration of 3 µmol/L) synergistically decreased the pErk/tErk ratio and accordingly
decreased Mmp9 in this hemorrhagic mutant at 24 hpf (Fig. 6L). Taken together, our data suggest
that miconazole suppresses brain hemorrhages in fn40a mutants by decreasing Mek/Erk-
regulated Mmp9 activity to protect the integrity of brain vessels.
Discussion
Strokes are consist of into ischemic and hemorrhagic types. Although hemorrhagic stroke has a
higher mortality than ischemic stroke, particularly with about double the incidence in Asian than
in other ethnic groups(van Asch et al., 2010), we have gained fewer mechanistic insights into
hemorrhagic stroke and so limited therapeutic interventions are available for this disease. To
identify the intervention targets for hemorrhagic stroke, investigators have identified several
mutations in monogenetic stroke syndromes, including NOTCH3 in cerebral autosomal
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
dominant subcortical infarcts and leukoencephalopathy, COL4A1 in COL4A1-related brain
small-vessel disease, and KRIT1, CCM2, and PDCD10 in cerebral cavernous malformations(Li
and Whitehead, 2010; Lindgren, 2014; Rost et al., 2008). Genome-wide association studies have
identified several SNPs associated with sporadic hemorrhagic strokes, including APOE (19q13),
SLC25A44 (1q22), COL4A1 (13q34), and KCNK17 (6p21)(Lindgren, 2014; Rost et al., 2008).
These genetic studies of stroke might eventually lead to the identification of therapeutic targets,
although none of them has been translated into the clinic yet. Mutations of COL4A1 in mice
recapitulate the clinical spectrum of disease including intracerebral hemorrhages, retinal vascular
tortuosity, and glomerular basement membrane defects(Gould et al., 2006). Other mouse/rat
models of hemorrhagic stroke by performing blood injection or collagenase injection are also
reported. The autologous blood injection model mimics intracerebral hematoma but represents
nothing about the vascular ruptures causing this disease(MacLellan et al., 2010; MacLellan et al.,
2008). The collagenase injection model mimics vascular rupture processes but does not mimic
the tissue-specific proteolysis that requires extreme caution of manipulation(Han et al., 2011;
Wang et al., 2003). While these mouse/rat models are widely used for studying the molecular and
cellular mechanisms of hemorrhagic stroke, no targets from these models have yet been
translated into therapeutics(Keep et al., 2012). Here, we report that a known mutant fn40a, with
significant intracerebral hemorrhages, is a hypomorphic allele to m292 (Chen et al., 2001; Liu et
al., 2007) but fn40a is capable of surviving to adulthood, a major feature for producing 100%
offspring from mutant crossings. Brain hemorrhages begin around ~40 hpf, but hematomas are
almost completely absorbed after 4 dpf. Inflammatory cells are also recruited to the hemorrhagic
loci in this mutant. Furthermore, β-pix, which genetically interacts with ccm1 and pak2a
(Buchner et al., 2007; Gore et al., 2008; Liu et al., 2007), is down regulated, while the
hemorrhage-induced inflammatory mediator mmp9 is up-regulated, suggesting that this mutant
recapitulates many aspects of hemorrhagic stroke in humans. As demonstrated in our work, this
mutant is suitable for high-throughput chemical screens for hemorrhagic suppressors. Identified
small-molecule suppressors can not only be used as chemical probes for addressing the
molecular mechanisms underlying hemorrhagic stroke, but also present a novel opportunity for
drug discovery in this devastating disease. Therefore, our work presents a valuable hemorrhagic
stroke model that can be exploited for both mechanistic studies and drug discovery.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Based on this study and previous work(Murthy et al., 2012), we propose that decreased
expression of β-pix in fn40a mutants (hypomorphic allele) decreases Rac1 function to activate
Mek1/2, and then increases the level of phosphorylated Erk and MMP9 transcription/translation,
leading to abnormal vascular permeability and disruption of vascular integrity. As blood
circulation is enhanced and blood pressure increases after 24 hpf, these fragile vessels are
inclined to break down and bleed, leading to brain hemorrhages in this mutant. By performing a
chemical suppressor screen, we found that miconazole suppressed the intracerebral hemorrhages
in fn40a mutants. Miconazole is an approved antifungal agent with a fungus-specific binding site
against ergosterol biosynthesis and is pharmacologically described as an inhibitor of sterol
demethylase and cell-wall synthesis(Pierard et al., 2012). We have demonstrated that miconazole
suppresses brain hemorrhages and vascular permeability by down-regulating pErk and Mmp9 in
fn40a mutants. More importantly, by injecting uPA into rats, we established a mesenteric
hemorrhage model that shares the same MEK/ERK/MMP9 pathway underlying vessel rupture
and hemorrhage. Like in zebrafish, miconazole rectified the over-expression of MMP9 and
suppressed mesenteric hemorrhages in rats. Therefore, miconazole works through highly
conserved target(s) to protect vascular integrity from zebrafish to mammals. Its direct target(s)
remain to be identified.
Clinical therapies for hemorrhagic stroke have not yet progressed beyond the introduction of
thrombin during the hematoma-forming process, and neuron protection or blood-clot surgery
after the formation of a hematoma(Keep et al., 2012). Generally, patients with hemorrhagic
strokes have a high percentage of capillary bleeding with tiny amounts of blood before a lethal
hemorrhage occurs. This opens a window for prevention and/or intervention. In clinical, patients
with ischemic stroke administrated by thrombolytic agent like tPA have a rather high probability
to induce hemorrhage transients, and miconazole may protect patients from such transients.
Based on work in zebrafish and rats, we propose that miconazole may be a good candidate for
these types of prevention of hemorrhagic stroke. Future clinical trials and mechanistic studies
need to bring this promising discovery from basic research into the clinic.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Materials and Methods
Animals
Zebrafish (Danio rerio) were maintained and bred as described in the Zebrafish
Book(Westerfield, 2000). The Tg(fli1a:nuc-eGFP)(Roman et al., 2002), Tg(coro1a:eGFP)(Li et
al., 2012), Tg(flk1:eGFP)(Beis et al., 2005) and Tg(gata1:DsRed)(Traver et al., 2003) zebrafish
were described previously. The fn40a mutant was isolated from a large-scale mutagenesis screen
of the zebrafish genome at Massachusetts General Hospital, Boston(Chen et al., 2001). Wild-type
Wistar rats (Rattus norvegicus, ~200 g, males) were purchased from Vital River, Beijing. The
animals were raised and handled in accordance with the animal protocol IMM-XiongJW-3 for
this study, which was approved by the Institutional Animal Care and Use Committee of Peking
University that is fully accredited by AAALAC International.
Chemical screen in zebrafish
We raised homozygous fn40a mutant embryos to adults, and selected those adult pairs that
consistently produced almost 100% hemorrhagic mutant embryos for this chemical screen. This
hemorrhagic suppressor screen consists of three major steps. We started to perform a primary
screen with 96-well plates. After collecting, sorting, and rinsing the embryos, we arrayed five
fertilized and healthy embryos before 6 hpf in each well of the 96-well plates, which were filled
with 200 µL E3 medium. Chemical compounds were dissolved by DMSO to form the stock
concentration of 20 mmol/L, and the compounds were diluted with E3 medium to form the working
concentration of 10 µmol/L in 96-well plates, and compounds were applied at 6 hpf. Treated
embryos were incubated in an incubator at 28°C, and compounds were washed off and replaced
with fresh E3 medium at 24 hpf. Each compound was repeated in 3 wells. Hemorrhage rate was
calculated at 2 dpf and 3 dpf, as the percentage of the number of hemorrhagic mutants after
compound treatment out of the total hemorrhagic mutants used. The morphological
changes/abnormalities of mutant embryos were also documented. We set up a criterion to choose
candidate compounds that the compounds were capable of decreasing the hemorrhage rate to lower
than 60% but caused minimal embryonic lethality. The primary screen led us to identifying 85
candidate hemorrhagic suppressors. Second, we verify each candidate compound for its efficacy
with more than 200 embryos in a 10-cm dish. We arrayed about 70 mutant embryos for each dish
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
and with 3 dishes for each compound. The fn40a mutant embryos were then treated with 10 µmol/L
compounds starting from 6 to 24 hpf, and compounds were washed off at 24 hpf, and the
hemorrhage rate/morphological abnormalities were documented at 2 dpf and 3 dpf. This second
verification screen resulted in 21 candidate compounds. Third, we carried out the final screen on
testing dose-dependent efficacy and toxicity of the 21 candidate compounds. For those compounds
that caused some embryonic abnormalities, we set up the concentration gradients as 0 (DMSO
control), 2 µmol/L, 4 µmol/L, 6 µmol/L, 8 µmol/L and 10 µmol/L. For those compounds that
caused minimal embryonic abnormalities, we first set up a wider range of concentrations as 0
(DMSO control), 1 µmol/L, 5 µmol/L, 10 µmol/L, 20 µmol/L and 40 µmol/L, and then gradually
narrowed down the range to the final concentration gradient from 0 (DMSO control), 1 µmol/L, 2
µmol/L, 3 µmol/L, 4 µmol/L to 5 µmol/L. The final candidate compounds were chosen based on
their efficacy and minimal toxicity by using the narrowest concentration gradients. About 40
mutant embryos were used for each experimental condition/concentration group in 6-well plates
and repeated for three times. After the three-step screen, we identified 7 candidate hemorrhagic
suppressors and 5 other compounds that caused different embryonic phenotypes such as abnormal
dorsal-ventral patterning and arrhythmia.
Chemical library and small-molecule inhibitors
An in-house small-molecule library containing 923 chemicals from the Chinese National
Compound Library Center, Beijing, China, as well as MMP9 inhibitor I (444278, Calbiochem),
MMP9 inhibitor II (444293, Calbiochem), MMP-9 inhibitor V (444285, Calbiochem), U0126
(U0120, Sigma-Aldrich), PD0325901 (PZ0162, Sigma-Aldrich), and SL-327 (S1066, Selleck)
were applied to siblings or fn40a mutant zebrafish embryos. Imidazole members of sulconazole,
tioconazole, voriconazole, fluconazole, and ketoconazole were also supplied by Chinese
National Compound Library Center. Treated embryos were collected and embedded in 1% low-
melting point agarose for taking live images, or fixed by 4% paraformaldehyde for in situ
hybridization, O-dianisidine staining, or fluorescence immunohistochemistry.
MMP9 Zymography Assay
Homozygous fn40a mutants were treated with DMSO, MMP9 inhibitor I (3 mol/L), MMP9
inhibitor II (3 mol/L), or miconazole (3 mol/L) from 6 to 24 hpf, and treated embryos at 24
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
hpf were collected for protein preparation with a non-reducing protein lysis buffer (C1050,
Applygen) that contains phenylmethyl sulfonyl fluoride (Sigma-Aldrich). Protein concentrations
were quantitated by Bicinchoninic Acid (BCA) protein assay kit (P1511, Applygen). We used
each of protein samples (40 mg) for MMP9 Zymography Assay (P1700, Applygen) according to
the manufacture’s protocol. Proteinase activity was visualized as clear bands on Coomassie blue-
stained gels (SDS-PAGE with substrate G from Applygen) as previously reported (Zhang et al.,
2013).
O-dianisidine staining and Tg(gata1:DsRed) for labeling erythrocytes
Applying 500 μL fresh staining solution (2 mL of 14% o-dianisidine; 500 μL of 0.1 mol/L
NaOAc, pH 4.5; 2 mL of deionized H2O; 100 μL of H2O2) to live, dechorionated embryos,
keeping in dark for 20 min, and washing three times with deionized H2O as previously
described(Wingert et al., 2004). Stained embryos were imaged. Alternatively, Tg(flk1:eGFP;
gata1:DsRed) double transgenic zebrafish were bred with wild-type or fn40a mutant to generate
fn40a+/-;Tg(flk1:eGFP; gata1:DsRed) and fn40a-/-; Tg(flk1:eGFP; gata1:DsRed). The embryos
treated with DMSO or miconazole were embedded in 1% low-melting point agarose, and live
images of brain regions at 2 dpf were taken with Zeiss LSM700 confocol microscope. The
regions of interest were restricted to the hindbrain, exactly posterior to the middle cerebral vein,
and the same depth of z-stacks was kept from all embryos. The Tg(gata1:DsRed) signals for
labeling erythrocytes were quantitated.
Fluorescent immunohistochemistry
Zebrafish embryos and rat tissues underwent cryosection at 5 to 7 µm and collected on gelatin-
covered slides. Sections were then processed for fluorescence immunohistochemistry. The
primary antibodies were anti-MMP9 (HPA001238, Lot D104322, Sigma-Aldrich, USA, 1:100)
for rat tissues, anti-ZO1 (339100, Lot 823765A, Invitrogen, USA, 1:100), anti-CD31 (DIA-310,
Lot, Dianova, Germany, 1:50), anti-laminin (Lama1, L9393, Lot 103M4779, Sigma-Aldrich,
USA, 1:100), anti-collagen type IV (ab6586, Lot GR193836-3, Abcam, UK, 1:100), anti-Mmp9
(55345, Lot OG2101, Anaspec, 1:50) for fish embryos, and anti-Phospho-p44/42 MAPK
(Erk1/2) (#4370, Lot 15, Cell Signaling Technology, 1:100) for fish embryos. The secondary
antibodies were Alexa Fluor 488 goat anti-rabbit IgG (A11008, Lot 1705869, Invitrogen, USA,
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
1:200), Alexa Fluor 555 goat anti-mouse IgG (A21424, Lot 1631208, Invitrogen, USA, 1:200),
Alexa Fluor 555 goat anti-rat IgG (A21434, Lot 1670155, Invitrogen, USA, 1:200), and Alexa
Fluor 555 goat anti-rabbit IgG (A21428, Lot 1608466, Invitrogen, 1:200).
Whole-mount RNA in situ hybridization
Digoxigenin-labeled antisense RNA probes were synthesized by in vitro transcription according
to a standard protocol, and the mmp9 RNA probe was made as previously described(Hillegass et
al., 2008). The mmp9 probe fragment was amplified from zebrafish cDNAs library with the
forward primer (5’-GGGGATTTTGCCCTGATCGTGGA-3’) and a T7-containing reverse primer
(5’-TAATACGACTCACTATAGGG TTCCAGTAGCGCCCGTCCTTGA-3’). T7 RNA polymerase
was used to generate the antisense mmp9 probe. Whole-mount in situ hybridization was performed
as described in the Zebrafish Book(Westerfield, 2000).
Microinjection of DAPI through the sinus venous
Basically, we used a protocol similar to that reported by the Watts group(Tam et al., 2012).
Tg(flk1:eGFP) embryos at 1.5 and 2 dpf were anesthetized and injected with 12 nL of 0.85
mg/mL DAPI through the sinus venous. Embryos were incubated at 28°C for 30 min, and then
imaged by laser scanning confocal microscope (LSM510, Carl Zeiss, Germany) with the
wavelength of 488 nm and 405 nm.
Western blot
Embryos (50 embryos at 2 dpf or 100 embryos at 24 hpf) were collected and rinsed. The yolk
proteins were removed and 100 µL RIPA protein lysis was then applied for isolating proteins.
Proteins were then electrophoresed on 10% or 12% SDS-PAGE gel and transferred onto
polyvinylidene difluoride membrane (Bio-Rad, USA). The primary antibodies were anti-Mmp9
(#55345, Lot OG2101, AnaSpect, USA, 1:500), anti-MMP9 (HPA001238, Lot D104322, Sigma
Aldrich, USA, 1:1000), anti-Phospho-p44/42 MAPK (Erk1/2) (#4370, Lot 15, Cell Signaling
Technology, USA, 1:2000), anti-p44/42 MAPK (Erk1/2) (#4695, Lot 14, Cell Signaling
Technology, USA, 1:2000), anti-α-tubulin (BE0031, Easybio, China, 1:2000), and anti-GAPDH
(#TA-08, Golden Bridge, China, 1:2000). The secondary antibodies were goat anti-rabbit IgG-
HRP (M21001, Abmart, China, 1:5000) and goat anti-mouse IgG-HRP (M21002, Lot 264474,
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Abmart, China, 1:5000). The stained membranes were imaged by BioRad ChemiDocTM MP
Imaging System (USA), and the bands were analyzed by ImageJ.
Quantitative real-time PCR
Total RNA from embryos at given stages was extracted using a TRIzol protocol as previously
described(Talbot and Schier, 1999). For each sample, 50 embryos were collected at 2 dpf, and
the experiments were independently repeated for three times. Primers of mmp2, mmp9, mmp13b,
mmp14a, mmp14b, and mmp23aa were synthesized to produce 100-150 bp fragments, of which
the sequences are shown in Supplemental Table S2. The quantitative real-time PCR was run with
the SYBR Premix DimerEraser (RR091a, Takara, Japan).
Transmission electron microscopy
As previously reported(Zhang et al., 2014), the embryos were harvested and fixed in 2%
paraformaldehyde and 2.5% glutaraldehyde in 0.1 mol/L PBS at 4°C for overnight. After a brief
rinse, embryos were postfixed with 1% osmium tetroxide in 0.1 mol/L PBS for 2 h at room
temperature, and then embedded in Epon 812. Ultrathin sections were stained with uranyl acetate
and lead citrate, and stained sections were observed and photographed in a transmission electron
microscopy (JEM 1230; JEOL, Japan).
Establishment of a rat model for urokinase-induced mesenteric hemorrhage and intravital
observation of mesenteric microcirculation
Rats were anesthetized by intramuscular injection of urethane (2 g/kg). Cannulae were inserted
into the left internal jugular and femoral vein for administration of urokinase (30,000 IU/kg,
Aladine, China) and miconazole (5 or 10 mg/kg) respectively. Miconazole or physiological
saline (NS) was injected into animals half an hour earlier than imaging the baseline. After
baseline recorded, urokinase solution or NS was injected as quickly as possible. Rats were kept
in 37°C incubator and mesenteric membranes were kept moisture with warm NS. Images and
records were documented at 30-min intervals for a total of 2 hours. Animals were placed in the
right-lateral decubitus position and the mesenteric microcirculation was observed under an
inverted microscope (BX51WI, Olympus, Japan) equipped with a color monitor (TCL J2118A,
TCL, China) and DVD recorder (DVR-R25, Malata, China). The hemorrhage spots and
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
hemorrhagic area of each spot were calculated by Image-Pro Plus and ImageJ. The total spot
number and the hemorrhagic area of the whole mesenteric membrane were presented for each
time point with statistic curves.
Statistical analyses
All experiments were repeated at least three times with more than 50 zebrafish embryos, and
with 5 or more rats. Two-tailed Student’s t-test, one-way ANOWA or two-way ANOVA with post
test was performed to evaluate the significance of differences between experimental and control
groups and was expressed by mean±SEM. *p <0.05 indicates a statistically significant effect.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Acknowledgements
The authors thank Dr. IC Bruce for critical comments and reading the manuscript; Drs. S Ding
and S Tang, as well as members of Drs. Xiong’s and Han’s laboratories for helpful discussion
and technical assistance; Dr. Jiulin Du for sharing the DAPI injection protocol for assessing
vascular permeability; Drs. BH Hu and X Chang for assistance with transmission electron
microscopy; and Dr. WH Wang at the Chinese National Compound Library for providing the
chemicals.
Competing interests
No competing interests declared.
Authorship Contributions
RY performed most of the experiments, analyzed the data, and wrote the manuscript; YZ and DH
performed some of the early experiments on the rat mesenteric hemorrhage model; X Zhu
supervised some of the experiments by RY and analyzed the data; X Zhang designed and
supervised some of the experiments and analyzed the data collected by RY; XL, LZ, and ZL
contributed the chemical library, synthesized miconazole and its analogues, and analyzed data;
JH designed the rat mesenteric model experiments, analyzed data, and contributed to manuscript
writing; and JWX conceived and designed this work, analyzed data, and wrote the manuscript.
Funding
This work was supported by grants from the National Basic Research Program of China
(2012CB944501), the National Natural Science Foundation of China (31430059, 81270164,
31271549, 81470399, and 31521062), and AstraZeneca Asia and Emerging Market Innovative
Medicine and Early Development.
Data availability
All data supporting this work are included in this manuscript (6 Figures, 8 Supplemental Figures,
2 Supplemental Tables, and 8 Supplemental Movies).
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
References
Beis, D., Bartman, T., Jin, S. W., Scott, I. C., D'Amico, L. A., Ober, E. A., Verkade,
H., Frantsve, J., Field, H. A., Wehman, A. et al. (2005). Genetic and cellular analyses of
zebrafish atrioventricular cushion and valve development. Development 132, 4193-204.
Buchner, D. A., Su, F., Yamaoka, J. S., Kamei, M., Shavit, J. A., Barthel, L. K.,
McGee, B., Amigo, J. D., Kim, S., Hanosh, A. W. et al. (2007). pak2a mutations cause cerebral
hemorrhage in redhead zebrafish. Proc Natl Acad Sci U S A 104, 13996-4001.
Chang, Y. M., Shih, Y. T., Chen, Y. S., Liu, C. L., Fang, W. K., Tsai, C. H., Tsai, F. J.,
Kuo, W. W., Lai, T. Y. and Huang, C. Y. (2011). Schwann Cell Migration Induced by
Earthworm Extract via Activation of PAs and MMP2/9 Mediated through ERK1/2 and p38. Evid
Based Complement Alternat Med 2011, 395458.
Chen, J. N., van Bebber, F., Goldstein, A. M., Serluca, F. C., Jackson, D., Childs, S.,
Serbedzija, G., Warren, K. S., Mably, J. D., Lindahl, P. et al. (2001). Genetic steps to organ
laterality in zebrafish. Comp Funct Genomics 2, 60-8.
del Zoppo, G. J. (2010). The neurovascular unit, matrix proteases, and innate
inflammation. Annals of the New York Academy of Sciences 1207, 46-9.
Egashira, Y., Hua, Y., Keep, R. F. and Xi, G. (2015). Intercellular cross-talk in
intracerebral hemorrhage. Brain Research 1623, 97-109.
Garavello, W., Maggioni, D., Nicolini, G., Motta, L., Tredici, G. and Gaini, R. (2010).
Association between metalloproteinases 2 and 9 activity and ERK1/2 phosphorylation status in
head and neck cancers: an ex vivo study. Oncol Rep 24, 1073-8.
Gore, A. V., Lampugnani, M. G., Dye, L., Dejana, E. and Weinstein, B. M. (2008).
Combinatorial interaction between CCM pathway genes precipitates hemorrhagic stroke. Dis
Model Mech 1, 275-81.
Gould, D. B., Phalan, F. C., van Mil, S. E., Sundberg, J. P., Vahedi, K., Massin, P.,
Bousser, M. G., Heutink, P., Miner, J. H., Tournier-Lasserve, E. et al. (2006). Role of
COL4A1 in small-vessel disease and hemorrhagic stroke. New England Journal of Medicine
354, 1489-96.
Han, Q. Q., Jin, W., Xiao, Z. F., Huang, J. C., Ni, H. B., Kong, J., Wu, J., Chen, B.,
Liang, W. B. and Dai, J. W. (2011). The promotion of neurological recovery in an intracerebral
hemorrhage model using fibrin-binding brain derived neurotrophic factor. Biomaterials 32, 3244-
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
52.
Hillegass, J. M., Villano, C. M., Cooper, K. R. and White, L. A. (2008).
Glucocorticoids alter craniofacial development and increase expression and activity of matrix
metalloproteinases in developing zebrafish (Danio rerio). Toxicol Sci 102, 413-24.
Keep, R. F., Hua, Y. and Xi, G. (2012). Intracerebral haemorrhage: mechanisms of
injury and therapeutic targets. Lancet Neurol 11, 720-31.
Li, D. Y. and Whitehead, K. J. (2010). Evaluating strategies for the treatment of
cerebral cavernous malformations. Stroke 41, S92-4.
Li, L., Yan, B., Shi, Y. Q., Zhang, W. Q. and Wen, Z. L. (2012). Live imaging reveals
differing roles of macrophages and neutrophils during zebrafish tail fin regeneration. Journal of
Biological Chemistry 287, 25353-60.
Lindgren, A. (2014). Stroke genetics: a review and update. J Stroke 16, 114-23.
Liu, J., Fraser, S. D., Faloon, P. W., Rollins, E. L., Vom Berg, J., Starovic-Subota, O.,
Laliberte, A. L., Chen, J. N., Serluca, F. C. and Childs, S. J. (2007). A betaPix Pak2a
signaling pathway regulates cerebral vascular stability in zebrafish. Proc Natl Acad Sci U S A
104, 13990-5.
Liu, J., Zeng, L., Kennedy, R. M., Gruenig, N. M. and Childs, S. J. (2012). betaPix
plays a dual role in cerebral vascular stability and angiogenesis, and interacts with integrin
alphavbeta8. Developmental Biology 363, 95-105.
MacLellan, C. L., Silasi, G., Auriat, A. M. and Colbourne, F. (2010). Rodent models
of intracerebral hemorrhage. Stroke 41, S95-8.
MacLellan, C. L., Silasi, G., Poon, C. C., Edmundson, C. L., Buist, R., Peeling, J.
and Colbourne, F. (2008). Intracerebral hemorrhage models in rat: comparing collagenase to
blood infusion. Journal of Cerebral Blood Flow and Metabolism 28, 516-25.
MacRae, C. A. and Peterson, R. T. (2015). Zebrafish as tools for drug discovery. Nature
Reviews Drug Discovery 14, 721-31.
Maddahi, A., Chen, Q. and Edvinsson, L. (2009). Enhanced cerebrovascular expression
of matrix metalloproteinase-9 and tissue inhibitor of metalloproteinase-1 via the MEK/ERK
pathway during cerebral ischemia in the rat. BMC Neurosci 10, 56.
Murthy, S., Ryan, A. J. and Carter, A. B. (2012). SP-1 regulation of MMP-9 expression
requires Ser586 in the PEST domain. Biochemical Journal 445, 229-36.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Peterson, R. T., Shaw, S. Y., Peterson, T. A., Milan, D. J., Zhong, T. P., Schreiber, S.
L., MacRae, C. A. and Fishman, M. C. (2004). Chemical suppression of a genetic mutation in
a zebrafish model of aortic coarctation. Nature Biotechnology 22, 595-9.
Pierard, G. E., Hermanns-Le, T., Delvenne, P. and Pierard-Franchimont, C. (2012).
Miconazole, a pharmacological barrier to skin fungal infections. Expert Opin Pharmacother 13,
1187-94.
Posada-Duque, R. A., Barreto, G. E. and Cardona-Gomez, G. P. (2014). Protection
after stroke: cellular effectors of neurovascular unit integrity. Front Cell Neurosci 8, 231.
Roman, B. L., Pham, V. N., Lawson, N. D., Kulik, M., Childs, S., Lekven, A. C.,
Garrity, D. M., Moon, R. T., Fishman, M. C., Lechleider, R. J. et al. (2002). Disruption of
acvrl1 increases endothelial cell number in zebrafish cranial vessels. Development 129, 3009-19.
Rost, N. S., Greenberg, S. M. and Rosand, J. (2008). The genetic architecture of
intracerebral hemorrhage. Stroke 39, 2166-73.
Shin, M., Beane, T. J., Quillien, A., Male, I., Zhu, L. J. and Lawson, N. D. (2016).
Vegfa signals through ERK to promote angiogenesis, but not artery differentiation. Development
143, 3796-3805.
Smith, H. W. and Marshall, C. J. (2010). Regulation of cell signalling by uPAR. Nat
Rev Mol Cell Biol 11, 23-36.
St-Pierre, Y., Couillard, J. and Van Themsche, C. (2004). Regulation of MMP-9 gene
expression for the development of novel molecular targets against cancer and inflammatory
diseases. Expert Opin Ther Targets 8, 473-89.
Talbot, W. S. and Schier, A. F. (1999). Methods in Cell Biology: Academic Press.
Tam, S. J., Richmond, D. L., Kaminker, J. S., Modrusan, Z., Martin-McNulty, B.,
Cao, T. C., Weimer, R. M., Carano, R. A., van Bruggen, N. and Watts, R. J. (2012). Death
receptors DR6 and TROY regulate brain vascular development. Developmental Cell 22, 403-17.
Traver, D., Paw, B. H., Poss, K. D., Penberthy, W. T., Lin, S. and Zon, L. I. (2003).
Transplantation and in vivo imaging of multilineage engraftment in zebrafish bloodless mutants.
Nat Immunol 4, 1238-46.
van Asch, C. J., Luitse, M. J., Rinkel, G. J., van der Tweel, I., Algra, A. and Klijn, C.
J. (2010). Incidence, case fatality, and functional outcome of intracerebral haemorrhage over
time, according to age, sex, and ethnic origin: a systematic review and meta-analysis. Lancet
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Neurol 9, 167-76.
Wang, J. (2010). Preclinical and clinical research on inflammation after intracerebral
hemorrhage. Progress in Neurobiology 92, 463-77.
Wang, J., Rogove, A. D., Tsirka, A. E. and Tsirka, S. E. (2003). Protective role of
tuftsin fragment 1-3 in an animal model of intracerebral hemorrhage. Annals of Neurology 54,
655-64.
Westerfield, M. (2000). The zebrafish book. A guide for the baloratory use of zebrafish
(Danio rerio). Eugene: Univ. of Oregen Press.
WHO. (2015). Health in 2015: from the Millennium Development Goals (MDGs) to
Sustainable Development Goals (SDGs), vol. Chapter 6. Noncommunicable Diseases: WHO.
Wingert, R. A., Brownlie, A., Galloway, J. L., Dooley, K., Fraenkel, P., Axe, J. L.,
Davidson, A. J., Barut, B., Noriega, L., Sheng, X. et al. (2004). The chianti zebrafish mutant
provides a model for erythroid-specific disruption of transferrin receptor 1. Development 131,
6225-35.
Wu, X., Huang, W., Luo, G. and Alain, L. A. (2013). Hypoxia induces connexin 43
dysregulation by modulating matrix metalloproteinases via MAPK signaling. Molecular and
Cellular Biochemistry 384, 155-62.
Xing, C., Hayakawa, K., Lok, J., Arai, K. and Lo, E. H. (2012). Injury and repair in
the neurovascular unit. Neurological Research 34, 325-30.
Zhang, Y., Huang, L., Wang, C., Gao, D. and Zuo, Z. (2013). Phenanthrene exposure
produces cardiac defects during embryo development of zebrafish (Danio rerio) through
activation of MMP-9. Chemosphere 93, 1168-75.
Zhang, Y., Sun, K., Liu, Y. Y., Zhang, Y. P., Hu, B. H., Chang, X., Yan, L., Pan, C. S.,
Li, Q., Fan, J. Y. et al. (2014). Ginsenoside Rb1 ameliorates lipopolysaccharide-induced
albumin leakage from rat mesenteric venules by intervening in both trans- and paracellular
pathway. Am J Physiol Gastrointest Liver Physiol 306, G289-300.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Figures
Fig. 1. The fn40a mutant is a hemorrhagic stroke model suitable for chemical suppressor
screens in zebrafish (seen also Fig. S1-2). (A-D) Live images of wild-type siblings (A-B) and
homozygous fn40a mutants (C-D) at 2 dpf; note evident brain hemorrhages in the mutant
(arrows). A, C: lateral view; B, D: dorsal view. (E-L) Live images of Tg(flk1:eGFP);
Tg(gata1:DsRed) double transgenic embryos of wild-type siblings (E-F, I-J) and fn40a mutants
(G-H, K-L) at 2 dpf. Note that gata1+ erythrocytes leaked from Tg(flk1:eGFP)-labeled vessels
and formed hematomas in the mutants (G-H, K; arrow) while erythrocytes remained within
vessels in wild-type siblings (E-F, I). Tg(coro1a:eGFP)-labeled macrophages and neutrophils
accumulated at the hemorrhagic site in this mutant (L, arrow) while leukocytes were evenly
distributed in the brain of a wild-type sibling embryo (J) at 2 dpf (n>8). (M) Overview of the
chemical screen with mutant embryos arrayed in a 96-well plate and given DMSO or compound
treatment, illustrating that miconazole suppressed brain hemorrhages in a dose-dependent
manner compared with DMSO treatment. Redhead fish indicates hemorrhagic mutants.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Fig. 2. Miconazole suppresses brain hemorrhages in fn40a-/- mutants (seen also Fig. S2-4
and Table S1). (A-D, A’-D’) O-dianisidine staining for erythrocytes showing hemorrhages in the
dorsal part of the brain (C’) that were suppressed by miconazole (mic) treatment (D’) in
homozygous fn40a mutants while no brain hemorrhages occurred in heterozygous fn40a mutants
in the absence or presence of miconazole (mic) (A-B, A’-B’) at 2 dpf. A-D, ventral view; A’-D’,
dorsal view. The numbers on the lower right show the number of hemorrhagic (C, C’) and non-
hemorrhagic (A-B, A’-B’, D, D’) embryos out of the total embryos scored. (E-L) CtA between
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
BA and PHBC labeled in Tg(flk1:eGFP) zebrafish were disrupted (G, K, arrows) in homozygous
mutants. This was rescued by mic treatment (H, L) in homozygous mutants compared with
heterozygous mutants with either DMSO or mic treatment (E-F, I-J). Images are representative
ones from more than 10 embryos per group (n>10). E-H, 2 dpf; I-L, 3 dpf. Scale bar, 50 m. (M-
T) Transmission electron microscopy uncovered the invasion of blood cells (red arrows) and
disrupted neurons (O) and the leakage of high-density blood plasma (red asterisks) into the
surrounding brain tissue (S) in fn40a homozygous mutants. This was rescued by mic treatment
(P, T), compared with heterozygous mutants with or without mic treatment (M-N, Q-R) at 2.5
dpf. Images are representative ones from more than 8 embryos per group (n>8). Q-T, dashed
lines mark blood vessels border. Red arrows point to invaded blood cells; yellow arrows point to
vacuoles after cell death. RB, red blood cell; Ne, neuron. Scale bars, 5 μm.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Fig. 3. Miconazole suppresses brain hemorrhages by preventing MMP9-mediated vessel
disruption in fn40a-/- mutants (seen also Fig. S5). (A-D, A’-D’) Fluorescence
immunohistochemistry revealed comparable expression of the tight junction protein ZO1 in
Tg(flk1:eGFP)-labeled brain vessels with or without miconazole (mic) treatment in wild type
siblings (A-B, A’-B’) and fn40a mutants (C-D, C’-D’) at 2 dpf. Images are representative ones
from more than 15 embryos per group (n >15). (E-L, E’-L’) Microinjection of DAPI into the
sinus venous revealed that neurons were stained by leaked DAPI before (G, G’, arrow) and after
(K, K’, arrow) evident hemorrhages, suggesting that vascular permeability increased in
homozygous fn40a mutants. This was rescued by mic treatment (H, H’, L, L’), compared with
much less leakage of DAPI in siblings with or without mic treatment (E-F, E’-F’, I-J, I’-J’). E-H
and E’-H’ were embryos at 1.5 dpf, I-L and I’-L’ were embryos at 2 dpf. The numbers on the
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
lower right show the number of phenotypical embryos out of the total embryos scored. Scale
bars, 100 µm. (M) Quantitative-real time PCR identified that mmp9, but not other mmp genes,
was up-regulated in homozygous fn40a mutants and that was suppressed to the normal level by
mic treatment at 2 dpf. *p <0.05 by two-way ANOVA with Bonferroni post-tests, compared with
sib-DMSO for each gene, mean±SEM. (N, O) Western blot (N) and qRT-PCR (O) confirmed that
both MMP9 protein and mmp9 mRNA increased in the mutants with DMSO control treatment
and this was normalized by mic treatment, compared with wild-type siblings with or without mic
treatment at 2 dpf. *p <0.05 by one-way ANOVA with Bonferroni’s Multiple Comparison Test,
mean±SEM. Total RNA was extracted from whole embryos and protein from whole embryos
without yolk. (P, Q) MMP9 inhibitor I suppressed brain hemorrhages rate to 50% in homozygous
mutants (P), which brain hemorrhage was scored with live mutant embryos treated by DMSO or
MMP9 inhibitor (Q). *p <0.05 by Student’s t-test, compared with controls, mean±SEM.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Fig. 4. Miconazole suppresses uPA-induced mesenteric hemorrhage in rats. (A-C) in control
group, neither physiological saline (NS) nor miconazole (mic) at 5 or 10 mg/kg caused
mesenteric hemorrhages 2 h after treatment. (D-F) uPA treatment induced mesenteric
hemorrhages from 1 to 2 h post-treatment (arrows) in the presence of NS (D). This was
suppressed by pre-treatment with miconazole at either 5 or 10 mg/kg (E-F). (G-H) Both the
number of hemorrhagic spots (G) and the hemorrhagic area (H) that induced by uPA treatment
were significantly decreased by pre-treatment with 5 or 10 mg/kg miconazole compared with NS
control without uPA treatment. *p <0.05 by two-way ANOVA with Bonferroni post-test
compared with baseline (n >5), mean±SEM.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Fig. 5. Miconazole ameliorates uPA-induced degradation of the extracellular matrix (ECM) of mesenteric vessels in rats. (A-F,A-F) Fluorescence immunohistochemistry showed that the fiber-like ECM structure, stained for collagen (Col) type IV (red), was disrupted by uPA treatment (D,D). This was partly rescued by either 5 or 10 mg/kg miconazole (mic) treatment (E,F,E, F), compared with NS or NS-mic without uPA treatment (A-C,A-C). (G-L,G-L) Fluorescence immunohistochemistry showed that the fiber-like ECM structure, stained for laminin (anti-Lama1, red), was disrupted by uPA treatment (J,J). This was partly rescued by either 5 or 10 mg/kg mic (K,K, L,L), compared with NS or NS-mic without uPA treatment (G-I,G-I). CD31 (green) was used to label the endothelial layer of the mesenteric vessels and DAPI (blue) was used to stain nuclei. The number of animals used each group is more than 5 (n>5). Yellow arrows point to non-specific staining in Fig 5E-E' and 5K-K. Scale bars, 100 µm.
Fig. 6. Miconazole suppresses hemorrhages via the Mek/Erk/Mmp9 pathway (seen also Fig.
S8). (A-F, A’-F’) Fluorescence immunohistochemistry showed that MMP9 (red) was induced in
the ECM by uPA treatment (D, D’). This was rescued by either 5 or 10 mg/kg miconazole (mic)
(E, E’, F, F’), compared with those without uPA treatment (A-C, A’-C’) in rat mesenteric vessels.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
CD31 (green) was used to label the endothelial layer of mesenteric vessels and DAPI (blue) was
used to stain nuclei. The number of animals used each group is more than 5 (n>5). Scale bars,
100 µm. (G) Western blots confirmed that the uPA-induced expression of MMP9 was rescued by
pre-treatment with mic. (H, H’) The pErk increased in homozygous fn40a mutants (DMSO
control), and this was reversed by mic (5 µmol/L), compared with wild-type siblings with
DMSO or mic at 24 hpf (H). Coomassie blue staining showed protein loadings (H’). (I-L) Like
mic, either the MEK inhibitor U0126 or PD0325901 suppressed the brain hemorrhages in
homozygous fn40a mutants in a dose-dependent manner (I). Mic and PD0325901 (J) or U0126
(K) synergistically suppressed brain hemorrhages in this mutant. Miconazole (3 µmol/L) and
U0126 (3 µmol/L) synergistically suppressed pErk, as well as Mmp9 in fn40a mutants at 24 hpf,
compared with wild-type siblings with DMSO or miconazole treatment (L). *p <0.05 by one-
way ANOVA with Bonferroni’s Multiple Comparison Test (G, H, J, L), or two-way ANOVA with
Bonferroni post-tests compared with control (I, K), mean±SEM, n >50.
Dis
ease
Mo
dels
& M
echa
nism
s •
DM
M •
Adv
ance
art
icle
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Materials
Supplemental Figure 1 (refer to Fig. 1). Chemical structures of the seven candidate hemorrhagic suppressors (identified in the small-molecule screen in zebrafish, including 3-methyl-4-nitro benzoic acid, clotrimazole, meloxicam, nisoldipine, N-cyclohexyl-p-toluenesulfonamid, oxyclozanide, and miconazole nitrate.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 2 (refer to Figs. 1 and 2). Miconazole suppresses brain hemorrhages in fn40a mutants. (A-D) Live image of embryos without PTU treatment. The homozygous fn40a mutant showed evident brain hemorrhage in the midbrain and hindbrain (C, arrow) compared with heterozygous mutant in the absence (A) or presence (B) of miconazole, and the hemorrhage was repressed in homozygous mutants by miconazole treatment (D). Embryos at 2 dpf were orientated in the lateral view. (E-H) O-dianisidine staining of embryos without PTU treatment revealed that erythrocytes accumulated in the brain region of homozygous fn40a mutant (G, arrow) compared with heterozygous mutant in the absence (E) or presence (F) of miconazole, and this hemorrhage phenotype in homozygous mutants was rescued by miconazole treatment (H). Embryos at 2 dpf
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
were orientated in the dorsal view. The numbers on lower right show the phenotypical ones out of total scored embryos. (I-L, I’-L’) Live images of Tg(flk1:eGFP; gata1:DsRed) double transgenic embryos of heterozygote siblings (I-J, I’-J’) and homozygous fn40a mutants (K-L, K’-L’) at 2 dpf in the hindbrain region. Note that gata1+ erythrocytes leaked out from Tg(flk1:eGFP)-labeled vessels and formed hematomas in the mutants (K, K’) while erythrocytes remained within vessels in heterozygous siblings (I-J, I’-J’) and mic-treated homozygous mutants (L, L’). (M) Intensity of the gata1:DsRed signals in the hindbrain region shown in panels A-D, noting a burst of gata1:DsRed signals in homozygous fn40a mutants compared with heterozygous mutants, which was reversed in homozygous mutants after mic treatment. *p <0.05 by one-way ANOVA with Bonferroni’s Multiple Comparison Test, n>10. (N) Miconazole suppresses brain hemorrhages in fn40a mutants in a dose-dependent manner. Note that miconazole at 10 mol/L completely suppressed brain hemorrhage in fn40a mutants. *p <0.05 by one-way ANOVA with Dunnett’s Multiple Comparison Test, compared with control, mean±SEM., n >50 for each independent repeat. (O) Application of miconazole from 6 to 24 hpf had more effective suppression on brain hemorrhage in fn40a mutants. At the concentration of 3.3 µmol/L, miconazole effectively suppressed brain hemorrhage (more than 50%) when its treatment from 6 to 24 hpf, but had little effect when its treatment either from 24 to 36 hpf, or 24 to 48 hpf, and had no effect when having treatment from 36 to 48 hpf. Similarly, at the concentration of 5 µmol/L, miconazole inhibited brain hemorrhage of fn40a mutants (better than that at 3.3 µmol/L) when having treatment from 6 to 24 hpf, and had less inhibition when having treatment either from 24 to 36 hpf or from 24 to 48 hpf, but had no effect when having treatment from 36 to 48 hpf. *p <0.05 by two-way ANOVA with Bonferroni’s post test compared with control, mean±SEM, n >100.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 3 (refer to Fig. 2). (A) Structures of six imidazole members that are miconazole analogues. The difference of each analogues from miconazole is marked by red-dot boxes. (B) Hemorrhagic suppression efficiency of the six imidazole members. Note that, like miconazole, sulconazole and tioconazole had dose-dependent inhibition of brain hemorrhage, while voriconazole, fluconazole, and ketoconazole had much less potency and were not dose-dependent in suppressing hemorrhage in fn40a mutants. The threshold of hemorrhagic suppression rate was set at 50% and dose-dependent inhibition was an important factor for scoring drug efficacy. Miconazole, sulconazole, or tioconazole at 20 mol/L was toxic to embryos and so the data were not included in this figure. *p <0.05 by two-way ANOVA with Bonferroni’s post test compared with control, mean±SEM, n >50 for each independent repeat.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 4. Ultrastructures of microvessels in the zebrafish brain shown by transmission electron microscopy at 2.5 dpf (refer to Fig. 2). (A-D) Endothelial cells wraps well around red blood cells in fn40a heterozygotes with or without miconazole treatment (A-B). Endothelial cells outside red blood cells were disrupted in fn40a homozygote (C), which was rescued by miconazole treatment (D). RB, red blood cell; E, endothelial cell; Ne, neuron; mic, miconazole. Arrows point to the tight junctions between endothelial cells. The lower panels are magnification of the red-square area in the upper ones. Number of embryos per group was more than 8. Scale bar, 2 µm.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 5 (refer to Figs. 3 and 6). Miconazole inhibits Mmp9 elevation in the brain of fn40a mutants. (A-H) RNA in situ hybridization revealed an increase of mmp9 expression in the brain region of fn40a mutants at 1.5 dpf (C) and 2 dpf (G) compared with heterozygote controls (A, E), and decreased mmp9 expression in fn40a mutants after mic treatment (D, H). A-D, 1.5 dpf; E-H, 2 dpf. A-H, lateral view. Red arrows point to the brain. (I-P, I’-P’) Fluorescence immunohistochemistry showed a higher level of Mmp9 (red) in the brain of homozygous fn40a mutants (K, K’; O, O’) compared with heterozygous controls in the absence (I, I’; M, M’) or presence (J, J’; N, N’) of miconazole; and miconazole decreased the Mmp9 level in homozygous mutants (L, L’; P, P’). Tg(flk1:eGFP) (green) was used for labeling brain vessels and DAPI (blue) for staining the nuclei. I-L, I’-L’, 1.5 dpf, scale bar is 20 µm; M-P, M’-P’, 2 dpf, scale bar is 40 µm. (Q) Intensity of anti-Mmp9 staining signals in the brain regions shown in panels I-P, confirming that the elevated expression of Mmp9 in fn40a mutants was suppressed by miconazole treatments. *p <0.05 by one-way ANOVA with Bonferroni’s Multiple Comparison Test. n >10 for each zebrafish independent repeat.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 6 (refer to Fig. 3). Alignment of protein sequences reveals that MMP9 is highly conserved from zebrafish (DANRE), mouse, rat, rabbit to human.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 7 (refer to Fig. 6). Either MMP9 or pErk inhibitors decrease brain hemorrhage in fn40a mutants. Live imaging is shown for fn40a embryos treated by DMSO or small-molecule inhibitors. The hemorrhage rate of fn40a mutants was suppressed by miconazole (1 to 4 mol/L (µM) mic, A), MEK inhibitor (2 to 8 mol/L U0126, B; 1 to 3 mol/L PD0325901, C; and 5 to 20 mol/L SL-327, D), or MMP9 inhibitor II (1 to 3 mol/L, E) and MMP9 inhibitor V (5 to 10 mol/L, F) compared control embryos treated with DMSO. Embryos were orientated anterior on left with the dorsal view. Red arrow indicates the hemorrhage region. The concentrations of each compound were shown on upper right. The numbers on lower right show the phenotypical ones out of total scored embryos. (G) MMP9 zymography assay revealed that either MMP9 inhibitor I (3 mol/L), MMP9 inhibitor II (3 mol/L), or miconazole (3 mol/L) decreased Mmp9 activities of homozygous fn40a mutants compared with DMSO-treated mutants. This represents one of two independent experiments.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Figure 8 (refer to Fig. 6). Miconazole inhibits pErk elevation in the brain vessels of fn40a mutants. Fluorescence immunohistochemistry showed that the phosphorylated Erk1/2 (pErk, red) in Tg(flk1:eGFP) endothelial cells (green) increased in homozygous fn40a mutants (C, C’; H, H’) compared with heterozygous siblings in the absence (A, A’; F, F’) or presence (B, B’; G, G’) of miconazole; and pErk decreased in homozygous mutants after mic treatments (D, D’; I, I’). Tg(flk1:eGFP) (green) was used to label vascular endothelia cells. The colocalization coefficient of pErk and flk1:eGFP to eGFP signals was shown in panel E (1.5 dpf) and panel J (2 dpf), noting that the pErk was up-regulated in mutant endothelium and the elevated pErk was rescued by miconazole treatments. A-D, A’-D’, 1.5 dpf, scale bar is 20 µm; F-I, F’-I’, 2 dpf, scale bar is 40 µm. DAPI (blue) was used to label nuclei. Yellow arrows indicated the double-positive signals (pErk/flk1:eGFP). *p <0.05 by one-way ANOVA with Bonferroni’s Multiple Comparison Test, n >10 for each zebrafish independent repeat.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Table 1. Seven hemorrhagic suppressors are isolated from a small-molecule screen in zebrafish (refer to Fig. 2). Brain hemorrhage was inhibited by seven candidate hemorrhagic suppressors in a dose-dependent manner; and miconazole was the very potent and least-toxic suppressor. Hemorrhagic rate was calculated as percentage of the number of hemorrhagic mutants after drug treatment from the total number of hemorrhagic mutants. #, no survival embryos; the percentage number, hemorrhage rate.
1 µM 2 µM 3 µM 4 µM 5 µM
3-methly-4-nitro benzoic acid
86.00% 72.00% 68.00% 22.00% #
Clotrimazole 91.84% 88.00% 84.27% 73.14% 72.05%
Miconazole
nitrate 93.02% 80.95% 81.82% 43.48% 30.51%
Meloxicam 92.31% 90.00% 84.15% 72.86% 67.86%
Nisoldipine 46.87% 8.27% 3.49% # #
N-cyclohexyl-p-toluenesulfonamid
87.10% 82.50% 68.68% 68.88% 71.93%
Oxyclozanide 45.00% 13.50% 6.75% # #
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Table 2. Primers for real-time quantitative PCR (refer to Fig. 3).
Genes Sequences (5’ to 3’)
beta-actin ATGCCCCTCGTGCTGTTTTC
GCCTCATCTCCCACATAGGA
mmp2 TGGCTCCTATCTACACATTCAC
ATGGCTTGTCTGTTGGTTCT
mmp9 GAACAGCAATGAAGCACCAT
TAGTGGAGCACCAGCGATA
mmp13b CGACGATGACGAGACATTC
GCGAGTGTTCCAGACCTA
mmp14a CTCTACTGTCACCGCTCTC
TCCAGATCCTTCCTCCTTGT
mmp14b ACTGAAGAGCAACCTGAGA
TTGGCGTGTAGTTCTGTATG
mmp23aa TGTTCATCACTCGTGCTCTAC
CTGTGCGTGTCGTCCATTA
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 1 (refer to Fig. 1). Gata1+ erythrocytes flowed within blood vessels (flk1+) in wild-type Tg(flk1:eGFP; gata1:DsRed) embryos.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 2 (refer to Fig. 1). Gata1+ erythrocytes leaked out and formed hematomas in fn40a mutant Tg(flk1:eGFP; gata1:DsRed) embryos.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 3 (refer to Fig. 1). Time-lapse images showing brain vessel development in heterozygous fn40a Tg(fli1a:nuc-eGFP) embryo from 36 to 48 hpf at 20-min intervals.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 4 (refer to Fig. 1). Time-lapse images showing brain vessel development in homozygous fn40a mutant Tg(fli1a:nuc-eGFP) embryo from 36 to 48 hpf at 20-min intervals. Note vessel rupture and hematoma formation.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 5 (refer to Fig. 4). Blood flowed normally in mesenteric microvessels at 0 h after saline treatment.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 6 (refer to Fig. 4). Blood flowed normally in mesenteric microvessels at 2 h after saline treatment.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 7 (refer to Fig. 4). Blood flowed normally in mesenteric microvessels at 0 h after uPA treatment.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n
Disease Models & Mechanisms 10: doi:10.1242/dmm.027268: Supplementary information
Supplemental Movie 8 (refer to Fig. 4). Hemorrhages formed in mesenteric micorvessels at 2 h after uPA treatment.
Dis
ease
Mo
dels
& M
echa
nism
s •
Sup
plem
enta
ry in
form
atio
n